96 SECTIONAL ADDRESSES. 



numbers of controls fed on untreated food, nor under natural conditions 

 in the district of origin of the parent individuals. 



Harrison and Garrett attribute the melanism which appeared under 

 these conditions to the direct effect of the metallic salts, either on the 

 soma or, as is perhaps more probable, on the germ cells. 



They showed by a very adequate series of breeding experiments that 

 the melanism which arose in this way is inherited as a simple Mendelian 

 recessive. Although these experiments have not yet been repeated by 

 other workers, there can be little doubt that their explanation is justified, 

 and that they have for the first time brought about by artificial inter- 

 ference a new mutation, dependent no doubt on a change in a single 

 definite gene. 



But no one will pretend that this mutation in its visible form has 

 arisen because it is valuable to the animal. Nor is there any evidence 

 that it is correlated with ph)^siological differences which render the animals 

 which exhibit it less liable to be killed by feeding on contaminated food. 

 There is no published evidence that such food results in a higher 

 death-rate than that in the controls. Thus there is at least one case 

 where there is very strong evidence that the environment may induce 

 the formation of mutations which are heritable. 



It is obvious that such a direct environmental effect, when taken in 

 association with the completely established fact of the common occurrence 

 of parallel or identical mutations in allied animals, provides a complete 

 formal explanation of such facts as that the coat-colour of a race of a 

 species of rodent from an arid region will in general be lighter in colour 

 than that of a race from a more humid and therefore more thickly vegetated 

 area. It is clear that such an explanation does not require that the coat- 

 colour has any adaptive significance whatsoever : it is in complete 

 contrast with the equally formally complete explanation by natural 

 selection. But it has the advantage that it can be submitted to experi- 

 mental confirmation. 



The neo-Darwinian would explain this occurrence by assuming that 

 the dark-coloured forms were less visible against the moist and therefore 

 darker soil of the humid locality than lighter animals would be, and 

 would thus escape the attacks of carnivors for a longer period. The light 

 forms would escape notice under the bright illumination and glitter of an 

 arid and especially a desert country. Such a view assumes without ques- 

 tion that the colour of the two groups is heritable, though it makes no 

 demands for any particular type of heredity. 



The only experiments which have been made with geographical races 

 of mammals are those which Sumner has carried on over many years. 



Sumner began his work by collecting considerable numbers of indi- 

 viduals of a certain species of the deer-footed mouse Peromyscus from 

 localities in California which present extreme variations in rainfall and 

 temperature. He subjected each group to analysis, measuring such 

 characters as the length of the tail and hind foot, and estimating the 

 . colour-coat by physical methods which alfow of a numerical statement. 



He thus showed that the mice from each locality varied, and that 

 the distribution of the variates for each character formed a unimodal 

 curve. He investigated by statistical methods the correlation between 



