ALBERT

Notes: Andalusite porphyroblasts are totally pseudomorphosed by margarite–paragonite aggregates in aluminous pelites containing the peak mineral assemblage andalusite, chlorite, chloritoid, margarite, paragonite, quartz ± garnet, in a NW Iberia contact area. Equilibria at low P–T are investigated using new KFMASH and (mainly) MnCNKFMASH grids constructed with Thermocalc 3.21. P–T and T–X pseudosections with phase modal volume isopleths are constructed for compositions relatively richer and poorer in andalusite to model the assemblages in an andalusite-bearing rock that contains a thin andalusite-rich band (ARB) during retrogression. Their compositions, prior to retrogression, are used in the modelling, and have been retrieved by restoring the pseudomorph-forming elements into the current-depleted matrix, except for Al2O3 which is assumed to be immobile. Compositional differences between the thin band and the rest of the rock have not resulted in differences in andalusite porphyroblast retrogression. The absence of chloritoid resorbtion implies either a pressure increase at constant reacting-system composition, or that its composition changed during retrogression at constant pressure, by becoming enriched in the progressively replaced andalusite porphyroblasts. T–X pseudosections at 1 kbar model this latter process using as end-members in X, first, the restored original rock and ARB compositions, and, then the same process, taking into account the change in composition of both as retrogression proceeded. The MnNCKFMASH pseudosections of rocks with different Al contents facilitate making further deductions on the rock-composition control of the resulting assemblages upon retrogression. Andalusite eventually disappears in relatively Al-poor rocks, resulting, as in this study, in a rock formed by chloritoid–chlorite as the only FM minerals, plus margarite–paragonite pseudomorphs of andalusite. In rocks richer in Al, chlorite would progressively disappear and a kyanite/andalusite–chloritoid assemblage would eventually be stable at retrograde conditions. The Al-silicate, stable during retrogression in Al-rich rocks, indicates pressure conditions and hence the tectonic context under which retrogression took place.

Notes: In statistically optimised P–T estimation, the contributions to overall uncertainty from different sources are represented by ellipses. One source, for a diffusion-controlled reaction at non-equilibrium, is diffusion modelling of the reaction texture. This modelling is used to estimate ratios, Q, between free-energy differences, ΔG, of reactions among mineral end-members, to replace the equilibrium condition ΔG = 0. The associated uncertainty is compared with those already inherent in the equilibrium case (from end-member data, activity models and mineral compositions). A compact matrix formulation is introduced for activity coefficients, and their partial derivatives governing error propagation. The non-equilibrium example studied is a corona reaction with the assemblage Grt–Opx–Cpx–Pl–Qtz. Two garnet compositions are used, from opposite sides of the corona. In one of them, affected by post-reaction Fe, Mg exchange with pyroxene, the problem of reconstructing the original composition is overcome by direct use of ratios between chemical-potential differences, given by the diffusion modelling. The number of geothermobarometers in the optimisation is limited by near-degeneracies. Their weightings are affected by strong correlations among Q ratios. Uncertainty from diffusion modelling is not large in comparison with other sources. Overall precision is limited mainly by uncertainties in activity models. Hypothetical equilibrium P–T are also estimated for both garnet compositions. By this approach, departure from equilibrium can be measured, with statistical uncertainties. For the example, the result for difference from equilibrium pressure is 1.2 ± 0.7 kbar.