NATURAL SELECTION 253 



abundant evidence, chiefly published by Poulton (Proc. Ent. 

 Soc. London, passim), that such attacks occur, so that it is 

 impossible thereby to dismiss the subject offhand. When we 

 consider the nature of the evidence the problem becomes more 

 difficult. Some of it has been derived from the experiments 

 on birds in captivity, but it is generally admitted that the 

 reactions of birds in this state are not very reliable guides to 

 their normal habits (cf. Swynnerton, 1919; McAtee, 1932). 

 We are bound to rely mainly on observations on birds enjoying 

 their freedom. 



We require evidence (a) not merely that predators attack 

 models and mimics, but that they gradually learn to reject 

 them ; (b) that the number of such attacks and rejections bears 

 a significant relation to the total number of individuals ; and 

 (c) that a significant number of the attacks is made before 

 the majority of the eggs have been laid by the female. With 

 regard to (a) it is obviously very difficult to obtain evidence. 

 There are undoubtedly some good observations showing that 

 certain supposedly protected forms, though often attacked, 

 escape or are only overcome with great difficulty. We may 

 instance Swynnerton's observations on the African butterflies 

 of the genus Charaxes (1926). Yet even here there is little 

 evidence that young or inexperienced birds at first attack pro- 

 tected forms, but later reject them at sight. Though no one 

 would expect that anything so difficult to observe would as 

 yet be directly established, yet the absence of the necessary 

 evidence is a definite gap in the argument for the selective 

 origin of mimicry. 



Another question which does not appear to have received 

 adequate consideration underlies the assumption that young 

 birds learn which foods are distasteful. Thus Fisher (I.e. p. 149), 

 speaking of Miiller's modification of the mimicry theory, says : 

 '. . . young birds, at least, do in fact learn much by experience, 

 and . . . during the process of self-education in what is and 

 what is not good to eat, the total destruction suffered by two 

 unpalatable species will be diminished and ultimately halved, 

 if they come gradually to resemble one another so closely that 

 the lesson of avoidance learnt from the one will be equally 

 applicable to the other.' This statement appears to overlook 

 the large extent to which young birds are taught what to eat 

 by their parents. Thus Perkins (191 2, p. 693), speaking of 



