RECAPITULATION. 499 



wanting in the Gadidae. In the latter the tail is per- 

 manently diphyceral, it is composed of dorsal and ventral 

 rays which are equal in number and size, and in fact closely 

 resembles the tail of the extinct Coelacanthidse, There can 

 be little doubt that even if the Gadidse cannot be direcdy 

 derived from the latter family they are descended from 

 Crossopterygian Ganoids with diphycercal tails, and have 

 never passed through a heterocercal condition. Although 

 the structure of the tail in the Gadidse was briefly and 

 correctly described by Alexander Agassiez in his paper on 

 the " Development of the Tail," he did not attach sufficient 

 importance to it, believing that a very slight apparent up- 

 bending of the termination of the notochord showed the 

 essential similarity in the development of this type of tail 

 with that seen in other Teleosteans. The proper classifica- 

 tion of the Anacanthini is yet to be worked out, but there 

 can be no doubt that the Gadidse and Pleuronectidae, instead 

 of being closely allied, are very remote from each other in 

 structure and descent. 



The recapitulative development of the tail in many 

 Teleosteans by no means proves that the larval condition 

 as a whole in those forms represents the condition of an adult 

 ancestor. We do not find that the scales in development 

 pass through an Elasmobranch and then a Ganoid stage. 

 On the contrary, they are developed directly. There can 

 be little doubt that the larvce hatched from pelagic eggs, 

 as well as the eggs themselves, are a direct adaptation to 

 the conditions of life of marine fishes. It is not possible 

 at present to say what are the conditions which tend to 

 increase the number and diminish the size of eggs, but 

 pelagic eggs are smaller and more numerous than those 

 which are attached to solid objects in fresh water or in the 

 littoral region. The pelagic larva presents many peculi- 

 arities which are obviously not ancestral, among them the 

 fusion of the veins of the yolk-sac into one continuous cham- 

 ber surrounding the yolk and communicating with a wide 

 opening at the hinder end of the heart. This may be 

 explained without difficulty on Lamarckian principles as 

 the result of the small quantity of mesoblast available at the 



