180 



THE AMERICAN NATURALIST [Vol. L 



Drosophila the descendants of each pair being allowed to inter- 

 breed freely. As to the results we said (p. 776) : 



These experiments show no appreciable effect of inbreeding (on vari- 

 ability). In every case the brood reared under the best and the most 

 uniform conditions has the highest average number of teeth (in the sex- 

 comb), irrespective of whether or not inbred. The same may be said 

 of variation in size. Inbreeding has diminished neither the average 

 size nor the variability in size. 



Walton considers these conclusions justified by our statistical 

 constants in the case of the sex-comb, but believes that a signifi- 

 cant difference is observed in length of tibia, which we found to 

 be both greater and less variable in the culture not inbred. He 

 criticizes our failure to calculate coefficients of variation for 

 tibia length (as we had done for sex-comb) and upon calculating 

 such coefficients finds the greater variability of the inbred lots 

 significant. But the same difference in variability was indicated 

 by the standard deviations (which we gave) and the calculation 

 of the coefficient of variation adds nothing to the force of the 

 demonstration. We considered then and still consider the dif- 

 ferences observed sufficiently accounted for by external condi- 

 tions, i. e., we considered them purely phenotypic. We showed 

 that length of tibia is greatest and its variability least when 

 food and temperature conditions are best. The difference be- 

 tween two inbred races (M and N) inbred practically the same 

 number of generations (viz., 31 and 30, respectively) but treated 

 very differently as regards food, was found to be several times 

 greater than the difference between the inbred culture M and 

 the not-inbred culture X. Hence it is not probable that the in- 

 breeding had anything to do with the differences found in 

 variability. 



It is difficult to understand how on any theory of heredity in- 

 breeding could be expected to increase variability within a single 

 inbred line, such as one of our inbred cultures of Drosophila. 

 On a Mendelian theory it would be expected that inbreeding, 

 brother with sister, for a large number of generations (61 in our 

 A series) would result in the production of a number of homo- 

 zygous lines, each of which by itself would be entirely devoid of 

 variability, except that due to environmental agencies. If all 

 the possible derived lines descended from a pair under inbreed- 

 ing were combined into one mass of material, it would seem 

 probable, on a Mendelian theory, that if any genotypic varia- 



