152 
PROFESSORS T. W. BRIDGE AXD A. C. HADDOX 
wall is firmly attached to the distal portion of the ventral process ; and (iv. ) the 
inner wall is similarly adherent to the dorsal lamina and its horizontal prolongation. 
In the condition of the dorsal walls of the two air-sacs Ghjiotosternum exhibits an 
even more degenerate condition than what we believe to be the case in Bagarius, 
with the possible exception of a thin stratum of connective tissue investing the inner 
surface of the dorsal wall of each bony recess, which may represent the degenerate 
dorsal wall of the enclosed air-sac, the latter does not appear to be indicated by any 
definite stratum of fibres. We could detect no fibres derivable from the outer or 
anterior wall of the sac and converging in the dorsal wall towards the crescentic 
process of the tripus, although this ossicle retains its normal relations to the rest of 
the sac, and is easily seen in the usual position when the ventral wall has been 
removed. The dorsal edge of the anterior wall is wholly attached to the anterior 
margin of the tubular transverse process, and no fibres could be traced therefrom into 
the dorsal wall ; an outer stratum of fibres in the posterior wall is similarly attached 
to the posterior margin of the same process, but the inner stratum thins away and 
disappears as a traceable layer without extending into the dorsal wall, as also is the 
case with the whole thickness of the outer wall. The inner wall, on the contrary, 
remains intact, and a few fibres are traceable therefrom to an insertion into the feebly 
concave inner margin of the crescentic process of the tripus. The latter fibres 
undoubtedly represent the remains of the characteristic radial fibres of other 
Siluridge, and share with the remainder of the inner wall a fixed insertion into the 
dorsal lamina. In some instances we were able to detect a few fibres at the inner 
and posterior portion of the sac, which at one extremity were inserted into the distal 
extremity of the crescentic process of the tripus, and in the opposite direction were 
traceable into the adjacent portion of the posterior wall. It is possible that in such 
cases these fibres represent the sole remains of the triangular sheets of a normal air- 
bladder. In our description of a young Bagarius we referred to the existence of a 
few scattered fibres radiating outwards from the crescentic process of the tripus, 
forming an imperfectly fibrous dorsal wall to the air- sac, and subsequently extending 
into the anterior and antero-lateral walls. These fibres were so feebly developed 
that it is almost impossible to regard them as having any functional significance, and 
in older and more mature specimens, it is by no means improbable that they entirely 
atrophy, leaving each air-sac in a condition precisely similar to those of Glyptosternum. 
We could detect no trace of any communication between the two air-sacs, or of 
the existence of a ductus pneumaticus. 
The tripus is a tri radiate ossicle, its three divergent divisions representing the 
anterior, articular, and crescentic processes (figs. 55 and 58, tr.). The anterior and 
articular processes (fig. 58, tr.a., tr.ar.) are similar in shape, and of approximately 
equal length, and both taper towards their distal extremities. The crescentic process 
(tr.c.) is slightly longer than either of the other two processes, and although some- 
what similar in shape has a faint inward curvature, but without forming the charac- 
