Letter to the Editor

Two theorems presented in this paper are incorrect. Theorem 2 stated by Hwang can be reformulated as follows. Theorem 2: tion only network if m 25 qiq2(n2 ~ 1) + Qi + rii - 1. Proof: u{m, under nu ru n2, r2) is nonblocking as a rx 2s q^rii and (71, q2) r2 2s multiconnecif and Strategy 2, for qqzn2, Sufficiency. Consider the connection of the pair (x, Y). The input switch that contains x can be connected already to at most Mi -- 1 distinct middle switches under Strategy 2. Each output switch in Y can be connected already to at most n2q -- 1 distinct middle switches under Strategy 2. Since | Y | ^ 72, we need q2 sets of n2qx -- 1 distinct middle switches, if the sets are disjoint. However, under Strategy 2 these sets are not disjoint and the number of middle switches must be replaced by 72(^2(71 - 1) - (71 ~ 1)(