Methods of classifying features of the lunar surface according to their relative ages are discussed. The origin of the lunar craters is thought to be meteoritic, and some new considerations are given concerning high velocity impacts. The investigation of relative ages suggests that a process of erosion has been operating, though of course much more slowly than on the Earth, and the hypothesis is then developed that eroded dust is subjected to an agitating action causing it to settle with the formation of flat, level surfaces. It is then unnecessary to introduce any hypothesis of lava flow for the generation of the flat surfaces, and a cold history for the lunar surface is then possible. The triaxial shape can then not be attributed to an equilibrium shape in a past epoch at another distance from the Earth, but merely as a shape that departs from the equilibrium by an amount compatible with the strength of the material.