Changes in Human Performance with Age 161 



proportion of servicemen in the youngest age range. Subse- 

 quent tests confirmed that men with recent army experience 

 performed the task significantly faster than those who had 

 been in civilian life throughout the second world war. It 

 seems impossible to control such influences in advance 

 without building up a systematic body of knowledge on the 

 nature and extent of occupational "transfer" effects. Until 

 this is done they must be reckoned as an unknown hazard for 

 the experimenter. 



(c) Control of family relationship by comparing different 

 generations of the same family would seem to be an obviously 

 desirable procedure in studies of ageing, yet it is one which is 

 seldom used. Doubtless this is in part due to the difficulty 

 that would be caused by the additional constraint in the 

 selection of subjects, but it may in part also be due to the 

 curious unwillingness of psychologists in the past to admit 

 the role of hereditary factors in determining human capacity. 

 Resemblances between parents and children have been 

 demonstrated in a number of performances, such as intelligence 

 tests (Jones, 1928). Family resemblances in behaviour have 

 been shown to be in part the result of heredity and in part the 

 result of similarity of environment. 



For a large-scale study we should take care to see that the 

 numbers in each age range are balanced for these and any 

 other background factors suspected as being of importance. 

 For a small-scale experiment aimed at establishing a qualita- 

 tive rather than an exact quantitative result it is probably 

 sufficient to concentrate on subjects from one particular 

 background — say from a single factory or a university 

 population. The inferences we can make will be restricted 

 owing to the restricted range of subjects, but not seriously so. 

 An alternative procedure which overcomes this limitation is to 

 take two or more groups of widely different background in each 

 age range — in effect to do the experiment twice on different 

 types of subject. By this method we not only achieve a high 

 degree of control but also throw light upon whether differences 

 of background are in fact important for our experimental task. 



AGEING — III — 6 



