ON THE ANATOMY OF FISHES. 
95 
of the chamber ; from thence they converge in the corresponding half of the dorsal 
wall, in the form of a broad triangular sheet, to their ultimate attachment to the 
tripus. The fibres of the outer stratum of the tunica externa in the mesial portion of 
the anterior wall, if traced ventrally from the anterior pillars and their skeletal 
attachments, become easily separable from those of the inner stratum, and ultimately 
extend into the ventral wall, where they assume a direction which, in the main, is 
longitudinal or oblique ; but this arrangement is somewhat modified by their tendency 
to describe curves and loops, and to become interwoven Avith one another. It is 
worthy of note that the attachments of the anterior and posterior pillars to rigid 
portions of the skeleton will not only render the anterior chamber inexpansible in the 
direction of its antero-posterior axis, but will also tend to limit, even if it does not 
entirely prevent, any expansion ventrally, inasmuch as the longitudinally-arranged 
fibres which form the inner layer of the tunica externa in the mesial portion of the 
ventral wall are directly continuous between the two pillars, and also share their 
respective skeletal attachments. In not a few Siluroids {e.g., Arius) these longitu- 
dinal fibres are unusually strongly developed, and form either a single ridge or a pair 
of ridges, which project from the median line of the ventral wall into the cavity of 
the chamber, and anteriorly and posteriorly are directly continuous with the anterior 
and posterior pillars. 
(c.) Over a somewhat lozenge-shaped area in the median portion of the dorsal wall 
of the anterior chamber, bounded in front and behind respectively by the anterior and 
posterior pillars, and laterally by the oblique posterior margins of the two triangular 
sheets which constitute the lateral portions of the dorsal wall, the tunica externa 
becomes so extremely thin as to allow the sub vertebral keel to be readily seen 
through it Avhen the tunica interna has been removed (figs. 19 and 20). Although 
extremely thin in this region the tunica externa is not free, but, on the contrary, is firmly 
adherent, like a periosteum, to the under surface and sides of the complex centrum 
and part of the body of the fifth vertebra, its area of attachment exactly coinciding 
with the extent of the superficial ossifications that invest the centra. It is scarcely 
necessary to mention that apart from the adhesion of the medio-dorsal portion of the 
anterior chamber to the skeleton, the close investment of the lateral portions of the 
dorsal wall by the expanded transverse j^rocesses of the fourth and fifth vertebrae 
must preclude all possibility of expansion or contraction in that direction. 
(2.) Attachments to moveable ossicles. 
(a.) The fibres forming both strata of the tunica externa in the lateral and antero- 
lateral walls of the anterior chamber converge as they pass into their respective 
halves of the dorsal wall towards the point a little to the outer side of the complex 
centrum, and there become inserted into the crescentic process of the tripus. The 
fibres constituting the inner stratum in these regions are vertically arranged ; traced 
thence into the ventral wall they pass more or less obliquely backw'ards, and 
eventually are traceable into continuity with the more laterally situated of the 
