STATISTICAL STUDY OP CONJUGATION IN PARAMECIUM. 185 



simplest prototype of sexual mating, namely the process of conjugation. 

 The form chosen for study was the common ciliate infnsorian, Parame- 

 cium candatum. The length and breadth of each individual in 200 conju- 

 gating pairs of Paramecia were measured. Correlation tables were then 

 formed relating given characters in the two individuals of the pair. From 

 these tables the correlation coefficients were deducted by well known 

 statistical methods. The values of some of the coefficients determined are 

 exhibited in the following table : 



Coefficient of 

 Character studied. correlation. 



Length of the conjugating- individuals .5514 ± .0337 



Breadth of the conjug-atine- individuals 2959 ± .0435 



Length of first* and breadth of second individual .1717 ± .0465 



Length of second and breadth of first individual 1606 ± .0465 



It is seen at once that in the case of direct correlations (length with 

 length, breadth with breadth) the values here are higher than those of 

 the corresponding coefficients in the case of man. This difference is 

 especially marked in the case of the character "length" where we have in 

 Paramecium roughly twice as close a degTee of "homogamy" as is found 

 in man. 



Before we can attach any significance to these results, it is, of course, 

 necessary to determine whether the relatively high correlations observed 

 are not spurious and arise mereh' because all of the Paramecia in a given 

 culture are presumably of very nearly the small size. If this were the case, 

 random pairing ought to give sensibly as high values for the correla- 

 tion coefficients as those actually observed, because on this view the 

 conjugation is random, and in the long run one random jjairing is the 

 same as another. To test this matter, the individual record cards were 

 shuffled together and drawn out in pairs entirely at random. The coeffi- 

 cient of correlation between the lengths of such fortuitously mated indi- 

 viduals proved to be .0223 ± .0471, or, in other Avords, there was no sensible 

 correlation between the members of random pairs. 



It is impossible in the space of a brief abstract to discuss fully the 

 precautions taken in the work to guard against possible sources of error, 

 both in fact and in interpretation, but we believe that our complete re- 

 sults justify the conclusion that there is a "real assortative mating" in 

 the conjugation of Paramecium. Furthermore, this tendency of like to 

 pair with like is more marked in Paramecium than in man. In the com- 

 plete paper the factors to which this is probably due will be discussed. 



(A complete account of this work, together with the statistical data, 

 will be published in another place.) 

 Raymond Pearl^ Ph. D., Instructor in Zoology, 



Mary J. Burr^ 



Zoological Laboratory, University of Michigan, 

 Ann Arbor. 



♦'•First" and "second" here refer to the individuals in a pair which happened to be measured 

 first and second respectively. 



24 



