Statistical methods in the development of apparatus life quality

Life tests on a new design of relay were so planned that the methods of statistical quality control could be used to separate assignable causes for variation from random causes for variation during life. The relays were inspected frequently at the beginning of the test and less frequently later. The results of these inspections were plotted in the form of control charts. From a study of points outside control limits, improvements were made in the design, the method of manufacture, and the method of test. From a study of the control limits and the number of points within them, predictions were made as to the useful life to be expected. Simplified methods used in recording and analyzing the data and in establishing trends are described.