NATURAL SELECTION 303 



food range than we have of the origin of completely new food 

 habits. It would probably be admitted that the first stage in 

 the former process was the formation of biological races 

 within the species, although we can, perhaps, never prove that 

 those races are not the result of a definite change in instinct 

 (as described in (a)) ; yet comparison with allied forms suggests 

 rather that a species of generalised habits has become more 

 specialised. We have dealt above (pp. 301-2) with the 

 question of the origin of biological races. For our present 

 purpose only two aspects of the problem concern us. First, 

 the instinctive basis of racial specialisation does not usually 

 appear to be hereditary, at any rate in its early stages. The 

 female returns to lay her eggs on the substratum on which the 

 larvae fed, and her response can be altered in a few generations 

 by experimental restriction to a different food : her reactions 

 may be due to the retention of a ' larval memory ' (Thorpe, 

 1930, p. 202) rather than to hereditarily fixed instincts. 

 In these circumstances the most that can be claimed for 

 selection is that it has favoured those species endowed with 

 the power of ' larval memory ' ; it has not been active in the 

 initial stages of the formation of biological races. Secondly, we 

 must consider how far each biological race is adapted to its 

 food. A certain confusion is liable to be introduced here by 

 the ambiguous use of the word ' adaptation ' (cf. Chapter IX). 

 Some authors have spoken of a race A as being ' adapted ' to 

 a particular host B, when meaning no more in reality than 

 that A is restricted to B. The true use of the term, however, 

 can be illustrated by considering a species with two races 

 A and A 1 , restricted to two hosts B and B'. These races are 

 spoken of as adapted to their hosts, only if in each case some 

 part of their structure or physiology makes each one better able 

 to live on its own host than on that of the other, so that not 

 only is each race restricted to its own host in nature, but that 

 A, transferred to B 1 , would be at a definite disadvantage com- 

 pared with A 1 . Unfortunately we have not nearly enough 

 evidence on this point. What evidence we have does not 

 suggest that there is necessarily a definite adaptation to the 

 preferred host. When one race is transferred to the food of 

 another, it is true that there is often (perhaps usually) a con- 

 siderable mortality. But a considerable number frequently 

 survive and appear from then onwards to be physiologically 



