IOO GIDLEY 



the derivation of this type of tooth is much simplified, it being 

 not so far removed from the primitive reptilian condition, and 

 though diverging on different lines, is no more specialized, as 

 a whole, than the Triconodon type of tooth, the differentiation 

 being carried on more rapidly in the latter in the special de- 

 velopment of the anterior and posterior lateral cones and their 

 accessory cusps, while in Dryolestes the specialization has 

 apparently been centralized in the development of the high, 

 narrow, heel-like cusp and its supporting fang on the inner 

 side of the molar. 



This view is strongly supported by the evidence obtained from 

 still another characteristic Atlantosaurus-beds type of molar 

 represented by Dicrocynodon. In this form, PL V, fig. 4, the 

 same primitive arrangement of three cusps and two fangs is 

 preserved in the outer portion of the tooth, while on the internal 

 side a large secondary cusp has been developed differing widely 

 in character from that of Dryolestes. This cusp is a laterally 

 compressed cone supported by two rudimentary fangs and is 

 joined to the outer portion of the tooth by a high, wedge-shaped 

 ridge. The base of the inner cone is greatly expanded antero- 

 posteriorly, curving gently outward toward the external portion 

 of the tooth. Thus the crown, as a whole, is greatly constricted 

 medially with the inner and outer portions superficially resem- 

 bling each other. 



From these observations two important conclusions may be 

 drawn : First, that, leaving out of consideration the multitu- 

 berculates, there are among the mammals of the Atlantosaurus 

 beds at least three distinct forms of upper molars representing 

 three primitive types of about equal specialization apparently 

 leading off in entirely independent lines. Probably only one of 

 these, Dryolestes, represents an ancestral type from which the 

 Upper Cretaceous and later forms possessing trigonodont molars 

 may have been derived. Second, that the evidence derived 

 from the Atlantosaurus beds mammals entirely supports the 

 evidence of embryology and agrees in general with the "pre- 

 molar analogy " theory. Thus, the evidence from all sources 

 points overwhelmingly to the conclusion that the primary cone 

 is to be found on the outer side in the upper molars of primi- 



