XXIV 



Tables for Statisticians and Biometricians 



The last two columns give, in terms of o- x , the means of the four arrays 

 measured from the upper and lower ends of h, the " normal health" range. This <r x 

 is the standard deviation in health of baby at first visit. 



We now draw the reader's attention to a point which is of some importance. 

 The mean of the whole population as judged from the marginal total in health at 

 first visit = xl<r x , but as obtained from the mean of the array means 



these are respectively l'3722ff x and l'3799cr x , or there is a small difference '0077 cr x 

 arising from the fact that the arrays are not really normal distributions*. 



If we adjust for this difference of mean we have by adding and subtracting 

 0077: 



Means of Arrays in terms of o- x , adjusted to their Mean. 



We have now to consider what values are to be given to the 7/-abscissae of 

 these #-means. The marginal totals for the health of baby at a year (y) are 

 treated as a normal curve and the z's found for the boundary of each partition. 

 We have : 



Column (i) gives the proportional frequency of the four arrays. Column (;'), 

 the value of |(1 a s ) from which y s>s +i/o-y, or the partitional abscissae in terms 



* The corresponding quantities measured from the lower end of range h, are -66820-,,. and '66060-3. 

 giving - '00760-,,. for the difference, agreeing with above as closely as four figures admit. 



