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



(2) the indicated straightness of the jaw anterior to the canine, (3) the 

 premolariform canine, (4) the compressed premolars, (5) the absence of 

 cingula, (6) the relatively high inner cusps (metaconid and entoconid). 

 He also points out that the extinct form differs from Myrmecobius in at 

 least three important particulars (i) the possession of the normal num- 

 ber of post-canine teeth, (2) the approximation of the paraconid to the 

 metaconid, (3) the more elevated and better defined trigonid. 



Comparison of Gidley's figures and description with several species 

 of Peramelidae shows at once that practically all the above enumerated 

 points of resemblance are to be found in that family and none of the 

 important points of difference. This is indicated by the following 

 parallel: 



Myrmecoboides and Perameles Myrmecobius 



Canine equal to or lower than pre- Canine higher than premolars. 

 molars. 



Protoconid and hypoconid well Protoconid and hypoconid small or 



developed. rudimentary. 



Paraconid and metaconid approxi- Paraconid and metaconid not approxi- 

 mated, mated. 



Trigonid elevated and distinct. Trigonid not differentiated. 



Post-canine teeth seven. Post-canine teeth more than seven. 



The important points of difference between Myrmecoboides and 

 Perameles are (a) the unicuspid anterior premolar in the extinct form, 

 (b) the large last molar with its well-developed hypoconulid, (c) the 

 exceptionally large paraconid of the first molar, and (d) the somewhat 

 more brachyodont condition of the molar series. The unicuspid pre- 

 molar is a specialization such as might be expected in an ancestral 

 diprotodont but the other characters are primitive and doubtless hark 

 back to a tritubercular stage. The resemblances to Perameles are 

 pronounced, however, and cannot be overlooked, although the classifica- 

 tion of such an ancient form on fragmentary material has much possi- 

 bility of error. In view of the many similarities to Perameles shown by 

 C&nolestes, the occurrence in the early American Tertiary of a form with 

 distinct leanings to Perameles becomes especially significant. The 

 hypothesis can scarcely be avoided that, provided Myrmecoboides is a 

 marsupial, it may well be ancestral to the caenolestids. A point of espe- 

 cial interest and significance is the unicuspid premolar which is paralleled 

 among marsupials only in caenolestids and diprotodonts. 



PHYLOGENY AND TAXONOMY. 



The great range of variation in structure and adaptive characters 

 shown by existing marsupials makes it possible without reference to 

 extinct forms to construct linear series in which the sequence from 



