404. It seems desirable here to say something of the uses of the digamma which are found on the older inscriptions of the chief Doric and Aeοlic dialects. The forms preserved on these inscriptions do not indeed prove anything directly as to the Homeric digamma. We cannot infer from them, for instance, that the symbol ϝ was ever used in any written copies of the poems, or that the sound which it represented in other dialects was known to the Homeric language. But they may serve by way of analogy to direct our conjectures on these questions.
The most striking examples of ϝ are found on the inscriptions of Corinth and its colony Corcyra (as ϝεκάβα, ϝιόλαϝος, ϝίφιτος, Δϝεινίας, Aἴϝας, Ξένϝων, Ξενϝάρεος, ὅρϝος, Τλασίαϝο, etc.). With these may be placed the Argive inscriptions (in one of which occurs Διϝί), and the few Laconian inscriptions. In the older monuments of these dialects initial ϝ is never wanting; but omission in the body of the word is occasionally found, as in Δαΐφοβος and Πολυξένα (on the same Corinthian vase), and several names ending in -κλῆς (for -κλέϝης), and -λας (for -λαϝος). The scanty Phocian inscriptions yield the important forms ϝέξ, αἰϝεί, κλέϝος, with no early examples of omission; and the little known Pamphylian dialect is equally constant, so far as it has been made out. The Locrian dialect shows more decided indications of falling off in the use of the digamma. On the inscriptions of that dialect (discussed by Prof. Allen in Curt. Stud. iii. 207 ff.) we find it in
- ϝαστός
- ϝέκαστος
- ϝεκών
- ϝέτος
- ϝεσπάριος
ϝοῖκος and its compounds (ἐπίϝοικος, etc.), also in καταιϝεί, ϝεϝαδηκότα, but not in δαμιωργός, ξένος, ἐννέα, Ὀπώντιος (for original Οποϝέντιος). The only initial ϝ which is wanting is in the wοrd ἱστίαι (we may compare the Laconian and Homeric ἐφέστιος). Similarly in the older Elean inscriptions initial ϝ is regular (ϝάργον, ϝέπος, ϝράτρα, etc); and we have also Ερϝαοῖοι (people of Herea?), ἐϝέρεν (probably an infinitive), but ξένος, Διός without ϝ. In the great inscription of Gortyn initial ϝ appears in
- ϝός
(suus)
- ϝίν
(= ῾ϝοῖ)
- ϝέκαστος
- ϝεκάτερος
- ϝέρξαι
- ϝεργασία
- ϝῆμα
(εἷμα)
- ϝεῖπαι
- ϝοικεύς
- ϝοῖνος
- ϝίκατι
- ϝεξήκοντα
and is only lost in ὠνά, ὠνάω (before ω, § 393). Theϝ is also fond in compounds, as
ἐνϝοικῇ
προϝειπάτω
δυοδεκαϝετίες
and in the body of the word ϝισϝόμοιρος, but disappears between vowels, as in λάω (genitive of λᾶος a stone), αἰεί, παιδίον, the oblique cases of nouns in -υς and -ευς (υἱέες, ϝοἰκέα, δρομέες, etc.), and the contracted words ἄτα (ἀϝάτη) and ἇς (for ἇϝος, = ἕως). It is also lost before ρ, as in ἀπορρηθέντι.[fn]Baunack, Die Inschrift von Gortyn, p. 37-39, 68.[/fn]
A somewhat later stage in the use οf ϝ is well exemplified by the numerous Boeotian inscriptions. In these the general rule is that initial ϝ is retained; the only word from which it is regularly absent is ἕκαστος. On the other hand the only instances of ϝ in the body of a word are the compound ϝικατιϝέτιες (εἰκοσιετέες), and a group of derivatives of ἀείδω (αὐλαϝυδός, τραγαϝυδός, etc.). The same rule applies to the Arcadian inscriptions, which however are too few to be of importance. The further progress of decay may be seen in the Doric dialect of Heraclea, of which a specimen remains in the well known Tabulae Heracleenses (of the 4th cent.). We there find
- ϝέξ
- ϝέτος
- ϝίδιος
- ϝίκατι
and the compound ἐγ-ϝηληθίωντι (= ἐξ-ειληθῶσι), but
- ἕκαστος
- ἶσος
- ἀφ-ερξόντι
- πενταἑτηρίς
- ἐργάζομαι
- οἰκία
- ῥήτρα
from which it follows that the use of ϝ even as an initial sound must have been fluctuating. A similar condition of at least partial loss of ϝ is found in inscriptions of Melos.
