184 JjeJcher's Digammated Text of Homer. 



biicher (Ixxxi, 681), Heinrich Rumpf, professes to have done so. He 

 says at least that he has looked at all the 2d aorist forms of the root 

 iS which by the augment begin with et, also at the forms of ii.v6.aab), 

 6.vd&v(a, and &yvvfii, which by the augment begin with rj. The number 

 of these, taken together, would be about 60, and there are 6 of them 

 ('P', 392, /, 305, t, 182, ;t, 373, A, 162, t, 539) which resist an initial di- 

 gamma. The proportion here is not decisive. But it is more import- 

 ant that he finds not a single case which requires digamma, and only 

 one which on Bekker's principles can be regarded as yielding it any 

 particular support. We must conclude then that there is no suffi- 

 cient warrant for Bekker's writing of these forms. 



In this connection I may speak of the form tJikto, a pluperfect mid- 

 dle of the stem i^ or etx. It occurs four times in the Odyssey in the 

 expression, dki^ag d' tjIxto yvvaixl, which does not allow an initial di- 

 gamma. '"HiKTo is most naturally exjilained as being equivalent to 

 eFeFtJ<To, the first e being the augment of the pluperfect, which after 

 the loss of digamma is contracted with the e of the reduplication : 

 epeFtf ro, eei'xTo, TJixro, like svnvaaae, eavocaas, I'lvaaae. Now it is remark- 

 able that the form eixTo, with s instead of rj, occurs once, in ^, 108, — 

 xttl fiot txaar' hniiellsv • eixio de diaxelov ccijrqj. This form is naturally 

 explained as being for vevi^ro, with the reduplication, but without 

 the augment, of the pluperfect. It will be seen that the passage 

 allows, though it does not require, an initial digamma. Bekker writes 

 it without ; in our judgment he should have inserted it : thus, xal fwi, 

 exnar^ inizslls' FfFttro ds OiaxeXov aim. 



If we have complained of Bekker for prefixing digamma to the 

 augmented forms of digammate verbs, we have to complain of him 

 for omitting digamma in some instances from their reduplicated forms. 

 The word just mentioned, in which he writes sfixto, not FtFixTo, is a 

 case in point. Another is seen in ;!r, 348 : 



navTolag svscpvasv • sFOixa di lot naqaslSeiv. 



On first looking at this, I tliought that perhaps the pause (colon) be- 

 fore eoixa might have had something to do with Bekker's retention of 

 the preceding v movable. But I found afterwards a passage (/, 70), 

 which in this respect is exactly similar, but is difierently treated by 

 Bekker : 



dalvv daTia ytQovav • piFoixi toi, ov rot, u^sixig. 



This inconsistency, I suspect, must be the result of inadvertence. In 

 all other cases, so far as I have observed, Bekker writes the perfect 

 and pluperfect active of this verb with digamma where the verse 

 allows it. The number of instances is very large, 125, if I have 



