“…Naively, one could suppose that we need a critical droplet with one face of supercritical size for the (1, 1) model, and two of supercritical size for the (1, c) model. However, we can do better than that, following essentially the same line of thought as in [17] which is similar to that in [2] and which was based on an unpublished observation of Roberto Schonmann. There a suitable critical droplet for the (1, 2) model was found to be a strip of length C p ln 1 p , and width 2, with C a large enough constant.…”