THE OEDER OF THE GENES 



121 



shows that such is the case, for such double crossing over 

 can be detected if enough points in the series are '' in- 

 volved '^ to catch all single crossovers. Now, as shown 

 in Fig. 52, whenever double crossing over takes place 

 between y and B, the two series that result, as marked 

 by their ends alone {y and B), are still y and B. The flies 

 will therefore be placed in such classes, which are the non- 

 crossover classes. A numerical increase in this class 

 will decrease the calculated percentage of crossovers. 

 Thus double crossing over by increasing the number of 



y 



B 



B 



Fig. 52. — Scheme illustrating how double crossing over between two distinct genes, y and 

 B, is not recorded, when only y and B are involved. 



apparent non-crossovers, decreases the observed per- 

 centage of crossovers. When enough points are marked 

 along the series to pick up all double crossovers, and 

 these are then referred to the proper single crossover 

 classes, the ^ ^ piece-by-piece ' ' per cent, estimate, and 

 the percentage obtained from the cross, are found in 

 complete agreement. 



The amount of double crossing over in Drosojjhila is so 

 large that the percentage of '' crossing over" is rarely or 



