346 J. A DETLEFSEN AND E. ROBERTS 
of the selected pair in each generation and the same data for 
en-masse matings from these selected pairs. Text figures 1 and 
2 give a graphic representation of the progress made in series B. 
The graphs are based upon the crossover values in the selected 
line; i.e., all en-masse matings except the selected one have been 
neglected in plotting the graph. In other words, the graph 
relates only to the actual line of selection, and all side Hues have 
no weight in determining the coordinates. It will be clear that 
those generations in table 3 which have any number of pairs 
entered under that heading were generations made up entirely 
of paired matings, while all other generations were en-masse 
matings. 
The first three selections had little or no effect, but it cannot 
be said that selection was very rigid during these generations. 
In Ft we selected a pair giving 36:251 = 14.34 per cent and 
made some progress, for the next seven generations (Fg-Fu) 
fluctuated between 10 per cent and 23 per cent. The subse- 
quent nine generations (F]5-F23) fluctuated around 10 per cent. 
Selection was carried on up to F29 and the last six generations 
(F24-F29) varied around 6 per cent."* After that we simply 
carried the stock without selection, and have found it to breed 
quite true to low crossover for twenty-two generations. The 
F29-F50 have given values around 6 per cent. These last twenty- 
one generations are shown in table 4. 
There are some features of tables 3 and 4 which require com- 
ment for the sake of clearness. Temperature conditions made 
it necessary to breed the offspring of the selected pair in the F13 
for two generations by the use of en-masse matings. Hence, 
the matings in the Fi4 and F15 show no pairs and selection was 
interrupted. This was the only case in which the usual sequence 
of selecting in alternate generations was not followed. The F33 
showed a rather abrupt rise in crossover value (12.50 per cent), 
"• An independent mutation of gray to yellow which occurred in the F25 should 
perhaps be put on record. One female ( 9 no. 30) proved to be heterozygous for 
yellow, and this gene was linked to white and miniature. Hence the mutating 
gene came through the spermatozoon from the gray white miniature father of 
9 30. This new gene for yellow proved to be identical with the original yellow 
mutation found by Wallace in 1911 (Morgan and Bridges, '16). 
