XXI XXII| 



CM 



..... ,.; VAHIA- 



Lnninnrpj| P 





Diagram 1. 



samples. The deviations are tabled for all sizes of sample up to 25; at intervals 

 of 10 from n = 30 to 150; and at intervals of 100 from n = 200 to 1000. For almost 

 the whole of the table linear interpolation will be adequate if accuracy to two 

 decimal places only be required. For three decimal place accuracy the simple 

 central difference formula* 



z = 



will be always sufficient. 



Illustration. In the mass production of a certain article a firm aim at an a v. i 

 breaking strength of 176 Ibs. and a variability in strength which should not 01 

 a value measured by a standard deviation of 12 Ibs. In order to ensure that the 

 production is kept under control, tests of breaking strength are applied at intervals 

 to samples of 20 articles, and a simple check rule is required to be given to the 

 foreman in charge of these tests. There is reason to believe that the distribution 

 of breaking strength of this product when the manufacture is properly controlled is 

 approximately normal. 



* It may be noted that the coefficients f^ = ,V0tf (l + ^>) and ( -).<p0 (1 + 0) have been tabled by 

 A. J. Thompson, Tracts for Computers, No. v. Cambridge University Press. 



