'"'y^ 
 
LIBR^RV 
 
 OF THK 
 
 University of California. 
 
 GIKT OF" 
 
 Class \ Y/ 
 
THE 
 
 NEGATIVE COMPOUNDS 
 IN GREEK. 
 
 A DISSERTATION 
 
 PRESENTED TO THE BOARD OF UNIVERSITY STUDIES 
 
 OF THE JOHNS HOPKINS UNIVERSITY FOR THE 
 
 DEGREE OF DOCTOR OF PHILOSOPHY 
 
 BY 
 
 HOLLISTER ADELBERT HAMILTON, 
 
 PROFESSOR OF GREEK IN ELMIRA COLLEGE. 
 
 ^NfVERsnv 
 . (s. or 
 
 BALTIMORE 
 1899 
 
JOHN MURPHY COMPANY, PRINTERS, 
 BALTIMORE. 
 
r\Z^uz:) 
 
 H3 
 CONTENTS. M/XIN 
 
 Page. 
 I. Introductory, 5 
 
 II. The Form of the Prefix, 6 
 
 The term alpha-privative. — Suggestion as to its possible origin. — 
 The Indo-European negatives. — Accent. — The form of the 
 prefix in Greek, lengthening of it, a- before vowels, the forms 
 I/7J- and 01/0-. ..._...-. 
 
 III. The Form and Classification op the Compounds, - - 15 
 
 IV. The Limitations on the Use of the Prefix in Composition, 17 
 
 The original sphere of the prefix. — Its use with the partici- 
 ple, — with the infinitive, — with the finite verb, — with nouns 
 in immutata, — with adjectives. 
 
 V. The Favorite Types of Negative Compounds, - - - 27 
 
 The compounds with the verbals in -tos. — The mutata of a- 
 privative -{- noun. 
 
 VI. Expressions which may replace the Negative Compounds, 29 
 
 Use of the sentence-negative, — of Svcr- and kuko-, — of preposi- 
 tions in composition, — of certain verb-stems, — of certain 
 adjective-stems.— Substitutes for the derivatives. 
 
 VII. The Semasiology of the Negative Compounds, - - 35 
 
 Development from original free negative to the negative in 
 composition. — Negative and contrary significations, thence 
 a positive content. — Prefix with the force of a sentence- 
 negative. — Privation and negation. — Hyperbole. 
 
 VIII. The Negative Compounds as an Element of Style, - 41 
 
 Stylistic character of compounds in general. — The artistic and 
 technical spheres. — Stylistic effects of the negative com- 
 pounds. — Massing of them together, alliteration, anaphora, — 
 feeling in privation, — triplets. — Figura etymologica. — Oxy- 
 moron. — The proportional metaphor. — Antithesis. — Litotes. 
 — Extension in form for phonetic impress! veness. 
 
 IX. History of the Formation of the Negative Compounds 
 
 IN Greek, 53 
 
 Compounds possibly inherited from the proethnic speech. — 
 Tables showing emergence of the compounds in the litera- 
 ture. — Contributions of different portions of the literature to 
 the number of negative compounds. — The various classes of 
 negative words, — d-privative -|- adjective, — growth of com- 
 pounds with verbals in -tos at the expense of the mutata, — 
 derivatives, especially the abstract nouns. — The vt)- com- 
 pounds. 
 
 3 
 
THE NEGATIVE COMPOUNDS IN GREEK. 
 
 I. INTRODUCTORY. 
 
 The study of the negative compounds in Greek may afford a 
 single chapter in the history of nonn-composition in that language, 
 a subject which has not yet received a comprehensive treatment, 
 nor one which is commensurate with its importance in more than 
 one department of philology. The compounds of alpha-privative 
 make one of the largest groups and one of the simplest types of 
 composite word-formation and the investigation has a varied interest 
 as we may approach our task from the formal, the semantic, the 
 stylistic, or the historical side. Thus we may consider first the 
 forms in which the negative prefix appears, the restrictions of its 
 use in composition, and the types of compounds which are formed 
 by it. Again, the semasiological character of these compounds 
 claims attention, both as to the various shades of negative concep- 
 tions and relations which they may express and as to the part 
 which they play in comparison with other means which the lan- 
 guage possesses for negative expression. In the third place the 
 employment of these compounds has a stylistic importance, espe- 
 cially where they appear massed together, where they belong to 
 certain poetic or technical spheres, or where they are used as the 
 vehicles of certain figurative forms of expression. Finally, we 
 have to give some account of the history of the formation and use 
 of the negative compounds in the various departments of the 
 literature and through the different periods of the language. 
 
The Negative Compounds in Greek. 
 
 II. THE FORM OF THE PREFIX. 
 
 The regular and productive form of the negative prefix in Greek 
 is the so-called a-privative (dX^a ar€pi]TCK6v). The negative 
 prefix VT)- is archaic and poetic and can hardly be said to be pro- 
 ductive in any period of the language. 
 
 Philologians now-a-days are wont to speak of the so-called 
 alpha-privative, and this is because that designation is misleading 
 in both its parts. For in the first place the ante-vocalic form dv- 
 undoubtedly represents more nearly the original form of the prefix 
 than does the ante-consonantal d-, and it is the nasal which is its 
 characteristic element and which is still common to most of its 
 forms in the various languages cognate with Greek ; cf. Lat. iw-, 
 Germ, un-, etc. The origin of the term a/^/ia-privative is doubt- 
 less due partly to the fact that in the great majority of cases the 
 prefix stands before a consonant, and so appears as a- rather than 
 du-, and partly to the accepted view of the ancients and of earlier 
 modern scholars that the v was inserted after the d for the avoid- 
 ance of hiatus. Secondly, the meaning of the prefix is by no 
 means merely privative, i. e. denoting the removal of that which 
 was before possessed or the absence of that which is aimed at or 
 expected, but it is rather negative in the widest sense, running the 
 whole gamut of possible shades of negation. A more proper des- 
 ignation, therefore, would be ar-negative, which, however, in defer- 
 ence to established usage we shall not venture to employ. 
 
 A comparison of the forms in the extant languages of the Indo- 
 European family points unmistakably towards the use of a nasal 
 element as a negative sign in the pro-ethnic speech. We are able 
 to discern also that the early language differentiated the negative 
 of the sentence from the negative which formed a close compound 
 with a noun, an adjective, or an adverb. We can tell too the 
 classes of compounds into which in this early period the latter form 
 of the negative entered. If we attempt to go further back than 
 this, we can no longer speak of scientific knowledge, but must be 
 content to ascribe to our surmises merely the character of a jprioH 
 probability or possibility. 
 
 Perhaps without claiming for the fancy any more than it is 
 
The Negative Compounds in Greek. 7 
 
 worth we may be permitted to indulge in a conjecture as to one of 
 the possible ways in which the use of a nasal element as a negative 
 sign may have originated. We must assume that the growth of 
 language has always been, as we see it now, a process of adaptation 
 of means to end. The end was and is the practical one of com- 
 municating thought to one's fellows; the means anything which 
 by association can effect the desired communication, whether it be 
 some spoken symbol originating in onomatopoeia, or interjection, 
 or gesture, or what not. 
 
 The earliest definite oral expression of negation by the child 
 would naturally often occur with reference to its chief object of 
 interest, its food. Now the refusal of food by one who has not yet 
 mastered articulate speech requires a closed mouth ; so, if this be 
 accompanied, as it is apt to be, by vocal utterance, a nasal is pro- 
 duced, generally the labial or dental nasal. The child's negative 
 is thus a 'vocal gesture,' or an interjection. A similar explanation 
 might account for the origin of shaking the head or throwing out 
 the hands sidewise as signs of negation. Languages widely sepa- 
 rated and outside of the Indo-European family also have what we 
 may call a nasal negative, e. g. in the American languages, Poko- 
 raan, Quichl, Maya, Haytian, Gvajiro, ma; Paez, me; Kechua, 
 mana ; Hidatsa, desa {d being interchangeable with n); cf. Douay, 
 Etudes Etymologiques sur I'Antiquite Americaine, Paris, 1891, p. 
 24. Note too the use of m in the negative verb in Turkish. 
 
 At any rate, whatever knowledge or theory we can hope to have 
 about the form of the negative in the glottogonic period of speech 
 must be reached by some such method, i. e. by studying the modern 
 phenomena which are observed either in the infancy of the indi- 
 vidual or in incipient stages in the development of linguistic pro- 
 cesses. In some such way as we have indicated, or indeed in any 
 one of various other ways, a negative sign might arise, as the act 
 expressing the negation became less and less instinctive and more 
 and more conventional ; and only by degrees would the accom- 
 panying sound become an articulate word with a definite place in 
 the sentence. Cf Paul, Principles of Language (Eng. trans.), p. 
 122, "One might very well imagine that negative sentences 
 might be formed in a primitive stage of development of language 
 
8 The Negative Compounds in Gh-eek. 
 
 in which the negative sense might be indicated by nothing else 
 than the stress and the accompanying gestures." 
 
 But in these matters the only safe attitude is that of an agnostic. 
 Such an attitude we must bear also toward the theory of the origin 
 of the negatives from a demonstrative root. This theory would 
 identify the original form of the negative with the pronominal ana, 
 meaning at first 'yonder/ then 'other/ and gradually acquiring a 
 purely negative force ; cf. Pott, E. F. ^, i, p. 382. 
 
 ne, ne, no, no, n- ("w), n- have been assumed as the forms in 
 
 which the negative appeared in the primitive Indo-European lan- 
 guage. Cf. Fowler, The Negatives of the Indo-European Lan- 
 guages, Chicago, 1896, p. 1. There is wide difference in the 
 character and weight of the evidence for the existence of each of 
 these various forms, but it is at least clear, as has been stated, that 
 even in the pro-ethnic speech there was a differentiation between 
 the negative of the verb or of the sentence and that of the noun. 
 Of these forms, if all were in existence, I-E. ne, ne, no, no belonged 
 to the verb or to the sentence, and n- (*n), h- to the noun (including 
 
 the adjective). The difference between n-, nn- and n- is purely one 
 
 of form, not of meaning ; cf. Kruzewski, Techmer's Zeitschrift, in, 
 p. 185, who says that "prefixes have so definite a meaning that 
 phonetic variations could not be used for any internal distinction 
 in signification." So too, as there is only one kind of negation 
 known to logic, the different forms of the I-E. negative must be 
 thought of as having originated through different accentual rela- 
 tions due to their position in the sentence and not as expressing 
 degrees or varieties of negative force. That the heavier forms 
 should have been used for the negative of the sentence is probably 
 due to the enclisis of the verb. The natural and common view is 
 to regard the negative prefix as a weak ablaut form of the stronger 
 particle. Bopp's identification (Ygl. Gram., § 537) of the a- 
 privative with the verbal augment is now nothing more than one 
 of the curiosities of the history of philology ; cf. Pott, E. F.^, ii, 
 p. 398. 
 
 This weaker form of the negative which appears in the privative 
 prefix must go back to conditions in which the negative was with- 
 out accent. Yet secondary causes operating in pro-ethnic times 
 
The Negative Compounds in Greek, 9 
 
 seemed to have caused the prefix to be accented in primary compo- 
 sition, i. e. in immutata (descriptives, karmadhdraya). In secon- 
 dary composition the mutata (possessives, bahuv7'lhi) seem to have 
 largely lost this accent (on the prefix) of the immutata from which 
 they were derived and to have become oxytone. See Knauer, 
 Ueber die Betonung der Composita mit a-privatum in Sanskrit, 
 K. Z. XXVI, pp. 65 ff. In the case of mutata with stems ending 
 in -es this loss would seem to go back to the pro-ethnic speech ; in 
 the case of the other mutata the loss, where it has occurred, belongs 
 only to the individual languages. Thus it has become a general 
 rule for mutata of all types in Sanskrit, while in Greek it has not 
 gone beyond the stems in -69. See Streitberg, I-G. Forschungen, 
 I, pp. 87 f., 94. 
 
 For Greek it may be stated as a general rule that, with the 
 exception of the stems in -e? and a few minor groups, the accent is 
 generally recessive in compounds with the negative prefix, both 
 immutata and mutata, though we can hardly tell whether this is 
 due to the general recessive law or to the tendency to accent the 
 first members of these compounds. Mutata in -e? (nom. -779) are 
 with few exceptions oxytone. 
 
 This theory, that the oxytonesis of the Greek mutata in -?;9 is a 
 remnant of an old bahuvrihi (mutatum) accent the tendency toward 
 which started before the separation of the languages, is that of 
 Knauer, who holds that this tendency was limited by other tenden- 
 cies in Greek to the mutata in -779, while in Sanskrit it became a 
 rule for mutata in general. But the view has also been held that 
 the peculiar accent of the compounds is due to a tendency to take 
 an accent like that of the simple adjectives in -779 ; cf. Schroeder, 
 K. Z., XXIV, p. 110, and Wheeler, Der Griechische Nominal- 
 accent, p. 46, n. 1, 
 
 While the other languages of the I-E. family have preserved the 
 nasal in their representatives of the I-E. negative prefix n- (*n) as 
 
 in Lat. in-, Germ, un-, Old Ir. an-, the Greek agrees with the 
 Sanskrit and Avestan in having the privative prefix in the form 
 of av- [an-) before vowels and of a- (a-) before consonants. In 
 Greek there are a few apparent exceptions to this rule. So for 
 
 ^ve(f)€\o<i, Od. 6, 45, some codices read avve(f)€Xo<; ; cf. aWo(f)o<;, II. 
 
10 The Negative Compounds in Greek. 
 
 10,258 (v. 1. dXo(})o^). There are a number of other phenomena ii» 
 the Homeric poems which seem to favor the view that these forms 
 may be due largely to metrical necessity in the hexameter. We 
 
 have d- regularly through the language in dddvaro<i, which is 
 doubtless a very old word and by far the most common of the pri- 
 vative compounds in early Greek poetry, constituting more than one- 
 seventh of all the occurrences in Homer and just one-fourth of those 
 
 in Hesiod. So we have d- in dKafiaroi; in dactylic poetry and 
 
 later d7raXafio<i (Hes.) and dTrapafivdc; (Aesch.). The otherwise 
 constant habit of the language rather forbids us to call these cases 
 of compensatory lengthening for the loss of the v of the prefix, but 
 it is more likely that we have here to do with a phenomenon par- 
 allel with '^ve/jioei'i from ave/xo^;, dvcovv/juo^ from 6vo/xa, the result 
 of a tendency toward the lengthening of short vowels of which the 
 epic poet availed himself under stress of metrical necessity, just as 
 he used always d'KToXefjbO^ for the unmetrical d7r6\efio<;. For the 
 rare lengthening of a-privative in Sanskrit, see Whitney, Sanskrit 
 Grammar, 1121 c. 
 
 The word 't]K€(Tro<; [ = 'untouched by the goad' (?)] is found 
 only in the Homeric line, Tjva r)Kecna<; iepevaeixev, at k e\.er]crri{i)y 
 II. 6, 94, 275, 309. If we have really to do with the negative 
 prefix here, it may be said that the phrase, iqvt^ r)Ke<TTa^ was prob- 
 ably a crystallized expression and that the metre requires the word 
 
 in question to be scanned ; ?; for d may be dialectic, the 
 
 consciousness of the relation to a-privative being lost, and the 
 alliteration may have had some influence. But a-privative 
 appears nowhere else in Homer as t; and the signification of the 
 word as given in the lexica rests apparently on the ancient expla- 
 nation of it as = dK6VTr}To<i. For a discussion of the word see 
 Froehde, Bezz. Beitr. yii, 328. 
 
 Uncertain too is the presence of the negative prefix in d/jb(})acrlT} 
 in II. 17, 695, Od. 4, 704, where there is, however, some manu- 
 script tradition supporting d(f)aaLr). Pott, E. F.^, p. 390 suggested 
 a confusion in d/j.(f)acrirj between the negative prefix and dvd or 
 dfji(t>L Froehde, Bezz. Beitr. xx, p. 212, thinks the analogy of 
 a/jb^poTo<; by the side of ^poro^ may have influenced dfx^aa-LT},. 
 while Brugmann, Vgl. Gram.^ I, p. 419, refers this form of the 
 
The Negative Compounds in Greek. 11 
 
 prefix to I-E. n. I am indebted to Prof. Dr. Hermann Collitz for 
 
 
 
 the suggestion that the proper division of the word may be afKJy-aa-iT) 
 and that we may here have to do with the root as, asd, to parch or 
 wither, and so the word may be cognate with Gk. a^a, a^w, dtalvco, 
 a^aKeo<i, Lat. arere, ardere, ardor (for asdor), Eng. ashes. Cf. 
 Fick, Vergleichendes Worterbuch der I-G. Sprachen,^ ii, p. 28. 
 
 With regard to the appearance of the form a- before vowels it 
 may be remarked that words like dvirvo^ (Hom.) and aoLKo^ (Hes.) 
 were formed in a period when initial a and F still existed in the 
 simple words ; in some cases the same compounds took av- later 
 when the traces of the original initial consonant had entirely dis- 
 appeared, e. g. dvocKo<; (Hdt.), dwirvo'^ ^y^-) ) but ciovro^ and 
 avovTaTo<i are both Homeric. dveaTio^ and dvetfjiwv are found in 
 Homer, but perhaps only in the later parts ; the latter word is 
 found only in the Odyssey, ddara and aacr^ero? are of doubtful 
 explanation. The true exceptions to the rule, such as dopvo<i, 
 do(Tfio<i, mostly with the prefix before a, w, are doubtless due to 
 the analogy of the earlier and authorized cases ; but most of these 
 have also forms with dv-. The differentiation of form before 
 vowels and consonants is perhaps Indo-European, as the agreement 
 of Sanskrit and Greek would seem to show, although in other 
 languages we find the same form for both cases, as in Lat. in-, 
 Germ. un-. 
 
 The proof for the existence of n as an I-E. negative prefix is 
 
 not nearly so abundant as is that for the existence of the shorter 
 forms. Schulze, K. Z. xxvii, p. 606, accepts the equation : 
 Greek vd-, vrj- = I-E. n, and correlates these with Italic an-] in 
 
 o 
 
 fact he gives the proportion, Lat, in- : Ital. an- = Gk. dp- : vt]- ; 
 while Kretschmer, K. Z. xxxi, pp. 405, 412, puts Greek vd-, 
 VT]- = I-E. *nd- and denies the necessity of positing a longnasalis 
 sonans for the parent speech. Persson, I-G. Forsch. II, p. 228, 
 puts Greek vd-, vrj- in relation with Lat. ne. 
 
 The matter is further obscured by formal difficulties, vt]- actu- 
 ally does appear as a negative prefix in epic, as in V7]'i<;, vrjKepSi]';, 
 v7)'Trevdr]<; and vt^Kepo<i. With regard to this prefix vrj- let us con- 
 sider the following possibilities : (1) that vr}- may represent n used 
 
 as a negative prefix in I-E. times, (2) that it represents the I-EJ/ 
 
12 The Negative Compounds in Gfreek. 
 
 sentence negative ne or ne, and (3) that it arose on Greek ground 
 but is to be derived from the I-E. prefix n (*n). 
 
