AS PROGRESS in the design and in the calculation of performance of synchronous machines has been made it has become necessary to specify their characteristics with continually greater accuracy. The predetermination of short-circuit currents, voltage dip on sudden application of load, stability characteristics during system disturbances, excitation requirements, and many other characteristics has been made possible by the development of an adequate theory, confirmed and refined by operating and test experience. The machine characteristics commonly used to predict performance usually are specified in terms of various reactances and time constants. As the number of these quantities and the accuracy to which they must be known has increased, certain inconsistencies among methods of definition, calculation, and test have become apparent. These should ideally all result in the same value. If they differ, one naturally would be inclined to consider the test value as the true reactance. In any case in which a reactance of a particular machine is specified, a method of testing for that reactance also must be specified, and any other definition of the reactance has little practical significance. In this sense the criterion of a good method of calculating the reactance is that it predict as nearly as possible the value which will be found by the specified test method. It, of course, can be asked legitimately whether the specified reactance is the proper one in view of the use to which it is to be put, and the definition is intended to describe the reactance so as to answer this question.