Many industrial plants judge the quality of their product by sending samples to laboratories for mechanical, electrical, or chemical tests. Such tests usually involve various instruments, different operators, and frequently more than one laboratory. The question often arises as to how much of the observed variation might be due to variability between instruments, between operators and/or between laboratories, and how much is experimental error. An example is carried through to show how to plan an experiment to check possible sources of variability. Emphasis is placed on the importance of careful planning of such an experiment before collecting the data.