236 



The Journal of Heredity 



ences. but more important from the 

 standpoint of heredity is a similar 

 table in which the marriages are 

 weighted by the number of relatives in 

 Table 3 connected with each. In 

 Table 5 the marriages were weighted 

 by the number of the children with an 

 additional weight of one-third in the 

 parental entries (as opposed to the 

 grandparental ) in order to balance ap- 

 proximately the fraternal correlation 

 among the children in Table 3. 



The results are presented graphically 

 below. In these diagrams, the solid 

 lines represent the per cent of cases in 

 which a given combination is found in 

 relatives of the various kinds, and the 

 broken lines represent the sinilar per- 

 centages for matings (weighted). 



NOXASSORTATIVE M.\TIXG 



The graphs on the whole tell a very 

 consistent story. The heredity of the 

 different combinations is obvious. In 

 each of the cases those of a given com- 

 bination form a larger per cent of the 

 relatixes of their f)wn color than of any 

 other. In a general way the colors 

 fall into the order of relationship in 

 which thev are here arranged. Dark 



eyed dark, dark eyed light, dark eyed 

 red. light eyed red, light eyed light with 

 light eyed dark comi)leting the cycle, 

 showing relationship to dark eyed dark 

 as well as light eyed light. A consider- 

 ation of merely the correlation among 

 relatives, however, gives a very imper- 

 fect idea of the connection between the 

 colors. There is approximately at least 

 equal transmission by the parents .so 

 that one would expect that the combi- 

 nation for which a given color shows 

 most preference in marriage would also 

 show preponderance among the chil- 

 dren. But it happens to a surprising 

 extent that the reverse is true. In gen- 

 eral these peojile seem to have preferred 

 to marrv those of the color combina- 

 tion most remote from their own, so 

 that where the line of preferential mat- 

 ing is depressed the line of relatives is 

 elevated and the reverse. In order to 

 gi\e a rough measure of the nonassorta- 

 ti\e mating, the rank of frequency as a 

 husband or wife was tabulated against 

 the rank of fre(|uenc\- as a relative for 

 each color in connection with each color. 

 This means tliat rank in the vertical 

 column in Table 3 was correlated with 



Table 4.— Per cent of marriages of each kind made by each kind of individual (reading hori- 

 zonlally). For measurement of degree of assortative mating {or reverse) read vertically. 

 Based on 182 marriages. 



'rAHi.K 5.— .^awjc as Tabled, except that the number of marriages of each kind is weighted by 

 the number of cffsprings produced and the parental generation is given more weight than 

 the grandparental. (See IcxL) 



