Our letter was motivated by experimental observations of evaporative dewetting of water films on a mica substrate [CITE]. In equilibrium the system shows complete wetting. On evaporating, holes nucleate in the film which retracts and finally collects into droplets. These droplets form patterns reminiscent of non-wetting fluids [CITE], [CITE], yet on return to saturated vapor conditions they spread with a smooth and convex foot. Indeed the finite contact angle induced by evaporation is implicit in the nucleation of holes in the liquid film. Unfortunately the analysis is in error. This was pointed out to us by Raz Kupferman, who showed that a stable drop cannot be explained by equilibrium arguments if f(L) is a convex function. We published a retraction at the first opportunity [CITE]. Bonn and Meunier go into detail but offer no new insights to explain the phenomena.