In the standard form of the phase-contrast test, the light-source is a pinhole comparable in size with the Airy disc of the system under test. As a consequence, it is difficult to obtain enough light through the pinhole to carry out the test with full efficiency. If a slit source and a phase-changing strip are used this difficulty is overcome, but the properties of the test are altered; for example, the brightness distribution seen on the surface of a true mirror under the test is no longer radially symmetrical.In the present paper the properties of the phase-strip test are examined in the mathematically simplest case where the light-source is a slit of negligible width, passing monochromatic light, and it is concluded that the test can safely be used in figuring an optical surface to an accuracy of about one twentieth of a fringe. Below this limit, the systematic error inherent in the usual interpretation of the test begins to be appreciable in the case of slow errors, though rapid local errors are correctly seen right up to the limit of sensitivity of the test