 (1). Certainly there is something to be said in favor of the cor- 
 relation of Greek vd-, vr]- with an I-E. n. For those who assume 
 
 o 
 
 n as an I-E. sound find its regular representative in Skr. a and 
 
 Gk. vd; thus the aforementioned rare lengthening of a- privative 
 in Sanskrit would be parallel to Greek vd-, vrj-. In the second 
 place I-E. n is assumed to result frequently from the reduction of 
 
 a dissyllabic root. Thus from I-E. * i/gene we have the propor- 
 tion, I-E. * genetor : * gn-tos = Skr. janilar : jdtds = Lat. genitor : 
 
 o 
 
 gndtus = Gh. yeveTcop (second e irregular) : yvrjTOfi in Ka<TLyv7)To<s. 
 This suggests a possible correlation of vrj- with the troublesome 
 ava- of dvdeSvofi (Hom.), ameXTrro? (Hes.), and the less certain 
 dvdyva>(TTO<i (Callim.), dvaTrvevarci (Hes.). Cf. 6dvaT0<i, 6vr)To<;, 
 etc. We should expect, however, iva rather than dva, and it is 
 more convincing to explain these forms of the prefix otherwise. 
 Cf. p. 14, where the existence of dva- as a simple form of the 
 negative prefix is shown to be improbable and this taken in con- 
 nection with the doubtful character of n as an I-E. sound renders 
 
 o 
 
 very unsatisfying the assumption of any such origin for the 
 prefix vr)-. 
 
 (2). From what is said hereafter showing the frequent use 
 of the Greek sentence negative as a substitute for the negative 
 prefix, it may not be thought improbable that the Greek vrj- may 
 represent the I-E. sentence negative ne or even ne -\- a following 
 initial vowel. Cf. Hartung, Gr. Part, ii, pp. 89 if. n^ in Lithu- 
 anian and ne in Old Bulgarian are used as negative prefixes, and 
 see Leo Meyer, Vgl. Gram.^, pp. 119, 594. This theory, that 
 vr]- is a survival of the I-E. sentence negative, ne, ne, is somewhat 
 favored by the very restricted scope of this prefix, for ne, ne were 
 otherwise practically lost in Greek. One may cite v^ as used in 
 oaths and vai, perhaps originally negative., veicrap seems to be a 
 Semitic loan-word and i^eTroSe? and ve/3p6<i have a different ety- 
 mology, cf. Fowler, op. cit. p. 11. The use of vr}- as a prefix is 
 confined to a few poetic words nearly all archaic or evident archaistic 
 imitations. Only about 50 compounds and derivative words appar- 
 
The Negative Compounds in Greek. 13 
 
 ently having this prefix are quotable ; of these only one appears 
 first in Attic prose, and this is vrjirotvei,, Andoc, after Homeric 
 V7]7rocvo<;. But aside from these words tliere is no example of the 
 survival of this I-E. sentence negative in Greek as a prefix or 
 otherwise, and to explain vco86<;, v(oSvvo<;, v(ovvfj,o<; on this basis 
 we should have to assume an I-E. no to have survived in Greek ; 
 which is very doubtful in any case. 
 
 (3). In vrjcrrtf (j/iS-), vi]V€/j,o<;, vr)fiepTi]<i, vri\er}<i, vrjKOvaTO<i, 
 vrjKe(TTo<;, vrjpL6fjLo<;, VT^yperci, and also v(ovvfxo<i the prefix 1/77-, if 
 such was its early form, has not been kept pure. Further, most 
 of these forms have by their side forms like avrja-rc^, dv7]vefx,o<i, 
 av7)\er}<i, dv7]K0vaT0<;, dvr}Kecno<i, dvi]pid/jbO<;, dvcovv/jbo^;. Clemm's 
 suggestion, Curt. Stud, viii, 14, that we have in these last forms 
 an instrumental dvrj-, from which vrj- results by aphaeresis, would 
 hardly now be regarded as tenable, but his alternative suggestion, 
 that " they have undergone the same prolongation which is wont 
 for metrical reasons to be found in other words," has more of 
 plausibility. There are other Homeric forms like dv7]fi€\KTo<;, 
 dvrjvvcno^, dvriporo<i, dvrjvopa, which have no forms in vrj- by 
 their side, but the existence of so many doublets like dvrjvefxof;, 
 vrjveixo<i, make it natural to infer an identity of origin for dv- and v-. 
 The words in question are all very old words and it would appear 
 that in one stage of the language there was a tendency to represent 
 the I-E. prefix before long vowels by v-. It was natural that 
 later this should succumb before the far more common ante- 
 vocalic dv-. 
 
 From words like v7]V€/j,o<i, i^T^yu-e/jr?;?, etc., it would be easy to 
 account for the development of a suffix vtj- by 'clipping;' whence 
 vryt9, vr^Kephrjq, v't]'rr6vdi]<;, etc. This seems to us more convincing 
 than Pott's idea that a form like dvT]\€7]<i might be due to a con- 
 fusion of dv- and vr)-, and so be a sort of syncretism of dve\€r]<i 
 and v7]Xei]'i. 
 
 The attempts at explaining what appears to be a fix dva- in the 
 words dvdeSvoi;, dvaekirro'^, dvdyvcoaro';, dvd7rv€vaT0<i, already 
 cited p. 12, have been many and various. The last two of these 
 words, being late, have been explained as analogical formations ; 
 cf. Froehde. Bezz. Beitr. xx, p. 212. Schneider, however, 
 changes dvdyvcoa-rov with some probability to dv dyvtaarov, where 
 
14 The Negative Compounds in Greek. 
 
 it occurs iu Callimachus, fr. 422, and avaTrveva-Ta iu Hes. Th, 
 797 was changed by Hartuug, Gr. Part, ii, p. 75, to ap' atrvevcrrotf 
 (cf. Od. 5, 456) and by Hermann, Opusc. vi, p. 164, to a^i 
 aTTvevara, Job. Schmidt, K. Z. xxiii, p. 271 ff., would 
 explain the second vowel in the first two of the forms mentioned 
 above {dvdeSvo<;, dvde\7rro<;) as developed from the lost F (cf. Har- 
 tung, 1. c), but in the other forms as due to the following double 
 consonant in the effort to avoid three successive consonants. Thus 
 the second a would be only a svarabhahti or parasitic vowel. Thi& 
 explanation, however, would assume the existence of a form dv- 
 before consonants, otherwise supported only by the Homeric 
 dfj,(l)acr[7), which has been differently explained above. 
 
 The occurrence of a dissyllabic form of the prefix has been 
 asserted too for other languages besides Greek. Pischel, Bezz. 
 Beitr. iii, p. 245, remarks on the extended use of ana- on Indian 
 ground, and on the strength of Indian ana-, Old Bactrian ana-, 
 Celtic ana-, Greek dva-, Old High Germ., una-, he assumes ana- 
 as the I-E. basic form, as already Buttmann, Gram, ii, § 120^ 
 anm. 1, and Olafsky (see Neue Jahrb. f. CI. Phil. u. Paed. Lxxiv, 
 p. 581). This assumed basic form has been identified with the 
 pronominal stem, ana, Pott. E. F.^, i, p. 384, or again with the 
 prepositional ana = dvd; Johansson, Bezz. Beitr. xv, p. 310. 
 Pott would explain the prefix an- and the adverb na as arising 
 from ana under different accentual conditions, thus, dn{a)-, {a)nd-. 
 
 But it seems needless to argue for ana- as the original I-E. 
 basic form of the prefix, for at any rate this cannot have been the 
 only form just previous to the separation of the individual lan- 
 guages. Otherwise it is difficult to explain, first, the form a-, 
 common to Aryan and Greek, and secondly, the very limited 
 scope in Greek of the prefix dva-, if such there be. It is likely 
 that the dissyllabic forms of the prefix arose during the separate 
 development of the languages in which they occur. For the 
 independent origin of ana- in Celtic before consonants see Zimmer, 
 K. Z. XXIV, p. 524. 
 
 It seems probable then that the forms with initial dva- are a 
 secondary development within the Greek language. It is not 
 necessary that the same explanation should hold for all cases. In 
 some instances the apparent prefix dva- may be due to 'clipping'; 
 
VM'.v 
 
 The Negative Compounds in Greek. 15 
 
 thus Froehde, Bezz. Beitr. xx, p. 213, assumes ai^aeSz/o? to be 
 for ay + * deSvo'i. Now we have in Homer the simple word in 
 two forms, eSva, eeSva, and Hesychius glosses clehvov by a(f)€pvov 
 rj TToXvc^epvov. If the ancient lexicographer really understood the 
 word, the second a in avdehvo<i was either negative or intensive. 
 The suggestion of Wharton, Trans. Philol. Soc, 1891-4, p. 331, 
 that dv- in these words = n- and is intensive, so that dvdehvo<i 
 
 o 
 
 means ' quite dowerless,' would presuppose a double prefix 
 here n-n-, of which the first part would be intensive and the 
 
 
 
 other negative. 
 
 The view of Pott, E. F.^ i, p. 389, that we have in dvdehvo^ 
 a doubling of the negative prefix seems more probable. Cf. 
 Bvo-dfi/jLopo<;, II. XXII, 428, where the scholiast remarks, Sva-dfx,- 
 lxopo<i, hehiTrkaalaKe irpo'i rrjv eirirao'iv • to rydp ova Kai a 
 ravrov SrjXova-iv ; also Et. Mag. sub voc, Svad/jifiopo<i, co? r/A-t/ce?, 
 6fi'^\cKe<;, Kai avvofjir^XtKe^. Cf. Skr. dur-a-dabhna, anavipra- 
 yukta (explained by na viprayukta), and the more doubtful ana- 
 vadya, and see Whitney, Gram. § 1121, b. We may add that a 
 confusion of dveeSvo^i and de8vo<; (* dFeSvo<i) both negative com- 
 pounds, may have led to a sort of blend in dvdeSvof;. 
 
 Froehde, 1. c, would connect dvd7rv€vaTo<; with dvairvea), being 
 shortened by Miaplology' from * dv-avdirvevarof;, as diroiva from 
 * diro-iroLva ; then again dvd'yvaxTTO'^ by analogy. But it is in 
 favor of the assumption of a doubling of the negative prefix that 
 dryvQ)aTo<; and dirvevaro^ are words in good standing in the 
 literature. 
 
 III. FORM AND CLASSIFICATION OF THE 
 COMPOUNDS. 
 
 The great productiveness of these negative formations and the 
 inseparable character of the first part might raise the question 
 whether these words are really to be considered as true compounds 
 and whether the first element is not to be looked upon as a mere 
 prefix, one of the few which the I-E. language possessed. Pre- 
 fixes have, however, in general a very definite meaning and func- 
 tion. They are not, like the suffixes, so liable to be reduced to 
 
16 The Negative Compounds in Greek, 
 
 the level of mere formative elements. Hence, there is no great 
 inappropriateness in the common designation of words, other than 
 derivatives, which contain these prefixes as compounds. 
 
 It has been found convenient for our purpose to classify as fol- 
 lows the Greek words containing the negative prefix : 
 
 I. Immutata (determinatives). 
 
 a. a-priv. + ordinary adj., as a^dp^apoq. 
 
 b. d-priv. + vbl. in -to<;, as a^uTd, d^Laaro^;. 
 G. a-priv. -j- other pples., as deKoov, dvdpp,evo<i. 
 
 d. a-priv. + noun, as dhutTrj^, dveLKaiorri<^. 
 II. So-called synthetica. 
 
 e. a-priv. -f- verbal root, as drpLylr, d<^avr}<;. 
 
 III. Mutata (so-called possessives). 
 
 /. a-priv. + noun, with or without new adjectivising 
 suffix, as d6vao<;, dvai^wv, d6€fi[aTto<i. 
 
 IV. Derivatives. 
 
 g. Negative compound + adjective, noun, or verbal root, 
 
 as d8afiavT07reSi\.o<;, dSiKOTrpayrjii. 
 h. Nouns, as depyia, mostly abstract. 
 i. Adjectives, as depyrjXo^. 
 k. Adverbs (except those regularly formed in -oo<i), as 
 
 deKrjTi. 
 I. Verbs, as dBvvaTeo). 
 
 For the basis of this classification see L. Schroeder, Ueber die 
 Formelle Unterscheidung der Redetheile, pp. x, 203, 287. All 
 the true negative compounds belong to class II in the classification 
 of Greek compounds given by Brugmann, Gr. Gram.^ §§ 153, 159. 
 
 In practice it is often impossible to discriminate between the 
 compounds of groups e and/, and occasionally of group a. Take 
 for example dl3\a/37j<; ; one may doubt whether we have here a 
 mutation from a noun /3\d^7] or ^\d^o<;, or a syntheticon from 
 the verbal root appearing in ^XdirTco. In many cases neither the 
 meaning nor the presence or absence of a suitable noun which 
 may have entered into composition is decisive. In many of the 
 early examples the formation is quite obscure. Often in late 
 Greek, as in other phenomena of the period, the formations seem 
 to be made on the analogy of established types, so that it is no 
 
The Negative Compounds in Greek. 17 
 
 longer possible to define precisely the separate elements of the 
 compound. The ancients were more inclined than later and 
 modern grammarians to look upon the second members of com- 
 pounds as verbal ; see Zacher, Breslauer Philologische Abhand- 
 lungen, i, pp. 10 ff., 48-61. So the Homeric lexicon of Apollonius 
 explains aSe^;? by ov SeSio)<;, while the Et. Mag. gives : d8ei<; 
 d(f)o^ov (iTTo Tov Seo9. Cf. too the frequent pairs of doublets like 
 d<f>0ovo<;, cKfidovrjTCi, et. al. 
 
 Outside of the dithyrambic and comic poets the moderation 
 of the Greek is shown in the general limitation of the compounds 
 to two parts. But while the compounds are generally bi-membral, 
 occasionally one of the members is itself a compound ; so as early 
 as Homer we find dvnrro-'jroSe';, d-'TrpoTi^iaara, then dv-eTrl-^eaTa 
 (Hes.), diSpo-BiKr)^, dSa/xavTO-'jTeScko'i (Find.), dKa/j,avT0-p6a<; 
 (Bacchyl.), etc. 
 
 IV. LIMITATIONS ON THE USE OF THE 
 PREFIX IN COMPOSITION. 
 
 The true sphere of the negative prefix is its combination with 
 nouns, adjectives and verbal stems to form adjective compounds. 
 The nouns and verbs with a-privative are in general not com- 
 pounds at all, but are derivatives from compound adjectives. 
 The prefix can never properly be combined with a verb to form 
 a negative verb, nor with a pronoun or pronominal word to form 
 an indefinite pronoun. In this respect the pro-ethnic speech 
 seems to have kept quite distinct the prefix and the independent 
 adverbs ne, ne, etc. The former was used only before nouns, 
 including adjectives, participles and infinitives, the latter with 
 verbs and also with pronouns to form the negative indefinites. 
 The negative adverbs never became productively used as prefixes 
 except in the Balto-Slavic, where for example Lithuanian n^ 
 (I-E. ne) quite usurped the place of the original negative prefix. 
 But it is to be observed that n^ in Lithuanian, having become a 
 true negative prefix, is not used to form negative indefinite pro- 
 nouns and adverbs. So in Greek, which has lost I-E. ne as a 
 negative adverb, the independent adverbs ov and /a?; are used to 
 
18 The Negative Compounds in Greek. 
 
 form indefinites, as ovtl^, /jbi]Ti^, oviroTe, ovS€t<; and the rest. 
 ovTa probably took the place of an I-E. * ne-qi-s when ov 
 had taken the place of I-E. ne (Brugmann, op, cit. Eng. trans. 
 II, pt. I, § 31). In a negative sentence containing an indefinite 
 pronoun or adverb it is immaterial, so far as the sense is concerned, 
 whether the negative is taken with the indefinite or the verb. 
 Thus the negative adverb, i. e. the sentence negative, and not 
 the negative prefix, came to be used in this combination, which 
 got to be felt and used as a compound, favored no doubt by 
 the fact that the accentless indefinite was attracted by the strongly 
 accented negative. Cf. Delbriick, Vgl. Synt. ii, p. 524. 
 
 Rare and decidedly irregular are the instances in Sanskrit of 
 the use of the negative prefix with demonstratives, as in a-sas 
 and an-esas. 
 
 The infinitives and participles being nominal in their origin 
 must at first have taken the negative prefix and not the negative 
 adverb ; cf. in Greek forms like aeKwv, aeKa^ofxevo^, aek7rreovTe<i, 
 a(f)poveovTe<;, avdpfi€vo<;, ari^cov, avofjLoXoyovfievo^;, aSdfia<;. Knauer, 
 K. Z, xxvii, p. 19, cites numerous examples of the composi- 
 tion of the Sanskrit participles belonging to the tense systems 
 with a(n). In the Rigveda na is not found with participles and 
 perhaps not in Sanskrit prose (Delbriick, Vgl. Synt. ii, p. 529). 
 There are a goodly number of examples of combinations of the 
 present participle in Gothic with un- and in Latin with in-. But 
 in general the composition with the negative prefix declines when 
 the participle, as in Greek, becomes more closely attached to the 
 verb and assumes more of verbal and less of adjectival character. 
 The participles ordinarily included in the verbal paradigm are 
 rarely, indeed only exceptionally, thus combined ; the forms 
 mentioned above, deKwv, etc., have all more or less of the 
 adjectival character which seems to have belonged of old to 
 the participle and is especially seen where the participle enters 
 into composition ; cf. proper names like FjvpvfieScdv, F^veXOfov, 
 Tlo\v(T7Tep')(a>V' By their side are found ovk eOeXwv et al.; dwrre- 
 (TTaXfjievo)'; is late. It is not surprising then to observe that 
 even as early as Homer the use of the negative adverb ov with 
 the participle is already well established (Monro, H. G., § 360), 
 although it is certainly not so freely combined with the participle 
 
The Negative Compounds in Greek. 19 
 
 in Homer as in later Greek. The second parts of the compounds 
 ABdij,a<i, aKdfxa<i, though no longer independently existing, have 
 lost their participial character and are practically adjectives ; so 
 ata-Twp, aKpdroop. Only with the participle in -to<} is the com- 
 bination with the negative prefix at all productive, but the verbals 
 in -To<; are not generally formed on a special tense-stem, are not 
 reckoned as a part of the verbal paradigm and are often merely 
 adjectives expressing fitness or capacity. They seem even in 
 Indo-European times to have been combined freely with the 
 negative prefix, as they are in all the separate languages. 
 
 Sanskrit, Latin, German and particularly English offer many 
 more examples than Latin or Greek of the combination of the 
 negative prefix with the present particijile, but in Latin the 
 ■examples are far less rare than in Greek and it is in Greek that 
 the participle has come most to be used as a substitute for a 
 subordinate finite verb. Cf. Sk. avidvdn, Lat. insciens, Germ. 
 unwissend, Eng. unknowing, but Gr. ovk elSox;. 
 
 The infinitive in Greek has become too distinctly verbal in its 
 nature to take the negative prefix ; even in Sanskrit it seems .to 
 take it very rarely except when used with the verb gak, Speijer, 
 Sanskrit Synt. § 404. On the other hand at an earlier period, 
 when the infinitive was more distinctly nominal in its nature, 
 the aversion of the language to combining the negative prefix 
 directly with a noun would render d-Svvao-dai quite as awkward 
 as d-Svvafji,i<;. 
 
