MAY, 1921. AMERICAN MARSUPIAL, C^NOLESTES OSGOOD. 145 



The accompanying diagram (p. 144), like others of its kind, has many 

 shortcomings and some inconsistencies. It is offered only provisionally 

 as an aid in indicating the relative position of Ccenolestes. Various prob- 

 lems in the phylogeny of marsupials must somehow be met in such a 

 diagram even though they seem properly beyond the scope of the 

 present study. So far as any conclusions are indicated, they are to be 

 regarded mainly as suggestive rather than decisive, as indeed they would 

 be in any case. Myrmecobius is separated from the main polyprotodont 

 stem and from the dasyurids in order to indicate its possible direct 

 descent from mesozoic types, but a position nearer the dasyurids is 

 almost equally probable. 1 The problem of the relationship of Propoly- 

 mastodon has not been studied and its uncertain position is merely 

 suggested (see Gregory, 1910, p. 212). The same is true of Pediomys 

 and Didelphops, while the multituberculates (Allotheria), which Broom 

 contends are monotremes rather than marsupials, are indicated as 

 probably proceeding from the unknown forms indefinitely termed 

 "generalized marsupials" and in reality meaning very little. 



DISPERSAL OF MARSUPIALS. 



Broadly speaking there are only two theories of the dispersal of 

 marsupials, one that the group originated in Holarctica and spread 

 southward, the other that it had its beginnings in Antarctica, the 

 hypothetical southern continent which formerly may have connected 

 part of present-day South America and Australia. Advancing knowl- 

 edge has from time to time favored one or the other hypothesis but 

 opinions are still divided. A review of the subject with many references 

 is given by Osborn (1910, pp. 64-80). 



Before the discovery of the extensive extinct fauna of Patagonia, 

 the distribution of marsupials offered no serious objection to a theory of 

 northern origin for the group. Practically the only fossils known were 

 didelphids from the Upper Miocene and Oligocene of Europe and North 

 America, and the conclusion was fairly obvious that these might repre- 

 sent the ancestral stock from which both Australia and South America 

 received their marsupial populations. The best known advocate of this 

 theory was Wallace. A reference to his work on the Geographical Dis- 

 tribution of Animals (1876), however, shows at once that the information 



1 Although this disposition of Myrmecobius is somewhat heretical, I am led to it 

 through the cumulative effect of encountering in this study and in the published 

 work of others the continual necessity of special theory or explanation to account 

 for first one and then another peculiarity of this animal. Much of this would be 

 obviated by carrying its origin farther back. The possibilities for longer lines of 

 descent than usually assumed is evidenced by Canolestes itself in its persistence 

 from Miocene times to the present in practically unchanged condition. 



