402 



THE AMEBIC AN NATURALIST [Vol. XLIX 



in the double cross-over class as abnormal unless there 

 could be no reasonable doubt as to the nature of the char- 

 acter. In case of doubt the flies were tested by crossing 



As before, yellow white normal (abdomen) females were 

 crossed to gray red abnormal males. These gave in Fj 

 YWN c? and GRAb $ which inbred gave the results shown 

 in Table VIII. 



The double cross-over class is GWAb. The combina- 

 tion did not appear once amongst the 2,690 flies that are 

 recorded in F . The percentage of crossing over between 

 Y and W is 1.0; that between W and Ab was 2.1. The 

 expectation of double crossing over on this basis (without 

 interference) would be .02 per cent., or about 1 in 5,000. 

 But the expectation would be far smaller than this be- 

 cause of a principle that we call interference. We mean 

 by this term that should a cross-over occur at one point 

 the chance of another occurring near it is greatly dimin- 

 ished, because if crossing over is due to twists of the 

 chromosome the length of a twist would usually preclude 

 the occurrence of two cross-overs near one another. In 

 other words, if the loop that makes the twist is more likely 

 to be of a certain length then the likelihood of the occur- 

 rence of a small loop necessary for a double cross-over is 

 very small. In two cases, B and C, the F 2 counts (from 

 pairs of flies) gave no YWN males as" shown in the 

 next counts. 



The absence of the YWN males, when the other classes 

 showed that no error in the experiment had been made, 

 was not understood until the occurrence of lethal factors 

 was worked out. Here clearly a lethal factor in the YWN 

 uT;m<lmother has been carried over into her GRAb 

 daughter. The lethal factor must have been closely 



