Page:Proceedings of the Royal Society of London Vol 60.djvu/303

Rh words, to judge from stature, the exceptional p aren t tends to have offspring of th e opposite sex.

(ii) Younger sons are taller and m ore variable th an elder sons, and elder sons are taller and more variable than fathers.

This conclusion, although less m arkedly, appears in the results on pp. 270 and 281, of my former paper. I t m ight be accounted for b y :

(а) A secular change going on in the statu re of the population, and even noticeable in the difference betw een th e stature of younger and elder sons.

(b) A fu rth er grow th of sons, and an ultim ate shrinkage, which will leave them at the age of th eir fathers w ith the same mean height and variation.

(c) Conditions of n u rtu re on th e average less favourable, and on the whole less varied in the case of elder th an in th a t of younger children.

(d) Natural selection. The difference between younger and elder sons and between elder sons and fathers represents th e selective death rate in m an due to causes correlated with stature in the years between youth and manhood, and m anhood and age. The difference is thus to be accounted for by a periodic and not a secular change.

Possibly (a), (6), (c), and (d), may all contribute to the observed results. It cannot be denied th at (d) has a special fascination of its own for the student of evolution, but prolonged study of the laws of grow th m ust precede the assertion th at we have here, or in any similar case, real evidence of an actual case of natural selection.

(iii) Younger daughters are taller than elder daughters and elder daughters than mothers.

This is in complete agreem ent w ith the result for fathers and sons. F u rth e r:

Daughters, as a class are fa r more variable than mothers, but wlnle in the earlier memoir younger daughters were sensibly more variable than elder daughters—and thus exactly corresponded with sons elder daughters are in this case more variable than younger. I lave been unable to find any slip in the tables or calculations, which mig t account for this divergence. I t exceeds considerably the probable error of the observations, and is not in accordance with t e geneial law connecting the variation of parent and offspring evienced for both sexes in the earlier, and for sons in the present memoir— e.g., the variation whether it be due to growth-change,