Using many-particle perturbation theory an attempt has been made to improve the author's previous calculation of the thermal expansion of sodium chloride and to assess the effect of including anharmonic terms in the crystal potential. The results are disappointing; agreement with experiment is poor except at low and moderate temperatures. One reason for this is the poor convergence of a method based on the series expansion of the crystal potential in terms of particle displacements, 8 point further. illustrated by calculations of specific heat and compressibility.