Abstract
For pt. I, see ibid., vol.6, 1393. It is shown that a Hartree-Fock calculation of the mean energy of the 2p53p configuration in Ne I is able to predict eight of the ten 3p levels accurately, the exceptions being 1S0 and 3S1. The deviations are shown to correspond to Hartree-Fock wavefunctions for the 1S and 3S terms which are appreciably different from that for the mean energy as well as from one another. While the peculiarity of the 1S term is expected in view of the importance of the exchange interaction for this term, the deviation for 3S does not have the same origin. The results are used to improve on the parametric energy fitting and it is shown that a correction of 3S results in a considerable reduction of the mean error. The resulting eigenvectors are identical to those obtained earlier by introduction of the alpha L(L+1) correction. It is shown that this result is not fortuitous but that an infinite number of corrections give exactly the same eigenvectors. Similar improvements are found for the other known spectra in the Ne I isoelectronic sequence and the empirically-determined 3S corrections are in good agreement with the Hartree-Fock predictions. In the Ar I sequence the 3S correction is less important.