EFFECT OF SELECTION ON CROSSOVER VALUES 
349 
There can be no doubt that these crossover values are signifi- 
cantly different from any ratio in the Fi in table 1, or from the 
usual ratios shown by random stock females. Furthermore, 
there is an interval of about 10 per cent between the lower and 
higher groups of table G, in which we found no crossover values. 
The natural inference is that any attempt to increase the amount 
of crossing over leads to double crossing over, and thus to very 
low crossover values (practically zero). That is, these nine 
females showed a marked decrease in crossover values, despite 
high selection, because they gave almost nothing but double 
crossovers. In other words, their low crossover values are, 
after all, the result of effective high selection. Mr.. L. E. Thorne, 
who had this series under observation, was called into military 
service and we did not make any further tests on this material. 
TABLE 6 
The distribution of crossover values in the Fi generation of series C 
NUMBER 
THE DISTRIBUTION OF CROSSOVER VALUES 
CROSSOVERS 
TOTAL 
CROSSOVER 
OF PAIRS 
lO 
o 
•o 
o 
•o 
5^ 
in 
lO 
CO 
"5 
O 
VALUE 
72 
5 
4 
1 
1 
6 
8 
15 
16 
8 
6 
1 
1 
3,893 
13 ,705 
28.41 
We hope, however, to repeat the high-selection experiment and 
test out the region between white and miniature in such females 
which apparently give uniform double crossing over in a region 
in which single crossing over is the rule. 
DISCUSSION AND SUMMARY 
As far as we are aware, there is only one record of a similar 
selection experiment. Gowen ('19) selected for high and low 
crossover values, but his results and conclusions are diametri- 
cally opposed to ours, since he found selection ineffective, and 
concluded there were no differences in modifying factors for 
crossing over in his experiment. He continued selection for 
only five generations in the low series and six in the high, using 
the region of the third chromosome between sepia and rough. 
