Page:Proceedings of the Royal Society of London Vol 60.djvu/307

Rh (6) Turning to the results for daughters, we have th e following table for the coefficients of correlation and regression :—

Mathematical Contributions to the Theory of Evolution. 281

Table I I I .—Inheritance of S tature by D aughters. Fathers and elder daughters. .. . Fathers and younger daughters. Mothers and elder daughters .. . Mothers and younger daughters 0 "4829 db 0-0220 0-4376 ± 0 -0 2 3 6 0-3953 ± 0 -0 2 5 0 0-4542 ± 0 -0 2 3 0 0 -4528 0 -396L 0 -4293 0 -4763

These results, more num erous than those for sons, are, for reasons which I am unable to explain, m uch more divergent. W e may note the following points :—

(i) There is a sensible difference between the coefficients of correlation for either parents w ith younger and elder daughters. Thus, the difference of the coefficients for fathers w ith elder and younger daughters is 0-0453, and the probable error of this only 0 0 3 2 ; while for mothers the corresponding difference is 0"0589, and the probable error of the difference only 0 -0328. The difference, however, is in the opposite sense. W e are thus face to face w ith an increasing maternal and a decreasing paternal influence on the stature of daughters. In other words, our statistics are entirely opposed to any steady telegonic influence on the statu re of daughters. If such a thing were conceivable, we should be confronted with the case of the mother influencing the father, the reverse of telegony.

(ii) The mean correlation of fathers and daughters is very slightly higher than th at of m others and daughters (0"4602 as compared with 0 4247). Thus, to judge by the mean coefficients of correlation, the father is slightly more prepotent than the m other in heredity. The mean coefficients of regression are for fathers 0 ’4'244, and for mothers 0"45*28, or in the ratio of 1 : l -067, b u t the ratio of the paternal to the m aternal stature is 1"083, or this slight prepotency is still preserved if we judge the m atter by regression coefficients. Again, we notice an immense increase (0-2841 to 04247) in the correlation between m others and daughters when we compare the present results with those of my earlier memoir. As an explanation of this, I have already suggested the possibility of a law exhibiting a relation between fertility and hereditary influence in mothers (§ 4 (ii) ). (iii) The mean coefficient of correlation in stature between either parent and a daughter may be taken to be—

0-44 ±0-02.