EFFECT OF SEX ON TYPE 521 



already disposed of, the remaining n - x tall women must have short 

 husbands, just as many as there are tall men with short wives. Hence 

 the generality of the proposition is only limited by two assumptions 

 tacitly made, viz. : 



1 i ) That there are equal numbers of tall and short. We can always 

 secure this by arranging the individuals in series and counting to the 

 half-way. But the separation will in general be rather arbitrary since 

 many men (and women) will be closely of medium height, and there will 

 be other striking features of distribution which are neglected in this 

 rough assortment. 



(2) We have assumed quite independent standards for men and 

 women. This is an assumption already familiar in statistical work. 

 Sir Francis Galton at the outset of his essay "Natural Inheritance" 

 adopts the artifice of "never dealing with female measures as they 

 are observed, but always employing the male equivalents in the place of 

 them " : and without this artifice he declares that he " hardly knows how 

 we should have succeeded in making a fair start." 



13. With these qualifications the condition (V.) is seen to be in 

 accordance with general experience : and hence it follows that on the 

 average the child's path will fall just midway between those of its 

 parents (VI.). 



14. But we must remember that this result is only the extension to 

 the case of two parents of the law assumed for one parent i.e. that a 

 child resembles its parent. It is therefore liable to modifications due to 

 any departure from this law such as we shall presently consider. 



15. Sexual restriction of deviation from type. Now suppose OP is 

 the general type of a species, and some individual follows the path 

 OQ differing from OP. The path of the 



mate may be on either side of OP, but 

 will on the average be along OP: and 

 hence the children will on the average 

 follow the midway path OR, which 

 diverges from OP only half as much as 

 OQ. Hence we see how the deviations 

 of the parents are reduced in their off- 

 spring, and the tendency to diverge 

 indefinitely, noticed in 9, is checked. 

 We shall presently show that definite 

 limits are assigned to the divergence, but 



first there is a special point calling for notice, which might modify the 

 conclusion just reached. 



1 6. Assortative mating. An assumption (italicized) was made in the 

 last paragraph, namely, that the mate is chosen indifferently from the 

 community at large: and that there is no systematic tendency, for 

 instance, for the parents to resemble one another. This is in rough 

 accordance with our experience : a tall man does not as a rule choose a 

 tall wife ; indeed there are often ludicrous exceptions, which strike the 

 attention so much that we should say that the tendency was the contrary. 



