S7^ MB. OLDFTELD THOMAS ON A NEW [DeC. 17, 



difficult to believe that the beds are quite so early as Senor 

 Ameghino supposes. 



Any Uugering doubt which may have existed among Naturalists 

 as to the correctness of Ameghino's reference of the Epanorthidte 

 to the Marsupials (and doubt has been thrown on it) is wholly 

 removed by the study of Gcemlesies, which is typically Marsupial 

 in every character. 



As to the general classification of the IMarsupials, a subject 

 already sufficiently difficult in view of the puzzling possession by the 

 Pemmelidce of polyprotodonty combined with syndactyly, Ccenolestes 

 apparently only adds to the difficulty, being non-syndactylous Uke 

 most Polyprotodonts, while it has by dentition nothing to do with 

 them. If anything, however, this fact tends to confirm the 

 tentative opinion expressed in the ' Catalogue of Marsupials,' 

 p. 220, that the primary division of the order should be by dentition, 

 and that syndactyly is a secondary character. Were syndactyly 

 the primary character, the Epanorthidce would be thrown with the 

 Dasijuridce and Didelphyidce, ^\-ith which they clearly have nothing 

 whatever to do, and separated from what appear to be their 

 nearest allies, the Fhalangeridce. 



If this A-iew be correct, the Marsupials as a whole might be 

 divided as foUows : — 



Order MAESUPIALIA. 



I. Suborder Dipeotodonta. 



A. Non-syndactylous. — American. 



1. Epanorthidse. 



B. Syndactylous. — Australian. 



2. Phalangeridse. 



3. Phascolomyidae. 



4. Macropodidse. 



n. Suborder Poltpeotodonta. 



A. Syndactylous. — Australian. 



5. Peramehdse. 



B. Non-syndactylous. — American and Australian. 



6. Didelphjidse. 



7. Dasyui'idae. 



8. Notoryctidae. 



It is, however, possible that, in spite of the resemblance of the 

 teeth of Cainolestes to those of certain Australian Diprotodonts, 

 the study of further material, including soft parts, skeleton, and 

 milk-teeth, will bring out differences of such importance as to 

 necessitate its subordinal separation from them. In this case the 

 name suggested by Aineghlno, Paucituberculata, will be available 

 for the suborder containing Ccenolestes and its fossil allies. 



Even in that case, however, in view of their many resemblances, 

 it does not seem possible that anything will show that there is no 



