Abstract
The 1P to 3P separation has been determined by minimizing the energies of the two states explicitly in the Hartree-Fock procedure and then taking the difference of the total energies determined for the two terms. This procedure gives extremely accurate results for K II and Rb II. The accuracy diminishes for more highly ionized atoms, but is still considerably better than that found using the conventional approach. The results of the Hartree-Fock calculations turn out to be qualitatively different for K II and Ca III (Rb II and Sr III). For K II, the 1P term is calculated to be below the prediction made by the usual approach in all 3p5nd configurations. In Ca III this result holds for the 3p53d configuration, while the calculations show the opposite result for all higher configurations. The reasons for this difference are discussed.