01' THE AUSTEALIAX .AIAKSUPiALIA. 209 



ment of syndactylous modifications of the pes, and Caluromys shows signs of the 

 ])arallel development of omnivorous characters in the molars as in the latter family. 

 These resemhlances are not indicative of affinity except in the hroader sense that 

 they possihly imply a similar potential of evolution carried over from common 

 ancestors. 



The case of the South -American Miocene and Australian groups is admittedly difficult 

 of discussion on account of the lack of definite information as to the pi'imary differential 

 characters of the former. The dental resemblances of Tlnjlaci)uis to the Spai'assodonta 

 are sufficiently close to warrant a belief in their common origin. Even so, however, 

 there is no evidence that the evolution of Thylacinus has been connected in any way 

 with that of the Dasyviridse, the reverse being indicated by the fact that the various 

 forms of the Dasyuridae present successive stages of a dental evolution, which, ^vhile 

 carnivorous in its character as in Thylacinus, is otherwise of a totally different facies. 



The fact that the diprotodont modification of the dentition occurs in the Soutli- 

 American as well as in the Australian forms has been regarded as a mark of affinity. 

 As indicated above, even the most primitive forms of the existing Phalangpridae are 

 well removed as regards dentition from the condition which miist have obtained at the 

 time of introduction of the diprotodont modification. Furthermore, the origin of the 

 peculiarly grooved, sectorials of the Bettongiinae is not illustrated in the Australian series. 

 In view of tlie presence of such gaps in the series, we could not wish to deny the possi- 

 bility that the prototypes of these forms may yet be found in the South-American group. 

 The mere fact of the presence of diprotodont modifications in each aeries, however, 

 means nothing. The development of a diprotodont modification has taken place in the 

 Soricidte, in the Rodentia, and in the Multituberculata. The bunodont molars of tlie 

 Epanorthidse resemble those of the Phalangeridte, but such teeth have been developed 

 in the Condylarthra and the Primates. If convergent development may occur between 

 groups which have little in common beyond their mammalian character, how much 

 greater is the chance of convergent development in smaller groujjs of common parentage ? 

 If resemblances between the Australian and the South-American groups are to be 

 recognized as indicative of affinity they must not be of a broad general kind such as those 

 already pointed out between the Epanorthidae and their allies and the Australian dij)ro- 

 todonts, but must be of much more special application. The forms presenting them 

 should not differ in a greater degi'ee from one another than do the successively specialized 

 genera of the AustraUan families. For all we are able to say at jjresent, the South- 

 American radiation may have proceeded on general lines of polyprotodonty and 

 diprotodonty or on some other uni-elated character. 



If the above interpretations are correct, how are the Marsupials to be classified ? 

 The farther the identification of intermediate types proceeds, the more difficult becomes 

 the systematic classification to which we are accustomed. We are now placed in the 

 somewhat paradoxical position of attempting to recognize and characterize concrete 

 groups which our knowledge of evolutionaiy sequence tells us could have had no 

 separate existence. If natural classification recognizes lines of descent, major classifica- 

 tion must recognize radiations, and distinctions of time and geographical distribution 



