1903. ] VERTEBRE OF THERIODONTS, 179 
arch and hypapophysis of the atlas, and the odontoid process and 
part of the body of the axis. Though the structure of the 
vertebra is essentially similar to that in Gomphognathus there are 
many points of difference. 
The proatlas (p.a.), though occupying the same relative position. 
as in Gomphognathus, is less specialized. It is present as a pair of 
curved, moderately thin, bony plates lying in front of, and probably 
somewhat overlapping, the arch of the atlas, There is no bony 
process developed as in Gomphognathus. 
The atlas (a¢.) consists of a well-developed arch and a distinct 
hypapophysis. ‘The arch is not so complete as in Gomphognathus, 
and must have differed considerably in regard to the arrangement 
of the anterior articular surfaces. In Gomphognathus the occi- 
pital condyles (0.c.) are small and moderately close together, 
whereas in Z'rirachodon the condyles (0.c.) are very large and as 
wide as the atlas. The arch, perhaps on this account, does not 
close in inferiorly to meet the hypapophysis, with which it can 
only have been attached by hgament. The ‘zygapophy sis (z.at.) 
for the axis is very small. “The hypapophysis (/.at.) is very 
sunilar to that in Gomphognathus. 
Of the axis only the odontoid process, imperfectly displayed, 
and a part of the body remain. 
The structure of the upper cervical vertebre is so imperfectly 
known in the majority of fossil reptiles, that there is some diili- 
culty in satisfactorily dealing with the aftinities of the structures. 
It is difficult to avoid being struck by the close resemblance of 
the axis, atlas, and proatlas in the Theriodont to those structures 
in the Crocodiles, more especially as there is no close aflinity 
between the groups. The explanation is probably to be found in 
the fact that both the Crocodiles and the Theriodonts, though far 
removed from each other, have retained with but slight modifi- 
cation the type of structures met with in their common ancestor. 
In Sphenodon, though the proatlas is small, the type is practically 
the same; and I have recently discovered that Procolophon has a 
well-developed proatlas. Procolophon has usually hitherto been 
associated with the Dicynodonts, Theriodonts, and Pareiasaurians, 
but as it has a persistent notochord, abdominal ribs, a plate-like 
pubis and ischium, and 2, 3, 4, 5, and 4 phalanges in the digits, it 
is manifestly much more closely related to Palwohatteria than it is 
to the Theriodonts and Dicynodonts. In most of its structures it is 
probably as primitive as the common ancestor of the Crocodiles 
and Theriodonts, so that there is good reason for believing that 
the common ancestor had a well-developed proatlas and probably 
an atlas and axis very similar to those found in later forms, 
but almost certainly with the odontoid process as a distinct 
element, as seen in Jcehthyosaurus. No proatlas has yet been 
detected in Pareiasaurus, but I have found one in Lystrosaurus 
(= Ptychognathus Owen). It will probably yet be found in most 
of the primitive reptilian types, 
12* 