 The negation of the finite verb in Greek by an a-privative is 
 to be looked upon as an anomaly. Professor Bloomfield, J, H. U. 
 Circ, 1882, p. 175, in objecting to the etymology of d/i^XaKelv 
 given by Curtius (Grundziige, 5, p. 463) and adopted by Vani9ek, 
 says, " The composition of a-privative with a simple aorist stem 
 would be an anomaly as great as a compound * d-ireTrovOa, ' I have 
 not suffered.' " When we have driei in Theogn. 621, 
 
 Tra? TC? ifkovatov dvSpa r/et, driei Be irevi'^^^pov, 
 
 the compound is evidently an artistic })roduct formed for the sake 
 of antithesis to riet preceding. Brugmann, Griech. Gram.^ § 590, 
 treats drUi as an extension of paradigm from uTiro^ (Hom.). 
 Parallel with this is a passage in Plut. Mor. 885 A, ware rd fxev 
 
20 The Negative Compounds in Greek. 
 
 eifidpOat TO, S' dveifidpOat,, where Usener with MS. D reads ra 
 fiev elfxapra ra 8' dvetfiapTa. So the SaDskrit dj^acasi is a purely 
 artificial form. The Homeric drl^cov, II. 20, 166, occurring as it 
 does in the participle, if it is really a negative compound, does not 
 count here on account of the original nominal character of the 
 participle, and dri^oiv is probably to be classed with deKu^ofMeva, 
 deKcov and dvofio\o<yovfji€vo<; (not a participle from a verb dvofio- 
 Xoyovfiai,). Later such a form might be expanded into a verbal 
 paradigm, and that the Homeric dTt^cov was so expanded by the 
 poets, as the tragedians and Apollonius Rhodius, does not excite 
 surprise. It is not found in prose until Galen, who uses dTL^6fievo<;, 
 and Greg. Naz., who uses the finite verb. Cf. Lat. indecet and in- 
 juro, probably extensions from indecens and injuratus, and ignosco 
 considered by V. Henry a re-formation from ignotus. So the 
 inseparable Sva- in hvcrdvrjaKwv, Eur. El. 843, Rhes. 791, 
 BvaOavovra, schol. to Luc. Icarom. 29, is somewhat legitimized 
 by being combined only with the participle. It may be, how- 
 ever, that art^ct) is a denominative verb, see Froehde, Bezz. 
 Beitr. xx, p. 221. On drifxaco, see Brugmann, Griech. Gram.^, 
 p. 529. 
 
 Before vowels there is the possibility that the verb is a com- 
 pound of the preposition dvd, which sometimes denotes a reversal 
 of the action of the verb and gives a negative sense. Thus 
 dvapdo/xai means ' to recall a curse ' and dvev^ofiai ' to unsay a 
 prayer.' Cf. Lat. resecrare, Germ, widersprechen. On the derivation 
 of dvaivofjuai from dva -f * aivofjiac (cf. alva, a saying) see Osthoff, 
 Bezz, Beitr. 24, 199 ff., esp. p. 205. dvev^oiiai, so far as we 
 know, occurs only in the participle, Plat. Alcib. ii, 142 D, 148 B, 
 and the articular infinitive, Poll, v, 130. The poet in Plat. 
 Alcib. II, 143 A, uses dvevKTOi<; as the negative of ev^ofievoi,^, 
 while the author of the dialogue himself says, 142 d, ev^acvTo av 
 
 oXtyov Se €7na^ovTe<; iviore iraXivwhovcr lv , dvev^o/juevoc 
 
 arr dv to TrpcoTov ev^covrai. 
 
 dv^So/xai in Hermippus, frag, 81 M, 77 K, was thought by 
 Shill'eto, Jour, of Philol. vii, 159, to be a compound of dvd, but it 
 seems not unlikely that it is an artificial compound made with the 
 privative prefix for the sake of antithesis to the preceding rjcrOrjv, 
 and so parallel to the examples already cited from Theognis and 
 
The Negative Compounds in Greek, 21 
 
 Plutarch. The fragment is : a Toff" rjaOrjv, ravra vvv dv^Sofiac 
 Comedy is just the place for such a formation, but it is not impos- 
 sible here that we may have a blending of both the preposition 
 dvd and the negative in these formations. 
 
 dvofjuoiow is probably a derivative from dvofioco^;, and dvav- 
 Spoofiac from dvav8po<;. 
 
 It may be found that some English verbs like ' undo/ ' unfasten/ 
 and the like, may have been first formed for the sake of opposition 
 to the single verbs, as in the Greek examples cited above. An 
 adherescent negative adverb might have been felt not to make a 
 close enough compound and again not to give sufficient opposition, 
 but only the negation of the positive idea. It is interesting to 
 note, however, that the use of un- with verbs in English seems to 
 have come in through confusion with the Anglo-Saxon inseparable 
 prefix ond, and, on, which is often = re-, denoting the reversal of 
 an action and quite indistinguishable from the negative in meaning. 
 Cf. Maetzner, Euglische Grammatik, i, p. 540. 
 
 In the case of verbal adjectives with initial dv- followed by a 
 vowel, as dvd\(OTo<i, dve^evpero^i, some ambiguity might arise 
 as to whether the verbal was derived from a simple verb and then 
 compounded with the negative prefix, or whether it was formed on 
 a verb already combined with the preposition dvd. Funck in 
 Curt. Stud., X, 41, has treated this subject and shown that in 
 Greek the external ambiguity of the words was hardly felt, for the 
 ' language never employed the same word at the same period in 
 both the different senses. 
 
 For the ambiguity in Latin between the prefix in- and the 
 preposition in-, as in invisus inauratus, see F. Vogel, Archiv. f. 
 Lat. Lex. u. Gram., iv, p. 321. Cf. inclinatus, bent, Juv. 15, 63, 
 but indeclinatus, unswerving, Ov. Epp. ex Ponto, 4, 10, 83, Yet 
 when Wolfflin, ibid., p. 400, says that in-privatum does not appear 
 before verbs on account of liability to confusion with the preposi- 
 tion in, he hardly gives the fundamental reason, for the absence of 
 ■composition of verb with the privative syllable was most certainly 
 Indo-European. 
 
 The negative adverb may sometimes form a quasi compound 
 
 with the verb making the combination so close that the, negative 
 
 is rarely displaced from its position immediately befcrfe the verl> 
 2 ( UNP'- 
 
22 The Negative Compounds in Greek. 
 
 and that the negative ov is retained even in cases where fxrj would 
 be required by ordinary usage. In these cases the particle usually 
 does something more than merely negative the meaning of the 
 simple verb, and the combination expresses an opposite rather 
 than a negative solely. Examples are: ov (J^tj/lll, '1 deny/ 
 ovK io), ' I forbid/ ovk ideXco, ' I am unwilling/ ovk eiraivoy, ' I 
 disapprove.' Cf. Lat. nescio — from which probably nescius — 
 nequeo, neclego. In ovk akiyo) we probably see more nearly a 
 merely negative combination. 
 
 In most of these cases the verb is one of saying or thinking and 
 often one which wonld naturally be followed by the infinitive, so 
 that the adheresceuce of the negative to the verb seems in part at 
 least due to the reluctance of the early language to combine the 
 negative adverb with the infinitive. 
 
 The necessity for forming such quasi compounds of the verb,, 
 which might at first thought seem to be far-reaching, was largely 
 relieved by the capacity of the language for forming denomina- 
 tive verbs from negative adjective compounds, as dyjrevBetv from 
 a-\|reu8779, and to some extent also by the fact that several of the 
 prepositions practically reversed the meaning of the verb with 
 which they were combined. Cf. for such a use of dvd, dvapdofxac, 
 etc., already cited. This reversing force of dvd is very natural 
 with a verb which denotes downward motion, e. g., epeiirw, throw 
 down, dv7)peL-\lrdfi7}v, snatched up ; /jlvco, close (lower) the eyelids, 
 dva/xvco, open the eyes ; see Lobeck, Rhematikon, p. 43. For dira 
 cf. direcrO'ieiv = fxr] iaBleiv, Theopomp. Com. Mein., ii, p. 813, fr. 
 62 K. Somewhat similar are diroa-LTeiv, diroKriheiv, diroirapOe- 
 vevea-dai which Meineke cites ; also dTroKaXvirrfo, uncover,, 
 d-TravBdo), forbid, and cf. especially dcfjavSavco, Soph. Ant. 501 . 
 With Btd we have Bia^evyvvfjiat, to be disjoined, cf. Halsey, Proc. 
 Am. Philol. Assn., 1888, p. xxiv. 
 
 The statement already made, that the sphere of direct combina- 
 tion of the privative syllable in Greek is its union with noun,, 
 adjective and verbal stem to form adjectives, contains the important 
 limitation that the prefix is only irregularly and exceptionally com- 
 bined with nouns to form nouns. This limitation holds also in 
 general for Latin of the best period, but not for Sanskrit, Englisb 
 or German. The privative syllable will, as a rule, be found before 
 
The Negative Compounds in Greek. 23 
 
 nouns in Greek only in mutata ; in irnmutata it can, generally 
 speaking, be combined only with the adjective. We find in Lid- 
 dell and Scott only 16 probable irnmutata with nouns. 
 
 Such immutata of a-privative -(- noun as do occur are plainly 
 exceptional and irregular artificial formations, mostly poetic or 
 late. Hesiod has dScorrj^, Op. 355, and dj3ovT7]<;, ibid., 451 ; vv. 
 354-5 are : 
 
 Kal ho[xev o? Kev Sm, koX fxrj Sofiev o? k€v jxrj 8c3. 
 hoiTr] fiev Tt<f eSojKev, dScoTj] S' ovrcs eScoKC. 
 
 Here Scorrj^; and dS(OTi]<; are both evidently formed for the 
 antithesis and so are hardly amenable to the ordinary laws 
 of formation. In Op. 451, KpaSiTjv S' eSuK dvSpo'i d^ovTeoo, 
 d^ovT7]<; is plainly the negative of ySour?/?, ' herdsman ; ' the 
 latter, however, does not occur in Hesiod, but is found in 
 Aeschylus and later. To one who had already just said dScoTTjf; 
 in antithesis to Scott;?, it would not seem so bold to say d^ovT7]<i 
 even if no form /3ovT7]<i was actually present, though ^ovrrj^i was 
 probably then in use. But d/3ovTr]<; here may have been felt as an 
 adjective ; cf. Eur. Hipp. 537 (lyric), /Sovrav (})6vov. Lobeck, 
 Path. Gr. Serm., i, p. 213, makes it= ^d/3ovq lengthened by addi- 
 tion of suffix -TT}-. 
 
 In Arist. Phys., i, 8, 3, and perhaps in Hippocrates is found 
 dviaTpa = a non-physician, quack, in opposition to laTpo'i,^ and 
 in Plut. Lye, x, a7r\ovTo<;, poverty, in opposition to ttXoOto"?. 
 
 Pollux, III, 58, censure three words used by Theopompus : 
 Tra/jbTTOi'Tjpot, Se ol @eo7ro/x7rov roy auyjpa(})e(o^ dTToXlrcu, Kal 
 d(f)6Taipot, Kal diraOrivaloi. But as the latter two words are 
 formed with the preposition diro, this may have been the feeling 
 also in the case of diroXlrai, without the necessity of assuming a 
 haplology for ^dTro-TroXcTai. 
 
 Priscian, Keil, ill, 211, uses two grammatical terms, av/x^afia 
 and davfi^a/jia, which he ascribes to the Stoics and which he 
 renders by covgruitas and incongruitas. 
 
 Other examples are : dhcKaiap'xp'; Cic, formed for the sake of a 
 
 1 But 6 1X7) larpSs, Plat. Gorg. 459 B ; cf. Sudhaus, Philodemiis, li, 6, rwy 
 fjLi) l)y]r6p())V. 
 
24 The Negative Compounds in Greek. 
 
 pun, ava^ia Zeno ap. Diog. L., a'n-eL6ap')(^ia Antiphon ap. An. 
 Bek. 78, a7r\7]po(j)opLa Byz., airpoahoKla Def. Plat., appoia and 
 axp^t-a Hippocr., and probably the Attic aa-rparela. 
 
 In Hippocrates, 379, 17 (Foes.), iii, 496 (Erm.), the authority 
 of the MSS. is for rf avapio-Trjaei {dv-apia-Trjcri^), but Ermerins 
 now reads tt} dvapio-Tij). 
 
 There is another group of compounds, sometimes referred to 
 this class, which are artificial formations and are distinctly poetic 
 and figurative in their use. This group comprises the compounds 
 which appear in the not infrequent poetic expressions of the type, 
 'I/jo? "Ai'/oo?, Od. 18, 73, TratSe? ctTratSe^, Aesch. Eum. 1034 et al. 
 These formations seem undoubtedly to belong to the mutata and 
 will be best discussed under the head of style. 
 
 W5efflin, Archiv, iv, 400 ff., has taken up the subject of 
 'Substantiva mit in-privatum' in Latin. He finds that such 
 formations are almost entirely absent from Latin of the Golden 
 and Silver Ages. From archaic Latin are cited only intemperies, 
 insoiietas, ingratiis, iniussu and irreligio. The later more pro- 
 ductive employment of such compounds, e. g., involuntas et al., 
 is confined almost exclusively to ecclesiastical and late Latin, is 
 characteristic of the Africitas beginning with Tertullian, and 
 arises in part from the license of vulgar speech and in part from 
 the exigencies of translation from the Greek. 
 
 In English a few words of the type, unbelief, undress, unrest, 
 are in use, but they are not to be coined at will, while in German 
 the type is quite productive, nouns like Unmuth, Ungeduld, being 
 rather numerous. 
 
 Even in Sanskrit, where compounds like adeva, non-god, akirti, 
 non-fame, disgrace, are very common, the mutata of a-privative + 
 noun seem relatively more frequent. In the immutata of this 
 type most often the second member seems to be an abstract noun. 
 Cf. Delbriick, Vgl. Synt., ii, p. 530. 
 
 Outside of the strictly bi-membral divisions which characterize 
 philosophical thought and which give rise to expressions like the 
 Aristotelian to ovk dv6po)7ro<;, Interp., x, 1, the need of a definite 
 compound to express the conception of negation of a concrete 
 noun is not very great and the requirements of expression are 
 more easily satisfied by the less permanent combinations of rela- 
 
The Negative Compounds in Greek. 25 
 
 tive and participial periphrases. Often there is an opposing 
 positive term, as ^evo<;, opposed to 7roX,tT.7;<?, or as t'StaJr?;? is 
 opposed to any professional designation, such as larpo^, dp^fov, etc. 
 
 With abstract terms, however, the case is different and one 
 naturally stops to inquire how the Greek expressed the negative 
 of terms like Slkt], ^o/3o<?, etc. A little examination will show 
 that these negative abstracts are most often derivatives from nega- 
 tive mutata, so that we liave 8if<r], aScKa, aSiKta ; (f)6^o^, a^o^o';, 
 a<^o^la, giving us the pairs, hiKri, aSiKca, (f)6^o<;, a(^o^ia. When 
 once the type is established, the intermediate mutatum may be 
 dispensed with and we may have aBwafMia (Hdt.) opposed to 
 SvvaiJit<;, though ahvvafjboq is not found till Dioscorides. 
 
 With ordinary adjectives (excluding verbals in -To<i) the combi- 
 nation of the negative prefix is by no means free and generally 
 productive, but on the contrary the number of direct combinations 
 of a-privative + adjective seem on examination to be surprisingly 
 small, especially in the earliest period. 
 
 Many undoubtedly old adjectives do not make any negative 
 compounds at all, but have their opposites expressed by positive 
 words, e. g., ixe<ya'i, fMiKpo^ ; a<ya66<i, KaKo<; ; iu,€\a<i, Xei//co9 ; 
 ^apv<i, Kov(fiO<i ; /jbaKp6<i, ^paj(y<; ; evpv<^, arevof; ; Kadapo<;, /xiapci. 
 So generally with adjectives expressing merely physical qualities ; 
 thus none of the adjectives in the list just given makes a negative 
 compound except /ca/co9, which has not generally a physical sense. 
 In the case of many such words no necessity for a negative com- 
 pound was ever felt. We have no *d/jid\aKO(; or ^d-y^oXeTro^ 
 {a')(aXe'Tr(x)<; in Math, Vett.), no *dfxiKpo<i or ^dfxeya^ {d/jLe'ye9i]<i 
 Aristotle). '^d-\lrvxpo<; and "^dvuypo^; are not found, and ciOepfioq 
 is perhaps a drra^ elprj/xevov, used as an abstract, to dOepfxov, 
 Plat. Phaedo, 106 a. Lobeck, Path. Gr. Serm., p. 213, states 
 that negatives are not found from adjectives of the second declen- 
 sion denoting color or taste ; aVt/cpo?, however, does occur in 
 Aristotle. 
 
 But this reluctance of the language to combine a-privative with 
 an adjective is by no means limited to words of any particular 
 semasiological category, and, though the number of such com- 
 pounds is substantially increased in the later language, this type 
 ■was never a favorite form of composition. Examples like ai8pt<;, 
 
26 The Negative Compounds in Greek. 
 
 avaiTio<; by the side of I'Spt?, aiTio<i are few in Homer, not more 
 than 10, and some of these are uncertain. 
 
 When we deduct from the total number of words containing 
 the negative prefix the compounds formed from the verbals in 
 -T09, the mntata, the derivative nouns and verbs, and a few other 
 types of formation, the proportion of the whole consisting of 
 combinations of a-privative + adjective, especially in the best 
 period, is small. 
 
 Thus many derivative adjectives find their opposites in a nega- 
 tive possessive compound, or rautatum. Just as the opposite of 
 BiKT] is not *a-hiK7], but ahuKla, so the opposite of BLKato<; is not 
 ^a-hiKaLo<i, but dSiKo^. Hence there is a larg6 number of such 
 groups of derivative words related to an abstract primitive, e. g., 
 
 Slkt} : SiKaiot; = ahtKia : aSiKO<i, 
 
 (po/So^ : (f>o/3epo<; = a(f)o/3La : a(f)ol3o<;, 
 
 /cocr/1.09 : KocrfiLO<; = aKocrixia : aKO(nxo<i. Cf. pp. 25, 31. 
 
 Valckenaer on Eur. Phoen. 397, denied the formation of de- 
 rivative verbs from compounds of a-privative + a dissyllabic 
 adjective of the second declension except verbals in -ro<i. But, as 
 we have seen, the number of such compound adjectives is at any 
 rate very limited, and Shilleto, Tr. Gamb. Philol. Soc. 1876, 
 p. 74, added dSrjXelv to the avvacroc^elv of Eur., 1. c. 
 
 The cacophony in the repetition of the syllable av may account 
 for the absence of some compounds of ay-privative and a word 
 beginning with av-. Thus from Greek of the best period, out- 
 side of Hippocrates, we have only avavhpo<;, avavBpia Aesch., 
 avdvSpcoTO^ Soph., dvavdr)<; Plato, dvavrajcovicrTO^ Thuc. *^dv- 
 avajKata is not found, nor any compound of the negative prefix 
 with av6p(07ro<; or its derivatives ; and note ovk dveKT6<;, which is 
 very frequent. Later Greek was not so fastidious, so in Cicero's 
 epistles we find : dvavTiXeKro<i, dvavrL(f)a>vr]TO<i, dvavT K^oovqala ; 
 the latter two are not found elsewhere. 
 
The Negative Compounds in Greek. 27 
 
 y. FAVORITE TYPES OF NEGATIVE 
 COMPOUNDS. 
 
 The favorite Greek types of compound with the negative prefix 
 are those with the verbals in -To<i and the mutata made up of 
 prefix + noun. Both of these types were Indo-European and 
 were apparently the most common types in the pro-ethnic speech 
 also. 
 
 The compound with the verbal in -to<; is the most productive 
 of all the types in Greek as in the other languages of the family. 
 This is doubtless partly due to the fact that it is susceptible of so 
 many variations in meaning. Thus the words which compose this 
 group are not simply past participles but they are also very often 
 modal verbal adjectives negativing the idea of fitness or possi- 
 bility, and hence are no longer past in signification. This modal 
 use is Indo-European and it is in combination with the negative 
 that the verbal seems first to have acquired a modal sense; whence 
 the usage became extended, particularly in Greek, to the simple 
 forms. Again these verbal compounds may be active, passive, or 
 neuter. Indeed the fundamental signification of the verbal itself 
 seems to have been only the predication of a verbal action as in 
 some way a quality or characteristic of the subject. Cf. Bishop, 
 A. J. P., XIII, 191 ; Brugraann, I. G. Forsch., v, 93. It is not 
 surprising that such a convenient type of formation should have 
 become so largely productive. A Greek negative compound with 
 the verbal may often be rendered variously into Latin, especially 
 by a compound of in- with a participle in -ns, -tus, or -ndus, or by 
 an adjective in -bills, e. g., airpaKTO'; by injioiens or infeclus, 
 arrLcno^ by ineredibilis, incredidus, and (late) inoredundus, inci'edi- 
 tus. Noteworthy is the large number of Latin negative compounds 
 ending in -bilis. 
 
