FOSSIL VERTEBRATES — REPTILIA 519 
interest, as the number is the same that occurs in the turtles, and 
this is one of the few clues that we have as to the origin of the 
turtles. The vertebrae were biconcave, but were not perforated 
for the passage of the notochord, which was consequently 
divided into intervertebral segments. In the skin of the back 
directly over the spinal column there were developed three rows 
of bony ossicles, the middle row lying directly over the spinous 
processes and the other two lying upon each side; this develop- 
ment of dermal plates is regarded as the first step toward the 
formation of a carapace, and as an additional evidence of the 
connection of the Pareiasauria with the turtles. The thoracic 
girdle is stout aud heavy and retains the central element, the 
interclavicle, which is lost in the majority of the reptiles; the 
pelvis was large and massive. In life the animal must have pre- 
sented the appearance of a large amphibian; the legs were too 
short to lift the body off the ground, so the belly must have 
dragged as in the crocodiles; the whole body was short and 
squat and the tail was short; the skin was without scales and 
was probably thickened and folded. 
Scattered bones indicate the presence of at least two other 
genera, Propappus and Anthodus, in the South African deposits, 
but not enough of the skeleton is known to warrant a description. 
In the American Permian deposits of Texas and Illinois there 
have been found a large number of forms belonging to this 
order; the best known are Diadectes, Empedias, Chilonyx, Pan- 
tylus, and Pariotichus. 
Diadectes and Empedias are the best known of the forms; 
they are similar to Paretasaurus, though they did not reach the 
size of that form; they had the same broad, flat skull with few 
openings; the teeth are different from the African form, in 
that, although they are similar in size and arranged in reg- 
ular order around the edges, they are not simple in form, but 
are expanded laterally so that they exhibit broad grinding sur- 
faces instead of cutting edges; they were undoubtedly herbiv- 
orous forms; the number of the presacral vertebre is unknown, 
but it was probably greater than in /Pareiasaurus, the verte- 
