AN ANALYSIS OF THE EFFECT OF SELECTION. 



23 



transference to a Dichset gene and must materially affect the result 

 produced by that Dichset gene. 



The hypothesis that modifying genes are responsible for the result 

 meets with no such complications. 



LOCATION OF MODIFYING GENES. 



40 



20 



The selection experiments recorded 

 above have demonstrated that Dichset 

 lines exist that are genetically different 

 with respect to bristle number. The 

 cross between the 1002 plus line and 

 1331 minus line showed that there is an 

 increase in variability in F2 when two 

 such lines are crossed. Both these facts 

 are consistent with the view that modi- 

 fying genes, other than the Dichset gene 

 itself, have influenced the bristle num- 

 ber of Dichset flies. But it would also 

 be possible to interpret the result as 

 due to variations in the Dichset gene it- 

 self, and to contamination of that gene 

 in crosses. (See above.) 



It has been pointed out by MuUer and 

 Altenburg (Morgan, Sturtevant, MuUer 

 and Bridges, 1915, p. 191), by Dexter 

 (1914), and by Muller (1917) that there 

 is a method of distinguishing between 

 these two possibilities. The truncate 

 race of Drosophila with which Muller 

 and Altenburg worked is characterized 

 by shortened ' 'truncated" wings. The 

 race does not breed true for the trun- 

 cate character, but the percentage of 

 truncates produced and the degree of 

 truncation shown are both capable of 

 modification by selection, Muller and 

 Altenburg showed that this race contains a mutant gene in the second 

 chromosome that is primarily responsible for the truncate character. 

 By means of linkage experiments involving readily classifiable mutant 

 characters they were able also to show that there are modifying factors 

 for the truncate character in the first and in the third chromosomes. 

 Furthermore, when the stock was by special methods kept uniform in 

 constitution with respect to the truncate gene itself and with respect to 

 these modifiers, selection was without effect. In this way the genetic 

 variabiUty of the race was shown to be due to modifying factors. 



Fig. 12. — Results of crossing 1002 in- 

 bred plus and speck (1331) minus 

 lines. The Pi curves represent the 

 last few generations of each parent 

 race. All four curves are reduced to 

 the percentage basis; the ordinates 

 represent percentages and the ab- 

 scissae bristle numbers. 



