352 H. S. JENNINGS 



pared with those between the diverse Hnes derived from the 

 conjugants. Furthermore, taking the most extreme case of 

 hne 3 ax, with but 18 fissions, we find that the sister hne, 3 ay, 

 derived from the same parent, does not show a low rate of fission; 

 so that the slow rate is not characteristic of this entire line. In 

 the conjugant hnes, on the other hand, the rates of the two or 

 more sets derived from a single individual we found to correspond 

 closely, showing that the characteristic is an inherited one. It 

 would then appear on the whole probable that all the differences 

 seen among the lines derived from the non-conjugants are simply 

 the slight fluctuations unavoidable where a large number of lines 

 are cultivated. 



The question may be tested for both sets in another way. If 

 the differences between different lines are matters of inherited 

 differentiation, then of course lines having a fast or a slow rate 

 in one part of the period of the experiment should have a cor- 

 responding rate in the other parts. That is, the rates of fission 

 for earlier and later periods should be correlated. We may there- 

 fore determine the coefficients of correlation for successive 

 periods, in both the conjugants and non-conjugants; this will tell 

 us whether the rates of fission are, as a rule, inherited in the 

 different lines. 



I have worked out for both sets the correlation between the 

 numbers of fissions in each line (1) in the first half of the experi- 

 ment compared with the second half; (2) in the second third 

 (September 3 to 8 or 4 to 10) compared with the last third (Sep- 

 tember 10 to 16 or 12 to 18). This latter comparison was made 

 owing to the fact that the direct physiological effect of conjuga- 

 tion appears to obscure the characteristic differentiations, for 

 some days after conjugation. 



Furthermore, I have worked out, for the entire period of the 

 experiment, the correlation between the sister lines, x and y, 

 derived originally from a single member (of a pair or split pair). 

 If the differences in rate of fission are inherited, these two sister 

 lines should of course be similar, giving a positive coefficient of 

 correlation. The correlation, for both conjugants and non- 

 conjugants, is given in table 25. 



