1899.] BLOOD-VESSELS OP TELEOSTBAN FISHES. 943 



involving the second, and later the third and fourth efferent 

 branchial vessels ; and secondly, by the progressive suppression in 

 length of the median aorta, bringing about an approximation of the 

 dorsal or proximal ends of the last three efferent branchial vessels 

 on each side. 



Owing to the fact that the one variety of specialization may 

 occur quite independently of the other, or in conjunction with 

 it, it becomes very difficult in some cases to compare the ultimate 

 degree of specialization attained, since there is no evidence to show 

 whether the inclusion of the efferent vessels into the circulus or 

 the approximation of the efferent vessels by the suppression of the 

 aorta is the more important. In the Salmon (fig. 7), for instance, 

 the second efferent vessel opens into the circulus cephalicus — an 

 indication of speciahzation ; but a length of aorta persists between 

 the circulus cephalicus and the third efferent vessel — a primitive 

 character. In Balistes (fig. 5) the second vessel is free from the 

 circulus cephalicus, and yet there is obvious specialization in the 

 complete suppression of the aorta in the branchial region, resulting 

 in the second, third, and fourth vessels opening close together, 

 immediately behind the circulus. Who shall say whetlier, in the 

 disposition of the efferent branchial vessels, the Salmon or the 

 File-fish is the more primitive ? Having recourse to the other 

 anatomical features of these two forms, one would conclude that, 

 the Salmon being in general structure the more primitive, the 

 abbreviation of the aorta is as a mode of specialization more 

 important than the backward extension of the circulus to include 

 the second efferent branchial vessels. The conclusion is further 

 justified by the fact that Albula (fig. 11), which is undoubtedly 

 allied to Megalops (fig. 4) and Ohiroceiitrus, differs from these 

 genera in this latter respect. 



This hypothesis, however, opens up the further question as to 

 how far a backward extension of the circulus cephalicus is due to 

 the longitudinal splitting of a part of the median aorta. Has, for 

 instance, the condition found in Megalops smd Chiroceutrus, in which 

 the circulus cephalicus extends back to the second branchial vessels, 

 been brought about by the longitudinal division of a median vessel 

 such as exists in Olupea (fig. 2) and Engraulis (fig. 1) between the 

 first and second efferent branchial vessels? The suggestion has 

 much to recommend it ; more especially as the suppression of 

 the median aorta cannot have operated here, or the two anterior 

 carotids would be arising close together at the bottom of the fork 

 of the first efferent branchials. 



Another line of specialization, independent of the two former, 

 can be traced in the confluence ot the third and fourth efferent 

 vessels. Having assumed that the separation of two consecutive 

 efferent branchial vessels by a portion of the median aorta is 

 a primitive feature, it follows that the separate entry into the 

 aorta of the third and fourth vessels in Engraulis (fig. 1) is an 

 indication of less specialization than the debouching of the two 

 vessels together, as in Clupea (fig. 2) ; and further, that this latter 



61* 



