LEVELLING EFFECT OF MARRIAGE 191 



they not controlled by the effects of marriage. 

 From one point of view, as we have seen, sexual 

 generation may be regarded as a cause of change, 

 since it results in the production of individuals 

 that differ from their predecessors, exactly re- 

 sembling neither of their parents. But from 

 another point of view it is a conservative influence, 

 since its effect is, so to speak, to dilute any 

 peculiarities possessed by either parent, and by 

 intermixture to produce offspring that diverge 

 less markedly from an average type. This 

 tendency has been formulated as the law of 

 " filial regression." Tallness in the father will 

 be represented in his sons, but in a less degree. 

 Galton found that in the families of 286 judges, 

 who sat on the bench between 1660 and 1865, 

 eminence was shown by 12.6 per cent, of their 

 sons, 3.7 per cent, of their grandsons, and only 

 0.5 per cent, of their great-grandsons. Their 

 talents fell rapidly to the normal in their offspring. 

 So also children tend towards the normal when 

 their parents are defective. It has been shown 

 by a large body of statistics that parents who are 

 both congenitally deaf may expect three out of 

 four of their children to be free from this defect ; 

 and when only one parent is deaf the chance of 

 deafness amongst the children is only one in eight. 

 The levelling effect of marriage is no doubt due 

 very largely to the fact that the sexes are not 

 systematically mated so as to bring together males 

 and females that agree in possessing some definite 

 peculiarity. By selective breeding on these lines 

 an immense variety has been produced amongst 

 cultivated plants and domesticated animals. But 

 if selection is relaxed, reversion to type follows. 

 And under normal conditions, whether of men or 

 the lower animals, there would be very little 

 chance that pairing would occur between two 



