312 MR. G. UDNY YULE ON THE ASSOCIATION 



66. We have no material, as already stated, for testing the effect of pure 

 selection with any absolute certainty. The greater number of the deaf and dumb, 

 however, possess their defect from birth or early childhood, and the same statement 

 holds good for imbeciles or idiots as distinct from the insane. Hence the asso- 

 ciation between deaf-mutism and imbecility will be affected by selection (selective 

 mortality) to a large extent, at all events, as compared with those associations that 

 were given in Table VIII. The necessary data for discussing the association in this 

 case are given in the English Census for 1881 (loc. cit., note on p. 298), and the 

 results are tabulated in Table IX. The figures for males show a steady and con- 

 tinuous decrease in association, without a break ; for females there is, on the whole, 

 a decrease, but it is less regular.* Taking both sexes together, I think the decrease 

 in association is rather greater for dumbness and imbecility than for dumbness and 

 mental derangement. Thus Table IX. seems to point to selection causing decrease of 

 association. This is the only evidence we have of at all a direct character, and so its 

 results should be accepted pending the production of anything better. At the same 

 time the material is obviously not unimpeachable, and an endeavour should be "made 

 to get reliable statistics for the special purpose, t 



Differences between the Sexes. 



67. The differences exhibited by the sexes as regards association are so marked 

 that they can hardly have failed. to have struck the reader of the foregoing tables. 

 In an immense majority of cases the association is greater for females than for males 

 dealing only with the total associations that is to say4 This is true for all divisions 

 of one material, and for the Census defects as well as for those dealt with by Dr. 

 WARNER. The evidence is collected in Table X., which is based entirely on the 

 preceding tables. In 87 cases out of 101, or 86 per cent., the associations are 

 greater for females than for males. There seems some indication of a decrease in 

 the difference with advancing age ; thus in Standards IV. -Ex. VII. the females are 

 only greatest in 3 cases of 6. In the age groups over 25, pooling Tables VII. and IX. 

 together, the females only exceed in 71 per cent, of the cases, or 15 out of 21, instead 

 of 86 per cent., or 18 out of 21. 



68. Besides being more highly associated, women are also in general less defective 

 than men. They exhibit a smaller percentage of individuals with development 

 defects, nerve signs, or mental dulness, but a slightly higher percentage with low 



* It will be noted that the age groups are not the same as in the last case, the figures being grouped 

 more coarsely in the 1881 Census. 



t Professor PEARSON informs me that unpublished material in his hands goes to show that correlation 

 decreases with age ; theoretically also he would expect selection to decrease correlation. 



J This is again in accord with the evidence for correlation females being more highly correlated than 

 males. 



