Comment on relativistic transition-probability calculations for the Be isoelectronic sequence

Recent calculations of oscillator strengths by Armstrong, Fielder, and Lin employing relativistic multiconfiguration Hartree-Fock (MCHF) methods give a disparity between length- and velocity-gauge results. It is shown here that the disparity is due in part to the neglect of certain "exchange overlap" terms in the evaluations of transition-matrix elements in MCHF calculations. Including such exchange overlap terms brings velocity-gauge values into substantial agreement with length-gauge results for high-$Z$ members of the Be isoelectronic sequence. The disparity for lower members of the sequence is increased and the resulting difference between length- and velocity-gauge values reflects the changing reliability of the MCHF calculation as $Z$ increases along the sequence.