A PHYSIOLOGICAL CHARACTER. 59 



means of 344.8 seconds for the plus strain and 331.8 seconds for the 

 minus strain. The difference, +13.0 ±12.9 seconds, is not of statis- 

 tical significance. A second test series conducted in November 1914 

 contained 1,055 individuals in the plus strain and 1,089 in the minus 

 strain. The difference in the mean reaction-time was +18.9 ±7.72 

 seconds, 2.46 times the probable error. A third test series was con- 

 ducted during March 1915. The numbers were large, 1,123 and 1,133 

 individuals in the two strains. The mean reaction-times were 421.9 

 and 396.6 seconds. The difference was +25.3 ±8.0 seconds and is 

 3.16 times the probable error. 



It will be seen that there is a rather striking similarity in the 

 differences obtained for the 3 test series and for the entire selection 

 data for this year-period, the differences being +13.0, +18.9, and 

 +25.3 for the 3 test series and +15.8 seconds for the selection data 

 as a whole. Such harmonious results seem to indicate that the plus 

 strain of Line 714, during this year-period, was actually less reactive 

 in spite of selection, although the same-day-brood data oppose this 

 conclusion. 



For the next year-period of selection with Line 714 (August 1, 

 1915-July 31, 1916) the means were 301.1 and 303.5 seconds. The 

 difference was —2.4 ±10.9 seconds. There remains a short period 

 of 13^2 months during which selection was continued. There were 

 only 66 individuals in the plus and 53 individuals in the minus strain. 

 The difference (+63.8 ±32.4) is not statistically significant. 



Considering the data for Line 714 as a whole, it is clear that 

 there is not an effect of selection. 



Figure 7c, showing the reaction-time curves for the two strains 

 of Line 714, presents curves with large minor irregularities. It will 

 be noted that throughout the first 6 months of selection with Line 

 714 there is a considerable divergence, the minus strain having much 

 the higher reaction-time. Such consistencies in reaction-time differ- 

 ences for limited periods also occur in Line 695 for a period of 8 

 months and in Line 740 for a period of 10 months. It would seem 

 that these are more than mere chance results, but in each of these 3 

 cases further selection not only did not result in increasing the 

 divergence, but in all 3 of these cases the divergence disappeared. 

 Assuming for the moment that an effect of selection was present in 

 these cases, it was later lost; but in what manner this loss could 

 have occurred is not clear. Selections were not relaxed, and there 

 seems to be no reason to suppose that a mutation had affected one 

 of the strains in each of these 3 lines. 



However suggestive of an effect of selection the early portion of 

 the curves for Line 714 are, it is clear from the remainder of the 

 curves that a selective effect is not present (figure 7c). The detailed 

 analysis of the data has already shown this. 



