RECAPITULATION. 501 



of the movement of a fish in the water are not very definitely 

 known. But the selectionist must assume that the position 

 of the pelvic fins is a mechanical advantage, and if that is 

 the case the change of position would necessarily be pro- 

 duced on Lamarckian principles by the directions in which 

 the muscles were exerted. 



Among particular cases of metamorphosis in fishes none 

 is more remarkable or has attracted more attention than 

 that of the fiat-fishes. This has always been cited as a 

 conspicuous case of recapitulation, as it certainly is. But it 

 also illustrates very forcibly the principle here maintained, 

 that recapitulation of modifications of structure depends not 

 merely on heredity, but to a very important extent on reca- 

 pitulation of changes of condition. The pelagic larva of 

 Pleuronectes, Solea or Rhombus is very closely similar to 

 that of the Gadidai or of many Acanthopterygians. We 

 may well believe that this pelagic larval condition had been 

 acquired before the peculiar condition of the flat fishes had 

 been evolved. The larval symmetrical condition represents 

 therefore not the adult condition but the larval condition of 

 the ancestor. New conditions of life and consequent 

 modifications of structure have been introduced into the 

 adult stage of life, while the conditions and the structure of 

 the larva have remained unchanged. Whether the habit of 

 lying on the bottom on one side was originally acquired at 

 an early or a late age, it is certain that it has come to be 

 assumed at a very early age, and consequently the structural 

 changes which we know to be associated with it are now in 

 great part simultaneous with the ordinary development of 

 the adult structures of the fish. For this reason it is by no 

 means easy to ascertain from the ontogeny of the fiat fish, 

 much as the process has been studied, the characters of the 

 adult ancestors. The assumption of the special Pleuro- 

 nectoid characters is most retarded in species of Rhombus 

 and their allies, such as the turbot and brill. In these cases 

 we find stages in which the eyes are still symmetrical when 

 the fin-rays are well developed, and we see a symmetrical 

 fish with a morphologically heterocercal tail, a single dorsal 

 and a single ventral fin, neither extending so far forward as 



