60 ME. W. S. EOWNTEEE ON THE 



disparity in size is less marked, though still considerable. The tube connecting the two 

 is not axial, as in most Characiuidae, but at the ventral border of the bladder. The 

 ductus jDneumaticus arises from this tube so obliquely that it appears to spring from the 

 anterior chamber ; in reality its commvmication is as usual with the posterior chamber. 

 The condition in respect of the relations of these tubes is somewhat suggestive of that 

 inet with in the Gymnotidae, where, however, the two chambers of the bladder are much 

 farther ajoart. The posterior chamber in Serrasalmo is concave in front, fitting over 

 the rounded oviform anterior chamber in the manner of a ball-and-socket joint. The 

 margin of its concave front face is raised on each side into some five or six small caecal 

 appendages, corresponding to internal pouches. A thick sheet of muscle on each side of 

 the anterior chamber connects the outer tunic with the dorsal body-wall. Some of these 

 features are shown in PI. 3. fig. i, together with internal characters to l)e described later. 

 This peculiar air-bladder has been briefly described by Valenciennes, but only as regards 

 its most obvious external chai'acters ; its marginal appendages and its internal structure 

 are not mentioned by this writer. It is shown partially dissected in fig. 4. In 

 Anacyrtus also, in which the body-cavity is short and deep, the air-bladder is peculiarly 

 shortened and rounded, its entire length being only one and a half times its depth. The 

 posterior chamber is, however, as usual, much the larger. Internally, also, there are 

 points of interest (PI. 3. fig. 8). 



In the diminutive Nannoeharax niloticus the bladder is much reduced. It does not 

 extend more than half the length of the body-cavity. There is the usual division into 

 two chambers, of which the posterior is considerably the longer, but the cavities appear 

 to be largely obliterated. The anterior chamber is globular, with hard tough walls ; I 

 believe, indeed, that it is partially ossified, but its diminutive size renders its condition 

 doubtful. The posterior chamber is narrowed almost to a filament. The ductus 

 pneumaticus, though fine, has its usual relations. In no other Characinid have I met 

 with so rudimentary a condition. 



Internally, the walls of the anterior chamber of the air-bladder are plain and smooth 

 in all the Characinidae. With the posterior chamber, however, it is otherwise : its walls 

 are often ridged by longitudinal bands of a ligamentous nature, some of which may be 

 of considerable depth, so as to appear like septa dividing the cavity peripherally into 

 longitudinal furrows or pouches. Most of the septa are restricted to the anterior part 

 of the chamber, becoming rapidly shallower backwards, and finally disappearing after a 

 comparatively short course. There are, however, usually tAvo, one dorsal and one 

 ventral, which extend the entire length of the chamber so as to meet one another at its 

 posterior extremity ; but these also become shallower as they proceed backwards. The 

 others are more or less symmetrically arranged laterally, but in varying numbers and 

 varying degrees of development. In a few cases such bands as are visible are but slightly 

 raised, the wall of the chamber apjiearing nearly smooth. In only about two cases, 

 however, have I observed the dorsal and ventral median bands to be absent or much 

 reduced in length : in the Ichthyoborince the two chief lateral septa extend the whole 

 length of ttie chamber, exactly like the dorsal and ventral pair, excejiting that they are 

 somewhat less prominent. 