If we do not confine οur view to the character ϝ, but look to the other indications of the sound which it represented, the most important evidence is that furnished by the Cyprian inscriptions. The forms which they yield belong, generally speaking, to an earlier period of the language than is known from alphabetical inscriptions. Yet the use of the sounds answering to ϝ is not uniform; we have Διϝός and Διός, βασιλέϝος and βασιλέος.
An original ϝ is represented by β in several parts of Greece, especially Laconia, Elis, Crete, but probably the β is merely a graphical substitute for ϝ. It is found in the inscriptions of later times, when β was probably = our v.
The substitution of υ for ϝ is characteristic of the Aeolic of Lesbos, as εὔιδε (for ἔϝιδε), αὐως, δεύομαι, ἐνδευής (= ἐνδεής). In these forms the ϝ is vocalized; cp. Homeric αὐίαχος (= ἀ-ϝίαχος), εὔαδε, ταλαύρινος.
It is necessary here to notice a group of uses of the ϝ in which it seems to have been developed from a neighboring vowel (υ or ο). The vowel usually precedes, as in Laconian ἐδήδοϝας, ἐδήδοϝε, Corcyrean ἀριστεύϝοντα, Boeotian Eὐϝαρα, βακευϝαι, Cyprian Eὐϝέλθων, Eὐϝαγόρω, κατεσκεύϝασε; but we also find Τλασίαϝο (Cοrcyr.), Γίλγαϝος (Cypr.), Τιμοκάριϝος (Cypr.), ϝότι (Locr). So perhaps the Boeotian αὐλαϝυδός, τραγαϝυδός, etc., (see above). With the former instances we might compare Italian Genŏνα, Padŏνα (for Genua, Padua) ; with the latter the u of Italian uomo, uopo, the w of whole, the provincial English wuts for οats, etc. With ϝότι we should compare the form Nαϝπάκτιος, also Locrian. Both are exceptional and indeed must be considered as mere errors,[fn]The ordinary form Ναύπακτοs occurs on the inscription 19 times, the form with Ναϝ- only once. Similarly against the single instance of ϝότι are to be set 2 instances of ὅτι, and 5 others of the relative ὅς, in the older Locrian inscription. See Allen in Curt. Stud. iii. p. 252; Brugmann, ibid. iv. p. 133, n. 57: Tudeer, De digammο, p. 45.[/fn]: but they help to show how near ϝ was to a pure vowel sound. It is evident that this redundant ϝ, growing out of the vowel υ or ο, is a parallel phenomenon to the loss of ϝ before these vowels which was noticed above as a characteristic of Homer (§ 393).
405. Ϝ in Ionic. There remains the interesting question whether the existence of the ϝ in Ionic can be traced in inscriptions. The evidence appears to be as follows (Tudeer, De digammo etc. pp. 5 ff.)
- The form AϜYΤO (= αὐτοῦ) on a Naxian inscription of the end of the 6th century B. C. But, as has been pointed out[fn]By Brugmann, Curt. Stud. iv. p. 132, n. 55, and Tudeer, p. 7.[/fn] the ϝ of ἀϝυτός indicates at most a special way οf pronouncing the υ, and is to be compared with the erroneous Νάϝπακτος noticed above.
- The name of the city οf Velia, which was founded by exiles from Phocaea (ϝέλεα marshes; but see § 393).
- The forms ϜlO, ΓAPΥϜONEΣ, OϜAΤIEΣ—all proper names—on vases found in Magna Graecia, and supposed to have come from Chalcis in Euboea, or one of its Italian colonies.
