Summary: Wherever the necessary data are available with adequate accuracy, the most satisfactory presentation of a chemical analysis is as empirical unit-cell contents, calculated from the cell-sides, density, and chemical data. When empirical unit-cell contents cannot be calculated, some assumption is necessary for a presentation having claims to any structural significance; the assumption to be made should be carefully chosen and explicitly stated. A number of assumptions suitable in special cases are discussed, as is the technique of calculating empirical unit-cell contents, and the effects of errors in the chemical analysis on the several forms of presentation.