Photometric error due to non-uniform responsivity in a photometer aperture is theoretically analyzed for illuminance measurement. Such error is only caused when neither responsivity nor illuminance is distributed uniformly in the aperture of an illumination meter. This error is given by the product of the fractional-standard-deviations in both distributions and the correlation coefficient between them. The fractional-standard-deviation is consequently proposed as a parameter to describe the non-uniformity of each distribution. This theory is validated by applying it to error estimation for luminance measurement of a color TV picture tube using a luminance meter composed of a diaphragm, a tube and an illumination meter. The spacial resolutions required in the measurements of both distributions, which make it possible to precisely estimate the above error, are also discussed. It is revealed that, if the distribution of either responsivity or illuminance is more gradual than the other and sinusoidal, the resolutions should be less than 1/8 the spacial period in which the sinusoidal patterns are repeated.