γ. f ^ f Ύ. REESE tr RARY UNIVERSITY OF CALIFORNIA, -fc^ Λί . 7 rvr, Received 'C^j y^/, ^y Accessions N,.A^^yj~f- Shelf No. '/>f-3/cL OK β PRONUNCIATION ANCIENT GEEEK, Eonlion: C. J. CLAY and SONS, CAMBRIDGE UNIVERSITY PRESS WAREHOUSE, AVE MARIA LANE. CAMBRIDGE : DEIGHTON, BELL, ΑΝΌ CO. LEIPZIG : F. A. BROCKHAUS. PEONUNCIATION OP ANCIENT GEEEK TRANSLATED FROM THE THIRD GERMAN EDITION OF DR BLASS WITH THE AUTHOR'S SANCTION W. J. PURTON, B.A. PEMBEOKK COLLEGE. ' Of THE 'r \, univeesity) CAMBRIDGE AT THE UNIVERSITY PRESS 1890 [All Rhihtx reserved.] Camliriiigc : PRINTED BY C. J. CLAY, M.A. AND SONS AT THK UNIVERSITY PRESS. PREFACE. rr^HE present translation of Dr Blass' work on ancient Greek -L pronunciation represents the third and latest German edition, and the translator has throughout its production had the advantage of the advice and help of the author, who kindly undertook to read all the proof-sheets. A few words are necessary touching the system of trans- literation adopted by the translator. As regards the consonants little difficulty presented itself He was able here simply to adopt the transliteration used by the author, only making the necessary changes of y for j, ch for tsch, j for dzh, and so on, according to the different values of the letters in German and English. With regard to the vowel sounds however his course was not so plain. As, in spite of the labours of Mr Sweet and Mr Ellis, no artificial system of phonetic representation has obtained sufficient acceptance to be really familiar to English scholars, he has resolved to retain the vowels with what may roughly be called their continental values. The alternative plan, namely to represent them by their approximate English equivalents, presented great difficulties. To take an instance : to represent the continental long ι sound by ee, not to speak of its cumbrousness, labours under the additional disadvantage that the short sound must still be represented by i, thus obscuring the identity of the two sounds. Again Dr Blass has in the case of the e and ο sounds adopted diacritic marks to distinguish the open and closed sounds, and it therefore seemed especially desirable here to vi PREFACE. retain simple symbols. In all cases therefore where the Greek vowels are represented by Roman letters, these must be under- stood to have their continental sound, that is to say roughly speaking : — d must be pronounced as h\ father. η „ as in vian. 7 ,, as in second syllable of quinine. / „ as in first syllable of quinine. e „ as in fete*. e „ as in ebb. δ „ as in note*. 6 ,, as in not. a ,, as in lute. a „ as in i)ut. The translator has already mentioned his indebtedness to the author for his kindness in reading the proof-sheets ; he has also to express his gi-atitude to Mr Κ A. Neil, Fellow of Pembroke College, for similar help. * It ouglit to be remarked that these two sounds in English contain a diphthongic element which phoneticians call a glide — in the case of e an /-glide, in the case of ό a ?i-glide — which gives tliem a decidedly different sound to that heard on the Continent. The nature of this difference may be suggested by saying that in the case of ϋ the continental sound often tends in the direction of our aio in saw etc. June, 1890. -\DDITIONS AND COKRECTIONS. P. 37, note δ. δινάρια for Αινάρια. P. 47 (text) 1. 12 after \ψτουρ-γία add " /cXet's for KXfis " and substitute for end of sentence "in which cases even inscriptions shew et and the grammarians designate ηί as old Attic." P. 52 (text) 1. 12, έπίτήδΐο^ for first έπιτήοειοί. P. 72 (text) 1. 7, i; for o. P. 77 (text) 1. 1-1, a-no-si-ya for a-no-si-ja. P. 118 (text) L 23 after Auramazda add Mafaios, Μα^άκ?;?, Mazdai, Mazdak. Section 1. The theoretical and pr-actical sides of the subject. The investigation of the pronnnciation of Ancient Greek may be considered from the point of view of theory and again from that of practice. In the former case its object is the phonetic value, which the Greek letters and combinations of letters had in the living ancient speech ; in the latter the point nnder discussion is, what phonetic value are we to give to those letters and combinations in reading and teaching Ancient Greek ? The answer to the question of theory will influence the answer to the question of practice ; not however exclusively, for in the case of the latter appropriateness and feasibility must be taken into consideration. I intend in the present work to enter but little into the practical question. For the Germans are not in need of reform either in the case of Greek or in that of Latin in the same degree as the English, and even if they were, the welfare of Greek and Latin instruction does not depend on the abolition of this misusage and this only. Our object is contact with the spirit of classical antiquity ; but for the purpose of such a contact it is by no means a hindrance to me, if I say something like TsUserd, while the actual man called himself Ktkerd. And there is according to my conviction nothing in our pronunciation of Greek so positively and stupidly wrong as the ordinary pronunciation of Latin c. If however anyone feels himself bound in the interest of what we may call a more workmanlike prosecution of classical studies to pay scrupulous regard to such things, and can in so doing guard against the reproach of straining at gnats and swallowing P. 1 2 THE PRONUNCIATION OF camels, for such a man I have of course nothing but praise. But the attempts, constantly repeated here as well as in other countries, to introduce in practice the modern Greek pronuncia- tion for ancient Greek, must be withstood in view of not only practical but also theoretical and scientific interests. For even the champions of the modern Greek pronunciation appeal not to a practical superiority, Λvhich it obviously docs not possess, but to a supposed scientific accuracy. A short history of the whole contest from the beginning of Greek studies in the West may conveniently be introduced here. Section 2. History of the contest about the pronunciation of Ancient Greek. The knowledge and study of ancient Greek came to the countries of the West towards the close of the Middle Ages through the medium of Byzantine scholars, λυΙιο naturally brought with them and introduced theii' own pronunciation, that is to say that current among the Greeks of their day. As however these studies were prosecuted more indepen- dently and thoroughly in the countries of the West, there arose against the traditional pronunciation a reaction Λvhich started with some support in the fact, that quite a different pro- nunciation was customary in the case of the sprinkling of Greek words in Latin, such as ecclesia, ethice, alpJiabetuni. Moreover the Byzantine pronunciation deviated so mdely from the writing and confused so many sounds, that it of necessity not only appeared unpractical but also called forth doubts as to its originality. Finally, many passages in ancient authors spoke so plainly for a different ulterior ])r()nunciation, that the fact of an alteration having taken place could not by any possibility escape classical scholars. Accordingly so early as Aldus ]\lanutius we have his little 7r(tpepyop\ whit-h has appeared in many forms in » Akli Manutii de vitiuta vocalium first printed (1Γ)12) in the appendix to rt diphthoiitjuruin prolatione irapfpyov, tlie Aldino edit, of tlie 'Επίτομη των ANCIENT GREEK. 3 print, relating to the diphthongs, η and v, and some con- sonants. A short treatise on the pronunciation of all the letters was furnished by Jacobus Ceratinus\ professor at Lou vain, who died in 1530. But the most celebrated of these early com- batants was the renowned Desiderius Erasmus, in a dialogue de recta Latini Graecique sermonis promuiciatione^, Avhich appeared first at Basel in 1528. Although the author was pleased to clothe his subject in the facetious, or more correctly the rather insipid, dress of a dialogue between a lion and a bear, nevertheless his treatment is so thorough and comprehensive, that there can be no doubt whatever of his scientific seriousness. The fact is not altered by our knoAvledge that Erasmus himself continued to use the traditional pronunciation^: a reformer he certainly was not. A greater stir was made by some English scholars at Cambridge, John Cheke and Thomas Smith, moving the condign wrath of Stephen Gardiner, bishop of \Vinchester, at that time Chancellor of the University, whom we know in Church History also as a fierce persecutor of heretics. In 1542 he issued an edict for his University, in which e.g. it was categorically forbidden to distinguish ai from e, et and 01 firom c in pronunciation, under penalty of expulsion from the Senate, exclusion from the attainment of a degree, rustication for students, and domestic chastisement for boys. Choke's corre- spondence with the Bishop on pronunciation appeared at Basel o/cTtb ToO λόγου ytteptD:/ by Const. Lascaris atione comnientarios reliquerunt, ed. (as E. Meister shews, z. griech. Sigeb. Havercampus, Lugd. Bat. 1736, Dialektologie, Progr. Nikolaigymn. p. 355 — 376. Title, de sono litter- Leipzig, 1883, p. 13), then repeated in arum, praesertim Graecarum. It is the Cologne pirated reprint of the dedicated to Erasmus, but does not Erasmian Dialogue (1529), also in the make the smallest reference to his Ortliographiae ratio Aldi (published labours on this subject, so that the by his grandson, 1566). priority is e\-ident. 1 His proper name was Teyng, - Eeprinted 1530, pirated 1529 at born at Hoorn in Holland, died Cologne (vid. supra) ; see further in 1530. The treatise was printed at Havercamp's StjUoge altera scriptormn Antwerp 1527 (vid. E.Lohmeyer,P/io?!. qui, etc. (Lugd. Bat. 17-10), p. 1 — 180. Stud. I. 183), reprinted in the above- ^ g_ Vossius, Aristarch. 1. c. 28 mentioned Cologne piracy of Erasmus, (0pp. vol. 11. p. 36) ; EUissen, GiJttinger also in Sylloge scriptorum, qui de Philologenversammlung (1853), p. 108 linguae Graecae vera et recta pronunci- ff. 1—2 4 THE Γ no Ν UNCI AT ION OF in 1555, published by Coelius Secundus Curio*; the Bishop uses for the most part the weapon of authority, Cheke on the other hand that of respectable learning and intelligent critical discussion. He was seconded by his friend Thomas Smith, whose missive to the Bishop is dated in the year of the edict I At this point the movement began also among the French scholars, among whom Petrus Ramus and Dionysius Lambinus^ must be mentioned as the first combatants. Before the century had closed, the victory of the Erasmians was decided in all the chief centres of classical philology. A pretty thorough exposition Avas Avritten by the well-known reformer Theodor Beza, de germana pronunciatione Graecae linguae*. He as well as Cheke was made use of in a somewhat questionable manner by the Dutchman Adolph van Metkerke (Mekerchus) in his work de linguae graecae veteri pronuntiatione", Bruges 1565, the most complete confirmation of the Erasmian system that had been written. Finally in 1578 the famous Henr. Stephanus entered the lists in the same cause, Apologeticus pro veteri ac germana linguae Graecae p)iOnuntiatione^. Stephanus is already able to say, that in France, England, the Netherlands and elsewhere the reformed pronunciation was eagerly learnt and practised. In this there is nothing to cause surprise ; for not only had the Erasmians, on the Avhole, the better cause, but the opposite party Λvere very weakly represented. Joh. Reuchlin, from whom the pronunciation of the latter takes its name in Germany, gave the impulse to it only in so far as he was the founder of Greek studies in that country ; for although he used and taught the modern Greek pronunciation, he could have no object in establishing and defending it, inasmuch as he never lived to see Erasmus' treatise. Bishop Gardiner cannot be reckoned a scientific combatant ; and the short treatise directed 1 rrinted in Hav. ii. p. 181 — 4G8 contest on pronunciation is learnt (the Chancellor's edict p. 205—207). from H. Stephanus in the work to be 2 Hav. p. 401)— 574. AccorilinR to cited below (p. 391 f.). Hav. 's Praefatio this was published in * Printed in Hav.'s first SyUotje, 1508 by Hob. Stephanus. p. 305—352, appeared (ace. to EUissen) 3 Both directly or indirectly victims 1554. of the massacre of St Bartholomew * Hav. p. 1—170. (1572). Their participation in the " Id. p. 377 — 470. ANCIENT GREEK. against Mekerchus by the Englishinau Gregory Martin' (died 1582) was of trifling importance. Accordingly the Eras- mian pronnnciation prevailed thronghont the West, and the counter-efforts of Erasmus Schmidt of Wittenberg (1560 — 1637') and of Job. Rud. Wetstein of Basel (end of the 17th century^) failed to make any alteration in this result. There Avas now a lull in the contest, and the interest in the question waned, until the revival of grammatical studies in our century gave it new life. All our great grammarians have entered the arena either entirely or essentially on the side of the Erasmian pronunciation, e. g. G. Hermann, August Matthiae, Phil. Butt- mann, R. Kiibner, K. W. Kruger, G. Curtius*. Seyffarth and Liscovius, who published special works on the subject in 1824 and 1825 respectively^, affect an independent attitude towards both schools, and arrive at mixed results. About the same time the Dane S. N. J. Bloch", who was refuted by his country- man R. T. F. Henrichsen in a justly valued book, was a zealous champion of the modern Greek pronunciation. The matter was next treated of in the Gottingen and in the Frankfort Philologenversammlimg in the years 1852 and 1861, EUissen supporting the modern Greek pronunciation and Bursian a mixture'. The hottest and most persistent combatants are the Greeks 1 In the Sijll. altera p. 575—622. ^ Id. p. 631—674. ^ Joh. Rod. WeMcnii pro graeca et (jemdna linguae Graecaepronunciatione orationes apologeticae, edition. Basileae 1686. ^ G. Curtius, Erlauter. p. 15 ff., and more thoroughly Ztschr. f. d. osterr. Gijmn. 1852, p. 1 ff. ^ Seyiiarth, de sonis Utterariim gr. turn genuinis turn adoptivis, Leipz. 1824; Karl Fr. Sal. Liscovius iiber die Aus- sprache des Griechischen, Leipz. 1825. β S. N. J. Bloch, Revision der Lehre von der Ausspr. des Altgr., Altona and Leipz. 1826; additions in Seebode's Archiv, 1827 and 1829; also three Copenhagen Schiil-Programme, 1829— 1831 ; Zweite Beleuchtung der Mat- thiae'schen Kritik, die Ausspr. des Altgr. betr., Altona 1832. E. J. F. Henrichsen, iiber die Neugriechischen Oder sogen. Reuchlinische Aussprache d. Hellen. Sprache, iibersetzt von P. Friedrichsen, Parchim and Ludwigs- lust 1839. '' Verhandl. der xiii. Vers, dent- scher Philologen, Gott. 1853, τρ. 106 — 144; id. d. xx. Vers. Leipzig, 1863, p. 183 — 195. EUissen's treatise is valu- able on account of its thorough treat- ment both of the history of the Greek nation and the history of the contest over the pronunciation : an index of the Uterature of the subject is given p. 137 f. note. 6 THE PRONUNCIATION OF themselves, who, now that the German pronunciation has been adopted even in Russia, are in fiict the only people who still cherish itacism. Among them however there are not wanting enlightened investigators of language, who do not refuse to take a scientific view even of this subject. Section 3. Genuine and counterfeit Erasmian pyincvple. It is however worthy of remark, that the Erasmian pro- nunciation, in the actual form which it has taken in various countries, is by no means identical with that theoretically developed by Erasmus and his adherents. In reality the axiom which has been more or less followed is this, that the symbols and combinations of symbols are to be pronounced as the corre- sponding symbols in the various languages ; but this is an axiom of convenience not of science. The genuine teaching of the Erasmians is on the contrary really scientific ; they endeavoured, independently of the modern Greek tradition, to recover the ancient pronunciation from direct evidences, from transcripts into and out of foreign languages, and from linguistic precedents. They also, as was right and fair, called in to their help the analogy of modern languages ; Erasmus heard the sound of ai, i.e. a + L•, in the German Kaiser, that of oi, i.e. o + i, in the nwi toi soi of certain Frenchmen, while Beza expresses the pro- nunciation of these words by moae toae soae (triphthongal), and recognizes the genuine ol {o + i) in soin and besoin. The train of thought then is this, various modes of writing such as i, η, υ, ec, ol, vl cannot possibly from the beginning have stood for the same sound, but rather, when the Avriting was diph- thongal, the pronunciation also was diphthongal, i.e. the mem- bers of the diphthong were pronounced distinctly but united into one syllable, as they are heard in numerous instances in living languages. But finally in practice only so much, as was convenient, was retained from those scholars' scientific discovery, namely the freedom from modern Greek tradition and the cni- pluyincnt of West Eurojican analogies, the most obvious being ANCIENT GREEK. J of course unconsciously adopted. Accordingly the Germans pronounce ξ" as ts, οϊνου<ί like evvov<;, both syllables of elvat with the same vowel sound, and call this the Erasmian pronunciation, although the ancient Erasmians requh'ed the pronunciations ds for ζ, €-{■ V for ev, e + t for ei. Section 4. Relation of Sound and Writing. However, as I have said before, I shall here disregard prac- tice and keep to scientific discovery ; for as such, and indeed as a very great discovery, I regard the achievement of Erasmus and his predecessors and followers. The theoretic and scientific significance of these researches can indeed be far more easily undervalued than overvalued. The history of Greek pronuncia- tion is the history of that phonetic change, which took jDlaco in the language so to speak covertly, but which is on that account by no means less real and important than the alteration, Λvhich became apparent in the writing. It is indeed the case with all languages, that the writing does not keep pace with the changes of sound, but remains more or less in the lurch. Writ- ing is no conscious translation of sound into symbols, but, after this has been done once and originally, habit has stepped in, and one race hands on this habit to the other. Hence arises the well-known variation between pronunciation and writing in modern languages, which is nowhere greater than in English. Not that the present English orthograjDhy is the same as that under Henry the Eighth : but we should be entii'ely misled, if we were to estimate the deviation of the language of that period from that of the present day by the deviation in the writing. The matter is well known to and treated of by specialists^ ; that however need not prevent us citing here the results of the above- mentioned treatises of Cheke, Smith and others. They tran- scribe Engl. 77iane μάν, gate ηατ ; Erasmus ascribes the pro- 1 H. Sweet, History of English A. Ellis on E. English Pr., ib. extra Sounds, Transactions of the Philol. vol. v. 1869—1870, 1809—1878, 1871, Society, 1873—1874, p. 461 (517). 1875. 8 THE nWNUNCIATION OF nunciation of a as ae to the Scotch. Further, mean μην, meat μητ, heat ήτ, 'wheat ούητ ; the η signifies the open sound, the closed sound in vie, bee being called e italicum. The Scotch according to Erasmus pronounced this e as i. Btr bite, φΐΧ file, βΐ It buy it. Των gone, yo ο ν go on. Αυκ Χυτ ρββυκ duke lute rebuke, the long French u, which was also attested for 7'ude, rue ; the corresponding short sound, says Smith, is heard more frequently in central than in southern England, but Avould be general in ruddy, bloody (written at that time blady), muddy. Latin u is heard according to them in bow the verb βου, gown jovp,foul φουΧ; in bow the substantive, boiul etc., the sound of the Greek ωυ (the modern ou). For the diphthong ai, i.e. a + i, way, pay are cited (in these cases however in more cultivated pronunciation more of an ei, in Scotch and north English almost a monophthongal ae was heard), for ei neigJi, for au claiu, for eu few, dew. ϊυ sum up, Λve find, that an ex- traordinary alteration has taken place in the actual language, quite as great as that established for Greek by the Erasmians. French also of that period was pronounced quite differently to what it is at the present day : mute e had its value, the mute final consonants were perceptibly dwelt upon at all events before a pause, in beau Smith heard the Greek diphthong ηυ, Erasmus and Stephanus a triphthong, all three vowels being- heard. So shifting is pronunciation, and so stable writing, juggling away as it does the most important changes. But the enquirer must not allow himself to be juggled with, not even to the extent of regarding what is apparent as more important owing to its transparency than that which conies to pass covertly. But if these sound-changes are not apparent, how can we knoAV anything at all about them and about the earlier sound- stage of Greek i' I might answer at once : in the same way that we do with regard to the earlier sound-stage of English ; for Greek too there is a whole series of similar evidences in ancient authors. But Erasmus was perfectly right in inferring a variety of sound from the application of various symbols, and a diphthongal pronunciation from diphthongal writing. The sim])le and natural rule, write as you speak, has never from the beginning been infringed without special reason. Such a reason ANCIENT GREEK. 9 existed in many instances for the Romance languages in the deference paid to the Latin mother language ; French modes of writing such as corps, doigt, at an earlier period also faict for fait and so on, where the penultimate consonant was always mute, could never have existed but for the Latin corpus, digitus, factum^. For the ai in aimer, faire etc. Erasmus and Beza attest the living dialectal use of the diphthongal pronun- ciation in their time ; eu is according to them universally a diphthong, =e + (Fr.) u, in like manner au {=a + o according to Beza); eau and oi have been already mentioned, and for the latter the original pronunciation as ο + ΐ is guaranteed by the living English voice from voix and choice from choix'\ Similarly English orthography, disregarding the mixture of different systems of sound-notation, has arrived at its present incongruity with the sound through deference to Latin and the permanence given by writing to sounds formerly — but now no longer — really heard. Since then the ancient Greeks were not in a position to pay deference to a previous language in a higher stage of culti- vation, they must consequently have originally striven to bring their writing as near as possible to the sound. As the language underwent further development, it may well have happened both in Attic and in the other dialects that the orthography did not progress evenly ; but this must have consisted much more in what was old not being entirely crowded out by what was new, than in the retention of the old to the absolute exclusion of the new. For a crystallization of orthography can only occur where the word forms have stamped themselves firmly 1 Diez, Gramm. d. roman. Spr. 1^, si una sola enuncietur, velut quaelibet p. 442. ex tribus voealibus?" Modern Pro- ■^ Stephanas, p. 41-1, ed. Haverc, venial still retains diphthongal ai makes the universal statement as re- (faire, paire, maire = pere, mere), «u, e« gards the French: " non solum diph- (Dieu, casteu = chateau) etc. Cp. Diez, thongos et triphthongos, hisque longi- p. 429 ff., who adduces for au from ores reete pronuntiamus ; verum etiam Beza's treatise de francicae UiKjuae nullames voealibus devorantes, indisso- recta pronimtiatione (1584) a somewhat luta voce plane distinguhnus beau, lieu, discrepant testimony to the effect that ioijaux, ioyeifx.... Quotum enim quern- the pronunciation hkeao was Norman, que Gallorum hodie reperias, qui aequo the ordinary pronunciation much like animo ferat μονοφωνίαν suarum diph- o. thongorum et triphthongorum ? Id est. ΙΟ THE PRONUNCIATION OF by much reading and writing ; Avhcrc there is but little reading and writing, as in Greece in the classical period and in western Europe in the Middle Ages, unless the sound is very stable and well defined, the orthography is extremely shifting. Now it is actually the case that in Attica towards the close of the fifth century the entire system was absolutely changed. Here Λν33 the opportunity in those cases, where the living sound had here and there deviated from the writing, to bring them again into harmony. Moreover, since the Athenians and also the other races did not yet possess any grammarians or etymologists to attach importance to a historical mode of writing, the only principle which could have weight was the phonetic. Ac- cordingly it is actually tlie case that on Attic inscriptions of the fourth century the orthography is by no means estab- lished in all points : ret Ttyu-et and τψ τίμήι are written promis- cuously. When in the course of time the Attic dialect ex- tended itself beyond the boundaries of Attica, and became essentially the standard for the κοίνή of Hellenizcd countries, and at the same time habits of composition and literary culture increased to an extraordinary degree, fluctuation in orthography must most certainly have become far less easy. To the Mace- donians, the Egyptians, the Carians and Lydians, and also the Dorians of the Peloponncse, Attic Greek was an acquired tongue, and that in part by means of its literature, so that sound and writing impressed themselves simultaneously. We soon have to add to this the influence of the learned gram- marians. However even at that period the orthography did not yet crystallize : the c of the diphthongs a, r;, ω, which had gradually disappeared in the spoken language, was in the time of Augustus consciously omitted by many in writing also, as Strabo says', -ποΧλοΙ εκβάΧλουσί τό e^orritten tradition, which they despise. The Reuchlinian therefore ought to say efta, ocJito, niffi (nifi) for νύμφη etc., and arrange everything under proper rules the number of which must certainly be very great; otherwise he transgresses at every step his own principle. Finally there is no lack of points, as regards which the testimony of oral tradition is entirely at variance, according to dialects and localities; for example with respect to the pronunciation of κ before e and t (%e, tye, chye, die, tsye, tse = «:e), or that of χ after ρ (k or ch) : Avhere consequently as a matter of fact we have no evidence. This is all emphasized by Psichari, and the necessary inference to be drawn from it is that the Reuchlinian principle neither is nor can be carried out in practice. 14 THE PRONUNCIATION OF Section 5. MetJiod of ascertaininf} the ancient pronunciation. The matter then stands thus ; fur the original sound writing is our evidence, for the present sound (and for this only)^ the living representatives of the nation, and the point to be investigated is, how long the original sound has stood its ground, and when the present sound began. This investigation must be carried on separately for every single sound, for the results may be very various. The sum of these is a piece of sound-history of the Greek language, to be supplemented from the alterations which become apparent in the Avriting, which latter however belong more to the prehistoric than the historic period. Looking at it in this light Ave first see the whole of the significance of the subject, and, it must be confessed, the whole of its difficulty. It is true the general rule, by Avhich to decide, whether a sound at a given time retained its original value or had already passed into another, may simply be taken over from allied fields of enquiry. E.g. the fact that French en in the golden age of old French literature was identical Avith an, is inferred among other proofs from its confusion Avith an which already took place at that period^ ; conversely if such a confusion did not appear, it would be concluded with equal certainty that en still had the e sound. If then in like manner we say with regard to the Greek at ; it was in the Attic period a real double sound, since it is exchanged neither with η nor with e ; this is a mode of reasoning, the justness of which no one would impeach in the domain of any other language. In feet it is quite clear that, if at was identical with e and also η, even in the case of a much more learned people than the ancient Athenians some confusion in \\riting would infallibly have occurred, especially during the course of so many centuries. We have only to notice in comparison, how shifting and un- certain the Latin writing is in the period of the Republic in spite of the exertions made by the grannnarians from an early date to regulate it. Even if we suppose that at was an e tres ' Lucking, Λ. Ultestcn fmnz. Mundnrten, p. lOO ff. ANCIENT GREEK. 