GENES MODIFYING NOTCH. 349 



2 to give the expected number of Notch females (viz, 714), the results 

 would mean that about 300 of the Notch females had varied into the 

 normal class of females. 



We may make the comparison in another way. If the number of 

 the males be taken as the measure of each class of females, there will 

 be over 400 too few Notch females, and about 200 too many normal 

 females. 



It was the offspring of some of these lots, viz, the SS lots, that later 

 furnished the materials for selection (SSO, SSO 1, etc.). If the above 

 interpretation be accepted as plausible, then at the beginning of the 

 experiment either different genes for Notch were present or modifying 

 genes were there. The later tests proved the presence of a modifying 

 gene, but since this is not sex-linked, it may have been present in certain 

 of the females or males either in heterozygous or homozygous condition, 

 hence, until the stock could be made homozygous for this gene, random 

 selection would be expected to give for some time variable results. 



SELECTION OF FEMALES HAVING NOTCH IN ONE WING 



ONLY. 



If the somatic characters were an index of the condition of the differ- 

 entiating factor for a character, it would appear that those flies in which 

 the character appeared in only one wing should indicate a change 

 towards the phenotypic normal end of the variation curve. Hence by 

 selecting in successive generations as parents those flies that had the 

 character only in one wing, and amongst these only those in which it 

 was developed to the slightest visible extent, then one might expect to 

 bring about a change, but of course this would be equally true whether 

 the selection was based on a changing factor or on the more frequent 

 presence of one or more modifying factors. An experiment of this 

 sort was begun in the third generation after SSO (viz, in SSO 112)and 

 continued through 11 generations, with the result shown in the table 3. 

 In the first column are given the flies in which both wings are notched, 

 in the second the flies with a notch in only one wing, in the third the 

 females with normal wings, and the fourth the males. I have indicated 

 by the star (*) those records in which it appears that a considerable 

 portion of the potential Notch females fall into the phenotypic normal 

 class as shown by the excess of normal females and the deficiency of 

 Notched females over the number of the males. This change is notice- 

 able in the sixth to the eleventh generation. In the last 4 generations 

 this relation holds for all the cultures, with two exceptions only in the 

 eighth generation. It is probable, therefore, that at this time the full 

 force of selection has been accomplished and there is nothing to indicate 

 that unless some new sort of change were to occur, selection would 

 accomplish anything further after the ninth generation. 



