Linkage Data. 49 



interrupted flies here appear to be normal, hence it is unsafe to use 

 the total count. Using only the interrupted flies, a cross-over value 

 of 19 per cent is obtained. 



Fused and branched. — Back-crosses of females heterozygous for 

 these two characters, in opposite chromosomes gave 49 cross-overs 

 in a total of 271 flies, a cross-over value of approximately 18 per cent 

 (experiment 65). 



Fused and approximated. — Experiment 66 gives the results from 

 back-crosses involving these two characters alone. Owing to the 

 difficulty of classifying approximated in the presence of fused, 

 only the not-fused flies are used in calculating the cross-over value. 

 These give a value of 35 per cent. Breeding tests of flies in both the 

 approximated, and the fused (not approximated) classes, were made 

 to verify the classification. 



Interrupted and branched. — Experiment 67 gives the results of 6 

 back-crosses involving these two characters. The data are very 

 unsatisfactory, for both characters are irregular in appearance, 

 making the wild-type class very large, at the expense of the two 

 non-cross-over classes. Using only the interrupted flies a cross-over 

 value of 2.4 per cent is obtained. This seems to indicate close linkage 

 between the two, but it needs to be checked by the opposite type of 

 mating, using both mutant genes in one chromosome, to make sure 

 that branched does not tend to conceal interrupted. 



Interrupted and approximated. — The regularity in appearance of 

 interrupted varies considerably in different bottles (see description), 

 and for this reason many of the data can not be used for linkage 

 calculations. Furthermore, it is often difficult to classify approx- 

 imated in the presence of interrupted. Experiment 68 gives the 

 results of 6 back-crosses involving interrupted and approximated 

 (both genes in the same chromosome). In the first part of this 

 experiment, the flies were all put in the four regular classes, the 

 doubtful ones being classified as accurately as possible. Values have 

 been calculated here by using (1) only the not-interrupted flies, 

 (2) only the interrupted flies, and (3) all flies. These are respectively 

 21, 13 and 17 per cent. The second part of the experiment is from 

 counts in which especial attention was paid to separating the inter- 

 rupted flies into three classes — those that were approximated, 

 those that were not approximated, and those that were doubtful. 

 When only the not-interrupted flies are counted, a cross-over value 

 of 16 per cent is obtained. 



Another method of handling the data in the second part of the 

 experiment has been used to obtain the value given. This permits 

 the use of all flies by dividing the doubtful flies between the inter- 

 rupted and the interrupted approximated classes in proportion to the 

 sizes of the latter. On this basis 16.2 per cent (35) are added to the 



