INFLUENCE OF ORDER OF BIRTH, ETC. 317 



ranks, and finds that the sons of old fathers live longer than the 

 sons of young fathers. He also studied the longevity of 1,104 

 persons from families of four or more children who lived to adult 

 life. From these persons "among whom those having high birth 

 ranks were brothers and sisters of those having low birth ranks, 

 it was found that there was a very uniform increase in length of 

 life as birth ranks grew higher," an addition of four years to the 

 age of the father added one year to the life of the child. 



In regard to the parentage of great men, Redfield remarks: 

 "It may be argued that the sons of old men are necessarily the 

 sons of long lived parents, while the sons of young men are the 

 sons of both long lived and short lived parents, and consequently 

 cannot be expected to live so long on an average." This objec- 

 tion, while sounding reasonable, Redfield attempts to show is 

 fallacious. In order to do so he selected from the Redfield gene- 

 alogy "every family which had four or more sons who reached 

 maturity and who did not lose their lives because of war or 

 accident." The average life of the different sons is indicated as 

 follows : 



Years . 



"There can be no selection in this case," says Redfield, " because 

 the different sons of the family are sons of identical parents, and 

 not sons of different or selected parents." 



Despite the plausibility of his contention I cannot feel sure 

 that Redfield has succeeded in avoiding our deceptive enemy, the 

 statistical fallacy. If he has averaged together the ages of sons 

 belonging to fathers of certain age groups without regard to date 

 of marriage or other circumstances, he may have obtained quite 

 misleading results. Young parents marry early and older parents 

 as a class must contain many who married late and whose four 

 children, therefore, belong to the later part of their reproductive 

 period. It is possible to have a number of families in each of 

 which the age of successively born children regularly diminishes 

 and yet when the ages of the children are averaged together there 



