380 



Dr. K. Pearl. A Biometrical 



[Nov. 15, 



same character in each member, e.y., length of A with length of B (direct 

 homogamic correlation), and (b) a different character in each member, e.g., 

 length of A with breadth of B (cross homogamic correlation). A part of the 

 results for direct homogamy in respect to the character length are shown in 

 Table II. 



Table II. — Direct Assortative Pairing in the Conjugation of Paranueeium. 



Lengths only. 



Series. 



Characters. 



Coefficient 

 of correlation. 



Kuinber of 

 pairs. 



A 



Length of A* 



Length of B* 



0-5327 ±0-0333 



105 



C 



A 



B 



0-7249 ±0-0225 



101 



D 



„ A 





0-4302 ± 0-0972 



16 



B 



A 



B 



0-7941 ±0-0509 



12 



AA 



» A 





0-5882 ±0-0221 



200 



It will be seen that these coefficients are relatively very high. The only 

 quantitative determinations of the degree of homogamy in any organism 

 which have hitherto been made, are those for assortative mating in man, 

 made by Pearson and his collaborators.! Por the characters so far investi- 

 gated in man, the coefficients of direct homogamy fall uniformly below 03. 

 It is evident that we are dealing here with a much more intense assortative 

 pairing. 



For the other characters, breadth and index, the direct homogamic correla- 

 tions are not so high as for length, though still relatively large. Taking 

 the means of the results for the different series, I find for the direct 

 homogamic correlation of breadth with breadth v = 0-3028, and for the 

 correlation of index with index r = - 4291. 



The cross homogamic correlations (length of A with breadth of B and viee 

 ccrsCt) were found to be low in all cases. The average value of the coefficients 

 for these cross correlations was. - 1082. 



It will be noted that the homogamic correlations are hi all cases positive. 



5. In order to test whether the observed homogamic correlations were real 

 or spurious, various sorts of random pairings were tried. The first question 

 which might be raised is as to what would be the value of the coefficients 

 measuring homogamy, provided we experimentally paired together conjugants 

 or non-conjugants quite at random. This has been done several times for 



* A and B refer respectively to the first and second individuals of a conjugant pair to 

 be measured. 



t ' Biometrika,' vol. 1, p. 373, aud vol. 2, pp. 481 — 498. 



