Λ. 
 
THE 
 
 NEW CRATYLUS 
 
 OR 
 
 CONTRIBUTIONS TOWARDS A MORE ACCURATE 
 KNOWLEDGE 
 
 GKEEK LANGUAGE. 
 
 BY 
 
 JOHN WILLIAM DONALDSON, D.D., 
 
 CLASSICAL EXAMINER IN THE UNIVERSITY OF LONDON; 
 
 AND FORMERLY FELLOW AND CLASSICAL LECTURER OF TRINITY COLLEGE, 
 
 CAMBRIDGE. 
 
 FOURTH EDITION. 
 
 LONDON: 
 LONGMANS, GREEN, AND CO. 
 
 1868. 
 

 TO 
 
 THE MASTER, FELLOWS, AND SCHOLARS 
 
 OF 
 
 TEIOTTY COLLEGE, CAMBRIDGE, 
 
 OEIGINALLY COMPOSED WITHIN THEIR PRECINCTS, 
 
 IS RE-INSCRIBED 
 
 AS 
 
 A RECORD OF THE AUTHOR'S GRATEFUL ATTACHMENT 
 
 TO THAT 
 
 ILLUSTRIOUS FOUNDATION. 
 
•^^s;& 
 
 ,^: 
 
PREFACE TO THE THIED EDITION. 
 
 TT is a great satisfaction and gratification to me that I have 
 been permitted to complete a third edition of this work. The 
 period of exactly twenty years which has elapsed since it was 
 first published, would have been materially diminished, if I 
 had not postponed the re-appearance of the book, both on 
 the present occasion, and when it was last reprinted, until 
 I had fully availed myself of the renewed opportunity of 
 surveying the wide field which I had endeavoured to embrace. 
 But long as the interval between the first and third editions of 
 my book may seem, when compared with the rapid sale of writ- 
 ings of a more popular character, it is worthy of remark that 
 the other principal treatises on comparative philology have 
 afforded their authors still rarer occasions for revision and re- 
 production; that with the exception of a part of the first volume, 
 which came out in 1840, Grimm's Deutsche Grammatik remains 
 in the form which it had reached when the first edition of this 
 work was published ; and thatBopp's Vergleichende Grammatik^ 
 and Pott's Etymologische Forschungen^ which appeared, in the 
 first part of each, in 1833, have just reached the commencement 
 of a second edition respectively. 
 
 In thus acknowledging the favourable reception which has 
 been accorded to this book, I may seem to place it on the same 
 footing as the three German works to which I have referred, 
 and to provoke a direct comparison of my labours and theirs. 
 On the other hand, the fact that these works were at least com- 
 menced before mine, and the references to these and other trea- 
 tises on philology which abound in the following pages, may 
 have countenanced the opinion, which some ill-informed or in- 
 
VI PREFACE TO THE THIRD EDITION. 
 
 considerate readers have been led to entertain, that it has been 
 my object rather to report the result of other men's inves- 
 tigations than to produce a work which would be entitled to claim 
 a distinct and original position in this department of literature. 
 For the credit, therefore, of English scholarship, which is too 
 often subordinated to the learning of Germany, I think it ne- 
 cessary, in publishing a third edition of this work, to direct 
 attention to some of the features which have hitherto distin- 
 guished this book from all other treatises on comparative 
 grammar. The numerous contributions to Greek lexicography, 
 and to the interpretation or correction of particular passages in 
 the Greek authors, do not require any mention, and may be left 
 to speak for themselves. But the independence of the work, 
 as a new contribution to comparative philology in general, 
 rests on the same foundation as that of the primary German 
 treatises, and I venture to think that it has done as much 
 as any other book of the same kind to interpret the facts 
 of language, to classify the phenomena, and to discover the 
 laws which regulate the transmutations of sounds in cognate 
 forms of human speech. In support of this conviction, 1 may 
 appeal to the fact that the most accomplished English phi- 
 lologer of the present century — the late Mr. R. Garnett — has in 
 more than one instance made the principles which were first 
 indicated in this book, the acknowledged starting-point of his 
 own profound and accurate investigations, and that German 
 writers on comparative grammar have tacitly accepted my 
 positions, or have written essays in proof of the combinations 
 which had been already made in my first edition; to say 
 nothing of the numerous and important details in which I 
 have corrected the errors of my most eminent contemporaries. 
 Above all, it must be remembered that this work was, at the 
 time of its first appearance, the only complete treatise on 
 inflected language then in existence either in England or on 
 the continent, and that it inaugurated a new method and a 
 new application of comparative philology, for it was the first 
 attempt to combine speculations afi'ecting the whole fabric 
 
PREFACE TO THE THIBD EDITION. Vll 
 
 of human speech with the established system and well-tried 
 materials of the old classical scholarship. 
 
 But although I am entitled to claim complete originality 
 both for the conception to which this work owes its existence, 
 and for the new combinations, which are exhibited in almost 
 every page, I am most willing to admit the accidental influ- 
 ences which directed my attention to comparative philology at 
 a very early period. It is more than probable that a natural 
 tendency would have led me to engage in lexicographic re- 
 searches, like those of Ruhnken and Lobeck, and to speculate 
 in Greek and Latin etymology, after the manner of Butmann 
 and Doderlein, even if I had never heard of Grimm or Bopp ; 
 but it so happened that, as a student of University College, 
 London, during the first two years of its existence (1828, 1829), 
 I had been made aware of the advantages which might be 
 derived from a study of Sanscrit, and had acquired some 
 knowledge of that language; and having become a contri- 
 butor to the Journal of Education immediately after pro- 
 ceeding to my first degree in 1834, I could not but be struck 
 by Rosen's admirable reviews of Bopp's Comparative Gram- 
 mar^ and Pott's Etymological Researches, which appeared by 
 the side of my own papers on subjects relating to classical 
 scholarship. The time, at which I was thus once more attracted 
 to comparative philology, was the epochal period of that study ; 
 and, for the success of my own special labours, it was for- 
 tunate, as I remarked on a former occasion, that I was enabled 
 to abstain from all general speculations in linguistic science 
 until I had passed through the schools of Bentley and Porson, 
 of Buttmannn and Hermann, of Niebuhr and K. O. Muller, and 
 had enjoyed the advantages common to all those who have 
 encountered the competitive discipline of the University of 
 Cambridge, advantages which the philological students of 
 Germany are quite unable to appreciate. 
 
 The general design of this work was sufficiently stated in 
 the preface to the first edition, and its antagonism to the 
 principles of Home Tooke is intimated without any reserve 
 
Vlll PREFACE TO THE THIRD EDITION. 
 
 in the introductory chapters (§§ 60 — 62, 126). Nevertheless, 
 I have from time to time observed a tendency towards a mis- 
 understanding of my general title, and as this may be the last 
 occasion on which I shall have to write a new preface to the 
 book, it may be worth while to explain why I have called this 
 work "The New Cratylus," and what was its intended form. 
 
 When I was first led to a study of general philology, the 
 "Diversions of Purley" was the standard book of reference 
 in this country; anew edition had been recently published; an 
 English dictionary had been commenced in accordance with its 
 theory; and it was generally understood that its principles were 
 unquestionably sound and valid. At an early period, however, 
 I had convinced myself that HorneTooke's method was not only 
 vicious in itself, but also a mere reproduction of the linguistic 
 sophistries whichPlato had confuted in his Crafi/Ziis.When, there- 
 fore, I had become persuaded that the time had arrived for a 
 radical reform of the current English philology, and saw my 
 way to the attainment of satisfactory results, by making the old 
 classical scholarship of the country my basis and substratum, it 
 was not an unnatural consequence of so wide and ambitious a 
 design that I should follow the established precedent of Francis 
 Bacon, and as he called his treatises in opposition to Aristotle 
 and in imitation of Plato by the now familiar names of Novum 
 Organon and "New Atlantis," I felt myself justified in adopting 
 a similar designation for my onslaught on the Cratylus redivivus 
 of Purley, and the winged words of hisHeracleitean ultra-nomi- 
 nalism. It was at first my intention to make the parallel 
 complete by assuming, as Home Tooke had done, the form of 
 a dialogue ; and I wrote in this way the two chapters from which 
 I proposed to develope the whole theory of language, namely, 
 those in which I discussed the particles denoting a motive 
 (book III. chapter iv.)* and the verbs signifying will and choice 
 (book IV. chapter v.). But I found this machinery too cumbrous 
 
 * It was intended that the whole theory and its details should flow 
 from an inquiry into the origin and structure of ενεκκ. 
 
PREFACE TO THE THIRD EDITION. IX 
 
 for the extent of my proposed undertaking, and without carry- 
 ing the experiment any farther, I adopted the methodical and 
 didactic form in which the work first met the public eye. 
 
 Such then are the claims of this work to a primary and 
 independent position. Such are the circumstances in which 
 it originated, and the design which its author proposed to 
 himself. But its permanent usefulness, as a sufficiently popu- 
 lar introduction to the study of Comparative Grammar and 
 Ethnography, must depend on the systematic completeness 
 with which it treats of the necessary details, and its adap- 
 tation for continuous perusal. With regard to the former, 
 the analytical tables of contents prefixed to the successive 
 chapters will enable the reader to see whether it neglects 
 any part of the subject. And with regard to the mode of 
 exposition, it has certainly been my wish to write a book 
 which might be read from beginning to end with as much 
 ease as the student would listen to a series of lectures on the 
 same heads. In this particular, at least, I do not fear a com- 
 parison with my fellow-labourers in Germany. For I doubt 
 whether any one, who is not already acquainted with phi- 
 lology, would attempt to use the Grammars of Grimm and 
 Bopp except as books of reference, and the new edition, which 
 Pott has commenced of his etymological researches, is a mere 
 farrago of crude materials, a confused lumber-room of ill- 
 arranged information, which the most determined student would 
 not enter without dismay or traverse without wearisomeness. 
 
 In the present edition I have carefully revised every page 
 with reference, not only to the general progress of philological 
 knowledge in the last few years, but also to my own studies 
 during the interval, and I hope that the result will be found 
 in a great number of little improvements. There is not so 
 much enlargement as there was in the second edition com- 
 pared with the first. Still there is a considerable increase of 
 matter; for which I have in some measure made room by 
 a more extensive use of the smaller type, by condensation 
 and by the omission of passages quoted from other authors 
 
χ PEEFACE TO THE THIRD EDITION. 
 
 or adequately represented in other works of my own. And 
 while the principles of linguistic philosophy advocated in this 
 book remain unaltered, I venture to hope that the competent 
 critic will find them in many places confirmed by new argu- 
 ments and illustrations. 
 
 In conclusion, I repeat here, what I said in the preface 
 to the second edition, that continued experience and re- 
 flexion have convinced me of the increasing importance of the 
 task which was for the first time attempted in this work — 
 namely, the prosecution of comparative philology as the safe 
 and ascertained basis of the old classical scholarship. And 
 though a German philologer has at last ventured on a similar 
 undertaking, in which I detect many traces of servile imita- 
 tion, I have no reason to believe that this book has been 
 superseded by any similar treatise either English or foreign. 
 I have reason to know that it has been hitherto of some use 
 in stimulation, guiding and assisting the studies of young 
 philologers in this country, and it has been formally or vir- 
 tually adopted as a text-book by more than one great Uni- 
 versity. Having, then, bestowed great pains on this revision, 
 I venture to renew the hope, with which I concluded the 
 preface to my last edition, that by the increased precision of 
 its results and the greater accuracy of its details it may now 
 contribute in a higher degree to establish a consistent theory 
 of linguistic philosophy, and may connect Greek scholarship 
 by firmer bonds with the general study of human speech and 
 of the co-ordinate laws of thought. 
 
 J. W. D. 
 
 Cambridge, 
 
 June 10, 1859. 
 
PEEPACE TO THE FffiST ΕϋΙΉΟΝ. 
 
 TN writing this book it has been my object to combine an 
 -'"investigation of general principles with an exposition of 
 particular results; I have endeavoured, on the one hand, to 
 establish a consistent and intelligible theory of inflected lan- 
 guage, considered in its most perfect state, that is, at is appears 
 in the oldest languages of the Indo-Germanic family; and, on 
 the other hand, I have attempted to place the Greek scholar- 
 ship of this country on a somewhat higher footing, by rendering 
 the resources of a more comprehensive philology available for 
 the improvement of the grammar and lexicography of the Greek 
 language, and for the criticism and interpretation of the authors 
 who have written in it. If it is thought strange that 1 have 
 not confined myself to one or other of these two sufficiently 
 difficult tasks, I may answer, that in the present state of phi- 
 lology it would be impossible to make any real contribution 
 to Greek scholarship without some sound theory of the philo- 
 sophy of language, and a certain acquaintance with the leading 
 members of the family to which the Greek language belongs ; 
 and, conversely, it would not be easy to write an instructive 
 treatise on the internal mechanism and organization of inflected 
 language, without taking some inflected language, by way, at 
 least, of exemplification. Now of all the languages with mono- 
 syllabic roots the Greek is the most fitted for this purpose. 
 It is, in the first place, a dead language, and therefore fixed 
 and unchangeable ; it is the most copious and expressive of all 
 languages ; it stands mid- way between the oldest form of the 
 Indo-Germanic idioms and the corrupted modern dialects of 
 that family, in other words, it has attained to a wonderfully 
 clear and copious syntax without sacrificing altogether, or in- 
 deed to any considerable extent, its inflexions and power of 
 composition; it has been more studied and is better known 
 than any other dead language, that is, that facts and phenomena 
 are more completely collected and more systematically arranged 
 than is the case with any other, so that allusions to it are 
 
Xll PREFACE TO THE FIEST EDITION. 
 
 more generally intelligible, and deductions, or new combina- 
 tions of laws, derived from it, are safer and more convincing; 
 above all, the value of the literature and the actual demand 
 for a knowledge of the language, should induce us to turn 
 upon the Greek, rather than upon the Gothic, the Latin, or 
 the Sanscrit, any new light which the doctrine of words may 
 have gained from investigations in the philosophy of language 
 or in comparative grammar. 
 
 Every didactic work is or ought to be adapted to the wants 
 of some particular class of readers, and should presume, in 
 them, a certain amount of preparatory knowledge and no more. 
 I have written, then, first, for Englishmen, who are not sup- 
 posed to be intimately or extensively acquainted with the phi- 
 lological literature of the continent: and secondly, I have written 
 for persons who possess at least some slight knowledge of the 
 Greek language, and would rather increase it by investigating 
 the principles of the language and endeavouring to discover 
 the causes of its grammatical peculiarities, than by overloading 
 the memory with a mass of crude, incoherent facts, which can 
 neither be digested nor retained. I have also wished to give 
 those who come to the study of Greek with no higher aim 
 than to make it the means of obtaining University distinctions, 
 an opportunity of learning from it the dignity of human speech, 
 of perceiving how little of the casual and capricious there is in 
 language, and of convincing themselves that in this, as in other 
 things, there are laws to combine, regulate, and vivify the 
 seemingly disjointed, scattered, and lifeless phenomena. It is 
 possible that the novelty of some of my speculations may induce 
 maturer scholars to take up this book. If so, they will under- 
 stand from this statement, why I have here and there entered 
 upon long explanations of peculiarities, which can occasion no 
 difficulty to the philologer or have been already discussed by 
 German or French writers, and, on the other hand, why I have 
 despatched with a hint or a reference some really difficult 
 questions, in which the young student could take no interest, 
 while the scholar would comprehend my meaning from a single 
 word. 
 
 Many people entertain strong prejudices against every thing 
 in the shape of etymology, prejudices which would be not only 
 just but inevitable, if etymology or the doctrine of words were 
 such a thing as they suppose it to be. They consider it as 
 
PEEFACE TO THE FIRST EDITION. XIU 
 
 amounting to nothing more than the derivation of words from 
 one another; and as this process is generally confined to" a 
 perception of some prima facie resemblance of two words, it 
 seldom rises beyond the dignity of an ingenious pun, and, 
 though amusing enough at times, is certainly neither an in- 
 structive nor an elevated employment for a rational being. 
 The only real etymology is that which attempts a resolution 
 of the words of a language into their ultimate elemen^^ by a 
 comparison of the greatest possible number of languages of 
 the same family Derivation is, strictly speaking, inapplicable, 
 farther than as pointing out the manner in which certain con- 
 stant syllables, belonging to the pronominal or formative ele- 
 ment of inflected languages, may be prefixed or subjoined to 
 a given form for the expression of some secondary or depen- 
 dent relation. In order to arrive at the primary origin of a 
 word or a form, we must get beyond the narrow limits of a single 
 idiom. Indeed, in many cases the source can only be traced 
 by a conjectural reproduction based on the most extended com- 
 parison of all the cognate languages, for when we take some 
 given variety of human speech, we find in it systems and series 
 of words running almost parallel to one another, but presenting 
 such resemblances in form and signification as convince us that, 
 though apparently asymptotes, they must have converged in 
 the form which we know would potentially contain them all. 
 This reproduction of the common mother of our family of lan- 
 guages, by a comparison of the features of all her children*, is 
 the great general object to which the efforts of the philologer 
 should be directed, and this, and not a mere derivation of 
 words in the same language from one another, constitutes the 
 etymology that is alone worthy of the name. 
 
 As far as this work is a contribution to the better know- 
 ledge of Greek in particular, I wish it to be understood, that 
 I have by no means confined myself to etymological researches, 
 but have endeavoured to avail myself of every resource of 
 scholarship, as well old-fashioned as new. The words, which I 
 have attempted to explain, are those which have either occa- 
 sioned peculiar difficulty to the young student, or the meaning 
 
 * I am told that some similar idea is to be found in Campbell's 
 Gertrude of Wyoming ^ a poem which, I am ashamed to say, I have 
 never read. 
 
XIV PEEFACE TO THE FIRST EDITION. 
 
 of which has been considered doubtful by scholars. Where I 
 have thought proper to make a digression for the purpose of 
 interpreting or emending a particular passage, I have always 
 had in view that class of students with whom my experience in 
 teaching has brought me most immediately in contact, and with 
 whose wants and difficulties I am best acquainted. It will be 
 seen, too, that in the selection of passages for this purpose, I 
 have generally confined myself to those authors who are most 
 read in the great schools and Universities of this country. In 
 this part of the work, I have been guided mainly by considera- 
 tions of practical utility, namely, by a wish to assist those whose 
 business it is to construe Greek authors, and to write Greek 
 exercises. It is for this reason that I have preserved, as far 
 as possible, the old grammatical nomenclature: the young 
 student regards with a sort of mysterious reverence the uncouth 
 terms of his grammar ; they are little household gods to him : 
 and, though, like the Lar familiaris of old, they are unseemly to 
 look upon and unavailing to help, there appears to be no good 
 reason why one should take them down from the niches, which 
 they have so long and so harmlessly occupied. 
 
 It is painful and humiliating to reflect, how much, after all 
 one's thought and labour, the execution of a task like this must 
 fall short, not merely of the exactions of a rigorous criticism, 
 but even of one's own imperfect conceptions. It may be, indeed, 
 that what I have attempted in this book is not yet to be 
 effected by one man and at one effort, and perhaps, in reference 
 to its wider scope, all that 1 can hope to do, is to awaken the 
 dormant energies of some young student, who may be qualified 
 at a future period to solve completely and finally the great 
 problem of inflected language; — άλλα καΐ εταχείρονντί τον τοΐζ 
 καλοΐζ καλόν καΐ τίάύχείν ο τι αν τω ζνμβ^ ηα^ύν. 
 
 J. W. D. 
 
 Trinity College, Cambridge, 
 4.th February, 1839. 
 
CONTENTS. 
 
 BOOK I. 
 GENERAL INTRODUCTION. 
 
 CHAPTER SECTIONS PAGES 
 
 I. The utility of philological studies ... 1—17 3— 25 
 
 II. The history and present state of philology. 17 — 40 26 — 64 
 
 III. The philosophy of language 41— 62 65—111 
 
 IV. The ethnographic affinities of the ancient 
 
 Greeks 63— 97 112—166 
 
 V. The theory of the Greek alphabet . . . 98—122 167—217 
 Appendix to Chap. V. § 110. Extracts from 
 
 Bentley's MS. on the digamma .... 218 — 225 
 
 VI. The parts of speech 123—128 226—238 
 
 BOOK II. 
 
 PRONOMINAL WORDS. 
 
 I. The personal and other pronouns . . . 129 — 152 241 — 278 
 
 II. The numerals 153—167 279—311 
 
 III. The prepositions 168—187 312—355 
 
 IV. The negative and other particles . . . 188—205 356—384 
 
 BOOK ΠΙ. 
 
 THE NOUN. 
 
 I. The roots of nouns and verbs .... 206—226 387—420 
 
 II. The case-endings of the noun .... 227—251 421—448 
 in. The pronominal terminations of the unin- 
 
 flected forms 252—270 449—477 
 
XVI CONTENTS. 
 
 CHAPTER SECTIONS PAGES 
 
 IV. Nouns used as prepositions 271 — 292 478—501 
 
 V. The adjective 293—306 502—523 
 
 VI. Compound words 307—344 524—566 
 
 BOOK IV. 
 
 THE YERB. 
 
 I. The person-endings 345—366 568—592 
 
 II. The tenses 367—387 593—616 
 
 III. The moods and participles 388—424 617—653 
 
 IV. The conjunctions 425—443 654—671 
 
 V. The use of auxiliary verbs in Greek . . 444—481 672—724 
 
BOOK I 
 
 GENERAL INTRODUCTION. 
 
THE 
 
 NEW CRATYLUS. 
 
 BOOK L 
 
 GENERAL INTRODUCTION. 
 
 CHAPTER I. 
 
 THE UTILITY OF PHILOLOGICAL 
 
 STUDIES. 
 
 1 Motives for a preliminary inquiry respecting the practical usefulness of phi- 
 lological learning. 2 Education, information, and knowledge, often confused. 
 1. Philology necessary to education. 3 Definition of Philology. 4 Liberal 
 and professional education. 5 Philology contributes to liberal education by 
 teaching deductive habits. 6 Study of dead languages recommended by their 
 fixity. 7 Advantage of learning any foreign language. 8 Value of ancient 
 literature. 9 Comparativegrammar leads to extensive acquisitions. II. Phi- 
 lology an important branch of general knowledge. 10 Worth and dignity of 
 ethnographical science. 11 Changes of population and government clearly 
 indicated by language. 12 Study of language belongs to a great branch of 
 inductive Philosophy. III. Philology valuable as the method of interpretation. 
 13 Historical criticism derived from Philology. 14 The philologer mediates 
 between reason and tradition, and pleads for a maximum of belief. 15 Im- 
 portance of Philology for the divine, both as the method of interpretation, and 
 a branch of ethnographical science. 16 Classical education, to whatever ex- 
 tent it is carried, ought to be rational and philological. 
 
 1 TT may be stated as a fact worthy of observation in the 
 •^ literary history of modern Europe, that generally, when 
 one of our countrymen has made the first advance in any branch 
 of knowledge, we have acquiesced in what he has done, and have 
 left the further improvement of the subject to our neighbours 
 on the continent. The man of genius always finds an utterance, 
 for he is urged on by an irresistible impulse— a conviction that 
 it is his duty and his vocation to speak: but we too often want 
 those who should follow in his steps, clear up what he has left 
 obscure, and complete his unfinished labours. Nor is it difficult 
 
 B2 
 
4 THE UTILITY [book I. 
 
 to show why this should be the case. The English mind, vigorous 
 and healthy as it generally is, appears to be constitutionally 
 averse from speculation ; we have all of us a bias towards the 
 practical and immediately profitable, generated by our mercantile 
 pursuits, which make all of us, to a certain extent, utilitarians, 
 and stifle the development of a literary taste among us; or, if 
 the voice of interest fails to control the vanity of authorship, 
 there is still another modification of self-love, a cold conventional 
 reserve, induced by the fear of committing one's self, which im- 
 poses silence upon those who have truths to tell. 
 
 To this general fact, however, there is one very remarkable 
 exception. The regulations of our grammar-schools, and, per- 
 haps, somewhat of the old custom and antiquated prejudice, of 
 which w^e hear so much, have made classical studies not only the 
 basis, but nearly the whole of a liberal education in this country; 
 and circumstances , which we shall point out in the following 
 chapter , have created for us a thriving philological literature. 
 Although the rewards and encouragements held out by our 
 great Universities have been considered by many as a sufficient 
 justification of such studies, it is the spirit of the age to inquire, 
 what advantage a young man derives from so protracted a study 
 of Latin and Greek, in addition to and independent of the Uni- 
 versity distinctions and emoluments which he may have the good 
 fortune to obtain. There is much of reason in this demand, and 
 it is doubtless incumbent upon those who have devoted them- 
 selves to such pursuits to point out to others their importance 
 and utility. Hitherto this has not been done in a satisfactory 
 manner; and therefore, although our object is rather to add 
 something to philological knowledge than to justify philological 
 pursuits , we deem it a necessary preliminary that we should 
 endeavour by some plain arguments to recommend to our readers 
 the sort of learning which we wish to increase and the studies 
 which we design to facilitate — that we should make known at 
 the very outset the nature and value of the subject on which 
 we write. And in doing this we disclaim any wish to perplex 
 ourselves with the polemics of the question, as it has been treated 
 by other writers. It is not our purpose to discuss the merits 
 or demerits of our collegiate institutions, still less to impugn or 
 exculpate, as the case may be, the conduct of those who are 
 
CHAP. Ι.] OF PHILOLOGICAL STUDIES. 5 
 
 intrusted with the management of them : least of all would we 
 assert that there is no room for improvement in the present method 
 of our classical studies; on the contrary, we hope and indeed 
 expect that, sooner or later, they will be pursued in a healthier 
 and a manlier spirit, that much that is superfluous will be re- 
 trenched, much that is useful added, so that even the educational 
 theorist may at length admit that there is something more in 
 nouns and verbs than was dreamt of in his philosophy. Our 
 only aim in this place is to satisfy the practical sense of our 
 countrymen with regard to the real uses of philology, properly 
 pursued: how it is and has been prosecuted will appear in the 
 next chapter*. 
 
 2 The cause of all the unprofitable discussions which have 
 arisen respecting the utility of particular branches of study is to 
 be sought in the vague and erroneous manner in which we use 
 the terms education, information, and knowledge. We are in 
 the habit of speaking of mere information as though it were the 
 same thing as exact knowledge, and we still more frequently 
 allow special or professional knowledge to assume the honours 
 which are due to general education. It is surely desirable that 
 these terms should be properly defined, and used only according 
 to their true signification; for there is no realism more oppressive 
 than the dominion of terms which stand as the representatives 
 of indefinite ideas. We believe that the following distinctions 
 will be found to agree with the opinions of the majority of re- 
 flecting men in this country. The term Education, which signifies 
 "a leading out," or "bringing up," is particularly applied to the 
 training of the young: but is is equally applicable to any process 
 which is calculated to discipline an uncultivated mind, whether 
 the bodily growth be matured or not. The idea conveyed by the 
 "word might be explained in metaphorical language as a bringing 
 forth from darkness into light, — it is a leading up from some 
 
 * The author has fully discussed all the questions immediately con- 
 nected with the modern developments of intellectual discipline in a sepa- 
 rate work, intitled: Classical Scholarship and Classical Learning considered 
 with especial reference to competitive tests and University teaching : a prac- 
 tical essay on liberal education. Cambridge, 1856. 
 
6 THE UTILITY [book I. 
 
 narrow and confined valley to the summit of a lofty mountain, 
 whence the elevated soul obtains a Pisgah view of truths and 
 duties — it is a careful survey of the domains of intellectual and 
 moral principles, which stretch before us when the sun-light of 
 reason has cleared away the mists of vulgar prejudice. We fall 
 into a mistake if we suppose that education is limited to mental 
 culture; it may be social and moral, as well as intellectual; and 
 we even give the name of spiritual education to that higher moral 
 training which emanates from the schooling of Christianity. 
 But to confineourselves,for ourpresentpurpose,toits intellectual 
 province , we may say that Education is properly a cultivation 
 and development of those reasoning faculties, which all men 
 have in common, though not all in the same degree. The term 
 Information^ on the contrary, although, according to the origin 
 of the word, it ought to be synonymous with intellectual educa- 
 tion, is generally understood to signify only an accumulation of 
 particular facts. When we speak of a well-informed man , we 
 generally mean some one who is able to return plausible answers 
 to the catechism of ordinary conversation; and the common 
 phrase '^a smattering of information on all subjects" shows that 
 the term is not supposed to imply a profound or extensive ac- 
 quaintance with any one branch of knowledge. In fact, so long 
 as information is onli/ information, it merely denotes an accumu- 
 lation of stray particulars by means of the memory. On the 
 other hsind^Knowledge is information appropriated and thoroughly 
 matured. It implies experience and practice, and it differs from 
 information as the food, which is taken into the system, and to 
 which we owe our strength and growth, differs from the gar- 
 ments which hang loosely about us, and which may be laid aside 
 or worn out. We must not however forget that information 
 maybe concentrated and ripened into knowledge; for knowledge 
 begins with and presumes information ; though information does 
 not presume or include knowledge. Our common phrases show 
 that this is the meaning of the term. We speak of knowledge 
 of the world, knowledge of our profession or business, knowledge 
 of ourselves, knowledge of our duties — all of which imply a 
 completeness and maturity of habit and experience. When 
 knowledge extends to a methodical comprehension of general 
 laws and principles, it is called science. It is the natural and 
 
CHAP. I.] OF PHILOLOGICAL STUDIES. 7 
 
 proper tendency of information tb ripen into knowledge, just as 
 knowledge itself is not complete until it is systematized into 
 Science* ; but as the dijfference between information and know- 
 ledge is one of kind rather than of degree, it is clear that no mere 
 accumulation of useful information, not even though it equalled 
 all the stores laid up in Mr Maunder's treasure-houses, would 
 amount to exact knowledge or scientific acquirement. We do 
 not think it worth while, therefore, to show that philology is a 
 branch of useful information. If it does not contribute to valu- 
 able and important knowledge, or if it is not amnllary to the 
 best kind of education, we shall waste our time in pleading for 
 the utility of a study which necessarily demands minute atten- 
 tion and laborious research. 
 
 3 We maintain, then, first, that a certain amount of phi- 
 lology is necessary as the basis of a liberal education; and 
 secondly, that, cultivated to its fullest extent, philological 
 scholarship furnishes valuable and essential contributions to 
 general science^ and to some of the most important applica- 
 tions of human knowledge. 
 
 Under the name philology we include the two great branches 
 of a scientific inquiry into the principles of language; — the 
 theory of the origin and formation of words, which is generally 
 called the philosophy of language; and — the method of lan- 
 guage, or, as it is more usually termed, logic or dialectic, which 
 treats of the formation of sentencesf . As a specific department 
 
 * We have a striking exemplification of this in the series of works 
 published by the Society for the Diffusion of Useful Knowledge. Beginning 
 with an attempt to disseminate popular information on a variety of sub- 
 jects, these treatises very soon aimed at communicating exact and scientific 
 knowledge, and some of them are even replete with learning derived from 
 the older schools of classical philology. 
 
 f This appears to be the real extent of the term philology. While 
 W. von Humboldt, however, would confine it to that department which 
 is conversant about the interpretation of the written monuments of a 
 language, as distinguished from the analysis of its structure and compa- 
 rison with other idioms, which he calls Ldnguistik {iiber die Verschied. d. 
 menschl. Sprachhaues, p. 202) ; his most enthusiastic disciple has claimed for 
 philology a domain which includes all recorded knowledge. "It appears 
 to me," says H. Steinthal (Z)e Pronomine Relativo Gommentatio Philoso- 
 
8 THE UTILITY [book I. 
 
 of study, philology deals with language for its own sake; and 
 though it originated in the verbal criticism and interpretation 
 of a literature which had become classical or sacred, it does not 
 properly comprehend all that falls within the province of the 
 editor and commentator. The philologer, as such,, does not deal 
 with the subject-matter of the authors, whose diction he ex- 
 amines with such minute and searching accuracy; nor does he 
 undertake to expound the rules of rhetoric and the theory of 
 taste. He has exhausted his proper functions, when he has in- 
 vestigated scientifically all that relates to the ultimate analysis of 
 the separate terms, and the construction of the subject and predi- 
 cate. These two departments are comprised in general grammar, 
 which is therefore identical with philology, and have also their 
 representatives in the etymology and syntax of every particular 
 
 pkico-Philologica, Berolini, 1847, pp. 4, 5), "that it is the business of the 
 human understanding, or of literature in general, to comprehend those 
 simple and absolute laws which appear in the world or in nature on the 
 one hand, and in the history of the human race on the other hand. As, 
 therefore, there are two forms of literature, — one, the history of nature^ 
 or physiology; the other, the history of the human mind, — philologers 
 undertake the examination of all that the λόγος, or human reason, has 
 produced. Now, whatever the human reason produces is some idea, 
 something recognised and discerned by the mind, although it may be 
 clothed in some outward form, whether it be a form of government con- 
 stituted by human society: or some monument of hewn stone: or some 
 type of mythology and religion: or some demonstrative result of philo- 
 sophical acuteness: or some outpouring of poetical genius or oratorical 
 eloquence. So that even the history of philology belongs to philology, 
 with this limitation, that, e.g.^ the history of classical philology is the 
 specialty of those who consider modern life from a philological point of 
 view. Accordingly, the only true definition is Bockh's, that philology is 
 the teaching and learning of that which is already discovered {philologiam 
 esse cogniti cognitionem)-, which is not to be understood, as though philolo- 
 gers were always doing over again the work done to their hands; but all 
 the products of the human mind which remain as recorded facts have to 
 be submitted afresh to the crucible of human thought, to the end that, 
 being recognised, not as the arbitrary acts of individuals, but as sprung 
 from the necessary laws of minds individually free, they may be regarded 
 as a mirror or picture of the human reason in general." We have fully 
 considered the various applications of philological research, as indicated 
 by its origin or its procedure, in the Encyclopedia Britunnica, ed. 8, Vol. 
 XVII. article "Philology." 
 
CHAP. I.] OF PHILOLOGICAL STUDIES. 9 
 
 grammar. Although they are but component parts of one 
 science, it is of the utmost consequence that they should not be 
 confused or interchanged: for it is not too much to say that the 
 most signal mistakes of philologers may be traced to the prac- 
 tice, hitherto so common, of supposing that the formation of 
 words may be discussed on a logical basis. In endeavouring, 
 then, to estimate the importance of philology we must consider 
 as separate questions, what is the use of etymology or the 
 doctrine of words, and of logic or the doctrine of sentences: 
 including under the latter all that belongs to the method of 
 language, and under the former whatever pertains to its origin 
 and generation. And in the first place it is to be shown, that 
 the rudiments of philology in both its branches are or ought to 
 be the basis of the intellectual training of man, or of that edu- 
 cation which is alone worthy of the name. As logic, or the 
 method of language, though properly secondary to etymology, 
 is of more ancient discovery, we shall consider it first. 
 
 4 From what has been already said it will be seen that we 
 distinguish between education properly so called, and the train- 
 ing which is necessary for the successful prosecution of any 
 profession or business. The former, as has been already said, 
 is designed for the cultivation of the intellect and the develop- 
 ment of the reasoning faculties. The latter is intended to adapt 
 a man for some particular calling, which the laws of society, on 
 the principle of the division of labour, have assigned ^to him 
 as an indiAddual member of the body politic. Now the training 
 of the individual for this particular purpose is not an education 
 of man as such; he might do his particular work as well, or 
 better, if you deprived him of all his speculative faculties and 
 converted him into an automaton ; in short, the better a man is 
 educated professionally the less is he a man; for, to use the 
 words of an able American writer* , " the planter who is Man 
 sent out into the field to gather food, is seldom cheered by any 
 idea of the true dignity of his ministry. He sees his bushel and 
 his cart, and nothing beyond, and sinks into the farmer, instead 
 
 * See An Oration before the Phi-Beta-Kappa Society, by Ralph 
 Waldo Emerson, p. 5. 
 
10 THE UTILITY [book I. 
 
 of Man on the farm. The tradesman scarcely ever gives an 
 ideal worth to his work, but is ridden by the routine of his craft, 
 and the soul is subject to dollars. The priest becomes a form; 
 the attorney, a statute-book; the mechanic, a machine; the 
 sailor, a rope of a ship." It was for this reason that the clear- 
 headed Greeks denied the name of education (τζαιδείοί) to that 
 which is learned, not for its own sake, but for the sake of some 
 extrinsic gain or for the sake of doing some work; and dis- 
 tinguished formally between those studies which they called 
 liberal, or worthy of a free man , and those which were merely 
 mechanical and professional*. In the same way Cicero speaks 
 of education properly so called, which he names humanity 
 (/mmanitas)f^ because its object is to give a full development 
 to those reasoning faculties which are the proper and distinctive 
 attributes of man as suchj. Now we do not pretend that 
 philology is of any mechanical or professional use, unless the 
 business of the teacher is to be regarded as a professional 
 employment: we do not say that philology will help a man to 
 plough or to reap ; but we do assert that it is of the highest 
 use as a part of humanity, or of education properly so called. 
 
 5 The test of a good education is the degree of mental 
 culture which it imparts; for education, so far as its object is 
 scientific, is the discipline of the mind. The reader must not 
 overlook what is meant by the word "mind" when used in 
 reference to education. That some dumb animals are possessed 
 of a sort of understanding is admitted; but it has ne\^er been 
 asserted, by those who pretend to accuracy and precision of 
 language, that they enjoy the use of reason, Man, however, has ' 
 the faculty called reason in addition to his understanding; lie 
 has a power of classifying or arranging, abstracting and gene- 
 
 * See Plato, hegg. i. p. 643 e: τκντην την τροφην (την προς άρετην ίκ 
 ίΐούδων παιδείαν) άφοριαύμενος 6 λόγος οντος, ώς έμοί φαίνεται, ννν βού- 
 λοίτ' αν μόνην τζαίδείανηροΰαγορενειν, την δε εις ;ι;ρ?;^αΓ0{ τείνον- 
 ααν η τίνα τίρος ίοχνν η καΐ προς αλλην τινά οοφίαν ανεν νον τιαϊ δίκης βά- 
 νανΰόν τε είναι yial άν ελενϋ' ερον κκί ονκ άξίαν το παράπαν παι- 
 δείαν τιαλεΐϋ&αι. Similarly Aristotle, PoUt. viii. c. 2. 
 f Pro Archia Poeta, 1; De Oratore, i. 9. 
 + Aul. Gellius, xiii. 16. 
 
CHAP. I.] OF PHILOLOGICAL STUDIES. 11 
 
 ralizing, and so arriving at principles*; in other words, his 
 mind is capable of method : and thus it has been well said that 
 we at once distinguish the man of education, or, among men of 
 education, the man of superior mind, by the unpremeditated and 
 evidently habitual arrangement of his words , grounded on the 
 habit of foreseeing, in every sentence, the whole that he intends 
 to communicate in the particular case, so that there is method in 
 the fragments of his conversation even when most irregular and 
 desultory! . Accordingly, what we mean by saying that the 
 object of education is the cultivation of our minds, or that the 
 goodness of an education varies with the degree of mental cul- 
 ture, amounts simply to this, that we better perform our functions 
 as rational creatures in proportion as we carry farther the dis- 
 tinction between ourselves and the brute creation, that is, in pro- 
 portion as we are the better fitted for the discourse of reason. 
 
 There are two ways in which we carry on the process of 
 reasoning, just as there are two relations out of which all method 
 or science is made up. The relations are, that of Law, by which 
 we lay down a rule of unconditional truth which we call an 
 Idea, and that of Observation, by which we get to a distinct 
 knowledge of facts. By the former Λνβ know that a thing must 
 be; by the latter we see that it is. Now when we reason from 
 the facts to the law, we call it analysis or induction; when 
 we reason from law to law, Avhen from a known truth we seek 
 to establish an unknown truth, we call the process deduction or 
 synthesis. As then all science is made up of Law and Observa- 
 tion , of the Idea and the Facts , so all scientific reasoning is 
 either induction or deduction. It is not possible, however, to 
 teach inductive reasoning, or even to cultivate a habit of it 
 directly ; we all reason inductively every moment of our lives, 
 
 * As the general reader may not perhaps be familiar with the Kantian 
 distinction of reason and understanding, it may be mentioned, that, accord- 
 ing to the critical philosophy, understanding is the faculty of rules, derived 
 from experience, and proverbially subject to exceptions, but reason the 
 faculty of principles or laws, to which there is no exception: the former 
 is the faculty of the unity of phsenomena by means of rules, the latter 
 the faculty of the unity of the understanding-rules under principles (Kritik 
 der reinen Vernunft, pp. 258, 260. 7th edition). 
 
 •}■ Coleridge's Friend, Vol. in. pp. 133 foil. 
 
12 THE UTILITY [book I. 
 
 but to reason inductively for the purposes of science belongs 
 only to those whose minds are so constituted that they can see 
 the resemblances in things which other men think unlike, in 
 short, to those who have powers of original combination and 
 whom we term men of genius. If, therefore, we can impart by 
 teaching deductive habits, education will have done its utmost 
 towards the discipline of the reasoning faculties. "When we 
 speak of laws and ideas we must not be misunderstood as 
 wishing to imply any thing more than general terms arrived at 
 by real classification. About these general terms and these alone 
 is deductive reason conversant, so that the method of mind, 
 which is the object of education, is nothing but the method of 
 language; and this is the reason why, as we have said, the 
 educated man is known by the arrangement of his words. 
 Hence, if there is any way of imparting to the mind deductive 
 habits, it must be by teaching the method of language; and this 
 discipline has in fact been adopted in all the more enlightened 
 periods of the existence of man. It will "be remembered, that 
 in this method of language it is not the words but the arrange- 
 ment of them which is the object of study; and thus the method 
 of language is independent of the conventional significations of 
 particular words; it is of no country and of no age, but is as 
 universal as the general mind of man. For these reasons we 
 assert that the method of language, one of the branches of 
 philology, must always be, as it has been, the basis of education 
 or humanity as such, that is, of the discipline of the human 
 mind. We may even go farther, and assert, that, when 
 Geometry is added to Grammar, we have exhausted the known 
 materials of deductive reasoning, and have called in the aid 
 of all the machinery which is at our disposal. 
 
 AVith regard to the importance of etymology as a part of a 
 liberal education very little need be said. It is just as necessary 
 that the educated man should be able to select and discriminate 
 the words which he employs, as that he should be able to 
 arrange them methodically. We acquire our mother-tongue 
 insensibly and by instinct, and to the untrained mind the words 
 of it are identified with the thoughts to which they correspond 
 in the mind of the individual, whereas he ought at least to be 
 taught so much of their analysis as to know that they are but 
 
CHAP. I.] OF PHILOLOGICAL STUDIES. 13 
 
 outward signs, the symbols of a prima facie classification, and 
 to employ them accordingly. In this simplest form etymology 
 is nothing but an intelligent spelling lesson, which the most 
 violent utilitarian would hardly venture to discard. When, 
 however, we remember that the most important result of intel- 
 lectual education is the overthrow of one-sided prejudices, and 
 when we reflect how apt we are to fall into practical Realism, 
 and "to apply the analytical power of language to the interpre- 
 tation of nature*," we cannot value too highly that habit of 
 dealing with words, which leads us to distinguish accurately 
 between the mere sign and the thing signified. 
 
 6 But, though perhaps every one will at once allow that 
 such a knowledge of language as we have described is an essen- 
 tial element of intellectual training, it may still be asked, IVhat 
 has this to do with the study of two dead languages? In the 
 first place, then, to study one branch at least of philology,namely, 
 Etymology, we must have some particular language in which to 
 study it; and although the method of language is independent 
 of any particular language, yet, like every other method or 
 science, it must have its facts as well as its laws. It will be 
 conceded that if we would go beyond the rudiments of spelling 
 and speaking, if we would catch a glimpse of what speech is in 
 itself and as detached from ourselves , it would be desirable to 
 select some foreign language, and if possible one no longer spoken 
 or liable to change : for idioms still in use are so fluctuating and 
 uncertain, that an attempt to get fixed ideas of the general 
 analogy of language from them is like trying to copy the fan- 
 tastic pictures of an ever-revolving kaleidoscope. The classical 
 languages lie before us in gigantic and well-preserved remains, 
 and we can scrutinize, dissect, and compare them with as much 
 certainty as we should feel in experimenting upon the objects of 
 any branch of natural philosophy. They are, therefore, well 
 adapted to supply us with the facts for our laws of speech or the 
 general analogy of language; and we might make them the basis 
 of our grammatical study, even though they had nothing to 
 recommend them but their permanence of form and perfection 
 of grammatical structure. 
 
 Hampden, Bampton Lectures, p. 88. 
 
14 THE UTILITY [book I. 
 
 7 This, however, is not all: it is indeed necessary to study 
 some language, and that too a dead language, in order to give 
 the mind a full grammatical training; but the mere fact of learn- 
 ing another language, whether dead or living, is in the highest 
 degree beneficial. We learn our own language from the lips of 
 a mother or a nurse, it grows with our growth, and strengthens 
 with our strength, so as to become a sort of second self; and 
 the words of the uneducated are household gods to him. This 
 idolatry is shaken, the individual is brought away from his own 
 associations to the higher truths which form the food of the 
 general mind of man , whenever he has learned to express his 
 thoughts in some other set of \vords. It was a great mistake of 
 Ennius to say that he had three hearts because he understood 
 three languages (Aulus Gellius, Nodes Atticce^ xvii. 17); the 
 heart of a people is its mother-tongue only (Jean Paul, xlvii. p. 
 173). The Emperor Charles the Fifth was nearer the truth when 
 he said — autant de langues que VJiomme sgait parler, autant de 
 fois est-il homme; — for every language that a man learns he 
 multiplies his individual nature and brings himself one step 
 nearer to the general collective mind of Man. The effect of 
 learning a language, then, consists in the contrast of the asso- 
 ciations, which it calls up, to those trains of thought which our 
 mother-tongue awakens. In this again the dead languages 
 possess a great advantage over every living one. It has been 
 well remarked "that our modern education consists in a sfreat 
 measure in the contrast between ourselves and classical an- 
 tiquity*;" it is a contrast produced by a sleep of more than a 
 thousand years between the last of the great men of old and the 
 first of the great moderns, when the reawakened world looked 
 with instructive astonishment upon its former self. 
 
 8 In addition to the two reasons which we have stated as 
 grounds for preferring the two classical languages as materials 
 of grammatical study, there is a third reason which has generally 
 been thought to be alone sufficient, — the value of the literature 
 to which they are a key. On this particular subject we do not 
 intend to dAvell; books without number have been written upon 
 
 * W. von Humboldt, uber die Verschiedenheit des menschlichen 
 Sprachbaites, p. 27. 
 
CHAP. I.] OF PHILOLOGICAL STUDIES. 15 
 
 it, and there does not seem to exist a doubt as to the paramount 
 excellence of the Greek and Latin writers. To those who still 
 argue the old question about the comparative merits of modern 
 and ancient literature, it is sufficient to answer, that, if the old 
 classical literature were swept away, the moderns whom they 
 so admire would in many cases become unintelligible and in 
 all lose most of their characteristic charms*. And, indepen- 
 dently of the influence which Greek literature has produced, 
 both directly and through the Latin, on that of Modern Europe, 
 and of those special causes, which have made it, as a whole, 
 the inalienable heir-loom of the highest civilization, the greatest 
 inheritance of genius and wisdom, and the most effective in- 
 strument of liberal culture that the world has ever produced, 
 we must recollect that all ancient records have a value, which 
 no modern efforts can replace , in linking the thoughts of the 
 present to the recollections of the past, and in laying a firm 
 foundation for the hopes of the future. Literature, as we have 
 elsewhere said, does not admit of perpetual recommencements and 
 repetitions, and when perfection has been once attained in any 
 department of intellectual productiveness , subsequent genera- 
 tions and other races of men, who have access to the treasures of 
 recorded wisdom, feel themselves constrained to abandon a fruit- 
 less rivalry, and to work out the expression of their own thoughts 
 according to the established model and exemplarf . 
 
 9 Lastly, the introduction of that branch of philology which 
 we call comparative grammar offers a great recommendation to 
 the careful study of these two languages. Notwithstanding the 
 beneficial contrast which they present, they are aged sisters 
 of our own mother-tongue, and studied according to the true 
 philological method in combination with the Asiatic mem- 
 bers of the family, they open the way to an easy and speedy 
 acquirement of every one of the Lido -Germanic languages, 
 and are thus a key to the greatest treasure which the mind of 
 man has collected, — the recorded wisdom of the Caucasian race. 
 
 * See Sedgwick, Discourse on the Studies of the University, 4th edit. p. 36 ; 
 and Whewell, On the Principles of University Education, p. 35. 
 
 f See the article Philology in the Encyclopedia Britanmca, Vol. xvii. 
 ed. 8, and Literature of Greece, Vol. iir. p. 409. 
 
16 THE UTILITY [book I. 
 
 10 From what we have said it appears that for the mental 
 training of the individual some philology is necessary; that 
 grammar is best studied through the classical languages; that 
 the study of these languages is also recommended by their con- 
 trast to our own, by the value of the literature to which they are 
 the key, and by their place in the family of languages to which 
 our own tongue belongs. These are reasons why the individual 
 who is to be liberally educated, should study Greek and Latin. 
 But the advantages of philological studies are not confined to the 
 individual. They may be cultivated to a higher degree than is 
 necessary for the mere purposes of education , and be made to 
 contribute to some of the most valuable and interesting applica- 
 tions of human knowledge. The claims of ethnological philology 
 to rank as a principal branch of general science have been suf- 
 ficiently vindicated of late years. The British Association for 
 the Advancement of Science, at its meeting in 1847, was thus 
 addressed by Bunsen*: "If man is the apex of the creation, it 
 seems right, on the one side, that a historical inquiry into his 
 origin and development should never be allowed to sever itself 
 from the general body of natural science, and in particular from 
 physiology. But , on the other hand, if man is the apex of the 
 creation, if he is the end to which all organic formations tend 
 from the very beginning; if man is at once the mystery and the 
 key of natural science; if that is the only view of natural science 
 worthy of our age — then ethnological science, once established 
 on principles as clear as the physiological are, is the highest 
 branch of that science for the advancement of which this Asso- 
 ciation is instituted. It is not an appendix to physiology or to 
 any thing else; but its object is, on the contrary, capable of 
 becoming the end and goal of the labours and transactions of a 
 scientific institution." Those who are jealous for the dignity 
 of man will not fail to echo these sentiments. Ethnology, which 
 treats of the different races into which the human family is sub- 
 divided, and indicates the bonds which bind them all together, 
 has not only appropriated to itself all the functions of the anthro- 
 pology, which discussed the natural and moral, the physical and 
 metaphysical history of man, but has exacted contributions from 
 
 Reporty p. 257. 
 
CHAP. I.] OF PHILOLOGICAL STUDIES. 17 
 
 other sciences which were once independent of it. Anatomy, 
 chemistry, geography, history, grammar, and criticism have 
 each brought a stone to this great fabric; and it is reasonable 
 that this should be the case. For when the very Kosmos finds 
 in man the most beautiful exemplifications of its own perfect 
 harmony and order, universal science should recognise in the 
 science which treats of man, its object, its aim, and its end. 
 
 11 There is in fact no sure way of tracing the history and 
 migrations of the early inhabitants of the world except by means 
 of their languages ; any other mode of inquiry must rest on the 
 merest conjecture and hypothesis. It may seem strange that any 
 thing so vague and arbitrary as language should survive all other 
 testimonies, and speak with more definiteness, even in its changed 
 and modern state, than all other monuments however grand and 
 durable. Yet so it is ; we have the proof before us every hour. 
 Though we had lost all other history of our country, we should 
 be able to tell from our language, composed as it is of a substra- 
 tum of Low German with deposits of Norman-French and Latin 
 — the terms of war and government pertaining to the former of 
 the superinduced elements,the terms of ecclesiastical and legal use 
 to both of them — that the bulk of our population was Saxon, and 
 that they were overcome and permanently subjected by a body 
 of Norman invaders ; while the Latin element would show us how 
 much that language had been used by the lawyers and church- 
 men. We know too that the inhabitants of Wales, of the High- 
 lands of Scotland, and of the Isle of Man, speak a Celtic dialect; 
 and from the position of these people we should infer that they 
 were the earliest inhabitants of the island, and were driven into 
 the mountains by the Saxon invaders. Even the names of 
 places would tell us as much. When we hear a stream called 
 Wans-beck- water, and know that the three words of which the 
 compound is made up all signify "water," the first being Celtic 
 (as in Wan*s-fo7'd^ A-von)^ the second German (beck=hacJi)^ the 
 last English, we at once recognise three changes of inhabitants 
 to whom the older name successively lost its significance*. It 
 has been the same with other countries also, Persia, for 
 
 * See Varronianus, p. 45. 
 
1 8 THE UTILITY [book I. 
 
 instance, has been under the dominion of Mohammedan con- 
 querors for twelve hundred years, and we find an immense num- 
 ber of Arabic words naturalized in the country ; but the language 
 which forms the basis of the whole, and the general organization 
 and grammar, are as entirely Indo-Germanic as if the country 
 had never had any intermixture of an Arabian population. But 
 above and beyond these results, to which historical records di- 
 rectly contribute, the comparative philologer is enabled, by an 
 examination of the common elements of language, to ascertain 
 the nature of the civilization which men enjoyed, and of the 
 religious belief and worship which represented their spiritual 
 aspirations, at the time, otherwise beyond the reach of human 
 research, when the undivided family of a race was still collected 
 round its hearth and home, and had not yet sent forth its colo- 
 nies to people distant lands. These two applications of philo- 
 logical research , the former of which we have proposed to call 
 "the linguistic records of civilization," and the latter of which is 
 known as "comparative mythology," are the most recent results 
 of philological ethnography, and they promise, if pursued with 
 sobriety and .caution, to lead to discoveries at once certain and 
 important*. 
 
 12 The study of language, therefore, in its wider range 
 may be used as a sure means of ascertaining the stock to which 
 any given nation belonged, and of tracing the changes of popu- 
 lation and government which it has undergone. It is indeed 
 perfectly analogous to Geology; they both present us with a 
 set of deposits in a present state of amalgamation which may 
 however be easily discriminated, and we may, by an allowable 
 chain of reasoning, in either case deduce from the pr^^e^ii the 
 
 * See Encycl. Brit. ed. 8, Vol. xvii. pp. 537, 538. The subject of 
 Comparative Mythology is very ably treated by Prof. Max Miiller in the 
 Oxford Essays for 1856. The linguistic records of civilization are illus- 
 trated in the introduction to Theodore Mommsen's History of Rome^ , 
 which has been translated into English (JThe Earliest Inhabitants of Italy ^ 
 translated from Mommsen's History of Rome, by G. Robertson, Lond. 1858, 
 pp. 9 sqq.), and in A. Kuhn's essay on die Sprachvergleichimg und die 
 Urgeschichte der indogermanischen Volker ( Zeitschrift f. d. Vergleich. 
 Sprachf. iv. pp. 81 sqq.). 
 
CHAP. I.] OF PHILOLOGICAL STUDIES. 19 
 
 former condition, and determine by what causes and in what 
 manner the superposition or amalgamation has taken place. 
 The excellent historian of the Inductive Sciences* would group 
 these and other speculations together in a separate class, con- 
 sidering them ail "as connected by this bond, that they 
 endeavour to ascend to a past state of things, by the aid of the 
 evidence of the present." He would term them palcetiological 
 sciences, and the sanction of his distinguished name will per- 
 haps give currency to this coinage of his private mint. In that 
 case, the classical scholar will wish that he had been induced to 
 select some designation more strictly in accordance with analogy 
 and more plainly expressive of his meaning. As the word 
 archceology is already appropriated to the discussion of those 
 subjects of which the antiquity is only comparative, it would be 
 consistent with the usual distinction between aQxalogsinanalaLOs 
 to give the name of palceology to those sciences which aim at 
 reproducing an absolutely primeval state or condition; or if 
 we were anxious to express that the objects of our science are 
 not only absolutely old, but, in the particular cases, constitute 
 the originals and beginnings of their class, we might indulge 
 in the combination palce-archwo-logyf. But whatever deno- 
 mination we may agree to employ, it is clear that linguistic 
 ethnology is entitled to the most prominent place among its 
 sister sciences. 
 
 13 But the application of philology to the case of ethno- 
 graphy is by no means its only use. Language is the oldest of 
 historical monuments; indeed, it enables us to go back to a 
 period long antecedent to the first beginnings of history, and to 
 
 * Vol. III. p. 481. 
 f The distinction between παλαιός and aqxatos is well given by Reiske, 
 nd Lys. p. 107, 41: "τταλα^όν et antiquum est solummodo rationem habet 
 temporis: άρχαΐον autem est quod ab initio rerum aut reipubliccs cujusdam 
 semper ita fuit actitatum, ut semel antiquo ritu fuit institutum.^' This dis- 
 tinction is well marked in the compounds παλαιόπλοντος , applied to a 
 place, Thucyd. viii. 28, § 3, and άρχαιόπλοντος , applied to a person, 
 -SJsch. Agam. 1013. The. ancients constantly used these synonyms in 
 juxta-position, and a very slight parody of Sophocles would aptly describe 
 a bone of the Dinotherium as όατονν παλαών αρχαίου ποτέ ϋ'ηρός {Trachin. 
 555). 
 
 C2 
 
20 THE UTILITY [book I. 
 
 trace the migrations of a people among whom history has never 
 existed. But philology is also conversant with the interpreta- 
 tion of historical documents. It is philology which has ex- 
 tracted a wonderful array of chronological data from the 
 hieroglyphical monuments of ancient Egypt. It is philology 
 which has recognised the contemporaneous history of Darius in 
 the cuneiform inscriptions ofBehistun. It is philology which 
 promises important revelations from a survey of the long-buried 
 sculptures of Nineveh. And while the science of language deals 
 thus familiarly with the contemporaneous records of ancient 
 history, which modern research has discovered, or which have 
 long been exposed to the careless eyes of an unobserving world, 
 it belongs to the same instrument to test the genuineness and 
 accuracy of traditionary annals which have been embellished 
 and diffused by a more modern and popular literature. Histo- 
 rical criticism is the legitimate offspring of philology. Its 
 functions are not destructive, but rather, in the highest degree, 
 conservative : for its chief aim is to ascertain and establish the 
 granite basis of history which is overlaid by the more recent 
 strata of poetical, philosophical, and religious mythology*. It 
 is true that there are still persons, especially in this country, 
 who plead for the undisturbed enjoyment of an ignorant and 
 childish credulity, and whose acceptance of historic truth is so 
 intimately connected with their adoption of the legendary in- 
 gredients which enter so largely into all ancient records, that, 
 for them, the kernel and the shell are irrevocably identical, 
 and facts and fictions must stand or fall together. It is true 
 also that those whose feeble conscience leans for support on 
 
 * "Denkmaler bilden das Zifferblatt der Geschichte; so lange dieso 
 nicht vorhanden sind, gehort einem Volke nur seine Gegenwart, nicht 
 seine Vergangenheit, es lebt ohne Geschichte. Verliert ein Volk seine 
 Denkmaler, sei es durch eigne Schuld oder die der Verhaltnisse, so wird 
 es auch seine Geschichte nicht retten konnen, sie gerath in Unordnung, 
 wird zur Tradition, und gewinnt im hesten Falle statt des verlorenen 
 rein geschichtlichen ein anderes Prinzip der innern Ordnung, ein poetisch- 
 mytholoyisches wie bei den Griechen, ein philosophisch mi/thologisches wie 
 bei den Indiern, oder ein religioses wie bei den Israeliten, verliert aber 
 stets ihren urspriinglichen zeitgeschichilicheii Werth." Lepsiug, Chronologic 
 der ^gypter^ pp. 1, 2. 
 
CHAP. I.] OF PHILOLOGICAL STUDIES. 21 
 
 some authority supposed to be infallible, as well as those who 
 are interested in the maintenance of such a tribunal of final 
 appeal, are still as heretofore occupied with endeavours to check 
 the inquisitiveness of our philisophical instincts. But the time 
 is long passed and gone, when timid science, wearied with 
 fruitless investigations, could be induced to sell its birthright of 
 expectations for the tasteless mess provided and prepared by an 
 eager and suspicious bigotry. Convinced of the truth of its 
 own conclusions, inductive philosophy needs no support from 
 without; and it has become incumbent upon those who advocate 
 the claims of an assumed authority, to find some common ground 
 on which it may succeed in reconciling its pretensions with the 
 demonstrable truths of science, 
 
 14 Now it is philology alone, acting principally through 
 its chief instrument, historical criticism, which can effectually 
 mediate between tradition and reason; for it is philology alone 
 which finds its materials in the former, and derives its principles 
 from the latter. To the mere votary of abstract science, it 
 matters not what opinions have been held by the most civilized 
 nations of antiquity; he is not interested in attempts to indicate 
 the first beginnings of his own speculations ; satisfied with the 
 possession of truth, he cares little who first discovered it. But 
 the philologer, whose main principle is a recognition of the 
 unity of human nature and of language as the necessary asso- 
 ciate and exponent of reason, is as much concerned with the 
 opinions of primitive Egyptians and Aramaeans as with those of 
 his own contemporaries; and he is predisposed to believe that 
 there must be some portion of divine truth in that which man 
 has in all ages accepted as binding on his faith and conscience. 
 He is anxious therefore that a maximum of ancient history 
 should be established on a scientific basis; and while he ex- 
 amines ancient documents with the rigorous accuracy which phi- 
 losophy demands, he handles the recorded utterances of the past 
 with a veneration which satisfies every enlightened believer. If 
 the Christian religion is to maintain its distinctive position, if it 
 is to enjoy other homage than that which must be always paid 
 to its intrinsic truthfulness, its final triumph over the dangers to 
 which it has been exposed by the ignorance and prejudices of its 
 
22 THE UTILITY [book I. 
 
 teachers , will be secured by the scientific philology which has 
 cleared away the obstructive suburbs, and has thus shown the 
 fortress in its true and naked strength. 
 
 15 That philology, as the method of interpretation , is of 
 essential importance to the protestant divine, is nearly self- 
 evident. According to his principles, no theology can be true 
 which does not rest upon a sound exposition of particular Books. 
 He must therefore regard his system of divinity as merely a 
 branch or application of philological science. That this is so, 
 was the unanimous opinion of those old writers to whom all 
 Protestants appeal as the foremost champions of their cause. 
 Luther thought that trvie theology was merely an application of 
 grammar*; Melancthon maintained that Scripture could not be 
 understood theologically, unless it had been previously under- 
 stood grammatically f ; and Scaliger said with great truth, that 
 ignorance of grammar was the cause of all religious differences t. 
 And not only so in regard to the exegesis from which the pro- 
 testant theologian derives his practical doctrines. In his contro- 
 versies also , he would do well to limit himself to the same safe 
 criterion , and his triumphs would leave no room for a second 
 fight, if, dismissing all perplexing references to the uncertain 
 echoes of ecclesiastical tradition, he were content to employ no 
 weapons save those of Biblical Criticism against adversaries who 
 have raised a fabric of error on their misconception of the dif- 
 ference between πετρος and πέτρα §. Nor is the tradition of 
 
 * "Lutherus — theologiam veram et summam nihil aliud esse quam 
 grammaticam — h. e. Graecarum Hebraicarumque literarum scientiam — 
 putabat." Ernesti Opera Philol. p. 199. 
 
 f "Melancthonis hoc dictum est: Scripturam non posse intelligi 
 theologice, nisi antea sit intellectam grammatice." Ernesti Op. Phil. 
 p. 223. 
 
 J "Utinara essem bonus Grammaticus; sufficit enim ei, qui omnes 
 auctores probe intelligere vult, esse bonum grammaticum. Non aliunde 
 dissidia in religione pendent, quam ab ignoratione grammaticse." Scali- 
 gerana, Prima, p. 86. 
 
 § Matth. xvi. 18. To those who argue for the pre-eminence of the 
 Apostle Peter it is sufficient to refer to ver. 23 in this same chapter, which 
 shows tfcat the address has reference to his words and not to his person; 
 
CHAP I.] OF PHILOLOGICAL STUDIES. 23 
 
 the Italian Church the only form in which an assumed infalli- 
 bility is made a pretext for degrading the intellect of man, and 
 contravening the instincts of conscientious morality. Protestant- 
 ism has submitted to a tyranny even more revolting to our 
 nobler aspirations than that of a Papal priesthood. And it has 
 become a duty for laymen, no less than for professed theolo- 
 gians, to examine certain dogmas which are as prejudicial to 
 true religion and as untenable in themselves as the theory of 
 transubstantiation or the practice of indulgences*. But not to 
 speak of the uses of philological criticism, it may be shown that 
 
 and the writer of the Apocalypse at all events considered the other 
 Apostles equally foundation-stones of the Church {Apoc. xxi. 14; cf. 
 Ephes. ii. 20). The linguistic argument of the Romanists, that in the 
 original languages of Palestine "ista generis differentia quae Graece et 
 Latine est inter Petrum et petram non reperitur, sed uno eodemque 
 nomine, sive Hebraice sive Syriace, Christus dixit: S"^n hy\ αζαζ) '<n 
 ^t^~yh ns-azx ssss, tu es petra, et super hanc petram cedificabo ecclesiam 
 meam'' (Maldonat. ad /.), seems to be supported by the Syriac version, but 
 will not stand the test of philological criticism; for πέτρος is a single 
 stone, but πέτρα is a rock, considered as including many Λετροι, and this 
 opposition is implied by the context; for otherwise he must have said 
 καϊ επΙ αοί. Moreover, in biblical Hebrew we have only the plural 
 13•^£5 "stones" in the signification of a rock, and the analogy of all 
 the Scriptures shows that the rock on which the Church was to be built 
 would be properly designated by i^is and not by i<B"^5. The fact that 
 neither the Peshito nor Mr Cureton's newly discovered Syriac version 
 distinguishes between the "rock" and the "stone," must be added to the 
 arguments against the supposition that the Syriac New Testament has 
 incorporated the lost Aramaic original of St Matthew's Gospel. 
 
 * See a calm and dispassionate pamphlet, entitled Free Theological 
 Inquiry the Duty of the Laity. Lond. 1858. Bunsen has well remarked 
 {JEgypten V, a, p. 52), that "a mere belief in authority, which is nothing 
 but an unthinking, and therefore spiritually feeble acceptance of mis- 
 understood traditions, must produce either a childish superstition, which 
 destroys the childlike truth of that belief, or a denying and deadening 
 incredulity, whether it speaks out honestly or hides itself behind the 
 mask of hypocrisy." He adds (p. 53): "It is almost more necessary to 
 give a strong expression to this in protestant than it is in catholic 
 countries. For the superstition, which the protestants practise with their 
 Bible, in order to conceal their laziness and want of thought, is now 
 employed to obtain the same object which was attained centuries ago in 
 southern Europe — an absolute retrogression in the scientific study of the 
 sacred records. Ignorance goes step for step along with perversity." 
 
24" THE UTILITY [book I. 
 
 linguistic ethnography contributes in no small measure towards 
 establishing the grounds of Revelation. According to the theo- 
 logical system, which lays claim to exclusive orthodoxy, all the 
 truth, or, at least, all the intelligibility of the Christian dispen- 
 sation, depends on the derivation of the human race from one 
 stock. If mankind had not a common origin, there must be 
 branches of our race which have no more share in hereditary 
 corruption or transmitted sin, than the supposed inhabitants of 
 another planet. Now it is by philology alone that we can 
 attempt to demonstrate the primeval unity of man. We are 
 already so far advanced as to be able to divide all the known 
 languages of the world into two main classes; and although 
 we do not yet possess sufficient knowledge of the whole body 
 of languages to be able to say what affinity exists between 
 the two great divisions, approximations have been made to the 
 conclusion that there are certain points in which they osculate; 
 and, judging from the progress of linguistic studies hitherto, we 
 may fairly hope that, as in the case of languages now known to 
 be cognate, we were impressed with the differences long before 
 yve perceived the similarities which are now the most prominent 
 features , so it will be hereafter with all the languages of the 
 world; and investigation will fully confirm what the great Apo- 
 stle proclaimed in the Areopagus, "that God has ordained that 
 from one common parentage all the different tribes of men should 
 spread themselves over the whole face of the earth, having de- 
 termined the particular times of .their successive emigrations, 
 and the boundaries of their respective settlements*." 
 
 16 On the Λvhole then it may be asserted, that philology is 
 essential to a liberal education, and useful as a branch of science ; 
 that it is the great link which connects the past with the present, 
 and the indispensable instrument of theological interpretation 
 
 * Έηοίηοέ TS ^1 hvog αίματος πάν t&vog άν&ρωΛων ν,ατοιγ,ύν έπϊ πάν 
 το πρόαωπον της γης, ορίαας προτεταγ μένους τΐΰαρονς yiccl τάς ορο^εοίας της 
 χατοικίας αυτών {Act Apostol. xvii. 26). Here it is clear that έποίηαε is 
 construed with the accus. and inf. as in Matth. v. 37, Mark vii. 37, and 
 that the έξ ενός αίματος is the main point in the passage. The other 
 words indicate the manner in which ethnical distinctions really arose. 
 
CHAP. I.] OF PHILOLOGICAL STUDIES. 25 
 
 and criticism. We do not, however, maintain that every one 
 should make himself a scientific philologer. In the first place, 
 it is not every one who is qualified by capacity and taste to 
 become a scholar: some particular faculties are needed for the 
 successful study of dead languages , as well as for a profitable 
 employment of one's time in the docks of London andLiverpool, 
 or in the manufactories of Birmingham and Elberfeld : it is also 
 a question of time, for no one can make himself acquainted with 
 the wide range of subjects which philology in its scientific cul- 
 tivation includes , so thoroughly and perfectly as to become a 
 competent teacher of immediate learners, and, by his writings, of 
 the world, without years of patient study and earnest thought; 
 in a word, he must make philology his profession, and if his 
 fellow-men set any value upon his labours he will live by this 
 as he might by any other business. But in a civilized state of 
 society every one ought to learn so much of philology, that is, 
 he should have such an acquaintance with the vehicle of his 
 thoughts, as may enable him to acquire a habit of method in the 
 way of practical teaching. The mistake into which we have 
 fallen in this country does not consist in our making classics and 
 mathematics the basis of our education, nor even in obliging all 
 to attempt what few can attain, but in making boys learn by 
 rote like parrots instead of learning by reflexion like men. We 
 repeat that man is a thinking being, and that his education 
 as such consists in giving him a power and a habit of arranging 
 his thoughts. The learning of Latin and Greek is profitable 
 so far as it is made a lesson on the analogy of language; and 
 we are convinced that the youth of this country would gain 
 more from two years exercise in the rigorous discrimination of 
 the like and unlike in the Greek language alone, than by 
 spending ten years, as many do, in overloading their memory 
 with a mass of crude facts , into the chaos of which the untu- 
 tored intellect cannot penetrate. And, if any one's profession is 
 to be that of a scholar, he will not be the longer in getting to 
 the end of his journey because he has spent more time than 
 some of his fellow-travellers in making himself thoroughly ac- 
 quainted with the route. 
 
 AVe now proceed to inquire, what is the state of scholarship 
 in reference to the objects which give it importance and value. 
 
CHAPTER Π. 
 
 THE HISTORY AND PRESENT STATE OF 
 
 PHILOLOGY. 
 
 I. Classical Philology. 17 Origin and causes of philological studies. 18 Scho- 
 lastic philosophy. Nominalism. 19 Occham's grammatical logic. 20 Refor- 
 mation connected with nominalism. 21 Revival of literature in Italy. 22 In- 
 vention of printing. 23 Progress of learning in Germany. 24 In France. 25 
 In England. 26 In Holland. 27 Progress of scholarship. 28 Bentley. 29 Mo- 
 dern German literature in connexion with scholarship. Heyne. F. A. Wolf 
 and Niebuhr. Reiz. Buttmann and Hermann. K. O. Muller and Welcker. 
 Schleiermacher and Savigny. 30 Merits and defects of German literature. 
 31 Verbal criticism in England. 32 Prospects of classical scholarship in this 
 country. II. Comparative Philology. 33 It did not emanate from the old 
 classical scholarship. 34 The true method of proceeding was first pointed 
 out by Leibnitz. 35 Study of Zend and Sanscrit. 36 Sanscrit scholarship, 
 and its connexion with comparative philology. Rask, Bopp and Grimm. 
 William Humboldt and A. W. Schlegel. 37 Progress of comparative philo- 
 logy in England. F. Rosen. Prichard and Garnett. Prinsep and Ravv- 
 linson. III. Combination of Classical Scholarship with Comparative Philology. 
 
 38 Advantages which the former would derive from such a combination. 
 
 39 An application of the old classical and critical spirit would be not less 
 advantageous for comparative philology. 40 Philological design of the 
 present work ; how far anticipated by Buttmann's Lexilogus and Grammar, 
 
 17 rpO form a proper estimate of the present condition of 
 philological knowledge, we should at least be acquainted 
 with the circumstances in which the study originated, and with 
 the principal changes which it has undergone up to our time. 
 An adequate discussion of this subject would, however, far ex- 
 ceed the limits of an introductory chapter*. All that we pro- 
 pose to attempt in this place is first to point out, as briefly as 
 possible, the more prominent and striking features that have 
 marked the progress of those grammatical and critical studies 
 which constitute our modern scholarship; then to indicate the 
 rise and progress of comparative philology; and lastly to plead 
 for a closer connexion between these two departments of lin- 
 guistic study. 
 
 The importance, which, for the last three hundred years has 
 been attached to philological studies , is at once accounted for 
 and justified by that cessation of all literary exertion for a long 
 
 * For a more complete history of philology the reader is referred to a 
 special article on the subject in the eighth edition of the Encyclopedia 
 Britannicat Vol. xvii. 
 
 / 
 
CHAP. II.] PRESENT STATE OF PHILOLOGY. 27 
 
 period, which ensued upon the overthrow ofthe Western Empire. 
 The rude vigour of barbarism had triumphed over the languish- 
 ing energies of civilization, and it seemed as if the last rays of 
 mental culture had faded away in the prison of Boethius, as if 
 some sudden paralysis had seized upon the general mind of 
 Europe , as if men had drunk of the waters of oblivion, or had 
 lost irrecoverably the key to those treasures of learning, wisdom, 
 and genius, which the old world had amassed for their use. 
 Rapid as was the downfall of literature , it must be recollected 
 that many causes had been gradually operating to produce it, 
 some of which continued their agency to prevent its revival. 
 Among these not the least efficacious was the influence of the 
 Romish Church, the subjection ofthe less cultivated laity to the 
 growing poAver and greater intelligence ofthe clergy, and the 
 prejudices which these last entertained, as well from motives 
 of interest as from scruples of religion, against the learning of 
 the heathen world ; so that, after the crisis had taken place, the 
 Church carefully appropriated to herself the little learning that 
 still struggled for existence, and the papal authority was openly 
 opposed to the diffusion of secular knowledge. Accordingly, 
 when , notwithstanding this prostration of mental culture , the 
 literary spirit revived after nearly a thousand years of darkness 
 or doubtful light, and the mind, awaking like Epimenides from 
 its long slumber, found all things altered but itself, men turned 
 eagerly to the written monuments of the former waking and 
 thinking world; and the grammatical studies necessary for the 
 understanding of these works constituted that philology or scho- 
 larship which has ever since formed the basis of education. 
 
 As the Romish Church throve by the ignorance which it 
 fostered, so it fell by means of the learning which it had always 
 opposed. The causes which produced modern scholarship were 
 identical with those which brought about the Reformation of 
 religion. Of these the most important were the three follow- 
 ing: the overthrow ofthe scholastic realism, and the introduction 
 of rationalism, or a philological spirit, by the Nominalists; the 
 reproduction of the classical authors in Italy , and the revival 
 of the study of Greek, which created a learned class in Europe ; 
 and the invention of printing, which by multiplying books im- 
 parted to the bulk of the laity the effects of the two former 
 
28 THE HISTOEY AND [bOOK I. 
 
 causes, the rationalism and learning of the thinking part of 
 mankind. 
 
 It will beproperto make a few remarks on each of these causes. 
 
 18 The grammatical studies, which the Romans had bor- 
 rowed from the Greeks, and which they had reduced to such an 
 excellent system , were utterly lost in the dark ages. If one 
 needed a proof of this, it would be sufficient to refer to the fact 
 that the languages of those nations which had been most exposed 
 to the influence of Roman literature, and which had for the most 
 part adopted the Latin idiom, degenerated into a barbarous 
 jargon without inflexions or syntax. A certain amount of educa- 
 tion was necessary for the clergy, but, though Grammatica 
 formed a part of the trivium in the seven arts which were then 
 taught, this term did not signify rational grammar, but merely 
 an acquaintance with the Latin of the schools. The nature, 
 however, of some of the Romish doctrines necessitated a mixture 
 of metaphysics with theology. Li this mixture originated the 
 scholastic philosophy, which was simply an attempt to prove, 
 by what they knew of Aristotle's logic, the necessary union of 
 reason and orthodoxy. It was only by such an instrument as 
 the quasi-realism of their Aristotle , that they could establish 
 those points of faith which constituted the difficulty of the 
 Romish creed. It was necessary that the mere abstractions 
 of common language should be considered as objective realities, 
 for it was about these abstractions alone that they argued ; and, 
 as there were some clear-headed men among them who could 
 not accept this position, there arose a dispute in the schools, 
 of which the real point under discussion was, whether the scho- 
 lastic philosophy had any authentic basis or not. This dispute is 
 commonly known as the question about Universals, or the con- 
 troversy between the Realists and the Nominalists; the former 
 of whom considered universal ideas as pre-existent in the mind 
 of God and man (^ante rem)^ and the latter as simple abstrac- 
 tions of the understanding from the objects of the senses (post 
 remy Now as it was about the universals alone that the 
 scholastics reasoned,it followed, that, if the universals were mere 
 words, they could not pretend to establish anything by their 
 argumentation , and consequently the whole system would be 
 
CHAP. Π.] PBESENT STATE OF PHILOLOGY. 29 
 
 overthrown. We can, therefore, easily understand why the 
 nominalists were so violently persecuted, and why they wxre 
 considered as little better than heretics ; for, if orthodoxy rested 
 upon scholastic realism, those who undermined the one contri- 
 buted not a little to the downfall of the other. Roscelinus, 
 the first opponent of the realists, went into the opposite extreme 
 of ultra-nominalism*, and as he was confuted without much dif- 
 ficulty by Abelard and the other conceptualists, his efforts did 
 not avail much to the overthrow of the system. It was reserved 
 for our countryman, William of Occham, to effect this. Without 
 running into any paradoxical absurdities,he showed in a straight- 
 forward manner that words are instruments of reasoning, not 
 objects of science, and set up the grounds of rational grammar 
 against realism , which had overlooked the fact that logic has 
 nothing to do with the particular significations of words, but only 
 with their methodical arrangement. 
 
 1 9 The Nominalism of Occham, however, was not merely a 
 reassertion of grammar; it was an overthrow of that w^orship of 
 words which was so important a part of the idolatry of the time. 
 As it is one of the objects of this work to maintain the opinions 
 which Occham advocated, both against the symbolical realism, 
 which is still prevalent, and against the ultra-nominalism which 
 now and then makes its appearance, we shall offer no apology 
 for giving a statement of views so much in accordance with our 
 own as his are; at the same time it is but due to the sturdy 
 Franciscan, who is little thought about by the thousands who 
 are reaping the benefit of his labours, to give as nearly as pos- 
 sible in his own words the important principles which he so 
 opportunely revived, and by which he earned the titles bestowed 
 upon him by his editor Marcus de Benevento — logicorum acutis- 
 simus, sacrce scholce invictissimorum Nominalium inceptor, in 
 omni disciplinarum genere Doctor plusquam suhtilis. It will 
 be sufficient to give the statements in his Summa totius logicce 
 
 * We apply this term to the doctrines of those philosophers who, like 
 Cratylus, Hobbes, and Home Tooke, not only reduced universale to mere 
 names, 
 has called Hobbes plusquam nominalis. 
 
30 THE HISTOEY AND [bOOK I. 
 
 ( Venet. 1522), one of the best of his works, and indeed as sound 
 a treatise on the philosophy of syntax as any which has come 
 to our knowledge. Arguments and syllogisms , says Occham, 
 are made up of propositions, and these again of terms. Of 
 terms there are three kinds, (1) that which is conceived, (2) that 
 which is spoken, (3) that which is written: the last two are 
 parts of a proposition meant to be heard by the bodily ear or 
 seen by the bodily eye ; the first is merely an intention of af- 
 fection of the mind signifying or consignifying something as a 
 part of a mental proposition*. AVe say that words are signs 
 subordinated to the conceptions or intentions of the mind , not 
 because, in the strict acceptation of the term "sign," words 
 primarily and properly signify the actual conceptions, but be- 
 cause words are imposed to signify those very things which 
 are signified by the conceptions of the mind: for the word sig- 
 nifies secondarily what is primarily signified by the conception, 
 and is instituted to denote something implied by a conception, 
 so that if the conception changes its meaning , the word would 
 also change its meaning unless it were altered to suit the change 
 in the conception. The intention, conception, or affection of the 
 mind, is defined to be something in the mind naturally signify- 
 ing something^ . As writing is the secondary sign of speech, 
 so speech in the secondary sign of the intention or conception, 
 which is the primary sign of the thing signified as forming a 
 part of a mental proposition, which belongs to no particular 
 idiom^. An intention is either primary or secondary: primary, 
 when it is the mental representative of the thing signified; 
 secondary, when it is a sign of such primary intentions : genus, 
 species , &c. are secondary intentions §. With regard to uni- 
 versals, under which are included the five predicables, genus, 
 
 * Cap. I. fol. 2. Terminus conceptus est intentio sen passio animce all- 
 quid naturaliter significans vel consignificans , nata esse pars propositionis 
 mentalis. 
 
 f Cap. XII. fol. 6. Intentio animce vocatar quoddam ens in anima natum 
 significare aliquid. 
 
 "^ Propositio mentalis quce nullius idiomatis est. 
 
 § Stride vacatur intentio prima nomen mentale natum pro suo significatu 
 supponere. Intentio autem secunda est ilia quce est signum talium intentio- 
 num primarum: cujusmodi sunt talis inteniiones: genus, species i!fc. 
 
 m 
 
CHAP. IL] PBESENT STATE OF PHILOLOGY. 31 
 
 species, difference, property and accident, he adopts Avicenna's 
 definition with the following explanation* : — an universal is a 
 single intention of the mind made to be predicated of many, not 
 for itself but for the things themselves, and because it is thus 
 predicabje of many it is called universal; but it is called singular 
 because it is one form really existing in the mind. Of the argu- 
 ments which Occham advances to prove that an universal is not 
 a substance, it will be sufficient to mention one. This intention 
 which we call an universal is a perfectly arbitrary sign, for this 
 is the only thing that is predicable of many. Substance, how- 
 ever, cannot be thus predicated, otherwise a proposition would 
 be composed of particular substances. Nowapropositionis either 
 mental, or spoken, or written. But these are not particular sub- 
 stances. Therefore no proposition can be composed of substances. 
 They are however composed of universals. Therefore universale 
 are not substances. Occham says the same of the categories or 
 predicamentsf, respecting which, in his important treatise de 
 Sacramento altaris, he adopts the grammatical views of Joannes 
 Damascenus. The ten predicaments, he says J, are the most 
 general signs, under which are included all mere words. For 
 every word which can be placed at the end of a proposition, or 
 employed in answering a question, may be arranged under some 
 one predicament; for example, all words which make answer to 
 the question "How much?" are placed in the predicament of, 
 quantity. And the words thus placed in the predicaments are 
 not only nouns, but also verbs, and other j)arts of speech : and 
 even phrases, such as prepositions with their cases. So that 
 predicaments are only certain predicables and uncombined signs 
 of things, out of which true or false combinations may be made. 
 
 * Cap. XIV. fol. 6, 2. Vult dicere quod universale est intentio singidaris 
 ipsius animce nata prcedicare de pluribus non pro se sed pro ipsis rebus : ita 
 quod per hoc quod ipsa nata est prcedicari de pluribus, non pro se sed pro 
 illis pluribus, ilia dicitur vniversalia: propter hoc autem quod est una forma 
 existens realiter in intellectu, dicitur singulare, 
 
 t Cap. xLi. fol. 13. 
 
 J Tractatus Venerabilis Inceptoris Guilgelmi Ocham de Sacramento al- 
 taris, Cap. xxxv. (This book is very scarce: our copy is a little black- 
 letter duodecimo, vsrithout a date, but stated to be impressus Parisiis per 
 Petrum Levet, impressorem.) 
 
32 THE HISTORY AND [BOOK I. 
 
 A difference of inflexion may make an irreconcileable difference 
 of signification between two predicaments; for example, angelus 
 cannot be angeli^ nor e converso. He adds, that such predi- 
 caments are not merely arbitrary signs, but rather meanings 
 or intentions of the mind, because they are signs naturally 
 denoting objects; and as words may be distinct, notwithstand- 
 ing the identity of the things signified, so meanings or intentions 
 may be distinct, although the objects signified are identical. 
 Thus the words substance, quality, and quantity, are not syno- 
 nymous because they can be predicated of the same object. 
 
 20 It is obvious, from the way in which Occham speaks 
 throughout his logic, that his object was to draw a strongly- 
 marked line of distinction between the method of language, as 
 the instrument of deduction, and the science.or truth of things. 
 By doing so he not only contributed to overthrow the scholastic 
 theology, but also laid the foundation for that system of inter- 
 pretation which became philology in the hands of the scholar: 
 and therefore it was not without reason that Luther, who so well 
 estimated the importance of grammatical studies to the ministers 
 of his Reformation, paid so much attention to Occham while he 
 despised and neglected the other schoolmen*. Indeed it appears 
 from Luther's early works that he had embraced very heartily 
 the Occhamistic nominalismf . 
 
 But the adoption of Occhamistic theology by the Reformers, 
 whatever may have been its effect on the sacramental doctrines 
 of Luther, or on the fate of scholastic dogmatism in general, is 
 important to us principally as furnishing collateral testimony to 
 the increasing influence at this period of that critical and ration- 
 alistic spirit, which is the essential characteristic of modern 
 
 * "Diu multumque legit scripta Occam; hujus acumen anteferebat 
 Thomae et Scoto." Melancthon, Historia de vita et actis Liitheri. Vite- 
 berg. 1545. fol. 5. 
 
 f Laurence {Bampton Lectures^ Serm. in. note 6) after quoting Me- 
 lancthon, Op. I. p. 414, Loci Theol. p. 113, says, "William of Occam here 
 alluded to (an English Scholastic of great reputation) had been a pecu- 
 liar favourite of Luther and Melancthon; the former styling him earns 
 magister mens, and the latter delicice quondam nostrw.^' Ranke {Hist, of 
 Reform, i. p. 310, Engl. Tr.) says, "Luther and Melancthon are the off- 
 spring of nominalism." 
 
CHAP. II.] PBESENT STATE OE PHILOLOGY. 33 
 
 times*. To deprive words of their mysterious attributes, to 
 rescue the mind from the thraldom of metaphors, and to subject 
 every sentence to the fearless searching of the critical faculty, is 
 the greatest triumph of iconoclastic reason, which has yet to cele- 
 brate the last of a long series of victories. This was the result 
 in part effected by the early success of the nominalistic school. 
 
 21 The overthrow of the scholastic philosophy, however, 
 and the proper cultivation of grammar, though good in them- 
 selves, would have had but little effect in dispelling the darkness 
 of the middle ages, had it not been for the classical enthusiasm 
 of the poet Petrarch, and the subsequent exertions of Poggio 
 Bracciolini, who laboured incessantly, and to a certain extent 
 successfully, in saving and bringing to light the remains of the 
 great Latin writers. The increased study of pure Latinity, pro- 
 duced by the gradual diffusion of these books, naturally created 
 a desire to become acquainted with the sisterliterature of ancient 
 Greece, which the overthrow of the Eastern Empire by the, 
 Turks accidentally favoured ; because those of the Greeks who 
 hadretained a knowledge of their ancestors'language hastened to 
 Italy, where the patrons of learning offered them protection and 
 reward. For some time, however, the number of Greek scholars 
 was very limited, and nothing was done for scholarship beyond 
 the translation of Greek authors into Latin, which, as the per- 
 formers were mostly modern Greeks who were indifferently 
 acquainted with Latin, and had but a traditional knowledge of 
 the language of their ancestors, have little to recommend them 
 beyond the merit of the attempt. But although the influence of 
 the new learning was somewhat limited, still a beginning was 
 made, a learned class was formed, comparisons were inevitably 
 suggested between the enlightened views of the old world and 
 the ignorance of the day, and the first seeds were sown of that 
 
 * Even the late Dr. Mill, the uncompromising advocate of "old paths," 
 was willing to admit that the perfection of critical skill "is undoubtedly 
 one of the real boasts of the later generations over the most intellectual 
 and able of the ancient world" {Four Sermons, 1849, p. 114). AVith 
 this admission, we need not wonder at the ill success of any attempts to 
 restore the authority of a dry and scholastic dogmatism in our chief 
 seat of learning and science. 
 
 D 
 
34 THE HISTOBY AND [bOOK I. 
 
 freedom of inquiry which shook the papal throne and changed 
 the tendencies of the world. 
 
 22 The third and greatest cause of the revival of letters 
 was the invention of the art of printing, which, by the multipli- 
 cation of books, spread the influences of revived learning over 
 the whole of Europe, and, it is not too much to say, produced 
 immediately and at one birth the Reformation and Philology. 
 This invention is due to Germany or to the Low Countries ; and 
 though it must not be overlooked that its immediate effects and 
 its application to the diffusion of classical learning were first felt 
 in Italy, yet the origin of philology or sound scholarship is to be 
 sought on this side of the Alps; its beginner was JohnReuchlin 
 (Capnio) a German (born in 1455, died in 1522); the two men 
 who brought it forward most prominently and tangibly were 
 William Bude(Budaeus), aFrenchman, and Desiderius Erasmus, 
 a Dutchman (both born in the year 1497); and while the com- 
 bined influences of scholarship and printing, as far as they con- 
 duced to the overthrow of the papal system, were developed in 
 Germany by Martin Luther (born in 1483, died in 1546), and 
 by Philip Schwarzerde (Melancthon) , a kinsman of Reuchlin 
 (born in 1497, died in 1560), the more profound and speculative 
 studies of Reuchlin in the later Jewish and Greek philosophy, 
 in the occult sciences, and in the Cabalistic art, were prosecuted 
 with great ability and learning, but with a growing consciousness 
 of their futility, by Henry Cornelius Agrippa von Nettesheim 
 (born in 1486, died in 1535), who, like Budaeus and Erasmus, 
 lived and died in communion with the Church, but, unlike them, 
 was persecuted and reviled as a magician and conjurer*. Under 
 the more immediate domination of the papal chair, and shrinking 
 from German sway with all the aversion produced by their his- 
 
 * The character of this eminent man has been vindicated by Mr. 
 Morley in his spirited and interesting book, The Life of Henry Cornelnis 
 Agrippa von Nettesheim^ Doctor and Knight^ commonly known as a Magi- 
 cian, London, 1856. It is to be regretted that the life of a scholar, 
 written by a well-informed man, and derived from Latin authorities, 
 should be deformed by such indications of laxity or carelessness (e. g. 
 we have litera humaniora! in Vol. i. p. 64, and Gregory of Nazianzen, 
 in Vol. I. p. 165). 
 
Β 
 
 CHAP. II.] PRESENT STATE OF PHILOLOGY. 35 
 
 torical recollections, the Italians were not likely to be affected by 
 that free spirit of inquiry, which had been for some time at work 
 among the Germans, and which seems to be still their natural 
 characteristic. It was, however, to this free German spirit, and 
 the materials furnished to it by the revival of letters in Italy and 
 the invention of printing among themselves, that we are to attri- 
 bute the great German Reformation. Luther himself has said 
 as much*, and we see that, in the expression of their alarm at 
 the probable effects of printing, the catholic hierarchy of the time 
 had reference to Germany alone f. 
 
 23 Intimate, however, as was the connexion between the 
 rise of philology and the Reformation, it cannot be denied that 
 the country which gave birth to the latter was for many years 
 afterAvards far behind other countries of Europe in its cultivation 
 of the former. Nor shall we wonder at this, when we recollect 
 that all intelligent Germans who embraced the principles of the 
 Reformation were absorbed in theological studies, and that in 
 consequence of the religious differences which were produced by 
 this engrossing change, Germany became one great battle-field, 
 and was prevented from attaining to a full literary development 
 till after the results of the thirty years' war had restored peace 
 and tranquillity to the Protestants. 
 
 24 The connexion of France with Italy, in consequence of 
 the ambitious projects of Charles VIII. and his successors Louis 
 and Francis, made the former country an important instrument 
 in the diffusion of the learning revived in the latter. The first 
 approach to really accurate scholarship is due to Budaeus (Bude) 
 and the family of the Stephenses (Estienne), who brought the 
 knowledge of Greek and Latin to a point of perfection which has 
 induced many to believe that there was little more to be done 
 by the modern scholar than to study the commentaries of the 
 
 * In his 239th letter (Erster Theil, p. 494, deWette), where he talks of 
 Germanorum ferocia ingenia, quce, nisi capta sint Scripturis et Ratione, non 
 est vel multis Papis irritare tutum; proesertim hoc tempore^ ubi in Germariia 
 regnant literce et linguce et sapere incipiunt Laid. 
 
 t See the Proclamation of the Archbishop of Mainz in 1486, quoted 
 by Hallam, Literature of Europe^ Vol, i. p. 348. 
 
 D 2 
 
36 THE HISTOBY AND [bOOK I. 
 
 former* and reprint the Thesaurus of Henry Stephens. Their 
 successors, Casaubon and Salmasius, exhibit in their works an 
 extent of reading and an accuracy of scholarship which few of 
 our modern philologists can rival, and Joseph Justus Scaliger, 
 who, though of Italian extraction, must be reckoned among the 
 glories of French literature, may fairly be considered as still 
 entitled to precedence above all scholars, Bentley, perhaps, ex- 
 cepted. Even Muretus, who was the most learned man in Italy 
 at this time, was a French emigrant. 
 
 25 Our own country occupies a very distinguished place 
 among the first nurseries of philology. Erasmus taught Greek 
 at the University of Cambridge as early as 1510, and many of 
 our native scholars were sought as instructors in Germany and 
 elsewhere on the continent "j*. Roger Ascham, whose "School- 
 master" was the earlist vindication of philological learning that 
 appeared in a vernacular form, says in that workj that "good 
 Mr. Redman §, and those two worthy stars of that University, 
 Mr. Cheke|| and Mr. Smith H with their scholars, had brought 
 Aristotle, Plato, TuUy, and Demosthenes to flourish as notably 
 in Cambridge as ever they did in Greece and in Italy: and for 
 the doctrine of those four, the four pillars of learning, Cambridge 
 then gave place to no University, neither in France, Spain, Ger- 
 many, nor Italy." And long before any continental scholar had 
 undertaken to explain in his mother tongue the classical lan- 
 guage of ancient Rome, Sir Thomas Elyot of Jesus College, 
 
 * Porson intended at one time to re-edit the Commentaries of Budaeus, 
 with such additions and improvements as would make it serve as a manual 
 for the Greek student. 
 
 f The fragmentary translation of the New Testament by Sir John 
 Cheke, which has been published from the MS. {Cambridge, 1843), is a 
 curious specimen of the accurate philology of the time. The controversy 
 between the same writer and Stephen Gardiner on the pronunciation of 
 Greek is also well worthy of notice (Basilece, 1555). 
 
 :}: Works, ed. 1815, p. 304. See also his letter to Joannes Sturmius, 
 Familiares Epistolce , edit. Edvardi Grant, 1576, especially pp. 17b, sqq. 
 
 § Dr. John Redman or Redmayne, archdeacon of Taunton, and one 
 
 the compilers of the Liturgy. 
 II Sir John Cheke, Tutor and Secretary of State to Edward VI. 
 
 ^ Sir Thomas Smith, Secretary of State to Queen Elizabeth. 
 
CHAP. Π.] PRESENT STATE OF PHILOLOGY. 37 
 
 Cambridge, had drawn up an excellent dictionary of Latin and 
 English, and an Oxford school-master, Thomas Cooper, after- 
 wards Bishop of Lincoln, so improved this work in successive 
 editions, that it became equal in value and authority to the 
 standard Lexicon of Robert Stephens, which was not similarly 
 adapted to the use of French students*. 
 
 26 The noble demand of the citizens of Leyden, that an 
 University might be established there as a reward for their 
 heroic defence in 1574, led to the institution of many similar 
 establishments in the country, and to the foundation of a school 
 of philology, which in accurate knowledge and extensive read- 
 ing soon left the rest of Europe far behind. Here it was that 
 Scaliger, the great epochal scholar of the latter half of the 16th 
 century, not only "laboured at restoring the work of Eusebius 
 with the confident strength of genius and with boundless learn- 
 ing•)•," and so laid the foundations of scientific chronology, but 
 threw a new light on every field of philological research, and 
 without dictionary or grammar mastered the minutest details of 
 classical and oriental learning, leaving to his successors to gather 
 up into sheaves the harvest, which he had reaped with the im- 
 patience of gigantic vigour, but had left lying on the ground J. 
 And although he was of Portuguese extraction, Benedict Spi- 
 noza was born at Amsterdam, and it was in Holland that he 
 taught the principles of a philosophy which has influenced all 
 subsequent speculations in metaphysics and theology §. 
 
 27 Whatever may be the merits of these different countries, 
 the progress of scholarship was every where the same. Atfirstthe 
 object of the learned was to publish as many of the Greek and 
 
 * See Mr. Mayor's paper in the Journal of Philology^ March 1857, 
 pp. 1^ sqq. 
 
 f Niebuhr, Hist, of Rome, Vol. i. p. 243•, see his remarks in note 
 660•, and compare his nachyelass. Schriften nicht pkilolog. Inhalts, p. 141. 
 
 :|: See the list of Scaliger's works in the Biography of this great 
 scholar by Jacob Bernays, Berlin, 1855, pp. 269 — 316. 
 
 § His translator, Emile Saisset, has pointed out his importance, and 
 Amand Saintes has called him "le fondateur de la philosophie moderne." 
 
38 THE HISTORY AND [bOOK I. 
 
 Latin authors as they could collect, to supply the former with 
 Latin versions, and both with such commentaries as might help 
 the less perfect scholars to an understanding of the grammatical 
 construction of the text, and the various allusions contained in it. 
 To further this object collections of miscellanies or varice lee- 
 tiones^ as they were called, were from time to time published by 
 different scholars. As grammatical knowledge advanced, the 
 editors were induced to remark the corruptions of the texts, pro- 
 duced by the ignorance of those who had copied the manuscripts 
 in the dark ages; and from this verbal criticism arose. But in 
 all that was done for nearly a hundred years after the death of 
 Scaliger, we perceive a want of method combined with much 
 individual industry, nor do we discern any marks of command- 
 ing genius in the many who directed their attention to the 
 elucidation and correction of ancient authors : they passed their 
 time in polishing the ancient authors and abusing one another; 
 and, like the rhapsodes of old, while they corrected their authors, 
 they needed much emendation themselves *. 
 
 28 The real beginning of a more enlightened scholarship is 
 due to Richard Bentley , whose unrivalled ingenuity and boundless 
 learning enabled him not merely to correct the words of the 
 ancient writers, but also to apply his intelligent criticism to the 
 literary history and antiquities of Greece and Rome. His actual 
 performances, and the example which he set to those who came 
 after him, cannot be too highly estimated. Indeed, he may be 
 regarded as the founder of historical philologyf . In verbal criti- 
 cism he found most able successors among his own countrymen : 
 D^wes, Markland, Toup, Tyrwhitt, and Person, who followed 
 in his steps, have advanced this department of scholarship to 
 its utmost limits. Meanwhile the scholars of Holland were not 
 idle, and to them we owe a most accurate examination of the 
 old Lexicographers and a collection of every thing that could 
 contribute to the illustration of what they advance: besides, the 
 
 * Τόνξ μεν γάρ ρα^ψωδονς oidcc τάμεν ϊτιη άκριβοϋντας, αντονς δε ηάνν 
 Ύ^λι^ίονζ οντάς (Xen. Mem. ιν. 2, § 10). 
 
 Ι "Historical philology — the discovery of Bentley, and the heritage 
 and glory of German scholars." Bunsen, Egypt, i. note 22. 
 
CHAP. II.] PEE SENT STATE OP PHILOLOGY. 39 
 
 first step in the systematic etymology of the Greek language 
 was taken in that country, and, though the principles on which 
 the undertaking was based were not sufficiently well founded, 
 the attempt was not unproductive of benefit. 
 
 29 The Germans did not contribute much to the advance- 
 ment of philology before the end of the seven years' war. Their 
 general literary condition was very low, as may be inferred 
 from the fact that the few great authors whom they had before 
 this period, for instance Leibnitz, usually wrote either in Latin 
 or French. When, however, they were released from the dis- 
 tractions of war and turned themselves in earnest to read and 
 write, they showed that the active spirit which had made Ger- 
 many the scene of the Reformation, was still alive among them, 
 and they advanced with such giant steps that within a century 
 they had placed themselves at the head of literary and learned 
 Europe. About the middle of the 18th century Gesner and 
 Ernesti came forward as prominent philologists in Germany, but 
 their style of scholarship had little in common with that of their 
 countrymen at the present day, being rather formed on the 
 model of the Dutch school ; indeed, the two most learned Ger- 
 mans of this time, Ruhnken and Drakenborch, confined the 
 sphere of their usefulness to the Universities of Holland. The 
 beginning of German, or, to use a more comprehensive but sy- 
 nonymous term, of modern scholarship, is to be referred to Les- 
 sing*, who created German literature by casting off the trammels 
 
 * See Stahr's G. E. Lessing, sein Leben und seine Werke, Berlin, 
 1858. A reviewer of this book in a weekly Journal has well remarked: 
 "As all roads led to Rome, so must the historian of nearly every de- 
 partment of intellectual labour in Germany look back to Lessing. Who 
 converted the Germans from an idolatrous worship of French classicism, 
 and set them about the formation of a national literature? — Lessing. 
 Who started that enthusiastic admiration for Shakspeare, that has gone 
 on increasing till the present day? — Lessing. Who commenced that 
 philosophy of art, that is such a peculiar emanation of the German 
 mind? — Lessing. Who laid the foundation of that earnest tendency to 
 free inquiry in matters of religion that is equally remote from old- 
 fashioned orthodoxy and Voltairian levity? — Lessing. Who wrote 
 the only German comedy that is considered classical? — Lessing. Who 
 wrote the only poem in Germany that anybody cares to read for the 
 
40 THE HISTOEY AND [bOOK I. 
 
 of the French school, andto Winckelmann, who laid the founda- 
 tions of the archieology of art, a most important application of 
 philology in the wider sense of the word. The successive ap- 
 pearance of the literary giants Wi eland. Herder, Schiller, Gothe 
 and Eichter, who were all to a certain extent contemporaries, 
 and the searching philosophy of Kant, which arose at the same 
 time, furnished the active intellect of Germany with abundant 
 materials for speculation; and the people became at once re- 
 markable for bold theories and uncompromising inquisitiveness. 
 This was not long without its effect upon scholarship, which, as 
 we have said, existed among them as an offshoot of the Dutch 
 school, and German philology properly so called arose with 
 Heyne. The difference between him and his predecessors con- 
 sisted mainly in this, that he did not limit his investigations to 
 the narrower field of classical criticism, but combined with it 
 all the newly applicable resources of the archaeology of art, of the 
 principles of taste, and of literature in general. In fact^ the line 
 of demarkation between literature and learning no longer existed 
 for the Germans. We see this in the acknowledged influence of 
 the translations of Yoss, in the classical spirit which pervades all 
 the great German writings^ and indeed even in the adoption of 
 classical metres for vernacular poems. The school which Heyne 
 founded was combined with a revival of the Bentleian spirit in 
 the person of F. A. Wolf^ a man of the most purely literary 
 genius that ever marched with the heavy baggage of book- 
 learning. Whether he owed any thing to Vico or not, he must 
 certainly be considered as the literary descendant of Bentley; 
 and his treatise on Homer produced an effect upon scholarship 
 analogous to that of the Dissertation on Phalaris. It is no dis- 
 paragement to the undeniable genius and prodigious learning of 
 Niebuhr to say, that the History of Rome, the master-work of 
 historical criticism which will ever mark our age as distin- 
 guished in the annals of literature, is due in a great measure to 
 
 sake of its moral import? — Lessing." It should not be forgotten that 
 it is to Lessing that we owe the term bibliolatry, by Λvhich we stigmatize 
 the degrading hypothesis of an infallible literature. For the sense in 
 which he used the term, see his collected works (Berlin, 1839), Vol. xi. 
 pp. 537, 538 sqq. 
 
CHAP. II.] PEESENT STATE OF PHILOLOGY. 41 
 
 the example of Wolf; and we may join the illustrious historian 
 of Rome in hailing him as the Hero andEponymus of the race of 
 German philologers*. But with all these wonderful extensions 
 of the domain of philology the Germans were not inattentive to 
 the details. F. W.Reiz, whose treatise on the Greek and Latin 
 verb appeared inl766f,laid the foundations of a rational method 
 of grammatical criticism; his pupil Godfrey Hermann at the 
 beginning of the present century published a little essay |, Avhich 
 placed the study of Greek grammar in particular on an entirely 
 new footing; Philip Buttmann, the first edition of whose Greek 
 Grammar, published in 1782, was not much in advance of its 
 forerunners, lived to make it a complete store-house of the facts and 
 forms of the language, and to prove himself the most ingenious 
 etymologer of the day; and the Dictionnaries of Schneider and 
 Passow constitute an epoch in the department of Lexicography. 
 Lobeck has proved himself a worthy successor of Hemsterhuis 
 and Ruhnken. The editions of classical authors published by 
 Spalding, Heindorf, Bockh, Dissen and others, are a proof that 
 the more enlarged views of the age have not been without their 
 influence on the old system of note writing; and the overthrow 
 which Hermann experienced some years since, in his opposition 
 to the more extended views of Miiller and Welcker, is a 
 proof that philology has, in Germany at least, taken its place 
 among the subjects of general literature. In addition to the 
 causes enumerated above, a great and immediate influence is 
 attributed § to the new system of interpretation introduced by 
 Schleiermacher, a most distinguished metaphysician and divine, 
 who commenced at the beginning of this century a complete 
 translation of Plato arranged according to a comprehensive view 
 which he had taken of the general connexion of thought and 
 mutual dependence of the separate dialogues: perhaps few 
 German books are a better example of the influence of the 
 
 * Rheinisches Museum for 1827, p. 2hl {Phil. Mus. i. p. 176). 
 
 f De temporibus et modis verbi Grceci et Latini, Lips. 1766. 
 
 J De emendanda raiione Grcecoe Gramm. Pars i. Lips. 1801. 
 § See Dissen in Bockh's Pindar, Vol.iii. p. 7; and Bockh himself says 
 (Abhandl. Berl. Akad. 1822, 3, p. 264), "und ich will mich auch nicht 
 riihmen es erfunden zu haben, da ich es von Schleiermacher gelernt 
 habe." 
 
42 THE HISTORY AND [bOOK I. 
 
 general literary spirit on philology*. In other fields equally 
 important advances were made. The connexion of philology 
 with the study of the civil law produced a most striking effect 
 on the latter, and a fortunate discovery of the fragments of 
 Gains has given an entirely new complexion to the history of 
 Jurisprudence: in this department Savigny stands foremost as 
 the greatest of lawyers and one of the greatest of philologers. 
 But we should far exceed our limits were we to particularize 
 the different German authors who have written first-rate books 
 on mythology, geography, or art. Suffice it to say, that by the 
 exertions of the Germans alone, philology has made more pro- 
 gress in the last fifty years than in the preceding two hundred. 
 
 30 It was not, however, to be expected that such a sudden 
 advance should be made without some corresponding disadvan- 
 tages. And here we may be permitted to make a few general 
 remarks on the merits and defects of German literature. The 
 merits of the German writers may be inferred from what we 
 have already said. They have a large share of originality; great 
 honesty and consistency of purpose in the maintenance of their 
 own opinions; indefatigable industry; and an entire freedom 
 from prejudices and one-sided views. Their defects, so far as 
 they are constitutional, arise from the excess of those qualities 
 which constitute their chief merit : they are liable to give way 
 too much to their imaginative enthusiasm; like the student An- 
 selmus, in Hoffmann's tale f , they are fascinated by the dark- 
 blue eyes in the elder -bush, and the crystal bells are ever 
 sounding in their ears. The circumstances under which their 
 literature so rapidly created itself have also produced their evils : 
 every young man who has any pretensions to learning or ability, 
 feels himself called upon to write something novel; the result of 
 which is, that without sufficient sobriety of mind to keep them 
 from error, their disposition leads them continually to make rash 
 attempts upon subjects which ought never to be treated without 
 the greatest care and circumspection, so that they have given 
 much scandal to people of more timid minds, and created preju- 
 
 * See the Journal of Education^ Vol. ix. pp. 118 foil, 
 f Der goldne Topf. 
 
CHAP. Π.] PRESENT STATE OF PHILOLOGY. 43 
 
 dices against that freedom of investigation^, which, when soberly 
 exercised, is at once the most imperative duty and the highest 
 privilege of civilized and educated men. There is perhaps no 
 literature in the world which furnishes so decided, so strongly 
 marked a contrast as the German does between the first- and 
 second-rate authors : which, while it presents to us more of the 
 true and the valuable, more that is kindled into sunlike splen- 
 dour by the fire of genius, more that is arrayed in the secure 
 panoply of learning, more that is dressed in the gay but not 
 inelegant trappings of a rich and luxurious fancy, can yet on the 
 other hand display so much frigid pomposity, so much flippant 
 ignorance, and such audacious dulness. True it is we have 
 but few of the first class among our writers : but then how very 
 few we have of the second. The same principle of cautious 
 reserve or of a persevering pursuit after the profitable, which 
 has doomed to the desert air many a glowing thought and many 
 a golden truth, has also saved us from sundry proofs of the 
 danger of trying to be ingenious, and has deterred many a 
 literary Curtius from leaping into the gulph, ever open in the 
 forum to swallow up those whose destiny it is to write and be 
 forgotten. Another feature in the German literary character is 
 the air of consciousness and pretence with which the great bulk 
 of writers bring forth their works. The mark of real originality 
 is, that the author is before his age. The great writer, through 
 his own modesty, is generally the last to discover this, or, per- 
 haps, through the tardy appreciation of his contemporaries he is 
 left in ignorance of the noble service he has done. Like Spen- 
 ser's Chrysogone, he has unawares conceived and borne children 
 of the sun, and they lie unheeded by the parent's side in some 
 rude wilderness untrodden by men, till the goddess of Beauty 
 claims them for her own and rears them in her own fair garden, 
 that they may hereafter delight and instruct mankind*. The 
 Germans, on the contrary, make an exhibition of their genius, 
 and do not scruple by themselves or their friends to point out, 
 to those who might fail to discover it, the value of their Λvares. 
 There is perhaps no other country in which the intercourse of 
 literary men is so much a commerce of compliments, and where 
 
 '^ Faerie Queen, Book iii. Cant. 6. 
 
44 THE HISTOEY AND [bOOK I. 
 
 friendship is so likely to cool whenever it is not fed with the 
 fuel of commendation*. An excessive love of applause, when 
 gratified, is apt to generate a self-complacent indifference to the 
 merits of foreign labourers in the same field; and when we con- 
 sider the almost extra\^agant commendation which is bestowed 
 upon German learning in this country, it may seem the less sur- 
 prising that the Professors of Berlin and Leipzig should never 
 take the trouble to inquire whether they are not bestowing their 
 industry on some investigation to which the acuteness of En- 
 glish scholars has already been successfully applied f. This 
 habit of ignoring the works of contemporary writers in another 
 country assumes the form of a grave moral obliquity, when it is 
 traceable, as it is in certain instances, to a wish, on the part of 
 some German philologer,to conceal his obligations to an English 
 labourer in the same field J. It is the inevitable result of these 
 arrogant pretensions, that the professed scholar in Germany will 
 
 * This was the case with Wieland and Jacobi. Wolf's ill treatment 
 of his friend and pupil Heindorf may be traced to wounded vanity; see 
 Schleiermacher's Werke: Philosoph. i. 697 sqq. 
 
 •\ These remarks were made with a general reference; but some 
 special illustrations will be found below, § 370, and in the present author's 
 essay on Classical Scholarship and Classical Learning^ pp. 232 sqq. An 
 eminent German professor lately told a learned friend of ours that he 
 had not thought it worth while to look at Grote's History of Greece^ 
 because he took it for granted that it was entirely borrowed from Ger- 
 man writers on the subject! 
 
 "^ The most painful example of this international plagiarism is fur- 
 nished by the Semitic scholar, Henry Ewald. There is no one in 
 Germany who has written more contemptuously of English learning 
 than this orientalist. In fact, our miserable inferiority to the Germans 
 is his favourite theme. And yet he has been unable to defend himself 
 successfully from the charge, brought against him by the late Dr. Samuel 
 Lee, of having intentionally concealed his obligations to the Hebrew 
 Grammar of an Englishman, who, at the instance of Gesenius, had 
 received from a German university the greatest compliment that could 
 have been bestowed on a foreigner (see Dr. Lee's Examination of Ewald's 
 Grammatical Principles, London, 1847). Not contented with assuming 
 the position of a dictator in regard to Semitic literature, Ewald has un- 
 dertaken to set us right in Greek criticism. Of his competency to write 
 on this subject, it is unnecessary to say more than that he translates 
 άτίε-Λδνβάμενος in Col. ii. 15, as if it had been written άτίεκδνοαςί (Send- 
 schreiben d. Ap. Paulas, Gotting. 1857, ad loc). 
 
CHAP. Π.] PEESENT STATE OF PHILOLOGY. 45 
 
 never acknowledge his errors: his reputation for infalHbility is 
 dearer to him than the truth; and while he receives even the 
 fairest criticisms with an impatient irritation, which not unfre- 
 quently explodes in an outbreak of insolence, he sometimes en- 
 deavours to sustain his untenable position or to cover his defeat 
 by audacious misstatements or disingenuous sophistry*. These, 
 is must be confessed, are serious drawbacks. Nevertheless, all 
 honour be to the Germans, for they have made the mind of 
 Europe what it is; and, though much that many a one of them 
 has written is neither wise nor true, though now and then we 
 may meet with something which startles us even in the pages of 
 the wisest and most true-hearted among them, yet let all those 
 who love knowledge and admire genius apply themselves dili- 
 gently to the treasures of German literature and scholarship, 
 with caution indeed lest they mistake dross for gold, but yet 
 with earnestness of purpose and with full confidence that if they 
 seek aright they will not be disappointed. 
 
 31 The names of the verbal critics who succeeded Bentley 
 in this country have been already mentioned : in their perform- 
 ances is included all that was done by Englishmen for the 
 scholarship of the eighteenth century. The very limited nature 
 of the instructions given at the classical schools at that time, 
 and the insufficient learning of the masters, necessitated some 
 arbitrary method of employing the ten years which it was 
 thought right to spend upon Latin and Greek. The principle 
 of competition which was introduced very early in this country, 
 and the facility of determining the relative merits of such exer- 
 cises, induced the masters of the schools to direct their attention 
 
 * A slight specimen of the manner in which Germans answer criti- 
 cisms, may be seen in some of Bunsen's recalcitrations (e. g. in the preface 
 to the last "volume of his ^gypten, pp. γ sqq.) Although Bunsen lived 
 in England, he did not consider himself bound to acknowledge his obli- 
 gations to Englishmen. At least he did not tell us, that the true opinion, 
 that Hippolytus wrote the book on heresies attributed to Origen, was 
 suggested to himself by Dr. John Barrow of Oxford. In our own de- 
 partment Hermann used to praise Elmsley for his candour and his 
 willingness to admit his mistakes; but when did he follow the good 
 example which he commended? 
 
46 THE HISTOEY AND [bOOK I. 
 
 principally to the composition of Latin and Greek verses, which 
 had been the amusement of scholars at the revival of learning; 
 or perhaps the custom was derived immediately from these early 
 scholars together with the grammars which long maintained 
 their place in our principal schools. The effect of this narrow 
 range of reading was to incapacitate our scholars from making 
 any real advances in philology, while the great facility of verse- 
 making, which every one brought up to the University with 
 him, was a very good preparative for the application of verbal 
 criticism to the correction of the ancient poets. Hence we find 
 that, in knowledge of the old metres, and in conjectural emenda- 
 tion, the scholars of this country were always infinitely superior 
 even to the most learned of their continental contemporaries. 
 The last great scholar of this class was Person, who was indeed 
 a man of very decided genius, and, though he lost himself in 
 trifles unworthy of his power, it must be admitted that, in the 
 little which he directly contributed to Greek learning, he showed 
 a knowledge of the language in all its details, to an extent which 
 can hardly be surpassed. The great mischief of which he is 
 guilty was not so much in omitting to do all that he might have 
 done, though this is deeply to be regretted, but in investing 
 with exclusive importance a merely sub ordinate branch of Greek 
 philology, and thus creating a school of criticism in England 
 which has stood seriously in the way of our philological progress. 
 
 32 At the termination of the war with Napoleon a very 
 great change took place in the intellectual state of the better 
 classes in this country. Excluded as we had been, more com- 
 pletely than at any former time, from all intercourse with the 
 continent, we were suddenly seized with a passion for foreign 
 travelling ; Englishmen were to be met with in every corner of 
 the world; an acquaintance with other languages was no longer 
 thought wonderful ; and in the process of time many returned 
 home impressed with the novel conviction that there was some- 
 thing good out of England. This influence of foreign intercourse 
 was felt latest in its effect upon classical scholarship ; the preju- 
 dice against German scholars, which Porson's quarrel with Her- 
 mann had produced, and his authority with his imitators, kept 
 us for some years after the peace in the same state as before it. 
 
CHAP. II.] PRESENT STATE OF PHILOLOGY. 47 
 
 and the pert mediocrity of the Museum Criticum was the only 
 representative of our philology; some of Person's successors, 
 especially Elmsley, were able critics, and accurate correctors of 
 minutiae, but not philologers. At length, in 1819, Mr. Valentine 
 Blomfield's translation of Matthiae's Greek Grammar made its 
 appearance, and its great superiority to anything of the kind 
 which we had in England soon recommended it to general use; 
 this w^ork and the German editions of classical authors, which 
 were now imported in considerable numbers, began to wean us 
 from our unreasonable prejudices against the scholarship of other 
 countries. In 1824 appeared the first part of Mr. Clinton's Fasti 
 Hellenici^ the most important work which had been published in 
 this country since Bentley's time; and the number of editions of 
 classical works, variorum editions, reprints of old lexicons, &c. 
 which appeared about this time, were a sufficient proof of a 
 reviving spirit of philology. The most important services, how- 
 ever, were performed by the almost simultaneous appearance of 
 translations of Niebuhr's History of Rome^ of Bockh's Public 
 Economy of Athens, and of Miiller's Treatise on the Dorians. In 
 fact, English scholarship became every day more and more ger- 
 manized^ to such an extent that our philologers hardly dared 
 think for themselves. Gradually, however, the characteristic 
 elements of the English mind have asserted themselves, in this 
 as in other departments ; and our apprenticeship to German phi- 
 lology has ended in producing a number of original workmen at 
 least equal to the majority of those in whose school they have 
 been trained. In some of the highest applications of scholar- 
 ship we should impartially prefer the English to the German 
 expositor ; and books have been published in this country during 
 the last twenty years, which stand in favourable contrast to any 
 continental works on the same subject*. Although Dr. Arnold's 
 edition of Thucydides, of which the first volume appeared in 
 1830, was not distinguished by profound or even very accurate 
 learning, we believe that the manly tone in which the notes are 
 written, and the large and unprejudiced views which they mani- 
 fest, and which were never seen in any former commentary on 
 
 * We may mention particularly The Egypt of Herodotus: with notes and 
 preliminary dissertations, by John Kenrick, M. A. London, 1841. 
 
48 THE HISTORY AND [bOOK I. 
 
 an ancient author, produced a most beneficial influence on clas- 
 sical literature in general. The two periodicals which were 
 established shortly afterwards, — the Journal of Education and 
 the Philological Museum — at least gave indications of the fact 
 that a truer scholarship, a more comprehensive criticism, and 
 more extended views of general literature, were beginning to 
 supersede the meagre, one-sided notions of a few years before. 
 And if we must pronounce an opinion with regard to the pro- 
 spects of British philology at present, we must needs confess that 
 they are far from discouraging, and that, although there is still 
 much to be done before we can say, that the educational and 
 literary results of our philological training are accurately and 
 fairly balanced, and that the scholarship, which has always been 
 the distinguishing mark of the English gentleman, has uni- 
 formly developed itself, whenever the occasion has arisen, into 
 the learning, which is too often supposed to have been exclu- 
 sively appropriated by the German professor, there is every 
 reason to hope that the labourers will not long be wanting; and 
 that while we shall retain the practical skill, which is calculated 
 to produce such important effects on the moral and intellectual 
 character of our younger students, we shall be able fully to meet 
 the increasing demand for a larger amount of comprehensive and 
 scientific knowledge. AVhen two statesmen, who have held in 
 succession the office of Minister of Finance in this great com- 
 mercial country *, have found time in the midst of their political 
 avocations to discuss with an overflowing abundance of erudition 
 the early history of Rome and the first beginning of Greek 
 poetry, there is little ground for the apprehension that classical 
 philology will not be duly prosecuted by those, who, as teachers 
 or writers, have undertaken professionally to expound the dead 
 languages and the literature to which they are the key. En- 
 lightened and accomplished scholars have been placed at the 
 head of all our great schools, so that the stumbling-blocks of 
 prejudice are no longer to be dreaded in that quarter; the Uni- 
 versities, it is well known, have never opposed themselves to 
 any necessary change in the method or extent for their classical 
 studies ; and in the recent histories of Greece and Rome, which 
 
 * Sir G. C. Lewis and Mr. W. E. Gladstone. 
 
CHAP. II.] PEESENT STATE OF PHILOLOGY. 49 
 
 have been completed or commenced in this country*, and which 
 will mark this epoch in the annals of our philology, we may 
 clearly discern what may be expected from an union of the mul- 
 tifarious learning and chastened scepticism of the better class 
 of German scholars with the good taste, sober judgment, and 
 straight -forwardness for which this country is so eminently 
 distinguished. 
 
 33 ΛVhile classical scholarship has been making this pro- 
 gress in England and on the continent, a new branch of philology 
 has grown up by its side, and is now so matured as to be entitled 
 to a place among its primary elements. This is the comparative 
 study of languages, which, while it also serves as an important 
 instrument in the investigations of ethnography and early his- 
 tory, seems likely to engross the whole grammatical element of 
 the older philology. It is curious to trace this study back to the 
 time when the first glimmerings of light broke upon the scho- 
 lars of Europe, when they first saw the little blue flame which 
 pointed out to them the treasures still in the mine. It must be 
 admitted, in the first place, that classical scholarship can claim 
 no share in the first production of the comparative grammar, 
 which is now its most important adjunct. Theology was the 
 original, though not perhaps the voluntary cause of this import- 
 ant addition to human knowledge. Linguistic science beganf 
 in the examination of the claims of various dialects to be con- 
 sidered as the original language of the world ; and the oldest 
 collections of materials for the erection of this new superstructure 
 were made by Propagandas and Bible Societies. Most of the 
 earliest inquirers favoured the pretensions of the Hebrew, and, 
 as they proceeded by derivation rather than by comparison, 
 they were obliged to have recourse to a number of very forced 
 etymologies in endeavouring to establish the affiliation of the 
 
 * By Dr. Thirlwall, Mr. Grote, Mr. Maiden, Dr. Arnold, Mr. Donne, 
 and Mr. Merivale. 
 
 f Dr. Wiseman has given a brief history of comparative philology in 
 his first and second lectures. Those who wish for further information 
 may consult Dorn, uber die Verwandtschaft des persischen^ gennanischen und 
 griechisch-lateinischen Sprachstammes y pp. 91 sqq., and Pott, in Ersch und 
 Gruber's Encyclop. s. v. Indo-Germanischer Sprachstamm^ pp. 6 sqq. 
 
 Ε 
 
50 THE HISTORY AND [bOOK I. 
 
 European languages to a Semitic mother. Even when they per- 
 ceived the connexion between sundry of the Asiatic and Euro- 
 pean members of the Indo-Germanic family*, as when Lipsius 
 in 1599f or Salmasius in 1643 J made a comparison of modem 
 Persian with German and Greek, they were utterly unable, from 
 the want of a wider induction, to see the important conclusions 
 to which such comparisons might lead, nor indeed had they 
 any fixed principles to guide them in their search. 
 
 34 The first who pointed out the true method of proceed- 
 ing was Leibnitz, whose comprehensive genius seems to have 
 suggested the beginning of almost every improvement in science. 
 This great man saw that the Hebrew had no claim to be consi- 
 dered as the mother of languages ; he showed that we must take 
 the widest possible inductions, and compare with one another the 
 languages of nations most remote in geographical position, and 
 that the words to be selected for comparison were the most sim- 
 ple and necessary terms in each language§; in fact, he pointed 
 out the course by pursuing which succeeding philologers have 
 realized most of the hopes which he entertained. 
 
 35 But, although the right method was thus suggested, 
 nothing of any great importance was performed till the discovery 
 of the Sanscrit or sacred language of India. The spirited re- 
 rearches of Anquetil du Perron , about the middle of the last 
 century, had introduced a knowledge of the old languages of 
 Persia, the Zend and the Pehlvi, into Europe; but, though it 
 has since been established that the Zend is but a variety of the 
 Sanscrit , no attempt was made to connect the study of Zend 
 
 * It has been erroneously supposed (Camden, Britannia^ p. 123; 
 Wilkins in Chamherlayn. orat. domin. p. 7^ that Joseph Scaliger was the 
 first to notice the resemblance of Persian and German; but although he 
 could not help observing the identity of some very common words in the 
 two languages, he was far from thinking this an argument for a sameness 
 of origin (EpistoL p. 489). If any one preceded Lipsius, it must have 
 been Pontanus, but we are not acquainted with his writings. 
 
 t In a letter to Schott {Lipsi 0pp. Vol. i. pp. 282 sqq. Edit. 1G13). 
 
 ;J; De lingua Hellenistica Commentarius, pp. 331 sqq. 
 
 § Otium Hannoveranum, p. 80; Collectanea Etymologica et Meditationes 
 de Originibus Gentium (passim); Commerc. Epist. Vol. iii. p. 79; Nouveaux 
 Essais sur fEntendement Hiimain, pp. 236 sqq. 
 
CHAP. II.] PEESENT STATE OF PHILOLOGY. 51 
 
 with philological researches. Anquetil du Perron himself was 
 no scholar, and people were much busier in endeavouring to 
 settle the claims of the Zend to be at all considered as a lan- 
 guage than to profit by it. In the year 1765, however, the 
 treaty of Allahabad made the East India Company sovereigns of 
 Bengal, and their first proconsul, Warren Hastings, calling in 
 the assistance of some Brahmins , drew up an epitome of the 
 native laws, by which the new rulers had resolved to govern 
 their subjects. In the preface to this book, which was translated 
 into English through the Persian, and was published in London 
 in 1776, with the title Code of Gentoo Law^ Mr. Halhed, the 
 editor, gave at second-hand the earliest account which had ap- 
 peared in Europe of the Sanscrit language, the original vehicle 
 of these laws. Two years after the appearance of this book, Mr. 
 Halhed published at Hoogly his Bengal Grammar, in the preface 
 to which he remarks (p. iii), "I have been astonished to find the 
 similitude of Sanscrit words with those of Persian and Arabic, 
 and even of Latin and Greek : and these not in technical and 
 metaphorical terms, which the mutuation of refined arts and im- 
 proved manners might have occasionally introduced; but in the 
 main ground-work of language, in monosyllables, in the names 
 of numbers, and the appellations of such things as would be first 
 discriminated on the immediate dawn of civilization.'^ These 
 remarks were followed up and confirmed by Sir \Villiam Jones 
 and Sir C, Wilkens, the latter of whom translated the Bhagavad- 
 gita in 1785, and the Uito'padei^a in 1787, and the former, besides 
 other services, awakened a European interest in Indian literature 
 by his translation of the Sakuntald in 1789. A German Mis- 
 sionary, named John Philip Werdin, but more commonly known 
 as the Pater Paulinus a SanctoBartholomaeo, deserves the credit 
 of being the first to publish Sanskrit books in Europe, and to 
 make a formal comparison of the Zend and Sanscrit with the 
 German and Latin; but as he had only a vague acquaintance 
 with the Sanscrit, and did not print his books in the Deva- 
 ndgart or Sanscrit character, he can hardly be said to have been 
 the founder of Sanscrit philology in Europe*. 
 
 * Werdin heaped all kinds of abuse upon our Calcutta scholars, and 
 therefore has been somewhat roughly handled by them. Dr. Leyden says 
 
 E2 
 
52 THE HISTORY AND [bOOK I. 
 
 36 The real beginning of this study is due to our country- 
 men in India, who acquired a thorough and critical knowledge 
 of the language from the native Pandits, and made it by their 
 writings accessible to European students. In the first rank of 
 these stand the two pioneers whom we have already mentioned, 
 Sir William Jones, the founder of the Asiatic Society at Cal- 
 cutta, and Sir Charles Wilkins, the first who translated Sanscrit 
 books, and printed Sanscrit in Europe, and the author of an 
 excellent Sanscrit Grammar*; and the work was completed by 
 two other Englishmen, distinguished alike by their abilities and 
 their industry: Mr. Colebrooke, the author of the first good San- 
 scrit Grammar, the beginner of Sanscrit Lexicography (by his 
 admirable edition of the Ama7'a Cosha)^ and, in fact, theScaliger 
 of Sanscrit scholarship ; and Professor Wilson, the compiler of 
 the great Sanscrit Dictionary. From England the knowledge of 
 this language passed into Germany. Frederick Schlegel, who 
 visited this country during the peace of Amiens, learned a smat- 
 tering of Sanscrit irom Mr. Hamilton (afterwards Buchanan), an 
 English officer who was detained as a prisoner of war in France; 
 and he subsequently gained further instruction fromM.Langles, 
 at Paris, where, since Anquetil du Perron's time, the Asiatic 
 languages had been cultivated with some success, and where the 
 
 (Asiatic Researches, Vol. x. p. 278), "his Vyacarana, seu locupletissima 
 Samscrdamicce linguce Institution RotncB, 1804, has given a death-blow to 
 his vaunted pretensions to profound oriental learning; and shown, as was 
 previously suspected, that he was incapable of accurately distinguishing 
 Sanscrit from the vernacular languages of India." And Professor Wilson 
 has pointed out the ludicrous fact that he mistook the Amara-Cvsha, a 
 common Sanscrit vocabulary, for a ritual and liturgy coexistent with the 
 origin of the Hindu Idolatry and the basis of the Brahmanical super- 
 stitions (Preface to Sanscrit Dictionary, 1st edition, p. 22). 
 
 * The Edinburgh reviewer of this grammar (Vol. xiii.) deserves the 
 credit of being one of the very first of our countrymen who perceived the 
 extensive affinities of the Sanscrit: he has given a comparative table of 
 Sanscrit, Persian, Latin, and German, which at once established the com- 
 mon origin of these languages. It is an interesting fact that the great 
 Edmund Burke had paid considerable attention to the filiation of lan- 
 guages long before the study of Sanscrit became known in Europe. See 
 Bisset's Life of Burke, ii. pp. 390, 1. Prior's do. p. 427: quoted by 
 Buckle, Civilization in England, p. 415. 
 
CHAP. II.] PEESENT STATE OF PHILOLOGY. 53 
 
 old language of Persia has been submitted to an analysis as 
 exact and scientific as that which has elucidated the history and 
 explained the structure of the other Indo-Germanic idioms*. The 
 little work on "the Language and Wisdom of the Indians,'' 
 which F.Schlegel published in the same year with Colebrooke's 
 Amara-Cosha and Wilkins's Grammar, was the first to awaken 
 the Germans to a sense of the value of these studies. The mere 
 knowledge of Sanscrit, however, would not have been sufficient 
 to create or even to suggest the comparative philology of the 
 present generation.Polyglot-collectionsofPaia'-nosi^rs furnished 
 but scanty materials for a wide induction; and though the Mith- 
 ridates ofAdelung and Vater undoubtedly contributed to awaken 
 a taste for linguistic studies, little advance would have been made 
 in the right direction had not some men of genius turned their 
 attention to the great capabilities of this untried field of specula- 
 tion. The establishment of Sanscrit scholarship as a branch of 
 European philology is due to Francis Bopp and Augustus Wil- 
 liam Schlegel. By the exertions of these two scholars, seconded 
 in no slight degree by the illustrious William Humboldt, a sound 
 and accurate knowledge of the Sanscrit language has been intro- 
 duced into Germany; and Schlegel, in particular, brought to 
 the study of the old Indian poems the truly literary spirit and 
 the highly-finished taste, which are so conspicuous in his cele- 
 brated lectures on the drama and in his inimitable translation of 
 Shakspere. The interest which the study of Sanscrit has created 
 in Germany results entirely from the greater diflfusion of philo- 
 logy in that country; and a wonderful impulse to the study of 
 comparative grammar had been given by Rasmus Rask's treatises 
 on the origin and affinities of the Scandinavian and Teutonic 
 idioms, and by the subsequent publication of James Grimm's 
 Deutsche Grammatikf, by far the most important book of the 
 kind which has made its appearance since the revival of letters. 
 William Humboldt's posthumous work on the Kawi dialect 
 
 * We allude to M. Eugene Burnouf's most meritorious labours in editr 
 ing and explaining the fragments attributed to Zoroaster. 
 
 t Grimm's last work (Gesch. der deutschen Sprache, Leipzig, 1848) con- 
 tains in its ethnographical department some very ingenious and learned 
 combinations. But the purely philological chapters appear to us singu.- 
 larly deficient in critical discrimination. 
 
54 THE HISTORY AND [bOOK I. 
 
 applies the results of a truly comprehensive survey of human 
 speech to the solution of the most important problems respecting 
 the origin of language. And at an earlier period, Ray nouard had 
 shown, in his scientific examination of the Romance languages, 
 the causes which lead to the disorganization of grammatical struc- 
 tures. By means of these masterpieces of philological criticism, 
 and of Bopp's numerous writings, to the efficacy of which the 
 subsidiary or subsequent researches of Pott, Eugene Burnouf, 
 Lassen, Lepsius, Diez, Diefenbach, Zeuss, Weber and Max 
 Miiller, each working diligently in his own field, have in no 
 slight degree contributed, comparative philology has been raised 
 to the rank of a science, the mechanism of the difi'erent branches 
 of the Indo- Germanic family has been investigated and ex- 
 plained , the pedigree of the various dialects included in it has 
 been enrolled, and their written monuments have been rendered 
 accessible and intelligible by the application of the sound criti- 
 cism created and fostered by the classical studies of Europe. 
 
 37 In the first beginnings of this new branch of inductive 
 science, England, we are sorry to say, did little that will bear 
 comparison with the performances of our continental neighbours, 
 in regard either to comparative philology in general, or to Indian 
 scholarship in particular. Indeed, with the exception of the 
 great Colebrooke and Sir Graves Haughton, no one of our San- 
 scrit scholars can be called a philologer in the higher sense of 
 the word, and even these eminent orientalists have confined their 
 attention to the languages of Asia. Accordingly, as we borrowed 
 our philology in its literary spirit from the Germans, we were 
 compelled to import also the raw materials at least of their 
 comparative grammar. But when the good work had once 
 commenced amongst us, our philology made very rapid progress, 
 and we can point to conceptions more original, and to results 
 more important, than any which have signalized the efforts of 
 the learned elsewhere. It is not to be denied that we had great 
 advantages at starting, and that it would have been very dis- 
 graceful if we had not learned to profit by them. Bopp's 
 System, which appeared in its first outlines in 1816*, was pub- 
 
 Conjucjationssystem der Sanskritsprache, Frankf. 1816. 
 
CHAP. II.] PEESENT STATE OF PHILOLOGY. 55 
 
 lished four years afterwards in an English journal*, and one of 
 his most able and distinguished pupils, the late Dr. Rosen, 
 became naturalized among us, as Professor of Sanscrit in Uni- 
 versity College, London, more than thirty years ago, and so 
 consecrated his learning and abilities to the service of our phi- 
 lology. It is difficult to estimate the loss which learning in 
 general sustained in the too early death of this admirable person: 
 but we must not forget that we really owe to him indirectly 
 the first application of comparative philology to the public 
 teaching of the classical languages, a merit which has been too 
 eagerly claimed for and too readily conceded to the Greek and 
 Latin Professors, who merely transmitted to their pupils the 
 ideas and information which they had derived from their Ger- 
 man colleague, and who, in the long period, more than a quarter 
 of a century, which has since then elapsed, have not proved 
 themselves capable of building on the foundation which he 
 had laid. The same influence was soon conspicuous in the pages 
 of the Penny Cyclopcedia , to which Dr. Rosen was himself a 
 frequent contributor, and which owes its decidedly philological 
 character to writers more or less connected with the London 
 University. Independently, however, of this immediate relation 
 to Bopp's philological, school, there appeared, about the same 
 time, two writers whose services to philology have been of the 
 utmost value. Dr.Prichard, who may be regarded as almost the 
 founder of the Science of Ethnography, was not long in per- 
 ceiving the important aid which his favourite speculations might 
 derive from a consideration of the affinities of language. He 
 made no insignificant contribution to pure philology in his vin- 
 dication of the claim of the Celtic languages to a place in the 
 great Indo-Germanic family f; and in his Natural History of 
 
 * Annals of Oriental Literature^ Vol. i. London, 1820. 
 f The Eastern Origin of the Celtic JNations, London, 1831. Although 
 we think that Dr. Prichard has on the whole proved his point, we must 
 take the liberty of saying, that his little book shows a great want of phi- 
 lological exactitude. He has not attempted to distinguish between those 
 words which the ancient Britons might have derived from the Roman 
 conquerors, or from the Anglo-Saxons who subsequently established 
 themselves in the island, and those which must have belonged to the 
 Celtic dialect from the first. Accordingly, many of his instances, where 
 they are merely comparisons of Celtic with the Latin, or with the German 
 
56 THE HISTORY AND [bOOK I. 
 
 Man^ and other works, he has done a great deal towards classi- 
 fying and grouping the varieties of human speech. The late 
 Mr.Garnett, whose comprehensive and truly philosophical ana- 
 lysis of the constituent elements of language was first made 
 known in a notice ofDr.Prichard's Celtic work*, developed his 
 views in various contributions to the records of UiQ London Phi- 
 lological Society; and we do not know where to look for sounder 
 or more instructive examples of linguistic research f. In regard 
 to palaeography also , English scholarship may claim the hon- 
 our of having made the first and most important, or the last 
 and most complete discoveries. Young guided Champollion 
 to that systematic examination of the Egyptian hieroglyphics, 
 which, in the hands of Schwartze, Bunsen, and Lepsius, has 
 produced, and is still producing, such important results for com- 
 parative grammar, chronology, and historical criticism J. James 
 Prinsep, by his unaided industry and acuteness, first discovered 
 the clue to the two alphabets in which the Buddhist Inscrip- 
 tions of A 9oka are written , and thus opened the way to safe 
 and important investigations respecting the development of the 
 old languages of India §. Sir H. Kawlinson, with the in- 
 
 dialects, prove nothing; his strong point is the comparison with Greek 
 and Sanscrit, and his analysis of the pronouns and inflexions; but of 
 these particulars a further and more accurate examination has been under- 
 taken by Zeuss and others. 
 
 * Quarterly Review, Vol. lvii. Essays, pp. Ill sqq. 
 
 t The Philological Essays of this accomplished and able man have 
 been published in a collected form by his son (London, 1859). The 
 student of comparative philology will find in this volume a most scien- 
 tific analysis of inflected language illustrated by the happiest applications 
 of critical skill and ingenuity. 
 
 J Many English writers have made valuable contributions to the de- 
 tails of the important subject of Egyptology; e. g. Mr. Osborn, Dr. Hincks, 
 Sir G. Wilkinson, Mr. Birch, and Mr. Sharpe. Nor must we omit to 
 mention the labours of Mr. G. R. Gliddon, who has made the wonders of 
 ancient Egypt familiar to our brethren in the United States of America: 
 see Otia jEgyptiaca : Discourses on Egyptian Archceology and Hieroglyphical 
 Discoveries. London, 1849. 
 
 § See the recognition of his merits in Lassen's Indische Alterthums- 
 kunde, II. p. 222; and in Weber's Indische Skizzen, pp. 25, loO. Prin- 
 sep's remains have been very ably edited by Mr. Edward Thomas {Essays 
 on Indian Antiquitie?, by the late James Prinsep. London, 1858, 2 Vols. 8vo.) 
 
CHAP. II.] PKESENT STATE OF PHILOLOGY. 57 
 
 telligeni aid of Mr. Norris^ has decyphered and translated the 
 contemporaneous records of the first Darius, and the same in- 
 dustry and acuteness will probably extract historical truth from 
 the cuneiform chronicles which the enterprise and indefatigable ., 
 zeal of Mr. Layard have brought from the ruins of Nineveh. 
 To come nearer home, Mr.Kemble and Mr. Thorpe have prose- 
 cuted the study of Anglo-Saxon philology with no less devotion 
 than Grimm. Mr.Kemble, in particular, formally enounced the 
 leading principles of comparative philology in his lectures on the 
 history of the English language, delivered before the University 
 of Cambridge in 1834; and has since published more than one 
 original work indicating the depth and extent of his acquaintance 
 with Teutonic lore. Dr.Bosworth also has connected his Anglo- 
 Saxon researches with comprehensive investigations in general 
 philology, and Dr. Latham, who had first distinguished himself 
 in this field, especially by the publication of a treatise on the 
 English language, has since become very generally known as a 
 collector of philological facts, mainly with reference to the lan- 
 guages of Africa. When we look to the activity of the Asiatic, 
 Geographical, Philological, and Syro- Arabian societies in this 
 country, to various publications which appear from time to time*, 
 and to the efiects which may be expected from the adoption of 
 ethnographic philology by the British Association, we cannot 
 allow ourselves to entertain any fears respecting the successful 
 cultivation of linguistic science in this country. 
 
 38 From this survey it will easily be seen what is the con- 
 dition of scholarship in reference to the higher objects and more 
 extensive applications of which we have spoken in the preceding 
 chapter. To these we need not return: but it is desirable that we 
 should inquire, whether the knowledge which we have recently 
 
 * We ought particularly to mention the late Mr. Winning's Manual of 
 Comparative Philology^ London, 1838. The first and second parts of 
 this work, which are composed, in a great measure, of well-selected ex- 
 tracts and translations from other writers, with intelligent criticisms on 
 their opinions, are worthy of almost unqualified approbation. The third 
 part is rather at variance with the other two, and is deformed by refer- 
 ences to Rabbinical authorities, on which we do not set the slightest 
 value. 
 
58 THE HISTORY AND [bOOK I. 
 
 gained with regard to language in general, and the Indo-Ger- 
 manic family of languages in particular, may not now he applied 
 by the classical scholar in gaining a more correct insight into the 
 structure of the Greek language, in classifying more accurately 
 its grammatical forms, and in interpreting more satisfactorily 
 the authors who have written in it. To this inquiry we may add 
 another: whether the general study of comparative grammar as 
 applied to thelndo-Germanic languages would not gain by such 
 an examination of the most perfect member of the family. 
 
 The advantages which classical scholarship would derive 
 from a more intimate union with comparative philology may 
 easily be enumerated ; and though some eminent verbal critics 
 were dispos'ed at first to reject the proffered assistance of their new 
 ally, the more genial and enlightened students of Greek and 
 Latin literature were prompt in their recognition of the coming 
 reform of grammar. In the year 1836 — only three years after 
 the first, instalments of Bopp's Vergleichende Grammatik and 
 YoiVs Etymologische Forschungen were giyento the world — CO. 
 Miiller, who perhaps took a more completely literary view of 
 classical scholarship than any German of our days, said distinctly, 
 with reference to the narrower views entertained by G.Hermann: 
 " it has now indeed come to this, that philology must either re- 
 nounce altogether any historical knowledge about the growth of 
 language, and all etymological researches into the form of roots 
 and the organism of grammatical structures, or trust herself in 
 these matters entirely to the guidance and counsels of the compa- 
 rative study of language*." The true scholar is of course not 
 merely a student of the Greek and Latin languages and an inter- 
 
 * Gotting. Gel. Anz. 1836, pp. 169 sqq.; Kleine Schriften, I. p. 12. 
 The following are the words in which Hermann expressed his prejudices 
 against comparative philology: "alii non magis multa justaque ratione 
 exculti, lucem sibi inde ubi sol oritur, repercussam aurora boreali, affulsu- 
 ram sperantes, ad Brachmanas et Ulphilam confugiunt, atque ex paucis 
 non satis cognatarum linguarum vestigiis quae GrEecorum et Latinorum 
 verborum vis sit explanare conantur. Qui ut hie illic alicujus vocabuli 
 formaeve originem inveniant, tamen ad Graeese Latinaeque linguae rationem 
 explicandam vereor ne non plus lucrentur, quam si Germanus aliquis 
 gentis suae linguam pluriraa vocabula communia cum Graecis sciat: quo 
 ille sua lingua nihilo rectius utetur, quam si id nesciat." 
 
CHAP. IL] present STATE OF PHILOLOGY. 59 
 
 preter of the authors who have written in them. It is his busi- 
 ness to lift the curtain which has fallen on the glories of the 
 past: to bring Athens and Rome again upon the stage: to enable 
 the modern reader to regard the old authors and the events of 
 which they write with the eyes of a contemporary. With re- 
 gard to this latter function the study of comparative philology 
 is of little avail. The Greek and Latin authors must be read 
 together and in connexion, and we must endeavour to peruse 
 them with as little interruption as possible from modern and 
 extrinsic associations. But for the study of the Greek language 
 alone and for the critical interpretation of Greek authors, com- 
 parative grammar is indispensable. And first, with regard to 
 the explanation of particular passages, in which daily experi- 
 ence teaches us that much remains to be done even after all 
 the labours of preceding scholars. The method of most ex- 
 tensive application is indeed purely a literary one: it is to 
 deduce the meaning of the words in question from a general 
 survey of the connexion of thought in the whole work, and, 
 for the language, from a comparison of the passage with other 
 similar ones in the same or contemporary writers. Cases, how- 
 ever, frequently occur in which the difficulty or misapprehen- 
 sion results entirely from an ignorance of the meaning of some 
 particular word ; and though, as even Thomas Aquinas has told 
 us*, the signification of a word is not identical with its ety- 
 mology, yet the latter is sometimes essential, on the principle 
 of suggestion, in order that we may arrive at the former, which 
 in most instances will also be determined, 'prima facie, by the 
 context. Now in such cases we must have recourse to compara- 
 tive philology combined with, and regulated by, the old method 
 of scholarship, and we hope to show, in the course of the 
 following pages, that something may be effected by such an 
 
 * Aliud est etymologia nominis et aliud significatio nominis. Etymoloyia 
 attenditur secundum id a quo imponitur nomen ad significandum : nominis 
 vero significatio secundum id ad quod significandum imponitur, quce quando- 
 que diversa sunt; nomen enim lapidis imponitur a Icesione pedis ("!), non 
 tamen hoc significat. Alioquin ferrum, cum pedem Icedat, lapis esset. Similiter 
 etiam nomen superstitionis non oportet quod significet illud a quo nomen est 
 impositum. — Thomas Aquinas, Summa Theolog. ii. 2, quwst. 92. Artieulus 
 primxis: conclusio ad 2, p. 201. Edit. Paris. 1631. 
 
60 THE HISTORY AND [bOOK I. 
 
 union. But secondly, it is also the object of the Greek scholar 
 to anatomize the forms of the words, to classify them according 
 to their etymological coherency, and so to simplify the practical 
 grammar of the language. For this purpose again comparative 
 philology is indispensable. If we were confined to the Greek 
 language we should know absolutely nothing of the principles of 
 its verb-conjugation, of the declension of its nouns, of the value 
 of its particles, and of the real causes of its peculiar conformation. 
 A sufficient proof of this is furnished by the old grammars and 
 lexicons. But after we have once taken up the higher ground 
 of comparative philology our difficulties on these points, be they 
 ever so great, immediately vanish, and our perplexities, how- 
 ever intricate, are at once unravelled. In consequence of the 
 facility with which all this is effected by the true scholar, many, 
 who knew little either of the language which they Avished to 
 illustrate or of the aids which they sought to employ, have been 
 led to attempt a solution of all the difficulties of Greek Grammar 
 by a comparison with Latin, German, or Sanscrit. It is not to 
 be wondered at that such persons should fall into great errors 
 and incur the just reprobation of mere Greek scholars. At the 
 same time, however, it is not to be supposed, as some of these 
 last have inferred from the failure of the would-be philologers, 
 that he who would increase our knowledge of the Greek language 
 must confine himself to it, and that the study of comparative 
 grammar is rather injurious than beneficial*. It is true that an 
 accurate study of the Greek language alone is more profitable to 
 the educated man and to the scholar than a smattering in twenty 
 others: for knowledge, which may be made the subject of thought, 
 is always better than general information, which can only be ren- 
 
 * These remarks are pointed at Lobeck (AglaophamuSj p. 478, note i; 
 Paralip. p. 127, note; Pathol, praef. p. vii) and his pupil Ellendt {Lex. 
 Sophocl. praef. p. iii). See Varronianus <, p. Ii4. They also refer to 
 the Programme of the Imperial Academy of Sciences at St. Petersburg 
 (29 Dec. 1836), proposing a reward for a treatise on the Greek dialects 
 from which all consideration of Sanscrit affinities was to be excluded. 
 This Programme is printed in Seebode's Neue Jahrhucher^ vii. Jahrg. xx. 
 Band. 3. Heft, p. 341. As usual Mr. G. Curtius follows in our steps in 
 his review of Lobeck's Rhematikon (Zeitschr. f. d. Alterthumsw. 1848, 
 pp. 151 sqq.). 
 
CHAP. II.] PRESENT STATE OF PHILOLOGY. 61 
 
 dered available for conversation in society. It is also true that 
 our means of elucidating the difficulties of the Greek language, 
 from itself alone and independently of comparative grammar, 
 are greater at the present than at any former time. The in- 
 scriptions which have been lately collected and explained, the 
 remains of Greek grammarians and lexicographers which have 
 been published from manuscripts in the various public libraries 
 of Europe, and the labours of Lobeck and others in examining 
 the forms of the Greek language as they appear in the whole 
 range of authors, would have enabled us to arrive at more accu- 
 rate conclusions, than was before possible, with regard to the 
 earliest state of the Greek dialects, even though the other lan- 
 guages of the Indo-Germanic family had been unknown or neg- 
 lected. But, though the comparative philologer would be much 
 to blame if he failed, as some have done, to avail himself of these 
 and similar resources, those are not to be listened to who would 
 tell us, on the other hand, that the mere Greek scholar is more 
 to be trusted than one whose studies, while equally accurate, have 
 taken a wider range. It is as if a man, who might illuminate a 
 room with a number of lamps, should find out that some one of 
 them gave more light than any one of the others, and should 
 therefore content himself with this one alone. The industrious 
 observer of phenomena in the Greek, or any other language of 
 the family to which we refer, is but a hewer of wood and a 
 drawerof water for the architectonic philologer: he brings some 
 of the materials necessary for the work, but cannot lay claim 
 to any share in building up the mighty fabric of general scho- 
 larship ; for, dig as he may in his own narrow quarry, he will 
 never catch a glimpse of the ground-plan and elevation so long 
 as he remains there. One-sided views are of little use to the 
 philologer of our day; and if he who forsakes the specialties of 
 Greek for the generalities of comparative grammar has made 
 a false step, neither is he to be commended, who, from prejudice 
 or want of resolution, obstinately refuses to read more than one 
 page of the great book of language which lies open before him. 
 
 39 With regard to the other question, namely, as to the 
 benefits which would result to the general study of compara- 
 tive grammar from a combination of it with accurate Greek 
 
62 THE HISTOEY AND [bOOK I. 
 
 scholarship, very little need be said. The majority of those 
 who have hitherto written on comparative philology have re- 
 garded the subject from the side of the oriental languages or 
 of the German dialects, and occupied by the extent and novelty 
 of their subject, have not paid sufficient attention to the old 
 classical languages of Europe. In fact, no one of the great 
 comparative philologers who have done so much for the science 
 is a professed classical scholar, and, as might have been ex- 
 pected, they occasionally fall into errors with regard to the 
 structure of the Greek and Latin languages, which are suffi- 
 ciently obvious to the scholar who has been able to study those 
 languages with the advantages which may be derived from an 
 acquaintance with the results of their laborious researches*. For 
 
 * It would be easy to give examples of the mistakes into which com- 
 parative philologers have been led by the habit of starting invariably 
 from Sanscrit. Perhaps the best instance of the advantage, which they 
 might occasionally derive from the converse process, is furnished by the 
 word vidua compared with the Sanskrit vidkava. Bopp (Gloss, p. 321) 
 and Benfey (Wvrzellex. ii. p. 272), adopt the opinion of the Sanscrit 
 grammarians, that this word is compounded of the prefix vi, "without," 
 and dhava, "a husband," and Max Miiller, who accepts this interpretation 
 (Oxford Essays, 1856, pp. 21, 22), — with the importaat deduction, that "if 
 the custom of widow-burning had existed at that early period, there would 
 have been no vidhavus, no 'husbandless women,' because they would all 
 have followed their husband unto death," — says afterwards (p. 38), "it must 
 be confessed that the old Latin viduus, a name of Orcus, who had a temple 
 outside Rome, makes it doubtful whether the Latin vidua is really the 
 Sanscrit vidhava, however great their similarity; unless we admit that a 
 verb viduare was derived from vidua, and that afterwards a new adjective 
 was formed with a more general sense, so that viduus to a Roman ear 
 meant nothing more than privatusy No comparative grammarian, start- 
 ing from the safe basis of classical philology, could have fallen into these 
 difffculties and contradictions. There is not the slightest doubt that 
 vid- or id- is a genuine Latin root, signifying to "divide" or "separate;" 
 we have it in di-vido and idus (διχόμηνις ημέρα) ; perhaps also in video (cf. 
 dis-cerno with di-vido); and if so in οΐδα, εΐδον, &c.; and even in the 
 Semitic Via (see Maskil le Sopher, p. 38). This root, in the same sense 
 as in viduus, appears in Indo-European languages in which it is admitted 
 that there is no trace of the Sanscrit dhava, e.g. the O. H. G. witaioa, 
 Goth, viduvo, Old Sax. widowa, Anglo -Sax. viduva, Old Pr. widdewu, 
 Sclav, vdova, Irish feadhh. From this we should infer that the Sanscrit 
 is the corrupted form, and that the root of vidhava is really that which 
 
CHAP. Π.] PRESENT STATE OP PHILOLOGY. 63 
 
 example, Bopp'sCbwj^araiive Grammar is conspicuously deficient 
 in that critical tact which is rarely found in any one who has not 
 passed through the regular training of the older classical scholar- 
 ship; nor indeed does this excellent etymologer give any evidence 
 of an extensive familiarity with the Greek or Latin authors. 
 Intimately acquainted with the old languages of India and 
 Persia, and well disciplined in Grimm's Teutonic philology, Bopp 
 has not been able to acquire either the knowledge or the habits 
 of mind which characterize the ripe and elegant scholar. His 
 own field is wide, and he has well surveyed it. But he has not 
 crossed its boundaries. It must not be forgotten, however, that, 
 although the science of comparative philology advances so rapidly 
 that every succeeding writer, if competent to add any thing to 
 the stock of knowledge, is also able to correct many mistakes 
 and supply many deficiencies of his predecessors, no one ought 
 to make this the ground of any assumption of superiority ; for it 
 would well become every one who follows in the steps of Grimm 
 and Bopp to recollect that he must himself have fallen into much 
 graver errors had not these men gone before him: the κελενΟΌ- 
 TtOLol Ttaideg '' Ηφαίΰτον, χ%•όνα άνήμερον Tid'evTsg ημερωμένην, 
 should be held in honour even by the constructors of rail roads. 
 
 40 To pass from these general considerations to the subject 
 at present before us, it may, we think, be concluded, that the 
 time is at length come when the Greek language at all events 
 mustbesubjected to the same scrutiny, absolute and comparative, 
 to which the great body of German dialects has been submitted 
 by Grimm, and the Sanscrit, Zend, and Sclavonic, in addition to 
 these, by Bopp. This examination, however, should be lexico- 
 graphical as well as grammatical. Buttmann was well aware of 
 this, when he added his Lexilogus to his admirable Grammar. 
 And here let us express our regret that a man so wonderfully 
 gifted, combining as he did all the learning of the old school 
 with sound views and unexampled ingenuity, was placed in an 
 age preceding though by a few years only the full establishment 
 
 also appears as bhid, li&t.fid- (in findo), the termination bearing the same 
 relation to the Latin -uus that tava does to tuus. 
 
64 THE HISTORY AND PRESENT STATE OP PHILOLOGY. 
 
 of comparative philology*. It must be obvious to any reader of 
 Buttmann's works, that, had he possessed a sufficient knowledge 
 of the other languages of the Indo-Germanic family, especially 
 had he been acquainted with the Asiatic branch, the work which 
 we consider as incumbent upon the scholars of our age would 
 not be still unperformed. As it is, he was in spirit a comparative 
 philologer, and succeeding scholars must make his works, the 
 Grammar and the Lexilogus, at once the model and the ground- 
 work of their labours. 
 
 The object of this work, so far as it is confined to the ad- 
 vancement of our knowledge of Greek, is to give to the Grammar 
 and Lexicography of that language all the aid that may be 
 derived from the present state as well of Greek scholarship as of 
 comparative philology ; and for this purpose to combine in one 
 body a series of contributions to the better classification of the facts 
 of Greek grammar, and also to the explanation of those words 
 which appear most frequently and prominently in the best writers, 
 and the meaning of which is still doubtful or but half understood. 
 
 We proceed to set forth its object so far as it has reference 
 to the general philosophy of language. 
 
 * Much the same remarks had been made by H. A. Hamaker, in a 
 valuable work with which the Author did not become acquainted until 
 some years after the publication of his first edition {Akademische Voqr- 
 lezingen^ Leyden, 1835, p. 3): "hoe dikwerf heeft niet Buttmann in 
 dienzelfden Lexilogus zijne toevlugt moeten nemen tot gissingen en on- 
 waarschijnlijkheden, waar het voile licht der waarheid hem zou hebben 
 bestraald, zoo hij met de vergelijkende studie der Germaansche dialekten 
 en haar vasten grondslag, de aloude taal der Brahmanen, ware bekend 
 geweest?" Mr. G. Curtius, whose diligently concealed obligations to the 
 present writer have been noticed in another place, nearly copies the words 
 in the text in his recent work ( Grundzuge der Griechischen Etymologies 
 Leipzig, 1858). He says (p. 17): "P. Buttmann possessed the genuine 
 spirit of a discreet and keen-sighted philologer in such a degree that we 
 must deeply regret that he made no use of the rich treasures discovered 
 by Grimm and Bopp even in his life-time. He was just the man to have 
 diffused new light by this means." Again in p. 19: "We may still always 
 take him as our model in his mode of treating Greek words, especially 
 in what concerns the distinctions of signiiication." 
 
CHAPTER III. 
 THE PHILOSOPHY OF LANGUAGE. 
 
 41Etymology and Syntax — how they ought to be discriminated. 420riginal unity 
 of language, which is necessarily co-ordinate with humanreason. 43 The Book 
 of Genesis is in accordance with the results of philosophy in this respect; for 
 it teaches: 44 (1) that language is an endowment, and not an invention; 
 45 (2) that differences of language are the effect and not the cause of dis- 
 persion. 46 Monumental writing connected with idolatry. 47 Spiritual 
 abstraction favoured by alphabetic writing ; this was manifested in the highest 
 degree by the invention of printing. 48 Effects of literature on the structure 
 of language; syntax and prose. 49 Passage of language from a primary to a 
 secondary, and from this to a tertiary state. The latter presumes ethnical ad- 
 mixture as well as literary cultivation. 50 Degraded languages; these also 
 capable of literary cultivation. The Chinese an example. 51 Outline of lin- 
 guistic psychology. 52 Two elements of speech, («) the organizing, {h) the 
 material. 53 Abstraction and association. 54 Space and time. 55 Algebra. 
 56 Realism and nominalism. 57 Plato a nominalist. 58 Outlines of Plato's 
 dialectics. 59 He was opposed to ultra-nominalism. GO Design of his Cra- 
 tylus. 61 Home Tooke the modern representative of the school controverted 
 in Plato's Cratylus. 62 Philosophical design of the present work. 
 
 41 rpHOSE who have hitherto written on the philosophy of 
 language have generally fallen into one of two errors ; 
 — they have either omitted altogether the consideration of that 
 department which relates to the formation of sentences, or, what 
 is worse, they have failed to discriminate the two divisions of the 
 subject, and, conducting their etymological analysis on strictly 
 logical principles, have necessarily taken a perverted view of the 
 nature and object of their inquiries. In the present work we 
 have endeavoured to remedy this defect, by showing that the 
 resolution of a sentence into its elements is a totally different 
 process from the analysis of those elements themselves — that in 
 a scientific investigation of the general speech of man our prin- 
 cipal concern is with the word, its structure and development ; 
 that the same causes which create syntax, or logical sentences, 
 tend to corrupt and destroy the original forms of speech, so that 
 the attempt to derive the elements of the word from the elements 
 of the sentence is absurd, as seeking the whole in its part, and 
 must lead to conclusions utterly false and contradictory. 
 
 A formal discussion of the philosophy of language attempts 
 the solution of two problems; —it purposes to ascertain, first, the 
 
 F 
 
66 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 origin of language ; and secondly, the connexion of our words 
 with our thoughts. But, although this may be adopted as a 
 methodical division and for form's sake, the two questions, ac- 
 cording to our view of the former of them, are in fact one and the 
 same ; for, if language is, as we have no doubt it is, a necessary 
 result of the constitution of man as a rational being, if the gifts 
 of reason and speech are necessarily co-ordinate, then there must 
 be no discussion, but simply an explanatory statement, with 
 regard to the connexion between language and mind. 
 
 42 The primitive state of mankind has been a favourite 
 subject of inquiry both in this country and on the continent, and 
 some theory of the origin of human speech has generally formed a 
 part of such disquisitions. Tillthe introduction of the comparative 
 study of languages the theorists wanted their facts, and therefore 
 met with the fate of those who advance unsupported hypotheses — 
 they did not arrive at any convincing results. The researches 
 of the present century, hoAvever, have given an entirely new turn 
 to this subject; the right method has been adopted, and it is 
 this, — that the only safe conclusions, with regard to the primi- 
 tive condition of language, are to be derived from a rigorous 
 scrutiny of all the various forms which it exhibits in its existing 
 state; and though philologers have not yet examined all the 
 dialects of the ΛνοΓΜ in a complete and scientific manner, they 
 have advanced so far as to be able to divide them all into a few 
 great families, and have moreover examined the difiPerent mem- 
 bers of the class, to which our own language belongs, with a 
 minute accuracy which leaves little to be desired : the facts with 
 regard to this class have not only been carefully collected, but 
 also scientifically arranged, so that the utmost reliance may be 
 placed upon any conclusions logically deduced from them: and 
 from a comparison of this family (considered in its unity, which 
 is thus established), with the other great classes of the general 
 language of mankind, a comparison guided and illustrated by 
 sound psychological views, the most profound and highly-gifted 
 of those philosophers who have devoted themselves to this study 
 have inferred, that language is the necessary and spontaneous 
 result of man's constitution, that human speech and human na- 
 ture are inseparable, and consequently that language was origi- 
 
CHAP. III.] THE PHILOSOPHY OP LANGUAGE. 67 
 
 nallyone*: physiology has made some important approximations 
 to a similar result with respect to the bodily structure of man- 
 kind f; and thus external probability leads us to the conclusion, 
 that the varieties which we distinguish as well in the form as in 
 the language of man must have been produced by the dispersion 
 of the human race from some one home over the whole surface 
 of the earth, and by the subsequent operation of the multifarious 
 causes to which the different parts of the separated family would 
 be exposed. 
 
 The result of investigations of this nature is generally more 
 satisfactory to our inquisitive spirit than any written testimony, 
 however authenticated, with regard to the creation and early 
 state of man: for the facts, to which such a testimony relates, 
 occurred long before the invention of writing; they are traditions 
 which appear either as scattered fragments floating on the stream 
 of time, or as remnants of the abandoned ark incorporated by 
 more recent speculation in some pretended semblance of the ori- 
 ginal structure. When, however, we find that the most distinct, 
 and perhaps, in its original form, the oldest of these traditions 
 agrees exactly and entirely with the result of our anthropological 
 studies so far as we have been able to prosecute them with 
 
 * William von Humboldt, the most eminent of those who have made 
 the philosophy of the word their study, has stated his opinion to this effect 
 in the most explicit terms. In a paper in the Berlin Transactions for 
 1820-1 (Werke, vol. in. p, 252), he expresses himself as follows: "Ac- 
 cording to my fullest conviction speech must be regarded as immediately 
 inherent in man; for it is altogether inexplicable as the work of his under- 
 standing in its simple consciousness. We are none the better for allowing 
 thousands and thousands of years for its invention. There could be no 
 invention of language unless its type already existed in the human under- 
 standing. In order that man should understand a single word truly, not 
 as a mere perceivable utterance, but as articulate sound denoting a con- 
 ception, he must have already in his head the whole connexion of speech. 
 There is nothing individual in speech; every one of its elements an- 
 nounces itself as part of a whole. Natural as the belief in a gradual 
 formation of speech may appear, the invention of it could only happen 
 at once. Man is man only by means of speech; but in order to invent 
 speech he must be already man." 
 
 -}- See Dr. Prichard's Researches into the Physical History of Mankind, 
 and Dr. Wiseman's Third and Fourth Lectures. 
 
 F 2 
 
68 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 safety, the most exclusive votary of inductive science will hardly 
 refuse to welcome such a confirmation of his independent re- 
 searches. Every genuine philologer, who has adequately studied 
 the subject, must of course reject as untenable the claim of infal- 
 libility, which has been made on behalf of a certain branch of 
 Semitic literature ; every honest man must indignantly protest 
 against the mendacious sophistry with which those pretensions 
 have been supported ; and every sincere follower of Him, who 
 both taught and lived a religion of loving-kindness, must shrink 
 with loathing from the narrow and malignant passions inevitably 
 fostered by bibliolatry. But though the philosophical scholar 
 knows that the noblest records of ancient learning are but the 
 products of that divine genius which God has given to man; 
 though he cannot think that any old book is a ^ίοπετ eg "άγαλμα, 
 which fell in perfect form from heaven, and should therefore be 
 regarded with the homage of unquestioning and slavish idola- 
 try*; though he can see that even the most venerable relics of 
 the past are completed by additions of later growth, which have 
 given an external finish to what was but a fragmentary torso ; 
 still he can look upon these time-honoured remains with a chas- 
 tened respect, which is as far removed from irreverent deprecia- 
 tion as it is from irrational enthusiasm, and he can listen, like a 
 disciple, to the words of ancient sages, without imagining that 
 he hears the unearthly utterances of an oracle. To him, who 
 cultivates classical and sacred learning with the one object of 
 discovering the truth, it matters little that the Hebrew Penta- 
 teuch is really a compilation from a greater or less number of 
 older books, and received its present form from an editor who 
 was well acquainted with the traditions of the Babylonians and 
 Assyrians among whom he lived in exile; he is not disturbed 
 by the reflexion that in all probability the chronicles of Berosus 
 andSanchoniathon,if theyhad come down to us complete, would 
 have rendered articulate and perspicuous much that is now ob- 
 scure and enigmatical. He thankfully accepts what has been 
 saved from the wreck of time, and while he applies the resources 
 
 * The cry μ^γάΚη η "Αρτεμις ΈψΒοίων (Acts xix. 28, 34), is really a 
 representative of the only argument by which the corresponding hypo- 
 thesis is maintained in modern times. 
 
CHAP. ΠΙ.] THE PHILOSOPHY OF LANGUAGE. 69 
 
 of his criticism to discover and eliminate the authentic ingre- 
 dients and to interpret the true meaning of the restored docu- 
 ments, he feels that he has enough of internal evidence and 
 extrinsic confirmation to justify*his belief, that the books which 
 still bear the name of the lawgiver of Israel contain the resi- 
 duary substratum of those ancient and venerable traditions of 
 the Aramaean race, which descended by an unbroken chain from 
 the primitive and highly-favoured men who still enjoyed the 
 higher inspiration of an unsophisticated nature, or, as the Jewish 
 allegorist expresses it, heard the voice of Jehovah Elohim, as it 
 floated to and fro on the evening breeze*. 
 
 43 Now the results of our philosophy are as follows. We 
 find in the internal mechanism of language the exact counterpart 
 of the mental phenomena Avhich writers on psychology have so 
 carefully collected and classified. We find that the structure of 
 human speech is the perfect reflex or image of what we know of 
 the organization of the mind : the same description, the same 
 arrangement of particulars, the same nomenclature would apply 
 to both, and we might turn a treatise on the philosophy of mind 
 into one on the philosophy of language, by merely supposing 
 that every thing said in the former of the thoughts as subjective 
 is said again in the latter of the words as objective. 
 
 From this we should infer, that if the mind of man is essen- 
 tially and ultimately the same, — in other words, if man, where- 
 ever he lives, under whatever climate and with whatever degree 
 of civilization, is still the same animal, — the only reasoning and 
 discoursing animal, — then language is essentially the same, and 
 only accidentally different, and there must have been some com- 
 mon point from which all the different languages diverged, some 
 handle to the fan which is spread out over all the world, some 
 first and primeval speech; and that this speech was not gradu^ 
 ally invented, but necessarily sprung, all armed like Minerva, 
 from the head of the first thinking man, as a necessary result 
 and product of his intellectual conformation. 
 
 But it is clear that the mind of man is essentially and ulti- 
 
 * Genesis iii. 8. See Kennicott, Two Dissertations^ Oxford, 1747, 
 p. 47, note k. 
 
70 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 mately the same in kind. Whatever may be the form of his 
 features or the colour of his skin, man is everywhere distin- 
 guished from the rest of animated creation by the godlike faculty 
 of reason; and the scarcely less godlike attribute of speech is 
 enjoyed by all human beings, from the broad-browed European, 
 who speculates upon the high things of heaven, to his woolly- 
 haired brother, who leads an unreflecting life in the arid plains 
 of Africa. But this is not all. As the only combination of the 
 two lives, — the animal and the spiritual, — as the only veritable 
 amphibion in the world*, as the only union of immortal mind 
 with corruptible matter, as the only being gifted with the co- 
 ordinate faculties of reason and speech, man is not merely one; 
 he stands alone among living creatures. And jealous for the 
 dignity of his species and proud in the consciousness of exclusive 
 privileges, he is led by his philosophical instinct to reject with 
 disdain any attempt to classify him with the animal tenants of 
 this lower world; and, that he may make the line of distinction 
 between himself and them more definite and palpable, he claims 
 for himself an aboriginal unity, and traces back the pedigree of 
 his scattered families to one common ancestor and to one com- 
 mon home. It is true that there are great outward bodily 
 differences between the different races of men, and that there 
 have been found some advocates for materialism, who ignore the 
 spiritual indications of unity, and deny the claim of the inhabit- 
 ants of Africa to rank with Europeans as the same animal. But 
 a more enlightened research has triumphed over all these diffi- 
 culties, and it is now seen that the physical differences of the 
 races spread over the earth's surface are explicable from secon- 
 dary causes, on the hypothesis of a primeval identity of origin, 
 and a subsequent dispersion of emigrants from the home of their 
 family ; and that we may account in the same manner for those 
 differences in intellectual development which correspond to the 
 physical differences of nations. 
 
 These are the results of our philosophy. And for the sake 
 of that union between ancient tradition and inductive reasoning, 
 which the philologer is always anxious to maintain, but which 
 the insane advocates of an infallible literature are perpetually 
 
 Sir T. Brown, Religio Medici, xxxiv. 
 
CHAP. III.j THE PHILOSOPHY OF LANGUAGE. 71 
 
 endeavouring to disturb*, we will step aside to show, that, in 
 this case at least, there is no conflict of authorities; that here 
 there is no occasion for mutual concessions, or a feigned recon- 
 ciliation; but that the Book of Genesis itself describes the 
 speech of man as an endowment, not as an invention, and attri- 
 butes the differences of human language to a constrained emi- 
 gration from the aboriginal settlement of our race. 
 
 44 "And out of the ground," says the ancient Jehovist, 
 "the Lord God formed every beast of the field and every fowl 
 of the air ; and brought them unto Adam to see what he would 
 call them: and whatsoever Adam called every living creature, 
 that was the name thereof. And Adam gave names to all cattle, 
 and to the fowl of the air, and to every beast of the field; but 
 for Adam there was not found an help meet for him" {Genesis^ 
 ch. ii. vv. 19,20). These words imply that the power of speak- 
 ing merely, and not language, was given to man, and therefore 
 there are no grounds for the inference which a modern writer 
 would draw from the passage, that the language of Adam was 
 an immediate revelation from the Divinity f. According to the 
 plain construction of the passage, aboriginal man was so con- 
 stituted that he had the power of speech, and this power he 
 exercised first of all in giving names to the different species 
 
 * Mirabeau has well remarked: "Voila ce que nos theologiens igno- 
 rans et vains devoient nous apprendre. Le grand art est de lier toujours 
 la science de la nature, avec celle de la theologie, et non de faire heurter 
 sans cesse des choses saintes et la raison, les croyans fideles et les pbi- 
 losophes." 
 
 •{• Quand on lit dans la Genese, que toutes les creatures passerent en face 
 de Dieii devant Adam, qui leur imposa des noms, espece de bapteme de Vesprity 
 qu'il leur con/era on οοηςοϋ pourquoi, dans le systeme des philosophes de 
 V Orient, qui est aussi celui de Pythagore et de Platan, Vhomme est envisage 
 dans son origine comfne un second createur, comme tin verhe incarne e'voquant 
 au moyen du discours les mysteres de la creation; espece de Mage en rapport 
 avec le monde ideal et le monde terrestre, avec la nature et Dieu. Oest le 
 langage primitif des hommes qui est Varbre de la science, c^est la, comme le 
 disent les livres Indiens, le Veda Celeste; et quelque corruption que les infir- 
 mites de notre nature y aient introduite par dans son essence, c^est une revela- 
 tion de la Divinite meme {Le Catholique, Tom. i. p. 418, quoted by C. J. 
 Sachs, de statu generis humani originali, p. 19 sq. Berol. 1831). 
 
72 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 of animals ; but, says the historian, although he had this power 
 of speech, he had no one to converse with, no one to share his 
 sympathies, no one gifted like himself with the wonderful powers 
 of reason and speech, no help meet for him, among all the 
 numbers of animated creation which thus passed in review be- 
 fore him ; and so God created him a wife* as the partner of his 
 joys and sorrows, and the living echo of the thoughts to which 
 he gave an utterance in spontaneous articulations. 
 
 This short passage actually contains the outline of all that 
 philosophy and philology have taught us of the probable origin 
 of language. The ultimate results of human consciousness are, 
 that the thinking subject *s, and that there is something without 
 him; that there is, in the language of the German philosophers, 
 a me and a not -me ^ or, if you will, he knows that he himself 
 exists, and believes that there is something which is not himself. 
 In those two results of all consciousness, in the consciousness of 
 self and of not-self, is comprehended all the world as it exists 
 for the individual. In the former are included all the thoughts, 
 feelings, impressions and ideas which a contact with the outward 
 world and the consequent sensations produce upon the thinking 
 subject. But the human mind is naturally impatient of pure 
 thought: it strives ever after objectivity, and endeavours to com- 
 plete and fix its inward conceptions by some species or other of 
 outward manifestation ; the thought comj)letes itself in the ex- 
 pression. Even if a man were placed alone in the world with 
 all the faculties which he now enjoys, he would give names to 
 the different objects of animal creation as they passed in review 
 before him, he would seize upon some one prominent attribute 
 
 * The manner in which the woman is created from the man in the 
 Jehovisfcic account — for the Elohist, or, as we think, subsequent editor 
 assumes that the human race was created in pairs like the other animals 
 (Gen. i. 27) — belongs to a class of allegories, found in most nations, of 
 which we have the Greek philosopheme in Plato's Srjmpos. p. 191, and a 
 separate application in the myth that Minerva sprang from the head of 
 Jupiter, as Eve here springs from the heart of Adam. Those, who 
 adopt the literal interpretation of the passage, may read, if they please, 
 how two of their own prophets have explained the flesh which was sub- 
 stituted for the flesh and bone taken from the side of the Autochthon 
 (see Hofmann, Weissagung und Er/ullung, i. pp. 65 sqq., and Delitzsch, 
 Genesis, p. 113, who takes a somewhat diflferent view). 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 73 
 
 in each class and mark it by a name of distinction*. This 
 name he would no doubt express by that which is the only 
 natural and obvious method, namely, by articulate sound. But 
 if such an effort of language might be expected from a solitary 
 man, it would be the inevitable consequence of his meeting with 
 some other thinking and^speaking being ; he would then neces- 
 sarily seek to transfer his thoughts to that outward objective 
 world which was cognizable to his fellow as well as to himself, 
 by the most natural and obvious method, which is, as we have 
 said, by articulate sound ; and if there were at first but two 
 such persons in the world, their communications, regulated by a 
 convention based upon a community of reason and necessitated 
 by a community of wants, would constitute the first language, 
 and, by transmission, the language of all mankind. 
 
 45 The same striving after outward expression, which ne- 
 cessarily produced spoken language, as its primary effect, led in 
 the course of time to the invention of letters or writing as a more 
 durable manifestation of the thoughts , which Λvas , however, 
 strictly artificial, and must therefore be carefully distinguished 
 from the natural language which necessarily preceded it. The 
 first writing was not alphabetical; each symbol was an inde- 
 pendent and significant term, and the huge masses of stone 
 which they set up for monuments, the walls and temples which 
 they built, and the rude figures which they carved and painted 
 upon them , were each and all of them distinct words. The 
 pyramids, arches, and obelisks on which the traveller still gazes 
 with wonder, the ruins of Egypt, Babylon, and India, are not 
 merely, as a quaint writer f has called them, the irregularities of 
 vainglory, the wild enormities of ancient magnanimity ; — they 
 are the huge chronicles by which the men who built them tell 
 to posterity the wonderful history of their industry and of their 
 art, — the writing of a race of giants, traced with enduring 
 characters on the great page of nature, which neither the rage 
 
 * "Without looking to the communication between man and man, 
 speech is the necessary condition of the thought of the individual in 
 secluded loneliness." Humboldt, iiber die Verschied. des menschl. Sprb. p. 53, 
 t Sir T. Brown, Hydriotaphia, ad fin. 
 
74 THE PHILOSOPHY OF LANGUAGE. [bOOK L 
 
 of the elements, nor the passions of men, nor even the slow sure 
 hands of time haA^e been able as yet to convert into a palimpsest. 
 The primary impulse to these rude writings was a hankering- 
 after durability, a desire to leave a lasting memorial of their 
 history, which should at the same time serve as a rallying point 
 to their descendants. According to the sacred writings, they 
 wished to build themselves a city and a tower, and to make 
 themselves a name, lest they should be scattered abroad upon 
 the face of the whole earth : as a punishment of this attempt 
 the dispersion and consequent varieties of human language took 
 place {Genesis^ xi. 4 foil.). It does not in any way appear from 
 the words of the sacred narrative that the common language of 
 man was violently and suddenly broken up into a number of 
 different speeches or dialects. Indeed it has been more than 
 doubted by some of the most learned commentators whether the 
 confusion of tongues really means anything more than the sudden 
 manifestation of a spirit of dissension among those who were 
 previously united in a common design*. The words of Scripture 
 
 * This view of the passage was first proposed by Vitringa (Observ. 
 Sacr. Tom. i. p. 106), who is followed by Robertson (Clavis Pentateuchi, 
 pp. 93, 96) and opposed by Perizonius, Orig. Bahyl. c. ix., whose views are 
 adopted by Dathe and Rosenmuller. It 'is the obvious intention of the 
 writer of Genesis to make Babylon the scene and the starting point of the 
 dispersion of men. This is clear from the use of the particle a» v. 7, 
 which is emphatically repeated in v. 9, and from the etymology proposed 
 for the name Vaa, Bavel, which the writer connects with the root hhzy 
 "he poured forth," though the word would be more naturally explained 
 as Aj «.«jLj, Bab Belj =•. ^a asa, porta vel aula^ civitas (Beli Winer s. v.). 
 As Robertson rightly observes, all depends on the meaning of the words 
 n£iy and fi'^'^an in v. 1. He says: "etiamsi nrb significat aliquando 
 linguam, dialectum, ut in Jes. xix. 18; xxviii. 11; xxxiii. 19; Ezech. iii. 5, 9; 
 Ps. Ixxxi. 6, Scriptores fere sacri sermonis dialectum^ et linguam per γ•ΰ^ 
 exprimunt; iis in locis ubi omnibus constat de dialectis, non vero de 
 sermonum sententiis, agi. Vide Gen. x. 5, 20. Vox na- verti possit 
 ratio aeque ac sermo, uti λόγος apud Grgecos; hie igitur reddi possit tDi-nan 
 per ejusmodi sententias quibus inest consilium et deliberatio." But even 
 if we take the words literally, and consider ηεώ, "lip," a synonym of 
 •jS»^, "tongue," it is clear from Ps. Iv. 10, which perhaps contains an 
 allusion to this very passage, that according to the Hebrew idiom a dis- 
 traction of counsels might be spoken of as a division and confusion of 
 language: for David says with reference to Achitophel and his brother 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 75 
 
 are (vv. 7, 8): "Let us go down, and there confound their lan- 
 guage, that they may not understand one another's speech. So 
 the Lord scattered them abroad from thence upon the face of all 
 the earth ; and they left off to build the city." The plain con- 
 struction is, that as the offence of the Noachidae consisted in 
 their reluctance to emigrate, their punishment was the dispersion 
 which they sought to avoid ; and this dispersion might be, and 
 probably was, a cause of the difference of tongues, but could 
 hardly have been an effect of it ; for if any two sets of men had 
 a common object in view, they would not be long in finding a 
 medium of communication. The statement therefore in the Book 
 of Genesis is simply this ; that when the whole or a considerable 
 portion of the early inhabitants of the world were settled in 
 Mesopotamia, their attempt to contravene the decree of Pro- 
 vidence, that man should multiply over the whole face of the 
 earth, was punished with so complete a dispersion that large 
 gaps were left between the settlements of the different races, 
 and by the operation of secondary causes the languages of 
 the earth became different. On the supposition that the lan- 
 guage of man was originally one, it is hardly necessary, as 
 Niebuhr suggests*, to suppose also a miraculous divulsion. If 
 the dispersion took place at a time when man had not passed 
 beyond the pastoral condition, and it is clear from philology, 
 that even the Iranian branch was still in this condition at 
 the time when its tribes ceased to live together on their central 
 plateau, — it might have been occasioned by natural causes, 
 such as the want of pasturage for their increasing flocks and 
 herds, and would inevitably lead to a wide-spread and im- 
 mediate separation of the different families, who would go 
 
 conspirators: 'av\'dh 5^& "^a'-TX yW, "Swallow up, Ο Lord, divide their 
 tongue," where the root r^^s clearly points to the name 5^.?. given to 
 the Patriarch in whose time the dispersion is placed; Gen. x. 25. With 
 regard to v. 4, the phrase tj^tosa Vi5s''i is merely an exaggeration like 
 the epithet ovqavoμΎμηζ\ and Le Clerc has shown that aui, "a name," means 
 simply a monument or pictoral commemoration; cf. 2 Sam. viii. 13, and 
 Ennius, Annal. xvi.: 
 
 Reges per regum statuasque sepulcraque quaerunt, 
 
 iEdiiicant nomen, summa nituntur opiim vi. 
 * H. R. I. p. 53, Tr. 
 
76 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 farther and farther apart, under the influence of the exigencies 
 which led to their primary disintegration. As far, however, 
 as the statement in the Book of Genesis is concerned, the so- 
 called confusion of tongues resolves itself into a case of disunion 
 and dispersion ; and science admits that this is the most pro- 
 bable cause of physical and linguistic differences. 
 
 46 One of the earliest and most important results of that 
 hankering after outward manifestation, which produced the first 
 l-ude monument and picture-writings, was the introduction every 
 where of idolatrous forms of worship. The first great fact of 
 consciousness is, as we have seen, the existence of an external 
 world in contrast to the thinking subject. To this belief the 
 mind attains on the evidence of the perceptions excited and 
 called into being by contact with the world of sense. But 
 there is another belief, to which the reason comes almost as 
 soon on the evidence of its own reflexions, the belief in a superior 
 being who created the subject as well as the object of conscious- 
 ness, the great point of union to the two contrasted realities*. 
 It is, however, much easier to contemplate the common objects 
 
 * In stating that the belief in a creator and governor of the world 
 and of man is an instinctive and inevitable conviction, we do not mean 
 to question the necessity for a revelation to teach us the personal exist- 
 ence of the Deity, and the nature of sin and its remedy. We have pur- 
 sued this subject elsewhere {Christian Orthodoxy^ pp. 100 sqq.). Still the 
 revelation must be within the limits of our intellectual capacity, and 
 must be capable of approving itself to our reason and conscience. 
 Whatever is unintelligible is simply not revealed. Whatever runs 
 counter to our moral instincts is ipso facto disallowed as not of divine 
 origin. Mr. Hansel's examination of this question {Limits of Religious 
 Thought^ Oxford, 1858) is utterly inconclusive. He is a dexterous logi- 
 cian of the Hamiltonian school, and is well read in certain departments 
 of metaphysical literature; but he has yet to learn the first elements of 
 theology, and he seems to be quite unable to appreciate the results of 
 biblical criticism, even if they are not unknown to him. He writes as if 
 the adversaries of bibliolatry were not driving with loosened reins over 
 the flimsy ruins of the structure which he regards as still intact. In re- 
 ferring to the present writer (note 13, p. 421), Mr. Mansel perverts and 
 misrepresents an argument, which he would perceive to be the same as 
 the only tenable portion of his own position, if he could only grasp the 
 consequences to which his reasonings inevitably conduct the intelligent 
 readers of his book. 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 77 
 
 of consciousness, than to think of and regard unceasingly this 
 reality of the reason. And thus, impatient of abstraction, the 
 reasoning being gives an outward manifestation to this as well 
 as to his other thoughts ; he writes God on the world as he 
 wrote other things, with picture and statue imitations, and ere 
 long worships the type instead of the reality; he falls down on 
 his knees before a mere memoria technica^ he pays homage to 
 an object of sense, forgetful that the essence and definition of 
 God, his own idea of a supreme being, is, that he is some- 
 thing without the subject, which is yet not a part of the ex- 
 ternal world. Picture-writing, and indeed all the arts, are but 
 so many different indications of that feeling which gave rise to 
 the worship of images ; they are all different species of idolatry, 
 different symptoms of man's aversion to abstract thought, of his 
 love of dresses and disguises, of the imphilosophical tendencies 
 of his lower nature. For what is philosophy but an undressing 
 of the world*? It is to deprive our thoughts of all those 
 outward veils and vestments in which they are generally too 
 prone to wrap up the objects of their contemplations, it is to 
 strip eternity of the robe of time , to divest existence of the 
 accident of mortality, to let fall the many-coloured cloaks of 
 individuality, in which the genus is enveloped, to see the soul 
 unclothed and unencumbered with that garment of flesh which 
 weighs it down to earth, and brings it to the near contact of 
 death and decay. All this is difficult to the untutored intellect ; 
 as difficult as to gaze on the noon-day sun without a cloud, or 
 a mist, or even so much as a bit of coloured glass, to break the 
 intensity of his light f; and yet it is what we ought to do, what 
 we must do if we would live as creatures that enjoy reason and 
 hope for immortality. 
 
 47 The invention of alphabets, or of writing, in the modern 
 
 * "The beginning of all Wisdom is to look fixedly on Clothes, el- 
 even with armed eye-sight, till they become transparent^' {Sartor Re- 
 sartus, p. 66; see also pp. 74, 210). 
 
 f Plato, Sophist, p. 254 a: ό δε γε φιλόΰοφος , rij τον υντος άεΐ δια 
 λογιαμών προϋκείμενος ίδέα^ διά το λαμττρον αύ της χώρας ουδαμώς ενπετης 
 οφΟ'ηναυ' τα. yaq της τά.ν πολλών ιρυχης όμματα καρτερεΐν προς το &εΙον 
 αφορώντα αδύνατα. 
 
78 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 sense of the word, was the first step towards the overthrow of 
 idolatry; and it is a remarkable fact, that Europe owes her alpha- 
 bet to the only nation, which, in the remote ages, preserved itself 
 to any considerable extent from the worship of symbols : but still 
 it was only a partial remedy; for books, those sworn foes of 
 all idols, of all worship of the world of sense, were but few, and 
 even "the old man eloquent" preached his mellifluous wisdom 
 to a small audience. It is true that wherever they went they 
 were fraught with a real vitality; they sped like the knights 
 errant of old, releasing many an oppressed mind from captivity, 
 and here and there lighting the lamp of truth in a land of dark- 
 ness ; yet their influence was very limited, and even after the 
 Christian religion had appeared, causes, which are well known, 
 operating with it, crumbled the old fabric of civilization into 
 minute fragments, and the mind of man was again a worshipper 
 of images and of art. At length came the invention of printing, 
 the most important event, perhaps, in the history of the world. 
 From this time forth, the book was not a solitary hero, a Her- 
 cules or a Theseus, striving for the liberation of men from the 
 giants who tyrannized over them. Their name was legion; in 
 infinite hosts they spread themselves over the world, conquering 
 and to conquer. First of all, the idolatry of popery fell before 
 them ; then art as the instrument of idolatry was overthrown ; 
 philosophy was by them reinstated in her rightful dominion ; 
 philology came forth as her hand-maid ; feudality and tyranny 
 gave way to their victorious march, and to this day the despots 
 of the world tremble before them. Let us not be deceived : 
 Luther was great, butGuttenberg was still greater*. The letter 
 
 * Luther himself called the invention of printing "das letzte Auflo- 
 dern vor dem Erloschen der Welt" (Falkenstein, Gesch. der Buchdrucker- 
 kunst, Vorrede, p. 1). The reader will find in Victor Hugo's Notre- Dame 
 de Paris (Lib. v. eh. 2, in the chapter headed, eeci tuera cela — le livre 
 tuera Vedifice)^ some remarks on this subject, written in a wild and ex- 
 travagant tone, as is the wont with modern French authors of that 
 school, but yet very striking and true. A more recent French writer 
 has taken up the same theme quite in the spirit of the text above, which 
 was written 20 years since, "lloger Bacon," says Arsene Houssaye {le 
 Roi Voltaire, 1858, p. 13), "qui pile dans sa cellule le soufre et le sal- 
 petre, servira les haines religieuses qui donnent la fievre a Voltaire ; mais 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 71) 
 
 did not kill but gave life, for it was by the letter that the spirit 
 lived again, and it is the aid of the letter, it is philology, in one 
 of its forms, which we must have recourse to whenever we would 
 struggle with those idols of the forum*, the realized ideals that 
 ever and anon usurp the throne of reason, and tyrannize over 
 the misnamed free-will of man f. 
 
 48 From these reflexions on the influence produced upon 
 literature, and, through it, upon the opinions of the world, by a 
 greater facility of writing, and an unbounded circulation of writ- 
 ten documents, we come naturally to consider the effects of the 
 invention of writing on the spoken language of which it is the 
 secondary expression. The art of writing was a mere invention ; 
 it stands on precisely the same footing with the other useful arts : 
 hence it was at first rude and uncouth; and as it did not for a 
 long time arrive at perfection, or become so easy as to fall into 
 general use, it has produced, by its want of completeness, great 
 and lasting evils on the mind of uneducated man. But language 
 was a spontaneous result of our organization, and thus, like every 
 production of nature, was as perfect at the beginning, indeed 
 much more so, than it is now, when literature or the written word 
 has developed itself in a thousand different ways. If any one 
 thing more than another can show the absurdity of those who 
 speak of an invented language, it is simply this fact, that the 
 oldest languages are always the richest in materials, the most 
 perfect in analogy, the most uniform in etymological organization. 
 Philology too instructs us that those very words, which the 
 believers in an invented language regard as the most difficult to 
 invent, and therefore as the last introduced, are, in fact, the basis 
 of all language ; for instance, the pronouns and numerals, which 
 Adam Smith J considers of recent introduction, are knoΛvn to 
 have been the very oldest part of every tongue; for it is just 
 
 Gutenberg va donner les armes a la raison. L'Evangile de Voltaire va 
 courir sur le monde comma si les millions d'oiseaux I'emportaient sur 
 leurs ailes: I'imprimerie tuera la poudre. Ceci tuera cela." 
 * Bacon's Novum Organon, Lib. i. Aph. 59, 60. 
 -j- Carlyle's Hist of the French Revolution, i. p. 13. 
 % Considerations concerning the Formation of Languages, at the end of 
 the Theory of Moral Sentiments (Vol. ii. p. 431). 
 
80 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 these words which retain their identity in languages which have 
 been longest separate, and have therefore become most unlike in 
 other particulars. The effect of increased use upon the structure 
 of inflected language is rather to weaken and corrupt than to im- 
 prove or amplify; and it may be laid down as a general rule, 
 that, as such languages remove themselves from their origin, the 
 love of what is called euphony gains ground more and more, the 
 elements or roots are no longer clearly discriminated from the 
 terminations, and the meaning of the separate parts of the word 
 becomes less distinguishable, till at last all inflexion is super- 
 seded by a system of prefixes and auxiliaries. The monosyllabic 
 languages, which are the most imperfect of all, appear to be 
 degenerated forms of older and more complete idioms. It may 
 seem strange, but it is nevertheless true, that this corruption of 
 the forms of language has arisen not in spite of, but directly in 
 consequence of literature; and the invention and diffusion of 
 writing, which have produced such important results in literature 
 and science, and, through them, on the general mind of man, 
 have thus contributed to undermine the mighty and perfect 
 structure of spoken language, the immediate production of that 
 reason of which writing is so important an instrument. This 
 has not been generally remarked*, and it will be worth our 
 while to bestow a little consideration upon it. 
 
 The beginning of literature has been prior to the beginning 
 of writing in all those countries in which literature has subse- 
 quently attained its greatest development. As the want of 
 writing materials necessitates the adoption of metre, the first 
 composition in every language is poetry. Had the invention of 
 writing and printing been coeval with the first beginnings of 
 language, we should certainly never have had an epic poem, 
 perhaps never a line of poetry in the world. Besides, there 
 appears to be something in the nature of early man, full as he is 
 of sublime inquiry, and impressed with the wonders of the earth 
 and the sky, which he gazes on with awe and veneration, that 
 
 * It has been noticed, however, that writing could never produce any 
 change upon the spoiten language otherwise than through the literature. 
 (See A. A. E. Schleiermacher, de ^Influence de VEcriture sur le Langage. 
 Darmstadt, 1835, p. 101). 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 81 
 
 leads him on to poetry. The language of the old world streamed 
 freely from the breast, swelling with infinite redundancy of ex- 
 pression, replete with the richest and most significant compounds, 
 and ever bursting forth into song. " We may suppose ," says 
 William von Humboldt*, "that there was hardly in any desert 
 a wandering horde which had not its lays. Man, as a species, 
 is a singing animal, connecting, however, thought with his 
 melody." The sage, who discoursed to his disciples on the 
 mysteries of man and the world, set before them "the sweet 
 food of sweetly-uttered knowledge f," and the chronicler, who 
 wished to perpetuate the past deeds of his warrior-race, sang to 
 the harp the verses he had composed. Prose can only arise after 
 a long period of civilization, when writing has become tolerably 
 easy, and writing materials sufficiently abundant J; it keeps pace 
 with the logical or syntactical development of a language ; so 
 that writing, which can produce no efi'ect in the way of improve- 
 ment on the forms of a language, exercises a most important 
 influence on the construction and connexion of its sentences, and 
 therefore on the science of the people who use it. The method 
 of language gains at the expense of its materials. It is observ- 
 able that the first literary productions of a nation, their epic 
 poems and lyrical hymns, are either entirely devoid of syntax or 
 but inadequately provided with it. In the earliest poems of the 
 Indians, for example, the Ramayana and Maha-Bharata, and 
 still more in the hymns of the Rig- Veda, there is no syntax or 
 construction properly so called §; and as we do not know to what 
 extent prose composition was cultivated by the Brahmins while 
 the Sanscrit \vas still a living language, we cannot say how far 
 they ever arrived at a logical syntax. And this of itself is some 
 argument for the inference, which is strengthened by other con- 
 siderations, that the Brahmins had no alphabet when they brought 
 
 * tlber die Verschiedenheit d. menschl. Spi-b. pp. 59, 60. 
 t Sir P. Sidney, Defence of Poesy, p. 495. 
 "^ Scripturam tentare et communi usui aptare plane idem videtur fuisse, 
 atque prosam tentare et in ea excolenda se ponere. Wolf, Prolegomena ad 
 Homerum, p. 72. 
 
 § By syntax we mean the strictly logical conformation of sentences, 
 including that accurate discrimination of subject and predicate, to which 
 the article is more or less necessary. 
 
 G 
 
82 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 their Vaidic hymns into Hindostan, but that they borrowed the 
 Phoenician writing, which they found already in use among the 
 earlier Arian inhabitants of northern India, and subsequently 
 improved and extended it, until it assumed that most elaborate 
 and ingenious form, which is known as Diva-nag art*. In the 
 Greek literature, however, we possess an excellent specimen of a 
 language developed through all the successive stages, from the 
 rude luxuriance of the Epos to the careful but barren elegances 
 of logical prose ; for Hellenism, after it had secured its predomi- 
 nance over the Pelasgian or older element, was subjected to no 
 external interference; its changes and progressions took place 
 within itself; and it may therefore be cited as a good example 
 of the influence of literature and civilization on the syntax of a 
 pure and higly-cultivated idiom. In this language, before all 
 others, we see the article, that great implement of logic as dis- 
 tinguishing the subject from the predicate, disunite itself from 
 the pronoun or general designation of locality; in this language 
 we see the differences of mood developing themselves from dif- 
 
 * On the relations between the Brahmins and other Arians of India, 
 see below, § 84. And for the different opinions entertained respecting 
 the Indian alphabets, see § 105. Strabo tells us, on the authority of 
 Nearchus, a contemporary of Alexander (ρ.716ϋ): τονξ νόμους άγραφους, 
 είναι τους μ\.ν -ΛΟίνοΰς, τους δ\ Ιδίους — and on the authority of Megasthenes, 
 who was resident for Antiochus at the court of Chandragupta (Sandro- 
 eottus), he mentions (p. 707 b): ουδέ ya.Q '{ράμματα είδεναι αυτούς, αλλ' 
 από μνήμης εκαατα διοιχεΐο^^αι. Now as the name of Antiochus actually 
 occurs in the inscriptions of A9oka as Ant'iayako Yona raja, "Antiochus, 
 king of the Greeks" (Prinsep's Essays, ii. p. 15, ed. Thomas), these 
 statements must refer to the Brahminical as distinguished from the 
 popular language; and there are other reasons for concluding that the 
 Vedas were taught orally, and retained in the memory without the aid of 
 books. It is expressly stated that krama-patha is smritiprayojanah, i. e. 
 "quod memoriam adjuvet in ediscendis ac tenendis hymnis Vedicis." 
 See Weber, Ind. Skizzen, p. 133. Bunsen carries back the grammatical 
 studies of the Indians and the formation of prose to the middle of the 7th 
 century b. c. The following is his table of dates (^'^y/)ien,V,a,pp.204,5): 
 
 B.C. 
 
 Panim . 350 
 
 Yaska's Nirukta, or interpretation of the Vedas . . 450 
 
 The Pratisakya ....... 550 
 
 Older grammarians ...... 650 
 
 This table rests upon conjecture rather than induction or evidence. 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 83 
 
 ferences of tense, and all the syntactical modification of the 
 subordinate or accessory verb expressed by the participle, an 
 etymological modification of the verb; in a word, the Greek 
 language, like Mahomet's coffin, which hangs between earth 
 and heaven, has taken the middle place, between the synthetic 
 and analytical languages, combining the perfection of the word 
 with the regularity of the sentence, to a degree which no other 
 idiom can parallel. The language of Homer is totally different 
 from that of the later poets, and although his lines are not so 
 devoid of logical structure as the glokas of the Indian poems, 
 (and this is perhaps explicable from the fact that our present text 
 of the Iliad and Odyssee is little more than a rifacimento of the 
 original works), it is still obvious on the most hasty perusal that 
 the logical structure of sentences, for which prose Greek is so 
 remarkable, had not yet established itself in the language. The 
 same is also evident from the old Attic prose of Thucydides, 
 which is full of what we should call bad grammar, arising of 
 course from his inability to correct and polish his style by writh- 
 ing his sentences over and over again. Thus we often find that 
 he has forgotten at the end of a sentence how he commenced it, 
 or has purposely changed the construction, without being able, 
 from want of facility in the mechanical part of writing, to re- 
 touch the beginning of the period. When Plato and Demos- 
 thenes flourished, the materials and habit of writing must have 
 improved wonderfully, as we may infer from the correctness and 
 polish of their style ; it is stated, too, that the former used fre- 
 quently to rewrite his works, and that a tablet was found after 
 his death in which the words at the beginning of the Republic 
 were transposed in a number of different ways* ; and we are told 
 that Demosthenes copied out the history of Thucydides eight 
 times f . Even in the Attic dramatists there is a great difference 
 between the construction of the choruses, written after an old 
 model, and the more prosaic dialogue ; and yet this last is by 
 no means so conspicuous for that discrimination of the subject 
 and predicate by which the prose works are distinguished. 
 
 * Quintilian, viii. 6, § 63; Dionys. Halic. De Compositione Verbormrif 
 p. 208, Reiske. 
 
 jf Lucian, adv. indocium^ p. 102. 
 
 G2 
 
84 THE PHILOSOPHY OP LANGUAGE. [bOOK L 
 
 Whenever a language has once arrived at a full syntactical 
 development, this distinction between prose and verse ceases to 
 exist; the constructions in poetry then possess the same logical 
 exactness as those in prose. But the Attic idiom, though pro- 
 gressively approximating to this state, did not attain to it till 
 the time of Xenophon and Plato, the latter of whom gave the 
 first hints of the proper analysis of the sentence*, which he 
 could not have arrived at had not the Greek language been by 
 that time capable of logical prose : for in order that the theory 
 of syntax may be discovered, the language itself must have be- 
 come syntactical. As Plato discovered this theory from the 
 logical texture which his own language had assumed, conversely 
 Aristotle, when he had formally and methodically set forth the 
 principles of the Platonic analysis of the sentence, adopted his 
 own style to this method; and thus he is not only the great 
 expounder of the method of language, but also the most metho- 
 dical of writers ; a circumstance which has induced an eminent 
 author f to compare his style to a table of contents. Thus we 
 see that the history of Greek literature exhibits the develop- 
 ment of a language originally the most copious into one con- 
 fessedly the most syntactical, one in which the discovery of logic 
 or of the principles of syntax was first made. And the wonderful 
 fact about it is, as we have said, that it should have arrived at 
 this ultimate state with a smaller sacrifice of its original form 
 than any other language in the world. In general, however, it 
 may be laid down that languages fall ofi" in perfection of form as 
 they gain in distinctness of literary composition, and that the 
 same causes which destroy the symmetry and regularity of the 
 structure of words, as a product and counterpart of the mind, 
 promote the efficacy of language, as an instrument of science. 
 Examine the analytical languages of modern Europe — our own 
 for instance ; you will find that in the arrangement of their 
 words in sentences they are absolutely confined to the logical 
 method. And what is the state of their etymological structure? 
 In the English language we have no distinction of genders by 
 means of inflexion, no declension, no facility of forming com- 
 
 * See below, chap. "vi. 
 The poet Gray (Letter to Dr. Wharton, Dec. 1746). 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 85 
 
 pound words, and but a few fragments of the Anglo-Saxon con- 
 jugation. In fact, the most perfect language for the purposes of 
 deduction would be one, the words of which have no individual 
 signification, but are merely general symbols; for the method 
 of language, as we have before observed, is independent of any 
 particular language; but as such a language can exist in writing 
 only, it follows that writing must have an important influence on 
 science. And this we know to be the case : for it is clear that 
 the greatest advances in science have always been preceded by 
 some great improvement in written language, whether it be the 
 step from picture-writing to the alphabet, from the rude manu- 
 script to the printed book, or from the abacus to Algebra. 
 
 49 These considerations lead us to expect in the languages 
 of all nations in which we find an early use of writing and an 
 early cultivation of pure literature, some indications of the tri- 
 umph of syntax over etymology. To repeat here, what we have 
 elsewhere stated*, this tendency is not so much a war of lan- 
 guage with itself, as a contest between two modes of expression, 
 one of which is best adapted to the memory unaided by written 
 words, and the other best suited to the formal statement and 
 registration of our connected thoughts. Accordingly, when we 
 speak of languages as being in an old and new state or con- 
 dition, we speak of them as more or less affected by the cultiva- 
 tion of prose literature and by the common use of writing. As 
 we have not, by the nature of the case, any ancient language 
 which is altogether unaffected by the written records which have 
 transmitted it to us, we can only speak of these differences as 
 differences of degree. But we may divide all languages known 
 to us into three states or conditions, thus differing in the degree 
 of detriment which their cultivation of syntax has caused to 
 their etymological structure. We shall call these primary, 
 secondary, and tertiary states. 
 
 (1) Languages in a primary, or highly etymological state, 
 are those which have few or no syntactical contrivances; but 
 complete and regular inflexions, and a living power of deriva- 
 tion and composition. In such languages, writing has been 
 
 Maskil le Sopher, pp. 3, 4. 
 
86 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 cultivated at a late period, and circumstances have not favoured 
 the logical development of the language. The most remarkable 
 specimens of languages in a primary state are, the Sanscrit, 
 Sclavonian, and old Latin. 
 
 (2) Languages in a secondary state are those, which, 
 without sacrificing, to any very considerable extent, their inflex- 
 ions and power of composition, have still attained to a clear 
 and copious syntax. The most remarkable specimens of this 
 class are, the ancient Greek, and the modern High German. 
 
 (3) Languages in a tertiary state are those, which have 
 all but lost their inflexions and power of composition ; which 
 substitute syntactical contrivances for those variations of form, 
 which, in the older languages, characterize differences of declen- 
 sion and conjugation ; and which enjoy all the resources of logic 
 in the construction of their propositions. To this class we must 
 refer all the Semitic languages, the dead no less than the living, 
 together with a considerable number of modern idioms, including 
 the Romance languages*, and our own. 
 
 It is to be observed, however, that the passage of a language 
 from a secondary to a tertiary state generally presupposes, in 
 addition to the influences of writing and literature, some con- 
 siderable infusion of heterogeneous ingredients produced either 
 by conquest or emigration. Thus all the Semitic languages 
 have lost their inflexions and their living etymology in conse- 
 quence of a very early admixture of ethnical elements, to which 
 the Book of Genesis bears satisfactory and circumstantial testi- 
 mony. The Franks, when they conquered the Latinized in- 
 habitants of Gaul, and the Normans, when they settled as a 
 military aristocracy among the Anglo-Saxons, found, in the 
 countries to which they migrated, languages capable of inflexion; 
 and in each case the language which resulted from a compro- 
 mise between the victors and the vanquished, sacrificed all the 
 characteristics of etymological vitality. The same effects were 
 produced in a minor degree by the Mohammedan conquest of 
 Persia; and it appears that when the Etruscans subjugated the 
 
 * As we have already intimated (above, § 18) the original patois was 
 utterly ungrammatical. The syntax was restored and extended by tlio 
 literary efforts of the Troubadours. 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 87 
 
 Umbrians in Italy, the result was a mixed language, mainly 
 that of the vanquished, in which the declensions and conjuga- 
 tions were nearly if not entirely annihilated. 
 
 50 It cannot be said that the passage of a language from 
 one state to another, and the improvement of syntax at the 
 expense of etymology, is in any case tantamount to a degeneracy 
 of idiom. On the contrary, as we have already suggested, the 
 tertiary state generally accompanies and promotes an advance in 
 science and social culture. The degradation of a language is 
 a different process, and it is attributable to a widely different 
 cause. Speech is degraded when it loses its etymological struc- 
 ture without gaining the compensating advantage of a syntactical 
 development; and this is occasioned by a retrogression in the 
 social and intellectual position of the people, as when emigrants 
 from a civilized community are widely dispersed, and reduced 
 from an agricultural or political state to that of nomads, espe- 
 cially when this is accompanied by privations, and by the 
 deteriorating influences of a worse soil or climate. All the 
 sporadic or Turanian idioms of High Asia, of which we shall 
 speak in the following chapter, are instances of a degradation of 
 language : they are all probably depravations of the Iranian 
 type. Similarly, the languages of Africa must be considered as 
 successive products of Semitic disorganization : the Syro-Arabian 
 tongue passes from the Abyssinian to theGalla and Berber, from 
 this again to the Caffre, from the Caffre to the Hottentot, who 
 is finally caricatured by the savage Bushman. 
 
 Any state of a language may become liable to this degrada- 
 tion. But, in by far the greater majority of cases, the idioms 
 which have been subjected to this falling off were in their 
 primitive state, or at least in a primary condition, when the 
 causes which we have mentioned led to this depravation of their 
 structure and capabilities. It is scarcely necessary to mention 
 that a return to civilization is by no means denied to any 
 degenerate or sporadic tribes, and that they may not only re- 
 sume their social and political state, but may also be liable to 
 the same influences of writing, and ethnical admixture, which 
 produce the transition from the secondary to the tertiary state 
 of a language. We have a conspicuous example of this in the 
 
88 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 case of the Chinese. It cannot be doubted that the population 
 of this immense empire is made up of successive stratifications 
 of sporadic or Turanian immigrants, closely packed together, 
 and reinitiated, at an early period, into the arts which their 
 ancestors had cultivated in the original abode of the human 
 race. The consequence of this revival has been to make the 
 disintegrated remains of their degenerate idioms an artificial ap- 
 pendage to a system of written symbols. And to such an extent 
 is this carried, that two entirely different spoken languages are 
 represented by one single convention of arbitrary signs *. 
 
 * If there is any truth in the linguistic principles which are advo- 
 cated in this work, it must be a philological paradox to maintain, as 
 Bunsen does {Brit. Assoc. Report^ 1847, p. 299), that we have a "monu- 
 ment of antediluvian speech" in the Chinese language, which, it is ad- 
 mitted (p. 284), has lost its etymology, and retains only a formal and 
 meagre syntax. It seems also inconsistent with our psychological ex- 
 perience to hold that a root or crude -form, with a separable appendage, 
 is more ancient or original than the complete etymological structure, 
 which presents the object of conception in subordination to the thought- 
 forms of space and time. Still more inadmissible, in our judgment, is 
 the supposition that a language, which — like the Egyptian — has a defi- 
 nite article, and other purely syntactical substitutes for an enfeebled ety- 
 mology, is in a more primitive condition than those languages, which — 
 like the oldest members of the Indo - Germanic family — still exhibit a 
 perfect system of inflexions. We know by positive experience that ter- 
 minations of all kinds may be worn out or become insignificant, and 
 that, when this takes place, various syntactical contrivances are the in- 
 evitable results or concomitants of the change: and it is the tendency of 
 our widest researches to convince us that this always occurs, when con- 
 quest or migration has introduced a fusion of foreign elements. But we 
 have absolutely no single example of the converse state of things; there 
 is no case, in which an etymological condition of language has sprung up 
 from a crude series of monosyllabic juxtapositions: and it seems to us 
 that we cannot make such an assumption without ingnoring the obviously 
 scientific procedure. Above all, we think that any success in an attempt 
 to claim for the Turanian languages, and especially for the Chinese, a 
 principal or even a representative position among the original develop- 
 ments of speech, must more or less weaken our belief in the spread of 
 the human race from one common birth-place. The division of languages 
 into two great groups — the Central and the Sporadic — and the recog- 
 nition of the Chinese as a peculiarly aft'ected member of the latter class, 
 appear to us to indicate the only course, which can lead to a satisfactory 
 olution of all the problems suggested by modern Ethnology. 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 89 
 
 These are the only changes to which language appears to be 
 liable. It has a tendency to pass from an etymological to a 
 syntactical state ; and this process is facilitated by the cultivation 
 of writing, and is carried to its fullest limits by the admixture 
 of new ethnical elements. Language top may be degenerated 
 or depraved, and in this condition it may, by the sacrifice of its 
 few remains of vitality, become the instrument of literature and 
 science, and minister to the intercommunion of civilized man. 
 But if we believe that languages had a common origin, and that 
 the aboriginal inhabitants of the world enjoyed a complete intel- 
 lectual organization, we must conclude that the highly etymolo- 
 gical condition of a language must have been its original type, 
 and that all deviations from this type are of subsequent introduc- 
 tion, and should be explained by a reference to the operations of 
 secondary and external causes. 
 
 51 We have now given a general sketch of the first part of 
 the philosophy of language; we have sought to point out the 
 original unity of speech, to show that spoken language is na- 
 tural, but written language artificial, and to draw a bold and 
 intelligible outline of the eifects of the latter upon the literary 
 development of a nation. It remains that we turn to the 
 second part of the subject, and state, by way of explanation, 
 the connexion between the results of psychology, or the science 
 of mind, and of the philosophical analysis of inflected language. 
 That such a connexion, or rather identity of results, should exist 
 is necessary, if we are right in maintaining that language springs 
 naturally and spontaneously from the mind of man. 
 
 The results of all that writers on the philosophy of mind 
 have collected, with regard to our thoughts and the constitution 
 of our intellectual powers, may easily be summed up , so far as 
 they accord with our own convictions. Every man has one pri- 
 mary belief; that he exists, and that there is something without 
 him, full of realities, animate and inanimate; he sees too an 
 infinity of beings like himself, who live in the same belief. This 
 something without him is known to him from his sensations, 
 which, acting in the first instance on his bodily organs, pro- 
 duce an impression on his mind which we call a perception. 
 These perceptions survive the presence and the influence of the 
 
90 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 substance which caused them, they become a part of the mind, 
 and are called conceptions. Now the mind of man is so consti- 
 tuted that, whenever a perception is recollected or a conception 
 arises, it instantly awakes some other similar conception, or 
 perhaps a whole train of them, connected by the relations of 
 resemblance or contrast. This habit or tendency is called 
 association or suggestion. We can also combine those concep- 
 tions at pleasure, so as to form new conceptions existing only 
 in the mind, and this faculty is called imagination. Now all 
 these powers, with the exception of the last, are confessedly 
 enjoyed by the lower animals, and we class them all under the 
 name Understanding, the faculty of rules, or the faculty of 
 judging according to sense*. But there is also a higher faculty, 
 Avhich we alone possess, which presides over and regulates the 
 understanding, and which we call Reason, or the faculty of 
 principles. By this faculty we compare our conceptions with 
 one another, we estimate their similarity or incongruity, we 
 arrange the objects of our perception in classes, and these classes 
 again under more general subdivisions; we compare these ulti- 
 mate generalizations with one another, and so arrive analytically 
 at absolute truth: or, in some cases, we seize upon the prin- 
 ciples of science synthetically, a priori, and at once. It is this 
 faculty which constitutes our humanity ; it is to this that speech 
 ministers as an indispensable, but subordinate, adjunct. 
 
 The knowledge of his own existence and the simultaneous 
 belief in an external world, — this is the first act of man's con- 
 sciousness. But this consciousness is itself subjected to two 
 
 * It will be observed that we vise the term "understanding" in a more 
 limited sense than others, Coleridge for instance, give to the "human un- 
 derstanding." Coleridge attributes to the understanding many operations 
 which we consider as peculiar to the reason — discourse, abstraction, 
 generalization, &c. (Aids to Reflection ^ p. 215.) We adopt the Kantian 
 4istinction, in general, but we are rather disposed to comprehend under 
 ihe term reason every faculty which is peculiar to the mind of man, 
 .excepting the imagination, which, however, in its truest and highest form 
 can exist only in a reasoning and speaking creature. For imagination, 
 when it really deserves the name, is intimately connected and blended 
 with the reason. It is in fact the poetical reason, or the realistic element 
 in the reason. In its lower form it constitutes the fancy, which ministers 
 to the hope and fear of ipfants and dumb animals. 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 91 
 
 other primary intuitions : it is subordinated to the intuition of 
 space, for he is here^ and everything else is there^ and these are 
 two positions; it is subordinated to the idea of time, for the very 
 belief in his own existence presumes a continuance. 
 
 This then is the sum of psychology. Man is, and the world 
 is, there is a here and a there^ a me and a not-me^ — the know- 
 ledge of this fact is consciousness. He has perception, con- 
 ception, association, which constitute his Understanding. He 
 compares, generalizes, knows, and discourses; these are the 
 operations of his Reason. And all his thoughts are modified by 
 and subordinated to his primary intuitions of space and time. 
 
 52 Now if language be, as we say it is, the genuine 
 product of the reason, we should expect to find traces of all 
 these conformations of the mind in the structure of our speech. 
 And so it is. 
 
 Our analysis of the Greek and cognate languages has taught 
 us that there are two primary elements of speech ; the first, an 
 organizing element which enters into all words, and which we 
 call a pronoun; the second, a material element which consti- 
 tutes the basis of all significant terms which are not pronouns. 
 The pronoun expresses in the first instance the relation of the 
 thinking being to the external world, of the subject to the object, 
 of the me to the not-me; and this is formally put as an oppo- 
 sition of here to there. The first general and vague idea of 
 there is soon split up into a number of modifications, of which 
 the first is a distinction of objects in the there or outward world, 
 according as they are nearer to or farther from the subject, and 
 subsequently a designation of all the different directions in which 
 they stand with regard to the subject. The pronoun therefore 
 in its different forms is an expression of the first great fact of con- 
 sciousness, — thatweare and that there is something without us. 
 
 53 The material element of language includes the names 
 of all the objects which present themselves to us in the outward 
 world, and to our contact with which we owe the experiences 
 that are the staple for our understanding. We find on exami- 
 nation that all names of things are generic terms, that they 
 describe some particular quality or attribute of the object, whicl^ 
 
92 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 strikes us as most remarkable in it, and by which we at once 
 see its resemblance to the other objects of the same class. 
 We observe, too, that even the words which we call proper 
 names were originally generic terms, designating some qualities, 
 and consecrated to certain particular objects possessing those 
 qualities in a remarkable degree. It is, therefore, clear that 
 the very act of naming implies classification and abstraction, or 
 reasoning power; and when Adam is said to have named all the 
 animals, this is only another way of expressing the fact, that 
 by his reasoning power, which is identical with the power of 
 speech, he divided them according to the prima facie classes of 
 natural history. Of course, this use of general instead of special 
 names has a great effect on the conciseness and perfection of 
 language as an instrument of thought. But the process does 
 not stop here; not only are individuals described by general 
 names, but all the relations which bear any resemblance to the 
 attribute from which the body of the name, or the root as it is 
 called, is derived, are expressed by words into which that root 
 enters ; nay more, very many Avords expressing contrasted rela- 
 tions have the same root perhaps slightly modified. This is an 
 exemplification in language of the principle of association or 
 suggestion, which all psychologists recognise as one of the most 
 important operations of the mind. All writers on suggestion or 
 the association of ideas admit, either directly or by implication, 
 that contrast or contrariety is a species of connexion among 
 ideas ; indeed. Brown makes it one of the primary laws of sug- 
 gestion. Now, if we recollect that suggestion or association 
 depends upon previous coexistence or previous proximate succes- 
 sion, we shall not wonder, that, in this natural and necessary 
 process of expressing the greatest number of thoughts ormodifi- 
 cations of thought with the fewest possible Avords or modifica- 
 tions of words, ideas of contrast, as well as ideas of resemblance, 
 should be expressed by words, into which the same, or a slightly 
 modified root enters; for all contrasts and resemblances are rela- 
 tions, and no idea of a relation could be formed unless we had 
 seen the related objects together, or experienced the related 
 feelings in close succession ; but in this case, where the percep- 
 tions have taken place together, the recollection of one percep- 
 tion awakens a remembrance of the other; consequently, if we 
 
CHAP, in.] THE PHILOSOPHY OF LANGUAGE. 93 
 
 have got a word to express one of these related ideas, that word 
 suggests the other idea to our mind; therefore, the root of that 
 word, or a slight modification of it, would naturally be adopted 
 to express the other idea, whether it be an idea of contrast or 
 an idea of resemblance. And thus we find that a word may 
 bear two contrasted significations, or there may be two or more 
 words, containing the same or slightly modified roots, which 
 denote contrasted or contrary objects or feelings, when the 
 objects or feelings have been seen, felt, or experienced, always 
 or generally, in connexion or in immediate succession*. 
 
 54 Every word containing a root, or belonging to the 
 material element of language, also contains by way of prefix, 
 suffix, or both, a pronominal element. This is the counterpart 
 in language of the psychological fact, that every act of con- 
 sciousness is subordinated to the two conditions of thought, the 
 intuitions of space and time. The old Epicureans maintained 
 that the only real existences in the world were matter and 
 space f, and that every thing else was either a property {ροη- 
 
 * The following are a few instances of the principle of association 
 as it manifests itself in the same or a cognate language. 
 
 Contrast. Cause and Effect. 
 
 havere, wish, habere, have. aio, speak, άΐω, hear. 
 
 cupio, desire, capio, take. ανδάω, speak, audioy hear, 
 
 λάω, wish, λάω, take. καλί'ω, call, κλνω, hear. 
 
 χρήαιμος, χραΐΰμεΐν, assist, χήρος, χρψ- video, see, οίδα, know. 
 
 ζπν, want assistance. "set" (place), "sit" (be placed). 
 
 cams, possessed and valued, carere, want, δάω, bind, 8αίω, burn, 
 "dear" (prized), because you have it, δήμος, δημυς, do. do. 
 
 "dear "(expensive), because you want it. αητω, fasten, απτω, set on fire. 
 
 XQficc, use, XQsicc, need. δεκ-αως, . δεκ-ομαι,\ the 
 
 gestire, wish, gerere, carry with one. X^iQ, I αίρ-έω, f conse- 
 
 trachten, look at eagerly, tragen, to carry, hand, \ hin-than, \ quent 
 
 ' \ J' 1 nieut, . 7 
 
 μνήμη 1 finger, ι fanqen, ' act. 
 
 μένος \ a moving ^^, '^ ' / remain- •' ^ ' / c/ , 
 
 μέμονα ) force, desire, , l ing. 
 
 'Ο'ίω ) quickness of τίΟ"ημί ) 
 Ooo's i motion, -O-axog i ^^^ ' 
 
 "fast" (rapid), "fast" (fixed), 
 t Lucretius, i. 446 : 
 
 propter inane et corpora, tertia per se 
 Nulla potest rerum in niimero natura relinqui. 
 
94 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 jtmctmii) or an accident (eventum) of these*. Time, for instance, 
 was an accident of matter, not perceptible in itself, but to be 
 inferred from the rest or motion of things f. With what con- 
 nexion with this materialistic view we know not, but all people, 
 whether philosophers or not, seemed to have made up their 
 minds, till Kant appeared, that space at all events was some- 
 thing external, empirical, and real. Kant, however, deduces his 
 critical philosophy from the position that space and time are 
 a priori intuitions, because we cannot form a conception of out- 
 ward objects without a presupposition of space and time ; they 
 necessarily form the basis of all outward phenomena; they are, 
 both of them taken together, pure forms of all perception, and 
 consequently make synthetical positions a prion possible J. It 
 is true that the intuitions of Space or Position, and of Time or 
 Continuity, are equally original and equally necessary_, but if we 
 analyze them more rigorously, we shall find that the intuition of 
 Time is only a refinement and modification of that of Space. 
 These two primary notions may be otherwise stated as an intui- 
 tion on the one hand of positon or fixedness of objects with 
 isolations or intervals, which is the intuition of Space ; and an 
 intuition on the other hand of continuousness or motion of ob- 
 jects, or of such a closeness and proximity in their positions 
 that the intervals are not perceived, or not taken into account, 
 and this is the intuition of Time. Now it is clear even from 
 common language, that this is the whole distinction between 
 space and time; for the words which we use as indications of 
 position, such as "before" and "after," "backwards" and "for- 
 wards," are also indicative of time. We shall, however, make 
 our meaning clearer by an example. 
 
 55 That these primary forms of thought necessary to per- 
 ception are the basis of pure mathematics, is distinctly stated by 
 Kant§, and it is indeed obvious to every one who agrees with 
 
 * Lucretius, i. 450: 
 
 Nam qucequomque cluent aut his conjuncta duabus 
 Rebus ea invenies aut horum eventa videbis. 
 
 t V.463: Nee per se quemquam tempus sentire fatendum est 
 Semotum ab rerum motu placidaque quiete. 
 
 X Kritik der reinen Vermnft, pp. 28—43. § Ibid. p. 41. 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 95 
 
 Plato in considering the exact sciences as derived from percep- 
 tion by the intellectual faculties. The two first invented of the 
 exact sciences were Arithmetic and Geometry, which are both 
 referable to the intuition of Space. The latter was always, in 
 the hands of the old geometers, the science of position ; in the 
 former, all the principles are derived from the notion of inter- 
 vals, and the primary names of the numbers are, as we shall 
 hereafter see, pronominal words signifying position. For con- 
 venience in reckoning, it soon became customary to substitute 
 for these arithmetical words a set of symbols, all of them single 
 letters, and people were not long in inventing concise methods of 
 combining these according to the principles of the science. But 
 even these abbreviations Avere not enough, and a sort of short- 
 hand was invented in different parts of the world, which Euro- 
 peans have agreed to designate by the Arabic name Algebra*. 
 This written language, for it was only a set of symbols, and 
 therefore could not be spoken, was, in process of time, extended 
 to the expression of geometrical results : but only imperfectly ; 
 because the geometer sometimes encroached upon the domain of 
 the other intuition; and a science of pure time had not been deve- 
 loped from the sciences of Space. In fact, the intuition of Time 
 or continuity was much more difficult to deal with; like the old 
 Heracleitean doctrines, it presupposed a continual flowing or 
 change, and escaped from the grasp of expression. The great 
 difficulty to be overcome was a philological one, — the construction 
 of a language to express motion, time, or continuous change. 
 This obstacle was surmounted at nearly the same period by 
 both Leibnitz and Newton, and their discovery of the language 
 of change was one of the greatest importance immediately for 
 physical science, and ultimately, we doubt not, for philosophy 
 in general. No one has been at the pains to point out the 
 natural progress of this discovery : it will, we apprehend, be 
 easy to do so. The most obvious example of continued change, or 
 melting down of intervals, is that of physical continuous motion: 
 
 60- 
 
 w -- 
 
 * This word is referred to γΛ:2^, and signifies "reductio partium ad 
 
 totum, sen fractionum ad integritatem" (Golius, c. 462; Freytag, i. 
 p. 239 b). 
 
96 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 SO obvious, indeed, that the ancient philosophers included under 
 the name motion (κίνηόις) all that we include under the term 
 change; thus Plato, Thecetet. p. 181 D: dvo δη λέγω τοντω 
 εϊδη κινήΰεως, άλλοίωβιν, την δε τνερίφοράν (read φοράν). 
 Parmen, ρ. 138 C: κίνονμενον η φέροοτο η άλλονοΐτο αν. 
 Aristot. Nat, Ausc. νιι. 2, §. 1 : επεί δε τρεις εΐΰι κίνήύείς, η τε 
 κατά τόπον, και κατά το τιοιόν, και κατά το τίούόν, ανάγκη 
 και τα κινούμενα τρία. η μεν ονν κατά τότίον φορά, η δε 
 κατά το τίοών άλλοίωΰις, η δε κατά το τίούον ανξηΰις κσΧ 
 φ^ίύΐξ. 1η accordance with this, then, the earliest language or 
 science of change borrowed all its terms and even its name 
 from physical motion: though from the very first it was applied 
 to the investigation of problems in change or continuity in 
 general. The natural division, therefore, of the exact sciences 
 is this. (1) The science of positions or intervals, which includes 
 geometry and arithmetic. (2) The science of time or of conti- 
 nuous change, which comprehends mechanics, dynamics, and the 
 great problem of physical astronomy. When Algebra, or the 
 symbolical language in which the sciences of space were ex- 
 pressed, was applied to the science of time, it was called 
 Fluxions or Differential Calculus ; but it might in fact be called 
 by the name of the older language, of which it is merely an 
 extension. We are aware that an eminent mathematician, in 
 the sister island, has asserted that Algebra, by which he means 
 all that is included in the unphilosophical use of the word 
 analysis, is the science of pure Time *, and he even goes so far 
 
 * Sir W. Rowan Hamilton, in the introductory remarks to an essay 
 "On Conjugate Functions and on Algebra as the Science of Pure Time" 
 (Transactions of the Royal Irish Academy, Vol. xvii. pp. 293 foil.), states 
 "that his object is to enquire whether existing Algebra offers no rudiment 
 which may encourage a hope of developing a science of Algebra, properly 
 so called, strict, pure and independent, deduced by valid reasons from its 
 own intuitive principles ; and this not less an object of ά priori contem- 
 plation than Geometry, not less distinct in its own essence from the rules 
 which it may teach or use, and from the signs by which it may express its 
 meaning; and that he has been led to the belief that the intuition of ti7ne 
 is such an element." This is not the place to enter upon a formal exami- 
 nation of so profound a subject: but wc are sure that any one who will 
 look into Sir W. R. Hamilton's paper, and compare it with the explana- 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. ' 97 
 
 as to say, that arithmetic is a part of the same science*: but 
 with all deference to him and another distinguished analyst f, 
 who defines Algebra as the science of general reasoning by 
 symbolical language, we must insist that Algebra can never be 
 called a science, when separated from its applications, which are 
 all so many distinct sciences. If the science of pure time is 
 coextensive and identical with Algebra, as the former scholar 
 asserts, then must Geometry, which is the science of pure Space, 
 become the science of pure Time whenever it is expressed in 
 analysis. Algebra should be defined as the method or art of 
 combining symbols, as a language in which we can carry on 
 the most abstract and general reasonings about sensible objects, 
 considered in their relation to one or other of our original 
 intuitions. 
 
 It appears, then, from the progress and extension of Algebra, 
 that the intuition of Time, though necessarily co-ordinate with 
 that of Space, may be derived from it by adding the idea of 
 motion or change, or by melting down the intervals which con- 
 stitute position, and that in scientific language, at all events, the 
 expression of Time is posterior to that of Space J. In the com- 
 mon languages we find two classes of the material words, which 
 
 tion given in the text, must admit that he has confused the method of 
 Algebra with one of its applications. 
 
 * JJbi supra^ p. 308. f Professor Peacock's Algebra, § 1. 
 
 J Mr. Mansel (Limits of Religious Thought, pp. 83, 204), says that all 
 consciousness is subordinate to the general law of time, and that subor- 
 dinate to this general law there are two inferior conditions, to which the 
 two great divisions of consciousness are severally subject— our knowledge 
 of body being governed by the condition of space, and our knowledge of 
 mind by that of personality. "From these," he maintains, "are derived 
 three corresponding systems of necessary truths in the highest sense of the 
 term: the science of Numbers being connected with the condition of 
 Time; that of Magnitude with Space; and that of Morals with Person, 
 ality." We place this statement in juxtaposition with that in the text, 
 which seems to us to expose before-hand Mr. Mansel's somewhat con- 
 fused apprehension of the subject. The idea of number is an idea of 
 fixity, for every number is a sum and a standing-point. The idea of time 
 is that of continuous change, and we cannot express time in number 
 without transferring it for the moment to the domain of space and mag- 
 nitude. This is only one instance among many of the sham science 
 which abounds in Mr. Mansel's book. 
 
 Η 
 
98 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 we call nouns and verbs. The former are capable of expressing 
 relations ofSpaceonly : the latter denote actions or express rela- 
 tions of Time. Yet we find that both are made out of the same 
 materials; the roots or stuff of language enter into each set, and 
 they are each of them combined with pronominal elements, 
 which denote the case-relations in the former, and the person- 
 relations in the latter; the cases of the nouns expressing the 
 position of some object with regard to other objects, the persons 
 of the verb the point from which the action begins, or at which 
 it ends. These, we shall see, stand upon exactly the same foot- 
 ing, and the expression of agency, whether effected by a case, 
 a preposition, or a person-ending, is still strictly pronominal or 
 derived from the intuition of space. 
 
 From this examination we see that the law according to 
 which the words of a perfect, or, what is the same thing, an 
 inflected language, are formed, that is to say, their anatomical 
 structure, or internal mechanism, is the counterpart of what we 
 know of the operations of the mind. Here, however, the parallel 
 is at an end, and we must be careful to recollect that the words 
 themselves, when once formed into a whole, are nowise repre- 
 sentatives of any thing in the mind. They may go on through 
 all possible shades of meaning, and even be used by abstraction 
 without any regard either to their structure or primitive signifi- 
 cation, and without in the least affecting the mind with a com- 
 prehension of their import: nay, it is, as we have shown, 
 the natural process in language, as it developes itself syntacti- 
 cally, to destroy the fulness and significance of its individual 
 words ; and it is highly beneficial to science that such should 
 be the case. A very pregnant example of this is furnished by 
 those general abstract terms of which so much has been said by 
 metaj^hysicians. 
 
 If we examine the abstract and general names in any lan- 
 guage, we shall find that they are only tropical or figurative 
 words properly referring to sensible objects; and the reason of 
 this is obvious, for the whole end of language is to transfer our 
 inward feelings to the outward world, so that they may become 
 cognizable to others, and objective to ourselves ; now in order to 
 attach a name to a thing, it is necessary that the name and the 
 thmg should be presented to the observation a certain number of 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 99 
 
 times together: but it is easier to present a material object to 
 the observation of another for the purpose of naming it, than to 
 describe to him an impression or a thought ; consequently, ma- 
 terial objects are first named, and thoughts or ideas are described 
 by a metaphorical reference to them. Of course, this method 
 of forming our abstract terms, though necessary under the cir- 
 cumstances, is productive of serious inconveniences ; by using 
 metaphorical words, we are apt to reason vaguely in consequence 
 of the dijBPerent significations which the words bear in common 
 language. Hence, for the purposes of science, it would doubtless 
 be desirable to have a set of words which bear no specific mean- 
 ing. But this is impossible in spoken language, except in the 
 case of merely pronominal words, denoting not things but the posi- 
 tions of things : therefore it is only in Arithmetic, Geometry, and 
 Music that we can have a spoken language perfectly general. In 
 symbolical written language, however, it is possible to put down 
 marks or signs, and invent laws for their combination with- 
 out at all troubling ourselves about their interpretation; and 
 it is to the invention of such a language, and its subsequent 
 extension to subjects beyond the arithmetical calculations to 
 which it was at first applied, that the great advances in pure 
 mathematics, and the sciences depending on them, are to be 
 attributed. 
 
 56 As abstract general terms are merely the names of sen- 
 sible objects used tropically, they could not be considered as the 
 representatives of any ideas in the mind, even though it were 
 true that the words of a language, and not the mode of forming 
 them only, might be regarded as the representatives of mental 
 operations. The controversy between the realists and nominalists, 
 of which we have given a short account in the last Chapter, 
 could not arise at the present day; every one is now aware that 
 words, as the signs of generalization, are the only objects about 
 which general reasoning is conversant. If any question of this 
 sort could be agitated at present, it must be one between the 
 nominalism of Occham, or conceptualism as some might be 
 pleased to call it, and the ultra -nominalism of the school of 
 Hobbes, Home Tooke, or Bentham. Some of these have gone so 
 far as to seek for general truths in the words of a particular 
 
 Η 2 
 
100 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 language; but no one, now-a-days, would conversely assert the 
 objective existence of general ideas, as something independent of 
 the general terms which we use in reasoning. 
 
 It is true, indeed_, that general terms presume generalization ; 
 it is true that there is such a thing as general^ necessary, abso- 
 lute truth, and that synthetic judgments a priori are possible ; 
 it is true that there are genera and species of things, and, 
 in short, representative or abstract knowledge as opposed to 
 perceptive or intuitive knowledge; but it is not true that, 
 because we can abstract and generalize, therefore we have in 
 our mind general abstract ideas or images of the absolute and 
 vmconditioned, still less that our general terms are representa- 
 tives of such ideas, and least of all that such abstract ideas 
 have an independent existence. As a great philosopher has 
 remarked, in speaking of the metaphorical meaning of general 
 terms, our necessities have obliged us to depart from the natural 
 order of our ideas; we have been obliged to attach ourselves 
 to one furnished by the occasions and accidents to which we 
 are liable, and this order gives us not the origin of our notions, 
 but the history of our discoveries*. To adopt the words of the 
 same philosopher, there are two different kinds of ideas, — the 
 real and the nominal. The nominal idea of a thing is but its 
 definition; and thus a simple idea is only real, for it cannot 
 have a definition, that is, a new simple idea cannot be raised 
 in the mind by means of words. The nominal idea or essence 
 of a thing is simply that quality or attribute which we remark 
 in it as the point of similarity between it and other individuals 
 which we class with it, and which is therefore the cause of its 
 name. This definition, like all classification or naming, is of 
 course to a certain extent arbitrary; for, as Dugald Stewart 
 observes f, it does not necessarily follow that this quality is 
 more essential to the existence of any thing as an individual 
 than various other qualities which we are accustomed to regard 
 as accidental. The real definition enables us to see the possi- 
 bility of the thing defined, and it is this definition alone that 
 can be madethebasis of science, for which the nominal definition 
 
 * Leibnitz, Nouveaiix Essais sur I'Entendement Humain, p. 324. 
 t Elements, p. 130. 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 101 
 
 is not sufficient, unless it can be shown by experiment that the 
 thing defined is possible, in which case the definition becomes 
 real. There is, perhaps, no such thing in the world as a perfect 
 circle, but the definition of the circle enables us to see the pos- 
 sibility of the thing, and therefore the definition is allowed to 
 rank among the first principles of the science of Geometry. The 
 essence of a thing is but the possibility of it, and therefore does 
 not depend upon ourselves ; the merely nominal definition is arbi- 
 trary, and though there is but one essence, there may be several 
 nominal definitions of the same thing, while the real definition 
 must be justified by the reason, which shows that it is possible, 
 or by experience, which shows that it actually is, and is therefore 
 possible*. 
 
 57 The doctrines of the Realists in the middle ages seem 
 to have been suggested by a misconception of the philosophy of 
 Plato, which has prevailed to the present time•)*. It has been 
 all along supposed that Plato was a realist in the strictest sense 
 of the word, that he believed in the independent existence of 
 universal ideas, that he had a great passion for the marvellous 
 and mysterious, and so forth. We believe nothing of the kind. 
 Plato may have been a bad citizen, — in his heart a traitor to 
 his country, and an enemy to her institutions, — but he was not 
 a mystical dreamer, or a wild enthusiast; he was the very 
 greatest of all true philosophers, because he was the first; he 
 was a sober, clear-headed thinker, and not the less so because he 
 had the most brilliant fancy — a mind teeming with the most 
 poetical imagery that ever gilded the page of abstract specula- 
 tion. The business of philosophy, as we have before said, is to 
 undress the objects of sense; to take the thought away from the 
 particular, and turn it to the general. In the beginning of real 
 philosophy this was the great thing to be done. The first phi- 
 losophers, so called, were materialists and ultra-nominalists ; and 
 therefore it was Plato's object, as a true philosopher, to esta- 
 blish at least the position that truth and science cannot be found 
 
 * Leibnitz {ubi supra, pp. 252 foil.). 
 t It is perhaps right to make an exception to a certain extent in favour 
 of Bishop Berkeley (see Siris^ § 338). 
 
102 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 in the individimls, but must be sought after by general reason- 
 ing; that we must take general terms, the names of classes 
 and not of individual things, if we would arrive at any valuable 
 conclusions. If he had written, as Dugald Stewart might have 
 written, on the same theme in the nineteenth century, after the 
 world had enjoyed for many hundred years the lights of philo- 
 sophy, science, literature, and a true spiritual religion, he would 
 have had no occasion to use allegories about chariots and 
 winged horses, and ideas dwelling in the world of intelligence, 
 and metempsychosis*, and so forth. But living as he did in 
 an idolatrous country, where every association was opposed 
 to abstraction, and the human soul made an image worship of 
 its every thought, where there was no literature except poetry 
 and annals, and these too read by few, he was obliged to set 
 up idols against idols, to make the imagination, which had 
 created all the elements of Greek polytheism, its own iconoclast 
 in favour of a rival worship ; and so he spoke of ideas as things 
 real, objective, and independent, dwelling with God in the 
 heaven of heavens, and making other things what they are by 
 participation. Nevertheless, no one knew better than he did 
 
 * The allegory in the Phiedrus is borrowed entirely from the circum- 
 stance, that, in the Attic dialect, words referring to the use of wings are 
 employed to signify the emotions of the mind (see, for instance, Sophocles, 
 Ajax, 693); a metaphor so obvious that Aristophanes makes a very length- 
 ened joke upon it in the Aves^ 1436 — 1450. If the reader wishes for an 
 instance of the way in which Plato could spin an allegory from the com- 
 mon terms of poetical language, let him compare Phcedrus, p. 251 a — d, 
 where we have ηρώτον μεν έφριξε — οίον L• της φρίκης — ϋ^ερμότης — ^ερμαν- 
 ^εντοςδε έτάτιη — ζεϊούν εν τούτω ολη,ηαΐάναν,ην,ίει — τ7]δίεξ,όδω εγ χρίει 
 εκαοττ] — α>ΰτε πάοα κεντονμένη κύκλω η ιρνχη οίοτρα ν,αϊ όδννάταί, with 
 Sophocles, Trachinioe, 831 foil.: 
 
 εί γάρ οφε Κενταύρου φονία νεφέλκ 
 χρίει δολοποιος ανάγκα 
 πλευρά ηροοτ ακέντος ίου 
 
 δεινοτάτω μεν ϋδρας προατεταζώς 
 
 φάαματι; 
 
 άμμιγά νιν αίκίζει 
 
 ύποφόνια δολόμυϋ'α κέντρ 
 
 ^φ ιζέααντα. 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 103 
 
 that this was but philosophy speaking in parables; as will 
 appear from the consideration of a very few facts*. 
 
 58 Plato bas6s his whole system on dialectic or lo'gic, the 
 art of general reasoning. He kneΛv that there could be no 
 general reasoning leading to philosophy, or general principles, 
 without real definitions. Now, the definition necessarily includes 
 two things, generalization and division, or, in the words of 
 modern logicians, it is made either per genus or per diffe7'entiam. 
 The former process is the base of the second; the second is the 
 development of the former. Accordingly, dialectic, and there- 
 fore philosophy, depend upon generalization ; and Plato's theory 
 of ideas, as it is called, is merely the assertion of the principle, 
 that in order to general reasoning we must generalize and 
 classify, κατά γένος διάκριναν and τίατ εϊδη ΰτίοΛεΐν^ which he 
 explains very clearly in the Phcedrus (p. 249 b): δει γαρ ανϋ'ρω- 
 Ttov ξυνιέναυ κατ εΐδος λεγόμενον, εκ πο?Λών ίον αίΰ^'ηΰεων εΙς 
 εν λογίΰμώ ξνναίρονμενον^ — andthis we presume is now generally 
 admitted f. It is strange that this should have escaped the notice 
 of so many writers on the history of philosophy ; one would have 
 thought that the connexion between him and the Pythagoreans, 
 who made the same use of numbers, — the first abstract terms in 
 language, — which he did of his ideas, would have taught them 
 that Plato's object was only to bring forward the principles of 
 science or necessary truth, to draw the first outlines of a system 
 of logic or general reasoning, by laying down the rules of classifi- 
 
 * See Cousin, Nouveaux Fragmens Fhilosopkiques, pp. 160 foil. 
 f Professor Thompson, in his essay on the Sophista of Plato {Trans. 
 of the Cambr. Phil. Soc. Vol. x. Part, i,), makes the following remarks on 
 Plato's method of definitions: "Two methods are marked out (in the 
 Phcedrus) for the dialectician to pursue in searching for definitions. 
 Either he may start from particulars and from them rise to generals: 
 or he may assume a general and descend by successive stages to the 
 subordinate species (the species specialissimd) which contains the thing or 
 idea, which he seeks to define. The first of these processes is styled 
 by Plato ΰνναγωγή^ Collection: by Aristotle, fjrayroyry, Induction: the 
 second is called by both Plato and Aristotle διαίρεοις^ or the διαιρετική 
 με^Όδος, Division or the Divisive Method. Whoso is master of both 
 methods is styled by Plato a Dialectician, and his art the Art of Dia- 
 lectic." 
 
104 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 cation and generalization. His pupil Aristotle, who has griev- 
 ously misrepresented his meaning, did but fill up his scheme*; 
 and it may be shown from the words of both, that, in talking of 
 genera and species, categories, and universals, they meant only 
 general terms, the necessary instruments of reasoning, the main 
 part of the definition real, which is perpetual because it speaks 
 only of the possible f. 
 
 59 We need not search long in Plato's works without 
 finding indubitable proofs of his nominalism, expressed in the 
 most direct terms J. For instance, in the Republic (x. p. 596 a) 
 
 * In the paper just quoted Mr. Thompson says: "Aristotle objects 
 to the term μέϋ-εξις on the ground that it is metaphorical. Now as a 
 logical term, the Platonic μέ^εξίς is but the counterpart of νπαρξις, the 
 Aristotelian word denoting the relation of subject to predicate. The 
 one term is as metaphorical as the other and not more so." 
 
 t Leibnitz, u. s. p. 254: les Essences sont perpetuelles paree qiCil ne s'y 
 agit que du possible. 
 
 J Mr. Dyer in a paper On the noun, or name, as an instrument of reason- 
 ing, read before the Philological Society, 14. Jan. 1848 {Proceedings, Vol. iii. 
 No. 65), has combated this view of Plato's philosophy. He maintains that 
 nominalism would have been totally inconsistent with Plato's particular 
 tenets; that he was in fact a realist. But he admits that Plato's "realism 
 did not prevent him from making use of general terms for logical pur- 
 poses, precisely in the same way as the most thorough nominalist." He 
 tells us too that "the germ of Plato's philosophy lies in the well-known 
 passage of the Pha;drus (245 d sqq.) in which the soul is likened to a yoke 
 of winged horses;" and he maintains that "fanciful as this sketch may 
 appear, it in reality contains the leading principles of the Platonic philo- 
 sophy, such as we find them worked out in a more serious manner in the 
 later dialogues." The question therefore between Mr. Dyer and ourselves 
 lies within a very narrow compass. He admits that Socrates was a nomi- 
 nalist, and that Plato adopted the dialectic method of his master in a 
 thoroughly nominalistic manner. Accordingly, it only remains that we 
 should decide whether his idealism was a fanciful play with words and 
 metaphors, which might be perfectly consistent with the merest nomi- 
 nalism : or whether a clear-headed man, who understood the meaning of a 
 general predication in language, was so besotted in his word-worship that 
 he was obliged to give an external and objective existence to the signi- 
 ficance of every verbal abstraction. In spite of the vagueness of his poe- 
 tical phraseology, it is sufficiently clear from the passages quoted in the 
 text that Plato regarded the general term or name as the only result of 
 abstraction. 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. 105 
 
 he begins an investigation by taking the generic name (ρνομά) 
 as a representative of the genus (είδος, Idea, which are in this 
 passage used as identical words*) and states that this is his 
 usual method — βονλει ovv εν^Ίνδε άρξώμε^'α εταΰκοπονντες , εκ 
 της εΐω^νίας μεθόδου; είδος γάρ που τι 'εν εκαΰτον εΐώ&αμεν τί- 
 %ε6^αυ Ίίερί εκαβτα τα πολλά, οίς ταντον όνομα εταφερομεν — 
 \)~ώμεν δη και νυν οτι βονλεί των Λολλών. οίον, ει θέλεις, τΐολλαί 
 jtov εΐΰι κλΐναι και τράτιεξαι. — αλλ' ιδεαι γε τΐον τΐερι ταύτα τα 
 ΰκενη δύο, μία μεν κλίνης, μία δε τραπεξης. And in the Laws 
 (χ. pp. 895, 6) he gives in plain words the distinction, which we 
 have given above, between the name and the definition, the 
 former being a prima facie^ the latter a scientific classification, 
 the former a nominal, the latter a real description: αρ ουκ αν 
 εΰ-έλοις περί εκαΰτον τρία νοεΐν; — ϊν μεν την ονΰίαν, 'εν δε της 
 ονΰίας τον λόγον, 'εν δε όνομα' και δη κάί ερωτηθείς είναι περί 
 το ον aitav δνο. — τότε μεν ημών εκαβτον τοννομα τΐροτεινόμενον 
 αντο τον λόγον άτΐαιτεΐν, τότε δε τον λόγον αντον τίροτεινόμενον 
 ερωτών αν τοννομα. — εότι τΐον δίχα διαιρονμενον εν άλλοις τε 
 καΐ εν άρι^^μω. τούτω δη τω κατ άρι^'μον όνομα μεν αρτιον, λόγος 
 δε άριΰ-μος διαιρούμενος εις ϊόα δύο μέρη. — μών ονν ον ταντον 
 εκατερως προΰαγορεύομεν, αν τε τον λόγον ερωτώμενοι τοννομα 
 άτίοδιδώμεν, αν τε τοννομα τον λόγον, αρτιον ονόματι και λόγω, 
 δίχα διαιρούμενον a^tO'^bf τίροβαγορεύοντες ταντον ον; — ων δη 
 φνχη τοννομα, τις τούτον λόγος; εχομεν άλλον τΐλην τον ννν δη 
 ρηΟ'έντα, την δνναμένην αντην αντην κινεΐν κίνηύιν; on which it 
 is asked, το εαντο κινεΐν φης λόγον εχειν την αντην ονΰίαν ηνπερ 
 
 * When είδος and ιδέα are distinguished by Plato, the former de- 
 notes the mental apprehension, and the latter its counterpart in nature. 
 See Thompson's note on Butler^s Lectures, ii. p. 127. In common Greek 
 ιδέα means the outward form or manner of any thing, while είδοξ denotes 
 the class or species. We have them both together in Thucyd. iii. 62 : 
 ■ημΒΪς δε μηδίβαι μεν αυτονς ον φαμεν, διότι ούδ' 'Αθηναίους, ry μεντοι avvfj 
 ιδέα νΰτερον ιόντων Ά&ηνούων ίτά τονς"Ελλην<χς μόνους αν Βοιωτών άττικί- 
 ΰαΐ' καίτοι οκέιραο^'ε εν οΐω εΐδει εηάτεροι ημών τοϋτο έπραξαν. Here it is 
 obvious that whether we take ττ] αντγι ιδέα with ιόντων or with άττιζίΰαι, it 
 means "in the same way or manner," eadem ratione, auf dieselbe Weise. 
 But έν οΐω εΐδει must mean "in what kind of government," "in what 
 specific form of constitution," in qua 7'eipubliccB forma;, in welcher Verfassung, 
 in was fur einer Stellung ; cf. viii. 90, § 1: oi τών τετραν,οοίων μάλΐΰτα 
 ενάντιοι οντες τα τοιοντω εΐδει. 
 
106 THE PHnjOSOPHY OF LANGUAGE. [bOOK I. 
 
 τοννομα d δη TcavTsg 'ψνχηνΛροΰαγο^ενομεν; and this is assented 
 to. If we compare these two passages with those which we have 
 quoted above from Occham, the chief of the Nominalists, we 
 shall see that their opinions on the value of universals coincide*. 
 Plato, although no philologer, had convinced himself of the 
 fact which philology has made certain to us, that although the 
 structure of language is a counterpart of the organization of the 
 mind, the individual words are only arbitrary signs, and there- 
 fore do not contain the truth of things. But the great talkers, 
 by whom he was surrounded, and whose writings constituted the 
 intellectual food of Athens, had arrived at the extremest point of 
 ultra-nominalism, and had asserted that truth was to be found, 
 not only in the fleeting phenomena of the visible world, but 
 even in the individual words of a particular language. Plato is 
 not to be charged with realism because he opposed this abuse 
 of nominalism, any more than a man is to be considered an in- 
 fidel who is opposed to the excesses of religious zeal. But he 
 has been called so, because, as Aristotle says, those who are in 
 one extreme of wrong, class in the opposite extreme of wrong 
 all who hold to the golden mean of right. 
 
 60 The work in which Plato directly opposed the philo- 
 logical application of this ultra-nominalism, the Cratyhis, was till 
 very lately altogether misunderstood; we shall therefore give 
 some account of it, and of the modern work which stands in 
 prominent opposition to it, the Diversions of Pmiey, by John 
 Home Tooke, as well on account of the contrast between 
 them, and our decided opposition to the latter, as because the 
 serious truths for the first time announced in the Cratylus, its 
 connexion with the rest of Plato's system, and consequently with 
 that philosophy which is the beginning of human knowledge, 
 have induced us to borrow from it the title of this book. 
 
 The utterly ridiculous and unjustifiable etymologies brought 
 forward in Plato's Cratylus^ and the strange mixture of joke 
 
 * We have elsewhere pointed out more fully the verbal and gramma- 
 tical reasoning on which Plato's system depends: see Penmj Cydopcvdia, 
 s. Λτ. Plato^ p. 236 ; Literature of Greec, ii. p. 230. 
 
CHAP. III.j THE PHILOSOPHY OF LANGUAGE. 107 
 
 and earnest which one finds in every page, formerly rendered 
 this dialogue a great stumblingblock to all the admirers of the 
 philosopher. They were generally unable to determine what 
 place in Plato's works should be assigned to it, and what was 
 its real object. At present, however, scholars are nearly agreed 
 as to its general meaning. That Plato, whose main object was 
 to establish a system of dialectics as a means of inquiring after 
 truth, should have been ignorant of the importance and neces- 
 sity of establishing the connexion between ideas and words, and 
 should not have had some sober theory of language, the dialec- 
 tician's instrument, cannot be believed. On the contrary, he 
 seems to have been continually impressed with the conviction, 
 that his chief business was to solve, in part at least, the problem 
 of language, for he says that language is the counterpart of the 
 mind (Phcedrus^ p. 276 a), and that the word-maker must have 
 a dialectician set over him {Gratyl. p. 390 d). The great object 
 of Plato in all his works was to lead the mind away from its 
 continual attraction to objects of sense, to teach us, that, if we 
 ΛνοιιΜ find truth and science, we must ascend to laws or general 
 principles, and not confine our attention to the multiplicity of 
 facts and individual objects (or, as he expressed it in his poetical 
 language, we must seek for them not in the world of matter, but 
 in the world of mind, for the former contains only shadowy 
 representations of the realities displayed by the latter), that there 
 it something more in man than a mere congeries of recollected 
 experiences, and that he ought to have higher thoughts and 
 more exalted pleasures than those which the outward world can 
 furnish. This is the substance of his arguments with his con- 
 temporaries, on all the great questions then agitated in philo- 
 sophy, and it is well known that he thought banter and irony 
 as good a vehicle as any other for his purpose. Accordingly, 
 when he found that words, like other outward objects, instead 
 of being considered as merely symbols of reasoning, were them- 
 selves made the objects of examination, as if truth and science 
 were to be discovered in sounds and signs, which had no mean- 
 ing, save as interpreted from within ; when he found too that this 
 examination was carried on in the most arbitrary and capri- 
 cious manner, without any regard paid even to the most obvious 
 principles of etymology, and solely for the purpose of supporting 
 
108 THE PHILOSOPHY OF LANGUAGE. [bOOK L 
 
 some specific dogmas; he added to his works an exposure of 
 these absurdities, which only diflPers from his other bantering 
 treatises in having more ludicrous and extravagant theories to 
 combat. The Eleatics and Heracleiteans in particular had made 
 use of etymology to establish their contradictory positions , as- 
 serting that it appeared from the words themselves, the former 
 that every thing was fixed and stationary (ΐοταΰΰ'αϋ)^ the latter 
 that every thing was in motion (κίνεΐύΟ'αί). This furnished an 
 excellent opportunity for ridiculing themethodof both, by show- 
 ing that both their systems were alike demonstrable from ety- 
 mology. The Cratylus who gives his name to the dialogue, 
 was a disciple of Heracleitus, and, according to Aristotle* (or 
 whoever is the author of the first book of the Metaphysica)^ 
 Plato had, when young, some intercourse or acquaintance with 
 him. The other interlocutor is Hermogenes, the brother of Cal- 
 lias, who is introduced as a supporter of the Eleatic doctrines. 
 When we remember how Protagoras, the Heracleiteans, and the 
 Eleatics, are all introduced together in the Theaetetus, and how 
 in that dialogue Plato combats the two former sets of doctrines 
 most especially, and in conjunction with one another, on grounds 
 similar to those which he advances against Cratylus in this, we 
 cannot avoid considering this treatise as a supplement to the 
 Theaetetus. The doctrines of Protagoras and Heracleitus coin- 
 cided in many points, and particularly in their views on the 
 nature of language; it is for this reason no doubt that Hermo- 
 genes, as the representative of the Eleatics, is made to speak 
 contemptuously of the philological part of Protagoras's work, 
 called '^Ai^O-ata (Cratyl, p.391c), and the Homeric etymologies 
 in this dialogue have been thought to be a hit at Protagoras ; for 
 it appears from the Theaetetus (p. 152 e), that the disciples of 
 Protagoras and Heracleitus supported by quotations from Homer 
 the doctrine of the perpetual motion of things; also, as in the 
 Theaetetus, the Eleatics are treated with much more considera- 
 tion, and all the weight of the ridicule is made to fall upon the 
 representative of the Heracleiteans; the banter is carried to the 
 
 * *Εκ vhov τε γάρ οννή^-ης γενόμενος πρώτον Κρατνλω %αΙ ταΐς Ήρακλει- 
 τείοίζ ό'όξαίς, ώ^ απάντων των αίΰΟ-ητών άεϊ ρεόντων και. επιστήμης περί 
 αντών ουκ οϋΰης, κ. τ. λ. Aristot. Metaphys. ι. c. 6. 
 
CHAP. ΠΙ.] THE PHILOSOPHY OF LANGUAGE. 109 
 
 greatest length, when Socrates tells Cratylus that he owes the 
 absurd derivations which he brings forward, and to all of which 
 Cratylus assents, to the inspiration which had come upon him 
 from his morning's talk with Euthypron, a mad and ridiculous 
 quack. The object of the Thesetetus is to overthrow entirely 
 the doctrines of Protagoras and the Heracleiteans, to show that 
 the grounds of science are not to be sought in the province of 
 the senses, that in fact science is neither perception nor right 
 conception, nor even right conception combined with reasonable 
 explanation. Now the second of the three things which science 
 is not, namely, right conception, is one and the same thing 
 with language*; and these sophists had actually made language 
 an object of inquiry,as if science had been to be found in words: 
 therefore it was necessary to show, not only that science was 
 not identical with right conception, but also that there were 
 no grounds of science in language, which, although intrinsically 
 the same with right conception, was extrinsically so far different 
 as to merit a separate investigation ; this, however, could not 
 well have been introduced as a digression into the Theaetetus, 
 and therefore the Cratylus was written as a distinct work sup- 
 plementary to that essay. The general conclusion is given at 
 the end of the dialogue (p. 439 a) ; that as words are merely 
 the images of things, it would be much better, even if we could 
 most perfectly learn the nature of things from their names, to 
 make the truth a criterion as well of itself as of its image. 
 
 61 The celebrated work of Home Tooke presents in many 
 ways a striking resemblance to the sophistical philology against 
 which the Cratylus was written. It was suggested more immedi- 
 ately by some legal quibbles originating in the author's trial for 
 high treason, just as the sophistical play upon words seems to 
 have been recommended as a part of the juggling rhetoric with 
 which the Athenian pleaders threw dust into the eyes of the 
 dicasts; and as Cratylus was a partizan of the materialism of 
 Heracleitus and Protagoras, so Home Tooke professedly adopts 
 the sensualism of Locke. In his philological method too he nearly 
 resembles those old etymologers; he endeavours to establish 
 
 * Schleiermacher, Einleitung zum Kratylos^ p. 15. 
 
110 THE PHILOSOPHY OF LANGUAGE. [bOOK I. 
 
 his views by an examination of his mother-tongue, chiefly, 
 if not entirely, unaided by a comparison of other languages. Of 
 his fundamental error with regard to the parts of speech we have 
 spoken in another place. His object is to establish nominalism 
 in its lowest and worst form, as an instrument in the hands of 
 materialism ; he endeavours to show that, in the English lan- 
 guage at least, all words, however abstract or general their pre- 
 sent use may be, are ultimately traceable to a meaning derived 
 from sensible impressions, and from this he concludes that these 
 words must still beunderstood, not in their present metaphorical, 
 but in their primitive literal sense, and consequently, that as 
 words are the signs of ideas, and all words refer only to sensa- 
 tions, we have no knowledge but through our sensations. But, 
 as Sir James Mackintosh somewhere asks, would it be just to 
 conclude that, because all words seem to represent, originally, 
 visible objects, there are no impressions of touch, smell, sound, 
 or taste in the human mind? This author, however, has no 
 deductions more unwarrantable in logic, or more truly con- 
 ceived in the spirit of the old Sophists, than those in which he 
 attempts, by twisting and materializing the meaning of some 
 of our most abstract terms, to subvert the principles of our inner 
 subjective morality. For instance, when he says, that '•^ truth is 
 nothing but what every man troiueth; that there is no such thing 
 as eternal, immutable, everlasting trutli^ unless mankind, such as 
 they are at present^ be also eternal, immutable, and everlasting ; 
 that two persons may contradict each other, and yet both speak 
 truth, for the truth of one person may be opposite to the truth of 
 another" (Vol. Ii. pp. 402, 3) — what is this but to reassert the 
 old docrma of Protaoforas, that the individual man is the standard 
 of all truth (πάντων μίτρον ανΟ'ρωπος) ? what is it but to leave 
 us to the dreary conclusion, which the follower of the So]3hists 
 must needs be contented with, that he has no community either 
 with men or with God, but remains, like another Prometheus, 
 bound to the isolated and comfortless rock of his own personal 
 consciousness, with all his social longings and irresistible first 
 convictions preying like a vulture on his soul*? 
 
 * See Schleiermacher's remarks in the Introduction to liis transla- 
 tion of the Theoitetm (p. 172 ad fin.). 
 
