AbstractThe theoretical treatments of the problem of image formation given by Scheffler, Obuchow, Ellison, van Isacker and the present author are compared in a number of respects. It is contended that the angular power spectrum and autocorrelation function of the electric vector do not vary with height in regions free from disturbances. This position is in agreement with BOOKER, RATCLIFFEand SHINN, HEWISHand the author's own work, but in disagreement with SCHEFFLER.