264 THE SECOND-CHROMOSOME GROUP 



then, that in this case there had been crossing-over in the male between 

 the loci purple and curved. It is true that there are two or three 

 possible escapes from this necessity. Thus, "deficiency" for the 

 curved locus occurring in the star speck gamete of the father would 

 give precisely this result. Like^-ise, mutation to curved occurring 

 in the germ-tract of the star speck male would give this result. " Du- 

 plication" of the curved locus in the . purple curved speck mother 

 would answer as well. All three of these processes have been met 

 with and amply established elsewhere in Drosophila. (See especially 

 Bridges, 1917, Genetics, 2, p. 454.) 



Two of these alternative explanations, deficiency and duplication, 

 were capable of differentiation from the other two and from each other 

 by proper tests, but these were not made. There was one previously 

 well-established case of crossing-over in the male (Muller, 1916), but 

 this occurred in a very early embryonic stage and hence affected all 

 the gametes. It is to be doubted if even the case just described is to 

 be considered as brought about by a mechanism analogous to that by 

 which crossing-over in the female is regularly effected. 



As the first broods of the star purple curved speck back-cross began 

 to hatch it became apparent that the position of star is very far to 

 the left of purple — even further than streak. The completed counts 

 (table Ho) showed that star and purple gave a total of 3,010 cross- 

 overs in the 6,766 flies, or 44.5 per cent of crossing-over. Streak had 

 given only 33.1 per cent of crossing-over with purple, so that when 

 allowance was made for double crossing-over, star was calculated at a 

 position fully 16 units to the left of streak. With this addition to the 

 total length of the map of the second chromosome, it was about 110 

 units, or very nearly twice as long as the X-chromosome map. 



The crossing-over between curved and speck proved to be slightly 

 greater than the previous data had indicated, for there was 30.5 per 

 cent of observed crossing-over between curved and speck. The total 

 available data (table 115) gave 3,037 cross-overs in a total of 10,042 

 flies, or 30.2 per cent. When a correction is made for double crossing- 

 over according to the probable coincidence of 20, the locus of speck is 

 found to be about 31.6 units to the right of curved or at 105.1. 



With respect to the third problem involved, that of the relation of 

 age to the amounts of crossing-over and of coincidence, these data 

 proved rather unsuitable, because of the large interval between 

 star and the other three loci. Because of this fact none of the inter- 

 \'als worked with were short enough to exclude double crossing-over 

 and furnish an uncomplicated measure of the effects of the age change. 

 This demerit was partly compensated by the fact that nearly the entire 

 length of the chromosome was covered by the four loci. 



In general, these cultures and their totals showed the usual drop in 

 the cross-over values of the second broods (table 116) and a slight 



