44 THE VARIATION OF ANIMALS IN NATURE 



examples of this kind of difficulty. In general the principle of 

 physiological economy on which the selective explanation of 

 atrophy from disuse is based, seems to us very unsatisfactory. 

 Where an organ or structure is definitely inconvenient in a 

 new mode of life its disappearance may be expected ; but 

 when it is merely useless it is very difficult to see how slight 

 variations in the direction of reduction could be effectively 

 selected, more especially if they are infrequent. 



We do not consider that this line of evidence is particularly 

 helpful, as it seems incapable of exact examination. Experi- 

 ment may show us that a given organ does or does not atrophy 

 through disuse and, if it could be experimentally proved that 

 complete atrophy took place in conditions in which selection 

 could be excluded, it would go a long way to proving that the 

 results of disuse are progressively inherited. Up to date no 

 such experiments are available. It may be pointed out that 

 Payne (191 1) subjected sixty-nine generations of Drosophila 

 ampelophila (melanogaster) to total darkness without any modifi- 

 cation of the eyes or the reaction to light. 



(B) Correlation of environmental differences with structural 

 divergences known or presumed to be hereditary. — Under this 

 heading we have a very large body of facts summarised in a 

 very able fashion by Rensch (1929). This author maintains 

 that there is much evidence tending to prove that lines of 

 structural differentiation are very frequently correlated with 

 environmental ' trends, 5 that there is a ' parallelism ' between 

 ' phenotype ' and genotype of such an order that ' modi- 

 fications ' can be artificially induced which are the same 

 as, or more or less the same as, characters known to be ' geno- 

 typic,' and that there is an inference that such externally 

 induced * Phanovarietaten ' become genotypically fixed. 

 He admits (p. 161) that the latter stage in the thesis has never 

 been quite unexceptionally proved ; but he holds that this 

 parallelism is of the highest significance. He gives great 

 weight to the production of identical variants in natural and 

 comparable experimental environments, but actually there 

 are very few instances of such parallelism. His evidence, 

 indeed, consists only of Sumner's (1915) experiments, already 

 shown (p. 39) to be of dubious value. He does not, however, 

 refer to the later experimental work of Sumner (cf. 1923), in 

 which it was shown that ' environmental ' forms of Peromyscus 



