BIPARENTAL INHERITANCE IN PARAMECIUM 447 



other experience. Hence no conclusions are warranted as to the 

 comparative mortaUty rate of the detecendants of pairs, and of 

 spUt pairs, from comparing Experiments 16 and 17. 



2. This is merely one example of a general principle which is 

 impressed on one throughout all the many experiments on which 

 the present series of papei's is bascid. For comparison as to mor- 

 tality, vigor, fission rate and the like, one must always compare 

 two lots that are cultivated together, so that all external conditions 

 are the same for each set. Comparisons of such matters in sets 

 cultivated at different times, and therefore necessarily under 

 different conditions, are bound to give fallacious results. 



Distribution of survivals and deaths. The experiments which 

 precede the present one show that after conjugation the two mem- 

 bers of a gi\'en pair tend to have the same fate (either survi\'al 

 or death for both). In the present experiment with split pairs 

 we have the same conditions before us, save that the two paired 

 members have not conjugated. Do such split pairs show the 

 tendency to have the same fate? 



In determining this, certain facts are to be noted. It is neces- 

 sary to use a certain amount of \'iolence in separating some of 

 the split pairs, and this may cause injury. If so, this injury 

 will probably affect both members of the. split pair, so that in 

 consequence both may die. Thus one might expect to find among 

 any that die shortl}^ after separation, and without having divided, 

 a greater number of split pairs than would occur if the distribu- 

 tion of deaths were not influenced by this common injury to the 

 two members of the split pairs. It is therefore to the period 

 following the first three or four daj'S of the experiment that we 

 must look for evidence as to anj^ intrinsic relation tending to 

 induce the same fate in the two members. In the case of pairs 

 in which conjugation was completed, we found that the tendency 

 for the two members to have a similar fate increased greatly in 

 the later periods of the experiment. 



In the first four days of the present experiment with split pairs 

 there died 26 lines, all save one without fission. Among these 

 26 lines were both members of 5 split pairs, all dying with- 

 out fission. Applying our formula (1), we find that when 26 



