FORT UNION OF CRAZY MOUNTAIN FIELD, MONT. 



77 



lower teeth they must be left incertae sedis wdth respect to them. It 

 was, however, possible to place all the lower jaws with P4, Mi, or both 

 in one of these species. 



It was next necessary to compare these eight species with those 

 previously named and described. Ptilodus mediaevus and troves- 

 sartianus, the onl^^ named approximately contemporary species, 

 demanded closest comparison. The probably later species Edypodus 

 musculus, E. cochranensis, Paredypodus simpsoni, and P. tardus were 

 also compared. No others are sufficiently close to warrant detailed 

 comparison. 



At this point the attribute of provenience becomes of essential 

 importance. As an example, the number of serrations on P4 of the 

 montanus group from the Fort Union, counted on 29 individuals, 

 varies from 13 to 15, the mean being 13.8 and the median 14. In the 

 six available individuals of comparable size (mediaevus group) from 

 the Torrejon, five have 12 and one has 13 serrations, mean 12.2, 

 median 12. If all these be considered as one sample the distribution 

 is: 



12 serrations — 5. 14 serrations — 19. 



13 serrations — 9. 15 serrations — 2. 



The distribution shows no bimodaiity, and it might be assumed that 

 the sample is homogeneous with respect to this character, but this is 

 fallacious, as the sample is not, in fact, drawn from one population. 

 Table 6 shows the distribution of 35 individuals with serration count 

 of P4 and provenience considered as attributes. 



Table 6. — Contingency tabic of serration counts and provenience for 35 specimens of 

 Pi of the Ptilodus mediaevus-montanus group 



Although the data are inadequate for the calculation of coefficients, 

 it is clear that there is positive association of Montana and 14 serra- 

 tions and of New Mexico and 12 serrations, and this association is of 

 significantly greater degree than would be expected from effects of 

 random sampling. In spite of the fact that the two samples overlap 

 in this respect and that they could not be separated if they were from 

 one population, it is clearly very probable that the two populations 