CHAP. III.] THE PHILOSOPHY OF LANGUAGE. Ill 
 
 62 The Diversions of Purley still maintains its ground, 
 censured by few*, and admired by many. To oppose the ex- 
 travagant nominalism and false philology of that work, and 
 others of a similar stamp, and to find the mean between an 
 excess of philological speculation and the superstitious realism, 
 which shrinks from all contact with philology, — this is the more 
 general object of the following pages. We bring forward 
 against vulgar materialism, a truer and more congenial philo- 
 sophy; we oppose to a narrow induction draAvn from a mixed, 
 wavering, and still spoken language, the carefully collected 
 results of the labours of three generations of scholars, applied 
 to a language copious, fixed, and comparatively pure^ aided by 
 the lights of comparative grammar, of a new era of the his- 
 tory of philology; in a word, we oppose to chimerical conjec- 
 tures the results of a science founded on facts. On the other 
 hand, our careful dissection of the whole body of inflected 
 speech will make it plain that, while words are merely outward 
 symbols, designating certain notions of the mind, those notions 
 do not stand related, in all cases, just as the words or inflexions 
 which express them, and that we cannot by means of mere 
 words convert into physical truth all that is logically and meta- 
 physically true. It is time that some attempt should be made 
 to show that the philosophy of language refuses its ministering 
 aid both to gross materialism and to superstitious fancy, and 
 that it stands forth as the chief confirmation of those systems, 
 by which human reason contributes to the support of religion 
 and morality. The word is destined to teach; let it cease to be 
 the instrument of deception. 
 
 * A Dutchman, who seems to have anticipated Home Tooke, was less 
 fortunate in the result of his experiment: " Un certain Hollandais, pen 
 affectionne a la religion, avoit abuse de cette verite (que les termes de The- 
 ologie, de Morale, et de Metaphysique sont pris originairement des choses 
 grossieres) pour tourner en ridicule la Theologie et la foi Chretienne dans 
 un petit dictionnaire iiamand, ou il donnoit aux termes des definitions on 
 explications non pas telles que I'usage demande, mais telles que serabloit 
 porter la force originaire des mots, et les tournoit malignement; ct comme 
 d'ailleurs 11 avoit donne des marques d'impiete, on dit qu'il en fut puni 
 dans le Raspel-huyss " (Leibnitz, Noiiveaux Essais sur rEntendement Hiimain, 
 p. 235). One might almost fancy that this was a description of our Eng- 
 lish etymologist, if the date and the punishment were more suitable. 
 
CHAPTER IV. 
 
 THE ETHNOGRAPHIC AFFINITIES OF THE 
 ANCIENT GREEKS. 
 
 C3 Ancient Greece must not be isolated. 64 Origin of the human race in Armenia. 
 65 Primeval civilization. QQ Mankind first spread into Asia Minor, and then 
 into Mesopotamia. 67 Widely-dispersed emigrations from the plain of Baby- 
 Ion. 68 Separation of the Aramaean and Iranian families in the vicinity of 
 the original settlements. 69 True classification of the human race presumes 
 an opposition between the central and sporadic branches only. 70 Old divi- 
 sion according to the descendants of Noah's three sons — how to be explained. 
 71 Spread of the Japhetic or Indo-Germanic race. 72 Order in which this 
 family entered Europe; (1) Celts, (2) Sclavonians, 3 {a) Low Germans, 
 3 (6) High Germans. We trace them back to Asia in the reversed order. 73 
 I. Germans, (a) Low Germans. 74 Saxons derived from the Sacie. 75 (b) 
 High Germans. 76 Origin of the name Ger/nan, 77 II. Sclavonians. Their 
 extensive diffusion. How connected with the Low Germans. Lithuanians 
 and Scandinavians. Getse and Daci. 78 Relations of the Sclavonic and 
 Teutonic tribes in general. 79 III. Celtic tribes. The two great dialects of 
 the Celtic. Causes of the insignificant ethnical position of the Celts. 80 IV. 
 Eastern members of the Indo-Germanic family. Iran defined. High and 
 Low Iranians. 81 Median origin of the Hindus shown by their ancient name. 
 82 The Low German tribes also derived from Media. 83 Meaning of the 
 term Sanscrit. 84 Antiquity of the Sanscrit language and literature. 85 The 
 High German tribes connected with the Persians or High Iranians. 86 The 
 Zend language a genuine remnant of old Persian. 87 V. The Latin and 
 Greek languages. 88 The Pelasgian or common element in Greek and Latin 
 was allied to the Sclavonian. 89 The additional or distinctive elements were 
 Lithuanian or Gothic in the Latin, and High German in the Greek language. 
 90 Ancient proofs of resemblances between the Greek and Persian. 91 The 
 Greeks and Germans had many features in common. 92 Their characteristic 
 designation may be traced in its course through Asia Minor and Eastern 
 Europe. 93 Proper classification of the Scythians. 94 Influence of the 
 Phoenicians on the early culture of the inhabitants of Southern Greece. 95 
 The name "Pelasgus" was not of Phoenician origin; but other names con- 
 nected with the arts of ancient Greece may be traced to the Phoenicians. 96 
 Characteristics of Hellenism. 97 Diflerences of dialect due to the prepon- 
 derance of Hellenic or Pelasgian elements respectively. 
 
 63 T>EFOE.E we commence our researches in the Greek lan- 
 guage, it will be as well to mention, for the information 
 of those readers to whom comparative philology is a new subject, 
 in what relation this language is supposed to stand in respect to 
 the other languages which we are about to compare with it. 
 The time is long past when we could surround Greece with a 
 
CHxVP. Π.] THE ETHNOGEAHIC AFFINITIES, ETC. 113 
 
 Chinese wall*, and content ourselves with surveying only as 
 much of its language , religion , and history as could be disco- 
 vered within these arbitrary limits. We cannot now content 
 ourselves with meagre disquisitions about ^olian or Dorian 
 dialects , or vague stories of Pelasgian serfs and Egyptian in- 
 vaders; we must look forth upon the great stage of universal 
 history, and consider whether these Greeks may not have had 
 some near relationship with those barbarians of Europe whom 
 they enlightened by their genius, and with those barbarians of 
 Asia whom they conquered by their valour; whether, in fact, 
 this very distinction of barbarian, or other-tongued , be not 
 after all the mere offspring of ignorance, which always perceives 
 the different before it can recognise the similar. It is now 
 incontrovertibly established that most of the inhabitants of 
 Europe, and a great number of the most ancient and civilized 
 tribes of Asia, speak, with greater or smaller modifications, the 
 same language; and the time may perhaps come when it will 
 appear as probable philologically, as it is certain historically, that 
 every language in the world has sprung from one original speach. 
 
 64 If we collect into one focus all the scattered informa- 
 tion respecting the birth-place of the human race, which we 
 can gather from tradition, from physiological considerations, and 
 from the exhaustion of contradictory hypotheses, we must feel 
 convinced that man originated in the temperate and fertile 
 regions which lie between the Southern extremities of the 
 Euxine and Caspian seas. Independently of all special induc- 
 tions , we should be inclined a priori to conclude , in accord- 
 ance with the general and systematic arrangements which we 
 notice in the procedure of creation so far as we are able to 
 trace its successive stages, that the human race would not be 
 planted upon the surface of the globe until life had become 
 both possible and easy to a creature so endowed, until the 
 earth had assumed its present , and , as we may conclude , its 
 permanent form, until the conditions of soil, atmosphere, vege- 
 table production, and animal life, to which our existence is still 
 liable, had been established on their present footing. And it is 
 
 * Krnse's Hellas^ Th, i. p. 395. 
 
114 THE ETHNOGEAPHIC AFFINITIES [bOOK I. 
 
 reasonable to think that man would be first cradled on some 
 plateau, which, — while it was raised above the lacustrine impu- 
 rities of the alluvial plains — was likewise free from an over- 
 o-rowth of wood, and well adapted for the cultivation of those 
 fruits and grasses, which furnish the necessary food of man. 
 There is no region in the world, which combines all these recom- 
 mendations so fully as the Armenian table-land lying to the 
 South and East of Mount Ararat. All tradition points to this 
 district. On the supposition that mankind originated there, we 
 may harmonize every linguistic phenomenon, and explain every 
 ethnographical fact. And the farther we depart in any direc- 
 tion, the greater are the difficulties in which we find ourselves 
 entangled. As for those on the other hand, who, recognising 
 Armenia as one birth-place of the human family, contend that 
 man was created independently in different parts of the globe 
 as they became favourable to his continued existence, we hold it 
 sufficient to say that such an hypothesis is unnecessary, since 
 the spread of population can be accounted for in a very satis- 
 factory manner without the assumption of more than one start- 
 ing point; and the differences of race, which we observe in 
 different parts of the globe, are not difierences of species incon- 
 sistent with one common origin. Besides, the hypothesis, that 
 man was created at different times and in different parts of the 
 world, would leave unexplained and inexplicable those proofs 
 of an original identity of language to which philology is daily 
 making additions of the greatest weight and importance. 
 Nothing short of necessity should induce us to seek for an 
 autochthony in different parts of the globe, which would break 
 the ties of blood-relationship that bind all men together; and 
 so far are we from being able to point out any such necessity in 
 this case, that all the attainable evidence clearly points in the 
 opposite direction. 
 
 65 "We conclude then that the first family of men lived in 
 the high but fertile country of Armenia, bounded to the North by 
 the true temperate zone, which there coincides with the fortieth 
 parallel of latitude. Little or no advantage is to be derived 
 from fanciful speculations respecting the so-called "ages of the 
 world," whether, with the old mythology we speak of a golden, 
 
CHAP. IV.] OF THE ANCIENT GBEEKS. 115 
 
 silver, brazen, and iron age*, or, with Grimm, arrange the dif- 
 ferent developments of society according to periods of stone, 
 brass , and ironf . Armenia was always a fertile and prolific 
 country. It abounded in corn, wine, and oil, and in those 
 animals which minister most directly to the comfort of man. 
 We cannot doubt therefore that the first society of human 
 beings, having every advantage of climate and situation, would 
 make a rapid advance in all the arts of life, and would soon lay 
 the foundations of civilization and citizenship. The earliest 
 records of the Semitic race tell us of the use of firej, of the 
 fabrications of metals§, of the computation of time||, and even of 
 navigation^. We read of cities built**, of fields cultivatedff, 
 of herds collectedJJ; and even the fine arts were not un- 
 known; at least, these early men were able to accompany their 
 native poetry with the sweet strains of instrumental music§§. 
 We may derive a similar picture of established social relations 
 and the adecxuate possession of material comforts from the 
 linguistic records of Arian civilization||||. The names of kindred, 
 in the oldest languages of the Indo-Germanic family , present 
 to us, when traced to their origin, a complete είδνλλίον of family 
 life in primeval IranH^. The father is "the protector," the 
 mother "the arranger" — "die kluge, verstandige Hausfrau," as 
 Goethe calls her***; the brother is "the helper," the daughter of 
 the house is called in relation to her brothers and sisters sva-sfri, 
 
 * Hesiod, I κ. ή. 108—199; cf. Ewald, Gesch. d. V.Isr.i. pp. 305 sqq. 
 
 f Gesch. der deutschen Sprache, i. p. 3. 
 
 J This is implied in the name Vs"^^"^ i. e. the light or splendour of 
 God {Gen. v. 15), if we seek its interpretation in the analogies furnished 
 by the other names: see Ewald, Gesch. d. V. Isr. i. p. 316. 
 
 § Gen. iv. 22. 
 
 II On the analogy between 'Hanok and Janus, and on the significance 
 of the number 365 attached to the former name, see Ewald u. s. p. 314. 
 
 ^ A comparison of the name of ί^;^ Jared {Gen. v. 15), with that of 
 the river Ύ}^^,, Jordan^ would seem to point to the first beginnings of 
 sea-faring. 
 
 ** Gen. iv. 17. ft iv. 2. 
 
 :li Ibid. §§ iv. 21. 
 
 nil Max Muller, Oxford Essays, 1856, pp. 14 sqq. 
 
 ^{^ Albrecht Weber, Indische Skizzen, 1857, pp. 8 sqq. 
 
 **'"•' Hermann und Dorothea, i. 22. 
 
 12 
 
116 THE ETHNOGRAPHIC AFFINITIES [BOOK I. 
 
 "the nearly related one," xolvov αυτάδελφον κ«ρα*, but in 
 regard to her functions she is duhitri, "the milkmaid," and, as 
 has been well remarked, "it discloses a kind of delicacy and 
 humour, even in the riidest state of society, if we imagine a 
 father calling his daughter 'his little milkmaid,' rather than 
 8uta^ 'his begotten,' or βΐία, the sucklingf.'" The family 
 group thus brought before us is surrounded by all the adjuncts 
 which can furnish us with a back-ground either of sedentary or 
 active occupation. "The house," says Weber J, "was secure, 
 provided with doors. Carriages and boats served for conveyance 
 over field and stream. The fields were equipped with ploughs. 
 Barley and wheat supplied meal and bread. Clothes, domestic 
 utensils, and weapons were there in abundance. Swords and 
 spears , knives and arrows were manufactured of bronze. In- 
 toxicating mead led to the merry song; large shells and reeds 
 served for music. Battle was a delight, and the feeling of race 
 so strong, that the word barbarian — the inarticulate speaker — 
 was used even in that primitive age as a designation of the 
 foreigner who spoke another language. The conquered enemy 
 became a slave. At the head of a commonalty stood a ruler, 
 protector, lord, the leader in battle and the judge in peace." 
 
 66 How many years elapsed before this first establishment 
 of social life spread beyond the limits of Armenia, we have no 
 means of guessing. But tradition distinctly tells us that prime- 
 val civilization first extended itself to Asia Minor , and after- 
 wards to Mesopotamia. Thus the earliest emigrant is carried to 
 Lydia§; and the city of Iconium|| in Lycaonia claims for its 
 
 * This is our interpretation of the word: Journal of Phtlologt/, 
 Vol. II. p. 357. According to MuHer svasar is "she who pleases or con- 
 soles," from svasti, "joy or happiness" {Oxford Essays, 1856, p. 16), and 
 Weber (Ind. Sk. p. 9) renders it "the careful one," die sorgliche. 
 
 t Muller, Oxf. Ess. p. 16. 
 
 X Ind. Sk. p. 9. 
 
 § The name ^ιλ {Gen. iv. 16) seems to be only another articulation 
 of τη? [Gen. x. 22); see Ewald u. a. p. 315. 
 
 II Steph.Byz.s. v.: '/κόνίον, ηόλιςΑνΛαονίαςπ^οςτοΐςο^ΟίςτονΤαν- 
 ρον.φααΐ d' οτι ην τις Άννατιός, ος ίζηαεν νπίρ τα. τριακόΰΐα ίτη. τους δε 
 TTt'^il ^αι/Γίΐ/σασΌ-αί, ί'ωζ τίνος §i06cG^at. έδόϋ-η δε χ^ηομός, οτι τούτον τελευ- 
 
CHAP. IV. J OP THE ANCIENT GEEEKS. 117 
 
 foundfir, Annacus, or 'Hanok, the first author of an improved 
 calender. In Mesopotamia, again, we can trace the stream of 
 primitive civilization, as it descended the Tigris, skirting the 
 mountains of Kurdistan, until it established itself, in full-blown 
 luxury, at Babylon on the Euphrates. These facts are sup- 
 ported by consistent tradition; but we might have inferred, 
 from general considerations, that such was the case. It was 
 likely that the first migrations from Armenia would spread 
 towards the West, because Asia Minor was not only very ac- 
 cessible, but presented also the same conditions of soil, climate 
 and elevation as the parent-country; whereas the rich alluvial 
 plains of the "Two Rivers" would not at first invite those who 
 had been accustomed to a higher region, a more moderate tem- 
 perature, and a purer air. When, however, the change of abode 
 had once taken place, it is easy to understand how the growth 
 of wealth , the formation of mightier empires , and the erection 
 of gigantic cities , with their usual accompaniments of tyranny 
 and vice, would flow from the new practice of living in open 
 plains, and from a superabundance of employment and of the 
 rewards of industry. 
 
 67 So long as the primitive population of the globe was 
 confined to Armenia and its two colonies in Asia Minor and 
 Mesopotamia, we find no traces of any differences of nation or 
 language. It was on the lower Euphrates that the multitudes 
 became too numerous for the soil; and from thence they 
 streamed away in successive parties, scattering their detached 
 and isolated bands over the whole surface of the globe. Hun- 
 dreds, nay, thousands of years may have passed away, while 
 these emigrants were wandering farther and farther from home, 
 and becoming more and more forgetful of the civilization and 
 social enjoyments which they had left behind them. It is rea- 
 sonable to suppose that many of them who set out on this dreary 
 and endless journey had committed crimes which made them 
 
 τήααντος πάντεζ διαφ^αρήβονται. ol δε Φ^νγες άτιονααντες ί&ρήνονν ΰψο- 
 δρως. ο&εν καΙ παροιμία, ''το επΙ Άνναν,οϋ τίλανϋειν" ίπϊ των λίαν 
 οίγ,τυζομύνων. And then he proceeds to connect this with a legend about 
 the deluge of Deucalion. Meineke suggests Ναννατιός; but the old reading 
 seems to be the best. See Ewald u. s. p. 314. 
 
118 THE ETHNOGEAPHIC AFFINITIES [bOOK I. 
 
 anxious to shun communion with their fellows, and many a Cain 
 transmitted to his wandering descendants the indelible impress 
 of degeneracy and sin. Not unnaturally those who went farthest 
 would fare worst, except in those cases where the ocean inter- 
 posed a barrier to all further progress, and where the stream of 
 population was dammed up in some well -watered and fertile 
 country, which soon brought man back to the city-life and social 
 habits of his forefathers. Perhaps the earliest case of this kind 
 was the empire of China. At a later period the narrow isthmus 
 of Darien produced a similar effect in Mexico. In general, Ιιολυ- 
 ever, the dispersion went on widening itself, and men whose 
 ancestors had been on the same footing in regard to speech, 
 colour, and frontal development, became Mongols, Tungusians, 
 Mantchoos, and Samoy eds in Asia ; Finns, Lapps, and Euskarians 
 in Europe; Negroes and Caffres in Africa; and Red Indians in 
 America; to say nothing of the Papuans, the Tasmanians, and 
 the more widely-scattered Polynesians. 
 
 68 Meanwhile, modifications were taking place nearer 
 home. Close to the original birth-place of man, two sister-races 
 formed themselves, with equal qualifications both of body and 
 mind, and divided between them, in nearly equal proportions, 
 the great work of developing the human intellect. The geogra- 
 phical line of demarcation, the boundary-line and wall of parti- 
 tion between their first abodes, is furnished by the mountains of 
 Kurdistan and by the Persian Gulf. To the South and West of 
 this, the Aramaic race occupied at a very early period Mesopo- 
 tamia, Syria, Palestine, Arabia, Egypt, and all the North of 
 Africa. To the East, the Iranian race was more slowly develop- 
 ing itself on the great Western plateau of Asia, from whence it 
 sent off successively streams of colonists, who carried the original 
 language and the original appetences for high mental cultivation 
 into India to the South East, and round by the North coasts of 
 the Caspian and Euxine seas into Europe. We are precluded 
 by the nature of this work from considering all the questions in 
 physical geography, psychology, and history, which are connect- 
 ed with the ethnology of these civilized races ; and in the philo- 
 logical part of the question, on which alone we can enter, we are 
 obliged to limit our investigations, as far as possible, to those 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 119 
 
 parts of the subject which are most immediately connected 
 with the illustration of the Greek language in particular. But 
 even with this restricted range of speculation, it will be neces- 
 sary to engage in a survey which a few years ago would have 
 been thought extravagantly wide and foreign to the main ques- 
 tion. Accessions of knowledge bring with them expanded and 
 comprehensive views. There was a time when it was perfectly 
 natural to regard the varieties rather than the affinities of human 
 speech. It was seen that there were differences ; but the points 
 of contact were unobserved. The time is rapidly approaching 
 when the discrepancies will appear inconsiderabl^i, and when 
 the marks of a common origin and of a family -likeness will 
 engross all our attention and interest, 
 
 69 At present , however , the languages of the earth are 
 divided into great families, which present remarkable points of 
 difference. Some years ago two eminent philologists concurred 
 in recognising three great classes or families of languages. They 
 are thus distinguished by A. W. von Schlegel {Observations sur 
 la langue et litterature Proven^ales, p. 14): Les langues sans 
 aucime structure grammaticaie ; les langues qui emploient des 
 affixes; et les langues a inflexions; and this arrangement is 
 adopted by Bopp (yergleicL• Gramm. p. 112, 3) with the follow- 
 ing explanation: (1) Languages with monosyllabic roots, but 
 incapable of composition , and therefore without grammar or 
 organization: to this class belongs the Chinese, in which we have 
 nothing but naked roots, and the predicates and other relations 
 of the subject are determined merely by the position of the 
 words in the sentence ; (2) Languages with monosyllabic roots, 
 which are susceptible of composition, and in which the grammar 
 and organization depend entirely on this. In this class the 
 leading principle of the formation of words lies in the connexion 
 of verbal and pronomial roots , which in combination form the 
 body and soul of the language: to this belongs the Sanscrit 
 family, and all other languages not included under (1) and (3), 
 and preserved in such a state that the forms of the words may 
 still be resolved into their simplest elements ; (3) Languages 
 which consist of dissyllabic verbal roots, and require three con- 
 sonants as the vehicles of their fundamental signification: this 
 
120 THE ETHNOGEAPHIC AFFINITIES [BOOK I. 
 
 class contains the Semitic languages only; its grammatical 
 forms are produced not merely by composition, as is the case 
 Avith the second , but also by means of a simple modification of 
 the roots. 
 
 More recently, it has been thought convenient to divide the 
 known languages of man into five difi*erent groups or dynasties. 
 (1) The Indo-Germanic, corresponding to the second family in 
 the above classification. (2) The Syro-Arabian, corresponding 
 to the third family. (3) The Turanian, or Ugr ο - Tartarian. 
 (4) The Chinese and Indo-Chinese , corresponding to the first 
 family. (5). The languages of Central and Southern Africa. 
 We still prefer a tripartite division , which in effect is capable 
 of further arrangement in two groups of languages ; and we 
 think that the following is the simplest nomenclature. The 
 two groups may be called (A) the central, and (B) the sporadic. 
 Group (A) contains (1) the Iranian languages, corresponding 
 to the Indo-Germanic, or Sanscrit family; and (2) the Aramaic 
 languages, corresponding to the Semitic or Syro-Arabian fa- 
 mily. Group (B), or the sporadic family, includes (3) the 
 Turanian, the Chinese, and all those other languages which 
 w^ere scattered over the globe by the first and farthest wan- 
 derers from the birth-place of our race. . According to this ar- 
 rangement, the first two families are classed together as con- 
 stituting one group of languages closely related in their material 
 elements,, and differing only in the state or degree of their 
 grammatical development. The third family stands by itself, 
 as comprising all the disintegrated or ungrammatical idioms. 
 By the researches of Dr. Pricliard and others, approximations 
 have been already made to the establishment of family affinities 
 between the different members of this sporadic group of lan- 
 guages. At present, however, they must be regarded as belong- 
 ing to a region of phenomena not yet completely explored by 
 science, and surrounding like a cloud the clearly-developed and 
 central mass of Aramaic and Iranian idioms. 
 
 According to a mode of classification which we have else- 
 where introduced*, these central languages differ rather in re- 
 gard to their state or condition than in regard to the materials 
 
 Maskil le Sojpher, pp. 3, 4. Above, §. 49. 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 121 
 
 of whicli they are composed. By the state or condition of a 
 language we mean, as we have already explained the term, the 
 degree of detriment which the cultivation of syntax has caused 
 to its etymological structure. The old languages of the Iranian 
 or Indo-Germanic family belong to the first and second classes 
 mentioned above. The Aramaic, Semitic orSyro-Arabian idioms 
 all belong to the third class. 
 
 70 The relations between the two great branches of the 
 central mass of languages may be established by a theory rest- 
 ing on scientific inductions* ; and the result is in close accord- 
 ance with the ethnographical pedigree given in the tenth chapter 
 of Genesisf. That ancient record divides the nations then known 
 to the Israelites into three classes, derived resjoectively from the 
 three sons of Noah, — namely, — Shem, Ham, and Japheth. But 
 although the subdivision is formally tripartite, the slightest exa- 
 mination of the document will convince us that a more inti- 
 mate affinity is presumed between the descendants of Shem and 
 Ham, than between either family and the tribes which claimed a 
 descent from Japheth. For example, the Arab tribes desig- 
 nated as Havilah and Sheba are derived from Shem as well as 
 from Ham. 'In fact, as we have elsewhere said, the relationship 
 between the Shemitic and Hamite nations is fully recognised, 
 but the latter are described as the previous occupants of the 
 difi'erent countries into which the Aramean tribes afterwards 
 forced their way. To repeat what we have stated on former 
 occasions^, the diffusion of the Aramaic race seems to have been 
 according to the following stages. After the aborigines of Armenia 
 had extended their territory into Asia Minor, and while the 
 
 * On the Sclavonians, as forming the point of contact between the 
 Semitic and Indo-Germanic races, see our essay On two unsolved Problems 
 in Indo-Germanic Philology, Report of Brit. Assoc, for 1851 , p. 138. And 
 compare Mr. W. K. Sullivan's paper On the Influences of Physical Causes 
 on Languages, Sfc, Atlantis, Jan. 1850, p. 121. 
 
 f Renan maintains {Histoire des Langues Semitiques, p. 38) that the 
 xth chapter of Genesis groups the diiFerent tribes not by race but by 
 climate; that its basis is geographical and not ethnographical: that Ja- 
 phet, Shem, and Cham represent three zones, the northern, the inter- 
 mediate, and the southern; and that no one of these names can desig- 
 nate a race in the scientific signification which we give to that word. 
 
 J Quarterly Review, No. xlv. p. 173; Maskil le Sopher^ p. 35. 
 
122 THE ETHNOGRAPHIC AFFINITIES [bOOK I. 
 
 population of Iran was beginning its development, two streams 
 of population descended from the mountains ; and, leaving the 
 desert between them, founded, in Mesopotamia to the left and in 
 Palestine to the right, wealthy and civilized communities, which 
 cultivated at an early period all the arts of city-life and practised 
 not a few of its attendant vices. From the left-hand colony, 
 which included the empire of Nineveh, and subsequently that of 
 Babylon, a further stream proceeded Southwards; and having 
 on its way established the rich kingdoms of Havilah and Sheba 
 in Arabia Felix, it ultimately carried its traditionary religion 
 and social culture into Upper Egypt, where it came in contact 
 with a kindred empire founded in Lower Egypt by those who 
 had taken the right-hand course. All these great diffusers of 
 sensual comfort and irreligious civilization are classed together 
 in the Old Testament as Hamites^ or descendants from Noah's 
 godless son, and are opposed to the ShemiteSj that is, to the 
 Hebrews, Assyrians, Syrians^ and Arabians, who subsequently 
 descended from the mountains of Aram. But there is every 
 reason to believe that all these nations spoke languages which 
 exhibited the same peculiarities, and differed only as dialects 
 of the same idiom ; and , as we have elsewhere shown* , their 
 apparent trigrammatism, their etymological disintegration, and 
 the tertiary condition in which their oldest remains arc found, 
 must be referred to the constant intermixtures , re-unions, and 
 confusions produced by the emigrations and conquests of the 
 different sections of this important family. 
 
 By means of a scientific analysis it is possible to point 
 out the existence of monosyllabic roots in Hebrew and in the 
 other Syro- Arabian languages no less than in the members of the 
 Iranian or Indo-Germanic family (§209). But though we must 
 not neglect the various contacts and affinities of the two branches 
 of our first and central group , the present is not the proper 
 occasion for a full discussion of the Semitic idioms; and we 
 must content ourselves with a survey of the branch to which the 
 Greek language belongs. 
 
 71 In describing the spread of the descendants of Japheth 
 the Book of Genesis enumerates only those tribes whose settle- 
 
 * Maskil le Sopher^ p. 36. 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 123 
 
 ments were in AsiaMinor, in the South-eastern parts of Europe, 
 and on the Mediterranean. The immediate offspring of Japheth, 
 in other words, the main divisions in this family of nations, are 
 the Cimmerians (Gomer)^ Scythians (^Magog)^ Medes (Mddai)^ 
 lonians (Jdvan)^ Tibareni (Tuhal)^ Moschi (^Meshek)^ and Thra- 
 cians (Tirag). Besides these, the Bithynians (^Ashkenaz)^ Sar- 
 matians (JRiphatli)^ and Armenians (TogarmaJi)^ are mentioned 
 as sons of Gomer, or offshoots of the Cimmerii; and not only 
 Hellas C'Helishdh')^ but other places in the Mediterranean, with 
 which the Phoenicians trafficked, even the distant Tartessus in 
 Spain, are said to be peopled by sons οι Jdvan, or lonians. 
 This of course is a one-sided survey of the spread of this great 
 family, though very valuable as far as it goes; and we must 
 take a much more comprehensive view of the population of 
 Europe, if we wish to understand the relation subsisting between 
 the Greek language and the other members of the class to 
 which it belongs. 
 
 This great class of languages , extending from India to the 
 British Isles, has been called the Japhetic, Arian, Iranian, San- 
 crit, Indo-European or Indo-Germanic family. We shall adopt 
 the last of these names, because it points at once to the two 
 most important branches of the family, the Indian and Teu- 
 tonic languages, and is free from the vagueness Avhich attaches 
 to the term Indo-European; for there are languages in Europe 
 which have no established affinity with this family. Besides, 
 we believe that all the members of the family are deducible 
 from two great branches corresponding to these, and the rigor- 
 ous examination to which they, in particular, have been sub- 
 jected, places them in a prominent position in regard to the other 
 idioms, which are not only less important, but also less known. 
 
 72 If we consider the elements of the population of Europe, 
 according to the order in which they were successively added to 
 the first sprinkling of scattered Turanian tribes which they drove 
 before them to the mountainous extremities of the continent, we 
 can hardly fail to arrive at the following results. The first 
 emigrants from Asia were sons of Gomer,— Celts and Cim- 
 merians, — who entered this continent from the steppes of the 
 Caucasus , and passing round the northern coasts of the Black 
 
124 THE ETHNOGRAPHIC AFFINITIES [bOOK I. 
 
 Sea, not only spread over the whole of Europe, specially to 
 the South and West, but also recrossed into Asia by the Helles- 
 pont, and conquered or colonized the countries bordering on 
 the South of the Euxine. The next invaders were the sons 
 of Magog, — Scythians, Sarmatians, or Sclavonians, — who are 
 generally found by the side of the Celts in their earliest settle- 
 ments. They more fully occupied the East of Europe, but 
 though they contributed largely to the population of Greece and 
 Italy, they do not appear to have spread beyond the Oder in the 
 North, or to have established themselves permanently in the 
 Alps, or in the Middle-highlands of Germany. The final settle- 
 ment of Iranians in Europe was that of the Teutonic races, con- 
 sisting first of the Low Germans, who, starting from the regions 
 between the Oxus and the Jaxartus, burst through the Scla- 
 vonians , and formally settled themselves in the Nord-west of 
 Europe ; and secondly of the High Germans, who subsequently 
 occupied the higher central regions, having also contributed an 
 important, and perhaps the most characteristic, element to the 
 population of Hellas. In considering these tribes separately, 
 we shall travel back to their original abodes in Asia, in an 
 order the reverse of this, and shall take as our starting-point 
 those who entered Europe last, and travelled farthest. 
 
 73 We begin, then, with the German languages, which are 
 of the highest interest to us , because our own language in its 
 fundamental element, and the oldest part of the Greek, to the 
 elucidation of which our present efforts are mainly directed, 
 belong to the oldest branch of this set. The German languages 
 are divided into two great branches, usually known as Low 
 German and High German. The former, which is the older, 
 was spoken in the low countries to the north of Europe ; the 
 latter was the language of the more mountainous districts of the 
 South : whence their distinctive names. There is every reason to 
 conclude that the Low Germans entered Europe from Asia long 
 before the High Germans, and that they were driven onwards 
 to the north and east by the overwhelming stream of the sub- 
 sequent invasion : this appears not only from their geographical 
 position , but also from the internal evidences of relative anti- 
 quity, furnished by the languages themselves. 
 
CHAP. ΙΥ.] OF THE ANCIENT GEEEKS. 125 
 
 The Low German includes (1) the Scandinavian languages, 
 Icelandic, Swedish, and Danish*; (2) the Low German dialects, 
 peculiarly so called, Anglo-Saxon, Frisian, Flemish, and Dutch; 
 (3) the old Gothic, or, as Bopp calls it, the German Sanscrit. 
 We mention the languages in this order, namely, those farthest 
 from Asia first, not only on account of the position, but also 
 because the languages in their internal structure stand in this 
 relation of antiquity. 
 
 74 With regard to our own language, the Anglian and 
 Scandinavian elements have overpowered the Frisian. The 
 Anglo-Saxon conquest of England was mainly effected by two 
 great branches of the Saxon race— the Frisians and Angles. 
 The former were more specifically termed "Saxonsf," but this 
 was the general designation of all the Six Settlements. While 
 therefore , with reference to the ultimate predominance of the 
 Anglian branch, we rightly designate ourselves as English (An- 
 glians), the Celts, whom the Teutonic invaders drove into the 
 mountains, not unnaturally applied to their conquerors the name 
 of Sassenach (Saxons) , which included both the Frisians and 
 the Angles. The Saxons, like the Germans, seem to have 
 derived their names directly from Asia. A tribe of the Sacse, 
 who dwelt by the Caspian, and were therefore, as will be seen, 
 Low Iranians , occupied Bactriana and the most fertile part of 
 Armenia, and extended in a westerly direction towards the 
 Euxine ; they were called Sacassani (according to Pliny, Hist. 
 Nat. VI. 11), and their country Σακαόηνή (Strabo, p. 511); and 
 it is supposed by the most eminent antiquaries that these were 
 no other than the Saxones, i. e. Sacasimu, or "Sons of the 
 Sac^." 
 
 75 The High German is simply divided into three classes, 
 or rather three stages of existence, the Old, Middle, and New 
 High German. The latter^ which took its origin in Uj^per 
 Saxony, and which owes its present position, as the written 
 language of all Germany, to the influence of Luther, who was 
 
 * The eastern afiinities of the Scandinavians have been well indi- 
 cated by Dr. G. W. Dasent in the Introduction to his Popular Tales 
 from the Norse, London, 1851. 
 
 f See Cambridge Essays, 1856, pp. 45 sqq. 
 
126 THE ETHNOGEAPHIC AFFINITIES [BOOK I. 
 
 from Upper Saxony, is probably the modern representative of 
 the language which was spoken on the confines of Upper and 
 Lower Germany, and this may account for its presenting, in some 
 degree, the combined features of the two sets of languages. 
 
 76 Many of the ancients believed that the epithet Germa- 
 nus, Γερμανός, by which they described the cognate inhabitants 
 of central Europe, was merely the Latin adjective, which denotes 
 brotherhood and kindred*, and the Romans often indulged in 
 a play of words arising out of this misconception respecting a 
 renowned ethnical appellationf . "We need not trouble ourselves 
 with the conjectures of a period, when philology was non-existent, 
 and w^hen it was natural for proud and ignorant men to seek an 
 interpretation of foreign words in the nearest corresponding 
 sounds of their own language. It is desirable, however, to in- 
 quire into the origin of this celebrated name ; and we ho^ie to 
 be able to show that it is not only an indigenous title, but that 
 it was brought by the High Germans from their settlements in 
 Asia, and left by them, both in itself and in a synonym, among 
 their earliest European colonists — the Dorian Greeks. This lat- 
 ter part of the investigation we will reserve till its proper place, 
 when we come to speak of the Persians and Greeks: in the 
 mean time, it will be right to show that the name borne by all 
 the Teutonic tribes was itself a Teutonic word. 
 
 In the valuable essay on the Germans, which Tacitus wrote 
 as an Excursus or appendix to his Historice, we are informed 
 that the name Germanus was originally confined to a particular 
 branch of the Teutonic race Qiationis nomen non gentis, c. m.), 
 from which, like the Hellenic name in Greece, it spread by 
 conquest or imitation, to the other neighbouring and cognate 
 tribes. As warriors, the Teutons took particular pride in calling 
 themselves emphatically "mm." According to Tacitus, they 
 
 * Strabo, 290 : διο δίν,αιά μοί δοκοϋοι Ψοτμαΐοί τοντο αύτοΐς &έΰϋ•(χι 
 τοϋνομα, ώς αν γνηαίονςΓαλάτας φράζειν βονλόμενοί' γνηαιοι γαρ οΙΓε^μανοΙ 
 κατά την 'Ρωμαίων δίάλεκτον. We find a reference to this error of the 
 Romans in Plutarch, Mnriiis, c. xxxiv., where Gcrmani as applied to the 
 Τεύτονες is rendered αδελφοί. 
 
 ■[ Quinctil. vni. 3, § 29: Cimbri hie fuit, a quo fratrem nccatum hoc 
 Ciceronis dioto notatum est, Germanum Cimber occidit. Velleius Pater- 
 culus, II. 67: De Germanis non de GaUis, duo triumphant consules; where 
 there is a double pun. 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 127 
 
 traced back their origin to Mannus^ the son of Tuisco^ the son 
 of Earth, that is, to a brave warrior, the child of their abo- 
 riginal god of war. The three great divisions of the nation 
 were referred to three sons οι Mannus, from whom they were 
 styled Iscwvones, Ingcevones, and Herminones, corresponding to 
 the Franks, Saxons, and Thuringians of a later age. What- 
 ever conclusion we may adopt respecting the origin and signi- 
 fication of the two former designations*, we can hardly doubt 
 that the ancient name of the Thuringians, as given by Tacitus, 
 is merely the compound Eerr-mann augmented by a formative 
 syllable: and both the modern and ancient title of the same 
 tribe are combined in the name of the Her-mun-duri. We 
 find the same element in the names of the Ala-manni, or " all 
 men," and the Marco-manni, or "border-men;" and we can 
 hardly doubt that as the term ^^ma?i^^ is thus appropriated 
 to the free Teutonic warrior, the word Ger-man is strictly 
 analogus to Gar-dane, which we find in Beowulf, whether the 
 first word signifies "a spear," or is merely the intensive par- 
 ticiple gar, "quite, entirelyf." In the latter case the Gar- 
 manner might be defined, in a certain sense, as the γνήΰίΟί 
 Tamovsg; and though this would not justify us in considering 
 the term as a foreign epithet imposed by the Romans, it might 
 serve to confirm the view which we have elsewhere taken| of 
 the old Roman name Herminius, namely, that this was a word 
 of Teutonic origin. 
 
 * See Grimm, Gesch. d. deutsch. Sj)r. pp. 824 sqq. We believe that the 
 element Isk- in Tscoevones is identical with the signilieant syllable Ask- in 
 the Ash-kenaz of Genesis x. 3, in the Phrygian Ascanius, and the ethnical 
 names Πελ-αΰγός, Oscus, &c., and we think that the element Lig- is that 
 which is found in the designation of the Ang-li. With regard to the letter 
 % in Isk and Ing^ it is worthy of remark, that though we use α or e in writing 
 the words Angles or English, we speak of England as Ing-land, and this 
 pronunciation is supported by the Runic inscriptions (Camhridge Essays, 
 1856, p. 43). Are the words Mann-ing and Mensch = MaJin-isk, indica- 
 tive of contacts between the Germans or Manner emphatically so called, 
 and their neighbours the Ingwvo7ies and Iscwvones f 
 
 t See the different opinions collected in Weishaupt's edition of the 
 Germania, pp. 135 sqq. Grimm has lately added to these a Celtic etymo- 
 logy — namely, from gairm, plur. gairmeanna = ruf, ausruf, so that Ger- 
 mani = βοην άγα&οί (Gesch. d. deutsch. Spr. p. 787). 
 t Varronianus, p. 25. 
 
128 THE ETHNOGRAPHIC AFFINITIES [bOOK I. 
 
 On the whole, we feel disposed to consider the term Manner 
 in general, and the compounds Ger-mmmer and Herr-mdnner 
 in particular, as originally derived from the last, most concen- 
 trated, and most warlike invaders of Europe — the Thuringian 
 or Eastern Teutons. But as they dispossessed or conquered 
 and settled amongst the Sclavonians on the one side, and the 
 more nearly-related Saxons on the other, they imparted to the 
 whole district the name which they brought with them from 
 the mountains of Iran. In the time of Tacitus it is fair to 
 conclude that, although the Herminones were the only pure 
 High Germans, the i\NO other representatives of the sons of 
 Mannus were not free from many admixtures and contacts with 
 these vigorous warriors. 
 
 77 The most widely-extended idiom of the Indo-Germanic 
 family is the Sclavonian : it is spread over a broad surface of 
 Europe and Asia, from the Pacific to the Baltic, from the Adriatic 
 to the Arctic sea. The different tribes who spoke this language 
 Λvere known to the ancients under the names of Rhoxolani*, 
 Krobyzi-j-, Sarmatse or Sauromatse, Pannonians, Illyrians, and 
 Veneti or Wenidae: at present it is spoken in Europe by the 
 Russians and Rusniaks, the Bulgarians, Servians, Bosnians, 
 Dalmatians, Croats, the Wends and Sorbs in Lusatia and 
 Saxony, the Slowaks in Hungary, the Bohemians, Moravians, 
 Poles, and Silesians. 
 
 Closely connected with the Sclavonian, but not so widely 
 diJBiused, are the Lithuanian languages; this set comprises the 
 Lithuanian proper, Lettish, and old Prussian. From grammatical 
 considerations, which we cannot here enlarge upon, we have no 
 hesitation in placing Sclavonian and Lithuanian, the agreement 
 of which is universally acknowledged, in the same class with the 
 
 * The Rhoxolani are mentioned by Strabo (pp. 114, 294, 306) as the 
 last of the known Scythians. From them Russia derives its name. "The 
 Finns distinguish the Muscovites by the name of Rosso-lainen, or Russian 
 people, and call themselves and nations of their own kindred Suoma- 
 lainen. The word Rosso-lainen, heard and written by a Greek, would be 
 Rhoxolani" (Prichard, Celtic Nations, p. 16). 
 
 -{- The Krobyzi mentioned by Herodotus (iv. 49) are supposed to be 
 the same with the Russian Kriwizen. 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 129 
 
 oldest Low German dialects. In fact, wherever the Low Ger- 
 man has escaped the overruling influence of the sister Teutonic 
 dialect, it has been placed in such close contact with the Sclavo- 
 nian, that it is often easier to pass from the Low German to the 
 Sclavonic form, than, from the former, to reproduce the High 
 German. Of the Low Germans who were thus Sclavonized, the 
 Lithuanians were almost incorporated in the older race. The 
 Scandinavian tribes, though they had escaped all direct contacts 
 with their High German brethren, were much less tainted with 
 Sclavonism than the Lithuanians, and exhibited in as pure a 
 form as possible the distinctive characteristics of their progenitors 
 the Getae or Goths. But the Gothic affinities of the Lithuanians 
 have not been forgotten. Their proper name is Saino-getce; 
 and the Prussian branch of this tribe call their neighbours, the 
 Polish Lithuanians , by the name Gudas or Guddas. Not to 
 speak at present of their Asiatic abodes, we find the Sclavonians 
 and Lithuanians side by side on the very threshold of Europe. 
 For we have no hesitation in recognising the Sclavonian race 
 in the original Thracians, and the Gothic name appears in that 
 of the Getse. Grimm has shown* that the neighbouring Daci 
 may have borne the original name of their northern descendants, 
 the Danes, who are an important scion of the Low German race. 
 
 78 If these opinions are well founded , we shall have no 
 difficulty in settling the relations of the Teutonic and Sclavonian 
 tribes. The latter, it appears, originally occupied the greater 
 part of Eastern Europe. They were first encroached upon by 
 the Gothic orLow German tribes, who left them in uninterrupted 
 possession of Thrace to the South and of Sarmatia to the North, 
 but deprived them of all their central and western habitations. 
 The High Germans finally pushed their way through the Low 
 Germans , and , first occupying in force the eastern part of the 
 district which had been already Teutonized, gradually extended 
 themselves to the West , where they were assimilated more or 
 less to the Low Germans who had gone before them. Along the 
 coasts of the Baltic Sea and German Ocean, and in Scandinavia, 
 the Gothic branch remained more or less independent of the 
 other Teutonic race, but in Lithuania especially they were very 
 
 Gesch. d. deutschen Spr. i. p. 192. 
 
130 THE ETHNOGRAPHIC AFFINITIES [BOOK I. 
 
 much influenced by their Sclavonian neighbours. The High 
 Germans had in some districts to give back what they had taken 
 from the earlier tribes , especially in Bohemia, but to the West 
 they carried forward their predominance till at last they crossed 
 the Rhine, and bestowed the name of the Frankish confederacy 
 on the most thoroughly Latinized of the Roman provinces. We 
 may therefore say that the Lithuanians were Low Germans highly 
 Sclavonized ; that the Scandinavians contained the Gothic ele- 
 ment in its purest form; that the Saxons or Ingcevones were Low 
 Germans untainted by Sclavonism, and but slightly influenced 
 by High Germanism ; that the Franks or Isccevones were Low 
 Germans, over whom the High Germans had exercised con- 
 siderable control*; and that the Thuringians or Hei^minones were 
 pure High Germans, in the full vigour of their active opposition 
 to the tribes among which they had settled. 
 
 79 The Celtic nations , the claim of whose speech to a * 
 place in thelndo-Germanic sisterhood is now fully establishedf , 
 appear to have been the oldest inhabitants of Europe, but, by 
 the pressure of subsequent immigrations, they have been thrust 
 out to the extreme corners of the continent; and Arndt has 
 endeavoured to show that they were also connected, to a certain 
 extent, with the Finns, the Samoiedes, and the Mongols, nations, 
 like themselves, detruded to the uttermost parts of the earth. 
 
 * It is worthy of remark that the Jutes, who settled in Kent, referred 
 the foundation of their kingdom to a mythical ^sc (i. e. Isk); and few 
 scholars will fail to recognise, in the dialectical synonyms Hengist and 
 Horsa, the last faint traces of a combination or fusion of High and Low 
 Germans, the two tribes represented being really the Frisians and Angles 
 (see Cambridge Essays^ 1856, p. 47). 
 
 f In spite of occasional outbreaks of the rash and unscientific specu- 
 lation, which has too often been the characteristic of Celtic philology, 
 the relations of the Gaelic and Cymric tribes, and the Indo- Germanic 
 affiliation of these ancient and interesting languages have at last been 
 brought under the control of a sober and accurate discussion. The admi- 
 rable papers by Mr. Garnet {On the Languages and Dialects of the British 
 Isles, Essays, pp. 147 sqq,), the Grammatica Geltica of Zeuss, and the 
 useful little work by Roget de Belloguet (Ethnogenie Gauloise^ Paris, 185&), 
 have placed the whole subject within the reach of the general student 
 of comparative Grammar. We have suggested some special combinations 
 in the Cambridge Essays for 1856, pp. 33 sqq. 
 
CHAP. IV.] OF THE ANCIENT GHEEKS. 131 
 
 There are two great dialects of the Celtic, which are thus 
 exhibited by the most recent writers on the subject*. 
 
 (I) The Gallic or British, comprehending 
 
 (a) the Cymric or Welsh; 
 
 (b) the Cornish, which is extinct ; 
 
 (c) the Armorican, or dialect of Brittany (Bas Breton). 
 
 (II) The Gaelic (Gadhelic) or Erse, comprehending 
 («) the Fenic or Irish; 
 
 (b) the Highland Scottish (Gaelic) ; 
 
 (c) the Manx in the Isle of Man. 
 
 From this enumeration it will be seen that the remains of 
 the Celtic language are now found only in nooks and corners of 
 western and insular Europe. But the same evidence which 
 establishes the'Asiatic origin of the sons of Gomer, proves also 
 their original diffusion throughout the whole of this continent. 
 Being, next to the Turanian tribes, the first inhabitants of this 
 part of the globe, they were either absorbed or driven onwards 
 by the subsequent streams of population. In Spain to the South- 
 west, and in the North of Scandinavia, they were swallowed up 
 in the more closely-packed Turanian tribes who preceded them 
 in those directions. The Basque or Euskarian language, in par- 
 ticular, which still remains isolated in the North-west of Spain, 
 may be called a Celto-Finnish language. In Gaul, however, 
 and in the British Isles the Celtic element predominated over all 
 preceding ingredients, and long kept itself free from subsequent 
 admixtures. Indeed, Celtic tribes appeared among the Germans 
 during the better known historical periods. The Marcomanni, 
 a High German tribe, drove the Celtic Boii from the country, 
 which, though since occupied by Sclavonian Czechs (i. e. "those 
 farthest in advance;" Dobrowsky, apud Adelung, Mitlwid. ii. 
 p. 672) is still called the home or land of its original inhabitants 
 (JBohemia=^Boien-HeimatJi). The Gauls conquered from Scla- 
 vonian and German tribes the fertile plains of Lombardy, and, 
 spreading to the South-east, sacked Rome and plundered Delphi; 
 and the oldest writing in the New Testament is an epistle ad- 
 dressed to a tribe of Galatae, or Gauls, settled in the North of 
 
 * Meier's Report to the Brit. Assoc. 1847, p. 301. 
 
 Κ 2 
 
132 THE ETHNOGRAPHIC AFFINITIES [bOOK I. 
 
 Asia Minor. These migrations, however, are all to be referred 
 to retrogressive movements of the unconquered Celtic tribes of 
 the West. In general, the Celts do not appear as distinct 
 nationalities in Asia or in the East of Europe, where they are 
 assimilated to theSclavonians and German tribes who conquered 
 them; and except in the regions already indicated, the Celts have 
 disappeared in the ethnical masses which they immediately pre- 
 ceded or followed. The two great dialects of the Celtic — the 
 Welsh and the Erse — exhibit difiPerences corresponding to those 
 between the High German and Low German , and may be dis- 
 tinguished by similar epithets as High and Low Celtic. We 
 attribute this difference to the early prevalence of High German 
 admixture in the case of the Cymric dialect. 
 
 These are all the European languages which belong to the 
 great Indo-Germanic family, with the exception of the Greek 
 and Latin, which we have purposely omitted, till after we have 
 spoken of theAsiatic members of the family, which contribute so 
 much to the accurate classification of the European idioms. 
 Arguing from what we know of the etymology and grammatical 
 structure of the languages w^e have mentioned, we should not 
 hesitate to class together with the Low German in its oldest 
 form the Low Celtic or Erse, the Lithuanian, and the Sclavonic 
 languages; and, with the Old High German, the High Celtic 
 only. By this we mean, that, though all these languages spring 
 from , the same Asiatic source, the idioms which we find in the 
 extremities of Europe, in the peninsulas, and on the northern 
 and western coasts, are due to tribes who entered Europe at an 
 earlier period, and were driven onwards by subsequent emi- 
 grants ; and we are able to ascertain from these languages them- 
 selves that such is the case. 
 
 80 If we turn to the Eastern members of the family, we 
 shall easily find a rational explanation of this division. It ap- 
 pears, then, that the origin of these languages is traceable to 
 Iran, a country bounded on the north by the Caspian, on the 
 south by the Indian Ocean, on the east by the Indus, and on 
 the west by the Euphrates. Within these limits were spoken, 
 so far as we can discover, two languages which bore the same 
 relation to one another that we recognise as subsisting between 
 
CHAP. IV.] OF THE ANCIENT GREEKS. 133 
 
 Low and High German, a language analogous to the former 
 being spoken in the North and East of the district, and one 
 analogous to the latter in the South. Although the latter ex- 
 tended to the sea-coast, yet, as the inhabitants who spoke it 
 were mostly mountaineers (Flerod. i. 71), we are justified in 
 adopting, as applicable to these two languages, the same dis- 
 tinctive epithets which use has conferred upon the two great 
 divisions of the German languages; and we will call the southern 
 High Iranian , the northern and eastern Low Iranian. The sur- 
 rounding nations to the North and East belonged to the Tura- 
 nian, or sporadic family ; but, when the mighty people confined 
 within these comparatively narrow limits had become too nu- 
 merous for the country they lived in, the eastern and northern 
 tribes sent off emigrations to the South-east and North-west, 
 breaking through or driving before them the tribes by which 
 they were hemmed in. Those, however, who went off to the 
 North-west were more powerful or more enterprising than the 
 emigrants who took a south-easterly course; for while the for- 
 mer carried the Low Iranian dialect over all Asia and Europe 
 to the islands of the West, the latter mastered only the northern 
 part of Hindostan, and perhaps also, to a certain extent, a few 
 of the islands of Polynesia. The proof of this colonization of 
 Europe and Northern India by the inhabitants of northern and 
 eastern Iran, rests upon the agreement of the languages spoken 
 by the oldest inhabitants of India and Europe, and on the ob- 
 vious derivation of the names of the earliest tribes, in both from 
 the country which afterwards became Media. The former of 
 these grounds confirms the other: for when we find that the 
 ancient Indians spoke the same language with the Low German 
 tribes in Europe, and that the names of both are derivable from 
 the same district, we are forced to conclude that they are both 
 the offspring of a people who dwelt in the country to which 
 their names point, and spoke a language which was the mother 
 of their sister-idioms. 
 
 81 And first, with regard to the Median origin of the old 
 name of Northern India, it is to be observed, that, according to 
 Herodotus (vii. 62), the Medes were in ancient times called 
 Arians by all the world — εκαλέοντο δε τίάλαυ Ttgog πάντων 
 
134 THE ETHNOGBAPHTC AFFINITIES [BOOK I. 
 
 "Jqlol. Now dryas is a Sanscrit word signifying "noble," 
 "splendid," "well born;" and the Hindus applied this epithet 
 to themselves in contradistinction to the rest of mankind, whom 
 they called Mlechch'has*, just as the Hellenes distinguished 
 themselves from the Barbarians (^Asiatic Researches, vii. p. 175, 
 and Schlegel, Etudes des Langues Asiatiques, p. 70). That this 
 name bore the same signification out of India, appears from the 
 fact, that those kings of Cappadocia who boasted of Median ex- 
 traction, called themselves Aria-rathes; this is obviously the 
 Sanscrit adjective drya-rathas, "mounted on a splendid chariot," 
 used as an epithet of warriors and kings, as mahd-rathas, 
 "mounted on a great chariot," is constantly applied by the 
 oldest Indian poets (see e. g. Bhagavad-Gita, I. si. 4, 6, 17, &c.). 
 Moreover, the name Αΐ'ΐαηα, in ancient times, undoubtedly in- 
 cluded the whole of the northern provinces of the Persian em- 
 pire: even when Strabo wrote it extended over part of Persia, 
 Media, Bactria and Sogdiana (p. 724, comp. Steph. Byz. s. v. 
 "AqloC). This name appears as Airaiene in the Zend books, and 
 is now contracted into Irdn, much in the same way as Ayodhyd, 
 the name of the old kingdom of Ramas, is shortened into the 
 modem Oude. The same name may be recognised in Arya- 
 avarta, "the country of the Arians," which is the classical 
 name for the old country of the Hindus, and which is defined 
 as lying between the Vindhya and "snowy" (Ilimdlayd) moun- 
 tains, and extending from the Eastern to the ΛVestern Oceanf . 
 This definition excludes theDeccan, or "country to the right" 
 (dakshina) ; and the language of the country , its geographical 
 features, its oldest traditions, and the physical characteristics 
 of the inhabitants, sufficiently show that the Arians or Iranians 
 entered Hindostan by the Panjab, and did not extend them- 
 selves far towards the South J. To the present day, though the 
 northern tribes of India speak languages more or less corrupted 
 
 * It has been suggested that the name Belooch is the modern repre- 
 sentative of this epithet. It is written \jOyX^] in Abnlfeda. 
 
 •j- Arya-avartah: punyu-hhumir (i. e. "the region of sanctity"); madhyan 
 Vindhya-Himalayoh {Am. Cosh. p. 66, Colebrooke). We may compare the 
 description punyn-bkumir with the epithet vaejo, "pure," given to the Arian 
 mother-land by the Medo-Persians (below, § 85). 
 
 ;[l Schlegel, sur tOrigine des Hindous, p. 415. 
 
CHAP. IV.] OF THE ANCIENT GREEKS. 135 
 
 from the Low Iranian or Sanscrit, such as the Bengali and Hin- 
 dostani, the southern languages are more akin to the Mongol 
 idioms, which entered into the languages of middle and northern 
 Asia*. The scenes of their oldest poems, the Maha-hharata and 
 Ramayana,2iVQ generally confined to the neighbourhood of Delhi 
 and Oude. In the latter, the exiled hero travels to the extreme 
 South, where he finds, among other things , innumerable hosts 
 of apes, who do him considerable service. We consider this 
 fable as proving that there was a striking physical difference 
 between the Hindus and the population of southern India in 
 the very earliest times. It appears that the aborigines of India, 
 whom the Hindus or Arians invaded and conquered, had most 
 of the characteristics of the negro tribes : at least, the supposed 
 remains of these earliest inhabitants, still found in the North of 
 India, have woolly hair, low foreheads, and flat noses. We 
 venture, then, to conclude that these "apes, with foreheads vil- 
 lainous lowf ," were merely the ill-formed natives of the South J, 
 who appeared to the handsome and well-proportioned Hindus 
 as little better than monkeys, just as the Greeks described the 
 negroes of Africa as Pygmies or Cercopes, because they differed 
 in form and stature from themselves, or a« Virey would class 
 the Hottentot with the baboon§. 
 
 82 Secondly, as to the Median origin of the Low German 
 tribes, the following examples may suffice. That the Medes 
 extended themselves to the North-west appears from the position 
 
 * Mr. Caldwell (in his Comparative Grammar of the Dravidian or South- 
 Indian Family of Languages, London, 1856) says, "the Dravidian lan- 
 guages are to be affiliated, not with the Indo-European, but with the 
 Scythian group of tongues; and the Scythian family, to which they ap- 
 pear to be most closely allied, is the Finnish or Ugrian." See also 
 Norris' Scythian text of the Inscriptions at Behistun, As. Sac. Vol. xv. 
 f The Tempest, Act iv. Sc. 1. 
 
 X In the Gem, a copy of which Schlegel has prefixed to his edition 
 of the Ramuyana, the attendant apes of Rama appear as men with the 
 faces and tails of apes. 
 
 § Hamaker {Akadem. Voorlezingen, p. 9) considers the Rumayana as a 
 poetical description of the complete triumph gained by Brahamanism and 
 its votaries over the autochthones of India , who had sought a retreat in 
 Lanka, or Ceylon, 
 
136 THE ETHNOGEAPHIC AFFINITIES [bOOK I. 
 
 of Media in the historical ages. The names of many of the 
 Low German nations point to a derivation from the north of 
 Iran. We have seen that the Saxons or Sacasunu are trace- 
 able to Bactria. The Sarmatie or Sauromatie, an old Sclavonian 
 nation, are expressly mentioned as descendants of the Medes 
 (Plin. Hist. Nat. Yi. 7: Sarmatce Medoricm, ut ferunt, soholes. 
 Diodor. Sic. ii. c. 43, p. 195, Dindorf: δύο όε μεγίΰτας άτΐοί- 
 κίας γενέΰ^αι, τγιν μεν . . . Τϊ^ν δε 1% της Μηδίας τϋαρα τον 
 Τάναϊν κα%'ίδρν&εΐΰαν, ης roi;g λαονς Σανρομάτας ονομαΰ^ηναί) : 
 and their name indicates that they too claimed the North of 
 Media as their father-land*. The Sigynnse, whose territory 
 extended from the north of the Danube to the country of the 
 Heneti or Veneti (Sclavonian Wends), on the Adriatic, in dress 
 resembled the Medes, from whom they derived themselves; 
 "how they could be colonists of the Medes ," adds Herodotus 
 (v. 9), "I cannot understand; but any thing may happen in the 
 long course of timef ." Now the abode which Herodotus assigns 
 to the Sigynnas falls within the limits of the Sauromatse , who 
 were a Sclavonian tribe, and also derived from the Medes. Ac- 
 cordingly, the Sigynnae must have been themselves Sclavonians, 
 and could not have been connected with the Huns, as some 
 suppose. Besides, Strabo describes the Sigynnas as living near 
 the Caspian, with habits similar to those which Herodotus 
 ascribes to them (p. 520). Therefore, we cannot doubt that 
 they were a low Iranian people. In the same manner we might 
 
 * Bockh, Cwpus Inscn'pt. ii. p. 83: "Sauromatie, Slavorum hand dubie 
 parentes, . . . e Media imniigrarunt ad Tanaim (Diod. II. 43, extr. Plin. 
 H. N. Ti, 7), unde Gatterer (Introd. in hist univ. S^fnchron. T. i. p. 75) 
 nomen derivat a Matenis s. Matienis s. Medis et voce Lithuanica Szaure, 
 qnse septentrionem designate tit Sauromatse sint Medi Septentrionales. 
 lidem recte visi sunt SyrmatcB, quos Plinius prope Oxydracas Indo vicinos 
 collocat (cf. Ritter, Vorhalle d. Gesch. p. 283) et eodem nomine Scylax 
 ad Mseotidem. Mox vero Sauromaticie gentes latius evagataj sunt." 
 
 t γένοίτο ό' αν παν iv τω μακρω χρόνω. Valckenaer quotes Soph. Aj, 
 655. for a similar sentiment. He might have said more aptly that Hero- 
 dotus was almost repeating Pkiloct. 306: πολλά yciQ τάδε εν τω μακρώ 
 γένοιτ αν άνϋ^ρώπων χ^όνω, and γένοιτ' αν πάν is also Sophoclean; cf. 
 AJ. 86 : γένοιτο μίν ταν πάν &Εον τεχνωμένον. That Herodotus often quoted 
 Sophocles we have endeavoured to show elsewhere (Proceedings of the 
 Phil. Soc. Vol. I. p. 164). 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 137 
 
 point out traces of a North Iranian pedigree in the case of every 
 nation of the Low German class of which any mention is made 
 by ancient writers. We consider even the invasions of the 
 Scythians by the Persians, mentioned by the Greek historians, 
 as traditions of the pressure of the High on the Low Iranians ; 
 for the identity of the names Scythians, Getae, Jutes, and Goths, 
 has been long recognised. 
 
 The argument from the language is decisive of the whole 
 question. The resemblances between the old Low German 
 dialects and the Sanscrit, even after a separation for thousands 
 of years , are so striking that an eminent philologer has re- 
 marked that "when he reads the venerable Ulphilas he could 
 believe he was reading Sanscrit*." On the whole, then, we 
 consider it as nearly certain that the Hindus in India and the 
 Low Germans in Europe are emigrants from the country about 
 the southern extremity of the Caspian sea. We do not pretend 
 to say when the emigration took place, nor do we suppose that 
 it took place at once. As the population became too numerous 
 for the country, or as they were pressed upon from without, 
 they would naturally send off streams of invaders to the right 
 and left in search of other settlements. 
 
 83 The term Sanscrit, by which we distinguish the old 
 Iranian idiom that formed the basis of the North Indian and 
 Low German dialects, is an epithet used by the Brahmins to 
 designate the language in which their books of law and religion 
 are written, the depository of their ancient poetry and philo- 
 sophy; it implies that this language possesses all its flexions and 
 grammatical forms, that, in fact, it is removed from the corrupt- 
 ing influences of every-day use. The original word Saii-s-krita 
 is a compound : the first syllable is the preposition sam, " with," 
 (ύνν) ; the second, the passive participle krita Q-tas^ -td, -tmn)^ of 
 the crude verb kri^ "to make" (creare, ceremonia, κραίνω) ; and 
 a silent s is interposed; its literal meaning is "done, made, 
 or formed completely" (con-fectus)^ "perfect," "highly polished," 
 
 * Bopp, Conjugationssystem der Sanskritsprache^ Frank. 1816. Vorrede 
 von Windischmann, p. x. : "Der Verfasser sagt (in einem Briefe) cr glaube 
 Sanskrit zu lesen, wenn er den ehrwiirdigen Ulphiia lese." 
 
138 THE ETHNOGEAPHIC AFFINITIES [bOOK I. 
 
 "regularly inflected," "classical*." This epithet seems to have 
 been applied to the old language ofNorthernlndia to distinguish 
 it from another class of old languages not so elegant and com- 
 plete, called the Prd'krtta, Ά word composed in a similar manner, 
 and signifying "low," "vulgar," "commonf." In fact, there 
 are three divisions of the written languages of India; these are, 
 to use the words of Colebrooke (Asiatic Researches, Vol. vii. 
 p. 209, Engl, reprint), "(1) Sanscrit, a polished dialect, the 
 inflexions of which, with all its numerous anomalies, are taught 
 in grammatical institutes. This the dramatic poets put into the 
 mouths of Gods and of Holy Personages. (2) Prdcrtt^, con- 
 sisting of provincial dialects, which are less refined, and have a 
 more imperfect grammar. In dramas it is spoken by women, 
 benevolent genii, &c. (3) Mdgadht, or Apa-bhranga^ a jargon 
 destitute of regular grammar. It is used by the vulgar, and 
 varies in different districts : the poets accordingly introduce into 
 the dialogues of plays a provincial jargon spoken by the lowest 
 persons in the drama." The word apa-hhranga , derived from 
 bhrag, "to fall down," signifies a word or dialect which has 
 fallen off from correct etymology ; the native grammarians use 
 it to signify "false grammar," as opposed to Sanscrit, in the 
 sense of "duly formed," "regularly inflected." The force of 
 the grammatical terni Sanscrit as a name for a sacred language 
 will be duly appreciated by all who know that the old gram- 
 
 * On the meaning of this term see Pott s. v. Indoyerm. Spracht. 
 p. 34, Hoefer, de Prakrita Dialecio, § 1. 
 
 •j- Compare Hamaker, Akadem. Voorlezingen, p, 213. 
 J For the various meanings of the word Prakrita, see Lassen's Institut. 
 Liny. Pracrit. pp. 23 sqq. The signification of the term, according to its 
 etymology, is "derived," i. e. from the Sanskrit; but, as we have men- 
 tioned in the text, this name does not correspond to the facts. Vararu- 
 chi treats of four kinds of Prakrit, and there are three forms used in 
 the Indian dramas. The heroine and principal female characters speak 
 Saurasmi, the royal attendants speak Mayadhi., and servants, &c, speak 
 Ardha-Magadhi, "half Magadhi" (Wilson, Hindu Drama, p. 68; Lassen, 
 Inst. Pracr. p. 28; Ind. Alterth. ii. 506). The Sauraseni has the greatest 
 claim to be considered the language of the country in which the drama 
 arose, and Krishna, who stands in intimate connexion with the origin of 
 the drama, was specially honoured in this district. Arrian, Ind. viii. 5; 
 Lassen, Inst. Pracr. pp. 35, 37; Ind. Alterth. i. 616, 625; ii. 507. 
 
CHAP. IV.] OF THE ANCIENT GREEKS. 139 
 
 marian Panini was esteemed as a sort of demigod, and was said 
 to be grandson of the inspired legislator Devala; and its appli- 
 cation from the description of the kind of language to the de- 
 signation of a particular ancient language is perfectly analogous 
 to the use of the word gramatica by the early Italian writers 
 to signify the Latin language: thus Boccaccio {Oecam, vii. 6) 
 describes a good Latin scholar as un gran valentuomo in gra- 
 matica, and Yarchi gives the following distinction (^Dialogo 
 sopra le lingue, p. 335): Tutte le lingue, die non sono Latine 
 gramaticali, si chiamavano e si chiamano volgari. In the 
 same manner the classical schools in this country are technically 
 designated as grammar schools. The meaning of the term 
 Banskrtta is plainly implied in what Dante says (^Convito, i. c. 5, 
 p. 21): il Latino e perpetuo e non corruttihile, e il Volgare e non 
 istabile e corricttihile. 
 
 84 Keverting to the doctrine maintained in the preceding 
 chapter, we may be surprised to find that any language laying 
 claim to great antiquity and to an early application of the art 
 of writing, should deserve the title oi Sanskrita, or "regulary 
 inflected." The oldest form of Sanscrit, that which is found in 
 the Vedas , is very like that of the ancient Persian as it has 
 been recovered from the Cuneiform monuments ; and it is highly 
 probable that the Brahminical people who spoke this language 
 were not the first of those Arians, who descended into the plains 
 of Delhi from the Himalayas or the Panjab; but that, on the 
 contrary, the peninsula had long before been occupied by a 
 cognate tribe, who eventually adopted the faith of Buddha, and 
 whose ordinary language, the Pracrit or Pali, is always recog- 
 nised in the oldest inscriptions. It has been maintained by 
 Weber ( Vdjasaneya Sank. Spec, ii. p. 203) that the Sanscrit 
 never was the language of an entire population, but that it was 
 limited to a learned class , who formed it from the Vedas. But 
 the general tendency of modern researches is opposed to this 
 hypothesis ; and it seems to be clear that, however difierent the 
 style of the ordinary spoken language may have been from the 
 written and poetic diction of the sacerdotal Brahmins, the forms 
 and inflexions of the language must have been the same in both, 
 and that the Bhashd or ordinary discourse of the Brahminical or 
 
140 THE ETHNOGRAPHIC AFFINITIES [BOOK I. 
 
 later Arian invaders of India, which Panini opposes to the 
 Ch'andas^ "metre," or Rik^ "hymn," of the Vedas (see Boeht- 
 lingk's edition, ii. p. 523), differed from the latter rather in the 
 diction or form of expression than in the furniture of words, or 
 καταΰκενή, as the Greeks significantly called it. It is true that 
 the pure Sanscrit of the conquerors, instead of becoming the 
 prevalent idiom, like the Attic Greek, the Tuscan Italian, the 
 Castilian Spanish, and the new High German, gradually shrank 
 within narrower limits, and constituted itself, like the Latin of 
 the middle ages, the dialect of the Brahminical aristocracy 
 (Lassen, Lid. Alterthumsk, ii. p. 1152); and that as the Pro- 
 testants opposed themselves to the Catholic hierarchy by adopt- 
 ing the vernacular in opposition to the learned language , as a 
 medium of communication in spiritual matters, so A^oka and 
 the other Buddhist kings adopted the popular language on their 
 coins and in their inscriptions, as an obvious instrument in fur- 
 therance of their anti-sacerdotal innovations. But this proves 
 nothing more than the distribution of dialects in the plays ; it 
 merely shows that the language of the priests and of the people 
 had become as different as it was in mediaeval Europe, and that 
 the Brahminical conquerors from the north-west spoke a lan- 
 guage in an older or more etymological condition than that of 
 the Sauraseni and tke other cognate tribes, who had previously 
 established themselves to the south of the Himalayas. If this 
 was the case, it would be a reasonable inference that the Brah- 
 minical conquerors derived from these older Hindoos the -ori- 
 ginal basis of the Devanagajn character, in which they recorded 
 their own traditions and mythology, and which those more 
 ancient Arians had borrowed from the Semitic inventors of the 
 syllabarium. But it is inconsistent with the rules of a scientific 
 philology to assume that the more perfect Sanscrit has been 
 made by refinement out of the less regular Pali, or that the 
 language of the Brahmins is more recent than that of the Bud- 
 dhists*, because the older alphabet does not contain all the 
 letters necessary to express the characteristic word-forms and 
 
 * Hodgson, As. Soc. of Bengal, vi. p. 682, has shown that probably 
 the Buddhist doctrines were drawn up in Sanscrit, though the practical 
 teaching of the priests was in the ordinary language of the country. 
 
CHAP. IV.] OF THE ANCIENT GREEKS. 141 
 
 syntax of the Sanscrit*. The following appears to us the only 
 sound theory derivable from the premises. Every argument 
 that is adduced to prove that the Pracrit, Pali, and other Ma- 
 gadhi dialects were connected with literature at an earlier period 
 than Sanscrit, tends to show — not that these languages are more 
 ancient, or are found in an older condition than the Sanscrit, — 
 but precisely the reverse, namely, that these are dialects of the 
 Sanscrit in a secondary state as far as their structure is con- 
 cernedf . And though the first Arian inhabitants of India may 
 have used the art of writing for a much longer period than the 
 Brahminical tribes, this is rather in favour of the belief that the 
 Sanscrit and Vaidic poems belong to that more ancient species 
 of literature which is preserved in the memory of a literary 
 caste long before the invention or common use of an alphabet. 
 And we are disposed to think that the learned grammarians of 
 Malwah and Pataliputra, and especially the artificial school of 
 King Vikramaditya's court, were for many years engaged in an 
 office not unlike that of the learned men of Athens and Alexan- 
 dria, who, from time to time, put together and published the 
 scattered lays of the Homeridae. 
 
 But although we have no good reason to doubt the great 
 antiquity of the Sanscrit language, and though the writings in 
 which it is contained are the modern representatives of a school 
 of hymnic, epic, and didactic poetry, probably older than the 
 earliest specimens of Greek literature, we must not suppose that 
 it was, as we have it now, the same old Iranian idiom which was 
 taken into Europe ; on the contrary , it bears evident marks of 
 those changes which long usage introduces into every language, 
 and which have not operated to so great an extent in some of the 
 sister-tongues of Europe, for instance, in the Low German, Latin, 
 and Greek. However, as we do not possess any memorials of 
 the primeval language from which it sprung, although we might 
 
 * This seems to be the view adopted by Colonel Sykes in an elabo- 
 rate paper "On the religious, moral, and political state of Ancient India," 
 which appears in the Journal of the Royal Asiatic Societi/, 1841, No. xii. 
 pp. 248—484. See especially pp. 272 and 411 sqq. 
 
 t See Rawlinson, Journal of R. As. Soc. x. part i. p. 41; Dr. Max 
 Miiller, Report of Brit. Assoc. 1847, pp. 321, 326. 
 
142 THE ETHNOGEAPHIC AFFINITIES [bOOK I. 
 
 be able, from a comparison of all the languages of the family, to 
 make a probable reproduction of its grammatical system, and as 
 the Sanscrit does present most remarkable correspondences with 
 the oldest European languages of the Indo-Germanic family, we 
 must be content to take it as the representative of the old Low 
 Iranian, and therefore in the following pages have made more 
 use of it than of any other language of this family in our en- 
 deavours to restore the oldest forms of Greek words. 
 
 85 When history tells us that the Median empire was 
 overthrown by the Persian, this is a distinct announcement of 
 the fact, which we might derive from philology alone, that the 
 Southern tribes of Germanii or High Iranians pressed upon and 
 mastered the Low Iranians, who are known to us as Medes 
 in their Arian home, and as Sauromatae, or northern Medes, in 
 Europe. The establishment of the kingdom of Cyrus was in 
 fact the final development of a tendency which had continued to 
 exhibit itself in the same manner for many centuries previously; 
 and in this we must recognise the counter pressure by which, as 
 we conceive, the streams of emigration to India and Europe were 
 increased. We have stated that the Medians or Low Iranians, 
 spoke the primeval tongue of which the Indian Sanscrit is an 
 offset, and which forms the distinctive element of those Euro- 
 pean dialects which are connected with the Low German and 
 Sclavonian. There is reason to suppose that the Low Iranian 
 emigration entered Europe by the North of the Black Sea, that 
 is, from the original abode of the Median race, the Airyanem 
 vaejo^ "the pure Arian land*," in Bokhara, from which they 
 had descended to Khorassan on their right, and to the Hapta- 
 Hindu^ or Punjab, on their leftf; whereas the mixed tribes of the 
 South and West, or those in which the Persian element pre- 
 dommated, must have extended themselves through Armenia 
 into Asia Minor. Not only the geographical position of the 
 country, but a singularly interesting tradition J, seems to prove 
 
 * See Haug, apud Bunsen, jEgypten, V, a, p. 126. 
 f Id. ibid. pp. 119 sqq. 
 :j: It is a tradition, which cannot be easily set aside, that the singular 
 story in Plato's Republic, p. 614 β sqq. was due to Zoroaster (Clem. Alex• 
 
CHAP. IV.] OF THE ANCIENT GREEKS. 143 
 
 that the province of Armenia, which, as we have intimated, was 
 probably the first seat of the whole human race, must also have 
 been the first stage in the journey of emigration for all the Ira- 
 nian tribes which started from the south and west of the Caspian. 
 In its present state the Armenian language cannot be traced 
 farther back than the fourth century a.d.; but even in its mo- 
 dern form it yields to a philological analysis the most convinc- 
 ing proofs of its Indo-Germanic structure, and the scattered 
 notices in ancient writers show that it must have been one of 
 the Medo-Persic idioms. At the commencement of the 4th cen- 
 tury B.C., the country people in Armenia understood Persian*, 
 and their deities bore Persian namesf . According to Herodotus 
 the Armenians and Phrygians were closely allied in origin, 
 though he probably inverts the fact when he calls the former a 
 colony of the latter. And the scanty remains of the Phrygian lan- 
 guage admit of immediate comparison with the Persian as well as 
 with the Armenian J. The Cappadocians, who have many affinities 
 with the Medo-Persians§ , are said to have spoken the same lan- 
 guage as their neighbours the Armenians]]. On the other hand, 
 the Sauromatae, who, as we have seen, were of Median origin, 
 have many Armenian affinitieslF. Finally, it has been shown that 
 some of the oldest European languages correspond to the Ar- 
 menian in many terms, which have no longer their counterparts 
 in the conterminous idioms, and that even the ancient Etruscans, 
 whom we have identified on other grounds with the oldest 
 
 Strom. V. p. 710, Potter), and we have shown in a special communication 
 to the Camhridye Philosophical Society (21 Febr. 1859) that it must have 
 been derived by Plato from Heracleitus, whose philosophy was Zoroastrian 
 (see Gladisch, Herakleitos und Zoroaster^ Leipzig, 1859). Now the author 
 of this apologue is called ' Hq δ 'Αρμενίου το γένος Παμφνλον ; and this 
 can only mean that the Arians, as they appeared in Pamphylia the most, 
 western province of Persia, called themselves descendants of the Ar- 
 menians. 
 
 * Xenophon, Anah. iv. 5, § 34 : άνηρώτων τον κωμάρχην διά τον περ- 
 ϋίζοντοζ ερμηνέως τις εΐη ή χώρα' δ δ' ελεγεν, οτι Αρμενία. 
 
 f Gosche, de Ariana linguce gentisque Armeniacce indole^ Berolini, 1847, 
 pp. 8 sqq. 
 
 X VII. 73. § Gosche, 1. c. pp. 55, 56. 
 
 II Moses Chorenensis, i. 13. 
 
 ^ Windischmann , die Grundlage d. Armenischen im Arisch. Sjpntch^ 
 stamme^ p. 14. 
 
144 THE ETHNOGEAPHIC AFFINITIES [bOOK I. 
 
 branch of the Low Iranians*, were connected also with the 
 Asiatic Thracians, the Phrygians, and the Armeniansf . 
 
 When these Persians, or High Iranians, had intruded them- 
 selves upon the Medes, or Low Iranians, it is probable that the 
 language of the latter became tinged with the peculiarities of 
 the Persian idiom, which was, however, nearly related to the 
 Median, and the mixed language constituted the speech of 
 those Persians with whom the Greeks had so much to do. The 
 connexion of modern Persian with modern High German, even 
 after many centuries of Arabian rule and the loss of its in- 
 flexions , was long ago perceived ; and in one of the tribes of 
 the Persians, the Γερμανίου mentioned by Herodotus (i. 125), 
 we still recognise the distinctive name of the Thuringians or 
 Herminones. We assume, therefore, that the High German 
 dialects of Europe are due to a final Iranian emigration con- 
 nected with the early expansion of the Persian race. And thus 
 if all the European members of the family can be assigned to 
 the two divisions of Low and High German, the former derived 
 from that old Iranian stock which gave to India its sacred lan- 
 guage, the latter from the great race of Persians or Germanians, 
 the name Indo-Germanic, which has been given to the family, 
 is doubly appropriate. 
 
 86 The opinion once entertained by the majority of Eng- 
 lish orientalists, that the Zend language, namely, that which 
 is found in the sacred books obtained by Anquetil du Perron 
 from the Parsis of Gujarat, is not a genuine dialect, but an 
 artificial and fabricated idiomj, or at best a corrupted Sanscrit, 
 
 * Varron. pp, 68 sqq. 
 
 f Ellis, Connexion of the Rhcetians and Etruscans with the Thracians, 
 Journal of Philology^ ii. pp. 1 — 20, 169—185; Contributions to the Ethno- 
 logy of Italy and Greece, Lond. 1858. 
 
 :{: Colonel Vans Kennedy went so far as to assert that the Zend and 
 Pehlvi are mere jargons made up of other languages (Asiat. Joiirn. \o\. 
 XXVI. Part I. p. 100); the late Dr. Leyden, Mr. Erskine, and Professor 
 von Bohlen, of Konigsberg, supposed that the Zend is genuine indeed, 
 but only a dialect of the Sanscrit, like the Pracrit, or Pali (the sacred 
 language of the Buddhists); and Mr. Romer, in a paper read before the 
 Asiatic Society {Journal^ iv. p. 363), says, "it appears to him far from 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 145 
 
 has been abandoned of late years by all scientific philologers. 
 Rask was the first to show that Zend is as much entitled to take 
 its place among the primitive languages of the Indo-Germanic 
 family as the Greek, the Lithuanian, or the Sanscrit, and 
 that the Avesta must have existed in writing previously to 
 the time of Alexander the Great*. The late Eugene Burnouf 
 submitted the text of the Ya^Mu to a minute grammatical ana- 
 lysis, and completely established the independent character of 
 the Zend language, and its great philological value; and he 
 was followed by Bopp, who included the language of the Avesta 
 among those Avhicli are compared with the other Arian idioms 
 in his great work. The recovery from the cuneiform inscrip- 
 tions of the language actually spoken by the Persians under the 
 Achaemenian kings, — a good work which was begun hj Grote- 
 fend and completed by Lassen and Rawiinson, — has given us 
 a form of human speech diiOPering from the Zend only in the 
 conditions of its development. And we can now see that the 
 language of the Zoroastrian books, no less than that of the 
 rock-inscriptions of the first Darius, belongs to the same class 
 as that of the Vedas or sacred books of the Brahmins, and that 
 even the names of the Vaidik deities, which appear with 
 strangely altered applications in the Avesta f, have survived in 
 the heroes of the Shahnameh, having passed "through the 
 Zoroastrian schism, the Achsemenian reign, the Macedonian 
 occupation, the Parthian wars, the Sassanian revival, and the 
 Mohammedan conquest J." So far then is Zend from being 
 
 improbable, that the Sanscrit supplied the framework upon which the 
 Zend has been constructed; whilst it is evident that, in the formation of 
 Pehlvij it is from the Arabic that assistance has been sought and applied, 
 but following, particularly in the infinitives, the form of the Persian 
 verb." A. W. von Schlegel, who was more sceptical on the subject than 
 most of his countrymen, was inclined to give the Zend an intermediate 
 place between the language of Darius Hystaspes and the Persian of 
 Firdousi {Etudes des Langues AsiatiqueSy p. 71). 
 
 * R. Hask, uber das Alter und die Echtheit der Zendsprache^ ubersetzt 
 von F. H. von der Hagen, Berlin, 1826. 
 
 f See the examples given in Christian Orthodoxy•, Loud. 1857, p. 128. 
 Cf. Bunsen, Jigijpten, V, a, p. 216. 
 
 % Max Miiller in Bunsen's Outlines of the Philosophy of Universal His- 
 tory, Lond. 1854, Vol. i. p. 123. 
 
 L 
 
146 THE ETHNOGRAPHIC AFFINITIES. [bOOK I. 
 
 a corruption of the Classical Sanscrit, that it actually ranks 
 itself with the primitive speech of the Brahmins. '^'The Veda," 
 says Roth*, "and the Zendavesta are two rivers flowing from 
 one fountain-head : the stream of the Yeda is the fuller and 
 purer, and has remained truer to its original character; that 
 of the Zendavesta has been in various ways polluted, has altered 
 its course, and cannot, with certainty, be traced back to its 
 source." Even the name Zend is, with a slightly altered pro- 
 nunciation•}•, the same as the Sanscrit word ChViandas (i. ;e. 
 metrical diction, cf the Latin scandere) which is used by 
 Panini and others to designate the language of the VedasJ ; and 
 it has been remarked §, that "when we read in Panini and 
 others that certain forms occur in Ch'handas but not in the 
 classical language, we may almost always translate the word 
 Chliandas by Zend^ for nearly all these rules apply equally 
 to the language of the Avesta." 
 
 87 We now come to the Latin and Greek languages, and 
 trust to be able to show, that the former is entirely referable to 
 the Low German class, whereas some of the most distinguishing 
 features of the latter are High German. That the Latin is the 
 older language of the two was recognised even by those who 
 wished to derive Latin from Greek ; for they sought a connexion 
 between the Latin as it stood and the oldest or ^olian dialect 
 of the Greek. The fact appears still more clearly from the 
 structure of the language. It is the tendency of all languages 
 built upon a system of inflexions to lose these inflexions and 
 substitute for them a system of auxiliaries and particles. Now 
 the Greek shows a much stronger bias to this than the Latin ; 
 indeed the mere use of the definite article and the particle av 
 
 * Quoted by Max Miiller, 1. c. p. 114. 
 f In Zend the letter ζ generally appears like the Greek ζ as the 
 representative of the softened guttural /. Accordingly in the preface to 
 the Sanscrit translation of the Ya<^na we find Ijisni JancU and Pahalavi 
 Janda^ signifying the ''Zend or metrical book Ijisni" and "the Pehlvi 
 book" (see Burnouf, Ya^na^ p. xvi. note). 
 
 J Amera-Cosha, p. 359, Colebrooke. 
 § Max Mi'iller, 1. c. p. 113; see also Brit. Associat. Report for 1847, 
 p. 330; Hamaker, Akad. Voorlez. p. 219. 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 147 
 
 in Greek would be sufficient to decide the question of their re- 
 lative antiquity. But this not only appears from the gramma- 
 tical structure, but may be established from the words themselves, 
 in consequence of a law which Grimm has derived from an ex- 
 amination of the German dialects and the old classical languages 
 (Deutsche Grammatik, τ, p. 584), and which has been extended 
 to the Zend and Lithuanian by Bopp ( Vei'gleich. Gramm, p. 78), 
 and placed in a very striking light by Mr. Winning (^Manual, 
 p. 36). As we shall have occasion to recur to the subject in 
 the following chapter, we shall only say here, that according to 
 this law High German uses tenues^ where the Gothic has medials^ 
 and the Sanscrit, Latin, and Zend, have aspirates; it has aspi- 
 rates where the Gothic has tenues^ and the last three languages 
 medials; and medials where there are aspirates in Gothic, and 
 tenues in the rest. The Zend sometimes corresponds to the 
 Gothic; the Lithuanian agrees entirely with the Latin and San- 
 scrit, except that it has no aspirates. The Greek sometimes 
 agrees with the Sanscrit, Latin, &c., at other times with the old 
 High German (AVinning, p. 40 foil.). In addition to this, the 
 Greek and High German have prefixes where the other languages 
 have the same word without a prefix (Winning, p. 35). Now there 
 can be no doubt that old High German stands lower down in 
 the scale of these languages than the Sanscrit and Lithuanian. 
 Therefore the Greek must have at least an important element 
 belonging to the younger or High German class -of languages. 
 
 88 This is fully borne out by all that tradition has told us 
 of the early population of Greece. It is distinctly stated that 
 the Pelasgians, as they were called, that is, the oldest inhabit- 
 ants of the country, spoke a language which was not Greek 
 (Herod, i. 57). But we must recollect that this does not imply 
 a radical difPerence of language. People, who, like the ancient 
 Greeks, never learn any language but their own, grow w^onder- 
 fully susceptible of the shghtest diflferences of writing or pro- 
 nunciation. Of this we have a striking instance in the expres- 
 sions^which the Greeks used in speaking of the different dialects 
 of their own tongue. It is well known that the different petty 
 tribes of Greece, in consequence of the numerous mountains and 
 rivers which kept them separate, used to speak a great many 
 
 L 2 
 
148 THE ETHNOGEAPHIC AFFINITIES. [bOOK I. 
 
 varieties of the same language even in the age of history. Now, 
 in mentioning these provincialisms, an Athenian would not hesi- 
 tate to call them all φωναί'^^ and even βάρβαροι φωναί•\; so 
 that even the epithet βαρβαρόψωνοξ is not to be understood as 
 implying that the difference of idiom is great or striking, but 
 only that there is a difference. But, what is of most importance 
 with regard to the Pelasgian languages, it appears that the old 
 inhabitants of Italy w^ere also Pelasgians, and there is certainly 
 no radical difference between Latin and Greek. "We are led, 
 then, to the conclusion that these Pelasgians were simply an old 
 or Low Iranian tribe who formed the basis of the population in 
 Italy and Greece. If it were necessary to ^^ upon some par- 
 ticular branch of the Low Iranian, we should be inclined to 
 select the Sclavonian. It must, however, be understood, that in 
 calling the common element of Latin and Greek a Sclavonic 
 language, we mean only that as the Sclavonians , the children of 
 the SauromataeJ, and the most widely-extended branch of the 
 Low Iranian family, may be traced to the immediate neighbour- 
 hood of Greece and Italy ; as there are singular coincidences 
 between Latin and the oldest Greek on the one hand, and even 
 the modern Sclavonian languages on the others and as the Greek 
 traditions point to the Hyperborean regions §, we may safely 
 call the Pelasgians by a name which, though now restricted, 
 properly describes all those Low Iranian tribes that came into 
 immediate contact with the people of whom we are Sj^eaking, 
 We do not exclude the claims of the Goths (or Getai), Scy- 
 thians ||, orThracians, but we consider all these tribes as more or 
 less affected by admixture or contact with members of the Sclavo- 
 
 * See Plato, Phcedo, p. 62 a; Protagor. p. 346 d. 
 
 t Protagor. p. 341 c : ατε Αέοβιος ων και h φωνγι β αρβάρω τεϋ-ραμμέ- 
 νος. According to Pindar {Isihm. v. [vi,] 24), παλίγγλωΰΰος is a stronger 
 term than βάρβαρος. * 
 
 ^ See Bockh, Corpus Inscript. ii. p. 83: Sauromatce , Slavorum hand 
 duhie parentes (above, p. 136). 
 
 § Diodorus Sic. (ii. 47, p. 198, Dindorf): ί-χειν δε τους Ύπερβορεονς 
 ιδίαν τινά διάλεν,τον -Λαϊ προς τονς"Ελληνας οίκειότατα διακεΐοΟ'αι. Clem. 
 Alex. Strom, ι. p. 225: ει δέ τις ττ^ν φωνην διαβάλλει των βαρβάρων^ 
 ^' έμοι δέ,'' φηοϊν 6 Άνάχαραις, ^' Λαντες "Ελληνες ΰν.ν^^ίζονΰι.'' 
 
 II When we identify the Sclavonians with the Scythians, we are speak- 
 ing only of those Scythians who were immediately known to the Greeks, 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 149 
 
 nian stock. The names of the Massa-Getae, Moeso-Goths, and 
 Mysians, are only various corruptions ofoneandtlie same original 
 designation. Now it appears probable that the Sclavonians in- 
 habited Mysia from the very earliest times. We are told by 
 Nestor, the oldest historian of Russia, that the ancient Sclavo- 
 nians were driven out ofMoesia and Pannonia by the Bulgarians : 
 he is perhaps wrong in placing this event so late as the fourth 
 orfifth century of the Christian era, but his testimony is valuable 
 as a tradition of the fact, that the Moesians, and therefore of 
 coursethe Asiatic Mysians, belonged to theSclavonian stock. If, 
 however, the old Mysians and Sclavonians were the same people, 
 it is pretty clear that the Pelasgians were also of Sclavonic 
 origin, for the inhabitants of Mysia were evidently of the Pelas- 
 gian race*, and the Pelasgian traditions of Rome all point to 
 that country. The argument from the agreement of even modern 
 Sclavonic with Latin and the oldest element of Greek, is still 
 more conclusive. The resemblance of the Russian to the Latin 
 is so striking that a modern traveller has not hesitated to assert, 
 that the founders of Rome spoke the Russian language f. It is 
 only in the most ancient monuments of the Greek language that 
 we can find the same coincidences, and then they are sufficiently 
 striking. Professor Dankovsky, of Posen, has shown this, in a 
 loose and unsatisfactory way, it is true, by an interlinear ap- 
 proximate translation of Homer into modern Sclavonic J, and a 
 
 and were therefore Sauromatae or Sclavonians. The original Scythians, 
 who were no doubt of the Mongolian race (Niebuhr, Kl. Schr. p. 361), 
 were invaded and conquered by the Getae and Sauromatas, that is, by the 
 Low Iranians, just as the old Mongolian population of India were subdued 
 by the Hindus: and it is these Sclavonians with whom the Greeks had 
 so much intercourse; see below, § 93. 
 
 * Niebuhr, Hist, of Rome, i. p. 33. 
 
 f Itahj and its inhabitants: an Account of a Tour in that Country, in 
 1816 atid 1817, by J. A. Galiffe, of Geneva, Vol. i. p. 356 foil. The con- 
 victions of this author on the identity of Russian and Latin are valu- 
 able, not because he is, but because he is not, a philologer. Mr. Galiffe 
 had no ethnographical theory to maintain, but, with only a superficial 
 knowledge of the two languages, could not help recognising a strong 
 family likeness between them. 
 
 J Homerus Slavicis dialectis cognata lingua scripsit; ex ipsius Homeri 
 Carmine ostendit Gregorius Dankovsky. Vindob. 1829. 
 
150 THE ETHNOGEAPHIO AFFINITIES- [bOOK I. 
 
 more extensive and formal comparison of Russian and Greek has 
 been instituted by Constantini*. There are^ indeed^ some archa- 
 isms in Greek which are hardly explicable, otherwise than by a 
 comparison with Sclavonic and the oldest Low German, "We 
 allude to the arbitrary insertion of i in some words in Gothic^ 
 Sclavonic, and the Boeotian and Thessalian varieties of the JSlo- 
 lian or oldest dialect of Greek. 
 
 The resemblance of Sciavonian to Latin and the oldest ele- 
 ment of Greek is not more remarkable than its dissimilarity, in 
 certain points, to the Greek of the classical ages. For instance, 
 there is a total absence of the article in the Latin and Russian^ 
 although this part of speech has generally become indispensable 
 to those languages which have obtained a full literary develop- 
 ment, and is especially prominent in the Greek. This is the 
 more singular as the Russians have never used the Roman law 
 or ritual, or, in fact, brought themselves into any immediate 
 contact with the Latin language, whereas the forms of the 
 Greek church have been long established in Russia (\Vinning^ 
 p. 121). With regard to the breaking up of the case-endings in 
 the Sclavonic declension, in which particular this branch differs 
 entirely from the other members of the family, we must refer 
 the reader to some good remarks by Bopp (VergL• Graiyim. 
 VoTT, Π. AbtL• p. iv. foil.). 
 
 89 It appears, then^ that the common or Pelasgian element 
 of Greek and Latin was allied to the Sciavonian, or Low Iranian 
 branch of the Indo-Germanic family. The additional or Hel- 
 lenic element of the Greek, which afterwards pervaded the 
 whole language, and gave a High German character to its 
 entire structure, seems to have come from the East by Asia 
 Minor; at any rate, we find that the Hellenes make their first. 
 appearance in the North-east of Greece. For reasons, which 
 we have already mentioned, we believe that this new element 
 was High Iranian or Persian. A question might be raised. 
 whether it belonged to the High Celtic or AVelsh, or to the 
 High German, which both seem to have entered Europe from 
 
 * ζίοτιίμιον τΐερί της πληοιεϋτάτης συγγενείας της Σλαβονο-Ρωϋοικης 
 γλΰΰΰβηζ TCQOS την Έλληνικην. εν Πετρουπόλει. 1828. 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 151 
 
 the same quarter. Our own opinion, drawn purely from philo- 
 logical and geographical considerations, is, that the first popu- 
 lation of both Italy and Greece was Erse or Low Celtic. After 
 them came the Sclavonian element in each country, and then 
 a Lithuanian or Gothic element was superadded in Italy*, and 
 a Persian, High German, High Celtic, or to speak generally, 
 High Iranian, in Greece. We think the only difference between 
 th^ Welsh or High Celts, and the High Germans was, that the 
 Welsh pushed farther towards the West and lost much of the 
 German type by mixing with the uncivilized and unadulterated 
 3rse tribes settled in that part of Europe. It would be absurd 
 to attempt any precise solution of all these ethnographical diffi- 
 culties, but as much as w^e have stated seems to be sound in 
 theory. We cannot determine, except inferentiaUy, which of 
 the numerous early tribes mentioned by the Greek historians was 
 Celtic, which Sclavonian, and which High German; but there 
 is every reason to believe that the Carians were Celtic, the Pel- 
 asgians Sclavonic or Sarmatian, and the Hellenes and historical 
 Thracians of Germanic origin. To examine at length ail the argu- 
 ments which lead to these conclusions, would involve a discussion 
 incompatible with our present limits. It will be sufficient to 
 indicate the main steps of the induction. That the Carians were 
 Celtic appears from the fact that they are said to have included the 
 Leleges and the Caucones. To say nothing of a presumed con- 
 nexion between the Caucones and the Cauci^ a Celtic tribe of 
 northern Germany, the Leleges^ with, a reduplication of the initial 
 I peculiar to the Celtic language, are the same race as the Ligyes 
 or Ligunans (see Vaoron. p. 63; Cambridge Essays^ 1856, p. 35 ; 
 
 * See Varronianus, ed. 2, pp. 59 sqq. The theory which we have 
 developed in this Treatise on Latin philology, and which was first stated 
 at the meeting of the British Association in 1851, namely, that the 
 Etruscans were a Gothic or Low German race, still seems to us to rest on 
 the best induction that can be obtained. The great philologer, James 
 Grimm, has been since then led to a similar view {Gesch. d. deutschen 
 Sprache, p. 115, ed. 1853). And these speculations have received con- 
 siderable support from the acuteness and learning of Mr. R• Ellis (Con- 
 nexion of the Rhcetians and Etruscans with the Thracians, Journal of Phi- 
 lology, No. IV. pp. 1—20; No. v. pp. 169—185; Contributions to the 
 Ethnology of Italy and Greece, Lond. 1858). 
 
152 THE ETHNOGEAPHIC AFFINITIES. [bOOK l, 
 
 below, §121), who are known to have been Celtic. The root of 
 the name Carian^ or 6«?', is obviously Celtic, and the same root 
 is found in the derivative Cretan (^K^g = Κρψτ -g), In the 
 island of Crete we find, first iheEteo-cretans or genuine Cretans, 
 a Carian race also appearing in Cos, Rhodes^ and other islards; 
 secondly, aPelasgian stock; and finally an Hellenic tribe. The 
 Carians, therefore^, were neither Pelasgian nor Hellenic. But 
 they were not Phcsnician : for David's body-gnard of foreign 
 mercenaries are distingnished as GretM and Plethi (2 Sam. vii. 
 3 8), that is, as Carians from Crete, and Philistines from Cyprus ; 
 but as the Philistines were Semitic, the Cretans, as distinguished 
 from them, must have been Indo-Germanic*. And as they were 
 not Germanic, Hellenic, or Sclavonic Pelasgians, they must have 
 been Celts. That the Pelasgians were Sarmatian or Sclavonic is 
 shown by the obvious resemblance between the Sclavonic lan- 
 guages and the oldest or Pelasgian element in Greek and Latin, 
 and by the consistent traditions which place Sarmatian tribes in 
 all the regions from which the Pelasgians appear to have been 
 derived. That the Hellenes, and the Thracians with whom they 
 claim immediate affinity, were of Germanic origin is proved by 
 the identity of the two races in regard to all those national cha- 
 racteristics which are generally distinctive, by peculiarities of 
 articulation and construction which are equally remarkable in 
 High German and in the Doric or purest Greek, and by a com- 
 munity of name, which is observable alike in the TAwr-ingians 
 and Hermun-i^i^r-i of High Gem:kany, and in the TAr-acians and 
 Dor-ians of northern Greece. Whether these considerations are 
 or are not satisfactory as indicating the ethnographic affinities of 
 the different tribes, we may venture to affirm the general fact, 
 that there was first a Celtic, then a Sclavonian, element: and 
 that the original language, in which the Sclavonian preponde- 
 rated, was subsequently infected and pervaded by a High Ger- 
 man dialect, to which the Greek language owes the most re- 
 markable points in its wonderful structure. 
 
 '90 The striking similarity between High German on the 
 
 * On the inferences derivable from the use of the Indo-Germanic root 
 of μάχαιρα, sec Christian Orthodoxy, pp. 251 sqq. 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 153 
 
 one hand and the ancient Greek and modern Persian on the 
 other, was pointed out in the infancy of comparative philology*. 
 The resemblance which Greek bore to the Persian in particular 
 must have been much greater formerly; so much so indeed, that 
 a Greek could learn Persian without any difficulty; Democedes 
 makes a witty remark in Persian before he has been long at 
 Susaf, andThemistocles, an elderly man, who had never learned 
 a foreign tongue in his life, made himself a proficient in the 
 language within a year J. 
 
 91 With the Teutonic race the ancient Greeks had many 
 points in common. The same love of freedom^ the same martial 
 qualities, the same tendency to the formation of a considerable 
 number of small independent states,, and the same prevalence of 
 federalism, characterizes both of these races. The Germans and 
 the Greeks alone have been distinguished among Europeans by 
 a fearlessness and subtlety of metaphysical speculation. Colonial 
 enterprise is a feature which marked the ancient Greeks, and it 
 is so conspicuous in the modern Germans that the whole world 
 is filled with scattered members of one family. Even in their 
 literary tendencies we observe the same agreement. It has been 
 well said by an eminent philologer§, that "the drama, or the 
 combination of the lyric and epic elements, and the complete 
 representation of the eternal laws of human destiny in political 
 society, is entirely unknown to the Semite. It is exclusively the 
 creation of the Hellenic mind, feebly imitated by the Roman, 
 reproduced with originality by the Germanic race. Nor is 
 Iranian India entirely wanting in this last of the three species 
 of poetical composition." If we turn to the languages them- 
 selves, we shall see that it is only the Greek and the German 
 which have combined a perfectly refined syntax with an etymo- 
 
 * See above, § 33. f Herod, m. 130. 
 
 :|: Plutarch, Themistocl. xxix. : ενίαντον αιτηβάμενος καΐ την Περοίδα 
 γλώτταν άτίοχρώντως §%μαϋ^ών. Cornel. Nepos surely exaggerates when 
 he says: ille omne illud tempus {minum) litteris sermonique Persarum dedit^ 
 qiiibus adeo eruditus est, ut multo commodius dicatur apud regem verba fe- 
 cisse, quam hi poierant, qui in Perside erant nati. Thucydides says merely: 
 της Περοίδος γλώβΰης οΰκ ήδύνατο κατενόηΰε (ι. 138). 
 
 § Bunsen, Report to the Brit. Assoc, for 1847, p. 270. 
 
154 THE ETHNOGEAPHIC AFFINITIES. [bOOK I. 
 
 logical structure more or less complete, and a living power of 
 derivation and composition. And even in the details of articu- 
 lation we observe striking coincidences. The evanescence of η 
 and s is j^articularly observable in German and Greek; and the 
 Dorian or peculiarly Hellenic Greek especially affects the final 
 r, which is so marked a characteristic of new High German*. 
 But perhaps the most decisive correspondence of articulation is 
 found in the consistent repudiation by both languages of all the 
 soft palatal sounds suggested by the Sclavonic and Pelasgian 
 idioms, with which the Greek and German languages were 
 thrown into contact at a very early period. 
 
 92 These resemblances are still farther confirmed by the 
 appellations in which the Greeks and Germans equally delighted. 
 We have seen above that the titles Mann^ Ilerr-mann^ Ger- 
 mann^ adopted by the eastern Teutons, indicated a predominance 
 of the manly character, or that this race adopted a name par- 
 ticularly significant of their warlike temper. The same is the 
 meaning of the word "Ελλην^, Another special designation 
 of the Eastern or High Germans is TJmr-mg^ which signifies 
 "highlander" or "mountaineer." AVe have found it combined 
 with the former appellation in the name of the Her-mun-cluri: 
 and it appears by itself in the words Tyr-ol, Tatir-us, Dicro- 
 triges, Dorset, and Taur-ini, Now this name again is a dis- 
 tinctive title of the genuine northern Greeks, as opposed to 
 the Pelasgians: for the ζ/ωρ-ίεΓ^, or "highlanders,'' are repre- 
 sented as descended from zl&Qog the son of "Ελλην, as well 
 as their brethren the Αίολεϊς^ or "mixed men," and the'Twi/gj, 
 or " coast-men." We can trace back this correspondence of eth- 
 nical nomenclature to the original seats of the Greek and Ger- 
 man race in Asia. Immediately to the north of Greece, in the 
 highest mountain-land of Epirus, we recognise in the rqal-oi 
 or J~(oat-xot J about Dodona the element ger- of the Avord Ger- 
 mann; and in the Θρα-κεζ to the west we have again the 
 
 * Sec the Instances in Matthias's Gr. Gr. p. 46; Ahrens, de dialecto 
 Dorica^ p. 71 sqq. 
 
 ^ "ΕλληνΒζ, "tlic warriors;" comp. the name of their god Άπέλλοίν ; 
 Muller, Dor. ii. 6, § 6. 
 
 X See Niebuhr, H. R. i. note 1G2, p. 55, Tr. 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 155 
 
 element Tor or Dor^. It has been already mentioned that the 
 Γερ-μάνίοι, were a tribe of the ancient Persians. ΛVe find the 
 other element in the proper name Darius or Darayawusli. And 
 we may, with a fair amount of probabihty, maintain that the 
 stream of High German or Greek emigration entered Europe 
 by way of Asia Minor, and that its course may still be traced 
 through the dry bed of obsolete proper names and shadowy 
 tradition f. Thus, to begin with the Hellespont, where Asia 
 
 * That, on the other hand, the original inhabitants of Thrace were 
 Sclavonians (above, pp. 129, 148), i. e. the same as the Pelasgians, is 
 similarly shown by the occurence in Thrace of names pointing to Media 
 and northern India. For it cannot be merely an accident that the 
 Thracians, whose range of snowy mountains was called Hcemus (Αίμος), 
 a name recalling that of the great chain in northern India, reckoned 
 among their tribes the "Indians" (Σίντοί) and the "Medes" (Maldot). 
 Thucyd. II. 98, § 2. That the original form of the word "Ivdog, which 
 designates the great western river of India, and the whole people, in 
 the language of the ancient Persians, was Sindhus, ΣίνΟ^ος, or Σιντός 
 is a well-known fact. See Schlegel, de VOrigine des Hindous, pp. 441, 
 2 {Essais Lit. et Hist. Bonn, 1842). 
 
 f The few inscriptions, which preserve fragments of the languages 
 of Asia Minor, exhibit unmistakable traces of Indo-Germanic affinities, 
 and occasionally approximate to the Greek language in its vocabulary 
 and grammatical forms. The tomb of king Midas, the son of Gordias, 
 (Herod, i. 14) has the following inscriptions : 
 
 1 Ates Arkiae/as akenanogafos Midai gafantaei fanactei edaes. 
 
 2 Baba Memefais Proitafos kfi ganafegos sikeman edaes. 
 
 (See Miiller, Dor. i. 9, note /; Texier, Asie Mineure, i. p. 155), where we 
 have the gen. in -J^o-g and the 3 pers. sing, in g, as in the Corcyraean 
 inscription, and may recognise the actual Greek word ^Γάνατιτί, and the 
 augmented verb e-daes. And the bilingual Lycian inscriptions have 
 decidedly an archaic Greek character. For example, we have the fol- 
 lowing renderings of the otherwise unknown words in the Levise inscrip- 
 tion {Trans. Phil. Soc. Vol. i. n. 18, p. 197): ewuinu itatii me7ie prina- 
 
 τοντο-τό μνήμα [ο] είργά- 
 fiitu Polenida MoUeueseu se Lapara Polenidau Poreuemeteu prineze- 
 οαντοΆπολλωνίδης ΜολλίοίοςηαΙ Λαπάρας'ΑηολλωνίδονΠνριμάτίος οίτιεΐοί 
 yeue urppe lada eptteue se tedeeme se ey e tese reti detee 
 
 επΙ ταΐς yvvuL^l ταΐς εαντών και τοΐζ εγγόνοις καϊ αν τις άδικήΰΐ] 
 itatu eweue me ey e oete ponamakke adadauade ada. 
 
 το μνήμα τοντο εξ,ώλεα και πανώλεα εΐη αντω πάντων. 
 
 The words beginning with me ey e and εξώλεα, to the end of the in- 
 scription and its version, do not correspond. Mr. Daniel Sharpe says 
 (Trans. Phil, Soc. p. 200) "in place of the curses denounced in the 
 
156 THE ETHNOGRAPHIC AFFINITIES. [bOOK I. 
 
 Minor and Europe are divided by a narrow strait, we find the 
 well-known name of Tqolcc^ in which the element To7' is still 
 conspicuous, and in connexion with the same region we have the 
 hero Dar-danus. Then again the Teutonic name appears in 
 Teuta-wus, Teuthras, and the like. And Priamus and Paris^ 
 whose common name is best explained from the Persian, appear 
 as the leaders of a confederacy which extended throughout the 
 whole of Asia Minor, and gave a hand to the western borderers 
 of Iran. "Priamus," says a modern philologer*, ''is simply a 
 vassal of the Assyrian King Teutamus, who sends him a body 
 of auxiliary troops out of the heart of Persia." The evidence 
 for this chain of ethnographic connexions is necessarily of a cu- 
 mulative nature. Language, tradition, history, mythology, and 
 as far as this is applicable, those features in descriptive geo- 
 graphy which influence the spread of population, enable us to 
 trace the Gra?co-German race from the mountains of Karmania 
 and Kurdistan through the north of Asia Minor and across the 
 Hellespont into Thrace and Illyria. Nor do we stop here: for 
 we may see how, in a strong but narrow stream f, this warrior- 
 band forced its way through the Sclavonian and Low German 
 tribes into the march-land of Vienna^ and from thence gradually 
 expanded itself along the Danube until it had peopled or con- 
 quered the whole, of the central plateau. 
 
 93 There are two ancient names of constant occurrence^ 
 which seem to mix themselves up with the traditions from 
 which we derive the theory respecting the origin and progress 
 of the Helleno-Teutones. We refer to the Scythians and 
 Pelasgi. It appears to us certain that the Pelasgians were the 
 great southern branch of the Sclavonian stock, which, starting 
 from Khorassan in an age long anterior to chronology, spread 
 itself over the whole of Sarmatia, and eventually furnished a 
 
 Greek against the violator of the tomb, the Lycian threatens him with a 
 fine." 
 
 * Hamaker, Akademische Voorlezingen, p. 14. 
 f The systematic and military form which characterized the emigra- 
 tions of the High German and High Celtic tribes, may be seen in the 
 accounts given of the movements of the Teutonic and Cimbrian tribes 
 conquered by Marius little more than 100 years b. c. 
 
CHAP. IV.] OF THE ANCIENT GBEEKS. 157 
 
 large substratum of population to Thrace, lUyria, Greece, and 
 Italy. It is also pretty clear that these Pelasgi recrossed into 
 Asia by the Hellespont, and colonized the Western coasts of 
 Asia Minor and the islands of the Archipelago long before the 
 Helleno-Teutones appeared on the stage. We conceive that the 
 Scythians, properly known under this name, were the great 
 Low German tribe of Getce, Guths^ or Goths. The prefix de- 
 notes that they were^sa- iroiAs, or points to their Asiatic origin; 
 and we conclude that they were identical with the Sacse*, 
 who gave their name to the other great subdivision of the low 
 German family — the Saxons. We trace them to an original 
 settlement a little to the East of the Sclavonian or Sarmatian 
 Pelasgi, namely, to Bokhara, or Hindu-kuh; and we entertain 
 no doubt that it was the same branch of the Iranian race which 
 invaded the Panjab and Hindostan, and established there the 
 Sanscrit language and the Brahminical religion. In Europe we 
 find the Getse or Scythae occupying the lower Danube, and 
 stretching in a North-westerly direction to the Baltic and Ger- 
 man ocean. It is easy, therefore, to distinguish between the 
 Getse and the Helleno-Teutones. But we have [to guard our- 
 selves against the risk of vagueness in regard to other tribes, 
 which is likely to be produced by the very lax and general 
 manner in which the ancients employed the name Scythian. It 
 is made to include all the tribes to the North of the Euxine and 
 Caspian, and may therefore point to branches of the Turanian, 
 Celtic, and Sclavonian stocks, as well as to the Low Germans, 
 whom it strictly and appropriately indicates. We ought there- 
 fore to adopt a classification which would distinguish between 
 the Scythians properly so called, namely, the Getae and Sacse, 
 whom we may term the Teutono-Scythians ; and the pseudo- 
 Scythians, i. e. (1) the Mongols or Turano- Scythians ; (2) the 
 
 * The term Sacce was used by the Persians to include all the Scy- 
 thians : 0^ γαρ Πέρσαι πάνταξ τονς Σκνϋ•αζ καλεονβι Σάν,αζ (Herod, τιι. 64), 
 and similarly Greece in general was the land of the lonians {'Ιαόνων γη, 
 iEsch. Pers. 182). It is remarkable that the Indians also included under 
 the same names of Sacce and lavani all the nations living to the north and 
 west of their neighbours the Pahlavi or Persians. See Rdmayana, Lib. i. 
 c. 54, 9I. 20, and Schlegel's note on his Latin version, p. 168. Gorresio 
 says {Rumuyana, vi. p. 443): "il nome di Yavani venno dopo I'eta 
 d'Alessandro il grande applicato ai Greci." 
 
158 THE ETHNOGEAPHIC AFFINITIES. [bOOK I. 
 
 CimmeHi or Celto- Scythians; and (3) the Sauromatce or Slavo- 
 Scythians'^. In the great country of Thrace we must admit the 
 presence of both Getae and Sarmatae ; and as the name Θραξ in- 
 volves the root Tor or Dor ^ we must also recognise an admixture 
 of the Helleno-Teutones. In fact, in all countries which have 
 been the highway of migration, we must expect that ethnical 
 elements will be fused together in an entanglement which no 
 modern knowledge can be expected to unravel. 
 
 94 Although we entertain no doubt whatever as to the 
 ethnical affinities of the ancient Greeks and old High Germans, 
 we do not overlook the fact that the inhabitants of Hellas 
 owed their early civilization and some very peculiar features of 
 their literary culture to a source with which the Teutonic races 
 had no connexion except through a Greek medium. While the 
 Indo-Germanic tribes were sj)reading to the Ganges on the one 
 side, and to the Atlantic and North Sea on the other, the Syro- 
 Arabian or Semitic family was gradually difiusing the primitive 
 civilization of our race from the river-lands of Mesopotamia and 
 Egypt along the whole of the south coast of the Mediterranean. 
 Under the name of Phoenicians they were the earliest navigators 
 and colonists of that great island seaf. In some of the islands 
 they constituted the largest part of the population. But every 
 
 * That the Scythians were at least in part Slavonians is proved by 
 the fact mentioned by Linde {Slownik Polski, u. p. 1042), and Garnett 
 {Essays, p. 248), that the Sclavonic kolo, "a wheel," and its Polish deri- 
 vative kolasa, " a wheel-carriage," French caleche, existed in Scythia in the 
 days of Ovid, who says: 
 
 Gens inculta nimis vehitur crepitante colossa; 
 
 Hoc verbo currum, Scytha, vocare soles. 
 
 That by "Scythian" he means " Sarmatian" is clear from TVisi. in. 12, 30. 
 
 Stridula Sauromates plaustra bubulcus agit. 
 
 f Appian says {Hist. Rom. viii. 1) : Καρχηδόνα την tv Αίβντ] Φοίνικεβ 
 φκιΰαν, 'ίτεαι ττεντήκοντα ηρο άλώΰεως Ίλίον, \. e. in β. c. 1234. It is proba- 
 ble, therefore, that they had navigated the Mediterranean for a long time 
 before they undertook this colonization on a large scale. And the ex- 
 treme antiquity of tlie name of Mount Atlas, which is clearly of Semitic 
 origin (Wetter, der Mythus von Atlas, Mainz, 1858, pp. 30 sqq.), shows 
 that the Phoenicians must have established themselves on the north- 
 western coast of Africa from the very earliest period. 
 
CHAP. IV.] OF THE ANCIENT GREEKS. 159 
 
 where they communicated the cognate arts of architecture and 
 writing, and imparted not a few of the religious and philo- 
 sophical dogmas which form the basis of European mythology. 
 Indeed, when, at a later period, Pythagoras and Plato sought 
 instruction from the fountain-head of Asiatic wisdom, they did 
 but confirm the belief which was floating vaguely on the surface 
 of Hellenic tradition. The influences of Phoenician culture must, 
 however, be traced back to a time when the Thuringian Greeks 
 had not begun to descend upon Thessaly and the rest of the 
 peninsula, and when the Sclavonian Pelasgians were still the 
 ruling caste. In fact, it was in the islands of Crete, Rhodes, and 
 Cyprus, and on the West coast of Asia Minor, that the Phoeni- 
 cians first taught the Thraco-Pelasgians those arts which made 
 the tower-builders of Argos and Italy look back to Lydia with 
 mysterious reverence, or which the Greeks themselves subse- 
 quently derived from the Southern Islands of their narrow sea. 
 
 95 We are not however to suppose with Dr. Roth*, who 
 has ably advocated the claims of the Phoenicians to a large 
 share in the early cultivation of Greece, that the name Πελαΰ- 
 yog is therefore of Phoenician origin. He maintains that this 
 word is merely another form of the designation of the PJiilis- 
 tines^ namely, "^niibs P,lishti; and that, as this is a deriva- 
 tion from n^bs, Pjlesheth^ the last letter of which is a servile, 
 the original form of the ethnical name must have been "'ώ^ε, 
 PyldsM^ "the wanderer," which is analogous to the jEthiopic 
 falasi^ "peregrinator." Consequently, the Kari, Krethi, Plethij 
 and P^ldshi^ were synonymous names for the Phoenician rovers 
 who colonized the islands of the Mediterranean. Now to omit 
 all strictures on the details of this etymology, which is faulty in 
 itself, we remark that, with the fullest admission of the many 
 contacts between the Phoenicians and Pelasgians, we are bound 
 to conclude that the name given to the foreigners by the 
 Greeks, who spoke about them, would more probably be a term 
 significant in their own language, than a foreign word which 
 conveyed no meaning to those who used it. Besides, this 
 
 * Geschichte unserer ahendlandischen Philosophie (Mannheim ^ 1846), 
 Vol. I. notes 25, 28. 
 
IGO THE ETHNOGEAPHIC AFFINITIES, [bOOK I. 
 
 etymology does not explain the cognate term Πελο^^ which is 
 synonymous with Πελαΰγός^ and points to an emigration from 
 Asia Minor to Argolis, indisputably connected with the progress 
 of Phoenician civilization. All nouns ending in -οψ refer to 
 the colour or shape of a substance, in fact, to that which most 
 strikes the eye in its outward appearance. Πελο'φ is therefore, 
 like Αΐ%ίο-φ^ an epithet descriptive [of the complexion*. We 
 have shown elsewhere that it signifies "swarthy of face," and 
 that Πελ-αΰγός means "the swarthy Asgian," or " Asiatic f." 
 
 * Although no one will deny this position in the case of Αίϋ-ί-οψ, άΙΟ•- 
 01/) and οΐν -oip, it is not thought so obvious that the same mode of interpre- 
 tation applies to ην-ο'ψ, μέρ-οτρ, vccQ-oip and οτέρ-οιρ. Hesychius renders 
 ■ήνοιρ, πάνν ενηχος, λαμπρός, διαφανής. As the word is an epithet oi χαλκός, 
 ουρανός and ηνρός, it can only imply a bright, yellow, golden colour, and 
 the word must be the Hellenized form of the original jan-ops or diav-oip, 
 referring to the light of day. Νώρ-οτρ is also an epithet of χαλκός, and is 
 considered by Plutarch, who makes some bad jokes on the subject, as 
 equivalent to λαμπρός and διανγής (^Conviv. Disput. Qb^B, 692 f = m. 689, 
 847, Wyttenb.) We believe that the root νωρ- is simply νηρ- with the usual 
 change of weight {Gr. Gr. 22), and that the latter, found in Νηρενς, νήρι- 
 τος, implies a perpetually floating motion, as of waves or leaves, and is 
 therefore applicable to the flashing and flickering άοτραπή of polished 
 metal. We have the same combination of meanings in the Latin micare 
 and coruscare: see Doderlein, Etym. u. Sy7i. ii. 79 sqq. It has generally 
 been thought that μέρ-ο'ψ refers to speech, and means "having an articulate 
 utterance." We have no hesitation in recognising in this compound the 
 root μαρ- oi μαρ-μαίρω, μάρ-μαρος, μορόεις, "bright," "shining;" and we 
 think, that, as an epithet of men, it implies a fair as opposed to a dark 
 complexion: so that the Μέροπες (of Cos and elsewhere) were opposed to 
 the Πέλοπες. According to the usual interpretation of ατέρ-οιρ it is syno- 
 nymous with ηνο'ψ and νώρο'ψ (cf. Soph. Antig. 1114; Eurip. Phceniss. 235 c, 
 schol.) : so that it will be connected with ατεροπή, άοτραπή, and ultimately 
 with άοτήρ, ά-οτέρ-ος. Lobeck, ρηματυαόν, p. 41, says: ^^βτεροιρ arbitror a 
 ΰτέρω sive ατρέω ductum cum significatione torti et volubilis; certe ΰτέρο'ψ 
 λιγννς apud Sophoclem nihil diifert a λίγννς οτρόβιλος in Dosiadae Ara i. 
 V. 5 , quod epitheton pluribus verbis explanat Apollonius i. 438 : λιγννς 
 πορφνρεαις ελίκεβοιν άΐαοοναα id est gyros agens, wirhelnder Ranch (curling 
 smoke)." If so, ατέροιρ is a synonym of καλαΐίροι/> = κραΓροι/). Doderlein, 
 Etym, u. Syn. i. p. 21. 
 
 ■j- Varronianus , p. 29; Kenrick, Phil. Miis. ii. 353. We may add that 
 πέλος is translated νπόφαιος in a gloss on a Fragment of Sophocles, 122 
 Dind.: κννος πέλες τε μηκάδος βοος ρινόν. It is somewhat comical that 
 with these evidences in favour of an appellation significant of a dark and 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 161 
 
 The last part of this name is clearly connected with the ethnical 
 designation Τ53ώϊί, liashk^naz^ "Ascanius," which the book of 
 Genesis assigns to one of the sons of Gomer; we have "Αβαύ- 
 γοι and "Άβαΰγία in Asiatic Sarmatia, on the eastern shore of the 
 Euxine; and as we find the Anglians by the side of the English, 
 (pronounced Ing-lisJi) and Ing-cevones, so we may recognise this 
 name in the Frankish Isc^cevones (above, p. 127, note, and p. 130, 
 note). But there is not the same objection to a Semitic etymo- 
 logy for the name of the TelchiT^es^ who introduced the useful 
 
 swarthy hue, Dr. Hitzing should have been led to suppose that the Pelasgi 
 owed these names to their fair and white complexion. He says {Urge- 
 schichfe u. Mythologie der Philistaer^ p. 44) : "I consider πελαογός identical 
 with the Sanscrit world valaksha, or halaksha 'white.'" This etymology 
 does not seem to us to need refutation. Its intrinsic improbability is a 
 sufficient obstacle to its reception. Dr. Karl Meyer proposes a Celtic 
 etymology. He says {Gel. Anzeig. d. bayer. Akad. d. Wiss. 1843, p. 
 709): "I refer this name Πελα6γυς , with perfect certainty, to the Celto- 
 Sanscrit (also Hebrew) root beh, 'high,' 'proud/ S. balh, valh ('eximium 
 esse, excellere') Cymbr. balch, gwalch, Irish balach, bale, a derivation which, 
 besides its obvious propriety as expressing the race of Niobe, is rendered 
 more probable because by its double reference in Celtic to bodily and 
 mental elevation, especially the flight of birds — Cymric gwalch^ 'soaring,' 
 aar, 'Falke' — it gives a natural explanation of the assumed connexion 
 between Πελααγόζ and πελαργοί." Another Celtic explanation has been 
 lately given in the North British Revieiv^ No. lix. p. 101, namely that 
 Pelasgia is Beil-as-ce, "the land of permanent habitation," an excellent 
 name doubtless for a tribe specially distinguished as wanderers! The 
 same writer has conclusively proved his philological incompetency by 
 maintaining confidently that the nomen Cincinnatus, which is a common 
 Latin epithet, is the original name for dictator, i. e. in Erse Gean-cean-eat- 
 eis, "the head over the head of the country people!" {lb. p. 98). Craine 
 {Philologus for 1855, p. 576 — 590) connects Πελαβγός with niXayog in 
 the sense of wald. We not only adhere to the derivation which con- 
 nects the ΙΊελαΰγοί with the ΙΊέλοτζες, but also recognise a confirmation 
 of this view respecting the first syllable in the world ηελκργός. Dr. Hitzig's 
 objection (p. 46) that πελ-αργός could not signify a juxtaposition of two 
 distinct colours in the plumage, but must signify a blackish white, 1. β. 
 "grey," does not appear to us at all valid. The main colour of the bird 
 is white, which seems to be superinduced upon a lower coat of black, so 
 that πελαργός, "the black but whitened bird," is quite analogous to πν- 
 γαργος, "the white-backed eagle," which ^Eschylus, Ag. 114, describes 
 thus in opposition to an eagle entirely black; οιωνών βαΰΐλεύζ, δ κεΧαινος 
 ο τ εξοτίίν άργάς. 
 
 Μ 
 
162 THE ETHNOGEAPHIC AFFINITIES [bOOK I. 
 
 arts among the Cretans, Cyprians, and Rhodians, and who gave 
 to the city ofSicyon one of its most ancient names; and we 
 feel disposed to accept Dr. Roth's suggestion that Tel-cJiin, the 
 son of EuropSy was another form of Tubal-qain^ and that the 
 same word appears in the name of the Tihareni or Chalyhes^. 
 Consequently, the Tel-chines were merely the ''copper-smelters ;" 
 and though the name of the Dak-tyli is formally grecized, 
 we can with him recognise in this a Semitic compound con- 
 taining the same root bri or bari, and signifying the "copper- 
 miners." So that the two names occur in a very natural juxta- 
 position. We think there are also good grounds for a belief in 
 the Phoenician origin of the names of the Plethi and Idaei, 
 names more or less connected with the Phoenician traffic with 
 the islands of the Mediterranean. It may be remarked, that 
 Cretan and Phoenician legends are usually commutable. 
 
 96 The difference, therefore, betw^een the old or Pelasgian 
 and the classic or Hellenic language, must have been the same 
 in kind with that which constitutes the distinction of the whole 
 Indo-Germanic family into two great branches; and in investi- 
 gating the origin of the Hellenic forms our problem is, from the 
 given Greek, to reproduce the Pelasgian, word, — to pass from a 
 language, which in its known state gravitates towards the High 
 German or Persian, to one which was entirely Low Iranian in 
 its structure. Although we shall be careful to point out in the 
 following pages the peculiarities which distinguish the Greek 
 words, as we have them, from their original type, as it may be 
 restored from a comparison of the oldest languages of the family^ 
 it may be convenient to state here, briefly and generally, what 
 are the laws regulating the use of the consonants in the Hellenic 
 idiom, as contrasted with what we can infer with respect to the 
 Pelasgian or older state of this language. The Hellenic or 
 classical Greek tolerates only four consonants at the end of 
 words, — κ, ν, ρ, g. The first of these forms the termination of 
 two words only — ουκ and εκ; in the former it is a mutilation of 
 
 * It is right to mention that a very different view is maintained by 
 Knhn, Zeitschr. f. vercjl. Sprachf. i. pp. 193 sqq. 
 
CHAP. IV.] OF THE ANCIENT GEEEKS. 163 
 
 κε, in the latter of κίζ. It will be observed, too, that ονκ never 
 occurs before a consonant, and εκ never before a vowel. Of the 
 other three consonants which may be finals, ν is often a repre- 
 sentative of g, as in τυτίτομεν. Moreover, when ς is the final 
 letter it must not be preceded by a dental or a liquid. An 
 analysis of the Greek language, and a comparison with other 
 members of the Indo-Germanic family, assure us that in the 
 more perfect form of the inflexions these rules could not come 
 into application: for in the old language no consonant would 
 ever be required to stand at the end of a word, or before another 
 consonant, because no consonant is ever articulated in the primi- 
 tive state of a language without a vowel following it. With a 
 liquid, as we shall see, the case is somewhat difi'erent: the arti- 
 culating vowel may be placed either before or after it; and this 
 is the reason why the only three consonants in common use as 
 finals are liquids. The Greek language first cast away its final 
 vowels, and then the consonant of the suffix yielded to the laws 
 of euphony. Besides these rules touching the final consonants? 
 there are also others respecting initial consonants, and those in 
 the middle of a word, which are just as distinctive of Hellenism 
 as the others. One of the most remarkable of these appearances 
 is the tendency to reject the digammaor aspirated labial, whether 
 at the beginning or in the middle of a word. This sound has 
 completely vanished from the Greek with which we are most 
 conversant, though traces of it still remain in the poems attri- 
 buted to Homer. The ^ίJ-sound seems frequently to have ap- 
 peared in the ante-Hellenic language preceded by d, κ, 6, τ. 
 In these cases either the first letter has been dropped , and the 
 w somewhat modified, as in φίν from (5φίν; or, what is more 
 common, the w has been omitted, as in δε for δνε. In Hellenic 
 the initial β is almost always changed into Λ, a principle also 
 observed in Zend as compared with Sanscrit, and in Welsh as 
 compared with Erse: similarly j is either changed into ξ or 
 Λ, vocalized into t, or assimilated, and this also takes place in 
 Zend as compared with Sanscrit: in Sclavonic the j appears 
 under the form 5Λ, which, we shall see, is a step towards z^ 
 though still a step from it. Finally, the following combinations 
 of consonants cannot take place in the Hellenic language — 
 μρ, μλ, βν, δλ, vq; but wherever they appeared in the older 
 
 Μ 2 
 
164 THE ETHNOGKAPHIC AFFINITIES [bOOK I. 
 
 language we have μβρ or βρ, μβλ or βλ^ μν, λλ and νδρ. It is 
 unnecessary to add, that these prohibitions against the use of 
 certain consonants and combinations of consonants interfered 
 materially with the discrimination of the root and termination, 
 and, by ruining the inflexions, gave occasion to some of the 
 most remarkable peculiarities of Greek syntax, such as the use 
 of the article and of the prepositions. 
 
 97 After what has been said, it is scarcely necessary to 
 mention that the diflPerent degrees in which the old Pelasgian or 
 Slavo-Phoenician language of the South was affected by the 
 Hellenic or Teutono-Persic language of the North, constitute the 
 differences of dialect about which Grammarians have written so 
 voluminously. Consequently, all dialectical distinctions in the 
 Greek language must resolve themselves into one or other of 
 two great classes ; and the ancient Greeks were well aware of 
 this when they regularly opposed the Dorians to the lonians*. 
 The former, as Λνβ have seen, were the representatives of the 
 High German warriors, who gradually forced their way, in an 
 united and distinct body, from the North of Thessaly to the 
 promontory of Taenarum. In Greece, as in Germany, these 
 Thuringians were remarkable for the military concentration 
 which kept them from being absorbed by the populations of 
 the invaded countries. They formed everywhere a distinct 
 caste, an aristocracy of conquest. But as every army, however 
 well organized, has its train of undisciplined followers, and 
 leaves crow^ds of stragglers on its line of march, so we find the 
 Dorians, in their progress through Thessaly, leaving behind 
 them similar detachments of their forces; and these stragglers, 
 having combined themselves with the Pelasgians of that district, 
 were called ΑΙολείξ^ or "mixed menf ," a name which was re- 
 
 * This opposition was not neglected by the author of Gen, x. 4, who 
 makes 'Helishuh or Hellas the eldest son of Javan or Ίάων. 
 
 f Grimm supposes that the name Αιολείς refers to their parti- coloured 
 clothing (Gesch. d. deutsch. Spr. p. 296): ^'faioXsLg d. h. die hunten; gleich 
 Britten und Picten fuhrten iEolier den Namen der buntgekleideten." 
 But if the name had referred to the clothing of the tribe, it would surely 
 have been in a compound form like αίολομίτρης, αίολο&ώρηξ, κορνϋ-αίολος. 
 As an epithet of a tribe, αίολενς must be compared with its other form 
 
OHAP. IV.] OF THE ANCIENT GEEEKS. 165 
 
 tained by the Thessaliang and Boeotians long after the opposi- 
 tion of Dorian and Ionian had established itself in other parts 
 of Greece. The ethnographical fact is preserved in the legend* 
 that "Hellen left his kingdom to ^olus, his eldest son, while 
 he sent forth Dorus, and Xuthus the father of Ion, to make 
 conquests in distant lands." This mythical genealogy makes 
 Ion not the son , but the grandson of H^Uen ; and it has been 
 shown by Mr. Kenrickf , that the name of Xuthus, which is in- 
 terposed, is simply an epithet of the Dorian God Apollo, who 
 was the O^foS τίατρωος of the lonians. From all the circum- 
 stances known to us, we are entitled to infer, that the lonians, 
 wherever they retained their independence, were only partially 
 influenced by the Dorians : the Pelasgian element in their com- 
 position remained for a long while in full force , though they 
 adopted the religious tenet of the Dorians, and paid homage to 
 the conquering God under whose auspices the invaders marched 
 and fought. We have shown above that the Dorians, accord- 
 ing to the primitive meaning of their name, were called "High- 
 landers or mountaineers," and Mr. Kenrick, who has derived the 
 same result from a Greek etymology of the name, has shown 
 that the lonians were emphatically the "Men of the coast" 
 QHCovLo)^ and that they were also called the "Beach-men" 
 {ΑΙγιαλύζ)^ or•" Sea-men" QA%aLoi^i and he remarks also that 
 "the distinction between Doric and Ionic in later times an- 
 swered very well to that which has been observed to prevail 
 between the speech of mountaineers and of littoral nations, — 
 one being harsh and broad, the other smooth and liquidj." 
 
 aoXlris, i. e. "pressed together, standing side by side," just as ^foAog, the 
 god of the different winds, expresses the meaning of his name in αελλα, 
 Ά turbo, or whirling together of objects from all quarters. From the 
 idea of juxtaposition without fusion, we get the signification of αίόλος 
 with reference to stripes or bands of alternate colours, as distinguished 
 from ποίΥ,ίλος, which denotes variation of colour by way of spots or cir- 
 cles (below, § 266). The meaning of uiolog, as indicated by the epithet 
 y.OQvd^aioXog is well illustrated by the alternate black and white in the 
 crest of an armed figure represented on an ancient vase in the British 
 Museum (see Gerhard, Athenem Gehurt. Berlin, 1838. Taf. ii. 2). 
 
 * ApoUodor. i. 7, 3, 1; Thirlwall, i. p. 101. 
 
 t The Egypt of Herodotus, p. lix, note 2. 
 J Ibid. p. Ixi. 
 
166 THE ETHNOGEAPHIC AFFINITIES, ETC. [bOOK I. 
 
 We must not forget, however, that there were other differences 
 of a more important and extensive nature ; and that the Doric, 
 or purely Hellenic element, at length so completely asserted 
 itself, that we can only by a laborious process succeed in par- 
 tially reproducing the articulation and structure of the old 
 Pelasgian speech. The broad distinctions therefore are not to 
 be expected in the four dialects, which, at a later period, were 
 rather names of different branches of literature, than four 
 varieties of spoken language. The ^olic dialect (η Αιολίς)^ in 
 this sense, referred to the lyric poetry cultivated at an early 
 period by the jEolians of Lesbos ; the Doric (ji ζ/ωρΐ;?), to the 
 choral poetry of the Dorians; the Ionic {η 'lag)^ to the epic 
 poetry of the lonians ; and the Attic (η ^Ατ%ίξ)^ to the universal 
 literature of that branch of the Ionian race which had settled in 
 the "Promontory-land" (Ύΐ^Αττιζη, οτΆκτίκη). The conquests 
 of Alexander carried this last, in a less pure and vigorous form, 
 into Asia and Egypt, where it incurred various corruptions, and 
 became Hellenistic rather than Hellenic. An investigation of 
 this κοινή διάλζκτοξ, as it has been called, does not belong to 
 our present purpose, which is rather to reproduce the more 
 perfect and complete state of the Greek language than to scru- 
 tinize its decayed and feeble condition. 
 
CHAPTER V. 
 THE THEORY OF THE GREEK ALPHABET. 
 
 98 Difficulty of the subject. 99 Every alphabet originally a syllabarium; so that 
 the distinction between vowels and consonants is quite arbitrary. 100 (i) Se- 
 mitic origin of the Greek Alphabet. Semitic alphabet consisted originally of 
 16 letters organically arranged. 101 These 16 letters formed the original 
 Greek alphabet. 102 Subsequent additions to the Greek alphabet. 103 
 Poverty of the Egyptian hieroglyphic alphabet. 104 Artificial arrangement 
 of the Arabic characters. 105 (2) Analysis of the Greek Alphabet. Prelimi- 
 nary examination of the Deva-Nugan. 106 Inferences deducible from the 
 shape of the Sanscrit characters. 107 Formation of the vowel-signs and ori- 
 gin of the liquids. 108 Theory of the aspirates, sibilants, and secondary 
 vowels. 109 Main difficulties in regard to the Greek alphabet. 110 The 
 Greek digamma. Ill The Latin F. 112 The dental sibilant %. 113 The 
 Greek aspirate. 114 Evanescence of ν and g. 115 Double value of |. 116 
 Etymological analysis of 77 and ω. 117 General review of theGreek alphabet. 
 118 (3) Interchange of mutes in the Greek and cognate languages. Grimm's 
 law. 119 Exemplifications. 120 Exceptions in the case of the Greek lan- 
 guage. 121 Law of divergent articulations. 122 Corresponding consonants 
 in Sanscrit, Greek, and Latin. Appendix to § 110. [a) The digamma as it 
 appears in inscriptions. (6) Extracts from Bentley's MS. on the digamma. 
 
 98 ΠΠΗΕ necessary prelude to an attempt to increase our know- 
 ledge of a dead language is, an inquiry into the value of 
 the symbols or letters which have preserved and transmitted to 
 us its written remains. All languages are made up of sounds, 
 and of these sounds the letters are the only representatives in 
 the case of a language no longer spoken ; unless, thert lore, we 
 can to a certain extent ascertain to what sounds these symbols 
 corresponded, we shall hardly be able to draw a profitable 
 comparison between the language in question and the others to 
 which it is related; nor will it be possible to explain and justify 
 those regular permutations of letters, which time and use have 
 occasioned in languages of the same family, ifAvedo not discover 
 what was the value of this notation in the first instance. To 
 obtain this knowledge, the great philologers of the present day 
 have laboured diligently ; but though they have collected an 
 immense mass of facts, and have heaped up materials for the 
 future labourer to work upon, they have left so much room 
 for arrangement and construction, that this subject is the 
 most difficult part of our task. The Greek alphabet presents 
 
168 THE THEORY OF [bOOK I. 
 
 peculiarities of a most embarrassing nature. It derives its charac- 
 ters and their arrangement from a family of languages with 
 which it has no immediate connexion, and the whole develop- 
 ment of its system of writing is at variance with the notation 
 on which it is based. We must, therefore, consider as inde- 
 pendent questions (1) the Semitic origin of the Greek alphabet, 
 (2) the actual value of the different letters as used by the 
 Greeks, and (3) the changes which take place in consonants of 
 words as represented in the different idioms of the Indo-Ger- 
 manic family. It will, however, be as well to begin with a few 
 remarks on alphabetical writing in general. ^ 
 
 99 According to the grammatical system which has de- 
 scended to us from the Greeks, we are taught from our earliest 
 years to distinguish between vowels and consonants, and to re- 
 gard them as necessarily having a separate existence. This is 
 a notion which must be at once discarded by every one who 
 would make any progress in philology. Language is a transfer 
 of the thoughts to the outward world of sense: when this is 
 effected by sounds, it is speech; when by symbols, it is writing; 
 but as men speak before they write, every symbol is a repre- 
 sentative of some sound : it is in itself an element of language. 
 There are some languages in which each symbol represents a 
 whole word ; such is the case in the Chinese. But in all lan- 
 guages every symbol must have been significant in the first in- 
 stance. Consequently , there could not be any distinction into 
 vowels and consonants, but the alphabet must have been a 
 syllabarium, the elements of Avhich might or might not be in- 
 dependent words. "By words^'' says W. von Humboldt (iib&r. d, 
 Versch. d. menschL Sprachb. p. 74) , " we understand the signs 
 of individual conceptions. A syllable forms a unity of sound, 
 and becomes a word when it obtains an independent significa- 
 tion; but for this a combination of several syllables is sometimes 
 necessary. A doubled unity — of sound and conception — meets 
 in a word." The distinction of these syllables into consonants 
 and vowels is perfectly arbitrary. Neither a vowel nor a conso- 
 nant can have any separate existence in spoken language: the 
 consonant always requires a vowel-appendage in order to be 
 pronounced; the vowel cannot be pronounced without an initial 
 
CHAP, v.] THE GREEK ALPHABET. 169 
 
 breathing, which is sometimes so strong as to become a definite 
 consonant. In either case the vowel can be regarded only as a 
 modification of its fulcrum. Hence, in all ancient alphabets, we 
 find that the vowels are not in the first instance expressed by 
 separate symbols, but, as the indistinct a or e, which originally 
 accompanied every consonant, was in process of time developed 
 into distinct vowel-sounds, these were denoted by various hooks 
 or points attached to or written under the consonants to which 
 they referred, or, at the beginning of the word, to the mark 
 denoting the breathing with which they were pronounced. At 
 first, then, there were only two sorts of letters, — breathings and 
 consonants, — both of them accompanied by short vowels which 
 were not expressed, or by modifications of these vowels expressed 
 by certain marks pertaining to the original symbol. The first 
 deviation from this original state would take place in those 
 languages, which, like the Indo-Germanic, did not use many or 
 very various breathings, and in which the vowels assumed to 
 themselves at an early period important functions in the gram- 
 matical organization. But even then no new symbols were 
 invented for the vowels. It was thought sufficient to adopt for 
 their expression more or less mutilated forms of those breathings 
 or consonants with which they were found most constantly com- 
 bined. We shall presently show, from a palseographical ex- 
 amination of the Greek and Sanscrit alphabets, in what manner 
 this was efiected. 
 
 100 (1) Semitic origin of the Greek Alphabet, 
 The traditionary history of the Greek alphabet is well known. It 
 is said to have originally consisted of only 16 letters, which were 
 brought from Tyre by Cadmus, and to which 4 were added by Pala- 
 medes at the time of the Trojan war, and subsequently 4 others by 
 Simonides of Ceos (Plin. Hist. Nat. vii. 56*). Other inventors or 
 importers of the alphabet are also mentioned (Schol. Dionys. Thr. 
 Bekk. Anecd. p. 784), perhaps with as much reason as those to whom 
 it is ordinarily attributed ; for all that we are to understand by these 
 
 * A vase found at Agylla has inscribed on it, in alphabetical order, all the 
 letters except ii, the digamma and koppa being inserted in their proper places. 
 The age of this vase is doubtful. See Franz, Elementa Epigraphices Grcecw, 
 Berolini, 1840, p. 22. Cf. Lepsius, Annal. Arch. InsHt. Vol. viii. 1836, 
 pp. 188—203. 
 
170 THE THEORY OF [bOOK I. 
 
 traditions is, that the alphabet was of Semitic origin, and this we can 
 discover for ourselves from an examination of the characters and their 
 arrangement. A knowledge of this fact, however, is of the utmost 
 importance; for the chief difficulties occasioned by the Greek alphabet, 
 have arisen from the circumstance, that its whole organization is 
 adapted to a language as widely different as possible from the Greek, 
 and that while the names and shape of the letters have been retained, 
 their value has been materially altered. It will be instructive to 
 inquire, what were the original 16 letters which the Greeks derived 
 from their intercourse with the Phoenicians, and how they came to 
 adopt in the first instance a part only of the Semitic syllabarium ; for 
 there are certainly more than 16 of the Greek letters which agree in 
 name and shape with the Phoenician and Hebrew as they are known 
 to us. The fact is, in our opinion, that the original Semitic alphabet 
 contained only 1 6 letters. This appears from the organic arrangement 
 of the characters*. The fundamental elements of a syllabarium are 
 the mutes, the breathings, and the liquids. Of these the most neces- 
 sary are the first two ; after these would come combinations of strong 
 breathings with mutes, or aspirated mutes ; and the liquids, which are 
 always secondary sounds, would be introduced last of all. In most 
 alphabets we find the mutes divided into three classes: ienhesp, k, t; 
 aspirates ph (f), kh (h), th; medials b, g, d. The first and third 
 orders do not, however, necessarily coexist. There are some nations, 
 as for instance our own Highlanders in Wales and Scotland, who can 
 never, even in English words, pronounce the medials, and even in those 
 languages which have both orders in constant use, as in the Greek, if a 
 tenuis becomes a medial, an adjoining tenuis is also changed into a 
 medial, as ετίτά, έβδομος. The old Italian and Runish alphabets had 
 no medials, and the Semitic nations seem to have little need of tenues ; 
 there is no ρ in the Arabic language, and, at the present day, most of 
 the Arabs pronunce their (^ caf as ga (Lepsius, Abhandl. p. 16). 
 Some European nations have adopted a set of vacillating middle 
 sounds, which sometimes approach to the tenues, at other times to the 
 medials ; for example, it requires a very practised ear to distinguish 
 whether a Saxon says Leibsig or Leibsik. If, therefore, the tenues 
 
 * This organic arrangement of the alphabet has been more or less 
 noticed by several philologcrs, of whom the earliest seems to have been 
 the acute and learned Dr. Richard Lepsius, in his essay uber die Anordnung 
 und Verwandtschaft des Semitischen, Indischen, Athiopischen, Alt-Persischen, 
 und Alt-Agyptischen Alphabets (Zwei Abhandl. Berl. 1836). It is rather 
 vsurprising that so obvious a phenomenon should have escaped the notice 
 of any observing grammarian. The deductions, however, in the text do 
 not appear to have been anticipated by any former writer. See Latham, 
 English Language^ p. 200. 
 
CHAP. V. THE GEEEK ALPHABET. 171 
 
 were so little used by the Semitic nations, we may presume that the 
 signs for them, as distinguished from the medials, were of later intro- 
 duction, and that they would take up the remaining order of mutes, — 
 the aspirates, — and even the liquids, before they introduced thetenues. 
 Besides the mutes and breathings, the Hebrew alphabet, as it now 
 stands, has four sibilants T, D, U, 115. Now it is quite clear that all 
 these four sibilants could not have existed in the oldest state of the 
 alphabet. Indeed we have positive evidence that the Ephraimites 
 could not pronounce ID, but substituted for it the articulation D 
 {Judges xii. 6). We consider it quite certain, that at the first there 
 was only one sibilant, namely, this or samech. Finally, to reduce 
 the Semitic alphabet to its oldest form, we must omit caph, which is 
 only a softened form of coph^ the liquid resJi, and the semivowel jod^ 
 which are of more recent introduction, as will be shown by and by. 
 The remaining 16 letters appear in the following order: N, i, Λ, i, Ji, 
 1, Π, c:, b, 12^ 5, D, 3>, D, p, n. If we examine this order more mi- 
 nutely, we shall see that it is not arbitrary or accidental, but strictly 
 organic according to the Semitic articulation. We have four classes 
 each consisting of 4 letters : the first and second classes consist each 
 of 3 mutes preceded by a breathing, the third of the 3 liquids and 
 the sibilant, which perhaps closed the oldest alphabet of all, and the 
 fourth contains the three supernumerary mutes, preceded by a breath- 
 ing. The N, which heads the first class, is a simple breathing corre- 
 sponding to the spiritus lenis of the Greeks, the Π which is placed 
 before the second is a hard aspirate, the spiritus asper of the Greeks, 
 It has been found difficult to determine the precise value of 3^, which 
 precedes the third order of mutes : it appears, however, to have been 
 a kind of nasal breathing, of less frequent use, even in the Semitic 
 languages, than either of the others, and therefore more easily cor- 
 rupted in the pronunciation. The principles of the arrangement will 
 better appear if we place the characters first vertically, and then in 
 horizontal classes. According to the first system we have: 
 
 First breathing (mere exspiration), 
 
 Mediae. 
 
 Second breathing (guttural aspiration), 
 
 Aspiratm. 
 
 Liquids. 
 
 Aleph, 
 
 Ν 
 
 'Λ 
 
 Beth, 
 
 η 
 
 Β 
 
 Gimel, 
 
 Λ 
 
 G 
 
 Daleth, 
 
 "Τ 
 
 D 
 
 He, 
 
 ii 
 
 h 
 
 Vav, 
 
 τ 
 
 ΒΗ 
 
 Chetk, 
 
 η 
 
 GH 
 
 Thet, 
 
 Ώ 
 
 DH 
 
 Lamed, 
 
 h 
 
 L 
 
 Mem, 
 
 Ώ 
 
 Μ 
 
 Nun, 
 
 . 5 
 
 Ν 
 
172 
 
 THE THEOEY OE 
 
 [book I. 
 
 Samechy D S The sibilant. 
 
 Ain, y ''h Third breathing (nasal asiDiration). 
 
 Pe, D P| 
 
 Koph, p Q I Tenues. 
 
 Tav, η TJ 
 
 In the horizontal arrangement we shall, for the sake of greater 
 simplicity, omit the liquids and the sibilant, and then we have : 
 
 Breathings. 
 
 Labials. 
 
 Palatals. 
 
 Linguals. 
 
 Ν 
 
 η 
 
 Λ 
 
 "f 
 
 r; 
 
 Τ 
 
 η 
 
 Ώ 
 
 3> 
 
 D 
 
 Ρ 
 
 η 
 
 In this we see, that, while the horizontal lines give us the arrange- 
 ment of the mutes according to the breathings, the vertical columns 
 exhibit them arranged according to the organ by which they are pro- 
 duced. Such a classification is obviously artificial; it is entirely 
 Semitic, and if, as we shall now proceed to show, these sixteen letters 
 constituted the original Greek alphabet, and were so arranged in that 
 alphabet, it is an additional proof of the Semitic origin of the Greek 
 characters; for, although it would be perfectly natural for a Phoeni- 
 cian to arrange his letters in such an order, no Greek could have 
 thought of placing the tenues, of which he made the most constant 
 use in his primary articulations, in the third order of mutes, and after 
 the liquids. 
 
 Before we proceed to the Greek ifiphabet, it will be proper to men- 
 tion an objection which might be raised to the completeness of the 
 classification which we have pointed out. It may be said that 'ύ has 
 no right to be considered as an aspirate of T, and that, therefore, there 
 is at least one objection to our systematic arrangement of the alpha- 
 bet. The Greeks considered their Ό* as an aspirate, not of d but of τ*, 
 because, in their system, the tenues were antecedent to the medials ; 
 but we are convinced that %' differed from 'ύ only by the difference of 
 Greek and Semitic articulation, and that η corresponded in value as 
 
 * In pronunciation, however, -O" corresponded rather to δ and q than 
 to t: see Greek Grammar, 18, g^ note. Some important etymological facts 
 result from this observation, as will be shown in the proper place. 
 
CHAP, v.] THE GEEEK ALPHABET. 173 
 
 in name to r. This view has been warmly advocated by Ewald (Krit. 
 Gramm, der Hehraisch. Spr. § 30, 1), whose arguments have been 
 violently combated by Redslob (Seebode's neue Jahrhucher , Vol. 
 XX. p. 72). The following reasons seem to us to prove that 
 originally Ώ must have been an aspirate and η a tenuis. In the 
 first place, neither the Jews of the present day, nor the Septuagint 
 translators, are a valid authority for the original pronunciation of 
 the Semitic letters. Again, at the time when the sixteen letters 
 mentioned above constituted the whole Hebrew alphabet, there was 
 only one sibilant. Now all aspirates are approximate assibilations, 
 and the aspirated dental, in particular, often degenerates into a 
 sibilant. There are reasons, therefore, for supposing that 'ύ assumed 
 eventually a hard sound, its original functions being discharged by 
 some one of the newly-introduced sibilants. Again , nothing is more 
 common than for th to be supplanted by t : this substitution has taken 
 place in every language of Europe, except our own and that of Greece ; 
 and any one who has read Lucian's joke about the usurpations of r 
 will be aware that even the Greek language was not altogether ex- 
 empt from it {Judicium VocaUum, § 10, pp. 95, 6). The occasional 
 assibilation of the proper i, n, is well accounted for by Ewald. But, 
 besides these arguments, the very analogy of the arrangement, the 
 correspondence of name, and the identity of the oldest Greek and 
 Phoenician forms of Γ3 and &, η and τ (Gesenius, Script. Linguceque 
 Phoenic. Monumenta, pp. 30, 74, and plate i.), are a proof to us that 
 & corresponded to LD and r to n, although u was rather an aspirate of 
 d than of t. 
 
 101 It must now be shown that these 16 letters which constituted 
 the basis of the Semitic alphabet were the same which were in the first 
 instance introduced into Greece. At first no doubt they wrote from 
 right to left, like their Phoenician instructors and other oriental 
 nations, and we have still inscriptions in proof of this; afterwards 
 they wrote βονύτροφηδον, or as the oxen plough the field, first to the 
 right, and then to the left ; and ultimately they wrote consistently from 
 left to right. As they adopted one or the other method they turned 
 their letters round in the direction of the writing; thus they 
 wrote SBTAIMOIVIiBH in the earlier times, and subsequently 
 ΗΕΡΊνΐΟΚΡΑΤΕ5 (I^epsius, PaUlographie, p. 11). Otherwise the 
 letters differed very little from the Phoenician , as may be seen from 
 a comparison of the 1st and 2nd plates in Gesenius. But of course, 
 as the Greeks made very little use of rough breathings, and a great 
 use of vowels , they would at an early period consider the signs of 
 the three primary breathings in the Semitic alphabet, as marks for 
 
174 - THE THEOEY OF [bOOK I. 
 
 the vowels with which they were pronounced, just as they subse- 
 quently vocalized another, the digamma. In the original syllabarium 
 of the Semitic nations, each symbol represented some consonant or 
 breathing with a short vowel-sound attached to it. According to the 
 weight of the consonant or breathing, would be the weight of this 
 vowel-sound. Now it is a philological fact, which will be developed 
 in its proper place, that there are three varieties of the primitive 
 vowel-sound, differing only in weight. The heaviest in a, the lightest 
 e, and ο is intermediate in point of weight. Therefore as He was the 
 heaviest, Ain the next, and Aleph the lightest of the three breathings, 
 they would be uttered by the vowels e, 6, a respectively, and, when 
 the breathings were omitted, as would very soon be the case, these 
 three vowels would stand in their place. The alterations which the 
 Greeks introduced at an early period in the letters which they bor- 
 rowed from the Phoenicians are distinctly alluded to by ancient writers : 
 Herodotus says (v.5 8): οί^εΦο^ν^κες οντοιοίΰννΚάδμφατακόμίνοί — 
 Ιϋηγαγον — ε^ τονς "Ελληνας— γράμματα — τΐρώτ α μίν, τοΐΰί καΐ 
 άπαντες χρεωνταί Φοίνικες' μετά δε, χρόνου προβαίνοντος^ αμα 
 ry φων]] μετεβαλον καΐτον ρνΟ'μοντών γραμμάτων — περιοίκεον δε 
 όφεας τα πολλά των χώρων τούτον τον χρόνον "Ιωνες, di παραλα- 
 βόντες διδαχτ} πάρα των Φοινίκων τα γράμματα μεταρρν^μίύαντες 
 ϋφεων δλιγα εχρέωντο. Similarly, Diodorus (iii.c. 67, p. 297Dindorf): 
 φηβϊ τοίννν (ζίίοννΰίος) παρ "Ελληύι πρώτον ενρετην γενεό^αι 
 ^ίνον ρνΟ'μών καΐ μέλους, ϊτι δε Κάδμου κομίΰαντος εκ Φοινίκης 
 τα καλούμενα γράμματα πρώτον εις την Έλληνικην μετα^εΐνοα 
 δίάλεκτον, καΐ τας προςηγορίας εκάότω τάξαί καΐ τους χαρακτήρας 
 διατνπώΟαί. It is obvious that these authors allude neither to any 
 change in the order of the letters, nor to any difference in the way of 
 writing them, as from left to right, instead of from right to left, but 
 to a slight alteration in the form {ρυ^'μός , Herod, comp. Athenseus, 
 HI. p. 125 F, διατνπώΰαι, Diodor.) and pronunciation (φωνή) of some 
 few of them (ΰφεων δλιγα). With regard to the change of form, it 
 must be supposed that these authors rather spoke from a comparison 
 of the Greek letters of their own time , with those of the Phoenicians, 
 than from any minute antiquarian researches on the subjects. But of 
 course the change of pronunciation principally refers to the substitu- 
 tion of vowels for breathings. 
 
 The grammarians tell us that the original 16 letters of the Greek 
 alphabet were a, β, γ, ά, ε, ι, κ, λ, μ^ ν, ο, π, ρ, 6, r, ν (ScJiol Dion. 
 Thr. p. 781). That this was not the case will appear from the follow- 
 ing considerations. It is well known that the sixth letter of the old 
 Greek alphabet was not t, but J^, βαν^ vau, or the digamma, as it was 
 indifferently termed, which, under the form g , was used to the latest 
 
CHAP, v.] THE GREEK ALPHABET. 175 
 
 period as a mark for the number 6; therefore v, which is said to have 
 been a sort of substitute for this letter (Marius Yictorinus, pp. 24, 
 68), could not have been included in the old alphabet. We shall show 
 presently that t, like the Hebrew jod, was a secondary and derived let- 
 ter, as was also the letter ρ. The oldest k, which was koppa, Q, stood 
 next to 7t, and there could hardly have been two ^'s in a primeval 
 alphabet. We assert, then, that l, ρ, and ν did not form a part of the 
 old Greek alphabet of 16 letters, and that κ or rather q stood after 7t. 
 Omitting l, ρ, ν, and κ from the letters mentioned by the grammarians, 
 we have with Γ and Q only fourteen. Which were the remaining 
 two? The letters which stood next to ε in the complete Greek alpha- 
 bet were η and Ό*, and we are convinced that they followed p in the 
 original 16, though the former had subsequently a very different value 
 from that which it originally possessed. In old written monuments 
 which have come down to us , η or Η is used as the common mark, of 
 aspiration; and therefore corresponded to the Hebrew ii or he*, i. e. 
 the double was used for the single aspirate after the latter had become 
 a mere ε ipiXov ; but this is sufficient to prove the antiquity of the 
 character; Ό* also occurs in very old inscriptions. A further confir- 
 mation of the opinion that p, η, 0", occupied the same places in the 
 original Greek alphabet that vav, clieth^ and teth did in the old Semi- 
 tic, is furnished by the fact, that, when the Greeks left off writing 
 the Ρ and emjDloyed Η to represent a long ε, they added to r, the 
 last letter of their old alphabet, ν and φ as approximate representa- 
 tions of p, and χ to replace H. ' The two corresponding alphabets of 
 16 letters were, then, as follows: 
 
 Vi ΒΓ^ Έ ΓΗΘ AM Ν Σ Ο 
 
 D ρ η 
 
 Π Q Τ 
 
 102 In the Greek alphabet, as it is now given in the grammars, 
 J^ and Q are omitted, and 1 other characters added to these. When 
 and by whom they were invented or introduced is of little importance 
 in regard to our present purpose. Thus much may be conjectured 
 Λvith safety. As soon as the Greeks ceased to employ p, and Η (as 
 an aspirated consonant), which was very early, v, φ, and ^, must have 
 come into use; they all occur in the oldest inscriptions; indeed it is 
 only on the columna Naniana that the two latter are written Piif, 
 ΉΚ, and the genuineness of that tablet has been doubted; at all 
 events they were antecedent to | and ψ, which are written ΗΣ, Φ Σ in 
 
 * Thiersch supposes that Η corresponded to the Hebrew rr, cheth-y he 
 says (p. 24 Sandf.) : "thus the liver is named in Hebrew chapar (~ΐ~), 
 Greek rjnccfj , which was written ΗΕΠΑΡ (ΗΑΠΑΡ).''' Can he have 
 mistaken the daleth for resh, the heth for pe, and the caph for cheth 'm 
 the Hebrew "as kaved? 
 
176 THE THEORY OP [bOOK I. 
 
 old inscriptions *. The Semitic Tsade (it) and Cajpli (r:) of course sug- 
 gested ζ and κ ; and it is also clear that l and ρ were derived from 
 their Semitic equivalents lod (■>) and Resli (ί). Besides Tsade the 
 Hebrew alphabet had another dental sibilant, Zain (τ), and the Greeks 
 borrowed this under the name Σάν. It is not known what was the 
 shape or value of this letter as used by the Greeks. It seems to have 
 represented a modified articulation of Σίγμα, for which the Dorians 
 used it as a substitute (Herod, i. 139). Pindar, in speaking of the 
 αΰί,γμ,ου ώδαί of Lasus, says {Fragm. 47), that in these artificially con- 
 structed and longspun Dithyrambs the (jav was falsified (κίβδηλον) ; 
 by which he means merely that the sibilant in general was intention- 
 ally omitted or slurred over•)*. But whatever may have been the dis- 
 tinction between ΰάν and ξητα or ΰίγμα, it is obvious that it very soon 
 fell out of use, and as ζήτα stands in the place occupied by the He- 
 brew Zain., it may be inferred that Zain and Tsade were borrowed 
 at the same time, and placed side by side in the gap occasioned by 
 the loss of Vav or P. As in the Hebrew alphabet lod and Kaph are 
 placed in the interval between the aspirates and the liquids, so in 
 Greek their representatives l and % stand between O' and λ: and ρ 
 stands in the gap left by the omission of Q, which corresponds to the 
 place of Besh in the Hebrew alphabet. It appears to us that ΰίγμα is 
 a pure. Greek word, derived from ύίζω, and expressing the hissing 
 articulation of the sibilant. Although its place in the Greek alphabet 
 after ρ corresponds to that of Shin in the Hebrew alphabet after Besh, 
 this is not to be taken as any evidence of the derivation of ΰιγμα 
 from 123, which is represented in name, form, and, originally, in pro- 
 nunciation by |r, whereas the shape of a and the use of ΰίγμα as the 
 oldest and simplest sibilant should induce us to derive it from Samech. 
 When first imported it was undoubtedly called ΰάμ: but as the Hel- 
 lenic articulation changed the final ^a to v, it became identified with 
 ύάν, from Zain; and while it assumed this name among the Dorians, 
 the lonians substituted the Greek term ΰίγμα. If, however, ΰίγμα 
 was originally (5άμ, or Samech, it must have formed a part of the Cad- 
 mean syllabarium of 16 letters, and therefore, as we have seen, must 
 have stood immediately after v. This place is now occupied by |Γ, 
 the representative of Shin, and conversely ΰίγμα has taken the place 
 of Shin after ρ. The only reason for this interchange, which occurs 
 
 * The reason of this combination appears to be, that a was considered as 
 a sort of aspirate, and therefore like other aspirated letters communicated 
 its rough breathing to the preceding letter: thus we have ΕΠΙζίΕΧΣΟΝΙ 
 for έπΐ^έξωνί, ΈΧΣΑΜΟΥ for k Σάμον, ΓΡΥΦΣ for yQvip, &c. The 
 same is the case with ρ: thus we have ά'Ό'ρακτοί for άτάρκκτοι (Hesycli.) 
 and φροίμιον for προοίμιον, &c. 
 
 t See The Theatre of the Greeks, 6th ed- p. [27]- 
 
CHAP, v.] THE GEEEK ALPHABET. 177 
 
 to us, is the following. It appears that βάμ was called 6^ before it 
 got the name of ύίγμα. Thus, the combinations d-Tt and τί-ΰ are 
 called όαμ-πΐ and Λ-ΰι respectively. Now Shin, with the point on the 
 left corner (ώ), is pronounced ΰΐ; and when this letter was intro- 
 duced into the Greek aljahabet, it is conceivable that ξι, with the 
 original power of ώ, i. e. sh, took the place of 6l, which was trans- 
 ferred, with its new name βίγμα^ to the place of SMuj which, in its 
 second value ia, no less than its first value "ώ, stood in the Hebrew 
 alphabet between Resh and Tau. We have already mentioned that ν 
 and qp were added to the original alphabet, i. e. placed after r, to 
 represent approximately the obsolete digamma, and that χ followed 
 them when the double aspirate was used as a simple h. The three 
 remaining letters ψ, ω, and TTi probably stand in the order of their in- 
 vention, as mere matters of convenience, at a comparatively late 
 period. In fact they are ligatures or contractions rather than let- 
 ters; and the last soon went out of use again except as a numeral 
 sign. 
 
 103 Some interesting deductions may be made from the later or 
 phonetic hieroglyphs of the ancient Egyptians, which must be con- 
 sidered as a form or application of the Semitic alphabet. The pho- 
 netic syllabarium of Champollion and Lepsius may be reduced ulti- 
 mately to the representatives of 15 different articulations. The Coptic 
 language, which is taken as the standard of pronunciation, has, strictly 
 speaking, no medial mutes, b and d for instance being expressed by mp 
 and nt. But it does not follow from this that the ancient Egyptian 
 abandoned the genuine Semitic preference for the medial articulation. 
 As it stands, the hieroglyphic alphabet comprises (a) 3 original mutes ; 
 in Coptic π:, z, r; (b) 3 liquids, λ or ρ, μ^ ν, (c) 3 sibilants and an 
 aspirate, s, ch, sh, h; (d) 2 ultimate breathings or vocalized gutturals, 
 a and i; (e) 3 supplementary labials, v, f, u. It seems probable that 
 a represented the ain as well as the aleph of the Hebrews, and that 
 it sometimes approximated even to he, which however has a repre- 
 sentative of its own. The Hebrew alphabet, in its original form, is a 
 complete expression of the necessary sounds of the language. We see 
 in this hieroglyphic alphabet, on the other hand, an ambiguity as to 
 the original mutes, and also as to the liquids I, r; and while it is 
 almost redundantly supplied with secondary labials and sibilants, it 
 has only two vowels, a and i ; and as the former is expressed by one 
 and the latter by two reeds, it is clear that the derivation oi jod or 
 chirik from aleph, which is so common in Hebrew, is fully recognised 
 in old Egyptian. All this shows us how fortunate the Indo-Germanic 
 races have been in the possession of a complete alphabet, which the 
 
 Ν 
 
178 THE THEOEY OF [bOOK I. 
 
 Egyptians with all their combinations of ideographic, determinative, 
 and phonetic signs, were never able to reahse in a satisfactory 
 manner. 
 
 104 The Arabic syllabarium, on the other hand, is one of th^ 
 most complete and systematic collections of phonetic signs which the 
 ingenuity of man has hitherto collected. Originating in mutilations 
 or corruptions of the Cufic, and ultimately of the Syriac characters, 
 it has received a new and highly artificial arrangement, depending 
 rather on the shapes of the letters than on the organic connexion of 
 the sounds which they represent. And a distinction by means of points 
 placed above or below the character has taken the place of an original 
 difference in the form of the character itself. Thus, b and t, which 
 belong to different organs, are placed side by side, and represented by 
 the same character, the former having a point below, the latter a 
 point above. In the middle of a wordc, n^ t, th, h, and y are discri- 
 minated only by means of the punctuation. The addition of points 
 above the line converts the liquid ^j ?i, into the mutes v:d = ί and 
 ^ii = th, and by similar additions we can aspirate or assibilate ^ =z h 
 
 into ^ z= hhj o = d into ό =dz, \=r into V == 2^ and j = zh, \j*^=^s 
 into Jm = shj [jO = ς into ^ = dh, L• = t into ^z=zz^o = ain into 
 c τ=ι ghain, ^ = w into ο = Λϋ = /, and ^ = hhv = q; whereas an 
 addition of points belo\v the line converts the sonant ^ =^ j into the 
 surd -- = cJi in the Persian use of this alphabet. It is worthy of 
 
 remark, too, that in the Syriac alphabet the cognate r and d are dis- 
 tinguished only by points placed above and below the sign respect- 
 ively; and in an ancient inscription belonging to the Royal Asiatic 
 Society, in which the name of Sapor seems to occur, the mute d is 
 distinguished by a point from the liquid r*. 
 
 105 (2) Analysis of the Greek Alphabet. 
 
 We have now seen in what state the Semitic syllabarium was 
 imported into Greece. Before we proceed to consider how the Greeks 
 modified and adapted to their own language a notation which, though 
 organically perfect in respect to the Semitic articulation, was but a 
 poor instrument for the expression of the language of Homer, and 
 what value they put on the different characters, it will be as well 
 to examine with some minuteness the Sanscrit alphabet, which has 
 
 * For this observation we are indebted to Mr. Edwin Norris, Secretary 
 to the Royal Asiatic Society. 
 
CHAP, v.] THE GREEK ALPHABET. 179 
 
 lost all traces of the Semitic arrangement, and has been made, by an 
 elaborate exercition of Brahminical ingenuity, a most suitable and 
 complete exponent of a language the same in kind with that of ancient 
 Greece*. The Sanscrit alphabet, called by the native Grammarians 
 Deva-ndgari or "the writing of the Gods' city," consists of forty-eight 
 characters, which are arranged according to an admirable system. 
 First of all are placed the simple vowels, then the diphthongs, and 
 the marks representing the final sounds of η and h. After these 
 come the consonants, divided into three classes, mutes, semivowels, 
 
 * There has been some difference of opinion among Sanscrit scholars 
 respecting the Semitic origin of the Indian alphabet since James Prinsep 
 first indicated the striking resemblances between these characters and the 
 ancient Greek. Dr. A. Weber, who has examined the question with some 
 minuteness (Ueber den semitischen Ursprung des indischen Alphabets. In- 
 dische Skizzen, pp. 125 sqq. cf. p. 77), has come to the conclusion that 
 the letters of the oldest Indian incriptions are of Phoenician origin, those 
 of the Bactrians being probably Aramaic (p. 145). 
 
 Mr. Prinsep (Vol. i. p. 434, 5, ed. Thomas) had identified the oldest 
 letters of the Greek alphabet in particular with the corresponding charac- 
 ters of the oldest Sanscrit. He says, "We might almost dare to advance 
 that the oldest Greek (that written like the Phoenician from right to left) 
 was nothing more than Sanscrit turned topsy-turvey." But his editor, 
 Mr. Thomas (ii. p. 42), considers Weber's theory to be "altogether unten- 
 able," and says (p. 43), that "to judge by internal evidence the Pali 
 alphabet of Asoka's day bears every impress of indigenous organization 
 and local maturation under the special needs and requirements of the 
 speech it was designed to convey." And this is also the opinion of Barthe- 
 lemy St. Hilaire , whom he quotes. Mr. Monier Williams (in a letter to 
 "The Times," 31 Dec. 1858) seems to distinguish entirely between the 
 alphabet of the inscriptions and the Deva-nagari. He says: "The spoken 
 Sanscrit which is found on the rock inscriptions of 300 years b. c, and 
 which is the direct source of all the languages now currrent among the 
 Hindoos, was a kind of patois as different from the learned Sanscrit as 
 Italian is from Latin, and this patois had its own written character quite 
 distinct from the present form of the Deva-nagari." And in a note he 
 adds: "The Sanscrit-speaking immigrants must have brought with them 
 their own character, or invented it very early, reserving to it the exclu- 
 sive privilege of expressing their sacred language." 
 
 In our opinion the intercourse of the Semitic nations with Ophir (that 
 is, according to Lassen, the land of AbMra at the mouth of the Indus) 
 about 1000 B. c. is sufficient to account for the acquaintance of the In- 
 dians with the Phoenician character at a very early period; and the sub- 
 sequent changes in the direction of the writing and in the position of 
 the characters is quite analogous to what took place in Greece. With 
 regard to the Deva-nagari we have already suggested that the Brahmins, 
 whose Vaidic period is placed about 1400 b. c, had no writing to serve 
 as a vehicle for their sacred literature, and that they adopted the charac- 
 ters of the cognate tribes already settled in Hindostan. That they inge- 
 niously extended and modified these characters , introducing the uniform 
 frame opening to the left, may be inferred from the elaborate ingenuity 
 of this syllabarium, and the little artifices for the expression of cognate 
 sounds to which we have called attention in the text (§§ 106 — 108). 
 
 N2 
 
180 THE THEOEY OF [bOOK I. 
 
 and sibilants. The mutes are subdivided into five orders, according 
 to the organs by which they are uttered. Besides these divisions 
 the whole alphabet forms two great classes, surds and sonants'^\ 
 "The term 5wrc?," says Wilkins {Grammar^ p. 15), "is applicable 
 to such letters as, in the first effort to form them, admit of no vocal 
 sound: and the term sonant to such letters as are attended by an 
 audible murmuring, as it were, of the voice." The surds are the first 
 two letters of each of the five orders of mutes, one being aspirated 
 and the other unaspirated, together with the sibilants, and the aspirate 
 which is classed with them. All the rest of the consonants, and all 
 the vowels and diphthongs, are sonants. The table which follows gives 
 the Sanscrit letters with the transcription in English which we have 
 adopted in this work. 
 
 VovTELS ; all sonant. 
 Simple vowels ^ a, ^fJJ a; ^ ^, "sT ^; ^ it, "^ u; ^J ri or r, 
 '^J fi or rr; ^J Iri or Ir, ^ Iri or Irr. 
 Diphthongs TJ ^, ^ ai; ^^ o; ^|[ 
 
 Consonants. 
 (1) Mutes. 
 
 Surd. Sonant. 
 
 au. 
 
 Gutturals cR /b, ^ k'h; Tf ^, ^ g'h; ^ ng 
 
 Palatals '^ cA, l^ch'k; '^ j, ^/^/^^y 
 
 Linguals ^ f', "Z'fh; . "S 'd, '^'άΊι; Tf^ 'n. 
 
 Dentals J{ t, T^ fh; ^d, ^ d'h; 7\ n. 
 
 Labials Xf^?, ^p'h; '^h^ }^b'h;J^m. 
 
 (2) Semivowels; all sonant. 
 
 * Sir Graves Haughton distinguishes them also as sharjys and fats 
 {Beng. Gram, p. 151). 
 
CHAP, v.] THE GREEK ALPHABET. 181 
 
 (3). Sibilants; all surd. 
 ^ f, ^ ^h ίΤ 5. 1 ^^• 
 
 Colebrooke in his Grammar gives the value of these letters as follows : 
 a is pronounced as e in her, i in iir, u in sun; α as α in hall; i as in 
 fit; i as ee in feet; u as in pull; ύ as oo in pool: ri as in merrily; 
 rt the same long; Iri as in revelry; Iri the same prolonged; e as e 
 in there; ai as i in fine^ or y in my; 6 as in go; au as ow in thou; 
 Jc as c in cause or A; in kin; kh as in ink-horn or as c'^ in cachexy; 
 g as in gain; g'^h as in log-house; ng as in sm^; cA as in church; 
 cKh the same aspirated; j as in judge ; fh the same aspirated; wz/ as 
 in onyon and as η in singe; t as in ί^7^; fh as in nut-hook or as 
 ie -Λ in White-hall; d as in deaZ; c?A as in red-haired; η as in noble ; 
 ρ as in pew; ph as in hap-hazard; h as in δαΖΖ; έΛ as in abhor; 
 m as in man; y as in ?/ei; r as in rwn; Ζ as in hull; ν as in 
 value; g is a palatal; "the proper sound of this letter is produced," 
 says Wilkins, "by applying the tip of the tongue to the forepart of 
 the palate, and passing the voice, as in pronouncing our 5;" sh is pro- 
 nounced as in shoe^ except in the West of India, where it is nearly 
 equivalent to kh^ and it is a Ungual; s is pronounced as in sin, and is 
 reckoned as a dental; h as in hair ; ksh as cti in fiction. The un- 
 guals 't, 'th^ 'd, 'dh, 'n, are sunds peculiar to the Indian articulation. 
 "This series of consonants," says Wilkins, "is pronounced by turning 
 and applying the tip of the tongue far back against the palate ; which, 
 producing a hollow sound as if proceeding from the head, it is distin- 
 guished by the term murddhanya, which Mr. Halhed, in his elegant 
 grammar of the Bengal language, has translated cerebral." 
 
 The marks ♦ (w) and * (h) are called anu-svara, "after- sound," 
 or "after- vowel," and vi-sarga, "leaving out," ecthlipsis; the former 
 is an abbreviation of nasal consonants at the end of a syllable, the 
 latter a substitute for the letters s or r at the end of a word. The 
 process by which the diphthongs are formed from the simple vowels is 
 of the most extensive application in Sanscrit, and is also of importance 
 in Greek. "When a short a is placed before either of the vowels i, w, 
 we have the diphthongs e = at and ο = aw; this change is called guna 
 
 * We adopt this transcription because ^ so frequently corresponds to a 
 Greek κ, and we are desirous of pointing out that it is but a softened 
 guttural. If what we have said in this chapter with regard to the real 
 value of ζ in Greek and Zend were as well known as we could wish it 
 to be, we should not hesitate to adopt ζ instead of ρ as a representative 
 of the first Sanscrit sibilant. 
 
182 THE THEORY OF [bOOK I. 
 
 or "strengthening;" when, again, the a is placed before the diph- 
 thongs e,d^ we have the other diphthongs at = «at, and aw=aazi; this 
 change is called vriddhi or "increment." It must be remarked that 
 a is also a guna of a, and the other vowels, r, Ir become ar, al by 
 gutia, and dr, dl by vriddhi. "The vowel Iri is only found in the 
 verb klrip and its derivatives. We remember no instances of the 
 long Iri in any Sanscrit word: Bhattoji, in commenting on Panini, 
 1. 1, 9, says that it is not used. Siddhdntakaumudi, fol. 1 6." (Rosen, 
 Journal of Education^ viii. p. 340, note). 
 
 106 An examination of the figures, which compose this wonder- 
 fully systematic alphabet, will lead to very important conclusions with 
 regard to the subject now before us. It is by this means alone that 
 we can ascend from the very artificial order in which the Hindu gram- 
 marians have arranged it to its primitive state, and to the order of its 
 formation. In the first place it must be remarked, that, like the 
 Semitic alphabet, it was originally a syllabarium ; in other words, it 
 had no vowels, and was written from the right to the left. A proof of 
 the latter is afforded by the fact, pointed out byLepsius {Paldographie, 
 p. 10), that with few exceptions all Sanscrit letters have a frame which 
 opens towards the left; and of those too which have no frame, the 
 vowels i, t, u, u, the guttaral ng, and the lingual d' are all turned 
 towards the left ; so that the ancient order of writing must have been 
 in that direction. The Greeks, and other nations who have borrowed 
 the Semitic alphabet, turned round the letters when they altered the 
 direction of their writing, whereas the Indians have left unaltered 
 those letters which were invented or introduced before they changed 
 their manner of writing. The only letters which were turned towards 
 the right were the diphthongs e, ai^ three of the Unguals 't, 'th, 'dh, 
 the aspiration Λ, the semivowel r, and the suffixes, which form the 
 vowels r, rr from the sign generally used to denote a, and the vowels 
 Ir, Irr, from the sign for I. This is at least prima facie evidence 
 for the conclusion that these last characters are of more recent intro- 
 duction. It may seem strange that the lingual 'd should be older than 
 the dental d\ while the other letters of the dental class are older than 
 
 the rest of the Unguals. Perhaps 5 ^^^ ^ have interchanged their 
 pronunciation like the Hebrew t3 and n. It seems probable that 
 5 ^ and 5• were the corresponding characters in the dental class 
 for rf, dh, and n, and that rT was originally the η of the guttural 
 class : 3^ differs from ^ only by a kind of anusvdra mark, and ψ\ 
 is only a reversed Xf. 
 
CHAP, v.] 
 
 THE GREEK ALPHABET. 
 
 183 
 
 107 After what has been said on the origin of alphabets in 
 general, no one will suppose that the vowels were from the first dis- 
 tinguished from the consonants in the Deva-ndgart alphabet. Palaeo- 
 graphy enables us to point out their origin. The charaters which we 
 have given for the vowels are used only at the beginnings of words; 
 for their expression in the middle of words a number of marks 
 analogous to the Hebrew points are substituted, as follows : 
 
 The short original vowel, with which every consonant is articulated, 
 is not written when it follows a consonant, because in the primitive syl- 
 labarium that vowel was always presumed in the first instance, and the 
 vowel-marks were subsequently invented to point out that a different 
 vowel-sound was intended in the particular case. The bar which desig- 
 nates the long a is merely a fulcrum to show that the voice must dwell 
 on the syllable. It cannot be a representative of the vowel a, for it is 
 
 written in cases where no a is implied, as in |, ο = au. The figure 
 which appears in the complete signs of a, a, r, rr, 6 and au, is obvi- 
 ously a mark of the gentle breathing with which all initial vowels were 
 pronounced, and probably corresponded as nearly as possible to the 
 Hebrew Aleph. From this it appears that r , f r , were originally indis- 
 tinct vowel-sounds of r preceded by a breathing. Bopp considers r as 
 a mutilation of ar ( Vocalismus, p. 157), andLepsius thinks that the con- 
 sonant r was formed from this vowel (Palaographie, p. 49). In Zend 
 there is no character answering to the Sanscrit r, but in corresponding 
 words the Sanscrit r is represented by ere (Burnouf, Yagna, p. 1). It 
 matters very little whether we place the vowel before or after a liquid, 
 and we constantly find the same root with the vowel sometimes before 
 the liquid and sometimes behind it. It is remarkable, that, in our own 
 alphabet for instance, while we articulate the consonants by means of 
 vowels placed after them, we always utter the necessary vowel before 
 the liquids and s : thus we say be, ce, &c. but em, en, &c. To adduce 
 another very familiar instance, the sign of our plural is -ne as well as 
 -en, -se (written -ce) as well as -es: thus we have bretNr-en, child'r-en^ 
 ox-en, but ki-ne, swi-ne; and churcli-es, speech-es^ but dice, pence. The 
 letter r is altogether a secondary one. In the articulations of some 
 provinces among ourselves, it is still unpronounceable, and in almost 
 
184 THE THEOBY OF [bOOK I. 
 
 every language it stands for some earlier letter ; most generally for Z, 
 as in the Sanscrit surya^ comp. sol, ήλιος-, Goth, sauil ; Sanscrit pur^ 
 comp. TtXeog, plenus, full; and even in modern languages, as in the 
 words apotre from apostolus, epitre from epistola, titre from titulus, &c. ; 
 it stands for s, as in the Laconian ul6q for 9"£0J, and in the words Vale- 
 rius, FuriuSj arbor, labor, vapor, clamor and lares, which Quintilian 
 tells us (i. 4, § 13) were originally written Valesius, Fusius, arbos, labos^ 
 vapos, clamos and lases. After a guttural, especially in Latin, r ap- 
 pears as the representative of an original sh (see below, § 160). When 
 it appears in the old Italian languages, as a substitute for d ( Varro- 
 nianus, pp. 82, 255), e. g. in ar-cesso by the side of accerso for ad-ced- 
 so=^aGcedere sino {Varron. p. 352), it must be regarded as having 
 passed through the intermediate articulation th. Palaeography shows 
 that r is connected both with d and its assibilation z. In the Hebrew 
 and Syriac alphabets d and r are represented by slightly different 
 forms of the same character (ηΐ, ?j), and in Arabic there is a similar 
 correspondence between Dal and Dzal (^t>), and between Ba and 
 Za (jO• For other changes see Lepsius (Abhandlungen, p. 12) and 
 Grimm (Deutsche Gramm. i. p. 581). It is a proof of the more recent 
 state of the Zend in the monuments which have come down to us, 
 that the r has entirely superseded the I in that language (Burnouf, 
 Υαςηα, p. Ixxviii) ; and in the Behistun Inscription r and s are re- 
 presented by the same character turned in different directions, for 
 ^ J is r, but y^ is s. The nasal liquids m and η are in fact modifi- 
 cations of the medials b and d, to which they sometimes revert; a 
 person who has a cold in his head, or a country actor trying to be im- 
 pressive, will always pronounce his m's as ^'s and his ti's as d^s. We 
 should therefore expect that the m Avould be derived from the δ, 
 and the η from the c?, if the principle of association held. This ap- 
 pears to have been the case in Sanscrit, as will be seen by comparing 
 •1 b?i with ^ 771, and 5 d with 5*• I^ seems that I was a vowel 
 in Sanscrit before it became a consonant; how this could be may be 
 inferred from the use of the I mouille in French. It was in fact the 
 first form of the r, or rather they were both produced from a sound 
 between the two (like* the Chinese ewZ), \vhich was the more like a 
 vowel the older the language was (Lepsius, AbJiandl. pp. 9, 10). We 
 have mentioned above that the ancient Egyptians had only one sound 
 for Λ and ρ. 
 
 108 The first thing which strikes us in the Deva-ndgari conso- 
 nants is the contrast which they present to those of the Semitic alpha- 
 bet. Omitting the palatals and Unguals, the former of which arc 
 
CHAP, v.] THE GEEEK ALPHABET. 185 
 
 immediately derived from the gutturals, and the latter peculiar to the 
 Indian organs of speech, we have the same three sets of fundamental 
 mutes as in the Hebrew alphabet. We remark, however, this striking 
 difference: in the Deva-ndgari alphabet the tenues, which are most 
 saited to the pronunciation of those who speak the Indo-Germanic 
 idioms, are placed first, the medials last. Besides, the Hindu gramma- 
 rians have begun with those letters which are pronounced in the back 
 part of the mouth, namely the throat, and have gone on through the 
 others in order, ending with those letters pronounced by the lips. This 
 is of course very good as a technical arrangement ; it seems, however, 
 that the order of creation is that given by the Hebrew alphabet, namely 
 labials first, then gutturals, and lastly dentals. Again, it is observable 
 that there are two orders of aspirates in this alphabet, sonants as well 
 as surds, whereas the Greek and Hebrew have the latter only. At first, 
 probably, the Hebrew aspirates were, as we have shown, modifications 
 of the medials, but, if we are to place any reliance on the assertions of 
 modern Hebraists, they all approximate to the tenues, and one of them, 
 the Teth^ has actually become a dental tenuis. It is very certain that the 
 Greek aspirates were ultimately modifications of the tenues and not of 
 the medials ; nevertheless, in words of the same origin, the Sanscrit hh, 
 dh, are represented by the Greek and Latin φ, / and &: thus the Sanscrit 
 root bhr corresponds to the Greek and Latin φέρω, fero, and madhu to 
 μΒ%ν: ph seldom occurs, and tJi is consistently represented byr. This 
 indicates in our opinion a change in the value of the Oeva-ndgart cha- 
 racters, analogous to that which we have hinted in the case of d and 
 d\ η and ng ; for there can be no doubt, that, if the tenues were really 
 anterior to the medials in this alphabet, their aspirates would be first 
 employed: besides, we find in these secondary aspirates a continual 
 shifting and interchange as well in form as in usage, which shows that 
 they had not a distinct independent existence from the first; ^ which 
 is in form only a modification of Xf, is often represented by the same 
 
 letter in cognate languages, thus dadhdmi = τίΟ'ημι, gharma = ^'kQμη; 
 \I and V| are also interchanged ; thus dhr and bhr both signify "to 
 carry *." If we revert to general principles we can hardJy doubt that 
 the aspirate must have been prefixed in the first instance to the arti- 
 culations which it modified. Thus, if we compare the first and second 
 lines of the original syllabarium (above, § 100), we shall conclude that 
 as Ν differs from ϋ by the h prefixed to the latter, so 1 must have 
 
 * This may perhaps be explained ty the fact that in the Prakrit there 
 is a tendency to substitute the aspirate alone for an aspirated letter, as 
 gaJiira for gambhtra, "deep," sahci for sabhd, "an assembly" (Wilson, 
 Hind. Dram. p. 70). 
 
186 THE THEOEY OP [bOOK I. 
 
 been originally hv or hh, t\ must have been %, and U hd. We shall 
 see that in the Greek βαν the aspirate always preceded the labial ; 
 and there is no reason why 0" should not have been originally hd^ just 
 as ρ, the cognate dental liquid, was undoubtedly hr as well as rh. In 
 fact, as we shall see, the metathesis of aspirates and sibilants is com- 
 mon in all languages. The aspirates of the guttural class are very 
 seldom used, but, when they are, kh, not gh, corresponds to the Greek 
 χ, as in κόγχη compared with gankha; οννχ-ξ {οννχοξ) compared with 
 nakha^ &c. "We think that th must have been originally an assibila- 
 tion rather than an aspirate of t. It will be shown in another place 
 that the Greek ζ or assibilated d is a representative of a sound re- 
 sembling sh or the French ch, produced by combining a guttural or a 
 dental with y. Such a sound is the Sanscrit p, for this is almost 
 always a representative, under a softer form, of the Greek κ, and of 
 the Lithuanian sz, which is pronounced in the same way as the San- 
 scrit letter. It is the tendency of all languages to soften or assibilate 
 their hard sounds. We have plenty of instances of this even in the 
 modern languages of Europe; in French it is particularly common; 
 thus, from camera we have chambre, from audere, oser, from canis, 
 chien, &c.; in England we have in the North, where the older Saxon 
 is rife, kirk, wick (Alnwick), brigg, dyke, &c., which in the South are 
 softened to church, wich (Greenwich, Brom-which-ham), bridge, ditch, 
 &c. In the ancient languages the same thing is observable: thus the 
 older forms preserved in da^qv, lacryma; ϊκκος, ΐτίτίος, equus; δέκα, 
 decern-, &c. are softened into the Sanscrit agru, agva, daga, and the 
 Lithuanian aszara, aszwa, deszimtis, &c. This change of the hard 
 pronunciation of c has taken place without a corresponding change of 
 form in the modern Italian, and is regulated by the appearance of the 
 vowels e ov i after it. The same is the case with the g in English, 
 Italian, and French. A good instance of the change in the pronuncia- 
 tion of a dental caused by the addition of i or y is furnished by our 
 way of pronouncing such words as nation, revolution. Different lan- 
 guages have various methods of expressing the sound sh, as resulting 
 from an aspiration or assibilation of the gutturals and dentals : and 
 sometimes the same language has several symbols for it. The San- 
 scrit, for instance, has a direct representation, or rather, two distinct 
 signs for it ; in other alphabets it is represented by z, j, y, or, in the 
 case of the dental, by %'. The symbol j often degenerates into the 
 simple vowel i, just as the symbol v, which represents the labial aspi- 
 rate, degenerates into the vowel u: in fact, this is the way in which 
 these vowels are formed, and in this case it may be said, that all that 
 part ^f the softened consonant, which bore any relation to the origi- 
 nal consonant, is lost; a phenomenon which often presents itself in 
 
CHAP, v.] THE GREEK ALPHABET. 187 
 
 language, and wliich is also an explanation of the change of aspirated 
 consonants into h, and of their interchange with one another. We 
 may take this opportunity of correcting a theory which has been 
 brought forward by two of the most distinguished philologers of the 
 present day, and which, though highly ingenious, appears to us to rest 
 upon a false principle. Grimm {Deutsche Gramm. i. p. 187) justly re- 
 marks that j : i = V : u, and that the row of labials p, b, /, v, u is 
 parallel to the row of gutturals, k, g, ch, j, i; but then he supposes 
 that each of these series is derived from the vowel which forms the 
 basis of it, whereas we are convinced that no mute was ever derived 
 from a vowel, still less the original mutes ρ and L• He asks (1) why 
 the dentals do not also rest upon some vowel as a basis? and (2) how 
 we are to reconcile with the above parallelism, the obvious analogy of 
 the row of dentals, t, d, th, s, to the row of gutturals, k, g^ cJi, h, and 
 to that of the labials p, h, f, ν ; in other words, why do the gutturals 
 possess a double support, j and Λ? Bnr noni (Yagna, p. cxiv, note 
 46) has attempted to remove the former difficulty by establishing from 
 the Zend language an analogy between a and Λ, so that the dentals 
 are derived fom a through Λ, according to the following table: 
 
 k — a 
 
 t—a 
 
 ρ — a 
 
 g—a 
 
 d—a 
 
 b—a 
 
 %—^ 
 
 ^—a 
 
 φ — a 
 
 y—a 
 
 z—a 
 Ji-a 
 
 V — a 
 
 i 
 
 a 
 
 u 
 
 To derive h from a is somewhat difficult; it is certain, on the con- 
 trary, that the character for a is derived in all alphabets from that 
 denoting a weak breathing not amounting to h. The vowels i and u 
 are formed from consonants, but it is in vain to attempt the establish- 
 ment of a parallelism throughout the three orders of mutes in refer- 
 ence to the formation of the three fundamental vowels a, ^■, u. The 
 vowel a is presumed in the existence of every letter, and, as we have 
 shown, assumes the two lighter forms of e, ο in heavier combinations. 
 The vowels i, u are of totally different origin ; they are derived from 
 the ultimate vocalization of weakened consonants. The latter results 
 from the vocalization of ϋ, the weakest form of the labial ; the former 
 is the common offspring of the dentals and gutturals, the softened 
 dental (= ty) and the softened guttural {=^ gy) being equally represented 
 by ζ or j, and equivalent to sh. The dental aspirate tJi passes at once 
 into 5; conversely we meet with people every day whose articulation 
 does not admit of their pronouncing a final s otherwise- than as th. 
 This view of the connexion between th, s, sh, y, j is borne out in a 
 
188 THE THEORY OF [bOOK I. 
 
 remarkable manner by the corresponding characters of the Deva-nd- 
 gari alphabet. That ^, the sign for j, was originally the same as 1\^ 
 ij, is clear from VJJ^, the older sign for jJi : ^, th, differs from ^, p, 
 only by the tail, which seems to be the distinguishing mark of the s : 
 this tail is clearly seen in "^, and that 'CJ, sh, originally had it, may 
 be inferred from ^, k-sli. The same confusion, which we have before 
 pointed out in the gutturals. Unguals, and dentals, seems to have con- 
 verted into a simple d of the dental class the figure ^, which appears 
 from its tail to have been the assibilated d of the lingual class ; in fact, 
 the lower part of 1^, which includes sh^ the sibilant of the Ungual 
 class, is merely this same H. We have already pointed out the simi- 
 larity of ^ and ^; there is an equally striking resemblance between 
 '^, δ, and of, v, which are similarly connected. The vowels i and u 
 which are derived from j and v, are designated in the Deva-ndgari by 
 derived symbols. The initials ^, % and ^, m, are composed of the 
 tail a. and the hook SD joined to the bar at the top by an unmeaning 
 line of connexion (see Lepsius, Paldographie, p. 16). The latter is 
 the essential part of of reversed, and the former is that tail which 
 distinguishes the sibilants and '^5. "With regard to Grimm's second 
 
 question, it is to be observed that not only from the analogy of all 
 languages, but also from a consideration of the form in Sanscrit (for 
 
 ^ has no frame and is turned towards the right), the aspirate must 
 
 be considered of subsequent formation. In Greek it resulted from the 
 digamma, from the gutturals, and from s. In Zend it is generally 
 derivable from the sibilant. In German it mostly comes from guttu- 
 rals, and we may consider it as more particularly attached to that 
 class of consonants. As all sounds are of two primitive kinds, 
 breathings or consonants, from a combination of which the aspirated 
 consonants are produced, we may consider the aspirate as the final 
 state of an aspirated consonant. In fact, consonants may be reduced 
 to four ultimate states without becoming vowels ; a dental or a gut- 
 tural may become j (y) or s (Λ), a labial may become ν ; vocalization is 
 effected in the former case by converting s into h, and then omitting 
 the aspirate, or by turning j into i; in the latter, by simjDle conversion 
 of V into M. When a dental or guttural is reduced to j, it may 
 always become i; when to s, it may always become h by visarga; 
 
CHAP, v.] THE GREEK ALPHABET. 189 
 
 when a labial is reduced to v, it may always become u ; and when a 
 consonant is composed of s and v, it may become indifferently either 
 h or j (y) from the one element, or u from the other. 
 
 109 We are now prepared to discuss the various difficult points 
 connected with the Greek alphabet, and to estimate the real value of 
 those characters about which so much has been said. It will be found 
 that in this as in other questions people have fruitlessly perplexed 
 themselves with details, when a proper consideration of the principles 
 would have disentangled all the confusion, and left no real ground for 
 doubt or uncertainty. The pronunciation of the unaspirated mutes 
 and liquids may fairly be presumed to be the same as that which all 
 nations have adopted for those letters ; for there is no reason whatever 
 to suppose the contrary. The only characters which we have to con- 
 sider are those representing, either in their earliest or in their subse- 
 quent state, breathings, or aspirated, or assibilated consonants. These 
 are a, ε, ξ;. Ρ, η, «θ', ί,, |, ν, φ, χ, ω. Of the first we have already 
 spoken : it is simply the Aleph N, or A-kdrah ^, stript of its breath- 
 ing, and is therefore the fundamental vowel with which every con- 
 sonant in the old syllabarium was articulated. The Greek a being 
 pronounced with a considerable opening of the mouth, and with an 
 approximation to the same curvature of the tongue which accompa- 
 nied the articulation of d, Ό", λ, ρ, and ν, it is generally found to take 
 or keep its place after the first four of these letters, and it habitually 
 replaces the last in those cases Avhere ν becomes evanescent. Similarly 
 u^ which is so often inserted before I in old French, is the only repre- 
 sentative of that liquid in the modern spelling; thus alter y through 
 aultre^ passes into autre {Varron. p. 259). That ε and ν were origi- 
 nally pronounced with an aspiration is clear from their names g jpiloVi 
 V i\)lX6v. The former is, as we have seen, derived from the Hebrew 
 He Ϊ1, the sign of the aspirate, which, as the hardest breathing, is 
 articulated with the lightest form of the fundamental vowel: this 
 aspirate being omitted, the vowel becomes ε ψίλόν, or the He without 
 aspiration. We must consider υ 'ψίλόν, which always retains its 
 original aspirate at the beginning of a word, in connexion with J=, the 
 most troublesome letter of the old Greek alphabet. Indeed, a full dis- 
 cussion of this obsolete character will exhaust nearly all that remains 
 to be said respecting the Greek alphabet. 
 
 110 It has been shown, that, in name and form, p, the vau or 
 digamma, corresponds to the Hebrew Vav , as Η does to Cheth and 
 Θ to Teth. It als appears that these three Hebrew characters were 
 originally the aspirated medials, though subsequently they approached. 
 
190 THE THEORY OF [bOOK I. 
 
 nearer to the tenues. The Hellenic organs of speech were, from the 
 first, more favourable to the tenues, and it is clear that their aspirated 
 mutes, with the exception of Q' (above, § 100), not only ultimately 
 belonged to that order, but must be regarded as tenues with a dis- 
 tinct aspiration added (below, § 111). For the rough breathing, whether 
 purely guttural or amalgamated with some other sound, the Greek 
 ear, as distinguished from the Pelasgian, had no toleration. As the 
 language advanced from its oldest to its most classical written state, it 
 lost most of those aspirations which it originally possessed, and gene- 
 rally substituted simple for compound articulations. If we compare 
 the Latin alphabet with the Greek, we shall see some remarkable 
 proofs of the truth of this observation. In the Latin we have not 
 only F with its original complex sound (§ 111), and Η with its strong 
 guttural articulation (§ 112), but J and V appear with consonantal 
 values, as a palatal and labio-dental respectively, and Q always re- 
 tains its semi-labial power, or demands the juxtaposition of u *. In 
 Greek, on the other hand, F and Q appear only in certain ancient in- 
 scriptions f , the mark of aspiration is converted into the symbol of a 
 vowel, and there is hardly a single case in which I and Τ have re- 
 tained or assumed a consonantal value. It is clear that f must have 
 been originally the aspirate of the labials, namely bh or hb] but it 
 assumed a different value, fell out of use, degenerated into a breath- 
 ing, or was vocalized into u, and therefore φ was introduced as the 
 proper aspirate of the labial tenuis. 
 
 With regard to Q or κότίΛα it is to be remarked that when this 
 letter appears in Greek inscriptions it is always followed by o. For 
 example, we have λυροδορκας (Bockh, C. I. No. 166), which shows 
 that TCOJtTta and κάτΐτΐα were distinguished as compound and simple 
 articulations, the former, with its accompanying o, being a residuum 
 of a syllable pronounced Kwa, so that it was equivalent to the Latin Qv. 
 The labial in the Latin lupus, compared with the Greek λύκος, shows, 
 in accordance with a principle which will be discussed below (§ 121), 
 that these two forms must have sprung from one, which, like λνοος, 
 contains the elements of both articulations. The Latin form qui- 
 squil-ice, shows that there must have been a κότΐτΐα represented by 
 the combinations κο and κν respectively, in the first two syllables 
 of κοΰκνλ•λω, κο-όκυλ-μάτια. And thus the δίγαμμα and κόπτία ap- 
 pear to be only varieties of the same complex sound, the labial, as 
 
 * See Varroniamis, pp. 248 sqq. 
 f The inscriptions in which κόππα appears are the following: Bockh, 
 C. I. Nos. 29, 37, 166. Those in which J= is found are enumerated in 
 
 the Appendix (A) to the present chapter. 
 
CHAP, v.] THE GBEEK ALPHABET. 191 
 
 a general rule, predominating in the former, and the guttural in the 
 latter. 
 
 It is not our intention to tire the reader's patience with an enumera- 
 tion of the various opinions which have been entertained respecting the 
 digamma: the scholars who have written about it have for the most 
 part been unable to avail themselves of the resources of comparative 
 philology, by the aid of which alone we can hope to solve the problem ; 
 and therefore our knowledge of the subject has advanced but little 
 since the point was first mooted by Bentley*. It will be far better in 
 this place to state plainly and at once what are the results at which 
 a comparison with the cognate languages has enabled us to arrive, 
 results consistent not only with themselves, but also with all that the 
 ancient writers have told us in regard to the power and functions of 
 this letter, and with the phenomena which it presents in the Greek 
 language. » 
 
 In all the languages of the Indo-Germanic family we find in some of 
 the most common and important words, as well at the beginning as in 
 the middle, traces more or less distinct of a letter compounded of the 
 two consonants, one of which represents the guttural, the other the 
 labial, in its ultimate state. These two consonants are s (sometimes 
 reduced to Λ), and ν ; and from their combination, and the different 
 changes which they separately and together admit of, arises that great 
 variety of letters which are etymologically traced to an original iden- 
 tity. In by far the greater number of cases this sv or hv represents 
 
 * This great scholar was the first to discover the traces of a lost di- 
 gamma in the frequent recurrence of an hiatus in Homer. It has been 
 made a matter of complaint by continental scholars that so little is known 
 of Bentley's opinion about the digamma. A copy of the Poetce Greedy 
 fol. 1566, with his marginal notes, was lent to Heyne by Trinity College, 
 Cambridge, but not the copy-book in which he had written his views on 
 the subject more at length. This manuscript is not, as Thiersch calls it 
 {G/-. Gr. § cLxii.), "a full and elaborate treatise, in which he goes through 
 the digammated words in alphabetical order, and overthrows all apparent 
 objections to his doctrine:'' it is merely a set of rough notes, in which 
 the words supposed to have had the digamma are enumerated, the pas- 
 sages in which they appear copied out, and, in some cases, the necessary 
 emendations are suggested. But there is a total absence of order or 
 arrangement, and it is not fit for publication. We have given, in the 
 Appendix (B) to this Chapter, all of it that appears to be of any value 
 or interest, as well to show how little could be done for the doctrine of 
 the Grςek alphabet without the aid of comparative philology, as to afford 
 another proof how far Bentley was in advance of his age in this as in 
 other points. On Bentley's Homeric studies the reader may now refer to 
 Dr. Wordsworth's note on his Correspondence^ p. 820. A specimen of the 
 vague and unsatisfactory manner in which modern scholars have spoken 
 of the digamma, may be seen in Hermann's Opiiscula, i. p. 131, where he 
 treats this letter as the single representative of three distinct sounds. 
 
192 THE THEOEY OF [bOOK I. 
 
 an union of the original guttural and labial kp ; in those cases where 
 a dental makes its appearence, it must be considered as having arisen, 
 by a fault of articulation, from the sibilant. The regular series of 
 transitions, which such a combination of the guttural and labial would 
 present, may easily be described; the guttural may be represented by 
 k^ q^ g^ y, 5, li\ the labial by p, b, v; and these sets of letters may be 
 permuted with one another to any extent. Then , either the one or 
 the other may be dropt, and the remaining one vocalized into i (ij) or 
 u, according as the one retained is the guttural or labial. This pro- 
 cess will be best shown by numerous examples. The root of the 
 reflexive or relative pronoun (which we shall show to be the same in 
 a future chapter) is properly kpa or kva in all the Indo-Germanic 
 languages. This appears as pwy in Breton, as qvis, qvi, svus (suus) in 
 Latin, as sva in Sanscrit, as (5ςρε* in Greek, as hvas in Gothic, as 
 quhay in Old Scotch, «ind, by a transposition of the letters, as who in 
 modern English. By an omission of the labial element, this becomes 
 sa, softened into ya or hardened into ka in Sanscrit ; og, xog, ε, Γ, 
 in Greek; se in Latin; in English who (when pronounced hoo); and in 
 French que (pronounced ke), &c. By an omission of the guttural ele- 
 ment it becomes in Greek πίς, tcov, &c., in German iver, in English 
 ivhat (pronouncd wat), &c. Similar changes are the following ; San- 
 scrit ςνά (ςυαη), Zend ςραη, old Persian gpaka •)*, Russian sabac, Greek 
 Τίνων, Latin, with a loss of the labial, cams; French, with a softening 
 of the guttural, chien; German, with a substitution of the aspirate 
 (the labial being only retained in the vowel «) hund: Latin eqvus, 
 Zend agpa^ Lithuanian aszwa; ^olic Greek, by assimilation to the 
 guttural, ϊκκοζ] common Greek, by assimilation to the labial, iTtTtog; 
 by omission of the labial, Scandinavian oek; by softening the guttu- 
 ral, Erse each : Sanscrit svadu, Latin suavis (the dental being omitted 
 as helium from duellum); by omission of the labial and substitution 
 
 * This word alone might have sufficed to teach our Greek scholars 
 that the digamma was occasionally something more than a mere labial 
 breathing. In such lines as Πηλείδης δε αάηος άπό pf'o (hvfo or αφέο) 
 χειρί ηαχείτ} {Iliad, χχ. 260) it is clear that the digamma represents a 
 double consonant. It is, however, generally true that in the Hellenismus, 
 with which alone we are acquainted, the digamma very seldom makes its 
 appearance as a double letter, yet the guttural element is as often retained 
 as the labial, as we may see, among other things, in the number of words 
 originally digammated which are written Avith a y in Hellenic. To look for 
 the digamma in its full form we must go back to the old Pelasgian language. 
 
 f Herod, i. no: οννομα δ\ rfj γννεακϊ ην rfj ΰννοίκεε Κννώ, κατά 
 την ^Ελλήνων γλώοΰαν %ατά δε την Μηδί•Η7]ν, Σπακώ' την γάρ κννα 
 ^οίλέονβι απά-κα Μηδοι. Justin, ι. 4: Nutrici Spaco postea nomen fuit, 
 quia canem Persce sic vocant. Gramm. apud Hermann, de Emend, gr. Gr. rat. 
 p. 434: απάξ, 6 ηνων Tcaqa Πέροαις. 
 
CHAP, v.] THE GREEK ALPHABET. 193 
 
 of the aspirate for s, ηδύς : Sanscrit gvagura, Gothic swaihra, Russian 
 svekor, German schwager^Welsh cliwegrwn\ by omission of the labial 
 socer^ by substitution of the aspirate εκυρό^: Sanscrit svapna; by aspi- 
 ration of the s and vocalization of the v, vTtvog; by omission of the 
 labial somnus : Sanscrit svid^ Gothic hweits, English sweat; by vocal- 
 ization of the labial sudor; by aspiration and vocalization νδωρ] by 
 aspiration and omission of the labial ίδρώξΐ Sanscrit svana-s, "a tone," 
 Latin sonus for svonus, Greek φωνή for ΰφωνή,&ο. In comparing the 
 Romance languages with the Teutonic, we find the labial w repre- 
 sented either by a combination of the guttural and vocalized labial 
 guj or by the guttural g only: thus from wer, war, we have guerra, 
 guerre; from ward^ guardire, guarder; from warrant, guarantir; 
 from Wilhelm, William, Guillaume; from wehr-wolf , loup-garou= 
 lupus- g ar-ulpJius ; from weise, guise; from (Jqp?J|, vespa, wespe^guepe, 
 &c. "The French writing," says Grimm (Gesch. d. deutsch. Spr. 
 p. 296), "still retains GU before E, I, as in guerre, guise, but lets it 
 pass over into a simple G before A, as in garder, gant. So Galli 
 seems to be placed for Gualli, 0. H. G. Walah. In Welsh I find the 
 pi. Gwalwys 'the Gauls.' What was the sing.? The Irish substitute 
 F for the Welsh GW, as βοη, gwin, vinum; fear, gwyrdd, viridis, 
 &c." It is also remarked that Neriosengh, who translated into San- 
 scrit the Pehlvi version of the Yagna, represents the Zend ν by the 
 Sanscrit ghv or gv; thus for vdhumano, Mvani, gdvangh he writes, 
 gJivahmana, Mguana, gdguamgha (Burnouf apud Lepsium, Abhand- 
 lungen, p. 100, Note). Similarly the old Persian name Hystaspes, 
 which is probably the Greek representative of the sound Hvistaspes, 
 appears as Gushtasp in the more modern Persian historians, but as 
 Vistagpa in the Zend books (Burnouf, Yagna, p. cvi), and Vashfdspa 
 in the Behistun inscription. Rawlinson has clearly established the fact 
 
 that the cuneiform letter ^YY , u, had an inherent power of aspiration 
 
 (Journal of the B. As, Soc. x. 2, pp. 69 sqq.). Indeed, an examination 
 of his alphabet will enable us to remark, that all the letters into 
 
 which the elemental sign / enters, are more or less affected by aspira- 
 tion ; and we should be inclined to infer that whenever / Υ Υ appears 
 
 alone it is a real combination of a strong guttural with the vocalized 
 labial, u; but that when it is combined with other letters the guttural 
 alone is retained. Thus, — as Rawlinson says (p. 76) that there are nu- 
 merous cases in which / VY replaces the Sanscrit W, su, and / YY-<y^ 
 is the substitute for ^1^, sva,— the former ought to be expressed by 
 ^hu and the latter by 'hva. For this reason, and because the Greek 
 transcriptions must be allowed to furnish a certain amount of evidence, 
 
 Ο 
 
194 THE THEOEY OF [BOOK I. 
 
 we should write '■Hvakshatara, not ' Uwakshatara, and ^Hvaspa, not 
 ' Uwaspa, as the Behistun forms of Κναξάρης and XoaoTtfjS- The fol- 
 lowing are instances of the improper substitution of dentals for com- 
 binations of this sort: the Russian svera, Lettish svehrs, old Prussian 
 svirs becomes fera in Latin by the omission of the guttural element, 
 unless, as Mr. Winning supposes {Manual, p. 83), the Latin /had a 
 pronunciation approximating to this compound: in the Greek ^^q, 
 the guttural sibilant has passed into the dental, by a false articulation 
 such as we often hear in English, and similar to that by which ύά- 
 λαΰΰα (from αλξ, sal) has become ^άλαΰΰα, &c.: this becomes the 
 tenuis-dental in the German thier (pronounced teer)^ and the medial- 
 dental in the English deer, Anglo-Saxon deor: other instances of the 
 same corruption are furnished by rig, τε &c., compared with κε, Ttov, 
 og, sa^ qvis, &c., and Teouagsg compared with jtioVQeg, chatur, qvatvor^ 
 &c. This is also the right way of accounting for such double forms 
 as ψλαν^ d'Xav; φλίβω, Ο'λίβω; φλιά, Ο'λίά; which Thiersch (Gr. Gr. 
 § CLn. 9) supposes to have arisen from the union of φ and O*, on the 
 analogy of φΟ^ηρ ; this form, however , has no real existence , but is 
 only assumed by the author of the Etymologicum Magnum (p. 451, 
 13) to be the original form of ^ηρ, as deduced from φ^Έίρω, accord- 
 ing to the second derivation which he proposes. It appears, then, that 
 where there has been at the first a combination of two letters, one 
 of the guttural , the other of the labial class, this compound is repre- 
 sented in the cognate languages by any letter which can serve as a 
 substitute for either of the elements, or by a combination of any two 
 of such substitutes. Now , those words, which are said to be affected 
 with the digamma, present continual variations from the guttural to 
 the labial and vice versa, and also in very many cases exhibit a com- 
 bination of letters belonging to both orders. We may therefore infer 
 conversely, that this digamma was originally a compound of the gut- 
 tural and labial, presenting, however, one at least of the elements in 
 its ultimate state, or in the nearest approach to vocalization. Compa- 
 rative philology has shown us how such a combination may stand in 
 one language for a simple labial or guttural in another, generally for 
 a labial; and the examination of the Greek alphabet, in its relation to 
 the Semitic, has enabled us to perceive that the Greek digamma 
 occupied the same place as the aspirated labial 1 in the Hebrew sylla- 
 barium. We must, then, conclude that the articulation of the Pelas- 
 gians necessitated a substitution of this compound sound for the He- 
 brew vau, just as the Sanscrit translator Neriosengh substituted gv 
 for the Zend v. The Persian or High German element, which pre- 
 vailed over the Sclavonian in the Greek language, expelled this hard 
 sound and the sibilants from the beginnings of words, and substituted 
 
CHAP, v.] THE GREEK ALPHABET. 195 
 
 a simple aspirate, just as we see in the Zend language as compared with 
 the Sanscrit , and in this way the real digamma vanished from the 
 pronunciation of the Greek language, and was consequently omitted 
 in the writing. The letters which most generally represent a lost 
 digamma in Greek, are β, ςρ, and ν from the labial class, the sibilant, 
 aspirate, and γ from the gutturals, and the combination 6φ. If ί ever 
 stood in its place it must be as a substitute for y, as in the Prussian 
 pronunciation o£ yott for Gott, &c.; and if we are to consider βάλλω^ 
 Βάκχος as identical with Ιάλλω, Ίάκχος (Thiersch, § clii. 9), the only 
 way of explaining the transition is, by supposing that these words 
 originally began with a digamma, or a sound composed of guttural 
 and labial, and that β is the representative of the labial, t of the gut- 
 tural element. There are traces of such a compound sound ίη'Ίακχος, 
 as we see from the fact, that the cognate words ίάχω, Ιαχή, are always 
 digammated when they appear in Homer, And with regard to Ιάλλω 
 likewise, we see traces of the labial, which in βάλλω appears by itself, 
 in the form φιάλλω (see Aristoph. Vesp. 1348; Pax, 434; Hesych. 
 εφιαλλεν. ετΐεχείρηόεν, ηφίαλεν); compare also the German qual, 
 qualm. In ζάλτι the guttural is represented by J, as is often the case. 
 The vowel ο occasionally stands for the labial element of the digamma, 
 just as ο has the force of w in our " one," or the French oiseau. Thus 
 the name of the town Oaxus, in Crete, is spelt Fd^og on the old coins 
 (Mionnet, Vol. n. p. 263); in Scylax, if the reading is correct, the 
 name is written Ila^og; and we find Fav^imv in an inscription from 
 the neighbourhood of Teos (Bockh, Corpus Inscript. ii. p. 638). The 
 name is said to point to the ruggedness of the place, and to be de- 
 rived from ^άγνυμί (Steph. Byz. v. "Οα^,Οξ — τυνες δε δια το κατα- 
 γηναί τον τότΐον καΐ κρημνώδη ντίάρχειν καλονΰυ γαρ τους τοιού- 
 τους τόπους αξους, κα^'άπερ κα\ ημείς αγμούς). So also Ίη'Όαύις, 
 οΐότρος, and other words, the ο must have represented the sound of 
 w {Varronianus , pp. 49, 251). A question might be raised, whether 
 the initial ο in the words οίκος, οίνος, was itself a representative of 
 the labial element of the digamma, or whether that letter was prefixed 
 to the whole word as it stands. By comparing ^οΐκος with the San- 
 scrit vegan, Latin vtcus, and Ι^οΐνος with the Latin vinum, we might 
 infer that the ο is merely a guna of the root- vowel in each word ; and 
 a comparison of οΐδα with the Latin vidi, Sanscrit veda (root vid)-, 
 2nd pers. sing, olod'a, Sanskrit vettha, Zend vaesta (Burnouf, Υαςηα^ 
 p. 451, note 315) ; 1st pers. plur. ϊδμεν^ Sanscrit vidmas, might lead us 
 to conclude that the diphthong in the singular is a guna of the root, 
 and that the initial v, which belongs to the root, is absorbed (see 
 Pott, Etymol. Forsch. i. pp. 246 foil.). On the other hand, the conso- 
 nantal value of the initial ο is established by the cases mentioned above; 
 
 2 
 
196 THE THEORY OF [bOOK I. 
 
 and we find in Hesychius the glosses youvog and γοΐδα. If, then, we 
 compare these forms with the Welsh gwyn and gwydd, which has the 
 full compound initial, and with the Armenian gini and the Anglo- 
 Saxon witan, which have lost respectively the labial or guttural ele- 
 ment of the prefix, we may fairly conclude that the ο in οίνοξ and 
 olda is really a residuum of the digamma. When the dental sibilant 
 %' appears as a substitute for the digamma, it results from false arti- 
 culation, as we have shown above. 
 
 Ill The Latin language, in which the Sclavonian element was 
 not, as in the Greek, overpowered by the introduction of High German 
 ingredients, retained the digamma sound to the last, though the Greek 
 figure was not introduced until the reign of Claudius, and then again 
 disused (Quintil. i. 7, § 26; xii. 10, § 29; comp. Tacitus, Annal. xi. 
 14). This digamma sound is attributed by the ancient writers, not to 
 the letter F, which so nearly resembles the Greek digamma in form (the 
 only difference in the digamma introduced by Claudius being , that it 
 was inverted), but to the letter u or «?, when used as a consonant, in 
 which case it seems to have had the sound of our Wy as appears from 
 the following passages. Dionysius of Halicarnassus (i. c. xx. p. 52 
 Beiske): καΐ didoaucv αντοΐς χωρία — τα ΛερΙ την ίεραν λίμνην-, iv 
 olg ην τα τΐολλα ελώδη, α ννν, κατά τον άρχαΐον της διαλέκτον 
 τρότΐον, Ο ν -ε λ La ονομάζεται. (5ννη^εζ γαρ ην τοις άρχαίοΐζ"Ελ- 
 λη6ι , ώ^ τα Λολλά , Λροτι^εναι των ονομάτων , οτίόβων αί άρχαΐ 
 ατά) φωνηέντων εγίνοντο^ την ον όνλλαβην ενί 6τοιχείω γραφο- 
 μενην. τοντο δ^ ην ώΰπερ γάμμα διτταΐς επΙ μίαν ορ&ην εταζενγνύ- 
 μενον ταϊς τίλαγίαυς, ως Γελένη,καΙ Pά^'α|, καΐ Γοΐκος και Γανήρ, καϊ 
 τίολλα τοιαύτα. Julian (Orat χι. ρ. 7 1 , edit. Spanheim) : "Ενετοί δε οΐμαυ 
 το 7ίρ6(5%εν ώνομάξοντο, ννν δε ηδη, 'Ρωμαίων τάς τιόλεΐξ εχόντων^ 
 το μεν εξαρχής όνομα ΰώζονόι βραχεία Λροβ&ήκη γράμματος εν 
 άρχη της επωνυμίας, εύτι δε αυτόν βύμβολον χαρακτηρ εΙς. ονομά- 
 ξουΰι δε αυτόν ΟΤ, καϊ χρώνται αντί του ΒΗΤΑ, τίολλάκις τίροβ- 
 πνεύΰεως οΐμαί ttrog ένεκα καϊ ιδιότητος γλώττης. Priscian (ρ. 560): 
 Hahehat autem hcec F litera hunc sonum quern nunc hahet U Iogo conso- 
 nantis iiosita. Quintilian (i. 4, § 7): Desintne aliqum nobis necessarian 
 liter ce — utinJiis, Seruus et Uulgus, ^olicum digamma desideratur. 
 (xii. 10, § 29): ^OLiCiE quoque literce qua seruum, CERUUM-^-we di- 
 cimus, etiamsi forma a nobis repudiata est^ vis tamen nos ipsa perse- 
 quitur. From these passages we see that the later writers considered 
 the digamma only in its labial element. It appears, however, from Quin- 
 tilian, that there was another Latin letter, F, which represented the di- 
 gamma in form and name , which occupied the same place in the Roman 
 alphabet , and which corresponded to the older power of the Greek di- 
 
CHAP, v.] THE OREEK ALPHABET. 197 
 
 gamma, in combining, as it would seem, a sibilant with the labial. Quin- 
 tilian says (I. 4, § 14), that fordeum and fxdus were written instead of 
 hordeum and hoedus in old Latin: quin fordeum, F(EOi]sque {dicebant), 
 pro aspriratione vav velut simili liter a utentes: nam contra Grceci as- 
 pirare solent (Φ), ut pro Fundanio Cicero testem, quiprimam ejus li- 
 ter am dicer e non posset, irridet. Again he says (xii. 10, § 27, 29), that 
 it was a great disadvantage to the Romans to have only Υ and F instead 
 of the Greek Tand Φ, quihus nullce apud eos (Grcecos) dulcius spirarit, 
 — Nam et ilia, quw est sexta nostrarum, pcene non Tiumana voce, vel 
 omnino non voce potius, inter discrimina dentium efflanda est: quce 
 eiiam cum vocalem proxima accessit quassa quodammodo,utique quoties 
 aliquam consonantem frangit, ut in hoc ipso frangit, multo fit hor- 
 ridior. From these passages Mr. Winning has inferred, as we have 
 already mentioned, that the Latin / corresponded to the Lithuanian 
 or Sclavonian double consonant sv or zw. He is also inclined to sup- 
 pose that two distinct sounds were represented by the Latin /; one 
 corresponded to the Greek ςρ, and was used in words connected with 
 the Greek, such as fero, fama, &c. : the other, this double sound, ap- 
 pears only in words to which the Greeks have no corresponding forms, 
 as in fariolus, fasena , &c. , in which the / was subsequently changed 
 to h {Manual, p. 266). We are willing to adopt the former opinion, 
 that / was, at least originally, a double sound, consisting of a sibilant 
 and a labial: from the manner in which Quintilian speaks of it, 
 there must have been something very peculiar about the Latin /, 
 and the description which he gives of its pronunciation would of it- 
 self lead us to suppose a sort of sibilation. But we are sure that it 
 never approximated to the Greek φ; Quintilian most expressly tells 
 us so, and the fact is clear from the following phenomena which are 
 applicable to words connected with the Greek, as well as to the other 
 words in which / appears. That the Greek φ is a distinct ph, as 
 in shepherd, and that it is not equivalent in sound to the Sanscrit 
 hh, which, as well as the Latin /, etymologically corresponds to it 
 (compare the Greek and Latin roots φν, /μ, with the Sanscrit equi- 
 valent hhu)y appears from reduplications like τίεφυκα, and from con- 
 tacts like Σατίφώ. That the Latin ν and / are cognate sounds , and 
 have nothing in common with the Latin h^ p^ or with the Greek |3, 
 ίΤ, ςρ, is clear from the fact that ν and / may be immediately pre- 
 ceded by n; which is always changed into m before the other letters: 
 compare conferre, conviva, with combibere^ componere, imprimis, 
 εμβάλλειν, ψπας, αμφί (Pott, Etymol. Forsch. i. p. 79). In our 
 opinion, the letters / and ν in Latin corresponded, in the time of 
 Quintilian, to hv and w respectively; at all events, the former was an 
 originally compound sound made up of a guttural (whether under the 
 
198 THE THEOEY OF [bOOK I. 
 
 form of a sibilant or an aspirate) and a labial; for we hold that 
 no labial can pass directly into a guttural, but that, in every in- 
 stance where this has apparently occurred, the labial has been 
 originally connected with a guttural: thus fircus must have been 
 Jivirquus or svirquus (Sanscrit vrikas, Zend vehrkas) before it could 
 become Urcus or liirpus; and we think that the change of a per- 
 manent /, at the beginning of Latin words, into Ti in the Romance 
 languageSj is a proof, that, to the last, / contained some guttural 
 element; thus the Norman hawk stands for falco; hors, in French, 
 represents the Latin /oris, which is fuori in modern Italian (com- 
 pare guerra and guerre)] and the Spanish hacer , hambre, hado, 
 Jdgo^ Jiijo^ hilo, hermoso, humo, hoja, hongo, huesa Qivesa), hurto, 
 are the modern substitutes for the Latin facere^ fames, fatum, ficus, 
 filius, filum, formosus, fumus, folium, fungus, fossa, furtum. Grimm, 
 we are aware, thinks that there is a real connexion between the 
 labial υ and the sibilant and aspirate s, h. "I infer," says he 
 {Deutsche Gramm. I. p. 581), "the intimate connexion of the three 
 breathings v, h, s, partly from the change of ei into e (p. 91), of 
 au into 6 (p. 94), which takes place before them, partly from the 
 interchange of h and v, w (pp. 148, 403), h and s (pp. 318, 416), 
 and the contact of the aspiration with the assibilation (ίΛ, ts, z)\ 
 between v, w, and s^ there is no immediate interchange; h and v, 
 the softest of all consonants, occasionally fall off without substitu- 
 tion, even when they stand at the beginning, and especially before 
 liquids." It is strange that this great philologer should not have 
 perceived that the arguments which he has drawn from the voca- 
 lization of the German languages, rather prove the original union 
 of a guttural and labial in v, Λ, s , than the identity of the gut- 
 tural and labial elements: the change of ei into e=ee before v, h, s, 
 is an absorption of i, the guttural element of the diphthong, into 
 the following guttural element h, s, or into the guttural element 
 combined with and presumed in the υ; and it is for this reason 
 that the guttural part of the w is often dropt in this case, the 
 change of the diphthong remaining as a trace of the guttural part 
 of the compound sound: conversely, the change of au into ό=οο 
 before the same letters, shows a corresponding loss of the labial 
 in h, s, the guttural part being alone retained. Grimm admits that 
 there is no immediate interchange between the v{w) and 5, and 
 that V and h (wliich is only a further reduction of s) sometimes fall 
 off without leaving any traces : this is a proof that a compound like 
 hO=^8v might become either Λ, 5, or v, not that these letters are 
 themselves identical. The fact is, as we have before mentioned, that 
 Grimm has not a proper notion of the origin and value of the vowels 
 
CHAP, v.] THE GEEEK ALPHABET. 199 
 
 e, Uj which, we have shown, are the vocalization of ^' and v, the 
 ultimate forms of the mutes, the former being the point of conver- 
 gence to the gutturals and dentals, the latter to the labials. The name 
 V φίλόν, and the fact that this letter never appears at the beginning of 
 a word without a rough breathing, show how natural the union of the 
 aspirate and labial are to some articulations ; this vowel is to the pre- 
 sent day pronounced like υ by the Greeks even before consonants; 
 and the analogy of the Sanscrit language, in which i and u are always 
 pronounced as the consonants y, ν before vowels, even when they 
 form part of the diphthongs e, 6, as well as the testimony of the old 
 Grammarians, assures us that the Greek υ was v, or rather hv, before 
 it was w, and that i, was y or j before it was L 
 
 112 We may now leave the digamma, or rather its labial ele- 
 ment, and direct our attention to the sibilants, the aspirate, the gut- 
 tural χ, and the vowel l considered as the ultimate state of the gut- 
 turals and dentals. The gutturals are softened as follow — k into ch, 
 sh, s; g into j^ sh: the dentals thus — t into th, sh,s; d into sli. These 
 changes may also be produced by subjoining to each guttural and 
 dental the letter y{i)^ and pronouncing the combined letters as one: 
 the sound which results in each case is sh. As, then, this softening 
 might have been effected by the addition of i, if that vowel had existed, 
 conversely the vowel ^, not having existed previously, naturally sprung 
 from this change of the aspirates of the dental and guttural orders, 
 the aspirated dental or guttural element being omitted; just as the 
 vowel u arose from a similar softening of the aspirate of the labial 
 order, the aspirated labial element being omitted. The Greek ζ remains 
 as a representative of the guttural or dental combined with i{jj), as 
 will hereafter be shown*. In the mean time it is sufficient to state 
 that it appears, from the forms of the comparativef and of the 
 verb conjugation^, and from other phenomena in Greek§, that ζ was 
 equivalent to %y ^ yy ^ ry, dy : we know from the analogy of modern 
 languages that these sounds may be equivalent to sh, and it is clear 
 
 * The theory, which we started in this work twenty years ago, re- 
 specting the influence of i on a preceding consonant, has been made the 
 subject of a special investigation by A. Schleicher {zur vergleichenden Spra- 
 chengeschichte , Bonn, 1848). He gives this phenomenon the name Zeta- 
 cismus, with the following explanation (p. 39): "I derived this name from 
 the most familiar example of the amalgamation of two consonants of which 
 the second is j. In the narrower sense then it denotes the complete amal- 
 gamation of j with the preceding consonant (as in the Greek), in the 
 wider sense, it signifies any influence of a contiguous j (or of the vowels 
 i, e), on the preceding consonant." 
 
 t Book II. chap. 2 J Book iv. chap. 4. § Book in. chap. 1. 
 
200 THE THEORY OF [bOOK I. 
 
 from the words of Dionysius of Halicarnassus (de Compos, c. 14, p. 81, 
 Reiske) that ζ must have been sometimes pronounced like sh=dy=^gy. 
 He says : τρνών δε των άλλων γραμμάτων j a δη δίτΐλά καλείται, το 
 ξ μάλλον ηδύνει την άκοην των ετέρων, το μεν γαρ ξ, δια τον κ, 
 το δε ψ, δια τον Λ, τον ύνριγμον άττοδίδωΰί, ι^ιλών όντων αμφο- 
 τέρων, τοντο δ' ηύνχη τω πνενματυ δαύννεται^ καΐ εβτι των ομο- 
 γενών γενναιότατον. In the Doric dialect*, ζ is written (jd, which is 
 merely a result of Hellenic euphony; for the place of ζ in the alphabet 
 shows that it must have been considered originally as a modification 
 of δ; and as | represents both %(3 and ^κ, so there is reason to think 
 that <3δ is a transposition of an original d^f, which corresponds to the 
 Hebrew Tsade, and also indicates a sound nearly connected with at, γι 
 or ^=j. Conversely, in the East- Anglian dialect of our own language, cZ^ 
 becomes dge=j, as is shown by the Suffolk proper names Ether edge, 
 Cocksedge, &c., which represent the original genitives Ether eds, Cocks- 
 heads, &c. The vowel c is placed next to θ, the simple aspirate of τ 
 or δ, because it is the last result of the softening of those letters. In 
 some cases a Greek -O" represents a Sanscrit y, as in χ&ές^ hyas. 
 
 When the Greek ξ stands as a representative of the guttural in 
 its softened state (thus, ζάω corresponds to the Sanscrit /ivawi, Sclavon. 
 schivu, Lithuanian gyws=vwus, gywenu=vwo'^, and ξέ^α to the 
 Sanscrit Java, Lith. jawai^)^ it is equivalent to the Zend z^ which was 
 nearly our j , as appears from the circumstance that Neriosengh uses 
 the Sanscrit j as a transcription of this letter: thus he writes Ahura- 
 majdasya, "of Ormuzd," and Jjisnijanda, "the Zend, or book called 
 Ύαςηα,^'' Majdaiasnyah, "the Mazdaiasnas" (Burnouf, Υαςηα, p. xv). 
 In related words this Zend letter stands for the Greek κ, γ^ οτ χ^ or 
 for a Latin and Sanscrit h when that letter represents a strong gut- 
 tural breathing: the following instances are given by Burnouf ( Yapwa, 
 p. Ixxxi): 
 
 * So say the Grammarians {Scholl. Dion. Thr. 780, 16; Etym. M. 412, 
 &c.) •, but Ahrens {de dialect. Dorica, p. 95) considers this as an occasional 
 imitation of the Lesbian dialect by the lyric and bucolic poets, 
 
 * A curious metathesis of tv into pt has been noticed in the Buddhist 
 inscription of Girnar. See Lassen, Indische Alierthumskunde, ii, p. 227, note 4. 
 
 J Bopp has not seen the true analysis of this root, although he has 
 furnished the materials for it (Vergl. Gramm. p. 128). A comparison of 
 the Gothic quivs=vivuSy with the O. N. quikr, O. S. quic, Engl, quick, and 
 the perfect vixi (Schneider, Elementarlehre, i. p. 243), shows that the root 
 originally involved a reduplication of the digamma or qv: so that the 
 Latin vivus stands for qviqvus {Varronian. p. 250). 
 
 § According to Ptolemy Java was "the barley island;" vii. 2: Ίαβκ- 
 δίον ο ΰημαίνει κριΘ^ης νήοος. See W. Humboldt, {'iber die Kawi-Sprache, 
 I. pp. 60 — 63; A. Humboldt, Kosmos, i. p. 559 note. 
 
CHAP, v.] 
 
 THE GEEEK ALPHABET. 
 
 201 
 
 Sanscrit. 
 
 
 
 Greek. 
 
 Zend. 
 
 Lith. 
 
 mih 
 
 Latin, 
 
 mingo 
 
 ομίχω 
 
 miz 
 
 myzu 
 
 hima 
 
 
 hiems 
 
 χύμα 
 
 zydo 
 
 ziema 
 
 hansa 
 
 ί Gothic, 
 \ Latin, 
 
 gans 
 
 %Ψ 
 
 
 zasis 
 
 
 anser 
 
 
 
 
 h-Mmi 
 
 ί Latin, 
 I Gothic, 
 
 humus \ 
 gaui 1 
 
 χαμαί 
 
 zem 
 
 zeme 
 
 
 
 
 
 hrdaya 
 
 ί Latin, 
 I Gothic, 
 
 cor 
 
 καρδία 
 
 
 szirdi 
 
 
 hairto 
 
 
 
 
 dham 
 
 ί Latin, 
 I Gothic, 
 
 ego 
 
 'εγώ 
 
 azem 
 
 isz 
 
 
 ik 
 
 
 
 
 Jiasta 
 
 ) Latin, 
 
 hand 
 
 hir 
 
 pre-hend-ere 
 
 %εφ 
 
 zasta 
 
 
 mahat 
 
 I Gothic, 
 
 magnus | 
 ' mikels / 
 
 μίγαξ 
 
 maz 
 
 
 
 
 
 
 hari 
 
 Latin 
 
 viridis 
 
 ωχρός 
 
 zairi 
 
 
 113 The reader will be careful to distinguish from this appear- 
 ance of the h in Sanscrit and Latin those cases in which the initial 
 aspiration appears in Greek as the representative of a sibilant in those 
 languages. We have already shown, from the form of the character 
 in the Deva-ndgari alphabet, that h is not an original letter. It has, 
 however, two sounds. According to one it is a hard guttural breath- 
 ing, and appears as the representative either of an original guttural 
 consonant or of the digamma. In this use it may stand either at the 
 beginning or in the middle of a word in Latin, Sanscrit, or Low 
 German , and it frequently resumes its original form in certain com- 
 binations. It is this h which corresponds to the Zend z=^j. To the 
 instances given above we may add the following: Sanscrit root vah, 
 Latin veh-it^ vec-sit, vectus^ vac-ca, JΓ6χ-og, ^οχενειν , ανχ-ήν, Zend 
 vaz-aitij Sclavonian vezeti, Lithuanian vezu, vessti; Latin trah-ere, 
 trac-si, German trag-en^ Lithuanian traukti; Latin hostis from f ο stis 
 (=svostis or hvostis)^ German gosts; Gothic haupit, Latin caput, κεφ- 
 αλή; Latin homin (homo)^ Gothic guma, old High German komo, Li- 
 thuanian zmones, old Prussian smunents; Latin hortus, Gothic gards, 
 old High German karti, χόρχοξ; Gothic taihun, δέκα, Sanskrit daga, 
 &c. From this it is clear that the strong h is the immediate represen- 
 tative of the digamma or of a primitive guttural; that it is harder 
 and more original than J=2; (Zend) appears from the use of the latter 
 in reduplications from roots beginning with the hard Λ, as in jahami, 
 juhava, &c. ; compare chakdra, bibharmi, &c. The other sound of h is 
 merely a weakening of the sibilant, similar to the visarga at the end 
 
202 
 
 THE THEOBY OF 
 
 [book I. 
 
 of a word, and generally makes its appearance as an initial. This 
 breathing, as a substitute for 5, is consistently found in those lan- 
 guages of the Indo-Germanic family, which we infer from other reasons 
 to be more recent than those of the same family in which the corre- 
 sponding words begin with s: its appearance may, therefore, be used 
 as an argument to prove that languages , of which we do not know 
 the age from other sources, are recent in comparison with those which 
 present the initial s. We have h for s in Greek as compared with Latin 
 and Sanscrit, in "Welsh as compared with Erse, and in Zend as com- 
 pared with Sanscrit : now we have reason to believe, independently of 
 this, that Greek (in one element at least) and Welsh are younger than 
 Latin or Sanscrit and Erse respectively: we therefore conclude that 
 Zend is younger than Sanscrit, or belongs, at least in one element, to 
 the High Iranian class*. The following are a few instances out of a 
 great number which might be collected. 
 
 More ancient languages with s. More recent languages with h. 
 
 
 Latin ( 
 
 or 
 
 '' 
 
 
 
 ^ 
 
 Sanscrit. 
 
 Gothi( 
 
 c. 
 
 Erse. 
 
 Zend. 
 
 Greek. 
 
 V^elsh. 
 
 sd 
 
 st 
 
 
 si 
 
 hd 
 
 h 
 
 hi 
 
 saptan 
 
 ί sibun 
 I septem 
 
 / 
 
 secht 
 
 hapta 
 
 ετΐτά 
 
 
 
 
 
 
 
 isam 
 \ sakrt) 
 
 simul 
 
 
 
 jham 1 
 Xhaker^etf 
 
 αμα 
 
 evo 
 
 
 
 
 
 
 
 sal 
 
 
 salan 
 
 
 
 halen 
 
 svapnas 
 
 ί somnus \ 
 
 suan 
 
 
 vTtvog 
 
 hyn 
 
 
 \ slips 
 
 / 
 
 
 
 
 
 svar 
 
 sol 
 
 
 saul \ hvare 
 ^ sornean f 
 
 ivhog) 
 
 i heol \ 
 \ hinon ί 
 
 
 
 
 
 santi 
 
 ί sunt \ 
 \sint 1 
 
 
 
 henti 
 
 
 
 
 
 
 
 
 
 βαταβυαΗ 
 
 
 
 
 Ilaraqaiti (i. ( 
 Arachosia) 
 
 3. 
 
 
 Sarayu 
 
 
 
 
 Harohu (i. e. 
 Herat) 
 
 
 
 Sindu 
 
 
 
 
 Hindu 
 
 
 
 soma 
 
 
 
 
 homa 
 
 
 
 Asura 
 
 
 
 
 Ahura 
 
 
 
 * Schleicher remarks {zur vergl. Sprachengesch. p. 34): "the Zend, 
 which is so rich in vowels and changes the dental sibilant into a breathing, 
 exhibits a strong resemblance to the Greek in this and other particulars, 
 whereas the more ancient Latin shows rather the habitus of the old In- 
 dian." It is perhaps unnecessary to remind the reader that this is another 
 proof of the affinities of the ancient Greeks and Persians (above, § 90). 
 
CHAP, v.] THE GEEEK ALPHABET. 203 
 
 114 In general it is to be remarked that the letters ν and ς, 
 which are of more frequent occurrence than any other consonants at 
 the end of words, are peculiarly liable to become evanescent not only 
 at the end of a word, but also at the beginning or end of any syllable. 
 This is an extension of the principle of anusvdra and msarga, and 
 some of the most remarkable phenomena of Greek etymology are ex- 
 plicable only with reference to this tendency. The evanescence of 6^ 
 or its subsidence into an aspirate at the commencement of a word, 
 has been sufficiently exemplified above. It may be considered as al- 
 most a rule of the Greek language when a vowel follows. Thus, the 
 s of silva is always represented by the aspirate of νλΓη, except in the 
 old compound Σ'κα7ίτη-6νλΎΐ-=^'^αΙά-0βνοάβ , the name of a place in 
 Thrace (cf. Sila in Bruttium). Σνζ is found by the side of vg^ and ^ 
 sometimes stands for 6 by the side of the aspirate, as in ^άλα(56α for 
 ΰάλαύΰα by the side of aXg; Ο'έλω, from the root ελ-; &c. We have 
 Sarpedon, "the rohher''=aQ7C7jdov (Welcker, Kret.KoL p. 9 ; Ruckert, 
 Troja's Ur sprung, p. 34). The Greek Έρυνννς, 'EQlveog have been 
 compared with the Sanscrit Saranyu, Sdranyava (Kuhn, Zeitschr. f. 
 vergl. Sprachf. I. pp. 439 sqq.). Similarly, the Οτββ^ί'Ερμης/ΕρμΕίας, 
 and the Tuscan Turms, are equally connected with the Sanskrit Sa- 
 rameya (Dr.Trithen, Proc. Phil. Sog. m. No. 71, pp.201 sqq.): compare 
 ημέρα, ΰήμερον, τημερον ; ετίτά, τετίτά; ερμίς, τερμίς, &c. ( Varronianus, 
 p. 150). The medial d appears also as an occasional representative of 
 the sibilant; compare δείλη with εϊλη, όέλ-ας, ηλυος, &c. The abso- 
 lute and total evanescence of g and ν is most conspicuous in the cases 
 in which they were originally flanked on both sides by short vowels. 
 They are then changed into the rough or nasal breathing respectively, 
 and thus they pass off from the want of characters to express them. 
 Compare rvTtry, τντίτεαι, with τί^'εΰαα, and μείξους, μείζoεg, with 
 μείξovεg. The simple a, which the Hebrew grammarians appropriately 
 call patha^h or " opening," seems to have been peculiarly fitted to serve 
 as the vehicle of the residuary nasal breathing. Thus when ν is fol• 
 lowed by r, O", or stands by itself at the end of a word, it is very often 
 represented by a*. Hence we have τετνψαταυ for τετνφνται^ όω- 
 ξοίατο for ΰώξοίντο, Λατερα for ττάτερν, or τζάτρεν, &c. The plural 
 εχόμε^Όί by the side of the dual εχόμεΟΌν, ^ol. εχόμε%'εν, is 
 an instance of the frequency of this interchange, and some roots 
 are regularly affected by it ; compare βένΟΌς with βά^^og ; τίέν^^οξ 
 with TtadOg, πείβομαι, on the analogy oi χαρίεις from χαρίεντς, for 
 
 ^ On the evanescence of ν in particular we have collected a numher of 
 illustrations in a special paper. Trans, of the Philological Society for 1854, 
 pp. 282—295. 
 
204 THE THEORY OF [bOOK I. 
 
 πεν^'-ΰομαί ; τίίτίον^α as the perfect of τΐάΰχω for τίένΟ'ύκω ; γίγονα 
 by the side of γέγαα\ the roots μα- and μεν-^ &c. A similar prin- 
 ciple, namely the affinity of the aspirate and sibilant, will explain 
 a remakable phenomenon in regard to the vanishing letter, ς. 
 Although the Greeks particularly delighted in the combination ^Ό'-, 
 which is partly shown by the purely Hellenic change of the more an- 
 cient J=d(j into (5δ] and though there are many structures, as we 
 shall see in the course of the following pages , which are really due to 
 this predilection, yet we find on the other hand some cases in which 
 the (J of the combination (JO* becomes absolutely evanescent when it is 
 preceded by an aspirate ; thus, we have τετνφ-ΟΈ for τέτνφΰΟ'ε, λέ- 
 λεχ-Ο'αί for λελέχ-ΰ&αί, &c. If the combination ντ is left at the'end of 
 a word it regularly becomes a. All neuter plurals are examples of 
 this, and we shall see below that δέκα must have been originally 
 δ^έ-κεντ. The loss of the ν is very common in the separative particle 
 ανά or av-, signifying remoteness in space or time, and used with 
 nouns as a negative prefix, and with verbs as a temporal augment. In 
 the former case ava or va always becomes a- in Attic Greek when 
 prefixed to a word beginning with a consonant. When it precedes a 
 word beginning with a vowel the ν is retained, and the form is av - 
 or άνη-. In Homeric Greek the form νη- appears whether the 
 word begins with a vowel or a consonant. In the other case re- 
 ferred to — ^that of the temporal augment — ava is invariably sof- 
 tened into £-. There is a curious analogy in this case to the Hebrew 
 "•iii, 'hantf which is always curtailed into iji, 'he, when it appears as 
 the personal prefix of the continuous tense: and in the same way "όίΝ, 
 ^henesh, "a man," is usually shortened into ώ'^Ν ^hish, and the wis 
 omitted in ηίί 'haph, from η^Ν 'heneph, just as in αω by the side of 
 ανεμοξ and animus. 
 
 115 The letter ξ, which is generally a direct union of the gut- 
 tural κ ov γ with the sibilant, does not require much notice. In San- 
 crit it is often represented by the softer form sJi^ just as κ is almost 
 consistently represented by ς. The explanation of this is not difficult. 
 We shall show immediately that, when an aspirate or sibilant is com- 
 bined with a consonant, it very often changes its place, that is, it may 
 stand either before or after the consonant. Thus the root of the rela- 
 tive pronoun, which is sv in Sanscrit and Greek, is qv in Latin, hv in 
 Gothic, but wh in English; so Q=hr becomes rh in Latin; and we 
 have already seen that ξ was both δΰ and όδ. Similarly ξ was both 
 κξ and (?κ, just as the sound, which is hs in Gothic, became sh in 
 Sanskrit, so that ξ occasionally corresponded in power to the Semitic 
 Shin, the name of which its common denomination ξι seems to repre- 
 
CHAP, v.] THE GEEEK ALPHABET. 205 
 
 sent. Thus ξίφος was written ΰκίφος (Hesych. s. v.), and perhaps even 
 κίφος (Pausan. m. 26, 9); and in the same way the Eastern subjects of 
 Alexander seem to have transposed the elements of the | in his name, 
 if we may judge from the forms of Αλέξανδρος and Άλεξανδρία, e. g. 
 Iscander, Scanderon, Candahar*. The same has been the case with ip 
 (Bekk. Anecd, p. 815), and the pronominal forms ψε, ψ^ν, -yse, must 
 be considered as transpositions of 0φε. It has been mentioned above, 
 that |t appears to have come in as a substitute for the Doric 6άν^ the 
 older representative of Shin. The Latin χ seems frequently to have lost 
 the ^-sound altogether, as in Ulyxes compared with ^Οδ\)β6Βνς^ rixa 
 compared with ε-ρίδ-ς^ ε-ρίξω^ &c. In the Doric dialext | appears for 6 
 in the fut. and 1 aor. of verbs in -ζω (Ahrens, dial. Dor. pp. 89 sqq.). 
 It can hardly be doubted that it is here, like 66-, a representative of the 
 the sound sh resulting from (?t. In the name βερξτις it seems that | re- 
 presents both ks and also sk softened into sh, for the Persian original 
 was Khshaydrsha (see below, §§ 160, 470). 
 
 116 We must now show how H, the old mark of the aspiration in 
 Greek, came to be used as a sign for the long e , the first half of it being 
 alone retained to denote the rough breathing. It is an etymological 
 principle of great importance , that an aspirated short vowel is equiva- 
 lent to an unaspirated long one. To this may be added another principle 
 of equal importance and fundamentally the same, that an initial di- 
 gamma or aspirated labial, represented by the secondary vowels i or u 
 may be placed behind the initial vowel so as to form a diphthong, or 
 even be transferred to the second syllable of the word. We have a 
 simple exemplification of the first principle in the word ωμός "a shoul- 
 der," which is clearly another form of ομος from the pronominal com- 
 pound ο-μ- (ομον, &c.), expressing the equilibrium of the shoulders: so 
 that ώμος=^ομος is related to humerus, as νόμος is to numerus. Of both 
 principles we have the following instances in the Greek language itself, 
 and from one root — ελκ in έλκω, which begins properly with the di- 
 gamma, and of which we shall speak more at length in a subsequent 
 chapter. From this root we have ηλεκτρον for ελκτρον, ηλακάτα and 
 ηλακάτη for ελκτά and ελκτή, and ώλαξ for ολ| according to the first 
 principle, and ανλαξ for the same word, according to the secondf . We 
 recognise the same etymological fact in the comparison of ae-quus with 
 ί-κα-νός, which has nearly the same meaning. In compounds we see 
 
 * In the inscription of A^oka at Kapur di Giri (b. c. 253), the name 
 of Alexander seems to be represented by Alikasunari, in which the ele- 
 ments of the I appear in their proper order. 
 
 t See Buttmann, Mythol. ii. pp. 355 foil., which we have translated in 
 our notes on the Antigone of Sophocles, pp. 213—219. 
 
206 THE THEORY OF [bOOK I. 
 
 that ae, originally «i, is equivalent to Γ. Thus from ces-timo we have 
 ex-istimo, from ce-quus, in-iquus, &c. (see Varronianus, p. 262). Con- 
 sequently ae-qua-nus=i-ka-nus=l -xcc-vog. So that we ought not to 
 connect ixavog with Γκω, ίκάνω as Passow does in his Lexicon, but with 
 the pronominal compound hi-c found in the affix -L-Kog, &c. : and thus 
 ίκα-νός corresponds in origin as well as meaning to idoneus=ideo-neus 
 (cf. Buttmann, Lexil. i. 46; Doderlein, Etym. u. Syn. iii. 276). The 
 second principle is very frequently applied in transitions from Greek 
 to Sanskrit: thus we have in Sanscrit devas=^^sS^og (dhevos); ekataras 
 {aikataras) =■ εκάτερος ; the Sanscrit eka=aika is yak in modern Persian, 
 and the Sanscrit vedmi (=vaidmi), phena (=phaina), gveta (=zgvaita) 
 correspond to the Sclavonic vjemi, pjena, svjet. But it is not at all 
 uncommon in Greek: thus from the last-mentioned root (βκαζ^ Sanscrit 
 ekas^ Latin cequus and secus)^ we have ενκηλος for εκηλος, a change 
 which it is certainly not necessary to account for by the supposition of 
 two digammas, as Thiersch and Buttmann have supposed. The gloss in 
 Hesychius (γέγκαλον, ^'(9i>;^ov), which would seem to point to a form 
 f^f κάλος, is evidently wrong from its position between γειώρας and 
 γεκα^ά (1. γεκονΰα) in that vocabulary : we should read γεκαλον with 
 Pearson and Guyet. Sometimes the digamma, which should have 
 begun the word, was transferred not only to a place behind the first 
 vowel, but even to the beginning of the second syllable , as in the fol- 
 lowing instance: ολος=^όλος, "full," became first ονλ6ς=ο^λθ£, 
 then oXfog, as we see in όλβος and ολβαχψον (as it should be read 
 in Hesychius, s. v. εϋτίλοντον) from ονλι^ and %έω. The same prin- 
 ciple explains the shifting of the aspirate in such cases as εχω (^ek-ho), 
 εξω (hek-so), τρέχω (trek-ho), ^ρέξω (threk-so), &c. We conceive 
 then that the adoption of Η as the sign for the long vowel 7j=e , is 
 due to the fact that he was actually considered as equivalent to e. 
 The vowel η really contains, in many cases, not merely the common 
 rough breathing, but also the digamma hv, and even the softened 
 dental or guttural dy, j, as in ItVTCriv for Irvjcyav, or, when aspirated 
 at the beginning of the word, as in ήμερος for δίάμερος, &c. This J 
 or y is also represented by ε in the middle of a word, as in 7ΐ6λεως= 
 7tόλyως ; and we often find that ε ε presumes a single ε preceded by 
 some guttural breathing (Buttmann, Ausfuhrl. Sprl. § 112, 17, Amn. 
 23). Compare εε'ρ^τ; with the Sancrit ί^ατ-^Λβ. We shall not be sur- 
 prised, therefore, to find that η is also in many cases the representa- 
 tive of ε ε : thus δέελος is another form of δήλος and yδεε, of '^δη; — 
 άναλνεταο γονν το η εΙς δύο ε ε, ώς τίαρα τω ποίψγ) — δεελον δ' επί 
 ύημά τ εχεναν — καΐ όνναίρεΐται πάλιν τα δύο ε ε εΙς το η, ώς το 
 ^δεε, ^δη{3οήοΙ. Dionys. Thr. p. 797). The form of ω shows that it 
 is a similar combination of oo, and there is every reason to believe 
 
CHAP, v.] THE GREEK ALPHABET. 207 
 
 that this was its real value. As the Sanscrit α=«+α regularly cor- 
 responds to ω as well as to a, and as the Sanscrit a represents the 
 lighter δ no less than the heavier a , we may fairly conclude that ω 
 is the reduplication of ο just as α is of a, or that in the longer as well 
 as in the shorter vowels the Greek alphabet denotes those differences 
 of weight, which the Sanscrit neglects. In this scale of weights f is 
 the lightest vowel. But ύι is heavier than ω, which is substituted for 
 it in derivative forms or heavier words; thus from jCafriQ we have 
 άτΐάτωρ, from φρην, ΰώφρων ; and we have the heavier words 'Jrcc- 
 λίώτης, ύτρατιώτης by the side of τζολίήτης. The statement, therefore, 
 that η=εε^ requires the explanation given above, namely that εε in 
 this case must be regarded as a fainter articulation of the ta to which 
 rj is etymologically equivalent. The pronunciation of η takes it out of 
 the category of the mere articulation vowels a, ε, ο. It corresponded 
 to the Hebrew tsere, i. e. to our long a as in mate, or to our long e as 
 in there. The passage from ta to this sound may be seen in a compa- 
 rison of the German ja with our yea. The act of utterance in o, ω, no 
 less than in a, a, ε, is consistent with a fully-opened mouth, and this 
 is indicated by the Hebrew names patha^h^ qametz^ and ^holem^ and 
 by the relation between ο and the nasal 3>; while the narrower open- 
 ing and the formation of the lips into a mere fissure , by which the 
 tsere and the cognate or included ^hirik are articulated, are suffi- 
 ciently expressed by these names. This difference is recognised by the 
 Greek grammarians, who give the following description of the sound 
 of 7] : δει το μεν η εκφωνονντοί μηκννει,ν το ΰτόμα ώς ετά τα ώτα 
 εκατέρωθεν, το ώ εκφωνονντα μηκννευν τα χείλη Sg ετά την ρίνα 
 καΐ τον Λωγωνα (Bekkeri Anecd. p. 797), and is confirmed by the 
 fact that Cratinus and Aristophanes represented the bleating of the 
 sheep by the syllable βη (see Meineke, Fragm. Com. Ant. p. 40 ; Hesych. 
 s. V. βη λέγεο; Etym. Μ, 196, 7; and Bekker. Anecd. p. 86). 
 
 117 Although we must reserve for subsequent chapters some 
 further discussion of certain letters, we shall perhaps consult the con- 
 venience of the student by stating briefly in this place the results at 
 which we have arrived respecting the Greek Alphabet in general. We 
 enumerate in the established order all the characters at any time 
 employed by the ancient Greeks. The original syllabarium is distin- 
 guished by the employment of capitals and Roman numerals, and we 
 have added the Hebrew letter when the character is of Semitic origin. 
 
 (1) I. A^ N, at first a mere breathing, afterwards a broad, 
 open sound, frequently used as a representative 
 of the nasal breathing, just as 5> passes through ;i 
 into N. 
 
(2) 
 
 Π. 
 
 Β, 
 
 (3) 
 
 III. 
 
 Γ, 
 
 (4) 
 
 IV. 
 
 ^, 
 
 (δ) 
 
 V. 
 
 Ε, 
 
 208 THE THEORY OF [bOOK I. 
 
 '2j generally like our δ, but sometimes, at it seems, 
 employed as a v. 
 
 Λ, a hard ^, sometimes accompanied by a nasal 
 breathing. 
 
 % often pronounced with a lisp, and then approxi- 
 mating to d" and ρ. 
 
 Ϊ1, at first an aspirate, afterwards the residuary light 
 vowel with which that aspirate was articulated; 
 often pronounced like y when followed by an- 
 other vowel. 
 
 (6) VI. /, 1, a combination of the guttural breathing with the 
 
 labial, most usually under the form hv or hu•, in 
 its original value the labial predominated. 
 
 (7) g, it, originally ds, transposed in some dialects to sd, 
 
 and softened generally into the sound j or sh, 
 which is equivalent to di or gi. 
 
 (8) VII. H, n, a hard aspirated guttural, pronounced hg or gh, 
 
 afterwards a long vowel like the Hebrew tsere 
 and our a in ale^ but always implying some ety- 
 mological absorption, especially the syllable La. 
 
 (9) VIII. Θ, Ώ, originally hd or dh, afterwards softened through 
 
 th into an approximate sibilant, and always closely 
 allied to d. 
 a vocalized guttural. 
 
 a substitution for q: occurs twice as a final letter, 
 sometimes approximated to the soft French I. 
 ?3, \ did not usually differ from their representatives 
 D, Mn other languages ; they came nearer to the 
 mediae than to the tenues; thus μ delights in con- 
 tacts with |3, V with d\ and in later applications 
 of the alphabet, μτί represents j3, and vr, d; the 
 same appetency for a quasi-medial articulation is 
 observable in the other dental liquids λ, ρ, which 
 often represent v, d, or 0"; ν is one of the most 
 frequently used of the final consonants; and in 
 this employment it has often taken the place of 
 an originally final ^ , or of ^ which by apocope 
 has become final; both μ and ν may approximate 
 to the nasal breathing. 
 (15) ξ, ^, originally {?κ from (5χ or sh; afterwards trans- 
 
 posed to ks; in some dialects it retained its softer 
 sound. 
 
 (10) t, 
 
 \ 
 
 (11) κ, 
 
 ^^ 
 
 (12) IX. A, 
 
 b, 
 
 (13) X. M, 
 
 52/ 
 
 (14) XL N, 
 
 3, 
 
CHAP, v.] THE GREEK ALPHABET. 209 
 
 (16) ΧΠ. O, 3>, at first a mere nasal breathing; afterwards au 
 
 intermediate value of the articulation vowel; 
 sometimes pronounced as w before another 
 vowel. 
 
 (17) XIII. 77, δ, did not differ from its modern equivalent. 
 
 (18) XI Y. Q, p. properly a combination of guttural and labial, 
 
 like /; the guttural however predominated in 
 this case. 
 
 (19) Q, Ί, approximated to d and -9•; and is occasionally 
 
 found as a final letter. 
 
 the most usual sibilant; very often occurs as a 
 final. 
 
 did not differ from the ordinary dental tenuis, 
 the residuum oi f = hu, when the letter became 
 ifjiXov by the omission of the aspirate, 
 an imperfect substitute for another value of the 
 lost p. 
 
 a substitute for Η after its disuse as a conso- 
 nant. 
 
 an arbitrary combination of π and 6. 
 a double o. 
 
 an arbitrary combination of 6 and jc, afterwards 
 obsolete, except as a numeral sign. Its name 
 was Σαμπί, i. e. όάν and πι. The former repre- 
 sents the only Hebrew letter omitted in the above 
 list, namely, Τ or Zain, which was once used by 
 the ancient Greeks. 
 
 118 (3) Interchange of mutes in the Greek and cognate languages. 
 
 We now come to the third subject of inquiry which we have pro- 
 posed to ourselves in this Chapter — the changes which take place in 
 the mutes or fundamental consonants of related words in the different 
 languages of the Indo- Germanic family. The liquids usually remain 
 unaltered in the corresponding words, and the breathings we have 
 already considered. It has been perceived that the changes of the 
 mutes generally follow a very striking law, which was first pointed 
 out, in a partial and imperfect manner, by Rasmus Rask, and after- 
 wards established, in its application to the Greek (Latin, Sanscrit), 
 the Gothic, and Old High German, by James Grimm {Deutsche Gramm. 
 I. p. 584 foil.). This law has been extended by Bopp (Vergl. Gramm. 
 pp. 78 foil.) to the Zend and Lithuanian. Some of the exceptions to 
 which the rule is liable have been indicated by Dr. Guest (Prov. Phil. 
 
 Ρ 
 
 (20) XV. 
 
 Σ, C, 0, 
 
 (21) XV] 
 (22) 
 
 [. Τ, η, 
 ν, 
 
 (23) 
 
 φ. 
 
 (24) 
 
 ζ. 
 
 (25) 
 (26) 
 (27) 
 
 ψ, 
 
 ω, 
 
 Φ, 
 
210 
 
 THE ΤΗΈΟΕΥ OF 
 
 [book I. 
 
 Soc, III. pp. 179 sqq.). The following table will afford the best expla- 
 nation of the canons as given by Grimm. 
 
 Greek (Latin, Sanscrit) 
 
 Gothic 
 
 Old High German 
 
 Oi 
 
 Greek (Latin, Sanscrit). 
 
 Tenuis 
 Medial 
 Aspirate 
 
 Labials. 
 
 Ρ h f 
 
 f Ρ b 
 b (v) f ρ 
 
 Gothic. 
 
 Aspirate 
 
 Tenuis 
 
 Medial 
 
 Dentals. 
 
 t d th 
 th t d 
 d ζ t 
 
 Gutturals. 
 
 k g ch 
 ..kg 
 g ch k 
 
 Old High German. 
 
 Medial 
 
 Aspirate 
 
 Tenuis 
 
 It must be remarked, that the Gothic aspirate, to which the Greek 
 tenuis corresponds, is not οΛ, for that combination does not exist in 
 Gothic, but either h or g with a strong guttural aspiration. The same 
 remark applies to the Latin, which, however, consistently employs the 
 strong A for the Greek χ (see above, p. 200). In Old High German, 
 b is superseded by v, a circumstance which has also taken place in the 
 modern Greek and other languages, and instead of th we have z=:ts 
 by assibilation instead of aspiration. 
 
 119 The following exemplification of the law is taken with some 
 slight modification from Bopp; we have subjoined his comparison of 
 the Zend and Lithuanian with the languages included in Grimm's 
 canon : 
 
 Sanscrit. 
 
 Greek. 
 
 Latin. 
 
 Gothic. 
 
 Old High German. 
 
 pdda-s 
 
 7ίοδ-6ζ 
 
 pedis 
 
 fotus 
 
 vuoz 
 
 panchan 
 
 τίεντε 
 
 quinque 
 
 fimf 
 
 vinf 
 
 purna 
 
 Tclkog 
 
 plenus 
 
 fulls 
 
 vol 
 
 pitri 
 
 τίατηρ 
 
 pater 
 
 fadrein (pi.) 
 
 vatar 
 
 upari 
 
 νπέρ 
 
 super 
 
 ufar 
 
 ubar 
 
 hhanj 
 
 Γρηγ-νν-μί 
 
 frango 
 
 brikan 
 
 prechan 
 
 bhratr 
 
 φράτωρ 
 
 frater 
 
 brothar 
 
 pruoder 
 
 bhri ' 
 
 φέρω 
 
 fero 
 
 baira 
 
 piru 
 
 kapdla 
 
 κεφαλή 
 
 caput 
 
 haubith 
 
 houpit 
 
 tvam 
 
 τυ 
 
 tu 
 
 thu 
 
 du 
 
 tray as 
 
 τρεις 
 
 tres 
 
 threis 
 
 dri 
 
 antara 
 
 έτερος 
 
 alter 
 
 anthar 
 
 andar 
 
 danta-m 
 
 οδόντα 
 
 dentem 
 
 thuntus 
 
 Zand 
 
 dvau 
 
 δύο 
 
 duo 
 
 tvai 
 
 zuene 
 
 dakshina 
 
 δέξίος 
 
 dextra 
 
 taihsvo 
 
 zesawa 
 
 uda 
 
 νδωρ 
 
 unda 
 
 vato 
 
 wazar 
 
 Gvan 
 
 κνων 
 
 canis 
 
 hunths 
 
 hund 
 
CHAP, v.] 
 
