In this note we explain why the three-dimensional counterexample of section 7.2 of the paper under consideration [E. Zuazua, SIAM J. Control Optim., 34 (1996), pp. 1473-1495] is incorrect. It concerns the existence of eigenfunctions for the Dirichlet Laplacian of a special form. We claimed the existence of particular three-dimensional domains in which these eigenfunctions do exist. However, in a recent joint work with G. Sweers [G. Sweers and E. Zuazua, J. Elasticity, to appear], we have proved that they may not exist in any bounded Lipschitz domain in dimensions n greater than or equal to 3.