80 



BULLETIN 16 9, UNITED STATES NATIONAL MUSEUM 



Table 7. — Distribution of 23 specimens of Ptilodus montanus and P. mediaevus 

 on the basis of the number of cusps of P^ 



« Where it is doubtful whettier a cuspule should be counted or not, I have counted the specimen as one- 

 half for each of the groups to which it might belong. 



The median for montanus is 6 and for mediaevus 4, and this may be 

 a specific character, but the distributions fully overlap and the 

 samples do not suffice for adequate determination of the significance 

 of this dift'erence. Until larger samples are at hand, this character 

 cannot be used with certainty to distinguish species. 



Passing other characters in more rapid review, the number of cusps 

 of Ml, particularly in the outer row, shows a range of not more than 2 

 in each species, even in the largest samples (which are, however, 

 small, not exceeding 9) and differ markedly from one species to another, 

 probably afi^ording good specific or even generic characters when the 

 means differ by tw^o or more. The external cusps of P^ are highly 

 variable (range 0-3 in montanus) and the data inadequate for proper 

 evaluation, but probably a marked deviation in the medians has 

 specific significance. The inner row varies less in this material (9-10 

 cusps) and probably has specific value (8 in mediaevus), and the width 

 of this tooth is a valuable although variable character, coefficient of 

 variation 8.4±1.7 in montanus. The one available specimen of 

 mediaevus shows a deviation from the means of montanus nearly three 

 times the standard deviation of the latter, almost certainly significant. 

 Other teeth are known in so few cases that their characters cannot be 

 evaluated. 



Family PTILODONTIDAE Simpson, 1927 



Six valid genera of Ptilodontidae have so far been described from 

 the American Paleocene. Their characters are summed up in the 

 following artificial keys: 



