373. The final ι of the dative (locative) singular is so frequently long that it may be regarded as a "doubtful vowel." The examples are especially found in lines and phrases of a fixed or archaic type.
ἦ ῥα, καὶ ἐν δεινῷ σάκεϊ ἔλασʼ ὄβριμον ἔγχος
οὕτω που Διῒ μέλλει ὑπερμενέϊ φίλον εἶναι
(thrice in the Il.)
τὸ τρίτον αὖθʼ ὕδατι
(Od. 10.520, 11.28)
αὐτοῦ πὰρ νηΐ τε μένειν
(Od. 9.194, 10.444)
ἤλυθον εἰκοστῷ ἔτεϊ ἐς κτλ.
(6 times in the Od.)
So in
Αἴαντι δὲ μάλιστα, Oδυσσῆί δὲ μάλιστα, etc.
and the fixed epithet Διῒ φίλος. Considering also that this vowel is rarely elided (§ 376), it becomes highly probable that ῑ as well as ῐ was originally in use.[fn]The priority in this as in so many inferences from Homeric usage belongs (as Hartel notices) to H. L. Ahrens (Philologus, iv. pp. 593 ff.).[/fn]
It is an interesting question whether these traces of -ῑ as the ending of the Homeric dative are to be connected with the occasional -ῑ of the locative in the Veda (Brugmann, Grundr. ii. § 256, p. 610). The Vedic lengthening appears to be one of a group of similar changes of quantity which affect a short final vowel, and which are in their origin rhythmical, since they generally serve to prevent a succession of short syllables (Wackernagel Das Dehnungsgesetz der griechischen Composita, p. 12 ff., quoted by Brugmann l. c.). The same thing may evidently be said of the Homeric -ῑ in many of the cases quoted, as πατέρι, σάκεϊ, ἔτεϊ. Hence it is probable that the lengthening dates from the Indo-European language, and is not due in the first instance to the requirements of the hexameter. But in such a case as Oδυσσῆϊ it may be that the Greek poet treats it as a license, which he takes advantage of in order to avoid the impossible quantities ˘ ˉ ˉ ˘ (cp. ὀϊζῡρώτερος for the unmetrical ὀϊζῡρότερος).