ABSTRACT
Specifications are frequently written in terms of statistical parameters which describe the product to be inspected. For attributes inspection, the parameter to be controlled is, of course, the proportion nonconforming in the lot or process. When specifications are written in term of measurements, other parameters may be of importance, such as the average (mean) level of a certain characteristic of the process which produced the units to be inspected, or in some instances its standard deviation. This implies Type B sampling. Examples of such specifications are mean life of a lamp, average amount of discharge of an impurity into a stream, average emission of carbon monoxide from cars of a certain make and model, and the standard deviation of an electrical test on semiconductors for use in a ballistic missile. Specifications of this type are in contrast to those on the individual measurements themselves which relate to individual units of product; variable sampling plans for such specifications will be covered in a later chapter. It is characteristic of specifications on a process parameter that certain levels are acceptable and
should be protected from rejection, while other levels are objectionable and should be rejected by the plan. This was recognized by Freund (1957) when he distinguished two critical levels:
u1: Acceptable process level (APL). A process level which is acceptable and should be accepted most of the time by the plan. u2: Rejectable process level (RPL). A process level which is rejectable and should be rejected
most of the time by the plan.
