MORPHOLOGICAL REVISION 85 



carpus and tarsus slightly. A small fragment of shale from the same region in 

 which this specimen was found carries the impression of two feet of a smaller animal, 

 but indicates that the foot was elevated, as only the impression of the claws and 

 of a flat pad are shown. 



I believe these animals to have been harmless, sluggish, terrestrial herbivores, 

 possibly fossorial in habit, at least to the extent of excavating burrows for their 

 protection. The attitude was habitually prone and they could not elevate the body 

 on the limbs even so much as the Crocodilia, nor could they develop any speed even 

 for short intervals. 



PHYLOGENETIC POSITION. 



The Diadectosauria have long been regarded as the nearest known form to the 

 primitive ancestor of all the reptiles. This idea must be completely abandoned 

 for the suborder. Aside from the extreme degree of specialization in all parts of the 

 body, which shows how far it is from the primitive line, the structure of the skull 

 makes it impossible that such types as double and single arched reptiles could have 

 been derived therefrom, at least in the simple manner long believed. In fact, the 

 greater our knowledge of these primitive forms becomes, the less certain are we of 

 the primal origin of two great stems from a form with a complete roof. The Rhyn- 

 cocephalian double-arch structure of the skull can not be derived from that of 

 Diadectes or any other known Permian reptile which is not^far too highly specialized 

 in other characters to make its consideration worth while. 



In a previous discussion of these forms (32) the author has suggested the possible 

 ancestral relationship of the suborder Diadectosauria to the turtles. It is not 

 assumed that this relationship is direct, but the specific resemblances, as well as the 

 general effect of the mounted skeleton, indicate very strongly that they were not far 

 from the ancest ral form of the turtles. The specific points of resemblance as first 

 stated are as follows: 



1. The form and relations of the quadrate. 

 II. The degenerate palate and the disappearing transverse bone. 



III. The absence of teeth on the palatines and pterygoids. 



IV. The absence of a parasphenoid rostrum attached to the basisphenoid bone. 

 V. The absence of prevomers and the presence of an anteriorly placed single 



vomer (parasphenoid.) 

 VI. The method of entrance of the internal carotids into the brain cavity. 

 VII. The presence of paired descending plates from the roof of the skull anterior 

 to the brain case. 

 Of these numbers, IV, V, and VI must be modified. 

 IV. Several specimens show no traces of an attached parasphenoid rostrum; 

 this one. No. 4684 Am. Mus., shows a short and degenerate process. 

 Versluys has recently brought evidence to show that Dermochelys has 

 a distinct parasphenoid (36). 

 V. This is wrong. There is a separate parasphenoid (vomer) and distinct 

 paired prevomers in the median line. 

 VI. In some specimens there are no foramina perforating the lower surface of 

 the basisphenoid; in this there is a single large foramen. 



The original conclusion seems still to be justified. Moreover, it will be at 

 once seen how close is the resemblance of the squamosal and quadrate bones to the 

 same elements in the turtles. If the quadratojugal and with it the quadrate fora- 

 men were to disappear, the condition of the turtles would be exactly realized; for in 



