The frequency of geomagnetic reversals and the symmetry of the nondipole field

The average frequency of geomagnetic reversals seen through a sliding window 10 m.y. long typically remains constant or nearly so for intervals of the order of 5 × 10 7 years and then increases or decreases to another level. Frequency shifts of this type are known from paleomagnetic studies to have occurred at times 107, 86, and 45 m.y. B.P. and at many other times earlier in the earth's history. The most likely cause of the frequency shifts is a change in the topography of the core‐mantle interface or changes in some physical property such as temperature along the interface. If such lateral variations at the base of the mantle exist, if they have changed with time, if they are large enough to create eddies or Taylor columns in the fluid motions of the core, and if these fluid motions affect the operating characteristics of the dynamo, then a simple model is at hand to account for the observed changes in reversal frequency. The generation of fluid eddies in the outer core by irregularities at the core‐mantle interface would also probably induce variations in the nondipole field. These variations should be observable if they include departures of the nondipole field from the zonal symmetry expected solely from rotational effects. Among the candidates for such asymmetrical components of the field are the standing waves that are currently being reported by geomagneticians analyzing the historic field and the apparent northward shift of the main dipole currents of the core dynamo that is suggested by time‐averaged global paleomagnetic data. On the basis of a reexamination of these data the conclusion is here drawn that the apparent northward displacement of the earth's main dipole moment is not the result of a standing zonal field but is rather the result of time averaging individual small excursions of the field caused by the westward drift of major features of the nondipole field. Paleomagnetic results indicate that these features of the nondipole field are not randomly distributed with respect to latitude, as is usually accepted. Instead, it appears that during times of normal polarity, flux usually emerges from the core in localized regions near the equator and reenters the core in localized regions located about 55° from either pole. During times of reversed polarity this entire pattern is reversed, suggesting that when the dipole field reverses, the toroidal field reverses also. The latitudes at which the large features of the nondipole filed are generated by convective cells or other types of fluid motion thus appear to be linked directly to the polarity of the dynamo. These findings are supportive of the reversal mechanism suggested by Parker and Levy.