348 PATAGONIAN EXPEDITIONS: PALEONTOLOGY. 



skull is viewed from above or below ; the maxillaries end abruptly at the 

 anterior nares and have no such forward prolongation as do those of 

 Scelidotheriuni. Posteriorly, the palate, which extends a short distance 

 behind the last tooth, is very narrow and quite decidedly convex, broad- 

 ening and flattening anteriorly between the divergent rows of teeth ; at 

 the front border is a narrow and deep V-shaped notch for the premaxillary 

 spines ; the facet for the external branch of the premaxillary is similar in 

 shape and position to that of Hapahps. 



Of the premaxillae only the spines are known, but from the shape of 

 the anterior ends of the maxillaries, the facets upon them, and from the 

 predental beak of the mandible, the shape and size of these bones may be 

 approximately determined. In all probability, the premaxillae of Nenia- 

 therium were, like those of Mylodon, each composed of two branches, one 

 internal, or the spine, and the other external, but having no distinct 

 anterior branch, such as is found in the Santa Cruz Megalonychidcc ; they 

 were more elongate than in Mylodon, very much less so than in Scelido- 

 theritim, but, as in the latter, the spines taper posteriorly to a point, filling 

 the V-shaped notch in the palate. The posterior nares are narrow in front, 

 broadening rapidly behind ; the side-walls, formed by the pterygoids and 

 alisphenoids, have much the same shape as in Hapalops. 



The mandible, which is known only in this genus, not in the succeed- 

 ing one, has already been described in the account of the family. 



No vertebrae have yet been identified, except a few in connection with 

 one individual (No. 15,893), the reference of which, though very probable, 

 is not entirely certain, because no tooth or part of the skull is associated 

 with the specimen. The posterior thoracic vertebrae resemble those of the 

 Santa Cruz Megalonychidce and have the centra perforated in the usual 

 way by paired canals opening on the ventral surface ; the caudal vertebras 

 also are not especially characteristic. On the other hand, the only lumbar 

 associated with this specimen has a very peculiar type of transverse process, 

 different from anything known in either of the preceding families ; the 

 process is quite short, but the distal portion is very broad antero-pos- 

 teriorly ; it is deeply notched on the anterior border and its base is perfo- 

 rated by a large vertical foramen ; the shape of the process is thus 

 deceptively like the inferior lamella of a cervical, except that it projects 

 horizontally, not ventrally; the accessory articular processes are, as in 

 , a small anterior pair, placed on the neural arch below the true 