 THE GREEK ALPHABET. 
 
 211 
 
 Sanscrit. 
 
 Greek. 
 
 Latin. 
 
 Gothic. 
 
 Old High German. 
 
 hridaya 
 
 καρδία 
 
 cor 
 
 hairto 
 
 herza 
 
 aksha 
 
 oKog 
 
 oqvulus 
 
 augo 
 
 ouga 
 
 agru 
 
 δάκρυ 
 
 lacrima 
 
 tagr 
 
 zakar 
 
 'pagu 
 
 τΐώϋ 
 
 pecus 
 
 faihu 
 
 vihu 
 
 ςναςηνα 
 
 εκνρόξ 
 
 socer 
 
 svaihra 
 
 suehur 
 
 dagan 
 
 δέκα 
 
 decern 
 
 taihun 
 
 zehan ^ 
 
 jnd 
 
 γνώμι 
 
 gnosco 
 
 lean 
 
 cJian 
 
 jdti 
 
 γένος 
 
 genus 
 
 kuni 
 
 chuni 
 
 jdnu 
 
 γόνυ 
 
 germ 
 
 kniu 
 
 chniu 
 
 mahat 
 
 μεγάλο- 
 
 magnus 
 
 mikils 
 
 mihil 
 
 hansa 
 
 %Ψ 
 
 'anser 
 
 gans 
 
 kans 
 
 Jiyas 
 
 Χ^έζ 
 
 heri 
 
 gistra 
 
 kestar 
 
 lih 
 
 λείχω 
 
 lingo 
 
 laigo 
 
 lekom. 
 
 The following comparison shows that the Lithuanian ranks with 
 the first three languages in the interchange of mutes. 
 
 Lithuanian, 
 
 Sanscrit. 
 
 ratas, "a wheel" 
 
 rathas, "a chariot'' 
 
 M5M, "Ishallbe" 
 
 bhavishydmi 
 
 kas, "who" 
 
 kas 
 
 dumi, "I give" 
 
 daddmi 
 
 pats, "a husband or master" 
 
 patis 
 
 penki, "five" 
 
 panchan 
 
 trys, "three" 
 
 trayas 
 
 keturi, "four" 
 
 chatvdras 
 
 ketwirtas, "the fourth" 
 
 chaturthas 
 
 szaka, "a branch" 
 
 gdkU 
 
 In most cases the Zend also agrees with the Sanscrit as opposed 
 to the Gothic. 
 
 Zend. 
 
 Sanscrit, 
 
 Gothic, 
 
 turn 
 
 tvam 
 
 thu 
 
 chatJiwdro 
 
 chatvdras 
 
 fidvor ■ 
 
 pancha 
 
 panchan 
 
 fimf 
 
 pereno 
 
 purna 
 
 fulls 
 
 paitis 
 
 patis 
 
 faths 
 
 pagus 
 
 pagu 
 
 faihu 
 
 pddha 
 
 pddas 
 
 fStus 
 
 peregaiti 
 
 prachch'hdti 
 
 fraihith 
 
 kd 
 
 kas 
 
 hvas 
 
 dashina 
 
 dakshina 
 
 taihsvo 
 
 Ρ 2 
 
212 THE THEORY OF [bOOK I. 
 
 The Zend has no bh, and therefore agrees with the Gothic in the 
 use of the medial for the aspirate. 
 
 Ze»d. 
 
 Sanscrit. 
 
 Gothic. 
 
 baraiti 
 
 bibJiarti 
 
 bairith 
 
 brdtarem 
 
 bhrdtaram 
 
 brothar 
 
 uba 
 
 ubhdu 
 
 bai 
 
 abi, aiwi 
 
 abhi 
 
 bi 
 
 maidhya 
 
 madhya 
 
 midja 
 
 bandh bandh bindan 
 
 There are other cases in which the Zend corresponds to the Gothic 
 rather than to the Sancrit : thus we have 
 
 Zend. 
 thri 
 
 '' Gothic. 
 
 thri 
 
 Sanscrit. 
 
 tri 
 
 thwoi 
 
 thus 
 
 tre 
 
 fra ^ 
 
 dfrindmi 
 
 dfs 
 
 fra 
 
 frijo 
 
 ahva 
 
 pra 
 
 prindmi 
 
 ap 
 
 120 We have before pointed out that, in the use of the soft aspi- 
 rate for s, the Zend corresponds to the Greek as contrasted with the 
 Sanscrit and Latin. We now proceed to show that the Greek lan- 
 guage, too, in addition to these agreements with the younger class of 
 languages, also presents a peculiarity, in the use of the mutes, which 
 belongs to the third rather than the first class of languages, according 
 to this arrangement. This peculiarity consists in employing an aspi- 
 rate of the labial or dental order instead of the corresponding medial 
 or tenuis which appears in the Latin, Sanscrit, and Sclavonian. In 
 some of these cases the Sanscrit and Greek stand together against Zend 
 and Latin, so that we must suppose that the effect of time in softening 
 and aspirating has been more felt upon the pliant frames of the former 
 than upon the tough antiquity of the two latter languages. The San- 
 scrit, in particular, presents many instances of softening and aspiration 
 which are not found in any of the other more ancient languages of 
 this family, as in the substitution of ch and ρ for k, of j for g, &c. The 
 following instances among others will show that the Greek sometimes 
 forfeits its claim to a place among the oldest languages : the Latin or 
 Sclavonian b is represented by /, as in Old High German, in ορφανός, 
 Latin orbus; άλφός, Latin albus; ομφαλός, Latin umbilicus, Lettish 
 nabba; άμφώ, άμφί, Latin ambo, ambivium, Zend uba, Gothic bai; in 
 these last two cases the Sanscrit nabhis, ubhdu agree with the Greek : 
 the form άμηί is still found in remains of the ^olic dialect: ρ is repre- 
 sented by/ as in Gothic, in κεφαλή, Sanscrit kapdla, Latin caput; in 
 
CHAP, v.] THE GEEEK ALPHABET. 213 
 
 ϋοφοξ, Latin sapiens; β often becomes φ, and vice versa, in Greek 
 itself, as κορυφή, κόQvμβog; ΰνρ£φω, ΰτρέβλος, &c.; the d of the old 
 languages in represented by θ, ιη^-υγάτηρ compared with Sanskrit du- 
 hitar, Sclavonian dotsher, Lithuanian dukter; in ^νρα, Sanscrit dvdr, 
 Sclavonian dver, and so forth. Mr. Winning would infer from the 
 appearance of φ, ^, and χ, for bh, dh, and gh or h in Sanscrit, that the 
 Greek language actually presents an aspirate instead of a medial in 
 these cases (Manual, p. 42), and fancies that he has discovered in this 
 a very curious interchange between Greek and Gothic, as in the fol- 
 lowing table (p. 111). 
 
 Older Greek. 
 
 Proper Gothic. 
 
 More recent Greek. 
 
 Older Gothic. 
 
 νδωρ 
 
 vato 
 
 ^υγάττιρ 
 
 daughtar 
 
 δεξιός 
 
 taihsvo 
 
 ^ύρα 
 
 daur 
 
 Ttodfg 
 
 fotus 
 
 οφρυς 
 
 bra 
 
 άγρόζ 
 
 akrs 
 
 νεφέλη 
 
 nibl 
 
 γόνυ 
 
 kniu 
 
 %Ψ 
 
 gans 
 
 μεγάλο- 
 
 mikils 
 
 %^k 
 
 gistra 
 
 But it must be remarked that in every instance which he has 
 adduced as an exception, the consonant objected to is an aspirate, and 
 that the Greek aspirates are only of the tenuis order, while those in 
 general use in Sanscrit are only of the medial order, and that, there- 
 fore, no argument can be drawn from this discrepancy, which, indeed, 
 admits of an explanation derivable from the vacillations and incom- 
 pleteness of the aspirates (above, p. 201). 
 
 121 Grimm's law applies only to the interchange of mutes of the 
 same order considered according to their distinction as tenues, aspi- 
 rates and medials. The interchange of mutes with others belonging 
 to different organs, e. g. of labials with dentals or gutturals, is not an 
 exception to the law, as Grimm calls it (i. p. 589), for it belongs to a 
 different principle, which we propose to term " the law of divergent 
 articulations," and which is illustrated and proved by the following 
 induction. 
 
 When ρ is changed to t we must consider it as having arisen from 
 a false articulation, which has formed a dental out of the sibilant 
 originally attached to the labial in the particular case. Thus from 
 the root /cc or (37ta are formed both τε and jtov. Similarly when ρ 
 becomes^, as in ΛοΓο^,κοΓοί, there has been an union of a guttural and a 
 labial in the original sound ; compare the Latin quis. When b becomes 
 dj or d becomes g^ the original sound must have comprised both con- 
 sonants ; thus bis and dig spring from d/tg, like bellum from dvellum, 
 and γη and 8η from γδή, like γυμνός for εγδνμενοξ. This principle 
 
214 THE THEOEY OF [bOOK I. 
 
 extends to combinations of mutes and liquids as well as to combina- 
 tions of mutes with mutes; thus κελαινόξ and μελαν spring from 
 αμελαν,ΆΒ appears from τα κμελεΟ'ρα quoted from theglossary of Pam- 
 philus, by Herodian, and from him by the author of the Etymologicum 
 Magnum (see Buttmann, Lexilog. u. p. 265). 
 
 We have sometimes double examples of these changes in the 
 same word. Thus the Sanscrit paktas corresponds to the Greek 
 τίετίτόξ in its initial, and to the Latin coctus in its second syllable. 
 But coquo, from which coctus is formed, gives us a compound of 
 guttural and labial in the latter case, and as we are told that coquus 
 was pronounced quoquus even in Cicero's time (Quintilian, /. 0. 
 VI. 3, § 47), we may infer that the original form of the verb was 
 quoquo. So that ιίετίτόξ contains in both syllables only the labial 
 part of the compound qv=kp; coctus contains in both syllables only 
 the guttural ingredient; and paktas impartially omits the alternate 
 elements. We have a very similar case in vivus compared with quick 
 (above, § 1 12, note J). The Greek digamma furnishes constant examples 
 of this principle, — that the archetype of incongruous articulations in 
 cognate words must have been a compound sound containing, or 
 capable of containing, both of the heterogeneous ingredients (above, 
 § 110). But the digamma is limited to the combinations of the gut- 
 tural with the labial, and the law which explains the ramifications in 
 this particular case, is equally applicable, as we have just seen, to 
 other combinations, such as the dental and labial, or the guttural and 
 dental. Among the most usual of these exemplifications of " the law 
 of divergent articulations," we must mention the occasional appear- 
 ance of a resolvable compound sound before the interchangeable liquids 
 I or r. The more usual instances are given in a subsequent chapter 
 (§ 212). One or two examples will suffice to explain the principle. 
 If we take the Latin lac^ we observe no trace of any prefixed or 
 initial sound before the I. But we get a guttural prefix as soon as 
 we compare the Greek γάλα{7ί)^ γλάγοξ^ and a labial prefix appears in 
 the Gaelic bligh, Sclavonic mliek, Latin mulgeo, Lith. melzu, Angl. 
 Sax. meolc^ English "milk." As these prefixes could not be derived 
 from one another, there must have been a compound sound at the 
 commencement of the word in its original form, and analogy suggests 
 that this must have been, like the digamma, a labial preceded by a 
 guttural. Now we have in Greek the verb άμέλγω, and that the short 
 initial vowel is the residuum of a guttural articulation is shown by 
 the Greek ερετμός compared on the one hand with the Latin remus 
 and the Gaelic ramh, and on the other hand with the Sclavonic grehlo^ 
 h-reblo, and the Welsh rhwyf. There are many other instances of 
 the representation of an initial guttural by ε or ο (Benfey, Wurzellex. 
 
CHAP, v.] THE GEEEK ALPHABET. 215 
 
 u. p. 120). But the most interesting is ε-λεν^'ερος compared with the 
 Latin libe7' and the Sanscrit g-ridh and g-ribh (Benfey, l, c. p. 140), 
 because this furnishes us with another example of the principle under 
 consideration, and leads to the important identification of the Latini 
 and Lavini with the Lithuariians, whose name contains both forms, 
 broken up again into the Lettonians and Livonians ( Varron. p. 6 1). The 
 aspirate presumed in the ε- of B-kEV^EQog is transposed in the Welsh 
 rhydd, and the same is the case in rhwyf compared with hreblo. To 
 return to lac, the Welsh equivalent is not only hlith, but llacth, and 
 there are very many instances in which a prefixed guttural or labial, 
 or even a compound sound, is represented by tliis double /, the most 
 familiar being the surnames Floyd or Fludd and Fluellen, as repre- 
 sentatives of the Welsh Lloyd, Llwyd^ and Llewellin. And this leads 
 us to another interesting comparison of ethnographic import. For 
 while we may compare the Αελεγες (Lieges) with the Αίγνες ( Varron. 
 p. 63; Cambridge Essays for 1856, p. 35), we have a trace of the 
 original prefix in the Φλεγνεξ or Φλεγναι of Bceotia. The extent to 
 which this double I may represent a compound prefix is shown by 
 the Welsh Hid compared with the Latin lis (lit-), originally stlit-, Old 
 High German strtt, Old Norse strida, Anglo-Saxon fly tan, and Greek 
 i.-Qi(S)g. The analogy of stlatus (stlatarius) and stlocus for latus and 
 locus, seems to show that t is only an euphonic insertion, and it is 
 omitted in the epitaph of Cn. Cornelius Scipio, where we have si. 
 judik. for slitibus judikandis (Varron. p. 224). That the s is the resi- 
 duum of a compound articulation, which, besides the sibilant as a 
 representative of the guttural, contained a labial element, may be 
 inferred from a comparison of stlocus with the Breton leacJi, the 
 Lithuanian plecus and the Lancashire pleck. The constant inter- 
 change of I and r, of which stlit and strtt furnish an example, will 
 hardly allow us to distinguish between the cases which are now under 
 consideration, and those in which r is preceded by a moveable articu- 
 lation. Of these cases there are many examples in Greek. Thus, for 
 ρίζα we have the ^olic form βρίζα or βρίΰδα, and this may be com- 
 pared Avith the German vaurts, the Welsh gwraidd, and the Sanscrit 
 hradJina. If the ζ represents an original γ^ we may farther compare 
 <5(ρριγάω and αβτίάραγοζ. Similarly, we know that ρηγννμι had the 
 digamma, for Ρρη^ΐξ is quoted from Alcaeus ; and this is confirmed by 
 the analogy of the Latin f-rango^ German b-rechen. 
 
 On the whole we are disposed to regard this "law of divergent 
 articulations" as involving a principle of scarcely less importance than 
 Grimm's rale for the interchange of mutes of the same order, to which 
 it may be regarded as a necessary supplement. The Greek gram- 
 marians were content to designate all interchanges of consonants by 
 
216 THE THEOEY OF [bOOK I. 
 
 the general name of μετάλη^Ι^ίς (Heraclid. ap. EustatTi. 1654, 19). 
 Scientific etymology requires us to regard as entirely distinct cases 
 the passage of one cognate sound into another, e. g. that of ρ into b; 
 and the appearance of incongruous articulations in words confessedly 
 of the same origin ; and while we recognise the possibility of a direct 
 transition in the former case, v/e must say of the latter, that "none of 
 the known forms are strictly speaking original, but that all have 
 branched out of some still older element, capable, according to known 
 phonetic laws, of producing them all *." 
 
 It is right to add that the same phenomenon is observable in forms 
 in which the complex sound is not the original articulation, but has 
 arisen by some subsequent process of consonantal mutation. Thus in 
 Corfu and Negripont, the corrupt modern names of Κερκνρα and Ig 
 τον EvQiTtoVj we have the labial and guttural elements respectively of 
 the more recent articulations Korkvra (according to the modern sound 
 οι v=^v) and EvJiripon = Egvripon. Similarly in the French we have 
 cage from cavea, Dijon from Dibio, rage from rabies^ sacJie and sage 
 from sapiam and sapiens (Varron. pp. 241, 244), where the complex 
 sound, thus represented by one of its elements, has arisen from a mere 
 synizesis of the vowel e or i, which has been thrown back on the pre- 
 ceding labial, has combined with it the palatal j, and afterwards 
 superseded the original middle-sound, or converted the head of the 
 firm into a sleeping partner. Something of this kind must be the 
 true explanation of the fact that rudere, which has the first syllable 
 short in the classical poets of the best age (Virg, Georg. ni. 374; ^n. 
 VII. 16; Ovid, Fast. i. 433; vi. 342; de Art. Am. iii. 290), has [the 
 u long in Pers. m. 90. The perfect rudivi shows that there must 
 have been a by-form rudio like rvgio, and it is probable that rudere 
 was pronounced rudjere or rugere. 
 
 The interchange of aspirates of different organs we have before 
 explained (above, § 108). 
 
 * This description of our principle is taken from the valuable paper by Mr. 
 Garnett, "On certain initial letter-changes in theEuropean languages " (jEssaj/s, 
 p. 253). .Mr. Garnett (p. 242) states it to be the object of his paper "to bring 
 further evidence in favour of the general correctness of the above theory from 
 some collateral sources of illustration, which it did not enter into Mr. Donaldson's 
 plan to notice;" and he may be regarded as having demonstrated scientifically 
 the proposition, which was briefly stated in the original edition of this work, not 
 merely under the head of the digamma (above, § 110), but also in the present 
 section. Mr. Garnett mentions that "nearly the same view of the subject," as 
 far as the digamma is concerned, has been taken by Hofer in his Beitrage ziir 
 Etymologik, a work which we have never seen, but which was published soon 
 after the first appearance of the present book. 
 
CHAP, v.] THE GEEEK ALPHABET. 217 
 
 122 We conclude this Chapter with a table of the consonants 
 which correspond in related words of the Sanscrit, Greek, and Latin 
 languages. If the reader desires to see this table immediately con- 
 firmed by examples, he may consult Pott's Etymologische ForscJiungen, 
 I. p. 84 and following. 
 
 Sanscrit. Greek. Latin. 
 
 k κ, Λ c (qv). 
 
 ksh I, σ, κ, κρ (jr) . . X (c-s) c, s, cr. 
 
 kh χ. qv, gv. 
 
 g • . . . y, j3 g, b. 
 
 gh • % gv. 
 
 .n (like the final η in French) γ (nasal) η (adulterinum). 
 
 ch jc^T c (qv). 
 
 ch'h <S%T 6k sc, c. 
 
 J ί g• 
 
 jh no example. 
 
 η (palatal) η (guttural) . . . some nasal. 
 
 u 
 
 m 
 Oh 
 
 
 The cerebrals have few if any representatives among the Greek 
 and Latin letters. 
 
 Sanscrit. Greek. Latin. 
 
 t r, (? t, s. 
 
 th τ t. 
 
 d d, O" d, e. 
 
 dh θ, (? f, d. 
 
 η V) λ, ρ 11)1• 
 
 Ρ Jt, φ Ρ, 1 (qv). 
 
 ph no example. 
 
 b • . 13, Λ b. 
 
 bh .' φ {β) f, b. 
 
 m μ {β bef. liquids) . m. 
 
 y (palatal) ξ, t, ε, -θ", aspirate j, i. 
 
 r (lingual) ρ, λ r, 1. 
 
 1 (dental) λ 1. 
 
 V (labial) J^,t;,£,j3,qp, aspirate v. 
 
 'g (s palatal) κ, <J, aspirate . . c (qv) s. 
 
 sh (lingual) (?, aspirate . . . . s. 
 
 s (s dental) <J, aspirate . . . . s. 
 
 h (guttural) X^V^^ h, g, c. 
 
APPENDIX 
 
 To Book I. Chap. V. § 110. 
 
 (A) 
 
 The Digamma as it appears in Inscriptions. 
 
 The words, which exhibit the digamma in extant inscriptions 
 have been collected and arranged in the order of the dialects by 
 J. Savelsberg {de digammo ejusque immutationibus dissertation Aquis- 
 grani, 1854). They are as follows: 
 
 I. -^olic Inscriptions. 
 
 a. Bceotic: fvKtas(C.7.n. 1562, 1563, 1564); f loot ελίαν{η. 1662, 
 1563); Γάρνων nom. pr. (n. 1569, an.) ; Γελατίη nom. urb., Γελατιήϋ, 
 Jrksai ^i^ccTL, Jrautiog (n. 1569, a. m.); ]^άδων nom. pr. (n. 1574); 
 afvd6g=^aoLd6g, in compounds in ρα^α^νδΟ£^=ραί1^ωδόζ (η. 1583); 
 BaxevFaL• nom. pr.(n. 1639) ; ^ιλαρχιόντων (Leake, Travels inNorth- 
 em Greece, ii. n. 31); Γάύκων (lb. n. 33); JFvKLag (Bhein. Mus. new 
 series n. p. 107, n. 7) ; PuorivLog (3. p. 108, n. 8) ; FaΰτvμεLδόvτL•og 
 and perhaps ^ETtLfalreig (Ulrichs, JReise in Griecheril. Vol. i. p. 247) ; 
 EvFaga and Γάΰτν (Eckhel, Doctr. Num. ii. 196). 
 
 b. Elean: Ρράττρα, Faldoig, EvfaoiOLg, Γετεα, Hitog, ^άργον, 
 Fkrag (Bockh, C. J. n. 11; StaatsTiaush. ii. p. 390). 
 
 II. Doric. 
 
 a. Crisscean: εχοι %khFog ait^itov αΙΓεί (Bockh, C. I. n. 1; Ul- 
 richs, Eeise, i. p. 31). 
 
 b. Argive: Γίκα[τι] (Bockh, C. I. n. 18); ττεδάΓοικοί (η. 19); 
 ^lH (η. 29). 
 
 c. Spartan: εδεδοΓα (η. 15); ^ίκατι (η. 1511); Γαΰΰτνοχο 
 (η. 1520; Ahrens, dial. Dor. p. 100). 
 
 d. Cretan: Fίκaδίovog nom. propr. (C. I. n. 2598); Γανξίων (η. 
 3050); Γαξίων (Eckhel, Doctr. Num. iv. p. 388); J=ελχάvog (Bull, 
 del Inst. Arch. 1841, pp.61, 174, cf. Hesych. fελχάvog, b Zεvg πάρα 
 Κρηΰίν). 
 
 e. Corcyrcean: J^a6(5[tv] (C. I. n. 20); ΤλαόίαΓο (cf. the form of 
 the gen. in the Phrygian inscription, above, § 92), τΐρύξενΓος, ροΓαΐύί, 
 ΰτονό^εϋύαν and perhaps άροΰτενΡοντα and ά^ντάν (Trans. Philol. 
 Soc. Vol. I. n. 14; Zeitschr. f. vergl. Sprachf. i. p. 118). 
 
CHAP. Vj APPENDIX TO § 110. 219 
 
 /. Italian (in Magna Graecia): AlS^ag (Monum. inedit. del Inst. 
 Arch. Bom. Tom. ii. tab. vui. ix.); ταξ Ηήρας Ηιαρός Ιμι rag εν 
 πεδίωι' Θύνίΰκός με άνέ^ηκε/Όρταμος Γερίον δεκάταν(οη a votive 
 axe^Ephem.Arehceol.n.Gl; Bullet. Ναρο1.η.18,'ρ.Α22)', Ιέτος, Ιετεος^ 
 PsTTft, Γετη, ^ετεων, Γεξήκοντα, Γεξακάηαί^ J^idiav, fidiaL (εν τα 
 ftdCa γα), εγ^ηλη^ίωντι, (ε^είληΟ'ώβί), /ίκατί, ^ίκοίτίδειον, Ιικατί- 
 Λεδον,^είκατι, /εξ, f έκτα (Tabulce Her acleenses); ^ουκίαν{αη.^. 
 
 III. Ionic: \τ\ο a f ντον λί^ο εμι ανδριαξ και το ύφελας (Bockh, 
 C. Ι. η. 10), i. e. ταντον λί&ον είμ άνδρίας κάί το ςφέλας, "of the 
 same stone I am a statue and its basis " (Bentley, Correspondence, ed. 
 Wordsworth, pp. 589, 598). 
 
 (B) 
 
 Extracts from Bentley's MS. on the Digamma. 
 
 Bentley first quotes the following authorities: Dionys, Halic. i. 
 
 20. Servius, ad jEneados vi. 359. Julianus, Orat. xi. p. 71; and the 
 
 following passages from "^ Grammaticce Latince auctores antiqui ; edidit 
 
 Putschius." Diomed. Gramm. p. 416. Priscian, pp. 546, 7 (where, 
 
 on the words — inveniuntur etiam pro vocali correpta hoc digamma illi 
 
 usi, ut Alcman: κάί χεΐμα τΐνρ τε δά^Ρυον^. — / digamma uEolis est, 
 
 quando in metris pro nihilo accipiebant, at αμμεξ d' Ιειράναν το δε 
 
 ταρ Ο'έτο μώύα λιγεια'^. — Sciendum tamen quod hoc ipsum JEoles 
 
 quidem uhique loco aspirationis ponehant, effugientes spiritus asperi- 
 
 tatem^. Hiatus quoque causa solebant illi interponere J^ digamma, quod' 
 
 ostendunt etiam poetce JEolidce ; uti Alcman: καΐ χεΐμα τΐνρ τε δάΡυον 
 
 et epigrammata quod egomet legi in tripode vetustissimo Apollinis, qui 
 
 Stat in Xerolopho Byzantii, sic scripta: ^EMOTIHOFON^j AA- 
 
 FOKOFON. Nos quoque hiatus causa interponimus U loco (τον) 
 
 digamma F, ut DaJJus, ArgiOi, PaVo, OOum, OVis, BoVis^ — he 
 
 remarks (I) Si locus sanus est, err at Priscianus: nam in iambico di" 
 
 metro, cujus hcec forma est: 
 
 ερώ τε δητα κουκ ερώ — 
 yial μαίνομαι κοΰ μαίνομαι — 
 Phaselus ille quern vides — 
 
 δάΡυον facit longam primam syllabam. Sed scepe apud Homerum cor- 
 repta syllaba est, ut 'Of ίων, "ΟΡιας, &c. (2) Non pro riihilo hie / 
 accipitur, sed pronunciatur δΡειράναν, ut nos possumus dwiranan. 
 (3) No7i ubique et in omnibus JEoli inserebant F, sed in certis tantum 
 verbis, ut ex Homer ο constabit. Et sic Dionysius supra, πολλά non 
 πάντα; et Sergius infra, "m quibusdam dictionibus :" et Terentianus 
 Maurus ''nominum multa." (4) Sic Codex MS. Cantabrigice, non 
 
220 ': APPENDIX TO § 110. [book I. 
 
 ^ημοφόων, unde constat tripodem ilium vetustiorem esse Uteris Simoni- 
 deis^ 7], φ, ^c, ut Homerus quo que scribebat MEN IN AEI^E THE A 
 ΠΕΑΕΙΑΑΕΟ ΑΚΗΙΑΕΟΣ. (5) Constat ex hoc loco JEoles 
 scripsisse, ^ά^ος^Άργεΐ^Οί, τα^ώς, oJ^ov, o/tg, βοΓόο). — Valerius 
 Probus, p. 1507. Sergius Grammat. p. 1827 (where, on the words — 
 j^olenses enim Grceci quibusdam dictionibus, ut pinguescant, istam 
 digammon apponunt, ut pro Elena dicunt Uelena. Sed Donatus hie 
 arguitur^quodapudGrcecos digammos Jianc legem habeat^ ut υ detracta, 
 nihil absit nomini. — Bentley remarks — nihil abest nomini in ceteris 
 dialectis Grcecorum, at apud jEolenses deperit totum: nam illi pro 
 foLVog nunquam dicebant οϊνοξ, non magis quam Latini pro vinum, 
 INUM, aut pro vulgus, ulgus). Velius Longus, p. 2217: Qui igitur 
 illam literam (H) vindicant, ostendunt ejus eandem vim esse quce con- 
 sonantis est; nam et in metro asserit sibi hanc potestatem, Unde et 
 apud Homerum non videntur esse vitiosi versus, qui hac aspiratione 
 supplentur: η ολίγον οί τΐαΐδα εοίκόνα γείνατο Τνδεύς. Et hoc am- 
 plius adeo litera est, ut possit videri etiam vicem duarum consonantium 
 implere, ut Sg είτίων τόξον μεν άπο εο Ο'ηκε χαμάξε. Et tale quidem 
 exemplum apud nos non animadverti. (On which Bentley observes — 
 Velius ille hie fallitur, qui id ascribit aspirationi Η quod debetur τώ 
 f; ολίγον HoL pro ολίγον foi, et άπο Ηεο pro άπο /εο. Sic 
 dedit Homerus : η ολίγον foL παΐδα Γεοίκοτα γείνατο Τνδεύζ. et og 
 J'εLπώv τόξον μεν άπο /εο d -ηκε χαμάξε). Idem, ρρ.^ 2222, 2235. 
 Scaurus Gramm. p. 2254. Annceus Cornutus, p. 2282. Cassiodorus, 
 p. 2292. Terentianus Maurus, pp. 2387, 2397. Marius Victorinus^ 
 ' pp. 2461, 2468. He then proceeds: Claudius Ccesar pro U consonanti 
 scribi jussit f AEolicum, sed inversum ne confunderetur cum V, quod 
 alium tunc sonum habuit. Qualia visuntur in inscriptionibus cevi Clau- 
 diani apud Gruterum, Suetonius in Claudio, c. 41. Tacitus, Annal. 
 XI. 14. Aulus Gellius, xiv. 5. (xvi. 17) Donatus, ad Andr. Terent, i. 2. 
 Although Bentley was well aware that / has occasionally the power of 
 a double consonant in Homer, it never seems to have struck him that 
 the original sound might have been made up of a guttural and a labial, 
 and therefore he presumes thatHesychius must be wrong when he writes 
 a number of digammated words with a guttural. Hesychius sane, he 
 says, ridiculus est quipluribus verbis Solids pro f digamma simplicem 
 γ posuit. His notion was, that the digamma was nothing more nor less 
 than our w, as appears from the following note in this MS : U Latinorum 
 olim pronuntiabatur ut W hodie. U consonantem eandem vim et sonum 
 habuisse quam Ρ JEolicum omnes testantur. Dionysius Ονελία, Ρελία, 
 Uelia, etJulianus OT. Gruter.p. 1027. ΟΚΤΑΟΤΙΟΣ, Octavius, 
 ΣΕΟΤΑΡΟΣ, Severus, ΟΤΕΙΒΙΟΣ, Vibius, ΟΤΙΒΙΑΝΟΣ, 
 Vibianus, ΟΤΑΑΕΡΙΟΣ, Valerius, OVENEPIA, Veneria, 
 
[book ι. appendix to § 110. 221 
 
 et in historicis Greeds nomina infinita. Idem ostendunt nomina qum 
 nos, hoc est, Germani veteres, ex Latinis sumpsimus. Uinum, hoc est 
 Winum, "wine;" Valium, Wallum, "a wall;" Uolo, wolo, "I will;" 
 Uentus, wentus, "wind;" Uectis^ Wectis, "weight;" Uellus, wellus, 
 "wool;" Uidua, Widua, "Widow." 
 
 The greater part of this MS. is filled with a number of lines from 
 Homer, in which digammated words appear. We quote those in 
 which he has introduced any emendations. 
 
 "Αναξ, άνάββω, &c. 
 
 Iliad. VII. 162. ώρτο Λολυ πρώτος μεν aval•,, lego ττρώτόζ γε 
 Fava^, vel πρώτιστα. 
 IX. 73. τΐολεεόΰί ό' άνάβΰεις, lego πόλεβιν 8ε ^ανά(3(5εΐζ, 
 χ. 33. ^Αργείων ηναΰΰε, lego ε^^άναΰόε. 
 XX. 67. εναντα Ποΰείδάωνος ανακτος, forte αναβτάς. 
 XXIV. 449. την Μυρμιδόνες τίοίηΰαν ανακτά^ lego δείξαντο 
 ^τάνακτα. Cf. 452. 
 Od. XIV. 438. κνδαινε δε %νμον ανακτος — versus spurius (Per- 
 son suggests d -νμον d' ευφραίνε ^άνακτοζ). 
 
 Άνδάνει. 
 
 Iliad. VII. 45. βονλην η ρα Ο'εοΐβίν εφηνδανε, lego ^'εοΐς εταΓάν- 
 
 δανε. 
 Od. XVI. 387. εΐ d' νμεΐς οδε μνΟΌξ άφανδάνει. (This seems to 
 have puzzled Bentley: we read ά^^ανδάνει with 
 Passow.) 
 
 "Α6τν; inde ί^αΰτννοος^ J^aurvaXog, J^aurvfava^, ^αΰτνβοωτης. 
 Iliad. III. 140. ανδρός τ ε προτεροιο καΐ αύτεος^ lego τίροτέρον 
 και ^άΰτεος. 
 XI. 732. άμφίύταντο δη αΰτν, lego το Γάΰτυ. 
 XVII. 274. νύκτα μίν είν άγορ^ ΰΟ-ενος εξομεν, αΰτν δε 
 τίνργοι, lego ε|ετε, Γάβτν. 
 
 Εϊδω, video. 
 
 Iliad. XXII. 450. ^εί^τε, δνω μοι ετίεό&ον, ϊδωμ , lego επεΰϋ'ε, Ηδωμ . 
 Od. VI. 160. ον γάρ ττω τοιούτον ϊδον, lego τοίονδε ί^ίδον. 
 IX. 182. εν^αδ' επ' εβχατίη ΰτΐεος εϊδομεν, lego ενρομεν. 
 Qucere de εύιδων-, &c. 
 Έΐΰκω. 
 Iliad. XXI. 332. μάχη ηΐΰκομεν εϊναι, lego ε^εΐΰκομεν. 
 
 379. ου γαρ εοικεν, lego ουδέ Ηοικεν ut ν. 435. 
 Od. IV. 247. φωτι κατακρντΐτων ηϊύκε, lego εΗϊΰκε. 
 
 IX. 321. το μεν αμμες εΐΰκομεν εΐΰορόωντες — αμμε semper 
 casus est accusativi (why not ε^ίΰκομεν'^). 
 
222 APPENDIX TO § 110. [book I. 
 
 XIX. 283. τόγε κερδοον εϊϋατο Ο^νμώ (read τόγε Ρείΰατο κερ- 
 διον είναι). 
 Inde mendosum est d' ψκτο quod quater venit pro δε Γεΐκτο. 
 "Εκαΰτος. 
 
 Iliad XVI. 151. μέγα uQ'hog εμβαΧ εκάΰτω, lego ώρΰε Γεκάΰτω. 
 Od. IX. 468. άνα δ' οφρνβι v&jov εκάβτω, lego νενβα. 
 XV. 377, φάΰΟ'αί καϊ εκαΰτα πν^'έΰΟ'αί, lego άπαντα- 
 XVII. 70. τοί δ εξερέεινον εκαΰτα, lego ατίαντα. 
 Cf, Od. XI. 228. οτίωζ ερεοιμι εκάΰτην et mox 233, εγώ ό' ερεεινον 
 άτΰάύας, ubi Eusthatius ταντόν, Sg καϊ aXkayov^ 
 το εκάΰτην καϊ τας ατίάΰας. 
 Iliad XIX. 302. Πάτροκλον τίρόφαΰιν, ΰφώλν δ' αυτών κήδε 
 εκάΰτη, lego ατίαΰαι. 
 332. δείξ,ειαζ εκαΰτα, lego ατίαντα. 
 Od, VIII. 15. ^υμον εκάΰτον, lego άτίάντων. 
 IX. 127. τελέοιευν εκαΰτα, lego ατίαντα. 
 "Εκ α g. 
 
 Od. VII. 321. μάλα πολλον εκαΰτέρω εΰτ Evβoίηg (read πολλά 
 ^εκαΰτερω). 
 'Eκήβoλog. 
 
 Iliad Ι. 21. άξόμενοϋ ^Cog νίον Έκήβολον Απόλλωνα, lego 
 via Ι^εκήβολον. 
 438. εκ δ' εκατόμβην βηΰαν εκηβόλω "Άπόλλωνι, lego 
 βηΰε. 
 Έλίΰΰω, &C. είλεω, είλύω. 
 
 Iliad XVIII. 522. IV^' αρα τοίγ ΐζοντ είλύμενοι^ lego ίζον^είλνμενοί. 
 XXIII. 320. άφράδεωg επϊπολλον ελίΰΰεται, lego πολλά ^ελίΰ- 
 ΰεται. 
 "Ελπομαί. 
 
 Od. π. 91. πάvτag μεν ρ ελπει, lego μεν Ηλπει. 
 IX. 419. οντω γάρ πον μ ηλπετ ενί φρεΰϊ νήπιον είναι, 
 lego πον Γέλπετ . 
 "Eπog, εϊπω, &c. 
 
 Od. I. 10. %-νγατερ ziiog είπε καϊ ημίν. (The MSS. omit Zliog, 
 and we might substitute ΰν δε', Bentley offers no 
 emendation, but refers to his note on the passage, 
 which we cannot find.) 
 91. μνηΰτήρεΰΰιν άπείπεμεν, lego μνηΰτήρεΰΰ' απο- 
 ^είπεμεν. 
 II. 269.) και μιν φωνηΰαΰ' ϊπεα, legout smpius φωνηΰαΰα 
 VII. 216.) ^^έπεα per crasin. 
 
 II. 331. αντ εΐπεΰκε, lego αν ^είπεΰκε. 
 
 III. 264. ^ελγεΰκ επέεΰΰιν, lego %ελγεΰκε Ηπεΰΰιν. 
 
CHAP, v.] APPENDIX TO § 110. 223 
 
 427. μενετ' avtov άόλλεες εΐτίατε d' θΙ'(?ω, lego άολλεΐς 
 ^είτίατε. 
 IV. 637. αντίκα d' ηγ ετΐέεΰύίν^ lego ηγε ^ετίεΰΰιν. 
 
 159. το πρώτον ετιεύβολίαξ άναφαίνεον, lego τα πρώτα 
 
 Γεπεΰβολίας ut scepe. 
 682. η είπεμεναυ δμωγιβίν, lego η ΡειπεΙν. 
 νπ. 275. ώδ' εΪ7ί^(5υ^ lego ωζ ^είτΐΐ^ύι, 
 
 VIII. 27. Dele τ. 91. lego τέρποντο Ηπεύύιν. 
 
 IX. 224. lego λίΰύοντο/έτΐεβΰί. 258,363. lego ως^επέεΰΰίν. 
 IX. 279. αλλά μου εϊφ οπη εύχες, lego αλλ' αγε ί^είφ . 
 
 XI. 296. %ε6φατα τίάντ είπόντα^ lego ^οι ^είτΐοντα, 
 
 560. αλλ' αγε δενρο αναξ ΐν έπος καΐ μν%Όν άκονό^ς^ 
 lego Lva που κα) — . 
 Iliad VII. 349. οφρ εϊπω, lego ώς ^είπω. 
 IX. 61. ε^είπω, lego εκλείπω. 
 
 376. lego εξαπάφοι ^επεεΰόι' ^άλΐξ δε /θ6 * άλλα ^έκηλος. 
 χ. 425. ενδονβ' η άπάνεν^ε; δίειπέ μοί, lego δια^είπεμεν. 
 XI. 790. lego τα ^είποίξ. 
 
 XV. 398. όλοφνρόμενος δ' έπος ηνδα, lego δε προΰηνδα. 
 Ι. 555. δείδοικα κατά φρένα μη ΰε παρείπγ, lego μη 
 παρ^είπ-^. 
 XIX. 35. lego μηνίν άπο^είπων'Αγαμεμνονί. 
 "Εργον"'. 
 
 Iliad XIX. 245. γυναικός άμυμονος εργ είδυίας, lego άμύμονα 
 Ι^έργα ^ίδυίας. 
 Od. XIV. 344. εϋδείελου έργα, lego ενδείελα ίεργα. 
 
 XI. 473. ετιμεΐξον ένϊφρεϋΐ μή(^εaL•εργov,lego μήΰαοΡέργον- 
 χνπ. 313. ημεν δέμας ηδε καϊ έργα, lego ηδέ τε /έργα vel τι. 
 XXII. 422. τας μέν τ έργα δυδάξαμεν έργάζεΰ%'αί, lego μεν 
 /έργα δίδάξαο Γερ^άξεΰ^αί. 
 Έρύω. 
 
 Od. χ. 402, 422. νηα μεν αρ πάμπρωτον έρύβΰατε, lego πάμπρωτα 
 /ερύόβατε. 
 XVI. 348. αλλ' αγε νηα μέλαιναν έρύΰβαμεν ητίζ αρίύτη 
 (perhaps we may read αλλ' αγε δη καϊ νηα 
 /ερύΰόαμεν; at any rate the epithet μέλαιναν 
 seems unnecessary). 
 Ίάχω, ιαχή, &c. 
 
 Od. II. 428. μέγαίί ϊαχε, lego μέγα Ηαχε. 
 
 IV. 458. ημείς δ' αΐφ' Ιάχοντες (omit alif) , and read ^έ 
 /ίάχοντες). 
 
 * We may add Iliad vi. 289; where we should read iv&' tcav oi 
 πέηλοι, παμηοίκιλα ^Ιέ^γα γυναικών , the old reading παμποίκιλοι being 
 objectionable on syntactical grounds also. 
 
224 APPENDIX TO § 110. [book I. 
 
 IX. 395. ΰμερδάλεον δε μεγ ωμωξεν τΐερί δ' ϊαχε τΐετρη (read 
 ΰμερδάλεον δ' ωμωξε, ΛερΙ δ' εΛί^ίαχε πετρη). 
 Iliad XIII. 835. ^Αργεΐου δ^ ετερω^εν επίαχον, forte έ^ίαχον vel 
 -ρωΟ"' ετΐί^ίαχον. 
 'l6og. 
 
 Iliad XXIII. 736. άεΟ'λοα ό' ΐΰ' άνέλοντες, lego αε^λα δε H(j . 
 Οίδα. 
 
 Od. Ι. 428. τίεδν είδνΐα, lego κεδνα ^ίδνΐα. 
 
 II. 111. ντΐοκρίνονται,, ίνα είδγις, lego -νονΟ'' ίνα ^είδ^ς. 
 IX. 534. ονκ είδότ , lego ου J^εL•δότ . 
 XI. 432. λνγρ είδνία, lego λνγρα Γίδνΐα. 
 XV. 417. καΐ αγλαα εργ εΙδνΐα, lego αγλαα ^εργα Ηδνΐα. 
 Οϊκος. 
 
 Od. XIII. 42. άμνμονα δ' οϊκοι ακοίπν — an ένδον? 
 XIV. 318.) r , τ J s^r ^ 
 
 XVII. 84. ^^^^' ^^ Οίκον, lego δομονδε. 
 
 χπι. 121. ώτίαΰαν οΧκαδ^ Ιόντυ, lego ^οίκαδ^ οτίαΰΰαν Ιόντι. 
 Iliad Ι. 19. ευ δ' οϊκαδ' ίκεύ^αι, lego ευ δ' "Αργοξ ίκεύ^αι. 
 And in a note on this passage he says — Homer ο 
 semper est Ιοΐκοξ, Ι^οίκαδε. Ergo hie scribendum 
 ευ δ' Ιοίκαδ\ DWOIKAz;' ut Anglice DWELL. 
 Priscianus: est quando in metris pro nihilo acci- 
 piehant, ut"Aμμεζ d' Ιειράναν. 
 Olvog. 
 Iliad VII. 467. τίαρεόταύαν olvov άγοντες, lego τίάρεβύανίοϊνον. 
 IX. 224. τΐληΰάμενος δ' οϊνοω δετΐας, lego πληΰάμενος 
 foLVOLO vel πληΰας δε Ιοίνοιο. 
 Od. XX. 255. εωνοχόευ δε Μελαν%'ε6ξ, lego είοινοχόει. 
 
 He also quotes the following exceptions, without proposing any 
 emendations: Iliad xviii. 545; Od. ii. 46. μελίΐήδεος Ιοίνου. Od. xi. 
 61. ά\^ε6φατοζ lolvog. 
 Oi, sibi, £, se, εο, sui. 
 
 Iliad V. 338. ov oi χάριτες κάμον αύται. Versus spurius. 
 VI. 90. Ίίετίλον og oi δοκεεί, lego ο lot. 
 101. μαίνεται ου δ ε τις οι, lego οϋτις Ι οι. 
 "Ο ς, SUUS, ην, ώ, &C. 
 Iliad XIII. 561. Άϋιάδην og οι οϋτα, lego ο /oi. 
 
 ' XIV. 407. οττι δη οι, lego ut MSS. οττι ρά foi. 
 XIX. 384. ϋτειρή&η δ' εο αι^τοι; (read δε Ιε αυτού). 
 
 The following passages are quoted without emendation : Iliad xx. 
 282. αχός oi χύτο (read άχλυς χύτο). xxiv. 73. ή γάρ foi. Od. ix. 
 360. αϋτάρ loi αϋτος εγώ. 
 
CHAP, v.] APPENDIX TO § 110. 225 
 
 Iliad VI. 474. αντάρ ογ ov φίλον υΐον, lego ανταρ ο fov. 
 χπ. 162. ωμωξεν τε καΐ ω Λεπλήγετο μηρώ, lego ωμωξεν 
 
 κ«1 εώ τίεπλ. μ. 
 XI. 330. ονδε εονς τΐαΐδας εαβκε, lego ούδε fovg (or 
 rather βφονς). 
 Iliad XVII. 90, xvm. 5, xx. 343. εΪΛε προς ov μεγαλψορα &υ- 
 
 μόν, lego εφη προς fov (or rather ΰφόν). 
 Od. IV. 4. %υγατροξ άμνμονος ω ενί οίκω, lego άμτψονα /ώ ad 
 γάμον. 
 ^ίΐξ, tamquam. 
 
 Iliad III. 196. αϋτοξ δε, κτίλος Sg, lego ipikog εών έταπωλεΐται, 
 and in a note on the passage he says — Quot sunt 
 in hoc versiculo peccata, licet ita citatus a vetere 
 Scholiaste Nicandri, et Timone de Cleanthe apud 
 Diogenem Laertium in Cleanth. rig δ' ovtog, κτί- 
 Xog cig, ετίίτίωλεΐται ΰτίχας ανδρών. Primum ex 
 j^olismo: oportet enim, — ut semper Sg "sicut," 
 ''• tamquam^^ — κτίλος /ω^ esse, metro repugnante. 
 Quale vero illud ^'-obit ordines virorum tamquam 
 Aries^^f Nondum vidi Arietem virorum ordines 
 moderantem. Quce vero ταυτολογία! '^Obit, 
 tamquam Aries; et comparo eum Arieti." Ex 
 ipsa sententia locum restituo. Versu prior e dixerat, 
 ^'Arma ejus humi posita sunt:'''' quorsum hoc^ 
 nisi ut inferret inermem eum obire ordines mili- 
 tumi Lego igitur 
 αυταρ i\)vXog εών ετίίπωλεΐται ύτίχαζ ανδρών. 
 Sic δ, 230, simili orationis filo : 
 
 LTtJtovg μεν γαρ εαΰε καΐ άρματα ποικίλα χαλκω — 
 αυταρ δ πεζός εών ΙπιπωλεΙτο ΰτίχας ανδρών, 
 et ξ, 214: 
 
 εγχος μεν κατεπηξεν επϊ χ^ΌνΙ πουλυβοτείρΊ;], 
 αυταρ 6 μειλιχίουβι προΰηύδα ποιμένα λαών. 
 
 Ceterum nihil refert quod est αυταρ 'ψιλός non 6 'ψιλός, ob aspe- 
 ritatem duplicis liter ce. Sic enim supra γ, 18, αυταρ ο δουρε δυω: 
 meliores εκδοθείς habuerunt sine articulo αυταρ δοϋρε. 
 
 It is hoped that these extracts will gratify any curiosity which 
 may still be entertained with regard to the manner in which Beutley 
 proposed to restore the digamma to the text of Homer. 
 
 Q 
 
CHAPTER VI. 
 THE PARTS OF SPEECH. 
 
 123 Different arrangements of the parts of speech. 124 Their syntactical di- 
 vision derived from the Dialectic of Plato. 125 Aristotle's Categories con- 
 sidered with this reference. 126 Home Tooke's fallacious use of the syn- 
 tactical parts of speech. 127 The empirical arrangement is mainly syn- 
 tactical. 128 The real etymological distribution of words. 
 
 123 TT has been already mentioned more than once, that 
 there are two divisions of philology , the etymological 
 and the syntactical , and that it is of great importance to keep 
 distinct these two departments. The distribution of words into 
 the parts of speech, as they are called, has been a fruitful 
 source of error to those philologers who have failed to observe 
 that there are two distinct methods according to which this dis- 
 tribution may be effected, the one syntactical, and the other 
 etymological; of which the former considers words only accord- 
 ing to their distinction as parts of a logical proposition , while 
 the latter analyzes the words themselves, and sets forth the 
 primary elements from which the different kinds of words have 
 sprung. There is a third method based on the former of the 
 two which we have just mentioned; but as its object is merely 
 to facilitate the acquirement of particular languages, and as it 
 differs with those languages , it has never been thought worthy 
 of discussion in formal treatises. 
 
 We propose to examine the syntactical arrangement of the 
 parts of speech, before we set forth that etymological distribu- 
 tion, according to which the investigations in the following 
 pages are carried on. 
 
 124 The syntactical division of the parts of speech may be 
 traced to the first beginnings of dialectic or logic, in other words, 
 to Plato. The formation of a system of logic is, in fact, simply 
 a discovery of the principles of syntax , or of the structure of 
 sciitences ; for, as far as the reasoning faculty is concerned, logic 
 is nothing but the nomenclature and method of the process which 
 
CHAP. VI.] THE PAETS OF SPEECH. 227 
 
 every man carries on in his discourse. Logic is conversant with 
 the truth or falsehood of propositions and not with single words 
 (Aristot. de Interpret, I.; Cicero, Tuscul. Disputt i. 7; Aulus 
 Gellius, XVI. 8). The first step , therefore , in logical analysis, 
 is the division of a sentence or proposition into its fundamental 
 parts. These fundamental parts we call the subject, copula, 
 and predicate; in other words, the proposition must contain 
 either a nominative case -f- verb-substantive -\- some predicate, or, 
 a nominative case + (verb —verb-substantive -f- some predicate). 
 Thus "I run" is equivalent to "I am running." The Greek, 
 however, does not make much use of the copiila, the article 
 being considered sufficient to distinguish the subject from the 
 predicate ; thus ό iJtJtog λευκός is fully equivalent to ό ΐτΐτιοζ 
 ΒΰτΙ λευκός. In by far the greater number of cases the Greek 
 verb contains both copula and predicate. Accordingly, it was 
 natural enough that, in analyzing the sentence into its primary 
 elements, Plato should consider these as consisting simply of the 
 noun (όνομα) and the verb (ρήμα) *, for as Plutarch observes 
 (Qucestiones Platonicce, p. Ill Wyttenb.): ρήματος ονόματι όυμ- 
 τΐλεκομενου, το γενόμενον εύΟ'υς δι,άλεκτός εΰτί καΐ λόγος ^ 
 and Apollonius Dyscolus says (de Syntaxi, p. 19 Bekker), that 
 the noun and verb are τα εμψυχότατα μέρη του λόγου. 
 Plato brings forward this division most directly in his SopMstes 
 (p. 261 Ε — 262 c): εβτί γαρ ημΐν τίου των ry φωνγ] τίερί 
 την ούΰίαν δηλωμάτων διττον γένος — το μεν ονόματα^ χο 
 δε ρήματα κλη^εν — το μεν ετά ταΐς πράξεΰίν ον δήλωμα 
 ρημά τίου λεγομεν — το δε γ lit! αυ τοις εκείνα ττράττουΰι 
 ΰημεΐον της φωνής ετίίτε^^εν όνομα, ούκοϋν εξ ονομάτων μεν 
 μόνων ΰυνεχώς λεγομένων ουκ εύτι Λοτε λόγος οϋδ' αϋ 
 ρημάτων χωρίς ονομάτων λεχθέντων — ουδεμίαν γαρ — τΐράξιν 
 
 * That the ancient Greeks did not make much distinction between 
 όνομα and ρήμα, is clear from Thucydides, v. Ill, where we find the words 
 used as synonymous in an emphatic passage: όνόματοξ επαγωγού δυνάμει 
 — ήΰΰηϋ'εΐβί τον ρήματος. We remark in passing that for ην—εοται at 
 the end of ihe chapter, we ought to read ην—Ϊ6τατε. The MMS. have 
 Ι'ατε and ιαταται. Plato himself sometimes uses ονόματα and ρήματα as 
 synonyms; cf. Gorg. γ. 4iSd d: ονόματα Χηρεύων. 489ε: ονόματα λέγεις. 
 In the singular he uses το ρήμα to signify "common parlance:" Resp. 
 340 d: λέγομεν τω ρήματι όντως. p. 490: ον ^jy'jiiOfra θηρεύω. 
 
 02 
 
228 THE PAETS OE SPEECH. [bOOK I. 
 
 ovd' άτίρα^ίαν ονδε ονοίαν δντος ονδε μη ovrog δηλοΐ τα 
 φωνη^εντα, τίρίν αν τΐξ τοΐξ ονόμαΰι τα ρήματα κεράΰτί^' 
 τότε δε ηρμούε τε %αΙ λόγος εγενετο εν&νς η ΐίρώτη βνμ- 
 πλοκή, ΰχεδον των λόγων 6 τΐρώτος εΐ καΐ ύμικρότατος. It 
 must be remarked on this passage, that Plato included in the 
 word ρήμα all that could be called a predicate or descriptive 
 word, as distinguished from the subject or word of designation, 
 namely the verb and the adjective as distinguished from the 
 substantive; for he could not have overlooked the obvious fact, 
 that in the Greek language a verb may alone constitute a whole 
 sentence: thus, τρεχεο means "he is running." Similarly in 
 the Cratylus (p. 431 b), he speaks of falsifying the meaning of 
 ονόματα and ρήματα as the same process: εϊη αν καΐ ρήματα 
 ταντον τοντο ποιεΐν, and adds: εΐ δε ρήματα και ονόματα εΰτιν 
 οντω τι^εναυ, ανάγκη καΐ λόγους' λόγοι γάρ tcov, ως εγωμαι, 
 η τούτων ^νν^εβίς εΰτιν. That the adjective is regarded as 
 predicable because it is generally descriptive is clear from ano- 
 ther passage in the Cratylus, in which he says (p. 399 b), that 
 we change ^ii φίλος into ζίίφιλος, ίνα αντί ρήματος όνομα 
 ημΐν γενηται: for ^ίφιλος is properly a subject, and ζί it φίλος 
 is calculated to be a predicate. Some logicians according to 
 Plutarch (Qucest. Platen, p. 108) substituted κατηγόρημα — "pre- 
 dicable" — for ρήμα, in this division of the λόγος, or logical 
 proposition, into its distinct parts. Did Plato make this divi- 
 sion, he asks, on ττρώτον λόγον οι %αλαιοί, την τότε καλον- 
 μενην τΐρόταΰιν, νυν δε αξίωμα, τίροβηγόρενον , ο τίρώτον 
 λέγοντες άληΟ'εύονΰιν ή φενδονται; τοντο δε εξ ονόματος κα\ 
 ^ijfiarog ΰυνείίτηκεν; ών το μεν τιτώ 6ιν οί διαλεκτικοί, το δε 
 κατηγόρημα καλονΰιν. That this Platonic analysis of the 
 sentence into its two main elements, the subject and predicate, 
 was accepted as a sufficient classification of the parts of speech, 
 is distinctly stated by the Pseudo- Apuleius (de dogmate Platonico.^ 
 Lib. III. 267): "Ceterum est propositio, ut ait in Thewteto 
 [Sophista] Plato, duabus paucissimis orationis partibusconstans 
 nomine et verbo, ut: Aptdeius disserit, quod aut verum aut 
 falsum est, et ideo propositio. Unde quidam rati sunt has duas 
 solas orationis partes esse, quod ex his solis fieri possit perfecta 
 oratio, id est quod abunde sententiam comprehendant — Porro ex 
 duabus prgedictis partibus altera subjectiva nominatur velut sub- 
 
CHAP. VI.] THE PARTS OP SPEECH. 229 
 
 dita, ut Apulems; altera declarativa ut: Disserit vel Non disseint, 
 declarat enim quid faciat Apuleius." 
 
 125 Aristotle, following in the steps of his master, adopted the 
 same division of the parts of a sentence. He says {de Interpret atione, 
 c. 1 — 5): TtQOtov δει ^ίύ^αιτί όνομα καΐ τι ρήμα, έπειτα τι εΰτιν 
 άτίόφαΰίς καΐκατάφαΰις καΐ άτΐόφανΰις καΐ λόγος — τα μ^ν ονν ονό- 
 ματα αντα καΐ τα ρήματα εοικε τω άνευ ΰνν^εΰεως καΐ διαιρεόεως νοψ 
 ματίοίοντο ανΟ'ρωτίος ή το λενκόν. — όνομα μεν ονν εότΐ φωνή 
 ϋημαντική κατά ΰννΟ^ήκην ανεν χρόνου Tjg μηδέν μέρος εΰτΐ όημαν- 
 τικον κεχωριΰμενον — ρήμα δε εότι το τκρούβημαΐνον χρόνον , ου 
 μέρος ουδέν ΰημαίνεο χωρίς, καϊ εΰτυν άεΐ των κα^ ετέρου λεγο- 
 μένων ΰημεΐον — λόγος δέ έΰτι φωνή Σημαντική κατά ύυνΟ'ήκην ης 
 των μερών τι ύημαντικόν έύτι κεχωρί(5μένον ως φάΰις, αλΧ ουχ ως 
 κατάφαύις ή άπόφαΰις — εότι δε εΙς πρώτος λόγος αποφαντικος κα- 
 τάφαΰις, είτα άπόφαύις' οί δε άλλοι πάντες ΰυνδέΰμω εϊς. Here 
 again it is clear that λόγος is the logical proposition , όνομα the sub- 
 ject, and ρήμα the predicate — "the sign of things predicated of another 
 thing;"— and that the ρήμα includes adjectives as well as verbs, 
 appears, as well from this place in which λευκόν is given as an instance 
 of a ρήμα, as from another passage in the same treatise (c. 14): ^ε- 
 τατι^έμενα τα ονόματα και τα ρήματα ταντον ύημαίνει, οίον εΰτι 
 λευκός άνθρωπος, εΰτιν άνθρωπος λευκός. The ρήμα is a predicate 
 broken off from an actual sentence, and it is only in this opposition 
 that it gets its distinctive name: otherwise it is merely an όνομα as 
 Aristotle says {de Interpret, c. 3): αϋτα μεν οϋν κα%^ εαυτα 
 λεγόμενα τα ρήματα ονόματα έΰτι"^. The philosopher has ex- 
 
 * This passage alone is sufficient to justify Ammonius from the criti- 
 cisms of Mr. Mansel, who says (in an article which he has elsewhere 
 acknowledged, North British RevieWyYol. xiv. p. 53): "Ammonius, comment- 
 ing on De Interpretatione , chap, i, supposes that Aristotle includes under 
 the name of verb an adjective in the predicate of the proposition, i. e. the 
 mere expression of an attribute without assertion ; and this has led Harris 
 to speak of the 'verb in its most comprehensive signification, as including 
 not only verbs properly so called, but participles and adjectives.' But 
 the explanation is erroneous. The qij^a of Aristotle has one uniform 
 signification, that of a combination of attribute and assertion — the predi- 
 cate and copula united. In the passage misunderstood by Ammonius, the 
 word Isvyiov is, by a common idiom, put for Xevnov έΰτί." With all his 
 acuteness Mr. Mansel has failed to observe that, according to the Greek 
 writers on logic and grammar, the ρήμα has no predicative value except 
 in the αυμπλοκή with the όνομα as subject; and the use of the definite 
 article, which limits the latter, enabled the Greeks to dispense with the 
 copula whenever the predicable word was sufficiently distributed; espe- 
 cially in the case of those expressious which were calculated of them- 
 selves to be used as adverbs or secondary predicates. 
 
230 THE PABTS OF SPEECH. [boOK I. 
 
 plained his meaning with regard to the predicate more fully in his 
 treatise on the Categories : as this little work has never been properly 
 understood*, it will he as well to point out its object. It commences 
 by distinguishing homonyms, synonyms^ and paronyms. Two things are 
 synonymous when they agree not only in name but also in nature, as 
 when we say that a man and a horse are both animals ; but they are 
 homonymous when they agree in name only, as one might say that a 
 horse and the picture of a horse were both animals, and that the man 
 who painted the latter was an animal-painter, although the real defi- 
 nition of the horse and its picture would not coincide. Hence Aristotle 
 speaks of ra άτΐο τνχης ομώνυμα (EtJi. Nic. i, 6. § 12), when words 
 are used in an equivocal or ambiguous sense, or when there is merely 
 an accidental coincidence in nomenclature , as when the word xXslg is 
 used ομωννμωζ to denote either the clavicular bone in animals , or 
 that with which we lock doors (Eth. Nic. v. 1, § 7). On the other hand 
 he says (Ibid. v. 2, § 6) that there is a particular kind of injustice 
 distinct from injustice in general, but having the same name , because 
 it falls under the same genus (ΰννώννμος, oxl δ OQioabg Iv τω αντώ 
 γένει). Words are paronymous when they differ only in inflexion : thus 
 "courageous" is a paronym of "courage." Aristotle proceeds: "Some 
 words are spoken in connexion (%ata ύνμτίλοκήν) ; others separate^. 
 Examples of the former case are — ανΟ'ρωτίος τρέχει, άνθρωπος νι,κα : 
 of the latter, ανΰ'ρωτΐος, βονς, τρέχει, νίκα.'' After distinguishing 
 between those things that are said of, and those things that exist in, 
 a subject (το δτίοκείμενον), and particularly asserting that individuals 
 (τα άτομα) cannot be said of a subject, he (c. 4) enumerates ten classes 
 of those words which are spoken separately ; they are thus described : 
 των κατά μηδεμίαν ΰνμτίλοκψ^ λεγομένων εκαύτον ητοί ονΰίαν 
 βημαίνει tj τιό^ον ij Λοών η Λρός τι η κεΐύ^'αί η εχειν η τίοιεϊν ij 
 τΐάΰχευν. εΰτι δε ονβία μεν ώς τνπω είτιεΐν οίον αν^ρωτίος, ΐτντΐος ' 
 τίόΰον δε οίον δί^ηχν, τρίτΐηχν' tcolov δε οίον L•vκόv, γραμματικόν - 
 Λρόζ τι δε οίον δίΛλάΰων. ημιΰν, μείζον * ποϋ δε οίον εν Αυκείχρ, 
 εν αγορά' τίοτε δέ οίον εχ%ές, τίέρνύιν κεΐύ^αι δε οίον ανάκευται, 
 κά^Ύΐται' εχειν δε οίον ντΐοδέδεται, ωτίλιΰταΐ' τίοιεΐν δε οίον τέμνει, 
 καίει' Λαΰχειν δε οίον τέμνεται, καίεται, εκαύτον δε των είρημένων 
 αντο μεν καθ"' αντο εν ουδεμία καταφάΰει λέγεται η άτίοφάΰει, ty 
 δε Λρος άλληλα τούτων ΰυμπλοκί] κατάφαόις η άπόφαΰις γίγνεται. 
 Now it is sufficiently obvious from these last words, that the ten sorts 
 of words thus described do not mean predicates, but simply the diffe- 
 
 * Since the above was first published the Categories have been fully dis- 
 cussed by Dr. Adolph Trendelenburg, in his Geschichte der Kategorien- 
 lehre, Berlin, 184G, p. 384. His conclusions are not materially different from 
 those given in the text, but he has gone into all the details of the subject. 
 
CHAP. VI.] THE PAETS OP SPEECH. 231 
 
 rent parts of a sentence, whether subject or predicate; for it is by the 
 joining of these with one another, that the sentence, whether affirmative 
 or negative, is to be formed*. Aristotle institutes a more particular 
 examination of the first four, which have descended to us from the 
 scholastic philosophy under the names substance (or quiddity), quan- 
 tity^ quality-) and relation: the other six he has hardly illustrated at 
 all. If we take a general view of these categories , according to the 
 instances which Aristotle has given, we shall see that this is merely a 
 grammatical or rather syntactical arrangement of certain parts of 
 speech: the first category includes nouns substantive, the three next, 
 different sorts of adjectives, the fifth and sixth, adverbs of place and 
 time, and the last four, verbs, considered as active (9th), passive 
 (10th), intransitive (7th), and in the perfect tense, or as representing 
 the effect of something which has been done or has occurred (8th) ; the 
 Greek perfect cannot be considered as denoting merely past time, which 
 is included in the fifth category. This is of course a very rude approxi- 
 mation to a scientific division, the number ten being in all probability 
 borrowed from a similar classification among the Pythagoreans. The 
 object of the philosopher in enumerating these classes is shown by his 
 subsequent explanation: these ten sorts of words do not in them- 
 selves constitute either an όνομα or ρήμα as distinguished from one 
 another, but only when they can be considered as general terms; for 
 instance, it is only as a synonym that substance can become a predicate; 
 in other words, the τίρώταί οϋΰίαι, or individuals cannot be predicated, 
 but only the δεντεραί ονύίαι, or genus and species-, thus he says: ta 
 εϊδη καΐ τα γένη μόνα δηλοΐ την τΐρώτην ονβίαν των κατηγορου- 
 μένων , and ντίάρχει ταϊζ οϋόίαΐξ καΐ ταϊ^ δίαφοραΐς το τίάντα 
 ύνϋωννμωξ α% αυτών λεγε6^αι' τιάβαι γαρ αί απ αυτών 
 κατηγορίαι ήτοι κατά τών ατόμων κατηγορούνται η κατά 
 τών ειδών, άτΐο μεν γαρ της πρώτης ούΰίας ουδεμία εΰτΐ κατη- 
 γορία' κατ ουδενος γαρ νποκεψένου λέγεται' τών δε δευτέρων 
 ούΰίών το μεν είδος κατά του ατόμου κατηγορείται, το δε 
 γένος καΐ κατά του εϊδους κο:1 κατά του ατόμου, and similarly 
 of the differences (δίαφοραί): ώΰτε τίάντα το oltco τών ουβιών 
 
 * Dr. Trendelenburg, u. s. p. 19, seems inclined to refer the first ca- 
 tegory or substance to the subject, the others to the predicate. It appears 
 from the important passage from the Analyt. Post. i. 22, p. 83, a, i, which 
 he quotes in p. 15, that the word κατηγορεΐν did not in Aristotle's sense 
 truly and properly apply to any predications except those contained in 
 general, abstract, and distributable words. Thus it is a true κατηγορία 
 to say, το ξνλον εβτί λεντιόν, but το λεναόν εβτι ξνλον is either μηδαμώς 
 τιατηγορεΐν or •αατηγορεΐν μεν μη απλώς, γ,ατά βνμβεβηκος δε ν,ατηγορεΐν. 
 But this does not prevent substance from being predicable in the abstract: 
 thus Σωκράτης ην ανϋ'ρωπος is as good a proposition as ό ανϋ'ρωπος 
 ην ΰοφός. 
 
232 THE PAETS OF SPEECH. [bOOK I. 
 
 και τα άπο των διαφορών ΰννωννμωξ λεγεταυ {Categ. 5). Hence, 
 κατηγορία is elsewhere used by Aristotle to signify genus ^ even as 
 distinguislied from species; tlms he says (de Partibus Animalium, i. 
 c. 1, p. 639, 1. 29): έτερα δε ϊΰως εότίν oig όνμβαίνει, την μεν κα- 
 τηγορίαν εχειν την αντην, διαφερειν δε τ]] κατ είδος διάφορα, 
 οίον η των ζώων πορεία' ον γαρ φαίνεται μία τω εϊδει' διαφέρει 
 γαρ Ίίτηύις και νενΰις και βάδιβις καΐ ερ'φις. And the categoremata 
 or predicables, as supplementary to and descriptive of the categories 
 or predicaments, are the universale : ορός, γένος, είδος^ διαφορά, 
 ϊδιον καΐ ΰνμβεβηκός {Topica, ι. c 6). So then Aristotle's treatise on 
 the categories is a first attempt to consider which of the \vords that 
 appear in a simple sentence {λόγος) may form the predicate or ρήμα of 
 that sentence, and neither he nor Plato meant to say that όνομα and 
 ρήμα were parts of speech in the etymological sense, nor had they any 
 other object than to lay the foundations of a system of logic, which 
 of course depends on the syntax of individual propositions. We must, 
 therefore, be careful to distinguish from this logical division of the 
 simple sentence, the distribution of words into the parts of speech, 
 also commenced by Aristotle, and completed by the Stoics and later 
 grammarians. This distinction is pointed out by Plato and Aristotle 
 themselves in their discrimination of λόγος and λε^ις. The former is 
 a logical sentence, the όνομα and ρήμα, — κατά ΰνμτΐλοκήν; the latter 
 is the whole outward form of language, whether expressed by articulate 
 sounds or in writing; or, to use the words of Ammonius Hermias (on 
 Aristotle de Interpret, p. 9 9 Brandis) : λόγο ν μεν ovv ταντα (the parts 
 of speech as they are called) ov μέρη, λέξεως δε μέρη ης καΐ ο λόγος 
 αντος μέρος, κα&άττερ εν τοΓ^ τίερί Ποιητικής (c 20) εϊρηται. δια- 
 φέρει δε 6 λόγος της λέξεως, οτι ό μέν Ιβτι τίληρωμα τΐροηγονμένως 
 των ΰημαινονύών τα τιράγματα φωνών, η δε τίαΰών των παραλαμ- 
 βανομένων εις την διαλεκτικήν. έχεις δε τον λόγου τηι* προς την 
 λέξιν διαφοραν και νπο Πλάτωνος εν τω τρίτω της Πολιτείας ττα- 
 ραδεδομένην (ρ. 392 c) εν οίς φ7]6ί "r« μεν δη λόγων τΐερί έχέτω 
 τέλος, το δε λέξεως, ώς έγώμαι, μετά τοντο όκετζτέον, και ημΐν α 
 τ ε λεκτέον και ώς λεκτέον παντελώς έβκέψεται.'' δι ων δηλός εΰτι 
 λόγον μεν την διάνοιαν καλών, λέξιν δε την άπαγγελίαν. This 
 also appears from the words of Plato (Politicus, p. 277 c): γραφής 
 δε και ξνμπάόης χειρονγρίας λέξει καΙ λόγω δηλονν πάν ζωον 
 μάλλον τρέπει τοις δνναμένοις επε(5%αι. In the passage oithQ Poetics, 
 referred to by Ammonius, Aristotle, if he has not been interpolated 
 here (see Ritter, pp. 221 sqq.), divides λέξις into the following parts; 
 the letter (6τοιχεΐον% the syllable (ύνλλαβή), the conjunction (6vv- 
 δεΰμος), the noun (δνομα), the verb (ρήμα), the article (αρ%^ρον), the 
 inflexion (πτώοις) , and the sentence (λόγος). From the explanations 
 
CHAP. VI.] THE PAETS OF SPEECH. 233 
 
 which follow, it appears that noun and verb are here used in the mo- 
 dern signification, that λόγος does not here mean a logical sentence 
 only (ov γαρ ατΐας λόγος εκ ρημάτων ΰύγκΗταή, but any set of words, 
 a definition for instance (oiov δ τον άνΟ'ρώτΐον δριΰμός), and that 
 τΰτώΰίς does not mean merely the case of a noun, but any inflexion of 
 a noun or verb. For instance, the distinctions of words noticed by Pro- 
 tagoras were only inflexions or χτώβΗξ^. It is clear that the only 
 parts of speech, according to the meaning which we attach to the term, 
 here mentioned by the philosopher, are the noun, article, conjunction, 
 and verb ; and Anaximenes (if Spengel has rightly attributed the trea- 
 tise to him) specially mentions the three former in the so-called Bhe- 
 torica ad Alexandrum (c. 25). Now it is stated by Dionysius of Ha- 
 licarnassus {de Compositione Verborum, c. 2, de Demosthenis Prcestan- 
 tia, p. 1101, Reiske) and Quintilian (i. 4, § 18), that Theodectes and 
 Aristotle acknowledged only three parts of speech, the verb, the noun, 
 and the conjunction; and that the article was subsequently added by 
 the Stoics. It appears then that Quintilian, and Dionysius, whom he 
 copies in this place, paid attention only to the passage of the de Inter- 
 pretatione quoted above , in which it is^said that the logical sentence 
 consists merely of όνομα and ρήμα, and that the different propositions 
 are united by means of the conjunction {βννδεΰμος), so that they also 
 have taken merely the logical division. The separation of the article 
 is also due entirely to its logical importance in the Greek language ; 
 every Greek word , however general , may be rendered by the article 
 so individual and definite, as to be adapted to form the subject of a 
 proposition: in fact, the whole distinction between the subject and 
 predicate in many cases is, that the former has, and the latter wants, 
 the article. But, although Aristotle felt this importance of the arti- 
 cle, he does not appear to have said any where that it was a part of 
 the λόγος, and therefore Dionysius and Quintilian are right in attri- 
 buting its addition to the Stoics, if, as we suppose, they were speaking 
 of the logical division. 
 
 126 "We have stated thus minutely the origin of the syntactical 
 parts of speech, because it has not been pointed out before, so far as 
 we know, and in order that our readers may more easily detect the 
 fallacy, by which Home Tooke has, perhaps designedly, built up his 
 whole system of etymology on this syntactical distribution of lan- 
 guage. The resolution of the sentence into subject and predicate, or, 
 what generally comes to the same thing in Greek, into the noun and 
 
 * He is said to have been the first to distinguish the different moods 
 of verbs (Aristot. Poet. c. 21; Quintil. iii. 3, § 10; Diogen. Laert. ix. 53; 
 Suidas, Πρωταγόρας) and the genders (Aristot. Rhet. in. 5, § 5), for which 
 last he is ridiculed by Aristophanes, Ntib. 656 foil. 
 
234 THE PAETS OF SPEECH. [bOOK I. 
 
 the verb, was undoubtedly of great importance at the time when it 
 was first effected: it was the beginning of a syntactical analysis of 
 language, when etymology, or the doctrine of the matter and form of 
 language, was not and could not be in existence. But to make this 
 arrangement the basis of etymology, and to deiive the elements of the 
 word from the elements of the sentence, is a mode of proceeding 
 which can only lead to error and confusion. What then shall we say 
 of a modern philologer, who not only reproduces this old logical 
 division of the parts of speech, applying, however, to the words όνομα 
 and ρήμα the limited signification of noun and verb, which they had 
 only as a part of λεξίς in the method of Aristotle, but has even made 
 this division the basis of a system of etymology, virtually supposing 
 that language was formed according to that system of logic, which 
 only the mighty genius of Plato and the unfailing subtlety of Aris- 
 stotle succeeded in extracting from the clearest and most syntactical 
 language ever spoken by man? And this is just what Home Tooke 
 has done. His system of ultra-nominalism rests upon the hypothesis, 
 which his contemporaries incautiously allowed^ that the two primary 
 sorts of words, from which all others are derived, were the nouns and 
 the verbs. "In English and in all languages," says he (Vol. i. p. 45), 
 "there are only two sorts of words which are necessary for the com- 
 munication of our thoughts; and they are (1) Noun and (2) Yerb, 
 In the strict sense of the term, no doubt both the necessary words 
 and the abbreviations are all of them parts of speech ; because they 
 are all useful in language, and each has a different manner of signi- 
 fication. But I think it of great consequence both to knowledge and 
 to languages, to keep the words employed for the different purposes of 
 speech as distinct as possible. And therefore I am inclined to allow 
 that rank only to the necessary words : and to include all the others 
 (which are not necessary to speech, but merely substitutes of the first 
 sort) under the title of abbreviations ^ Proceeding from this assump- 
 tion, he has not hesitated to derive all the indeclinable words, whether 
 conjunctions , prepositions , or adverbs, from nouns or verbs , and thus 
 contrived to disguise the few remains of etymological structures in 
 our language*. As the best answer to an erroneous system is a plain 
 statement of the antagonistic truths, and as this is the final object of 
 the present work in reference to the English representative of the 
 Ileracleitean school, we will simply offer our shield to those who 
 
 * It is right to mention, that most writers on philology hefore com- 
 parative grammar was brought to its present state of perfection fell into 
 the same error of taking a logical view of etymology, but we select Home 
 Tooke as an object of animadversion, because his book is designedly 
 wrong, and actually pernicious. 
 
CHAP. VI.] THE PAETS OF SPEECH. 235 
 
 dread the power of the ejtsa Λτερόενταή-; and will here take our leave 
 of this celebrated piece of sophistry, with the remark that, although 
 it contains some very happy explanations of English words, and some 
 very ingenious argumentations, written in a lively and attractive 
 style, it should never be oj)ened by any one who has not sufficient 
 philological knowledge to guard him from the errors into which it 
 will inevitably lead the ignorant or unwary. 
 
 127 The logical or syntactical distribution of words was, as we 
 have already stated, made the basis of that division of the parts of 
 speech which has been adopted for practical convenience in learning 
 particular languages , according to which it varies. In the case of the 
 Greek language the usual arrangement is a follows : 
 
 ( Substantive. 
 
 1 Noun, < ... ,. 
 
 ' I Adjective. 
 
 2 Pronoun, including the Article. 
 
 3 Yerb, with which is connected, 
 
 4 Participle. 
 These are all capable of inflexion. 
 
 5 Adverb. 
 
 6 Preposition. 
 
 7 Conjunction. 
 
 8 Interjection. 
 
 These are not mflected, and are generally called particles. 
 
 f There can be little doubt that in taking this title for his book, 
 Home Tooke merely meant to imply that the wings of Mercury, or our 
 haste in expressing our meaning, occasioned those abbreviations to which 
 he traces all the corruptions of language — consequently, by enea τίτερόεντα 
 he understands merely "hurried or hastily uttered words." It is, however, 
 worth while to remark that here, as elsewhere, he is in error; for the 
 metaphor in the epithet πτερόεντα it borrowed from the winged arrows, 
 to which words are so often compared in ancient writers; and, in spite 
 of the joke, it may be truly said, that he has robbed the phrase of its 
 point. See Pindar, 0. ix. ii: ητερόεντα δ' ΐει γλνκυν Πν&ώναδ' όϊβτόν 
 (cf. Soph. P/w7. 166: TiTccvolg ίοΐς). Ν. ιχ. 55: άτιοντί^ων αηοποϋ αγχιΰτα 
 Μοίΰάν, and the note on Ο. vi. 82. In the application of the epithet 
 'winged' to the arrow itself, there was a sort of metaphor or quasi-perso- 
 nification: see JEschyl• Eumen. 181: μη y.ccl λαβονΰα ητηνον άργηοτην 
 δφιν, χρνβηλάτου Ο'ώμίγγος έξορμώμενον. And so lightning is not only 
 the dart of Jove {βέλος τιεραννον, πνρπάλαμον βέλος), but is represented 
 as furnished with eagle wings (see the coins from Elis, in Bronsted's 
 Voyages dans la Grece, i. Livr. p. 112): and from this figure the spread 
 eagle of modern heraldry is obviously derived. By a further personi- 
 fication, the Eagle itself, as z/iog άγγελος, is the bearer of the thunder- 
 bolt; and it is ^vell known that the Semitic poets passed in a similar 
 manner from the powers of nature to an army of MaP hakim. This tran- 
 sition is shown even by the form of the word: "neque enim ullum nomen, 
 cui Mem prsefixum est, vim personie innatam habet, sed per translationem 
 abstracti in concretum miituatur" (Fiirst, Concord, p. 581). 
 
236 THE PABTS OF SPEECH. [bOOK I. 
 
 The old grammarians adopted this arrangement, with the excep- 
 tion, that they classed the interjection with the adverb and made the 
 article a distinct part of speech (Dionys. Thr. BeJck. Anecd. p. 634). 
 For the purpose of learning a language, the syntax of which is logical 
 and for the most part well understood, this arrangement is highly 
 convenient, and from its constant adoption the ear has become so 
 familiar with it, that its nomenclature can hardly be banished even 
 from etymology, with which it has no concern. We, therefore, 
 retain the names of these eight parts of speech, though, to avoid con- 
 fusion, we render them subordinate to the real etymological division 
 of words, which we have already mentioned , and which we now more 
 formally set forth. 
 
 128 In the Indo-Germanic languages, all words may be 
 comprised in two general classes; the pronouns, or those words 
 which indicate space or position ; and the words containing 
 ROOTS, which express the positional relations of general attri- 
 butes. The former are words , declinable or indeclinable as the 
 case may be, without any admixture with the other element of 
 language. The latter require the addition of at least one pro- 
 nominal suffix to make them words. It is for this reason that 
 we term the pronouns or positional words, the organizing, 
 constituent, or formative element of inflected language, and the 
 ROOTS the material element. By pronominal additions of a per- 
 fectly analogous nature, the same root becomes either a noun or 
 a VERB, that is, it expresses either a thing, or an acting, or result 
 of acting. The only etymological difference between the noun 
 and the verb is this — that the pronominal suffixes, which mark 
 the inflexions of the noun, are fixed or adverbial, while those 
 which mark the persons of the verb are themselves capable of 
 inflexion. When they lose this independent power of inflexion 
 they become intermediate affixes, and the crude verb is then ca- 
 pable of receiving a set of case-inflexions, so that it becomes a 
 noun. This sort of noun is called a participle. We cannot, 
 therefore, consider the noun and verb as, etymologically, different 
 parts of speech, but, on account of their prominent importance in 
 the syntactical scheme, we have in the following pages classed 
 them under separate heads; which arrangement is farther jus- 
 tified by the fact, that in the case of the verb the idea of space 
 has developed itself into the idea of time. That the formative 
 
CHAP. VI.] 
 
 THE PAET8 OF SPEECH. 
 
 237 
 
 element of language is prior to the material, appears from what 
 we have just mentioned, — that the pronoun is a word without 
 extrinsic addition; but the noun and verb are such only by the 
 addition of pronominal element. It also appears from psycho- 
 logical considerations. Every thing is conceived as happening 
 in space or time, the idea of space being however antecedent to, 
 and the parent of, that of time. The first conception about any 
 thing is that" it has a position, that it is somewhere without us, 
 and, as it is our conceptions that we express in words, the first 
 words must be those which indicate position, that is, pronouns*. 
 The next conception with regard to the particular object, is of 
 some particular quality with which sensation has invested it, and 
 this quality of course gives it a name according to a sort of prima 
 facie classification. But still it is somewhere^ and therefore the 
 pronoun is tacked on to the end of it, in order to constitute it a 
 word: the same would be the case, though in a more obvious 
 manner, when the thing conceived was not a quality but an 
 action. 
 
 That, in the Greek language in particular, all words may be 
 resolved into and deduced from these elements, will be abun- 
 dantly shown in the following pages. As we use the word 
 "pronoun" or "pronominal" in a sense somewhat wider than 
 that which it generally bears, it will be as well to give compa- 
 rative tables of the etymological and ordinary arrangements of 
 the parts of speech. 
 
 Etymological. 
 
 Formative element or Pronoun, 
 
 Material element com- 
 bined with Pronoun, 
 
 Syntactical. 
 
 {Preposition, 
 Conjunction, 
 Numeral, j 
 {Adjective, J 
 Substantive, 
 ί Participle, 
 Verb. 
 
 JLibrm 
 
 California 
 
 * W. Humboldt, though he admits that the pronoun is an original 
 element of language, says very emphatically: "according to my most 
 profound conviction, any attempt to define the chronological sequence of 
 the essential elements of speech, is a chimera (em Undingy (jiber d. 
 Verwandtsch. d. Ortsadverhien mit dem Pronomen, 1830, p. 3). 
 
238 THE PARTS OF SPEECH. [bOOK I. 
 
 The interjection is either an unmeaning cry, a residuary 
 pronoun, or the vocative case of a noun*: the adverbs are gen- 
 erally pronominal words, but some of them are merely cases 
 of nouns. The numerals are adjectives in syntax, though ety- 
 mologically pronouns. 
 
 * A writer in the Proceedings of the Philological Society, in. No. 72, 
 p. 211, says: "Interjections had originally a distinct meaning and dis- 
 tinct origin," and by way of proving this, he derives eja from audin! Of 
 course, he never heard of the Greek εΐα, and he is prepared, we presume, 
 to find a new parentage for ejulare. As the writer in question is the only 
 person in this country, who is styled Professor of Comparative Grammar^ 
 he might perhaps have been expected to inform himself of the fact, that 
 the change of aud- into ec?-, or rather cec?-, could only occur in a com- 
 pound, like ob-oedio] and even in this word the later writers seem to 
 have preferred ob-audio. Experience teaches us that those who have 
 neither the industry to learn nor the sagacity to discover the truth, have 
 often the courage to invent, and that there are no limits to the extrava- 
 gance of perverted ingenuity. 
 
BOOK II 
 
 PRONOMINAL WORDS. 
 
THE 
 
 NEW CRATYLUS. 
 
 BOOK II. 
 
 PRONOMINAL WORDS. 
 
 CHAPTER I. 
 TEE PERSONAL AND OTHER PRONOUNS. 
 
 129 Ultimate analysis of pronominal words discovered by Bopp.lSOPrinciples of 
 arrangement and classification of the pronominal elements first suggested by 
 the present Author. 131 General view of the pronominal combinations in the 
 Greek Language. 132 Objective cases of the three personal pronouns. 133 
 Nominatives of the first and second pronouns. 134 Lengthened forms of the 
 objective cases, supplied by Bentley's view of the Homeric possessive. 135 
 Demonstrative force of first and second personal pronouns. 136 Plural and 
 dual forms of the first and second pronouns. 137 The nominative, masculine 
 and feminine, of the third personal pronoun. 138 The combinations αν-τόξ 
 and ov-Tog. 139 The nominative t'and its affinities. 140 Demonstrative use 
 of the Greek reflexive; 141 Its confusion of number and gender; 142 Its 
 vague application to different persons. 143 Identity οίαφέ and ε. 144 Doric 
 inversions ipe, ipiv. 145 Indefinite, interrogative, and relative pronouns all 
 referable to the second element: 146 Their forms in Latin; 147 In Sanscrit. 
 148 Connexion of the relative and demonstrative in Greek. 149 Guttural 
 origin of riff. 150 Demonstrative use of the guttural pronouns. 151 Gen- 
 eral conclusion respecting the pronouns. 152 Pronominal adjectives. 
 
 129 TN the Indo-Germanic languages, considered in their most 
 ancient form , we can always resolve the pronouns into 
 the shortest possible words, monosyllables for instance, or even 
 single vowels ; but in the latter case we have, of course, derivative 
 forms, for, as we have shown before, no single vowel can exist in 
 the first instance without at least an initial breathing: in fact, 
 the primitive pronouns must have been very simple words, for 
 the first and easiest articulations would naturally be adopted to 
 express the primary intuition of space. These little vocables 
 denote only the immediate relations of locality; and to designate 
 
242 THE PEKSONAL [bOOK IL 
 
 all the subordinate varieties of position and direction, it is the 
 custom , especially in the Greek language, to join together the 
 different pronouns, or different modifications of the same prono- 
 minal stem, till at last we arrive at long words , like δϋ-τις-δή- 
 πο-τε, every syllable of which is a distinct pronoun 
 
 The first principles of this ultimate analysis of all pronominal 
 words were discovered by Francis Bopp. In his essay ueher 
 den Einfluss der Fronomina auf die Wortbildung (Berlin, 1832), 
 he remarks (p. 13): "From the dissection of the pronouns and 
 the prepositions connected with them, we get the following 
 monosyllabic stems, partly consisting of a mere vowel, which 
 either occur in Sanscrit only, or are found in the connected 
 European languages with more or less exact correspondence 
 in form: a, i, u, e; ka, hi, ku; na, ni, nu; ma, mi, (-μή, 
 7nu; ya^ yu; va, vi; ta, da, sa. The compound pronouns — 
 meaning thereby not derivatives like ta-vat, '-so much,' but pri- 
 mitives, which the grammarians consider as simple, but which 
 we have endeavoured to reduce into their real elements — show 
 as their first member, in Sanscrit, a stem consisting of a single 
 vowel ; they are the following, a-va^ i-va, e-va, a-na, e-na, a-da, 
 i-da, e-ta, e-ka, e-sha,''^ But, although this analysis was pointed 
 out many years ago, and though the importance of these re- 
 searches was soon afterwards recognised and explained by Mr. 
 Garnett (^Quarterly Review^ Vol, LVii. p. 80 sqq.; Essays, pp. 96 
 sqq.), no one* has thoroughly examined, compared, and classified 
 these monosyllabic stems and others which Bopp has omitted 
 to mention; we shall therefore endeavour to show which of these 
 pronominal words have a common origin, and thus to arrange 
 them according to their natural coherency. (See below, § 169.) 
 
 130 It is reasonable to suppose that the primitive pronouns 
 would be designations of here and thet^e, of the subject and 
 object as contrasted and opposed to one another. As soon as 
 language becomes a medium of communication between two 
 speaking persons (and it is useless to consider it before it arrives 
 at this point), a threefold distinction at once arises between the 
 
 * These words, which appeared in the two former editions, are still 
 applicable. 
 
CHAP. I.] AND OTHER PRONOUNS. 243 
 
 here or subject, the there or object, and the person spoken to, or 
 considered as a subjectin himself, though an object in regard to 
 the speaker. We find traces in the Indo-Germanic languages of 
 an application of the first three consonant-sounds belonging to 
 this family of languages, namely, the three tenues, to denote 
 these three positions of here, near to the here, and there, or 
 first, second, and third personal pronouns, as they are generally 
 called. These tenues, articulated with the usual short vowel, 
 are the three pronominal elements pa (found in τΐα-ρά, &c.), qva 
 or ka (found in κε, &c.), and ta (ro, &c.). The two former are, 
 however, more usually expressed by the cognate sounds ma or 
 t-a, and by fa whether the digamma is represented by one or by 
 both of its members (above, § 110). By a similar change of ar- 
 ticulation the third element appears as na. If now we take the 
 elements enumerated by Bopp, we shall find, that, according to the 
 principles stated in a former chapter, ma, mi, mu, va^ m, belong 
 to the first of them; ka, ki, ku, ya^ yu, e = ai = ya, i, u, to the 
 second ; and ta, nu^ ni, na, to the third of the original pronouns. 
 The second also appears under the forms ga, ha, immediately 
 derived from qva or fa. The syllables da, tha, and 5a, for reasons 
 which will be stated in a subsequent chapter, generally belong to 
 the second pronoun, though the two former would seem to be only 
 slight variations of the element ta, and the pronoun sa is actually 
 used for the third pronoun when that pronoun denotes a person 
 or subject. From va, as a variation oima the first pronominal 
 clement, we must carefully distinguish the same syllable when it 
 appears as a mutilation of /a, the second element; but there are 
 some cases in which this distinction cannot be made without the 
 most refined etymological analysis. .And here we will anticipate 
 what will be stated hereafter more at length. According to the 
 principle mentioned above, any one of these simple elements 
 may be compounded with any one of the others so as to form 
 new modifications of the idea of position. There is, as we 
 shall see, a pronominal element -la or -ra, derived from na, but 
 indicating motion, or conveying the idea of "beyond." When 
 this is added to the first pronominal element, it combines the 
 idea of closeness with that of removal, as in ΰΐα-ρά, tcb-ql, 
 which express motion from or to the side of, and motion close 
 round an object. When with the second, it expresses the third 
 
 R2 
 
244 THE PEESONAL [bOOK II. 
 
 position as opposed to the second , and tva-ra^ as we shall see 
 in the following chapter, is equivalent to ta^ which indicates the 
 third position, i. e. the second position in a state of further 
 removal. On the contrary, by combining the third pronominal ele- 
 ment with the first, under the form of ma, we obtain a signification 
 of nearness approaching to, or even coinciding with that of the 
 second element: thus ta-ma, the suffix of the superlative, ex- 
 presses the approximation of the end of a series to the speaker, 
 and, conversely, ma-ta denotes the approach of the speaker to 
 a distant object*. We shall see in the next chapter that the 
 first three numerals are the primary pronouns under the forms 
 ma, tva, and tva-ra. The elements va and na are both em- 
 ployed to designate the first person, though always in the dual 
 or plural number. They are also used to convey the strongest 
 signification of the demonstrative pronoun , that of distance or 
 separation. This coincidence in meaning between the first pro- 
 nominal element under the form ma., va, with the third per- 
 sonal pronoun na, is explicable psychologically, from the fact 
 that the ideas of self, unity, separation, distance, solitude, and 
 negation, all spring from a common source. The element na is 
 an emphatic expression of the there, or distance ; ma, &c. of the 
 here or self as a separate individuality. Hence, the strongest 
 predication of self in these languages is, in Sanscrit and Latin, 
 a-ha-m, e-go-met, "that which is here," but in Greek Ι-γώ-νη, 
 "that which is by itself or separate." The apparent coinci- 
 dence of va and na^ as expressive of negation , is found only in 
 the longer forms a-va or a-u, and a-na or a-w, ultimately re- 
 presented by a- or even e-. But we hope to show in the follow- 
 ing pages that na or ana is actually prefixed to va when the 
 compound a-va or a-pa bears the negative signification. 
 
 131 We have thus stated beforehand the results of our ana- 
 lysis of the pronominal words, in order that the student may be 
 provided with a general map of the country which he is about 
 to survey, and, knowing what to look for, may not be con- 
 fused by the multiplicity of details. We now proceed to examine 
 
 * The student will find these principles categorically set forth and 
 illustrated in our Greek Grammar, articles 63 — 79, 356, 357. 
 
CHAP. I.] AND OTHEE PEONOUNS. 245 
 
 each of the pronouns more minutely, taking them in the order 
 in which they appear in the common grammars, and beginning 
 in each case with the existing classical form, and so ascending 
 to its primary state. 
 
 It will be found that our analysis has conducted us to the 
 following general results. 
 
 (1) All pronouns are demonstrative, or indicative of par- 
 ticular position in space. Accordingly, in their original ap- 
 plication, there could not have been those distinctions between 
 the personal, demonstrative, and relative pronouns, which are 
 of so much importance in syntax*. 
 
 (2) The elements of these positional words very rarely occur 
 in their simple and unaffected state; for the craving after an exact 
 definition of locality, which necessitated their employment in 
 the first instance, has generally led to the superaddition of other 
 distinctive syllables. Consequently, in the forms of language, as 
 they appear to us, it will very often be found that the shortest 
 pronominal words or affixes are not the primitive types in the 
 particular instances, but are mutilated relics, which an accurate 
 investigation will often enable us to restore to their original 
 completeness. 
 
 (3) As distinct words the pronominal elements appear in 
 the form of demonstrative substitutes for the noun, or of 
 numerals, with the corresponding adjectives in either case, and 
 also as prepositions and other particles. 
 
 (4) As part of the formative machinery of language, the 
 pronominal elements appear regularly in the case-endings of the 
 noun and in the person-endings of the verb, and variably in 
 the termination of the crude-form of the noun, and as the ex- 
 tension of the root in the verbal-derivative. The person-endings 
 are merely certain functions of the objective cases of personal 
 pronouns, and the cases are connected with a special develop- 
 ment of the second and third elements. In the derivative forms 
 we find the converse. Those from nouns use all three pro- 
 nominal elements, in their distinctive senses, and in combination 
 
 * The student will find in the Greek Grammar ^ articles 229, 408 sqq. 
 the proper discrimination of all the Greek pronouns according to their 
 syntactical use. 
 
246 THE PEESONAL [bOOK II. 
 
 with one another: while the verbal derivatives are limited to 
 that special development of the second and third elements, which 
 we find in the cases of the noun. 
 
 132 The objective cases of pronouns, as well as of nouns, 
 are always older than the subjective. This appears from the 
 fact that there are many nouns which have no subjective case 
 (for instance, all neuter nouns), but no one, so far as we know, 
 which has the nominative only. It might also be inferred, 
 from a priori considerations, that it must be so. All things are 
 to us parts of an external world, and must needs be spoken of 
 as such long before the mind of man can invest the not-me with 
 the powers of agency and will, which we experience in our- 
 selves. We feel that even the spot on which we stand, and 
 Avhich is for the moment identified with our description, for we 
 are the here, is nevertheless a not-me, and is spoken of as some- 
 thing without, as an object, and therefore must continue to be 
 called one till language begins to assume a logical structure. 
 
 The common forms of the objective or accusative cases of 
 the personal pronouns, in Sanscrit, Greek, Latin and Gothic, 
 are as follows : 
 
 2nd 
 3rd 
 
 The Cretans appear to have said r /ε instead of (jg, as may 
 be inferred from the following glosses of Hesychius: τίορ. 
 ύον. Κρψες. — τρε. ΰέ. Κρψες. Here we should read τδο/ for 
 rsop, and r/f for τρε, for immediately under the former gloss we 
 have τεόζ-, ίχηάτεονς. ύονς — and the resemblance between Ρ and 
 / might easily cause the mistake. Besides, a labial is mani- 
 festly included in the nominative tv. We shall discuss hereafter 
 this corruption of the second pronominal element fa. The Latin 
 turn occurs only in composition or as a particle. 
 
 The termination of the objective case of the third person is 
 strictly a case-ending, i. e. expressive of a relation of place. But 
 the first element is obviously appended to the first and second 
 pronouns in Sanscrit, and the second element to the same pro- 
 
 mscrit. 
 
 Oreek. 
 
 Latin. 
 
 Gothic. 
 
 mam, ma 
 
 με 
 
 me 
 
 mik 
 
 tvdm, tvci 
 
 ΰε 
 
 te 
 
 thuk 
 
 tarn 
 
 τοΛ 
 
 turn 
 
 thana 
 
CHAP. I.j AND OTHEE PRONOUNS. 247 
 
 nouns in Gothic, and, as we shall see presently, in Greek and 
 Latin also. This is a metaphysical rather than a grammatical 
 phenomenon*, and indicates that the here or the near is implied 
 in the second, no less than in the first personal pronoun. 
 
 The ε, which, under certain circumstances , is found at the 
 beginning of the Greek με and the other oblique cases of the 
 first personal pronoun, may be compared with the prosthesis of 
 a vowel in such words as ο-νομα, 6-dovg, 6-φρνς, ε-λαχνς, by the 
 side of ndma, danta-s, bhru-s, laghu-s. We find it also in Ικέίνοξ 
 compared with nuvog. There is no reason to doubt that it is the 
 residuum of a pronominal element, and most probably, like the 
 verbal augment, it represents the evanescent -v , when it is 
 followed by a guttural, but stands for / or one of its constituent 
 parts, when the following syllable begins with a nasal or dental 
 sound. 
 
 These forms, then, of the objective cases of the personal 
 pronouns are identical in the four languages compared above : 
 this identity they also maintain as the personal endings of the 
 oldest class of verbs in Sanscrit, Greek, and Latin; but then, as 
 we shall show hereafter, the vowel is generally subjected to 
 certain changes which we shall discuss in the proper place. 
 
 133 In the nominative also, the resemblances of the first 
 two, though not so striking, are sufficiently certain. 
 
 
 Sanscrit. 
 
 Greek. 
 
 Latin. 
 
 Gothic 
 
 1st person 
 
 aham 
 
 Ιγών 
 
 ego 
 
 ik 
 
 2nd 
 
 tvam 
 
 tovvf 
 
 tu 
 
 thu 
 
 It is worthy of remark, that although the nominative of the 
 first personal pronoun is distinguished in every language of the 
 Indo-Germanic family by an initial vowel, whereas the objective 
 case invariably commences with the characteristic m , the no- 
 minative of the second pronoun manifestly contains the same 
 
 * Grimm {Gesch. d. deiitsch. Spr. i. p. 262) seems to write rather con- 
 fusedly on this point. 
 
 t The word τάν (and in the vocative α τάν, "Ο you,") is an old form 
 of τονν (Buttmann, Ausfi'ihrl. Sprachl. § 57, Anm. i.), and is strikingly 
 like the Sanscrit tvam. 
 
248 THE PERSONAL [bOOK II. 
 
 element as its objective case*. From this fact we are entitled 
 to conclude that while the nominative and objective cases of the 
 second pronoun had a common origin , the nominative of the 
 first person was formed independently of its objective casef. 
 This, however, is by no means inexplicable. Even after the 
 adoption of nominative pronouns the person spoken to would 
 stiU continue to be an object, and therefore the nominative and 
 objective cases of that pronoun wovild contain the same elements ; 
 but when the speaker could detach his notion of himself from 
 the idea of space, from the here^ which before constituted his 
 definition, and consider himself as the /, the real subject, he 
 would adopt some word more emphatic than the mere mono- 
 syllable μ2 to express himself by , and this word was aham in 
 Sanscrit, and ego in Latin and Greek, by a common transition 
 from the h to the g (Pott's Etym, Forsch. I. p. 144). 
 
 If we compare aJia-m, tva-m with a-ya-m, "this man," i-ya-m, 
 "this woman," sva-ya-m, "oneself," va-ya-m, "we," yu-ya-m^ "you," 
 ma-h-ya-m, "to me," &c., we must conclude that the termination is 
 simply m. The Greek εγών was written εγων by the ^olians, εγώνγα 
 and εγώνη by the Dorians (ApoUonius Dyscolus, de Pronom. p. 64 β 
 
 * In addition to the four languages adduced in the text, the following 
 have heen cited by Grimm (Gesch. d. deutsch. Spr. pp. 257, 8). 
 
 
 Nom. 
 
 Object. 
 
 
 Nom. 
 
 Object. 
 
 Nom. 
 
 Object. 
 
 Zend 
 
 azem 
 
 manm 
 
 0. Pruss. 
 
 as 
 
 mien 
 
 Osset. az 
 
 ma 
 
 Lithuan. 
 
 asz 
 
 mane 
 
 Slav. 
 
 az 
 
 mja 
 
 0. H. D. ih 
 
 mili 
 
 Lett. 
 
 es 
 
 man 
 
 Pol. 
 
 ia 
 
 mie 
 
 A. S. ic 
 
 met- 
 
 
 
 
 Bohem. 
 
 v« 
 
 me 
 
 Eng. / 
 0. N. ek 
 
 me 
 mik 
 
 t The Professor of Comparative Grammar at the London University 
 College, in a paper which appears in the Proceedings of the Philological 
 Society, Vol. iv. No. 78, has the courage to maintain that /, me, we, us; 
 that ego, me, nos ;ana that iyoj, νίόϊ, and ημεΐς , have all arisen from one 
 common stem. For example , the German nominative ich is a mutilation of 
 the corresponding accusative mich, and the h in mihi is the same guttural 
 which appears as g in ego. But perhaps any thing might be expected 
 from a writer, who derives the composite pronouns κεϊνος, &c. from a 
 supposed verb ken =. "to see" (Phil. Soc. in. p. 67)! who tells us that βαν 
 is the radical syllable of βαίνω (Ibid. u. p. 147)! who connects dives whh 
 bigce {Phil. Soc. 1854, p. 27, cf. Joiirn. of Phil. ii. p. 354)! and considers 
 eja as another form of audin! (above, p. 238). 
 
CHAP. I.] AND OTHEE PEONOUNS. 249 
 
 Bekker). Of the Boeotian ιών , ApoUonius writes as follows : Botcorol 
 "^ISIN, ώζ ^'£V Τρύφων φήΰιν, νφέΰεί ενλόγω τον γ, ίνα καν τα τηζ 
 μετα^'εόεως τον ε εΐζ ϊ γενηται , ετιεί φωνήεντος ετΐίφερομενον το 
 TCLOVTOV τΐαρακολον&εΐ. άλλα μην καΐ εδαΰνν^'η , επεί δασύνεται 
 τα φωνήεντα εν ταΐζ αντωννμίαις , οτε ττρο φωνηέντων τί%^ενται, 
 εόξ, εον, εαντώ, εαντόν , εοι. ως δε ενίοι, ών καΐ 6 "Αβρών, 
 ^έμα εύτίν δ ΰνζνγως οι αντοί φαύυ tjj μεν εγών την ιών, εϊγε το 
 τίαρα Ζίωριενύιν ε εΙς ϊ μεταβάλλεται , Ty δε εγωνγα την ΐωνγα. 
 Κόριννα' 
 
 μέμφομαι δε και λιγονράν Μνρτίδ' ΐωνγα 
 οτι βάνα* φονΰ' f εβα Πινδαρίοιο ποτ εριν. 
 Και ετι 
 
 ϊωνει ήδ^ ηρώων άρετάς χειρωάδων 
 (read, partly with Sturz, ΐωνγα ονδ' ηρώων άρετάς χείρω αδω). In 
 like manner for ΰν or rv, we find τννη , and the Boeotian forms τον, 
 Tovi^, τονγα (Apollon. de Pronom, p. 69 c); also in the oblique cases 
 the Cohans wrote εμοι, the Boeotians εμν^ the Dorians εμίν and εμινγα, 
 and the Tarentines εμίνη and τίνη (Apollon. de Pronom. p. 104 c, and 
 105 c). In Hesychius too we have ε μήνη (read εμεννη), εμον, and 
 ε^εχέμεναι (read with Hemsterhuis εξεχ εμεννη)^ %ωρΙ^ εμον, 
 Αάκωνες. When we remember that the Dorians wrote δνων for δύο 
 (Hesychius), that εγώ appears as ego in Latin, that the third person 
 dual was sometimes -rov, sometimes -την or -των, that we have me, te, 
 se, in Latin, amd mam, ma, tvdm, tvd in Sanscrit, by the side of ^έ, 
 ΰέ, ε; also μήν, μή, δήν, δή, as well as μεν, μα, δεν, δε, and in the case- 
 endings -δην, as well as -δα, -δον, we shall not be perplexed by the 
 variation in quantity which takes place in these pronouns, and shall 
 
 * This word, whicli is the Boeotian form for γννή, deserves some re- 
 mark. The following are the notices of the grammarians respecting it. 
 Herodian (ηερί μον. λε|• p. 18, 1. 26): το γάρ παρά Κορίνντ] βάνα (1. 
 βανά) ον Kotvbv ονδε εις νη ληγον, άλλα Ι'διον ^έμα Βοιωτών ταοοόμενον 
 άντϊ τον γννή. Hesych. Βάννα (1. βανά). γννή νηό Βοιωτών. Now 
 the Dorians said γάνα instead of γννή (Gregor. Corinth, de Dialecto Dor. 
 § CLiii.). It is, therefore, obvious that the original form must have com- 
 menced with a compound of the guttural and labial, otherwise the two 
 forms are inexplicable. The old word was perhaps γβάνα or fava. This 
 digamma initial is split tip in the usual way in the different Indo-Germanic 
 languages. The whole letter is preserved in the old English quean, the 
 labial only in the Boeotic βάνα, Sanscrit vanita, Erse hemi or vean, Welsh 
 henw, the guttural only in the Sanscrit jani, Erse gean, Russian jena. In 
 the common Greek γννή the labial is vocalized into v, as is the case with 
 ■κ,νων, κννόζ, Sanscrit ς.ναη, &c. On the probable significance of the words 
 of Corinna, see History of Greek Literature, in. p. 446. 
 
 f It may be doubted whether this is the Boeotian form for φνοα (ov 
 being regularly substituted for v, whether long or short, in that dialect), 
 or the participle of the verb φέω , supposed to be the old form of φ i; ω 
 (see J. H. Voss on Virgil, Eclog. i. 49). 
 
250 THE PEE80NAL [bOOK II. 
 
 perhaps be disposed to recognise, without any hesitation on that ac- 
 count, the identity of the first two syllables of aha-m and Ιγώ-ν, 
 εγών-η. The termination of the former is undoubtedly the clement 
 of the first personal pronoun, which appears at greater length in the 
 Latin ego-met; the final -v, -νη of the latter is a variation of the ele- 
 ment ta, which we shall frequently have occasion to speak of. The 
 bulk of the pronoun aha-m is a compound of the pronominal elements 
 a, ha, the former being here of demonstrative, the latter of relative 
 import, and the whole signifying "that which," or conveying, with 
 the affix, the meaning "that which is here," while the gloss in Hesy- 
 chius {Ιγών, αντοξ Ιγώ) seems exactly to express the meaning of 
 Ι-γώ-νη, i, e. "that which is by itself." We have the same combination 
 in s-xag, eka-s, and, with the casal affix tur=z-tus, in igi-tur, "from 
 or after that" (see below, § 362). If we may be allowed to state now 
 what we shall prove in the course of the following chapters, the rela- 
 tive element, which here in Sanscrit and generally in Greek appears 
 under the form ha, o-g, also assumes (1) the form γα, γε in Greek, 
 under which it corresjaonds in meaning to the Sanscrit particle ha; 
 (2) the form que, qui in Latin, under which it corresponds in meaning 
 to the Greek γε, Sanscrit ha (qui-dem = γε-δην) ; (3) the form ka, oca, 
 in Sanscrit and Greek. All these, it will be seen, belong to the second 
 pronominal element, which of itself is capable of expressing the mere 
 relative pronoun, and the cognate signification of the genitive case. 
 In order, however, to give greater emphasis to this sense of proximity, 
 derivation, and relation, it is not unusual to find a reduplication of 
 the second element. As Mr. Garnett has remarked, in his instructive 
 essay on this subject ("On the origin and import of the genitive case," 
 Phil. Soc. Vol. Π. nr. 39, p. 168; Essays, pp. 214 sqq.), we find cases 
 in which, "for the sake of greater precision, the demonstrative element 
 na is doubled to form a relative, much as in Norse and Anglo-Saxon: 
 sa-er, se-the = 'who,'' lit. 'the-the,' or 'the-that:' the object of this 
 duplication appears to be to establish a more precise connexion be- 
 tween the antecedent and the relative clauses , a portion of the com- 
 plex expression being referred to each." Although, therefore, the 
 existing import of a-ha, e-go, &c., may be "that which," there is 
 every reason to believe that we have here the remnant of the redupli- 
 cation ha-ha, the stronger form being retained for the relative expres- 
 sion. This is confirmed, as we have elsewhere shown*, by the analogy 
 of the composite Hebrew relative "nm ^hasher', compared with the 
 personal pronoun ^Dbi< Ίια-ηό -ki, in which the elements of ε-^ω-ντ; 
 
 Maskil le Sopher, pp. 12, 24. 
 
CHAP. I.] AND OTHER PBONOUNS. 251 
 
 appear in a different order ; and we shall see that the word fa-va -ξ 
 involves a similar combination of pronominal roots. For a-ha-m we 
 have a-da-m in the Behistun inscription; and there are other in- 
 stances in which the d appears in this inscription as the representa- 
 tive of a Sanscrit h (Rawlinson, As. Soc.x.2j p. 114). Moreover, the 
 character itself indicates some connexion with the old Persian di- 
 
 gamma which is made up of / and γ γ or «i (above, § 110). We 
 trust to be able to show in the following chapter that the pronominal 
 element ma is also one of the representatives of the first numeral in 
 Greek. There is another expression for unity in Sanscrit e-kas^ in 
 Greek ε-κας^ in Latin cequus (secus), in Hebrew "τΠΝ ^elidd, both 
 syllables of which represent the second pronominal element. These 
 expressions for unity are, in our opinion, identical with the Greek 
 Ι-γώ^ the Latin e-go, and the bulk of the Sanscrit a-Jia-m. In the 
 BcBotian ίώ-ν the evanescent guttural is represented by the aspirate, 
 on a principle before explained. 
 
 134 With the final letter suppressed, a-ha is in Sanscrit "a par- 
 ticle and interjection implying (1) commendation; (2) rejecting, send- 
 ing; (3) deviation from custom (improperly); (4) certainly, ascer- 
 tainment" (Wilson's Sanscr. Diet, s.v.); and it is interesting to know, 
 that the Latin language has two interjections, ehem and eJio, cor- 
 responding to aham and ego respectively, which convey a meaning 
 very nearly akin to what we conceive to have been the original force 
 of the first personal pronoun: compare such passages as ehem, Demea^ 
 hand aspexeram te: — eho puer, curve ad Bacchidem. It is not impos- 
 sible that eja and Bia may be residuary forms of the same kind. These 
 are not the only instances of the use of the particle lia = fa in Latin. 
 It appears to have been the regular suffix of the objective cases of the 
 first pronoun in old Latin. Quintilian says {Inst. Oral. i. 5, § 21): 
 Inde du7'at ad nos usque vehementer, ei compkehendere, et Mim: et 
 MEHE quoque pro me apud antiquos, tragoediarum prcecipue scriptores, 
 in veteribus libris invenimus (see Lepsius, Tahb. Eugub. pp. 92 sqq.). 
 It is not improbable too that the same termination was employed to 
 form the objective, genitive, and dative cases of the first personal pro- 
 noun in the oldest Greek. This dative appears generally under the 
 form μοί or Ιμοί. Bentley, however, perceived that the metre in 
 Homer occasionally required μεον, μεοί, instead of εμον, εμοί, and 
 μεός instead of εμόg in the possessive, after the analogy of the Latin 
 met, mihi, and mens. We give the passages in which he has noticed 
 this, from a MS. in the Library of Trinity College : 
 
252 THE PERSONAL [bOOK II. 
 
 Uiad IV. 426. ην vvv εφράΰύαντο, μεοϋ άττομηνίΰαντος. 
 Odyss.x. 4:25. αυτοί δ' ότρννεΰ^'ε μεοϊ αμα πάντες ετιεΰ^&αι. 
 XX. 364. Ενρνμαχ ,οντυό' ανωγα μεοϊ τΐομτΐηας οπάζευν. 
 
 In the two latter passages the editions have εμοί, in the first εμεν. 
 In the following, they have different cases of the possessive εμόζ- 
 
 Iliad IX. 412. τεττα, ΰιωτι^ ύι<3ο, μ ε φ δ' εταπείΟ'εο μύΟ'ορ. 
 
 IX. 57. η μην καϊ νεοξ εββί, μεοζ δε %ε καΐτΐάΐς εϊης. 
 XIX. 194. δώρα μεης τίαρα νηοξ ενεγκέμεν. 
 XX. 20. εγνως/Εννοβίγαιε, μεην εν ΰτή&εΰί βουλην. 
 
 XXIII. 278. πατρί μεω Πηληϊ. 
 
 Odyss. IX. 459. καδδε μεον κηρ 
 
 λωφηΟειε κακών. 
 χ. 178, and 428; χπ. 222. οί δ' ώκα μεοΐς ετιέεύΰί πι- 
 &οντο. 
 XII. 258. οϊκτοΰτον δη κείνο μεοΐς /ίδον οφ^αλμοΐΰι. 
 397, and χιν. 249. εξημαρ μεν ετίειτα μεοϊ ερίηρεζ 
 εταίρου. 
 χιπ. 305. ^οίκαδ' οτίαόΰαν ϊοντι,, μεη βονλτ] τε νόω τε. 
 
 XXIV. 327. εΐ μεν δη ^Οδνύενξ γε, μεοςτίαϊξ, ενΰ'άδ' [κάνεις. 
 
 As we have the forms εμίω (Ahrens, dial. Dor. p. 249) and εμέ&εν, 
 and even με^'έν in Sophron (ετί μεΟΊν a καρδία τίαδη. Fr. 46, ap. Apoll. 
 de Prcn. 83 c, 98 a), we are not obliged to assume the form μεον^ 
 though the analogy of τεον (Apoll. 96 b) renders it justifiable. It 
 must be remarked in general that all these case-endings of the pro- 
 nouns belong to the full development of a language, and are as recent 
 as the case-system of the nouns themselves. For we get beyond the 
 merely distinctive use of the three original elements themselves , as 
 soon as we begin to speak of case-affixes^, which, as we shall see, are 
 connected with a special development of the second and third ele- 
 ments. The Boeotian εμν for εμοί points to an original εμε /t, just as 
 τν represents an original τ/ε; and with regard to Bentley's assump- 
 tion of μεοϊ for εμοί, we think the ο in both forms is a substitute for 
 the lost digamma or aspirate , which is otherwise represented in the 
 common suffix -φο, and that μεοί = με-φί = με^ί is fully equivalent to 
 mi-hi. The same stem is represented by the k, or ch appended to the 
 Gothic mik, thuk. New German mich^ dich; and the long vowel in 
 the Latin me^ te, may indicate, as we have seen, an original me-he, 
 te-he; so that in the Greek, Latin, and German languages the second 
 element was probably appended to the first and second personal pro- 
 nouns in the objective cases. We have already adverted to the meta- 
 physical significance of this phenomenon. 
 
CHAP. I.] AND OTHER PRONOUNS. 253 
 
 It may be objected to this explanation of the identity of ha^ γα, 
 &c., that, if so, this element is repeated in such combinations as εγωγε, 
 ϊγωνγα, ΐωνγα, &c.; but it will be remembered that nothing is more 
 common than such repetitions of the same root in pronominal com- 
 pounds, and especially when it appears under slightly modified forms. 
 "When it is placed after the termination -v-, as in εγωνγα, the word is 
 a compound of two compound pronouns, into each of which similar 
 pronominal elements enter : for νγα or ναγα is a compound analogous 
 to me-he, mi-hi. 
 
 135 As the three primitive personal pronouns are expres- 
 sions for the three relations of place, we ought to find in the 
 Greek and cognate lenguages traces of the use of all the primi- 
 tive forms as demonstrative pronouns ; and we do so. Of the 
 third it is unneccessary to speak. With regard to the first; in 
 Sanscrit we have i-ma, "this-here," from which Bopp derives 
 the Latin words im-ago and im-itor (I)emonst7'atwstdmme , p. 
 21). The Greek demonstrative μίν, and, as we shall hereafter 
 show, the preposition με-τά^ the particles μεν, μά^ and the verbs 
 μάω=^μενω, maneo, &c., all contain this element*. It appears 
 as a suffix to the second and third personal pronouns : in San- 
 scrit, tva-m^ aya-m, as well as aha-m; in Latin, tu-met, vos-met, 
 se-met, ipse-met, as well as ego-met^ nos-met. The second element, 
 under the form dva^ is used as a demonstrative in the numerals. 
 Of its use under other forms derived ifrom J^a we shall speak by 
 and by. 
 
 If now we look into any practical Grammar of the Greek 
 language, we shall see that, while the first and second persons are 
 always expressed, when they require pronominal expression, by 
 the forms εγώ, 6v or their oblique cases, the pronominal expres- 
 sion of the third person varies with the particular reference, 
 which is intended in the construction of the sentence. All three 
 persons may be denoted by demonstratives, that is, in terms of 
 the third person, if we think fit to employ the indicative pronouns 
 οδε, "he who is here, i. e. by me," ovrog, "he who is near, i. e. by 
 you," and εκείνος^ "that other man, i. e. he who is there." Thus 
 
 * Mlv from the first person and νίν from the third are so completely 
 synonymous that, as we have shown elsewhere (ad Pindar, p. Iviii), 
 euphony alone determines their employment in the lyric poets. 
 
254 THE PEESONAL [bOOK II. 
 
 οδε signifies the speaker and his client in a law-suit, but οντοζ 
 denotes the opposite party, usually the defendant, as "the per- 
 son before you, the person in your court." Or, more generally, 
 ods άνήρ = εγώ, and, in addresses, ώ ovtog = (5v, For example, we 
 have Eurip. Ale. 690: μη ^ντιόχ νπερ τονδ^ ανδρός, ονδ' εγώ τίρο 
 ύοϊ), where the opposition between οδε and εγώ is immediate 
 and palpable; and Soph. CEd. C. 1623: ώ ovrog, ovrog ΟΙδίτϋονς, 
 TV μελλομεν; where the repeated ovrog (Jieus tu!) is an emphatic 
 substitute for όν. On the other hand, εκείνος always denotes 
 something "other" and "distant" either in space or time, as in 
 Arist. ^qu. 390: rovg ΰτάχνς εκείνους, ovg εκείθεν ηγαγεν, "those 
 (other) ears of corn which he brought from thence (that distant 
 place)." And so of time Thucyd. i. 20: εκείνη ry ήμερα, "on 
 that day (long past and gone)." It will be shown below that 
 άλλος = αλίος, Sanscr. any as, and the Latin ille or alius and 
 alius correspond ultimately, in form as well as meaning, to 
 κείνος = κένίος=^^άνως (§ 166). And it is well known that the 
 distinction between οδε, οντος, and εκείνος, as representing de- 
 monstratively the three personal pronouns, is exactly given by 
 the Latin hie, iste and ille^ and the Italian questo {= ecce isfe^^ 
 cotesto {= ecce tibi (eecoti) iste)^ and quello (= eeeeille) (see Diez, ii. 
 369). But besides the indicative pronouns, which, though al- 
 ways construed with the third person of the verb, may thus 
 denote or refer to the three persons in this distinction, the third 
 person may be expressed by the reflexive pronoun t, accus. ε, by 
 the pronoun oiself or same^iess, αντός^ by the distinctive pronoun 
 ος (δ), and by the indefinite τις. It is worthy of notice, that al- 
 though αυτός is combined directly with the first two personal 
 pronouns and with the reflexive in the form εμαυτου, βεαυτου, 
 εαυτού^ and may always stand in apposition to εγώ and (Jv, or 
 take their place with \^erbs of the first and second person, its use 
 in the oblique cases is simply that of a pronoun of reference in 
 the third person, like the oblique cases of the Latin is^ and while 
 I) αυτός άνηρ^ "the same man," corresponds in use to one deriva- 
 tive from is, namely, i-dem, the other derivative i-'pse is repre- 
 sented by an apposition of αυτός to a noun sufficiently defined, 
 as ύ αντ^ρ αυτός = viripse^ " the man himself" Similarly, although 
 αϋτός^ even in an objective sentence, may appear as the apposi- 
 tion of the subject, when that subject can be expressed in the 
 
CHAP. I.] AND OTHER PRONOUNS. 255 
 
 nominative, it is regularly opposed to the reflexive, as object to 
 subject, when that pronoun appears in the accusative; compare 
 Thucyd. IV. 28, § 2: ονκ εφη αντόξ, αλλ^ εκείνον ΰτρατηγεΐν, 
 " Cleon said that not he himself Qt amog)^ but the other (Nicias), 
 held the office of general ; " with Thucyd. iii. 24 : νομίξοντεζ 
 ηκΐΰτα (5φαξ ταντην avrovg νττοτοτΐηβαί τραπεϋ^αι, "thinking 
 (the Platseans) that they (the Thebans) would by no means 
 suppose that they (the Platseans) had turned that way." 
 
 From all this, it is clear that the personal pronouns, whatever 
 position theydesignate, are necessarilydemonstrative,andthatthe 
 variations in their use depend rather on the usages of syntax than 
 on the original significations of these primary words themselves. 
 
 136 The common dual and plural of aJiam and tvam are in 
 the nominative and accusative as follows: nom. dual, dvcim, 
 yuvdm; accus. dual, dvdm or nau, yuvdm or vdm; nom. plur. 
 vayarriy yuyam; accus. plur. asmdn or nas^ yusJimdn or vas. It 
 will, of course, be understood by every one, that the plural of 
 the first personal pronoun, of which the dual is only a modifica- 
 tion, could not be formed from the singular as the plural of any 
 noun might be. The plural of this pronoun must signify one of 
 two things, — either a collection of persons united in the idea of 
 here, and, as such, separated for the moment from the rest of the 
 world, or, as between two speakers , the idea of / + you- On 
 analyzing these Sanscrit forms we shall find that one of these 
 meanings is always implied. The plural va-ya-m is a combina- 
 tion of two modified forms of the first and second pronouns re- 
 spectively, to which the common element of the first is added to 
 imply more strongly that the notion of here is intended. There- 
 fore, va-ya-m means "I + you here." Similarly, the plural 
 yu-ya-m is a repetition of the second element , with the same 
 suffix. Of the first syllable we will speak presently. We have 
 already mentioned the employment of the pronoun expressing 
 distance or separation to denote the first person, and explained 
 how, in fact, there is not that absolute difference between 
 pronouns of different persons which habit leads us to imagine. 
 It is true that there was and is a distinction in meaning between 
 the stems ma and ία, as signifying the opposition of here to there. 
 But they are, both of them, essentially demonstrative, and there 
 
256 THE PEBSONAL [bOOK II. 
 
 is no reason whatever why modifications, in fact, stronger forms 
 of them, should notlbe used to convey the notions of unity, dis- 
 tinctness, and separation, which run into one another. At any 
 rate, there is no doubt of the fact, that these stems are so used ; 
 and we shall see abundant proof of it when we come to a discus- 
 sion of the negative particles. We have an instance of this 
 phenomenon in the pronouns before us. The dual d-vd-m is 
 evidently composed of the pronominal stem a, in our opinion a 
 degenerated form of the third pronoun na, and vd an entirely 
 different element , Avhich is unquestionably a corruption of the 
 second pronoun under the form sva^ and appears as vdm and vas 
 in the accusative dual and plural. Consequently d-vd and 
 the accusative nau = na-va represent the same combination. 
 To the whole is appended the suffix w, so that this word signi- 
 fies "you + I by ourselves ^''^ which is equivalent to va-ya-m 
 = "I -f- you here.'''' In order to analyze the plural accusatives 
 asmdn^ yushmdn^ we must take the Veda forms of the nominative 
 plural, asme and yushme. The former is written amha in Pali 
 and Pracrit. It is obvious that the termination of these forms is 
 sma-i; the aspiration of s in the second word is caused by the u 
 which precedes, a phenomenon common enough in Sanscrit : 
 compare the datives amushmai^ from amu^ and tasmai from ta. 
 This suffix δ?7^α, compounded of the stems sa-ma and also appear- 
 ing as the preposition sam (<5vv)^ is used to form some of the 
 oblique cases of all pronouns of the third person; indeed, asmdt^ 
 the ablative singular of a demonstrative, differs only in the 
 quantity of the last syllable from asmat^ the ablative plural of 
 the first pronoun: sa-ma signifies "all taken together," "whole," 
 "entire," "complete;" and in this sense of completeness, it is 
 used to give verbs in the present tense a past signification : thus 
 we have hanti sma Rdvanan i^awa7i = "Rama killed" (instead 
 of "kills") "Ravana" (Wilson, Diet.). The initial vowel a is 
 the pronominal element na in the last state of mutilation, and 
 thus the compound asme = a-sa-ma-i (the final vowel being the 
 mark of plurality in the case of pronouns ending in a, Bopp, 
 
 Vergl. Gramm. p. 262) signifies "the here taken altogether," 
 with a note of plurality appended. The first syllable of yushme 
 
 = yU'Sa-ma-i^ is obviously the same stem that appears in yuyam. 
 As the dual accusative nau = na-va seems to have the same 
 
CHAP. I.] AND OTHER PEONOUNS. 257 
 
 origin as the form a-vd'm=na-vd-7}i^ it is reasonable to conlude 
 that the plural accusative na-s is a similar mutilation of a-sman 
 ^na-sman. And the same reasoning applies to the abbreviated 
 forms vdm and vas. 
 
 We may now compare these Sanscrit forms with those which 
 occur in some of the other languages of the Indo-Germanic family. 
 The Latin plurals nos and vos agree exactly with the Sanscrit 
 accusatives plural nas and vas^ and the Greek duals νώϊ^ ϋφώϊ 
 correspond pretty well to the duals 7iau and vdm^ the latter of 
 which, as we have said^ has lost its initial sibilant or guttural. 
 In the Greek νώΐ, the most predominant idea must have been 
 "separation" "unity," as appears from the adverb ν6(5-φι, "by 
 himself*," ν6(5-τοζ, "a return" (literally, as we shall see below, 
 "a coming from a distance"), and in the pronoun viv, which 
 always stands alone, though we often have μιν αντόν. The 
 long vowel however shows that here, as in the Sanscrit nau^vdm^ 
 we have a mutilated or abridged form, and there can be little 
 doubt that νώϊ, βφώΐ were originally va-fa~i, J^a-fa-i, The 
 German languages have very short forms for the plural of these 
 pronouns ; generally, in fact, modifications of the simple root. 
 Thus the Gothic has weis^ yas^ Lithuanian mes^ yus^ English we^ 
 you. The Zend has also the short forms vaem^ yus. The 
 Greek forms ημείς, νμεΐς agree with the Yeda-words asme^ 
 yushme; for the >^olic αμμες, νμμεξ (by the ^olian 'φίλωβίξ 
 for νμμεξ) are assimilations of αΰμεξ, νΰμες, just as εμμί stands 
 for εΰμί^ Sanscrit asmi (Bopp, Vergl. Gramm. p. 473), and 
 the assimilation is represented by a long vowel in the possessive 
 αμόζ, νμός (Ahrens, dial. Dor. p. 262). With regard to the 
 first syllable of νμεΐς, the following remarks may suffice. The 
 Greek aspirate often stands for the Sanscrit y : thus yas^ yaj^ 
 yakrit {jecur)^ correspond to og, αξω, ήπαρ. We have no hesita- 
 tion, then, in comparing the Homeric vfova νΰμίνη, "battle," 
 with its Sanscrit synonym yudhma. When we recollect phrases 
 like conserere pugnam^ and compare yudhma with yugma (Lat. 
 jugum), "a pair," "a brace," we shall be disposed to seek for 
 a connexion of meaning, ^ow jugum^ another form of djugum^ 
 
 * The grammarians, absurdly enough, consider this words as an abbre- 
 viation of νόθτο-φι (Hermann, Opuscul. i. p. 222). 
 
258 THE PERSONAL [bOOK II. 
 
 as Janus is of Djanus, Juturna of Djtiturna, &c., contains the 
 element of the second numeral*, as does also the word duellum, 
 "battle." In the next chapter we shall show that the second 
 numeral is identical with the second person singular. As, then, 
 the second numeral is contained in the first syllable of yugma^ 
 yudhma, νβμίνη^ so is the second pronoun in yushme, νμεΐς. 
 The suffix -sma is assimilated into -mma in the singular dative 
 of the Gothic pronoun, just as it is in αμμ^ξ, ϋμμες; thus, the 
 Gothic thamma, hvamma and imma correspond to the Sanscrit 
 tasmai, kasmai and asmai (Bopp, Annals of Orient, hit. p. 1 6, 
 and Grimm, Deutsche Gramm. i. p. 826). It appears as smu 
 in old Prussian: thus antar-smw, ha-smu correspond to the San- 
 scrit antara-smaiy ka-smai (Bopp, Abh. Ak. Berl. 1824, p. 143). 
 
 137 The nominative masculine and feminine of the third 
 personal pronoun are as follows: 
 
 Sanscrit. Zend. Greek. Gothic. 
 
 Masc. sa, sak, so ho 6 sa 
 
 Femin. sd ha a or η so 
 
 The Greek and Zend aspirates are of course derived from 
 the sibilants preserved in Sanscrit and Gothic. The nearest 
 Latin forms corresponding to these are the compounds hi-c, si-c, 
 the latter of which is used only as a conjunction. We shall 
 speak of these in connexion with the forms i, &c. Perhaps the 
 original J^a is preserved in its most genuine form by the 
 Hebrew Nin, though we might be disposed to compare this 
 rather with the compound av, of which we shall speak directly. 
 It will be observed that all these forms belong to a different 
 
 * Compare § 180. Plato had an instinctive perception of the true 
 etymology when he said {Cratylus, p. 418 d): καϊ τό γε ζνγόν otod'a ore 
 δνογόν ol παλαιοί έκάλουν. πάνυ γε. και το μεν γε ξνγον ονδεν δηλοϊ, το δε 
 τοΐνδνεΐν ένεκα της δέϋεως ig την άγωγην ίπωνόμαΰται δνογόν δικαίως' ννν 
 δε ζνγόν. καϊ άλλα πάμπολλα οϋτως έχει. Where Schneider supposes that 
 we ought to read δναγόν, in accordance with the Etym. M. p. 411; but 
 the second syllable in that case would have been long: and Plato is 
 speaking of an old word which might have passed into ζνγόν. If he 
 refers to the natural change of pronunciation which led to the ^, we do 
 not agree with Pott {Etym. Forsch. n. p. 35) that ^'δνογόν ist sicher nur 
 blosse Fiction." 
 
CHAP. I.] AND OTHEE PRONOUNS. 259 
 
 element from the neuter nominative, tat, το, thata, turn or 
 is-tud. In fact, as will hereafter be shown, it is only a mas- 
 culine or feminine noun that can have a nominative case pro- 
 perly so called. The reason for the adoption of a form manifestly 
 connected with the second pronominal element as a nominative 
 of the third personal pronoun, will be obvious on the slightest 
 consideration. The person spoken to, or designated as near, is 
 invested with a subjectivity and personality which is denied to 
 the object spoken of, or designated as there. Now, whatever 
 is spoken of as in the nominative case, is considered as sub- 
 jective in itself, though not a part of the speaker, and therefore 
 can only be designated by a pronoun which expresses the 
 greatest degree of nearness to the here. We shall return to 
 this subject when we come to the case-endings. 
 
 138 There are two stronger forms of the demonstrative or pro- 
 noun of the third person, both compounds with the simple δ, η, τό : 
 namely, ο-δε, ψδε, τό-δε, and ου-τοξ, αυ-τη, τον-το. The former we 
 shall consider in the next chapter. The latter we will now notice in 
 connexion with av-rog, another pronoun of the third person. 
 
 The first part of αύ-τόζ occurs as the separate particle av, express- 
 ing removal, distance, negation, &c. And we shall see that the same 
 particle is involved in the negative ον-κ, and the illative ov-v. It is a 
 prefix in av-^i, αν-τάρ, and, in a weaker form, in ά-τάρ. In the dis- 
 syllabic form it appears in the Sanscrit ava^ ava-k, and the Sclavonic 
 ovo (Bopp, Vergl. Gramm. pp. 400, 544). We consider αν-τός, -ή, -ό, as 
 a combination of the particle av and the inflect^ element το-, which 
 is found in the oblique cases of the distinctive pronoun, also used as 
 the definite article. In the Delian inscription (Bockh, C. I. n. 10) 
 the first syllable appears in the strong form ΑΡΥ, which corresponds 
 very nearly to the Sanscrit and Sclavonic varieties. Bockh supposes 
 (p. 25) that ΑΡΥ was not a dissyllable, that in olden times αντόξ was 
 pronounced afftos, as in modern Greek, and that the lapidary ought 
 to have written AYP. If this had been the reading we should have 
 had a parallel case to the French u, as a substitute for I, that vowel 
 having been prefixed to the liquid, before it actually took its place 
 (cf. autre with alter and the old French aultre) ; but we have here the 
 converse process, which is represented by the change of quojus, quoi, 
 into Gujus, cui, and must suppose that, at the time when the Delian 
 inscription was set up , the digamma had lost its labial element , or 
 the full expression of it, and was subsiding into an aspirate. 
 
 We cannot agree with Bopp {Vergl. Gramm. p. 491) in regarding 
 
 S 2 
 
260 THE PEKSONAL [bOOK II. 
 
 ov-Togj αϋ-τη, τον-το as a combination of αντόξ-, αντή^ αντό^ with the 
 primitive pronoun δ, η τό. Indeed, this suggestion must be added to 
 the many instances which prove that classical scholarship is the safest 
 basis for comparative philology. Every Greek scholar is aware that 
 when 0, &c. are combined with αυτός, &c., the crasis is avtog αυτή, 
 ταντό, not ovrog, αντη, τούτο. And it is quite impossible that a com- 
 bination of αυτός with o, η, τό could give the meaning of omog, which, 
 as we have seen, is a representative of the second personal pronoun, 
 whereas αυτός means that which is away (ανά) and by itself (above, 
 § 135). The true explanation of ούτος undoubtedly is that sug- 
 gested by Max Schmidt (Comment atio de pronomine Grceco et Latino, 
 p. 38), namely, that as δ-άε, ψδε, τό-δε are compounded of the 
 simple demonstrative and the enclitic particle -δε, so οϋτος, αυτή, 
 τούτο give us the same simple pronoun in combination with its 
 objective form -τος, so that ου-τος is formed from 6, just as αυ-τός 
 is formed from αϋ] but it is not right to compare this words with 
 τοιούτος, &c., as Schmidt does, for these words are really compounds 
 of τοϊος, &c., with the pronoun αυτός, as appears from the feminine 
 forms τοιαύτη, &c. In the same way αυτός appears compounded with 
 itself in αϋταυτος (Sophron, apud AppoUon. de Pronom. p. 339 b). 
 Schmidt professes his inability to explain the υ in οϋτος, and admits 
 that according to his analysis it ought to be οτος. It appears to us 
 that οϋ-τος is really an older and more genuine form than ο-δε, and 
 that the υ represents the digamma of the second pronominal element, 
 transposed according to the law which we have illustrated above 
 (§ 116). If this is the case, the first syllable in the objective forms 
 τοϋ-το, &c. represents the same variety of the second pronominal element 
 as that which is found in the second numeral. 
 
 139 The nominative case b=<jo also appears under the subsidiary 
 form i, in which the effect of the digamma has been to convert the 
 vowel into i (compare φύω,υίός, ϋ7ΐερφύης,^αΐα.βο,βΙίη8, υπερφίαλος)^ 
 and this vowel being, as we have before seen, itself a representative 
 of the guttural, the aspiration has been omitted when less emphasis 
 was intended. Both forms appear in Latin, the stronger as hi-c, si-c, 
 the weaker as is, i-terum, i-tem, &c. The latter appears as i~ma, i-taras 
 in Sanscrit. Most scholars, as well as the old grammarians, consider ϊ 
 to be the nominative case of ov, ol, ε', thus we find in Bachmann's Anec- 
 dota, II. p. 72, 1. 3 : ort at τίρωτότυτΐοι άντωνυμίαι, οίον το εγώ, ΰύ, ΐ; 
 andagain,ii. ρ. 66,1.28: το εμείο καν 6εΐο καΙ εΐο, ει μεν πρωτότυ- 
 ποι εΐύιν άντωνυμίαι άπο τον εγώ καΐ 6υ καΐϊ κλινόμεναι,διατης ει 
 διφθόγγου γράφονται. So also Dionysius the Thracian (Bekk. Anecd. 
 p. 640): πρό(3ωπα πρωτοτύπων μεν εγώ — ύύ — " — άρι%μοΙ δε πρω- 
 
€HAP. I.] AND OTHEE PRONOUNS. 261 
 
 τοτνπων eviKog μεν εγώ — 6v — ζ δνικος δε νώΐ — όφώϊ,Λλη^^νντίκοζ 
 δε ημείς — νμεΐς — ΰφεΐς — πτώόεος δε πρωτοτντζων μεν ορΟ'ης εγώ — 
 ΰν — ΐ,γενίκης δε εμον — βον — οϋ,δοτικης δε εμοί — (5οί — οΐ^ αίπατίκης 
 δε εμε — όε — δ. The Scholiast on this passage says(p.916Bekker): τον 
 τρίτον 7tQo6w7tov εΰτίτοί,καΐ ΰημαίνειτο ovTog η εκεΐνος,κα^' εαντο 
 μεν κειμενον ενρεΟ'η ονδαμον, εουκε δε άτΐοτηςτών'Αττίκων χρήόεως 
 εΙρηΟ^αυ τΐαρα τούτω τω τεχνογράφω. εκεινοόΐ γαρ λέγονΰι, και ον- 
 τοΰ.ί. That this Scholiast is mistaken, appears from the words of Apollo- 
 nius Dyscolus (dePronomine, p. 69 c): t ταντην oi μεν φαΰί τϋαράλο- 
 γον, οτι ον δια τον ν ' άτίοβολ^ γαρ τον ΰ κατά το τρίτον άποτε- 
 λεΐύΟ'αί όον, ον, dog, og. He proceeds (p. 70) : άξωττίΰτότερος δε 6 Σο- 
 φοκλής μάρτνς χρηΰάμενος εν ΟΙνομάω' εΐ μεν ώΰεΐ Ο'άΰύονα είδώς 
 el τέκοί τΐαΐδα. The Venetian Scholiast, on the Iliad xxii. 410, reads 
 this fragment of Sophocles somewhat differently, thus: η μεν ώΰεΐ^'άΰΰο' 
 ηδε ώβΐ τεξον τΐαΐδα. From a comparisonof these two readings, William 
 Dindorf has emended this corrupt lina as follows : rj μεν ως Υ d'aouov^ , 
 7] δ' ώς ϊ τεκοί τΐαΐδα, adding, " it is said of two women, each of whom 
 was boasting of her son's fleetness. Why they did so, is clear from the 
 well-known story about (Enomaus." It is hardly necessary to add that 
 Τΐαΐδα is either followed by a vowel in the next line of which it is the 
 first word, or else that it is added by ApoUonius to explain the line. 
 Bekker thinks [Comment. Crit. p. 337) that l might properly be sub- 
 stituted for ε in the following passage of Plato (Sympos. p. 175 c): 
 μετά ταντα εφη ΰφάς μεν δειτΐνεΐν, τον δε Σωκράτη ονκ εΙ(5ιεναί. 
 τον ονν ^ Αγάθωνα τΐολλάκις κελενειν μετατΐεμαβ^αυ τον Σωκράτη^ 
 ^ δε ονκ εάν, and appears also to suggest the insertion of this obsolete 
 nominative in a passage at the end of the same dialogue (p. 223 d): 
 τον ονν Σωκράτη — άναΰτάντα άτΐνεναι, καΐ αντός, ωΰτΐερ είώϋ-ει 
 £Τίε6%^αι, where almost all the MSS. omit «t'rog (Bekker's Comment. Crit. 
 p. 362). With regard to the former passage it may be observed, that, 
 according to the rules of Syntax, Twould be better grammarthani'; that, 
 in the passages quoted by Heindorf, on the Euthydemus^ § 72, and on 
 the Sophistes, § 42, the verb is always οϊμαί or ηγονμαο, and that 
 οίμαί με by no means justifies φημί με, as Stallbaum (ad loc.) seems to 
 imagine. In the latter passage the ϊ might so easily have been absorbed 
 by the last letter of the preceding και, that we cannot doubt it would 
 be a better reading than αντός, which is disallowed by the MSS. In 
 the Attic dialect we find the pronoun ΐ used very frequently as an 
 inseparable and uninflected affix to the indicative pronouns odf, 
 οντοζ, εκείνος. This termination is long and carries the accent; 
 thus, οδί, ovtog, εκεΐνοβί; also in other cases, as τοντονί, 
 ταντηΰί; plur. οίτοΐΐ, avralt, ταντί, &c. And so also adverbs, as 
 
262 THE PEESONAL [bOOK I. 
 
 Ιν!&αδί, &c. It is right to remark in passing that the form ενταν^ΌΪ 
 has been imported into the texts of the Attic writers, and even of 
 Homer, by grammarians, who were misled by a prima facie appear- 
 ance of analogy. The fact is that d'a, θεν, Ο'ην, dOV, O't, &ε are the 
 only allowable forms of the affix -%•-. 'Evtav%ol is just as impossible 
 as λοΟόΓ would be. There can scarcely be any doubt that when Ινταν^οΙ 
 is found in the Attic writers, we should substitute ενταν^ί, and when 
 it has been intruded upon Homer we should write εντεν^'εν=εντανΟ'α 
 (above, § 114), which he uses in the same sense. The two following 
 are important passages relative to the pronoun t Apollonius Dyscolus 
 (de Pronom. p. 10b): άντωννμίαι άναφορίκαϊ η τε Γ, ον, ε. Priscian, 
 XIII. 2, § 7: Quceritur etiam illud, cur, quum apud Grcecos tertia per- 
 sona pronominum primitivorum et in singulari numero nominativum 
 habeat /t, et in plurali όφεΐς, apud Latinos sui et nominativo deficit, et 
 pluralia separatim non hahuit. § 8: Apud Grcecos nominativus su- 
 pradicti pronominis, id est, fi, rarus est in usu. These authorities, 
 combined with those quoted above, are sufficient to establish the fact, 
 that there was originally a nominative to the reflexive pronoun ov, ol, 
 θ, and that this nominative was Ϊ. We need not wonder that this ϊ 
 should be of rare occurrence. There is very little demand for the nomi- 
 native of the reflexive pronoun, which is, in most cases, used objectively. 
 The nominative of the Latin reflexive pronoun sui, sibi, se, never occurs, 
 nor has that pronoun any plural number. Now what is the reflexive 
 pronoun? Nothing more, we believe, than a form of the second pro- 
 noun, pointing at once to some person or thing close at hand, and it 
 is always used in connected speech to refer to the person last spoken 
 of, when that person is considered in any way as a subject, especially 
 in relating the words of that person. The idiom of the Latin lan- 
 guage admits of such phrases as di:uit se venturum, but the Greek, 
 which is much more accurate in its syntax, always requires that, if the 
 subject is expressed in the nominative in one member of the sentence, 
 the same case should be continued in the dependent member; accord- 
 ingly, the pronoun is either suppressed, as is generally the case, for 
 example, Μοιροκλης ονΟ'εν εφη Λονηρότερος είναι (Aristot. Bhet. πι. 
 10, § 7), or, if emphasis requires it, as in the line of Sophocles quoted 
 above, or distinctness in a complicated sentence renders it necessary, 
 as in the two passages from Plato's Symposium, a nominative case of 
 the reflexive pronoun would of course be used if there were one ; and 
 we have clear testimony that there was one. The very fact that 
 there was no plural of the Latin reflexive pronoun shows, as all un- 
 declined parts of speech always show, that the word itself, and the 
 peculiar use of it, belong to the oldest state of the language. As, 
 
CHAP. I.] AND OTHER PKONOUNS. 263 
 
 therefore, it is manifest η 'priori that the reflexive is nothing more in 
 its nature than a demonstrative pronoun indicating nearness of posi- 
 tion, we must seek for its equivalent among the oldest forms of the 
 demonstrative pronouns. . We have irresistible evidence that there 
 was a pronoun Ϊ as well as a pronoun Ϊ. The Scholiast on Dionysius 
 (quoted above) says as much, though he confused it with the reflexive 
 pronoun mentioned by the author on whom he was annotating, and 
 whom, like most other Scholiasts, he made a point of misunderstand- 
 ing. Its existence is farther proved by Hesychius' Glosses: IV. αντήν^ 
 αυτόν. KvitQioL. — siv. SKHVog (leg. εκείνον); by what Lascaris says 
 (de Pr onomine, III. τρ. 3 A4): τον τρίτον Λροβώτίον η ονομαότίκη κατά 
 JtaXaLOvg ϊ (leg. Γ) και og, and by the remark of Draco (p. 106) : ^ I' 
 άντωννμία η ΰημαίνονόα τρίτον τίρόβωΛον βραχν εχευ το l ; for if 
 ϊ was short, it could not have been the same with if, which we know 
 from the passage of Sophocles, and from the terminations εκεινοξ^ν^ 
 Όντοξ -t, &c., to have been a long syllable. Moreover, the conjunction 
 €t, "by this that" = "on this condition" — is, as we shall show, the 
 dative of I', just as the Latin si is the dative of sis; the pronoun also 
 evidently enters into ϊ-γνητεξ^ the Rhodian word for αν^ίγενεΐζ, i. e. 
 *'born in that place,''^ into l-(Jog, ϊ -og, ϊ-ωρος, and ϊ-διος, and we have 
 seen its appearance in Latin and Sanscrit. The evidence indeed for 
 this pronoun is so strong that Hermann and Bockh have not hesitated 
 to introduce IV, as the dative, into the text of Pindar (Pyth. iv. 36; 
 Nem. I. 66). It is quite clear, then, that the Greeks had two pronouns, 
 the stronger aspirated form ϊ being used to express the reflexive rela- 
 tion, i. e. relation to something near and immediate; the I', a synonym 
 for αντόξ-, to denote something in which the idea of nearness was not 
 so directly implied ; and, therefore, to return to the common terminology, 
 ϊ is a complete, and ϊ a shortened or mutilated form of the second 
 personal pronoun, which appears in the ultimate analysis as the stem 
 J^a. The same was the case in Latin. The common pronoun i-s ex- 
 presses the weakest demonstrative relation, the equally common pro- 
 noun hi-c signifies "that which is near : " the termination c is a mutila- 
 tion of the ordinary affix ce, so that we should write ^wn-ce, not hunc-ce. 
 This termination is by no means confined to the singular, as some 
 scholars have imagined. In good writers we have hi-c for hi (Varro, 
 VI. 73), and hce-c for hw (Plautus,^«ZwZan«,in.5,59; Terence, Eunuch. 
 HI. 5, 34 ; Phormio, v. 8, 23; Yarro, v. 75, &c.). Now this pronoun hi-c 
 stands for the Latin si-c by the common change from the sibilant to 
 the aspirate, and to the same root belong se-d, si, the forms in Ennius 
 sa-m, sap-sa, su-m, so-s, and, we have no hesitation in saying, also the 
 reflexive su-i, si-hi, se. It is highly interesting to know that in the 
 
264 THE PEESONAL [bOOK II. 
 
 Rig-veda we possess the weaker forms i-m (also an old Latin form) 
 and i-d, and also the stronger form si-m. 
 
 140 This supposition that the reflexive pronoun is identical with 
 that of the second person, and merely indicates nearness of place, is 
 fully borne out by the use of ε in Homer, where it occurs as a demon- 
 strative pronoun implying nearness, and is used in the singular and 
 plural of all genders: thus Iliad i. 236 : 
 
 vol μα τόδε όκητΐτρον το μεν οϋποτε φνλλα καΙ oiovg 
 
 φνΰεί, ετίείδη τΐρώτα τομην εν ορεόύι λελοϋττεν, 
 
 οϋδ' άναΟ'ηλήόεί, ττερί γαρ ρά ε (nam circa hoc) χαλκός ελε'ψε 
 
 φύλλα τε καϊ φλοίόν νυν αυτέ μιν vhg 'Αχαιών 
 
 εν Λαλάμτ[]ς φορεονΰϋ' 
 
 when we see that τόδε, ε, and μυν are used with reference to the same 
 object and in the same manner. And in the Hymn to Yenus, 
 V. 268: 
 
 τγjΰL δ' αμ η ελάται ηε δρνες νιρικάρηνοι 
 γεινομεν]ΐ6ιν εφνόαν ετά χ^ονί βωτιανείρ^ 
 καλαί, τηλε^'άουΰαι, εν ουρεόυν ν^Ι^ηλοΙόυν 
 εΰτάό' ηλίβατοί, τεμένη δε ε κίκληΰκονΰίν 
 αθανάτων {hcec autem vocant). 
 
 141 With regard to the use of ε both as singular and plural, it is 
 to be observed, that while we have αμμε, νμμε, constantly in the 
 plural, we have εμέ, (ίέ, invariably as singular forms. We may re- 
 mark, too, that the endings of the datives plural αμμυν, νμμιν, of the 
 datives singular εμίν, τίν, γίν=^ίν{=ΰοί, Hesychius), and of the ac- 
 cusatives μίν, VLV•, IV, are the same : 6cpLV is both singular and plural : 
 for its plural use see Matthise's Note on Herodotus, p. 285, and for 
 its use in the singular, see Horn. Hymn. Pan. 19; ^schylus, Per see, 
 761; Sophocles, (Ed. Col. 1847, and Reisig, Enarr. p. clxxxi; Butt- 
 mann, Lexilogus, Vol. i. p. 60. It is very easy to conceive why there 
 should be identities in the case of pronouns, where there are such 
 marked distinctions in the other parts of speech ; the fact is certain, 
 and it is an additional proof, if proof were needed, of the exceeding 
 antiquity of these little elementary words. If the pronouns were, as 
 will be shown hereafter, used for the purpose of distinguishing the 
 cases and numbers of nouns, it is obvious that in the original pro- 
 nouns there could be no consistent distinctions of that kind. 
 
 142 It will be observed by every reader of the Greek writers, 
 that the ordinary dual and plural forms of the reflexive pronoun were 
 
CHAP. I.] AND OTHEB PRONOUNS. 265 
 
 identical witli the second persanal pronoun, in form, and with both 
 first and second personal pronouns in signification. Thus we have 
 ΰφωέ, "they two," όφώϊ, "ye two;" μετά ΰφίΰίν=^με^^ νμΐν (Horn. 
 Iliad X. 398); ΰφέτερος, (1) "yours," Hesiod, ε. κ. η. 2, and else- 
 where; (2) "mine," only in Theocr. Id. xxv. 163; (3) "thine," 
 Theocr. Id. xxii. 67; (4) "ours," Xen. Cyr. vi. 1, § 10. In the same 
 way we find words into which the element ε enters, used to express 
 the first and second persons; thus Moschus, Idyll, iv. 77: μηδέν όε 
 χερευότερον φρεΰΐν ηΰυν βτεργειν ("my own heart"); Horn. Iliad. xvs.. 
 11 L• (5v δε φρεβίν ηΰιν lavd'yg ("your own heart"); Odyss. i. 402: 
 κτήματα δ' αντοζ εχοΐξ καΐ δώμαΰιν oIolv άνάΰΰοις ("your own 
 house "). The same is the case with the cognate Sanscrit word svayam, 
 and its possessive adjective svas. Now it may be asked with regard to 
 this fact in Greek and Sanscrit, why it happens that such an extension 
 of meaning is given to the reciprocal in those languages, while in Latin 
 the same pronoun is never used, except in the third person. The 
 fact is, that the Romans made but a sparing use of their pronominal 
 words in comparison with the Greeks, as indeed appears from their 
 unacquaintance with that particular sort of pronoun called the article, 
 and from the infinitely greater number of pronominal particles in 
 Greek and Sanscrit. As a natural consequence, we find in Latin a 
 greater restriction upon the free use of these particles and pronouns, 
 and a greater uniformity in the employment of them, for, not possess- 
 ing an extensive variety of pronouns, they frequently used as distinct 
 words what were only difiOrent articulations of the same form. Thus, 
 while they employed suus (in Ennius sus) = svus (Greek (?φό^, San- 
 scrit svas)^ as the possessive of se^ they turned the by-form όφετεροζ 
 into vester (compare ΰφηξ with vespa), just as they converted 6φώϊ 
 into vos. It must be observed, that the use of the reciprocal in the 
 third person, is infinitely more frequent than its use in the first and 
 second persons ; and the reason is plain : for as all pronouns denote 
 relations of place, and the primitive pronouns difiOr only in the 
 degree of nearness to the hei'e, a pronoun expressing, as the reciprocal 
 does, a special nearness, would be more naturally and more frequently 
 superseded by the first and second personal pronouns than by the 
 third, in other words, there might be a necessity for the expression of 
 a particular kind of nearness in the third person, which could arise 
 but seldom in the case of the other two persons. 
 
 143 We have before shown how the stem of the second personal 
 pronoun (element fa) enters into ΰφώϊ; it will be equally easy to 
 point out the etymological connexion of 6φέ and ε. That the latter 
 was a digammated word is well known, and we have showTi before 
 
266 THE PERSONAL [bOOK II. 
 
 how often the digamma was a representative of the double sound sv^ 
 as in ηδνς^^ηδνξ, compared with svddus (Sanscrit), suavis=svadvis 
 (Latin) ; and SKVQOg = fbKVQog: compared with ςναςηνα (Sanscrit). 
 The Latins dropt the labial in se, and vocalized it in sus; or omit- 
 ted the sibilant as in vos and vester; similarly the ordinary Greek 
 omitted the labial, and softened the sibilant into an aspirate. This 
 intimate etymological connexion between the reflexive or reciprocal 
 pronoun, and that of the second person, throws very great light on 
 both. It was to express an idea of relative nearness that the second 
 of the old pronominal roots was originally employed: it is the same 
 idea of relative nearness that constitutes the distinction between ε 
 and εκείνον, between hie and tile, between questo and quello. 
 
 144 The inverted Dorian forms ψέ, ^tV, deserve notice from their 
 constant appearance in Latin. "We have in Ennius sapsa, and in 
 Plautus eapse; in the former word the enclitic is inflected as well as 
 the personal pronoun, and in more modern Latin the form i-pse 
 always retains its first syllable unaltered, while the -pse is subjected 
 to inflexion in every case. This alone should be a confutation of 
 those who fancy something essentially accusative in ε. The same 
 may be said of ε-τερος, which must be connected with this pronoun, 
 and cannot be derived from gV, as Bopp seems to suppose (Demon- 
 strativstammej p. 14). 
 
 145 We come now to the most important part of this sub- 
 ject: namely, the discussion of the indefinite, interrogative, and 
 relative pronouns. In the Latin language these three pronouns 
 contain the same element ; the question is whether this is also 
 the case in Greek and Sanscrit, a question to which we should 
 be inclined to seek for an affirmative answer, as well from the 
 analogy of the Latin, as from general considerations. In our 
 own every-day language we constantly use a demonstrative for 
 a relative, and we must be conscious to ourselves of frequently 
 using a demonstrative sentence, with a difference of tone, in an 
 interrogative sense. This alone would lead us to believe that 
 there cannotbe any radical difference between the demonstrative 
 and these particular kinds of pronouns. A little investigation 
 will satisfy us that in the Greek and Sanscrit languages they are 
 all etymologically identical, and all connected with the pronoun 
 of the second person (element /a). 
 
 146 The Latin interrogative and indefinite are both writ- 
 ten quis; the relative is written qui. In these words therefore 
 
CHAP. I.] AND ΟΪΗΕΚ PRONOUNS. 267 
 
 the root is qv- or kv-. It has been already shown (§§ 110,121) 
 that a double consonant-sound like kv may be superseded by a 
 single representative of one of its two constituent parts. This is 
 particularly the case with the digamma sound, which, we have 
 shown, was this same compound sound kv=^kp. There are 
 many instances in which this compound sound in Latin words 
 is represented in Greek, Sanscrit, and Gothic, or one of them, 
 by one of its elements; for example, we have Latin co-qv-o 
 compared with the Sanscrit pach^ and the Greek τίίτίω ; o-qv- 
 ulus compared with the Sanscrit aksha^ and the Greek όμμα 
 =οππα; lin-qv-o compared with λείτίω; qv-atuor^ and qv-inqv-e 
 compared with the Sanscrit chatur^ and panchan^ Greek %k- 
 τορε^, reroQEg (τεΰΰαρες)^ and πεντε^ πέμπε', a-qv-a compared 
 with the Sanscrit ap^ Gothic ahva; se-qv-or^ compared with 
 the Sanscrit sajj^ and Greek έπομαι. We may also compare 
 the Latin an-gv-is with the Sanscrit ahis^ and the Greek εχις. 
 Similar changes have taken place even in the same language; 
 thus, to take an instance in point, the Oscans, according to 
 Festus, WYoiepitpid for qv-id-qv-id., and the terminations -qvam., 
 -ce were identical with -piam^ -pe. The guttural element ce^ 
 which thus appears as a substitute for the labial pe., was 
 further softened into hi, as is shown by a comparison of hi-c 
 "this," a-5, ci-tra., "on this side," ci-terio and ci-timo. In 
 fact, qui., si-c^ hi-c, is., are four forms of the same pronominal 
 root, signifying relative proximity, in which the guttural element 
 has successively degenerated. Accordingly, if all the Sanscrit 
 and Greek forms of the relative, interrogative, and indefinite, 
 are resolvable into one or other of the elements of this compound 
 Latin consonant, we are entitled to conclude in favour of their 
 original identity with one another. The full form is preserved 
 in the Gothic hver., hva; we pronounce the labial only in which., 
 what, and the guttural only in who., how. 
 
 147 We now turn to the Sanscrit forms. This language 
 has three interrogative stems, ka^ ki., ku: thus, from the first 
 and second, to, ka., kim=^quis, quae, quid? from the second, 
 kiyan, kiyati, kiyat=quotus, quota, quotum? and from the third, 
 kutas ==: undef kutra and kva = ubif From the second of these 
 interrogative stems comes, by the softening process which is 
 
268 THE PEKSONAL [bOOK II. 
 
 always going on in languages, the indefinite chit, just as church 
 from kirk^ chamhre from camera^ &c. This particle, also writ- 
 ten chana when added to the interrogative, gives it the sense 
 " any one," " whosoever," " a certain person/' just like quis-quis, 
 &c. in Latin: thus kach-chit (from kat-chit^ used as an inter- 
 rogative particle like the Latin an and nuni)^ kag-chit, kag- 
 chana^=^quis'p%am. The copulative conjunction cha^ also from 
 this root, agrees as well with the Latin qvc as with the Greek 
 τε, the connexion of which is otherwise shown by a comparison 
 of the forms ore, ττότε with δκ«, ιττόκα. It appears unsoftened in 
 the Veda-forms makis^ nakis=nequ{s (softened again in the Zend 
 7ndchis^ nae-chis)^ in ma-Mr, na-kir, rtia-kim, na-kim=msi^ non 
 (Colebrooke, Gramm. p. 121); and so also in the old Pelasgo- 
 Etruscan, if we are right in supposing that ne-ke=^ne-que in 
 the Hexameter inscription at Naples first printed by Lepsius 
 (die Tyrrhenisch. Pelasger^ p. 42), which we divide thus: mi 
 ni Mulve neke Velthu ir Pupliana, and render, "I am not Mulva 
 nor Volsinii, but Populonia." A comparison of these words 
 with ni-hil, ne-qvid^ together with the analogy between hi-c 
 and ci-s before pointed out, can leave no doubt in our minds 
 as to the connexion of these terminations with the second pro- 
 nominal element /a. The Sanscrit relative is yas, ya, yat; the 
 y standing for the aspirate in og, ^, δ, according to what we 
 said upon yushme und νμ,Βΐξ. That the demonstrative meaning 
 entered largely into this relative, appears from a comparison, 
 of the Zend demonstrative yim, Latin jam, "at this time," 
 with the relative sense preserved in yadi, " when," and in yadi, 
 "wherein" = "if" (comp. bI and si). It will be remembered 
 that the German wenn signifies both "when" and "if," and 
 we shall see (below, § 205) that bI may, in the combination 
 ΒΪ Λοτε, bear a temporal meaning. It may be concluded, then, 
 that the Sanscrit interrogative, indefinite, and relative, spring 
 from the guttural part of the digamma, which is the initial of 
 the second pronominal element. 
 
 148 If we even confined ourselves to the Greek language 
 alone, we should have no difficulty in recognising the connexion 
 between the relative and demonstrative. Greek Syntax teaches 
 us that the relative oV, ri, δ, is only a later and more emphatic 
 
CHAP. I.] AND OTHEE PEONOUNS. 269 
 
 form of the distinctive pronoun or definite article δ (og), ^, τό. 
 In Homer we find the latter both as relative and antecedent ; 
 thus (//. I. 125): άλλα τα μεν πολιών ε| ετίράΟΌμεν^ τα δέδαΰ- 
 ται: and even in Attic Greek, the distinctive pronoun is occa- 
 sionally used for the oblique cases of the relative*, just as our 
 "that" appears instead of "which." Generally, however in 
 the more fully developed syntax of the language, og, ^', o, as 
 relative pronoun, is limited in its application to some sentence 
 containing a finite verb, in close conjunction with which it 
 forms a periphrastic definition or description of some object 
 considered as otherwise well known, or else, which is the highest 
 refinement of syntax, it makes some general assumption or 
 supposition. But whether the antecedent, or object referred to, 
 is definite as in the former case, or indefinite as in the latter, 
 the relative sentence exists only by virtue of its antecedent; in 
 other words, it is a syntactical contrivance which plays the same 
 part as the adjective or genitive case; and Mr. Garnett, in the 
 paper already quoted {Essays^ pp. 214 sqq.), has collected in- 
 stances from various languages in which the affix of the genitive 
 case is manifestly identical with the relative pronoun. The 
 Semitic languages, which, as we have before remarked (above, 
 § 49), are in a tertiary or merely syntactical state, and have 
 consequently lost their apparatus of inflexions, show, more clearly 
 than even the inflected languages, that a demonstrative or in- 
 dicative pronoun is the vehicle or instrument by which human 
 speech expresses the connected, if not concurrent, notions of a 
 relative sentence, an adjectival epithet, and a genitive case. It 
 is well known to every Hebrew scholar that the noun to be 
 expressed in the genitive case is placed unaltered after the 
 governing noun, which, being affected by the contact, is said 
 to be in the construct state. Not unfrequently the qualifying or 
 genitive noun has prefixed to it the distinctive pronoun hal, 
 which serves as a definite article, and sometimes a relative sen- 
 tence takes the place of the genitive. Thus mizmor being '^a 
 
 * Some have attempted to limit this use to the neuter gender, as well 
 as to the oblique cases. But there are authorities for the use of the mas- 
 culine {xovi Eurip. Bacch. 712, τον?, Androm. 810) and feminine (την, Soph. 
 Track. 47). 
 
270 THE PEKSONAL [bOOK ΙΓ. 
 
 psalm," we might express the phrase, "a psalm of David" by 
 (a) mizmor-David, (b) mizmor had-David, (c) mizmor ^hasher or 
 she le-David^ which would be in Greek, (a) ψαλμο-^αβιδ^ (b) 
 'φαλμοξ b ^αβίδ, (c) τ^αλμος og τω ζίαβίδ. In Chaldee, Syriac, 
 Samaritan, Ethiopic and Arabic, the demonstrative pronoun di^ 
 de, za, and dsa, which is regularly used as the relative in these 
 idioms, is as regularly employed to mark the genitive relation, 
 and we have elsewhere* pointed out instances in which the 
 cognate Hebrew pronoun zeh is used in the same manner. For 
 example the LXX translate mi-p^ne-' helohihi zeh-^Jinai^ by 
 άτΐο τίροόώτίον τον Θεον τον Συναί. The most instructive fea- 
 ture in this usage is the occasional corroboration of the distinc- 
 tive particle by another pronominal element, which is properly 
 synonymous with it, but which is prefixed by way of antecedent 
 when the demonstrative import is superseded by the relative. 
 We have seen instances of this in the relative ^hasher itself, and 
 in the compounds a-ha-m^ e-go. And the genitive may be ex- 
 pressed not only by the relative ^hasher or she^ but also by the 
 article and relative ha she^ and even with a double determi- 
 native, as in ''heth ha she. It has therefore been rightly sug- 
 gested by Mr. Garnett that the termination a in Ethiopic 
 construct nouns, and the -i and -w in Hebrew and Arabic, 
 are derived from pronominal elements of a demonstrative and, 
 ultimately, a relative nature. As the guttural in the Semitic 
 languages constantly subsides into the vocal chirik^ there is no 
 reason why Abi-meleh should not represent an original Ah-ham- 
 melek. We must return to this important subject when we come 
 to speak more at length of the adjective; at present enough has 
 been said to show that there is no reason for doubting the con- 
 nexion between the relative and the second pronominal element 
 indicating proximity. Indeed, there are some languages, in the 
 ultimate condition of departure from the original etymological 
 structure, which express the strongest form of the relative sen- 
 tence either by the correlation of two pronouns expressing near- 
 ness, or by placing one of these in the relative clause: thus in 
 
 * Proelectio Philologica in Deborce Canticum Triumphale, Cantabrigiae, 
 1848, p. 11; Jashar, Berolini, 1854, p. 337. 
 
CHAP. I.] AND OTHER PRONOUNS. 271 
 
 the instance* quoted by Buschmann (Humboldt, uber die Kawi- 
 Sprache^ ni. nr. 720), either of the relative sentences, quifidelis 
 amicus est^ is fidelis in rebus adversis est; or quifidelis in r. a. 
 est^ is est fidelis amicus; is expressed, is fidelis amicus est, is 
 fidelis in rebus adversis est. And in the celebrated Chinese say- 
 ing, usually attributed to a greater than Confucius, we find the 
 demonstrative used instead of the relative only: J&isopuyo^ ue 
 sJa yu jin, which is literally: "ipse hoc non cupis, ne inferas 
 hominibus." In the most interesting and important application 
 of the relative construction, namely, to the formation of hypo- 
 thetical propositions, we shall see that the Greek not only 
 opposes the antecedent civ to the relative f/, but occasionally, in 
 the older poets, introduces κε in the protasis as well as in the 
 apodosis. 
 
 149 The indefinite and interrogative pronouns in Greek 
 are both written rtg, originally τυ-ν-ς^ the distinction between 
 them being that one is an enclitic, the other accentuated; the 
 one being written after, and the before, the word to which 
 it refers: in the former case, the want of accent unites the noun 
 and its corresponding indefinite so closely, that they may fairly 
 be considered as one word. That the first part of nv-g is not 
 connected with the third pronominal root ta, as might be sup- 
 posed on a casual inspection, appears, to a certain extent from 
 the fact, that their uses in Greek are absolutely and diametri- 
 
 * We have taken this and the other example from Dr. Steinthal's 
 tract entitled "de pronomine relativo commentatio philosophico-philolo- 
 gica cum excursu de nominativi parti cula" (Berolini, 1847), p. 87. This 
 author, who is an enthusiastic disciple of W. von Humboldt, applies his 
 master's principles to an investigation of the use of the relative chiefly 
 in the Chinese and African languages. With regard to the Indo- Ger- 
 manic idioms he states specifically (p. 101), "relativum in his Unguis e 
 demonstrativis et interrogativis ortum esse." But he defines the relative 
 generally as the demonstrativum formate (p, 84), and says that whereas the 
 personal suffixes denote coincidence^ and the demonstratives Cincluding case- 
 endings and prepositions) signify dependence^ the proper use of the relative 
 is to express inherence (p. 21). In all this there is rather an accumulation 
 of words than a clear statement of definite notions. The relative is 
 neither more nor less than a particular logical employment of the pro- 
 noun signifying proximity. 
 
272 THE PEESONAL [bOOK II. 
 
 cally opposite. We are not s^^eaking here of the nominative, 
 masculine and feminine, of the Greek article, which we believe 
 to be of the same origin with the relative, &c., but of the ob- 
 jective cases which tlq externally resembles. The article, o,^, ro, 
 is a pronoun which would not in all cases express with sufficient 
 definiteness any particular object, even though that object may 
 have been mentioned before: the name of the object is there- 
 fore added to avoid a vague generality. Conversely, when it 
 is desirable to express that some class is known, but not a par- 
 ticular individual of that class, the general attributive noun is 
 put first, and the indefinite word after it. And thus, if it is 
 necessary to lay particular emphasis on the class, the individual 
 of which we do not know, a Greek would not hesitate to prefix 
 an article and affix an indefinite to the same word; thus we 
 have, in a distributive sentence, in Eurip. Med. 1141 : 
 
 KVVBL d' δ μεν Tig χειρ ^ b ds ξαν^'ον κάρα 
 Ίίαίδων — 
 
 and similarly in Herod, n. 60 : ai μεν τυνεξ των γυναικών κρό- 
 ταλα εχονύαι κροταλίξονύί, ot δε ανλεονΰί^ αϊ δε λουτταΐ γυναίκες 
 καΐ άνδρες αείδουύι^ and αί μεν τίνες των γυναικών ποιευΰι τάπερ 
 εϊρηκα^ αί δε τω^άζουόυ.) κ. τ. λ., where the τίνες implies, that the 
 women are not divided into two sets, previously arranged and 
 regularly distinguished, as if the indefinite pronoun were omitted, 
 and the sentence were simply distributive ; but that the women 
 with the castanets are any members of the party who happen to 
 be so furnished. In the same way we have the indefinite after 
 a noun with the article prefixed, as in Soph. (Ed. Tyr. 107: 
 
 τούτου θανόντος νυν εταΰτελλει ΰαφώς 
 τους αυτοέντας χειρί τιμωρεΐν τινάς^ 
 
 i. e. "the murderers (for we know he was murdered)^ whoever 
 the particular persons may be (for we do not know that) : " and 
 thus CEdipus immediately asks οϊ δ' εΐΰΐ που γης ; " where are 
 they?" a question which shows how the interrogative might 
 have arisen from this use of the indefinite, with merely the 
 change of tone indicated by the accent: γυνή τις — τις γυνή; 
 "a woman somewhere" — "where?" But, if τις is not con- 
 nected with the stem ta^ how is it to be explained? It is 
 sufficiently obvious that it is not the primary and genuine, but 
 
CHAP. I.] AND OTHEE PEONOUNS. 273 
 
 a secondary and corrupted form of the original interrogative 
 and indefinite. We find traces both of the labial and of the 
 guttural element of the original fa even in the existing state 
 of the language : we have the former in the Attic words %ον^ 
 jtol, Λοτε, Tto^L, ΰΐόΟΈν, 7to(jog, Ttolog, τΐότερος, &c., and of the 
 latter in their Ionic equivalents καν, κοΓ, κότε, yod'i^ κόΟ^εν, 
 κό(?05, κοΓοί, πότερος, &c. The latter of these two forms 
 appears immediately in the Latin word cur, and also in (c)ubi, 
 (c)unde, (^c)uter^ (c)ut^ as we may fairly infer from the com- 
 pounds ali'Cuhiy ali-cunde and from a comparison of the 
 Latin amo with the Sanscrit kamayami. That these words, 
 however, are only the older and more genuine forms of the 
 interrogative rtj, appears from a comparison of τετορε^, τε, 
 οτε &c. with ττέτορε^, και, δκα, &c., and from the identity of 
 TcivtB with quinque, Sanscrit panchan^ &c.; so that rtg is 
 merely a corrupted form of Ptg, the course of the changes 
 being from sv through hv to du, τν, and rt. We have shown 
 that the Sanscrit relative yas is another form of the Greek og 
 (= (Jog), and it appears that the indefinite and interrogative rtg. 
 Tig are identical with chis, Ms. In fact, ng is nearer to the 
 Latin synonym quis than ττε is to que^ for the guttural element 
 is very sufficiently represented by Ti==j. The full crude form 
 of TLg is of course τι-ν-^ the latter part being the third element 
 under the type na *. In other words, it is a compound of two 
 pronominal elements, like elgz=B-v-g^ κεΐ -vog^ rrj-vog «-νά, &c. 
 It bears, therefore, the same relation to og, yas, qui^ that Zt^V, 
 Janus^ Tuscan Tina^ does to Ζενς^ Ju-piter, Jov-is. As this 
 termination -v- is emphatically demonstrative, it might, at first 
 sight, seem strange that the interrogative and indefinite pronoun 
 should be distinguished in the Greek language only by such an 
 affix. But this very affix appears as the regular interrogative 
 particle in the Latin -ne^ and as a proclitic we have en^ not only 
 as an interrogative particle, in en-unquam, ec-quis =^ en-quis^ but 
 also to call attention, in the sense "look here ! " which is, in 
 itself, a form of question — e.g. in en and ecce = en-ce^ Greek ηνί. 
 
 * This addition, which Schomann regards as unessential, in order to 
 bring in his strange etymology, T-ig = is-te (Hofer's Zeitsehrift, i. 2, pp. 
 250, 1), has caused some difficulty to Lobeck also (see Varronianus, 143, 
 note). 
 
 Τ 
 
274 THE PERSONAL. [bOOK II. 
 
 It is on this account that the Greek interrogative is less adapted 
 than the Latin to take the place of the relative in a dependent 
 interrogation ; it generally assumes the compound form oerig when 
 it is thus used, and there are very few passages in which τίζ 
 appears for the relative indefinite. In iEsch. Sept. c. Theh. 417 : 
 τοίώδε φωτΐ πεμτε rig ξνΰτήΰεται, τίξ άνδρα κομτΐάξοντα μη 
 τρεβαζ μενεϊ^ the editors suppose a recurrence to the direct inter- 
 rogation; and in Callimachus, Epigr. xxxi. 2: εχ^αίρω το ποίη- 
 μα το κυκλικον ονδε κελενΟ'ω χαίρω, τις itoXXovg ώδε καϊ ώδε 
 φέρει, we ought perhaps to read χαίρω οτυξ. It is difficult to say 
 why the tenuis τ in this word, in the pronoun τν — τΡε-, in the 
 terminations -ng, -τνς^ and in τετορεξ^ τε, &c., has taken the 
 place of the medial δ or medial aspirate -O", which are generally 
 the dental representatives of the sibilant : but there is abundant 
 reason to believe that this substitution has been made ; indeed, 
 as we shall see in the following chapter, the more original form 
 of rtff, namely, δεις — δεν-ξ or δειν-ς^ has not quite vanished from 
 the language, and δε==^ δΓε still stands by the side of τε = τΓε. 
 In general, we remark a tendency to substitute the tenuis for the 
 medial, which is especially manifested in the altered value of O". 
 And, in regard to the particular case, as the natural changes, 
 which are always at work upon a language as long as it is a 
 spoken one, continued longer in operation in Greek than in 
 Sanscrit, we are entitled to conclude that the Greek relative was 
 also, in the Pelasgian state of the language, identical with the 
 indefinite and interrogative. As an illustration of the affinities 
 between the Greek and old Persian, to which we have already 
 adverted, it is worthy of remark, that the relative in the Behistun 
 inscription, as deciphered by Rawlinson, it written both tya 
 and hya. 
 
 150 The original demonstrative power of the Greek inter- 
 rogative stem κ («) is still preserved in a number of words in 
 common use, such as κα-τα, κεΐ -vog^ καί^ κεν (κα Doric), and γε 
 (γα Doric). This root may also be recognised with a demon- 
 strative power in the modern French ce, ci, ce-tte, ce-ci, ce-lle, 
 and in the Italian cL The use of ci, as the accusative plural of 
 the first personal pronoun, is an additional illustration of the 
 demonstrative nature of that pronoun. 
 
CHAP. I.] AND OTHER PRONOUNS. 275 
 
 We must bestow a little consideration upon some very important 
 words indicating relations of time into which this root kv (under 
 certain regular varieties of form) enters with the demonstrative sig- 
 nification. The Sanscrit word for "yesterday" is hjas^ the Greek 
 χ%'Εξ^ the Latin her-i from hes-i (hes-ternus, Sanscrit hyas-tanas), and 
 the Gothic gis-tra. All these words manifestly contain the same 
 element kv-^ represented by hy- in the Sanscrit word, by χ^•- in Greek, 
 by the common aspirate in Latin, and the guttural in Gothic. A 
 comparison of χ&ών^ χ&αμαλός, χαμαί, and humus, shows that χ^' is 
 occasionally represented by ;^ or Λ only. We do not believe that the 
 syllable -as is, as Bopp suggests (Vergl. Gramm. p. 568), a mutilation 
 of divas. In a word of such common use, an adjective pronoun 
 signifying nearness might be used without any substantive, just as 
 bruma (= brevima) is used for "the shortest day" without any addi- 
 tion of dies. The same root also enters into the Sanscrit word gvas 
 (^=kvas), Latin cras=.Gsas ; these words imply nearness as well as 
 hes-, &c., but as the nearness is predicated with a prospective and not 
 a retrospective reference, a different form of the same root has been 
 adopted. The word ^erew- die, "on another day" (para, "another," 
 Sanscrit), should also signify "tomorrow," but all-powerful custom has 
 assigned to it the meaning "on the day-after-tomorrow." The word 
 vesper, Βΰτίέρα, "evening," is made up of the pronoun ves (=hes=ichthes 
 z=gis=hyas), and the pronominal adjective para, per a, which we 
 have seen in peren-die, only here para is used in the sense of "late" 
 or "after," as in pardhna, "the afternoon," "the latter part of the 
 day" (from para, "after," and alian, "a day") ; the initial pronoun, 
 retaining its signification of nearness, is applied by another change in 
 the association to a part of the present day, and vesperus means "this 
 day late" or "after this day." These transitions by association are 
 all so many facts; the reason for them, though easily explained, are 
 most easily felt; and it is better to investigate these curious time- 
 adjectives by the application of such a simple principle, than by the 
 hypothesis of almost impossible mutilations, as Bopp does. The word 
 ήμερα itself, the second part of which (μερ-) is evidently the element 
 of μερθ£, may be connected with the Sanscrit root dya=ja (com- 
 pare 'η7ΐCίρ,jeGur, yak-rit, &c.): so that η-μερ-α will signify "the light," 
 "the part of the twenty-four hours during which the sun shines." 
 Or rather, to go farther back, it may be the preposition δίά, dya=ja, 
 which appears in ημίΰν^^δίά-μεΰος, and which we have no hesitation 
 in recognising in the adjective ημερog, the regular antithesis to αγρυοζ. 
 Thus, too, we must compare ηγεμών, ί^γεΐϋΟ'αί, with διάγω, δίοτίοζ 
 (iEsch. Pers. 44), &c. ; for the preposition δνά seems particularly 
 applicable to the functions of a leader or guide. We consider ήμερος 
 
 Τ 2 
 
276 THE PERSONAL [bOOK II. 
 
 as originally predicable of a country through which there was a road 
 or passage, a country divided by a road (^LάμεQog), just as άγριος was 
 properly applied to a rude open country with nothing but αγροί. 
 The Grammarians distinguish between άγριος and άγρύος^ and confine 
 the local meaning to the latter (Bachmann, Anecd. ii. p. 375, 29 : 
 άγριος καΐ αγρύος^ τίοιψίκόν, διαφέρει' άγριος ο ωμός' άγρεΐος 
 δε δ εν τώ άγρώ); but this does not affect the etymology of the older 
 form. And the primitive meaning of the correlative word τ^μερος is 
 shown by the following passages: -^schylus, Eumenid. 12 — 14: 
 
 Λεμτίονΰι δ' αντον καΐ βεβίξονόιν μέγα 
 κελεν^οτίοιοϋ Λαίδες ^Ηφαίΰτον^ ypova 
 άνήμερον τι^εντες ημερωμενην. 
 
 Plato, Legg. p. 761 A: οδών τε ετίΐμελουμενονς^ οτίως ώςημερώτατα^ 
 εκαΰται γίγνωνται; Philostr. vit. Soph. p. 552 : ώνηΰε δε και το εν ry 
 'Ιταλία Καννβιον ημερών ας νδατι μάλα τούτον δεόμενον: com- 
 pare Pindar, Isthm. πι. 75 (ιν. 97): 
 
 και βα%νκρήμνου τΐολιάς άλος εξενρών ^εναρ, 
 ναντιλίαιΰί τε Λορ^μον άμερώΰατο. 
 
 which refers to the passage of Hercules through the straits; and for 
 the etymology, οοτα^Β,γβ'Ήτΐειρος with its epithet δίατΐρνΰιος (Pindar : 
 Nem. IV. 51), on the principle pointed out by Lobeck (ad Soph. Aj 
 254, p. 193). The effect of such road-making on civilization appears 
 clearly enough from what Aristotle says of the Herculean way, and of 
 the protection afforded to those who travelled upon it (τίεριΰ'ανμαβίων 
 άκονβμάτων, c.85,p. 837 Bekker)*. We consider dies, dya^ &c. of pro- 
 nominal origin, like the particle δη, which generally refers to time. 
 
 151 From the facts here stated, we are entitled to con- 
 clude, with regard to the pronouns in general, that they were 
 all originally demonstrative ; that there are three primitive pro- 
 nouns ; that the second of these, which indicates nearness to the 
 here^ gave birth under the form fa to the reflexive pronoun and 
 the relative, which are identical with it; and that different modi- 
 fications of this same second personal pronoun were subsequently 
 used to express all relations of nearness, till at last in modern 
 Italian a word formed from the same root came to be used to ex- 
 press the here itself, that is, the first personal pronoun. 
 
 * We have treated this important subj ect at greater length in the Varronianus, 
 pp. 268 — 271. On the connexion in meaning between ayqag (which contains 
 the same roof as άγων, ayoqa) and χο^ός, χ&^ος, see below, § 229. 
 
CHAP. I. AND OTHER PEONOUNS. 277 
 
 152 The adjectives formed from pronouns constitute an interest- 
 ing subject immediately connected with that which we have been 
 discussing. In regard to the possessives formed from the personal 
 pronouns, it appears singular that, while those from the second and 
 third person are regularly formed from the genitive case, the posses- 
 sive of the first pronoun, εμός, though it contains the adscititious ε 
 which we have noticed as an occasional prefix to the objective cases 
 of the pronoun itself, omits the ε in the second sylla-ble : there is, 
 however, reason to believe that Homer used the form μεος from 
 μεο, like the Latin mens from mei, which is quite regular, and εμύς is 
 only a corruption of an original εμεός^ from εμεο for εμεΐο (§ 134). The 
 Sanscrit possessives are madiya, "mine," tvadiya, "thine," tadiya^ 
 "his." The first syllable of the two former is identical with mat^ 
 tvat, the ablatives of the pronouns. The only ablative of the third 
 person which is in use is tasmdt, but it is very likely that a shorter 
 form, analogous to the ablatives of the other two pronouns, was once 
 in existence; at all events, constant use would easily generate such 
 an abbreviation in the compound. It will be shown in a future 
 chapter that dia or tia is the full or original form of the ablative 
 ^ffix, from which, of course, these possessives are derived. This ter- 
 mination is also found in the Greek adjectives of quality olog (b-uog), 
 utolog {7to-Log)i rolog (το -Log)^ the d or s having dropt out, just as the s 
 has been absorbed in the cognate genitives in -oto, in which we should 
 expect -ΰίο on the analogy of the Sanscrit genitive ending -sya; com- 
 pare εδίδούο, εδίδον, δίδοίΰο, δι,δοΐο (Βορρ, Vergl Gramm. p. 220). In- 
 deed, as we shall show hereafter, -δLog and -uiog are by-forms of the 
 genitive or ablative case. The d is preserved in ϊδLog from the pro- 
 noun Ϊ. This termination in Greek is properly and regularly used to 
 denote derivation^ kind, or quality : we might, if we liked, consider 
 the Sanscrit affix as compounded of the demonstrative and rela- 
 tive, and signifying "that which ; " but it is better to refer this form 
 to the genitive or ablative ending. The Greek pronominal adjectives 
 signifying quantity are o-^og, Tto-uog, t6-6og, &c. We have elsewhere 
 pointed out that the termination -6og signifies "a collection" or "ag- 
 gregate," even when found as a part of nouns, and it obviously bears the 
 same sense in these pronominal words. But how did it get that sense? 
 It cannot be doubted that ϊ -uog has the same termination. Now it is 
 well known that this word was originally pronounced "uvog, and even 
 Pi(5pog, as appears from the necessities of the metre in Homer, and 
 from the gloss γίβγον, ϊΰον in Hesychius. The labial was, however, 
 dropt in the course of time, and 'iuog appears in all the later poets 
 with the first syllable short, so that this is no reason for denying its 
 connexion with o-uog, &c. in point of termination. As Y-oog and 
 
278 THE PEBSONAL AND OTHEE PEONOUNS. [bOOK II. 
 
 τόΰοξ are synonyms, both signifying "so great," i. e. "equal," an ex- 
 amination of the common part in these words will lead to a proper 
 interpretation of the affix 6og=J^os=^6φόg. This word implies "all 
 that belongs to the person near us," a meaning which includes in it 
 the idea of a collection or aggregation ; and this is just what we want. 
 The Greek terminatic^ -λίκος or -At|, (ήλί-ξ, ομηλυξ) runs through most 
 of the sister-languages. Thus we have in Gothic svaleiks, hve-leiks, 
 Anglo-Saxon tliy-lic, German ahn-lich, so-lcher (so-like, such) and in 
 English like. In Latin many very common adjectives are formed with 
 the termination -Us : e. g. ce-qua-lis (from ce-quu$, Sanscrit e-kas, "that 
 which"), rega-Us, viri-lis, missi-Us, humi-lis, simi-lis, fame-licus^ &c. 
 The vfovdife-Ux contains the full form of this termination; the first part 
 of the word is connected, as Bopp suggests ( VergleicJi. Grammatik, 
 p. 606) with the Sanscrit hhaj, not, as others have supposed, with 
 fe-tus, fe-tura, fe-mina, &c. It is related to fau-stus, fav-or, &c., and 
 the Greek φάΓος (ντίο-φαν-ΰις). It might be written φαϋλίξ in Greek 
 characters*, like φαυ-λο^, "light," "blown about by every ' breeze" 
 {^ά^ω, Fav-onius)', and signifies literally "light-like," i. e. "brilliant," 
 "splendidf." 
 
 * We are aware that the surname of Claudius Felix is written Φηλιξ 
 in Josephus, xx. 6; Act. Apostol. xxiv. •, Suidas s. v. Κλαύδιος. This is 
 not, however, an etymological transcription, but only an attempt, like the 
 ώκ αγε (hoc age) of Plutarch, to represent the sound of the Latin word. 
 
 t On the connexion of "light" and "air," see Book iv. ch. 5. The 
 Latin adjective dives conveys the same idea as felix: see Journ. of Philol. ii, 
 p. 354. And perhaps we may also compare he-atiis with φαητός. 
 
CHAPTER Π. 
 TFIE NUMERALS. 
 
 153 Why numerals have lost their original signification. 154 Connexion of the 
 first numeral with the first personal pronoun. 155 Similar affinities of the 
 second numeral and second pronoun. 156 Origin and explanation of δπξ 
 and δείνα. 157 The tJiird numeral and its use as a comparative affix. 
 158 The fourth numeral compounded of the first and third. 159 Why the 
 first four numerals are declined in Greek, and the others undeclined: eight 
 the dual of four, 160 The sixth and seventh numerals how related. 161 The 
 fifth, ninth, and tenth numerals are to be referred to a decimal system of 
 computation. 162 Views of Lepsius on this subject, and on the higher 
 numbers. 163 Vague expressions for large numbers. 164 Ordinals, and their 
 connexion with comparatives and superlatives. 165 General comparison 
 expressed by the affix -ίων. 166 Comparative words, Βηο)Ά.2(.^ημίΰυ,μέ-6θξ, 
 άλλος. 167 Superlatives in -ια-τοζ derived from adverbs in -lq. 
 
 153 OINCE it is the tendency of inflected languages to be- 
 come more and more abstract, as they develop them- 
 selves syntactically by means of writing, and, by striving after 
 generalization, to lose the immediately perceivable meaning of 
 their individual words, we might expect that this tendency 
 would soonest be realized in the numerals*. The use of numerals 
 at all is an abstraction, and one of the highest kind; it is 
 stripping things of all their sensible properties and considering 
 them as merely relations of number, as members of a series, as 
 perfectly general relations of place. Hence it is, that the short- 
 
 * There have been many important treatises on the subject of the 
 numerals. The most valuable are those by Lepsius (Ueher d. Ursprung u. d. 
 Verwandtschaft der Zahlworter in der indogermanischen , semitischen u. d. 
 koptischen Sprache, Zwei Sprachvergl. Abh. Berlin, 1836) and by Pott (Die 
 quinare u. vigesimale Zahlmethode hei Volkern aller Welttheile, nehst ausfuhrl. 
 Bemerkungen uber die Zahlworter Indogerm. Stammes u. einem Anh- ilber 
 Fingernamen, Halle, 1847). The second of these treatises is to a certain 
 extent controversially opposed to the former; and though we are in- 
 debted to Dr. Lepsius for many interesting details in the present chapter, 
 we think his leading principles (pp. 92 sqq.) untenable, and we have here 
 followed up the views respecting the classification of the pronominal ele- 
 ments, which we have set forth in the preceding chapter. We have made 
 the Hebrew numerals the subject of a special investigation in a tract en- 
 titled Maskil le Sopher, London, 1848, pp. 41 sqq. 
 
280 THE NUMERALS. [bOOK II. 
 
 hand of written language has arrived at its completion in nu- 
 merals sooner than in any other words; for while all other 
 words are expressed by combinations of letters, the words ex- 
 pressing abstract number have, in all languages, a set of distinct 
 symbols or cyphers for their expression. In the language of 
 Algebra the same method has been carried so far, that we can now, 
 by a systematic combination of single letters, carry on the most 
 comjDlicated analytical reasonings in all sciences based upon one 
 or other of our primary intuitions of space and time. 
 
 154 It was hinted in the last chapter, that the first three 
 numerals are the three personal pronouns; this we shall now 
 endeavour to prove by considering them in detail. 
 
 The root of the first personal pronoun is ma; it signifies 
 "that which is here." The natural connexion between the ideas 
 of here, that which is near to the here^ and there^ and the num- 
 bers "one," "two," "three," needs no formal exposition: the 
 vulgarism "number one" as a synonym for the first person, 
 and the proximus sum egomet mihi of the Latin comedian, 
 speak in the plainest terms for this identity. Our business is to 
 establish the etymological fact. 
 
 The Greek word expressing the number one was a regularly 
 inflected adjective; in the ordinary writers we find εΐ^, μία^ fV, 
 but in Homer the feminine is written ϊα (Iliad iv. 437, xiii. 
 354, XXI. 569), in which, as well as in the ordinary masculine 
 and neuter, we discern no traces of the first pronominal root 
 μα, Doderlein (Lat. Synon. iv. p. 52) supposes that ϊα is con- 
 nected with μία as eo is with meo^ and ολαί with mola. The 
 following ivestigation will show that there is no connexion be- 
 tween la, or, as it should be written, t«, and μία. 
 
 Man is naturally led to adopt one of two methods of arith- 
 metical reckoning : the decimal, suggested by his own hands 
 and feet, and the duodecimal, derived from the twelve moons. 
 The latter system was of more frequent use in ancient times 
 than it is now, though we still have our dozen as a distinct 
 term, and still divide the day into two portions of 12 hours 
 each, and carry the same division into our tables of weights and 
 measures. Among the Greeks, the prevalence of this system is 
 more strongly shown by the 12 gods, most of them clearly con- 
 
CHAP. Π.] THE NUMEEALS. 281 
 
 nected with the months, and by the frequent occurrence of the 
 number 12 in political subdivisions. We find that most ancient 
 states had some regulative number vi^hich was the basis of their 
 social organization. "Twelve," says Niebuhr, "was the funda- 
 mental number of the lonians, which appears in their towns in 
 the Peloponnesus and in Asia, as well as in the Attic τρίττύεζ. 
 Their primary number was four; then each quarter was sub- 
 divided into three" (Hist, of Rome^ ii. p. 20). He should 
 rather have stated that the primary number was twelve, a num- 
 ber suggested by the changes of the moon, and that this number 
 was divided into three tetrads. The influence of this subdivision 
 upon the formation of the numerals will be seen by and by. 
 But if twelve was the regulative number of the lonians, and if 
 this number was, as is highly probable, suggested by the num- 
 ber of the moons or months, should we not expect that the 
 words for an "unit" and a "month" would be identical? Now 
 the Ionian word for a month or moon is μείς (Homer, Iliad 
 XIX. 117; Hymn. Merc. 11; Hes. L κ. η. 559; Herod, π. 82), 
 and this is also found in ^olic (Pindar, Nem. v. 82, comp. 
 Suidas and Zonaras under the word μείς). The feminine μία 
 therefore perfectly corresponds with this form. And we shall 
 proceed to show that the common particle μεν is the regular 
 neuter of μείς = μέν-ς. 
 
 A full discussion of all the usages of μεν^ as a conjunction, 
 belongs rather to the syntax of the conjunction than to the 
 present subject*. We shall now insist only on those of its uses 
 which most strikingly show that it means "the first thing," "in 
 the first place." This amounts almost to a certainty when it is 
 considered, that, in its regular use, it is always opposed to ds, 
 which can be proved to mean "in the second place." It is also 
 proved by this circumstance, that μεν never stands alone without 
 suggesting the idea of something that is to follow. Thus, when 
 Socrates is going to catechize Meno's slave, he asks the master, 
 "Ελλην μεν Ιότυ καΐ ελληνίζεί', "He is a Greek, I suppose, and 
 talks Greek" (Plato, Meno, p. 82 b). Here an εΐ δε μή is ob- 
 viously implied : "if he is not, he will not serve my purpose of 
 questioning him:" so also in the answer jcavv μεν ovv, which is 
 
 * See Greek Grammar, Art. 559-568. 
 
282 THE NUMEEALS. [bOOK II. 
 
 SO common in Plato's dialogues, there is a manifest suspension of 
 part of the sentence: "you are right, but what then?" (τι ά' 
 hTtSLta;) and there is always an expectation of something ulte- 
 rior in the use oi μεντοί in answers: e. g. Aristophanes, Eguites^ 
 890: τον κανλον oL•^ εκείνον του βιλφίον τον αξίον γενόμενον; 
 "Do you remember the fall in the price of laserpitium ? " where 
 the Demus answers, otOa μεντοι — " To be sure I do, and what of 
 it?" In such phrases as η ύοΐ μεν ημείς πανταχη δρώντεξ 
 φίλου (Soph. Antig. 634), the other part of the opposition {εΐ 
 μηδενϊ αλλω) is so obvious that it is omitted. There are also 
 cases where μεν standing alone recalls the idea of the first per- 
 son, and is nearly equivalent to the Italian phrase in quanta a 
 me^ e. g. Plato, Crito^ p. 43 D: οντου δη άφίκται^ άλλα δοκεΐ 
 μεν μου ηξευν τήμερον. In other passages it means "first of 
 all," "above all others," as in Homer, Iliad v. 893: "Ηρης^ 
 την μεν εγώ ΰπονδη δάμνημ επέεόΰίν. From all this it 
 appears certain that μεν is the neuter of μείς^ which bears the 
 same relation to men-sis that %•είς (βέντς) does to Ο'εντος (Bockh, 
 Staatshaush. ii. p. 395). 
 
 We have therefore μείς^ μία^ μέν^ as complete in all its parts 
 as slg, fcci, έν, and containing the elementary pronominal form 
 με> We find the same root and with the same signification in 
 fiovoff, ^^only," "one-ly" (Ionic μοννος^ Doric μώνος)^ which 
 answers also to the Gothic possessive meins. The ϊα men- 
 tioned above is obviously connected with the second prono- 
 minal element: compare t, Λ^-c, &c. : there are many coinci- 
 dences in use between the first and second elements; see above, 
 §§135, 150, &c. 
 
 The first Sanscrit numeral ekas is, as we have before sug- 
 gested, related to aliam^ the nominative of the first pronoun; 
 it is represented by the Greek εκας^ εκάτερος^ εκαΰτος^ and is 
 probably formed, as Bopp supposes, from the demonstrative 
 stem e, and the relative or interrogative kas^ with the meaning 
 "that which:" we shall speak again of this word in a future 
 chapter. Bopp has attempted to find a further remnant of the 
 Sanscrit numeral in the word cocles^ which is explained as "one- 
 eyed" (Plin. N. H. xxxvii. 55: codites qui altero lumine orbi 
 nascuntur), and which he would consider as a compound of ca 
 for eka and oculns. In a similar manner he would explain the 
 
CHAP. Π. THE NUMERALS. 283 
 
 Gothic JiaiM^ and the Latin ccecus, which he writes caicus. 
 Grimm also compares κυκλω^ρ {GescL• d. deutsch. Spr, p. 255). 
 It appears to us that cocles^ which has the same formative ending 
 as aries, miles, paries, &c., is derived from cceculus^ a diminutive 
 of ccecus', and we have no objection to consider κύκλ-ωφ an- 
 other derivative of the same kind : cf. τΐρύλεες with proelium^ &c. 
 LuscKS seems to be λοξός with the common inversion of the ele- 
 ments of ξ; and both Codes and Ao^iag may be proved to mean 
 "an archer." The ordinary Greek dg = evg is connected with 
 the Sanscrit demonstrative e-na (aind), with the Gothic aina^ 
 and with the Latin unus^ most anciently written oinos^ by 
 the substitution of an unaspirated long for an aspirated short 
 vowel before explained, just as ehas and εκαξ are connected. 
 The same word also occurred in Greek (see the Commentators 
 on Hesychius, suh vv. οϊνη and οίνίξειν)^ and we have it with 
 an s instead of the aspirate in the Latin words sem-el'^ sim- 
 plea;, sem-per, and sin-gulus, just as εκαξ appears under the form 
 secus in the same language. 
 
 155 It is clear that the first Greek numeral contains the first 
 pronominal element; it is no less so that the word expressing 
 the number two is identical with the second personal pronoun. 
 In the last chapter it was suggested that the three original pro- 
 nouns would probably be the three tenues^a, ka=^qa^ ta; that 
 the first might be represented by the cognate sounds ma or να, 
 and the second by that double sound J^a, a combination of the 
 guttural and labial, which so often appears in certain languages 
 of the Indo-Germanic family, where we have only a labial or a 
 guttural in the others. We have seen that in some cases the 
 second element is represented only by rP, τν, or τι. It was also 
 mentioned that we might extend or modify the signification of 
 these elements by combining them with one another, or with the 
 element 7'a, denoting motion or beyond. Thus, the compound 
 tva-ra might represent the third position, which might also be 
 expressed by the third element alone. Now it is the corruption 
 tv which constitutes the usual form of the second personal pro- 
 noun ; and this form of the second pronoun exactly coincides 
 
 * Pott (Zahlmethode, p. 156) derives α-πα| from πήγννμί; cf. επιμίξ, 
 αναμϊξ, Sec. 
 
284 THE NUMERALS. [bOOK II. 
 
 with the second numeral. In most cases, however, the more 
 ancient d has not been superseded in the numeral by the tenuis 
 i, which takes its place in the personal pronoun. Thus, we have 
 in Sanscrit dvau, in Zend dva^ in Greek δνω^ δνο {=^dJro)^ in 
 Latin duo (dvo)^ but in Gothic tvai. In this, as in other cases, 
 where two consonants make one sound, we frequently find one of 
 them standing as a representative of both (§ 121). Sometimes 
 the dental is omitted, as in αμ-φω (αν a άνΡω), in Μκατί^ vi-ginti 
 and vin-gati instead of άΓεί-κατί^ dvi-ginti^ dvm-gati: so also in 
 helium^ bonus, Mn% his, hes*, from dvellum, dvonus, dvini, dvis, 
 dves. At other times, on the contrary, the labial is dropt, as in 
 ^f, άέω, dig^ δι,βόόΰ^ δώ-δεκα^ δίμήτωρ (Sanscrit dvimdtri). 
 
 It has been remarked that the origin of the second personal 
 pronoun is the idea of comparative nearness. An examination of 
 the second numeral whill show that the same is the case with it. 
 That δε is the shortest form of this numeral, is proved by its 
 constant use in the obvious sense of "in the second place," and 
 by the verb δεω=^δεΓω^ "to bind" (compare twine, two). Be- 
 sides, the numeral δυο was also written δνε^=δΡε'. this might be 
 inferred from the Attic form δυεϊν\.^ and we clearly read it in an 
 Arcadian inscription (1511, 1. 7, Bockh): μνάζ δύε και τριάκοντα. 
 Now this particle δε is often used in composition to express com-, 
 parative nearness. In this sense it appears in ο-(ϊε, "the man 
 near" (§ 135), &c. It is also used to express motion towards, or 
 a tendency to become near, as ΊηΟλύμτίον-δε^ "towards Olym- 
 pus," οϊκον-δε, "homewards," ^A^^qvat,ε=^Ά^r^vaq-δε., "towards 
 Athens J." It is found with the same meaning in άενρο^άδΡρο, 
 a word which requires some explanation. We have before re- 
 marked on the change of place to which the digamma is liable: 
 
 * See Salmas. d. Mod. Usui'arum, p. 252. 
 
 f On the difference between δνεΐν and dvolv see Bachmann, Anecd. ii. 
 p. 372, 2: δνεΐν δίχα τον ccq^qov γενικ'ης επαγόμενης, ε ν,αϊ Τ' τοΐν 
 δ ν οι ν δε μετά. τον αρϋ'ρον, ο καί Ι (cf. π. ρ. 390, 31; Blomfield ad 
 Prom. 803; Dorville, Chariton, p. 527; Person, Advers. p. 94). 
 
 J Since the above was first published we have seen an ingenious at- 
 tempt by the late Professor Hunter to connect δε and -δε, at and ad, two, 
 to, and too ("A grammatical Essay on the nature, import, and effect of 
 certain Conjunctions; particularly the Greek δε: read June 21, 1784." 
 Trans, of the R. Soc. of Edinburgh, Vol. i. pp. 113—34). On the relation 
 of -δε to -&εν, see below, § 263. 
 
CHAP. Π.] THE NUMEEALS. 285 
 
 there is nothing singular, therefore, in the change from dJre- to 
 deF-. That such a change has taken place in this root, is mani- 
 fest from the fact, that δεύτερος in the only ordinal of δνω^ and 
 that άΡεω, to bind (which we have shown to be immediately 
 formed from this root), is intimately connected with δέομαι = 
 δένομαι. The word δε^^ρο signifies "in this direction," δεύτερος^ 
 "a man who is nearer to us than another man," and δεντατος "a 
 man who is nearest to us of a series of men," i. e. "the last," 
 and thus it is used as a synonym for νΰτατος. The Latin se- 
 cundus is only a lengthened form of sequens.^ as will be shown 
 in a future chapter. 
 
 156 There is another word of the highest interest connected 
 with the second numeral, which these combinations will enable us to 
 explain : we mean the pronoun δ, r], το δεΐνα^ or, as we would write it 
 (after the analogy of οδε, ηδε, τόδε), — οδεινα, ηδεινα, τόδείνα. This 
 word signifies, that, though we know perfectly the particular person 
 or thing we are speaking of, we either cannot or will not mention 
 the name: it was, therefore, natural enough that a word, signifying 
 proximity, should be added to the personal pronoun, to refer to a 
 person or thing definitely conceived, but indefinitely mentioned. Now 
 we have seen that the first personal pronoun, when used to express 
 the first numeral, was lengthened from ^ε- into μείς=^μένς. We 
 should expect, therefore, that the second pronominal root άΡε or τ^ε 
 would be analogously lengthened into δΓείς = δΓένς when used to 
 express the second numeral. This termination -vg (-v\o\g) was, as we 
 shall see under the prepositions, a strong expression of locality, and 
 this sense is highly appropriate for a transfer of the weaker relations 
 of space which constitute persons, into those stronger ones which 
 originated the numerals. Let us inquire then, if there ever was such 
 a word as δεις. The author of the Etymologicum Magnum says 
 (p.639,1. 11, Sylb.): Ουδείς. Ίύτεονοτυτονονδείς,οτεΙύοδνναμεϊτω 
 οντις, δύο μέρη λόγου εΐύί,τό τε ου, καΐτο δεις. ουδέ γαρ έΰτΐ ΰνν^ενον. 
 ει γαρ τ,ν ύύν^ετον, ημελλε ττρο μιας εχειν τον τόνον. τίαν γαρ όνομα 
 μονοΰύλλαβον ΰνντι^^έμενον αναβιβάζει τον τόνον — τταΐς, εϋπαις' 
 χΟΌον, αυτόχθων Θράξ, Σαμό^'ραξ' χο^ρίς tol• τιτώξ, τΐολυπτώξ. — 
 αύτοΰ δε του ουδείς τό ούδέτερον, δέν, χωρίς της ου παραΟ^έΰεως 
 εχομεν πάρα Άλκαίω εν τω ένάτω, Κούδεν εκ δενος γένοιτο, Ζηνό- 
 βιος. See Mullach, Qucest. Democrit. p. 362. So also Choeroboscus 
 (Bekkeri Anecd. p. 1362): δεν, οτίερ Ισοδυναμεί τω τι. In fact, 
 as we have suggested above (§ 149), δείς = δέ-ν-ς is really the 
 older form of τίς=^τί-ν-ς. The word μείς has the flexion ^ε^- 
 
286 THE NUMEEALS. [bOOK II. 
 
 vog (Bockh, C. I. i. p. 741), as well as μενός (Choerobosc. in 
 Theodos. i. p. 200, ed. Gaisford). Similarly, we find dstvog^ as well 
 as dsvog, from dsig. There is no more difficulty therefore in the adverb 
 δείνα from δεις, than in ίνα from stg*. But, besides this adverb, we 
 find traces of a regular declension : thus we have gen. δεινός, dat. δεΐνί, 
 accus. δείνα, in the singular, and nom. ^gii'fg, gen. δεινών, accus. ^εΓι^α^, 
 in the plural. The form of the dative plural may be inferred from 
 the forms τοΐς-δεΰί, τοΐς-δεΰΰί. These forms are all regular inflexions 
 of δεις, just as μεινί, which is found in an inscription, is formed from 
 μείς. We may, therefore, reasonably infer that there was originally 
 such a Greek word αΒδείς=δένς corresponding to μείς=μενς, and that 
 when δ δείνα is used in the nominative case, the second part must be 
 considered as an adverb. In regard to the genitive δεινός as com- 
 pared with δενός, quoted by Zenobius, we may remark that there was 
 also a form είνα for ενα. See Lex, de Spir. p. 240, and Buttmann's 
 Mythologus, Vol. ii. p. 142. 
 
 From these two words μείς=μενς and δείς=δένς, we have μ'ήν, 
 δήν, and μη, δη, which we shall discuss in a future chapter. We 
 have also μην, μηνός, "a month," as well as μείς, μεινός. It may be 
 thought singular that while μεν preserved the final consonant, it is 
 dropt in the correlative δε. It is to be remarked, however, that in 
 words of such common occurrence, the shortest forms would naturally 
 be preferred, unless there were some reason to the contrary, as there 
 is in the case of μεν, which would otherwise be confounded with the 
 pronoun ^£, whereas no confusion could take place between the second 
 personal pronoun and such a modified form as δε. Of the omission of 
 V in such cases, we have other instances in κεν, κε, ένεκεν, ενεκε, 
 πρόΰ^εν, πρόΰΟ'ε, and probably -δε, -Ο'εν. 
 
 An objection has been made by Buttmann (Ausfilhrl. Sprl. § 70, 
 Anm. 7, note) to the derivation of ουδείς from ov and δεις, namely, 
 that the forms ουδεμία, ουδέτερος and ουδετΐοτε, manifestly contain 
 οϋδε. Now it is also a theory of Buttmann's that ουδείς, ου%εν are 
 the masculine and neuter of this same ουδεμία, the δ being turned 
 into a 0' by the^contact of the aspiration, just as is the case in oO"' 
 Έρμης (found for οό' Έρμης in an old inscription, Bockh's Corp. In- 
 script. I. p. 32), and as Thiersch would write in Pindar πενταετηρί& 
 οτζως, Έλλά^ ευρηβεις (Thiersch's Pindar, π. p. 349). It is, there- 
 fore, unnecessary to suj)pose, because an ουδέ μία implies an ουδέ εΙς, 
 which indeed occur as two words in the older Avriters, and as one 
 
 * Schumann (Hofer's Zeitschr. i. 2, p. 249) suggests that ό δείνα is a 
 combination of o, δέ, and ha: and Mehlhorn (Gr. Gr. § 110) identifies 
 δεις with εΐς, to which it ultimately reverts, although the use is widely 
 different. See Choerobosc. in Theodos. p. 199, Gaisford. 
 
CHAP. Π.] THE NUMERALS. 287 
 
 word — ovd^eig — in the more recent authors, that there could not be 
 such a compound as ov-deig. The only question is, whether there is 
 such a word as dsig. If so, and it appears clear that there was, ov-deig 
 is just as allowable as ον-δέ or ov-dl-eig*, 
 
 "With these uses of the particles βε, ddg, &c., we may compare the 
 collocation δη XLg=:nescio quis (Heindorf ad Platon. Phced. §. 130). 
 
 157 The root of the third numeral in the Indo-Germanic 
 languages is t-j- r with a short vowel either interposed or sub- 
 joined, according to the etymological rule that a vowel may be 
 sounded either before or after a liqoid. In Sanscrit we have 
 trayas, tisras, trini, in Greek τρεΓ^, τρία^ in Latin tres, tria. We 
 do not know the nominative (threisf) of this numeral in Gothic, 
 but the genitive, dative and accusative are thrije, thrim^ thrins. 
 In Latin we have also ter^ ternio^ and tertius. 
 
 If the second numeral has arisen from the idea of nearness, 
 the third must be the expression for that which is farther. The 
 third personal pronoun ta does indeed express the there^ but for 
 the third numeral a form was required in which a relation to 
 the second was definitely given, and therefore the particle ra 
 was added to the form denoting the second numeral. This 
 particle, which we shall examine hereafter more minutely, 
 expresses the idea of motion from or beyond, the point from 
 which the motion is supposed to begin being indicated by the 
 pronominal element to which the ra is subjoined. If then, as is 
 most probable, the composite t-{-r is the corruption of an original 
 tva-ra^ the second numeral is the parent of the third. To this 
 point we must return, when speaking of the fourth and sixth 
 numerals. The combination t{^yra is used in Sanscrit as an affix 
 to pronominal roots, when distance, whether definite or indefinite, 
 is implied : thus we have amutra^ "on the other side," ku-tra^ 
 "where?" It also denotes direction or tendency, and in this 
 sense it is found in the Greek adjectives oQeu-TSQog^ ayQO-reQog^ 
 δημό■τεQog^ &c. In Latin this root appears in the preposition 
 
 * The existence of δεις is still questioned by Pott {Zahlmethode, pp. 
 152, 3), on grounds which seem insufficient in themselves. To say nothing 
 of the passages which prove that this word was actually used, it appears 
 to us that a sound theory respecting the particle δε would almost lead us 
 to assume this inflected form. It is surely a most unscientific proceeding 
 to suppose, as Pott does, that δε is a mutilation of ώδε. 
 
288 THE NUMEEALS. [bOOK II. 
 
 tra-ns, signifying "beyond," and it is also affixed to pronominal 
 stems as in Sanscrit; thus we have iil-tra^ "on that side," 
 ci-tra^ "on this side." It appears too in the word terminus, 
 "a limit," which has the form of a passive participle, and may 
 perhaps be referred to the verbal root tr which is formed from 
 this pronominal word, cf. τραω, in-trare, &c. The most im- 
 portant, however, of the uses of this word is as a suffix, indicating 
 the comparative degree in Sanscrit, Greek, and Latin. Thus 
 we have ha-taras, Tto-TSQog^ u-ter. In this use the idea of 
 "beyond" is also involved. Thus we are told that in the 
 Chinese language, which has no inflexions, hou yoong kwo gno 
 ("he is more vehement than I"), may be translated literally 
 "he is vehement beyond me" (^Quarterly Review, Vol. L. p. 187). 
 The Hebrew method of expressing the comparative degree is 
 not altogether dissimilar, e. g. pbln^j lia, melior Balaqo, is lite- 
 rally "good above or beyond Balaq." The suffix ta-ra, as is 
 well known, is used when only two things are to be compared, 
 and this was its original force when employed as the third 
 numeral : for the first numeral signifies, like the first personal 
 pronoun, "that which is here," the second "that which is near," 
 the third "that which is farther." Now far and near are rela- 
 tive terms ; and though, for the purpose of expressing a person 
 who is neither / nor you, an indefinite there would suffice, the 
 number "three" must be considered more distinctly in its 
 relation of contrast to the number two. Hence it is that the 
 idea of there was extended to that of relatively greater distance, 
 when applied in direct and particular contradiction to the idea 
 of nearness contained in the number two. This comparative 
 ending sometimes appears under a form still more like the 
 common third numeral, as in άλλο-τριοσ, for which, however, 
 the ^olians also wrote αλλό-τερο^ or «λλό-τερροί;, as they also 
 wrote TBQTog for τρίτος (Choeroboscus, apud Cramer. Anecd. ii, 
 p. 275, 23), κόττερα for κόπροα and Περαμος or Πέρραμος for 
 Πρίαμος {Etymolog. Magn, p. 529, 1. 22, p. 665, 1. 40; Gregor. 
 Corinth. 639 and 907). 
 
 158 That the Indo-Germanic word for the number four is 
 composed of those for one and three is clearly proved by the 
 following combinations. The oldest Greek form was τΐί-τορες 
 
CHAP. II.] THE NUMEEALS. 289 
 
 or TtL-ovQeg; the first syllable bears the same relation to ^ε- 
 that 7ΐέ-δα does to με-τά^ and the remainder of the word is only 
 another form of τρεΐς=^τΓα-ρες or ΰ^α-ρες. The Sanscrit form 
 for this numeral is, masc. chatvdras^ fern, chatasras^ neut. chat- 
 vdri, where the feminine appears to be anomalous ; now the same 
 anomaly is found in the feminine tisras of the third numeral; 
 it is therefore clear that the last two syllables of the fourth nu- 
 meral comprise the third. The same appears also from a com- 
 parison of the Latin ter with qua-ter, ter-nus with qua-ter-nus, 
 and tri-duum with qua-tri'duum. With regard to the first syl- 
 lable of the Latin numeral , it is a mutilation of the Sanscrit 
 numeral ekas^ "one," which, as we have seen, is synonymous 
 with ce-qV'US and secus: this u will show why -qva stands for 
 'ka in the Latin word for "four." We have before pointed 
 out how ki became softened into chi (§ 147) ; such a softening 
 would most naturally take place in an abbreviated form like 
 chatvd^'as. By the side of the strong form chatvdras we have 
 a weaker form chaturas. In Gothic we have fidvor and fidur^ 
 dogs, just as we have qvatvor and quaternus in Latin. It will 
 be observed that we generally have o, v, or u in the second part 
 of the word signifying "four," although the labial does not 
 appear in the common word for "three." We have suggested 
 before that the relation of "three" was expressed by adding 
 the particle ra to the second numeral: and tva-ra^ "motion 
 from that which is near to the here,'''' might signify the third 
 position as well as ta, Hhere^ Indeed, this compound would 
 be more intelligible than ta-ra, for it bears outwardly the form 
 of a comparative extension of the numeral "two," and this is 
 the proper idea of "three." 
 
 159 It is a remarkable fact, that the first four numerals in 
 Greek and Sanscrit, and the first three in Latin, are declined, 
 while all the others remain without inflexion. There must be 
 some reason for this. Now we know that the oldest Greek 
 year was divided into three seasons of four months each : and 
 the subdivision of the fundamental number in the state-division 
 into the factors 3χ4, of which the four was the basis, needs 
 not to be insisted on. The first four numerals, therefore, would 
 be more frequently used as adjectives than any of the others, 
 
 U 
 
290 THE NUMEEALS. [BOOK II. 
 
 and for this reason would have inflexions, which the others, 
 whose use would be more adverbial, might want without so 
 much inconvenience. The same remark applies to the corre- 
 sponding fact with regard to the Roman numerals. The funda- 
 mental number of the Romans was three ; they had three tribes, 
 just as the lonians had four. Besides , the old Etruscan year, 
 which was the basis of their civil and religious arrangements, 
 consisted often months, not of twelve, and therefore the division 
 into tetrads would not hold with them. That this division into 
 tetrads was observed not only in the old Greek and Egyptian 
 year of twelve months, but also in the Greek and Sanscrit 
 system of numbers, is clear from the following facts. The 
 numbers two and eight in Sanscrit and Greek have the ordinary 
 dual-ending which is found in the dual number of nouns in 
 those languages ; they are written dv-ati, δν-ω, — asht-au^ όκτ-ω. 
 The meaning of this termination is clear in the former case: 
 can we then deny its force in the latter? But if the number 
 eight is really in the dual number, it can only be so as denoting 
 "twice four;" therefore the root of the number eight in these 
 languages must be the number four. This root in Sanscrit is 
 ash-t-. We have seen that the first part of the Sanscrit nume- 
 ral, four, is a mutilation of e-ha aspirated into -cL• Here the 
 whole word is shortened and assibilated into ash-. The second 
 part wants the letter r, which gives the third numeral its parti- 
 cular meaning , as distinguished from the second. That it is 
 wanting here is no argument against the identity of the latter 
 part of the root of the number eight with the number three. 
 In words of common use , when they exceed a certain length, 
 and especially in those which are compounds , the process of 
 shortening and softening always takes place, sometimes to an 
 extent which renders it difficult to discern the elements of which 
 they were originally made up. Even the accentuation of όκτω, 
 as compared with the other dual-form 8vω^ seems to indicate 
 a mutilation in the last syllable. The same remark applies to 
 mxa^ which is not dual in its inflexion. And there can, we 
 think, be little doubt that the Indo - Germanic forms of the 
 numerals "one," "two," "three," "four," "eight," must have 
 been, in their fullest development, eka, svau=dvauy sva'ra-s= 
 dva-ras = tva-ras , eka'tva7'a-s = ka-tvara-s, eka-tvarau =^ ash- 
 
CHAP. II.] THE NUMERALS. 291 
 
 tvarau. An additional reason for the hypothesis of a subdivision 
 of the duodecimal basis into tetrads is derived from the fact that 
 in Greek, in which this division seems to have been most called 
 for, the numbers eleven and twelve are single words ένδεκα and 
 δώδεκα, whereas the succeeding numbers up to twenty are made 
 up of separate words, connected by και: thus tqIq και δέκα, 
 τε6<JaQεg και δέκα, &c. The same appears still more clearly 
 from the Teutonic ain-lif, "e-leven," tva-lif^ "twe-lve," which 
 mean respectively "one" and "two left" or "over" (Bopp, 
 VergLGramm. 450; Pott, Zahlmethode, 172 sqq.). 
 
 160 There are only two other numerals which appear to 
 contain jthe roots of the primitive pronominal numbers; namely, 
 six and seven, which commence with the same letters respectively 
 in Greek and Latin. In these words, however, the process of 
 abbreviation and softening has been carried so far that we must 
 enter on a more minute examination of them. On comparing 
 ε-ξ, ε-7ίτά, with se-x, se-ptem, and the Sanscrit shash, saptam, it 
 appears exceedingly probably at first sight that the initials in 
 each had a common origin. Setting aside, then, this first syllable, 
 we have in all three languages the letters -pt- as the element 
 of the second part of the numeral seven, and these letters point 
 at once to the elements of the old τΐετορες, "four." The first 
 part, therefore, must be some mutilated form of the number 
 three, so that ε-τίτ-ά, &c. will be 3-\-4:=7. This also appears 
 from an examination of the corresponding syllable in the number 
 six. Bopp remarks ( Vergleichende Grammatik, p. 443) that as 
 the Zend word for six is written khshvas, and as sh is neither 
 an original letter nor the beginning of any other word in San- 
 scrit, we may infer that the Sanscrit word should be written 
 kshash. A comparison of the Greek and Latin would incline us 
 to believe that the more ancient form would be ksha~ksh, for 
 there is an evident reduplication. And similarly the Greek and 
 Latin words would be written εξ = ΰεξ = (κ6)ε-κ6-, and sex=^ 
 (Jc)se'ksy which are perfectly analogous to, and equally indica- 
 tive of a reduplication with the Sanscrit. If, therefore, there is 
 a reduplication it must be that the word is composed of two 
 co-ordinate parts , and as the word is a numeral, this must ex- 
 press that it is a number added to itself, and in the case of the 
 
 U2 
 
292 THE NUMERALS. [BOOK 11. 
 
 number six, this number must be "three." Accordingly, shash= 
 s-7cg=se-cs=S-{-S=6. But although it appears highly probable 
 that the numeral is composed of two words , each signifying 
 three, it would be desirable to know how the element ksh could 
 bear this meaning. And here again we find the clue in the Zend 
 ksvas^ which gives us not only a guttural before the sibilant, 
 but also a labial after it. That the full form of this numeral in- 
 volved a labial element is proved by the regular prefix of the 
 digamma in the Tables of Heraclea, where we have /a| , ^εξψ 
 κοντά, Ηκτος, &c. And this articulation appears in other Arian 
 languages. "It is curious," says Mr. Garnett*, "to observe how 
 the component elements of the word appear and disappear in the 
 cognate dialects. The Welsh chivech has preserved the guttural 
 and labial; the Afighan shpaj or spash the sibilant and labial; 
 the Albanian giast the mere guttural ; while the Armenian wetz 
 corresponds pretty closely with the digamma form of the Tables." 
 According to the law of divergent articulations (above, § 121) 
 the fullest form is the original matrix from which these modifi- 
 cations have ramified. And the analysis of this group of letters 
 is immediately suggested by palaeographical considerations. For 
 in the Behistun inscription the combination kh-sh is represented 
 by two distinct charactersf , and we have seen above (§110) that 
 'hva is a regular substitute for sva. In the particular case before 
 us, it seems clear that shva represents the original form of dva, 
 for, as we shall see directly, the Zend khsh occasionally corre- 
 sponds to cr in Latin , and while the Greek ρ regularly re- 
 presents a (>, the Latin r is regularly a substitute for d. Sup- 
 posing then that the initial guttural in the Zend numeral is a 
 residuum of eka (and the common Zend aevo does not prevent 
 us from supposing an original form which recurs even in the 
 modern Persian ek or yak) , we shall see that kh-8hva=kha-dva 
 =3 is perfectly analogous to qua-tuor=kha-dva'ra=z4:. And 
 khshvas is the mutilated reduplication of the former. 
 
 If we examine the compound articulation khsh^ as it appears 
 
 * "On certain initial letter-changes" (Essays, p. 244). 
 * See Rawlinson, As. Soc. x. 86, 157, who observes that the aspira- 
 tion in each character is developed by the mutual influence of the gut- 
 tural and sibilant. 
 
CHAP. II.] THE NUMERALS. 293 
 
 in the remains of old Persian, we shall see that it corresponds 
 (a) to .r, as in the Greek εξ, (δ) to s, as in the Latin sex, and 
 (c) to cr, as in the assumed residuum of the dental in the case 
 oi kh-shva=zkh'th-va=kh'rva=kh'dva (above, § 107). 
 
 (a) We find ξ for khsh in the Greek Ξέρξης for KhsJiayarsha, 
 "the venerable king*;" and ^Αρχαξερξ'ηζ for Artakhshatra , "the 
 honoured warrior." When Herodotus tells us (vi. 98) that the Greek 
 word which translates ^^ρξης is agrjCog, and that the words which 
 translate '^ρτο^-ξερίί;? are μέγας άρήϊος, he has, as RawHnson re- 
 marks (As. SoG. XL p. 120), followed the orthography of the one 
 word and the etymology of the other; for there can be no doubt that 
 khshatra signines "a warrior," and Rosen supposes (Journal of Educat. 
 IX. 336) that arta is the perf. pass, partiple of n, which, as appears 
 from a comparison of the Sanskrit sakrit, krinoti^ mrityUy with the 
 Zend hakeretj kerenoit^ merethyu, would be written in Zend ereta 
 (comp. Bahr ad I. Herod.) \ arta, therefore, means "honoured," and 
 Arta-khshatra ^ "the honoured warrior or king" (like maha-rajah in 
 Sanscrit), is rather the epithet than the name of a king, as indeed 
 appears from Ctesias, Pers. 49, 53, 57: βαβυλενει δε ^Αρόάτίτις δ 
 μετονεμαΰ^'εϊς ^Αρταξέρξης, and Curtius, νι. 6: Bessus veste regia 
 sumpta^ Artaxerxem appellari se jusserat (quoted by Pott, Etym. 
 Forsch. I. p. Ixv). In Ό-ξά%'ρΎΐς (Diod. xvn. 34; Plut. Artax. c. i.), 
 the Zend ksathra is exactly preserved. The ο is to be explained like 
 the first syllable of Otanes , which , according to Pott (Etym. Forsch. 
 I. p. XXXV ), is equivalent to the Sanscrit su-tanu, "having a beau- 
 tiful body," from the old Persian su, Zend hu, Sanscrit su= ευ, and 
 the Zend and Sanscrit tanu, "a body." 
 
 (b) Malcolm (Hist, of Persia, i. p. 271) translated Satrap "um- 
 brella-carrier." We think this far-fetched, and consider Σατρά-Λης ^^ 
 be the nearest approximation a Greek could make to what would be 
 in Persian ksMtra-bdn (Σατρα-πηνός', Plut. Lucull. xxxi. 4), or in 
 Sanscrit kshetra-pd, "ruler of the country," for, according to Xeno- 
 phon (Cyrop. vra. 6, § 3), the Satraps were persons οΐτινες αρτονόι 
 των ενοίκονντων (Pott, Etym. Forsch. i. Ixviii, and see below, § 213). 
 Here ksh is represented by s only; comp. sex with kshash. Similarly, 
 in the root kship, "to throw," the initial guttural is left out in the 
 Latin equivalent sip- (in-sip-ere=injicere, dis-sip-are=disjicere)^ but 
 in the Greek ρίτΐ-τειν, anciently ^ρίτΐ-τεον, as appears from ερείτί-ειν 
 (Pott, Etym. Forsch. i. 257; n. 167), we have ρ substituted for sh. 
 
 * This is Rawlinson's interpretation ; but another analysis of the word 
 has been suggested by Benfey and Oppert; see below, § 479. 
 
294 THE NUMERALS. [BOOK II. 
 
 (c) The following instances, in which the Sanscrit ksh is repre- 
 sented by cr in Greek or Latin, have been pointed out by Rosen 
 {Journal of Education, vm. p. 345, cf. Big-Vedce Spec. Annot. p. xi): 
 kshapd, "night," Zend ksafna^ ksafne^ ksapanem, Persian shab^ cor- 
 respond to the Latin crepusculum; kshura^ "the hoof of an animal," 
 to the Latin cms (crur-is)-, and kshpira, "swift," "quick," to the 
 Greek KQaiTtvog. 
 
 From all this we may safely infer that as the numerals 
 "four" and "eight" in the Indo-Germanic languages were fully 
 expressed by the combinations eka'tvaras=^l-\-3^ and eka-tvarau 
 =(1-|-3)χ2, so "six" and "seven" were signified by eka-tva- 
 eka-tva=:^3-\-d^ and eka-tva-eka-tvara^S-^-^. 
 
 161 It appears then, upon the whole, very probable that 
 seven of the first ten numerals may be traced to the three pri- 
 mitive pronominal elements. The numerals nye^ nine, and ten, 
 cannot be derived from the same source. 
 
 Although the duodecimal system of notation was forced 
 upon the notice of man by prominent and ever-recurring objects, 
 it must not be forgotten that there was still another mode of 
 counting no less obvious and necessary. We mean the decimal 
 notation suggested by the number of the fingers and toes. That 
 this system of notation should be mixed up with the duode- 
 cimal, in suggesting the names of the numerals, is natural enough ; 
 and we see such a mixture in the fact that the Romans had two 
 years, one of twelve months and the other of ten. One would 
 fancy, indeed, without any particular investigation into the sub- 
 ject, that the number five would have some connexion with the 
 word signifying "a hand," and the number ten with a word 
 denoting the right hand; for in counting with our fingers we 
 begin with the little finger of the left hand and so on till we get 
 to the little finger of the right hand. In Greek and Latin, 
 epecially, it is impossible to overlook the resemblance of δεκ-α, 
 dec-em to dU-6Log, dec-s-ter; and with regard to πέντε, quinque, 
 we have already seen (above, § 146) that the % of the former is 
 duly represented by the labial included in qv, and its Greek 
 representative P. The same interchange might be presumed 
 in the second syllables τε and que, for the identity of which 
 we have abundant examples , and this might seem to be sup- 
 
CHAP. II.] THE NUMEEALS. 295 
 
 ported by the dialectical form τίεμτΐΒ. A more accurate exa- 
 mination, however, ought to convince us that the nq in the La- 
 tin numeral is merely the representative of an euphonic nasal 
 which took the place of the original n, for the ordinal is quintus 
 not quinctus, and the derived proper names (according to the 
 true orthography ; see Forcell. s. v. Quintius) are Quintius (Samn. 
 Pontius'), Quintilius^ Quintilianus, &c. The change of i/r into 
 μπ in πψΛΒ, τΐεμτΐάς, Λεμτίάξω, ΛεμΛ-τός, &c. must therefore 
 be regarded as euphonical and arbitrary; and the original form 
 of the fifth numeral in Greek and Latin must have been πεντε= 
 ^εντε and quinte; and this view is confirmed by the Greek 
 δέκα : for we have shown above that κα may represent either 
 κεν or κεντ (§ 114), and we have just seen that άέ=ά/ε. If, 
 therefore, κ in δέκα stands for an original κότΐτία (above, § 110), 
 the compound must denote "twice-five;" and as we shall show 
 hereafter that κa=J^εvτ is the root which expresses "a hand," 
 it will appear that δΓε-^εντ originally meant "two hands," i. e, 
 the ten fingers held out together. 
 
 With regard to the ninth numeral, in Greek at all events, it 
 is difficult to resist the first impression that εννέα ^ which must 
 have been originally εννέ^α^ owes its origin to the Greek mode 
 of speaking of the end of a month, as ενη καΐ νέα, i. e. "the old 
 and the new." That the word ενός (also written 'εννος and ενός) 
 means "old" is proved by the gloss ^έννος, άρχαΐθ£ (Ahrens, 
 Dial. Dor. p. 54), and may be inferred from Aristoph. Nub. 
 1183, where the words γρανς τε και νέα γυνή are cited to justify 
 the expression ενη καΐ νέα. Kuhn (^Zeifschr. f. Vergl. Spr. u. 
 p. 129) compares the word with senea;^ Goth, sineigs, Sanscrit 
 Sana, and he quotes {Ibid. iv. p. 44) passages from the Vedas in 
 which Sana is opposed to apara, navy a, nutana, just as ^og is 
 opposed to νέος. The idea attached to ενός is that of "just 
 passed, completed," as in the phrase εναυ άρχαί, αί Λαρωχμέναι 
 (Harpocrat.); and it seems to us, that kviavrog means simply 
 ένί-Γέτος, "the past or completed year." In this sense έννέ^α 
 =έννέ-Γεντ will mean " the last of the completed hand," before 
 the two hands were held up together. At any rate Plato seems 
 to have recognised the possibility of ενοξ τε καΐ νέος being 
 respresented by έν-νεο; for it is idle to correct his intentionally 
 ludicrous compound ΰελα-εν-νεο-άεια^ in the G^atylus, 409 b. 
 
296 THE NUMEEALS. [bOOK II. 
 
 where the philosopher explains it as meaning on βελας νέον τε 
 καΐ ενόν έχει αεί. And it may reasonably be inferred that the 
 orthography εν-νοζ, which some critics reject, has arisen from 
 the constant combination of ενός and νέος to signify the last day 
 of the month. This must at all events be the meaning in the 
 line of Hesiod, Op. et D, 408: μηδ^ άναβάλλεΰ^'αί εg τ avqiov 
 ες τ εννηφί, where it denotes the last of an assumed period, i. e, 
 three days. This explanation of the ninth numeral must of 
 course be limited to the Greek language. But the Latin, San- 
 scrit, and Teutonic novem, navan^ niun, admit of an explanation 
 which involves the same idea, though it implies a slightly dif- 
 ferent origin. For we agree with Bopp and Benary in referring 
 those names to the adjectives noviis, nava, niiijis, i. e. "new," 
 and the interpretation of their use is simply this — that "nine" 
 can only be contemplated with reference to preceding numbers, 
 and as something later, subsequent, and new. In the Lithua- 
 nian and Sclavonian languages no doubt this numeral has im- 
 mediate reference to the succeeding "ten;" thus de-wyni means 
 "therefrom one" (like the Latin do-drans^^de-quadrans); and 
 Pott proposes (Zahlmethode , p. 142) to consider the Sanscrit 
 navan as a compound of na, "not," and ναη=ύηα, "diminished," 
 which seems self-contradictory. 
 
 On these and kindred subjects Lepsius has collected a great 
 deal of valuable information combined with much ingenuity and 
 acuteness: and though we disagree with him on many points 
 we think it right to give our readers an opportunity of judging 
 for themselves, and therefore subjoin an extract from the essay 
 to which we have referred above*. 
 
 162 "It is not difficult to perceive," says our authorf, "that the 
 same stem [i.e. that of five] recurs in the number 10 of the Indo-Ger- 
 manic languages ; it is preserved most entire in the Latin de-cem. The 
 final m, which has fallen off in the nominative of the Sanscrit and Zend 
 da -ςα^ shows itself still in the declension, and therefore may be sup- 
 posed in the original form of the Greek δεχα. In the Gothic taihun, the 
 k is changed into Λ according to the usual law: in this it differs from 
 
 * Dr. Richard Lepsius, Zwei vergleichende Abhandlungen. Berlin, 183G. 
 t p. 116. 
 
CHAP. II.] THE NUMEEALS. 297 
 
 the form admitted in fimf^ without, however, justifying any doubt as 
 to the identity of the two forms. Moreover, we find the Gothic form 
 hun^ with an addition of d (see Grimm, ii. pp. 231, 232), in the com- 
 pounds sibun-tehund, 70; ahtau-tehund, 80; niun-tehund, 90; in which 
 we find tehund as an equivalent to taihun. Indeed both forms are 
 combined in taihun-tehund=^10y:^\0=100^ and it is not till the com- 
 binations which follow, tva-hunda, 200; thrija-hunda, 300; &c., that 
 the simple form hunda appears, in which of course we must recognise 
 the same stem as in tai-hun and te-hund. It is certain, from a mere 
 comparison, that hunda is again found in centum , tva-hunda in du- 
 centi, &c. The radical wi or η is thrown out, as is frequently the 
 case before i, in the Sanscrit p«iiz, for which eka-gata is also used 
 (compare ε-κατόν). As hunda reappears in the tens , so also we have 
 centum in (d)vi-ginti, tri-ginta, &c.; and although the η has fallen out 
 in the Greek εκατόν, it is preserved in τρίά-κοντα, τεΰΰαρά-κοντα, 
 it has fallen out only in (df)8ίκaτi', the ordinary Attic form εϊκοόι 
 has gone still farther, and has softened t into s ; so also in δοα-κόΰ-οοί, 
 τρία-κόΰ-ίΟϋ (comp. the Boeot. δία-κάτ-ιοί, &c.), and in the Latin 
 ordinals vi-ces-imus, tri-ces-imus. In these, therefore, the same stem 
 appears as κοξ and ces. In Sanscrit the η is quite dropt in the tens 
 also: vin-gatij 20; the three following have lost the i also; trin-gat, 30; 
 chatvdrin-gat, 40; panchd-gat, 50; in the following gati loses its first 
 syllable, and ti, originally nothing but an affix, alone remains : shash-ti^ 
 60; sapta-ti^ 70; agi-ti, 80; nava-tij 90. It is precisely the same in 
 Zend, except that 30, 40, 50 are formed with gati, instead of gat; 
 those which follow however also take -ti. With regard to the Gothic 
 we have only the additional remark te make, that we find in the first 
 four tens tvai-tigus, thrija-tigus, fidvor-tigus, fimf-tigus, a third form 
 gus of the same stem: this comes very near the Greek κο^, and has 
 besides retained the softened guttural instead of the h. Thus we 
 find in the Gothic the remarkable phenomenon of one and the same 
 stem, which is written /m in 5: hun in 10 and the higher tens, and 
 ^w(n) in the lower tens: and it is the business of definitive, and, at 
 the same time, extensive comparisons, like those which are possible in 
 the numerals , to establish such facts as must necessarily be exposed 
 to objections, when the investigation is confined within the limits of 
 a single language. 
 
 "How then are we to interpret this widely-diffused stem, which 
 we see recurring in the five, the tens, and hundreds of all Indo-Ger- 
 manic languages ? We observe that this stem contains precisely the 
 most essential numbers of the decimal system. In general, how have 
 mankind arrived at the decimal system, which is so inconvenient for 
 all minute reckoning, and especially for division? and yet the earlier 
 
298 THE NUMEEALS. [bOOK II. 
 
 the period, the less was the occasion for large numbers, iu which the 
 fundamental system becomes less important. Finally, why did they 
 not go back to the number 5, the lowest basis of the decimal system? 
 We find both systems together among the aborigines of America , as 
 well as among the most polished nations of all ages. Whence came 
 this decimal system which has everywhere taken its place by the side 
 of the far more natural duodecimal system? From what else but 
 from the 10 fingers of the two hands ^ on which every child at the 
 present day begins to count? 
 
 "In this simple consideration we must be struck with the sur- 
 prising resemblance between hunda and handus, the hand, in Gothic: 
 in fact, a narrow scrutiny of both stems , which we will now attempt, 
 will easily convince us that this similarity is not merely external and 
 accidental, but that the two words are etymologically one and the 
 same. 
 
 ^^ Handus is immediately connected with hinthan, capere^^ which 
 we also find in the isolated, and, I might almost say, Germanized 
 form pre-hendo. Grimm {Gr. ii. p. 35) is quite right in also referring 
 to this stem hund-s, canis^ the catcher, qui capit feras. In this too 
 we see that in the whole stem d is really nothing but an affix, of 
 which Grimm (ii. pp. 231 foil.) has very fully treated: for hund-s, 
 with the usual changes, but without d, is found in the Greek κνν-ός^ 
 Latin can-is^ Sanscrit gvan (gen. abl. gunas, dat. gun-e^ instrum. gun-a, 
 locat. gun% nom. pi. gvd, accus. gvan-am). We find the same stem in 
 the Homeric form yiv-to ά' ίμάύ^λην , which points to an ancient 
 form ysv-SLV, instead of ελεΐνγ. As hund-s, ca7iis, refers to the stem 
 /mw, Greek yew, Latin can, similarly we may trace also hunda, centum, 
 to the stem hun, Latin cen, Greek καν. Consequently, both stems 
 
 [* Hente, which so often occurs in Chaucer, is the same word.] 
 f "Buttmann is unquestionably right in comparing γέν-το immediately 
 with ελετο, just as the iEolic form κε'ντο for γ,έλετο is adduced from Ale- 
 man. It is this transition from η to ί which prevents us from recognis- 
 ing the stem hinthan in the Greek language. We find the same stem with 
 r for I in the Sanscrit hri (capere), to which belong hasta {manus) (see 
 Burnouf, Yagna, Tom. i. p. Ixxxi, and note H), Latin hir, Greek χειρ 
 and αίρέείν, also, with an addition of j), κάρ πος (the wrist), carpus, άρ- 
 πάζειν, Gothic hreiban, greifen (see Grimm, ii. p. 45); nay, as it appears, 
 also in the Sanskrit kara (ma?ius), and consequently the whole wide-ex- 
 tended stem kri, the general signification of which (facere) cannot be the 
 original one. The stem hri, as we must infer from the letter h, which 
 is always a later one, cannot represent an original form, but we must al- 
 ways seek for this in kri, which therefore corresponds to kri, facere. 
 'That which is identical as far as the letters are concerned cannot be 
 diverse as concerns the meaning' (see Grimm, ii. pp. 76 foil.). The stem 
 of manus is different, though it is probably connected with the Sanscrit 
 pani, the hand, and with the Greek μην-νω, mon-stro.'' 
 
CHAP. Π.] THE NUMEEALS. 299 
 
 are identical even in this form. An u in a stem, as in hund^ κννόζ, 
 often points to an original v, which in this case brings us nearer to 
 the stem kvan. In fact we still find both consonants in the Sanscrit 
 ςυαη, dog, Zend gpari^ ; we must therefore in the case of canis also 
 suppose an older form qvanis. Just so the stem of the number 5 
 was originally kvam; only in this word the m is still preserved, as 
 might be established by certain forms (τΐέμτΐε, fiwf), though in most 
 words expressing this numeral the m has been softened into n. This 
 m is probably a softening of p., which we still find in cap-ere, the 
 connexion of which with can-is is as certain as that of hinthan with 
 hunds. Lastly, we find the same stem in the Hebrew qomez, 'the 
 full hand,' qdmaz , 'to take,' kaf, 'the hand,' and in the Coptic gop 
 (capere)^ whence comes giig\ ' the hand.' 
 
 "So much for the stem from which hunda and handus are de- 
 rived. It appears to me fully established , that , in all the languages 
 referred to, the number 5 was expressed by the hand with its 5 
 fingers, and was thus made the simplest and most obvious basis of 
 the old system of numeration. The Greek word πεμΛάξειν means 
 what we call 'counting on the fingers.' It was possible, however, 
 when the one hand was finished to go on with the other, and thus a 
 higher unity was naturally made of the number 10; so that instead 
 of the quinary scale, which is still in use among some nations, they 
 formed a scale of which the radix was 10. There are people in 
 America who count with their feet, and have thus arrived at the still 
 higher radix 20t. 
 
 "It is clear that, if the principle of continued composition of the 
 same elements was applied to the higher numbers, it would soon lead 
 to forms of intolerable length. Even the simple numbers up to 10 
 are abridged and mutilated in a most violent manner. "We should 
 expect to find the same in the higher numbers formed on the digit 
 system, though in Gothic we may still point out this system in almost 
 its original completeness , for in this language the outward similarity 
 of hunda and handus seems to have preserved the feeling of their 
 identity for the longest period, while in other languages this feeling 
 was sooner lost because they had no form for handus so similar to 
 that for 100. 
 
 "In the number 5 we find not only no mutilation, but even a 
 reduplication of the stem J. The Latin is the only language in which 
 
 * "SeeBurnouf, Χαςηα, i. p. Ixxii, and corap. Herodotus, i. 110: την 
 yccQ wvcc απάκα -καλέοναι Mrjdot.'' 
 
 f Voyage de Humboldt et Β onpland. lere Partie, a Paris, 1810, p. 193. 
 [+ βορρ and Benary think that the last syllable of the numeral five in 
 
300 THE NUMEEALS. [bOOK II. 
 
 we find traces of the simple stem, namely in quim-atus, quin-i, quin-io, 
 quin-arius, quin-decim, quin-genti, &c. : perhaps also in the old Norse 
 fimm, Danish and Swedish /ew (see Grimm, i. p. 762), unless these 
 forms have arisen from a mutilation which seems to be indicated by 
 the double m in the old Norse. 
 
 "In taihun, 10, we easily recognise tvai with an omission of the 
 v: 'two hands.' Just so in da -ςαη, decern, δέ-κα. 
 
 "In tvai-ti-gus, 20, 'twice two hands,' the first tvdi is still entire; 
 ti is a further mutilation of the tai in tai-hun. We find that in the 
 other languages even this ti has fallen out. Instead of (d)vi-gati, we 
 ought to have (d)vi-da-gati from da-ga; instead of dvi-ginti, dvi-de- 
 ginti; instead of εΐ-καη, εΐ-δεκατι. 
 
 "The same relation remains in thrija-ti-gus, '3 times 2 hands,' 
 fid-vor-ti-gus, ' 4 times 2 hands,' sibun-te-hund, ' 7 times 2 hands,' &c. 
 
 "In Gothic the number 100 is written at full length taihun-te- 
 hund, '2 hands X 2 hands.' But this exactness does not extent far- 
 ther in Gothic ; instead of the difficult composition taihun tehund, the 
 following hundreds return to the simple stem, and we have tva-hunda, 
 200, instead of tvai-ti-gus tehund. In the other languages, as also in 
 the later dialects of the Germanic language, the simple stem is put for 
 100, and only distinguished by the ending, so that εκατόν properly 
 signifies 'one hand,' and as far as the letters are concerned, du-centi 
 and dvi-ginti are perfectly identical, and denote 2 hands, just as 
 tai-hun does. 
 
 Sanscrit, Latin, and Greek', is the copulative conjunction, and that the 
 nasal, which, in Sanscrit and Zend, appears at the end of this numeral, 
 is a later excrescence. Bopp (Vergl. Gramm. p. 443) considers pan-cha 
 as signifying "and one," the lirst syllable being the neuter form of pa 
 which appears as the number "one." Benary remarks {Berl. Jahrh. July 
 1833, p. 48), that pan-cha is easily explicable as a mutilation of pdni-cha^ 
 "and the hand," because with this number they began to count with the 
 other hand; and he thinks this derivation confirmed by a comparison of 
 pan-cha^ quinque^ and ηέν-τ8, the last syllable in each being the regular 
 conjunction in each language. This last suggestion is not to be despised; 
 but if the termination of these words is the conjunction, it implies simply 
 that after counting four the whole hand was opened and held up.] 
 
 * "Grimm {Gr. ii. p. 17) is perfectly right in connecting the following 
 words : Gothic teihan {nuntiare^ dicere), old High German zihan {accusare), 
 zeigon (indicare), zeha (digitus, i.e. index)', Gothic taihun, old High German 
 zehan (decern), Gothic tigus (decas, numerus index), &c. A confirmation of 
 this will appear in the following development. 
 
 "In counting with the fingers one naturally begins with the left hand 
 and so goes on to the right. This may explain why in different languages 
 the words for the left refer to the root of five, those for the right to the 
 root of the ten, and why expressions like finger, fangen, zeigen, zdhlen, 
 refer sometimes to 5, and at other times to 10. To omit any strict de- 
 velopment of the ideas, — that there is a connexion between 10 (the second 
 
CHAP. Π.] THE NUMERALS. 
 
 "Lastly, in Gothic the word tJiusundi, 1000, seims td^^^e^T^aflMsei 
 stem, and appears to be composed, how we know not, oJ 
 hundi^. 
 
 "I subjoin here the explanation of the Indo-Germanic expression 
 for 9, which is also, I think, though not so distinctly, derivable from 
 the stem kvam. It has here, as in τΐέμ-πε, fimf, old High German vinf, 
 thrown off the ^, and appears as vam. "We start here from the Greek 
 εννέα, which stands for Ιν-^εναμ, as we see from novem, navan, niun. 
 The Greek form is distinguished by the prefix εΐ'-, which is wanting in 
 the other languages. In this we are immediately reminded of ε-κατόν, 
 which appears more entire in the Sanscrit eka-gata, 'one hundred.' 
 There is a form ekona or -una peculiar to the Sanscrit ( originally it 
 was eka vina, 'one without,' 'one less') which subtracts one from the 
 
 hand) and the right hand, appears from the words: Sanscrit c?apa — dak-sha, 
 dak-shina; δέκα — δέ-κ-βιος; decern — dec-ster; Gothic taihun — taih-s-vo (on the 
 derivation in νΰ, see Grimm, ii. p. 189), old High German zehan — zeso, 
 ze-se-wa, old High German zeswe (dexter). All the languages have also 
 formed from this a distinct feminine substantive, to signify the right 
 hand. This transition to the idea of the right hand will enable us to 
 understand how the ideas of pointing, taking, directing, could proceed 
 from the same root: Sanscrit dig (mon-strare), δείκ-νυμι, δέκ-ομαι, dic-ere, 
 in-dic-are, in-dec-s, dig-niis, &c., Gothic teihan (accusare), zeig -ύη (monstrare). 
 Let it be observed here, how these verbal roots preserve, by abbreviation, 
 an appearance -of simplicity and originality which by no means belongs to 
 them: this is a phenomenon of frequent occurrence, which has been hitherto 
 but little attended to. Language, like the Indian fig-tree, lets its branches 
 sink into the ground again, all round its root, and these strike root afresh 
 and become new stems, like the old one, whose relative originality can 
 only be estimated according to the degree of their removal from the com- 
 mon middle-point. From δέκα is farther derived δάν,-τ-νλος, from decern, 
 dig-it-US, and from zehan, zeha (the toe). Lastly, we refer to taihun, the 
 old High German zehan, also old Norsa ta-la (instead of tahi-la), old High 
 German za-la, zahl (number), ζαΐυη, zahlen (to count), just as πεμπάζειν 
 comes, though with still greater clearness, from τίέμπε (5), and just as 
 the Sanscrit gatai (numerare) is derived from gata (100)." 
 
 [We do not translate the remainder of the note, which seems to in- 
 volve some precarious comparisons. For instance, we cannot agree with 
 Lepsius that sinister and aqiGTBqos ought to be connected with the old 
 High German vinstar. On the contrary, it seems infinitely more natural 
 to conclude that, as in the phrase ^π άοπίδα, "to the left," so in aQie- 
 τϊρόί we allude to the weapon of defence carried on the left arm, so 
 that the root will be that of /^άρης, βαρετή, ^/^ηρως, &c. , Germ, war, 
 wehren, &c. (Graif, i. p. 906), Sanscrit vri, Welsh gwared (Garnett, Essays, 
 p. 233), &c. Similarly sinis-ter refers to the sinus togce, which was on 
 the left hand (Pott, Zahlmethode, p. 139)], 
 
 * "Just as there is a break after taihun tehund and a return to the 
 simple AMWc?a, the Romans, when they got beyond 100,000 in their money- 
 reckonings, left out this sum and said only decies ceris, instead of decies 
 centena millia ceris, and sestertium was in the reckoning equivalent to 1000 
 sestertii, when it was joined to decern, undecim, &c., and to 100,000 sestertii, 
 when connected with decies, undecies, &c." 
 
302 THE NUMEEALS. [bOOK II. 
 
 number which follows : ekona mn^ati ov una viiigati , 19. Similarly, 
 there might have been an ekdna dagan or una dacan, for 9; the da 
 fell out, as in vingati for vin-da-gati, and there remained ekona-kan 
 or ekonavan, which corresponds to the Greek Ivvifav or unavan, 
 which by dropping the u becomes navan, 7iovem and niun.''^ 
 
 This disquisition anticipates all that remained to be said on 
 the tens and hundreds. 
 
 163 The words χίλίοι, 1000, and μνρυοί, 10,000, are merely 
 expressions for large but indefinite numbers, like the Latin 
 mile, i.e. 'ΐη-ίΙβ=ομ-ίλία, "a great crowd;" whence miles (mil-it)^ 
 "one who goes in or belongs to a large body" (cf. equ-es,ped-es, 
 cocl-es, &c., and see Varron. p. 24). The connexion oi χίλιοι 
 with ;^tA6^, " a heap of fodder," is self-evident: and it is equally 
 clear that χιλόζ is connected with χίω (%έ^ω) , just as χύλοξ is 
 with χάω (χά^ω), and καυλός with καίω, κ«/ω, κάνω. That it 
 has nothing to do with the words χλοή, χλωρός, &c., as Pott 
 supposes (Etym. Forsch. i. p. 141), is shown by the length of 
 the first syllable. The intimate relationship in meaning which 
 subsists between χίω and χιλός will be felt by any one who 
 reads such passages as Odyss. Xl. 588: δενδρεα δ' ν-φιτίέτηλα 
 κατάκρη^εν χεε καρπόν. The same is the case with μύριοι, 
 which, with a difi'erence of accentuation, is used in the best 
 writers in a general and indefinite sense. Thus we have μάλα 
 μύριοι, "a great many," μνρία βτίονδη, "excessive eagerness" 
 (see Buttm. AusfilhrL Sprl. § 70, Anm. 15). This word is 
 connected with μνρω^ a verb which expresses the falling of 
 water, and is especially applied to a flood of tears. Compare 
 Hesiod, άΰπ/Ηρακλ. 132: τνρόύ&εν μεν θάνατον τ είχον και δά- 
 κρνϋι μνρον,ν^ίύϊ Soph. (Ed. Col. 1253: δι όμματος άΰτακτΐ λεί- 
 βων δάκρνον. The derivation of the idea of a large number 
 from the sight of water falling in infinite drops is too obvious 
 to require any remark. 
 
 164 We must now turn to the ordinals, and, in discussing 
 them, we will include an inquiry into the modes of expressing 
 a superlative common to the Greek and cognate languages; an 
 inquiry which might indeed be postponed to the third book, but 
 which may be conveniently discussed in this chapter, as the com- 
 
CHAP. II.] THE NUMERALS. 303 
 
 paratives have also been touched on here, and as this will give 
 us an opportunity of explaining three words intimately con- 
 nected with the numerals — μεΰος, ημυΰν, and άλλος. 
 
 It has been mentioned that the ordinal of the second number is 
 dsv-TEQog, and it will be observed that this word contains the compa- 
 rative suffix -TEQog explained above. This comparative suffix is , as 
 we have seen, from its origin peculiarly adapted to the expression of 
 a relation between two persons or things, especially of the relation 
 between farther and nearer. Hence , the ordinal of the number two 
 would naturally be expressed by affixing to that numeral this com- 
 parative termination, for in that case a relation between two only is 
 implied. But when the relation of nearness is applied to one out of 
 a greater number, we find that a difiOrent termination is affixed, and 
 demarog is the word used when we are speaking of the nearest out 
 of a given series, that is, "the last," considering them as in a state of 
 motion from the terminus in quo. Now the ending -rccTog is the most 
 common of those which are used to express the superlative degree in 
 Greek. That this form, however, is an arbitrary extension appears 
 from the following considerations: there is obviously an appended 
 Tog in the epic forms εβδόμα-τος, oydo-arog compared with the com- 
 mon ordinals έβδομος, όγδοος', there is clearly a reduplication in τρί- 
 τατος by the side of τρί-τος ; the form -τατος is never found in the 
 cognate languages ; it does not appear in those Greek superlatives, in 
 which there is still a trace of the included adverb (for even τΐρώτος 
 makes τΐρώτιΰ-τος and not τίρώ-τατος; see below, § 167); and, with 
 two casual exceptions ( δεντατος and τρίτατος) it does not serve as 
 the ending of the Greek ordinals. For the ordinals in the common 
 Greek writers are τΐρώτος (or πρότερος when only two are spoken 
 of), δεύτερος (or δεντατος when more than two are spoken of) , τρί- 
 τος (ravelj τρίτατος), τέταρτος, τίέμτίτος, έκτος, έβδομος, όγδοος 
 (or δγδο^ος, as the analogy of the Latin octavus would lead us to 
 infer), gVi/cirog, δέκατος, &c., είκοβτός, &c., εκατοΰτός, δίακοΰίοΰτός, 
 &c., χίλίοβτός, μνρωΰτός. In all these, except εβδο-μος, 6γδο-ος= 
 ογδο^ος (which, like octavus, exhibits υ for m; cf. δράω, δρατίετης, 
 δρόμος'"^), the termination is τος. The same termination is found 
 in the Sanscrit chatur-thas, shash-thas ^ and in the Latin quartus, 
 
 * On the change of m into ν see Kuhn, Zeitschr. f. Vergl. Sprachf. 
 VII. I, p. 80. In Sclav, we have osmyi and in Lith. aszmas for octavus, 
 where the m is retained. Grimm {Deutsch. Gramm. in. 640) considers that 
 the full form tarn, tim, or torn is really retained in sap-tamas, sep-timus, 
 εβδομος= ere- τόμος, and in ash-tamas, ογδοο£=ογδομος~6'κ-τομος. 
 
304 THE NUMERALS. [bOOK II. 
 
 quintus, sextus ; all the other Latin ordinals, except s'ecundus (which 
 is merely the participle of sequor), octavus (for octimus), and nonus 
 (for novimus), end in -mus, an equivalent to which is found in 
 the Sanscrit pancha-mas^ sapta-mas^ ashta-mas, nava-mas, daga-mas. 
 As the endings μος and rog can have no connexion with one an- 
 other, we must conclude, either that the meaning of the superla- 
 tive and ordinal might be equally expressed by the simple elements 
 μος and rog, — that is, they both imply that the thing specified is the 
 last of a series ending with the speaker or the object specified, — or 
 they must be fragments of a compound affix capable of expressing 
 that relation. We cannot recognise the necessary meaning as com- 
 mon to two elements so distinct as μog and tog, and must therefore 
 fall back on the other supposition; and comparative philology fur- 
 nishes us with a form containing both of them. The common San- 
 scrit terminations for the comparative and superlative are -tara, -tama 
 (Latin -timus) , but the Sanscrit and Latin forms of some pronominal 
 developments show that -μog and -rog have the same value as -ta-mas 
 or -ti-mus. Thus we have pra-thama by the side of pri-mus , and 
 πQώ-τog or 7tQGirLu-rog\ eka-taras and eka-tamas by the side of £κα- 
 rsQog and axoLo-Tog ; ka-taras and ka-tamas by the side of πό-τερος 
 and 7t66-rog\ ya-taras and ya-tamas by the side of ojto-TSQog and 
 OTtoo-rog The Latin superlative ending is -timus, as in op-timus, 
 "uppermost" (from οδ), in-timus, "innermost" (fromm), and this ter- 
 mination is universally assimilated in the superlatives of ordinary ad- 
 jectives, as in duris-simus for dured-timus , cf. ses-sum for sed-tum 
 Varron. p. 329). But the Latin language has not only ordinals in -mus 
 like primus, and in -tus like quartus, but it has superlatives also in 
 -mus like extre-mus, postre-mus , infi-mus or i-mus and sum-mus for 
 supi-mus. The simplest explanation of these interchanges is the ap- 
 plication to the consecutive syllables ta-ma, of the principle, which 
 explains the divergency of articulation from a combined sound in a 
 single syllable (above, § 121). Now -ta-mas would imply "motion 
 from there to here," so as virtually to coincide with the second po- 
 sition , as appears from the force of such a word as fini-timus (above, 
 § 130). And while the full combination ia-wa seems necessary to 
 give the requisite signification of the ordinal and superlative, namely, 
 the last term in a series ending with the specified person or object, 
 there are several indications of the manner in which the two syllables 
 might have collapsed into one. The loss of the m and the represen- 
 tation of the whole syllable by ta-s, -rog, -tus, is explained by a com- 
 parison of ashtamas with octavus and oydoog. And the substitution 
 of ma-s, -μog, -mus for the full termination ta-ma is supported by the 
 analogy of comparative forms, in which we have ra alone for -tara. 
 
CHAP. II.] THE NUMEEALS. 305 
 
 Thus we have adha-ra ''lower," as well as adha-ma "lowest*;" ρα is 
 used as well as a-Qa\ and the adjectives in -r<z, -ρο^ have undoubtedly 
 a comparative or relative value (see below, §§ 167, 204). If this con- 
 clusion is valid, μος and -rog are only portions of the full termination 
 ta-ma^ and -τατοξ is merely a reduplication of the latter. The force of 
 the termination -το^, as it appears in the ordinals, is easily shown by 
 a few examples. We have Hom. Od. xxiv. 289 : τΐοβτον δη erog εύτίν 
 by the side of (v. 308) : τόδε δη πέμπτον ετοζ εΰτίν. In Aristotle, Pol. 
 ir. 3, § 5, we have εχαύτος εμοξ λέγει τον εν πράττοντα των πολιτών 
 η κακώς, οπόβτος τυγχάνει τον άρι^'μον ών, "as each one out of the 
 number is prosperous or the reverse, each one of the citizens will 
 claim or repudiate him as his son, whatever may be the number of 
 which he forms one." And generally we have the word πολλοστός, 
 which means "one taken out of many," and hence, by a very natural 
 transition, "very small;" for, when a given whole is divided into 
 many units, the smallness of the units will of course depend on the 
 number of them. The word also signifies "the last of a long series," 
 as in Aristoph. Pax, 559: 
 
 άβπάΰαβΟ'αι Ο'νμος ημΐν εΰτι πολλοβτώ χρόνω, 
 i. e. "after so long a time," "in the last of so long a series of years;" 
 the converse of which is ολιγοότός χρόνος (Soph. Antig, 619). These 
 meanings arise naturally from the signification which we have given 
 to the affix, and we do not know of any other means of explaining 
 the word. We have also a very striking proof of the correctness of 
 this view in the ordinary use of the superlative in Greek, where we 
 should expect a comparative according to our idiom. Thus we have 
 in ^schylus (Per see, 180): έδοξάτην μοι δυο γνναϊκ ενείμονε — εΐζ 
 bi^LV μολείν — μεγέ&ει τών νυν ενπρεπεΰτάτα πόλν, where these two 
 visionary women are supposed to be the first of a series including all 
 the actually existing women, with whom they certainly would not be 
 classed were there not something in the nature of a Greek superlative 
 which renders a construction like this necessary. The same also appears 
 from the common Greek idiom πέμπτος αυτός, "with four others," &c. 
 
