The semiannual variation in geomagnetic activity is generally attributed to the Russell-McPherron effect. In that picture, enhancements of southward field B_S near the equinoxes account for the higher geomagnetic activity in March and September. In a contrary point of view, we argue that the bulk of the semiannual variation results from an equinoctial effect (based on the theta angle between the solar wind flow direction and Earth's dipole axis) that makes B_S coupling less effective (by ~25% on average) at the solstices. Thus the semiannual variation is not simply due to "mountain building" (creation of B_S) at the equinoxes but primarily from "valley digging" (loss of coupling efficiency) at the solstices. We estimate that this later effect, which clearly reveals itself in the diurnal variation of the <i>am</i> index, is responsible for ~65% of the semiannual modulation. The characterstic imprint of the equinoctial hypothesis is also apparent in hourly/monthly averages of the time-differented <i>DST</i> index and the <i>AE</i> index.