Fifty experiments are presented which explore ferric/ferrous redox equilibrium in melts in the CaO-FeO-FeO 1.5-SiO 2 system. Plots of ln( FeO 1.5/FeO) against ln(ƒ o 2) for compositions equilibrated at a single temperature have slopes of 0∼.21. This result is consistent with observations in a wide variety of silicate liquid compositions, including compositions corresponding to those found in nature, but contrasts with the slope expected if ferric-ferrous mixing were ideal (0.25). Compositional trends for redox equilibrium in the lime-iron-silicate system contrast with those found in the Na 2O-FeO-FeO 1.5-SiO 2 system. It is shown that minor amounts of quench crystals in the glasses significantly increase ferric/ferrous ratios from their equilibrium values. The results of quench crystal-free ferric/ferrous equilibration experiments are used to constrain a simple empirical equation by which one can estimate the distribution of iron between the ferric and ferrous species at a given composition, temperature, and oxygen fugacity. This predictive model reproduces measured ferric/ferrous distribution within analytical error. The simplest thermodynamically valid model by which the novel systematics of the iron redox reaction in silicate melts can be rationalized involves partial association of FeO and FeO 1.5 to a hypothetical melt species with FeO 1.3 stoichiometry. We apply our empirical ferric/ferrous estimation model to recalculate the ferric/ferrous distribution in the liquid-density experiments of Dingwell and Brearley (1988) and Mo et al. (1982). A simple (ideal) volume of mixing model is derived by which the volumes of CaO-FeO-Fe 2O 3-SiO 2 melts with more than 20 wt% silica can be estimated to within ±0.16 cc/gfw ( 2σ $∼0.7% ). This average residual in our linear volume model is well below the reported experimental uncertainty ( DINGWELL and BREARLEY, 1988), indicating that partial molar volumes can be considered independent of composition within the experimental resolution in melts in this composition range. Density data in melts with very low silica ( CaO/SiO 2 less than 0.3) suggest a negative volume of mixing between CaO and SiO 2 components.