338 Wortman — Studies of Eocene Mammalia in the 



more typical structure of these teeth, but the evidence is very 

 strong in favor of their derivation from ancestors in which the 

 carnassials were well developed.* In the Creodonta, on the 

 other hand, if carnassial teeth are developed they are not con- 

 fined to the fourth superior premolar and the first inferior 

 molar, but usually consist of a varying number of molars in 

 each series. In some of the older types there are no carnassials 

 developed, the molars being intermediate in structure ; while in 

 others they are of a pronounced tubercular form. The three 

 suborders would then be divided and defined as follows : 



Suborder Creodonta. 



Carnassial teeth present or absent ; when present, not consisting 

 exclusively of fourth superior premolar and first inferior molar. 

 Scaphoid, lunar, and centrale of the carpus very generally free. 

 Ungual phalanges broad, depressed and fissured, or laterally com- 

 pressed and pointed. The following families are included : 

 Oxyclaenidae, Arctocyonidae, Mesonychidae, Oxyaenidae, and Hyaeno- 

 dontidae. 



Suborder Carnassidentia. 



Carnassial teeth present and always consisting of the fourth 

 superior premolar and first inferior molar. Scaphoid, lunar, and 

 centrale of the carpus, very generally united. Ungual phalanges 

 compressed and pointed. 



The following families are included : Viverravidae Viverridae, 

 Hyaenidae, Protelidae, Palaeonictidse, Felidae, Canidae, Procyonidae, 

 Ursidae, and Mustelidae. 



Suborder Pinnipedia. 



Limbs modified for progression in the water ; no carnassials ; 

 scaphoid, lunar, and centrale united. Ungual phalanges greatly 

 modified by enormous development of subungual processes. 



Families : Otariidae, Trichechidae, Phocidae. 



Suborder Carnassidentia. 

 Family Canidm. 

 The study of the Eocene Canidae is attended with no little 

 difficulty, owing in large measure to the insufficient and frag- 

 mentary materials upon which the types of the respective 

 genera have been based. While fortunately these are not 



* Matthew has (loc. cit., p. 17) quite recently discussed the relationship of the 

 Arctocyonidse, a family of the Creodonta from the Torrejon and Wasatch of this 

 country and Europe, to the modern Bears. Arguing from the structure of the 

 feet and teeth, he believes that they make certain distinctive approaches towards 

 the Ursidse and may have been ancestral to them. Had he taken the trouble to 

 compare the feet of Clc&nodon with a living Opossum or a Dasyure, he would have 

 found such a striking similarity of structure in every detail, with the possible 

 exception of the astragalus, that he would have concluded that the Arctocyonidae 

 are much nearer to the Marsupials in these characters than to the Bears. 



