Linkage Data. 35 



by double, gave: confluent 343, double 113, confluent double 83, 

 wild-type 435. Counting only the double flies (to avoid errors due 

 to viability and to inability to exclude all double flies from the 

 other two classes), this gives a cross-over value of 42 per cent, which 

 is reliable at least to the extent of showing that confluent and double 

 are not closely linked. No attempt was made to determine the 

 amount of crossing-over between double and concave, because of the 

 difficulty of distinguishing the two characters. 



ConUuent and broken. — Two back-crosses of heterozygous females 

 (experiment 55) gave 139 non-cross-over flies and 93 cross-overs, 

 from which we obtain a cross-over value of 40 per cent, showing 

 very loose linkage. 



CONSTRUCTION OF SECOND-CHROMOSOME MAP. 



(Figure^?). 



Unfortunately, broken appeared so recently that it has only been 

 possible to secure counts from the one type of mating given above; 

 hence broken can not yet be located on the map. 



Since the '^ position" of double is likewise uncertain, only the 

 genes for confluent and concave can be located. These are probably 

 at least 50 units apart, and perhaps much farther, since double 

 crossing-over would probably prevent the detection of more than 

 about 42 per cent of crossing-over, even if more occurred. 



