This work reviews site diversity concepts and models and presents the results of a site diversity experiment, in which radiometers were used to measure attenuation on Earth-space paths, performed at Columbus, Ohio, for a period of one year. Attenuation statistics are presented first in the form of the joint probability density function of attenuation at the two sites. From these are derived the cumulative two-dimensional distribution of attenuation at both sites, the cumulative distribution of attenuation at each site, the effective cumulative distribution of attenuation for a diversity system that always switches to the better of two signals, and the effective cumulative distribution of attenuation for a system with maximal-ratio signal combining. The switched-system diversity gain is compared with D.B. Hodge's (1982) predictive model, and the diversity improvement is compared with L. Boithias' (Radiowave Propagation, McGraw-Hill, 1987) model. Fade-duration and interfade-interval statistics are presented for each site and for systems that use switching and maximal-ratio combining. Rain-rate and interrainfall interval statistics are included.>