1 5 ouvert, while η was an ordinary open e, such a trifling difference as that Avould not long have been adequate to hinder confusion. This then is the first and most general method : investigate up to what period the writing is constant and when it begins to be no longer so. Next we have direct information and descriptions ' in the Avorks of the grammarians, and can also draw inferences indirectly from the grammatical nomenclature and classifications of sounds, from directions as to orthography and so on. Further phonetic transitions Avithin the Avord and especially in the combination of Avords have weight ; for if eVt ω becomes εφ' ω, and καΐ έ'στί becomes καστι, this teaches us something about the value of φ and at, since this fact is utterly irreconcilable with certain values of these symbols. Of great importance too are transcriptions from and into other languages, and here Latin is of primary value for Greek, just as Greek is for Latin. ΚέΧβρες Κικέρων, Cimon Gyrus, are in themselves adequate evidence for the fact, which is established by other considera- tions, that Latin c Avas always k in the classical language ; for no one can doubt that this Avas the value of κ^. In like manner transcriptions such as Athenae, ecclesia, κήνσος, Αουκρήτιος are alone sufiicient proof that η Avas equivalent to e ; for that Latin e was not equivalent to i is doubted by none except those who have given their verdict after having boAved their necks once for all to modern Greek authority. Such people are doubtless skilled to throw doubt on that which is most firmly estab- lished, and give a plausible appearance to that which is most questionable, according as it falls foul of or is at harmony with this authority^ Much light can be obtained for Greek from 1 It is true that in the 16th century pluribusreceptus est, ilium frequeutato. the point was not considered to be - Ellissen, p. 136 : " we do not know settled; Bishop Gardiner prescribes: how the Romans pronounced ;.•{;)Γ(/(•/ΐί' otlicr nictliods of proof the moritorious des Clricch., Leipzig, IHSS. ANCIENT GREEK. 17 sound of ?; as e ; for there are two sorts of e's, the open and the closed. If however I say η was the open e, I ought not to be asked further, which open e ? although, as is well known, the French distinguish three sub-varieties in their language: an ordinary open e, a more open, and a very open one. This is by no means a matter of indifference for harmony and correctness of pronun- ciation : but no one can expect to know anything about such subtleties in the case of a dead language. Lastly there are not merely three open e's, but a numberless series, and the same holds good with regard to the other sounds and combina- tions of sounds ; for instance a diphthong can be spoken with greater or less preponderance of one or the other vowel, without regard to the possible variety in the single elements. I am perfectly convinced, that, if an ancient Athenian were ( to rise from his grave and hear one of us speak Greek, on the basis of the best scientific enquiry and with the most delicate and practised organs, he would think the pronunciation , horribly barbarous. But if he heard a modern Greek, he would not indeed be so loud in his censure, simply because he failed to observe that this is supposed to be his own language. For where, not to mention all the other points of difference, acute and circumflex are not differentiated, and every accented vowel is pronounced long, every unaccented vowel short (e.g. jevotro yenUo), there the language has suffered a change affecting its very essence and something absolutely new has been developed out of the old. Nor would the ancient Athenian think the language especially agreeable• to the ear, I mean ancient Greek in the mouth of the modern Greek. His taste would probably coincide with that of Dionysius of Halicarnassus and Hermo- genes, who both declare i to be of all vowels the least agreeable to the ear and the most wanting in dignity\ But in ancient Greek, spoken according to the fashion of the modern Greek, this vowel has an unnatural preponderance. Finally, if a German came with his Reuchlinian pronun- ciation, observing quantities with pedantic care, the ancient Athenian would probably stop his ears at such disfigurement of 1 Dionys. ττ. συνθέσ. p. 77 Κ. (ίσχα- p. 225 W. 291 Sp. (το i ηκιστα σεμνψ TOP δί πάντων το t) ; Hfrmof^. π. ιδ. ττοιεΐ την λβ^α' πΧΐονάσαν). iS THE J'RONUNCIATION OF his language (if iiidced he rccognizcrl it as such) and at such discordant sounds. For who (to take an instance from Herodo- tus) would put up with tlS alithlUs τ;;9 ηΧηθηΐ'ης, tls ii/ilis της ύ'γίβίης and all the similar monstrosities, such as never appear in any real language ? The ancient Greeks, as soon as ei became simple t, no longer said vyteia but vyela, and in like manner ταμβΐον for ταμιβΐον, irelv for irielv, just as at an earlier date TToku was contracted to ττόλί, Δϋ' in many cases to Δ/. However we are at liberty by all means to pronounce as we please ; we are perfectly secure against the censure of the hypothetical ancient Athenian, and this fiction only illustrates the fact, that we can attain perfect accuracy neither in practical pronunciation nor in theory. After this rather long introduction I reach my subject, and first in order the history of the vowels and diphthongs. I. Vowels axd Diphthongs. Section 7. System of Vowels. The relation of the voAvels to one another is excellently illustrated by modern authorities, for instance R. Lepsius', by the well-known triangle, having at its corners a i and u. Between a and i come the two e's, the open (French e e, Lepsius' e) nearer to a, the closed (French e, Lepsius' e) nearer to i. Both es are found both long and short; the German laniTuaere however Avants the short closed e, which must be sought in the short i of certain dialects. In like manner between a and u come two o's, an open and a close. A. ii. p. S{);$, into ί'φωνοι ANCIENT GREEK. 23 must also be meutioucd, that Sextus Empiricus' quotes from 'certain philosophers' the statement, that there are other ele- mentary sounds, different from those usually taught, for instance ai, ου and all similar sounds. For these sounds are, according to their statement, unlike a syllable such as pa, the same from the beginning to the end of their duration, and this is the charac- teristic of an element. He afterwards mentions ei also as be- longing to this class, which indeed will coincide with the six diphthongs of Dionysius and with the diphthongs κατά κρασιν according to the original numeration, to which therefore ai and ei also belonged. More discrepant, than at first appears, is the distribution of the musician Aristeides'•^: κατά κρασιν, κατά συιχττΧοκήν, κατ βττικράτβιαν:, of the diphthongs κατά συμττΧοκην he says, that coming at the end of a word they are less easily shortened before a following vowel than the others, since the tone is stronger owing to the clear pronunciation of both vowels. Now since ην ωυ vt scarcely ever occur at the end of words, we must understand this to refer to ev and αυ {αν, ev, Zev etc.), and the corrupt statement about these diphthongs των κατά συμττΧ., λέγω δε των hui του (a poor variant δί' αυτών) συντιθεμένων, must be emended by the repetition of a letter, διά τον <ϋ>. The class κατά κρασιν would thus be limited to ai, ei, 01, except in so far as ei, having already become long i, had now to be counted in the class κατ βττικράτβιαν. The expressions κατά (the six oiOionys.), κακόφωνοί {ηυ ωυ υή ^^γ^ι καΐ ουτο% στοιχεΐον. If then in aud άφωνοι (a τ] φ) 1 pass over as ^^jg t^j^g of gextus (about 200 a.d.) at having no importance by the side of ^^^s pronounced ever so decidedly as ά, the other. ^.g gQ^ jjq ,;^,,^ element out of this or out 1 Sext. Emp. adv. matheiii. p. 625 of et==t. Accordiugly the philosophers Bk. : και ανάστροφων ίσεσθαί ηνά φασιν referred to in the sentence, in whose evioi των φιλοσόφων πλείονα στοιχύα, ^[^^q ^^ ^y^s still a diphthong, must be οιάφορον έχοντα δύναμιν των συνή- earlier. θωνπαραοίδομένων,οΐον καΐτο at καΐτόου - Ai-ist. Quintil. p. 44 Meibom. καί πάν ο τψ όμοιας έστΙ φύσεων.— έπειουν (ρ. 29 Jahn) (at οίφθο^^οι, as τ^τοι κατά ό του at και ei φθογγοί άπλοΰί ίστι και κρασιν η κατά συμπλοκην <η> κατ έπι- μονοΐώψ, ΐσται και ταύτα στοιχεία. κράτειαν -γί-γνεσθαί φαμεν). Afterwards Afterwards 626 after a discussion on p_ j^(j^ εύτονωτέρουν yap αΰται ποιούνται at: — τούτου δε οϋτων ίχοντον, έπεί και ο rovs ηχούν, αμφότερα φανερών έκβοώσαι του et φθόγγον και δ τοΰ ου μονοειδην και τα φωνήεντα, άσύνάετον και άμετάβυΧον λαμβάνεται, 24 '/ν/Λ' riiOXUNClAT'lON OF κρΓισίν and κατά συμ,ττΧοκήν are a marvellously happy (lefinition of the (listiuction intended ; for in proper diphthongs, as Rum- pelt says', the voice sounds during the movement from one vowel-position to the other and only during this movement, so that an actual ' mixture ' takes place as between water and wine ; in improper diphthongs on the other hand the relation of the sounds one to the other is an ' interweaving '. We are un- fortunately not in a position, with the means at our command, to follow up to its sources with any certainty the ancient theory of diphthongs. Section 9. Ε and ϋ nuands, their oldest development and representation. As regards the value of these vowels and diphthongs, since a admits of no doubt whatever, we will begin our in- vestigation with a discussion of the Ε and Ο sounds. Originally, and in most local alphabets up to the year 400, e\ery e Avas written Λvith E, every ο with Ο I The Greeks of the East however, and especially the lonians of Asia Minor, at a very early period employed the symbol H, Phoenician Cheth, properly used to signify the rough breathing, as a vowel-symbol for a particular kind of e. This Avas in fact very readily done in Asiatic Ionia where the breathing was lost ; the symbol in con- sequence of this Avas now called ητα instead of Cheth ^Ητα, and began with this vowel, exactly as άλφα with a. At a somewhat later time, about the sixth century, various attempts appear in various localities, to distinguish the corresponding Ο sounds by the introduction of a new symbol. The symbol Ο was differen- tiated by leaving the circle open (C), or by a point in it (Θ), or by leaving it open below and annexing two feet (Ω) ; this last form ultimately prevailed, and was applied in the manner adopted by the lonians of Asia, according to which the new symbol corre- ' lUmijielt d. luitiirliche Si/stcm des reader once for all to the classical book Sprachlaute p. •17. of ,\. KircliholT: Stiidiiii :ur Gcschiclite • For facts of epigraphy I refer the den ijiicchischvit Al2)hubctii. ANCIENT GREEK. 25 spoiuled to H, the old symbol Ο to E'. But that, which was so carefully distinguished in the cases of e and 0, was by no means, as has been assumed since the days of Greek grammarians, the quantity. For, although Η almost never and the corresponding Ο symbol in no instance whatever represents a short sound, Ε and Ο are as late as the fourth century used for long sounds, for those namely, which in the developed orthography are written diph- thongally eu and ov respectively, without however being really by origin diphthongs arising from e + i, o + v respectively. In λείττω and 7ez/ci the t is radical, as is also the ν in ου and ούτος ; on the other hand in έστειλα, στεΧλειν, τιθείς, φίΧεΐτε the ec is merely lengthened e, and in βουΧη, δίΒονς, μισθοντβ, λόγου the ου lengthened ο. On the one hand, therefore, the Greeks distin- guished e and ο together tuith their lengthened forms, and on the other the sounds η and ω \vhich were always or almost always long, and furthermore it never occurred to anyone in ancient Hellas to distinguish in script a and a, I and I, ΰ and ϋ, the natural way to do which would have been to donble the vowel, just as the consonants were written doubled for similar reasons. Consequently the distinction betAveen Η and E, Ω and Ο was originally one of quality^ and the only qualitative distinction Avhich can have been intended is that Avhich the Italians make prominent both in pronunciation and in grammatical writing in the case of these two vowels and only these, namely the distinction between open and closed e and 0. The quantitative distinction came to pass accidentally and secondarily, after e and ο had been distinguished from their lengthened equivalents by the diphthongal writing of the latter, and it became the more obvious and finally as early as Aiistotle^ the only distinction recognized. But which e did the ancient lonians intend to re- present by H, and which by Ω, the open or the closed ? On this point the old inscriptions of Keos Naxos and perhaps 1 In Faros, Thasos, Siphnos con- application, by Dittenberger on the versely Ω was written for (oi'), for ω: subject of the old Naxian and Kean in- ΣΩΙ σοι, ΤΩ του, TON των, see Ivirchhoff scriptions [Zum Vocalismus des ioni- p. 60 ff. schen Dialekts, Hermes xv., 225 ff.)• 2 This was first explicitly stated, ^ gee Arist. Foet. c. 21. though not with .the necessary general 26 THE rUONUNClATION OF Aiiiorgos also are especially instructive ; in them Η and Ε only partially coincide with ordinary Η and EV For there Η is only written i'or that e, which corresponds to old Greek (Doric) a, and also that arising from contraction of ea: OIKIH, ΔΗΜΟ^, EllllN, ΘΤΙ1,(τα^ι^εα from το θνο^;)'-; the η on the other hand which is common to the Greek dialects together with e and e is denoted by E, without admixture of diphthongal writing': ME μή, Ε11113ΛΕλΙΑ βττίβλημα, ΦΙίΡΕΝ φέρβί,ν, ΕΝ ΑΙ elvai. The Naxians represent the short sounds also with H, if they have arisen from long a\ ΔΗλΙΟΔΙΚΙΙΟ Δ.ημο8ίκ€ω, ΑΛΗΟΝ άΧΧύων*; in Keos e is written in these cases. If then in these • dialects that sound is written with 11, which elsewhere has the value of a, and previously had that value universally, we nmst give to Η the value of open e, that is, the e which stands nearer to a, and to Ε that of closed e, that is, the e which stands nearer to i, especially as this corresponds to the Avriting El cur- 1 Cp. Ditkub. 1. c; Mitth. iU'n arclucol. Iiislit. i. lii'J li'., (Keos U. KoLler) — Ituhl, Iiiacr. Gr. tDitiquissimue uo. iJ'Jo a. ; liulletin de correipondcnce Hellciiique, in. 1 ff. (Bustropliedon Inscr. on the offering of a Naxian woman) = liohl 407; Bechtel 23, Bull. VI. 187, Mitheil. xi. 97 (Amorgos) ; ilcclitd, 2'J if. ; Kirchhoff •* 32. - The two last cxamijles ou line 17 of the longer Keuu inscr. (derived from a correction ou the stone) ; in the same Ijlace occur also διαραρΰψ and line 23 θάνηι; thus in the diphthong ci (24 έ^ΐνιχϋά) a mixture of the two Ε sounds appears. But this occurs in Attic also and elsewhere : TEI for ttj side by side with Till. Dittenberger's endeavours on this head are in my opinion misplaced, liohl's restoration TII[/\oC ffrdjNTA 1. 1(3 1 consider wrong ou the score of meaning ; for a lustratiou of the interior of the house {διαρραίναν) cannot be accomplished from a dis- tance. The Naxiau luscr. oilers only one stumbling block IlKllDOAUl έκη- /3ύλψ, which 1). is certainly right in ex- plaining as agraver's error for HEKHB.; for Η here still keeps the value of the breathing as well as the other. On the Naxian bronze published by Friiukel Arch. Z. 1879, «4 ff. ( = Hohl 408) we find EKHBOAOI. I may here remark, that Merzdorf (Curtius Stud. ix. 202 ff.^ tries to prove a double value of Η in ordinary Ionic : from Xaos, Χηόί (open i•) came Xews ; from βασιληοί on the other hand (e original and closed) βασιλ^ί. νόλΐωί however occurs twice ou the tolerably old inscr. of Chios ; Caucr no. 133, Εϋΐιΐ no. 381, Bechtel 174, cp. id. p. 107. « But in C. I. Gr. 23G3 b, Bechtel 44 (Keos) ΕΙΣ occurs twice in proper names of the 3rd declons. alongside ΕΣ (according to the earlier copies, while the later shew lacunae in the places in (question). ■• Conip. 7ΓΟλί;α5 in verso, Abdera liulil 349, Bechtel 102. IIII'UN Thasos Kohl Itiutij. 52, no. 4 is explained by Bechtel (Ion. Inscr. δϋ) as a mistake for HIP., since Ipos is found elsewhere in Thasos. ANCIENT GREEK. 2/ ri'iit clsowhcre, ami the latter as early as the Alexandrian period had become i. Conse(|uently ω also is open ο (ρ), and ο the closed : in fact the lengthened equivalent of the latter became at an early period a ti. Those then pronounce correctly, who give to μη a sound similar to French mais — in Keos and Naxos, it is true, the pronunciation was me, but the mode of Avriting was also different — ; no German on the other hand pronounces e/ie correctly, inasmuch as the short closed e, frequent in Italian (fendere, elmoY is strange to that language. Ω, in ωρα must be pronounced as ο in French encore ; but ο in οράν neither with this sound nor yet with German short o, but one tending more to a, although not the same as u. Here again the Italians might be our instructors"''. I am not of opinion however, that we ought in practice to exercise ourselves or our pupils in this mode of pronunciation ; there could not be a more mischievous Avaste of time. A striking proof, that the foregoing statements are true in an especial degree for Attic, is, that the Boeotians, when about the beginning of the fourth century they appropriated the symbol H, employed it to represent their dia- lectal sound arising from common Greek ai {\\ρίστηχμο<ί, κή). That is to say they gave to the symbol derived from their neigh- bours in Attica, the value, which it had there, and from at came e (d), not e. Moreover in the comic poets of Attica the cry of the sheep is represented by βτ} βή : υ δ' ήΧίθίος ωσττβρ ττρόβατον β η βή λεγωι/ βα8ίζ€ί^. The next evidence touching e and η is the contraction of ea to η, in Attic, Doric and Ionic : τεί;)^?/ Att., ην, βιτήν, θύη, lonic, Χτρατής='^τρατ£α^ and Χαλκ7'}=Χ.α'\.κέα on Uhodian inscriptions*. For we cannot get from ea the mixed sound e, which lies nearer i than either of the two elements, 1 Diez Gr. i^. p. 333 (every un- of Etacism. Further cp. Ai-istoph. fr. accented e in Italian is both short and 642 K. θύειν με μέλλει καΙ κελεύει, βη closed). λέ-γειν. Lohmeyer {Phon. Stud. ι. u9j - Diez 33G: "every unaccented ο compares Hesych. βηβην πρόβατον. is closed." ■• Inscr. of Kamiros and lalysos ^ Kratinus fr. 43 Kock (from the published by C. T. Newton, Transac- drama Dionysalexandrony assigned by tions of the R. S. of Literat., Vol, ix. Meineke to the younger Kr.). The line N. S. Liscovius, p. 19, recognizes in is used by Aldus Slanutius and after- this phenomenon an argument for wards by Cheke (p. 288) in support open c. 28 THE I'liONUNCIATION OF but ea readily gives e, standing as it does midway between the two. The same follows for Doric from the contraction of ae to η : νικην, τημά i.e. τά €μά. In fact for dialectal η in general we must everywhere assume the same sound of open e, and accor- dingly the following history of the Ε sound for Greek becomes evident*. The short e had at that prehistoric time, when forms such as τΓΟίησαί ττατήρ ησθίον arose, still an open sound ; for the lengthening gave η e. This open sound may have been re- tained in those dialects, Avhich in later formations also, such as contractions, keep η as lengthened equivalent of e, that is in Arkadian Elean Lakouian Lesbian etc. The Dorian dialects coming under this category having ηχον for €Ϊχοι/, ζφίΧητο, ης, are united by Ahrens under the name of the stricter Dorism. These then, and the Lesbian etc., had everywhere only one sort of e, the open, at least in the long sound, for the short may indeed subsequently have had the same development in these too, which it had long before elsewhere. In the milder Dorism, in Boeotian, Thessalian, and Ionic, e became at an early period e, hence its lengthened ecpiivalent ei. Further the old long sound as in ττατήρ remained in most dialects open ; but among the Ionic Keans and Naxians and also in Boeotia and Thessaly it got the closed sound: MHTEP meter (Keos), MATEP and from the fourth century onwards MAT EI Ρ in Boeotia and Thessaly. In the last two dialects therefore there was also only one kind of e, that is the closed, except in so far as an open e had been newly developed out of ai. Lastly the special Ionic η was everywhere e. The case is partly analogous, partly diiferent, with respect to the ο sounds. Since ω was open, ο must have been so too at the time when the nominative -ων arose from -οντ and the augment ω from ο ; the open sound maintained its ground still longer in those dialects, which nuule λόγω? out of λόγοζ/9 and λόγω uut of Xajoo, that is, roughly speaking, the same, which shew η fur et, and also Boeotian. In the rest ο became at an early period υ, hence the lengthening ov. Finally the original long sound as in Χβων remained open every- where except in Thessalian, where it was represented by ου. ' I follow bore the excellent essay Spnich,•, ΚηΙιιι'.-! ZiitKchr. xiv. μ. 18 1Ϊ. ufDic'tricli, Xiiiit Vocalisinun der griecli. ANCIENT GREEK. 29 Section 10. EI and OT from Ε and O. I have intentionally deferred to this point the important question, Λvhat the sounds are, which are represented by EI = e and OT = 0. First of all there is no doubt on this point, that the real el as in Χείττω and the real ου as in οντος were origin- ally the diphthongs ei (more accurately ei) and ov (more accurately o?0 ; with these diphthongs at a later period, lengthened e and ο are universally confounded in writing, and were so, in many places, even at an early period. This levelling took place earliest in Corinth and its colonies, in the sixth century or even earlier. By the Corinthians the local symbol Β was employed for e and η, the ordinar}'^ Ε for e and er. ΔΡΕΝΙΑ (real et) ^eiviov, ΠΟΤΕΔΑΝ (do.) UoreiSav, KAETOAAS KXeiroXa^, but Η^ΝΟΚΛ^Σ ΒιβνοκΧης^. In Corcyra ^ is the only form, and both ets are written diphthongallyl In both places and also in the Sicilian colonies of Corinth spurious ου is denoted by OT, while Ο serves for ο and ω I This OT is found also instead of Τ in the diphthong eu : ΆχιΧλεούς on a Corinthian vase*; correspondingly Corinthian E = et as second element of the diphthong ai: ΑΘΑΝΑΕΑ Άθαναβία Άθαναία^. All these forms of writing are not perfectly constant ; for example here and there the Corinth- ians resolve their Ε into ^S (ei), as ΠΟΤ^^ΔΑΝ, once we find even Άμφιτρίτα written with Ε et in the penultimate 1 KirchhofP, p. 88 ff. ; Eohl Inscr. yopvs Naxos E. 408 ; ΔΕΙΝΟ Melos K. Gr. antiqu. no. 15, 20, 16, 23. As a 433; Aavias C. I. A. i. 299, 433, 447, rule I intentionally refrain from giving 483. the epigraphic forms of the symbols. ^ Epitaph of Meneki-ates (Rohl 342) That Aeivt'as has the real et is shewn by ivoiei. Epitaph of Xenvares (E. no. the fact that archaic inscriptions every- 344) Μ e/^ios (real et) etV. where else WTite EI in names derived ^ Kirchhofl'•', no. 104 f. from SeiiO's : Α€ΐ.ροδίκηο and A€lvoμ^veos * Colhtz Dial. Iiischr. 3122 (An- Bustrophedon Inscr. Naxos; Α€υΌμέ- imli deW Inst. 1S&2, 5f> ff.). i/eoy Hiero's helmet, R5hl 510 ; Aeiva- ^ Eohl no. 20, 4 comp. 5. 30 THE rnOXUNCIATIOX OF syllable \ From all this it is quite clear, that the lengthened equivalents of e and ο had become so near to i and u respec- tively, that a need was felt of differentiating the real and spurious e, and in like manner the real and spurious o, while on the other hand no such need was felt of separating original diphthongal et and ου from the newly developed mixed sound. The mixed sound was thought to be heard in diphthongs such as ai and eu also, and a corresponding mode of writing was adopted. This sound might be represented by e' H, g" "71 ; the * i pingue' of Lucilius, which he wrote ei (puerei nom. plur.), will be nothing else but the Corinthian E. For the other Doric dialects our material is not at present adequate. ; but the diph- thongal Avriting of et and ov is to be found on one of the Lokrian bronzes of the fifth century ^ The old Ionic and Attic inscrip- tions nowhere or almost nowhere shcAv Ε for real EI ', but at a very early period EI for e^ although the Athenians in particular in by far the larger number of cases do not separate e and e in script. We must here state our opinion : the sound Avhich is constant in writing, that is real ei, was constant also in pronun- ciation ; that which was shifting in writing was shifting also in 1 On the Corinth, clay tablets, pub- lished by Eohl under no. 20, Τίοτειδανι is wi-itten 26 times with E, 4 times with ^Σ (once also ΠΟΤ^...), twice with B=f) twice with Σ = ι, and once with E2 = fu. The last three forms are rightly considered by Kirchhoff^, 103 (note) as errors (omission); in fact Ποτ- <Ε>δά>', 'A0ca etc. are also found. Wc have a certain example of == ου in ATTO RiJlil no. 329 (Anaktorion ac- cording; to Kirchhoff). ■- Cauer, Del- no. 229; Riihl no. 321; Kirchhofip. 146; v. Wilamowitz ZUvhr.f. Gijmn.-Wesen. 1877, 042. ■' For Attic see Cauer (in Citrthis Stud. VIII. 231); he produces as examples of Ε = real fi only OAEZON (so C. I. A. I. 37 (9 ??); IV. Γ)3\ with OAEIZON 1 η 33 ; IV. 27 1' 18. There arc found be- sides ΙΙΕΙΙΛΟϋ IIf/iri5os C. I. A. IV. 373"; ΕΧ^'ΛΛΚΨΑΤΟ? dn. Γ>3»22; also (Kretschmer ZUchr. f. vgl. Sprachf. N. F. nc. 154) ΗΡΑΚΛΕΔΗΣ C. I. A. IV. 4911»; KETAI κείται do. 491=7; ΜΕΝΕΚΑΕΔΕΣ 373 "^, These are almost all private inscriptions. — But άτΓοδβκηψτΐί Eiihl no. 381 B, 13 shews the Ionic shortening of this verb. •» Teos C. I. Gr. 3044 = Eohl 497 ΚΕΙΝΟ Β, 7; in the same place 6 in- stances of Ε in this word. Halikarnassus R. 500 at least 4 times ΕΙΝΑΙ (with Ε only two certain instances) ; EIXON ; on the other hand φίΐ^γαν and ένικαλεΐν with E. The Sigean Inscr. R. 492 has ειμί in the Ionic part with E, in tlic Attic with EI. Miletus 6th cent. (Kirch, p. 19 ff., Rohl 488, 485): el.ul, Κλέσιο^ i.e. Κλίίσιοί, iiroUv i.e. (iroUiv. Athens C. I. Att. I. 1 thrice ΕΙΝΑΙ; Hull, de corresp. IleU. in. 179 ΕΙΜΙ. Comp. Cauer C. St. vin. 230. ANCIENT GREEK. 31 prouunciatiuii. Consequently λβιττω did not tend to be pro- nounced as lepo, but φέρειν (ΦΕΡΚΝ plteren) did tend to the prommciation pherein, without however the i in this case being very prominent. For the different treatment of the two sounds is a proof that they were not quite similar in the fifth century : etymological scruples about original i were obviously foreign to those writers. I am consequently opposed to the opinion, which is tolerably general at present, being held by Brugman^ and after him by G. Meyer, according to which the spurious ei never had the value of a diphthong among the Athenians and lonians, but was only an orthographic expression for β; A. Dietrich^ seems to me rather in this respect also to have seen the truth. For distinction of quantity cannot be regarded either in this case or elsewhere in ancient times as the cause of difference in writing: consequently the second syllable of φβρβι,ρ was dis- tinguished from the first in quality. The levelling of et and e, that is the passing of both of them into the mixed-sound described above, takes place for Athens and Ionia in the fourth century ; after the first decades of this century Ε is very seldom foimd for spurious ei, although this mode of writing can be traced beyond the middle of the century^ The Boeotians write their long closed e (— Att. η and βή even in the fourth century very frequently with E*; the thickened pronunciation can scarcely here be traced back beyond the beginning of this century^. Subsequently the i everywhere prevailed over the e in the case of the later (spurious) et of the various races, just as had long before happened in Boeotia in the case of the real e<. The view of Zacher (p. 30 of the treatise referred to on p. 16), that real 1 Brugman C. St. iv. 82 ff. menos Bull, de corr. Hell. in. 454, Bia- 2 A. Dietrich Kuhn's Ztschr. xiv. lekt. Inschr. 470, composed soon after 67 ; Eodiger Prorjr. Berl. (Liiisenst. 330, has in five instances EI only once, Gymn.) 1884 p. 6. Ε 4 times ; that from Thespiaj id. p. 382, 3 The latest Attic examples known Bialel't. Inschr. 798, never has EI. to me are Έστιαΰί 'Orpwes πρντάνε^ ^ Examples on the Theban inscrip. (341 / B.C.) C. I. A. II. 872; άποδώσΐν Eohl no. 300, which shews in essentials and is do. 804 A» 33, ''13, B.C. 334/3. the Boeotian alphabet; here EI comes Άλικαρί'ασσε (dative) is found in the four times, Ε thrice; and Καλλίκ-ρατΕΙ? inscr. Bull, de corr. Hell. 1888, 173 on the archaic inscr. of Akraiphia, (B.C. 354 / 3). Lolling Monatsber. d. Berl. Alcad. 