The discrepancy between the reproducibility of the ellipsometric parameters Δ and ψ (0.01° in the best cases) at any one particular setting of the polarizer, compensator and analyzer, and the difference in Δ and ψ values measured at different settings of these components (even as high as 2-3°) causes some doubts about the validity of the optical parameters determined from ellipsometric measurements. Some reasons for this discrepancy have been mentioned previously in the literature and the present paper attempts to complete these ideas. It is possible to prove that when the angle of incidence is carefully adjusted at fixed polarizer and analyzer azimuths and the measurements are done at two compensator azimuths ( 1 4π and 3 4π) for four different sets of the polarizer and analyzer azimuths, then the average values of Δ and ψ are identical with the true values of these ellipsometric parameters within the experimental error. © 1969.