NATURAL SELECTION 223 



mutation has shown greatly lowered vitality during its earlier 

 generations, accompanied by marked abnormalities in breed- 

 ing.' Once established, however, the mutant strains ' tend to 

 become healthier with each generation.' 



The value of calculations and theories based on the muta- 

 tion rates and types in Drosophila and Gammarus seems to us 

 to be very questionable. In these forms we are dealing with 

 a type of variation which is in all probability of an exceptional 

 order. Wright {I.e. p. 143) speaks of gene-mutations as 

 ' generally injurious,' and suggests that they must necessarily 

 be of this nature. Fisher {I.e. p. 19) assumes ' that we may 

 reasonably infer that other less obvious mutations occur which 

 are not necessarily harmful or lethal.' The position, then, is 

 that many gene-mutations which have been exactly observed 

 are disadvantageous, but there may be others which are not. 

 Surely it is a reasonable inference that, whatever may have 

 been their frequency of original occurrence, very many viable 

 mutations of the same magnitude as those in Drosophila and 

 Gammarus must have occurred. Sturtevant (192112, p. 120) 

 even records the natural occurrence of eye colours resembling 

 those of the mutants observed in cultures. From the only 

 exact sources of information on the subject it seems that we 

 can draw very few useful conclusions as to either the frequency 

 or the nature of gene-mutations. If our theories as to the 

 process by which evolutionary change has been effected are 

 to be rigorously held to exact evidence, then we have no option 

 but to admit candidly that, as far as the frequency ] and nature 

 of observed changes in the gene are concerned, we know 

 nothing that entitles us to erect a general hypothesis. 



{b) The survival value of mutants. — We have already discussed 

 the small (or negative) survival value of most of the best- 

 known mutants. We wish here, however, to deal more 

 generally with the whole conception of an average survival 

 value as applied to the minor variants which may arise in 

 any species. 



Apart from the uncertainty as to mutation-rates, the 

 mathematical treatment of the early stages of the spread of 

 mutants does not seem to be very satisfactory. The particulate 



1 The observations of Goldschmidt (1929), Jollos (1930) and others on the 

 induction of mutation by high temperatures suggest that in exceptional environ- 

 mental circumstances high mutation-rates might actually be observed. 



