Average features of geomagnetic disturbance are represented by atmospheric‐electric current‐systems. These cannot be uniquely inferred from the magnetic observations at ground level. Using the Balfour Stewart dynamo theory, these electric current‐systems are then tentatively replaced by equivalent atmospheric windsystems, using order of magnitude calculations. It is found that these supposed wind‐systems will differ in accordance with the degree of electric polarization of the atmosphere, and with the intensity of toroidal or solenoidal electric currents likely to be present, as a consequence of zonal winds.A simple monthly mean wind‐system of geomagnetic disturbance shows good general agreement with a wind‐system derived by Kellogg and Schilling from meteorology. A qualitative version of a possible wind‐system for the main phase of a magnetic storm is derived and found to show some measure of agreement with diurnal features of atmospheric motions deduced from radio‐star scintillations and auroral motions.The causes of such wind‐systems, and the effective transverse conductivity of the atmosphere, are briefly but inconclusively discussed. It is shown that the flux of X‐rays producing ionization of theE‐region is apparently the same on days of magnetic storm as on days prior to the storm, and that the dynamo air‐flow at theE‐region yielding the quiet‐day diurnal variation also appears inappreciably affected by the p