 165 Although the Greeks generally expressed the comparative 
 by -τε-ρος, there was another method which they adopted most 
 frequently in the case of dissyllabic adjectives terminating in -ρος or 
 -νς. This was by affixing the termination -ιων. There was, however, 
 a great difference in the etymological structure of these and the other 
 
 * Bopp divides these words a-dhara and a-dhama (Demonstrativsiamme^ 
 p. 9) ; but the formation indicated in the text is supported by Pott iEtym. 
 Forsch. I. pp. 61, 254, 2nd ed.). 
 
 X 
 
306 THE NUMERALS. [bOOK II. 
 
 comparatives. For while the termination -τερος is appended to a 
 fixed case or adverbial inflexion of the positive, the suffix -ιων= Lov-g 
 is added, like other formative adjuncts, to the crude or uninflected 
 form of the noun. This fact , which we first pointed out in another 
 place (Gr. Gr. Art. 269 sqq., cf. Varron. p. 328), explains all the pecu- 
 liarities in the form of the syllable immediately preceding -τεροξ , in 
 which the traces of the original adverb are more or less distinct ac- 
 cording to the influence of euphonical and other like considerations. 
 The quantity of -υων is variable , the first syllable being short in the 
 old epic poets, and long afterwards; in this variableness it stands 
 between the analogous Sanscrit terminations -lyas, -iydns^ and the 
 Latin -tor, one of which has the first syllable always long, and the 
 other always short. Some of the comparatives thus formed admit of 
 an anomalous contraction, which requires some notice. Thus ταχνξ 
 makes comp. ταχίων, superl. τάχιστος, but ταχίων is often contracted 
 into Ο-άβόων, neut. d-a(56ov. Similarly βα^'ύς makes βάόΰων ; βραδύς^ 
 βράόύων; γλνκνζ,γλνόόων ; μακρός, μάΰβων; τταχνς, τίάΰΰων. Το 
 these may be added άόΰον from αγχί, and μάλλον from μάλα. In all 
 these cases there has been a process of assimilation, like that which 
 has taken place in the barytone verbs in -ΰόω *. In the case of Ο'άύ- 
 6ων, γλνύΰων μάύύων and ηάΰύων, there has been an assibilation of 
 the κ sound, the aspirate in the first word being transferred from the 
 end to the beginning of the syllable , according to a principle before 
 explained. The second letter in all these words is a representative of 
 the L, as is also the case in άλλος, alius, μί(5<5ος, medius, φνλλον,/ο- 
 Hum. From the accentuation of άΰΰον, ^άΰΰον, and μάλλον, it may 
 be inferred that in Attic at least the first vowel is long by nature, 
 and it may be doubted whether these words ought not to have the 
 iota subscriptum, the i being transferred to the first from the second 
 syllable, as is indeed clearly the case in κρείύΰων, μείζων. 
 
 To ascertain the meaning of this comparative suffix, we must recur 
 to what was said in the last chapter on the connexion of the Greek and 
 Sanscrit terminations — ίος, tya. It appears that these endings express 
 a quality or relation, and so differ from the comparative-ending only 
 in this, that the comparative properly denotes a relation between two 
 only, whereas these terminations express a relation generally. There 
 is reason, therefore, to believe that these terminations are only weaker 
 forms of the comparative ending in -ιων, just as the terminations of 
 med-ius, al-ius, clearly relative words, are only slight variations of the 
 comparative-endings in -ior (more anciently -ios), neuter -ius. In fact, 
 as we shall show in a subsequent chapter, the original form of the 
 
 See Book in. chap, i, and Book iv. chap. 4. 
 
CHAP, il] the numebals. 307 
 
 genith'^e was -lov = -6loVj which was more usually apocopized into -to, 
 and as the qualitative adjectives in -cog were derived from the latter, 
 so the comparatives in -ίων = -Lov-g were inflexions of the former. The 
 meaning of the ending -ίων is quite in accordance with this derivation. 
 It does not imply excess, like the suffix -ZEQog, but only a considerable 
 amount of the quality indicated by the adjective — rather more than 
 less — and this is often the force of the adjectives in -uog and of the 
 Latin comparatives in -ior, which may be added to formations in -ter; 
 cf. dex-ter-ior , ex-ter-ior^ &c. We recognise the same force in our 
 ending -ish, as brack-ish, "rather salt than otherwise." 
 
 166 The word medius is perfectly analogous to Xdiog; the first 
 syllable of medius is the first personal pronoun, just as the first syllable 
 oiXdiog is the third. The word "middle" expresses that the subject, 
 the here^ is considered as equidistant from two other localities , and is 
 therefore properly expressed by a variety of the first personal possessive 
 pronoun. The possessive in Sanscrit is written madiya, and the word 
 signifying "middle" is madhya. In Greek the dh or th of the Sanscrit 
 is assibilated into (?, as in Ccog from &£0g, and the ior y is represented 
 by the second ύ of μk(^(5og , an assimilation which we often find. We 
 do not hesitate to recognise this word in the second part of ψμίβν. 
 The first part bears the same relation to δίά, Latin di (di-midius, 
 "through the middle"), that the first syllable of ν-μέίς does to δυο. 
 The same preposition also appears in the first syllable of η-μερα and 
 7j-μεQog (§ 150). By an assimilation very like that which has taken 
 place in με6ΰog for με^Logi we have allog for ahog^ with she same 
 ending signifying quality or relation. This is clearly identical with 
 the Sanscrit anuya (for aniya^ from ana the demonstrative) and the 
 Latin alius. The interchange of the liquids Z, n, in Greek and San- 
 scrit, is well known ; it is also common among the dialectical varieties 
 of the Greek language itself (Buttmann, Ausf. Sprachl. Vol. i. p. 74). 
 If the connexion between αλλο^, alius, and any a is admitted, we can- 
 not hesitate about referring to the same origin the Latin distinctive 
 pronoun ille or ollus, which stands therefore for inia or inius, and 
 differs from alius = anius, only as the Latin prefix in differs from the 
 Greek ανά\ of. άv-ήρi%'μog, in-numerus, &c.; and as ανά ^^J^ ανά is 
 only another form of κεν (below, § 183), akXog=favLog, and therefore 
 alius = ille, may be regarded as identical in origin no less than in 
 signification with 'H£ivog^=^'iihviog (above, § 135). We see the original 
 η in the Umbrian hon-dra = ul-terior ^ hon-domu = ul-timus, which 
 must be compared with our yon, yon-der, be-yond, the Gothic jains, 
 jaind, N. H. G. jener, &c. ( Varron. p. 3 1 5). The u of ul-lus, ul-tra, &c. 
 is merely a result of the contact with I: see Varronianus, pp. 250, 266. 
 
 X2 
 
308 THE NUMERALS. [bOOK II, 
 
 The long ο of oUm is confined to those cases where one of the liquids 
 is absorbed ( Varron. p. 201). In αλλήλων (from άλλοι άλλων) one 
 of the A's is absorbed in the long vowel, just as the i is omitted in 
 solus formed from the separative particle se and alius. The Sanscrit 
 synonym anyonya = anya-anya is melted down into one word in a 
 somewhat similar manner. 
 
 It may be objected to this resolution of μΈύοξ and αλλοξ into 
 forms so analogous to the qualitative ending in -ιοξ on the one handy 
 and the comparative in -ίων on the other, that there is actually a form 
 of αλλοξ {αλλοΐοζ) ending with the former affix, and a comparative of 
 /LtS(3og, namely, με6aίt8Qog (superl. μεβαίτατος). With regard to άλλοΐοζ 
 we have only to say that a reduplication of the ending would be quite 
 in accordance with the analogy of the Greek language , even though 
 we were not entitled to suppose that this elongation took place after 
 the word had, by assimilation, lost the outward features of its original 
 form. In Sanscrit an-ya and an-ya-tara are synonyms for "other," 
 "different." In fact αλλοξ is rather a distinctive word, than a com- 
 parative like έτερος: thus the grammarian says (Bachm. Anecd. ii. p. 
 376, 1. 4): αλλοζ μεν ετά πλειόνων λέγεται καΐ επί άλλοίον κατά το 
 είδος, ετεροξ δε ετά δύο. The same may be said οί μεΰαίτεροξ' for 
 although με6ίθ£, medius, have unquestionably a comparative meaning, 
 yet, in the general use of the word , it may have become necessary to 
 have a separate form to express more strongly that one of two objects 
 was nearer to the middle point than the other. In Sanscrit , madhya 
 has a superlative madhyama, like the Latin medio ximus. We have be- 
 fore remarked on the connexion of ϊ<5θξ and ϊδίΟξ-, which are both ana- 
 logous in meaning, as the latter is in form also, to μεόοζ {μεδϋος). It is 
 observable that Ydog has a comparative Ιβαίτερος, and ϊδίΟξ a compa- 
 rative Ιδιαίτερος. The former might be explained by the existence of 
 the word iuaiog (the name of the orator), which bears the same re- 
 lation to i6og that τριταίος, &c. do to τρίτος, &c. But we are inclined 
 to believe with Buttmann (Ausfiihrl. Sprachl. § 65, 5 , Anm. 6), that 
 the particular form of the antepenultima (that is, according to our 
 theory, the particular adverb to which the ending -τερος is attached) 
 was often chosen by the speaker or writer arbitrarily and to suit his 
 own ear. Thus we have from αΰμενος, a perfect passive participle, 
 άΰμενώτερος, αΰμεναίτατα, and άύμενεύτατα. 
 
 167 When the Greeks formed the comparative in -ίων, they em- 
 ployed a superlative in -ιβ-τος. Similarly, in Sanscrit a superlative 
 in -isht'Kas corresponds to a comparative in -iyans or -iyas. It will 
 be worth our while to examine these forms a little more minutely. 
 Most of the Greek , and many of the Sanscrit adjectives which form 
 
CHAP. Π.] THE NUMERALS. 309 
 
 their comparatives and superlatives in this manner, end in -us or 
 -r(l)-s. Why the former should prefer this method it is difficult to 
 say, but one may easily see why the latter, which already contain 
 part of the ordinary comparative-ending, should repudiate the common 
 termination. The fact, however, is, that it is only by a substitution 
 of the ending -us that adjectives in -ros pass to the qualitative form 
 in -ίων. That is to say, those Greek adjectives which already end in 
 -ρ-05, and those Sanscrit adjectives which have this or any other con- 
 sonant-suffix, drop it in the comparative, which is formed as if from 
 an adjective in -us. Thus, αΐΰχ-ρόζ naakes αΐΰχ-ίων, Βγβ -QOg, εχ^-ίων, 
 &c. In Sanscrit the vowel of the new positive is always affected by 
 guna: thus, kship-ra, "swift ," makes kshep-iy.as; kshud-ra, "little," 
 kshod-tyas; dura, "far," dav-iyas; yu-van, "young," yav-iyas; sthu- 
 la, "thick," sthav-tyas, &c. We think the long syllable in the penul- 
 tima of these Greek and Sanscrit comparatives, is occasioned by the 
 coalition of the final vowel w, from which they are always formed, 
 with the i of the ending. In the words %ά60ων , γλύΰΰων, &c. the 
 final u has been elided, and the short i as usual changed into y, whence 
 the assimilation. There are three instances of a somewhat similar 
 kind in Sanscrit: pri-ya, "dear," sthi-ra, "firm," and sphira, "swol- 
 len," form their comparatives by adding -yas to a guna of the roots, 
 thus, pre-yas, sthe-yas, sphe-yas. The reason of this is obvious: for 
 otherwise there would be a concurrence of eiy = eiiy, which would not 
 be allowable. 
 
 In compliance with the custom of all writers on grammar we have 
 talked of a superlative in -uu-rog or -isJi-th'as. If it were true, as 
 Grimm supposes (Deutsch. Gramm. m. p. 583), that the superlative 
 is properly formed from the comparative, it would be necessary to 
 infer that these two syllables constitute the ending of the superlative 
 as such, and that, in the Sanscrit at least, the syllable ish was a con- 
 traction of -lyas; and this is Bopp's notion {Krit. Gramm. der San- 
 skrita-Sprache, p. 11 3). To us it appears quite unnecessary to derive 
 the superlative from the comparative in any case. Nobody will assert 
 that the more common comparative ending -ta-ra, -τε -Qog, is contained 
 in the corresponding superlative ; why then should we suppose that 
 the other comparative is? The comparative expresses a relation 
 between two things; one is farther than or beyond the other (ta-rajj 
 or bears a certain qualitative ratio to it (i-ydns). The superlative 
 expresses that the thing is the last in the series, proceeding from (-ία) 
 the object nearest to {-ma) the subject, and so expressed by both 
 elements (-ta-ma). These ideas are surely distinct, and need not be 
 derived from one another. We believe that this superlative-ending 
 -L0-Tog is simplx the termination -tog, which we have already spoken 
 
310 THE NUMERALS. [bOOK II. 
 
 of, appended to an adverb in -tg. The adverb in -ig, by the side of 
 which we frequently find an adverb in -a (cf. μέγα, μόγις, μάλα, μόλις, 
 &c.), seems to be an abbreviated inflexion of adjectives in -i;- or -εν-. 
 The V is turned into i, which probably always followed it in the com-- 
 plete termination -vis or -hvis before the ν was vocalized ; thus the 
 Sanscrit root svid is represented both by νδ-ωρ and ίδ-ρώς ; hve^ the 
 root of viog, appears in φν ω and filiuSj εμοί = £με-φί^ appears as εμν 
 in Boeotian Greek (Apollon. de Pronom. p. 364), 'O'Ckevg = ^^ίλενς appears 
 as'/Aavg or ΟΙλενς (Hermann, de Emend. Gr. Gr. 415), &c. At any 
 rate all comparisons in -lcov and -Lijtog seem to be ultimately referable 
 to positives in -vg, -ενς, or to adverbs in -ig. Although these forms in 
 -vg or -vis do not exist, they may be presumed as possible from the 
 analogies pointed out by Buttmann (Ausfiihrl. Sprl. § 67, Anm. 6): 
 namely, by the side οι μακ-ρός, ο^Ιύχ-ρός^ and νεκ-ρός^ we have the 
 forms μηκ-ν-νω (in which the root-syllable is guna'd as in the Sanscrit 
 superlatives, and in μήκίΰ-τος, μήκος), αΐΰχ-ύνη, Άηάνεκνς. Similarly, 
 καλλίων, κάλλιύτος, though immediately connected with καλός ■=■ καδ~ 
 λός, most probably refer to a by- form καλ-λνς\ cf καλ-λν-νω after the 
 analogy of τίολνς, τίολλός, πρανς, τίραος, &c. "We do not, however, 
 agree with Buttmann in regarding the termination -vg and -og as 
 originally the same. On the contrary, the terminations in -vg, as will 
 hereafter be βΙιΟΛνη, contain a distinct pronominal addition to the 
 crude form of the word, while in the ending -og the nominative suffix 
 is immediately attached. It is no slight proof of the durability of these 
 suffixes, that Sanscrit words in -us or -u are always represented in 
 Greek by corresponding adjectives in -vg or substantives in -v. thus 
 for the Sanscrit svddus, pnthus, urus^ dgus, laghus, mridus and madhu, 
 we have the Greek ηδύς, τίλατνς, ενρνς, ώκνς, ε-λαχνς, βραδύς and 
 με^ν. The Sanscrit bahus has two Greek representatives , βα^'ύς and 
 ^(^χνς, just as ahis stands between εχις and ogptg, or han between φαν 
 (φόν-ος,εΛεφν-ον) and ξ^αν (Ο'άν-ατος^ ^'ν-ήΰκω,ΒβηΆτγ, Berl.Jahrb. 
 August, 1834, p. 229). There are cases, in which the superlative termi- 
 nation is -ο<5-τος = -ως-τος, as in πολλ-ού-τός, for which we might have 
 had 7toL•(^τoς', as may be seen by a comparison of ολίγοΰ-τός with ολίγιύ- 
 τος. For the form πολύς admits of an adverb in -ις, as τίολλός does of 
 an adverb in -ω^, and perhaps, after all, the difficult word τριπόλιβτος 
 in Soph. Antig. 857, may contain a remnant of this other superlative 
 oi πολύς. The name' Οδν66εύς=^ Ολνΰϋεύς ον'Ολυΰϋεύς points to an 
 original ολιγενς, which would fully account for ολίγυύτος (see Kenrick, 
 Herod, p. 281, and Varron. p. 142). The terminations -τερος, -τατος, 
 are sometimes appended to adverbs in -ις from by-forms in -ενς, as in 
 ληλίύ-τερος, πτωχίΰ-τερος, άρπαγίΰ-τατος, &c. We must not confuse 
 this with the fact that adjectives, in ^ης or -ψς and -εντς, and even 
 
CHAP. Π.] THE NUMEEALS. 311 
 
 some in -ων {-ov-g), -t| (-L7C-g), regularly form the comparative and su- 
 perlative in -8u-t£Qog, -Eo-Tatog, for in these cases we have remains 
 of the adverb in -ωg. Nor does the apparent analogy of magis favour 
 the derivation of μέγL6-τog from an imaginary comparative μέγLg for 
 μεγίων ; the comparative of mag-nus has suffered all sorts of abbre- 
 viations; in major the gi has become J as usual, while magis has lost 
 the u of the ending : the g has become ζ in the Zend mazista, and has 
 vanished altogether from the French mats. There is no reason what- 
 ever why there should not be a form μεγε -vg, as well as μεγε -QOg. On 
 the contrary, the existence of μέγL•ό-τog is a reason for presuming the 
 existence of μεγεvg, just as we might presume the existence oVAqEvg, 
 if we did not know it, from the superlative ccQio-Tog, and as a positive 
 vasus has been inferred from the Zend superlative vahista (Benary, 
 Berl. Jahrb. August, 1843, p. 230). In fact, the adverb μόγL•g, as we 
 have already mentioned, is itself an argument in favour of the same 
 supposition. We find £Ka-T8Qog by the side oi SKau-Tog, because, as 
 we shall see, έκα is an adverb as well as sxag. So also some adjectives 
 in -vg add the terminations -τερο^, -Tarog to their neuter used adver- 
 bially, as well as employing the other adverbial form in -i,g. Cf γλν- 
 KvraTog ^vith yXvK-iu-Tog. 
 
CHAPTER III. 
 THE PREPOSITION. 
 
 168 Independent value of prepositions as positional words. 169 General view of 
 the Greek prepositions according to the elements of which they are composed. 
 170 Eig=iv~g and έν. Apparent use of ώς as a preposition. 171 ΤΤρο', πρό-ς, 
 7tQ0-TL, άν-τί, αν-τα. Expression of space and time respectively by the same 
 prepositions. 172 'Επί and άμφί identified with abhi and ob. 173 Their 
 agreements in meaning with each other. 174 The use of επί in composition 
 to signify mutuality or interchange , also points to its affinity with άμφί. 
 175 The same appears also from its use with the dative to signify combination 
 or coexistence. 176 Etymology of ε|. Use of έκϋ'νηβκω and εκφέρω. 177 
 'Από and παρά. Connexion of the latter with περί, προς, proe , prceter^ and 
 per. 178 Περί and the Sanscrit pari, para. 179 Ύττό and νπέρ. 180. ^lU 
 and δνς-. 181 Σνν and μετά. 182 Κατά, another form of άντα. 183 'Ανά 
 and ν,ατά are properly correlatives, and are most satisfactorily examined 
 together. 183 'Ava, av-, ά-, and άπο- as negative prefixes; κατά as an af- 
 firmative. 185 The intensive άλφα. 19>&"Av and κεν related to ανά and γ,ατά. 
 187 Position of av and γ,έν in the sentence. 
 
 168 OEEPOSITIONS are pronouns or positional words in 
 the strictest sense of the term*. They express rela- 
 tions of place, and in their ordinary use are employed to denote 
 the relative positions of visible objects. Grammarians tell us 
 that they govern cases, and it is the prevailing practice to 
 arrange them according to the cases which they are said to 
 govern. But this is palpably erroneous : for in all languages 
 which have inflected nouns, a case may express by itself any re- 
 lation which the addition of a preposition could give to it and, 
 
 * The preposition is fully discussed in the new edition of Pott's 
 Etymological Researches, the first part of which has just appeared {Ety- 
 mologische Forschungen auf dem Gebiete der Indo-Germanischen Sprachen; 
 Zweite Auf lag e in vollig neuer Umarbeitung ; erster Theil: Prapositionen. 
 Lemgo und Detmold, 1859). The whole of the volume (859 pages) is 
 devoted to this subject. As usual, Pott exhibits an inexhaustible abun- 
 dance of lexicographical details and illustrations, but he seems to be still 
 unacquainted with the analysis of these pronominal forms, which was 
 first expounded, twenty years ago, in the present work, and without 
 which, as we conceive, no definite results can be obtained. In general 
 we think he has rendered himself liable to the happy remark, which he 
 made upon Benfcy {Jahrb. d. Wiss. Krit. 1840, p. 629), that "it always 
 excites a feeling of dissatisfaction, when the head of the nail is missed, 
 though we see the hammer falling with busy haste on all sides of it." 
 
CHAP. III.] THE PKEPOSITIONS. 313 
 
 in languages which, like the Sanscrit and the modern Russian, 
 have a complete assortment of cases, many relations of place 
 are invariably expressed by the cases without any particle pre- 
 fixed. Such would have been the fact in the Greek language 
 too, but the rules of euphony, convenience, the influence of 
 writing, and a multitude of other causes, have contributed to 
 mutilate the terminations of the nouns as well as of the verbs, 
 and thus prepositions, the force of which was originally included 
 in the case-endings, have come to be prefixed for the sake of 
 greater distinctness, just as the particular noun is placed after 
 the pronoun, called the article, in repetitions, and just as the 
 nominative case is prefixed to the verb. 
 
 169 There are eighteen Greek words which are commonly 
 reckoned as prepositions: six monosyllables, εΙς, εν, ff, tcqO, 
 TtQog, 6vv, and twelve dissyllables, άμφί, ανά, αντί, από, δι,ά, 
 επί, κατά, μετά, παρά, περί, υπέρ, νπό. We shall consider 
 these according to the relations which they express, and not ac- 
 cording to any arbitrary division of former grammarians*. Since 
 the prepositions retain their original meaning, as words indicating 
 positions and directions in space, more characteristically than 
 any other pronominal words, and also present the simplest com- 
 binations of the original elements of the pronouns, we will, pre- 
 viously to examining these Greek forms separately and in detail, 
 endeavour to point out their etymological analysis in a sum- 
 mary manner, and to explain the general principles of their 
 composition. 
 
 It has been stated before that the primitive pronouns are 
 three in number, expressing respectively the positions here, near 
 to the here^ and there ^ and that diflPerent modifications of di- 
 rection or position may be denoted by combining these original 
 stems with one another or with the particle la or ra. On 
 examination it will appear that aU the Greek prepositions, with 
 the exception of δυά which is a form of the second numeral, are 
 compounds of at least two of the primary elements, or of one of 
 
 * The object of this chapter is to discuss the origin and signification 
 rather than the syntax of the Greek prepositions: as a supplement to the 
 eases they are fully treated in the Greek Grammar, articles 470 — 488. 
 
314 THE PREPOSITIONS. [bOOK II. 
 
 them with -ra. We have already adverted to the principles 
 according to which we would arrange and classify all pronominal 
 compounds (§ 130). After a careful dissection of all the pro- 
 nominal forms with which we are acquainted, w« have arrived 
 at the conclusion, that if any one of the elements of position 
 is combined with -r«, it indicates motion and continuation in a 
 direction of which the element in question represents the point 
 nearest to the subject; and that, by subjoining any one of the 
 pronominal elements to any other of them, we denote a motion 
 or continuation from the position denoted by the first element 
 towards that indicated by the second. Thus we have seen, that 
 the second element when prefixed to -ra (as in tva-va) expresses 
 motion onwards from the position indicated as near^ so as, in 
 fact, to coincide with a word indicating the third position {to) ; 
 and that the first element subjoined to the third (as in ta-mii) 
 expresses motion or continuation from the third position to- 
 wards the first, so as to coincide with the second position 
 (cf. fini-timus, &c.). We shall find this method most amply 
 illustrated by the Greek prepositions. Of these, πα-ρά, ττε-ρί, 
 π-ρό, 3Τ-ρό-5, are compounded of the first element and -ρα. In 
 the first, which is also written τια-ραί^ we find both elements 
 in the simplest form. In the second, in which traces of a heavier 
 ending still remain, the vowel of the first element has assumed 
 the lighter form f, according to a principle which will be more 
 fully explained hereafter. In τί-ρό and ;Γ-ρ6-?, which are in 
 fact one and the same word, another element has been subjoined 
 in the s , indicating motion or transitiveness , and probably a 
 shortened form of the affix -og, -(jto, which plays an important 
 part as the sign of the genitive case. In consequence of this 
 addition, the root-vowel has been dropt before the liquid, and 
 a medium weight given to the vowel of the termination. The 
 forms π-ρο-τί^ τΐο-τί^ also used for 7ΐ-ρό-£ , are compounds , one 
 of the'preposition π-ρό , the other of the simple element of the 
 first pronoun, with the element of the second under the form 
 ^t=S (§ 1^2); and both, therefore, denote (the former more 
 strongly) motion from the first to the second position. A similar 
 form is με-τά^ which is compounded of the first and third elements, 
 and signifies "with" as a connexion between the here and the 
 there^ and "after" as denoting an approximation to their union. 
 
CHAP. ΠΙ.] THE PEEPOSITIONS. 315 
 
 The third pronoun is twice subjoined to the second in %α-τά== 
 κεν-τά^ as it is to the first in με-τά^ and the meaning which re- 
 sults is analogous. In the Sanscrit sa-ma^ the first element is 
 appended to a form of the second; the meaning "with," which 
 results, is explicable in much the same way as the similar, sense 
 oi με-τά. If the latter expresses "with" as implying a junction 
 of the here with the tliere^ so sa-ma may convey the same sort of 
 idea as implying an union of the near with the here. We shall 
 see by and by how this differs from <jvv = Fa-v. The preposi- 
 tions V'TCO , ν-πέ-ρ , are related as ta is to ta-ra ; the latter ex- 
 presses a continuance of the direction implied by the former. It 
 must be observed, however, that the final vowel of νπό^ like that 
 of άτΐό, tcqO^ indicates the loss of the genitive s, so that the full 
 form was ι;ψ , or vTtog = νπό^εν. The first syllable of νπό is 
 a vocalization of the second pronoun /« or sva. In Sanscrit it 
 appears as 'U in ii-pa^ and in Latin as su in su-h. Of the suc- 
 cessive mutilations of this stem from sva to hi, ΐ, we have 
 already spoken , and we have also hinted the general tendency 
 of the aspirated labial to vocalize itself into i as well as u. 
 We shall, therefore, have little difficulty in recognising it in the 
 Greek v, the Latin sii^ or even in the Sanscrit lo. According to 
 this analysis ν-τίό should signify a'direction from the immediately 
 near to the here., and ϋ-τιε-ρ a continuation of motion in that 
 direction ; we shall see presently that this agrees very well with 
 the common use of these prepositions as the correlatives "under" 
 and "over." The element a (e), must be the ultimate form 
 either of ha from /a, or of na. There are many cases in which 
 this element appears in combination with an existing na, and we 
 must suppose, in those cases, that it is the pronoun J^a prefixed 
 to 7^α, especially as there are occasionally traces of a lost digamma. 
 Now there are three Greek prepositions in which we have the 
 compound of α or e and na: — ά-νά, sig = e-vg., and ε-ν. The idea 
 of farness is strongly conveyed by the first; the last and 
 shortest form conveys simply and explicitly the idea of locality ; 
 and in the second this is combined with the idea of motion. 
 The element a (e) appears in three other prepositions, α-τίό, 
 8-7CL, and E-^=8-Kg. In these compounds there is reason to believe 
 that the ά or ε is the residuum of the element na. For other- 
 wise we must have an unmeaning reduplication of the idea of 
 
316 THE PKEPOSITIONS. [book II. 
 
 nearness in the preposition εξ, and ά-ΛΟ would, on that sup- 
 position, be identical with v-Tto = fa-Tto. The lighter vowel is 
 chosen in the two latter because their terminations are heavier. 
 The two former are indicative of a direction from the distant to 
 the here : the latter of motion from the distant to that which 
 is near. The first and third have the cognate meanings "from" 
 and "out of:" that which comes from the distant hitherward 
 comes "from," or, if through an intermediate spot, "out of" it. 
 The signification of επί is "upon" or "across;" the vowel of 
 its termination, as well as that of τΐε-ρί^ points to an original 
 expression of locality by way of case-ending. A comparison of 
 άτΐό, αψ (a-jrog), τί-ρό, Tt-QO-g, convinces us that both ά-τίό 
 and Tt-Qo originally possessed the final s , which we have men- 
 tioned before as the index of transitiveness. Άν-τΙ and άμ-φί 
 are compounds, of which the first part is ά-νά. The latter part 
 of άν-τί is the same as the termination of τίρο-τί = Λρό -s, and 
 is therefore equivalent to the sign of the genitive case. As 
 we have τίαρά by the side of περί, so we have αντα as well 
 as αντί^ and these again are represented by ει-τα and ε-τι 
 (below, § 193). The termination of αμ-φί is a mutilation of the 
 second pronominal stem, which is often used to form the lo- 
 cative or dative case. In ^olic and Doric αμ-φί is written 
 άμ-πί^ but we regard this as merely a dialectical corruption. It 
 is clear from αμ-φω = ava 8νω, that φι for <5φι is the more 
 genuine form. A similar but more general corruption has taken 
 place in s-jrt, the last syllable of which does not appear to be 
 a case-variation of the ending of a-jro, but a substitute for φι, 
 since in its proper meaning ε-πί is strictly locative. If therefore 
 the first syllable ε is a residuum of av-^ we may trace αμ-φί and 
 Ιπί to a common origin, and we shall see that their use is not 
 so very different. 
 
 170 We commence our separate examination of the Greek pre- 
 positions with εΐξ and εν, which we believe to have been originally 
 identical, and which express the simplest and most elementary notion 
 of locality — the being in a place. In the Sanscrit system of cases 
 there is one which by itself expresses this relation, and therefore it is 
 not to be wondered at that there should be no Sanscrit preposition 
 corresponding to εν. Now in the Sanscrit pronouns the termination 
 of this locative case is i-n: thus, tasmin^ "in this," kasmin, "in 
 
CHAP. III.] THE PREPOSITIONS. 317 
 
 what?" etasmin^ "in the same," yasmin^ "in what," sarvasmin, "in 
 every thing." The pronouns of the first and second person are excep- 
 tions: these form their locatives in i, like the bulk of the Sanscrit 
 nouns*. Bopp supposes (Vergl Gramm. p. 231) that the termination 
 in -i-n is of later origin than that in -i, and that the final η is merely 
 a V εφελκνΰτικόν , or an arbitrary addition. We do not believe that 
 V εφελκνβτίκόν in Greek words is a merely capricious termination: 
 there is no reason why an η should be added ; and it is very easy to 
 see, on the contrary, how time might have caused an abbreviation of 
 the ending, which is so liable to become evanescent (§ 114). Besides, 
 the use of the full termination in old words, like the pronouns, in 
 Sanscrit, and its appearance in the Greek and Latin pronouns are to 
 us sufficient proofs of its being the most ancient form of the locative. 
 The Greek pronouns in which this termination is found, are εμίν, 
 τεΐν, τίν, IV, εϊν, 6φίν, φίν, 'φίν, ημΐν, νμΐν, Λρίν, &c. In Latin it is 
 even more extensively used , though in accordance with the genius of 
 the language it is generally changed into i-m, except in compounds. 
 Olim is obviously the locative of ille or olle, and means "in that 
 time.''^ The same appears with regard to the other adverbs of time, 
 quom and interim. Nam is the locative of some lost pronoun analo- 
 gous to tarn and jam, and means "in that" = "for," just as sin 
 means "in this " = "if," the one pointing to something that has been 
 mentioned, the other to something about to be mentioned. Enim, 
 which is related to nam^ just as tango is to contingo (Bopp, Vergl. 
 Gramm. p. 534), is a compound word like the Sanscrit a-na, ena, 
 e-ta, e-ka : compare the Greek vvv, vCv, νόΰφι, with κεΐ-νο£,τη-νος, ά-νά. 
 The same termination with the same meaning is found in istim, instin- 
 c(e), illim, iUin-c(e), hin-c(e), intrin-secus, extrin-secus, quin, in-de^ 
 utrin-de, un-de, subin-de, dein., &c. The first part of perendie, which, 
 as we showed before, is formed like the Sanscrit paredyus, or apare- 
 dyus^ from a pronominal adjective signifying "other" or "further," 
 and the common word "a day," exhibits the locative of the adjective 
 very clearly, as does also the Sanscrit equivalent apare = apara-i. It 
 is obvious that this termination -in or -en is identical with the Latin 
 preposition i-w, which from the disuse of the locative case is always 
 employed to express the simple notion of locality. The difference of 
 vocalization need not prevent us from identifying εν with in. The 
 Latin in may express the meaning of all the Greek compounds of a- 
 or ε- with να, and even the negative prefix ava•, which is a formation 
 of the same kind: e.g. ccva μέρος =^in-vicem; εις την 7tόL•v=^in urhem; 
 
 * Those Sanscrit nouns which form their locative in e are no variation 
 from the general ending in ?', for in all those nouns the crude form ends 
 in «, and e = ai. 
 
318 THE PREPOSITIONS. [bOOK II. 
 
 εν rfj TtoXst = in urbe ; άνήρι^μοζ = in-numerus. We must therefore 
 be contented to suppose that the first part of Iv ^=^ i-n is the ultimate 
 weakened form of the second personal pronoun, the stronger form of 
 which appears in the Greek locative i-va^ " where," and both stronger 
 and weaker in the forms analyzed in the last chapter but one. There 
 are traces of this locative of the second pronoun in the prepositions 
 dv, dvl; and in this pronominal sense Iv itself is often used in the 
 best writers. Thus, Sophocl. CEd. Tijr. 27: ε^ <5' ό πυρφόρος %-Eog 
 ΰκή'φας ελαννει. And thus we see why the first and second personal 
 pronouns, when it was necessary to give to them that more marked 
 and definite expression of locality which is conducive to their dis- 
 tinctness as numerals, were occasionally combined with this form of 
 the third demonstrative pronoun. 
 
 That εΐζ and εν are essentially the same word has not escaped the 
 notice of Greek scholars. In Pindar we have εν with an accusative 
 case expressing motion to a place , where we should certainly find εΙς 
 in Attic (Pyth. ii. 11 and 86; v. 38; Nem, vii. 31), and there are 
 passages in which εΙς with an accusative is found instead of εν with a 
 dative (Person ad Eurip. Phceniss. 1381). It might be suggested that 
 εΙς = Ivg was the basis both of Ig and εν, the former omitting the v, 
 the latter the g. But it is not necessary to regard them in this light. 
 We consider them to be related simply as μείς, μεν, δεις and δεν; 
 namely, μεν, δεν, and εν are three locatives, and μείς, δεις and εΙς 
 are three transitive forms incorporating the idea of locality. That 
 £tg expresses motion to a place (not unlike δε), while εν expresses 
 nothing but locality, or the abiding in a place, is to be explained 
 from the addition of the element s, wliich is also a mark of the 
 nominative or relation of subjectivity. The termination n(a)s appears 
 also in the Latin preposition tra-ns. 
 
 There is only one passage, so far as we know, in which the pre- 
 position £tg has occasioned any difficulty. It is in Euripides, Bacchce, 
 742: 
 
 τανροί, δ' νβρίόταΐ κείς κέρας Ο'νμονμενοί 
 το 7ΐρόΰ&εν — 
 from which Yirgil has taken his irasci in cornua {Georg. iii. 232; 
 jEneid. xii. 104), and which is also imitated by ^lian (Hist. Anim. 
 II. 20 and iv. 28, where we have υβρίζειν εΙς κέρας). We think the 
 line in Euripides explicable from the idea of " looking towards ," im- 
 plied in the classical use of εΙς , and from the passage in his Helena 
 (1558), where a bull is described as κυρτών τε νώτα κείς κέρας 
 Λαρεμβλέπων. 
 
 With regard to the so-called usage of ώς as a synonym for εις, it 
 need only be observed that this adverb is constantly employed by the 
 
CHAP. ΠΙ.] THE PREPOSITIONS. 319 
 
 best writers with prepositions of motion like εΙς, ετίί, TCQog, especially 
 the latter, followed by an accusative, so that the three words are 
 equivalent to cog with a participle, and the construction's explained 
 in the common grammars by a supposed ellipse of the i participle. 
 This, of course, is an unnecessary hypothesis ; but it is generally easy 
 to see that there is an ellipsis of the preposition jtQog in the apparent 
 use of Sg for eig with the accusative, "which is generally restricted to 
 the names of persons. Thus we read (Demosth. Phil. in. p. 113): elg 
 ΦωκBag wg TCQog ΰvμμάχovg sjtOQeveto, where elg Φωκ8ag strictly 
 speaking denotes the name of the country, whereas og JtQog <Jvμμάχovg 
 is a personal reference: from which mode of speaking arose the use 
 of ag alone with the names of persons, in nearly the same sense as 
 elg with the names of things; e. g. TCQ^G^Big πετΐομφεν Sg βαΰίλεα 
 for og TtQog βαΰίλεα (Demosth. Phil, i. p. 54). 
 
 171 The most elementary notion of situation next to that of 
 mere locality is the notion of that which is before us, in which the 
 notion of forwardness, or motion forwards, is included. The simplest 
 word for expressing this in Greek is τίρό, which is a compound of the 
 first pronominal stem under the form pa with the termination r«, sig- 
 nifying, as we have seen, motion, and conveying the idea of beyond. 
 This word, like gtg and εν, appears among the numerals; for a super- 
 lative form of it (TtQOTog) is used as the ordinal of the first number. 
 Another form of jtQO is προτί (Sanscrit prati) or πο-τί, where rt has 
 taken the place of the genitive -g. That jt-QO-g is only the fuller and 
 more genuine form of τΐ-ρό appears as well from the comparison of 
 a-7to, «1^, suggested above, as from the fact that Tt-Qog, under the still 
 more complete form Ttaqog., is used as a synonym for tcqo. The idea of 
 motion, conveyed by 7C-q6g or π-ροτ-ί, has been before explained; the 
 form 7tρόg combines with a word expressing here, the syllable ra indi- 
 cating motion, and the termination s, which has much the same force 
 (compare gtg) : Ίίρο-τί contains the same affix in a fuller form , and 
 αν-τί, which is all but a synonym of τίρό, also terminates in this geni- 
 tival τι = Ο'εν = g. As ρα ultimately = va, and as ά-νά = Fa -νά, the 
 only difierence between ττ-ρο-τί and α-ν-τί is that which is indicated by 
 the first syllable; namely, the starting-point in τί-ρο-τί is the here, in 
 ά-ντ-ί, the near. In the Latin ante we have e for i, as in mare, leve, &c. 
 And here we must notice the fact, that, whereas the notion of priority 
 in time is expressed in Greek by Λρό or 7tάρog and not by αντί, the 
 same meaning is conveyed in Latin by a7ite and not by j9ro. This 
 interchange in meaning belongs to a class of phonomena not uncommon 
 in cognate languages, and is by no means inconsistent with the etymolo- 
 gical identity of πρό, pro, and αντί, ante. In the sense in which αντί 
 
320 THE PREPOSITIONS. [bOOK II. 
 
 most nearly corresponds to πρό, i. e. "in front of," we have more 
 usually the form ccvta, though elisions often prevent us from saying 
 which vowel is represented under the apostrophe (see Greek Grammar ^ 
 474, (aa)). We shall see below κατά = κεν-τά, which is really a form 
 of ίίντα, has a temporal meaning, which is not in accordance with the 
 temjDoral values of Ttgo or ante^ though their local meanings ultima- 
 tely coincide. 
 
 In general there is some difficulty in reconciling the local with the 
 temporal values of the prepositions. That the conception of time is a 
 refinement or modification of that of space has bieen mentioned in a for- 
 mer chapter (§ 54). And it is clear that position in space is used for 
 the measurement of time. The dial-plate of a clock with its moving 
 hands is a sufficient indication of this. And the words καιρός (from 
 κείρω) and tempus (from temno) show that the measurement of time is 
 derived from that of distance or magnitude. The word %αιρ6ξ in parti- 
 cular is a synonym oi μετρον. Thus Plato (PoUticus, p. 284 e), when he 
 divides the metretic art into two portions, defines the second as con- 
 taining those arts οτίόύαι τίροξ το μέτριο ν καΐ το τίρετίον κοίΐ τον 
 καιρόν και το δέον καΐ τΐάν^' οτΐόΰα εις το μετρον άτΐωκίΰ&η των 
 εΰχάτων. Hesiod says, Op.et D. 692: μέτρα φνλά<56ε6^αι, καιρός δ^ 
 ετά τΐάΰιν αριΰτοξ. Pindar also combines the two words, Pyth. iv. 286 : 
 καιρός προς άν^'ρώπων βραχν μετρον έχει. It seems to refer es- 
 pecially to the distance in aiming at a mark. Thus in -^schylus, Agam. 
 363 we have %ρο καιρόν, "short of the mark;" and in Pindar, Nem. 
 vm. 4, καιρόν μη τίλανα^εντα seems to be equivalent to βκοτίον μη 
 πλανα%εντα (vide ad Pind. 01. ix. 38 ; Varronianus, p. 392). In Pind. 
 Fragm, 150, κκ^ρο^ is actually opposed to χρόνος^ for we must construe 
 the words διακρΐναι εν καιρώ together; "there was plenty of time for a 
 person looking on to discern the whole proceeding accurately" (ijV δε 
 Ιδόντα διακρΖναιπολλος εν καιρώ χρόνος). But although time is gene- 
 rally expressed in language denoting primarily the relations of space, 
 the analogy is not always maintained. Future time is generally re- 
 garded as before us, and as approximating. Hence, as we shall see 
 (below,§§370,371), it is expressed in the tenses of the verb, by the pro- 
 nommal element denoting nearness in space, whereas past time is indi- 
 cated by the augment, which is the fragment of a preposition denoting 
 distance. When, however, we express these ideas by prepositions and 
 their cases , we find a good deal of confusion. For while future time 
 IS expressed by μξτα with the accusative, which really means motion 
 with a view to conjunction, as βη μετ Ίδομενηα, "he went after (to 
 overtake) Idomeneus," compared with εγενετο μετά ταντα, "it came to 
 pass after these things," past time is expressed ])y πρό and in Latin 
 by ante , both of which indicate an object as immediately before us. 
 
CHAP. III.] THE PBEPOSITIONS. 321 
 
 Again, although κατά and μετά both express motion in the track of 
 someone proceding (as in κατά jtodag^ "at the heels," μετ αντόν, "after 
 him"), we find that καθ"' ημεραν means "day by day," and μεΟ"' ημέραν, 
 "in the day-time." And while επί with the dative signifies "after," 
 as ετΐί %l6vl τΐεόονΰ^^ "after snow has fallen," επί with the genitive 
 denotes priority in time, as ετά ζίαρείον, "in the time of Darius," i. e. 
 at a former period. As ετΐί has this meaning because past time de- 
 notes a reckoning upwards (below, § 173), we find that νπερ την φΟΌ- 
 ράν means "before, earlier than the destruction" (Plat. Tim. 23 c); 
 and this may be in part the meaning of ανά in the augment. On the 
 whole, then, we see that the ideas of "before" and "after" in space 
 and time are qualified by the special reference in the particular case, 
 and that while τΐρο εμον in space means "before me," i. e. in front of 
 me, ττρο εμον in time means "before me," i. e. in the progress of 
 events, in the torch-race of life, which is "behind me" in regard to 
 the appointed post of the successive runners. 
 
 172 ΛΥβ have already suggested that ε-πί and άμ-φί=άνά-φυ are 
 by-forms of one and the same preposition. If so, the primary mean- 
 ing of both must be superposition combined with an idea of juxta- 
 position, arid this resolves itself into the ideas of upon and around. 
 Before we examine this meaning we will point out the connexion 
 between these particles and their Sanscrit and Latin synonyms abhi 
 and oh. Wilson's account of abhi is as follows : ^'■Abhi is a prepo- 
 sition and particle implying (1) superiority in place, rank or power 
 (over, upon, against, above), (2) proximity (near), (3) separation (seve- 
 rally), (4) wish, desire, (5) conjunction, particularizing (to, with re- 
 spect to). Thus abhi-kramihm ^ to overpower; abhi-ganiuti^ to ap- 
 proach, abhi-khydtwi^ to speak to; abhi-bhava^ disgrace; abhi-ldsJia^ 
 desire; abhyagnin^ on the fire." From this abhi we have the adverb 
 abhitas, "near." The conjunction and inseparable preposition apt, 
 which is a still nearer approach to tho Greek ετίΐ, occurs as a prepo- 
 sition before a few roots, and seems to signify "over;" the roots dhd, 
 "to place," nah, "to fasten," receive the meaning "to cover," when 
 this particle is prefixed. The a of api is usually omitted ; e. g. pinad- 
 dha, "covered," "clothed." As a conjunction, we very often find api 
 by itself, without any rejection of the a, in which case it signifies 
 "even," "though," "yet," "assuredly," "moreover," "therefore," 
 "also;" it is likewise an interrogative particle; the compound Icim- 
 api is equivalent to qtdspiam, or, if preceded by the relative, to qui- 
 cunque (Wilson sub v.; Bopp, Kritische Grammatik der Sanskrita- 
 Sprache^ p. 55 ; and Glossar. Sanscr. p. 73). So that api and abhi 
 seem to be modifications of the same word. This view of the con- 
 
322 THE PREPOSITIONS. [book it. 
 
 uexion of ετίί, abhi, and apij is supported by the common usage of 
 Ιτίί^ which always conveys the idea of nearness or approximation, even 
 when it is followed by an accusative and signifies motion. Now we 
 may fairly conclude that the nasal has fallen out in abhi for ambhi = 
 αμφί, just as in abhra = ομβρος, in ubhau =: αμφω, ambo, &c. The 
 identity therefore of STti and abhi farther confirms the connexion be- 
 tween £7tL and άμφί. On the other side ετύ coincides so remarkably 
 in some of its applications with the Latin ob, that it is difficult to 
 suppose that the Greek and Latin prepositions can have had different 
 origins. Thus ojptimus from ob manifestly denotes uppermost, and 
 therefore stands like supremus and summus in complementary oppo- 
 sition to infimus and imus. Here we have ob with the common 
 meaning of Ιτίί^ "upon." But it corresponds to the other meanings 
 οίΐτίί and αμορί-, as Festus tells us (p. 178 Miiller): '•'ob prsepositio 
 alias ponitur pro circum, ut cum dicimus urbem obsideri, ob-vallari, 
 ob-signari...dliaB pro ad ponitur, ut Ennius: ob Bomam noctu legiones 
 ducere ecepity et alibi ob Trojam duxity So also in the fragment of 
 his Telamon quoted by Cicero (Tusc. Disp, iii. 18): Hicine est ille 
 Telamo, modo quern gloria ad coelum extulit, quern adspectabant^ cujus 
 ob OS Grail ora obvertebant sua. (Compare the ετίί-ΰτρεΛτος αιών of 
 ^schyl. Choeph. 350.) The same may be said of the use of obeo, "to 
 go to," "to visit." The English preposition "upon" conveys much 
 the same idea as sjtL in such passages as Hom. Od. v. 17 : ol κέν μιν 
 τίψτίοιεν lit ευρέα νώτα ^αλάΰΟης. Now it is pretty clear that ob is a 
 shortened form of amb. Thus obba stands of the side of αμβίξ, and 
 Festus has told us that there are many usages of ob in composition in 
 which it corresponds in meaning to άμφί or τίερί, even more than to 
 επί: compare obsidere urbem with the Greek εφέξεΰ^'αί on the one 
 hand, and τίερικαΟ^ηΰ^'αί κνκλφ το τείχος on the other: if ob-scurus 
 reminds us οίετίίΰκιοξ and ob-edio of ετζακονω^ ob-esus (bassus) suggests 
 άμψίλαφηζ^ and ob-erro may be translated by περυτίλανώμαι ; and if 
 occupo corresponds to ετΐίλαμβάνω^ so does ob-liquus to άμφίλοξος. 
 
 173 But we must now show that ετΐί and άμφί agree occasionally 
 in meaning with one another. The most striking instance of this is 
 furnished by the syntactical usage of ετίί with the genitive and dative 
 as corresponding to that of άμφί or Λερί with the accusative and 
 dative. With the genitive case, επί denotes, as we have elsewhere 
 shown*, superposition with separation. The latter part of the mean- 
 ing belongs more immediately to the case itself, but of course the pre- 
 position must be able to reconcile itself to such an apphcation. Now 
 
 * Gr. Gr. Art. 483. 
 
CHAP. III.] THE PEEPOSITIONS. 323 
 
 there are two modes of connecting the idea of superposition with that 
 of separation. The first is when we imply that the separation is 
 total, but that a line drawn over one object will pass over the other. 
 Here then we denote direction or motion at a certain height or dis- 
 tance : e. g. a ship at sea was spoken of as up in the air (μετέωρος) ; 
 hence such phrases as τΐλεΐν επΙ Σάμου (Thucyd. ι. 116); and post or 
 distant time supposes a reckoning upwards, as we shall show farther 
 when we come to the temporal augment; hence we have such phrases 
 as επΙ ζίαρείον Ιγενετο (Herod, vi. 98). It is clear that we might say, 
 with a slight difi'erence of meaning, %λεΙν άμφΐ Σάμον, and άμφΐ τον 
 Δαρείου χρόνον. The other mode of considering superposition and 
 separation together is where we signify that the whole of a super- 
 imposed object does not rest on the supporting surface. Thus we 
 imply only a partial superposition when we say that planks are laid 
 across piles fixed at intervals (Herod, v. 16: ϊτιρια ετά των βτανρών 
 εΰτηκε)^ or that burdens are placed on the head or shoulders, so as to 
 extend beyond them on both sides (Herod, n. 35 : oi μεν ετά των 
 κεφαλεων φορέονόί, αϊ δε γυναΐκεζ ετά των ωμών). Hence such phrases 
 as ετά ^•ρόνου κα^^ίζεΰΟ'αί^ εφ ϊτιττου οχεϊΰ^αι^ because in sitting and 
 riding the legs hang down by the side, and the Spartan woman gave 
 her son his shield with the words : η ταν η ετά τάς (Plutarch, Lac. 
 Apophth. p. 241 ε). Much in the same way, we find άμφί used with 
 the accusative, as inEurip.P^cBTi. 122: άβτΐίδ άμφϊ βραχίονα κονφίζων. 
 With the dative επΙ and άμφί give many traces of a cognate meaning; 
 thus we can say with the same signification of proximity οικεοντε^ επϊ 
 Σίΐρυμόνυ (Herod, vii. 75), or άμφΐ divaig Ενρίτιον (Iph. T. 6). As 
 however άμφί has retained its original form more completely than 
 ετίί, we must expect that the combinnd meanings "up and about, or 
 around," will be more consistently retained by the stronger word. 
 
 1 74 There is a very remarkable reference to the original meaning 
 of επί and άμφί, when the former is used in composition to signify 
 mutuality, an interchange, a running of one thing into another. As 
 this meaning of εττΐ has not been sufficiently noticed by Greek scho- 
 lars, we shall illustrate it by examples. In this sense ετΐί is frequently 
 prefixed to άλλος, or some word like it, and the origin of the meaning 
 seems to be the same as that of αλλήλων, with which indeed it is 
 combined, if we may adopt Hermann's very probable emendation of 
 Sophocles {Antig. 57): 
 
 τρίτον δ' άδελφω δνο μίαν καθ' ημεραν 
 αϋτοκτονονντε, τώ τίαλαυτίώρω, μόρον 
 κοινον κατειργάύαντ ετιαλλήλοιν χεροΐν. 
 The expression of mutuality or interchange by juxtaposition seems to 
 
 Υ 2 
 
324 THE PEEPOSITIONS. [bOOK II. 
 
 have been the result of a natural love of brachylogy or the shortest 
 mode of expressing our meaning. When we say "they hurt one an- 
 other," we mean that A hurt B, and also that Β hurt A: which would 
 certainly not be expressed by saying "the one hurt the other." Simi- 
 larly in Greek, if we wrote αλλοϋ SKTELvav αλλονξ, we should merely 
 express that the one party killed the other, but if we put the two 
 pronouns together and write άλλοι άλλους (άλλήλονο) s^teivav, we 
 express that the slaughter was mutual, that there were killers on both 
 sides. It is therefore by an obvious contrivance for the purpose of 
 saving a superabundance of words, that, whenever reciprocity is in- 
 tended to be expressed, the subject and object are placed in immediate 
 contact, in order that the hearer or reader may combine them both 
 into one idea of agency. The fusion which has taken place in αλλήλων 
 is only a further extension of the juxtaposition. In Plato, Resp. iv. 
 p. 444, we have V7t αλλήλων twice by the side of άλλο υπ άλλον. It 
 is precisely the same case with phrases like tcqo δ tov ενόηΰεν and jCQog 
 άλλότ άλλον. This expression of reciprocity, by creating an idea of 
 contiguity or contact, is sometimes extended in Greek to an expres- 
 sion of identity. This appears from the use of εαυτούξ, αϋτονς, for 
 αλλήλους (see Hesychius and his commentators sub v. εαυτούς). In the 
 passage of Sophocles quoted above, we have αυτοκτονονντε ΐοτάλληλο- 
 %τονονντε, and farther on in the same play (145), we find %α\^ αυτοΐν 
 δυκρατεΐς λόγχας ΰτήΰαντε for κατ αλλήλων. In Plato (Parmenides, 
 p. 133 ε) we see the grounds of this usage more clearly: άλΧ ου τα εν 
 ήμΐν TtQog εκείνα την δύναμυν εχεο ουδέ εκείνα τΐρος ημάς, αλί , ο λέγω, 
 αντα αυτών καΐ προς αντα έκεΐνά τέ εΰτι,καΐ τα παρ ήμΐν ώΰαύτως 
 προς εαυτά^ — that is, αλλήλων καΐ προς άλληλα. We do not very well 
 understand what the author of the ΰυναγωγη λέξεων χρηΰίμων (Bekkeri 
 Anecdot. p. 378) means by asserting that the converse is the case: άλλί]- 
 λων αντί του εαυτών, οϋτως Ευριπίδης καΐ Θουκυδίδης- κ αϊ πού 
 τίνες αλλήλων έγεύΰαντο, άντϊ του εαυτών. The passage of 
 Thucydides here alluded to is as follows (ii. 70) : δ τε ΰΐτος επελελοίπει, 
 καϊ άλλα τε πολλά επεγεγένητο αυτό%υ ήδη βρώΰεως πέρι αναγκαίας, 
 %αί ηνες καϊ αλλήλων εγέγευντο. It is here simply stated that the 
 Potidseans were reduced to the necessity of feeding on human flesh: 
 of course αλλήλων is not used in its ordinary signification, for there 
 could certainly be no reciprocity in such an action as that referred to 
 in the text; but still less can any reflexive meaning be intended. 
 Thucydides perhaps considers the Potidseans as one body, and intends, 
 by the use of αλλήλων, to indicate that they fed upon the corpses of 
 their fellow-citizens : this is the only way in which we can comprehend 
 the interpretation given in the ΰυναγωγή. There are other passages 
 in which αλλήλων cannot be interpreted with any reference to reci- 
 
CHAP. III.] THE PEEPOSITIONS. 325 
 
 procity. In Odyss. xii. 102, αλλήλων must be equivalent to Ιτεροι» 
 if the present punctuation is retained : 
 
 τον δ' stBQOv ΰτωτίελον χ%'αμαλώτερον ο'φεί, Όδνΰΰεν, 
 τίλήΰίον άλλτιλων και κεν δίοϊύτενόείας. 
 But it is better, perhaps, to understand it as if it were written oi 
 ΰκότίελοί πληβίόν εΐΰιν αλλήλων, putting a full stop at the end of the 
 first line. In such words as άλληλονχος, αλληλουχία, the idea of reci- 
 procity is merged in that of contiguity or union. 
 
 Besides ετίάλληλοζ we have ετίαλλόκαρποξ and ετίαλλόκαυλοξ in 
 Theophrastus {Hist. Plaut, ni. 18), with the same meaning of reci- 
 procity. The use of ετίαλλάΰβείν and its derivatives, in the sense of 
 alternation or interchange, is very common. Thus in Homer [Iliad 
 XIII. 359) we have: 
 
 τώ ό' ερίδθ£ κρατερής και ομοιΐου ΐίολεμοιο 
 τίεΐραρ ετίαλλάξαντες, ε% άμφοτέροιΰί τάννβόαν, 
 ί. e, "alternating the rope of war, pulling it now to one side, now to 
 the other, fighting with various success." The metaphor is taken 
 from a game, in which two people tried their strength by pulling at a 
 rope, which is also alluded to in Iliad xi. 336 : 
 
 εν^'α ΰψί κατά ϊΰα μάχην ετάννβΰε Κρονίων. 
 The Homeric ετΐοίχεΰ^αι also expresses reciprocity, especially when it 
 means "to walk backwards and forwards in weaving:" comp. ίΰτον 
 ΒΛοιχομενην (Iliad i. 31) with Pindar's ίΰτών τΐαλυμβάμονς οδονς 
 (Pyth. IX. 18). 
 
 The word ετίαλλαγή is used by Herodotus (i. 74) to express an inter- 
 change of matrimonial relations between two families: καΐγάμων ετίαλ- 
 λαγηνετίοίηϋαν ^Αλυάττεα γαρ εγνωΰαν δονναίτηνΟ'νγάτερα'Αρνή- 
 VLV Άΰτνάγεϊ τω ΚναξάρεωτΐαοδΙ. On the similar word επάλλαξί.^, see 
 Wyttenbach on Plutarch (Moral, i. 2, p. 885). In this sense ετίιγαμία is 
 used (Herodot. ii. 147); it also denotes &, jus mutui connuhiihei-ween two 
 states or parties in a state (Wolf ad Demost. Leptin. p. 282). The word 
 ετίαλλάττειν is used to express an interchange or interlacing in a ma- 
 terial sense in the following passages: τα μεν ε(5τι καρχαρόδοντα αυτών 
 — τα δε άνετίάλλακτα — καρχαρόδοντα γάρ εΰτιν ούα ετιαλλάτ- 
 τ £t τον^ οδόντας τους όζεις (Aristot. Hist, Anim.i. c.i, p. 501 Bekker); 
 άρχέων των νεύρων η ετίαλλαγή, "the interlacing of the ends of the 
 muscles" (Aretseus, p. 34 β Boerhaave) ; and a line or two lower down in 
 the same page, άλλήλοι<5ι ετίαλλαξάμενα εις χιαΰμον Οχήματος, "inter- 
 lacing with one another like the letter χ." In a metaphorical sense 
 ετίαλλάττειν is applied to express verbal ambiguities ; thus, Xenophon, 
 Mem. III. 8, § 1 : μι} τίη b λόγος επαλλαχ^^, "lest your words be per- 
 verted." Aristot. Polit. i. 6 (p. 1255, 1. 13 Bekker): αϊτιον δε ταύτης 
 
326 THE PREPOSITIONS. [book ii. 
 
 άμψίΰβητήΰεως καν ο tiolbl rovg λόγον£ Ιπαλλάτχπν^ oti...B7ta δια- 
 ΰτάντων γε χωρίς τούτων των λόγων, κ.τ.λ. "that wliicli makes the 
 aro-uments run into one another and interchange. ..for if they were kept 
 distinct,... &c." In this sense ετΐαμφοτερίζευν is very frequently found : 
 ετίαμφοτερίϋαί. εΙς αμφιβολίας άγαγειν τον λόγον (Timasus, Lex.Pla- 
 ton. S.V.): see also Suidas and Hesychius. Plato, Euthydemus, p.500 d: 
 αδελφός ΰον εξ,ημφοτερικε τον λόγον. — Besp. v. p. 479 Β: τοις εν 
 ταΐς εΰτιάΰεόιν επαμφοτερίζονΰιν εοικε — και γαρ ταντα επαμφοτε- 
 ρίζειν. — Scholiast. adAristoph. Pac. 849 : επαμφοτεριζόντως λέλεκται 
 καΐ τίρος το τΐράγμα και προς το αΰελγες. Id. adPlutum, 635 : παίζων 
 γαρ ετΐαμφοτεριξονΰας λέξεις ε^ηκεν. The word also occurs in the 
 sense of "to fluctuate," "to waver between two parties or opinions:" 
 επαμφοτερίζειν, το μη παγίως εν τι βονλενεΰ^αι καΐ πράττειν, άλλα 
 και τόδε και τόδε διανοεΐύΟ'αι (Bekkeri Anecd. ρ.ΑΙ), as we see in 
 Tlmcyd. viii. 85 : φχ^είροντα των Πελοποννηΰίων τα πράγματα μετά 
 ^Αλκιβιάδου καΐ επαμφοτερίζοντα. Plat. Phcedrus, p. 257 β: ίνα κάί 
 δ εραβτης οδε αντον μηκετυ επαμφοτερίζτβ. If it be objected to the 
 connexion between επί and άμφί that the compound επαμφοτερίξω 
 could not in that case be formed without tautology, we need only 
 remark that επί and άμφί had become in time difi'erent, though cog- 
 nate, prepositions, and their use in the same compound is not more 
 remarkable than that of ob and amb- in the verb obambulo. 
 
 The same idea of reciprocity is conveyed by επαμείβειν and επαμον- 
 βαδίς. It may also be inferred that επίχειρα in the sense of reward 
 or punishment contains the same notion; επίχειρον. άνταπόδούις, 
 Hesychius; τάπίχειρα. τον μιΰ^^όν, τάς άμοιβάς, Suidas. Hence 
 Mschyl. Prometh, 327 : τοιαύτα μέντοι της άγαν νΊΡηγόρον γλώόβης, 
 Προμη^εν, τάπίχειρα γίγνεταΐ• The 28th line of the same play; 
 τοιαντ άπηνρω τον φιλανΟ'ρώπον τρόπου is explained by the Scho- 
 liast: τα της φιλανθρωπίας £π;^;^ ειρ α τοιαύτα ΰοιεγενετο, probably, 
 as Hemste^huis suggests (ad Lucian.i.-p.lOG, p. 370 Lehm.), with re- 
 ference to line 327. Plato,i?esp.x. p. 608 c: τά γε μεγιΰτα επίχειρα 
 της άρετης και...ά%^λα. In Sophocles, Antig. 814, ούτε ξιφέων επί- 
 χειρα λαχονΰα, punishment is implied: but in the Greek idea of punish- 
 ment, that of an equivalent or quid pro quo was always included. 
 The word επίχειρα is of course intimately connected in meaning and, 
 ultimately, in origin with επιχρηόθ'αι, "to have an habitual interchange 
 of kindnesses with any one." Thus, Herod, m. 99: ην δε γυνή κάμ^, 
 ώΰαύτως at επιχρεώμεναι μάλιότα γυναίκες ταύτα τοιύι άνδράΰι 
 ποιεύΰι. Thucyd.i. 41 : άξίωΰιν χάριτος τοιάνδε ην ουκ εχΌ^ροΙ οντες 
 ώΰτε βλάπτειν ονδ' αν φίλο/, ωύτ επιχρηύθαι άντιδοθηναι ημΐν εν 
 τω πάροντι φάμεν χρηναι. Plato, Legg.xu. p. 953 A: αναγκαία μεν ώς 
 δλίγιΰτα δε επιχρωμενους. The same meaning of reciprocity is con- 
 
CHAP. III. THE PHEPOSITIONS. 327 
 
 yeyed by Ιπιμαχία^ "an alliance for mutual defence," as distinguished 
 from ξνμμαχία. Thus Thucyd. i. 44 : ξνμμαχίαν μεν μη τίοίηβάΰ^'αί, 
 SoTB'Covg αντονςεχΟ'ρονςκαΙφίλονζνομίξείν...87ΐίμαχίαν δε ετίοίή- 
 ϋαντο,ττ] αλλήλων βοη&εΐν.^ΛΒ: άρκεΐν την ετΐιμαχίαν, άλλήλοις 
 βοη^'είν, ^ννεΛίΰτρατενείν δε μηδενί. Ammon.^;£9l διαφ. λε|.: Σνμ- 
 μαχεΐνκαϊεταμαχεΐν διαφέρει' ΰνμμαχεΐν μεν γαρ λεγονύυντο 
 ϋνν εαντοΐς, φηϋΐ δίδυμος, εϊτ αντοί ετίίοιεν πολέμιοι^ εϊ^ έτεροι 
 Ιταύτρατενοιεν. εταμαχεΐν [δε] οτετονξετίιόνταξ αμύνονται μόνον 
 (according to Valckenaer's emendations). It is probable that the diffi- 
 culty which has been occasioned by the description of the breadth of the 
 long wall in Thucydides, i, 93, is entirely due to a misconception of the 
 meaning ultro citroque agere, which is conveyed by ετΐάγω. The words 
 are : δύο γαρ αμαξαι εναντίαι αλλήλαΐζ τονύ λί^Όνς ετίηγον, "two 
 wains passing one another on the top of the wall brought the stones 
 in different directions," i. e. like the trucks used in making the viaduct 
 of a railway. Words formed with the preposition ετά are also used of 
 actions which take place on the borders of two countries, or on deba- 
 table land: for then a reciprocity, a motion in both directions, a Mn 
 und her, as the Germans say, is implied. Thus, εττεργαύία is "a com- 
 munion of husbandry between two neighbouring states," "a mutual 
 right of ploughing in one another's lands:" and ετίινομία is the same 
 with regard to pasture. Xenoph. Cyrop.iii.2, §23: ετΐιγαμίας δη είναι 
 και ετΐεργαΰίας καΐ ετίΐνομίαξ και ετΐΐμαχίαν δε κοινήν, εϊτις 
 αδικοίη οποτέρονς. Flstto, Leg g.YiJi. p. 843 c: των τ ε άλλων τιερι και 
 δήκαΐ ετίεργαΰίαξ όνμτίάύηζ.,.ος δ^ αν ετΐεργοίξηται τα τον 
 γείτονοζ υπερβαίνων τους ορονς. Aristot. ΡοΙϋ.ΎΛ, 5 : των ευπόρων 
 τα κτήνη άποΰφάξας λαβών πάρα τονπόταμον επινεμοντας. The 
 verb επινεμειν is also used when a common trespass on the part of a 
 neighbour is spoken of, as in Demosth. in CalUclem, p. 1274: καΐτών 
 γειτόνων επινεμόντων αμα καΐ βαδιζόντων δια του χωρίου, την 
 αίμαβίαν περιωκοδόμηΰε ταύτην. From the fact that the debatable 
 land between two countries was generally left \πήΐΆ\&ά,επεργαύία is also 
 applied to the cultivation of such land, or generally to the appropriation 
 of sacred or public property. Aristot. Bhet. i. 13 (p. 1374.ABekk.). 
 επεργάΰαΰ^αι μεν αλλ' ου δημόύια. Diodor. Sicul. χνι. 23: οι δε 
 Φωκεις έπεργαΰάμενοι Λολλήν της ίερας χώρας. From not ob- 
 serving this use of επί in composition, every one, so far as we know, has 
 misinterpreted an interesting passage in ^schylus (Agamemnon^ 444) : 
 
 γυναικός αίχμα πρέπει 
 
 προ του φανέντος χάριν ξυναινέΰαι. ,, 
 
 πιθανός άγαν 6 ^ήλυς ορός έπινέμεται 
 
 ταχύπορος. άλλα ταχύμορον 
 
 γυναικογήρυτον ολλνται κλεός. 
 
328 THE PEEPOSITIONS. [bOOK II. 
 
 Without dwelling on the absurdities of the commentators, it will 
 be sufficient, after what we have already said on the meaning of 
 ΙηινεμΗν, to translate the passage. The Chorus says : "It is in accord- 
 ance with the disposition of a woman" (so Choeph. 630 : yvvaixdav 
 ατολμον αίχμάν. αΙχμή is connected with αΐΰΰω root άϊκ-, as δραχμή 
 is with δράΰΰω root δρακ-) "to express her gratitude before the good 
 luck really appears. From excessive credulity, the boundaries of a 
 Avoman's mind are easily encroached upon. But a good report set on 
 foot by a woman perishes by a speedy fate." (For the sentiment we 
 may compare Tacitus, Annal. xiv. 4: facili feminarum credulitate ad 
 gaudia). Here, οροζ is used by a strong, but very intelligible meta- 
 phor, just as in V. 1077 of the same play {τύό^εν ορονςεχεος ^εϋΛεύίας 
 δδον κακορρήμονας). Similarly Menander says : τονςτης γαμέτης ο ροι;^ 
 ντιερ βαίνεις J γνναι {'Ιέρεια, 80) ; cf. Plat. Legg. 843 c, quoted above. 
 Another word of some difficulty, in which we believe ετίί is found 
 with this sense, is επιραβδοφορείν-, which signifies to gallop, as applied 
 to a horse. This meaning is generally supposed to be derived from 
 the ράβδος, or riding-whip, with which the rider struck his steed in 
 order to quicken its motion. Passow compares it with ετΐίΰείειν, which 
 is used in much the same sense : but we are convinced that this last 
 word is derived from the act of shaking the rein in order to urge on 
 the horse, a practice to which allusion is frequently made in the Greek 
 writers. Thus Soph. Antig. 109 : 
 
 φυγάδα πρόδρομον οξ,ντέρω. 
 κινήύαόα χαλινώ, 
 
 which the editors have generally misunderstood. Euripides, Iphig. 
 Taur. 909 : ω<5%^ αΐματηρα ΰτόμΐ επεμβαλεΐν εμοί. The Furies put 
 their bits into the mouth of Orestes, and, by shaking the rein till the 
 snafile was covered with blood, urged him to furious flight: seeAlcest. 
 495 : χαλινον εμβαλείν γνάθους. The passage in Xenophon in which 
 the word επυραβδοφορεΐν occurs, is as follows {de Be Equestri, vii. 1 1) : 
 μετά δε ταντα τον αυτοφυή δίατροχάξων δοαχαλφη τ αν άλυτΐότατα 
 τοΰώμακαΐεΐςτο ετίιραβδοφορεΐνΎ^διΰτ αν άφίκνοΐτο,εττείδήτΰερ 
 και άπο των άριΰτερών αρχεόΟ'αί ευδοκιμώτερον, ώδ' αν μάλιβτα άπο 
 τούτων αρχοιτο εΐ δυατροχάζοντος μεν^ οτίότε άναβαίνοι τω δεξίο^, 
 τότεΰημαίνοί τω ΐτίτίω το ετΐίραβδοφορεΐν. το γαρ άρίΰτερον μέλλων 
 αϊρευνεκ τούτου αν αρχοιτο. Now it is quite clear from this passage that 
 εταραβδοφορειν applies to an action, not of the rider, but of the horse, 
 and this action is the gallop, for the whole passage is about the change 
 from the trot to the gallop. The Greek work for "to trot" is, as we see 
 here, δίατροχάζειν-, "to make two wheels," for, as every one knows, 
 when a horse trots, he makes semicii'cles with his legs first on one 
 
CHAP. III.] THE PEEPOSITIONS. 329 
 
 side of the body, and then on the other: so that the hind and fore 
 feet on the same side occasionally touch. Hence Aristotle mentions 
 it as a fault in sculpture if the artist represented tov XitTtov αμφω τα 
 δεξία ττροβεβληκότα (Poet. χχιν. 4). To this Virgil also alludes in 
 the Georgica, iii. 192, where he speaks of breaking in the horse: 
 At, tribus exactis, ubi quarta access erit sestas, 
 Carpere mox gyrum incipiat, gradibusque sonare 
 Compositis, sinuetque alterna volumina crurum. 
 
 Which Voss, with his usual accuracy, translates "wncu erhehe die wech- 
 selnde Krilmme der Schenkel.^^ That Virgil is here talking of the trot 
 is farther obvious from his allusion to the gallop, which immediately 
 follows : 
 
 turn cursibus auras 
 
 Tum vocet, ac per aperta volans, ceu liber habenis, 
 
 -^quora, vix summa vestigia ponat arena. 
 
 It appears then that the word which expressed the action of 
 trotting was derived from the appearance of the horse's legs in trot- 
 ting: accordingly, we should expect that the same would be the case 
 with the word expressing the gallop. Now the primary idea in 
 ράβδος is "beating," "striking," "an instrument to strike with" — ράύ- 
 6SLV, ραπίξείν ; and it appears to have been one of the chief functions 
 of the ραβδούχος or ραβδοφόρος, to inflict the punishment of stripes ; 
 see Thucyd. v. 50: V7tb των ραβδούχων τΐληγας ελαβεν; and when a 
 horse gallops or canters he strikes the ground alternately with his fore 
 and hind feet. This therefore, is expressed by ετίοραβδοφορεΐν. 
 
 175 The use of ετΐί to signify combination or coexistence may 
 also be considered as a trace of this preposition's original identity 
 with άμφί. Thucyd. ii. 101 : ντίοβχόμενος άδελφην δώβειν καΐ χρή- 
 ματα Bit αντγ. Soph. Antig. 555: ον% lit άρρήτοις γε τοις 
 εμοΐς λόγοίζ. Ibid. 759: επί ιΐ^όγοιβί δεννάξείν. Eurip. Ion, 235: 
 ετί άόφάκτοίς μήλονύν. In this way ετά is especially used when dishes 
 are spoken of as eaten together: thus Aristoph. Equit. 101: ετά τώ 
 φάγοίςηδίύτ αν; ετΐΐ βαλαντίω; Pax, 123: εζετ εν ώρακολλνραν 
 μεγάλην καΐ κόνδνλον ο^ον ετί αντω. Acharn. 835: παίευν εφ 
 αλίταν μάδδαν. Xen. Mem. ηι. 14, § 2: έΰ^ίευν επί τω ΰίτω δφον, 
 Cyrop. IV. 2, § 27: litl τώ οίτω τΐίνείν, and even in metaphorical 
 expressions, as when Pindar says {Pyth. iv. 187 Bockh): 
 
 ημίΟ'εοίύίν TtodOV Λρόβδαυεν'Ήρα 
 ναός ^Αργονς, μη τίνα λειτίόμενον 
 
 τ αν άκίνδννον τίαρα ματρί μενειν αΙώνα Λε66οντ, αλλ ετάκαΐΟ^ανάτο 
 φάρμακον κάλλιύτον εάς άρετάς αλίξιν ευρεό^αυ ΰνν άλλους. 
 
330 THE PREPOSITIONS. [bOOK I. 
 
 i. e. "Juno kindled in the minds of the heroes a longing for the ship 
 Argo, so that no one might brook to be left behind and remain by 
 his mother's side, leading the sodden, insipid life which is free from 
 danger, but that each might strive to obtain in company with his 
 mates a seasoning or relish even for death itself, in his own glory and 
 renown." (On this sense of agstrjj see Pindar, Olymp. vii. 163; 
 Thucyd. 1.33; Sophocl. Philoct. 1406; Bekker Anecd. p. 443, 33, &c.). 
 It is quite a mistake of Matthise {Gr. Gr. § 586) to suppose that Ιτίί 
 has this force in sentences like og litl ^υγατρί άμητορυ — ετνΐ ταντ^ 
 εγημε αλλην γυναίκα, ηδε ετίεΰελΟΌνΰα εδικαίεν είναυκαΐτωεργφ 
 μψρνιη ty Φρονίμτ^} (Herod* ιν. 154). Here ετίί has the sense of 
 coming into a family as a stranger, in which sense Schiller, in The Song 
 oftheBellf most naturally calls a step-mother "the stranger:" 
 
 An verwaister Statte schalten 
 Wird die Fremde, liebeleer. 
 
 That such is the meaning oi ετίί when applied to a step-mother is 
 proved by the η επεΰελ^ονΰα in Herodotus, and by the following 
 passage of Euripides (Alcestis, 305): 
 
 Tcal μη ' Λίγήμ-^ς τοΐΰδε μψρνίαν τέκνοις — 
 εχ^'ρα γαρ η' τΐίονΰα μητρνια τέκνοις 
 τοις τίρόό^'ε. 
 
 176 We now proced to the considerations of those prepositions 
 of which the leading idea is motion from or out of some given place. 
 The shortest and simplest of these is εκ or J|, which is written e or 
 ex in Latin. Various conjectures have been made with regard to the 
 origin of this little word. Pott supposes (Etym. Forsch. ii. p. 183) 
 that it is connected with the Sanscrit vahis (extra); that the -his is 
 represented by the Greek -|, and that a digamma has fallen out in 
 the Greek word; he recognises the same root in the Sanscrit ava (off, 
 from), and considers ονκ to be the same word with avak (deorsum). 
 Hartung (Partikeln, n. 81) looks upon £κ as a subsidiary form of ονκ. 
 It is true that ονκ and εκ are the only words in the Greek language 
 which ever end in κ, but it must be recollected that one of the words 
 is written with a κ just where the other throws off the κ, and as this 
 κ is the only letter they have in common, it would be rather rash to 
 assume their identity on such a ground as that. We reserve the con- 
 sideration of ov, ονκ for the next chapter. It is clear to us that the 
 L•, εξ (έκς) bear the same relation to one another that subsists between 
 Iv and εΙς (Ivg), and that εξ is the original and proper form of the 
 word. It is perfectly analogous to αψ, the old Homeric form of άτΐό, 
 and there are words of great antiquity into which εξ enters. It has 
 
CHAP. III.] THE PBEPOSITIONS. 331 
 
 been the lot of this preposition more than any other to suffer muti- 
 lation when used in composition. Thus we have κ -QOvvog and κ-ρήνη 
 from εκ-ρέειν, and, what is more to our purpose, ^svog from εξ, and the 
 locative εζ = ενς = εν(ο)ς. Comp. etranger, extraneus^ straniero. We 
 have no doubt that εξ is, as we have mentioned before, a compound 
 of the demonstrative stem a or e, in this case a mutilation of na, with 
 the second pronominal element under the form -κΐζ, so that, according 
 to the principle of composition before explained, it expresses a re- 
 moval from the there through that which is near to the here, and 
 therefore naturally expresses "out of." We have still more mutilated 
 forms of it in the Gothic us, Slavonic iz, Lithuanian isz, old Prus- 
 sian and Lettish is (Grimm, iii. p. 253); for we have already seen 
 that the Lithuanian sz at any rate is a representative of the Greek 
 guttural (above, p. 201). The adverb vahis, with which Pott connects 
 ε-κ-ζ, should rather be compared with α'ψ, ahs, and the Sclavonic bez. 
 For our purpose this preposition presents little that is deserving 
 of notice. Its meaning is generally fixed with great accuracy, and it 
 seldom occasions any difficulty. The only word in which its usage 
 appears to be really anomalous is εκ^'νήΰκείν, which, though at first 
 sight it might be thought a synonym of emori, "to die the death," is 
 always used to signify "fainting," "seeming dead." Thus Homer, 
 Odyss. xviii. 100: a%aQ μνηύτηρεζ άγανοίχεΐρας άναύχόμενοί γελώ 
 εκΟ'ανον — which is precisely our idiom "died with laughter." Soph. 
 Trachin.^eS : εκ^'νήΰκων d' δ Ο'ηρ τοΰοντον είτίε, "the monster, as 
 he was fainting away, said just so much." Plato, Legg. xii. p. 959 A: 
 tag dl 7tρo&έ6εLg τίρώΐον μεν μη μακρότερον χρόνον ένδον γίγνε6%'αι 
 τον δηλούντος τόν τε εκτε^νεώτα καϊ τον όντως τε^νηκότα, "the 
 apparently dead," as opposed to "the really dead." Hippocr. Morb. 
 Vulg. V. 7: καϊ εξεΟ'ανε Λεντάκις, ωύτε τε^νάναι δοκεειν (comp. 
 Abresch. ad j^schyl. Agam.h^^). Plutarch, de sera Num. Vind. p. 563 d : 
 κατενεχ&είς εξ v^povg τινός, ου γενομένου τραύματος άλλα τίληγης 
 μόνον, εξεΟ'ανε καϊ τριταίος ηδη τίερίταςταψας αυτάς άνηνεγκε. 
 ^^lian, Hist. An. νιιι. 7 : τον άι^άμενον λειτίο^νμείν καϊ εκΟ^νήΰκειν 
 τα πρώτα, είτα μεντοικάί αΛο%νη6κειν. Hesychius: εξε^ανεν' ελευ- 
 τίο^υμηβεν. The origin of this meaning is the opposition of the outward 
 and seeming to the intrinsic and real. It is very well explained by Eusta- 
 thius in his note on the passage of Homer just quoted : δηλον δε ώς αφε- 
 λώς καϊ γλυκεως το άγαν γελάόαι γέλωτιεκΟ'ανεΐν λέγεται, ήγουν 
 εξω, καϊ, ώς είτΐαΐν, ετΐΐτΐολαίως ^ανεΐν, και ου κατά το κυρίως %α- 
 νεΐν. The verb εκφέρω also presents some peculiarities of usage ; for 
 while in a very common application it signifies to carry out a corpse to 
 burial, the effect of which, by the nature of the case, is a short conceal- 
 ment and seclusion (as is shown by the Latin words se-pelio, &c.; Gothic 
 
332 THE PREPOSITIONS. [bOOK II. 
 
 filhan, Gr. φνλ-άό68ίν)^ on the other hand, it means, with more continuous 
 reference to its origin, a publishing abroad or promulgation of that which 
 ought not to be concealed. The former meaning needs no illustration ; it 
 is as old as Homer (Il.xxiv. 786) ; it is recognised by the legal phraseo- 
 logy of Athens (εκφΕρειν άτίο^ανόντα^ Demost. p. 1071, 2); and is re- 
 presented also by the Latin effero. With refererce to this sense εκφορά 
 means "a funeral," and perhaps there is truth in the suggestion that 
 ^schyl. Eumenid. 910: των δνΰΰεβονντων δ' εκφορωτερα πελοις, 
 means "mayest thou rather carry off the impious as corpses" (see Muller, 
 Eumeniden, p. 178). The other meaning of εκφέρω is found in such 
 passages as Eurip, Hippol. 650 : vvv d' ai μεν ένδον δρώΰιν αί κακάΐ 
 κακά βονλενματ , εξω ά' εκφέρονόιτΐρόΰΛολοί. From this sense we pass 
 easily to that of commemoration or celebration, as in Soph. Track. 791 : 
 TLV εξήνεγκα£^ ώτέκνον^ λόγον, hence it means in the middle voice, "to 
 carry off /or oneself, to get the credit of, or to be celebrated on account 
 of some thing." So we have in Soph. Electr. 60 : τι γάρ με λντΐεΐ τον^'' 
 όταν λόγω &ανών εργοίΰί 6ω0^ώ, κά^ενέγκωμαι κλέος ; Trachin. 497 : 
 μέγα τι ύ^ένοξ α Κντίρις εκφέρεται νίκας αεί. Demoth. ρ. 178,7 : του 
 δοκεΐν εν λέγειν την δόξαν εκφέρονται. Xen. de venat. 1,15: δό^αν 
 εϋόεβείαςέ^ηνέγκατο. In the signification "to lead out of a crowd, to con- 
 duct from a confusion of surrounding objects to a definite end or goal," 
 εκφέρω is used both transitively and intransitively. Thus, of a road or 
 path, Plato, Phced. p. 66 β : κινδυνεύει τις ωΰτΐερ ατραπός έκφέρειν 
 ημάς μετά του λόγου εν τ]] ΰκέίΐ^ει: of a scent in hunting, Soph. 4/• 7 : 
 ευ δέ σ' εκφέρει κυνος Αακαίνης ως τις ευρινος βάΰις: and intransi- 
 tively of the fulfilment of oracles, Soph. (ScZ. Co/. 1424: οράςτατοϋδ^ ούν 
 ώΰ ες ορ^^όν εκφέρει μαντεύμα^ , ος ΰφων θάνατον εξ άμφοΐν ^'ροεΐ. 
 So we say inEnglish, "a road leads to such a place;" "this conduct leads 
 to certain ruin," &c. For the similar employment of διά, see above, 
 §. 150. Another use of εκφέρω is to express that something has been 
 cast on shore, from, or out of the sea, as Eurip. Hec. 701: πόντου viv 
 έξήνεγκε πόντιος κλύδων. We mention these significations, not because 
 they are at all unknown or recondite, but merely because their con- 
 nexion is inadequately recognised by commentators and lexicographers. 
 
 177 The prepositions από and παρά also express motion from a 
 place. The former corresponds to the Sanscrit apa; its original form 
 was ά-πό-ς, for άν-πός or να-πός, and as such it denotes motion from 
 a distant object to the subject. This etymology is in accordance with 
 the distinction always observed between από and έζ as denoting mo- 
 tion from the surface and interior respectively of a distant object; for 
 in ά-πό we go straight from the remote to the hei^e; while in ε-| we 
 pass tlirough intermediate proximity. Hence άπο-δύομαι means "I 
 
CHAP. III.] THE PEEPOSITIONS. 333 
 
 put off my ίμάτιον or outer garment ; " Ικ-δνομαί, "I put off my χυτών 
 or inner garment" (Lysias, c. Theomn. § 10); and άτΐεκδνομαί, "I strip 
 or divest myself altogether" (Paul, Col ii. 15). The preposition TtaQcc 
 is related to the Sanscrit para, and as a compound of the first -ργο- 
 noun pa=ma (compare ite-da, με-τά) with -ra denotes primarily motion 
 from the subject, and, by a secondary sense, conveys generally the 
 idea of motion connected with that of closeness, and may even signify 
 motion to the subject. For para we also find apara, which is only 
 the comparative of apa, that is, apa with the suffix -r«, which we have 
 before explained; there is also a superlative par ama (see Schlegel's In- 
 dische BibliotheJc. i. p. 362). In its ordinary use, para is an indefinite 
 pronoun, and is equivalent to άλλος ^ alius, the relative meaning of 
 which we have before discussed. When we compare ^ara^ par, itaqa^ 
 πέραν (Sanscrit param), τίείρείν, per, peren-die, τίέρνόί"^, with π-ρό-ζ, 
 τΐάρος (Sanscrit pwms), prce, pro (Sanscritpra) τΐρω-ΐ] (whence pru-ina), 
 on the one hand, and τίερί^ Sanscrit pari, on the other hand, we shall 
 find it impossible to believe that the Greek prepositions ττρο, Λαρά 
 (τΐαροα), and περί, and the Lation pro, prce, per, are not etymologically 
 connected. Nor is there much difficulty in reconciling their various 
 meanings. The essential part, the expression of the here, is the same 
 in all the words which we have compared above; the only variation 
 is in the affix, which is written ro-s, ra, ri. Let us examine the force 
 of these terminations in the Greek: (1) τΐ-ρό or τίά-ρο-ς, signifies 
 "that which is before the subject;" τί-ρό-ζ, τΐ-ρο-τί, denotes "motion 
 towards that which is before the subject" when joined with the accu- 
 sative; "mere direction" when joined with the genitive -j-; and "close- 
 ness" when joined with the dative; in this last collocation it also signi- 
 fies, by a very natural transition, "adding" or "superimposing:" (2) 
 τίαρά is found with the same three cases, and in its general use cor- 
 responds pretty nearly to τί-ρό-ζ, except in its use with the genitive, 
 when it invariably means that something is taken away from some 
 other thing ; it is true there is an appearance of the same force in 
 such phrases as τίρος Λατρός, μητρός, "on the father's, mother's side," 
 OL τΐρος αίματος, "blood-relations," αΐτε Ο^εμνΰτας προς ζίώς είρνα- 
 ται, "by commission from Jupiter" {Iliad i. 239), ονδ' ακλεης vlv 
 δό^α προς ανθρώπων νποδεξεταί, "proceeding from men" (Eurip. 
 Her ad. 625), and in the other examples cited by Matthise; but the 
 more general use of π-ρό-ς is unquestionably to denote motion, not 
 
 * On the connexion of πέραν and πέρνΰΐ compare "Ηπειρος — ή δια- 
 τζέραν χώρα with its epithet δίαπρνβιος (above, § 150). 
 
 f On the vague uses of προς and παρά with the genitive, see Gr. Gr. 
 Art. 486, obs. i. 
 
334 THE PREPOSITIONS. [bOOK II. 
 
 frorriy but ίο, a place before us: (3) its-Qi generally signifies, when joined 
 to the genitive, "relation;" when joined to the dative, "closeness," 
 "on," "about;" when joined to the accusative, it denotes "motion or 
 extension around," and answers to the question "whither?" "where is 
 it moving?" We see, then, that the general difference in meaning be- 
 tween •jt'QO', 7ΐα-ρά, TCe-Qi, is very slight; indeed in some cases their 
 meaning is so nearly the same, that one of the three might be substi- 
 tuted for either of the others without materially affecting the sense. 
 Thus, with the dative, in the sense of apud, we have TtQog με(5^ άγορα 
 (Sophocl. Track. 371, comp. ev μίς^ άγορα, Ibid. 423), τΐαρα τνραν- 
 vidi (Find. Pyth. ii. 159), τΐερί Σκαι^ύο TtvXyoL (Iliad xvm. 453); 
 with the accusative, in the sense of secundum, "with respect to," we 
 have τελεος τίρος άρετήν (Plat. Alcib. i. p. 120 e), τίονηροζ περί τι 
 (Plat. Besp. v. ad init.) ; and in the sense of propter^ "on account of," 
 we have Λροζ ών την OTpLV ταντην τον γάμον τοιτοντον εΰττενΰα 
 (Herod. 1. 158), εκοίΰτος ου it αρά την εαυτόν άμελείαν οϊεταυ βλά'φευν 
 (Thucyd. Ι. 141). We have another proof of their identity in the way 
 in which the Latin prepositions are used to express the meaning of 
 these three. Thus prw^ which bears more outward resemblance to 
 Λα-ραί, is used as a synonym for τΐερί in prce metu, περί τάρβει : per, 
 which externally resembles περί, is used as a synonym for παρά in 
 such phrases as πάρα τοΰτο, per hoc: also proe-ter, the comparative 
 of prce, in πα-ρά δό^αν^ prceter opinionem: prce is also used for περί 
 in such sentences as περί πάντων, prce omnibus; also prce and per 
 are sometimes synonyms; cf. non possum prce fletu (Cic. Att. xi. 7), 
 and ne que per cetatem potis erat (Ter. Eun. i. 2, 32): per is used in 
 Latin, where προς is in Greek, to express the person called upon 
 to witness an oath: and προς and per are used in the same way 
 with a neuter adjective in an adverbial sense: compare προς τάχος 
 for ταχέως (Plato, Legg. vn. p. 810 b) with per taciturn for tacite 
 (Virgil, JEneid ix. 30) : περέ and per agree in the intensive sense ; 
 compare περυκαλλής with perpulcer. 
 
 178 There is one use of the preposition περί which is not found 
 in any of the cognate prepositions, except in the Sanscrit pari. These 
 words frequently mean "round about," like the Greek άμφί, the Ger- 
 man wiw, and the Latin circum; and περί designates, by an association 
 which we have already explained, not only that which surrounds, but 
 also that which is surrounded. The former meaning is due to a con- 
 nexion of the ideas of closeness and removal in this word; the last 
 vowel seems to point to the fragment of a case denoting rest, and the 
 termination ra indicates motion. It may, therefore, be surmised that 
 the whole word denotes motion confined to a sphere of action not far 
 
CHAP. ΠΙ.] THE PREPOSITIONS. 335 
 
 removed from the subject. The other meaning has, perhaps, arisen 
 from the wider signification of ττα-ρα, which seems to denote simply 
 motion from the subject; and this has generated the ideas of "going 
 through, "piercing," "boring a hole," conveyed by τίερί and its de- 
 rivatives τΐείρω, τίερονή, τίορττή. That ττερί does bear this meaning 
 appears from a well-known line in Sophocles (Ajax, 890) : 
 
 Iv γάρ οί χ^ΌνΙ 
 τΐηκτον τόδ' εγχος περιτιετεξ κατηγορεί. 
 
 Lobeck approves of the interpretation of Eustathius (p. 644, 7). Σο- 
 φοκλης εγχος περιτίετες είτίεΐνετόλμηΰεν^ ωτΐεριπεπτωκεν Αϊαξ. This 
 interpretation has obviously been suggested by two other lines in the 
 same play — τίετίτώτα τ ώδε τίερί νεορράντω ξίφεί (8 1 5), and κείται κρν- 
 φαίω φαόγάνω τΐερίτΐτνχής (883). But these passages have nothing to 
 do with the proper interpretation of τΰεριτΰετεζ in the former one : the 
 construction of that line is obvious, and though it is true that τίερΙ 
 generally denotes that which is round any spot, and not a round hole, 
 it does not follow that it may not have borne the other signification 
 also. Lobeck quotes the following passages in illustration of the line 
 in Sophocles; MMim^ Hist. An. xv. c. 10: αγκίΰτρα τίερυτίαγεντατοΐζ 
 Ιχ%ν(5ί\ Liban. Decl. Tom.iv.p. 1081 : οδόντες τ^ δείρτ[ΐ%ερι%είρονται\ 
 Chrysostom, 0pp. Tom.iii. 85 A: εαντω το ξίφος τΐεριέπεορε. But the 
 age of the authors quoted renders their authority of little value: 
 indeed the passages prove no more than that these Avriters understood 
 the line in Sophocles as we do, and had probably met with other pas- 
 sages like it. The following analogies are more to the purpose. It will 
 not be denied that the ideas of "boring," "piercing," "roundness," 
 &c. are mixed up in the family of words which includes τόρ-Οζ, τόρ -vog, 
 τορ-ενω,τόρ-μος,τί-τρά-ω,τί-τραίνω, τρώω, το-τρώΰκω, τραύμα, 
 τρύω, τρντΐη, τρντΐάω, &c. Now these words are unquestionably con- 
 nected with one of the simplest words expressing beyond or mo- 
 tion: we mean the ending -tara, -τερος, discussed in the last chapter. 
 The analogies between this set of words and those we are discussing ex- 
 tends in many directions, both in Greek and Latin. In the first place, 
 we have τέρμα, τόρμος, terminus, trans, by the side of τίερας, τίεραν 
 (περάω), "the object of a journey," "the place gone to." The words 
 τέρμα, Λεραζ, also signify "an end," "a termination" in general, and 
 verbs signifying "to end" or "finish" are frequently used in Greek to 
 express a journey: thus αννΰειν τον "Αιδαν (Soph. Aj. 666), where 
 Lobeck quotes καταννΰας εξ'Ε.ες A. (Herod, vi. 140); τελεΐν ετάτο 
 τέρμα (Lucian, Trin. § 20); lAd'rjvag εκπεραν (Athen. ii. p. 47 c); 
 τέρμα βίοω καΐ ες βαλβίδα τΐερήΰας (Oppian, Cyn. ι. 512) ; οβτουν εΙς 
 την ντΐερώαν περαινον (Galen, de Usu Partium, ix. 3, 508). By the 
 
336 THE PBEPOSITIONS. [bOOK II. 
 
 side of 7t8Qag, τέρμα, we have the by-forms TteXag, τέλος. In Latin the 
 adjective teres is applied to signify roundness in general ; in se ipso 
 totus, teres, atque rotundus is said of a sphere in Horace [Sat. n. 7, 86). 
 Compare Ausonius, 7(i. 16: mundi instar habens, teres, atque rotundus. 
 It is also applied to a net, either because the ropes which composed it 
 were cylindrical, and drawn through holes, or because it was full of 
 holes, namely, the meshes. If the former is the true explanation, 
 which is more probable, we may compare teres with ττεΐραρ, the Ho- 
 meric word for a rope (Iliad xiii. 359). There are two Latin words 
 which involve the combination pa-ra, and which are interesting, not 
 only in reference to the combinations just mentioned , but also as 
 bringing us back to the liquid from of the first element — we allude to 
 pars (par-t-s) and paries (pari-et-sj. We cannot doubt the affinity of 
 the former to its Greek synonym με-ρος or μέ-λος, which we shall 
 discuss in a future chapter; division or partition, as we shall there 
 show, resolves itself into the idea of a line proceeding from the subject 
 and cutting another fixed line. The analogies, which we have men- 
 tioned above, are sufficient to show that the same or a similar idea 
 may be conveyed without any reference to the subject, by the mere 
 combination ta-ra, and it is easy to pass from ττε-ρά-ωίο τίείρω, ηόροζ^ 
 &c. The notion of piercing through and separating into parts very 
 soon suggests that of distribution and division; and from this again 
 we get the sense of discussing and declaring at length: we see this 
 especially in the verb ενδατεΐΰ^αί, which Hesychius translates by 
 ατίομερίζειν^ but which is used by Sophocles {(Ed. T. 205) to signify 
 celebration or commemoration by word of mouth, and by ^schylus 
 with the similar implication of repeated mention (Sept. c. Theh. 578, 
 and apud Platon. Bepuhl. ii. 383 E; cf. Soph. Tr. 791). The same 
 is the case with the Latin disputo, and we are disposed to seek a 
 similar explanation for the solitary form Ίίεπαρεΐν (V'ma.Pyth. ii. 57), 
 which Hesychius explains as signifying "to display or exhibit" {%ε- 
 παρείν ενδεί^αυ, ΰημηναί. πεπαρενΰίμον ενφραότον, ΰαφές), 
 and which we regard as an obsolete aorist of τΐείρω. We recognise the 
 sense of separation and division in paries, "that which goes, or is a 
 divider" (cf. se-paro, &c.) ; for paries is properly a party wall common 
 to two chambers, or standing between two houses : thus it is applied 
 to the divisions in a labyrinth, e. g, in Virgil, JEneid v. 588: fertur 
 Labyrinthus habuisse parietibus textum ccecis iter; and to the walls 
 between two houses as distinguished from the outer wall ; e. g. in 
 Tacitus, Annal. xv. 43 : nee communione parietum sed propriis quce- 
 que muris [cedificia] ambirentur, where the last Λvord refers to the 
 technical term ambitus, which was probably used thus in the seventh 
 of the Twelve Tables : inter vicinorum cedificia ambitus parietum ses- 
 
CHAP. III.] THE PREPOSITIONS. 337 
 
 tertius pes esto (Dirksen, Zwblf Tafel Fragm. p. 565). We must 
 not be led by an apparent similarity of sound to seek for the mean- 
 ing of paries in the Greek words Ttagodois, τίαραΰκήνίοί, wbere the 
 τζαρά stands in a sort of opposition to jteQi (Theatre of the Greeks, 
 6th edition, p. [152] sqq. notes), and where there is rather an im- 
 plication of lateral motion. The same is signified by the military 
 term τίαροΛτΐενω, which means "to move cavalry along a line of 
 battle," either for the purpose of changing its position from one wing 
 to the other (Arrian, Anab. ii. c. 9, § 1), for the purpose of extending 
 the flank (Id. Ibid. iii. c. 13, § 2), or for the purpose of masking the 
 movements of a line of infantry (Id. Ibid. v. c. 16, § 1: eg κνκλονξ 
 τίαρίΤίΛεύων avsTtave τους τΐεξονς). Doderlein, who has seen the con- 
 nexion between paries and pars, has not observed the relationship be- 
 tween the latter and μέρος, but has endeavoured to trace an affinity 
 between murus and μοίρα, which contains the same root as μέρος 
 {Syn. und Etym, v. p. 350). Now it is clear that this meaning does 
 not apply to murus or moerus, which denotes the outer wall of a city 
 or house; see Virg. yEn. ii. 234: dividimus muros et moenia pan- 
 dimus urbis, from which Niebuhr would infer a distinction of build- 
 ings within and walls around a city {H. B, ii. note 80), a distinction 
 which is not borne out by the practice of the best writers. We 
 believe that murus and mcenia come from the same root ά-μννω,ηιωηιο, 
 which we have discussed in another part of this book, and that the 
 only difference between the words is thsit murus is "a wall" generally, 
 but mcenia a city- wall or fortification in particular. 
 
 If we place this meaning of separation by the side of the other 
 significations of jt -ρό, τΐα-ρά, Λε-ρί, we see that the force of the San- 
 scrit pronoun pa-ra = alius is more or less conveyed by all of them, 
 and that they are all merely modifications of the expression of diversity 
 or beyond considered in immediate connexion with the subject; from 
 this ground-meaning all their uses may be explained, the separate 
 words being only different cases of the Sanscrit pronoun. This pro- 
 noun occurs directly in the Latin par, which is equivalent to alter 
 (Pott, Etym. Forsch. ii. p. 230), and we may recognise it in the 
 compounds perendie ("on another day") per-(^)grinus ("of another 
 country") (Bopp, Vergl. (rr^wwi.p. 541),and,perhaps, in the Greek ττερ- 
 φερεες. Herodotus says (iv.33): τονς^^Ττίερβορεονς πεμ^φαυ φερονΰας 
 τα ίρα δυο κόρας - αμα δε αντ^βΰ άΰφαλίης εΐνεκεν Λεμ^αι τονς 
 "Ττίερ^ορεονς των άΰτών άνδρας τίεντετίόμτίονς τούτους dt νννΠερ- 
 φερέες καλεονταυ. Some suppose this word to be another form of τΐερυ- 
 φερέες: Guyet (on Hesychius sub v.) suggests that it is a synonym of^ro- 
 ceres : and Niebuhr (Hist, of Borne, i. p. 8 5) connects it withperferre. " If," 
 
 Ζ 
 
338 THE PEEPOSITIONS. [bOOK II. 
 
 he says, "it be but allowed that the people called Hyperboreans rxiay 
 have been a Pelasgian tribe in Italy, the possibility will, perhaps, be 
 nearly turned into certainty when it appears that the title of the car- 
 riers was almost a Latin word." Now there is no objection to con- 
 sider TtSQ as standing for τίερί in this compound; see Bockh, Not. 
 Crit. ad Finn. 01. vi. 38, ad Fragm. p. 631; and compare πέρ-Ο"» 
 with per-do, which is related to per-eo as ven-do is to ven-eo. We 
 might, however, be contented to suppose that the word merely signi- 
 fies "the strange" or "foreign carriers." That Λ£ρφερε£5ΐ8 connected 
 with φέρω appears from the word φερονΰας in the passage of Hero- 
 dotus, and by the names Amallophori (Porphyr. de Abstin. n. 19), 
 ονλοφόροί (Servius ad Virg. JEn. xi. 858), also given to these per- 
 sonages. There is yet another Greek word in which we recognise this 
 root, namely Tteg-TCSQ-og (τίερΛερενομαι,, περτίερεία). This word, 
 which is evidently a reduplication, means a strange, out-of-the-way, 
 absurd person: compare the Latin perperam, perperitudo. 
 
 The intensive meaning, which we find in τίερί and τΐερ, may be 
 thought due to the idea of perfection and completeness suggested by 
 the circle or sphere: thus Plato, Timceus, p. 33 b: xal όχημα δε, 
 'εδωκεν (δ Ο'εος) αντω (τω κόΰμω) το πρετΐον καΐ ξυγγενες- τω γαρ 
 τα Ίίάντ εν αυτω Ζ,ώα περιεχευν μέλλοντν ζώω τΐρεπον αν εϊη βχημα 
 το τίεριειλΎίψοξ εν αυτω τίάντα οτΐόΰα ΰχηματα' δώ και ΰφαυροευδες 
 εκ μεΰου τίάντη ηροξ ταξ τελεντας ϊΰον ατΐεχον καΐ κvκλoτερεg αντο 
 Ιτορνενβατο τΐάντων τελεώτατον ομοίότατόν τε αντο εαντω οχημά- 
 των. But the same idea of completeness is found in πέρας, which 
 derives it from the other and perhaps earlier sense of περί, namely, 
 that of "going through," "piercing," "faring to the end;" and we 
 believe that this is also the origin of the intensive meaning of περί. 
 Compare the case of δίά (mentioned below) and the English use of 
 "throughly," "thorough-going," "through and through," &c. 
 
 179 It is easy to see that ν-πό, ν-πε-ρ, and their Sanscrit and 
 Latin equivalents u-pa^ u-pari, su-b, su-per, are related to one another 
 precisely in the same manner as the Sanscrit a-pa, a-pa-ra; that is 
 to say, the latter preposition in each case denotes a continuation of 
 the direction indicated by the former. If νπερ is actually formed 
 from νπό, the vowel o, which represents the loss of an original g, as 
 in από, Λρό, &c., has only suffered the usual attenuation in point of 
 weight. We have already endeavoured to show that ν-πό (su-b, u-pa), 
 is a compound of the second pronominal stem with the first, implying 
 a motion or continuation from that which is immediately near in the 
 direction of the subject. From this analysis it appears that ν-πό 
 
CHAP. III.] THE PREPOSITIONS. 339 
 
 differs from ά-πό only in supposing the point from which the motion 
 commences to be near to instead oi far from the subject, and, in fact, 
 the most prominent signification of upa is "near," or "by the side 
 of." It does indeed also denote "inferiority," but in a great many 
 cases it approaches nearer to the meaning of Ιτΐί as implying super- 
 position, or indeed position generally. We believe, then, that v-jto 
 primarily implies only a motion to the speaker from that which is 
 near to him, the idea of "under" not being in any way directly con- 
 veyed by it, except so far as the Greeks habitually spoke of things 
 near them as τα εμΛθδών=τα εν Jtoulv οντά, or τα εν τίοβΐ είλεύ- 
 μενα (Herod, ιι. 76). With the same vagueness of reference the Sanscrit 
 adhi means both "over" and "under." But when by the addition of 
 the suffix -ra this direction is continued beyond the subject, the ideas 
 of "upper" and "under" arise from the correlation. 'T-Jto, su-b, u-pa 
 are related to ν-πέρ, su-per, u-pari pretty nearly as positives to com- 
 paratives. A similar relation subsists between the Gothic w/, ufan 
 thus in Ulfilas, Mark iv. 32, ντώ την ϋκυαν αντοΰ, ^^ under the 
 shadow of it," is translated "w/ skadau is," and Matth. x. 24, ουκ 
 ϊΰτι μα&ψης vtcIq τον διδάΰκαλον, ονδε δούλος ϋπερ τον kvqlov 
 αντον, " the disciple is not above his master, nor the servant above his 
 lord," stands in the Gothic version "nist siponeis ufar laisarja: nih 
 skalks ufar fraujin seinamma." Now "over" and "under" are both 
 comparatives, the former being, in fact, identical with ufar. The simple 
 method of explaining this difference is, to suppose that υπό, sub, upa^ 
 uf originally signified, "up," "upon," like επί and ob; and then, 
 according to the proper use of the suffix ra, υπέρ, super, upari, ufar, 
 would mean "upper," as the comparatives of the other set of words. 
 The meaning of the words "over" and "under" is this, that the sub•? 
 ject consider himself as a point in a vertical line, every point in the 
 line, reckoning from his feet, being considered as "under," and every 
 point in the line, reckoning from his head, being considered as 
 "over," so that the subject is the positive, and those two words are 
 comparatives, not in relation to one another, but to him; thus it is, 
 that when the relations of "over," "under" are expressed by com- 
 parative forms, as in English, German, and Sanscrit, they contain 
 different roots: for "over," ilber, upari, have no etymological con- 
 nexion with "under," unter, antar. But "over" and "under" are 
 really opposed to one another; they are relative terms, and are ex- 
 pressed as such, when, as in Greek, Latin, Sanscrit, and Gothic, they 
 stand as positive and comparative degrees of the same word : so that, 
 although "under" is expressed in these three languages by a word 
 signifying "up" or "upon," it must be recollected that what is "up" 
 in regard to one thing, is "under" in regard to that which is "upper;" 
 
 Ζ 2 
 
340 THE PREPOSITIONS. [bOOK II. 
 
 just as τό TLVL ΰννελΟ'ον καί tcolovv, αλλω αν 7tQO0jtBuov, τΐάΰχον 
 άνεφάνη (Plato, Thecefet. p. 157 a)*. 
 
 1 80 The preposition dia is merely a form of the second numeral : 
 compare δϋάκόΰίΟί for δυα-εκόΰιοί, with τριακόΰίοι, &c. Accordingly, 
 it generally conveys the idea of "penetrating," "dividing into two 
 parts." That the same root should express both the division and 
 union of two objects is an instance of the association by contrast to 
 which we have before drawn attention (§ 53). ζ/εω=ά-/εα5 means to 
 "couple two things," and the converse idea of separation is mani- 
 festly presented to us in the word δνω^ "to penetrate," which, of 
 course, belongs to the same origin, for in this word the vowel is 
 always wavering between ν and t, as we might expect, according as 
 the labial is omitted or vocalized. 
 
 In ^ί-δvμog, a reduplicated form, we have both ways of writing 
 the root. We have remarked before on the Sanscrit dvimdtri, which 
 stands between bimater and δυμητωρ. It is from this connexion with 
 the second numeral that δίζω is a synonym with άμφίΰβΊ^τεω. The 
 separative force of δίά is strongly expressed in such compounds as 
 διαφέρω, διαΰχίξω, δυάνδίχα, &c. In Sophocles, Ajax, 511, it is clear 
 that ΰον δυοίΰεταί μόνος νπ ορφανυΰτών μη φίλων cannot be re- 
 ferred to such phrases as διαφερενν βίον αιώνα. We have elsewhere 
 shown (Proceed. ofPhilol. Soc. i. p. 164) that Herodotus, who was a 
 constant imitator of Sophocles, has indicated the true meaning of the 
 passive διαφερε6%αυ here, namely, that it refers to the embezzlement 
 and spoliation of a minor's goods by his dishonest guardians: cf. Herod, 
 ui. 53 : βονλεαο την τε τυραννίδα ες άλλους πεβεειν και τον οίκον 
 τον π;ίϊΤΓρο2 §ιαφορη%εντα μάλλον η αντός 6φε άτίελ^'ών εχειν; with 
 1. 88 : ττόλίντετην ΰην δίαρτϋάξείκαίχρήματατα ΰα δι,αφορέει,; andLo- 
 beckhasquotedDioChr.Or.xLi. 506 c: vit ορφανιΰτών διαΰτΐαΰΟ'ήίίε- 
 ταί-, which is obviously an imitation of the passage. The preposition di« is 
 sometimes used emphatically, with this separative force, in the verb δία- 
 φαίνω. Thus we haveinPind.PyiA. iii. 44, καιομενα d' αντω διεφανε 
 
 * The Professor of Comparative Philology, who not only renders his own 
 science ridiculous by the extravagantabsurdity of his etymologies, but also takes 
 every opportunity of treating its first principles with contempt, has endeavoured 
 to evade these difficulties in the usage of the prepositions by laying down the doc- 
 trine, "that prepositions of different origin and power frequently assume an 
 identity of form" {Trans. ofPhilol. Soc. 1854, p. 69). To suppose that there was 
 one unter=oxxr under and that another unter was akin to avcc, that one vn- was 
 i=av- privative, and another =(χι;ά, and so on, is to run counter to all the prin- 
 ciples on which scientific philology depends. We have protested against this 
 monstrous paradox in the same volume {Trans, of Philol. Soc. 1854, p. 289). 
 
CHAP. III.] THE PREPOSITIONS. 341 
 
 Λνρά, of a cloven flame; and in Thucyd. iv. 108, ε'φενβμενοίς της 
 Ά%-ην(χίων δυνάμεως επϊ τοΰοντον οΰη νΰτερον δίεφάνη, all the 
 
 emphasis falls on the preposition; the meaning is that the subject-states 
 " were mistaken in regard to the power of the Athenians, by as much 
 as that power afterwards appeared different from their notion of it" 
 (on the force of the οΰη after ετά τοΰοντον, see Greek Grammar, Art. 
 555). In Latin the numeral adverb dig is written bis, the labial sound 
 being alone retained, but, as a preposition in composition, the Greek form, 
 dig or da, is retained, the final letter when not preserved being repre- 
 sented either by a lengthening of the i, or by an assimilation with the 
 first letter of the word with which it is compounded. Thus the Greek 
 form is preserved in dis-cedo, dis-rumpo, &c. ; it is softened into r in 
 dir-imo {dis-emo)\ it is assimilated in dif-fero (δια-φερω), dif-findo 
 (δία-ΰχίζω); and represented by the long i in di-lanio (δία-ύτίάω), di- 
 midius (δυα-μεΰος, ημιΰν), &c. The form in s is preserved in Greek, 
 but only in one sense, that of separation, weakness, and, consequently, 
 in general opposition to that which is good. In this sense the con- 
 necting vowel is V in Greek, though in Latin the form is generally 
 not altered in consequence of this signification. In Gothic tus is used 
 in this sense (Grimm, ii. p. 768); this is because tvis is the form of 
 dig in that language. In Galat. v. 20, διχούταΰίαι is translated tvis- 
 stasseis. The extensive use of the preposition or second numeral, call 
 it which you will, in this sense cannot be better shown than by com- 
 paring δνΰ-μενής with the Sanscrit dur-manas (comp. dir-imo) and 
 Persian dush-men, which have the same signification. In Latin we 
 have dif-ficilis (δυΰ-χερης), dif-famo (δνΰ-φημέώ), &c. ; and the same 
 prefix appears in French words like des-astre, "evil-stars," des-ordre, 
 "dis-order," des-accord, "violation of harmony," &c. There is one 
 case in which des has sprung directly from the Greek δνβ-; for Des- 
 demona is merely the Italian form of Αν<5-δαίμονα, the accusative of 
 the name given to this unfortunate heroine in the original Cyprian 
 story from which Cynthio borrowed his novel. The accusative form 
 is of course the usual one in Italian. So Shakspere's Cressida is 
 merely Homer's Chryseis, represented, however, as the daughter of 
 Calchas, and not of an Asiatic priest of Apollo. Sometimes, as in bis^ 
 the labial only is represented, as ve-cors, ve-sanus, unless we prefer to 
 consider these words as representing the element va, in a-^m, "out 
 of," which seems at any rate to be contained in ve-stibulum (see 
 Becker, Gallus, Eng. Tr. p. 189). Under the form ξα we have δοά in 
 an intensive sense, for the same reason that τίερί, per, bears that 
 meaning also. Thus ζάχρνΰος = δίάχρνΰος. It is a strong proof of 
 the connexion of this ξα with δίά, δις, dvg, that ξα-μενής is generally 
 used as a direct synonym for δνό-μενής. Compare also δίαυ-τα with 
 
342 THE PREPOSITIONS. [book II. 
 
 vi'ta, on the one hand, and ζωή on the other. So too we have ζενγος^ 
 jugum (diugum), Ζευς (gen. z/i-og), Ju-piter (Diu-pater), &c. The den- 
 tal aloneis retained in some words, as in δά-6κίος=ξα-ακίθς{δίά6νΛθθ), 
 and dia itself is used in this sense uncompounded with any word ; see 
 Iliad XII. 104: ό d' ^TtQBTtL• %ai diix τίάντων. In Sanscrit this prepo- 
 sition also assumes the form vi-, like the Latin his^ ve, as well in the 
 more original signification of disjunction or separation (thus vi-yoktun, 
 disjungere), as with this intensive meaning (thus vi-mahat^ "very 
 great"). 
 
 181 The preposition 6vv, ξύν, always signifying "in conjunction 
 with," does not require much discussion. There can be no question 
 about the connexion of its first two letters with the Sanscrit prepo- 
 sition sa-m, and consequently with the pronominal stem sa (Grimm, 
 II. p. 1018). We have the former in ομον, αμα, simi-lis: we have the 
 latter in (ja-^ijg, from ^a and φώξ (compare ευγενής from γένος, ά-ΰ&ε- 
 νής from ΰΟ^ενος) (Bopp, iiber den Einfluss der Pronomina, p. 10) ; and 
 Λve have the same prefix represented by a simple a in a-KOLng and 
 α-λοχος-, "a sharer of the same bed;" α-κ6λου%ος^ "a sharer of the 
 same way" {κελεν%θζ)\ ά-γάλακτες, "sharers in the same milk" {οι 
 αδελφοί' οίονεϊ ομογάλακτες τίνες οντες, Orion, Etym. Μ. 5, 42); ά- 
 ά^λφό^ (or more correctly in the epic ά-δελφεός or ά-δελφειός) "sprung 
 from the same womb (δελφνς) ; " compare the Sanscrit so-daras = sa- 
 udaras; ά-νέψιος, "a first cousin, i. e. one with whom we are grand- 
 sons (νετΐοδες) together" (Max Muller, Oxf. Essays, 1851, p. 21); 
 ά-ελιοι, "brothers in law" {pi άδελφάς γυναίκας έχοντες, Hesych.), 
 connected with the Sanscrit sydla, "a wife's sister," for "having married 
 sisters they would have their sydlis in common" (Max Muller, /. c); 
 ά-πατουρία, " the festival of the paternal unions," i. e. of those who 
 have the same τΐατερες (C. 0. Muller, Dorians, i. 5 § 5, note) ; α-Λτερος-, 
 "equally winged" {Ιύόπτερος^ Hesych.), whence ατΐτερος ψάτις, "a 
 fleeting rumour" (^sch. Ag. 267); α-οζος, "belonging to the same 
 stem" (below, § 267; ατάλαντος, "of equal weight," &c. The iden- 
 tity of the prefix a- or a- and the fuller form άμα- is shown by the 
 fact that nymphs of trees were called not only Hamadryades, but 
 also Adryades or Hadryades (see the commentators on Propertius, i. 
 20, 12). The Latin cu-m is of course connected with ^ύν=^κ(5νν^ as 
 κοινός is with Jwog, and the Gothic ga is another form of the Sanscrit 
 sa (Grimm, ii. p. 751). The u in (3υ-ν•, cu-m is a remnant of the 
 original labial in the element Fa, 6(pa> 
 
 We must not identify the final -v of 6υ-ν with the m of cm-w, 
 sa-ma, da, ο-μού, α-μα, &c. ; nor must we forget the distinction be- 
 tween ύύ-ν and μετά. The Sanscrit sa-ma is obviously a compound 
 
CHAP. III.] THE PEEPOSITIONS. 343 
 
 of two pronominal roots, the one belonging to the second element, 
 the other to the first. Consequently, while the full force of the two 
 elements was perceptible the word expressed a conjunction of the 
 near with the here; and this was the simplest form in which the 
 mind of man could conceive an union. Just so με-τά was a compound 
 of the first and third pronominal roots, that of the first person being, 
 however, put foremost; thus, although the ide% of an union in 
 general was still conveyed by the word, this idea was combined 
 with that of a motion from the here to the there^ and it is in this 
 sense that μ^τά signifies "after," when joined with the accusative, 
 the case of motion. But, even with the accusative, the idea of 
 companionship or continuity is retained. Thus we have με-θ'' ημεραν, 
 "in the day-time" (Eurip. Orest. 58), μετά ννκτάζ, "in the night" 
 (Find. Nem. vi. 12). The Homeric με-6φα, a synonym for μέχρις 
 is used with the genitive in the sense of "until" (Iliad vm. 508). 
 This word is a compound of the first and second pronominal roots, 
 just as με-τά is of the first and tliird. It has no affinity with μεχρί, 
 which is connected with μακρός, as άχρι is with άκρος. Μετά is found 
 as mit in German, by a mutilation not unlike that of cum from 
 sa-ma. On the other hand, 6vv=5^a-v is merely a combination of the 
 second element with the third under that form of the latter, which is 
 used as the expression of the mere locative. So that it approximates 
 in meaning to εν•, with which it is ultimately identical. This is con- 
 sistent with the distinction which we have elsewhere laid down (on 
 Soph. ^71%. 115, 6) between 6vv and μετά; namely, 6vv denotes con- 
 junction or union on the same spot; whereas μετά implies companion- 
 ship or juxtaposition. In other words, Jw is used when we wish to 
 express that completeness of conjunction which enables us to regard 
 the combined elements as forming one whole, whereas μετά always 
 implies that the combination is separable. And here the Greek is 
 much more distinct than the Latin, which has only one preposition 
 cum to express ^,ύν, μετά, and τίρός. Thus in Thucyd. i. 18, we have 
 ξ,νμμαχοϋ and oi ξνμτΐολεμήΰαντες to express a regular confederacy; 
 but the writer says: ετΐολεμηόαν μετά των ξυμμάχων %ρος άλΧτιλονξ•, 
 "they fought against (with) one another in conjunction with their re- 
 spective allies," though aRoman would say: "pugnabant cum hostibus." 
 
 182 The preposition κα-τά is a form perfectly analogous to μετά. 
 The first part, however, is itself the compound form κεν (above, § 114); 
 and as κείνοξ=^κενυος is ultimately identical withαλλos=άVtog,κά-rcί 
 =^κεν-τά must be ultimately the same word as αν-τα., "opposite to." 
 And this is really the case. If we examine all the uses oi κατά we 
 shall see that they resolve themselves into an expression of parallelism 
 
344 THE PREPOSITIONS. [book II. 
 
 and correspondence, such as would be suggested by the juxtaposition 
 of two objects, placing them on the same level, in the same line, face 
 to face. Such, for example, is the use of κατά in those phrases, to 
 which Bopp has called attention in his vague and confused remarks 
 on this Greek preposition (ilber den Einfluss der Pronomina auf die 
 Wortbildung, p. 5). For ov κατά Μι^ραδάτην means "not according 
 to the standard of ^ithradates," and μίίζων rj κατ ανΟ'ρωτίον signifies 
 " greater than in accordance with the standard of a man." Cf. Horn. 
 7/. XXIV. 630: d'BOLoL γαρ αντα εώκεϋ. That this usage really springs 
 from the sense of juxtaposition and contrast contained in κατά is 
 clear from those passages in which κατά is really a synonym of its 
 congener αντα^ and signifies " in front of," " opposite to," e regione. 
 Thus in Herodotus, VIII. 85, we have κατα^Α^ημαίονζ ετετάχατο Φοί- 
 νικες; and in IX. 31, κατά Λακεδαιμονίους εύτηΰε Περοας is explained 
 by the immediately following words: ort μεν ην αυτόν δννατώτατον 
 άπολέξας εΰτηΰε άντίον Αακεδαινονίων. And similarly in a geogra- 
 phical description, Herod, n. 75: εΰτι χώροξ τηςΆραβίης κατά Βου- 
 τούν τίόλιν μάλιΰτά κη κείμενος^ "there is a place in Arabia almost 
 exactly opposite the city of Buto." And this city itself is described 
 (ii. 155) as κατά το Σεβεννυτικον καλεόμενον ΰτόμα τον Νείλου, 
 "over against the Sebennytic mouth of the Nile," i. e. on the other side 
 of the Buticus lacus. Cf. Hom. 11. ii. 626: dl ναίουΰι περην αλός, 
 "Ηλιδθζ*αντα, "the islands which lie across the sea, over against Elis." 
 
 183 In order to understand properly the various uses οι κατά, 
 thus identified with αντα, we must consider it in immediate connexion 
 with ανά, which is found as its correlative or counterpart in almost 
 every one of its significations. Thus, if we have κατά. τον τίόλεμον 
 (Herod. VII. 137), we have also ανά τον πόλεμον τοντον (Herod, νιιι. 
 123) with but a slight difi'erence of meaning: we have both ανά 
 ότρατόν (Eurip. Phceniss. 1309), and κατά ΰτρατόν (Iliad \ιι. 370): 
 and both ανά and κατά are used with numerals to give them a distri- 
 butive signification. As counterparts, ανά, ανω are used to signify 
 "up," "motion up;" κατά, κάτω, "down," "motion down." 
 
 If we examine ά-νά more minutely, and compare it with κατά, we 
 shall arrive at a satisfactory explanation of their correlative use. We 
 have already remarked that the ultimate pronominal form a- must be 
 referred either to the second element fa or to the third va. When 
 therefore it is prefixed to this latter element, as in the prepositions 
 ά-νά,εΙξ=^ε-νξ, and εν, it seems probable at first sight that it repre- 
 sents the other element S^a; for a compound particle, indicating a 
 relation between two positions, could hardly be made up of a redupli- 
 cation of the same element. But, in addition to this a priori con- 
 
CHAP. ΠΙ.] THE PllEPOSITIONS. 345 
 
 sideration, a comparison of ίνα and the Sanscrit e-na=ai-na=ya-naj 
 furnished a strong presumption in favour of the opinion that in the 
 case of 8V and εΙς^Ινξ the initial element is a mutilated remnant of 
 the second pronoun ^a ; and when we discover that a-va-l•,^ which, as 
 we shall see afterwards, is derivable from α-νά, was J^a-val in Homer's 
 time, and that in all probability it contains the same elements as the 
 Hebrew ^ha-no-M, which with the exception of the reduplicative first 
 syllable corresponds to its synonym Ι-γώ-νη^ we have as much evi- 
 dence as we can expect in this ultimate refinement of etymological 
 analysis, to convince us that the full form of α-νά was a compound of 
 J^a and va. Considered under this point of view, there ought to be 
 no difi'erence of meaning between l-v^ ΰν-ν, and ά-νά, which are equally 
 compounded of the second and third elements ; and, in point of fact, 
 there are many correspondences in the use of these prepositions. As 
 far as sv and ανά are concerned they do not difi'er in signification 
 otherwise than in and on, their English equivalents. For the same 
 reason there ought to be an identity of meaning between ανά and the 
 first syllable of κατά, namely, κα=κεν, and there really is this equiva- 
 lence in the particles av and κεν. It must therefore be the termi- 
 nation -τα which constitutes the difference between ανά and κατά, and 
 a little consideration will show that the affix operates in the same 
 way and with the same results as the ρ which distinguishes ντΐέρ from 
 νπό (§ 179). For iiava means "up," but κατά, "down;" it ανά means 
 "backward," but κατά, like αντα, "in front," the subject must be 
 regarded as a point in the middle of a line, whether vertical or hori- 
 zontal, of which one extremity is indicated by ανά and the other by 
 κατά. The termination of κα-τά and άν-τα must, like that of εΐ-τα, 
 ετί-ει-τα, εν^αν-τα, &c., be a corruption οί-τεν=-%'εν', compare επ:Β- 
 τεν,εντευ-%εν; and this termination is generally ablative, i.e. it denotes 
 separation or motion from a place; so that the relation between 
 κα-τά=κεν-^εν andafa^p^i/a^xfi/ismuchthe same as that between 
 super, sub-ter, sub, νπερ, ντΐό. And we have a further analogy in 
 α-τερ, αν-τερ, Lat. inter, Sanscr. antar, Germ, unter, which in some 
 of their applications give the sense opposed to ανά, "up," namely, 
 "down," "between two points on the surface," "out of the way," 
 and generally "separate," "apart" (below, § 204). That the anti- 
 thesis of "backwards" -and "forwards," "above" and "below," may 
 coincide with that of "far" and "near," is shown by our common 
 language ; for we say " here in front of us," but " behind there,'' and 
 "up there,'' but "down here." 
 
 184 The preposition ανά occurs, either separately or as a prefix, 
 in almost every language of the Indo-Germanic family, and there are 
 
346 THE PREPOSITIONS. [bOOK II. 
 
 few words which have more varied functions to perform. It is found 
 even in the Semitic languages; for the negative ixb and the prohibi- 
 tive bfi< are clearly connected with the prepositions bN, ~b = Iv, in 
 (see Maskil le Sopher, p. 15)*. In Greek this particle appears not 
 only as the dissyllable ανά, but also under the monosyllabic forms va 
 or vrj, and av-^ and even without the characteristic nasal as the prefix 
 «- or ε. Similarly the Sanscrit ana is reduced occasionally to the 
 initial a, and na occurs separately in Pali, though it is used only as a 
 termination in Sanscrit and Zend (Bopp, Vergl. Gramm. p. 531). 
 This latter element stands independently in Latin, in the words nam, 
 (Bopp, Vergl. Gramm. p. 534), num, nun-c, we, we, ni, m, non; in 
 Greek, in the words vv^ νυν (compare 6v-v, sa-m), vat, νή, νη-, &c. In 
 Sanscrit the full form a-na is used as a negative prefix : thus anapa- 
 kara, " harmlessness," "freedom from hatred and malice," is com- 
 pounded of ana and apakara, "evil doing." It is both in this sense, 
 and as an augment indicating past time, that it is shortened into a. 
 The same is the case with the Greek ανά. We have both av- and va 
 or V7J as negative prefixes; we have ανά so used as a prefix, and sepa- 
 rately, in the form άνεν^ with a very similar signification; we have a 
 for av as a negative prefix f, and we have a- for ανά as the verbal 
 augment. That va is not a mere abbreviation of ά-νά, like the modern 
 Greek δεν for ονδεν, appears from the fact, that av- is found with the 
 same privative meaning as va and ανά ; and in the Greek ά-%6 {a-il))^ 
 α-τερ (a comparative form), &c., as well as in the Sanscrit a-pa and 
 a-va^ we have a for na. In a difierent application of the same pro- 
 nominal combination we have seen above that the Hebrew ^ha-ni, 
 when used as a verbal prefix, is shortened into 'Ae, and that ^henesh 
 becomes 'hish. With regard to -va we must remark that the primi- 
 tive meaning of this stem is sufficient to account for its negative use, 
 without assuming that when so used it is merely an abridgment of 
 a-vd. We have before pointed out the idea of separation, removal, 
 distance, conveyed by the words vlv, νόΰ-φί, v00-rog, &c. ; this power 
 
 * It is to be remarked that V? bears the same relation to ανά, "up," that 
 ■^ay does to super, νπέρ, iiber, ufar, "over," and thus completes the chain of 
 analogies between the Greek and Hebrew particles involving h and v. 
 
 f On the supposition that the first syllable of ά-νά is the element ofj^a, it is 
 obvious that the negative prefix cannot be this syllable only: itmustbe a remnant 
 either of άνά the fuller form, or of να, the ν having evanesced according to the 
 general principle. We prefer the former alternative, and the same may have been 
 the view ofLepsius when he proposedthefollowingthesis: "particulamprivati- 
 vam apud Grgecos a principio unam av fuisse; ceteras, quae reperiantur, formas 
 secundum euphonise leges inde natas esse" (de Tahulis Eugubinisy ad calc). 
 
CHAP. III.] THE PEEPOSITIONS. 347 
 
 of the element -na is so forcible, that when appended to the element 
 ka, which is the strongest expression of proximity, it converts it into 
 a pronoun (%sl-vog), denoting distance in a very marked manner. It 
 is to this same idea of distance that we owe the meaning of negation 
 contained in va ; for after all, what is a negation but an expression 
 of farness or removal? and what way of negativing have we in our 
 own language more decided than the common "far from it"? 
 
 It is on this account that we also find άτΐό and apa=na-pa with a 
 negative meaning both in Greek and Sanscrit. Thus the word apa~ 
 kdra, mentioned above, which is composed of apa, "from," and kri^ 
 "to make," signifies "evil-doing," "injury," in perfect analogy with the 
 Greek : ovdlv ατίο τον άνΟ'ρωπείου τρότΐον πετΐοίήκαμεν (Thucyd. 
 I. 76). It is this use of από which has given occasion to the employ- 
 ment of άττοκαλεω in a bad sense : thus Soph. Aj. 727 : τον τον μανεν- 
 tog κάταβονλεντον ΰτρατον ξνναίμον άτίοκαλονντες, "calling him 
 by way of abuse." Aristoph. Aves, 1263: άτιοκεκλήκαμεν δι,ογενεΐς 
 %'εονξ (unless we ought to read άτίοκελγικαμεν). Xenoph. Metn. i. 2, § 6 : 
 tovg δε λαμβάνοντας της ομιλίας μιΰ^ον άνδρατίοδίότας εαντών 
 άπεκάλεί. ι. 2, § 57: άργονς άτΐεκάλεν. ι. 6, § 13: βοφιΰτας ωΰτίερ 
 πόρνονς ατίοκαλονύυν. Plato, Gorg, p. 512 c: cug £v ονείδεϋ άτίοκα- 
 λέΰαίς αν μηχανοτΐοίόν. Andoc.c.^Zcii.31,10: αλλονς ολιγαρχικούς 
 — ατίοκαλεΐ. Eurip. Iph. Λ. 1354: οι με των γάμων άτίεκάλονν 
 ηΰΰονα. Plut. ilforaZ. ρ. 204 F : άναιΰ&ήτονς και βαρβάρους άπεκάλει. 
 TAeceiei.p. 168d : χαριεντιΰμόν τίνα «ττοκαλών, which Schleiermacher 
 properly translates "nannte er nicht dies einen schlechten Scherz;" 
 Demosth. Fals. Leg. 47: λογογράφονς τοίνυν και ύοφιΰτάς άτίοκα- 
 λών. Aristides,Tom.ii.p.383: «ττοκαλίΓι/ ά/1«ξονα (see also Stallbaum 
 on Plato, Gorg. u. s.). The meaning " up," with which ανά is so often 
 found, is only another modification of the same idea, for highness and 
 farness are related notions. This is shown by the word άνεκάς, which, 
 we conceive, is a compound of ανά and εκάς, although Doderlein sup- 
 poses that it is formed from ανά, as τίερίξ is from περί, and that it is 
 related to εκάς in sound only (De άλφα intensivo, p. 12). Bockh has 
 endeavoured to show {Notce Crit. in Find. Olymp. n. 23) that άνεκάς 
 always means "upwards." Plutarch says that the Attics used άνεκάς 
 for ανω, and ανέκαθεν for άνωθεν (in vita Tfiesei, cxxxiu). The 
 grammarians allege that άνεκα^^εν can only be used ετΐΐ τότίον-, but 
 that ανω%εν may be used ItcI χρόνον, and this is partly true (see Phry- 
 nichus,p.270Lob.; Thomas M. p. 77; Bachmann, Anecd.Y ol.ii. p. 393). 
 One would think that there must be some distinction between άνωθεν 
 and άνεκα\^ενι and that the latter was the stronger of the two, from their 
 use in-^schylus, Choeph. 421, though Lobeck (adSoph.Aj. 145, p. 148) 
 looks upon this as one of many instances of a mere redundancy of 
 
348 THE PEEPOSITIONS. [bOOK II. 
 
 expression: "neque negari potest," he remarks, "de vocabulis idem 
 valere, quod de foliis dici solet, nullum unum alteri perfecte simile 
 esse, sed hsec discrimina plerumque delitescunt." The Scholiast on 
 Aristophanes (Vesp. 18) says: άνεκας da άντΙ του ανω, τΰάνν sKag 
 καΙ dg v^og, which we believe to be a correct statement; the word 
 implies both height and distance, as in the line of Pherecrates emended 
 by Yalckenaer (Diatrib. in Eurip. p. 285): τοντί τι εβτυν] og av^KCcg 
 to χρίβανον : and this is also implied in the passage of Pindar on 
 which Bochh is commenting : όταν dsov μοίρα πέμτΐΐ;] avsKccg δλβον 
 ν^ηλόν, i. e. πψπΎ} avEnag ωότε ν-φηλον είναι. 
 
 As ανά is used negatively, so is κατά used affirmatively. But in 
 this case, at least in composition, the opposite of %ατά is often άτΐό: 
 thus, κατά-φημι., "to say yes," άτΐό-φημί, "to say no." As opposed 
 to κατα-νενω, "to express assent by nodding," we have both άτΐο- 
 νενω and άνα-νενω with the contrary signification. The simple a- 
 was also a direct opposition to κατά, as Thucydides plainly announces 
 (1.123): ov γαρ dy] 7tεφεvγότεg ταντα επΙ την 7tλεί6τovg δη βλά-φα- 
 6αν καταφρόνηΰιν κεχωρήκατε, η εκ τον TtoXXovg ύφάλλεον το 
 εναντίον όνομα ά-φροΰννη μετωνόμαΰταί. 
 
 185 That ά-νά is used not only in a negative or privative sense, 
 but also with an intensive meaning, has been asserted by the old 
 grammarians, but the instances adduced have been doubtfully receiv^ed 
 by modern philologers; and in some cases, when this prefix appears 
 in its shortest form a-, it cannot be determined whether it is a rehc of 
 ανά, or a representative of α=όα (above, § 181). The words, which 
 are supposed to commence with the intensi\'-e ανά, have been examined 
 by Doderlein in a special treatise (Commentatio de άλφα intensivo ser- 
 monis GrcecL Erlang. 1830). The following, commencing with a-, 
 seem to be intensive and yet not collective; namely, ά-τενης, "exces- 
 sively stretched, obstinate, stubborn;" ά-χανής, "widely opened, ex- 
 tremely gaping;" ά-ΰ7tερχέg, "very eagerly;" ά-ΰκελέg and ά-ΰκελεωg, 
 " exceedingly hard ; " ά-κήδεία, "very great sorrow" (Buttmann, ^W5/. 
 Sprl. II. p. 358); and that these may presume the prefix ανά is clear 
 from the words which have the prefix νη- with an intensive significa- 
 tion: such are ι^τ^^^άτεος (ayad'og), vηδvμog {ηδύg), vηλL•τήg (άλLτηg\ 
 vη7tεδavόg (ηπεδavόg), vητρεκώg (άτρεκώg), vrjxvtog (jtokvxvtog), 
 vωλεμεg (είλεΐν, comp. ovλcίμόg, Doderlein, de άλφα intensivo, pp.21 
 foil.). We do not refer vηρiτog to this class. It seems to us rather to be 
 connected with the root oi vηρεvg,&G., and thus i\iQvήρLτog νλη of He- 
 siod {Op. et D. 511) will imply a floating fluctuating mass of foliage 
 (above, p. 160, note). Supposing, however, that the intensive a- and νη- 
 belong to the same pronominal element as the negative prefix of the same 
 
CHAP. III.] THE PREPOSITIONS. 349 
 
 form, it becomes necessary to inquire how the same form could have 
 borne meanings so opposite. 
 
 Doderlein attempts to reconcile the negative and intensive uses of 
 a as follows {de ίίλφα intensivo, p. 24): "I consider that by a sort of 
 abuse the proper force of a privative has been turned into an expres- 
 sion of excess {nimietas) ; just as in those words in which a is put for 
 dvg\ for the depravation of a thing is nearly the same as its defect, 
 and excess is nearly the same as depravation. Thus, among the Greeks, 
 άπάλαμοξ is, in Homer, he who is without skill, in Pindar, he who 
 makes a bad use of his skill; and in German, Unlust is used in a 
 privative sense, Unstern in a depravative sense, and Unstier in an 
 exaggerative and intensive sense. If greatness lies between immensity 
 and littleness^ immense and little things are alike excluded from the 
 category of greatness." Pott {Etymol.Forsch. zweite Auflage, i. Theil. 
 p. 387) explains the negative prefix with an intensive signification 
 in such words as Unzahl^ "an enormous number;" ungross, "very 
 great;" Unkuh^ "a great cow," &c., by the conception of a magni- 
 tude without comparison or proportion. In our opinion the negative• 
 and intensive significations of a- are alike due to the notion of "far- 
 ness," which we have pointed out as the primary one of α-νά^ να-, and 
 «-; for distance suggests magnitude or extent quite as much as sepa- 
 ration and incompatibility ; and we shall see, in the following chapter, 
 that in the form ναι or vri this negative prefix appears as a most 
 emphatic affirmation. With regard to Doderlein's supposed connexion 
 between "excess" and "depravity," it cannot be doubted that μάλα 
 and μόλΐξ are connected with mains, and that μόγις and μόχ^οξ spring 
 from the same origin as μ^γαξ. So, too, we understand Virgil's iniquo 
 pondere rastri as referring only to excessive weight; and in-gens = 
 υπερ-φνης, means " over-grown," i. e. of excessive magnitude. 
 
 186 But by far the most important of the correlative uses of ανά 
 and κατά is their appearance in an abridged form as av and κεν, with 
 much the same functions to perform. That the av, which is found in 
 conjunction with verbs, expressing the apodosis of a condition, is in 
 reality the preposition ανά, appears clearly enough from the use of 
 that preposition in the old poets under the same shortened form 
 (Bockh, Not. Critt. in Find. p. 387). It is also sufficiently obvious 
 from the use of the prepositions, άν-τί, which bears the same relation 
 to ανά that προ-τί does to tcqo, and άμ-φΐ, which stands for άναφί, as 
 αμφω does for άνα-φω=άνα-δ^ω. It is altogether a matter of indif- 
 ference whether we regard a-v as the locative of the mutilated pro- 
 noun a, or consider the ν as part of the suffix -va, for the ν of the 
 locative is simply this suffix in a mutilated state. 
 
350 THE PREPOSITIONS. [bOOK II. 
 
 The identity of the preposition ά-νά and the particle av was recog- 
 nised many years ago by Hermann, who, in his laborious treatise on 
 the particle av (Opuscul. Vol.iv. p. 6), explains the use of the particle 
 from the sense of secundum, " according to," in which ανά is found ; 
 thus, he says, Ββονλόμην av d εδυνάμην is equivalent to εβονλόμην 
 άνα tovTO εΐ εδυνάμην. He also connects κεν with και, on the ground 
 that og κε Ο'εοΐς εΛετίεί^'ψαι is equivalent to οξ καϊ %'εοΐς ετίετίείΰ'ηταί. 
 With regard to the meaning of the particles av and κέν, he thinks 
 that these two words, like ϊΰως, τΐού, and τέ, are only so many modifi- 
 cations of the idea of probability. This explanation is quite insufficient, 
 but Hermann is undoubtedly right in comparing av with ανά. That 
 κεν also is connected with κατά, and that the first syllable κα was 
 actually used by the older Greeks instead of the more lengthened form 
 in which the preposition appears in the classical authors, was first sug- 
 gested by Adolfus Weber, Professor of Mathematics in the Gymnasium 
 at Torgau, who has shown that κατά is shortened into κά, not into 
 κάτ, and that κάτ has arisen more from usage than from the nature of 
 the case. Welcker, in his notice of Weber's pamphlet {BJieinisches 
 Museum for 1835, p. 638), has given an instance of the use of κά for 
 κατά from an inscription of the 69th Olympiad (in Chishull, Antiq. 
 Asiat.^. 49), where we have ΕΖΗΤΗΣΕ ΤΑΣ ΠΟΛΕΙΣ Τ ΑΣ 
 ΚΑ ΤΗΝ ΣΕΑΕΤΚΙζ/Α. Weber therefore infers the derivation 
 of av and κά(κε), from ανά, κατά *, and we agree with him in thinking 
 that ανά, κατά, and av, κεν are cognate and parallel correlatives. 
 We believe, however, that κα=κεν is not apocopated from κατά, but 
 that the latter is a subsequent and lengthened form, bearing the same 
 relation to κα=κεν that αν-τα does to av. For if κα had not resulted 
 from the independent κεν (according to the principle often referred 
 to), the longer form would have retained the ε, as in με-τά, εΐ-τα, 
 κεΐ -vog, κεΐ-^Ί, &c. 
 
 It is in accordance with all that we know of the origin of prepo- 
 sitions, to suppose that they would naturally and necessarily be used 
 by themselves as cases of pronouns before they were employed as 
 supplements to the cases of nouns, when those cases had, by the 
 mutilation of their endings, lost their original significance. We ob- 
 serve remains of this use not only in classical Greek, but even in 
 
 * De κατά Prcepositionis Apocope, scr. Ad. Weber, Gymnasii Toryovani Suh- 
 conrector et Disciplince Mathem. ac Physicce Doctor. 1835. Since the publication 
 of the first edition we have succeeded in obtaining a copy of this valuable little 
 essay, with the following notice respectingtheauthorfromDr.G. W. Miiller,the 
 Rector |of the Torgau Gymnasium: "Obiit Suerini a. d. IV. Jan. anno p. Chr. 
 1842.G.W.M." Thereal effect of Weber's argument, although it does not seem 
 to have occurred to himself, is to prove the independentexistenceof xa = xfi/as 
 a preposition equivalent to its oiFspring γ,α-τά. 
 
CHAP. III.] THE PEEPOSITIONS. 351 
 
 modern French and Italian. In the former we find almost all the 
 prepositions used in their primary sense as adverbs of place : thus we 
 have εν, "at the same time" (Sophocl. (Ed. Tyr. 27, quoted above, 
 §170); 87th "in addition" ((Ed. Tyr. 18S: "εν d' αλοχοί, πολιαί τ ετΐί 
 ματερες. Antig. 789 : τιαί ΰ' οντ αθανάτων φνξίμος ουδείς, οϋ%'' 
 άμερίων ετΐ άνΟ'ρώτΐων; see however, Matth. Gr. Gr. § 584?^); ΐίρόξ, 
 "in addition" (Plato, Besip. p. 466 E: καΐ Ίΐροξ γε αξονΰι, Demosth. 
 Philipp. I. p. 47: τάλαντα ενενήκοντα καΙ μικρόν τι τΐρός): or even 
 two at once in Homer, as in Hiad y. G6: η δε δια τΐρο αντίκρυ κατά 
 κύύτυν υπ όΰτεον ηλυ^ άκωκή. That κα-τά and ά-νά are also used 
 in this way is well known: thus we have in Herodotus (i. 208) : η μεν 
 δη εξαναχώρεε, κατά υτΐέΰχετο πρώτα, and (ιπ. 86): οί εξ, κατά 
 (5υνε^'ηκαντο, παρηόαν ετά τα ν ίππων: — and in Homer (Iliad xvm. 
 562): μέλανες d' ava βότρυες ήόαν. In Italian and French the pre- 
 positions ne and en, both signifying "in," are used as general adverbs 
 of relation with the meaning "of this," "from this," "with regard to 
 this." It is generally supposed that these particles are derived from 
 inde, as y is from ibi (Grimm, D. Gr. iii. p. 746 ; Raynouard, Gr. d. I. 
 langue Rom. pp. 86, 268), and there is no doubt that this was the 
 origin of their use ; but the forms themselves show that these adverbs 
 were eventually superseded by the preposition included in the Latin in-de. 
 The use of av and κεν, like that of all the other particles in Greek, 
 is a proof of the early tendency and striving of that language after 
 clearness of logical expression. It is true that the moods of verbs, as 
 well as the cases of nouns, are capable of expressing, without any out- 
 ward helps, all the necessary modifications of meaning. But they can- 
 not always do this with logical distinctness, even when they retain the 
 fall force of their inflexions ; and when, in process of time, the ending 
 is overbalanced by the body of the word, or sacrificed to the laws of 
 euphony, it becomes impossible to express the different local relations 
 or cases of nouns without prepositions, and the difierent modal rela- 
 tions of verbs without particles, to designate the dependence or sub- 
 sequence of the secondary sentences. One of the great beauties of 
 the Greek language, as it stands, is its frequent use of particles or 
 pronominal words for this purpose, but no one of these particles is 
 employed with more efficacy than those two fragmentary prepositions 
 of which we are now speaking. The older Greek writers used both av 
 and κεν to express the apodosis of an hypothesis or condition: the 
 later authors employed only the former. According to what we have 
 stated above (§ 166), κεν would, if not an enclitic, have precisely the 
 same meaning as α-νά, '^in that." As, however, κεν is but a dependent 
 word, it stands on the same footing as τε and τι, που and πως, which 
 are all etymologically connected with it, and signifies "in some place 
 
352 THE PREPOSITIONS. [bOOK II. 
 
 or other," "in any way," and, by implication, '^perhaps." In fact, 
 7CBV bears the same relation to TLg=tLV-g that ava=Fava does to 
 κείνος (§ 149). Κ τε is an enclitic form of και, the relation between 
 them is altogether the same as that which subsists between κεν and 
 av. Indeed there is a much closer connexion than is generally sup- 
 posed between the copulative sentence and the hypothetical proposi- 
 tion. In either case we have a relative with an indefinite antecedent. 
 Thus in άνδρες τε καϊ ΐτίτζοί, ^^ where horses, there men," τε is an in- 
 definite antecedent to the relative καί', and in ει Tig ταντα δρώϊ], 
 ayad'og civ εϊη, ^^as often as any one did these things, he would so 
 often be a good man," the particle av is the indefinite antecedent to 
 the relative εΐ (see Gr. Gr. Art. 397, and elsewhere). The particles 
 Τ\Γ\:> and Ί, which are both of a demonstrative nature, the latter being 
 in fact a residuum of NiM, are similarly used in the apodosis as the 
 antecedents of dN, which is clearly connected with the roots d^, t3|, 
 cum, αμα, &c. (see Maskil le Sopher, pp. 14, 30). In Sanscrit the 
 intimate connexion between the hypothetical and relative sentences 
 is shown not only by the obviously relative origin of yadi, but also 
 by the occasional parallelism of the relative and cAei = "if:" thus in 
 the Hitopadega, gl. 29, we find: ^'■yad abhavi, na tad bhavi; bhavi 
 chen na tad anyatha," L e. ^'•what Avill not be, that will not be; if it 
 will be, this [is] not otherwise." When we find re in the first sen- 
 tence, and καί in the latter, which is the common construction, the 
 meaning conveyed is, that what is affirmed generally (τε = "in any 
 way ") of the former, is affirmed in the same way of the latter {και 
 =z "in this"). When τ ε appears in both sentences, the meaning is, 
 that what is affirmed in any way of one is predicated in some way of 
 the other. Similarly, we should expect (1) that κε would appear in 
 the hypothesis and av in the apodosis, with this meaning — if such 
 were "in any way" (κε) the case, then "in that case" or "farther" 
 (ά-νά, av) such things would follow: or (2) that κεν would appear 
 in both, with this sense — if such were "in any way" the case, then 
 "in some way" such things would ensue. We frequently find both of 
 these constructions in the epic and lyric poets, as in the following 
 examples; (1) Homer, Odyss. viii. 353: τΐώς av εγώ 6ε δεοιμι μετ 
 ά%ανάτοί<5ι %εοΐ6ιν, εϊ κεν'Άρης οϊχοιτο, χρέος καϊ δεΰμόν άλνξας; 
 Pindar, Nem. ιχ. 34 : Χρομίω κεν ντΐαΰτΐίζων — εκρινας αν κίνδν- 
 νον ονείας άντάς. (2) Homer, Iliad νι. 50: των κεν τοι χαρίΰαιτο 
 τίατηρ άτΐερείόυ αποίνα, εϊ κεν ^με ξωον τίετίύ^^οιτ ετά νηνΰΐν 
 'αχαιών. Hesiod, apud Aristot. Eth. v. 5 : εϊ κε 7ta%0L τά κ ερεξε, 
 δίκη κ ί&εΐα γένοιτο. There appears to be a particular attraction 
 of the indefinite κεν into the protasis, as might be expected from the 
 generally vague nature of hypothetical sentences. Even av is ap- 
 
CHAP. III.] THE PEEPOSITIONS. 353 
 
 pended to relative, or, what is the same thing, hypothetical words in 
 the Attic dialect; thus, we constantly have εάν, όταν, og αν, &c.: and 
 ϋεν seems to have been similarly appended to the conditional particle 
 by the Cretans, as we may infer from the gloss on βαί-καν=^αί-καν, 
 in Hesychius , and from the Doric collocation αϊ-τία. In Homer and 
 Pindar we often find κε in the hypothesis without any corresponding 
 κε or av in the apodosis. Thus in Iliad, xix. 321: ov μεν γάρ τυ κα- 
 κώτεροι; άλλο τίά^ουμι, ονδ' εϊ κεν τον τίατροζ ατίοφ^ιμενοιο πν- 
 %οίμην. Pindar, Pyth. ιν. 263: εΐ γάρ τις οζονς οξ,ντόμω τίελεκει 
 εξερεί'ψαί κεν μεγάλας δρυός, αΐΰχννοι δε οί ^αητον είδος, καΐ 
 φΰ'ίνόκαρΛος εοΐΰα δίδοΐ 'φάφον τίερ αντάς, — where, however, %αί 
 stands as a sort of substitution for the κεν which might have appeared 
 in the apodosis. The fact is, that the hypothetical particle, in its older 
 and stronger form, is itself a relative word, as will be shown in the 
 following chapter, and even ε^, which is generally its representative, 
 and which is more immediately connected with i', where the idea of 
 nearness is not so strongly expressed, may always be referred both in 
 origin and meaning to the second pronominal element. 
 
 As there are instances in which και is found in the first of two 
 correlated sentences, and τε in the second, the enclitic κεν might occa- 
 sionally be expected to appear in the second sentence in opposition to 
 av in the protasis : the instances of this construction must be very 
 few; the only example, with which we are acquainted, is in Pindar, 
 Nem. VII. 89 : εΐ δ" αντο καΙ %εος αν εχοο, εν τίν κ ε^ελοι— ευτυχώς 
 ναίειν. It would be better, however, to read ανεχοι, with Thiersch 
 and Bockh, for the meaning clearly is — "if a God would condescend 
 to, would put up with, would not disdain, the law of good neigh- 
 bours," a sense which άνεχω bears in Euripides, Hecuba, 119: Ka<5' 
 ΰ άνδρας άνεχων λεκτρ' ^Αγαμέμνων, and Sophocles, Ajax, 212: επεί 
 6ε, λεχος δουρυάλωτον, 6τερξας άνεχει θούριος Αϊας. 
 
 187 One of the best proofs of the correctness of this view, with 
 regard to the meaning of av and κεν , is the place which these par- 
 ticles occupy in the sentence. Neither of them can appear as the 
 first word, but they are always placed in immediate connexion with 
 the conditional word , when they appear in the protasis , or in that 
 part of the apodosis, in which the reference to the hypothesis is most 
 distinct and prominent, in other words, where the antecedent would 
 be most likely to stand. The formation of the compound conditional 
 particles βαίκαν, αϊκα, εάν, όταν, &c. is a sufficient proof of the at- 
 traction of κεν and av to the conditional words: the following will serve 
 as instances of the mode of placing av in the apodosis. When there 
 is no reason for its appearing in other positions, av always follows the 
 
 AA 
 
354 THE PREPOSITIONS. [bOOK IL• 
 
 predicate: thus Sophocles, Ajax, 550: ώ ΛαΓ, ykvoio πατρός εντνχε- 
 6T£Qog,Tcc δ' αλλ' ομοίος' καΐ γενον αν ον κακός. Here is a wish 
 expressed by the first γενοι,ο, and the second signifies the result of a 
 condition; in the opposition therefore of the two repeated words, 
 the av should immediately follow the second: "may you be, &c., and 
 you will be in that case, &c." But in the following passage, where 
 there is a similar opposition of the optative proper to the optative 
 with av, the antithesis is between the two negatives, not between the 
 two verbs, and therefore the av appears immediately after the direct 
 negative ov, to which the indirect μη is strongly opposed ; Sophocles, 
 Antigon. 686: oik' av δνναίμην, μψ εΛίΰταίμην λέγειν, "I should 
 not even in this (i. e. if I knew how) be able, and I pray that I never 
 may know how to say, &c." The negative ov, and the cognate par- 
 ticle ovv, exercise an attraction upon av in the apodosis similar to 
 that which it experiences in the protasis from the relative and con- 
 ditional words. Thus, we very often find the collocations ονκ av, ονδ' 
 av, OVT av, ovtcox av, &c., and av is often drawn away from its 
 verb by the influence of ovv: compare Demosthen. OlyntJi. i. 13: rt 
 ovv av τις εϊποι ταντα λέγεις ημίν ννν; Olynth. πι. 14: τι ονν αν 
 τις εΪΛοι ΰν γράφεις ταντ είναι Στρατιωτικά ; Plato, Sympos.-p. 202 d : 
 τι ovv αν, εφην, εϊη δ "Ερως. The reason for this is obvious; the par- 
 ticle ovv refers directly and specially to what has preceded, and the 
 particle av must of course have the same reference in questions like 
 those which we have quoted. In general , whatever word in the apo- 
 dosis is to be expressed with most emphasis in reference to the con- 
 ditional sentence, whether that conditional sentence is expressed or 
 understood, this word is followed by av, Herodotus, m. 119: τΐατρος 
 δε και μητρός ονκ ετι μεν ξωόντων, άδελφεος αν άλλος ονδενι 
 τρόπω ^^ένοί,το. Thus also qualitative adverbs, like ηδέως, εΐκότως^ 
 τάχα, μάλιϋτα, &c., on which the emphasis always falls, are invariably 
 followed by av, unless some other word with a stronger attractive 
 power appears in the sentence: Plato, Protagor. p. 318 A: ηδεως av 
 φηΰι TCvd'iod'ai. As words expressing opinion about or information on 
 a subject have all the efiect of qualifying adverbs or predicative words, 
 we find that they also attract the particle into their immediate neigh- 
 bourhood; Plato, Phcedo, p. 101 ε : ΰν δ' εϊπερ ει των φιλούόφων 
 οΐμαι αν ως Ιγω λέγω ποιοΐς. It is prefixed to the word of thinking 
 when the emphasis falls upon a word before it, as in Plato, Respub- 
 lica, I. p. 333 A: τίρός γε νποδημάτων αν οΐμαι φαίης κτηΰιν, or 
 if a relative word precedes, as in Thucydides, i. 22: ώ^ δ' αν εδόκονν 
 εμοί — τα δέοντα μάλιΰτα εΙπέίν , or ονκ, as in the same author, π. 
 89: ονκ αν ηγούνται — άν^^ί6ταΰ%'αι ημάς. If emphatic adverbs 
 occur in the same sentence with the verb significant of opinion , the 
 
CHAP. III.] THE PEEPOeiTIONS. 355 
 
 av as a particle of reference is naturally enough repeated with the 
 different predicative words; thus in Thucyd. ii. 41, we find δοκεΐν αν 
 μοί τον αντον άνδρα ετά πλεϊ^χ αν εϊδη^ καΐ μετά χαρίτων μά- 
 λίΰτ αν ευτραηελωξ το ύώμα ανταρκεζ τίαρεχεύ^αυ. The most 
 curious and most instructive instance of this hyperbaton is the in- 
 trusion of aV, which belongs to an optative following, into the phrase 
 ουκ οϊδ' εΙ=ν6νβον tit. Euripides has ονκ otd' civ εΐ τίείΰαιμι in two 
 passages {Medea ^ 911, Alcestis, 49), but it is clear that the neces- 
 sities of the metre have obliged him to misplace the particle, which 
 certainly ought to follow the negative, as appears from Plato, Timceus, 
 p. 26 b: εγώ, a μεν χ^Ίς ηκονΰα, ονκ αν οίδ' εΐ δνναίμην ατΐαντα 
 εν μνήμγ Ttahv λαβείν, and from the somewhat similar passages in 
 Demosthenes, de Fals. Legat. p. 441, 21: ουδ^ άξ ευ oW οτι φή- 
 ΰειεν, and Prooem. p. 1423, 14: ονδεν αν τα νμετερ εν οίδ' οτι 
 βελτίον ΰχοίη. The αν appears unattracted in Aristoph. Aves^ 1018: 
 ονκ Οίδά γ* εΐ φΟ'αίης αν. 
 
 The particle κεν is distinguished from αν by its tendency to as- 
 sume an early place in the sentence. It is put before many words to 
 which av is regularly subjoined; thus as Hermann justly remarks 
 (Opuscul. IV. p. 7), if Syagrus (Herodotus, vn. 159) had not been de- 
 sirous of making a line η κε μεγ ώμώξειεν δ Πελοπίδας "Αγαμέμνων 
 in imitation of Homer's η κε μεγ' οΐμώξε^ε γέρων [τΐτΐηλάτα Πηλενς 
 {Iliad VII. 125), he would have said ή μέγα αν ώμώξειεν. 
 
 ΑΑ2 
 
CHAPTER VI. 
 THE NEGATIVE AND OTHER PARTICLES. 
 
 § 188 Grimm's remarks on negative particles. 189 Ma, μη, νψ ναι, and 
 ον-κ. 190 Interrogative particles, and their connexion vrith the nega- 
 tives. 191 Particles used in answers to questions. 192 Inferential par- 
 ticles. 193 Other words indicating progression or continuance. 194 Co- 
 pulative conjunctions sometimes due to the same connexion of thought; 
 195 but generally derived from the indefinite and relative pronouns. 196 
 Identity of κα and que. 197 Use of τε as an affix to relative words. 
 198 Different origin of τε and tol. 199 Disjunctive particles. 200 Com- 
 parisons. 201 Distributive particles. 202 Pronominal origin of 8η. Tem- 
 poral particles. 203 The concessive particle ys. 204 Faq and uqcc. 
 205 The hypothetical d. 
 
 188 TN the last chapter we were led, by an investigation into 
 . the origin of the prepositions ανά and %ατά, to make 
 some remarks as well on the particles aV and %iv, as on the 
 negative uses of the word α-νά , and its abridged or mutilated 
 forms. We shall commence our inquiries, upon the important 
 subject of the Greek particles in general, by a reference to what 
 we said there, as a natural introduction to the extensive question 
 respecting the words which express interrogation, negation and 
 inference, which, we shall find, are all connected in the Greek 
 and cognate languages. 
 
 Grimm, at the end of the third volume of his great work, has dis- 
 cussed this question with that extraordinary diligence and learning 
 which he everywhere displays. We refer our readers to what he says 
 with the greater pleasure, as this part of his work has been rendered 
 accessible to the English reader , and commented on , by a scholar of 
 considerable ability (Philol Museum ^ n. pp. 315 foil). We shall 
 make the German philologer's inquiries the basis of our own on the 
 present occasion ; for, although we do not think that he has seen the 
 general principle by which all the phenomena are to be explained, 
 and although he has in consequence fallen into some particular errors, 
 such are his learning and indefatigable industry, that we could not 
 hope to add much by our own researches to the vast induction of 
 particulars which he has collected and arranged. 
 
 He commences by stating (m. p. 708) the distinction between a 
 negation and an opposition; the latter includes the former, but not 
 vice versa. "The essence of the proper negation consists in the 
 logical denying of a position. By the expression not mountain , not 
 
CHAP. IV.] THE NEGATIVE AND OTHEB PARTICLES. 357 
 
 good, the position mountain, good is excluded, but it is left indefinite, 
 whether the opposite valley , evil , or the intermediate notion plain^ 
 middling, is to be supplied." He adds , " All negation proceeds from 
 the grounds of the position, and presupposes it. The position is in- 
 dependent, the negation necessarily refers to a position, and cannot 
 be expressed as anything new, but merely as a modification of the 
 position. This modification results from an insertion in the positive 
 position , which insertion in the first instance consists of the smallest 
 possible particle, producing an effect both rapid and sure. By de- 
 grees, however, this negative particle is usually connected very closely 
 with other words. Along with it we often find substantives that 
 strengthen the sense, which can even take the negativing power from 
 it to themselves." He then divides the simple negation in the Teu- 
 tonic languages into two kinds — the consonant-form and the vowel- 
 form. The fundamental letter of the consonant-form is N. Thus, 
 in Gothic it was ne, in High German nein=^ni-ein (so non, anciently 
 nenum, from ne-unum^)^ and in old English ne. The German nicht, 
 English noi, are compounds signifying no-thing; compare the old 
 High German neowiht^ niowicht^ nieht; middle High German nicht^ 
 niht; Anglo-Saxon naviht, nauht, nauht; English nought^ not (Phil. 
 Mus. u. p. 326). Of a similar formation is the Latin nihil=^ne•• 
 hilum. The English no is a compound of ne and the Anglo-Saxon ά 
 (Gothic ai, aio, comp. alfd-, aevum)^ which signifies always; ev-er 
 contains the same element. The middle High German prefix en- 
 does not we conceive arise from the old High German w' for ni, nor 
 do we think it is analogous to the formation of Ιμον, εμοί, ε^έ, from 
 μον, μοί, με (Grimm, p. 711). It is, we believe, the fuller form of the 
 negation (compare ά-νά\ and is connected with the German un-, ent-, 
 Latin in-, Grimm himself compares the old High German interro- 
 gative innu, inu, eno with the Gothic annu, and sees nothing strange 
 in the substitution of i in old High German for a in Gothic (in. 
 p. 757). The vowel-form of the simple negation is a suffix -at, -a or 
 -t, one or other of the two component letters being occasionally 
 omitted. This suffix is peculiar to the old Norse. It is probably, 
 as Grimm supposes (p. 718), a corruption oivdtt^ which is used to 
 strengthen the negation, like the German wicht, and the fundamental 
 negation is omitted, as the French negative is before pas, point, rien, 
 
 * Whether we accept this etymology or not, there is no objection to 
 it from the use of the word as a mere negative, having no relation to 
 unity in particular, e. g. in non multi, for in these compound negatives the 
 adjunct is very little regarded. Thus, although it is clear that nemo — 
 ne-homoy Virgil does not scruple to write nemo divom {jEneid ix. 6). 
 
358 THE NEGATIVE [bOOK II. 
 
 in pas un mot, point du tout, rien du tout. The prohibitive negation 
 is in Gothic ni as in ni gret I (μη κλαϊε) ; old High German the same 
 as in ni churi! (noli); in middle High German the prefix en as in 
 en ruoche I (noli curare) ; in Anglo-Saxon it is we, frequently strength- 
 ened by a following na=ne-a (never), thus ne vep J)w na, "weep not." 
 Grimm's general conclusion is as follows (p. 743): "On the whole, 
 then, there are two sorts of negation. The one quite formal and 
 abstract, which, though at first the soul of all negation, vanishes by 
 degrees in its separate use, and only continues its influence in con- 
 nexion with other particles. It is superseded by words which pro- 
 perly contain the idea of less, little, small, either quite concretely, or 
 perhaps in a more abstract sense. At first they are only united as 
 companions to the negative particle, and coalesce into an equally 
 abstract form, of which our new High German nicht, new Netherland 
 nietj English not is the most striking example. Frequently, however, 
 they dispense with the simple negative , and make a formal negation 
 out of their diminutive-sense, as is especially shown in the Norse icke. 
 Tliis interlacing of the formal, and, as it were, material negation, 
 explains to us two phenomena: on the one hand, the repetition of the 
 negative-particle, and, on the other, its complete dispensahleness. If 
 our new High German weder (neque)=:^o\di High German niwedar, the 
 middle High German wan (nisi)=7iewan, the Gothic 'ibai=inibai: in 
 like manner we have seen that also stoup ('an atom'), wint, tuivel 
 ('devil')*, and the old Nordish vcetr ('demon,' 'genius') serve as nega- 
 tives without any preceding abstract negation. The formal negation 
 is therefore unessential." 
 
 If now we compare this ni, ne, of the German dialects with the 
 Greek negative prefix νψ and with the second syllable of ά-νά , we 
 shall have no difficulty in recognising their identity. We have before 
 mentioned, that the middle High German prefix en- points to a fuller 
 form corresponding to the whole of ά-νά. The form (ne) of the simple 
 negative occurs in Latin in the combination ne-quidem with a word 
 interposed, and also in the compounds wow, neque, &c. 
 
 189 In the German dialects we have seen that the prohibitive 
 does not differ from the simple negative : the same is the case in the 
 Latin ne, except that the vowel is long. In Hebrew also the pro- 
 hibitive bN involves the same element as the negative i<b (above, 
 § 184). In Greek the prohibitive is μη, in Sanscrit md, and in Per- 
 sian me. The analogy of the German dialects might lead us , at first 
 
 * Just as ^'Devil a bit," is used in vulgar English to signify "not at all." 
 
CHAP. IV.] AND OTHER PARTICLES. 359 
 
 sight, to seek for some connexion between μή, and ne, as Grimm has 
 done (p. 745). But, when we consider that μά and νή are used in 
 direct opposition to one another in oaths , and compare μίν and vCVf 
 the distinction between which we have pointed out before, we are 
 compelled to seek for some way of explaining the word μη less obvious 
 but less objectionable than that of a transformation of η into m. 
 
 In the ordinary use of μά and νή^ the former refers to a negative 
 oath, the latter to a positive one ; moreover val, which bears the same 
 relation to νή that dai does to δη, is always used in a positive sense, 
 like the Latin nee. The question τι μην., is generally used with a 
 negative application; ή μην, which is used as a form of swearing, is 
 mostly found in a positive sense. With regard to μά we believe, with 
 Passow, that it is, in itself, neither affirmative nor negative, but gains 
 either the one sense or the other according as it has vcci or ov prefixed 
 or understood. In our opinion μά contains the element of the first 
 personal pronoun; it represents an original ^εν, which is used for ^ijv 
 in Herodotus , and bears the same relation to με-τά that κά or κεν 
 does to κα -ta : so that the leading idea is that of absolute nearness to 
 the subject. Κ ναί, νή, are, as we have no doubt they are, connected 
 with the second syllable of ά-νά, and the negative prefix νη-, the 
 idea conveyed by these particles must be quite the reverse; for the 
 leading meaning of vlv, ~va, νη- is, as we have already shown, that 
 of " distance," "separation." Grimm says (m. p. 767), "the seemingly 
 negative form of the affirmative vai (Lat. nae!) is worthy of notice; 
 we might compare ναι and ov with the Gothic ne and jai, except that 
 the meaning is reversed. Κ we take the Hessian ά=.ηα, in connexion 
 with the Swabian et=net, and the identity between the negative and 
 positive expression which occasionally presents itself, there results 
 apparently a deep-founded identity between the negative and affirma- 
 tive particle , which I purposely forbear to investigate farther." All 
 prima facie difficulty occasioned by this fact vanishes when we recol- 
 lect that the prefix νη- is used with an intensive or affirmative signi- 
 fication, and in general "yes" and "no" are only emphatic expletives, 
 which may be expressed by the two most definite pronominal words 
 ma and na, signifying separation and distance, as well as by any one 
 of those simple words by which we affirm or deny in our common 
 conversation. Our own "yes" is simply the second pronoun, denoting 
 "here," opposed to "no," the strongest form of the third element, just 
 as γε and κα-τά are used with an affirmative sense , in opposition to 
 ά-νά and ajto=av-vo or va-7to. 
 
 In order to explain μή , we must turn our attention to the other 
 and more direct negative ov, and consider what are the leading and 
 fundamental distinctions between the two particles. 
 
360 THE NEGATIVE [bOOK II. 
 
 There can be little doubt as to tbe derivation of ov , ουκ. The 
 only question that could be raised, is whether we ought to derive it 
 at once from the Sanscrit avak (deorsum)^ considering that an apocope 
 has taken place in the form ov, which is Pott's opinion (Etym. Forsch. 
 Π. p. 134, comp. i. 273, ii. 64, 183); or rather with Bopp (Vergh 
 Gramm. pp. 547, 8) connect it with the Indian-Zendic ava, Sclavonic 
 ovo, and take the final consonant as a mutilation of -xl, -yi^ the San- 
 scrit -cAi, -cJia, Latin -que ; so that ov is related to ον-κ, ον-χί, as ne 
 is to ne-Gj ne-que. We have no hesitation in adopting this latter 
 opinion, which, we think, is confirmed by the appearance of the muti- 
 lated root au (av) (§ 138), with the signification of "removal," "sepa- 
 ration," "contrast," which gives rise to the negative, in the Greek av, 
 av-9'Lg, αν-ΟΊ, av-rig*, αν-ερνω, &c., and the Latin au-tem, au-t, hau-t 
 (haud), au-fugiOy &c. (see Bopp, I. I. p. 546). A question might arise, 
 whether we are to consider av, αυα, as simply the element va, a form 
 of the first pronominal stem , with the semivowel transposed, or this 
 same element appended to another in a similar state of mutilation. 
 A comparison of aufero, with abs-tuli, ab-latus, and of a-va with ά-τΐό, 
 also used in a negative sense , and the difficulty of explaining the 
 second vowel otherwise, induce us to believe that a-va-k, ο-ν-κ-, are 
 the compound preposition a-va=^a-'Jto=^av-7c6 , with the element ka 
 affixed; so that a-va-k=ovκ combines the meaning of από and εξ. 
 The particle ovv=^avam=:navam is an additional confirmation of this 
 etymology of ον-κ. Hartung justly remarks (Partikeln, ii. 3), that the 
 double form ovv, ών, leads us, according to the analogy oiovg=og or 
 auris, to a ground-form avv. It is doubtful whether the Cretan and 
 Laconian word avg, quoted by Hesychius, is a synonym of avrog or of 
 ovg. He says, avg. avtog. KQψεg καΐ Aάκωvεg. Now we might sup- 
 pose with Soping that avTog is the genitive of avg, as orog is of ovg. 
 But in either case we have an analogy for the connexion of ovv with 
 av, and the Zend neuter accusative aom. If this be true, ον-κ ex- 
 presses a total disjunction, separation, denial; for as it means both 
 "from" and "out of" (§ 177), it is even stronger than the Latin 
 h-aut and non, and the Greek ά-νά, in which the fundamental mean- 
 ing is the same — i. e. distance, removal. Accordingly, ii μά and μή are 
 connected, ον-κ must be put alongside of ά-νά, νη-, and we must ex- 
 pect to find the same distinction between ον-κ and μή as we have 
 already observed to exist between νή and μά. The connexion in mean- 
 ing between ov, ανά, and ajto, is farther shown by the former being 
 
 * Bachmann, Anecd. Vol. ii. p. 376, 34: το αν^ις μετά τον 6 το πάλιν 
 ό αυϋ•ι δε Λοιητικόν, άντι τον αντό^ΐ' και αντίζ ποιητικόν^ μετά 
 
 το 
 
 τον τ, το νΰτερον 
 
CHAP. IV.] AND OTHER PAETICLES. 361 
 
 used like the other two , as a sort of negative prefix , and even in 
 proper names; thus we have (1) Οντίξ-, "Mr. Nobody," Horn. Od. ix. 
 366 ; Eurip. Cycl. 549, 672. (2) Ούκαλεγων, "Dreadnought,'' or "Care- 
 fornought," Horn. 11. m. 140; Virg. ^n. ii. 312; Juv. iii. 198. (3) Ov- 
 tOTcla^ Utopia^ "Weissnichtwo," "Kennaquhair, "Lord-knows-where." 
 All scholars who have written on the distinction between ο-ύ-κ 
 and μη , have made it consist in this , that while the former denies 
 absolutely, the latter denies relatively to some thought or opinion of 
 the speaker. In other words, ov-x denies objectively, μη subjectively*. 
 And what could be better suited to express subjectivity than a word 
 which contains, under a strong form, the simple element of the first 
 personal pronoun? Such a word is μη^ a stronger form of μά\ for μά 
 is opposed to νη as nearness is to farness, and μη to ου-κ as subject 
 is to object. This root appears with the same notion of subjectivity 
 in the verb μάω=μενω^ of which more hereafter. 
 
 190 We proceed to the interrogative particles. As we have dis- 
 tinguished negation into subjective and objective, so Grimm considers 
 that questions are either subjective or objective (m. p. 751): "In- 
 trinsically considered," says he, "there are two sorts of questions, the 
 one requires in the answer an expression of knowledge or opinion 
 about something, without the necessity for any direct affirmation or 
 negation; the other sort of question, on the contrary, desires the 
 affirming or negativing determination of the answerer. In the former 
 case the question turns upon the thing asked about, in the latter on 
 the will or knowledge of the person questioned. The former may be 
 called the objective^ the latter the subjective question. The following 
 are instances of the objective question: ^who is there?' 'whom lovest 
 thou?' 'how did that come to pass?' 'when did you see him?' The 
 following are subjective questions: 'is he there?' 'lovest thou me?' 
 'will you go with me?' 'is he alone?' 
 
 "This distinction according to the content of the questions is con- 
 nected with the formal distinction which has been premised. Objec- 
 tive questions are generally expressed by the sound, subjective by a 
 suffix. 
 
 "Inasmuch as the questioner does not know beforehand what 
 answer will be given, all questions are naturally doubtful. The sub- 
 jective question, however, is doubtful in a higher degree. The objec- 
 tive questioner desires information , and may receive it in an infinite 
 variety of ways. The uncertainty of the subjective questioner is tied 
 
 * For the distinctive syntactical use of ου and μ-ή, we must refer to 
 our Greek Grammar (articles 528—545). 
 
362 THE NEGATIVE [bOOK II. 
 
 to one of two possible answers, yes or no. Accordingly, we may add 
 an ^or notV to every subjective question; in other words, it may be 
 taken either positively or negatively. Instead of — 'is he there?' 
 *lovest thou me?' it would, in many cases, equivalent to say — 'is 
 he not there?' 'lovest thou not me?' (the expectation of the ques- 
 tioner being the cause of preferring one mode of expression to the 
 other). From this follows generally and at once a sort of connexion 
 between the subjective question and the negation, which the forms of 
 the interrogative also confirm." 
 
 The objective question is asked by the interrogative pronoun, or 
 some word derived from it. We have before shown that this pronoun 
 in Greek is, in its shortest and oldest form, J^- (κ /tg, τίζ). The sub- 
 jective question is expressed by interrogative particles, of which, 
 according to Grimm (in. p. 760), there were three forms in the Indo- 
 Germanic languages: (1) the Gothic suffix -w, which he considers as 
 perhaps related to the Greek oi) ; (2) the Gotlfic suffix -nu , old High 
 German -ww, Sanscrit -ww, Latin -we, Greek μη^ old High German 
 -na; (3) the Gothic an- , old High German m-, Latin an, Greek ovv, 
 which, he remarks, are all connected with the simple negation. To 
 these may be added the important particle aga or αρα, which asks the 
 questions apparently in a negative as well as in a positive way: this 
 also seems to have been the case with ονκ ovv. It is highly impor- 
 tant to know that all the roots mentioned by Grimm , and the aga 
 adduced by us, are used not merely as interrogative and negative, but 
 also as inferential, particles. 
 
 We have pointed out before the co-ordination which subsists be- 
 tween α-νά and κατά, in their various uses. We find them here 
 again employed in expressing, the former, a subjective, the latter, an 
 objective question. Or rather the root which enters into τια-τά , and 
 which is found more simply in κα, κεν, κε, is the same as that of the 
 interrogative pronoun , which in some form or other is generally pre- 
 fixed to objective questions: and both the successive mutilations of 
 α-νά^ as well as the whole word, are found as the interrogative par- 
 ticle in subjective questions in the Indo-Germanic languages. When 
 the ultimate β;=αν- occurs in this use in Greek, it is strengthened 
 by the particle ρα, and becomes αρα=αν-ρ(Χ. In Latin we have an 
 as a direct interrogative, and also -ne and num, the former being 
 used as a suffix only. 
 
 Although there is no occasion to connect the Gothic u with ov, as 
 Grimm has done, it is true enough that the Greeks used the objective 
 negation in these subjective questions, sometimes in a doubled form, 
 as ουκ, ovv, sometimes alone, and frequently in conjunction with the 
 subjective negation ^ij, as in μών^μψονν, μη ονχ οντωζ ^χ^, &c. This 
 
CHAP. IV.] AND OTHEE PARTICLES. 363 
 
 is easily explicable. We do not speak of the subjective negation in 
 the same sense in wliich Grimm applies the epithet to questions which 
 may be expressed negatively. The subjective negation refers to the 
 opinion of the speaker, the subjective question to that of the person 
 interrogated. But although these subjective questions always refer 
 to some opinion , will , or knowledge of the person questioned , they 
 may express in themselves , though faintly, an opinion on the part of 
 the questioner. We find that in these cases the subjective negation, 
 which refers to the speaker, is always prefixed. As a general rule, 
 whenever ov is found in an interrogation, a positive answer is ex- 
 pected; and the appearance of μγι in a question presumes a negative 
 reply. 
 
 191 When there is such a vacillation between the affirmative and 
 negative modes of putting a subjective question, we need not wonder 
 that there should be a similar ambiguity in the answer. Thus it is 
 that ναι, nee, obviously connected with the negative root, -va-, νψ, are 
 always used as affirmative answers, corresponding to the English 
 "yes." The writer in the Philological Museum above referred to, 
 explains the fact thus: he asks (p. 323), "may not this connexion 
 between the positive answers, and the negative root have arisen from 
 the use of the negative, like our 'nay,' immo, or anzi (ante), in Italian? 
 which are negative, inasmuch as they object to the preceding phrase 
 as not being strong enough, whilst they agree with its general mean- 
 ing and enhance its force? Certainly, in these cases, the negative and 
 affirmative senses often approach very near to one another, as, for in- 
 stance, in the following passage of Ben Jonson: Ά good man always 
 profits by his endeavour, yea, when absent, nay, when dead, by his 
 example and memory.'" The opinion of Pott {Etym. Forsch. n. p. 65) 
 is much the same: ^Hmmo," he says, "nicht bejaht sondern nur negirt, 
 oder (steigernd) opponirt." There is, it must be confessed, much of 
 truth in this explanation , but it is not sufficient. The proper view, 
 we conceive, is to derive this sense of ναι, like the other significations 
 of the same root, from the original idea of farness, distance, separation, 
 conveyed by it, and in which the meaning assigned by Pott and the 
 English critic are contained. It is curious, however, that in Latin 
 the application of the correlative words is inverted. For imo or 
 immo is an adverb derived from imus or immus for infimus^, and is 
 
 * A writer in the Classical Museum (Vol. in. pp. 291 — 297) proposes 
 to consider immo as a mutilation of in modo. This unhappy thought seems 
 to have been suggested by the fact that modo is counted as a monosyl- 
 lable in Terentian scanning, and that quomodo is represented by the Spanish 
 
364 THE NEGATIVE [bOOK II. 
 
 regularly opposed to ad summum^yaiich. corresponds to most of the uses 
 of ys or γονν^ and this is concessive and affirmative, whereas immo is 
 at least corrective, like the Greek μίν ovv (Grammar, Art. 567), if it is 
 not even negative. In the correlation, however, of the affirmative 
 κατά and the negative ava, the latter signifies "up" and the former 
 "down," which is just the reverse of the negative immo and the affirm- 
 ative ad summum. The fact is that the Roman spoke as if standing 
 between two distant points, the one below and the other above him: 
 whereas the Greek was contented with opposing the near to the dis- 
 tant; and negation is as much implied by a-7c6=av-7t6 and ον-κ=^αη- 
 va-k jwhich in the Sanscrit a-va-k signifies "down," as by a-va=fa-va, 
 which denotes "up." 
 
 192 From the signification of farness or distance we also explain 
 the use of words containing the negative element as inferential parti- 
 cles. We have the negative root a- for va- or avcc- in α-ρα, the second 
 part of which is the particle ρα , also used separately , which, as we 
 have before remarked, denotes motion, and is found in comparatives; 
 so that αρα will signify distance, or progression to another step in the 
 argument, which is the idea of an inference. It is owing to this that 
 καΐ απ εκείνου is used for αρα in the nearest approximation to a re- 
 gular syllogism which is found in Thucydides , namely , in vi. 89 : (a) 
 tolg γαρ τυράννους αεί τίοτε διάφοροι εβμεν (b) πάν δε το εναντιού- 
 μενον τω δνναΰτενοντί δήμος ώνόμαΰταυ' (c) καϊ an εκείνον 
 ξυμπαρεμευνεν η προΰταΰία ημΐν του πλήΟ^ους. We have also a dis- 
 tinct reference to the original meaning of αρα in those cases in which 
 
 como, Italian comey and French comme. But in the Latin authors themselves 
 modo and quomodo are never written mo and quomo, and conversely we 
 never find immodo for immo, to say nothing of the fact that, although we 
 have in modum, ad modunij and in morem, ad morem, in modo would be as 
 surprising a piece of Latinity as in more. Any good Latin scholar must 
 see that the citation of illico for in loco is no justification of the assump- 
 tion that immo=:inmodo ; for though we may say, dulce est desipere in loco, 
 we cannot say, in more modoque apis Matinee. Then if immo were felt to 
 be equivalent to in modo, Cicero would never have written et quid dico 
 nuperf immo vero modo ac plane paullo ante vidimus, &c. Lastly, which is 
 the main point, ad modum is used in answers to questions, in the sense 
 *'yes, by all means," which is surely the antithesis of immo, "nay;" and it 
 would be somewhat remarkable if in modo, supposing the existence of 
 such a phrase, were the exact opposite of ad modum=in modum. As the 
 writer in question is distinguished by a confident assumption of superio- 
 rity, we have thought it worth while to show that in this case at least 
 he has proved nothing except the laxity of his own Latin scholarship, 
 which he has elsewhere compromised, by asserting that C. Verres had 
 probably no gentile name; by saying that arcesso is derived from arcieo 
 in the same way as facesso is from facio; and by treating repetundce (which 
 never appears except in the genitive and dative plural of the attracted 
 gerundive) as a regular Latin substantive 1 
 
CHAP. IV.] AND OTHER PARTICLES. 365 
 
 it expresses that the existing state of things is at variance with our pre- 
 vious expectations. Here it implies that at the advanced point at which 
 we have arrived, we are enabled to detect a fallacy which had pre- 
 viously eluded our observation : so that after all^ the state of the case 
 appears widely different from our antecedent notion respecting it. In 
 this usage, it very often amounts to a declaration of regret and disap- 
 pointment, in accordance with our collocations "so then," &c. The fol- 
 lowing passages will make this sufficiently clear; Soph. Electr. 934 : 
 
 ώ δνβτνχήζ' Ιγώ δ\ ΰνν χαρά λόγους 
 τοιονύδ' εχονΰ' εότίενδον, ουκ είδνΐ' α ρ α 
 ϊν 7] μεν α τ η ξ' άλλα νυν, δθ' ίκόμην, 
 τά τ οντά τΐρόύΟ^εν, άλλα -9'' εύρίΰκω κακά— 
 
 "Ι brought the news not knowing, as it seems j in what misfortunes 
 we were all the while involved, but now, &c." Id. Ibid. 1175: ώ^ ουκ 
 αρ ^δη των εμών ούδεν κακών. Eurip. Troad. 414: ουδέν η 
 κρεί0(5ω των το μηδέν ην αρα^. Cf. Id. Ibid. 109. Herod, iv. 64: 
 δέρμα δε άνΟ'ρώτίου ην αρα βχεδον δερμάτων τίάντων λαμτίρότα- 
 τον λευκότητί, "so then it seems that after all the human skin is, what 
 we should not expect, nearly the whitest of all skins," meaning that if 
 we had not laboured under that error, tanners would have made some 
 use of our hides. Thucyd. i. 69, § 4: έλέγεΰ&ε άΰφαλεΐς εϊναί, ών αρα 
 δ λόγος του έργου εκράτει, "you were said to be safe persons, but it 
 seems that after all your character was better than the reality." This 
 usage ought to be restored to a very corrupt passage of JEschylus, 
 where we obtain a simple metre and an intelligible meaning by read- 
 ing (Choeph. 73 sqq.): 
 
 φοί ό' — άνάγκαν γαρ άμφίτίτολυν 
 Ό'εοΙ προΰηνεγκαν εκ γαρ οϊκων 
 πατρώων δουλίαν εύάγον αϊΰαν — 
 δίκαια καΐ μη δίκαι αρχάς βίου 
 πρέτίοντ άρ ην αΐνέύαι 
 %υκρον φρενών ύτύγος κρατούβτ:]. 
 
 The phrase τΐυκρον φρενών 6τύγος is given in the best MSS. and is 
 justified by Agam. 1279. The φερομένων of the MSS. is a corruption 
 of φρενών, βία of βίου ; and the missing άρ' ην were absorbed by the 
 first syllables of αρχάς which follows τΐρέτΐοντ in the manuscripts, 
 and of αΐνέΰαί, which must have followed the whole phrase in the 
 original text. Hence we have άρα as a term of sorrow. Soph. Aj. 
 
 * The phrase ην ccqa may be illustrated by the combination το τι ην 
 είναι, by which Aristotle so often signifies the formal cause. 
 
366 THE NEGATIVE [bOOK II. 
 
 1025 : πώς β' αποβπάόω τούδε κνώδοντος, νφ' ου φονεως αρ ε|ε- 
 πνενΰας, "by means of whicli, as it seems, (or to my sorrow) you have 
 been slain." Similarly in Latin, Hor. i. Carm. 24, 5 : ergo Quintilium 
 perpetuus sopor urget, "So then Quintilius is dead after all!" Cf. 
 Serm. n. 5, 101 (Heindorf, p. 376). From this we may see that 
 Dr. Cyril Jackson was not so far from the truth when he translated 
 Τρώες αρα, "the Trojans — God help them I ^^ 
 
 We have the negative root va- , with a weaker inferential sense, 
 in vVj vvv=na-va, ηα-να-Ίΐ, Latin nam, and the direct negative οΐ'-κ, 
 as we have before observed, appears in ovv=^na-va-n, with the same 
 ending, and nearly the same meaning as vvv. This last particle 
 always refers to something that has gone before; it takes up what has 
 been said, and continues it; so that the ground-meaning is still the 
 same , namely, that of farness. We shall see below (§ 202) that the 
 same explanation applies to the use of vvv, nun-c as particles of time. 
 
 193 There are still other modifications of the idea of distance or 
 progression, to which we will now advert. Words with much the 
 same signification appear in weaker or stronger forms in the Greek 
 language. Thus we have both ϊχι and είτα (comp. αντί and αντά), 
 αταρ and αϋταρ, αρα and αρα, εν and είν. The same is the case in 
 Sanscrit. We find pronominal roots ana and ena, ava and eva, ati and 
 eta, which have the same force, for they all alike signify distance. 
 Bopp is unquestionably right in identifying ετο and ati (Demonstra- 
 tivstamme, p. 16), which are to be considered as by-forms of είτα and 
 eta. A comparison of the Sanscrit and Greek enables us to explain 
 the Greek and Latin particles In, ην , ηνίδε, which have occasioned 
 much difficulty to scholars. That en and ην are identical can scarcely 
 be doubted, and we must also recognise the affinity between these 
 particles and the Hebrew ϊΐ5!ΐ and N5. The only question is, how 
 to explain ecce, ecquis in Latin, and ηνίδε in Greek. There is no 
 instance, so far as we know, of a change of η into c in the former 
 language, therefore ecce which is a synonym for en, cannot be a com- 
 pound of en and ce: and the same remark applies to ecquis. We 
 adopt, without hesitation, the suggestion thrown out by Pott {Etym. 
 Forsch. II. p. 138), that the first part of ecce, ecquis, is a pronominal 
 root analogous to the Sanscrit eta: comp. iccirco for idcirco, accingo 
 for adcingo, &c. : in fact , etquis is often found in old MSS. The pro- 
 nominal roots eta, ena, equally signify distance: they point to the 
 there as removed beyond some other point. Now this is the ground- 
 meaning of en and ecce: they are particles which are used to attract 
 the hearer's attention to an object distinct from him and the speaker, 
 and for this purpose a word strongly marking distance would natu- 
 
CHAP. IV.] AND OTHER PARTICLESc 367 
 
 rally be used, and the element of the second pronoun is appended in 
 the case of ec-ce, in order to mark the approximation or import- 
 ance of the distant object to the speaker or hearer. As Ιδεΐν de- 
 notes merely "to see," but Ideud'aL is "to view for oneself," "to 
 see with interest," "to gaze upon" (Henrick, Herod, p. 48), we 
 may understand why Idov is so frequently used as an interjectional 
 word. The particle ηνί is related to ijv as vvvi is to vvv, and 
 daC to df] (Pott, 1. c). Hartung supposes {Partikeln^ I. p. 273, note) 
 that ηνίδε is merely this ηνί with the suffix ^ε. Although this is 
 possible , a comparison of ην Idov , which also occurs , of the French 
 voila^ of the German sieh da! and of our "lo you there" (look 
 there), inclines us to suppose that this word is nothing but a com- 
 pound of ην Yds, "see there." This shows us too that the real 
 meaning of ην is that of ena; that in fact it is only a pronoun 
 like the Gothic ^aruh, "there," which is used by Ulphilas to 
 translate ίδον (Grimm, iii. p. 172). 
 
 194 To the idea of distance or progression some of the copu- 
 lative conjunctions are also due. The Latin ei, at are to be compared 
 with ETL•, Sanscrit ati; for εη, είτα, are used in Greek very much in 
 the same way as the copulative conjunctions. As a prefix we have 
 at- in the Latin at-avus, and in Sanscrit we have ati- in ati-matra-s, 
 "in excessive measure," ati=rdtra-s, "in the depth of night," &c. 
 That these particles contain the element va appears from the analogy 
 of εν and ανά (§§ 170, 183), and might be inferred from the Behistun 
 synonym u-td (Rawlinson, As. Soc. xi. 1, p. 80). But the affinity of 
 H-tL, εϊ~τα, to εν, ανά, is by no means limited to this Anlaut or initial 
 articulation. There is every reason to conclude that the two former 
 contained also the characteristic nasal of the two latter. For if s-tcI 
 and ob are by forms of άμ-φί and amb- (above, § 172); if the augment 
 ε-=άνά (§ 370); if α-τερ, compared with αν-εν, belongs to the same 
 class with the Sanscrit antar, old Persian atar, Lat. in-ter, German 
 un-ter (§§ 170, 204); then ^tl, είτα, Sanscrit ati, Latin ad, at, et, must 
 be classed with αντί, αντα, und, and. Nor is it at all difficult to see 
 the connexion in meaning. The idea of εν, ανά, English in, on, is 
 that of placing something away from ourselves. Hence the idea of 
 distance, and from this the use of av- {a-), Latin in-, as negative pre- 
 fixes. The idea of motion is conveyed by the affixed -g, -d, -τυ. Thus 
 we have είς=εν-ς, "unto;" ad=:and, "to or thereto;" αντί, ετυ, αντα, 
 είτα, "in front of" or "besides." As a copulative conjunction, the 
 form, which appears as ati, ετυ, et and at in Sanscrit, Greek, and 
 Latin, is retained complete in all the Teutonic languages, except the 
 Gothic and Scandinavian. Thus we have in old High German anti 
 
368 * THE NEGATIVE [bOOK II. 
 
 (with the variations enti, endi, inti, indi, unte, unde)\ middle High 
 German unde^ unt; old Fr. ande, and; mid. and new Netherl. en; 
 Anglo-Saxon and English and; and the same form appears, as a pre- 
 fix, in the Gothic and Norse languages also (Grimm, m. pp. 271, 2; 
 Grafif, Sprachsch. i. 362, cf. 352). In this fuller form the copulative 
 conjunction appears as a verb of motion in most of these languages, 
 which also give us the movable digamma, as in ava^=J^ava. Thus we 
 have Goth, vandjan, A. S. vendan, Engl, wend, went, by the side of 
 enden, Gothic andeis, "end;" and the 0. H. G. gi-want, 0. S. gi-wand 
 mean "finis," "terminus." Grimm (GrenzaltertJiumer , p. 6) suggests 
 that the old High German anti and andi, "end," are derived from 
 the particle and and denote the "end," as "ausserste Entdeckung in 
 Raum und Zeit." The original meaning then of the verbs enden and 
 wenden must have been "to go," "to go up," "to complete the act of 
 going," "to arrive at the turning-point." The Semitic languages 
 furnish us with a striking analogy in Jib:j "he went up," ^ΠΝ "he 
 came," manifestly connected with the particles b^ "up," and ΠΝ, 
 which not only means "with," but also denotes the definite object of 
 the transitive verb. The instances given above (§ 114 ad fin.) 
 make it more than probable that ΠΝ stands for nSN, and the 
 dagesh ϊιη;^ represents iinsy, in which the full form of αντα is 
 preserved. This latter particle corresponds in usage to the apo- 
 dotic av {Maskil le Sopher, p. 30). The Hebrew ly corresponds 
 both in form and meaning to the Latin ad, and the change of 
 breathing from Ν to y is not at all uncommon. We have discussed 
 this subject in general in the Transactions of the Philological So- 
 cietij, 1854, pp. 291 sqq. 
 
 195 The Latin language has not only the preposition ad and 
 the conjunction ei, — as a mere particle of addition and progress, — and 
 at, — as signifying in an adversative sense "still," "yet," "continuing 
 this state of things," — but also uses the enclitic particle -que as a 
 copulative conjunction, and combines this with ad or at in the form 
 atque or ac (compare neque, nee) to imply that there is not only an 
 addition , but also an intimate connexion between the things coupled 
 together, such as that of cause and effect. The Zend at-ga^=at'que 
 seems to be rather a compound of the same kind, than the neuter of 
 the pronoun a, with the enclitic conjunction appended, as Pott sup- 
 poses (Etym, Forsch. zw. Aufl. p. 253). The Greek language does 
 not use hi for et, still less εη τε for at-que, but it uses a form cor- 
 responding to que and the Sanscrit -cha, both as an enclitic and as an 
 independent word, both as an indefinite antecedent and as a relative 
 particle. We refer of course to τε and και, which are ultimately iden- 
 
CHAP. IV.] AND OTHER PARTICLES. 369 
 
 tical with κεν and «V, and perform corresponding functions. When 
 we compare οκ-κα, &c. with ο-τε, &c., and τεύβαρες^ τε, with the San- 
 scrit synonyms chatur, cha, and the Latin quatuor^ que, and remember 
 the connexion which subsists between cha and the interrogative stem 
 ka, and between the first part of cha-tur, and e-ka, EngHsh each 
 (comp. qua-tuor, and ce-quus), we shall find it difficult to deny the 
 relationship of τε to και*. Hermann, with a different view, has 
 endeavoured to establish the identity of τε, κεν (which he derives 
 from και), and jtov (κον) in the sense of "perhaps" (de Particuld «V, 
 Opusc. IV. pp. 4, 9 foil.) : their correspondence, in this respect, can only 
 be explained by the fact which he has overlooked, that they are all 
 forms of the indefinite pronoun, and are all used to convey that inde~ 
 finite idea of locality to which their meaning of doubtfulness is due : 
 (κά) κέ, κεν, bear the same relation to και, that δε, δεν, do to δαί, which, 
 like καί^ is never used as an enclitic. The distinction between τε and 
 και consists merely in this, that τε is an indefinite enclitic word, al- 
 ways placed after the word to which it refers, whereas καί is generally 
 placed before the words to which it belongs, and is used rather in a 
 relative than in an indefinite sense. Though we constantly find 
 τε — καί, we seldom havexat — τε: the former collocation answers pretty 
 jiearly to "somewhere — where." The use of the combination και 
 τε is of itself a proof of the relative power of καί, for τε is continually 
 found in immediate connexion with relative words, as og τε, Sg τε, 
 olog τε, ouog τε, &. (below, § 197). 
 
 196 The investigation of the use of copulative conjunctions, or ot 
 the connexion of sentences , is a question of syntax, on which we must 
 make a few remarks here, though it may appear at first sight to be in 
 some measure foreign to our leading object. When we wish to speak 
 of something that happened as subordinate to , or in connexion with, 
 some other thing that we are speaking about, we may express this 
 occasionally by a participle or infinitive mood ; but in the majority of 
 cases a greater degree of definiteness is required, and then we in- 
 variably call in the aid of some word of pronominal origin to connect 
 the two statements together. These auxiliary words are relatives, 
 whether they appear in the form of adjectives or of adverbs. The 
 correlation of two sentences is effected either by placing the relative 
 word in both clauses, as in the Latin qua — qua, Greek καί — καί; by 
 placing an indefinite in both sentences, as viri-que,foe'mince-que, avδQεg 
 
 * In the relative value of the copulative conjunction it may be worth 
 while to refer to L. Tobler's paper "On the relative use of the German 
 uncV {Zeitschr. f. vergl. Sprachf. vii. pp. 353 sqq.)• 
 
 Β Β 
 
370 THE NEGATIVE [bOOK II. 
 
 te ΟέοΙ τε; by placing the indefinite in the first clause and the relative 
 in the second, as in -τε — καί, or, by an inversion common enough in 
 Greek and Latin, putting the relative clause first, as in quum — turn; 
 or, finally, by putting a demonstrative in each sentence, as turn — turn, 
 which was probably the original formula (above, §. 148). The primary 
 mode of balancing sentences appears to have been by placing the 
 same particle in each clause in order that the similarity of sound 
 might help the ear. But, by the same process which led to the di- 
 vision of pronouns into definite and indefinite, a distinction was made 
 between the forms as stronger and weaker; the latter preceded, and 
 the office of uniting the two parts of the sentence devolved upon the 
 former. 
 
 Like the Latin -que and the Sanscrit -cha, the enclitic -τ ε is 
 often rather emphatic or introductory of an explanatory circum- 
 stance or apposition, than strictly copulative, as combining things in 
 themselves distinct and independent. In ^schylus, in particular, we 
 have observed that τε may often be rendered "namely," "that is to 
 say," "even," rather than "and." Thus to take the Agamemnon only 
 we have the following examples, 10: εκ Tqoiag φάτιν άλώΰίμόν τε 
 βάζιν, "a report from Troy, even the tidings of its capture;" 123: 
 εδάη λαγοδαίτας πομπούς τ αρχάς, "he recognised the devourers of 
 the hare, that is to say, the leaders of the expedition;" 210: Λανβ α- 
 νέμου ^υόίας, παρθενίου θ' αίματος, "a sacrifice to lull the winds, 
 to wit, the blood of a virgin;" 229: βία χαλινών τ άνανδω μένεις 
 " by violence , yea , even by the voice-constraining force of muzzles." 
 Similarly in the Eumenides, 690: "Αρει d' εΟ^νον, fVO'fV εβτ^ επώνυ- 
 μος πέτρα , πάγος τ "Αρειος, " they sacrificed to Ares , whence the 
 rock derives its name, that is, the hill of Ares." 
 
 In its usage the Latin enclitic -que corresponds to the Greek -τε, 
 but it bears more external resemblance to %αί, or rather to the %a 
 which appears in οκ-κα = δ-τε , πό-κα = πό-τε, εϊ-κα = εΐ-τε, Latin si- 
 qua. We find this older and more genuine form of the Greek enclitic 
 also in the adverbs αυτί-κα, πηνί-κα^ ηνί-κα, τηνί-κα, &e. Βορρ sup- 
 poses that the termination of these words is νίκα, αύτίκα being a 
 contraction for αϋτηνίκα. The termination -νίκα he compares with the 
 Sanscrit nish, nom. nik, "night," which is found in anisham, "for 
 ever," i. e. "without night" {uber den Einfiuss, &c. p. 10; Vergleich. 
 Gramm. p. 614). This is quite inadmissible. Nor can we approve of 
 Buttmann's supposition that the termination is -ίκα, the accusative of 
 an obsolete Ι'ξ, analogous to the Latin vix , vices (Lexil. ii. p. 227). 
 When we compare thewi/, νυνί, ην, ήνί, mentioned above, and remem- 
 ber that δήν is often used to signify time, and that αντί, αύτε, αύ%Ίς, 
 &c. occur separately, we must conclude that the termination is -κκ• 
 
CHAP. IV.] AND OTHEB PABTICLES. 371 
 
 When we recollect, too, that the Latin done-c, doni-cum, deni-que, 
 nam-que (nem-pe), quando-que, and the Greek ο-τε, πό-τε, το-τε, ευ-τε, 
 &c. are also particles expressing time, we are driven to the conclusion 
 that the last syllable of αντί-κα, πηνί-κα, &c. is also the interrogative 
 stem, used as an enclitic like the Latin -que. The same is to be said 
 of the termination of τετρά-κίς^ &c. The compounds οκως (ο-τΐως) and 
 ο-φρα must not be confused with one another : the former is a com- 
 pound of the ordinary relative with the indefinite, the whole being 
 expressed in the ablative case , while οφρα is the same relative with 
 the suffix 7ΐα-ρά, the jt being changed into φ either from the contact 
 with ρ or by a transfer of the initial aspirate. 
 
 197 Although our analysis of the original forms of the pronomi- 
 nal roots has led us to consider τε as belonging to the interrogatives 
 and ultimately to the relatives, we find that this enclitic is constantly 
 used in combination with the stronger relative forms og, και, 7to-, κο-, 
 OLOg, &c., and with the demonstratives ro-, τοΓο^, &c. We need not 
 wonder at this, for the stronger form is often placed before the weaker 
 in the languages with which we are concerned, especially when the 
 weaker form has by mutilation or otherwise lost to any considerable 
 extent its external resemblance to the stronger form with which it is 
 connected. The addition of the indefinite to the relative occurs more 
 frequently in the older than in the more recent Greek writers. The 
 effect of it is to give greater emphasis to the relative sentence. We 
 have before referred to the mode of expressing the correlation of two 
 sentences by putting the same particle in each. In process of time 
 different variations of this practice came in. One of the first would 
 be to place the two correlative words together (for instance καί-καί, 
 qua-quo)^ — the natural consequence of which would be to shorten one 
 of them — and then to cause its omission altogether. The sentence, 
 is tibi notus est ceque ac mihi^ might be written is notus est qua tibi, 
 qua mihi^ or is notus est tibi at-que (and besides) mihi. Another step 
 would be to put the indefinite or weaker form into the first sentence, 
 and the relative or stronger form into the other. An instance of this 
 is the common use of τε and και. If, however , it was desired to give 
 particular emphasis to the relative clause, the word which was found 
 in the antecedent clause would be repeated in the relative clause, though 
 the relative word alone was sufficient to express the general correlation. 
 Thus Iliad x\i. 9 : ηντε κονρη νηπίη, rfi' αμα ματρί %'εονΰ' άνελεΰ^-αΰ 
 άνώγει είανου α%τομενη και τε εόΰνμένη κατερνκεν. Xenophon, 
 Anab. ιι. 1, §. 22: και ημίν ταντα δοκεΐ ατίερ κ αϊ βαΰίλεΐ. Lastly, 
 just as the relative adjective is put alone in certain cases without a 
 formal expression of the antecedent, the antecedent particle also is 
 
 BB2 
 
372 THE NEGATIVE [bOOK II. 
 
 omitted; but to give the Ml definiteness required by tlie relative 
 sentence, not only the relative, but the antecedent particle is retained 
 in it. Of this usage of και we have an infinity of examples even in 
 the Attic writers ; of the separate use of r£ with the relative words 
 the instances in the Attic writers are comparatively rare, with the 
 exception of the combinations α-τε, ωζ-τε olog τε, and to these may be 
 added the analogous compound og-Tig, which is to be explained by an 
 application of the same principle. Olog τ εΙμι always means , in the 
 Attic writers, "I am able." In the older writers ol6gτε signifies "pre- 
 cisely such," and chiefly appears in comparisons. Thus Odyss.xxi. 173: 
 TOLOV — οΐόντε ρυτηρα βίον τ ϊμεναι καΐ οϊβτών. It is quite clear 
 that, though the meaning is idiomatically restricted to the idea of 
 ability in the Attic dialect, this notion of a comparison is the basis. 
 In that dialect the omission of the antecedent clause is more common 
 than in any of the other varieties of the Greek language. Such an 
 omission has taken place here, and oi6g τ είμι is equivalent to rolog 
 εϊμι>, olog τε ποιεΐν ταντα^ "I am the particular kind of person to do 
 these things," i. e. "I am calculated for the performance of such and 
 such things." In the same way olog και is also used. We have strik- 
 ino• instances of the definiteness thus communicated to the relative 
 
 ο 
 
 clause by the addition of τε, in οΰον-τε used in definitions of numbers : 
 e. g. Herod, in. 5 : εον τούτο ουκ ολίγον χωρίον, αλλ' ούον τε επί τρεΐg 
 'ημeρag οδόν. And in εφ' ω τε, which is used to express the terms of 
 a condition. Herod, i. 22 : ^ δοαλλαγή ΰφο εγίνετο, ε% φ τε ξεLVOvg 
 άλληλοίύί είναι καϊ ^υμμάχovg. Like ωύτε we find εφ' ω τε with the 
 indicative as well as with the infinitive: comp. Thucyd. i. 103, 113, 
 where we have the future indicative. We also find ετΰΐ τοΐΰδε, ωβτε 
 (Thucyd. III. 114). The same expression of definiteness is conveyed 
 by -τ ε affixed to temporal particles, such as ΛΟ-τε, ο-πότε, ο-τε, τό-τε, 
 εκάΰτο-τε, &c. And when the relative is used, even in Attic Greek, 
 with this temporal reference, the τε is sometimes appended, as in 
 ^schyl. Pers. 748 : εξ οντε τιμήν Zεvgεμolτήvδ' ώτίαοεν. Eumen. 25 : 
 ε| οντε Baz%aig εύτρατήγηΰεν &εόg. 
 
 198 The enclitic τε appears to be used for the relative in ε^ τε, 
 &c. ; the same is the case in the Latin us-que compared with 'iog ού, 
 μέχρι ου, &c. Τοι has no connexion with τε ; it is simply a case of the 
 third personal pronoun. Its perfect identity with the demonstrative 
 appears from the fact that και τοι and καϊ ταντα are synonyms. 
 English scholars generally translate the latter "and this too;" a 
 translation which applies to very few of the passages in which this 
 combination occurs. It is of course the proper rendering in those 
 cases where the pronoun is regularly declined as an adjective, as, for 
 
CHAP. IV.] AND OTHEB PAETICLES. 373 
 
 instance, in Aristoph. Plutus, 545 : 7ΖίΟ•άκνης τΐλενράν, ερρωγνΐαν καν 
 τ αν την (where see Dobree's note). In general, however, it corre- 
 sponds exactly to και τον and means "although," "and yet." For 
 instance, in JEschylus {Eumen. 113) it is used as a particle with a 
 finite verb: και ταύτα κονφως εκ μεόων άρκνΰμάτων ώροναεν. 
 When it comes after other words, rot is written as an enclitic. We do 
 not believe that it ever approached to the indefinite meaning which 
 is proper to enclitics , and its appearance at the beginning of a sen- 
 tence in the combinations τοίγαρ, τοιγάρτοι^ and even the use of 
 Tolvvv, prove that it ought not to be considered as a merely depend- 
 ent particle. It is strictly a demonstrative adverb. 
 
 199 The ordinary disjunctive in the Greek language is the simple 
 vowel η. Pott justly remarks {Etym. ForscL• n. p. 321) that the 
 number of words of different origin which this same vowel is used to 
 represent is a striking proof of the corruptions which have crept into 
 the Greek language. He enumerates the following: η=^αίΐ, Sanscrit 
 aha; 'η = 'ηα (eram), Sanscrit dsa; from the same root y=isiet, sit; η 
 interjection; ^ = Sanscrit sd; but η (^wce) = Sanscrit yd; and fj or 
 sjj = suce, which is quite different from y = cut feminine. This is suffi- 
 cient, he adds, to show that the Greek etymologer cannot expect 
 much success if he pays no attention to the cognate languages. Ac- 
 cording to the principle which we have often referred to, we are 
 justified in comparing ij and ην (in ηντε) with the Sanscrit disjunctive 
 va or vd. With regard to the substitution of η in Greek for va in 
 Sanscrit, Hartung (i. p. 217) properly compares svddu (suavis), ηδνς; 
 vdch (vox) , ηχή. The following instances belong to what we have 
 called the change of place of the digamma : vdmi, α^ημι, ανρα, ά^ήρ 
 or ηήρ, ηερ; α,Γώξ, ηώξ\ vdnkh, ενχομαι; to which we may add ηνς, 
 ενζ, ην, εν, compared with the Sanscrit synonym vasu (see Benary, 
 Berlin. Jahrb. Aug. 1834, p. 230). The Sanscrit vd also signifies 
 "like," "as," and in this it coincides with the Greek φή or φη = ώς, 
 which bears the same relation to η that φημυ does to η μι,. Buttmann 
 supposes that φή is connected with jcij, as φανός with jtavog, φάρβος 
 with pars, φαινόλης with pcenula, flagrum with πληγή {Lexil. i, p. 
 241). The same seems to have been the opinion of C. 0. Miiller 
 (Grimm, Deutsche Grammatik, in. p. 770). This leads us to the 
 second pronominal stem Jra, or, as it is more generally written in 
 Greek, ΰφε. The shorter Sanscrit form va corresponds to the Latin 
 -ve. That vel contains the root of velle, as Hartung (ii. p. 63) and 
 Pott (ii. p. 317) suggest, seems to be proved by the similar use of 
 heris in the Umbrian, as heris vinu, heris puni, "either with bread 
 or wine;" for this particle of choice is obviously derived from the 
 
374 THE NEGATIVE [bOOK II. 
 
 root An, "to choose," αιρ-έω, cf. ελ-εΐν, old Latin hir, "a hand;" we 
 have the imperative heri-tu in the Eugubine tables, and the same 
 verb occurs in Oscan (Varron. p. 92). This conclusion might seem 
 to be favoured also by the assumed connexion between the Hebrew 
 "iii "or," and the root ^i]N or uiN (of. ηϊΐΝ: and the Latin aveo) 
 "he desired." But there is no reason to adopt this etymology of TN; 
 which is much more likely to have been an extension oi ^=:ve or turn, 
 to which it bears the same or nearly the same relation as bii does to 
 jsb (Maskil le Sopher, p. 15). In this way we get back to the second 
 pronominal element of which 1 is a residuary form. With regard, 
 however, to vel, we must remember that it sometimes appears in 
 another sense of the verb vol-o, as when it means " for example," i. e. 
 "take this" (Plant. Miles Glor. v. 59). And its use, by the side of 
 its synonym libet, as an affix to the relative, in qui-vis, qui-libet, 
 seems to justify its application as an alternative conjunction. The 
 particle quam, which is the Latin representative of ^' as a particle of 
 comparison, stands in direct opposition to -vis, -libet^ as a relative 
 affix. The difierence between qui-vis or qui-libet and quis-quam is 
 simply this; the former means "any one selected from any given 
 number," "any one you please,^ so that all are included in the range 
 of choice; but quisquam^ like ullus , means "any one at all," the 
 selection not being supposed; in other words, quis-quam h exclusive*: 
 thus Seneca, de Tranquill. 11: cuivis potest accidere^ quod cuiquam 
 potest. That -piam may be identical with quam in signification, as it 
 is in origin, would appear from Cicero {in Verrem Actio, ii. Lib. i. 
 c. 10) : nego esse quicquam a testibus dictum, quod aut vestrum cuipiam 
 esset obscurum aut cujusquam or at oris eloquentiam queer eret. Prac- 
 tically the substantive quisquam, like the adjective ullus, is confined 
 to sentences which are formally or virtually negative. As in Ovid, 
 Metam.u.bS: placeat sibi quisque licebit; non tamen ignifero quisquam 
 consistere in axe Me valet excepto. It derives this meaning, however, 
 from that other use of quam in comparisons, in which it so exactly 
 corresponds to ^'. For quis-quam^ "any one at all," means "a person 
 of any manner or kind." Similarly per -quam is "in a very high 
 manner, kind or degree;" ne-quam is "in no manner or degree," i. e. 
 
 * It is a remarkable proof of the laxity of modern Latin scholarship 
 that all the recent editors, so far as we know, acquiesce in et latus Oceano 
 quisquam Deus advena junxit (Ovid, Fast. v. 21). It is manifest that quis- 
 quam is inadmissible here, and we propose to read quisquis, with the punc- 
 tuation, et latus Oceano, quisquis Deus advena, junxit, i. e. "whatever God 
 happened to come up." Cf. Plautus, Amphitr. i. i, 156: quisquis homo hue 
 venerity puynos edet. 
 
CHAP. IV.] AND OTHER PARTICLES. 375 
 
 utterly worthless; neuti-quam^ "in no way" (nuUo modo^ &c.) (Pott, 
 Etym. Forsch. zw. Aufl. p. 149). It is a mistake to suppose that 
 ali-quis can ever be rendered by the English word "any," or that it 
 is ever equivalent to quispiam^ as Heindorf supposes (on Hor. Sat. i. 
 4, 35, p. 95). All compounds with ali- (ali-quis, ali-quot, ali-quando, 
 ali-Gubi, &c.) are definite, and must be rendered by the English word 
 "some." So that aliquis approaches more nearly in meaning to qui- 
 dam than to qiiis-piam, which usually means ^^any one in general," 
 rather than ^'■some one in particular," which is the force of aliquis 
 and quidam. It is remarkable, however, that aucun, which must 
 have been originally aliquis unus, performs the same functions as 
 quisquam: for nan vidi quenquam might be rendered: je n'ai vu 
 uucune personne 
 
 200 There can be no doubt that the ij used in comparisons is 
 the same word as the disjunctive gj, and there is every reason to be- 
 lieve, on the other hand, that the disjunctive and comparative η (epic 
 ■ηέ, ήέτΰεο) = να is simply the second or relative pronoun. In Latin 
 the clause compared is connected with the clause on which it depends 
 by quam, a case of the relative, and the Sanscrit vd^ when it means 
 "in which manner," is clearly relative. In the same way waii, also 
 connected with the relative, is used in middle New German (Grrimm, 
 ΠΙ. pp. 183, 283). In Greek and Latin there are in fact two ways in 
 which the standard of comparison may be expressed. It may either 
 be represented by the case of ablation, i. e. the genitive in Greek and 
 ablative in Latin , as ουτοζ 6οφώτερός εΰτίν εκείνον, iste est illo sa- 
 pientior, "this man is wiser , stands in a higher grade of wisdom , in 
 relation to that man ; " or the standard of comparison is introduced by 
 the relative particle -^', quam, signifying "in the manner in which" 
 (like the German als, a mutilated form oi al-so)^ as άρείοΰυν ηετΰερ 
 νμϊν άνδράΰίν ώμίληβα, cum fortiorihus quam vos estis viris versatus 
 sum, "I have kept company with men brave in a higher degree, as com- 
 pared with the manner or degree in which you are brave." Our 
 than, only another way of spelling then, is more nearly connected 
 with the demonstrative, and expresses, like the Greek comparative 
 termination, that in the given relation that which is mentioned in the 
 second clause comes after that which is mentioned in the first: "Peter 
 in greater than John," i. e. "Peter is greater, then (comes) John." 
 Although relative words are not directly used to connect the terms of 
 a comparison in Greek, we see traces of the feeling which led to their 
 use in other languages, in such phrases as μείζων η κατ αν&ρωτίον, 
 ^'άύβονη ώς — , κακίων η ωό τ ε — , where the difference is expressed 
 as well as the likeness. 
 
376 THE NEGATIVE [bOOK II. 
 
 201 'Ή is found in immediate conjunction with μ,ίν, δε, δη. Ήμεν 
 
 — ηδε are not disjunctives, but copulatives, signifying "both — and;" 
 η in this combination is, therefore, to be compared with the use of 
 the Sanscrit vd for "as;" — "as in the first place — so in the second 
 place." Oi μεν — δε it is unnecessary to add any thing to what has 
 been already said (§§ 154, 155). They are generally and properly 
 correlatives, though sometimes τίλην, "farther," άλλο, "another," are 
 substituted for δε , to which their meaning is very much akin. As a 
 general rule, αλλά is opposed not to μεν, but to oi», just as sondern in 
 German expresses the opposition to a negative, and we often find 
 a sentence in which ovK , αλλά , are opposed , including two others 
 which contain an opposition of μεν, δε. Thus Eurip. Med. 555 : 
 
 ονχ, y 6v κνίζει — ΰον μεν εχΟ'αίρων λέχοζ 
 καινής δε νύμφης ίμερω τίετίληγμενοζ — 
 αλλ' ώ^ κ. r. λ. 
 
 202 "Ηδη is almost always used as an expression of time, but 
 this is not its inherent signification or the primary meaning of its 
 second syllable. Hartung, whose account of this word (Partikeln, i. 
 pp. 222 — 322) is very unsatisfactory, considers ηδη as prior to δη 
 (p. 245), refers all the meanings to a temporal one, and derives the 
 second syllable from the Sanscrit root signifying *'day" (pp. 223 foil.), 
 which is as objectionable as Bopp's derivation of the ending -vi -κα 
 from nisham, nox, or Pott's comparison of ya-di with dies. "We have 
 before shown that the idea of intervals or positions belongs to the 
 primary thought-form of space, and that the words by which this 
 idea is conveyed are pronouns, that is, words indicating position. 
 Such a pronominal word is δε , which, we have seen, indicates relative 
 nearness, and therefore enters into the second personal pronoun, the 
 second numeral, and a number of other pronominal words conveying 
 the same idea. That such a pronominal word should be used to 
 express "what is near to the now" as well as "what is near to the 
 here," is only what we should expect; and this is the full extent to 
 which the idea of time enters into δη and ηδη. We have before 
 pointed to the connexion between this root and the preposition δυά. 
 The idea of duration is derived from that of division or passing 
 through, and thus it is that diu signifies both length of time gene- 
 rally, and also the particular length of a day (inter-diu, compare inter- 
 dum, where the root retains its more general meaning); it is, there- 
 fore, more than probable that the Latin dies, the Sanscrit diva, dyu, 
 &c., are connected with this pronominal root, just as we have already 
 seen that other words of the same kind are manifestly of pronominal 
 origin ; but to say conversely that a word like δη, which enters into 
 
CHAP. IV.] AND OTHEE PARTICLES. 377 
 
 such a vast number of combinations, which very often has not the 
 slightest reference to time, and which is so obviously connected with 
 an extensive class of pronominal roots, is derived from such a second- 
 ary idea as that of "day," is opposed to all the principles of a 
 systematic analysis of language. Hartung supposes (i. p. 230) that 
 jam is connected with dies^ Sanscrit dyas^ on the analogy of Jovis for 
 Djovis, &c. It is clear, on the contrary, that jam is merely the loca- 
 tive of the stem fa , and bears the same relation to piam that ίάλλω 
 does to φίάλλω. It appears also as the dissylable 'iam. 
 ' Besides the combination ηδη we have also the phrase Tcal δη as an 
 expression of time; thus in Soph. Ajax^ 49: %al δη ' Ttl δίόΰαΐζ ην 
 ΰτρατηγίβίν Λνλαΐζ, "he had just got to the gates of the two generals." 
 Theocritus, v. 83: τα δε Κάρνεα καί δη εφερ^ει, "the Carnea are 
 just coming on." Whence the Gloss. Paris, ad Arist. Plut. 227: κ«1 
 δη' η δ η. From this primary signification springs the use of κ«1 δη 
 in a sort of supposition, as in Eurip. Med. 386: %al δη τε^νάβι, 
 "they are just now dead," i. e. "suppose them dead." Dem. in Aphoh, 
 p. 856, 16: %ai δη λέγει, "suppose that he does say so." 
 
 As δη corresponds in meaning to jam, which is traceable ulti- 
 mately to the same pronominal origin, we have exact equivalents also 
 in nun-c and vvv, vvvL It has been already mentioned that vv^ vvv 
 is an inferential particle, and that it derives this meaning from the 
 sense of farness and progress (§ 192). The same explanation applies 
 to vvv, nun-c as particles of time ; for it is felt that the present time 
 is an advanced point , in comparison with any preceding time. Hence 
 the best writers call present time to ετίειτα^ "that which supervenes." 
 For example, Sophocles, Antig. 605 : τό τ έπειτα και το μέλλον καί 
 το TtQiv, "the present, the future, and the past." Eurip. Iph. T. 1263: 
 τά τ ε τίρώτα τά τ ετίει^^ ο6α τ έμελλε τνχεΐν, "the first, and the 
 present, and all that was to happen." 
 
 In the temporal sense the element δ- appears also under the form 
 δήν, which generally signifies "long." The same word is found in 
 Latin, under the lengthened form den% which bears the same relation 
 to δήν, that τ^νί, νυνί do to ην, νυν; and denique might be written 
 δηνίκα, in conformity with αντίκα, ηνίκα^ τίηνίκα, and οτίηνίκα, though 
 it appears from a comparison of the synonyms τημος, demus (demum), 
 that τηνί-κα is the Greek form of deni-que, by a corruption similar to, 
 but not so great as, that by which the enclitic τε has been formed 
 from the regular stem of the second element, ζ/έ, δεν, δή^ δήν, are 
 related just as ^ε, μεν , μή, μήν; and it is singular enough that the 
 last word of each set has the most direct reference to time, and the 
 third words of each set are particularly opposed to one another, for 
 μη expresses a doubt, and δ^, confirmation. The opposition oi μεν. 
 
378 THE NEGATIVE [bOOK II. 
 
 μέ, δεν, δε, bas been interfered with by causes to which we have be- 
 fore directly referred. 
 
 The first part of the compound ηδη is the same as that of τιμέν, ηδέ; 
 also, we believe, as that of ημος, which has no immediate connexion with 
 ημαρ, ημέρα. In the other compounds δη stands first. Of δήτίς, and its 
 connexion in meaning with δεις, we have already spoken : δητοί is of 
 perfectly analogous formation: we may compare εϊτε, είτα; εηειχε, 
 ϊτίειτα', ενεκε, ένεκα, which fully explain the termination -τα. ^η&ενίΒ 
 generally used in a sort of ironical signification (:τρο(>:7Τθί'7^(9^ν αλ7ΐ^ϊείαξ 
 ϊχευ, δυναμιν δε φενδονς, Suidas) ; this signification is also generally 
 borne by d'rjv, which is obviously connected with Ο'έν as μεν with μην, 
 and ^ivwith δην. If we compare ενΰ'α, ενταύθα, εϊ^ ε, ά^Ό'α, with the 
 words before us , we must be convinced that these terminations differ 
 from the second syllable of δη^'εν only in the absorption or absence 
 of the locative ending ν (above, § 114). That there is some sympathy 
 between Ό" and y appears from χΰ'ές compared with hyas , and that 
 d'a is in fact a representative of the second personal pronoun is clear 
 from the forms OLu-d'a, %λν-^ι, &c. The ironical use may be easily 
 explained by the sense "only," which is nothing but a mark of position, 
 and which seems to pervade the Greek as well as the Latin words 
 into which this root enters: e. g. εϊ-^ε, "I wish that" = "if only!" = 
 " if in this particular." Compare the similar use of τίώξ αν. The ter- 
 mination of μέταζε is the same as O'e, &a : comp. Ζευς with ^εός. We 
 may class δηντε, δεϋτε, with ηντε, εντε, and χρώνμα, τρανμα. There 
 is no need to suppose, with ApoUonius and Hartung, that the first is 
 a contraction of δη and άντε, although such compounds were common 
 enough; take, for instance, τηνικαντα from τηνίκα. 
 
 The Latin set of words into which this root enters very often have 
 or w instead of the Greek ε, η. They use donee by the side of deni- 
 que, dum with δην (dum-taxat involves a verb; see Varronianus^ 
 p. 231), and -do as a termination in quan-do, which may be compared 
 with tada, "then," yadd, "when," kada, "when," in Sanscrit. The 
 coincidence in meaning between the Greek and Latin appearances of 
 this root is very striking : it is to be remarked, however, that in Latin 
 it is very often placed after the word to which it immediately refers, 
 while in the same case it precedes in Greek : thus δήτΐ£ answers exactly 
 in meaning to qui-dam. In tan-dem and deni-que we see that this 
 root may be placed before , as well as after , the syllable to which it 
 refers, and that too in synonyms. Another word of the same signifi- 
 cation is demum, which, according to Festus, was also written demus 
 (comp. rursum, rursus), and which corresponds exactly to the Greek 
 τημο£ opposed to ημος. The use of i-dem, tanti-dem, toti-dem, is just 
 the same as that of αντός δη, οΰον δη, οίον δη, &c. in Greek. All the 
 
CHAP. IV.] AND OTHEE PABTICLES. 379 
 
 meanings, therefore, of δη, dem, &c. may be deduced from the naturally 
 intensive use of a word marking location with emphasis. 
 
 In negations of time the Greek enclitics τιοτε and τίω correspond 
 in meaning to the Latin -quam and -dum respectively, and while 
 ovTtOTEj 7iun-quam mean "never," i. e. "at no time at all," whether 
 past, present, or future (cf. quis-quam above, § 199), ου-τίω, non-dum 
 mean "not yet," i. e. "at no time up to the present," "at no present 
 time," and we may also say ονπω Λοτε, "never yet," "never at any 
 previous time." While ποτέ has a complete family of correlatives 
 {τότε, "then," πότε, "when?" οπότε, "whenever," ποτέ, "at a certain 
 time"), πω is used only after a negative, with the exception of one 
 passage (-Slsch. Agam. 1507 : πώ; scil. civahcog ει (jv), where it is ob- 
 viously a synonym of jrtog, and in the combination πώμαλα, where it 
 bears the same meaning with a little more emphasis. For πώμαλα 
 really signifies, "how can it possibly be so?" i. e. "it cannot be so at 
 all," as the following passages will show. Aristoph. Cocalus ap. Har- 
 pocrat. s. v. : a. λοίδορία τΐζ εγενε^' νμίν ; β. πώμαλα' οίδ' είπον ονδεν. 
 "-4. Have you been quarrelling? Β. Certainly not (how could it be 
 so?). I did not even say a word." Id. Plut. 66 : Πλ. ώ τάν, άπαλλάχ- 
 ^ητον απ εμού. Χρ. πώμαλα. ^^ΡΙ. Depart from me. Ch. By no 
 means (how can you expect it?)." So also in Dem. Fals. Leg. 357, 
 and in the passage quoted by Suidas s. v. The grammarians, who 
 rightly render ττω^αλα according to its ordinary use by ονόα^ώ^, adopt 
 the statement of Apollonius {Bekk. Anecd. pp. 604, 622, 623) that 
 πώ is another form of πό^εν. This statement is combated by Mr. Shil- 
 leto {ad Dem. de fals. leg. p. 34), who, appealing to the Doric forms 
 αντώ, τοντώ, &c., considers πώ as a variety of πον. But πώμαλα be- 
 longs to Attic Greek, and while πώ bears the same relation to πώς 
 that οντω does to οντωξ, the true theory of the cases shows that the 
 adverbial ending -ojg is really derived from an original -οΰ^εν (below, 
 §§ 247 — 249). With regard to the meaning, Suidas is clearly right 
 in saying that πώμαλα is άντίτον πό^εν — οίον ουδαμώς. For example, 
 in the following passage we might substitute πώμαλα or ουδαμώς for 
 πό^εν; Arist. Ban. 1455: Ai. τί(5ι χρηταο, πότερα τοΐς χρηΰτοΐς; 
 z/t. πό&εν; μιΰεΐ κάκίότα. ^'■^sch. Whom does the city employ? the 
 good? Bacch. By no means — it hates them abominably." Mr. 
 Mitchell, in his note on this passage, well remarks that πόΟ'εν is like 
 the French comment, a civil interrogative instead of a direct negative; 
 and the same may be said of πώμαλα. 
 
 A great many of the uses of δη correspond to those of περ^ 
 which we have explained before as a word signifying "throughly," 
 "entirely," and thus it is that καί περ corresponds so exactly with 'εμ- 
 πας {εν πάΰι). The m, ν or vl, which is found in words from this root 
 
380 THE NEGATIVE [bOOK II. 
 
 is the mark of the locative case. The form -de which is found as a 
 termination in Latin {in-de, quam-de) corresponds to -%a in ϊν^α, &c., 
 and is another proof of the connexion of that termination with the 
 root of the second personal pronoun. 
 
 203 If we compare ye, γα, with κε, %a, we must feel disposed to 
 seek for some connexion between these forms; we shall now endea- 
 vour to establish by proofs our assertion, that such a connexion does 
 in fact subsist. The Latin qui-dem^ which corresponds to γΕ, or 
 rather to γε, δη, in almost every respect, and which is obviously a 
 compound of the relative stem quQ) and the de-stem which we have 
 just discussed, proves that there is some affinity between the mean- 
 ings of these concessive particles and that of the relative, which 
 indeed might be shown directly, for εϊ κε and siqua are as nearly as 
 possible equivalent to εϊ γε and si quidem. In fact, we believe that, 
 in Attic Greek, γε, which was much more frequently used by the 
 more modern than by the more ancient writers, actually stands as the 
 successor and substitute of κε, which never appears in that dialect. 
 In other languages of the Indo-Germanic family, the words which 
 correspond to γε are manifestly connected with the relative and 
 demonstrative stems. Thus in German we have ja as a synonym of 
 γε in its common sense of "yes," by the side οι jener ^ a synonym of 
 κείνος, the connexion of which with κε has been before pointed out. 
 We do not concur with Hartung in recognising the identity of άγαν 
 and the Sanscrit saha (i.p. 228); for the first syllable of the Greek word 
 is not the collective but the intensive a. The second syllable in the 
 Sanscrit word is a pronominal root intimately connected with the 
 relative, and is the same which appears in the first personal pronoun 
 a-Jia-m, which we have before explained: so that, in their existing 
 state at least, a-ha-m, ε-γω-ν, and α-γα-ν are only different forms of 
 the same word. It is remarkable that in Sanscrit there is a particle 
 hi which corresponds very nearly in its use to γε. That this is only 
 another form of ha cannot be doubted, and thus we have an additional 
 proof of the connexion of the relative with γε. We shall have further 
 remarks to make on this particle hi when we come to the verbal- 
 endings. Hartung's supposition (i. p. 352), that -ξα is a by-form of 
 άγαν and an older form of γε (p. 357), is not to be entertained; for we 
 think we have shown pretty clearly that ξα- is only another form of 
 δίά. The Anglo-Saxon gea, our "yea," is, like the German ja, another 
 form of γε, and indeed a tolerably near approximation to it. 
 
 204 That the common inferential particle γάρ involves the same 
 root as γε,γά, is sufficiently obvious, and scholars have generally 
 
CHAP. IV.] AND OTHER PARTICLES. 381 
 
 endeavoured to explain it as a compound of γε and αρα, just as γονν is 
 made up of γε and ovv. We have before pointed out the meaning of 
 the pronominal stem «-, av- or va, signifying distance, removal, &c., 
 and its use as a negative; also we have said generally that the termi- 
 nation -ρα indicates motion, and is consequently joined to the third 
 pronominal root τα-, to denote "beyond :" in which sense it is used as 
 a comparative ending. As an ending with this sense we find not 
 only -τε -Qog^ but also the simple -Qog, and we have not only αρα, but 
 also the simple ρα. Thus we have τΐα-ρά and tva-ra instead of tara 
 in the fourth numeral. Similarly, in the older English writers, we 
 have where as well as whether : e. g. "but where I be as true begot or no" 
 (King John, Act i. Sc. 1). With our modern where, signifying merely 
 the relation of place , we may contrast the Sanscrit ku-tra, and the 
 Latin trans, ul-tra, &c., in which the stronger comparative ending is 
 used without any immediate reference to the comparative degree. 
 As a general rule we have the shorter form in re, er as the sign of 
 the comparative in the Teutonic languages , so that ra may fairly be 
 considered as a comparative suffix. For this reason also, as we have 
 seen, words like λντί-ρόζ, μακ-ρός^ which have a comparative meaning, 
 throw off" the termination when they assume the comparative ending 
 -ίων. Thus too we have α-ρα = αν-ρα as well as α-τε-ρ = αν-τερ. It 
 is the merest trifling to say that ρα is a contraction of αρα\ it is found 
 in the oldest authors by itself, and has every right to be considered 
 as ancient as the compound into which it enters. Nay, more; it is 
 compared through its degrees ραον, ραβτοξ, just like μάλα^ μάλλον, 
 μάλίύτα. The words, in which it is found as a stem in the different 
 Indo-Germanic languages, show at once that its meaning must be 
 that which we have assigned to it. In Greek it is found in ρεω, 
 ράδιος, ρετίω-, &c., in Latin in rapio, rapidus, prefix re-*, &c., in the 
 old Norse ras^ Anglo-Saxon raes, Swedish rasa^ German rasch, rau- 
 schen. The difference in meaning between αρα and ρα appears to us 
 to be merely this, that in the former the notion of farness is combined 
 with that of motion, in much the same way, though more emphati- 
 cally, than in the comparative ending -τερος, while in ρα the idea of 
 motion, proceeding, &c. remains alone. It is for this reason that ρα 
 is never employed to express a direct inference, i. e. a going farther in 
 reference to, and in consequence of what has gone before, which is 
 
 * It is worthy of observation that re-, which corresponds to άνα- in 
 some of its uses as a prefix, exhibits a singular interchange of meaning 
 in the later as contrasted with the earlier Latinity. Thus while one word, 
 "re-veal," gives us the classical value of re-velo, our "re-cluse" gives us 
 the later, as opposed to the classical signification of re-cludo. See Casau- 
 bon's note on Suetonius, August, c. 78. 
 
382 THE NEGATIVE [bOOK II. 
 
 the common use of ccQa. We have before mentioned the intimate 
 relationship of the negative and inferential particles. It arises, as 
 before stated, from the idea of distance implied in the former. 
 Another form of ανά, namely, ανεν (-3Eolic ccVi-g), is used as a synonym 
 of α-τε -Q, which, as we have just remarked, is only a stronger form of 
 α-ρα. If we are right in our conclusion that cc-tsq was originally 
 av-tBQj it must be the same preposition as the Sanscrit an-tar, Latin 
 in-ter (above, § 170), German un-ter*, and thus α(ν)τερ will bear the 
 same relation to αντί that τΐρότερος does to προτί. That the ideas of 
 distance, separation, and evanescence, are cognate, is sufficiently ob- 
 vious. Compare, however, the uses of the Latin prefix se- = sed and 
 sine; and for the affinity of mier and unter, "under," we have the 
 cognate significations of inter- ficio and pessum-do, and of inter-eo and 
 per-eo. To say of anything that it goes, or is made to go, through or 
 between in a downward direction, or that it vanishes through the 
 floor, is a very simple description of its being destroyed or made 
 away with. In a similar manner , a slight modification of the second 
 syllable of ανεν, namely, vv, vvv, is used as a synonym of αρα, i. e. in 
 an inferential sense, just as nam, nempe=namque, are used in Latin. 
 The use of the second syllable of α-νά , under the lengthened form of 
 ri}, vai, as a strong affirmative particle, corresponds to the similar use 
 of δη, δαί, just as the strong form of vv , vvv , i. e. vvv corresponds 
 to δη when that particle is used to mark a point of time. 
 
 205 The hypothetical particle ει is intimately connected in mean- 
 ing with the relative. The meaning "if" = "in which" would at once 
 lead us to imagine a connexion between the relative and demonstra- 
 tive and the "if;" and there is at least one Homeric usage, in which 
 εΐ appears as the relative particle of time. This is the phrase εϊ ποτ 
 ϊην or εϊ τίοτ ϊην γε-, which occurs in the four following passages. //. 
 m. 180: δαηρ am εμος εΰκεκννώταδος, εΧηοτ ϊην γε, "he (i.e. Aga- 
 memnon, who is before her eyes) was my brother-in-law, shameless 
 woman that I am , at least when he was so (for he is so no longer)." 
 Ib.-xi. 762; ωξ εον, εϊ τίοτ εην γε μετ ανδράύιν, "such was I, when I 
 was engaged among warriors," which his extreme old age prevented 
 at the time when he was speaking. Od. xv. 268: Λατηρ δέ μοί εβτυν 
 Όδν6(3ενζ, εϊ ποτ ϊην ννν δ' ηδη άπέφ^Ίτο λνγρω όλέ-θ-ρφΙ," my 
 father is (was) Ulysses, when he formerly existed, but now he is dead." 
 Jb, XIX. 315 : olog Όδνόΰενς εΰκε μετ άνδράΰιν, εϊποτ ϊην γε, "such 
 
 *Thi8 view is farther supported by the Behistnn atar for inter \ see 
 Rawlinson, As. Soc. xi. i, p. 4. 
 
CHAP. IV.] AND OTHER PARTICLES. 383 
 
 as Ulysses was among men, when lie formerly existed." The context in 
 each case shows that there is no condition , least of all such a con- 
 dition as would be expressed by the past tense of the indicative*. So 
 that we have merely an emphatic and mournful reference to that 
 which no longer exists , as in the singular passage at the end of the 
 Ajax of Sophocles: Αϊαντοξ oV ην τότε φωνώ. But although we re- 
 cognise this temporal meaning in ft, we do not withBopp {Gloss. Sanscr, 
 p. 275) imagine an original form idi) still less with Pott do we sup- 
 pose that the Sanscrit ya-di involves the root of dw-a, dies. The use 
 of clearly demonstrative or relative words in the Indo-Germanic lan- 
 guages, as hypothetical particles, is an additional proof that ύ must 
 be a case of the pronoun Ϊ. This etymological fact is proved by a 
 comparison of the Sanscrit ya-di, "if," ya-di-va, "or" (si-ve), with 
 yat=-quod, ya-tas = 6%^ev, ya-da = or8'. also by the Gothic ja-bai, 
 Frisic jef, Anglo-Saxon gif^ Old German i-bu, ubi, Lettish Ja^ Finnish 
 yos, all signifying "if" = "in which or in that case or supposition." 
 The termination bai^ appears to be the same as the Latin bi, Greek 
 -φί, both locative endings (Quarterly Review , Vol. lvii. p. 105). It is 
 an important principle of syntax that the conditional clause or pro- 
 tasis is a relative sentence, the antecedent, when expressed, being the 
 particle av in the apodosis (above, § 186; Greek Grammar, articles 
 392, 501); and we thus see that the etymological result corresponds 
 exactly to the syntactical or logical deduction. 
 
 The only difficulty is occasioned by the appearance of the form al, 
 as an equally old particle with εΐ. "We might be inclined to suppose, 
 from a comparison of the ^olic forms κταίνω, φ^'αιρω, with the com- 
 mon forms κτείνω, φΟ'είρω, that the difference is merely a dialectical 
 one: but the forms κταίνω, &c., seem to have arisen from an insuf- 
 ficient perception for the finer shades of vocalization, which requires 
 that in heavy forms like these the lightest vowel should be employed, 
 whereas ai seems to be a really stronger and less degenerated form 
 than ει. That there is some difference between εΐ and «t, farther 
 appears from the fact, that Homer uses both forms, and with a slight 
 but decided difference of meaning , at being used principally in con- 
 nexion with γάρ, κε, and ^s, and not by itself as a mere ordinary 
 hypothetical particle. In the Boeotian dialect, we find, singularly 
 enough, that not only is η occasionally written £t, as in ονε^εικε for 
 ανέβηκε, μεί for μή^ ετίίδεί for ετΐείδή, &c. (Bockh, Corpus Inscripi. i. p. 
 720), but also conversely, ει is written η, as in η δε κά τΐξ [ηράτ]τ^ το 
 
 * The rendering si quidem olim fuit suggested by Nagelsbach , Anmerk. 
 z. Ilias^ Niirnb. 1850, p. 240, seems not to explain the phrase either com- 
 pletely or accurately. 
 
384 THE NEGATIVE AND OTHER PARTICLES. [bOOK II. 
 
 Ιννόμιον Ενβωλον (Bockh, Corp, Inscript. i. p. 741, No. 1569, 1. 48); 
 moreover, η is also a representative of «t (Bockh, Corp. Inscript. i. 
 p. 722) , so that no argument for the difference can be derived from 
 this dialect. K, however, we may read βαίκα, εάν Κρψες*, for the 
 gloss βαΐκαν, Κρψες], in Hesychius (see Kon, ad Greg. Cor. p. 114), 
 and compare the old Latin form suce, = uFai (sucepis = siquis; Miiller, 
 Etrusker, Vol. i. p. 31, note 62) with the other Latin forms si, se, sei^ 
 we may perhaps be disposed to conclude that the form al is derived 
 more from the stronger form of the second element /a, εΐ from the 
 mutilated form I', which, however, is ultimately traceable to the same 
 origin. This will account for the more common union of at with 
 relative particles in Homer, though, as we have seen, the weaker form 
 εΐ occasionally exerts an attractive influence on κεν , and even on av. 
 
 * This reading is due to Heringa (ap. Valcken. ad Herod. 350, 21). Is. 
 Voss conjectures that the true reading is βαΐζαν , αΐγα, and this reading 
 is adopted by Meineke. On the corrupted forms ulyav^ χεΐραν &c., see 
 Lobeck, Paralipom. p. 142. 
 
BOOK III. 
 
 THE NOUN. 
 
 cc 
 
THE 
 
 NEW CRATYLUS. 
 
 BOOK III. ff^r-T^ 
 
 J^thrary, 
 
 THE NOUN, 'V.Or 
 
 California• 
 
 CHAPTER I. 
 
 TEE ROOTS OF NOUNS AND VERBS. 
 
 206 Definition of a root. 207 The roots of words did not exist separately 
 and before the words in which they are found. 208 Quasi-monosyllabic 
 roots are really compounds. 209 The same remark is applicable to the 
 triliteral roots of the Semitic family. 210 Internal modifications of Indo- 
 Germanic roots. 211 I. Reduplication. 212 II. Prefixes consisting of 
 single letters. 213 Some of these are prepositions. 214 III. Assimilation. 
 215 (1) Assimilation proper and absorption. 216 (2) Substitution of ca 
 or ζ for a consonant and i. 217 (3) Introduction or interpolation of sym- 
 phonic consonants. 218 Digression respecting αμβλύς and κατηρτναώζ. 
 219 (4) Affections of the final consonant of the root. 220 Arbitrary 
 duplications of liquids. 221 Dissimilation and metathesis. 222 IV. Vowel 
 changes. (1) Weight of vowels. 223 (2) Adscititious vocalization. 
 224 Significance of roots. 225 Metaphysical and historical differ- 
 ences. 226 Dissection of words, in order to arrive at the root. 
 
 206 TN a language which, like the Greek, admits of inflexion 
 and composition without limit, we find in every word 
 that expresses a conception, whether it be a noun or a verb, 
 some prefix, suffix, or both, common to it and to a great num- 
 ber of other words, from which it essentially differs in meaning; 
 and, when these adjuncts are removed, there generally remains, 
 if the word be not a compound, some single syllable which 
 constitutes its meaning, and which again, with occasional slight 
 modifications, runs through another set of words, differing from 
 the one in question in prefix, suffix, or both, but agreeing with 
 
 CC2 
 
388 THE BOOTS OF [book III. 
 
 it in the fundamental signification. This ultimate element we 
 call the root^ or, if we may be permitted to borrow the termi- 
 nology of mathematical analysis, and apply it to philology, we 
 may say that every word is a function^ the root being the in- 
 dependent variable^ and the prefixes and suffixes the constants. 
 
 207 When we talk of the roots of words, we do not mean 
 to say that words are derived from them, or that they ever 
 existed separately*. If we did we must fall into the absurdity 
 of deriving all languages from a few primitive syllables, an 
 absurdity for which Murray has been so justly derided. Like 
 the common parts in things generically the same, they are 
 created by our powers of abstraction and generalization, they 
 have only a subjective existence, and to speak otherwise of them 
 would be the excess of realism. Everything is conceived as 
 existing or happening in space or time, and therefore, as has 
 been shown, the element indicating the conception must always 
 have, subjoined to it, some element denoting position, that is, at 
 least one pronominal stem, before it can be considered as a word.. 
 That any hypothesis of the separate and primary existence of 
 roots must lead to the merest trifling, is clear from the absur- 
 dities into which Lennep and Scheide have fallen, in their 
 attempt to carry out Hemsterhuis's principle, that the primitive 
 verbs consisted of two or three letters, from which the complete 
 words, as we have them, were formed. It is, of course, of the 
 utmost importance that we should analyze and compare words, 
 so as to arrive at their primary elements, just as it is necessary 
 that the philosopher should seek for the real definition ; but 
 there is no more truth in saying that the bare roots, which 
 form the materials of inflected language, ever existed sepa- 
 rately, than there would be in asserting that the world was 
 once peopled with αντοέκαΰτα, whose fossile remains, forsooth, 
 the geologers have as yet failed to discover. 
 
 * The subject of roots has been very ably discussed by Pott in his 
 review of Benfey's Wurzellexikon (Juhrbucher f. Wissensch. Kritik, 1840, 
 pp. 623 sqq.). He says with great truth (p. 631) that a root is only a 
 figurative expression, and that it is merely the germ of a family of words, 
 without being a word itself. 
 
CHAP. I.] NOUNS AND VEEBS. 389 
 
 When we thus deny the separate existence of roots, it may- 
 be objected to us, that some languages, the Chinese for instance, 
 are entirely made up of naked roots. But then it must be 
 recollected, that these roots are mutilated words which have in 
 all probability lost their original inflexions, and that we are not 
 speaking of tertiary idioms in which there is no such thing as 
 flexion or etymology, but of perfect languages like those of the 
 Indo-Germanic family, which are based upon monosyllabic 
 roots, adapted for composition, and only appearing in connexion 
 with at least some one pronominal element*. 
 
 208 Many syllables terminating with a consonant are called 
 roots of Indo-Germanic words. It must be recollected, however, 
 that no consonant can be pronounced without a vowel, and that 
 every such final consonant of a root was originally a distinct 
 syllable ; so that all roots terminating in consonants must be 
 considered as dissylabic, and, therefore, as compound roots. In 
 such compounds not only is the second vowel suppressed, but 
 also, in some cases, the first, and thus too there are apparently 
 monosyllabic roots ending in a vowel, which are nevertheless 
 dissyllabic (see Lepsius, Palaographie^ p. 65). An instance to 
 the point is fiirnished by the root κτε-, the first vowel of which 
 is never inserted in Greek, though it appears distinctly enough 
 in the Semitic synonyms bt^p, aap, iai*, JJai*, v,;^iai', &c. When 
 the second consonant is a liquid, nothing is more common than 
 the metathesis of the vowel, according to the principle men- 
 tioned above (§ 107). We have an example in the root yfv-, 
 which sometimes appears as γνε-'. compare ε-γεν-ό-μην, γί- 
 γνομαι; genitus^ gig^o; and the Sanscrit jan, jajnati. In the 
 form γα- of γε-γαα, &c. the ν has evanesced, according to the 
 common practice (above, § 114). It will be understood, then, 
 that when we call ^ακ- the root of δάκ-ν-ω, tvTt- of τντί-τ-ω, 
 φνγ- of φενγω^ Λραγ- of τίράΰΰω, κτε- of κτείνω^ &c., we are 
 
 * It is right to mention that, while the true theory, as we consider 
 it, has been carried too far by Bopp, other modern philologers have en- 
 deavoured to find a philosophical foundation for what is in fact a repro- 
 duction of the antiquated hypothesis (see for example Buns en, Brit. Assoc. 
 1847, p. 293;. 
 
390 'J^HE ROOTS OF [book III. 
 
 not speaking of the ultimate element of these words, but of that 
 root, which, though a secondary one, forms the base of a large 
 class of words, and of which the real ultimate elements are not 
 known or not taken into the account. It is these quasi-mono- 
 syllabic roots which it is so difficult to recognise in consequence 
 of the changes which the terminations produce upon them. 
 
 209 It is usual to speak of the triliteral elements of the 
 Semitic languages as roots or ultimate forms of words. On this 
 subject we have only to repeat what we have elsewhere had 
 occasion to state (^Maskil le Sopher^ pp. 36 sq.), namely, that 
 if, assuming, as a general view of all languages entitles us to 
 assume, that the languages of the Semitic tribes were originally 
 endued with a living power of flexion from single significant ut- 
 terances, we were to inquire what process of mutilation would 
 most naturally ensue from the early use of writing, and those 
 ethnical admixtures to which we have adverted in a former 
 chapter (§49), we could hardly imagine any other state of things 
 than that which is exhibited by the Hebrew language. We 
 observe here a regular preference for words of a certain length, 
 and we must suppose that this craving after a symmetry of 
 rhythm could only have arisen, when the gradual loss of a living 
 etymology had reduced compounds and derivative forms of words 
 to the mere units or terms of a logical proposition. Under these 
 circumstances, it is probable that euphony would take the place 
 of any previous feeling for the significance of the constituent 
 parts of a word ; and in the pollarding process, which was gra- 
 dually introduced, the parts omitted were often at least as sig- 
 nificant as those which were retained ; for when the etymological 
 value of a language is at a discount, all syllables are alike except 
 so far as they minister to or oppose a conventional euphony. 
 To the same obtuseness of etymological discrimination we must 
 attribute another phenomenon, which is very observable in the 
 Hebrew language, namely, that different meanings of the same 
 word, from being at first indicated by mere modifications of sound, 
 are ultimately represented by different words. The same euphony, 
 however, which leads to the mutilation of more complex words, 
 also obliges us, in certain cases, to elongate, or otherwise make 
 the most of simple crude-forms; and thus it happens that the 
 
CHAP. I.] NOUNS AND VERBS. 391 
 
 Hebrew language often preserves the truest and fullest forms 
 of certain important roots, or quasi-roots, which it has in 
 common with the Indo-Germanic idioms. We will give a 
 few examples of these linguistic phenomena, which are very 
 instructive with reference to our immediate object, (a) We 
 should not at first sight recognise the identity of Ti^h and 
 μαν^άνω^ but it is demonstrable. In its pi^'hel^ or causative 
 form, niab, the Hebrew verb signifies "to teach," or "to cause 
 learning to pass on to another;" but this leaves the inference, 
 that in its primary use it signified "to learn." Supposing, 
 then, that the first syllable is the preposition b=am, the 
 extensive use of which we have already mentioned, the mono- 
 syllabic root of the verb is 173, which is found with the same 
 meaning in μα^-εΐν^ μηδ-ομαί, μηδοζ, μψις^ 7tQoμη^'εvg^ &c., in 
 the Latin med-itari^ and in the Sanscrit medhd. In the word 
 n7?b, "he took," which we at once identify with the Greek 
 λαχ-εΐν^ euphony has called into play the residuum of some 
 pronominal adjunct, analogous to that which supports the 
 conjugation of λα-γ-χά-νω^ and in some of the forms, e, g. the 
 imperat. Π]? , this affix is retained at the expense of the radical b. 
 If we compare ydovjtog with dovTCog^ xrvTCog with τυτΐτω, κρηνη^ 
 XQOvvog with ρέω, &c., we shall probably recognise in these prefixes 
 a remnant of the preposition εκ or εξ. The same may have been 
 the case with bi'l, "great," compared with ooylt^og, Russian 
 dolgye^ &c. ; but the Hebrew euphony employs the prefix as 
 equally significant with the root of the word. Again, the very 
 common verb ^ni, "he gave," must be considered as a compound 
 of the preposition 5=b and the root in found with an affix in the 
 synonym tiin, "porrexit," and in the Indo-Germanic τείνω^ 
 ταννω^ teneo^ tendo^ Sanscrit tanomi^ German dehnen, which again 
 are formatives from a root τα- stiU found in Greek. The im- 
 perative 1?! exhibits the simpler root without any prefix, and the 
 construct-infinitive nn entirely disguises the common or elongated 
 form of the Hebrew verb. In b^-i we have the same root as in 
 fallo^ "fall," but, as in the Greek ύ-φάλλω^ the prefix is in- 
 separably connected with the root. The same remark applies to 
 Nl"5, Na"5, which may be compared with φψμΐ', fari, fatum^ 
 vates^ 7tQ0-(p'riT'rig. In yn-b, yt!"^? ^^^ ^^''h ^^ have the same 
 root yn, which is also found in the Latin quatere, the German 
 
392 THE BOOTS OF [book III. 
 
 quetschen, and in our quash, s-quash, s-queeze. The preposi- 
 tional affixes, of which the first and third must be considered as 
 identical, modify very slightly the signification of the compounds, 
 and yet, according to the Jewish grammarians, they are three 
 separate roots, (b) This brings us to the second class of pheno- 
 mena. An example of this is furnished by the root md-.^ which, 
 in Latin, denotes "to divide" and "separate" (as in di-vido, 
 idus, vid-uus) ; hence "to distinguish" or "to see" (video). We 
 have the latter sense in the Greek Ιδεΐν=Ηδεΐν^ and an exten- 
 sion to the sense "to know" in olda. Now in Hebrew we have 
 b^S, "he divided," and by the side of this the entirely distinct 
 word 5>-5:;, "he knew," containing the same root slightly dis- 
 guised. Then, again, there can be no doubt that the important 
 verb 3>ΐώ-^, "he was open and unconfined," contains the same 
 root as ^-^ώ, "he opened," and b-i^'a, "hollow," κοΓλο^. And we 
 must suppose that ϊιν'ί, "he saw" (δ-ράω), is really connected 
 with JiiJ'i, "he looked after a flock of sheep, went about with 
 them, and tended them," the more immediate analysis of which 
 leads us to s^'n (cf. erro, ερρείν^ and the roots ρα or ρε/); and 
 we find a further transition in '^J';}"'], "a road" (Maskil le Sopher^ 
 p. 40). (c) The cases in which the Hebrew euphony has pre- 
 served the fullest forms of Indo-Germanic roots or quasi-roots, 
 are very numerous. A few specimens must suffice. By the 
 side of κ«λέω, κελομαί^ κελεύω^ κλνω^ κλέ /og, &c., calare, cliens, 
 in-clytus^ &c., we have the Hebrew bip or biip^, which shows 
 us that the initial must have been originally a compound of 
 guttural and labial — in fact, the sound / or q. The same 
 inference might, perhaps, be derived from the Anglo-Saxon 
 gyllan and onv yell: and we are thus enabled to pass on to 
 the connexion between N'lp^, κρά-ξω^ κηρνΰΰω^ γηρνς^ &c. 
 Again, we have probably the more original initial in ^ί):, com- 
 pared with κεραξ=κερεν-τ (above, § 114), for this q sound is 
 necessary to explain the ο in cornu, horn (Goth, haurns) : see 
 Varron. p. 250. The Latin p-recor^ the German f-ragen^ 
 Sanscrit p-rich-chhami^ &c. exhibit the p-r in close proxi- 
 mity, while rogo omits the labial, and posco for proc-sco 
 nearly annihilates the root-syllable; but the Hebrew χ\"^ ex- 
 hibits the root rek or reg- with its prepositional prefix complete, 
 and carries us at once to the original idea of the word, namely, 
 
CHAP. I.] NOUNS AND VEEBS. 393 
 
 approach and supplication (cf. ίκε-της). From these examples, 
 which might be multiplied to any extent, we see, on the one hand, 
 that Hebrew words might be reduced to monosyllabic roots, 
 like those which form the basis of the In do-Germanic languages*; 
 and, on the other hand, we observe that if the quasi-monosylla- 
 bic roots of our family were liable to the same extensions which 
 we see in the Semitic languages, we should recognise the fact, 
 that they are compounds, and should seek their explanation in 
 a further analysis of the elements of which they are composed. 
 
 210 Although therefore the distinction between the Indo- 
 Germanic and Semitic languages consists principally in the 
 fixity of the compound crude-form, and the more frequent in- 
 ternal modifications to which it is liable, in the latter family, yet 
 we must not suppose that the roots in the Indo-Germanic lan- 
 guages are always simple and incapable of internal modification. 
 Every root consists either of a vowel, or of at least one con- 
 sonant and a vowel. Now if we alter the consonant or conso- 
 nants we alter the root, as far as its immediate identity is 
 concerned, for we are not speaking here of cognate roots as li 
 and ri; but the vowel of a root containing consonants admits of 
 many modifications, and even when the vowel constitutes the 
 whole root, it is liable to a few regular changes. The reason 
 is, that the vowel is merely a voice or sound produced by a 
 greater or less opening of the mouth, and entirely dependent on 
 the consonant or breathing which it articulates, whereas the 
 consonants are all produced by distinct intentional motions of the 
 tongue or lips, or both, and therefore constitute the intended 
 meaning of the word. The dijfference between composition with 
 constant prefixes and suffixes, and modification of the root, be- 
 tween the external and internal modifications of the significant 
 element, consists in this, that the former being pronominal 
 
 * The existence of monosyllabic roots in the Hebrew language has 
 long been maintained: see Adelung, Mithridates^ i. pp. 301, 2; Klaproth, 
 in the Baron de Merian's Principes de Ρ etude comparative des Langues, pp. 
 212 sqq.; Fuerst, as explained by Delitzsch, Jesurun^ Grimmae, 1838; and 
 others. '' 
 
394 THE BOOTS OF [book III. 
 
 additions can only affect the expression of local relations of 
 the thing intended, whereas the latter expresses a modification 
 of the thing itself. As the roots in the Indo-Germanic 
 languages are ultimately monosyllabic, the internal modi- 
 fication must of course be of less frequent occurrence and 
 less various than the external. 
 
 The changes which it will be proper to notice in this chapter are, 
 (I) reduplication of the radical consonant, or, in some cases, of the 
 whole or greater part of the root; (Π) prefixes consisting of single 
 letters; (III) in the case of quasi-roots, or roots ending with a con- 
 sonant, assimilation produced by contact with the pronominal part of 
 the word ; (IV) modification of the root-vowel. 
 
 211 (I) Between the roots of nouns and verbs as such, there is 
 no essential difference. The verb and noun are distinguished by the 
 constant pronominal endings which are attached to each, and occasion- 
 ally by the manner in wliich these endings are connected with the root. 
 Reduplication, however, takes place in the verb much more frequently 
 than in the noun, and with reason, for the intention of the reduplica- 
 tion is to convey an idea of repeated action, which is of course more 
 frequently applicable to a verb than to a noun. It will be understood 
 that the reduplication, of which we speak here, is essentially different 
 from that which we have pointed out in the pronominal word desig- 
 nating the number "six." In that case the pronominal word, denoting 
 position, and as such representing a certain number, signifies, when 
 repeated or written twice, the double of that number, just as the word 
 thus reduplicated itself consists of two pronominal words, each repre- 
 senting a number. But the reduplication of the root is merely an 
 internal modification, intended to enable the sense to dwell longer on 
 the word, and as such does not differ in kind from the guna or anus- 
 vdra, of which we shall speak in this chapter. 
 
 212 (Π) Many roots in the different Indo-Germanic languages 
 have occasionally single letters prefixed, which do not belong to the 
 root, for it generally appears without them, so that even when these 
 letters are vowels we are not to conclude against the assertion that all 
 proper roots are monosyllabic. The following are some instances (see 
 Quarterly Beview^ LVii. p. 107). 
 
 1 a;=e=o prefixed. 
 
 a-^TOOS Ιβλ,χρόί,μάΐ 
 
 I μαλκκοί 
 
CHAP. I.] NOUNS AND VERBS. 395 
 
 ά-μανρόζ- • • "• μανρός 
 
 {Latin mulgeo 
 Germ, melken 
 Eng. milk 
 {Old Lat. nero 
 Sansc. narah 
 Zend nairya 
 
 a-JtOLVCi Λοίνη 
 
 α~<57ίάραγοξ ότΐαργή^ ΰφρίγώ 
 
 ά-λείφω λίτία 
 
 a-perio porta 
 
 , / ί Latin stella (for sterula) 
 
 ,Γχ. τ'?, χ ^• ? • < Zend stero 
 
 (Emh German) ge-stirni ^ 
 
 ^ ^ ^^ VEng. 5i«r 
 
 ε-κεΓι /os κεLVog 
 
 ' λαόζ 
 
 ϊ-λζν%ίροζ 
 
 Germ, leute 
 Sclav, liudi 
 ^ Anglo-Saxon leod 
 εμε, εμοί Lat. me, mihi 
 
 ( Eng. ruddy 
 
 ε-QV&Qόg { Lithuan. ruddus 
 
 [ Lat. ruber, rufus 
 Sanscrit k-ship\ ^, 
 
 l-QliTtHV ] P''''^"'' 
 
 8-Qετμόg Lat. re-mus 
 
 ε-νεροι νερ^ε 
 
 ε-ρενγω Lat. ructo 
 
 ε-λαννω Idmma 
 
 ε-λαχνξ Sansc. laghus 
 
 ε-ρεφω Eng. roof 
 
 o-βρLμog βρL^'όg 
 
 6-φρvg Sansc. bhrus 
 
 ' ^ ' / ' ^ ' \ ί Lat. dens 
 
 o-oovg {o-ooi/Tg) - - - - { ^ 
 
 [ bansc. dantas 
 
 6'βελog βsλog 
 
 6-δάξ δάκ-νω 
 
 » ^ \ Lat. nomen 
 
 ο-νομα { ^ 
 
 \ Sansc. naman 
 
 o-xlog τλημι 
 
 6-ρεγω Lat. rego 
 
 b -νύξ I 
 
 Sansc. nakha 
 Germ, nagel 
 
396 THE ROOTS OF [book III. 
 
 o-vsidog Sane. nid,^to despise" 
 
 6-κρνοείς κρύος 
 
 2 Labials. 
 
 Eng. b-reak \ 
 
 Icel. b-raka > . . . . ρηγ-νν-μι 
 
 Lat. f-ra-n-go j 
 
 Eng. b-road Λ 
 
 Germ, b-reit !-.... Lat. latus 
 
 Greek τί-λσχνξ j 
 
 Germ, f-ragen Λ 
 
 Lat. p-recor > . . . . Lat. rogo 
 
 Sansc. 'p-racKh J 
 
 Lat. f-luo \ ( Lat. liio 
 
 p-luo i I λούω 
 
 Lat. p-iscis \ /-« τ . 
 
