150 Walter Petersen, 



masculine because of the long ultima, the change took place in 

 the reverse direction. It must be borne in mind too that only a 

 few cases were affected : the nom. sing, [xstpaxicmoc, the ace. (jxtpo*- 

 xicntov, and the nom. pi. [j,£ipaxi(7xot, while the long ultima of all the 

 other cases, e. g. the gen. sing. p-stpaxiT/.ou, caused the masc. and 

 fem. to agree from the beginning. Similarl}' apxrJ}.OG could have 

 followed *apxTuXY), etc. After the masculine had followed the fem- 

 inine, the neuter in -icrxov in turn followed the mascuHne. For 

 -(crxo- this tendency might also have been helped by primitives in 

 -iq, e. g. xo7ci(7XO(; after xotoi;, though it does not seem that this 

 could have been of great importance. 



15. The question whether the old accentuation of -ktxo- has left 

 any traces in the Greek language may be answered in the nega- 

 tive. The only possible case of an appellative with the accent on 

 either the ultima or propenultima is )(e>.i,(7xov (Hipp. ap. Erotian. 

 p. 137 Klein), but the fact that the text is suspected also for other 

 reasons should not make us place too much faith in it. On the 

 other hand it would certainly be conceivable that a word which 

 had lost completely all association with any primitive (it designated 

 a kind of pot, see § 41 A) should have resisted the tendency to 

 change the accent to the penult, since it was by all means not 

 felt as a diminutive. Other examples of irregular accentuation of 

 the suffix are all non-Greek proper names, which therefore do not 

 concern us. I might mention the Thracian river 'ApTt(7x6?,^ the 

 Dacian river TijiiTxoc, the Dacian town Ttpicrxov, the Parthian town 

 <I>t}a(Txov. 



16. The gender of words in -t(7xo-, which may be either mascu- 

 hne, feminine, or neuter, like that of the I. E. -ko- and -lo- di- 

 minutives, followed the gender of their primitives ^ : 6 6^tk\.(TMc : 

 6^zk6<;, b vsavLtrxo? : 6 vsavtac, yj TCapQ-svicTtv) ; yj xapO-svo?, y] xopiirxTj : 

 T, xopY), Y) oixtcTXYj : ■/) oixia, TO [Kzkirry.ov : xb ix-ilon;, to (7a[jL(3a).{ffxov : to 

 (7a[j.(3a>.ov. When, liowever, we count the extant words in the suffix 

 we find that the mascuHnes are overwhelmingly predominant,^ the 

 feminines are much less frequent, and the neuters are exceedingly 

 rare. In round numbers there are found considerably over two 

 hundred masculines, about forty feminines, and only about a dozen 

 neuters. 



17. This preponderance of the masculine is not due to any extended 

 tendency to form masc. derivatives in -tcjxo? from fem. and neut. 



^ Also in the form 'AQiriaxcg. 



^ Cf. Brugmann, op. cit. 669 f. 



3 Cf. Schwabe, De Dem. Graec. et Lat. 50. 



