BIPARENTAL INHERITANCE IN PARAMECIUM 431 



fissions in each strain for the period set forth; it is entirely possible 

 that two weeks should when considered separately each give a 

 positi\'e coefficient, while taken as one period they should give a 

 negative one, or vice versa, so that the coefficients for the longer 

 periods furnish independent data. In all cases except the three 

 entries where the words ' all pairs' are added, there is included in 

 each case only the strains in which both members of the pair 

 lived through the period in question. In the three where the 

 words 'all pairs' are added, I have given the coefficient which 

 appears when we include the number of fissions for all the strains 

 that entered the experiment, whether they lived through the 

 period in question or not; taking thus, in the case of Experiment 

 1, all the Hotals' of table 29 of the preceding paper. 



In every case, as table 43 shows, there is a positive coefficient 

 of correlation between the progeny of members of pairs. In 

 one or two cases this is small and would hardly be significant 

 in comparison with its probable error, if it stood alone; but in 

 most cases the coefficient is of very considerable value. The 

 coefficient persists and is even in Experiment 1 increased, when 

 we include the number of fissions up to the time of death in the 

 strains that died before the end of the period in question, as in 

 the last two entries for Experiment 1. This is due to the fact 

 that there is a similarity in the two strains as regards length of 

 life; a point more fully brought out in the study of the distribu- 

 tion of mortality, already made. 



The data from Experiments 2 and 3 of the paper on the effects 

 of conjugation are less full than those for Experiment 1, the 

 members of pairs being still smaller. The coefficients are given 

 in table 43 for what they are worth; they confirm so far as they 

 go the results from Experiment 1. 



The question which next arises is as to the cause of the fact that, 

 as shown by these positive coefficients, the progeny of the two 

 members of a pair resemble each other in rate of reproduction 

 more than do those of unpaired individuals. Here there are two 

 possibilities : 



1. The resemblance may be due to inheritance from both par- 

 ents. This would fully account for the facts, and appears a highly 



