ON THE ANATOMY OF FISHES. 
233 
that our observations were necessarily somewhat incomplete. The condition of the 
membranous labyrinth, and its relations to the cavum sinus imparis and to the atrial 
cavities, were investigated in a large number of Siluridee normales, but with the 
purely negative result that we could detect no variations of any importance from the 
arrangement of these structures, already described for Amiurus catus by Eamsay 
Weight, and for Silurus glanis by Webee, and in Macrones by ourselves. In 
all cases we found a transversely disposed ductus endolymphaticus connecting the 
two sacculi, and, attached to the ductus, a median pear-shaped sinus endolym- 
phaticus projecting backwards into, and almost completelj filling, the “ cavum sinus 
imparis.” 
With the exception of the intercalarium, the Weberian ossicles exhibit but little 
variety in shape, or in their relations to one another, or to the atrial cavities and air- 
bladder. The variations in the condition of the tripus relate principally to the degree 
and shape of the curvature of its posterior or crescentic process. In some genera 
[Aucheni'ptci'us, Oxydoras) the “ crescentic process ” is almost straight ; in others 
almost hook-shaped (Plotosus) ; and between these extremes the process may exhibit 
almost every degree of curvature. A ventral ridge on the root of the crescentic 
process, to receive the insertion of a slip of fibres from the adjacent anterior wall of 
the bladder, is very generally present, and varies in size according to the thickness of 
the walls of the bladder. In some Siluroids [Macrones, Liocassis) the outer convex 
margin of the process may be increased for the purpose of fibrous attachment, and 
possibly for leverage, by the addition of an outwardly "directed heel-like process, 
which occasionally may be of considerable length [e.g., Bagroides). The articular 
process of the tripus is usually distinct from the complex centrum, with which, how- 
ever, it articulates at the bottom of a deep pit-like depression. It is very rare, as in 
the two genera Oxydoras and Auchenipterus, for the process to be flexible and elastic, 
and directly continuous by an oblique origin with the anterior part of the neural arch 
of the complex vertebra, like the adjacent and similarly elastic root of the “ elastic- 
spring” apparatus. The proportional lengths of the anterior and crescentic processes 
vary somewhat in different forms ; generally, the two processes are of approximately 
equal length, but when otherwise it is the anterior which is the longer. 
The intercalarium varies greatly in development. Usually a small osseous nodule 
imbedded in the interossicular ligament, the intercalarium may, in addition, be pro- 
longed therefrom as a horizontal spicule which terminates in the fibrous wall of the 
neural canal, between the arch of the complex vertebra and the ascending process ot 
the scaphium, near the dorso-lateral margin of the anterior portion of the complex 
centrum, with which, however, it is in no way directly attached [Cryptopterus, 
Ccdlichrous). In a few genera [Macrones, Liocassis, Pseudohagrus, &c.) the hori- 
zontal process is prolonged upwards into a vertically disposed or ascending process, 
which also lies in the fibrous wall of the neural canal, behind and parallel to the 
ascending process of the scaphium. In all cases where an ascending ])rocess is 
MDCCCXCIII. — B. 2 11 
