The object of this paper is to show how a well-known method of demonstrating the independence of one member of a system of axioms can be incorporated into a systematic method of testing for independence, suitable for solution on a logical computer of the type now being built at the University of Nottingham.