An error in the proof of Corollary 6.2 of [1] has been pointed out by Gerard Ben-Arous. The computation of M(x,λ) in the case λo = 0 on p. 243 is incorrect, because M(x,λ) = 0 when λ0 = 0 and λjg (x) = 0 for all k. (There is also a factor of \ missing in the formula as stated for λ0 φ 0, but this is not significant.) Thus when applying the Pontryagin Theorem (Theorem 6.1) it is necessary to consider one additional case: λo = 0 and M(x9λ) = 0 along the curve. The conditions of the theorem in this case lead to the equations