Evidence is presented to show that high-level tropical cyclones originate over the Marshall Islands as tropospheric responses to sudden increases in the radiative output of the sun. Corroboration of the hypothesis that the effect occurs at other longitudes in the tropics is derived from the work of Riehl, Yeh and La Seur, and the implication is made that changes in the general circulation of middle latitudes follow the poleward migration of the cyclones. The first response of the troposphere, following solar explosions with a delay probably less than thirty-six hours, is a sudden rise of temperature in the layer between 300 mb and the tropopause over the equatorial convergence zone. The rate of rise is 7C per day at 100 mb. The changes cannot be explained by advection. Cyclones develop on the poleward side of the upper warm ridge produced by the local temperature rises. It is suggested that, during solar flares, ultra-violet radiation of greatly enhanced intensity penetrates the ozonosphere near the equator; the absorbent is tentatively identified as water vapor.