138 



apparently this anastomosis, shifted posteriorly onto the lateral dorsal 

 aorta, that persists in the adult elasmobranch. It accordingly seems 

 probable that this anastomosis of the mandibular aortic arches will 

 be found to also exist in early, embryonic stages of Amia and 

 teleosts. Assuming this to be the case, and assuming this anastomosis 

 to disappear immediately after hatching, as it does in Acipenser, the 

 later anastomosis of the lateral dorsal aortae in certain teleosts 

 might be accounted for by a simple enlargement of the orbits. 

 For if such an enlargement were to take place, in a fish like Amia, 

 the internal carotids of opposite sides, as they traversed the myo- 

 dome or its orbital opening, would be pressed backward and mesially, 

 carrying with them the dorsal ends of the efferent mandibular 

 arteries, and if they were to be pressed together in the median line 

 and there fused with each other this fusion would quite inevitably 

 take place in the region where the carotids are joined by the efferent 

 mandibular arteries. By slight changes the dorsal ends of these latter 

 arteries would then fuse with each other and become separated from 

 the internal carotids, and the teleosteau condition would arise ; the 

 conditions in Gadus, in particular, suggesting this origin. If this be 

 the origin of this teleostean condition, it is evident that this myodomal 

 anastomosis of the internal carotids (lateral dorsal aortae) in certain 

 teleosts is not homologous with the anastomosis found in elasmobranchs, 

 although occurring in what seems strictly the same region, for in 

 elasmobranchs the anastomosis is said to be of very early, if not of 

 primary origin, while in teleosts it is of later, and probably of 

 secondary origin. This secondary anastomosis, if it be such, would 

 then seem to be a strictly teleostean characteristic, but it is not found 

 in all teleosts, for I did not find it in larvae of Ameiurus (Allis, 

 1908 b), and it is not found in a large adult specimen of Silurus 

 glanis that I have had examined for this purpose. Ameiurus and 

 Silurus are accordingly ganoidean in this respect. Ameiurus is 

 furthermore transitional between ganoids and teleosts in the manner 

 of innervation of its eye muscles, and it agrees with Lepidosteus in 

 possessing a rudimentary myodome that does not, and probably never 

 did communicate with the orbit (Allis, 1910). 



The arteria ophthalmica magna, as given in existing descriptions 

 of adult fishes, is always shown either as a dorsal branch of the ef- 

 ferent mandibular (pseudobranchial) artery (elasmobranchs), or as 

 representing that entire artery together with a direct anterior pro- 

 longation of it that extends to the eye-ball (teleosts, Amia). In young 



