contained only one vowel-sound apiece; these were: a, n, w, vɩ, nu, wv. And leaving out at and ot, which, as we shall see, had assumed a weaker pronunciation, they arranged the other ten combinations as follows (Anecd. Bekk. 1214, 804; Theodos. p. 34; Moschop. Titz. p. 24):

ευ

Kaтà κρâσw (i.e. when the two sounds were really mixed): ov, av, KAT' ETIKρáтELAV (i.e. when the former vowel predominates): eu, ą, ?, w Kaтà diéĝodov (i.e. when both vowels are heard in succession): nu, vi, wv.

These distinctions are quite inapplicable to the sounds of the letters in the best ages of the Greek language. The long vowels η and o involve such an effort in the pronunciation that they necessarily absorb any following sound; this is well expressed by the grammarian's definition : δεῖ τὸ ἢ μὲν ἐκφωνοῦντα μηκύνειν τὸ στόμα ὡς ἐπὶ τὰ ὦτα ἑκατέρωθεν, τὸ δὲ ῶ ἐκφωνοῦντα μηκύνειν τὰ χείλη ὡς ẻπì tỷv pîva kaì тòv πwywva (Anecd. Bekk. p. 797). The value of η, in accordance with this description, is established by the passages from Cratinus (Etym. M. 196, 7):

ὁ δ ̓ ἠλίθιος ὥσπερ πρόβατον βῆ βῆ λέγων βαδίζει :

and Aristophanes (Anecd. Bekk. p. 86):

θύειν με μέλλει καὶ κελεύει βῆ λέγειν.

Cf. Hesych. s. v. Bî Xéyer Bλnxâтaι ǹ Oveтaι (Meineke, Fragm. Com. Ant. p. 40).

In Doric ʼn is represented by a, and in Boeotic by &; and in heavier forms η passes into ω ; thus, from πατήρ we have ἀπάτωρ, from φρήν, σώφρων; and we have the heavier words, ἀσπιδιώτης, Ιταλιώτης, στρατιώτης by the side of ὁπλιήτης, πολιήτης. When these long vowels are prefixed to v, it can hardly be said that the result is a diphthong: on the contrary the accentuation of Ocupa for Oatua shows that in wu the v was not sounded, and aov is contracted into w, and not into wv. And with regard to nu the habit of omitting the temporal augment in verbs beginning with ev, and the pronunciation of such words as us, lead us to the conclusion that when 7 appears as the augment of verbs beginning with av, it was either written rather than pronounced; or, if pronounced, that the augment was syllabic as well as temporal. At any rate, when ā, ŋ, w are prefixed to, the latter is so completely absorbed that it is not generally written in the same line, but becomes

what is called Iota subscriptum, or "Iota underwritten;" thus: q, ?, w. This iota, however, is placed on the same line when it follows a capital ; thus: "ΑΙΔΗΣ="Αιδης = ᾅδης, Ἤιθεος =ᾔθεος, Ωιδή

=

on. In the combination v one or other of the letters has assumed or resumed a consonantal value, unless the is merely absorbed as in the cases just mentioned. It is laid down by the grammarians (Eustath. p. 1047, 53; Etym. M. 72, 26; 775, 24; Choroboscus, apud Anecd. Bekk. pp. 1220, 1292) that this combination is never found before a consonant. Hence we have oπvíw, oπúow; and the drops out in the optative-endings of verbs in -vu, and in diminutives like βοτρύδιον, ἀφύδιον, ὕδιον, ἰχθύδιον, whereas a is inserted in viáow. That the combination was not dissyllabic is proved by the accentuation of αἴθυια, ἅρπυια. The accentuation of participles like TeTUQUîa shows that the retains its value as a vowel. And a careful review of all the usages leads to the conclusion that, when followed by a vowel, the v in this combination was pronounced like our w, and that, when followed by a consonant, the v was vocalized and the absorbed and omitted. This conclusion is fortified by the change of cui-rei into cur, and of the diphthong ot, pronounced wee, into v in ἐμύ for ἐμοί, πρύλεες for προίλες, &c. How the Romans pronounced their quadrisyllable Ilithyia, from Eixelovia, we do not know: but the analogy of fio and fui would lead to the conclusion that yi became in effect a long i.

