[RightWhale]

It seems like, and this is a gut feeling, core convection would be too slow and too stable to change quickly enough to correspond with fluctuations and known reversals in the magfield. Something more mobile would be needed. Something that can change quickly and often.

[inquest]

What about the second possibility I raised? It appears that there are at least two main gyres circulating inside the globe in opposite directions from each other. That would explain why at the crust is splitting in two directions at the mid-Atlantic ridge. If they're moving in opposite directions, then the fields they're generating would largely cancel each other out, unless one is a bit stronger than the other. So all it would take for a reversal of polarity would be a change in the relative strength of the two gyres, so that the other one becomes dominant.