AbstractIt is not possible to express the experimental relation between the proportions of the two oxides of iron and the equilibrium oxygen pressure over the single oxide phase magnetite, by the usual type of mass action law. A constant may be obtained, however, valid over the entire composition range in which the mixed oxide has a spinel structure, if an expression for crystal defects is included in the equation. When the assumption is made that the iron oxides form an intermediate spinel, the defect term may readily be calculated in terms of experimental concentrations. This constant expresses the experimental results very closely over the entire concentration range in which the spinel structure is stable.The treatment also includes the effect of adding other metal oxides to iron oxide, in so far as the data are available. The same structural hypothesis appears to be valid for both the magnesium‐iron spinels and the chromium‐iron spinels.The success of the same treatment when applied to molten iron oxide leads to interesting conclusions about the liquid structure. The defect concentration is much greater, but the equilibrium constant is nearly the same as for the solid. Only a preliminary discussion of the effect of other metal oxides on the liquid equilibrium seems justif