ECONOMIC QUALITY CONTROL OF PRODUCT 383 



we proceed to eliminate assignable causes, the quality of product usu- 

 ally approaches a state of stable equilibrium somewhat after the man- 

 ner of the two specific illustrations presented in Fig. 11. In both 

 instances, the record goes back for more than two years and the process 

 of elimination in each case covers a period of more than a year. 



It is evident that as the quality approaches what appears to be a 

 comparatively stable state, the need for inspection is reduced. 



2. Reduction in the Cost of Rejections 



That we may better visualize the economic significance of control, 

 we shall now view the production process as a whole. We take as a 

 specific illustration the manufacture of telephone equipment. Picture, 

 if you will, the twenty or more raw materials such as gold, platinum, 

 silver, copper, tin, lead, wool, rubber, silk, and so forth, literally col- 

 lected from the four corners of the earth and poured into the manu- 

 facturing process. The telephone instrument as it emerges at the end 

 of the production process is not so simple as it looks. In it there are 

 201 parts, and in the line and equipment making possible the connec- 

 tion of one telephone to another, there are approximately 110,000 more 

 parts. The annual production of most of these parts runs into the 

 millions so that the total annual production of parts runs into the 

 billions. 



How shall the production process for such a complicated mechanism 

 be engineered so as to secure the economies of quantity production and 

 at the same time a finished product with quality characteristics lying 

 within specified tolerances? One such scheme is illustrated in Fig. 12. 

 Here the manufacturing process is indicated schematically as a funnel, 

 at the small end of which we have the 100 per cent inspection screen to 

 protect the consumer by assuring that the quality of the finished 

 product is satisfactory. Obviously, however, it is often more econom- 

 ical to throw out defective material at some of the initial stages in 

 production rather than to let it pass on to the final stage where it would 

 likely cause the rejection of a finished unit of product. For example, 

 we see to the right of the funnel, piles of defectives, which must be 

 junked or reclaimed at considerable cost. 



It may be shown theoretically that, by eliminating assignable causes 

 of variability, we arrive at a limit to which it is feasible to go in reducing 

 the fraction defective. It must sufiice here to call attention to the kind 

 of evidence indicating that this limiting situation is actually approached 

 in practice as we remove the assignable causes of variability. 



Let us refer to the information given in Fig. 6 which is particularly 

 significant because it represents the results of a large scale experiment 



