MAY, 1921. AMERICAN MARSUPIAL, C^ENOLESTES OSGOOD. 129 



and so far as that alone is concerned there could be no possible objection 

 to considering C&nolestes directly prototypal to the Australian diproto- 

 donts. It may be said, however, that the ancestor of diprotodonts by 

 merely a priori reasoning would at some point have the general charac- 

 ters of C&nolestes, that is, it would have a diprotodont dentition while 

 retaining a number of polyprotodont characters. This will be discussed 

 elsewhere, but in the series of diprotodont dentitions we can at least 

 place that of Canokstes as the first and most generalized among living 

 forms. 



The uncertainty lies in the steps preceding the development of the 

 caenolestid type of dentition, for the gradations between generalized 

 polyprotodont dentitions and the caenolestid stage of diprotodonty are 

 not recognizable with certainty among known living or extinct forms. 

 Hence the most constructive stages in the development of diprotodonty 

 in this case must be supplied theoretically. Bensley, whose opinion is 

 most important, states that "the diprotodont modification, although 

 characteristic of the herbivorous section of the Marsupials, is the result 

 of an insectivorous adaptation which must have been developed in the 

 minute ancestors of the Phalangeridae during the incipient stages of the 

 omnivorous evolution, but after the separation of the peramelid stem." 

 He therefore finds the beginnings of diprotodont tendencies in small 

 polyprotodonts such as Phascologale and Peramys which have undiffer- 

 entiated lower incisors but show a slight differentiation of the median 

 upper ones. Since primitive diprotodonts like Dromicia have upper 

 incisors of this same character he concludes that such incisors are part of 

 the course toward diprotodonty. With enlargement of the median 

 lower incisors in a form like Phascologale 1 he would expect a shortening 

 of the lower jaw and consequent disturbance and reduction of the 

 "intermediate" antemolar teeth. These conditions he finds exemplified 

 in Dromicia and C&nolestes. This explanation by Bensley may be the 

 true one, for it is difficult to gainsay; but if applied to Casnolestes it 

 practically assumes a direct didelphid ancestor and therefore needs 

 careful examination. Before granting that diprotodonty arose ex- 

 clusively in this manner, it may be inquired whether there is any other 

 way in which it might have originated among marsupials. The didel- 

 phids and caenolestids were- already well distinguished in the Patagonian 

 Miocene. Hence the origin of the diprotodonty of Canolestes is still 

 farther back, perhaps in the Mesozoic, and it is certain that the ancestral 

 form was not the present day didelphid but an ancient generalized type, 

 all the characters of which are not preserved in any recent forms. 



1 A much greater development of the anterior incisors is found in Oxygomphius 

 frequens, a European Miocene form usually referred to the Didelphiidae. 



