CASTLE. — INBREEDING, CROSS-BREEDING, AND SELECTION. 737 



A brood reared a month later (in January, 1903) contained 178 

 young, but was probably incomplete, since it was produced in a very 

 small jar containing an insufficient amount of food, and the food was not 

 renewed, as was the practice in ordinary cultures. 



Four control pairs taken from the stock-jar about four months later, 

 in May, 1903, produced broods numbering 174, 142, 79, and 157 respec- 

 tively. Tbese numbers are very low compared with those obtained from 

 control pairs in the previous fall, but it should be observed that the num- 

 bers obtained from the A series during this same period (generation 23) 



tionO 10 ! 15 ; 20 j p5 30! 35 40 | 45 50 j 55. j .59 



Oct. 1002 Jan. 1903 Apr.isXWJune 1903 Oct.l903Jan.'l904 Apr.1904 Julie 1904 Oct.1904Jan.1905Apr.1905 



Figure 1. Productiveness of fertile pairs in the A series. 



are likewise low. Very probably the falling off in both cases was due to 

 low temperatures in the building, at about the time when the steam heat 

 was discontinued, resulting in imperfect fermentation of the food, or 

 directly affecting the egg production of the females. For the broods 

 produced by the A series were larger both previous to this period and 

 subsequent to it. 



Broods of the control series were throughout this year several times as 

 large as broods of the A series reared simultaneously under identical 

 conditions. Further, no sterile pairs were encountered in the control 

 series, though a considerable percentage of sterile pairs was constantly 



VOL- XLI. — 47 



