NATURAL SELECTION 229 



of a gene and not the remainder, but in that case also this 

 part of Fisher's argument is invalidated. 



(c) We cannot consider Fisher's evidence as to polymorphic 

 species (grouse-locusts, land snails, butterflies) in detail. All 

 the examples are highly complicated and admittedly in need 

 of further investigation. In order to support the theory of 

 the evolution of dominance it is necessary to assume that a 

 selective process has been favouring the heterozygotes at the 

 expense of the dominants. There is no direct evidence that 

 such selection occurs, and in the case of land snails (Cepea) 

 there is some evidence that the attacks of birds on the different 

 colour-forms are indiscriminate. The number of such poly- 

 morphic species is much larger than is perhaps realised (cf. 

 Chapter IV, p. 94), and the development of an ad hoc explana- 

 tion for each of them would be a thankless task. 



Ford has also pointed out (1931, p. 55) that selection in 

 the direction of suitable gene environment will be going on 

 in many different directions at once, some of which may be 

 antagonistic. He argues that the number of relevant environ- 

 ments for any one gene may be relatively small, so that a 

 number of selective processes could proceed simultaneously 

 without interference. We find this argument unsatisfactory, 

 and must regard the theory of evolution of dominance as 

 still in need of verification. There is no direct evidence that 

 most mutants are not recessive ab initio. 



Summary of Section 



The preceding paragraphs may be summarised as follows. 

 If we examine the little we know as to the causes and frequency 

 of new variations, we find the data are far too scanty to warrant 

 any generalisation. We are not able to say whether muta- 

 tion-rates in nature are as low as suggested. This, of course, 

 has no direct bearing on the value of the Natural Selection 

 theory, but it does mean that extensions of the original theory 

 should not be made to depend on the mutation-rate of Drosophila 

 as observed in laboratory conditions. The data for a con- 

 vincing mathematical treatment of Natural Selection are not 

 yet available. The formulae at present proposed rely to a 

 large extent on assumptions which have to take the place of 

 the missing evidence. None of the formulae seems likely to 

 approximate to the actualities of fluctuating environments and 



