1907.] OF MAMMAL-LIKE KEPTILES. 1057 



has the usual foui- elements in the proximal row, two centralia, of 

 which the outer is the larger, and four distal carpals. It differs 

 from that of Dimetrodon and agrees with that of Sphenodon and 

 most reptiles in having the pisiform articulating with the ulna. 

 There is also evidence of speciahsation in the 5th carpale being 

 lost, though retained even in S'phenodon. 



In the recently discovered GcdecMrus the carpus is preserved 

 in perfect condition. In the proximal row are a large radiale and 

 ulnare, with a smaller but well- developed intermedium and 

 pisiform. In the distal row are five cai'palia, but the 5th is 

 small. In the centre of the carpus are two centralia, of which 

 that to the radial side is the smaller and lies between the radiale 

 and the 1st carpale. It will be seen that this small animal with 

 a mammalian digital formula has nevertheless a carpus almost 

 exactly similar to that in Sphenodon. 



The only Therocephalian carpus at present known is that of 

 Theriodes')imis{^2), which, though well preserved, has unfortunately 

 the elements somewhat displaced. Restorations have been 

 attempted by Seeley, Bardeleben, and myself. We may feel 

 pretty certain that there are the usual four bones in the proximal 

 i-ow, and also that there are only four in the distal row. There 

 are apparently two centralia, one of which is rudimentary. 



In the Anomodontia the carpus is much better known, that of 

 both Oudenodon and Opisthoctenodon (23) being represented by 

 perfect specimens. In Oudenodon the arrangement is almost 

 precisely similar to that in Galechirus, there being two large 

 centralia, of which the inner lies between the radiale and the 

 1st carpale. There is a small but distinct 5th carpale. In OjnstJio- 

 ctenodon there is no 5th carpale and the inner centrale is not so 

 markedly between the radiale and 1st carpale. 



Until recently very little has been known of the Oynodont 

 carpus. A very imperfect carpus of Microgomphodon (24) has 

 been figured by Seeley, but it is too badly preserved to help us 

 much. About six months ago I figured the carpus of a new 

 Oynodont J^lurosuchus (18), which shows at least the distal 

 carpals and the centralia in true position, and though the 

 proximal elements are somewhat displaced, we may feel fairly 

 sure of their relations. There are but four distal carpals, the 5th 

 being lost as in Mammals. Of the centralia that to the radial 

 side is rudimentary, the other being large, from which we may 

 infer that the centrale of the mammalian carpus corresponds to 

 the outer of the two centralia of the primitive reptilian carpus. 

 In the proximal row there are probably the usual four bones. 

 We thus find that the Oynodont carpus is identical with that of 

 the typical Mammal, except that whereas in the latter there is 

 only a 0", the former has, in addition to a 0', a rudimentary 0'. 



The evolution of the pelvis is now pretty well known. Perhaps 

 the most primitive type of pelvis in any land animal is that seen 

 in some of the Stegocephalia, such as Discosaurus. Here we 



