Appreciable experimental data exist for the amplitude of basilar membrane displacement as a function of sound frequency. Data on the phase of displacement vs frequency are relatively meager. At low frequencies the latter do not relate well to the phase predicted by recent mathematical models for membrane displacement. To examine the difference, we programmed a digital computer to calculate: (1) minimum phase functions corresponding to the experimental amplitude vs frequency responses; (2) inverse Fourier transforms of the experimental amplitude and phase data; and (3) inverse transforms of the experimental amplitude and calculated minimum phase data. The results were: (1) At low frequencies the calculated phase differs from the experimental phase by about12πradians or more, with the membrane displacement leading that of the stapes. Constant delay (i.e., linear phase) does not account for the difference; (2) inverse transforms of the experimental data yield nonrealizable impulse responses, i.e., responses which are nonzero for negative times; and (3) inverse transforms of the experimental amplitude and calculated phase yield proper impulse responses. At low frequencies, therefore, the experimental amplitude and phase data do not seem completely compatible. For several reasons, the phase data seem more likely to include a small imprecision than do the amplitude data.