72 BULLETIN 169, UNITED STATES NATIONAL MUSEUM 



Matthew (first in 1915) and Granger (1915 and subsequently), 

 however, reviewed all the evidence of Gidley and Broom and also 

 some very important new evidence (chiefly hind limb of Eucosmodon) 

 and reached a conclusion similar to Broom's first opinion, that the 

 multituberculates are not closely related to any other known mammals. 

 I reached this same opinion independently, and I have reviewed all 

 the evidence and added to it in several studies. 



The new evidence from the Fort Union specimens, which I had 

 seen but not studied (except through Gidley's publication) before 

 writing my previous papers on multituberculates, is not very exten- 

 sive. I confirm Broom's opinions that the jugal is probably small in 

 Ptilodus and not entering the glenoid, that there is no evidence for 

 an ahsphenoid bulla, and that there may be an uncoiled cochlea 

 (although I consider this unproved), as well as the point already 

 checked on other material that there is no true angular process. On 

 the other hand Broom's evidence regarding the shoulder girdle was 

 certainly incorrect,^^ and I have elsewhere indicated that the teeth 

 do not support monotreme relationships.^^ The few new details 

 regarding skull structure, foramina, etc., that I have been able 

 to make out show a rather generalized structure with no special 

 characters either of monotremes or marsupials. The humerus, the 

 only known skeletal element not fully considered in my previous analy- 

 sis, seems to me to be neither distinctly therian throughout, at least 

 in a taxonomic sense, as Gidley believed, ^^ nor in its articular portion 

 nearer to the monotreme than to any higher type, as Broom believed. 

 Of the two, it seems to me superficially and adaptively much nearer 

 the Theria, but fundamentally distinct from both. 



In conclusion, the present study confirms and to a slight degree 

 strengthens my former opinion, anticipated by Matthew and Granger 

 and still earlier by Broom but abandoned by the latter, that the mul- 

 tituberculates are a distinctive group not ancestral or closely related 

 to any later mammals and of extremely ancient separation from the 

 main mammalian stock. The very real resemblances to the Theria, 

 pointed out by Gidley, seem to be superficial and adaptive and to 

 indicate analogous stages of evolution, not blood relationship. The 

 likewise real resemblances to the monotremes, pointed out by Broom, 

 seem to be in part adaptive, in part due to the retention in two fairly 

 conservative but not especially related lines of a few very primitive 

 characters, inherited from the mammal-like reptiles. 



25 His conclusion might (but improbably) prove correct, but the evidence is not. The shoulder girdle is 

 known only from one fragment (Djadochiatherium) , not conclusive but rather opposed to Broom's view. 



28 This was the basis of Cope's behef in such relationships, and Broom added it to his table of evidence 

 but placed no great weight on it. 



" Gidley says "eutherian", which was made the subject of a correction by Broom. In fact, Gidley was 

 right, historically, in his use of the term, since Eutheria was originally defined to include both marsupials 

 and placentals, and he was using it in that sense. It is surely less misleading at present, however, to follow 

 the more current usage of Metatheria for the marsupials, Eutheria for the placentals only, and Theria for 

 both together. 



