AIR-BLADDEB IN NOTOPTKKTJS BOENEEXSIS. 509 



the dorsal and ventral walls are also firmly attached to the 

 skeleton. Along its inner or mesial dorso-lateral margin, where 

 the relatively thick dorsal wall becomes continuous witli the 

 attenuated iuner wall, the tunica externa becomes somewhat 

 thickened and terminates by becoming firmly attached to the 

 bases of the haemal spines and also to the dorsal extremities of 

 the two radial elements with which each spine is associated. 

 Yeutrally also the tunica externa thickens and forms a series of 

 strong transversely disposed ridges projecting from the floor of 

 the cajcum, and separating the orifices leading into the ventral 

 diverticula. Traced towards the inner wall of the caecum, the 

 ridges cease by becoming inserted into the adjacent sides of two 

 contiguous radial elements and to the connecting fibrous mem- 

 brane which extends between them. On the other hand, if traced 

 ventrally, the fibres of each ridge split to form the povsterior w^all 

 of one diverticulum and the anterior w'all of the next succeeding 

 diverticulnm, both walls nevertheless retaining their mesial 

 attachments to the contiguous radial elements between which 

 the diverticulum is situated. 



Near the dorsal wall of each caudal caecum, the fibrous sheet 

 between the radial elements presented a series of oval vacuities 

 (d.v.), which were disposed in regular order between the succes- 

 sive pairs of radial elements which are attached to the vertebral 

 haemal spines. Over most of these vacuities the thin inner wall 

 of the caudal caecum appears wanting, and hence the cavity 

 of each caecum seems to communicate with that of its fellow 

 at these points. In one or two instances, however, it is 

 clear that no such communication exists, for, notwithstanding 

 the vacuity in the fibrous sheet, the opposed inner walls of the 

 two caeca remain intact over the area of the vacuity. It is 

 difficult, therefore, to be quite certain that these vacuities are 

 associated with normal perforations in the opposed walls of the 

 caeca, or that tlie cavities of the latter really intercominunicate 

 during life, or in perfectly fresh specimens, more especially as, 

 owing to its thinness and fragile character, the tunica interna is ex- 

 tremely likely to break down and disintegrate where unsupported 

 by the much stronger and more resistant fibrous sheet, unless 

 more than ordinary care is taken with the preservation of the Fish. 



A similar series of small, oval, or rounded vacuities (v.v.) in 

 the interradial fibrous membrane was also present towards the 

 ventral side of each caudal caecum, where the membrane in 



