THE DETERMINATION OF DOMINANCE. 309 



which clearly indicates that through some cause the hibernating 

 conditions of the location were favorable for signaticollis, but were 

 decidedly unfavorable for the undecimlineata and for the interme- 

 diate hybrid type. These individuals were allowed to interbreed 

 freely and produced a numerous progeny, in which the larvae 

 were of four different types ; white without spots, white with spots, 

 yellow without spots, and yellow with spots. The second genera- 

 tion emerged from the middle to the end of July, 1905, and 

 showed a huge preponderance of the signaticollis type. The 

 census of a random sample taken the last week in July gave the 

 following count: 



These individuals were not removed from the colony ; the census 

 of the sample was made, the individuals put back, and the colony 

 allowed to encounter the conditions and behavior which it would 

 meet in a state of nature. Nine pairs, taken at random, of the 

 undecimlineata type were bred out as pedigreed cultures during 

 August and part of September, 1905, and gave uniformly an 

 undecimlineata progeny. Seven pairs of the signaticollis type, 

 which were bred out, gave uniformly a signaticollis progeny, and 

 out of five other pairs there appeared individuals of the mid type 

 and of the undecimlineata type, showing that some of the signa- 

 ticollis type were hybrid in character. Six pairs of the mid type 

 were also bred out as pedigreed stock, and showed themselves to 

 be in every case hybrid. 



The third generation was produced in August and early Sep- 

 tember, 1905. In this the larvae were of the same four classes, but 

 showed a huge preponderance of yellow larvae (YIS). A count 

 made late in August, when perhaps the bulk of the larvae had 

 entered into pupation, gave the following results: 



The adults of Generation III. emerged early in September; 

 a census made about the middle of September gave the following : 



