354 H. S. JENNINGS 



Whether any significance is to be attached to this is doubtful, 

 since the value of the coefficient is but two-and-a-half times its 

 probable error; and a coefficient of this amount would occur once 

 in ten times as a result of chance distribution. Further, the two 

 sister lines x and y were kept in the two concavities of the same 

 slide, and one was changed immediately after the other. The 

 result of this may have been to keep the two under slightly more 

 uniform conditions than prevails for two individuals in different 

 moist chambers, giving rise to the slight correlation. The matter 

 will be investigated farther, but in any case it is clear that any 

 differentiation that may exist between the non-conjugant lines 

 is extremely slight; so that correlating the fissions of successive 

 periods gives no trace of it. 



The present experiment therefore clears up the difficulty left 

 by the results of Experiment 13. In that experiment, as shown 

 in table 33, the 'non-conjugants' exhibited inherited differentia- 

 tions, as did the conjugants. It seemed practically certain how- 

 ever that these 'non-conjugants' had gone through previous 

 conjugations, so that the observed heritable differentiations were 

 probably due to these previous conjugations. On page 337, 

 I pointed out the necessity for an experiment in which this matter 

 should be controlled. Our present experiment supplies this need; 

 we know that our non-conjugants here have not conjugated since 

 they came from a single parent individual. And our results show 

 that the inherited differentiations in Experiment 13 were indeed 

 due to conjugation; they do not appear when we deal with actual 

 non-conjugants (lines which have not conjugated since they were 

 all derived from a single individual). 



Even if it should turn out that the slight correlation shown by 

 X and 7j in the non-conjugants of the present experiment is due to 

 real differentiations between the lines, this result would not modify 

 our present conclusion in any essential way, since the differentia- 

 tion so indicated would be so slight as to be of quite a different 

 order from that produced by conjugation, the latter giving rise, 

 as we have seen, to coefficients as high as 0.9. Even a slight dif- 

 ferentiation arising during vegetative reproduction would be of 

 the highest interest, but it would not alter the positive fact of the 



