214 PROF. D. M. S. WATSON AND MR. E. L. GILL ON THE 
the bone behind the squamosal. So far as the skull-roof is concerned, Uro- 
neimts might be derived from Ctenodiis, Conchofioma from Sagenodus. 
In the palate the two genera differ very widely. Uronemus has a small 
parasplienoid, with a long posterior stem and a lozenge-shaped palatal part 
completely devoid of teeth. The pterygoids form an exceptio'nally large 
part of the palate, have a single series of large teeth along their outer margin, 
and an area covered with small teeth which show no trace of arrangement in 
radiating ridges. Concliopoma has a unique parasplienoid with a long, slender, 
posterior stem, and an enormously enlarged palatal part extending far forward 
and covered with an irregular development of small denticles. The ptery- 
goids form narrow slips along the lateral borders of the parasphenoid, bear 
only a few snaall teeth, and present no trace of the enlarged marginal teeth 
of Uronemus. 
Both forms are unexpectedly primitive in retaining dentaries in the lower 
jaw, in addition to those splenials which have usually been called by that 
name in Ceratodus. 
It now falls to be consi<lereJ whether all the ancestors of these fish had 
only isolated denticles, or whether they have arisen separately by the breaking 
up of dental plates. 
Semon showed that in development the dentition of Ceratodus begins as a 
series of isolated denticles supported by a net-work of bony spicules, and that 
the tooth-plates of the adult are built up byithe confluence of such denticles. 
This mode of development is consistent with the view that Uronemus may 
have arisen from a form with tooth-plates, because its isolated denticles 
mav merely result from the carrying on to adult life of a structure which 
occurred in larval stages. It is probable that a stage with distinct unfused 
denticles formed a larger part of the life-history in early Dipnoi than it-does 
in Ceratodus ; indeed, the small plate figured by Pander as D. tuherculatus 
(Taf. 5, figs. 20-21) seems actually to consist of individual denticles placed 
on a bony base. 
The whole structure of the Dipnoan skull, the short mouth and forwardly- 
directed quadrate, the rigid attachment of tiie pterygoids to the basis cranii, 
and the reduction of the hyomandibular, all point to the Dipnoan stock being 
specially modified (or the use of a highly-developed crushing dentition. All 
these changes can be paralleled in those other groups of fishes which have 
developed analogous tooth-plates. 
Thus we believe that a typical tooth-plate like that of Diptems was probably 
of very early introduction into Dipnoan structure, its production having in 
fact gone on jxiri passu with the other correlated changes in the head. 
We are thus led to believe that Uronemus and Concliopoma are derived 
from fish which had tooth-plates, and as no such fish either has or needs a 
parasphenoid dentition, that the well-developed denticles on that bone in the 
latter fish are new developments. 
