86 SELECTION IN CLADOCERA ON THE BASIS OF 



opposite direction for the selection data as a whole and the larger 

 difference in the opposite direction for the same-day-brood data for 

 this year. Another test series was conducted during this period, 

 during June 1916; it contained 1,231 individuals in the plus strain 

 and 1,282 in the minus strain. The mean reaction-times were 420.2 

 and 525.7 seconds. The difference ( — 105.5 ±10.3 seconds) is 10.24 

 times the probable error. The result of this test series, conducted 

 only 3 months later than the earlier one, is quite contradictory to the 

 result of the former. 



The results of these test series are somewhat puzzling, but table 

 39, giving the selection summary, shows that for the three-month 

 period February- April 1916 (including the month in which the 

 earlier test series was made and that just preceding and just succeed- 

 ing) the mean reaction-time was much higher for the plus strain. The 

 similar period of 3 months, May-July 1916, during which the June 

 test series was conducted, shows a mean reaction-time in the minus 

 strain 135.7 seconds greater than that for the plus strain. One sees, 

 in these test series and the selection data for the contingent periods, 

 that the relative reactiveness had changed here in a comparatively 

 short time. Reference to the curve, figure 15, shows that for the 

 two-month period, February-March 1916, the two curves met, but 

 for the later two-month periods the minus strain for a considerable 

 time was much less reactive than the plus strain. 



For the final longer period of the experiment with Line 740 

 (August 1916 to the close of the experiment, May 1, 1917) the mean 

 reaction-times were 311.1 seconds and 395.8 seconds for the two 

 strains. The difference (— 84. 7 ±16.2 seconds) was 5.23 times the 

 probable error. There were only 4 same-day broods, for which the 

 mean reaction- times were 358 and 221 seconds, the difference being 

 + 137 ±34.03 seconds. Here the number of same-day broods was 

 too small to be given serious consideration, but it is interesting to 

 note that with these same-day broods the plus strain was so much 

 less reactive than the minus, in opposition to the difference shown 

 by the data as a whole for this period. Two of these 4 same-day 

 broods, however, occurred during the final 3 months of the experi- 

 ment, when the data as a whole showed that the minus strain was 

 temporarily the more reactive. 



Reference to the curves of the mean reaction-times (figure 15) 

 shows that for 10 months (from April 1916 to January 1917, inclusive) 

 the minus strain was the less reactive and that the margin of differ- 

 ence increased until in the December 1916 and January 1917 periods 

 the reaction-time for the minus strain was more than double that 

 for the plus strain. The data for a portion of this ten-month period 

 was not very satisfactory. We were troubled with poor food con- 

 ditions, and, at times, when the stock was not well nourished, the 

 selections were not made by tests, but individuals were chosen at 



