Abstract
The van der Waals, Cahn—Hilliard theory of interfacial tension is reformulated for a fluid in the neighborhood of its critical point. The reformulated theory becomes equivalent to the Ornstein—Zernike, Debye theory of molecular correlations when the interface thickness is identified with the correlation length. When account is taken of the nonclassical behavior of the compressibility and coexistence curve, the theory is found to be in good agreement with independently known facts in three‐dimensional systems, yet slightly but unambiguously wrong in two dimensions. When one of the hypotheses of the original theory is replaced by an alternative hypothesis, the resulting theory is found to be correct in both two and three dimensions.