VISCERAL ANATOxMY OF THE CHAEACINID^. 61 



It may well be that the structure above described serves to maintain this chamber of 

 the bladder, winch is comparatively thin-walled, in a state of due disteusion, and to 

 guard against its collapse. Svich, indeed, one cannot but suppose, must be its effect. 

 But that it may also possess a vestigial significance is suggested by a further very 

 curious modification, which is met with in the genera Lebtaslna and Erythrinm. In 

 these two forms, and in these only amongst the Characinidae, the longitudinal septa are 

 connected for a certain distance by transverse septa, in such a way as to constitute a 

 true cellular air-bladder — a suggestion of Ganoid affinities which harmonizes with 

 deductions from the cranial characters. This structure, which is found only in the 

 anterior third or two-fifths of the chamber, has been described by Valenciennes for 

 Erijthrinus as follows * : — " La vessie posterieure est conique, pointue ; sa tunique 

 fibreuse est plus confondiie avec la tunique interne, et on y remarque quatre brides 

 longitudiuales : une superieure, une inferieure et deux laterales. Le tiers anterieur de 

 cette seconde vessie offre des parois celluleuses ; les cellules sont determinees par des 

 brides trausversales nombreuses, serrees, paralleles entre elles, et j)erpendiculaires aux 

 grandes brides tangentes a la surface du cone. Ces brides elles-memes sont reunies par 

 d'autres phis petites, excessivement plus nombreuses et perpendiculaires aux trausversales 

 que nous venons d'indiquer. Entre ces mailles on aper^oit de nombreuses lamelles 

 entrelacees, auxquelles est due la cellulosite des parois de la vessie." 



Of the bladder of Lebiashia this writer simply says that it resembles that of 

 Erythrinus. My own observations, however, lead me to state that the cellular structiu-e 

 is distinctly more pronounced in Lebiasina than in Erythrinm. Figs. 5 and 6 (PI. 3.) 

 show that in the former the cellular part of the wall is of some thickness, materially 

 narrowing the lumen of the chamber, whilst in the latter the bands or septa are com- 

 paratively slightly raised, and do not appear to encroach appreciably on the central cavity. 

 It is remarkable that in Jlacrodon, which otherwise so strongly resembles Uryihrmus, 

 and in Pyrrliulina, which seems i-ather to be related to Lebiasina, there are none but 

 the longitudinal septa (fig. 7). But in Jlacrodon, at least, these latter are strongly 

 suggestive of cellular structure. 



Regarding the resemblance in detail presented by this type of air-bladder to that 

 of Amia, I can express no opinion, and observers are at variance. Sagemehl f considers 

 the similarity to be closer than in the case of any other Teleost ; whilst, according to 

 Valenciennes :i:, it is not so great as in some species of JHemirhamphiis. 



In all the other Characinids, so far as they have been examined, cellulatiou does not 

 exist, transverse septa being entirely absent. Nevertheless, the longitudinal sejsta, 

 constantly present but presenting varying stages of development or degeneration, are, 

 as it seems to m.e, suggestive of a former cellular condition of the bladder. In the short 

 statement given below, the order represents, according to my observations, the degree of 

 development of the longitudinal septa, 



* C'uvier & Vuleucionnes, ' Histoire Xaturelle des Poissons,' vol. xix. p. 4S5. 

 t Sagemehl, Morphol. Jahrb., Bd. x. 1884-5, p. 108. 



* Cuvier & Valenciennes, ' Hist. Xat. des Poissons,' vol. xix. p. 493. 



