124 PROCEEDINGS OF THE ACADEMY OF [1894. 



proportional breadth in arctos being .786 and the longest molar .121^ 

 This very wide arctos skull, however, presents the striking feature 

 of having the smallest molar of any examined by me .094 — and, 

 in fact, these measurements show that in this species an inverse pro- 

 portion pretty constantly exists between these characters; the widest 

 skulls have small molars, while the narrower ones have them of large 

 size and grade by a regular series into the proportions of horribilis. 

 Thus, in the eight arctos skulls given above, the four widest give an 

 average length of .1011 to the molar, while the four narrowest give 

 .112 J for the corresponding ratio. In contradistinction to this inverse 

 relation, the exact opposite exists in cinnamomens, — great breadth 

 being here conjoined with great size of molar. 



In some old skulls of arctos a considerable hollowness of the fore- 

 head is found, but in no case equal to that of 3,308 and the Ozark 

 skull. These features of great breadth, frontal concavity and great 

 raassiveness are, of course, among those which increase with age, but 

 it is a significant fact that we find them associated here in skulls of 

 very moderate size. The individuals of cinnamomeus which have 

 come under my notice were very old, three being males and the 

 fourth perhaps female ; they are so nearly equal in size that they 

 may be justly assumed to represent a fair extreme of growth for the 

 species. In 3,308 the skull is 288 mm. in greatest length, while to 

 find anything like an approximation to the same chai-acters in arctos 

 or even to get correspondingly old skulls, it is necessary to go to those 

 from 50 to 80 mm. greater in length. 



Comparing the above proportions, cinnamomeus, while greatly ex- 

 ceeding in both, the average of arctos, is yet surpassed in breadth by 

 one example and in length of molar by one; — arctos has two speci- 

 mens narrower than the broadest horribilis, one of them being the nar- 

 rowest of either species and six have lai-ger molars than the shortest of 

 the American form. The difference in average breadth between arctos 

 and americanus is comparatively small and they overlap greatly, 

 while the largest molar of the latter species is equal to, or larger than 

 four of arctos. 



It is impossible to avoid the conclusion that these proportions are 

 too irregular to be of importance except in groups of averages, and 



3 This skull is not fully ailult and is the shortest examined. With maturity 

 the proportions of this tooth would somewhat decrease. 



