VISCERAL ANATOMY OF THE CHAEACINIDJil. 63 



superficial. In its first-mentioned characters this air-bladder faintly suggests that of 

 Sermsalmo, which it also resembles in the somewhat vinvisual thickness of its walls. 



Upon the question wliether or no the internal features of the air-bladder summarized 

 in the preceding pages possess any real significance bearing on the affinities or classi- 

 fication of this puzzling family of fishes, it is difficult to speak with confidence. Certain 

 facts, however, seem to be in some degree suggestive. Thus, of the Erythrinoids, two 

 forms — Erijtlirinus and Lehlaslna — present a similar cellular structure, but in different 

 stages of development or degeneration, whilst 3Iacroclon appears to follow suit after a 

 much greater interval. Of the I'ourth form — Fi/rrlmlina — I have no satisfactory 

 observation. Again, the members of the group Ichthyohorince agree, so far as they have 

 been examined, in a unique extension of the two main lateral septa. Also the allied 

 African mud-eating fishes — Disticlwdvs, Xenocharax, and Citharimis—Tpresent conditions 

 similar to one another. 



The resemblances above pointed out are sufficient to inspire some confidence in the 

 persistence and constancy of the structural features of the air-bladder ; and it seems not 

 unlikely that the examination of that organ in a larger number of forms might here and 

 there yield evidence of value as an assistance in classification. I do not, however, feel 

 justified in proceeding further with any such deductions from the slender material 

 before me. The groupings found in my description may be in some degree accidental. 

 Perhaps the most noteworthy points are :— (1) the isolated and aberrant conditions in 

 Serrasalmo (PI. 3. fig. 4), and to a less degree in Anacyrtits (fig. 8) ; (2) the fact that, of 

 the African forms, Sarcodaces seems to stand nearest to the archaic Erythrinoids ; and (3) 

 the agreement of the IclithyohorincB in possessing two more completely developed lateral 

 septa than are found in any other forms. 



In conclusion, it is worthy of mention that in Cyprinoids I have observed two 

 longitudinal bands in the posterior air-bladder, which are visible as streaks, but not 

 appreciable to the touch, tlie surface being quite smooth. This condition sugo'ests a 

 further stage of degeneration from a primitive cellular air-bladder. A still further step 

 seems to be recognizable in the Gymnotidte, in which these bands are absent. 



The Ductus Pneumaticus and its Relations in the Phtsostomi generally. 



The ductus pneumaticus has already received a brief mention. There remain, 

 however, certain points in connection therewith which merit consideration. 



The duct takes its origin in all the Characinidse from the posterior chamber of the 

 air-bladder, which is thus placed in communication with the alimentary tract in the 

 transitional region between the oesophagus and the stomach. In some cases the 

 opening appears to be in the stomach, in others in the oesophagus, and in others on 

 the border line between the two regions. This border, it is true, is sometimes difficult 

 to define ; but, on the other hand, there may be a sudden change of calibre or consistency 

 which one cannot but interpret as marking the transition, and this may be variously 

 situated in relation to the opening of the duct. 



In Physostomes generally the duct opens dorsally, and I am not aware of anything in 



SECOND series. — ZOOLOGY, VOL. IX. 10 



