480 AUSTIN RALPH MIDDLETON 



In order to be sure of the uniformity of the bacterial con- 

 tent of the shdes, on March 18, an equal small quantity of the 

 liquid from each of the mass cultures was added to the fresh 

 medium. It was added two or three drops at a time and each 

 few drops carefully studied in a watch glass under the binocular 

 before it was added to the fresh culture medium. On March 20, 

 each animal transferred to the fresh slides was washed in a watch 

 glass in a mixture of equal quantities of the culture medium 

 from the mass cultures instead of being washed in fresh culture 

 medium. 



Now at the end of this fifty-day period of balanced selection 

 the two sets had experienced eighty days of opposite selection 

 immediately followed by one hundred and two days of no selection. 

 If the difference remains after this test it is evident that selec- 

 tion has in this case produced an heritable difference in fission rate 

 within the clone. Table 9 shows that this difference of average 

 fission rate does persist. Figure 11 shows the rate of division 

 of the two sets of lines, averaged for ten-day periods, during 

 this balanced selection test. 



Experiment 1-D. Reversed selection, April 13 to June 1, 1914. 



A second experiment in reversed selection was started on April 

 13, 1914, from lines derived from Experiment 1-B, the ninety- 

 day balanced-selection experiment described above. Hence they 

 had been subjected to eighty days of opposite selection and this 

 was followed by eighty days of balanced selection before re- 

 versed selection was started. This experiment was prolonged 

 through five ten-day periods with daily transfer to fresh slides 

 and during .the whole time the average fission rate of the fast- 

 selected 'slow' lines was higher than that of the slow-selected 

 'fast' lines. And on the whole there was a gradual increase of 

 this average difference as reversed selection proceeded. Here 

 again selection has altered the fission rate and the new rates 

 were hereditary. Figure 12 and table 10 show graphically the 

 results of this experiment. 



Attention is called in table 10 to the large average number of 

 selections per line for the fifty days of this reversed selection. 