 Often the negative compound with the verbal is equivalent in 
 meaning to a niutatum of a-privative + noun. Thus, aTrypwro? 
 (TrvpoQ)) II. 23, 270 = dirvpci, ibid., 267 ; a(f)d6v7}TO'i [(pdoveco) == 
 a(j)6ovo<;, both in active (Pind. Ol. 13, 25) and passive sense (ibid., 
 
28 The Negative Compounds in Greek. 
 
 10, 7); dvavSrjTO^ (^avSdco) = dvavSo<; ; dTdpaKTO<; (rapdcraoo) = 
 drdpa'xo'^. Sometimes, however, no intermediate verb from which 
 the verbal may be derived seems to exist, so that the verbal is in 
 effect, though not in form, a sort of denominative participle, e. g., 
 aTreStXo?, Aesch. Prom. 135 = aTreS/Xfyro?, Callim,, VI, 125, but 
 there is no verb "^irehCKow ; so d7rvv8dKa>To<; {trvvha^), Soph. frag. 
 554 = dirvOjjbevo^, Hesych. = aTrvOficov, Theognost. = d7rv6/j,e- 
 vLCTTOf, Eustath. See Schneider on Callim., Hymn in, 213. The 
 above are not to be confused with adjectives which are formally 
 denominative, and in which the suffix -to- is appended immedi- 
 ately to a nominal stem, as in dyepaa-roq from <yepa<?. They are 
 to be distinguished too from verbals derived from denominative 
 verbs, which latter are in turn derivatives from true negative 
 compounds. Thus d\6yrjTo<i from dXoyeo), but dXcyeco from 
 aXoyci. 
 
 The other very frequent type of negative compound in Greek 
 is that of the mutata of a-privative + noun, the so-called posses- 
 sives. In its origin this type seems not to have been a true 
 compound at all, but a derivative adjective made by giving to an 
 immutatum that variation according to gender which characterizes 
 an adjective. But the formation is a very old one and inde- 
 pendently productive, so that for very few of the existing mutata 
 can underlying immutata be cited. The fact that compounds 
 whose second member is a substantive in -09 or -ov do not regu- 
 larly make a special form for the feminine is significant. The 
 type is perhaps older than the use of the d-forms to denote the 
 feminine gender, and later forms like dre'xvo'i, -ov, from rexvrj 
 may be explained as made on the analogy of the pre-established 
 type; see Brugmann, K. Z. 24, 39 ff., Wheeler, Trans. Am. 
 Philol. Assn. 30, xxi ff. 
 
 This type of negative compound is well represented in Greek 
 and Sanskrit, is not very common in Latin (cf. iners, inops, et al.) 
 and is foreign to German and English. On the other hand in 
 the last two languages, as already remarked, the immutata are not 
 at all uncommon. 
 
The Negative Compounds in Greek. 29 
 
 VI. EXPRESSIONS WHICH MAY REPLACE THE 
 NEGATIVE COMPOUNDS. 
 
 The negative compound may be replaced by a complex with a 
 negative adverb or by other forms of expression which, though 
 not negative in form, are so or nearly so in effect. 
 
 The negative of the sentence and of the verb may in various 
 ways become extended in its use to other parts of speech — the 
 so-called ou-privative, Thompson, Gr. Synt., §§ 280-1. First, 
 the close and constant combination of the negative adverb with 
 the verb may persist with other words, particularly participles, 
 which are derivatives from the verbal root or stem. Thus we 
 have an old Greek compound, OvKuXeymv, to be connected with 
 the fact that the verb dXeyw is used in Homer most often with a 
 negative. Other illustrations are ovk eOeXwv, ov ^(a'ipwv, ovk i(ov, 
 to be referred to the close association of negative and verb in ovk 
 ideXo), ov 'xciiprj(T€t„ ovk ew, and also of course to the intimate 
 connection of the participle with the verbal paradigm. One may 
 compare Latin nescius for inscius by the side of nescio. 
 
 Secondly, by confusion or displacement of the negative relations 
 in the sentence from which a nexus of the negative and noun, 
 adjective, or adverb might result. Peculiar conditions of emphasis 
 and metre often bring the negative into a position in the sentence 
 which seems very arbitrary ; so the position of the negative in the 
 poets, especially Pindar, is apparently very free. In sentences 
 where the predicate is in the form of a noun or adjective with 
 the sentence negative, especially where either the necessity for 
 a copula is not felt or the substantive verb is only a copula, 
 a sort of adherescence may take place ; cf. cases like ovk ayaOov 
 Trep iovra, II. 9, 637, ovS' akiov eiro'^ eao-erat, II. 24, 92 
 (so often with oti;;^ a\io^, a common litotes in Homer) ; also 
 OVK €pavo<; rciSe yecTTiv, Od. 1, 226. , The Latin well exemplifies 
 the working of such a process in nefas and nimirum. Nefas proba- 
 bly started in the sentence, nefas (est), at a period when ne had not 
 yet given place to nan as the sentence negative in Latin ; cf. the 
 common use of fas est and the frequency of nefas and nimirum as 
 
30 The Negative Compounds in Greek. 
 
 parenthetical exclamations. Ne/astus, nefandus, nefarius are of 
 course deriv^atives. Again such confusion or displacement of the 
 negative might occur in other relations within the sentence, as in 
 the case of an object, or an adjective, or an adverbial modifier. 
 Note expressions like ovk dirocpcoXta €lSco<i, Od. 5, 182, 6S' ov^ 
 aXiov /3e\o<? rJKev, II. 15, 575, aij S' ov Kara /juolpav eeiTre?, Od. 2, 
 251, ov Koa-fio) irapa vav(f)iv iXevaofMed' avra KeXevOa, II. 12, 225; 
 so ov Kara Kocrfxov, II. 2, 214, and frequently. Note the Sanskrit 
 adverbs naciram and mdcirani, and cf. Whitney, Gram. 1122 e. 
 
 Such an extension of the use of the negative originally belong- 
 ing to the sentence or to the verb to combinations with parts of 
 speech other than the verb is probably inevitable in any language 
 and, as we have seen (p. 17) is regular in the case of the indefinite 
 pronouns. But some of these uses of the negative adverb with 
 parts of speech other than the verb may date back to the period 
 when the negative adverb and the negative prefix were as yet 
 undifiPerentiated. Various conditions, however, must have oper- 
 ated to make this use of the negative adverbs a part of the 
 language's ordinary means of expression. Such a combination 
 was less permanent and more readily produced than a compound 
 with the negative prefix and would often be used where no corre- 
 sponding compound with the prefix was in familiar use or, if it 
 was, could not be used in certain metres, e. g., avcovofiao-TOf, 
 which appears first in Euripides and which was barred from the 
 hexameter, is represented in Homer and Hesiod by ovk ovofiacrTOf;. 
 So doubtless melri causa, ov Oefjua-rov, Aesch. Sept. 694, is used = 
 aOefxicrrov. Then for various reasons most adjectives never 
 formed compounds with the negative prefix ; e. g., *dv-d7rapvo<i, 
 *dv-avdyKaio^, ^d-Kd0apo<; seem not to have existed. Hence we 
 must have in Antiphon, 1, 9, ovk ovcrav dirapvov ; 5, 11, rol'^ /xr] 
 Ka6apoL<i ; Xen. Rep. Lac. 5, 4, rd<^ ovk dva'yKala'^ Troaea. In 
 general the compounds combining directly the negative prefix and 
 the adjective, being not at all numerous in Greek, must be to 
 some extent replaced by the combination of negative adverb and 
 adjective, especially where no positive of precisely opposite signifi- 
 cation was in use. Sometimes the combination {7rapd6eai<i) of 
 negative adverb and adjective is so close as to be almost equal 
 to composition {a-vvOea-is:). In such cases ov may be retained 
 
The Negative Comfpounds in Greek. 31 
 
 where the generic /a?) would be expected, aud thus we may explain 
 some, though of course not all, of the cases like rov<; ov')(^ olou<; re 
 ovra^, Lys. 20, 19 = Tov<i a8vvdTov<i ovtw; ; tov<; ov-^ ofJiO(^v\ov^, 
 Just. Mar. Apol. 1, 14; rov ovk opdco^ 'x^poifievov, Plat. Gorg. 
 457 c. 
 
 The participle, showing its close connection with the verbal 
 system, regularly takes the negative of the verb even when the 
 participle has almost a purely adjectival sense, e. g., ov irpoarjKwv, 
 as in Antiphou, tetral. V, a, 3 ; Or. v, 2. The readiness with 
 which occasional complexes of negative adverb and participle 
 could be formed may account in part at least for the reluctance 
 of the language to form compounds of negative prefix + adjective, 
 cf. p. 25. Cf. Prof Giklersleeve, A. J. P., ix, 139, n. 3, end. 
 The rarity of the compounding of the negative prefix with parti- 
 ciples belonging to the tense systems is noticed elsewhere (p. 18), 
 and the use of the sentence negative with the participle is a matter 
 of course where the participle is consciously used as a substitute 
 for the verb of a subordinate sentence. Yet an approximation to 
 a strictly verbal use of the participle compounded with the nega- 
 tive prefix is seen in Latin me indieente, me insoiente (abl. abs.), 
 and a few others, see Delbriick, Vgl. Synt., ii, 529. 
 
 The reluctance of the Greek lansruao-e to form immutata of 
 a-privative + noun left it often with pressing needs of expression 
 which the use of the negative articular infinitive, the formation 
 of new abstract derivatives, or the employment of periphrases 
 with the relative or participle could not relieve. So we have : 
 Si airetpoavvqv kovk airohei^LV, Eur. Hipp. 195 f.; ev ov Kutpw, 
 id. Bacch. 1287; ?; fxr] ^fjuTreipia, Ar. Eccl. 115; t7]v jxtj iTTLTpoTrrjv, 
 Plat. Legg., XII, 966 c. A number of examples are found in 
 Thucydides : t7]v tmv <ye(j)vpMV . . . . ov SiaXvaiv, I, 37, 4 ; rijv 
 ov TrepLTei')(^icnv, III, 95, 2 ; rhv tmv '^^copLoov aX\7]\oi<; ovk 
 aTToSocTLV, V, 35, 2 ; rrjv ovk e^ovcrtav Tr]<i aycovLaeco'i, V, 50, 4. 
 Cf. Lat. negotinm = nee otium ; *in-otium was hardly possible. 
 Cf. p. 24. So even with concrete nouns ; tmv ov')(l Sov\a>v, Eur. 
 frag. 831 ; tmv fxt] pr^ropwv, Philodemus, Rhet., ii, 6, 18 ; Plat. 
 Gorg. 459 B, has o fir] larpoi;. Cf. non-orator, Quint., il, 15, 17. 
 Some of these expressions are due to the need of terms to express 
 the negation of abstract philosophical concepts, e. g., to ovk 
 
32 The Negative Compounds in Greek. 
 
 avOpoiira, Arist. Interp. 10, 1 ; cf. non-corpus, Cic. Ac, i, 11, 40 
 = incorporeum or incorporale, and the Hegelian nicht-ich [non-ego, 
 non-moi), nicht-sein (ro firj 6v, non-tire). Old French seems to 
 have been fairly productive in these compounds with non, but 
 in- was early substituted for this, doubtless originally a literary 
 reversion to the Latin. Cf. Darrasteter, De la creation des mots 
 nouveaux dans la langue fran§aise, p. 140. So in English, nou- 
 regardance, Shak. Twelfth Night, Act. V. 
 
 For expressions like ov KaKovpy6<i elfMc, Autiphon, v, 9, the 
 Sanskrit might use a negative compound, cf. abrdhmana = any- 
 body but a brahman. 
 
 Often too the combination with the negative adverb had to be 
 employed to denote the negation of a quality where the compound 
 with the prefix had come to signify its opposite. Note the differ- 
 ence between aKaKo<i (Sappho and Aeschylus) and ov kuko^;, Ar. 
 Pax. 430. Expressions like the latter lend themselves readily ta 
 litotes. In that form of litotes in which an adjective which 
 is already a compound of the negative prefix is again negatived, 
 the usage of the language would tolerate nothing else but the 
 negative adverb; so ovk d^pojv 6 ^elvo<i oterai, Od. 17, 586, 
 where ^avdcfipwp, besides being unmetrical, is formally impossible 
 in Greek. There is no litotes, however, in ovk akrjdri, Antiphon, 
 V, 26 ; VI, 28, for aXrjdrj'i has really lost the force of a negative 
 compound and has come to be felt as a positive, so that in vi, 28, 
 cited above, ovk oXtjOtj does duty as a quasi negative compound 
 opposed to a preceding oXtjOtj. Sanskrit has less repugnance to a 
 doubling of the negative prefix; cf. Whitney, Gram. 1121, b. 
 Perhaps we are to see a double negative in dvdeSvo<i, where, how- 
 ever, one prefix does not nullify the other (cf. p. 15). 
 
 Further examples of the frequent use of the negative adverbs 
 instead of the prefix are due to the tendency of the sentence 
 negative to attach itself to certain adverbs of time and certain 
 conjunctions with which the negative prefix, belonging as it does 
 strictly to noun formations (including nouns, adjectives, and 
 adverbs derived from these), is never used. Such combinations 
 are : ovKen, oviray, ovTrore, ovBi, ovre, etc. So too in the corre- 
 sponding combinations with the negative /xi], though the question 
 is here complicated by the special functions of fnj as a sentence 
 
The Negative Compounds in Greek. 33 
 
 negative. In Thuc, vii, 34, 6, such an adverb is used even with 
 a noun, hca rrjv rwv K.opiv6icov ovKeri eTravajcoyT^v, cf. p. 31. 
 
 Regular is the use of the sentence negative with adjectives like 
 Tra?, 'rrdvTe'i, ttoXv^, ttoWol, which are allied to the pronouns or 
 numerals in their meaning and are never in the best period com- 
 pounded with the negative prefix ; airoXv; occurs in Damascius ; 
 aTTocof;, Democr. ap. Stob., and airoaa, Eccl., are technical meta- 
 physical terms. The negative adverb is used also with adverbs 
 like fxaXa, adc^a, Stjv, and in general with those which were not 
 felt distinctly as derived from nouns or adjectives. 
 
 We are told by Choeroboscus that the negative adverb cannot 
 have a distinctly privative force ; cf. Et. Mag. 639, 50, ?; fiev 
 (TTeprjcn^ eVi/SaXA-OfTo? iarcv cnev^ia, cb? aTe')(yo<i prJTcop, 97 8e 
 d7ro(fiaai<; co? eVu^e Xa/x/SdveTai .... 01) reT^fi/co? 6vo<i- 
 aTe')(yo<i he 6 ovo^, ov Svvarov Xiyeiv. But the difference is rather 
 to be explained as due to the characterizing force of compounds in 
 general, dre-xva applied to the orator affects his character, not so 
 when it is applied to the ass. 
 
 When the negative prefix appears in a mutatum as dre'y^i^o'i, the 
 negative adverb can be used in its place only by combining it with 
 a derivative adjective, as in ov Te')(}>t,K6<i. A mutatum formed 
 with the negative adverb would be a decided anomaly. Other 
 equivalents for the mutata may be found in phrases with dvev, 
 ')(a)pi'i, or their equivalents ; e. g., irovov roi ^copl? ovBep evrv^ei, 
 Soph. El. 945; dvev uKoXovdov, Plat. Symp. 217 a, drep KopvOof 
 T€ Kol dairiho^, ouS' e^ef e'y')(p<i, II. 21, 50. So Suidas, dvdifxoTOfi, 
 
 %(W/>t.9 opKov. 
 
 ovSevoa-copa (neut. pi.), II. 8, 178, is a solitary example of a 
 dependent negative compound with a case form in the first member, 
 belonging thus to the class of improper compounds {irapdOera). 
 In ovTi8av6<; -avo- seems to be a suffix, and -rih- is perhaps = 
 Tt = Lat. quid (Brugmann, Vgl. Gram. 2, 135). 
 
 Just as we shall see that by a kind of hyperbole the negative 
 compounds were used in cases where the ancient grammarians 
 attribute to the prefix the sense of KaKov or oXljov, so we find 
 that by the figure of understatement, of which the Greeks were 
 rather fond, the compounds in Sva- were sometimes employed 
 where a strictly negative compound would not be improper. So 
 
34 The Negative Compounds in Greek. 
 
 Bv(T6^7]VvaTo<; Secr/xo?, Eur. Hipp. 1237 ; kukwv <yap 8vcraXo)TO<; 
 ouSet?, Soph. O. C 1723; SuaSfjiota (= dvo/xoca) cited from 
 Strattis by Suidas ; also Suaajvo<;, hva^arc;, hvavorjTo^. The 
 use of the negative compounds in hyperbole and the employment 
 of the hvcr- compounds in litotes makes it possible for compounds 
 of either class to do duty as substitutes for those of the other. 
 Cf. the use of Latin male in phrases like statio malejida carinis, 
 Verg. Aeu. 2, 23. The potential force of the verbals in -To<i is 
 predominant in their composition with hva--, as it is very common 
 in the strictly negative compounds. 
 
 Similar is the use of kuko^; in composition ; e. g., KaKOTV')(r]^, 
 Eur. Med. 1274 = drv'xri'i and cf. KaKorv)(^cov, Thuc, il, 60, 3. 
 Sometimes in the scholia negative compounds are explained by 
 compounds with KaKo- and 6\i<yo-. So darj^wv KaKocnfjiJbwv, 
 schol. Soph. Ant. 1013; Eustath. on Od. 11, 490, ' a/cXT^po? 8e 
 dvr]p oXtyoKXrjpo'i. Cf. Eustath. on II. 5, 800, to Se oXlyov 
 ravTov ian vvv t<x> rJTrov, avrb 8e dvrX rov ou8afia)<i. So Suidas, 
 sub voc. oXlyov, dvrl rov ovSo\co<; (ouS' oXco?). 
 
 Compounds with certain prepositions too may be in eflPect nega- 
 tive. So with diTO, diro^evof;, Soph. O. T. 196 = d^evoq, id. 
 Philoc. 217. Examples are numerous : aTroSetTri^o?, ciTrepyo^, 
 dirodpi^, d7r6dvfMO<;, d7r6cnT0<;, d.TroaTop'yo'i, diroTiixo^. Cf. Schmid, 
 Atticismus, 4, 696. dirrjXiaaTd (dual), Ar. Av. 110, immediately 
 after rfXtaard, being a noun immutatum, is an irregular and comic 
 formation made for the sake of the antithesis ; cf. on dBcorrj^, p. 
 23. Examples with i^ are : e/cStKo?, cKvofxa, eKvov;, eKc^pcov. 
 So even irapd. dvojxo'i' 6 irapdvoiMd, Suidas, sub voc. So vtto 
 in composition with an adjective has often a minimizing force 
 comparable with that of the diminutive suffix in nouns, as vtto- 
 \evKo<i, v7rofj,eXa<i, viro'rrvppo'i, somewhat white, etc., a usage 
 especially common in scientific writers. 
 
 Compounds with certain verbal stems as first members are often 
 practically negative. As in the case of other compounds with 
 verbal first member, they are mostly poetic in the classical period. 
 Examples, (^vyoirroXep^o'^, Od. 14, 213, ^vyofMwx^o^, Simon. 106; 
 compounds of Xtiro- are found in tragedy and abound in late Greek, 
 especially in Nonnus, as XiTroyXwaa-ci, Xnroa-Oevrj'i, Xlttoctkioi;. 
 So Xvcn/j,epi,/jivo<; = dfjiipi/jLva in its sense of ' driving away care.' 
 
