182 Belcher's Diganxmated Text of Homer. 



ing : with t^to to drcm, there is no such side-form. Of these points 

 of difference, Buttmann does not notice the first two : as to the last, 

 he says with much force that the substantive ^vTijopole of a wagon 

 (drawer) gives j^roof of a verb ^t^w =: iqiw to drato. If we turn from 

 the form to consider the meaning of these verbs, we find something 

 of a step from dravnng to xoatching. Buttmann, however, bridges 

 over the gulf: from drawing to oneself (the proper sense of the mid- 

 dle form) comes the idea of rescuing ; from rescuing that of guard- 

 ing ; from guarding that of watching and even of vmtching against. 

 The development is certainly possible, and if it stopped at the point 

 of rescuitig, we might regard it as probable : but from rescuing to 

 watching against, watchi7ig to injure, there is still a long journey, 

 which we cannot assume without hesitation. But these reasons for 

 separating the words gain almost irresistible confirmation from a cir- 

 cumstance which Buttmann has not noticed, viz., that the indications 

 of the Homeric verse show very clearly that igvco to draw began with 

 a consonant, and almost as clearly that igdojuai to guard began with a 

 vowel. I have collected the Homeric passages which show middle 

 forms of igvu to drav:: I find 60 in all, many of which give strong 

 proof of an initial consonant, while only 3 (i. e., 5 per cent of the 

 whole number) oppose its introduction. Of the deponent ^qiofiai I 

 find 43 instances in all, among which 23, or more than half, resist the 

 introduction of a digamma. Forms which begin with e followed by 

 qq I of course do not reckon, as they obviously belong to giofiai, not 

 ^vofxai. It might be said, however, that some forms in which s is fol- 

 lowed by single g, such as bqvto was guarding, igvaaaro guarded, coiild 

 also be taken from gvofiat, the g being left single after the augment, 

 as often happens in the aorist of ^^^w to do. Assuming this, we shall 

 have in all 29 instances of igiofiai, of which 12, or more than 40 per 

 cent, will resist the introduction of digamma. Again, it might be 

 said that such a form as elgvaaaio is to be explained from sFegvaaato, 

 by omission of digamma and contraction of the vowels, so that we 

 could not expect to see in Elgvaaaxo the digamma which belongs to 

 the verb-root. If we admit the justice of this reasoning, we shall 

 still further reduce the number of instances to be considered, bring- 

 ing them down to 19, of which, however, 9, or nearly half, will still 

 oppose the digamma. It is possible that two or three of the cases 

 which I have regarded as middle forms of igvoi, might be assigned to 

 hgiofiat in the sense of rescuing ; but if we should trajisfer them ac- 

 cordingly, this would not materially aflect the immerical relations 

 just -exhibited. Observe then that in the middle of igiu only one- 

 twentieth of the instances resist the insertion of disrarama, while in 



