“…Finally, for a fixed L > 2R, as A 1" oo, So, by Harris' FKG inequality (see Roy [11] for a generalization which allows us to apply the inequality in our model), (a), (b) and (c) along with the preceding discussion, imply that for A > A0, some point in S(0, 0) is in an infinite occupied cluster with probability at least 1/2. Since, with positive probability, we can connect a ray in S(0, 0) to the ray ~0, we have, for ), > A0, ~0 is in an infinite occupied cluster with positive probability, i.e., A ~ A0 < oe.…”