The Negative Compounds in Greek, 35 
 
 For certain adjective first members equivalent to the nega- 
 tive prefix compare eprj/uiOTroXi^; Eur., eprjfioKOfMrjq Anth. Pal., 
 KevavSpa Aesch., Soph., Kevohovri^ Anth. Pal., K€v6o-apKo<i Et. 
 Mag., /jLovo/jbdrcop Eur. 
 
 With reference to the derivatives it may be observed that in 
 some few cases the range of usage of a negative verb might over- 
 lap that of a denominative verb derived from a negative compound. 
 Thus under some circumstances aTria-reco might = ov Treido/xac and 
 ayvoeco might =^ ov yiyvooaKo) ; on the other hand the common 
 ovK edeXwv and ovk iOeXo) quite take the place of a negative 
 adjective and its derived verb. 
 
 A similar partial coincidence in function is true also in the case 
 of the negative abstract noun and the articular infinitive ; but the 
 coincidence is only partial. The articular infinitive is much the 
 more readily productive ; it does not, as does often the abstract 
 noun, acquire transferred significations different from that of its 
 verb ; the variation in form according to voice and tense makes 
 the meaning of the infinitive more explicit; finally, the infinitive 
 may carry with it a subject and all the verbal modifiers. Cf. 
 Prof. Gildersleeve in Trans. Am. Philological Assn., 1878, p. 18. 
 On the other hand the abstract noun is more implicit and terse ; 
 it has more of a€fxv6Tri<; ; and, where it is in common use, it is 
 less clumsy than the articular infinitive. 
 
 VII. SEMASIOLOGY OF THE NEGATIVE 
 COMPOUNDS. 
 
 The simplest and readiest theory for the entrance of the nega- 
 tive into language is the assumption that the negative was originally 
 free and formed a sentence by itself, a phenomenon still not un- 
 common in Classical Greek ; cf. Plat. Phaedr. 236 D. Later by 
 the most primitive sort of parataxis it might come to be used just 
 before or after a positive sentence, thereby negativing the whole. 
 For the effect cf oXolto fiev fi^, Eur. Med. 83 ; also Ar. Aves 
 1219-20, IP. TTOta yap dWj] ■^prj TrereaOaL T01/9 deovf ; IIEI. 
 OVK ol8a fid At" e7<07e* rrjBe fiev yap ov. Cf. too Prof. Gilder- 
 sleeve's explanation of the origin of oy fi^ in independent sentences 
 from the free negative, A. J. P., in, pp. 203-5. A sentence thus 
 
36 The Negative Compounds in Gi^eek. 
 
 negatived as a whole is simply " the expression of the fact that 
 the attempt to establish a relation between the two ideas has 
 failed," Paul, Prin. of Lang. (Eng. trans.), p. 122. This attach- 
 ment of the negative to the sentence, by which the latter is annulled 
 as a whole, would naturally be followed by the attachment of the 
 negative to some part of the sentence to which the negative particu- 
 larly applies. Thus the subject may be felt to be the same, while 
 it is the predicate which is altered by the negative, so that the 
 negative finds its way into the predicate. Now if the predicate 
 denotes only an act or state which is somewhat temporary in its 
 nature and which is generally expressed by the verb, the combi- 
 nation of negative and verb is naturally loose and not more 
 permanent than the act or state which is expressed by the verb. 
 The adjective, however, — and this includes the participle in its 
 earlier use, — has more of a characterizing function and expresses a 
 more or less permanent quality or property of the subject. So 
 when the subject is characterized by the absence of a certain 
 quality, an habitual combination of negative and adjective which 
 is used to express this will necessarily become a close compound. 
 Cf. Hartung, Gr. Partik. ii, p. 74, " So wie diese Negation den 
 BegrifF selbst, und Nicht seine Beziehung bestimmt, so kann sie 
 auch korperlich nicht als ein besonderes Wort existiren sondern 
 wird mit dem zu negirenden Worte eine Zusammensetzung bilden 
 miissen." Delbriick, Vgl. Synt. ii, p. 533, seems to suggest that 
 the negative may have been first used as a prefix with the parti- 
 ciple, from whence the usage was extended to adjectives and nouns, 
 but it is hard to see any basis of probability for this. A predicate 
 adjective or noun in a negative sentence without a copula might 
 easily furnish the conditions for starting a compound. 
 
 The necessary and sufficient condition for a compound is that 
 the individuality of the component element be lost to the conscious- 
 ness and the combination itself become the conventional sign for a 
 single notion. When this state of affairs has been reached the 
 compound adjective may become used as an attribute or as a mere 
 epithet. That which denotes the absence of a quality may easily 
 come to denote its contrary and may finally through usage acquire 
 a positive content of its own. 
 
 Whether a negative compound shall pass through the inter- 
 
The Negative Compounds in Greek. 37 
 
 mediate stages between mere negation and a positive content 
 naturally depends very much on the needs and resources of the 
 language. Thus the contrary of KaKo^i is pretty well expressed 
 by dyaOo'i, so that dKaKo<i is used as a merely negative word = 
 ■* guileless/ and if it suggests more than this, we have an instance 
 of litotes ; aKaKo<i never ^= the simple ' good,' aSiKd, however, 
 is regularly opposed to 8LKaio<;, for there is no other word which 
 expresses the contrary of SlKaco<i so precisely. Again oXtjOtj^; 
 has gone a long way from its original negative signification and 
 has acquired a very distinct positive notion of its own. 
 
 In a stage of thought and language where there is little occasion 
 to distinguish the negative from the contrary, there would be but 
 little impulse toward the formation of negative compounds and 
 the needs of expression would be satisfied by the primitive pairs 
 of positive adjectives of opposite meaning, as fxiyafi fjLiKp6<;, 
 ^frvxpo<i depix6<i. This is a further exemplification of the general 
 principle emphasized by Osthoff in his recent academic address, 
 Vom Suppletivwesen der Indo-Germanischen Sprachen, that 
 there is a stage in man's use of language when his capacity 
 for classification is weak and when differences of tense, gender, 
 degree of comparison, etc., are expressed rather by means of 
 words formed on different roots than by words formed by suffixal 
 or other modifications of the same root. So the use of words 
 radically different to express the two genders of the same 
 species seems to be older than the use of the so-called substantiva 
 mobilia ; e. g. pairs like brother and sister, son and daughter, 
 horse and mare, are undoubtedly older than formations like 
 a8e\0o9. d8e\(jii] ; jilius, jilia ; he-goat, she-goat. Words of this 
 class which are material (stoffiich) rather than formal opposites 
 seem to occur mostly to express the closest of family relationships 
 and the sex of the most familiar of domestic animals. In English 
 we say bull and coiv, but in the case of animals which are not 
 native to British soil or which are less familiar we are apt to 
 resort to mere formal variation of suffix, prefix, or compound 
 to express gender, so lion, lioness, bull-buffalo, coiv-buffalo, buck- 
 rabbit. The same principle holds true of diminutives; thus we 
 iSay baby, colt, puppy, but baby-elephant, lion-cub, etc. 
 
38 The Negative Compounds in Greek. 
 
 To Osthoff's examples of the working of the principle in the 
 defective verbs (so-called), irregularly compared adjectives, sex- 
 words, numerals and pronouns, might be added the expression 
 of qualitative opposites by words from different roots rather than 
 by a word and its formal negative. While a language is still 
 growing and productive in the formation of new words, it has 
 uo need to avail itself of the device of the negative compound. 
 Of negative compounds which may fairly be assumed for the 
 pro-ethnic speech, Fick, I. G. W5rterbuch, i, pp. 94 ff., gives 
 only about eighteen, and many of these have a secondary char- 
 acter, as have also the great majority of the compounds in Homer 
 and in Greek generally. Ten of these eighteen words are found 
 in Greek and in almost every one of them the negative force i& 
 still very prominent and not easily expressed by a synonym that 
 is positive in form. 
 
 The introduction of new negative compounds is doubtless most 
 often due to a desire for precision of expression, for distinguishing 
 between negative and contrary, for expressing the opposite of some 
 new positive adjective for which the language offers no contrary^ 
 or for expressing the contrary of one of a pair of original con- 
 traries which have shifted their meaning so that their original 
 polarity is no longter preserved. Doubtless too a certain desire 
 for euphemism in expression or litotes is the cause for the intro- 
 duction of many negative words, but the consciousness of the 
 figure soon fades away with use. Thus the production of new 
 compounds is often an index of reflectiveness or of an affected 
 striving for precision. 
 
 The negative formations seem to have proved so convenient in 
 the growth of the language that they were used as vehicles of 
 expression for ideas which seem to have arisen independently 
 of any formal positive contrary. So dvapxo<; can be opposed 
 only by evTaKra or Koafxco^; ; a/nefiTTTo^ is far more common 
 than /i.6/i7rT09. dveXeyKTo^; and ave^e\eyKTo<; are not uncommon, 
 while eXeyKTo-i is cited only from Hesychius, and so with many 
 verbals in -T09. 
 
 Rarely we find a positive adjective which seems only a 'pendant' 
 to its negative ; e. g. Tnj/juav, * baneful/ Orph. Hymn., opposed to 
 
The Negative Compounds in Greek. 39 
 
 airrj fimv , Horn. ; imfjv, ' winged/ Herodian, opposed to aTrrijv, 
 Horn. Cf. the Sanskrit sura, ' god ' by popular etymology from 
 asura, 'demon.' 
 
 The tendency of the negative compounds to acquire a contrary 
 or positive content prevents any but a new or unusual compound 
 from being used unmodified in the common Greek figure of litotes. 
 In a new or occasional compound the force of the negative is 
 more strongly felt and the word has not acquired by use a positive 
 notion of its own ; so with words belonging to a highly poetic 
 sphere. The litotes can be felt in a/ca«:o9, afxei\i'xp<i, a\afi'rrrj<i, 
 dS6Kt/jio<; ; but the examples are not numerous. 
 
 The positive content of many of these compounds is shown too 
 by the readiness with which they may form derivatives or extend 
 their inflection into the comparative or superlative degrees. The 
 large number of denominative verbs and derivative nouns is note- 
 worthy and examples of comparison are frequent, as airoriJbOTaTO';, 
 Od. 1, 219, dfjuadearare, Ar. Ran. 933, dvoaKorepov koX dirtcFTO- 
 repov, Andoc. 1, 23. 
 
 A quite exceptional instance of the reinforcement of the negative 
 prefix by another and independent negative is to be seen in Eur 
 Androm. 745-6, aKia yap dvTi(TTOL')(o<i mv (fxovrjv e%€t9 | dBvvarof; 
 ovSev aXko TrXrjv Xeyetv [xovov, where the context shows that 
 V. 746 means ' unable to do aught but speak,' and not ' feeble in 
 naught but speech,' which would be the normal interpretation. 
 (i8vvaro<i is here used like ov 8vvaT6<; and the poet seems to have 
 made the odd experiment of trying to resolve the compound by 
 treating the inseparable prefix as separable and an independent 
 word (cf. p. 36). 
 
 In Soph. Ant. 175-7, the negative sense of d/x'^x^vov justifies 
 the following irplv dv with the subjunctive : 
 
 d^ir]'^avov Se 7ravT0<; dv8po<; eK^aOeiv 
 yfrv^^yjv re kuI (ppovrifia Kol yvco/xTjv, irplv dv 
 dp')(al<i re Koi vofioiaiv ivTpt/Srjf; cfjavfj. 
 
 Cf. Eur. I. A. 489-90 and see A. J. P. ii, 469 ; G. M. T. 633. 
 
40 The Negative Compounds in Greek, 
 
 The Greek name for the common negative prefix was ak^a 
 
 a-reprjTCKov = alpha-privative. Yet cf. Chrysippus ap. Simplic. 
 ad Arist. Cat. ed. Basil, f. 100 A, to ^yap aOdvarov a-repTjTLKov 
 eyov TO (T')(r]/jba r?}? Xe^eto? ov crrjfjbaLveL arepr]cnv' ov yap iirl 
 ire^VKOTO'i airoOvrjCTKeLV, elra fjurj aTrodvyaKovTCi y^pdifxeda rcS 
 ovo/jiaTC. . . . Kal to kukov Se SrjXovTat TroXXa/ci?, co? d(f>a)VOV 
 iXeyofiev TpaycpBov tov KaKocj^covov. Kal d7ro(f)daet<; Se hrjXovvTai 
 Sm Tcbv aT€p7}Ti,Kcov cf)(ov(ov, oicnrep to Sid<popa, dhLd(^opa Kal 
 XvatTeXrj, aXvaiTeXij. So Choeroboscus, Can. 361, and Et. Mag. 
 639, 50, dfieXet rjVLKa rj a crTepr]ai<; fxrj crrjfiaivet iiTL^dWovTO'i 
 d'iroTV')(iav, totc uvtX diro^daeoa^; avTrjv Xijovatv eo? d6dvaT0<;, 
 aXoyot;. Thus it is plain that even the ancients saw that the 
 prefix was not always privative in the strict sense of that word, 
 but was often a mere negation (aTro^atri?) as in dBLd(j)opo^, 
 d\v(ri,Te\i]<i, and so, in a sense, in the mutatum dddvaTo^. Cf. 
 pp. 6, 33. But the distinction between privation and mere nega- 
 tion is generally outside of language. 
 
 Again we have a sort of hyperbole when d^covo<i is used of a 
 bad actor, dKdpBio<i of a coward, d7r\evpo<i of a man with weak 
 lungs, dfiadrj'i of a man of little learning. Cf. Lobeck, Path. Gr. 
 Serm., i, p. 29, where Choeroboscus and Theodosius are quoted 
 as follows : to ak<^a eiTTa arj/jbalveL' (TTeprjcriv &)? d(f)iXo<i, iTTLTacriv 
 O)? TO d^v\o<i vXt), ofjbov 6i<i d\o')(o<i Kal dhe\<^6<i, KaKov &)? to 
 d^covo<i 6 KaK6<^ci>vo<i, oXljov to? d/jLadr)<i 6 oXiyofiaOijii, ddpoiaiv 
 0)9 TO diraf;, irXeovaa-fiov co? to dcrTa(j)o<i. It is evident that what 
 the grammarians call a-Tepr)(Tt<;, KaKov and okiyov belong to our 
 alpha-privative, though the latter two do not at all express the 
 true force of the prefix, but show simply that the negative com- 
 pounds may be used in hyperbole. On the use of a negative 
 compound to denote what is disagreeable or monstrous, cf. Soph. 
 El. 492, where the scholiast explains d\€KTpo<; by 8va\eKTpo<i, 
 and Soph. Trach. 1060, 07X0)0-0-09, schol. KaKoyXwaao^ ; so 
 d/jLop(f)ou in Cic. ad Att. 7, 8 fin., which Tyrrell translates ' bad 
 form ' ; cf. the German Ungewitter. This use of the negative 
 prefix has been thought by some to form the connecting link 
 between it and the so-called alpha-intensive, the iirLTaac^; of the 
 grammarians quoted above. Pott, E. F.^, i, p. 387, cites 0701/09, 
 which is glossed by Hesychius with iroXvyovo^. So Key, Trans. 
 
The Negative Compounds in Greek. 41 
 
 Philol. Soc, 1865, p. 64, compares a-privative and intensive to 
 Latin male, which may be privative or intensive according to 
 circumstances. But it is simpler to connect most of the cases 
 of so-called a-intensive with the a-copulative, the o/xov and 
 dOpoKTif of the grammarians. 
 
 VIII. THE NEGATIVE COMPOUNDS AS AN 
 ELEMENT OF STYLE. 
 
 In the study by us moderns of a language no longer accessible 
 to us through the medium of living speech analytical methods 
 play an important part in assisting our appreciation of the ele- 
 ments of style. The observation from a syntactical and lexical 
 point of view of the development of modes of expression has 
 therefore a certain value in enabling us better to estimate the 
 stylistic character of the various Greek authors. 
 
 The consideration of the negative compounds with reference to 
 style falls naturally into two divisions. We may observe, first, 
 the effect upon style which the negative compounds have in 
 common with the compounds in general and, secondly, the stylistic 
 character which is peculiar to the negative compounds. 
 
 The creative activity of the users of language in the produc- 
 tion of new words is nowhere more apparent than in the coinage 
 of new compounds. The best Greek prose with its characteristic 
 moderation used compounds for the most part to convey some defi- 
 nitely crystallized idea which required a permanent and unitary 
 form for its expression. Thus the compounds were not used in 
 Greek as in Sanskrit prose to give expression to a passing fact or to 
 a merely incidental attribute, but for these the Greek would use the 
 more readily extemporized syntactical combinations of words for 
 which it had abundant facilities, especially in its large use of the 
 participle and the articular infinitive. So the common use in 
 the later Sanskrit of compounds to express a mere occasional 
 complex of ideas makes it impossible to ascribe to a Sanskrit 
 compound that more definite characterizing force which a Greek 
 compound has. We are to recognize then in the use of a new 
 compound by a Greek author a distinct exercise of the creative 
 faculty in language and we should expect to see the compounds 
 
42 The Negative Compounds in Greek. 
 
 entering the language in those spheres where the creative faculty 
 has the greatest activity. One of these spheres is that of the 
 7roir)Ti]s, whose function it is to beget new combinations of ideas 
 or at least new expressions for old ones. In using a new com- 
 pound the poet has a form of expression which is his own creation 
 and which puts the idea in a brief and implicit form. His 
 language thus acquires a cre/zi/or?;? which is above the level of 
 ordinary speech. This is the o-eyavoT?;? which is a common 
 characteristic of a nominal as contrasted with a verbal expression ; 
 cf. crooTTjp 'ladc, Aesch. Ag. 512, awTrjp jevov [mol, id. Cho. 2, for 
 <T<x>^e and awaov ^e respectively, also Hdt. 7, 139, 3. If the poet 
 uses, as he sometimes does, a compound to designate a particular 
 action or relation, he nevertheless characterizes the subject through 
 this act or relation, and this makes his statement of the fact much 
 more impressive. Aeschylus' 7ro\vdvopo^ aixc^l >yvvaiK6<;, Ag. 62, 
 characterizes Helen more than any less implicit expression could 
 do. In (TV fiev <pdl/jb€vo<; aXalvei'^ dda7rTo<;, dvv8po<;, Eur. Tro. 
 1085, the compounds state condition as well as fact ; so droKo<i, 
 Eur. El. 1127, ^r^elroiv, ibid. 1130. 
 
 The more implicit character of compounds as contrasted with 
 the corresponding syntactical combinations lent itself readily to 
 that striving after vague expression which was characteristic of 
 the dithyramb. The dithyramb was the great ' Tumvielplatz ' for 
 all that was bold in the formation of compounds, and tragedy 
 shows traces of its dithyrambic antecedents in its tendency toward 
 fullness and boldness in the use of compounds, especially in the 
 choral parts. The comic poets, as creative as any others, but with 
 a different purpose, either parodied the more serious poets in the 
 use of compounds, or else carried their formations to the extreme 
 of clumsiness for comic effect. 
 
 The second great sphere of creativeness in the production of 
 compound words is the language of those who deal with subjects 
 more or less technical. Such are the philosophical and scientific 
 writers, men who must have new terms to describe new con- 
 ceptions and discoveries. In the endeavor to give accurate 
 expression to new ideas they tax to the utmost the capacities of 
 the language and sometimes do violence to its laws of composition. 
 The negative compounds are especially prominent in these depart- 
 
The Negative Compounds in Greek, 43 
 
 ments of literature, as for example in many of the writings of the 
 Hippocratean corpus, and we may compare the large use of non- 
 and a- as prefixes in modern technical and scientific writings. A 
 glance at an English dictionary will show how numerous are 
 words like non-contagious, anodyne, anaesthetic, etc. The wonder- 
 ful fertility of the Greek in evolving new technical terras by 
 composition of elements native to the language, when compared 
 with the barrenness of English in this respect, need occasion no 
 surprise, for it will be remembered that the Greek could not, as 
 does the English, draw for its technical vocabulary on two ancient 
 languages so long the vehicles of science and culture. 
 