 „;. 1 if } (jaelic tasg 
 
 Welsh p-ysg ) '^ 
 
 3 Dentals. 
 
 ,,j f Germ, reiben 
 
 '^■P'^"" tEng.ru* 
 
 d -άκρν 1 r Sansc. agru 
 
 (l-acryma) t I Lithunan. aszara 
 
 ^ r ί Lat. ros 
 
 I Sansc. r«5« 
 
 d-QbTtG) Germ, rupfen 
 
 δ-ερ-κω i , 
 
 Sansc. d-npai/ ^''^^^ 
 
 4 Gutturals. 
 
 i'-i^c? Lat. Icetus 
 
 High. Germ, ^e-siirw . . . | ^''^• *^^'' 
 I Zend stero 
 
 γ-λψη 1 Γλαω 
 
 γ-λανκόξ \ J λευκός 
 
 γ-λαύΰύω J I λεΰίίίίω 
 
 γ-λάγος ^ i Zacfi; 
 
 Sansc. A:-am Lat. amo 
 
 κ-λυτο^ /oMci 
 
 Lat. g-lubo Lettish lobit 
 
 Sansc. g-rabh\ 
 
 Icel. g-ripa f ^^*• ''«i'*^ 
 
CHAP. 
 
 I. NOUNS AND VEEBS. 397 
 
 Υ orkshire dialect c-lubstart Norfolk dialect lobster 
 (=zclubtail, a name for the stoat) 
 
 κ-νίΰβα nidor 
 
 χ-λιαρόζ λιαρός 
 
 χ-λαΐνα Icena 
 
 g-raf reeve 
 
 C-lanius Ital. Lagno 
 
 H-lodwig Ludwig 
 
 k-nut Lat. nodus 
 
 , ι Lat. aiper 
 
 ^-«^^^^ { Germ, eber 
 
 5 s, 
 
 ύ-κετΐαρνον (q.xs τταρ- λ 
 
 νον, Od. V. 237) >. . κόπτω 
 ΰ-κάπτω J 
 
 β-τελλειν τελλευν 
 
 β-τολγ] άνα-τολή 
 
 Lat. sorbeo ] r / 
 
 Litn. s-reb-jul 
 
 ΰ-τενω • . τείνω {δνζ-τψοξ}) 
 
 6-κάμανδροζ κάμανδροξ 
 
 ίκεδάννυμι 
 κεάζω 
 cut 
 
 s-masJi mash 
 
 s-lash, s-p-lash lash 
 
 s-pike pike 
 
 s-cocp cooper 
 
 , -. , ( plenus (comp. candidus with 
 
 s-nlendidus [^ ^. ^ 
 
 \ canus) 
 
 s-pcnte pondus 
 
 s-queeze quetschen 
 
 213 In many of the vowel protheses it is easy to see that they 
 correspond to the elements of prepositions. Thus in α-μεργευν, 6-μόρ- 
 γννμι^ cc -μάω, ά-μερδειν, ά-μείβευν {=^άμενειν^ άμεν(3α6^αυ, Pindar, 
 Pyth. I. 45 ; Hesych. s. v. αμενβίτίοροξ ; Find. Pyth. xi. 38 ; a-moveo)^ 
 ά-μαρνΰΰειν (=άπο-(3τίλβείν, comp. μαρ-μαρ-ν-γη), α-Λοινα, α-6κοζ, 
 α-ύ%ί{δ)ξ (comp. 6κντοζ\ &c., the prothesis is obviously the first part 
 of the preposition ά-πό, or the pronominal element να, which, we have 
 seen, is frequently used to express removal, distance, and, by inference, 
 negation. The prefix seems rather to bear the meaning of the compound 
 α-νά in α-6πάραγοζ ("that which sprouts i^p"), α-ΰταχνς, ά-λείφω 
 (comp. άναμάΰΰω), 6-φρν£, &c. And we have clearly o- for ανά in 
 
398 THE BOOTS OF [book II. 
 
 ο-ΰκάπτω, ϋ-ΰταΰαν,6-6τα%'είς^ ο-μνάΰ&ψ (Ahreiis, de dial.^ol. pp. 
 77, 149). The same is probably the true explanation of 6-πνίω, com- 
 pared with φνίω, φύω, fac -ίο, fi-o, ποί-έω, τίών, ποί-μήν, pu-er, pu-bes, 
 Sajiscr. pu-tra, &c. In J^a-vriQ•, as it was written in Homer, we have 
 the older and fuller form va of the element a, which, as we have before 
 suggested, is probably in this case a mutilation of the second pronominal 
 stem. The initial a has a collective meaning, or stands for sa, in a-oTca- 
 ξομαί, ά-κόλου&θ£, &c. But many of the vowels, and most of the 
 gutturals and sibilants thus prefixed, are only added for the sake of 
 euphony. Indeed this appears to be invariably the case with the short 
 e prefixed to a word beginning with s and a consonant, as in a great 
 number of French words; e-space=spatium, esprit =spiritus, e-stomac 
 ^istomachus^ e-cuyer, old French escuyer, Ital. scudiero, Latin scutiger 
 scutarius, in English both Esquire and Squire*. In the same way Bar- 
 tolomeo della Scala, of Verona, becomes Prince Escalus in Borneo and 
 Juliet. In this country we have a habit of prefixing such a vowel in 
 pronunciation, whether we prefix it in writing or not. Thus many people 
 pronounce Xerxes as if it were written Exerxes , and the people of 
 Asia Minor, and others to whom the Persian was a foreign language, 
 seem to have been driven to the same use of a prothesis in articulating 
 the designation of the provincial governors. The word ΰατραπηΐνρ 
 defined by Herodotus (i. 192) as η άρχη Tfjg χώρης, and explained by 
 Xenophon's mention of the ύαχράτίαξ OLTiveg αρξονΰι τών ενοικούν- 
 τωι/ (C?/r. vm. 6, § 3), is obviously explicable from theSanscYii kshetra-m, 
 "a field," and pa, "to rule." Now in an inscription found in Asia 
 Minor we have εξαίΟ'ρατΐενειν , or εξόατραπενείν , for ΰαχρατΐενείν, 
 (Bockh, C.Zii. pp.470, 583). Theopompus (Phot. Cod. cLxxvi. pp. 120, 
 24) gave the name ΰατράτίτβ as εζατράτίης; and Scaliger (^ad Euseb. 
 nr. MDCxi) conjectured, from the form in Esther ix. 3; Esr. vin. 836, 
 that the foreign pronunciation of the word was often '^ξαδράτίαζ. This 
 is obviously the result of a pronunciation of the initial X similar to 
 that of which we have been speaking in the case of Xerxes. It might 
 be supposed that we have another example of the same kind in Όξά^'ρης ; 
 but here we can revert to the obvious analysis of 'OTavTjg (above, 
 § 160), and explain the name as compounded of the old Persian 'w, 
 Zend hu=εv and kshatra, "a warrior" or "chieftain." "We recognise 
 the same prefix, under a weakened form, in Άξάνηζ (Herod, vii. 66) 
 compared with the name Ozines (Curtius, ix. 41, § 19); and also in 
 Omartes (A then. 575 b) compared with the ethnic name Amardi borne 
 by tribes on the Caspian Sea (Strabo, pp. 507, 508, 514), on the north 
 
 * Similarly we have i before s and another consonant in Latin MSS. 
 and later inscriptions; see Lachmann ad Liicret. iv, 283; Corssen, Ausspr. 
 d. Lat. Spr. i. p. 289. 
 
CHAP. I.] NOUNS AND VERBS. 399 
 
 side of Taurus (Id. p. 510). in northern Media (Id. p. 523), and in Per- 
 sia, where they were also called Mardi (Id. pp. 523, 4). This last word 
 probably signifies "a man" (tJir) or "a warrior" (Zend wiereio, Old Persian 
 martiya), and is a proper name not only in the adjectival iovm^Mardon 
 (JEsch. Per s.bl) J Mardonius {Mar duniya^ i.e. virilis, strenuus), Mar- 
 dontes (Mardavanta), but also with the prefix dry a, "noble," in Ario- 
 mardus (-^sch. Pers, 38, 323). This ethnical name is interesting as 
 showing the same tendency to the general assumption of a warlike cha- 
 racter among the Persian tribes in the East, that we have remarked in 
 their kindred the Ger-mans and Her-min-ones^ compared with the Per- 
 sian Γερμάνιοί (above, §§ 76, 85), and it is a significant fact that ' Umar- 
 tiydjEvavOQog, "having good or brave inhabitants," is an epithet par- 
 ticularly applied to the province of Persis in the Behistun Inscription 
 (No. 3, 1. 8, 9, p. 274 4 see Rawlinson, As. Soc. xi. p. 85). AVith these 
 combinations within his reach, it is really surprising that Rawlinson 
 should say confidently that "the first element in Amardi cannot 
 possibly be hu^ 'good' " (Herod, iii. p. 550). The sibilant seems to be 
 prefixed chiefly for the sake of additional emphasis; at least such is 
 the general effect of this prothesis in our language; but it some- 
 times belonged to the fuller and more original form, as in λίπα^ 
 ά-λείφω^ s-a-lbon; latus, t-latus (τλητόζ\ s-t-latus; rip, t-rapping, 
 s-t-rip; lis, s-t-lis; lentus, s-t-lentus; locus, s-t-locus, &c. (see above, 
 p. 215). The labials appear for the most part to be prefixed to words 
 the meaning of which would admit of such additional force as might 
 be derived from composition with ατίό-, a-pa, a-va, ab, off, to the latter 
 part of which the prothesis corresponds. A similar explanation is 
 applicable to the dentals (comp. έτί-, a-ii^ a-d, &c.), and to the Latin 
 prefix re (jta-ga, pa-ra, &c.). We do not, however, believe there has 
 been any aphaeresis, as Pott suggests {Etym. Forsch. ii. p. 156): the 
 monosyllabic elements are shortened into single letters, in Greek as in 
 the German dialects (see Grimm, Deutsche Grammatik, n. p. 700), but 
 we cannot admit the propriety of stating that, when a prefix presents 
 traces of one of the elements of a compound worth, the other part of 
 that compound prefix must have been originally there: this amounts 
 to a denial of the separate use of these elements, which, however, is 
 too well known to be doubted. 
 
 214 (III) When a root, or rather quasi-root, terminates with a 
 consonant, and is combined with some element beginning with a con- 
 sonant, we frequently lose all immediate traces of the original form of 
 the root in consequence of its fusion with the termination. This 
 process is generally called Assimilation; it may sometimes be termed 
 Absorption, and in most cases would be included under those changes 
 
400 THE BOOTS OP [book III. 
 
 which we are accustomed to call euphonical. "We may trace it in the 
 melting down of two words into one by crasis, which we may observe 
 in our common conversation, a crasis which in Sanscrit is sometimes 
 extended through a whole sentence. 
 
 It will be convenient to consider the different kinds of assimila- 
 tion separately. They are, (1) a. Assimilation proper, or doubling one 
 of two consonants which meet in a word, and leaving out the other. 
 h. Doubling a consonant to replace a vowel which folloAvs it. c. Ab- 
 sorption and compensation in either of the two last-mentioned cases. 
 (2) Substituting (56 or ξ for a consonant and i. (3) a. Change of a conso- 
 nant to one more symphonic with a succeeding consonant, h. Insertion 
 of a symphonic consonant in certain cases. (4) Omission of the final 
 consonant of the root when it is identical with one in the termination. 
 
 215 (1) a. Assimilation of the first consonant to the second: 
 εν-νν-μί for εΰ-νν-μί (Ης-ννμι, Sanscr. vas, ves-tis); αμμες, νμμες^ 
 for αΰμεΰ,νΰμες (Veda asme, yushme) ; κομμώ from κόΰ-μος^ κα-Ο'-β-ρό^, 
 Bceot. 7C0^-a-QOg, Sanscr. gudh; flam-ma irom flag-ro, φλέγω; il-lus- 
 tris for in-lustris^ &c. 
 
 Assimilation of the second consonant to the first: κορρί} for κορ- 
 d?}, ολ-λνμι for ολ-ννμι, ϊτί-ποξ^ ϊκ-κοζ^ for eq-vm, Sanscr. agvas, 
 Pers. aspas. 
 
 b. The vowel l is frequently represented by a doubled liquid or ΰ, 
 and that too, whether the t follows or precedes : αλ-λος, ol-lus, il-le, 
 al-iuSj Sanscr. an-ya (above, §§ 135, 166); φνλ-λον^ fol-ium\ μάλ-λον, 
 μάλίον; μεόΰος, μεβίΟξ; κννοκεφαλλοξ ίον κυνοκεφάλυοξ (Α,νϊ^ίοφ. 
 Equ. 416; Phryn. Append, p. 49; Photius, p. 188 Pors.); ελλαπίνη 
 (είλαπ-), ώτελλά (οντει,λή), μάγερρος (μάγειρος)^ ^hvog (ξεΐνος), φά- 
 εννο£ {φaεLv6g)^ μέλλοχος {μείλυχοξ\ χολλάδεξ for χολιάδεξ (Phryn. 
 Append, p. 72), &c. Frequently the doubled ύ represents a guttural or 
 dental succeeded by t, as in ^άύΰον for τάχιον, κρέύΰων for κρατίων. 
 
 c. Absorption of and compensation for a consonant : λέγων for 
 λέγοντξ ; ϊηπονξ for ϊππον-ξ ; χαριεϊξ for χαρίέντς. 
 
 Absorption of and compensation for a vowel : rv^ratg for τνπτεΰί ; 
 γενέτειρα for γενετέρια ; μέλαινα for μελάνια ; εξαίφνης for έξαπίνης. 
 
 216 (2) The present tense of a verb, or that on which its pecu- 
 liar conjugation depends, is generally a longer form than the other 
 tenses : it contains besides the root an unorganic and generally a pro- 
 nominal addition. This addition is very often one of the equivalent 
 pronominal elements τ, ν, or the relative element -?/a, and we have 
 many derivative verbs in -ζω or -ύκω. Thus from the roots τυτΐ-,ταμ-, 
 we have the present tenses τνπ-τ-ω αηάτέμ-ν-ω'. we have both ατιμάω 
 
CHAP. I.] NOUNS AND VEEBS. 401 
 
 and ατιμάζω; both γηράω and γηράΰκω. The element ya appears to 
 be concealed in the contracted verbs, as they are called, and in the 
 very large class of verbs which form the present in -όΰω-) or, what is 
 only a dialectical variety, in -ττω. In all these last it is obvious that 
 assimilation has taken place : the only question is , with which of the 
 terminations mentioned above the last consonant of the root has been 
 consolidated. This question cannot be answered without a removal 
 of the difficulties which have always been experienced by etymologists 
 in fixing the value of the letter ξ. This letter, as we have already 
 shown (§ 112), is an assibilated or softened d. As an assibilated d it 
 was originally equivalent to d6', and it is frequently found in the 
 dialectical varieties of the Greek language under the transposition ΰδ. 
 As a softened d it is equivalent to dy, or the English soft g or j, repre- 
 sented in modern Italian by gi or ge, which may be derived from d(?, 
 and often passes into a mere sh. This appears from investigations in 
 the Greek language only ; thus Zsvg (=Ζε^ς), genitive difog, ξα-=δίά, 
 μίξων or μείζων = μεγίων, όλίζων = ολιγίων^, Λρωϊ-ζόξ, χ^ι-ζόζ^^- 
 τίρωϊ-δίοζ, χ^υδίοξ, "the before-day," "the near day" (for according to 
 Macrobius, Saturnal. i. 15: Jovem accipimus lucis auctorem, unde et 
 Luctetium Salii in carmine canunt , et Cretenses zlia trjV ημεραν vo- 
 cant, ipsi quoque Bomani Diespitrem appellant^ ut diei patrem). In 
 some Latin forms di is actually written for ζ (see Schneider, Elementar- 
 lehre, i. p. 386; Lobeck, Aglaoph. i. p. 296), and conversely, W. Din- 
 dorf would write κάρζα for καρδία in three passages of -^schylus 
 (Steph. Thes. Vol. ii. p. 1106). A comparison with other languages 
 leaves no doubt whatever on the subject: compare ζνγόν with San- 
 scrit yuj-^ Latin jw^-wwt; juturna with diuturna; Janus with Diana; 
 jubar with Sanscrit div, dyu^ Latin dies^ Italian giorno ; ζεα with San- 
 scrit yava; ζογγίβερί with "ginger;" ζνζνφοζ with "jujube;" ζάω with 
 Sanscrit /w, Lithuan. gyvs, gyvenu, gyvatd, Sclavonic schivu, &c. 
 Similarly, in the formation of the Kussian comparative kye=chj as in 
 kryepkie, " strong," A^ryepc^e, "stronger;" gye=j, as in dorogye, "dear," 
 doroje, "dearer;" cliye=sh^ as in suchye, "dry," sushe, "drier." The 
 dentals d, i, st are changed into j (French), ch, and shch, when followed 
 by the comparative affix ye. When it is represented by the transpo- 
 sition οδ, ζ admits of both modes of proper assimilation : the is assi- 
 milated to the δ in μάδδα for μάζα, in πλάδδω for τΐλάζω, &c. ; the δ to 
 ί\ιβ6ΪΏ.νί66ω ίοννίζω, άοΰΰεΐν from αοζος, λαπάΰόω for λαπάζω, &c. 
 
 * The remarks made by Bopp (Vergleichendes Accentuationssystem , pp. 
 224 — 226) on these comparatives are partly erroneous, and partly derived 
 from the former editions of the present work. He refers to Corssen in 
 the neue Jahrh. f^Phil. u. Pad. lxviii. p. 244, a periodical which we have 
 never seen. 
 
 DD 
 
402 THE ROOTS OF [bOOK III. 
 
 Those verbs which are assimilated in -ύβω very often return to the 
 former assimilation, thus for ΰφάζω we have ΰφάττω. 
 
 Having now determined the two values of J, and therefore of the 
 termination -ξω, we shall be able to establish with more accuracy the 
 nature of the assimilation of verbs in -ΰΰω, -ττω. Buttmann stoutly 
 denies the possibility of such an ending as -6ΰω Avithout an included 
 guttural {Lexilog. n. p. Ill), and would even go so far as to derive 
 άλλάΰΰω from άλλαχον (π. p. 198). But, besides the fact which we 
 have just mentioned, that -ξω may be assimilated to -ΰύω, the other 
 view is established by the instances quoted by Lobeck {Paralipomena^ 
 p. 403): namely, άφά(56ω, κλαδάΰΰω, κλνδάόΰω, τταλάΰΰω, uita- 
 ράΰϋω, άλ^εβΰω, έδρηΰΰω, λαφνϋΰω, λψώΰΰω, &c. Leaving out 
 these words, and such as φράξω, root φραδ-•, οξω, root 6(5-, which are 
 never assimilated, we must consider those in which a final consonant 
 of the root is actually contained and concealed in the assimilation, as 
 in the following examples : (1) Gutturals : Λράΰϋω, root Λραγ- ; τάΰύω, 
 root tay-; άΐόΰω^ root αϊκ-, αίγ-; φρίΰόω, root φρικ-; ^66ω,Υθοϊ 
 ρήγ-; πλήΰΰω, root πληγ-; δράόΰω, root δραγ-. (2) Labials: πεύΰει,ν^ 
 root jtETt-; lvL66sLVi root bvltc-; λάζομαι, root λαβ-] νίπτειν^ νίόΰειν, 
 νίζειν; κόϋόειν χότιτειν. (3) Dentals : κορνβΰειν, root κορν%'-; ερέύ- 
 ΰειν^ from ερετ-. 
 
 Now it is quite clear that these verbs are not formed by a mere 
 addition of -6ω to the root : by this addition the futures of all of them 
 are characterized, and it would be absurd to suppose that the only 
 difference between a present and future should be an assimilation of 
 one of them. The termination of a present tense, if strengthened at 
 all , must involve one of the pronominal endings , r, v, ΰκ , ξ, or ya. 
 Now the termination in this instance cannot be t or v, for there is no 
 example of the assimilation of these letters with any preceding mute. 
 Nor can it be -6κ appended to a root ending with a guttural, for 
 then the assimilation takes place according to (4), as in λάΰκω for 
 λάκ-ΰκω, μιΰγω for μίγ-βκω. The only way in which the ending -όΰω^ 
 -ττω can arise, is from an assimilation of the ending ξ affixed to roots 
 terminating with a vowel , or of a guttural or dental with the prono- 
 minal ending ya. It has been mentioned before that both gutturals 
 and dentals, when followed by y, are softened into sibilants, a softened 
 guttural being equivalent in many cases to a softened dental, and thus 
 both are represented by ζ (compare ολίξων with ολίγος, andz/toff with 
 Ζευς)• Accordingly, it is probable that in the majority of cases the 
 guttural or labial has been softened into a sibilant by the addition of 
 the pronominal element icc, Sanscrit ya, which is represented by a 
 doubling of the sibilant, so that πράύΰευν may stand for πραγεειν, or 
 7tράγyεL•V, as ^άΰΰον stands for ταχίον, Β,ηάερεΰΰεΛ for ερ£τεε^ν or 
 ερέτyεL•V, as κρέΰΰων for κρατίων. This appears more clearly in the verb. 
 
CHAP. I.] NOUNS AND VEEBS. 403 
 
 roots which have liquid endings : here the iota is occasionally trans- 
 posed to counterbalance an omission of the vicarious liquid : compare 
 φχί'είρω, γείνω, κτείνω, with the other forms φθερραι, γεννώ, κτεννω. 
 Compare also χραιΰμεΐν with χρήΰίμ>ος. The appearance of this ending 
 after roots terminating in labials cannot be explained in this way. 
 In all the labial verbs which have by-forms in ξ and 66, the labial 
 seems to stand for a digamma, which has been in some cases omitted 
 without compensation, and in others represented only by its guttural 
 element. Thus in νίτίχειν, the root is m^-, judging from χερνίβ-α, &c. 
 That however this β is the representative of a digamma, and that the 
 real root is νε^=ηβςν, appears from the verb νεω (νε/ω), ενεν6α, νεν- 
 6ομαι, &C., vavg, navis, &c., which convey the general idea of "being 
 in the water," compared with the by-form νήχω, which contains the 
 guttural only. Hence Nep-tunus, "the god of the sea," stands by the 
 side of the Etruscan Ne-thuns (Varron, pp. 148, 168), and! the Greek 
 Νη-ρενς, where the guttural perhaps is absorbed. In many of the forma- 
 tions from this root, as νεω, vrj6og, νίτρον for νίτίχρον, &c., the digamma 
 has vanished; in the case of m' -ξω, νί-66ω, then, we might suppose that 
 these terminations are merely suffixed to the root denuded of its termi- 
 nating digamma-sound. When, however, we advert to the l in ναίω, and 
 to the same letter either directly or by implication contained in the other 
 forms, it seems more reasonable to conclude that the guttural element 
 became predominant in these forms, and that it is simply combined 
 with t in the verb νίξω=νί'γίω : comp. the Latin rabies with the French 
 rage, Butupium with the modern Bich-horough , pcedagogium with 
 page, &c. The same may be said of λά-ζομαι, from the root λαβ- or 
 λαΡ, AfP, for it is clear that λαχ- (in Χαγχάνω, &c., Hebrew ri/ib, 
 ldqa%) contains the same root, and this of course indicates a combi- 
 nation of the guttural with the labial in the original form of the word. 
 In %ε66ευν, from τίεΐί-, it will be recollected that the labials , between 
 which the root- vowel of the Greek form is confined, are represented in 
 the Latin co-qOO=qvoqvo by twogv's (above, § 121); the secondlabial 
 is still a softened guttural in the Sanscrit pack-. The same is to be 
 observed οίΙνί66ευν, ε^τίΤΒν (whether it is connected with ϊτίοξ, voc-s, 
 Sanscr. ναοΛ; with ετίομαι, sequor, Sanscr. sack; or with νίκ-η, νεΐκος); 
 and of 666ομαι, οτΐτομαί] ο66ε, ο'ψις, τριοττίζ, τρίοπίξ) comp. Latin 
 oculus, Sanscrit iksh. The idea of striking conveyed by κότίτείν is 
 derived from that of butting with the head (κεροτντΐεΐν, &c.) , and all 
 the cognate words point to this union of stooping and striking in the 
 meanings of the word: compare κνΛ-τείν, κυβερνάν (to direct the head 
 of a ship), cub-are (to lay down the head), cap-ut, κεφαλή, Sanscrit 
 kapdla. But besides the root cap-, κοΛ-, for "a head," we have also 
 an accessary root κορ-, κορ-6Ύΐ, κόρν^-ξ, Sanscrit ^iras, &c,, with the 
 
 DD2 
 
404 THE ROOTS OF [bOOK III. 
 
 same additional meanings of butting, striking, as in XBQ-ag. From this 
 second root comes by assimilation κό(?(?ο^, %o(56blv ,%6tra,%OTta^og,&c., 
 which therefore have no immediate connexion with %67trBLV, &c. With 
 regard to 7iOQv6(5w, κορνΛτω, κορνττολος, κορνΛΤίλος, it will be re- 
 membered that there is a form KOQvd'-s, as well as κορυφή. On the 
 whole, then, it appears, that the barytone verbs in -6ΰω or -ττω, are 
 properly assimilations either of the suffix -ζω, or of a guttural or 
 dental at the end of the root with the pronominal suffix ya. 
 
 This assimilation of a guttural or dental succeeded bjiory into 
 06 or ττ, which we have been endeavouring to establish in the case of 
 barytone verbs, appears also in the case of certain feminine nouns, in 
 which Bopp, erroneously as we believe, supposes that the original 
 feminine l is not represented (Vergl. Gramm. p. 140). Such words 
 are Θρα(?^α for Θράκια, βαϋίλίύΰα for βαΰίλίδία, ίίναΰΰα for άνάκια^ 
 μΕλίΰΰα for μελίπα. To these feminine forms we must return in the 
 following chapter. Before t the δ and τ are naturally assibilated, as 
 in ξα from διά, ^ανάβίμος from >ϋ'άνατος ; the κ becomes s through 
 the intermediate sound, ch, sJi, as in modern languages; compare ca- 
 mera, chamber y chambre ; and when two vowels follow a ί or c the 
 sound in both may be the same; compare redemption^ Lucien. 
 
 217 (3) a. This rule in most frequently exemplified in compounds 
 with prepositions: as εμβάλλω ίονίνβάλλω; άμφί for άναφί, &c. We 
 have 6 for δ in certain words, as άΰμα from αδω, olu-d'a for οΙδ-%'α, 
 η6'%'ην from ηδομαί', εΰ-&ίω, root εδ-, &c.; and this takes place in 
 some cases when there is in fact no actual composition, but only an 
 immediate succession : thus we have in inscriptions εμπυρί, ύνγκαρτίώ, 
 εΰτήλτ;] for εν τΐυρί, Gvv καρτίώ, εν ΰτήλ^. 
 
 b. The Greek ear seems to have been particularly averse to the 
 immediate concurrence of ^λ, μρ^ νρ, v6, ΰρ, 6λ (above, § 96), and 
 whenever, by contraction or otherwise, any of these pairs of liquids 
 have come in contact, the mute which bears the nearest relation to 
 the first of them is inserted, unless assimilation takes place, or one 
 of the liquids is omitted, which generally happens in the combinations 
 v6, βρ: thus β is inserted after μ, δ after ν, and Ο' after ^; for exam- 
 ple, we have μεμ-β-λεται for μεμελψαι.; μεμ-β-λωκα from μολείν; 
 αμ-β-ροτοξ for α-μορτοξ ; γαμ β-ρόξ for γαμερόξ; μεΰημ-β-ρία for 
 μέύη ήμερα, and 6μ-β-ρος compared with humor; we have άν-δ-ρός 
 for άνέρος, and εύ-%•-λός for εβλός from εδ-λός; cf. edel, &c. When 
 μ is inserted before λ or |3 as in ά-μ-τίλάκημα, πί-μ-ττλημί, ο-μ-βρυμος, 
 &c., it is merely an instance of anusvdra. 
 
 218 There is one word in which this insertion of β has not been 
 observed by Greek scholars, and as it has occasioned some difficulty 
 
CHAP. I.] NOUNS AND VEBBS. 405 
 
 and misconception, it will be worth while to explain it more at length. 
 "We allude to αμ-β-λνξ^ which appears to us to be merely another 
 form oi αμα-λόζ and άμαν-ρόζ, with both of which it coincides to a 
 certain extent in signification. The primary meaning seems to be 
 ''smooth," "reduced to a level," as opposed to any word signifying 
 "pointed," "projecting," "sharp;" hence, by a natural transition, it 
 implies want of vigour or energy, — the weak, timid, or tender. It 
 is also applied to express the fading of colours, loss of the vivid 
 freshness which once distinguished them, &c. The first syllable seems 
 to be one of those prefixes of which we have spoken above. At least, 
 the word μαν -QOg appears separately, and μα-λα-χΟ£^ its synonym, 
 seems to be only another form of ά-μ-β-λνς, the second pronominal 
 stem being appended under the shape -κ-ός instead oi-v-g. Μα-ρα-ίνω, 
 which in its earliest use signifies "to extinguish fire" (Homer, Hymn. 
 Merc. 140), and which is particularly opposed to φλέγω (Soph. AjaXj 
 700) , seems to be a derivative of μανρός. As we have άμβλνΰκω^ 
 άμβλνωΛω, άμβλόω, from αμβλύς, so we have also βλώβκω, βλω&ρός; 
 and we find άβληχρός, as well as βληχρόζ, βλάξ. The verb βλώβκω 
 and its aorist f'^oAoi' (comp. %ρώ6κω, ε^ορον ; ^νήύκω, ε^ανον;) signify 
 "to go;" this meaning arises naturally from the sense of levelling, 
 smoothing, clearing away, so prominent in the adjectives which we 
 have mentioned first. It is curious to observe generally how words 
 denoting sharpness, a point, acuteness, &c., are applied to convey the 
 ideas of rapid motion, and, metaphorically, readiness of wit, while 
 those indicating a level, smoothness, bluntness, &c., are used to signify 
 slowness, dulness, &c. This is particularly the case in Greek. The 
 reader will remember such phrases are άμβλν ξυφίδίον, αμβλν οραν, οί 
 αμβλντεροι την φύουν (to which meaning we refer the gloss in Hesy- 
 chius ιαβλας, άΰννετος, αγνώμων), αμβλνξ τΐρος δρόμον, αμβλύτεροξ 
 τίρος την μάχην, on the one hand, as opposed to o^v φάόγανον, όξ,ύ- 
 τατον δέρκεΰ^'αί, o|vs ετΐυνοηΰαυ, οξύτατοι ΐτΐτΐοο (Herod, ν. 9). We 
 also find αμβλύς opposed to τΐίκ-ρός, the original signification of which 
 is undoubtedly "piercing," "penetrating," "sharp." As Λίκρός in its 
 metaphorical use is particularly applied to denote that which is pain- 
 ful or galling to the feelings , so αμβλύς is employed to express the 
 effect of appeasing or removing disagreeable or distressing sensations : 
 thus Thucydides (ii. 6b): ών τίερϊ τα οΙκεΐα έκαστος ηλγει, άμβλύτεροί 
 ηδη οντες. We beUeve that the root of άμ-β-λύς, άμα-λός, αμαυρός, is 
 found in αμάω ; the opinion of the lexicographers that this verb is 
 connected with αμα, and means to collect, to gather together, appears 
 to us very erroneous , and we cannot conceive how modern scholars 
 could adopt this view, which seems to be quite at variance with the 
 common usage of αμάω and its compounds. In all the passages in 
 
406 THE ROOTS OF [bOOK III. 
 
 which it appears, άμάω means "to make a level surface," "to lay 
 flat or even," and, when applied to corn, which is emphatically said 
 φρίΰΰείν, "to stick up" or "stand on end" (horrere) (Iliad xxiu. 
 599), it means "to cut down," "to lay down." The following pas- 
 sages will make this clear. When earth is said to be laid smoothly 
 over any thing, as on a grave, the phrase is ετΐαμάΰ^'οα γην; Herodot. 
 vm. 24 : τάφρους ορνξάμενος ε^'α'ψε, φυλλάδα τε εταβαλών καΐ γην 
 ετΐαμηβ ά μένος , on which Valckenaer quotes from Plutarch: το λείο- 
 τατον εΛαμαται της Q'Lvog αϋτοΐς καΙ μαλακώτατον. The two epi- 
 thets in this passage of Plutarch show clearly that smoothness, and 
 not collection or congeries, is intended. In the same way Homer speaks 
 of smoothing or making level a bed of leaves, with which, he says, 
 the ground was covered, so thas there was no need of collecting them, 
 Odyss. v. 483: 
 
 αφαρ d' ευνην ετίαμηβατο χερΰί φίλ'^ΰίν 
 
 εύρεΐαν φύλλων γαρ εην χύόος ηλι^α τεολλή. 
 Also of milk laid out in broad dishes for the purpose of forming 
 cream, Odyss. ix. 247 : 
 
 αυτίκα d' ημιβυ μεν ^ρε'φας λεύκοιο γάλακτος 
 
 τίλεκτοϊς εν ταλάροιβιν άμηΰάμενος κατε^ηκεν. 
 It is only by perceiving this sense of the word that we can translate 
 a well-known passage of Sophocles (Antig. 600), where Askew's 
 emendation, κοτΐίς, is absurd: 
 
 νυν γαρ εΰχάτας υΛερ 
 
 ρίζας ο τετατο φάος εν Οίδίτΐου δόμους, κατ οώ νιν 
 
 φοινία ^εών των 
 
 νερτέρων άμα κόνις, 
 Α light had beamed upon the root (see our Notes on the Antigone^ 
 p. 181), but the dust was levelled over it, and made it άμαυρόν again. 
 As the Greeks said καταμά6%^αι κόνι/ν, so also conversely they could 
 say καταμα η κόνυς. Hence it is, that, when they spoke of penetrating 
 a surface lying flat or level over something they wished to get at, they 
 used the verb διαμαβΟ'αί, as in Thucyd. iv. 26: διαμώμενοι τον κά- 
 χληκα, or διαμάν in poetry, as in Eurip. Bacchce, 701 : ακροιύι δακ- 
 τύλοίΰί διαμώΰαι %%όνα. 
 
 It will be proper, as well for this as for other reasons, to examine 
 minutely a passage in JEschylus, in which the meaning of αμβλύς has 
 not been properly understood. It is in the Eumenides, 229 Miiller: 
 
 ηκω, δεχου δε τίρευμενώς άλάότορα, 
 
 ου Λροΰτρότίαιον, ούδ^ άφοίβαντον χέρα, 
 
 αλλ' άμβλυν ηδη^ προβτετρψμενον τε Λρος 
 
 αλλοιϋιν οϊκοις και πορεύμαΰυν βροτών. 
 
CHAP. I.] NOUNS AND VEEBS. 407 
 
 We do not agree either with Miiller or Hermann in their reading and 
 interpretation of this passage. Hermann's τίροΰτετρνμμενον μν6θξ is 
 quite unjustifiable; it appears to us entirely unmeaning, and is sup- 
 ported by no evidence, either internal or external. We have no 
 hesitation in saying that Miiller is wrong in taking tcqOq adverbially : 
 it is obvious from v. 429, Ttakai JtQog άλλοις tarn άφιερώμεΟ'α οϊ- 
 κοίϋι, καΙ βοτοΐΰί καΐ ρντοΐς πόροι,ς (comp. 272, TtQog εΰτία Ο'εον)^ 
 that TtQog agrees with the following datives ; its position at the end of 
 the line is sufficiently supported byjDindorf's quotation from Soph. 
 CEd. C. 495 : λείτίομαι γαρ εν τω μη δννα6^αι μηδ^ οράν δνονν κά- 
 κουν. Hermann seems to think that the last words of v. 430 contain an 
 explanation of the τίορεύμαΰυν βροτών (Opuscul. vi. p. 46). But the 
 καϊ βοτοΐόί καί ρντοϊς πόροΐξ are not connected with οΧκοι,ΰι; they 
 are datives of the instrument: "I have been purified at other places 
 by means of sacrifices and the pouring out of water." The τίορενματα 
 βροτών is only a general expression — "where men go*." We have 
 νάϊον πόρενμα in Euripides {Iph. Aul. 300), and Hermann, who had 
 read ^schylus so often, might have recollected τέκνων κελεν^οι 
 (Choeph. 350), and τρίβοι ερώτων (SuppL 1042). MuUer again has 
 lost sight of the connexion, and has mistranslated άμβλνξ, which does 
 not here mean abgestumpft^ i. e. "blunted," but implies a fading* or 
 loss of colour, i. e. of the colour of blood, as below, v. 270 : 
 
 βρίξβΐ γαρ αίμα καϊ μαραίνεται χερός, 
 μψροκτόνον μίαόμα δ' 'εκτίλντον ηελει. 
 Λοταίνυον γαρ ον ηροζ εύτία Ο'εοϋ 
 Φοίβου κa^'aρμoΐg ηλά&η χoLρoκτόvoLg. 
 We have a reference to the same sense of άμβλvg in Plato , Respuhl. 
 
 IV. p. 442 : μη πη ημΐν άτίαμβλννεται άλλο τι δικαιοσύνη δοκεϊν είναι 
 Tj οτίερ εν τ^ τίόλει εφάνη ; "justice has not lost any of its fair pro- 
 portions, has it? its colours have not become faded and dim, so that it 
 should appear to be something different from that which it appeared 
 to be in the state?" 
 
 In the passage of -^schylus now under consideration we read 
 ϋΐροβτετραμμένον in the sense in which τΐροΰτραπεβ&αι occurs in 
 
 V. 200: 
 
 καΐ Λρούτραπεΰ&αι τονΰδ^ ετίέΰτελλον δόμovg, 
 for ΛροβτρατΐέύΟ'αι δόμovgJ in the aorist, is the correlative phrase to 
 τΐροΰτετραμμενον 7tρog oϊκoιg in the perfect ; and the sense of these 
 lines is as follows: "I am not a polluted person, i. e. an applicant or 
 
 * Dindorf has since given the same interpretation (Steph. Thes. Vol. 
 1482): "de locis quibus homines viam faciunt, jEsch. Eum. 239." 
 
408 THE HOOTS OF [book III. 
 
 supplicant for purification; nor is there the stain of blood upon my 
 hand ; but that stain is already (ηδη) washed out and faded away 
 (αμβλύς ειμί = άμβλννομαυ); and I have prayed for purification (τίροβ- 
 τετρκμμαή at other temples and in the haunts of men." So that αμ- 
 βλύς is synonymous with ουκ άφοίβαντος χέρα, and τίροΰτετ ραμ- 
 μένος with ον τΐροΰτρότίαιος, a sort of tautology very common in 
 iEschylus. See, for instance. Prom. 613: ονκ εμΊΐλεκων αΐνίγματ 
 αλί! ατίλω λόγφ. Ibid, 951: κ«1 ταντα μεντοι μηδέν αΐνίκτηρίως 
 άλλ' αν%•' εκαΰτ εκφράζε. Pers.'QSA: μψι μακεβτηρα μν^ον άλλα 
 ύνντομον λέγων. Sept. c. Theb. 866 : ονκ ετά φιλία αλλ' ετά φόνω 
 διεκρί^ψε. See also Eumen. 436—439, 762, 3; Prom. 654—7, &c. 
 In a fragment of the ^olus of Euripides {apud Galen. Charter. 
 p. 418 Kuhn) we have 
 
 ει μεν τόά' ημαρ τίρώχον ην κακονμενφ, 
 και μη μακράν δη δια τίόνων ενανΰτόλονν, 
 εΙκος ΰφαδάξειν ην αν, ώό νεόξνγα 
 τίώλον χαλινον άρτίως δεδεγμενον. 
 νυν δ' αμβλύς εΙμι και κατηρτνκώς τίόνων, 
 
 where αμβλύς είμι = αμβλύνομαι is applied, in the other sense of μα- 
 ραΐνεβ^'αι, to a taming or quenching of the fiery spirit of a young 
 horse, as appears even from Cicero's loose translation (Titscul. Dis- 
 put. III. 28): 
 
 Sed jam subactus miseriis obtorpui. 
 
 In the same sense we find ατίαμβλύνω in ^schylus , Se^ot. c. Th. 
 697: τε%ηγμενον τοί μϋ ονκ άτίαμβλύνεις λόγω; Prom. 868: μίαν 
 δε τζαίδων ίμερος ^'έλξει το μη κτείνειν ύύνεννον, αλλ' απαμβλνν- 
 %η6εται^ "she will be tamed." 
 
 We may take this opportunity of explaining the participle κατηρ- 
 τνκώς^ which occurs as a quasi-synonym of αμβλύς both in this pas- 
 sage of Euripides, and in the Eumenides^ v. 145: 
 
 ίίλλως τε και ύν μεν κατηρτνκώς εμοΐς 
 ικέτης προΰηλΟ'ες κα^'αρος άβλαβης δόμοις. 
 
 The word καταρτύομαι and the perfect participle κατηρτνκώς are 
 applied to signify arriving at maturity, and the effects of age in so- 
 bering the passions of youth. Thus Solon says (p. 66 Bach.): 
 
 T^ d' έκτη τΐερί τίάντα καταρτύεται νόος ανδρός, 
 ονδ' ερδειν ε%'' ομώς εργ' άτιάλαμνα ^ελει, 
 which should be translated, "At the age of forty -two the mind of man 
 is matured in every respect (for the use of τίερί, comp. Plat. Resp. v. p. 
 449 A : κακας καΐ ημαρτημεναι ηερί τίόλεων διοικήΰείς), nor does he 
 now commit the intemperate actions of his youth" {ομώς, i. e. as he 
 
CHAP. I.] NOUNS AND VERBS. 409 
 
 did before). For tlie use of άτίάλαμνα, see Theognis, v. 481. From 
 this meaning of 'naτaQτvω we derive its use to signify the effects pro- 
 duced by training, especially upon horses, to which the passage of the 
 j^olus particularly refers. See also Sophocl. Antig. 478: 
 
 ΰμίκρφ χαλίνω ά' old a τουζ ^νμονμένουζ 
 
 LTtTtovg καταρτνΟ^ενταζ. 
 Plutarch, Themist, c. ii. : rovg τραχντάτονς Λωλονξ αφβχους LTtTtovg 
 γίγνεόΟ'αί φάΰκων, otav, rjg τίροΰήκευ, τνχωβί Ttaideiag καί καταρ- 
 τv(^εωg. Vol. ι. ρ. 31 d: LTtTtovg — κaτaQτvovτεg ετΐί rovg ayovag 
 ayovuLV. p. 38 d: καταρτντ^ την φνβυν. From this is derived the pe- 
 culiar meaning of κaτηρτvκώg — "a horse or ass which has cast its 
 teeth," i. e. "aged," because his age can no longer be known from his 
 teeth, which are therefore called γvώμovεg or φρa6τηρεg, "the index 
 of age." Suidas : '^ίβολήτωρ καΐ a^olLg καΐ aβoλog, ovog 6 μψ 
 8επω βεβληκώg odovrag, εξ ov γvωρLζετaL η 'ηL•κLa τον ζώον. εκ δε 
 τούτον ο viog, ονδ^ω γνώμονα έχων. γνώμονα δε ελεγον τον βαλ- 
 λόμενον οδόντα, δL ον Tag iik'Miag εξήταζον τον δε αντον καΐ κα- 
 τ η ρτ ν κότα ελεγον, εκ μετaφoράg τώντετραΉ:όδων. κάί άπογνώ- 
 μovag, Tovg άτtoγεγtjρaκότag, oig ελελο^ευ το γνώριύμα. καΐ 
 άβo'λovg τtώλovg Tovg μηδ^ω βεβληκότag 6δόvτag. Hesychius: 
 "Aβoλog. viog, ονδετιω ρL'ψag οδόντα, τον δε αντον καΐ κατηρ- 
 τνκότα ελεγον. Γνώμ α τον βαλλόμενον οδόντα, δL ον Tag ηλLκLag 
 εγνώριξοντών τετραπόδων, κάί 6 κaτηρτvκώg ηδη, λειτίογνώ- 
 μ ω ν. λεγετaL δε καί γvώΰLg' Κατηρτνκώζ. τελεLώΰag. κvρLωg 
 δε επί των αλόγων ζώων, όταν εκβάλτ^ TtavTag Tovg 6δόvτag. Αει- 
 τtoγvώμωv, δ μ'ηκετL βόλον έχων ο δε τελεLog, καί γεγηρaκώg μη 
 έχων γνωρίσματα Ttjg ηλLκίag. Malala,^C/irow. ρ. 379 (quoted by Toup, 
 ΠΙ. p. 539): τtρωτoβόλog. Pullus^ qui primos denies emittit. From this 
 it appears that in the passage of Euripides κaτηρτvκώg is used in 
 its secondary and more limited sense as indicating the age of a horse, 
 and so the Antiatticistes {Bekkeri Anecd, p. 105, 1. 25) understood it: 
 KaτηρτvκεvaL ελεγοντο oi μηκετL βόλον εχovτεg LTtTtOL. EvρL•τιLδηg 
 ΑΙόλω. In the passage of ^schylus (Eumenides, 451) κaτηρτvκώg is 
 taken in its wider signification, "having performed or completed," i. e. 
 all the necessary rites, just as τελog, τελεω, τελεLog, τελεLόω are used 
 absolutely in speaking of the performance of sacred rites, although 
 they only denote completion or fulfilment in general , and τέλεLog is 
 used like κaτηρτvκώg■, in speaking of the age of animals, in opposition 
 to aβoλog. Plato, Legg. vin. p. 834 c: μovLTtτtOLg τε ά&λα τι^εvτεg, 
 τtώλoLg τε aβόλoLg καί τελείων τε καί άβολων Tolg με60Lg καί aircoig 
 δη Tolg τέλog εχovΰL. The Scholiast, on the passage of ^schylus, felt this, 
 when he wrote κaτηρτvκώg.τέλεLog την ηλLκίav. τοντο δε απο των 
 ζώων ι and Hesychius too : κaτaρτv6aL.κaτa6κεvάύav.τελεLώ6aL. 
 
410 THE ROOTS OF [bOOK III. 
 
 ΰτερεώόΜ. Also St. Paul {Corinth, ii. xiii. 11): καχαρτίξε6%^ε, '^be 
 perfect." Καταρτνω is used in this more general sense in Soph. (Ed. 
 Col. 71 : og TtQog tl λέξων η καταρτύΰων μολεΐν; on which Suidas 
 writes: καταρτύΰων. τίαραϋκενάόων. εντρετΐίΰων. And thus we 
 have, in the use of αμβλύς and κατηρτνκώξ by ^schylus and Euri- 
 pides, a fruitful instance of the manner in which two authors nearly- 
 contemporary may employ two words in connexion with one another 
 with a marked difference , but yet with an affinity of meaning that 
 cannot be mistaken. » 
 
 219 (4) Of this rule we find frequent exemplifications in verbs 
 and verbal compounds, both in Greek and in Latin. Thus διδαχή, 
 διδάβκειν^διδάχ-βκειν ; docere, discere=dic-scere ; δυκεΐν, δίΰκος= 
 δίκ-όκος-, λακεΐν, λά6κω=λάκ-6κω; εϊκω, εΐύκω = εϊκ-ΰκω ; είττεΐν, 
 ϊΰκεν=ΪΛ-6κεν] έ'χω, ϊόχω (εχ-ΰκω); λέγω, λέΰχη=λέγ'ΰχη ; μι,γψ 
 vaij μίΰγω=μίγ-ΰκω, misceo =micsGeo ; πα^'εΐν, 7τάΰχω = Λάϋ'- 
 βκω = πεν%'-6κω; μόγ-ος, μόΰ-χος=μόγ-6κος (comp. vacca, vehere); 
 αϊδως,οίϊΰχος=αϊδ-ΰκος', ί| (εκ) εΰχατος = εκ-ΰκατος; precor, pro- 
 cusj posco =proG-sco. 
 
 220 We may mention as outwardly connected with assimilation, 
 though in fact by no means proceeding from the same cause, the very 
 common practice of doubling liquids in Greek words where no com- 
 pensation is necessary or intended. This is particularly remarkable in 
 some proper names which occur in the Attic dramatists ; as ΊτΐΛομ- 
 μεδοντοζ (^sch. Sept. c. Theb. 488), Παρ^εννοτίαΐοζ (Id. Ibid. 547), 
 Τελλεύταντοξ (Soph. Ajax, 210),^Αλφεΰ6ίβθίαν (Soph. ap. Priscian.]p. 
 1^28)/ΐΛ7ΐοδ άμμου (Axistoi^h.. Equit. 328), in some of which passages 
 editors of the Porsonian school have introduced alterations as forced 
 as they are unnecessary. To these arbitrary reduplications we may 
 add βρόκχον for βρόχον in Theognis, v. 1095. (See Scaliger ad Eusel•. 
 p. 119, quoted by Gaisford, Poet. Min. n. p. xxix.) 
 
 221 An inquiry into the rules of assimilation might lead us to 
 an investigation of a converse phenomenon in the Greek and other 
 languages, which Pott (Etym. Forsch. n. pp. 65 foil.) calls dissimila- 
 tion. This consists in the avoidance of a concurrence of similar syl- 
 lables and letters, hiatus of vowels, and so forth: and euphony is 
 alleged as the cause of this as well as of its opposite. We do not 
 think, however, that this so-called dissimilation is due to anything 
 beyond accident or caprice, a wayward choice or an accidental mis- 
 pronunciation. That the Greek ear often repudiated such concur- 
 rences, and that many Greek words, especially compounds, have lost 
 in consequence essential consonants, is well known (see Lobeck 
 Paralipom. Dissertatio prima)-, but it is equally certain that they 
 had, when the fit took them, a strong predilection for alliteration, an 
 
CHAP. I.] NOUNS AND VERBS. 411 
 
 irresistible propensity to assimilation in words and homoeoteleuta in 
 sentences, of which their constant use of the figura etymologica is a 
 sufficient proof. Of their dislike to a concurrence of aspirates, or 
 indeed to an appearance of two aspirates in the same word, and of the 
 transpositions occasioned by this taste, we have already spoken. Of 
 the other changes which fall under the head of dissimilation it is 
 scarcely worth while to speak, for they seem incapable of a reduction 
 to rules and systems; besides, the phenomena are to be collected from 
 so wide a field that the enumeration would far exceed the limits to 
 which these discussions are necessarily confined. 
 
 One of the most common changes of dissimilation with which a 
 root may be affected without losing its identity is that called Meta- 
 thesis. The metathesis of vowels takes place most frequently in the 
 case of liquids, because it is a matter of indifference whether a vowel 
 is placed before or after them. Instances of this change meet us con- 
 stantly in every language. Thus in English we have bird^ hrid; 
 third, thridde; bordel, brothel; burst, brust; board, broad; &c. {Di- 
 ver sions\of Purley, n. pp. 83 foil.); and there is a curious instance of 
 the same kind in Drayton's Nymphidia; for when he says, 
 "By the mandrake's dreadful groans. 
 By the Lubricants sad moans,", 
 it is clear that he is alluding to the Lubber kin, that lazy fiend, who 
 piteously resisted his brother fairies' attempts to awaken him. In 
 Oreek we find Η^'ανον,&νήϋκω ; καρδία, κραδία ; ρίξει,ν, ϊργον; εμολον, 
 βλώΰκω, and even in the same word εγρ-ηγορα (Journal of Education, 
 V. p. 305). And so also in comparisons of different languages, as Eng- 
 lish horse, German Boss, French roussin; English fright, German 
 Furcht for Vor-acht; English folk, German Volk, Latin volgus, Greek 
 όχλος, Cretan Λολχος, SclsiYomc plok , polk, pulk, Lithuanian ^jw^^as 
 (as we read of a "pulk of cossacks"), old Norse flockr, Anglo-Saxon 
 floe, English flock; &c. Metatheses like 'ψίν for 6φίν; -φάλυον for 
 (ΐηάλίον, φάΰγανον for όφάγανον, &c., are analogous to that transpo- 
 sition of the aspirate and semivowels, which we have already stated 
 and explained. We may also compare ^ίφος = ΰκίφος (scldef) with 
 the old Egyptian sef, and the Semitic analogies pointed out by Bun^ 
 sen, JEgypten, i. p. 582. There does not seem to be merely a meta- 
 thesis in (ίφτ}! compared with vespa; but rather a concealed redupli- 
 cation in the latter, like that of vivus compared with quick (above, 
 § 121); for vespa=hvespa=:svespa=spe-spa, and ΰφήξ contains only 
 one of these elements, with the formative affix |. 
 
 222 (IV) (1) The most systematic of the changes which a 
 root undergoes is the modification of its vowel. This is produced by 
 
412 THE BOOTS OF [BOOK III. 
 
 influences similar to those to which we have attributed the assimila- 
 tion of the consonants, namely, the contact with the prefix and suffix, 
 the greater or less weight of which induces a less or greater weight of 
 the root-vowel. 
 
 It has been shown in a former chapter that there are properly- 
 speaking only three distinct vowels : a or the common sound by which 
 all consonants are articulated; % and u the ultimate state, the former 
 of the gutturals and the dentals, and the latter of the labials. The 
 vowels and e differ from a in weight only. 
 
 The Greek vowels a, 6 , υ , as opposed to one another, in most 
 cases correspond to the three original Sanscrit vowels a^i^u, and there 
 is seldom , if ever, any interchange or confusion between the different 
 vowels of the two sets; whereas the two Greek vowels, f, o, very often 
 share with a in the representation of the Sanscrit a, f being the most 
 common substitution for it, ο the next, and a the least usual. Of 
 these three representatives of the Indian short vowel, the heaviest is 
 a, the lightest ε, and ο stands between them. As the Indian a is the 
 mere articulation of the different consonants which form the Sanscrit 
 syllabarium, and is therefore the shortest possible, it is of course 
 obvious why the lightest of the Greek vowels stands in its place. 
 Any one of the three forms of the Greek articulation - vowel may 
 stand as a representative of the Sanscrit a; but practically it appears 
 most frequently as ε, less so as o, and most rarely as a; thus we have 
 pancha, jtsvts, gankJias, %6γχοξ, aldbliam, ελαβον. It is a great ad- 
 vantage to the Greek alphabet that it has these distinctions in the 
 weight of the articulation-vowel. In Sanscrit we have sometimes 
 only the accent to distinguish between two different cases, as padas, 
 τίόδεζ, padaSj τίοδός. Κ the Greek had not these three vowels, there 
 would be no means of discriminating the three proparoxytones, εχον- 
 tog, εχovτεg, έχοντας. Of the three distinct vowels a, i, w, it is clear 
 that i is lighter than a though it is heavier than e. This appears, as 
 far as the Latin language is concerned, from the fact pointed out by 
 Bopp (Vergl. Gramm. p. 5), that in secondary formations the radical 
 a is turned into ί in syllables terminating with a vowel , and into e 
 when the syllable is followed by two consonants, or the consonant 
 which follows is deprived of its vowel and thrown back upon the root- 
 vowel, as in the following instances adduced by Bopp (I. c.) and Rosen 
 (Journal of Education, viii. p. 244*). 
 
 * Rosen asks, "How does it happen that pario deviates from the ana- 
 logy, forming its preterperfect tense pep eri instead of 2)epm?" We answer, 
 Because in this word the r is thrown back upon the root-vowel. 
 
CHAP. I.] 
 
 NOUNS AND VERBS. 
 
 413 
 
 E. 
 
 apiscor \ 
 aptus i 
 
 adipiscor 
 
 adeptus 
 
 amicus 
 
 inimicus 
 
 
 arma 
 
 
 inermis 
 
 harha 
 
 
 imberbis 
 
 
 occiput 
 
 (biceps 
 
 caput 
 
 . sinciput 
 
 < prceceps 
 
 
 principium 
 
 [princeps 
 
 cado 
 
 j cecidi 
 
 \ stillicidium 
 
 
 cano 
 
 cecini 
 
 concentus 
 
 fallo fefelli 
 
 jacio abjicio abjectus 
 
 tuba, cano tubicinis tubicen 
 
 Now it is quite clear that in all these cases the i is introduced into 
 a heavier form than the «, and the e than the i, consequently i is 
 heavier than e and lighter than a. A similar analogy shows that the 
 Latin u is heavier than i. We have shown elsewhere ( Varronian. pp. 
 262 sqq.) that there were three values of the Latin i and u respec- 
 tively. (1) The long τ represents, in composition, the diphthong ai==: 
 ae, as in in-iquus from cequus; (2) the medium i is that which stands 
 for a in the instances given above, and also in inter for av-tSQ (§ 204), 
 in for ανά (§ 170), ille for aXlog (§ 136), &c.; (3) the short ^ approxi- 
 mates to the sound of the shorter u^ and was chiefly used where we 
 should expect e before r and another consonant, as in vir-tus from vir. 
 Again, (1) the long ύ represents the diphthong oi=oe, as in munus=z 
 moenus, and, in composition, the diphthong au, as in in-cludo from 
 claudo; (2) the medium u stands for a Greek ο as in lup-us, λνκ -og; 
 (3) the short ύ is nearly the same as the shortest i, and is chiefly used 
 before I and another consonant, where we should expect e , as in con- 
 culco from calco^ which, according to the table, ought to be con-celco. 
 Now in the first and third cases it is obvious that there can be no 
 diiference in weight between i and u: indeed, t is sometimes written 
 for oi= u^ as in cimeterium for κοιμητήρίον ; and in ob - edio from 
 audio ^ e takes the place of u. But the medium or ordinary u must 
 have been heavier than the medium or ordinary i, for the Greek 
 passes through u into i; compare the Greek τν7ί%-ο-μΒν=^τνΛτ-ο-μΒξ 
 with the old forms sumus, volumus, and their more recent counter- 
 parts in -imus: so also the Greek KauTOQog passes through the old 
 Castorus into the classical Castoris^ and some genitives in -us never 
 became obsolete, as hujus, ejus , unius, &c. Again, in old Latin the 
 vowel of the crude-form is preserved in the inflexions , as in arcu-bus, 
 
414 THE ROOTS OF . [bOOK III. 
 
 op-tumuSj pontu-fex<, &c., in all of which the later Latin exhibits an 
 i (see Lepsius, Paldo graphs p. 53). From these instances we should 
 infer that the medium u is lighter than ο and heavier than i. That 
 u is lighter than ο is farther shown by the change from colere to cul- 
 tuSy from columen to culmen^ though the u here may have been partly 
 occasioned by that affinity between u and /, of which the French 
 furnishes so many examples , and which we see also in the transition 
 from the Greek Άβκλήτΐίος^ 'Ηρακλής to the Latin jEsculapius, Her- 
 cules. "We have perhaps the lightest form of u in the reduplications 
 cucurri, tutudi, &c. ; for a becomes e in the reduplicative syllable, just 
 as 7], ω become l. It is probable that momordi is a corruption of an 
 original memordi or mu-mordi. We observe the same retention of u 
 in Sanscrit desideratives, as in yuyuts, "to desire to fight." In Gothic, 
 a is obviously heavier than w, for we have hulpum, "we helped," in 
 the plural or heavier form, but halp, "I helped," in the singular. We 
 observe the same change from -thus, -tas^ the ordinary dual-endings in 
 Sanscrit, to thus, -tus, in the longer and heavier forms of the preterite 
 (Bopp, Vocalismus, p. 227). In Greek, not only is ο lighter than a, 
 but ω is lighter than 9^ (§ 116) ; and the change from -tog to -εω^ proves 
 that V is heavier than ε. That ν is heavier than l appears from the 
 fact that in the weakest forms of words containing labials, whether 
 the labial is vocalized into ν or not, we find l as the last faint trace of 
 the original form : compare φνω, φνίω, vtog, fio, filius ; hva , J^t, 1'; 
 γραμμα-τεν£^ γραμμα-τίξω; <ivg, 0ίαλος = 6ί-Ραλος (below, § 641); 
 "^(^X^S) τάχιΰτος, &c. The Boeotian substitution of ν for oi shows that 
 V was equal to i and something more. Λ fortiori ν is also heavier than 
 ε. On the whole then it may be stated that the following tables re- 
 present the comparative weight of the Greek and Latin vowels: 
 
 Greek: a; η, ω; a; ο; ν, ι; ε. 
 
 Latin: a; ύ, i; a; ο, u, i; e; m, ^. 
 
 223 (2) Besides the euphonical change produced by substituting 
 a lighter for a heavier vowel in a heavier form, there are three other 
 vowel-changes of very frequent occurrence, which we may term adsci- 
 titious vocalization. 
 
 (a) The first is that which the Sanscrit grammarians call guna 
 and vriddhi; guna or "corroboration" takes place when a is put be- 
 fore either of the last four of the simple vowels i, u, r, Ir; vriddhi or 
 "increment" when a is placed before one of these vowels after it has 
 been guna'a ; thus 
 
 % u r Ir 
 
 guna e 6 ar al 
 
 vriddhi ai au ar al 
 
CHAP. I.] NOUNS AND VEBBS. 415 
 
 That the conversion of the vowels f , Ir into the syllables ar and 
 al is a guntty appears from the fact that e, o, became ay, av before 
 vowels. This insertion of a new vowel into the root is not in itself 
 significant; it is purely dynamical, takes place in verbal roots only, 
 and, like reduplication, &c., is designed to give that extension to the 
 root which is necessary to adapt it for the expression of duration. In 
 Greek the guna is never effected by a, but by ε in the heavier, and ο 
 in the lighter forms. Bopp has rightly Yemarked (Vocalismus, pp. 193 
 foil.) that the Greek av corresponds to a vriddhiy not to a guna of the 
 Vj and that although ai does sometimes stand for the Sanscrit e=ai, 
 it never does so in cases where e is a guna of i, with the exception 
 perhaps of αϊ&ω. At the same time he is not correct in stating that 
 whereas εί and ol are both gunas of t, the only allowable guna of ν is 
 £V, for ύπονδή is a guna as well as ύτίΒνδω^ and άτίόλου^οξ as well 
 as χΒλενχίοξ. The greater weight of the vowel ν is the reason why the 
 lighter vowel ε is preferred in the guna of v. 
 
 (β) The second of the changes to which we allude must be care- 
 fully distinguished from guna. This is when ^ or i; is subjoined to 
 the ci or ε of the root, so as to make apparent guna, as, for instance, 
 when we have νεα-ι-ρα by the side of νεαρά, or ελα-νν-ω by the side 
 of ελά-6αί. In this case the intruder is the second, not the first vowel 
 of the diphthong ; and as this phenomenon takes place in liquids only, 
 we may conclude that it is due to the nature of the liquid which ad- 
 mits a vowel indifferently before or after it. The doubling of liquids 
 to which we have already referred is a phenomenon of the same kind. 
 
 (γ) It seems not unreasonable, after what Lepsius has said (Pa- 
 laographie, pp. 73 foil.), to consider the anusvdra, or nasal insertion, 
 as a part of vocalization. The name anusvdra or "after- vowel" shows 
 that it is reckoned among vowels by the Sanscrit grammarians , and, 
 like the vowel r, it is capable of guna. In the conjugations it serves 
 very much the same purpose as guna, and we find the same root 
 strengthened by guna in one language, and by anusvdra in another. 
 Thus, to take the instances given by Lepsius (p. 79), we have from 
 the Sanscrit chid, Latin scid, cM-na-dmi and sci-n-do by anusvdra, 
 whereas the Gothic makes sk-a-ida by guna, and if α-ϊ%ω is a guna'd 
 form, we may compare it with the Sanscrit anusvdra-iorm. i-n-dh. 
 The Latin is most partial to anusvdra, the Gothic least to, as will be 
 seen from the following instances: 
 
 Sanscrit. Greek. Latin. Gothic. 
 
 (root Uha) lehmi (root λυχ) λείχω lingo 
 
 {λιπ) λείπω linquo 
 
 (str) strnomi (βτορ) ΰτόρννμυ sterno strauja 
 
 (tud) tuddmi tundo stauta 
 
 (uda) νδωρ unda vato 
 
416 THE BOOTS OF [bOOK III. 
 
 The vowel of the guna is softened into i in Gothic , a fact which 
 was first pointed out by Bopp, and to which we shall return when we 
 come to a discussion of the verb-conjugation. We may compare with 
 it the insertion of j before vowels in the Sclavonic languages , some- 
 times instead of guna as in vjemj (Sanscrit vedmi) ; sometimes as an 
 arbitrary insertion; compare jesmj with the Sanscrit asmi. Of this 
 letter insertion we have some remarkable instances in ^olic Greek. 
 Thus, in a Boeotian Inscription (Bockh, No. 1564, 1. 1) we have 
 τιονχαν αγά^αν for τνχΎΐν aya%iiv; in a Delphian Inscription 
 (Bockh, No. 1688, 1. 11), μιφί δώρα δεξίάβ^-ω for δεξάΰΟ'ω; and in 
 the Fragment of Corinna quoted above (p. 249), we have Πινδαρίοίο 
 for Πινδάροιο. 
 
 224 Before we quit this subject, it will be proper to add 
 a few remarks on the absolute significance of roots, on which 
 some very gratuitous assumptions have been made. Roots 
 being the centres around which the words of a language are 
 grouped, the elements from which the noun and verb develop 
 their multifarious forms, the points of convergence from which 
 they spread themselves out with infinite ramifications, it is un- 
 necessary to suppose that they should all have a distinct mean- 
 ing when taken by themselves. The fact is , that most of them 
 obtain a significance, recognisable by the understanding, only 
 when combined with those terminations and flexion-forms which 
 make them into words , and in these words they must be exa- 
 mined if we would know them. 
 
 The root of a word points to the conception, to the selection 
 of some particular quality of the object which makes most im- 
 pression upon us, and by which we classify it with the other 
 objects, possessing or appearing to possess the same quality. 
 Why particular combinations of letters should be chosen for the 
 expression of certain qualities, is a mystery which cannot always 
 be explained. It has been hinted that the three primary posi- 
 tions in space were indicated by the first three consonant-articu- 
 lations, namely, the three tenues, and that these constitute the 
 three fundamental pronouns. Farther than this we cannot go. 
 It appears that certain of these pronominal stems, or modifica- 
 tions of them, have become verbal roots; thus, we have the 
 first in μά-ω = με-νω, and in τΐεράω, Λράγος, &c. (see Greek 
 Grammar, Art. 79); from the second in its two forms Pa and 
 tva^ we have 0ενω, OOog, τί-^ψμι^ κεΐιιαι, δέω, δνω^ &c.; all 
 
CHAP. I.] NOUNS AND VERBS. . 417 
 
 of which preserve the meaning of the pronominal words with 
 which they are connected. We shall see too, that ^άναξ is 
 only the preposition ά-νά, with a suffix, and that there might 
 be some reason for selecting the particular syllables which ex- 
 press the relations of father and mother. But, generally speak- 
 ing, the choice is either arbitrary, or depends upon principles 
 of which it would be idle to seek an explanation. 
 
 225 In considering the roots of words, we must be careful 
 to distinguish them according to the metaphysical or historical 
 differences of the same root. A metaphysical difference between 
 two roots etymologically equivalent, is when they express two 
 ideas connected by the relation, not of resemblance, but of con- 
 trast : an historical difference is when , with the same meaning, 
 they have suffered those systematic changes, which time and use 
 are continually producing upon the consonants of a language as 
 long as it is spoken. It will be recollected that both these dif- 
 ferences are daily taking place : for every man uses every word 
 of his own language according to his own mode of thinking or 
 habits of life; and the pronunciation of words is also subject 
 to continual variation*. 
 
 It is easy to illustrate these remarks by examples. Thus many 
 of our English vulgarisms are merely examples of changes, which take 
 place so regularly in certain languages, that they may almost be 
 reduced under general rules: for instance, the addition of a dental 
 as in gown-d for gown, varmin-t for vermin , Negripon-t for Ig τον 
 EvQLTCOVj is the same variation that appears in hun-d compared with 
 canis, tyran-t compared with tyrannus, &c. Changes occasionally 
 take place in the secondary applications of terms, which have no 
 reference whatever to their original or radical meaning. Sometimes, 
 to adopt Mr. Cobbett's expression , we have the same combination of 
 letters J but not the same word. For example, the word "page," when 
 it signifies the side of a leaf of paper , plainly recalls its origin , the 
 Latin pagina. When, however, it means a youthful domestic, whether 
 at court or in a private family, it is the mutilated representative of 
 the Greek word τίαιδαγώγίον^ "a little τίαιδαγωγόξ^^ i.e. one of those 
 
 * "Adeo enim cerea est vocabulorum natura, ut in ipso loquentium 
 ore aliter atque aliter figureutur." Lobeck, Paralipom. p. 148. 
 
 EE 
 
418 THE ROOTS OF [bOOK III. 
 
 servants who were especially appointed to wait upon the young 
 master. The French term chetif, and our old "caitiff," are derived 
 through the Italian cattivo from the Latin captivus, as if all baseness 
 and misery were the natural result of misfortune in war. And thus 
 the Sclavonians , whose name signifies "glorious," "illustrious," and 
 the root of which constitutes the key-note to the laconic epinicium of 
 their celebrated warrior Suwarrow* , from having merely furnished a 
 large number of prisoners of war, have given us our modern name 
 "slave," esclave, schiavo. The Bulgari (according to the French pro- 
 nunciation Boulgres and Bougres) owe the horrible degradation of 
 their national name to their early connexion with heresy (Gibbon, x. 
 p. 177 Milman). The word Gypsy, which signifies "vagabond," and 
 "impostor," is merely a corruption of Egyptian; and in the land of 
 the Nile itself, gins el Farauni, "descendant of Pharaoh," is an 
 abusive designation of Christians. Perhaps the most remarkable in- 
 stance of these changes is furnished by the adjective τΐολως. There 
 can be no doubt that its primary meaning is whiteness superinduced 
 on a surface previously regarded as darker. It is therefore connected 
 in origin with Λελυός, "swarthy," i. e. imperfectly black, πελίδνόζ, 
 "livid," and the Latin pullus and ρ allidus. Most usually Ttokiog is 
 applied to hair , which has become white , or to old age as indicated 
 by white hairs; then it is applied to animals, e.g. wolves (τΐολίόί 
 λνκοί) or kids (ΛολίοΙ 'εριφου), which have white hairs in their coats ; 
 then it is applied to the sea, as broken into white foam (TCoXiog τίόν- 
 Tog)', to the air as bright with light (πoλώg αίϋ'ήρ)', to the spring 
 {πολίον εαρ) as a bright season in contrast to winter; to metal when 
 it reflects light from its polished surface (ττολώ^ uLdfjQog, xakaog); and 
 finally to the human skin when it appears white from distension ; for 
 it is clear that in Pindar, Tyth, iv. 98 : rig ανθρώπων 6ε χαμαιγενεων 
 noUag εξανηκεν yauxkqog; we cannot understand the epithet as refer- 
 ring to the age of Jasons's mother, either as implying that he was τψ 
 λvγετog, which is Hermann's idea, or as conveying a sarcasm, which 
 is Bockh's interpretation; but must explain the phrase by a reference 
 to the whiteness of the distended skin (cf. Pers. iii. 98: albo ventre la- 
 vatur. Hor. 2 Serm. n. 21 : pinguem vitiis albumque, and our phrase 
 "a white swelling"). From the word geist or geest, we have both gas, 
 which represents the highest flight of modern science, and ghost, which 
 
 * Suwarrow's letter to the Empress Katharine on the taking of Ismael 
 was the following couplet: 
 
 "Slava Bogu! slava vam! 
 Krepost vzata, y ia tarn." 
 "Glory to God, glory to thee! The fortress is taken, and I am there." 
 
CHAP. I.] NOUNS AND VEEBS. 419 
 
 suggests the most degrading of medieval superstitions. The word 
 "quarrel" leads us back through querelle to querela^ which means a 
 complaint from the weaker or injured party; but our English word 
 signifies rather to take a high ground in asserting one's rights , and 
 even to assume the initiative in a dispute. These meanings find their 
 common ground in the forensic application of the term : for the humble 
 complainant is naturally antecedent to the litigious suitor. When we 
 speak of a tapster in modern English, we always imply a burly cellar- 
 man ; but our ancestors left this office to women , and tapster is the 
 regular feminine of tapper, as spinster is of spinner'^. Many of these 
 feminine forms are preserved only in proper names, as Baxter^ 
 Brewster^ Sangster, Webster ^ &c., but they are not the less genuine 
 remnants of ancient employments of the weaker sex , which are now 
 more appropriately transferred to men. The examples which we have 
 given, and to which almost any number might be added, are sufficient 
 to show that even an obvious etymology has often no connexion with 
 the existing acceptation of a word; and from this the inference is 
 plain, that the dissection of words , though uniformly valuable as 
 a department of grammar, is not the only source of information which 
 the lexicographer must render available to his purposes. 
 
 226 When we wish to dissect a word in order to arrive 
 at its primary element or root, our first object is to inquire with 
 what other words it agrees in termination or prefix. The latter 
 is stript off at once, but the removal of the affix is often a double 
 operation. To take that set of words called nouns, with which 
 we are in the present part of this work more immediately con- 
 cerned , we find that every one ends with a short termination, 
 often a single letter, which marks its immediate relation to the 
 other objects in connexion with it, and which we call the case- 
 ending. But, in the majority of words, we find, between this 
 and the root, an affix consisting of one or more pronominal stems, 
 which marks the definite class and quality of the noun, and 
 points out the restriction with which the general force of the 
 
 * This feminine termination -ster, which appears in its fullest form in 
 the Sanscrit stri^ "a woman," and with a loss of the t in the Latin so- 
 ror=so-sor and uxor=ug'Sor=jug-sor {Journal of Philolog. ii. 357), will not 
 explain all English words with the same ending. Thus, to say nothing 
 of words derived from the Latin, like monster from monstrum, we have 
 dempster for doom-master, gamester {= gamesome, in Shakspere) from game- 
 master, &c. after the analogy of " master of the requests," " master of the 
 revels," &c. 
 
 E!E2 
 
420 THE BOOTS OF NOUNS AND VERBS. [BOOK III. 
 
 roots is applied in the particular instance. When the case-end- 
 ing alone is removed, the remaining part of the word is called 
 its crude-form, whether it has another pronominal affix or not. 
 In most nouns the crude or uninflected form must be still farther 
 denuded before we can arrive at the root or skeleton of the 
 word. Accordingly, in the following analysis of the noun, we 
 have first considered the case-endings or absolute terminations 
 of the noun , and have then examined those pronominal inser- 
 tions before the case-ending, which may be considered as the 
 terminations, not of the noun, but of its crude-form. The young 
 student will thus more clearly discern by what successive steps 
 he must proceed in dissecting any given noun in order to arrive 
 at a definite conception of its meaning, so far as the significa- 
 tion has remained unaffected by the arbitrary or capricious ap- 
 plications to which we have adverted. 
 
CHAPTER Π. 
 THE CASE-ENDINGS OF THE NOUN 
 
 227 Definition and arrangement of cases. 228 Formation of the feminine 
 noun. Criticism of Bopp's opinion. 229 Reasons for the assumption of 
 the feminine gender in certain nouns. 230 Dual number. 231 Sanscrit 
 case-system. 232 Latin and Greek Declensions. 233 (a) Latin case-forms. 
 234 (b) Greek case-forms. 235 Origin and mutual relations of the cases. 
 236 Detailed examination of the cases : (i) Accusative. The dental η its 
 original affix. 237 Neuter nouns have no nominative. 238 Distinction 
 of the accusatives and locatives in n. 239 Accusatives plural of mas- 
 culine and feminine nouns. 240 Origin of neuter plurals in a. 241 
 Dual forms. 242 (2) Nominative. The sibilant s its proper suffix, but 
 often absorbed or dropt. 243 Explanation of the nominative sign. 244 
 Nominative plural formed by a reduplication of this affix. 245 (3) Im- 
 plementive or Instrumental, b. Dative, c. Locative. Sanscrit forms. 246 
 Greek Dative includes these cases. 247 (4) a. Ablative. Its form in 
 Sanscrit and Latin nouns and in Greek adverbs. 248 (4) b. Genitive. 
 Only a longer form of the ablative. 249 Identity of the terminations 
 -Ο'εν and -olo-v. 250 The genitive plural formed from the genitive sin- 
 gular. 251 (5) Vocative. A mere crude-form. Its accentuation. 
 
 227 Τ\7ΉΕΝ we say that a noun is the name of a thing, we 
 mean that it is a word by which we express our 
 conception of some object; now the conception of a natural 
 object is the recollection of the most prominent quality or at- 
 tribute which we have perceived in it; the name, therefore, 
 points out or refers to this quality or attribute. We have shown 
 in the last chapter that the part of a noun which conveys its 
 meaning to our ear, and which is called its stem or root, never 
 appears by itself in those languages which have inflexions; even 
 the crude or uninflected form is never found alone, except when 
 it stands as the vocative case. To the crude-form, in ail other 
 instances, is affixed a termination, which constitutes it a word, 
 and gives it the signification of a noun ; for the same root, with 
 a difi'erent termination, and perhaps slightly modified, might be 
 a verb. These endings, which make the crude-form into a noun, 
 and which we call the case-endings , it is now our business to 
 discuss separately and in detail. The designation "case," casus, 
 i. e, "falling," is derived from the Latin version of the Greek 
 Ttrooig. Now this word is used by Aristotle to signify not only 
 a case of the noun, but any inflexion either of a noun or a verb, 
 and indeed any word-form, whether declinable, as a comparative 
 in -τερο^, or indeclinable, as an adverb in -ωg. Nay more: not 
 merely forms of words, but even forms of sentences, are, accord- 
 
422 THE CASE-ENDINGS [BOOK III. 
 
 ing to his phraseology, Λτώύεις λόγου j see Aristot. (?) Poet. 20, 
 10; and for the πτώΰις λόγου, compare Topic, vi. 10, 1: ετί si 
 των ομοίων του ονόματος Λτώΰεων ai αμοίαι του λόγου Λτώΰει,ς 
 εφαρμόττουΰον ^ οίον εΐ ώφέλψον τα τϋοιητικον νγυείαξ κάί ώφε- 
 λυμον το Λετίοίηκος υγίειαν^ where we have a change in the 
 sentence introduced by a change of tense. In this wider sense 
 of the word τίτώΰυξ.) it seems to approximate in meaning to the 
 word τίτώμα, and to signify the accidental state or condition of 
 an object presented to the senses. Hence the old logicians, ac- 
 cording to Plutarch, used τίζώβΐξ as a synonym for όνομα, 
 namely, as signifying whatever was the subject of a predication; 
 Qucest. Platon. 1009 c, p. 108 Wyttenb.: τούτο δε (sc. the τΐρώτος 
 λόγος, formerly called τίρόταόυς or "proposition," and after- 
 wards αξίωμα or "enunciation") εξ ονόματος καϊ ρήματος 
 ΰυνέΰτηκεν (above, § 124), ών το μεν τίτώύιν διαλεκτικοί, το δε 
 κατηγόρημα καλουΰιν. In this sense we call that part of 
 grammar which refers to the forms of individual words by the 
 name of "accidence" (accidentia), which seems to point to the 
 Greek ΰύμτττωΰίς. Supposing then that πτώΰις originally de- 
 signated any change of form to which the individual word was 
 liable, it is easy to understand the transition by which the Stoics 
 limited its use to the declensions of the noun. For όνομα, in 
 its logical sense, was equivalent to τΐτώΰίς: accordingly, when 
 oVo^a was merely "the noun," τζτώΰείς would designate merely 
 the inflexions of the noun. But along with this limited applica- 
 tion the Stoics introduced a different explanation of the term. 
 The Peripatetics understood by it merely the corpus mortuum 
 of an individual word, the όνομα, before it was vivified by its 
 connexion with the ρήμα, and became a part of the λόγος. But 
 the Stoics considered the nominative as indicated by a per- 
 pendicular line, from which the other cases fell away or de- 
 clined at different angles. Hence the nominative was called the 
 πτώΰίς ορ^-ή or ευθεία, i.e. casus rectus, and the others the 
 7ίτώ<5ευς τιλάγυαι, i. e. casus obliqui (see Diog. Laert. Vita 
 Zenonis, vii. 65). That this, however, was not the original 
 meaning of πτώΰίς, is sufficiently shown by the objection of 
 Georgius Chocroboscus (ad Theodos. pp. 9, 35 Gaisford): δηλον 
 οτι η ευ^'εΐα ουκ εύτΐ πτώβις κυρίως' ει γαρ ην κυρίως τΐτώΰις 
 εν παρα&έΰεί είχεν είναι, μετά των ηρο^εύεων. How Chrysippus, 
 
CHAP. II.] OF THE NOUN. 423 
 
 in his book TtEql των πέντε πτώΰεων^ would have dealt with 
 this difficulty, we have no means of knowing: but in all pro- 
 bability the original and secondary meanings of the term were 
 somewhat blended and confused. It is clear that Choeroboscus 
 did not understand the terms ορ^ή and svd'ela as opposed to the 
 term τίλαγία, for he says (u. s. pp. 10, 26): del γίνώόκειν οτι η 
 μεν ορ^η ονομαΰτίκη λέγεται καΐ ευθεία' %αί ορ&η και ένδεια 
 λέγεταί, επειδή ορ^'ώς ΰημαίνεϋ την ονΰίαν του πράγματος- — 
 ονομαστική επειδή δι αντης τας ονομαΰίαξ ποιούμεθα*. 
 
 We have before adduced reasons for the opinion, that the 
 accusative or objective case is the primitive form of the pro- 
 nouns; the same, we believe, holds with regard to the nouns: 
 for if the primary expression of self is objective, much more so 
 must be that of any object in the external world. The primary 
 noun is the object of speech, and the correlative expression for 
 the subject is necessarily a subsequent abstraction. In syntax 
 — that is, according to the logical arrangement — there are only 
 three forms of the objective case, expressed by the Latin adverbs 
 unde, ubi, quo, and corresponding to the Greek genitive, dative, 
 and accusative, as the cases of motion from, rest at, and motion 
 to, respectively. Our present object, however, is not the syntac- 
 tical explanation, but the etymological origin of the forms. 
 In analyzing the cases, therefore, we shall consider the accusa- 
 tive or general objective case first; the others we shall discuss, 
 as nearly as possible, in the order in which they are placed in 
 Sanscrit, which has the fewest prepositions, and therefore the 
 most complete case-system of any of the languages with which 
 we are immediately concerned•}•. 
 
 In Sanscrit there are three genders, masculine, feminine, and 
 
 * On the subject of the dispute between the Peripatetics and Stoics 
 respecting the applicability of πτώϋίζ as a designation of the nominative, 
 the reader may consult Ammonias, p. 104 Brandis. And for the designa- 
 tions by which the cases were known to the Greeks, from whom we have 
 borrowed them, see Choeroboscus, I. c. There is a paper on the γενική 
 τιτώΰις by Schomann in Hofer's Zeitschrift, i. 1, pp. 79 sqq. 
 
 t In Finnish there are fourteen cases besides the nominative, namely, 
 seven simple and seven composite cases (Pott, Etym. Forseh. zw. Aufl. 
 p. 6); but it is clear that this diffusive enumeration is due rather to a 
 confusion than to a scientific distinction of forms and significations. 
 
424 THE CASE-ENDINGS [bOOK III. 
 
 neuter; three numbers, singular, dual, and plural; and eight 
 cases, nominative, accusative, implementive or instrumental, da- 
 tive, ablative, genitive or possessive, locative, and vocative. 
 
 228 Of the feminine and neuter genders, as distinguished 
 from the masculine, Bopp says rather quaintly ( Vergl, Gramm. 
 p. 135): "In Sanscrit the feminine, as well in the stem as in the 
 case-endings, loves a luxuriant fulness of form, and vrhere it is 
 distinguished in the stem or in the ending from the other genders, 
 it is marked by broader, more sounding vowels. The neuter, 
 on the contrary, loves the utmost brevity, but is distinguished 
 from the masculine, not in the stem, but only in the most pro- 
 minent cases, in the nominative and in its perfect opposite, the 
 accusative, also in the vocative, where this is the same as the 
 nominative." The fact is, that in order to mark more strongly 
 the relative and collective nature of things conceived as feminine 
 or maternal, the merely subjective s of the nominative masculine 
 is generally expanded by a broad vocal utterance into a form 
 which reproduces the collective value of the same element 
 (§152); whereas the neuter, which has no nominative, appears 
 only in the objective case, which is most liable to mutilation. 
 This explains the circumstance that, in masculine and neuter 
 nouns, the vowel which terminates the crude-form, and to which 
 the case-ending is attached, is generally and properly short; 
 while in feminine nouns , the vowel is long. There are excep- 
 tions to this rule, more frequently however in Greek than in 
 Sanscrit. Thus, instead of the ο which stands for the Sanscrit 
 masculine a in λόγ-ο-ξ^ &c., we have a long α or τ; in παιδο- 
 τρίβη-ς^ &c., and in the numerous class of nouns ending in -της. 
 In our opinion the η here, as elsewhere, includes the lost ^, 
 which is used to form derivative verbs, and which seems by no 
 means out of place in words expressing an action, as the nouns 
 in -ηςΒ,ηά -r?^g invariably do: and thus τιαίδοτρίβ-ης, ενεργέ-της 
 are equivalent to παίδοτρίβ-τ/ας•, ενεργέ -tyag, just as the corres- 
 ponding verbs would be παυδοτρυβέω (τίαίδοτρίβι/ω), ενεργετέω 
 {ενεργετνω). That we have here the second pronominal ele- 
 ment under the form τί,-, appears more clearly, and throughout 
 the cases, in the feminine forms of nouns in -της; compare 
 ηροδό-της, τιροδό-τις {-τιδ-ς)\ ικέ-της, ίκε-τις (-τίδ-ς) &c. In nouns 
 
CHAP. Π.] OF THE NOUN. 425 
 
 like ταμ-ίά-ς^ root -ταμ, the second element is clearly seen under 
 the double form -La=La-6cc. A different explanation is re- 
 quired in the case of those nouns in -ης of which the crude-form 
 is -£g=-og=-oT•, for example, ^ημοΰ^'ενης, crudeform ^ημό- 
 6%'£veg (below, §251) from δήμου od'ivog. Here it seems ob- 
 vious that the new masculine nominative is formed by adding 
 -g with, of course, some connecting vowel, to the crude-form 
 'Sg, probably another -sg; the sibilant at the end of the crude- 
 form naturally becomes evanescent between the two vowels; 
 and these vowels are represented by η; so that ^ημοΰ%'ένης= 
 ^ημοόΟΈνεό-ες becomes ^ημοό^Ένε-ες = ^ημοΰΟ'ενης. Bopp's 
 remarks ( VergL Gramm. p. 139) on the long t, which appears 
 most frequently in Sanscrit as the characteristic addition for 
 the formation of stems of the feminine genders , seem to us to 
 be altogether erroneous. According to the principles developed 
 in a preceding chapter, the vowel i is a secondary form of the 
 sibilant s. It appears therefore to us a sort of philological 
 soloecism to say that -ύα is a corrupt and more recent form of 
 -i,. On the contrary, we are convinced that the oldest and most 
 genuine method of forming the feminine from the masculine 
 is that which is still preserved in many Greek words, namely, 
 by substituting -ΰά for the -g of the masculine nominative. The 
 participial words cited by Bopp furnish simple examples. Thus, 
 from the masc. χαρίείς=^χαρίεντ-ς, we have the fem. χ(χρίεΰ-ΰα= 
 χαρίεντ-6α; from, τντΰτων^τνπτοντ-ς and δίδονς=^δίδόντ-ς, we 
 have TV7ttov-ua, δ^δον-ΰα; from δείκννς=δείκ-ννντ-ς, we have 
 δείκνν-6α, and so forth. We consider the forms in -La, -ρα to 
 be only secondary states of these original forms in -6a, whereas 
 the forms in -lcc, -ρά are contractions of -La -ΰά , and -ρά-βά , in 
 which a formative syllable is inserted, just as in the nouns in 
 -d'-g, -δ-ς; for we have words in -τριά=-τρία(3α by the side of 
 words in -τρι-δ-ς. That the same is the case in the words which 
 end in -η^ namely, that these are contracted from older forms 
 in εα=^ά-6α, appears from words like ύνκέα, ΰνκη, where the 
 uncontracted form is still extant. The feminine adjectives μέ- 
 λαινα, &c. merely exhibit the secondary forms μελάν-ια, &c. 
 with the absorption and compensation noticed above (§ 215, c); 
 and the same is the case with nouns like χλαίνα , which have 
 no corresponding masculine forms. In nouns like μέρψνά, 
 
426 THE CASE-ENDINGS [BOOK III. 
 
 ϊχυδνα, which we must compare with ηάτνια•, τίότνα, the l of 
 the termination is either lost or appears in the penultima only, 
 while it is represented by the doubled λ or ν of άμιλλα, αελλα, 
 KOQLVva^ &c. (above, § 215, δ), and contained in the ζ of φζα, 
 according to the proper power of that letter (above, § 216). 
 In the words which end in -^a and -da we must consider these 
 dentals as representing an original <3- (above, § 149). As the 
 feminines τέκταίνα, Αάκαονα, stand by the side of masculines in 
 -ων, which in the one case represents ov-g and in the other -ων -g, 
 we cannot consider them as entirely analogous to μέλαινα, &c., 
 unless we presume obsolete masculines in -av or -ag. The ethni- 
 cal name Άκαρνάν would justify an original Αακάν=Αακψ, of 
 which Αάκων is after aU only a lighter form (§ 116): and the 
 verb, τεκταίνω, points to an original τεκτΐί}ν=τέκτων, cf. φρήν, 
 ευφραίνω, &c. The same assumption of obsolete masculines is 
 also required by λέαινα, λέαινα and is easily justifiable: for the 
 Tiτάvεg presume as their opponents the Θεάvεg; and Θεάνω must 
 be derived from Θεάν, or Θεavεvg: the extant λL•g, accusative 
 λΐν, may lead us to an original λέαν, of which the participial 
 λέοντ- for λάοντ- is a by-form; compare μονΰα with the parti- 
 ciple μώύα from μάω. The words λνκάων and λvκά-βag would 
 suggest λνκάν, according to the combinations noticed by Miiller 
 (Dor. 11. § 6), and from this masculine, λνκαίνα would be the 
 analogous derivative. With regard to the very peculiar form 
 δί(5ποινα,^ο> must remark that dB<3-7t6trig and τΰότ-νια correspond 
 to the Sanscrit ^αί^s, "a master," and pat-ni, "a mistress" 
 (Rosen, Journal of Ε due. viii. p. 346), and consequently, that 
 we need not trouble ourselves to find in jtotvia the feminine for 
 δεΰτίότης. The analogy of Ο^εράτΐων , ^εράτίαινα would conduct 
 us to an obsolete δελτίων, a degenerate participle, of which we 
 have other examples. It may seem an open question whether 
 we are to explain «VacJija, βαΰίλίΰΰα, Ο-άλαύΰα, Θραΰύα, με- 
 λίΰϋα^ &C. , with reference to the primary form in -ΰα or to the 
 secondary form in -t«. There can be no doubt that Θραΰβα 
 might result from Θράκ-υα^ βαόίλιΰΰα from βαύιλίδ-ια, and so 
 forth. But the analogy of Ulyxes^ compared with Όδv<5<5εύg^ 
 might justify our assumption of the original -6a in the case of 
 the gutturals, and the assimilation of the dentals δ-, t-, to a fol- 
 lowing 0- might seem not imnatural. As, however, we have 
 
CHAP. II. J OF THE NOUN. 427 
 
 seen that the barytone verbs in 6ΰ exhibit the assimilation of a 
 guttural or dental succeeded by i=y•, it is more reasonable to 
 suppose that the same contact has produced the same result in 
 the feminines in -06a ; for it would be strange if the explanation 
 of άνάΰβω did not apply also to ava6(5a, especially as the future 
 άνάξω and the dative αναξί are opposed to the assumed assimi- 
 lations of κ or κτ to 66. When we see the termination -lcc thus 
 brought back by contact or assimilation to the form -6cc from 
 which it originally started, we seem to prove our etymological 
 rule by a process of inversion which is so frequently applicable 
 in arithmetic; and on the whole we cannot but regard Bopp's 
 explanation of these feminine forms as singularly deficient in 
 critical tact and accuracy. That the d or c of the Greek and 
 Latin feminines is not unorganic, as he supposes, will appear 
 in the next chapter. 
 
 229 If it be inquired what is the reason why so many in- 
 animate objects are called by names which are considered mas- 
 culine or feminine, it will be sufficient to answer, that this may 
 have arisen partly from the idea of comparative strength or 
 weakness (Hermann, de Emend, rat. Gr. G7\ p. 125), partly also 
 from association; for if one word of a class be considered as 
 feminine, all other words of a similar signification would be so 
 considered likewise. For a great many words the gender depends 
 upon something included in the idea of the word; a tree, in 
 reference to its branches, and most collective words, would be 
 feminine, from the included idea of mother (comp. Buttmann, 
 Ausfuhrl. Sprl. § 32, Anm. 3). It is for this reason, we con- 
 ceive, that η LTtTCog signifies "a body of cavalry" (Thucyd. I. 
 62), η βονς, "a herd of oxen" (Thom. Mag. in v.)^ and η κάμη- 
 λος, "a troop of camels" (Herod. I. 80). We observe the same 
 collective meaning in πέτρα, "a rock," i.e. a collection of stones, 
 as opposed to πέτρος, "an individual or single stone" (§ 15, 
 note) ; also in χώρα, an extensive tract of country, as opposed to 
 χώρος, and its synonym χορός, which signify "any separate 
 piece of land not built on," i. e. either the open space in a town, 
 which is the proper meaning of χορός (Theatre of the Greeks, 
 ed. 6, p. [11]), or a field in the country, which is the ordinary 
 signification of χώρος: so Herod, ii. 154: δίδωΰί χώρους Ινοικψ 
 
428 THE CASE-ENDINGS [BOOK III. 
 
 ύαί: cf. I. 126. We might say that χώρα, "a territory," was 
 an adjective agreeing with the suppressed noun γη, and that 
 χώρος referred to αγρός ; but there is no occasion to call in this 
 machinery. The diminutive χωρίον of course belongs imme- 
 diately to χώρος. The large meaning of χώρα is still farther 
 shown by its use to denote the room or space, the vacans provin- 
 cia, which ought to be filled by some one: see Xenoph. Anah. 
 IV. 8, § 15: επειδή εν ταΐς χώραις εκαύτου εγένοντο, and cf. 
 Blomfield, Gloss, ad Agam. j^schyli, 77. 
 
 230 We have remarked that the Sanscrit nouns are in- 
 flected in the dual as well as in the singular and plural numbers ; 
 the same is the case with the Greek , and , to a certain extent, 
 with the Gothic. There is every reason to believe that, in 
 Greek at least, this dual is nothing more than an older and 
 weaker form of the plural, restricted in the newer and more 
 refined speech to the expression of two instead of more; for, 
 first, in many of the pronouns we find the oldest forms of the 
 root in the dual number; secondly, we find in Homer, and in- 
 deed in later writers, this dual form used as a plural; lastly, we 
 have the analogy of our own and other languages in support of 
 the opinion, that of two forms of the same word, the older may 
 be confined to vulgar use as a plural , while in the more po- 
 lished language it is restricted to the number two (comp. Butt- 
 mann, Ausfuhrl. Sprl. § 33, Anm. 1; Fenny Cyclop, article 
 Dual Number). Besides, it appears that some old grammarians 
 considered the forms dixere^ &c. for dixerunt, &c. as duals 
 (Quintil. I. 5, § 42); on the other hand Cicero {Orator, c. 47) 
 admitted them as allowable, though antiquated, forms of the 
 plural. 
 
 231 As we are about to base our detailed inquiry into the Greek 
 cases upon the more complete case-system of the Sanscrit language, 
 we may introduce the subject by laying before our readers an exam- 
 ple of the declension of some simple and regular noun in that lan- 
 guage. The name of the divinity, giva, is thus declined : 
 
 
 
 Sing. 
 
 Dual. 
 
 Plur. 
 
 1 
 
 Nominative 
 
 givas 
 
 ςιυαη 
 
 givas 
 
 2 
 
 Accusative 
 
 givam 
 
 do. 
 
 ςίνάη 
 
kV 
 
 .11.] 
 
 OF 
 
 THE NOUN. 
 
 ί 
 
 
 
 
 Sing. 
 
 Dual. 
 
 Plur. 
 
 3 
 
 Implementive or 
 instrumental* 
 
 } 
 
 givena 
 
 gwdbhydm 
 
 givais 
 
 4 
 
 Dative 
 
 
 givdya 
 
 do. 
 
 givebhyas 
 
 5 
 
 Ablative 
 
 
 givat 
 
 do. 
 
 do. 
 
 6 
 
 Genitive 
 
 
 givasya 
 
 givayos 
 
 givdndm 
 
 7 
 
 Locative 
 
 
 give 
 
 do. 
 
 giveshu 
 
 8 
 
 Vocative 
 
 
 giva 
 
 givau 
 
 givds 
 
 429 
 
 It is not necessary to remind the student of Sanscrit that every 
 final s in this scheme is changed by visarga into h , and every m by 
 anusvdra into the nasal n. 
 
 The crude-form of the word which we have taken as an example 
 is giva, which ends in short a. Separating this from its affixes, we 
 have the following scheme of case-endings for a noun the crude-form 
 of which is terminated by a. 
 
 
 Sing. 
 
 Dual. 
 
 Plur. 
 
 1 
 
 -S 
 
 -6= a — u 
 
 -as 
 
 2 
 
 -m 
 
 do. 
 
 -an 
 
 3 
 
 -ina 
 
 -abhydm 
 
 -is 
 
 4 
 
 -a-ya 
 
 do. 
 
 -ibhyas 
 
 5 
 
 -a-t 
 
 do. 
 
 do. 
 
 6 
 
 'Sya 
 
 -yds 
 
 -andm 
 
 7 
 
 -i 
 
 do. 
 
 -ishu 
 
 8 
 
 crude-form 
 
 -u 
 
 -as 
 
 It will be remembered that this is only one of many forms of declen- 
 sion in Sanscrit, and that even this form diflfers when applied to 
 feminine or neuter nouns. In most of the other declensions the 
 instrumental and dative very nearly resemble one another: thus, the 
 dative of dhard, *' earth," is dhardydi, and the instrumental dharayd; 
 the dative oi prttis, "love," is pritaye, the implementive is prityd, 
 and so forth. To this we shall recur hereafter. The general form 
 of the cases in other than the short a declension may be derived from 
 the following paradigm (vide Bopp, Grammatica Sanscrita, p. 85, or 
 Kritische Grammatik, p. 82): 
 
 Singular. Dual. 
 
 Nom. s masc. fern, w neut.f au m. f. i n. 
 Ace. w, am m. f. m n.f au m. f. i n. 
 
 Instr. a m. f. n. a m. n.f bhydm m. f. n. 
 
 Plural. 
 
 as m. f. n. i n. 
 
 Sf as m. f. η m. i n. 
 
 bhis m. f. n. 
 
 * Wilkins calls this case the implementive; Bopp and other Germans 
 term it the instrumental. We have mentioned both, but we much prefer 
 the latter designation. 
 
 f Only in the a declension. 
 
430 THE CASE-ENDINGS [BOOK III. 
 
 Singular. Dual. Plural. 
 
 Dat. e m. f. n. ai f. ay a m. n.* bhydm m. f. n. bhyas m. f. n. 
 
 Abl. t m. n.* as m. f. n. bhydm m. f. n. bhyas m. f. n. 
 
 s m. f. as f. 
 
 Gen. sya m. n.* as m. f. n. os m. f. n. am m. f. n. 
 
 s m. as f. 
 
 Loc. i m. f. n. dm f. 05 m. f. n. su m. f. n. 
 
 232 It is customary to divide Greek nouns according to three, 
 and Latin nouns according to one consonant and five vowel declen- 
 sionsf. The differences which constitute the declensions are differ- 
 ences of root and crude-form, not differences of termination. The 
 case-endings must have been originally the same for all nouns ; indeed 
 we can observe striking resemblances between them even in the Greek 
 language, as it exists in the writings which have come down to us. 
 For instance, the dative singular and the dative and genitive plural 
 are always distinguished by the same endings, as is generally the 
 accusative singular also. A writer in the Journal of Education (Yol. 
 V. p. 19) remarks, "That at one period of the language, probably 
 prior to any written books that have come down to us, all the Latin 
 and Greek nouns had an incremental syllable in the genitive and 
 oblique cases, we consider to be nearly demonstrable." We do not 
 think there are sufficient grounds for this generalization. There is 
 no good reason, so far as we know, for supposing that, in the oldest 
 state of the language, the accusative was ever a longer form than the 
 nominative. If we consider the Latin and Greek nouns in the oldest 
 forms which we have of them, and extend our observations by ana- 
 logy to all cases, we shall have the following schemes for the case- 
 endings in Latin and Greek. It will be recollected that we take 
 merely the case-endings, and do not trouble ourselves with the roots 
 or crude-forms of particular words. 
 
 233 (a) Latin declension. 
 
 Sing. Plur. 
 
 {sometimes absorbed, 1 
 assimilated, or dropt > -[s]es / ^^.^ ^ 
 by visarga J ^ 
 
 !the singular 
 m constantly 
 absorbed 
 
 * Only in the a declension. 
 f The student will find the Greek declensions arranged according to 
 our views in the Greek Grammur, 157 — 193, and the Latin in the Var- 
 ronianus, pp. 293 sqq. 
 
ΪΑΡ. II. 
 
 J 
 
 OF THE NOUN. 
 
 Dat. and Loc. 
 
 Sing. 
 
 -i or -ibi 
 
 Plur. 
 
 -ibus or 'ebos 
 
 Abl. 
 
 
 -d or -tus 
 
 
 Gen. 
 
 
 -ts^ -jus, -sis 
 
 Example. 
 
 -[r]um 
 
 Nom. 
 
 Accus. 
 
 Dat. 
 
 Abl. 
 
 Gen. 
 
 Sing. 
 
 lapi[d]-s 
 
 lapid-e-m 
 
 lapid-t-[b{]=lapid-i 
 
 lapid-e-[d] 
 
 lapid-is 
 
 , Plur. 
 
 lapid-ls^es 
 
 lapid-e[m]s=lapides 
 
 lapi-ibus 
 
 do. 
 lapid-e-rum 
 
 431 
 
 It is not necessary for our present purpose to consider the differ- 
 ences of gender in the Latin noun, but our scheme for the Greek 
 case-endings must have regard to these alterations. 
 
 234 
 
 (b) Greek declension. 
 
 
 
 
 
 Singular. 
 
 
 
 Masc. 
 
 
 Fern. Neutr. 
 
 Nom. 
 
 
 -s 
 
 
 ί-ΰα,-ιά, -&-g,-d-s\ 
 
 Accus. 
 
 
 -V 
 
 
 -OaVi -O-a, -da, &c. -v, -τ 
 
 Dat. 
 
 Ablat. 
 
 Gen. 
 
 or\^ 
 
 Masc. 
 
 d'BV, -1 
 
 -ua-L=6yj &c. same as masc. 
 ως -6α-ίον=ΰης, &c. do. 
 
 Plural, 
 
 Fern. Neutr. 
 
 Nom. 
 
 
 -(J£S 
 
 
 -ua-sg wanting 
 
 Accus. 
 
 
 -vg 
 
 
 -<5av-g, &e. -v-t=cc 
 
 Dat. 
 
 Ablat. 
 Gen. 
 
 1 
 
 -i-ui-v 
 
 '6L0V-g, 
 
 -ων 
 
 -ua-L-oLV, &G. same as masc. 
 -6a-L0V-g, ΰα-ων, &c. do. 
 
 Nom. 
 Accus. 
 Dat. 
 Abl. or I 
 Gen. / 
 
 Examples. 
 {a) Masculine. 
 
 Sing. 
 
 K6yo-g 
 
 λόγο-ν 
 
 λόγω =λόγο-φν 
 
 λόγον=λυγο-ίθ \ . , ^ , 
 
 ==λογό-0ω-ν / ^^^^^^ λογων=λογο-αον-ς 
 
 Dual. 
 
 Ι λόγω 
 
 Plur. 
 
 λόγο -L =λόγo-6εg 
 λόγους = λόγο -vg 
 λόγο-ιΰυ-ν 
 
432 THE CASE-ENDINGS [bOOK III. 
 
 (β) Feminine. 
 
 Sing. Dual. Plur. 
 
 Norn. uov-6a \ , _ μον-(5αι^μον-(5αίζ 
 
 Ace. μον-6αν I ^ μον-6ας=μον-ΰαν-ς 
 
 Dat. μον-ΰγι μον-όαιΰί-ν 
 
 =μον-ΰα-ν λ 
 
 =μού-6ΰί-φο }μον-ΰαίν μου-ΰών =^μον-6άων 
 
 ύ-6ης ) 
 Gen. } ==:μον-ύά-ιον ^=μον-0α-ιον-ξ 
 
 {γ) Neuter. 
 
 Nom. \ , , , f / 
 
 > κερα-^^^κερεντ κερατ-ε κερατ-α ^=%BQBVt-VT 
 
 Dat. κ8ρατ-ί=κέρεντ-ί]κεράτ-θίν κέρα-ΰο =^κερεντ-ε66ι 
 Ah\. ογ\κέρατ-ος ' κεράτ-ων=^κερεντ-ε6ίον-ς 
 
 Gen. f =κερεντ-ε6ίνον 
 
 235 If we examine these cases, in the fullest forms to which they 
 can be expanded, we shall not have much difficulty in convincing 
 ourselves that the following is the simplest account of their origin 
 and mutual relations. While the nominative is merely designated by 
 the subjective -s, as a remnant of the pronoun expressing relative 
 proximity, the accusative is simply marked by the objective -w, as a 
 residuum of the pronoun denoting distance. A combination of these 
 two is used to form the genitive or case of removal, which thus indi- 
 cates motion from the near to the distant, and the dative, or case of 
 rest , is indicated by an affix virtually the same as the genitive , but 
 formally appearing as the preposition in {Fava, φυν^ cf. -ψϋ in άμ-ψί, 
 or %i of the gen. -^εν), which is the regular expression of definite lo- 
 cality. The breaking up of the genitive into two distinct cases — the 
 ablative and genitive — in Latin and Sanscrit, and the resolution of 
 the dative in the latter language into a dative, instrumental, and lo- 
 cative, or, if we please, the converse process of absorption in Greek, 
 must be explained in the same way as the variety of prepositions, 
 many of them identical in origin , to express the different modifica- 
 tions of direction and position in the case of those languages, which 
 make a sparing use of inflexions, or have lost them altogether. 
 
 It will be observed that the plural masculine and feminine are 
 formed by adding to the singular the letter -g, and that the plural 
 neuter is merely a reduplication of the singular ν or r, the resulting 
 combination -vt being invariably softened in α (§ 114). As the accu- 
 sative plural, masculine and feminine, does not reduplicate the objective 
 -?i, but adds to it the nominative -s, it is a reasonable conclusion that 
 
CHAP. Π.] OF THE NOUN. 433 
 
 in the plural, as in the feminine distinguished - from the masculine 
 (§ 228), the s is collective (§ 152) rather than merely subjective. 
 And this prepositional value of the plural sign is illustrated by the 
 fact, which we may claim to have discovered, that the Hebrew 
 plural was similarly formed by means of the prepositions dy and πϊί 
 both signifying "together with" (see Maskil le Sopher^ Cambridge, 
 1848, p. 13). 
 
 The dual presents abridged forms of the plural, the nominative 
 and accusative being distinguished by a vague -ε, which is often 
 absorbed, and the genitive or ablative and dative being both expressed 
 by the same residuary ending -lv. The abridgment of a-ses into ae 
 in the nominative plural of the Latin α -declension shows how the 
 nominative and accusative dual have been merged in a single repre- 
 sentative. But it is impossible that the genitive and dative dual can 
 have sprung from any disintegration of those two cases in the plural 
 or singular. It is clear that -tv for -φυν or -φις is the plural form of 
 the locative in l or φ^, just as the Latin plural in -bis or -bus stands 
 by the side of a singular in -bi. But as the Greek genitive is strictly 
 and properly an ablative, and as the ablative and dative plural are 
 uniformly expressed by the same locative case-ending in Latin, we 
 need not be surprised to find a similar neglect of case-distinctions in 
 the mutilated dual of the Greek nouns. 
 
 236 We shall now discuss in detail all the cases of the Sanscrit 
 declension, comparing them with the corresponding Greek and Latin 
 cases, and pointing out what are the substitutions in those two lan^ 
 guages for those Sanscrit cases which they have not, 
 
 (1) Accusative. 
 
 The regular sign of this case is m in Sanscrit and Latin, and ν in 
 Greek. Ifc is well known that the laws of euphony, which prevail in 
 the Greek language, do not permit the appearance of any labial at 
 the end of a word. It would be of little use to seek for an explana- 
 tion of this rule; and in the present instance there is reason to 
 beUeve that the Latin and Sanscrit m are weaker forms of an original 
 dental more truly represented by the Greek v. The Sanscrit m of 
 the accusative is generally transformed by anusvdra into a nasal ή ; 
 it is probable that the Greek final ν occasionally had the same sound, 
 and that it then subsided into the broad a, which is so frequently its 
 representative (above, § 114). Some such view is also necessary to 
 explain the fact that the Latin final m is disregarded in prosody, and 
 the vowel preceding it elided, when the following word begins with a 
 vowel; so that this m is merely the nasal liquid in its ultimate state 
 
 F F 
 
434 THE CASE-ENDINGS [bOOK III. 
 
 of obscuration. Indeed Quintilian distinctly explains the ecthlipsis 
 as a kind of anusvdra. He says {Inst. Orator, ix. 4, § 39): Inde 
 BELLiGERARE, po' MERIDIEM: et ilia Censoru Catonis Diee Hanc, 
 ceque Μ litera in Ε mollita: quce in veteribus libris reperta mutare 
 imperiti solent: ei, dum librariorwn insectari volunt inscientiam, 
 suam confitentur. Atqui eadem ilia litera^ quoties ultima est et voca- 
 lem verbi sequentis ita contingit^ ut in earn transire possit, etiamsi scri- 
 bitur, tamen parum exprimitur : ut, Multum ille, et Quantum erat: 
 adeo ut pcene cujusdam novce literce sonum reddat. Neque enim ex- 
 imitur, sed obscuratur, et tantum aliqua inter duas vocales velut nota 
 est, ne ipsce coeant. As an accusative case-ending, therefore, we must 
 conclude that the Latin m and the Greek ν are traceable to a common 
 origin, which is more truly represented by the Greek affix. 
 
 237 In Latin, Greek, and Sanscrit, the nominative and accusa- 
 tive of neuter nouns have the same termination. There can be little 
 doubt that the true explanation of this phenomenon is that given by 
 the late Mr. Coleridge, especially in its connexion with the fact that 
 in Greek the neuter plural is generally followed by a singular verb. 
 "The neuter plural governing, as they call it, a singular verb, is one 
 of the many instances in Greek of the inward and metaphysic gram- 
 mar resisting successfully the tyranny of formal grammar. In truth, 
 there may be Multeity in things; but there can only be Plurality in 
 persons. Observe also that, in fact, a neuter noun in Greek has no 
 real nominative case, though it has a formal one, that is to say, the 
 same word with the accusative. The reason is — a thing has no sub- 
 jectivity or nominative case: it exists only as an object in the accu- 
 sative or oblique case" {Table Talk, Vol. ii. pp. 61, 2). It would 
 perhaps have been better to say at once that both these facts depend 
 upon the same principle, that there is, namely, no nominative case of 
 neuter nouns, either in the singular or in the plural. The reason of 
 this we shall see better when we come to explain the meaning of the 
 termination of the nominative. 
 
 238 We must be careful to distinguish between the accusative 
 in -V or -m and the locatives in -iv or -im {am), which have been men- 
 tioned above (§ 170), and this discrimination is the more necessary 
 because the accusative form is often used with a locative meaning. 
 In Greek the adverbs δίκην, ακην, &c. are clearly locatives in meaning, 
 though in form they are perfectly analogous to accusatives. We may 
 also compare the accusatives μίν, νίν, with the recognised datives φ^ν, 
 rtV, 6cpiv^ &c., and the Sanscrit instrumental in α ==: ana and ΐ7ΐα 
 (§ 245). Moreover, it is well known that in a multitude of instances 
 
CHAP. Π.] OF THE NOUN. 435 
 
 the meaning of tlie Greek accusative is entirely locative: e. g. when 
 it follows a neuter or passive verb. We find other examples of an 
 apparent interchange of case-endings in the use of -d as the sign of 
 the neuter accusative-nominative in id, istud, and the appearances of 
 a similar ending in οτ-τι-, &c. (Bopp, Vergl. Gramm. p. 183), whereas 
 the -d was the ancient termination of the ablative in Latin, and also, 
 as we shall show, in Greek. In all these cases, however, we must 
 notice that the meaning of the particular case is explained by the 
 syntactical idioms of the language ; but that the accusative form^ as 
 distinguished from the locative form, has only a single letter, whereas 
 the locative properly so called appends this letter, indicating ob- 
 jectivity, to the second pronominal element, expressing an approxi- 
 mate position. Thus in Greek, while the accusative is marked by 
 -v, or in neuter nouns by -r, the locative in its full form is ex- 
 pressed by -φι-ν or -%i-v. 
 
 239 With regard to the accusative plural we adopt without hesi- 
 tation Grimm's opinion, that it is merely the accusative singular with 
 the plural s superadded. The stems which end with a short vowel 
 in Sanscrit form the accusative plural in n, with a lengthening of 
 the final vowel of the stem: thus vrtkas, "a wolf," makes accus. plur. 
 vrikan. If we compare this word with the Gothic vulfans on the 
 one hand, and the Latin lupos on the other, we shall perceive that 
 the Gothic is the complete form, the Sanscrit and Latin having lost, 
 one the 5, the other the w, and both having supplied the loss by 
 strengthening the final vowel of the crude-form. If we now take the 
 Greek λνκο?;^, which bears the same relation to λύχονς that odovg does 
 to dens, &c., it will appear, we conceive, that the same holds in 
 Greek. We may add that τνΛτων for τνπτον{τ)ξ h analogous to 
 vrikan for vrikans. It will be observed, too, that -in those cases 
 where the accusative singular has lost its final m, n, and indeed in 
 some others, the plural s is merely subjoined to the weakened form of 
 the accusative singular; thus τν7ίτοντα{ν) makes τντίτοντα-ζ. 
 
 240 In neuter nouns the accusative and nominative plural, 
 which are, for the reasons above given, the same, always end in -a in 
 Zend and in the old European languages of the family; but in San- 
 scrit we find an i, which, according to Boop, is only a weakened form 
 of an original a {Vergl. Gramm. pp. 5 and 269); the final vowel of the 
 crude-form is lengthened, and "an euphonical 7i," as Bopp calls it, is 
 inserted between it and the case-ending i. Thus madhn (μεΟ'υ), 
 "honey," makes in the plural madhu-n-i instead οι μί^ν-α. 
 
 This appears to us a rather unscientific way of accounting for the 
 
 F F 2 
 
436 THE CASE-ENDINGS [bOOK III. 
 
 Sanscrit inflexion. A more accurate examination of the phenomena 
 will enable us to reconcile the different forms by reprociucing the 
 structure in which they all originated. 
 
 We have already shown generally that the broad a represents an 
 anusvdra or suppressed η (§ 114), and the final η in particular is 
 constantly so represented in nouns of the third declension, as. in 
 φλεβ-α, τίατερ-α, &c. Moreover, we have shown that even -ν-τ may 
 be represented by a solitary «, as in δε-κα for δ^έ-κεντ (§ 161), and 
 we shall see that the formative ^ο:τ=^^ει^τ becomes -^a, as in ύώ-μα,&ο. 
 There would be no objection then, a priori, to regard the plural -a as 
 a relic of ντ', and if the objective ν or τ of the singular had to be 
 formed into a plural analogous to that of the masculine nouns in -g, 
 which, we shall see, form their plural by a reduplication of the ending, 
 we should be led at once to the assumption that the result would be 
 the combination ντ, or the reduplication vv. Now we have positive 
 authority for the assertion that the neuter plural in Latin originally 
 ended in -ad; thus we find in the Senatus Consultum de Bacchanali- 
 bus, 1. 24 : quei advorsum ead fecisent. Again, we find in Sanscrit, 
 as we have seen above, an interpolated η in the terminations of neuter 
 plurals, and the ^, which follows it, is most probably the vocalization 
 of a second n, just as conversely nn is substituted for ni (§ 215, b). 
 Putting all these considerations together, we can hardly avoid coming 
 to the conclusion that the proper and original plural of i-d was e-ad=^ 
 e-nd; that the genuine plural of madhu =r. madhu-n was madhu -ήή; 
 and that ΐΕ^νλ-α from ^,νλο-ν represents a primitive ξνλ-εντ. Our view 
 is still farther confirmed by the fact, that while the Erse plural of the 
 third personal pronoun is siad (for swiad), the Welsh form of the 
 plural is hwynt (for swynt) *. 
 
 241 We have before stated that the dual is merely a by-form of 
 the plural. The nominative and accusative dual in Sanscrit are, as 
 in Greek, the same. In some neuter nouns the dual nom., accus. are 
 the same as the plural ; in others there is an omission of the charac- 
 teristic n. Thus ddna, "a gift," makes in the dual ddne = dana-i, 
 the plural being ddnd-n-i; vachas, "speech," makes in the dual 
 
 * That the anomalous iorms hcec and quoi represent the original ha-ee,qua-ce 
 has been proved elsewhere (FarronmnMS, p. 317), and it has been shown that 
 this also explains the quantity of antea, quapropter, posthac, &c. A writer, who 
 can neither discover the truth nor recognise it when discovered, obstinately 
 maintains that the long a in antea,&>c. results from an absorption of m, and that 
 the original forms were anteafn, &,c.,^^onthe analogy οι postquam,antequam,&c.^' 
 Every Latin scholar is aware that quam is not here a case after posi or ante, but 
 the particle of comparison, so that the full form is, in fact, posteaquam, antea- 
 quam, &c. 
 
CHAP. II.] OF THE NOUN. 437 
 
 vachas-i, the plural being vachdns-i, where we have dififerent compen- 
 sations for a final n. 
 
 242 (2) Nominative. 
 
 The projDer sign of the nominative case is s. In Sanscrit this 
 sibilant is usually softened by visarga into h. In Greek and San- 
 scrit it is often absorbed in an «-ending in feminine nouns; in Latin 
 and Zend it is frequently dropt altogether in this case. When we 
 recollect how constantly the final s is mute in modern French, we 
 shall not wonder at these appearances of a similar insignificance of 
 the same letter in the ancient languages. We find instances in old 
 Latin of s concluding a short syllable though the succeeding word 
 begins with a consonant, as in the senio confectus quiescit of Ennius, 
 which can only be explained by supposing a visarga of the nomina- 
 tive analogous to the anusvdra of which we have spoken above. 
 There are reasons for supposing that this was the case in Greek also. 
 Thus it is possible that the rule about the pause is not violated in 
 ^schylus (PerscB, 321), as Person thinks {Suppl. ad Prcef. Hec. p. 33); 
 we can easily imagine that Άρίόμαρδοζ Σάρδεΰί might be pronounced 
 '^ρίόμαρδο Σάρδεΰί. As to the objection that Ariomardus was a 
 governor of Thebes and not of Sardis, we might as well object to 
 ^schylus for saying in v. 301, that Arcteus was ΐίηγαίξ Νείλου γει- 
 τόνων Αίγντίτ ίου, because the sameArcteus,byan amusing conversion, 
 is called in v. 41 a governor of Lydia: άβροδοαίτων 0' ετίεταν Αυδών 
 όχλος — Tovg Μυνρ. Άρκτενς τ άγα%•ός — εξορμώΰίν. Quintilian, too, 
 seems to have thought that the chief reason for the visarga in old 
 Latin was to avoid a concourse of consonants similar to that in this 
 passage of ^schylus. He says {Inst. Orator, ix. 4, § 37): Ceterum 
 Gonsonantes quoque, earumque prcecipue, quce sunt asperiores, in com- 
 missura verborum rixantur, ut X ultima cum S proxima, quarum 
 tristior etiam^ si bince coUidantur, stridor est: ut Ars studiorum. 
 Qucefuit causa et Servio {ut dixi) subtrahendce S liter ce, quoties ultima 
 esset aliaque consonante susciperetur, quod reprehendit Lauranius, 
 Messala defendit. Nam neque Lucilium putant uti eadem ultima., 
 quum dicit, Serenus fuit, et dignus locoque, et Cicero in Oratore 
 (c. XL VIII.) plures antiquorum tradii sic locutos, (We have here 
 adopted the emendations of RoUin and Gesner ; the books have et S 
 ultima cum X proxima, which is nonsense.) If this view is well 
 founded, we shall not wonder that a final letter, of little force in pro- 
 nunciation, should in some cases be dropt in writing also. Such 
 we find to be the fact in the feminine nouns of the first declension, 
 where the termination sa is represented only by the length of the 
 
438 THE CASE-ENDINGS [bOOK III. 
 
 final vowel, and in many masculine and feminine nouns of the third 
 or imparasyllabic declension. 
 
 243 The explanation of this nominative sign is obvious and easy. 
 The simplest form of the third personal pronoun in the Indo-Ger- 
 manic languages is, we have seen, in Sanscrit sa-s, sd, tad; in Greek 
 (())o, {6)ri•, το; in Goth, sa, s6^ thata. It is observed by Bopp {Abh. 
 Ak.Berl. 1826, p. 66), that although there is such a great number of 
 crude-forms in the Greek language ending in -o, there is not one 
 which wants the nominative sign s except this pronoun. A similar 
 remark applies to the Gothic language. There must be some very 
 good reason for this exception. We notice that, in modern German, 
 when the adjective has an article prefixed, it is declined according to 
 a weaker form, in other words, it does not preserve the declension 
 which it exhibits when not preceded by an article; thus we have 
 gute-r Wein, gute-m Wein, "good wine," "to good wine;" but de-r 
 gute- Wein, de-m gute-n Wein, "the good wine," "to the good wine;" 
 the case-ending of the article not being repeated in the adjective. 
 Conversely, we find in Greek that the case-endings are preserved in 
 the noun or adjective, but not in the nominative masculine of the 
 article. Thus we have 6 (= (jo) αγα%ο-ξ αν&ρωτιο-ς, not dg άγα%'ο 
 αν^^ρωτίο. When, however, this pronoun assumes a distinct personality, 
 it takes to itself a nominative ending like another noun; thus, δ 
 αν^'ρωτΐοζ οζ ταχηα TtOLSi, "the man who makes these things" := '■'■the 
 man, he {%. e, this man) makes these things." We have before re- 
 marked that in the article it is only the nominative, masculine and 
 feminine, of the demonstrative pronoun which exhibits the form o, ij, 
 or, substituting the sibilant for the aspirate, <jo, 67] ; the other cases 
 present a form beginning with τ, as nominative-accusative neuter ro; 
 gen. rov, r^g, rot), &c.: whereas in the relative, as it is called, the 
 form beginning with an aspirate is used for all cases and genders. 
 Now this relative expresses distinct personahty, or a subject, in the 
 masculine and feminine genders (that there is no subject or nomina- 
 tive in neuter we have before seen); but when this pronoun is used 
 merely as a prepositive article, it of course expresses personality or 
 subjectivity only in the nominative case, for in that case only can the 
 name of any object be considered as a subject; this is the reason why 
 all cases and genders of the article δ, ή, except the nom. masc. and 
 fem., assume the objective form in r-. We have shown before that, 
 of the three fundamental pronouns, the first represents the here^ the 
 second that which is near to the here, the third the mere object; we 
 liave also seen how in the idea of nearness the second pronoun in- 
 
CHAP. Π.] OF THE NOUN. 439 
 
 eludes the relative and reflexive pronouns as opposed to the mere 
 demonstrative, and that it is also used to form a nominative case to 
 the third pronoun; we now see that this element under the form sa 
 represents subjectivity as opposed to objectivity^ and is for that reason 
 used as a sort of post-positive article to note the nominative case. 
 
 244 The plural nominative ending in Sanscrit is -as^ which we 
 might be content to explain with Bopp (Vergl, Gramm. p. 261), by- 
 considering it "as an extension of the singular nominative sign 5, so 
 that there lies in this extension of the case-suffix a symbolical indica- 
 tion of plurality." For this we generally find the ending -Βξ in Greek. 
 In the first and second declensions, however, the nominative plural is 
 'CCL or -01, It is nearly demonstrable that the final i in these in- 
 stances, as well as in Latin, has supplanted an original s, a remark 
 which, applies also to the Latin genitives of the first declension ; thus 
 for familice, familial, the common forms, we have also familias=^fami- 
 lia-is. "We have no hesitation in asserting that λνκο-ί-, χώρα-Λ, stand 
 for Avxo-£g, %ώ^α-ίξ^ and that the Latin fifth declension in s differs 
 from the first declension in a only in the termination of the crude- 
 form (Bopp, Vergl. Gramm. p. 263). That this -eg presumes an 
 original -G£g, appears, in part at least, from the following considera- 
 tions (see also Varronian. pp. 278 sq.). As the accusative plural is 
 formed from the accusative singular by adding 5, we may be justified 
 in supposing that the plural nominative is formed from the singular 
 by the same addition; and as a double s is not allowed at the end of 
 a word in these languages, we must conclude that the plural s was 
 joined to the singular by the intervention of a short e or «, so that in 
 all probability the original form of the plural nominative of Ιχ^ύ-ζ^ 
 for instance, would be Ιχ%ν6-Βς^ which is practically shortened into 
 Ιχ^ύξξ^ as έλέγεΰο is into έλέγΒΟ=^ελεγον^ &c. (comp. Bopp, Vergl. 
 Gramm. p. 220). 
 
 In our own and other modern languages we find η as well as s 
 the sign of the plural. In the verb-endings in Greek we find the 
 plural -μεν instead of -μεζ. We shall discuss this in the proper place. 
 
 245 (3) a. Implementive or Instrumental. 
 
 b. Dative. 
 
 c. Locative. 
 
 We shall consider all these three cases together, because we be- 
 lieve that their meanings spring from the same source, and because 
 these meanings are all represented by the same case in Greek, that, 
 namely, which we call the Dative. 
 
440 THE CASE-ENDlNGS. [boOK III. 
 
 In Sanscrit tlie instrumental singular is marked by an ending a 
 or ina, the latter ending being appropriated to nouns the crude-form 
 of which ends in a short vowel. The termination a is considered by 
 Bopp (Abh. Ak. Berl. 1826, p. 77; Vergl Gramm. p. 188) to be iden- 
 tical with the preposition a, which is elsewhere used as a prefix, with 
 the signification " on," and which must be a residuum of a-na. The pre- 
 positions in, ανά (cf. ίνα), have the same meaning, and involve the same 
 pronominal elements (above, § 170). Hence, the instrumental in -ana 
 or -ina and the locative or dative in -m (§ 238) are traceable to a 
 common origin. That the idea of an instrument or cause may be 
 included in that of position or containing, is obvious ; and that a pre- 
 position, the intention of which is originally (as indeed is the case 
 with all prepositions) to denote position, can be used as a word mark- 
 ing an instrument or cause, is clear from the functions of δίά and ντΐό 
 in Greek, and per and ob in Latin. The phrase εν χερΰί, cominus, 
 "in close fight," sufficiently points to the connexion between imme- 
 diate proximity and the active use of an implement. The sign of the 
 dative in Sanscrit is e=m or one of its longer fo^ms ai or ay a. The 
 termination of the locative singular is in most cases i, so that the 
 dative is only an extension of the locative. The masculine nouns in 
 i and w, and sometimes also the feminines, have an anomalous loca- 
 tive ending in du; in this case the final vowel of the crude-form is 
 •thrown off, or changed into y, as in ρ atyduj sakhydu, from pati, "a 
 master," sakhi, "a friend." Bopp supposes (Vergl, Gramm. p. 228) 
 that du stands for as, and he recognises a connexion between the 
 genitive and locative both in Sanscrit and Zend. We prefer con- 
 sidering the w as a relic of the ending bhi=^cp. Compare the Greek 
 locatives avtov^ ov, &c. for αντό-φί or ανζό-&ί^ o-O't, &c.; and in the 
 Latin humi, domi, militice, CorintJii, &c., the locative has accidentally 
 assumed the same form as the genitive. In the Sanscrit dual, the 
 genitive and locative are identical ; they both end in 6s. The instru- 
 mental and dative dual are identical with each other and with the 
 ablative. The termination is always -bhydm. In the plural, the 
 implementive ends in -bhis, the dative and ablative in -bhyas, and 
 the locative in -su or -shu. These confusions are more than paralleled 
 in Latin. 
 
 246 It now remains to inquire what is the relation, in form or 
 otherwise, between these cases and the Greek dative, in which their 
 meanings seem to be included. The characteristic of the Greek dative 
 singular is -t for -φι or -O"t; of the dative dual -tv, it being identical 
 with the genitive of that number; of the dative plural -lol or -ig. 
 Bopp has remarked (Abh.Ak,Berl.lS26j p. 78) that the Sanscrit pre- 
 
CHAP. II.] OF THE NOUN. 441 
 
 position a {==.and) and ahhi {=^anahh% § 172) are synonymous; that 
 the former constitutes the singular instrumental ending, the latter, 
 with an end-syllable -άτη, the dual -bliyam=^bhi-am, and, with the 
 end-letter -s or end-syllable -as^ the plural -hhis or -bhyas. The dative 
 and locative singular present abhi under the different forms of ai and 
 i, the former standing for abhi just as tais does for tdbhis, and the 
 latter for bht just as avtco-L does for αντόφί, avto-iv for αντό-ψίν, 
 and avTo-ig for αντό-ψίζ. The terminations of the Sanscrit implemen- 
 tive, dative, and locative, are therefore resolvable into synonyms ana- 
 logous to the Greek ανά^ άμ-φί,Άηά -φι. The hsLimti-bi^vo-bis present 
 us with the form -ςρί, and its plural variation; and the termination -(φ)ίΐ; 
 bears the same relation to -qptg, that -μεν, in the verb-ending, does to 
 -με£, the other form. We have before remarked, that the letter η is 
 the most striking mark of the locative, and that we find this letter 
 at the end of locatives of all numbers. The ν Ιφελτινβτνκόν^ then, 
 which we so often see at the end of Greek datives plural, is not 
 a merely arbitrary addition, but a real part of the word, dropt 
 according to laws of euphony in the newer language. The anusvdra^ 
 or nasal at the end of the plural locative in Pracrit, points to a simi- 
 lar final η in that language. For the Sanscrit su, shu we find shva, 
 hva, in Zend, from which Bopp not unreasonably concludes (Vergl. 
 Gramm. p. 288), that the original form of the Sanscrit plural-locative 
 ending may have been sva, and this leads us at once to the reflexive 
 pronoun sva, Greek 6φε. The relationship between ΰφι,ν, the locative 
 of this pronoun, and the locative ending -ςρί,, -φιν, is clear from what 
 we have said in a former chapter (§§ 110, 121). The termination of 
 the dative plural in Greek is generally -l-olv. We have a similar 
 form in Sanscrit. Thus vnka-s makes locative plural vrike-shu=vrika- 
 i-shu. At other times the crude-form is not thus altered. This 
 is the rule in Sanscrit feminine nouns. Thus jiTivd^ "a tongue," 
 makes jihvdsu. It also holds in certain Greek words, as in the femi- 
 nine ΆΟ'ήνηβί, Όλνμτΐίάΰί, &νρα6ί, &c., and even in masculines, as 
 ταμίαΰο (Bockli, Corpus Inscript. i. p. 80). As the penultimate vowel, 
 however, is invariably long, it is probable that the dative singular 
 is included or absorbed in the form to which the plural affix is 
 appended. Putting all these facts together, we may come with toler- 
 able safety to the following conclusions : 
 
 (1) That the locative or dative singular in Greek is formed of the 
 same elements as the prepositions εν or ava, namely, S^ava or ΰφα-να^ 
 which may appear as φί, ^t, or i. 
 
 (2) That the plural of this form is really a reduplication of the 
 singular, or at least includes its characteristic i; for the termination 
 
442 THE CASE-ENDINGS [bOOK III. 
 
 '6lv is shown by the Zend forms to have been ΰφυν. The objective v, 
 however, was not repeated, but appeared only at the end of the com- 
 plex plural form. It is to be observed that the termination -ψί is 
 used by the epic and sometimes by the lyric poets to denote a geni- 
 tive, dative, and even an accusative, with or without a preposition. 
 Thus κεφαληφίν is a genitive in Hom. J/.XI.350; we have ajtb νενρη- 
 φιν in //.viii.300; danQVOcpi is a genitive plural in //.xvii.696; and 
 we have εκ Ο'εοφί- for εκ θεών, lb. 101, xxm. 347 ; as a dative singular 
 we have αμ ηοΐ φαίνομεντ^φίν in 11. ix. 618; as a dative plural with 
 a preposition we have 6vv οχεϋφι in //. xvi. 811; and as an ac- 
 cusative we have ετά δεξίόψίν for ετά δεξία in II. xiii. 307. Perhaps 
 as in the adverbs νόύφί, λίκρίψί£,&ο. the general expression of locality 
 had in these instances superseded the proper distinction of case, and 
 it is remarkable that this confusion is particularly apparent in the 
 syntactical use of this case-ending. Otherwise we could hardly have 
 had the identification of genitive and dative in the Greek dual, of 
 the dative and ablative in the Sanscrit plural, and of the dative and 
 ablative in the Latin noun of both numbers. 
 
 247 (4) a. Ablative. 
 
 The plural ablative in Sanscrit has the same termination as the 
 dative; the dual ablative is identical with the instrumental as well as 
 with the dative. The connexion in meaning between the dative and 
 instrumental cases we have just shown. The Sanscrit ablative pro- 
 perly expresses removal from a place, i. e. it answers to the question 
 "whence." It will easily be perceived how this might be resolved 
 into the idea of a cause or instrument, and also how the same mean- 
 ing might be made applicable to the ordinary use of a dative; for 
 instance, "I give to him" might be represented by "I give through 
 him," or "he is the cause as well as the object of my giving," for in 
 such cases the object to whom is very often the cause by which (Gr. Gr. 
 Art. 457). In Greek, the use of the dative to signify the occasion or 
 instrument is but little different from that of the adverb of manner ; 
 and this adverb, as we shall see presently, is a residuum of the 
 ablative, which is similarly used in Latin. 
 
 The characteristic of the ablative singular is -t, when the crude- 
 form of the noun ends in a; in other declensions it more nearly 
 resembles the genitive, to which also it corresponds in meaning. The 
 English writers on Sanscrit grammar consider -at as the termination, 
 but Bopp rightly concludes ( Vergl. Gramm. 209), from the analogy of 
 mat, tvat^ the ablatives of the first and second personal pronouns, and 
 of the Zend ablatives, that the ablative-ending is merely the letter -t. 
 
 We find this termination in the Latin met = Sanscrit matj which 
 
CHAP. Π.] OF THE NOUN. 443 
 
 appears in the combinations egomet^ memet, &c., and in the conjunc- 
 tion se-d^ more anciently written se-t. Under the form -d, this end- 
 ing appears as the regular characteristic of the ablative in old Latin. 
 Thus, on the Columna Rostrata we have : ^jrcesented sumod dictatored 
 olorom in altod marid pucgnad vicet (Varronian. p. 229). 
 
 It is to be remembered, that in Latin the same letter appears as 
 the characteristic of the neuter- objective singular as in id, illud, &c.; 
 also, as has been mentioned above (§ 239), in the neuter plural; thus, 
 in the Senatus Consultum de Bacclianalibus, we have quei advorsum 
 ead fecisent ; and as we, te (anciently met^ tet or med, ted) are used 
 both as accusatives and ablatives, so sed, which appears as an ablative 
 in its conjunctive use, is an accusative in the senatus consultum just 
 mentioned, where it appears after inter. 
 
 The only instance in Greek of a near approximation in form to 
 the Sanscrit ablative in -t, Latin -d^ is to be sought in the adverbs in 
 -ωξ as they are called. We find the ablative -d at the end of adverbs 
 like benedj facillumed^ which are clearly the old ablatives bon(^^)d, 
 faciUum{^)d. In the same way οντως^ καλώς, &c. are the old ablatives 
 of ovTogi καλός, &c.; for in the Greek language euphony will not 
 allow the appearance of a ά or τ at the end of a word, and όμως bears 
 the same relation to its Sanscrit synonym samdt that δίδωΰι does to 
 daddti (Bopp, Vergl. Gramm. p. 205). 
 
 We have before seen this termination of the ablative under the 
 Latin form in the adjective ϊδ-ίος, and we shall meet it again in the 
 iovms μοίρί-δι,ος, &c. and in patronymics Ιι[ί6Άτρείδης='^τρΒί-δως, 
 &c. A comparison oi 'Αφροδίτη with the Sanscrit abhrdditd, "she 
 who comes out of a cloud " (from the ablative abhrdt, written abhrdd 
 in composition, and ltd), shows that the first part of the word is an 
 adjective equivalent to the ablativa of αφρός, which should be αφρωδ 
 or αφρως ; the included form άφροδΰ- from άφροδυος is indicated by 
 the succeeding long vowel, which would otherwise have been short: 
 comp. ϊτης, ιταμός, &c. (Bopp, Vergl. Gramm. p. 216, note). We shall 
 see directly that this fuller form of the ablative is neither more nor 
 less than the genitive. 
 
 248 (4) b. Genitive. 
 
 The endings of the genitive singular in Sanscrit are s, sya, as, and 
 as. The latter is appropriated to feminine nouns which end with a 
 vowel; thus priti-s, "love," makes prityds.. When the final vowel of 
 the feminine stem is long, this termination is invariably preceded by 
 yovv, ViBJihvdyas, vadJivas; when the final vowel is short it is optional 
 whether we write a guna of the vowel and the simple 5-ending, or 
 the full ending in yds. Thus, we may have from priti-s not only 
 
444 THE CASE-ENDINGS. [bOOK III. 
 
 prityas, but also prites=prit-a-i-s. The termination ?/-a5 is represented 
 under a weakened form by the Latin j-us in cujus for quo-i-us, &c. 
 The terminations s or as are used according as the stem ends with a 
 vowel or a consonant. In the former case the termination may still 
 be considered as -as^ for the vowels i and u are always affected by 
 guna in the genitive: thus pati-s, "a master," mokes pates (=pat- 
 a-i-s\ and sunu-s^ "a son," makes sunos (=sun-a-u-s). In all these 
 instances the genitive corresponds to the ablative. When, however, 
 the ablative ending is -at, the genitive ends in -sya: thus vnka-s 
 makes ablative vrikdt, genitive vrikasya. We observe that this geni- 
 tive ending, which is peculiar to the Sanscrit declension in a, does 
 not affect the final syllable of the stem with guna: it is also probable 
 that the guna of the final syllable in the ablative of this declension, 
 and the genitive and ablative of other declensions, is occasioned by 
 the loss of some final vowel or vowels, as in ΆφροδΙ -ϊτη mentioned 
 above. Since, therefore, there is no reason why the genitive and 
 ablative should be different in the first, and identical in the other 
 Sanscrit declensions, we should be induced, a priori, to conclude that 
 the genitive and ablative of the first declension are to be considered 
 as merely by-forms of one another. And as the Greek genitive, 
 either with or without a preposition, expresses all the significations of 
 the Sanscrit genitive and ablative, it is reasonable to expect that the 
 forms must ultimately coincide. An examination of these forms, 
 compared with those of the Sanscrit ablative and genitive, will enable 
 us to prove, in opposition to Bopp (Abh. Ak, Berl. 1826, p. 97), that 
 these cases spring from a common origin. 
 
 In classical Greek there are three forms of the genitive which 
 are more prominently different, that of masculine nouns of the first 
 declension, of which the Homeric form is -cl-o {BoQBa-o,u4lvBicc-o, &c.)*; 
 that of nouns of the second declension, of which the Homeric form 
 is -OiO (XvK-OLO, &c.) ; and that of l and ν nouns of the third declen- 
 sion in -εως (τίόλ-εως,βαΰίλ-έως,&ο.)•]'. That the first of these originally 
 included a digamma is proved by the Corcyrsean inscription (Trans, 
 of Phil. Soc. I. p. 150, above, p. 218), in which we have the genitive 
 TXciuiafo, and we find the same form with an s appended in the 
 
 * In these forms the α belongs to the crude-form of the declension, and 
 there is no evidence whatever for Bopp's assertion ( Vergl. Gramm. p. 220) that 
 a-o stands for a-io=^a-aLO. 
 
 f That the shorter genitive in -og, which is found with most nouns of the 
 third declension, is a mutilated form, may be inferred from the genitive and 
 dative of monosyllabic nouns, e. g. ηοδόξ, τΐοδί, ποδών, τιοΰί, as distinguished 
 from πόδα and πόδες. The Sanscrit padas and padas are distinguished only 
 by the accent (see above, § 222). 
 
CHAP. Π.] OF THE NOUN. 445 
 
 inscription on the tomb of Midas (above, p. 155). The appearance 
 of Αα^οκόίων on an old inscription known to Priscian (i. pp. 23, 
 265 Krehl), a comparison of A«og, ναόξ^ &c. with λεω^, vmg^ &c., and 
 the obvious fact that in Homer εωξ must have been aS^og (Varron. p. 
 288), would lead us to conclude that ΛΟλεω5,/3α<9^λεω5,&ο. are similarly- 
 related to an original TtoXefog, βαΰίλά^ος^ &c., and we actually find 
 the genitives πόλψς, βαΰίληοζ- Then, again, the derivative adjective 
 δημό-ΰίΟξ would induce us to conclude that the genitive δημο-ιο was 
 originally δημό-ΰίο (Bopp, Vergl. Gramm. p. 294, note), the ΰ being 
 dropt, as in εδίδον compared with Mtdo(90, &c. (above, § 114) ; and thus 
 roto, λναοίΟ would ultimately correspond in form as well as meaning 
 with the Sanscrit tasya, vrikasya. We have therefore only two forms 
 for the Greek genitive, one in -/og or -Po and the other in -(jto, and 
 these unite in the form 6(po-, Sanscrit sva. That 6l may represent 
 / appears as an undoubted fact, after the analysis to which the pro- 
 nominal elements have been submitted. And if we compare δημόβίος 
 with I'dtog, a possessive of the same kind, and remember that ϊδίος 
 bears the same relation to Ϊ6ος=ϊΰΡθζ that medius does to μίύ6οξ=^ 
 μίβυοξ^ we shall see that the ablative vrikat=^vrikatya=:vrikasya,Q,na 
 that prttyat=prttyasj so that the genitive and ablative are identical 
 in Sanscrit as they are in Greek. 
 
 249 The genitive-ablative relation is also expressed by -0•εν in 
 Greek. The connexion of this suffix with the Latin -tus, Sanscrit 
 tas, thas, das, dhas, we have already pointed out; and it will easily 
 be seen what relationship subsists between humanit-us and the proper 
 adverb humaned (comp. bened, &c. in the Columna Bostrata), or be- 
 tween ονρανό%'(ΐ'ς) and the genitive ονρανό(6)ίθ. In point of fact, the 
 terminations -d'sv and -ΰιο are only different modifications of one and 
 the same form. We often find that %•ε-=ΰί-, e.g. in d'e6g=(jLOg. And 
 while the vowel, which followed the palatal of which &ε or 0i is the 
 representative, has been lost in -Ό'εν, ΰιο has parted with its final 
 nasal, which is so often seen to be removable. This observation points 
 the way to an interesting and instructive analogy. We cannot doubt 
 that adjectives in -Log=-6Log are formed from the genitive -ίο=-ΰίθ, 
 and we have seen (above, § 165), that these adjectives correspond in 
 signification to the quasi-comparatives in -Lmv=-LOV-g. If, as there is 
 every reason to believe, these latter are older, stronger, and more 
 complete forms of the adjectives in -to = -ΰιο, it is a just inference 
 that -ίον=-6ι,ον=^ο-ν was the fullest and most original form of the 
 genitive case. 
 
 250 Tins view is confirmed by the Greek genitive plural, which 
 is otherwise an inexplicable phenomenon. According to all reasonable 
 
446 THE CASE-ENDINGS [bOOK III. 
 
 expectations, the genitive plural ought to be derivable from the 
 genitive or ablative singular by the addition of -g. In common 
 Greek the genitive plural consistently ends in -ων, which is imme- 
 diately attached to the crude or uninflected form ; the circumflex how- 
 ever in the first declension points to an original form in -ά-ων, which 
 is still found. But iVAtQud-a-o =Άτρείδ-α-ΰίον, still more easily do 
 we pass irom'ATQELd-a-mv='j4tQeid-a-ov-g to^ArQud-a-uiov-g. This 
 result receives a further support from the analogy of the Latin and 
 Sanscrit, Although in common Sanscrit nouns the genitive plural 
 ends in -nam or -am only, the pronouns, which generally preserve the 
 authentic forms longer than the nouns, give us -sam or sham, as in 
 te-sMm = horum, td-sam = Jiarum. Now this fully accords with the 
 Latin; for -rum, which is the proper and genuine termination of the 
 genitive plural, not only in the vowel declensions, but also in the 
 others (Varro, L. L. viii. § 74 ; Cn. Gell. ap. Charts, i. 40), must have 
 been originally -sum, and this points to an older -som corresponding 
 to the Sanscrit -sam (Muller ad Varron. L. L. u. s. p. 192). It is not 
 easy to arrive at any certain conclusion respecting the η which 
 appears in many Sanscrit genitives; but when we find the more 
 easily explicable form in the pronouns, it is reasonable to conclude 
 that this nasal or nasal guttural is, like the Latin r, a corrupt repre- 
 sentative of the original sy-, Persian hy-. The passage from the 
 palatal y or j to a sound like the Hebrew Ayin is very conceivable. 
 At all events, any thing is better than, with Bopp, to import a 
 foreign difficulty into the Greek language, by supposing that λνκων 
 is a representative of λνκό-ν-ων, the ν being dropt as in μείξω from 
 μείζονα (Vergl. Gramm. p, 286, note). 
 
 251 (5) Vocative. 
 
 The vocative is either the crude-form of the noun, the vowel 
 being generally represented by the lightest sound ε, e, or the same 
 as the nominative. Thus in the first declension, ταμία-ξ makes 
 ταμία, κρίτη-ς, κρίτά in the vocative. In the second declension, λόγο-ζ 
 makes λόγε. In the third declension the vocative Σώκρατεζ implies 
 the included κράτοζ of the true crude-form, and its relation to the 
 nom. ΣωκράτηςΊ8ί\ι& same as that of the neuters εϋηΟ'ες, ενδαίμον, &c. 
 to their masc. and fem. ενη%•ης, ευδαίμων, &c. The nominative case 
 corresponds to the third person of the verb; it presumes a subject 
 spoken of. The vocative corresponds to the second person; it pre- 
 sumes a subject spoken to. We believe that the nominative was 
 secondary to the vocative; the idea of a subject in its objectivity 
 being suggested by conversation with another subjective reasoning 
 being; and for this cause the sign of the nominative case is identical 
 
CHAP. Π.] OF THE NOUN. 447 
 
 with the element of the second personal pronoun. In the second 
 person of the imperative mood, where the person spoken to is most 
 directly and impressively addressed, we find sometimes the crude-form 
 of the verb as τντίΤΒ^ at other times the element of the second per- 
 sonal pronoun more strongly expressed, as in %λν-^ι\ for the other 
 persons of this mood a stronger form is invariably adopted. The 
 vocative λνκε stands related to the imperative τύτίτε just as the nomi- 
 native λνκο -g does to the indicative τντίτεις for τντΐτε-ΰί. 
 
 'The accentuation of the vocative in the third declension, when 
 the crude-form is terminated by a consonant, and the nominative pro- 
 perly had its last syllable long by position or absorption, presents 
 some phenomena, which are very instructive when properly examined 
 and analyzed. The technical writers on accentuation are contented 
 with saying (Gottling, Elements of Accentuation, § 26, p. 53, Engl.Tr.) 
 that "no vocative of the third declension is oxytone, if it be really 
 distinguished from the nominative by a peculiar form, because, being 
 the oldest form of the nominative, it follows the oldest iEolic accen- 
 tuation, which recognises no oxytone. Those only in ig and vg remain 
 oxytone." To the comparative philologer this rule is simply an eva- 
 sion of the difficulty, coupled with a misrepresentation of the facts. 
 That the vocative is not in any sense a form of the nominative, and 
 that the ^olic tendency to draw back the accent has nothing to do 
 with the difierence between the intonation of the vocative and that of 
 the nominative, may be inferred from the following circumstances. We 
 do not find that the accent is drawn back, even when the last syllable 
 becomes short, in those cases in which the crude-form was originally 
 terminated by two consonants. Thus a participle like παίδενων was 
 Ttaidsvov in the vocative, not παίδενον*, because the crude-form was 
 τΐαι,δενοντ, just as the third person plural of the imperfect was accen- 
 tuated ελέγον, εφάβαν, &c. (Ahrens, de dial. Dor. pp. 28 sqq.) on account 
 of the original termination in VT. Again, although barytone com- 
 pounds regularly draw back their accent in the vocative — as Άγάμεμ- 
 vov, ^ημό6^•εvεg from Αγαμέμνων^ ^ημo6^εv7]g — this does not apply 
 to compounds in -φρων and -τωρ, because the secondary crude-form does 
 not revert to the original form of the included word, but retains enough 
 of the modified syllable to affect the intonation ; hence we have βαρύ- 
 φρον, καρτερόφρον from βαρνφρων^ καρτερόφρων, not βάρνφρεν^ 
 καρτεροφρεν, that form being lost in the composition; and similarly, 
 τιροΊίάτορ^ αυτοκράτορ from Λροτΐάτωρ, αντοκράτωρ,ηοί τΰρότΐατερ, 
 
 * G. Curtius, who as usual follows us, adds that we have υπάρχον not 
 υπάρχον (Sprachvergleichung, p. 16). He does not seem to be aware that this is 
 owing to the parathetic composition. 
 
448 THE CASE-ENDINGS OF THE NOUN. [bOOK III. 
 
 αντόκρατερ, because the primitive form has yielded to the formative 
 process. Exceptional instances in the accentuation of the vocative 
 seem to belong to words, in which the accent was affected by a pecu- 
 liar urgency or familiarity of the invocation. Thus we have the 
 precative exclamations/'AjtolXov, ΠόβΒίδον, ΰώτερ, ηοιη.Άτΐόλλων^ 
 Ποΰείδών, όωτήρ; the naval cry to the steersman, κνβερνάτερ, nom. 
 κνβερνητήρ; the address of near relatives, as ττάτερ, ανερ, ^νγατερ, 
 δάερ^ εϊνατερ, nom. πατηρ^ άνήρ, Ο'νγάτηρ, δαήρ, εΐνάτηρ. With this 
 change of accent we may compare the exclamation, αλη^'εξ, "indeed!" 
 from αλΥΐ^Ύΐξ"^. 
 
 * This drawing back of the accent is observable in Sanscrit also, where it is 
 similarly explicable (Bopp, Vergl. Accentuationssystem, p. 20}. 
 
CHAPTER III. 
 
 THE PRONOMINAL TERMINATIONS OF THE 
 UNINFLECTED FORMS. 
 
 252 Derivation-suffixes of pronominal origin. Their classification and mean- 
 ing. 253 (1) Forms with the first element only. 254 (2) Forms with 
 the second element only. 255 (3) Forms with the third element only. 
 
 256 (1 a) Terminations compounded of the first and other elements. 
 
 257 (2a) \Terminations compounded of the second and other elements. 
 Nouns in -lig, -ώς. 258. Nouns in -βν-νη, -i-vog, &c. Άλη^Ίνός. 
 "259 Nouns in -ιτης, -ιητης, &c. 260 Collective nouns in -ων. 
 
 261 (3 a) Reduplications of the third element. 262 (2 b) Second element 
 as -8 or -^. Patronymics , comparatives, and participles. 263 Ad- 
 verbs in -δην^ &c. 264 Adverbs in -rt, &c. 265 Connexion of these 
 forms. 266 (3 5) Third element as λ- or ρ-. Identity of these forms. 
 267 Also of the forms τ-λος, τ-ρος- &c. 268 Reasons for believing 
 that the forms λ- and ρ- are derived from -va, both from their use 
 as pronominal syllables ; 269 and from their employment as verbal 
 roots: 270 especially in expressing the ideas of progressive time and 
 recurrence. 
 
 252 ΠΠΗΕ root of a noun or verb, which properly speaking 
 never exceeds a single syllable, may, as we have said 
 before, be considered as the independent variable ; the noun or 
 verb is a function of this variable, and contains, besides the root, 
 in the former instance, a set of case-endings, in the latter a set 
 of person-endings, both of pronominal origin. But between 
 these case- and person-endings we find, frequently in the verb, 
 and generally in the noun, a derivation-suffix. Having discussed 
 the case-endings of the noun, we proceed to the consideration of 
 these derivation-suffixes, which determine the class the noun be- 
 longs to, and declare the particular modification of meaning with 
 which it is used. Of the prefixes, which express the relation in 
 space that the noun is supposed to bear to the other words in 
 the sentence, we have spoken under the head of the prepositions. 
 We do not intend here to mention those compound words which 
 contain two or more distinct verbal or nominal roots. We shall 
 confine ourselves to those suffixes which are of pronominal origin, 
 and which may therefore be considered as the constants of the 
 function ; compound words are, as it were, functions of two or 
 more variables. 
 
 In these pronominal suffixes we observe all the peculiarities 
 which we have noted in the pronominal elements separately 
 
 GG 
 
450 THE PEONOMINAL TEBMINATIONS [bOOK III. 
 
 considered; there is not one monosyllabic or primary pronoun 
 which does not appear as a termination, and scarcely one combina- 
 tion of monosyllabic pronominal elements which is employed in- 
 dependently but is applied to this purpose. But though we can 
 most clearly trace the pronominal elements in these formations, 
 it is scarcely possible to assign to them a definite meaning on all 
 occasions, any more than we can translate literally all the long 
 compound pronouns. That they had a distinct signification at 
 one time cannot be doubted, and it is at least useful, for the 
 purpose of classification, to arrange the terminations according 
 to the significant pronouns to which they externally correspond, 
 even though we are not always able to assign to them, in their 
 use as suffixes, a meaning perfectly and perceivably identical 
 with that which they bear when used independently. 
 
 The three primary monosyllabic pronouns are, as we have 
 seen, (1) 'pa (ma); (2) qva or J^a; and (3) ta: the first expresses 
 the idea of "here ;" the other two are modifications of the idea 
 of "there," distinguished according to the relative nearness or 
 farness of the object. These primary words are, we have seen, 
 compounded with each other, and otherwise modified, in a great 
 variety of ways. The second admits of the greatest number of 
 modifications; its original and proper element J^« or ^'va becomes 
 sva, va^ sa, hva, ha, ha^ pa, ga^ and ya; and the sibilant passes 
 into the dental articulations dva, da, tva, tu, &c. The thiiid is 
 changed only into na. Any one of these stems may be strength- 
 ened or rendered more emphatic by the addition of the affix -ra 
 or -la derived from na. The Sanscrit vowels and diphthongs 
 a, a; i, ^; e; which are all used as pronouns in composition, 
 are only initials. 
 
 In the formation of nominal derivatives in Greek, we ob- 
 serve generally that the first pronominal element expresses that 
 the thing proceeds from, or immediately belongs to, the subject; 
 the second, that it has a relation to the subject; the third, that it 
 is a mere object, or something removed from theproximity of the 
 subject. 
 
 253 (1) Forms with the first pronominal element only. 
 
 The termination -μο-ξ masculine, -μη feminine, is generally found 
 with a class of nouns which represent the action of the verb proceed- 
 
CHAP. III.] OF THE UNINFLECTED FOBMS. 451 
 
 ing from the subject, and may be expressed by the infinitive active 
 used as a noun : thus (5ει6-μό-(5, "the shaking," = το ϋείείν, ποτ-μός, 
 "the falling,"=ro τΰίτίτειν (root πετ-) ; or the thing in which the action 
 of the verb is exemplified ; as δη-μος, δψμός, "that which binds," i. e. 
 either an inclosed field or village (like the English "toune" from 
 "twine," which signifies both a field and a town), or the fat in which 
 the thighs were wrapt up for sacrifice; ερετ-μός, re-mus, "that which 
 rows," &c. The same meaning may be observed in μνή-μη^ "a recol- 
 lecting," εΛίΰτή-μη, "an understanding," &c. We cannot agree with 
 Buttmann (Ausfuhrl. Sprl. § cxix. 22) in thinking that the ending 
 -μη is ever really equivalent to -ματ=μεντ] but cases may occur, in 
 particular nouns, in which the termination -μη or even -μog, repre- 
 senting, as these endings do, the mere action of a verb, expressed by 
 its infinitive, may seem tantamount to the complex and secondary 
 affix μα=μεντ, expressing the act or result of the action. Such for 
 example is the true explanation of the phrase, Phil. ii. 6 : ονχ άρτίαγ- 
 μον ηγηΰατο το είναι Υΰα &εω, "he did not think that the being equal 
 with God was an object of eager desire," where some would expect 
 αρτίαγμα (Heliodorus, ^eth. vii. 20). For we have precisely the same 
 idea in the phrase, Pind. Pyth. x. 62 : τυχών κεν αρτίαλεαν ΰχέϋΌί φρον- 
 τίδα ταν παρ τΐοδός, "he would make much of, be contented with the 
 thought of the moment;" and we have άρτίαλέα δόβις in Pind. Pyth. 
 vni. 63, meaning "a gift to be eagerly desired : " and though the ter- 
 mination -6Lg is quite as abstract as -μος, this word docjtg, which has 
 become fixed in our English "dose," is used by the oldest and best 
 writers to signify "a giving" with reference to the recipient rather 
 than to the giver; thus we have of the alms given to a beggar (Horn. 
 Orf.vi.208): δόόις ολίγη τε φίλη τε. In the same way nouns in -μog 
 are used to express an act quite as much with reference to its accom- 
 plishment as with reference to its origin or continuance, for χρηό-μός, 
 "an oracle," is something actually delivered, and ^εβ-μός, "a law," is 
 not so much the act of law-giving as its consequence. We find the 
 common ground of these meanings in the verbals in -τέος, -Tog, which 
 though originally expressions for the active infinitive are used in a 
 manner which leads to their being confused with the passive. Thus, 
 to return to άρτίαγμός, we have just the same idea in άρτΐακτός, as 
 used by Hesiod, ε. κ. η. 320: χρήματα δ' ονχ άρπακτά' ϋ^εόΰδοτα πολ- 
 λον αμείνω-, "money is not for snatching, i. e. to be eagerly snatched." 
 While therefore we think that we may always reproduce the abstract 
 or active meaning of nouns in -μοξ and -μη^ we admit that there are 
 cases in which the latter especially are practically expressive of the 
 thing done rather than the act of doing. Thus though τι-μη means 
 really "an honouring" or "paying of honour," it is practically used 
 
 GG2 
 
452 THE PRONOMINAL TERMINATIONS [bOOK III. 
 
 to denote "the price" paid, much like τίμψμα, which is formally dis- 
 tinguished from it ; and though γράα-μη means "a writing or draw- 
 ing," as distinguished from γράμ-μα, "a thing written or drawn," it is 
 practically used to signify "a line" drawn (cf. the two meanings of 
 γραφή)' So also δεϋμη means "a bundle" rather than the act of 
 binding, and δεΰμός means a thing for fastening, like δεβμω-μα. The 
 termination -μος is generally preceded by 6, as in ^εβ-μός from τί^ημι^ 
 δεβ-μός, δεΰ-μή from δεω (compare however δη-μός\ or by its repre- 
 sentativeO', as in άριΟ^-μός, άρ&ψός, ρν^'-μός (compare however ρν-μός). 
 A κ is occasionally aspirated, as in πλοχμός from ττλεκω^ ίωχμός from 
 Ιώκω, or even inserted instead of 6 or 0", as in αυχμός from ανω. The 
 force of this ending is well shown by a comparison between the Latin 
 al-mus^ "the nourisher," and alu-m-nus^ "the nourished." 
 
 254 (2) Forms with the second pronominal element only. 
 
 Nouns in -Gig, -tvg, both feminine, express a relation to the sub- 
 ject, and therefore approximate in meaning to those in -μΟ£'• thus we 
 may compare Λρακ-ΰίς^ "a doing," μίμψΰίς, "an imitating," φά-ΰις, "a 
 showing," &c., with εδη-τνς^ "an eating," ετίψτύς, "a showing com- 
 passion," ορχηΰ-τνς^ "a dancing," Λρακ-τύζ, "a doing," &c. These two 
 endings are related as 6v and tv : between them stood the old form in 
 -rtff, as φά-ης, "a speaking," μψτίς^ "a thinking," 7tv6-TLg, "an in- 
 quiring," &c. Compare the Latin pes-tis^ "a destroying," ves-tis, "a 
 covering," &c.; po-tus, "a drinking," spiri-tus, "a breathing." When 
 we compare por-tus with Λορ^'-μός^ ορ%ψ(5ίξ and ορχηΰ-τνς with 
 6ρχη6-μΟ£, we see that, as the meanings of the first and second ele- 
 ment often correspond in separate words (§§ 135, 150, 154), so there 
 is scarcely any difference in signification between these endings. To 
 this class belong the Latin supines in -turn and ~tu, and the Greek 
 verbals in -τέος^ -tog. By the side of the words in -rvg, expressing an 
 action, we have a set of masculine nouns in -τηg denoting the corre- 
 sponding agents: thus with h7t7]-tvg, 6ρχfj6-τvg, we have επη-τής, 
 ορχΎΐύ-ΤΎΐς., &c. We shall show directly that these endings are con- 
 nected in origin as well as signification with the verbals of which 
 we are speaking. 
 
 The termination -ΰog=^(5φog^ implying a collection, is a represen- 
 tative of the full form of the root Fa. We have spoken above of its 
 use in forming pronominal adjectives like o-6og.) 'i-uog, &c. (§ 152). It 
 also appears in substantives, as in d^la-oog, "a collection of sacred per- 
 sons," %'vρ-0og, "a collection of leaves," πvρ-ΰ6g, "a gathering of fire.'' 
 
 Among the modifications of the second pronominal element, the 
 form -κό -s is very common as an adjective ending. The vowel before 
 the termination is most frequently l, especially when the adjective is 
 
CHAP. ΠΙ.] OF THE UNINFLECTED FOEMS. 453 
 
 derived from a verbal noun of agency in -της, as in τίΟίψν-κός from 
 Λοοητης^ &εατί-κός from Ο^εατης, &c. ; and there is little doubt that this 
 is an articulation indicating a pathology similar to that of the redu- 
 plications τί^ημι^ δίδωμί, and that it most generally represents an ψ 
 If the word from which the adjective was derived ended in -Log or -ta, 
 the termination -xog is preceded by -cc, as in ηλία-κος from ήλιος, 
 παροιμια-κός from Λαροϋμία. If the noun ends in v, the termination is 
 immediately appended to this semi-consonant, as in ά6τυ-κ6ς,%''ηλν-κ6ς, 
 λφν-κός. The isolated form άλν-κός from αλς seems to have owed its 
 V to the influence of the λ*. The terminations -α-κ-ς, -ε-κ -g, α-κ -g, 
 -ι-κ-ς-, -ν-κ -g•, and the Latin -i-c-s, are only modifications of this ending, 
 as the adverb-ending -κ -g is of -κυς, comp. άλώτΐηξ with the Sanscrit 
 lopaka, κάλνξ with the Sanscrit kalaga ("a water-jar") ; and the forms 
 μοννάξ, ενράξ, πολλάκις, τετράκις, &c.f From these we must distin- 
 guish those words in which the guttural belongs to the root and the 
 -ς constitutes the ending, as in ετίίμίξ (root μιγ-), οδάξ (root δακ-). A 
 few nouns have an euphonic nasal before the guttural ; such are λνγξ, 
 ΰαλτΐίγζ, τίλάότιγξ. This termination κό-ς expresses a quality in an- 
 swer to the question "what?" — as in Λουψι-κός, "like a poet," άνδρι- 
 -κός, "hke a man," &c.: just so bala-ka=puer qualisf in Sanscrit 
 (Pott, Etym. Forsch. ii. p. 458). Still more common is the termina- 
 tion in -ίο-ς answering to the Sanscrit ya-s : in fact there is hardly a 
 root or termination to which it may not be joined with a qualitative 
 meaning. Under the form -ίας, -ίου, it is of extensive use in the con- 
 struction of qualitative nouns, such as νεανίας, κουρίας, &c., and of 
 proper names expressing a quality, as Καλλίας, Αο^ίας,^ ΙτίΛίας, Νικίας, 
 &c. We recognise this suffix also in the numerous class of abstract 
 nouns ending in -η^ and in those, denoting agents, in -ης: thus αρχή, 
 τριηράρχης, stand for άρχyά, tριηράρχyaς, just as τριηραρχεω repre- 
 sents τριηράρχyaμι. It is not, however, subjoined to the ending -κός, 
 to which it appears to be perfectly equivalent (compare ανδρείος with 
 
 * A young philologer, Dr. Jos. Budenz, has written an elaborate essay 
 on this termination (das Suffix κός, (ixdg, axoj, νηός) im griechischen. Got- 
 tingen, 1858). His object is to show that the termination is not -y.og but 
 lyi-og, a theory which is at once overthrown by the accentuation, for no 
 merely vowel ending is oxytone. But although his analysis is altogether 
 faulty, he has collected a number of stray particulars, and he has not 
 failed to arrive at the conclusion, which we indicated 20 years ago, that 
 β is ultimately identical as an affix with κ, both being forms of the second 
 pronominal element. We may claim the same priority in regard to his 
 remark (pp. 53, 54) that "considered from the point of view occupied by 
 comparative philology, the Greek language is better suited than any other 
 of the cognate idioms to serve as the basis for many researches into the 
 formation of words." 
 
 t The comparative philologer will be amused by Lobeck's difficulties; 
 Paralip. p. 123. 
 
454 THE PBONOMINAL TEKMINATIONS [bOOK III. 
 
 avOQiKog), nor can we persuade ourselves that it is ever appended to 
 the simj^le element of the second person, though some scholars have 
 suggested that terminations like -Clou in Greek and -sya in Sanscrit 
 are compounds of sa and ya. We believe that the terminations -uig, 
 -ΰία, are perfectly equivalent in form as they certainly are in mean- 
 ing, and differ from the form in -Ca only by being more complete 
 (below, § 258). Our previous investigations have led us to the con- 
 clusion that there is an immediate connexion between the formative 
 elements 6i- and d-, and we have seen this identity, and the con- 
 nexion of both of them with p- in the analysis of the genitive (§ 247 
 — 249). As in the case of the relative pronoun, so also in these 
 affixes it is clear that the mere guttural -κ may appear as the repre- 
 sentative of the complex articulation (§ 146, cf. § 121); and there is 
 a very remarkable identity in the use of the formatives κ- and d-. 
 Thus while the Greek feminines are regularly formed in τρίδ- from 
 τηρ^ the Latin are as regularly formed in trie from tor; compare γενε- 
 τρίδ (-τρις) with gene-trio {-trix). Similarly it cannot be doubted that 
 τνραννί-κόςί^ an extension οίτνραννί-δ-{ς), that Ίλία-κός, Κορινθια- 
 κός, &G. are really derived from the collective nouns Ίλίάδ-{ς), Κορυν- 
 %ιάδ-{ς), &c., and that the names of diseases, such as αρθρΙτίδ-{ζ), νεφρΐ- 
 τίδ-(ο), &c. are the basis of the corresponding adjectives άρ^-ρίτι-κός, 
 νεφρίΤί-κός, &c. And as the nouns of agency in -της(-τηρ,-τωρ) have a 
 corresponding form in rt5-, we must recognise the same interchange of 
 δ and κ in the almost invariable adjectives in -τικός. A third class of 
 qualitative endings is connected with the modification -va. These 
 appear in a very mutilated form, and therefore the real termination is 
 not discernible at first sight, at least in the uncompounded state; 
 with the addition of the third pronominal element this suffix becomes 
 more obvious in Greek : without that addition it is not used in San- 
 scrit. In Greek we have γραφε-ν-ζ=γροίφ{'-'ϋ6-ς, ίΛΛε-ν-ς=ί7ί7α-κό-ς, 
 &c. With reference to what has been said of the identity of the 
 affixes δ- and κ-, it is worthy of observation, that while ζίωρίδ-{ς), 
 Φωκίδ-(β), stand by the side οί^ωρε -vg^ Φωκε -vg, we have secondary 
 forms like Ζίωρια-κόζ-, &c. The word βαΰυ-λεν-ς does not belong to 
 this class. It is compounded of λαός (λεΓος, λενς), and a root βαΰϊ-^ 
 of which we can only say that it seems to be the same as that of 
 βαγός, βά, a vocative (see Valcken. Adoniaz. p. 383): compare Οϊλευς 
 (=zvis ρορηΙί=ιΣΘ'ενε-λαος), '^ίγηύί-λαος, &c., and see Pott, Etym. 
 Forsch.u. p. 278. The feminine corresponding to this masculine ending 
 is -ω; thus77ρorε-υ-g,77ρωr-ω; Νηλε -vg, Νηλ-ώ',θ'εμίΰτε-νς,&εμίΰτ-ώ, 
 &c. The words εν-εβτ-ώ^ κακεότ-ώ, αει-εβτ-ώ and εύτ-ώ point to the con- 
 nexion between this class and the nouns in -c)g=for-g, fem. -via. 
 For it is clear that we ought to derive ενεΰτώ not from εν εύτί, which 
 
CHAP. III.] OF THE UNINFLECTED FOEMS. 455 
 
 is the derivation given by H. Stephens, after Hesychius, but from the 
 participial ev εύτώς. Thus Thucydides, v. 46 : (5φί(5ι μεν γαρ εν εύτώ- 
 των των τίραγμάτων αροΰτον εϊναι δοαβώΰαβ^αί την ευτίραγίαν. We 
 shall point out the connexion between the masculine and feminine of this 
 suffix when we come to the compound terminations. The nouns in -πόζ 
 seem to contain the second pronominal element in its Attic form {tcov., 
 πτόΟ'εν, &c.), as μαβτρο-τίόξ (seeEustath. 308, 2), χεδρο-τίόξ^ χαρο-πός, 
 and perhaps άΰτερο-τΐος (Schol. Eurip.313), and κάρδο-τίος- The soli- 
 tary form εχ^^ο-δο-τίός contains this ending, subjoined to the element 
 do-] of which hereafter. It is possible that this ending may also lurk in 
 the Homeric ητίε-ρο-ττε-ν-ς, ηΛε-ρο-τΐε-υ-τής, the root of which seems 
 to be contained in ατία-τή, Doderlein {Homer. Gloss, p. 243, No. 376) 
 connects ητίεροτίεύειν with άπρετΐής^ comparing Ο^εοπρόπος. 
 
 It might be supposed that the termination -τη-ζ^ gen. -του, which 
 signifies a male agent, as %ρί-τ7ΐξ> "a judge," &c., consists of the third 
 element only. There is reason, however, to believe that this ending 
 either presents the second element under the form rt, which is the 
 case in all nouns like χρυ-τής of the first declension, or contains a com- 
 pensation for a consonantal crude-form in the nom. of the few nouns 
 which end in -της, -τονς, just as -ης appears for -εό -ες in the nominative 
 of compounds like εντείχης from τείχος. Words formed with the end- 
 ing -της are sometimes passive; thus γενε-τής signifies both "father," 
 which is the more common meaning, and "son" (Soph. (EcZ. Ti/r. 470 ; 
 Eurip. Ion, 919). We have also άτί-της, "dishonoured" (^schyl. 
 Agam. 72; Eumen. 246), αειγενε-ται &εοί in Komer^aihv εόντες, and 
 Pindar calls Bacchus κυΰβοδέ-ταν &εόν (/r.45,9). But this meaning is 
 more generally found with words in -τός, which termination appears 
 not only in a large class of words with a passive signification, but also 
 in the ordinals and superlatives. In the two latter cases it is probable 
 that the termination is connected with -ϋ'εν the mark of the genitive 
 case, and, therefore, with the second pronominal element, and the 
 verbals in -τεος, -τνς, -τις, &c. 
 
 255 (3) Forms with the third pronominal element only. 
 
 There is a large class of neuter substantives properly terminating 
 in the element -τ-, which is however softened in various ways by the 
 process of declension (see Gr. Gr. Art. 182). One of the most common 
 of these alterations is the substitution of -ο-ς for -o-tr, and the omission 
 of this sibilant in the oblique cases: thus τίράγο-ς for τΐράγο-τ means 
 a thing done, the genitive being τΐράγεος for πράγε-ΰος (according to 
 § 114). The dental tenuis also appears alone in a number of par- 
 ticipal adjectives, such as γρατΐ-τός, "written," χρίύ-τός, "anointed," 
 δεί-νός, "dreaded," &c. Although the termination of the passive par- 
 
456 THE PEONOMINAL TERMINATIONS [bOOK III. 
 
 ticiple in Latin and Sanscrit is identical with that of the supine, as it 
 is called, in Latin, and the Sanscrit infinitive, and though it is clear 
 that these supines and infinitives are of the same origin with the 
 verbals in -rf Off, &c. it is perhaps an open question whether we ought 
 not to regard the termination of the passive participle as resulting 
 from the third pronominal element only, and therefore as difiOrent 
 from that of these verbals. One reason for inferring a difference be- 
 tween these endings and the verbals formed from the second stem is, 
 that the former perfectly coincide in meaning with the words formed 
 with the suffix -vog, which certainly has no connexion with the second 
 element, and contains only the third in its strongest form. Thus 
 both -Tog and -vog are used to form passive participles ; compare ple- 
 nus and re-jple-tus, bha-nu-s and bhd-tu-s (both signifying the sun in 
 Sanscrit), the pronouns e-na and e-ta in the same language, and (5ίμ- 
 vog, 687t-TOg, in Greek. On the other hand, the termination -vog may 
 be referred, like -τεο^, -Tog, to an active infinitive and there are ana- 
 logies to explain the passive or objective meaning of adjectival words 
 thus connected with transitive verb-forms (see below, § 415). The 
 words dsL-vog, ΰτνγ -vogj TtoQ'ei-vog, ελεί -vog, &c. are all expressive of 
 objects conceived under certain relations; and the same may be said 
 of the corresponding nouns in -vov, such as δργα-νον. In ΰπλάγ-χνον 
 from ΰτίλήν, we have probably the verbal root χα-ν-. If so, the 
 όπλάγχνα or viscera majora were regarded as an extension of the 
 (StcXyiv. It will be remembered too that η and t are inserted in the. 
 present tenses of verbs in precisely the same manner, thus we have 
 τν-7ί-τ-ευν and τεμ-ν-ειν in Greek, si-n-ere Άηά flec-t-ere in Latin (Pott, 
 Etym. Forsch, ii. p. 467). The qualitative nouns in -τητ -g {τΎΐ -g), 
 -Tfjtog^ Latin 'tat-s(ta-s), -tatis, appear to be formed either from abla- 
 tive cases oi the third pronominal root, or by combination of the 
 ending τη^=^τ^α with that of the third element. This termination -νη 
 is found with a similar meaning in rjdo-VTj, άγχο-νή: more clearly 
 developed when compounded with the first and second elements as in 
 χαρ-μο-νή^ δίκαοο-6ν-νη, and the adjectives XQi&a^L-vog^ πνρά-μο- 
 vog, γη^'ό-όv-vogJ κη^ό-ΰv-vog. The words κ -QOV-vog, κ-ρή-νη from 
 £κ-ρεω, deserve particular notice on account of the aphseresis of the 
 preposition. See above, § 176, and compare the Greek ϊχ -^og^ εχ-%- 
 Qog^ '^X^OL (=£|ω. Hesych.), &c. 
 
 256 (1 a) Terminations compounded of the first and other pro- 
 nominal elements. 
 
 Of these the most common are the compounds of the first and 
 third pronouns. They express the action as proceeding from the 
 subject, but with especial reference to its results, and generally appear 
 
CHAP. III.] OF THE UNINFLECTED FORMS. 457 
 
 as extensions of the ending -μο6 (above, § 253). The simplest form of 
 this extension is the noun of agency in -μην (-μεν£-) or -μών- (μόν-ζ), 
 as in 7Ζ0ί-μήν^ "the feeder" (from πα-, pa-scor^ πών, Sanscr. pagu, &c.), 
 ηγε-μών^ "the leader" ('ηγε=^δίαγ-). There are also numerous adjec- 
 tives with this formation, as αΐδψμών, "bashful," μνή-μων, "mindful" 
 (cf. μνή-μη)^ ίτΐτίο-βά-μων, "going on horseback," &c. We have also 
 participles in -μενοξ (Latin -men, -mnus, Sanscrit -mdna-s), and nouns 
 
 in -ματ=μεντ. Thus < ^ ^ i -μ»; / i equally signify a thig done, 
 ^ ^ \Λε-πραγ-με-νον) 
 
 and differ from Ttgayog only in their reference to the subject as the 
 doer in the former, and in the reduplication of the participial form. 
 That this secondary form in ματ-=μεντ- implies an anterior condition 
 of the noun in μεν-* is shown by the derivative verbs, e.g. κν-μαίνω= 
 κν-μεν-ίω from κν-μα=κν-μεν-τ compared with 7ίου-μαίνω=^ποι- 
 μεν-ίω from ποί-μήν=Λθί-μέν-ς. This is farther shown by adjectives 
 like άκνμων, άπράγμων, άνοάμων. The Latin and Sanscrit give the full 
 affix, as in ar-me-n-t-um, mo-me-n-t-um^ &c., in the former language, 
 ^ri-ma-n-tas, &c. in the latter. The Greek compound suffix μεν or 
 μον often assumes the form of -μω-ν, as in λει-μώ-ν, γνώ-μων, πλεν- 
 μων, &c., to which ser-mon, pul-mon^ &c. correspond in Latin: it is some- 
 times extended into -μο-νή, as in χαρ-μο-νή, τΐείΰ-μο-νή^ φλεγ-μο-νή, 
 τίληύ-μο-νη^ τίψμο-νή (οοτητρ.Λψμα-τ), and άρ-μο-ν-ία{άρ-μο-νι-κό-ς). 
 Corresponding to these forms we have derivative verbs in -αω, as 
 δαι,μον-άω from δαίμων through δαυ-μονψ, we are disposed to refer 
 the active infinitives in -μεναι to the noun of agency in -μονη^ and we 
 regard the passive participle in -μενοζ as a secondary derivative. 
 
 257 (2 β) Terminations compounded of the second and other 
 pronominal elements. 
 
 Under the form -ya or -va or -ha the second pronoun forms with 
 the third the termination -ya-t^ va-t, or va-n-t, in Sanscrit, Ροτ or 
 hot- in Greek. Like the shorter form in -v-g, this ending generally 
 confers a qualitative meaning: thus sinhn-vat, "lion-like," hhaga- 
 va-n-tj "prosperous," bha-va-n-t, "a lord (from bhd, "splendour"), 
 φως, φωτός, for φα^^οτ-ς, τίάχρωξ for τίατρυ-^οτ-ς (here the Sanscrit 
 pitrwyas presents a longer form of the second pronoun), and the parti- 
 ciples in -ωξ=^^οτ-ς (compare the neuter and oblique cases). The 
 Greek compound terminations in -iog=/or-5 have a feminine in -vla^ 
 in which the v, or labial part of the digamma, is still seen. The 
 
 * Bopp's idea (VergL Gramm. § 801) that -ματ=-μαν is quite erroneous. 
 The form carmentis shows that carmen was originally Garment. 
 
458 THE PEONOMINAL TERMINATIONS [bOOK III. 
 
 shorter but analogous termination in -νς has a feminine in -ω. Here 
 the case is reversed, the masculine having kejit, while the feminine 
 has lost, its characteristic breathing. We are disposed to infer from 
 the feminine ending, -via, and from the appearances to which we have 
 before called the reader's attention, that the termination -v-g was ori- 
 ginally written -νιξ or -J^iff, and that the termination -ώ is a contrac- 
 tion of fa. The change from vis to us is shown in the Oscan ke-us 
 for Gi-vis; see Varronianus, p. 125. We have endeavoured to show 
 before that the vowels i, u^ never appear but as the representatives of 
 some lost or vocalized consonants, and we think that whenever they 
 are found in a Greek termination, we may conclude that the element 
 of the second pronoun is comprised in it. The Sanscrit feminine cor- 
 responding to the Greek in -via^ is -ushi. It is to be remarked that 
 the feminine noun ^Ηώζ corresponds to the Sanscrit ush-as•^ and the 
 analogy between Ήώζ"Εως, and the particle fog, originally afog ( Var- 
 ronian. p. 288, above, § 248), may convince us that the former in- 
 volved a digamma, which is farther confirmed by the form Ανωξ. In 
 the same way, αιδώς must be regarded as derived from aldo-fig. The 
 masculine ηρως, which we shall discuss more at length in a future 
 chapter, includes φώζ-ς mentioned above, for the original form was 
 r]Q-faOT-g. It is also curious that in the Sclavonic languages o-va cor- 
 responds to the Greek ending in -ώ, as Janova=Joannis uxor (Pott, 
 Etymol. Forsch. ii. p. 486) 
 
 258 With the third pronominal root under the form -νη the 
 second makes a class of abstract nouns in ΰν-νη, as ΰωφρο-ΰν-νη, 
 δίκαίθ-6ν-νη, καλλο-ΰν-νη^ &c., which are nearly equivalent in meaning 
 to those in -νη and -μο-νη. Indeed, καλλοόννη, which means "that 
 which is of the quality of beauty," differs very little from κάλλο-ςζ=ζ 
 κάλλο -t^ "objective beauty." We have no hesitation in classifying 
 the termination -u-vog with -ΰν-νη, to which it bears the same relation 
 that -ί-μος does to -ΰι-μος^ -log to -uiog^ -la to -(Jtg, and the genitive in 
 -10 to that in -6lo. The adjectives in -L-vog express the material out 
 of which any thing is made, or rather they imply a mixed relation, of 
 quality and origin, to the object denoted by the substantive from 
 which they are derived. Thus ξνλ -L-vog means "of wood," "wooden," 
 οΰτράκ-ινος, "of earthenware," testacetis, υάλ -LVog, "of glass," "glassy," 
 κήρ-ίνος^ "of wax," "waxen," δοφ&ερ-υνος, "of tanned leather," "lea- 
 thern," and similarly when there is only a metaphorical reference to 
 the materials, as ΒότΐΕρ-υνός, "late," i. e. belonging to the eventide, 
 Tto&s-i-vog, "full of desire," ελε -L-vog, "pitiable," &c., and άληϋ'-ί-νός 
 signifies "genuine," i. e. "made up of that which is true." This last 
 adjective is particularly applied to express that which is all that it 
 
CHAP. ΠΙ.] OF THE UNINFLECTED FOEMS. 459 
 
 pretends to be, for instance, pure gold as opposed to adulterated metal. 
 Compare the αλη^ινον ΰτράτενμα of Xenophon {Anal•, i. 9, § 17) with 
 the κα^αφξ ΰτρατός or το καΟ'αρον του ΰτρατον of Herodotus (ι. 
 211, IV. 135) and Thucydides (v. 8), where the reader will remember 
 that the better class of citizens are contrasted to those of lower origin 
 as good coin is to bad (see Aristoph. Acharn, 517; Ban. 719 foil.). 
 This force of άλη&ίνός seems to have escaped the notice of all the 
 commentators on Theocritus (xm. 14, 15*): 
 
 ώ^ αυτώ κατά %'νμον δ Ttalg ττετΐοναμενος εϊη, 
 αντω δ' εν ελκών ες aXad'Lvbv ανδρ άτΰοβαίη, 
 where ελκών does not refer to oxen drawing the plough, as the editors 
 suppose, but bears its common sense of "weighing," "being heavy," 
 "drawing down the scale." So that the passage means that Hercules 
 brought up Hylas with a view to expel all dross and adulteration 
 from him, in order that he might, by "weighing well," like pure gold, 
 turn out a genuine man : just as Plato, speaking of the military caste 
 in his state (i?e5p.iv.p.428E), says: τΐότερον ovv Iv τγ] πόλει οϊει ημΐν 
 χαλκέας πλείονς ενεβεϋ^αυ η τους αληθινούς φύλακας τοντονς ; be- 
 cause, according to his fiction, the artizans were made of copper or 
 steel, but the guardian soldiers of pure silver (ni. p. 415 a). In the 
 same way Theocritus says (Epigr. xYii. 3) : ώ Βάκχε, χαλκεόν viv 
 άντ άλα^υνον τιν ώό' ανε^Ύΐκαν\. The adjectives in -ι-μος or -(5l- 
 μος express a quality by virtue of the first part of their termination, 
 and also an action like the nouns in -μος. In fact, by this appendage, 
 the relative word becomes sul•ject^υe\ thus αλω-ΰυς signifies "a cap- 
 ture," and άλώ-όί-μος Ttaiav, "a song of triumph from the captors." 
 
 259 It is difficult to believe that the large class of words in -ϊχης, 
 -ίάτης, -ατής, -ι^της, -ωτης, ought to be classed among those terminat- 
 ing with the affix -της only : for, if so, how are we to account for the 
 penultimate syllable? It may seem strange that the Greeks should 
 have written both τίολίήτης, or τίολίτης, and δημότης, φνλετης', but it 
 cannot be denied that the former contains something in addition to 
 the termination of the two latter. We must endeavour to ascertain 
 
 * Dr. Wordsworth, very ingeniously, proposes to alter εν ελκών into 
 εξ (χ'άιλων, but we still prefer the old reading with the interpretation given 
 above, which is farther confirmed by the additional passage from Theocritus. 
 
 f Trench has given some additional illustrations of the ending -ivng 
 {Synonyms of the New Testament, pp. 28 sqq.); but although his object is 
 theological, he has not remarked that Θεον άληΟ^νον εκ Θεον άλη&ινον 
 in the Nicene Creed refers really to the statement that the Son is όμοονϋίον 
 τω Πατρί, "of one substance with the Father,'' which is just the point 
 to be noticed. 
 
460 THE PEONOMINAL TERMINATIONS [bOOK III. 
 
 what this addition is. The termination -trT^g, which is probably 
 always an abbreviation of -ίάτης or -ιητης, is appended to substantives 
 of all declensions; thus, χωρίτηςίνοπιχώρα^ OTtXLtrjg fromoTcXov, itoXi- 
 τηζ from τίόλίζ. "With the exception of ητίείρώτης from ήπειρος^ the 
 termination -ώτης is appended to those nouns only which end in -ta 
 and -εία. Now these nouns are combinations with the second pro- 
 nominal element under the form i/a•: thus, Ίταλ-ό-ς, ^ItaX-ia, Ίταλ- 
 ιώχηξ. In tins case, therefore, we conceive the termination is com- 
 pounded of the second and third pronominal elements. May not the 
 others be so likewise? In the Latin terminations -as (for cits) genitive 
 -atis, the third pronominal element does not appear in so full a form 
 as it does in Greek, but the length of the penultima points to a com- 
 bination of the two elements as in the Greek. The whole question 
 will be set in a clearer light, if we consider in general what is the 
 origin of the ethnic names. Now, either the name of the country is 
 derived from ^that of the people or vice versa. When the former is 
 the case, the name of the country generally ends in -ία or -t-xij, which 
 are relative endings affixed to the gentile name : thus, ^Ιταλόξ makes 
 ^Italia, Αάκων makes Λακωνική. But there were two classes of inha- 
 bitants in countries of which the Greeks were wont to speak and 
 write; the native inhabitants, and the Greek settlers. Thus, if 'Ιταλία 
 is the country of the 'Ιταλοί, a person living and acting there would 
 be Ίταλί{^)της, which is therefore a secondary formation, or includes 
 both pronominal endings. The Romans, in like manner, would call 
 Hispan-ia the land of the Hispan-i, but a Roman living there would 
 be called Hispaniensis (see Ruhnken ad Sueton. Ccesar. § 37). If all 
 the Greek nouns of which we are speaking are secondary formations, 
 we can now understand why we have τίολυήτης, τίολίτης, but φνλέτης, 
 and δημότης. The two latter are derived from the substantives φυλή, 
 δήμος, which are themselves formed from the verbal roots φν-,δΓε, by 
 the suffixes -λη^ -μος; the introduction of the syllable ya^ ta, would 
 therefore be quite superfluous. But τίόλ-ις from the root τιολ- {τίολνς, 
 &c.) is equivalent to πολία, just as the suffixes -ΰις and -(?ta are iden- 
 tical ; the form τίολίήτης^ τίολίτης, is therefore necessary for the second 
 derivative. For ή^ευρώτης, οτίλίτης, we must suppose intermediate 
 forms ήηείρίος, οτίλιος. The latter is presumed in the secondary deri- 
 vative, and may be inferred from the words ενότΐλυος, τίανοτίλία. We 
 might therefore suppose the original existence of a word οτίλιήτης or 
 οτίλιώτης corresponding in form and meaning to άΰτηδίώτης (Horn. M. 
 XVI. 167, cf. Eurip. Here. F. 159—161): for the οτίλον or "thing 
 moved about in defence" (επω), and the ρόπαλον or "thing brought 
 heavily down to strike" (ρέττω), would constitute the two arms, pro- 
 tective and ofi'ensive, of the primitive warrior (see on Antigone., 115,6). 
 
CHAP. ΠΙ.] OF THE UNINFLECTED FORMS. 461 
 
 There is the same parallelism between ώκεα -vog^ the swift river that 
 bounded the earth, and ovQa-vόg^ the broad expanse which rested upon 
 it, according to the ancient idea. There is no occasion therefore for 
 the derivation proposed by Bopp {Gloss. Sanscr. p. 334). In like 
 manner for the patronymics -ίων, -ίώ-νη•) -ί-νη, we must presume inter- 
 mediate proper names in -lag, -ία. It is important to remark that the 
 nouns in -ίccg have occasionally by-forms in -ίτηg', thus, we have both 
 vsβQ-ίag and vεβQ-ίτ'ηgi the latter having a compound, the former only 
 a simple ending. Pott's supposition (Etymol. Forsch. ii. p. 559), that 
 -la-trig contains the Sanscrit root i or gd^ "to go," is founded on what 
 we consider a misconception respecting the nature of these forma- 
 tions. 
 
 260 The large class of nouns in -ών, -G)Vog^ must be referred to 
 the same origin as the genitive plural, and therefore, as we have 
 already seen (§ 250), are derived from a combination of the second 
 element under the form Cl- with the third element v-. They denote 
 a place of collection or aggregation : thus, avδQ-ώv is "a place for 
 men," icaQ^iv-mv^ "a maiden's chamber" (hence the temple of the 
 virgin goddess Pallas), άγ-ών, "a general place of meeting," αμτίελ-ών, 
 "a place of vines," and even ai-cov, "a collection of periods." To all 
 these the derivation from the same source as the genitive case is very 
 appropriate, for the same idea is involved in each. This community 
 of origin is farther shown by the form -εών, which is found in Ionic 
 and old Attic (Lobeck, PhrynicJius, p. 166). The further affix -ία is 
 sometimes found, as in ροδ-ων-ία, μν-ων-ία. 
 
 261 (3 a) BedupUcations of the third pronominal element. 
 Except in the nominative case of the few nouns which end in -τηgy 
 
 -roug (above, § 288), and in words in -νη, -VLog, -VLKog (§ 255), the third 
 pronominal element can hardly be said to be in itself the vehicle of 
 any ulterior formations, although, when appended to the other ele- 
 ments, it is often followed by additional syllables. The forms -τε-ρ, 
 -τε -Qog, &c. must, like the third numeral, be considered as corruptions 
 of a compound of the second element under the form tv or rt, and ρ-. 
 But the third pronoun under the form na is very often followed by 
 the other form ία, and in Latin and Sanscrit this combination is 
 reversed. Thus we have cras-ti-nus, pris-ti-nus, in the former, and in 
 Sanscrit we find both -tana and -tna, as in hya-tanas=hesternus, and 
 nu-tnas=novus. In Latin and Sanscrit, but not in Greek, we have 
 a combination of the third and first elements, as in in-ti-mus, "in- 
 nermost," punya-ta-mas, "purest." The superlative-ending -ra-Tog 
 is a direct repetition of the third element. 
 
462 THE PBONOMINAL TEEMINATIONS [bOOK III. 
 
 262 (2^) The second pronominal element under the form da or Q^a. 
 
 We have already mentioned that the element d- or Q•- is to be 
 regarded as a corruption of the second pronoun. At first sight these 
 articulations might seem to stand rather for the third than for the 
 second element. But the question is easily determined by an exami- 
 nation of their employment as flexional endings. For while it is 
 quite clear that d- is equivalent to κ- and v- as a derivative suffix 
 (§ 254), there can be no doubt as to the identity of ΰί, J^, d- as the 
 signs of the genitive or ablative (§§ 247 — 249). The simplest shape 
 in which d- or θ- appears is as the last letter of feminine crude-forms, 
 as in φνγάδ-ς {φνγά -g), Έλλάδ-ς (Ελλάς), ληΰτρίδ-ζ (ληβτρίς), ogvLd's 
 (oQVL-g), %OQV%'-g (κόρν-ς), &c. According to Bopp (Vergl. Gramm. 
 p. 139), the d- is merely a secondary, unorganic addition, intended as a 
 vehicle for the case-endings, and yet he says (p. 147), that κόρνΟ'-ς and 
 ogvL%^-g are compounds, the one denoting "what is placed on the head," 
 as from τί^'ημί, and the other "that which goes in the wood," as from 
 !&8ω, "to run." We believe that in all cases d- or 0'- is as distinct and 
 significant a pronominal suffix as any other: for instance, why is λο- 
 γά-δ -g not a derivative, if λεκ-τός is ? In the first and most general of 
 patronymics, παΐ-δ-ς, this ending appears in the shortest form, and 
 also in the patronymic Θέο-γνι-δ-ς (comp. Θον-κνδί-δης). The root of 
 ΰΐαΐ-δ-ς is that pronunciation of the first tenuis which in all the lan- 
 guages of the Indo-Germanic family is one of the first sounds uttered 
 by an infant, to express at once one of the first persons whom he sees 
 — his father, — and one of his first wants — food ; a similar combination 
 of the mother and the breast is found in the cognate sound ma, which 
 is merely a modified utterance οι pa; or perhaps ma is the first sound, 
 and pa the second, uttered by an infant. Comp. TtarCTtag, ραρρα, 
 "papa," "pap," papulla, with μάμμα, mamma, "mama," μύξευν, 
 "mouth," μαξός, &c.: see Varron^an. p. 49. Combined with the ter- 
 mination expressing an agent, we have τία-τηρ, μά-τηρ in Greek, and 
 similar words in all the cognate languages ; from the former root with 
 a more general ending we have Λα-οι=ε7ακτητοΙ dvyysvelg, then τίώ- 
 λος, &c.m Gr eek ; pue-r (hac. τίόϊρ), pu-sus, puUus, pusillus^ disci-pulus, 
 disci-p-lina, &c. in Latin; Sansc. putra, Pers. pussr, pur (whence 
 Shahpur=regis filius, and so on. The same derivative sense may be 
 recognised in the feminine nouns in δ- or ^-. "That which comes 
 from or belongs to the wood" (Sanscrit arani, Lat. ornus) would be 
 as good an explanation of ορνι^ -g as any other, and the importation of 
 a verbal root is quite gratuitous. The termination δ- appears in a 
 longer form in the verbal substantives, like βά-δος, χλί-δη, &c., and in 
 the common patronymics, as Κρονί-δης, Θονκνδί-δης{ίΙΐ6Ώΐΐτη6 Θεοκν- 
 
CHAP. III.] OF THE UNINFLECTED FOEMS. 463 
 
 δης occurs in Herod, viii. 65, &c.). To the same class with these patro- 
 nymics we may fairly refer the derivative endings -devs and o'fog= 
 δε fog to indicate a young animal, or as secondary relative; thus we 
 have άλωτΐεκο-δενς, "a young fox;" άεη-δενς, "a young eagle;" d'vya- 
 τρί,-^εο^, "a daughter's son;" άδελφί-δεος, "a sister's son." In the 
 ^olic dialect this suffix appears as -δiog(Bekkeri Anecd.Tp.GS4:: δ των 
 Αίλεων ϊδιος rvTtog 'TQράδLog. "Τρραγαρ TtaCg δ Πίττακός), and the 
 same affix appears in the diminutives βοί-δίον^ γαΰτρί-δίον^ βοτρν-δίον, 
 &c. The second pronominal ending precedes the -δ7jg in ΆγLάδηg from 
 'Ayig^ Baκχiάδηg from BάκχLg, &c. Compare 7toλ-L•ψτηg with δημό-τηg. 
 As some of the female patronymics are formed in the same way as 
 the feminine nouns mentioned above — thus from ^oρεag we have 
 masc.patr. Boρεά-δΎιg^ femin.patr. Boρεάg{gQn. Βορεάδος)^ Soph. Antig. 
 985, from Τάνταλος we have masc.patr. Τανταλίδης, iemin.-psitr. Tav- 
 taXCg= Τανταλιδς (gen. Τανταλίδος), &c. — we cannot consider the δ 
 in the feminine nouns mentioned above as an unorganic fulcrum, and 
 must regard it as the elemental letter of the termination da., which we 
 have recognised in the ablative or genitive case ; and we have seen, that, 
 in the fullest form of this case, -ulov, -lov, it serves as the inflexion of 
 the comparative degree. The relationship of these inflexions is abso- 
 lutely proved by an examination οί"Ιων=Ίά^Γων compared with the 
 feminine form Ίάδ-(ς), and the adjective ^'Iccoog (Ίαύον'Άργος). For 
 while the former preserves the digamma, the two latter stand related 
 as medius to μέΰος, originally μέβΰος. Similarly we have "l7Cna6og 
 {ίτίτίάδ-^ Πήγαΰος (μηγάδ-), &c. And Ίαΰων appears as a substitute 
 ίονΊ()(6ΰων=Ία6ίων=Ίαδίων (Budenz, I.e. p. 69). Hence it is inter- 
 esting to remark, that we find -ιων as well as -δηg=δyag used for the 
 expression of a patronymic. In the patronymics in δ- the feminine 
 form is shorter than the masculine; in those in -ίων=ίον-ς the con- 
 verse is observable, the feminine being -ιώνη and -ϊνη. The fair 
 inference from this is, that the feminines in -δ -g are anterior to the 
 masculines in -δης, but the feminines in πωνη^ -υνη later than the 
 corresponding masculine nouns. As it is pretty clear that the patro- 
 nymics in -δης and -ίων find their common origin in the sign of the 
 genitive case, we may expect that, with a little vagueness occasion- 
 ally, their significations will correspond. Indeed, the distinctions 
 which we observe are casual or arbitrary, and the vagueness is shown 
 by the accumulation of one ending after the other. The terminations 
 -ίονί-δης^ ιά-δης, ίων-υάδης, would sometimes express the son, some- 
 times the grandson ; thus from "Ατρευς we have ^Ατρείων, Άτρείδης 
 (Agamemnon or Menelaus, the grandsons of Atreus); thus also we have 
 the forms [Ταπετ-ίον-ί-δηςίνοϊη'Ιάτίετος, Ταλα-ων-ί-δηςίτοτη TaXaog, 
 Άκρίΰίων-ίά-δης from Άκρίΰως, &c. In the names of tribes, supposed 
 
464 THE PEONOMINAL TERMINATIONS [bOOK III. 
 
 to be the extant representatives of remote ancestors, we always have 
 -δης or -d«t, never -Lovsg; for example, the Athenian tribes are called 
 Άργαδης^ Βοντάδαυ, EvTtatQidai, &c., and we have clans or castes called 
 '^Ομηρίδαι, Άΰκληπιάδαι,, &c. The termination -δενς, whence Άργαδης, 
 expresses also general derivation without reference to any proper 
 name, as in the words νίδενς^ sing., λεονηδης^ χ^νιδης^ plur. There is 
 a particular class of patronymics, principally found in the Boeotian 
 dialect, in which the second element reappears in the guttural form; 
 such are Άΰώταχος^ Βοβπόρίχος^ Θείβιχος^ Ίΰμτ^νιχος^ Καβίριχοξ, 
 Λεόντιχοζι Φρυνιχοζ (Ahrens, Dial.^ol. p. 216); and these forms 
 may no doubt bear the same relation to those in -δ that ορτάλιχοζ^οβΒ 
 to 6ρταλίδ-(ς)^ άρνΰτιχοςϊο άρν(3τίδ-(ς) from άρνότηρ, and 'φιξ ('φϊχός) 
 to 'φί£('ψϊδός). The Boeotian patronymics in -ώνδας seem to be derived 
 from participial names, as Χαιρώνδαζ from χαίροντ, Χαρώνδας from 
 χάροντ, Κρεώνδαςίνοπίκρέοντ^ ΊΊαγώνδα^ίτοτΆτίάγοντ^Ετίαμείνων- 
 δας ον^Ετίαμινώνδαζ^ from ετίαμννοντ. The participle άμίνων^ αμυ- 
 νών, involved in the last word, is particularly interesting from its outward 
 identity with the comparative άμείνων. When we place the correla- 
 tives αμείνων and χείρων side by side, we are led to the conclusion, 
 that, standing as they do for ά-μεν-ίων^ χερ-ίων, they must be formed 
 from some such words as ά-μεν-ενς, χερ-ενς. Now the former of 
 these, on the analogy of tt-oζog, ά-οόΰητηρ, would imply some one who 
 stands or remains (Εητψ. Here. F. 163; Soph. Aniig. 671) by us in 
 battle: while χερ-εvg would denote a handicraftsman or labourer; and 
 thus the usual opposition (Varronianus, p. 24) between the better 
 and the worse, between the warrior and the workman, would be ex- 
 pressed in the terms of the language itself. The more common form 
 of the word expressing assistance in battle is άμννω, and we have the 
 same form with the same meaning in the Latin munia^^mcenia^ and 
 murus=moerus. The explanation of this long u is not difficult. "We 
 have already seen that comparatives in -ι,ων presuppose a positive in 
 -vg or -ρόg. We have no trace of the latter here, and must therefore 
 assume the former. Accordingly, if the positive was ά-μεv-εvg, the verb 
 would be άμεν-νω, and this, on the analogy of ελαννω, would pass into 
 αμεύνω=^αμυνω, just as άμενίων would pass into αμείνων. The change 
 of άμννω into αμίνω would be natural enough in the JEolic dialect, 
 which often substituted ι for ν (Ahrens, Dial. ^ol. p. 81). Besides 
 the participle in -ων, which forms the basis of the proper name ' Επα- 
 μειvώvδag, we have also the ΐoττΆJμειv-ίag corresponding to KaXliag, 
 
 * That this is the genuine Boeotic spelling is clear from the inscrip- 
 tions; see B5ckh, C. 1. i. p. 723. Thus we Ιιανβ'Αμινίας, Nos. 1584, 1608. 
 '^^μινόκλ'εις, 1563 b. Έηαμινώνδοις, 1574. 
 
CHAP. III.] OF THE UNINFLECTED FORMS. 465 
 
 &c. Consequently, the form ^Επαμεινώνδας may be considered as a 
 variation of the double form 'ΕΛαμεινίάδας. Aristophanes plays on 
 the interchange οίΆμευνίας Άηά'^ί4μννίας (SchoLad Nub. 31)', and a 
 comparison of ταμ-ίαξ and ταμ-ών may show us the correspondence 
 between the forms in -tag and -ων. Besides, '^fiwioi' itself occurs as a 
 proper name (Aristoph. iJcde*. 365). The names' Ττίερίων aud'Evdv- 
 μίων, which represent the rising and setting sun (see for the latter 
 Max Miiller, Oxford Essay s, 1856, p. 49), have a quasi-participial 
 value, though their terminations are merely relative or comparative. 
 With this explanation, the comparative αμείνων will stand in good 
 parallelism to its synonym άQεiωv ίνοτα^'^ρης or Άρενς, and both will 
 signify pre-eminence in war. Similarly, %ρείαΰων^ another synonym, 
 refers to the possession of greater strength or power, and perhaps 
 there may be some connexion between bonus (anciently duonus\ βελ' 
 τερος=βεν-τεροςΒ'ηάδνναμαί*. At least there seems to be little doubt 
 that duonus must be akin to the Gaelic duine, "a man," in the em- 
 phatic sense, i.e. vir^ άνΎΐρ•,νάϊ&τίο,&^θί\ί&Ν& duineadas=.viriUtas, virtus . 
 Cicero says (deFinibus, iv. 26, § 73), ^^bonum, ex quo appellatum sit, 
 nescio." And if it really belonged to the Celtic ingredient in the 
 Latin language his perplexity was very excusable. To return, how- 
 ever, to Έτΐαμεινώνδας^ we may regard this word as a proof of the 
 contacts between the participles in -ων and the comparatives in -ιων, 
 and we may explain the addition of the patronymic ending -δας to 
 these Theban names derived from participles by a reference to some 
 other forms in which a verbal signification is distinctly included. 
 
 263 We have a long list of adverbs terminating in -δήν^ which, as 
 Grimm rightly observes (ni. p. 239), are to be classed with the Latin 
 in -im, -Urn (see Varronianus, p. 289), and the German in -ingen^ 
 -ling en; thus he compares βάδην, gradatim; βλήδην, wurfiingen; ΰνλ- 
 
 * There has been a great deal of vague writing on the subject of these com- 
 paratives. For instance, Bopp (Vergl. Gramm. p. 421) proposes to consider 
 ά-μείνων as compounded oi a privativum, and μείνων=ηιιηον, and he finds the 
 same compound concealed in omnis ! Ooaer\em(Syn. u. Etym. v. p. 349) derives 
 άμείνων from μένος, μέμονα, in the sense of "willing," and finds the same idea 
 in the connexion vs^hich he assumes between βελ-τερος and vel-le. And the Pro- 
 fessor of Comparative Grammar, to whose ludicrous performances we have oc^ 
 casionally adverted, has put together a tissue of absurdities in his attempt to 
 trace the Greek, Latin, and English synonyms for good^ better, best, and well, to 
 a common origin. Yov\ns,t2ince,opti'musi& o-pet-umus=o-bet-umus=^bet-est=best ; 
 the initial vowel being there merely to furnish the astonished reader with the 
 necessary exclamation. Some cruel wag will suggest that the author of such 
 derivations ought to be promoted at once to the professorship of superlative 
 philology! 
 
 Η Η 
 
466 THE PEONOMI^AL TEKMINATIONS [bOOK III. 
 
 ληβδην, Gonjunctim; γράβδην, rizilingen; άρπάγδην, raptim; κρνβ- 
 δην, clam; δρομάδην, lauflingen; φυγάδΎΐν^^ηηοΗτη; &c. These ad- 
 verbs sometimes appear under the shorter forms -δα^ -δον, -δίζ^ some- 
 times unter the longer forms -νδα^ -νδην, and -vd'a^ the last however 
 in two instances only, μίνννΟ'α and ολίγιν^α. We occasionally find 
 nearly all these terminations appended to the same root, as κρνβδα-, 
 κρυφηδόν,κρνφάδις,κρνβδην,κρνφανδόν. The explanation of these 
 forms is much facilitated by their appearance in a special class of 
 words, namely, the adverbs used as secondary predicates with παίζειν 
 to describe some particular kind of game ; such are βαΰίλίνδα, δρα- 
 πετίνδα, ελκνβτίνδα, εφετίνδα, κρντίτίνδα, κνβιότΜα, ληκίνδα^ 
 μνίνδα, ούτρακίνδα, ΰτρετντίνδα, φαυνίνδα, φρνγίνδα, χαλκίνδα. 
 It is clear from the instances in which we perfectly understand the 
 formation of the adverb, that these words are the cases signifying 
 direction and tendency (like οϊκονδε) of verbal abstract nouns. Thus 
 ελκνύτίζ must be assumed as meaning "a pulling," δραηετίς, "a run- 
 ning away," οβτρακίξ, "the use of a potsherd" {νΐ.οΰτρακίζώ)'•, so that 
 τίαίζειν ελκνότίνδα must mean " to play in the way of, in the direction 
 of, a pulling," and so of the others. The termination -δα is another 
 form of δεν=&εν, which, as we have seen (§§ 245, 248), is ultimately 
 equivalent to the merely locative ending '9't(v) οΐφί(ν)•, and the genitive 
 and locative sign are really traceable to a common origin (§ 235). 
 Consequently the difference between a form like κρνφανδόν and one 
 like κρνβδα or κρυφηδόν, is really the same as that between the two 
 adverbial case-forms οϊκονδε and οΙ'κοΟ'εν, which have the same ter- 
 mination signifying motion from a place, but differ in the accusative 
 -V retained in the former word, so that in οϊκονδε the inflexion deno- 
 ting removal is added to that indicating direction or motion onwards, 
 and the whole form takes its signification from this element. A care- 
 ful examination of all the adverbs now under consideration would 
 convince us that the meaning which they convey, whether they are 
 more immediately connected with nouns or with verbs, is simply that 
 which would be produced by the suffix -^'εν, or the patronymic suffix 
 -δης, -δ -g, that, namely, of proceeding from, being deduced from, caused 
 by, in the manner of, &c. Thus, to take those formed from nouns, 
 κλαγγψδόν is equivalent to κλάγγψΟ'εν, καναχψδά to κανάχη-^'εν^ 
 &c. With regard to those formed from verbs, we must first consider 
 what would be the meaning of a noun formed from a verb-root by the 
 suffix -δης, -δ -g. Thus, from the root βα-, "to go," we have βά-δος, 
 "a going," also βά-δί-ΰις, &c., and εμβά{δ)ς, "a shoe:" from φνγ-, 
 "to fly," we have φνγή, "flight" or "fleeing," but φνγά{δ)£, "a fugi- 
 tive;" so that these words express that which comes out of the action 
 of the verb, i. e. the manner of it. Just such a meaning we have in 
 
CHAP. III.] OF THE UNINFLECTED FORMS. 467 
 
 the adverbs βά-δην, εμ-βα-δόν, φνγ-δα, where the forms -δον, -δα, -δην, 
 diifer only as τντΐτόμε^Όν, τν7ίτόμ£%'α, τύτίτετον, τντίτετην, in the 
 verbs, which, as we shall hereafter show, were originally identical. The 
 relation between these adverbs in -δα,&ο. and the corresponding forms 
 in -νδα, &c. is just that which subsists between the inflected verbal in 
 δ- and the common participle in -ντ, e. g. between φνγάξ=^φνγάδ-ξ•, 
 and φυγών =φνγόντ-ζ, or between the Latin synonyms cupi-dus and 
 cupi-en(t)s. That the τ (ή, which appears in the ordinary form of the 
 Greek and Latin participle, represents a residuum of J^=dv or tv is 
 proved by the ordinary Latin gerund in nd-, by the by-forms of the 
 Latin participle, e. g. oriundus=^orien{t)s, &c., by the verbals in -τόξ, 
 -τεός, which must have been originally -rsPdg, and by the abstract 
 verbals in -xvg corresponding on the one hand to the supines in -iw-, 
 and on the other to the verbals in -(jig, -tio^ &c. Accordingly the par 
 ticipial adverbs in -νδα, -νδον, &c., are really inflexions of the cor- 
 responding participles, and the participle itself is thus proved to be a 
 further inflexion from the simplest form of the verbal noun, just as the 
 τίαιδίών ονόματα in -ίνδα (Julius Pollux, ix. 110) are formed from 
 abstract nouns in -ΐξ. The two in -v%^a must be compared with εν^α, 
 IVO'fV (cf. i-nde\ and the interesting word λαβυρυν^Ός shows that they 
 belong to the same analogy as the adverbs in -νδα. For λαβύρινθος 
 is evidently formed {νοτηλαβνρινΟ'α,ι.β.λανρίνδα ονλα^ρίνδα, "shaft- 
 wise," i.e. a place constructed of shafts, ducts, adits, or narrow passages, 
 whence Αανρεΐον, '*the place of shafts," as a name for the silver mines 
 in Attica (see Welcker, Tril. p. 212; Kenrick, Egypt, p. 190). There 
 can be little doubt that a similar explanation is applicable to the 
 proper names Τίρυν{θ')ς, Κόριν^Ός, &c.^ though their origin is more 
 obscure. The word ελμίνθ-ς, "a worm," which Bopp explains as 
 " winding itself" ( Vergl. Gramm. § 803), seems to include the simpler 
 form exhibited in the Latin vermis, Germ. wurm. 
 
 264 It is well known that these adverbs are not formed from verbs 
 which take a J in their derivatives, with the exception of βνξην, βνξόν, 
 from βνω (Buttmann, Ausfuhrl. Sprl. § 119, 83). From verbs of this 
 kind we have generally adverbs in -(jirt, as ονομα-(3τί from ονομά-ζω. 
 This form is most particularly common in connexion with verbs in 
 -^-ξω, as ελλγιν-ί-ζω, ελλην-ί-βτί, άνδρατίοδ-ί-ξω, ανδραποδ-ι-ΰτί, &c. 
 In some of these adverbs κ is substituted for 6, on the same euphonical 
 ground which has produced such forms SiS βαΰτά-κ-της, ΐτοιηβαΰτά-ξω, 
 though from κτί-ξω we have κτί-ΰτης, and εδψτνς by the side of εδέ- 
 6της, and though we have όρ;^ι^(9τν^, ορχήΰτης, ΰωφρονίύτνς, ΰωφρονί- 
 ύτης, and conversely both ετεητνς and ετΐητης (see Lobeck, Paralipom. 
 p. 19). According to this principle, we have άότα-κ-τί from ΰτάξω, and 
 
 Η Η 2 
 
468 THE PEONOMINAL TEEMINATIONS [bOOK ΠΙ. 
 
 άνοιμω-κ-τί from οΐμώ-ξω (Hermann, ad Soph. Aj. 1206). It will be 
 remarked, however, that most of these verbs have y or κ in the noun- 
 derivates, as ΰτενά-ξω, ΰτενα-γ-μα (not 6τενα-6-μά), άΰτενα-κτί, 
 κηρνΰΰω^ κήρυξ, κήρυ-γ-μα, άκηρν-κτί; οιμώ-ξω, οΙμωγή, άνοιμω- 
 κτί; ΰτάξω, ΰτάγμα^ άΰτα-κτί; and the truer account undoubtedly is, 
 that the ξ of the indicative is a representative of yy or xy. Many ad- 
 verbs of this class have neither β nor κ before the -rt, as αμελλητί, 
 άμεταΰτρετίτί, ανυδρντί, &c., especially when the root terminates with 
 ρ, as αρ-τι^ εγρηγορ-τί, εγερ-τί, &c. These terminations belong ori- 
 ginally to the same class with those which we have just discussed; 
 namely, to the verbals in -rtg, -χνξ-,- τεοζ. They are all locative cases, 
 and bear the same relation to the Latin locatives in -Ητη, -ter, that 
 the ordinary locatives in i do to the older locatives in -iv, -im. Those 
 in -ωΰτί, as μεγαλωύτί, δημίωΰτί, ίερωΰτί, νεωΰτί, &c., are very sin- 
 gular forms ; they comprise, in fact, an union of the old ablative in -og 
 with this locative suffix, an union similar to that which we have pointed 
 out in οϊκον-δε, βαύίλίν-δα, &c. Besides these locatives with the suffix 
 -TLj -κ-τι^ -(?-rt, from verbs, a great number of adverbs appear as the 
 immediate locative cases of nouns, with the ending εί, or i ; thus we have 
 άμιΰ^Ί, αντοβοεί, τίανδημεί^ άμαχεί, &c. It appears quite impossible 
 to settle the orthography of these endings. Blomfield (ad^sch.Prom. 
 216) would write -i in all those to which there are corresponding 
 nouns in -og, on the analogy of οϊκοί, τίεδοΐ^ &c. ; and -ει in the others. 
 But the traditionary orthography on which the varieties depend is too 
 consistent to admit of any such alteration; nothing is to be inferred 
 from the analogy of οϊκοι^ for οϊκει is recognised as a genuine form 
 by Theognostus, and the Dorians wrote χοντεϊ^ τηνεΐ, αντεΐ•, εκεί, as 
 general locatives, without any particular expression of gender (Bekkeri 
 Anecdot. p. 1404). The variation in the orthography and also in the 
 quantity of these endings {Bekkeri Anecdot. p. 5 71; Gramm, ap. Her- 
 mann. de Emend. Bat. Gr. Gr. p. 448) must be reckoned under those 
 anomalies which are due only to caprice and accident, and which are 
 so numerous as to defy all the Procrustean ejfforts of the Porsonian 
 school. With these locatives in ευ, ι, we must of course class those in 
 av, as χαμαί, τίάλαι, Λαραί, &c. We have also older locative forms in 
 -V corresponding to these adverbs: comp. αΐεν, αΙεί\ sometimes even 
 αΐεξ, comp. -^εν, -%ες, -tus, -dhas ; τιάλιν ,τίάλι, τίάλαι ; πριν, 7ΐρό(δενρο, 
 ετνπτετο), περί, πέραν, πέρα, παραί, πρώϊ, παρά, &c. The forms in 
 'η belong also to this class, for in the Boeotian dialect μη, νή, επειδή^ 
 &c, were written μεί, νεί, επιδεί, &c. (Bockh, Corp. Inscript. i. p. 720). 
 So that, on the whole, strange as it may appear, we are compelled to 
 admit an original identity of terminations apparently so different as 
 -ov, -ην, (compare the secondary person-endings of the passive voice 
 
CHAP. ΠΙ.] OF THE UNINFLECTED FORMS. 469 
 
 -μην^ -(30, -ro, &c.), -ϊ, -Γ, -lv, -ft, -at, -η, -ες. To such a distance 
 from an original form in the ending of a word will the arbitrary or ac- 
 cidental divergences of human utterance lead those who speak the same 
 language! or shall we say that the principle of association, working 
 and fermenting in the mind, has generated these by-forms in language 
 to preserve in the outward symbols of thought the idea of likeness in 
 dissimilarity ? 
 
 265 To return, however, to the suffix da. We have before shown 
 on more than one occasion, that, in spite of the obvious suggestion of a 
 simple change of the tenuis into the medial, this element is not a repre- 
 sentative of the third pronominal stem ta, but a shortened form of that 
 word which appears as the second personal pronoun and the second 
 numeral *. The nature of the present researches and the wide field in 
 which they are carried on, does not allow us to bring forward all our 
 proofs at once ; we are now, however, enabled to set forth vcdth addi- 
 tional confirmation, some of the statements which we made in the pre- 
 ceding chapters. It appears from the investigation which we have just 
 concluded, that there is an obvious connexion between the termination 
 -Ti^j, expressing agency, the patronymic -drj-g, where the rj includes y as 
 in the passive aorist £ri;7i7^i'(comp.the-3EoUc patronymics in -δίος^ also 
 dij-kog^ for δείε-λοξ, dyaKog, "as clear as day"), the adverbial termina- 
 tions in -δον, -δην, -δα, %'a, -tl, -tim, -δυς, and the verbals -δος, -δς, 
 -τνς,-τέος. The person- endings of the passive verb may convince us that 
 the terminations -δα, -δε, must have emanated from -δαι through -δην^ 
 -δον; comp. ετντίτό-μην^^Άτνπτο-μαι, τνπτε-ύ&ην ^ϋΙιτντίτε-ΰΟΌν, 
 and τυΛτό-με^Όν with τντΐτό-με&α. "We have before pointed out the 
 identity of -&εν-, ^'ες, with the ablative -d or -t, and the patronymic 
 -δης. The adjectives in -δίος, which generally express immediate proxi- 
 mity in space (Lobeck^ Phryn. pp. 555 foil.), evidently belong to this 
 class, as does also the Sclavonic ending dje, de, or du (Bopp, Vergl. 
 Gramm. -p. 394:). There is only one common ground on which all these 
 forms can meet, namely, the element used for the second personal pro- 
 noun, tva, dva, dya, or tha; and one or other of these natural varieties 
 is represented by every one of the above suffixes, which in meaning 
 and use seem to be equivalent. The Greek /O' is a softened δ almost 
 verging upon y or j. It is found where y appears in Sanscrit, and in 
 some cases it appears to be equivalent to ξ, which is either δ(5-, or y 
 with a guttural or dental prefixed : 'compare Ζενξ, ^εόξ, διός, αίξηός^ 
 
 * The relationship of da to the second as distinguished from the third 
 element, of which the strongest form is na, is illustrated by the fact that in Ar- 
 menian sa=hie, da—iste, and na=Ule. 
 
470 THE PEONOMINAL TERMINATIONS [bOOK III. 
 
 ήΐ&εοζ; x^d^t-^og ίονχ^ί-διος] μετα-^ε with ^έι;, &c. In the terminations 
 'tvg, -τέος (for raJ^og), the original tva is more clearly discerned, the 
 va being vocalized in the former. In the endings -τυ, -tim, the i is the 
 only representative of the additional element by which in these cases 
 the second pronoun is distinguished from the third. In general it may 
 be laid down that the appearance of either i or w in a syllable is the 
 representative of some lost element. These letters, as we have taken 
 some pains to show, are the ultimate vocalization of certain consonants, 
 and not simple articulation- vowels, like a and its lighter forms e, o. 
 We have seen that i sometimes stands as the sole representative, not 
 merely of w, but even of the digamma or a compound of the guttural 
 and labial. It is this letter alone which is left to distinguish rtg-, the 
 corrupted Hellenic form of the interrogative and indefinite, from the 
 common pronoun of the third person, and it is also this which alone 
 remains in -im, -τίζ-, as ν alone remains in -tvg^ to indicate that they 
 belong to the second pronoun. A similar remark may be made with 
 regard to η. This da being the ultimate form of tva^ we shall not be 
 surprised to see it combined with the more original and simple form 
 of the second pronominal stem, in -La-drjg, -do-7tog (lx^o-do-7tog), &c. 
 Its appearance in composition with the element -na is perfectly analo- 
 gous to the compound terminations -μο-νη, -μω-ν, -ΰν-νη, -ΰν -vog* 
 Thus we have άλγψ6ώ-ν, άχ%ψδώ-ν^ κοτνλψδώ-ν, &c., to many of 
 which adjectives in -da-vog correspond, as τηκε-δών, τηκε-δα -vog ; 
 but of course there are many adjectives in -davog — ovTL-da-vog for 
 instance — which have no corresponding substantive in -δ(όν.\ A long 
 series of Latin words in άο{η)-, dinis, may be classed with the Greek 
 nouns in -δών: the Latin termination seems to have the same force 
 as the Greek; compare grave-do{n) with αχ^ψδών^ &c. In Greek, 
 -δav6g^ -δών^ appear to be sometimes equivalent to one another and 
 to -trig; tli^swe have μaκε-δvόg^ μaκε-δώv18i>nάμaκέ'τ7}g, as synonyms. 
 
 266 (S b) The third pronominal element va under the form λα 
 or Qa. 
 
 There are two terminations of most extensive use, -λο^,-ρο?, which 
 seem to agree in meaning. The former is found in a number of ad- 
 jectives expressing objective relations, as tV(p-X6g•, δεL-λ6g^ <^τvφ-λ6g^ 
 μεγά-λog^ or substantives denoting things of a certain Idnd, as κ^ότα- 
 λον^^υμέ-λη^νεφε-λη : sometimes under a longer form, as όμεQδa-λέogf 
 λεvγa-λεogy vricpa-Xiog^ δαιτα-λε -vg'., sometimes compounded with the 
 element -^Off, as in πευκά-λL-μog', sometimes with the element -κogy as 
 in ψλLκ-g (ηλίξ\ τή-λi-κog, &c. In Latin it presents itself in all these 
 forms and some others ; thus we have tremu-lus^ faci-lis, vincu-lum, 
 scapu-la, fi-lius (/tog, φν-ειν\ fe-li-c-s (Jelix), fame-li-cus, &c. The 
 
CHAP. III.] OF THE UNINFLECTED FOEMS. 471 
 
 compound-endings -Xixog, -licus have been preserved in the Gothic 
 and German languages, and even in modern English. Thus we have 
 leiks=:-YiTig\, "like," from a verb signifying "to see: " and hwe-leiks is 
 "what like," German lue-Z-c^er; compare 50-/-cAer, "so-like," "su-ch." 
 The Gothic ga-leik$, German gleich, is analogous to the Sanscrit sa- 
 drigas; compare όά-φης from ΰνν (ΰα-μά) and φως (above, § 181). 
 
 The termination -Qog seems to be equivalent in value to -Xog. 
 Compare ΰκλψρός with 6τνφ-λός, λενγα-λΕος with λν^^-ρό^, μακ-ρός 
 with μεγά-λος, &c., δώ-ρον with κρότα-λον^ &c. The Latin words da- 
 rns, glo-ria (κλε-), prima-rius, Mla-ris, exhibit correspondences to all 
 the simple forms of -kog. In the compound-endings the coincidences 
 are still more striking : compare ficu-l-nus, &c. with hodie-r-nu9, &c., 
 doct-rt-na, text-rl-na, &c. with canti-lena, stercu-li-num, &c., simula- 
 c-rum with peri-c-lum, &c. ; and in regard to the compound -λi-κog we 
 may observe that the Greek and Sanscrit have with the same meaning 
 δέρκω (ε-δρακον), -d-rtg, where the d is one of those prefixes, probably 
 pronominal, which so often appear before simple roots : compare δάκρν 
 with the Sanscrit agru, Lithuanian aszara. 
 
 In fact there can be no doubt that -Xog-, -ρog are etymologically 
 identical, the latter being only a modernization of the former, as is so 
 often the case; com.ipa.YQ glisco,GresGO ; celeher,creher; apostolus, apotre, 
 &c. (see above, § 107). The very same word with modified meanings 
 presents both endings. Thus we find τίά-λαί, Tta-XiV, and τνρίν, from 
 τίαραί: and similarly we have both πoLκί-λog and 7aκ-ρόg, from the 
 same root Λίκ-, "to pierce." It will be remembered that ΊtOL•κίλog and 
 6τiκτ6g and even 7tOL•κιλ6-(^tL•κτog are synonyms : the root 6τιγ- like 
 ^ίικ- means "to pierce" (Buttmann, Lexil. i. p. 18); and JtOLκίλog, as 
 distinguished from aloXog, means "spotted," or marked with circles 
 or points of a difi'erent colour (see e. g. Plato, Eesp. p. 616 e, where 
 it refers to the heaven, as spangled with " patins of bright gold "). 
 Πiκρόg seems to have its proper meaning in Soph. Ajax, 1024. 
 
 267 The identity of the terminations -Aog, -ρog, is still farther 
 shown by the correspondence in meaning of the compound endings 
 -T-Xog, 'Τ-λη, -Ο'-λο^, -d' -λη with those in -τ-ρog, -τ-ρα, -^'^^ρog^ -Ο'-ρα ; 
 compare for example, εχε-τ-λη, "the plough-tail," with αρο-τ-ρον, "the 
 plough" itself; and see Pott, Etym. Forsch. n. p. 555. 
 
 We have already made some remarks on the combination t-{-r 
 when speaking of the numerals and comparatives (§ 157), and have 
 indicated the probability that the element r- is not the third pro- 
 noun, but the second, under the form τν- or -τυ. As a termination we 
 have not only the forms -τε-ρο -g, τ-ρο-^, but also -τερ-?, -τηρ -g, -τορ-^. 
 In import these forms differ little from the simpler ending 'trjg. 
 
472 THE PRONOMINAL TERMINATIONS [bOOK III. 
 
 According to Buttmann {Ausfuhrl, Sprl. § 119 b) the latter is more 
 general, and is used as a kind of participle; thus, while οΙχΎμο^^ζ 
 means, "the inhahita7its of a country," we have in Plato, Fhcedo, 
 p. 1 1 1 c: κ«1 δη καΐ ^εών εδη re καΐ ίερα αντοΐς είναι εν οΐξ τω οντι 
 οίκητας είναι %εονζ^ "in which the gods really dwell." In some cases 
 the difference seems to have been merely that the Attics preferred 
 the stronger form in -ττ^ρ : thus they wrote γνωΰτηρ for γνωύΤΎΐξ-, 
 δοτήρ for δότης, &c. (Pierson ad Moerid. v. γνωύτηρας). Nouns in 
 -τηρ may even be used with a passive signification, like some of those 
 in -της: thus, we have ενδντηρ τΐέτΐλος in Sophocles (Trachin. 671). 
 The termination -τήρ does not differ at all in value from -τωρ. Words 
 compounded with the latter are invariably paroxytones, with the 
 former oxytones, a fact which we might thus express; the nouns ex- 
 pressing agency, which, either from being compounds, or from requir- 
 ing emphasis on their root-syllable, draw back the accent, change the 
 termination -τηρ into -τωρ- Thus, τιατηρ makes ατΐάτωρ., because the 
 accent is thrown back towards the negative a, and we find ρήτωρ not 
 ρητήρ, because the emphasis lies upon the verb-syllable. Just in the 
 same way we have άφρων and ύώφρων from φρην (seen § 1 1 6). Some- 
 times this compound termination appears under the forms -τ-ρό-ς, -τρα, 
 and -τρο-ν, as in Ιατρός, παλαίστρα, κέντρον, which must be supposed 
 to be corrupted from -τηρ, as the more common -ta-ra is from -tva-ra 
 found in qua-tvor (above, § 158). In Sanscrit we have both tri and 
 tar, in Latin tor, -turns, the latter generally as a future participle; 
 also in the feminine as a noun expressing the office or function of a 
 person designated by the ending -tor; thus, prce-tor, prce-tura, &c. 
 The forms -ττρο-^,-τρα, sometimes appear as --Ο'ρο-ν, -Ο'ρα, as ίηολε&ρος, 
 ονρή-Ο'ρα (not connected with Ο^νρα, as Home Tooke supposes, Diver- 
 sions of Purley,u. -p. 316), Λλέ-^ρον,&ο. Giese suggests {^ol.Dialect, 
 p. 108), that this aspiration of the τ is caused by the ρ. This, how- 
 ever, is not to be considered as inevitable, otherwise we should have 
 no terminations in -τρο-ς. A similar change has taken place in Ο^ρΐον 
 compared with τρεις, and in "three" compared with drei. The iden- 
 tity of the terminations -%'ρον and -τρον is manifest on a comparison 
 of αν-τρον with βάρα-Ο^ρον. The former is not connected with άνεμος, 
 αημι, as Pott supposes, but with ανά, and it signifies a passage above 
 ground in a solid substance — e. g. a rock — as opposed to βάρα-^ρον 
 (=βάΟ•α-0'ρον), which implies a passage in the ground below us. We 
 may also compare αν%'ρωπος from άναδρώτίτειν or άναδράν=άνα- 
 βλετίειν (seeLobeck, Paral. 118). The feminine of these terminations 
 in Greek is -τειρα, -τρία, -τριδ, -τηρίδ, and -τραίνα, in Sanscrit tri, 
 and in Latin trie. These forms have been explained in the last chapter. 
 By the addition of the second pronominal element, we obtain the further 
 
CHAP. III. J OF THE UNINFLECTED FORMS. 473 
 
 forms T^QLog in Greek, and -tonus in Latin, both for substantives and 
 for adjectives; thus, δ^αΰττι^ιοξ^ τίοτήρων, senatorius, victoria. By a 
 similar addition the Sanscrit verbal-ending tav-ya is formed from the 
 second pronoun tva. The neuter forms in -τήριον denote the place 
 where the work of the agent, who is designated hj-T7]g, -τηρ^ is carried 
 on. When we wish to speak of a similar place in reference to an 
 agent defined by the ending v-g^Fig, it is only necessary to give the 
 word an adjective form, and put it in the neuter gender. Thus, from 
 τρoφB-v-g=τρoφε-HgJ we have τροφεΐον (compare -uLg, -6iog, &c.). 
 The comparatives in -icov from adjectives in -vg are other instances 
 of this formation. The Latin language has terminations -ber, -bra, brum; 
 -cer, -oris, -crum, corresponding to the Greek suffixes which we have 
 been considering: cre-ber^ verte-bra, mem-brum; pul-cer, volu-cris, 
 sepul-crum. The latter seem to be another form of the ending -cuius, 
 composed of the second element and -lus (Xog). A comparison of 
 u-ber with ον-Ο^αρ, and ru-brum with Ιρν-^ρόν^ makes it possible that 
 some connexion may subsist between these endings similar to that 
 which we find between fera and %'Ύΐρ, so that the b will be a remnant 
 of the labial involved in the second element J^a, just as the c in cr^ cl, 
 represents the guttural portion of that compound articulation. We 
 refer to this class of nouns the names of months ending in -ber or 
 -bris, such as Septem-ber. We cannot suppose with Bopp (Vergl. 
 Gram. p. 436), that there is any necessity for having recourse to the 
 Sanscrit substantive vara, signifying " time." Bohlen (Das alte Indien, 
 II. 445) considers the ending -brum connected with the Sanscrit bhri, 
 "to carry," so that candela-brum would signify "the light-bearer." 
 If the importation of a verbal root were necessary, why should we 
 not go at once to the Latin root feri The large class of verbals in 
 -hi-lis, to which nouns like fa-bula might be added, should induce us 
 to explain the nouns in -ber, -bra, -brum, like those in -cer, -cris, -crum, 
 namely, by a reference to the constant interchange of I and r. This 
 is particularly exhibited in Latin in certain assimilations arising from 
 abridgment. We have shown elsewhere {Varronian. p. 435), that 
 sacellum was originally sacra-culum; it is well known that puella 
 is a corruption of puerula; and no one will doubt that castellum is 
 derived from cas-trum. The affinities of this last word are very 
 interesting, and we will digress from our immediate object to trace 
 a few of these ramifications. The root cas- conveys the cognate 
 significations of "purity" and "protection," which are related as 
 effect and cause. When religious reverence throws its shield over any 
 person or locality, it becomes, according to the Greek notions, ouLog ; 
 but it is βBβηλog when that protection is withdrawn. The idea of 
 order and arrangement is similarly opposed to that of confusion and 
 
474 THE PEONOMINAL TERMINATIONS [bOOK ΠΙ. 
 
 license. We can therefore understand why the same root cas- should 
 enter into cas-nus=canus, "white" (Varronian. pp.53, 106), cas-tus, 
 "religiously pure," casa, "a covered building," cas-trum, "a fortified 
 enclosure," xad'a-Qog, "undefiled," κεΰ-τός, "the ornamented garment 
 of Venus," κάδ-μος, "a suit of armour" (mythically the husband of 
 αρμονία), κάό-τωρ, "the mailed warrior" (o χαλκομίτρας Pind.), κ6<5- 
 μος, "ornament and order," Sanscr. gud, "purificare" (Humboldt, 
 Kosmos, Vol. I. note 27), &c. Combining all these and many other 
 parallelisms which might be adduced, we shall see that the Latin 
 caS'trmn, like the Greek τέμενος and its cognate temp-lum, combined 
 a signification of safety with that of sanctity, and held out a warning 
 to all intruders. The plural, which is most generally used, merely 
 indicates the collective nature of a camp, and is therefore quite analo- 
 gous to mcenia, &c.* 
 
 268 From all these comparisons, it must appear pretty obvious 
 that -λος and -ρος are identical terminations; and when we consider 
 the manner in which they are combined with pronominal elements, 
 we cannot doubt that they must themselves belong to the formative 
 element of inflected language. But it is not so easy to determine in 
 which of the pronominal roots we are to seek the common origin of 
 these particles. At one time we were disposed to connect them with 
 the dental degenerations of the second element; and the principles 
 of etymology are not opposed to this conclusion; nevertheless, after 
 sifting all the evidence, and balancing one induction against the 
 other, we have come, with full conviction, to the result, that the 
 elements λ- and ρ- are by-forms of the third pronoun, and imme- 
 diately derived from the other dental liquid v. The following are 
 the leading proofs. 
 
 * TheProfessorof Comparative Philology, to whom we have more than once 
 alluded, in his anxiety to gain a character for originality, has ventured ^.o propose 
 that castrum means "an axe," and that itis derived from cac?o, which, borrowing 
 a principle from this book, he connects with ccedo {Proceed, of the Phil. Soc. ii. 
 pp. 249 sqq.). His grand reason for this conclusion is, "that the suffix -irwrn 
 denotes always an instrument/' Now to say nothing of plaustrum ana clmistrum, 
 which he quotes, are antrum, theatrum, monstrum, fefiestra, veratrum, and many 
 like words, properly designated as instruments? It is palpably absurd to press 
 the phrases movere castra, ponere castra with this view. No one ever felt any in- 
 consistency, when it was still left for despairing ingenuity to convert a camp 
 into a tool-chest. Besides, castrum occurs in the singular, and who would think 
 of translating castellum, " a little chopper" ? An absurd etymology provokes our 
 mirth; but those who are anxious that scientific grammar should take root in 
 this country will regret to find such crude puerilities recorded in the annals of a 
 learned society. 
 
CHAP. III.] OF THE UNINFLECTED FORMS. 475 
 
 Although dei-Xog seems to stand in a certain opposition to δεί-νός, 
 these words are easily reconcileable according to the principle of 
 contrast, which cannot depend upon the termination, for 8κΛθίγ-λος= 
 ϊκπλαγ-λοξ corresponds in meaning to δεί-νός; and they are brought 
 together in the Latin di-rus, which answers to δεί-νόξ, just as con- 
 versely mag-nus does to μεγά-λο-. Then, in the Greek language itself 
 we may compare oQya-vov with %QOra-lov, ΰεΐ6-τρον; ορφ-νός and 
 όρφα-νόζ with τvφ-λog; λνγ -QOg with ύτυγ -vog; εχ^'-Q6g with [ε]|ε- 
 vog ; xa-Xog and xad'a-Qog with xaL-vog; ohv-Qog with ελεεL•-vόg', and 
 many others, which clearly show that the terminations -log, -Qog, -vog, 
 agree in expressing objective relations, or in denoting that an object 
 presents itself to our observation a shaving a certain capability or use. 
 To these special instances we must add the general fact that the use 
 of ρ- as an affix indicating motion or conveying the idea of " beyond " 
 (§ 130) is in strict accordance with the use of -v, to mark the accusa- 
 tive, as the case of motion towards an object. 
 
 It will perhaps be regarded as an important confirmation of this 
 view, that the only Latin and Greek pronouns which exhibit the 
 liquid i, λ — namely, ille or oUus and alius, aXXog — are manifestly 
 derived from ava^=Fa-va (§ 116), so that κεΐ -vog and ille correspond 
 not only in meaning (§ 135), but, ultimately, in form. We have a 
 slighter evidence of the same kind in the use of the liquid b by the 
 Hebrews to express the most emphatic employment of the Indo- 
 Germanic pronoun n- (above, § 184). 
 
 269 But we rest our demonstration of the identity of the prono- 
 minal elements λ, ρ, ν chiefly on the extensive and essential corre- 
 spondences of their use as verbal roots. And, first, with regard to 
 the identity of λ- and ρ- as they appear in verbs. 
 
 The intensive particle ρα, which belongs to this family, seems to 
 convey the idea of facility, easy motion, and so forth. We have ac- 
 cordingly recognised its connexion with ρε-ευν, ρά-δίog, &c. Now 
 there are two Sanscrit verbal roots with the same meaning, ri and 
 m, both signifying "to go." We do not conceive that the sibilant 
 prefixed to the second interferes with its relationship to the first. 
 The present of sri is sarami=^adeo aliquem. This word is of course 
 related to deserere, salire. We consider too that conserere and con- 
 sulere are the same word. "Without doubt," says Niebuhr {Hist, of 
 Bome^ I. p. 512), "the name consides means nothing more than 
 simply colleagues: the syllable sul is found in prcesul and exsul, where 
 it signifies one who is : thus consules is tantamount to consentes, the 
 name given to Jupiter's council of gods." This is not altogether 
 
476 THE PRONOMINAL TERMINATIONS [bOOK III. 
 
 accurate*: the word consentes means "tliose who are together" 
 (compare ab-sentes, prce-sentes): consules, "those who go together," 
 prcesul, "he who goes before," exsul^ "he who goes out." That the 
 Romans habitually spoke of "going," where we should rather in- 
 dicate "being," is sufficiently proved by the words in it-, as paries, 
 aries, miles, pedes, eques, &c. If sa-li-re, se-re-re are the same word, 
 li and re must be the same root, and therefore lev-is and rap-idus are 
 connected. The former contains the root AstF, which we shall show 
 in a future chapter in all its various uses. It signifies both "to see" 
 and "to take;" we have the former meaning, e. g. in ^-ρ(2)κω; the 
 latter in rap-io, rap-idus, and lev-is, lev-are. We may also compare 
 Sanscrit vrtkas (Sabine hirpus=-virpus, or vripus or irpus, Lithuan. 
 wilkas, Latin lupus, Goth, vulfs) with the Greek Kv-%og, the connexion 
 of which with λευ-κο5=λεΡκο^, λν^,η, lux, λνκάβας, and the root keF, 
 "to see," is well known. Thus the old difficulty about Apollo's 
 epithet kvxsLog vanishes at once. 
 
 270 With respect to the identity of ρ- and v- in this reference, 
 we will take the liberty of repeating here what we have incidentally 
 stated on a former occasion, when we endeavoured to show that the 
 ideas of progressive time and recurrence are connected with our 
 conceptions of the regular flowing of water. As the cycles of hours 
 and seasons are always recommencing, it is plain that our first notions 
 of progression in time must be nearly allied to, if not identical with, 
 those of recurrence ; and there is no object presented to the senses 
 which is more likely to suggest the idea of the course of time, than 
 the noiseless, but unceasing flow of the running stream. Not to 
 enter upon any metaphysical discussion of this point, etymology 
 renders it sufficiently plain, that the words which imply "flowing," 
 "progression," and "recurring," are connected with the same ele- 
 mental root. 
 
 The particle, which, in the Latin language, expresses return, 
 reversion, and recurrence, is that which forms the first syllable of 
 these very words — namely, the prefix re. In the Greek language, 
 the functions of this particle are performed by the preposition ά-νά, 
 while the particle and affix -ρα, -ρ denotes, as we have seen above, 
 motion in a direction previously indicated. Now the idea of motion 
 in a direct line is the idea of perpetual recurrence, for the line is a 
 series of points with evanescent intervals; and this again is the idea 
 of progressive time. For our purpose, therefore, it only remains to 
 
 * Lobeck too has subsequently expressed his dissent from this remark of 
 Niebuhr's (Paralipom. p. 128, note 16). 
 
CHAP. ΠΙ.] OF THE UNINFLECTEB FORMS. 477 
 
 show that in their employment as verbal roots, λ-, ρ-, ν- are indif- 
 ferently used to express these connected ideas. 
 
 It will not be denied that while νεομαι, νίΰΰομαι, νόΰτος, &c. 
 express return and recurrence, and while veog implies change, which 
 is included in the idea of motion (above, § 55), the words νεω, vavg, 
 νάμα, &c. convey the meaning of "being in the water, being borne 
 along the stream," &c. Now a very similar conception is expressed 
 by the cognate roots ρεω, ρενμα, Λ-λέω, τί-λν-νω, λονω ; rapio, p-luo, 
 f-luo, lavo, &c.; compare '/^;^ε-λωο5 with aqua lavans^ fluens. We 
 should naturally expect to find these roots in the Indo-Germanic 
 term for "a year" — a period which includes all the changes of the 
 seasons, which is always progressive, yet always recommencing; always 
 changing, but always resuming its identity. Now in the ancient 
 Etruscan, which we believe to have been pure Pelasgian in its calen- 
 dar, we find the word ri-l signifying "a year." As the termination 
 corresponds to the patronymic I- in servi-Uus, fi-lius, which in Greek 
 is 8l- or d-, as in Άτρεί-δης, νΐ-δίον, we may compare ri-l with ^εΓ- 
 ^QOVy which denotes the motion of water. Similarly, the Latin annus, 
 more anciently dnus^must denote at once "the ever-flowing " (αε-ναο^), 
 and "the ever-returning" {άεΐ νεόμενος). Accordingly, anus ^ anus 
 stands on the same footing as d-ril, who seems to have been the God 
 of the Tuscan year. 
 
 It will be admitted, we trust, that the result of this investigation 
 is to identify the element λ-=ρ- with the third pronominal root v-. 
 And if any one seeks to undervalue the importance of these inquiries, 
 he may be told that these combinations have enabled us to supply the 
 only link wanted to complete the chain of evidence, which proves the 
 wonderful and systematic perfection of the formative contrivances of 
 inflected language. 
 
CHAPTER IV. 
 NOUNS USED AS PREPOSITIONS. 
 
 271 (1)''Ένεκα. Its true meaning suggested by its apparently pleonastic use. 
 272 Connexion of iVfxa=£i/f>£a and δκατί. 273 Words containing fxa- ; their 
 cognate meanings. 274 Ideas of separation and will meet in that of unity. 
 21b Analysis of s-yia. 276 Proper names which include these syllables. 
 277 Compounds of gWxojand a pronoun. JStVfxa should be written for owfxa, 
 when the latter appears as a preposition. 278 (2) Χάριν. Distinctive use of 
 χάριν and ένεκα. 279 Meaning οΐτίρος χάριν. 280 Examination of the class 
 of words to which χάρις belongs. Χώρος, χορός, Άπάχώρκ. 281 Χερΰος, 
 χοιράς, χειρ, Sic. 2S2 Χάροιρ,χάρνβδις,Άηάχάρων. 2SZ Χάρων and Γ7]ρνων. 
 284 Military applications of the words χάρμη, χρεία, &c. 285 The same idea 
 conveyed by άρης, τιρα, &c. 286 A similar reference discovered in the primi- 
 tive meaning oi χάρα^. 287 Associations by contrast in the acceptations of 
 the root χαρ-. 288 More doubtful affinities of χάρις. 289 (3) ^Urjv. Its 
 prepositional use. 290 Meaning of δίκη. 291 Connected applications of;^άρίS 
 and δίηη. "The Graces" and "fair dealing." 292 This is supported by the 
 etymology of δίκη. 
 
 271 "DY the side of the regular prepositions, which, as we 
 have seen, are reducible to the simplest pronominal 
 elements, the fixed inflexions or adverbial forms of many nouns 
 and pronouns are used syntactically as prepositions, that is, they 
 are employed in connexion with some case, usually the genitive, 
 or, in common language, they govern that case. Thus we have 
 αγχυ and Ιγγνς, εγγνΟ'ί, *' near," one evidently the dative of αγξ^ 
 the shortened form of αγκάλη, the other exhibiting 8V prefixed to 
 the dative of a word synonymous with yvi^. The idea of separation 
 is expressed by ατερ=^αντερ, by χωρίς, the fragmentary dative of 
 χώρα, and by τηλε, τηλον, τηλό'^ι, and τηλό^'εν, different cases of 
 a word denoting growth and extension (below, § 344). Among 
 the cases of nouns, which contribute in this way to the definite 
 syntax of the Greek language, there are three words which 
 deserve special notice for etymological considerations. These 
 words are (1) ένεκα or εκατί, (2) χάριν ^ and (3) δίκην. As 
 these quasi-prepositions have a sort of connexion with one ano- 
 ther, and as the first two belong, each of them, to an extensive 
 family of words which has not been sufficiently explained, we 
 shall devote a separate chapter to their consideration. 
 
 (1) It is generally laid down that ένεκα signifies "on account 
 pf," "for the sake of;" but it is proper to state that tlie genitive case, 
 
CHAP. IV.] NOUNS USED AS PREPOSITIONS. 479 
 
 with which ένεκα is generally found, may stand alone with the same 
 signification, as when Thucydides says (i. 4) that Minos cleared the 
 ^gean sea of pirates as far as he could, τον Tag JtQOoodovg μάλλον 
 Ιεναυ αντω, and also that the genitive case may be accompanied by 
 some additional preposition conveying a similar meaning, or by χάρνν : 
 as will appear from the following passages ; Sophocles, PAi/ociet 554; 
 a τοΐόίν'^ργείοίόίν άμφϊ ΰον'νεκα 
 βονλενματ εΰτί. 
 Thucydides, νιιι. 92 : και 6 μεν Θηραμένηζ ελ^ών ες τον Πειραιά... 
 ο6ον και άτίο βοής ένεκα ώργίζετο τοις οπλίταις' 6 δε Άρί(5ταρχος 
 καΐ οι ενάντιοι τω τΐλη&ει (read τα άλήΟ-ει) εχαλετίαινον. Xenophon, 
 Hellenic. U. 4:, § 31: τίεμτίων δε τίρεόβεις 6 Uavoaviag προς τονς εν 
 Πειραιεΐ εκελ^νεν άΐίίεναι ετά τα εαυτών επεί δ' ονκ ετΐεί^οντο., 
 τίροΰεβαλλεν οΰον aito βοής ένεκεν, οτΐως μη δήλος εϊη ευμενής αυ- 
 τοίς ων. Lysias de Evandri probatione, p. 176: υ ^εις τον τΐερι των 
 δοκιμασιών νόμον [ονχ ηκιΰτα] τΐερΙ τών εν ολιγαρχία άρξάντων 
 ένεκεν ε^ηκεν. Plato, Pc»/^ί^cw5, ρ. 302 β: ου μην αλλ' εϊς γε το όλον 
 ϊΰως απαν%^ ένεκα του τοιούτου τίάντες δρώμεν χάριν. Legg. ιιι. 
 ρ. 701 D: άλλ' ετίανερωταν το νυν δη λεχ^εν, το τίνος δη χάριν 
 ένεκα ταύτα ελέχθη. In Aristophanes, Thesmo'ph. 372^ 
 η Μηδους ετίάγουβι της 
 χώρας οϋνεκ [leg. εΐνεκ] ετά βλάβη^ 
 it is probable that the last three words are a mere repetition of those 
 inv.367: κεράών οννεκ' [εΐνεκ] επι βλάβη. We think, however, that 
 it is unnecessary to place them between brackets, as Dindorf has 
 done. In the other passages it is easy to show that ένεκα is neither 
 superfluous nor insignificant. The phrase οΰον άτίο βοής ένεκα, used 
 by Thucydides and Xenophon, is probably a military expression: for 
 a Greek battle generally began with a shout, and if the parties did 
 not go farther than that, it was of course only a sham-fight. And 
 thus Xenophon says that Pausanias attacked the Peiraeus merely so 
 far as shouting went ; he made a false attack : and Thucydides states 
 that whereas Aristarchus and the young oligarchs who accompanied 
 him were sincerely indignant (we read either τω άλη%'ει or τω 7ίλη%ει 
 τω άλή^ει, the latter having been merged in the former from its simi- 
 larity of appearance), Theramenes only afiPected to condemn the con- 
 duct of the soldiers, — οΰον ά%ο βοής ένεκα ώργίξετο, he showed his 
 anger only so far as making an outcry went. In these two passages, 
 then, ένεκα clearly means "only." In all the other instances of al- 
 leged pleonasm the signification obviously is "especially," "in par- 
 ticular." Indeed, it is probable that, in the passage of Lysias, we 
 ought to bracket, not ττερί, as Bekker has done, but ούχ ηκιύτα, which 
 seems to be a gloss upon ένεκεν. The etymological analysis, which we 
 
480 NOUNS USED AS PBEPOSITIONS. [bOOK ΠΙ. 
 
 sliall now attempt, will show us that both these adverbial meanings, 
 "only" and "especially," are included among the primitive significa- 
 tions of ένεκα. 
 
 272 The relationship between ένεκα and εκατι or εκητο, as it is 
 written in the Ionic dialect, is the same as that which subsists be- 
 tween the Italian synonyms in fuori and fuori, which are used indif- 
 ferently as prepositions signifying "without." '"ϊί'κατί'} the older word, 
 has the complete case-ending, and is used without the preposition εν, 
 which supplied the place of the «locative in the more recent language ; 
 ένεκα contains the preposition εν prefixed to a mutilated locative of 
 εκας. The formation of ενεκα=εν εκα is perfectly analogous to that 
 of εναντα=εν αντα and εμ7ία=^εν παβιν. In ένεκα the aspirate of 
 the noun has been transferred to the beginning of the word, according 
 to a principle mentioned before, of which the Greek language fur- 
 nishes many examples: thus ό Ttqo odov makes φρούδος, and ό JtQO 
 ορών, φρουρός. Indeed, it would be absolutely necessary that when 
 εν εκα became a single word, the aspirate should be so transposed, for 
 the analogies of Hellenism do not permit an aspirated vowel any 
 where but in the first syllable, and it is remarked as a strange pecu- 
 liarity of the Athenians that they said ταως instead οΐταώς — φίλαρ- 
 χος γαρ ov6a καΐ ηγεμονική την φνύιν η δαΰντης τοις τελευταίους 
 μερεόυ των ονομάτων ουδαμώς εγκα^'είργνυταυ (Athen8eus,p.397F). 
 We find a similar transfer even in the case of aspirated consonants : 
 as in τρέφω, ^'ρε^ρω; Βχω, εζω, &c. From these instances, and from 
 mere erases like ^οίμάτιον for το ιμάτυον, we must distinguish (a) 
 those in which it appears uncertain whether there is a transfer of the 
 aspirate or not, as in τε^'ρυτΐτίον from τετ ρ a- and ϊτίτΐος•, (b) those in 
 which the aspiration seems to result merely from a contact with the 
 ρ, as in Ο^ράΰύω for ταράόΰω, and (c) those words (like θόρυβος 
 compared with τύρβη, ^-ρεομαί with τ ρέω, and ^'ρυΛτω and %'ραύω 
 with τρυφή and τρι;ω) in which the aspirate seems to result merely 
 from a kind of vacillation and uncertainty of use (see above §§. 100, 
 164). "Εκα as a mutilated though old form of the dative or locative 
 may be compared with λίτΐα in the phrase χρίειν λίπ ελαίω, where 
 ελαωςίΒ a regular adjective from ελαα, αηάλίτίά ελαίον signifies "olive- 
 oil" (Buttmann, Ausf. Sprl. Vol. i. p. ^29); κέρα for κέρατο gives us 
 the intermediate stage. But we have the proper ending of the locative 
 in the form ένεκεν, from which ένεκα is derived (above, § 114), and 
 which is often used even in the more recent Attic writers: compare 
 κα, κεν, IW«, ενΟ'εν; ετίειτα, επειτεν; ττρόό&α, Λρόΰ&εν, &c. 
 
 273 The element εκα, which is the essential part of the synonyms 
 εκατι and ένεκα, occurs also in the following words: έκάβη, έκάς, 
 
CHAP. IV.] NOUNS USED AS PREPOSITIONS. 481 
 
 εκα^^εν, εκάλη, εκαμήδη^ εκαΰτος, εκάτερος, εκάτη, εκατος, εκηλος, and 
 εκών, besides a great number of derivatives, as Ικα-ερ^ο^, εκατί^-|3όλο^, 
 εκψβόλος) &c. To classify these words we will first set apart the 
 proper names 'Εκάβη, Εκάλη, and Έκαμήδη. The remaining words 
 are a substantive εκας, genit. εκαΟ'εν, dat. εκα or εκενίοηηά in ενεκ[^.,). 
 with which are connected the two adjectives εκα-τος (fern, εκάτη), and 
 εκών (εκό-ντ -g) ; the comparative and superlative δζά-τερο^,ίκαί) -TOg; 
 and the dative εκητί of a substantive εκης {εκ-ι/α-χ-ξ) no longer in ex- 
 istence, by the side of which we have the adjective εκηλοξ. Such is 
 obviously the proper grammatical classification of this set of words, so 
 far as regards the forms. We must now investigate their significations. 
 Έκάς, whi#h is used as an adverb, denotes distance, whether in space 
 or time; as κα(5τ ονχ εκάς Jtov (Sophocl. Philoct. 41), "he is not far 
 off;" ουκ εκας χρόνου τίάρεΰται (Herod, νπι. 154), "he will be here at 
 no distant period." The word belongs to the oldest state of the language. 
 A grammarian under the head ποΐαι, γλώοΰαο κατά τίόλεΐξ remarks, 
 θεύύαλών — εκάζ. πόρρω {Bekkeri Anecd. p. 1095, note), which is much 
 the same as calling it a Pelasgian word (Niebuhr, Hist, of Borne, i. p. 30, 
 note 69). "Εκαϋ'εν generally means "from or of that which is distant : " 
 it may be used as a synonym for εκαζ (Homer, Odyss. xvii. 25: εκα^'εν 
 δε τε αύτν φάτ είναή, as the genitive of other words is also used to 
 denote locality. The idea of distance is also conveyed \>ΥΈκατοξ,' Εκάτη, 
 which are epithets of the Sun and Moon, the two distant powers {Theatre 
 of the Greeks, 6th edition, p. [23]), and by the epithets εκά-εργος, 
 εκατη-βόλος especially applied to them. That the numeral εκατόν has 
 no real connexion with these words may be seen by the accentuation ; 
 and it has been shown before that it belongs to entirely a different class 
 (§ 162). The idea of distance is exchanged for that of separation in the 
 words εκά-τερος, "either one of two;" εκαΰ-το£, "each one by himself 
 out of many;" the latter is analogous to όλιγοβ-τόξ , τΐολλοΰ-τός , of 
 which we have spoken before. A comparison of εκά-τερος with άμφό- 
 tερog shows how completely the former represents the sense of unity 
 conveyed by the Sanscrit e-kas. Plato defines the two comparatives by 
 saying (ΤΛβοΒίβί. 185 β): ort άμφοτέρω δύο, εκάτερον δε εν. The gram- 
 marians understood the distinction. Thus Maximus Planudes says (apud 
 Bachmann. Anecd. u. p. 88, 1. 25): εκάτεροί καϊ αμφότεροι, επ:ί< δύο, 
 εκαΰτοί καϊ πάντες επΙ πολλών, διαφορά δε οτι το μεν εκάτεροο 
 και εκαΰτοο επί των διηρημενως ενεργούντων τι νοείται, το 
 δε αμφότεροι καϊ πάντες επί των ΰννημμένως. And this dis- 
 tinction is illustrated by another grammarian (Bachmann. Anecd. ii. 
 p. 376, 1. 10): αμφότεροι δοκον φερονβιν, ήγουν ενα δοκόν 
 εκάτεροί δοκον φέρονΰιν, ήγουν δύο δοκούς, χωρϊς εκαΰτος. 
 This idea of separation and unity is clearly exhibited in the phrase 
 
 II 
 
482 NOUNS USED AS PREPOSITIONS. [bOOK III. 
 
 6s εκαβτοί, "severally," as in Tliucyd. i. 3: οΓ ως εκαότοο^Ελληνες^ 
 "the Hellenes, taken severally or as separate tribes," where it is op- 
 posed to ξνμτίαντες. So also in Thucyd. i. 36, we have τοις τε ξνμ- 
 TtauL καΐ καθ•' εκαύτον, and in Aristotle, Politics ^ m. 15, § 16, we 
 read εκάΰτον καΐ ενός καΐ ΰνμτιλεωνων κρείττω. We have a still 
 further modij&cation in εκών, εκηλοξ, which imply, acting by oneself, 
 without interference on the part of any other person, acting according 
 to one's own will and pleasure, doing anything of one's own accord, 
 without consulting or being influenced by any other person. Of the 
 two adjectives εκών refers rather to the free-will of the agent, εκηλος 
 to his freedom from disturbance, and his consequent pleasure and 
 quiet. Yet both of them equally convey the idea that the person des- 
 cribed is some one by himself, that is, considered without reference to 
 any one else. Hence εκών is particularly opposed to βία, as in So- 
 phocles, (Ed. Col. 939: βίατε κονχ εκών, and it is used in speaking 
 of errors which men have committed with a full knowledge of the 
 consequences, as when Prometheus says (^schyl. Prom. 265): 
 
 εγώ δε tavxt' aitavz ητΐίοτάμην 
 εκών, εκών -ζ'^ιαρτον ουκ άρνήΰομαί. 
 
 In the phrase εκών είναι, this adjective points still more directly to 
 the independent agency, as in Herodotus, vii. 164: εκών τε είναι κ«1 
 δεινού ετίΐόντος ονδενός; so that the phrase is nearly equivalent to 
 the Atticism το ετά ^φαξ είναι (Thucyd. ιν. 28). "Εκηλος generally 
 implies uninterrupted rest and quietness, the results of non-inter- 
 ference from without. We have already explained the principle ac- 
 cording to which ενκηλοζ is only a by-form of εκηλοξ (above, § 116): 
 it is expressly stated by ApoUonius (Bekkeri Anecd. p. 558) that 
 ενκτιλοζ is related to εκηλος in the same way as εντε to οτε. Homer 
 invariably uses εκητι in connexion with the name of some divinity, to 
 express that the action in question has been effected by the aid or 
 special favour of the protecting power. Thus Odyss. xx. 42, Ulysses, 
 addressing Minerva, asks: 
 
 εϊπερ γαρ κτείναιμι, ^i6g τε ΰε%'εν τε εκψι, 
 
 7ty κεν νΛεκτίροφνγοιμι; 
 
 It is used in this sense by Pindar, e.g.Pyth.Y. 9: εκατι χρυύαρμάτον 
 Κάβτοροξ, or as expressing human agency, as in the old proverb: 
 εκητυ Σνλοΰώντοζ ενρνχωρίη , "room enough, thanks to Syloson" 
 (Strabo, p. 638); and finally, which is much the more common usage, 
 as a synonym for ένεκα, whether as signifying "by means of," "on 
 account of," "for the sake of" (propter), as in Pindar, Pyth. x. 58: 
 εκατι ΰτεφάνων ^'αητον εν αλιξι &η6εμεν εν και τίαλαιτεροις, or with 
 the meaning " as far as," "in regard to" (quoad), as in ^schylus, Pers. 
 
CHAP. IV.] NOUNS USED AS PREPOSITIONS. 483 
 
 343 : τΐλήϋΌνς μεν αν βάφ' Ϊ6^^ ξ%ατι βαρβάρονζ νανβΐν κρατηβαι 
 
 — "if it had depended only on the number of their ships, the Persians 
 would no doubt have gained the victory." 
 
 274 We have now considered all the simple words into which 
 εκα- enters, and it appears that there are only three meanings, distance 
 or separation, will, and unity. It is, we conceive, easy to show that 
 these are only modifications of one and the same idea. And first, the 
 ideas of separation, distance, and unity are identical. That which is 
 absolutely distant or separated is said to be alone, i. e. all one: and 
 thus the first numeral is expressed either by the pronominal element 
 maj denoting the subject as opposed to the object, or by some pronoun 
 e-na (unus), or e-ka, which combines the idea of definite locality with 
 that of distance. Again, the idea of will is immediately derived 
 from that of self. A man's personality, individuality, or character, 
 depends upon his will, as Schiller says {Wallensteins Tod, iv. 8): 
 "den Menschen macht sein Wille gross und klein," or rather the will 
 and the character are the same, for, as Novalis remarks: "a character 
 is a completely fashioned will" (ein Character ist ein voUkommen 
 gebildeter Wille, ii. p. 284). Now the very idea of a distinct person 
 or individual is that he cannot be divided, that he is an unit; hence 
 Paschasius rightly asserts — in Deo et homine, gemina quidem substan- 
 tia, sed non gemina persona est, quia persona personam consumers 
 potest, substantia vero substantiam consumere non potest (quoted by 
 Hooker, Vol. n. p. 288 Keble). So that the meanings "separation" 
 or "distance," conveyed by sxag, εκατος, and "will," borne by εκών 
 and εκ7ΐλθ£, as well as the sense "only" or "especially," which we 
 have extracted from ένεκα in the cases of alleged pleonasm quoted 
 above, are all derivable from that of "unity," which is the meaning 
 of εκάτερος and εκαβτοξ. "Ενεκα and εκατι, in the ordinary use, bear 
 all these modified but nearly connected significations. 
 
