On the consistency of CI calculations using MCHF orbitals

A recent calculation of the interaction between 3s3p 6 and the 3s 2 3p 4 md/ epsilon d series in Ar II by Mitroy et al. (1984) made use of a variationally determined d orbital to represent the complete d series in a fixed basis CI calculation. The authors show that this approach gives the correct wavefunction only for the particular state for which the d orbital was optimised. This result explains the differences between the calculations by Mitroy et al. (1984) and previous calculations by the authors.