AbstractThe key to the understanding of the gross changes of climate throughout geological time indicated by the geological record in the various continents depends on finding a physical method of determining latitudes in the geological past. The palaeomagnetic method seems to be the most hopeful, but for this one must assume that the average geomagnetic field is a dipole along the axis of rotation. The palaeomagnetic evidence shows this to be true for late Tertiary times – the poles derived from the mean palaeomagnetic directions of Tertiary rocks cluster about the present geographical pole and are distinctly different from the present position of the magnetic pole. There are strong theoretical grounds for believing that it has been true throughout the earth's history. The theory of the irregular fluctuations in the length of the day indicates that the rotation of the core and mantle are only weakly coupled. The electro‐magnetic torque to which this coupling is ascribed is variable because of the short term changes in the geomagnetic field. These fluctuations appear to accumulate with time and so there is always a small differential angular velocity between core and mantle, probably of either sign. Therefore, to an observer fixed on the earth's crust, the earth's magnetic field is axisymmetrical when averaged over periods of the order of the secular variation, i.e., extremely small times geologically speaking.The palaeomagnetic surveys of the various continents show systematic differences which, it appears, cannot be explained even by assuming that the earth's magnetic field was non‐dipolar, though axial, in times earlier than the Tertiary. Continental drift, as well as polar wandering, must therefore be inferred.There is a prima facie case that the latitudes thus inferred for the various continents from palaeomagnetic observations do agree with the commonly held inferences concerning the past climates of the earth made by geologists from the study of evaporite deposits, red beds, zonation of fossils, and glacial deposits. A new important check has been provided by palaeowind observations. The inference from the palaeomagnetic studies of Europe and North America that these continents were in low latitude between the Devonian and Triassic is consistent with the wind directions inferred from dune sandstones in Britain and the western United States. The consistency of the directions in the western United States over large distances seems to show that the winds were not affected by local geography.It appears that an explanation of Continental drift involves the hypothesis of convection currents in the earth's mantle which requires that the mantle is fluid when subjected to small stresses existing for many millions of years.