2i8 THE VARIATION OF ANIMALS IN NATURE 



to some purpose. In the first place, we have to-day enough 

 evidence from experiment to convince us that much variation 

 is purely somatic and non-heritable. Darwin's unlimited 

 variation no longer appears as an inexhaustible fund for 

 selection to draw upon, and the question begins to shape itself 

 in our minds — with this reduction made, are heritable varia- 

 tions frequent enough to provide a reasonable chance that they 

 will coincide with the crises that supposedly lead to selection ? 



The initiation of a mathematical treatment of Natural 

 Selection was due to Pearson and his collaborators. Pearson 

 himself (1903) contributed an attempt to give a mathe- 

 matical expression to the action of selection, and studied 

 the special effects of selection in reducing variability and 

 causing correlation. As regards his main theory, ' the 

 calculations,' as Haldane (1932, p. 171) has pointed out, ' rest 

 on the particular theory of genetics held by Pearson, and the 

 results are not in harmony with experimental results obtained 

 in other organisms.' Of recent years several attempts have 

 been made to develop a mathematical theory of selection 

 which is based on our experimental knowledge of the laws 

 of heredity. These studies (Hardy, 1908 ; Fisher, 1930 ; 

 Haldane, 1932 ; Wright, 1931) do not in fact provide any 

 proof of the efficacy of selection, though Fisher and Haldane 

 imply that selection is the only means of accounting for the 

 spread of variants that occur as single or few individuals. 

 Selection is always taken as a vera causa, and the various mathe- 

 matical expressions of its activity are based on this assumption. 

 Moreover, although most authors are aware of the fact 

 that ' all-round adaptiveness ' cannot be neglected, the action 

 of selection is sometimes considered rather in vacuo as a 

 unitary process affecting single genes, whereas in nature 

 survival and extinction are probably issues in which the 

 organism as a whole is involved. 



As we have already indicated, we do not think any 

 deductive argument can really replace the crucial direct evi- 

 dence that a selective process actually occurs in nature. But 

 if for the moment we neglect this point, we believe that it is 

 easy to be misled by concentrating too much on the genetical 

 evidence, which is necessarily drawn from a few intensively 

 studied species (for many purposes a single species of Drosophila) . 

 After all, what we have to explain is the normal cause of 



