142 FIELD MUSEUM OF NATURAL HISTORY ZOOLOGY, VOL. XIV. 



and giving rise through another line to the Dasyuridae and modern 

 polyprotodonts. Interesting in this connection is the fact noted by 

 Bensley (1903, p. in) that the modern peramelids agree with the 

 Oligocene Peraiherium more closely in the character of the important 

 styles of the molars than they do with the Australian dasyurids. 



The fossil record is very incomplete but its testimony accords in a 

 surprising degree with the hypothetical requirements of Bensley's 

 scheme. Myrmecoboides might without serious objection be looked upon 

 as in the direct line between the " Properamelids " and the diprotodonts, 

 since it has begun to reduce its premolars, the first being small, unicuspid, 

 and single-rooted as in many diprotodonts. Furthermore, Canolestes 

 shows numerous resemblances to Perameks, so many that a common 

 ancestry subsequent to the generalized polyprotodont stage is by no 

 means impossible. The material representing Myrmecoboides is too 

 imperfect to warrant more than a provisional conclusion that it may 

 be an early stage in the line of the caenolestids. Nevertheless the evi- 

 dences of affinity between Myrmecoboides, Perameles, and Canolestes 

 cannot be overlooked. Moreover, if Myrmecoboides is in truth a repre- 

 sentative of the early differentiation of a diprotodont stem then it is 

 the first form of the kind to be discovered outside the southern continents 

 and goes far toward explaining the present day occurrence of diprotodont 

 types in both Australia and South America. 



Further points of interest in connection with the resemblances of 

 Ccenolestes to the peramelids are found in the slight indications of com- 

 mon relationship to Notoryctes. This peculiar form is so highly spe- 

 cialized that its affinities are very problematical, but Bensley is inclined 

 to regard it as an offshoot of the "properamelids." Certain features of 

 the myology (see p. 60) and osteology of Canolestes and Notoryctes are 

 similar and favor belief in a common (properamelid) derivation rather 

 than otherwise. The actual course of evolution must be inferred to a 

 considerable extent even under the best of circumstances. The view 

 that Canolestes is a primitive diprotodont is not proved but is strongly 

 supported by its resemblance to the peramelids which of all polyproto- 

 donts are the ones most suggestive of the incipient stages leading from 

 one large group to the other. It is probably not too much to say that if 

 the caenolestids had been discovered in Australia instead of South 

 America they would have been accepted without question as ancestral 

 diprotodonts. This geographical difficulty is at least partly overcome 

 by the presence of Myrmecoboides ', also with peramelid affinities, in 

 North America. Hence it is a hypothesis of considerable probability 

 that the caenolestids, like the Australian diprotodonts, developed 



