Application of the Gibbs adsorption isotherm to the available data for the pH-dependence of the surface tension of water leads to two conclusions. First, a strong hydroxide surface affinity implies the surface tension nearly pH-independent near pH 7. Second, surface tension data at high bulk hydroxide concentration are consistent with a large surface excess of hydroxide at low bulk concentration. These in part contradict the claims made by Winter et al., but also allow a reinterpretation of their spectra, revealing the essential importance of their conclusion that hydroxide is absent from the top few molecular layers near the water-vapour interface. (C) 2009 Elsevier B.V. All rights reserved.