198 H. S. JENNINGS AND K. S. LASHLEY 



the correct coefficient of correlation is found to be 0.5744 + 

 0.0487. This coefficient is considerably higher than that found 

 in the senior author's extensive study of the matter ('11), to exist 

 between the conjugants in any of the eight wild cultures examined. 

 The mean of all the 86 means is found to be 40.33 units (161.32 

 microns); the standard deviation of the. means is 13.60 microns; 

 their coefficient of variation 8.43 per cent. 



The question may be raised as to whether the proper method of pro- 

 cedure for determining the coefficient of correlation would not be, to 

 mate in the correlation tables every specimen of a of a given pair, with 

 every specimen of h, of the same pair, thus giving a very large number 

 of pairings (in this case 41,900), from which to compute the coefficient. 

 This method of work might be useful for certain purposes, but our pre- 

 cise object is to determine whether the mean sizes of the lines a and 

 6 of the pairs are correlated and how much; this the method suggested 

 would not do. By including among the matings those of young, small 

 specimens of a with old, large specimens of b (and vice versa) the coeffi- 

 cient would be much reduced; so that even if the mean sizes of the mem- 

 bers pairs were perfectly correlated (mean size of a and b equal in each 

 pair), this method might give a low coefficient of correlation. We have 

 worked out the coefficient by this method; mating each of the descen- 

 dants of a of each pair with those of the corresponding b (41,900 pairings) 

 gives a coefficient of correlation of 0.2871 — almost exactly half that of 

 the correlation between the means of a and b. 



Our result, then, is that in consequence of conjugation, the co- 

 efficient of correlation of 0.3881, due to assortative mating, is 

 increased to 0.5744 — an increase of 48 per cent. This increase 

 can be attributed onlj^ to biparental inheritance. As a result of 

 conjugation, the progeny of the two members of the pairs become 

 more alike in size — just as our previous study had shown them to 

 become more alike in their rate of reproduction. 



It may be of interest to compare the similarity in size of these 

 members of pairs, with the similarity in rate of fission of the same 

 lines. The lines were, taken for measurement merely as they 

 happened to come to hand, without any knowledge of their rates 

 of fission. In table 3 each line is given the same number as in 

 table 51 of our previous paper ('13), so that any comparison 

 desired can be made. We will select for examination the number 

 of fissions for each a and b of table 3 for the first twenty days of the 