 275 Comparative grammar shows us that the element ε-κα- is, in 
 fact, identical with that compound of two pronominal stems which 
 forms the first Sanscrit numeral e-ka-s, and the Latin adjective 
 cE-quu-s , denoting unity or sameness. The identity of ε and e is esta- 
 blished by a comparison of devas with \Ιτεόξ, &c., and by the frequent 
 metathesis of the breathings and semivowels (above, § 116). It hap- 
 pens in some cases that a class of words containing a common element 
 preserve their external resemblance more completely in Greek than in 
 Sanscrit, although the changes which the root has undergone may be 
 greater in the former than in the latter language. This is one of 
 those cases: for, although the element εκα- originally began with an 
 
 112 
 
484 NOUNS USED AS PREPOSITIONS. [bOOK III. 
 
 aspirated labial, as appears from the fact that all words containing 
 this root are diganunated, and from the by-form ενκηλος, they have in 
 every other instance been consistent in the change to the aspirate , or 
 in the suppression of the labial element ; whereas the Sanscrit has 
 merely transposed the guttural semivowel in the numeral e-Jcas = ai- 
 kas, and has retained the labial alone in νας -mi, "I will," vagas^ 
 "will or power ," a-vagas^ "unwilling," d-vagyam^ "by compulsion." 
 The ablative vagdt of vagas is used to signify "on account of," just 
 like Βκατι. After what has been said about the semivowels in a for- 
 mer chapter, we shall have little difficulty in admitting the relation- 
 ship of these Sanscrit words denoting "will" to the first numeral in 
 that language. This presumes, however, that the first part of the 
 compound now before us is the element Ρα=Λϋα, denoting relative 
 nearness. The same must be the case with the noun aval•,, anciently 
 pronounced ^άναξ, which, as we shall see, is derived from fa-va, the 
 older and more correct form of ά-νά. In ε-γώ , ay αν, aham, which 
 seem to be connected in meaning with the first Sanscrit numeral, the 
 first syllable is always found without any aspirate. That the aspirate 
 or ultimate guttural, however, was really an essential part of the first 
 syllable of this element, appears from the Latin secus, which is clearly 
 the representative of εκας. We are obliged to infer, then, that although 
 ε-γώ^ α-γα-ν, α -ham ^ &c., seem to be connected in meaning with the 
 most emphatic demonstrative, for they imply distance or separation, 
 and though ε%αξ^ &c., bear the same or a very similar sense, all these 
 words are related, in their first syllable as well as in their second, to 
 the second pronominal element, and signify "this which;" for, although 
 a may come from va as well as from Pa, it is impossible that the 
 different elements ε-, ai-, se-, w-, να, ευ, should have any other common 
 origin than the element J^a. And thus the simplest demonstrative 
 root a, though in all its appearances il seems to correspond in value 
 to the elements of the first or third person, must in this instance be 
 connected in origin with the element of the second; the idea of 
 proximity to, having merged in that of identity with, the subject, as 
 in the Italian ci mentioned above (§ 150). The etymological fact is 
 certain; the explanation depends upon the exclusively demonstrative 
 nature of the original pronouns. There is, in fact, no reason Avhy 
 the ideas of separation, distance, and unity, should not be expressed 
 by the combination signifying "this which," as well as by one de- 
 noting "that which;" and we have seen other instances of this 
 reduplication (§ 133). 
 
 276 It will perhaps be as well to explain the three proper names 
 into which the element εκα enters. 'Εκάβη means either an only child, 
 
CHAP. IV.] NOUNS USED AS PBEPOSITIONS. 485 
 
 or one born among the last of her father's family ; in either case it is 
 a title of endearment. With regard to the first part, it may be com- 
 pared with the Sanscrit eka-ja =: qui solus natus est (Bopp , Gloss. 
 Sanscr. p. 58): its termination seems to be analogous to that of 
 λνκά-βας. The name 'Εκάλη was borne by a mythical old woman 
 who was very kind to Theseus in his childhood, and as a by-form 
 of 8K7]Xog expresses her good nature. This appears from the words of 
 Plutarch (in vita Thesei, cxiv.) : την Έκάλην Ιτίμων^ Έκαλίνην ντίο- 
 κορίζόμενοί, dik το κάκείνην νέον οντά κομιδη τον Θηΰέα ξενίξον- 
 (5αν aoTtduaOd'ai τίρεΰβντίκώς και φίλοφρονεΐό%'αί τοιοντοος ντΐο- 
 κορυΰμοΐς. Έκα-μηδη is the name of an active and willing female 
 servant in Homer (Iliad xi. 623), and may be compared with Περο- 
 μηδη£, and with Γανν-μήδης, the name of a heavenly menial. 
 
 277 ΎΙιβΐονταΒείνεκα,τοννεκα,οννεκα, and οΰΌννεκα aXso require 
 some remark. In the first, the preposition tv appears in the stronger 
 form είν (above, § 1 70) , which is used by the Attic writers , not only 
 by itself, as in Sophocl. Antig. 1226 : είν "^ίδον δόμοίς; ^schyl. Suppl. 
 872: αραίαΐξ εΙν ανραΐξ (according to Lbbeck's emendation; see, how- 
 evnr, below, § 475) ; but also in composition, as in Sophocl. Antig. 346 : 
 Λοντον τ είναλίαν φνόιν; Sophocl. /r. 480: της είνοδ lag 'Εκάτης 
 εγχος. This form of ένεκα occurs in the Attic prose-writers, with the ex- 
 ception of Thucydides , as Thomas Magister tells us: καΐ ένεκα και 
 είν εκ a, Πλάτων, ^ημοβ^ενης, καΐ οί άλλοι. Θουκυδίδης δε άεϊ 
 ένεκα. Of its use in Plato, the following instances may suffice: Legg. 
 p. 778 D, p. 916 A, p. 949 d. It has been remarked by F. A. Wolf (ad 
 Demosth. Leptin. p. 338) that the form εΐνεκα never occurs in the Attic 
 poets, but that they always write ένεκα or οϋνεκανίϋΐα. the same signifi- 
 cation. That is to say, when εΐνεκα occurs, there are almost always va- 
 rious readings, whereas we often find οϋνεκα without any variation in 
 the MSS. It seems, however, quite impossible to suppose that οϋνεκα 
 could be used as a mere preposition. It stands precisely on the same foot- 
 ing with τοϋνεκα and bQ^oύvεκa,which. are compounds of του and ότου 
 with δνεκοί, as οϋνεκα is of ου and ένεκα. Accordingly, the first should 
 signify "on this account," the other two, "on which account;" τοϋνεκα 
 also =τίνος ένεκα, "why?" (Steph. Thes. r.p. 1204g). It would be bet- 
 ter, perhaps, to write τοϋνεκα, on the analogy of the other two words. 
 Matthia (Gr. Gr. § 624, 2 obs.) adopts the old derivation of ο&ούνεκα 
 from οτυ and οϋνεκα, objecting to the obvious etymology from οτον and 
 ένεκα, given by Lobeck and Buttmann, that if this had been the case, 
 it should have been written οτούνεκα, like τοϋνεκα. But τοϋνεκα is 
 purely Ionic, and the lonians did not throw the aspirate forwards, 
 for they wrote ουκ έκάς, ε% οΰον, κατάτΐερ, and so forth, whereas the 
 
486 NOUNS USED AS PREPOSITIONS. [bOOK III. 
 
 Attics would write 6&οννεκα just as they wrote Ο'ώτίλα for τα όπλα, 
 &7]μέρα for rfj ήμερα, &c. Reisig in his exposition of the (Edipus 
 Coloneus (p. cxxviii) advances an opinion still more untenable, for he 
 considers ο%Όννε%α as a compound of oO't and οννεκα, and translates 
 it ubi id est, cajus causa quidque fit. The proper use of οννεκα and 
 οΰΌννεκα is, as conjunctions, nearly equivalent to ore, and signifying 
 "that" or "because," a meaning which they seem to have obtained 
 by a kind of attraction or brachylogy, like their synonym άνΰ'^ ών: 
 Sophocl. Antig. 1050: 
 
 άν%^ ών έχεις μεν των ανω βαλών κάτω. 
 
 The fuller form may be surmised from a former line of the same play 
 (237): 
 
 τι d' 'εΰην, «ι^θ' ov τ^νδ^ έχεις α^νμίαν; 
 
 that is to say, av%'^ ov stands for άντΙ τούτον οτυ, just as οννεκα is 
 put for τοντον ένεκα, οτι. Ammonius has given the distinctions be- 
 tween οννεκα and εϊνεκα correctly enough. He says: οννεκα καΐ 
 εΐνεκα διαφέρει, οννεκα μεν (ji^fitxtVat το οτι. εϊνεκα δε χάριν. 
 "We agree, therefore, with Ahlwardt (Zweiter Beytr. zu Schneiders 
 Worterb. 1813) in ^thinking, that, as the MSS. in many cases, and 
 common sense in all, authorise the change, we should substitute εϊνεκα, 
 which is acknowledged to be good in Attic prose, for οννεκα, when- 
 ever it stands for ένεκα in Attic verse. Conversely, we have proposed 
 to read οννεκεν for εϊνεκεν in Pindar, Isthm. vii. [vm.] , 33 , because 
 we do not believe that ένεκα can be a conjunction, any more than we 
 think that the compound οννεκα can perform the functions of a mere 
 preposition. 
 
 278 (2) The difference between ένεκα and χάριν, in their use 
 as prepositions, has been correctly stated by Ammonius: "Ενεκα καΐ 
 Χάριν διαφέρει' δ μίν γαρ^Ενεκα ι^ιλην την αίτίαν δηλοΐ, οίον — 
 ενεκα'^λε^άνδρον και ενεκα^Ελένης ε6τράτεν(5ε Μενέλαος' 6 δε Χά- 
 ριν μετά της αίτιας δηλοΐ και την χάριν — χάριν Μενελάον^Αχιλλενς 
 εΰτράτενΰε. τοντεΰι, Μενελάω χαριξόμενος. In other words, the 
 strength of the motive only is implied in ένεκα, while in χάριν we are 
 told the action is intended to please some one, or to benefit him in some 
 way; and, to express the distinction with reference to the original mean- 
 ing of the two words , ένεκα or εκατι states that the action has taken 
 place according to the will of a certain person, or λνϋΐι particular or 
 exclusive reference to some person, thing, or action; whereas χάριν 
 always indicates that the action is done to promote some thing or 
 other, or to please or benefit some person ; just as the Germans would 
 say urn meinet Willen in the one case, and mir zu gef alien in the 
 
CHAP. IV.] NOUNS USED AS PEEPOSITIONS. 487 
 
 other. This distinction is obviously preserved in the following line 
 of Euripides {Helen. 1254): 
 
 πλούτου λίγ' οννεχ (leg. εΐνεχ), οτι %Βλεΐζ ταύτης χάριν 
 " with regard to wealth in particular, as far as wealth or expenditure 
 is concerned , say what you would have , for the gratification of this 
 lady." Something of this kind seems to be intended in the opposi- 
 tion of χάριν to evsTCEV, which we find in Aristotle, Polit. i. 8, § 11: 
 ωότε ομοίως δηλον οτι και γενομενοις οίψεον τά τε φυτά των ζώων 
 ένεκεν είναι και τ άλλα ζώα των άν&ρώτίων χάριν. 
 
 279 It will be remembered that the Greeks said not only'^O*?^- 
 ναίων χάριν, but also την ^ Αθηναίων χάριν (Herod, v. 99) ; also εμην 
 χάριν^ ΰην χάριν, like mea gratia, tua gratia. Besides these modes of 
 expression, we find εν %«ρ6η τινόξ or τινί, like ενεκα=εν εκα, and 
 Λρος χάριν τινός. The last phrase, Λρος χάριν, has created some dif- 
 ficulty in two passages of Sophocles, as to whether it should be taken 
 with the genitive cases with which it is found, or absolutely, in the 
 sense of Mi volupe est; it will not, therefore, be irrelevant to attempt [a 
 settlement of the question. The two passages are as follows; Antig. 29 ; 
 
 εάν δε (Πολυνείκους νεκυν) ακλαυβτον, αταφον-, οίωνοΐς γλυκυν 
 ^ηβαυρον ειύορώΰι τΐρος χάριν βοράς. 
 Philoct. 1155: 
 
 ερτίετε, νυν καλόν, 
 
 άντίφονον κορεύαι ύτόμα τΐρος χάριν 
 εμάς ύαρκος αίόλας. 
 It would be absurd to say that τΐρος χάριν cannot be taken abso- 
 lutely; in which case it means "to please oneself," as in Sophocl. 
 apud Athenceum, p. 220 : Λρος χάριν τε κού βία, Eurip. Supplices, 385 : 
 προς χάριν ^aijjai νεκρούς. And so Λρος ηδονήν is put absolutely in 
 JEschyl, Agam. 262 (unless we read Ιχ%ϋς)\ Eurip. Medea, 771; al- 
 though it is found with a dative in ^sch. Prom. 502 ; Eurip. Iphig. in 
 Aul. 1022. We have τίρος εχ^ραν (Dem. c. Apatur. 900, 12), and 
 τίρος άτίεχ^είαν (lb. 901, 21). In the second passage of Sophocles 
 we should be inclined to take Ίίρος χάριν in this adverbial sense, but 
 in the first we are convinced it stands in the relation of a preposition 
 to βοράς. The following reasons will perhaps make it clear that such 
 is the case. First of all, it must strike any one who has any feeling 
 for Greek construction that the words τΐρος χάριν βοράς come natu- 
 rally together; for χαρίζεΰ&αι is particularly applied to setting food 
 before people, as when Homer says (Odyss. i. 140): 
 
 6ϊτον δ' αΐδοίη ταμίη Λαρε&ηκε φερουΰα, 
 εϊδατα ΐίολί! ετίΐΛίειβα, χαριζομενη τίαρεόντωυ. 
 
488 NOUNS USED AS PBEPOSITIONS. [bOOK III. 
 
 And this seems to have occurred to the Scholiast on the passage of 
 Sophocles, when he wrote: ηγονν rig αντοΐξ βοραν χαρίΰεταυ, and to 
 Euripides (Suppl. 282): μη ατάφονξ χάρματα ^ηρών TCaidag κα- 
 Tidyg. That Jtρog χάριν can stand with a genitive as well as by itself 
 is known to all scholars; in Pind. 01. viii. 10, ανεταυ δε 7tρog χάριν 
 εv6εβίag, we must clearly take the genitive with Jtρog χάριν, "it is 
 accomplished by prayers on account of — for the gratification of — in 
 return for piety;" and in Sophocles, Antig. 908, we have: Tivog νόμου 
 7tρog χάριν. Similarly, we have in Eurip. Med. 541 : 7tρog ίΰχvog χά- 
 ριν; and so Eustathius evidently construes the passage quoted from 
 the Antigone : καΐ avtoig ov TOig άττλώ^, οίον και toig 67tερμoφάγoιg, 
 άλλα Toig 7tρog χάρ^ν ορώΰι βoράg Tfjg άτώ των ΰαρκών (ad II. Θ. 
 ρ. 719, 9), only he has not seen the force of εΐΰοράν, which means 
 "to look at anything with longing eyes," as in Xen. Cyrop. v. 1, § 15 
 (quoted by Sturz, Lex. Xen.) : οντε τovg κaλovg εΰορώ, ουδέ γε 6οΙ 
 ΰνμβονλεύω εν τoΐg κaλoΐg εάν την δ'φιν διατρίβειν. The sense of the 
 lines of Sophocles evidently is " Creon ordered that the body of Po- 
 lynices should be left unwept and unburied, a welcome store for birds, 
 when they are looking out with greedy eyes for a dinner," 
 
 280 The numerous and important family to which χάριg belongs 
 is deserving of a more minute attention than it has hitherto met 
 with. There are, indeed, few sets of words in the Greek language to 
 which researches, such as those in which we are engaged, could be 
 more profitably applied. We may divide the words with which χάριg 
 is connected into two classes ; first, those which contain the root χα-, 
 with or without an affix ; secondly, those which contain the quasi-root 
 χαρ- or exhibit the termination -ρα. To the first belong χάω, %aog, 
 χαίνω, χάΰκω, χανδάνω, χάζομαι, χάννος, χηλή, χειά, χείλος, χέλνς, 
 χαλάω, χατεω, ΰχάζω, and ΰχίξω. Το the second, χάρις., ;^£ΐρ, χερμάς, 
 χοιράς, χερβος, χορός, χώρα and χώρος. If we examine the words of 
 the first class , we shall see that the prevailing and prominent mean- 
 ing is "opening" qrv^'openness." In this the idea of "hollowness" 
 is implied, and, as that which is hollow may be either full or empty, 
 the contrasted notions of content and vacuity are also conveyed by 
 words of this class. When the termination -ρα, which implies motion 
 or continuance, is appended to this root χα- signifying " to lay open," 
 the idea of extent or surface naturally results. And thus we find 
 that the words of the second class imply a surface, something laid flat 
 or open, and by inference, a support or basis, something to rest upon. 
 This meaning appears most clearly in the words χώρα, "a tract of 
 country," χώρος or χορός, "a piece of land," which, though differing 
 in the extent or space signified (§ 229), equally denote a hard, level 
 
CHAP. IV.] NOUNS USED AS PEEPOSITIONS. 489 
 
 surface. The word χορός specially designates a square or public place 
 in which the military people of ancient Hellas met to celebrate their 
 gods with songs and dances of a military character: hence the epi- 
 thet ενρνχοροζ which is applied to the ancient cities. The use of 
 χορός to denote the people assembled on these occasions is quite a 
 secondary one. "We have hinted elsewhere the connexion of χορός 
 with χώρος, χώρα {Theatre of the Greeks^ 6th edition, p. [11], note); 
 that this etymology is the true one is clear from what the King says 
 to the Chorus, in iEschylus, Supplices, 796: λαών εν χώρω τάΰΰεΰΟΈ 
 (see Introduction to the Antig. p. xxix, note). In speaking of the open 
 sea, Thucydides uses ευρυχωρία, in direct opposition to Στενοχώρια, 
 and as synonymous with τΐελαγος (vn. 49). The latter word, which is 
 connected with π:λά|* (Find. P. i. 24: ες βαΟ'εΐαν τίόντου πλάκα), 
 and not with τΐλέω, as Scott and Liddell suppose {Lexic. s. v.), always 
 means "the high sea;" whence jTfAaytog means "out at sea," Thucyd. 
 VIII. 44 ; Polyb. iv. 41, § 2. 
 
 281 We find the same root in χερ-(5ος, which has the collective 
 ending -0ος subjoined, and designates the hard, dry land, as opposed 
 to the sea. It is also an adjective used as an epithet of yij, and 
 meaning "hard," "untilled," "unbroken by the plough:" Sophocl. 
 Antig. 25 1 : ΰτυφλος δε γη καΐ χερβος , αρρώ^ ούδ' ετίημαξευμένη 
 τροχοΐΰον, whence it is applied to unmarried women, (Ed. Tyr. 1502: 
 δηλαδή χερΰους φ^αρηναι κάγάμους υμάς χρεών: by a metaphor 
 similar to that which Creon uses (in the Antigone , 569) in answer to 
 the question of Ismena: 
 
 Ιύμ. άλλα κτενεΐς νυμφεΐα του ΰαυτοϋ τέκνου; 
 
 Κρ. άρώΰυμοϋ γαρ χάτέρων εΐόΐν γΰαι. 
 The use of χοιράς is much the same as the ordinary one of χερ<5ος. 
 The collective ending of the latter renders it more applicable to 
 signify an extended, continuous surface of dry land, whence χερόό- 
 νηΰος means a quasi-island connected at one end with the main-land, 
 a peninsula, whereas χοοράς signifies an island entirely surrounded by 
 water; thus Delos is called ζίηλία χοϋράς (^schyl. Eumen. 9). It 
 does not signify a rock under the water, as the Scholiast on Pindar 
 says {ad Pyth. x. 81), but merely something hard and fixed against 
 which a ship might strike, and thus it is used as an epithet of ηέτρα 
 
 * We may mention obiter that the τιλαγζτοί δίτζλακες of ^schyl. Pers. 
 269, appear to us to describe the άντίΛληγ^ς άκτού of the Straits of Sa- 
 lamis: see our note on the Antigone, 588, p. 181. The epithet πλαγ-κτόζ 
 is well explained by K. O. Miiller, {Klein. Schrift. i. p. 309) by a reference 
 to the appearance of motion assumed by a coast, when the line of breakers 
 keeps oscillating backwards and forwards. 
 
490 NOUNS USED AS PREPOSITIONS. [bOOK III. 
 
 in the passage of Pindar on whicli the Scholiast is writing : ταχν δ^ 
 αγκνραν ερειόον χ%ΌνΙ τΐρώραϋΈ, χοιράδοξ αλκαρ πέτρας. In the 
 plural χοιράδεζ signifies "scrofulous tumours," ^'glandular swellings," 
 from the general meaning "hard," "projecting," borne by %οιράξ\ 
 just as the Latin name struma^ for the same disease, is obviously 
 derived from struo. It is singular, that another Latin word, by 
 which the king's evil is designated, namely, scrofula^ is a diminutive 
 of scrofa, "a sow," just as χοιράξ is connected with χοϊροξ^ the com- 
 mon name for a pig (cf. grice 0. E. and Sc, gris Icel.); and scrofa, 
 scrofula (scrophula) are connected with scrupus, scrupulus and rupee, 
 which are synonymous with the more usual meaning of χοιράζ» 
 There is very little reason, therefore, for deriving χουράς from χοίρος^ 
 as Blomfield does (Gloss, in jEschyl. Pers. 427). The names of 
 animals seem always to be connected with those of certain qualities 
 which they possess in an eminent degree. When we remember that 
 the hedgehog was also called χΊ^ρ or ΰχνρος, we might suppose that 
 the bristly skin of the hog was described by the name. But as this 
 attribute would be better expressed by another root, Sanscrit vrih, 
 Greek φρυκ-, which appear in vardha, φρίΰΰείν, and verres, and as 
 the whole outer form of the pig and the hedgehog suggests the 
 more general idea of a hard projecting object, it is more reasonable 
 to conclude that the name refers to that appearance of the hog's 
 back , which has given rise to the name of the Surrey hill , mentioned 
 by Blomfield in the note above referred to. The same idea is con- 
 veyed by porcus, porca^ which we are inclined to connect immediately 
 with the Sanscrit root vrthj "to grow up" (Pott, Etym, Forsch. ii. 
 p. 53); the derivation from porricere suggested by Yarro and Festus 
 does not seem probable. The root pore- signifies not only a pig, but 
 also a balk or high ridge between two furrows ; and we have a similar 
 resemblance between the English "farrow"="a litter of pigs," A. S. 
 fearh, "a pig," and "furrow," A. S. furh. For another meaning in 
 which χοίρος and porca coincide, and which is welL known to the 
 readers of Aristophanes , we must refer to the application of χερΰος 
 to unmarried women, mentioned above. Although it is ciear that 
 χερμάς is connected with χειρ, it does not therefore follow that it 
 is immediately derived from it, with the limited signification of λί^ος 
 χείροπλήΟ^ηα, as the grammarians and Blomfield (Gloss, in j^schyl. 
 Sept. c. Thebas, 287) suppose. Χερμάξ itself indicates only the hard- 
 ness and roughness, not the size of the stone. Indeed, it appears that 
 the by-form χεράξ rather implies a collection of small stones, shingle, 
 &c., whence τίάμφοροζ χεράς (Pind. Pyth. vi. 13), than any one large 
 and heavy stone (see Journal of Philology, ii. pp. 204 sqq.). The 
 German hartj English "hard," are evidently connected with the 
 
CHAP. IV.] NOUNS USED AS PREPOSITIONS. 491 
 
 quasi-root χαρ- in this signification. With regard to χειρ itself, the 
 primary meening seems to be a combination of the ideas of extension 
 and support, out of which the secondary one of taking, holding, &c. 
 very soon developed itself. In this latter sense it is connected with 
 the Sanscrit verb-root Λη, "to take," the Latin hir, and the Greek 
 aigesLV, κάρ-πος,άρτίάζευν^&ο. (above,p. 298, note). The ΎβτΙοΒχρανω, 
 χρίμπτω, &c., the primary meaning of which is "to touch" (Ruhnken, 
 Timceus^ p. 104), are also secondary to χειρ. The words yhv-td, hin- 
 than, hand, pre-hend-ere, &c., although bearing the same signification, 
 seem rather to be connected with the anusvdra form χανδάνω. 
 
 282 The idea of "opening" conveyed by the root χα- would very 
 naturally be applied to yawning, a wide opening of the jaws, or, in 
 general, to the mouth, the fissure which most frequently meets the 
 eye. Hence we have, as connected with this 'root, the words χεΐ-λος, 
 "a lip," χά-ΰκω, "to yawn," χά-6μα^ "a yawning," χά-ΰμημα, "the 
 wide opening of a bird's mouth" (Aristoph. Av. 61), χήν, "the gaping 
 bird" (χην κεχηνώζ, Athen. 519 a), &c. By a further transition, the 
 secondary root χα-ρ- is employed to denote the noise proceeding from 
 a widely opened mouth, the roar of a lion for instance. Hence it is 
 that χaρo7tόg and χάροφ are common epithets for the lion, and Hesy- 
 chius tells us that χάρων was a name for that animal : χάρων b 
 λέων άτίο της χαροτίότητος. The reason that he gives for it is absurd, 
 for no one would derive χάρων from χαροτίός, though he might derive 
 χαροτίόξ from χάρων. The idea that the epithet χαροτίόξ refers to the 
 colour of the lion is quite erroneous, else how could Lycophron call 
 Achilles περκνοξ αίχμψης χάρων'? for περκνός means black. Hesy- 
 chius and his commentators had a distant inkling of the truth , as 
 appears from the glosses and notes in Alberti's edition. Vol. ii. col. 
 1544: Χαροτΐός. περυχαρής (we should read Λθρίφδρ?^^ with Suidas), 
 γλαυκός, ξαν^'ός, φοβερός, on which Schrevelius writes: χαροττός — a 
 χάρω, id est, χάΰκω, χαρόω, χαρότίω, χαροπός id est δ χάόκων. G. 
 ΑροΙΙοη. Schol. p. 62. Χάρνβδίς. χάΰμα ^αλάΰϋης, Χάρνβδις, 
 ώμόβροτος. η άνατΐίνομένη %'άλα6όα. Χαροτίόν. ^avd'ov. γλανκόν. 
 φοβερόν. περιφερή, and Salmasius asks : An χκροτίός est δ χάόκων, 
 idem τω χάρων, unde φοβερός exponituri We believe that χάρων 
 and χάρνβδις originally meant "the open-mouthed animal" and "the 
 sea that sucked every thing in ; " as Hesychius says a few lines lower 
 down: Χάΰμα ^'ηρός' ο^ις ^ηρός. η [χάό μα πελάγους το της] 
 Ο'αλάΰΰης τΐρόΰωτίον (according to Faber's reading), but that after- 
 wards χαροτίός was used with that signification in regard to the lion's 
 mouth, while χάρων came to signify the lion in general as a wide- 
 mouthed roaring beast. 
 
492 NOUNS USED AS PEEP08ITI0NS. [bOOK III. 
 
 283 There are two reasons which lead us to the conclusion that 
 the lion's roar is particularly referred to in the name χάρων. The first 
 is, that Charon (δ χάρων, " the roarer") is so constantly spoken of in 
 connexion with the χ^Όνία βροντή; the following passages among 
 many others will prove this. Diog. Laert. vii. 28 (in vita Zenonis): 
 ετελεντα δε όντως, εκ της ΰχολης ατίΐών τίροΰεπταιύε , καΙ τον 
 δάκτνλον περίέρρηξε. τίαίύας δε την γην τ^ χειρί φηΰο το εκ της 
 Νιόβης. 'Έρχομαι, τί μ' αν εις; Photius, ι. ρ. 301 Porson: Νοβακ- 
 κίζειν: το ορχονμενον τοις δακτύλους επο'φοφεΐν' 6εΐ6μος Νιόβη. 
 Athenseus, ρ. 341 c: 
 
 αλλ' εΛεΙ 
 6 Τιμο^εον Χάρων ύχολάξείν μ ονκ εα 
 ονκ της Νιόβης, χωρεϊν δε τίορ^'μον άναβοα. 
 Euripides, Alcestis, 252: 
 
 ορώ δίκωπον ορώ ΰκάφος, νεκνων δε πορ&μενς 
 
 'έχων χερ ετά κοντώ Χάρων μ ηδη καλεί' "τι μέλλεις; 
 
 ετίείγον ' 6ν κατείργεις.^^ τάδε τοί με (ΐττερχόμενος ταχννεο. 
 
 The many points of contact between Χάρων, "the roarer," and Γηρνων, 
 " the caller," are also to be added to the evidence which establishes the 
 etymology of the former name: see Varronianus, p. 149, note. 
 
 284 Our other reason for drawing this inference, is the perfect 
 analogy that subsists between the root χαρ- and the root βο- or j3oP-. 
 Thus we have χάρ-ων, "the roaring animal," βονς=βό^-ις, "the bel- 
 lowing or lowing animal" (see below, § 470); and here the roots hri 
 and vrih, which so often come in contact, present another parallelism, 
 for as χάρων is "the lion," so is vrisha "the bull" in Sanscrit, from 
 whatever quality the name may be derived ; χάρ-μη and βοΓ-ή, "the 
 battle-shout," by an extension of usage "the battle itself;" hence βοή ν 
 αγαθός, δ κατατήν μάχην ανδρείος; βοήΟΌον. κατά την μάχην 
 ταχνν (Hesych.), from which comes the word βοηΟ'εΐν, "to assist:" also 
 βοήθεια and βοηδρομία, "a running to a man's assistance in battle." 
 With the same reference we find in Pindar, Pyth. ix. 64: άνδράΰι 
 χ ά ρμαφίλοις αγχιΰτον, where the epithet signifies "nearest at hand to 
 protect." So also we have in ^sch. Ag. 237: αγχιΰτον μονόφρονρον 
 ερκος. We have seen the same connexion of the ideas of neighbour- 
 hood, presence, and assistance in the word άμείνων (§ 262). We have 
 it also in the Latin prcesens deus (Cic. Tusc. Disp. i. 12; Hor. ii.. 
 Carm. v. 2): also in propitiuSj which is the antecedent of prcesens, 
 unless we take the view suggested above (§ 269): cf. Domitius, &c., 
 llythyia (old fern, of είλεν^ώς) might be rendered propitia. From 
 χάρ-μη J in a perfectly similar way, we may derive χραιΰμεΐν, which 
 
CHAP. IV.] NOUNS USED AS PEPOSITIONS. 493 
 
 the Greeks used as a synonym for βoη^'εiVJ and also for επαρκεΐν, a 
 word which we will discuss presently (Schol. on Apollon. ii. 218). In 
 immediate connexion with this words we have χάρις ^ "help," χραν, 
 "to offer help or assistance," and χρηΰίμθ£, "a person capable of 
 offering help or assistance." To this also belongs the use of χρεία in 
 ^schyl. Sept. c. Theh. 49: εί,υότοργιβαυ μοΐραν εν χρεία τνχηξ, and 
 Soph. Aj. 963: %αν6ντ αν οΐμώξειαν εν χρεία δορός. The Greeks, 
 therefore, would have understood why Sir "William of Deloraine was 
 called "good at need." 
 
 285 A most remarkable confirmation of this etymology will be 
 found in the word ηρα, which Buttmann has so fully, and, upon the 
 whole, so satisfactorily discussed. The root of this word appears in 
 αρ-ης^ "war," άρ-είων, αρ-ιΰτος, αρ-ω, άρ-έΰκω,Ο'νμ-ηρ-ης, ερί-ηρ -og, 
 αρ-μενα, άρ-κεΐν, and άρ-ήγευν. The element αρ-, ι^ρ-, which forms the 
 basis of this last set of words, has lost an initial digamma, as appears 
 from a comparison of αρετή, αρρην, arma, ηρως,'Οαρίων, with "war," 
 Wehr, wehren, vir, virtus, "warrior;" as m is often only another form 
 of υ (comp. Mulciber with Vulcanus^ the first pronominal element un- 
 der the forms ma, va, and the German meinen with wahnen; Minne 
 with Ven, Winnesjafte , "friend," "friendship," and Venus); we may 
 also compare the words "Αργίζ and αρρην with Mars and mas (maris) 
 (Buttmann, Abh. Ak. Berl. 1826, p. 58). Now the element /αρ- is 
 obviously related to the Sanscrit root vri, "to protect" or "shelter/' 
 from which comes vrih, "to grow up," as may be seen by comparing 
 vira, "a hero," with ηροζ and vir, and variyas and varisMh'as with 
 αρείων, αριβτος (see Pott, Etymol. Forsch. i. p. 221). As the root hri. 
 "to take," which we shall see has other points of contact with νηΛ, 
 the secondary form of this root, appears in χειρ, &c., so does this root 
 vrt appear in Faρί6τερos, "the left or shield hand" (above, p. 301). 
 In the words from the root Γαρ- , which we have enumerated above, 
 there is precisely the same transition of ideas as in the two sets of 
 derivatives from χάρμη and βοή, which we have just been considering. 
 This would be sufficiently clear from Hesychius only, if we had no 
 other means of showing it. See the following glosses: Έπίηρα. την 
 μετ ετΐίκονρίας χάριν μεγάλην, η εκ τήξ τΰεριονδίας ' Sg Αντίμα- 
 χος.— Ε π ίη ρ ο g. ετίίκονρος. ετα^νμητής. — Έπίηρος. βοη%^ός. χά- 
 ριν ccjtodidovg. — ' Ετίΐήρεα. ετΐίχαρίτια. — Ήρα. ήτοι, όντως, ή 
 χάριν, βοή^^ειαν, Ιτίΐκονρίαν , τΐατρϊ φίλω ετά ήρα φέρων ζ/ί/. ή 
 εφη. And the words of Apollonius; Έτΐίηρα' την μετ επικουρίας 
 χάριν. ΜητρΙ φίλ^ ετίίηρα φέρων, εν δε τω, ουδέ τι μοι ποδάνιπτρα 
 ποδών ετίίηρ άνα %νμω, τα ετίΐκονρητικα της -ψυχής, οϋτως Άρί- 
 6τ άρχος. If, in addition to all this, we compare ετίίηρα φέρειν^^έπα- 
 
494 NOUNS USED AS PREPOSITIONS. [bOOK III. 
 
 ρήγείν = επαρκεΐν , with ετίίβοη^'εΐν, χάριν φερειν , and χραιΰμεΐν, 
 we must feel an irresistible conviction that these expressions are all 
 due to the same train of ideas in a Greek mind ; that, in fact, the ideas 
 of good , assistance, favour , and pleasure , were evolved in the Greek 
 language from the military feelings of the heroic age, 
 
 286 The secondary root χα-ρ appears with the pronominal affix 
 -κ- in the word χάραζ and its derivative χαράόΰω. It will be de- 
 sirable to explain these two words. We are told that when Ιχάραξ 
 signifies "a vine-prop," it is feminine, when it denotes "a palisade," 
 it is masculine: thus Moeris says (p. 372 ed. Koch): Χάραξ, τ] μεν 
 7Τρο8 ταΐς άμτιελοΐξ, ^ηλνκώξ. δ 8ε εν τοις ΰτρατοτιέδοις άρΰενικώς. 
 And Phrynichus (p. 61 Lobeck): ^H χάραξ ερεΐς ^νλνκώς το τηξ 
 αμπέλου ΰτήρίγμα, ου κάζα το αρρενικόν. Α similar remark is made 
 with regard to κάμαξ in the Etymologicum Magnum, From this we 
 conclude that the two significations of the word belong to different 
 ages of the Greek language. Now αμτΐελος was feminine from the 
 first; and as the vine leans upon and twines round its prop, like a 
 sister, for support, it may be believed that the oldest signification of 
 χάραζ was "the supporter or helper of the vine," in which sense its 
 connexion with the family of words we have been discussing is indis- 
 putable. In confirmation of this, let us observe that we find in Homer 
 the word άοΰΰητήρ, in the sense "a person who stands behind us to 
 help us up" (Iliad xv. 735): 
 
 ηε τίνάξ φαμεν είναυ άοΰΰητηρας οτίίΰΰω. 
 Now άοΰΰητήρ is obviously another form οί αοζος, "a servant" or 
 'Hielper" (-^sch. Agam. 209), and aoiog means "a branch which 
 grows up by the side of another branch," for, according to Theophrastus 
 (Hist. Plant, i. 1 3), οξοξ is that part of a tree from which the branch 
 sprouts out, and consequently, as ά-δελφός means "he who springs from 
 the same δελφύς or womb," so α-οξος means " the sister branch," "the 
 branch which derives its origin from the same nodus :^^ therefore ao6- 
 ΰψήρ, which the Scholiast on Homer (loc. cit.) explains by βοη^'ός, 
 conveys the same idea as χάραζ, and conversely χάραξ may imply a 
 βοή^'ευα^ and so it is naturally connected with χάρις, according to our 
 former investigation. Χαράΰόω is of course formed from χάραξ, just as 
 καμάΰΰω is from κάμαξ; and as καμάβΰω means " to make like a reed" 
 (κάμαξ), that is, "to cause to shake," so χαράΰόω properly signifies 
 "to make like a stake," that is, "to sharpen," and in this sense the 
 word frequently occurs in the oldest writers. It also signifies to pro- 
 duce the effects of a sharp instrument on some substance, just as 
 άνάΰΰω means "to perform the functions of a king" (αναξ); and it is 
 to this latter sense that we owe the important word χαρακτήρ, both 
 
CHAP. IV.] NOUNS USED AS PREPOSITIONS. 495 
 
 as it is applied to signify "the stamp on a coin," and as it is figura- 
 tively used to denote the stamp of mind which distinguishes one man 
 from another. In Hesychius we have the gloss: κεχαραγμενος. 
 ώργιβμενος, which is supported by Herodotus vu. 1 : μεχάλως κεχα- 
 ραγμενον τοΐΰί Ά%ηναίθί6ί, and Eurip. Med. 157: κείνω τόΰον μη 
 χαράΰΰον. This meaning might seem to be derived from the first sig- 
 nification of χαράΰόω, for ^ήγομαυ and οξύνομαι are used in a simi- 
 larly metaphorical manner. If, however, the ingenious emendation of 
 Matreas (apud Athen. p. 19 d) suggested by Blomfield {Gloss, ad Pers. 
 689) is to be received, we must conclude that the metaphorical use of 
 χαράΰΰείν is derived from the second of the primitive meanings. The 
 passage in Athenseus stands thus : Έποίηΰε δε ovTog (δ Ματρεαξ) 
 καΐ τίαρα τας^^ριΟτοτελονς άτίορίαξ, και άνζγίνωόκε δημοΰία δοα 
 τι 6 ηλίος δννει μεν^ κολνμβα δ' ον. κα\ δια τι οι βτΐόγγοι Ονμτίί- 
 νουβι μεν, όνγκωΟ'ωνίζονται ό' ον' καΐ τα τετράδραχμα κατ αλ- 
 λ άτ τετ αν μεν J οργίζεται ά' ον. Blomfield reads χαράττεταί, 
 which, as a synonym for οργίζεται, is more in place here than τίαταλ- 
 λάττεται, which bears just the contrary signification. 
 
 287 It has been mentioned before (§ 53) that associations by 
 way of contrast are often expressed by the same root or element in 
 the languages with which we are concerned. This is particularly 
 remarkable in the class of words into which the element χα-ρ- enters. 
 That the notions of emptiness and containing are both expressed by 
 the simple root χα- has already been shown. Now we find precisely 
 the same conversion in the derived root ^c^-^,,the primary acceptation 
 of which denotes "protection," "good," "benefit," and the feeling of 
 joy which the possession of such things imparts. By the law of asso- 
 ciation mentioned above, it also expresses the feeling of desire created 
 by the want of such things. Thus %ρ]ίξω and χηροζ belong to the 
 same family with χάρΐξ^ χραιΰμεω, χρη(5^αι. The ideas conveyed by 
 both these sets of words are included in the single word χρεία, which 
 denotes both "use" and "need;" and the uncertainty, which arises in 
 some cases as to the proper way of translating this word, shows how 
 nearly these ideas are connected with one another. . Hesychius uses 
 χρείαν έχων as an interpretation of χατεων, χηροξ^ and χρηΐζων, and 
 the doubt entertained by the critics as to whether we should read 
 (^schyl. Pers. 815) ΰωφρονεΐν κεχρημένοι, "in want of teaching," or 
 ϋωφρονεΐν κεχρημενον, "having wisdom," is a sort of proof that the verb 
 also conveys these two ideas (Wellauer and Dindorf ad I. j^schyl. and 
 Elmsl. ad Eurip. Heracl. 801). We observe the same connexion in the 
 Latin cdrus, gratus, grates, and gratia, which Passow considers to be 
 connected with this element. Cdrus may be a corrupted participle 
 
496 NOUNS USED AS PEEPOSITIONS. [bOOK ΠΙ. 
 
 from careo, just as purus is a participle oiputo; verus of vereor; pro- 
 cerus oi procello (where the e is short by nature, compare columen^ 
 s-celus, proculus, βον-ηόλ-οζ,&ο,.)] obscurus oi ob(sJculo=^occulo. It is 
 to be observed that the ending rus=sus (§ 107) is not at all uncom- 
 mon in Latin words of this class : thus we have cla-rus, glo-ria (κλέθ£, 
 κλνω, &c.); ca-rus, ca-reo (χάος)] seve-rus (βέβω, Umbrian sevum); &c. 
 We have a by-form of cams in cassus, which seems to be the proper 
 form of a participle from some verb like χατεω ; compare fateor^ par- 
 ticiple /a^^ws. Gratus=^cardtus might be the participle of a deriva- 
 tive verb like carare, if it existed. It would be foreign to our present 
 purpose to enumerate all the Latin words of this family ; otherwise it 
 would be easy to show that the ideas of value , preciousness , conse- 
 quent difficulty in obtaining, or even striving in vain to get, and 
 therefore being without , are developed from one another in that lan- 
 guage also, like the two meanings of the English adjective "dear." 
 Doderlein, therefore, is mistaken when he derives careo and cassus 
 from KBLQEiv, καρηναί, carper e^ κάρφειν (Lat. Synon. und Etym. iii. 
 p. 114, note), as opposed to cams and grains, which, he admits, arc 
 connected with χάρΐξ and χαίρω (p. 254). 
 
 288 Bopp (Glossar. Sanscr. p. 404) and Pott {Etymol. Forsch. i. 
 p. 272) are inclined to connect χαίρω with the Sanscrit hrish. The 
 primitive meaning of this Sanscrit root is "to be erect," and it is par- 
 ticularly applied to the hair of the head, whence the epithet hrisJita- 
 romd^ "with the hair of the body on end" {Bliagavad-Gita, xi. 4; 
 Lassen, Anthologia Sanscritica, p. 4, 1. 9). In a secondary sense, it 
 signifies "to rejoice," "to be elated," "to exult," "to be exceedingly 
 pleased," " to have the hair of the body erect with pleasure ;" so that 
 it seems to agree in all its meanings with φρίΰύω (root φρικ-) rather 
 than with χαίρω. Whether it is applied to the projecting spears of a 
 body of soldiers, to the erect hairs of the head, to the standing corn, 
 to the first ripple on the sea, or to the appearance of the skin when 
 one is shivering from cold (which we call "goose's skin"), the primary 
 meaning of φρίκη, φρίζ: φρίΰΰω, is always projection or unevenness 
 in a physical sense; the mental emotion sometimes implied being of 
 course secondary and metaphorical. We have the same idea in the 
 Latin horrere (=^horsere'^)^ hirsutus, &c. Although these words are so 
 nearly connected in meaning, it is difficult to pronounce a decided 
 opinion as to their etymological relation to one another. It is clear 
 that hrish-, hirs-, and φρΖκ-, must be secondary formations, and it is 
 obvious that the first two are the same root. We are inclined to 
 connect φρΐκ- like pore-, with the Sanscrit root vrih, "to grow," so 
 that this root , the meanings of which bear a great resemblance to 
 
CHAP. IV.] NOUNS USED AS PREPOSITIONS. 497 
 
 those of %cc-Q-, again approximates, in its secondary lengthened form 
 ςρρΓκ-, to a secondary and lengthened form of the other root hri. It 
 is singular that not only does this root hri agree with χαρ- in its 
 military use (ior pra-?iri signifies "to fight," and pra-hdra, "a com- 
 batant"), but we find the Homeric χάρμη even in the modern languages 
 of Europe, as a remnant of the warlike Goths. Thus we have the 
 German Schirm^ Italian Schermo , with one of the primitive meanings 
 of the element ;^α-ρ-, namely, defence, protection, reliance, &c.; and 
 German Scharmutzel , Italian Scherma, English Skirmish, with the 
 common Homeric signification of χάρμη j to which these words are 
 related, as the German Schaum to χυμός, Schelm to χάL•μogJ χαλί- 
 μαδες^ and Schief to χαβός (see Doderlein, Vocabulorum Homericorum 
 Etyma, p. 14). 
 
 289 (3) Before we consider the remaining significations of ;^a^tg 
 it will be proper to discuss δίκην, the third of those nouns which are 
 used as prepositions, for it bears a remarkable analogy, in some of its 
 applications, to χάρις. As a preposition with the genitive case, δίκην 
 is equivalent to the Latin instar, and signifies "like," "after the like- 
 ness of." Thus in Pindar {Fyth. n. 84), λύκοίο δίκαν is "just like 
 (i. e. justly) a wolf;" in^schylus {Agam, 3), κννος δίκην means "just 
 like a watch-dog." The use of δίκην as a preposition seems to be 
 for the most part confined to the older poets ; for, although it occurs 
 even in Plato and Aristotle, it is generally used when an air of quaint- 
 ness or a poetical colouring is designed. For instance, Plato, Legg. 
 VI. p. 773, ου γαρ ράδιον εννοεΐν, on τίόλιν είναι δει δίκην κρα- 
 τηροξ κεκραμενην, seems to be quoting some line from a play, such 
 as jcoXig δίκην κρατηρος ην κεκραμενη ; just as, in Legg. x. p. 886 e, 
 λόγοιΰι δε ταύτα ευ τίως εις το τίΐ^ανον τίερίΛετΐεμμενα ("well in- 
 crusted, covered or concealed with words, so as to appear probable"), 
 he seems to have have had in his head some line of an old comedian — 
 perhaps λόγοιΰι δ' ευ τνως ταύτα περιπε7ίεμμενα\ compare Aristoph. 
 Plut. 157: ονόματι τίεριτιεττουύι την μοχ^ηρίαν. Vesp. 668: ρημα- 
 τίοις ηεριπεφ^είξ. 
 
 290 The sense of δίκη^ which has given rise to this use of its 
 accusative or old locative case as a preposition , is found in Homer, 
 Odyss. xvni. 274 : μνηβτήρων ουχ ηδε δίκη το πάροικε τετυκτο, and 
 in Pindar, Pyth. ι. 50: ταν Φιλοκτήταο δίκαν εφέτίων, which the 
 Scholiast rightly explains : τον Φιλοκτητου τρότΐον μετερχόμενος, for 
 τρόπον is also used in the same way as δίκην, as in -^schyl. Agam. 48 : 
 τρόπον αίγυπιών, "like vultures." So also δέμας, "the outward build 
 or body of an object," as in Horn. II. xi. 595: ώ^ οι μεν μάρναντο, 
 
 κκ 
 
498 NOUNS USED AS PBEPOSITIONS. [bOOK III. 
 
 δίμας TtVQog αΐ&ομενοω, which Hesychius explains as τρόπον TtVQOS 
 καίομίνον, and with this reference he renders δέμας by μορφή, 
 ιδέα, τρόπος. Hesychius also recognises the meaning of likeness or 
 similitude here implied, as will appear from the following glosses: 
 δίκψ 6 τρόπος — μνηΰτήρων (he refers to the following of the Odyssee 
 above quoted); δίκηλον. (1) εκτνπωμκ. ομοίωμα, εϊδωλον^ άνδριας, 
 ξώδίον. πάρα Αάκωΰιν. (2) φάΰμα, Ο'φίς, εϊδωλον, μίμημα. o^tεv καΐ 
 δ μιμολόγος πάρα ΑάκωΟι, δίκηλίΰτας. (3) άγαλμα ανδρίαντος ; 
 δίκη ν. (1) κα&άπερ, ωόπερ, ομοίως. (2) τρόχον (read τρόπον); 
 δίκης, τρόπον, dt κ ?yff (this word should be inserted), ομοίωβις. η 
 κρίΰυς. But this is only a secondary sense of δίκη. The following con- 
 siderations will convince us that its primary meaning was "an equi- 
 valent," that is, not only a similitude, but an identity. This appears 
 most clearly from the uses of δίκαιος in the best writers. Thus we 
 have in Herodotus, ii. 149: ai d' εκατόν οργνιαΐ δίκαιαί elut ΰτάδων 
 ε^άπλε^'ρον, i. e. " one hundred fathoms are exactly or just equivalent 
 to a stadium." Referring to which, as it seems, the Antiatticistes says 
 {Bekk. Anecd. p. 90, 1. 20): δίκαιον μέτρον. το ϊΰον:' Ηρόδοτος 
 δεντερω. Similarly, Xenophon, Cyrop. π. 2, § 26: οντε γαρ άρμα 
 δήπον ταχν γενοιτ αν βραδεων ίππων ενόντων, οντε δίκαιον αδί- 
 κων ύννεζενγμενων , "when the horses are not a pair;" cf. Soph. 
 Antig. 292: ονδ' νπο ζνγω λόφον δικαίως εΐχον. Ibid. 662: μενειν 
 δίκαιον κάγα^ον παραότάτην , where we have endeavoured to ex- 
 press this idea in our version. That this is really the meaning is shown 
 by the phrase in Plato, Phcedrus, p. 247 b: τα μεν ovv %'εών οχή- 
 ματα ίΰορρόπως ενήνια οντά, because in the allegory the chariot of 
 the human soul is represented as drawn by two horses of different 
 quality, which did not pull well together. In the same way we have 
 (Hippocr. de Art. p. 787 b) δίκαιον ΰώμα, "a body equal on both 
 sides;" (Id. deFract. p. 783 e) δικαιόταται αντιρροπαί, "perfect equi- 
 librium;" (Id. Ibid. p. 772 a) κατάταΰιν δικαίην και ομάλην, "an equal, 
 level extension." It is to this primary sense that the moral, legal, and 
 political use of δίκη is due, just as from the similar application of the 
 X^atin cequus and iniquus spring the sense of counterpoise or equi- 
 valence. Thus δίκην δονναι, λαβείν, εχειν, di-ioxftT/, &c., "to give, ob- 
 tain, have, or endeavour to get, satisfaction, or an equivalent for some 
 injury," on the principle of the lex talionis, which the old Greek 
 legislators considered to be perfect justice. Aristotle, who does not 
 admit of the universal applicability of retaliation {EtJi. v. 5), and 
 would rather consider δίκη as something proportional {ανάλογόν τι), 
 than as an equivalent (Eth. v. 3, § 8), was nevertheless perfectly aware, 
 that, according to the ordinary acceptation of the term in Greece, 
 δίκη conveyed the idea of a quid pro quo: for he says that inequality 
 
CHAP. IV.] NOUNS USED AS PREPOSITIONS. 499 
 
 and injustice are synonymous terms, and that to have more than one's 
 share (τΐλεονετίτεΐν) is to commit an injury ; the same appears from his 
 ingenious but false derivation of δίκη from δίχα (Eth. v. 4, § 9): το 
 δ' ΐόον μέβον ΙύτΙ της μείζονος κάί ελάχτονος κατά την άρίχί'μετυκην 
 άναλογίαν δυα τοϋτο και ονομάζεται δίκαιον^οτι δίχα εβτίν ωότΐερ 
 αν εϊ τις εΧτίοι δίχαιον. και 6 δικαΰτης διχαβτής. As καιρός often 
 denotes the proper measure or proportion (Pind. 01. ix. 38; JEsch. 
 Ag. 760), it is interesting to observe that άκαιρος is used as a sy- 
 nonym for αδίκως in iEsch. Ag. 781; cf. Choeph. 615 [1. άκαιρων]; 
 Theogn. 899 (919). 
 
 291 We now return to ;^άρΐ5, which, besides the meanings we 
 have already discussed (namely, good, protection, benefit conferred, 
 and the feeling of joy which such things create), has some applications 
 intimately connected with the primitive sense of δίκη. In its widest 
 sense χάρις expresses every thing that is graceful, amiable, and charm- 
 ing, and it is therefore very difficult to find any one English word 
 Λvhich will generally translate it. As it includes the kindness, which 
 confers a boon or favour, no less than the gratitude which desires to 
 make a suitable return, it may often be rendered by "mutual good 
 will." Perhaps both of these feelings are implied in the use of χάρις 
 to denote the worship of the gods ; for while the religious act involves 
 the free-will offering of gifts , by which the gods are presumed to be 
 gratified (Plato, i^wiA^^^Aro, 15 a: τι δ' οϊει άλλο η τιμή και γέρα και, 
 όπερ εγώ άρτι ελεγον, η χάρις ; κεχαριΰμένον αρα εότι το ούιον), it 
 certainly indicates the gratitude of the worshipper in certain cases, as 
 when we have ευκταία χάρις, "a votive offering" (^sch. Agam. 1360). 
 But χάρις also means the worship of sacred objects regarded as a 
 reverence due to them (JEsch. Agam. 363 : οόοις ά%ίκτων χάρις Λα- 
 τοΓτο) ; and in a very difficult passage ^schylus seems to say that the 
 reverence due to the deities is forced upon man by suffering {δαιμόνων 
 δε Λον χάρις βίαιος, Agam. 176 : for the meaning of βίαιος see Plato, 
 Έββρ. 603 c; and for the sentiment, cf. Pers. 409 — 491, and Andrewes' 
 Sermons, Vol. i. pp. 305 sqq. Oxf. ed.). Mythologically, η Χάρις, or 
 more generally in the plural at Χάριτες, are the goddesses who preside 
 over all that imparts a charm to the social relations of man (Pindar, 
 Olymp. I. 30 : Χάρις, απερ άπαντα τενχει τα μείλιχα Ο^νατοΐς. Where 
 the Schol. Vratisl. says truly : η χάρις φηόΐ της ποιητικής γραφής. 
 Olymp. XIV. 5 : Χάριτες — 6νν νμμιν — τά τε τερπνά και τα γλυκέα 
 γίγνεται πάντα βροτοΐς). The ordinary names of these divinities, 
 Aglaia, Euphrosyne, and Thalia^ refer specifically to the kindness of 
 social festivity (see note on Pind. Nem. v. 38); and they had an inti- 
 mate connexion with Dionysus (Schol. ad Find 01. v. 10; Pausan. v. 14, 
 
 KK 2 
 
500 NOUNS USED AS PEEPOSITIONS. [BOOK III. 
 
 fin.; Plut. Qu. Gr. 36 ; Apoll. Rhod. iv. 424). And their worship was 
 designed to sanction religiously that interchange of good offices which 
 is the foundation of δίτιη-, or "give and take." Thus Aristotle says 
 {Ethic. V. 5, § 6): τω αντιΛοιέιν γαρ άνάλογον ΰνμμένευ η τίόλις' 
 η γαρ το κακώς ζψονϋιν (εΐ δε μη, δουλεία δοκεΐ είναι, ει μη αντι- 
 Λοοήΰεϊ) ' η το εν {εΐ δε μη, μετάδούυς ον γίνεται, τ^ μεταδόόει ΰνμ- 
 μενονΰι). δώ καϊ Χαρίτων ίερον εμποδών ποιούνται, ίνα άνταπό- 
 δοΰις y' τοντο γαρ χάριτοζ ϊδιον αν^υτίηρετηϋαι γαρ δει τω χα- 
 ριύαμενω, καΐ πάλιν αντον αρξαι χαριΰάμενον. And hence the 
 Eumenides, praying that there may be no factions at Athens, say 
 (^schyl. Eumen. 970): 
 
 χάρματα δ' άντιδίδοιεν κοινοφελεΐ διάνοια, 
 
 καΐ ΰτνγεΐν μια φρενί. 
 
 πολλών γαρ τόδ' εν βροτοΐς ακος. 
 
 At Athens, according to Josephus (Antiquit. xiv. c. 8, § 5, p. 699), 
 a common temple was erected to them and the Demus : ΰτηΰαι αντον 
 εικόνα χαλκην εν τω τεμένει τον ζίήμον κάί τών Χαρίτων ', and their 
 statues stood at the entrance to the Acropolis, where they were wor- 
 shipped with mysterious rites: Pausan. ix. 35, § 3 : ^Α^ηνηύι προ της 
 ες την ^Ακρόπολιν εβόδον Χάριτες είΰι και ανται τρεις' πάρα δε 
 ανταΐς τελετην αγονΰιν ες τους πολλονς απόρρητον. There was a co- 
 lossal statue of the patroness Juno by Polycletus in the Herseum at 
 Argos,on whose crown the Hours and Graces were sculptured, and their 
 statues were in the Pronaus (Pausan. ii. 17, § 3, 4, and v. 11, § 7). The 
 chief attribute of the Graces was sociability : they are represented as 
 inseparable from one another , and as promoting all kinds of unions 
 among mankind; — that of matrimony, of the family (πάτρα), of the 
 civic phratria at the εύτίαόις, of the whole state or race at the public 
 festivals (Miiller, Orchomen. p. 180). It is with this feeling of the po- 
 litical significance of their w^orship that Pindar says (Pyth. viii. 21): 
 
 επεΰε δ' ον Χαρίτων εκας 
 
 α δικαιόπολις άρεταϊς 
 
 κλειναΐΰιν Αίακιδαν 
 
 ^ίγοιΰα νάΰος. 
 for he would hardly have used the epithet δικαιόπολις had there not 
 been some connexion in signification between χάρις and δίκη (cf. 
 Thucyd. m. 67 : άνταπόδοτε χάριν δικαίαν); his meaning is "the fair- 
 dealing and glorious island of ^gina is not disregarded by the Graces, 
 — for they preside over the intercourse of men, and are also the 
 givers of glory." The epithet δικαιόπολις, — which is properly appli- 
 cable to a man (it is the name of the hero of the Acharnians of Aris- 
 tophanes, and is analogous to απολις,νψίπολις,&ο.), but is here applied 
 
CHAP. IV.] NOUNS USED AS PEEPOSITIONS. 501 
 
 by personification to JEgina, — refers to the fairness which character- 
 ized the commercial dealings of that island, and for which Pindar 
 elsewhere extols its inhabitants. In the same sense, we have ex- 
 plained (not. ad I.) Olymp. vin. 20 sqq.: where Σώτπρα ^ώζ ξενίου 
 τίάρεδρος Θεμίξ is only another name for zJiari (see Sophocl. (Ed. Col. 
 1384 : 7} τίαλαίψατοξ ^ίκη ξννεδρος Zrivog αρχαίους νόμους). In the 
 Pythian hymn quoted above ετκεΰε does not refer to the insular posi- 
 tion of ^gina, as Dissen supposes, but is used in the same sense as in 
 Sophocl. Aj. 620: 
 
 τα 3ΤρΙι^ δ' έργα χεροΐν 
 
 μεγίΰτας άρετάς 
 
 αψίλα τΐαρ' αφίλοίζ 
 
 ετίεύ' , επεβε μελέοις ^Ατρείδαίξ. 
 
 as the Scholiast perceived: η δε δικαιότίολίζ vrj6og Αϊγινα ουκ εξε- 
 πεΰε των Χαρίτων, i.e. "^gina has not fallen or been thrown aside, 
 banished far from the presiding deities of mutual kindness and good 
 
 will." 
 
 292 The etymological connexion of δίκη and χάρις is even more 
 remarkable than their analogy in signification. As χάρίς is connected 
 with χειρ, the general name for a hand , and with a number of words 
 signifying "to take or hold" (above, p. 298, note), δίκη is obviously 
 connected with δεκ-ΰιος, the name for the right hand, with δάκτυλος, 
 "the finger," and with a number of words denoting "to receive" 
 (δεκομαι, &c.), or "to point out" {δείκνυμι, &c.), (above, p. 301, note). 
 The element of the word is, as we have seen (above, pp. 294 foil.) , a 
 compound of the numeral "two" with a root signifying "to take," 
 and probably connected with the root hrt, which appears in χειρ. The 
 form δι- of the first syllable is on the analogy of διά, δί-δυμος, &c., and 
 is more regular and original than the δε- of δέκα or the δα- of δάκτυ- 
 λος. We have this t in δείκνυμι, where it is afiected by guna, in the 
 Sanscrit dig, and in the Latin dicis causa, dicere, digitus, dignus, &c. 
 Herodian remarks (π:ερΙ μονήρους λέξεως, p. 14) that the accentuation 
 of δίκη is very singular, because a barytone noun of this kind ought 
 either to have a consonant before the κ, like κίρκη, δίρκη, τρίκκη, or 
 the ι should be long, as in νίκη, φρίκη — ή τοίνυν δίκη όημειώδες. 
 λείττει γαρ η χρόνω η ΰυμφώνω. Nouns in η, formed from adjectives 
 in -Ικός, are naturally oxytone : but there is no reason whatever why 
 δίκη, — which has no connexion with the pronominal termination -κός, 
 but it is formed directly from the verb-root δνκ- (δίκη for δίκya), — 
 should not follow tlie analogy of ελίκη, εκάτη, μελέτη, &c. 
 
CHAPTER V. 
 THE ADJECTIVE. 
 
 293 Etymological distinction of adjective and substantive. 294 Under what 
 circumstances an adjective or participle may become a substantive or 
 definitive name. 295 This transference is particularly common in Latin. 
 296 It is also found in Greek. Connexion of the participle with nouns 
 denoting agency and agents. 297 Digression respecting the substantive 
 use of δνσφρονα and γεραρά. 298 The adjective often represents the 
 genitive case of a substantive, and is sometimes of adverbial origin. 299 
 Hence the adjective in its distinctive use is merely a syntactical con- 
 trivance. 300 Predicable nature of the adjective and participle. 301 The 
 three different kinds of predicates may be expressed by adjectives. 302 
 Secondary predicates may also be expressed by oblique cases of nouns 
 or adverbs, and to these the tertiary predicate is often attached. 303 
 Certain adjectives especially used in this way. 304 Epithets and pre- 
 dicates have been confused by great scholars. 305 Familiar illustrations 
 of the general principle. 306 Convertible propositions. 
 
 293 Tj^TYMOLOGERS have found or created for themselves 
 very great difficulties in the nouns adjective. We do 
 not intend to set forth all the explanations which have been 
 offered with regard to their nature and functions. It will be 
 better to state at once that the adjective differs etymologically 
 from the substantive only in being capable of flexion through 
 the different genders of the substantive to which it is joined. 
 Otherwise it is as much the designation of a quality or attri- 
 bute, and therefore as truly a noun^ or the name of a thing, as 
 the substantive itself. As for the compound adjectives, they 
 are in many languages merely substantives subjoined to ad- 
 jectives. In general, the explanation of the adjective belongs 
 to syntax rather than to etymology, 
 
 294 To the student of Greek the adjective is particularly in- 
 teresting, and especially in its connexion with the participle, a kind 
 of word of which more use is made in Greek than in any other lan- 
 guage, insomuch that the Greeks have been emphatically called φίλο- 
 μίτοχοι, or lovers of participles. While in this language — more per- 
 haps than in any other — adjectives and participles are employed to 
 express all the adverbial or accessary relations of the sentence, a great 
 number of adjectives have taken their station amongst the most com- 
 mon of the substantives, and there is no single Greek adjective or 
 even participle which may not become a substantive if it only has the 
 
CHAP, v.] THE ADJECTIVE. 503 
 
 definite article prefixed, if, in a word, it has that accompaniment 
 which is necessary for the conversion of a substantive, as the name of 
 a quality or attribute, into the name of a particular thing*. Indeed, 
 to such an extent has this been carried, that many adjectives, espe- 
 cially those ending in -χή, which have obtained a substantive use by 
 prefixing the article and omitting the substantive τέχνη, have at last 
 become so completely substantives, that the article is always omitted, 
 except in those cases where a substantive would require this auxiliary 
 (see Middleton On the Greek Article, pp.xxi, 50, note, edit. Rose). This 
 restricted employment of a general attribute may be compared with 
 the use of βαβυλενζ without the article , when a particular king, the 
 king of Persia, is meant, so that the general term becomes a proper 
 name or appropriated word. 
 
 295 The use of participles as mere adjectives is undoubtedly 
 much more common in Latin than in Greek, and the reason is this, 
 that as the Latin language has no definite article, the distinction 
 between ό φιλών, "the lover," and φιλών, "if he loves," cannot be 
 expressed by this part of speech. It becomes therefore a matter of 
 indifierence whether we use the Latin participle as a definitive or as a 
 hypothetical word; but in all cases where a distinct protasis was 
 intended, the participle would give way to the conditional sentence. 
 Conversely, all the functions of an adjective would be assumed by a 
 participle whether active or passive. The passage from this to the 
 substantive use of the participle is immediate. Thus the active par- 
 ticiples adolescens, parens, and sapiens, are constantly found as sub- 
 stantives; secundus is always an adjective, and is generally used in a 
 metaphorical or applied sense. The passive participles acutus, argu- 
 tusj &c. are almost always employed as epithets, and the neuters prce- 
 ceptum, dictum, scriptum, consultum, placitum, furtuni, &c. are to all 
 intents and purposes substantives. Some of the participles in -ns 
 have their comparative and superlative degrees like the ordinary 
 adjectives. Thus we find parentior, utentior, appetentior, and appe- 
 tentissimus (Lubker, de participiis Grcecis Latinisque, p. 12). In 
 some cases we have actually to reproduce the participial meaning by 
 a close examination of some common noun. There is an interesting 
 exemplification of this in the words pons and fons. From the root 
 pos, strengthened by w in the present of po[syno, pos-ui, we have the 
 
 * When the Anglo-Saxon present participle is used as a noun, it is 
 distinguished by a weaker form of inflexion (Latham, Eng. Lang. p. 71, 
 ed. 2). This appears to us to indicate the absence of a definite article: 
 comp. the H. G. der gute Mensch with guter Mensch. 
 
504 THE ADJECTIVE. [bOOK III. 
 
 participial noun pons=^pos-nts, which had a primitive form pos 
 (Varro, L. L. Vol. i. p. 3 Muller), and this conveyed the idea of laying 
 down heavily, whether this signified that a mass of stones was thrown 
 into the water (γε-φνρα), or generally that there was a weight which 
 caused an inclination of the scale. This, no douht, is the origin of 
 s-ponte, which refers to the momentum of moral inclination, and thus 
 we get the explanation of the ponti-fex^ who settled the atonement 
 by the imposition of a fine, i.e. a certain weight of copper, as opposed 
 to the carni-feXj who took satisfaction on the body of the delinquent. 
 Hence we have the secondary forms pendo^ pondus, &c. Similarly, 
 from the root fo=svo or 7ιυο=χΒ^- for %/ε-, "to pour out," we have 
 the participial noun fons, "that which pours forth water," i.e. "a 
 fountain," and from this the secondary forms /undo and fundus. That 
 these roots fo=svo or hvo, and po or spo, "to pour forth" or "cause 
 to fall," have an identity of signification as they probably have a com- 
 munity of origin, may appear from the connexion of meaning between 
 χώμα and pons, between ΰΛενδω and /undo. 
 
 296 The Greek participle is not used as a substantive except in 
 cases analogous to those which we have just discussed , namely, when 
 the hypothetical use of the word has become quite indistinct. Thus 
 the nouns αγνια, αρηπα, οργνια, &c., though obviously participles 
 from the roots άγ-, άρΛ-, ορεγ-, &c., never occur except as feminine 
 substantives. The same remark applies to the proper names EUsi^VLa 
 and Κάλχας, though the former signifies the pains of childbirth as well 
 as the goddess: soHesych.: eiXst^vicig ενίοτε μεν rag &εάς, ενίοτε δε 
 τας οδυνας-, see Hom. Β. χιχ. 119, where the Scholiast explains ΰχεΟ'ε 
 δ* είλει^νίας by ώδΐνας εΛεΰχεν. The form ελεv^'ώ shows us the con- 
 nexion between the feminines in -vLa and -ω ; see above, § 257. (For 
 the participial origin οι Κάλχας, see our note on the Antigone, p. 136, 
 and compare the subsequent remarks of Pott, Zeitschr. f. vergl. Sprf. 
 VI. p. 114, VII. p. 244). It seems that ελέφα(ντ)ς is not a participial 
 word , but a corruption of the Semitic name of the elephant, as aleph 
 hind, "the Indian ox," just as tamarind is tamr hind, "the Indian 
 date;" cf. bos lucanus (Weber, Indische Skizzen, p. 74). The nouns in 
 -της and -τνς (§267) are strictly of participial origin; they are in fact 
 connected with the verbals in -τεος. These verbals very nearly corre- 
 spond in meaning to the Latin gerundiva in -ndus, which are merely 
 lengthened forms of the participle in -nts (§ 265). The idea of action 
 is explicitly conveyed by the nouns in -ττις, -τνς, which denote a doer 
 and a doing respectively. It has not been sufficiently remarked that 
 from the longer form -τηρ, -τωρ , into which the ending -της is occa- 
 sionally expanded, the Romans formed not only an active future par- 
 
CHAP, v.] THE ADJECTIVE. 505 
 
 ticiple, but also an abstract noun equivalent in meaning to those 
 ending in -Tvg. Thus, if scrip-tus meant "a writing" (cf. the supine 
 scrip-turn, "to write"), scrip-tor meant "a writer," scrip-turus, "a 
 person about to write," and scrip-tura again, "a writing." We can 
 scarcely imagine a more interesting subject of speculation to the lin- 
 guistic philosopher than that which is presented by this class of 
 words. It enables us to see how from the idea of proximity conveyed 
 by the second pronominal element (-tyg^ -tlq^ 6ig) that of emanation 
 is at once derived (-di]g, -Q'sv), and how we pass from this to an expres- 
 sion of agency and an agent (-Tr]g, -ttjq, -τωρ). Then again we observe 
 how the notion of continuity, which is involved in that of agency, 
 connects itself with that of futurity ; whence the continuous tense in 
 Hebrew is so often a future (Maskil le Sopher, pp. 23 sqq.); and from 
 this we get the idea of obligation, or the feeling that the performance 
 of an act is continuously incumbent upon us (-reog). It is only by 
 considering the matter thus that we can understand the concidence 
 in meaning between the first and last term of the series -tus, -Tvg, 
 -TBog, -Zf^g, -ζηρ, τωρ , -tor, -tura. The Latin and Greek verbals in 
 -tus^ -xvg strictly correspond in meaning to one another and to the 
 infinitive. In Greek, the adjective in -τεο^, derived from -rvg, corre- 
 sponds in meaning to the adjective in -ndus derived from the par- 
 ticiple in -nts (above, § 265). But the latter is strictly equivalent in 
 meaning to the infinitive active, of wloich the so-called gerunds are 
 only inflected cases (see Varronianus, pp. 361 sqq.). Every thing 
 therefore tends to confirm the opinion that these expressions all sjpring 
 from a common metaphysical origin. 
 
 297 There are instances in Greek where scholars are still uncer- 
 tain, whether a particular substantive use or a general adjective use 
 is intended. We will select one or two which affect the proper inter- 
 pretation of certain passages in the Greek poets. 
 
 In Pindar, 01. ii. 51, we read: το τυχεΐν ττειρώμανον aymviag jta- 
 qalvBL δυΰφρόνων , which must mean "to succeed when one makes 
 trial of, engages in, a public competition, sets a man free from gloomy 
 thoughts" (Schol.rwv δνΰκόλων φροντίδοιν των επί rjj vlxjj άηολνει)^ 
 so that ^v(^φρ6vωv^iho\xφ. in form an adjective, is equivalent to a sub- 
 stantive. Dindorf, whose remarks (Steph. Thesaur. ii. 1801 b, s. v. 
 δνΰφρονή) have been borrowed without acknowledgment by the Eton 
 editor of Pindar, reads δνΰφρονάν, because "non apparet quomodo 
 δνΰφρονα Λρο δνόφρούνναι vel φρovτίδεg dici potuerit." He quotes 
 Hesiod, Theog. 102: αΐ-φ' ογε δνΰφρονεων εταλτι^εται ονδε τι. κήδεων 
 μεμνηταί^ as another example of the feminine form δνύφρονή ; but here 
 δνΰφρονεων is the participle, and if Hesiod had intended to use the 
 
506 THE ADJECTIVE. [BOOK III. 
 
 noun, lie would have said δνβφροΰννη, as in Theog. 528 : xai ελνΰατο 
 δυΰφροΰννάων. That the neuter δνΰφρονα may be used as a noun 
 equivalent to δνΰφροΰννη is proved not only by ^sch. Choeph. 272: 
 τα μεν γαρ εκ γης δνΰφρόνων μειλίγματα βροτοΐς^ " the fruits of the 
 earth which soothe gloomy thoughts for mortals," i.e. drive away dull 
 care; but also by the converse use of ενφρονα for ενφροόννη in the 
 Eumen. 602: τα Λλεΐϋτ αμεινον εϋφροΰιν δεδεγμένη (see note on 
 Pind. Nem. v. 38). And the sentiment that success drives away gloomy 
 thoughts is echoed in another passage of Pindar, Nem. iv. 1, where it 
 is said that the merriment of the banquet is the best physician for 
 labours when they are brought to a decision {αρνΰτος ενφροΰύνα 
 Ίίόνων κεκριμενων ϊατρος). 
 
 The word γεραρός is in Homer an adjective , and an epithet of 
 honour. Thus Priam, describing Agamemnon {II. iii. 170), says: 
 
 καλόν d' οντω εγών οϋπω ϊδον οφΟ'αλμοΐΰίν, 
 ονδ^ οϋτω γεραρόν βαβιληϊ γαρ άνδρΐ εοικε. 
 
 and Antenor, contrasting Ulysses with Menelaus (//. iii. 211), says: 
 
 βτάντων μεν Μενέλαος ϋτίείρεχεν ευρεας ώμους, 
 αμφω δ' εξομένω, γεραρώτερος ψν Όδνβΰενς. 
 
 The Scholiast's interpretation of the former is εντιμον , of the latter 
 εντιμότερος τίρος otjJLVj and we have no doubt he is right; γέρας, γέ- 
 ρων, and γερήνίος or γερην are explained in the same way by He- 
 sychius: γέρων. έπΙ μεν του εντίμου — κίκληΰκεν δε γέροντας άρι- 
 ^τηας (Iliad u. 404), γέρας γαρ η τυμη' γερήνοος έντιμος, γέρων 
 γέρην, έντιμος. Γεραρός is formed by the common suffix -ρό^ from 
 the word γέρας, "the privilege or peculiar gift of a person in autho- 
 rity," — e.g. the first share of the booty and so forth — especially "the 
 hereditary privileges and prerogatives of a king : " in which sense it 
 was equivalent to έρανος (see Welcker, Trilog. p. 381, note). Hence 
 Thucydides says of the old kings of Greece (i. 13): τίρότερον δε ηΰαν 
 επί ρητοΐς γέραύι πατρικοί βαΰιλεΐαι: accordingly γέροντ-ς (γέρων) 
 was a person holding such privileges or authority, and γερόντια or 
 γερουύία was the name given to the ruling Senate at Sparta. The root 
 of the word is ^n-, "to take" or "receive," one of common occurrence 
 in all the languages of the Indo-Germanic family, and probably the 
 same with Λη, of which we have said so much in the last chapter (see 
 also p. 298). We do not believe that it is connected with γηρας^ "old 
 age," which seems to be related to the Sanscrit root jri (jard)^ "to 
 wear away;" and the use oi γέρων, γεραιός, as an apparent synonym 
 for γηραιός, is to be explained from the connexion of the ideas of age 
 and dignity in the Greek mind. So that Homer says, very appositely 
 for our purpose (iv. 323): 
 
CHAP, v.] THE ADJECTIVE. 507 
 
 κελενΰω 
 βουλ^ καΐ μν%οι0ΐ' το yccQ γέρας εΰτϊ γερόντων. 
 In Euripides (Suppl. 42), 
 
 ικετεύω βε-, γερανά, γεραρών εκ 
 
 ΰτομάτων προς γόνν πίπτουύα το ΰόν, 
 we beKeve that the idea intended to be conveyed by the words γεραιά 
 and γεραρών is not that of age, though both -^thra the person ad- 
 dressed, and the chorus who are speaking, are represented as old 
 women, but that of veneration or respect: "I beseech you, hon- 
 oured dame, with a mouth paying you due respect, and falling down 
 at your knee." As γεραρός is used as an epithet of the person who 
 receives the γέρας, so γεραίρειν is employed to denote the act of be- 
 stowing the γέρας, as when a greater share of any thing is given to a 
 distinguished man (Homer, Iliad vn. 321): 
 
 νώτοιΰί δ' Αϊαντα δίηνεκέεΰΰί γέραιρεν 
 
 Ύίρως Άτρείδης. 
 Or when the worship of a divinity is spoken of (Plato, Legg. vir. 
 799 a): χορείαις τίοίαιΰι γεραίρειν την τότε %ν6ίαν. There is no 
 doubt, then, as to the meaning of the adjective or epithet γεραρός. 
 But there are two passages of ^schylus in which this word is clearly 
 used as a substantive, and in both we have a dative plural. They 
 are, SuppliceSj 726: 
 
 καΐ γεραροΐΰϋ ττρεΰ- 
 βντοδόκου γεμόν- 
 των %νμελαι φλεγόντων θ' 
 ώ^ τίόλις εν νέμοιτο. 
 and Agamemnon.) 722: 
 
 ε%'ρε'ψεν δε λέοντα 
 ΰίνιν δόμους άγάλακτον 
 οϋτως άνηρ φυλόμαότον, 
 εν βίότον τίροτελείοις 
 αμερον.) ενφιλότίαιδα, 
 καϊ γεραροΐς ετίίχαρτον. 
 
 In consequence of the word ηρεββντοδόκοι, which precedes in the first 
 passage, and ενφιλότίαιδα in the second, every one has been led, not 
 unnaturally it must be confessed, to translate γεραροϊς in both pas- 
 sages "the aged men." A little examination will show that the word 
 is in both passages a synonym for γέραΰι. That γεραροΐς does not 
 mean "old men" in the first passage is clear; for how can altars, or 
 rather the terraces round the altar (^νμέλαι: see Muller, Anhang zu 
 den Eumeniden, p. 35) be said to be loaded like a ship with freight 
 
508 THE ADJECTIVE. [bOOK III. 
 
 (γεμΕίν), arid to blaze (φλέγεον) with old menf That old men crowded 
 round the altars is sufficiently stated by the epithet τΐρεΰβντοδόκοί: 
 and the addition of a synonym for τΐρεΰβνταΐς would be very unlike 
 ^schylus. That γεραροΐς does not signify persons in the second pas- 
 sage is clear from this, that when ετΐίχαρτοζ governs a dative case of 
 the person it always has the meaning "rejoiced over as by an enemy:" 
 thus ^schyl. Prometh. 164: IxxtQolg επίχαρτα τίεπον^α (see the pas- 
 sages quoted in Blomfield's note); Thucydides, m. 67: οΧκτον δε άξίώ- 
 τεροί τνγχάνειν οΓ άτιρεπεξ τι τΐάΰχοντεξ των άνΟ-ρώτΐων' οί δε δι- 
 καίως, ωότίερ οΓάε, τα εναντία εττίχαρτοι είναι,; and so επιχαίρω, 
 when it governs the dative, as in Soph. 4/. 940: oi δ' ovv γελώντων 
 κατίίχαυρόντων κακοΐς ; but when the verb governs the accusative, it 
 expresses merely a symple act of joy, like χαίρω, γη^ώ, &c. with the 
 same government (see Sophocl. Aj. 136, andLobeck), and so ετίίχαρτος, 
 without a dative of the person, signifies "joy-causing, agreeable," as in 
 Sophocl. Trachin. 1262: ώ^ ετίίχαρτον τελεονό* άεκονΰων έργον, 
 Avhere the Scholiast rightly compares the Homeric εκών άέκοντί γε 
 %νμω. If, therefore, in the passage of ihQ Agamemnon we take γεραροϊζ 
 in the sense "by means of gifts," we shall have the natural significa- 
 tion, "the young lion was tame, it gambolled with children, and was 
 agreeable or pleasing, it caressed any one — when it was fed," just as 
 he says afterwards (725): φαιδρωτΐος ποτΐ χείρα βαίνων τε γα(5τροξ 
 άνάγκαΐζ. That γεραροΐζ, in the other passage, means "with sacrificial 
 offerings," is shown by the use of the word φλέγευν (so in the Aga- 
 memnon, 91: βωμοϊ δώροίβί φλέγοντ ai), and by the imitation of 
 the whole passage in the Electra of Euripides (720 and following): 
 
 χοροί δ^ Άτρευδαν εγεραιρον οϊκονξ' 
 ^ ν με λ αν δ' επίτναντο χρν- 
 ΰήλατοί, ΰελαγεΐτο δ' αν αΰτν 
 τίνρ εταβώμυον^Αργείων. 
 
 We conclude, therefore, that in these passages γεραροΐξ represents the 
 dative plural of γεραρόν , which has become permanent as a neuter 
 substantive equivalent to γέρας just as the still longer, but analogous, 
 adjective λντήριος is used by Pindar (Pyth. v. 99) in the neuter, as a 
 synonym for λιηρον (see below, § 304). 
 
 298 We have stated that the etymological distinction between 
 the adjective and substantive is, that the former is generally capable 
 by its inflexions of being attached to substantives of every gender. 
 It will be easy to show that this sole etymological difference is the 
 result of the syntactical use of adjectives. A great number of pos- 
 sessive adjectives are nothing more than genitive cases attracted by 
 juxtaposition into a variety of inflexions. For instance, if, as is most 
 
CHAP, v.] THE ADJECTIVE. 509 
 
 probable, an older form of the genitive of δήμος, δημοίΟ, was δημόΰω, 
 what is this in relation to δημόΰίος, but the crude-form of a new 
 system of inflexions? The same may be said of the comparative in 
 -iG)V=-LOV-g, which refer to the still older genitive in -L0V=-6i0V 
 (§ 165), and the weakened genitive may still be recognised in such 
 adjectives as χρνύεος from χρνΰός. Mr. Garnett, who has referred to 
 this principle in his instructive paper " On the Formation of Words, 
 by the further Modification of Inflected Cases" (Proceed, of Phil. 
 Soc. Vol. III. pp. 9 sqq.; Essays, pp. 260 sqq.), seems to have over- 
 looked the distinction between those nouns which are formed from 
 oblique cases, by the mere appendage of a new system of inflexions, 
 and a different class of secondary structures , which affix to the new 
 crude-form the pronominal terminations enumerated in a preceding 
 chapter. Thus, it is plain to see on the one hand, that δημό-βίο-ς is 
 merely the genitive δημό-<3ίθ made the vehicle of a new set of case- 
 endings, and that %QVoeog, χοόνεος, &c. are similarly derived from 
 weaker forms of the genitive. But it is equally clear, on the other 
 hand, that a form like ϊφίος contains something more than an oblique 
 case and a new system of case-endings; and a comparison οίΊφί-κλης, 
 Οί-λενς, &c. would lead us to doubt whether the first part is to be re- 
 garded as merely the dative of ϊς. The same remark applies to the 
 forms ημερηΰίος and ^^artog, which Mr. Garnett would derive from the 
 datives ημερ^ΰι and ημαη, but which appear to us to be formed from 
 the nominatives ήμερη and ηματ^ημαρτ,ΐογ the addition of the affixes 
 -6Log and -Log (above, § 254). Nor does there seem to be any necessity 
 for deriving οικείος from οϊκο or οϊκει, when the appendage of tog to 
 the regular crude-form οΙκ{11) of οϊκος suggests itself at once. With 
 regard to forms like βίαιος, we should compare them with liSa-iog, 
 τριτα-ΐος,&ο. from 1(97; [μοΐρά],τρίτη [ημέρα], &c. j and though, accord- 
 ing to the principle laid down above (§ 165), the comparatives μεβαί- 
 τερος, Ιδιαί-τερος, Ιΰαί-τερος, &c. presume a derivation from the loca- 
 tive adverbs μεβτ}, Ιδία, Yoy, &c., we must not apply this to the very 
 different case formations in -ιος from the nominative in Ύΐ^^ -ya (above, 
 § 254). This is still farther indicated by the accent of τίαλαιός, com- 
 pared with that of Ιύαϊος, &c., for this shows that the affix, by which 
 the adjective was formed from τίάλαυ, contained at least an additional 
 L, SO that 7ίαλαι-ός=^παλαι-ί6ς=^7ίαλαυ-κός. There seems to us to be 
 the same objection to Mr. Garnett's theory respecting the derivation 
 of the participle from an ablative of the verbal root. It has been 
 shown above that the ablative sign was something more than a mere 
 -r, that it was in fact the affix -6iOV, which appears as -^εν, -χΙ^ην, -%a, 
 -δα. Now it is clear that this, like other forms of the genitive, may 
 become the vehicle of a new set of case-endings , as we see in patro- 
 
510 THE ADJECTIVE. [bOOK III. 
 
 nymics in -δης, and in the forms in -OLOg, &c. But we have no such 
 forms in the participles, which merely interpose -/or- or -j;r- between 
 the root of the verb and the case-endings. The adverbs in -doVj -δην, 
 -da are in fact cases of the participles in -ντ=νδ, or of the verbal 
 nouns in δ- ; and it is not consistent with sound philology to suppose 
 that a mere crude-form is equivalent to a case formed upon it. The 
 ablative forms of currendo and cursu do not prove the ablative nature 
 of currens, which is really equivalent to currendus (§ 265). 
 
 There are many adjectives which are immovable or not capable 
 of a variety of flexion (Lobeck, Paralipom. p. 189), and these, when 
 placed by the side of the noun- substantive, constitute it to all intents 
 and purposes one of those compound words in which the genius of 
 the Greek and Sanscrit languages is most strikingly developed, the 
 only difference being, that in the former instance the inflexions of 
 case are preserved, while those of gender are neglected; whereas in 
 the latter the crude-form only is prefixed. Most adjectives, however, 
 vary in gender, number, and case with the noun to which they belong, 
 and are either the predicate of the sentence of which the noun is the 
 subject, or stand as the representative of some case (mostly the geni- 
 tive) of another noun dependent upon the substantive in question, just 
 as, conversely, the genitive case of a substantive may stand as a sub- 
 stitute for an epithet; see Soph. Antig. 114: χιόνος πτερνγυ for χίο- 
 νέη-, Electra, 19: αΰτρων ενφρονή for άΰτερόεΰϋα, &c. In much the 
 same way we have. Soph. Antig. 795: εναργής βλεφάρων ίμερος. 
 
 299 From this it will be seen that the use of adjective at all is 
 entirely logical or syntactical, for an adjective when distinct from the 
 substantive, that is, when not a mere epithet, is a predicate as con- 
 trasted with the subject (above, § 124). Although we do not profess 
 to discuss the syntax of the Greek language in these pages, yet as the 
 very essence of the adjective is syntactical (its etymological distinction 
 being merely an accident), and as its different value, according as it is 
 used with or without the article, is a point which teachers of Greek 
 find it most difficult to impress upon their pupils, we shall indulge in 
 some remarks on the subject. 
 
 300 Every noun in the Greek language, however vague and 
 general its signification may be, is capable of forming the subject of 
 a proposition, if it only has prefixed to it that simplest form of the 
 demonstrative, which we call the article; and if two nouns occur in 
 connexion , one of which has , and the other wants , the article , the 
 former is related to the latter as subject is to predicate. This is 
 a fact which must be always kept in mind by every one who would 
 
CHAP, v.] THE ADJECTIVE. 511 
 
 translate the Greek authors correctly. An adjective, in our accepta- 
 tion of the name, is an epithet of the substantive with which it is 
 joined, and the difficulty which we experience in practically teaching 
 the Greek language is, to convince the learner, that no noun, though 
 with variable gender, can be considered as an adjective when it stands 
 alone, while the substantive to which it refers has the article prefixed, 
 but that it is always a predicate , or asserts something of the noun, 
 even though they should both of them be in oblique cases. Another 
 assertion of the same principle is , that no participle or verbal adjec- 
 tive in -vtj -/or-, or -μένος, can be considered as an adjective, unless it 
 be subjoined to an article, in which case it is equivalent to the relative 
 pronoun with a finite verb and a definite antecedent. Etymologically 
 considered, the article, the relative, and the affiLx of the genitive case, 
 are equally derived from the pronoun denoting proximity (§ 148), 
 and the definiteness which is implied by all three is due to the ex- 
 pression of nearness. The difference between the relative and the 
 definite article consists, as we have already seen (§ 243), in this, that 
 the former preserves the subjective form throughout all cases and 
 genders, and has the case-ending in the nominative, whereas the case- 
 ending s of the nominative is wanting in the article, because that 
 suffix is appended to the noun with which it is connected, and the 
 element ta is used for the neuter and for the oblique cases of the' 
 itiasculine and feminine. Thus we see that the relative is especially 
 subjective, or that its function, when it has a definite antecedent, is 
 to express by periphrasis a definition or epithet as distinguished from 
 a predication or assertion. And this view is not to be qualified by 
 the fact that with an indefinite antecedent the relative sentence 
 becomes hypothetical or virtually adverbial. Considered with refer- 
 ence to its origin, the relative sentence is necessarily definitive or 
 descriptive , and its own true nature is not affected by the occasional 
 circumstance that the antecedent, to which it stands in an adjectival 
 relation , is an adverbial or general term : so that the definition is 
 like all hypotheses, in that case , of a frequentative nature. Thus, if 
 we say, Σωκράτης, ος ταϋτα εδίδαόκε, άγαΟ'ος ην, we make a single 
 predication of an individual specially defined ; but if we say, ος αν 
 ταντα didaoKjj, άγαΟ'ον νομίζω, we repeat the predication of all who 
 may answer to the definition, which is, in itself, as much a definition 
 as before. Now the Greek participle active, whether it ends in vt=vd 
 or in -/or-, exhibits the second pronominal element, and implies the 
 mere action of the verb (above, §§ 263, 265), which a set of adjectival 
 inflexions render applicable to its association with a given noun as 
 attribute or predicate. Similarly, the participle passive exhibits the 
 combination of the first element with the third in the form which 
 
512 THE ADJECTIVE. [bOOK III. 
 
 expresses the result of the verbs active (§ 410, (1)), and the adjectival 
 inflexions render it available for the same purposes as the active par- 
 ticiple. By itself, however, the participle, whether active or passive, 
 defines nothing. It is a mere assumption of the agency implied , and 
 is equivalent to a conditional sentence, or, what is the same thing, to 
 a relative sentence with an indefinite antecedent. Thus τντίτων^τνη•- 
 τοντ-ξ means "a person striking, if there is one"="if any one 
 strikes "=£l'rtg τνΛτΗ=^οξ civ TVTtTjj. But both the verb and the par- 
 ticiple constitute the subject of the sentence , when we prefix to the 
 former the relative pronoun with a definite antecedent: and to the 
 latter the definite article alone. Thus, sKslvog og δίδω-Τί, "he who 
 gives"=o δίδόντ-ς {δίδον£), "the giver;" and hence it is that the 
 epithet or a'djective in the proper sense of the term is equivalent either 
 to the verb with the relative, or to the participle with the article; 
 for δ ayad'b-g av^ρω7ΐo-g is identical in signification with both ό 
 ccv^'Qω7Co-g o-g ccyad'o-g Ιβτιν and ό av^Qω7to-g^ 6 ευ Ttoiyo-vt-g 
 (ποιών). So that, in fact, the adjective, which, from its variety of 
 flexion, requires the guidance of an article, before it can be regarded 
 either as an epithet or as a substantive, that is to say, before it can 
 be used as a subject, is to be considered in the same light with the 
 participle, which differs from the verb only in having variable in- 
 flexions. Whereas, conversely, when the adjective and participle stand 
 after an article, and unconnected with any substantive, they are sul^ 
 stantives to all intents and purposes , for their variation of gender is 
 excluded by the nature of the case; and adjectives or participles which 
 have been long used in this way may become regular nouns appella- 
 tive, like the adjectives in -κή^ or the word γεραρόν^ mentioned above, 
 or even proper names , like Χάρων, KaXyjag, and the patronymics 
 referred to in the last chapter (above, § 267). 
 
 301 An application of these principles will enable us to classify 
 and explain all the various uses of the Greek participle and adjective. 
 "We have seen that the etymological difference between the adjective 
 and substantive is limited to the parallelism of inflexion to which the 
 former and the participle are generally liable: and we have shown 
 generally that this is a formative process arising from the contrivances 
 of syntax. The main business of syntax, as the handmaid of logic, is 
 to distinguish accurately between the subject and the predicate. Now 
 we have endeavoured to show in another place {Greek Grammar^ Arts. 
 400, 417), that there are three different kinds or classes of predicates, 
 which we have termed (A) Primary , when there is nothing between 
 the subject and the predicate, except the copula, either expressed or 
 implied; thus, in the phrase oi λόγοι '^pεvδεΐg είΰίν, "the Avords are 
 
CHAP, v.] THE ADJECTIVE. 513 
 
 false," the adjective ιΙ^ενδεΐς is a primanj predicate ; (B) Secondary, 
 when the predicate is connected with the subject through a verb, 
 which already contains a primary predicate ; thus , in the phrase oi 
 λόγου ψενδεΐζ ελεχ^ηβαν, '^the words were spoken, and they were false 
 =:= the words which were spoken were false," two circumstances are 
 predicated of λόγοι, first, the utterance, and, secondly, the falsity; 
 consequently, -^ενδείξ^^ a secondary predicate; (C) Tertiary, when in 
 the second case there is also a Λρόληιρος, or anticipation of a distinct 
 predication of something additional ; in other words , when the whole 
 of the secondary predication is subordinated to a primary predication, 
 which refers to a different subject; thus, in the phrase δ [lavTig rovg 
 λόγους Ίρενδεΐς λέγει (Soph. (Ed. Τ. 526), "the prophet speaks words, 
 and they are false "=" the words, which the prophet speaks, are false," 
 we have the secondary predication ot λόγοΰ 'ψευδείς λέγονται attached, 
 by τΐρόλη'ψις, to the primary predication δ μάνης λέγει, i. e. έύτΐ λέ- 
 γων, "the prophet is speaking." Now the vehicle of this τίρόλ'η'φις is 
 the accusative case λόγους ; and oblique cases of nouns, as predicating 
 specially some secondary relation, are themselves secondary predicates. 
 Therefore, the τΐρόλη^Ρις is rightly termed a tertiary predicate, and 
 this τΐρόλη^ρις is contained in the adjective 'ψευδείς, here used in the 
 accusative case. 
 
 302 From this we see, that all three classes of predicates may be 
 expressed by the adjective, according to different syntactical usages; 
 but a primary predicate might be expressed not only by an unappro- 
 priated adjective or participle, but also by an attributive or general 
 substantive, as μάντις 'ην 6 Κάλχας, and by a finite verb considered 
 as including a participle, as δ μάνχις λέγει^=έΰύ λέγων. Again, the 
 secondary predicate might be expressed not only by an adjective in 
 the nominative case, as in the instance given above, but by a similar 
 use of the substantive, as in the line of Homer (7Z. ii. 673): Νιρενς, 
 ος κάλλιΰτος άν^ρ υτίο'Ίλιον 7]λ^εν, where it is predicated of Nireus, 
 not only that he went to Troy, but also, which is the main point, 
 though the secondary predication, that he was the handsomest man 
 among those who went thither. And not only have we the nominative 
 with verbs which admit of this apposition. The oblique cases of nouns 
 are used with all classes of verbs to convey the idea of a secondary pre- 
 dicate; and we have seen that the tertiary predicate is a τΐρόλη'ψις 
 springing out of this usage. If we say for example, δ Σωκράτης έχει, 
 we have a primary predication; for we speak of Socrates as having; 
 but the sentence is incomplete ; because the transitive verb conveys 
 no definite meaning without an expression of the object. When 
 therefore we add the accusative case 'ψυχ'ήν, we state what it is that 
 
 L L 
 
514 THE ADJECTIVE [bOOK III. 
 
 Socrates possesses, and, however little we may be in the habit of 
 regarding the fact from this point of view, we add a fresh predication; 
 for the sentence, δ Σωκρκτη£ έχει, ^υχψ, is really equivalent to the 
 two sentences, "Socrates is possessed of something, and the thing 
 which he possesses is a soul." The accusative, then, is like the other 
 oblique cases, an abverbium or εηίρρημα properly so called ; i. e. it de- 
 rives its significance from and through the verb by which it stands, 
 or by which it is governed, as the phrase is. The particles which we 
 term adverbs are merely oblique cases of nouns, pronouns, or adjec- 
 tives, which express generally the time, place, cause, form, or manner 
 of an action. It is this generality of reference which constitutes the 
 distinction between the adverb and the oblique case of a noun. The 
 latter implies a special object; the former is applicable to all objects. 
 Thus we may say with reference to different objects, ό TCalg τίατάύβει 
 τον ovov, or xov τιννα j but we may add the expression of manner or 
 degree to any such special statement; as τΐατάόΰευ ίΰχνρώς^ or αφρον- 
 τίύτως. So also, a dative case expressing the instrument may occa- 
 sionally become attached to the verb in such a way as to complete its 
 signification, or to take the place of a merely adverbial adjunct ; thus 
 κτείνεί ξίφεν is virtually one word, quite as much as ξιφοκτονεΐ. The 
 Greek idiomatic usage will enable us to explain this satisfactorily with 
 reference to the predication quoted above. The verb εχω is constantly 
 used in a neuter sense, or the reflexive pronoun is dispensed with, 
 whenever this verb is construed with εν or any adverb in -mg. Thus, 
 it is a complete sentence if we say, ό Σωκράτης εχεί καλώξ-, "Socrates 
 is (= has himself) well," Socrates bene se Jiabet. To this phrase we 
 may add the specification of a particular object, which Λνϋΐ be ex- 
 pressed by the genitive or accusative, according to the reference 
 intended. If we say, δ Σωκράτης καλώς εχευ την "φνχήν, we give the 
 immediate object of the verb: "Socrates has his soul well." But if 
 we say, δ Σωκράτης καλώς έχει, της 'ψνχης, we mean that "as to his 
 soul, he is well," where the sentence is doubly adverbial. The former 
 of these phrases may be expressed equally well by the commonest form 
 of the τΐρόλη'φίς or tertiary predicate. Thus, if we affirmed the jocular 
 hypothesis, which the philosopher is represented as making in his 
 argument with CalHcles (Plato, Gorgias, p. 486 d), we might say, δ 
 Σωκράτης χρνΰην είχε την 'φνχήν, which would imply, not only that 
 "Socrates had a soul," but also that "the soul which he had was 
 golden." This tertiary predication is particularly common in those 
 cases in which the secondary predication is assumed in the very terms 
 of the expression. For instance, we might assume that "Socrates had 
 a soul/' or that "a prophet, if he speaks at all, speaks words." But 
 the same assumption in regard to the objective case is equally obvious 
 
CHAP, v.] THE ADJECTIVE. 515 
 
 in those passages in which the unwary student is most liable to con- 
 vert the ulterior predication into an epithet. Thus, in the description 
 of the lines around Platsea , Thucydides speaks of the circumvallation 
 itself as something known and assumed, but he finds it necessary to 
 state that the lines were double, one wall being intended to check the 
 sallies of the besieged , the other to resist the attacks of a relieving 
 army. He says, therefore (iii. 21): το τεΐχοζεϊχε δύο τονς περιβόλους, 
 which most readers would be contented to translate "the wall had 
 two circles," but which must mean that 'Hhe circles, which — as a 
 metter of course — it had, were two in number." 
 
 303 As all additional references in a proposition are connected 
 with the subject through the root, it is clear that they are all adverbs 
 or secondary predicates, if they are in the same case with the subject, 
 but tertiary predicates , if they are connected with some object of the 
 verb, which is itself a secondary predicate. It is also clear that a 
 verb may be the vehicle of any number of such additional and acces- 
 sary statements. Thus we might ask not only "how Socrates does," 
 but "how he does as to body," or "how he does in regard to health" 
 (Plato, Georg. p. 514 d), and we might combine the answer to these two 
 questions in one proposition : %αλώξ έχει δ Σωκράτης το ΰώμα TtQog 
 νγίευαν, where it is clear that the verb έχει is assisted by three ad- 
 verbs or adverbial phrases expressing the manner^ the object, and the 
 relations of the existing state of Socrates. To return then to our 
 immediate object — the adjective: — as on the one hand, there are 
 adjectives and participles, which have fixed themselves in use as sub- 
 stantives, so on the other hand there are many words with movable 
 inflexions, which have a confirmed tendency towards an adverbial 
 usage; and some of them are then used specially and in a different 
 sense from that which they bear as epithets. Such are the pronouns 
 and adjectives which denote separation, locality, quantity, &c. ; e. g. 
 amog, μόνος, μέΰος, πάς, άλλος, εκαύτος, &c. Not to trouble our- 
 selves with a discussion of the usages of all these words, which would 
 be necessary in a more elementary treatise , it will be sufficient if we 
 take αυτός as an example of the principle to which we refer. It is 
 well known that if αυτός has the article it is merely definitive — in fact, 
 it merely strengthens the article. By a little emphasis we can make 
 "ίΛβ man," δ άνήρ, equivalent to "the same man," δ αυτός ανηρ. 
 When αυτός stands by itself and in an oblique case without the ar- 
 ticle , it is the pronoun of unemphatic reference , like the Latin is or 
 the Hebrew affix η or ίη-. Thus 7] γυνή αϋτον is perfectly equivalent 
 to uxor ejus or nnpN. But if αυτός stands by the side of a noun 
 already defined, and is not itself, by means of the article , included 
 
 LL2 
 
516 THE ADJECTIVE. [bOOK IIL 
 
 in the definition, it becomes adverbial, or serves as a secondary pre- 
 dicate; thus υ avTjQ avrog means "the man considered by himself," or 
 "alone." The full force of this adverbial usage is perhaps nowhere 
 so clearly seen as in the idiomatic employment of the dative plural to 
 signify a collective accompaniment. In such phrases as jtsvTS vavg 
 ίλαβον, και μίαν τούτων αντοΐς ανδράόιν (Thucyd. ιν. 14), "they 
 took five ships, and one of these together with its whole crew" or 
 "men and all," we see that the adjunct αντοΐξ άνδράΰιν is as much 
 a secondary statement as if we had said, in a distinct proposition, καΐ 
 ot άνδρες εληφ&ηΰαν ώΰαντως. 
 
 304 Although all this is obvious enough when stated plainly 
 and directly, and though the fact must be known to every one who 
 has any pretension to the name of a Greeek scholar, it is, as we have 
 before said, very difficult to impress these distinctions upon the young 
 student ; we shall , therefore, make no apology for showing by a few 
 examples the application of the principle to the commonest con- 
 structions in Greek. We feel the more justified in doing so as even 
 the most eminent scholars have occasionally fallen into the mistake 
 of confusing the epithet with the secondary or tertiary predicate. 
 These errors may be divided into three classes, (a) When the com- 
 mentator has mistranslated the existing text. (b) When a true 
 reading is altered from a misapprehension of the construction, (c) 
 When, for the same reason, a corruption is left in the text, (a) The 
 most singular blunder of this sort is that which Brunck , Blomfield, 
 and Wellauer have committed in construing της ενΛραξίας ΰωτηρος 
 in ^schyl. Sept. c. Theb. 309 : 
 
 Πειθαρχία γάρ εβτι της Εντίραξίας 
 μήτηρ, γννη Σωτηρος, ώδ' εχευ λόγος, 
 
 the meaning of which clearly is " Obedience is the mother of good 
 fortune, and the wife of Jove the Saviour; such is the saying." The 
 last part of the λόγος we find in another form in Soph. Antig. 676: 
 
 των δ' όρΟ^ονμένων 
 ΰώξει τα τΐολλα όώμα^'^ η Πειχ^αρία. 
 
 Hermann, in his edition of Aristotle's Poetic, c. iv. § 16, where we 
 have %al τον λόγον τίρωταγωνιύτην τϋαρεΰκεύαβε, writes as follows 
 (p. 109) ' Sextam tragcedm formam instituit u:Eschylus, secundo addito 
 actore, unde primarum partium actor exstitit, quern Aristoteles λόγον 
 τΐρωταγωνίΰτην vocat, male a Twiningio et Buhlio intellectum. So that 
 he makes τίρωταγωνιύτήν an epithet, whereas it is a predicate, as it 
 is correctly rendered by Twining ("he made the dialogue the principal 
 
CHAP, v.] THE ADJECTIVE. 517 
 
 part of Tragedy"), whose interpretation is adopted by Buhle. On 
 Pindar, Pyth. v. 99, το κακλίηκον λντηριον δατίαναν μέλος χαρίεν, 
 Bockh writes as follows: "junge μέλος καλλίνίκον χαρίεν λντήρίον δα- 
 Tiavav : καλλίνυκον est adjectivum ad μέλος, ut Nem. iv. 1 6, atque etiam 
 λνττί^ρίον δατίαναν adjectivum est ; " and his construction is adopted by 
 Dissen. But', as we have shown in our note on the passage, το λντή- 
 QLOV is put for λντρον, so that το λντήρων δατίαναν is analogous to 
 λντρον καμάτων (Isthm. vii. 1), and μελού χαρίεν is an explanatory 
 apposition: "the triumphal guerdon of his costs, a sweet song." We 
 might expect to find examples of the same inadvertence in Dr. Arnold's 
 notes on Thucydides, for perfectly accurate scholarship was not one 
 of the many excellences of that great teacher. Thus in iv. 86 , he 
 translates: ονδε άΰαφη την ελεν^ερίαν νομίζω ετίυφέρείν, ''nor am I 
 minded to offer you a dim and doubtful liberty," just as he had rendered 
 the parallel passage in the preceding chapter : αδυκον την ελεν^ερίαν 
 Ιτίίφερειν^ "I shall be charged with offering you a false liberty," al- 
 though the preceding passage : την αΐτίαν ονχ εζω Λΐότην άτίοδει- 
 κννναι is correctly given in his version: "the reason of your not joining 
 me I shall never be able to make out to men's satisfaction ;" for it is 
 clear that in all three cases there is a prolepsis or tertiary predication 
 — "no one will believe the alleged reason;" "the freedom which I 
 offer will be thought a cloke for meditated injustice;" "I do not 
 think that the freedom which I offer need be the cause of any misap- 
 prehension." But we are more surprised to find an instance of simi- 
 lar carelessness in Dr. C. Wordsworth, who might have been expected 
 to inherit a special regard for the position of the Greek article. In 
 his Athens and Attica, p. 180, we find an extract from Philostratus 
 {Vita Herodis Soph, 11): κάκεΐνα τίερϊ των Παναθηναίων τούτων 
 ηκονον, τίέτίλον μεν ανηφθαι της νεώς ηδίω γραφής ύνν ονρίω τω 
 κόλτΐω, which is thus translated: "I have heard this description of the 
 Panathenaic festival: they "tell me that a Peplus, more lovely than a 
 picture, was hung from the ship wafted by its swelling bosom." Now 
 it is clear from the passage that the ship was not wafted by the sail, 
 but moved by machinery on the ground (ετίίγείοις μηχαναΐς) ; conse- 
 quently, it was necessary to predicate of the Peplus that it was artifi- 
 cially distended, as if filled by a favourable wind; and Dr. Wordsworth's 
 translation, which substitutes an epithet for the predicate, conveys no 
 such idea, (b) In Longinns de sublimitate, § viii. we find that the first 
 and most important of the five sources of sublimity is described as το 
 Τίερϊ τ ας νοήύείς άδρετίηβολον, where the correction αδρεπίβολον is 
 fully justified by the parallel cases of ενετίίβολος and μεγαλεπίβολος 
 (see Dindorf ad Steph. Thes. iii. 1502). Instead of this Kuhnken has 
 not hesitated to propose an unintelligible soloecism. He says: *'Lon- 
 
518 THE ADJECTIVE. [bOOK III. 
 
 ginus, ni fallor, scripserat: τα tCbqI Tag vorjoeig aOQccg εηήβολον^ 
 Those who have objected to this emendation have not remarked that 
 it is opposed to the vital principles of Greek syntax. Conversely, Elms- 
 ley, and Burges, from not perceiving the construction, have extruded 
 the article from the following passage of Euripides {Troad. 398): 
 
 nagig d' ϊγημε ταν Zliog^ yri^ag δ\ μη 
 ΰίγώμενον το xrjdog είχεν εν ^όμOLg. 
 
 Here Elmsley, following in the steps of Mr. Burges, proposes to read 
 
 the second line thus : 
 
 ύιγώμενόν το Krjdog είχεν αν δόμoLg'^. 
 
 Mr. Burges says: "nihil hie habet arti cuius." It seems to us com- 
 pletely at variance with the spirit of the Greek language to omit the 
 article here. For a participle like ΰίγώμενον could not be a mere epi- 
 thet, when used without the article, though it naturally follows the 
 verb είχεν, as the expression of a continuous result. With regard to 
 the av, which is substituted for εν, we think, in the first place, that 
 the preposition is required here; and that the av , so far from being 
 necessary, would actually weaken the meaning. For as the matter 
 was all past and gone, the only apodosis allowable here would be the 
 aorist with av. No one would say of Paris after his death , ειχεν αν, 
 "he would have," but εΰχεν αν, "he would have had." As it is 
 Euripides, referring no doubt to the humble connexion between Paris 
 and (Enone, makes Cassandra say that "Paris married (aor. i. e. as 
 one act) Jove's daughter; but by not having married her (i. e. if this 
 act had been omitted), he thereby continued to keep his marriage 
 affinity in the obscurity which originally belonged to it." (c) We have 
 removed a gross soloecism from Pindar, Isthm. m. 23, by reading ^va- 
 τον δυερχονταο βυότον rikog instead of rb βίου rikog, in which the 
 editors acquiesce, although the article is omitted in several MSS. 
 And we have similarly expunged the article which Bockh had 
 inserted in Isthm. viii. 39, where he reads, contrary to all syntax, 
 
 * Elmsley's note {ad Med. 416), is as follows; ^^ Troad. 398, Πάρις 
 d' εγημε την z/tog, γήμας δε μη βιγώμενον το οιηδος (recte Burgesius ol- 
 γώμενόν τι κηδοζ) εΐχ' αν iv δόμοις. Ita Burgesius, S chief erus, et Matthiae. 
 Vulgatam είχεν iv δόμοις retinuit Seidlerus. Certissime reponendum 
 είχεν αν δόμοίς. Noster Helen. 765: τους ϋ'εονς έχων τις αν φίλους, 
 άρίϋτην μαντιν,ην εχοι δόμοις.'^ Mr. Burges has favoured us with a com- 
 munication on the subject of his conjecture. He says, very truly, that 
 the iv is indispensable, "as is shewn by Burney, or rather Porson, in the 
 Monthly Rev. 1789, p. 245," and that to obviate all difficulties, he would 
 now read, Σιγώμενόν y' av κηδος εΐχεν iv δόμοις, for that av is absolu- 
 tely requisite. We have stated above our reason for dispensing with av. 
 
CHAP, v.] THE ADJECTIVE. 519 
 
 ^^εόμορον οτίάύΰαί το ysgag. In -^schylus, Choeph. 489 , all the edi- 
 tors, so far as we know, silently accept the reading: 
 
 άρ' oQd'ov αϊρείς φίλτατον το <Sov κάρα; 
 
 which is an indefensible soloecism; for it is clear that the tertiary pre- 
 dicate is expressed by δρΰ'όν, and that φίλτατον has no place here 
 unless as an epithet to κάρα^ which, from the position of the article, 
 it cannot be. We believe that the true reading is φυλτάτοίς, just as 
 we find in the reply of Orestes which follows : ήτοι δίκην ϊαλλε 6νμ- 
 μαχον φίλοίξ^. We remark in passing that we should infer from the 
 word Ιτεινάτην in v. 503 , that the last ten lines should be equally 
 divided between the two interlocutors , Electra and Orestes , namely, 
 that we should assign to Electra, vv. 493 — 496, and to Orestes, vv. 
 497 — 502: so that he will begin his little speech with οντωγάρ, just 
 as in 476. In Sophocles, Ajax, 135, the commentators cannot see the 
 necessity of accepting Bothe's emendation, 
 
 της αμφιρύτον Σαλαμίνος 'έχων 
 βά^ρον άγχίαλον, 
 
 for άγχιάλον, nor did that somewhat rash critic propose it on account 
 of the position of the article , but merely on account of the distribu- 
 tion of the epithets. That they should thus strain at the gnat and 
 swallow the camel, is a striking proof of the grammatical laxity on 
 which we are now remarking. Again, in Sophocles (Ajax^ 572): 
 
 καΐ τάμα τεύχη μψ άγωνάρχαν τινεζ 
 ^ηΰονΰ' Άχαίοΐξ, μη^' δ λνμεών εμός — 
 
 all the editors (except Schafer, who proposes to omit the article) have 
 passed over the soloecism in the last words ; and it is even defended 
 by Schneider in his edition of Plato's Bespublica, Vol. ii. p. 319, and 
 by W. Dindorf. We read ό λνμεών εμοί. The word λνμεών stands 
 on the same footing with άτίατεών, &c. It is the weakened form of a 
 participle used as a noun; for as we have άτίατη, άτΐατάω, άτίατεών, for 
 άτίατάων, we might have λνμη, λνμάω, λνμεών ίοτλνμάων. The verb 
 λνμάω does not exist, nor is the lengthened form λνμαίνω used in the 
 
 * We have read with much surprise a note on this line by Professor 
 Conington in his edition of the Choephorce (London, 1857). "There are," 
 he says, "other passages where the adjective, though in the position of 
 a predicate, cannot be taken predicatively (e. g. Soph. Aj. 135, 1167; 
 Phil. 394), so that we must suppose that metrical necessity occasionally 
 induced the Greek poets to violate a rule, which in general they unques- 
 tionably observed." If this principle is once admitted, it will be a mere 
 imposition to talk of accurate scholarship. 
 
520 THE ADJECTIVE. [bOOK III. 
 
 active by the best Attic writers ; they employ only the deponent λυ- 
 μαίνομαι, which is properly followed by the dative, as we are told by 
 the Scholiast on Aristophanes (Nub. 931, λνμαίνόμ&νον τοις μείρα- 
 KLOig): λνμαινόμΒνον , άντΙ τον εννβρίξοντα, όντως δε αντοΐξ 
 6ννη%'ες, ονχί, τα μευράκία λνμαινόμενον , ηροζ ο καΐ το χ. 
 (though Xenophon and others make it govern the accusative also); 
 and even in the later writers, when the active λνμαίνω occurs, it is 
 construed with the dative, as in Libanius, iv. p. 350: τα λνμ^ναντα 
 τοις τίράγμαΰί. It is, therefore, by no means unlikely that λνμεών 
 should be construed with the dative (for such nouns govern the same 
 case as the verbs from which they are derived), and that the ignorant 
 transcriber should not perceive it and write εμός. A similar solcecism 
 has been remedied by a similar correction in Eurip. Hippol. 683 : Ζενς 
 σ' δ γεννητωρ εμος τίρόρριξον εκτρί'ψείεν, where the Copenhagen MS. 
 •supports the reading εμοί, and in Theocritus, xxvn. 58: τώμτίεχονον 
 τίοίηΰας εμοί ράκος, where the common editions have εμόν, the final 
 V, 6, L are very like one another in the MSS. If any one objects that 
 λνμεών is usually found with the genitive (as in Eurip. Hippolyt. 1068), 
 there would be no objection to the emphatic εμον at the end of the 
 line. But the rule is , to prefer the facilior lectio , especially when it 
 involves the difficilior interpretatio , and the transition from εμός to 
 εμοί is easier than that to εμον^. In another passage of the Hippo- 
 lytus (605), VOL τίρόξ 6ε της 6ης δεξιάς ενωλενον , we must consider 
 ενωλενον as the genitive in agreement with (jov implied in ύης, unless 
 we prefer the reading τηΰδε δεξιάς, which is found in one of the MSS., 
 but would hardly suit the context ; for τηΰδε w^ould refer rather to 
 the Nurse than to Hippolytus. 
 
 We might bring forward a great many other instances of the igno- 
 rance or inadvertence of scholars with regard to this fundamental 
 principle of Greek construction; indeed, the remark which Valckenaer 
 made, when he stated the rule, is still applicable — credi via; potest 
 quam frequenter in minutis hisce fuerit ab hominibus etiam Greece 
 perdoctis peccatum (ad Herod, i. 1 80). That English scholars should 
 not have observed this, after the publication of Middleton's elabo- 
 rate treatise, is still more wonderful ; for, although that book is based 
 upon a theory opposed to all sound views of the philosophy of lan- 
 guage, it at least stated distinctly enough the rule that the subject 
 is generally found with the article and the predicate without it (pp. 53 
 foil. ed. Rose). 
 
 '■^ Mr. Shilleto, like S chafer, proposes the simple expedient of omitting 
 the article before λνμεών in the one passage and γεννήτωρ in the other 
 Journal of Philology^ ii. p. 85). 
 
CHAP, v.] THE ADJECTIVE. 521 
 
 305 The following examples will explain to the young student 
 the influence of the article in determining whether a given adjective 
 or participle is to be considered as an όνομα or as a ρήμα. ^O βαΰι- 
 λεύων is a synonym for δ βαΰίλενς, but βαΟίλενων means "during 
 his reign," "when or if he is reigning:" for example, KvQog δ βαο^- 
 λεύων is "Cyrus the king," but βαΰιλενων 6 KvQog, "Cyrus, when 
 he was king." ""O λενκος iTCTCog means "the white horse," but λενκος 
 6 LTtitog, "the horse is white." ""0 φρονών is "the wise or prudent 
 man," but, Sophocl. (Ed. Tyr. 316, 
 
 ςρεν, φεν' φρονεΐν Sg δεινον ev&a μη τέλη 
 λνει φρονοϋντο, 
 means "what a sad thing it is to be wise, in cases where it is unpro- 
 fitable (does not pay) to be wise ," so that the participle is equivalent 
 to the infinitive; comp. (Ed. Tyr. 863: εϊ μοι ξννείη φεροντι μοίρα 
 ταν ενΰετντον αγνείαν λόγων, έργον τε navxog, and Lysias (de eversa 
 re^mblica, p. 174, 1. 14): ονκ a^iov 7toλλάκLg χρηΰ^αι <5vμβovλoLg olg 
 ονδε ατΐαξ ελνύίτεληΰε 7tεi^oμεvoLg. Ή βεβaιog χάρίg would mean 
 " the lasting obligation," but the Corcyreans say "it is their business to 
 show that their gratitude will be lasting," — og καΐ την χάριν βέβαιον 
 εξονΰιν (Tliucyd. ι. 32), and thus they tell the Athenians (in the fol- 
 lowing chapter): "that by receiving as allies people whose dearest 
 interests were at stake, they would confer the favour with as indelible 
 a record as possible," og av μάλιότα μετ' άειμνήΰτον μαρτυρίου την 
 χάριν κατα^εΐΰΟ'ε. The following passage contains a good exemplifi- 
 cation of the uses of the participle, both as όνομα and ρήμα ; Thucy- 
 dides, I. 36: και οτω τάδε ξ,υμφεροντα μεν δοκεΐ λεγεΰ^'αι, φοβείται 
 δε μη δι αυτά Jtεώ'όμεvog Tag 0πovδag λνΰη, γνώτω το μεν δεδιog 
 αυτοΰ, ίΰχνν εχον^ roi'g εvavτίoυg μάλλον φοβηύον, το δε Ο'αρβονν 
 μη δεξαμενου, ά6&εvεg ον 7tρog i(5χύovτag Tohg εχ^ρoύg, άδεέΰτερον 
 εύόμενον. Here it is obvious, that Συμφέροντα , πει^'όμεvog , έχον, 
 φοβηΰον, δεξαμενου, ον, ι6χύovτag, and εύόμενον, are all predicates, 
 the two futures being equivalent to infinitive moods of the same tense, 
 while το δBδιόg and το &αρΰούν are subjects, or equivalent to nouns 
 substantive: the meaning is: "and if any one thinks that what has 
 been said is for his interest, but is afraid, lest, if he listens to our ar- 
 guments, he may be induced to break the treaty, let him know that 
 his fear, if it brings him strength, will rather be alarming to his ene- 
 mies, whereas his confidence, after having refused our aid, will be less 
 formidable, because it will be weak as compared with his enemies, who 
 will be strong." Again, oi aφεβτηκότεg Σύμμαχοι would mean "the 
 revolted allies;" oi άφεΰτηκότεg των ξνμμάχων^ ["those of the allies 
 who had revolted;" but Thucydides, ii. 65: άντεΐχον τών ξυμμάχων 
 ετι Tolg τΐλείοΰιν άφεύτηκόΰι means "in addition to all their other 
 
522 THE ADJECTIVE. [bOOK III. 
 
 enemies they kept their ground against the majority of their allies, 
 they having revolted," or "for they had revolted," or "after they had 
 revolted;" not "who had revolted," as some people would translate it 
 (cf. Thucyd. i. 18, § 7). Aristoph. Equites. 280: 
 
 val μα jdia καγωγε τοντον, on ^svy ry κοιλία 
 ίΰδραμών eg το τΐρντανεΐον, είτα τίάλιν εκ^ει πλεα. 
 "I'll inform against this fellow, for going into the town-hall with 
 his belly empty, and coming out again with it full;" i. e. his belly 
 was empty when he went in, but full when he came out again. In 
 Sophocl. Antig. 360, 
 
 ατίοροζ ετΐ ονδεν άρχεται 
 το μέλλον, 
 the Scholiast and Hermann suppose that ετΐ' ονδεν το μέλλον ought 
 to be taken together , with the sense εΛ ονδεν των μελλόντων ; but 
 it appears to us that το μέλλον is in apposition to the whole of the 
 preceding line, "in regard to the future, he comes to nothing unpro- 
 vided with resources." At the beginning of the following strophe, the 
 article is properly explained by "Wex. In another part of the same 
 play, the chorus, after stating that, when misfortuness once begin in 
 a family, they go on till the race is extinct, exclaim (v. 594): 
 
 αρχαία τα Ααβδακίδάν οϊκων ορώμαι 
 τίήματα φ^ιμένων έτά τΐήμαΰί τΐίτίτοντ, 
 
 ονδ' ατίαλλάύύει 
 γενεαν γένοξ — 
 which signifies, "the calamities of the house of Labdacus, which I 
 behold in the act of being added to the calamities of those members 
 of the family who are dead and gone, are only the old misfortunes re- 
 suscitated and revived." 
 
 306 These instances will be sufficient to show the natural con- 
 nexion of the article with the subject of the proposition. From these 
 cases, the student will be careful to distinguish those in which the 
 proposition is convertible or reciprocating, such, namely, "that of either 
 term taken as the subject, the other may be affirmed as a predicate" 
 (Middleton, p. 54). In these last cases, which are, of course, not very 
 numerous, the article either appears before both subject and predicate, 
 as in Plato, Thecetet.ip. 145 ε : αρ' ov το μανΟ'άνειν έΰτϊτο όοφώτερον 
 γίγνεύ^αί περί ο μαν^άνεΐξ; (cf. Heindorf ad Gorg. § 102, p. 49 1 ε*) or 
 
 * The instance given by Middleton from Aristot. Eth. ii. 9, eauv 7) 
 UQBTi} y ήΟ-ικη μεοότης, is erroneous; η ήΟ-ι-κτ] is the epithet to y αρετή, 
 and μεΰότης alone is the predicate. 
 
CHAP, v.] THE ADJECTIVE. 523 
 
 is omitted in both, as in the aphorism of Protagoras: τίάντων χρη- 
 μάτων μίτρον αν^ρωτίος. "With the latter, we must compare such 
 cases of tertiary predication, as λεγόμενον ερέω (Pind. Pyih, v. 101), 
 which may be rendered indifferently "every body will say what I say;" 
 and, "I will say what every body says." Matthise (Gr. Gr. § 264, 
 obs.) quotes some instances, in which he says the predicate has the 
 article, but the subject wants it. In all these the predicate, as he 
 calls it, is the subject, nor can we conceive the possibility of such a 
 conversion as he supposes. On the omission of the article with 
 the subject and its appearance with the predicate, see the Greek 
 Grammar y articles 392 sqq. ; and for full illustrations of the different 
 kinds of predicates, see articles 417 — 498. 
 
CHAPTER VI. 
 COMPOUND WORDS. 
 
 307 Analogy between the Greek and Sanscrit compounds. 308 Orthographic 
 affections of contiguous words in these languages approximate to compounds. 
 309 The six classes of Sanscrit compounds compared with the Greek. 310 
 Parathetic and synthetic compounds in Greek. 311 Parathetic compounds 
 occasionally overlooked. Case oi χρόνω-κλντός and λόγω-παλαιός. 312 Syn- 
 thetic compounds. Their euphonical affections. 313 Accentuation of those 
 which terminate with a verbal. 314 Compounds commencing with a verbal. 
 315 The signification and construction of compound words has sometimes 
 very little reference to their constituent parts. 316 The compounds διθύ- 
 ραμβος, καλοκάγ(χΰ•ός , and εντελέχεια, important terms in literature, poli- 
 tics, and philosophy. 317 (1) ζ}ί%^νραμ§ος. The termination connected with 
 ίαμβος. 318 The middle syllable contains the root οίΟ•νρ-αος. Explanation 
 of this symbol. 319 The first syllable is the dative of Ζενς. 320 More recent 
 opinions respecting the Ο^νρΰος. 321 (2) Καλογ,άγαΟ-ός. General meaning 
 of the compound. 322 ΆγαΟ'ός refers to nobility of birth. 323 Derivation 
 of άγα%•ός. 324 Καλός expresses educational accomplishment. 325 Quali- 
 ties attributed to the aristocracy; 326 Connected with their social position, 
 as opposed to that of the populace. 327 Moral excellence denoted by %αλο- 
 "ϋάγαΟ-ός. 328 Also by the Latin gentilis. 329 The woro ηρως as a title of 
 rank. 330 Connexion of this term with Άνριος, κόροδ, τίονρίδιος, &c. 331 
 Digression respecting τινρηβος and ^νρηβια. 332 "Αρης and vir. 333 'Ανηρ 
 = /'α-νήρ and Nero. 334:"Αναξ =^άναξ, connected with ανά. 335 ''Ύβρις 
 similarly connected with νπέρ. Contacts between this word and τίόρος. 
 336 Further analogies between %όρος, αδρός, χλιδή, οργή, &c. 337 Konig 
 and "king" not immediately connected with /"άναξ. 338 "Lord" refers to 
 elevation. 339 (3) *Εντελέχεια. Difficulties occasioned by a confusion be- 
 tween this Aristotelian term and the older word ενδελέχεια. 340 Opposition 
 between δνναμις and εντελέχεια. 341 ζΐνναμις also opposed to ενέργεια. 
 342 Distinction between εντελέχεια and ενέργεια. 343 Aristotle uses εντε- 
 λέχεια to signify the absolute definition of a thing. 344 Signification and 
 etymology of ενδελεχής. 
 
 307 ANE of the most striking peculiarities, and indeed one 
 of the greatest beauties, of classical Greek, is the fre- 
 quent and varied use of compound words by the best authors. 
 Our own language cannot make the most distant approximation 
 to the Greek in this, the German falls far short of it, the Latin 
 still more so. There is, however, one language of our family, 
 the Sanscrit, which bears a strong analogy to, and even excels, 
 the Greek in this respect; it will be proper, therefore, before we 
 engage in an inquiry about the principles which regulate the 
 formation of compound words in Greek, to consider the laws 
 according to which this process is carried on in the old language 
 of India. 
 
CHAP. VI.] COMPOUND WOEDS. 525 
 
 308 A person not well skilled in Sanscrit always experiences 
 great difficulty in distinguishing the words in a line of poetry from 
 one another: the whole line will appear to him to be formed into one 
 mass , the end of every word being altered , on euphonical principles, 
 to suit the commencement of the word which follows; in fact, as 
 Colebrooke has remarked (Asiatic Researches, vni. p. 201), it is an 
 euphonical orthography, which consists in extending to syntax the 
 rules for the permutation of letters in etymology. The same is ob- 
 served to a certain extent in old Greek inscriptions (Bockh, Corpus 
 Inscript. i. p. 126). The feeling which gave rise to this orthographi- 
 cal anomaly, and certainly to the formation of the long compounds 
 also (see the instances in the Gr. Gr. Art. 365), may be traced in the 
 peculiarities of Greek syntax; for instance, that construction which 
 we call attraction is the simple effect of a striving after brachylogy, 
 of the attempt to compress the meaning of a sentence into a closely- 
 connected group of words. The only distinction, between a real 
 compound and syntactical phenomena like this, is that in the com- 
 pound the separate words have so entirely coalesced that the inflexion 
 of the last word alone is regarded. 
 
 309 The Sanscrit grammarians have discriminated six kinds of 
 compound words or samdsa. They give the following names to the 
 different species (see Wilkins' Grammar, pp. 556 foil.): (1) avyayi- 
 bhava, (2) tatpurusha, (3) dvandva, (4) dvigu, (5) bahuvrihi, (6) 
 karmmadharaya. We shall consider these one after the other. 
 
 1st class. A compound of this kind is indeclinable. The first 
 member is some preposition or particle, and the last is a noun termi- 
 nating in the sign of the neuter gender; for instance, nirmakshikan, 
 "without flies," from nir^ "without," and makshika-s, "a fly." Simi- 
 lar compounds in Greek are εμτίοδών, &c. Sometimes , however , the 
 substantive appears in the instrumental or locative case, like ένεκεν; 
 thus we may write either upa-kumbhan^ or upa-kumbhena krttan, 
 "done by the jar," and either upa-kumbhan, or upa-kumbhe nidhe-hi, 
 "place it in the jar." This looks very like the government of a case 
 by a preposition, which the Sanscrit grammarians repudiate, but of 
 which we have found other similar instances: thus in the Edmdyana 
 (Lib. II. 66, gloka 29, ed. Gorresio) we find nanu te aham prdnebhyo 
 'pi priyd^ "am I not dear above your life?" priyd apt (ετΐί) prane- 
 bhyas, for priyatard pranebhyas, is like vitd super cara for vitd carior. 
 Sometimes we have a bahuvrihi^ or declinable adjective, where we 
 should expect the avyayibhdva or adverbial compound : thus, close by 
 the passage just quoted we have, in two following glokas , sabhdryas^ 
 "with my wife," in the nominative case, and tarn aham spargayd- 
 
526 COMPOUND WORDS. [bOOK III. 
 
 mdsa sa-bMryam (accus. agreeing with tarn) patitam {τί^βόντά) sutam, 
 "I made him, together with his wife, touch his fallen son." 
 
 2nd class, or tatpurusha. These compounds are formed of two or 
 more nouns, the first set being in some oblique case, governed by the 
 last, which may be a substantive, an adjective, or a participle in -ία: 
 the following are instances, rdja-purusha^ " a king's man," Jiasty-agva- 
 ratha-ghesha^ "the noise of elephants, horses, and chariots," svarga- 
 patiia, "fallen from heaven." These correspond, of course, to the 
 Greek compounds εγχεΰί-μωρος, ^εο6-εχ%ρία, τΐαρ^ενο-ττίπης, ΰιδψ 
 ρο-μήτωρ, &c. The construct state of the Hebrew noun represents 
 the tatpurusha compound in the inverse order of arrangement, cf. 
 Tl'i "ΐώίίί with rdja-purusha. 
 
 3rd class, or dvandva. This class forms substitutes for collections 
 of nouns in the same case and joined together by a copulative 
 conjunction. The last noun alone is declined, and is (1) in the dual 
 or plural number according as two or more nouns are joined together; 
 or (2) it is a neuter singular, showing that the aggregate is considered 
 as one: thus (1) guru-gishydu, "master and scholar" (dual); hrdh- 
 mana-kshatriya-vit' -guards, the names of the four Indian castes (plur.) : 
 (2) chhatro-'pdnaha'm (cliliatra, updnaha) , "parasol and shoe" (neut. 
 sing.). The Greeks do not distinguish the dual, but dvandva com- 
 pounds are not uncommon in Aristophanes; see, for instance, the 
 long word with a collective ending in the Ecclesiazusce (1169 foil.), 
 and proper names like Τίΰαμενο-ΦαίνιτίτίΟί (Acharn. 603). 
 
 4th class, or dvigu. Collectives, of which the first part is a nume- 
 ral, belong to this class; the noun is either feminine in ^, or neuter 
 in a: thus tri-rdtra^ "three nights," tri-loki, "three worlds." The 
 Greeks have no dvigu compounds, unless we can consider adjectives 
 like δίδραχμον, χιλιον αυττίξ-, &c., as belonging to the class. 
 
 5th class, or bahuvrthi. In this class are contained compound 
 adjectives ; the last part is a substantive, the first is any other part of 
 speech. The following are instances: bahu-dhanas , -a, -an, "rich," 
 from bahu, "much," and dhana , "wealth;" baliu-padas ^ "having 
 many feet;" bahu-mantangan vanan^ "a forest with many elephants." 
 These compounds are very common in Greek, as τίολντίονζ, φυλόκα- 
 λος, &c. 
 
 6th class, or karmmadhdrya. This class forms compounds, of 
 which the last part is a substantive or adjective, modified in meaning 
 by some preceding adjective in an uninflected state. Thus, mahdrdjas, 
 "a great king;" param-dha, "a fine day;" su-mahat (εν μέγας) 
 «very great." The interrogative km is also used in this sort. of com- 
 
CHAP. VI.] COMPOUND WOllDS. 527 
 
 pound as an expression of contempt: as kinvira, "what a hero!" The 
 karmmadhdrya compounds in Greek are mostly proper names, as 
 Ίεροΰόλνμα, ΜΈγαλόπολίξ ; in the poets, however, we often find them 
 as common epithets; but the adjective does not always precede; thus 
 we have ορ^όμανης for ogd'og μάντις, t ν μβόχωΰτ og ior t ν μβθ£ χωύ- 
 τόζ, αρΐ(5'&άρματοζ for αριότον άρμα, άνδροφ^'όρος for άνηρ φ^α- 
 Qstg, &G. One of these compounds, which is ajta^ λεγόμ^ενον, ought to 
 be banished from the lexicons. For we entertain no doubt that the 
 line (iEsch. Agam. 1143), εγω δε ^ερμ6vovg τάχ εν πεδω βαλώ, ex- 
 hibits merely a confusion of the true readings, εγώ δε ^ερμ()cg ΰτα- 
 yovag εν Λεδω βαλώ. Cf. Agam. 1249: Ο'ερμώ %07Cεί6ηg φοι,νίω 
 τΐροόφάγματί. Choeph. 394: (poviag urayovag χvμέvag εg ηεδον. 
 
 310 Greek grammarians distinguish between two sorts of combi- 
 nations : the one synthetic or organic — τίατα βνν^εΰιν; the other para- 
 thetic or unorganic — 'κατα τΐαράΟ'εΰίν (Apollon. Dyscol. Syntax, p. 310 
 Bekk.). The former is when words are so combined that the first of 
 them loses all inflexion , and the last word is the pivot of the mean- 
 ing : the latter is when both words retain their inflexion, but are joined 
 together so intimately and habitually that they may be written as 
 one word. The parathetic compound is a natural prelude to synthetic 
 combination, and we sometimes find words oscillating between the two 
 sorts of composition. Regular parathetic compounds are often found 
 in proper names, as Κννόΰΰημα ; in particles, as ονκέτί, τοπρώτον ; 
 or in epithets, as vavoL-KXvTOg , γaΰτρί-μaργog, νεώΰ-οίκοϋ; or we 
 find that, although the two words are not absolutely melted down into 
 one, the former has sufl'ered some modification in its vowels on account 
 of the weight of the word, so that neither part could stand alone: such 
 words are 6aκε6-7tάλog^ μελεΰί-πτερος^ τελε(^-φόρog, 6δoί.-7tόρog, and 
 a number of words compounded with %'εog- for ^'εoLg (Pott, Etymol. 
 Forsch. I. p. xxxviii; Rosen, Journal of Education, ix. p. 334), such 
 as ^εο6-ε%^ρία, %•ε6(3-δotog , ^εού-κυνη , ^'έ(J-κελog (^'εoΐg ϊκελog), 
 ^ε6-'JtLg, ^'εΰ-τΐέΰίος,^'εβ-τη-έτΐεία (in which the root /f^r- occurs twice; 
 so that the word means "saying again what was said to her by the 
 gods "), &εΰ-7tρωτόg (^'εoΐg 7tε7tρω-μεvog), ^•εόΰ-6vtog (εκ &εοϋ δρμη- 
 %'εΐΰα, Schol. ad jEschyl. Prom. 1 1 6), ^ε6-φaτog, ^εύ-πιαί, %ί6-βη, &c. 
 In άργεL•-φόvτηg, άνδρει-φόντης, the possessive adjectives ^Αργεΐος, άν- 
 δρεΐog partly explain the genitive presumed in the original parathesis. 
 In ένεκα, φρovδog, &c. the breathing only of the first part is afiected : 
 in βόβ-7toρog, vΰ-7toρog, λυκόΰ-ονρα, we have abbreviated forms of 
 the old genitive. A verb combined with a preposition is not only a 
 parathetic, but also a separable compound: at least in the older state 
 of Hellenism they could always be parted by tmesis ; when , howeve;-, 
 
528 COMPOUND WOEDS. [bOOK III. 
 
 a, verb combined with the preposition assumes a secondary derivative 
 form, the compound becomes synthetic, and such words as 6vvoiKog, 
 ετΐίκονρος, uvvdovXogf &c, are one and indivisible, as indeed appears 
 from the way in which the accent is thrown back (Apollon. ubi 
 supra). 
 
 311 We must take this opportunity of pointing out two passages 
 which have been erroneously interpreted, in consequence of a mis- 
 understanding of a juxtaposition , virtually amounting to a parathe- 
 tic compound, which occurs in both. The passages are, Pindar, Pyth. 
 XI. 32: 
 
 d'avev μεν αντος ηρωζ 'ATQBidccg 
 ΐκων χρόνω-κλνταΐς εν ^Αμνηλαυξ. 
 and ^schylus, Choeph. 641 : 
 
 τέκνον δ' επειόφέρευ δόμοιύι, 
 δωμάτων Λαλαιτέρων 
 τίνειν μύβοζ, 
 
 χρόνω-κλντη βνόΰόφρων ^Ερινύζ. 
 In both of these passages the commentators take χρόνω by itself, as 
 signifying "after a long absence," "at last." Pindar, however, would 
 hardly have used χρόνω in this sense, as he says immediately after- 
 wards χρονίω όνν "Αρει (v. 36), and this meaning in the line of 
 .^schylus would leave no tolerable sense for κλυτή. But κλντός is 
 constantly used in parathetic compounds, as νανΰίκλντός, δονρίκλυτός, 
 &c., implying merely eminence in that which is expressed in the dative 
 case preceding. Now the particular always considered as most remark- 
 able in Amyclse is its extreme antiquity, and the Erinyes are espe- 
 cially in this very trilogy of -^schylus distinguished from the other 
 deities by their greater age (see Eumen. 701, 748, 810, and comp. 
 Λαλαιόφρων ^ Eumen. 833, with βνόΰόφρων in this passage: also 
 Miiller, p. 181). Therefore, χρονοκλντός, as it might be written, signi- 
 fies only "time-honoured" or "ancient" in both places. The same is 
 the case with the collocation λόγω-ΛαλαίΟ£ in the two following pas- 
 sages; -^schyl. Agam, 1198: 
 
 εκμαρτνρηΰον %ρουμ06αζ τό μ είδεναι 
 
 λόγω-Λαλαιαξ τάΰδ' αμαρτίαξ δόμων. 
 \. e. "that I know the sins of this house, although they are old in 
 story." Soph. (Ed. Tyr. 1395: 
 
 ώ Πόλνβε και Κόριν^ε, καί τα τίάτρια 
 
 λόγω-Λαλαια δώματα. 
 \. e. "Ο Polybus and Corinth, and ye paternal, in-word-original man- 
 sions" — which were formerly called my home. The construction of 
 
CHAP. VI.] COMPOUND WORDS. 529 
 
 παλαιός with, λόγω in both these passages, and the similar use οΐχρόνω 
 in the two former instances of quasi-parathesis , is confirmed by a 
 passage in Sophocles, where the dative of the one collocation is com- 
 bined with the adjective of the other; (Eed. Col. 112: 
 ΰίγα, TtOQBvovtaL γαρ οΐδε δη nveg 
 χρόνφ-τίαλαιοί^ 6rjg έδρας Ιτΐίΰκοτΐον'^'. 
 312 In treating of the synthetic compounds we will first take 
 those which have the verb-element in the ast place, and then those 
 which begin with the verb. When two nouns are joined together we 
 find them connected by the short vowel o, which is elided when the 
 second word begins with a vowel : where the first word ends in l or 
 V these vowels generally keep their place before consonants: the same 
 remark applies to av and ov, and, in certain cases, to the liquids ν 
 and ρ. When the first noun has two pronominal elements affixed to 
 the root the latter is frequently dropt. This is particularly the case 
 with nouns ending in -μα-τ-', thus we have αυ-μο-ΰταγης for αί-μα-το- 
 όταγης, ΰτο-μ-αλγίαίον 6τομα-τ-αλγία, &c. The ν of the comparative 
 may be omitted even when the second part of the compound begins 
 Avith a consonant, as in ^sch. Sept. c. Theh. 783 : των κρενύόοτέκνων 
 ajc ομμάτων ε7ίλάγχ%'η, where we ought to read κρείβΰοτεχνών (cf. Pind. 
 jPr.29: άριΰτοτέχνα πάτερ). And in the strophe whe should read μη 
 τελ^ τάΰδε καμ'ΐ\}ίπουξΈρινύς. We have similarly 'Ελπινίκη for Έλπι- 
 δονίκη. After the compound word had become so well established in 
 use, that its individuality was alone considered, and its separate parts 
 forgotten, the fineness of the Greek ear induced certain euphonical 
 changes made with a view to blending the elements into one mass. 
 This generally took place at the point of junction, but, according to 
 the principle before laid down with regard to the change of place in 
 the aspirate, this euphonical change was shifted to different syllables 
 according to the length and general weight of the word. One of the 
 common changes was that repetition of a liquid, of which we have 
 before spoken: this was always the case with the ρ, as in περίρροος, 
 Ιΰόρροπος, &. ; it also occurred very frequently with the other liquids, 
 as in άπολλήγειν^ ^Ιππομμεδων, Παρ%'εννοπαίος, '^λφεβόίβοια. In a 
 compound like ^Ιππόδαμος euphony might require a lengthening of one 
 of the last three short syllables, and as this could not take place at 
 the junction, where there was no liquid, it is passed on to the liquid 
 μ in the next syllable, and thus we ΆΏά'Ιππόδαμμος (see above, § 220). 
 
 * The second passage, which was first adduced in our Greek Grammar, 
 Art. 378 (ed. i.; 367 ed. ii.) shows that Mr. Paley was rather hasty in 
 questioning the interpretation of λόγω -παλαιός in the Agamemnon; see 
 also Journal of Philology, iii. p, 210, 
 
 MM 
 
530 COMPOUND WOEDS. [bOOK III. 
 
 This euphonic lengthening, however, most frequently happens in the 
 first syllable of the second part of the word, where we often find an a 
 or ε changed to η, and ο to ω, as in the following instances : ενήνεμοζ 
 {ανεμοξ}^ δυ6ήρετμος{ερετμος\ όννώννμοξ (όνομα), &g. In some com- 
 pounds it is a matter of indifference whether we lengthen this joining 
 syllable or some other: thus we may say indifferently άνήρΐ^μος or 
 άνάρΐ&μος. A case of the same kind is the lengthening of the first 
 vowel of a compound or derivative word, as ηνορέη (άνηρ), ώλεΰίοίκος, 
 &c., and to this case also the last-mentioned variation applies; thus 
 we have both Ύίλνΰίξ and ελενβίξ (Buttmann, Ausfuhrl. Sprl.^. 120, 
 Anm. 1). These vowel-changes fall under the head of guna; before 
 β and TC however we often find anusvdra, as in τν-μ-τίανον, τίειβί-μ- 
 βροτοζ, ο-μ-βριμθξι &c. When the second part of the compound does 
 not begin with either a liquid or a vowel, we find the final vowel of 
 the first part, which should properly be a short o, the medium weight 
 of «:, converted into r^ or a, and that too, not only in the case of nouns 
 of the first declension as χοηφόρος, άγορανομοζ, but also in nouns of 
 the second and third declensions, as ελαφηβόλος, θανατηφόρος, τίολιά- 
 νόμοξ, which seem to refer to obsolete forms of the included noun 
 (compare verb-forms like %'ανατάω). 
 
 313 When we say that some synthetic compounds begin or end 
 with a verb, we must be understood as meaning, that they contain 
 the root of a verb, or rather the crude-form of a verbal substantive. 
 It cannot be too strongly impressed upon the student's mind, that 
 verbs are never directly compounded with any thing but prepositions, 
 in which case they are separable compounds. All apparently com- 
 pound verbs are derivatives from compound substantives, of which 
 the last part is a verbal root. Thus, from βάλ-λειν is formed the 
 verbal βολή, from λίθ^ον βάλλευν the juxtaposition or parathesis λίθ^ον 
 βολή, from this the compound verbal λι^οβόλο-ς, and from this again 
 the derivative verb λί%Όβολέ-ω. According to the accentuation, such 
 a word as λίθοβόλοζ might mean either "a person who pelts" or "a 
 person who is pelted" with stones ; that is, according as the accent is 
 on the verbal element or not. Thus , μητροκτόνος would apply to 
 Orestes who killed his mother, but μητρόκτονοι to the children of 
 Medea who were slain by their mother. This rule of distinctive 
 accentuation is not observed in words compounded with prepositions 
 and adverbs or with the adjectives Ttag and τίολνς, which are always 
 proparoxytone: for instance, we find ομόΰΛορος proparoxytone when 
 applied to Jocasta as the wife both of CEdipus and Laius (CEd. Tyr. 
 260), and to Qi]dipus himself, as the husband of the same wife with 
 his father (460) : in both cases as a synonym of ομόγαμος (Eurip. Here. 
 
CHAP. VI.] COMPOUND WOEDS. 531 
 
 F. 339), and οι ομογενής {(Eed. Tyr, 1361); and we have the same 
 accentuation when the word signifies "a brother" or "sister," as in 
 Trachin.212 : βοάτε ταν ομόοτΐορον'Άρτεμιν Όρτνγίαν. But besides 
 this exception depending on the nature of the first part of the com- 
 pound, instances are numerous in which compounds terminating with 
 verbals, and having only one, and that an active signification, are 
 nevertheless consistently proparoxytone : such words are (1) the epic 
 compounds ί'Λ7ί68αμος^εγχε(57ίαλος^(5α'αε6%αλθξ^ηά^τολί7ίορ^ος\ (2) 
 compounds with άρχω and ΰνλάω, as ιπτίαρχος, ίερόΰνλος-, (3) com- 
 pounds with εχω, as ηνίοχος, δαδονχος. It is to be remarked that 
 words compounded with έργο- are oxytone when they signify a bodily 
 action, thus, we have λί%Όνργ6ς, γεωργός, &c. ; but proparoxytone, or 
 by contraction properispome, when they denote merely an operation 
 or habit of the mind, and thus we have Λανονργος, τίαντουργος, &c. 
 Exceptions to this rule are found in ραδιονργός-, λίτονργός, λεωργός, 
 which last is a synonym of λοτονργός, and derived from λεώς, not 
 from λέως, as Hermann suggests (ad Soph. Antig. 1261). 
 
 314 AYhen the first part of the compound is a verb-root, it gene- 
 rally appears under the form of a verbal in -(Jt-, as in εγερϋίχορος, 
 λνύίτίονος, or the vowel is elided, α^ίηρί'φοίΰτας,τίλΎΐ^ίτίΛος,φνξάνωρ. 
 Very frequently, however, the shorter verbal form in η is preferred, 
 the crude-form of which ends in ε : as in άρχέ-κακος, δακε-0•νμος, &c. ; 
 or with elision, as in φεραύτΐίς, where we have the lighter vowel ε 
 instead of the o, which we find in the lighter form φορά. Sometimes 
 the connecting vowel is ο or i,, as in φνγό-μαχος, άρχι-^έωρος. The 
 compounds like λείτίο-ταξία^ φαίνο-μηρίς, in which the strengthened 
 form of the present tense is preserved, seem to owe this peculiarity to 
 a speciality elsewhere observable in these two roots. Thus we have 
 λείτίοΰτράτίον, λειτΐοιί^νχεΐν, λείτίογνώμων, &c., though Cobet (Ilyperid. 
 p. 57) would write λίτΐ- on the authority of Callimachus (ap. Schol. 
 AristopL• Av, 873) and the verbals in -ΰίς from this root {ϊκίει-φις, 
 ελλεΐ'φι,ς, &c.) belong to the class which always retains the diphthong 
 (cf. αμεΐ'ψίς, δεΐξις, ξεν^ις, &c. And we have the strengthened root 
 of φαίνω in Φαΐνοφ, φαυνοτΐροβωτίώ, Φαίνιπτίος, Φαινιππίδης, Φαυ- 
 νοκλης, Φαινόκριτος, &c. Forms like ταμεΰί-χρως, φερεΰ-βιος, ελκεύί- 
 πεπλος, ταλαΰί-φρων, λιτίεύ-ηνωρ, ^αμάβ-ιτΐτίος, Άγηόί-λαος, &c, 
 present the verbal in rather an unusual form, but there is no necessity 
 on this account to have recourse to the theory of Rosen (Adnot. ad 
 Big- Ved. p. xxii), that the 6 represents the r of a participial crude- 
 form, and that the l is an euphonic insertion before consonants. 
 
 315 Sometimes a compound is considered as conveying only a 
 simple notion, and the separate meaning of its constituent parts is 
 
 MM2 
 
532 COMPOUND WOEDS. [bOOK III. 
 
 consequently disregarded. Thus, in Sanscrit, go-sht'ha is literally "a 
 cow-stall;" but from constant use, it came to signify any stable for 
 cattle; hence we have agva-go-sht'ha, "a horse-cow-stall," and even 
 go-go- sht' ha, "a cow-cow-stall." This, of course, is more common, 
 when the compound belongs in part to an obsolete dialect; thus we 
 have loup-garou=lupus garulphus, where the second word is equiva- 
 lent to wehr-wolf (sibove, p. 193); and nag-poor =nagara-pura, where 
 both words signify "a city" or "town." So too cor-morant is com- 
 pounded of the Latin corvus and the Bas-Breton mor-vran, "sea- 
 robber." The same redundance is observable in phrases also, and 
 sometimes appears rather ludicrous, as in the French aller a cheval 
 sur un dne. These instances and others have been collected by Pott 
 (Zdhlmethode, p. 127, note). "We have examples of this in Greek 
 also. Thus Aristophanes uses the phrase tqlklvovv r jj δίκελλη το 
 χωρίον (Pax, 571), although the τρίαινα and the δίκελλα were un- 
 doubtedly not identical implements. And compounds are occasionally 
 significant only in reference to the epithet which they jinvolve; thus 
 τΐαλαίφατος and αρείφατοξ are used in the same sense as παλαιόζ and 
 αρειοζ alone (Hermann ad Soph. Aj. 221). It has been supposed that 
 ορ^ότίονς, as an epithet of jtayog (Soph. Antig, 985), conveys merely 
 the meaning of 6ρ%'ός (Matth. Gr. Gr. § 446, 10, obs. 3 a); but the 
 use of the phrase ορ&ος Tcovg to denote the foot outstretched with 
 unbent knee, e. g. as when Minerva is represented in the attitude of 
 hurling a spear (^sch. Eumen, 284, and Muller, p. 112), or when a 
 bull resists an effort to drag it on board a vessel (Eurip. Hel. 1555), 
 shows that both parts of the compound are significant in this case. 
 Compound verbs are often placed on a parallel footing with complete 
 phrases, as in ^sch. Sept. c. Theb. 261, where we ought to read 
 μήλοίΰίν ανμάΰΰοντ ag εΰτ lag ^εών τavρoκτovovvτάg θ', οΙ<5ιν ωδ^ 
 lπεvχoμaι,m^iQΆ^i οι τavρoκτovovvτag^εoΐ6LV' and Plato (cf. Sophist. 
 p. 365 Ε; PoliticuSj p. 302 ε) uses indifferently the uncompounded τέμ- 
 νειν δίχα, and the compound verb δίχοτομεω. When the force of a com- 
 pound verb is reduced to a simple notion, which does not definitely 
 express the meaning of its constituent parts, we find the construction 
 altered accordingly. Thus εκΰτηναυ, "to stand out of," should govern 
 the genitive case ; but when it signifies simply "to avoid," it is followed 
 by the accusative, as in Sophocles, Ajax, 82 : φρονονντα γαρ νυν ουκ 
 αν Ιξέΰτην οκνω (see the parallel passages quoted by Lobeck). Some- 
 times the compound word is used absolutely, and without any case 
 after it, to express some simple notion. This is the usage of the aorist 
 άνεγνων in Homer, as in //. xiii. 734, it is said of the man endued 
 with wisdom: και τε πόλεLg εΰάωΰε, μάλιύτα δε κavτog άνεγνω, "he 
 saves cities, but, what is the chief matter, he has knowledge himself." 
 
CHAP. VI.] COMPOUND WORDS. 533 
 
 The common reading, 8ύι % αντόζ is manifestly wrong. Similarly, the 
 compound verb αφαιρούμαι, bears the sense of "prevent, hinder, put 
 a stop to," without any immediate reference to the literal signification 
 of its component parts; generally with the infinitive; Pindar, Isihm. i. 
 62: πάντα δ' Ιξείτίεΐν — αφαιρείται βραχν μετρον έχων ϋμνος. Some- 
 times with μη also: Eurip. Troad. 1145: το δεΰΛοτον τάχος αφείλετ 
 αντην Jtaida μη δονναι τάφω (see Heindorf, Plat. Protag, p. 260 a). 
 Hence it is used absolutely, without any accusative or infinitive, to 
 signify mere hinderance or prevention, as in ^schyl. Fers. 428: έω^ 
 κελαίνης νυκτός ομμ άφείλετο, "until the darkness of night inter- 
 posed a hinderance." So also Xen. Hell. i. 2, § 16: μέχρι ΰκότος 
 άφείλετο. This is imitated with an explanatory addition by Thucyd. iv. 
 134, § 2: άγχωμάλον της μάχης γενομένης καΐ άφελομένης νυκτός 
 το έργον. Arrian, Anab. ιι. 11, § 5: η νυξ ου δια μακρού ετίεγε- 
 νομενη άφείλετο αυτόν το τίρος ^^λεξάνδρου άλώναι. Josephus, 
 Antiq. ι. 16, ρ. 47 Bekker : του τίροΰωτέρω χωρεΐν της νυκτός αυτόν 
 αφαιρούμενης. Similarly in Demosth. adv. Steph. i. p. 1128, 1: τον 
 άφηρημένον is used absolutely with something like εμε το λαβείν 
 δίκην Λαρά τούτου to be supplied. 
 
 316 On the terminations of compound words, we must refer the 
 reader to Lobeck's Parerga (ad Phrynich. pp. 487 foil.), as we do not 
 mean to add anything to what we have said on terminations in 
 general. It will be better in this place to discuss, with some minute- 
 ness, three compound words, which have given much trouble to philo- 
 logers, and which are respectively of great interest in the literary, 
 political, and philosophical history of Greece : these three words are 
 δι%'ύραμβος, καλοκαγαθίας (along with which we shall consider the 
 other Greek words of a cognate meaning), and εντελέχεια. 
 
 317 (1) ^1%'ΰραμβος. We have before expressed our opinion 
 with regard to this much-disputed word, but as we were then obliged 
 to confine our remarks to the limits of a note, we may be permitted 
 to repeat them here, in a more expanded and systematic form. 
 
 In analyzing a word which we do not understand, but which be- 
 longs to a language, the etymological principles whereof are reducible 
 to order and system, the first step naturally is to discover what is the 
 termination of its crude-form, if it has any constant pronominal affix 
 between the root and the case-ending; if not, whether there is any 
 compound word, the last part of which corresponds to the word in 
 question. Now, although the ending of διθ'ύρ-αμβος does not coincide 
 with any of the pronominal suffixes which we have mentioned above, 
 there are two words which are strikingly analogous to it in termina- 
 
534 COMPOUND WOEDS. [bOOK ΠΙ. 
 
 tion, namely, ϊαμβος and Ο-ρί-αμβοζ- It is incumbeut on us, then, in 
 the first place, to inquire what is the force of the termination -αμβος. 
 Nothing is more common in Greek than the appearance of μ before 
 labial-endings. It seems to be a sort of anusvdra insertion, which is 
 peculiarly agreeable to the Hellenic ear. The following instances 
 will make the fact sufficiently evident. We have ^ά-μ-βος as well as 
 ^αυ-μα (0•α/-μα), τε-^ητΐα-, ^-ρό-μ-βος as well as τρ8φ-ω; κόρν-μ-βος 
 as well as %ορν-φή ; κό-μ-7tog (e. g. οδόντων) as well as κότΐ-τω ; κν-μ- 
 βος, Sanscrit kumbha, as well as κντΐ-τω and κνφ-ός] 6τρ6-μ-βος as 
 well as (ίτρεφω; andinHomer(Oii2/55.iv.84), a certain people are called 
 Έρεμβοί, probably the Indians, as the Scholiast writes: ot δε Tovg 
 ^Ινδονς τίαρα το έρεβος, μέλανες γαρ, ο&εν καΐ ΚράτηςΈρεμνονς 
 γράφει. If so we may compare the word with έ-ρέφ-ω, ο-ρφ-νόξ^ 
 ο-ρφα-νόξ,ερεμνοζ. By a similar insertion of ^, i,'ci-^-|3og is formed from 
 the root of Ιάτί-τω, but not with the secondary meaning, which some 
 people have given to it, namely, that of satirizing and abusing, as 
 Sophocles says (^^«λ?, 496): λόγοίζ Ιάπτειντινά, "to make a person the 
 butt of one's jeers" (comp. %ενειν ονείδει, iEschyl. Sept. c. Theb. 388). 
 It w&s first pointed out by Lobeck, that the verb Ιάτίτω is merely 
 another form of ατΐτω, "to join," in which sense it is used of joining 
 hands in the dance, or of the dance generally, as in Sophocles (4/^•^, 700) : 
 
 φάνη^, ώ ^'εών χοροτΐοί αναξ, 
 οτίως μοϋ Ννύια Κνώΰυ ορχήματ 
 αντοδαη ξ,ννών iatl^jjs• 
 
 The authorities which Lobeck has collected in his note on this pas- 
 sage leave no doubt whatever as to the fact, that Ιάτίτειν is here used 
 in the same sense in which we have χορον ατίτευν (^schyl. Eumenid. 
 307), χείρα χειρί ύννάτίτειν (Aristoph. Thesmoph. 995), and nectere 
 brachia (Ovid, Fast. vi. 329). That Ιάτίτειν was used as a synonym 
 for ατίτω, appears from the words of Hesychius, "Ια^εν. ετίεμιΐ^εν. εβα- 
 λεν. ϊδωκεν. εδηΰεν, where the editions wrongly read εδευΰεν: com- 
 pare Hesych. α-ψω. τΐροβεγγίΰω. δήύω. Κρατίνος. That the ideas of 
 "joining," "touching," "throwing at," &c., are related, needs no proof; 
 it is sufficiently clear from the use of μετά to signify both "with" and 
 "following after:" and that the υ may be prefixed, even when another 
 form of the word wants the aspirate, is established by the instances 
 which Lobeck quotes : namely, ανω, ίανω ; αΐβοΐ, Ιαιβοΐ] ατίάλλω — ατίο- 
 Λεμπω compared with υάλλω and εφιάλτης = δ εΛίτίηδών; αειρε, Ιάειρε — 
 Λρόΰφερε; ίώλκα=^ αύλακα', ηδοναί, ίηδόνες=ενφρο6νναΐ', ονρος, 
 ίωρός; αν%Ός, ϊον^ος-, ώκνς, Ιωκψ, ^'ώμη, Ι^ώμη-,^νω, ίΟ'ύω. Com- 
 pare also the Persian yd/ten with the Sanscrit root dp-. In the sense "to 
 kindle," "to set on fire," we may compare α;ττω with deco, "to bind," and 
 
CHAP. VI.] COMPOUND WOEDS. 535 
 
 δαίω, "to burn," with δήμος and δημός^ and with the English phrase "to 
 catch fire." From all this, we conclude that ϊαμβος, or αμβος, is sim- 
 ply a word designating a procession or dance of people in close order, 
 and, by implication, a song or hymn performed by such a body. 
 
 318 Having now ascertained the value of the termination, we 
 proceed to investigate the rest of the word. It appears that ^l^v- 
 ραμβος is not only a name of Bacchus, the god in whose honour the 
 song or hymn was chanted (Eurip. Bacch. 526), but also a very common 
 proper name (comp. Herodot. vn. 227, with ^lian, Var. Hist. vi. 2). 
 We believe that in this use it was only an epithet derived from the 
 song, the subject of which was originally the birth of Bacchus (z/to- 
 νύβου γ8νε6ίς...δι^νραμβος λεγόμΕνος, Plato, Legg. m. p. 700 b), and 
 we think the following considerations will show that the name itself 
 properly refers to this mythological incident. The old legend states, 
 that Bacchus, as soon as he was born, was surrounded with ivy-boughs, 
 and so prevented from sharing in the fate of his mother. Thus 
 Euripides says (Phoeniss. 650): 
 
 Βρόμίον svd'a τεκετο 
 
 μάτηρ, z/tog γάμοίς, 
 
 κίΰϋος ov περιΰτεφης 
 
 ελίκτος εν^νς ετυ βρέφος 
 
 χλοηφόροίΰίν 'ερνεύίν 
 
 καταύκίουΰίν ολβίΰας ενώηΰε, 
 
 Βάκχων χόρενμα 
 
 7ίαρ%ενοι6ι Θηβαίαίΰί 
 
 καΐ γνναίξίν Ενΐαις. 
 We look upon this passage as an approximate explanation of the word 
 ^υ^νραμβος. The Scholiast says : οντυνα^υόννΰον κίβΰος εξωΟ'εν τίερι- 
 %λακε\ξ ετί βρέφος οντακατα τον νώτου εκάλνφεν. ίΰτορεΐγαρ Μνα- 
 ΰεας, otl, Καδμείων βαΰίλείων κεραννω^'εντων,κοΰβός, ττερί τους κίο- 
 νας φνείς, εκάλνφεν αντον οτίως μη αυθημερόν και εν μηδενΐ το βρέ- 
 φος δίαφ^'αρ^ καλνφ%εν κι66ω, δώ καΐ Λερίκιόνιος 6 ^εος εκλήθη 
 τίαρα Θηβαίοις. The following are further illustrations of the tradition, 
 that the young Bacchus was enveloped in ivy. Homer, Hymn xxv. 9. : 
 ανταρ ετίειδη τόνδε Ο'εαΙ τΐολννμνον ε^ρει\)αν, 
 δη τότε φουτίξεΰκε κα^' νλήεντας ένανλονς, 
 κιύβω καΐ δάφνη πεπνκαΰ μένος. 
 Nonnus, Dionysiaca, ιχ. 1 1 : 
 
 τον μεν ντΐερκύ'ψαντα %'εηγενέος τοκετοΐο 
 ΰτέμματί κιύβήεντι λεχωΐδες 'εΰτεφον ^ίΐραί, 
 έύόομένων κήρυκες. 
 
536 COMPOUND WORDS. [bOOK III. 
 
 Philostratus, Imag. i. c. 14 : ό δεζίίόννβος τηζ μίν μψρός εκ^'ρώΰκεί, 
 ραγείΰης την γαότέρα, το δετίϋρ άχλνώδεςεργάξεταί^ φαιδρός αντοξ 
 οίον άΰτήρ τΐξ αΰτρά^των. διαΰχονΰα δε η φλοξ αντρον τι τω Αιο- 
 ννΰω ΰκιαγραφεΐ, παντοζ ηδων Άΰόνρίου τ ε και Ανδίον. έλικες τ ε 
 γαρ ττερί αντω τε^ήλαβι, και κιττον κόρνμβοι, καΐ ηδη αμτίελοι και 
 %^νρΰου δένδρα, ούτω τι εκονΰης άνάβχονται της γης, ώς και τω 
 τίνρΐ είναι ενια. As it is stated that the Dithyramb was a song relating 
 to the birth of Bacchus, and as it appears that, according to the tradi- 
 tion, the infant god was surrounded with ivy, it would not be unnatural 
 to seek for some connexion between the ^νρΰος, or ivy-staff of the Bac- 
 chanalians, and the Dithyramb, which Simonides of Ceos calls κκίίΐοφόροί 
 (fragm. 205, cd. Schneidewin). The word 0"'ΐ;ρ-(9θ^ evidently consist of the 
 root &νρ=^^ρυ or &ρι, and the termination -ύος, and the analogy οΐτίνρ- 
 60ς. This termination, we have seen, implies a collection, as in ο-ΰος, 
 %ία-(3ος, &c. Now the %νρ-6ος was a light staff, surrounded with ivy and 
 vine-branches, and terminating in a pine-cone ; i. e. %ύρ6ον χλοερον 
 κωνοφόρον κάμακα, as the epigram aptly describes it. It would be 
 necessary, in order to explain fully the significance of this symbol, that 
 we should investigate the whole question of the worship of Bacchus. 
 As this would be foreign to our present object, we shall merely men- 
 tion, that Bacchus or Dionysus, as the Sun-god, coincides with the Baal 
 or Moloch of the Phoenicians ; as the god of generation with Baal-Peor, 
 whence the symbol of the Phallus used in his worship. Like Hermes, 
 he presided over the friendly intercourse of strangers with one another, 
 as Diodorus says, in, c. 64: {τον ^ιόννΰον) πανταχού πανηγύρεις 
 αγειν και μονβικούς αγώνας ΰνντελεΐν, και το ύύνολον ΰνλλνοντα 
 τά τε νείκη των ε^νών και πόλεων αντί των ΰτάόεων και των πολέ- 
 μων ομόνοιαν και πολλην είρήνην παραΰκευάζειν ] and he is ex- 
 pressly represented as a travelling god, surrounded by attendants 
 bearing thyrsi (Diodor. ubi supra). In the oldest ages of Greece, 
 when a stranger came to the coast, or in general one person applied 
 to another for protection or hospitality, — and both these classes were 
 designated by the same name, ικέτης, "the comer," — it was customary 
 for the suppliant to bear in his hand an olive-bough, surrounded with 
 Λνοοίΐβη bands. And thus the Danaides are introduced by ^schylus 
 (Supplices, ad initium) as singing, on their arrival in Greece : 
 
 Ζευς μεν άφίκτωρ επίδοι προφρόνως 
 
 ύτόλον ημέτερον — 
 
 τίν αν ονν χώραν ενφρονα μάλλον 
 
 τηύδ' αφικοίμε^α 
 
 6νν τοΐΰδ' ικετών έγχειριδίοις 
 
 έριοότέπτοιΰι κλάδοι6ιν; 
 Nothing could be more natural than that a person landing on a strange 
 
CHAP. VI.] COMPOUND WORDS. 537 
 
 coast, where every foreigner was expected to be a pirate, or, generally, 
 an enemy, and was consequently received and treated as such, should 
 seek for some symbol of peace, some token which would prove to the 
 inhabitants of the country that his intentions were friendly. The 
 most obvious symbol would be the bough of some tree, the one pre- 
 ferred being of course some one of sacred use, the olive for instance; 
 and as fillets of wool were generally employed in the religious rites of 
 the Greeks, it would not be unnatural that they should add these to 
 the bough. In the course of time, when the conventional sign was 
 better understood, they would carry with them a white or gilded stafif, 
 which they would surround with wool of foliage, as the occasion 
 served. "This," says Bottiger (Amalthea, i. p. Ill), "is the proper 
 and original κηρνκαον, the staff of Hermes and the herald, and the 
 only one which occurs in the oldest writings and statues. By degrees, 
 several slight alterations were introduced. In common use the green 
 bough was confined to the suppliants for purification. Wound round 
 the staff it becomes the Bacchic thyrsus. The bands, however, re- 
 mained; only they did not flutter round at random; they were tied 
 up into two bows or loops, and these bows gave rise to the idea of two 
 serpents tied into a knot and looking towards one another." Hence, 
 the Romans called the symbols of the suppliant supplicia; Sallust, 
 Jugurth. c. 46: legatos ad consulem cum suppliciis mittit. These sup- 
 plicia were boughs of some sacred tree (verbena) bound (supplicata) 
 round a staff, so that supplicare does not mean "to bend the knee," 
 but is derived from these bandages on the suppliant's staff (see Virg. 
 ^n. vn. 237; Hor. m. Carm. xiv. 8). This connexion with the cadu- 
 ceus, or symbol of peaceful intercourse between strangers, is, however, 
 only a part of the significance of the thyrsus. Bacchus was not only 
 a travelling god and the god of travellers, but also the god of genera- 
 tion; and it is natural to suppoee that his peculiar symbols would refer 
 to his functions or to the legend about his birth. The thyrsus was of 
 two kinds: it was either a staff surmounted by a cone and surrounded 
 with ivy, in which case we might suppose that it was a rude imitation 
 of Bacchus Pericionius, the staff representing the body, and the cone 
 the head of the infant god ; or it was a bare staff surmounted by a 
 cone which is wrapped up in leaves, in which case we must adopt the 
 explanation of Eustathius (p. 84), that it was the heart of the embryo 
 god which was carried to Jupiter by Pallas — την τον άμβλώματοζ 
 καρδίαν ηνεγκε (Πάλλαζ) τω ^d' — and must, therefore, suppose that 
 this heart was represented by the cone of the thyrsus. According to 
 this latter view, which we regard as the true one, the staff and cone 
 constituted the thyrsus, or emblem of the birth of Bacchus: the ivy 
 round the staff or νάρ^'ηξ was perhaps an adjunct borrowed from the 
 
538 COMPOUND WORDS. [bOOK III. 
 
 Λερϊ κίονας φνείς of the legend, and also imitative of the olive-bough 
 round the wand of peace. We consider the root of %^νρ-6θ£, namely, 
 ^VQ=d^QV- or d'Qi-, to be the same as that which we find in Q'qlov. The 
 word '9'ρίό|3ολθί isused as a synonym ίοΟ^νρΰόφοροςοτναρχ^ηκόφορος 
 (see the passages quoted by Lobeck, AglaophamuSj pp. 813 sqq.), and 
 Ο'ρίαμβος ajopears in the same sense as διθύραμβος-, so that it is at 
 least probable that the elements 0•ρί- and Ο'νρ- found in these words 
 respectively, are in fact, as they may be according to the laws of ety- 
 mology, one and the same. Now the word %ριαί denotes the sooth- 
 sayer's lots, which, like the dice, were probably used in triplets, and 
 perhaps consisted of leaves (Hesych. ^ριάζειν' φνλλολογεΐν) ; ^'ρΐον 
 means "a leaf of the vine or fig," and "an olio wrapped up in such 
 leaves;" Hesychius; θ ρ Γα*, φύλλα όνκης η άμπελον. καΐ τα εν αύτοΐς 
 δεΰμούμενα βρώματα. Both words therefore are connected with the 
 numeral τρεις, — ^just as Ο'ρΐνα^ stands as a synonym for τρίαινα, — 
 and Ο'ρΐον may point, in this application, to the shape of the vine and 
 fig-leaf. For this reason, perhaps, the Cretans called the vine %'ρίνία ; 
 Hesychius: %ρινία. αμτίελος. εν Κρήττ}. In the plural, ^ριοί means 
 the three stays {εκφοροι) going from the topmast to each side of the 
 ship and to the bowsprit. The third pronouns appears still more 
 obviously in the reduplicated synonym τερ-^ριος. In its application 
 to the thyrsus, this word points to the mystical number "three" 
 (Bottiger, Amalthea, i. p. 137), which is also referred to in the epi- 
 thet τρντΐετηλος, applied to the mystical wand which Apollo gave to 
 Hermes: Homer, Hymn, in Mercur. 530: 
 
 αντάρ ετίευζα 
 ολβον κα\ πλούτου δώύω περικαλλεα ράβδον, 
 χρνόείην, τριπετηλον, άκήριον, η ύε φυλάξει, ' 
 
 in the τριετηρίς of which Bacchus was the inventor (Cic. Nat. Deor. 
 III. 23), and perhaps also in the epithet τρικέφαλος applied to Hermes 
 (Harpocration, s. v. p. 334; Etymol. Magn. s. v. p. 694). If, then, 
 %ύρ6ος is connected with ^ίρίον, it will signify "a gathering of three 
 leaves, and that which they contain, at the end of a rod," just as τίνρ- 
 ΰύς denotes "a torch," i. e. "a gathering of fire at the end of a piece 
 of wood." Hartung (Class. Mus. vi. 372 sqq.) proposes to connect 
 διθύραμβος with Ο-όρυβος. If the one were really a by-form of the 
 other, we should have χΙ'όρυμβος, not ^ύραμβος, (cf. κόρυμβος, ϊϋ'υμβος, 
 &c.). As, however, the dithyrambic dance was called τυρβαύία (Jul. 
 Poll. IV. 104), and as the roots %υρ-, %'ρι-, as well as ΟΌρ-, Ό'ρο-, might 
 be connected with that of τύρβη, turba, from which τυρβαύία is formed, "" 
 a question might arise, whether the name of the %ύρ-ΰος was derived 
 from the tumultuous clamours {^ρόος, %'ροεω, %'ρύλλος, &c.) of the %ία(5ος 
 
CHAP. VI.] COMPOUND WOEDS. 539 
 
 of Bacchus ; or whether it was expressive of the symbolical meaning 
 of the Bacchic staff with its accompaniments: and this would lead to 
 the inquiry, whether d'vQ-oog signified primarily the distinctive staff 
 (of. 7tVQ-(jog) or the party who bore it (cf. ^ice-dog), according to the 
 transition in the Latin manipulus (Ovid, Fast. ni. 117). In our 
 opinion, we must refer to three distinct origins, the ^VQ-uog or three- 
 leaved emblem (Ο'ρί-), the &ία-ΰος or sacred band (d'sog), and the τνρ- 
 βα-6ία or noisy crowd (d'QO-, dOQ-). 
 
 319 The quantity shows that the first syllable oi zίL-Q'VQaμβog is 
 a contraction oizlit, like that ofz/t-JtdAta, ζ/ί-φί-λο^, andz/t-(Jωr?Jρ^o^' 
 Bekker. Anecd. i. p. 91), and thus the whole word will signify "a 
 chorus or song celebrating the birth of Bacchus," i. e. "the bringing 
 to Jupiter of the %'qiov or leaf-enveloped heart or body of the god." 
 The termination, as we have explained it above, is particularly applic- 
 able to the dithyramb, which was performed by a chorus of fifty 
 persons dancing hand in hand {αλλήλων επί καρπω %BlQag sxovTsg, 
 Homer, Iliad xvni. 594) round a blazing altar in honour of the 
 sun-god. 
 
 320 It is right to mention an explanation of the thyrsus adopted 
 by some of the more recent Greek writers and by the Roman poets. 
 They supposed that the thyrsus was a spear concealed in ivy, or 
 having the point covered by the cone: thus Justin Martyr (quoted 
 by Schneider): ώβτίερ ai βάκχαί δια ύχήμaτog είρηνίκον tag λ6γχag 
 εν rolg %ύρ(5οι^ Λεριφερονβι. Catullus, lxiv. 257 . 
 
 Horum pars tecta quatiebant cuspide thyrsos. 
 Seneca, Here. Fur. (quoted by Schneider) : 
 
 Tectam virenti cuspidem thyrso gerens. 
 
 But Ovid {Metamoi^h. xi. 9) distinctly implies that the thyrsus had 
 no point: 
 
 Hastam 
 Quce foliis prcesuta notmn sine vulnere fecit. 
 
 and the learned Yirgil says {jEneid vn. 396) : 
 
 Pampineasque gerunt, incinctce pellibus, hastas^ 
 which he calls (v. 390) molles thyrsos ', Lucian, too (Bacchus, c. i. 
 p. 292Lehm.), describes the Bacchanalians as κίττω εΰτεμμέναί, νεβρί- 
 dag ενημμεναι, δόρατα ηνα μικρά 'εχονΰαι, άΰίδηρα, κιττοΛΟίητά και 
 ταντα-, where, although he calls the thyrsus a spear, he does not give 
 it a point; and Diodorus (iii. c. 64) distinctly says that they some- 
 times had spears covered with ivy "instead οι thyrsi:'''' άναδίδόναί γαρ 
 talg βάκχαι^ αντί των ^υρύων λόγ%ag τω κιττώ κεκαλνμμένας 
 
540 COMPOUND WORDS. [bOOK III. 
 
 την άκμην τον ΰίδηρον. So that it appears probable that this notion 
 of the thyrsus Avas a misconception resulting from a legend, that the 
 covered spear was sometimes substituted for the peaceful emblem 
 of the Bacchanalians. The νάρϋ'ηξ itself would be much too light 
 for a spearshaft. 
 
 321 (2) KaXoKuyad'og^^. The general meaning of καλδ^ κά^^ο:- 
 ^og has long been well known, and Englishmen are in the habit of 
 congratulating themselves that their word "gentleman" is the only 
 modern term which precisely expresses the meaning of the Greek 
 compound; an opinion in which foreign writers seem, on the whole, 
 to concur. But it does not appear to have been satisfactorily shown 
 how a comination of these two simple adjectives came to denote such 
 a complex idea, nor have the moderns sufficiently distinguished be- 
 tween the political and moral use of the word. For instance, one of 
 the greatest and most learned of our theologians supposes that the 
 use of καλοκάγα^Ία, to signify moral perfection, is derived from the 
 literal subsequent meaning of its two component parts. "That 
 which is good in the actions of men," says Hooker (Ecclesiast. Polity, 
 I. § 8, p. 281 Keble), "doth not only delight as profitable, but as 
 amiable also. In which consideration the Grecians most divinely 
 have given to the active perfection of men, a name expressing both 
 beauty and goodness; because goodness, in ordinary speech, is, for the 
 most part, applied only to that which is beneficial." Writers of our 
 own time have given still more vague and general interpretations'}*. 
 We think, therefore, that it will not be unadvisable to examine, with 
 more minuteness than has generally been done, what is the primary 
 meaning of each of the adjectives καλός and ayad'og, what is the sig- 
 
 * There is a difference of opinion as to the proper mode of writing 
 and accentuating this combination of three words. Some write καλοκά- 
 yadOS, with the accent drawn back and without any mark of crasis. Lobeck 
 has καλοηάγαϋΌζ in his text (ad Phryn. p. 603) and ν,αλον,αγαΟ^ός in his index 
 (p. 805). Ast has y.aXov.uyaQ^ia in his text (Plat. Def. 412 e), but ■Λαλον,άγαΟ'ία. 
 in his Lexicon Platonicum (ii. p. 133). Meineke has v.aXoy.aya%Biv in his 
 Fragm. Com. ii. p. 1033. But Jacobi in the Comicce Dictionis Index, p. 505, 
 has ν,κλοκάγα&εΐν. We prefer the orthography given in the text, for as 
 long as the second adjective retains its inflexions, the word is not only 
 parathetic, but separable in the plural and in the other cases. And we 
 would retain the apostrophe even in the adverb ηαλογ,άγά&ως and in the 
 derivatives καλοκάγα&ία and καλο%άγα&ι>ιός. 
 
 f Thus Passow says (Meletemata Critica in jEsckyli Persas, p. 31, ed. 
 Bach): '■'■Interior civilis bellicceque prudentice concentus, qui sub nomine καλο'κά- 
 yad-iag uno vocabulo cotnjjlectebatur." Delbriick's definition is still looser 
 (see Philol. Mus. i. p. 503). 
 
CHAP. VI.] COMPOUND WOEDS. 541 
 
 nificance of this compound as a political term, and what its applica- 
 tion as a moral epithet. 
 
 322 With regard to the second adjective, ayad'og, little remains 
 to be done: Welcker, in his admirable introduction to Theognis 
 (pp. xxi fol.), has collected nearly all the passages bearing on the 
 subject, and has clearly shown that the Greek ayadOi, as also αριΰτοί, 
 «ρκίτζε^, &c., the Latin boni, optimi, optimates, and the old German 
 Rachinburgi, gute Manner, Gudemanner, guden Manne, and Herrn 
 von Bechte, are names of the nobles, the men of rank, and of good 
 family, in a state. To which, in Greek, κακοί, δειλοί, as epithets of 
 the common people, are regularly opposed ; an opposition which has 
 taken such deep root, that it is even preserved in compounds : e. g. 
 κακότίαχρίς (Alcaeus, fr. 9), and κακογείτων (Sophocl. Philoct. 688, 
 Herm.). Of the last word Welcker writes as follows (Rhein. Mus. for 
 1833, p. 450): ''^κακογείτων is a low-born, common neighbour^ a poor 
 native dwelling by a man of the noblest extraction, which is also 
 mentioned v. 180: 
 
 οντοξ τΐρωτογόνων ϊύως 
 οϊκων ovdevog νβτερος. 
 
 In the verses below, the low-born native finds his opposite, when the 
 chorus says: vvv δ' ανδρών αγα%ών τίαιδοζ ν7ίανΧΎ\6αζΓ In the 
 same political sense as oi αγα%οί, the Greeks used ol ε6%λοί, oi 
 βελτίονς, το βελτι6τον, οί βελτιύτου, as opposed to the δημοζ (Xe- 
 nophon, passim). An older word of the same import was άχαΐα (Αά- 
 κωνες, άγαΟ'ά, Hesych.), χαός, χαως, χάϊος, whence, according to 
 some, the names of the Achseans (= αγαθοί, άρίύτψζ' Muller, Prole- 
 gom. zur Mythol. p. 291. Comp. Journ. of Educat. iii. p. 87; Philol. 
 Mus. II. 88: see however Phil. Mus. ii. 367: above, p. 165) and Chao- 
 nians (Welcker, ad Theogn. p. xxviii, note) were derived, just as the 
 name of the Goths was derived from goths, goda, "good" (Savigny, 
 Gesch. Rom. Rechts, i. p. 194). 
 
 f 
 
 323 The derivation of αγα%όξ is a great stumblingblock to 
 
 etymologers. Bopp would connect it with the Sanscrit agadha-s, 
 "deep" (Vergl. Gramm. p. 441); this we consider undoubtedly erro- 
 neous. Pott's suggestions* {Etymol. Forsch. ii. p. 299) do not merit 
 the slightest attention, nor can we say much in favour of Passow's 
 derivation from άγαν. We consider the first letter to be one of those 
 
 , * Surely he is joking when he proposes to regard it as a compound 
 of κγαν and καϋ-αρόςΐ 
 
542 COMPOUND WOEDS. [bOOK III. 
 
 movable initials, of which we have already spoken more than once, 
 and we class all the following words together; ά-γα-^ός, η-γά-^Έος, 
 γη-^'έω, ά-γανός, α-γα-μαί, ά-γαίο-μαυ, ά-γαν-ρός, γαν-ρος, γαν-ριάω, 
 Latin gaudeo, γά -vog, γά-ννμυ, yaUiv^ γα^^υάδας{ηρωθ£ όνομα οξ καΐ 
 τους καταφεύγοντας dg αντον ρνεταο εκ θανάτου, Hesychius), ^^«^εΓι; 
 γάδεΰ^'αί, γαδεω (χαρά^ Hesych.) The meaning which runs through 
 most of these words is that of "pleasure," *'joy," "delight:" α-γαμαι, 
 into which the idea of "wonder," &c., frequently enters, derives this 
 meaning from a primary one of pleasure, for the wonder implied is 
 always considered as a pleasurable sensation; and the word really 
 signifies in an infinity of passages, as well in the most ancient as in 
 the more recent authors, "to be pleased with," "to delight in," "to 
 think highly of." We have before shown how the synonymous root 
 χα-ρ- derives all its meanings from the primary one of "containing;" 
 thence, "support," "firmness," &c. We find this root with a set of 
 formations corresj)onding in the main to those of the root γα-. The 
 primary meaning of the root χα- or χαΓ is "containing" (χά-ω, χάν- 
 νος, &c.), thence, "firmness," "hardness," the earth (χε-ρ-6ος, χο-ρός, 
 χώ-ρα-, &c.), thence, help or assistance in battle, and pleasure in such 
 assistance (χάρις-, &c.), then it becomes the epithet of a person who 
 can so help us (χρ'ή(3ΐμος), and finally of an order in the state, com- 
 posed of the best warriors or chief men (χρηβτοί , &c.). To this last 
 meaning belongs the old word χαός (or χαίός, Lobeck, Phrijn. p. 404), 
 where the termination -ρος is omitted. Similarly from γα-^ we have 
 the primary idea of firmness or support, the earth (γαία, γη); assist- 
 ance in battle (as in the patronymic γα-Ο^άδας), joy, pleasure in gene- 
 ral (in most of the words quoted above); and thence the epithet of a 
 warrior, a person able to help in battle, and the upper class in society, 
 which was composed of such warriors (άγα%Όί, &c.). We do not pre- 
 sume to say that the roots γα- and χα- are identical, though this is 
 not impossible ; it is, however, important to observe this correspond- 
 ence in their applications. The class of nobles being continued by 
 hereditary descent, άγαΟ'ός came to express not only the bravery, 
 which was one of their qualifications, but also their good descent, 
 which was another, so that άγαΟ'ός became a synonym for ευγενής', and 
 thus Hesychius (s. v. άγα&όν) rightly defines the epithet as signifying 
 both ανδρείος and γενναίος', conversely, ^ει/ι/αΓο? is used as a synonym 
 for άγα&ός, in the meanings "brave," "strong," "great" (in Sophocl, 
 Aj. 938 : γενναία δνη is explained ίόχνρά by the Scholiast), just like 
 gnavus, ingens, from gigno, in Latin. Another obvious quality of 
 the nobles was their wealth, and thus we find as synonyms for the 
 άγα%^οί, such adjectives as oi ηλούβιοι, oi ενΛοροι^ οί τα χρήματα 
 έχοντες.) οι τΐάχεες, and (because it was considered a great mark of 
 
CHAP. VI.] COMPOUND WOEDS. 543 
 
 wealth to keep a horse, Aristot.PoZii.iv. 3), ίτΐτΐεΐς^υτΤΛοβόταί, ίτίπόδα- 
 μοο,&ο. InThucydides theyare called δυνατοί (i.2 A), ονχ οι άνδννατώ- 
 χατοι (ι. 5), and wealth is called δύναμίξ (ι. 2). In opposition to these 
 terms, the lower orders are called τίένητες, and thus, in the words of 
 Aristodemus, as reported by Alcaeus, — "it was money that made the man, 
 there was no goodness nor honour in the poor" (χρήματ άνήρ, Λενυχρος 
 δ' ουδείς τΐέλετ leXog ούτε τίμιοξ. Αρ. Schol Find. Jsthm.ii.ll). But, 
 in the matter of wealth, that a man might be really a gentleman, he must 
 have inherited his riches, so that the other qualities of the nobles are 
 presumed in their epithets implying opulence. For instance^ in JEschy- 
 lus {Agamemn. 1010) we read : άρχαυοπλούτων δεβποτών πολλή χάρυς, 
 which is thus explained by Aristotle {Bhet. ii. 32): διαφέρει δε τοις 
 νεωΰτΐ κεκτημένους καϊ τοις τνάλαο τα ηΟ'η, τω ατίαντα μάλλον καΐ 
 φαυλότερα τα κακά εχειν τους νεόπλουτους^ ωΰπερ γαρ απαιδευβία 
 πλούτου εύτι το νεόπλουτον ειναυ. From the tithes, &c., which they 
 received, the nobles were called δωροφόροι, from the lands which they oc- 
 cupied ^fcofto^Oi-jfrom their conspicuous position γνώριμοι or '^notables." 
 
 324 The adjective καλός^ which has the penultima long in Homer 
 and the old epic poets, stands for καδ-λός, and is connected with κε- 
 καΰ-μαι, κε-καδ-μαι^ κε-καδ-μένος, κάδ-μος. Doderlein, to whom this 
 etymology is due, justly remarks (Lat. Syn. und Etym. m. p. 38) that 
 it may be compared to εάνός for εαδνός-, and the more so as δλ does 
 not belong to the Greek ecphoneses : and in another place (iir. p. 97) 
 he shows that κα-λός and καί-νός (καί-νυ-μαι is the ordinary form of the 
 present for κεκαύμαι) are connected, as canus, candidus with re-cens 
 and as δεί-νός with δει,-λός from δείδω. It is possible that κάλος may 
 have been written originally with a doubled λ like κάλλος: compare 
 bellus for henulus. The primary meaning of the v»rord is in strict ac- 
 cordance with this derivation; it signifies "furnished with outward 
 adornments," in general, "that of which the outword form is pleas- 
 ing," and thus it is regularly opposed to αίΰχρός, especially in Plato 
 (Hipp. Maj. 289 A, Protagor. 332 c, Sympos. 183 d, &c. &c.), and 
 aί6χρoςκalκaκόςiso^lposeάtoκaλόςκaiάγa&ός(PlεL•io,Sympo$.20lΈ,). 
 But to the Greek notion of κάλλος something beyond mere outward 
 garnishing of the person was required; it was not a languishing 
 beauty, a listless though correct set of features, an enervated volup- 
 tuousness of figure, to which the homage of their admiration was 
 paid. It was the grace and activity of motion which the practice of 
 gymnastic exercises was calculated to promote — the free step, the 
 erect mien, the healthy glow, combined with the elegances of conver- 
 sation and the possession of musical accomplishments ; it was in fact 
 the result of an union of the μουβική and γυμναστικής of which their 
 
544 COMPOUND WOEJDS. [bOOK III. 
 
 education was made up. It was this that constituted beauty in the 
 Greek sense of the word — the educated man alone was considered 
 καλός; thus Jischines says(m Ctesiph, ad βη.) : Λαίδεία, y δοαγινώΰκο- 
 μεν τα καλά καΐ τα αίΰχρά, and in the Attic writers in general this 
 adjective is used as an epithet of persons distinguished by their ac- 
 complishments (see the passages quoted by Heindorf ad Plat, Hipp, 
 ^^j' § 1)• Now the people of rank and wealth were always then, as 
 they are now, most able to obtain the advantages of education ; they 
 had more leisure than the common people to devote themselves to 
 those exercises which were calculated to produce grace and ease of 
 motion and the other accomplishments necessary to the gentleman; 
 and as the aristocracy, like the knights of the middle ages and the 
 duinhe -was sals in a Highland Clan, owed much of their reputation 
 for superior valour to their being better furnished with arms, and, 
 from leisure and practice, more skilled in the use of them, so they de- 
 rived their superior accomplishments in music, dancing, &c., from the 
 same source; the best dancer and the best fighter were synonymous, 
 the first in the chorus and the foremost in the battle array were the 
 same persons ; they were the nobles, the pre-eminently καλοί και άγα- 
 QoL It Λvas from this that they are called χαριεντες, and the same 
 idea is clearly seen in the use of κόόμος as an old political term. 
 
 325 If there were any doubt with regard to what we have just 
 stated, it would be removed by the following passages. In fact the 
 καλοί κάγα&οί are actually described according to this definition by 
 Euripides (Ion, 598): 
 
 060L• δε χρηβτοί, δυνάμενοι τ είναι ΰοφοί, 
 where χρηΰτοί is, as Λve have seen, another name for the ayadOi, and 
 δυνάμενοι είναι βοφοί expresses that facility of acquiring knowledge 
 and accomplishments by the attainment of which the nobles became 
 καλοί. Again, in the same author (Alcestis, 605): 
 
 το γαρ ευγενές εκφέρεται τζρος αΐδώ^ 
 
 εν τοις άγα%Όΐ6ι δε 
 
 πάντ ενεΰτιν βοφίας. 
 
 Here the αΙδώς is that sense of honour which is the natural accompa- 
 niment of gentle blood, and to which the nobles owed their innate 
 valour; thus Ajax says {Iliad xv. 561): 
 
 'Si φίλοι, ανερες εύτε καΐ αΙδώ %'έΰ^' ενϊ 0Ί>μω, 
 αλλήλους δ' αΐδεΐΰ^'ε κατά κρατερας ϋΰ μίνας, 
 αίδομενων ανδρών τΐλέονες ΰόοι ηε τΐεφανται, 
 φευγόντων δ' οϋτ αρ κλέος ορνυται οϋτε τις άλκή^ 
 
 and Archidamus, in Thucydides (ι. 84), distinctly attributes the 
 
CHAP. VI.] COMPOUND WORDS. 545 
 
 bravery of the Spartan aristocracy to the possession of this quality : 
 "our good κόΰμος,'' says he, "makes us πολεμικοί, on αΙδώς όωφρο- 
 uvvTjg τΐλεΐΰτον μετέχει, αίύχννης δε εν^υχία^'' where αΐβχύνη is 
 used as a synonym for αΙδώς, as in ^sch. Sept. c. Theb. 394 : 
 
 μάλ' ευγενή τε και τον αΐβχννηζ Ο'ρόνον 
 τιμώντα, καΐ ότνγονν%'' ντίέρφρονας λόγους — 
 αίΰχρών γαρ αργός, μη κακός δ' είναι φιλεΐ. 
 
 We take this opportunity of explaining a passage in Pindar (Olymp. 
 VII. 44), vt^here Bockh and Dissen, misled by the Scholiast, have 
 mistaken the sense of αιδώς. Pindar is saying that valour (αρετή) 
 and warlike spirit (χάρματα) are produced by that sense of honour 
 (αιδώς) which springs from provident forethought (τίρομηΟ'εύς), i. e. 
 from a careful attention to the rules which regulate the conduct of 
 honourable men ; he expresses the sentiment thus : 
 
 εν αρεταν 
 εβαλεν και χάρματ άνΟ'ρώτΐοιΰι Προμα^έος Αιδώς. 
 
 For the phraseology compare Horn. II. xm. 82: χάρμγ} γη&όΰυνοι, 
 την ΰφιν Ο^εος εμβαλε ^υμω, and see Pind. 01. xm, 16. Προμηθεύς 
 here is not the Titan, but the more general word equivalent to πρό- 
 νοια or φρόνηΰις (see Welcker, Trilog. p. 70, note), and Αιδώς is 
 called the daughter of Προμηθεύς by an allegorical genealogy similar 
 to that in ^sch. Sept. c. Theb. 208 (above, § 305) : 
 
 Πειθαρχία γάρ εΰτι της Εϋτΐρα^ίας 
 μήτηρ, γυνή Σωτήρος' ώδ' έχει λόγος. 
 
 "Obedience produces good fortune, and helps men as much as the 
 preserving Jupiter." The passage in Pindar is introduced as an old 
 saw like this (ώά' έχει λόγος), and should perhaps be printed between 
 inverted commas, like many other aphorisms in that author. The 
 stress is to be laid upon the word Προμα&έος, the fact spoken of 
 being only the want of care on the part of the Rhodians, in not oflPering 
 burnt sacrifices to Zeus and Athena. Similarly, Pindar (Pyth. v. 25) 
 makes Excuse (Πρόφαΰις) the daughter of tardy-witted Afterthought 
 (Επιμα^εος ό'φινόου θυγατέρα)*. 
 
 326 As the wealth of the nobles enabled them to provide them- 
 selves with a more expensive panoply, to appear, in fact, as heavy- 
 armed men with the δόρυ και ξίφος και το καλόν λαιύήιον πρόβλημα 
 
 * By a strange oversight Max MuUer in referring to this language substitutes 
 Apophasis forProphasis, though he uses the proper English Avord Excuse (Ox/. 
 Ess. 1856, p. 44). 
 
 NN 
 
546 COMPOUND WOKDS. [book III. 
 
 XQCOTog (Hybrias ap. Athen. p. 695 f), arms wliicli the poorer classes 
 were unable to buy, and which they were not generally even per- 
 mitted to possess (see Thucydides, in. 27); and as their leisure allowed 
 them to acquire skill in using their arms, and to spend most of their 
 time in the open air, which was essential, in their opinion, to the full 
 development of the bodily powers, all these attributes would become 
 mixed up with the definition of a man of rank, and by taking a part 
 for the whole might be considered as constituting his definition. On 
 the contrary, the poorer people were taunted with epithets derived 
 from their sedentary employments, which took away from the grace 
 of the person; thus Aristot. PoUt. vm. 1 : dio rag ts toiavrag Tsxvag 
 o6ai το ύώμα Λαραΰκενάξονόϋ χείρον δίακεΐόχ^αι, ^avav(5ovg κα- 
 λούμενη cf. I. 11: βανανΰόταταυ [των εργαΰίών], εν aig τα ΰώματα 
 λωβώνται, μάλίΰτα : and it seems that βάvaυΰog was almost regarded 
 as signifying "unmilitary," or at least as denoting some occupation 
 inconsistent with a devotion to warlike exercises ; thus Sophocles makes 
 Menelaus say to Teucer (Aj. 1096): δ τo^ότηg εοίκεν ov ΰμικρον 
 φρονεΐν, to which he answers, ov γαρ βάναυϋον την τεχνην εκ- 
 τηΰάμην; see the defence of the bow in Eurip. Here. Fur. 188 sqq.; 
 cf. Herod, ii. 165: καΐ τούτων βavav(5ίηg ovδεlg δεδάηκε ονδεν^ άλΧ 
 ανεονται Ig το μάχιμον. But although the light-armed troops were 
 necessarily composed of the lower order of people, yet all these were 
 not βάνσνβοί', atleastAristotlesays(PoZii.iv. 3): καΐτών εντίόρωνβε 
 καϊ των ατίόρων το μεν οτίλιτικον το δε ανοπλον,καΐτον μεν γεωρ- 
 γικό ν δημον ορώμεν οντά, τον δ' αγοραίο ν, τον δε βάνανβον. 
 
 As a political term, then, κaλog κaγa^6g implied no particular 
 moral excellence: it is merely the name of the upper class, "the 
 accomplished and well-born," as opposed to the δΎιμog (Thucyd. viii. 
 48; IV. 40). Hence, the Scholiast on Aristides (m. p. 446 Dind.) 
 says: oi μεν γαρ ηβαν καλοί καϊ άγα^οϊ οι καλούμενοι ολιγαρχικοί, 
 οι δε δημοτικοί. Sometimes, indeed, it would seem to imply nothing 
 more than good descent, as when Herodotus uses it as a synonym for 
 Piromis (n. 143). 
 
 327 The application of this compound to denote moral excellence 
 is thus explained by Aristotle (PoUt.iY.S3): εΐώΰ'αΰι δε κaλεΐv—τag 
 7tρog την ολιγαρχιαν μάλλον {άπoκλιvoύ6ag) άρι6τoκρaτίag δια το 
 μάλλον άκολον^'εΐν τίαιδείαν και ενγενειαν τoΐg εντίορωτεροις. 
 ετι δε δοκονβιν ϊχειν οι εντΐοροι ών ένεκεν οι aδικoύvτεg άδικου- 
 ΰιν'ο^ενκαΐ κaλovg κάγa%•ovg και γvωρίμovg τoύτovg7tρo(J- 
 αγορεύονΰιν. The philosopher, however, has reversed the natural 
 order, for the rich were not called καλοί κάγαΟΌΐ on account of their 
 respectability, but conversely, the name of the upper classes, from 
 
CHAP. VI.] COMPOUND WORDS. 547 
 
 their general respectability, came to be used as synonym for "respec- 
 table." A more apt instance would have been the use of εταεικης as 
 a synonym for καλοζ xayad'og. It was because the better classes, 
 having no temptations like their poorer brethren, abstained from those 
 vices which common opinion reprobated, that their regular name be- 
 came an epithet descriptive of good moral conduct : thus Aristophanes 
 says (Ban. 7, 8): των τίολιτών τον ξ καλούς τ ε ή, αγαθούς — ους 
 μίν ϊβμεν ευγενείς %αί βώφρονας άνδρας οντάς κ αϊ δίκαιους^ 
 and iEschines opposes it to φανλος (in Ctesiph. p. 65, 1): ονδ' ο(ίτις 
 ΙβτΙν οϊκοι φαύλος^ ονδεΛοτ ην εν Μακεδονία, κατά την πρεύ- 
 βείαν, καλός καγα%ός. For this sense of φανλος, see Thucyd. vi. 
 18: TO re φανλον και το μέΰον καΐ το πάνν ακριβές. Eurip. Bacch. 
 431 : ro πλήθος οτι το φανλότερον ενόμίβε χρηταί τε. 'Pla.to, JResp. 
 IV. p. 431 c: καΐ των ελεν&ερων λεγομένων εν τοις πολλοίς τε καΐ 
 φανλοις. The older Greeks did not imagine that good descent really 
 proved the possession of moral excellence, though they believed in the 
 transmission of luck, and so forth ; they looked only to the advantages 
 which the better classes possessed; they would have had little sympathy 
 for the Christian chivalry which adorns the enthusiastic pages of Mr. 
 Digby (Godefridus, pp. 225 foil.) ; and though Theognis and the ad- 
 vocates of the aristocracy strove to instil into their readers or hearers 
 that goodness was innate in the nobility, their doctrines were rejected 
 by the democratical spirit of the great literary nation of Greece, and 
 overthrown by the philosophy of Socrates ; so that after all they only 
 succeeded in inserting in the philosophical vocabulary one of the old 
 titles of rank, of which the original political meaning was, as we see 
 from Aristotle, soon merged in the new moral use. 
 
 328 Much the same has been the fate of the Latin gentilis. 
 This word originally signified "one who belonged to a patrician gens 
 or clan," in fact, a patrician, and from this it has gone through the 
 Italian gentiluomo, the French gentilhomme, to our "gentleman," a 
 word which combines the old political meaning of rank with an ex- 
 pression of those moral and social qualities, which we consider, though 
 generally foiind along with rank, to be attainable by every one. The 
 adjective has diverged in our language into two, namely, "gentle," 
 expressing the moral meaning; and "genteel," conveying the idea 
 of rank. 
 
 329 The word ηρως, in its old Homeric use, did not imply any 
 deification or super human qualities; it was merely a title expressing 
 military pre-eminence applied to all the heavy-armed fighters men- 
 tioned in the old poems ; it meant, in fact, nothing more than a good 
 
 NN2 
 
548 COMPOUND WOEDS. [bOOK III. 
 
 soldier : it was onginally a title of rank, and had become indiscrimi- 
 nately assigned to all distinguished soldiers, just as the word knight 
 was extended in the middle ages to all fighters, and translated miles 
 by the monks. We shall not quote from Homer to prove this; all 
 the passages have been collected and the general fact established by 
 a writer in a work which we hope is accessible to most of our readers 
 (Philolog. Mus. ii. pp. 72 foil.). That ΎlQωg was originally a title of 
 rank we think appears from the following considerations. The ter- 
 mination points to a derivation from riQa. That the genuine form of 
 the word was 'ήρΡαο'Τ-^, i. e. "the noble warrior," may be inferred 
 from the form ΉρΓαοΐοζ, which is so written in the Olympic tablet : 
 Bockh, C. 1. I. pp. 26 sqq.; and thus "Ήρακλ^?, whose connexion with 
 the goddess ""Ήιοο: does not appear to be a sufficient cause for his name, 
 may have been so called as the representative of the race of Heroes 
 (see Vott, Etymol.Forsch. ii. p.324). Hesychius says thai^HQauog was 
 another name ίοτΉρακλης, αηάηραίΟζ bears the same relation ίοηρως 
 that γηραιόζ does to άγήρωζ. However, it is obvious, as we have just 
 shown, that ^ρω^ and ηρα are themselves connected; and how they 
 are related will appear from an investigation of the latter name. The 
 goddess Hera is always spoken of as presiding over or connected 
 with marriage-rites ; the chief feature in her mythology is her sacred 
 marriage (ίερός γάμος) with Zeus (Diodor. Sic. v. c. 72); her three 
 names, τΐαρ^^νία (Pindar, Olymp. vi. 88), rslua (Nem. x. 18), χήρα 
 (Pausan. viii. 22, 2), show that she represented marriage and its two 
 periods of negation, according to the principle of contrast which we 
 have pointed out on a former occasio^ (Theatre of the Greeks, 6th 
 edition, p. [22]). The name τέλεια, as applied to Juno, refers to the 
 γάμοιο teXog spoken of by Homer (Odyss. xx. 74): 
 
 em ΆφροδίτΎΐ ^^^ προβεΰτίχε μακρόν 'Ό λνμπον 
 κούρας αΐτήβουύα τέλος ^'αλεροΐο γάμουο, 
 which is called γαμήλίον τέλος by ^schylus (Eumenid. 838), and 
 refers to the marriage-rites as an initiation into a new life (Ruhnken, 
 Timceus, p. 224). "Ηβη appears as the wife oi 'Ηρακλής, and the 
 daughter οί"Ηρα (Pind. Nem. x. ubi supra). 
 
 330 The common epithet expressing lawful marriage in Homer 
 is κουρίδίος, as κονρυδίη αλοχος, κονριδίη γυνή, κονρίδιος φίλος, &c. 
 (Buttmann,Le^:i7o^.i. 32), and κύριος was the Athenian name for the 
 husband in reference to the wife, as δεΰπότης was of the master in 
 reference to his female slave. Thus Aristophanes (Equites, 969): 
 
 χρνΰοϋ διώ^ευ Σμυκνϋ'ην καΐ κνριον, 
 alluding to the custom of including the husband in actions against 
 
CHAP. VI.] COMPOUND WOEDS. 549 
 
 the wife, as the Scholiast observes: tov Σμυκν^ην κωμωδεΐ ώς κί- 
 vaidov, κνρυον δε λέγει τον άνδρα, ούτω γαρ εΛεγράφοντο εν 
 τοις δικαΰτηρίοΐξ^ Άΰτίαΰία και κύριος-, τοντέβτιν ο Περικλής. Also 
 κόρος, κούρης, κνρβας, κυρΰάνιον, κνρνος^ were names of children 
 born in lawful marriage, especially those of the upper classes. Thus 
 in Critias (apud Athenceum, p. 432 f): 
 
 oi Λακεδαιμονίων δε κόροι πίνονβι τοιούτον 
 ώβτε φρέν εις ίλαραν άΰπίδα τΐάντ άτίάγειν. 
 and in Plutarch (i^?/cwr^. XIV. 47): ονδεν ήττον εϊ^ιΰε των κόρων τάς 
 κόρας γνμνάς τε Λομπεύειν και προς ίεροις ncitv 6ρχεΐ6&αι και αδειν 
 (Welcker in Theogn. p.xxxiii). But κύριος and the cognate κοίρανος 
 were also used as titles of honour, as signifying a lord and master. 
 There is, therefore, on this side at least, a connexion between the 
 words referring to marriage and those expressing rank and dignity. 
 Such is also the case in the German Ehe, "marriage," and Ehre, 
 "honour." 
 
 Buttmann rightly observes (Lexil. i. p. 35, note), that on the 
 analogy of κοίλος, Germ, hohl, κάλαμος, Germ. Halm, Germ, κνων, 
 Hund, &c., we might fairly place κονρίδιος by the side of the German 
 Heurath, more anciently written Heurde, and compare κύριος with 
 Herr. Now the Latin herus is a perfect synonym to κύριος, and its 
 analogy to Herr cannot be denied. Moreover ''^5^^ρο^ was another 
 name for Ζευς (Hesych. s. v.), and as the old Greek gods went in pairs, 
 and Zeus and Hera were conjointly worshipped as presiding over the 
 marriage-rites (Diodorus, ubi supra), we may well suppose that this 
 is but another way of writing the masculine of "Ηρα. From these 
 analogies alone it is probable that ηρως ΰ^τιάκνριος may have a cognate 
 origin. It is to be observed that even in Homer κούρητες are synony- 
 mous with Ύίρωες', thus J/.xix. 193: κρινάμενος κούρητας άριΰτηας 
 Παναχαιών. Compare the Latin Quirites, Curiatii, &c. ( Varron. p. 24). 
 On the connexion between κύρβας, κορύβας, and the helmets of these 
 personages, see Lobeck ad Soph. Ajac. 817; AglaopJiam. 1144. 
 
 331 There is a word in Aristophanes which it seems impossible 
 not to connect with κύρβας ("crested"), κύρβεις ("pyramids"), κνρψ 
 βάζω ("to butt with the horn"), κνρβαΰία, κνρηβαΰία, &c., and which 
 still remains unexplained. We refer to the genitive plural κνρηβίων^ 
 which we find in the well-known passage about Eucrates, Equites,263 : 
 
 ενλαβον δε μη 'κφύγτ] ΰε' και γαρ οίδε τάς οδονς 
 
 άΰπερ Εύκρατης εφενγεν εν&ν των κνρηβίων. 
 
 And the same allusion is contained in the fragment quoted by the 
 
 Scholiast on V. 254: και ΰν κ^ί^ρηβιοτίώλα Εύκρατες ΰτύτ^τιαξ. AH 
 
550 COMPOUND WOEDS. [bOOK III. 
 
 the grammarians explain κνρηβιων with reference to a neuter plural 
 κνρήβια^ by which they understand "bran," or "the husks of barley." 
 ThustheScholiast,onv. 253, says: κνρήβια da εΐΰι ταττίτνρακαϊ αχνρα 
 τώνΐΐνρών η κριθών. Similarly, Bekkeri Anecd. p. 272,1. 24. Hesych. : 
 κνρήβία' των κριθών τα ατίοβρίγματα καΐ κνάμων λετΐνρα και τα 
 των κεγχρων αλφιτα. The Scholiast, on v. 2 5 3, understands by κνρήβία, 
 "the kiln," εν&α αίκάγχρνξ φρνγονται, and this is so far right; since 
 iv^v, with the genitive, always expresses motion to or towards a place. 
 We do not, however, see very well by what etymology κνρηβυα can 
 lead to this result; for if we suppose that it refers to the beards, as 
 crests of the barley, we do not obtain a very perspicuous definition of 
 the place of refuge which Eucrates found so convenient. An entirely 
 different interpretation has suggested itself to us. We infer from 
 Xenophon {Memorab. ii. 7, § 6) that Cyrebus was the chief baker at 
 Athens ; in an enumeration of the most successful tradesmen in the 
 city, Socrates is made to say: ατώ άέ άρτοτίοιΧ'αζ Κύρηβος την τε 
 οΐκίαν τίάβαν διατρέφει καΐ %y δα'φιλώξ. Such being the case, nothing 
 is more natural than that the favourite loaves at Athens should be called 
 κνρήβιοί αρτοι^ and the place where they were made κνρτίβια. Nay, 
 even the waste of the flour may have got its name from the chief 
 dealer. Now Eucrates was properly a hemp-seller {Equit. 129); and 
 it is to be supposed that he is called κνρηβιοΛωλης^ because he bought 
 up a quantity of bread and distributed it to the poor, according to 
 the method adopted by the άλλαντοτίώλης in regard to the κορίαννα 
 (Equit, 676). And if this was his usual way of getting out of a diffi- 
 culty, it might be said of him that he used to fly straight to the 
 Cyrehia, i. e. to the establishment of the great bakers. As the name 
 of a man, Κνρηβος is a complimentary title; and Epicrates was known 
 by the patronymic Κνρηβίων (Demosth. Fals. Leg. p. 434, 21). 
 
 332 The Sanscrit vivas (Latin vir) signifies "a warrior, a cham- 
 pion, a hero;" as a feminine, vird^ it denotes "a matron, a wife, and 
 mother;" and as an adjective it expresses the qualities "excellent, 
 eminent, heroic, powerful," &c. From this is formed the abstract 
 vaira^ "heroism." In the same language we have, as a synonym to 
 viras, the word guras, of which the first letter, as we have often seen, 
 corresponds to a Greek κ ; so that this word may fairly be compared 
 with κύρνοξ' Pott thinks (Etymol. Forsck. i. p. 131) that it is iden- 
 tical also with ^ρω^, on the analogy of εκνρος, Sanscrit gvaguras, Latin 
 socerus. In another place (i. p. 221) he makes guras a compound 
 βη=εν and vtras, which, however, we think quite unnecessary: we 
 believe that viras, guras are but modifications of the same word, and 
 both connected with the Sanscrit root vri or var^ "to protect," 
 
CHAP. VI.] COMPOUND WORDS. 551 
 
 modern German Wehr, wehren (above, pp. 301, 493). Further modi- 
 fications iivevaras^ subst. a "husband or bridegroom," adj. "excellent;" 
 compar. vartydn, superl. varisht'lias, and urus, " great ; " with the latter 
 of which we may compare svQvg, ovga-vog (§ 259), with the former 
 "Αρης (=^άρης\ βαρειών, ^άρίβτος, ^^αρετη^ Γάρ67}ν = Γάρ-ρ7]ν, 
 Latin Md-vor-t-s, "man-protector," like Αα-^^έρ-της', conversely, the 
 Oscan Md-mers, "man-killer." With these may also be compared the 
 Erse, /ear, "aman," Welsh ^ii;r, i^r, "a man," gwraig,wraig, "a woman," 
 Latin virgo, virago, Erse/ra^, German Frau: and the first of these 
 words, fear, comes very near ^ηρωξ (Prichard, Eastern Origin of the 
 Celt. Nations, p. QQ). 
 
 333 A similar connexion of the ideas "warrior," "husband," 
 "man of rank," we find in the Sanscrit naras, nris, "a man," "a lord," 
 "a husband;" Celtic iier, "a lord," Greek ά-νήρ (=ά-νερ-£). This 
 word finds its fullest development in the old Latin or Sabine language. 
 Nero, as a cognomen of the Sabine family of the Claudii, is familiar 
 i,o every one; it signifies "a brave man," and its derivatives neria, 
 nerio, neriene, nerienes, signify "valour," also the wife of Mars, as in 
 the following passages quoted by Aulus Gellius (xiii. 22). Plautus 
 {Truculent, ii. 6, 34): 
 
 Mars peregre adveniens salutat Nerienem uxorem suam. 
 
 Cn. Gellius {ΑηηαΙ.ιΐί.): Neria Martis te ohsecro pacem dare, uti liceat 
 
 nuptiis propriis et prosperis uti, quod de tui conjugis consilio contigit. 
 
 Licinius Imbrex (in Nemra Com) : 
 
 Nolo ego Neceram te vocent, sed Nerienem: 
 Quum qiddem Marti es in connubium data. 
 
 Ennius (Annal. i.): 
 
 Nerienem Mavortis et Herclem. 
 
 It is well known that α-νί^ρ, as distinguished irom αν^ρωτίος-, invariably 
 means "a brave man" or "a husband;" and to the latest period ανδρεζ 
 was a complimentary address (see Yalckenaer ad Herod, vii. 210. 
 
 Again we find the same combination in TtoOLg^ πότνια, τίότνα, δεΰ- 
 Λ6rfjg,^ε6-7tOLva,Lsiimpotes (sis Dii potes),potenteSjpot-sum (possum), 
 &c. Sanscrit patis, patni, &c. (above, § 228). 
 
 334 The title αναξ does not seem to imply any thing beyond 
 mere siiperiority, though it has a domestic as well as a political appli- 
 cation: thus, we have in Homer οϊκοιο aval•, of the pater familias 
 (Odyss. I. 397). It is probably, like av-^og, connected with the pre- 
 position ανά. The simpler form is ava-Kog or ανα-κ -g, preserved in 
 Hesychius as an epithet of the Dioscuri — avazag, Tovg zίLogxόρovg. — 
 άνάκείον. το ^ωςκονρίον — άνάκουν. τοίν ^ιο^κόροιν — and per- 
 
552 COMPOUND WORDS. [bOOK III. 
 
 haps ^ενας. rovg άνάκονς (instead of ά%ά%ονξ) %Βθνς. Μάγοι. 
 The common form ανακτ-ς, gen. avanrog, has the double pronominal 
 ending : άνάκτο7ρ is a still longer form. 
 
 335 If αναξ is simply a formation from the preposition ανά, it 
 may be compared with νβρίς (νττερ), "uppishness," ντΐέρφεν, ν%ερ- 
 φνώς, ντΐερφίαλοζ (=^ν7ίερφναλος, Buttmann, Lexil. π. ρ. 213), 
 νπεροτίλοζ, &c. Comp. ντίεραντλοξ νβρίζ (Hemsterhuis ad Lucian. ι. 
 p. 341). In immediate connexion with νβρίζ, we constantly find %όρog. 
 Sometimes κόρog is the parent oi vβρLg, as in Theognis, v. 153 (p. 7 
 Welcker) : 
 
 τίκτει, τον κόρog νβρυν όταν κακω 6λβog ετνηταί 
 άνΟ'ρώτίω και οτω μη voog aρτωg fj. 
 
 V. 751 (p. 12 Welcker): νβρίξ^ τίλούτω κεκoρημεvog. Diog. Laert. 
 (ι. 59), quoting Solon, says: τον μεν κόρον τον πλούτου γεννα6%αι, 
 την d' νβρυν ντίο κόρον. Proclus (^7^0αί2/^•P•59Boissonade): νβριν 
 γάρ φαΰον (οί Λοιηταϊ) τίκτει κόρog (quoted by Welcker, p. 93). Py- 
 thagoras (apud Stobceum Serm. xli. p. 247): πρώτον τρνφήν, έπειτα 
 κόρον•, είτα νβριν, μετά δε ταντα ολε%'ρον. Conversely, vβρLg is the 
 mother οι κόρog. Pindar {Olymp. xm. 10): 
 
 ε^ελοντι δ' άλεξειν 
 "Τβριν, Κόρον ματερα ^'ραΰνμυΟ'ον. 
 (with which may be compared Solon, p. 88 Bach: 
 
 δήμον θ' ηγεμόνων aδικog voog, οΐβιν ετοϊμον 
 
 vβρLog εκ μεγάλης αλγεα πολλά πα^'εΐν^ 
 ον γαρ επίΰτανται κατέχειν κόρον). 
 
 Bacis (apud Herodot. νιιι. 77) ; 
 
 δια ^ίκη ΰβεΰΰει κρατερον Κόρον /'Τβριος νϊον, 
 
 δεινον μαιμώοντα, δοκενντ άνα πάντα τί&εϋΟ'αί. 
 
 And in accordance with this genealogy we would emend a passage of 
 
 JEschylus {Agamemn. 741 foil.), which the editors have regarded as 
 
 hopelessly corrupt. 
 
 We read and arrange the strophe and the antistrophe as follows: 
 
 ύτρ. δί. 1 φιλεΐ δε τίκτειν^Τβρις μεν πάλαια νεά- 
 
 2 ζονΰαν εν κακοΐς βροτών 
 
 3 '''Τβριν, τότ η τοΟ"', οτε το κνριον μόλ^' 
 
 4 νέα δε φύει Κόρον, 
 
 5 δαίμονα τε ταν αμαχον, άπόλεμον, 
 
 6 άνίερον Θράΰος, μελαί- 
 
 7 να μελά%ροι6ιν"Ατα, 
 
 8 είδομένα τοκενΰιν. 
 
CHAP. VI.] COMPOUND WOBDS. 553 
 
 avt. δ\ 1 δί%α δ β λάμτίει, μίν εν δυΰκάτίνοίς δώμαΰον, 
 
 2 το δ' εναίΰίμον τίει. 
 
 3 τα χρνΰότταΰτα δ' εΰΟ^λά ΰνν τΐίνω χερών 
 
 4 τίαλιντρότΐοίς ομμαόυ 
 
 5 [δώματα] λιπον6\ οόια ττροΰεμολε, 
 
 6 δνναμιν ον ύεβονΰα πλού- 
 
 7 τον παράύημον αϊνω' 
 
 8 %αν δ' επί τέρμα νωμα. 
 
 The meaning of the first lines is: ^^0\ά"Τβρίς is wont to bring forth 
 new'''ϊJ3ρ^g, this new^'Tj(3ρ^g brings forth Κόροζ ο,ηάΘράΰος-, two black 
 fates to houses, like their parents." Νεαρά φνεί κόρον was first 
 adopted by Butler, who was guided by Wakefield's νεαρον φνονΰα 
 κόρον; οτε for όταν was introduced by Klausen. The substitution of 
 ταν for τον in the fifth line is due to Hermann. It appears to us that 
 ανίερον Θράύοζ is in apposition with ταν άτίόλεμον, αμαχον δαίμονα. 
 For the duals at the end of the strophe compare Soph. Antig. 529 : 
 
 τρέφων δν "Ατα κάπαναΰτάβεις χ^ρόνων. 
 
 In the antistrophe we have inserted δώματα, which was perhaps 
 omitted in consequence of the similarity of the preceding ομμαΰϋ, and 
 this would be a still easier corruption, if the last syllable of προΰε- 
 μολε, which is due to Hermann, found its way into the space above, 
 so that ομμαύί λε and δώματα λι got confused. 
 
 Now this κόροξ, which we see in such close connexion with νβρις, 
 is in our opinion a word of cognate meaning. It is, we conceive, 
 connected with κόρ-ν(%)ς,κορ-νφή, &c. "the head or top of any thing." 
 The idea of "satiety," which κορο^ often conveys, is subordinate to 
 that of "fulness," "up to the top," as appears from the well-known 
 passage of Sophocles ((Ed. Tyr. 874): 
 
 νβρις φυτεύει τνραννον. 
 
 νβρις, εΐ τίολλών ντίερτΐλιρϋ'γι μάταν 
 
 α μη'τίίκαι,ρα, μηδέ ΰνμφέροντα, 
 
 άκροτάταν εΐΰαναβαβ' άηότομον 
 
 ώρονΰεν εΐξ άνάγκαν 
 
 IW oi) ΛοδΙ χρηΰίμω 
 
 χρηταυ. 
 
 The first line reminds us of Shakspere, Macbeth, iv. sc. 3 : " Bound- 
 less intemperance in nature is a tyranny." 
 
 We observe that χλιδή in v. 888 is a synonym of κόρος. Thus 
 κόρος comes to have the sense of having got as far as one can go, con- 
 sequently it suggests that fixedness, to which νβρις never attains 
 till it has produced κόρος. This view of the case is confirmed by the 
 
554 COMPOUND WORDS. [bOOK III. 
 
 following passage of Plato (Philebus, p. 26): νβρίν γάρ που καΙ ξνμ- 
 Ttauav πάντων πονηρίαν αντη κατυδονΰα η ^εόζ-, πέρας ουδέν 
 οϋΟ^' ηδονών οϋτε πληΰμονών ενόν εν αύτοΐς, νόμον καΐ τάξίν, πέ- 
 ρας εχοντ , ε^ετο. It is from the sense of fixedness and limitation 
 implied in κορο^ that we derive the meaning "uncertain," "unsteady," 
 "wayward," "always changing its place," with wliich ακόρεβτος is 
 found in the dramatists. Thus ^schylus {Agamemn. 1304): 
 
 το μεν ευ πράόΰειν άκόρεΰτον εφυ 
 
 πάοί βροτοΐΰυ. 
 1461 : φευ φευ κακόν αΐνον ατψ 
 
 ρας τυχας άκορέότου. 
 972: μάλα γάρ του τας πολλάς ϋγιείας άκόρεότον τέρμα, 
 in imitation of Solon (p. 80 Bach): 
 
 πλούτου δ' ούδεν τέρμα πεφαβμένον άνδράβι κείται, 
 οϊ γαρ νυν ημών πλεΐΰτον εχουύι βίον 
 
 δίπλάΰίον ΰπεύδουβί' τις αν κορέΰειεν απαντάς] 
 (where we may remark by the way that in Solon's laws πεφαύμένως 
 stands for φανερώς\ see Lysias in Theomnest. p. 117 [363]), Sophocles, 
 (Ed. Col. 120: 
 
 που κυρεΐ εκτόπιος όυ^εις 6 πάντων 
 6 πάντων άκορεΰτότατος-, 
 In this last passage this epithet is applied to (Edipus whom the chorus 
 cannot find ; and they charge him with leaving the place where he was 
 before from mere caprice — "Where," they ask, "is the most unsatisfied, 
 most place- changing of men?" This interpretation is strikingly con- 
 firmed by a very similar passage of the same poet, namely, that in the 
 Ajax (871 sqq.), where the Salaminian sailors complain that they can- 
 not find the hero: "It is hard," they say somewhat peevishly, "that 
 we with so much wandering about should not succeed in approaching 
 him, but should fail to see where the fleeting man is" (άλ^! άμενηνόν 
 άνδρα μη λεύΰΰειν οπού). If ά-μεν-ηνός is derived from μένω, as is 
 most probable, its application to the flitting shades of the dead would 
 be doubly appropriate, and we have elsewhere {Journal of Philology , 
 m. p. 209) quoted the words of our funeral service: "Man fleeth as a 
 shadow, and never continueth in one stay," which exactly gives the 
 meaning of άμενηνός^ required by the passage in the Ajax. It is to 
 be remarked that Pindar uses ακορος to signify "restless," "unceas- 
 ing," "never stopping;" Pyth.Vf.^O'^'. είρεβία δ' υπεχώρηύεν ταχευάν 
 εκ παλαμών ακορος. 
 
 336 The outward resemblance of this κόρος to the similarly writ- 
 ten members of the family, which we have discussed above (κύρ-ίος, 
 
CHAP. VI.] COMPOUND WORDS. 555 
 
 zoLQ-avog^ &c.), is obvious. Let us examine if there be not also a con- 
 nexion in meaning. That the idea of "a head" or "completion" is 
 nearly connected with that of "king," "lord," or "master," cannot 
 be denied; therefore in this sense of κόρ-ος it may fairly be assigned 
 to the same family. But what is the connexion between this word 
 and κόρος, "a young man"? This again can be easily shown. The 
 idea of "fulness," "growing up," enters into our conception of a tall 
 youth, and this the Greek expresses very strongly. Thus άδ-ρός (con- 
 nected with adivos, Buttmann, Lexil. i. p. 206) is used as an epithet 
 alike of a young man, of a tree, of fire, of snow (Herod, iv. 31), of 
 any thing in fact in which the idea of fulness, growth, or strength, is 
 implied : αδροτής is constantly found in connexion with ηβη or μένος 
 in Homer, and αδροΰύνη is used of the top-heaviness of ripe corn in 
 Hesiod (ίίργα καϊ ημέραϋ, v. 471): 
 
 ώδε κεν άδροΰνντ;} ΰτάχνες νενοιεν εραζε. 
 Again, αν%ος — connected with ανά,ανω,άνήνοχ^ε {VoiiEtymol.Forsch. 
 I. p. 211), and, if the derivation proposed above be the true one, with 
 «VaJ— is also used in connexion with ηβη. Thus Pindar (Fyth. iv. 
 158, 281): ΰον δ' αν%ος ηβαξ άρτι κυμαίνει. The last word carries 
 the metaphor only a little farther : a swelling like a wave, an excessive 
 fulness even to overflowing, being also attributed to the lusty vigour 
 of youth. Accordingly we have in the same ode of Pindar (170= 
 318): κεχλάδοντας ηβα. Now κεχλάδοντας is an irregular form of the 
 perfect participle of χλάξω, or rather a new present formed from the 
 perfect, according to a custom not very uncommon in Greek: in fact 
 we have a reduplicated present καχλάξω as a synonym οΐχλάξω, which 
 means "to swell, to be exuberant or full;" hence the rushing, loud- 
 sounding noise of overflowing water is sometimes included in the notion 
 of the word. In a dithyramb of Pindar {Fragm. 48) we have : 
 
 (?ol μεν κατάρχειν, 
 μάτερ μεγάλα, τίάρα ρόμβοι κνμβάλων, 
 εν δε κεχλάδειν κρόταλα, 
 
 where κεχλάδειν follows the analogy of κεχλάδοντας. Similarly in 
 0/2/wj?.ix.init. we have: καλλίνικος 6 τριτίλόος κεχλαδώς, "the triple 
 song of victory when it pours forth its loud full tones," and in the 
 passage under consideration κεχλάδοντας ηβα, "in all the exuberance 
 of youth," " swelling with youthful strength." We do not agree with 
 Buttmann (Ausfilhrl. Sprl. ii. p. 255, note) in supposing that there is 
 no connexion between κεχλαδώς and καχλάξω; the useoi καχλάζονύα, 
 of a cup filled with mousseux wine, sparkling, bubbling, and running 
 over (Pindar, Olymp. \u. init.: φιάλαν ένδον αμπέλου καχλάξοιόαν 
 δρόΰφ) shows that the words have precisely the same force, for κεχλά- 
 
556 COMPOUND WOEDS. [bOOK III. 
 
 δονύαν or κεχλαδυΐαν might have been used here. There is no doubt, 
 however, that it is also connected with χληδος, χλιδή (a perfect 
 synonym οι κόρος, (Ed. Tyr. SSS, and above, § 335), and χλίδάω, and 
 thus χλίδώβα μολπά is used, like καλλίνικος κεχλαδώς, in Pindar 
 (Olymp. XI. 84=100). In precisely the same manner ΰφριγάν and 
 όργάν are used as synonyms of άκμάξειν (Ruhnken, ad Timwum, p. 
 244). It is remarkable that όργάν, which thus conveys the sense of 
 fulness implied in κόρος, also, under the forms regere {ο-ρεγειν), rex, 
 Sanscrit rajas^ corresponds to the political meaning of κύριος, κοίρα- 
 νος, &c. The same metaphor, with respect to the exuberance of youth, 
 is found in the Latin language : thus Quintilian (Instit. Orator, xi. 3, 
 § 28): Illud non sine causa est ah omnibus 'prceceptum ut parcatur 
 maxime I'Oci in illo a pueritia in adolescentiam transitu, quia natura- 
 liter impeditur, non, ut arhitror, propter calorem, quod quidam puta- 
 verunt; nam est major alias; sed propter humorem potius ; nam hoc 
 cetas ilia turgescit. 
 
 337 We see thus how the ideas of fulness, height, and com- 
 pleteness, are connected in the Greek language with that of political 
 superiority. We should be perfectly authorized, then, in connecting 
 aval•, with αν^ος and ανά, as far as the meaning is concerned, even 
 though the words were not etymologically related to one another. 
 We believe, however, as we stated above, that they spring from a 
 common origin. It is clear, indeed, that aval•, was a digammated 
 word (Dawes, Miscell. Grit. pp. 144 sqq.), whereas it does not appear 
 that ανά was ever digammated. But this does not create any real 
 difficulty, for the element a, which forms the first part of ά-νά, is only 
 the second pronominal stem under the form Fa (§ 183); so that 
 JFaval•, is connected with JFava, an older and more genuine form of the 
 preposition. There are traces of the digamma in the form ϊον^ος for 
 αν^ος. At any rate, we cannot agree with those who connect ^άνακτς 
 with Konig, a word with regard to which we rather adopt the 
 opinion of Thierry. It appears that Konig, more anciently Koning^ 
 was the name of any person under authority. Thus the converted 
 centurion bears this name, — ein Koning gieiscot iz in war (Otfrid, 
 Lib. ii). Alfred applies the term Cyning to Csesar as general, to 
 Brutus as the head of a party, and to Antony as consul ; sometimes 
 he designates the particular office of consul by the compound Gear- 
 Cyning, "King for a year." In the Danish language a chief of 
 pirates was called Sie-Konong, the leader of an army Her-Konong, 
 and so forth. In the Saxon language we also find Ober-Cyning, 
 Under-Cyning^ Half- Cyning, In fact the word is merely the par- 
 ticiple of the verb Konnen or Kennen, for they were originally iden- 
 
CHAP. VI.] COMPOUND WORDS. 557 
 
 tical (GraiF, Sprsch. iv. 408), and denotes a person who kens, or can, 
 who has superior knowledge or superior power*. 
 
 338 The original meaning of our English word "Lord" is pre- 
 cisely similar to that which we have endeavoured to point out in 
 several of those Greek names significant of rank. Home Tooke says 
 {Diversions of Purley, n. pp. 155 foil.) that it was originally written 
 Loverd or Hlaford^ that it is compounded of the participle hlaf, from 
 hlifian, "to lift," and of the word Ord, ortuSy "source," "origin," 
 " birth," and that it consequently signifies High-born, or of an exalted 
 origin. That he is right in connecting this word with a verb sig- 
 nifying "to lift up," is sufficiently clear. Similarly, "Lady" is de- 
 rived from Hlafdig, which signifies "lofty," i. e. "raised" or "ex- 
 alted ; " it is written levedi in an old EngUsh MS. in the Cambridge 
 Library (see Gentleman's Mag. Dec. 1838, pp. 619 sqq.). We enter- 
 tain some doubt, however, as to the supposition, that the termination 
 of "Lord" refers to "birth." We would rather connect it with 
 A. S. Ord, 0. H. G. Ort, "a place," and thus Hlaford will mean "a 
 person in high places," perhaps in reference to the heo6e or "dais" 
 in the dining-hall, where the nobles sat (see Beowulf, v. 804 Kemble). 
 This last name is also a general designation of height ; it means any 
 thing elevated, e. g. a shore, as in our names Clay-hithe, Queen-hit he^ 
 applied to the banks of a river where they rise a little. 
 
 The German words Tugend, taugen, signifying "virtue," or 
 "goodness" in general, seem to have had originally a^ political mean- 
 ing, like the Greek καλοκάγαΟ'ός. Thus in Beowulf (v. 716 Kemble) 
 we have cu^e he duguSe peaw, "he knew the manners of the court;" 
 where dugude, obviously connected with Tugend, means "the better 
 part of his followers," "the elders," the γερονΰία. 
 
 339 (3) 'Εντελέχεια. All the questions which have been 
 raised with regard to the celebrated Aristotelian word εντελέχεια, 
 and all the difficulty which it has caused to philosophers (see Tren- 
 delenburg, ad Aristot. de Animd, pp. 319 sqq.), have been occasioned 
 by an inability to discriminate between this and the compound ένδε- 
 
 * "Cette variete d'applications du meme mot n'etonnera point quand on 
 saura que ce titre Koning, n'est probablement que leparticipe actif d'un verbe qui 
 signifie savoir oupouvoir, que par consequent il ne signifie, lui-meme,rien autre 
 chose qu'un homme habile, ou capable, a qui les autres obeissent par la convic- 
 tion de son habilete reconnue " (Thierry. Dix ans d'Etudes Historiques, p. 248. 
 Comp. Lettres sur rHistoire de France, pp. 62—73). "In such Acknowledged 
 Strongest (well-named King, Kon-ning, Can-ning, or Man that was Able) what 
 a symbol shone now for them — significant with the destinies of the world " 
 (Carlyle, French Revolution, i. p. 14). 
 
558 COMPOUND WORDS. [bOOK III. 
 
 λέχΒία, which so nearly resembles it in sound. It will be worth while, 
 then, to explain these words once for all. 
 
 340 The meaning of εντελέχεια may be derived without much 
 difficulty from Aristotle himself. The philosopher is in the constant 
 habit of using a double antithesis to the word δύναμυζ, which he 
 opposes both to εντελέχεια and to ενέργεια. The δνναμίς in this oppo- 
 sition corresponds to νλη, the material out of which any thing is 
 immediately made : the εντελέχεια to the είδος or form which con- 
 stitutes the definition of the thing. "Substance itself," says Ari- 
 stotle, "is reckoned a sort of entity, and in this we discriminate, 
 1st, the material, which by itself does not constitute an individual; 
 2nd, the shape and form, by which the individuality is determined ; 
 3rd, a compound of the two. Now the material is a δνναμις, but the 
 form an εντελέχεια-, and that in two ways, either as science (εΛίότημη) 
 or as contemplation (το ^εωρεΓν)" (de Animd^ ii. 1, § 2): and shortly 
 afterwards (§ 4), he says "the soul is a substance, as the form of a 
 natural body alive •^ννάμει\ but the substance is an εντελέχεια; 
 therefore the soul is the εντελέχεια of such a body. But εντελέχεια is 
 predicated in two ways, as science and as contemplation; accordingly, 
 it is clear that the soul is an εντελέχεια in the same way as science, 
 for sleeping and waking presuppose the soul, and waking is analogous 
 to contemplation, but sleep to the having and not exerting" (i. e. to 
 science, το γαρ ηρεμ'τ^ΰαί και ΰτηναιτην διάνοιαν ετίίΰταύ^αι καΐ 
 φρονεΐν λεγόμε^'α, Phys. νπ. 3). " Wherefore the soul is the primary 
 εντελέχεια of a natural body δυνάμει alive.^^ The body is a δύναμίξ; 
 it contains the outward conditions necessary to the life of the soul : 
 the soul is the εντελέχεια of the body; it is that which makes the 
 body perform its functions ; so sight is the εντελέχεια of the eye, be- 
 cause it is its perfection and consummation. He calls the soul a 
 pritnary εντελέχεια., because it corresponds to science, the first of the 
 two kinds of εντελέχεια. We must not consider the opposition of 
 δνναμίζ and εντελέχεια as equivalent to that of matter and form ; it is 
 merely analogous to it; the εντελέχεια is not a form as something 
 distinct from matter and adscititious ; it is the acting and efficient 
 principle which makes the thing what it is, which individualizes it — 
 το γαρ δυνάμει ov και μη έντελεχεία^άόριΰτόν έΰτιν (Metaphys.ui. 4). 
 
 341 We adopt in our every-day conversation the Aristotelian 
 distinction between δύναμίζ and ενέργεια. The schoolmen, from whose 
 barbarous Latin we have borrowed many of our common words, used 
 to translate δυνάμει hy potentid, or virtute, or virtuaUter; ενεργεία by 
 actUj or actualiter, which the French have adopted as virtuellement, 
 
CHAP. VI.] COMPOUND WORDS. 559 
 
 actuellement, and we oppose that which exists "virtually" or "poten- 
 tially" to that which "actually" is. Aristotle thus defines ενέργεια 
 {Metaphys. vm. 6, p. 1048 Bekk.): "ενεργεία supposes the existence 
 of a thing, but not in that way in which we talk of its existing 
 δυνάμει. We talk of a thing existing δυνάμει^ when, for instance, we 
 say that the statue of Hermes is in the wood, and the half in the 
 whole, because it can be deducted, and that the person who does not 
 speculate is scientific, provided he has the power of speculation : but 
 as for that which is ενεργεία (now our meaning is clear from an in- 
 duction of particulars, and it is not necessary to seek a definition of 
 everything, but we must also take analogy into the account), it stands 
 related to the other as that which builds to that which is capable of 
 building, as waking to sleeping, as seeing to having the eyes shut, 
 but being able to see, as that which is separated from matter to the 
 material, as that which is worked out to that which is not. Of 
 these contrasts let us call the former ενέργεια, and the latter το δννα- 
 τόνΓ We learn from this that the opposition of δνναμίξ to ενέργεια 
 stands much on the same footing with that of the same word to 
 εντελέχεια. 
 
 342 We are not, however, to suppose that εντελέχεια and ενέρ- 
 γεια are synonyms, though it must be confessed that the distinction 
 between them is not always strongly marked. The word ενέργεια 
 signifies an acting, exertion, or operation. Thus, in the celebrated 
 definition of the summum honum in the Ethics (i. 7, § 14) it is said 
 to be an operation, exertion, or acting of the soul, according to virtue 
 (το άν^ρώτίΐνον άγαΟ'όν 'φνχης ενέργεια γίνεται κατ άρετηνγ. Now 
 
 * Sir W. Hamilton says (Lectures on Metaphysics^ i. p.l81) : '■'■power Jaculty, 
 capacity•) disposition^ habit, are all different expressions for potential or possible 
 existence; act,operation,energy, for actual or present existence." Sir Alexander 
 Grant has ably discussed the meaning of ενέργεια as applied to the definition of 
 happiness (Aristotle's Ethics,Y o\.i. pp. 181 — 201). After examining the difference 
 between δννκμίς, εξίζ and ενέργεια, and maintaining that the ιρνχής ενέργεια 
 κατ άρετην must mean a deep and vivid consciousness of life according to the 
 law of virtue, he says (p. 195) : "life, according to his (Aristotle's) philosophy, 
 is no means to any thing ulterior; in the words of Goethe, life itself is the end of 
 life. The very use of the term ενέργεια, as part of the definition of happiness, 
 shows, as Aristotle tells us, that he regards the chief good as nothing external 
 to man, but as existing in man and for man, — existing in the evocation, the 
 vividness, the fruition, of man's own power (Eth. Nic. i. 8, § 3). Let that be 
 called out into actuality which is potential or latent in man, and happiness is the 
 result." And he then quotes the following passages, in which ενέργεια may be 
 rendered by consciousness, conscious state or mood of mind: Eth. Nic. i. 10, 
 §2,§9,§15;vii.l4,§8;ix.9,§5. 
 
560 COMPOUND WOEDS. [bOOK III. 
 
 lie says at the very beginning of the same work, that "this summum 
 bonum is an end (τέλος) ', but there is a diiference between ends, for 
 some are operations (ενέργειαί), and others are certain works (εργά) 
 besides these operations; and in cases wiiere there are any ends 
 collateral to the actions (τΐρά^εις), in these cases the works (έργα) are 
 better than the operations {Ινεργειαι) ; still it does not follow that an 
 operation, which terminates in itself and produces no έργον, should be 
 inferior to one that does, in other words, that a τίρακηκη ενέργεια 
 should rank lower than a τϋοίητικη ενέργεια.''^ From this it appears 
 that the ενέργεια is a mere operation or act, whereas the εντελέχεια 
 is never spoken of as an act, but as a state ensuing upon an act : thus, 
 Aristotle says (Phys. m. 3) that motion is in that which is moved, 
 for it is the εντελέχεια of that which is moved, and is produced by 
 the moving force, and the ενέργεια of the moving force is the same. 
 Now the passage from the δνναμις to the εντελέχεια consists in motion, 
 which is a sort of imperfect ενέργεια (η τε κίνηΰΐζ ενέργεια μέν τις 
 είναι δοκεΐ, ατελής όέ. Phys. πι. 2), and again, motion is the εντελέ- 
 χεια of that which virtually exists, so far as such a thing can be called 
 an εντελέχεια (Phys. m. 2), and motion differs from ενέργεια in this, 
 that the former implies change, the latter continuance (Metaphys. viii. 
 6): therefore, ενέργεια is not εντελέχεια, but only tends to it, as 
 Aristotle distinctly explains it from the primary meaning of the two 
 words : το γαρ έργον τέλος, η δε ενέργεια το έργον, διο και τοννομα 
 ενέργεια λέγεται κατά το έργον κο:1 Συντείνει προς την έντελέχειαν 
 (Metaphys. νιιι. 8, ρ. 1050 Bekk.), that is, the work (έργον) being 
 the end (τέλος), and being implied in the word έν-έργ-εια, this last may 
 be considered as tending to the έν-τελ-έχ-εια, in which the τέλος is con- 
 tained*. Again, he says (Metaphys.Yiii.S,!^. 1047): έλήλυ^ε δ' η ενέρ- 
 γεια τοννομα, η ττρος την έντελέχειαν (5νντε^ειμένη και έτΐΐ τα άλλα, 
 εκ των κινήΰεων μάλιΰτα' δοκεΐ γαρ η ενέργεια μάλιΰτα η κίνηβις 
 είναι, dtb καΐ rotg μη ονβιν ουκ άποδιδόαΰι το κινεϊύ^αι, αλλάς δε 
 
 * On the term εντελέχεια Sir Α. Grant says (Arist. Eth. i. p. 184) : "since 
 δνναμις has the double meaning οι possibility of existence as well as capacity of 
 action, there arise the double contrast of action opposed to the capacity for ac- 
 tion; actual existence opposed to possible existence or potentiality. To express 
 accurately this latter opposition Aristotle seems to have introduced the term 
 εντελέχεια, of which the most natural account is, that it is a compound of fv 
 τέλει εχειν, being the state of perfection, Ιντελεχής being constructed on the ana- 
 logy of νουνεχής. But in fact this distinction between εντελέχεια and ενέργεια 
 (cf. Metaph. viii. i, § 2) is not maintained. The former word is of comparatively 
 rare occurrence, while we find everywhere throughout Aristotle ενέργεια, as he 
 says, προς έντελέχειαν ονντιΟ•εμένη, mixed up with the idea of complete ex- 
 istence. It is contrasted with δνναμις, sometimes as implying motion, some- 
 times as form opposed to matter.'' 
 
CHAP. VI.] COMPOUND WORDS. 561 
 
 tLvag κατηγορίας, οϊον δυανοψα καΐ εΛί%'νμψα είναι τα μη οντά, 
 κινούμενα δ' ου. τοντο δε οτι ουκ οντά ενεργεία εΰονΐαυ ενεργεία, 
 των γαρ μη όντων ενια δυνάμει εΰτίν ουκ εΰτι άέ, οτι ουκ εντελέ- 
 χεια εΰτίν"^. "The name ενέργεια^ ί. e. operation or action, which is 
 joined to εντελέχεια, and occurs also in other combinations, is derived 
 principally from motions; for motion and action are generally iden- 
 tified. "Wherefore motion is not attributed to nonentities, but some- 
 thing else is predicated of them , for instance , that they are conceiv- 
 able or desirable, but not that they are moved. And the reason is, 
 that if we attributed motion to them , we should attribute action to 
 things which do not actually exist. Some nonentities do indeed exist 
 virtually or potentially, but not actually , for they do not exist εντε- 
 λεχεία^ Taking therefore the definitions and illustrations which Aris- 
 totle has given iff this word of his own coinage, we see that he means 
 by it "an organizing force," and that the only English term which 
 can be accepted as an equivalent for the compound is the word "com- 
 pleteness" {Hist, of Greek Lit. n. p. 281). 
 
 343 From all this it clearly appears that Aristotle derived εντε- 
 λέχεια from εν, τέλος, and εχειν, on the analogy of νουνεχής, &c., and 
 that he meant by it the acting and efficient principle of all those 
 things which exist potentially (δυνάμει) and may be otherwise; that 
 is to say, it is their absolute definition — ετι του δυνάμει οντος λόγος 
 η εντελέχεια (de Anim. π. 4, § 4), and το δε τι ην είναι ουκ έχει 
 ϋλην το τίρώτον εντελέχεια γάρ (Metaphys. χπ. 8)|; whereas 
 ενέργεια is the act of that which cannot be otherwise (de Anim. m. 7); 
 it is a kind of motion tending to εντελέχεια , but not attaining to it, 
 except in those cases in which the τέλος is the ενέργεια itself. We 
 must not overlook the distinction between ενέργεια and εζις, which 
 are also opposed to one another, not, however, as δύναμις is to ενέρ- 
 γεια, for ε^ις is much nearer to ενέργεια than to δύναμις. Actions 
 (ενερ^^εί-αι) proceed from, and return to, the universal action (έξις) 
 which is the origin and end of all action: thus, a brave action pro- 
 
 * On this passage Bonitz writes as follows, in his edition of the Me- 
 taphysica, p. 387: ^^ εντελέχεια , ut descendit alj adjective έντελεχης, i. e. 
 plenus, perfectus^ perfectionem rei significat; ενέργεια vero derivatum a v. 
 ενεργεΐν, earn actionem et mutationem, qua quid ex mera possibiliatate ad 
 plenam perducitur essentiam. Quare, ένεργειαν suum et peculiarem locum 
 habere dicit ubi agitur de mutatione et motu, eandemque dicit pertinere 
 et tendere (ΰυντείνειν, ΰυντε&εΐαϋ-αι) ad έντελέχειαν, perfectum rei statum, 
 qui inde conficiatur. Sed licet alterum proprie viam, alterum finem vice 
 significet, tamen h^c duo ita inter se cohgerent, ut facile appareat cur 
 saepissime nuUo usurpentur discrimine." 
 
 t See Michelet, de la Me'iaphysiqne d'Aristote (Paris, 1836), pp. 165, 294. 
 
 00 
 
562 COMPOUND WOEDS. [bOOK III. 
 
 ceeds from the habit of bravery (εξυς, i. e. ανδρεία) and bravery is the 
 end (rsAog) sought by the brave man (Arist. Eth. vii. 10): thus too, 
 αΧΰΛΪψίς is a ε^ις, but κίνηΰος, as we have before shown, is an ενερ- 
 γεία— η μεν γαρ αϊΰΟ-ηόις έξις, η δε κίνηΰίξ ενέργεια (Topica, ιν, 5, 
 ρ. 1^5 Bekker). Το return, then, to the point from which we started, 
 the soul is the εντελέχεια and not the ενέργεια of the body, because 
 the soul is incapable of motion — εν τι των ανδννάτων το ντίάρχειν 
 φνχ^ κίνηΰιν {de Anima, ι. 3, § 1). 
 
 344 We trust that the meaning of this Aristotelian term is now 
 established, from the writings of the philosopher; so that Hermolaus 
 Barbarus would have done better if, instead of consulting the Evil 
 Being about the meaning of the word , according to the absurd story 
 told by Crinitus {de Honesta Disciplina, Lib. vi. c. 2), he had been 
 content to look through his Aristotle. We now proceed to show, 
 notwithstanding the contrary opinion of some great Greek cholars, 
 including Cicero, that εντελέχεια is, etymologically as well as in sig- 
 nification, totally distinct from the older compound ενδελέχεια. It is 
 remarkable that Cicero, who was aware that the word was coined by 
 Aristotle, should have given a translation of it applying exactly to 
 the older word , which he must have met with in his Plato. He says 
 {Tusculan. Disput. i. 10): Aristoteles, — quum quatuor nota ilia genera 
 principiorum esset complexus^ e quibus omnia orirentur, quintam quan- 
 dam naturam censet esse^ e qua sit mens. Quintum genus adhibet 
 vacans nomine; et sic ipsum animum εντελέχειαν appellat novo no- 
 mine, quasi quandam continuatam motionem et perennem. Now the 
 word ενδελεχής does signify perennis, and ενδελέχεια implies con- 
 tinuance and duration, as appears from the following passages, some 
 of which are adduced by Nake {in Choerilum, p. 177). Ohoerilus 
 (p. 169 Nake): 
 
 πέτρην κοιλαίνει ρανίς νδατος εντελέχεια (read ενδελεχείς). 
 
 Euripides {Perithous apud Valcken. Diatrib, p. 39) : 
 
 άκριτος δ' α<3τρων 
 όχλος ένδελεχώς άμφιχορενει. 
 
 Crobylus {apud Athenceum, p. 429 d, repeated p. 443 f): 
 
 το δ' ένδελεχώς με^ειν τίν ηδονήν έ'χδί-; 
 
 Diodorus {apud Athenceum, p. 431 d): 
 
 εΐ το %αρ εκαύτον ένδελεχώς Λοτήριον 
 %ίνειν το λοιτίον τονς λογισμούς αν^,άνει, 
 
 Plato {Legg. ιν. ρ. 7 17 ε): μνήμην ενδελεχή Λαρεχόμενον. {Tim. ρ. 43 
 c): μετά τον ρέοντος ένδελεχώς οχετον. (ρ. 58 c): την άει κίνηΰιν 
 
CHAP. VI.] COMPOUND WORDS. 563 
 
 τούτων ονόαν εβομενην τε ενδελεχώς τίαρεχετοα, exactly Cicero's 
 Gontinuatam motionem et perennem. {Respubh vii. 539 d): άρκεΐ δη 
 ml λόγων μεταλήτΐ^ει μεϊναι ενδελεχώς καΐ ξυντόνως μηδέν άλλο 
 τΐράττοντί. From which passages of Plato, it is quite certain that we 
 ought to read in Legg. x. p. 905 e: άρχοντας μεν άναγκαΐόν τίον 
 γίγνεΰ%'αι τους γε δίΟίκηόονταζ τον αΛαντα ενδελεχώς ονρανόν. 
 Besides these passages, we have in the lexicographers the following 
 notices; Bekker. Anecdot. p. 251, 24: εν δ ελεχεβτ άτης' ΰννεχεΰτά- 
 της καΙ αδιαλείτίτον ; Hesy ch. -.ενδελεχει' τΐνκνάξεί, Αάκωνες (where 
 the last word, as Ruhnken observes in the auctarium, belongs to the 
 gloss ενδεκαδίκορ) J ενδελεχιΰμός' έτΐίμονή (on which Toup, Vol. iv. 
 p. 260, quotes Josephus, xi. 4, p. 555 : ηγαγον δε και την όκηνοπη- 
 γίαν κατ εκείνον τον καυρόν^ — καΐτουξ καλουμένους ενδελεχιύ- 
 μούς)'^ ενδελεχώς' αδναλείΛτως^ ΰυνεχώς, επιμόνως. And Nake 
 quotes from Basilius : το γαρ ενδελεχές το τΐυκνον καΙ ΰυνεχές. From 
 all this it is clear that ενδελεχής , ενδελέχεια , &c. , were in Cicero's 
 thoughts when he gave the translation of εντελέχεια, which we 
 have quoted, and that he probably thought the two words were iden- 
 tical. In this he is borne out by only two authorities; the first is a 
 joke of Lucian's, who makes the letter δ complain that τ has robbed 
 him of his place in ενδελέχεια: άκούετε φωνήεντα δικαΰταί, του 
 μεν δ λέγοντος^ άφεί-λετό μου την ένδελέχειαν , εντελέχειαν άξιοϋν 
 λέγεΰ^αι τίαρα τίάντας τους νόμους {Judicium Vocalium, p. 95 
 Hemsterh.); the other is a gloss of Gregorius of Corinth , who says 
 (p. 155) that the Attics write εντελέχεια for ενδελέχεια. But these 
 two authorities only prove that the Aristotelian word had become 
 more common than the older compound , not that it was synonymous 
 with it or had superseded it, for we find the older word in the Sep- 
 tuagint and in Josephus. The earlier scholars, however, were quite 
 misled by Cicero: Politian {Miscell. cap. 1) and Andreas Schottus 
 (Tull. Qucest. iv. 12) tried to justify his interpretation; Scaliger 
 thought the only difference between the two words was one of pro- 
 nunciation, probably like that between 'εντός and Ενδον : in fact, it was 
 an open question at the revival of letters how the word ought to be 
 written. Rabelais, with his usual learning and discrimination, saw that 
 in Aristotle, at all events, the proper orthography was εντελέχεια. 
 In the chapter headed Comment nous arrivasmes on royaulme de la 
 Quinte Essence, nommee Entelechie (Lib. v, ch. 19), he says: Aristo- 
 teles, prime homme, et paragon de toute philosophy e, feut parrain de 
 nostre dame royne: il, treshien et proprement ^ la nomma Entelechie. 
 Entelechie est son vrai ^nom — qui aulirement la nomme erre par tout 
 le del. Yet, notwithstanding this denunciation, some, even of modern 
 scholars, have thought that the words εντελέχεια and ενδελέχεια were 
 
 002 
 
564 COMPOUND WOEDS. [bOOK III. 
 
 the same. Such appears to have been the opinion of Hemsterhuis, 
 and a living scholar has endeavoured to establish their equivalence by 
 etymology. Doderlein asserts {Lat. Synonyme undEtymol. i. p. 22) on 
 theanalogy oicreber, celeher; κρντίτω, καλντίτω; φράόΰω, φυλάόΰω; 
 cresco, glisco; &c,, "that εντελέχεια is only another pronunciation of 
 εντρεχεια, the Platonic expression for solertia, and he agrees with 
 Hemsterhuis {ad Lucian. Tom. i. p. 95) in thinking that Greg. Cor. 
 p. 155, has rightly explained ενδελέχεια as only a dialectical variety, 
 of which perhaps there is some vestige in τρεχευν and δραμεΐν" In 
 the first place, we believe there is no such word as εντρεχεία in Plato, 
 and the adjective εντρεχης, which occurs only once (Besp. ύιι. p. 537 a), 
 does not, in the slightest degree, approximate to the meaning either 
 of ενδελέχεια or of εντελέχεια: in the latter, the idea of motion is 
 expressly excluded by Aristotle. And , in the second place , even if 
 the sense did not guide us to a discrimination between these two 
 words, the etymology would. Aristotle himself has hinted the deri- 
 vation of his own coinage εντελέχεια from εντελγίξ and εχειν-, and ενδε- 
 λεχής is, without doubt, a compound of εν with δολιχόξ^ an adjective 
 used, indeed, to signify length in distance, but more frequently appli- 
 cable to express length in duration, as δολιχη vouog, δόλιχος τίλόος, 
 &c., and δολιχόν is used as an adverb to signify simply continuance 
 and length of time, as in Homer, Iliad x. 52: 
 
 'έργα δ' έ'ρε|' οΰα φημί μελήΰεμεν Άργείοιΰι 
 δηΟ'ά τε και δολιχόν. 
 
 So that in meaning it is very near akin to ενδελεχής. With regard to 
 the form of the compound, it may be observed that the termination is 
 regularly altered from -ος to -ης in such words ; thus, from γένος we 
 have εγγενής; φίλος, Λροΰφιλής] καλός, τίετίΐκαλλής ; μακρός, ενμψ 
 κής, &C., not as grammarians say, because it has passed through a 
 substantive in -ος , -εος, but by the common addition of the element 
 -ya. The change from ο to ε, in the first syllable of the adjective, is 
 due to the greater weight of ενδελεχής (above, § 222).. The element 
 of δόλιχος is found in all the languages of the Indo-Germanic family, 
 and may be detected also in the Hebrew hh^ (above, § 209). In Sans- 
 crit, it appears as the root drih, "to grow," adjective dirgha, "long," 
 Zend darega, Behistun daraga (Rawlinson, As. Soc. xi. 1, p. 188), Per- 
 sian dira, Splavonic dolgui , Bohemian dlauTiy , Polish dlugi, Servian 
 dyg, Lithuanian ilga, Lettish ^7^, old Prussian ilga, and, probably, by 
 dissimilation, the Latin longus, Gothic laggs, German lang, English long. 
 Just in the same way the adverb τήλε, which appears in the difierent 
 cases of a last noun, τηλον, τηλνΐ, τηλό&ι, τηλοΐ, τηλό&εν, and indi- 
 cates extent or distance, should be connected with the root οίΟ^άλλω, 
 
CHAP. VI.] COMPOUND W0BD8. 565 
 
 άτάλλω, ατιτάλλω-, τηλεΟ'άω, all denoting growth and cultivation. The 
 ^olic form of the adverb is Ληλνυ (Ahrens, Dial. ^ol. 41), and this, 
 on the analogy of τεΰΟαρες, Λίύνρες^ &c. , points to a combined arti- 
 culation of guttural and labial in the original form (§ 121), so that we 
 may recognise another representative of the same root in the Sanscr. 
 phala, "fruit," in ςρ /Loos, fios, flojan^ φνλλον, folium , "bloom," &c. 
 There can be no doubt, we think, that Doderlein is right in referring 
 the much disputed adjective τηλνγετος rather to the primary meaning 
 of the verb Ο'αλλω , than to the secondary signification of the adverb 
 τηλον {Comment atio devocabulo Ti;Ai;y£TOS,Erlang. 1825; Homerisches 
 Glossarium^ Erlang. 1850, no. 351). All the ancient and original 
 passages , in which this word is found , accord best with the sense 
 "blooming in youth," "delicately nursed or cherished," "childlike." 
 The meaning and etymology together are given in the lines {II. ix. 
 255): τί(5ω δε μιν ίΰον'Ορέΰτγ, og μοι ττιλυγετοξ τρέφεται ^aXiy 
 ενί τίολλη. And that it may signify "childlike," "childish," "babyish," 
 in a reproachful sense, is clear from Jl. xiii. 472: αλλ' ονκ'Ιδομενηα 
 φόβος λάβε, τηλνγετον cos- The termination -γετος, as Doderlein has 
 shown, merely denotes the nature or constitution (Hesych. s. v. βαρν- 
 γετας' βάρος μεν έχοντας, -γετας δε οντάς. Id. s. v. τανγεταις %ν- 
 λαις' ταΐς μεγάλαις), as in ταλ%νβίος. We have the same reference 
 to childhood and childish sports in αταλός and its derivative ατάλλω ; 
 compare Hesiod, Op. et D. 130: τίαΐς Λαρα μψερι κεδν^ ετρεφετ 
 άτάλλων. Theog, 989: τίαΐδ' άταλα φρονεοντα. Hom. 7/. χνπι. 567: 
 παρθενικά! δε καΐ ηΐ^εοι άταλα φρονεοντες. νι. 400: %αΐδ' ετά 
 κόλτίον εχονύ' άταλόφρονα, νήτηον αντως. Pind. Nem. νπ. 90: 
 άμφετΰων άταλόν Ο^νμον τιατρί. Sop. Aj. 556: τέως δε κονφοις 
 Λνεύμαύιν βόύκον, νεαν ιΐ^υχην άτάλλων μητρί τ^δε χαρμονήν. 
 Hesych. : άτάλματα ' τιαίγνια. And ατιτάλλω is a synonym of τρέφω, 
 "I nourish or increase in the way of growth," //. v. 271 : τονς μεν 
 τεββαρας αυτός έχων άτίταλ)^ ε%ί φάτνγ]. χχιν. 60: ην εγώ αντή 
 %'ρέ'φα τ ε και άτίτηλα. Od. χνπι. 323 : παϊδα δε ως άτίταλλε, δίδον 
 ά' αρ* αθύρματα ^νμω. With reference then to the idea of size or 
 extent indicated by δόλιχος and τηλου, we may compare τρόφι and 
 τραφερός.^ derived from τρέφω ; for τραφερη κέλεν^ος, Apoll. Rhod. 
 Π. 545, is exactly equivalent to δολίχη κέλεν%ος, Hom. Od. iv. 493. 
 Pott suggests (Etym, Forsch. i. 87) that in-dulgere belongs to the 
 root of δόλιχος , with the original signification of laxius or longius 
 reddere, remitiere, just as languere seems to be connected with longus, 
 and we admit the derivation as at least probable: the meaning to 
 wliich he refers seems to be found in Yirgil, Georgic. n. 276 : 
 
 Sin tumults acclive solum collesque supinos, 
 Indulge ordinibus; 
 
566 COMPOUND WOEDS. [bOOK III. 
 
 and in the preface of Aulus Gellius: Animus, inter stitione negotiorum 
 aliqua data, laxari indulgerique potuisset. The Sanscrit, Zend, and 
 Persian forms approach nearly to the Greek doQV, δρνζ, in which some 
 meaning of growth and length is obviously contained. The Lettish 
 verb ilgt, "to^delay," also belongs to this class : also, the Anglo-Saxon 
 telg, "a plant," the Gothic tulgjan, "to strengthen," and the Latin 
 largus (Pott, Etym. Forsch. i. p. 251). The idea of prolongation and 
 continuance, which is conveyed by ενδελεχί^ς, and with which this 
 etymology entirely accords , is directly opposed to the notion of ante- 
 cedent completeness suggested by εντελέχεια. As we have seen above 
 (§ 343) , the εντελέχεια belongs to the same class of definitions with 
 the το τι ην εϊναι, which, being a ηρώτη ονΰία or abstraction, pre- 
 cedes the actual or concrete existence of the thing, and carries us back 
 by inference to what it was (§ 192): ώΰτε το τι ην είναι εΰτιν οΰων 
 δ λόγθ£ εΰτίν οριΰμός (Aristot. Metaphys. νι. 4, § 9). Accordingly, 
 the words in question stand opposed in much the same way as if they 
 were different predications of tense; and while ενδελεχής points to 
 the present and the future (Plat. Tim. 58 c, quoted above), which 
 imply continuous motion , εντελέχεια is an expression of the prcete- 
 ritum ac perfectum tempus, and belongs to the category of the im- 
 movable. 
 
