The distributions of light incident upon the windows of classrooms and offices were measured by a spatial illuminance meter designed by the authors. An experiment was also carried out to evaluate the modelling of human face at the window which simulated the typical pattern of vertical illuminance distribution. The relation between the vertical illuminance ratio and the subjective scale value is systematically separated by the window size and luminance, and is rather different with that proposed by Tabuchi. The relation between the semi-cylindrical illuminance ratio and the subjective scale value is separated mainly by the window luminance, and the ratio may be an index suitable for the modelling at a window.