THE MOST PKIMITIVE KNOWN REPTILE. 293 



14. Intercentra with reduced and unossified dorsal halves 



occur. Rachitomi parallel. 



15. A distinct sacral vertebra is present. Rachitomi parallel. 



16. A cleithrum is absent. Rachitomi not parallel. 



17. An additional bone occurs on each side of the cartila- 



ginous shoulder-girdle. Some Rachitomi parallel. 



18. An entepicondylar foiamen is present in the humerus. 



Rachitomi not parallel. 



19. The presacral part of the vertebral column is shortened. 



20. An articular not fused with the surang-ular occurs. 



Rachitomi not parallel. 



21. The odontoid is a single bone. Rachitomi not parallel. 



The only characters which show that Seymouria is a reptile 

 are : 8, 10, 12, 13, 16, 18, 20, 21, of the foregoing list. 



*It is specially to be noted that some of these [10, 18, 20] 

 are features apparently of very minor interest, and that the 

 really important characters, in the absence of knowledge of which 

 no one would be justified in recognising the reptilian nature of 

 'Seymouria, are those of the vertebral column. Nothing like the 

 large swollen neural arches with horizontally placed articulations 

 is known in Temnospondyls, and the small notochordal centra 

 when taken in conjunction with the crescentic intercentra are 

 equally distinctive. 



It thus appears that the vertebral column affords the best 

 evidence for the reptilian nature of any Palaeozoic tetrapod. 



It is of interest to compare Seymouria in more detail with 

 other Cotylosaurs and putative Cotylosaurs. 



Sauravits. — The small, rather imperfectly-known animal from 

 the upper part of the Stephanian of France described by 

 Thevenin as Sau7-avus costei was regarded by its describer as a 

 reptile, and referred by Case to the Cotylosaviria. This attribu- 

 tion has not yet been challenged ; it appears to have been 

 founded on the presence in the hind foot of two proximal tarsals 

 and a digital formula 2, 3, 4, ? ?. 



The structure of. the vertebral column shows beyond all doubt 

 that this form is a Lepospondylous amphibian. There are no 

 intercentra in any part of the column, and in the caudal region 

 the haemal spines, which in all reptiles are supported by inter- 

 centra (which in Mososaurs may fuse with the centra), project 

 from the lower surface of the whole length of the " centrum." 

 They have peculiar fluted ends, exactly similar to Ceraterj)eto7i 

 and other types forming Miall's family IsTectridia. The dorsal 

 ''centra" ai^e slender hourglass-shaped bones shown by the 

 section figured by Thevenin to be continuous with the neural 

 arch. The presacral neural arches bear a flat lamellar expansion, 

 the middle of which forms the rib-cai'rier, the whole strvicture 

 being identical with that of a nectridian vertebra. In fact the 

 whole column difiers so fundamentally from that of a Cotylosaur 



Proc. Zool. Soc— 1918, No. XXI. 21 