1885, * For instance the inscr. of Orcho- 1031 no. 4, 2. 32 ΤΠΈ mONUNCTA Τ10Ν OF and unreal et were united in the 4th century into a pure closed e (e), seems untenable. For if -eiv, as Λνβ are bound to assume, Λvas in the 5th century e'n, but in the 2nd or 1st %n, it is quite certain that it cannot in the meantime in the 4th and 3rd have been en. With regard to the Attic-Ionic ov=d the case stands thus : the mode of writing was for a long time almost exclusively O, nay, isolated instances occur, Avhere it is written for όυ diph- thong, as in TOTON τοίτων'^. Even after the reform of the Attic orthography the simple Ο held its ground with great persistency, (and got more and more to be used quite indifferently for ov and o), isolated examples occurriiig up to the end of the fourth century^ In this case then tho designation of the diphthongal sound is at an ancient period no more constant than that of the lengthened sound, and accordingly the diphthong ου had as early as the fifth century coalesced Avith a sound, which arising from ο approximated to u, and finally became an undoubted v?. When the Boeotians in the fourth century adapted their own to the ordinary orthography, they employed the combination OT in this value, that is for their old T, for which unlike most of the other Greeks they had preserved the old w-sound. In the first quarter of the fourth century however the difference in quality betΛveen ο and its lengthened form cannot have been great at Athens, since to take an instance on the document of• the new 1 Dietrich 1. c. p. 51 ίϊ. Cauer Curt. Stud. VIII. 241 ff. OT is always writ- ten for 5 on the inscr. of Keos R. .395. In this dialect therefore the coalition took place very early. The Asiatic-Ionic inscr. generally distin- guish correctly (Chios E. 380 ; Halik. 500), in Chios 382 however we have TOTo; Teos 497 b, 20 βαρβάρονί. Comp. Erman Cvrt. Stud. v. 284 f. On the Attic treasurer's account C. I. A. i. 128 (01. 91, 2), TOTON and TOTO stand almost without exception, though it is true the older documents of a similar nature and also most of the later ones shew TOTTON and TOTTO quite wiMiout exception (s. no. 117 — 17(5). Other examples of for real ov from the 5th and 4th centuries are given by Meisterhans Gr. d. att. Inscr. ed. 2, p. 49. For ον = ό the oldest example on stone is C. I.A.i. 36 Λ0 (end of the 0th century?) Ήρακλ^ουί, Meist. p. 21, n. 121 ; exx. on vases Kretschmer, p. 154 (cp. p. 30, n. 3). - The latest Attic exx. C. I. A. ii. 830 c — k MaXi'a/fio(i')and other genitives in -ov, Meisterh." p. 6, n. 21. The inscription dates from the time of the Chremonidean War (circ. 202). See also Bull, de corr. Hell. ni. 513 κοινό and μνλωθρΒ (n.c. .302, ,301). =' Ace. to Dietrich p. (!(». ANCIENT GREEK. 33 maritime alliance (37 ; 132ίΙ II. ί("' (Tliessaly shortly after IW> B.C.) Χρισίμον for Χρασ. an accidental omission of the E. •* In Thespiae itself θίΐσττ. is the regular orthography, in the writing of the dialect; likewise Orchom. Bull, dc coir. h. III. 463 in the Boeotian part of the document always θασιτ. and actu- ally (line 91) θ€ΐσπΐ(ΐα$ το, in the same place θώφειστοί i.e. Att. Θΐόθίστο^: on the other hand in the part composed in the κοινή both name.s are written witli e. — As to ΐϊσχηκα see G. Meyer Gramm.- § 112; (ΐσχον Telos Bull, de corr. h. HI. 42; €ίστή\ψ, C. I. A. ii. 563 (else- where ίστήληι i.e. ev σττιΧηι). ' a. I. Att. IV. 373»», II. 1G8, 2C3, 352, 553,115''. Meisterhans2nded. p.35. In like manner we find on the Ionic inscr. of Zeleia Mitth. d. arch. luxt. vi. 229 (Ditt. 113) fiav and ivveia, and I have no doubt, tliat .Kolic πρίσβ€ΐα=:πρ(σβ^α πρΐσβΐντήν and in general -etos as gen. to fi'j assigned by grammarians to the later lonians and ..liolians, are so to he (■x])laiiied (Meyt i- § 14',)). Λ NOT Ε NT GREEK. 35 developed after a preccdiui^ vowel from the sound of tlic; .s-, and in like manner a weak i (or y) is a natural residt when the voice passes directly from e to another vowel. For the quality of e as ^ in the Hellenistic and Roman period we can cite, besides the confusions with ei, of which there are isolated occurrences before other consonants also\ the fact that Latin % in many positions was expressed by the Greeks in the earlier period with e; Ύββέρίος and Te/3epi9, Αομέτως, Κ-αττετώΧιον, Kat/ceXioos etc. cp. above (but Κλυτώ 7459), 9λυτίο5 7382 ; that p. 21, n. 3, G. Meyer, § 73. is, the interposition of a consonant 1 Καθημερίσια C. I. A. IV. p. 76 does not remove the influence on the (col. III. 25) ; for νυκτερήσωί vuKTepeiaLos K-sound (or on its representative). Cp. also has mss. authority (as in Aristoph. Άρίι^λτ;? (?) Εϋΐιΐ no. 520 (Chalc.) and Thesm. 201). Cp. νυκτερινοί ή /xepiros. from Doric Magna GriEcia do. 513 In Plin. xxxv. 11, § 124 mss. Bamberg. QwiaQos. Corinth Euhl 7, 47 (^vXoidas, and Voss. have liemerisioe. Dial. Insclir. 3123 (3129) (^uWapos, - Foy p. 86; Meyer- § 93. Cp. Ap- 3135 iXiiros lleptiXyyueiOs etc. Cyrene pendix. lluhl 506* Qvpa{vai(xiv). But in Attica ■^ RYllY Bechtellnsckr. J. io)i.Dial. Κννόρτψ side by side with (}όραξ, 3 (Kirchh. 121) ; ςιύςνοί λή^τυθοί C. I. Meisterh.- p. 22. Gr. 7611, 8337 ; likewise (^λυτώ 73Sl ^ Dial. Imchr. 3130, 3135. 40 THE PROXrXCIATWX OF remain to tliis clay. That the «-.sound was preserved in the neighbouring^ eountry of Buiotia, we know from the translitera- tion with ου, which became usual there after the adoption of the common Greek modification of the Ionic alphabet ; this ου was iu the course of the fourth century already employed for the short sound also: Πουρρΐνος, Φάοϋλλο•?'. This is another proof, how little the ancient Greeks troubled themselves about the differentiation of short and long vowels in script. The popu- lar Lakonian also still possessed the i7-sound, as is shewn by glosses '■', such as κάρουα, τούνη i.e. σύ, ούμ,αί νμέτβραί. It must be stated however that on inscriptions and in the literary monu- ments of this dialect no such form is found ^ ; accordingly the cultivated language of the Spartans may have had the ordinary a, in support of which the intercliange of υ and l on inscrip- tions such as Ύίν8αρί8αί, 'ΚΧευύνία {ΈΧευσίνίαΥ may be cited. In Cyprian and Pamphylian also the sound appears to have been the original one I But in general the a was modified at a very early period in the same way as Latin ii in France and northern Italy®: this i^ronunciation is established for the Attic of the fourth century in particular by the Boeotian manner of writing ; for the ου would not have been introduced, if the Athenians had given the same value to Τ as the ancient Boeotians. Moreover if that had been the case, as ο became nearer and nearer in sound to a, a confusion between the symbols Ο (OT) and Τ would have been inevitable. But on Attic and other inscriptions of the fifth and fourth centuries it is rather I and Τ which interchange : βιβΧίον antl βυβλίον, 1 Orchoin. Bull, dc corr. hell. in. incuts Alcmau's poems come particu- \o\ — DiaL Inschr. 470 (about Ά'όΟ larly uiuler consideration, since the ου B.C.) ; V is written here as well. The has been introduced into the Bceotian Thcban inscription on the contribu- poems of Corinna ; the Lakonian iu tions to the Sacred War (Αθήναιον iii. the Lijsi.it nitu indeed shows thronjih- 479, Dial. Insclir. 705) has ov only for out υ, but the same may be said of the V. Cp. 11. Meister Or. Dial. i. 231 f. BaOtian in the .iclianiian.i. - Ahrens D. D. p. 124 ff. ; G. * Ίινδ. liohl 02" ; also Bull, dc con: Meyer- p. 103 f. /wiZ. in. 365 (Cythera). 'EXevvvia 1\. ■' But Koi'ooKpets=Ki'»'offoi;/)ets on the 79, 11, cp. ΈΧΐνσννίω Crete Bull, dc very late Lakon. inscr. C. I. Gr. 1347 corr. hell. in. 292 1. 8 (name of month), and 1388 comes under this head "* G. Meyer- p. 105 f. (Ahrens 1. c). Amont; literary nionu- " Diez Ur. p. So f. AJCIENT GREEK. 41 τρίβΧων iuul τρνίβΧων, Μονιηγ^ιων and Μουνυχιών, ημυσυ very i're([uently ibr ημισυ\ Άμφίκτύονβς and ' λμφίκτίονβ<ί. Added to this the treatment of the diphthong vi, which at Athens in the fourth century was siniplitied to υ ahnost without exception, would be perfectly incomprehensible, if the latter had been not u but a: the BcEotians write ovio'i. But in Athens even the archaic inscriptions shew vv is τρίβλιον for τρύβλιον Delos 36J: B.C. (Bull, de con: hell. x. 461, 1. 16, 23). Μουννχ. first C. I.A.n.2i7 (306 B.C.); ημυσυ II. 17 A, 45 (378/7 B.C.) and in all later Attic examples; further. Bull, de corr. hell. 11. 580 (Delos), Pap. Louvre 1, col. 4 and in general here without exception ; but in the more correct documents no. 22 aud Pap. pears first 410 B.C. (Bull, de corr. hell. VIII. 283). On the inscr. of Hahcar- nassus Bull, de corr. hell. iv.-295 (cu'c. 400) Σι.ούλημΐ5 a.na Συδύλημί3 ; the stone of Sigeion Kohl 492 (6th cent.) Συκε- ίΰσί Σί-γευΐΰσί 2t7eies. Examples from Dehan inscr. (cii'c. 180 b.c.) HomoUe Bull, de corr. hell. vi. 114 (κυλύχνιον cp. κυΧίχνη, Κυνθυκωί -ικώι, XoipyXos assimilation like ημυσυ). ^υνδυμενίηί Aitake Bull, de corr. hell. xii. 108. Megariau αίσιμνάτα$, Bechtel n. on Dial. Lischr. 3016. - ΗΤΣ C. I. A. IV. 373»^; HTTS scanned as one syllable do. i. 398, as two syllables iv. 373^'^ (-ir). Cp. § 14 below. 4^ THE rnONUNCiATION OF pronimciatioii is proved even in the case of the later Boeotian, only there it takes a different form which coincides with the present English representation of French u. That is to say an i is prefixed, and there arises an improper diphthong, which so far as the writing is concerned was in Greek actually a triphthong, capable of being scanned either short or long: ΥΙόλιούστρατο<ί, TLov -χα, Αιωνωύσίος. This mode of writing is however never , constant, and is generally only found after S τ θ ν \^. A similar development of sound has taken place also in popular dialects of modern Greek, for instance in that of Trapezus, and in the descendant of the ancient Lakonian, the Tsakonian, from the latter of which are cited Xcovko Χνκος, κίουρε τυρός, νιοντα νύξ and others ^ In this case the modification, which is strange to the ancient dialect and even in the modern has by no means become general, seems to have established itself in a manner analogous to that in Boeotia; but in Trapezuntine and in the other localities, where a similar phenomenon is found^, the transition may have been similar to that in English, that is yii may have aj)peared in the place of a ii which was disappearing. In ordinary Greek however the ii has maintained itself for a very long time, not only through the Roman period, where the Latin representation with a and then with y is in evidence against its identity with i, but also on into the Byzantine era. For long after the extinction of the diphthongs and the transition of η to l, ν and ot (which by that time coincided with υ in sound) kept themselves distinct from t η et, even the most uneducated masons never confusing them. Accordingly in iSuidas' Lexicon, Avhere ειη ι stand together after Ζ and before 1, 01 and υ are put by themselves in the alphabetical position of the latter ; at that time every one knew by the light of nature, that οίκος and vypc^ were not to be looked for under t or η*. ' 11. Moister O';•. Dial. i. ΊΛΆ. IJut voiiic the boriOwed words, kyuminu also ίοΐ'ίώ (Tetraplitlionx!) Cluicroiifa ϋνμινον,ηιγαϊο μύρον, zmyuiua, σμύρνα. Dial. Iiiiicltr. 382. '* This is not coutradictetl by the - Foy p. 80 ; Dellner C. Stud. iv. I'aet, that coul'usioii between η- οι- ν- ι 208 1ϊ. ; G. Meyer- p. 108. etc. appears occasionally in a l'ai>yrus •' Meyer produces from tlic modern of a much older date, see rsiehari Greek of southern Italy χyuuo = χι;ι'ω liev. crit. 1888, 381. For the Byzan- χέω, ax;>nro άχιφον, from Church Sla- tine period, where we have such abund- ANCIENT GREElt. 43 Lastly we iimsL not omit to iiieiition, that Quiutiliaii sees an especial euphonic superiority of the Greek over the Latin language in the possession of" the f-souud*. Section 13. DipldlLongs having the first vowel long (HT, ilT, AI, il\, HI). So much for the simple vowels ; we have now to speak of the diphthongs formed from them. These have in the lapse of time altogether lost their distinctive character, by no means however simultaneously, but rather one sooner another later. I begin with those improper diphthongs, the first element of which is a long vowel, that is cu ηο ωι (άυ) ηυ ωυ ; these indeed were the first to lose their distinctive character. We must in accordance with what has been said^ before lay down as their original value : ili, ei, oi, m, on ; the u when occurring as the second member of a diphthong having in general, as will be shewn later, the value of u proper. Now these semi-diphthongs are one and all inconvenient to pronounce, because the com- ponent parts do not coalesce to a proper unity, and hence the tendency of the language, either to fuse them more closely together by shortening the first element, or to simplify them by rejecting the second. In the former way we may suppose that άυ, if this sound was indeed heard in such a word as Attic ^ρανς = Ionic Ύρηΰς, at an early period was identified with άυ ; in the Attic ναΰς in spite of the Ionic νηΰ<; a short vowel must be assumed, since an ά would in these cases in Attic also have become η^. ΩΤ hardly occurs in Attic (7Γρωυ8άι/=-7Γροαυδάι>, ance of evidence, it is not only allowed who draAvs the conclusion that the 9th but incumbent on us, to sift this arid 10th centuries were the period, in evidence more than elsewhere ; it is which the transition from ot and υ to impossible that an obscure Egyptian the I- sound was completed. Hatzi- scribe should be taken as an authority dakis Άθψαιον χ. 42 ff. for the general cultivated prouuucia- i Quint, xii. 10, 27 : jucundissimas tion in the Byzantine realm. The ex Graecis litteras non habemus, voca- great mass of evidence certainly shews lem alteram, alteram consonantem (i; the long continuance of a separation and φ), quibus nullae apud eos dulcius between ii and i. tiee also Ivium- spirant. bacher Ber. d. hai/er. Ak. 1886, 444, - G. Meyer-, p. ia4. 44 ^v//; I'lULMixcTATroN of (ovpLTrLhi] — ω ϊ'Λ'ριττίΒη), is more tre([uenl iu Ionic and Doric, but even here is almost contiucd to crasis : εωυτοΰ, ώυτός. Now we Hud on an Ionic inscription €ουτώρ\ with shortening and at the same time also approximation of" the first sound to the second (ou instead of qv) ; κον = και ου also may be ascribed to this shortening, since κωύ would have been the regular crasis and is actually recorded in Sappho and Epicharmus^ In the middle of the fourth century the Athenians retained ην in the augments of verbs with initial ev, and it is therefore rightly re- placed iu texts ; at a later period these verbs were augmentless, that is ηυ ea had passed into eu^. This also may be regarded as an accommodation of the first element to the second, in so far as e lies farther than e from the original sound a, although not in the direction of ii but of i ηυ maintained itself as augment of αυ, chiefly perhaps owing to the Grammarians, if an inference may be drawn from the augmenting of ai to et, to be mentioned immediately, and from €ύ-χρΰμ,ην (from αύ-χ^εω) βΰξησα (from αυξάνω) of later inscriptions*. Far more important in the language are the corresponding diphthongs with t, in which the other method also, that of simplification, is employed more vigorously. We have an instance of this at a very early period in the nominative of feminines in -ώ, originally -ooc, as is shewn by the testimony of the Grammarians from old manuscripts and by a few inscriptional examples ; as a general rule on quite old vases and stones we find only -o (ω)^ The next instance to be produced under this ' C. /. Gr. 2"J0u (Mykale);=Beclitel Kohler C. J. A. ii. 01^ (u.c. 3G2/1), but Iiinchr. d. ion. Dial. 144. The reading only owing to an error, since the of this inscription is however by no stone, as Rieuiann assures us, has means certain. HT. On the other hand we have cv- - yajjpho I. 21 κωΐ'κ iOeXoiaa, ερ•γέτηκε and ΐνερ-γΐτήκασι, id. 271 and Epich. 1',) Ahreus κωϊ'οέν od. On the 283 (end of the fourth century), other hand κονκ on a Papyrus of the •* Kaibel Kpiijr. no. 11(2 (Thora, in Ionic dialect editid by Pelrettini, Pap. Koni.in j)eriod, in epic dialect) ; also Grecu-l'Ujizj (Vienna, lH2()j line 1") (cp. v. Λ ΕΤΧΩ is certainly to be emended p. 55, n. 2), and more accurately by to ίί'χο(ΐΊ') not ΐσχον or Ισχω. Ει'ξτ/σα Wessely, d. (jr. I'djajri d. kais. Summ- ίπ(όζησα in (ireek text of Monum. luwjeiL Wiem (Vienna, 1885). Ancyranum col. 4, 8; 14, 4. ^ O. lliemann Hull, de con: U. iii. '' Meyer- p. 315. In lUihls ///.•<(•/•. ϋϋυ f. (after Wecklein Cur. cpiijraph. dr. anl. I find only three instances of :<;5 ff,). We liiid, it is true, (Γ'χί?αι in c no. 415, 433 (Melos), 558 (Akrai ANCIENT a REEK. 45 head is the -ησι -ασι of early Attic inscriptions by the side of -7)iai -αισι (i.e. αισι) in the dative of the first declension: this -Γασι -ασί occurs only after t or p, while the form -at? Avhich appears subsequently following all sounds alike has the a short*. The Dorians, Boeotians, etc. had -ai? already in the earliest period ; the ancient Boeotians having also ϊη in the dative singular, as we may gather from the analog}^ of their oi in the dative of the second declension ; the same is true of the Arcadians and Eleans. In the case of these races indeed the diphthongs di ωι had in general become at oi ; for ττατροίος is recorded by the Grammarians as a Boeotian form^ Or again, they kept the vowels separate, Boeot. ΐΐτω'ί'ων ΌμοΧώΐχ^ος Έ,ίρωί'Βας, like Καραίων Έρμάϊος Νί/ίολαϊο?. Among the Thessalians we find in like manner Είρονίδας and ττατρουβαν^, where no one will assume diphthongs ovc and ove ; in the other cases this dialect rejects the i tolerably early; τα for rai, του for τώί*. The Lesbians also as early as the fourth century begin to dispense with the t of the dative •\ Conversely the Euboean and Oropian Ionic of the fourth century weakened final ωι and ηι to ol and ei, reducing internal ηί before a vowel to the simple sound : Ιβρηον'''. In the a colony of Syracuse). But the Cor- iuthian vases {T)ial. Inschr. 3130, 3187, 3143, 3146, 3148, 31.52, 3156) furnish 12 examples of 01 and none of 0. Conversely the vases in the Chalcidian alphabet (Kirch.•* 124) in eight examples of such names have only one with t (Ξα/'θώι), the Attic vases not one (Meisterh.2 p. 109). 1 Cauer Curt. Stud. viii. 403 ff. ; Meisterhans p. 94-5. I may remark that δραχμαΐσι C. I. A. I. 48 is by no means certain, since the remainder of the line after ΣΙ is wanting. But μυριησι C. I. A. IV. 53% 20 (418 b.c.) is an error ; χιλίαισι stands do. 10, ταμίαισι 17. Also Ionic δεσπόνησίν Eohl 501 ; but elsewhere -ηισί. 2 Ahrens D. A. 193 f, Meister Gr. Dial. i. 249. 3 Dial. Inschr. 326, 4; m. ,50; iv. 9 ; EipoviSaios 345, 86 f. With B»ot. ττατροΐοζ compare Thessal. Κερδοίον, Ahrens, p. 221. * With the article still earlier than elsewhere : Eohl, no. 327 τάφροδίται τα τίειθοΐ (Dial. Inschr. 325). s Ahrens D. ^. p. 99 ; Meister 87 ff. Still earlier in the case of the article : Rohl, no. 503 TO for τωι twice. Like- wise in the dat. plur. of the article TOiS rats, in other cases -οισι, -αισι. "Άρωίδα3 Dial. Inschr. 281a, 37; 262; Έροίδα Assos Arch. Inst, of America I. p. 75. " Bechtel Inschr. d. ion. Dial. p. 9, 13 (Inscr. of Eretria no. 15, Olyn- thus no. 8, Oropus 18). 'leprjof Crop. 18, 33, 36. Bechtel would assume quantitative metathesis, as in the case 46 THE rnONUNGIATIOK OF case of the Athenians on the other lianfl wt and ai hold their ground almost entirely in the classical and also in the period immediately following; with \ωοι> and σωω, i.e. no doubt σωίώ from σωίζω as νομιω from νομίζω\ we must compare the numerous instances where at and ol lose their c before a vowel ; ot for ωι is hai'dly more freijuent than the converse ωι for oi^. The A.siatic lonians distingnisli correctly the conjunctives Χήβωισιν (Aorist II. with long thematic vo\vel) and πρηξοισίν (Aorist I. with short)^; exception might be taken to κοινοττίδης on the same Chian inscription, since καΐ ot- must by rights give K(vt ; on the other hand τοΙκόττεΒον on the same is correct, as + ot cannot give ot*. The case stands otherwise however Avith HI both for Attic and the other dialects. Dorian inscrip- tions shew very early for ηι, in the conjunctive for example, sometimes η (Ε) sometimes et^' ; in Bosotia €i is indistinguish- ably confounded with η and ηt ; on Ionic inscriptions the dropping of the ι in the dative, and the use of €t in the of εω from ηο : ηι to εΓ, and then to (L. In any case in this dialect it is impossible to consider the oi to have been an original locative, as many do in the case of Boeotian etc. (Eretria Έφημ. 1888, 8.3 ff. 1. 180 c ΣΟΙΝΑΡ- rO[T], i.e. Σοιμαύτου=Σωίν.2). 1 C. I. A. 11. 162; 12^ 7. (Others, as Cauer, p. 416 ff. and G. Meyer-, p. 470, take σώω as present form with future sense ; cp. σωον, θωά, Meister- hans p. 52.) 2 τοΐ δήμοι, C. I. A. II. 277 (ΤΩΙ- ΔΗΜίίΙ Pittalds) ; τοΐ δήμοι τοΐ Αιοι>ύ- σοι τρα^οιδοΐ$ by side of six instances of ωί Έ^. αρχ. 1884, 69 ff. (in the same are two instances of ο for ov ; ft alwnys for ψ) ; κωμοιδίαι, Kaibel no. 38 (ivth Centurj') ; conversely οίκοσί- τω«, C. I. A. 11. 834''; ii. 24 (n.c. 329/8), ίκάστωΐί (for ίκάστωι sing. ? Meisterh. p. 52) 258 (n.c. 304), ωΐνοχό-η 403 (iiird Century) ; στεφανώι 3rd sing. ind., HuU. de ctirr. hell. in. 120 (ivth Century). F,i)idauruR'R77/i. af)\ato\oy. 1886, p. 147, line 56 Άξιόχοι. 69 τοΐ. 238 Ύιμασιθέοι. 254 Άριστόνοι. There are however in this inscription other instances of confusion between Ο and Ω, as λυσίονι alongside λυσίωνι. 3 E5hl, no. 381 (Bechtel, no. 174) ; the correct explanation for the altera- tion of ωί, -ηι and οι, ei in the conj. on inscrip. of Asiatic Ionia (and Crete) was given by Schulze, Herm. xx. 491. ■* The modes of writing such as φνοχόο^ and φνο%, Eur. Cyd. 560, I cannot consider correct. Comp. μίσ- θιοι, μισΟόβι = μισθοί, διττλόοι = δίττλοΓ. 5 Ahreus Ι). D. 293 ff.; G. Meyer = p. 86. On the Xuthias insciiption (Sparta?) liiihl, no. 68; ΖΩΕ, ^ώτ; occurs twice by ΑΠΟΘΑΝΕΙ. On the pillar of Damonon (Sparta) R. 78, ΙΙΕΠΟΚΑ. The inscription of Gortyn however, which always has ι in the subjunct., has ΟΠΕ, so that these adverbs must bo jilaced in a separate category. ANCIENT GREEK. 47 conjunctive, is strikingly frequent^; at Athens from about 376 B.C., although the usage of Ionic Η Avas in other respects correct, EI (in isolated instances even E) was often written, and this orthography at the close of the fourth century actually prevails for every HI, e.g. ret βουλβΐ, Alj€l6 f. on 'Ολύμπιοι ϊπί ; οι' σνίΌλείφΐται ταντα αΚλ-ήλοΐί (the ι with the f). σιωπΐ] δι) τΐί η μίταζν άμφοΐν ylv(Tai κτέ. '' (iardtliausen Gr. I'tihcogr. S. 1',).'5, '20;{. 1 found the o'at ω'ω, in η Papyrus nis. of about the 7th cent., see Xtsclir. f. .Eiiijpt. Spr. 1880, p. 35. ANCIENT GREEK. 51 Section 14. Value and treatment of TI, EI, AI, OI. The remaining diphthongs with i, that is vt at et ol, have all of them this in common, that from a very ancient time they tend to simplification before following vowels: the l then in many cases disappears in script, and in pronnnciation had at most only the value of a weak ?/\ Of the spurious diphthong vt, which from the earliest period and during the whole history of the language appears only before a vowel, there is in Attica so early as the fourth century no trace whatever : there vo, 24 (iv. p. 8 Mein., ir 20 Kock). ■• Mfisterlians, p. 31 Π'. Examples for α e from old Attic inscriptions given by Cauer C. Stitil. viii. 2(!S ff. •■^ Ahrens 7). .1. 100; Mcister Or. Dial. I. 90 f. '' Ahrens D. D. 187 f. " K5hl I. dr. ant. 372^• ■'■"-, Bechtel In.-'ch. p()nax. '" C. I. A. II. 50 Κΐ'δ, twice; 573 llcip. four times witli a, once witli at. Meisterhans, \). 2."). ANCIENT GREEK. 5. aierosi''. Lastly for ot the best known and most frequent example is iroelv ποητψ, from Avhich comes Latin j^oeta ; στοιά also became στοά^, and in the Attic poets scansions such as τοιούτος, οίος o'iet with the first syllable short are frequent. If then in their final development the diphthongs at oi ei coalesced into the simple sounds e (that is ancient Greek η), U (i), Ϊ, yet it follows from this fact of the alternation between ai and a, at and o, et and e, that so long as this took place so freely, the first elements a ο e were still clearly present. A shifting also occurs in the converse direction : on inscriptions of the third century and on Papyrus we find written βοιηΒρομιών βοιηθόν 6•γ8οίης^, in Ionic are found Ααναίη ΐΐαμφαίης Φαι- €ννός, and ei stands for e in eweia and the examples, Attic and others, mentioned above*. Less frequently before con- sonants (στ, σ8 — ζ); τταλαστη in Attic, not ττάλαιστή^ ; Γβρα- στός and Τβραίστός in manuscripts^ Ύροζήνωι has inscriptional warranto Should any one on the other hand be inclined to 1 ΈλαΓαι, C.I. A. iv. 299% 7 (before 403) ; elsewhere with simple a, also IV. 53», 33 (B.C. 418). According to Cauer C. Stud. viii. 270 the origin is Aai-i'a, cp. also on au = at a. Wackernagel K. Z. xxvii. 278. For κάω, κλάω (Voemel Dein. contiones p. 36) we have no examples on inscript. (the mss. waver) ; on aieros Meist. p. 24, n. 142. The proper name'A€rtωί' occurs as early as 4th cent, at lasos (Bechtel d. Insch. d. ion. Dial. 104, 16) ; but nothing obliges us to take the a in this case as long. - στοιά Αγ. Eccl, 684, 688 ; στωιά Mitylen, Dial. Imchr. 273; Curt. Etijm.^ 216. Attic inscr. have only στοά, Meisterhans, p. 44, u. 8^4 ; στοιά Chalkid. Ditt. Syll. 369, 25. •* Meisterhans, p. 45 f. ; βοιηθόν Pap. L. 27 (2ud cent, rather incorrect); κατα- βοιηί by δότ; on the Papyr. in Ionic dial. (p. 44, note 2), which I place in 4th cent. Also ΈΙραξινοίη Bull, de corr. hell. x. 340 (epitaph of a woman of Halikarnassus in Khodes), βοίηθοί Calymua, Gr. Inscr. Br. Mus. ii. 298, 9. ■^ Bechtel d. Inschr. d. ion. Dial. uo. 99 (Miletus) ; do. Thas. inscr. in the Louvre, p. 26, 28. it for e ceases according to Meisterh. p. 46 as early as B.C. 250, accordingly much earher than Oi for ο ; Μ. refers this to the degeneration of the et to a simple sound. '" C. I. A. II. 167, I. 321 f. etc.; also 834'', 11 ; in 16, 33 at in Kuhler's transcription is only a mistake. " 0. Kiemann Bull, de corr. h. iii. 497. " Besides appearing on the snake- pillar at Delphi (cp. Thuc. i. 132) it is also constant, C. I. A. ii. 614; Wescher-Foucart, Delph. 4, 50 ; C. I. Gr. 1. 106 (whence?) ; Dial. Inschr. 3014 (Megara). Coins shew Ύροιξ: not be- fore Empire, earlier TPO, which points to T/)of., Foucart on Le Bas, ii. 33". 54 THE PRONUNCIATION OF infer from the Attic llorethedrai from TioreLhaia^, that ai had the same sound «is e, the answer would be obvious, that YloreLhaarai is wanting in euphony and out of all analogy, and that in Ionic also έτηστβαταί eSwearo were used instead of €7Γΐστάαταί i8vvaaTo'\ In like manner we arc not by any means to conclude from the censure which occurs in Aristopha- nes of an inelegantly broad pronunciation of κρύμαω, that the elegant pronunciation was already at that time κρέμηυ^, but rather that a drawling of the diphthong is the object of the reproof, or perhaps, since the latter is in this case followed by a vowel, a strictly diphthongal pronunciation as opposed to the more careless, which allowed the ι to become more evanescent. It is certainly possible to pronounce ai as well as ii in very different ways. For that ai was so early pronounced e and had become identical Avith η, appears to me in the face of the constant separation in script a pure impossibility, as also an identity of ei and i, of oi and υ ; a historical mode of writing running counter to the pronunciation is only possible, where there is a strict grammatical code, which at that period did not exist, and isolated blunders and shiftings make their appearance in spite of such a code, especially in the course of so many centuries. The only examples however which are brought are TloreiSeaTai, a mode of Avriting Avhich is as invariable as UorecSaia on the other side, and next in the third century a supposed inscriptional ^ένητβ, which does not exist, as the right reading is οττως ^γένητ €φρόντίσ[€*. This leads then to the 1 C. I. A. I. 240, 241, 242, 244; ηλίθων ΐφθί-γξατο καΐ τοίσι χΐίλΐσιν but no less also in 238, where the no. δκρρνηκόσιν. Curtius argues agaiust of letters shews that Kirchhoff is wrong Biich. in Stud. i. 2. 