MATE SELECTION. 



45 



In applying the test to the hypothesis we may assume in turn that 

 the groom has done the selecting and that the bride has doi 

 selecting. We then compare, in the selections made by the groon 

 the expected proportion of the classes of build on the assumption of 

 no assortative mating, with the proportions actually found in the 

 brides. Similarly, with suitable changes for the selections made by 

 the brides. The results are given in table 20. 



Table 20. — Percentage distribution of build of consorts selected In/ grooms and by hr </. | belonging to each 

 of the classes of build, and comparison with the standards of table 19, 



P, percentages found or expected. E, percentage excess of found ov. tad. 



Selections made bt Grooms. 



An inspection of tables 19 and 20 shows that the hypothesis that 

 wives and husbands of men of each different class of build are merely 

 random samples of the whole population of parents is not supported 

 by the facts. Thus on the part of both very fleshy grooms and bri 

 over 70 per cent more consorts, who will ultimately be very fleshy 

 are selected than are expected on the hypothesis of random sam- 

 pling. Also, among fleshy fathers there is a marked : very 

 fleshy wives. Slender parents have an excess of similar const 

 Medium parents have selected consorts nearly at random B0 far 

 regards build. Slender parents have selected a smaller proportion 

 of very fleshy consorts than expectation on random choice, and \- 

 fleshy parents have selected less than the average of very slender and 

 slender consorts. In a word, there is some d of assorta' 