 While very many of the compounds remained confined in their 
 use to the poetic and technical spheres, others descended to the 
 level of the more ordinary prose. Some of the compounds which 
 appear in ordinary prose are really technical in origin, as when an 
 orator employs a legal expression or an historian a word borrowed 
 from the vocabulary of military science, nor is it surprising if an 
 orator occasionally uses a lofty and perhaps tragic word in the 
 epilogue. In later Greek the poetic or technical feeling of the 
 older types of compounds was largely lost and their use became 
 a part of the language's common stock of expressions, so that 
 there was an approach to the facility of the German in composi- 
 tion, yet never to that of the Sanskrit. The poets, especially in 
 the later periods, made their use of compounds more poetic by 
 multiplying compounds of unusual types ; cf., e. g., the large use 
 of compounds like \t7r6(TKio<i (= dcrKia) in Nonnus (cf. p. 34). 
 
 The line between the artistic and technical compounds and those 
 of every-day speech it is of course impossible to draw ; the techni- 
 cal language or cant of one period may be a part of the popular 
 idiom of the next. The compounds of the poets do not so readily 
 tend to sink to the common level as do the technical words, the 
 former being the creations of art and feeling rather than due like 
 the latter to the necessities of expression. The artistic compounds 
 may be classed with the luxuries, the scientific and technical ones 
 among the necessities of language. It is noticeable that a very 
 large number of the more common compounds, including many 
 negative words, belong to the ethical sphere, a department of 
 
44 The Negative Compounds in Greek. 
 
 Greek thought in which poetry and philosophy may be said to 
 meet and one where antithesis is particularly frequent. 
 
 The form and meaning of the negative compounds lent them- 
 selves readily to particular stylistic effects. 
 
 This is very noticeable where these compounds are used two or 
 more together. Formally this makes a species of that figure 
 which is variously called alliteration, annomination, homoeopro- 
 phoron, parhomoeon ; in Greek 6fj,oi6apKTov, ofioLOKarapKrov, cf. 
 Max. Plan, in Walz, Rhett. Gr., v, 511, 6, 551, 4. Doubtless the 
 full sound which belongs to the vowel a gives it a phonetic effec- 
 tiveness suited to pathetic situations. Here, however, the figure 
 resides in the sense as well as in the sound, and we have really a 
 species of anaphora. Now a negative compound characterizes its 
 subject, as the name a-a-repTjTtKov implies, by attributing to it a 
 deprivation or lack of something and, if this something be one 
 of the chief objects of human interest or desire, the privation 
 connotes feeling and the compound has necessarily a pathetic or 
 drastic character, the effect of which is heightened by repetition. 
 So II. 9, 63-4, 
 
 a<^priTwp d6efMiaT0<i ave<7ri6<i icrrcv eiceivo<i, 
 09 TToXifMov eparai iTnBrjfjiiov 0Kpv6evT0<i, 
 
 where the effect is made greater, as often, by the asyndeton ; 
 cf. Od. 1, 242, ot'xeT ai(TTo<; a7rvcrTo<i. So also II. 2, 201, 
 aTTToXe/xo? koI dva\Ki<;, and the same combination in II. 9, 35 
 and 41 ; Simon. 36 (51), ovSe .... airovov ouS' d(f)6irov ouS* 
 aKivhwov ^iov \ t'? y^pa^ i^iKovTo Te\eaa-avre<i. For such ex- 
 pressions of a more felicitous character note the frequent combi- 
 nation, ayijpaov aOavcLTTjv re, II. 2, 447, et al. 
 
 This figure is a great favorite with the tragic poets. Next ta 
 the TT-alliterations the a-alliterations are the most numerous in the 
 plays of the three great tragedians and, as is the case with 
 alliteration generally, a considerable majority of the cases occur 
 in the lyric parts. Many of these instances of a-alliteration are 
 due to the massing of the privative compounds, though the 
 alliteration is sometimes extended by the addition of other words 
 beginning with a. See Riedel, Alliteration bei den drei grossen 
 
The Negative Compounds in Greek. 45 
 
 griechischen Tragikern, diss., Erlangen, 1900. Examples are: 
 a')(opov aKidapiv .... "Apr], Aesch. Supp. 681 ; afia')(ov, arro- 
 Xefiov, aviepov, id. Ag. 769 ; cf. Cho. 55 ; uKXavra, acfiiko^, 
 dvv/j,6vato<i ep')(^ofjLat rav Trvfidrav oSov, Soph. Ant. 876 ; ibid. 29, 
 1071 ; "Ai'So? ore Moip' dvvfj,evato<; \ aXvpof; a')(opo'i dvanrecfirjve, 
 id. O. C. 1221; cf. ibid. 1236-7; ciTrai^ re Kayvvat^ Kavkcmo'i, 
 id. frag. 4 N. ; dvd8e\(po'i, dirdrbyp, d(f)cXo<i, Eur. Or. 310; d<yafio<;, 
 dT€Kvo<;, aTToXt?, d(])i\o<i, id. Iph. T. 220 ; ddeov, civo/jlov, uSckov, 
 id. Bacch. 995, So Ar. Ran. 204 is doubtless paratragedy : Kara 
 TTfti? Sw^crofiac \ dneipa d0a\drTcoTO<;, daa\afA,Lvto<; \ mv, eir 
 iXavveiv, of. Vesp. 729-30, and Phrynichus, fragg. 18, 19 k (ii, 
 pp. 587, 592 m). For the orators cf. Autiphon, 1, 22, dOe/jiiara koX 
 dvotjLa Kal dTeXeara /cal dvrjKOVcrra ; in the Palaniedes ascribed 
 to Gorgias, 36, heivov ddeov dSiKov dvop,ov epyov, recalling Eur. 
 Bacch. 995, quoted above ; Demos. 25, 52, dXX' daTreicrTo^, dvi- 
 SpvTO^, dfiCKTOt;, ov ')(dpiv, ov ^cXiav, ovk dXX" ovSev oiv dvOpcorrro^ 
 fierpoo^; ytyvoxxKcov ; cf. Lys. 12, 82. So Plat. Phaedr. 240 A, 
 dyafxov diraiSa doiKov. The frequency of the phenomenon in the 
 English poets has long been a subject of remark. Two stock ex- 
 amples are, "unwept, unhonored and unsung," Scott, Lay of the 
 Last Minstrel, 6, 1, 16, and "unhousel'd, disappointed, unaneled," 
 Shak. Hamlet, 1, 5, 77. The large use of negative words is thus 
 often a mark of an elevated style ; cf. remarks hereafter on their 
 use in Autiphon. 
 
 As in some of the examples given above, the compounds are 
 frequently massed in threes. Whether we are to see in this fact 
 any connection with three as a sacred number and hence with the 
 threefold repetition common in old religious poetry and formulae 
 is doubtful. The compounds occur very often in pairs and occa- 
 sionally in a series of four or more. Yet cf. Karl Frey, Homer, 
 progr. Bern, 1881, p. 31, "Die Dreizahl ist in der Kunst eiue 
 feierliche Form. Und ihre Verweudung ist unendlich. Und im 
 Grossen und im Kleinen wird sie angewendet ; drei Epitheta sind 
 ein beliebter Redeschmuck, und der griechische Tragiker theilte 
 sein grosses Festspiel ein in drei Acte; est ist eine Trilogie. Auch 
 die greichische Komodie beruht auf der Dreizahl, insofern zwei 
 Parabasen die Handlung trennen ; auch sie ist eine Trilogie." 
 
46 The Negative Compounds in Greek. 
 
 This whole subject belongs to the great chapter of reduplication 
 in language. A different phenomenon is the Greek fondness for 
 reinforcing one negative by another one or more as in Plat. 
 Phaedo, 78 D, ovheirore ovhajjif) ovZajjbw^ aXXolcoatv ovhe^lav 
 ivBex^Tai ; for a reinforcement of the privative prefix by another 
 negative see p. 39. 
 
 Another form of reduplication is the jigura etymologlca, a figure 
 which doubtless had its beginnings in the oldest popular poetry 
 and which survived in tragedy, and was used occasionally by 
 most of the orators. With some negative compounds where the 
 negative or privative force is still very strong we have what we 
 may call a cognate genitive, but cognate only with the second 
 member of the compound, e. g., a^i\o<? cjjlXcov, Eur. Hel. 524 ; 
 reKca TraTpo^ airaTopa, id. Her. Fur. 114; tov rjBlcTTOv aKov(r/xa- 
 T09 avrjKoo^ el koI tov rjSiaTov 6ed/jbaTo<i ddearci, Xen. Mem. 2, 
 1, 31 ; TifjLi]<; drcfji0<; Tracr?/?, Plat. Legg. 6, 774 B; aTraa dppevoiv 
 iralScov, Andoc. 1, 117; cf. Xen, Cyr, 4, 6, 2. The genitive is 
 generally limited by an adjective. So in the rhetorical use of the 
 cognate accusative in the orators we have a host of examples of 
 dBiKTjfia dSiKetv, but here the negative force is perhaps hardly 
 present to the consciousness. A combination of figura etymologica 
 and antithesis is often possible with the negative compounds, e. g., 
 Aristarch. Trag. frag. 2 n, koI tov daOevi) aOevetv \ TiOrjat koI 
 TOV airopov evplcTKeLV iropov. 
 
 There was developed in Greek a form of the jigura etymologica, 
 a peculiar type of oxymoron, which created for itself a special set 
 of negative compounds. The earliest example of the use of this 
 form of expression is in Od. 18, 73,'Ipo9 'Atpo?, where the com- 
 pound seems obviously formed for the purpose of the word-play. 
 We may compare Ai/o-Trapt?, II. 3, 39, and KaKoiXtov, Od. 19, 
 260. The Homeric 'Ipo? "Ai'po? is followed by more than a score 
 of similar expressions in later authors, chiefly the tragedians, as 
 Aesch. Pers. 680, vde<; dvae^, et al. 
 
 '1/309 'Atpo9 has usually been intrepreted to mean ' Irus, who is 
 no longer Irus,' and so vde<i dvaef = vde<; ov vde<i ovaai. Yet 
 Froehde, Bezz. Beitr. 20, 216, would read dFipo<i = Sanskrit 
 avlras, ' unmanly.' Curious indeed it would be if a chance pun 
 of the epic poet had furnished the model and authorization for the 
 
The Negative Compounds in Gh^eeh. 47 
 
 rather numerous brood of later imitations, which, however, are 
 not puns but examples of the figura etymologica. Some, as Curtius, 
 Or. Gram., § 360, and Wheeler, Der Gr. Norainalaccent, p. 46, 
 have inclined to look upon these negative formations as immutata 
 and they were perhaps influenced by the analogy of the Sanskrit 
 negative noun immutata. The accent in the Greek words does not 
 seem decisive in one way or the other. Brugmann, Vgl. Gram., il, 
 1, p. 89, considers theui to be on the border line between adjective 
 and noun and takes refuge in the statement that the distinction 
 between adjective and noun is not a thoroughgoing one in the 
 Indo-European languages. Lobeck, Paralipomena, 229 ff., argues 
 stoutly in favor of taking them as adjectives. A case like vvfju^r^v 
 avvficfjov (v. 1. -?}), Eur. Hec. 612, if the common reading is 
 correct, favors this view, otherwise one might expect *d-vvfju(f)7]. 
 So "^v-^r] dy\rv')(o<i, Ar. Ran. 1334, and fjbrjrrip a/njTcop, Soph. El. 
 1154, where otherwise '^d'\lrvxv and *d/j,7]Tr]p might- have been 
 expected. The change in ending undoubtedly helped the feeling 
 of change in part of speech. As has been shown already (p. 22), 
 all Greek analogy is against the combination of the negative })re- 
 fix with nouns to form nouns, and when this occurs in Sanskrit, 
 the ending generally suffers no change, so from vagci, cow, is made 
 avagd, non-cow. The true parallel to this last in Greek is found 
 in expressions like yuvatKa ov yvvat/ca, Soph. O. T. 1256 ; cf. the 
 popular riddle in Bergk-Hiller-Crusius, Anthol. Lyr., p. 131 : 
 
 AiVo? Tt? iaTtv, &)? dpr]p re kouk dvi]p 
 opviOa KovK opviO^ l8(ov re kovk l8o)v 
 eVi ^vXov re kov ^vKov Kadrj/jbevrjv 
 Xldfp re KOV \i6(p ^dXot re kov j3d\oi. 
 
 So in Latin, imitations of the Greek figure are not precisely 
 similar formations, but employ either negative adverb -f noun, 
 or else a clearly adjectival formation, asfunera nee funera, CatuU. 
 64, 83, insepulta sepultura, Cic. Phil. 1, 2, 5 = rd(f)ot aTacf)OL. So 
 these formations nyist, I think, be looked upon as adjectives and 
 hence mutata. The mutatum aTraa, meaning 'one who is no 
 child,' must of course be distinguished from tlie more common 
 mutatum, dirat^, meaning ' childless.' 
 
 /^ ^ or TV. 
 f UNIVER^ 
 
48 The Negative Compounds in Greek. 
 
 These compounds, as is shown alike by the range of literature 
 in which they appear and by the striking figure in which they are 
 employed, are quite artificial and artistic formations, perhaps, as 
 suggested above, modelled on a misunderstood Homeric pun, and 
 the examples are numerous enough to show that they were a part 
 of the regular stock of figurative expression belonging to Greek 
 tragedy. The figure in its unmodified form is rare in classical 
 prose ; a few examples have been noted in the prose of the later 
 period. 
 
 The following examples of the figure have been collected : 
 'Ipo9 "Ai/309, Od. 18, 73 ; vde<; avae<;, Aesch. Pers. 680 ; ci'x^apt'^ 
 Xdpt<i, id. Prom. 544, Ag. 1545 (cf. %apti/ axapirov, Cho. 42) 
 a7ro\e/i09 TroXe/io?, Prom. 904; vo/xov dvo/juov, id. Ag, 1142 
 aTToXiv TToXiv, id. Eum. 457; TratSe? avratSe?, id. Eum. 1034 
 dBcopa Bcjpa, Soph. Aj. 665; fi'^rrjp a/XTjrcop, id. El. 1154; dyafiov 
 <yd/xov, id. O. T. 1214 : v7rvo<i dv7rvo<;, id. Phil. 848 (cf. docKou 
 elaoiKTjacv, Phil. 534) ; ttot/jLov aTroTfiov, Eur. Phoen. 1306, 
 Hipp. 1144; vvfjL(f)7]v dvvfji(f)ov, wapOevov dnrdpOevov, id. Hec. 612; 
 aTToXefjiov TroXe/xov, id. H. F. 1133; Secrfjuov dSea/jLov, id. Suppl. 
 32; X'^P^^ dxapiv, id. Iph. T. 566 (cf. Phoen. 1757); SaKpv dSaKpv, 
 id. Iph. T. 832 (dub. lect) ; 6801;? dv68ov^, id. Iph. T. 888 ; ydf^ou 
 dya/xov, id. Hel. 690 (cf. Phoen. 1047); cf. d Se /xeyaXoTroXii; 
 dTToXii; oXroXev .... Tpota, id. Tro. 1291 ; "yjrvxv d'y^rvxo^, Ar. 
 Ran. 1334 (Aesch. loq.) ; aSee? Seo?, Plat. Symp. 198 A; TroXt? 
 diroXcq (pred.), id. Legg. 766 D ; /3to9 dj3io<i, Leon. Tar. Anth. 
 Pal. 7, 715, 3; %apt9 dxcf'pt,<;, id. ibid. 9, 322, 2; ic6a/j,o<i dKoa-fici, 
 Jul. Aeg. ibid. 7, 561, Antip. Sid. ibid. 9, 323, Epigr. Gr. 365; 
 KrJ7ro<i dKTjTToq, Greg. Naz. (v. Thesaur.) ; SelTrva dBetTrvo^, Nonn. 
 17, 51 ; KO(Tfio<i dKoa/jbo<;, id. 6, 371 ; oIko<; doLKCi, id. 17, 42. It 
 will be noted how a number of the expressions are repeated and 
 seem to have become a part of the literary tradition of the 
 language. 
 
 The effect of this bold figure is to show that the object to 
 which the epithet is applied contradicts its own true nature and 
 so belies its name. The force of these expressions may be highly 
 pathetic and their use is out of place except in the most im- 
 passioned speech, ylrvxv dyjruxo^, Ar. Ran. 1334, is put into the 
 mouth of Aeschylus, and the fact that the figure is thus used 
 
The Negative Compounds in Gu'eek. 49 
 
 in travesty shows that it was a recognized element of tragic 
 diction. 
 
 There are a number of examples of equivalent oxymora which 
 are not however precisely parallel in form ; e. g., /xijrep ifir) 
 Sv(T/jL7]T€p, Otl. 23, 97; diropa 7r6pi,fio<;, Aesch. Prom. 904; 6eXyoi<; 
 av dOeXKTov, id. Suppl. 1056; drXrjTa rXdaa, id. Ag. 407; 
 TjKova dvTjKovara, Soph. El. 1407 ; (f>(o<; dcfieyyef, id. O. C. 1549 ; 
 ^dfi,ov<i 8vcr<ydfjiOV<;, Eur. Phoen. 1047; avrj^acara) nrvpi, id. Or, 
 621 ; 7rt(TTtv rSiv iv dvdp<o7roi<; dTrLaTordTrjv, Andoe. 1, 67 ; /xrjSe 
 ra? '^dpcTa<; d')(^apiar(ii'i ')(api,^6ixevo<i [Isoc], 1, 31 : d^to3TO<i ^co<;, 
 Philemon, frag. 90, 7 ; 93, 7 k. (87, 90 M.), and freq. ; cf. Autiphon, 
 tetral. B, yS, 10. 
 
 Aristotle in the chapter of his Rhetoric on 07/C09, iir, 6, 7, 
 1408 a, teaches that it is advantageous to describe a thing by the 
 qualities which it does not possess, — ef mv fxr) e^ei Xeyecv. ' For 
 thus,' he says, ' the amplification may be carried to infinity,' — 
 av^erai yap ovrcc; €t<; direipov. He continues : ccttl he tovto 
 KoX iirl dyadoov koX KaKMV, ottco? ovk e^et, OTrorepw? au y ■)(^prjcnfiov. 
 69ev Koi rd ovofjuara ol TroLTjral (f)€povcri, to d')(ophov Koi to 
 dXvpov fieXo^' GK rwv a-repiqcreoyv yap e7n(f)epovaiv' evSoKCfiel yap 
 TOVTO iv Tat<; fieTa(f)opai<; Xeyofjuevov Tai<i dvdXoyov, olov to (f)dvai 
 TYjv adXinyya elvat /j,eXo<{ dXvpov. The figure thus belongs to 
 Aristotle's class of proportional metaphors, fieTacf^opal e« tov 
 avdXoyov. In Rhet. 3, 11, 11, 1412 b, fin., Aristotle gives 
 another example : to^ov (sc. io-TX) (f>6pp,cy^ d')(opho'i, cf. Theogn. 
 Trag. frag. In.; so in the example in Poet. 21, 15, the shield is 
 called (fiLaXr} "Apeew? or (f)idX7) doLvo<i ; in Aesch. Sept. 82, the 
 dust (kovl';) is called dvavBo<; dyyeXoi;. See Cope, Arist. Rhet. ad 
 locc. The following proportions may illustrate the character of 
 the figure : 
 
 To^ov : vevpd, bowstring (omitted term) 
 
 = (pop/jity^ : x°P^V> lyrestring ; 
 k6vi<; : its appearance (omitted term) 
 
 = dyyeXof; : avhrj. 
 
