ABSTRACT
Voltage regulators fitted to private motor cars were required to operate within the range of 15.8 to 16.4 volts, and the following investigation was conducted to estimate the pattern of variability encountered in production. Normal procedure was for a regulator from the production line to be passed to one of a number of setting stations, where the regulator was adjusted on a test rig. These regulators then passed to one of four testing stations, where the regulator was tested, and if found to be unsatisfactory, it was passed down the production line to be reset. The primary objective is to isolate that part of the observed variation that is 'real' variation between regulators, as contrasted with testing or measurement error. Graphical plots confirm the assumption of normality of the between-regulator and between-setting station variation. The analysis of variance does not require normality; the specific question posed does require the approximate normality of the relevant portions of the variability.
