EFFECT OF HETEROZYGOSIS ON DROSOPHILA 463 



secondly because these were most easily accessible. The 'branded' 

 flies were thoroughly mixed with those to be compared with 

 them before the record was taken. Each wing was first graded 

 and then identified as stock or outcrossed type. In this process 

 of classification a diagram of the grade types was constantly 

 referred to so as to make the judgment as uniform as possible. 

 Each wing was recorded separately, there being no satisfactory 

 way in which to determine the variation of the fly as a whole. 



As far as possible the crosses for comparison were started on 

 the same day, using food from the same batch, of the same age 

 and consistency, and the bottles were placed in the same con- 

 ditions of environment. Approximately the same amount of 

 food was used in each case; and with the exception of a few 

 instances the same number of parents were allowed to breed in 

 each bottle and have access to this food. 



The food and rearing problem was without doubt the most 

 difficult question encountered. It was necessary to start many 

 crosses in order to insure a sufficient number of offspring. Un- 

 fortunately, owing to the infertility of the flies, it was often 

 impossible to obtain counts on the same date from stock and 

 outcrossed bottles that had been started simultaneously and 

 with similar food. But the results include observations on a 

 number of bottles and are consistent enough to show that varia- 

 tions in the character due to environmental differences between 

 the bottles were not responsible for the main outcome of the 

 comparison. 



The flies for each day were recorded in a correlation table, 

 one mark representing both wings, the modification in the right 

 wing being placed in its proper column within the row labeled 

 for the modification of the left. The data obtained in this way 

 gave a correlation between the right and left wings which justified 

 the order in which the grades had previously been classified. 

 For example, that 'small blister' is a greater de\'iation from 

 normal than is 'marked' is proved by the fact that 'small blister' 

 (in one wing) shows a greater tendency to be associated with 

 'marked' (in the other wing) than with 'moderate.' This may 

 be seen by consulting table 1. Similarly each of the other 