275. in supplying \\οτίΐοαι\α.ται. There is * C. I. A. ii. 379, 18, where Kiihlcr absolutely no example for the latter reads -γ^ητΐ φροντίί ; cp. Illi. Muk. reading; for ΙΙοπώαι-, 230, can just xxxvi. 617. In like manner v. Wila- as well be completed as IloretSaia (as mowitz in the letter of Attalus to the in the preceding list, 235). priest of Pessinus (Domaszewski Arch. " Merzdorf C. St. viii. 188; Cauer Kjyigr. Mitt. a. Oest. 1884, 05) c. IG do. 268. has corrected the ίπιστραφήσΐσθί kcI- •' Biicbcler ll]i. .l/iix. xx. 302 ; vovi of the editor into -σεσθ' ϊκ. (Lt'ct. Aristoph. Nuhcs 870 IT. : ΦΕΙΔ. — ti i'pi<}r. p. 16). κρέμΛίό ye. Ι^ΩΚΙ'. — ιδού κρέμαί, ώ$ ANCIENT GREEK. 55 argviments to be drawn from elision crasis etc. The at of most verb-endings is, as is well known, not only in Homer but also in the Attic comic poets and indeed in prose subject to elision \ This fact is explainable without diflficulty from the pronuncia- tion ai: in the first place legetaen was pronounced as kaen" and as Peiraeus ; but afterwards the a of this legeta was treated like that of ταύτα, λεγετ' ev like ταυτ ev, though ka Avas as will readily be understood not in general allowed to shrink up into L•', but here crasis was employed. If on the other hand it had been legetil (λεγεττ/, as in Bueotian), I see no possibility of the long vowel being elided. Further, crasis furnishes, as G. Curtius shews^ especially strong arguments for diphthongal pronunciation. Καί, ev = καν, ku en becoming kan ; how could ke en become kdn ? The same applies to κάκ€Ϊνος, κατά, καστιν etc. Now no doubt, where there is a frequent occurrence of a certain word-combination, a definite form of crasis might be handed down to a period, in which its elements, having in the intervening time suffered change, ought properly to give a different result : for instance θάτβρον is good Attic, though the form in use there is no longer ατβρος but €Tepo<;. But this is clearly not applicable to the crasis of καΐ with any chance word beginning with e or ei. For ot we have to consider, μονστίν from μοί eaTLv, μου^όκβι, μού'^κώμίον, σοϋόωκβν, καίτούστιν, then ωζυρέ, eycvSa, also θοΐμάτιον like θαΐμάτια, all phenomena as easily comprehensible on the assumption that ol = gi 0, as they are absolutely incomprehensible supposing ot to be u. With regard to the other dialects, giving a passing notice to the Lesbian dicereses such as 6ϊ8α ούκην (οΙκβΙνΥ I call especial attention to the Cyprian writing. This peculiar script, which is entirely inde- pendent of that of the ordinary Greek, being not an alphabet but a syllabarium, nevertheless expresses all the diphthongs in a manner entirely analogous to the ordinary script, a clear indica- tion, that this Avas conformable to the pronunciation, and a ' E.g. Deinarch. 1, 40 τταρακροΰονθ' the Ion. Papyr. (see p. 44, note 2) ύμαί (according to cod. Ν and A pr.) ; KEN i.e. κ iv twice (1. 6). 2, 3 ■γΐΐ'ήσ€σθ' αυτόν (according to Ν •* Gr. Curtius Stud. i. 2, 277 ff. pr., A pr.). ■* Ahrens D. A. p. 106. Meister - κα ev found in C. I. A. ii. 50 : on Gr. Dial, i. 96. 56 . Τ Η Κ ΙΊίΟΧΓΛΧΊΛΊΊϋΧ OF cuiLain proof, that the piOiiuiiciatioii Wiis diplithuiigal in Cyprus. Fur instance a-i-ve-i alPet (dei), tu-i τα, pe-i-se-i Treiaet (i.e. τβίσβί from τίνω) {rj is wanting), ma-to-i Μαδοί, tu-i τω, o-na-t>a-ko-ra-u ^Ovaaayopau, a-ne-u civeu, a-ro-a-ra άρουρα. Section 15. Tramfomiation of El, AI, OI, iii, Boeotian. The above however does not hold good for all dialects, and it is the Boeotian, in which we have already recognized the beginning of itacisni in the case of H, that in the case of these diphthongs also has anticipated by centuries the development as it took place elsewhere. The Boeotian sound-system, as referred to the -^olic, shews the following changes : .^ol. 7] ai at 6i T/t OL ωι Boeot. €t η η t €ί υ υ. These alterations however did not all arise simultaneously, and it is not till the inscriptions of the third century that we find them all complete, et, wherever it is really e + 1 and not e, is in many cases in the earliest monuments and at a later period without exception simplitied to ι ; in those cases, where it maintains itself, as in Baveiov ΑορκβίΒας, ηι appears to be original, and accordingly we find also eit (= ηϊ) in such w'ords, μάντευα WvTtyei/eii.o'i^. AI is retained in the earlier period, for instance the older coins of Thebes shew ΘΕΒΑΙΟΝ ; only in Tanagra and llysia• AE is written for ai and Γη (piite in the Latin manner : ^Αβα€08ορο<; i.e. -8ωρος ; eVi ' Afietro/cXetae*. The old Corinthian writing also had this diphthong, there how- ever the Ε was equivalent to ei; ΑΘΑΝΑΕΑ, ΠΒΡΑΕΟΗΒΝ 1 Ahrens D. A. 185. Moister Gr. Gr. Dial. i. 238; Plataiai Lolling Dial. I. 228 f. AopKUdas Orchonicnos Berlin. Motuitsber. 1885, 1031 no. 22 about 330, ])ial. Dixchr. 470 (ib. 502; ΐττΐ λaμaev^τoe. Terent. Scaur, vii. as regards the formation cp. Meister 1ϋ Κ. : autiqui quoiiue Graecorum in lii-zzenherijer'a Itiitr. vi. (11). banc syllabani per ae scripsisse tra- ■•' Foucart Bull. in. 130; Meiater duutur. Λλ^ΙΕΝΤ GREEK. 57 WOavatia UepaeioOel•^ ; ΑΕΘΡΑ" also is fuuiui on a, vase which is probably Attic; on the other hand the Thessalian Ααρισαέων on coins of Larisa comes not from Λαρισαίος, but from Aα/^ίσα(t)ey9^ At Tanagra we find also corresponding to AE the writing Ο Ε for ot and ωι: Μοεριχος, UoXvaparoe ; even Priscian compares this Boeotism with Latin ue*. I consider AE OE as real diphthongs; for not only in Latin bnt also in Cymric (Welsh) there is a diphthongal ae oe, distinct from ai oi, thongh nearly approximating to them^; the Corinthian Avriting however corresponds exactly to the Oscan HA, HV, -| beino- the sound midway between i and e, just as V expresses that between ο and u, — Afterwards however the Thebans adopted the Ionic Η for ai, even before the introduction of the common alphabet, Te\eaTrjo<;, Άρίστηχμος'^, and this is subsequently the regular mode of writing everywhere in Boeotian At this time therefore te time was pronounced with the simple sound e both in the dative singular and in the nominative plural ; for at and at also at this time Avere not distinguished, ot on the other hand remains, if we except Tanagra, not only in the fifth but even in the fourth century, and even subsequently was not ousted by the simple writing L'^ On very late Boeotian inscriptions we find et, which in other instances appears on these with the evident value of I, alternating with υ as in ret, At ret βασιΧα, (for βασιΧύί pron. basili), κή τΰ Ύρ€φωνί[υ] (or Ύρβφί ,r-> ? )• ^ Eubl I. Gr. ant. no. 20, 4, 5 ; cp. above, p. 29. - C. I. G/-. 774(3; on the other hand 9o'pae (Ahrens i. 199, 3) is a wrong reading for (}όραξ, C. I. Gr. 7374, Dial. Imchr. 3127. 3 Fick Dial. Imchr. 360 ; Beer- mann, Curt. Stud. ix. 34, compares the two forms ΙΙεΧίνναιέωυ and l\e\i,vvaiwv from ΙίίΚιννα. ^ Prise. Instit. 1 § 53. Kpoecros on a vase, probably Attic, C. I. Gr. 7756, Welcker Alte Denkm. in. 481 ff. ^ E. Lepsivis Standard AlpliaJtct p. 172. 6 Foucart Bull. in. 136, 140 (Eohl no. 300, Dial. Imchr. 700). '' Accordingly it must be regarded as due to intermixture of the κοινή, that in the Theban Proxeny decree in favour of a Carthaginian (Dial. Inschr. 719) at is written throughout. '' Ahrens D. A. 194 1ϊ. , shewing the local differences ; Foucart 1. c. 133 and iv. 88 ; Meister Gr. Dial. i. 235. « Dial. Inschr. 382 ff. (Chaironeia), 429 f. (Lebadeia) ; these are all dedi- catory documents relating to slaves. The example cited 429, Ύρεφωνΐ Meist. 5δ ΊΊΙΕ PRONUNCIATIOX OF ΕΙ i'or ΟΙ occurs also sporadically in late Attic ; οϊκ^ι, for υϊκοι (icciirs in Mcnander, Zvelv is frcuncnt, roh Χοιττβΐς is found on an inscrip. dating 100 a.v^ The Boeotians did not readily admit ν instead of ol before a vowel, e.g. in Βοιωτοί^; according to the Grammarians also ot not ν was substituted in Boeotian for ωι'. All this is very mysterious and perplexing. If the Boeotians finally pronounced ti Di, one cannot understand, why in the first word they always added the E, against the pro- nunciation and against custom. The EI must it would seem have been an attempt to imitate the sound, Avhich appeared to their ear something like ei, just as in the Attic hveiv. If this is the case, Ave shall have for the foundation of this ei in a preceding stage a diphthongal oi, not a monophthongal ii, and we must suppose, that the early fluctuation between OI and Τ represents a fluctuation of pronunciation. For ol and ν are closely related to one another both in ancient Greek pro- nunciation and that of the κοινή, as we shall shew hereafter But EI is confined to the endings: ποαομβνβί or TTo-LOfieveL = ποιούμενοι, ; in these endings (as indeed also in the sten)s) in Latin also υί has become ei {i); Nom. plur. oi -ei -i, Dat. υϊα -eis -is. — The view held by Curtius and Dietrich*, that in Boeotian oi first became ui and then il, is contradicted both by the ancient OE, as also by the fact that TI was never written', although, in accordance with the value of the Τ prevailing there, this would have been the adequate expression for ui. (383 oiwv stands for νιων, with omission of υ, which we meet with frequently elsewhere, JltKoXoj Έ,ΰφροσόναν 386). 1 ακΐί Herodian i. 504, IC ; ii. 4(53, 31. Αυύν on Attic inscr. Meisterhans, p. 124. Toh \oLTr(h C. I. A. ii. 467, 12 f. = Dittenb. Si/ll. 347. Cp. Φαληρί, p. 32, n. 2. - Βυωτων, title of Athena Itonia, Ihtll. de con: hell. ix. 430. The case form TOIL can become τνΐ as well as Tot. ■• Ahrens D. A. 1!)3 f . ; Meister (Jr. Diiil. I. 21'.( f. Tlie case will be found to stand thus ; ττατροΓοϊ etc. (ol before a vowel) was always found in Corinna ; that in the same rot for τφ could have stood by the side of ΐμύ τν (=τοί, οι) is not credible. •• Curtius Gott. ydclirichtcn 1862, p. 495; Grdz.^ 706; Dietrich Fleck- ti'.st'ft'*• Jahrl). 1872 p. 24. On the other hand Beermann (Stud. ix. 41 f.) decides for the transition oi, i). ^ M^ri'iKos is said to be found on one of the tablets of Styra (Kulil, no. 372-'"'); see liowevor Bechtel liisclir. d. ion. Dial. p. 18 (no. Ii»""). ANCIENT GREEK. 59 Section 10. Later Situplification of EI tu i (e). We imist now with reference to the remaining dialects and the Greek kxnguage as a whole separate the diphthongs which have hitherto been treated in common and first of all give our especial attention to EI. Outside of Boeotian the examples for an early simplification of this diphthong to l are not numerous nor are they sufficiently trustworthy. We must of course place in a separate category abbreviations in particular words, as for instance in the month-name Yloathtcov, which in Attic also is always so written ; the same is true of recurring forms such as Χίρων for the Centaur, which may lay claim to pass as correct \ The ending -/cXtS?;? for -«λε/,'δ?;? may be derived from -κ\ος not κλη<;, accordingly even on the lead tablets of Styra^ in Euboea, which seemed to furnish the most numerous examples of t for ei, nothing more is left which can be considered trust- worthy. But Avithout doubt fiOm the end of the third century onwards, EI, and that both genuine and spurious without ex- ception, was simplified in this way in the most diverse regions of Hellas. Our evidence for this is drawn from the same documents, from which we gained our information on the fate 1 C. I. Gr. 7400, 7687, 8185, 8287, 8359, (7870). For Χαίρων there is only one untrustworthy example. Cp. Meis- terhans, p. 43. - Kretschmer Ztschr.f. vghSpmchf. IX. 159, on Χ]αρικλίδαί of a Corinthian vase (Dial. Inschr. 3121) ; Bechtel Inschr. d. ion. Dial. p. 36 (Lead tablets). On Φιλαι-γίρψ on a tablet (Rohl 372^82; B. 19 "S) it must be noted, that Aiyeipa in Achaia, whence the name comes, api^ears when cor- rectly written to have an t just as Στά- ytpos, ΑΙΓΙ, Alyiparav on coins (Fried- lander Z. f. Numixmatik p. 6; Cata- logue of Greek coins Peloponn. p. 17), Αϊ'γφα Inscr. of Lagina Bull, dc corr. h. IX. 444 frg. M, 2 col. 3 (81 n.c); Myipbiv proper name Inscr. Epidaur. "Άφ-ημ. αρχ. 1887, p. 9 ff. 1. 38; Αίγι- ράτα Wescher-Foucart 109. On the other hand Αίγεφάττ?? Oropus Έφ. 1885, p. 97 ff. 1. 2 and 31 (soon after 115 B.C.). The Bceotiau Φιλαί7ίρα[ο] Eohl no. 382 (Dial. Inschr. 566) is not evidence for either side of the question. On the lead tablets there still re- mains UipLUos (ΙΙειρίθου$) Ε. 312, but guarantee for this reading is wanting, cp. Bechtel on no. 297 (p. 29). I may also cite: δαρίκ05 Sparta E. no. 69 (only in a bad copy of Fourmout) ; 'ApiffTLoas Sparta ib. 84; 'Ή.ρακ\ίδηί Halic. Unll. de corr. h. iv. 297 B, 8 (elsewhere the same occurs Λvith ei ; cp. Ά/ίχαγόρω for -ρεω ib. 1. 3). 6ο THE PRoXUXriATKJX OF of the dijihthongs at wt, ηι, namely the Delphian manumission documents and the Egyptian Papyri. These writers of the second century were in perfect ignorance, where they ought to put t and whore ei, and wrote Έ,Ιρις, ret/ia? and conversely τταραμινάτω and ί€ρΐς\ and if the Delphian masons at least left the short t to itself, the Egyptians allowed this to be mixed up in the universal confusion, ίμί and iareiv, σνμ- φωνουσ€ΐΐ', ηΧβως, oret, μβίζονβν^. There are however two exceptions. First a ρ i'oUowing exercises a certain protective power over the E-sound, which is seen also in Latin and in modern Greek (^epo?) ; accordingly Σωτήρα and %epa are written, just as εκεχηρία is found on an early Delphian record Avith what appears to be strict Doric η, for which however the Delphian dialect has ei^ Secondly the ordinary ccjuivalent before vowels is η or e, μηνιήα ττΧηάς τταιΒήα οΙκηότη<;, TTopeav evBea^ ΒιΒασκαλέα*, and this holds its ground for a long time, so that e.g. in a decree from Byzan- tium of the time of Tiberius^, ■χ^ρήα'ζ, ττΧήονας, ετητάΒηον, 1 Wcscher-Fouc. no. 108, 435, 82, 365. Cp. Hull, de coir. Ii. v. 42,= Dial. In.schi: 153!), Ditteiib. 291, Pho- kian official record from beginning of 2nd cent, (after 181 b.c, Dittenb.), in which Στίρι occurs (by Στεφίων) and κλαμωσϊ, and further what was said above on the final confusion of Boeotian et with t. — For Athens cp. Meisterh. p. 38 f. - Papyr. 1 of the Louvre, which is by no means the most incorrect, fur- nishes these and other examjjles. In the Papyrus imblished by Π. Weil 1871), which contains fragments of Euripides and other poets, we find inei^fTri, irtpd, χάραν (χάρίν), βασα- Χύσσηί, Κύπρ€ΐδοί and others without number. ^ 'Σωτήραι (-pat>) ( '. /. .(. II. Kill, 22; III. 308; χ^ρα Papyr. L. no. όΟ τή Xfpei, x^po-i, Xfp^^ (•ΐ"ΐ^ X'P"•") I *J1 ^ol. ,'> xipo.; do. I'ap. Lond. (Wessely Wivner St. 1880, 2U3) -X-Liv. 11. C}). άηηρον Scott Frarjm. llcrculan. p. 219 f. (col. 15, 29); but the same gives also δηνα for odva twice Append, xxxviii. 1. 9 (Philod. 7Γ. θανάτου Δ col. xxxvii.). This calls to mind the Lat. i pingue, written ei c i. — Lat. cy2)erit^ (-um) Kvw€(i)pos ; but Epirits pirata etc. — "Ε,κίχ-ηρία. Ani- phictyonic decree C. I. Λ . π. 545 1. 48, 49 (the Η which was suspected by Ahrens has been confirmed for both places by U. Kohler and by myself); Έϊύχήρου Delph. Dittenb. Sijll. 198, 73. •* The examples are from Pap. 1. both sides (on reverse side = Pap. 03, col. 2 xpia^). 'ίιΐηνι-ήαν I'ap. 1, col. 15, wrongly read μηνικαν by editor and wrongly emended to μψιαίαν ; cp. μηνιύοΐί Pap. 01, col. 8. — At Delphi άΐ'δρΐον, yvvaiKCov, Κα\\ικράτηα etc, '• C. I. dr. 2000, Dial, lim-hr. 3059. (Επαινησθαι ib. not^eTron'fiX^at but a perfect, cp. Dial. In.iclir. 3l)7S, Dittenb. 240, 92 etc.) οϊκήον ΐΓολιτ-ηαν (.\murg. ) Bull, dc coir. Ii. vm. 445. ANCIENT a η Ε Ε Κ. 6ΐ άσαμ,ήωτον are written consistently. In like manner an inscription of Cos shcAvs Καισάρηα, ^Χ^ρίτΓΤΓηα, ΉρήκΧηα, on the other hand ^ΑττοΧλωνβια, ΆσκΧαττβΐα, Αιονύσαα, Αώρβία^, where the pronnnciation was -ia, being in some cases original, in some cases owing to contraction, ^ΑσκΧαττίβια AsJdapia. For in those cases where this ei is preceded by an t, either ιη can arise, as in μηνιηαν and Ί,αραττιηον on the Papyri, or again %, as in vulgar v'yeia (hi/gla) instead of vyteia, ταμ€Ϊον instead of ταμίεΐον'\ There is no especial degree of consistency to be found in the Latin representation of et before vowels; Aeneas, Medea, Alexandrea and Alexandria, Darens and Darius, Clio, Iphigenia ; in words taken over at an early period shortening occurs : platea, halineum βάλαν€Ϊοι>, in general even at a later period e predominates^ The un- certainty of the Greek pronunciation is sufficiently established by 67Γίτ/;δίο account of the derivation from ρίω; Schneider Ausf. Gr. i. 69 ff. also deh instead of θ'α on account of •» Gomperz Wiener Akad. Bd. 83, θύνεσθαι. As regards the supposed con- p. 91 f. fusion of et and η there is need of greater ' Terentian. Maur. v. 441, 458. discrimination; for instance I cannot 6 Meisterh. p. 37, n. 30G (we must admit, that on the Pap. in Wessely however notice here ν(ών = νηόν, ναόν W. Stud. 1882, 175 ΰπόλήψΐων ίνβκα on the Koman inscript. of Herodes (sic) is = ύ7roλetΊ/'eωI' (cp. ib. p. 19C). Of Attikos, C. I. Gr. 6280, v. 96; also von Herwerden's examples some may be 62 77//•; PRONUNCIATIOX OF even if in the second century H.C. care and culture were still able to give not only tut and (u, but also ei its due and no more than its due', nevertheless this soon ceased to be a possibility, and the distinction of et and ί had become a crux o)iliographica^. In many cases the resource adopted was to write ei in all cases for long ?', as on the inscription of Byzantium, Avhich lias been cited, τ€ΐμ(7<; and ττολβ/τα?^ regularly; this however never became a universal and fixed mode of writing, and the Gram- marians, especially Herodian, took pains in the opposite direction everywhere to ascertain and carry out the historical method. Even at the present day an orthographic correction is nowhere more frequently necessary than in the case of ι and et. For instance we Avrite ι wrongly instead of the diphthong in the following words, βτβισα τβίσω and in all the derivatives of τίνω'^; μζίηννμι, μείξω, Mei|^ia?, etc.", ΦλείοΟ?, ΦΧβιάσωί, explained grammatically, very many contain η for et before a vowel (p); ητα for dra C. I. A. iii. 39 is found in an inscription which is very imperfectly handed down ; finally Χολλήδη? π. 82 shews η for ηί. — In Latin there are certainly some examples, where the Ε sound has remained even when fol- lowed by consonants (liypotenusa, Po- lycletus), ^ Of the Papyri of the 2nd century the following are correct and trust- worthy in disputed questions : Louvre 2 (dialectics), 15 (judicial verdict), 22 (petition); Taur. i. (verdict). Also on inscriptions: Olbia C. I. Gr. 2058; Delphi null, de con: h. v. 157 (State- record). The inscription of the Mys- teries of Andania (93 n.c.) has only one blunder άποησάτω 1. 78; for the writing ύμάτιον (Ιματισμός is conform- able with the dialect. With regard to Attica in 2nd cent, see Dittenbcrgor Ilenn. i. Ill ; Meisterh. p. 38, according to whom the confusion properly begins there about 100 n.c. - Mar. Victor, p. 17 K. says, ortho- graphia (iiiicfOruiii ex parte maxima in ista littcra consistit. nam...et in quibusdam mediis interponitur verbis, ut "λίδηί, et in extremis, ut (ί'χηι et τΓορεϋψ, et dativis casibus adjungitur, quamvis non enuntietur; et eadem subjecta e litterae facit longam sylla- bam et. •' In like manner e.g. C. I. Gr. 1798 (Epirus), 2059 (Olbia), 2335 (Tcnos). Cp. Quintil. i. 7. 15 (cp. p. 10, n. 2 above); Priscian i. 50: quam (ei diphthongum) pro omni / longa scribebant more antique Graecorum. (Fairly regular in the Greek text of the Moman. Ancyr.) ■* Sauppe de duobus iitiilta Teeculiiir jihuiiitir .iin;;le. develnpinent.) ANCIENT GREEK. 65 (iWa ττρίν el-Keiv τούτο σαφώ^ '^ί%^' Φ'?^''' '^'''^ άλλο? €-)(^£iV. Accordingly the mention of echo applies to the repetition of κα\ό<ί and there is no longer any question of a harmony of soniul between ναιχί and βχβι. Should any one however prefer to take it as an instance of parechesis, nothing is easier than by reading φησί τα ίίλλο? "€χω" to restore such between 'Ηχώ and βχω. But a positive refutation can be given in the fol- lowing manner. If in the time of Callimachus there was no distinction in the most cultivated court speech between the sounds at and e, in the vulgar speech of the second century there can have been absolutely no difference whatever. In that case however uneducated writers must of necessity con- found at and e (or η) in the same degree, that they intermix ei and I ί, and ω and so on. What then are the facts of the case ? The somewhat incorrect astronomical papyrus in the Louvre has opuTe^ for όραται once. The fragments of writing on the reverse side of the same shew no error. The same may be said of papyrus No. 23, where besides έστβίν etc., άσττασάμενος την μάγαιραν stands for σττασάμβνος and ττίνοντβς for ττεινώσαι. On the other hand on No. 43 we find 'έρρωσθαί for -σθβ and elSPjTai ; on 40 α>γορασ€Βωκ6 = dyopaaat εΒωκβ, with which may be compared γ^ρωνθ^ ώ?, τΓβριωκοΒομηκεν' αυτονς on another Pap^rus'l On Weil's large papyrus" τηστβύσβταί stands for -σατ€, 6ΚΤ6τατ€ for -ταται, βαίνβταί for -re ; for του μβν ξζναικβιν for του (το) μη '^eveyK€ti>* is an unintelligible cor- ruption and cannot be regarded as evidence. And never- theless these bungling copies bristle with the most crying confusions of et and ΐ and such like errors. Accordingly it is quite plain that the at of the verb-endings -σθαί -ται sounded in the speech of the uneducated like the e of the endings -σθβ -re ; but then these are cases, where the diphthong was from of old liable to elision and had no influence on the accent ; the representation by e not by η may be to some extent connected with this weakening. But we nowhere read ημέρη {-pe) for -pai, or κή (/ce) for και, or 7]ρώ for αίρω ; on the contrary eXav is the 1 Pap. L. 1, col. 17, 11. ^ Col. 5, 13; reverse side col. 4, 5, 19. 2 Wessely W. Stud. 1886, 206. •* Col. 4, 17. P. 66 THE PRONUNCIATION OF shortened form of eXacov, as Έαραττιήν of -Trtetoy'; it is there- fore quite impossible, that ai was at that time universally confounded with e η and had ceased to jjreserve the yl -sound. The contemporary inscriptions are perfectly free from examples of interchange, even those from*"' Delphi in (jther respects so incorrect ; on those from Attica the confusion of at and e cannot be proved before the second century a.d." It may be mentioned that where in the period of the Empire at is written as e, this is expressed not only by e but also by η, for instance on an inscription from the Thracian Chersonese we find κή tAvice side by side Avith jvveKt /Ήφηστος* on a pap}Tus. Dionysius of Halicarnassus furnishes an unmistakeable testimony for the correct pronunciation of the Augustan period ; he says that καΐ 'Αθηναίων in Thucydides is a case of harsh compo- sition, since the sounds of the t of καΐ and the a of 'Αθηναίων could not blend into one^ Demetrius the rhetorician declares the name Αίαίη to have a particularly harmonious sound®, surely however not pronouncing it eee. In the next place the Grammarians describe at in contra-distinction to a as η at Si- φθογγος ή εκφωνούσα το ι\ an expression which, to say the least of it, is very ill suited to at^e; for in that case why should it not be η Ισοδυναμούσα τω η ? This description caused even Aldus Manutius^ to recognize and insist on the distinction between the modern Greek proniinciation of the diphthongs and the genuine ancient sound. — If then in spite of all this the ' Pap. L. no. 31. cvvexes τηί αρμονία^ και δι^στακί, πάνυ - That I may pass over nothing, αίσθητόν τον μ(ταξν λαβονσα χρόνον, Ι notice the Rhodian verse inscription άκέραστοί re yap ai φωναί τον re ι και τον Ατταιλου (^ΆττελλοΟ?) in Άθήν. ιιι. α, καΊ άποκύπτονσαι τον ηχον. 22G. On the inscr. of IMylasa C I. *"' Demetr. π. ΐρμ-ην. % lii): πολλά δί Gr. 2693*^ (Rhodian money ; no Roman και δια μονών των φωνηέντων σνντΐθησιν names), 20!Ι3^ κέ, Άριστενέτον, Έπένετοί (scil. ή σννήθ€ΐα) ονόματα, οίον Αία/ι; do not occur at all; see more correct καΐ Enos, ονδέν τε δνσφωνότερα των copy Le Bas v. 416, 414. αΚΚων cotI ταΟτα, αλλ' ίσωs και μονσι- 3 Meisterhans, p. 26-. κώτερα. * Bull, de corr. hell. i\. 5li; K&ihe\ ' B. A. p. 1211; more correctly Epigr. 372 ; Pap. L. no. 19. Cp. how- elsewhere ή ai δίφθ. ή ίχοΐ'σο το ι ίκ- ever ρ. 38, η. 1 above, ρ. 69, η. 2 below. φωνονμκνον. ^ Dionys. ν. συνθ. ρ. 167: η των *• Cp. above, p. 2. φωνηέντων ιταράθίσΐί — διακίκρονκί τό ANCIENT GREEK. 67 o])iiii(jn ]trevails with regard to this very diphthong, that it had l)t'C()iiie simple e at an early period, the I'cal reason must be sought in the fact, that it represents Latin ae and is represented by ae^; for even Corssen gives to this ae the value of German d. But it is just as reasonable to draw inferences from Greek at wdth regard to the pronunciation of Latin ae, as the converse, since express testimony to ae = e is only to be produced from the period of the late Empire^. In the first place it seems to me certain, that A Ε was originally intended to represent a diphthong, just as much as the Cymric ae mentioned above. In old times the spelling at prevailed . in Latin also ; afterwards however an e-sound Avas thought to be heard in the second element, or rather the intermediate sound between e and i, ofteii written ei] hence arose about 200 B.C. the spelling ae, about 130 B.C. aei as in conquaeisivei, Gaeicilius^. This latter corresponds exactly to the Oskan, where the i tending to become e (S) stands as the second element. Now the difference between such an ae and ai is sufficiently slight, to cause the one to be readily substituted for the other in transcription. Moreover the Greeks are not the only people who have heard in ae a diphthong similar to ai, but also the Ancient Germans, as is unmistakeably shewn by the living pronunciation of German Kaiser derived from Caesar. If so early as Varro's time there w^as a fluctuation in isolated words between e and ae, sceptrum scaeptrum, faenerator fenera- trix^ (and pretor and Cecilius are given even by Lucilius as examples of countryfied language^), this is in no way different from the fliictuation prevailing at the same period between au 1 Except in words borrowed at an ae, formerly written with ai) e yiovis- early period such as Aia.t, Ma'ia, era- sima sonat. At that period then (that ΐηύα κραίπάλη. of Hadrian) it was not yet sounded as - So Terent. Maur. v. 490: hanc a simple e, but a followed by e. Seel- enim (the diphthong e«)siprotrahamuR, mann Ausp7•. d. Lat. 224. a sonabit, e et u (that is ae (e) the •* Corssen ^ii.s.s;2^)•.- 1. 676; Seelmann lengthening of e (e) will be the first 167. element). Sergius in Donat. i. 520, 28 * Varro L. L. vii. § 96 (cp. v. 97); κ of e ; quando correptum est, sic sonat in fen. e is original, Corssen^ 327. quasi diphthongos. But Terent. Scaur. ^ ib. VII. IG κ : sed magis in illis (words with 5—2 68 THE PRONUNCIATION OF — (u): planstTiim plostrum caudex codex, claudo cludo, Claudius Clodius. Whoever then does not deny, that the Romans pronounced mi as a diphthong, must allow to ae the vahie of a real diphthong. Moreover Varro by no means says, that the writing fluctuates between sceptruin and scaeptrum, but : partim dicunt sceptrum, partim scaeptrum, and we must interpret what precedes in accordance with this : in phiribus verbis A ante Ε alii ponimt (in pronunciation) alii non'. Should the question be asked, why the Romans made scaep- trum scaena out of σκητττρον σκηνή, if they did not pronounce the sound as skena, but rather as skaena, I suggest that these forms shew an intermediate form between the σκητττρον σκανά of Magna Graecia, which the Romans received first, and the σκητττρον σκηνή of the Koun] which reached them at a later period. For although η = e, no Roman of ancient times thought of writing Daeruaetrias or thaesaurus, but ae for η is confined to the two words in (juestion, in these however and especially in scaena the writing is almost without exce])tion. Diphthongizing has also taken place sporadically in austrum = ostrum {οστρβίον) Άπά in Latin words such as auscidttm (faenus faenum); just as ai — e, so an — ο lie very near together in sound, and foreign words adapted to popular use are especially liable to peculiar treatment^. It is also worthy of mention, that Latin poets occasionally scan Phaethon as a dissyllable, by no means however with a pronunciation so remote from the original sound as PhetJwn : Quintilian calls this συναίρ€σι<ί^. At the period then, in which Latin ae became the simple sound, that is in the third and still more in the fourth century*, the Greek at also had suftered the same fate''; but up to that time at and ae may be considered to have preserved their 1 See also Gellius xvi. 12. 