MAMMAL-LIKE REPTILES, 781 
directly derived from the connection of the distal end of the 
hyomandibular and the quadrate in its far-off fish-ancestor, and 
that this connection has never been lost in the phylogeny of the 
group. 
The most primitive lower jaw known amongst Therapsids, that 
of Dimetrodon, as 1t has been described and figured by Case, 
Williston, and Broom, is differentiated from that of a member of 
any other Reptilian supererder by the characters of its posterior 
part, and particularly the angular. In Dimetrodon the angular 
is essentially a flat plate whose upper border overlaps the sur- 
angular and is itself overlapped by the dentary in the usual way, 
whilst its lower border forms the lower border of the jaw. The 
posterior end of this bone is separated from the outer side of the 
prearticular by a notch, so that the border stands out freely. 
The whole jaw is very narrow from side to side, the Meckelian 
vacuity being reduced to a very narrew slit between the pre- 
articular and the surangular. The flat plate-like angular and 
the notch in its posterior border are the characteristic features of 
this jaw, and are found in all Therapsids which are properly 
known and in no other reptiles whatsoever. 
The Deinocephalian jaw only differs from that of Dimetrodon 
in the further enlargement of the notch, which in this type 
extends forwards so as to form a deep pocket in the substance of 
the angular. In such Therocephalians as Seylacosaurus the con- 
dition of the angular is almost exactly as in Deinocephalia, there 
being a deep but very narrow pocket in the substance of the 
angular which forms a large reflected lamina outside it. The 
foramen shown in the angular in many of Broom’s figures of 
Therocephalian skulls has no existence in fact, as such, but is 
merely the anterior end of the notch. 
In the Gorgonopsids there is the usual notch and reflected 
lamina, which is, however, small, never reaching the size of that 
of Scylacosaurus. This notch in the higher Gorgonopsids, e. g. 
Arctognathus, seems to move relatively further forward. Finally, 
in the Cynognathids, which have a “‘ look” of the Gorgonopsids, 
of a character that cannot be intelligibly expressed in words but 
is very striking to anyone accustomed to handling the two types, 
and suggests that they are in all probability genetically related, 
the notch has moved still further forward, and the reilected 
lamina now stands downwards from the rest of the bone as a 
very slender process. The extraordinary resemblance of this 
bone to the tympanic of the pouch feetus of Perameles (or of 
Dasyurus) seems to me to raise a very strong case for believing 
that Cynognathus and its allies had a tympanic membrane spread 
between the divergent branches of its angular. ‘The actual shape 
of the bone makes such a, position possible, the hinder end of the 
membrane being carried by the ridge on the lower side of the 
squamosal, which lies just outside the end of the paroccipital 
process, and on the musculus depressor mandibuli, if such a 
muscle be present. A membrane in this position stands nearly 
53* 
