310 MR G. UDNY YULE ON THE ASSOCIATION 



boys increasing by half; the percentage of dull also increases largely. One can only 

 conclude that these defects, whether inherited or not, are not " innate " in the simple 

 sense of being observable at birth or during early infancy ; possibly they may be 

 brought on in part by the school attendance itself, a theory which would account 

 for the sudden rise after " Infancy," or rather early childhood. As for the subsequent 

 decreases in the percentages of all defects, I am inclined to attribute them, in part 

 at all events, to the selection by capacity that must take place. This would reduce 

 the percentage of " mentally dull," and, owing to the association between defects, 

 would also reduce indirectly the percentages of those with other defects. A proportion 

 of the decrease one would expect to take place from natural selection, but probably 

 a small proportion. We have, in fact, selective mortality, selection by mental 

 capacity, and growth (change in the individual), all acting together as causes of 

 change and very probably also initial differences between the groups from which 

 the older and younger children sprang precisely as was indicated in the previous 

 discussion on pp. 299-300. 



63. In the association coefficients there is,however, no irregularity in the group 

 Standards I. -III. ; the most cursory inspection of Tables IV. and V. shows that the 

 total coefficients and partial coefficients with negative universes decrease when we 

 pass from " Infants " to the next group, and again from Standards I.-III. to the next 

 group. The changes in partial associations with positive or defective universes are 

 not so obvious, partly, no doubt, because these coefficients are small and have large 

 probable errors, but the majority of the changes are in the same direction. Table VII. 

 shows the total number of changes in either direction. Thus comparing Standards I.- 

 III. with " Infants," we see that in the case of the boys all the total associations and all 

 the partials with negative universes have decreased ; of the partial associations in 

 defective universes 8 only decreased while 10 have increased. Taking all cases 

 together there are 88 decreases to 32 increases, but if we cut out the partials 

 22 decreases to 2 increases only. However we take it there is overwhelming 

 evidence that association in general decreases as age advances. 



64. That this is a law of pretty general application seems to be borne out by the 

 entirely different class of statistics drawn from the English Census.* In Table VIII. 

 are given the associations derived from these figures between blindness and dumbness, 

 blindness and mental derangement, and dumbness and mental derangement. The 

 figures do not run quite regularly, but the associations for the age group 5 15 are, 

 in every case, greater than the " all-ages " associations, and if we draw curves showing 

 the change of association during life the downward trend is quite obvious (curves of 

 figs. 1, 2, 3, p. 314). 



Such decrease is again not, of course, the result of selection alone but of selective 

 mortality, growth (change in the individual), and initial differences in the child- 

 populations from which the successive age groups were formed. That changes in the 



* Loc. <., on p. 42. 