し

23 In general, it is not easy to recover the true pronunciation of the diphthongs, whether proper or improper. It is probable, that e and ev were articulated nearly as in English, but that av, ol, ou were sounded like the German au, ö, ou, or the English ow, wi, oo in how, wig, choose. A dog in Aristophanes (Vesp. 903) barks aù, au; the bird's note p'wee, as in pewit, is written Toi (Aves, 227); and our natural exclamation ugh is expressed by the common interjection ioù. Considerable difficulties, however, are occasioned (a) by the manner in which the Greeks themselves speak of the pronunciation of some of their vowels and diphthongs; (b) by the Roman transcriptions; and (c) by the modern Greek articulation: and we may infer that changes took place at different epochs.

(a) Plato says, Cratylus, p. 393 D: άσTTEρ Tŵν σTOIXEίwv oiola ὅτι ὀνόματα λέγομεν, ἀλλ ̓ οὐκ αὐτὰ τὰ στοιχεῖα πλὴν τεττάρων, τοῦ

Ε καὶ τοῦ ῦ καὶ τοῦ ὁ καὶ τοῦ ὦ, from which it is clear that the vowels e, v, o, o must have had fixed sounds, which would justify their designations à ψιλόν, &c.; whereas ἄλφα, ἦτα, and ἰῶτα were expressed by names rather than by definite utterances, because their sounds were variable. In Aristophanes, Nub. 872, Pheidippides is ridiculed for his broad pronunciation of ai:

ἰδοὺ κρέμαι, ὡς ἠλίθιον ἐφθέγξατο

καὶ τοῖσι χείλεσιν διεῤῥνηκόσιν.

from which it is clear that Attic taste had begun to drop the full articulation of both vowels in the diphthong. That a and or made single sounds is also shown by their liability to elision, and by their being counted short for purposes of accentuation; the Alexandrian poet Callimachus actually makes exe the response of an echo to vaixi (Epigr. 30); and Sextus Empiricus, who flourished under Hadrian, says that at, et, and où, were not diphthongs, because they had but one sound from beginning to end (adv. Gramm. 1. 1. c. 5, p. 241). From all this it appears that there was a gradual loss of the full pronunciation of the combined vowels.

(b) The same results from the Roman transcriptions; for at and o are generally written ae and oe, as in 'A¤ñvai, Athenæ, BowTía, Boeotia. The improper diphthong @ is sometimes expressed by o, sometimes by e, as in paywdía, rhapsodia, éπwdós, epodus, but Tpaywdós, tragedus; ou is sometimes shortened into i, as in Inomaus, cimeterium, from Оivóμaos, кoιμητýpɩov; et is sometimes i and sometimes e; thus we have Evğeivos, Euxinus, and generally before consonants, except in Polycletus, Helotes, Cuperus; Aiveías, Æneas, Movσetov, Museum. It is to be observed, however, that all these transcriptions belong to a time when the Romans had lost their own pure diphthongs; for aule was once even aulāï.

η,

(c) The modern Greek pronunciation confuses between oɩ, et, n, and, pronounces at like a in male, and gives the value of v to the second vowel in eu, ou, nu. That Reuchlin should have adopted this articulation from the learned Greeks, who taught him their language, was natural enough, but it is certain that no such confusion prevailed among the ancient Greeks, who could not have had such words as κεκέλευσμαι, πεπαίδευνται if u had been a consonant, and would never have relinquished the power of distinguishing between the root syllables in πείθω, πέποιθα, and ἐπίθησα:

and that a never sank from æ, however that was pronounced, to a mere French ē, which was the value of 7, is clear from the fact, that even the latest Romans gave at the value of a, as in Plutarch's πрaípekтоs for præfectus, and always represented ʼn by e, as in the same writer's transcriptions: carere, κарnрe, majores, païúpns, sapiens, σαπίηνs, Rhenus, Ρῆνος.

η

24 On the employment of a and e to represent vowels, the breathings were expressed by the second or first half of H written over the vowel affected by it: this notation, which is due to Aristophanes of Byzantium, has been revived by the Oxford Press; but generally the smooth breathing (spiritus lenis, πveûμа iλóv) is marked'; and the rough breathing (spiritus asper, πveûμa daσú) is marked. If the word commences with a diphthong, the breathing is placed over the second vowel; not so when the second vowel is Iota subscriptum, or absorbed: compare éy, olos with "Aions and adw. When a word or syllable begins with p the rough breathing is always placed over that letter: when p is doubled, the first p takes the smooth breathing according to a general principle, which will be explained in its proper place (97): compare pýτwp, rhetor with Iúppos, Pyrrhus.

