Secondary natural magnetizations have been examined by thermally demagnetizing a variety of rock types of mainly Tertiary age from Britain, India and Iceland. The conclusion is that the secondary magnetizations are most likely to be viscous magnetizations slowly acquired isothermally in the earth's magnetic field. A theory is developed which predicts an upper limit of temperature needed to erase thermally in the laboratory any viscous magnetization acquired by a rock at or near room temperature. This upper limit is near to 300°C, and is observed to be correct for all except four of the 123 specimens investigated.
It also appears that if, during laboratory heating, the Curie point of a specimen rises due to oxidation before complete demagnetization of the NRM, then the NRM direction may be retained by the specimen unchanged, and only disappears at the new, higher, Curie point.