392 Bulletin American Museum of. Natural History. [Yo\. XXVIl, 



trihedral character by the outgrowth of its external ridge and the flattened 

 spatulate character of its postero-superior glutfieal surface. In Phenacodus 

 this external ridge was already produced into a long sickle-shaped surface. 

 In the ancestral Perissodactyl the ridge was probably not pointed at the 

 tip but slightly expanded and flattened while the internal superior border 

 of the ilium w^as produced inward and upward above the level of the sacrum. 

 These two ridges, the internal and the external, caused the ilium in obli(iue 

 postero-superior view to be Y-shaped, the antero-superior border being 

 concave. This type of ilium is preserved in the primitive cursorial types 

 Eokippns and Hijrachyus but is lost in the very heavy slow moving types. 

 The pubo-ischiadic part of the pelvis was somewhat smaller than in Phena- 

 codus and much shorter and less stout than in the Mesonychidje and Artio- 

 dactyla. 



Fem,uT. The shaft was slightly flattened and retained a moderate 

 third trochanter, which was perhaps placed higher up on the shaft than in 

 Phenacodus. 



The manus and pes of the ancestral Perissodactyl had already advanced 

 beyond the Condylarth stage. In the manus the pollex was much reduced, 

 in the pes the hallux disappeared very early and the fifth digit w^as greatly 

 reduced. The functional digital formula was thus 4-3, as it is in all Lower 

 Eocene Perissodactyls. The middle digit was the longest, especially in the 

 pes, the proportions of the several digits were approximately as in Tapirus 

 but the manus was very likely narrower. The manus in all known Lower 

 Eocene Perissodactyls was very narrow, with a rather high carpus and there 

 is good reason to believe that this is an inheritance from unknown small sized 

 cursorial ancestors. No truly cursorial Creodont or Condylarth known has 

 a broad manus. The broad manus of later Titanotheres, Palaotheres, 

 I>ophiodonts, Rhinoceroses, etc., is very probably a progressive adaptation to 

 increasing weight. 



The Perissodactyl character of the feet may be more directly derived 

 from the Condylarth than from the Creodont condition. In none of the 

 Creodonts does the tendency to tridactylism become nearly as pronounced 

 as it does in Phenacodus, where both in manus and pes the digits are sym- 

 metrically arranged on either side of the third. Phenacodus itself, of course, 

 is not ancestral to the Perissodactyls and may be more Perissodactyl-like 

 than its own ancestors; but the development of perissodactylism in both 

 orders argues a similar type of foot in the ancestors of each, a type of foot 

 which is not reached in the more remotely ancestral Insectivore-Creodont 

 stock. The perissodactylism of the Litopterns ofi'ers no obstacle to this 

 reasoning, for they themselves were probably also derived from Condy- 

 larthra (p. 448). 



