NATURAL SELECTION 255 



as to the extent to which selection of the kind required is 

 really operative. 



A somewhat different argument has been applied by 

 Heikertinger (1929a) to the supposed warning colours of many 

 Hymenoptera. He maintains that, so far from being pro- 

 tected, such species are the favourite food of many birds. It 

 is perhaps significant that Myers (1931) found that unpleasant 

 taste appeared to disgust a Coati (S. American mammal, 

 largely insectivorous) far more than stings. Heikertinger 

 bases his statement on the analysis of stomach-contents made 

 in Hungary and U.S.A. The literature on the food of birds 

 is vast and requires an adequate quantitative investigation 

 from this point of view. Heikertinger entirely ignores the 

 possibility that birds may have a scale of likes and dislikes ; 

 they may perhaps eat only Hymenoptera when very hungry 

 or when other food is scarce. There is certainly a prima facie 

 case for Heikertinger's contention, but only quantitative data 

 can settle the question {cf. Protective Resemblance, p. 233). 



In the early stages of genetic inquiry it was thought that 

 every mutation must always have produced as big an effect as 

 it is seen to produce at the present day. On this basis, Punnett 

 (191 5, p. 141) and Nicholson (1927) have suggested that, as 

 the patterns of some of the mimetic forms of butterflies are 

 known to be inherited as units, it may be assumed that they 

 arose in a single step. It is now known that effects of a given 

 gene depend on the gene-complex which forms part of its 

 environment. If this environment is altered, so will be the 

 effects of the gene, and we have no reason, therefore, to assume 

 that a mimetic pattern, now inherited as a unit, tells us what 

 effect the controlling gene had initially. In this way it can be 

 assumed that selection has acted, not on the controlling gene, 

 but on the genetic environment with which it reacts. It 

 may be noted that there is no more evidence for this theory 

 than there is for the simpler assumption. 



(2) It is not very difficult to find a few cases of close resem- 

 blance between animals living in entirely different countries. 

 We may instance Bombus terrestris xanthopus of Corsica and 

 B. eximius of the Himalayas, which belong to different sections 

 of the genus. Berg (1926, chapter viii) quotes several 

 additional examples of more or less widely separated species 

 resembling one another in colour-pattern ; and Dewar and 



