372 On the Lcurs of Inheritance in Man 



If we compare the two sexes, we see that except in the matter of stature the 

 iiiarrierl woman is relatively as variable as the inarried man, whilc in all tliree 

 chaiacters the young woman is relatively morc variable than the yonng man. The 

 supposed preponderance of male variability is thns agaiu very fuliy iiegatived, 

 for large statistics of typical physical characters in mankind*. 



Tnrning to the correlations we see («) that in the older generation the niother 

 is Icss highly corrolated than the father, (6) that in the yonnger generation the 

 soll is less highly correlated than the daughter, (c) that the younger generation 

 of both sexes is niore iiighly correlated than tiie older generation. Now the effeet 

 of selection is to rediice cürrelation, hence if selcction — a selective death-rate — 

 be a real fai-tor in tlic oasc oC uiaii am! \vr kimw it to be so, we should certainly 

 ex|>eot the correlations betweeii the ages of yoiith and of inidille life to be redneed. 

 They arc thus reduced, biit fhr more markedly so in the case of woman than in that 

 of man. Now as far as our data at ])reseMt reaeh wo kiiow that the male baby 

 is more variable and more highly correlated than the femalef. In youtli the 

 woman is more variable and more highly correlated than the man ; in adult age 

 after child-bearing she is less highly correlated and perhaps very slightly less 

 variable. It would thus seem that between birth and manhood the male is 

 selected and falls in both variability and correlation below his sister, With 

 womanhood comes her period of selection, sexual selection for wifehood, natural 

 and ir]iro(luctive selection fnr niotherhood. These act witii a little expected 

 iutfnsity and leave motliers of adult familics with fai- less variability and corre- 

 lation than tlieir husbands have. 



Of course these changes in varialiility and coirelation niay be partly growth 

 changi'S, but since on the average the man reaches his maximum size four er 

 five years later than the woman and at least four or five years beyond the average 

 age of our group sons, it is ditfieull tt) account for the wide difference in Variation 

 and correlation between daughters and niothers as compared with that between 

 sons and fathers by growth changes only. 



I am inclined to think Table I. is very illustrative of the natiire of selection 

 amoug mankind, and further that it is also hopcful, not as regards the quantity, 

 of which it takes no account, but as regards the ipiality of the ofi'spring of a fair 

 sample of the English inidillc cla.sses. 



(v) Direct Assortative Mating in Man. 



We have seen above that all women, if they indeed become wives, de not 

 beconie the mothers of adult children, i.e. the mothers of the second generation are 

 not a random sample of their own generation. However it may arise there is 



• See The Chances of Vcutli, Vol. i. pp. 250 — 377. A recent criticism by Mr Havelock Ellia of my 

 view that there is no prepondeiating variability of man over woman seems to nced no reply, for the 

 author does not appear to understand wliat weicht is to be given to scientific evidencc as compared with 

 vague gencralities. 



t li. S. Proc. Vol. G6, p. 25. 



