192 THE VARIATION OF ANIMALS IN NATURE 



extensive than that of another, and that more time may be 

 required to exhaust it. Selection may be carried on success- 

 fully over a certain number of generations and then stopped 

 before improvement has ended. All that we are entitled to 

 infer from this is that selection has been successful up to a point. 

 We are not entitled to assume that it will continue to be so. 

 Castle (I.e.) considered that the extensive changes in pattern 

 which he produced in rats were due to the effects of selection 

 on the ' residual heredity ' and ' not to any change in the gene 

 for the hooded character.' That this interpretation is correct 

 is shown by the result of back-crossing both the selected types 

 to unselected ' selfs.' But even so, the modification produced 

 was very extensive, whatever the underlying cause of variation 

 may have been. Even if selection had ceased eventually to 

 be effective (' the variability of the stock had not been dimin- 

 ished during twenty (selected) generations'), the amount of 

 change wrought by it was very large, and it seems quite irrele- 

 vant whether it was due to a change in the hooded gene or to 

 residual heredity. It should also be noticed that in this case 

 selection brought about substantial results without any fresh 

 stock being introduced. 



The negative results cited certainly show that the initial 

 variability of a stock may be easily exhausted and its capacity 

 for improvement by selection may be very limited, unless 

 reinforced by new gene mutations. But it is equally clear 

 that in other heterozygous stocks there is a large opportunity 

 for selective modification. This conclusion shows that the 

 effect of selection is entirely a question of the initial variability 

 of a stock and its subsequent mutations, and that Darwin's 

 general assumption of unlimited variability is scarcely justified. 

 It also points our way to the really crucial question — viz. how 

 frequent in nature are species which are heterozygous for many 

 characters ? As we saw in Chapter II (p. 26), we are still far 

 from being able to give an answer. 



II. Direct Evidence for Natural Selection. 1 — The inci- 

 dence of death-rates in nature. — The facts and arguments dealt with 

 in the preceding section do not, of course, cast any light on what 

 is, after all, the most important question — viz. Is there a selective 

 process in nature ? As we have already pointed out, for Darwin 



1 In the present chapter we use the term ' adaptation ' in a comprehensive 

 sense. In Chapter IX it is subjected to more detailed analysis. 



