28 



AN ANALYSIS OF THE EFFECT OF SELECTION. 



1002 line contained a dominant plus modifier in the region of the 

 rough locus. 1 



There can be no question that the lines studied do differ in their 

 constitution with respect to definite modifying genes that affect 

 bristle number. In the case of the 1002 and 1331 lines there is at 

 least one modifier, and probably two, located in different chromosomes. 

 This gives the explanation of the increased variability observed in F 2 

 when these lines were crossed. The only other available explanation of 

 that phenomenon — contamination of allelomorphs — has already been 

 shown above to lead to complications in this case. (See also below.) 

 Since it is both improbable and unnecessary, it may safely be dis- 

 carded. 



Table 21. — 9 Tests, Chromosome III. 



THIRD-CHROMOSOME LETHALS. 



Culture 1264, belonging to the third generation of the 1002 inbred 

 plus line, produced, in the last 6 days it was counted, 60 Dichsets and 

 no not-Dichsets. It seemed possible that one of the parents was 

 homozygous for Dichset, so the line was continued. It was finally 

 bred through about 18 generations, and produced 2,735 Dichsets and 

 only 4 not-Dichsets. The 4 not-Dichsets suggest the hypothesis that 

 all the Dichsets are really heterozygous as usual, but that they carry 

 a lethal in the other chromosome that kills the not-Dichsets. 2 That 

 they are heterozygous has been shown by out-crossing them. When 

 mated to Dichsets of other strains the result was 211 Dichsets to 103 

 not-Dichsets (4 cultures), the 2 : 1 ratio usually obtained when Dichsets 

 are mated together. When mated to not-Dichsets the result was 207 

 Dichsets to 209 not-Dichsets (6 cultures) — a normal 1 : 1 ratio. That 

 there is a lethal in the stock has been shown by mating Dichsets of 

 this strain to Extended flies and inbreeding the not-Dichset offspring, 

 which were found to carry a lethal as expected. (See below.) 



1 The second row of table 18 seems to contradict the conclusion that the 1002 and 1331 lines 

 differed with respect to a modifier near rough. However, the experiment represents only a few 

 flies, and did not give a significant result. Moreover, it was carried out before the 1002 line 

 had been very long inbred (F&), and involved a not-rough chromosome from that line, which had 

 not then become homozygous for rough. 



2 See Muller (1917) for a discussion of autosomal lethals. 



