MATSUI and I have estimated partition coefficients for some elements1–4 and discussed the significance5,6 of estimating them. More recently7, I have estimated partition coefficients for rubidium and strontium. However, we have little knowledge about what conditions affect partition coefficients and to what extent they can be influenced. For example, it is conceivable that temperature and pressure have some effects on them. But results so far obtained suggest that the effects of temperature and pressure are not serious. A difficult problem, however, is the influence of calcium. If magnesium and/or ferrous iron are the overwhelmingly predominant metallic cations in a liquid and a solid separated from it, the relationship between the chemical ‘environment’ and the partition coefficient is very simple. But if calcium is abundantly present its effect is superimposed on that of magnesium. The lower limit of concentration required for the appearance of this calcium effect is, as yet, unknown. However, since the effect is thought to arise when calcium becomes one of the host elements of rock-forming minerals, it is most probable that it occurs when the concentration of calcium in the ‘liquid’ (magma) is enriched to some definite degree. Furthermore, the combined or concurrent effect of calcium and magnesium may reflect the simple atomic ratio in which they are present in crystals. (These possibilities suggest that the apparent change of partition coefficient due to calcium is somewhat discrete.)