IN his article elsewhere in this issue (p. 200), Prof. E. W. Brown makes out a strong case for the belief that the chief outstanding anomalies in the motions of the moon, the sun, and the inner planets are due to errors in time-keeping arising from variations in the earth's rate of rotation. A set of the most puzzling departures of astronomical observation from gravitational theory has therefore been brought into harmony. The iatio of the chief inequalities in the longitudes of the sun and moon is such as to indicate that the variations of the rate of rotation are not associated with any corresponding transfer of angular momentum to the moon's orbital motion, and therefore Prof. Brown attributes them to changes in the earth's moment of inertia. Symmetrical swellings and contractions of the earth would certainly give rise to effects of the type observed. But elastic vibrations of the earth of this type would have periods of the order of minutes, not years, and it is difficult to think of any geological or seismological process that could give simultaneous expansion or contraction over the whole earth. An alternative explanation resting on a phenomenon already known to occur may be found in variations of the thickness of the polar ice-caps. Partial melting of the polar ice and redistribution of the water over the ocean would give changes of the moment of inertia such as are required, and the amount does not seem prohibitive. A small secular increase in the amount of ice would also give a secular acceleration of the earth's rotation, which would go some way towards explaining the outstanding discrepancy between theory and observation with respect to the ratio of the secular accelerations of the sun and moon. There is evidence of a warm climate in Europe about 2000 B.C., which would fit this suggestion.