1 88 Bekker's Digamniated Text of Homer. 



sense of desiring, aftei' it had been lost, and that with uniformity, in 

 the sense of sending. That this is a possibility we admit, but it is 

 nothing more. The probability is that the two words are radically 

 distinct; and if so, then for the same reason as before, the f has 

 more in its favor than the y. The fact that later dialects furnish no 

 support to it is of little significance, as the deponent ?£,"«'■ belongs 

 only to the early language. 



The remaining case in which Curtius recognizes initial y as exer- 

 cising the power of a consonant in the Homeric verse, is the relative 

 stem, which appears in 6?, % o, (bj, oloq, oaog, onaig, onolog, oqpp, rifiog, "ya, 

 ea>g, etc. Savelsberg, in the eighth volume of Kuhn's Zeitschrift, has 

 given a series of citations from Homer, showing traces of an initial 

 consonant for this class of words. Unfortunately he has not fur- 

 nished exact numerical data, by which we might see the comparative 

 frequency or infrequency of the phenomenon. I have not myself had 

 time to sujDply the deficiency. I have only run hastily through the 

 first book of the Iliad, noting all words which show the relative stem 

 — omitting, however, the relative adverb wij, which in this respect 

 stands by itself The whole number of instances noted was 72, of 

 which 30, or about 42 per cent, testify against a consonant initial. 

 Of the remaining instances, many were found at the beginning of a 

 line, or in other indecisive positions ; and in fact there are only 8, or 

 one-ninth of the whole number, which give any indication of a con- 

 sonant initial. Even of these, the majority, either from the part of 

 the verse which they occupy, or from the pause which precedes them, 

 are of but little weight : only 2 or 3 give decided indications of an 

 initial consonant. I strongly suspect that a more extended compari- 

 son would not essentially change the proportions derived from this 

 first book. They seem barely sufiicient to give plausibility to the 

 conjecture that the relative stem did once begin with a consonant, 

 but had nearly or quite lost it in the Homeric time. I say, " nearly 

 or quite .•" for if the letter had wholly died out from common use 

 shortly before Homer's time, the force of epic tradition would proba- 

 bly have caused some traces of it still to appear in his verse. But 

 the adverb w; diifers in this respect very remarkably from the other 

 forms of the relative stem. According to Bekker, as cited by Cur- 

 tius, the instances which indicate a consonant initial are three times 

 more numerous than those which indicate a vowel initial. It can 

 hardly be doubted that this word, as pronounced in the Homeric 

 time, began frequently, if not generally, with a consonant. 



It must be owned that our condition as regards the etymology of 

 .the Greek relative, is an unsatisfactory one. We are less confident and 



