SAMPLING INSPECTION TABLES 3 



from accepted standards of good workmanship is termed a "defect." If, in the 

 inspection of the "end illumination" of 1000 lamps, it were found that 10 of the 

 lamps had illumination less than the minimum value specified, and the remaining 

 990 had illumination equal to or greater than the minimum value, we would 

 say that 10 defects were observed, and the lot of 1000 was 1% defective (fraction 

 defective, p = 0.01). 



Sampling Inspection — The tables are applicable where, under normal con- 

 ditions, it will be satisfactory to inspect only a portion of the pieces in the 

 lot and to accept the lot if the inspection results for this sample of pieces 

 meet certain criteria. This, in effect, imposes the condition that it is not 

 the purpose of this inspection to make sure that each piece in the lot con- 

 forms to the requirements for the characteristic inspected. 



Such a situation is common, for example, in the process inspection of com- 

 ponent parts of product units, where it may be the purpose of inspection to make 

 reasonably certain that the quality passing on to the next stage is such that no 

 extraordinary effort will be expended on defective parts. This situation is also 

 common for various characteristics of finished units of product, such as some 

 adjustment and dimensional items, items of condition, finish and workmanship 

 that can be covered by a "surface" inspection, as well as items for which 100% 

 inspections or tests have been made previously during process or are to be made 

 in subsequent operations before delivery to the ultimate consumer. Characteris- 

 tics, whose conformance to specified requirements is of vital importance to the 

 functional quality of the product, and for which 100% inspection is feasible, may 

 not of course be candidates for sampling inspection. 



Acceptance Based on Observed Number of Defects — The acceptance criterion 

 used in the tables is a stated allowable number of defects in a sample of 

 stated size. 



If only one defect is allowed in a sample of n pieces selected from a lot, then 

 the "Allowable Defect Number" is 1 (referred to as the "Acceptance Number" 

 in an earlier paper^). The criterion for the acceptance of a lot is the finding of 

 a number of defects equal to or less than the Allowable Defect Number. 



Random Samples — The theory used in the development of the tables as- 

 sumes that each sample drawn from a lot is a random sample. 



A random sample is one selected by a random operation,^ such as would obtain 

 if a number of physically similar chips, numbered to correspond to the pieces of 

 product under consideration, were thoroughly mixed in a mixing bowl, and a 

 number of them, equal to the desired sample size, were withdrawn to identify 

 which pieces of product should be included in the inspection sample. When, 

 in practice, there are indications that individual lots may be stratified in quality, 

 it is of course best to select a "representative" sample, one such that each stratum 

 or subportion of the lot is proportionately represented by a subsample that is 

 selected by a random operation. 



