D.— ZOOLOGY. 97 



pairs of the characters with which he worked, showing that for many 

 of them the correlation was small. He showed that the curves for different 

 subspecies might overlap, so that no one individual could fairly represent 

 its race. 



By a series of breeding experiments carried on with caged animals 

 Sumner showed that, when allowance was made for certain bodily changes 

 clearly caused by the artificial conditions of life, the races bred true in 

 the sense that the modes of the curves of variation of the characters con- 

 sidered remained stationary. 



The results of crossing individuals selected from different subspecies 

 and treating in a biometric manner the offspring resulting from these 

 crosses were uniform, in so much as that the fi generation were always 

 intermediate in character between the parents, and the range of variation 

 they exhibited was less than that of the parent stocks. In later genera- 

 tions there was no obvious segregation, and the range of variation increased 

 again. 



Sumner at first regarded these results as evidence of a blending inheri- 

 tance without any Mendelian character ; but subsequently concluded that 

 they could be explained on a multiple factor hypothesis, like that which 

 is accepted for Castle's hooded rats. The reduced variability of the fi 

 generation is thus accounted for. 



Although as a palaeontologist who has seen the extraordinarily small 

 magnitude of the steps which separate successive members of a phyletic 

 line I am temperamentally indisposed to do so, I am forced to accept the 

 multiple factor hypothesis as an account of the majority of cases of 

 blending inheritance. Castle's experiments on hooded rats, carried as 

 they have been over very many generations, seem conclusive for that 

 particular case. It seems clear, furthermore, that any change in a 

 spermatozoon which results in a change in the adult which arises from its 

 conjugation with an egg, must be a chemical change ; and chemical 

 changes are all particulate, there are no intermediates between a hydrogen 

 atom and a methyl group ! 



It follows therefore that the light-coloured mice of the arid interior of 

 California differ from those of the coast because in them have been accumu- 

 lated a number of genes for light pigmentation, much more sparsely 

 present in the dark races. 



Such a differential distribution of genes is of course what is assumed to 

 occur under the influence of natural selection. 



It is not perhaps very easy to believe that the direct action of the- 

 environment would result in the production of a series of mutations all 

 independent, and all in the same direction, yet this assumption is necessary 

 for the alternative explanation of direct environmental effect ! 



But Sumner went further, and attempted to investigate the possibility 

 of such environmental influences by direct experiment. He transplanted 

 a small colony of mice into a very different environment, enclosing them 

 in a small netted area and leaving them to breed. The offspring which 

 appeared during the course of the experiment showed no tendency to 

 approach the local races in their characters. 



This experiment has been criticised because the numbers of individuals 

 were small, and because they were unnaturally crowded in a small en- 

 1929 H 



