450 Berry and Gregory — Prorosmarus alleni. 



The mandible of Alachtherium (figs. 3, 4B), Las been 

 identified as positively Odobaenid, while at the same time 

 retaining certain characters of the Otaries. It resembles 

 that of Prorosmarus in the dental formula, in the general 

 form and arrangement of the teeth, in the persistent separa- 

 tion of the opposite rami of the jaw at the symphysis, in the 

 long sloping chin and the slender coronoid. It differs from 

 the mandible of Prorosmarus in being relatively longer, 

 more slender, and shallower, and much more upwardly curved 

 anteriorly, the long axis of the symphysial face being almost 

 at right angles to a line drawn through the front border of 

 the coronoid. This implies a shorter, more upturned facial 

 region. The distance from the last molar to the condyle is 

 relatively greater and the bony " lip " of the jaw is much 

 more produced anteriorly and lacks the eversion of the anterior 

 margin. There is no narrowing of the upper border of the 

 jaw immediately behind the canine, the inferior mental pro- 

 tuberance is much less marked, the posterior dental foramen 

 is farther forward, and the jaw as a whole is much larger than 

 that of Prorosmarus, the total length being about 37 cm from 

 the incisive border to the condyle as against an estimated 

 length of 21 - 5 cm in Proromarus. The molar series in Alach- 

 iherium measures about 9 - 5 cm as against 8 cm in Prorosmarus, 

 that of the latter being thus relatively longer. The molars 

 are also considerably more widely spaced and thus relatively 

 smaller in Alachtherium, especially p 4 . In Alachtherium the 

 lower canine seems to be in a fair way towards being taken 

 over into the molariform series and the projection and nar- 

 rowing of the bony "lip" beyond the canine may foreshadow 

 the compressed rostrum of Odobcenus. 



Trichechoclon huxleyi Lankester is known only from imper- 

 fect upper canine tusks which were very large but not equal 

 to, and more recurved than, those of the existing Odobcenus. 

 These tusks are correlated by Tan Beneden with the 

 remains described by him as " Trichechoclon " koninckii. 

 The latter animal agrees with Prorosmarus in possessing four 

 lower " molars," but, as represented in Van Beneden's plates, 

 was extremely short-jawed, with very short mandibular sym- 

 physis, stout ramus and very large transversely expanded 

 molars. In front of the molars was a still larger excavation 

 or sinus which apparently opened externally and served to 

 lodge the presumably immense upper canine. 



All these forms strengthen the inference that the Walruses 

 have been derived from some primitive member of the Otariidse, 

 probably during the middle Tertiary. 



