Abstract. A theory is presented, which is based mainly on dimensional analysis (but also on gravity wave theory), that attempts to explain all the types of gravity wave power spectral densities (PSDs) now being measured. This theory is based on two concepts, namely, wave saturation and wave cascade. The immediate result of the simultaneous presence of these two processes is that there should exist a unique relation between the vertical (or horizontal) wavelength of a gravity wave and its period (provided the Brunt Period and dissipation rate are given and Doppler effects are omitted). This relation provides a way to derive all of the intrinsic spectra from the fundamental one which is the vertical wavenumber PSD of the horizontal winds. The most important suggestion to emerge from this theory is that e, the dissipation rate, is the main controlling independent variable for the amplitude of all but 3 of the 12 spectra predicted. It would also control the wavelength-period relations. Comparisons are made between observations and theory, and important experimental tests are proposed. This model presently appears to be useful in the analysis of gravity wave data obtained by means of lidars, radars, interferometers, and imagers. In addition, it raises a number of new scientific issues for future research. [1984, 1988]. These data were obtained from ground-based time-lapse photographs of rocket-laid smoke trails. Prior to the spectral analysis of the velocity profile, a cubic trend was removed. Some important features to note about Figure 1 are (1) these spectra have not been normalized (that is to say, the intrinsic "universality" of the wave behavior is the only cause of the close overlap of these PSDs), (2) the 10-m resolution of these data is the highest known to date for a simultaneously obtained velocity profile (i.e., same time for all altitudes), and (3) the velocity profiles, upon which of gravity. This creates the familiar "white caps" so often observed on the ocean. Such "saturated" waves are regarded as being in their "equilibrium range" due to the fact that the input power from the wind is equal to the output power of the wave breaking.