xxii Tables for Statisticians and Biometricians [I 



arrays, i.e. cr y VI r 2 . We are not compelled to give any value to x, although it 

 would be reasonable to take it equal to the mean x of the array, i.e. 



* n x /N ' 



We simply treat the array of T/'S as a normal curve, find with a tripartite 

 division the mean of the array of ys in terms of its own standard deviation, 

 and afterwards in terms of the marginal standard deviation. We can later com- 

 pare the mean thus found with the true mean as deduced by the formula 



(T y n x jN 



which would be the y-point at 



on a theoretical regression straight line. 



Illustration (i). Find the boundaries and means of the following system of 

 broad health categories for yearling babies, on the assumption that health follows 

 a normal frequency distribution*. 



Total ... 1436 



The permilles are obtained by a continuous process with the reciprocal of 1436 

 on the machine. The differences are obtained from the permilles by subtraction, 

 and only differ in one case, that of the Normal Health, which if found directly 

 from the frequency would be '354, and this is of no importance for our present 

 purposes. 



Table I, Part I, and the present table provide at once : 



-oo 



The whole work is very simple as there is no interpolation necessary. 

 * The continuous variate may be looked upon as physiological fitness for environment. 



