BIPARENTAL INHERITANCE IN PARAMECIUM 433 



jugants 0.2635; so that that for the conjugants is about 20 per 

 cent greater. 



It would seem therefore that assortative mating accounts for 

 at least a part of the similarity between the progeny of the pairs. 

 Whether any part of it is due to inheritance can hardly be deter- 

 mined from our data up to this point, since the probable errors 

 are in all cases large, owing to the small numbers of pairs in the 

 experiments. 



^Vhat we need clearly is, data based on larger numbers. Such 

 will be supphed in Part II. 



PART II 



H. S. JENNINGS AND K. S. LASHLEY 



EXPERIMENT 16: 241 PAIRS OF CONJUGANTS PROPAGATED SEPAR- 

 RATELY FOR FORTY-SEVEN DAYS 



On the work presented in Part I the criticism may be made that 

 the numbers dealt with are hardly sufficient to place beyond all 

 doubt the conclusions to which they lead. This applies particu- 

 larly to the results shown in the measurements of the similarity 

 in rate of reproduction between the descendants of the two mem- 

 bers of pairs; the coefficients of correlation are variable and their 

 probable errors are large. It hardly applies to the study of the 

 distribution of the deaths among the excon jugants, since here we 

 had the relatively large number of Miss Cull's valuable experi- 

 ment (93 pairs) to work with, but these results should be tested 

 by others. It is true that the difficulty in dealing with large 

 numbers is very great when the rate of reproduction for each 

 strain must be recorded for long periods. But the fact that it is 

 difficult to work with large numbers does not lessen the insecurity 

 of results drawn from small ones. 



Furthermore, all the results thus far have been drawn from 

 experiments designed for other purposes; the question arises as 

 to whether all conditions have been fulfilled for getting accurate 

 results on the present problems. One point in particular suggests 

 itself. In the experiments thus far, the two members a and b 



