ON THE ANATOMY OF FISHES. 
165 
but a pair of rudimentarj processes which are widely separated from those of the 
fourth and sixth (Zoc. cit., Plate 2, figs. 14-16). 
The air-bladder is partially constricted into two portions, of which the anterior is 
cordate in shape, and subdivided internally into the usual anterior and lateral com- 
partments by transverse and longitudinal septa {loc. cit., Plate 4, fig. 43). The posterior 
portion {loc. cit., Plate 3, fig. 44), is a long internally sacculated caecal appendage, closely 
resembling the corresponding structure in Malapterurus. Unlike all other Siluridm in 
which an “ elastic- spring ” mechanism is present, the anterior chamber is fringed on each 
side by a series of variously-sized more or less branched, tubular caeca. Over the circular 
or somewhat oval areas, where the terminal discs of the two elastic springs are applied 
to the walls of the anterior chamber, only the thin, delicate, transparent tunica interna 
is present. On this point, and with reference to the mode of formation of the discs, 
Sorensen {loc. cit., p. 139) remarks : — “ Chez le genre Doras, il* se termine en un disque 
rond et epais qui est une ossification de (la plevre et de) la membrane externe de la 
vessie natatoire dans toute son epaisseur, de sorte qu’en ouvrant la vessie natatoire on 
aper 9 oit ce disque a travers la membrame interne pellucide.” According to Sorensen, 
therefore, the thickened discs in which the transverse processes of the fourth vertebra 
terminate distally are to be regarded as owing their formation to the extension of 
ossification from the transverse processes into the transverse membrane (la plevre), 
and the whole thickness of the tunica externa. 
The crescentic process of the tripus {loc. cit., fig. 15) has apparently the same shape 
as in Auchcnipterus. 
Oxydoras brevis. 
We are not acquainted with any previous reference to the air-bladder of Gxydoras, 
nor are we aware that the presence of an “elastic sprmg” apparatus in this genus has 
hitherto been recorded. 
In all essential features the modified anterior vertebrse and their processes are 
similar to those of Auchcnipterus nodosus. The transverse processes of the fourth 
vertebra are modified to form an “ elastic spring ” mechanism. The deflected distal 
portion of each process (fig. 64, t.p.'^) forms, however, but a slightly expanded and 
inwardly curved plate with a thickened outer margin and an almost nodular extremity. 
Near the junction of each plate with its elastic and flexible root there is a slight groove 
into which the outer margin of the crescentic process of the tripus is received. As in 
Auchcnipterus, the modified transverse process has no trace of a posterior division, 
but, unlike that genus, the fifth vertebra has scarcely any trace of transverse processes, 
and, as is usually the case where the lateral surfaces of the body are protected by an 
armature of dermal bony plates (e.^., the Loracaroid Siluridoe), the first pair of ribs are 
exceptionally massive. The radial nodule (r.n.) is an elongated spicule of bone, some- 
what loosely attached by fibrous tissue to the lateral surface of the complex centrum. 
* The ti’ansverse process of the fourth vertebra. 