8: deiectum fulmine Phaethoti. Nam si (Varro) M. Catonem et ceteros aetatis esset prosa oratio, easdem litteras eius fenenitoirin sine ti littera pro- enuntiare veris syllabis licebat. nuntiasse tradit. •• Corssen i.-' p. 692 f. Seelmann ■-i Prise. I. 52 ; Seelmann p. 1G8 f. 224 f. 3 Quintil. i. δ. 17: quod aivaipfcif '' In Coptic loan-words f was writ- et σνναΧοιφήν Graeci vocant — , quails ten, Storn Kopt. Gr. 36. est apud 1'. Varronem: tuin tv jUigmnti A NCI Ε XT GREEK. 69 character of double sounds, not indeed in the mouths of the people*, nevertheless in the cultivated speech. The oldest testimony as regards at = e, corresponding to that of the later Latin Grammarians on ae as the lengthened form of the open e, is to be found in the treatise of Aristides Quintilianus irepl μουσι,κή<;, which is placed by some in the second, by others in the third or even the fourth century, but which judging by the names of those to whom the author dedicates it, Eusebius and Florentius, certainly cannot belong to the second I The evidence drawn by the followers of Reuchlin from transcrip- tions in the Septuagint is quite worthless. For the fact of Bethel being written ΒαιθηΧ and Elam Αίλάμ"' does not shew that αϊ = e, but rather, if indeed it shews anything at all, that Hebrew Tsere with Yod quiescens was represented by at. In the first place it ought logically to have been written ΒαιθαίΧ, if the sound were the same in both syllables, and in the second place the combination of Cholem with Yau quiescens is perfectly analogously represented by av : Χύνάν Onan, Ναβαύ Neho*. Finally this point too does not appear to me prove u, that so early as the second century A.d. Herodian had given ortho- graphic rules on at and e^. For why not also on η and at ? Η was at that period certainly still e. There are moreover at the 1 The wall inscriptions of Pompeii shew the greatest confusion, both be- tween ae and e, and between at and e. For example, sometimes c/?!rtffZi/*• some- times cinedus ; no. 1684: etati maeae, haberae ; 733 ΐνθάδαι κατοικεί, μηδέν eiaeiairw (i.e. ΐίσιέτω, (ίσίτω) κακόμ (here too it is evident that Lat. e Gr. η — e, Latin e Gr. e=e, cp. p. 37, n. 5 above). 2 Aristid. π. μονσ. p. 56 Jahn (93 Meibom.): το 5e e θη\υ μέν έστι κατά το ττΚΐΐστον ώ5 ττροΐίρηται ("has a femi- nine character in contra-distinction to the masculine and the neutral α"),τφ δε τον ομοίον ηχον ϊττιφαίνειν, ei έκτα- θείη, Τ| αϊ δίφθό-γ-γφ, -γραφομένΎ} δια τοΰ ά, €ιγ' ίλάχιστον ( " in a very slight degree") -ήρρένωται. — As regards the period of Ai'istides, cp. Jahn in the intro- duction ; what the latter says p. xxx. f. against Casar's argument from the names, has not the least significance. ^ Frankel Vorstudien zur Sejitita- ginta p. 115; O.de Lagarde Onomastica aacra. Βηθ- (Bei^-) is fdund for Βαιθ- in other names, but -η\ (simple Tsere) is never written -aiX. * Frankel ib. p. 116. ^ I must here run counter to the authority of Lenz, who tries to prove (Herod, p. ci.), that H. has given such rules, and who accordingly collects from the Byzantine writers everything having reference to this in the frag- ments Trept 6ρθο•γραφία$, while he sets aside their• rules on 77 -ei -i, 01 -1;, -ω (cp. p. cii. f.). But the proofs are neither numerous nor sufficiently strong. 70 Til κ PRONUNCIATION OF present time hardly any instances of uncertainty of writing with regard to at and the E-sounds. It is a ridiculous thing, that the name of the well-known Athenian, Avho fell at Marathon, is written Κνναί<γ€ΐρος instead of Kvveyetpo^, in Avhich latter spelling it gives the intelligible sense " urger of the hounds " and may be compared with Κυνόρτας. According to Moeris tooth-ache is in Attic ημωΒία, in Hellenistic αΙμωΒία^; but the Attic form is perhaps an invention of someone who found the imperfect of the verb αίμωδιαν written HMΩΔIA^ The form σημαία {standard) for σημαία is erroneous : all the older inscriptions such as the Monumentum Ancyranum, and also the oldest manuscript of Polybius, shew either -et- or, which comes to the same thing, -η- or -e-, which latter form explains the false -αί-^. The extraordinary contrast to the confusion in the case of El -I is uumistakeable. Section IS. Subsequent history of Ο I. 01 appears to have become confounded with ν at about the same time, that ai was confounded Avith e. It had never been very far removed from this sound ; if the attempt is made to Steph. Byz. Άβάκαινον : πάλα ^tuiXLas, ούδβτέρωί καΐ προπαροξυτόνωί, καΐ -η παραλήγουσα δια διφθοΎ^/ου, ώϊ Ίίρ. ev iy irepi ουδετέρων. Ai'e these the ipsis- sima verba of Herodian, or has he not rather merely set Άβάκ. under the neuters in -aivov? Theogn. xii. 20 (Lenz II. 409) etymology of χαίτη from 'Up. iv TTJ όρθο-γραφίφ. Is it really likely that he intended by the ety- mology (from κρατώ κράτη) to guard against the barbarous writing χ^τη? P. 410, an etymology of άχρι is cited from the same work. Eustath. 1H'.I2. 23 (L. ib.) on -γαιήοχοί and γεοΰχο? -γηοΰχοί, from Didymus and Herodian. This is an isolated case if one at all. The 4lh passage (Jo. Alex. 18. 23) Lenz himself ceases on mature con- sideration to reckon as belonging to the fragments of Herodian. And now with these compare the abundance of instances, even out of nepi μονήρονί λ^ξβω^, in the case of et -i, φ -α etc. ! In the same way Marius Victor, (see above p. 62, n. 2) says that the orthography of the Greeks had to do for the most part with t mute and « ; there is no mention of ai. ' Mocr. l',)8. 15; ai,a. is inmany cases the traditional reading in Aristotle. - Timokles in Ath. vi. 241 a uses the form •ί]μωδία in such a context, that any one might well take it for the sub- stantive. ^ Ditteubcrger Sijll. p. 4s'J. ANCIENT GREEK. 7 1 pronounce ol really Avith the closed o, as must be done in accordance with what has been said above, the small interval separating it from iJi will be remarked. Consequently Eustathius may be right in seeing intentional alliterations in the Homeric — Λτύλλ?; κοίΧης, χίάρνβ8ι<ϊ avappoi^SeV, and there is a close connection between words like Xoiyo^ Χνγρός, κοίρανο<; κνρίος'\ Accordingly there is no more need to assume any intermediate step, in order to explain the common Greek transition of ol to v, than to assume such a step between ai and e. The transition through id assumed by Curtius and others was destitute of actual traces even in Boeotian ; that through ο must be decidedly rejected both for that dialect and for the Greek dialects taken as a whole ^ For it is always open to suspicion to enrich a language with a new sound taken from other languages ; moreover ο that is the sound intermediate between ϋ and e is no nearer to gi than is il, Avhich forms the middle point between u and i. Latin oe, by which ot is regularly represented except in Troia and anquina {dyKotvaY which were taken over at an early date, was in my opinion^ just as much as ae and for as long a time as the latter a real diphthong, but afterwards passed not like ae into an open but into a closed e^. Whether it was at any intermediate period 0, I do not venture to decide; still it seems dangerous even here to assign this special sound to such an extremely small number of words in the language. As regards the time of the transition of oc to v, we find isolated examples of the simple spelling so early as a papyrus of the second century B.C., but only where it is accompanied by very negligent orthography and grammar: avvyere, άν^νω\ The later inscriptions in general interchange ol with ν in the same degree as at with 1 Eustath. on II. A. 406, Od. μ. 104 Seelmanu Auspr. 226 f. hold the same (long ago cited by the followers of view. Keuchlin). ^ 'jjjjg jg shewn by its representa- - Curtius Etijmol.^ p. 658 f. tion in Romance bye (Diez Gravun. i.- ^ This transition is favom-ed by 170), Avhile «c con-esponds to Romance Beermann, Curt. Stud. ix. 41 f. ie ; oe and e are treated entirely alike, * On anquina see Boeckh Seeiuesen and c was closed, Schuchardt Vulgar- 1,52. latein iii. 151. Seelmann 227. 5 K. L. Schneider Gramm. i. 1, 77, ^ Pap. L. 50 (160 b.c), 51. 72 THE rRoyuxciATiox OF e 77' ; the orthographic rules on ot, υ belong to the period of the Byzantine writers* ; this statement however according to what has been said before applies e(]ually to the case of at e. 01 has shared with υ the fate of becoming first ii and finally i. Section 19. Pronunciation of genuine OT. υ Of the three corresponding diphthongs Avith ^ AT ET OT, we have already had occasion to treat of the rarest and the first to disappear, namely ov. It is self-evident that its second element was u not ii, and that accordingly simplification took place by βττικράτβια as in the case of ei. An ou occiu's as is well known in old Latin (douco, ious), in old German, in English, in Portuguese and other Romance dialects ; it is nearly related to ail, Avhich arises from ^ it as in German, or forms its origin as in Portuguese, cousa ouvo*. This ou however is related rather with the Greek ων (ou) than with ov (gii); moreover in the case of the latter there is hardly any appearance of contact with av'. The genuine diphthong ov is found in ov, οντος τοντο etc., where it is formed by the addition to ο of the same v, which in αντη ταντα produces with a the diphthong av ; also in τοωντο<{ τοσοΟτο9 τηΧικοντος ; next in σττονδή (cp. σττβύδω), in άκύλον- θο<ί (cp. κύΧ^νθος), in βονς (βούτης) ΒοντάΒης, in 8ον\ος (written so in Bceotian too, not δώλο'. -1. i>. 1204 ANCIENT GREEK. 73 etc.'; all these instances rest on the testimony of ancient, especially ancient Attic, inscriptions, which continue to dis- tinguish ου and 0. It must be admitted, as we have said before, that the line of demarcation is not exceedingly sharp, and accordingly we find both TOTON τούτον and BON βουν'\ and in the case of ΦΡΟΤΡΟΣ ΦΡΟΡΟΧ' (from -npoVopaw) it is difficult to say which is correct. In άρουρα the genuine diphthong is shewn by the Cyprian writing a-ro-u-ra\ Section 20. Pronunciation of AT ET. There remain AT ET, diphthongs, whose fate was notably different from that of all the others, inasmuch as here there took place not a simplification, but a hardening of the second element into a consonant. The Greeks of the present day pronounce them as av ev before vowels and soft consonants {βη^, Χμ,νρ, ζ) that is according to their usual writing αβ ββ, but before hard consonants (τγκτ, φχθ, σ) ^ af ef, — αφ βφ, e.g. βββρ'^βσία, €β8ίΐ> (evSetp), άφτός, βφκοΧος, ίφκρατο^ etc. This sound-develop- ment forms a decisive proof, that in ancient Greek the ν in this diphthong, at least in general, had preserved its original i/-sound free from modification", and accordingly must be transliterated by au, eii and not by ail eii. For the development of ν from ύ would be as difficult as that from u is easy. At the same time in the case of ev traces are not entirely wanting of a modifica- tion of the second element : ev interchanges with et in 'iXet- θυια 'JLXevOvia Έλευ^ω' ; further we find on an inscription of '■ Λουθίαί ancient Doric (Sparta?) 60 b, 20. Rohl no. 68; Στρουθίη? also with OT ^ Before σ only in cultivated pro- Styra E5hl 372*''•'. nunciation ; the popular pronunciation - Inscrip. of Eleusis C. I. A. iv. is ^js (see Appendix). 27^, 40. "^ The same opinion is held by G. ' Dietrich in K. Zeitsclir. xiv. 56 ; Curtius, G. Meyer and others. Cauer L c. Also in the late inscrip. '' "EXevdvias Cret. (Le Bas v, 67, 74, C. I. A. IV. 22^ ΦΡΟΤΡΟΝ and ΦΡΟ- Bull, de con: hell. rii. 293, 1. 13), ΡΙΔΕ side by side. Cretic φρώριον, "ΕΚΐνθώ in the Anthology (A. P. 7. 604, BttU. de con: hell. ix. 8, 1. 8. 9. 268), 'Έ,λενθίαΈλΐυσία Sparta Mitth. ■• Inscrip. of Idalion. Dial. In.schr. arch. Instit. i. 162, Dittenb. Syll. 191. 74 THE PRONUNCIATION OF Mantiiiea belonging to the iir.st century B.C. αίτάν and έτησκβιάν, side by side with frequent instances of αυ and eu\ The lonians however were so far from tending to such a pronunciation, that in the fifth fourth and third centuries they wrote with more or less consistency AO, EO : raora, α6τ6<;, Καοκασίων, Εοελ^ώι^, λ€οκοΐ<;'\ This need imply no difference of pronunciation from the Attic, for au could be equally or more coiTectly represented by ao i.e. ao as by αυ i.e. aii, and this mode of writing was also made easy by the treatnu'iit in Ionic of original eo, which became in pronunciation and for the most part also in writing ev : KaXevvTe'i, eVot'eui/. Ίllere is an isolated instance of eov, Ένρυσθβνβους from Samos'. This very contraction into ev was in many places usual in Doric ^ and wherever it occurs furnishes a proof, that in the district in question ev was not eil. ao also in many places became av: Arcadian and Cypriot -av in the Gen. of the 1st Declension ; "EavKp/irei^; 'S.aύμeiXo<ί Ώρανχ^α^ in Boeo- tian, from Σαο-, Πραόχα. The lonians on the other hand made do first into 770 then into era : ττολιτεω, λεώ?, χpe(ϋμevo<{, certainly implying a sort of diphthong (eo), since this €ω decidedly resists separation into tAvo syllables*'. The process is this, the second vowel is lengthened and approximates to a, while the first loses some of its α -sound and is shortened. In the Doric Έρμοκρηνν Ύιμοκρηνν from Έρμηκρ4ων wo have the converse process^ But, to close this digression ; the close relationship of ei^ av to corre- sponding combinations of an 0-sound is sufficiently made cleai', and to return to the point from which we started, the value of this V has been thereby established as distinct from the ordinary 1 Lc 13as II. 352^ 35, 27. vowel are notably frequent, 1. 4 τντο- - Ernian in Curt. Stud. \. 294; σαυτο = ταύτοσαντου,6 ει>θντα = ίι>θαντα, C. Curtius Proijr, Wesel 1873; Hau- i) £Κ67•ΐΌΐ'σα, also 14 KeXcet. souUier in Bull, de corr. hell. iv. ")1 ; •' Bechtel Inschr. d. ion. Ditil. 217. G. Meyer- p. 135 f. The examples are li. would also explain thus (p. 58) from Chios, Samos, Erythra', Hali- ' .\piaroK\(ovs (Thasos) no. 72, cp. p. 35, carnassus and other towns of the n. 4 above. Asiatic mainland (also a coin of the •• Alireus D. D. 213 ff. Doric Cnidus has Εδβω\ο$, Hauss. ^ (I. Meyer- )). 13(); Ιϊραύχαί Kohl 1. c); from Phanagoria ('. /. Gr. 2121, no. 127, who explains rightly. Εόπάμονοί, Amphipolis (ib. 200Η). The « G. Meyer» p. 148 f. Ionic papyrus so often mentioned has ^ Cauer Del,- 16i). υ always, but omissions of the preceding ANCIENT GREEK. 75 value. Additional proof is turuished by spellings such as Άχίλλεοι/? ancient Corinthian ; aveo avev Attic, period of the Empire ; \αυΖίκβού<ί Olympia, period of the Empire, and other similar instances \ Moreover in this case alone Latin has not retained the Greek υ, but has represented it by its own ?i^ At the same time the other point too, namely that ancient Greek av ev were not av ev, has been pretty nearly proved already by Avhat has gone before. For how could καλβοντες yeveo'i have been contracted to kalevntes genevs {genefs) ? Or how could av af have come to be written with aol It is indeed just as hard to say, how if the pronunciation was av AT came to be Λvritten and not AF, especially as the digamma continued for a long time in use in so many dialects. Nevertheless, except in the case of Crete, as far as we know, it occurred to only one man among all the engravers and stone-cutters, to Avrite digamma here, namely the cutter of a Locrian inscription^, and even he did so only in one word ^α^ττακτίων, and that only once in twenty possible instances where he had to bring in Naupactus or its inhabitants. So fixed was the stupid " historic " or " traditional " orthography among the Locrians ! In like manner "EF^ero? on a Corinthian clay-tablet is isolated, Avhile on others Άχιλει;?, Ένρυμ,η^η<ί, Zeii?, αύτο- are so written*. In Crete on the other hand such a multitude of examples of aF eF (oF) have recently come to light ^ owing to the excavations of Halbherr, that the matter deserves serious consideration. In 1 Cp. p. 29 above; G. Meyer- p. 135 Beitmge vi. 78, Dial. Inschr. 68). (after Dittenberger Herm. vi. 306); A\/\TAI2I is found on 1. 7 of the great Arch. Ztg. 1877 part 2, no. 68. Ααυδι- Pamphylian inscrip. of Syllium (Eohl Kei's and many others [Empire], Meyer 505 Dial. Inschr. 1267) with a symbol 136. ToveoOaL Assos Archceolog. Inst. which on other Pamphylian inscrip- of America i.p. 33 '^ also ΕΘΤΤΤΧΟ[Σ tions stands for Digamma (Eohl p. 143 BuU. de c. h. vii. 52 (Thess.) must be Bezzenb. Dial. Inschr. 368); here too Εόύτ. we find VVOIKT oUof, Σ(Ε)ΑΤ\ΛΙΙΟΣ ■^ Cp. below p. 81, n. 4. Seluviyos. The digamma it is true ap- '^ Eohl no. 321 (Cauer- 229) B. 15. pears as well: ρίτ[ι]ία (vetiija, erea), •» E. no. 20, 101, 43, 48, 66, 68. τψάΡεσα. I willingly leave undecided the new ^ Comparetti Musco Ituliano ii. reading o-vo i.e. ov for ου on a Cyprian 131, 162 f., 194, 211, 215, 218, 222, 231, inscription (Deecke in Bezzcnherger's etc. 76 THE I'KONUKCIATWN OF the first place then on archaic Cretan inscriptions also we find as a rule αυ ey, and on the Gortynian law code without exception. In the next place examples are not wanting of a writing which was evidently in a state of fluctuation, corres])()nding to the instance cited Άχί,λλβού?, for example άμε^νσαθαι [ά]^υτάν^, just as an old Naxian inscription also shews AFVTO αυτού, an example of F in Ionic to which exception has long been taken though to no purposed Now this fluctuation points to the fact, that the sound au was adequately represented neither by αυ i.e. ail nor by a?. In the third place it may be erroneous to give to the F the value of the English and Romance ν and not rather that of the English w, Avhich as is well known belonged to the Latin v. For on a later Cretan inscription, dating from the time when the digamma was disap- pearing, υβρ'γων i.e. Ρβργωι/, βρ'γων is found repeatedly, and the name of the town Axus, properly Fa|o9, appears more than once as "Οαξος : while on the other hand it is true, that the sound might be thickened to a spirant, written β, instead of being resolved into a vocalic syllable : Βιαββιττάμβιως, ΒοΧοβΡτίοι = OXoi/Tioi". Moreover, the digamma, had it had the flxed sound of v, would hardly have disappeared so generally from the language, nor indeed would it have been likely to have existed in it before, as the only spirant of this sort, Avithout / etc.; but conversely, if it was a semivocalic u, and the language in general gave up the w-sound, it is easy to understand, that it did not follow suit in undergoiug the change to il and consequently had to disappear. Accordingly there will be to a certain extent a connection between this sound-change and the disappearance of F, and we also see dialects such as the Boeotian retaining ' ib. 20i, 221 (cp. tlic doubtful light : Hull, de con: hdl. 1888, IGl : TITOTfE^eO 157, Avhile in 215 we ^ΒΔΙΤ^ΑΗΙΟΙΊ Ρ^φ^καρτίδη^. have TITOrro:i, 208 TITTfOi:). •. ij -ui (no /n A- •. ^tpyuv toinp. ii. 678, col. ii. 1. a, - libhl no. 408 (the reading quite ^, . , ,..„ ,, , , . , „ .. . . , «; 3ta/ifi7r. (Jo'J, no. 21, 11; on Ηολο• nertain). Many attempts at explana- . ,^ ., .,.,., „ , ,, . • ' , , , % , , t, , ,, ffTioi Meyer- 2.^3. Cp. also Eretna tion have been hazarded (as by Kohl), ■,,, , ....^ ^,., „ , ,_, .^ , „ , , , , , • , . ^'^Φνμ- αρχ- 1^^'^H, 83 ff. 1. 174 c Οαλι- see however liechtel liixchr. d. ion. χ; r i. .. • .. χ ^, ,•»,» , ν . . Οίο[ I'Jnom. i)iOpnuin = fa\i5/oi;( HXeioii) />!«/. p. Hy. A new instaiiuu ol f in a „.i,;i„ ;κ iqo •* • •»χ -.χ ,., V • , , , while lb. 182 \ it is written λΧώίον. Naxian inscription has lately come to ANCIENT GREEK. 77 the digaiiiina with the true «,, whihi those like the Attic and Asiatic-Ionic gave botli of thein up at an early period. If now the digamma was a semivowel, no inference can be made from the writing aF eF for a modern Greek pronunciation, any more than in the case of the Oscan, which writes the corresponding diphthongs regularly av ov, that is to say with the semivower. The interpolation of a digamma or of a β representing a digamma after ev when followed by a vowel which occurs regularly in Cyprus and occasionally in various localities: — ΕυΤάγορα?, Ba/ceuFa? Bo3ot., Κνβανδρος Dodona, Έ^ύβάλκης Lakon., e-u-ve-r-{e)-ke-si-a evepyeaia, e-u-ve-le-to-to-se Έιν€λθοντο<ζ Cypriote admits of easy explanation. For in this case a semi- vowel V was developed out of a -a just as easily as a semivowel y ■ from an i, which likewise appears in Cyprus : a-no-si^a ανοσία, and in neighbouring Pamphylia, where two i's are written : ΔΙΙ A, ESTFEAIITS 'A^7reV3to9. The same holds good naturally not only of ev but also of υ = u; hence we have in Cyj^rus tu- va-no-i 8vfavoi i.e. Βώοίη (formed from ΔΤ instead of Δ0), and on a Chalcidian vase Ταμν^όνης Τηρνόνη'^^. In case however any should be inclined to infer from what has been cited, that the V of these diphthongs tended from an early period in these dialects to harden into a consonant, it must at least not be forgotten, that it was precisely in the Cyprian dialect that the customary pro- nunciation was really diphthongic ; for the manner of writing- is pa-si-le-n-s{e) βασιΧεύς o-na-sa-ko-ra-u Ovaaayopav. The Cyprian dialect also shews by the coexistence of forms such as e-v{e)-re-ta-sa-t'U and e-u-v{e)-re-ta-sa-tu {βϊρητάσατν, €ν?ρητά- σατν i.e. ώμόΚό^ησβν, from ^ρητα — ώμόΧο^ία, ρήτρα)*, how Lesbian ανρηκτος for άρρηκτος άρρηκτος, ΕνρυσίΧαος for ΈF/^Lισ.^, and similar instances are to be explained. In these the ' B. Krnczkiewicz, d. altlnt. u. άριστεύτοντα, Corcyra B. no. 343, may oskiscJie Diphth. ou, Ztschr. f. ost. just as well be an error for άριστΐύοντα Gymn. 1879, 1 £f. as for -evfovra. 2 Dial. Imchr. 648, 458 (cp. 1040, ^ Inscr. of Idalium, Dial. Inschr. 60 1146) ; Karapanos Dodone Tab. 34, 3; (Cauer^ no. 472) 1. 6; C. I. Gr. 7582. Mitth. d. archaeolog. Imt. i. 231; Dial. * Idalium 1. c. 4, 14. Inschr. 71, 165 ff. (cp. e-u-va-ko-ro ^ Ahrens D. A. p. 37; Inscr. of Ei)f αγόρω 153 ff., e-u-va-te-vo-se Έΐύάν- Eresus, Dial. Dischr. 281 c. deos 161 ff.). On the other hand 78 THE PRONUNCIATION OF F was changed into a vowel before the r, for which process the Cyprian writing contains the middle step ; the υ however must by no means be considered as the representation in writing of a digamma still heard in pronunciation. In many cases a digamma in the middle of a word also has in the dialects become combined with the ]jreceding \'()wel into a diphthong : e.g. C}^rian ke-ne-u-vo-n{e) KevevVov Kevevov (κενβόν), Lesbian νάνος (ναός), αϋως (i}u> (Rome). ANCIENT GREEK. 85 not reckoned in the alphabet, but belonged to the fricatives, the new formation of the fricative y not only from γ but also from vowel i, in many cases diminishing the number of syllables (Ιατρός yatras, ττοΐος pyos); lastly the universal abandonment of the lengthening of the consonants represented in writing by their doubling : αλλά pronounced aid, μέλλω meW^. I think therefore, that the transformation of the sound- conditions could hardly have been greater, especially as even the explosive sounds which have remained have in certain cases a special pronunciation, conflicting with the writing. — We Λνϋΐ begin our more detailed examination with the ημίφωνα, under which head we shall reckon the spiritus asper. Section 22. Pronunciation of the nasals λΙΝΓ. The Greeks have and had three nasal sounds, corresponding to the three classes of mutes : the labial nasal μ, the dental v, and the guttural, Avhich has no especial symbol in the alphabet and is represented by γ (?Ί in Lepsius), called by certain grammarians ά'γ/χα or αγγμα. Only ν can be used as a final, but final ν was assimilated in the context to following consonants, i.e. it became yu. or γ respectively, and more rarely λ ρ σ : το ρ 'PoStov, ώλ λέ'γονσι,, e? ΧιΒώνι, έσστήλβι or βστηληι^. Inscriptions preserve abundant testimony to this, and in many, at least before mutes and μ, assimilation is consistently carried out^ ; even manuscript authority is not 1 In the modern dialects according 1885. The Ionic Inscript. of Halicar- to Psichari (cp. Rev. crit. 1887, 26J: n. uassus Bull, de corr. hell. iv. 303 has 4) the vanished nasal has developed a sometimes e\ Ανρισσωι sometimes iv doublingof the consonant: aedosavdos, Χνρισσωί; the older one K. 500 1. 41 niffi νύμφη, Ιϋχχιτο τόν χοΐρον, toyyero tws συμπάντων, τον yepov{Ta), tonafti τον ράφτην. "^ Consistently carried out e.g. on - C. I. Λ. II. 9. 14''. 86, 14, 31. 369 the Megarian inscrip. C. I. Gr. 1052 etc. So also έστήσαντι i.e. ένστ. (Dial. InscJir. 3003) : in the rescript 834''", 28. Cp. Giese Jco/. Difl/. s3ff.; of Cu. Manlius to the inhabitants of Cauer in Ciiri. Sii/rf. VIII. 295 ff. ; Meis- Heraclea Latmi, C. I. Gr. 3800, Le terhans ed. 2, p. 86 ; Hecht Ortliogr. Bas v. no. 588, Dittenb. 209 (only 1. 9 dial. Forsdunigen 1, Progr. Konigsb. ττρόνοιαν ττο^ΐσθαι). 86 THE PRONUNCIATION OF Λvanting on some papyri*, and doubtless in the Attic and Macedonian periods this mode of writing was largely made use of in the texts of authors. But it appears, that in time the general tendency Avas, in the cultivated speech, to isolate words more and pronounce each distinctly by itself^ as is shewn in an especial degree by the dropping of elision and crasis. In any case very few traces of assimilation have remained in our best manuscripts, and in our present manner of writing none ; the modern Greek popular pronunciation on the other hand retains certain traces of it, although in general it rejects hnal ν alto- gether^. Conversely with us assimilation in the interior of words is regular, with the ancients this is not so much the case : it is not only that συνΧαμβάνω, evKoXetp and in general eV- συν- τταν- before all sounds is on papyri the more common writing*, but also on inscriptions 'OXwiria, Χανβανέτω, ανκυρα, €7Γ€νψ€ν and such like appear at all periods Avith greater or less frequency^. To infer from this, as some have done, that the Greeks pronounced the nasal before consonants in the French way, is an extraordinary piece of perversity^; however 1 Pap. L. 2 (Dialectics) col. 2 τωμ ποιητών, 3 ούθίμ πημα, 5 ey yvvai^i, 7 Toy ye, 8 προσίδοΐσαμ φάο$, 9 ay ylvoLTO, 11 τωμ ποί-ητων. However it is not frequent on this careful and very old manuscript. Ι^ψ. 1 has only μey yap col. ϋ, and ΐμ βραχεΊ in the acrostic v. 2. On the other hand a Herculanean ms. (Gompertz Wiener Akad. Bd. 83, 87 ff.) which also shews et for Tjt : οταμ ττ6ρρωθέμ ποθΐν. €πιτήδιομ πραξιν. τώμ προλήψεων y^yvόμevov καΐ τωμ φαινομένων, το \€yύμev6μ ποτέ. KaTayi- λαστογ yap etc. - Hecht 1. c. p. 32 cites (after G. Hermann de emend, rat. Crete Bull, dc con: hell. in. 290. Cp. G. Meycr^ § 289. ^ Cp. ταρριφεντα on a papyrus of the Ptolemaic era, Wessely Wicn. Stud. 1880 p. 205 ; in Homer forms like άνάρρώ-γαί, καταρρόον, τόρρα, La Eochc Horn. Textkr. 389, though Aristarchus certainly wrote not only τό pa, but διαραΐσει (- ), πολύρην^!. The same fluctuation however appears in Homer in the case of the other liquids. '' /J. A. II. 693: oi αρχαίοι -γραμμα- τικοί τό μίν μ(τα ψιΧον εϊψισκόμ(νον ρ ΙψίΧοχψ, το δέ μίτά δασέοί (δάσΐΊΌν' οΙοι> τό Άτ/)€ΐ'ϊ καΐ κάπροί έψίΧονν, τό δί Xpovos άφρόί ΰρόνοί ίδάσννον. Λ NCI EST GREEK. 91 out by its treatiiiont in prosody, since χρ no more than κρ makes length by position. Among the dialects, aspiration of the ρ as well as the vowels was unknown in ^olic*; in other places it was omitted in the few Avords, where the second syllable began with ρ (aspirated ?) : 'P«/3o Aristot. El. Soph. p. 177 b 3 on o/)os and Spot : ίι> μέν rotj '/('^ραμμ^νοΐί TaUTuf όΐΌμα, όταν (κ των αιτών στοι• χΰων -^(.