Transfer of angular momentum between the core and the mantle seems to be the only quantitatively satisfactory explanation of the irregular fluctuations in the length of the day, although some small part, hitherto supposed to be noise in the data, may have an origin in the atmosphere. Correlation between the westward drift, as determined from the motion of the off-centre dipole, and the changes in the length of the day since 1820 seems convincing support for the theory. The nature of the coupling between the core and the mantle is controversial: if it is electromagnetic, the quantitative difficulty has not been resolved. I have shown that if the torques resulting from electric currents induced in the lower semiconducting mantle are of an impulse character, those parallel to the axis (required to change the length of the day) are of the same order of magnitude as those in the equatorial plane, which may be the long-sought origin of the excitation of the Chandler wobble. Recent observational evidence of the geomagnetic secular variation, and from the Earth’s rotation, supports this view: the short timescale changes in the geomagnetic field at the core surface have been greatly underestimated.