738 PROCEEDINGS OF THE AMERICAN ACADEMY. 



found in the A series. The inference naturally drawn from these facts 

 was that inbreeding had decreased the fertility of the A series, but 

 subsequent investigations throw doubt on this conclusion. 



The A series was continued from June, 1903, to October, 1904, by 

 the senior author. The warm weather of the months June-September, 

 1903, gave the flies optimal conditions for reproduction, and naturally 

 the average number of young in a brood rose somewhat, reaching its 

 highest point in the 29th generation, 153.5 young to a brood. But 

 throughout this period, notwithstanding the favorable conditions, a con- 

 siderable proportion of the pairs proved sterile, amounting on the average 

 to 28 per cent of the pairs formed. The average brood of a fertile pair 

 was 124 young. 



Ten control pairs from outdoor sources, kept under the same condi- 

 tions and at the same time as generation 27 of the A series, were all 

 fertile. In only four out of the ten cases was a full record made of the 

 number of young produced. These four broods numbered 164, 204, 

 304, and 241 respectively, an average of 228 to a brood, which was just 

 100 greater than the average-sized brood produced by fertile pairs in 

 generation 27 of the A series. 



In October, 1903 (31st generation), the productiveness of the A series 

 fell off sharply and the series seemed on the verge of extinction. Four 

 out of seven pairs were sterile, and the three fertile pairs produced in all 

 only twelve young. At the same time control pairs from outside were 

 producing vigorous broods containing usually 200 or 300 individuals. 



But in the subsequent (32d) generation, the A series seemed to 

 recover its normal condition, and so continued for six generations (32-37), 

 fertile pairs producing an average of 41 young. During this period 8 

 out of 59 pairs, or 13 5 per cent, were sterile. 



For the next five generations (38-42) the productiveness of the A 

 series was more than doubled. The rise occurred suddenly in generation 

 58, the parents for this generation having been mated in the last days of 

 February. As this was by no means the most favorable season of the 

 year, and the possibility of an accidental cross seems for a variety of 

 reasons excluded, it is difficult to account for the sudden change, unless 

 it is the expression of a cyclical variation in fertility. In this period 6 

 out of 51 pairs, or 11.7 per cent, were sterile, while the fertile pairs 

 produced on the average 137 young. Two other series, M and N, 

 started in the previous October, and kept under similar conditions, were 

 at the same time producing between 200 and 300 young, or even 

 more. 



