506 PROCEEDINGS OF THE NATIONAL MUSEUM vol. loi 



It seems clear, then, that we are dealing with a different strain of 

 cats, one that was intermediate between P. atrox and P. onca, and one 

 that, for reasons of priority, should be called P. augusta. 



It cannot be determined by a study of P* alone whether our P. 

 augusta was an independent species, or a subspecies of P. onca or of 

 P. atrox. Even if good fortune should provide us with a large num- 

 ber of these intermediate cats, and the size of their premolars should 

 overlap the range of P. atrox on the one side and of P. onca on the 

 other (the above statistics are not concerned with this possibility and 

 do not make it at all improbable) , this would not indicate that they 

 were all one species. The species P. atrox is characterized by many 

 other consistent peculiarities of form, and enough specimens showing 

 such features have already been assembled to make it clear that the 

 species possessing this particular constellation of morphological char- 

 acters was restricted in range of size of premolars within fairly clear 

 limits. No specimen of P. atrox thus far found possesses premolars 

 of the dimension of the one we have found, and the implication of the 

 mathematics is that none existed that did possess such premolars. 

 The same is true for all known races of P. onca. In other words, the 

 probability is that a cat having a carnassial measuring from 31 to 

 34 mm. will be found to differ in other morphological features from 

 atrox and onca. 



To explore further this possibility we have constructed the graphs 

 shown in figures 97 and 98, using Simpson's (1941b) convenient method 

 of plotting logarithmic differences. The characteristics chosen for 

 reference are those used both by Merriam and Stock in their table 98 

 for comparison of P. atrox and P. spelaea and by Simpson (1941b, 

 fig. 10) for P. atrox and P. onca? 



' The virtue of this kind of graph is that it facilitates the comparison of relative pro- 

 portions of different parts of the body in different animals. If the proportions of P. atrox 

 be talcen as the standard for comparison, as in figure 97, the base or zero line represents 

 these, and is straight. Another animal with the same relative proportions would appear 

 on the graph as another straight line, to the left if smaller in all parts, to the right if 

 larger in all parts. A crooked line represents an animal with different proportions, irre- 

 spective of the actual dimension of any of the parts. These dimensions may, of course, 

 be calculated if such data as those in the next paragraph are provided. 



The base line in figure 97 represents the average values for all the measurements listed 

 on the right. As given in Merriam and Stock they are : 



Structure Millimeters 



Condylobasal length of skull 353.0 



Bizygomatic breadth of skull 250.4 



Length of P* 89.4 



Length of humerus 368.5 



Length of ulna 416.0 



Length of third metacarpal 126.6 



Length of femur 421.3 



Length of tibia 361.7 



Length of third metatarsal 146.1 



Such dimensions, which determine the zero line, should always be recorded in connection 

 with any such graph, as they enable the reader both to compute any other dimensions 

 indicated, and to add to the graph additional lines representing other material for 

 comparison. i 



