146 BULLETIN 16 9, UNITED STATES NATIONAL MUSEUM 



tooth should be a canine, that Pi should be present in an upper Eocene 

 tarsioid when it is absent, as far as surely known, in every other laiown 

 tarsioid even in the Paleocene, or that an enlarged lower canine should 

 occlude against an enlarged first upper incisor and anterior to a second 

 upper incisor. The occlusion in itself is so suggestive of an incisor 

 that only the strongest contrary evidence would warrant any other 

 conclusion, and the other considerations seem to place this almost 

 beyond doubt. It is true that in Tarsius the largest anterior mandi- 

 bular tooth is the canine, but as Stehlin (1916) himself has shown the 

 analogy with Necrolemur is very distant, and in Tarsius the lower 

 canine occludes between I^ and C as would be expected, not between 

 I^ and I^. Furthermore, the actual formula in Tarsius is yjj;^, as 

 I believe it was also in Necrolemur. ^^ 



If this formula be accepted for Necrolemur, its anterior dentition 

 could be derived from that of Paromomys by further enlargement of 

 the already enlarged incisor and great reduction of the canine, still of 

 moderate size but not enlarged in Paromomys. The premolars of 

 Necrolemur are broader, lower, and more proclivous than in Paro- 

 momys, and there is a distinct metaconid on P4. The lower molar 

 structure is closely similar throughout except for details in Necrolemur 

 like the complete loss of separate paraconid on M2.3, which are the 

 logical continuation of tendencies clearly present in Paromomys. 



In the upper jaw, the 3-rooted P^ of Necrolemur is also progressive. 

 P* and the molars are less transverse than in Paromomys, a feature of 

 no clear significance. The protocone of P* may be smaller in the 

 later genus; if so, it is the only character that suggests, and it does 

 not prove, that Necrolemur could not be derived from Paromomys. 

 Aside from their proportions, the upper molars of Necrolemur dift'er 

 in having stronger conules (the metaconule double) and distinct 

 hypocones, but the basic plan is remarkably similar, as already 

 suggested. 



As far as the dentition goes, it must be concluded that Necrolemur 

 and Paromomys are probably rather closely related, and the latter 

 could be ancestral to the former. The conclusion is obviously un- 

 proved and open to doubt. Corresponding with their great separation 

 in space and in time, the genera do differ markedly and annectant 

 types are unknown, but the fundamental similarity is striking. 



Trogolemur, Uintalestes, and Phenacolemur are all incertae sedis;, 

 but all show some special resemblances to Paromomys and its Fort 

 Union allies. Trogolemur has the same dental formula as Paromomys, 

 jx^. The incisor is relatively larger, and the next tooth, presum- 

 ably the canine, is very small. The premolars are more expanded 



«' Although the ease is less clear, by analogy it seems probable that the enlarged median mandibular 

 teeth of Tetonius are also a pair of incisors, not canines as generally supposed following Matthew. I should 

 write the Tctov'ms formula either 1.0.3.3 or 1.1.2.3, of which the second is perhaps slightly more probable. 

 (Matthew wiote 0.1.2.3 but this is an evident lapsus, and he clearly meant to write 0.1.3.3-) 



