Nova Terra 52

space (TW diagram, Tera and Wasserburg 1972) inter- preted to be a mixture of initial common-Pb and Pb that formed due to in situ decay of U since mineralization. The age of formation is defined by the lower intercept with the Concordia. Plots and ages were calculated using Isoplot 3.71 (Ludwig 2007). All uncertainties are reported as 2σ. 4.4. Mineral chemistry Four garnet-bearing amphibolite samples were inves- tigated for mineral composition analysis and thermo- dynamic modelling (samples 117,344 to 117,350 in Supplementary Table 1). These samples were ana- lysed using the same routine applied to the previous samples, but in this case the Fe 2 O 3 and FeO contents were analytically differentiated. The composition of these samples differs greatly from that of the com- mon metagabbros of the unit, as they show higher contents of Al 2 O 3 and Na 2 O and lower MgO content (Supplementary Table 1). It is unclear if this composi- tional difference is primary or due to chemical changes induced by deformation. However, the com- positions are in equilibrium with the mineral assem- blage of the amphibolites, according to the representation of the compositional space using Cspace software (Torres-Roldán et al . 2000; Supplementary Figure 1). Sample 117,350 was selected for thermodynamic modelling. To achieve a better understanding of the mineral compositions and their textural variations and relationships, ele- mental X-Ray images of a representative area in sam- ple 117,350 were acquired with a JEOL Superprobe JXA-8900 M at the Universidad Complutense of Madrid. Operating conditions were 20 kV accelerating voltage, 300 nA beam current, 2 μm beam diameter, 12 μm pixel size, and 20 ms counting time. The elemental counts were obtained by means of WDS, including Si (Kα), Ti (Kα), Al (Kα), Fe (Kα), Mn (Kα), Mg (Kα), Ca (Kα), Na (Kα), K (Kα), and P (Kα). The images have been treated with Imager software (Torres- Roldán and García-Casco 2003, unpublished; see García-Casco 2007) and consist of the XR signals of the elements (colour-coded; expressed in counts) in the mineral(s) of interest (with polish defects, voids, and all other minerals masked out), set on top of a grey-scale base-layer that contains the basic tex- tural information of the scanned areas calculated with the expression Σ[(counts/nA per s)ixAi] (where A is atomic number, and i is Si, Ti, Al, Fe, Mn, Mg, Ca, Na, K, and P). Figure 4 depicts the phase-map of a representative area obtained after treatment of all X-Ray maps that shows the textural distribution of the mineral phases (colour code). Spot analysis was acquired on points selected using the X-ray maps. Additional spot analyses and garnet profiles were performed in other significant areas of the thin sec- tions (Supplementary Tables 6–9). The operation con- ditions were 20 kV, 20 nA, and the PAP correction procedure was used (Pouchou and Pichoir 1985). Structural formulas were calculated using the follow- ing oxygen normalizing basis (oxygens in brackets): garnet (12), amphibole (23), plagioclase (8), epidote (12.5). Fe(t) content is included as FeO. For calcula- tions in garnet profiles and compositional diagrams, Fe 2+ /Fe 3+ ratios were calculated using the stoichio- metric criteria of Droop (1987). Fe 3+ = Fe total is assumed for the epidote group minerals, and Fe 3+ content in amphiboles is calculated following Hawthorne et al . (2012). Mineral abbreviations are after Whitney and Evans (2010), with the exception of the pistacite end-member (Ps). The atomic concen- tration of elements per formula units is abbreviated as apfu. 4.5. Pseudosection modelling The phase assemblage history and P-T path of the garnet amphibolites were investigated using phase diagrams cal- culated for the specific bulk compositions (pseudosec- tions) in the system MnNCFMASHTO (MnO-Na 2 O-CaO- FeO-Fe 2 O 3 -MgO-Al 2 O 3 -SiO 2 -TiO 2 -H 2 O). The inclusion of K 2 O in the system favours systematic development of biotite in the pseudosections, but this mineral has never been found in the metabasites. Therefore, a system without K 2 O has been modelled. All calculations were performed with the software Theriak-Domino (De Capitani and Brown 1987; De Capitani and Petrakakis 2010) and con- sidering H 2 O saturation. Measured FeO and Fe 2 O 3 contents of the samples were used to calculate the total oxygen in the rock. The activity-composition models used in thermo- dynamic calculations were epidote and talc (Holland and Powell 1998), staurolite and cordierite (Mahar et al . 1997; Holland and Powell 1998), chlorite (Holland and Powell 1998), chloritoid (Mahar et al . 1997; White et al . 2000), white mica (Coggon and Holland 2002), carpholite (Kelsey et al . 2004), feldspar (Holland and Powell 2003; Baldwin et al . 2005), orthopyroxene (White et al . 2002, 2007; also in Baldwin et al . 2005), amphibole (Diener et al . 2007), clinopyroxenes (Green et al . 2007), and ilmenite, spinel, biotite, and garnet (White et al . 2007). Different pseudosections were calculated for samples with the most characteristic composition and mineral assemblages (samples 117,349 and 117,350), using the internally con- sistent databases 5.5 THERMOCALC (tcdb55c2d; Holland and Powell 1998; updated November 2003) and 6.2 INTERNATIONAL GEOLOGY REVIEW 9 Mérida Ophiolite (SW Iberia)