-^ραμμένον τι καΐ ώσαντωί, κάκ(ΐ δ' 7}δη παράσημα ττοιοΟνται' τα δί φΟί^• -^όμΐνα οΰ ταντά. ANCIENT GREEK 93 the same line with the other letters, but written above as a . . . I- diacritic mark A. At a subsequent period the corresponding^ symbol -\ was invented for the spiritus lenis, i.e. the absence of the breathing^ and the rounding of these symbols gave our present mode of representing the spiritus. Its representation in Latin shews that the h was still heard in the Hellenistic dialect ; moreover the aspiration of the tenuis in elision was consistently observed, although not always in a way identical Avith our own ; for we find for example, μβθοττωριρός, KaO'eroq and δωδβχεττ;?, βφ'Γστ;, άφβσταλκα^. Similar fluctuations are Avell known in Latin from the first century B.C. and onwards both in the case of consonants and vowels^ ; Catullus' poem on Arrius and his cJiommoda, hinsidiae illustrates this best^. In the case of consonants aspiration came in about this time from the Greek, in the case of vowels it must conversely from this time onwards have lost ground in the popular language, so that it Avas in the cultivated language that uncertainty prevailed, where to pronounce and write h and where not. That educated people continued to pronounce the Ji even during the Empire is shewn, to take an example, by a passage of Quintilian, where he laughs at those people as affected, who greet one another with ave instead of have on account of the derivation from avere^. ^ The definitions προσφδία φι\ή or πνεύμα ψϊΚόν (the latter properly speak- ing an unsuitable expression) can mean absolutely nothing else: ipCKo^ is devoid of breath, and Seelmann p. 262 is mis- taken, when he takes the expressions δασεΓα and ψιΚη to mean not something absolutely opposite, but only different degrees of aspiration. Latin writers have been (as so often) awkward in their translation of the terms, and the passages spoken of by S. from their grammarians, Λvhich would not allow to h the value of a letter, have no value for phonetics whatsoever, but only shew like countless others the dependence of Latin gi'ammaron Greek. For my part I see no reason for tlie as- sumption, that Greeks or Komans pronounced the unaspirated vowels differently from the Germanic and Romance peoples of to-day. 2 G. Meyer2 p. 244. Dittenberger Si///. Ind. p. 781 f. 784. ΜεόΌττ. is the regular spelling PajD. L. 1 ; δωδεχέτ-η^ Kaibel EpUjr. 112, cp. 190, 205, 222 ; Rich. Wagner de epigr. gr. (Lpz. 1883) p. 90 ; on άφέσταΚκα etc. cp. Keil Sclicdae epigr. p. 7 ff. ^ Corssen Ausspr. τ." 104. ■* Catullus carm. 84. 5 Quintil. i. 6, 21 : multum enim litteratus, qui sine aspiratione et pro- ducta secunda syllaba salutarit (avei-e est enim). In the whole section he is speaking only of correctness of pronun- 94 THE PRONUNCTATIOX OF But after the second half of the second century a.d. h in inscriptions is more and more frequently wrongly put in and wrongly omitted* ; the letter was therefore evidently dis- appearing, and the same development took place in Greek. The Copts, it is true, continue to represent the spiritus in Greek loan-words almost without exception with• their Ζ (Λ): hoste, hina, hole etc.^; it cannot therefore have disappeared in the second century. The cultivated pronunciation certainly retained it much longer, just as in Latin, where we find Augustine testif}ing to the offence taken in his time at pro- nunciations such as ominem^. Modern Greek however knows the aspiration no more than the Romance languages ; ior the French owe their h aspire to the Germans. If however we infer from the growing uncertainty in the use of the symbol in Latin that the sound Avas beginning to disaj^pear, are we not bound to make the same inference Avith regard to the Attic of the fifth century B.C.? For here too the cases are very numerous, Avhere Η ought to stand and does not*. The converse of this is of less frequent occurrence, except on one inscription Avhich was evidently cut by a foreigner, where ev, οΙκών etc. are written in the most surprising manner^ It has indeed actually been maintained, that the breathing was no longer heard among the Athenians of the 4th century*, and this view receives support from passages of Aristotle, Avhere the ciation ; he comes to orthography in c. 7. The question is also settled by c. 5, 17 ff. ; Vel. Long. K. vii. G8 f., etc. ' Coisseu 1. c. p. 110 ; Seelmaun p. 265 f. (the wall inscrip. of Pompeii shew the same uncertainty as early as the 1st cent., cp. on the confusion of (le and e in the same, p. 69, n. 1). - Stern Kopt. Gr. p. 19. 3 August. Confess, i. c. 18 § 29 (Seelmann p. 265). — Among Greeks compare (Oros) Prolegomen. Hcphaest. p. 93 W.: yiverai βραδντηί rts τοΰ χρόνου, ώί καΐ tv rrj δασύς. Xiyerai, δια τψ όξύαί (λοι in καλοί longer than in 4 ff.; G. Moyer Gr.- p. 242. ANCIENT GREEK. 95 distinction between ου where and ου is designated as one of pitch without the least mention of breathing\ But on the opposite side we have another passage of the same author, according to which opo'i and opo (ΐττή^ησι^^ Moreover Plato's Kratylus contains two important passages, of Avhich the reading is.it is true corrupt, but the sense of which cannot be mistaken. Socrates derives the word εττιστήμη on one occasion, starting from the philosophical standpoint of Heraclitus, from βττομαι, according to which it must be έτηστ., on another occasion how- ever, considering it from the Eleatic standpoint, from ΐστημι, that is €7Γ-Ιστημη. The former is expressed according to the recorded text thus : Bio Βή βμβάΧλοντας Set το ei (e) έττιστήμην αύτην ονομάζειν; the latter: όρθότβρόν earw wairep νυν αυτοΰ την αρχήν Xejeiv μίιΧλον ή βμβάΧλοντας (βκβ.) το el (e) βττιστη- μην, άλλα την εμβολην ττοιήσασθαι άντΙ τη<; iv τω ei (e) iv τω ίωτα*. Since βμβάλλειν often occurs in the Kratylus of the interpolation of a letter, and that which is here interpolated is the breathing, the object to εμβάλλοντας in both places must have been the name of the breathing. I suggest therefore that the symbol Η was already known to Plato, as a τταράσημον written over letters, and that the name answering to its form was the first half of ?)τα, accordingly η. If then we substitute TO η (or TO h) in both places for το ei, I think we shall have restored these much abused passages. To return, the chief point is, that such an inference proves too much. For there is scarcely a dialect, where there is not fluctuation*; even on the tables 1 Aristot. El. Soph. p. IGG b 1. 178 this word in tlie same more general a 2 {to μέν ο^ύτερον το U βαρύτΐρον sense as later writers (ep. Schmidt 1. c. ρηθερ). Κ. Ε. Α. Schmidt Beitr. ζ. 187 f.). Gesch. d. Gravimatik p. 155 f. wishes ^ Ruhl no. 500, 1. 19 (Halic), 384 to explain this on the assumption, that (Samos), Bull, de corr. h. iv. 115 = in the combinations μ^ν ου (ine-nu) and Dittenb. 349 (Teos). TO ού mentioned by Ar. the spiritus ^ Plat. Kratijl. 412 a, 437 a. was not perceptible; but in that case ^ Hiero's helmet, B. 510, Ίάρων Avhere was it? Only at the beginning and ο — ; Locris E. 821, Όποντίων of a sentence ? and 'Οποντίων. Thespias E. 140, μ' 6 - Here also Aristotle is speaking of and 6s (o[?] E.). προσφδία, and must therefore have used 96 THE rRONUXCTATIOX OF of Heraclea M-e find ίσος and "σος side by side. But if the breathing began to disappear at an early period in all the dialects, it could not very well have continued to exist in the Alexandrine and Roman periods in the common Hellenistic language. We must therefore seek for some other explana- tion ; such an explanation is furnished by the weakness of the breathing, Avhich also serves to make the great inconsistency and capriciousness in the aspiration of isolated words more intelligible. We say ίττττος but Γλαυ/ίίττττος AevKnnro^, and as the cognate languages shew, the spiritus has no etymological warrant whatever. We find too side by side αγω (in Locrian it is true αγω) and ηΎέομαί, ημαρ and ήμερα, ήως and εως ; there is no etj^mological reason for the fact, that initial υ is always aspirated*. This \veakness of pronunciation also made it natural, that the Athenians and most of the other stems on adopting the Ionic alphabet should not trouble them- selves about any new symbol for the sound of the breathing. In the interior of words in Laconian and other dialects the breathing was a late development from σ : ' Κ'^ηίστρατο'^, iiroi- Fr;e = ζττοίησβ^ ; according to the Grammarians the Attic dialect knew this internal spiritus only in the foreign \vord ταώ<{^. In composition it was not generally Λvritten in Attica*, on the Heraclean tables not ahvays'; Latin as a rule represents it even here : exhedra {exedra), parhippus, Panhormus, Euhemerus^. It had undoubtedly in this position a still slighter sound than at the beginning of words ; the Alexandrine Grammarians themselves, Λvho wrote the ' interaspiration ' in the texts of the poets for the sake of clearness, renounced the rough breathing, if the real significance of the word lying hidden in the compound appeared to be no longer felt: ωκναΧο<ς νηΟ'ί from αλ?, Έ,ναϊμωι> from αΐμων. 1 G. Meyer 2 p. 243. ΚΑΘΗΑΠΕΡ iv. 51% 43), although - The latter is Argive, E. no. 42, Giese Aeol. Dial. p. 333 maintains, 44 a. that the aspirate in this case was quite •' Athen. ix. 397 ef. (Attic vase inaudible, inscription vlvs, C. I. Gr. 8202, cp. '' wapf^ofn once by παρΐξόντι (the H203.) preposition in this dialect took the * Cauer Fitud. viii. 240 f., Meistor- form πάρ). bans ed. 2, p. f)7. In Elision IIAPH- '• K. L. Schneider p. i;t2 f. Also ΕΔ1ΌΙ C. I. A. I. 34 and iv. IIU•, 10, Coptic <ι/ί(.;•.//(»>•. Stern Kopt. Gr. p. H). 31ΕΔΙΙΕΝΙ i. 77, C (alao with pleonasm ANCIENT GREEK. 97 Section 26. Pronunciation of the Teniies. Among the nine mutes the Tenues (i.e. ψιΧά, the surd letters) have on the whole retained their pronunciation. At the present day however the media appears in pronunciation after a nasal : Χαμττρός pr. lainhros, αηρέττομαι endrepome ; άναΎκάζω ανά^γκη anangcizo anangi^. The same thing takes place in close combination of words : τον πόΧβμον torn halemo, τον -τόττον ton dopo, TO/' κόσμον toil goznio^. The assumption of a similar pronunciation in ancient Greek leads at once to pure impossibilities : hoAv could the ancients have kept eVro? and evhov, αναφανδόν and -φαντο- so strictly distinct, as they certainly did i For we are not entitled to appeal to the Aristotelian ivreXexeta by the side of €ν8ε\€χης : the word must have been eVSeXe^eta, but being of infrequent usage it was remodelled on the analogy of τέλο^. Next we are con- fronted with αμττΧακβϊν and αμβΧακβΐν, Άμττρακία and Ά/χ- ^pa/cta, finally the Aristophanic pun βΧέττβιν BaXXrjvaSe (UaX- Χηνάδβ, ΐΙαΧΧηνΐ] and βάΧΧβινΥ, those who cite these instances not jjerceiving that the very infrequency with which they occur contains a full refutation of the inference they draw. For tenuis and media or as we now say surd and sonant explosives approxi- mate so closely in sound, that to say nothing of the license of word-plays, actual instances of interchange are not wanting in Greek any more than in other languages, for instance on Attic inscriptions τότω for δότω, aypoTroXec, Me/cαΛ:λ^s■^ Above all in Egypt τ and δ could not be kept distinct OAving to the peculiarity of the national language, which did not possess a d, although it had h ; accordingly mistakes such as τίΒυμοι, ToBe for τότε, Έντοξος are among the commonest on papyri^ 1 The pronunciation of κ\ as 7>v Cuitius Gott. Nachr. 1857 p. 303. which has often been maintained is ■* C. I. ^. 11. 603, 272. Bull, dc con: denied by Psichari for the peneral h. 11. 5.52 (ib. iii. G4 Scyros κυνή for language. ywrj). - Foy p. 47. " Praefat. Hyperid. p. xvii. 3 Aristophanes Achnni. 2.33; E. 98 THE PRONUNCIATION OF But the position in Avhich the sound occurs, makes in these cases no difference whatever. Apart from this in the case of h a twofold pronunciation is current in modt'rn Greek' : guttural before consonants and before α ο ίΐ, and inclining to palatal before e i (i.e. k' according to Lepsius' alphabet, being to k as ch in ich is to ch in ach). Consequently iu the καί of the present day a sound is heard somewhat like ki/e, in which the k is produced so far forward on the palate, that it approximates to t. In many cases this palatal k like the c in Romance Avas and is further developed to cJi tfi, so that Psichari gives four further pronunciations for /ce και: — chye die tsye tse', and this pronunciation as Italian ce, although at the present day it is not considered worthy of imitation*, nevertheless made itself distinctly felt side by side Avith the other at the period of the revival of letters*. Something analogous to k k' might be found in ancient Greek in the contiguous use of 9 (koppa) and Κ ; this however seems in point of fact to have been more a matter of orthography than prommciation. The syllables κο κρο κτο Avere Avritten Avith 9, because the letter Avas called koppa, κα κρα etc. Avere on the other hand Avritten with kappa for the same reason'; the rest of the Avork fell to the share of the latter, as standing before the other in the alphabet, except Avhere a u still retaining its proper w-sound appeared to demand similar treatment to o". Subsequently 9 was given up as superfluous, just as A* in Latin gave place to c. Section 27. Aspirates and mediae; contrast between ancient and nwdei^n Greek. The pronunciation of the aspirates Θ Φ Χ is one of the most dilHcult points. The name aspirata Uttcra, Saav γράμμα points 1 Foy p. 5. p. 530. 2 Psichari Jiei•. C;/i. 1887 p. 2GO. « ΒΟ9ΑΣ (?) Ba^ot. Ruhl 183 =* Foy p. 5G. stands alone ; Meister Dial. -Iiischr. SSI ■» Cp. the edict of Chancellor Gardi- /3w[\]as ? ner (p. 3 above), which on this point '' Cp. p. 35 above, allows a certain licence; Smith Si/lliiije ANCIEXT GREEK. 99 to tlio addition of a breathing, i.e. an h ; accordingly they are Avritten in Latin th x>h ch. In pronouncing ju as f, and ch in the German lashion, we make out of the aspirate a spirant, and such also is the English th. Modern Greek also makes them spirants, θ being pronounced as English th in think, φ as / χ before consonants and before a u with a guttural sound like the ch in German ach (χ in Lepsius' alphabet), before ei on the other hand with a palatal sound like ch in German ich (χ). It has hoAvever also made the mediae into spirants in the same way. Media, μέσον, denotes the intermediate pronunciation betAveen ψίΧόν and 8ασν, that is to say neither quite Avithout breathino• nor yet with a particularly strong breathing \ But in modern Greek δ is the soft English th as in this ; β is v, that is to say the soft sound corresponding to the hard /; γ either a soft guttural ch or a soft palatal, being wholly analogous to the χ ; Lepsius Avrites these sounds too Avith Greek letters : y y\ The Germans give the g this pronunciation in many cases, especially in the interior of words, and make γ γ' Avhen medial correspond to χ χ' at the end, just as in German other consonants which are soft when medial are pronounced hard when final: Tage, Tag {Taye -Ταχ), Berge Berg (Bery'e-Berx) corresponding to Leid pronounced Lett while leiden has the proper fZ-sound. Palatal 7' is identical with English i/ German j, and accordingly the Greeks now pronounce yhotro yenito, yfj yi. The explosive pronunciation, as a media in the Latin sense, only remains to the modern Greek mediae where a nasal precedes, consequently at the present day ντ v8, μττ μβ, and partly also jk 77 are identical in sound ^. Such a sound- 1 So Diou. Thr. B. A. 631; Dionj-s. komvos (κ6μβos), andros is simply arti- de compos, γι. 83: ψιΧα μΐν τό re κ και τό ficial and owes its existence to the 7Γ και TO T, δασέα δέ τό re χ καΐ το φ καΐ written form. Another instance of TO Θ, KOLva be αμφοίν (repeated subse- artificial pronunciation according to quently as μίσον άμφοΐν, τοΰ μίν yap Ps. is the sounding of the ν of the ψίλότερον τοΰ 5e δασύτΐρον) τό re 7 καΐ article in τον -γάμον, την -γυναίκα : dia- τ6 β καΐ το δ. Cp. also Aristid. Quintil. lectally this is assimilated {toyyamo, p. 89 f. Meib. (54 Jahn), 44 (29 J.), thjyineka), in the ordinary language p. 101, n. 1 below. it disappears without any compensa- 2 Psichari Ber. Crit. 1887, 2GG, tion (to yamo, ti yineka). according to whom the pronunciation 7—2 lOO THE PRONUNCIATION OF system cas this transferred to the ancient language must of necessity alter its character most violently. It is however perfectly impossible to transfer it. For all spirants are frica- tives, that is according to the ancient nomenclature ημίφωνα, having even without the addition of a νοΛΥβΙ a certain percep- tible sound ; but in ancient Greek /3 7 δ are always and φ χθ generally reckoned among the άφωνα. That the latter were by some, apparently by the Stoics, considered as ■ημίφωνα\ is fully explained by the fact, that the added breathing is of itself a ήμίφωνον ; in like manner ξψ ζ are reckoned as ημίφωνα owing to the σ which forms one of their component parts. In the modern Greek pronunciation on the contrary no one could ever maintain these letters to be mutes. Moreover Dionysius of Halicarnassus gives a closer description of the pronunciation'' ; he says that in the case of ττ β φ the mouth is shut and then suddenly opened, in like manner in τ ^ δ the tongue is pressed against the teeth ; in /c χ 7 it is raised to the palate, and there is no further distinction between these letters according to him beyond that of the breathing^ Aristides Qiiintilianus also writing in the third century expresses himself to the same effect : — in the case of the media β and the related sounds 7Γ and φ the stream of air, he says,• breaks through the closure of the lips in the centre, and so on, he too making the only ditference betΛveen the related sounds to consist in the fact, that the tenues were articulated in the front part of the ^ Sext. Enipir. p. 021 f. represents β are prononncetl, οται> τον στόματοί the aspirates as ημίψωνα, adding, that πκσθ^ντοί τό προβα\\όμ(νον ίκ τηί 'some' reckon them as άφωνα; Pris- αρτηρίας πΐ'€νμα λύστ] τόν δ(σμ6ν αντοΰ. cian Ι. 14 says conversely : hie quoqiie — ΤΘΑ : ttjs -γΧώττηί άκρψ τφ στόματι error a quibusdam antiquis Graecorum ττροσΐρΐώομένη^ κατά. toi>j /uerew^oi't grammaticis invasit Latinos, qui φ et όδόνταί, ίπειθ' ύττό τον πνΐύματοί άπορ- θ et χ semivocales putabant, nulla alia ριτηζομένψ καΐ την δι^ξοδον αύτφ κάτω causa, nisi quod spiritus in els abnu- nepl τοι /s όδόνταί άποδιδούσηί. — ΚΧΓ: det, inducti. The Stoics according to ttjs -γλώττη^ άνισταμ{νη% πρόί τόν οϊφα- Dion. L. VII. 57 reckoned only six v6v iyyvs τηί φάρΐ'-γ-γοί και Trjs άρτηρίαί άφωνα, β y δ κ π τ. Dionys. Halic, Οπηχονσηί τφ ιτνΐνματι, ονδίν ονδί ταντα Dion. Thrax etc. reckon the aspirates διαφέροντα τφ σχήματι άΧΧήλων, irXijv as mute without expressing any doubt δτι τό μίν κ ψιλώί \^-γεται, τδ δι χ on the subject. δασίωί, τό δί y μ(τρΙωί καΊ μ(ταξιΊ - Dion. Hal. Comp. γ. 7Η 11. άμφοΐν. ■' Dion. Ihil. ('. p. Η'Λ {.: w φ ANCIENT GREEK. ΙΟΙ inuuth and softly, the aspirates energetically from the larynx, the mediae with moderate force in the central part*. Acc(ji-(1- ingly all these sounds were instantaneous and explosive ; f c/i etc. on the other hand are fricatives, being produced by a contraction not amounting to complete closure of the vocal [)assage ; for neither are the lips closed in producing f, nor in making the ο$, Ύρόνιμος, 6k\os\ - lluschur I.e. p. 98. So tvOavUoi Sterret Arcli. Innt. of Ainen'cu in. no. and άν(βέθΊ) on the Eleusiuiau inscrip- 3u(i 1. iio, 100, 31) Ιωπάνη^ Ίίΐμότΐο^, 1. tion C. /. A. IV. 27(). Cp. also Meister- lO'.t KTt^€»'7;fos, for which subsequently hans p. 78- f. ΧΛμ. ^ lloscher p. 711 if. ; Schmitz p. 114 ■* There are naturally here and if. — The βάρβαροι in Aristophanes, the there violations of this rule, lloscher Scythian in the Theaiitdpli. and the p. 89. ANCIENT GREEK. 105 to proiiuuucc ekiitlios (βχθο•^), phthctro with doubled breath- ing \ % also has a similar aspirating power, at all events at an early period ; hence arises the writing Φ%, XS for ψ and ξ ; Plato says, that φ ψ σ ξ are letters with a strong breathing'•'. Is it possible then to pronounce j) h s in succession ? We must however be on our guard against speaking too readily of impossibility ; for to others, as for example to Lepsius, khth, jus appears perfectly possible, and only khkh impossible, since here the organ is the same : where the organ is different on the other hand, the breath, according to them, comes out simul- taneously behind the first letter, before the mouth assumes the new position. Accordingly we have no need of the way out of the difficulty, which was adopted by G. Curtius' following the lead of others. This was that the breathing heard after the t or with the s in combinations such as j^th jjs was liable to be transformed in the sensorium of the hearer and consequently also in script to the ρ Avhich was equally susceptible of aspira- tion, and these combinations being of frequent occurrence habit did the rest to establish an orthography ΦΘ etc.^ This form of writing is as a matter of fact much too well established for such an explanation to hold Avater ; the four or five exceptions on archaic and later monuments: ΑΓΊΘΙΤΟΝ, ΚΑΤΑΠΘΙλΙΕΝΗΣ etc. can hardly count^ On the other hand the entirely different treatment of such combinations in modern Greek must be made prominent. The modern Greek spirants shewing an exactly opposite tendency combine with the teuues : φτάνω φθάνω, κ\εφτης κΧ€7Γτη<ς, οχτώ οκτώ : neither a combination of hard (surd) spirant with spirant nor of tenuis with tenuis is in accordance with the genius of the language. In like manner σ admits of a surd spirant neither ^ Ebel in Kuhn's Zeitxchr. xiii. who uses it (for want of better proof) 266 ff. to establis]i a spirantic element in φ 2 Plato Kratyl. •127 a: δια τοΰ φΐΐ in Plato's time, και τοΰ ipei καΐ τοΰ σΐ-γμα καΐ τοΰ ξητα, '^ Curtius Grdz.^ p. 41-i ff., after OTL πνευματώδη τα. -γράμματα, πάντα τα W. von der Mtihl A.^iration der Teuues τοιαύτα μεμίμηταί avTots ονομάξων (the (Lpz. 1875) p. 21 f. See on the other giver of the names), οίον το ψυχρον side J. Schmidt A'. Z. xxviii. 179 ff. Kai TO ieov καΐ το ceieadai κτλ. The ■* Kohl no. 314 (Phokis), 382 (Chios), passage is (jaotetl by v. Raumer p. 101, I06 THE PRONUNCIATION OF iiiiiiR(liatcl\ incccdiug nur ioUuwing it: -ei /σα i.e. -€φσα becomes eyfra — ejisa, σχ^ίζω becomes σκίζω, μισθό^; μιστός, αισθάνομαι στάνομαι\ In the same way a surd spirant does not allow a preceding na.sal : either this is assimilated and in some cases expelled as (Ίνθος αθθο>< aOos, νύμφη ηίβί nifi'\ or a tenuis took the place of the aspirate and then a media the place of the tennis, as in the word Κόρινθος which I have myself heard pronounced Kurindos (written Κόριντος). Finally we must remark the etieet jjroduced in many cases by a preceding ρ : ηρτα for ήρθα ήλθα, Κόρτο popular name for Corinth, €ρ•χ^ομαι pronounced erkonie or er-^ome^. The same applies to the voiced spirant in combination with a nasal, neither is this spirant allowed without exception to stand combined with p. Where the phonetic laws are so different the sounds themselves of ancient and modern Greek must be fundamentally distinct. In the next place there remains to be produced in support of the long continuance of the real aspirates not only Quintilian's testimony, who regarded φ as a dulcissime spirans littera, Roman f and also the ν in servus on the contrary as odious and offensive sounds \ but also that of the Coptic mode of writing which arose at the end of the second or the beginning of the third century. The Egyptian Christians, when they devised a new alphabet, mainly borrowed from the Greek, for their national language, employed the symbols Θ Φ X for the real aspirates which are found in Egyptian ; on the other hand for the sounds / and cJi, which they likewise possessed, they adopted peculiar symbols Avhich were annexed to the Greek alphabet. In the numerous words borrowed from the ' Foy p. 131. Σφ holds its ground and adds : (june si nostris litteris scri- according to I'sicbari {Mem. de la Soc. bantur, suiduni quiildam et barbanun liiKjiiint. VI. 305) in tlie ordinarj' Ian- eflicient, et veliit in locum eaiuni siic- guage, but in Trapezus has become cedent tristes et honidao, quibus (iiae- (77Γ. cia caret (/ and ii). Nam et ilia, ({uae '■' See p. 8ϋ, u. 1. est sextanostrarum.paeuenonhumana * Psichari liev. crit. 1887, 265. voce vel omnino non voce potius inter •» Quint. XII. 10, 27: jucundissimas discrimiua dentium etUauda est. — ex Graecis litteris non habemus (v and Aeolicae quoque litterae, qua gervum φ) — , quibus nuUac apud eos dulcius cfrntiinine dicimus, etsi forma (/") a spirant. He goes on to speak of the nobis repudiata est, vis tamen uos ipsa grecizing spellings Zepliynis, Kplii/itt, persequitur. ANCIENT ΙΙΙΙΚΙ^κ. loy Gi'i'ck Φ Χ uccur, but instances also are found υί" the resolved spelling, EllSOSON as well as ΕΦΟΣΟΝ, just as in the case of native words the combination of tenviis Avith h alternates with the aspirates \ At A\'hat time then did the spirants appear in the ordinary speech ? Priscian about 500 A.D. evidently found it difficult to adjust to his satisfaction the difference between φ and Latin f, which he found emphasized by the earlier grammarians ; it appears to him (jiiite absurd, that φ should be a mute and / a semivowel, and accord- ingly he ends in making _/ likewise a mute^ A spirantic pronunciation of φ χ ^ is unmistakeably described by the Byzantine scholiast of Dionysius Thrax, who brings into pro- minence the decisive absence of closure in contrast to ττ κ τ^ The description is entirely at variance with that of Dionysius of Halicarnassus. But even Ulfilas makes no difficulty about representing φ by Gothic /", θ by \y. Latin monuments Λvith verj^ few exceptions shew ^;/< and / unconfused up to the time of Severus ; after that however the alternations become numerous, and after the middle of the fourth century there is no longer any distinction even in the best documents*. Before this and ^ Lejisius Stand. Alph. p. 202; Schwartze Kopt. Grainm. p. 79 ff. ; Steiu CojJt. Gr. 16 f Ou hierogly- phics the name of Philip Aridffius was written according to Lepsius with D ΓΠ ph, at a later period Φίλωτίρα, Ύρύφαίνα only with p. - Prise. I. 13 : quare cum /' loco mutae ponatur (in fama φήμη etc.), miror hanc inter semivocales posuisse artium scriptores — (14) sciendum ta- men, quod hie quoc^ue error a quibus- dam antiquis Graecorum grammaticis invasit Latinos, qui ^ et ^ et χ semi- vocales putabant. — Hoc tamen scire oportet, quod non fixis labris est pro- nuntiauda /, quomodo ph, atque hoc solum interest. This sounds quite differently from what Quintilian says, although Priscian also, true to his predecessors, makes φ pronounced with closed lips. '^ B. A. II. p. 810, on φ X θ in suc- cession. Π is pronounced with closure of the lips ; άνοι^ομίνων 5e τών χ€ίΧΐων ττάΐΊ', και τη/ΐύματο$ ττολλοΟ ίξίόντοί, έκφωνΐΐται το φ. In the case of κ the tongue is pressed against the palate, in t against the teeth ; χ on the other hand is pronounced ttjs ■γΧώττηί μη ΤΓροσπιΧονμένη^ μηδ ό'λωϊ συναπτόμενης τω ούρανίσκφ, and θ απο- χωρούσης της "γΧώσσης των οδόντων και παρεχούσης ίξοδον τφ πολλφ πνεύματι. V. Kaumer puts a right value on this testimony, p. 103 f. ^ W. Schmitz p. 122 f. ; Th. Momm- sen Hertu. xiv. 70 ff. The graffiti of Pompeii furnish only four exx. Or- thogra^jhic precepts on / ph are given by Caper vii. p. 95 K. (time of Trajan but not preserved in its genuine form) and Diomedes p. 423 f. K. (4th. cent.), Schmitz p. 126 ; cp. also Mar. Plot. [ iJDisHaAiJua ^8Π io8 THE I'UONUNCIATION OF ;is long lis [)h and f were distinguished pit and p, t/i and /, cli and c had been liable to be interchanged : the contrast between the earlier and later pronunciation is therefore evident. This later pronunciation however will not have arisen all at once, it must have needed time to have made its way I'rom the lower to the upper stratum of the people and to have become general. But its beginning or, if you prefer it, its prelude, is perhaps already to be found in the ancient Gi'eek dialects : on this point we go on to speak in connection with the transformation of the mediae. Section 29. Pronanciatwii of the Mediae; dialectal pivnnnciation of the Mediae and Aspirates. We have seen above, that the name media denotes a half aspirated sound, and not by any means a weak or voiced sound, with which names b d g are now denoted in contradistinction to ρ t k. The Greeks then heard a certain breathing in their β y 8: and who shall maintain, that their ears deceived them ? Moreover there is this confirmatory fact, that the mediae as well as the aspirates became spirants. It certainly may be maintained that the name mediae suits the present pronunciation also, in so far as the breathing in β ν is really weaker than in φ f\ On the other hand, since Latin b(j d and Greek /:? 7 δ correspond to one another with perfect regularity, and the value of the Latin mediae is certainly identical with that of the present Romance and German, the pronunciation of Greek /S 7 δ must have been approximately the same as that of our mediae. In the case of δ this is made especially clear by the itict, that it is so frequently confused with t by Egyptian scribes ; consequently there can have been no such wiile difference as that between SaccrJos (Hid. cent.) K. vi. 451. — uotac Tironianae, also Sehitderico for Hchniit/, p. 1:54 furnishes examples for Tlwodeiico on an inscrip. the conlusioii of /// ami s from the ' Cp. al.