A PHYSIOLOGICAL CHARACTER. 97 



Detailed Analysis of Data for Line 757. 



The curve of the reaction-times for Line 757 (figure 18b) begins 

 with the December 1912-January 1913 period, though it includes a 

 few reaction-time records obtained in November.^ The means for 

 this period are identical, but in the following period there is a slight 

 divergence (the minus strain being the less reactive) followed by- 

 irregular but in general increased divergences in the later two-month 

 periods, so that the course of the curves is from the start suggestive 

 of an effect of selection. 



There is one period of two months (June-July 1914) for which 

 the minus strain had the lower reaction-time and the difference was 

 fairly large (132 seconds; table 44). In 3 other two-month periods 

 the minus strains had mean reaction-times only 2, 20, and 48 seconds 

 higher than those for the plus strains for the corresponding periods. 

 But in all the other two-month periods of the experiment after the 

 first 5 months of selection the mean reaction-time for the minus 

 strain was from 55 to 629 seconds higher than that for the plus strain 



FiouEE 16. — Line 757. Relative rates of descent of the two strains. (See figure 2a for description.) 



(tables 43 and 44). Hence the mean reaction-time differences by 

 two-month periods are nearly all in the direction sought in selection, 

 and most of them are differences of statistical significance. Calculated 

 by longer periods, all the differences are of large statistical value, 

 the minus strains being the less reactive. This will be made clear 

 in the following detailed analysis. 



Reaction-time Means Compared by Longer Periods. 



For the first 9 months of the experiment (November 1912-July 

 1913) the means for the plus and minus strains (29 generations and 

 320 individual reaction- time records, and 28 generations and 261 



^ Data for two abnormally small broods of each strain were obtained during November. 

 These data consist of the reaction-times of only 6 individuals of the plus strain and 8 individuals 

 of the minus strain, and of course give averages of little value, bo that it seems best to combine 

 these data with the data for the following two months in constructing the reaction-time curves. 

 The means (see tables 43 and 44) for the two strains for November alone are 685 seconds for 

 the plus strain and 900 seconds for the minus strain. All 8 of the minus strain and 4 of the 6 

 of the plus strain were "over-time" individuals. These meager data have little significance, 

 except to indicate that both strains were very slightly reactive at the beginning of the experi- 

 ment. For the December 1912-January 1913 period alone (excluding the few November 

 records) the means are 747 and 718 seconds. Hence the curves with the data for November 

 separate would start with the minus strain, the less reactive by 215 seconds, to be followed by 

 the December-January period, when the plus was the less reactive by 29 seconds. In tables 

 43 and 44 the data are presented with the November data separate from those for the December- 

 January period. 



