368 Bulletin American Museum of Natural History. \Yo\. XXVII, 



(3) Diploe of the skull not markedly inflated. Primitive. 



(4) Narrow supraorbital region. Primitive. 



(5) Naso-premaxillary and palatal regions very different in Mo?ri- 

 tlierium and Pakpomastodon. 



(6) Xeck longer in Ma'ritheriimi; atlas with peg- like odontoid (An- 

 drews). Primitive. 



(7) Dorso lumbars (Andrews) at least 23 (D. 19, L. 4, S 4), i. e. more or 

 less intermediate in number between Proboscidea (D. 19-20, L. 4-5, S 4) 

 and Sirenia (D. 17-19, L 2-5). 



(8) Referretl humerus {op. cit., pi. xi) somewhat more primitive than 

 that of Pala^omas-fodon in smaller supinator crest and larger internal condyle 

 (cf. Taligrada). 



(9) Referred femiu' lacking third trochanter (retained, low down on 

 shaft, in Paloeomastodon). 



Conclusion in regard to Mwriiherium. The genus represents a very 

 primitive offshoot from the Proboscideo-Sirenian stock. Its dentition and 

 certain other characters indicate a nearer alliance with the Proboscidea than 

 with the Sirenia, but it is far more primitive than any other known repre- 

 sentative of either order. 



Points of similarity to the Hy races also are not lacking, especially the 

 detailed relations of the premaxillaries, maxillaries, nasals and frontals, the 

 areas for the insertion of the masseter muscle, the backward extension of the 

 malar, reduction of the mastoid, general shape of the lower jaw, shape of 

 the ilium and other characters. 



Genetic Relations of the Proboscidea. 



Assuming that Mwriiherium is a relatively primitive offshoot of the 

 Proboscidea, allied also to the Sirenia, and perhaps more remotely to the 

 Hyracoidea what light does it seem to throw on the derivation of these orders ? 



There can be no question of the derivation of Ma>ritherium from typical 

 Placentals Avith a dental formula of |xil:' ^ normal milk dentition (retained 

 in Mwriiherium, cf. Andrews, p. Ill), and normal cranial foramina. The 

 premolars could not have been molariform and p' was probably bicuspid 

 {cf. Mwritherium). The premolars seem to suggest the quadrate contour 

 of the premolars of Megalohyra.v. The upper molars were derived from a 

 tjuadritubercular, incipiently bilophodont type in which the proto- and 

 metacomdes were ridged transversely. 



The jientadactyl manus may have resembled that of Ilijra.v in a general 

 way and also in certain detailed relations of the carpals (p. 448). The pes 

 was probably not dissimilar to that of Pantolambda (p. 358). With the 



