The estimation of triplet invariants from multi-wavelength data

A two-step procedure is presented for the estimation of triplet invariants from multi-wavelength data. In the first step wavelength-independent structure-factor magnitudes of both the total structure and the light-atom substructure, together with associated structure-factor phase differences, are calculated explicitly via a modified Singh & Ramaseshan [Acta Cryst. (1968), B24, 35-40] procedure. In the second step these quantities are employed as conditional information in the probability distribution of a triplet invariant which can be derived from Hauptman's [Acta Cryst. (1982), A38, 289-294] joint probability distribution for an isomorphous pair of structures. Test calculations on the structure of the iron-containing protein ferredoxin show the feasibility of the procedure.