 This device is often employed by the poets and gives a special 
 sphere for the use of the negative compounds. Examples abound 
 
50 The Negative Compounds in Greek, 
 
 ill the tragedians, as dirrepa (j)drt<;, Aesch. Ag. 276 (cf. eVea 
 TTTepoevTo) ; the sceut of blood is /jLtjvvtvp a^6e'yKT0<i, id. Eura. 
 245 ; the eagles are Zrjvbf} aKpayel'i (v. 1. aKXayyel'^) Kvve<;, id. 
 Prom. 803 ; the sting of the gad-fly is apBt<; a7rvpo<;, ' a barb not 
 wrought with fire/ ibid. 880; the Pleiades are called aTrrepoc 
 7re\€idSe<;, Aesch. frag. 312 N.; midnight is daa\.7ri,HTo<; copa. 
 Soph. frag. 357 N. ; an animal's hide is daTrddijro^; ')(\aiva, id. 
 frag. 793 N. So Kcofxov dvavXorarov, Eur. Phoen. 791 ; alSov<^ 
 d'X^akKevTotcnv e^evKTui TreSat?, id. frag. 595 N. 
 
 Antithesis is a figure native to the Greek, and expressions like 
 the negative compounds, which have an opposition in their very 
 nature, are very readily employed in such a figure. Antitheses 
 of this kind, where two expressions are opposed not only in mean- 
 ing but also in form, afford besides a species of paronomasia. Cf. 
 p. 46. In the ethical and philosophical sphere such pairs of 
 negative compounds or derivatives with their positives are of very 
 frequent occurrence, but much less often is the antithesis a rhe- 
 torical one. Expressions like : irepl tmv BtKaicov koL twv dSiKcov, 
 Isocr. 15, 255; rou? art/iou? i7riTi/xov<; iroirjcraL, Andoc. 1, 109, 
 are too simple and unavoidable to be considered as conscious elements 
 of style, though their very common occurrence would be an index 
 of an author's cast of mind. More rhetorical is the balancing in 
 epyov S' ovSev oveiSci, depjir) Se r oveiSoq, Hes. Op. 311 ; Scott; 
 Ijuev Tt<i eB(OK€v, dBcorj) S' ov rt? eScoKe, ibid. 355 ; Tra? ti<; irXovcnov 
 dvSpa Ttet, aTLei. Se irevi'x^pov, Theognis 621 (in the latter two 
 examples the antithesis accounts for the anomalous forms, dScorr), 
 drieL, see pp. 19, 23) ; Gorg., Palamedes, 30, iiroirja-e tqv dv- 
 OpwTrivov ^Lov TTopifMov i^ dtropov koI KeKoafiTj/xevov i^ aKocr/jLov. 
 So ovT(o Oi re rjfuv acTLOC eleven SovXtov ^vyov oi re dvalTioi, 
 Hdt. 7, 8, 3 ; irpoaipovjxevcov /juerpiov /3iov p,erd StKacoavvrjii 
 fiaXXov T] fiiyav irXovrov p,€T dSt,KLa<;, Isocr. 8, 93 ; el tov<; fiev 
 ■^ScKrjKOTai; Ti/jbcopla^ dcjielvaL Kvptac yevoLVTO, e0' rj/juv Se Tol<; ev 
 TreTTotrjKoa-tv aKvpoi KaTacTTaOelev, id. 18, 68. 
 
 Here belongs mention of the negative compounds as used in the 
 (T')(r}iJba KUT dpaiv koI Oicriv, the figure by which both the nega- 
 tive and positive form of expression are used together for the same 
 thing. This is a favorite usage in the older literature ; see 
 Bekker, Homerische Blatter, ii, 222. Rehdantz, Ind. Dem. 
 
The Negative Compounds in Greek. 51 
 
 Phil, sub voc. apai<i, distinguishes the two cases, where the 
 negative expression precedes and where it follows. The former 
 is a case of litotes followed by its antithetical positive to guard 
 against a misunderstanding, as in ov i^r)i<? aedXcov .... aXX" iv 
 TrpcoToccrcv oto) efifxevat, Od. 8, 179 ; ovk aKXrjroc, 7rapaK\r}0€VT€<; 
 Se, Thuc. 6, 87, 2 ; ovk a<j)p6vco'i dWa vovv e')(ovroi<i, Isocr. 
 Ep. 5, 2. This sort of antithesis is frequent in Antiphon, e. g., 
 ovhe d<j>avr]'i dXka koI \[av (j)avepo<i, tetral. B, 7, 7 ; oyre .... 
 ocria aXX' dvoaC av TrdOoifxev, ibid. 11 ; cf. Blass, Attische 
 Beredsarakeit^, i, 143. 
 
 The case where the negative follows is somewhat different, for 
 here the first expression logically includes the second. At/ca^o? 
 necessarily implies ovk uSlko^, but the converse is not true. 
 Examples are : eoiKe roi, ov rot d€iKe<;, II. 9, 70 ; jvco p ouS' 
 r)<yvoir}cre, Hes. Theog. 551 ; aoc^irj . . . ., ovk d'yvwiMoavvr), 
 Hdt. 2, 172 ; ryvcora kovk diyvcora, Soph. O. T. 58 ; eKovra kovk 
 ciKovTa, ibid. 1230, conversely, uKovra koI fir) eKovra, Lys. 13, 19 ; 
 wairep d'tropia^i ovctt]'? .... dXk" ov 7roWrj<i d(f)dovia<i virap'^ovcrr]';, 
 Isocr. 12, 90. Such is often the force of litotes that an adjectival 
 expression with the double negative is stronger than that of the 
 opposite positive ; thus rhetorically the effect is a climax ; cf. 
 Gildersleeve, note on Justin Martyr, I, 22, 10. 
 
 Litotes, the figure of understatement, often appears in the form 
 of a merely negative expression of that which could be more truly 
 expressed positively. As has been shown already, the negative 
 compounds in common use could seldom be used alone in this 
 figure. Ordinarily, for instance, the use of the term d(Tej3rj<i 
 would be no litotes at all, for the word has by use acquired a 
 positive content which makes it the proper contrary,* not the mere 
 negative, of evaej3y]<;. One might well suppose that ao-e/St^?, 
 when first used, if applied to a downright opponent of the worship 
 of the gods, might be felt as a true instance of litotes. 
 
 But it is coupled with the negative adverb that these words 
 appear so frequently used in litotes ; indeed it is probably true 
 that litotes in Greek is much more frequent with negative adjec- 
 tives than with positive and that some negative words are used 
 only in this figure. The reason for this is doubtless that the 
 positive adjectives in common use are more numerous than the 
 
52 The Negative Compounds in Greek. 
 
 negative, and since this form of litotes consists in negativing 
 the opposite of the word which would be used in simple direct 
 statement, we should expect the opposites of the more common 
 adjectives to predominate here and not the common adjectives 
 themselves. 
 
 The force of the figure seems to lie partly in the suggestion 
 of reserve on the side of the speaker or writer and partly in the 
 fact that the hearer or reader is compelled to make for himself 
 the true judgment which the language used does not state but only 
 implies, so that the thought becomes more a part of his own 
 thinking. This latter fact accounts largely for the great force 
 in general of the implicit in expression, a phenomenon already 
 noticed in connection with the use of compounds. 
 
 The figure is very common in Greek from Homer down. Cf. 
 the Homeric ovk dcjipcov, ovk deKcov, ovS" d'rrldrjae, ov rot aet/ce?, 
 oit . . . . dK\r)€i<i. djueXeo) is always used with a negative in 
 Homer. The figure is not so common in prose as in poetry, but 
 it is absent from no department of the literature ; cf. ovk dhrfK,ov, 
 Antiphon, 1, 13; ovk drLjxwpriro'i, id. tetral. B, h, 8.. 
 
 The poets sometimes increase the phonetic impressiveness of 
 their compounds by adding an element which, while making 
 greater the number of syllables, contributes little to the sense, 
 so that the result hardly differs in force from a simple negative 
 compound. So for the simple negative prefix a negative adjective 
 is sometimes substituted, e. g. dlhpoh'iKiqq, Find., and many with 
 airetpo-, as direcpor^afio'i, Eubul. Com. frag. 35, direLpohaKpvi , 
 Aesch., d7retpofid')(r)<;, Find., direipoTToXepbO'i, Dion. H., direipo- 
 TOKO'i, Antip. Sid. in Anth. Pal. and many in late writers; so 
 Hesych. d'rroXepLor d7recpo7r6\€fjbo<i. Similar are some compounds 
 in dijiLko- particularly affected by Cyrill of Alexandria, e. g. 
 d(j)L\66eo<i and d<f>t\6KOfji7ro<;, hardly different in force from 
 a^eo9 and dKoiJ,7ro<;. Again a final element is sometimes added 
 which, while formally the second member of a bi-merabral 
 compound, is hardly more than a mere suffix in force. The 
 phenomenon is frequent in the language of Aeschylus and may 
 belong to the solemn diction of the old religious poetry. Cf. 
 such series of names of divinities as : "E/caro? (Apollo), 'E/caep- 
 709, 'E/c77^oXo?, 'E«:aT7//3oX,o9 ; 'Apiarr) (Artemis), 'Apta-To^ovXr], 
 
The Negative Compounds in Greek. 53 
 
 'ApKTTo/jbdxV' (Usener's Gdtternamen, pp. 49 f. ; A. J. P. xvii, 
 357-8.) In the case of the negative coniponnds those which have 
 an active sense are occasionally extended, mostly in late Greek, by 
 being again compounded with -ttoco'^, -rrpay^'i, or -epyo';. Soa^o/3o- 
 TTotof, schol. Aescij. = a(^o/3o9, Aesch. Prom. 902; aSiKOTrpayt]^, 
 Stob. nearly = aSiKO<i ; so d6efjitTOvp'y6<;, Eccl., (i6ep,LT0<; ; dvocn- 
 ovpyo';, Aristotle, Ep. Plat., dvo&ia. 
 
 IX. HISTORY OF THE FORMATION OF NEGATIVE 
 COMPOUNDS IN GREEK. 
 
 As we have seen, the Greek negative prefix is the formal equiv- 
 alent of the corresponding element in the original pro-ethnic speech, 
 and it is an interesting question, but one hardly to be satisfactorily 
 answered, how many of the pro-ethnic negative compounds the 
 Greek inherited and preserved. Fick in his Worterbuch der Indo- 
 germanischen Sprachen,* I, pp. 85 f., gives a list of 10 such words, 
 aytTTTTO?, dvayij'i, dvv8po<i, aKvpo^, djv(oTO<;, dfi^poro'i, airov^i, 
 dvdFaro^, d(f)diTo<;, dv7rvo<i. Such a list has some slight value but 
 must not be accepted as at all certain. In the case of such a pro- 
 ductive type of formation it is very possible that two or more in- 
 dividual languages may have combined phonetically equivalent 
 elements quite independently. Again, some of the old words may 
 have survived in a single language only or the form may have 
 become altered in one or more of the languages owing to analogical 
 influences. A comparison of the phrase KXiFa dc^dirov, II. 9, 
 413, with the Sanskrit gravas aksitam gives strong proof of the 
 high antiquity of d(j)diTo<;, and the retention of the /u, in d/x0poTO<; 
 marks the compound as old. Fick compares *d^poT'r]aio<i with 
 the Zend amerefdt. 
 
 It is unquestionably true that the readiness with which new 
 compounds w^ere made in the Greek language and the types of 
 formation which were prevalent varied much in the diiFerent periods 
 of the language and the different spheres of the literature in which 
 they were employed. 
 
54 The Negative Compounds in Greek. 
 
 The following tables are the result of an attempt to determine 
 the growth in the use of the negative compounds and derivatives 
 as far as is shown by the emergence of new words with. the develop- 
 ment of the literature. To undertake to examine the text of the 
 various authors with reference to the use which they made of the 
 negative compounds, new and old, would carry us beyond the limits 
 of this dissertation. 
 
 It cannot be claimed for the following tables that they are be- 
 yond the need of revision. With the exception of a few additions, 
 notably words from the new Bacchylides, the tables include only 
 the words found in Liddell and Scott's lexicon. To ascertain the 
 first occurrence of the words in the literature Stephanus' Thesaurus 
 has been used. The corpus of inscriptions has not been taken into 
 account except in the instances where a word is cited only from 
 inscriptions, and there is a possibility of error in the uncertainty 
 as to the genuine and spurious portions of the Hippocratean corpus. 
 There are included, too, a few of the inevitable cases of doubtful 
 date and etymology. Absolute accuracy in such a set of figures is 
 perhaps hardly to be thought of. The main outlines, however, of 
 the tendencies of the language are made clear and can be given so 
 satisfactorily in no other way. 
 
 Our collection of the negative compounds is more complete than 
 the one included in the general lists of Greek compounds made by 
 Schroeder in his book "Ueber die formelle Unterscheidung der 
 Redetheile," pp. 198 ff. A collection of the Latin compounds and 
 derivatives containing m-privative has been made by F. L. Vicol, 
 Programm des Gr.-Or. Obergymnasiums in Suczawa, 1890 and 
 1891. 
 
The Negative Compounds in Greek, 
 
 65 
 
 TABLE I. 
 
 Showing the emergence of new negative words in the literature by 
 classes and by authors, periods, or departments. See classification on 
 p. 16. The column headed m includes anomalous forms which could 
 not be otherwise classified. 
 
 
 
 
 
 
 g 
 
 TO 
 
 o 
 
 u 
 
 
 
 O 
 
 
 
 
 
 
 
 
 
 ^ 
 
 XI 
 
 
 
 •*-" ^ — » 
 
 
 
 
 
 
 
 
 
 o . 
 
 O o 
 
 a . 
 
 n 
 
 J3 
 
 TO 
 
 
 a.3 
 
 
 
 
 
 
 
 a, 
 
 
 
 a> c 
 
 a P- 
 ° £ 
 
 TO " 
 
 o 
 
 a 
 
 C 
 
 a: 
 
 to E 
 
 .s 
 
 te 
 
 s 
 
 
 
 93 H 
 
 It 
 
 <0 
 
 s 
 
 e v^ 
 
 I- 
 
 O 
 
 c 
 
 -a' 
 
 TO 
 
 
 > 
 
 ^ 
 
 
 
 « e 
 
 > 
 
 o 
 
 s 
 
 o V 
 
 If TO 
 
 <u 
 
 Q) 
 
 a^ ^ 
 
 <u 
 
 tc 
 
 
 
 4- ^ 
 
 + 
 
 + 
 
 + 
 
 ^ 
 
 
 
 _> 
 
 "13 bt 
 
 
 o 
 
 
 
 
 
 
 
 
 
 Oj 
 
 
 cj ^ 
 
 c3 
 
 
 
 
 '3 
 
 
 > 
 
 > 
 
 B 
 
 03 
 
 
 _> 
 
 _> ^ 
 
 > 
 
 o 
 
 
 
 A 
 
 & 
 
 a, 
 
 O- 
 
 
 <u 
 
 4J 
 
 <V 
 
 lU 
 
 V 
 
 s 
 
 
 
 « 
 
 s 
 
 8 
 
 8 
 
 ^ 
 
 ^ 
 
 M 
 
 Q 
 
 Q 
 
 P 
 
 -< 
 
 
 
 ac 
 
 6 
 
 C 
 
 d 
 
 ef 
 
 .9 
 
 A 
 
 i 
 
 k 
 
 / 
 
 m 
 
 Tot. 
 
 Indo-European (?), v. Fick.. 
 
 1 
 
 4 
 
 
 
 5 
 
 
 
 4 
 
 10 
 1 
 
 19 
 
 2 
 
 10 
 
 Homer 
 
 13 
 
 6 
 
 71 
 15 
 
 (5) 
 (1) 
 
 2 
 
 84 
 11 
 
 6 
 1 
 
 17 
 3 
 
 9,9fi 
 
 Hesiod 
 
 39 
 
 Early elegiac and iambic po- 
 
 
 etry 
 
 5 
 
 11 
 
 
 
 IV 
 
 
 6 
 
 
 1 
 
 
 1 
 
 41 
 
 Homeric hymns and later 
 
 
 
 
 
 
 
 
 
 
 
 
 
 epic 
 
 ?: 
 
 n 
 
 
 
 10 
 
 1 
 
 3 
 
 
 
 1 
 
 
 ?R 
 
 Early melic poetry before 
 
 
 
 
 
 
 
 
 
 
 
 
 
 500 B. c 
 
 2 
 
 11 
 
 
 
 17 
 
 J 
 
 
 1 
 
 2 3 
 
 
 37 
 
 Frags, early hist., phil., com. 
 
 
 
 • 
 
 before 500 B. c 
 
 13 
 
 11 
 
 56 
 
 
 
 3 
 
 46 
 
 6 
 
 4 
 13 
 
 
 3 
 
 2 
 6 
 
 
 ^0 
 
 Aeschylus 
 
 143 
 
 Aeschylus and Pindar 
 
 1 
 3 
 
 8 
 15 
 
 
 2 
 
 9 
 12 
 
 209 
 
 11 
 
 26 
 
 2 
 
 6 
 
 53 
 
 1 
 
 6 
 
 1 
 18 
 
 31 
 
 3 
 
 9A 
 
 Pindar and Bacchylides 
 
 48 
 
 Total before 500 b. o 
 
 45 
 
 210 
 
 (6) 
 
 603 
 
 
 
 Attic philos. and tech. lit. 
 
 
 
 
 exc. Plat. Aristot. and 
 
 
 
 
 
 
 
 
 
 
 
 
 
 Hippocr 
 
 3 
 
 20 
 
 
 
 2 
 
 2 
 
 / 
 
 
 1 
 
 3 
 
 
 38 
 
 Sophocles 
 
 10 
 
 43 
 
 
 
 5'' 
 
 3 
 
 4 
 
 
 o 
 
 q 
 
 
 123 
 
 Euripides 
 
 8 
 
 35 
 
 
 
 36 
 
 5 
 
 I'' 
 
 
 
 4 
 
 
 100 
 
 Tragic frage , 
 
 
 1 
 
 
 
 9 
 
 
 
 
 
 
 
 3 
 
 Herodotus 
 
 6 
 10 
 
 13 
 34 
 
 
 3 
 
 14 
 49, 
 
 1 
 
 13 
 
 40 
 
 1 
 
 9 
 
 1 
 
 5 
 
 17 
 
 1 
 
 53 
 
 Hippocrates 
 
 150 
 
 Historians, lesser and lost.... 
 
 
 2 
 
 
 1 
 
 2 
 
 
 2 
 
 
 
 
 
 7 
 
 Old Com. fragg 
 
 2 
 
 14 
 
 
 
 18 
 
 5 
 
 7 
 
 ^ 
 
 1 
 
 3 
 
 1 
 
 53 
 
 Aristophanes... 
 
 3 
 
 16 
 
 
 1 
 
 6 
 
 3 
 
 7 
 
 
 1 
 
 3 
 
 
 39 
 
 Antiphon 
 
 1 
 
 10 
 
 
 1 
 
 9, 
 
 
 6 
 
 
 9 
 
 
 
 ?,^ 
 
 Andocides 
 
 5 
 
 1 
 ?8 
 
 
 
 1 
 6 
 
 
 3 
 
 1 
 
 4 
 
 ? 
 
 
 3 
 
 Thucydides 
 
 4K 
 
 Lysias 
 
 
 5 
 
 
 
 3 
 
 
 
 
 2 
 
 1 
 
 
 11 
 
56 
 
 The Negoiive Compounds in Greek. 
 
 TABLE I (Continued). 
 
