I20 PATAGONIAN EXPEDITIONS: PALEONTOLOGY. 



Vertebral Column, Ribs and Sternum. — In one individual, the type of 

 T. garrettorum (No. 15,164) the vertebral column appears to be complete 

 to the end of the sacrum and gives the formula: C.7, Th. 13, L.6, S.7. 

 This is a surprise in view of the widely held opinion that Macrauchenia 

 had the perissodactyl formula of 23 trunk-vertebrae, and, at first sight, this 

 would seem to militate strongly against the derivation of the Pampean 

 genus from TJieosodon. It is, however, very doubtful whether Macrau- 

 chenia actually possessed so large a number of trunk-vertebrae. Bur- 

 meister, it is true, definitely states ('91,483) that in the La Plata Museum 

 skeleton the number of thoracic vertebrae is 18 and of lumbar vertebrae 5, 

 but there is some inexplicable mistake involved in the statement, as the 

 La Plata skeleton is much less complete than Burmeister seems to have 

 supposed. In fact, that skeleton, which Lydekker has figured ('96, 76, 

 fig. 10) makes it very probable that the number of trunk-vertebras could 

 not have exceeded 19 or 20. On the other hand, it is not altogether 

 certain that the number 19 attributed to TJieosodon is correct, for the speci- 

 men above referred to (No. 15,164) was not found with all the bones in 

 their natural connections and it is possible that some of the vertebrae are 

 missing. Still, the vertebrae, when put together, seem to form a complete 

 series without breaks. 



In TJieosodon the neck is very long and its proportions have much 

 resemblance to those seen in the llama. The atlas (PI. XVIII, fig. 2) 

 is quite elongate in proportion to its width, though very much shorter 

 than the succeeding cervicals, and has a large, subcircular neural canal. 

 The anterior cotyles for the occipital condyles are large, widely separated 

 dorsally, where the neural arch is quite deeply emarginated between them ; 

 ventrally, the cotyles are much more closely approximated. The poste- 

 rior cotyles for the axis are small and nearly plane, while the facet for the 

 odontoid is quite deeply concave transversely and is not confluent with 

 the posterior cotyles. The neural arch is broad in both directions and 

 moderately curved ; the spine is a very low ridge, the height of which 

 varies somewhat in different individuals. The inferior arch is quite nar- 

 row antero-posteriorly and has but a very small hypapophysis, or none at 

 all. The transverse processes are very short, with regularly curved free 

 border ; anteriorly, they do not reach the outer wall of the cotyles but are 

 notched and posteriorly, the cotyles also project behind them. At the 

 base, the transverse processes are perforated by the vertebrarterial canal, 



