440 PROCEEDINGS OF THE ACADEMY OF [1892. 



rium the cusps have united to form a pair of transverse crests, but 

 a reminiscence of the tubercular stage is preserved in P3. In this 

 tooth the posterior crest is complete, formed by the junction of the 

 trito- and tetartocones, but the anterior crest is unfinished, the deu- 

 terocone remaining isolated from the protocone. 



The foregoing brief survey of the steps of premolar development 

 in those mammalian groups which enable us to follow it out, shows 

 very clearly that the premolars follow an order of differentiation 

 quite at variance with that attributed to the molars, even when the 

 final results are the same. It also shows that in the fourth upper 

 premolar the order of succession in the appearance of the cusps is 

 remarkably constant, while in the other superior premolars and in 

 all of the lower ones, the order is very much less regular. But 

 even in these cases the homologies are clearly given by the position 

 of the new elements with reference to the primary cusp or proto- 

 cone. Some of the Artiodactyla, however (e. g., Procamelus), 

 appear to depart altogether from the typical order, in that the inner 

 crescent of P2 or Pa ( or both) is formed, not by the extension of a 

 single element, the deuterocone, but by the coalesence of two. Even 

 in these cases, however, we found reason to believe that the depar- 

 ture from the normal was less radical than at first sight seemed to 

 be true, for the posterior element of the crest is with great proba- 

 bility to be homologized with the deuterocone. Another result 

 which follows from our survey of premolar development, is the 

 steadiness with which it is followed out when once undertaken. 

 The cusps do not apj^ear hap-hazard, or in a tentative way, but 

 definitely may be traced from the faintest beginnings to the final 

 result. It is obvious, further, that the same result may be indepen- 

 dently attained many times in widely separated groups, and by pre- 

 cisely similar steps. Nothing could be more eloquent of the danger 

 of constructing phylogenies from the teeth alone. Often, it is true, 

 they would turn out to be right, but not infrequently they would 

 lead to the grossest error. 



The Milk Dentition. 



Palreontology is not fitted to throw very much light upon the 

 relation of the milk molars to the permanent set, i. e., Avith regard 

 to the homologies of the cusps, for it is exceedingly difficult to 

 obtain phylogenetic series of the temporary dentition. Some facts 

 may, however, be gleaned which are not altogether devoid of value. 



