AN ANALYSIS OF THE EFFECT OF SELECTION. 



25 



between them must be due to second-chromosome differences between 

 the two Pi races. This experiment may be continued further by 

 mating the not-speck Dichset males produced by the back-cross to the 

 speck not-Dichset females. Such a mating should give the same result 



n 



n in 



^^ irrr: 



B.C. 



m 



D' 



Test 



D' 



I 11 



in 



Fig. 13. 



as the first back-cross and does in fact do so. In table 19 both tj'pes of 

 experiment are treated together. Figure 14 is a graphical representa- 

 tion of the result of one experiment of the tj-pe here described. 



It should be noted that such experiments are suited only for the 

 detection of dominant modifiers present in one Pi race, or, stated 

 conversely, recessives present both in one Pi race and in the speck race 

 from which the test female came. 



Table 19. — cf Tests, Chrovxosome II. 



Table 19 gives the results of the experiments of this type that have 

 been carried out. (For the raw^ data see Appendix.) The first two 

 columns give the Pi races, or in cases where the tested male came 

 himself from a back-cross test, the original source of his not-speck 

 and speck chromosomes. The third column gives the source of the 

 test female. All tests in which the data for these three columns were 

 identical have been lumped. The next three columns give the mean 



