SELECTION IN BAR-EYED RACE OF DROSOPHILA. 389 



latter part of this work and on account of the lack of time no 

 special effort was made to determine the exact relation between 

 temperature and facet number. The fact that the change 

 usually occurred one to three days sooner in bottles that just 

 began to hatch than in old ones can be explained by tRe assump- 

 tion that the effect was produced at an early stage in the develop- 

 mental period. Since the first flies hatching from a bottle must 

 have a shorter developmental period than the later ones they 

 would be the first to show the effect. 



It does not seem possible, however, that temperature is the sole 

 cause of somatic variations. In the VBa selections it was ob- 

 served that the two lines throughout averaged. higher than the 

 stock. In spite of that fact an examination of the extremes 

 shows that most of the high flies were eliminated from the low 

 lines. The rank of the average flies must, then, have been raised. 

 The same can be said of the low lines in Ba. Here the elimina- 

 tion of high flies is much more pronounced and still the mean 

 remains very near that of the stock. It is possible that the 

 crowded condition of the larvae in the stock bottles reduced the 

 facet number, but the real cause may be something very different. 



The results of the selections in VBa indicate that there is in 

 this race only a single hereditary factor involved in the modi- 

 fication of the facet number. In spite of the fact that the mean 

 of the stock is much lower than that of either line we must 

 assume that under the same conditions it would lie somewhere 

 between them, in other words, if the lines could have been reared 

 under the conditions of the stock, the low lines would have gone 

 slightly downward due to the elimination of high numbers and 

 the high lines would have gone slightly upward due to the elimina- 

 tion of low numbers. Practically pure lines were established in 

 the first generation. This, however, may be merely apparent, 

 the lack of further effect of selection being due to the interference 

 of somatic factors. 



The male with 34 facets, giving rise to the brood l6fi-6, must 

 be regarded as a mutant. This conclusion is based on the great 

 difference between this male and its brothers in 16, the definite 

 relation between the male parent and the female offspring, and 

 the extremely low grade of the resulting females. 



