OF MUTANT CHARACTERS. 



225 



This very free crossing-over between streak and morula indicated 

 that streak was far away from the right end of the chromosome, where 

 the gene for morula had been located. It was thought thiit the locus 

 of streak was in the neighborhood of bhick (which was at that time 

 considered the left end of the chromosome) or purple, which was not 

 far from black. It did not seem probable that an accurate classifi- 

 cation of streak and black could be made at the same time, so purple 

 was used instead. 



Table 71. — Pi, streak 9 X morula cf; Fi streak 9 X morula cf of stock. 



A three-locus experiment partly balanced for inviability was carried 

 out. A streak female was out-crossed to a purple curved male and Fi 

 streak females were back-crossed by purple curved males. Since 

 streak is a dominant, the triple recessive not-streak purple curved was a 

 double mutant form, which was a great saving of labor over the ordinary 

 case, in which the triple mutant multiple recessive has to be made up. 



Table 72. — Pi, streak 9 X purple curved cf ; B.C., Fi streak 9 X purple 



curved d^ of stock. 



The result of the back-cross (table 72) was surprising, since it upset 

 the ill-founded notion that black was at the left end of the second 

 chromosome. The cross-over values (streak purple = 36.0, jnirple 

 curv'ed 24.5, streak curved 45.7) and the double cross-over classes 

 (streak purple versus curved) showed that streak was fully 40 units to 

 the left of purple (allowing for double crossing-over). No trouble in 

 classifying streak was met in this experiment, so that all of the flies are 

 available for the calculation. 



From the streak purple and the curved flies that appeared in the 

 back-cross just described a Pi mating was made for the second type of 



