MAMMAL-LIKE REPTILES. 783 
It will be convenient here to leave this particular line of 
argument, and starting from the normal conditions of the mem- 
brana tympani in Reptiles, try to reconstruct the arrangement in 
Dimetrodon, and so work back to the termination of our original 
argument at the Deinocephalia. 
In Lizards the insertion of the tympanic membrane is on to 
the back of the quadrate, the squamosal just outside the par- 
occipital process, and the retro-articular portion of the lower jaw. 
In Pelycosaurs and other Therapsids the retro-articular portion 
of the lower jaw is of insignificant proportions, at its largest in 
Dimetrodon. In Deinocephaha and Dicynodon it is represented 
solely by a small process from the articular directed as much down- 
wards as backwards. The definite presence of this process in these 
more primitive forms and its still further reduction in later types, 
seem to show that some Therapsid ancestor had a relatively 
large retro-articular process. As in all reptiles which have such 
a process its outer side is covered by the angular, we may assume 
that in this hypothetical ancestral Therapsid the retro-articular 
portion of the angular played some part in the support of the 
tympanic membrane. If now in such an animal we make the 
tubo-tympanal cavity grow forwards and the membrane to keep 
pace with it in such change, we must also move forward the upper 
edge of the retro-articular part of the angular to which the 
membrane is attached. As it is essential to keep the stretched 
membrane clear of other bones, and the articular and quadrate 
are by hypothesis fixed, we can only do so by separating the edge 
of the retro-articular part of the angular from other bones and 
moving it slightly outwards: this on the theory which I am at 
present expounding is the origin of the Therapsid notch, the 
upper edge of the reflected lamina being phylogenetically the 
upper border of the retro-articular portion of the angular, and 
retaining that connection with the lower edge of the membrana 
tympani which occurs in Lizards. By this method we arrive at 
a, position of the tympanic membrane in Deinocephalia identical 
with that deduced by tracing the conditions down from Cyno- 
gnathids, the arrangement in which was determined by assuming 
that the great resemblance between the angular of these types, 
with their long, slender, downwardly directed “ reflected lamina,” 
and the tympanic of a pouch-young of Perameles with its slender 
lower limb, was a real one. The fact that it is possible to trace a 
sequence of stages with only very few hypothetical intermediates 
between the actual condition of the membrana tympani in lizards 
and that in embryo mammals, and that these stages are in the 
correct time order, and are each self-consistent, seems to me to 
establish a probability that the assumption on which they depend, 
i. e. that the reflected lamina of the Therapsid angular carried the 
tympanic membrane, is a justifiable one. 
T am quite aware that the shape of the edge of this lamina in 
some types, as for example in Hndothiodon and Anomodonts 
generally, is inconsistent with the view that it carried the 
. 