^u U. A. 810, n. 2. A NOTE NT GREEK. 109 modern Greek τ and h\ Strangely enough it is only the pronimciation of the β which has really been made a matter of controversy. However that this was during the Attic period not V appears suflficiently proven, in case there is still any doubt, by Plato ^, Avho calls it a mute, and by the βη βή of the comic dramatists, and it is by no means the case, as has been stated, that in the Roman period it was employed without scruple for v. On the contrary the inscriptions of the time of the republic shew almost without exception ΟναΧβριος, Φολουίος, and this mode of writing, tedious though it was, even in the period of the empire was never quite ousted by the far more convenient β^. There existed then a pretty consider- able difference between β and v, greater than that between semivocalic ν (English w) and consonantal ν (English v), for this Avould not have prevented the universal adoption of the Avriting with β. In the time of the Empire, especially from the second century onwards, this difference must have become smaller ; otherwise the earlier usage would have been pre- served. The Latin b too in many places had a similar develop- ment, being pronounced in the same Avay that survives at the present day among the Spaniards and many of the French of the south, whose vivere is according to the well known witticism hihere*. This indistinguishable confusion of the two sounds gave rise next to such spellings as ^6οναστό<ί, which is often met Avith on Greek inscriptions in Italy". But in the fact, that even at the present day β is an explosive sound when following ^ See Plat. Crat. 427 a: t^s του ly. The same writing was used in δΑτα συμπιέσεων και τον ταυ και αττερει- verse also ; C. /. Gr. G7 sq. Ξϊλδιιίοΐ' ffews τ•^5 ^\ώττην. (ύξάμενον with consonantal pronuncia- ■^ Theaet. 203 b: toO δ' αΰ βήτα. tion. The name of L. Verus is com- oiVe φωνή οΰτε ψόφον (cp. Dion. συνθ. monly written Ouijpos, much more 72). rarely B^pos, Dittenberger p. 304. In •* S. Dittenberger Henn. vi. 302 ff., many exx. also ν internal is omitted, who has only two exx. from the time Φαώνιον, Βόιλλαι, in short it is quite of the republic of β for ν (yet in Delos evident that the Greeks possessed no about 180 B.C. Bull, de corr. h. vi. 88, quite appropriate expression for r. 43, Dittenb. Syll. no. 367, 86, 130 ^ Corsseu 1-, 131; Diez Gr. 1, 280. λιβ'ιου Βιβίου) ; the Monumentum An- 376 ; Seelmann p. 239 f. cyranum also still shews ov consistent- ^ Dittenberger p. 304. no THE PBONUNC/AT/OX OF a nasal, Psiclimi* liglitly finds a proof, that it was originally this in all cases ; for hviiivos could not have produced koinhos, but an original υ woidd have done away with the nasal. As regards γ, this letter seems at all events when between vowels to have become a spirant at a very early period in the popular pronunciation. For a fre(|uent misuse of it on ])apyrus is to bridge over a hiatus: vyiyaivt.<; = vyiaiv€i<:, κΧαί'γω = κλαίω, Ύανηη<; Τίίγτ;? for Ύανη<; Ύάη<ζ, 'SapaTriyijov', and conversely it is frequently wrongly omitted : νιαίνη<;, 6\ίο<ί^, which latter form is also attested as Tarentine and is cited by the Attic comic poets as a barbarism of the demagogue Hyperbolus*. Compare further Φιάλ^ια = ΦcyάXeιa, ayr^o^a for ny^yoya, Boeotian ιών for εγώ, ayeOXa in Pamphylian^ All this points to a softening of the guttural explosive to a y, or in the case of a back-vowel to the g, which the Germans usually pronounce in Tage ; but the sound was so undefined and weak, that it was thrust in and left out at will''. The phenomenon was however in any case strange to the standard Attic, as is shewn by the sneer at Hyperbolus and probably neither Hyperbolus nor any one else at Athens who pronounced όΧίο^;, on the same principle pronounced λέγω as λεω, any more than a Boeotian said Χίω because he had ίων for eycov. Such cases as these have their source in isolated words of frequent occurrence — compare Italian io from ego, but not lio from lego — and may subsequently develope into a principle of universal application. In some of the dialects however other mediae also and not less other aspirates to all appearance became at an early period 1 Psich. Rev. crit. 1887, 2C7. 39 1. 22 (Peiiaicus, decree of Mace- - Pap. L. 03 col. 1 vyiyaivis and doninn Period) ; Bull. vii. IGG (Im- conj. L'7i7a/:'7js; κλαί-γω 51; Ται'γ?;? bros) Meisterhans, p. 59'-. Τάγηί 23. 55 Bis ; Σαραττ. 40. 41. » G. Meyer Gi:- p. 218; dy(0\a •' ΰίαίνομΐν and inaivis 42, όλ /os (13, Riihl no. 505 1. 24. 4 ; 20 twice. '' Cp. Wessely Wimier Stud 1882 ■' HcTodian 1, 141, 19; Plat. com. p. 197, wlio draws the general con- frg. 108 K. (in Herod, ii. 920) ; v. elusion, that y was pronounced like ij Herwerden 00 (C I. A. ii. 594, 8). before an E- or /-sound. The indica- Also on the inscr. of Chersonesos on tions or snggestions of such a pro- the Crimean Peninsula (Bull, de corr. nunciation arc however found with licll. V. 70 Dittenb. 252), όλίωι is found much greater frequency in the earlier ut 1. 10; see further Έ. άρχ. 1884 p. than the later period. AXCIENT GREEK. i I Τ spirautic. lu the first place so early as the pre-Roman ])oriu(l Laconian employs β in the place of the digamma : Boii;e[tSa9], Βίΐστ/α twice ; but always σθ. Getf/ioi' 121 (I). -I. 1147, 1157, 1159, llGl, Epidaurus Έφημ. 1885, 65/(J(J. G. 11G8), πάσκοι 1\. 112 (D.-I. 1152). Meyer- p. 291. ' G. Meyer2 p. 2()2 is much too ^ H. 552 (/).-/. IIGI). preciiTitate : "wbicli proves a pronun- " Dion. Thrax 11. A. p. ()32: {avy- ciation as in modern Greek." κΐΐταή τό ξ (κ τον κ καΙ σ, το δί ψ ίκ τοΰ 3 Β. 321, 322 (D.-I. 1478, 1479). π καΐ σ. Dion. Hal. Comp. p. 82 r to Also 321'••, at κ 6; but the aspirate is ξ διά, τοΰ κ καΐ τό φ διά τον π τον σνρι- never found written in the case of the •γμον άποδίδωσι, ψιλών όντων αμφοτέρων article. (cp. p. 78, 79). •« Ε. .321•»β; 113'• (]).-!. 1151). r Syrian Schol. Ar. Met. p. 910: Also on tlu! Xulhias inscrij). (R. (IM) τανττ) δί rij άττοδόσα (that these are ANCIENT an Ε Ε Κ. 115 'π-νβυματώΒες^, χ φ were liable to be heard instead of κττ; but that actual assimilation also took place dialectically and absolutely destroyed the explosive, is shewn by the ancient boustrophedon inscription of Naxos where €\-σοχος, Να|-σίο (Ν αξίου) are written with the symbol for the sj)iritus asper^. Unfortunately λ/γ or a substitute for it does not occur on the inscription ; on the island of Amorgos, which was colonised by Naxians, on inscriptions in the same alphabet the tenuis occurs in the case of ξ as well as ψ : κσ, ττσ^. Section 81. Pronunciation of Z. The third double-consonant Ζ presents a most difficult problem. In modern Greek this is always a simple sound, namely a soft or sonant s (French zero, German sagen, English zeal) ; the ancients from the time of Archinus and Aristotle* ahvays regard it as a double-consonant, just as much as |^ λ^ ; and indeed the grammarians make it consist of σ and δ^ (in καΐ TO A'i'Z συμφωνίας φασίν elvai (the three double cousonants are compared with the three musical chords, octave, fifth and fourth), καΐ δτι έκΐΐναι rpeis, καΙ ταΰτα τρία, δτι δέ μυρία αν €Ϊη τοι- αύτα (that the possible number of double consonants would be countless) ουδέν μέλει• το yap Γ και Ρ €Ϊη αν iv σημεΐον (a simple symbol might be devised for yp). ei δ' οτι διπλάσιον των άλλων (as the others) 'έκαστον (scil. of those three), άλλο δ' οϋ, αϊτιον δ' οτι τριών όντων iv €φ' 'έκαστον επιφέρεται τφ σ (ν.1. το σ, cp. Schol. p. 381), δια τοντο τρία μόνον εστίν, άλλ' ούχ οτι αι σνμφωνίαι TpeTs (the construction and argument are confused here, the schol. certainly had a different reading). ■' Dion. Thr. I.e. : to f έκ του σ και δ; cp. Schol. p. 780, 814, 815. Dionys. Halic. p. 78 : δίπλα δε λέγονσιν αύτα ήτοι δια το σύνθετα εϊναι, το μέν j" δια το σ και δ, το δί ξ' δια τοΰ κ και σ, το δέ ψ δια τοΰ π καΐ σ συνεφθαρμένων ιδίαν φωννν 8—2 three double consonants, owing to the thi'ee positions of articulation, as laid down by Aristotle himself Metaph. 1093 A, 23) καΐ 'Apxivos έχρητο, ώ? ιστορεί Θεό0ραστο5 • έλtyε yap 6 Ά. η έξω τι παρά, την μύσιν των χειλών εκφω- νεΐσθαι, ώσπερ το π, και δια τοΰτο το ψ προί τφ άκρφ γεν^ίάσ^αι ttjs yλώττηs ώ; έκ τοΰ π σ σvyκείμεvov • η τφ πλατεΐ ttjs y\ώττηs παρά. τού5 όδόνταί, ώσπερ το δ, καΐ δια τοΰτο το ^ κατά, ταντην γεί'^'ά"' ^«t την χώραν ' η τφ κνρτφ καΐ πιεζομένφ εκ τοΰ έσχατου, ωσπερ το κ, όθεν το ξ προιέναι (cp. for inscriptional forms note 4 below, and Styra Bechtel 19"^ ΧάροτΓϊ, though on 263 Μοφσίδη^ ; ώα- ypάπψaι Mykale Β; no. 144 does not tell us much). 1 Plato, cp. p. 105 n. 2 above. 2 Eohl no. 407 = Bechtel no. 23. ^ Bechtel Inschr. d. ion. Dial. no. 29 Aaμπσay6ρεω ; ib. note (Diimmler Mltth. XI. 99) Άλεκσοΐ. ^ Aristot. MeUiph. 1093 a, 20: ίπεί ii6 THE PRONUNCIATION OF this order). Archinus also says that it contains a h, and on this point certainly there ought to be no dispute. The German pronunciation giving it the sound (^9) of their own ζ is of course a mere misuse and is not defended, but many modern philologists imagine its sound to have been somewhat like zz (double sonant s) and endeavour not without a little violence to bring the authorities into harmony Avith their theory*. Such speculations as these I cannot follow but rather believe, that the sound, Avhich men like Aristotle and Dionysius of Halicarnassus heard, must have really existed. But with refer- ence to the sequence of the two elements G. Curtius also has entered the lists against the ancients supporting the pronuncia- tion (Is (more correctly dz, with the French value of ζΥ. This pronunciation too can bo designated as traditional; for in Italian the ζ of Greek words has still this sound {zelo, zeta), and it is easy to shew that the tradition goes back to an early period''. On the other hand, according to that excellent authority Psichari, the pronunciation of ζ as dz which is at present current among the Greek islands is not to be regarded as in any Avay traditional, any more than the pronimciation of σσ σ as ts (τβτσαρα, άτσήμι = ησ. " silver "). Psichari states that in Chios, the various stages of this modern development may be observed side by side : nomi"zo, nominzo, nomindzo*. Moreover, as dy is etymo- \αμβάνοντα, ij δια το χωράν ϊπέχειν δνΐΐν -γραμμάτων ΐν ταΪ5 συ\\αβαί$ τταραλαμ- βανόμΐνον ΐκαστον. — ρ. 82 : τριών 5k των άΧΚων γραμμάτων α. δη διττλα καλεί- ται το ϊ" μάλλον ήδύνει την άκοην των έτερων το μ^ν yap f δια τοΰ κ και το ψ δια τοΰ 7Γ τόί' σνρΐΎμον άποδίδωσι, ψιλών όντων αμφοτέρων, τοΰτο δ' ήσυχη τφ πνΐύματι δασϋνΐται (on account of the media δ contiiined in it), καΐ ίστι των όμο^ΐνων "γενναιύτατον (the noblest, most euphonious sound). This passage is wrongly interpreted by Ascoli Krit. Stud. p. 3G5 f. of the derman trans., who finds in it an indication of tlie sound z'z'). — Scxt. Empir. j). (i()2, Bk.; Bekk. Antr. p. 1175 (j" cannot like ξψ stand as a (inal, οώτι ΐκ τοΰ σ και δ δοκΰ aiiyKeiaOai, ουδέποτε δί λέξΐί "Ελληνική eh άφωνον καταλη•γ(ΐ). The evidence from Greek sources is therefore un- animous except the scholia on Aristotle, in which certainly (p. 331 it, 33, 42) the σ is denoted as the second sound for all three double letters. For the Scholiast thus understands the ivi- φέρΐται of Arist., which however in this author (s. Bonitz Index) by no means has the later meaning 'follow'. 1 Ascoli (see preceding note). - Curtius Grdr''. p. 615. ■' We have also the testimony of the liatin grammarians, see below. ■' Mondry Beaudouin Hull, dc rorr. lull. IV. p. 3()(i (Carpatluis). ANCIENT GREEK. I i 7 logically at the root of ζ, dz may easily have been devulopeel from this just as in Italian mezzo i.e. meddzo comes from medius {medyus), orzo fi-om liordeum {ordyum); diurnus giorno (dzorno) also is essentially analogous. Accordingly this pronunciation too has its claims, and moreover the origin of the modern Greek pronunciation as simple ζ requires illustration ; the third and not the least Avarranted pronunciation is that main- tained by the grammarians, namely sd or more accurately, since s must be soft before the media, zd. Let us endeavour then to do justice to each one, assigning to it its province and period. It is a well known rule that in Attic and Hellenistic Greek the preposition συν loses its ν in combination Avith initial ζ: συζητζΐν, συζβν^νύναι, σνζήν. If now d was the prior element in the compound letter ζ {syn-dsen), there was no reason for the rejection of the ν ; we find συηξέω, σύμψηφος. But if the pronunciation in Attic was sd, sy{n)sden is perfectly analogous to συ{ν)σ7Γαν, συ{ν)σκ€υάζβι,ν. Here then we have our first confirmation of the tradition of the grammarians \ In the next place the preposition e| must of necessity lose its s before δ; before σ it need not. Now we find on the Attic maritime documents in big letters as a title βξ 7ika; 'ITD coins of the Satraps, 700 (li. 300); Thossaly 345'', Λ///ί/«•ι7. Mazdiik head of a sect 500 a.d., see ΑΛΊ ΊΑ' Λ' τ ah• Ε Α' Κ. Ι 1 9 ingly iroiii all this I infer that in Attica and central Greece generally ζ, so far as it occurred in the dialects, had the sound of zd, and that this pronunciation was circulated and main- tained at least in the case of the grammarians in the Hellenistic period up to the second century a.d. But we must now go back to the oldest form of Greek and especially of Ionic, as we find it in Homer. Here έ'ρδω is found beside ρβζω from the stem fepy; why not €ρζω, if ζ = els'? It is on the other assumption quite intelligible \ that 'έρσΖω should not remain, but that the σ should fall out. Moreover initial ζ does not make length by position in ZeXeta Ζάκυνθος, from metrical necessity ; the same is true of σκ in Χκαμάνδριον, σκέτταρνον, σκίή. Now here the reading ΙίαμάνΒρων has authority", and virelp aXa κίΒναται ηώ<ς is read universally; moreover σ often falls out before ττ and φ, and accordingly it could easily be di'opped in 'ΧΒάκυνθος also^ I do not think its omission in Ασάκυνθος would be so easy. — Δα stands in the place of the prefix ζα in Βαφοίνός and Βάσκιος. This is quite natural and easy, if ζα = σδα, σ8αφοίνό<ί just as Ζάκυνθος and σ^άσκιος is difficult to pronounce. — Also it is admitted by all that όζος comes from οσΒος (Germ. Ast); according to my view there is here no change of sound whatsoever. Μαζός is the Homeric and Ionic form for μαστός, μασθός of other dialects : the former is μασδός. We must add ΐζω i.e. ισΒω, βζόμην i.e. βσΒόμην which are analogous to Ισγω, βσχον, where likewise, in Homer as in Attic, the original sound has simply remained. And, I think, in Ionic also; €ρ8ω at all events is used by Herodotus, and the latter writes "Αζωτος, where, if the sound was shd and Greek ζ Avas ds or z, there Avas nothing certainly to jDrevent him writing "Ασδωτο9. If the undoubted origin of ζ in dij be brought forward in opposition to this view, it would at once be easy to shew that zd too may originate in this, since it is a fact that in old Nuldeke Ber. Wien. Akad. 1888, 414 f. time of Herodotus downwards. ' λρταουαςηί Plut. Crass, c. 19, 22, ac- i Osthoff Perf. 596, 1. cording to cod. Ν Matrit. (see Charles - La Eoche Horn. Uiitcrs. p. 42 f. Graux, de Plut. codice Matrit., Paris s Cp. Thiersch Gi: Gi: § 146, n. 8. 1880, p. 55). "λ^ωτοί universal from 120 THE riiONUNCIATION OF Slovenian dya regularly becomes zda\ It is true that tlieie tija also becomes sta, whilst in Greek rya becomes -ττα or -σσα according to the dialect. But those who adopt Curtius' assumption are eijually unable to shew any analogy between the treatment of ty in Greek and that of dy, and they have to explain what is absolutely surprising, namely that the same language admitted dz but not ts^. According to my view, it has (at least universally) admitted even dz, since not only has this in many dialects become δ δδ, as ts has become ττ in Thessaliau, Boeotian, Attic' and also Cretan, but also other dialects have transposed the two elements. Attic might very well reject δδ, although it preferred ττ to σσ, and the Doric of Delphi, Avhich had σσ for ty ts, might nevertheless avoid the corresponding assimilation in the case of dz, especially as the sound of the soft s only existed in the language in combination with a consonant, while here it would have been independent. On this side then there is really no obstacle; on the other hand it is certainly perplexing to meet with σδ, i.e. the Attic sound of ζ, as a dialectic peculiarity of the Lesbian and some other poets as Alcman and Theocritus*. Be it remarked however, this is only in books, not on in- 1 Rliklosich Altslovcii. Lautl. p. 270. '^ I would however suggest, that the sound ts is hidden beneath the writing T, which occurs in Halicarnassus li. 500 (5th cent.): ΑΛΙΚΑΡΝΑΤ[ΕΩ]Ν and ΑΑΙΚΑ[ΡΝΙΙ]:ϊΣΕΩΝ AAIKAl'N- ΙΙ^;Σ0Ν, ΟΑΤΑΤΙΟΣ, ΙΙΑΝΤΑΤΙΟΣ; also on coins of the Thracian Me- Bcmbria: ΜΕΤΑ:\ΙΒΡΙΑΝΩΝ (Kirch- hoff•* p. 12), and according to Eiihl's suggestion (p. 13'J) in GAAATHi: Teos 497 B. 23. For in these Carian proper names on other later inscriptions -ttffffis -αξ« is written for -ATIw, BuJl. ile con. hell. iv. HIO, v. 580, vi. H»l (Bechtel lOi, 231), 240). "^ I still have no doubt in spito of Ascoli (Krit. Stud. 321 ff.) and in spite of (J. Curtius' recantation (Klijnt.'' (IGl)) that TT and σσ both go back to tn. AscoU's proofs of the origin of ττ in σσ are all of a very problematical character. It seems to me also suffi- ciently certain, that ττ was a peculi- arity of Euboca and Oropus, although Bechtel Iitschr. d. ion. Dial. p. 13, 37, still doubts it. Κίσυ$ Styra no 1ί),•'**^Β. and Κισα- do.•"^'- are too obscure in their derivation, to be of any use as instances. — On Crete see p. 122, n. 4 below. ^ Ahrens 1). A. 45 ff. ; Meister Gr. Dial. i. 129. This usage is not constant either in the Aeolic poets or in Theocritus; the rule which .\hrens tries to institute is doubtful. Cp. Morsbach dial. Tluocr. Curtiiis Stud. X. 31 II. ANCIENT a REEK. 12 1 scriptions ; tlic Lesbian inscriptions as early as the fourth century have always ζ"*. But an antiquated spelling might easily be transmitted in the manuscripts of poets ^ and be adopted by artificial poets like Theocritus. It appears to me, that considering the few fragments which we possess of the Lesbian poets and the almost entire Avant of early ^olian inscriptions, we cannot yet expect a satisfactory solution of this riddle. I would suggest however, that the ^olians pronounced sd as the Athenians, but wrote this with two symbols, employ- ing ζ for that sound, which in their dialect arose from δί- before a vowel : ζά = Scd, κάρζα ; this sound must have been ζ {dzY, and for such a ζ no one cites any instance of the writing σδ. A difficulty of a different sort is the Delphian καταΒουΧιζμώι*, evidently pronounced -zmoi, in a dialect which we have claimed for the pronunciation zd. This orthography Ζμύρνα ζββννύναί is, as was mentioned under σ, very Avidely circulated in the Hellenistic and Roman period^; in itself however it by no means proves the simplification of the ζ. For Σμύρνα did not represent the actual pronunciation zmyrna (with soft s) with greater propriety than 'Δμνρνα, in which latter spelling the d became mute spontaneously. Thus in the Attic period also we find beside 'Ω,ρομάζης, Avhere ζ — zd, Φαρνάβαζος Ύιρί- βαζο<;, Avith ζ= Persian z'^. It is however noteworthy, that ζμ appears so often subsequently, and moreover the alterna- tions between σ and ζ are not entirely limited to this case. We find on an inscription of Cnidus ζήζαζα (ζήσασα)\ on 1 προσονυμΑσοεσθαι on an iuscrip- ■' σ^ also occurs occasionally before tion of Cyme of the Roman period μ : ένδέσ'ςμουί Ath. (Macedonian period) (Cauer no. 127=Di«/.-I?!.sc/ii•. 311) is "Άφ. 1883, 125 f. y, 12; 'Έ,ρασξμία of com-se only an affected archaism. C. I. A. in., 1553; χρησ'ςμόν Cos - The grammarians themselves re- Bull. dc. corr. h. v. 228 (to be divided gard it merely as a matter of spelling, as χρησ-ξ'μόν). putting it in the same category with '' Coins of the Satraps 1T3313, ^olic Kaevos Πελοττϊ lepaKs (Ahrens p. 1T!l''"in, Niildeke Ber. Wien. Akad. 48 f.; Meister 127, 1; K. Schneider 1888,415,419. Bodleiana p. 43). '' Kaibel Eiiigr. 20^^; a few other 3 With διά — ζά cp. τια {τίνα) σά in exx. Keil Bullet, de I'acad. de St Pet. Doric (Ahrens D. D. 277). 1857, p. 179 (Mel. Greco-rom. ii. •» Wescher-Foucart 433, 13. 38 f.). I 22 THE PKONUXCIATIOX OF ordinary pajiyri ύβριζαν, eaoyi] {€ζύ<γηνΥ : as a general rule it is true the writers of the papyri know how to distinguish the two letters. In the next place, against the value zd we have the Hellenistic spellings ΆσΒρούβας, ^ΕσΒρας, Άσδώδ, Άρτα- ουάσ8ης, ^Ω.ρομάσΒου^ ; for in the case of ks ps ξ ψ are ahvays used in these transliterations and adaptations, and I would also confidently suggest, that the presumably Carthaginian name ASOTJiH (gen.) on a Theban inscription is really ΑΧΡΟΤΒΩ^ So far then we should conclude that the modern Greek pronunciation prevailed in the Hellenistic iwpidar language», while for the preceding era we have as yet only found the sound zd. And certainly zd could be simplified to ζ by a graduall}^ weakened pronunciation of the d ; but this is true to a still greater degree of dz, the claims of Avhich must now be put to the test. Now ζ occurs to all appearance with such a value, ts or dz, on old Cretan inscriptions : οζΌ? i.e. οσο^ (irom Ότι/ο<; = οτσος), άνΒάζαθαί = άνΒάτσαθαι άν{α)8άσσασθαί {ζωώ= ζωοΰ)*. But this disappeared in Crete at an early date, and ττ or δδ according to the circumstances, and initially δ was written for it. Thus the Gortynian inscription ; later on we 1 Pap. L. 40, 41 (ϋβρ.) ; pap. Weil col. 4, 14 ; ib. 5, 1 φορντί'ζ^ίν φροντί- σιν. All these pieces are more than avcragely faulty ; e.g. the Paijyrus of Hyperides on the contrary shews no- thing of the kind. (The attic ψηφίαΐαΟαι for -ζίσθαι Boeckh Sec-Urku)ulen p. 467 does not exist; see C. I. Λ. ii. 809•', 35.) - We find on the Monum. Ancyr. col. 5, 20 Artavitsdis Greek Άρταονάσ- δον, 29 Art(ibii{zi) ' λρταβάξου, 30 Arta- vasdi Άρταουάσδη, G, 11 Artavazdis ' λρταβάξου. Cp. Mommsen p. 110, 1 ; p. 118, n. 6 above. — Ώρομάσδου Inscr. of Antiocbus of Conimagenc(09 — 34 B.C.), Puchstein JScrl. Muitatsbei: 1888, 49 ff., col. 1'' 19; 11" ΙΟ.— Αριοβαρ^άνηί is written by Greeks and Latins with 2 ; tlic pronunciation was probably here, wliere in any case there was position-length, generally simplified. In Herodotus however (7, 2 f.) we find ('λρτο)βαί;άνη$, and I think the Athe- nians wrote it thus, though now we find in the texts Άρωβαρί'. (the latter also C. I. A. II 481 c^ 1st. cent. B.C.) •' Νώ/ίαι/ (Accus.) Ά. Dial.-Iiixchr. 719 ; Meister writes here 'Α(σδρ)ούβω. The inflection according to the 2nd decl. is certainly strange, especially beside 'Sώβav. * Comparetti Mtis. Ital. ii. 131, 142, 162, 172, 194, 202 f., 210, 212, 224, 674; hitlierto f had not occurred on old Cretan inscr. Further discoveries are certainly pressingly wanted, in order to tln-ow light on ήλιζίαι { = i]\iKlq,) Ροί'ζηα ( — Ροίκηα) and such monstrous forms. ANCIENT GREEK. 12 uiid iiLsu θύΧαθθα ;iud alsu Ύήνα TrPjua Αηνα {'Δήνα)^. Of these θάΧαθθα with spirantic θ appears to be a sort of compromise between real Orotic θάλαττα and the ordinary θάλασσα ; but Ύτήνα might be something like ddena, with a stronger articulation of the initial letter, to express Avhich the tenuis was brought into requisition. In any case the original ts dz in Crete disappeared in later times Avdthout leaving any trace, and where in the Greek-speaking world can it be said to have continued its existence ? As a matter of fact we find no trace of it on Greek soil, but only on Italian. For the Italian peoples, the Latins, Faliscans, Oscans and Umbrians, employ Ζ always either for the soft >S^-sound or, and this must be the more original, for ds ts ; consequently those Greeks too, λυΙίο brought the alphabet to them, that is especially the Chalcidians, must have possessed the ζ with the value of dz^. Whether they retained it with this value, or later on like the Cretans rejected the sound-combination dz, is of course another question ; in Italy however ζ may have been maintained as ds ts, whatever the Chalcidians did, and this would explain the fact, that this value appeared again in the time of the Empire and has continued to the present day. The Latin Grammarians, although sometimes under Greek influence they resolve ζ into sd^, nevertheless maintain elsewhere, that it is equivalent to 1 Meyer- p. 217, 256, 273. It is best however not to venture on such far-fetched solutions of the riddle as Meyer, who finds a palatal {j in the initial sound of Ττηνα. - On this cp. Corssen Ausjir. 1- 395 (Osk. horz hortus, -azitm -arum; Umbr. inliaz piatus, menzarii mcnsa- rum). The whole subject of f in the Italian languages has been worked out by L. Havet, Memoires de la societe dc linguist, iii. 192 — 196 ; I owe my knowledge of this to a kind communi- cation of the author himself. Besides the three values of f which we have discussed he gives as the fourth and most universal zz, as an instance of which he gives the Latin badisso; but this I cannot at once follow. The curious Oscan Ί^ίνμσδιηί^ (Messana, Mommsen Unterital. Dial. p. 192), where σδ represents the simple soft s- sound (the name is written elsewhere Niumsi-, Niunipsius, 'ι^ύ{μ)ψίο$), is ex- plained by Havet by the assumption, that with certain Greeks σδ too had become the simple sound z. It is quite as possible however to say that the Greeks not possessing the simple ζ without δ interpolated a d after it, zd being famiUar to them. (For σδ instead of f cp. on the same inscr. ττσ in ονπσενί for ψ.) ^ Victoriu. K. vi. 196, on the Virgilian Mezentius with the vowel of the first syllable long by position : quae 124 THE PRONUNCIATION OF ds {ts), it lliL'y do iiul actually deny altugethcr the cunipouiid nature of the lcttcr\ Moreover in the vulgar writing of the later empire ζ appears representing di followed by a vowel : Aziabenicus or Azabenicus, zeta (diaeta), and also for j (y): — cozugi, Zanuari'', no doubt in the same way and having the same value as in the common Italian mezzo and the Venetian viazo7'e. To sum up then, the following seems to be the result of the whole investigation. In ancient times the Greeks possessed the sound-combination zd, in όζος βζόμην etc., and beside it a dz which was developed from di/, to which corresponded a ί6• from ty. The latter sound-combinations however did not hold their ground, the result being that hizdo and nomizdo, the former original, the latter from nomidzo, coincided in sound. To denote zd the Phoenician Sain was taken, ^^^hich in Semitic signifies simple ζ (soft s), partly also as it seems dz ; similarly Samech (s) had to serve for ks. In those places where dz (z) si adsumiJta non esset, per s et d Mesdcntiiim scriberemus. Cp. Tereut. Maiir. v. 021. ^ Mar. Yictorin. K. vi. p. G: sic et z, si modo latino sermoni necessaria esset, per d et s litteras faceremus (obscurely p. 34). Vel. Long. K. vii. 51 : atque has [tres] litteras (x also as well as z) semivocales plerique tradiderunt. Vcrrio Flacco (time of Augustus) placet mutas esse, quo- niam a mutis incipiant, una a c, altera a d (mss. a p). quodsi quos movet, quod in semivocalem dcsinant, " sciant," inquit "s litteram per sd scribi ab iis qui putant illann.ex s et d coustare, ut sine dubio muta finia- tur." niihi videtur esse aliud z, aliud σί-γμα και δέλτα, uec eandem potestatem nee eundem sonum esse, sed secundum divcrsas dialectos enuntiari. Dores cnim Kcimus dicere μ(\ίσδ(ίν, alios μίλίΐ'ίΐι/, nee idco tamen eadem littera est, non magis (juam cum alii κ(βα\ήν, alii κίφαλήν, alii 6ππατα alii όμματα, alii ΟάΧατταν, alii (θάλασσα;' dicunt, cum idem dicant. He goes on to deny that 2 is according to its actual sound a double consonant ; for it is, he saj's, susceptible of being doubled and in pronunciation it has not, like .r, a dis- tinct sound at the beginning and end of its utterance. This gi'ammariau then (time of Trajan) evidently pro- nounced a simple modern Greek ξ. For the very reason that c in itself was not a double consonant, some wished to write Mezzentius in Virgil, K. L, Schneider p. 380. — Martian. Cap. in. § 257 considers the sound of Greek ξ to be ΤΣ. — Against Seelmann Auspr. 