144 TEETIAE1 VERTEBEATA OE THE EATtTM. 



found in the immediate neighbourhood of the type mandible of P. headnelli, and 

 from their size probably referable to that species. Other specimens differing only 

 in their, smaller dimensions will be found mentioned in the Catalogue, where they 

 are referred provisionally to other species. 



The atlas of P. headnelli (PI. XV. fig. 3) is of very large size, equalling the same 

 vertebra in a small Indian Elephant. The condylar cups (cond.) for the occipital condyles 

 are deep ; they are wide dorsally, but narrow considerably towards their lower end. The 

 articular surface for the skull is rather wider and lower than in Mastodon or Mephas, in 

 which the long axis of the condylar cups is nearly vertical. The arch forms a low nearly 

 straight bar, which bears on its dorsal surface a median transverse ridge (n.sp.) separating 

 an anterior rugose surface for the attachment of muscles from a nearly smooth posterior 

 one. Anteriorly the base of the arch is perforated by a foramen for the transmission of 

 the first spinal nerve. In Elephas and Mastodon the arch is more elevated and shorter 

 from side to side. The ventral bar is likewise longer and straighter than in those genera, 

 though some species of Mastodon (e. g., M. americanus) approach the present form 

 in this respect. The notch for the odontoid process {od.) is broader and shallower than 

 in Mastodon, ElepTias primigenius, and E. maximus, in which it is comparatively narrow 

 and deep ; E. africanus resembles Palceomastodon most nearly in this respect. The 

 distinct separation of the odontoid notch from the neural canal distinguishes the atlas 

 in this genus from that of Arsinoitherium. The posterior surface for union with the 

 axis differs widely from the form seen in the atlas of other Proboscidea. In these 

 the lateral surfaces for articulation with the centrum of the axis are continuous both 

 with that for the lower surface of the short blunt odontoid process and with one 

 another beneath it. In Palceomastodon, on the other hand, the two lateral Surfaces, 

 which make only a small angle with the transverse plane, are distinct from one 

 another and are not continuous with the odontoid facet, which is a cylindrical surface 

 on the upper face of the ventral bar ; this latter bears on its postero-ventral border 

 a blunt hypapophysial process. The transverse processes (tp.) are about the same 

 size as in the Indian Elephant ; they are perforated at their base by a foramen {v.c.) 

 which is much smaller than that found in the atlas of Mastodon and Elephas. The 

 upper pedicle of the transverse process is much wider than the ventral one; the 

 anterior face of the transverse process looks downwards and forwards. On the whole, 

 this vertebra approaches very nearly to the atlas of the Elephant, differing from it 

 mainly in (1) the form and relations of the surfaces for articulation with the axis ; 

 (2) the smaller size of the vertebrarterial canal ; (3) the more oval outline of the 

 surface for articulation with the skull. 



The axis (PL XVI. figs. 1, 1 a) differs much more widely from the modern Pro- 

 boscidean type than does the atlas, and at the same time approximates more nearly to 

 the form seen in some Ungulates (e. g., Hippopotamus) with necks of moderate length. 

 The odontoid process {od.) is large and somewhat compressed laterally ; its ventral 



