BIPARENTAL INHERITANCE IN PARAMECIUM 417 



had no relation to the pairing. The most probable number of 

 pairs when 22 hnes die is 4 ; the actual number 5. The probability 

 for four is 0.306; for 5 it is 0.214, or about two-thirds as great as 

 for 4. Five pairs might therefore well occur purely as a matter 

 of chance distribution. 



Thus in this experiment the number of pairs included among 

 those that died is a little greater than would be expected if the 

 deaths had no relation to the pairing. The relation is again the 

 reverse of that which the theory of sexuality reauires. 



Experiment 2: Paramecium caudatum 

 This Journal, vol. 14, 1913, page 293 



In this experiment, lasting for eight weeks, 56 Unes derived from 

 members of pairs, 38 from split pairs, and 58 from unpaired 

 individuals, were cultivated. The designations of the unpaired 

 lines were arbitrarily grouped in pairs at the beginning, in order 

 to discover whether such arbitraiy grouping would giv'e the same 

 relations between its members as appear in the actual pairs. 



The death rate was very high in this experiment, for reasons 

 set forth in the preceding paper (p. 293) ; on this account it does 

 not furnish the best of data for determining whether there is any 

 relation between hability to death and the pairing. It will be 

 worth while however to set forth the facts in the case ; this is done 

 in table 37. I have included the facts, not only as to the pairs, 

 but as to the split pairs, and the free specimens. In the split 

 pairs, the two individuals that had begun pairing are called a 

 pair. In the free specimens, the pairing is purely artificial; these 

 are for control purposes. 



As the table shows, in no case does the actual number of pairs 

 dead differ from the most probable number by more than a single 

 one. This is precisely such a result as might be expected in 

 chance single cases. The deviations from the number that is 

 absolutely the most probable one occur, as will be seen, in the 

 case of the free specimens, in which the pairing is purely arbi- 

 trary^, as well as in the other classes. I have given in the table the 