BOOK IV. 
 
 THE VERB. 
 
THE 
 
 NEW CRATYLUS. 
 
 BOOK IV. 
 
 THE VERB. 
 
 CHAPTER I. 
 ^ THE PERSON-ENDINGS. 
 
 345 Original comprehensiveness of the verb. 346 Causes which led to the 
 mutilation of the affixes. 347 The person-endings are objective cases of 
 the personal pronouns. 348 Differences of voice are only different cases 
 of the pronominal affix. 349 A. Primary forms. 1st person singular. 
 Reasons for believing that verbs in -μι are older than verbs in -ω. 350 
 Successive mutilations of -μ,ι. Quantity of the original penultima. 351 
 In the active, the person-ending represents the instrumental or ablative 
 case. 352 In the middle, it represents the locative. 353 2nd person sin- 
 gular. Usual form in -a-. The by-form is -&-, not -ΰ-ϋ: 354 Eva- 
 nescence of the characteristic of this person. 355 3rd person singular. 
 Active in -tl, -τω : middle in -ταί, -την. 356 1st person plural. Active 
 -μες for -με-αι. 357 Middle -με^α, -με^Όν, for -μεϋ'αι, -με&ην. 358 
 2nd person plural. Active -τε for -τες: middle -α&ε for -ad-at and -αϋ'ην. 
 
 ' 359 3rd person plural. Difficulties occasioned by the dual. 360 Plural 
 in -VTt and -νται. Explanation of -ciai. 361 Origin of ντ. 362 Β. Se- 
 condary forms. Active suffixes. Explanation of -aav. 363 Middle suf- 
 fixes. They are not reduplications. 364 Table of existing forms. 365 
 Latin person-endings. Difficulties of the passive forms. The second 
 persons. Are participial predications without a copula allowable in 
 Greek? 366 Influence of the weight of the person-endings. 
 
 345 rpHOSE students, whose notions of the nature of a verb 
 are derived from the appearance of that part of 
 speech in our own and indeed in most modern languages, will 
 not be able to understand very well the meaning of the term as 
 applied in the grammars of the ancient languages. A verb in 
 English can only express the copula or the copula and predicate 
 of a sentence: it can never contain the subject; in other words. 
 
570 THE PEESON-ENDINGS. [BOOK IV. 
 
 no English verb can really b« used impersonally , as the gram- 
 marians say, except in such obsolete phrases as me-thinketh^ me- 
 seems^ &c. But in the ancient languages, verbs are often found 
 in the third person without any subject or nominative case ex- 
 pressed: and unless some particular emphasis is required, the 
 nominative of the first and second persons is regularly omitted; 
 so that the whole logical proposition may be included in a 
 single word. 
 
 346 The explanation of this follows from what we have 
 said of the substitution of prepositions for case-endings, &c. ; the 
 original verb contained the pronominal elements or symbols of 
 the relations of place, wliich constituted at once the case-endings 
 of the noun and the persons of the verb. A bare root or stem 
 without a pronominal suffix could no more form a noun, than 
 a modification of it could form a verb without a corresponding 
 person-ending. By the lapse of time, the introduction of writ- 
 ing, and the other causes which are always at work upon a 
 language as long as it retains its vitality, the old forms degen- 
 erated into those naked shapes in which we find but scanty rem- 
 nants of the original clothing. A desire for greater distinctness 
 in the applications of the verb led to the introduction of a system 
 of nominative cases, or the express statement of the subject, and 
 then, as the person-endings became less necessary, they were 
 gradually dropt; just as the same causes produced an analo- 
 gous efi'ect upon the cases of the noun. We have shown that 
 the Sanscrit language, which had no prepositions in the ordinary 
 sense of the word, exhibits a most complete system of case- 
 endings ; the person-endings of the verb are also more strongly 
 marked in Sanscrit than in Greek, because the Indians were 
 less accustomed than the Greek to state the subject of the pro- 
 position, and in general the language had not attained to a 
 full logical development. With the exception of the verbs in 
 -/Ltt, and some of the secondary forms of the common verbs, the 
 person-endings are, as we shall soon show, absorbed in the more 
 modern state of the Greek language. 
 
 347 It has long been perceived that the terminations of 
 the verbs in -μι are personal pronouns; but it has been sup- 
 
CHAP. I.] THE PEBSON-ENDINGS. 571 
 
 posed by modern scholars (for instance, by Thiersch, in his 
 Grammar, and Pott) that these personal pronouns must needs 
 be nominative cases, the root of the verb constituting the pre- 
 dicate, and the connecting syllable the copula. Nothing, in our 
 opinion , can be more unphilosophical than such a supposition. 
 Even if man , in the rudest and earliest times , had excogitated 
 all the rules of logic which were adumbrated by Plato and set 
 forth as a novelty by Aristotle, it appears to us inconceivable 
 that he should have arranged predicate, copula and subject in 
 an order converse to that which every logician knows to be the 
 natural one. Besides , when the system of nominative cases 
 was introduced, the nominative of the first person was in Latin 
 ego, in Greek Ιγών^ in Sanscrit aham; of the second in Latin 
 tu^ in Greek TV, subsequently 6v^ in Sanscrit tvam; and, as 
 we have shown in the preceding book, the nominative sign in 
 nouns expressing the third person was -s. Now the oldest forms 
 of the singular person-endings in these languages, so far as they 
 have come down to us, appear to have been -mi^ -si, ~ti, of 
 which the first and third manifestly refer to the objective cases 
 of the same pronouns : and when the third person appears as 
 -si, this is manifestly only a dialectic softening of the objective f. 
 In Hebrew it is well known that the pronominal affixes both of 
 nouns and verbs perform the functions of objective cases (see 
 Mashil le Sopher, p. 17); and the late Mr. Garnett proved many 
 years ago that the person-endings of all the Indo - Germanic 
 verbs are in statu regiminis (Quarterly Review, LVii. pp. 93 
 sqq.)*. The fact is, that the original verbs stands on precisely 
 
 * This admirable philologer is entitled to the credit of this important 
 discovery. In the first edition of the present work, the author, not hav- 
 ing the Quarterly Review of Prichard before him, had attributed the true 
 theory to that distinguished w^riter, and not to the reviewer, whose name 
 at that time had not been mentioned. This inadvertence was corrected 
 by Mr. Garnett, in a letter dated "British Museum, May 3, 1842," in 
 which he introduced himself to the author of the present book, and 
 claimed as his own the remark referred to in the text. With Mr. Gar- 
 nett's consent we printed in 1844 (Varron. p. 290) an extract from this 
 letter, to which he has referred in one of his papers (Essays, p. 269). 
 It may be interesting to the reader to see in Mr. Garnett's own words 
 the vindication of his right to this linguistic deduction. " If Dr. Prichard," 
 
572 • THE PERSON-ENDINGS. [bOOK IV. 
 
 the same footing as the noun ; it is a word, of which the element 
 or differential part might be found in a noun, the constant part 
 consisting, like that of the noun , in a pronominal element, ex- 
 pressing some relation of place. Thus δίδω-μι would signify 
 "giving here," i. e. where the speaker is; δίδω-ΰί^ "giving 
 where the person addressed is;" δίδω-Τί^ "giving there," i.e. at 
 any other place; and similarly with regard to τί^'ψμί. Now 
 the roots of δίδωμυ and τί^ημι are δο- and %b- respectively*, and 
 both of them represent a Sanscrit a, for they correspond to the 
 verbs daddmi and dadhdmi. But in each case the root seems to 
 be connected with the person-ending by an intervening a, and 
 it is then reduplicated to express more vividly the continuity 
 of the action ; a custom which we find in the unformed dialects 
 of rude tribes even at the present day. The root δο appears 
 with a similar prolongation in δώ-ρο-ν^ "a giving away" (-ρα-), 
 the case-ending of which implies mere location, and does not, 
 like the person-endings of the verb, markka particular relative 
 place. As the verb gradually receives its development, we 
 find that the differences of mood and tense affect the termina- 
 tions as well as the root; but this is sufficiently intelligible, for 
 of course the conjugation would not receive its completion till 
 
 he says, "ever did advance that the personal-endings of the verbs are 
 objective or oblique cases of pronouns, I have done him an injustice, which 
 I ought to repair, for in the Q. R. I expressly affirmed that he had over- 
 looked the fact altogether. Indeed, on looking carefully through his book, 
 I find he speaks of the suffixes in question as abbreviated pronouns, but 
 never, as far as I am aware, as oblique cases. Nor can I discover that 
 they were ever considered in that light by any philologist prior to the 
 publication of the paper in the Quarterly. Gesenius, and other orien- 
 talists, speak of them decidedly as nominatives; and Lee, though he re- 
 gards verbal roots as nouns, speaks indeed of the suffixes as abbreviated 
 pronouns, but never that I know of as being in statu obliquo. The only 
 writer I am acquainted with who has at all adverted to the fact, is the 
 late William Humboldt, and he speaks of it as an occasional pheno- 
 menon, in a few languages, without appearing at all aware of the exten- 
 sive and important application of which it is capable." 
 
 * The roots da, δο-, dha, Οέ-, are confused in Latin, for do means 
 not only "to give," but "to put," as is seen from the compounds abdere, 
 condere, dedere, &c. Pott, Etym. Forsch. ii. 114; Bopp, Vergl. Gr. p. 886; 
 Benary, Rom. Lautl. p. 175. 
 
CHAP. I.] THE PEESON-ENDINGS. 573 
 
 the formation of sentences , when the expression of the nomi- 
 native or subjective case had become necessary and common; 
 and this, by rendering the person-endings less important, vi^ould 
 also make them yield more readily to the laws of euphony, 
 which required modifications of the termination corresponding 
 to those of the root or body of the word. 
 
 348 In treating of the person-endings we must consider 
 them as the oblique cases of personal pronouns. The number 
 designated, whether singular, dual, or plural, is to be explained 
 in the same manner as the numbers of those positional words. 
 The divison of verbs into different voices , one of which ex- 
 presses that the action is considered as affecting the speaker 
 or person spoken of, and the other as affecting some other 
 person or thing, depends upon a change in the case-endings; 
 the former exhibits that relation of case which we call the loca- 
 tive, the latter that which is termed the instrumental. But we 
 must also examine the person-endings, thus affected by number 
 and voice, in their appearances as primary or secondary forms ; 
 that is, as they appear affected or unaffected by the modifi- 
 cations of tense and mood to which the verb in its full de- 
 velopment is subjected. We shall, therefore, first take the 
 primary or simplest forms, explaining them successively as 
 they appear in the different numbers and voices, and then pro- 
 cend to the discussion of the secondary or subordinate forms. 
 Throughout we shall presume a reference to the chapter on the 
 pronouns. 
 
 A. PKIMARY FORMS. 
 
 349 1st Person Singular. 
 
 That the conjugation in -μι is the original one may be proved 
 from the Greek language alone, without the aid of comparative philo- 
 logy. In the first place, those verbs which in the classical ages of 
 Greek literature were still conjugated in -μι, such as είμί^ ^L^ωμLi ΐΰτψ 
 μι, φτ^μί, &c., all convey the most elementary ideas ever expressed by 
 verbs: "being," "giving," "standing," "s^ing," &c. They are words 
 which must have existed in the oldest and rudest state of the lan- 
 guage, and therefore could not have owed their existence to the obser- 
 vation of analogies which had arisen subsequently to that earlier state. 
 
574 THE PERSON-ENDINGS. [bOOK IV. 
 
 Again, the conjugation in -μι is departed from only in a few tenses 
 (principally the present and imperfect active) of the ordinary verbs ; 
 the other tenses all retain traces, more or less distinct, of the original 
 form. Thus, though we have τντΐτω we have τντίτομαί (which is per- 
 fectly analogous to δίδομαή, τντίτοί-μι^ ετνφά-μην, &c. Finally, the 
 change from -μο to -ω is explicable, and may be supported by ortho- 
 graphical analogies ; the converse is not. In all languages, we find a 
 tendency to abridge words as far is consistent with the preservation 
 of their meaning, and in those which exhibit systematic composition 
 we observe a continual conflict for mastery between the body of the 
 word and the suffix. The original verbs were very short and simple, 
 and, even when the person-ending was retained at full length, did not' 
 fatigue the voice of the speaker; there was, therefore, no immediate 
 reason for abolishing the person-endings even after they had forfeited 
 their claim to indispensable utility. In other roots, which the neces- 
 sities of language required, the verbal element would be longer, some- 
 times composed of two distinct stems or a stem and a preposition, 
 sometimes of a heavy, hard-sounding stem, with many consonants, 
 or in general the present tense would be strengthened by insertion, 
 whether of guna or anusvara^ or by the addition of some pronominal 
 element; this of itself, on the principle we have mentioned, would 
 interfere materially with the termination , which when it became less 
 necessary, would be dropt altogether. After this custom of dropping 
 the ending in the present tense had become common, new verbs would 
 be formed on the new, rather than on the old system, and so at length 
 the number of verbs in -μυ would become comparatively inconsider- 
 able, 
 
 350 Supposing -μι to be the original ending of the first person, 
 the most natural method of avoiding an additional syllable, while the 
 meaning of the ending was still retained, would be by keeping only 
 the consonant and omitting the final short vowel; this plan we find 
 adopted in Latin, though su-m and inqua-m are the only verbs which 
 exhibit it in the present indicative ; in the other tenses and moods m 
 is the regular ending, as in amaba-m, ame-m, &c. Its omission at the 
 end of the present indicative is perhaps only another instance of that 
 use of the final anusvara in Latin, which we have pointed out in 
 treating of the accusative case; for it will be recollected, that the 
 final m is liable to ecthlipsis in verbs as well as in nouns. But in 
 Greek -μ cannot, according to the laws of euphony, stand at the end 
 of a word; in shortening the ending, therefore, in the way we have 
 supposed, the m must either have been struck out, or some represen- 
 tative must have been substituted for it. We find both methods 
 
CHAP. I.] THE PEESON-ENDINGS. 575 
 
 adopted. The former is the common one in the present tense, where 
 we have τντίτω for τύτΐτο-μί or τύτίτο-μ \ the latter however appears 
 in the secondary forms, as UrvTiXO-v for Ι-χντίτο-μι. This interchange 
 of m and η we have also observed in the accusative case. Bopp is in- 
 clined to suppose (Vergl. Gramm. p. 626), that τύ%τω-μι^νίθϊτνητο-μί, 
 would be the original form of η;Λτω , in which he is guided by the 
 analogy of the Sanscrit verbs in -dmi^ and the Greek δίδωμί, τίΰ'ημί, 
 &c. It seems better to conclude, that, in cases where the verb-root 
 ends with a consonant, the vowel used to connect the stem with the 
 suffix would be a short ο or ε, especially in forms like τντΐ-τ-ω, τέμ- 
 ν-ω-, where the root is strengthened by a consonantal addition. As we 
 ^*Jiave suggested above (§ 347), there is reason to believe that the roots 
 of δίδωμο, τί^ημί, &c., are strengthened by an inserted a, which is still 
 seen in the third person plural in -άΰυ (below, § 360): and it is a fair 
 inference, that this a is the residuum of some pronominal adjunct 
 analogous to the r or i; of τύτΐ-τ-ω, τέμ-ν-ω, which, therefore, do not 
 need any further corroboration. The long vowel at the end of these 
 barytone verbs is due to the principle of compensation which we so 
 often find in the Greek and other languages. Thus τντίτω would stand 
 for τνΛΤομν, as τντίτείζ foi^ τνπτεΰι,, and τΰπτει for τιντίτετι. The 
 reader will recollect the formation of the comparative - endings -ων 
 from -ovg^ μάλλον from μάλιον^ &c. The cases of δίδωμι &c., are quite 
 different; in these the verb-root itself is lengthened, as in the nouns 
 δώ-ρον, &c., from the same root. 
 
 351 Upon the whole, we may safely conclude, that the first per- 
 son singular in Greek, Latin, and Sanscrit, was always designated by 
 m-, in the present indicative of the old forms of those languages. 
 That this m- was the element of the objective cases of the first per- 
 sonal pronouns is obvious. It is also clear enough, that when the verb 
 in active, the person-ending must needs express an agent; in other 
 words, the action or doing implied by the root must be set forth as 
 preceeding from him; this is effected in the flexion-system of the lan- 
 guages we are considering, by putting the name of the agent in the 
 instrumental, ablative, or, what is equivalent in Greek, the genitive 
 case. The strong resemblance which subsists between the instru- 
 mental and ablative of the third personal pronoun in Greek , and the 
 termination -τω of the third person imperative active, cannot be over- 
 looked: and it has been remarked by Mr. Garnett (Quarterly Eeview, 
 Vol.Lvn.p.99,note),that "the ancient Latin imperatives, estod, vivitod, 
 and the analogous Yeda-imperative, jtvatat=vivito, are unequivocally 
 in the ablative form." See also Curtius, Sprachvergl. Beitrdge, pp.270 
 sqq. In all probability, the ending -mi is merely an abbreviation of 
 
576 THE PERSON-ENDINGS. [bOOK IV. 
 
 the instrumental me=mai, to which it stands related as Λερί does to 
 TCaQai-, whereas the third person of the imperative prefers the stronger 
 inflexion of the ablative in -ω[ς]=^-ωδ=-ο%Εν^ a difference of case which 
 does not produce any real difference in the meaning of the pronoun 
 affixed. Indeed, as the instrumental and locative are often used with 
 the same application (§ 246) , and as the differences of voice are indi- 
 cated by the contrast of their distinctive meanings , it would appear 
 more reasonable that the ablative inflexion should have been used 
 throughout the moods as an indication of the active verb. 
 
 ^352 In the passive or middle, however, we should expect to find 
 indications of a locative-case in the personal suffix : for in this voic^ 
 the action is supposed to end with (i. e. upon) the agent, as indeed 
 is implied in the name dtmane-padam or "self-form," given to it by the 
 Sanscrit grammarians. That the passive verb, in the languages which 
 we are considering, had not in itself any reflexive meaning, is an 
 obvious fact, if the explanation we have given of the personal suffixes 
 be the correct one, and it may be shown that the Greek middle verb 
 in particular is only an idiomatic application of the intransitive pas- 
 sive (Gr. Gr. Art. 432). 
 
 The middle or passive person-endings are distinguished from the 
 active by a greater weight and fulness of form. The first person 
 middle in Greek is -μαι\ in Sanscrit it is wanting, but the other per- 
 sons exhibit a similar alteration by guna of the active person-ending. 
 As the active -mi points to the instrumental we, so -mai must be due 
 to the locative -mayi, the person-ending being in each instance, an 
 abbreviation of the regular case of the pronoun. Thus , if δί-δω-μι 
 means "a giving effected by me," or "I give," δί-δο-μαυ will signify 
 "a giving of which I am the object," the giver being presumed. The 
 analogy of the secondary form -μην shows that the complete locative 
 affix -mai must have been men=mayina; compare εν, ίνα, with εΐ, 
 αϊ, and the common locative in t, with the more original form -l-v. 
 The Sanscrit third person imper. mid. tuda-tdm may induce us to form 
 the same conclusion with regard to the 2nd and 3rd person-endings 
 in -0, while those in -a and -ε are explained by the analogies pointed 
 out above (§ 363). 
 
 353 2nd Person Singular. 
 
 The characteristic of the second person in Greek is -g, and this 
 may be considered as a shortened form of -(5i (which is retained only 
 in the Doric εό-<5ί, though it appears consistently in the Sanscrit, 
 Zend, and Sclavonic), just as the Latin -m and Greek -v at the end of 
 other tenses are of the original -μυ. Besides this, we find a termina- 
 tion -&α, the connexion of which with the second personal pronoun 
 
CHAP. I.] THE PEBSON-ENDINGS. 577 
 
 we have before pointed out (§ 262 — 4). In the imperative it appears 
 as -Οί,. This ending corresponds to the Sansrit -tha, -dhi. Buttmann 
 will not allow (Ausfuhrl. Sprachl. § 78, 7, Anm. 3), that -xtcc can be 
 considered a termination; he joins it with the 6-, which generally 
 precedes it. That, however, -%'a not -ΰΟ'α is the ending, will appear 
 from the following considerations, in addition to the argument deriv- 
 able from the analogy of the Sanscrit. The two words in which this 
 termination most frequently occurs, are οϊΰΟ'α'^, ηΰ^'α, obviously words 
 of great antiquity. The first of these stands for old-^a, for the root is 
 /id-; the second for ϊ-ζϋ-^α, the root being Ιβ-. In these two instances, 
 then, '^a is obviously the termination, as is -^l in the imperatives l'(i- 
 %L•, (pa-^L, Ϊ -^L, xkv-d'i, ΰτψ&ί, &c. Although the 6 cannot be referred 
 to the root in such forms as τί%^η6^α, εφηΰ&α, &c., they admit of an 
 easy explanation; for in the course of time the pronominal nature of 
 the termination would be forgotten or overlooked, and (>, the ordinary 
 mark of the second person, would be inserted on a mistaken analogy, 
 just as we sometimes find οϊβ-Ο'ας for olu-d'cc. This is Bopp's opinion 
 (Vergl. Gramm. p. 655); he formerly thought with Buttmann, that 
 the termination was -6^a' in the Annals of Oriental Literature (i. 
 p. 42), he remarks, " The Greek language is very fond of prefixing <J 
 to 0", as is manifestly evinced by the passive participle formed by the 
 suffix -^ζΐς ; but prefixing 6 to the 0*, there is formed οϊόΟ'είς, μνηύ^Βΐζ^ 
 χρηΰΟΈίς, &c. Therefore, it is no wonder if the Greek has formed 
 6&a out of the Sanscrit termination ίΛα." In the second person of 
 the Latin perfect, we look upon -ti as the person-ending, the preceding 
 * being a representative of ka = sa = ha, the proper characteristic of 
 that tense. 
 
 354 The termination of the second person passive , in the Greek 
 verbs in -^t, is -6aL, as in τί&ε-ΰα^ in the ordinary verbs this is con- 
 tracted to -y, as in tVTtty for τυτΐτεΰαί. It is very remarkable, that, 
 though all Greek verbs keep the full form in -μαι> for the first person, 
 the second person is shortened in this way: and conversely, though 
 in Sanscrit the full form of the second person is invariably preserved, 
 the first person of the passive is abbreviated quite analogously to the 
 abbreviation of the second person in the passive of Greek barytone 
 verbs. Thus we have 
 
 bhar-e φέρ-ο-μαί^ 
 
 but bhar-a-se φ^9-ν* 
 
 * It seems that Sophocles wrote οΐα&ε in the 2nd person plural: οΐα&ε 
 Ιτά δευτέρου προαώτίου γράφει δ Σοφοκλής• nal ηέηοβΟ'ε Λαρά ^Ομηρω 
 άηο του ηεηόνΟ-ατε' ο^τωζ το οΐΰΟ'ε άπο του οι'δατε' κατά αυγκοπήν 
 καΙ τά δύο. Bachmann, Anecd. π. ρ. 358, 1. 20. 
 
 ΡΡ 
 
578 THE PEESON-ENDINGS. [bOOK IV. 
 
 This omision of the 6 in the middle of Greek verbs has been 
 more than once explained (see § 114). We have had a similar evan- 
 escence in the nouns (§ 244). 
 
 355 3rd Person Singular. 
 
 Of the third person singular we have little to say, after the gene- 
 ral remarks which we have made on the two preceding suffixes. In 
 the Doric forms of verbs in -μι it is written -rt,, which is either 
 softened into -6i, in the ordinary dialects, or represented by t, as is 
 the case in all the barytone verbs. For the imperative in -τω, see 
 above, § 351. 
 
 In the middle or passive, the regular ending -tai is invariably 
 preserved in the primary forms. We have already adverted to the 
 inference which may be drawn from the Sanscrit imperative tudatdm 
 (§ 352), namely, that -rai was originally -την. The form in -βΟ'ω 
 belong to a later falsification , which will be discussed in the proper 
 place. 
 
 356 1st Person Plural. 
 
 The arguments which have been advanced to prove the original 
 identity of the dual and plural of nouns , apply also to these num- 
 bers in the verb; besides, Buttmann has shown {Ausfuhrl. Sprl. § 87, 
 4, Anm. 1), that, in the case of the Greek verb, the dual is actually 
 nothing but an older form of the plural. We shall, therefore, con- 
 sider these numbers together. 
 
 In the active voice the Greek makes no distinction between the 
 dual and the plural of the first person ; the Sanscrit presents the form 
 vas or va for the first person of the dual in the active voice, but this 
 is only a modification of the plural mas; compare vayam for mayam, 
 &c. (Bopp, Vergl. Gramm. p. 331). 
 
 The ordinary characteristic of the first person plural in Greek is 
 -ftsi/, the older form is -/Ltgg, which is more analogous to the Latin 
 -mus^ and to the Sanscrit -mas. If our supposition that the singular 
 -mi stands for -me, the instrumental of the pronoun, be correct, then 
 -meSf the characteristic of the first person plural in old High German, 
 as well in the secondary as in the primary forms, sets this fact more 
 clearly before us. This -mes cannot stand for mds, as Bopp suggests 
 {Vergl. Gramm. p. 635). 
 
 When we reflect that the idea of "We," i. e. "I + you," cannot be 
 considered as contained in the plural of the first personal pronoun, 
 we shall refrain from adopting the simplest method of explaining this 
 characteristic, namely, by supposing that the final s is merely the 
 ordinary mark of the plural number. If, instead of this, we consider 
 
CHAP. I.] THE PERSON-ENDINGS. 579 
 
 the last letter as the characteristic of the second person singular, 
 which is of course allowable , we shall find this view harmonize with 
 all the other phenomena of the plural characteristics ; it is , too , the 
 only one which is consistent with our a priori expectations. In the 
 Veda-dialects we find the form -ma-si as the termination of the first 
 person plural of the active voice: this shows the two pronominal 
 elements at full length. 
 
 357 In the middle or passive voice, the terminations of the dual 
 and plural of the first person are in the oldest Greek forms -μεβ^^ον, 
 -μΒ6%•α, -με%'8ν, the more recent forms being -με^ον, -με^'α, where the 
 ϋ is omitted, as in the secondary form of the first person plural of 
 the active verb in Sanscrit. We are inclined, however, to consider the 
 dual form -με -dO-v, or the ^olic -με-ϋ'ε-ν, as the genuine charac- 
 teristic of the first person plural of the passive voice. The first 
 syllable represents the element of the first person singular, the ^f- 
 (Qo, Q'a) is, as we have shown , one of the forms of the second person 
 singular, and the final letter is -v, the oldest characteristic of the 
 locative case, which is necessary to the passive voice. 
 
 The Indian languages furnish analogies confirmatory of this view. 
 The characteristic of the Sanscrit active-dual is vas: that of the pas- 
 sive-dual va-ha-i. Here the ha stands for the second person (Bopp's 
 Vergl. Gramm. p. 651). Thus, Sanscrit va-ha=^Zendi να-ζα = Βχε; 
 Sanscrit de-hi = Zena daz-dhi=dido-^L, and conversely Zend hista-hi 
 = Sanscrit tishfh'a-si, and Zend dadhd-hi = Sanscrit dadd-si. The 
 syllable hai stands related to hi as -ύαυ does to -6i, that is, in the 
 relation of locative to instrumental. The same may be said of the 
 Zend -mai-dhe (for -dhi is one of the forms of the second personal 
 pronoun), and of the Sanscrit -mahe, -mahai. The forms -mahi, 
 -vahi, may be compared with the less genuine form -μεΟ'α•, they have 
 all lost the final w, the passive characteristic or locative ending, 
 according to the principle so often explained (§ 114). The full form 
 must have been -με-%'ην, which passed through -με^Όν and -με%αι to 
 
 358 2nd Person Plural. 
 
 In Greek the second person plural of the active voice appears in 
 a very mutilated state. It is invariably written -ra, or in the dual 
 -τον. We are enabled, however, by the aid of the cognate languages 
 to arrive at its real form. In Latin it is -tis or -tote. In the per- 
 fect, the plural in -s-tis must be explained in the same manner as 
 the singular in -s-ti. In Sanscrit we find -thas as a dual or older 
 form, and 'dha as the plural, which is mutilated like the Greek. A 
 
 PP2 
 
580 THE PEESON-ENDINGS. [bOOK IV. 
 
 comparison of the Greek dual -τον with the Latin plural -tis and the 
 Sanscrit dual -tJias, leads us to conclude that it stands for -rog, as -μεν 
 stands for -μες in the first person of the plural. "We should, however, 
 still be at a loss to explain the ending, were it not for the aid afforded 
 us by the sister languages. It appears from the Sanscrit tha-s , that 
 the second person dual is made up of a repetition of the second per- 
 son singular; and this is farther shown by a comparison of the Latin 
 imperative ending -tote=-tva-te, with the old Umbrian -tu-to. 
 
 This view is confirmed by the passive characteristic of the second 
 person plural -β&ε for -ΰ&αι, which in the dual or older form is -(?-0"o-v 
 for '6^ην, and in which the repetition of the second person singular 
 is manifest. 
 
 359 3rd Person Plural. 
 
 The third person plural presents us with difficulties considerably 
 greater than either of the other two. It is very hard to determine 
 with certainty the elements of which the dual characteristic is com- 
 posed, and still harder to reconcile the dual with the plural. 
 
 It is only in the historical tenses of the ordinary Greek, that the 
 second and third persons of the dual are distinguished from one 
 another: in the primary forms they are the same in both active and 
 passive, namely, both -τον in the former, and both -uQov in the latter; 
 in the secondary forms η is substituted for ο in the third person. In 
 Homer there are three passages in which we find the termination -ov 
 in the third person of historical tenses, and Elmsley has shown (ad 
 Aristoph. Ach. 733), that in the Attic writers the second person in the 
 historical tenses was occasionally written -την. It may be laid down, 
 with Buttmann (Ausfuhrl. Sprackl. § 87, Anm. 2, note), that in the 
 old Ionic the termination -ov was used for the second and third 
 persons of the dual in all moods and tenses; in the old Attic we 
 find the following distinction, — 
 
 Primary tenses and conjunctive, 2nd and 3rd . . . ov, 
 Historical tenses and optative, 2nd and 3rd . . . i^v, 
 
 while in the more modern Greek the distinction was. 
 
 Primary tenses and conjunctive, 2nd and 3rd . . . ov. 
 Historical tenses and optative, 2nd ov, 3rd . . . ην. 
 
 The same distinction was observed in the imperative; only in this' 
 case the third person was distinguished by ω instead of rj. 
 
 The letter n, in many modern languages an indistinct sound, has 
 made its appearance at the end of Greek words as the representative 
 of so many letters not much related to it, that it might seem strange 
 to any one not acquainted with the freaks of language. It has been 
 
CHAP. I.] THE PERSON-ENDINGS, 581 
 
 mentioned before (§. 86) that there are only four consonants which 
 ever stand at the end of the word in Greek, v, (5, ρ, and κ. Of these 
 ρ occurs very seldom in this position, and κ only twice, in f κ and ουκ, 
 which, as we have shown, are peculiar exceptions. With this fact 
 before us, we need not wonder that, when a consonant ought to stand 
 at the end of a word, in order to represent a significant suffix reduced 
 to its consonantal element, this consonant should so often be repre- 
 sented by v, which even takes the place of (?, the only other consonant 
 that frequently appears in the same way. Thus we have seen -μεν for 
 -^ε^, and -τον for -τοξ or -τεξ. On the same principle, we may conjec- 
 ture that the third person dual, -rov, stands for -τότε, or that, as the 
 second person-ending is made up of the element of the second per- 
 sonal pronoun twice repeated, the third should be constructed by a 
 similar repetition of the third person. We must, therefore, consider 
 the resemblance of the first and second persons of the dual in the 
 active as well as in the passive voice, as either produced by accident 
 or by a mistaken analogy. In the passive, as -udOV in the second 
 person stands for -ΰε-^Ο-ν^ a repetition, namely, of the second person 
 singular with a mark of the locative case, so in the third person the 
 same -ΰ%Όν must stand for -to-TO-v, namely, a similar repetition of the 
 third person singular with the same mark of the locative case. This 
 may seem wonderful, — perhaps, at first sight, hardly credible, — but it 
 is the only way of explaining the fact, and etymologically speaking 
 there is nothing against it. The Sanscrit presents the two character- 
 istics under a form in which we can more easily recognise the distinc- 
 tion of 'persons. In the active , the second person dual is -tha-s (for 
 -iha-tlia or tha-si), the third -ta-s (for -ta-ta)\ in the passive, the 
 second person dual is d-thd-m (for -thd-thd- with a locative ending) ; 
 the third, d-td-m or a-te (for td-td with a locative ending). 
 
 360 The genuine form of the third person 'plural of the Greek 
 active verb in -μι, is -vti, which is still found in Doric remains (Butt- 
 mann, Ausfuhrl. Sprl. § 107 , Anm. 7, note), and all verbs give -νται 
 in the middle or passive. The Boeotians wrote ~v%l for -ντι (Bockh, 
 Corp. Inscript. i. n. 1569, a. iii.). This is an approximation to the 
 ordinary Greek, in which the termination is -{^}(jt, or -6i with a 
 compensation for the -v. The Dorians wrote δώόντι, the Boeotians 
 δεδόαν^ί, the Attics dMa0L or διδονύι. We have mentioned before 
 that the roots of the verbs δίδωμι and τίΟ'ημυ are θε-, δο-, and that the 
 length of the penultima is occasioned by the insertion of a short a 
 which connects the root with the affix (§ 347). This addition to the 
 root is not found in the first and second persons of the plural, δίδομεν, 
 δίδοτ^,τί^εμεν-, τί&ετε, on account of the greater weight of these forms. 
 
582 THE PERSON-ENDINGS. [bOOK IV. 
 
 Now the first and second persons of the plural are proparoxytone, but 
 the third is paroxytone. As the contracted forms δίδονβι,^ tl^blol, 
 ίΰτάΰί, δευχννύι•, are properispome, we may conclude that the accentua- 
 tion of δίδόντί, TLd'BVTi,, is correct, that the fuller forms were δίδόα-ντί, 
 Tid'sa-VTLj and that the additional vowel was lost at a later period in 
 the third than in the two other persons of the plural. This view is 
 confirmed by theBoeotic δεδόαν^ι, and the Attic διδόαΰί,τιΟ'εάβί. That 
 a short a is frequently substituted for a v, we have shown in treating 
 of the declensions, and we find instances of it in the tenses of the 
 verb : thus we have κεκλίαταϋ for κέκλινται, ΰωΰοίάτο for όώΰοιντο, &c. 
 The V in διδόντι would therefore become a short «, and not a long one 
 as in διδόαβι, unless we suppose an original form διδόα-ντι, which 
 became δεδόαν^ι in Bceotic, and, substituting a for ν and 6 for ^ ac- 
 cording to the common practice, δίδοά-ά-6ί or δίδόαΰι, in Attic. That 
 a final a of the crude verb is implied in the termination -αΰί may be 
 also inferred from the perfect active , which invariably terminates in 
 a, and uniformly exhibits this form of the third person plural. "We 
 cannot believe that this final a was found in the present tense of 
 verbs which were otherwise strengthened by pronominal additions. 
 The form δευκνύάΰί^ to which Bopp adverts (Vergl. Gramm. p. 663), 
 appears to be the offspring of a later use of analogy, and was perhaps 
 suggested by the wish to avoid any risk of a confusion with the sin- 
 gular δείκννΰί. The greater weight of the person-suffix explains the 
 form τι^εαται, τίθενται. We must conclude that the ti passed through 
 %L into (5l before the ν was lost, and then the change of δίδόνΰί into 
 δίδονβί, τι^ένόί into τι^εϊΰυ, &c., is the same as that of τντίχονΰα 
 into τύτίτονβα, evg, into bIq, &c. 
 
 361 The Sanscrit presents us with the fullest analogy for this form 
 of the third person, both in the active and the passive — the former 
 being -nti, the latter -nte or -ntai. In Latin and old High German it 
 is -nt in the active. The Latin passive form in -ntur must be reserved 
 for a special discussion. Dr. Prichard (Eastern Origin of the Celtic 
 Nations, pp. 134 foil.) first called attention to the connexion subsisting 
 between the Welsh pronoun hwynt, "they" (written ynt when used as 
 a suffix) and the Welsh characteristics of the third person plural, -ant^ 
 -ent, or -ynt. He does not appear, however, to have taken the correot 
 view of the relation which subsists between the pronoun hwynt and 
 these suffixes in Welsh. The fact is, we conceive, that there are two 
 uses of wi considered as a reduplication of the third pronominal element. 
 It is (1) a sign of the neuter plural (§ 239), and as such appears in 
 hwy-nt; (2) a repeated demonstrative, as in the third person plural 
 of verbs. In this latter use, Mr. Garnett {Quarterly Beview, Vol. 
 
CHAP. I.] THE PERSON-ENDINGS. 583 
 
 LVii. p. 100) considers it analogous to the Esthonian need = iUi, and 
 derives it from a combination of the demonstrative roots na-\-ta. The 
 latter we have already; of the former he gives the following instances: 
 — na in the Finnish dialects "this" or "that;" Pali nam, "that;" 
 Greek vlv, "him," "her," "them;" Sanscrit ace. dual, ndu, "us-two;" 
 Gr. νώϊ; Sclav, dat. nawa; Plur. Sanscr. accus. nas; Zend no; Latin 
 nos; Welsh ni; Sclav, gen. nas; Pali ne, nd^ "those." We have 
 before stated our belief that all these have arisen from an obscurer 
 pronunciation of the demonstrative t. That η stands for this t in the 
 person-endings is shown by the secondary forms ετυτίτε-ν for ετύτίτετί, 
 &c., and we believe with Kuhn (de Conjugatione in -μι, pp. 23, 31), 
 that the first consonant in the plural termination -nti is a representa- 
 tive of the demonstrative element -i, so that the whole is a repetition 
 of the third person singular, perfectly analogous to that of the other 
 persons which we have seen used to form their plural. The voices 
 are distinguished by the same difference of case as in the other in- 
 flexions. The Doric dialect has preserved -ντω, the true form of the 
 imperative active (Ahrens, Dial. Dor. p. 296). The Attic -ντων, and 
 still more the later -τωβαν^ are due to corrupt analogies. 
 
 B. SECONDARY FORMS. 
 
 362 The secondary forms of the person-endings are generally 
 shorter varieties of the primary forms : this abbreviation is caused, as 
 we have already hinted , by the augmented length of the verb in the 
 historical tenses. The first person singular is marked by -i», a repre- 
 sentative of -μ or -μι. The same substitution takes place in the 
 Frankish language in the present tense: thus we have machon, "I 
 make," machos, "thou makest," machot, "he makes" and "ye make," 
 but machomes and machont for the first and third persons plural. We 
 find the same letter standing for -tt in the third person singular, and 
 for -VTL in the third person plural. It appears, indeed, to be the 
 regular abbreviation of the third person plural , for, though we have 
 beside it a longer form in -ύαν as the termination of the third person 
 in some of the historical tenses both active and passive , we consider 
 this as an illegitimate and later suffix. Eminent philologers (Butt- 
 mann, Ausfuhrl. Sprl § 107, Anm. 7, note; Bopp, Annals of Oriental 
 Literat. p. 60) regard this -6av as the third person plural of the verb 
 substantive; an opinion from which we entirely dissent, for the fol- 
 lowing reason: — those tenses in which this ending occurs do not offer 
 any trace of a periphrastic formation in the other persons. Let us 
 
584 THE PEESON-ENDINGS. [bOOK IV. 
 
 take a simple instance. The imperfect oi τί^ημι runs thus ίμ the 
 active and passive : 
 
 Active. Passive. 
 
 ετί%•ψν{=μ) εη^ε-μψ 
 
 Ιτί&η-ς εχί%•ε-6ο 
 
 ετί%'ψ(τ) ετί&ε-το 
 
 ετίΟ'ε-με^Όν 
 
 ετί&ε-τον ετί%•ε-ΰθΌν 
 
 Ιτι^ε-ΧΎΐν Ιτι^ε-6%ην 
 
 ετί^'ε-μεν ετιΟ'έ-μεΟ'α 
 
 ετίΟ'ε-τε ετί^'ε-ΰΟ'ε 
 
 ετί^ε-ΰαν έτί^'ε-ντο 
 
 Now if we compare these two forms with any common form of the 
 imperfect indicative, ϊτνΛτον, for example, we shall find that they 
 correspond exactly except in the third person plural. We can hardly 
 believe, therefore, that when such an absolute correspondence exists in 
 every other person, both active and passive, the third person plural of 
 the active alone should be really different in nature and origin. In 
 the analogies for such a variation, for instance, the third person plural 
 of the perfect passive τεχνμμενοι εΐβί^ the third person of the Sanscrit 
 periphrastic future, and the second plural of the Latin passive verb, 
 mentioned above, we have clear indications of a participle , which in 
 the last two cases appears alone; but in the case under consideration, 
 the first part of ετί^ε-ύαν for instance bears no resemblance to any 
 participle which could be joined with the substantive verb to form a 
 person of ετί^'ην. Accordingly we must seek to show rather that the 
 plural endings ετντίτο-ν, ετί^ε-6αν were originally the same, than that 
 they were different. And this we think possible. If we compare the 
 common ending of the third person plural in the present tense, namely 
 •■6l{v), with the old one in -vrt, we observe the following facts. In 
 the present tense a compensation has been made for the loss of the -v 
 in the original ending, and xvntov6i stands for τντίτον^Ί, and ultimately 
 for TVTttovri. The ν εφελκυβτι,κόν which is found at the end of this 
 termination, when the following word begins with a vowel, may have 
 owed its origin to an indistinct feeling that the η included in the 
 diphthong before the ending was still wanting, or from a mistaken 
 analogy : that, however, it was not essential is shown by the fact that 
 it never appears before a consonant. Supposing then that the original 
 forms of the imperfect were ετι,^'εα-ντΰ and ετύτΐτο-ντί, which may be 
 inferred from the middle ετί&εντο, ετντίτοντο^ and from the old Doric 
 accentuation of ελεγον, ετντίον, εφάΰαν (Ahrens, de Dialect, Dor. pp. 
 28 sqq.), we have only to inquire what abbreviations would most 
 probably result from the greater weight of the form. The existing 
 
CHAP. I.] THE PERSON-ENDINGS. 585 
 
 ϊχντίτον leads us to ετί%'εαν=ετίΟ'ην ; and as this would be identical 
 with the singular, the analogy of riuav=s-8(ja-VTi, and of aorists like 
 ετυφα-ν, assisted by the -ul-v of the present, would readily suggest 
 the insertion of 6. The thoroughly corrupt τυΛτέτω-ΰαν from tvtc- 
 rarco, shows to what an extent this secondary process might be car- 
 ried. In Sanscrit we have ahhd-n^ εφαΰαν, by the side of adadu-s, 
 εδίδοόαν; in the former case the -n of the suffix -nt is alone retained, 
 in the latter t is preserved and softened into s. 
 
 363 "We have thus seen that in the active voice, the secondary 
 forms, when they differ from the primary, are generally corrupted or 
 mutilated. The contrary is the case in the first person singular of the 
 secondary form in the passive , which is written -μην instead of -μοα, 
 and which, as we have seen above (§ 352), must have been the origi- 
 nal inflexion. Of the other person-endings , the first and second dual 
 and plural do not differ from the primary passive forms. The third 
 dual is, as we have mentioned , -6%'ην instead of -ΰ%•ον, which again 
 is more genuine. The second and third singular and the third plural 
 differ from the primary passive forms by being written -ΰο, -to, -ντο, 
 instead of -βαο, -rat, -νταί. The second person singular is generally 
 subjected to a contraction similar to that of the primary form; 
 namely, as τντΐτεΰαυ becomes tvTCry or τντΐτεί, so ετνητε-ΰο and ετν- 
 'ψα-6ο become IrvTCtov and ετνφω. 
 
 Bopp (Vergl. Gramm. 680) and Kuhn (de Conjugatione in -μι, 
 p. 25) have attempted to show that the middle or passive forms are 
 reduplications, namely, -μαι. for -μαμί or μαμά, and so on; so that the 
 agent as well as the object of the action are expressed by the affix. 
 The latter sees a confirmation of this view in the secondary form -μην, 
 which he thinks stands for μημ or μαμ. This opinion was probably 
 suggested by the erroneous belief that the person-endings are nomi- 
 native cases. We have already shown that this is an untenable hypo- 
 thesis. But there are other objections to this theory respecting the 
 person-ending. If we may consider the active forms as inflexions of 
 the affix, we may fairly conclude that the same explanation will apply 
 to the middle form, and there can be no more reason why part of the 
 active ending should be included in the middle, than there would be 
 for supposing that the locative ending must include the ablative. 
 Besides, in the remaining pronominal elements, which still maintain 
 an independent existence as particles, we have forms analogous to all 
 the person-endings of the verbs, and these particles cannot be ex- 
 plained as reduplications; why then should we attempt such an ex- 
 planation in the other case? The third pronominal root furnishes us 
 with the following analogies to the person-endings : rot (corresponding 
 
586 
 
 THE PERSON-ENDINGS. 
 
 [book IV. 
 
 to the tliird person singular active); to (corresponding to the third 
 person singular passive, secondary form) ; rs for τε-ν, comp. κε, κεν, &c. 
 (corresponding to the second person plural active for τε-ν or τε -g). 
 The first, the following: μοί=^μί (corresponding to the first person 
 singular active); μ'η=μαί, comp. d^ , δαί, &c. (corresponding to the 
 first person singular passive) ; μεν (corresponding to the first person 
 plural active, though the final letter is of different origin) ; μην (cor- 
 responding to the first person singular passive, secondary form. For 
 the change of (jat, rat, into (ίο, to, we may compare jt-QO with τίαραί, 
 νπό with vTtah &c. 
 
 364 The following table will enable the student to estimate at 
 one view the difference between the primary and secondary forms of 
 the person-endings in the two voices. 
 
 Primary form 
 Secondary form 
 
 Primary form 
 Secondary form 
 
 Primary form 
 Secondary form 
 
 1 
 
 -μυ 
 
 V 
 
 -μες, -μεν 
 -μες, -μεν 
 
 ACTIVE. 
 
 SINGULAR. 
 
 2 3 
 
 -ul•, -Q'a, '%Ί,, -ί£\ -τω, -ti, -<Jt, -ι 
 -ξ, -%'α \ -ν 
 
 OVAL. 
 
 2 
 
 -τον 
 -τον 
 
 PLURAL. 
 
 2 
 
 -τε 
 -τε 
 
 3 
 
 -τον 
 -την 
 
 -ντω, -ντί,-ν&ί,-\1\ΰυ(ν) 
 -ν, [-ΰαν] 
 
 PASSIVE. 
 
 Primary form 
 Secondary form 
 
 Primary form 
 Secondary form 
 
 Primary form 
 Secondary form 
 
 1 
 
 -μαυ 
 -μην 
 
 SINGULAR. 
 2 
 
 -(5ai^ -y^ -ει 
 -ΰο, -ου, -ω 
 
 -μεΟ^ον 
 -μεΟ^ον 
 
 -με^^α 
 -με^α 
 
 DUAL. 
 
 2 
 
 ■ύ^ον 
 
 6%0ν 
 
 PLURAL. 
 1 2 
 
 -ό^ε 
 '0%ε 
 
 3 
 
 -ταί 
 -το 
 
 3 
 
 -ΰ&ην 
 
 ■νταί, -ατ<χί 
 ■ντο, -ατο 
 
CHAP. I.] THE PEESON-ENDINGS. 587 
 
 365 The person-endings in Latin have been occasionally cited; 
 but they- involve some difficulties , which seem to invite a special dis- 
 cussion. In the active these are preserved more faithfully than in 
 the Greek, because the laws of the language do not forbid either m or 
 i, the signs of the first and third person, to stand at the end of the 
 word. In the present indicative , the sign of the first person singular 
 is invariably wanting, except in the case of sum and inquam. The 
 passive presents much greater difficulties. If we compare amo, amat, 
 amant, amare, with their passive forms amo-r^ amat-ur, amant-ur, 
 amari-er, we should be inclined at first to suppose that the termina- 
 tion r or er, ur, was sufficient to convert an active into a passive 
 form. But then, what does this termination mean, and how are we 
 to explain amaris and amamini, which do not receive it? Bopp 
 (Annals of Oriental Literature, i. p. 62; Vergl. Gramm. p. 688) and 
 Pott (Etymol. Forsch. i. p. 133, 6) suppose that this final r is a repre- 
 sentative of the reflexive pronoun, the termination of the second 
 person singular being an inversion for ama-sir: Kuhn (I. I. p. 26, 
 note) suggests that it may represent the radical consonant of the sub- 
 stantive verb. The latter conjecture rests on assumptions respecting 
 the nature and origin of the person-endings of which we have before 
 expressed our disapprobation. The opinion which Bopp first stated, 
 and which he has now resumed after having abandoned it, rests upon 
 nothing except the analogy of the Sclavonic, Lithuanian, and other 
 modern languages *. And this analogy does not furnish prima facie 
 evidence of the same value as the consistent harmony of Greek and 
 Sanscrit, which shows that the person-endings are inflexions of the 
 pronouns, the active representing an ablative or instrumental, and the 
 passive a locative case. In the active of the Latin verb, the case of 
 the pronoun is not fully expressed, but we should infer from the 
 Greek and Sanscrit person-endings that the original forms exhibited 
 the ablative or instrumental inflexions, and this inference is confirmed 
 by the imperatives in -to, -tud, -to-te (Garnett, Essays, p. 99, note). 
 
 * We cannot help regarding it as an Indirect testimony against the 
 soundness of this hypothesis, when we find it warmly espoused by a 
 writer who seems to be unable to take a correct view on any point of 
 grammatical criticism. The Professor of Comparative Philology whose 
 ludicrous suggestions have been noticed more than once in these pages, 
 has written a long paper (Proc. of the Philol. Soc. Vol. v. nos. 107, 108), to 
 show that the r which marks the person-endings of the Latin passive verb, 
 was nothing else than the accusative se of the reflexive pronoun, and he 
 even goes so far as to seek for a similar explanation of the Greek ter- 
 mination -ai, forgetting that it is the same as the Sanscrit e, and was 
 also represented by -ην, -cc, -ov, and -o. But, as Alcibiades says, ηερί 
 όμολογονμένης άνοιας ούδεν ctv γ,αινον λέγοιτο. 
 
588 THE PEESON-ENDINGS. [BOOK IV. 
 
 That the passive person-endings in Latin exhibit an oblique case of 
 the same force as that which marks the passive in Greek may be 
 proved to demonstration. The steps of the proof are as follows: 
 (1) The fullest forms of the passive inflexion in Greek are -αί, -ην 
 (like δαί, δήν)^ with weaker forms in -ov, -a, -o. (2) Corresponding to 
 these we have adverbial inflexions in -δην, -δον, -δα. (3) These ad- 
 verbial inflexions have the same force as the Latin adverbs in -tim 
 and -ter; compare βά-δην with pedetentim, and 6μο%'νμα-δόν with 
 concordi-ter. (4) In the isolated form igi-tur this ending appears with 
 the short u instead of e, as in the third person singular and plural of 
 the Latin passive verb. And there can be no doubt as to its locative 
 meaning. According to Festus (p. 105 Muller) igitur signifies inde, 
 postea, turn. It means "thereupon" in a fragment of the xii Tables 
 (Varron. p. 203): si in jus vocat, ni it, antestator ; igitur em capita. 
 And it denotes "then" as the antecedent to quando in Plautus, Miles 
 Glor. III. 1, 177: quando habeho^ igitur rationem mearum fabric arum 
 daho. From all this it is certain, if any thing can be certain in phi- 
 lology, that the -ur or -er of the Latin passive is a locative ending 
 equivalent to the -ην or -au of the Greek. It is only in the third 
 person, singular and plural, that we have the affix complete. The 
 first person singular has sufiered a contraction similar to that which 
 we have pointed out in the Sanscrit: amor stands for amomur, as 
 Mare does for bharame. The first person plural is also contracted 
 (amamur stands for amamusur) , but not to so great an extent. The 
 second persons singular and plural require a difi'erent explanation. 
 {a) The former, which ends in -ris or -re, is thought by Pott and 
 Bopp to be reducible to the analogy of the other persons, according to 
 the theory which they have adopted regarding their formation. Thus 
 Pott says (Etym. ForscJi. i. p. 135) that ama-ri-s is equivalent to 
 amas se, the final s being the reflexive pronoun, and the ri being the 
 usual alteration of the complete [second person-ending -si. Bopp 
 ( Vergl. Gramm. p. 688) doubts whether to adopt this view, or to sup- 
 pose that there is a metathesis of the ending, so that amaris stands 
 for amasir. There are several objections to this mode of considering 
 the subject, even though we should suppose that the final r of the 
 imaginary amasir could be taken as the sign of the locative 
 case, which would leave the vowel i unexplained.. In the first 
 place, this hypothesis creates a difficulty in the case of the ending -re, 
 which is found consistently in the imperative and generally in the 
 oblique moods and tenses , even in the best writers ; for, if this -re is 
 merely the mark of the second person, the second persons in question 
 are all active forms. Again , it will be recollected , that in the active 
 voice of the Greek and Latin verb the second person of the imperative 
 
CHAP. I.] THE PEBSON-ENDINGS. 58^ 
 
 difiPers from tlie others, in the absence of any sign of person, not in 
 the want of any other characteristic. Thus, we have τνπτε, ama, &c. 
 The principle is the same as that which has occasioned an omission of 
 the nominative sign in the vocative case (above, § 252). Analogy 
 would, therefore, favour the supposition that the -s which is omitted 
 in amare is the sign of the second person, and not a mark of the 
 passive voice. The word amare , which remains, is identical with the 
 infinitive active. Now the infinitive and imperative have other points 
 of resemblance in regard to form, as we shall see in a future chapter; 
 thus Tv^m, the second person singular first aorist imperative middle, 
 might bg thought the same word as τν^αι, the first aorist infinitive 
 active : it is also well known to every reader of Greek, that the infini- 
 tive is often used for the imperative. The Latin infinitive in -re cor- 
 responds, as we shall show more at length hereafter, to the JEolic 
 infinitive in -i,g=-(3i; they are both verbals, formed with the second 
 pronominal suffix, and used in the locative case. We shall also 
 endeavour to prove that a passive verbal is formed from the active 
 infinitive. Now the difference between amare and amarier is one of 
 case only : the more common form of the latter word is amari, in 
 which even this difference is neglected, for the i belongs to the verbal 
 in -ris = -sis, as appears from the forms amaris^ γελαϊξ. All things 
 considered, it appears to be far from improbable that the second per- 
 son passive, amare, is this verbal which is also used for the infinitive, 
 and that amaris was formed, by an addition of the characteristic of 
 the second person, on a mistaken analogy. If this supposition cannot 
 be admitted, we can only suppose that the r of amaris^ &c. represents 
 the ^ which is inserted in certain forms as a mark of the passive in 
 Greek. That r had this value in old Latin and the other Italian 
 languages is well known ( Varron. p. 82). And it is possible that the 
 r at the end of igitur, sapienter, &c. was identical in origin as in use 
 with the Greek --O't. (b) "We might suppose that the second person 
 plural of the Latin passive verb, which ends in -mini, bore a similar 
 relation to the Greek infinitive in -μεναο, and indeed this has been 
 suggested by Grafe (das Sanskrit Verbum^ p. 120). That this, how- 
 ever, is not the case, but that the second person plural is a participle 
 in the nominative case with the copula suppressed, formed indeed by 
 the same pronominal suffix as the Greek infinitive referred to, but not 
 used like it in the locative, appears from the following considerations, 
 which were first brought forward by Bopp (Annals of Oriental Lit. i. 
 p. 51; Vergl. Gramm. p. 689). It is clear that the Latins had a 
 passive participle in -minus: we find it rather shortened in alu-mnus; 
 Vertu-mnus (comp. κρήδεμνον, &c.) ; calu-mnus presumed in calu-mniay 
 and obviously the participle of calvor (Gains, L. 233, pr. D. de Verh^ 
 
590 THE PERSON-ENDINGS. [bOOK IV. 
 
 Signif.; Varron. p. 204); auctu-mnus; ceru-mna; da-mnum, &c.; 
 and at full length in ter-minuSj fe-mina, and ge-mini for genimini. 
 In the imperative it is written -minor ^ as in amaminor. Now the 
 sign of the plural nominative in the old Umbrian is- -r; thus we have 
 in the Eugubine tables suhator, screhitor for suhacti, scripti^ corre- 
 sponding to a singular form in -o, as orto, suhato for ortus, suhactus; 
 and, what is still more to the point, the Latins had a singular impera- 
 tive in -mino ; thus we find f amino = dicito, in Festus (p. 87) , prce- 
 famino=prcefato, in Cato (B. B. 141), and in an inscription, published 
 by Gruter (p. 204) , we have is eum agrum nei habeto nei fruimino, 
 where the participle in -mino is clearly used for the third person sin- 
 gular. There need be no confusion between the -r of -minor and that 
 which marks the other persons of the passive voice. As -r is the mark 
 of both genitive singular and nominative plural in the Umbrian, so 
 its substitute -s designated both of the same cases in old Latin, and 
 its vocalization -i again the same two cases in the more modern state 
 of the language. Nor need we be surprised that the termination is 
 fixed, and does not change according to the gender of the substantive; 
 for when the use of this participle without the verb estis became per- 
 manent and regular, the gender of course would not change : if it did 
 it would be unnecessary to investigate the form at all, since it would 
 then stand on the same footing with the Greek use of the participle 
 for the third person of the perfect passive, τετνμμενοί εΐύί being the 
 regular form. In this case, however, the copula is never omitted. 
 At least, it is clear that in the only instance which has been quoted 
 to prove the contrary*, namely, ^schyl. Eumen. 340, the true reading 
 is not ότΐενδόμεναι, for the immediately following εμαΐξ, καταφέρω, &c. 
 would then be intolerable ; and the μηδ^ ε^ αγκρίΰυν έλΰ'εΐν would 
 have no meaning, if applied to the Eumenides, whose special business 
 was to impeach the homicide. Having regard to the metre and to the 
 whole context we have ventured to read the passage thus in our re- 
 cension of the Eumenides (Lond. 1848): 
 
 ύΛενδόμενος δ' άφελεΐν τίνα τάΰδε μερίμναζ, 
 ^'εών δ' άτέλειαν εμαΐς λεΐταίξ ετακραίνείν, 
 μηδ' εζ αγκριΰιν ελ&εΐν, 
 Zεvg aiμoΰτaγεg άξώμίΰον 'd&vog τόδε λέόχα^ 
 ag άπηξοώόατο, 
 
 i. e. "Zeus, being anxious (making a business /or himself ^ cf. Agam. 
 147) to relieve every one from these cares, and by my ministrations to 
 
 * In ^sch. Suppl. 40, we ought to read ϊπνΛΐ%Χ6μΈ%α. The change of 
 number in the antistrophe merely indicates, as usual, a solo by the Coryphaeus. 
 
CHAP. I.] THE PEBSON-ENDINGS. 551 
 
 obtain an immunity for the gods , and not to be troubled with trials, 
 has renounced all intercourse or dealings with the odious race of 
 homicides." Although we do not believe that the participle in -μένος 
 is used as a primary predicate without the copula, we are aware that 
 with other participles this construction is not uncommon (see JEschyl. 
 Agam. 169, 515, ed. Klausen). And we need not trouble our- 
 selves to find Greek analogies for a phenomenon which is one of 
 the most singular features of the Latin language. The best syn- 
 tactical parallel for Bopp's explanation of the second person plural 
 passive in Latin, is one which he has pointed out in the Sanscrit lan- 
 guage. There is a periphrastic future in that language, of which the 
 third person, singular, dual, and plural, appears to be nothing else 
 than the nominative masculine of a participle having a future sense 
 and formed by the suffix tri. In the nominative case singular of this 
 participle the r is left out and a long a substituted for it, as in ddtay 
 "about to give" (daturus)', but the nominative of the dual and plural 
 are datarau, ddtdras. The first and second persons of this future are 
 expressed by a combination of this participle with the verb substan- 
 tive, as will apear from the following scheme: 
 
 
 Sing. 
 
 Dual. 
 
 Plur, 
 
 1 
 
 Ddtdsmi 
 
 Ddtdsvas 
 
 Ddtdsmas 
 
 2 
 
 Ddtdsi 
 
 Ddtdsthas 
 
 Ddtdstha 
 
 3 
 
 Ddtd 
 
 Ddtdrau 
 
 Ddtdras 
 
 In the first and second persons the participle and verb are sometimes 
 separated by the intervention of other words, as in kartd tad as mi 
 ie, "I will do this for you," — facturus hoc sum tibi (Mahd-BJidrata) ; 
 hatham trdtd parasutdn as if "why are you going to preserve the 
 sons of another?" — cur servaturus alius filios es'? (Rdmdyana). The 
 suggestion of Ewald, in his review of Grafe (Zeitschrift fur die Kunde 
 des Morgenlandes for 1837, pp. 110 foil.), that -mini is a corruption 
 of the Sanscrit -dhvam^ does not appear to us to merit any serious 
 attention. 
 
 366 We cannot leave the subject of the person-suffixes without 
 making a few remarks on an interesting phenomenon which Bopp was 
 the first to explain satisfactorily, and which he calls the "influence of 
 the weight of the person-endings." Every student of Greek grammar 
 must have remarked, that though the existing verbs in -μι are all 
 formed from roots ending in a vowel or -vv, which is regularly length- 
 ened when joined to the person- endings , yet this is the case only in 
 the singular of the active voice. The explanation of the phenome- 
 non is as follows. In the process of abbreviation always going on in 
 
592 THE PERSON-ENDINGS. [bOOK IV. 
 
 languages which admit of indefinite composition, there is a war between 
 the body of the word and its appendages; and when these latter 
 become very weighty, the curtailment falls upon the body of the word. 
 Now all the dual or plural person-endings are duplicates of the sin- 
 gular, and, therefore, twice as heavy, and the passive endings are, as 
 we have seen, necessarily fuller than the active. Accordingly, to keep 
 up the equilibrium of the verb , it was necessary that the root should 
 be left in its simple form in those persons where a heavier system of 
 terminations was adopted. In the third person plural active the long 
 syllable is retained, but, as we have shown, by a sacrifice of part of 
 the termination. For the rest, we may repeat what we have said 
 before, that a principle of equilibrium or compensation is observed* 
 throughout the Greek language, so that this "influence of the weight 
 of the person-endings" is not a new or isolated phenomenon, but a 
 good example of the general rule. 
 
CHAPTER Π. 
 
 THE TENSES, 
 
 3G7 Relations of mood and tense. Their proper arrangement. 368 Limi- 
 tations of Bopp's theory of agglutination. 369 Genuine forms of verbs 
 include pronominal adjuncts. 370 Augment and reduplication are not 
 identical. The augment i- from άνα expresses distance. 371 The future 
 G- an indication of proximity. 372 Connexion of aorist and future. Bur- 
 nouf's theory of the tenses. 373 How the aorist combines the expres- 
 sion of posteriority with that of past time. 374 The 2nd aorist. 375 
 The desiderative in -οείω. 376 The 1st and 2nd perfect. 377 Aorist 
 and perfect agree in termination, and differ in the reduplication of the 
 latter. This is shown by the Latin forms of the perfect. 378 Repeti- 
 tions of the termination in Latin. 379 Passive aorists: erroneous views 
 respecting these formations. 380 Supposed active aorists which must be 
 
 ; considered as belonging to this class. 381 Aorist in -ην due to the 
 insertion of the pronominal element ya. 382 Aorist in -^ην similarly 
 formed by the insertion of thya. 383 This last element is not imme- 
 diately connected with the root of τί&ημι. Middle forms with transi- 
 tive signification. 384 Middle futures to active verbs. 385 Abnormal 
 formation of passive futures. 386 Tenses in cnov. 387 Verbs in -σκω. 
 
 367 TT is difficult to discuss the various questions connected 
 ■*• with the development of the tenses in Greek , without 
 including some topics which properly belong to an inquiry into 
 the origin and meaning of the modal inflexions, because, as we 
 shall soon see, the inflexions of tense and mood are in fact 
 identical. In the progressive analysis , however, of the verbs, 
 it is customary to consider the expression of the relations of 
 time immediately after those of number and person, and distinct 
 from those of modality ; and since these must be considered as 
 subject to some sort of modality, to treat of them in the first 
 instance as they appear in the indicative mood, in that mood 
 which is always used in the oratio directa when nothing beyond 
 assertion is intended. For form's sake, we shall abide by the 
 old method. 
 
 The tense-system of the Greek verb is wonderfully compre- 
 hensive ; in number of tenses it far exceeds the Latin , and in 
 the preciseness of their significations it leaves the Sanscrit a 
 long way behind. The proper and most general division of 
 the Greek inflexions of tense and mood, is into definite and 
 indefinite forms. In accordance, however, with the old system, 
 we will first discuss the tenses agreeably to the three great 
 
 QQ 
 
594 1ΉΕ TENSES. [BOOK IV. 
 
 divisions, — according as they signiiy present,past^ or future time. 
 Of these divisions, we must omit the present tense: the affec- 
 tions of that form are not so much variations of tense, as pecu- 
 liarities of conjugation ; it belongs, therefore, to a future chapter. 
 Before we discuss in detail the different tense-forms, it may 
 be convenient that we should state the general results at which 
 we have arrived. It is hoped that the reader will be convinced 
 that the proper expression of past time is by means of the 
 prefix a- or e- from a-na, denoting distance or separation ; that 
 future time is expressed by the second element, under the form 
 'S implying proximity; that continuous action is denoted by 
 reduplication ; posteriority in past time by a combination of the 
 prefix e- with the affix -s; and continuation up to the present 
 time, by a combination of the reduplicated root with the latter 
 affix. These are the regular forms. Abnormal varieties will be 
 noticed in the proper place. 
 
 368 It will, perhaps, be proper that we should in the first place 
 make a few remarks on the theory of Bopp, according to which many 
 of the past and future tenses are formed by agglutination, or compo- 
 sition with the substantive verb. Thus he thinks (Annals of Oriental 
 Literature, p. 45) that -εΰω, middle -Ββομαί, is properly the character- 
 istic of the future tense, and that this is merely the present tense of 
 the substantive verb ^ΕΣ provided with -ω for the termination, to 
 which the usage of language has given a future signification. He 
 even goes so far as to assume that Ιύονμαι may be an abbreviation of 
 Ιΰεύομα^ and (p. 61) that εϊηΰαν is a compound of εϊη and ΰαντίΐ He 
 conceives there is an analogy for this in the Sanscrit future character- 
 istic -sydmi. "It may be supposed," he says (p. 47), "that the root 
 As would have had a future tense originally, and it seems to me 
 credible that sydmi is this future tense, being lost by lapse of time in 
 disconnected use, and being found at present extant only compounded 
 with attributive roots." In the same way he considers the first aorist, 
 as it is called [ετντΐ-ΰα, &c.), as a compound of the root of the attri- 
 butive verb with the first preterite of the substantive verb, and simi- 
 larly analyzes /w-erwwi ior fuesunt, fu-erim for fuesim, fac-sim, &c. 
 In favour of this general view, he adduces the Provengal compound 
 futures, aurai for aver ai, &c. (p. 46). But these are widely diflferent 
 formations from those which he imagines in Sanscrit and Greek. The 
 two parts of the compound are both existing words, and may be 
 written separately, as indeed appears from the instances which he 
 
CHAP. Π.] THE TENSES. 595 
 
 quotes from Sainte-Palaye : compatar vos ai for je vous compterai; dar 
 vos w' ai ior je vous en donnerai; dir vos aiior je vous dirai; dir vos 
 em for nous vous dirons; gitar m'etz for vous me jeterez. See also 
 Raynouard, Gramm. Bomane, p. 221. Whereas his supposed com- 
 pounds are made up of a root, which of course cannot exist separately, 
 and of a termination which never does appear as a distinct word ; for 
 there is no such verb as 86ω in Greek, and sydmi is equally imaginary. 
 There are instances of compound tenses in Sanscrit: namely, the 
 future, of which we have already spoken (§ 362), and which is com- 
 posed of a participle and the full verb asmi ; also, a preterite formed 
 of an abstract substantive in <5, used only in the accusative am, and 
 corresponding to an infinitive in Zend, and one of the three verbs 
 dsa, "I was," hahhuva, "I have been," and chakdra^ "I have made:" 
 thus from the root ις, "to rule," we have the abstract substantive ίρα, 
 accusative igdm, and by composition with the perfects of as-^ hhu, and 
 kr%, are formed the perfects igdm-dsa, igdm-babMva^ and tgdny-cJiakdra, 
 all signifying "I ruled" (Bopp, Krit. Gramm. p. 229). These com- 
 pounds might be divided, as is often the case with the former. To a 
 certain extent, we recognise the truth of Bopp's theory in its applica- 
 tion to the Latin tenses in -ho, -ham., and ~vi (see Vergl. Gramm. 
 p. 804, cf. Varron. pp. 353 sqq.), and we have elsewhere pointed out 
 an agglutination running through all the tenses of the Latin verbs in 
 -so, -sivi {Varron. p. 352). There are also instances of auxiliary or 
 periphrastic formations in Greek, of which we shall speak hereafter, 
 but they are all made up of distinctly developed verbs, and therefore 
 furnish no analogy for the support of Bopp's theory. In general, we 
 have no more right to presume a composition in etymology when the 
 elements never exist separately, than we have to infer an ellipsis in 
 syntax, when the supposed full form never occurs. 
 
 369 But perhaps the greatest objection to this comprehensive 
 theory of agglutination, arises from its contradiction to a mode of 
 developing the tenses natural in itself and supported by every analogy 
 of comparative grammar. In the verb, as well as in the noun, there 
 is a wide distinction between compound words and those which are 
 merely developments of a root by means of pronominal additions. In 
 pointing out the analogy between the verb and the noun, we have 
 mentioned that the person-endings in their modifications correspond 
 to the cases. We are convinced that the difi'erences of tense and 
 3nood , and , in some instances, of voice, were originally expressed by 
 pronominal adjuncts, the same in kind with those which constitute 
 the affixes between the root and case-ending of a noun. There are in 
 fact two ways in which the crude-form of a word, whether it be a noun 
 
 QQ2 
 
596 THE TENSES. [bOOK IV. 
 
 or a verb, may be afifected. It may either be affected internally, that 
 is, by reduplication, guna^ or anusvdra^ or externally, by means of 
 some prefix or affix. The first method is adopted in the two primary 
 tenses, the present and perfect, as will be shown in the chapter on 
 the conjugations. The second is applied to the formation of all the 
 other moods and tenses, and, in some cases, also to the expression of 
 the passive voice. This external pronominal affection is brought 
 about in two ways; first, by a simple prefix of the demonstrative 
 element a or ε, called the augment: secondly, by an affix which is 
 always some modification of the second pronominal element : thus we 
 have aorists and futures under the form sa', perfects under the form 
 ka or ha; the optative mood under the form ya; the passive voice 
 under the form ya or thy a; and sometimes two forms of the same 
 element are combined, as in the iterative s-ka, and, according to one 
 view, in the desiderative s-ya, and the aorist th-ya just mentioned. In 
 a subsequent chapter we shall refer to the same root the derivative 
 affixes in -ξω=dya, and -εω, &c.=^ya. We begin with the augment. 
 
 370 In the Greek system of tenses, past time is denoted by a 
 short ε prefixed to the verb, or, apparently, by a reduplication of the 
 first consonant and root-vowel, which, however, is generally altered 
 according to certain rules. Buttmann is inclined to consider the 
 latter as the original characteristic of past time, the former being 
 a mutilation of it {Ausfuhrl. Sprl. § 82, 3, note). Even though we 
 had no other objection to offer to this view, we should consider Bopp's 
 argument fatal to it. The historical tenses in the Sanscrit verb are 
 marked by an augment a : the perfect, in the same way as the Greek, 
 by reduplication : but, as Bopp remarks {Annals of Oriental Litera- 
 ture^ p. 41), "the Sanscrit augment has no connexion at all with the 
 reduplication, because the redoubled consonant is generally articulated 
 by the vowel of the root; tup forming tutup-, and lig, ΙιΙις; now, if 
 the first preterite of these roots were utopat, ilegat, instead of atopat, 
 alegat, then it might be said that there exists some connexion between 
 the reduplication and the augment, when there also would be a mere 
 inflexion , whilst , in its. actual state , I consider it as an affix which 
 had its proper signification." There are, however, other reasons for 
 believing that the augment and reduplication are essentially different. 
 Besides the repetition of the initial consonant with ε to form a per- 
 fect, there is another reduplication, frequently found in verbs in -μι, 
 of the initial consonant with a short i, to form the present and im- 
 perfect. Thus we have τό-Ο'η-μί in the present, and τε-ΟΈί-κα in the 
 perfect. We believe that there is no essential difference between 
 these two prefixes : the difference of tense is expressed by the suffix -κα 
 
CHAP. II.] THE TENSES. 597 
 
 and the change of the root-vowel. What the prefix or reduplication 
 means, we must now endeavour to find out. The root O'f- might 
 appear in a noun, — d^e-og for instance — as well as in the verb τί-Ο'η-μί. 
 In the noun it might imply that the person indicated was "a placer," 
 but in the verb it must convey an idea of an action — "a placing." 
 Now an act necessarily implies a continuance or duration, and what 
 more simple method of expressing this could we desire than by a 
 repetition of the root? At any rate, we are told that, in the broken 
 English of the Negroes , acts are expressed by repetitions of substan- 
 tives. Just so we believe it was with the original Greek present 
 tense. The present and imperfect both express a continuance — 
 τί^ημυ, "I am placing," ετί^ην, "I was placing." The perfect ex- 
 presses the continued state after the act had taken place: it is a 
 mixture of present and past time : it speaks of the person's state or 
 duration as present, of the act as past; τε-Ο-Β-κα, "I am in the state 
 of a person who has placed." The distinction between the aorist and 
 these reduplicated tenses consists in this, that the aorist always speaks 
 of the action as single, as confined to a point of time, and this mean- 
 ing it retains through all the moods, even the imperative, as Greek 
 scholars have fully shown. Now the aorist never admits of redupli- 
 cation, even in those verbs in which the present, &c., are reduplicated, 
 except in some cases, when it entirely loses its meaning. 
 
 In his earlier treatise on the verb , Bopp endeavoured to connect 
 the privative signification of the particle a- with the temporal aug- 
 ment. He wrote as follows (Annals of Oriental Literature, p. 27): 
 "What the a, prefixed to (Sanscrit) verbs in order to form a preterite, 
 originally signified, I do not know, but this I know, that it is pre- 
 fixed in the same manner to nouns with the sense of a negative and 
 privative particle; for instance, adina, happy (not miserable), anin- 
 dita, dear (not despised) , abala, weak (without strength), &c. It 
 would not by any means be contrary to the general practice of lan- 
 guages, if by the words adina, anindita, exceeding the primary sense 
 of the negative particle a, the Sanscrit had also signified one who has 
 been miserable, who has been despised — but who is not now miserable, 
 not now despised; in that case there might have been a closer con- 
 nexion between a negative and a preterite, than would be evident at 
 first sight; or, in other words, the particle a, expressing in its primi- 
 tive sense negation, can Jvery properly in a secondary sense indicate 
 past time, that is to say, deny the existence of the action or quality 
 v/ith respect to the present time. One might ask why in this way a 
 is not as well employed to form the future tense , for neither in this 
 tense does the action or quality expressed by the verb actually exist; 
 but the usage of language is despotic, arbitrarily employing its means 
 
598 THE TENSES. [bOOK IV. 
 
 \vithout controul. Another objection could be taken against the 
 original identity of the negative «, and the a expressing past time, 
 from the case of the first being employed in Greek under the form of 
 a, the second under that of an a, so that different forms answer to 
 different meanings. But it is very often the case, that one original 
 word produces, in languages originating from others, two, three, or 
 more words, with slight variations in form and meaning; and this 
 practice has particularly contributed to the copiousness of the English 
 language. For instance, to stay and to stand have the same origin; 
 both are variations of the Sanscrit root sthd, to stand. / stay agrees 
 with the German verb ich stehe, signifying Ί stand,' of which the 
 preterite is, ich stand (I stood), which is considered as a new root in 
 English." In our opinion, an opinion subsequently adopted by Bopp 
 himself and by other philologers, the augment is the pronominal root 
 1=:ια=άνά (§ 114), which we find elsewhere as denoting distance or 
 remoteness*. We have seen how this idea of separation is connected 
 
 * The opinion that the augment is a pronominal particle denoting dis- 
 tance or remoteness, that is, originally in space, and then in time (above, 
 §§ 54, 55), was stated, as in the text, in the first edition of this work 
 published in February, 1839, and we claim this theory as an important 
 discovery, which has been confirmed by the subsequent adhesion of some 
 of the best philologers. Bopp, -whose explanation of the augmented past 
 tense as a negation of the present, had been severely ridiculed by Lassen 
 in 1830 (Jndische Bihliothek, Band in. Heft i. p. 79), and by Grafe in 1836 
 (das Sanscrit Verbum, p. 103, note 57), made an attempt to defend it in 
 the 4th part of his Vergleichende Grammatik , published in August, 1842, 
 when he had been possessed of the first edition of this work for more 
 than three years. In § 540, however, without any acknowledgment, he 
 adopted our identification of the negative a-, ava, with the pronominal 
 word signifying distance, and recognised the possibility of the latter mean- 
 ing in the augment. He was followed in 1846 by G. Curtius, who in his 
 Sprachvergleichende Beitrage gives more than one indication of his ac- 
 quaintance with the New Cratylus, and who in pp. 128 sqq. formally 
 adopts our view on the suggestion, as he professess, of Bopp, considering 
 the augment as equivalent in meaning to the German da, especially in the 
 combination damals. A similar view is accepted by Steinthal, in his essay 
 de Pronomine RelativOy 1847, p. 62. He says: "augmenti a significationem 
 statuo esse proprie: illud, illic, tunc, {εδων==ίηηι dans ego, 7]v=tum sum)." 
 But by far the most important confirmation of our theory was that which 
 it received in May in 1844, from the excellent English philologer Mr. 
 Richard Garnett, who contributed to the Philological Society (Proceedings, 
 Vol. I. no. 23) a paper "On the Origin and Import of tlie Augment 'in 
 Sanscrit and Greek," in which his object was "to show that the expla- 
 nation which Bopp himself allows to be admissible — namely, that the aug- 
 ment may be regarded as a demonstrative particle, primarily expressing 
 remote place, and secondarily remote time — is the one which unites the 
 most probabilities in its favour," Mr. Garnett supports our view by argu- 
 ments deduced from an examination of a great number of languages. In 
 the second edition of his tract, entitled Sprachvergleichung in ihrem Ver- 
 hdUniss zur classischen Philologie, published in 1848, Mr. G. Curtius has 
 
CHAP. Π.] THE TENSES. 599 
 
 with that of the first personal pronoun (p. 244) , and we shall thus 
 understand how the separate particle sa-ma^ which denotes complete- 
 ness or all between the near and the here, is used as a mark of past 
 time in Sanscrit (p. 256). The definiteness of locality presumed by 
 these pronominal roots would render them unsuitable for the expres- 
 sion of future time; in looking back on a past action, we always, in 
 some measure, fix its position, or regard it as fixed, with a subordi- 
 nate idea of distance, — "that which is gone by;" in looking forward 
 to the future, our anticipations are always indefinite and confused, 
 Jbhough perhaps connected in all cases with a sentiment of proximity 
 or approach — the idea of "that which is coming." The pronominal 
 root, in the former case , is used as a prefix , and the accent is drawn 
 back on it to express that the time referred to is gone by, just as 
 conversely future time is expressed by a suffix. A striking analogy 
 in support of this is found in the Greek prepositions of two syllables, 
 which throw their accent backwards or forwards, according as the 
 noun to which they refer precedes or follows. We observe something 
 very similar in the shifting of the accent in common conversation. 
 It is scarcely necessary to refer to the argument for the difference of 
 the augment and reduplication which is afiOrded by the augmentation 
 of the reduplicated perfect to form the plusquamperfectum. 
 
 371 The addition of the letter -6- to the root of the verb is the 
 commonest method of denoting the future in Greek. This letter, as 
 we have mentioned above, represents the second pronominal element ; 
 its use to denote the future is due to the principle just alluded to, 
 
 the effrontery to say (p. 72), referring to his Beitrdge of 1846, "I have 
 since seen the explanation I have there given confirmed on perfectly dis- 
 tinct grounds. Mr. R. Garnett has arrived at the s^me opinion quiie in- 
 dependently of me." Independently enough no doubt; for he adopted in 
 1844, and Bopp accepted in 1842, a theory which we had put forward in 
 1839; so that Mr. Curtius had little right to claim as his own discovery, 
 a piece of information, which he either had acquired or might have acquired 
 exactly seven years before he wrote a line on the subject. On the whole, 
 we feel ourselves constrained to assert that Mr. Curtius, availing himself 
 of the ignorance of English philology, which is a characteristic of his con- 
 trymen, has endeavoured to appropriate to himself the merit of the attempt, 
 which was first made by the present writer, and that he not only bor- 
 rows from us many details, but gravely puts forward totidem verbis as his 
 own original thought the idea which led to the composition of this book. 
 Even his title, Sprachvergleichende Beitrage zur griechischen una lateinischen 
 Grammatik, "Contributions of Comparative Philology to Greek and Latin 
 Grammar," seems to have been suggested by the title " Contributions (i. e. 
 Sprachvergleichende Beitrage) to a more accurate Knowledge of the Greek 
 Language." We have given elsewhere a sufficient proof of our charge 
 against Mr. Curtius (Classical Scholarship and Classical Learning, pp. 234, 
 235). 
 
QQQ THE TENSES. [bOOK IV. 
 
 that the future, or, as the Germans more correctly call it, Zukunft, 
 properly refers to an approximation or "coming." Indeed, the idea 
 of union is not excluded from our notions of future time. "The next 
 day," means the day following, not the day past: and as the augment 
 points to ανά , indicating " distance," so the future sign leads us to 
 ijv-v, indicating conjunction. We find this illustrated by the phrase- 
 ology of the best authors. Thus, oi ανω^Έν, or ετΐάνωχΙ'εν, are those of 
 former times (Theocr. vii. 5), but το εχόμενον hog (Thucyd. vi. 3), 
 "the adjoining year," means the year which follows. In certain cases 
 the letter -6- is not immediately joined to the root, but a short vowel 
 e intervenes, just as we have both liebete and liebte in German, both 
 charmed and charm' d in English , or , to take a still more striking 
 analogy from our own language, just as the e is regularly sounded in 
 some participles, and as regularly dropt in others. Of this common 
 future in -(ίω, there are two leading modifications in Greek; they are 
 called by grammarians the Attic future, and the second future. The 
 distinction between these two futures , which both end in -ώ in the 
 Attic dialect, consists in this, that, whereas in the Attic future in 
 -εω, -ώ ; -αω, -ώ ; the εον a belongs to the root, — in the second future 
 the ε included in the ώ is added to the root: and that in the verbs 
 which form the future in -ιώ for -ίόω the characteristic of the verb is 
 dropt; whereas the second future keeps its characteristic unaltered. 
 The second future is the regular form for verbs ending in λ, μ^ ν, ρ. 
 It is to be observed that the lonians used the uncontracted form in 
 -έω instead of the Attic future, and that in Homer we have such 
 futures as ερνω, ταννω for ερνΰω, ταννΰω. In general, we may say 
 of the two shortened futures , that they are abbreviations , the Attic 
 future of a future in -^ω, the second future of a future in -ε-(>ω, the (J- 
 being omitted as in τύτίτει for τντίτεόαι, δημοω for δημόΰίο^ &c. (see 
 § llU). If we can understand that δημόόιος and χQύΰεog belong to 
 the same formations, we shall have little difficulty in identifying the 
 ordinary future in -6ω with its weaker form in -εω. 
 
 372 The circumstance most deserving of notice in the future is 
 its connexion with the tense called the aorist. Thus we have, cor- 
 responding to the regular future τντΰ-ΰω,ϋαο regular first aorist ε-τντί- 
 6a, actually differing from it only in the augment. In fact, ετν-φα^^^ 
 ετνψον (^.τυπτόμε&α^τντντόμεχϊΌν) stands related to %ν^ω=^τν^ομι^ 
 just as ετί%'εα=^ετί%'ην stands related to τί^ημυ, that is, as imperfect 
 to present. We shall show, in the following chapter, that the same 
 relation subsists between the subjunctive and the optative. The con- 
 nexion between the future and the aorist was first pointed out and 
 explained by J. L. Burnouf, in his Methode poiir etudier la langue 
 
CHAP. II.] THE TENSES. 601 
 
 Grecque^ § 255, from a comparison with the tenses of the French verb*. 
 As we think his way of considering the subject very good , and as it 
 has not been sufficiently, if at all, attended to by those who have 
 written on the tenses, we shall give his illustrations as nearly as pos- 
 sible in his own words. He divides the tenses of the Greek verb into 
 two classes, the principal, and the secondary. The principal tenses are, 
 (1) the present, (2) the future, and (3) the perfect. The secondary, 
 (1) the imperfect, (2) the aorist, and (3) the plus-perfect, each of 
 which is formed from the corresponding primary tense. The follow- 
 ing investigation of the French verb lire shows the connexion in 
 meaning between the primary and the secondary tenses: 
 
 I. Principal tenses, which express that the action has relation to the 
 time of speaking. 
 
 (1) Present, je lis, "I am reading," i. e. at the present moment. 
 
 (2) Future, je lirai, "I shall read," i. e. at some period suc- 
 ceeding the present moment. 
 
 (3) Perfect, fai lu^ "I have read," i. e. at some period pre- 
 ceding the present moment. 
 
 The whole duration of time is thus divided into three portions , the 
 present, which is fixed, so that if you say, "I am reading," no one 
 will ask you "when?" — and the future and perfect, which are fixed 
 relatively to the present. For the assertion "I shall read," or "I 
 have read," would convey a clear and intelligible idea, even though 
 you should answer, "I do not know," or "I do not remember," to the 
 question "when will you read?" or "when have you read?" So that 
 the primary tenses enable us to see at once to which particular point 
 of time, — present, future, or past, — the act relates, and are therefore 
 absolute and independent, and express only a simple relation to one 
 of the three points of time. 
 
 n. Secondary tenses, which imply a relation to some point of time 
 other than the present. • 
 
 (1) Imperfect, je lisais, "I was reading." 
 
 (2) Aorist, ye lus, "I read." 
 
 (3) Plus-perfect, favais lu^ "I had read." 
 
 All these assertions suggest the question "when?" and if you would 
 have your hearer understand you, the precise point of time, whe7i you 
 
 * This explanation has been adopted, with some additions which do 
 not add to its distinctness, by Mr. Mansel, in the North British Review, 
 Vol. XIV. p. 55. 
 
602 THE TENSES. [bOOK IV. 
 
 were reading, or read, or had read, must be stated. Accordingly, not 
 being determined by themselves, they require some additional state- 
 ment to fix their meaning; and thus they express a double relation or 
 two relations^ (1) to the past, generally, and this is determined by the 
 forms themselves ; (2) to some fixed point in the past. The primary 
 tenses, therefore, may be called definite {determine)] the secondary, 
 indefinite or half-definite (indetermine or semi-determine). The former 
 express only one relation, and this relation is determined by their 
 form; the latter, two relations, of which the form determines only ' 
 one. 
 
 DEFINITE TENSES. 
 
 The present expresses simultaneity -j ■,,.-,. 
 
 rp, Ρ , J. • -i. ( relatively to the time of 
 
 The future posteriority > "^ 
 
 The perfect anteriority ^ Ρ iiig- 
 
 INDEFINITE TENSES. 
 
 The imperfect expresses simultaneity — ^je lisais pendant que vous 
 ecriviez. 
 
 The aorist expresses posteriority — je lus apres que vous eutes fini 
 d'ecrire. 
 
 The plus-perfect expresses anteriority — j'avais lu avant que vous 
 eussiez ecrit. 
 
 Now all these tenses express anteriority alone, in regard to the time 
 oT speaking. The relation, in which they differ from one another, is 
 the only one expressed by the definite tenses. It is, therefore, by a 
 natural analogy that, in the Greek language, the imperfect is derived 
 from the present, the aorist from the future, and the plus-perfect from 
 the perfect, by prefixing the augment, which is the mark of past time, 
 to these tenses, which in themselves denote simultaneity, posteriority, 
 and anteriority. 
 
 373 This view of the case will contribute materially to the better 
 understanding of the whole system of moods and tenses in Greek. 
 We have seen, from what Burnouf says, that all three of the tenses 
 which express a double relation may be called άόρίύτοί, or indefinite. 
 But the second of them, which is formed from the future, is peculiarly 
 so, from the mixture of past and future time implied in it, and there- 
 fore the Greek grammarians have particularly distinguished it by this 
 name. Thus we find it used in cases where we should expect one of 
 the other indefinite tenses, though never, we believe, for a definite 
 tense. In the passage quoted from Xenophon by Burnouf (§ 357), as 
 
CHAP. Π. J THE TENSES. 603 
 
 an instance of its use for the perfect, it bears the proper aorist sense, 
 or implies posteriority in reference to a fixed point of time : rovg %'η- 
 0avQOvg των πάλαυ ΰοφών, ovs εκεΓι/Οί. κατέλίτίον εν τοις βφλίοίς 
 γρά^αντες, ύυν rotg φίλοίς διέρχομαι, i.e. they first wrote them down, 
 and then left them. When the aorist expresses repetition or contin- 
 uance, and thus seems to approximate to the present, it always has 
 reference to some fixed point or circumstance , which is necessary to 
 define it. In the same way, the optative, which is the aorist of an old 
 future, is used to express repetitions. In any mood except the indica- 
 tive the aorists are employed to denote single acts or transient time, 
 and the first aorist conveys this meaning occasionally even in the in- 
 dicative, especially in those cases where it implies that something fol- 
 lowed a given event, and is itself completed, so that we may dismiss 
 the subject. Compare the use of ^νεΰα and its compounds (Gr. Gr. 
 Art. 427 (dd)). AVhen deprived of its augment, as in the infinitive mood, 
 the aorist may be used for the future after μέλλω (see Porson, ad 
 Eurip. Orest. 929). The Greeks very frequently described historical 
 events in the present tense. Similarly, we sometimes find the imperfect 
 in a narrative, where we should expect the aorist, especially in Homer 
 and Herodotus, the narrator representing the action as if he had been 
 present at the time when it was going on (see Matth. Gr. Gr. § 505). 
 A similar feeling gave rise to the employment of the imperfect, in the 
 old αναγραφαί and in Pindar, to describe a victory gained at the 
 public games (see Dissen, ad Find. JS'em. v. 5 ; Thucyd. m. 8 ; v. 49, 
 and Arnold's note on the latter passage). The same is observable in 
 the inscriptions on works of art, as ^Ατίελληζ ετίοίει, where Pliny (H. 
 N. I. 20) gives a special explanation of the phrase, and refers it to 
 the modesty or caution of the artists. 
 
 374 The relation between the second aorist, as it is called, and 
 the second future, does not seem to be the same in all verbs. In 
 some the second aorist, like the corresponding future from which it is 
 derived, is merely a shortened form of the first aorist in -βα (above, 
 § 371); thus we have second future βαλώ, second aorist ϊβαλον\ and 
 as the former stands for βαλέω=βαλέΰω, so the latter represents 
 εβάλεον, as appears from the infinitive βαλεΐν (in Ionic βαλέειν), and 
 the participle βαλονΰα^ Doric βαλοΐόα, Ionic βαλενύα. Such a first 
 aorist as έμεινα is formed on the compensation principle for εμενι/α= 
 εμένεΰα. It will be recollected, that although we retain the names 
 first future, first aorist, second future, second aorist, we do not mean 
 that each verb was provided with such an apparatus of longer and 
 shorter forms. Some would have the more complete tense only; 
 others, only the contracted one; and in the same verb, certain writers 
 
604 THE TENSES. [bOOK IV. 
 
 would adopt the former, while others would prefer the latter. The 
 opinion of Herodian {Bekker. Anecd. p. 1290), an opinion adopted by 
 many modern scholars (PMlol. Mus. ii. p. 205), that there is no such 
 thing as a second future, is only so far true as this — the second future 
 is not a distinct tense, for it is only a contraction ; but there are such 
 contracted futures; indeed they are the regular forms for verbs the 
 roots of which terminate with a liquid; and we might go so far as to 
 say, that every tense of the subjunctive mood is an instance of the 
 shortened future, to which the corresponding tense of the optative 
 stands in the relation of aorist. But besides this second aorist, which 
 appears as the correlative tense to the second future, and may be 
 recognised in the optative as opposed to the subjunctive, there may 
 have been some cases in which this tense, as denoting single or trans- 
 itory acts, never had the future (j- of the proper or first aorist. 
 Whether this tense, which has an augment without reduplication or 
 affix, is to be regarded as an aorist, or as the imperfect of the oldest 
 semel-factive verb, it is clear that its meaning corresponds to that of 
 the aorist, though it is differently expressed. For while the form 
 which has the augment ε- as well as the affix -(5 implies that the act 
 was future and is past, or that it took place within limits, which 
 require to be defined, the shorter form expresses much the same 
 meaning when it indicates a transitory or momentary action com- 
 pleted in past time (see below, § 443). 
 
 375 From the future τύφω is formed the desiderative τν^Ι^είω, to 
 which again the form TvipSia stands as imperfect ; the latter is generally 
 considered as an optative aorist, and the desire or wish conveyed by 
 it has deprived it of all actual reference to the past , and therefore of 
 its augment. Besides this, the formation of a new present tense 
 from some past or future tense of a verb is one of the commonest 
 phenomena in the Greek language: thus we have from the aorist ηκα 
 the new present η}ΐω , and from the perfect τέϋ'νηκα the new present 
 τεΟ'νήκω, &c. 
 
 377 The Greek grammarians acknowledge a first and second per- 
 fect as well as a first and second aorist. The first perfect is thus dis- 
 tinguished. When the final letter of the root of the verb is β, jt, φ, 
 o^ 7i ^) Xi ^^is consonant either becomes aspirated or remains so. In 
 all other cases the characteristic of the first perfect is -κα. The second 
 perfect, though it occasionally admits of alterations of the root, prin- 
 cipally by guna, adds nothing but -a, -ag, -£, as a termination. Some 
 scholars think that the second perfect must be considered as older and 
 
CHAP. II.] THE TENSES. 605 
 
 more organic than that called the first, as it is formed out of the 
 resources of the root itself without foreign additions, and corresponds 
 exactly to the proper perfects in Latin and Sanscrit, and to the perfects 
 of the first six of the strong conjugations in old Low German. We 
 entertain very strong doubts as to the truth of this assertion (see Gr. 
 Gr. p. 185). The second perfect must be a mutilated form , for the 
 past time implied in this tense could not be conveyed by the redu- 
 plication alone. When we compare the aorists εΟ'ηκα, έδωκα, with the 
 perfects τεθεί-κα, δεδωκα, we perceive the only real difference to be that 
 the aorists have augments, the perfects reduplication. And if we 
 
 take the perfects ^ , ^ / r ,1 -, »-, / Λ we see that the termination 
 ^ y τετντΐ-ϊΐα ' ' I λελεγ-η3, ] 
 
 is simply Tia. That the Sanscrit ρ in most cases represents a Greek 
 
 κ we have already seen ; it has also been shown how the soft sound 
 
 of s is often substituted for the hard κ Of the substitution of h both 
 
 for k and s it is unnecessary to speak. 
 
 377 It appears then probable, from observation in the Greek 
 language alone, that the terminations of the perfect and aorist are iden- 
 tical. The analogy of the Latin language renders this all but certain. 
 The Latin conjugation is exceedingly incomplete. There are no means 
 of expressing past time by augmentation, and very few verbs have 
 a proper reduplicate perfect, as tetuli from tollo, pepuli from pello, 
 cecidi from cado. In some the syllable of reduplication is completely 
 lost, as infidi iromfmdo; in others, the length of the penultima is 
 all that remains of the original form of the perfect , as in legi from 
 lego &c. In a few cases we still have both forms : thus, we have both 
 tetuli and tuU, both tutudi and tudi; and these may be properly 
 termed perfects^ for they are equivalents , more or less mutilated in- 
 deed, to the corresponding Greek tense. Another perfect , so called, 
 ends in -si, which is never reduplicated, and therefore seems more en- 
 titled to the name of first aorist. It is true that the flexion is that 
 of the other perfects (-^, -isti, -it, -imus, -istis, -erunt), and it would be 
 easy to say that this is a form of the perfect which has lost its redu- 
 plication, just as the past tense in Latin has always lost its augment. 
 The fact, however, that we have a future in -sim, as faxim (fac-sim), 
 corresponding to this tense in -si, furnishes a sufficient reason for 
 believing that the latter was an indefinite tense or aorist. We shall 
 not be able to settle this question completely without examining the 
 inflexions of the real or reduplicated perfect. The most undoubted 
 and ancient form of this tense is furnished by the verb fuio (Gr. φνίω), 
 which appears both as fio and fuo {Gr. Gr. Art. 321). The latter, still 
 farther shortened into bo, furnishes the usual future of all vowel verbs ; 
 
606 THE TENSES. [bOOK IV, 
 
 thus αηια-δο=α7ηα-/ηο, &c.;its imperfect, originally e-fuarrij under 
 the form e-bam, appears as the adjunct to the imperfect of all verbs, 
 not excluding fio itself in the later condition of the language ; and 
 its perfect fuvi=^fufui^ under the weakened form fui or ui or vi, 
 furnishes a perfect to all verbs ending in a vowel; thus amavi=ama- 
 fuij &c. (see Bopp, Vergl. Gramm. p. 804; Varron. pp. 353 sqq.). 
 Now as fufui may be properly compared with πεφνκα^ as i is the 
 regular exponent of guttural vocalization , as the guttural , before it 
 subsides into t, is generally softened into s and Λ, and as we find s, k, 
 and h, in the aorist and perfect of Greek verbs, we have abundant 
 reason to believe that fufui stands for fufusa, which is again an off- 
 spring of fufuka. If then we restore the regular inflexions of the 
 assumed /w/M5a we shall get: 
 
 fufusa-\m\ =fufuis =: fufui 
 fufusa-tha=fufuis-ti 
 fufusa-t =fufui-s-t =fufuit 
 fufusa-mus =fufui-s-mus =fufuimus 
 fufusa-tis =fufuis-tis 
 fufusa-nt=fufue-s-nt=fufuerunt 
 
 If we admit this transposition and substitution, which seem to be 
 justified by general principles and by the analogy of the French 
 change of / through ul into u, we must apply the same explanation to 
 all regular perfects. Thus tutudi stands for tutudis=^tutudsa, &c. 
 But it is manifest that the forms in -si contain something more than 
 a mere s. According to the principle stated above (§ 372), the future 
 in -sim must be antecedent to the tense in -si, supposing that this 
 latter is an aorist. Now if we compare /ac-siw, for example, with 
 sim=:siem (Varron. p. 345), we shall feel justified in concluding that 
 fac-sim=zfac-siem is analogous to the desideratives in -ΰείω, and that 
 dic-si, for example, corresponds rather to δύξ,πα than to edei^a. The 
 early loss of the primitive system of augments and reduplications in 
 the Latin language, has introduced a mode of extending the affi^ or 
 person-endings, which we shall see also in the latter Greek imperative. 
 "We regard this as springing from a false sense of analogy : for the affix 
 5- does not denote past time but future. "We must not, however, 
 forget that these subsequent extensions invariably presume a neglect 
 pr ignorance of the original significance of formative elements ; con- 
 sequently, that the procedure is entirely conventional and arbitrary. 
 In point of fact, there is no more difficulty in understanding the 
 transition oi e-dixi^ e-dixis, &c. into dixi, dixisti, &c. than in seeing 
 the reasons for the change of the active τυτΐτε-τ-ω into the passive 
 τνπτε-ΰΟ'-ω^οη the analogy oi TvnTE-r-ov,TV7tT8-ud'-ov. On the whole. 
 
CHAP. Π.] THE TENSES. 607 
 
 then, we may fairly conclude that the suffix of the future, aorist, and 
 perfect, as well in Greek as in Latin, is the same, being always some 
 representative of the second pronoun; that in the case where this 
 characteristic appears to be lost, it has past into h or i; and that 
 while the Greek distinguishes the aorist and perfect from the future 
 by augment or reduplication, in Latin the proper distinction has 
 been lost, the only differences which remain being accidental and not 
 essential. 
 
 378 This view of the case explains not only such futures as ταννω 
 for χαννβω, &c., and βεΐομαι, for βί/^ομαί=βΐ λύομαι (root βιΓ^ La- 
 tin viv-o=qviqv-o, above, § 112) ; and such aorists as εχενα or εχεα for 
 εχεΓΰα (root %εΡ) ; but also the mutilated perfects like olda, and those 
 forms to which the participal nouns in -νια (αγνια, &c., § 296) \vould 
 naturally be referred. We are led, however, by the result of this in- 
 vestigation, to a special inquiry respecting the original form of the^/w5- 
 quamp erf ec turn. As εχεα and οϊδα presume a lost ΰ-, it might be in- 
 ferred that ετετνφεα stands for ετετνφε7ια = ε-τε-τν7ΐ-ύε6α 'from 
 τε-τνφα = τέ'τν7ΐ(^^)α. It has been observed above (§ 347), that in 
 the present τίϋ'ημί, and therefore in the imperfect ετί^εα for ετί^'ην, 
 the root Ό'ε- is connected with the person-ending by an intervening a. 
 In the other persons we generally find ετί^ης, έτίΰ'η. As ετί^'εεg, ετί- 
 %εε are presumed in the Attic forms srtO'fti, ετί%ει^ and as this com- 
 bination generally implies an included ya (^ 116), we may conclude 
 that the root %ε- is strengthened in the present and imperfect by this 
 pronominal addition. Now in the plusquamperfectum we not only find 
 ετετνφεα^ Ιτετνφεε, but the other persons also end in -εα^, -εαμεν, 
 -εατε, -εαν; so that ετετνφεα stands for ετετνφειν or ετετνφεαν=ετε- 
 τντί-ΰε-ύα, and not merely for ετετνφα-ν, for then the ε would be inex- 
 plicable. The same conclusion would be deduced from the form -εια for 
 -ευν, which is found in inscriptions. Now as we should expect a priori 
 that the plusquamperfectum would differ from the perfect only in the 
 augment prefixed, we must regard the double addition of the future or 
 aorist affix as the result of a later and abnormal analogy. The Latin 
 verb, however, fully shows us the possibility of such a procedure. If 
 we compare fui=fuesa-m with fueram^ji]fuesam, we shall recognise 
 the last faint traces of the legitimate formation ; and we see of course 
 the same regularity in the inflexions of those verbs which form their 
 perfect and plusquamperfectum with the aid of these tenses. Here, 
 however, the parallelism of the definite and indefinite tenses termi- 
 nates, and even in the substantive verb the transitions are effected by 
 accretions of the affix: thus from fuerim=fuesiem we ha,Ye fuissem= 
 fuesesiem; and in the verbs which have the aorist-perfect in -si^ the 
 
G08 THE TENSES. [bOOK IV. 
 
 same abnormal formation is found even in the indicative ; thus from 
 rexi we have plusquamperf. rexeram=reg-se-sam and rexerim=^reg-se- 
 siem, which is again lengthened into rexissem=^reg-si-se-siem. From 
 this use of the element s- to transfer the definite tense into the cor- 
 sesponding indefinite , it might seem that the same suffix is employed 
 to indicate both future and past time : but this is not exactly the case. 
 In the formation of the aorist from the future, it was intended to ex- 
 press posteriority in relation to some past event (above, § 372), and 
 we observe that this suffix is never used by itself to signify past time 
 in Greek; this is always done by the augment. The want of an augment 
 in Latin, and the gradual loss of a regular future by the substitution 
 for it of a subjunctive , gave this termination the improper influence 
 which it exerts in that language. 
 
 379 The form which some of the tenses present in the Greek 
 passive has occasioned difficulties which no philologer has hitherto 
 been able to surmount. We conceive that the general principles which 
 we have laid down at the beginning of this chapter will afiOrd a satis- 
 factory explanation of these troublesome phenomena. It has been men- 
 tioned , that in Sanscrit there are two forms of verbs , considered ac- 
 cording to their person-endings ; the one is called parasmaipadam, or 
 transitive, and has endings in the instrumental case; the other, called 
 dtmanepadam, is middle or deponent, and has endings in the locative 
 case. The passive voice is formed from the dtmanepadam^ by insert- 
 ing the pronominal syllable ya between the root and the ending. Now 
 we find that, in Greek, the present tense of the dtmanepadam is used 
 as a passive and also as a middle, and the passive forms of the other 
 tenses generally bear the same relation to the active forms that we 
 find in the present passive or middle, as compared with the same 
 tense in the active ; the difference, namely, is only in the ending. In 
 the perfect and plusquamperfectum the characteristic 6-, κ, or h is 
 frequently dropt altogether, as we have seen in the case of the second 
 perfect and second aorist. The form ετν^Ι^άμην from 8τν^α-{μ)^ which 
 we should expect as the passive aorist, is never strictly passive, except 
 in those cases where the reflexive and passive significations are inter- 
 changed or become commutable. The forms actually used as passives, 
 ετνφ^'ην, ετυτΐην, &c., have the active person-affixes throughout all 
 the moods , and philologers are quite at a loss to determine whence 
 they have got their uniformly passive signification. Bopp {Berlin. 
 Jahrb. 1827, pp. 284 foil; Vocalismus, pp. 53 fol.) has suggested two 
 methods of explaining the former. He conjectures that it is either 
 formed of the verbal adjective and the substantive verb (ετνφ&ην for 
 τνπτοξ ήν), or by an addition to the root of the aorist ε&ην, on the sup- 
 
CHAP. II.J THE TENSES. 609 
 
 posed analogy of the Gothic sokidedum, " seek-did-we," and the Latin 
 vendo. Pott (Etymol. Forsch. i. p. 47) prefers the former of these ex- 
 planations; Kuhn {de Conjugation in -μι^ p. 67) is inclined to adopt 
 the latter. On a former occasion {Annals of Oriental Literature, p. 39) 
 Bopp proposed a different conjecture, which shows to what extremities 
 despairing ingenuity may be driven. He suggested that ετ^φ^'-ι^ν, fiddO"- 
 gjv, &c., may proceed from the passive participles τνφχΡ-ύς^ do&sig^ sub- 
 stituting for the termination -scg the person-terminations -ην, -ης, -ι/, 
 &c. But how does the syllable τνφ%'-, with the active participial end- 
 ing, come to be a passive participle? And how does it happen that 
 τνψχί'είξ is explicable, and Ιτνφ%'ην not ? In fact, this explanation tells 
 us nothing: he might just as well have said that ϊ^-ην was formed 
 from O'-stg, by the substitution of -ην for -ftg! Perhaps, however, 
 Grafe's remark (p. 114) is even more absurd than any of these. He 
 says that ετντίον is an imperfect of the -co conjugation, Ιχνίίην of the 
 conjugation in -μι, as if the distinction between the conjugations in -ω 
 and -μυ could in any sense correspond to a difference of voice ! 
 
 380 That there is some foreign element in forms like ετνφΟ'ην, 
 every one is constrained to admit at first sight, but no one has at- 
 tempted to explain Ιτντίην otherwise than as a mutilated form of 
 ετνφΟ'ην, except Bopp, who considers it as a compound with the sub- 
 stantive verb ην ( Vocalismus, p. 54). This misconception of the second 
 aorist passive has led to some extraordinary classifications in the com- 
 monest verbs. To take one of the most obvious instances, the verb 
 ϊβτημι^ root ^ta, which means "I cause to stand," has a transitive 
 future and aorist 6χη6ω, εύτηΰα, regularly formed from it. Kow in 
 every grammar we find in the paradigm of the active voice an aorist 
 and perfect εότην, εΰτηκα, which bear a passive signification ("I stood" 
 or "was caused to stand ") throughout all the moods: whereas είίταΟ'ί^ι/, 
 which is a synonym of ϊβχην^ is not placed in the active but in the 
 passive paradigm. It is true that at first sight Ϊ6την seems to corre- 
 spond ϊοε%'ην,&,ο., but if we compare ΰτηναυ,ΰτηϋ'ί,εότημεν,ΰτάς, &c., 
 νήϋι^'εΐναι,,Ο'ες,εΰ'εμεν,&ο., and εΰτηχα with τεϋ'εικα, we shall be con- 
 strained to admit that these are widely different forms, and that the 
 root ατα has suffered some change in these tenses which has not been 
 experienced by the root -Ο'ε in the others. But ετέ%'ην, which stands for 
 ε^'έ^'ην (one of the aspirates being necessarily abolished, and the second 
 retained in preference to the first in consequence of the importance of 
 the termination), corresponds to the other aorist εβτά^ην; and εΰτην, 
 6τη(3θμαί; εβτά^ην, βτα^ηόομαι•, are perfectly analogous to Ιτντίην, 
 τνΛηΰομαο', ετυφ^ην^τυφ^ηύομαΐ) so that τί^ημι must be considered 
 as having lost its second aorist passive, and ϊύτην, ύτηύομαι, must be 
 
 R R 
 
610 THE TENSES. [bOOK IV. 
 
 placed by the side of εότά&ην, βτα^ήΰομαυ, in the passive paradigm. 
 We find other instances of a loss of the second aorist, and we must 
 determine from the meaning and the form in the plural and other 
 moods, whether the active or the passive aorist is the one wanting. 
 Thus ϊδων is not only active in signification, but we see from εδομεν, 
 dog, dovg, &c., that the form contains no foreign element; whereas 
 when we find φύω {φνμή, φνύω, εφνβα, active ; but εφνν, τΰεφνκα, 
 passive: δύω, δνβω, εδνόα^ active; εδνν, δεδνκα, passive: ΰβένννμΰ, 
 ΰβέΰω, active; εββην, passive: βήΰω^ εβηΰα, active; εβην, βέβηκα, 
 passive: &c., and also observe the ίοΥΏΐΒφνναί, εδνμεν^ εΰβημεν,βηναι^ 
 we conclude that in these cases the root has received some accession, 
 and that the verb has no short form of the active aorist. 
 
 381 Having vindicated the claim of these intransitive aorists in 
 -rjv, - w, to rank with the passive forms in -%^ην, we must endeavour to 
 point out the pronominal element which has given them their passive 
 signification. This is, in our opinion, the second pronoun, under that 
 form which indicates the locative case. It is used to form passive and 
 other derivative verbs in Sanscrit, such as denominatives and causals. 
 That it was of most extensive application in Greek, we shall see in 
 another chapter. In almost all cases, however, it has been absorbed 
 by one or other of the various euphonical artifices which the fineness 
 of the Greek ear necessitated. Among other instances of this, it may 
 be easily seen, that it lies hid in the derivative verb-endings -αω, -όω, 
 in many of those in -υω, and also perhaps in some of those in -έω 
 (Bopp, Vergl. Gramm. p. 727). A similar absorption has taken place 
 in the optative of verbs in -νω, -νμί(Β\ιατα&Ώ.η,Αη8/.8ρνΙ. ξ 107 ^Anm. 
 36), and we have seen something of the kind in ημεροζ for δι/ά-μερος. 
 We consider, then, that εΰτην, εφνν, εδνν, &c., stand (for εότά'υαμι^ 
 εφvyaμLJ εδνραμί, &c., respectively. The Latin language affords us an 
 excellent example of the way in which this pronominal formation can 
 give a passive sense without the addition of an dtmanepadam affix. It 
 seems that in Latin the contracted verbs in -ao agree in their uses 
 with the Greek in -έω, and those in -eo with the Greek in -άω or -ya. 
 The Greek conjugation in -εα> was, as we shall see, that which was 
 always adopted in forming verbs from compound nouns; thus, ευεργέ- 
 της made ενεργετεω, &c. The same was the case with the Latin verbs 
 in -ao: thus from Icetificus we have loetificari, &c. Again, we find 
 that many active verbs in Latin, either uncontracted or contracted in 
 a, have a neuter or passive verb from the same root which is uncon- 
 tracted, or, what is more usual, contracted in e. We will take a few 
 instances: 1st, active uncontracted, passive contracted in e; active 
 pendercy passive pendere; active scandere^ passive scatere; active 
 
CHAP. Π.] THE TENSES. 611 
 
 pandere^ passive patere; active sistere, passive stare (this is perfectly- 
 analogous to ΐύτημν, εΰτην)] actiYejacere, passive ^«cere; 2nd, active 
 contracted in a, passive contracted in e or uncontracted ; in this case, 
 it will be observed, the active verbs are all derivatives from the verbal 
 nouns of the neuter verbs: Sbctive sedare, passive «ecZere ; SLctive parare, 
 ■passive parere ; active liquare, passive liquere; active /w^are, passive 
 fugere. The same method which the Latin language has applied in 
 the formation of complete verbs has been adopted in Greek for the 
 construction of one tense, and the forms dependent on it; and there 
 are traces of the same insertion in certain verbs, for example, in 
 τίονάω, as distinguished from τΐονεω (Bockh, ad Find. Pyth. iv. 236 ; 
 Hermann, Dial. Find. p. xv). 
 
 382 The aorist in -Ο'ην is easily explained. It contains the ele- 
 ment -9' in the locative form Q'l. We might say then that the form in 
 L was the ultimate state of that in thi, just as αντω=αντό-υ stands for 
 avto-Q'L; for there is no reason, etymologically speaking, why Q'l and t 
 should not be considered as identical, any more than there is for dis- 
 tinguishing between the noun-endings -Gig, -La; the adjectives in -0l- 
 μος, -ίμος', and the genitives in -(ΰ)ίθ, -to. But we must be careful 
 not to confuse between the locative pronoun -9•^, and the simple ele- 
 ment '9•, which is used as a verb-root in Greek, Latin, and German, to 
 signify action ; nor must we consider the locative i, which is synony- 
 mous with O't, as identical with the verb-root e, which is used in a 
 similar manner in opposition to d-, Ό"-. When we compare per-eo 
 with per-do, and this last with τΐερ-Ο'ω, and when we recollect the 
 corresponding analogies, such as inter-eo contrasted with inter-ficio, 
 ven-eo with ven-dico, venum-do and verirdo, &c., we must be led to con- 
 clude that all these verbs are compounds in which eo, "I go," is opposed 
 to do, "I put," &c., and that the latter are entirely analogous to the 
 forms in -so, -sivi, which we first explained in another work ( Varron. 
 p. 352). In the indicative present of the verbs in -so, such as cap-es- 
 so=capere sino, arGes-so=^accedere 5wo,&c.,the form sino has lost the 
 inserted nasal which distinguishes this tense from the perfect, just as 
 the word pons^ discussed above (§ 295), refers us to the original form 
 of^d-wo=po5-fio, in which the formative η has not yet made its appear- 
 ance. The verbs in -do keep only the shorter form of the inflexions, 
 which has been extended in the separate verb do, das, dat, but the 
 perfect remains to vindicate the relationship. And the verbs com- 
 pounded with the separable -eo are distinguished from the neuter 
 verbs mentioned siboYe(pend-eo, pat-eo, sed-eo, &c.), both by the conju- 
 gation of the present, and by the form of the perfect, which correspond 
 to eo, and not to the ordinary verbs of the so-called second conjugation. 
 
 R R 2 
 
G12 THE TENSES. [bOOK IV. 
 
 But if per-do is truly represented by Λερ-θω, we must recognise a 
 verbal compound in the latter, and the same reasoning will apply to 
 the other cases in which the element Ό"- appears in its simple form, 
 such for instance as πελά-Ο^ω^ φλεγε-Ο'ω, νεμε-%'ω, ^αλέ-Ο'ω, φαε-χΙ'ω, 
 wd'LVV-&co,&c. Besides these we have, but always in the past tense, such 
 forms as εδίώκα&ον from dtcoxQ,&c., with regard to which the question 
 has been raised^ whether they are aorists or imperfects (Elmsley, ad 
 Eurip. Med. 186; Pierson, Moeris, p. 118; Ruhnken, Timceus, p. 87; 
 Hermann, ad Soph. (Ed. Col. 1619): we are quite sure, as well from 
 the meaning as from the analogy of the presents in -0"a), that they are 
 all imperfects of lost verbs in -^ω. Now it must be obvious on the 
 slightest 'consideration that causative verbs in -^ω can have nothing to 
 do with passive forms in -^ην. And it is also obvious that any true 
 explanationoftrt^gpO'i^i' must also be applicable to ετντΰην. We conceive 
 that we have correctly analyzed these aorists, when we identify the 
 interpolated elements with the synonymous locatives 0"t and ι respec- 
 tively. According to this, ε-τνφ-Ο-η { ^ J =ετν7ί-η { j^^ } will mean " there 
 was a single act of beating performed close at hand by me," so that 
 these tenses expressed the limitation of the agency by a formative 
 insertion instead of an inflexion of the person- ending. It is scarcely 
 necessary to add, that, while the (J-, κ, λ, of the future, aorist, and 
 perfect active, affect the verb-root itself with an expression of future 
 and approximate time, the interpolated -^-η, -η of the passive aorist 
 confine the instrumental ease of the person-ending to a home-circle of 
 limited agency, so that instead of the mere locality which is expressed 
 by the forms in -μαί^ -μ^ν^ we have an expression of locality added to 
 that of instrumentality signified by the case of the person-ending. In 
 fact, -η-μι or -^ψμι becomes equal to -μγιν. 
 
 383 It is of course possible that the pronominal element Ό"- may 
 be connected with the verbal root %^ (§ 224); but to imagine, with 
 Bopp and Pott, that any tense οίτί^ημί — ε^ην for instance — is added 
 to the root or crude-form of the verb, would be to reverse the natural 
 processes of language. The idea of location is intimately connected 
 with that of the verbal root ^£-, and we can easily conceive that the 
 locative O't has extended its influence to this verb as well as to the 
 aorist in question, and to the passive infinitive. It has been men- 
 tioned before that τί-%ψμι=τι-^εα-μι (§ 347), and ^η=^εε or ^la 
 (§ 116). Consequently, the root oi τίΟ'ημί is rather the locative ^i- 
 than the simple element 0'; and while we have δο=%•ο in δίδωμί= 
 daddmi, we must recognise %La=dha in τίξ^ημι —dadhami. We also 
 discern the simple element in μόχ-^ος, "a labour," "something great 
 to do;" compare the first ByYidihlQ ^ii}i μόγ-ΐζφογ-εω^μεγ-αξ','ηιαρηηβ, 
 
CHAP. Π.] THE TENSES. 613 
 
 μαί (Hesych.), made, Sanscrit mahat, mahita, "worship," &c. Also in 
 6ρμα-%'6ς, "a thing done in vows" (ορμοί). But we have the longer 
 form with a kind of passive sense in Βύ-^ηξ from εν-ννμι for ^εΰ-ννμι, 
 (Sanscrit vas, Latin ves-tis), and in εΰ-ΟΊω (root M-), where we find 
 the locative -O't, as in the first aorist passive. That this last has for its 
 future ϊδομαι, clearly of a middle form, would of itself be some proof 
 that the word is not altogether of an active nature. The following con- 
 siderations add much weight to this view. We find that the cognate 
 word 7ίί-νω has only the future πίομαυ. These two futures are, as we 
 shall show in the following chapter, nothing but subjunctives, like 
 Σείομαι for ^εωμαι-, βείομαυ or βεομοα, &c. We are convinced that 
 7ΐί-νω, as well as εΰ-Ο^ί-ω, is virtually a deponent verb; in other words, 
 that although the action may pass on to the object eaten or drunk, 
 yet, the agent being considered as the object benefited, he is spoken 
 of in the locative and not in the instrumental case. For this reason we 
 find that all verbs in Greek and Latin, which, though they may ex- 
 press an action, confine the benefits or results of that action to the 
 agent, are middle or deponent in form, and, for the same reason, the 
 Greeks use the middle voice to express that a person is not the 
 instrument, but the cause, of an action. This explains the middle or 
 deponent use of vescor, γεύομαι (as opposed to γενώ), utor, fruor, 
 χρώμαι, nanciscor, δέχομαι, adipiscor, sortior, &c. ; also of verbs 
 denoting the exercise of the senses, as αΐΰΰ'άνομαί^οοηίβηιρίον,^εώμαι, 
 όΰφραίνομαι; conspicari, intueri, άκροάΰΟ'αι, &c.; of words implying 
 mental emotions, as mirari, vereri, Icetari, &c. 
 
 384 In Greek we find certain words of tliis class with the present 
 tense of an active, but the future of a deponent_form; thus ακούω 
 makes άκούόομαι; •ϋ'ανμάξω,^'ανμάΰομαί^ ^'νήόκω,^'ανονμαι; τίάύ- 
 χω, Λείΰομαΐ', &c. It is easy to understand this: when we speak of 
 something that will make an impression upon our senses or feelings, 
 or, in general, befall us, as future, we consider ourselves as merely 
 the objects of these outward impressions or accidents; but when we 
 speak of their present efi'ect, we consider ourselves as agents or in- 
 choatives in respect of them. If any one says, "I am hearing," he 
 asserts that he is exercising that sense; if he says, "I shall hear," he 
 says merely that there will be a sound or noise: again, if he says, "I 
 am dying," he speaks of his being on the way towards death, and 
 therefore, an inchoative verb like ^'νήόκω might very well be used : if 
 he says, "I shall die," he merely states that his death λυΙΙΙ take place, 
 that he will be dead (ϋ'ανεΐται), in which event he cannot be considered 
 as an actor at all. The same principle has extended itself even to 
 the primitive verb of existence: for while εΙμί=εΰ-μί has lost its 
 
614 THE TENSES. [bOOK IV. 
 
 original future εό-ΰο-μί, which is now represented only by the sub- 
 junctive «=f^-ta5=£(?-to-^t, we find in constant use εΰΰομοαοΐεΰομαί, 
 which means, "there is or will be existence for me." Some such 
 method might be adopted to explain all those numerous instances in 
 Greek, where we find an active present with a deponent future {Gr. 
 Gr. 344). M. Burnouf justly remarks (Z. I. § 204) that the active 
 verbs with middle futures, are precisely those which, in the French 
 language, are reflected in form but not in sense; thus, ΰίγήόομαί or 
 όίωτίήΰωμαυ is je me tairai; βγιόομαι,^β m^en irai; ^ανμάβομαι^ je 
 ni'etonnerai; άμαρτήβομαί, je me tromperai; οίμώ^ομαυ, je me lamen- 
 terai; ΰτίονδάύομαί, je m'etudierai a; &c. 
 
 385 It is worthy of observation that the passive futures formed 
 from the aorists in -i^r and --O'^i;, have person-endings of the middle form. 
 It is an irregularity that they should be formed from the aorists at 
 all, and we can only explain it on the supposition that they were first 
 constructed when the future middle, as it is called, which is often 
 used in a passive sense (Monk onEurip.Hippolyt.labS, above, § 379), 
 was appropriated to the active verb, and an independent passive future 
 was necessary [Journal of Education^ iv. p. 158); at all events, they 
 must be considered as subsequent to the other forms of the future. 
 
 386 Another instance of the formation of a tense by the addi- 
 tion of a suffix used to form a set of verbs, is furnished by the itera- 
 tive tenses in -βκον. This ending is affixed to the imperfect and to 
 both aorists of the indicative, and the augment is omitted; thus ετυτί- 
 r Of makes τνΛτε^κον; ετντΙ^α^ τν-ψαόκον; and ελίΛον, λίΛεόκον, and 
 so also in the passive. This mode of forming tenses is peculiar to 
 the lonians, who were in the habit of omitting the augment of the 
 historical tenses in their descriptive poetry, and from them the custom 
 was introduced into the long narrative speeches which the epic ele- 
 ment of the Athenian drama permitted. These tenses, according to 
 Buttmann (Ausfuhrl. Sprl. § 94, Anm. 3), denote not a continued, 
 but a repeated action. The iterative formed from the imperfect im- 
 plies occasionally an action of some duration frequently repeated — as 
 in Herod, m. 119: η γννη κλαίεβκε και όάυρέσκετο— at other times a 
 momentary action repeated, Herod. 1. 185 : επιτείνεΰκε,οκως μεν ήμερη 
 γένοιτο^ ξνλα τετράγωνα — τας δε νύκτας τα ξύλα ταντα άπαίρέεβ- 
 κον : the iterative formed from the aorist conveys the latter meaning 
 only; Herod, iv. 130: οκως — καταλίτίοιεν^ αντοί αν ντΰεξήλαυνον, 
 οί δε αν Περΰαυ εΛελ%'όντεζ λάβεΰκον τα τίρόβατα. The aorist itera- 
 tives occur very seldom, and those from the first aorist are never found 
 in prose. There is one instance in Herodotus (iii. 17) where the word 
 
CHAP. Π.] THE TENSES. 615 
 
 α^ίδ^ΰκε is used as an imperfect, but, as Buttmann remarks, though 
 continuation in time is implied, there is still a signification of repetition 
 in space in this passage. It is also true, though Buttmann has not 
 remarked this, that the imperfect is used in Attic prose as an itera- 
 tive, in connexion with OTtots and the optative: as, for instance, Xen. 
 Anab. iv. 5, § 25 : έδει οτΐότε rig δν^ρφη, where certainly a repetition 
 is implied (Gr. Gr. 580). But the fact is, that the ideas of repetition 
 and continuance are intimately connected, the former being related 
 to the latter as the idea of a series of points is to that of a line ; and 
 therefore as the generating or suggesting idea is to the idea suggested 
 or generated. We have shown elsewhere that the Hebrew tense which 
 indicates continuous time, is often used as a future (Maskil le Sopher, 
 p. 28), and it is well known that esco appears as the future of sum. 
 No one needs to be told that the future is by its nature inchoative, 
 and that continued actions involve a series of recommencements. 
 
 387 There is, as well in the Greek as in the Latin language, a 
 numerous class of verbs ending in -ΰκω. Buttmann is inclined to 
 consider these terminations as totally different from the iterative pre- 
 terites of which we have been speaking (Ausfuhrl. Sprl. § 94, 4, note) ; 
 but there seems to be a sufficient similarity of meaning in the two 
 cases, to justify us in the belief that both formations owe their origin 
 to the same principle, although the original meaning of the ending 
 seems in the case of the verbs in -ΰκω to have been split up into a 
 number of subordinate significations. The sense of the Latin verbs 
 in -SCO is generally inchoative: cresco,gli-sco,quie-sco^na-scoryno-sco, 
 di-scOj sci-sco, ira-scor, paci-scor^ puera-sco, tenera-scoj illuce-sco, 
 grande-sco, mature-sco, expergi-sco, contice-sco, &c. It is to be re- 
 marked, that not one of these keeps the -5Co-form in the perfect; 
 which is cre-vi, quie-vi, contic-ui, &c. Pott has truly remarked {Etym. 
 Forsch. I. p. 56), that many Greek verbs in -ΰκω^ and more than 
 people generally suppose, are genuine inchoatives : he instances ηβά- 
 όκω,γηρά-ΰκω^γενειά-ΰκω, κνΐ-ΰκω and κνΐ-ΰκομαί, ^νψβκω (^olic 
 Ο'ναί-ΰκώ), "to be taken in death," μι,-μνή-(5κομαι.(ΜοΙιομί-μναί-ΰκο- 
 μαι)^ γυ-γνώ-6κω. In others this meaning is less clearly seen, as 
 άλΟ'ί-ΰκω^ "I make sound," ίλά-ΰκομαί, "I make myself gracious," 
 γαννν-ΰκομαί, "I become merry," άλί-ΰκομαί, "I fall into the enemy's 
 hands." The following have a causative signification, με^ν-ΰκω, 
 Λίπί-ΰκω, γαμί'ΰκω and γαμί-ξω, Λίνν-ΰκω and ττιννόΰω, δίδά- 
 ΰκω, εταβα-ΰκεμεν, Λίφαν-ΰκω and Λίφά-ΰκομαί, άναβιώ-ΰκομαί, 
 μιμνή-ΰκω, άρε-6κω, &c. Now it is sufficiently obvious that the only 
 difference between an inchoative and a causative is this, that the one 
 is a passive, the other an active relation. But the idea of iteration or 
 
QIQ THE TENSES. [bOOK IV. 
 
 repetition presupposes the idea of a beginning, and the very disti]|B- 
 tion between an imperfect and an iterative preterite is, that the former 
 presumes a line, the latter a sequence of points, the former unbroken 
 continuation, the latter a series of recommencements. 
 
 We find an analogous ending, with a similar meaning, in nouns 
 derived from verbs; thus we have δίΰχος from dixsiVj λεΰχη from 
 λέγειν, αϊΰχος from αΐδονμαι: in the last two instances the κ is 
 aspirated on the compensation principle, as in τίά-ΰχω, root ηεν^- ; 
 for the root of the former is λεγ- or λε/, as we shall show hereafter, 
 of the latter Ραιδ-, Lithuanian geda : the first seems to stand for δίκ- 
 0Kog, or the κ has been dropt, as in λά-6κω {λακείν), disco (doc-ere), 
 μί'βγω, mi-sceo (μι,γηναί), Ϊ6χω (εχω), (above, § 219). 
 
 The pronominal roots sa, Jca, are ultimately the same ; as tense- 
 endings we have established their identity. If, then, our view of the 
 termination -<?κ is correct, this is a junction of two forms of the same 
 element, just like the very common pronominal endings -n-t, -ta-na, 
 &c.; and the principles already explained will enable us to under- 
 stand how a compound signifying successive proximity may be used 
 to express the ideas of inchoation, continuance, repeated action, 
 growth and causation. As the locative O't may become a verb root, 
 and denote the act of placing, so 6κ- is found as the element of a set 
 of words which denote progression by successive steps. Such are 
 0κάλλω, ΰκάξω^ ΰκέλοζ, scando, scala^ &c. (See Kenriek, Herodotus^ 
 p. 24.) 
 
CHAPTER III. 
 THE MOODS AND PARTICIPLES, 
 
 388 Distinctions of mood are rather syntactical than etymological. 389 (1) Sub- 
 junctive and optative. Forms of the subjunctive. 390 Explanation of the 
 optative i. 391 Subjunctive and optative related as future and aorist. 392 
 Different forms of the future. 393 The relations of tense between the sub- 
 junctive and optative are indicated by their syntactical use. 394 Especially by 
 the construction of ov μη. 395 Correspondences of the future and subjunctive 
 in hypothetical sentences. 396 Their appearance in the same final sentence. 
 397 Use of the optative to express a wish. 398 The subjunctive and future 
 interchanged in Latin. 399 Sanscrit forms. 400 Subjunctive and optative 
 appear as differences of tense in modern languages. 401 (2) Imperative and 
 infinitive. The former distinguished by its person-endings only. 402 Active 
 person-endings of the Greek imperative. 403 Passive person-endings. 404 
 Sanscrit imperative. 405 Greek infinitive. 406 Latin infinitives. 407 San- 
 scrit infinitive. 408 Connexion between the imperative and infinitive. 409 
 This is especially manifested in the false analogy which led to the passive 
 forms of the imperative. 410 The three different forms of the infinitive active 
 are the locatives of three verbals. 411 The Greek infinitive passive. 412 Latin 
 and Sanscrit forms. 413 (3) Participles. Forms in -nts. 414 Perfect parti- 
 ciples. 415 Origin of the present participle. 416 Infinitive and participle, 
 how related. 417 Adjectives and participles. 418 Teutonic infinitive — a de- 
 clinable participle. 419 Latin gerunds and supines. 420 Participles used 
 absolutely in Greek, Latin, and Sanscrit. 421 Paradigms. 422 L Greek verb, 
 τνητω. 423 II. Sanscrit verb, tudumi. 424 III. Latin verb, tundo. 
 
 388 rpHE distinctions of mood and tense are due rather to 
 the methodical language of syntax than to any es- 
 sential varieties in the forms themselves. The imperative mood 
 is merely the indicative with a peculiar affection of the person- 
 endings, and the infinitive is merely a fixed or adverbial form 
 of the participle, which again is an adjective derived from the 
 third person plural of the present indicative, just as δημ06ίθζ 
 springs from the genitive of δημοζ. With regard to the sub- 
 junctive and optative, although it is customary in Greek Gram- 
 mars to class them as distinct moods, having tenses of their 
 own, it has long been felt by scholars, on syntactical grounds, 
 that, considered in their relation to one another and to the 
 other moods, they must be regarded as differing in tense only. 
 It has also been observed that the person-endings of the sub- 
 junctive correspond to the primary forms, or those of the primary 
 tenses, while the person-endings of the optative always agree 
 with the secondary forms, or those of the historical tenses 
 
618 THE MOODS AND PAETICIPLE8. [bOOK IV. 
 
 (Buttmann, Ausfuhrl. Sprl. § 88, 3, 4). A more exact etymo- 
 logical and syntactical examination of the whole question will 
 show clearly, that, in the oldest form of the language, these 
 moods have no right to a separate classification, and at the 
 same time enable us to point out the real connexion between 
 them. 
 
 389 (1) Subjunctive and Optative. 
 
 The form which the subjunctive generally presents diflers from 
 the indicative in the following points. Where the indicative has ω, 
 (w, 0, the subjunctive has ω\ where the indicative has ε, ει, -j^, the 
 subjunctive has η^ τ^] : the a of the first and second persons of the first 
 aorist, active and middle, become ω and η respectively. There is no 
 reason to believe that there is any more essential diff'erence between 
 the ω and η of the subjunctive, than there is between the ο and ε of 
 the indicative ; as the two latter represent the short a in Sanscrit, the 
 two former may be considered as substitutes for the long «, which is 
 the characteristic of the Let or subjunctive mood in the Veda-dialect ; 
 and in every case we must conclude that an original l- or vocalized 
 (9- is absorbed, or more or less imperfectly represented by this long 
 vowel. This a is found in the subjunctive of the Doric dialect: for 
 instance, we have Ιύαντι {Corpus Inscript. Yol. ii. p. 641, no. 3053); 
 επυΰτάντυ (lb. p. 413, no. 2556, 1. 68); iQCCtai (Pindar, Pyth. iv. 92= 
 164); ^αμεζ (Theocr. xv. 22); see Kuhn, 1. 1. p. 39. In the older 
 writers, the root-syllable alone is changed (sometimes by the addition 
 of an i), the connecting- vowel not being affected, though this seems to 
 take place only in the dual and plural; thus we have βείομεν^ ^'είομεν, 
 ότείομεν,τίαρΰτψτον,δώομεν, γνώομεν. In the passive this may take 
 place in the singular, as in ατίο^είομαι. In the active we occasionally 
 find both root-vowel and connecting-vowel affected by guna, as in 
 utYiys, Ιμβή^, ^^Ί^ζ-, εφείτβς, δώγ], &c. Buttmann supposes the exist- 
 ence of subjunctives without any mark to distinguish them from the 
 indicative (Ausfuhrl. Sprl. § 88, Anm. 3), but these could only be 
 corrupt and mutilated forms. We believe that all the instances which 
 he mentions belong to the analogy of βείομεν, the production having 
 fallen upon the first syllable. With regard to these epic forms in 
 general, it is clear, that if θ'ε/ω, ^εί^ς, ^εί^^ %'είομεν, άτίοΟ'είομαο are 
 to be considered as belonging to the same form, the short connecting- 
 vowel is due to the weight of the person-endings. It is equally clear, 
 that they are all futures, and perhaps very ancient forms of the 
 future: at least, forms answering to ^'είομεν, ϊομεν are used with a 
 future signification in Homer; βείομαι^ for instance, is a regular 
 
CHAP. III.] THE MOODS AND PARTICIPLES. 619 
 
 future. Perhaps in the original form the increment did not extend 
 to both root and connecting-vowel. There are other instances of this 
 superfluous strengthening of the form : thus on the analogy of κομιοϋ•- 
 μαι and κομίΰομαι we should expect only τίλεΓονμαυ or τίλενΰομαο 
 from τίλέ^ω: whereas we have also 7tλεvβovμaL: τιμάοιμί would be a 
 sufficient optative from τιμάω, but we have also ημφην. 
 
 390 The characteristic of the optative is t, which forms a di- 
 phthong with the connecting- vowel : to this the long vowel η is occa- 
 sionally subjoined. The person-endings are of the secondary form, 
 with the exception of the first person, which is generally -μι, though 
 it seems that the shorter form in -v was here, as in the imperfect, 
 occasionally substituted for it, in those cases where the first person of 
 the optative regularly ended like the indicative: thus we have in 
 Euripides: άφρων ανεϊην εΐ τρέφοίν τα των 7cilag{Etymol. Magnum, 
 s. V. τρέφουν). The optatives which add η to the characteristic t 
 always form the first person in -v. The reason for the difference is 
 obvious. In τρεφοιμι^ which has no augment, there is no reason why 
 the weight of the verb should drive off the full ending ; but the longer 
 form in -ιτιν of course would not admit of it. In the ^ulic dialect, 
 the characteristic l was sometimes omitted, as already included in the 
 ri. Thus we have in a passage, probably from Sappho, quoted by 
 ApoUonius {de Syntaxi, in. 22, p. 247 Bekker): 
 
 αϊ%^ εγώ^ χρυΰοΰτέφαν Άφροδίτα, 
 
 τόνδε τον τίάλον λα%6ΐ(\ν — 
 where we adopt the emendation proposed by Bergk {RheiniscJi. Mus. 
 for 1835, p. 218), which is confirmed as well by the metre as by the 
 remark in the Etymologicum Magnum, p. 558: Ααχόην εΰτο λάχοψί, 
 λάχοίξ, λάχοί. We believe that this word represents the original 
 formation of the optative; for λαχόην = λaχ6ya-μL•=[8]λaχό-6a-μί, 
 would be the regular indeterminate corresponding to the subjunctive 
 λαχόω=λαχο-ίω=^λαχόόω. Bopp has suggested (Annals of Oriental 
 Liter at. p. 23) an explanation of this characteristic i which at least 
 deserves to be mentioned. He says : " The reason why the vowel i 
 expresses the potential mood cannot be discovered in Greek, in Latin, 
 nor perhaps in any other European language, but in Sanscrit, the 
 radical element i expresses wishing, desiring; and what syllable could 
 be more properly employed to indicate an optative, than the one to 
 which the Hindu grammarians had given the primary signification of 
 Kanti — having desiret I will not affirm that this is the primitive 
 meaning of the root i, and that the grammarians had a sufficient rea- 
 son for putting Kanti at the head of their explanations, but certain it 
 is, that imas has, among other significations, that of we desire or wish. 
 
620 THE MOODS AND PARTICIPLES. [bOOK IV. 
 
 Now it is remarkable, that the sense expressed in Sanscrit, and the 
 languages here compared with it, by a syllable, signifying desire, in- 
 corporated into the verb, is in English, and often in German also, 
 expressed by detached auxiliary verbs, having the primary signification 
 of wishing. Thus, in Notker we read, I mahta baldur weinon — vellem 
 vehement er plorare. The German mo gen has frequently this signifi- 
 cation, and the English may is of the same origin, derived from the 
 Saxon mag an J in Gothic likewise magan^ We have already said, 
 that in our opinion the theory of agglutination, which Bopp has in- 
 troduced, must be received with great caution and subject to many 
 limitations. The inflexions of verbs may and do take place in the 
 same way as the modification of nouns; namely, by pronominal in- 
 sertion between the root and the person-ending in verbs, and between 
 the root and case-ending, in nouns. Of the pronominal elements 
 which may be so inserted, there is no one more common than -la^ 
 Sanscrit ya, corresponding to the second pronoun or the relative form. 
 We know that the s, which characterizes the future and aorist, may 
 degenerate into i, and we have had no difficulty in identifying this s 
 with the second element. As then we shall see that the subjunctive 
 and optative are virtually related as future and aorist, we must refer 
 their characteristic i to a pronominal insertion of the same kind, and 
 thus τ^ίφου-μι for τρεφέΰα-μϋ will be quite analogous to fui for fuesa. 
 
 391 Having now considered the forms of the subjunctive and 
 optative separately, we must endeavour to determine the connexion 
 between them. In the last chapter we mentioned that the desidera- 
 tive τνφείω was related to the so-called optative aorist τύψεία, as pre- 
 sent to imperfect, as definite tense to indefinite ; and the augment of the 
 latter is omitted like that of the iterative preterites. Now if we take 
 the verbs in -p, which present us with the oldest forms, we shall find 
 a subjunctive^fta3,^£t];g,&c. corresponding to an optative 0'£t?;i/,'9'a'iy^, 
 &c. The subjunctive has the primary form of the person- endiugs, the 
 optative, the secondary form: in other respects they do not differ, for 
 they both contain the verb-root afiOcted by the insertion of l : com- 
 pare Ο-ειω, ^eiyg, &c. ; &Είην, ^είης, &c. ; with τίΟ'ημυ (τι-θ'ε-ω), τί&ης, &c. ; 
 ετίϋ'ην, ετί^ης, &c. In the existing state, the subjunctive is just as 
 heavy a form as the optative; we must conclude, however, that, as 
 the optative has the lighter endings, it must have been originally a 
 heavier form than the subjunctive, and as this could result only from 
 its having some prefix which the subjunctive wanted, it follows that 
 it had the augment, or was related to the subjunctive as aorist to 
 future. When the optative ceased to be used as a past tense in pri- 
 mary sentences, it would naturally lose its augment, or mark of past 
 
CHAP, m.] THE MOODS AND PARTICIPLES. 621 
 
 time, retaining, however, its lighter person-endings as a trace of what 
 it once was. The other differences between it and the subjunctive 
 would spring up as time and use widened the gap which separated the 
 parent tense from its offspring. 
 
 392 On the whole, then, we may say that future time was 
 expressed by two varieties of the same pronominal insertion : in the 
 one, the second element was inserted under the form (j-, sa; in the 
 other, the same element appeared as t-, ya. These two forms were 
 occasionally united, as in the desiderative verbs in -ΰείω (Latin -no); 
 but in most cases only one of the equivalent insertions was employed; 
 thus we have the latter only in the old subjunctives or futures βείομαο, 
 βείομεν, Ο'είομεν, &c. ; and the sigmatic form only appears in the more 
 permanent futures indicative. Indeed it might be inferred from the 
 analogy of the genitives in -(>to and -to, that the so-called desiderative 
 is the primitive and genuine form, of which the future in (j-, and the 
 subjunctive in t-, are successive degenerations. To all these three 
 forms of the future there were corresponding forms of aorists or past 
 tenses; to the first, the so-called ^olic optative aorist in -όεοα; to 
 the second, the ordinary optative; to the third, the ordinary first 
 aorist. The last alone preserved the augment, because in the indica- 
 tive mood the idea of past time predominated in this form of the 
 indefinite tense; in the other two the augment was omitted, because 
 they are never used as direct expressions of past time, though they 
 always bear the preterite meaning in subordinate sentences. We do 
 not say that there ever existed a desiderative form of every tense of 
 the indicative mood to which there is a corresponding tense in the 
 optative; there might have been one, and there must have been one 
 originally; but afterwards the tenses of the optative were formed by 
 analogy, without the introduction of the intermediate form of the 
 subjunctive. The only very remarkable variety in the formation of 
 these optatives, desideratives, and futures, is, that the i is sometimes 
 appended to the root-vowel, as in Ο'είομεν, at other times to the con- 
 necting-vowel, as τντίτοίμί', and sometimes placed after the 5, as in 
 δραόείω ; at other times before the s or its substitute, as in Ο'είτβς^ 
 ετίαινιώ (Greg. Corinth, p. 229), and in some of the Sanscrit volitives 
 (Wilkins, p. 365). ^These varieties are due to subsequent analogies, 
 and not to any thing in the original principles of the language. 
 
 393 We shall now proceed to show, that the syntactical relation 
 of the optative to the subjunctive is that which subsists between in- 
 definite and definite tenses. It is well known to every student of 
 Greek, that, in connected sentences, the Latin subjunctive present 
 
622 THE MOODS AND PARTICIPLES. [bOOK IV. 
 
 corresponds to the Greek subjunctive, and the Latin subjunctive im- 
 perfect to the Greek optative; in other words, the Greeks used the 
 subjunctive in sentences dependent upon a verb in the present or 
 future tense, and the optative in those which were dependent on a 
 verb in the past tense : for instance, ^ρ«φω, ίνα μαν^'άντ^ς corresponds 
 to scribo^ ut discas; hut eyQCiipa^ ίνα μαν^άνοΐξ to scripsi, ut disceres. 
 Therefore, the subjunctive is a definite tense, for it has relation to 
 the present moment, and the optative is indefinite, for it must be de- 
 termined by some particular time or circumstance referred to. The 
 following considerations will show, that, as well in dependent as in 
 connected sentences, the subjunctive corresponds to a future, and the 
 optative to the aorist derived from it; in other words, the subjunctive 
 is a determinate tense, and signifies " the probable occurrence of some- 
 thing after the time of speaking: " and the optative is an indeterminate 
 tense, and signifies "the probable occurrence of something after the 
 time specified" (above, § 372). And first, the subjunctive appears as 
 an actual future in Homer. Thus we have in the Iliad i. 262: 
 
 ου γάρ τϋω toiovs ϊδον άνερας^ ονδε ϊδωμαυ. 
 
 Iliad VI. 459: 
 
 και τΐοτέ τις εϊτίτ^βυ^ Ιδών κατά δάκρυ χεουΰαν 
 '''^Έκτοροξι ηδε, γυνή, κ. τ, λ." 
 ωξ τΐοτε τις ερέει, 
 
 where the future, which follows, clearly shows that the subjunctive 
 before it is a future. Iliad "vn. 197: 
 
 ου γάρ τίζ με βίγ] γε εκών άεκοντα δίηται. 
 
 Odyss. VI. 201 : 
 
 ουκ sW omog άνηρ διερος βροτός, ούδε γένηταί, 
 og κεν Φαιάκων ανδρών ες γαΐαν ΐκψαι, 
 
 on which see Wyttenbach {Eclog. Histor, p. 343), quoted by Gaisford 
 (in his notes on Herodotus, Yol. i. p. 5). Odyss. xvi. 437: 
 
 ουκ εσθ•' o^rog ανήρ, ουδ' εβόεται, ούδε γενψαι, 
 οζ κεν Τηλεμάχω 6ω υίεϊ χείρας ετΐοίΰει. 
 
 It will be observed that we have og κεν with the future in this passage, 
 but og κεν with the subjunctive in that which precedes. 
 
 394 Instead of this direct future with ου, the Attics employed ου 
 μη with the aorist subjunctive; thus we have in Plato, Respubl. vi. 
 p. 492 Ε : οϋτεγάρ γίγνεταυ,οϋτε γεγονεν, ουδέ ουν μη γενηταν,κ.τ.λ. 
 (cf. Phcedr. p. 350 g), where οϋδε μη γενηται is perfectly synonymous 
 with the ουδέ γενηται of the two passages from the Odyssee. The 
 
CHAP. ΠΙ.] THE MOODS AND PARTICIPLES. 623 
 
 combination of the negatives ov μη is found not only before the aorist 
 subjunctive with a negative future sense, but also before the future 
 indicative, generally in a prohibitive sense. In the latter case, Elmsley 
 and others would take the sentence interrogatively, according to the 
 method adopted with regard to the positive use of ονκονν. Thus, e. g. 
 Eurip. Bacchce, 340: 
 
 ov μη TCQOoOLoSLg χείρα, βακχεύΰείς δ' Ιών, 
 μηδ' εξομόρ^ευ μωρίαν την όην εμοί; 
 
 should mean, according to these critics, "will you not keep oifyour 
 hand (will you not — not put your hand near me), and go and play the 
 Bacchanal, and not wipe off your folly on me?" But ού μη with the 
 subjunctive is explained as an ellipse — ov μη γενηται being equivalent 
 to ov δεοξ εότϊ μη γενηται, "there is no fear lest it should happen." 
 Neither .the one nor the other of these explanations is perfectly ac- 
 curate. With regard to ov μη with the future, it is not true that this 
 combination always implies a prohibition. There are passages in which 
 it is perfectly equivalent in meaning to ov μη with the subjunctive ; 
 as in Sophocles, CEd. Col. 176: 
 
 οντου μη ποτέ ύ' εκ τώνδ' εδράνων, 
 ώ γέρον, ακοντά τις αξει. 
 
 Electra, 1052: 
 
 αλλ' εϊΰίδ' ' ον ύοι μη με^ετ^ομαί ποτέ, 
 ονδ' ην ΰφόδρ ίμείρονύα τνγχάν^ζ. 
 Euripides, Phoeniss. 1606: 
 
 ΰαφώς γαρ είπε Τειρεόίας, ον μη ποτέ, 
 ΰον τήνδε γην οίκονντος, εν 7ίρά^,ειν πόλιν. 
 Aristoph. Ran. 508: 
 
 μα τον Άπόλλω, ον μη (?' εγώ 
 περιό^ομαί απελθόντα. 
 
 It is obvious that an assertion, not a prohibition, is implied in these 
 passages. The proper explanation of those from Sophocles and Aristo- 
 phanes has been suggested by Hermann (on Elmsley's Medea, v. 1120). 
 He says that the general meaning of ov μη with the future is due 
 entirely to the circumstance, that in the cases where a prohibition is 
 implied, the verb is always in the second person; — and "will you not 
 not touch," &c., is equivalent to "don't touch." But in the cases which 
 he is discussing, the verb is in the first or third person; and the infini- 
 tive, in the passage which we have quoted from Euripides, implies a 
 third person. Now when we say, as in the passage from the (Edipus 
 Coloneus, " will not a person not drag you from this seat against your 
 
624 THE MOODS AND PARTICIPLES. [bOOK IV. 
 
 will?" this is equivalent to "will he not suffer you to remain?" which 
 implies "of course he will;" and so in the other passages. With 
 regard to ov μη with the subjunctive, Hermann thinks, that, although 
 an ellipse such as we have mentioned is possible — for we have the full 
 form in Herodotus, e. g. in i. 84: ov γαρ ην δεινον μη άλω Λοτε— yet 
 this ellipse would be somewhat harsh and unnatural, and it would be 
 much better to follow the analogy of the future, and suppose that ov 
 μη with the subjunctive is also interrogative: thus ^schylus, Sept. c. 
 Thebas, 38 : 
 
 κάί τώνδ^ άκονΰας σν τι μη ληφ&ώ δόλφ ; 
 would mean, "Having heard of these things, shall I not not-be-caught 
 by stratagem?" i. e. "shall I not be safe from it?" which implies, "of 
 course I shall." We think this view a sound one ; the following illus- 
 trations will perhaps convince our readers that it is so. There can be 
 no doubt that ov, with the future or subjunctive taken interro'gatively, 
 is equivalent to an injunction or exhortation. The future is generally 
 in the second person: thus, ov μένεις; means "stop!" the subjunctive 
 in the first, as ονκ ϊω ; " shall I not go ? " We have a good analogy for 
 this in the use of quin by the best Latin writers. This particle, which 
 is equivalent to cur non, is constantly used with the indicative present, 
 taken interrogatively, but always implying an exhortation; thus we 
 have, Plautus, Mencechm. ii. 1, 22: Quin nos Jiinc domum redimus'? 
 Terence, Andr. iv. 4, 15: Quin dicis unde est claref Livy, i. 57: 
 Quin, si vigor juventce inest, conscendimus equosf where see Draken- 
 borch's note: so that Bentley is quite, right in reading quin redimusf 
 instead of quin redeamus'? in Ter. Eunuch, iv. 7, 41. It is also clear 
 that μη, with the subjunctive or future, is the expression of a direct 
 prohibition. We need not give any instance to show that μη τν-φηξ 
 differs from μη τντΐτε only in being particular instead of general. The 
 imperative use o£ μη with the future has been denied by Elmsley, who 
 would substitute the subjunctive for the future in Euripides, Med. 804 : 
 λεξείζ δε μηδέν των εμόί δεδογμένων, and would either emend or explain 
 away a number of other passages which he quotes in his note upon that 
 line, but which are, we think, sufficient to justify the construction. 
 Matthise (Gr. Gr. § 511, 3) quotes two or three others, and we may 
 add Soph. Aj. 572: και τάμα τεύχη μήτ' άγωνάρχαυ τίνες ^ήΰονβ^ 
 Άχαιοΐς, μη^'' δ λνμεών εμοί (above, ρ. 519). As, therefore, ov with 
 both future and subjunctive, taken interrogatively, may convey a posi- 
 tive injunction, and μη with either of the same inflexions, taken im- 
 peratively, may convey a negative command, it would not be un- 
 natural that, when command with regard to one act and prohibition 
 with regard to another were to be expressed at once, the first would 
 be 'effected by ov with the future or subjunctive, taken interrogatively, 
 
CHAP. ΠΙ.] THE MOODS AND PAETICIPLES. 625 
 
 the second by μή with the future or subjunctive, without any interro- 
 gation. We have an instance of this in JEschyl. Sept. c. Theb. 232: 
 
 ov ΰΐγα-, μηδέν τώνδ' sqbIs κατά τιτόλιν. 
 
 In most cases, however, the two sentences, which generally seem to 
 have referred to a command of something and the prohibition of its 
 opposite, would be joined together by some copulative conjunction, 
 and thus the whole would be included in the interrogation, as in the 
 passage from the BaccJice quoted above; in Soph. Aj. 76: 
 
 ov ΰΐγ' άνίξ,ει μηδέ δειλίαν ά^εΐξ, 
 
 and in Eurip. Hippolyt. 498 : 
 
 ώ δείνα λέξαό\ ονχΐ όνγκλείόείς ΰτόμα, 
 τίαι μη με^ήΰεΐξ avd'ig αΐΰχίΰτονς λόγους ; 
 
 From this custom of joining together an injunction of some thing 
 and a prohibition of its contrary, would arise the custom of employing 
 a combination of the two negatives to express in the strongest terms an 
 union of the two imperatives ; and this combination would always be 
 used, by implication, interrogatively, and with two shades of meaning. 
 As the future or the longer form was more used in connexion with the 
 direct negative ov and in the second person, this tense would be more 
 generally employed by the Attics to express a prohibition in the 
 second person by means of ov μη taken interrogatively: and as the 
 subjunctive, a shorter form of the future, was more frequently sub- 
 joined to the indirect or subjective negation μη, to express a direct 
 prohibition, it would be more usually employed, in connexion with ov 
 μή and in an interrogative sense, to express the direct negation of 
 something future, in the sense in which ov was used with the sub- 
 junctive by Homer. That in the collocation ov μή, whether with the 
 future or with the subjunctive, the notion of the verb is negatived 
 and reversed by μή, and the question expressed by ov, is clear from 
 the nature of the case. For as ov μένεις is equivalent to μένε, and 
 ονκ ϊω to μένω ; so ov μη μένεις is equivalent to ονκ εΐ and oi) μη 
 μείνω to ονκ Ιω; This is also shown by the fact, that, if by any 
 chance the combination ov μή is separated from the verb, the μή is 
 repeated immediately before the verb to which it belongs; thus we 
 find in Soph. (Ed. Tyr. 328: 
 
 εγω d' ov μή Λοτε, 
 ταμ ώς αν εϊτΐω, μή τα d' έκφήνω κακά. 
 
 In this syntactical peculiarity of the Greek language we see clear 
 enough traces of the original identity of the future and subjunctive, 
 in an actual and strongly marked divergency of use, 
 
 SS 
 
626 THE MOODS AND PARTICIPLES. [bOOK IV. 
 
 395 The employment of the subjunctive in dependent or con- 
 nected sentences, and its contrast here to the optative, on the one 
 hand, and to the past tenses of the indicative on the other, will also 
 show very directly its affinity to the future (see Gr, Gr. Art. 502, &c. 
 607 (c), 614). It almost invariably follows lav or any relative word 
 succeeded by av in the protasis, in which case it is equivalent to a con- 
 ditional future, and is generally followed in the apodosis by the future 
 indicative, as lav τυ εχ^£, δώοείς. There is scarcely one undoubted 
 instance in Attic Greek of the use of av with the future indicative. 
 In those which are cited by the grammarians, the most eminent 
 critics have either omitted the av, or changed the future into the 
 optative. In Aristoph. Nub. 466, where the critics and the MSS. are 
 equally divided between αρ and av — ο-φομαί, we ought to prefer the 
 former particle, as in Eurip. Bacch. 639: τί τίοτ' αρ εκ τούτων Ιρεΐ; 
 and in -^schin. c. Ctes. 543, the true reading is άνερεΐ. In Xenophon, 
 Cyrop. VII. 5, § 21, we have όταν δε αϊΰ^ωνταυ ημάς ένδον οντάς, 
 τίολν αν ετι μάλλον η ννν αχρείοι εύονται. Dindorf omits the αν, 
 though there is no variation in the MSS. A very similar passage is 
 found in Dinarchus {in Demosth. § 111): πολύ γαρ αν δίκαώτερον 
 Ιλεήύετε την χώραν, where Bekker conjectures Ιλεήΰαίτε, though Her- 
 mann does not think that the emendation is necessary (Opusc. iv. 
 p. 33). For similar corruptions and their obvious remedy, see the 
 latest editors on the following passages: Thucyd. i. 150; Plat. Phcedo, 
 p. 61 c; Crito, p. 53 c; Besp. p. 615 d; Eurip. Andr. 464; Xen. Cyr. 
 IV. 5, § 49. It is nearly certain that av cannot be used with the 
 future infinitive. In Thucydides, at all events, this solecism is 
 easily removed from the few passages in which it deforms the text 
 (see the Preface to our recension, p. xi). It is worthy of remark that 
 the construction with ov μη and the subjunctive is considered quite 
 equivalent to the future in the apodosis: thus we have in Thucyd. iv. 
 95: Tjv νικήΰωμεν, ov μητίοτε Ιΰβάλωΰιν, and in Sophocles, ElectrAS: 
 ov γάρ ΰε μη γήρ^ τε καΐ χρόνω μακρω γνώύ' ούδ^ ντΐοτΐτενΰονΰίν. 
 The subjunctive also follows ει or a relative word without av, but then 
 there is a difference of meaning: thus, lav τί εχϊ]ς, δώΰευς means "if 
 you shall happen to have any thing (which will probably be the case) 
 you will give it ;" but εϊ τί εχηξ would not have implied any proba- 
 bility, — "if you shall happen to have any thing (which is a mere con- 
 tingency)" (see Philol. Mus. i. pp. 96 foil.). If, however, we compare 
 either of these cases with the optative similarly used, we shall easily 
 perceive, that, while the subjunctive in the hypothesis implies only 
 one relation — that is, a relation to the time of speaking — and there- 
 fore stands on the same footing with the definite tenses, the optative 
 presumes a relation to some time or circumstance which it is neces- 
 
CHAP, m.] THE MOODS AND PAETICIPLES. 627 
 
 sary to define. Thus, εΪ ti εχοψί, δώοίψ civ means "if I had any- 
 thing under certain circumstances (i. e. as often as I had it), I would 
 give it," where the verbs are clearly in the indefinite tense, or express 
 a double relation — of past time in regard to the time of speaking, 
 and of posteriority in reference to the time or circumstance spoken of. 
 Perhaps the most direct proof of this is the usage of the subjunctive 
 or future, as contrasted with that of the optative in the apodosis. 
 Both the future and subjunctive were constantly used with KBV=av in 
 the apodosis of conditions in the older state of the language, and we 
 have just shown that even in the Attic writers there may have been 
 some traces of this usage of av with the future. But then the protasis 
 is always expressed, for the subjunctive and future being definite 
 tenses, and implying only a relation to the time of speaking, would 
 not require av when used independently, because av refers at once to 
 some other circumstance, which other circumstance, namely, the time 
 of speaking, is presumed in the future and subjunctive: when the 
 condition was expressed, the av might accompany the apodosis, though 
 even then it would be unnecessary, and would consequently be omit- 
 ted when the syntax of the language gained its full development. 
 We find the same correspondence between the future and subjunctive 
 in certain forms of the temporal sentence (Gr. Gr. Arts. 580, 582, 
 583, (β)). 
 
 396 The relation between the subjunctive and optative is farther 
 shown by their occasional appearance in the same final sentence, to 
 express a succession of consequences. This usage has been very well 
 explained by Dr. Arnold (Thucyd. iii. 22, p. 446). The following are 
 instances; Herod, ix. 51; sg rovrov δη τον χώρον εβουλενΰαντο 
 μεταναΰτηναί, ίνα καΐ νδατι εχωΰι χρηΰ^'αν άφ%'όνω^ καΐ οί ίτΐτΐεες 
 <5φεας μη όυνοίατο. Thucyd. νπ. 17 : νανξ ετζλήρονν, όπως ναυμαχίας 
 τε αΛοπείράΰωύί, καϊ η<5<3ον οί ^Α&ηναϊου κωλύοιεν άτναίρευν. Eurip. 
 Hecuba, 1120: 
 
 εδειΰά, μη 6οι τίολψιοζ λεί-φ-θ'εΐ^ δ Λαϊζ 
 Τροίαν α%ροί6η καϊ ξννοίκίύ^ τίάλιν' 
 γνόντες ά' 'Αχαιοί ξώντα Πριαμιδών τίνα 
 Φρνγών ες αίαν avd^ig αϊροιειν ΰτόλον, 
 κατίειτα Θρηκης τιεδία τρίβοιεν τάδε 
 λεηλατονντες' γείτοΰιν δ' εϊη κακόν 
 Τρώων, εν ωτίερ νυν, αναξ, εκάμνομεν. 
 "It seems to me," says Dr. Arnold, "that in all these cases the trans- 
 ition from the subjunctive to the optative mood is meant to show 
 that the several consequences are not contemporaneous, but that the 
 subjunctive mood indicates the immediate, and the optative the 
 
 SS2 
 
628 THE MOODS AND PARTICIPLES. [bOOK IV. 
 
 remote consequence of the action contained in the principal verb, the 
 second being a consequence upon the first: and that to mark this 
 gradation, different moods are employed, and the subjunctive is thus 
 used even when the principal verb is in the past tense, because other- 
 wise the distinction intended could not be marked." 
 
 397 "With regard to the separate use of the optative without aV, 
 that is, as a proper optative expressive of a wish, it need only be 
 remarked, that the entire dependence of the verb expressive of the 
 wish upon some circumstance or event is obvious, not only from the 
 fact that the past tense of the auxiliary is used in modern languages, 
 but also from the employment of the limiting particle θ-ε, "in this 
 particular," in connexion with d and the optative, and from the use 
 of the past tense ωφελε for the same purpose. It is remarkable that 
 the optative proper is accompanied not only by the conditional par- 
 ticle, but also by τΐώς aV, as the apodosis of a condition implied. 
 This shows how little reason there is to suppose with Bopp that the 
 optative intrinsically and primarily expresses a wish. It only does so 
 as an indefinite and dependent tense, having reference to some other 
 time or circumstance than the present. In our own language, "if I 
 only could manage to bring it about ! " and "how could I manage to 
 bring it about?" are expressions of the same wish. Inattention to 
 this latter usage has prevented all the commentators* from seeing the 
 force of a very natural passage in iEschylus (Agamemn. 1198). Cas- 
 sandra says wildly to the Chorus: εκμαρτνρηΰον τΐρονμόύας τό μ 
 είδεναι, λόγω-τίαλαιαξ των δ' αμαρτίας δόμων' — "give me a direct 
 testimony confirmed by oath that I am acquainted with the old 
 traditionary sins of this family" (see above, § 311); to which the 
 Coryphaeus replies : 
 
 καΐ τίώξ αν όρκος, Ληγμα γενναίως τίαγεν-, 
 τίαιώνιον γένοιτο; 
 
 "and would to God that an oath, strong as I could make it, might 
 serve as a remedy for them!" The words which follow show that we 
 have here rather an admission than a question on the part of the 
 Chorus; and the emphasis, implied in the cumulative τιηγμα γενναίως 
 
 * Hermann, in his posthumous edition of ^schylus, adopts our inter- 
 pretation of the passage. His note is: "inepte addidisset -Sischylus γεν- 
 ναίως Ttayiv, si quaereret chorus, quid prodesse jusjurandum posset. Hoc 
 potius dicit, atque utinam jusjurandum^ firmamentum generose firmatum, mede- 
 lam afferre possit! quo indicat, quamvis sanctissimum jusjurandum tamen 
 nihil profuturum esse." We have written farther in support of our view 
 in the Journal of Philology, iii. pp. 210 — 215. 
 
CHAP. III.] THE MOODS AND PARTICIPLES. 629 
 
 TCayiv, suggests tlie objection that no assent to prophecy, however 
 earnest, will serve as a cure for the evils predicted. This is in fact 
 the foundation of the idiom in question: for when we say, "how 
 could it be brought about?" we are seeking for some consummation 
 devoutly to be wished. 
 
 398 This connexion of the subjunctive and optative appears also 
 from an examination of other languages, as well those which form 
 their moods and tenses by inflexion, as those which use auxiliaries. 
 
 We cannot have a better proof of the correspondences in meaning 
 between the future and aorist «i the one hand, and the subjunctive 
 and optative on the other, than that which is furnished by the tense- 
 system of the Latin verb. It is true that we are obliged to call in a 
 sort of philological algebra, before we can restore the existing forms 
 to their proper shape and their legitimate functions. But this pro- 
 cedure is one which justifies itself, and leaves no ambiguity as to the 
 correctness of the results. To begin with the substantive verb sum= 
 esuMj we cannot doubt that the future ero, eris^ erii^ &c., is another 
 form of the subjunctive «m, sis, sit, &c. The one has lost its charac- 
 teristic i, which the other has retained at the expense of its initial 
 vowel ; but even in its monosyllabic form it is not complete, for we 
 find the fuller word stem in the older writers, so that the complete 
 future or subjunctive must have been eslem^ero. In order to apply 
 this result to the ordinary verb, we must set aside the futures in -bo, 
 -bis^ -bit, &c., which are generally found in the first two conjugations, 
 and here and there (as ibit and quibit) in the fourth. In the third or 
 consonantal conjugation, the future generally ends in -am, -es, -et, &c. 
 The first person in -am belongs to the present subjunctive of the last 
 three conjugations, which exhibit -am, -as, -at, &c. ; the other persons 
 in -eSj -et, &c. are found throughout in the present subjunctive of the 
 first conjugation, which gives us -em, -es, -et, &c. To begin with 
 regam, we might assume an original reg-iam on the analogy oinavalis 
 for navialis, funalis for funialis, &c. : and we have many old forms to 
 convince us that the Latin subjunctive ended in im; such are temper•- 
 im, ed-im, du-im, &c. Consequently, reg-am was originally reg-iam 
 or reg-im, and this harmonizes with the form /wi[wi] for fuesa: so 
 that the corresponding aorist ought to be e-reg-i. According to this 
 principle amem is equivalent to ama-im; and reg-em (which must be 
 assumed from reg-es, reg-et, &c.), presumes an original reg-a-im^=^reg- 
 ia-im=^reg-sim, of which we have a further extension in reg-sero=^ 
 reg-se-sim=.reg-se-siem ; and a third in reg-sis-sem=^reg-se-se-siem 
 (above, § 378). Now regem=reg-si-m is really nothing more than the 
 determinate tense corresponding to [e]reg-si the aorist; and as one 
 
630 THE MOODS AND PAETICIPLES. [BOOK IV. 
 
 performs its functions in the subjunctive, the other in the indicative 
 mood, we can plainly see that the differences of mood, as they are 
 called, are set at nought by this pair of tenses, and we may infer that 
 there is, after all, rather a conventional than a real distinction be- 
 tween the modal and temporal forms. The dissimilitude of the future 
 indicative in -bo and the present subjunctive of the verbs which 
 admit this formation, is due to the subsequent introduction of this 
 composite tense. It is easy to understand why the first person of the 
 subjunctive has been called in by the other future: at any rate no 
 EngHshman need wonder that a broader form should be used for the 
 first than for the other persons of the future, for we are in the con- 
 stant habit of saying "I shall," when we say "you will," and vice 
 versa, 
 
 399 The Sanscrit language seems to have lost the future corre- 
 sponding to the aorist. We find an aorist or perfect in -sha-m analo- 
 gous to the Greek in -ΰα: thus, from pack- (τΐετίτω, coquere) we have 
 the aorist apaksham, and from srip- (eQTCELV, serpere)weha,Ye the aorist 
 asarpsam. We should of course expect to find a corresponding future 
 in -sM-mi, and such a form occurs in the Vedas (see Rosen, Big- Veda 
 Sanhita, p. iv); but the Sanscrit future always inserts an i, which, 
 according to the laws of euphony, becomes y; and the future oi pach- 
 is not pakshami, corresponding to apaksham, but paksliydmi^ which 
 rather corresponds to the Greek desiderative in -όείω. Now it is 
 remarkable that the Sanscrit has an aorist corresponding to this 
 future, in which the radical vowel is not increased as in the other 
 aorist; and this aorist, like the similar Greek aorist in -oelcCj is very 
 seldom used singly (Wilkins, Sanscrit Grammar, p. 297), but occurs 
 chiefly as an optative in conditional sentences {Gr. Gr. Art. 502, m.) ; 
 as jndnan ched a-bhavishyat, sukTiam a-bhavis9iyat, "if there were 
 knowledge, there would be happiness" (Wilkins, Sanscrit Grammar, 
 p. 655), i.e. εί Ιτΐίϋτήμη εϊη, καΐ ευδαιμονία αν εϊη. Or sometimes for 
 the sake of greater emphasis, the demonstrative tadd will be prefixed 
 to the apodosis in correlation to the conditional relative yadi: as in 
 the following instance, where the hypothesis is assumed to be false, 
 and where both clauses would have the imperfect indicative in Greek, 
 and the imperfect subjunctive in Latin {Gr. Gr. Art. 502, iv. (a)): yadi 
 ^ild komala a-bhavishyat, tadd grigdlair eva a-bhakshishyata, "if a 
 stone were soft, which it is not, in that case it would certainly be 
 eaten up by the jackals." When, however, the time of speaking only 
 is referred to (Gr. Gr. Art. 502, ii.), the unaugmented form, which is 
 used as the regular future, is invariably employed. Thus, Krishna 
 says (Bhagavad-Gita, xviii. gloka 68): 
 
CHAP, in.] THE MOODS AND PARTICIPLES. 631 
 
 ya idari paraman guhyan madbhakteslivabhidhashyati^ 
 hhakiin mayi parafi kritvd^ mamevaishyatyasangayah. 
 
 i. e. ya idam paramam guJiyam mat bhakteshu abhi-dhdshyati^ 
 bhaktim mayi pardm kritvd, mam eva eshyati a-sangayas. 
 
 —"he who shall explain this most excellent mystery to my worship- 
 pers, having performed the highest act of worship to me, shall ap- 
 proach me without hesitation." Abhi-dhd-shyati (a compound of the 
 root dhd-, "to place," Greek 0'£-, with the preposition abhif Greek ξτίί, 
 and therefore signifying "to place near," "to lay before," "to ex- 
 plain"), is the regular future, and is so used independently of any 
 protasis in the Nalas, xn. 76: vistdrena abhi-ddshydmi^ "I will tell 
 you at full length;" eshyati (=aishyati) is a similar future of the root 
 i-; the 2nd person occurs in the apodosis to a participle in Bhag, 
 Git. vm. 7. The only difference, in fact, between this and the Greek 
 optative is in the use of the augment, which therefore marks the 
 tense of this optative very strongly. The potential or subjunctive, 
 on the other hand, has no augment, and is merely formed from the 
 present (next to which it is placed by the Sanscrit grammarians) by 
 the insertion of this desiderative i. It is either an optative, as in the 
 Nalas^ xui. glok. 28, 29 : yadi pagyema tarn pdpdh, avagyam eva 
 hanydma, "could we see that wicked woman we would assuredly kill 
 her ;" or a subjunctive, as in Nal. xvii. 45 : tad vachas mama dvedyan, 
 yathd 7io jdmydd^ "that discourse must be made known to me, lest 
 he discover, &c." (ne cognoscat). We see, then, that in the Sanscrit 
 forms no less than in the Greek, the so-called modal distinctions 
 resolve themselves into mere differences of tense. 
 
 400 In our own and other languages, in which the moods and 
 tenses are formed by auxiliaries, we find the relation between the 
 future and subjunctive, and the subjunctive and optative, expressed 
 as a relation between present and past time. Thus, future "I will;'' 
 subjunctive "I would;" future "I shall;" subjunctive "I should;" 
 subjunctive "I may;" optative "I might;" &c. German future ich 
 iverde, subjunctive ich wilrde; future ich soil, subjunctive ich sollte; 
 subjunctive ich moge, optative ich mochte-, &c. 
 
 401 (2) Imperative and Infinitive. 
 
 It may be doubted if the imperative is really entitled to the rank 
 of a distinct mood. The marks which characterize it cannot be con- 
 sidered as modal inflexions, as they affect only the personal termina- 
 tion, which is generally omitted in the second and strengthened in 
 the other persons. It appears natural, that, in issuing a command, a 
 
632 THE MOODS AND PAETICIPLES. [bOOK IV. 
 
 specification of the person immediately addressed should be omitted, 
 and a great emphasis laid on the person to whom the command is 
 meant to apply. This is effected in English and German, by placing 
 the pronoun after the verb, as "give ye," gehen sie, &c.: or, in the 
 second person singular, by omitting the pronoun altogether, as "give 
 me the book," sieh da! A command is expressed in the ancient lan- 
 guages in a precisely analogous manner : in the second person singular, 
 the personal inflexion is frequently omitted, or when expressed, a 
 stronger form is used, as is always the case in the other persons. 
 
 402 The person-endings of the Greek imperative are, in the 
 active voice, as follows : 
 
 Second person singular: omitted as in tv7ttB\ -g, as in d'e-g, do-g^ 
 &c.; -O't, as in L(jta-Q'i', for this the Laconians used the indicative form 
 -ύί, just aS'O'sogin the same dialect is written olog: thus we have 
 αχταβί for άνάόΐηΟ'ί', κάββαΰυ for κατάβη^ι: see Valckn. ad Adon. 
 p. 104 : and the -^i is changed into -rt when a %- precedes, as in τί^Β- 
 tl; for the same reason the -g is turned into -v when a (J- precedes, as 
 in τύφο-ν for rvipa-g^ or TV7t-6a-d'i. 
 
 Third person singular: -τω, the idea of instrumentality being 
 expressed by the ablative case (§ 351). 
 
 Second and third dual and second plural are the same as those in 
 the indicative, the third dual having, however, ω instead of the η, 
 which appears in the secondary forms of that person in the indicative. 
 
 The third person plural is either -vtco {^ίβτ-ντω-ν) ον-τωΰαν: the 
 former appears to be the more genuine ; it is the same as the indicative 
 with a lengthening of the vowel, corresponding to that which appears 
 in the third singular. Whether the final -v is merely adscititious, or 
 intended to mark more strongly the genitive plural or ablative case, is 
 uncertain. With regard to the form in -τω-6αν, it seems to have 
 arisen from a mistaken analogy derived from the secondary tenses in 
 the indicative: the ear led to the impression that τνΛτέτω-ΰαν was a 
 plural formed from the singular, like ετετνφεί-όαν from ετετνφεί, and 
 they did not perceive the compensation-principle in the latter case. 
 
 403 The passive-endings are, in the singular, second person -(Jo, 
 contracted as in the secondary forms of the indicative; third person 
 -ΰΟ'ω: in the dual, second person -odOV, third person ΰ^ων: in the 
 plural, second person -ΰ^ον, third person -ύΟ'ων or -6%^ω6αν. In the 
 first aorist middle the ending of the second person singular is not -ω, 
 as it should be if contracted from -α-ΰο, but -ao. The neuter aorists 
 in -ην form their imperative like the ordinary verbs in -μι in the 
 active voice ; of course we have τνφ^^ηη not τύφ^'η^ο, for the reason 
 mentioned before. 
 
CHAP. III.] THE MOODS AND PARTICIPLES. 633 
 
 404 The Sanscrit imperative differs from the Greek in having a 
 first person in all three numbers and in both voices. This is also the 
 case in English, though only in poetry and in the plural, as "Leave 
 we the theme," — "Charge we the foe." The characteristie of the first 
 person imperative active in Sanscrit is -ni instead of -mi. In the case 
 of verbs in -ami, the first person imperative differs from the first 
 person present indicative only in the substitution of -ni for mi: in the 
 case of verbs in -emi, - omi, and -aumi, it is formed like the imperfect 
 indicative, by resolution into -aydni and -avdni. In the middle this 
 is contracted into di, by an elision similar to that which transforms 
 the first person indicative dtmanepadam from -me into -e. The second 
 person of the imperative active is expressed either by the person-end- 
 ing -(iAe= Greek -«9't, occasionally abbreviated into -hi, as hru-hi, "say;" 
 or by the mere crude-form of the verb, as tanu, like δείκνν in Greek. 
 The other persons in the active are expressed much in the same way 
 as in the indicative; the third persons singular and plural are -tUy 
 -ntu, instead of -ti, -nti^ so that they correspond to the Greek -τω, 
 -ντω. The second person singular dtmanepadam is -sva; the other end- 
 ings -iam, -mahai, -thdm, -tdm, -mahai^ -dhvdm, -ntdm, present the 
 locative case of the personal pronouns under a very strong form. 
 
 405 The Greek infinitive has the following endings in the active 
 voice. In the more recent authors the infinitive of the verb in -μι, 
 ends in -vai, preceded by a short vowel in the present tense, as η-θέ- 
 -vauj L-ora-vdl•, δι-δό-ναί; but by a long vowel or diphthong in the 
 second aorist, as d'Ei-vab, ΰτη-ναί, δον-ναι. The accent seems to show 
 that the infinitive in -h-vai, &c. must imply the compound termination 
 -ί-νοξ=--6ινοξ (§ 258), so that the infinitive τετνφ-εναί represents the 
 locative of a word analogous to δίκαυο-ύννη. The verbs in -ω form the 
 infinitive of the present, future, and second aorist in -ει- ν or -εΐ-ν, of 
 the first aorist in -(Jat, of the perfect in -ε-ναι. The passive infinitive 
 of all verbs ended in -(jO'ai', that of the neuter aorists in -η-ναί. All 
 infinitives in -εί-ν or -vat have -μεναι or -μεν as their representatives 
 in the more ancient authors. For -ευν the Cohans and Dorians wrote 
 -ην, the Dorians also -εν. The Dorians and -^Eolians also substituted 
 -ην for the infinitive ending -vao; and for the contracted verbs in -«oj, 
 -όω, there was an JEolic infinitive in -g, as γέλαϊς^ vtpo'igj for γελάν y 
 vipovv (Buttmann, Ausfuhrl. Sprl. § 105, Anm. 21). 
 
 406 The Latin language has two active infinitives: the one ter- 
 minates in -re or -se {dic-e-re, dic-si-s-se, es-se)\ the other in -turn 
 {dictu-m), which, in the modern grammars, is absurdly enough called 
 the supine in -um. In the passive, -er is subjoined to the former 
 
634 THE MOODS AND PAETICIPLES. [bOOK IV. 
 
 infinitive, thus from videre we have viderier=.videsyer ; this full form, 
 however, is generally contracted by the omission either of the charac- 
 teristic r= 5 of the active, as in dici-er; of the last syllable -er, as in 
 videri; or of both at once, as in diet; the latter infinitive is written 
 -tu (dic-tu) : modern grammars call it the supine in -u, 
 
 407 The Sanscrit infinitive is perfectly analogous to the Latin 
 infinitive in -turn. Thus the root gru (Greek κλν-), "to hear," makes 
 qrutas^ "heard," and grotum^ "to hear." These infinitives in -turn are 
 cases of verbal nouns: another case is the Sanscrit gerund in -tva; 
 thus from hd-tum, "to leave," or "the leaving," we have M-tvd, "by 
 or in the leaving" =^τω Isiituv. 
 
 408 Those acquainted with Greek syntax are aware that the in- 
 finitive is sometimes used to express a command: it must be remarked 
 too by every one, that there is a great resemblance between the third 
 person singular imperative passive and the termination of the passive 
 infinitive. A modern philologer (Grafe, das Sanskrit Verbum^ p. 58) 
 has gone so far as to propose, that the first person singular imperative 
 in Sanscrit, as tisht'h'dni, and the second person singular first aorist 
 imperative in Greek, as τν-φον, should be considered as forms of the 
 infinitive in -vau: nay more, that the first person plural, as tish-t'h'dma, 
 is the same as the infinitive ίΰτάμεν for ίβτάναι. It would be difficult 
 to convince us that these resemblances are more than accidental, 
 though, as we have shown above, the Latin language presents some 
 analogies which favour the supposition (above, § 362). 
 
 409 With regard to the similarity between τνΛτίΰ^ω, &c. and 
 τντίχεβ^Ι'αι^ &c. we must explain ourselves at greater length. It is 
 remarkable, that, where t appears in the active person-endings, -C^f 
 appears in the passive: thus we have tV7Ctb-T-ov,TV7ttB-6^ov\ τντΐτε-τε, 
 TVTtTS-ud'-e; τυτίτε-τ-ω^ τντίτε-ΰ^'-ω ; &c. At first sight one might be 
 disposed to think that this 6d'- is merely an arbitrary insertion to 
 mark the passive voice. But this view is overthrown by the appear- 
 ance of the same combination -6^- in the infinitives, where there is no 
 corresponding r in the active voice. Besides, in some instances, we 
 have seen that the -ud" admits of an easy explanation; in fact, merely 
 the imperative third person singular and the infinitives remain unex- 
 plained. We must, therefore, seek for some solution in these two 
 cases, and, if a probable and consistent theory suggests itself, we must 
 take it on its own merits, even though it may not harmonize with the 
 account which we have given of the same combination of letters in 
 the dual and plural person-endings. 
 
CHAP. III.] THE MOODS AND PARTICIPLES. 635 
 
 On examining the passive imperative-endings we find, that, though 
 the second person has generally the same ending as the corresponding 
 person in secondary form of the ordinary suffixes (τνπτον for τντΐτεΰο, 
 comTpave hvTtrov ; τετνψο^τετντΓ-οϊο, compare £τδη;ι^ο), the third per- 
 son always end in --Οό) or -6&ω (as τντΐτε-ύ^'ω, τετνφ-Ο'ώ), though the 
 corresponding secondary form is -το ; nay more, that the aspirate is 
 even extended to the preceding consonant, though this is not aspirated 
 in the corresponding tense of the indicative ; thus Ιτετντίτο, imperative 
 τετνφΟ'ω, We must, therefore, conclude that there is something essen- 
 tial and necessary in this aspiration. Now we observe, not only that 
 the present imperative τντΐτέό^^ω is like the present infinitive τύτΰτεΰ- 
 %^ai in this respect, but also that there is the same resemblance between 
 the perfect imperative τεχνφ^ω and the perfect infinitive τετνφ%αί. 
 There is only one way of explaining these resemblances, namely, by 
 supposing that they arose from the use of the infinitive to express a 
 command, and by a subsequent efibrt of euphony which accommodated 
 the final sound to the person-endings of the active imperative. Be- 
 sides, the Sanscrit iwcZa-iam , shows us that if the analogy of the 
 affixes was carried out, the third person of the passive imperative must 
 have been τντίτε-την, and this was driven out by its resemblance to 
 the corrupted dual. On the whole, we cannot fail to recognise in the 
 forms of the imperative, both active and passive, the influence of a 
 later and abnormal analogy operating on the basis of a conventional 
 idiom which employed the infinitive as the vehicle of commands and 
 entreaties. "We do not, however, believe that there is any connexion 
 between the infinitive active τνψαί and the corresponding imperative 
 middle, which must be a corruption of τν^Ι^α-ΰαι. 
 
 410 An examination of the active infinitive endings will show 
 us that there are in fact three, and only three, distinct forms : 
 
 (1) -μεναί, contracted to -μεν or -μεον; 
 
 (2) -VKL, contracted to -εν or -ευν or -ην; 
 
 (3) -g or -eg. 
 
 Although the same verb in difi'erent ages of the language appears 
 v/ith infinitives in -μεναο and -είν, we think it erroneous to say that 
 the form in -ευν is a contraction of that in -μεναυ. These three termi- 
 nations are, we conceive, participial endings, and therefore it is just 
 as possible for a verb to have two of these infinitive endings, as it is 
 for the same verb to exhibit two difi'erent forms of the participle. 
 
 We will first produce other instances of these endings with an 
 adjectival or participial signification. 
 
 (1) -μεναί. This form seems at first sight to be an inflexion of 
 the regular passive participle in -μεvog ; but we must recollect that 
 
636 THE MOODS AND PAETICIPLES. [bOOK IV. 
 
 the termination -μήν=-μέν-ς expresses an active agent, as we may 
 see in such words as τίροβατο-γνώμων"^, Ίίολν-πράγ-μων=^πολν- 
 7ίράγ-μ{ΐ]ν-ζ^ τίΟί-μήν =7ΐθί-μεν-ς^ &c. (above, § 256); and that an 
 active sense is generally conveyed by the combination -με-ν, when it 
 is followed by the second element; cf. χαρ-μονή, άρ-μονία, &c. It would 
 be most reasonable therefore to assume that the infinitive in -μενοα is 
 the locative case of the verbal abstract in -μονή or some analogous form. 
 With regard to the participles in -μενοξ^ it is doubted whether the 
 meaning is invariably and necessarily passive, though it must be 
 owned that in most cases this force may be assigned to the word. 
 Under the shortened form μνο-^ we have this compound affix, with a 
 participial meaning, in such words as κρήδε-μνον, "a fillet," i. e. "that 
 which is bound round the head:" μίδι-μνοξ-, "that which is mea- 
 sured" (a certain quantity of corn, Latin modi-us) βελε-μνον, *'that 
 which is thrown" (a dart), γυ-μνός, "stript" (from εκδνω, Pott, Etym. 
 Forsch. II. p. 182); μερι-μνα^ "that which is thought of or recollected," 
 comp. μερ-μηρ-ίζω, μερ-μερ -og, μάρ-τνρ, Latin me-mor or mes-mor^ San- 
 scrit root smri. In Latin we have seen ama-min% ama-minor; we 
 have also the participles alu-mnus, "reared;" auctu-mnus, "increased;" 
 vertu-mnus , "turned;" da-mnum, "given;" ceru-mna, "a load or 
 weight;" not, as Yoss thinks, for αιρουμενον, but, as Pott suggests 
 (Etym. Forsch. i. p. 279), connected with the Sanscrit root yas=^ad- 
 niti (because Sanscrit ayas=^'L•εL•tm ces), so that ce-ger is qui cerumnam 
 gerit. We have, besides, [shortened forms in -men corresponding to 
 the Greek infinitives in -μεν\ thus, from the root coZ-, "to raise up" 
 (cel-sus, col-lis, Greek κολώνη κέλλειν, cul-mus, ex-cel-lere, &c.), we 
 have not only colu-mna, but cul-men. Bopp (Annals of Oriental Lite- 
 rature^ p. 52) mentions dis-cri-men (which means quod discernitur, 
 not, as he says, quod discernit), stra-men=quod struitur, legu-men= 
 quod legitur, prce/a-wew, "what is said at the commencement" (cf. 
 Vergl. Gr. p. 1115). He also compares car-men with the Sanscrit 
 karman, "a deed," from the root krij "to make," and with the Greek 
 
 * This word is a synonym for τίοιμήν (cf. Ιτίπογνώμων, Jaculair. fr. 219), 
 and is used figuratively to denote a king, who can see into the hearts of 
 his subjects, in a very contorted passage which all the editors have failed 
 to construe. Msch. Agam. 768: 
 
 οβτΐξ ^ άγαΟ-ος προβατογνώμων 
 ovy, ϊβτι λαΟ'εΐν όμματα φωτός 
 τα δοκοϋντ ενφρονος έκ διανοίας 
 νδαρεΐ οαίνειν φιλότητί. 
 
 Here φωτός is the antecedent to οϋτις, and the meaning is "it is not possible 
 that mere semblances of friendship should deceive the eyes of a man, who 
 is a good discerner of character." 
 
CHAP. III.] THE MOODS AND PARTICIPLES. 637 
 
 7ίοίημα=Λθίή-μ8ν-τ. From the corresponding forms carmen and car- 
 mentis it seems to us probable that all the Latin neuters in -men were 
 originally furnished with the objective affix -i, so that these words 
 really belong to the same class with those in -ματ=-μεντ and -mentum. 
 Other instances are su-men=^quod sugitur, volu-men=quod volvitur, 
 se-men=quod seritur, &c. The Sanscrit termination -mdna frequently 
 makes a participle, which has a middle or active signification, unless 
 preceded by the syllable -ya^ when it becomes passive ; and the words 
 ser-mon^ λεί-μών, &c. (§ 256), may be considered as implying action. 
 
 (2) -vai. This termination also has both an active and a passive 
 participial meaning (§ 255): an active, in such words as τεχ-νη {from 
 τενχώ), "that which makes," τερτζ-νός, "he who pleases," &c. ; a pas- 
 sive, in such words as τεκ-νον-, "that which is brought forth," ΰκψνη^ 
 "that which is covered" (comp. ΰκί-α, ΰκό-τος^ Sanscrit root cJiliad, 
 "to cover," Hebr. ^S^, Latin squd-ma^ ob-scu-rus, scu-tum^ &c.), 
 6ρφ-ν6£ (ερεφω), d'aXit-vog, &c. In Latin we have the same termina- 
 tion with an active meaning in poe-na (Greek Λον-νή), "that which 
 atones," comp. the Sanscrit root pu-, "to purify," pe(^yna, "that 
 which flies" (Greek Λετη-νός, τίτη-νός) ; but in by far the greater num- 
 ber of cases it has a passive signification, as in ple-nus, sd-nus, reg- 
 num, pug-na^ more anciently puc-na (Greek τυκ-νόξ)^ do-num, lig-num, 
 "that which is bound" (a faggot), vd-nus (comp. vacuus), pd-nis, 
 dig-nus, "what is shown" {βεικ-\ &c. 
 
 (3) -ξ or -ΐξ. The analogy of the other two terminations -μεναι,, 
 -ναι, shortened to -μειν and -ειν-, would lead us to expect that this end- 
 ing was originally -ijcit, an opinion which is confirmed by the distinct 
 i,, which is inserted between it and the last vowel of the crude-form : 
 thus, we have γελαϊζ, not γελαΐς, and γελαϊς must be considered as 
 analogous to TV7tT8Lg for τντΐτεΰί, &c. One of the terminations which 
 we have seen used to express the ahstractum verbale, or action of the 
 verb, is -6i,g, as τΐράκ-ΰίς, μίμψόί£, λητί-ΰις, φά-όίς, &c. ; another, not 
 quite so common, is -Tvg, as Idri-Tvg, l7C^-Tvg, alYi-rvg, 6ρχηΰ-τvgJ 
 6ωφρovίΰ-tvg. These two endings, in -Ocg and -rvg, are equivalent in 
 meaning ; and the reader who has observed the changes in the second 
 personal pronoun from tu, tva, to si, will not be disposed to deny the 
 possibility of their being originally the same in form. It so happens, 
 that the two Latin infinitives are verbals, similarly related to one 
 another; to one ending in -se is a case of the Greek verbal in -uig; 
 that ending in -tum corresponds to the Greek verbal in tvg. We con- 
 sider the ^olic infinitive in -(?- or -ug, as a genuine representative of 
 the former Latin infinitive : every one knows the resemblance of the 
 JEolic dialect to the Latin, and it would have been strange, if the 
 
638 THE MOODS AND PARTICIPLES. [bOOK IV. 
 
 Greek had retained not traces of likeness in an inflexion of so common 
 occurrence as the infinitive mood. 
 
 Upon the whole, then, we have no hesitation in asserting that the 
 three forms of the Greek infinitive active were originally -μεναί, -vai^ 
 and 6aL•' or -(Jat, and that these are the locatives of three verbals. We 
 have seen that even the two former are often used with an active 
 sense; but if it is objected that they are more generally passive in 
 signification, and that therefore they cannot well express the action of 
 a transitive verb, it may be answered that the participle in -ndus in 
 Latin is used more frequently in a passive than in an active significa- 
 tion, and yet the ablative of this participle is used like a case of the 
 active infinitive — thus, dandus means "to be given," but dando, "by 
 giving;" other cases may be used in a similar manner, as ad opes dan- 
 das or ad dandum opes. Besides, although the verbals in -μεναυ -vai, 
 which are used as active infinitives, were strictly passive, we might 
 remark conversely, that the active participles in -vt are occasionally 
 used in an infinitive sense, in which the difi^erences of voice seem to 
 be neglected (see Varron. p. 361, note 2). The reason for all this is 
 very plain and simple. In the indefiniteness proper to the infinitive 
 mood, it matters little whether we consider the verb as transitive or 
 intransitive. For instance, what diff'erence is there in our own lan- 
 guage between "the thing is doing" or "the thing is being done"? 
 All this would have been seen long ago, if the Greek and Latin infini- 
 tives had not lost their case endings and become mere crude-forms. 
 The consequence of which has been, that, although it is admitted that 
 the Greek infinitive is to all intents and purposes a noun, to be de- 
 clined by the article, the loss of the final -au has prevented gram- 
 marians from discerning its relation to forms which may be declined 
 without articles or prepositions. The Latin infinitive stands in the 
 same predicament, except that as the Romans had other verbals still 
 admitting of inflexion, and had no article to help out the infinitive, 
 its employment as a noun is confined to the general objective or accu- 
 sative case. As a nominative, it occurs only in later writers who 
 were familiar with the Greek idiom. Supposing that we had not only 
 pugna but pugnamen and pugnatus, we should be able to represent 
 from one root all forms of the infinitive; pug-nd (jtvK-vca), pug-na- 
 mine (^νκ-να-μενοα), pug-na-se=pug-na-re (τΐνκ-νά-ΰεή, pug-na-tu 
 (Λνκ-να-τνΐ), pug-na-ndo, pug-na-tum, &c., being all difi"erent ex- 
 pressions, in the way of cases, of the same idea, — "closeness for the 
 sake of fighting." 
 
 It is perhaps right to add that the first aorist infinitive active (as 
 TV7t-6ai is, in our opinion, a representative of the third form of the 
 infinitive, the final s having dropt off". 
 
CHAP. III.] THE MOODS AND PAETICIPLES. 639 
 
 411 The ending of the Greek infinitive passive, Λve have seen, is 
 invariably -od'ai. This we shall now be able to explain without diifi- 
 culty. The second person plural in -ud'S must have been originally 
 -Οΰ'αο for -6•^ην ; cf. TV7tTB-6d'e with τντίτο-μαι and ετνΛτόμην. It is 
 therefore not an unreasonable inference that the infinitive TVTtreud^ai 
 was originally τντζτεΰΟ'ην or τντίτεΰ^ηναί; the <j before the & being 
 one of those euphonic insertions which are due to the Greek predilec- 
 tion for the combination 00". Consequently, τντίτεΰΟ'αί is ultimately 
 identical in form with κελενόΟ'ήναί. If this explanation is not satis- 
 factory, it only remains to suppose that τντίτεΰ^'αι, is a representative 
 of τντίτε-Ο'-ΰαί,, the inserted d' being the usual mark of the passive, 
 and the termination being the same as that of tV7t-6aL. 
 
 412 The -er, which marks the Latin infinitive passive, is that 
 sign of the locative which we have seen used as the characteristic of 
 the passive voice in the other moods. And for this reason we con- 
 sider that forms like dicier are mutilated, and that the r, which 
 marks the active infinitive, has been omitted between the two vowels, 
 just as in the genitive plural of all consonantal nouns. It is obvious 
 that spargier for spargerer (Hor. iv. Carm. xi. 8) is abridged on the 
 same principle as lapidum for lapiderum. The r, which is retained 
 in musarum, is also retained in the vowel-verbs, and this fuller form 
 (as vocarier=^vocasyer from vocare:=voeasie) is perfectly analogous to 
 that of the indicative passive (as vacatur from vocat). 
 
 The Sanscrit gerund in -tvd stands between the Latin infinitive -se 
 and the Greek verbal in -τύς^ just as the Sanscrit pronoun tvam stands 
 between the Greek 6v and the Latin tu (above, § 133). 
 
 413 (3) Participles. 
 
 Although we have already spoken of the Greek participle in con- 
 nexion with the adjective, and have endeavoured to explain the prin- 
 ciples of its syntactical use (above §§ 300 sqq.), method requires that 
 we should again consider it in its etymological and syntactical relation 
 to the infinitive mood and other inflexions of the verb. 
 
 The characteristic of the participle active in the present, future, 
 and aorists, is -nt=-nd, both in Latin and Greek. This appears only 
 in the oblique cases of the masculine and neuters, as TVJtrovTog^ η^'έν- 
 tog, tv^avrog^ ^εLKVvvτog, amantis, docentis, &c. In the nominative 
 masculine -ντ and -nt are always abbreviated; the 5, which is the 
 characteristic of the nominative case, being, however, invariably 
 retained in the older conjugation: thus, for xL^ht-g we have τL•^'είg^ 
 on the analogy oi εlg for Ivg', hutiov rvTCTOVt-gwehsbve τντίτ ων, where 
 the -g is absorbed, as in καλλίων for }ialkiovg\ for rvipavrg we have 
 
G40 THE MOODS AND PARTICIPLES. [bOOK IV. 
 
 τπψας', for δει,κννντς,δει,κννς', for amants, amans; for docents,docens; 
 &c. The feminine is formed according to the principles laid down in 
 the last book; and the neuter, by leaving out the nominative s; the t 
 is necessarily omitted. In Latin we have occasionally a lengthened 
 form of this word, used as an active participle, though it is 
 generally employed as an equivalent to the Greek verbals in -reog: 
 we conceive that secundus, morihundus, amabundus, oriundus, are 
 formed from sequ(J^nSy mori{^7is, ama(l)ns, ori{i)ns, as Agrigentum 
 Tarentum, Buxentum^ &c. are from Ακράγας, Τάρα^,ΰυξεί-?, where we 
 have the transition step in the assertion that the -^Eolians formed 
 these names as masculines in -εντός (Niebuhr, Hist, of Borne, i. note 
 148). The Boeotian patronymics in -ωνδας, which are undoubtedly 
 connected with participles (above, p. 464) exhibit a similar extension 
 of the ending, for, as we have seen, the τ of the participle belongs, 
 like the δ- of the patronymic, to a form derived from the second 
 pronominal element. 
 
 414 The Greek participle of the perfect active ends in -ώς, -νΐα^ 
 -ός ; -ότος-, -νίας^ -ότος, &c. The passive participle ends in -μένος, as 
 we have said in speaking of the infinitives. The Latin passive par- 
 ticiple ends in -tus, and the Greeks have a corresponding verbal in 
 -τός. The Sanscrit active participle ends in -wi, like the Greek and 
 Latin, with the exception of the perfect participle, of which the ter- 
 minations are -us, -usM. 
 
 We cannot agree with some modem scholars (Pott, Etymol. 
 ForscJi. I. p. 92; Giese, Mol, Dial. p. 103; Bopp, Krit. Gramm. rule 
 186) in considering the Greek and Sanscrit perfect participles as 
 sprung from the same form with the present. There is no trace 
 throughout of the n, which appears so essential to the ordinary par- 
 ticiple form, and, though it may be said that there is an appearance 
 of a weaker as well as a stronger form in the declension of the San- 
 scrit participle, yet this does not go so far as to establish an identity 
 between this participle and one in which the weaker form is con- 
 sistently employed. It is very probable that the Sanscrit perfect 
 participle contains the Sanscrit ending -vat, which we find in td-vat, 
 "so much." dhana-vat, "rich," &c. The Greek perfect participle ends 
 in -/or, as appears from the neuter -^ος for -/or, and the oblique cases 
 -J^6r-og, &c. The nominative masculine in -ως is a representative of 
 -Porg, the length of the vowel being a compensation for the lost con- 
 sonants. The feminine in -via and the lengthened t would seem to 
 point at once to the loss of an aspirate or digamma, and of a sibilant, 
 so that this form is perfectly analogous to the Sanscrit. "We have 
 remarked above (§ 329), the connexion between ^ρ-/αοΓ-θ5 and ηρως= 
 
CHAP. III.] THE MOODS AND PARTICIPLES. 641 
 
 rjQ-faOT-S' The affinity between this termination -or or -Por, the 
 Sanscrit -vat, and the common adjective ending -xog, will appear from 
 the following considerations. That τΐατρίκός and τνατρωοξ are all but 
 synonyms is well known (see Hermann ad Eurip. Bacch. 1362; Neu- 
 mann ad Aristot. Rerumpuhl. Beliquias, p. 65). Now we find that 
 TCatQVLOS, TtaTQvog, τΐάτρως, τίατρωόξ, are also nearly synonymous; 
 and that these words are merely by-forms, no etymologist will deny. 
 We likewise find μητρνίά, apparently the feminine of μήτρως and 
 μητρvL•όg^, also the adjective μητρυκ6£. It appears to us that all these 
 words, as well as the Latin patruus, Sanscrit pitrwyas, belong to the 
 same class as the perfect participles. The feminine termination -vlol 
 sometimes occurs in words apparently participles, but evidently not 
 perfects, as αρπνιαι (αρπουβαή^ αγνιά (αγονβά), οργνια (ορεγονΰά)^ 
 (above, § 296). It has been already mentioned (§,257) that the Sanscrit 
 feminine corresponding to these forms in -νια is -ushi: thus, τετνφνΐα 
 =itutupushi. According to this analogy we may class the Latin noun 
 sec-uri-s=^sec-usi-s, "the cutter," with the participial forms just 
 mentioned (see Bopp, Vergl Gr. p. 1097). 
 
 415 The participle termination in -nt is an adverbial formation 
 from a verbal root, bearing the same relation to the forms in -d-, -dus^ 
 that the adverb i-nde opposed to i-hi does to avTO-d'sv opposed to 
 avt6-%L ( Varron. p. 290). It is formed, as we have already intimated 
 (§ 263), by appending the second pronominal element, in the form ^- 
 ='rJ^ or rt, to an inflexion in i;, of the simplest verbal noun, so that it 
 differs from the common verbal noun in -8, as οϊτιονδε differs from οϊτιο- 
 %BVj or the adverb %ρνφανδ6ν from its other forms κρνβ^ο; and κρνφη- 
 δόν. It is only by considering it in this way that we can harmonize 
 all the phenomena. On the one hand, the verbals in -τος, -Tsog, which 
 so nearly correspond in meaning (Gr. Gr, p. 190), and evidently belong 
 to an original from -τείχος, Sanscrit tavya, are thus brought into unison 
 with the verbals in -dus (cupidus, &c.) and the gerundives in -ndus^ 
 which are only lengthened forms of participles in -nt (cf. sequens with 
 secundus)] and on the other hand the exact correspondence in mean- 
 ing of the gerunds in -ndi, -ndo, -ndum and the supines in -tum^ ~tu^ 
 finds its explanation in the identical use of the adverbs in -νδον, which 
 connect themselves with the former, and those in -tim, which are 
 obviously cases of the latter. Even in its extended form -tio(n), this 
 verbal in -tus^ -tis, retains its active force, and we have constructions 
 like hanc digito tactio^ "a touching of her with the finger," hue re- 
 ceptio ad te meum virum, "a receiving of my husband here into your 
 house," &c. (Bopp, Vergl. Gramm. p. 1255). To the same class 
 belong the nouns of agency in -της, -τηρ, -tor, and their active force 
 
 TT 
 
642 THE MOODS AND PAKTICIPLES. [bOOK IV. 
 
 is shown in the future participles in -turns (cf. genitoVf &c. with 
 geniturus^ &c.). 
 
 The passive participle in -μενος^ to which we have already referred 
 in speaking of the infinitive in -μεναυ (§410, (1)), seems to us most 
 naturally explicable as a derivative from the noun of agency in -μεν-. 
 In itself therefore it is not essentially passive. But the analogy of 
 the Greek and Latin languages furnishes us with many examples of 
 this interchange of meaning when the noun of agency passes into an 
 attribute. Thus we have seen that even the noun of agency in -τηζ^ 
 when it is used adjectivally, may become equivalent to a passive par- 
 ticiple (above, § 254, p. 455). Similarly, the verbals in -τός, -τέος, 
 which properly express the meaning of the active infinitive, and 
 which not only govern a case, in such phrases as rovg φίλους ενεργε- 
 τητεον εβτί^ but express a capacity of action, as in Soph. Track. 466: 
 εΐ'τί. τω ' μω άνδρΐ μεμτΐτόξ εΙμί, are not only apparently passive, when 
 predicated directly in such phrases as ώφελητέα Col η %6Xig εΰτίν, but 
 actually passive in such instances as αίρετος^ which denotes not only 
 "eligible," i. e. "a man to choose, capable of being chosen" (eligendus), 
 but "chosen" (electiis). It may of course be said that in this case 
 the termination -rog is not a mutilation of -τέος, but a merely objective 
 affix of the third pronominal element like the Latin and Sanscrit 
 passive participles in -tus -tas^ and reference may be made to the 
 corresponding use of the verbal adjectives in -vog (above, § 253). 
 Allowing that this is still an open question, it may be observed on 
 the other hand that we have an active infinitive in -ναι as well as 
 -μεναο, and as the superlative-ordinals in -tus and -mus may have 
 sprung from the fuller form in -ti-mus (above, § 164), it is certainly 
 not impossible that -vao represents only a part of -μεναυ, the specially 
 objective part, and that these adjectives in -vog are explicable in the 
 same manner. At any rate, it is sufficiently clear that the participles 
 in -μένος and -tus are used with a passive signification, even when the 
 verbs, to which alone they can be referred, have a deponent sig- 
 nification. For example, we have the following passive participles 
 from deponent verbs: άτΐηγημένος (Herod, i. 207, ix. 29); ητυαμενος 
 (Thucyd. III. 61); κεκτημένος (Id. vii. 70); εξειργαΰμενος (Msch.Pers. 
 517; Agam. 1352); εντε^^νμημένος (Plat. Crat. p. 404 a); ληόόμενος 
 (Soph. EL 1248); complexus (Cic. Bosc. Am. 13, 37); partitus (Liv. 
 XXVII. 8, XXVIII. 19); adeptus (Sail. Cat. 7); opinatus (Cic. Tusc. iv. 
 6, m. 31; Offic. n. 6); meditatus (Offic. i. 8); interpretatus (Sail. Jug. 
 17 ; Cic. Divin. i. 52) ; periclitatus (Cic. Lcel. 17) ; detestatus {de Leg. u. 
 11, 28); confectus (Sail. Jug. 17). And conversely we have participles 
 in -μένος, -tus with a transitive or deponent signification, when there 
 are no deponent verbs to which they can be referred, as οΙκημενος 
 
CHAP. III.] THE MOODS AND PARTICIPLES. 643 
 
 (Herod, i. 27, vii. 21); consideratus (Cic. Ccecin, 1), and the common 
 instances, cautus, circumspectus, exosus, falsus, tacitus, &c. (Lubker, 
 de Participiis, p. 20). It seems to us that the simplest way of meeting 
 these difficulties is to connect the participial form with the infinitive, 
 and to direct the student's attention, as we have done elsewhere 
 {Varron. p. 361), to the fact that the difference between active and 
 passive really becomes evanescent in the infinitive use of a verb. 
 
 416 It will be inferred from what we have said, that we con- 
 sider the participle as ultimately identical with the infinitive. In 
 usage they are different, and the difference consists in this, that the 
 infinitive, originally a participle or verbal noun in the locative case, 
 has, in process of time, lost all traces of its inflexion, and may be con- 
 sidered as the most general and crude form in which the verb can 
 appear — γενίκώτατον xad'eoTog των άλλων ρημάτων, as Apollonius 
 says {Syntax. ΊΏ.. 13, p. 229 Bekk.): whereas the participle, instead of 
 being a particular case, the locative or instrumental, of a verbal form, 
 has received or retained a set of inflexions which constitute it a mix- 
 ture of adjective and verb. It will be remembered, however, that the 
 participle becomes an adjective only in the same way in which the 
 infinitive, and consequently the participle itself, when not joined to 
 some noun, becomes a substantive, namely, by the syntactical con- 
 trivance which we call prefixing the definite article (above, § 300). 
 
 417 There are adjectives which have the same termination as 
 the active participle, for instance χαρίΗς^^χαρίεντ-ς. These adjectivec, 
 however, do not express an action, but a quality, and therefore may 
 be joined to a substantive as an epithet without an article, or even, as 
 we have shown on a former occasion, become to all intents and purposes 
 a substantive. Thus χαρίεις αν&ρωΛος means simply "an agreeable 
 man;" τί^είς αν^ρωτίοξ, on the contrary, could not mean "a placing 
 man" or "a man who is in the act of placing," but "a man, when or 
 if he is placing," an hypothesis of some kind of action being implied. 
 In the case, however, when the article is prefixed, the participle and 
 adjective stand on the same footing: δ χαρίεις αν^ρωτίος, or δ χαρίεις^ 
 means "the agreeable man ;" δ τι^εΐξ αν^ρωποξ or δ rt^ftff, "the placer." 
 And in this relation the participle and infinitive are virtually iden- 
 tical: τω τι^^εναυ and τω τι^εντι equally signify "by the placing." 
 
 418 It is a theory of Grimm's {Deutsche Grammatik, i. pp. 1020 
 foil.), that the Teutonic infinitive was originally declinable, the 
 ordinary form in -an being the accusative, which is also used sub- 
 stantively as the nominative, and the forms in -annes, -anne, &c., 
 
 TT2 
 
644 THE MOODS AND PARTICIPLES. [bOOK IY. 
 
 representing the genitive and dative respectively. Grimm recognises 
 the genitive in the new High German forms meidens^ fragens^ &c., and 
 the dative in the common infinitives meiden, fragen^ on the analogy 
 of zeiclien^signo^ and regen=^pluvia. He adds (p. 1022), "an unor- 
 ganic participle in -nd, declinable like an adjective, and with a pas- 
 sive signification, has gradually developed itself out of the old nn 
 and the preposition ze prefixed, by an interchange with nd, just as 
 niemannes has become niemandes: thus we have ein zu lesender 
 (legendus), ein zu gebender (dandus). Perhaps there is still time to 
 expel from the language this stiff and unnatural formation." We 
 cannot agree with the great philologer in his disapprobation of this 
 form, which seems to us to be confirmed by every analogy. Our English 
 infinitive is the mutilated form of the dative of such a participle or 
 gerund. Thus, in Wicklifif's Bible we have thou that art to comynge^ 
 which corresponds to the Anglo-Saxon ^u ^e to cumenne eart (see 
 Diversions of Purley^ Vol. i. pp. xxxiii, 450, Vol. ii. p. 505, Taylor's 
 Edition). Rask says that the jaresent infinitive in Anglo-Saxon is 
 never used with the particle to as in modern English, though the 
 gerund always requires ίο, whence he concludes that the gerund is 
 nothing but the dative of the infinitive {Anglo-Saxon Grammar^ 
 § 400). In our opinion the infinitive was originally the participle in 
 -nt, which became -nd in German, while the flexion form of the old 
 Saxon, which inserted j before the case-ending, as in slapandjes= 
 dormientis, led to the English -nge and -ng. This participle, when 
 used as an infinitive, lost its inflexion, and could only be used as a 
 gerund or case by means of the auxiliary preposition zu or to. It 
 also suffered mutilation, being deprived of its final d in German, and 
 reduced to the crude-form of the word in English. The forms in -nn 
 are assimilations for which we have many analogies in the Teutonic 
 languages. 
 
 419 The Latin and Sanscrit languages, which have no definite 
 article, and have therefore retained their inflexions longer than the 
 Greek, afford us a direct proof of the view which we have taken of 
 this question. The Latin expresses by cases of the lengthened parti- 
 ciple in -nduSj and the verbals in -iwm, -tu, those relations, which, in 
 Greek, are generally conveyed by the infinitive with the definite arti- 
 cle: — Priscian calls all these cases gerundia: gerundia quoque vel 
 participialia , quum participiorum videantur habere obliquos casus ^ 
 nee tempora significent, quod alienum est a verbo (legendi, legendo., 
 legendum, lectum, lectu); infinitivi tamen vice funguntur. quod solet 
 apud Groecos articulis conjungi (p. 808). They had also an infinitive, 
 which, like the Greek, was a mutilated form of the locative of a ver- 
 
CHAP. ΠΙ.] THE MOODS AND PAETICIPLES. 645 
 
 bal in -sis^ but as they had no definite article, they could not treat 
 this as the Greeks did their infinitive ; it remains, therefore , as the 
 mere crude-form, expressing the action of the verb, and perfectly 
 equivalent to the verbal in -turn. The Sanscrit expresses all the 
 relations of gerund and infinitive by verbal-forms in -tvd, -ya, -turn. 
 The last of these is equivalent to the Latin verbal in -turn : deprived 
 of its final letter, it may form the first part of a compound word, like 
 any other crude-form ; thus we have such words as je-tu-kdma, "de- 
 sirous of vanquishing," &c. The verbal in -tvd is either the instru- 
 mental in -a, or a remnant of the locative in -dm; cf. tvam and τν. 
 The verbal in -ya, which is only formed from roots compounded with 
 prepositions, and has therefore lost its case-ending, owing to the 
 weight of the forms in which it is found, must be considered as the 
 corresponding case of a similar participle in -ya. All these three, then, 
 are verbals formed by means of the second pronominal element, which 
 we have seen of so much use in the formation of abstract nouns. The 
 elements -tva and -ya are combined to form the future ending -tavya, 
 which is equivalent in meaning to the Greek verbal in -thoQ (perhaps 
 rfPoff), and to one use of the Latin participle in -ndus. 
 
 420 The oblique cases of participles in Greek, Latin, and San- 
 scrit, are often used absolutely, that is, they form, in combination 
 with nouns in the same case, a distinct though subordinate sentence 
 without the intervention of any finite verbs. When we use the in- 
 finitive with an oblique case of the article, as in τω ri^ivai, "by the 
 placing," we employ it as an abstract noun with a general reference. 
 When we use the instrumental or genitive case of the participle, we im- 
 ply hypothetical placing, of which, however, the agent or instrument is 
 some individual, so that the \^erb-form has an adjectival, but predicative 
 value. When this individual is expressed in the same case, we under- 
 stand that a supposition is made with regard to his placing: and thus 
 ^ίοδώρω τι^ίνη or Διόδωρου tid'evTog^ τοιαντ οίττα γίγνονι «V, 
 means "that as often as Diodorus placed, certain things would ensue by 
 or out of that action of his." We endeavoured to show, when speaking 
 of the adjective (above, § 300), that the participle must always be 
 understood in this way, when it appears, not as an attributive word 
 or epithet, but as a predicate, that is, when it is unaccompanied by 
 the article. In modern German the predicate, whether adjective or 
 participle, is not only distinguished by the want of the article, but 
 also by mutilation in its inflexions. And here it must be remarked, 
 that the German language possesses a logical advantage over all 
 others; for it not only marks a distinction between the attributive 
 used without the article and the attributive preceded by it — as gute-r 
 
64.6 THE MOODS AND PABTICIPLES. [bOOK IV. 
 
 Mensch^ der gute- Mensch, — but also between the epithet in this last 
 case and the predicate — as in der Mensch ist gut (see Grimm, Deutsche 
 Grammatik^ iv. p. 577). This is surely the ultimate refinement of 
 a language in a logical or syntactical state. The German and Latin, 
 however, fall short of the Greek in one use of the absolute participle. 
 They rarely express the whole hypothesis by a single participle, in 
 other words, they do not often use the absolute participle or predicate 
 impersonally, but generally and as a rule add the subject. "The 
 German and Latin syntax," says Grimm (iv. p. 894), "generally 
 requires that the subject should be expressed by the side of the abso- 
 lute participle: the Greek, which is freer and less constrained, allows 
 us to use absolutely the participles of impersonal verbs, whose subject 
 is not defined : e. g. νοντοξ τίολλω (not Λολλον), 'when it was raining 
 hard;' ΰαλτίίξοντος, 'as we was sounding the trumpet.' I am not 
 able to adduce a Latin pluente, ningente, buccinante, in any such 
 sense; still less a Gothic rignjandin or atrignjandin, haurnjandin, or 
 at haurnjandin, although such absolute constructions are conceivable. 
 Least of all would it be allowable in the dialects which prefix the 
 indefinite es to their impersonal verbs. It was necessary always to 
 subjoin a definite subject to the absolute participle ; e. g. in old High 
 German reganontin himile; Gothic at rignjandin himina; Latin ccelo 
 pluente. It is only by way of exception that certain Latin and 
 German formulae allow the omission of the subject; e. g, comperto, 
 posito; new High German ^esei^i, kaum gesagt^ anders ausgedruckt.''^ 
 When a participle is used absolutely in Sanscrit it generally appears 
 connected with some noun in the locative case, which is the proper 
 inflexion for this construction; the genitive and ablative cases, which 
 are the most common forms in Greek and Latin respectively, seem to 
 owe their employment for this purpose to their having superseded the 
 old locative. The following instance is given by Wilkins (Sanscrit 
 Grammar, p. 632): 
 
 mr\te pitari, te vird, vandd etya sva mandiran, 
 na-chirdd eva vidvanso vede dhanushi-cha a-hhavan, 
 "their father being dead (locatives), those heroes, by the having gone 
 (verbal in -ya) from the forest to their own abode, after no long time 
 even, became skilled in the Veda and in the bow" (we observe, by 
 the way, that chira is the modern form oi jira from jara-Sj γηραζ- 
 ajira-s is found in the Big- Veda: see Lassen's Antholog. Sanscritica, 
 p. 97, 1. 10, and note 1). It is a remarkable instance of the looseness 
 of the Indian sjrntax, that the indeclinable verbals in -ya and -tva, 
 which are obviously fragments of cases (above, § 419), may be used 
 absolutely, though the nominative case of the sentence is not the 
 noun to which they refer: either with an active sense, as in the 
 
CHAP, in.] THE MOODS AND PARTICIPLES. 647 
 
 HitSpadega, p. 10, 1. 185 : iti uk-tvd, ganaih ganair upagam-ya, tena 
 vydghrena dhritak sa pdnthah achintayat, "when he (the tiger) had 
 thus spoken, and had come up to him slowly, slowly, the traveller, 
 caught by the tiger, began to reflect," — where uk-tvd and upagam-ya 
 refer to vydgJirena^ the instrumental of vydghra-s^ whereas pdnthah is 
 the nominative to the verb achintayat ; — or with a passive sense, as 
 in Nalus, xiv. 17: 
 
 krodhdd-asuyayi-tvd tan, rakshd me blavatah kritd, 
 
 "as I (the serpent) have cursed him in anger, thy preservation has 
 been effected by me." It is unnecessary to dwell upon these Sanscrit 
 forms : the reader who is curious about them will find acute reasoning 
 on the subject, fortified by an abundance of examples, in a paper by 
 W. von Humboldt in the Indische Bibliothek, Vol. i. pp. 433 foil.; 
 Vol. 11. pp. 72 foil. 
 
 421 Paradigms. 
 
 Having thus investigated all the inflexions of person, tense, and 
 mood, which distinguish the verb in the three classical languages of 
 the Indo-Germanic family, we shall, for the sake of the younger 
 student, select an example or paradigm from each of these languages, 
 in order that the principles of classification which we propose may be 
 more evident. We shall take the Greek language first, because it 
 possesses the most complete system of moods and tenses; then the 
 Sanscrit, which falls short of it in this respect ; and, lastly, the Latin, 
 which is the most defective. 
 
 In the arrangement of the tenses and moods in the following 
 paradigms, we have adopted the more general division into definite 
 and indefinite tenses, and, making the active voice of the Greek verb 
 the basis of comparison, we have five definite tenses and as many 
 indefinite. The order in which we have taken these tenses is chosen 
 rather from etymological considerations, than from regard to the con- 
 nexion of past, present, and future time. We have given not only 
 the forms which really occur or might occur in classical Greek or 
 Latin, but also those which we infer must have existed in the oldest 
 state of the language. The three roots we have chosen are all analo- 
 gous in point of meaning, and identical etymologically in the Sanscrit 
 and Latin. 
 
648 
 
 THE MOODS AJuD PARTICIPLES. [bOOK IV. 
 
 422 I. Greek root rvTt-', pronominal suffix -τ-. 
 
 (1) Active voice, or, verb with person-endings in the 
 instrumental case. 
 
 (a) 
 1st Definite Tense. 
 
 Indicative Mood. 
 
 1st Indefinite Tense. 
 
 (Imperfect). 
 
 Supposed original 
 forms. 
 
 τυτΐ-το-μι, 
 tvjt-Te-6v 
 
 Existing 
 forms. 
 
 (τντίτώ) 
 {tvittug) 
 
 (TVTttEi) 
 
 Supposed original 
 forms. 
 
 ε-τνΛ-το-μί 
 ε-τυτΐ-τε-ΰί 
 
 £-TV7t-TB-ti 
 
 Existing 
 forms. 
 
 (ετνπτον) 
 (ετντίτεξ) 
 (ετντίτεν) 
 
 (τνΛτετον) 
 
 
 (ετντΐτετον 
 
 (τντίτετον) 
 
 
 (ετνπτετην) 
 
 τυτί-τό-μεόι, (τντΐτομεν) 
 
 ε-τνΛ-τό-μεΰί 
 
 (ετντίτομεν) 
 
 τνΛ-τε-τε6ί (τντΐτετέ) 
 
 ε-τνΛ-τε-τεΰο 
 
 (ετντίτετε) 
 
 τνΛ-το-ντι (xv7txov(5i{v)) 
 
 ε-τνπ-το-ντί 
 
 (ετντϋτον) 
 
 2nd Definite Tense. 
 
 2nd Indefinite Tense. 
 
 {Perfect). 
 
 (Plus-Perfect). 
 
 τεζντί'φα-μϋ (τετνφα) 
 
 ε-τετυφει/α-μυ 
 
 (εχΒτνφειν) 
 
 --τε-τνφα-μυ 
 
 ε-τετνφε'υα-βι 
 
 (ετετνφεις) 
 
 τετνφα-όί (τέτνφας) 
 
 ε-τετυφεΐ)α-τι 
 
 (ετετνφεή 
 
 τετνφα-η {τεζνφεν) 
 
 
 (ετετύφείτον) 
 
 (τετνφατον) 
 
 
 (ετετνφείτψ) 
 
 {χετνφατον 
 
 ε-τετνφε'υά-μεβι 
 
 (ετετνφειμεν 
 
 τετνφά-μεύί (τετνφαμεν) 
 
 ε-χετνορε'υά-τεβι 
 
 (ετετνφείτε) 
 
 τετνφά-τεΰϋ (τετνφατε) 
 
 ε-τετνφε'υα-ντι 
 
 (ετετύφειΰαν) 
 
 τετνφα-ντν (τετνφαΰίν) 
 
 
 
 3rd Definite Tense. 
 
 3rd Indefinite Tense. 
 
 (Future). 
 
 (Aorist). 
 
 τυΛ-ΰο-μι (τνιρω) 
 
 ε-τντί-ΰα-μι 
 
 (kvjpa) 
 
 τνΛ-6ε-0υ (τύφείς) 
 
 ε-τνΛ-ΰα-6ί 
 
 (hvipag) 
 
 τνΛ'ύε-τυ (τν-φεί) 
 
 ε-τυτί-βε-τυ 
 
 (ϊτν^εν) 
 
 τντϋ-ΰΟ'μεΰί 
 τνΛ'ΰε-τεΰι, 
 
 TV7C-60-VU 
 
 (τν'φετον) 
 
 (τύ'φετον) 
 
 (τνχρομεν) 
 
 (τύτρετε) 
 
 (τύ'φονΰίν) 
 
 ε-τυη-ΰά-μεΰί 
 ε-τντί-ύά-τεΰί 
 ε-τντί-ύα-ντι 
 
 (ετύ'φατον) 
 
 (ετυφάτην) 
 
 (ετν'ψαμεν) 
 
 (ετύ'φατε) 
 
 (ετνψαν) 
 
CHAP. III.] THE MOODS AND PAETICIPLES. 
 
 649 
 
 (b) Subjunctive and Optative Moods. 
 4th Definite Tense. 4th Indefinite Tense. 
 
 (Present Subjunctive, or Present 
 
 (Present Optative, or Past 
 
 Desiderative). 
 
 Desiderative). 
 
 Supposed original 
 
 Existing 
 
 Supposed original 
 
 Existing 
 
 forms. 
 
 forms. 
 
 forms. 
 
 forms. 
 
 τνΛ-τΐί-ομι 
 
 (τνΛτώ) 
 
 (ε)-τvJt-τ-yo-μι 
 
 {τντίτοιμι) 
 
 rvTt-t-ya-uL 
 
 {τντίττ^ΐζ) 
 
 (e)-TV7t-T-ya-0i 
 
 (τντίτοίς) 
 
 TV7t-t-ya-TL 
 
 (TVTtty) 
 
 {lytvit-t-ya-ti 
 
 {τντίτοι) 
 
 
 (τντιτψον) 
 
 
 (tVTtTOLTOv) 
 
 
 (τντίτητον) 
 
 
 {τντίτοίτψ) 
 
 τνΛ-τ-ρό-μεΰί 
 
 (τντίτωμεν) 
 
 (£)-TV7t-T-yo^e6L 
 
 (τντΐτοιμεν) 
 
 tv7t-t-ya-Ts6L• 
 
 (τντίτητε) 
 
 (lyrvx-t-ya-Tsei 
 
 {τνπτοιτε) 
 
 tv7C-t-yo-VTL 
 
 (τυτίτωβιν) 
 
 {l)-rv7t-T-yo-vxi 
 
 {τντίτοιεν) 
 
 5th Definite Tense. 
 {Desiderative Verb in -6είω). 
 TVTt-uyo^L (τατί-ΰε-ίω) 
 
 &c. &c. 
 
 5th Indefinite Tense. 
 {JEolic Aorist Optative), 
 (ε)-τύ-ΰya-μL {τντί-όεια) 
 
 &c. &c. 
 
 And so on through the other tenses in the later state of the 
 language. 
 
 (c) Imperative. 
 
 Direct Command {Present Imperative). 
 
 Supposed original forms. 
 
 τνΛ-τε-{^ι) 
 τνΛ-τετω 
 
 Existing forms. 
 
 {τύπτε) 
 {τντίτέτω) 
 
 {τντίτετον) 
 {τυτιτετων) 
 
 τνΛ-τε-τε-ΰι {τνηχεχε) 
 
 τντί-τό-ντω {τντίτόντω-ν) 
 
 And so on through the other tenses. 
 
 {d) Infinitive. 
 
 The mere state or action. 
 
 τυΛ-τε-μεναι 
 
 τυτί-τε-ναι {τντιτειν) 
 
 τνΛ'ΰαι-ζ {τιτψαή 
 
 And similarly in the other tenses. 
 
650 
 
 THE MOODS AND PAETICIPLES. [bOOK IV. 
 
 (2) Beflexwe voice, or verb with person-endings in the locative case. 
 
 Two of the tenses of the indicative will be sufficient to show 
 the principle. 
 
 1st Definite Tense. 
 
 {Present). 
 
 1st Indefinite Tense. 
 {Imperfect), 
 
 Supposed original 
 forms. 
 
 τντΐ-τό-μην 
 τνΛ-ΤΒ-ΰην 
 τνΛ-τε-την 
 
 τν7ί-τθ'με-0•ην 
 
 τντί-τέ-ό-Ο'ην 
 
 tvTC-ro-vtYiv 
 
 Existing 
 forms. 
 
 (τνΛτομαι) 
 
 (τντίτευ) 
 
 (τνπτεται) 
 
 {τντΐτόμε^ον) 
 
 (τντίτεβ^ον) 
 
 (τνπτεΰΟΌν) 
 
 {τυτίτόμε^α) 
 
 (τντΐτεΰΟ'ε) 
 
 {τντίτονται) 
 
 Supposed original 
 forms. 
 
 ε-τντΐ-τό-μην 
 ε-τυτί-τέ-ΰην 
 ε-τυπ-τέ-την 
 
 ε-τυΛ-το-με-%'Ύΐν 
 
 ε-τνττ-τέ-ΰ-Ο'ην 
 
 ε-τυπ-τό-ντην 
 
 Existing, 
 forms. 
 
 (ετντίτόμην) 
 
 (ετύπτου) 
 
 {ετνπτεχο) 
 
 {ετντίτόμε^ον) 
 
 {ετύτίτεβ^ον) 
 
 (ετντίτέΰΟ-ην) 
 
 (ετνπτόμεΟ'ά) 
 
 (ετνπτεΰΟ'ε) 
 
 {ετνπχοντο) 
 
 (3) Passive voice, or formations with the element thya, or ya, the 
 person-endings being still in the instrumental case, 
 
 1st AoRiST or full form. 2nd Aorist or contracted form. 
 
 ε-τνφ-%'υα-μι {ετνφ^ψ) ε-τύτζ-'υα-μι {ε-τνηψ) 
 
 ε-τνφ -d'ya-oL {Ιτνφ^Ύΐξ) l-TV7t-ya-(5L• (ετύτΐης) 
 
 ε-τνφ-^^α-τι (ετνφΟ^η) kvTt-ya-u (ετυΛη) 
 
 ε-τvφ-^'yά-μεΰL 
 ε-τvφ-^'yά-τεύL 
 l-rv(p-%ya-vrL 
 
 (ετνφ%•ψον) 
 
 (ετνΟ'φήτην) 
 
 {ετνφ^ημεν) 
 
 (ετνφ&ψε) 
 
 (ετνφ^^ηΰαν) 
 
 (ετνπητον) 
 
 (ετντΐήτην) 
 
 ε-τv7t-yά-με(JL• (ετντΐημεν) 
 
 ε-τv7t-yά-τεΰi {ετντίψε) 
 
 B-TV7t-ya-VTL (hvTtrjuav) 
 
 From these are formed futures with person-endings in the locative 
 case, τνφϋ'ήόομαί, τντίήΰομαί, like τετν^ομαι from τέτνφα, and we 
 have recognised the suffix -Ο-ηίΤί the infinitive τν7ίτε6&αί=τνπτε(5^'η- 
 vai (§ 411). 
 
 The very complete, and indeed superabundant set of inflexions, 
 which this paradigm assigns to the Greek verb, hardly falls to the lot 
 of any single verb in the writings which have come down to us. Like 
 every other copious language, the Greek has more than one word to 
 express our every-day actions, and thus the verbs of very common 
 use generally work together in a short of co-partnership ; one furnishes 
 the aorist, another supplies the future, and so on: the verb signifying 
 "to go or come," for instance, has made up its tenses from four 
 
CHAP. 
 
 ΠΙ.] 
 
 THE MOODS AND PAETICIPLES. 
 
 651 
 
 different roots: present, ερχομα^ imperfect, jjsLV or ^a*; future, εϊμο; 
 aorist, ηλϋΌν, perfect, ηκω and οϊχομκί. The long list of defective 
 verbs, which we find in every Greek grammar, is a proof of the rich- 
 ness, not of the poverty, of the language. 
 
 423 Π. Sanscrit root tud-. 
 
 1st Definite Tense. 
 {Present Indicative). 
 
 Farasmaipadam. AtmanSpadam. 
 
 tudd-mi 
 
 tuda-si 
 
 tuda-ti 
 
 tudd-vas 
 
 tuda-thas 
 
 tuda-tas 
 
 tudd-mas 
 
 tuda-tha 
 
 tuda-nti 
 
 tude (for tuda-me) 
 
 tuda-se 
 
 tuda-te 
 
 tuda-vahe 
 
 tude-the 
 
 tude-te 
 
 tudd-ma-he 
 
 tuda-dhve 
 
 tuda-nte 
 
 1st Indefinite Tense. 
 {Imperfect Indicative). 
 
 Parasmaip adam. 
 
 a-tuda-m 
 
 a-tuda-s 
 
 a-tuda-t 
 
 a-tudd-va 
 
 a-tuda-tam 
 
 a-tuda-tdm 
 
 a-tudd-ma 
 
 a-tuda-ta 
 
 a-tuda-n 
 
 AtmanSpadam. 
 
 a-tude (for a-tuda-me) 
 
 a-tuda-thds 
 
 a-tuda-ta 
 
 a-tudd'Vahi 
 
 a-tude-thdm 
 
 a-tude-tdm 
 
 a-tudd-ma-ki 
 
 a-tuda-dhvam 
 
 a-tuda-nta 
 
 2nd Definite Tense {Perfect Indicative). 
 
 Parasmaipadam. 
 
 tu-t6da{m) 
 
 tu-tod-i-tha 
 
 tu-toda(t) 
 
 tu-tud-i-va 
 
 tu-tuda-thus 
 
 tu-tuda-tus 
 
 tu-tudi-ma 
 
 tu-tuda(tha) 
 
 Atmau§padam. 
 
 tu-tude (for tu-tuda-me) 
 
 tu-tud-i-sM 
 
 tu-tude (for tu-tuda-te) 
 
 tu-tud-i-va-he 
 
 tu-tudd-the 
 
 tu-tudd-te 
 
 tu-tudi-ma-he 
 
 tu-tudi-dhve 
 
 The Plus-Perfect, 
 or corresponding 
 2nd Indefinite 
 Tense, it want- 
 ing. 
 
 tu-tudus (for tu-tuda-nti) tu-tudi-re (for tu-tuda-nte) 
 
 * That ηρχόμην must be left to άρχομαι, and that the only imperfect 
 of ερχομαί is jjeiv or fjcc, is now a recognised fact. It was first pointed 
 out by Elmsley {ad Eurip. Heracl. 210), and Cobet has shown (Varice 
 Lectiones, 1854, p. 32), by an interesting proof, that we may safely restore 
 the true form in the few passages where the corruption is still retained. 
 For though there is no variation in the MSS. of Thucyd. iv. 121 : ίδια 
 δ' έταινίονν τε καΙ προαηρχοντο ώς άΟ'λητ^, we find in Julius Pollux, 
 III. 152 : ίΞίενοφών γάρ εΐρη-κεν εταινίονν τε καΐ ηροΰ'^εΰαν ωβπερ άϋ'λητ-η^ 
 where, although he writes Xenophon for Thucydides, he obviously refers to 
 the passage in question, and as our oldest MS. of Thucydides is not earlier 
 than the 14th century, we must prefer the reading which appeared in the 
 text in the reign of Commodus. The cause for the corruption has often 
 been the apparent identity of the forms γιμεν, ^τε, jjaav with those of the 
 substantive-verb. And in one case the scribe has not hesitated to sub- 
 stitute ήλ&ες for rjoQ-a, i. e. in ^sch. Ayam. 494, where the metre ne- 
 cessitates the true form of the imperfect. 
 
652 
 
 THE MOODS AND PAETICIPLES. [bOOK IV. 
 
 3rd Definite Tense, 
 {Future Indicative) wanting. 
 
 Parasmaipadam. 
 
 a-taut-sa-m 
 
 a-taus-si-s 
 
 a-taut-st-t 
 
 a-taut-s-va 
 
 a-taut-tam 
 
 a-taut-tdm 
 
 a-taut-s-ma 
 
 a-taut-ta 
 
 a-taut-sus 
 
 3rd Indefinite Tense. 
 {Aorist Indicative). 
 
 Atmanepadam. 
 
 a-tut-si (for a-tut-sa-me) 
 
 a-tut-thds (for a-tut-sa-se), 
 
 a-tut-ta and so on 
 
 a-iut-s-va-hi 
 
 a-tut-sd-thdm 
 
 a-tut-sd-tdm 
 
 a-tut-s-ma-hi 
 
 a-tud-dhvam 
 
 a-tut-sa-ta 
 
 It is doubtful whether the Potential, which follows, is the 4th 
 Definite Tense, or the 4th Indefinite Tense deprived of its aug- 
 ment (above, § 339). 
 
 
 Parasmaipadam. 
 
 
 Atmanepadam. 
 
 
 tude-ya-m 
 
 
 tude-yaime) 
 
 
 tude-s 
 
 
 tude-thas 
 
 
 tude-t 
 
 
 tude-ta 
 
 
 tude-va 
 
 
 tude-va-hi 
 
 
 tude-tam 
 
 
 tude-yd-thdm 
 
 
 tude-tdm 
 
 
 tude-yd-tdm 
 
 
 tude-ma 
 
 
 tude-ma-hi 
 
 
 tude-ta 
 
 
 tude-dhvam 
 
 
 tude-yu-s 
 
 
 tude-ran 
 
 5th Definite Tense 
 
 
 5th Indefinit Tense. 
 
 {Desiderative). 
 
 
 {Aorist Optative). 
 
 Parasmaipadam 
 
 Atmangpadam. 
 
 Parasmaipadam. AtmanSpadam. 
 
 tdt-s-yd-mi 
 
 tot-s-ye (for 
 
 a-tot- 
 
 -s-yam a-tot-s-ye (for 
 
 
 tot-s-ya-me) 
 
 
 a-tot- s-ya-me) 
 
 tot-s-yd-si 
 
 tot-s-ya-se 
 
 a-tot- 
 
 -s-ya-s a-tot- s-ya-thds 
 
 tot-s-ya-ti 
 
 tot-s-ya-te 
 
 a-toP 
 
 s-ya-t a-tot-s-ya-ta 
 
 tot-s-yd-vas 
 
 tot-s-yd-va-he 
 
 a-tot 
 
 -s-yd-va a-tot-s-yd-va-hi 
 
 tdt-s-ya-tJias 
 
 tot-s-ye-the 
 
 a-tot 
 
 -s-ya-tam a-tot- s-ye-tham 
 
 tot-s-ya-tas 
 
 tot-s-ye-te 
 
 a-tot 
 
 -s-ya-tdm a-tot- s-ye-t am 
 
 tot-s-yd-mas 
 
 tot-s-yd-ma-M 
 
 a-tot 
 
 -s-yd-ma a-tot- s-yd-ma-hi 
 
 tdt-s-ya-tha 
 
 tot-s-ya-dhve 
 
 a-tot 
 
 -s-ya-ta a-tot-s-ya-dhvam 
 
 tdt-s-ya-nti 
 
 tdt-s-ya-nte 
 
 a-tot 
 
 -s-ya-niti) a-tot-s-ya-nta 
 
CHAP. III.] THE MOODS AND PAETICIPLES. 
 
 653 
 
 
 Imperative. 
 
 Parasmaipadam. 
 
 tudd-ni 
 
 Atmanfipadam. 
 
 tudai 
 
 tuda 
 
 tuda-sva 
 
 tuda-tu 
 
 tuda-tdm 
 
 tudd-va 
 
 tudd-va-hai 
 
 tuda-tam 
 
 tudS-thdm 
 
 tuda-tdm 
 
 tude-tdm 
 
 tudd-ma 
 
 tudd-ma-hai 
 
 tuda-ta 
 tuda-ntu 
 
 tuda-dhvam 
 tuda-ntdm 
 
 Infinitive and Gerund. 
 tot-tum 
 tut-tvd 
 
 The passive is formed from the Atmanepadam by the 
 insertion of ya in the conjugational tenses. 
 
 424 III. Latin root tud- with anusvdra. 
 
 1st Definite Tense. 
 (^Present Indicative). 
 
 Active. Passive. 
 
 tu-n-do{m) tu-n-d-or (tundo-mer) 
 tu-n-di-s [tu-n-de-re] 
 
 tu-n-di-t tu-n-di-tur 
 
 tu-n-di-mus tu-n-di-mu\se\r 
 tu-n-di-tis [tu-n-di-mini or -minor] 
 tu-n-du-nt tu-n-du-ntur 
 
 1st Indefinite Tense. 
 Wanting. 
 
 2nd Definite Tense (Perfect Indicative). 
 
 Active. Passive. 
 
 tu-tudi {tutudsa-m) Wanting. 
 
 tu-tudi-sti (tutudsa-si) 
 
 tu-tudi-t (tutudsa-ti) 
 
 tu-tudi-mus^ &c. 
 
 tu-tudi-stis 
 
 tu-tude-runt 
 
 3rd Definite Tense (used under the form -aw, -es^ &c., as Future 
 Indicative ; under the form -am, «5, &c., as Present Subjunctive). 
 
 tu-n-dd-m tu-n-da-r (for tu-n-da-mer) 
 
 tu-n-dd-s (or -de-s) tu-n-da-ris (or -de-ris) 
 
 &c. &c. 
 
654 THE MOODS AND PARTICIPLES. [bOOK IV. 
 
 3rd Indefinite Tense {Imperfect Subjunctive). 
 
 tu-n-de-re-m tu-n-de-re-r (for tu-n-de-se-mer) 
 
 tu-n-de-re-s tu-n-de-re-ris 
 
 &c. &c. 
 
 The 4th Definite Tense would be formed in -rio^ if it existed in 
 this particular verb. 
 
 Imperative. 
 
 Active. Passive. 
 
 tu-n-de [tu-n- de-re] 
 
 tu-n-di-to tu-n-di-tor 
 
 tu-n-di-tote [tu-n-di-minor] 
 
 tu-n-du-nto tu-n-du-ntor 
 
 Infinitive and Gerund. 
 
 tu-n-de-re 
 
 tu-n-de-ri-er, tu-n-di-er, tu-n-di 
 
 tun-sum 
 
 tu-n-de-ndus 
 
 &c. 
 
 &c. 
 
CHAPTER IV. 
 THE CONJUGATIONS. 
 
 425 Differences of conjugation due to variety in the forms of the present 
 tense. 426 The Sanscrit conjugations, doubly classified. 427 Grimm's 
 arrangement of the Teutonic conjugations. 428 Analogy between the 
 German strong verbs and the principal Sanscrit conjugation. 429 Greek 
 conjugations. Objections to the «ordinary arrangement. 430 True 
 classification of Greek verbs. 431 I. Primitive verbs. (1) Reduplica- 
 tion. 432 (2) Addition of ya. 433 (3) Insertion of r- or v-. Case 
 of δίοτζτενω. 434 (4) Forms in &- and -σκ. 435 (5) Guna. 
 436 II. Derivative verbs. Their terminations. 437 Verbs derived from 
 compound adjectives. 438 Affections of the root-vowel. 439 (1) The 
 vowel permanent. 440 (2) Alteration of quality. 441 (3) Alteration 
 of quantity. 442 (4) Guna. 443 The simple aorist as a conjugatiorial 
 variety. 
 
 425 rpHE differences in the verbs of the Indo-Germanic 
 family, which have induced grammarians to divide 
 them into classes called conjugations, are produced entirely by 
 the various methods that have been adopted to give greater 
 strength to the root in the present tense and those dependent 
 upon it. The necessity for this addition to the root has arisen 
 in the wish to create an expression of duration, which, as we 
 have before remarked, is often effected by reduplication only. 
 Other methods of expressing it were brought forward under the 
 head of vocalization (§ 223): as, however, the subject of the 
 Greek conjugations has never been properly treated, we may 
 venture in this place to resume the question at greater length. 
 It will perhaps conduce to the better understanding of the nature 
 of those changes on which differences of conjugation depend, if 
 we begin by a brief survey of the Sanscrit and German systems. 
 
 426 The Indian grammarians divide their verbal roots into 10 
 classes or conjugations. They reckon the whole number of roots at 
 2352, to which they add 44 Sautra roots, or those which are taken 
 into the Sutrdni or grammatical rules for the purpose of deriving some 
 few nouns which cannot be traced back to any of the regular verbal 
 roots. The number of the distinct roots in actual use is much less 
 than that which the grammarians give; indeed not above 500 are 
 found in the existing Sanscrit writings. The distinctions of the 10 
 conjugation classes, which with the exception of the 9th class do not 
 extend beyond the present (indicative, imperative, and optative) and 
 imperfect (indicative), are of two kinds, first, vowel-changes, guna or 
 
656 THE CONJUGATIONS. [bOOK IV. 
 
 anusvdra; secondly, pronominal additions, na, 7iu, or ya. We shall, 
 therefore, give two classifications of the 10 conjugations; first, ac- 
 cording as they have pronominal additions or not; secondly, as they 
 have guna or anusvdra. 
 
 A. I. No pronominal addition. 
 
 a. Person-endings joined immediately to the root. 
 
 {2nd class, 70 roots, as pa-mi, "I rule." 
 3rd 20 roots — reduplication, as da-dd-mi, "I give." 
 7th 24 roots — anusvdra^ as bhi-na-d-mi, "I split," 
 
 (findo). 
 
 β. Person- endings joined to the root by the vowel a. 
 Γ 1st class, 1000 roots — gu7ia of root vowel, as bodhdmi 
 I (from budh\ "I know." 
 
 1 6th 130 roots — simple root, as tuddmi (from tud.) 
 
 II. Pronominal addition. 
 
 a. ya added to the root. 
 
 {4th class, 130 roots — mostly neuter verbs, as nag-yd-mi, 
 "I perish." 
 10th guna; mostly causals and derivatives, as mdna- 
 
 yd-mi, "I honour *." 
 
 β. nu added to the root. 
 
 5th class, 30 roots; guno of suffix, as dp-no-mi, "I ob- 
 tain." 
 
 8th 10 roots; all except kri, "to make," ending in 
 
 η or ή, probably adscititious in the other 9 
 roots, as ta-no-mi. 
 
 γ. na added to the root. 
 
 9th class — guna of suffix, which becomes ni before the 
 heavy endings, as mrd; mrd-nd-mi^mrd-ni-mas. 
 
 B. a. Both vowels guna'd. 
 
 1st and 10th classes, budha, bodhdmi; churi, choraydmi. 
 
 β. First vowel guna'd. 
 
 2nd and 3rd classes, dvi^ dveshmi; da, dadd-mi. 
 
 * The value of the distinction of vowels in Greek is shown, as' well 
 in other formations, as in the fact that the Sanscrit -ayumi includes the 
 three classes of vowel-verbs in -άω, ~έω, and -όω. 
 
CHAP. IV.] THE CONJUGATIONS. 657 
 
 γ. Second vowel guna'd. 
 
 4tli, 5th, 6th, 8th, 9th classes', gucJi-yd-mi ; dpnomi; tudd- 
 mi; ianomi; mrdndmi. 
 
 d. anusvdra. 
 
 7th class, yuj, yunajmi (Jungo). 
 
 427 Grimm divides the verb in all the German dialects into two 
 great classes or conjugations distinguished by the form of the preterite. 
 The first he calls the strong conjugation ; it forms the preterite from 
 the root without the addition of any foreign element. The second he 
 terms the weak conjugation ; it forms its preterite by the insertion of 
 the lingual d (in old High German t) between the root and person- 
 ending. Now it appeai# that the strong form is the original one in all 
 the German dialects: for all derivative verbs are conjugated according 
 to the weak form ; the roots which form the basis of the language are 
 confined to the strong verbs ; and though a strong verb may in the 
 course of time degenerate into a weak one, the converse never takes 
 place. Therefore, for the purpose of comparison, we may safely leave 
 alone the weak form of conjugation. The strong verbs are divided 
 into two classes, each consisting of six conjugations ; the first of these 
 leading subdivisions contains the verbs with reduplicated preterites, 
 the second , those in which the perfect is formed by vocalization only. 
 Two of the reduplicated conjugations (the 5th and 6th) adopt both 
 methods of forming the preterite; and as the vowel is the same in the 
 singular and plural preterite of the 7th conjugation, and as verbs 
 vacillate between this and the reduplicated conjugation, Grimm con- 
 cludes (i. p. 838) that the 7th must be supposed to have been origi- 
 nally reduplicative with a change of vowel like the 5th and 6th. 
 Grimm (i. p. 840) takes the following instances among others, from 
 the Gothic language ; he gives the first singular present indicative ; 
 preterite indicative singular; preterite indicative plural; and the 
 participle. 
 
 Reduplicated conjugation. 
 I. Salta (sdlio); sai-salt, sdi-saltum; saltans. 
 
 II. Haita (voco); hdi-hdit, hdi-hditum ; hditans. 
 
 III. Hldupa (curro); hldi-hldup (i), hldi-hldupum ; hldupans. 
 
 IV. Slepa (dormio) ; sdi-zlep, sdi-zUpum ; slepans. 
 
 Reduplicated with change of vowel. 
 V. Ldia (irrideo); Idi-lo, Idi-ldum; Idians. 
 
 VI. Greta (ploro); gdi-grot, gdi-grdtum; gretans. 
 
 UU 
 
658 THE CONJUGATIONS. [bOOK IV. 
 
 Change of vowel without reduplication. 
 
 VII. Fara (proficiscor) ; for, forum; farans. 
 
 VIII. Greipa (rapio); grdip, gripum; grip arts. 
 
 IX. Kiusa (eligo); kaus, kusum; kusans. 
 
 X. Giba (do); gab, gebum; gibans. 
 
 XI. JSHma (sumo); 7iam, nemum; numans. 
 
 XII. Ilil'pa (adjuvo); halp, hulpum; hulpans*, 
 
 428 These strong conjugations include 80 English verbs; the 
 second conjugation, however, has no representative in our language, 
 and all our verbs have lost the reduplication. All the primitive verbs 
 in the other Teutonic dialects are also formed according to the strong 
 Gothic inflexion. Now if we examine this form of conjugation, which 
 comprises 130 verbs out of the scanty remaina^f Gothic literature, we 
 shall find that it agrees with the first Sanscrit conjugation, which con- 
 tains nearly half of the Sanscrit verbal roots. Some few verbs present 
 traces of the fourth Sanscrit conjugation ; for instance, in the seventh 
 Gothic conjugation, we have vahs-ja (cresco), vash-ji-th (crescit); pre- 
 terite vohs, vohsum; participle vahsans, corresponding to the old High 
 German and old Saxon wahsu^ middle High German wahse, Anglo-Saxon 
 veaxe, English wax. The general analogy between the German strong 
 verbs and the principal Sanscrit conjugation was first pointed out by 
 Bopp {VergL Gramm. p. 115). There is, however, a remarkable dif- 
 ference between the strong German verbs and the first Sanscrit conju- 
 gation, which Bopp has not noticed ; namely, that in the first Sanscrit 
 conjugation, the connecting vowel is almost always guna'd in the first 
 person of the pres. , imperf. , and 1st pret., as well as the root-vowel, 
 whereas, in the German, the connecting vowel never admits of any 
 quantitative modification. There is yet another point which we must 
 not overlook, if we would make a profitable comparison between the 
 German and Sanscrit verb. We have seen that the Sanscrit verb has 
 guna as well as reduplication, so that it agrees with the fifth and sixth 
 strong conjugations of the German verb. But when the verb-root 
 consists of a short a, between two single consonants, only the first and 
 third persons singular of the preterite are reduplicated, all the other 
 persons being distinguished by a substitution of the diphthong e = ai 
 for the long a, and the first syllable being thrown ofi". Thus from the 
 root tan we have preterite first and third person singular tatana, se- 
 
 * The English reader will find the law according to which Grimm has 
 arranged the last six conjugations, very well stated in a paper in the Phi- 
 lological Museum (Vol. ii. pp. 375 foil.). The scheme of the vowels in all 
 the conjugations in Gothic and old High Herman is given by Grimm, 
 Deutsch. Gramm. i. 837. 
 
CHAP. IV.] THE CONJUGATIONS. 659 
 
 cond singular tenitha, dual teniva, tenathus, tenatus, plural tenima, 
 tena, tenus. Bopp thinks this i has arisen from assimilation of the 
 connecting vowels (Krit, Gramm. der Sanskr. /Spr.p. 224, comp. Vergl. 
 Gramm. pp. 40, 64). We do not see how this can be the case, for in 
 the second and third dual, at all events, the connecting vowel is not i 
 but a. We rather look upon it as a substitution for the reduplication, 
 of the same kind with that which takes place so commonly in Latin, 
 where reduplication is so seldom found : thus , we have capio , cejpi ; 
 venio, veni, &c. That these are substitutions for a lost reduplication, 
 may be inferred from a comparison of ρ epigi with compegij of fugi with 
 Ί^ίφενγα, &c. With regard to the distinction of the first and third sin- 
 gular in the Sanscrit preterite from the other persons , Grimm has 
 aptly compared the old High German and Anglo-Saxon singular, I. 
 las, II. Msi, III. las; plural, I. Idsumes, II. Idsut, ΙΠ. Idsun {Deutsche 
 Gramm. i. p. 1056). From this we infer that the last six of the Ger- 
 man strong conjugations were either originally reduplicated like the 
 first six, or that the strengthening of the vowel is to be considered as 
 perfectly identical with reduplication, as we shall see it is in Greek. 
 
 429 In the old Greek grammars (e. g. Dionys. Thr. Bekker. Aiiecd. 
 p. 638) the conjugations of the verbs are divided into three classes, 
 which are given in the following order; (1) those of the barytone 
 verbs , or of those which never admit an accent on tlie ending of the 
 present; (2) those of the contracted verbs in -εω, -«ω, and -οω; and 
 (3) those of the verbs in -μι. This classification is not scientific, and 
 cannot be recognised by the comparative philologer. We have before 
 shown that all verbs originally agreed in their person-endings with 
 those in -μι, as well from other reasons, as from the fact that those 
 which are still so conjugated express the most elementary notions, and, 
 therefore, must be considered as the oldest verbs. It is true that the 
 Greek verbs in -μυ are distinguished from the other verbs by certain 
 peculiarities, but they should be classed according to these peculiari- 
 ties , and not be discriminated from the rest of the verbs merely be- 
 cause they have their person-endings preserved in a more primitive 
 form. For the practical convenience of the learner the best arrange- 
 ment of the Greek conjugations is that which recognises two classes of 
 verbs, A. Primary Verbs in -^t, and B. Secondary Verbs in -ω; and 
 which subdivides class B. , according to the root or characteristic let- 
 ters , into {a) consonantal or semi-consonantal verbs ; and (b) vowel- 
 verbs, which admit of contraction (see Greek Grammar, Art. 299). 
 But comparative philology, having regard to the origination of the 
 forms themselves, demands a classification corresponding to that which 
 is adopted in the Sanscrit and Teutonic grammars : namely, according 
 
 UU2 
 
660 THE CONJUGATIONS. [bOOK IV. 
 
 to the difiPerent modes of strengthening the form of the present and 
 other principal tenses *. We cannot indeed divide the Greek conju- 
 gations into strong and weak forms, in the same way as the Gothic: 
 for, in the first place, all Greek verbs form the preterite by reduplica- 
 tions, whereas this is confined to the strong form in Gothic: and, con- 
 versely, all verbs form certain tenses by the addition of a foreign 
 element, which is the indication of the weak form in Gothic. We 
 may however say , that, in case the lengthened form of the present is 
 a consequence of its being a derivative or secondary verb, it cannot 
 form any of its tenses by an alteration of the quality or quantity of 
 the root-vowel, but must have recourse either to reduplication, addi- 
 tion, or both; and in this way it may be said that derivative or 
 secondary verbs in Greek are weaker forms. 
 
 430 One of the great mistakes which grammarians have com- 
 mitted in this department is, that they have considered the present 
 tense of barytone verbs as the primitive form of the root, and the other 
 tenses as derived from it. The consequence of this has been to load 
 our dictionaries and lists of defective verbs with an infinitude of so- 
 called obsolete forms, from which, forsooth, these tenses were derived. 
 The fact is, that the present tense of a barytone verb always shows the 
 root under some modification or increase, the genuine root being in 
 most cases preserved in the so-called second aorist. We proceed to 
 classify the verbs according to the various methods by which this cor- 
 roboration of the present is effected. And let us take (I) the primi- 
 tive verbs, i. e. those which are not derived from nouns or which do 
 not exhibit the root in any more simple form; (II) the derivative 
 verbs, i. e. those which are formed from nouns or from the other class 
 of verbs. 
 
 431 I. (1) In the simplest and apparently the oldest modifica- 
 tion, the root is immediately connected with the ending, the vowel 
 
 * The most recent treatise on the Indo- Germanic verb is that of Pro- 
 fessor Moriz Rapp of Tubingen {Der Verbal-Organismus der indisch-euro- 
 paischen Sprachen, Stuttgart und Augsburg, 1859, three volumes). This 
 work, which does not seem to have any original or independent value, is 
 merely a sort of collectanea in aid of the author's lectures. His theory 
 of the Greek verb in particular is professedly based on the grammatical 
 systems of Buttmann and Bopp. His enthusiastic appreciation of Buttmann 
 (Vol. II. p. 4) is quite in accordance with what we said long ago (above, 
 p. 64). With regard to the conjugations of the Greek verb, he is content 
 to adopt three main classes, (1) those which have no formative adjunct; 
 (2) those which have a formative vowel; (3) those which have a forma- 
 tive consonant. But he does not exhibit much analytical power, and fre- 
 quently confuses when he professes to distinguish. 
 
CHAP. IV.] THE CONJUGATIONS. . 661 
 
 being generally guna'd, and the initial consonant reduplicated, when 
 the root begins with a consonant. To this form belong the first three 
 conjugations of the verbs in -μι. There are cases in which we 
 find reduplication without guna, as in τίίτίτω for τίυ-τίεχω; in others 
 guna without reduplication, as in φη-μί; in others the roots are 
 joined to the ending without any change, as in εϋ-μί = 86-μί; and in 
 some we find anusvdra of the reduplication, as in τΐί-μ-πλημι. All 
 these verbs may be compared with the second and third Sanscrit 
 conjugations. 
 
 432 (2) A very large class adds to the root the pronominal 
 syllable (second element) ί,α= Sanscrit ya. These correspond, therefore, 
 to the fourth and tenth Sanscrit conjugations. In Greek, however, 
 the addition is absorbed, and represented only by a variety of assimi- 
 lations, as is often the case when the same syllable is used for the 
 purpose of forming a comparative degree. The following are the 
 transformations under which this adjunct appears in the verb-con- 
 jugation. 
 
 a. When the verb-root ends in a liquid, the l is frequently placed 
 before the liquid, by virtue of the fact, before mentioned, that the 
 liquid sounds its adjacent vowel indifferently before or after; thus 
 from the crude-form (jita-Qi- we have (Σπείρω for ΰτίάρι/αμί; from φα- 
 VL-, φαίνω for cpάvyaμL, &c. Or the liquid is doubled ; thus from ψα- 
 λυ- we have φάλλω for 'φάλραμί, like aXkog for aliog^ &c., and from 
 κα-ρ- we have both κέρρω and κείρω, from φΟ'α-ρ- both φΟ'έρρω and 
 φ^'είρω, and so forth. Or a short c in the root coalesces with the t 
 of the adjunct and becomes long ; thus from κρϊ- we have χρΐνω == 
 κρϊνι/αμι,. 
 
 β. When the root ends in j^, κ, ;^, or in ^, r, ^, this pronominal 
 adjunct is represented, as when it appears under the same circum- 
 stances in comparatives, by J, (>(?, or ττ, of which we conclude that 
 ξ is always the primitive change (above, § 216). 
 
 γ. The adjunct ya also appears to be contained in most of the 
 verbs in -εω , -άω. With regard to the latter , which often occur as 
 verbs in -ψμί, little need be said', as we have already shown in more 
 than one instance that the sound ya is included in η. We have also 
 had examples of the substitution of ε for y in the middle of a word : 
 that this is its use in the case of the verbs in -εω appears from the 
 Boeotian forms αγωνο%ετίον%οξ, είλαρχυόντων, Ο'ίΟΤΐροΛίοντος, χορα- 
 γίοντεξ, &c. ; for άγωvo^'ετyovτog, or in common Greek αγωνο^ετουν- 
 tog, &c. (Bockh, Corpus Inscript. i. p. 720). There are some verbs in 
 
662 THE CONJUGATIONS. [BOOK IV. 
 
 -«05, -εω, which must not be considered as containing the adjunct ya: 
 such are άρά-ω = ό'ρ«-Ρω , which must be connected with δρα7ΐ-= 
 ^ερατΐ- (cf. δρατΐέτης) ; καλέω = καλέσω connected with %lsfog, 
 κλνω, &c. ; cf. |εο9, ξ,νω, ξίφος, &c. ; βρνω, βρέφος, &c. 
 
 433 (3) Another pronominal adjunct is the element τ or v, 
 which, we have seen, are identical, the former being added to roots 
 ending in |3, 7t, φ, or κ, the first and third of which are, of course, in 
 this combination , changed to jc , and the liquid ν being adopted for 
 roots which end with the liquid μ, or with the dental r. We some- 
 times also find ν after -κ. Thus from the roots κρνκ-, τντΐ-, ραφ-, and 
 τεκ-, we have κρντί-τ-ω^ τντί-τ-ω, ράτί-τω, τίκ-χ-ω; ταμ-^ ττειτ-, and 
 δακ-, make τέμ-ν-ω, τΐίτ-ν-ω, and δάκ-ν-ω. This ν is also added to 
 vowel-roots, as in τίί-νω. It appears also under the longer form νη, 
 as in the 9th Sanscrit conjugation, in δάμ-νψμι, &c. ; under the form 
 vv^ as in the 5th Sanscrit conjugation, in ζενγ-νν-μυ (root ζνγ- , al- 
 ready strengthened by guna\ &c., and in this form the ν is often doubled, 
 as in (5βέ-ν-ννμι, χρώ-ν-νυμυ, ύκεδά-ν-νυμι (above, § 220), and 
 sometimes under the ordinary short form it is added to a root already 
 strengthened by anusvdra, as in λα-μ-βά-νω , root λαβ-, κυ-γ-χά-νω 
 (also written κυχανω), root κοχ-, λυ-μ-τίάνω (also written with a simple 
 guna, as λείπω), root λίτί-. The change from ίκανω to ίκ-νε-ομαι is 
 in accordance with the general principle with regard to liquids, already 
 so often mentioned. 
 
 There is one verb , with regard to which it may seem doubtful 
 whether there is an insertion of this r- or not. "We allude to the 
 Attic διοτίτενω, in the signification "to be the δίοτίος, or supercargo 
 of a ship" — "one who sailed in her as manager," also called the ετίί- 
 λωυς. The following are the passages in the grammarians referring to 
 it. Harpocration : δίοτΐτενων, ζίημοΰ&ένης εν τω εΙς την Αακρί- 
 του. δίοτίος λέγεται νεώς 6 δάτίων καΐ εποπτεύων τα κατά την 
 ναϋν, 6 κα%'' ημάς λεγόμενος επίπλους . Hesychius: δ ίο π ο ι. έπιμε- 
 ,ληταί. — άδίοπον. αναρχον κα\ άφνλακτον. Αισχύλος, Φρνξ,ιν. 
 δίοποι γαρ ot της νεώς φύλακες, ^lius Dionysius: δίοπος δ ναν- 
 φνλαξ ώς επιοκοπών αντην καΐ εφόρων άπο τον οπτω• Erotianus : 
 δίόπω, τω της νηος επιμελητή, πάρα το διοπτενειν. 'Αττική δε ί 
 λέξις κειμένη και παρ' Άριβτοφάνει εν Άττικαϊς λέξεΰι και παρ' 
 Αΐύχνλω εν Σιύνφω και Ευριπίδη εν Ίππολντω, It is quite clear 
 that the connexion of δίοπος with οπτομαι is merely a fancy of the 
 grammarians. As a political term δίοπος is naturally and immediately 
 referred to διέπω, the use of which is quite in aocordance with that of 
 its derivative. Thus , as we have βαΰιλης δίοποι of two generals in 
 iEschylus (Pers. 44) ; we have also in the same play (v. 108) πολέμους 
 
CHAP. IV.] THE CONJUGATIONS. 663 
 
 τΐνργοδαΐκτονς ddjtBLv; and Pindar (according to Strabo , p. 544 b), 
 
 φηβΐν OTL at Αμαζόνες, 
 
 Σνρίον ενρναίχμαν διεΐτΐον ΰτρατόν. 
 