It is inferred by Tudeer (l. c.) that the ϝ must have been a living sound in the Ionic dialect of Euboea at the time when the colonies of Chalcis were sent to Magna Graecia, i. e. probably in the 8th century B. C. On the other hand, since there is no example on the inscriptions of Euboea itself, the sound does not seem to have survived there down to the date of the earliest examples of writing, viz. the 6th century B. C. Hence Tudeer puts the loss of the ϝ in Ionic Euboea at some time between the 8th and the 6th centuries.
It has been recently pointed out by P. Κretschmer (K. Z. xxxi. 285) that the Ionic change of ᾱ to η cannot be placed very early. The name Μῆδοι underwent the change—the original ᾱ appears in the form Mᾶδοι on the monument of Idalion—and the Medes must therefore have become known to the Ionians before it was completed. The Persian names which reached Ionia later—Δᾱρεῖος, Μιθριδάτης, etc.—retain their ᾱ. Similarly the old Carian Mίλατος became the Ionic Μίλητος. Hence the Ionic η is later than the contact of Ionians with the nations of Asia Minor. Nοw the anomalous η after ρ in the Attic κόρη and δέρη is to be explained from the older forms κόρϝη, δέρϝη (cp. κόρρη from κόρση). Consequently the loss οf ϝ in Attic must be later than the change of ᾱ to η, and a fοrtiοri later than the Ionian migration. This inference is confirmed by the ο of the comparatives κενότερος and στενότερος, pointing as it does to the forms κενϝός, στενϝός (since the lengthening of the ε, as in Ionic κεινός, στεινός, never took place in Attic).
The former use of ϝ as a letter in all Greek alphabets is shown by its use as a numeral, and also by the existence of the first non-Phoenician letter, Υ, The addition of Υ, which was the earliest made, and perhaps contemporaneous with the introduction of the alphabet, shows that the Greeks felt the need of a vowel distinct from the labial spirant Vau. Otherwise the Phoenician Vau would have served for the vowel υ, just as the Yod was taken for the vowel ι. And as there is no Greek alphabet without Υ, it follows that the consonant ϝ was equally universal.[fn] As the Vau is written ? on the Moabite Stone, it has been suggested that it was the source of the Greek Υ. It seems not improbable that the letters Ϝ and Υ were at first only two forms of Vau, appropriated in course of time to the consonant and vowel υ—just as our u and v come from the two uses of Latin V. If this is so, the place of Υ at the end of the then alphabet is significant, as showing the importance attached to the original order of the letters. See Roberts, Greek Epigraphy, § 11: Taylor, The Alphabet, ii. p. 82.[/fn]
Combining these inferences with the independent evidence furnished by the meter, we may arrive at some approximate conclusions regarding the value of ϝ in the Ionic of Homer.
a. Initial ϝ had the value of a consonant, except before ο or ω (§ 393).
b. δϝ was retained, not only at the beginning of a word (§ 394), but also in ἔδϝεισα, δέδϝια, etc.; we can hardly suppose compensatory lengthening in these forms.
c. between vowels is more doubtful (§ 396). Since initial ϝ was lost as early as Homer before ο or ω, it probably vanished before most case endings of the 2nd declension, and before the -ος, -ων of the 3rd declension. Thus for λαϝός, etc., we should have λαός, λαοῦ, etc., (but ϝ possibly in λαϝοί, λαϝοῖσι), and again ἡδύς, ἡδέος, ἡδέϝι, etc., Πηλεύς, Πηλῆος, Πηλῆϝι, etc. Then other cases might fοllοw the analogy of the genitive singular and plural, and so drop the ϝ altogether. However this may be, it is clear that ϝ between vowels was generally lost much earlier than ϝ at the beginning of the word (cp. Italian amai for amaνi, etc.). The absence of contraction proves little, as we see from the Attic χέω, ἔχεα, ἔχεε, etc. At the same time we occasionally find a partial survival of ϝ in a vοcalised form, making a diphthong with the preceding vowel (§ 396).
Note— A parallel to the Naxian ΑϜΥΤΟ has now been found in the form ΑϜΥΤΑΡ on an Attic inscription of the 6th cent. BCE (see J. van Leeuwen, Mnemos, xix.21). Further instances of Chalcidian ϝ (ϝοικέων, σαϝοῖ?) are given by Roberts, Epigraphy, p. 204.)