A previous paper by Starace showed that the length formula for the electric-dipole matrix element is the only consistent one for calculations that solve for the exact eigenfunctions of an approximate Hamiltonian containing a nonlocal potential. It is emphasized here that disagreement between such a length-formula calculation and experiment indicates inadequacy of the approximate Hamiltonian and not that the velocity formula may be preferable.

• Received 30 November 1972

DOI:https://doi.org/10.1103/PhysRevA.8.1141