46 Chai'les E. Bennett, 



Karl, has been assumed by Deecke, as the full form of the 

 elided /car' 'and' in Kar ^}^haXiwv Coll. 59, i. If this is 

 correct, the form would belong with '^n and a(y)Ti. At all 

 events we are not justified in assuming that the form /ca? 

 'and' originated from Kan by the latter's becoming *Kda-L, 

 whence (before vowels) /ca?. 



So also TTo? Coll. 60, 19, 20, 21 is not to be explained as 

 the ante-vocalic form of ^ttoo-l (for ttoti), since ttotl so far 

 as known never assibilates its t. The s of 7ro9 must be 

 explained in some other way ; see § 33, 3. iror, which 

 Deecke reads in Coll. 68, i, by elision for ttotl, is perfectly 

 consistent with the existence of tto? in Cyprian (see § 33, 3), 

 but the context is so doubtful that small probability attaches 

 to this form. 



2. The indefinite 0-^9 (for tU) occurs Coll. 60, 10, 23 ; 

 and 67ri(n<i 60, 29. This is irregular, since initial t before i is 

 not assibilated ; yet the form is certain. Possibly, tU as an 

 enclitic, was so closely connected with the preceding word as 

 to be felt as a part of it. In this way the t became intervo- 

 calic and so changed to a-. This is the explanation of Meyer, 

 Gr. Gr?, § 299, and in support of it may be cited Att. a-rra, 

 which developed from the primitive nom. pi. neuter of rt?, 

 viz. rrd, in such phrases as -^pij/iiard rta. The two words in 

 such instances were so closely connected as to be treated 

 like one. Hence '^p/j/jbard ria became regularly xP^'lf^^'^"''^'^'^- 

 This was felt as ■^(^ptj/jbaT arra, so that drra arose as an inde- 

 pendent word. The only objection that can be urged against 

 this explanation of aU is that Hesychius gives us crt as 

 an interrogative pronoun in the gloss crt jSoXe • rl 6eXeL<i. 

 KuTr/Jiot. 



Deecke's rl in Coll. 68, 3, is to be rejected. The reading 

 is uncertain, and the form highly improbable by the side 

 of crU. 



3. Deecke reads ttotl Coll. 68, 3, as vocative of tt6tl<;, 

 ' lord.' The word occurs, however, in 26 as Troo-i?, with 

 regular assibilation of the t. The fact that we always find 

 TTo'crt? in other dialects would certainly tend under any cir- 



176 



