I agree with Henry Kolm that we need to “stand on the shoulders of giants” (Physics Today, October 2006, page 14), but he needs to put his giants in the right place, and get his facts right.
Much of my 1949–52 PhD work at the University of Manchester in the UK was a laboratory-model simulation and extension of an early idea of Edward Bullard’s that a single eddy in Earth’s core could perturb the dipole field sufficiently to give a local focus of the nondipole field. Toward the end of that work, my fellow student Arvid Herzenberg became interested and produced a formal algebraic solution; in 1957 the two aspects were reported together. 1 Herzenberg then extended his work on a single eddy to two eddies, and showed that such a two-eddy situation could give a self-exciting dynamo. 2
But Herzenberg’s paper was highly mathematical; it also considered only a steady-state situation, while the geomagnetic dynamo was reversing intermittently. By then I was teaching at Newcastle, so I selected a promising graduating student, Ian Wilkinson, and we set about building a laboratory-scale, self-exciting dynamo based on Herzenberg’s geometry. In 1963 we reported steady-state self-excitation. 3 Our dynamo was made of Perminvar, a magnetically soft steel alloy, and used 3.5-cm-diameter half-cylinders (not the “two meters” Kolm states). The cylinders rotated with their axes at right angles in a larger stationary block, with a thin layer of mercury (not just “equatorial”) between the rotors and the block. By going to a larger system made of annealed mild steel, with 5-cm-diameter rotors and variable geometry, we eventually produced a reversing dynamo. Kolm says that the field “reversed its direction every 20 minutes,” giving a period of 40 minutes. However, we reported that the reversals “mostly had a period of the order of 5 minutes, but that periods up to 20 minutes have been observed.”
In a separate approach in 1955, Bullard published the results of numerical integration of a very simple lumped-constant dynamo model. In that model, based on a Faraday-disk dynamo, the output current flowed through a coil to give positive feedback to the initial imposed axial magnetic field. 4 Essentially it was a single-stage amplifier with positive feedback; depending on the conditions, it could give periodic oscillation but no reversals. In 1958 Tsuneji Rikitake (in Tokyo, and never Bullard’s student) extended the lumped-constant model to two Faraday disks in series. 5 His model was equivalent to a two-stage amplifier with positive feedback; under certain conditions, it could give oscillations of increasing amplitude, which led to reversals. But the reversals were not periodic; that model is now recognized as an example of a chaotic system.
I have some other comments: The best-fit dipole is currently about 500 km from the geocenter—about 4%, not 10%, of Earth’s diameter. Also, the geomagnetic field does not reverse periodically; it is because the reversal record is so erratic that it can be used for dating rocks.
