“…Then one can show that there exists a sequence of coaxial cylinders C n with common axis l, radii going to infinity as n → ∞ and symmetric with a certain fixed small positive height h with respect to Π, such that M 1,n ∩ C n consists only of disks for each n (because for n large, the part Ω n of M 1,n ∩ C n outside certain cone with axis l centered at p 0 consists of a highly-sheeted double multigraph over an annulus in Π, hence Ω n is topologically a disk; from here one directly obtains that M 1,n ∩ C n is a disk for n large). Using a suitable modification of the proof by Colding-Minicozzi of Theorem 0.1 in [8] with the cylinders C n replacing balls with radii going to ∞, one deduces that after passing to a subsequence, that the disks M 1,n ∩ C n converge to the foliation L Π by parallel planes of a neighborhood of Π, with singular set of convergence S(L Π ) consisting of exactly one Lipschitz curve passing through p 0 . By Meeks' regularity theorem [24], S(L Π ) is a segment contained in l. After repeating this argument at the boundary planes of L Π , we see that L Π can be enlarged to the foliation L 1 of R 3 by planes parallel to Π and that a subsequence of the M 1,n (denoted in the same way) converges to L 1 (in particular, L = L 1 ), with singular set of convergence S(L 1 ) = l. This implies M 1,n intersects B(2) in disks for n sufficiently large, which contradicts that M 1,n ∩ B(1) contains a homotopically nontrivial curve.…”