41)0 Assortalice Matlur/ in Man 



uf a yoar's seuiority in the husband being accoinpauied by less thau half a year's 

 loDger survival on the part of the wife. 



To throw sonie light on this the purtial correlations /3//i and p,,-^ ai-e given. 

 These nieasure the correlation between seniority of husband and the lem'th of life 

 of husband or wife for a given length of life of the other. 



Thns jOicA is very substantiai, or \vc sce that among the array of wives who 

 niany husbands of a given duration <pf life, th;it wife lives shortest whosc husband 

 has the gie.itest seniorilv. In other words grcat disparity in age between husband 

 and wife tends to shorten the life of the younger of the pair. This must not be 

 confused with the notion of 'shock ' referred to on p. 486. For by the formula for 

 partial correlation : 



P DA = 



Vl-r=,p//Vl- 



j-,p//Vl-r'HA 



and this would have a finite value even if rir//=(), in fact the actual value of r^ji 

 only influences pu-^ in the seeond order of approxiniation. The essential part 

 of the value of pH-^ is due to the value of ch-a, and the sanie remark ajjplies 

 to pii\ and /-y/^. 



It would thus appear, tiiat while nien and wonien tend to select luatcs of 

 physical constitutions similar to their own and leading to correlated durations 

 of life, thero is within this geueral tendency a rather more subtle factor at work, 

 namely that wlien there is disparity of age, the older niember of the pair either 

 mates with an individual relatively less fitter to the environment thau the older 

 individual, — for example, if older men marry the less robust younger women 

 — or eise the disparity in age has a physically bad ctfect on the duration of life of 

 the younger member of the ])air. Whichever way we Interpret the result this 

 factor is one of very consideraljle iuterest and deserves investigation beyond the 

 limits of a paper devoted to assortative mating. The above table sutfices to show 

 the sensible existence of this factor in very difFerent sub-groups of the present 

 Euglish population. It also indicates how close the Statistical coustants we have 

 dealt witii remain when we pass from Yorkshire dales to Oxfordshire flats. We 

 believe that similar results obtained from the graveyards in different parts of Grcat 

 Britain would be of great value as throwing light on the influeuce of locality and 

 local race on vital statistics. 



(10) The niain result of this paper is to demonstrate a very sensible degree 

 of assortative mating in man, and we have already suggested the importauce of 

 this for the problem of evolution. But the field opened up leads us directly to a 

 number of actuarial problems, which seem deserving of special consideration. They 

 form another link in the chain which must ultiinately draw together the actuary 

 and the biologist. The actuary has been studying evolution without knowing it 

 for generations ; and the biologist to grasp evolution has got to do actuarial work 

 not only for man, but for all typcs of life. 



