426 THE GRAMMAR OF SCIENCE 



next, fertility may be correlated with the very character 

 or organ measured ; every mate is not a parent. Further, 

 we must be careful to see that reproduction itself is 

 unlikely to have influenced the quantitative value of the 

 character investigated.^ But with due precautions as to 

 the character and individuals chosen, there seems no 

 doubt that a quantitative answer can be obtained to the 

 general problem as to whether preferential mating is 

 really at work in any form of life. It appears to me just 

 one of those cases in which it is better to quietly collect 

 the statistics, than to enter into an endless argument as 

 to whether sexual selection is or is not a vera causa of 

 evolution. For example, in the case of man, let us take, 

 say, looo husbands and looo bachelors at the ages of 

 forty-five to fifty, from the same social class, and measure 

 their stature or classify their hair and eye-colour. Again, 

 let us take lOOO wives and lOOO spinsters from the same 

 period of life and class, and measure these or other 

 characters in them. If this be done, we shall soon 

 ascertain whether preferential mating, at any rate with 

 regard to these or correlated characters, is or is not at 

 work among mankind. Nor need we take bachelors 

 or spinsters alone. If our statistics are sufficiently ample 

 we may compare husbands or wives with the general 

 population of the same age and class, and notice whether 

 the differences mark them off as a distinct group.^ Un- 

 fortunately I have not hitherto had the opportunity for 

 collecting statistics bearing exactly on this point, but it 

 may not be without interest to examine as an illustration 

 some statistics collected for other purposes. 



In the first place let us consider stature. I find the 

 following results in inches for the middle class : — 



Type. Variability. 



Husbands .... 69. 136 + .126 2.628 + .089 



Males in general . . . 69.215 + .066 2.592 + .047 



Wives .... 63.869±.i-io 2.303 + .078 



Females in general . . 64.043 + .061 2.325 +.043 



^ It is asserted, for example, that childbearing alters women's stature. 

 ^ We must investigate whether the differences in type and variability are 

 several times as large as the " probable errors " of these differences. 



