Abstract
As known for the last ten years, calculations made with the linear theory agree poorly with the experimental results for the 3d6 configuration of Fe III, compared with the agreement obtained for the 3d54s configuration of that spectrum. It is shown that effects of 3s3d64s on 3s23d54s satisfy the necessary conditions for linear behavior, consistent with the close agreement obtained in the latter configuration of Fe III. Corresponding effects of 3s3d7 on 3s23d6 are not linear. When they are included in the calculation, close agreement is obtained in the 3d6 configuration as well; the mean error is reduced from ±364 cm1 to ±66 cm1. An additional parameter, γ, is introduced to define this nonlinear effect. The value of γ obtained from the experimental data by use of least squares is confirmed reasonably well by the value calculated from Watson's self-consistent field results. The parameters of the linear theory are changed considerably when this non-linear effect is accounted for.

This publication has 12 references indexed in Scilit: