THE L3 SYSTEM — QUALITY CONTROL REQUIREMENTS 955 



(c) The seven consecutive averages are not all outside the same 

 PA limit (not all above the upper PA limit or all below the 

 lower PA limit). 

 When eligibility to use the control chart method is established and so 

 long as it is maintained, sampling inspection may be used. For this in- 

 spection, periodic samples of 5 are selected in accordance with a schedule 

 which normally calls for measurement of about 10 per cent of the units 

 produced. The lots represented by such samples are considered as con- 

 forming to the intent of the distribution requirements, and hence ac- 

 ceptable for this feature. Provisions are made for further reducing the 

 inspection rate when consistently good control performance is evidenced. 

 During the period of sampling, the following criterion applies to the 

 sampling results: 



Criterion II — Maintaining Eligibility 



Ehgibility for use of the control chart method is maintained so 

 long as the results of the current sample satisfy all of the following 

 conditions : 



(a) The average either (1) meets the A5 limits; or (2) fails to 

 meet the A5 limits but at the same time all of the 6 preceding con- 

 secutive averages meet the A5 limits; and 



(b) The range either (1) meets the R5 limit; or (2) fails to meet 

 the R5 limit but at the same time all of the 6 preceding consecutive 

 ranges meet the R5 limit; and 



(c) Seven consecutive averages (for the current sample and the 

 6 preceding samples) do not all fall outside the same PA limit. 



(d) No major change is made in raw material, machine set-up or 

 personnel, which may have a significant effect on the quality of 

 the product. 



Fig. 4 gives a control chart for averages and ranges of samples of 5 

 such as might be obtained under the control chart method. The first 

 20 points (Series A) indicate what might be expected in a series of sam- 

 ples if the process were controlled with its average, X\ at N and its 

 standard deviation, a, equal to the standand value, 0.3^. The next 20 

 points (Series B) indicate the expected pattern of points if the process 

 average suddenly shifted to a level about 0.25^ above the nominal, 

 while a' remained unchanged. Both sets of points represent the result of 

 random sampling experiments simulating the conditions stated. During 

 Series A the first seven points meet Criterion I and would have estab- 

 lished eligibility for using the control chart method. Eligibility would 

 have been maintained for the balance of Series A. Starting with Series 



