Melting as a disorder phenomenon

Abstract The essential nature of the usual theory of melting-points is indicated by fig. 1. It is supposed that a property X (degree of order, density, heat content, etc.) of either phase is a continuous function of temperature, but that at a particular temperature, where the two phases have equal thermodynamic potential j) = u + pv — Ts, the stability changes over, this being the meltingpoint. This theory gives no explanation of the fact that the continuation of the crystal curve beyond the melting-point has never been realized, i.e. that whereas supercooling of a liquid occurs, superheating of a crystal is unknown. This theory, which seemed natural when a liquid was regarded as a dense gas and a crystal as a perfect space lattice, is now by no means so obvious, since our knowledge has increased in recent years (Mie 1903). We know on the one hand from X-ray diffraction experiments that a liquid has a fairly high degree of order, only with a limited range of coherence; and on the other from these, from dielectric experiments and observations on diffusion, electrical conduction and specific heat, that a fairly large degree of disorder may exist in a crystal.