290 Bulletin American Museum, of Natural History. [\q\. XX\^II, 



Trituberculates is the primitive type. But even in the latter group single 

 rooted caniniform canines also occurred {cf. Ambloiherium) and the small 

 double rootetl canine in modern Insectivores may be associated with an 

 exclusively insectivorous diet. In Hyopsodus, Ictops, Ericulus, Centetes 

 and Tupaia the canine is simple. If however the double rooted canine 

 shall prove to be an ancestral Insectivore character, then it will serve to 

 distinguish that order from the ancestral Polyjjrotodonts, Creodonts, Pri- 

 mates and Basal Eocene Prot-ungulates, which probably had normal, 

 small caniniform canines. 



(7) Premolars |, simple, p| 7iot molariform. P| are molariform in Ictops, 

 the Zalambdodonta and Macroscelides. In the Erinaceidte and Tupaiidse 

 p^ is subsectorial. 



(8) Upper molars simple, tritubercular, much as in Ictops, i. e., narrow 

 anteroposteriorly with high V-shaped protocone, small para- and metacones, 

 small para- and metastyles and metaconule, slight external and internal cin- 

 gulum, the latter possibly with an incipient hypocone. This type appears to 

 be the most primitive in the order; for, on the one hand it points back to the 

 type represented by the Jurassic Trituberculate Dri/olestes, on the other hand 

 it approaches the general type seen in the Eocene Pantolestids, Oxyclsenids, 

 Mixodectids, and "prot-ungidates." This primitive molar pattern may 

 have given rise to the modern types in the following ways (cf. Fig. 17, p. 23S): 



(a) The Zalambdodont molar (Fig. 17, 3-5) may have been derived by 

 the fusion of the paracone and parastyle, reduction of metacone, emphasis 

 of para- and metastyle, and of the internal cingulimi, including the pseudo- 

 protocone and the hypocone (p. 238). 



(b) The Erinaceid type (Fig. 17, 6-S) has been derived by the antero- 

 posterior broadening of the molar and emphasis of the hypocone (p. 260). 



(c) The Soricoid type (Fig. 17, 9) as represented by Proscalops may 

 have been derived by the development of the pseudoprotocone and its 

 fusion with the protocone and by the great emphasis of the para-, meso-, 

 and metastyles, which connect with the para- and metacones so as to form 

 the two external Vs. 



(d) The Ptilocercus type (Fig. 21, p. 273) may have been derived by the 

 slight emphasis of the metastyle, anteroposterior broadening of the crown 

 and oblicjue development of the hypocone. 



(e) The Tupaia type (Fig. 22J5) was formed by the great development 

 of the para-, meso- and metastyles so as to produce two external Vs, and the 

 division of the mesostyle in correlation with the enlargement of the hypo- 

 conid. The hypocone remained very small. 



(/) The Macroscelides type may have been formed after the manner 

 of the bilophodont m' of Eriuaceus. Here the protocone unites with the 



