In a recent paper, Curtiss and Powers claim to have shown that the phase shift η₁, for the case when the effective potential presents three turning points, contains in the classical limit a contribution from the classically inaccessible potential well between the inner two turning points, but is independent of the form of the barrier between the outer two turning points. It is shown that what Curtiss and Powers calculated is the semi-classical form of the average of η₁ over a small range of energies. If the correct semi-classical η₁ is used, the contribution of classically inaccessible regions to the averages of observable quantities such as cross sections vanishes in the classical limit, as one would expect.