Cope.] ^"0 [Oct. 7, 



with Elotheriiim and Poebrotherium and Oreodon, belong to Oligocene 

 beds. 



Tubercular or bunodont molars are of prior age to selenodont molars, 

 phylogenetically speaking. Of the former, the tritubercular type, it has 

 already been shown, is ancestral to the quadritubercular type. Panto- 

 lestidse are then clearly ancestral to all known Artiodactyla, and are 

 themselves probably the descendants of the lost Amblypoda Hyodonta, 

 whose existence I have anticipated on hypothetical grounds. Of the 

 remaining families which are constructed on the quadritubercular basis, 

 there are two types, as represented in divisions II and III of the pre- 

 ceding table. The intermediate or fifth lobe is especially characteristic 

 of Eocene Artiodactyla. The intermediate tubercles exist in the Pan- 

 tolestidaj, and one of them is preserved in the families of division II ; 

 but in group A it is the posterior one, and in group A A it is the 

 anterior one. In the Suidse and Elotheriidae, which are permanently 

 bunodont, the intermediates are either lost or so divided as to lose 

 their distinctive character. In Elotherium traces of both the interme- 

 diates are visible, but they are obscure. The genetic relations of the 

 families with five lobes to those with four are supposed by Schlosser to be 

 direct and ancestral. This looks probable in the case of the Merycopo- 

 tamida: of the latter group, which has inferior molars like those of Hyopota- 

 mus of the former group. Whether the remaining families of division 

 III AAA (see table) (four-lobed) came off from the families of division II 

 (five-lobed) is uncertain. It is probable that the fifth and si::th (or inter- 

 mediate) tubercles were present in all primitive Artiodactyla, but they 

 may have been lost, as in the Suida3, in the bunodont stage, wliich gave 

 origin to III AAA, so as to be wanting from the earliest four lobed seleno- 

 dont ancestors. Of the two types of II, the division A (Dichobunoidea) 

 is supposed by Schlosser to have been the ancestor of the true selenodonts 

 (III AAA), but excepting in the case of Merycopotamidge this has not yet 

 been demonstrated. Scott suspects with reason that the quinquetubercu- 

 lar Protoreodon is the ancestor of the quadritubercular Oreodon. 



Leaving this debatable question, I refer to the family of the Anoplo- 

 theriidae. The remarkable structure of the feet discovered by Gervais, and 

 shown by Schlosser to belong to this family, distinguishes it at once from 

 all families of this and all other orders. 



The second digit is well developed in both feet, and stands inwards 

 at a strong angle to the other toes. A rudimental fifth is present in 

 the manus, but not in the pes. The latter is therefore tridactyle. 

 The third and fourth digits are equal in the pes, but the third exceeds 

 the fourth in the manus, giving an entirely perissodactyle character. 

 Some didactyle forms have been placed in this family, but this is inadmis- 

 sible on ordinary taxonomic principles. The divergent inner toe is sup- 

 posed to have supported a web, useful in an aquatic life. As remarked by 

 Schlosser, the origin of the Anoplotheridse is entirely obscure as yet, the 

 only ancestor yet known being the Pantolestidae. It is probable that some 



