OF MUTANT CHARACTERS. 189 



of these problems in double crossing-over is out of all proportion to our 

 progress in their solutions. 



It seems likely that a more satisfactory method of expressing 

 these relations may be derived than is provided by the present form- 

 ula; the new formulation must take account of separate factors analyz- 

 able in the process and permit their adequate representation. The 

 conclusions based on the old formula must be regarded as provisional. 



THE USE OF PURPLE IN MAPPING OTHER GENES-CURVED. 



STREAK. ETC. 



The "map" of the second chromosome began to be useful when the 

 order and spacing for the three genes, black, purple, and vestigial, 

 were roughly established by the determination of the third value, that 

 for black purple (December 1912). The three-locus experiment just 

 given provided more accurate measures of the map-distances involved. 

 The preparation of the multiple recessive and of the Pi stocks delayed 

 the completion of this experiment for nearly a year (January 1914). 

 Meanwhile, these provisional locations were used as the basis for 

 locating other genes more closely. The first of these was curved. 

 Bridges and Sturtevant (Biol. Bull., 1914) soon found (January 1913) 

 that black and curved gave approximately 23 per cent of crossing- 

 over. The next point to be determined was the relation of curved 

 and one of the other two mutations whose loci had been mapped. 

 Both of these tests were used, since each offered advantages; the chief 

 disadvantage was that vestigial interferes with the classification of 

 curved, so that it is impossible to distinguish between the simple 

 vestigials and the vestigial curved class. 



The purple curved test was undertaken by Bridges, who prepared to 

 run a three-point back-cross involving black, purple, and curved. The 

 first step was the synthesis of the purple curved double recessive. As 

 soon as this was obtained it was turned over to Mr. W. S. Adkins, who 

 ran a preliminary back-cross test of the simple purple curved crossing- 

 over, and found that there was about 18 per cent of crossing-over. 

 This enabled us to determine the relation of curved to the other three 

 genes. The purple curved value of 18 showed that curved was closer 

 to purple than to black (black curved = 23) and that purple and vesti- 

 gial were therefore "to the right" of black. Curved was further to 

 the right than vestigial, since black and vestigial gave only about 18 

 per cent of crossing-over. 



The vestigial curved distance was tested by Sturtevant, who found 

 that there was about 8.5 per cent of crossing-over. Because of the 

 difficulty of classification already referred to, it was not thought worth 

 while to run these tests on a large scale. However, vestigial is itself 

 accurately mapped and is nearer to curved than purple is, and these 

 considerations are strong enough so that the vestigial curved tests will 



