124 SELECTION IN CLADOCERA ON THE BASIS OF 



Considered by longer periods, it is found that during the first 

 (9 months) and second (12 months) longer periods of the experiment 

 the plus strain of Line 757 was less reactive than the Line 740 pus 

 strain by 39.0 12.36 seconds and 86.613.11 seconds, differences 

 of statistical value, 3.15 and 6.6 times the probable errors. During 

 the following year the plus strain of Line 757 became the more re- 

 active, though the difference for the entire year-period was only 

 37 13. 61 seconds, or 2.76 times the probable error. During the 

 next year, August 1915-July 1916, the difference is 116.4 13.67 

 seconds, 8.51 times the probable error the plus strain of Line 

 757 being the more reactive. The averages for the final nine- 

 month period of selection differ by 1.9 15. 89 seconds, the plus 

 strain of Line 740 having the lower reaction-time. This on its face 

 distinctly controverts the existence of a significant difference in 

 reactiveness between the plus strains of lines 757 and 740. But, as 

 already noted, in one two-month period during this longer period 

 there was an exceptionally low average for the plus strain of Line 

 740 and an exceptionally high average for the plus strain of Line 

 757, and the striking influence of these two aberrant averages pro- 

 duces the difference noted above, a difference which is quite out of 

 line with the general behavior for the latter part of this experiment 

 of the two strains concerned. 



The curves for Lines 740 and 757 by six-month periods (figure 

 19) show the plus strain of Line 757 the less reactive until the May- 

 October 1915 periods, from which time the 757 plus strain is shown 

 the more reactive for the remainder of the experiment. The ex- 

 ceptional reaction-time periods (December 1916-January 1917) for 

 the two strains compared serve to cause the curves to approach for 

 the last six-month period, but the plus strain of Line 757 maintains 

 the lower level. 



Hence it is seen that, whereas on the start the plus strain of 

 Line 757 was fundamentally less reactive than the plus strain of 

 Line 740, during the latter half (nearly) of the experiment there is 

 fairly conclusive evidence that the plus strain of Line 757 was more 

 reactive than the plus strain of Line 740, which may fairly be con- 

 sidered as a check, since it was subjected to the same treatment as 

 Line 757 and throughout a contemporaneous period of time. 



Comparing the mean reaction-times by two-month periods for 

 the plus strain of Line 757 with the means for the plus strains of the 

 remaining lines of S. exspinosus subjected to selection, Lines 794, 

 795, and 796, the result is still more suggestive of a changed reactive- 

 ness of the plus strain of Line 757. The remaining 3 lines are lumped 

 in this comparison, a procedure justified only in the interest of 

 brevity and the securing of less fluctuating averages. The plus 

 strain of Line 757 was less reactive than the combined plus strains 

 of Lines 794, 795, and 796 in only 3 out of 15 two-month periods, the 



