IV] Introduction \liii 



(iii) Mental Capacity and Place in Class. The data concern 249 boys arranged 

 in four categories of intelligence Excellent, Good, Moderate, Dull by tea< 

 estimate, and by place attained in class*. The observed n*p.i = '525,045, and aa 

 N = 249, n = 4, we have from the table 



^ = -012,097, oy = 009,778. 

 Hence \ d = (-525,045 - -012,097)/-009,778 52'5, 



but since X 1 =3'20 gives a Pj= -012, we have no hesitation in asserting the signifi- 

 cance of the relationship, which will connote a correlation greater than '70. 



(iv) School Examination Marks and Intelligence Test Marks. Dr Isserlis gives 

 a table for the correlation of the above variates for 50 girlsf. 



Here there are n= 5 arrays and JV= 50. The data give 



*PTMJUI = '164,761. 



We find ^='081,633 and a^ = '054,221 



from equations (iii) and (v). Thus we have 



\ d = (-164,761 - -081,633)/'054,221 = 1-53. 



Now our table only starts at N 51, but the constancy of \z and P 2 shows us 

 that for \2 = 2-68, P a will be '019, also for N = 51 as for N = 50. Thus, since \ d is 

 far below \z, we can be quite sure that the numbers are inadequate for the purpose 

 of ascertaining any correlation, should it exist. 



(v) Distance of Nearpoint and Colour of Iris. The dataj were obtained from 

 770 boys in seven categories of iris colour determined by Martin's scale. 



For n = 7 and N= 770, our table provides 



7j a = -007,802 and o>, = '004,481, 

 and the data lead to r) 2 NP . cl = -021,727. 



Thus \ d = (-021,727 - -007,802)/'004,481 = 3'11. 



Here Pi='0lf) for \!=3-08, P 2 = '021 for \a = 2'58, and accordingly \ d will give 

 a value of PI slightly less than '010, say '009, or the odds will be about 110 to 1 

 against such a value of rj 2 arising, if there be no association between iris colour 

 and distance of nearpoint. It is reasonable therefore to assume that there exists 

 an association of the order of '15 between the colour of the iris and the distance 

 of the nearpoint. 



(vi) Rainfall in relation to Longitude, Latitude and Altitude. The data refer 

 to 57 recording stations in Hertfordshire, and FisherJ found the multiple cor- 

 relation of rainfall on longitude, latitude and altitude to be given by J2*=' 



* Diometrika, Vol. vrn. p. 544. 



t "The Relation between Home Conditions and the Intelligence of School Children," Medical 

 Research Council, Special Report Series 74. 



J Statistical Methods for Research Workers, pp. 135 and 228. 



