1910.] Amphitherium ] Oriyin aj the Tritubercular Molar. 181 



upper teeth but us iu all primitive tuberculoseetoriul dentitions served 

 chiefly to pierce and hold the small living prey. 



The posterior borders of the upper molars were probably more directly 

 transverse and less oblique than in Didclphis; because the short talonid 

 would not leave room for an oblicjue posterior shear; consefjuently the 

 metastyle shear in the upper molars must have been less prominent than in 

 Didelphis. Unfortunately it is not certain that the antero-external face of 

 the trigonid lacked the deep fossa for the metacone (Bensley's "anterior 

 cingulum") which is so characteristic of Didelphis and its allies; but at any 

 rate it seems reasonable to infer that the metacone in Aviphitherium was not 

 set so far internally as it is in Didelphis and was not greatly enlarged and 

 sharply V-shaped on its internal surface. From the inferred anteroposterior 

 shortness of the upper molar, and from the general resemblance of the 

 lower molar of Amphitherium to the lower molar of Dri/olesfes it also seems 

 reasonable to infer that, as in the upper molar of Dryolestes, the metacone 

 was smaller than the paracone and like it rounded rather than V-shaped. 

 As to the character of the outer or overhanging part of the upper molars, 

 the lower molars afford no direct inference ; but the conditions in Peralestes, 

 Dryolestes and Didelphops indicate the presence of a parastyle, a metastyle 

 and perhaps of an external cingulum indented at its mid-point. 



Summing up these inferential characters of the upper molars of Amphi- 

 therium we get a very primitive tritubercular molar of the same general type 

 as that in Dryolestes, as figured by Osborn (1907, p. 27, fig. 14) and by 

 Gidley (190G, pi. v, fig. 2 ^) : namely a slightly asymmetrical triangle, shorter 

 on its anterior leg and very broad transversely; antero-posterior diameter 

 })erhaps somewhat greater than in Dryolestes, a high protocone, a large 

 paracone directly external to the protocone, a smaller metacone, a parastyle, 

 an external cingulum indented at its middle })oint, and perhaps a small 

 metastyle (Fig. 10). Whether the paracone and metacone were crescentic 

 externally we have no means of knowing, but the rounded character of these 

 cusps in related genera does not favor the assumption. 



Origin of the Tritubercular Molar Type. 



The existence of Jurassic Therians with very generalized tuberculo- 

 sectorial lower, and tritubercular upper, molars raises the problem of the 

 mode or modes in which this highly effective molar type has been derived. 



It has been argued above that the "Cusp-rotation" theory, while possibly 

 valid for the Triconodonta (p. 177), remains entirely unproven for the Tri- 



1 Gidley's figure is also given in Osborn, 1907, p. 218, No. 2. 



