MAY, 1921. AMERICAN MARSUPIAL, C^ENOLESTES OSGOOD. 141 



generalized to specialized types is remarkably free from interruptions. 

 If their present distribution also be disregarded, that is, if both American 

 and Australian forms be included, the problem is simplified. As worked 

 out in great detail by Bensley (1903), the American Didelphiidae stand 

 in a prototypal morphogenetic relation to the -remaining groups of 

 marsupials. This conclusion is based upon their possession of a com- 

 bination of various generalized features which are distributed over at 

 least three different Australian groups. They have the primitive 

 tritubercular molar as in the Dasyuridae, the upper incisor formula and 

 upper molar stylar characters as in the Peramelidae, and the pedimanous 

 foot-structure (including incipient syndactyly in one form) as in the 

 Phalangeridae. 



As the probable ancestor of the modern Didelphiidae, Bensley finds 

 the Oligocene Peratherium fulfilling practically all theoretical require- 

 ments including a distribution in both Europe and North America 

 which might permit its dispersal either to South America or Australia or 

 to both. This dispersal he supposes to have taken place in at least three 

 radiations, the first being represented by the Australian fauna, the 

 second by the Miocene fauna of South America and the third by the 

 present day Didelphiidae of South America, this last being in its incipi- 

 ency. The evolutionary series so thoroughly expounded by Bensley is 

 an exceedingly plausible one in spite of its having been based largely 

 upon existing forms. The various modern polyprotodonts are obviously 

 reducible to a common type and, although the diprotodonts offer more 

 difficulty, they may at least be traced from one to another. Between the 

 two groups is a rather definite hiatus but there is little room for doubt 

 that diprotodonts proceeded from polyprotodont ancestors. Indications 

 of some of the steps leading from one group to the other are seen prin- 

 cipally in the Peramelidae, the Wynyardiidae, the Myrmecoboididae and 

 the Palaeothentidae. No one of these aberrant groups is free from 

 objection as an ideal transitional type, yet taken together they serve to 

 bridge quite thoroughly the gap between the generalized polyprotodonts 

 and the more primitive diprotodonts. Bensley dealt chiefly with modern 

 Australian forms, giving relatively little attention to Wynyardia or 

 C&nolestes, and Myrmecoboides was unknown to him. But the Pera- 

 melidae he regarded as proceeding from a hypothetical stem-group called 

 Properamelidae, one branch of which led to modern peramelids and 

 another to the diprotodonts. This group presumably had begun to 

 acquire subquadrate upper molars with incipient hypocones as in 

 modern peramelids but had not developed a terrestrial foot structure. 

 The Properamelidae themselves were supposed to have evolved from 

 the generalized polyprotodont having the characters of the Didelphiidae 



