From spectra of the equatorial photosphere taken near. the limb of the Sun between 1950 October and 1952 June, 480 determinations of line-of-sight velocity are made. A careful investigation shows, contrary to an earlier suggestion, that the effect on velocities, of light scattered in the Earth's atmosphere and in the telescope, is negligible. In order that each day on which observations are made should carry equal weight in the final solution, only 334 of the determinations are used to find the mean equatorial velocity. Mter the elimination of limb effect, the velocity due to solar rotation is found to be 1 ·91 km sec –1 for the mean epoch 1951·5. The individual determinations have a large dispersion about this mean and a statistical discussion shows that the chance is less than one in a thousand that the dispersion is due to errors of measurement. It is therefore highly probable that the velocity varies from point to point in the photosphere. The extreme range in measured velocity is approximately 0·3 km sec –1 and successive maxima or minima of velocity appear to be separated by a distance of the order of 75 000 km. Some evidence is found for bands of constant velocity occurring perpendicular to the solar equator, but this is less certain. The existence of such velocity changes may account for the discrepant rates of rotation found by earlier observers.