 Isocrates , 
 
 Isaeus 
 
 Plato 
 
 Xenophon 
 
 Aeschines 
 
 Demosthenes 
 
 Hypereides 
 
 Aristotle 
 
 Middle and New Comedy. 
 
 Total Attic period. 
 
 Alexandrian poetry. 
 Alexandrian prose.... 
 
 Total Alexandrian period.... 
 
 Roman period, exc. eccl. and 
 teclinical 
 
 Early Byzantine period (330- 
 622 A. D.), secular lit 
 
 Roman and Early Byz. peri 
 od?, eccl. lit 
 
 Later Byzantine lit 
 
 Anth. Pal 
 
 Oracles 
 
 Inscriptions 
 
 Math, and Astron 
 
 Medical lit 
 
 Rhetores, grammarians, scho 
 lia, lexica, etc 
 
 Total Byzantine, eccl., tech- 
 nical lit., etc 
 
 Uncertain. 
 
 Grand total 352 1631 (10) 15 883 
 
 78 
 
 17 
 
 66 
 
 15 
 
 57 
 13 
 3 
 2 
 3 
 3 
 14 
 
 34 
 
 144 
 
 402 
 
 113 
 
 214 
 
 50 
 
 250 
 62 
 19 
 
 3 
 12 
 
 7 
 64 
 
 216 
 
 683 
 
 (1) 
 
 (i: 
 
 (2) 
 
 (1) 
 
 (3) 
 
 «/ 
 
 264 
 
 70 
 
 83 
 
 27 
 
 51 
 
 17 
 
 24 
 
 6 
 
 3 
 
 3 
 
 15 
 
 107 
 
 253 
 
 27 
 
 20 
 
 95 
 
 2 
 
 177 
 
 3 
 1 
 
 37 
 6 
 1 
 3 
 2 
 
 34 
 5 
 
 193 
 
 4 
 37 
 
 41 
 
 81 
 
 190 
 
 561 
 
 13 
 
 49 
 
 23 
 
 2 
 
 13 
 
 25 
 
 38 
 
 94 
 
 11 
 
 70 
 
 25 
 25 
 
 28 
 
 35 
 9 
 1 
 
 3 
 1 
 6 
 
 43 
 
 101 
 
 2 
 
 257 
 
 10 
 
 Tot. 
 
 9 
 
 5 
 
 137 
 
 57 
 
 10 
 
 20 
 
 7 
 
 167 
 
 19 
 
 1081 
 
 79 
 204 
 
 283 
 
 518 
 
 128 
 
 508 
 146 
 53 
 16 
 28 
 20 
 112 
 
 511 
 
 1522 
 
 22 
 
 4029 
 
The Negative Compounds in Greek. 
 
 57 
 
 TABLE II (Summary of Table I). 
 
 Literature before 500 b. c... 
 Attic literature after 500 B.C. 
 
 Alexandrian literature 
 
 Literature of Roman period, 
 
 exc. eccl. and technical 
 
 Byzantine, eccl. and technical 
 
 lit 
 
 Uncertain 
 
 Grand total 
 
 45 
 
 78 
 17 
 
 210 
 402 
 113 
 
 (6) 
 (1) 
 
 2 
 
 7 
 1 
 
 209 
 
 264 
 
 70 
 
 26 
 
 27 
 7 
 
 53 
 
 193 
 
 41 
 
 6 
 15 
 
 7 
 
 18 
 
 23 
 
 2 
 
 31 
 
 70 
 25 
 
 3 
 
 2 
 
 66 
 
 214 
 
 
 3 
 
 83 
 
 20 
 
 81 
 
 8 
 
 13 
 
 28 
 
 2 
 
 144 
 2 
 
 683 
 9 
 
 (3) 
 
 2 
 
 253 
 4 
 
 95 
 2 
 
 190 
 3 
 
 13 
 
 38 
 
 101 
 2 
 
 3 
 
 10 
 
 352 
 
 1631 
 
 (10) 
 
 15 
 
 883 
 
 177 
 
 561 
 
 49 
 
 94 
 
 257 
 
 603 
 
 1081 
 
 283 
 
 518 
 
 1522 
 22 
 
 4029 
 
 TABLE III. 
 
 Showing percentage of new negative words by classes in various 
 groups of authors. 
 
 Indo-European (?), F'ick.... 
 
 Early Epic and Lyric 
 
 Frags, early hist., philos 
 
 and com 
 
 Aesch., Find, and Bacchyli- 
 
 des 
 
 Before 500 b. c 
 
 Attic philos. and technical 
 
 lit., exc. Plat, and Aristot. 
 Soph., Eur. and tragic frags. 
 
 Herodotus 
 
 Hippocrates 
 
 Historians, lesser and lost 
 Comedy, Old, Middle and 
 
 New 
 
 Thucydides 
 
 Orators 
 
 Plato 
 
 Xenophon , 
 
 Aristotle 
 
 Attic lit. after 500 b. c 
 
 Alexandrian lit 
 
 Lit. of Roman period, exc. 
 
 eccl. and technical 
 
 Byzantine and ecclesiasti- 
 cal 
 
 Anth. Pal 
 
 Oracles 
 
 Inscriptions 
 
 MatheiDatical and n^edical 
 
 lit 
 
 Rhetores, grammarians, 
 
 scholia, lexica, etc 
 
 All the literature 
 
 .100 
 075 
 
 .080 
 .075 
 
 .079 
 .080 
 .113 
 .06 
 
 ,054 
 ,104 
 .046 
 058 
 
 ,053 
 090 
 072 
 060 
 
 108 
 
 131 
 
 057 
 125 
 107 
 
 129 
 
 067 
 087 
 
 .400 
 .321 
 
 .550 
 
 .373 
 .348 
 
 .526 
 .350 
 .245 
 
 .227 
 .286 
 
 ,306 
 ,583 
 ,483 
 ,372 
 526 
 413 
 372 
 399 
 
 459 
 
 414 
 
 358 
 ,188 
 429 
 
 538 
 
 423 
 4051.004 
 
 .005 
 
 ,003 
 
 .020 
 .148 
 
 .009 
 
 ,011 
 .007 
 
 .006 
 .004 
 
 .004 
 
 004 
 
 ^f 
 
 .500 
 .375 
 
 .150 
 
 .316 
 .347 
 
 053 
 
 .398 
 .264 
 .280 
 .286 
 
 252 
 
 .125 
 138 
 .197 
 193 
 180 
 244 
 247 
 
 .135 
 
 .140 
 453 
 375 
 107 
 
 .024 
 
 .080 
 ,043 
 
 ,053 
 ,035 
 
 007 
 
 099 
 
 Oil 
 007 
 018 
 012 
 025 
 025 
 
 047 
 
 087 
 057 
 188 
 036 
 
 .078 
 
 ,200 
 
 ,013 
 
 ,099 .005 
 088 .010 
 
 136.045 
 
 209.043 
 2191.044 
 
 ,184 
 
 
 
 245 
 
 267 
 
 286 
 
 .171 
 .063 
 184 
 270 
 105 
 2>)4 
 179 
 .145 
 
 .151 
 
 .146 
 .057 
 .125 
 .179 
 
 .091 
 
 .108 
 .139 
 
 019 
 
 ,013 
 
 ,027 
 
 034 
 007 
 
 024 
 .014 
 .025 
 
 Oil 
 
 .008 
 
 016 
 .012 
 
 .038 .062 
 
 ,019 
 ,030 
 
 ,026 
 009 
 019 
 
 ,018 
 .083 
 ,057 
 ,029 
 ,035 
 012 
 021 
 007 
 
 021 
 
 015 
 
 036 
 
 008 
 
 049 
 023 
 
 .100 
 
 .028 
 .051 
 
 .079 
 .058 
 .094 
 .013 
 
 054 
 042 
 034 
 051 
 070 
 .066 
 .065 
 .088 
 
 .062 
 
 .051 
 .019 
 
 .107 
 
 .053 
 
 .084 
 .064 
 
 ,008 
 
 .005 
 
 .007 
 
 009 
 
 002 
 
 002 
 004 
 
 002 
 
 002 
 
 = 1.000 
 
58 The Negative Compounds in Greek. 
 
 The total number of compounds, not including derivatives, for 
 all the literature is 3058, while for the Sanskrit, Knauer gives 
 only 1475, and for Latin, Vicol counts only 846 true compounds 
 (including the mutata). As far then as the mere number of words 
 is concerned, it is not true, as Froehde says, 1. c. p, 214, that the 
 use of the negative prefix is in Greek more limited than in the 
 other languages which possess it. 
 
 15 per cent, of all the compounds given in the lexicon appear in 
 the literature before 500 b. c. The Attic literature adds 26.8 per 
 cent, and the Alexandrian 7 per cent., making a total of 48.8 per 
 cent, before the Roman period. The Roman period, excluding 
 ecclesiastical and technical literature, adds 12.9 per ceiit. and the 
 Byzantine, ecclesiastical and technical literature 37.8 per cent. Un- 
 certain are .5 per cent. It will be seen that, as in Latin, a very 
 large proportion of the negative words belong to the post-classical 
 period. Thus it would seem that the negative prefix was one of 
 those elements of language, both in Greek and in Latin, which 
 became much more productive as the making of the literature 
 passed more and more into the hands of those who were not born 
 to the traditions of the classical speech. Negative terms, for ex- 
 ample, are quite prominent in the list of Greek words used by 
 Cicero. 
 
 From Table I it may be seen that the only authors or groups 
 of authors of the classical period in whose writings 100 or more 
 of these negative words appear for the first time are : Homer, 
 Aeschylus, Sophocles, Euripides, Hippocrates, Plato, Aristotle. 
 In the later language the new words appear very largely in the 
 departments of medicine and philology. On the other hand the 
 conservative vocabulary of the orators, with all the bulk of this 
 department of the literature, furnishes only 87 new words altogether, 
 and no one orator as many as 20 except Antiphon (22) and Demos- 
 thenes (20). The number in the latter case is to be explained as 
 due largely to the great bulk of the extant orations. It is signifi- 
 cant, however, that Antiphon furnishes the largest number of new 
 negative words of any of the orators. The fact that he is the first 
 of the orators may be off-set by the small bulk of his extant ora- 
 tions. But Antiphon's elevated style, the sternness of his subject, 
 homicide, his religious, almost Aeschylean tone, his fondness for 
 
( UNIVL 
 
 The Negative Compounds in Gh-eek. 59 
 
 antithesis and parallelism, tlie fullness of significance, which he is 
 wont to compress into single words, all contribute to explain his 
 proneness to employ the negative compounds and their derivatives, 
 a proneness still more apparent after a glance at a few pages of his 
 text. Cf. A. J. P., XVI, 525. 
 
 In the search for new words a proportionately larger number 
 are of course to be expected in an earlier than in a later portion of 
 the literature. It is clear, however, that the classical literature 
 after Homer gained a very large proportion of its negative com- 
 pounds from the poetic sphere of tragedy and the more or less 
 technical works of Hippocrates, Plato and Aristotle. Very few 
 new compounds are to be cited from the fourth century. In later 
 Greek important accessions to the number of new words come from 
 the Anthology, Cicero, Dionysius of Halicarnassus, Diodorus, 
 Lucian, and Cyrill of Alexandria. Many are very late and are to 
 be found only in Hesychius, Eustathius, Tzetzes and the Etymo- 
 logicum Magnum ; many are cited simply as ecclesiastical or 
 Byzantine. 
 
 Table III gives the percentage of the whole number appearing 
 in any author or group of authors which is formed by any one 
 class. A glance at the first column in Table III shows that only 
 a small proportion of the total number of negative words is formed 
 by the compounds of prefix + adjective, especially in the classical 
 language. Herodotus, Thucydides and Aristotle show a slight 
 advance on the rest, while the proportion is lowest of all in the 
 orators. It becomes larger again in late Greek. 
 
 Next in interest is a comparison of the second and fourth 
 columns, classes 6 and ef. The great majority of the negative 
 compounds in Homer are either verbal compounds of the type, 
 aSdfx,acrTo<;, or secondary possessives (mutata) of the type, a0yX\o<?. 
 We observe relatively a quite steady advance of the later language 
 beyond the earlier in the making of the -to? compounds and a 
 still more striking falling off in the introduction of new mutata, 
 compounds whose second member is a noun or a verbal stem. 
 These changes are most noticeable in prose. Thus in Homer and 
 early lyric poetry the mutata of class efare the more numerous ; but 
 the relation is reversed in Hesiod, Aeschylus, Pindar and Bac- 
 chylides, while in the fragments of early philosophy, history and 
 
60 The Negative Compounds in Greek. 
 
 comedy the shift is very much more marked. For the whole 
 period before the fifth century the proportions are about equal. 
 In tragedy after Aeschylus the mutata again predominate, also in 
 the Ionic Herodotus and Hippocrates, but nowhere else within the 
 range of the table except in the very poetic spheres of the Anthology 
 and the Oracles. The mutata are slightly the more numerous in 
 Alexandrian poetry (not shown in Table III, but see Table I). 
 In comedy, Plato, Xenophon and Aristotle the -To<i compounds 
 predominate, and in Thucydides and the orators the percentages 
 of mutata are away down to about the level of late Greek. In 
 the fragments of philosophy (largely those attributed to Democri- 
 tus) and in inscriptions the proportion of the mutata is lowest. 
 The large use then of the mutata in preference to the verbals in 
 -ro<; is most clearly poetic and perhaps, though of this we do not 
 feel so certain, it is Ionic. Many of the mutata belong only to 
 the higher ranges of literature and never gained currency in prose. 
 It seems to have been distinctly an older habit of the language to 
 coin a mutatum from a noun than to make a new verbal in -T09 
 from the verb which corresponded to the noun. Yet as early as 
 Homer we find a number of pairs like the following : airvpof;, 
 airvpcoTOf;, areXr]<i, driXea-Ta, dTifjbO<;, dTi./xTjTO<i. The slight rise 
 in the proportion of mutata in the philological literature is largely 
 due to the citation of rare words in the lexica. In the later 
 literature the list of -ra compounds is swelled by the numerous 
 compounds with verbals having already a prepositional prefix. 
 The solitary example in Homer is diTpoTiixaaTo<i ; Hesiod adds 
 dve'7ri^earo<i and dveTrippeKTOf;. 
 
 In the column containing the percentage of compounds of class g 
 (as dvaiS6-fji,axo<i Bacchylides, etc.) the highest percentages are in 
 Aeschylus, Pindar and Bacchylides, comedy, ecclesiastical and 
 Byzantine writers, and most of all in the Oracles, which last speaks 
 especially for the high poetic tone of these imposing compounds. 
 Those in comedy may be credited to mere comic exuberance in 
 word-formation or to parody and paratragedy, those in ecclesiastical 
 and Byzantine writers to the general tendency of late Greek toward 
 making common the loftier diction of earlier times. 
 
The Negative Compounds in Greek. 61 
 
 The column headed h contains tlie percentages of tlie new deriva- 
 tive nouns, for the most part abstract. These next to the -to<; 
 compounds and the mutata are the most numerous of the negative 
 words ; see foot of Table III. Indeed, in the orators, Plato and 
 Aristotle, and in much of the late prose they rank next to the -T09 
 compounds and exceed the mutata in number. Disregarding the 
 fragments of the lesser historians — the numbers are so small that 
 the proportions may be only accidental — the highest percentage is 
 Plato's; the percentages of Herodotus and Hippocrates follow 
 close behind. The small number of new negative abstracts in 
 Thucydides is noticeable, but then Thucydides could say rj ov 
 8idXv(TL<i, etc.; cf. p. 31. The percentages are low in the poets. 
 This may have some relation to the use of the neuter of adjectives 
 as abstract nouns, which is more characteristic of the older language 
 (Wil.-Moell. on Eur. H. F. 75). On the relation between the use 
 of negative abstract nouns and the negative articular infinitive no 
 trustworthy conclusion can be drawn from this table. For the 
 orators the numbers of new negative abstracts are too small to 
 afford any reliable basis for inference. Better results would doubt- 
 less be gained by studying the use of all the negative abstracts in 
 the indices and texts of the Attic writers. 
 
 The derivative adverbs with the negative prefix, outside of those 
 regularly formed in -&)«?, are mostly those in -i (-el) and are shown 
 by the tables not to have been at all numerous, only 94 in all. 
 Yet a few new ones appear in every period of the literature (see 
 Tables I and II), notably 23 in good Attic, and 25 in the late 
 grammarians and lexicographers. 
 
 The VT}- compounds. Leaving out of account vrjiria and its 
 derivatives as being doubtful in etymology, the compounds in vtj- 
 make their appearance in the literature as follows : Horn. 13, Hes. 
 5, lyric poets 2 + 2 doubtful, Pind. 2, Emped. 1, Soph. 1, Old 
 Comedy 1, Com. Anon. 1, Hdt. 1, Hippocr. 1+1 doubtful, Andoc. 
 1, Alexandrian poetry 7, late Greek 13. 
 
62 
 
 Tlie Negative Compounds in Greek. 
 
 Of the 52 compounds and derivative words in v-q- only 13 ap- 
 pear first in prose, and only one, vrjiroivel (Andoc, after Homeric 
 vrj-TToivo'i), first in Attic prose. Considering the poetic and archaic 
 sphere of these vrj- compounds, the predominance of the mutata 
 {ef) was to be expected. This is very marked in the Early and 
 Attic literature but does not hold true for the later period. 
 
 
 a 
 
 b 
 
 ef 
 
 g 
 
 h 
 
 i 
 
 k 
 
 I 
 
 Tot. 
 
 Early and Attic literature 
 
 1 
 
 3 
 
 7 
 
 14 
 4 
 
 2 
 
 7 
 4 
 
 1 
 
 1 
 
 5 
 2 
 
 3'?, 
 
 Alexandrian and later literature... 
 
 20 
 
 Total 
 
 1 
 
 10 
 
 18 
 
 3 
 
 11 
 
 1 
 
 1 
 
 7 
 
 52 
 
 
 
LIFE. 
 
 HolHster Adelbert Hamilton was born in Wayne Co., New 
 York, Jan. 14, 1870. His preliminary education was received in 
 the public schools of Rochester, New York. In the autumn of 
 1888 he entered the University of Rochester from which he was 
 graduated with the degree of A. B. in June, 1892. He takes 
 this opportunity of expressing his sense of indebtedness to the 
 instruction in the classics and to personal influence of Professors 
 Geo. M. Forbes and H. F. Burton. During the years 1892-94 
 he was vice-principal of the high school at Waterloo, New 
 York, and during the years 1894-96 he was instructor in Latin 
 and Greek at the University of Rochester. During the years 
 1896-99 he was a graduate student at the Johns Hopkins Uni- 
 versity, making Greek his principal and Latin and Sanskrit 
 his subordinate subjects. Here he was successively appointed 
 university scholar and fellow in Greek, and in 1899 he received 
 the degree of doctor of philosophy. He attended the lectures of 
 Professors Gildersleeve, Warren, Bloomfield, K. F. Smith and 
 Miller, to all of whom he desires to make grateful acknowledg- 
 ment, and especially to Professor Gildersleeve, who by his kindly 
 interest, by the inspiration of his teaching and by the influence 
 of his own personality has imposed a debt of lasting gratitude. 
 
 63 
 
RETURN 
 TO— ^ 
 
 MAIN CIRCULATION 
 
 ALL BOOKS ARE SUBJECT TO RECALL 
 RENEW BOOKS BY CALLING 642-3405 
 
 DUE AS STAMPED BELOW 
 
 DEC 09 1996 
 
 RECEIVED 
 ^m 2 3 1997 
 
 8KCUUTI0N DEPT 
 
 FORM NO. DD6 
 
 UNIVERSITY OF CALIFORNIA, BERKELEY 
 BERKELEY, CA 94720 
 
U.C.BERKELEY LIBRARIES 
 
 CD^b^flflfl7fl 
 
 A