Ixii 



Tables for Statisticians and Biometricians [VIII IX 



The above value of d/N, for a value of h between '40 and '50, and of k between 

 10 and '20, lies between the values for r = '40 and r "45 : 



1^0,0= '099,2408, * lf0 = -085,5431, * 0fl = '087,0997, S M = -074,8716 ; 



r=--40 J S 2 * 0) o = H853, S 2 ^ 1)0 = 12327, S 2 z 0>1 = 11026, S 2 * M = 11382, 



U'%o = 7394, S /2 i >0 = 6895, S'%i = 8263, S /2 ^ 1>1 = 7616. 



f * ,o = '091,4776, s 1>0 = -078,2330, ^ 0>1 = '079,6341, 2 M = -067,8782; 



r = --45 | S 2 * 0) o= 12573, S 2 * 1>0 = 12899, S 2 s 0f i = 11675, 8 a * M = 11878, 



U'%o = 8255, S' 2 z li0 = 7680, B' 2 z Q>l = 9021, S /2 *n = 8289. 



Accordingly we have the following values for : 



z g> x = dlN= When r= - -40 



= -0869,1705 



{(1 + <)(^S 2 0)0 + % S 2 .2 0>1 )} 4524 



{(1 + e ) (^S 2 ^ 1>0 + %S 2 ^1,1)} 6605 



/2 ^ 10 ) 3167 



2281 



When r= - -45 



= '0795,5185 



4797 

 6909 

 3529 



2485 



Hence by linear interpolation 



0867,5128 



-0793,7465 



141 21 



73766 



or the correlation of Mother's increasing Outwork with Father's decreasing Wage is 

 4096. 



Had we used only the hyperbolic formula, i.e. the first line of the above 

 expression for ZB^, we should have found 



= '4107, a value fairly close to the above. 



Thus, from the examples worked in this Introduction, it would appear as if the 

 hyperbolic formula were adequate for either finding a cell content, or determining 

 the value of the coefficient of correlation. 



Illustration (v). An appreciation of sex was made by two different observers 

 on 216 femora. It is required to find a measure of the association in judgment 

 between the two observers. 



In appreciating sex by the examination of a bone the observer's opinion varies 

 from practical certainty of maleness through every shade of doubt to practical 

 certainty of femaleness. The strength of the judgment is therefore a continuous 

 character, although the actual sex forms a rigid categorical differentiation. It is 

 with the judgment, and not with the actual sex, that we are here concerned, and 

 we have simplified those judgments down to unique categories </" and $ , classifying 

 under each such category all queried values like </? and $ ?. 



