MAY, 1921. AMERICAN MARSUPIAL, C^NOLESTES OSGOOD. 143 



from "properamelid" ancestors and that the characters or tendencies 

 fundamentally distinguishing them from polyprotodonts were a heritage 

 which they had in common with the Australian diprotodonts. 



Increased knowledge frequently, perhaps usually, leads to increased 

 difficulty in defining taxonomic concepts. This is especially true if it 

 be attempted to express both morphogenetic^ and phylogenetic relation- 

 ships in one and the same scheme of nomenclature. The division of 

 modern marsupials into polyprotodonts and diprotodonts was made 

 originally upon a morphological basis and until the discovery of extinct 

 forms, which not only were connectant themselves but drew attention 

 to connectant characters in certain modern forms, the two groups were 

 clearly defined. In the case of the peramelids, a position with the 

 polyprotodonts was generally conceded, notwithstanding their syn- 

 dactyly which was otherwise characteristic of diprotodonts; Wynyardia 

 was left without definite allocation; and Canolestes, as detailed in pre- 

 ceding pages (pp. 6-15), was regarded by some authors as a diproto- 

 dont and by others as a polyprotodont. 



The foregoing study of Canolestes, while adding much to knowledge 

 of fact, does not make classification easier. It is believed that the 

 caenolestids and the peramelids, possibly with Notoryctes also, proceeded 

 from the same stem as the phalangeroid diprotodonts. Hence these 

 might be brigaded as one group. To characterize such a group on the 

 basis of the morphological characters exhibited by modern forms, espe- 

 cially with relation to degree of specialization, however, seems practically 

 impossible. After trying various alternatives and considering the present 

 state of knowledge, no more satisfactory primary division of marsupials 

 offers itself than the old one into Polyprotodontia and Diprotodontia. 

 In such a classification, as already shown, C&nolesies and Wynyardia 

 may be placed with the diprotodonts. The line must be drawn some- 

 where and the least objectionable place for it seems to be between the 

 caenolestids and the modern peramelids. The position of Myrmecoboides 

 must remain doubtful until better material is forthcoming, but its close 

 alliance with Perameles is scarcely to be doubted unless it should prove 

 to have placenta! affinities. The recognition of a superfamily Caenoles- 

 toidea, as in the classification of Osborn and Gregory, serves to divide 

 American and Australian diprotodonts and has its advantages as a 

 matter of convenience. The modern genus Canolestes, however, need not 

 take separate family rank but may be included in the Palaothentidae 

 with the very closely allied extinct forms from the Santa Cruz beds 

 of Patagonia. To rank it as a subfamily, Caenolestinae, would be 

 convenient. 



