MORRISON MAMMALS: MULTITUBERCULATA 23 



the opening of the alveolus would be in advance of the maxillo-premaxillary suture. 

 This would be at least anomalous for a premolar or canine, but it would be entirely 

 normal for an incisor. The roots of enlarged incisors often extend back into the 

 maxilla. Comparison with Psalodon and with Bolodon, both very closely related, 

 leaves little doubt that there was no canine and that there were only five premolars. 



The next question is that of the division of the seven cheek teeth into premolars 

 and molars. Marsh's view, supported by Gidley and others, was that there were five 

 premolars (reduced in number in the Cretaceous and Paleocene multituberculates) 

 and two molars (constant for the suborder Plagiaulacoidea). Osborn (i888a) sug- 

 gested a formula of P* M^ for Bolodon. Broom (1914) recognized the possibility of 

 this, but considered P* M' as probable for Bolodon and Ctenacodon. The whole ques- 

 tion is a complex one, and it has been treated in a general way elsewhere, so that a 

 brief resume must here suffice. Broom especially was probably influenced by the desire 

 to give Ctenacodon a normal mammalian dental formula. The fact is that Ctenacodon 

 is not a "normal mammal." It is removed by millions of years from the first mammals 

 with a fixed formula of either P' M* or P^ M^, and as Broom himself has conclusively 

 shown, it is not related to these mammals. Analogy with forms so far removed in time 

 and in zoological relationships must surely be an uncertain guide. If the morphology 

 of the teeth themselves were amenable to such interpretation, it would be gladly ac- 

 cepted, but this is not the case. The fourth and fifth upper cheek teeth are almost identi- 

 cal in structure, in function, and in degree of wear in the known specimens. It is 

 logically impossible to refer them to different dental series; the formula must be either 

 P' M*, as given by Osborn, or P^ M^, as given by Marsh. Now, the fourth and fifth 

 upper cheek teeth do not have the same structure as the two which follow. The latter 

 are grinding teeth (literally "molars") with two straight longitudinal rows of cusps 

 separated by a straight valley, while the former are grasping and shearing teeth, with 

 conical, rugose cusps arranged roughly in two rows but not adapted to fore-and-aft 

 motion, in other words exactly like the teeth which precede them except for having 

 two or three more cusps. Furthermore they act against the lower shearing teeth or 

 premolars." Their consideration as anything but premolars could only be justified in 

 the present state of our knowledge by actual demonstration of the extent of tooth re- 

 placement in these genera. This is not now forthcoming, and it would be surprising, 

 although of course not impossible, if it ran counter to the morphological facts. 



A few other formulae have been assigned to this group, such as 1 1 C J P4 M3 ," but 

 they are hardly worthy of more extended discussion. 



The first three premolars are identical in pattern and of almost the same size, 

 although the second and third are progressively a little smaller. Each has three low 



" Broom divides the lower shearing teeth also, considering the last one or two as molars. Without 

 depreciating Broom's ingenious argument (1914, p. 121), .study of the illustrations here given will show 

 at once that this view does violence to the known morphology. It can only be maintained by having 

 recourse to analogies which are either doubtful or demonstrably incorrect. 



*" Schlosser in Zittel's Gruridziige, 1923. This is for Plagiaulax and gives that genus two more 

 teeth in the lower jaw than it actually possesses, deprives it of two of its upper incisors, gives it a canine 

 of which there is no trace, and adopts an untenable division of the premolar-molar series. 



