28 Charles E. Bennett, 



ev^afeire Coll. 56 which Deecke reads as contracted form 

 for ev^afiere {Bezs. Beitr., vi., p. 148) cannot be regarded as 

 a Cyprian form. 



In the group -€€o- ee does not contract, but the group is 

 simplified by aphaeresis of the first «. Instances of this are 

 ©eo/cXeof for %eoK\keo<i {i.e. %€OK\epeo<i) Coll. 126, I ; TtfjLo- 

 K\eo<i 35 ((/i Arcadian 'BevoK\€o<i for Hevo/cXeeo?, Coll. 1246, 

 B, 12 ; XapcK\eo<i 1 246, B, 4) ; a'jreo';, gen. sing, for *cr7reeo? 

 (i.e. *cr7re-ecr-09) 3 1, 2; 32, 2. Deecke reads arrecof; here, 

 assuming contraction of €o to «, but €o does not elsewhere 

 contract to m in Cyprian (cf. ©eo/cA-eo?, TtfxoK\eo<; ; SraaiKpd- 

 reoq Coll. 18, 2, etc.), nor in the closely related Arcadian 

 (see above). Hence the Cyprian form must be o-ttco?, even 

 though the genitive thereby becomes identical with the 

 nominative. 



The Cyprian accordingly bears out the general principle 

 assumed by Spitzer {Laiit. Ark. Dial., p. 37), viz. that when 

 of three successive vowels the last two are incapable of con- 

 traction, in the particular dialect where they occur, the first 

 of the three disappears. 



In the same connection Spitzer formulates another general 

 principle intended to apply to all Greek dialects. It is this : 

 When of three successive vowels the two latter are capable 

 of contraction, they contract and no further contraction with 

 the first vowel takes place. This principle I believe to be 

 unsafe and to be contradicted by an undoubted illustration 

 taken from the Arcadian itself, viz. hayiiop^o<i. This is gen- 

 erally incorrectly referred to a form 8a/jiio€py6<i. But the 

 second member of the compound as a noincn agentis demands 

 the ablaut of the root, -pop'y- ; cf. Kkoir-c'i 'thief; o-kott-o'; 

 'spy'; TrofiTT-o^ 'attendant' etc. See Meyer Gr. Gr.^, §9. 

 The Homeric jjoems, it is true, exhibit Bi]/jiio6pyG'i r 383 et 

 pass. ; but this is to be regarded as of secondary origin by 

 the side of *ha^iio-op'y6'^. Cf. the similar relation existing 

 between Att. 7revTr)K6vT-opo<i 'fifty-oared galley' and Ionic 

 7r€VT7)K6vT-epo<i, — root ip- 'row'. So Attic inscriptions have 

 rpiaKOVT-epo^i by the side of the earlier TpiaK6vT-opo<i. See 



i=;8 



