A simplified analytical model of tropospheric random variations in angle measurements is described. This model is used to predict the minimum and maximum power density spectra between which the tropospheric random angle errors of observations on the Early Bird satellite are expected to lie. The apparent angular position of the Early Bird satellite was then measured at microwave frequencies with the large horn-reflector antenna at the AT&T station near Andover, Maine. Random variations in the azimuth and elevation angles have been observed and recorded. The analysis of these records results in a description of the observed random angle variations by their power density spectra. A comparison of the predicted power density spectra from the model with the observed spectra is made. It is concluded that the observed random angle variations are indeed caused by random tropospheric refraction. The feasibility of acquiring data on atmospheric propagation effects, particularly tropospheric angle errors, with the aid of geo-stationary satellites is therefore also demonstrated.