397. The traces of these sounds in the meter of Homer are chiefly of interest for the purpose of comparison with the facts relating to ϝ.
The effects of initial σ may be seen in a few cases of the non-elision of prepositions
ἐπι-άλμενος
(Latin salio)
ἀμφί-αλος
(Latin sal)
ἀμφί-επον
(Latin sequor)
κατα-ΐσχεται
(ἴσχω for σίσχω)
and the lengthening in πᾱρέχῃ (Od. 19.113) and σῡνεχές (Od. 9.74). Hiatus is also found twice before ὕλη (Il. 14.285, Od. 5.257, once before ὕπνος (Od. 10.68), and 18 times before ἑός (mostly in the principal caesura). These instances however are too few to prove anything.
Initial ι̯ or y is chiefly traced in the adverb ὡς, which when used after the noun to which it refers is allowed to lengthen the final syllable, as θεὸς ὥς, ὄρνιθες ὥς, etc. (so in 36 places). On the other hand there are nearly as many places which do not admit an initial consonant, as κτίλος ὥς (Il. 3.196), λέονθʼ ὥς (Il. 11.383, 12.293, 16.756), θεὸς δʼ ὡς κτλ. Probably therefore no spirant was heard, and the lengthening of the syllable before ὥς was a mere "survival" or traditional rule (§ 375.1).