The self-attenuation correction factor is used to relate the efficiency for a sample with a given matrix to the efficiency for an ideal sample with identical geometry but negligible photon attenuation. A certain linear relation for the efficiency for a given sample as a function of the efficiencies for a number of subsamples into which the original sample can be decomposed is established and experimentally validated. This relation can be used also in the case when the sample and the subsamples have different matrices. In this way the efficiency for volume samples with arbitrary compositions and densities can be constructed on the basis of the efficiencies (independently measured) for a number of basic geometries. Also a possibility to check the consistency of efficiency calibrations carried out with different standard sources (with different matrices) is provided.