Measurements are made of the initial permeability of iron and of certain high-permeability nickel-iron alloys over a frequency range extending to 30 Mc/s. The Polivanov-type eddy-current-loss calculations of Part 1 of the paper, intended to explain the eddy-current anomaly by considering a material with a magnetic-domain structure, are shown to be no more accurate than classical theory in predicting the behaviour of these specimens. In the thinnest Permalloys the high-frequency behaviour can be explained by a Snoek-type resonance phenomenon.