DERIVATION OF ORDER. 29 
in Diadectes the temporal roof is notcht where applied to the quadrate and the latter 
is excavated to form a tympanic cavity; instead of a pair of prevomers, there is a 
single vomer; and there is a sort of carapace composed of bony plates ov erlying 
ae pairs of the ribs. 
It seems evident that we are getting close to the base of the phylogenetic tree of 
the chelonians; but, likewise, et we have not yet reacht it. Taking into con- 
sideration the fact that no turtle possesses a temporal fenestra, that when the 
temporal roof is deficient either the zygomatic arch or the parietal arch or both are 
missing, it seems impossible to derive ake turtles from any group of reptiles in which 
even the beginning of such fenestra has been made. If this view is correct, there are 
excluded at once from the chelonian ancestry all those reptiles belonging to Professor 
Osborn’s Diapsida and all the Anomodontia, limited so as not to include the 
Cotylosauria. The Diadectide, too, would be excluded, if they really possest 
temporal fenestrae. If the presence of a pair of prevomers, instead of a vomer, has 
the importance attributed to it by some recent authors, the Cotylosauria, as limited 
by Case, would have to forego their claims. Of known reptiles we appear to be 
limited to the Diadectidz and the Otoccelide, in our search for the ancestors 
of the chelonians. 
It is improbable that any of the reptiles of the Permian of Texas or equivalent 
deposits were the ancestors of the turtles. When we consider that already in the 
Upper Trias the turtle Proganochel ys possest a typical shell, we must conclude that 
the earliest of the race must have been in existence as early as the Permian itself. 
It is, of course, possible, indeed quite probable, that some earlier, less differentiated 
form of one or the other of the two families just mentioned gave origin to the most 
primitive turtles. 
It is a serious objection to any of the Cotylosauria, using the name in the wider 
sense, that have been mentioned as possible ancestors of the fartlest that they appear 
to Rave possest no ventral armor. Almost certainly the turtles inherited this portion 
of the shell, in some form, from the Stegocephalia. The Diadectes described by 
Case presented no traces of a plastron, or of abdominal ribs. There are various 
other reasons why this reptile can not pose as the founder of the chelonian line. 
Among these are ‘the elevated spines of the vertebra, the great amount of motion 
between the several dorsal vertebra, their complex structure, the anchylosis of the 
sacrals, and the fact that the plates of the carapace lie between the scapula and the 
ribs. In turtles the scapula lies inside, not outside, of the costal plates. 
Dr. O. Jaekel, of Berlin, has described (Neues Jahrb. Min., 1, 1902, p. 127) an 
interesting genus named by him Placochelys and regarded by hin as standing in 
ancestral relations to the turtles. The body of the apanale is short and broad, and is 
covered by an armor of thick small plates, all closely joined. There is no ventral 
armor. The specimen was found in the lower portion of the Keuper. The well- 
preserved skull shows that there were large supratemporal fenestre and that the 
nasal openings were separate and far behind the end of the snout. These structures, 
especially the first named, make it improbable that this reptile is at all closely 
related to the turtles. 
The discovery of Otocelus and Diadectes and Placochelys is like the capture of 
so many strag glers and deserters from an army of heterogeneous composition. 
From these prisoners we may learn something of the personnel and the equipment 
of that army; but it would be unsafe to ee from the data obtained too wide 
conclusions. From Plac ochelys we learn that in Triassic times there were broad and 
short bodied reptiles of tortoise-like form and covered with an armor of small plates. 
From Diadectes and Otocelus we discover that in Permian times there were more 
