Bekker's Digammated Text of Hom&r. 191 



Ahrcns, in Kuhn's Zcitschrift, x, 65 ff., has sought to show that ap- 

 pearances in the Homeric verse similar to those produced by digam- 

 ma are in some instances to be ascribed to a lost sigma. An instance 

 of this kind — in which, however, the initial o is not lost but retained 

 in the written text — is presented by the word 6s hog. The simple 

 form of this word appeare in 55 passages with initial cr, as aO?, avix;, 

 etc. It occurs also in 21 passages where an initial consonant would 

 be incompatible with the metre, and in all these places the a is 

 dropped : we have H, ioq^ etc. In the later language both forms of 

 this word were in use, though the one with consonant initial was 

 comparatively rare. If Avhen the poems were reduced to writing 

 the form with o- had been wholly lost from use, it is probable that 

 our written text would have shown vg only, with initial vowel, in all 

 the 76 passages, though in many of them the metre would have 

 shown traces of a lost consonant. And it is quite conceivable that 

 in other words this may actually have been the case. Such an occur- 

 rence Ahrens recognizes with no little plausibility in the words vlri 

 wood, io; possessive, and i^aawg each. In regard to vlr], Lat. silva., it 

 is certain that it began originally with a, and equally certain that in 

 Homer's language it usually began with a vowel. But there are two 

 cases of a remarkable hiatus before the word (aelexo vlrj^ a, 285, and 

 ine/evato vi\]p, e, 257, where W'/ in each forms the sixth foot), which 

 seem to show that the initial cr was not wholly forgotten in the Ho- 

 meric time. The possessive !«?, in its relation to o?, is explained by 

 Ahrens in a way which has been quite generally received as probable. 

 He assumes in the earlier period two forms, a fuller ctefo?, and a 

 shorter ctfo? : from ae^dg came regularly the ^oj, from crF6g the o?, of 

 our common text. This explanation is supported by the analogy of 

 the possessive forms rsog and aog for the second person ; of which, in 

 all probability, teoj is for t£F(5s, and o6g for rpog. Now in 92 instances 

 of the pronoun 1(5?, there are 52 which do not allow an initial conso- 

 nant, and it is therefore certain that in the Homeric time it generally 

 began with a vowel. But there are 4 instances (v#, 533, /, 420, 687, 

 6, 524) of remarkable hiatus in the first foot, after the first short of a 

 dactyl [Zevg ds e6i^. /sTga Itjj/. ost£ irj;.), which seem to present traces 

 of the primitive initial c. And moreover this word makes hiatus 14' 

 times in the feminine caesura. We have seen that hiatus is readily 

 allowed in that place ; but its relative frequency is so great in the 

 case of this word (three times greater than in the analogous case of 

 ifi6g my) as to warrant the suspicion that it arises from a peculiar 

 cause, and is connected with the primitive initial o-. Bekker writes 

 the pronoun, wherever he can, w ith initial f : he appears to suppose 



