ECONOMIC QUALITY CONTROL OF PRODUCT 389 



lines in Fig. 14-J. The lengths of these are proportional to the allow- 

 able tolerance limits corresponding to a given probability. Customary 

 practice is to use the line of regression between tensile strength and 

 hardness. Note the improvement effected by using the plane of regres- 

 sion. By using the hardness and density together as a measure of 

 tensile strength in this case, the tolerance limits on tensile strength 

 corresponding to a given probability can be reduced to approximately 

 one-half what they would be if either of these measures were used alone. 



IV. Conclusion 



It seems reasonable to believe that there is an objective state of control, 

 making possible the prediction of quality within limits even though the 

 causes of variability are unknown. Evidence has been given to indi- 

 cate that through the use of statistical machinery in the hands of an 

 engineer artful in making the right kind of hypothesis, it appears 

 possible to establish criteria which indicate when the state of control 

 has been reached. It has been shown that by securing this state of 

 control, we can secure the following advantages: 



1. Reduction in the cost of inspection. 



2. Reduction in the cost of rejections. 



3. Attainment of maximum benefits from quantity production. 



4. Attainment of uniform quality even though inspection test is 



destructive. 



5. Reduction in tolerance limits where quality measurement is 



indirect. 



26 



