Coeval silicate melt and aqueous synthetic fluid inclusions were INTRODUCTION formed at 800°C and 2000 bars in the quartz-saturated haploIn recent years the study of silicate melt inclusions has granite–H2O–NaCl–KCl system. The equilibrium assemblage congained considerable popularity as a method to charsisted of Ab19·2Or31·1Qtz49·1 melt, quartz, and an aqueous solution acterize the PTX evolution of igneous processes (Andwith a composition of 7·4 wt % NaCl + 5·9 wt % KCl. The erson, 1974; Roedder, 1979; Lowenstern, 1995). Of melt contained 0·18 wt % Cl – and ~5·5 wt % H2O. The special interest has been the study of melt inclusions calculated partition coefficient of chloride between the melt and trapped in an immiscible fluid–melt system (Roedder, aqueous fluid (D m/aq Cl =C m Cl/C aq Cl ) is 0·021. The calculated dis1979, 1984, 1992; Hansteen & Lustenhouwer, 1990; tribution coefficient of Na and K between the melt and the aqueous Frezzotti, 1992; Naumov et al., 1992, 1996; Yang & phase D m/aq Na/K=(C m Na/C m K )/(C aq Na/C aq K )) is 0·40. Homogenization Bodnar, 1994; Student & Bodnar, 1996). Microtemperatures of synthetic silicate melt inclusions obtained by heating thermometric data from such inclusions can be used in 10·0°C/day increments in a tube furnace agreed with known directly to determine formation temperatures and presformation temperatures, with no size dependence. When measured sures, without the necessity of invoking difficult to prove in a high-temperature heating stage, a heating rate of 1°C/min assumptions or requiring an independent geobarometer produced homogenization temperatures that were about 10°C lower or geothermometer (Roedder & Bodnar, 1980). than those obtained from the same inclusions using a heating rate Roedder (1992) described specific types of magmatic of 3°C/min, although both heating rates produced homogenization immiscibility (immiscibility being defined as the existence temperatures above the formation temperature. A positive correlation of two or more non-crystalline, multi-component sobetween inclusion size and Th was observed for both heating rates. lutions which differ in properties and composition at Results confirm that microthermometric data from coeval silicate equilibrium), including silicate melt–silicate melt, silicate melt and aqueous fluid inclusions can be used to accurately predict melt–sulfide melt and silicate melt–aqueous fluid. In this P–T formation conditions if data from the smallest melt inclusions study we have investigated the behavior of synthetic fluid are used, or if the melt inclusions are homogenized using a slow inclusions trapped under conditions of silicate melt– heating rate (<1°C/min). aqueous fluid immiscibility in the haplogranite–