GENES MODIFYING NOTCH. 347 



concavity of the edge. The range of variation of the notching is very 

 wide. That the limit of variability overlaps in one direction the nor- 

 mal wing is certain, for amongst the daughters without notching occa- 

 sionally one is found half of whose daughters are notched. The not 

 unusual occurrence of a fly with one entire wing and one with notching 

 (fig. 91, c) indicates that the range of variation includes normal wings. 

 The low productivity of the Notch female appears to be an incidental 

 effect of the Notch factor, because the normal sisters of the same stock 

 are, whenever tested, much better producers. The viability of the 

 Notch females is fairly good, but they appear to run behind their nor- 

 mal sisters in nearly all cultures. Change in the viability will be dis- 

 cussed later. 



THE PROBLEM. 



Throughout the older literature dealing with selection, the idea that 

 the grade of any character shown by animals or plants is a criterion of 

 the condition of the genetic factor or gene responsible for the character 

 continually recurs, and the same idea appears occasionally in more 

 recent tunes, despite Johanssen's analysis showing the inconsequence 

 of such an argument, and despite the accumulated demonstrations 

 that the production of a given character depends on the environment 

 and on internal modifying genes, as well as on the principal gene itself. 

 The wide range of variability of the notching, the fact that the females 

 genetically Notch may be identified by the 2 to 1 sex-ratio in the off- 

 spring, even when the wings themselves have somatically the normal 

 margin, as well as the fact that it is a dominant, and therefore any 

 alteration in the gene may be tested directly by outbreeding; the fact 

 that linkage relations made it possible to identify any changes that 

 might follow selection in the individuals that were Notch, although 

 with normal wings; all these made Notch excellent material on which 

 to put to actual test some of the older as well as current views con- 

 cerning the nature of Mendelian factors and the influence of selection. 

 In each generation several (usually 2 to 10) virgin, notched- winged 

 females (of the derived type) were picked out and put into a new bottle 

 with one to 10 males. Occasionally pairs were used and then mass 

 selection followed in the next generations. This prodecure is not un- 

 like the rough procedure formerly practised by the breeder, but is not, 

 of course, to be recommended for a thorough understanding of the 

 changes that are taking place during the selection period. Moreover, 

 by such a method the end result is attained only after a long tune, 

 whereas the results here described could probably have been reached in 

 two or three generations; for, as the duplicate experiments show, the 

 modifying gene for "slight notch" did not arise in the course of the 

 experiment, but was present in some flies of the stock at the beginning. 



