466 AUSTIN .RALPH MIDDLETON 



merit 1. For the first period of this part (sixth period of the 

 entire experiment) the difference was 1.53 generations per hne 

 per ten-day period, for the next it was 3.41 generations and for 

 the last it was 7.51 generations. The small difference of 1.53 

 between fast and slow in period 6, following upon a period when 

 the difference was 3.57, is due to a slowing of the fission rate in 

 all lines during period 6, owing probably to low temperature, 

 and perhaps partly also to one of the rhythms emphasized by 

 Woodruff and his colleagues. The percentage of difference in 

 proportion to the total average fission rate is in reality greater 

 in period 6 than in preceding periods. Thus, for period 4 the 

 average number of fissions for all sets was 15.925, and the differ-, 

 ence betw^een the fast and slow was 2.43 — this difference being 

 thus 15.25 per cent of the average rate. In period 6, the small 

 difference 1.53 is actually 27.34 per cent of the average rate 

 for all. Table 5 gives the actual number of generations pro- 

 duced by each of the lines during the three consecutive ten-day 

 periods, the number of selections that were made in each line 

 and the average difference per line for each ten-day period. 



Table 5 and figure 3 demonstrate in this experiment a con- 

 tinued increase of the difference between the average rates of 

 fission of the two sets of lines. As has each of the previous 

 experiments, so also does this one show that the fission rate 

 within the clone may be changed by selection. During its three 

 ten-day periods each fast line produced on the average 0.415 

 generations more per day than each slow one. On only one day 

 during the thirty days of this experiment did the slow lines 

 produce more generations than the fast lines and on that day 

 only an average of 0.3 generation per line, as shown by figure 

 5. The two sets indeed hardly overlap at all in their rates, 

 practically all of the fast set being faster than any of the slow 

 set. This is shown by the curves of variation of the two sets 

 in figure 6. Only one slow line produced as many generations 

 as the slowest fast line. 



