242 



Sturtevant. 



there was crossing over between W and M. This phenomenon I shall 

 call interference M. 



Another way of expressing the relation is as follows. In the YWM 

 experiments the cross-overs between Y and W amounted to 1' 04^/0 of 

 the total, those between W and M to 32"56'' o. Had these two cross- 

 overs been entirely independent of each other we should have obtained 

 0-34''/o (-0104 X -3256) of double cross -overs. But we actually did 

 obtain only 0'07'',o. The ratio between the number of double cross- 

 overs expected ^\itliout interference and the number actually obtained 

 (in this case 0*34,0'07 = 4'9) is then the index of interference. 

 If the ratio is 1 there is no interference — the cross-overs are independent. 

 It grows larger as interference becomes greater. Table IV shows the 



Table IV. 



values for this index obtained from the data summarized in table III. 

 The counts on which YWR, YVBr, and WVM are based are too small 

 to be very significant, and some of the others also were not done on 

 as extensive a scale as would be necessary to get really reliable results. 

 However, in every case interference was found, and it is also obvious 

 that it is less when long distances are involved than when shorter ones 

 are used. The expected percent of double cross-overs (numerator of the 

 fraction in 2nd column of the table) is one measure of the lengths of 

 the two distances involved in any case. The smallest expected percent 

 of cross-overs is in the case of YWR, and interference was complete 

 here, but the count was so small that the case carries no weight. The 



^) This term was suggested by Mr. H. J. Müller, to whom is due much credit 

 for his part in the analysis of the phenomena involved. 



') Based on counts that are too small to be significant. 



