The semiannual variation in geomagnetic activity is generally attributed to
the Russell-McPherron effect. In that picture, enhancements of southward f
ield B-s near the equinoxes account for the observed higher geomagnetic act
ivity in March and September. In a contrary point of view, we argue that th
e bulk of the semiannual variation results from an equinoctial effect (base
d on the psi angle between the solar wind flow direction and Earth's dipole
axis) that makes B-s coupling less effective (by similar to 25% on average
) at the solstices. Thus the semiannual variation is not simply due to "mou
ntain building" (creation of B-s) at the equinoxes but results primarily fr
om "valley digging" (loss of coupling efficiency) at the solstices. We esti
mate that this latter effect, which clearly reveals itself in the diurnal v
ariation of the am index, is responsible for similar to 65% of the semiannu
al modulation. The characteristic imprint of the equinoctial hypothesis is
also apparent in hourly/monthly averages of the time-differentiated Dst ind
ex and the AE index.