399. The main question which arises on these facts evidently is : Hοw can the great number of passages in which the ϝ affects the meter of Homer be reconciled with the not inconsiderable number of passages in which it is neglected?
The scholars who first became aware of the traces of a lost letter in Homer assumed that in the original form of the poems this letter, or at least the consonantal sound for which it afterwards stood, was consistently used—that it was in fact one of the ordinary sounds of the language—and accordingly they directed their efforts to restoring it to the text. This was the principle on which Bentley made his famous series of emendations: and which was carried out by Bekker in his edition of 1858. Of late years, however, different views of the matter have been taken. Leskien seems to have been the first to maintain that the passages which do not admit ϝ are not necessarily corrupt or spurious, but are to be regarded as evidence of an original fluctuation in the use of the sound. His view is adopted and defended by Curtius (Grandz. p. 550, 5th ed.). Prοf. Hartel has more recently put forward a theory which agrees with that of Curtius in treating the apparent neglect of the ϝ as part of the original condition of the text. But he ascribes this neglect, not to irregularity in the use of the sound, but to the intermediate half-vοwel character of the sound itself.
400. If we are not satisfied that the ϝ had the value of an ordinary consonant at the time when the Homeric poems were produced (or when they received their present form), we may explain the influence which it has on the meter in several ways.
Hypothesis of alternative forms. We may suppose that each word that originally had initial ϝ was known to Homeric times in two forms, an older form with the ϝ—confined perhaps to the archaic or poetical style—and a later in which ϝ was no longer heard. Just as the poet could say either σῦς or ὗς, either πόλις or πτόλις, either τελέσσαι or τελέσαι, so he may have had the choice between ϝἄναξ and ἄναξ, ῾ϝηδύς and ἡδύς, etc.
In order to test the probability of this hypothesis, let us take a few common words of different metrical form, and which show no trace of ϝ, the words
These words, with their immediate derivatives, occur in the Iliad 1022 times, and the places that would not admit an initial consonant number 684, or just two-thirds of the whole. Again, take some of the commonest words with ϝ, ἄναξ, ἄστυ, ἔργον, οἶκος, and the aorist ἰδεῖν. These occur in the Iliad 685 times, and the exceptions are hardly 50, or about one-fourteenth. Compared with the other proportion this surely proves that the recognition of the ϝ in these words was not arbitrary, but was the rule in Homeric verse.