334 PATAGONIAN EXPEDITIONS: PALEONTOLOGY. 



large size of the processes gives them a very complex appearance. Very 

 large and concave prezygapophyses rise prominently above the neural 

 arch, while the rugose metapophyses are shorter and heavier than in 

 Hapalops; the accessory prezygapophyses are prominent, shelf-like pro- 

 jections below and external to the primary pair; the articular surface is 

 not plane, as it is in Hapalops, but concave, and is reflected upward upon 

 the external side of the primary pair, an incipient stage in the development 

 of the second accessory pair of articulations. The postzgapophyses are 

 convex semicylinders, separated from the transverse processes by deep, nar- 

 row notches ; the accessory surfaces are on the transverse processes and 

 chiefly on the ventral side, but partly reflected upon the dorsal side also. 



Four lumbars are associated with one individual, but only the third 

 is sufficiently well preserved to show the mode of articulation (Plate LX, 

 figs. 5-5*). In all of the lumbars the centra are of about the same 

 length, but become broader and more depressed posteriorly, especially 

 the hinder faces; the third and fourth have short and blunt, but quite 

 prominent projections from the infero-lateral angles of the posterior faces ; 

 except in the first, in which they are obsolete, the ventral foramina are 

 very large, and the opening in the floor of the neural canal is extraordi- 

 narily large in the second vertebra. The neural spines differ from those 

 of the hinder part of the thorax in being longer vertically, narrower antero- 

 posteriorly, but much thicker and more erect ; the successive spines appear 

 to have been nearly or quite in contact. The zygapophyses, at least of 

 the third lumbar, have an additional complication, as compared with those 

 of the posterior thoracic region, namely, that the accessory articular pro- 

 cesses are each divided into two, one dorsal and one ventral, the only 

 instance of the kind known among the Santa Cruz Gravigrada. What 

 remains of the transverse processes shows that they were long, broad and 

 depressed. 



The number of sacral vertebrae can hardly have been less than five and 

 may have been six. The first vertebra has a very broad and depressed 

 centrum, thinning backward into a mere plate ; all the other centra are of 

 similar plate-like form, except the last, which has a large posterior face ; 

 the penultimate sacral has a pair of low, parallel, infero-lateral keels, which 

 on the last become quite prominent and are separated by grooves from the 

 chevron facets ; very small ventral foramina appear on the first and last 

 vertebrae. Posteriorly, at least, the neural canal is relatively larger than in 



