1910.] Relation.^ of Creodonts to Insectivores. 305 



sectivores may be conceded to be, on the whole, more aberrantly specialized 

 than the majority of Eocene Creodonts. Admitting the partly secondary 

 character of the elongation of the premaxillaries in typical Insectivores^ we 

 still have left the more important question: which is more primitive, the 

 prototypal Insectivore family as reconstructed above (p. 2S7) or the rela- 

 tively primitive Creodont type, such as Sinopa'^ To put the cpiestion in 

 another way, were the Insectivores derived from terrestrial forms with short 

 premaxillaries, transversely arranged incisors and large canines, or were 

 the Creodonts derived from small semiarboreal insectivorous forms analo- 

 gous to Marmosa, i. e., with slightly elongate premaxillaries, opposite in- 

 cisors arranged in converging series, rather small canines. 



The Insectivora as an order possess two characters which have been 

 shown by cumulative evidence to distinguish the ancestors of many other 

 groups such as the Marsupials (p. 289), Rodents (p. 331), Ungulates, Pri- 

 mates (p. 321) and others, namely small size and insectivorous-omnivorous 

 habits. 



The Creodonta themselves seem to run back into small, probably insec- 

 tivorous-carnivorous forms. Certainly the very large forms Harpagolestes, 

 Pterodon and Patriofelis, representing three of the leading families, are far 

 more specialized in their dentition (figured e. g., by Osborn, 1907, pp. 131- 

 135) than their small relatives Dissacus, Sinopa and Oxifcena in which the 

 traces of former trituberculy are much clearer. These in turn are all more 

 liighly specialized than the small Basal Eocene members of the Oxyclsenida?. 

 The latter family includes forms {Tricentes, Chriacus) the molars of which 

 approach the Insectivore type illustrated in Leptictis and Pantolestes so that 

 their position in the Creodonts is somewhat doubtful (Matthew). The 

 very small Lower Eocene Pala^osinopa, which strongly resembles Didel- 

 phodus, a primitive member of the Hypenodontidfe, is related to Pantolestes, 

 a true Insectivore (Matthew). The latter, although an undoubted Insecti- 

 vore, approaches the Creodonts in its large size, heavy muzzle, long con- 

 stricted postorbital region, Dissacus-Y\he molars, etc. (Matthew, 1909). In 

 Palceosinopa and Pantolestes the canines are larger than in most Insectivores 

 but the incisors are arranged more antero-posteriorly than transversely 

 and the premaxillary is slightly elongate. 



The earlier Creodonts also retain a more or less divergent poUex and 

 hallux which, as remarked by Matthew (1904, p. 813-814), is strong evi- 

 dence of arboreal origin. The oldest Creodont skull known, that of Triiso- 

 don heilprinianus Cope (figured by Matthew, 1901, pp. 30-31) resembles 

 that of the lowest Insectivora in the possession of a long cylindrical mid- 

 cranial region and very small brain case. 



On their part the Insectivora are allied to the Carnivora by many prim- 



