SSO The American Naturalist. [June, 
chelyide represent a specialized branch of the Pinnata there cannot 
be any doubt whatever. The mosaik-like carapace and plastron of 
these forms is probably a secondary formation, which appeared 
after the dermal part of the ribs had disappeared entirely. The 
oldest Dermochelyide known are from the ,lower Eocene 
(Eosphargis). True Cheloniidz are known already from the Creta- 
ceous, and the intermediate Protostegidz are from the same form- 
ation ; zs probable that the Protestegide have to be considered as 
the ancestor of the Dermochelyide. 
After it has been shown that the Athecæ are an unnatural 
group, and belong to the Pinnata, we have to consider the other 
divisions proposed. Boulenger, Lydekker, Déderlein, Zittel, all 
accept the groups Pleurodira, Cryptodira and Trionychia; these 
groups are certainly natural, as will be admitted by everybody ; a 
new section was introduced by Lydekker under the name of 
Amphichelydia (Quart, Jour. Geol. Soc, XLV., p. 518, 1880). 
“They are characterized by having a shell constructed on the 
plan of that of the Cryptodira and Pleurodira, in which mesoplas- 
tral bones and an intergular shield are developed. The pubis 
may articulate, without sutural union, with the xiphiplastral.” 
The skull and neck are unknown. The coracoid and humerus 
(when known) are of a Pleurodiran type (Lydekker, Cat., pp. 204, 
205). This group is also natural, and corresponds to a sub- 
order to which I have given a different name in MSS., and the 
characters of which I can point out in full. The material on 
which these characters are based consists of nearly all parts of 
the skeleton, including skull and cervicals of Compsemys plicatu- 
lus Cope, the oldest American Tortoise, from the Jurassic of the 
Rocky Mountains. Most of the material was examined at the 
Peabody Museum, New Haven, Conn. Especially I have to 
state the interesting fact that Compsemys has a complete mesoplas- 
tron, and resembles very much Pleurosternum. This is another 
support for the view that the Camarasaurus beds correspond to 
the Purbeck and Oolite of England. 
_ I give now the characters of the five sub-orders of Testudinata, 
which I adopt: 

