Knowing the pressure of crystallization of peraluminous felsic rocks is essential to understanding their petrogenesis and implications for the tectonic histories of orogenic belts. Phase relations studies in the haplogranite system reveal that the stability field of quartz enlarges while the stability field of feldspars shrinks with increasing pressure. The result is that the compositions of the cotectic melts (saturated in quartz and feldspar) become less silicic with increasing pressure. Furthermore, the position of the quartz - feldspar cotectic is insensitive to water activity in the melt. This relationship provides an opportunity to estimate the pressure at which melt, quartz, and feldspar equilibrated. The accuracy of pressure estimation depends on accurately estimating the composition of the cotectic melt. However, additional components drive the composition of the cotectic melt away from that in the simple haplogranite system. For example, anorthite component (An) shifts the cotectic curve towards the Qz-Or join, and excess aluminum in the melt (i.e., peraluminous composition) drives the cotectic curve toward the Qz apex. A majority of felsic magmas (> ~ 70 wt% SiO.sub.2) are at least slightly peraluminous, and strongly peraluminous magmas are common in orogenic settings. A reasonable estimation of the pressures (depths) of crystallization of peraluminous magmas, which is critical for understanding accompanying tectonic processes, requires a better understanding of the effect of excess Al on the cotectic, and specifically on quartz saturation. We have compiled 32 experimental cotectic glass compositions in the felsic system, in which ASI (= Al.sub.2O.sub.3/(CaO + Na.sub.2O + K.sub.2O)(commonly defined identically to A/CNK), in mole content)) range from 1.10 to 2.13. The scheme of (Blundy and Cashman, Contrib Mineral Petrol 140:631-650, 2001) was chosen to correct for the effect of An on the position of the cotectic. The measured Qz of the experimental cotectic melts was compared to Qz calculated using the Blundy and Cashman correction at the experimental pressures. We assumed that the discrepancy between the corrected Qz content and the experimental Qz content was caused by excess aluminum in the melt. Based on this assumption, we can regress the relationship between excess aluminum (ASI-1) and 'excess' Qz content. The positive correlation between ASI-1 and increased Qz demonstrates that excess aluminum increases the cotectic Qz content, i.e. quartz solubility, in the peraluminous system. The deviation of Qz content systematically increases from 4 to 20 wt%, with ASI rising from 1.10 to 2.13; the increase is especially rapid and clearly defined between 1.1 and 1.5, the normal range for natural peraluminous melts. The departure from ASI = 1.00 (ASI-1) systematically affects normative Qz in melt at a given P. We calibrate this relationship as: [Formula omitted]where [delta]Qz is the deviation of normative Qz in quartz- + feldspar-saturated melts as ASI increases from 1.00. Uncertainty in our calibration is fairly high (~ 120 MPa (one sigma), based on a standard error from the regression). This uncertainty might be influenced by many factors: (1) the accuracy of the compiled experiments; (2) disequilibrium in the compiled experiments; (3) uncertainty in the estimation of quartz + feldspar-saturated melt compositions based on natural rock samples; (4) the effect of other elements on the position of the cotectic curve, such as F, P, B, Fe, Mg, Ti, Mn. The empirical barometer was applied to natural examples (compositions of peraluminous glasses and granite compositions that permit estimation of melt compositions) with other thermobarometric constraints (cordierite-saturation, GASP (garnet-Al.sub.2SiO.sub.5-quartz-plagioclase), GBPQ (garnet-biotite-plagioclase-quartz), TitaniQ barometers for estimated melts; host rock metamorphic conditions). We propose that, where natural samples provide good estimates of the compositions of peraluminous melts that were saturated in quartz + feldspars (ideally two feldspars), our calibration provides useful estimates of the pressure of equilibration. Note that whole-rock compositions of granites cannot be assumed to represent melt composition; hence, pressures calculated by our barometer may not be meaningful unless applied to rocks whose compositions are likely to represent melts on the quartz-feldspar cotectic. The most obvious candidates are rhyolite glasses that host quartz + feldspar phenocryst assemblages and phenocryst-poor, aplitic-textured granites with textures indicating simultaneous crystallization of quartz and feldspar. The pressure calculation can be easily performed with the user-friendly spreadsheet (online supplement Table S4). As anticipated, the calculated pressures for peraluminous compositions are significantly higher than current quartz-saturation-based barometers would otherwise suggest., Author(s): Lei Yang [sup.1] [sup.2], Calvin F. Miller [sup.3], Fu-Yuan Wu [sup.2] Author Affiliations: (1) grid.411288.6, 0000 0000 8846 0060, College of Earth Sciences, Chengdu University of Technology, , 610059, [...]