186 ON PRE- JURASSIC TETRAPODS. 



from the super-oi-dei' because of certain difierences in the brain- 

 case, which may, however, be due to the small size of Sphenodon, 

 in which alone is the region known. The exclusion is really for 

 the purpose of drawing attention to the extreme smallne.ss of our 

 knowledge of early two-arched reptiles. 



The small aquatic Mesosaurus and Stereostermmi are still un- 

 represented by well-preserved skulls. They seem to be definitely 

 ordinally distinct, but their affinities are quite obscure. 



Finally, the order Protorosaui-ia is I'etained merely because it 

 is already in. existence. Included in it is a series of small 

 Permian reptiles which resemble one another in having slender 

 limbs and a single coracoidal element. There is no evidence that 

 these animals ai'e in reality in any way related. 



In this classification I have refrained from throwing M'eight on 

 the peculiarly modified 5th metatarsal which occurs in Chelonia, 

 Rhynchocephalia, Thecodonts, Crocorliles, Deinosaurs, and Squa- 

 inates, because it is difficult to believe that all these forms can 

 have been derived from the same advanced Cotylosaurian ancestor. 

 It is perhaps an arboreal adaptation, which may have originated 

 separately. Pos-sibly the broadened ribs of Eimotosaurus, which 

 recall those of sloths, ai'e also to some extent an arboreal 

 adaptation. 



This classification is on the whole consciously conservative, 

 but contains many new features, chiefly in the definitions and the 

 super-ordinal grouping. An attempt has been made to make 

 the structural difi;erences separating orders approximate, having 

 regard to the total variation, to those used in separating the 

 orders of mammals, but in many,cases suborders should perhaps 

 be raised to ordinal rank. Super-orders are used to group together 

 orders which seem to have had a common origin. 



