Results from two 48-hour beam-swinging experiments in transhorizon microwave propagation are presented. A 3.2-GHz signal was transmitted over a 164-km path and received with a 12-element vertical antenna array (beamwidth

elevation by

azimuth). The beam was rapidly scanned in elevation. Experimental angular response patterns (antenna scans) averaged over about 3 min are compared with theoretical patterns computed from turbulent scattering theory. There is considerable variability in the experimental patterns for different time periods. The experimental patterns for different time periods are separated into three groups. Signal group 1 patterns resemble the response of the array to a point source. A propagation model based on partial reflection or refraction from a stratified atmospheric layer best describes these signal characteristics. Signal group 2 patterns can be described by a model predicting a smooth decrease in scattered power with scattering angle at a rate inversely proportional to the

th power of the angle with

between 4 and 10. A propagation model based on atmospheric turbulence can describe these signal characteristics only if it includes modifying factors to account for the difference in the exponent

observed at different time periods. Signal group 3 patterns are characterized by broadened maxima, by two or more maxima, or by a maximum which is significantly displaced above the horizon. These signals can be explained only by a model which contains a nonuniform scattering mechanism. A variation in signal characteristics as a function of time of day was evident in both experiments.