A framework is set up for carrying out approximate SCF π-electron calculations on complexes of conjugated ligands. For calibration, calculations were carried out on series of polynuclear aromatic hydrocarbons and some azaderivatives, and the parameters adjusted to secure the best least-squares fit to experiment. Intensities of absorption bands were calculated by use of both dipole lenght and dipole velocity operators. The differences between the two methods are discussed with particular emphasis on charge-transfer transitions, for which the predictions of the two methods vary considerably with respect to polarisation and dependence on ligand size. The differences may be eliminated by more complete configuration interaction.

The calculated spectrum of phenanthroline and the effect of a point charge upon it are shown to be in reasonable agreement with experiment, and the effects of methyl substitution on base strength are examined.

Calculations on a hypothetical low-spin d⁶ mono-complex of phenanthroline are described. The observed band envelope of the trisphenanthrolineiron(II) charge-transfer band includes all the transitions to the two lowest of the empty ligand orbitals. The value of the metal–ligand resonance integral derived on this assumption is very close to those derived from charge-transfer intensities in other iron(II) complexes, and to that derived from the dipole velocity operator by Linderberg's method. One of the transitions, mainly composed of charge-transfer to the lowest but one of the empty orbitals of the isolated ligand, is much the strongest. The charge-transfer energy is extremely sensitive to the charge on the iron atom (taken as a parameter in these purely π-electron calculations); a value of +0·3 or +0·4 is required before taking into account back-donation, which increases the net charge to ca. +0·6. The classification of bands into charge-transfer and π–π* is shown to be quite clear-cut. The intensities of the various charge-transfer components as predicted by the dipole length and dipole velocity methods depend on the extent of configuration interaction as expected. The origin of the dipole length intensity agrees with that suggested in our earlier work.