Obs. 1 Certain German editors, following Bekker and Dindorf, print the double p without any breathing'. That this is erroneous is shown not only by the Latin transcription, but also by the express statement of Arcadius (πeρì πvevμáтwv, 200, 21, ap. Valcken. Ammon. p. 242). It is held by one scholar (Lobeck, Paralip. p. 14) that the accents of eppupai ought to be expressed by pépiμai when the perfect assumes that form, and the rule for the transference of the breathing seems to justify this. Some of the ancient grammarians retained the breathing of the p even in the middle of a word, making it lenis or asper, according to circumstances; thus they wrote káπpos, 'Arpeús, but Χρόνος, ἀφρός, θρόνος (Anecd. Bekk. p. 693, 20).

Obs. 2 It is observed that the Eolians especially omitted the aspirate, whence this wors, as it is called, is regarded as Æolic wherever it occurs. This was carried so far that even words beginning with v or p were marked with the lenis ; thus we find ῢμμε, ύμμιν, ὔμοιος, ὖρχα, σκλος, ἔμφαλος, and Ρᾶρος with its derivatives; Arcadius, p. 242 Valcken.; Anecd. Bekk. p. 693, 11; Herodian, Tepi μový povs λéέews, p. 35, 6. The Æolians are also said to have written pp in the middle of a word (Anecd. Bekk. u. s.; Greg. Corinth. p. 588). Those who adopt the old-fashioned theory that the Latin language is connected with the Eolic dialect of the Greek, will find in this wors a special

1 In adopting this orthography, in the Cambridge text of Thucydides, we have sacrificed our own opinion to the necessity of making the series uniform.

difficulty, for the Latin, as distinguished from the Greek, retains the original sibilant of which even the aspirate is a weaker form: compare ¿§, sex, πтá, septem, &c.

Obs. 3 On the other hand, the Attic dialect, in its later forms, seems to have had a tendency to aspiration: thus we find authority for άδην ; ἁθρόος and ἁθροίζω in the orators; άθυρμα ; άμμος, whence καθαμμίζω in Aristotle; ανω and αὑαίνω to explain ἀφαναίνω in Aristophanes ; cipyw, "to shut in," as distinguished from epyw, "to shut out" (Lobeck, Αj. p. 338); ἔνη; ανύω ; ἡθμός and Αίσωπος on the Sigæan inscription ; άλλω to explain φιάλλω and ἐφιάλλω; even ιχθύς (Gellius, Ν.Α. II. 3); and 'Αβδηρῖται, ἐλπίς, ισθμός in Attic inscriptions. It is also remarked that the Athenians said raws instead of raws (Athen. p. 397 F).

Obs. 4 In the Ionic dialect the aspirate is sometimes represented by the lengthening of the syllable which bore it; thus we have οὐδός=οδός; οὖλος-όλος; οὖρος=δρος; ὤριστος=ὁ ἄριστος; ἄλλοι=οἱ ἄλλοι; ikμevos and iέados from ikw, perhaps because the first syllable in these words involved originally a reduplication; aμvdis for apa, because the whole word is lengthened, or because in all forms of Greek å for σa had a tendency to drop its breathing; cf. ἄκοιτις, ἀδελφός, ἀκόλουθος, &c. even in Attic; and ἠλέκτρον, αὖλαξ, &c., from ἕλκω, not only lengthen the first syllable, but introduce an articulation vowel after the liquid.

§ IV. Origin and Arrangement of the Letters.

25 Mythology attributes to the Phoenician hero, Cadmus, the introduction into Greece of an original alphabet of sixteen letters; and the old grammarians have supposed that these sixteen were the following:-a, B, v, d, e, i, K, X, μ, V, 0, π, P, σ, T, v (Schol. Dion. Thr. p. 781). There can be little doubt that the Greek alphabet is of Semitic origin, and there is every reason to believe that it originally consisted of four quaternions of letters: but it is a sound theory, which has been confirmed by the independent investigations of at least four or five different scholars, that for ɩ, k, p, and v in the above list we must substitute n, 0, and the two obsolete characters F (Ba) and Q (KóπTа), which are still retained as numerical signs after e and π respectively, and that the original arrangement of these sixteen letters was as follows:

Α. ΒΓΔ. Ε. FΗΘ. ΛΜΝ. Σ. Ο. Π Ω Τ.

This order is artificial and systematic; as we shall see, if we consider the original value of these characters. For A, E, and O were originally the representatives of breathings of which A was the