308 I remark, that the passage Quintil. XII. 10, 27 f. does not refer to ξ and υ, but to φ and ν : quos mutuari solemits refers to speaking, while he comes afterwards to writing, and in doing so speaks of /" (and m) as com- pensatory letters belonging to Latin, tpuittiiiii ill Greek. So tspulding and before him (rcsner. -■ Corssen 1-, 215 f. ; Seelmann p. 23'J, 320 11. ANCTKXT CREEK. I 25 was in use, as long• as it hold its ground, it too and also U were represented by Sain = Zeta ; with this value it reached the Italians. In other localities it was otherwise, according^ to the wants of the dialect ; in Elis Ζ was used for spirantic δ\ In the pronunciation zd however the sibilant gradually over- powered and extinguished the d; if in spite of this the sound continued to form length by position, the sibilant must have been doubled, and this certainly presents difficulties in the cases where it was initial. There is however no reason to assume that the simplification of the compound took place before the Hellenistic period ; possibly the Macedonians were the originators and propagators of the change, the sound zd being strange to them. During this period there is no cause for surprise, if we find ζ for Sain in transliterations, as in Γάζα and the numerous Hebrew names such as Ζαγ^αρίας, or for English j = dz, in Indian names such as Όζηνη Ujjaj/ini. Correspondingly on a bilingual Attic inscription we find Sain as the Phoenician equivalent of ξ" in Βυζαντία'\ Section 32. Assimilation in Word-nexus; Hiatus. We have yet to make some general remarks on the combination of words and on their accentuation. With regard to the first point the Greek language appears to stand midway between the Sanskritic method, where the single word is modified by the surrounding words in the main in the same way, as the elements of a single word are modified by one another, and the method of our own language, which allows single Avords, and indeed any separable parts of a word, entire independence. We have spoken above of the assimilation of the final nasal, probably this was carried out still more in pronunciation than in Avriting. On the other hand in the case of final ρ and σ, as Avell as ξ and ψ, assimilation does not take 1 See p. 113 above. cent. b. c.?): TUn η^Ί;3 Ν:")Π = Ε(ί)- - Corp. Inscr. Semitic, no. 120 (3rd. ρήνη Βν^'αντία. 126 THE PRONUNCIATION OF place or only in a very .slight degree. For instance the combination κσθ is not suffered in the interior of words, but σ is rejected (ττβττΧέγ^θαί for νβττΧβκσθαι) ; in the case of final ξ however thLs takes place only in very close combination, namely in the case of e| and at most also in βξ ττύξ Χάξ. As regards the prepositions we must remark beforehand that the language, and this is true of Latin as well as Greek, made no distinction between their combination with a verb, where we Avrite one Avord, and that with a noun; there was the same close connection and consecjuently the same assimilation'. The only way in \vhich we practise this in the case of έξ, is to write eK before a consonant, i.e. to reject the σ ; but the Greeks even in writing as.similatcd the mute to the folloAving sound Avith great regularity, the tenuis only standing before /c τ ττ χ σ, before θ φ, and at an earlier period before σ also, εχ was written, before media or liquid €7^ And this was so establi.shed as a u.sage in writing, that it is found regularly even on the papyri, though there in the case of iv and συν contrary to our custom the assimilation is omitted. "Εγγοζ/ος also comes under this head, i.e. eK'yovo^, certainly not to be pronounced eiigonos and derived from iv^. The Boeotians and Arcadians however assimilated the ξ in quite a different Avay, namely by rejection of the κ : eaheXkeiv (έκβάΧΧειν), i'iifcat' ipsao horridulac Catonis (in which etc. tliereforo a hiatus must frequently ^ § 3G : ct coeuntcs littcrao, quae have been suppressed even in script), σΐ';'ολοίο! dicnntur, etiani levioroni indicant onines poctae, etc. This is faciunt orationeni, ijuara si Diiiiiia ANCIENT GREEK. 1 29 case irmst have been the same with the Greek of that period. Dionysiiis of Halicaraassus found the hiatus μαΧΚον he o\ov in his Demosthenes, and imagined that this was really intended by the orator', evidently only because there were speakers at that period, who alloAved this in speaking and did not get rid of it by synalepha. Demetrius who is somewhat later considers it actually more euphonious, to pronounce the vowels separate in the sentence ττάντα μβν τα νέα καϊ καλά eariv, than with synalepha καΧά 'σην'^; the people however no doubt even at that period pronoiuiced in the latter way. For even the Greeks of the present day are accustomed to annul the hiatus, at all events in speaking. Section 33. Transference of final consonants. In ancient Greek, just as in French, though hardly to such an extent, final consonants were liable to be carried on. The teaching of the grammarians is^, that where elision of a final vowel has taken place the consonant preceding this must be given to the following syllable : κα-τβ-μον, ά-ττβ-κβί-νου, just as in French en-treux. Wherever in composition a consonant comes before a vowel it belongs to this vowel without any exception, even in the case of e^ et? ττ/οος δυσ- ; on the other hand, if a consonant follow, the final consonant remains Avith the preceding vowel ; thus e-^t-ivat, δν-σέλ-της, but Βνσ-μυρ-φο<ί. In the case of σ indeed, as has been already remarked, the right analysis even in the case of simple words was a matter of doubt ; hence these rules, which were of course capricious. verba siio fine cludantur, et iionnun- συναλήψα^ e'iirois καλά 'στιν, δυσφων6- quam hiulea etiam decent faciuntque repov ίσταί το λε•/όμενοι> καΐ ΐύτβλΐσ- ampliora quaedam : ut Pulchra etc. repov. 1 Dionys. Dem. 42. ■' Tlieodosius Bekk. Anec. 1127 f.; - Demetr. ir. έρμ. § 70: ττολλά δ^ ed. Giittl. p. G2; Lentz Herodian. 11. καΐ άλλα iv συνάΧοιφτ) μΐν λε•/όμενα 390ff., 407f. Vid. Κ, Ε. Α. Schmidt ούσφωνα ην, δίαιρεθίντα δε καΐ σν/κρονσ- Beitracje ρ. 134 ίϊ. θέντα (ύφωνότερα, ώ? το πάντα κτέ. ei δε Ρ, 9 I ςο πι κ ΙΊίΟΚυχαίΛΤΙΟΝ OF We are at liberty to doubt, whether the pronunciation really was so entirely established and certainly whether it continued the same through the different periods. The writer of the great Hyperides manuscript indeed always separates a \ 7re- στελλετε and so on where the line breaks off, but he writes more freijuently ela-ayyeXia than βί-σαγγελία, and moreover sometimes ταυ \ τούχ, sometimes ovS' | οστί?'. On the long Epidaurian inscriptions, which sing the praises of the miracles of healing worked by Apollo and Asclepius with classic men- dacity, the following exam])les of line-division occur : ώ \ aBe (beside τάχίσ | τα γασ | τρί'~ υ \ στρακα), βξ | έΧθηι, α \ ττά- youra, e | vvirvtov, e | κτούτου. The pronunciation and separa- tion οι; I κέστί ου | ■χι'^κιστα^ was certainly established. A transference between article and noun (τώ I ν'έρ^ων) and also between other looser connections may have taken place fre- quently'*, but they did not divide so (in Avriting) except in rare instances, which are paralleled by instances of the opposite such as οσ-09 of equally little significance. A peculiarity worthy of mention, which appears on the Gortynian inscription and elsewhere sporadically, is the doubling of final ν in short words in close connection, so that it belongs to both syllables: ταννημίναν, συννηι (i.e. avvfj), ώνναν, ηννεγ^ων^. Although Biicheler is of a different opinion, I think that this pronuncia- tion gives the explanation for corresponding instances of licence in prosody in the .^olian dialect: — ησννβτημι, ^νόχΧης, σνν ολ/γω". ' Hyper. Praefat. p. ix. - Έφ. άρχ. 1883 p. 199 ; 1885 p. 15. Cp. later Attic inscr., which also finish the lino with ii complete syllabic: ii. 409, 35 i \ V άστα, -108, 17 t \ κ των. — Inscr. of Antiochus (p. 122, n. 2 above) II." 2H TTpo I σώδουϊ ; iv." 15 ιτροσ | καρ- τερΐΐτωσαν is necessai'y, because on this inscription σ is always separated from τ θ etc. =* ib.; C. I. A. II. •1(•.7, 81 οι- | λί'άσαί, H79, 3 οι' | KUXiya. •* Several occur in the second Hyperides mss. Praef. p. xvi. •'■ Gortyn. Insc. 2, 49 ; 10, 41 ; Miiseo Itul. II. 599 col. vi., 9 ; also νσσ in τ6νσσίιηβα\\6νταν^ 7, 9 ; but not σννΐσσάζ,αι 3, 10, which comes from ϊκσάττω. — Samos Dittenb. Si/ll. 132, 12. 15. « Meistcr Gr. Dial. i. 148 (Biich- ch'r /.'//. .1/».•?. XI.. addititinal fasciculus 1•. ;•)• ANCIENT GREEK. 1 31 Section 34. Accent. Witli rcofard to the accent of words it is well known, that in Greek this consisted in voice-pitch, not voice-stress and still less voice-duration, although in both the classical languages the latter was united with voice-pitch in the period of their degeneration \ For the Greek of the present day pronounces accented vowels long, unaccented short : xentis (ξβνους), yenUo (ryevoLTo), άθνΰρυβ, αθτόρα8. The period, at which this extra- ordinarily important transformation took place, may be to a certain extent ascertained from prosody. For the versifica- tion of the classical period makes no account Avhatever of word-accent", and indeed, since the accent was purely musical, there was not the slightest reason Λvhy it should ; but even tunes, according to the testimony of Dionysius of Halicarnassus, Avere set without regard to the accent, that is the tune (cp. ττροσωΒία, accentus) of ordinary speech I But in the post- Christian period we find it the rule in Babrius' fjxbles, that the penultimate syllable of the choliambic always bears the accent, and Nonnus (end of the 4th century) never ends a hexameter with a l3roparoxyton^ In the case of the pentameter an ever-increasing effort can be traced right on from the Alexandrine period, to limit accented final syllables, and finally almost to banish them, and this Avas done for the most part in favour of the paroxyton termination which prevailed also in the Byzantine trimeter^. In just the same way in the Latin of the 1 This is not the place, to enter on 211 ίϊ.) ; I cannot however regard his the controversy with regard to Latin, attempts as successful, and see little for which Seelmann also maintains an trace of anything of the kind in Latin original predominance of the factor of versification. Cp. H. Weil Gi'ittinger stress over the musical. The evidences Philologenverg. (1852) p. 85 δ". ; Weil- of the grammarians, to which he Benloew Accentual, lat. p. 66 ff. 240 E. appeals (28 f.), are certainly all late, ^ Dionys. Compos, p. 63 R., with while the musical factor (and that examples from Eurip. Orest. exclusively) is testified to so early as ■* This law was discovered by A. by Nigidius. Ludwich (Fleckeis. Jw/irt. 1874, 441 ff.). - Attempts have indeed been made "' F. Hannsen 7?//. Mux. xxxviii. to shew that there was some such 226. regard (v. J. H. Heinr. Schmidt ^letr. 9—2 132 nil•: PRONUNCIATION OF same period the transformation in the pronunciation may be ascertained from tht• metrical phenomena'. Further accurate observation of the Greek poets has of late led to the assump- tion, that there existed in the language from the earliest period side by side with the variety of pitch a variety of stress following laws coinciding with those of Latin accentuation : namely the stress is said never to have rested on the last syllable and on the last but two only Λvhen the penultimate was short*. It appears to me however still doubtful whether this is the true significance of the observations. With regard to pitch and tone ^\Q are told by Dionysius, that the interval betAveen high pitch and low pitch syllables amounts pretty nearly to a fifth I Now our accentual system, based on the statements of Dionysius Thrax, Dionysius of Halicarnassus and others, distinguishes only three kinds of syllables, high pitch {οξβία προσωδία), low pitch {βαρεία ττροσωΖία) and those in which high and low pitch are united (in that order) {ττροσ. irepi- σττωμύνη, so called from the 'draAving round' the accent from high to loAv)*. This kind of accentuation or that corresponding to it, in Λvhich the secpience is from low to high, occurs in modem languages also ; for instance the ancient circumflex is heard in Italian in the case of double consonants (donna, stella). Since the time of Aristophanes of Byzantium the low pitch syllables have been denoted by -, the high pitch by -, the circumflexed by the combination of the two symbols ^ , which gradually became rounded. Originally every syllable had its \ / \ V accent: ΛΕΓΟΜΕΝΟΙ; but in course of time the notation • Weil-Benloew p. 2•')ό ff. χωρίου τούτον vXtlov tVJ to βαρό. On - Isid. Hilbeifi, das Prinzij) tier Sil- the contrary, afterwards page 02 : ή δί hemciiiiung, Wien, 1879. Cp. Hannson 6μ•γανικτι τ( και ωδική μονσα διαστήμασί Rh. Mtif. XXXVII. 252, who, thouj^'h τ( χρηται η\(ίοσιν, ον τφ δια irh'Tf agreeinR in principle, make; the law μόνον, ktL of accentuation run quite differently; * Dion. Hal. γ. (>() IT.; IMon. Thrax the last syllable, if long, has the strong p. 020 Bk. : τΟιόϊ ^στϊ φωνψ άττήχησίί stress ; if not, the j)enultimate. ϊναρμονίον, η κατά άνάτασιν «V τί} όζιίςί, ■' Dionj's. Comp. p. 58 : διαΧίκτον η κατά 6μα\ισμ6ν ΐν tjj βαρ(1η., -η κατά μίν ονν μ(\θί (νΐ μΐτρΰται διαστήματι π(ρίκ\ασιν ίν τί) τκριαιτωμίντΓΐ. Varro τφ \(•^ομ{νψ διά irivTt, ώί (γγιστα' και {{Scviiii\ c.rphiii. hi Ihm. Κ. iv. 581) οΰτε ίτΓίτίΙνΐται iripq. των τριών τόνων hands down to us several other names και ήμιτονίον t'lri το ΰξΰ, οϊ'Τ( άνίίται τοΰ for 7γιιΤ7γ. : δίτονοί, αι''μΐΓ\(κτο! etc. ANCIENT GREEK. 133 was simplified; the gravis being placed oidy on the penultimate syllables of oxy tones and perispomena instead of those accents being used, or on final syllables, Avhere the high pitch was partially su})pressed in the speaker's context, to serve to denote such suppression \ Here the imperfection of this accentual system becomes evident ; for it is obvious that in 6 h" djaOd^ ανηρ άύ... all the syllables do not really have the same pitch. Accordingly even in ancient times more accurate systems were put forward, Avhich, Ave may say, fortunately never attained general circulation, but unfortunately have not even been properly handed down to us''. For example many distinguish a μ.€ση, which was recognized also by the Roman Varro ; this middle pitch probably comprised besides the final syllables which properly speaking were oxytone all syllables foUoAving next after a high pitch and likewise the second half of a syllable having the circumflex ^ Glaucus of 8amos made the number as many as six: άνβίμίνη [— βαρζΐα), μβση, €7ητ€ταμ€νη (= of em), Κ€κλ.ασμ6.νη (= ττβρίσττωμέιη), άντανακΧωμίνη and a sixth accent, of which not even the name or indeed anything else concerning it is established, except that it belonged to the subdivisions of the circumflex'*. The άντ ανακλώμενη however has its origin in the union of gravis and acute on the same syllable : δίη'ς δας, eav ην, and since the high pitch never occupied more than one mora, appears to have been the 1 SeeiV^A•. J Hfc. 674; confirmation able, that this gravis is often pushed of the ancient writing in the Egyptian so far to the right; but this must not fragment of Alcman. The jDapyri of lead us to suppose that it belongs to the Iliad in London (Pap. Bankes and the last syllable ; for we also find Pap. Harris) have likewise examples IIOAIOXTE (the symbol being over of several accents on the same word: to), and in words with more than one EIIECCETONTO; still both in them gravis AMOIBHAIC, ΔΑΦΟΙΝΕΟΝ. and in the fragment of the Iliad in the - See Varro 1. c. p. ό28 f. Louvre (Pap. 3) the βαρύα is jnincipally '^ Weil-Benloew p. 13 ff. ; Misteli employed to represent the oxytone or Ueber rflu. This as will be seen does not agree Avith the notation given above, but Psichari states that this pronunciation of final ο as ο is very Avidely spread. On the other hand he gives y?^(J as his pronun- ciation of πταίω ; in the case oiflero (πτίρόν) he leaves the e without designation. (h) Accented i is almost always given as closed; when unaccented it appears to fluctuate; in both cases the origin of i (from ι τ; υ etc.) is perfectly indiftiMcnt. With reference to the dialectal pronunciation of ν (ot) as ii noticed above Psichari remarks that the statements of G. Meyi-r {(iirutiiii.'- p. lOS) are very accurate ; M. speaks there of tlif pronunciation as in and gives as examjiles APPENDIX. 137 from the dialect of Attica κιουλια κυιλια, u;^tou/)a α;^υρα, κιουρτος κυρτό?. Psicliari however is inclined to regard this ii in all cases as a modern development after palatals, not as a survival from an older period; τυρί will be found to be in the dialect, Avhcre such phenomena occur, not tiiri l)ut tsiivi witii palatalisation of the t. (c) The transcription πονηρόν honiro militates against the rule we have mentioned above, according to which unaccented ir (ip, ηρ, vp) must become er. I assume that the Chiot thought it necessary to pronounce this word witli its ecclesiastical associations ("the Evil One") in accordance Avith the writing. The apparent retention of the e-sound of η in the dialect of Trapezus is much doubted by Psicliari : reV = τψ, "EAXei'cs etc. might rest on modern phonetic laws; a scientific investigation of the matter has yet to be undertaken. (d) Both ^ασιλε6α and αμαρτία remain free from the detri- tion of L before a vowel following, Avhicli has been referred to above (vasilya, amarti/d). The reason again appears to be, that they are ecclesiastical words, which are not subject to popular treatment. (e) Νάγιαστ^ tmyasti is written by Psicliari with r, though as a rule in such cases the written form contrary to the pronunciation retains the Θ. The rule that two Λ -oiceless spirants, just as tAVO tenues, are not tolerated in immediate proximity, is in general extended to σ also, except that the ordinary pronunciation does not follow this out consistently in the case of σφ. On the other hand φσ is not allowable (except in the artificial pronunciation of the educated) : δουλεΰσω pr. δουλει/ζω, and so always in the interior of words, while in the case of final syllables tu?, i.e. £φς, becomes es : ^ασιλε?, 'Op^e's. There are indeed no words, which terminate with two consonants in the nominative. — ^ΝΌ exception is taken to the collision of voiced spirants (such as β^, ευδ ech). In order to place in a true light the contrast of the old and the new, I add myself a transliteration of the Lord's Prayer, according to the original text, in the Hellenistic pronunciation of that period, Avithout however venturing to denote the quality of the vowels; for the popular pronunciation of the first century A. D. is not known with sufiicient accuracy to render that possible. Only in the case of ot I have given the closed pronunciation of the 0. I denote the 138 APPENDIX. aspirates hy k\ j/, t' {~k + Λ, ρ + h, I + /<); s and ;; are tlie hard and soft ώ -sounds. I give the accents in the ordinary manner, except that I dispense with the grave in the case of monosyllabic Avords. PatCr liCinon 1ιυ on tois Qranois, ha(g)last'ctr> tu ύηδηιδ. (tOnijma,?) sfi, clt'c'tri lie Ijasiloa (biisilia) sii, gcnct'etu \x> t'elC-ma sii hOs δη Qrano kai cpi gcs. Τύη aitun heniun tcin Cplusion dijs liOmin scniCron, kai Tip'os liC'iiiiu ta op'ilcniata Ιιόηιόη (top'ilCniat'C'mun ?), hOs kai hemis (k'C'iiiis?) rip'ckainCn tois op'ilctais Ιιοηιύη, kai me isijnCi'ikC'S liGmas is pirazmun, Γιΐΐϊί rhiisai (rhiisf;?) hOnias apO tii puncru. GREEK INDEX. A/iatooopos (Tanagra) -Ju άγ(γ)/χα, name of uasal y 85, 88 dyeOXa (άεθλα) 110 άγω Locrian for άγω 96 AE BaOt.= ai 56 AE diphth. , its phonetic value and liis- tory 67 ff Af Ef for αυ ev 75 ff AfTTO αύτοΰ Naxian inscr. 76 "Α^ωτοϊ Άσδώδ 118, 1-22 Ά^ήί'α^ε 118 ΑΙ diphth. , phonetic vahic and history 52 ίϊ, 6i ff ; at in verb-endings 65 ; a.L for at Boeot. etc. 45 ; change of at to α before a vowel 5'2 ; at seldom confused with e in script 65 AI diphth. phonetic value and history 43 ff ; dt before vowel becomes ά 52 Αιαίη 66 Aiyeis for Αίγτ/? 47 ; Α^γτ^ί? poet. 47 η. 5 ; became later Aiyis ib. Aϊfγtpα not Ai'γetpα 59 n. 2 at'eros Att., not αετοί 53 η. 1 AIH (AIE) for AH (AE) Ionic 53 αιμωδία not ημ. 70 αλ ελ in Cretan became av iv 80 άλά^εα (Lesb.) 52 AAHON (άλλί'ωι/) 26 ΑΑΙΚΑΡΝΑΤΕΩΝ 120 η. 2 Ά/ατΓΐ^άλτ/ϊ 104 άμιτλακεΐν and άμβλ. 97 άναιραίρημαι ? Thasos 64 f. ac5aj"ai?at (Cret.) 122 dtceo (ευ) 75 &ντρωτΓο$ (Cret.) 113 ά^ω not &νφ 50 η. 1 AO EO Ionic for AT ET 74 ; ao con- tracted to αυ 74 AOT for AT 83 άπεΧέφτεροί 84 Άριστηίδηί -ε'ώη$ 48 Άρίστ-ηχμοί (BcEot.) 27, 57 Άρκεφών 104 αρονρα with real ου 73 Αρρενψδηζ -e'tOijs 48 άσάλεα (Dor.) 52 -acrt (atfft) dat. plur. Att. 45 άτοΰ earoO for αύτοΰ έαντοΰ 79 ff άτσήμί 116 AT diphth. , phonetic value and history 73 ff ; αυ contracted to ω 80 αύάτα, αύεφομεναι, αύ'ω? 78 αύκά (αλκή) 80 αϋρηκτοί (Lesb.) 77 αύσωτοΰ (Dor.) 80 Άφίτρίτα 87 άφτόί {avTos) 73 άχυρα 21, 42 η. 3 Β Β its phonetic value and history 108 ίϊ, in the dialects lllli'; β for f dia- lect 76, 111, 113 ; the same after εν 77 ; β for Latin υ 109 βαδύ 113 /SafftXeia (ace. of βασ^\εύ$) 34 βαστίαί 111 βδεΐι, distortion of Zeu 117 140 GREEK INDEX. βη βη κ;, 27, 1(11) βίβλίον, βΐ'βλίυν 41 11. 1 βλίπαν, βαλΧήΐ'αδί (Arislojili.) 1)7 βοιηΟόν ό3 ΰύιλλαι, 10'.) 11. 3 Βολοεντίοι (ΌλοντίοΟ 76 βραχέα. Ion. fur ^/<αχ(Γα 52 /3i)j-7;f 118 Βΐ'ωτώί» biiut. (nuu) i'ur HoiwrtDi/ 58 11. 2 Γ its i)liuiu'tic Viiluc unci history 108 11; 7 betwein vowels 110 f Γ nasal = ύ 85 11 IVpaffTus 5;5 7^ 71/ proiiiuiciation 88 ί Λ, its pboiic'tic value and history 108 1Ϊ; in the dialects (Elis) 1115 δα- for j'a- (δάσ /itos, δαφοινύ?) Ill) 1' δασύ -γράμμα 1)7 ΔΠΜυΛΙΚΙΐυ [λημοδίκίω) 20 Μ for Λα 18 διαβΐΐπάμ(νοί 70 διαρανϋψ 2() ιι. 2 δινάρια 37 η. 5 Αομέτιο$ 35 δο0λο5 with real ου 72 δυΐΐν 58 Ε E-SOunds, tlieir oldest develupiiieiit and representation 21 ff ; later develop- ment 33 1Ϊ . t for Latin t 35 φδΐν (€ϋδ(11>) 73 iyyovos 120 ίΡοΌχοί 113 ίδυνέατο (Ion.) 54 ί'ζων for ΐστωι> 118 -^T/j Att. noiii. plur. of -cvs 33 EI, real and unreal 25, 21ΗΪ, jplioiuaie value and treatment of real lil 52 IT, ehan},'e in BuOtian 50, later history of et 50 IT ; ei for e (before vowels, before 0•) 34 f, 53, sporadic 35 ; ft for ψ (middle and late Attie, l)i>ric etc.) 47 f. 57; ti and ψ confused in Ceos 2() 11. 2 ; «i Bteot. and Thessal. from η 28, 31, 50; ei for 01 (late lia'otian and late Attic) 57; ei in lloman (Hellenistic) period written for i 10, 60 f; uncertain representa- tion in Latin before vowels 01 -ti 2 pers. midd. for -5 47 tl μάι/ Dor., ei μήν Hellenist, for η μήν 34 ύαντόν (Attic) 34 ίΐσω not ΰσφ 50 η. 1 etVe'a ΥΛτΐοΙοι 03 ελάα 52 ?.\αν for ίΚαων 05 Έλεύΰι/ια Έλ£ΐ;ί^ώ for Ει'λαί^υια 73, La- con. ΈΧΐυθία ΎΛΐυσία 73 η. 7, 111 η. 3 'E/VuuVia 41 ϊμβ\(ύσαντ($ for ΐμβΧέψ. 81 η. 2 tvtLpoaia 34 ίνΰσαν 34 ivvtia 53 f.vT(.\txfLa and ίνδ(\ίχή^ 97 ΐξΐνιχθΐΐ 20 η. 2 Ευ Ιοη. for ΕΤ once in Attic 74 EOT for ET Ion. 74 ; Corinth. 29, 75 ; in lloman period 75, 82 eovTwv 44 ΐπύκΐία 63 έττιστέαται Ion. 54 ϊτΓίστήμη ΐπ-Ίστημη Plato 95 ί'τΓττασίί (Bteot.) 87 ίρδω from ϊρσδω — ΐρξω 119 f ΐρίκη, Έμκΐΐα 03 έσ- ίσσ- Ba'ot. Arc. for ίκ• ι'ξ- 120 f. E^Tf EAlITi: ("AffWcSios) 77 ET phonetic value and history 73 ff; ευ from ι\\) 44 f; eu before a vowel becoming e 80 ; in Cretan ου 80 tilaSe Honi. ίΐ'άλωΛ -e Lesb. 78 t{i<)^ ριγτάσατν Cypr. 77 ί'ύδομο<; (ίβδ. Bteot.) 81 (ΐ'θην (i\Ouv) 80 (ΐ'ξησα (αι'^άΐΌ») (νχούμην (ανχέω) 44 Εΐ'ρνσίλαοί Lesb. 77 (νφήβοισι in Attic epigr. of Κοιηαη lH.'rioil 82 GREEK INDEX. 141 ζ, phonetic value tx (> ΜΑΤΚ1Ρ (Baot. and Tlicss.), ΜΠΤΕΡ (Ceos) 28 μά'^ννμι not μίγ**• *^2 ΜΕΤΑΜΗΡΙΑΧΩΧ 120 η. 2 M^TWKOS (?) for -οικοί 58 η. 5 μ-ηνι-ίιαν i.e. ίίαι* G1 Μο^ριχοϊ (Tanagra) 57 Μουΐ'ίχιώΐ' and Μοι^/'ΐ'χιώΐ' 41 Ν Ν 85f; ν ίφΐΚκνστικΙη• 87 f '^αΡπακτίων 75 Naffl-io (Να^ίου) 115 ioOos (Lcsb.) 78 vi'/ire/JiVtos 39 ιί507; 87 S pronunciation 114 f ; ξ,σ written for f 118 ί,ΐί, not fT G3 ^epos 21 ζουθόί with real on 72 Ο 0- sounds, their oldest development and representation 24 1Ϊ, lat(M• de- velopment 35 f for Latin it 35 Όαλ /uios Άλίδ. ( = fo\.) 7Γ. n. 3 Όαξοί 7G όγδοίτ/ί 53 OB Buot. for υ I 57 Όί^ηνή (Ujja.viui) 125 «j'Oj (branch) 119 ofos (oVi/os Cret.) 122 01 diphthong, its phonetic value and history 51 ff, 70 f ; oi for ωι Boeot. etc. 45 f ; ot confused with ωι 46 ; οι before vowel becomes ο 53 ; confused witli υ 72 η. 1 οϊδα (Lesb.) 55 ΟΙ Η for οη 53 όίκην (oiKtif Lesb.) 55 οικτίρω not οίκτΐ'ιρω 03 6κχοί 103 6\ΐίζον 30 η. 3 όλίοί for όλίγοϊ 110 ΌλυτΓΤΓί'χα 87 ΟΤ real and unreal 24, 29 flf ; phonetic value and extension of real OT 72 ff ; both sorts of ου become ΰ 32 f; ov Thessal. for ordinary Greek ω 28, 41; ου in later Bceot. (Lacon.) written for υ i.e. u 40; ου from ωυ (?) 44 ; from ευ in Cretan 80 ; sporadic for ο 35 ; for Latin ν 109 ούώ? (Bceot.) 51 οφι% 103 Π παΚαστη not παΧαιστ-ή 53 Παοΐ'λλίΐ'α Η;! ■πατρουέαν 45 πίί not 7Γί 03 ΐΓ(Ιν for ττιεΓν 18, 03 TrfTTou'fi 37 η. 5 ■jroetv ποητήί 53 ποΐσαι for ποιησαι 37 η. 5 ttoXtjoj 20 η. 4 ΤΓολτ;! dat. of noKis 48 πολί from ττόλιι 18 Πολίοι'στρατοϊ (Bceot.) 42 Ποσιδίώ;' (month) 59 IIoTf/oato not llor /δ. 03 ΙΤοτίΐδ(δται not -αιαται 54 f. Πραι'χα 74 ττρέσβαα (fa) 34 η. 5 νρηξ,οίσιν 40 η. 3 προσωδία ψιλή 93 η. 1 προτίρω not νροτ^ρφ 50 η. 1 ττΰϊ .51 GREEK INDEX. 143 Ρ pronunciation 89 f 'Papos 91 paOSos, pai'Tos for ράβδοί 81 Σ Σ proiumciation 91 f. ; aspirating power 105, 114 f; σ Lacon. for θ lllf; σ doubled in writiug befoi'e κ τ etc. 91 f. Σαραιηην for -ττιεΐον (JU Σαυκράτείί 74 σδ Lesbian for f 120 Σβοΐ'αστό? 109 Σβου^ροί, Σίΐί^ροί 82 σ^ written for <; 117, 121 η. 5 σημΐία not στ^μαία 70 ZiXtji/os not Σειλ. C3 σίρόϊ not aetpos 63 σιώ 111 Σκνλλη κοί\ψ (Horn.) 71 σκύπφοί 103 σσ, ττ 120 f, 122 Στά'/ιροί not Στάγεφο? G3 στάνομαι [αισθάνομαι] 106 στοά from στοιά 53 Στρατή? (-^as) 27 συντελΐΐται (conjunct.) 24 σωω 46 Τ τ for δ in Egyptian frequent 97 f, 108; Γ for θ after σ in dialects 111, 113; the same in modern Greek 106 ταμΰον for ταμίΐΐον 18, 61 ταώϊ (Attic) 96 Ύεβέριοί, TejSepis 35 τ€Ϊ and τΰ for τφ (late Boeot.) 57 τΐΐρΐΐν 34 τύσω ίτίίσα not τίσω 62 τέτσαρα 115 Τι;'δαρίδαί 40 τιούχα (τί^χτ? Bceot.) 42 τοΓοΰτοϊ 53 ΤΟΤΟΝ (τούτωι/) 32, 73 τούνη (Lacon.) 40 τρίβλιον for τρύβ\. 41 Tpoi'^i'. ancient not Ύροιξ. 53 Ττήι/α Cret. for Ζ^να 123 τιιρί (τυρο'ϊ) 21 Υ τ phonetic value orif^inally 11 39 ff; generally modified at early period to η ib. ; in diphthongs {av m etc ) re- mained )/ 43, 73 ; i; for f 76 ; for r/ in foreign names 41 vyeia, vyt'ia from vyieia (-itia) 18, 61 uyiyatvis, viaivrjs [vyiaivei';) 110 'Ύδάρνη! Vidarna 41 vepyivp {fepywv) 76 TI diphthong, its iDhonctic value and history 41 f, 51 ff i>r vh (Dor. 'whither') 51 'Ύσμήνα^ΐσμήνη 39 'ΤστάστΓτ;; Yistaspa 41 ύύί (vl I's) ancient for vi6s 41, 96 n. 3 Φάβεννοί 111 n. 1 Φαρνάβαζοί 121 Φαώνιο^ 109 η. 3 φΐΚόσοφο'! 103 ΦλεωΟ; not Φ\ιοΰ% 62 Φόλοι/ίοϊ (Fulvius) 109 φρουρό? real or unreal οκ ? 73 Χ Χαιρεληίδτ;; -Χείδη? 48 ΧαΧκ-η Dor. from Χολλ -ea 27 χάρνβδίί άναρροιβδΰ (Hom.) 71 χάφτω etc. (mod. Grk.) 113 χει not χ? 63 Χίρων not \fipuv 59 Ψ Ψ^ pronunciation 114 f i/ziXos meaning 93 η. 1 Ω Ω from α υ 80 ώδε not φδΐ 50 η. 1 ΩΧ diphthong, phonetic value and his- tory 43 ff ; becomes ω before vowel 46, 52 ; confused with ot 46 -ώι ancient for -ώ nom. 3rd decl. 45 'ίΐρομά^η? -μάσδη? 118, 122 ΩΤ phonetic value and history 43 ENGLISH AND LATIN INDEX. A Accent 130 ff ΑΕΙ Lat. for Η V in 57 ; ζ for ts (horz =hortus etc.) 123 n. 2 Papyri, confusion of et and t on 60 parhippus 96 pa-si-le-u-s(e) (βασϊΚΐύί) Cyprian 77 Phaethon dissyllable, 68 Psalterium Verouense 38 Psichari, ,J., on modern Greek pronunc. 13 and passim Ptolemy's son, hermit of Serapeum 81 n. 2 R Ramus, P. on pronunc. 4 Keuchlin, Joh. do. 4 Romance languages, their representa- tion of Lat. oe and ae 71, n. 6 ; ciudad from civitat — cautivo from captivus etc. 81. Pablo 84 10 146 EXGLLSII AXD LATIN IXDEX. Sanskrit, genuine aspirates in 101 scaena, scaeptrum 07 f Schmidt, Erasm. on pronuuc. 4 Septuagint, transliterations in 69 Seyifartb on pronunc. 5 sh- sound, unknown in Greek 92 Smith, Thomas, on pronuuc. 3, 7 Spanish x, early pronunciation 12 f spirants formed from aspirates in modern Greek 101 ; also from medial in ancient Greek dialects and in colloquial language 110 f; spirant for tenuis before tenuis in mod. Grk. 13, 106 spiritus asper 92 ff, in compounds 9(3 f ; dialectically not regarded in elision 109 f; represents σ in Laconiau 96; spiritus + σ for ξ in old Naxian 115 Stephanus, Henr. , on pronunc. 4, 8 Suidas' Lexicon, ei 77 and s in 69 syllables, division into 89, 91 L for aspirate (τίθημι etc.) 104, 113 ; in modern Greek for spirants 106 Thraex, Lat. form of θρψξ 50 tragoedia 50 transference of iinal consonants 129 f U UpsiloD, the name a misusage 20 Latin r, phonetic value 7 DEC i ^ VjB] CIRCULATION DEPT i LD2IA-20m-3.'73 (Q8677sl0)476-A-31 l5Jun59PIVIX General Library University of California Berkeley ^V " M) ■-M-100m-7,'3:i υ C BERKELEY LIBRARIES I Hill I mil CD h^fifiDlfi