2<)1 



THE AMERICAN NATURALIST [Vol. XLIX 



The only known characters by which the skull of Canis lupus can 

 he distinguished from that of the larger domestic dogs is the greater 

 average general size and the relatively larger teeth. In a dog's skull 

 with condylobasal length of 230 mm. the length of upper and lower 

 carnassials is, respect ively. 21.6 and 25.0 mm. In ten skulls with con- 

 dylobasal length of mope than 200 mm. the average and extremes for 

 these teeth are: upper, 20.5 (19-22); lower, 24.0 (22.8-26.0). In all 

 the dog skulls which I have examined, representing such different 

 breeds as the pug, fox-terrier, bloodhound, mastiff, ancient Egyptian, 

 ancient Peruvian, Eskimo (Greenland and Alaska) and American In- 

 dian, the teeth are strictly of the wolf type, never showing any ap- 

 proach to that of the jackal (Fig. 62). 



This indicates that the profound differences of osteolog- 

 ical character which separate the larger breeds of domes- 

 tic dogs are chiefly in the proportions. 



No numerical, or presence and absence diameters are 

 used in Miller's definition of the wolf, arctic fox, and red 

 fox although a number of minor numerical characters are 

 clearly described and figured in his text, especially the 

 cuspules on the incisor and premolar teeth, as shown in 

 Fig. 1. These numerical cusp characters would have re- 

 ceived more attention from a paleontologist partly be- 

 cause of the paucity of material which comes into his 

 hands, partly because he is in a position to observe the de- 

 velopment of these cuspules. 



This contrast between proportional and numerical char- 

 acters brings out a fundamental law in the evolution of 

 the hard parts of mammals which is of great importance. 

 First, characters of form and proportion, without numer- 

 ical change, are constantly originating as a universal prin- 

 ciple and forming the chief distinctions between divisions 

 from the high rank of orders down to those of subspecies, 

 races, and even individuals. Second, numerical loss or 

 fusion of old characters of teeth, digits, or vertebrae is 

 next in frequency, the loss always following diminution 

 in form and proportion. Third, numerical gain of new 

 characters is the least frequent process; it is relatively 

 rare in the endoskeleton, that is, in added teeth, added 

 vertebra? and other segmental parts, added cranial bones, 

 added phalanges; it is more frequent in added cusps on 



