This thesis describes a new technique for determining the magnitude of the palaeomagnetic field, and the results of applying it to a transition from reversed to normal polarity. Several successful experiments were carried out on modern lavas and on archaeomagnetic material to test the accuracy of the technique. The results of these experiments demonstrate that this technique gives accurate, consistent results. The technique was applied to samples from lavas that were extruded during a palaeomagnetic field reversal. The variation of the magnitude of the palaeofield with time was successfully determined with a mean error (standard deviation) of only O.03T (3000 gamma). The results indicate that the palaeofield was large and st~ble during a period when the virtual magnetic north pole seems to have lingered at the geographic equator. This and other published results suggest that an intermediate state .of the geomagnetic field can exist which is sometimes as strong as the more usual normal and reversed states, and which endures metastably for short periods of time. This means', among other things, that any single transition may appear to be quite complex, although the average transition is known to involve fairly simple geometry. It also constrains future theories of the generation of the geomagnetic field to include the phenomenon of "intermediate metastable states".