104 SELECTION IN CLADOCERA ON THE BASIS OF 



that for the plus strain. The means for both strains were lower than 

 for any other longer period of the experiment (313.1 seconds for the 

 plus and 625.9 seconds for the minus strain, table 45). The difference 

 was 312. 8 18. 37 seconds. The mean difference in reaction-time 

 was slightly increased. Because of the much smaller total number of 

 individuals, however, the probable error of the difference was larger, 

 though the difference was still 17.01 times its probable error. There 

 was only one same-day brood. Although from a single pair of broods 

 little may wisely be assumed, it is interesting to note that the mean 

 for the plus brood was 93 seconds, while that for the minus was 

 814 seconds, the latter mean being almost 9 times the former. 



While the reaction-time curves by two-month periods (figure 

 18s) clearly show the wide and increasing divergence in reactiveness 

 between the two strains of Line 757, curves by six-month periods 

 (November- April and May-October) are also given, together with 

 similar curves for Line 740 for comparison with them (figure 19). 

 Naturally these curves are less affected by local fluctuations and are 

 smoother curves. 



OTHEB FEATURES OP THE DATA. 



Examination of other features of the data for Line 757 brings 

 out similar evidence of a marked effect of selection. 1 The average 

 minimum reaction-time 2 for the two strains is a point on which their 

 reactiveness may be compared. The average minimum reaction- 

 time for the plus strain for the longer periods of the experiment was 

 progressively lowered from 468 to 357, 282, 190, and 94 seconds 

 (table 45). For the minus strain there was likewise a lowering, 

 though not so great nor so consistent. The minus strain average 

 minima were 639, 548, 380, 467, and 245 seconds (table 45). For 

 the plus strain the average minimum for the last long period was 

 only one-fifth that for the first of the 5 longer periods; for the minus 

 strain the minimum for the last period was two-fifths that for the 

 earliest period. The average minima for the two strains show wide 

 differences throughout the experiment, the differences being 171, 

 191, 98, 277, and 151 seconds for the 5 longer periods. These differ- 



1 In a single minor detail the data for a small portion of the time do not seem consistent 

 with an effect of selection. This point is the relative numbers of negatively reacting individuals 

 in the two strains. From August to December 1915 there were considerably more of the nega- 

 tively reacting individuals in the plus (21) than in the minus (5) strain. This is of course con- 

 trary to one's expectation in a less-reactive strain. However, 19 of the 21 negatively reacting 

 individuals in the plus strain for this period occurre^ in four broods. Since in all strains negatively 

 reacting individuals are extremely irregular in their occurrence, and when they do occur are often 

 relatively numerous within a single brood, it seems probable that these reactions are in large 

 measure determined by environmental conditions and hence are of little, if any, real significance 

 (cf. page 15). If, however, they were to be ascribed an important significance here, this data alone 

 (and only a very limited portion of it) runs counter to the plain implication of all the rest of 

 the data for Line 757. During the remainder of the experiment the plus and minus strains of 

 Line 757 had approximately equal numbers of negatively reacting individuals (tables 43 and 44). 



1 This, of course, includes the reaction-time of only a single individual (the most reactive 

 one) of each brood. Like the data for the average maximum reaction-time, it has much less 

 value as a measure of reactiveness than the mean reaction-time for all the individuals tested; 

 nevertheless, it is highly significant. 



