A PHYSIOLOGICAL CHARACTER. 75 



broods of young produced in the laboratory and continued for 29 

 months, 91 generations in the plus strain and 90 generations in the 

 minus strain. 



For the first longer period (December 1914-July 1915) of the 

 experiment the means for the plus and minus strains (358 and 288 

 individuals) were, respectively, 529.4 and 451.8 seconds (table 33). 

 The difference (+77. 6 15. 6 seconds) was 4.97 times the probable 

 error. For the first single month of selection (December 1914) the 

 plus strain was more reactive by 40 seconds and again in June 1915 

 the plus strain was the more reactive strain, this time by a large 

 margin, 274 seconds; but for the other 6 months of the first longer 

 period the minus was more reactive by differences ranging from 88 

 to 219 seconds. For the same-day broods (table 34) the means 

 were 475 and 359 seconds, the plus strain having a higher mean by 

 116 25.78 seconds. Of the 7 same-day broods for this period, 5 

 occurred in succession in January and February 1915, almost im- 

 mediately after the beginning of the experiment. The differences 

 were +175, +207, +193, +140, and +152 seconds. Data for a 

 pair of same-day broods occurring just before these five are incomplete, 

 but the minus brood was the more reactive by at least 196 seconds, 

 while the pair of broods just preceding those last mentioned were 

 likewise same-day broods with the plus the more reactive by a small 

 margin (27 seconds). Thus there were in effect 7 successive same-day 

 broods, in all of which the minus brood was considerably the more 

 reactive. These differences are so large and so persistent as to indi- 

 cate for this early and limited period (January-February 1915) a 

 real difference in reaction-time, with the minus strain the more 

 reactive. In the last 2 months of this longer period (to August 1915) 

 the minus strain had much the larger reaction-time, and this differ- 

 ence was greatly increased in the next two-month period and con- 

 tinued very large for another two-month period. 1 



The second longer period (the year August 1915-July 1916) 

 gave averages of 483.8 and 589.1 seconds. The difference was 

 105. 3 14.3 seconds, or 7.36 times the probable error. For the 

 same period the 5 same-day broods had a difference of 27 seconds. 

 This year's data, taken as a whole, might at first thought be con- 

 sidered suggestive of an effect of selection ; but the great irregularities 

 of the curves, the small difference between same-day-brood means, 

 and the large differences in the opposite direction for the preceding 

 year do not favor such an interpretation. A test series conducted 

 during July 1916, consisting of nearly 1,000 individuals of each 

 strain, gave means differing by only +0.7 9. 7 seconds. This test 



1 The very great divergence in the two reaction-time curves, the minus strain being the 

 less reactive of the two, for the six-month period June to November 1915 is no more easily 

 accounted for than the earlier six-month period, during which the plus strain was markedly the 

 less reactive. 



