A PHYSIOLOGICAL CHARACTER. 85 



until the experiment was completed. Both strains are still in our 

 laboratory. 1 The data are summarized in tables 39 and 40 and figures 

 14 and 15. 



The first longer period of the experiment, approximately 11 

 months, contains the data for 37 and 36 broods of the two strains. 

 The mean reaction-times (427 and 357 individuals) are 735.2 and 

 757 seconds (table 39). The difference (-21.8±12.3 seconds) was 



1.77 times the probable error. There were 6 same-day broods for 

 this period (table 40) for which the mean reaction-times were 713 

 and 682 seconds, the difference being +31 seconds. 



For the second year-period of selection with Line 740, the mean 

 reaction-times (424 and 482 individuals) were 597.1 and 656.9 

 seconds. The difference was -59.8 ±13.5 seconds. This difference 

 was 4.43 times the statistical probable error. The 9 same-day 

 broods for this period gave as mean reaction-times 607 and 632 

 seconds, the difference being -25 ±31.18 seconds. 



A test series was conducted in April 1914. There were 653 

 individuals of the plus strain and 614 of the minus strain. The mean 

 reaction-times were 666.4 and 711.9 seconds. The difference was 

 -45.5 ±10.9 seconds, or 4.17 times the probable error. There is 

 a suggestion of an effect of selection for this year-period in that the 

 data as a whole, the data for the same-day broods, and the test- 

 series data all show higher reaction-times for the minus strain. But 

 in view of the later data this evidence is overruled. 



For the next year-period (August 1914-July 1915) the mean 

 reaction-times (452 and 454 individuals) were 642.5 and 607.7. The 

 difference (+34.8 ±13.8 seconds) is 2.52 times the probable error. 

 Hence the difference is of statistical interest, but is opposed to an 

 effect of selection. There were only 3 same-day broods for this year- 

 period for which the mean reaction-times were 834 and 815 seconds. 

 The difference was +19 seconds. 



For the year-period (August 1915-July 1916) the mean reaction- 

 times (471 and 519 individuals) were 574.6 and 600.0 seconds. The 

 difference (—25.4 ±14. 2 seconds) is only 1.79 times the probable 

 error. There were 14 same-day broods for this period, for which the 

 mean reaction-times were 558 and 640 seconds. The difference was 

 -82 ±21.53 seconds. 



A test series was conducted during March 1916 and contained 



1.078 individuals in the plus strain and 1,089 in the minus strain. 

 The mean reaction-times were 641.6 and 577.9 seconds. The differ- 

 ence (+63.7 ±9.5 seconds) "is 6.71 times the probable error. This 

 test series was conducted with great care, and the fact that the plus 

 strain was the less reactive of the two by a margin of statistical 

 significance largely counterbalances the small difference in the 



1 It was in one of the substrains of this line (740 - Special B) that the sex-intergrades were 

 first discovered. 