 Now there is, no doubt, a word δίότίτης or δίοτίτήρ, which is connected 
 
 with οτΰχομαι, and means "a spy," as in Homer, Iliad, x. 582: 
 τόν Qcc δίΟΛτηρα στρατοί) εμμεναί ημετεροιο 
 "Ετιτωρ tB τΐροεηκε κοίΐ αλλοί Τρώες αγανοί — 
 
 where διοπτηρ ΰτρατον is a very diiferent person from δ δίοτίος 
 
 ότρατον, and from this δοότίτης we haye a verb δίοτΐτενω, "to be a 
 
 spy," as in Kom. Iliad, x. 451: 
 
 η τε καΐ νΰτερον εί6^α ^οας επΙ νηας Αχαιών, 
 ηε διοΊίχεύβων , ^ εναντίβιον πολεμίξων. 
 In Sophocl. Ajaa;, 307, και πλήρες ατής ώς διοτίτενει ΰτέγος^ it means 
 simply "to perceive," just as we use the verb "to espy." There 
 would, therefore, be some ambiguity if δ ίοτίτενω were also formed from 
 δίοΛος. And as δίοπτενων in Demosthen. Lacrit. p. 929/ΐΛΉ:ίας 
 Ά^'ηνίτΐπον ^Αλικαρνα6(3ενς μαρτυρεί ΰνμτΐλεΐν εν τγ/'^Τβληβίον νηΐ 
 δίοητενων την νανν, does appear to be connected with δίοτΐος and 
 not with δίότίτης, it would be better, w^ think, to suppose that the 
 r has got in, in consequence of the false etymology which the gram- 
 marians have adopted, and therefore to read διοτίενων. 
 
 434 (4) The two pronominal insertions -%"- and -ΰκ-, which we 
 have seen performing such important functions in the tenses and voices 
 of the verb, also appear as corroborants of the present tense, though 
 generally with a distinct meaning; thus from the root ε^- we have 
 εύ-^ί-ω^ from φα-, φά-ΰκω, &c. These additions often affect the final 
 consonant of the root, as in τίάόχω from %α%'-^=^7ίεν%^- ^ the κ being 
 aspirated and the ^ lost in the similar sound of the (j; so also we have 
 δίΰκος from δίκ-; and the word ΐΰκε (Odyss.xix. 203, xxii. 31), which 
 clearly means "he said" (Buttmann, Lexil. ii. p. 83), is to be con- 
 sidered as a similar contraction from Γί7ί-ΰκε^=^εΪ7ίεϋκε, the root being 
 Ι^ετΐ-, Sanscrit vach., Latin voc-: a converse contraction has taken 
 place in ενιύπον for ενιτί-ΰκον: ϊ(5χείν, ΰχεΐν, from 'εχεόκον are more 
 analogous to τίάΰχω (see above, § 219). 
 
 435 (5) The remaining method of strengthening the present in- 
 dicative is by simple guna, which, as we have seen, consists in placing 
 e before the root-vowel. This method is only applicable to those verbs 
 in which the root-vowel is t or ν : instances are , however, sufficiently 
 numerous; thus, from the root 6τφ- we have ΰτείβω; from φνγ-^ 
 φεύγω; from λίτζ-, λείτίω; from λίχ-, λείχω; from Ttid'-, τζεΐ^ω; from 
 
664 THE CONJUGATIONS. [bOOK IV. 
 
 τνχ-, τενχομαί, also, with anusvdra and ν adjunct, τν-γ-χά-νω ; from 
 At/3-, λείκω; from Ttvd"-, τίενΟΌμα^ &c. 
 
 436 II. The derivative verbs are formed from their primitives 
 by the addition of elements, some of which are used also for the mere 
 purpose of strengthening the present tense in the primitives, from 
 which, however, they are distinguished by the way in which these end- 
 ings are joined to the crude-form, and by their possessing only those 
 of the secondary tenses which can be constructed by extrinsic addition, 
 namely, the first aorist active and passive and the perfect active in -κα. 
 The terminations are -έ-ω, -ά-ω, -ό-ω,-ά-ξω,-ί-ξω^ -ί-ΰκω, -λλω=λίω, 
 -εν-ω,-αί-νω,-ν-νω; as φιλ-ε-ω ΐτοτηφίλο-ς^τίμ-ά-ω ίτοπιτίμή^μίΟ- 
 Ο'ό-ω ίνοτημίύ%Ό-ς^ 6κευ-ά-ξω from ύκενή, νομ-ί-ξω trom νόμο -g, γαμ- 
 ί-0κω from γάμο -g, ίΐ^ά-λλω from φάω, τίαώ-εύ-ω from τίαΐξ (τΐαΐδ-ς), 
 ύημ-αί-νω from 6ημα=ΰήμεντ, εν&-ν-νω from ενΟ'ΐ^ς^ the termi- 
 nations -έ-ΟΌ), -ν^'ω, seem to be appropriated to derivatives from 
 simple verbs , for instance φλεγ-έ-&ω from φλέγω , φ&ίν-ύ-Ο'ω from 
 φ&ίνω: the ν in the latter case is due to the vv, which seems to be 
 added to the root φΰ'ί-, to form the present tense. The terminations 
 -ι>-ΰκ-, -ν-όκ-Ί are inserted between the root and the strengthening 
 pronominal adjunct v- in the verbs 6φλ-ι-6κ-άνω, άλ-υ-6κ-ά-νω-, αμ- 
 βλί-ΰκ-ά-νω. It is obvious that these terminations are of pronominal 
 origin , and we have already discussed most of them. The derivative 
 verbs themselves are, for the most part, either causative, inchoative, 
 or denominative, and they obtain these meanings from pronominal 
 affections of the primitive forms , which seem to be strictly analogous 
 to the case-endings of the noun : for example, we cannot overlook the 
 resemblance between -ay a, which forms the Sanscrit causative verb, 
 and the same syllables, which mark the dative case in that language 
 (§ 231). In Greek it is clear that the denominative verbs in -ξω= 
 -δya-μi are built on the foundation of the genuine ablative case, which 
 gives rise also to the analogous patronymics in -δης, and to the ad- 
 jectives in -diog (§ 247) : and we are justified in considering the verbs 
 in 'έω, -άω, -όω, -ενω, as similarly formed from the shortened geni- 
 tive, like the corresponding adjectives: cf. %QVuog, gen. χρνΰοΐο with 
 χQv6εog and χρνόόω (above, § 298). By the side of verbs in -λλω 
 and -ΰκω we have diminutives in -λog and -ύκog; thus βάλλω =βα- 
 λί'ω, "to cause to go away," may be compared with ^'ρvλλog from 
 Q'QOOg, and ηβά~ΰκω is analogous to τΐαίδί-βκη. We have already (p. 
 457) adverted to the connexion of such words as τίΟίμήν, &c. with 
 the corresponding verbs Λοιμαίνω , &c. It very often happens , as in 
 the case oi 6ημα=ΰημεντ- and ΰημαίνω, ενΟ'ν=ενχΙ'ν-τ and εν^'ύνω^ 
 that the noun of agency is no longer extant under the simple form in 
 
CHAP. IV.] THE CONJUGATIONS, "j 665 
 
 -v, but has to be reproduced under some longer form, such as ύημάν- 
 τωρ, ενϋ'νν-τηρ; but this will not prevent us from perceiving that 
 the full form of ματ- is μεν-τ, and that we have in the verb , as well 
 as in the noun, that combination of the first and third elements under 
 the form μ-\-ν, which expresses the action as proceeding from the 
 subject, and with especial reference to its results (§ 256). When the 
 objective ν or t appears alone it is difficult to say what precise relation 
 is implied, though we know from the parallel case of the nouns that 
 there must be some reference to a special objectivity (§ 255). When 
 we pass from fs/oj to ξίφος , and from ξνω to ξν-λον, we see the in- 
 strument in its objective expression', and we cannot mistake the same 
 fact as manifested in ξαίνω and ξανΟ'ός, by the side οίξον^ός and ξεΰτός. 
 The common verb αΐνέω merits particular consideration, even after 
 what Buttmann has written about it (Lexih ii. pp. 112 foil.). There 
 can be no doubt that the simplest form of the verb was αϊνω, as Hesy- 
 chius tells us: αϊνω v. βαρντόνως, ετΐαινών η. From this was formed 
 the substantive αίνος, and from it the derivative verb αΐνε-ω, and the 
 derivative noun αίνη=^αϊνρα (Herodot. m. 74). The primary significa- 
 tion of αΥνω is "to say" or "declare," and the meaning of alvog in 
 Homer is '^a speech" or "narrative." From this came the sense of 
 "praise," "commendation," just as the Latin laudare is connected with 
 lauty loud. That the verb is formed with the pronominal suffix v-, 
 and that its root is aL-, is proved by the comparison of αϊνω with mo, 
 and of ai-(5a with fa-tum, which Buttmann has suggested. We also 
 recognise the root in ημι^ "I say," and in the Sanscrit aha, "I said," 
 In these last two forms the guttural semi-vowel is absorbed; it is 
 transposed in at-, according to the rule, and perhaps represents the 
 digamma, the labial element of which appears in φψμί, for, according 
 to Thiersch, αϊνη is a digammated word. As the derivative αΐνεω 
 bears the secondary sense of "to praise," to extol," we might, per- 
 haps, change the accent, and read αϊνω, αϊνεις, in those passages of 
 JEschylus where the general and primary meaning is conveyed: thus, 
 in the Agamemnon, 98, we read 
 
 τούτων λέξαΰ' [1. λέξον -9^'] οτί καΐ δυνατόν 
 καΐ Ο'έμι,ς αΐνείν, 
 
 " of these things declare , as much as it is possible and right to men- 
 tion." And in the same play, 1460, we have 
 
 η μεγαν οϊκοίς τοΐόδε 
 δαίμονα κάί βαρνμηνίν αινείς. 
 φευ, φευ! κακόν αΐνον άτψ 
 ράς τνχας άκορέΰτον. 
 
 i. e. "you mention a divinity who has exerted great influence upon 
 
QQQ THE CONJUGATIONS. [bOOK IV. 
 
 this family, and has manifested his heavy indignation against it — an 
 ill-boding mention of misfortunes ever new" (for the force of άκό- 
 ρεΰτος see above, p. 554); and in the Choeph, 1000, we have 
 
 φόνον δε κτ^κΐ^ ξνν χρόνω ξνμβάλλεται, 
 
 τίολλας βαφας φΰ'είρονύοί τον ποί%ίλματοζ — 
 
 νυν αντον {scil. τον φόνον) αΐνώ^ ννν άποιμώζω τζαρών, 
 
 τίατροκτόνον Ό"' νφαΰμα τζροΰφωνών τόδε. 
 
 437 In the case of primitive words the terminations are joined 
 immediately to some root, even though the contact may absorb the 
 final consonant of the root , but in these derivatives the endings are 
 always affixed to some crude-form, and therefore, of necessity, a vowel 
 is inserted between the root and the termination. This enables us to 
 explain, why verbs, including a noun, or the a- privative, or dvg-, or 
 BV, and a verbal root, are generally formed in -ε-ω. The fact is, that 
 while such verbs as άτΐο-δίδωμι, ϋνν-τρεχειν, τΐαρα-λαμβάνειν , &c., 
 are strictly separable compounds, as is shown by the frequent tmesis in 
 the older writers, all compound nouns, whether made up of prepositions, 
 or of nouns, or of ά-, δνς-, or ev, and verbal roots, are actually melted 
 down into individual words incapable of divulsion, and it is from these 
 compound nouns that the verbs in question are formed; therefore they 
 are derivative verbs, and the length of the word would generally in- 
 duce a necessity for the shortest kind of derivation, which is in -ε-ω. 
 If a Greek had wished to express the idea of conferring a benefit on 
 any one, or of co-operating with him in a particular action, he would 
 say εν τίΟίεΐν, ϋνν-ερδειν^ but if he wished to express habitual bene- 
 faction or habitual co-operation, he would take the compound nouns 
 εν-εργετη-ς and 6νν-εργο-£ and make derivative verbs from them — 
 εν-εργετ-ε-ω 1 "to be a benefactor," and 6νν-εργ-ε-ω , "to be a co- 
 operator." It will be observed that we have precisely the same phe- 
 nomenon in Latin : from facer e , an uncontracted verb , we have the 
 compound noun Iceti-ficus, and from this the derivative or contract 
 verb Iceti-ficare; from gerere we have belli-ger, and from this belli- 
 gerdre. "With regard to the Greek verbs it seems very strange that 
 even modern scholars should talk of deriving them from the second 
 perfect of the barytone verbs, the roots of which they contain (see 
 Erfurdt ad Soph. Antig. 56; Lobeck ad Phrynich. p. 580), especially 
 after Scaliger had , with his usual penetration', discovered the truth. 
 Nemo Hellenismi paullo peritior^ says that great scholar {Phrynich. 
 p. 266 Lobeck), concedet, εναγγελλω grcecum esse. Nam το εν καΐ τα 
 ΰτερητυκα μόρια ηοη componuntur cum verbis., sed cum nominibus. Ita- 
 que ενάγγελοξ recte dicitur^ unde verbum εναγγελεω, ηοη εναγγελλω 
 
CHAP. IV.] THE CONJUGATIONS. 667 
 
 quod est absurdissimum *. But although in general the weight of the 
 compound nouns recommended the shortest form of derivation for the 
 verb, there is no absolute reason why one of the other derivative forms 
 should not be occasionally adopted, when there was any particular 
 reason for the preference. Such a reason seems to exist for the assump- 
 tion of άημάξω instead οίάτυμεω; for, τιμάω being itself a derivative, 
 a similar derivative would hardly point to the distinction of τιμγι and 
 άτιμος. The word άτίει which occurs in Theognis (621) is justified 
 by the opposition to tlbl: 
 
 πάς rig τΐλονΰοον άνδρα tlbl, άτίεϋ δε τίενιχρόν. 
 
 The word χερνίτιτεΰ^^αι is probably derived from χερνί'φ considered 
 as a simple word, like χαλεπτω from χαλετίός (Buttmann, Ausfuhrl. 
 Sprl. § 121); δνύΟ'νήΰκων, which occurs twice in Euripides (Bhes. 
 791; Electr. 843), is used only as an epithet or adjective, and might 
 be explained by the metrical impracticability of the legitimate δνύ^α- 
 νατεω, though if it had been in one of the other dramatists (see Miiller, 
 Hist. Lit. Gr. i. p. 483), we should have expected any form, — δνΰ^νψ 
 τέω for example, — rather than such a violation of all analogy ; and in 
 the Here. Fur. 863, it is clear that ΰταδίοδραμονμαι is a false reading, 
 probably due to the gloss ΰτάδοα δραμονμαυ which is found in Flor. 2, 
 and we ought to restore the genuine ΰταδίοδρομηΰω of which these 
 words are an explanation (see Lobeck, Phryn. p. 617): for ^ra^to^^o- 
 μεω actually occurs , and it is not unlikely that an ignorant copyist, 
 knowing by habit that δραμονμαι was the future of τρέχω , has bar- 
 barized the word into the form in which it appears. Some of the 
 
 * The necessity for some "contributions towards a more accurate know- 
 ledge of the Greek Language" in this country was signally proved in the 
 last months of 1857, by a controversy waged in the "Times" and other 
 newspapers as to the validity of the form τηλέγραμμα. It is believed that 
 no doubt on the subject was entertained by any really good scholar, but 
 graduates" of Oxford and Cambridge, laying claim to academical distinc- 
 tions, strenuously asserted in print that the barytone τηλεγράφω would 
 have been an allowable form ! The analogy of χειρογραφέω, τηλεβολέω, &c. 
 shows what the form would have been, and the two synonymous nouns 
 χειρόγραφον and χειρογράφημα prove that the telegraphic message might 
 be expressed by either τηλέγραφον or τηλεγράφημα; and while τηλέβολος 
 in the classical writers is generally used as an epithet of the missile dis- 
 charged from afar (Pind. Pyth. iii. 49: τηλεβόλω χερμάδι. Mnasalcas, 
 Anth. Pal. 6, 125, 3: τηλέβολος ιός), it sometimes means the instrument, 
 as in Anth. Pal. Appendix, ix. 97, we have τηλεβόλου ρντηρα of an ar- 
 cher; and the medieval Greeks had no better name for the cannon and 
 the gun than τηλέβολος and its diminutive τηλεβόλΐΰ-κος (Laonicus Chal- 
 condyles, p. 72 ed. Bekker). So that τηλέγραφος might signify the in- 
 strument or machine, while τηλέγραφον or τηλεγράφημα would denote 
 the message. 
 
668 THE CONJUGATIONS. [bOOK IV. 
 
 other instances in which this rule seems to be violated have been suc- 
 cessfully corrected by modern scholars. 
 
 438 Having now shown by what increments of addition or inser- 
 tion the present may be strengthened , we proceed to point out the 
 relation which subsists between the root-vowel and that which appears 
 as its locum tenens in the present tense. This subject has been already 
 touched on in the chapter on the roots : it will, therefore, be sufficient 
 in this place to give instances of the change of vowel in the Greek 
 verb according to an arrangement first pointed out by Pott (Etymol. 
 Forsch. I. pp. 11 foil.). This scholar has divided the Greek verb into 
 four classes according to the affections of the root-vowel in the lead- 
 ing tenses ; (1) the root preserves the same vowel throughout all the 
 inflexions 5 (2) the quality of the vowel is altered; (3) its quantity is 
 altered, generally by doubhng; (4) it is guna'd. The second aorist 
 active, middle and passive, generally exhibits the root , and the prin- 
 cipal changes are those of the second perfect , or noun containing the 
 verbal root which agrees with the second perfect, and the present. 
 The reason for the change of vowel in (2), (3), (4) is, as we have seen, 
 the greater weight of the perfect and present in consequence of the 
 methods adopted for strengthening them. The present is generally a 
 heavier form than the perfect or derivative noun , and, therefore, has 
 the lighter vowel. It will be observed that there are some verbs, placed 
 by Pott in the 3rd class, which are examples of guna: to prevent 
 mistakes we have always stated the method of corroboration adopted 
 in the particular case. 
 
 439 (1) This class is very numerous, as it contains all the 
 weak or derivative verbs', many of which have already received a 
 vowel modification in their crude-form. Thus from the root Ttsv- we 
 have 7tovo-g and from this τΐονέ-ω , in which the first vowel remains 
 unaltered through all tenses. The most obvious instance of the 
 primitive verbs of this class is second aorist ε-τντί-ον , second perfect 
 th-tV7t-a, verbal substantive τί-χντί-Οζ, present τντί-τ-ω. 
 
 440 (2) In this class are included verbs with a primitive a (or 
 aQ=^r Sanscrit) changed into ο and ε in the strong tenses. It agrees 
 with the 11th and 12th of the strong German conjugations, in which 
 the root-vowel is followed by a liquid, or a mute and liquid, or pre- 
 ceded by a liquid (conj. 11), or followed by a liquid and mute, or 
 a double liquid (conj. 12). This class is separated by Pott into four 
 subdivisions, the second and third of which we consider identical. 
 
CHAP. IV.] 
 
 THE CONJUGATIONS. 
 
 669 
 
 A. Roots ending in a liquid. 
 
 2nd Aorist. 2nd Perfect or Verbal Noun. 
 
 utoXo-g 
 
 ίβε-βόλ-ψμοα \ 
 
 8-ΰτάλ-ην 
 
 Εβαλ-ον 
 
 Β-ταμ-ον 
 
 /Β-ταμ-ον \ 
 \ταμ-ία-ς ί 
 καν-εΐν 
 
 8-7ίταρ-ον 
 
 ε-φ^^άρ-ην 
 
 έ-ύτΐάρ-ην 
 
 ε-δάρ-ην 
 φαρε-τρα 
 
 \βολή 
 
 τόμΟ'ζ 
 
 κονή Hesych 
 
 Λτόρο-ς 
 
 ε-φ^ορ-α 
 (ε<5Λορ-α ι 
 ι ύτίορά ί 
 
 φόρος 
 
 Present. 
 
 βτέλλω (adjunct ιά) 
 βάλλω (id.) 
 
 τεμ-ν-ω (adjunct ν) 
 
 καιν ω (adj. t«) 
 ί Ίίτάρ-νν-μαι (adj. νυ) \ 
 Ιπταίρω (adj. lo) ί 
 
 φ^'είρω (id.) 
 
 Οτίείρω (id.) 
 
 δείρω (id.) 
 
 φέρω (adj. absorbed). 
 
 Β. Roots ending in liquid and mute; the liquid of course may- 
 
 shift its place. 
 
 2nd Aorist. 
 
 ε-δρακ-ον 
 ε-Ίίαρδ-ον 
 ε-7ίρα^-ον 
 ίε-τρατί-ό-μηνΧ 
 ^ ε-τρατί-ον ^ 
 ε-τραφ-ον 
 ε-κλάτΐ-ην 
 
 2nd Perfect or Verbal Noun. 
 
 δε-δορκ-α 
 
 τΐέ-τνορδ-α 
 
 πτολί-7ΐορ^Ό-ς 
 (τροφή ) 
 ^τετροφ-α ί 
 
 τετροφα 
 
 κε-κλοφ-α 
 
 C. Roots without liquids. 
 
 2nd Aorist. 2nd Perfect or Verbal Noun. 
 
 Sanscrit had κέ-χοδ-α 
 
 ε-τεκ-ον 
 
 ίε-Λεό-ον 
 ε-Λετ-ον 
 Sanscr. pat 
 ε-^φεγ-ην 
 
 τετοκα 
 
 Λοτ-μο-ς 
 
 ipoyo-g 
 
 Presont. 
 
 δερκ-ο-μαο 
 
 περδ-ω 
 
 τΐερΟ'-ω 
 
 ίτερτιω χ '^' 
 Χτρετιω \ 
 
 τρέφω 
 
 κλετί^τ-ω (adj. τ). 
 
 Present. 
 
 χέζω (adj. lo) 
 τίκ-τ-ω (adj. τ-) 
 ίτΐί-τίτ-ω (redupl.) 
 ι τατ-νέω (adj. νε) 
 
 -φέγω 
 
 441 (3) This class comprehends verbs having a for their root- 
 vowel; this vowel is generally doubled or guna^'d in the perfect and 
 present, so that these verbs agree with the seventh strong German 
 conjugation. 
 
 * Pott considers τέρπω and τρέπω as belonging to diiferent classes; 
 we think that they are the same word, and regard them as but slight 
 modifications of τρέφω. 
 
670 
 
 THE CONJUGATIONS. 
 
 [book IV. 
 
 2nd Aorist. 
 
 ^-κραγ-ον 
 
 ϊ-%λαγ-ον ι 
 
 ε-7ΐάγ-ην 
 ^-λακ-ον 
 'ε-τακ-ον 
 ε-λαχ-ον 
 
 'ε-λαβ-ον 
 
 ϊ-δακ-ον 
 ε-φάν-ην 
 
 2nd Perfect. 
 
 κε-κράγα (yuna) 
 κε-κληγα {guna) \ 
 
 κε-κλαγγα (anusvard)) 
 ηε-πηγα {guna) , 
 
 λέ-λάκα (guna) 
 τέ-τηκ-α (guna) 
 εϊληχα (guna) 
 
 εϊληφα (guna and affix) 
 
 δεδηχα (guna and affix) 
 τΐέφηνα (guna) 
 
 Present. 
 
 κράξω (adj. la) 
 
 κλάξω (adj. ta) 
 
 τίήγ-ννμι (adj. vv) 
 λάβκω (adj. οκ) 
 τήκω (guna) 
 λαγχάνω (adj. ν, and 
 
 anusvara) 
 λαμβάνω (adj. v, and 
 
 anusvara) 
 δάκ-ν-ω (adj. v) 
 φαίνω (adj. la) 
 
 Pott subjoins to this class a number of perfects such as πεφρικα, 
 'ερρϊγα, βεβρϊ^α, τετρΐγα, &c. , aorists such as ε^'ίγον, κρίκε, δεκεΐν, 
 εκικον, &c., but they are, in our opinion, quite different. The ν in 
 πετίον^-α belongs to the root, which is πα^-=^7ίεν^- (above, § 114). 
 
 442 (4) In the last class we find verbs which have t or i; for 
 their root- vowel, and this is guna'' ά. in the present and perfect, the 
 ^wna-vowel being generally changed according to the law observed in 
 class 2. This class corresponds, therefore, to the eighth and ninth 
 strong German conjugations. 
 
 2nd Aortst. 2nd Perfect or Verbal Noun. 
 
 ϊ-6τιχ-ον 
 ϊ-λΐ7ί-ον 
 
 6τοΐχο-ς 
 λέλουπα 
 
 ε-ύτίβ-ην 
 
 λιβά-δ- 
 
 Ιδ-εϊν 
 
 ΰτοφή 
 
 λοιβή 
 
 οίδα 
 
 'ε -md'-ov 
 
 Λετίοι^α 
 
 ε-φυγ-ον 
 ε-ξνγ-ην 
 
 τίεφενγα 
 ξενγοξ 
 
 ε-τνχ-ον 
 
 τετενχα 
 
 ε-ηνΟ'-ό-μην Λεν^'ώ 
 
 Lat. stud-eo ΰτΐονδή 
 λνγ-ρό -g λενγ-α-λε -og 
 
 Present. 
 
 ΰτείχω 
 
 λείτίω 
 
 6τείβω 
 
 λείβω 
 
 εΧδω 
 
 τίεί^ω 
 
 φεύγω 
 
 ξενγννμί (adj. νυ, as well 
 
 as guna) 
 τυγχάνω (adj. ν, and^ 
 
 anusvara I 
 
 τενχω (guna) j 
 
 τΐενδομαί Λ 
 
 Λνν%άνομαι (adj. ν, I 
 
 and anusvara) J 
 
 ύτίενδω 
 lugeo 
 
CHAP. IV.] THE CONJUGATIONS. 671 
 
 443 In giving to the augmented tense, which presents the root 
 of the verb in its simplest form , the name of second aorist , we ac- 
 knowledge the difficulty of determining in every particular case, whe- 
 ther this tense is really a second aorist, correlative with the second 
 future, and produced by an evanescence of the affix <j- , or whether it 
 is a conjugational variety, that is, the imperfect of a semel-factive verb, 
 which never had any affix. The unaffected form in which the root 
 appears would lead to the latter conclusion, and , not to speak of the 
 semel-factive tenses in the Semitic languages, the Sclavonic languages 
 give us different classes of verbs distinguished as semel-factive or 
 monologous (in Polish iednotliwe) and frequentative or iterative (in 
 Polish GzestotUwe) ; so that this conjugational variety would not be 
 without precedent. If we look through any list of so-called second 
 aorists, in which the person-endings are joined immediately to the 
 root without even the intervention of a short vowel, or any other 
 fulcrum beyond guna (see Kuhn, De Conjugatione in -μι^ p. 60; Gr. 
 Gr. Art. 315, n.), we shall see that in most cases it is nearly impos- 
 sible to say, whether the form is not due to some absorption or apocope 
 of an affix. At any rate , it is clear that the shortest of these forms, 
 and those too which we should most confidently refer to a conjuga- 
 tional variety, are identical in use with the regular aorist. This is 
 shown in a striking manner by the fact , that in common verbs like 
 τίϋ'ημί, ΐημί, δίδωμί, &c., the first aorist (ε%'ηκα, ηκα^ έδωκα) is used 
 only in the singular of the indicative mood, the shorter and apparently 
 unaffected forms (^Ο'εμεν, είμεν, ϊδομεν) being substituted in the plural 
 and dual indicative, and in the other moods and participles. 
 
CHAPTER V. 
 THE USE OF AUXILIARY VERBS IN GREEK, 
 
 444 Auxiliaries are generally verbs expressive of will and choice. 445 Differ- 
 ences between the infinitive and participle in relation to the auxiliary. 446 De- 
 finition of an auxiliary verb. 447 Use of σωκώ; 448 Οί έ^έλω and μέλλω. 
 449 Reasons for examining the Greek verbs which signify will or desire. 450 
 General enumeration of results. 451 Αάω signifies "to wish," "to take," and 
 "to see." 452 Verbs derived from λάω bearing the last sense. 453 Verbs sig- 
 nifying "to take." 454 Proper meaning of βλάπτω. 455 Λεία, λέων, λάας, &c. 
 Idea of selection implied in Aaag. Ancient buildings of picked stones. 456 
 Derivations from λάω signifying "I wish." 457 Digression concerning α'ίγλη. 
 458 Etymology of this word. 459 Άγάλλω, αγλαος, γελάω, &c. 460 Con- 
 nexion of "light" and "speech." 461 Other analogies: άΰελγής and αιγα- 
 λοείς. 462 Etymology of ϋ'έλω compared with ϋ-έναρ and ^άλκΰβα. 463 
 Distinction of -ό-ελω and |3ου/Ιο^αί. 464 ΈλεΓν, f λκ£ iv, 'άπάϋ'έλγειν. 465 Θέ- 
 λεμοζ. 466 Etymology of βονλομαι. 467 The forms βονλήΰομαι, ηβουλή- 
 Q -ην, «&c. 468 The intensive prefix βον-. 469 Βοννός, &c. 470 The earth» 
 and the cattle. 471 Θυμός: its Platonic use; 472 Its etymology. Μάω. 
 Θοάξω. 473 Θεός, &οός, &c. 474 ^εαπότης. Alb ^εαπόοιος and δεαπο- 
 ΰίοναντης; explanation of a chorus jin jEschylus. 476 Θύμος and d'V- 
 μέλη. 477 'Οργή, ρέω, &c. 478 'Απορροή applied to colours, &c. 479 
 Other applications of οργή. Words denoting kingly power. 480 Mean- 
 ing of the phrase έττιφέρειν οργάς. 
 
 444 "ΙΤ/ΉΕΝ a verb in some finite tense is prefixed to some 
 
 VY 
 
 other verb in the infinitive, so that the two taken 
 
 together form one notion, and then combination is equivalent to a 
 periphrasis of some tense, we say that the finite verb is used as an 
 auxiliary. The verbs most frequently employed in this manner 
 are those expressive of the condition or power of the agent, of 
 his freedom from external hindrances, whether moral or physi- 
 cal, of his thoughts, intentions, will, or desire. That the employ- 
 ment of auxiliaries originated in the substitution of syntactical 
 contrivances for the etymological inflexions of the older language, 
 after these last had fallen into disuse or had become less obvious 
 and significant, has been already shown at some length. In most 
 of the languages of modern Europe , the system of auxiliaries 
 has superseded all inflexions of mood and tense, and, indeed, of 
 voice too, in the common verbs. In our own language, some of 
 the verbs employed for this purpose have lost their applicability 
 as independent verbs: thus, we never say "to may," or "to 
 can," or "to shall," or "to must," though "I may," "I can," 
 
CHAP, v.] USE OF AUXILIABY VEEBS IN GEEEK. 673 
 
 "I shall," "I must," are in constant use, as auxiliaries, for the 
 purpose of forming the potential mood, or the future tense, or an 
 expression of obligation. In German, however, the verbs cor- 
 responding to these, namely, mogen^ konnen, soUen, miissen, are 
 capable of inflexion throughout all their moods and tenses with 
 the exception of the participle present. It has been well remarked 
 by Max MtiUer (Oxf. Essays 1856, p. 38), that "our auxiliary 
 verbs have had to pass through a long chain of vicissitudes before 
 they arrived at the withered and lifeless form which fits them- 
 selves so well for the purposes of our abstract prose." And it is 
 a very vivid anticipation of their subsequent use as mere adjunclfe 
 for the expression of tense or time, when we find the verbs denot- 
 ing growth and obligation appearing in the names of the three 
 Fates, who were called Vurdh, Verdhandi^ Skuld, i. e. past, pre- 
 sent, and future time (Zeitschr.f. vergl. Sprf. iii. p. 449), whereas 
 in Greek the Destinies are represented as goddesses, who deter- 
 mine the fixed lot of the past (Lachesis)^ spin the thread of 
 present life (ClotJio)^ and decide unalterably what shall and must 
 come into being (Atropos)^ (see Plat. Resp. p. 617 C: Αάχεΰίν 
 μεν τα γεγονότα, Κλω^'ώ δε τα οντά, 'Άτροττον δε τα μέλλοντα). 
 
 4^4:6 We have said that the verb which the auxiliary modifies, or 
 which forms the pivot of the meaning, is in the infinitive mood. That 
 the auxiliary could not be so used with the participle is obvious. The 
 infinitive is, indeed, as we have shown in a previous chapter, the same 
 sort of word as the participle, but then it is a participle which has 
 lost all power of change, being a form, sometimes mutilated, of the 
 locative case, so that it expresses the locus of the action to which the 
 possibility or will denoted by the auxiliary has immediate reference. 
 The participle, on the contrary, being capable of inflexion, expresses 
 an hypothesis or some subordinate relation of the verb from which it 
 is derived, the subject being the nominative case of the principal verb; 
 in fact, participle and verb are equivalent to two verbs joined by a 
 copulative conjunction, and there is no more reason why the verb 
 should be considered as auxiliary to the participle, than the participle 
 to the verb. There are, indeed, some cases where the Attic idiom em- 
 ploys the participle of the verb on which the whole force of the sen- 
 tence depends, while the finite verb plays the part of a qualifying 
 adverb; as in the following instances -(given by Buttmann, GriecMsche 
 Grammatik, § 1 14, Anm. 8, and § 1 50) ; τυγχάνω : ώς δε ηλ%Όν, ετνχεν 
 αταών, "when Ι came, he went away directly''^ (schoolboys in this 
 
 XX 
 
674 USE OF AUXILIAEY VERBS IN GREEK. [bOOK IV. 
 
 country would be told to translate this "he happened to go away;" 
 but τυγχάνω implies hitting a mark, coincidence, especially in point of 
 time, just as we say in lowland Scotch, "I happened upon him," for 
 "I fell in with him," and the young student should be taught to 
 translate it "directly," "on the spot," "at the moment," or, in some 
 cases, "precisely," as in Isocrates, Areopagit. p. 140c: Ιγω δ\ 81 αντα 
 ταύτα τυγχάνω δεδίώς^ "it is precisely on account of these things," or 
 " on account of these very things, that I am afraid"). — λαν^'άνω: ταύτα 
 TtOiTjoag, sXad'BV ϋτίεκφυγών^ "after he had done these things, he got off 
 unobserved." — φΟ'άνω : εφ&αΰα αυτόν Ή:αρΒλ%'ών^^^ϊθΒ,τίϊΒ earlier than 
 he did," ουκ εφ^Ί^μεν ελ%'όντες, καΐ vodoig ελήφ^'ημεν, "we 7ΐο sooner 
 came than we were taken ill." — διατελώ: διατελεί τΐαρών^ "he is βΖ- 
 waystheγe.'' — χaίQεLv: χαίρουΰιν ετΐαινοϋντες, "they praise ^Ζαί?/?^." 
 The last word may be used in a finite tense with the participle of the 
 verb to which it refers. Thus we have not only ου χαίρων εττΐ ^όγοιβί, 
 δεννάύεΐξ ε με (Soph. Antig. 758), " you shall not abuse me with im- 
 punity," but also ούτοι χαιρη^ετον (Aristoph. Equit. 235), "you two 
 shall not escape unpunished." It will easily be seen, that in all these 
 cases the verb connected with the participle cannot be considered in 
 any intelligible sense as an auxiliary. It might be supposed, however, 
 that the use of εχω with a participle of the aorist, of which Valckenaer 
 has given so many examples {ad Phoeniss. 712, p. 310), is a peri- 
 phrasis for the perfect; and indeed, it is used in connexion with the 
 perfect in the line of ^schylus, which he quotes: ποικίλωκε, κάτίοδη- 
 λώΰαζ ϊχει. We believe that it is not merely equivalent to the perfect 
 (still less, as Valckenaer suggests, to the aorist), but that it has a 
 stronger sense than the perfect, expressing not only a state consequent 
 upon an action, but also a continuance in that state; thus, ^'ανμάύας 
 εχω means " I keep wondering," "I continue in a state of wonder " (Gr. 
 Gr. hll)' In this same sense εχω is used in the participle where we 
 should employ an adverb: thus, τι ληρεϊζ έχων; "why do you talk 
 nonsense continually ί" or "why do you keep talking nonsense?" The 
 only case in which a periphrastic tense is really formed by means of 
 the participle is, when the finite verb signifies "existence," that is, 
 when it is a substantive verb; but in that case the difference is not 
 one of tense, unless the participle be itself future, but of voice : compare 
 ich werde gelobt, "I am praised," with ich werde loben, "I shall praise." 
 In French and Italian, the verbs avoir, avere, " to have," form a future 
 tense with the infinitive and a perfect with the participle of the verb : 
 compare j^aimer-ai^ io amer-b (§ 368) with j'ai aime, io ho amato. 
 But it must be recollected that these modern languages do not use the 
 participle with such refinement of meaning as the ancient Greek, and 
 in this particular case the passive participle is used as an adjective 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GREEK. 675 
 
 agreeing with the object of the verb. In French, the past participle 
 and the object are in the same gender and number when the object 
 precedes, though the participle is not declined when the object follows: 
 thus, they sajj'ai regu voire lettre, but void les lettres que j^ai regues. 
 The principle has been fully explained by Du Marsais and other French 
 grammarians. In old Italian, the past participle is frequently in con- 
 cord with the object of the verb even when it follows; thus Dante: 
 
 Ed un cK avea perduti amho gli orecchi, 
 and Boccaccio : Messer lo geloso s^avea messe alcune petruzze in hocca. 
 The future participle may often be translated by the infinitive, but 
 this does not make the verb an auxiliary any more than leg at ο s misi 
 qui peterent is a periphrastic future. Even in phrases like οΛερ fja 
 ερών the use oi ερών is distinct from that of the infinitive, though we 
 should translate it in French or English, ce que fallois dire, or "as I 
 was going to say." The student who understands why φαίνεται είναι 
 means "it appears to be" (videtur), and φαίνεται ov, "it manifestly is," 
 i. e. " it appears so, and it is so " (apparet), need never feel any diffi- 
 culty in discriminating between the syntactical uses of the participle 
 and infinitive mood. 
 
 446 All verbs expressing power, will, &c., are, according to the 
 idiom of the Greek language, immediately followed by an infinitive 
 mood without any repetition of the subject of the primary verb when 
 this is also the subject of the infinitive: το γαρ βονλομαι φιλο- 
 λογεΐν, τζροαιρονμαι άναγιγνώβκειν ον δεΐται της Λροΰδια- 
 ύτελλονύΎΐζαντωνυμίαζ, ετίεΐ δυνάμει εν αντοπα^είαεχειτα της <3νν- 
 τάξεως, ε%εϊ τοι πάλιν ην 6 λόγος τοιοΐπ;ος, βονλομαι εμαντον πλου- 
 τεΐν, βονλομαι εμαντον Ίίερι%ατεΐν, rofr£(9'rt διατίΟ'εμαι εις το Λερι- 
 τίατεΐν, εις το πλοντεΐν (Apollonius Dyscol. de Syntax, πι. 32, p. 285 
 Bekker). It is not, however, to be concluded that every word so used 
 is use as an auxiliary. The finite verb is to be considered as an auxiliary, 
 and its combination with the infinitive as a periphrastic tense, only 
 when the two together express some one notion which might be ex- 
 pressed by an inflexion of the latter verb, and when the former is in 
 consequence used with a highly abstract meaning, much vaguer than 
 the sense which it bears when uncompounded. Thus, to take the words 
 expressing power or qualification, in such phrases as οΙός τε είμι τΐοιεΐν 
 τοντο^ ταντα πεφνκεν επϊ την ^υχην ίέναι, ϋρετίων εφνς τιρό τώνδε 
 φωνεΐν, τλαίης κεν Μενελάω ετίΐτίροεμεν ταχνν Ιόν, the finite verb 
 expresses only the abstract ideas, "I can," — "these things are suited," 
 — "you are qualified," — "dare you," — being quite subservient ;to and 
 dependent upon the verb in the infinitive mood ; so that these construc- 
 tions are entirely analogous to the Sanscrit periphrastic perfects, which, 
 
 XX2 
 
676 USE OF AUXILIARY VERBS IN GREEK. [bOOK IV. 
 
 though formed of a finite verb and the infinitive mood of another verb, 
 are equivalent to and are written as single words : thus, igam-hahMva 
 is written as a single word, though it is analogous in form as it is 
 equivalent in signification to αρχευν τΐεφνκα. 
 
 447 One of the Greek verbs expressing power or possibility is 
 
 deserving of some particular notice, as well from its rareness as from 
 
 its etymological connexion with a very interesting Sanscrit auxiliary. 
 
 •We refer to όωκεω, "to be strong," a derivative from ΰώκοζ, which 
 
 appears in Homer, Iliad, xx. 72, as an epithet of Hermes : 
 
 ΑψοΙ δ' άντεΰτη ύώκος Iqlovvlos Έρμης. 
 The verb occurs in two passages only; as an independent verb in 
 -3Eschylus, Eumenides, 36 : where, an old priestess, in great trepida- 
 tion, says: 
 
 η δείνα λέξαι, δείνα δ' δφΟ'αλμοΐς Ιδεΐν, 
 ηάλιν μ εΛεμίΙ^εν εκ δόμων των Αοξίον, 
 ώξ μήτε ύωκεϊν μήtε μ ακταίνειν βάΰον, 
 τρέχω δε χερΰίν, ον ποδωκία βκελών. 
 δείΰαύα γαρ γρανς οϋδεν άντίτίαις μεν ούν. 
 That ΰωκώ here is equivalent to ρώνννμαυ is clear from a passage of 
 Homer, which ^schylus ό φιλόμηρος had in his head when he wrote 
 these lines. In a case, precisely the converse of this, an old woman is 
 thus described as running to tell good news (Odyss. xxiu. init.) : 
 
 γρηνς δ' εΙς νΛερω άνεβήβατο καγχαλόωΰα, 
 δεΰτίοίν^ ερεονΰα φίλον τΐόΰον ένδον εόντα' 
 γοννατα δ' ερρώΰαντο, τίόδες δ' νπερακταίνοντο. 
 
 (The editions Υββ^άντΐερίκταίνοντο, an inexplicable word — the emenda- 
 tion we have introduced is supported by Hesychius, and the meaning 
 of the passage; see Ruhnken ad Timceum^ p. 20.) It must be clear 
 to every one that the third line in the first of these passages is the 
 negation of the third line in the second. But <5ωκώ occurs as a mere 
 auxiliary in Sophocles, Electra, 120: 
 
 μοννη γαρ αγειν ονκετι ΰωκώ 
 
 λύτίης άντίρροτίον αχ%•0£, 
 which Porson neatly translates " for I am no longer able by myself to 
 draw up the weight of grief which is in the opposite scale." It is in 
 this use that we are to compare ΰωκώ with the Sanscrit root gak, 
 which is not only the most general word signifying abihty or strength, 
 but also an auxiliary possessing this peculiar power, that, when pre- 
 fixed in the passive to an active infinitive (the Sanscrit language has 
 no passive infinitive), it gives the verb with which it is joined a passive 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GREEK. 677 
 
 signification, as in the Nalas, xx. 5, w« dhartun gak-ya-te^ wHcli is 
 as if you were to say in Latin afferre nequitur for afferri nequit; see 
 the examples of this idiom quoted by Lubker, de Participiis Grcecis 
 Latinisque^ p. 29, note; and compare the Latin infinitive future pas- 
 sive amatum iri (Bopp, Glossar. Sanscrit, ed. 1, p. 171). The root 
 gak is perhaps found also in the Irish ceach-t^ "power," and in go-na- 
 ri for coc-na-ri (Bopp, Gloss, ed. 2, p. 341). 
 
 448 The verbs ε&έλω and μέλλω are used more distinctly as auxi- 
 liaries than any others in the Greek language. They properly express 
 a will or intention, but are employed by the best authors to form peri- 
 phrastic futures, in which they merely modify the time of the verb 
 which accompanies them in the infinitive mood. Thus fO'fAco is used 
 in speaking of inanimate objects, as in Plato, Eespubl. ii. p. 370 b: 
 ονκ εΟ'ελξΐ το τίραττόμενον την τον πράττοντος ΰχολην τιεριμενειν, 
 where the sense of will or intention is so entirely lost that the gram- 
 marians consider it as equivalent to δνναμαι (see Gregor. Corinth, de 
 Dialect. Attica, § Lxvn. : Άττικον καΐ το ^ελει άντΙ του δύναται, 
 ώξ Πλάτων, "τά χωρία ουδέν μ ε&έλεί διδάβκειν'' (Phcedrus, ρ. 
 230 D)j άντΙ του δύναται)', and μέλλω is so constantly used in form- 
 ing the future tenses of verbs, that ό μέλλων χρόνος is the regular 
 name for future time, and τα μέλλοντα for future or expected events. 
 
 449 We purpose to devote the remainder of this chapter to an 
 etymological investigation of the most important Greek words de- 
 noting will or desire. It is a matter of some interest to point out 
 their primary meanings and their connexion with one another, and, 
 as it would be difficult to do this in a satisfactory manner without the 
 aid of the principles which we have endeavoured to establish and 
 explain in the preceding pages, a detailed examination of the question 
 will not perhaps form an inappropriate conclusion to this work. 
 
 The words which we shall more particularly examine in this place 
 are λάω, Ο'έλω, βούλομαι, μάω, ^υμός, and οργή, all expressive of 
 will, desire, or intention. As the discussion will be long, and encum- 
 bered with references and remarks upon cognate words, it will be as 
 well, for the sake of clearness, to state beforehand some of the general 
 results to which the investigation will lead us. 
 
 450 It has been mentioned in a former chapter (above, §§ 266 sqq.), 
 that there are two roots, la- and ra-, corresponding as weU etymo- 
 logically as in signification, which may be traced back to an identity 
 with the pronominal element na. The primary meaning of these 
 elements, in their use as particles or terminations, is — motion in a 
 
678 USE OF AUXILIARY VERBS IN GREEK. [bOOK IV. 
 
 given direction (above, §§ 130, 169, 204, 270, and elsewhere). From 
 this meaning results the idea of taking or seizing an object, just as 
 the preposition με-τά signifies both '^following after" and "companion- 
 ship" (§ 181). The meanings "to look at" and "to desire" are also 
 secondary ones, which association has attached to the root, but which 
 it could not by itself express in any strong or decided manner. We 
 have endeavoured to show, on a former occasion (above, § 169), that, 
 if the element -ra is subjoined to any pronominal stem, it denotes 
 motion or continuation in a line of which the first point is indicated 
 by the particular pronominal word. If, then, this element were appended 
 to the first or second pronominal stems, ma, Fa, it would denote pri- 
 marily a motion or emanation from the subject, or from that which is 
 near to the subject. Of the former combination we have the follow- 
 ing examples. The synonyms με-λος and με -Qog both express " divi- 
 sion," "separation of any object into its parts." Now, if we resolve 
 the sentiment or notion of division into its ultimate elements, we shall 
 find that it is reducible to the idea of a line proceeding from the 
 divider and cutting another line, the position of which is fixed. It is 
 for this reason that the adjective signifying "middle" is formed by the 
 adjunct ya from mat, the ablative of the first personal pronoun, which 
 case of itself denotes emanation, or proceeding from (§ 247). This 
 adjective is in Sanscrit madhya, in Latin med-ius. That the Greek 
 μεΰΰος was originally μεύ-ιοξ for μεδ-ιος, we have shown above, from 
 the analogy οίϊδ-ιοξ (§ 166); to which we may now add, that the form 
 μεΰίος at least is presumed in the derivative με6υτης(Β6β above, §259). 
 The primary meaning of medius, &c., "is the quality of that which 
 proceeds from the we." Sometimes the proposition διά, which sig- 
 nifies penetration or division into two parts, is prefixed not only to the 
 adjective μεβΰος, medius, as in ημυΰν^ dimidius (above, § 166), but 
 also to the substantive μέρος, as in ήμερος (above, § 150), in order to 
 express more strongly the idea of " bipartition " or "a passage;" cf. 
 meri-dies=:medius dies. For further coincidences in meaning between 
 these elements, we may compare μελ-λευν, μελέ-τη, μερι-μνα, μάρ-τυς-, 
 μερ-μηρ-ίξευν, me-mor, mora, &c., with meditari, mederi, re-medium, ^ 
 modus, moderari, &c., in which the idea of "thinking about or pro- 
 viding for any thing" predominates. Nor ought the connexion of 
 μέδί-μνα with μερι-μνα to be overlooked. For the meaning of ^«ρ- 
 rvg*, me-mor, &c., we may compare the poetical ενδατεΐ6%'αί, which 
 means literally " to divide," but is generally used in the signification 
 
 * The Etym. M. connects μάρτυς directly with μκρη = χειρ — μάρτυς ο 
 μάριρας καΐ είδώς το άλη&έξ. See Lobeck, ψηματίκόν, ρ. 8. With this view 
 we might compare the Latin mani 
 
CHAP, v.] USE OF AUXILIAEY VERBS IN GREEK. 679 
 
 "to commemorate" or "make mention of," as in -^schyliis {apud 
 Platon. in Eepubl. u. ad fin.) : τον Άτΐόλλω ενδατεΐβΟ^αι rag sag ev- 
 Ttaidiag (above, §178). If we now turn to tlie second pronominal ele- 
 ment we shall observe the following facts. This pronoun, in its oldest 
 element, is Fa, or it consists of a guttural and labial, articulated, of 
 course, by means of a short vowel. In combination with the element 
 -la or -ra, it always bears a meaning derivable from one analogous to 
 that of the similar formation with the first element : i. e. the primary 
 meaning oiFa-ra is "motion from the near,^'' as ma-ra means "motion 
 from the here.'''' It will be seen at once that these two meanings may 
 present many points of contact. We have indicated in a former 
 chapter, the many coincidences of the roots %α-ρ-, Sanscrit Jiri, and 
 Ρα-ρ-, Sanscrit vri (above, § 258, and elsewhere). The investigation, 
 on which we are about to enter, will render it probable that they are, 
 after all, only by-forms of the same root /a, the former representing 
 the guttural, the latter the labial element of the initial digamma. 
 "We have already discussed so thoroughly the secondary meanings of 
 these roots, that it will only be requisite in this place to collect the 
 forms in which they occur. With meanings intimately connected the 
 guttural element appears in ;υείρ, Xcc-Qug, xag-Ttogj aiq-Biv, ελ-εΐν, 
 grei-fan,yh'-SLV, can-is, hin-than^ hand; the labial in vin- star, fan- gen, 
 fin-ger, πεμ-ηε, pan-chan, five ; and both elements in gvan, gpan, κνων, 
 hun-d. In the sense of seeing and taking, borne by la by itself, and 
 occasionally with an additional fa subjoined, we have β-λέ-τίω, β-λά- 
 7ί-τω, y-kav-KOg, γ-λή-νη, &c. In the sense of willing or wishing, also 
 borne by -la, we have, from the guttural element, Ο'έ-λ-ευν for ελεΐν, 
 as ^άλαϋΰα from <5άλαΰΰα=αλα6ΰα, &c. And here the compound of 
 -la, -ra, with the second root, presents a most remarkable contact to 
 the same combination with the first; for as we have γεν -to, han-d, &c., 
 in connexion with ελεϊν, so we have Ο'εν-αρ^ "the hollow of the hand," 
 in connexion with Ο'έλ-ω, and manus, "the hand," in connexion with 
 μάρη=χείρ (Schol. Venet. Iliad.xv. 37, whence εv-μccQrjg=εύχεQηg); 
 and the words Ό'ερο^, αερίζω, express the idea of "smoothness," "cut- 
 ting down that which is sticking up " (namely, standing corn), just as 
 these same ideas are conveyed by the more general words μav-ρόg, 
 ά-μa-λ6g^i and ά-μά-ω (above, § 218). Finally, it will appear that as 
 the root la or ra by itself, or with the suffix fa, expresses motion in 
 general, and particularly the emotion of desire, so the first and second 
 elements under the forms ma and sa or tha, with or without the suffix 
 fa, express the very same ideas — motion in general, and desire in 
 particular ; as will appear from an examination of the verbs μάω and 
 %'ύω ; and thus it will be seen, that the ideas of will or desire, when 
 expressed by verbs (with the exception of βονλομαί, which is a totally 
 
680 USE OF AUXILIAEY VEBBS IN GEEEK. [bOOK IV. 
 
 different case), are resolvable into pronominal elements or words 
 significant of position, just as we have shown in the case of vagas and 
 εκα (§ 275). 
 
 451 We begin with λάα?, which means not only "to wish," but 
 also "to take," and "to see." How λάω can signify both "to have," 
 and "to want," has been already explained on general principles 
 (§ 53), and the analogies of capio and cu2nOj havere and habere, gestire 
 and gerere^ are also so many illustrations of it in this particular case ; 
 it will be recollected too that Λλεονεκτης, Λλεονεκτεΐν, τΐλεονεξία, imply 
 not only having more, but covetousness or desire to possess more (see 
 Herodot. vii. 149, 158, viii. 112). The two meanings "take" and 
 " see " will cause no difficulty to any one who recollects that the Ger- 
 man tragen, "to carry," and trachten, "to look at," are by-forms of 
 the same root, and that percipere, "perceive," is formed from capio, 
 "take." The connexion of "looking at," and "longing for," is ob- 
 vious, and the German word sehnsucht, which expresses an earnest 
 desire, is an instance of it. These meanings, "see," "take," and 
 " wish," run through a large class of word containing some modifica- 
 tion of the root λα-, by a series of transitions perfectly similar to those 
 we have remarked in the words belonging to the root χα-ρ-, and the 
 transitions may be explained in precisely the same way by a reference 
 to the principle of the association of ideas. This class is indeed a very 
 numerous and important one, for it contains all the old Greek words 
 beginning with λα-, λε-, or λί-. In the verb λάω itself, the succession 
 of meaning is, we conceive, first, simply "to see," then "to take," and 
 thirdly, "to wish." In Homer it is doubtful if it bears any meaning 
 but the first. In the Hymn to Mercury (v. 360), where we have 
 αίετος ό|ι> λάων, it evidently means " seeing." The word occurs twice 
 in the description of the cloak of Ulysses (Odyss. xix, 229), where it 
 is rather uncertain whether it means "to hold" or "to look at." 
 Either meaning would suit the context; Passow takes the latter; we 
 incline to the former. The words are as follows: 
 
 εν τΐροτεροίΰϋ πόδεύΰί κνων εχε τΐοίκίλον ελλον 
 
 άΰτΐαίροντα λάω ν το 8ε %'ανμάζε6κον άτίαντεζ 
 
 ώς οί χρύΰεοί oVreg, ό μεν λάε νεβρον άτίάγχων, 
 
 αϋταρ ό εκφνγεειν μεμαωζ ηΰτίαίρε τίόδεύΰιν. 
 
 We think the last two words favour the former interpretation : ό κνων 
 
 είχε τον νεβρον εν τΐροτέροίΰυ τΐόδεΰΰί, λάων αντον άύτΰαί- 
 
 ροντα, ο δε νεβρος ηΰτΐαίρεν εν τίόδεύΰί τον κννός. Α curious 
 
 confirmation of this view is furnished by the relation between the name 
 
 ofLais and her symbolical monument: Pans. ii. 2, 4: τάφοζ Ααΐδος, ω δ7^ 
 
 λέαινα ετίί^'ημά εΰτι κρίον εχονύα εν τοΐζ τίροτεροΐξ ito^iv. 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GREEK. 681 
 
 452 The words connected with λαω, in the sense of "seeing," are 
 β-λέπω, λενύβω, γ-λήνη, γ-λαν6ΰω, λαμτίρός, and λευκός. Hesychius 
 quotes λενΰευ (βλειί^ε^ Ο'εωρήύευ), apparently as the future of λάω in 
 this first sense; we might conclude that it is merely a mistake for 
 λενΰύεί, because in the next article he explains λενΰετε by οράτε, βλε- 
 τΐετε; but it appears from the Scholiast on Homer, that Aristarchus 
 considered it a future (see Alberti's note, 19). From the forms ajco- 
 λαν-ω=^άΛο-λα-μ-βάνω, β-λεΛω,Άηά λενΰΰω, we are inclined to infer 
 that the root was generally strengthened by the element pa, that is to 
 say, there is a secondary root λα -f- (formed of this root and the suffix 
 /a), which enters into the words in question : if so, λ«Ρω bears the 
 same relation to Χεν(5(5ω, that AaP«g, another word of the same family, 
 bears to λευζ, its synonym, and is related to β-λετΐω just as λά^ας is 
 related to the Latin lapis. The same may be said of γ-λανΰύω, λαμ- 
 πρός, and λευκός. 
 
 453 The suffix Fa also accounts for the labials which so often ap- 
 pear in words of this family bearing the second signification "to take." 
 Thus we have λα-μ-βάνω (where the μ is an euphonic insertion by way 
 of anusvdra, as in λα-μ-Λρός), λατΐ-άξω, λαφ-ύΰόω, λαύ-ρα, λαβύριν- 
 ^Ός, and ατίολαύω. The forms λεγ-ω, "to pick up," " select," "take one 
 by one," "utter articulate and continuous words," and its derivative 
 λεχος (properly, "a bed made up of gathered or picked leaves"); the 
 by-form λεβχη, a "speaking place," for λεγ-ΰκη (above, § 219), λάΰκω, 
 λάκω, "to speak," άδολέύχης, " a great speaker ;" also λα-γ-χ-άνω, "to 
 receive by lot," and its substantive λάχος ; and λάξομαί^ " to take hold 
 of;" must be regarded as containing the same root difi'erently modified. 
 
 454 A word more nearly connected with λάί^ω, λα-μ-β-άνω, 
 though we are not accustomed to vieAV it in that light, on account of 
 the β which is prefixed, is β-λάτίτω-, with its derivatives β-λάβη, &c., 
 which are perfectly analogous to λαβή, &c. Another reason for our 
 disregarding the connexion between β-λάτίτειν, λά/ω, and λαμβάνω, 
 is, that we attend' only to the derived sense of the former word, and 
 disregard its proper and original meaning. The word β-λάτΐ-τω does 
 unquestionably imply, in many cases, a certain degree of harm or mis- 
 chief, but even where it bears this sense, it is a sort of hindrance or 
 accidental harm, some mere pain or loss, that is denoted, and never an 
 injury of that kind which can cause resentment (see Butler's vmth 
 Sermon). In fact, this distinction is frequently pointed out in the 
 best Greek writers. In the following passage of Thucydides, for in- 
 stance (i. 71), where the Corinthian ambassador says to the Spar- 
 tans : οϊεΰ^'ε την ηβυχίαν ου τούτοις tov άν&ρωΛων επί πλεϊύτον 
 
682 USE OF AUXILIAKY VEEBS IN GEEEK. [bOOK IV. 
 
 άρκεΐν^ dt av xy μεν Λαραύκευ]] δίκαυα τΐράΰΰωΰι, ry δε γνώμ^, ην 
 άδυκώνταυ, δήλοι ώΰί μη ετατρέ-φοντες, αλλ" επί τω μη λνπεΐν τε 
 αλλονξ και αντοί αμυνόμενοι μη βλάτίτεϋ^'αι το ϊύον νέμετε — the 
 distinction between άδικεΐβ^αι and βλάτίτεύ^αι is clearly shown by 
 the opposition of δίκαια Λράύΰειν to the one, and μη λνπειν to the 
 other. This passage is considered a difficult one, and has been mis- 
 interpreted, we think, by all the commentators. The meaning is, "it 
 is not your opinion that those persons enjoy peace the longest who, 
 while they act justly, show that they have made up their minds not to 
 submit to injury ; but you observe the rule of non-interference, i. e. 
 you are strictly neutral or impartial, on the principle of not hurting 
 others and of avoiding the inconveniences to which reprisals would 
 subject you." For the μη βλάτντεβΟ'αι we may compare ^schylus, 
 Suppl. 577: νμϊν δ' άρήγειν ουκ εχω βλάβης ατερ. The preposi- 
 tion ετά here implies a principle or condition of action — as in Demos- 
 then. Philipp. II. p. 68 : ηγεΐτ ovv, ει μεν νμάς ελοιτο φίλους, ετά 
 τοις δικαίοΐξ αίρεΐ6%αι. The phrase το ϊύον νέμετε, means "to act 
 fairly or impartially to both of two parties." Thucyd. vi. 16: ωΰτΐερ 
 δυότυχουντεξ ου Ίίρο6αγορευόμε%'α, εν τω ομοίω τις ανεχεΰ^ω και 
 VTto των ευτίραγούντων υτζερφρονούμενος, η τα Ϊ6α νεμων τα όμοια 
 άντα^ιούτω — "if a man treats his inferiors as equals, he has a right 
 to claim the same treatment from his superiors." Herodot. vi. 11, and 
 109: 0•εών τα ϊΰα νεμόντων, "if the gods remain neutral." Aristotle 
 (JS^ei.i. 10,p,1368Bekker), by defining το άδικεΐν as βλάτίτειν, with 
 the addition of purpose and illegality, implies that βλάτίτειν was never 
 considered to imply an injury or any thing more than mere pain or 
 loss occasioned unwittingly, or by an inanimate object. He says — 
 Ϊ0τω δη το άδικεΐν το βλάτίτειν εκόντα τίαρά τον νόμον. It is 
 well known that the original meaning of the word is "lay hold of," 
 "stop," "retard," "impede;" as appears from the following passages. 
 
 Homer, Iliad vi. 38 : 
 
 ΐτΰτΐω γάρ ot άτυξομενω τίεδίοιο 
 οξω ενί βλαφ^^έντε μυρικίνω — , "caught in." 
 
 Odyss. I. 195 : 
 
 αλλά νυ τόνγε %εοι βλάτττουΰι κελεύ^^ου — 
 "stop him from his journey." On which Eustathius remarks: βλάτί- 
 τειν, κυρίως το εμτίοδίζειν τον τρέχοντα. 
 
 ^schyl. Agamemnon, 118: 
 
 βούκόμενοι λαγίναν, ερικύμονα φέρματι, γένναν, 
 βλαβεντα λοιβ^ίων δρόμων, 
 
 "things (i. e. the hare and her young) stopped from running any more 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GREEK. 683 
 
 races," as in Hor. iv. Carm. vi. 34: "Delise tutela Deae fugaces 
 Lyncas et cervos cohibentis arcu," i. e. "sistentis in fuga, dum eos 
 sagittis transfigit" (Orelli). 
 
 Sophocles, Electra, 696: 
 
 όταν δε ng ^εών 
 βλάτΐτγ], δύναιτ αν ονδ' αν ίΰχνων φυγείν, 
 
 "whenever any god stops one's flight, the best runner cannot get off:" 
 Ajax, 455 : 
 
 ε^ δε tig Ο'εών 
 βΐάπτου^ φνγου γ αν χώ %a%og τον κρείΰΰονα, 
 
 "if any one of the gods were to stop the pursuer." 
 
 We do not consider the word βρά'φαί, which is mentioned by 
 Hesychius {βράχΐ^αϋ, ΰνλλαβεΐν) as a synonym for βλά^Ι^αο, and which 
 certainly is very like it in sound, to be connected with this root. 
 Βρά^^αί, as Buttmann has remarked, is connected with^ci^^rro, just as 
 βλάξ is with μaλaκόg; μάρπτω is derived from μάρη=χείρ, and the 
 same root is found in ευμαρης^ a synonym for ευχερής. The word 
 μαρήγεί (=λαμβάνεΰ Hesych.) is, perhaps, as Alberti supposes, a cor- 
 ruption for μάρπτω. We do not look upon β-ραβεύς as connected 
 with βράφαι,: it is derived, as we have elsewhere surmised, from the 
 ράβδος, which was the umpire's mark of distinction. 
 
 455 In the word λεία, "a booty," "that which is taken," also 
 ληΐς, Dor. Xatg, the connexion with λάω need hardly be pointed out : 
 λίόπή and λι,ΰΰή are by-forms (see Hesychius). The common name 
 for the lion, λε^ων, " the seizing animal," clearly belongs to this second 
 class of the family of words into which the root XaF- enters. The 
 digamma in this name is preserved in the German Lowe, old German 
 Lev. We have stated above (§ 282), that χάρων is another name for 
 the lion, and that it means "the roaring beast." The Sanscrit name 
 sinha, according to Lassen (Ind. Alterthumsk. i. p. 295, note 5), means 
 "the slayer," probably for simbha, from sibh, "to slay." If so, it is 
 only by an accidental coincidence that JEschylus (Agam. 697) speaks 
 of the lion as a oivig or "destruction," for this cannot be the origin 
 of ΰυνέομαο, &c. That the lion should have several names is not at 
 all wonderful. "Of every thing in nature," says Bopp (^Annals of 
 Oriental Literature, p. 26), " of every animal, of every plant, speech 
 can seize only one quality, in order to express the whole by it. The 
 elephant is called in Sanscrit dantin (nominative danti) from his 
 teeth, or dvirada (endued with two teeth), or from his trunk serving 
 him as a hand, he is called hastin or karin (nominative hasti, kari) ; 
 
684 USE OF AUXILIAKY VERBS IN GREEK. [bOOK IV. 
 
 from his habit of taking water in his trunk and then drinking when 
 he pleases, he is called dvipa (twice drinking). "Were the Sanscrit to 
 express all these qualities of the elephant by one word, it would be 
 obliged to join all those mentioned together, and to add a great num- 
 ber of others. The serpent is called from his motion sarpa or pan- 
 naga, going not with feet (from pad, foot, wa, not, and ga, going) ; or 
 uraga, going upon the breast. Besides many other names, the serpent 
 has also in Sanscrit that oi ραναηάςαηα, wind-eating." In a passage 
 of the Nalas (xx. glok. 1), khe-charaJi, "going in the air," is used as a 
 name for " a bird," but the etymology is indicated in the comparison : 
 acMrena atichakrdma^ Iche-charah kJie char arm ina, "he passed [by 
 [the rivers, &c.] rapidly, like an air-farer faring in the air.''"' Besides 
 this it may be mentioned that cervus is nothing but κερεΓος, " the 
 horned animal," that lobster, clubster, or cluhstart {=clubtail), is the 
 English name, not merely for a thick-tailed shell-fish, but also for the 
 stoat, an animal with a tuft on his tail {Quarterly Bev. Vol. Lvn. 
 p. 90), just as aXkovQOQ=^al6XovQog and ΰκίουρος refer to the striped 
 (^αϊολος δράκων conveys the same idea; see above, §97) or thick tails 
 of the cat and squirrel, that άλωπηξ appears from the Sanscrit word 
 lopagaka to signify "the carrion-eater " (the other Sanscrit name for the 
 fox, lomaga, means "hairy"), that the dog is called "the taker," cam>, 
 hund, &c., in all languages of the Indo-Germanic family (above, § 269), 
 and so on (see Varron, p. 155). In XaJ^ag^ lapis, the idea of taking 
 up is clearly implied, for the idea of "a stone" is that it is something 
 detached and movable, and, if we are right in supposing that λέγω 
 also belongs to this root, it is an encouragement to the supposition 
 that Xafaq belongs to it also, when we find λέγω, λογάδην, &c., espe- 
 cially applied to the picking and placing of stones. This etymology is 
 much confirmed by the fact that the Sanscrit ςίΐά, fem. "a stone," is 
 manifestly of the same origin as ςιΐ, to "glean," and gilam, neut. "a 
 gleaning of the ears of corn." The oldest walls in Greece, especially 
 those which are called Cyclopean, were formed of picked stones, which 
 were adjusted together without cement, as they happened to fit, the 
 intervals between the larger blocks being filled up with smaller stones. 
 Hence the idea of selection, of placing the small with the great, be- 
 came identified with that of a stone, and it was customary to speak 
 proverbially of such arrangements, without alluding to the word λίΟ'θ£. 
 Thus Sophocles says {Ajax, 158): 
 
 και tOL• όμικροί μεγάλων χωρίς 
 ΰφαλερον τινργον ρνμα ηελονταυ' 
 μετά γαρ μεγάλων βαώς άριβτ αν 
 καί μέγας όρΟ'οίΟ'' νπο μικρότερων. 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GREEK. 685 
 
 ΙύΧ ού δυνατόν τους άνοψονς 
 τούτων γνώμαξ τίροδίδάύκειν'^. 
 The last line shows that the chorus is reciting a proverb, like the 
 
 Italian : 
 
 Duro con ditro 
 Non fa bon muro. 
 Or the German: 
 
 Hart gegen Hart 
 Nimmer gut wardf. 
 
 It is very strange that all the commentators have failed to perceive 
 this obvious interpretation, which is confirmed by a passage in Plato, 
 Legg. p. 902 d : ov μην ονδε κυβερνήταις ονδε ΰτρατηγοΐς ονδ' 
 οίκονόμοΐζ ούδ' αν τιΰΐ itoXuTLTioig ονδ' αλλω των τοιούτων ονδενΐ 
 χωρίς των ολίγων καΐ ΰμίκρών τίολλα η μεγάλα' ονδε γαρ ανεν 
 ΰμίκρών τονς μεγάλονς φαΰϊν οί λιγόλογοι λί^Όνς εν κεΐύΟ'αί. Of 
 the change of the vowel in λίΰτΐή, λίΰόή^ and λίΟΌς, we shall have 
 further examples in the third set of words from this root. 
 
 456 Of λάω, in the sense of "to wish," we have the following 
 forms, λώ, λης, λη, λώντι, and the optative λεωμι (Hesych.) ; also the 
 reduplicated forms λι-λαίε6^αι (=ε7ίί'^νμεΐν, ορεγεΰ^αι. ΰτίεύδειν. 
 Hesych.) and λι-λεϊ (=φ^Όνει, ετΐι^νμεΐ. Hesych.); λίΰόομαί and λί- 
 τομαι likewise belong to this root ; for their form compare the words 
 λι66Ύΐ and λί^ος. Another form is λιχάζει {=ε7ίι^νμει Hesych.), with 
 which may be compared γ-λίχομαι. To these we may add λίτίτω^ λε- 
 λιμμενοζ^ and λίί^ (=ε7ΐί%'νμία Hesych.), which appears to have the 
 same origin with the Latin libet and libido. We do not consider the 
 quantity of the first syllable as any objection to our classing λιμός and 
 λί-Λαρ-ίί? in this set of words. We have seen all through the words derived 
 fromAaP-, in all three significations an indiscriminate use of the vowels 
 
 * If we might adopt the quaint style of the sixteenth century, the proverbial 
 tone of the whole passage might be given thus : 
 " Great without small 
 Make a bad wall ; 
 For the help of the great 
 Makes the little go straight. 
 And the nobles endure 
 With the aid of the poor. 
 But wisdom may preach — 
 She never vrill teach 
 These maxims of good 
 To the minds of the rude." 
 f Lehmann's Florilegium in Lessing's Werke^ Vol. xi. p. 672 Lachmann's 
 edition. 
 
686 USE OF AUXILIARY VERBS IN GREEK. [bOOK IV. 
 
 a, ε, V. This has taken place on account of the connecting vowel 
 being short, and the root terminating in a digamma, which has been 
 represented in the derived words by a great variety of substitutes. 
 The lengthening of the syllables λιτί- and Xl- in the words λtπro:ρrIg 
 and λψόζ may perhaps be indicative of a lost guna. At all events, 
 this is a more satisfactory account of Xi7taQTfi$ than the old explana- 
 tion αΛο τον λίαν Λαρεΐναυ. It must be allowed too, that the connexion 
 which Passow points out (s. v. λίΛοίρός, 5) between the meanings of 
 λίτιαρός and λιιίαρηζ is sufficient to establish some sort of relationship 
 between the words; for λιτίαρός^ even when it is applied to bright, 
 shining substances, immediately refers to one meaning of its primitive 
 λίτία (cf. n. xvm. 596, Od. vii. 107; below § 461), and λίτιαρης may 
 have denoted originally the clammy, viscous, and adhesive nature of 
 oil. If there is really an affinity between kiTCagog and λιπαρής, and if 
 the supposition of a guna is not allowable, the difference of quantity 
 has arisen from one of those accidents in language which cannot be 
 properly accounted for by any causes known to us. 
 
 457 Before we proceed to consider Ο'ελω and βονλομαί, it will be 
 necessary to investigate the words αϊγλη, άγλαός, and άγάλλω, which 
 Passow assigns to this root. A careful investigation will show us how 
 far this is the case. The first of these words has excited some interest 
 from an ingenious attempt which was made some years since to give 
 a new meaning to it in the Philoctetes of Sophocles (v. 816). The 
 passage runs as follows : 
 
 "Tjtv odvvag άδαήζ/'Τπνε d' αλγεων, 
 
 εναης ημΐν ελ&οις 
 
 εναίων, εναίων ώναξ' 
 
 6μμα6ί ά' άντεχοίς τάνδ' αΐγλαν 
 
 α τεταται ταννν. 
 In the Eheinisches Museum (for 1828, p. 125, translated in the Philo- 
 logical Museum, i. p. 468), Welcker has endeavoured to prove from 
 Bekker's Anecdota (p. 354), from Hesychius, and from Pollux, that, 
 in the passage of Sophocles just quoted, αϊγλΎΐ signifies a band which 
 Sleep was begged to continue holding before the eyes of the slumber- 
 ing hero. In a subsequent paper (Bheinisch. Mus. for 1833, p. 454, 
 note 3) he has made some additional remarks on the same subject. 
 
 It appears to us, that though Welcker's interpretation of the pas- 
 sage in Sophocles is characterized by his usual ingenuity, there are 
 serious objections to it, and that he has proved nothing except that 
 α^ιγλΎΐ may signify a glittering band, just as it might describe any 
 other brilliant and splendid decoration, or as the phrase atj;A«£r;^^f(j£a3 
 ^νύάνω is used by Pindar to describe the golden fleece (Ρί/ίΛ.ιν.231). 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GREEK. 687 
 
 One great objection to the application of this sense of αϋίγλΎΐ to the 
 line of the Philoctetes is, that in the other two passages in Sophocles 
 where this word occurs it manifestly bears a different signification. In 
 the (Ed. Tyr. 207, 
 
 rag τε τίνροφόρονς 
 'Αρτέμιδος αϊγλας, ξνν alg 
 Αύκυ oQsa δυάΰβΒΐ — 
 
 the word refers to the two torches with which Artemis is represented 
 on ancient coins. In the Antigone, 610, 
 
 άγήρω δε χρόνω δννάότας κατέχεις Όλνμτΐον 
 μαρμ,αρόεΰΰαν αΐγλαν, 
 
 the epithet shows clearly enough that αίγλη implies nothing but the 
 splendour and brightness of the abode of the gods. It might be 
 thought that this last passage is a sort of confirmation of Welcker's 
 opinion, and thsbt μαρμαρόεΰβαν αϊγληντβίβΥΒ to the band of snow with 
 which Olympus is capped; but this cannot be: μάρμαρος was not 
 used in Sophocles' time to signify a white stone more than any other 
 bright, polished stone. The word μορόευς, which, as Doderlein justly 
 remarks {Lat. Syn. und Etym. ii. p. 81, note), is connected with 
 μαρμαίρω, is used as an epithet to earrings in Iliad xiv. 183 ; Od. 
 XVIII. 298, and it is certainly not hinted in either passage that the 
 stones in the earrings were white. Besides, a poet, so full of taste 
 and art as Sophocles, would never have expressed such an allusion in 
 so frigid a manner. Our chief objection to Welcker's interpretation 
 of the Philoctetes is this, and we think it is decisive: we can gather 
 from the context that Sophocles intended to use the word in its 
 primitive sense. The chorus says afterwards (847) in a parenthetical 
 way — αλεηξ VTtvog lud'Xog^ " sleep in the sunshine is good for our pur- 
 pose," i. e. because it is very sound. If this was a common opinion, 
 it was natural enough for the chorus to pray that sleep would keep 
 before the eyes of Philoctetes the light of the sun (αϊγλτ]) which was 
 spread over them and prevented him from waking. That αϊγλη refers 
 to the brilliant light of the sun in particular may be gathered from 
 the hints of the lexicographers. Anecd. Bekkeri, p. 354: καΐ η d'voia 
 δε η νπερ τον κατακλυσμού εΙς Ζίελφονς άΐίαγομενη αϊγλη εκαλεΐτο 
 (it will be recollected that the Delphian god was also the god of the 
 sun) άλλα καΐ η όελήνη. Hesychius has the following articles: ΑΙγ- 
 λαηρ. b ^Αβκλητίίος (because he was the son of Apollo). Αϊγλης 
 Χάριτες, πιθανώς έγενεαλόγηύαν τας Χάριτας, Αίγλης καΐ 'Ηλίου, 
 ετίεΐ τας Χάριτας λαμτΐρας είναι δει, and ΑΙγλητην. ετίί^ετον 
 ΆτίΛολλωνος, where Toup quotes Apollon. Rhod. iv. 1716: 
 
688 USE OF AUXILIARY VERBS IN GREEK. [bOOK IV. 
 
 αίγλητην μίν Ινΰχότΐον εΐνεκεν αϊγληζ 
 Φοίβο ν κεκλόμενοί. 
 
 Let us add a remark which we think is also of some importance. It 
 cannot have escaped any one that all writers are apt to repeat them- 
 selves. Now, although we are told that Sophocles published the 
 Philoctetes 31 years after the Antigone^ it cannot be believed that he 
 never read the latter again : the beautiful chorus, from which we 
 have just quoted (1.583 foil.), must have been constantly in his mind, 
 and we have no doubt was present to his memory when he wrote the 
 lines in the Philoctetes which we have been discussing. A line or two 
 before μαρμαρόεΰΟαν αϊγλαν the following passage occurs (600) : 
 
 vvv γαρ εόχάτας ντΐερ 
 
 ρίζας ο τετατο φάος εν ΟΙδίτίον δόμοΐζ κ. τ. λ. 
 
 It will readily be understood what train of thought led him to sub- 
 stitute for δ τετατο φάοζ in the one passage the perfectly synonymous 
 αϊγλαν a τεταταυ which we find in the other (cf. Plato, Besp. 616 β ; 
 δΐ(χ τίαντος τον ονρανον τεταμενον φώξ εν^ν\ and how the αϊγλγι^ 
 which occurs so shortly after in the Antigone, became mixed up in the 
 same sentence in the Philoctetes. 
 
 458 To return, however, from this disgression, which has little to 
 do with the etymology of αίγλη (and that is the point we have now 
 to discuss), we agree with Lobeck {ad Soph. Aj. 40, p. 97) in classing 
 αϊγλη among the derivatives from αω, or as it should be written, 
 FaFm. The labials may be recognised in Favonius and vapor, and 
 perhaps also in φάθζ'=φάΓος= fafog : compare the Sanscrit bha-va-t 
 with φως, φωτός (above, § 257). The ρ in vap-or as well as in the 
 cases of a similar insertion which he mentions (note 14): δράω, δρα- 
 τίετης; 0•«ω, Ο'άτΐω; μέλος, μέπλω; κείρω, carpo, κάρτίος; κοίλος, 
 κόλτίος; λάς^ lapis; χάω^ capio; τ ρέω, trepidus; περόνη, τίόρτΐη; luo, 
 lupercus; are all, we think, to be explained in the same way. The other 
 words connected like aϊγλ7ιwιih. αω are as follows — ανω, αυγή, ανρα, 
 αήρ, αΐ^'ήρ, αϊ^ω, αΐ^'νΰΰω, cestus, αϊολος, cestas, άΐΰΰω, all•, (genitive 
 αϊκος), αϊζ (genitive αϊγ&ς), and αίγίς. In all of these we find the cognate 
 ideas of blowing, flaming, shining, flickering, moving rapidly. That 
 these ideas are related to one another and to those of "blowing" or 
 "the motion of the air," and of "light" or "brightness," is clear 
 from a comparison oi flare with, flagrare ; οϊ φαύλος with felix, faus- 
 tus (above, § 152); of λεν-κός with lev-is; from the various meanings 
 of micare and "light;" and from the two uses of καταίΟ'νβΰω in 
 Pindar (Pyth. iv. 83, v. 11). "We have shown above that the stem 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GREEK. 689 
 
 λα- or λα -f-., which enters into words bearing all these meanings, 
 primarily signifies only motion in general. 
 
 459 To this class of words then, we agree with Lobeck in refer- 
 ring the first syllable of αϊ-γ-λγι^ "the light of the sun"; and we 
 entertain no doubt that άγάλλω and αγλαοξ are derived from it : that 
 ccyXaog, at least, is, appears from the fact that'^yAaw;, one of the 
 Graces, was called by Hesiod by the same name as her mother Αϊγλη 
 (Senecr. deBenefic. i. 3, § 6). We consider the ending to be a formation 
 of the pronominal root /a, under the form ga, with the element la, 
 which we have discussed above, and we proceed to show that whether 
 it appears as γε-λάω, ΰέ-λας, ε-λη, or α-γ-λα -og, it conveys the same 
 general notion of light or brightness. In Latin the ideas of "shining" 
 and "laughter" are mixed up with one another: remdeo, "to throw 
 back light," a by-form of mteo , is connected in meaning, perhaps in 
 origin with video, as is proved by the following passages (quoted by 
 Doderlein, Lat. Syn. und Etym. n. p. 73). Catullus, xxxix. 15: 
 
 Renidere usquequaque te nollem 
 Nam risu inepto res ineptior nulla est. 
 
 Tacitus, Annal. iv. 60: Tiberius torvus aut falsum renidens vultu, 
 and Acron ad Hor. Carm, u. 18, 2: Nidor a nimio odore dictum, seu 
 risu, unde et renideo. That the much spoken of ποντίων κυμάτων 
 άνήρί^'μον γελαβμα of ^schylus {Prom. 90) refers to the infinitude 
 of glittering spangles which one may observe on the sea when a gentle 
 breeze is passing over it in sunny weather , must, we think, have 
 struck every reader of taste. It is well known too, that in modern 
 French the epithet riant, or "smiling," is constantly used in speaking 
 of landscapes, &c., and the phrases " a cheerful prospect," &c., are not 
 uncommon among ourselves. The following glosses from Hesychius are 
 conclusive in favour of the opinion that the ideas of merriment and 
 brightness are included in the word γελάω. Γελάν, αύγην ηλίου. 
 Γελεΐν. λάμπειν-, άν%εΙν. There is no occasion to read here Γέλαν, as 
 Toup proposes (Vol. ni. pp. 400, 473), any more than in Βέλα. ηλίΟξ 
 καΙ αυγή υτώ Αακώνων, or ίΏ.''Έλη. ηλίου αυλαία η αυγή (Timajus); 
 the γ, |3, and aspirate in these words represent the digamma, as does 
 also the ΰ in ΰέλας, ϋελήνη. In the word γαλήνη the leading idea is 
 that of the sunshine or brightness which invariably accompanies fine 
 weather in the East ; the same idea of brightness accounts for the 
 other meaning given to this word by Hesychius: Γαλήνη, το ετΐυ- 
 Ίίόλαζον εν ty μεταλλεία του αργύρου χωνευομένου. The idea of 
 shining whiteness is conveyed also by the word γάλα, "milk," and by 
 the Sicilian word γέλα (Lat. gelus, gelu), "frost," which according to 
 Lennep (ad Phalarid. Ep. 106, p. 308) is alluded to in the following 
 
 YY 
 
690 USE OF AUXILIAEY VEEBS IN GBEEK. [bOOK IV. 
 
 gloss of Hesychius: Κιελλη. φέγγος, αυγή, φως, Λάχνη, ομίχλη, 
 where Rulmken (ad Tim. p. 96) reads βείλη. The reviewer of Niebuhr's 
 History of Borne (quoted by Goller, de Situ et Origine Syracusarum^ 
 p. 150) supposes that the Sicilian river Gela was so called from its 
 coolness. We entertain a different opinion. AVhen we remember that 
 the city Gela was founded by the Khodians (Thucyd. vi. 4), who were 
 near enough to the Triopian promontory to be influenced by the Trio- 
 pian religion ; that the Triopian rites were at an early period intro- 
 duced into that city (Herod, vii. 153); that one of the Triopian deities 
 was Apollo (Herod, i. 144); that an ancestor of Gelon, one Telines 
 of the island of Telos, was Hierophant of the Triopian rites (Herod, 
 vm. 153), and that this office remained in the family (Bockh, ad Schol. 
 Pindar, p. 314); finally, that the Athenian priest-tribe was called 
 Γελεόντες'*' (Arnold's Thucydides, Vol. i. p. 659), and that the patron 
 god of the old Athenians was '^τΐόλλων πατρώος^ considered as the 
 sun-god (φαΰί τινεξ^ Αθηναίους αντόχ^οναζ cpvvai καΐ τούτω γο- 
 νέας εχειν Την %aV'HlLOV, ος 6 αυτός έΰτίν Άτνόλλωνί; Scholiast, 
 in Plat. Euthydem. p. 369 Bekker); we cannot doubt that the city 
 and river Gela, as well as the two kings Gelo and Hiero, owed 
 their names to their connexion with the Triopian worship of the 
 bright sun-god. On the whole, then, αϊ-γ-λη=^ί^ά-γ-λη or φα-γέ-λη 
 (for the change of place in the semivowel see above, § 116, and else- 
 where) is a word strongly expressive of bright, shining light, and is 
 particularly employed to signify the sun. 
 
 460 It is worthy of remark , that , as the latter of the two ele- 
 ments which go to make up αϊ-γλη, refers not only to "light," but 
 also to the sense of "seeing," as in λάω and β-λετίω; so the first part 
 of the word expresses not only "light" or "brightness," but also 
 "speaking" (η-μυ, αϊ-νω), and by association "hearing" (άΐω; above, 
 p. 93). The connexion between "light" and the faculty of "seeing" 
 is sufficiently obvious: without φως there could be no οψίς (Plato, 
 Respub. VI. p. 507 d). The ideas of "speaking" and "light" both 
 belong to the more general one of manifestation. The etymological 
 connexion of the words expressing them has been shown by A. W. 
 Schlegel (in an article in the Indische Bihliothek, Vol. n. p. 284), who 
 has compared the Sanscrit root hhdf "to shine," with the Latin and 
 Greek fa-ri, φημί, "to say." The inflexions of the latter verb, in its 
 Doric form, are perfectly identical with those of the Sanscrit bha: 
 thus we have 
 
 Those who read Τελέοντες, must remember the gloss in Hesychius, 
 
 Γέλεα, τέλεα. 
 
hha-mi 
 
 φα-μί 
 
 bhd-si 
 
 φα-ΰί 
 
 bha-ti 
 
 φοί-τί 
 
 hJid-mas 
 
 φοί-μέζ 
 
 hJid-tha 
 
 φα-τέ 
 
 hJid-nti 
 
 φα-ντί 
 
 CHAP, v.] USE OF AUXILIARY VEEBS IN GREEK. 
 
 Library 
 
 5{^Califorma• 
 
 The Greek φαί-νω, "to shine," bears the same relation to φά-μι-, that 
 αϊ-νω does to the Latin aio or the Greek τιμι. In the sense "light" 
 we have the Sanscrit bhds and the Greek φάοξ. Although the root 
 hhd itself never signifies "to speak," we have with that signification 
 the root bhdsh, which bears the same relation to it that hrish does to 
 hri (above, § 288). Thus, we have bhdshate=loquitur ; bhdshd= 
 loquela, dialectus ; bhd-shitam=sermo ; bhdshyam^^commentarius; dvi- 
 bhdshin=bilinguiSj interpres; abhi-bhdshd=allocutio ; abhi-bhdshin 
 =aUoqui solitus^ &c. A writer in Blackwood's Mag. Feb. 1840, p. 
 208, compares the following cognate woids, which signify ^'■light''^ 
 and ^'• sound'''' respectively: 
 
 clarus κλως 
 
 dim dumb 
 
 swart surdus* 
 
 lauter loud. 
 
 Not altogether unconnected with this interchange of meaning we 
 have two compounds with ota, one of which, though referred by its 
 etymology to a root expressing distance or extent in space, is con- 
 stantly used to denote a pealing sound, and the other, though derived 
 from a root signifying a shout or cry, is almost invariably employed 
 as a synonym of μακρόζ. These adjectives are OLaTtQvuiog, connected 
 with δίατίεράω (see on Pindar, Nem. iv. 51, 52), and δνωλυγίος^ 
 which contains the same root as δλ-ολυγ-ή. That the ideas of "loud- 
 sounding," "heard afar ofP," and "exhibiting a lengthened vista to the 
 eye," are really interchanged in these words, may be inferred from the 
 following passages. On the one hand, we have Hom. II. vm. 227, 
 XVII. 247: ηϋΰεν δίατίρνβίον; Soph. (Ed. Col. 1479: δίαπρύΰίος 
 6τοβος\ Callim. Hymn, in Del. 258: δίατΐρνόίην όλολνγην. On the 
 other hand, δLωλvγLog, which Hesychius renders ηχονν ΙττΧ τίολν, is 
 explained with reference to its usage meaning Ιτά τίολν διηκον (Ti- 
 mseus. Lex. Plat, p. 88, where see Ruhnken): and though a later 
 
 * In Pliny, H. N. xxxvii. 5, we find surdus color for "faint," "dull:" 
 with which we may compare the meaning claimed for άμβλνς, above, 
 § 2J8. So also fusca vox is opposed to Candida or canora^ and we have 
 the phrase infuscatur ex inopinato of the nightingale's song (Plin. H. N. 
 X. 29, 43, § 82). 
 
 YY2 
 
692 USE OF AUXILIABY VERBS IN GREEK. [bOOK IV. 
 
 writer, like Libanius (Vol. iv. p. 149 Reiske: άνεβόηΰεν διωλνγίον), 
 may refer to the lengthened sound originally intended, there can be 
 no doubt that in the old Attic writers the word is quite equivalent to 
 the adjectives denoting merely length or extension in space : see e. g. 
 Plat. Thecetet. p. 162 A: μακρά καΐ δίωλνγως φλυαρία. Legg. -p. 
 890 Ε : μήκη κέκτηται δίωλνγία. 
 
 461 In the dualism of the Greek mythology the Goddess of the 
 Moon appears as the sister of Phoebus. Her name in connexion with 
 this worship is όε-λή-νη^ which is in fact only another form of γ-λή-νη. 
 Compare τίρονβελεΐν , τίρονγελεΐν, and their probably original form 
 τΐροΰ^ελεΐν (Buttmann, Lexilogus, ii. p. 159). In the Latin language 
 we have the shorter form which contains only the element λε -f- : com- 
 pare Lu-na, lu-c-s, with ^ε-λήνΊτι^ λεν-κόξ. We find a similar abbre- 
 viation in the Latin lac-t-, "milk," i. e. "white liquid," compared with 
 the Greek γάλα{-κ-τ) (above, § 212). "Whenever in cognate languages 
 synonymous words exhibit the same root, sometimes with and some- 
 times without a prefix, we may generally conclude that the longer 
 form is of later introduction, the additional syllable, which is generally 
 of pronominal origin (§ 213), being prefixed for the sake of greater 
 emphasis or distinctness. There can be no doubt, therefore, that the 
 word β-λέτί-ω, γ-λαυ-κόξ^ which exhibit the labial and guttural ele- 
 ments respectively of the pronominal prefix Pa-, are more recent than 
 the simpler derivatives from the root λε- or λε/-. There are two 
 words of considerable interest, which contain the combined elements 
 6ε-λα or γε-λα, in a form very likely to escape detection. These are 
 άΰελγής and ΰίγαλοείς. The former word was derived by the older 
 grammarians from the town of Σελγη in Pisidia , so that άΰελγεοα 
 was presumed to denote the reverse of the characteristics of that town. 
 This etymology needs no refutation; and there is not much to be said 
 for the derivation from Ο'ελγω, which is so confidently proposed by 
 modern lexicographers. The word denotes excess of• any kind; for it 
 not only implies moral extravagance and outrage, but even inordinate 
 size; thus we have Arist. Plut. 560: τίίονες αβελγώξ, "fat beyond all 
 measure;" and Plat. com. fragm, 24 (ii. p. 288 Meineke): κριοξ άβελ- 
 γόκερωg, "a ram with enormous horns." But these are secondary 
 usages, and we entertain no doubt that the primary meaning of άΰελ- 
 γής is properly given by Hesychius when he renders it άκόλαΰτος, 
 άκά%'αρτος the latter of these synonyms pointing to the form of ΰέλας, 
 which appears in ΰελαγέοι^ "to light up, to illume;" so that άΰελγής 
 primarily means "dark, dirty, foul, unclean, defiled." This etymology 
 is confirmed by the relation between άΰάλγη (Hesych. άΰάλγαν άμε- 
 λείαν) and βαλαγεΐ Ο^^^^άί.ταράύΰευ' η γαρ φροντίζ Σάλα λέγεται). 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GREEK. 693 
 
 whence άΰαλαγεΐν (Toup,m. 4:91)=άφροντίότΒΪν Qiesjch. s. v. άΰάλ- 
 λειν). Accordingly α-ύελγήξ and ύελαγ-είν contain really the same 
 elements as γάλακ-τ or γλαν%θζ. The other word ΰίγαλοείζ is un- 
 doubtedly derived from uC-JF'aXog , "the fat or grease of a hog," the 
 first syllable being the root of dvg (above, § 222), and the rest of the 
 word being this combination fa -λα or γάλα. The use of βυγαλοείξ 
 and the compound νεο-ΰίγαλοξ (Find. 01. m, 4) to signify generally 
 that which is bright, glittering, glossy and shining, must be compared 
 with that oi XiTtaQOg, which properly speaking means "shining from 
 being smeared over with oil," but which is used to signify "brightness" 
 in general, as in the epithet L•'JtaQάμ7tvl•,J of a bright gold frontlet 
 (above, § 456). The original meaning of the epithet was, however, 
 quite obvious to the Greeks themselves, for Aristophanes ridicules the 
 phrase λίΛαρας ΆΟ'ήνας, maintaining that the poet who used it ap- 
 plied to the city a compliment better suited to a jar of anchovies 
 (άφνων τιμήν περίάφας, Acharn. 640). And the older poets do not 
 hesitate to say that "glossy tunics are shining with oil" (Hom. M. 
 XVIII. 596: χιτώνας ηκα ότίλβοντας ελαίω), or that "liquid oil 
 trickles off the close warped linen" (Od^ vii. 107: και,ροΰέων odO- 
 νέων άτίολείβεται νγρον ελαων), just as the Latin poets say: ignis 
 relucet a galea; vestes ardent; and the like. 
 
 462 To return to the shorter form, we have the three, meanings 
 "to see," "to take," and "to wish," expressed by one word λά/ω. At 
 a subsequent period the root Fa was prefixed to that word, and different 
 modifications of it were employed to express the same three different 
 meanings. For the first sense "to see," we have β-λετζ-ω, γ-λαν-(5(5ω, 
 and for the cognate idea of light, we have βελα^ γέλα, ΰέλας, and ελη. 
 For the second meaning " to take," we have β-λάτί-τω, γεν-το {Iliad. 
 vni.43, XIII. 241, xviii. 476), ελεΐν^ and for the cognate word "hand" 
 we have %'εν-αρ\(ίον the ν compare Alcman's κεν-το for κέλεro,Eustath. 
 p. 658, 29). And for the third meaning "to wish," we have the com- 
 mon word Ο'έλ-ω, which, with the Ο'εν-αρ just mentioned, bears the 
 same relation to ΰέ-λας, ε-λεΐν, that the equally common ^άλαββα does 
 to the older form ΰάλαΰύα (Κδη ad Greg. Cor. p. 300), and to ΰάλος, 
 ΰαλενω, όαλάΰΰω, αλς, &c., still used by the best writers. We think 
 Σαλαμίν-ξ, as the name of an island, is also connected with &άλαΰύα; 
 so also ΰαλαΰΰομέδων in Alcman's Ionics a minore : 
 
 Ίνώ 6αλαό6ομέδθίΰ% αν άτι ο μαύρων 
 
 ρίτίτεν φάτίζ γαλαΰηνον Μελικερταν. 
 as emended by Porson (Gaisford's Hephcestion, p. 337). And we may 
 compare Σεμελη= Θεμελη-, Welcker, Gotterlehre, i. p. 536. The labial 
 element of the original digamma is still preserved in the Latin volo. 
 
694 USE OF AUXILIAEY VEEBS IN GEEEK. [bOOK IV. 
 
 463 Before we proceed to investigate the origin of βονλομαι, it 
 will be proper to inquire how far, in its actual use by the best writers, 
 it agrees with or diifers from Ο'ελω, or rather ε-^ελω, as the word is 
 written by the epic poets, by Pindar, and, with the exception of 
 phrases like εΐ ^ελεΐζ, by the prose writers in general. 
 
 The distinction between εΌ"έλω and βονλομαι-, given by Buttmann 
 (Lexilog. i. p. 26), will certainly not hold any where but in Homer, 
 and we do not believe that it obtains even there. He is quite right 
 in his general statement that ε%^ελειν meaos " to be willing," and βον- 
 λεΰϋ'αί, "to wish;" that is the distinction which prevails throughout 
 the Greek writers : in his application, however, he is not only wrong, 
 but inconsistent. Let us turn to the first passage which he quotes 
 from Homer; it is taken from Priam's answer to Hecuba, when she 
 tries to dissuade him from going to ransom the dead body of Hector. 
 At the very beginning of his speech (v. 217), he says: μη μ^ εΟ'έλοντ' 
 Ιεναι τιατερύκανε, which means " do not detain me, for I will go ;" and 
 in the passage quoted by Buttmann he says, "if it be destined that I 
 die by the ships of the Greeks, I am not merely willing, it is my wish : 
 
 αντίκα γαρ με κατακτείνειεν ^Αχιλλενξ 
 
 άγκαξ ελόντ εμον νίόν, επην γόου εξ ερον εΐην' 
 
 for Ι wish that Achilles would slay me , after I have taken my son in 
 my arms and wept my fill." In the passage from the Odyssey it is 
 not true that βονλεταί implies a mere acquiescence in the will of 
 others; the wise goddess Athena tells Telemachus, that it is a pecu- 
 liarity of a woman to feel a strong attachment to her husband , who- 
 ever he may be, and that it is her earnest desire to further his inter- 
 ests as much as possible. Why the third passage was quoted we 
 cannot see, for it so obviously means an active though fruitless desire, 
 and has nothing to do with the Bereitwilligkeit — "the being ready 
 and willing" — which he says is implied in βονλομαι. It certainly is 
 somewhat remarkable that βονλομαι should be used by Homer in 
 speaking of the gods, contrary to the sense of the word, and the use 
 of all other writers. One would think that will^ rather than desire, 
 would naturally be attributed to a superior being. We must recol- 
 lect, however, that Homer's gods were very second-rate personages, 
 who might, under certain circumstances, receive wounds from mortal 
 men; so that we need not wonder if we find desire, and all other 
 human feelings, attributed to them. In Demosthenes, av ^εοζ l^kky-, 
 "if it be the will of God," is a common phrase (see for example, 
 Philipp. I. 42, and Herald. Animadv. ii. 5) ; and in Olynth. i. 23 , he 
 puts ε%ελω, as applied to the gods, and βονλομαί, as applied to man, 
 in direct opposition: δοκεΐ d' εμοίγε, ώ άνδρες Ά^ηναΐοο, δεί^,ειν ουκ 
 
CHAP, v.] USE OF AUXILIAEY VERBS IN GREEK. 695 
 
 εΙς μακράν, αν οϊτε O'fol θέΑω^ί. καΙ νμεΐς βονληΰ^'ε. And so we have 
 in Plato, Legg. vii. p. 799 ε : ει &εο£ ε^ελοί-, and in Legg. iii. p. 688 ε : 
 εαν ^'εος ε^'έλι^. There are three passages, one in Euripides, and two 
 in Plato, in which ε^ελω and βονλομαι are so directly opposed, that 
 we cannot mistake the distinction between them if we would. In the 
 Iphigenia in Aulide (v. 336), Menelaus says to Agamemnon: 
 βονλομαι δε α' εξελεγξαι, καΐ 6ν μήτ οργής νπο 
 άτίοτρετίου τάληϋ'ες, οϋτε κατ ατ ενώ λίαν εγώ. 
 οισ-θ'' or' εΰΛονδαξες αρχειν ^αναΐδαις τίρος'Ίλυον^ 
 τω δοκεΐν μεν ονχΐ χργ^ξων, τω δε βονλεΰ^αί ^ελων, 
 
 κ. τ. λ. 
 The two passages from Plato are as follows; Legg. ix. p. 863 b: ηδο- 
 νην...φαμεν...πράττειν ο τι τίερ αν αντης η βονληΰις ε^'ελήό^. 
 Eespublica, ιν. ρ. 437 β: τί ovv] ην δ' εγώ' δι^Ι^ην καΐ τίεινην και 
 ολωξ τας ετίΐ^νμίαζ., καΐ αν το ε%ελειν και το βονλεΰΟ'αι, ον 
 τίάντα ταντα εΙς εκείνα π^ αν ^είης τα εϊδη τα ννν δη λεχθέντα; 
 οίον άει την τον ετΐΐ^νμονντος -ψνχην ονχι ήτοι εφίεΰΟ'αι φήΰεις 
 εκείνον ον αν ετίιΟ^νμη^ η τίροΰάγεΰ^αι τοντο ο αν βονληταί οιγε- 
 νεό^'αι, η αν, καθ'' οΰον ε%ελει τί ot τίοριΰ^ηναι, ετίΐνενειν τοντο 
 Ίίροζ αντήν, ώΰτίερ τινοζ ερωτώντος, ετνορεγομένην της γενεΰεως; In 
 all these three passages it is abundantly clear that βονλομαι and βον- 
 ληόις refer to the desire or wishing for a thing, while εχί'έλωΐΒ restricted 
 to the mere will or willingness. In regard to Ο'ελων αρχειν in the pas- 
 sage from Euripides, it is evident from the perfectly similar sentence in 
 Plato's PoUticus (p. 299 b), that the mere willingness or acquiescence in 
 the office is implied: "pretending to have no desire for the office, but 
 in real wishes, in regard to his ambition, being perfectly willing to 
 undertake it." The words of Plato are οΐ'κουν ο γ' ε%•ελων και εκών 
 εν τοΓ^ τοιοντοις αρχειν δικαιότατ αν δτιονν πάύχοι και άτΐοτί- 
 νοι, where the meaning is, "he who, willingly and of his own accord, 
 &c.," as in the Protagoras (p. 335 a): οτι ονκ ε^•ελήΰοι εκών είναι 
 άποκρινόμενος διαλεγεύ^αι — "will not be willing as far as he is con- 
 cerned." We do not recollect one instance in the good Greek writers 
 in which βονλομαι and ε^ελω are confused; they are as distinct in 
 meaning and origin as the German correlatives wunschen, which ans- 
 wers to the former, and wollen, which translates the latter. 
 
 464 It is a common opinion (see Doderlein, Syn. und Etym. v. 
 56, and Passow, s. v. Ο'ελω), that βονλομαι, another form of which is 
 βόλομαι, bears the same relation to Ο^έλω, that volo, volt, voleham, 
 volent, do to velle, velim, and vis for velis, and that the β and ^ are 
 interchanged like φ and %' in φηρ and %ηρ , and υ and 0• in venari 
 and %ηραν\ it is also suggested that the change from ε or ο to ov is 
 
696 USE OF AUXILIAEY VEEBS IN GREEK. [bOOK IV. 
 
 explained by the transitioil from the Italian volere to the French ν ouloir. 
 It is of course easy to add a comparison of the Teutonic roots, which 
 really correspond to J^sX-, but which have generally been traced to 
 a nearer relationship with βονλομαί', such are the Gothic viljan, A. S. 
 vilnian, 0. S. wilUan, 0. H. G. wellan^ N. H, G. wollen, Engl. will. 
 If we had no other means of proving it, this word βονλομαί alone 
 in its relation to -θ'δλω, might serve to convince us of the uselessness 
 of confining the functions of comparative philology to a mere juxta- 
 position of prima facie resemblances*. On all sides , 0•8λω presents 
 correspondences of signification to words containing the root /ελ-, 
 with which, we have seen, it has an obvious etymological connexion; 
 whereas βούλ-ομαι, with its two labials and heavier vowel o, is no less 
 distinct etymologically from the root /ελ- , than it is different in sig- 
 nification from the verb ^ελω. And first, let us consider the length- 
 ened form εΌ'έλω. It is a well-known fact, that, in the Greek language, 
 the oldest verbs were very frequently reduplicated forms of those in 
 
 * The comparison of βονλομοίΐ with these Teutonic roots and with 
 volo, and the reference'of ϋ'έλω to a forced Sanscrit affinity, which we find 
 in Benfey's voluminous work (Wurzellexicon, Berlin 1839 — 42, i. 320, ii. 
 328, 350), supply an illustration of the total want of any thing like a real 
 insight into the structure of language which that author everywhere dis- 
 plays. It is not to be denied that his industry has been very great, and 
 that he has collected an enormous mass of crude materials ; we hear too, 
 from our German friends, that he is a very good Sanscrit scholar; but he 
 exhibits no acquaintance with the higher departments of classical learning, 
 and he deals with the Greek words as if there were no means of dis- 
 tinguishing between the root and the formative affix. To take one example; 
 the interesting word νά-κιν^Όζ is referred to the root ι5 = "ίο bring forth,'^ 
 and the last part confidently is indentified with avd-og ("der letzte Theil 
 des Wortes ist ohne alien Zweifel αν&ος:" i. p. 413). Now there are 
 many purely Greek words ending in -ivdOg, which is merely a formative 
 affix of pronominal origin (above, § 263). The first part of the word is 
 therefore νακ- as in the Latin vac-cinium; and we recognise this in the 
 root of νακ-ίξω (=νετίξω, Hesych.), and in a number of Teutonic roots 
 signifying softness or pliability, e. g. weiche, A. S. wake, &c. We are 
 sure, as in the case of the cognate /m, that the plant derives its name 
 from the mythological personage. Now Benfey himself has seen that ΐρίς= 
 J^LQig means primarily the curved rainbow (ii. 302), and it is equally cer- 
 tain to us that in the old elementary religion of the Laconians, from which 
 the legend of Hyacinthus is derived (see Miiller, Dor. i. p. 374, who how- 
 ever derives the name from the flower), the beautiful youth slain by the 
 discus of Apollo is merely a type of the rainy spring, whose tender flowers 
 are wet with the moisture of heaven, and which falls a victim to the 
 powerful orh of the sun-god. So that the Iris or "rainbow" and the Hya- 
 cinthus or "watery flower" are equally symbolic of the triumphs of the great 
 God of day. If this interpretation is correct, and if the explanation of 
 νάτιιν&ος is a fair specimen of the Wurzellexicon (and we think that it 
 is), what is the use of the book, except to show that classical scholar- 
 ship is still the best and safest basis of operations for the general phi- 
 lologer? 
 
CHAP, v.] USE OF AUXILIARY VEEBS IN GEEEK. 697 
 
 common use. Now, it has been observed, that a number of verbs, 
 which in the oldest state of the language were digammated , also in 
 the old language appear with an initial ε in the present tense (for 
 example, Μλδομαο, εελτίομαι, εέργω, εείδόμενος^ εί'ΰκω] Buttmann's 
 Ausfuhrl. Sprl. § 112, Anm. 23); which is merely a mutilated redu- 
 plication (Buttmann, § 82, 3, note). To this class we refer εΟ•ελω= 
 εί^έλω , and consider it , on the principle mentioned just before , as 
 standing for the reduplicated form J^ε/ελω. The only difference, in 
 fact , between ε^'ελω and εελδομαι is , that Ι^ελω has lost only one 
 digamma, and εελδομαι has lost both; the preservation of the one 
 digamma in ε%ελω is due, we conceive, to the very common occurrence 
 of the word from the earliest times. 
 
 Again, no one can doubt that ελεΐν and ελκειν are connected. 
 That such is the case is obvious from a comparison of the glosses 
 Γέλλαυ, τΐλαΐ'/Ελλίξων, τίλλων (Hesychius); and the Latin vello 
 (all which bear the sense of ελκειν) , with the common uses of ελεΐν. 
 The connexion of ελκειν with ηλακάτη is acknowledged by every one 
 who has read Buttmann's paper on the -ζ/λεκτρον (compare κτένες ελκψ 
 TTjQeg, Phanias, Έρ. 4, 5). Now Hesychius has the following glosses : 
 Γελγη' δ ρώπος καΐ βάμματα, ατρακτοί καΐ κτένες (on the first 
 part see Toup, Emendationes, iv. p. 106); Γελγια. τίίνη. βτίά^η. 
 κουτάλια. From these we infer that there were other words connected 
 with weaving and spinning also derived from έλκω: for the change of 
 the κ into γ is hardly worth noticing. Besides all this, we have the 
 common word -Ο-εΛ^ω, the primary signification of which is "to enchant," 
 "to act upon by charms." That in this sense it was nothing more 
 than a synonym for έλκω appears from the following considerations. 
 The principal instrument in magic among the Greeks, especially for 
 love-charms, was the wry-neck, Xvyl•,^ a little bird which, when fastened 
 to a metal wheel and turned round like an άτρακτος^ was thought to 
 have a δνναμις ελκτική. Thus Pindar says (Nem. iv. 35): ϊϋγγι d* 
 έλκομαι ητορ νονμηνία ^ιγεμεν, and Theocritus, π. 17: ϊνγ^^ έλκε τν 
 τηνον εμον τίοτΐ δώμα τον άνδρα. Conversely, we find ελκειν ϊνγγα 
 επί τινι, and the word ελκειν is sometimes used absolutely for "to 
 entice," just like Ο'έλγω. Thus we have in Philostratus , Imagg. i. 4, 
 p. 769 : αρπάζεται τον θάνατον καλώ κα\ ηδεΐ τω ομματι και οίον 
 VTtvov ελκοντι. This passage is quoted by Jacobs, Anth. Pal. in. p. 664, 
 in a note on the words of Philippus : 
 
 ως εΛίτίερκάζεις μιαρα τριχί., νυν φίλον ελκών, 
 την καλάμην δωρ^, δονς ετέροις το %'ερος. 
 
 He also cites Lysias, de Eratosth. Ccede, p. 191 Bekker, where είλκε^ 
 bears a different meaning, i. e. that of the Latin veUicare. We have 
 
698 USE OF AUXILIARY VERBS IN GREEK. [bOOK IV. 
 
 the verb έλκω in connexion with τΐεί^'ω in Hato, Besp. v. p. 458 d, 
 and with this verb and κολακεύω in vii. p. ^%β d : ονκονν και άλλα 
 εναντία τούτων ετΐΐτηδεύματα ηδονας έχοντα, α κολακεύει μεν ημών 
 την -ψνχην καΐ έλκει εφ' εαυτά , τΐεί^ει δ' ον τονζ και οπηονν με- 
 τρίους. That this primary notion of ^έλγω and έλκω was connected 
 with the idea of ελεΐν, βαλεΐν, is shown by the ττ] 6y ληφθέντες ϊνγγι 
 of Aristophanes (Lysistr. 1110), where Pindar would have written 
 ελκόμενοι. The relation between ^ελγειν and ελκειν is farther shown 
 by a comparison of the forms %έλγητρον and ελκη^'ρον, θελκτικός 
 and ελκτικός, Ο^ελκτήρ and ελκητήρ. 
 
 465 We must also say something of the adjective %'ελεμοςι which 
 occurs only in the following passage of ^schylus (Suppl. 1006): 
 
 τΐοταμονς δ\ di δια χώρας 
 ^έλεμον τίώμα χεονύιν, 
 Λολύτεκνοι, λιπαροΐς χεύ- 
 μαΰι γαίαΟ 
 τόδε μειλίΰύοντες ονδας, 
 
 where the Chorus is not speaking of the Nile (as Passow supposes), 
 but of the rivers of Argolis, in opposition to the river of Egypt. The 
 glosses of Hesychius are (1) Ο'έλεμον. οικτρόν. ηΰνχον. (2) ^ε /La- 
 μως. ηΰύχως. οικτρώς. (3) ^'ελερόν. ^ελκτόν^ και το ^^ελγον τα 
 όμματα liti κακώΰει. In the first two, we must substitute φίλιον, φι- 
 λίως, for οικτρόν, οικτρών, just as %ελητρον ("errore pro ^ελκτρον,'' 
 Pors. not. MS.) is afterwards rendered φίλτρον. The third gloss is 
 also corrupt: we must read Ο'ελκτρον, Ο'ελκτόν, &c. The other 
 meaning, ηβνχον, is clearly that borne by Ο'έλεμον in the passage of 
 ^schylus; this sense has been derived from -Ο'έλω in much the same 
 way as that of εκηλος from εκών, which we pointed out before (above, 
 § 273): the word can hardly be compared with γη εΟ^έλονΰα and vo- 
 lentia rura, quoted from Xenophon (G^con. v. 12) and Virgil {Georg. 
 Π. 500) by Toup {Emend, in Suid. i. p. 285). 
 
 466 We believe, then, that βούλομαι has no etymological con- 
 nexion with ε^'ελω. This at least we consider certain, whatever may 
 be thought of the derivation we are about to propose for the former 
 word. Every student is aware that there are many words in the 
 Greek language which begin with the syllable βον-. In most of these 
 words it is customary to explain this prefix from a gloss in Hesy- 
 chius: Bov. το μέγα καΐ τίολν δηλοΐ. Αάκωνες. Thus βουλιμία is 
 translated "violent hunger," βούπαις, "a big boy." We remark, in 
 the first place, that Hesychius assigns this prefix to the Laconians. 
 
CHAP, v.] USE OF AUXILIAEY VEEBS IN GEEEK. 699 
 
 We attach no weight to the etymological guess-work of Plutarch 
 (Sympos. vi. 8) : το μεν ovv βονλιμον εδόκευ μεγαν η δημόΰίον ατίο- 
 ΰημαίνειν και μάλίΰτα itaQ ημίν τοις Αίολενβυ αντί τον β τω Λ 
 χρωμενοίς' ον γαρ βούλιμον, άλλα τίονλιμον οίον τΐολνλψον τίάλαν 
 ονομάζομεν. We believe that Hesychius had some good reason for attri- 
 buting this prefix to theLaconians ; what this reason was we will endea- 
 vour to show. The Spartan youth were divided into classes, which bore 
 the same names as their flocks and herds, that is, were called after the 
 first objects of classification in a primitive state. The larger divisions 
 were termed άγέλαι, a word generally applied to herds of oxen ; the 
 smaller Ιλαι, a word in its ordinary acceptation denoting a troop of 
 horse. There are two analogous adverbs corresponding to these two 
 words, άγεληδόν and ϊλαδον, both used by very old writers. Now it 
 appears that in Sparta the άγελη was called βονα (= αγέλη τίαίδων 
 Hesych.); and its chief was termed βονάγορ (=άγελάρχης, b της αγέ- 
 λης άρχων ηαΐς. Αάκωνες. Hesych.). From the form βαγός, which is 
 found in Laconian inscriptions, Bockh is disposed to infer that the ν of 
 βονάγορ, which is also written βοαγός, represents the initial digamma 
 of the termination (Corpus Inscript. Vol. i. p. 612). It appears to us, 
 from the form βονα, that the digamma must have belonged to the 
 first part of the compound. There are two other words referring 
 immediately to this political division: Bovoa. αγέλη — τετάχα^" αϊ 
 βονοαι. άντΙ τον βον(56αι%.τ.λ. Etym. Magn., according to the admi- 
 rable emendation of Hemsterhuis ; and βνμβοναδεΐ. ντίερμαχει. Αά- 
 κωνες (Hesychius). It will not be denied that the syllable βον- in these 
 words is the element of βονς. If so , the name given in Sparta to a 
 body of young men was literally the same as that borne by a herd of 
 oxen. The connexion of βο^ή, " the war-shout," with βονς and βοΓίς, 
 has been mentioned above. We have here a transition from agricul- 
 tural to pohtical ideas, just as the step in that case was from agricul- 
 ture to war. We have endeavoured to show on a former occasion 
 that military arrangements were the basis of all the organization of 
 a Doric state, so that the transition is the same in both cases. It 
 might be asked whether the word όνμβοναδεΐ means "to shout toge- 
 ther," from the one meaning, or "to herd together," from the other; 
 from whichever of the two meanings it is derived, it is evidently a 
 synonym for βοη%'εΐ, and as such is a striking confirmation of the 
 supposition, that the digamma appeared in βο/^ή. We recognise the 
 same meaning in βονγάϊος==^βο^^-γάϊς, "delighting in battle." That 
 the first syllable of βονβρωβτις and βοντίρηΰτις refers to cattle is 
 generally acknowledged, and the same connexion with βονς is admitted 
 in βονχιλος λειμών {M^ch.. Suppl. 540), compared with the ίτΐτΐομανης 
 λείμών of Soph. Aj. 143. A βοντΐαος was a boy of the βονα, or "of 
 
700 USE OF AUXILIABY VEEBS IN GEEEK. [BOOK IV. 
 
 the herd." Now when we reflect that the ^olian form of βονλη was 
 βόλληΙ(τηη βονλης ντώ ΑΙολ^ων βόλλας Λροβαγορενομενης^νΐηίατο}ι, 
 Qucest. Bom. p. 228), when we call to mind that we have in old Latin 
 both boo and bovo, both boarius and bovarius, both hoatim and bova- 
 tim^ and that there were two old towns on the Appian way, i. e. in 
 the midst of the old Pelasgian population of Italy, one called Bola, 
 the other Bovillce, and that bovile is the old Latin for an ox-stall, we 
 can hardly refuse to adopt an explanation of βονλη, which, while it 
 accounts for βόλομαι as well as βονλομαι, is consistent and intelligible 
 in every other respect. Only suppose that βονλη means an assembly, 
 that it is another form of βονα , which we have seen applied to men, 
 and we have every point about βονλομαι satisfactorily explained. It 
 may be asked , How can a word which means an assembly come to 
 signify "a desire of the mind"? We might just as well ask, why 
 αγών., the primary meaning of which is a place of assembly (for exam- 
 ple, Iliad VII. 298), afterwards came to signify not only the assembly 
 itself, but the object of a particular kind of assembly (public games); 
 also a fit and proper time for any thing (see Yalckenaer ad Phceniss. 
 591); and, finally, the violent emotions of an agitated mind (Thucyd. 
 vn. 71); or, to take a case exactly in point, it might as well be askec' 
 why consilium, which originally meant nothing but a coming togeth' 
 just as exsilium means a going out, should not merely signify κ 
 assembly of men who have come together to deliberate, but also bear 
 every other sense of βονλη. This derivation explains a great many 
 peculiarities about the word βονλομαι. In the first place, it shows us 
 why it is a deponent verb , why it has no active form. We think it 
 scarcely necessary to mention, that βονλη is antecedent to the verb: 
 βονλομαι, therefore, is properly "I am one of a βονλη,^^ i.e. βονλενω, 
 save that βονλενω always expresses a more decided, deliberate purpose 
 than βονλομαι, which, in its original sense, perhaps bore the same 
 relation to βονλενω that the deponent consilior does to the active 
 consulo. 
 
 467 From such a noun as βονλη one would expect to have a verb 
 βονλάω, as from τιμή, τιμάω, and indeed we find traces of such a verb 
 in all the tenses but the present; — thus we have βονληΰομαι, βεβον- 
 λημαι, ηβονλη^ην. The same sort of expectation is entertained and 
 justified with regard to ε^ελω, the oldest form of which (as we have 
 rendered probable) is J^ελώ : for we have ε%εληόω, η%ελη(5α. We be- 
 lieve the loss of the derivation-syllable to have been occasioned by the 
 very common use of the present tense of both verbs ; it is this tense 
 alone that is employed as a mere auxiliary. With regard to the pecu- 
 liarity in the augment of the aorist ηβονλη^ην, we think Buttmann's 
 
CHAP, v.] USE OP AUXILIARY VEEBS IN GREEK. 701 
 
 explanation is satisfactory. He observes (ad Flat. Georgiam ed Hein. 
 dorf. p. 522), that, in verbs of cognate signification, we find analogous 
 irregularities of form : thus we have the curious futures εδομαι and 
 τίίομαι^ the futures formed by the diphthong bv from verbs in εω, as 
 ρευίίω, %Βν6ω^ νενΰω, &c., and the anomalous infinitives κνηΰΟ'αί and 
 -ψην; thus also the verbs βονλομαι^ όνναμαο, and ^ελλω, which he 
 remarks (AusfilhrL Sprl. §. 83, Anm. 8, note), have something ana- 
 logous in their signification, make ηβονλόμην, ηβονλη^ην^ ηδυνάμην, 
 ηδννήΟ'ην, η μέλλον : and he conjectures that the augment may have 
 been suggested by the sound of η^'ελον, which is also connected with 
 them in meaning. 
 
 468 The use of βον- as an intensive prefix may be compared with 
 that of LTCTto- in ίτΖΛόκρημνος, ίτττίομάραΟ'ρον, ίτίτίοΰελίνον, ίπττοτν- 
 φία, LTtJtOTtOQVog. The same idea of weight or strength is conveyed by 
 the word βovg in the proverb βονς ετά γλώόΰ^ , as indeed appears 
 clearly from the words κρατερω ποδΙ in Theognis (815), and the 
 epithet μεγαζ, which is joined to it in -^schylus (Agamemnon, 36). 
 
 469 That the word βοννός was a strange and unusual one, ap- 
 pears from the explanation which Herodotus (iv. 199) thinks proper 
 to give of it, and, indeed, from the express statement of Phrynichus 
 (p. 355 Lobeck). Herodotus considers it a Cyrenaean word: των 
 νπερ^αλαύβίδίων χώρων τα μεΰα, τα βουνονς καλεουΰι. Valckenaer 
 thinks (ad Herod, iv. 158), that it was taken by the Dorians to Sicily, 
 and learned there by JEschylus, but from the manner in which it is 
 introduced by this poet (SuppUces , 101) we have no doubt that he 
 considered it an African word, and used it as such, because his chorus 
 consisted of African damsels. The passage , which is almost hope- 
 lessly corrupt, stands thus in the MSS. : ίλεομαι μεν Άττίαν βοννίν, 
 καρβανα δ' ανδαν ενγα κοννίζ Λολλακιδ' εμπιτνώ ξνν λακίδυ. As 
 we cannot believe that the second person of a verb would be inserted 
 in the adversative clause to ίλεομαι μεν, and before εμτατνώ, which 
 is the legitimate antithesis, and as the last two syllables of τίολ-λακίδ^ 
 seem to be suggested by the λακίδι which follows , to say nothing of 
 the feebleness of such a particle as Λολλάκυξ in a sentence expressing 
 the visible act of the suppliants, we would read and arrange the pas- 
 sage as follows: 
 
 ίλεομαί μεν ^A%iav βοννιν, 
 
 καρβανα δ' ανδαν ενάκοον 
 
 εΙς %6λιν χέονΰ' εμτίίτνώ ^νν λακίδυ 
 
 λίνοίΰιν η Σίδονία καλντίτρα. 
 In inscriptions we find ευήκοος as an epithet of protecting deities 
 (Bockh, C.I. II. p. 422). For the phrase χεονβ' ανδάν, we may compare 
 
702 USE OF AUXILIAEY VEEBS IN GREEK. [bOOK IV. 
 
 Sept. c. Thel•. 73 : φ%'όγγον %80v6av. Suppl. 626 : ενκταΐα χεον6ας\&ηβ. 
 for Eig τίόλιν χεονΰα we have Agam. 230 : dg πέδον χίονΰα ; and the 
 whole passage, thus altered, will be strictly parallel to Pers. 120: μτ] 
 TtoXig πν^'ηταί καΐ το Κί<5<5ινον πολιό μ' άντίδοντίον εόΰεται^ όά, 
 row' 87tog [this καρβάνα ανδάν, i.e. 6α] γvvaiκoπλη^7]g oμιλog 
 άτίνων^ βvύ(^ίvoig δ' εν Ttrnkoug τΐέΰτ] λaκίg. But though the Cyre- 
 nseans may have been remarkable for their use of the word ^ovvig, 
 it was rightly referred to a Greek origin by the old grammarians. 
 Thus we find in the Etymologicum Magnum: Bovvig και ^ovvltigf 
 η γη' εϊρηταυ δε TCaqa Tovg βovvovg^ βοννοί δε είΰιν οι ν^ΡηλοΙ και 
 6Qώδειg καΐ γεώλοφοι τόποι, πάρα το βαίνειν ανω. Though we take 
 the liberty of setting aside this etymology, we still think that the 
 origin of this word is to be sought in the Greek language. If it means, 
 as appears from all the grammarians, an elevation, it may be com- 
 pared with the German Buhne. We are inclined to believe that its 
 real origin may be derived from a comparison of the following glosses 
 in Hesychius; {!) Bovvig. γη. Ai(5%vlog^ (2) Bovvog. ΰτιβάg. Κύ- 
 πριοι. (3) Βοννοί. βωμοί. (4) Bo ν λάκα. βόλον όνομα (resid βώλον 
 with Toup, Emeiidat. νι. p. 30 and 274). (5) Βωμώ. ovτωg η Μά- 
 κριg ώνομάζετο. (6) Βω^'εΐν, ομιλεΐν. βοηϋΊιν. (7) Βώλαξ. βώλog. 
 γη. (8) Β ω λό ν αι. οι μεν κολών ag. οι δέ^ το Κίλλαιον άκονονΰι, δια 
 το άνακεχώΰ^'αι,παρα Σοφοκλεΐ. (9) Βώλοι. γη. (10) Βωλώρνχα. 
 την ύνν. Aάκωvεg. (lί)Bωμιηg. οι περί τovg Bωμovg κaλovμέvovg 
 λόφovg oικovvτεg. (12) Βών. βονν. οπλον. (13) Bωvίτag. τovg εν 
 άγρω, οι δε βovκόλovg,η άγρoίκovg. (14) Bog. άΰπίg. πελτη. βνρΰα. 
 (15) Βώΰομαι. βοήβομαι. επικαλέΰομαι. (16) Bω(Jτηρεg. voμεΐg. 
 (17) Βωύτρειν. βοάν. καλεΐν. επικαλεϊύ^αι. From (1) we learn that 
 βovvίg denotes "the earth" in general; from (2) that /3ovi^dg means a 
 heap of straw, for instance, litter for an ox ; from (7), (9) and (10), that 
 βώλog is a synonym for βovvίg in the sense of γη^ particularly among 
 the Laconians (10); from a comparison of (1) with (13), and of (4) 
 with (7), we see that βovvίg may be written βωvίg, and conversely 
 βώλα^, may be written βονλα^\ from (3) we see that βcoμ6g is a sy- 
 nonym from βovv6g'^, from (8) and (11) we infer that βώλog and βω- 
 μόg may be translated, the one κολώνη, the other λόφog: now it so 
 happens that both these words are used by Eustathius (ad Iliad. λ\ 
 p. 880) to explain βovvόg', he says: ^^έ κολών η και λόφog αν λέ- 
 γοιτο κάί βovvόg, όπερ 'Ήρόδoτog μεν Αιβνων λέξιν είναί φηβιν, 
 AUiog δε Aiovvdiog λέγει οτι Φιλήμων επιΰκώπτει το όνομα ag βάρ- 
 βαρον, λόφον γαρ καλονΰι; from (5) it appears that the island of 
 Euboea , which is signified by η Mάκριg (Strabo, p. 445) was called 
 Βωμώ : now we know that the name Ενβοΐα was connected with βovg, 
 either on account of its pastures or from the myth about lo : τάχα d' 
 ώΰπερ βoog ανλη λέγεται τι αντρον Ιντ^ πρog Αιγαίον τετραμμένιυ 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GBEEK. 703 
 
 TiaQaXla otcov την "Ιω τεκεΐν φα6ιν "Ετταφον, καΐ η νηΰος άπο της 
 αντηξ αΐτίαξ εΰχε τοντο τοννομα (Strabo, p.445),andIZbAi;/3ot« is an 
 epithet of Ceres (see Lobeck, Paralip, p. 465); from (12), (13), (16) we 
 observe that βώς is a form of βονς] from (6) we see that βω&εΐν may 
 stand as well for βοναδεΐν=6μίλεΐν as for βοηΟ-εΐν; and from (6), (15) 
 and (17), we discover that the Laconians could contract βοη- into βω-. 
 
 470 We think that, after this comparison, no doubt ought to 
 remain upon our minds as to the Greek origin of βοννόξ^ as to its 
 connexion with βονς, and as to its affinity with βωμός and βώλαξ. 
 We would, in addition, point out that the connexion between the land 
 and the cattle , which are used for tilling it , is immediate. There is 
 indeed reason to believe that in the oldest languages of the Indo- 
 Germanic family, the names of the cow and the earth are commutable, 
 the latter being derived from the former, which was the symbol of 
 fruitfulness and agriculture. (See the IndiscJie Bibliothek, n. p. 288, 
 and Bopp's Glossar, Sanscrit, p. 109, ed. alter.). The Sanscrit go, 
 nominative gau-s (masculine and feminine), signifies "a bull" or "a 
 cow." In the feminine it also denotes "the earth." There is another 
 Sanscrit word, hhu-s, which is confined to the latter meaning. Now it 
 is singular, that while the Sanscrit go , old Latin ceva (the name for 
 "a cow" at Altinum on the Adriatic, Columella, vi. 24), Persian ku., 
 Frankish cJiuo, and Anglo-Saxon cw, all meaning "a cow," agree with 
 one of the Sanscrit names for "the earth," the Greek βονς and the 
 Latin bos perfectly coincide with the other. Thus, to take the cases 
 which correspond in the three languages, we have 
 
 bos 
 bovis 
 bovi 
 abl. bove 
 3oFav) bovem 
 
 It is also remarkable that γη, Doric γά , the common name for "the 
 earth," coincides with the other Sanscrit name for the earth, which also 
 signifies "a cow"^." As the nominative of the latter word is gaus, we 
 should expect gavam in the accusative; whereas we have gam or gan, 
 which is identical with the Doric accusative γαν. The Greek student 
 will recollect that there is a longer as well as a shorter form of the 
 Greek word for "the earth," namely, γαία as well as γη. Lastly, it 
 should also be mentioned, that the Germans have Gau, "country," by 
 
 Nom. 
 
 bhus 
 
 βονς 
 
 Gen. 
 
 bJiuvas 
 
 βο^ός 
 
 Dative andl 
 Locative / 
 
 bhuve 
 
 βοΗ 
 
 bhuvi 
 
 
 Accus. 
 
 bhuvam 
 
 βονν ( 
 
 * On the connexion between the earth and the cow-horned virgin To, 
 see Mr. Scott's ingenious essay in the Classical Museum, No. 12, pp. 166 
 sqq. ; and compare Mr. Paley's preface to his edition of the Supplices of 
 ^schylus, 1851, p. vii. 
 
704 USE or AUXILIAKY VEEBS IN GEEEK. [bOOK IV. 
 
 the side oiKuh, "cow," and that our Saxon ancestors spoke of a '■'■hide 
 of land*." It has been mentioned above (§ 284) that 'βονζ in itself 
 means "the bellowing or lowing animal," and is therefore imme- 
 diately connected with βοη^ βοάω. In the same way, we may com- 
 pare gau-s with γοάω and the Hebrew ϊιί^5, mugire, "to low like an 
 ox" (1 Sam. vi. 12; Job vi. 5). But although there can be no doubt 
 as to the connexion of βονς and βοη, βοάω, it must be remarked that 
 the verb was used to express sounds which we consider very di£ferent 
 from the lowing of an ox or the battle-shout of a warrior. Thus the 
 serpent is said to utter a βοή, ^sch. Sept. c. Theb. 365: μίβημβ^ιναΐζ 
 κλαγγαΐΰυν ώς δράκων βοά. Find. 01. νιιι. 40 : sig (των δρακόντων) 
 εΰόρονβε βοάΰαις: and so of the goose, Diog. Laert. n. 36: ov και ΰν, 
 ^φη, χηνών βοώντων άνεχγ} ; Pindar also uses βοή indifferently for the 
 notes of the pipe (Nem. v. 38), the flute (01. iii. 8), or the lyre (Pyth. 
 X. 39). 
 
 471 The word Ο^νμός is particularly interesting from its use in 
 the Eepublic of Plato. It will be recollected that Plato, and Hooker 
 after him , consider the mind as performing the three distinct func- 
 tions, reasoning, willing, and desiring; Plato divides the mind into 
 three independent faculties, by which these functions are performed, 
 namely, Aoyog or λογίΰμός^&νμός^ and ετΐΐ^νμία (Eespubl.iY. p.439D), 
 the first belonging to the το λογυΰτικόν^ or rational part of the soul, 
 the last two to the το αλογον , or irrational part. The ^υμόξ or το 
 ^νμοειδίξ^ however, is not identical with the Ιτα^νμία^ though it is 
 classed with it under the same general head, for it often contradicts it, 
 and assists the reason in governing its unruly attempts to lead man into 
 the wrong path: ονκονν καΐ αλλο^ι^ εφην, πολλαχον αΐβ^'ανόμε&α, 
 όταν βιάζωνταί τίνα τίαρά τον λογιϋμον ετίί^νμίαι^ λοιδορονντά τε 
 αντον κάί ^νμονμενον τω βιαξομενω εν αντω, κα\ ωβτίερ δνοΐν 6τα- 
 ΰίαξόντοιν ξνμμαχον τω λόγω γιγνόμενον τον Ο'νμον τον τοιούτου 
 (Plat. U. S. ρ. 440 α); — τω %νμοειδεΐ τιροβήκει νΛηκόω είναι καΐ ξνμ- 
 μάχω τούτον (τον λογιβτικον) (Ibid. ρ. 441 ε). We translate the word 
 '9"i>/Ltog,"the will," because this term conveys to our minds the idea which, 
 in this passage, Plato evidently attached to the word he made use of; 
 Hooker, too, translates it "will," and Hemsterhuis the younger, la 
 velleite (see also Heber, Bampton Lectures, p. 178). It seems, indeed, 
 that Plato thought he was using the word in a somewhat unusual 
 signification; the first of the passages we have just quoted implies 
 that, in his opinion, "anger" was the primary meaning of the word. 
 
 * It is a remarkable circumstance that the legend of the Dun-cow has 
 arisen from a confusion with the Dena-go^ "Danish district" (cf. Dun-church, 
 &c.), conquered by the Anglo-Saxons ; Cambridge Essays, 1856, p. 54. 
 
CHAP, v.] USE OF AUXILIARY YEEBS IN GREEK. 705 
 
 as indeed is evident from his etymology in the Cratylus (p. 419 e): 
 %νμοζ ατά) της %'νΰεωζκαΙ ζεΰεως της-φνχης εχου αντοντο το όνομα; 
 and from the Timceus (p. 70b): ore ζε^ευε το του ^νμου μένος: he 
 also uses the word to signify disposition in general, as appears from 
 Legg. v. p. 731 b: τοντο ανεν ^νμον γενναίου ψυχή τίάΰα αδύνα- 
 τος δράν, and such is the usual and oldest meaning borne by Ο'υμός. 
 The meaning " anger," though certainly often conveyed by %'υμός, and 
 always by θυμούμαι, appears to be quite a secondary one, and we can 
 only explain Plato's addition of the participles λοίδοροΰντα and %•ν- 
 μούμενον, as an exegesis of ^'υμός, by supposing that, in his abstract 
 way of considering verbs before substantives , he got into a habit of 
 deceiving himself with the belief that the former necessarily preceded 
 the latter, and that ^^υμός actually derived its primary meaning from 
 ^υμονναυ. This was far from being the case ; indeed %^υμ6ς was a most 
 proper word for his purpose, and, if it had been understood according 
 to its old usage, he needed no addition to qualify it for the signifi- 
 cation of "will," with which we find it used in the Bepublic. That this 
 was its meaning in Homer, we see from the phrases, %'υμος άνώγεο, 
 κελεύει, κέλεταί με, η^'ελε or %'υμω η^'ελε, like ΐετο Ο'υμω (Herod, 
 ν. 59); and indeed this signification of %'υμός is sufficiently clear from 
 the word ετα^νμία-, which Plato employs in opposition to it: ετίΐ^υ- 
 μία means "a setting one's mind upon a thing," "directing one's 
 wishes to it," quite in accordance with the old sense of %'υμός. 
 
 472 We may settle the etymology of this word without any diffi- 
 culty: at the same time we will endeavour to decide some troublesome 
 questions which have been started regarding certain words of the same 
 family with O'l'^og. That the ideas of placing and being placed, of 
 setting and sitting, sedare and sedere, are intimately connected, will, of 
 course, be at once conceded; and we hope that those who are inclined 
 to adopt what has been said about the influence of suggestion by con- 
 trast on the formation of words , will be willing to allow that words 
 significant of remaining and moving may have a common origin: if 
 instances are required we may compare μένω, μνήμη, μένος, mens, 
 maneo, with μάω=μέν-ω, μέμαα^==μέμονα,αύτόματος=αύτόμεντος, 
 &c. The affinity of these forms is clear from έ-γεν-όμην, γέγαα, and 
 γέγονα (above, § 114). The relation between μέν-ω and μέλ-λω (for 
 μέλ-ι/ω) is the same as that which subsists between κέν-το and κέλ- 
 ετο, between γέν-το and ελ-ετο^ &c. That μέν-ω agrees with its other 
 form μά-ω in expressing an expectation of, or a mental impulse 
 towards, any object, as well as the meaning of fixity or continuance, 
 which it generally bears, will appear from the following passages. 
 Romer, Iliad xY. 599: 
 
 ZZ 
 
706 USE OF AUXILIAEY VERBS IN GREEK. [bOOK IV. 
 
 το γαρ μένε μητιετα Ζενς 
 νηοζ καιομενης ύελαξ 6φ%αλμοΙ<3ι Ιδεΰ&αι. 
 Sophocles, Philoct. 511: 
 
 εγώ μεν 
 
 ενΟ'ατιερ εηυμεμονεν 
 
 επ ενύτόλου ταχείας νεώς τίορενΰαιμ αν ες δόμους. 
 And the idea of remaining or abiding may be conveyed by forms 
 which have lost all traces of the original suffix v-: as in ^schylus, 
 Choeph. 464, where we ought to read with Butler: 
 
 δώμαβιν εμμοτον 
 
 τοΐΰδ' άγος, ονδ' νπ άλλων 
 
 εκτο^εν, αλλ^ νπ αυτών 
 
 αΙών αναιρεϊν. 
 
 That the form εμμοτος , which has nothing to do with μότος , " lint," 
 really belongs to this class of words , is clear from the antistrophic 
 αμοτον-, which is constantly used with words containing this root, 
 as μεμάαΰί, μεμαώς, &c. The most difficult word into which this 
 root enters is perhaps the adjective άμαυμάκετος, of which Gottling 
 (ad Hesiod. Theog. 319) and Lobeck {Pathol, p. 374) have proposed 
 erroneous derivations, the former connecting it with μάχομαι, and 
 the latter supposing that it is for άμαίμακτος, as αλάμτίετος for αλαμτΐ- 
 rog, with the intensive a-. This opinion is also adopted by Doderlein 
 {Gloss. Horn. No. 140), who renders the word "violent," "raging," 
 "impetuous." As this rendering is inconsistent with the applica- 
 tion of the epithet to a mast in the Odyss. xiv. 311, ίότον άμαι- 
 μάκετον νηος κυανοτίρώροιο , Doderlein supposes that the poet was 
 misled by a false analogy, and considered the word as a synonym and 
 derivative of μακρός^ just as Apollonius Rhodius calls a mast μέγας 
 (i. 563) or περιμήκης (iv. 1832). It appears to us that the word is 
 derived from μαομάόΰω, which is formed from μαιμάω, a reduplication 
 of μάω, and signifies "to quiver," "move rapidly," "pulsate," &c. 
 We have μαίμαξ (ταραχώδης Hesych.), and the fifth month in the 
 Athenian calendar, which commenced the winter, was called μαιμακ- 
 τηριών^ from Ζευς μαίμακτής, i. e. the disturbed state of the weather. 
 As applied then to the mast of Ulysses άμαίμάκετοςϊΆη8ί mean strong, 
 stiff, unyielding, in the midst of the waves. From this sense of im- 
 movable, unyielding, comes the meaning "irresistible," "invincible," 
 which is applicable to all the passages in the ancient poets in which 
 this adjective occurs. Thus it is an attribute of the irresistible chimasra 
 (Horn. II. VI. 179, xvi. 329), and of the fire which it breathed 
 (Hesiod, Theog. 319), or of fire in general (Soph. (Ed. T. 171); it is 
 predicated of the deep sea (τΐόντος)^ which no wind can shake to its 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GREEK. 707 
 
 bottom (Pind. Pytli. i. 14; cf. the epithet δνΰήνεμος, Soph. Antig. 587, 
 applied to the sand at the bottom of the sea); of the irresistible wrath 
 (μένος) of the goddess of destruction (Pind. Pyth. ni. 33) ; of the irre- 
 sistible collision (κι,νη^'μός) of the Symplegades (Pind, Pyth. TV. 308) ; 
 of the invincible trident (τρίόδονς) of Neptune (Pind. Isthm. vii. 35) ; 
 of the Eumenides, as immovable in their purpose or not to be med- 
 dled with, according to the two applications of ακίνητος (Soph. (Ed. 
 Co?. 12 7, where the Scholiast writes ακαταμάχητων η άτΐροΰτΐελάότων). 
 In the general sense of thinking or caring about any thing, μεν-ω, 
 μελ-λω, and the impersonal μέλει, may be compared with the Gothic 
 munan^ German meinen, old Nordish man (Grimm, i. p. 926). The 
 same idea of thinking about a thing enters into the subjective nega- 
 tion μη (above, p. 361). All these words, as we have said more than 
 once, are connected with the root of the first personal pronoun, and a 
 feeling of subjectivity or self lies at the root of them all: compare, 
 for instance, the very similar form μόν -og (above, p. 282). To return 
 to the question before us: the root %'ε- means "to place" {τί-%η-μι), 
 %a- or %o- means "to sit" (&α-ά(56ω, Ο^ώ-κο^), and Οό- or %'ε- means 
 "to run" (d'o-og, ^έω), and these are ultimately identical. Modern 
 scholars have not observed this fact, and have therefore got into great 
 difficulties about some words of this class. The word which has caused 
 them most doubt is ^οάξω. That this word may signify "to move 
 quickly," whether in a transitive or intransitive sense, appears from 
 the following passages of Euripides: (1) Transitively, Bacchce, 65, 
 ^Όαξείν Ttovov; Iph. Th. 1142, ^οάζειν τϋτέρνγας; Orest. 355, ^οά- 
 ζων 6ε τον μελεον; Her. F. 382, ε^όαξον κά^αυμα 6ΐτα γενυΰί. 
 (2) Intransitively, Bacch. 216, κλνω — γυναίκας ^οάζειν ^ιόννΰον; 
 Troad. 349, 507, μαινας Ο^οαζονβα — ^οάζει δενρο δρόμω; Orest, 
 1542, ΟΌαζων αϊ^'ερος ανω κατΐνός; Phoeniss. 800, ΐΛΛείαίόυ^οάξευς. 
 And its connexion with 0ΌΟς is indisputable. At the same time it is 
 equally clear that in ^schylus Suppl. 610: 
 
 vjt αρχάς d' ούτινος ^ϊοάζων 
 
 το μείον κρειΰόόνων κρατννευ. 
 
 ούτινος άνωθεν ημενου όεβευ κάτω. 
 
 and in Soph. (Ed. Tyr. 2 : 
 
 τίνας 7Co%^ έδρας τάΰδε μοι ^Όαζετε — 
 the meaning of the word is "sit:" and, indeed, Plutarch (de Audiendis 
 Poetis, p. 22 e) and the Etymologicum Magnum (v. ^ώκος, p. 460, 10) 
 acknowledge this meaning in Sophocles. But modern scholars, from 
 not perceiving that the same root may convey contrasted ideas, have 
 gone wrong about this word; for while Buttmann, on the one hand (in 
 his LexiloguSj ii. p. 105), thinks it necessary to suppose that ^Όαξω, 
 
 ZZ2 
 
708 USE OF AUXILIAEY VEEBS IN gIiEEK. [bOOK IV. 
 
 "to move violently," is derived from %ΌΟς, and that ^οάζω, "to sit," 
 comes from %'εω, τίΟ'ημί ; Hermann, on the contrary (ad Sophoclis lo- 
 cum), denies the existence of the latter signification, and twists the 
 two passages of ^schylus and Sophocles to a very forced and un- 
 natural meaning. The gloss in Hesychius shows, not, as Hermann 
 'supposes, that he was puzzled by the word, but that it possessed a 
 very extensive range of meanings : %•οάξεί' τρέχει, μαίνεται, ΰκιρτα, 
 κτείνεί (read εκτείνει), ότίενδει, ταράττει-, κά^ηται, χορεύεις αννει, 
 ηδεται, τελεί, τΐλάττεί, φοβείται, τίλανάταυ, Ο'εοφορεΐταί. That most 
 of these meanings are directly, or by implication, conveyed by the 
 word ^Όαξω, might easily be shown by a more minute examination 
 of the whole family to which it is related. It is generally agreed that 
 in ^schylus, Choeph. 853, and Euripides, Medea, 1409, we ought to 
 substitute corresponding forms of εΛυ^εάζω for the ετίί^οάζονύα and 
 ΙτίΐΛ^οάζω, which appear in the ordinary editions. 
 
 473 The root ^l^a-, %'ε- or ^o- , which we are now discussing, oc- 
 curs in another, perhaps an older form, as (5a-, 6ε- or όο-. From 
 many indications in the classes of Avords, into which one or other form 
 of this root enters, it is probable that the syllable was generally closed 
 by a /. We are disposed to believe that the ground-meaning of the 
 root is "place" or "make," with which the second meaning, "be 
 placed" or "seated," is intimately connected. The most important 
 word into which the root enters with this meaning [is ^'εός, and its 
 oldest form was ΰίός, which was used by the Lacedaemonians to the 
 latest time , and to which &εός , as has been already remarked, bears 
 the same relation that Ο-άλαβόα does to ΰάλαΰΰα, and ϋ^έλω to ΰέλας. 
 In compounds, the Laconians pronounced it tog. Thus Bacchus was 
 called ενίος for ενόίος, just as they said Λοιηαί ior jtoirjuai, and βονοα 
 for βονβόα {Etymol. Magn. p. 391). It may be interesting to know 
 that there are in the Sanscrit language representatives both of 6ίόξ 
 and %ε6ξ. To the latter the common word deva corresponds, to the 
 former gwa, the name, indeed, of a particular god in the Hindu my- 
 thology, the god of fire (Bohlen, Das alte Indian, i. p. 206), but still 
 only a general name for the deity (Bohlen, i. pp. 148, 206), as appears 
 from the neuter noun givam, which means "happiness," "prosperity," 
 the lot of the gods (Arjuni Reditus, v. 19). These two words and the 
 Latin divus , if, as we have no doubt is the case , they are identical 
 with the Greek ^εόζ, 6iog, would incline us to suppose that the 
 digamma had slipped out in their Greek synonyms. In attempts to 
 represent directly in Greek characters the Sanscrit and Persian names 
 for the Divinity, this digamma sound is represented by the i;, or a 
 lengthening of the root-syllable. Thus Hesychius: ^ενας, τον£ άκά~ 
 
CHAP, v.] USE OP AUXILIARY VERBS IN GREEK. 709 
 
 Kovg d'Eovg. Μάγοι (where Reland proposes to read avaxag^ soil, rovg 
 ^tooKOVQOvg); and Athenseus, i. p. 27 d: on παρ' 'IvdoLg τιμάται 
 δαίμων, cog φη0ι XaQTig 6 Μυτιλγιναΐος , og καλείται 2oQoad8iog' 
 ερμηνεύεται δε Ελλάδι φων^ oivoTtoiog, where he is evidently al- 
 luding to Suryadeva, "the sun-god," who was, as we have elsewhere 
 shown, the same as the god of wine. The Indians had, properly speak- 
 ing, no wine-god (Schlegel's Indische Bibliothek, i. p. 250). If we may 
 be allowed to hazard such a conjecture, the epithets in -^schylus, 
 Persce, 86, Σνριον άρμα διωκων, and in the oracle quoted by Hero- 
 dotus, VII. 140, Σvριηγεvεg άρμα διώκων, where Xerxes, or the Per- 
 sian war-god , is spoken of, refer to the sacred chariot (το άρμα το 
 ίρόν, Herod, νπ. 55), which always attended the Persian armies on 
 their march, and which, though called the chariot of Jove by Herodo- 
 tus (vii. 40), may have been identical with the chariot of the sun, 
 Ήλίον άρμα λενκόν, which Xenophon represents as following it (Cyrop. 
 VIII. 3, § 11); for it was drawn by white horses, which were conse- 
 crated to the sun (Herod, i. 189, vii. 113), and a remarkably beautiful 
 horse, sacred to that god, was led behind it (Q. Curtius, iii. 3, § 13). 
 The decision of Darius's claim to the throne by his horse's neighing 
 at sun-rise (Herod, ni. 84) seems to have reference to the same sym- 
 bolical connexion of the horse with the sun *. The epithet ΰvριηγεvέg 
 appears to be quite equivalent to the Sanscrit surya-ja or "sun-born." 
 It is likely that the Greeks would have heard of the Persian name for 
 this chariot, and that this name would be explicable from the Sanscrit 
 is consistent with all that we know of the old language of the Per- 
 sians. If this interpretation is not correct, we shall find it difficult 
 to explain why the Syrian chariot should be so specially mentioned in 
 connexion with the army of Xerxes. Still less shall we be able to 
 explain the distich , quoted from a Spanish MS. by Blomfield on the 
 passage of the Persce, in which it is assigned to an Indian : 
 
 "Ivδog οδ* άνηρ Τροί^ Σνριον άρμα διώκ,ων 
 Πovλvδάμag κεΐμαι vεκρbg επϊ τίατρίδι. 
 
 The substitution of the ethnical adjective Σνριος, which the Greeks so 
 well understood, for a foreign term which conveyed no significance to 
 them , is quite in accordance with a tendency which has often been 
 observed. Nothing in fact is more common than this transmutation f» 
 
 * See a paper on this subject which we have contributed to the Jour- 
 nal of the Royal Asiatic Society, Vol. xvi. Part i. pp. 1 — 7. 
 
 t "The desire of converting a strange into a familiar sound is a fre- 
 quent cause of corruption in all languages. Changes of this sort are 
 usually made without any reference to the meaning of the word. Thus 
 the French rondeau became round 0, and bourdon became burden (of a 
 
710 USE or AUXILIABY VEEBS IN GEEEK. [bOOK IV. 
 
 as when the Westphalian peasant, who was sent to the chemist's shop 
 for unguentum Napolitanum^ asked for angewandten Napoleons^ or to 
 take a similar illustration, when we find the poisonous hyoscyamus or 
 "hog-bean" metamorphosed through he^hen (see the commentators on 
 Shakspere, Hamlet^ Acti. Sc. 5, where we should read "with juice of 
 cursed he'ben in a vial "), into " hen-bane." The supposition that the 
 root we are discussing was %t^- is farther confirmed by the future 
 %'εν0ομαί=&εΓΰομαί from 0'εω=0'έ/ω, by the forms ^α^άϋόω (for 
 in words of this kind we may presume that there was once a digamma), 
 ^'αβακόν (^^^d'aFaxov), which Gregor. Corinth, (p. 354) quotes as the 
 Doric form of Ο'ακοι;, ΰε-β-ω=ΰε^ω=ΰ8ύω=6είω==ΰαί-νω, ΰονμοα 
 =ΰόΓομαί (which occurs in ApoUonius Rhodius), βοβεΐν=^(5οΡεΐν, and 
 Z8vg=dy8^g. Heronotus tells us (ii. 52) that the dame d'80g was Pe- 
 lasgian, and was derived from τί-^ψμι: ε^νον δε Jtavta Λρότερον οί 
 ΠελαβγοΙ ^'εοϊόί ετίενχόμενοι , ώς εγώ εν ^ωδώνγ, οΐδα άκονόας, 
 ετίωννμίην δε ονδ' οννομα ετίοιενντο ονδενΐ αυτών' ον γαρ άκη- 
 κόεΰάν κω. %'80vg δε τίροΰωνόμαβάν dφεag άτίο τοιούτον, οτυ κόΰμω 
 ^'έvτεg τα τίάντα ^ρήγματα καϊ itauag νομας είχον. Plato derives the 
 word from %'έει,ν^ from the apparent motion of the heavenly bodies which 
 were the objects of worship in an elementary religion {Cratylus^ p. 397 
 c, d). The Etymologicum Magnum (p. 445, 42) gives both etymologies. 
 Clemens of Alexandria (Strom, i. ad fin.), like Herodotus, derives ^fbg 
 τταρα την %ε6ίν καϊ τά^υν καν την δυακόΰμηΰίν, and Eustathius too 
 says (ad Iliad, p. 1148) that ^adg is δ τίάντα τL%ειg καϊ τίοίών. We 
 agree with this etymology, as well from its obviousness, as because the 
 analogy between ϋΤαά^ο^, who was the god of the Pelasgians of Boeotia, 
 and was identical with Κάδμι,λος, one of the Cabiri (Muller's Orchome- 
 nos,-p. 216), and κεκαβμαι, κέκαδμαί, κέκαδον, κεκάδηκα, κa^'aρ6g, 
 κό6μog, &c. (above, § 267), would at once lead us to it. It is not impos- 
 sible that the same root may lurk in the hitherto unexplained word Gott, 
 "God." There is good reason to believe that the etymology which would 
 connect (τοίί, "God," and gut, "good," is erroneous (see Blackwood, Feb. 
 1840, p. 205) : and we think that while the latter is related to the root 
 γα^"- or γη%'- of ά-γa^'όgJ &c., the former contains the element καδ- of 
 xaXog, so that the compound καλοκάγαθος actually unites the substan- 
 tive God with the adjective good. If this is the case, θεός and "God" 
 mean "the creator;" in immediate connexion with which we have 
 
 song); so humble bee became humble bee, kink-couyh became chin-cough, 
 and gorst-berries, gooseberries. The craig (i. e. throat) end of a neck of 
 mutton became the scrag end; and lustring, a shining silk, so called from 
 its lustre, was commonly called lutestring. Livorno was changed into 
 Leghorn, Coruna into the Groin, and a Prussian fir into a spruce fir.'^ Sir 
 G. C. Lewis, Gloss, of Prov. Words in Herefordshire, p. 89. 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GREEK. 711 
 
 τί^^Ύΐμ,ι, to "make" or "place," καξω, "to arrange," together with ^νω-, 
 *' sacrifice," and (j£j3o, "worship." The last word but one often ex- 
 presses violent motion, but we do not think, with Passow, that this 
 meaning is due to the notion of flaming, blazing , &c., derived from a 
 sacrifice of burnt-ofierings. The meaning of motion became attached 
 to words from this root according to the principle of suggestion from 
 contrast, and there is no class of words in which the meanings are 
 more mixed up together than in this. With regard to the form, ^vca 
 bears the same relation to θε/ω, ^ε^ω, &c., that|3ρ^!ω and εμβρνον do 
 to βρέφος, ^νω to ξίφος, &c. In the words d^ew aud Όόο^, the most 
 prominent meaning is that of motion. The latter is used in the sense 
 of "terrible," "dreadful," when applied as an epithet to ννξ; but the 
 idea of swiftness seems to be included in the word even in this appli- 
 cation of it, for most persons, whose reason has not the full command 
 over their other faculties, are accustomed to look upon that which is 
 sudden and startling as also alarming and terrible. Compare the use 
 oi καταΰτίερχω (Thucyd. iv. 126), &c. Buttmann has clearly shown 
 {Lexilogus, ii. p. 60) that %o6g has also the meaning " sharp," "point- 
 ed," as a synonym for ο^,νς: he appears, however, to be wrong in sup- 
 posing that this was the primary signification, and that ^ηγενν, 0"«(3- 
 <?ov, ταχνς, contain older forms and meanings of the word than^foj: 
 a comparison of the terms "sharp" and "quick" in English, the use of 
 the word "set," [as applied to sharpening a knife, the resemblance of 
 ώτινς and ο^,νς, and the English word "fast," which means both 
 "fixed" and "rapid," will show us that the ideas of fixedness, rapid- 
 ity, and sharpness, are frequently interchanged (see also above, 
 § 218). There can be Httle doubt that, as Buttmann supposes, the 
 ^ψες or lowest division of the subjects of Athens were so called from 
 their being the oldest inhabitants, the Sassen or Insassen (inhabit- 
 ants), and similarly we may recognise the rooot κε- (κεΐμαί, &c.) in 
 ci-viSj Osc. keus. The word Ο'νμός conveys the idea of an eager motion 
 towards any thing, an impulse; and in this sense it agrees pretty well 
 with the second meaning oi ^νω, from which also the idea of anger 
 may easily be derived: the idea of "motion towards" is implied in 
 anger as well as in desire, and hence we have such phrases as Q'v- 
 μονβΟ'αί εϊΰ rt, "to be angry with a thing," as in Herod, in. 52 : ig 
 τοϋξ τοκέας καΙ Ig tovg κρεΰΰονας ζεΟ'νμώό^αι,. The phrase eg κέ- 
 ρας &νμον6%^αυ is of a difierent origin (above, p. 318). 
 
 474 It has been mentioned above , that the initial syllable of a 
 number of words, such as ^ες-φατος, %ες-κελος, %ες-7ίυς, &c., is merely 
 a mutilation of ^εοΐς (§ 310). This explanation is not applicable to 
 ^εύ-μός, which is a simple derivative from the root Ο'ε- now under 
 
712 USE OF AUXILIAKY VEEBS IN GEEEK. [bOOK IV. 
 
 discussion, the <J- being an euphonical insertion, as in deC-^og from deG) 
 (above, § 253). For ^εΰμός we have the by-form τε%'-μός; Q• very often 
 appears instead of before the termination -μός, and vice versa; — 
 thus we have not only ρυΟ'-μός, QV-μόgi but ρν-{3-μ6ς was also used in 
 the sense of όχημα by Democritus (Aristot. Metaphys. p. 1042 Bek- 
 ker); — and rfO'-^og is written for %'Βθ'-μός, just as we have ετε^'ην for 
 ε^έ^'ην. A question may be raised, whether we should connect δεΰ- 
 Λοτης with δέω, like δέΰ-μος, or with τί^^ημι. That the latter part of 
 the word is connected with the Sanscrit patiy "a lord" or "husband," 
 Latin pot-ens, pot-ioVj &c., is sufficiently obvious. For the feminine 
 we have not only δεύ-ΛΟίνα, but also τίότνία, corresponding to the 
 Sanscrit patni, and Ttoxva» We have TCOtvia with a genitive case, as 
 an epithet, like the Diva potens Cypri of Horace; thus Homer, 7Z. xxi. 
 470: TtotVLCi Ο^ηρών. Pind. Pyth. iv. 213=380: TtotVLa οξυτάτων 
 βελών. Now the name for a master of a house in Latin is hos-pes 
 (hos-pit-s)j and it appears from a comparison of the IjSitmJios-tis, hos- 
 tia, Gothic Gasts, Sclavonian Gos-podar or Hos-podar, new High Ger- 
 man ge-gessen^ and the Sanscrit root gJias, "to eat," that hos-pit-s 
 means "the master of the feast." If, then, we remember how often τί- 
 %ημι is used in speaking of a meal (comp. Ίίαρε^ηκε τράτίεξαν, &c.), 
 and that the word ^οί-νη^ together with Ο'ώΰ^αυ and d'cooaud'aL, 
 which are quoted from ^schylus as synonyms for ενωχεΐΰΟ'ϋα, are 
 manifestly connected with the root 0•ε- (comp. %•ώκος, &g.), we must 
 conclude that δεΰτΐότης is a synonym of the Latin hos-pit-s, and that 
 the first syllable contains the root &ε- by a change from -θ" to d , like 
 that which we have pointed out in the inflexions (§ 249) and termi- 
 nations (§ 262) of the noun. "Whether δov-λog is connected with δεω, 
 or bears the same relation to the Sanscrit ddsa^ "a slave" (from the 
 root dds^ "to give"), that δαυλός does to δαβύζ (Pott, Etymol. Forsck. 
 1. p. 190), we do not venture to decide. 
 
 475 The common adjective from δεΰτίότης is δεΰτΐόόυνος. We have, 
 however, another form δεΟΛοΰίΟζ, in ^schylus, Supplices, 848 : δεύ- 
 τΐοΰίω ξυν υβρεί, where Tumebus reads ^εΟτίεβίφ, and Stephens δε(5- 
 τΐοΰύνω. We ihink the word is genuine , and that it is confirmed by 
 the name δεβτΐοΰίοναύτης, given by the Lacedaemonians to the helots 
 who were emancipated and sent on ship-board. Myron (apud Athe- 
 nceum,-p. 271 f) says : πολλάκις ηλευΟ'ερωΰαν Λακεδαιμόνιοι δούλους, 
 καΐ ους μεν άφέτας εκάλεΰαν, ους δε άδεϋπότους, ους δε ερυκτηρας, 
 δεύτίοΰιοναύτας δε άλλους, ους εΙς τους ότόλους κατεταόΰον. It 
 will be recollected , that in the passage under consideration, a herald 
 is endeavouring to force the Danaides and their father to go on board 
 a ship ; and ^schylus, who had often served in the same fleet with the 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GEEEK. 713 
 
 Lacedsemonians, might very well have understood that sense of δεΰτϋό- 
 0ιος with which it was applied to the " slave-sailors." We conceive 
 that the δεβτίοΰιονανται were so called, not because they were Tieri- 
 lium navium magistri, as Casaubon renders the compound, but because 
 they were still liable to the δεόΛοΰυνου αναγκαί (Pers. 479) on board 
 ship, though free from the other liabilities of helotism — in fact they 
 were on the same footing as the "pressed bakers," who are described 
 by Thucydides (vi. 22) as (jiTOTtOLOvg εκ των μνλώνων TtQog μέρος 
 ηναγκαΰμένονς εμμίό%^ονς, i. e. "hired bakers procured from the 
 mills by a compulsory levy" (Bockh, Staatshaush. Bk. ii. c. xxii. p. 
 286 Engl. Tr.). In accordance with this technical expression, JEschy- 
 lus makes the Danaides describe the insolence of the herald as similar 
 to that to which the pressed or helot-sailors were exposed. As the 
 whole passage has hitherto defied the efforts of criticism, we shall 
 make no apology for attempting to correct and explain it in a work 
 which professes to contribute to the more accurate knowledge of the 
 Greek language. We think that the antistrophic portion does not 
 begin until v. 843 ed. Dindorf. The preceding line should be read, 
 with Dobree, 0ον6%ε^ uov6%^ ολόμεναι ολόμεν εφ' αλίαδα, and 
 then the strophes will proceed as follows: 
 
 XoQog. 1 VtO"' ccva τΐολνρντον 6tQ. a. 
 
 2 άλμήεντα τζόρον 
 
 3 δε(77ΐοΰίω ξνν νβρεί, 
 
 4 γομφοδετω δε δόρεί ή-δίώλον. 
 Κηρνξ. 5 αϊμον *δ'(5ω (?' ετά "^βάριν. 
 
 6 *ή 6ν "^δοντίεΐξ ^ατϋυτα; 
 
 7 κελεύω βία μεΟ'έΰ%'αί. 
 Χορός. 8 "^ϊκταρ '^φρενός "^ata, 
 
 9 Ιώ Ιον. 
 
 Κηρνξ. 10 λεΐφ' έδρανα^ κι' ες δόρυ 
 11 anW *αμ7ΐαλίν ενβεβών. 
 
 .άάναος. 1 μήτίοτε ηάλιν Υίδοψ άντιύτρ. α* 
 
 2 άλφεΰίβοίον νδωρ^ 
 
 3 εν^'εν άεξόμενον 
 
 4 ξώφντον αίμα βροτοΐόϋ βάλλει. 
 
 5 *άγνος εγώ βα%νχαΐθξ^ 
 
 6 *γερον^ ■\ανά^,ιος *βίας. 
 Κη ρ ν ξ, 7 6υ δη vat ναι βάΰεί 
 
 8 τάχα %•έλεος άΟ'ελεο?, 
 
 9 βία βία. 
 
 10 βά^'ί τίρόκακα ϋΛρΙν] τία^εΐν 
 
 11 Υολομενα Λαλάμαις *{εμαΐς\. 
 
714 USE OF Auxiliary verbs in greek, [book iv. 
 
 Χορός. 1 αίαΐ^ αίαΐ^ 0tQ. β'. 
 
 2 καί γαρ δυβτίαλάμωξ ολοιο 
 
 3 δυ άλίρρχπον αλΰος 
 
 4 κατά Σαρτίηδόνυον 
 
 5 χώμα '\πολν'φαμμον άλα^ΒΪς 
 
 6 ή-εϋρεΐαίϋίν ανραίξ. 
 
 * * * * 
 
 Χορό ξ οι. 1 οι οι οι οι αντι6τρ. β\ 
 
 2 *λνμας ' ^αργια *γαρ *6ν λάΰκείς. 
 β\ 3 περίχριμτίτα βρνάζεις. 
 γ . 4 δ *6αρών γάν, δ μέγας 
 
 5 Νείλος, υβρίζοντα (?' άποτρέ- 
 
 6 'φείεν αϊύτον νβρυν. 
 
 'Γ" 'Γ• •ί• ^ 
 
 Χορός. 1 οι οΐ πάτερ ^βρετεων τΐρος α- 6τρ. γ. 
 
 2 ταν άλκά μ άγει, 
 
 3 αραχνος ως βάδην 
 
 4 οναρ οναρ ^μ' ελών. 
 
 * * * * 
 
 1 μαιμα τίέλας δίτίους όφις άντιΰτρ. γ. 
 
 2 *[0'εών ου φροντίβας], 
 
 3 εχιδνα δ' ως με τις 
 
 4 *πόδα δακονό' άγει. 
 
 The following is a literal translation. Chorus. "Would you had 
 perished outright on the rolling briny main, with your domineering 
 insolence and your nail-fastened ship." Herald. "I will set you bleed- 
 ing on board my bark. What ! are you beating your breast, bawler? 
 I bid you desist." Chorus. "Mischief is near my heart, alas! alas!" 
 Herald (to Danaus). "Leave these seats; go to the ship, revering 
 once more what you have dishonoured." Danaus. "Never again 
 may I see the enriching stream, whence the life-giving blood finds its 
 growth and vigour for mortals. 1 am pure, and of ancient descent, 
 undeserving of violence, old man!" Herald (to Danaus). "You shall 
 soon go to the ship, to the ship, whether you will or no, by force, by 
 force ! " (to the Chorus). " Go before you suffer the preludes of woes, 
 being struck to death by my hands." Chorus. "Woe, woe! for may 
 you perish helplessly out on the salt-flowing plain , driven to and fro 
 in the wide breezes off the sandy promontory of Sarpedon."...CAorM5 
 a. "Ah! what an outrage: for you utter savage words." Chorus b. 
 "Proudly you stalk nearer and nearer!" Chorus c. "May the 
 mighty Nile , that sweeps o'er the land , turn you back in perpetrat- 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GREEK. 715 
 
 ing this unheard-of insolence !"..." Oh! my father, the supposed secu- 
 rity of the altar is leading me to mischief, like a spider having taken 
 me, step by step, a dream, a dream !"..." The two-footed serpent is 
 raging near me, with no thought for the gods; and some viper, as it 
 were, having bitten my foot, is leading me away." 
 
 Although the corrections , which we have introduced on our own 
 conjectures or borrowed from other scholars *, are very considerable, 
 we beUeve that they are all in accordance with the soundest rules of 
 criticism, and they are confirmed by their results — the restoration of 
 the antistrophic metres and of a simple meaning to the most confused 
 passage in the whole range of Greek poetry. With regard to the 
 phraseology the following remarks will suffice. The peculiar appUca- 
 bility of the epithet γομφόδετος to an Egyptian βάρις is proved by 
 Herodotus, n. 96, and for the wish of the Danaides, we have only to 
 look back to vr. 33 — 36. The future εόω does not occur, but we find 
 the aorist eloa in the same sense as here in Hom. II. ix. 200 : εΐΰεν δ' 
 εν κλίΰμοΐΰί. χπι. 657 : ε? δίφρον άνεΰαντες. Od. ι. 361 : ες αβάμιν^ον 
 εΰαΰα. For the emphatic question η <5v we may compare Agam. 1523 : 
 7] 6v τόδ' ερ^αί τΰλήΰι;]; κ.τ.λ. cf. Prom. 745: 0v δ' αν κέκραγαζ. 
 The verb δοντΐεΐν is particularly applied to express the planctus, 
 which was a necessary part of female lamentation. Thus we have in 
 Eurip. Alcest. 103: ονδε νολαία δοντίεΐ χεΙρ γυναικών: in Soph. 
 Aj. 633: χεροΛλϊίκτοι d' εν ότερνοιΰί πεόοννταυ δοντίοί, και 7to^ 
 λιαξ αμνγμα χαίταξ : and in ^sch. Pers. 120, we have: κα\ το Κίΰ- 
 oLVov τΐόλίΰμ άντίδοντΐον εΰΰεται, οά, τουτ ετιοξ γυναικοτιλψ 
 ^ης όμιλος άτίνων, where the ideas of "beating" and "crying" are 
 placed together, as in the passage before us, if we have rightly intro- 
 duced ατίντα. With regard to this latter word, which is an Homeric 
 epithet for a herald, there does not appear to be any reason why it 
 should not be a general term for loud and clamorous speaking , and 
 Quintus Smyrnseus uses it as an epithet for the feminine noun ΰνριγξ^ 
 just as we have ηχετα τέττιξ in Hesiod. The phrase βία or βίας με- 
 Ο'έΰ^'αι is Homeric; thus, τρις δε με^'ήκε /3tjy (11 xxi. 177; Od. xxi. 
 126); and the quasi-preposition ϊκταρ occurs with the genitive case 
 
 * We are acquainted with those conjectures only which are recorded 
 in the editions of Butler {Cantabr. 1809), Wellauer {Lips. 1823), Dindorf 
 {Oxon. 1841), and Paley {Cantabr. 1851). The last of these has intro- 
 duced many of our emendations, and we have accepted some of his sug.- 
 gestions in return. To save the necessity of repeated reference we have 
 designated all the emendations which we have introduced for the first 
 time by an asterisk C*), and have placed an obelus (|) before those 
 which had been suggested previously. With regard to the emendations 
 of Bothe and Burges, we subscribe to Wellauer's remark: sed hcec risui 
 sunt cuivis sano. 
 
716 USE OF AUXILIABY VERBS IN GREEK. [bOOK IV. 
 
 in ^schyl. Ag. 114; Eumenid. 1000; and as an adverb in Plato, Besp, 
 IX. p. 575 c. The last two lines of the strophe are addressed to Da- 
 naus, to whom we also assign the first six lines of the antistrophe. 
 We conceive that αμτίαλιν is justified by the μήτΐοτε τΐάλιν which 
 follows, just as τΐαλάμαις (avt. cc. 11) suggests the following δνΰττα- 
 λάμως. The epithet άλφεβίβοίος seems to bear a general sense in the 
 present passage, without any immediate reference to the production 
 of oxen. In Homer άλφεόίβοιαί τίαρ^ένου (II. xviii. 593; Hymn, in 
 Vener. 110) are damsels who procure for their fathers large gifts of 
 cattle from their bridegrooms (Doderlein, Gloss. Horn. No. 2268); 
 and the epithet is therefore synonymous with the female names, Ήερί- 
 βοια, Περίβοία^ Ενβοία., Ώολνβοια, &c. The proposal to derive άλ- 
 φεΐν, like άλφα (Plut. Sympos. Qu. ix. 3, p. 1047 Wyttenb.), from 
 the Semitic word for "an ox," and the inference that ανδρε£ άλφηΰταί 
 are "traders in cattle," and αλφιχον "the food of cattle" (Paley, on 
 Propert. iv. 17, 19), is not likely to be generally accepted. It seems 
 clear, on the contrary, that Pott (Etym. ForscTi. i. p. 359) and Doder- 
 lein (u. s.) have arrived by different routes at the same true result 
 when the former compares άλφαίνω with the Sanscrit a-labh, "to 
 obtain," and when they both recognise in it the stem of λάτν-τειν, 
 λαβ-εΐν, &c.; cf. ά-λάβαΰτροζ. Benfey (Wurzellex. i. p. 75) compares 
 ολβ -ogy which belongs to a different root (above, § 116). We have 
 extracted the new reading of avt lot ρ. cc'. 5, 6, from the corrupt words 
 of the Scholiast: εγώ ^[δ?] βαΟ^νχαΐοζ άναξία [άνάξως?] ταύτης τηζ 
 β[α%'ρε]ίας, γερον. According to our view, Danaus signifies to the 
 herald that he is άγνόζ, i. e. religiously pure in his worship of the 
 foreign gods, (so he is called άνηρ ayvog, in v. 358), for that he is 
 an old noble in Argos , and therefore undeserving of violence on the 
 part of an Egyptian. Or the epithet may refer to his being on sacred 
 ground, as in v. 223 : εν αγνω δ' εΰμος ώς τίελειάδων ϊζεβ^ε. The 
 herald then turns to the Danaides, as appears from the change of gen- 
 der in ολομενα. Here we have omitted the superfluous words, under 
 the guidance of the metre. It seems to us that βία τΐολλα was a mar- 
 ginal gloss on βία βία, and φρονδα , which is not found elsewhere in 
 JEschylus, is an explanation of βά^ί. The necessary particle πριν 
 might easily slip out in a sentence which contained προ-, but μή^ 
 which is found in the Medicean MS., might take its place with an 
 allowable construction. We have added εμαΐ§ in 1. 11; it is merely 
 a repetition of the end of παλάμαις , by which it was probably ab- 
 sorbed. With one slight alteration ύτρ. β\ seems to be correct, 
 and guides us through the otherwise inextricable difficulties of the 
 άντίΰτροφη. We consider λνμαξ a genitive dependent on the in- 
 terjection, according to the well-known construction. We are willing 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GREEK. 717 
 
 to believe that τίερίχριμτίτα βρυάζευς, though not very easy, is justi- 
 fied by τίερίτΐεμτττα ^νοΰκεΐς (Ag. 87); compare also Soph. CEd. T. 
 192: φλέγει με Λερίβόητος αντιάζων. It must refer to the insolent 
 and boisterous manner in which the herald stalked round the group of 
 suppliants, continually drawing nearer to them: the same is obviously 
 implied in the αραχνος ώς βάδην (6τρ. y . 3), and μαίμα τΐέλας (άντ. 
 γ . 1), which follow. In line 4 of the antistrophe it is manifest that 
 0(5' έρωτας must be corrupt. It seems also clear to us that these 
 words refer in some way to the river Nile; and keeping close to the 
 last traces in the books, we read ό βαρών γάν, i. e. "he who inundates 
 the land." The word ΰαρόω in this sense has no better authority 
 than Lycophron, who writes (389): 
 
 τον d' ola δνατην κηρνλον, δυα ΰτενοΰ 
 ανλώνοξ οϊβεί κνμα γνμνίτην φάγρον 
 διπλών μεταξν χοιράδων ΰαρούμενον. 
 
 But that there must have been a very old use of the verb ΰαρόω 
 or ΰάρω in this sense, is at least clear from the name of the Σαρω- 
 νικός κόλτίος, the Σαρωνίδεςτίετραι, and the Σαρωνία'Άρτεμις, In 
 the two remaining strophes we have been guided chiefly by the re- 
 quirements of the metre. The Scholiast says of the former: η τών βρε- 
 τέων ετίικονρία βλάπτει με; the old editions have μελαν in v. 4; and 
 the Scholiast renders οναρ by μηδέν ονΰαν, it is clear therefore that 
 the true reading is that given above. In the antistrophe the supple- 
 ment &εών ov φροντίβας is suggested by the words of the herald : 
 οντοι φοβούμαι δαίμονας τονς εν^'άδε. The phrase πόδα δικονβα, 
 for which Paley reads πόδ' ενδακονϋα, contrary to the metre, is ex- 
 plained by the fragments \vhich he cites, 231: φαγέδαιναν η βάρκας 
 ΙβΛί'ίει ποδός. 234: δεινην βτομωτον εκφνΰιν ποδός λαβών. As to 
 the extraordinary scholium on v. 875 : εϊς ϋπερ τών Αιγυπτίων πρεΰ- 
 βενοι^ we can only suppose that, unless it belongs to something sug- 
 gested by V. 905 : πολλούς ανακτάς παΐδας Αιγύπτου τάχα ο'ψεβ^'ε, 
 it must have arisen from the corruption εΙς προ γάς ύλάβκοι, sug- 
 gested by λυμα6\_-ις η προγαβυλαβκει]. In v. 950, we ought to read: 
 εβται τάδ' ηδη πόλεμον αϊρεό^•αι νέον. cf. 335 : βαρέα βύ γ είπας, 
 πόλεμον αϊρεβ^αι νέον. With regard to the whole passage, it may 
 truly be said that any thing is better than the present state of the 
 text; and the old physicians' maxim — fiat periculum in corpore vili — 
 may be applied here without the slightest risk of injury to the patient. 
 
 476 With regard to ^υμος and θυμέλη, the quantity of the first 
 syllable is no reason against concluding that they are connected with 
 ^ύω, as the similarity of meaning, especially of the latter word, would 
 
718 USE OF AUXILIARY VERBS IN GREEK. [bOOK IV. 
 
 induce us to suppose. The quantity of the ν in words from this root 
 is continually varying, why we know not, any more than we can ex- 
 plain why we should have both λϊτίαρής and λϊτταρός. Why, for in- 
 stance, should we have ^'υω, ^νμα, ^vXaxog, and %OVQogj but %'νάω, 
 ^ύβία, ^νβανοζ, and ^νηλτι^ The word %νμοξ seems to have derived 
 its meaning from the use of thyme in fumigations, or because the 
 brushwood of the tree was employed for fuel in sacrifices: its resem- 
 blance to the Sanscrit dhumay Latin fumuSf is obvious, in spite of the 
 difference of quantity; the word thymus was, of course, borrowed from 
 the Greek. We do not find any reference to Q'vμog in the article of 
 Hesychius on O'v^Off, which gives the following meanings: ιΐ^υχη η 
 TCQoaLQBuLg. b^vTTig. τΰνευμα. επιθυμία, οργή, λoγL(^μ6g. Most of these 
 we have already considered: that %^vμόg might signify πνεύμα in its 
 physical sense, a comparison of θύελλα forbids us to doubt; but of 
 course the lexicographer refers to the moral or mental application. It 
 is curious that Hesychius includes among the meanings of ^'vμόg the 
 very two words which Plato opposes to it, ετίι^νμία and λoγίΰμόg. 
 The sense of auger is implied in o^VTTjg, not in οργή. The word 
 7tρoaiρεύίg, meaning a deliberate choice, is well known to the readers 
 of Aristotle's Ethics. 
 
 477 That 6-ργή is immediately connected with ό-ρέ^ω appears 
 from a comparison of όρόγυυα (Pindar, Pyth. τν. 406) with οργνια, of 
 άλεγοο with akyog, &c. The mental or moral meaning of οργή is clearly 
 seen in the common use of ο-ρεγομαι, " I reach out my hands eagerly 
 to take;" so that we have here again a reference to two of the mean- 
 ings of λάω. How intimately these ideas are connected may be seen 
 even in the later writers, who revert to the old usages of language, 
 without any feeling for the etymology. Thus Juvenal says (x. 139): 
 ad hcec se endoperator erexit, where we have the sense of ορεγεΰ^αί 
 nvog. The words ο-ργή, ο-ρεγω are made up of the elements ra-\-S^a 
 with the vowel prefix ο ; compare rego, &c. (above, p. 395). The same 
 compound root appears ΊΆρεJ^-ω{ίvii.ρεv-6ώ)-,ρεv-μa^ ρεΛ-ω, ρέ-μ-β-ω, 
 ρίτΐ-τω, ^ί-μ-φα, &c.; in all of which the idea of accelerated velocity is 
 conveyed, as also in the cognate Latin words rcfp-io, rap-idus, &c. We 
 have it also with a dental prefix (p. 396) in the secondary formation 
 δρακ-, *'I take" (cf. δράΰΰω, δραχμή, &c.; above, p. 328), in άε-ρκ-, 
 "I see," and in the Sanscrit termination driga, Greek -λi-κog\, which 
 imply the sense of seeing. That this is only a secondary meaning is 
 clear, the primary one being that of motion just mentioned. 
 
 478 That this idea of motion, the common meaning of ρέω, is 
 implied in the Greek notions of seeing, light, &c. , appears from such 
 
CHAP, v.] USE OF AUXILIAEY VEEBS IN GEEEK. 719 
 
 phrases as sutt χρόα απορροή οχημάτων (Plato, Meii. p. 76 d), and 
 from the Greek conception of love as a kind of ophthalmia, caught 
 by an efflux or emanation of particles from the eyes : thus -3Eschylus 
 says, Agamemnon, 4:14:: 
 
 πό^ω d' νπερτνοντίας 
 φάΰμα δόξει, δόμων άνάΰΰειν. 
 ενμόρφων δε κολοΰΰών 
 εχ^εται χάρις άνδρί, 
 ομμάτων δ' εν άχηνίαις 
 ερρεϋ πάό' Αφροδίτη. 
 
 ί. e. after the abduction of Helen, her image would seem to be mistress 
 of the house ; but Menelaus would take no pleasure in statues, how- 
 ever beautiful, and in the want of eyes (for the Greek statues had 
 none, or rather they were cold and inefficacious) all love would vanish 
 (see the imitation in Euripides, Alcestis, 348 foil.). And with a 
 corresponding sentiment, the same Helen is said {Agam. 719) to have 
 come to Troy as μαλ&ακον ομμάτων βέλος, δηξί&νμον έρωτος αν- 
 θΌς, "a tender dart from the eyes, a soul-consuming flower of love." 
 That ομμάτων refers to the eyes of Helen , and not , as some of the 
 commentators suppose (on the strength of the phrase in Herod, v. 18, 
 quoted below), to the eyes of those who gazed on her beauty, is clear 
 from V. 229: εβαλλ' ε^ίαβτον %ντηρων απ όμματος βελει φιλοίκτω. 
 Similarly, Sophocl. Antig. 795: 
 
 VL^a d' εναργής βλεφάρων ίμερος ενλεκτρου 
 ννμφας, 
 
 where the two genitives would be very awkward, unless βλεφάρων 
 ίμερος are to be taken together in close connexion (see Julius Pollux, 
 11. § 63: λεγοοντο δ' αν οφθαλμοί λάμποντες, κ. τ. λ., καΐ το απ 
 αυτών απορρέον ίμερος). Euripides, Ilippolytus, 525 : 
 
 έρως, έρως, ος κατ ομμάτων 
 ΰτάζείς πό^ον, εΐΰάγων γλνκεΐαν 
 'ψνχα χάριν. 
 
 Iph. Aul. 584 : 
 
 ος τας^Ελένας εν άντώποιβι βλεφάροις 
 έρωτας εδωκας. 
 
 Plato, Phoedrus, p. 251 β: δε^,άμενος τον κάλλους την άπορροην δια 
 των ομμάτων. Aristot. Eth. iV. ιχ. 12: τοις ερώύι το δράν άγαπη- 
 τότατόν έΰτι. Herod, ν, 1 8 : άντίας ϊζεΰ^αι γυναίκας, άλγηδόνας 6φι 
 οφθαλμών. Wesseling ad 1.: "Issei Rhetoris — qui rogitanti Ardui, εΐ 
 η δείνα αϋτωκαλη ^pc^/i/otro, modestissimi {μάλα ΰοφρόνως) respondit 
 
720 USE OF AUXILIARY VEEBS IN GEEEK. [bOOK IV. 
 
 πΒΊίανμαι οφ^αλμιών.^^ Plut. Alex. 21 : Longin. iv. 12; Ovid, Heroid. 
 XII. 36: ahstulerant oculi lumina nostra tui. There is a remarkable 
 application of this phraseology in two late writers, one of whom has pro- 
 bably copied from the other. Musseus, in his pretty little poem about 
 Hero and Leander (vv. 90 — 98), thus describes the first access of love: 
 
 6vv βλεφάρων d' άκτΐβιν άέξετο πνρΰος ερώτων 
 και κραδίη πάφλαζεν ανικψου τίνρος ορμγ]. 
 κάλλοξ γαρ τίερίτΰνύχον άμωμήτοω γνναικοξ 
 οξντερον μερότΰεΰόι πελει πτερόεντος δϊΰτοΰ' 
 οφ^'αλμος d' bdog εότυν α% οφΰ'αλμοΐο βολάων 
 ελκog ολιΰ^αίνει και ετά φρεναξ άνδροξ οδεύει, 
 είλε δε μιν τότε θάμβος, άναυδείη, τρόμος, αιδώς. 
 
 The same thoughts are expressed in prose by Achilles Tatius (de 
 Clitophontis et Leucippes amoribus , i. 4) : ώ^ d' είδον εν^νς ατίολώ- 
 λειν ' κάλλος γαρ οξντερον τυτρώβκευ βελονς και δια των οφθαλμών 
 εΙς την -φνχην καταρρεϊ. οφ^'αλμος γαρ οδός ερωτικω τρανματι. 
 πάντα δε μ είχεν δμον, ετίαινος^ εκπληξίς, τρόμος^ αιδώς, αναίδεια. 
 It is possible that the sense of hearing conveyed by the cognate words 
 κ-λν-ω, λν-ρα, Sanscrit g-ru^ &c., may have a similar origin. There 
 is also some trace of an implication of the sense of seeing in*! the root 
 O'ccP-, the primary meaning of which is motion or impulse. Thus, the 
 words expressing astonishment, %αν-μα, τέ-Ο'ητί-α-, %•ά-μ-βος, &c., 
 are clearly connected with Ο'άω, ^'ανω, and %'εάομαι, " I gaze at." 
 
 479 Hesychius gives οργή the three following significations: τρό- 
 πος. μανία. %'νμ6ς, the order of which should, however, be reversed ; 
 for &νμός is the word which contains the primitive meaning of οργή. 
 The ground-idea in both is the same, *^an impulse" or "moving in any 
 direction." The analogy between ^νμός and οργή is farther obser- 
 vable in the application of the roots of both to the designation of 
 sacred rites: that Ο'ύω and its derivatives are so applied, we have 
 already seen, and the same may be said of οργιά, οργεών, οργιάζω. 
 This meaning appears to be derived from the custom of holding out 
 the hands or parts of the offerings to heaven (comp. ενχετο χειρ' ορέ- 
 γων εις ονρανον άβτερόεντα with dupUces tendens ad sidera palmas, 
 and ότίλάγχνων μοΐραν δρεξον with porrigit exta hovis), just as adoro 
 derives its meaning from the custom of turning the face to the sky. 
 When δργάω is applied to vegetable productions, it refers to their 
 "sprouting," "rising above ground," "moving upwards," and from 
 this is deduced the sense of "ripe," "soft," which is found in δργάζω 
 (compare the passage in Kuhnken's Timceus, under the words μετρίως 
 ώργαΰμενος, and δργα). That this was the transition appears from the 
 
CHAP, v.] USE OF AUXILIABY VERBS IN GREEK. 721 
 
 conjunction of τίΒΤίαίνεται with oQya in Herodotus (iv. 199): ό Iv ry 
 κατντΐερτάτΊ;] της γης ΛΒτίαίνεταί ts καΙ δργα. We do not agree with 
 Passow in supposing that ύφριγάω and ύτΐαργάω, which are used as 
 synonyms for οργάω, are etymologically connected with this word, 
 but would rather connect them with ΰφαραγεω, άΰτΐάραγος (Attice 
 άΰφάραγος)^ and the German Spargel, Sprosse, and spriessen. 
 
 We have already adverted to the fact that as οργή, οργάω^ which 
 imply ripeness and fulness, are connected in meaning with κόρο^, so 
 ορεγω is connected through rego^ rex, Sanscrit raja, with the political 
 signification of the cognate κνρίος (above, p. 556) ; and we have traced 
 some of the trains of thought which are suggested by the names sig- 
 nificant of rank, power, and royalty (§§ 333, 336 — 338). The root 
 reg- in this application suggests some further comparisons of a very 
 interesting nature. If reg-num and reg-ius immediately refer to rex^ 
 it is equally clear that reg-io and recta regio, or regio viarum, point 
 to the primary meaning of reg-ere, i. e. " to make a straight line," in 
 which sense, as we have elsewhere suggested (Gr.Gr.-p.2d2), the root 
 0-ΡΕΓ, 0-PX^ or E-PX appears in 'ε-ρχ-ομαο, "I make a straight 
 line for myself," " I go forward ; " so that reXj after all, corresponds 
 pretty nearly in origin and meaning to the Homeric ορχαμος ανδρών. 
 The idea of dividing by rows androads is connected with that of passage 
 and direct progression; and in this sense we find the words ορχος^ 
 ορχατος, and ορχεομαυ. Although the Sanscrit root raj is generally 
 rendered splendere, we look upon this as a secondary meaning ; for the 
 word raji=linea, series, clearly points to the primitive and true sig- 
 nification of ορχ- and rig. Some mention has been made above (§ 160) 
 of the connexion between the Sanscrit ksJiatra, " a man of the military 
 caste," Zend ksatJira, and old Persian kJisJidyathiya, "a king." Raw- 
 linson {As. Soc. xi. pp. 115 sqq.) reads the following words in the 
 Persian cuneiform inscriptions: khshatram = corona v. imperium; 
 khshatra~pa=satrapes', khshayarshd= β^ρξης, Arta-khshatrd=Άρ- 
 ταξέρξης ; and khshdi/athiya=rex. We refer all these words to a lost 
 root khsha=ka-ra, which we regard as ultimately identical with kr'i, 
 "to make," as well as with kshi, "to rule" (pres. kshaydmi); cf. ενρν- 
 %ρείων with uru-kshaya (Rosen, Big- Ved(B Specim. p. xi) ; and thus 
 khsha-tram may signify reg-num, khshe-tra, regio, whence khsMtra-pa 
 =regionis dominus (above, §§160, 213); k'hsha-tra=^regius, i. e. miles 
 =rdja-purusha, kJishaya=rex, khshdy-a-thiya for khshdy-anthiya=^ 
 regens, or regentius, and khs}iay-arsha=^rex sanctus, like the Sanscrit 
 rdj-arshi=rex sanctus*. Now we have shown above the connexion in 
 
 * Benfey {Keilinschr. p. 79) and Oppert {Journ. As. xix. p. 174) compare 
 arsha with the Zend arsna, Sanscr. aksha, "oculus," and render Khshayurshd 
 by Konig-seher and ml dominatrice. 
 
 3 A 
 
722 USE OF AUXILIABY VEEBS IN GEEEK. [bOOK IV. 
 
 meaning of ka-ra and ta-ra^ and how the latter root signifies " to bore 
 a hole," "to pass over," &c. (§ 178), and how many contacts it has 
 with pa-ra. We may therefore understand how ksJii may signify not 
 only regere^ " to make a straight line through a country," but per-ire, 
 or, with a causative affix, perdere^ &c. (Bopp, Gloss, p. 93, ed. alt.). 
 In the Semitic languages, the two principal words denoting sovranty 
 are 'rrb^a and b^la. Fuerst is inclined to explain the former in the same 
 way as Sanscrit scholars illustrate the root raj, namely, by a reference 
 to the idea of light and splendour: "Radix hujus verbi compositi est 
 •^rb, ^b (ibs) vi splendendi; vis autem splendoris et excellentise ad 
 regnum transfertur" {Concord, p. 635). It appears to us that the 
 word is more truly referred to the root of Nb^i, TcXsog, TtoX-vg, "full," 
 which, through pol-leo, connects itself with val-eo, validus, and the 
 Sanscrit bala=vis, robur, and, like βελ-τίων, mel-ior, expresses per- 
 sonal superiority. In the Tyrian language it seems to have dispensed 
 with the affix *?:- which distinguishes the word in Hebrew and Arabic : 
 at least it is easy to explain Μελ-ί-καρ^ος, "the king of the city," 
 Nn'n')?'^b?3, in the same way as Ab-i-melek (above, p. 270), and in the 
 Maltese inscription we have rTnj:)b73 (Gesenius, Ling. Ph. Mon. p. 96), 
 where even the. connecting vowel is omitted, and the same is the case 
 in the well-known proper names Ha-mil-car, Bo-mil-car, &c. We do 
 not connect with this root the other word b^3, which is sufficiently 
 distinguished from b!^3 in the inscription just mentioned, where we 
 have the three quasi-synonyms in one address: n'ljpb^sb ^S"^?.^^^ 
 ΊΪί b^5 "to our Lord, to the city-king, the superior of Tyre." It seems 
 to us that b3>-a is merely a prepositional compound indicating su- 
 periority in place, so that it corresponds to ανα-ξ from ανά: we have 
 a perfectly analogous form in h^-12. ^schylus, no doubt, when he 
 makes the Persian Chorus address [their king as βαλήν {Pers. 663), 
 was thinking of this word b^a, which he had heard from Phoenician 
 sailors. In the Scholiast, we must of course read Τνρίων for Θουρίων, 
 though the word seems to have taken root in Phrygia also. We have 
 discussed the etymology of the Egyptian word Pharaoh in the Quar- 
 terly BevieWy No. clv. p. 168. 
 
 480 The meaning of οργγι, when it denotes an emotion of the 
 mind, is easily deduced from that of the verb ο-ρέγω. In all cases it 
 indicates "an upward striving," "a tendency to the surface," "an im- 
 pulse," "a fancy," "a first impression," "a prominent desire." Hence 
 we understand such phrases as όργαζ ευμενείς καταΰχε&εΐν (Soph. 
 Antig. 1166), "to keep down their angry impulses, and so to make 
 them gentle : " and we have the same idea in another passage of So- 
 phocles, which has given the greatest trouble to the commentators. 
 
CHAP, v.] USE OF AUXILIARY VERBS IN GREEK. 723 
 
 Electra is excusing herself for the irrepressible outbreak of joy with 
 which she welcomes her long-lost brother; she was silent, she says, 
 wheif the false message of his death reached her, but she cannot con- 
 tain herself now. The true reading and arrangement of the passage 
 (Soph. Electr. 1281 sqq.) appear to us to be the following: 
 
 ώ φίλαι [άνίκ] εκλνον 
 
 αν εγώ ονδ' αν ηλταΰ' αυδάν, 
 
 ϊοχον οργαν ανανδον, 
 
 ονδε ύνν βοά κλνονύ% 
 
 α χάλαινα' νυν δ^ εχω 6ε' κ. t. λ. 
 
 We think that ανίκ has been absorbed by its similarity to the con- 
 tiguous syllables φίλ[αίεκ]λνον, and with this insertion, there is no 
 further difficulty in the passage. 
 
 The sense of " anger," which Hesychius assigns to οργή when he 
 renders it μανία, is quite a secondary one; the transition is the same 
 as in ϋ'νμός. The third meaning, τρότίοζ, is a very old application of 
 this word (Hesiod, Op. 306). Whether it is derived from the con- 
 nexion of will and character, which we have before pointed out, or by 
 some transition to the meaning of όργάξω, similar to that which pro- 
 duced the words "humour" and "temperament," we cannot pretend 
 to say. There is a singular passage of Thucydides (vni. 83) in which 
 the word occurs in the plural number: τΐάντων τε^Α6χνοχον είναι 
 αϊτιον, επιφεροντα οργαξ Τυ6(5αφερνει δια ϊδια κέρδη. Dr. Arnold's 
 translation of this phrase ('* humouring," "supplying or ministering 
 tempers such as a man likes ") seems to show that he approved of the 
 interpretation of the Scholiast, who says: το ετίίψερειν οργην ετΐΐ τω 
 χαρίζεβ^αι καΐ ΰνγχωρεΐν εταττον oi αρχαίοι., μάρτυς Κρατίνος εν 
 Χείρωΰί λέγων την μουόι,κην άκορέΰτονς επυφερενν οργάς 
 βροτοΐς ΰώφροΰί. This passage of Cratinus, compared with the 
 words of Thucydides, assures us that the interpretation of the Scholiast 
 is incorrect. Besides, the change of number from οργάς to οργην en- 
 tirely alters the meaning of the phrase ; at least, St. Paul's expression 
 (Bom. ΠΙ. 5), δ Θεός δ ετΐΐφέρων την οργην, must, like that which 
 is quoted from Polybius (Leg. 28), την οργην φερεί επί τους Αιτω- 
 λούς, refer to the effects of anger (it is η οργή in both passages); and 
 with regard to the terms χαρίζεΰΟ'αί and όυγχωρεΐν, it seems clear 
 that the transitive phrase ετίιφερειν οργάς could not signify to bring 
 one's own disposition to suit that of another. It must mean, rather, 
 to suggest some humours, dispositions or habits of mind to another, 
 like εταφέρευν δόξας (Plato, Bespublica, x. p. 612 b). &c. Compare 
 also the common phrases ετίιφερειν αΐτίαν-, 'ψόγον, τιμωρίαν, &c., all 
 of which imply an importation of something from without. From these 
 
 3 A2 
 
724 USE OF AUXILIARY VERBS IN GREEK. [bOOK IV, 
 
 considerations, we conceive that Hanovius {Exercit. in Com. Gr. Hal. 
 1830, I. p. 60) and Meineke (Fragm. Com. Gr. u. p. 157) have given 
 to the phrase ετΖίφίρειν OQyag a meaning which could only have been 
 extracted from the middle voice. The former, who renders the phrase 
 in much the same way as Dr. Arnold — voluntatem accommodare^ i. e. 
 obsequiy — remarks, "quoniam οργή vel οργαί in cujusque animo cer- 
 nuntur, moveri et excitari possunt ab alio vel alia quadam re, afferri 
 non possunt; quod si esset, extrinsecus οργαί petendse forent" — which 
 seems to us to imply a misconception respecting the meaning of the 
 term οργή', and Meineke, who makes the fragment mean musicam 
 immense favere hominihus moderatis^ appears to have overlooked the 
 obvious force of the passage, in which άκορε^το^, " restless," " change- 
 able" (above, p. 554), is opposed to σώφρων, "sober," "contented." 
 Cratinus says, that music puts restless whims into the heads of sober- 
 minded people: and Thucydides, that Astyochus was charged with 
 suggesting caprices, or putting crotchets into the head of Tissaphernes: 
 that he induced those whims which prevented the satrap from dis- 
 charging his duty to the confederacy. That οργαί may be used in the 
 plural after such a verb as εΛιφέρευν, appears from Lysias (de Ccede 
 Eratosth. p. 94): οργίίξ τοϊξ άκούονΰυ Λαραόκενάξονβίΐ and that 
 it implies, when thus used, " habits of thought " or " a turn of mind/' 
 may be seen from Soph. Antig. 354 : άΰτννόμονς οργάς, and from 
 Thucyd.in.82 : ό τΐόλεμος τίρος τα Λαρόντα rag οργαζ των τίολλών 
 όμοιοι. In the same way, %νμοί is used in the plural when the temper 
 of one man only is spoken of; Sophocl. Aj. 716 : 
 
 εντε γ εξ άελπτων 
 
 Alag μετανεγνώΰ^η 
 
 Ο'νμών Άτρείδαις μεγάλων τε νεικέων. 
 
 Heracleides, Allegor. Homer. 19: ετΐΐόκοτονμένον τον κατά τήν κε- 
 φαλήν λογιΰμον τοις ττερί τα ΰτερνα ^νμοΐς. Id. 59: αϊ τΐρώται 
 της ίκεΰίας φωναί τονς αρΰενας αυτοί) ^νμονς εξεΟ'ήλυναν. These 
 two passages are quoted by Lobeck (ad Soph. Aj. 1. c. p. 488). 
 
Ι. INDEX. 
 
 PASSAGES FROM GREEK AUTHORS, ILLUSTRATED, EMENDED, 
 OR EXPLAINED. 
 
 [Unless it is otherwise specified, the numbers refer to the sections.] 
 
 Emendations are marked"^. 
 
 SEC. 
 
 ^SCHINES: 
 
 *c.Cies. 543 395 
 
 iEsCHYLUS : 
 
 Agamem. 36 468 
 
 114 95 
 
 118 454 
 
 176 291 
 
 237 284 
 
 267 181 
 
 363 291 
 
 414 478 
 
 444 174 
 
 *494 422 
 
 719 478 
 
 - 722 297 
 
 *741foll. .355 
 760,781.. 290 
 
 768 410 
 
 972 335 
 
 1010 323 
 
 1077 174 
 
 *1143 309 
 
 1198 310 
 
 1200 397 
 
 1304 335 
 
 1460 436 
 
 1461 335 
 
 1507 202 
 
 SEC. 
 
 • Cho'eph. *73 sqq. . 192 
 
 272 297 
 
 *464 472 
 
 *489 ..... 304 
 
 *615 290 
 
 630 174 
 
 641 310 
 
 853 472 
 
 1000 436 
 
 Eumen.12^14: 150 
 
 36 447 
 
 113 ...198 
 
 *229 218 
 
 270 218 
 
 *340 365 
 
 451 218 
 
 602 297 
 
 910 176 
 
 PerscB, 44 433 
 
 86 473 
 
 120 ..469,475 
 
 180 164 
 
 239 280 
 
 269p.455,note, 
 
 321 242 
 
 343 273 
 
 428 315 
 
 815 287 
 
 Prom. 90...... 459 
 
 SEC. 
 
 Prom. 265 273 
 
 327 174 
 
 Sept.c.Theb.^S...Zd4. 
 209.. 304, 325 
 
 232 394 
 
 *261 315 
 
 365 470 
 
 394 325 
 
 417 149 
 
 578 450 
 
 783 312 
 
 Suppl.init 318 
 
 *40 365 
 
 noifoll 469 
 
 577 454 
 
 610 472 
 
 672 297 
 
 796 .280 
 
 *848foU 475 
 
 *950 475 
 
 976 280 
 
 1006 465 
 
 AreT-«!uS: "^ 
 p. 34 174 
 
 Arrian: 
 
 Anab.ii.ll 315 
 
 V.16 178 
 
726 
 
 INDEX OF GEEEK AUTHOES. 
 
 SBC. 
 
 Aristophanes : 
 
 Achar.2bZ 303 
 
 640 461 
 
 835 175 
 
 Aves, lOlS 184 
 
 1263 184 
 
 Equit 253 331 
 
 280 305 
 
 707.. 175 
 
 890 154 
 
 969 330 
 
 Nub. 466 395 
 
 1183 161 
 
 Pax, 123 175 
 
 559 164 
 
 Pint 66 202 
 
 545 198 
 
 *Thesm. 372 721 
 
 Aristotle : 
 
 Anima, ii. 1 340 
 
 Cateyor. 1—5 125 
 
 £;iA.iVic.i.7,§14..342 
 v.4,§9...290 
 5,§6...290 
 Hist.Anim.i. 1, . . .174 
 Interpretl — 5. . . .125 
 Metaphys. viii. 6 . . 341 
 
 Poet. c. 4 304 
 
 Polit.i.G 174 
 
 8 278 
 
 II.3 164 
 
 IV. 8 327 
 
 V.4 174 
 
 i^Aei.i.lO 454 
 
 AXHENiBuS : 
 
 p.271 475 
 
 Callimachus : 
 Epigr. 29 149 
 
 CORINNA : 
 
 Fr.2l 133 
 
 Demosthenes : 
 
 Calliel. -p, 121 4: 174 
 
 *iyami. p.929 433 
 
 Steph.p.ll28 315 
 
 Euripides : 
 
 Mol.frag 218 
 
 Alcest 49 187 
 
 252 283 
 
 305 175 
 
 495 174 
 
 SEC. 
 
 Alcest. Q>Ob 325 
 
 Bacch.ZA.0 394 
 
 639 395 
 
 701 218 
 
 742 170 
 
 Electr. 112 ^97 
 
 Helen. *1254 278 
 
 1558 170 
 
 iTerc.Fwr. *863...437 
 
 Hippol. 525 478 
 
 605 304 
 
 *683 304 
 
 /o«, 235 175 
 
 598 325 
 
 Iph.Aul.2,2,Q 463 
 
 584 478 
 
 Iph. Taur. 909 ... . 174 
 1263... 202 
 
 Med. 1141 149 
 
 1409 472 
 
 Phoen. 650 318 
 
 Suppl 42 297 
 
 282 279 
 
 Troad.Z9S 304 
 
 Herodotus: 
 
 i.llO 162 
 
 II. 35 173 
 
 60 149 
 
 75 182 
 
 149 290 
 
 165 326 
 
 III. 99 174 
 
 119 187 
 
 IV. 33 178 
 
 64 192 
 
 154 175 
 
 199 479 
 
 V. 9 82 
 
 18....;.. 478 
 
 58 101 
 
 VI. 11 454 
 
 VII. 80 306 
 
 140 473 
 
 159 187 
 
 164 273 
 
 VIII. 24 218 
 
 Hesiodus : 
 
 ΓΛβο^. 102,528...297 
 
 319 472 
 
 Hesychius : 
 
 εκοίλος, *£κονσα. . 116 
 
 Hippocrates: 
 pp. 772, 783, 787.. 290 
 
 SEC. 
 
 Homer : 
 *Emend.var.loe. 
 
 pp.219sqq.l34 
 
 Iliadi. 31 174 
 
 11.673 302 
 
 IV.323 297 
 
 V.893 154 
 
 VI. 38 454 
 
 179 472 
 
 X. 52 344 
 
 XI. 336 174 
 
 XIII. 82. .•. 325 
 
 359 174 
 
 XVI.329 472 
 
 XVIII. 596 461 
 
 XXIV. 226 462 
 
 0^8.1.140 279 
 
 195 454 
 
 V. 17 172 
 
 483 218 
 
 VII.107 461 
 
 IX. 247 218 
 
 XI.588 163 
 
 XII.102 174 
 
 XIV. 311 472 
 
 XV.561 325 
 
 XVIII. 100 176 
 
 XIX. 229 451 
 
 *xxiii.m< 447 
 
 LONGINUS: 
 
 Sublim. 8 304 
 
 LySIAS: 
 
 DeErat. 0«?d 19 1.464 
 "^ DeEvan.Prol•. IIQ .211 
 c.Theomn.%10....111 
 
 Paul: 
 
 Goloss.ii.lh . 
 
 177 
 
 Philostratus : 
 V.HerodAl 304 
 
 Pindar: 
 
 Isthm. I.Q2 315 
 
 *23 304 
 
 III.23 304 
 
 75 150 
 
 V.24 88 
 
 VII. 35 472 
 
 39 304 
 
 Nem.w. 1 297 
 
 *vii.89 186 
 
 Olymp.ii.2Z 184 
 
I. INDEX OF GEEEK AUTHORS. 
 
 727 
 
 SBC. 
 
 Olymp.w.bl 297 
 
 III. 4 461 
 
 VII. 44 325 
 
 VIII. 20 291 
 
 XIII.16 325 
 
 Pyth. I. 14 472 
 
 50 290 
 
 III.33 472 
 
 44 180 
 
 IV.35 464 
 
 187 175 
 
 202 335 
 
 263 186 
 
 308 472 
 
 V.25 325 
 
 99 304 
 
 101 305 
 
 VI.13 281 
 
 VII.44 325 
 
 VIII. 21 foil... 291 
 40 470 
 
 IX. 64 284 
 
 X.58 273 
 
 81 281 
 
 XI.32 310 
 
 FragmAl 102 
 
 150 171 
 
 Plato : 
 
 Cmii/i. p. 391 60 
 
 399 124 
 
 418 136 
 
 CnVo, 43 154 
 
 Euthyd. 500 174 
 
 Gorg. 486 302 
 
 514 303 
 
 Legg. 701 271 
 
 761 150 
 
 773 289 
 
 834 218 
 
 863 463 
 
 886 289 
 
 895 59 
 
 902 455 
 
 *905 344 
 
 959 176 
 
 Meno, 82 154 
 
 Farm. 133 174 
 
 Phmd. 247 290 
 
 249 57 
 
 251... 57, 478 
 
 257 174 
 
 Phileh. 26 335 
 
 Polit. 299 463 
 
 302 271 
 
 Protag. 335 463 
 
 SEC. 
 
 Protag, 428 258 
 
 Resp. 437 463 
 
 439 471 
 
 444 174 
 
 458 464 
 
 479 174 
 
 538 464 
 
 596 59 
 
 603 291 
 
 614 sqq. . . 85 
 
 616 457 
 
 617 444 
 
 Soph. 261 124 
 
 Symp. 145 306 
 
 *175 139 
 
 Theoet. 145 306 
 
 168 184 
 
 Tim. 26 187 
 
 33 178 
 
 Solon: 
 
 p. 66 218 
 
 80 335 
 
 Sophocles: 
 
 4;"ax,*135 304 
 
 158 455 
 
 455 454 
 
 496 317 
 
 511 180 
 
 550 187 
 
 *572 304 
 
 620 291 
 
 693 57 
 
 700 317 
 
 727 184 
 
 871 335 
 
 890 187 
 
 938 32δ 
 
 1024 266 
 
 1096 326 
 
 ad fin 205 
 
 Antig. 29 279 
 
 57 174 
 
 109 174 
 
 114 298 
 
 145 174 
 
 237 277 
 
 251 281 
 
 292 290 
 
 360 305 
 
 478 218 
 
 505 175 
 
 569 281 
 
 594 305 
 
 600... 218, 457 
 
 8EC. 
 
 Antig. 605 202 
 
 610 457 
 
 634 154 
 
 662 290 
 
 686 187 
 
 795... 298, 478 
 
 857 167 
 
 1050 277 
 
 1166 480 
 
 1281 200 
 
 Elect. 43 395 
 
 120 447 
 
 696 454 
 
 934 192 
 
 *1281 480 
 
 (Ed.Gol.ni 218 
 
 112 311 
 
 120 335 
 
 127 472 
 
 (Ed.Tyr. 2 472 
 
 107 149 
 
 171 472 
 
 207 457 
 
 316 305 
 
 328 394 
 
 526 301 
 
 874 335 
 
 1395 310 
 
 1502 281 
 
 (Enom.fr 139 
 
 Philoct. 180 322 
 
 511 472 
 
 554 271 
 
 688 322 
 
 816 457 
 
 1155 279 
 
 TVacA. 568 176 
 
 831 foil. ... 57 
 
 Theocritus : 
 
 VII. 5 371 
 
 XIII.15 258 
 
 Thucydides : 
 
 1.18 181 
 
 22 187 
 
 36 305 
 
 41 174 
 
 44 .174 
 
 69 192 
 
 71 454 
 
 76 184 
 
 84 325 
 
 93 174 
 
 123 184 
 
 II. 41 187 
 
728 
 
 I. INDEX OF GREEK AUTHOES. 
 
 SEC. 
 
 II. 65 305 
 
 70 174 
 
 101 175 
 
 III. 21 302 
 
 24 135 
 
 36 200 
 
 62 59 
 
 67 291 
 
 IV. 13 315 
 
 26 218 
 
 28 273 
 
 85,86 304 
 
 95 395 
 
 SEC. 
 
 IV. 108 180 
 
 *121 422 
 
 V. 46 254 
 
 *111 124 
 
 VI. 3 371 
 
 16 .454 
 
 22 475 
 
 89 192 
 
 VIII. 83 480 
 
 85 174 
 
 *92 271 
 
 SEC. 
 XeNOPHON : 
 
 ^^αδ. I. 9, §17.... 258 
 
 iv.5,§35 85 
 
 8, § 15. . ..229 
 
 O/rqp. II. 2, §26. . .290 
 
 III. 2, §23... 174 
 
 v.l,§15...279 
 
 VII. 5, §21... 395 
 
 DereEquest.yii.il . 174 
 
 Hellen.i.2,^16 ...315 
 
 ii.4,§31...271 
 
 Mem. III. 14, §2... 175 
 
 8,§1...174 
 
π. INDEX. 
 
 GREEK WORDS ILLUSTRATED OR EXPLAINED. 
 
 ά negative, 184 
 ά intensive, 185 
 άβληχρόζ, 212 
 άβολος, 218 
 άγαμος, 322 
 άγάλακτες, 181 
 αγαμα,ί, 323 
 ayav, 203 
 ayavQOs, 323 
 ay^os, άγριος, 150 
 «yvia, 296, 414 
 αγχιϋτος, 284 
 α^ελφό^, 181, 286 
 άδρεπίβολος, 305 
 αό'ρο?,^ 336 
 άειγενέτης, 254 
 αελίο?, 181 
 αέναος, 270 
 α?7ρ, 199 
 Αίγιαλεΐς, 97 
 αίγλη, 451 
 αίγλητής, 457 
 α/άω^, 257 
 α/ε/, 188 
 a^^T^o's, 265 
 Αίϋ•ίο-ψ, 95 
 αΓ'9•ω, 223 
 αύ,ονρος, 455 
 cii^ocroryi^s, 312 
 «tVo), 436 
 Αιολείς, 97 
 a^'Aog, 97, 266 
 αίρέω, 269 
 αΓσα, 436 
 
 αΐοχος,αίοχννη, 219, 325 
 αίοχρός, 324 
 
 αίχμη, 174 
 «ί'ω, 513, 436 
 άκαιρος, 290 
 άκτ^άεία, 185 
 ακοιτις, 181 
 «κόλου-θΌ?, 181,213, 223 
 ακόρεστος, 335 
 ακορο5, 335 
 άζταίνω•, 4Α1 
 αλείφω, 213 
 «λτ^'ό'ίνό^, 258 
 άη^λων, 167, 173 
 άλλος, 138,166,215,268 
 αλλότριος, άλλότερος, 
 
 157 
 αλοχος, 181 
 άλφεΐν, άλφεοίβοια, 475 
 Άλφεααίβοιαν, 220 
 aAcojrjyl, 254, 455 
 a/tia, 181 
 άμαιμάτιετος, 472 
 «/[ίαλόί, 218 
 άμαρνββω, 213 
 άμανρός, 218 
 άμάω, 218 
 αμβλύς, 218, 460 
 αμείβω, 213 
 άμείνων, 262 
 άμέλγω, 212 
 άμενηνός, 335 
 άμέργω, 213 
 άμεύΰαοΟ'αι, 213 
 άμενοίηορος, 213 
 ά>με5, 136 
 άμννω, 262 
 α/ί*ςρ/, 172 
 
 άμφότερος, 273 
 
 αϊ/, 184, 186 
 
 ανά, 184 
 
 aVa^, 275, 334 
 
 άν^ροφ-θ-οροί, 30ϋ 
 
 άνεκάς, 184 
 
 ανεν, 204 
 
 άνε';ΐ;ω, 186 
 
 άνέ'ίριος, 181 
 
 άνήνοΟ'εν, 336 
 
 άΐ'τ^ρ, 333 
 
 αν^ο?, 334 
 
 «Vig, 204 
 
 άί/ΓΚ, 171 
 
 άντ/, 171 
 
 αντρον, 267 
 
 «0^05, 181, 262, 286 
 
 άοσσε'ω, άοΰβητήρ, 262, 
 
 286^ 
 άηάλαμνος, 218 
 άπαμβλύνομαι, 218 
 αηαξ, 154 
 άηατονρία, 181 
 άπο, 138, 177 
 απο, 184 
 άποδνομαι, 177 
 αποινα, 212 
 άποκαλε'ω, 184 
 άττο,Ιαυω, 453 
 'Απόλλων πατρώος, 459 
 άποψη μι, 184 
 απτερο5, 181 
 απτω, 318 
 αρα, 192 
 άρε(ϊκω, 285 
 άρεττ?', 174, 285 
 
730 
 
 π. INDEX or GBEEK WOEDS. 
 
 ιχρήγω^ 285 
 
 "Αρης, 285 
 άριϋ-μός, 272 
 άριβϋ'άρματος, 309 
 άριβτερός, 272, 285 
 αριΰτος, άριατενς, 285 
 
 322 
 άρκέω, 285 
 αρπάξω, 269 
 αρπυια, 296, 414 
 α^ρην, αρΰην, 285 
 'Αρταξέρξης, 160, 479 
 αρ;ΐ;αί()7τλθ'ϋΓθ5, 12 
 αρ;ΐ;αΓο5αη(ΐ7Τθίλαίθ'ί, 12 
 ασάλ)/?/, 461 
 άβελγής, 461 
 ασκελ£5, 185 
 ασκο5, 213 
 
 άαμεναίτατα, άβμενέα- 
 τατα,άαμενώτερος,ΙβΒ 
 αΰμες, 136 
 άβηάζομαι, 213 
 «(yTra^oryog, 212, 213 
 άΰηερχές, 185 
 citrTtis, 213 
 άΰταχυξ, 213 
 αστ?7ρ, pp. 160, 395 
 ατάλαντος, 181 
 αταλλω, ατίτάλλω, 344 
 αταρ, 138 
 ατε, 197 
 άτενής, 185 
 ατερ, 204 
 άτίτης, 254 
 ατίω, 437 
 !4ττ/κ77, 97 
 αν, a'Sd-tg, 138, 189 
 ίίΰερνω, 189 
 ανν, 189 
 ανρα, 198 
 αντάρ, 138 
 ανταυτος, 138 
 αντίκα, 196 
 αντοκτονε'ω, 173 
 αυτόματος, 472 
 avrds, 138, 303 
 άφαιρονμαι, 315 
 άφοίβαντος, 218 
 'Αφροδίτη, 247 
 '/i;j;aios, 97, 322 
 '^;^ελώθ5, 270 
 ά>, 169 
 
 βαγός, 254 
 |3a'9'v;uatos, 475 
 Βάκχος, 110 
 
 /?αλ7?ν, 479 
 /?άηω, 110, 436 
 |3aVa, 133 
 βάνανοος, 326 
 βαρβαρόφωνος, 88 
 βασιλεύς, 254 » 
 
 βείομαι, 378 
 βέλτερος, βελτίων, 262 
 
 479 
 /3/aios, 291 
 /3λά|, 218 
 βλάπτω, 454 
 /3λε7Ρω, 452 
 ^λ?7;^ρ05, 218 
 /3λω(?κω, 218 
 ^077, 284, 470 
 βοηΟ-έω, 284 
 |3όλλ?7, 466 
 βόβπορος, 310 
 |5ο{ία, 466 
 /?ονλτ/, 466 
 βονλομαι, 462, 466 
 βοννός, 469 
 βοντιαις, 466 
 ^ovg, 284, 470 
 /?oi5s εττί γλώαβΐ}, 468 
 βοναόα, 466 
 βονχίλος, 466 
 βρά'ψαι, 454 
 βρέφος, 432, 473 
 ^ρΰω, 432, 473 
 ^ώ /tAOS, 469 
 
 ya-O-m^ag, 323 
 
 yata, 323, 470 
 
 ya^or, 213, 459 
 
 γαλήνη, 459 
 
 γαμβρός, 217 
 
 yavos, 323 
 
 ya^, 204 
 
 yα1;ρds, 323 
 
 ydovirog 209 
 
 yε, 203 
 
 y8 ^Ty, 203 
 
 yg'Aci, 459 
 
 γέλααμα κνμάτων, 459 
 
 yελαω, 459 
 
 yελεΌvrεs, 459 
 
 Γε'λωι;, 459 
 
 γενετής, 2b4: 
 
 γενναίος, 323 
 
 yεVro, 272 
 
 γεραίρω, 297 
 
 γεραρόν, 297 
 
 yε9αρd(>, 297 
 
 y89as, 297 
 
 γερηνιος, 297 
 γέρων, 297 
 γέφνρα, 298 
 y^, 121, 323, 470 
 y^^ors, 297 
 yAavxos, 461 
 γλανοΰω, 452 
 γλήνη, 452 
 yz/Qffr?;^, 567 
 γομφόδετος, 475 
 γνμνός, 121, 410 
 ywij, 133j 
 
 ^άκρν, 212 
 ^άκτνλοί, 162 
 ^αμάΰιππος, 314 
 da0i;g, όανλο'?, 474 
 ^ε,^ 155 
 δείκννμί, 271 
 5είΐ/05, 255, 268, 324 
 άε/5, 156 
 δέκα, 161, 162 
 δέκομαι, 161, 271 
 ^f>orS, 290 
 ^ε|ί05, ρ. 301, § 292 
 δέομαι, 155 
 ^ε'ρκω, 152, 262, 269 
 δεομός, 253, 474 
 δέσποινα, 228 
 δεσποβιοναντης, 4:16 
 δεσπόσιος, 475 
 δεσπότης, 228, 474 
 ^εΰρο, 155 
 5εω, 155, 180 
 δη, 201 
 δ^-θ-εν, 202 
 δήλος, 265 ^ 
 δήμος, δημός, 253 
 δημόσιος, 298 
 ^τ?!/, 202 
 dijTis, 156 
 dm', 180 
 δίαιτα, 180 
 διαμάω, 218 
 διαπρνσιος, 150, 460 
 ^ίατελώ, 445 
 5ίαΓρο;ΐ;ά^ω, 174 
 διαφέρομαι, 180 
 διδάσκω, 219 
 δίδνμος, 180 
 d/^co, 180 
 διϋ•νραμβος, 317 
 δικαιόπολις, 291 
 δίκαιος, 290 
 όϊκ77, 289 
 δίοπενω, 433 
 
IT. INDEX 0Έ GEEEK WOEDS. 
 
 731 
 
 dioTtog, 433 
 δίοπτενω, 433 
 δίτίλαξ, 280 
 δί$, 155 
 δίΰκος, 219 
 ^ΐΰωτηριον, 318 
 διωλνγίθ£, 460 
 δόλιχος, 209, 344 
 δονπέω, 475 
 δραπέτης, 164, 432 
 δράχμη, 174, 477 
 5ράω, 164, 432 
 ^ρο>ο?, 164, 432 
 δνναμκι, 262 
 δύναμις, 323 
 ^νσ-, 180 
 δνβΟ'νηβν,ω, 437 
 δυαμενής, 180 
 δνοφρονα, 297 
 δνω, 155, 180 
 δώδεκα, 159 
 Δωριείς, 92 
 
 Ι, 140 
 
 iy^Tyyo^o;, 221 
 cytw, 133, 203, 275 
 εδομαι, 383, 467 
 ίελ^οί^αί, 463 
 ^ερστ?, 116 
 ί^ω, εΓσοί, εσω, 475 
 Ι-Ο-ελω, 463 
 ε/, 139 
 
 εΓ TtoT ϊην yε, 205 
 £t(Y, pp. 237, 251 
 εϊδος and \ .^ 
 /5εα, / ^^ 
 
 εΓ^ε, 202 
 εΓκαΓί, 162 
 εί'κω, 228 
 Είλείϋ-νίκ, 296 
 εΓν, 139 
 εΐνεκα, 277 
 ε^5, 170 
 sfs, 154 
 stacc, 475 
 ε/σορώ, 279 
 είτα, 169, 202 
 "Ετιάβη, 276 
 '£κάλ?7, 276 
 Έκαμήδη, 276 
 έκά?, 273 
 εκατί, εκτ^τ^, 273 
 εκατόν, 162 
 
 εκατοί, εκαιτί;, εκατερο^, 
 
 273 
 εκείνος, 135 
 
 εκδνομαι, 177 
 
 έκ?7/3όλθ5, 273 
 
 εκτ^λο?, 116, 273 
 
 ^κ-ό-ι^τ^ίΐκω, 176 
 
 εκνρό^, 110 
 
 εκφέρω, 176 
 
 έκφορώτερος, 176 
 
 εκα<ι;, 273 
 
 ελαύνω, 223 
 
 ελα^Κ^^'δ, 132 
 
 ^λεν-Ο-εροί, 212 
 
 ελεν^'ώ, 296 
 
 ελεφαρ, 296 
 
 ελκεΰίπεπλος, 314 
 
 έλκω, 464 
 
 "Ελληνες, 92 
 
 "Ελπινίκη, 312 
 
 εμβρυον, 473 
 
 εμμοτος, 472 
 
 'έμολον, 218 
 
 εν, 170 
 
 ένδατονμαι, 178, 450 
 
 εν5ελε;^?75, 344 
 
 ένεκα, 272 
 
 ενέργεια, 340 
 
 εν 166 ω, ένίτίτω, 216 
 
 εννέα, 161 
 
 εννος, 161 
 
 ε'νννμι, 215 
 
 ενοδ τε καΙ νε'ο^, 161 
 
 εντελέχεια, 340 
 
 εντός, 164 
 
 ε'Ι, 176 
 
 ε^, 160 
 
 ε'ξαί'Ο'ραΛενω, 160, 213 
 
 εξαίφνης, εξαπίνης, 215 
 
 εξί6ταμαι, 315 
 
 επάγω, 174 
 
 επαλλάσοίω, 174 
 
 επάλληλος, 174 
 
 έπαμάω, 218 
 
 επαμείβω, 174 
 
 'Επαμεινώνδας, 262 
 
 έπαμφοτερίζω, 174 
 
 επαρκώ, 284 
 
 έπεα πτερόεντα, 126 
 
 έπειτα of present time, 
 
 202 
 επεργα6ία, 174 
 επε6ελϋ•εΐν, 175 
 έπητνς, 254 
 επ/, 172 
 επιγαμία, 174 
 έπίηρος, 285 
 επι&εάξω, 472 
 έπιμαχία, 174 
 ίπιμίξ, 254 
 
 επινέμω, 174 
 έπινομία, 174 
 έπιραβδοφορώ, 174 
 έπιφέρειν οργάς, 480 
 επίχαρτος, 297 
 έπίχειρα, 174 
 έπιχρώμαι, 174 
 έποίχομαι, 174 
 έτττά^ 160 
 ίρείπω, 160, 212 
 έρεμβός, 317 
 έρετμός, 253 
 έ'ρί^, 115 
 Έρμης, 114 
 ερρω, 209 
 έρχομαι, 479 
 ^σ-θ-Ζω, 383 
 εσ^λο'ί, 217 
 ε6πέρα, 150 
 ε'ίίτε, 198 
 έ'σω, 475 
 έτερος, 144 
 ετί, 169, 193 
 εν, 160, 199 
 ενεβτώ, 2h4i 
 ενήκοος, 469 
 είίίο?, 473 
 ενκηλος, 116, 275 
 ενμαρής, 450 
 ευρυχωρία, 273, 280 
 ενδ, 199 
 εύχομαι, 199 
 εφιάλλω, 110 
 εφιάλτης, 317 
 εφ ω τε, 197 
 £;Cis/ 146 
 εω5, εώί, 257 
 
 έ:άλϊ7, 110 
 ζαμενής, 180 
 ^ά;ΐ;ρνσο?, 180 
 ^άω, 110, 216 
 έεά, 216 
 ^ε^'yos, 180 
 Ζε^ν, 202, 216 
 ζιγγίβερι, 216 
 ^vyoV, 136, 216 
 ^ω)7, 180 
 
 ^', i ί, ^7, &C. 199, 200 
 
 "Η^??, 329, 336 
 
 ^'ά?7, 201 
 
 T^dvs, 199 
 
 ^τίρ, 199 
 
 ί^λακάτ?;, 116 
 
732 
 
 π. INDEX OF GBEEK WORDS. 
 
 ηλεκτρον, 116 
 ημάτως, 298 
 ημείς, 136 
 ήμέν—ηδέ, 201 
 ημέρσ., 150 
 ■ημερήοίος, 298 
 ήμερος, 150 
 ήμι, 199, 436 
 τίμίΰν, 150 
 ^ί', ηνίδε, 193 
 ήνίκοί, 196 
 ήνοψ, 95 
 ήπαρ, 150 
 ήπειρος, 150 
 ήπεροπενς, 254 
 ήπντης, 475 
 ''/Γρα, 329 
 
 "Ηραιος, Ήρ/'αοΐος, 329 
 'Ηρακλής, 329 
 ^'ρω?, 329, 414 
 97i;S, 199 
 
 θάλασσα, 110, 473 
 ϋ'άμβος,&αϋμα, 318,478 
 <9•ε, 202 
 -δ-εαίνα, 228 
 -θ-εαω, 478 
 -Ο-ελνω, 464 
 -Ο-ελε/ΜΟδ, 465 
 -9•ελω, 463 
 &έναρ, 450 
 ^ε05, 473 
 
 -θ-εόσάοτοδ, &C. 310 
 ^ερ/^ω, 450 
 -θ-εσ/ΑΟ?, 253, 474 
 ϋ'έοπις, ^εΰτίρωτός, d'sa- 
 
 φατος, &c. 310 
 -Ο-εω, 473 
 Ο'ήγειν, 473 
 -Ο-τίρ, 110 
 %•τιτες, 473 
 ^/α(ίθ5, 318 
 θοα^ω, 472 
 %Όίνη, 474 ^ 
 
 θοό?, 473 
 Ο-όρνβος, 272, 318 
 Θρ«ξ, 92 
 Τραβάω, 272 
 θρίαμβος, 317 
 «θ-ρίναΙ, 318 
 Ο-ρίνια, 318 
 •θ-ριον, 318 
 -θ-ιίελλα, 477 
 0•νμέλη, 476 
 -O-v/uog, 471 
 
 Ο'νμοβ, 476 
 Ο'νμοναΟ'αι εις κέρας, 
 
 170 
 -θ-νρσό?, 318 
 «θ-ώκο?, 472 
 ϋ-ώΰαΰΟ-αι, 474 > 
 
 ί; £; 139 
 
 Γ(Κ, Γα, 154 
 
 "Ιαηχος, 110 
 
 ^άΠω, 110, 318 
 
 ί'αμβος, 317 
 
 ίάπτω, 317 
 
 Γ7ν7?τε5, 139 
 
 r^ios, 139, 166 
 
 ίδον, 193 
 
 2^ρώ?, 110, 167 
 
 Ικανός, 116 
 
 Ικέτης, 318 
 
 ί'κταρ, 475 
 
 Γι/, 139 
 
 Ίπποδάμμον, 220 
 
 ίππόκρημνος, &c. 468 
 
 ιππομανής, 466 
 
 '^Ιππομμέδοντος, 220 
 
 t'TTjrog, 110 
 
 r^ig, 464 
 
 ίϋαΐος, ίοαίτερος, 167 
 
 Γοκε, 219, 434 
 
 f(7os, 152 
 
 Γ(Γ;^ω, 219 
 
 t'iiy^, 464 
 
 2ων, 133 
 
 -ίων, 165 
 
 κά, 186 
 
 κά^^αοί, 267, 473 
 
 κα%-αρός, 267 
 
 καί, 186, 195 
 
 καϊ ταντα, και τοι, 198 
 
 καιρός, 171, 290 
 
 κακογείτων, 322 
 
 καλεω, 209 
 
 κάλλος, καλλοαννη, 258 
 
 καλοκα'/α'9'05, 321 
 
 καλο'ί, 324 
 
 κάλνξ, 254 
 
 Jfa;.;tas, 296 
 
 κάμαξ, 286 
 
 καρπός, 162 
 
 κατά, 182 
 
 καταί'9"ϋσ()ω, 458 
 
 καταρτυω, 218 
 
 κατάφημι, 184 
 
 κατηγορία, 125 
 
 κατηρτνκώς, 218 
 καυλός, 163 
 κεί-θ-εν, κεϊ'Ο'ί, 186 
 κεΓνο?, 135, 138 
 κεν, 186, 195 
 κέντο, ρ. 298 
 κεραί, 209 
 κεφαλή, 216 
 κεχλάδειν, 336 
 κιοοοδέτης, 254 
 κλυτο'?, 212, 478 
 κλνω, 209 
 κ^ελεΌ'ρα, 121 
 κόθεν, 182 
 κοίρανος, 336 
 κο'λττο?, 458 
 κολώνη, 410 
 κομμώ, 215 
 κόμπος, 317 
 κόπτω, 317 
 κόρο?, 335 
 κορν?, 216, 262 
 κορυφή, 216 
 κόί>/Μ0?, 215, 267, 473 
 κότταβος, 216 
 κουρίδιος, 330 
 κραίνω, 83, 479 
 κραιπνός, 160 
 κρείσσότεκνοδ, 
 
 -τέχνης, 312 
 κρήδεμνον, 410 
 κρήνη, κρουνός, 176,255 
 κτείνω, 208 
 κυβερνάω, 216 
 κύκλωιρ, 154 
 κυΛτω, 216 
 κύρβας, 331 
 κυρήβια 331 
 κυρίο^, 330 
 κυωι/, 110, 162 
 
 Aaag, 455 
 
 λαΐ^νρίν^ο?, 263, 453 
 λα)';^άνω, 209, 216, 453 
 Λαέρτης, 332 
 ^ai'g, 455 
 λαμβάνω, 453 
 λα/ΕΑΤτρό?, 452 
 λάσκω, 219, 387 
 λάω, 451 
 λεα^α, 228 
 λε'γω, 453, 455 
 λε/α, 455 
 λειποταξία, 314 
 λείπω, 223 
 λεί;ίω, 223 
 
II. INDEX OF GREEK WORDS. 
 
 733 
 
 Ae'lis, 125 
 λέϋχη, 219, 453 
 λευκός, 269, 456 
 Xtvs, 452 
 λεναΰω, 452 
 λέχος, 453 
 λέων, 455 
 λεωργός, 313 
 λτ^Γδ, λαΐζ, 455 
 A/Oos, 455 
 λίπα, 272 
 λιπαρής, 456 
 λίΛαρο?, 456 
 λογάδην, 455 
 Zoyos, 124 
 Aoyoo-TtaAaiog, 310 
 λοξίας, 154 
 λο|ο?, 154 
 λονω, 270 
 λντιαβας, 269 
 AuxoriiOr, 228 
 λνκείΟδ, 269 
 λιίκτ;, 269 
 λυμαίνομαι, 305 
 λυμεαν, 305 
 λύρα, 478 
 λυτήριον, subst, 297 
 
 /lia, 135, 189 
 
 μάλα, 185 
 
 μάλλον, 165 
 
 μανΟ'άνω, 209 
 
 μαραίνω, 218 
 
 μάρη, 450 
 
 μαρηγει, 454 
 
 μάρμαρος, 95, 457 
 
 μάρητω, 454 
 
 μάρτυρ, 450 
 
 μαυρός, 218 
 
 juaoo, 135, 224, 472 
 
 /tt£, 132 
 
 μέγεϋΌς, μέγιοτος, 167 
 
 μέδιμνα, 450 
 
 /ί^ε-Ο-εν, 134 
 
 μεΟ•ίημι, 4,1b 
 
 μειδιάω, 199 
 
 /Μεί'δ, 154 
 
 μέλας, 121 
 
 /ΐιελεί, 472 
 
 /[^ελλω, 448, 450 
 
 /«ελο?, 178 
 
 μέλπω, 458 
 
 μέμβλεται, 217 
 
 μέμβλωκα, 217 
 
 /Είεν, 154 
 
 ^ενω, 472 
 
 /ttfOi, /Μ-εόί, 134 
 μέριμνα, 410 
 μερμηρί^ω, 410, 450 
 ^ε'ροδ, 178, 450 
 μέροιρ, 95 
 μεΰίτης, 450 
 μέΰΰος, 166, 450 
 /Μετά, 130, 181, 450 
 ί^ετα^, 202, 265 
 μετέωρος, 173 
 ^ε;ΐ:ρί, 181 
 /tiTy, 189 
 iiijV, 189 
 μηνύω, 269 
 μητροτιτόνος, 313 
 μητρυιά, 414 
 /Mtor, 154 
 /u-iv, 135 
 (tioyiS, 167 
 /ttoAts, 185 
 μόνος, 154 
 μόΰχος, 219 
 ί*ο';Κ'^θ5, 383 
 μύριοι, μύρω, 163 
 
 mt, 189 
 
 ι^αίί?, 216 
 
 νέμειν το ΐβον, τα ιοα, 
 
 454 
 νεοοίγαλος, 461 
 νεω, 216 
 ν?/, 189 
 
 Νηρενς, 95, 216 
 νηριτος, 95, 185 
 νηαος, 216 
 ν/κτ?, 216 
 ν/έ:ω, 216 
 νίπτω, 216 
 νιφάς, 199 
 vdcrros, 136, 164 
 νόβφι, 136 
 νυ, ννν, 192 
 vvoff, 199 
 νώΐ-, 136 
 νώροιρ, 95 
 
 ξέννος, 215 
 Ιει/οδ, 176 
 ^ερ|?;&, 160, 479 
 |εω, 221, 432, 436 
 |/φοδ, 221, 432, 436 
 |w, 181 
 Iwog, 181 
 |νω, 433 
 
 ο, 137 ^^ 
 "Oo;|os,"Ocf<Jiff, οΓ(?τρο5, 
 
 &C. 110 
 όγδοος, 164 
 ο|ίε, 15J8 
 οάε, ovrog and κεΓίΌ^, 
 
 οδεινα, 156 
 o^ovg, 132 
 'Οδυοΰεύς, 115, 167 
 ό-θΌυνεκα, 277 
 οΐδα, 110 
 
 οίκητής, οίκητωρ, 267 
 oi'i'jy, οίνίζω, 154 
 οίνος, 110 
 ofog, 152 
 ofo's τε, 197 
 οΓα-θ-α, 110, 353 
 οκτώ, 159 
 όλβαχήϊον, 116 
 δλ/5θ5, 116 
 
 oAiyi(?TOS,oAiyo(ir(}S,164 
 ^ 167 
 
 ολλυμι, 215 
 Ολο?, 116 
 ομβρος, 217 
 όμόοπορος, 313 
 o^djg, 247 
 όνειδος, 212 
 oVoita, 124, 132 
 Όξάϋ-ρης, 160, 213 
 οπλίτης, 259 
 οπλον, 259 
 
 οπτομαι, οΰΰομαι, 216 
 όπυίω, 213 
 οττω?, 196 
 όράω, 209 
 όρ7?7, όργάω, 479 
 υργυια, 296, 414 
 όρέγω, 477 
 ορΟΌ'Λου^, 315 
 ορ^α-θΌ'ί, 383 
 δρνίί, 262 
 ορο?, 174 
 ορχαμος, 479 
 ορχηβις, ορχηβμός, ορ- 
 
 χηατύς, 254 
 OS, 148, 243, 300 
 όΰκάπτω, &c. 213 
 δσον τε, 197 
 ο σο 5, 152 
 δ'ττε, «&C. 197 
 ον, ουκ, 96, 176, 189 
 ου /α?7, 394 
 ουδείς, 156 
 ούΟ'είς, 156 
 ουκ— αλλά, 201 
 
734 
 
 π. INDEX OF GREEK WORDS. 
 
 Ούκαλεγων, 189 
 ovv, 189 
 οννεκα, 277 
 ουρανός, 259 
 ovg, 189 
 OvTig, 189 
 Ουτοπία, 189 
 οντος, 135, 138 
 οφρα, 196 
 οφρν?, 132 
 6';ΐ:λθ5, 221 
 
 παιδεία, 4 
 TTOfiff, 262 
 παλαίοΛλουΓΟί, 12 
 παλαιός and αρχαίος, 12 
 τταλοτίόί, 298 
 παλαίφατος, 315 
 παλίγγλωοοος, 88 
 παλίμβαμος, 174 
 πάλιν, 264 
 ττανόϊ, 199 
 πανούργος, 313 
 τταο/, 262 
 παρά, 177 
 
 Παρό'εννοπαΐος, 220 
 παρ%'ενών, 260 
 παριππενω, 178 
 πάροί, 171 
 πας, 265 
 
 πάΰχω,ίί4:, 219, 434 
 πατρικός, πατρφος, &c. 
 
 414 
 Λε^ά, 158 
 πείρω, 178 
 
 πελα)Ό5, πελάγιος, 280 
 πε>1αρ7()?, 195 
 ΙΓελο;ί)7ο'5, 95 
 Πέλο^, 95 
 π ε μπάζω, 161 
 ττεν^Όί, 114 
 πε'ντε, 161, 162 
 πεπαρεΐν, 178 
 πεπονΟ-α, 441 
 πέραν, πέρνΰΐ, 177 
 περάω, 178 
 περΟ-ω, 382 
 ττερί, 177 
 περιπετής, 178 
 περιπέττω, 289 
 περίχριμπτα, 475 
 πέρπερος, 178 
 τίερφερέες, 178 
 πέββω, 216 
 ττε'τορεδ, 158 
 πέτρα, πέτρος, 15, 229 
 
 πηλίκος, 152 
 Λίκρ05, 218, 266 
 πίομαι, 383, 467 
 πίπτω, 431 
 TcZayHTo's, 280 
 ττλάί, 280 
 πλε'ω, 270 
 πληγή, 199 
 πλτ/ν, 200 
 πό-θ-εν, 202 
 ποιέω, 213 
 ποίημα, 410 
 ποι-ϋίλος, 266 
 ποί'Λΐλόϋτιν,τος, 266 
 TTOi/UTyv, 133, 256, 410, 
 
 436 
 ποινή, 410 
 
 πολίτης, πολιήτης, 259 
 πολλοστοί, 164 
 ττολ^ϊ, 479 
 πόλ;^ο?, 221 
 πορενματα, 218 
 ττορ-θ'/αο^, 254 
 iToffrdg, 164 
 ποτμός, 253 
 jrorvm, 228, 474 
 3Τρα70&, πράγμα, 224, 
 
 256 
 προ, ττρό?, προτί, ποτί, 
 
 171, 177 
 προβατογνώμων, 410 
 προοτρέπομαι, 218 
 προυβελεΐν, 461 
 πρνλεε^, 154 
 πρωί^ο?, 216 
 πτώσί?, 227 
 πνρΰός, 318 
 3Γοο, πώμαλα, 202 
 
 ρα, 202 
 ^ά|3άθ5, 174 
 ρέπω, 477 
 ρεω, 270, 477 
 ρήμα, 124 
 ρίμψα, 477 
 ^ίπτω, 160, 477 
 ρυ-θ-ΖίΑό?, 253 
 ρνμός, 253 
 ρυΰμός, 474 
 
 βαίνω, 473 
 αάλαβαα, 461 
 οαλαοΰομέδοιΰα, 461 
 (ϊαρώ, 475 
 οατράπης, 160, 213, 475 
 
 ΰαφής, 181 
 σε, 132 
 ()ε|3ω, 473 
 οειομός, 253 
 αέλας, 461 
 ΰελήνη, 461 
 σειίω, 473 
 ΰημαίνω, 436 
 σ/αλθ5, 222, 461 
 ΰίγαλοείς, 461 
 σί05, 473 
 σκάλλω, 387 
 σκελθ5, 387 
 Σηαπτηΰνλη, 114 
 ο-κήνη, 410 
 ομικρός, 199 
 αονμαι, 473 
 βπάκα, 110 
 
 σπλα7;^νοί', σ^λτ^ν, 255 
 ΰπονδη, 223 
 6ταδιοδρομέω, 437 
 ΰτενοχωρία, 280 
 ΰτέρο'ίρ, 95 
 βτόρννμι, 223 
 σν, 133 
 σνν, 181 
 ί)νρί7^7εν7;5, 473 
 ΰνριον αρ/αα, 473 
 σφάλλω, 209 
 σφετερο?, 142 
 σφτ^Ι, 142, 221 
 σφ05, 142 
 6φώϊ, 136 
 σ;^είν, 434 
 σωκεω, 447 
 σώκθ5, 447 
 
 r«/ttiaff, 262, 440 
 
 τάΐΛ, 132 
 
 ταω?, 272 
 
 τε, 149, 186, 195 
 
 τεΌ-ί^ό^, 474 . 
 
 τείνω, 209 
 
 τελείό?, τελείόω, 218 
 
 Γελλενταντοδ, 220 
 
 τελθ5, 178, 342 
 
 τέμενος, 267 
 
 τεο/, 133 
 
 τερΌ^ρίΟ^, 318 
 
 τέρμα, 187 
 
 -Γερο?, affix, 157 
 
 τέρπω, τρέπω, τρέφω, 
 
 440 
 τέαβαρες, 146, 158 
 τ/^ε, 132 
 τ?ϊλε, &C 344 
 
π. INDEX OF GEEEK WORDS. 
 
 735 
 
 τηλέβολος, 437 
 
 τηλιξ, τηλίζος, 152 
 
 τηλνγετος, 344 
 
 τημος, 202 
 
 τηνίκα, 196 
 
 τί&ημι, 347 
 
 τιονχαν, 223 
 
 τις, 149, 156 
 
 ro re ^ν είναι, 192, 344 
 
 roi, τοίγαρ, τοίννν, 198 
 
 τόν, 132 
 
 τόρο?, τόρνος, 178 
 
 τόσο?, 152 
 
 tovv, 133 
 
 τραύμα, 178 
 
 τραφερός, 344 
 
 T^fts, 157 
 
 τρίΛολί()Τθ?, 167 
 
 r^trrufg, 154 
 
 τρόπος, 290 
 
 τρνω, 178 ^, 
 
 Tvypjavctj, 445 * 
 
 τνμβόχωοτος, 309 
 
 τνρβααία, 318 
 
 τνρ/3ί7, 318 
 
 ΰάκίί^ΌΌ?, 464 
 vj^^is, 335 
 νδωρ, 167, 223 
 rids, 167 
 νλη, 114, 340 
 v/[A£ts, 136 
 ντίέρ, νπό, 178 
 ντζερακταίνω, 447 
 υπερφίαλος, 139 
 υττνος, 110 
 υτζοφανοις, 152 
 Ύρρά^ίο?, 262 
 νβμίνη, 136 
 
 φαινόμηρις, 314 
 φαίνω, 460 
 φανός, 199 
 φάοί, 152, 460 
 φαρέτρα, 440 
 φάρσο?, 199 
 φάσ^αί/ον, 221 
 φανλθ5, 152, 327 
 φ?/, 199 
 qo?7/tAt, 209, 460 
 φ-Ο'άνω, 445 
 φιάλλω, 110 
 φοναα, ρ. 249, note 
 φρίΰβω, 218, 288 
 φροίμιον, 102 
 φρουρό?, φρονρο'?, 272 
 φν/ω, 377 
 φνλάσΰω, 176 
 φϋλλον, 165 
 φυω, 133, 167 
 φωντ;, 88, 110, 199 
 φώ^, 257 
 
 χαίνω, 280 
 ;^α/ρω, 288, 445 
 χανδάνω, 280 
 ;ΐ;άο?, 280 
 ;i;ads, 322 
 ;ΐ;αρακτ7^'ρ, 286 
 ;^:αραέ, 286 
 χαράοΰω, 286 
 χάριν, 278 
 ;ΐ;άρί5, 291 
 Χάρίτεσ, 291 
 ;ΐ:άρ^77, 288, 325 
 χαροτΐός, χάροτρ, 282 
 χάρνβδις, 282 
 ;^άρων, 282 
 ;ΐ;εαο5, 163, 280 
 
 ;^ε/ρ, 162, 28 
 χειρών, 262 
 χεράς, 281 
 χέρμας, 281 
 χερΰος, 281 
 ί^ί'ω, 163 
 
 ;ΐ;ε'ω qc)'9'dyyo7/,a'yiai',469 
 ;ΐ;?5ρθ5, 287 
 ;i^fS, 150 
 χϋΊξός, 216 
 χίλιοι, χιλός, 163 
 ;^λά^ω, 336 
 ;ΐ:λήδθί, 336 
 ;ίλίά77, 336 
 χοιράς, 281 
 χοίρος, 281 
 ;^ορ()5, 280 
 χραιαμέω, 284 
 χράομαι, 287 
 χρανω, 281 
 χρεία, 284 
 χρήοιμος, 284 
 χρίμητω, 281 
 ;^ρόί/ω-κλυτ05, 310 
 ;ΐ;ώρα, ί^ώρο^, ί^ορό^, 229, 
 280 
 
 ιράλιον, 221 
 ιράλλω, 432, 436 
 ψε, 144 
 i/;tv, 221 
 
 ώκνί, 473 
 ωλαέ, 116 
 ώ|Μ05, 116 
 ώ^, 170 
 ώ'στε, 197 
 ώ τάν, 133 
 
ΠΙ. INDEX. 
 
 MISCELLANEOUS MATTERS. 
 
 Abstraction, 55 
 ac and atque, 194 
 Acts of the Apostles, xvii. 
 
 26; 15 
 Adelung (J. C), 36 
 adoro, 479 
 ceger, cerumna, 410 
 Agrippa (H. Cornelius), 
 
 52 
 Airyamen vaejo, 81, 85 
 Algebra, 55 
 aliquis, quispiam, quis- 
 
 quam, 199 
 alius, 135, 138 
 almus, alumnus, 253, 410 
 Alphabet (Indian), 105, 
 
 note 
 "and," 194 
 Annakus, 66 
 Anquetil du Perron, 35 
 Antecedent, indefinite, 
 
 300 
 Arabic Alphabet, 104 
 arcesso, &c. 382 
 Architectural writing, 45 
 Ariana, 81 
 Ariarathe^i 81 
 Ariomardus, 213 
 Armenia, 64, 85 
 Arnold (Dr. Ϊ.), 32,305, 
 
 396,^ 480 
 Arya-avartah, 81 
 Ascham (R.), 25 
 Askenaz, 76 
 
 Association of ideas, 53 
 auctumnus, 410 
 
 Babel, 45 
 
 Basque or Euskarian 
 
 language, 79 
 Behistun Inscription, 13, 
 
 37,107,110,133,149, 
 
 194, 344, 479 
 Belooch, 81 
 
 Benfey (Dr. Theodor), 
 168, p, 696 
 
 Bentley(Dr.R.), 28,110, 
 pp. 219 sqq. 
 
 Berkeley (Bishop), 57 
 
 bes, 155 
 
 Bibliolatry, 13 
 
 Blomfield (Bishop), his 
 ingenious emendation 
 of Matreas, 286 
 
 Bohemia, 79 
 
 Bopp (F.), 39 
 
 strictures on, pp. 
 
 259, 344, 370, 401, 
 404, 424, 435, 444, 
 465, 473, 585 sqq., 
 594, 597 sqq., 608 sqq. 
 
 bruma, 150 
 
 Budaus (W.), 24 
 
 Buddhists, 84 
 
 Budenz (J.), p. 453 
 
 Bunsen,(C.C.J.), 10,15, 
 30, 37, 48, 208 
 
 Burnouf, E. (36, 108), 
 J. L. 372 
 
 Buttmann (P.), 40 
 
 Ccecus, 154 
 Caldwell (Mr.), 81 
 caleche, 93 
 
 Carians and Cretans, 95 
 carmen, 410 
 carnifex, 295 
 castra, 267 
 Chalybes, 95 
 Cheke (Sir J.), 25 
 Chinese language, 50, 
 
 148, 157 
 Cicero, 194, 344 
 Civilization, linguistic 
 
 records of, 11 
 clarus, gloria, 266, 287 
 Classical learning, value 
 
 of, 8 
 
 codes, 154 
 
 Colebrooke (H. T.), 36 
 Coleridge (S. T.), 5, 51, 
 
 236 
 conari, 447 
 
 Confusion of tongues, 45 
 consul, 269 
 Contrast (association 
 
 by;, 53 
 cormorant, 315 
 cornu, 209 
 
 Cratylus, Plato's, 60 
 crepusculum, 160 
 crus, 160 
 Curtius (M. George), 40, 
 
 370 
 Cyclopean walls, 455 
 Czechs, 79 
 
 Dardanus, ' 92 
 
 Deccan, 81 
 
 Deduction and induc- 
 tion, 5 
 
 Demonstrative used for 
 relative, 148 
 
 dempster, 225 
 
 Desdemona, 180 
 
 Deva-nugari, 105 
 
 Difi'erential Calculus, 55 
 
 Digamma, 110 
 
 dodrans, 161 
 
 Dorset, 92 
 
 dum-taxat, 202 
 
 Dun-cow, 470 
 
 Dyer (Mr. T.), p. 104 
 
 Education, information, 
 and knowledge, 2 
 
 Egyptian letters, 103 
 
 Egyptology, 37 
 
 eja, 237, 251 
 
 Elephant and tamarind, 
 296 
 
 eleven, twelve, &c. 159 
 
 3 Β 
 
738 III. INDEX OF MISCELLANEOUS MATTERS. 
 
 England, 76, p. Ill, note 
 English language, 74 
 English scholarship, 22 
 Escalus in Shakspere, 
 
 213 
 Ethnology, 10 
 Etruscan, 89, 147, 149 
 Ewald (H.), 100 
 
 farrow and furrow, 281 
 filius, 177, 270 
 fons, 295 
 Frisians and Angles, 78 
 
 Galiffe(Mr.),p.l49,note 
 
 gamester, 225 
 
 Garnett (Rev. R.), 37,79, 
 
 121, 129, 133, 148, 
 
 298, 347 
 gas and ghost, 225 
 Gelo and Gela, origin 
 
 of their names, 459 
 Genesis (Book of), 44 
 
 —46, 65, 66, 71 
 genteel and gentle, 278 
 German Literature, 30 
 Germanus, 76, 213 
 Getce, 77 
 Gladstone (Mr. W. E.), 
 
 p. 48 
 «God" and good, 473 
 Gomer, 72 
 
 Grammar-schools, 1, 83 
 Gray (T.), 48 
 Greeks and Germans, 92 
 Grimm(J.), 36, 108,118 
 corrected pp. 187 
 
 213, 247, &c. 
 Guest (Dr. E.), 118 
 guna, 106 
 Gypsy, 225 
 
 Hadryades, 181 
 Halhed (Dr.), 35 
 Hamadryades, 181 
 Hamaker (Professor), 40 
 Ha-milcar, &c. 479 
 Hamilton (Sir W. R.), 55 
 Hamites, 70 
 Hebrew language, illu- 
 strations of, 100, 102, 
 116, 133, 137, 148, 
 157, 184, 186, 189, 
 193, 199, 208, 209, 
 234, 296, 479 
 henbane, 473 
 Hengist and Horsa, 78 
 
 Hercules, 'Ηρακλής, 222 
 Hermann (G.), 29, 38 
 Herminones, Hermunduri, 
 
 &c. 76, 92, 213 
 Heyne (Prof.), 29 
 hie, 139 
 
 Historical criticism, 13 
 Homeric deities, 463 . 
 Houssaye (Arsene), 47 
 humanitas, 4 
 Humboldt (W. von), 42 
 humerus, 116 
 Hunter (Prof.), p. 256, 
 
 note 
 Hyacinthus, legend of, 
 
 p. 696, note 
 
 I termination, 139 
 
 Iconium, 66 
 
 Ideas (doctrine of;, 58, 59 
 
 idoneus, 116 
 
 "if," 205 
 
 igitur, 362 
 
 ilk, 166 
 
 Imagination, p. 90, note 
 
 immOf 191 
 
 Indo-Germanic, 71 
 
 indulgeo, 344 
 
 Ingcevones, 76 
 
 iniquus, ingens, 185 
 
 inter, 204 
 
 interficio, intereo, 382 
 
 Iran, 80 
 
 iri, 447 
 
 IsccevoneSylb, -p. 121, note 
 
 Jackson (Dr. Cyril), 192 
 Jones (Sir W.), 35 
 
 Kant (I.), 54 
 Kemble (Mr. J. M.), 37 
 Kenrick (Rev. J), 32 
 Key (Mr. T.H.), pp.238, 
 note, 248, note, 340, 
 note, 436, note, 465, 
 note, 474 sqq., 587, 
 note. 
 King, and kingly titles, 
 
 337, 479 
 Klaproth (H.J.), p. 360, 
 note 
 
 Language (its regular 
 
 changes), 50 
 Latin participles, 295 
 Law and Observation, 5 
 Layard (Mr.), 37 
 
 Leibnitz, 34, 5b 
 Lepsius (Dr. R.), 13, 37, 
 
 100, 153, 162 
 Lewis (SirG.C), pp. 48, 
 
 709 
 Light and sound, 460 
 Lithuanians, 77 
 Lobeck (Professor), 38 
 lobster, 212 
 longus, 344 
 lord, lady, 338 
 loup-garou, 110, 315 
 lubrican, 221 
 Lucretius, 54 
 Luna, 461 
 
 Luther, 15, 20, 21, 47 
 Lycian language, 92 
 
 Magog, 72 
 
 malus, 185 
 
 manifestus, 450 
 
 Μ annus, 76 
 
 Mansel (Mr. H. L.), 46, 
 
 55, 125 
 Marcomanni, 76 
 Mardi, Mardonias, Mar- 
 
 dontes, 213 
 -ματ- — -μεντ, 114,133, 
 
 256 
 Matthew (St.) xvi. 18; 15 
 Melancthon, 15 
 Melicarthus, 479 
 Mexico, 67 
 
 Meyer (Dr.K.),p .161 ,note 
 miles, 163 
 
 Miiller, K. O., 29, 38 
 MuUer, Max, 11,39, 444 
 murus, 178 
 
 Names of objects, 44,454 
 
 Nero, 332 
 
 Niebuhr (B. G.), 29 
 
 Nod (land of), p. 116, note 
 
 Nominalism, 18, 20, 56 
 
 non, 188 
 
 Norris (Mr.E.), 37, 104 
 
 novem, 161 
 
 numerus, 116 
 
 ob, 172 
 obambulo, 174 
 obedio, 128, 222 
 Occham (William), 19 
 octavus, 164 
 olimj 166 
 opidentus, 152 
 Ovid, Fast. v. 21; 99 
 
III. INDEX OP MISCELLANEOUS MATTERS. 
 
 739 
 
 page^ 225 
 
 palcetiology, 12 
 
 Pali, 84 
 
 paries, pars, 178 
 
 Paris and Priamus, 92 
 
 Pelasgians, 95 
 
 Perception and Concep- 
 tion, 51 
 
 peregrinus, 178 
 
 pereo, 382, 479 
 
 Personincation of the 
 powers of nature, p. 
 234, note 
 
 Philistines, 95 
 
 Philology, 3 
 
 Phoenicians, 94 
 
 Plato, 57 sqq. 
 
 pons, 295 
 
 pontifex, 295 
 
 porca, porous, 281 
 
 Porson (R.), 24, 31 
 
 posco, 209 
 
 posthac, quapropter, &c. 
 240 
 
 Pott (A. F.), 38, 168 
 
 Pracrit, 83, 108 
 
 prehendo, 281 
 
 Prichard (Dr. J. C), 37 
 
 Prinsep (James), 37, 105 
 
 Printing, invention of, 
 22, 47 
 
 Prolepsis, 300 
 
 propitius, 284 
 
 Prose,introduction,of, 48 
 
 pugna, 410 
 
 Pygmies, 81 
 
 quarrel, 225 
 quick, 112 
 
 quintus, not quinctus, 161 
 quispiam, quivis^quisquam, 
 199 
 
 Rabelais, 395 
 Ramuyana, p. 135, note 
 Rawlinson(SirH.C.),37 
 Realism and Nomina- 
 lism, 5, 18 
 Reason and Understand- 
 ing, 5, 51 
 refert=reifert, 240 
 rego, rex, &c. 479 
 Reiz (F. W.), 29 
 Relative Sentence, 148, 
 300 
 
 renideo, 459 
 Roads, 150, 479 
 Roget de Belloguet, 79 
 rogo, 209 
 
 Romance languages, 49 
 Rosen (Dr. F.), 55 
 Roth (Dr.), 95 
 Russians, 77 
 
 Sacse, 93 
 Samogetge, 77 
 Sanscrit, 83,84,108,399, 
 
 420, 455, 479 
 Satrap, 160, 213 
 Sauromatse, 82 
 Saxons, 74 
 
 Scaliger (J. J.), 15, 24 
 Schlegel(A.W.von), 36 
 Schleicher (Dr. Α.), 112 
 Schleiermacher (Dr.F.), 
 
 29 
 Sclavonians, 77, 88 
 Scythians, 88, 93 
 secundus, 155 
 securis, 414 
 ShiUeto (Mr. R.), 202 
 slave, 225 
 sonus, 110, 199 
 Sophists (old), 61 
 soror, 225 
 
 Space and Time, 54 
 Spanish h, 111 
 spinster, 225 
 squire, 213 
 
 Steinthal (Dr.), 3, 148 
 supplex, 318 
 surdus, 460 
 Suwarrow, 225 
 swart, 460 
 Sykes (Col.), 84. 
 Syntax, 48 
 
 Tacitus, German., 76 
 tapster, 225 
 Telchines, 95 
 Telegraph, p. 667 
 templum, 267 
 teres, 178 
 
 Testament (Old), 61,126 
 Teutamus, 92 
 Thiersch (Prof.), 101 
 Thirlwall (Bishop), his 
 
 History of Greece, 32 
 Thompson (Prof.W.H.), 
 
 58, 59 
 
 Thuringians, 76, 92 
 
 Time, 55 
 
 Tooke CMr. J. Home), 
 
 61, 126 
 Tower-builders, 45 
 Trendelenburg (Dr. Α.), 
 
 125 
 Triopian rites, 459 
 Troubadours, 49 
 Troy, 92 
 Tuhalqain, 25 
 Turanian, 69 
 
 Ulphilas, 82 
 understanding, 5, 51 
 Unity of man, 43 
 uxor, 225 
 
 valeo, 379 
 vapor, 458 
 vel, 199 
 Venus, 285 
 vestibulum. 180 
 vir, 332 
 Virgil, (reor^ .III. 192 ; 174 
 
 232; 170 
 
 jE7i. IX. 6; 188 
 
 vivo, 112, 378 
 
 Vriddhi and Guna, 105 
 
 Walls of picked stones, 
 
 455 
 Wansheck-water, 11 
 Weber (Α.), 105 
 Weight of vowels, 101 
 Welcker (F.G.), 29,457 
 Wends, 82 
 Werdin(J.), (ie.P.Paul. 
 
 a S. Barthol.), 35 
 Whewell (Dr. W.), 12 
 Williams (Monier), 105 
 Winning (Rev. J.), 37 
 Wolf (F. Α.), 29 
 Writing, origin of, 46 
 effects on language. 
 
 48 
 
 Xerxes, 479 
 
 Ya(^na, 86 
 
 Zend, 86 
 Zeuss, 79 
 Zoroaster, 85 
 
 THE END. 
 
Peinted by F. a. BrockhauS) Leipzig. 
 
 9.^4 
 
 92^ 
 
RETURN TO: CIRCULATION DEPARTMENT 
 198 Main Stacks 
 
 LOAN PERIOD 1 
 
 Home Use 
 
 2 
 
 3 
 
 4 
 
 5 
 
 6 
 
 ALL BOOKS MAY BE RECALLED AFTER 7 DAYS. 
 
 Renewals and Recharges may be made 4 days prior to the due date. 
 Books may be renewed by calling 642-3405. 
 
 DUE AS STAMPED BELOW. 
 
 1 
 
 3UL 1 8 2005 
 
 - 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 . 
 
 
 
 
 
 
 
 
 
 FORM NO. DD6 UNIVERSITY OF CALIFORNIA, BERKELEY 
 50M 1-05 Berkeley, Califomia 94720-6000 
 
C03asm4as