BI PARENTAL INHERITANCE IN PARAMECIUM 



437 



TABLE 44 



Experiment 16. Paramecium caudaium. Distribution of deaths among the mem- 

 bers of the 241 pairs of table 51, for five periods following conjugation, with the 

 probabilities of the observed results, if the distribution of deaths has no relation 

 to the pairing. The table is divided into two parts: in the upper half are given 

 the data when we include all of the 4^2 strains that were cultivated {m = 482) . In 

 the lower half are given the data in case we omit entirely the members of the eight 

 complete pairs that died during the first period, in order to exclude any possibility 

 that their deaths may have been due to the handling of the pairs while still united 

 (see text). This leaves a total of 466 to be considered, in place of 482. See the 

 detailed explanation following the table. 



27 ! 57 



37 j 63 



47 j 76 



68 145 



(Omitting the 8 pairs of the first period, m = 466) 



3 



4 



6 



29 



4 i 1 



7 I 3 



10 I 4 



32 ! 10 



1 to 3.1 

 5.5 to 1 

 10.3 to 1 

 247.6 to 1 



give the number of individuals dead, and alive, respectively, at the end 

 of the periods given in column (1). Columns (3) and (7) give the most 

 probable numbers of complete pairs that would be included among these 

 if there were no relation between pairing and the distribution of deaths; 

 this is determined by formula (1), page 461. 



Columns (4) and (8) give, on the other hand, the actual number of 

 complete pairs included ; while columns (5) and (9) show how much the 

 actual number exceeds the most probable number. (It is to be observed 

 that this excess is the same for those dead as for the survivors.) Column 

 (10) gives the probability that any deviation so great as this (either by 

 excess or by deficiency) should occur. This is determined in accord- 

 ance with rule (5), in the Appendix. (If we should compute the proba- 

 bility for so great an excess, it would be much smaller than the proba- 

 bility given in column 10.) This probability shows directly in how 

 many case out of a thousand or a miUion, et cetera, so great a deviation 



