How the Neoproterozoic S-isotope record illuminates the genesis of vein gold systems: an example from the Dalradian Supergroup in Scotland

Abstract The genesis of quartz vein-hosted gold mineralization in the Neoproterozoic–early Palaeozoic Dalradian Supergroup of Scotland remains controversial. An extensive new dataset of S-isotope analyses from the Tyndrum area, together with correlation of the global Neoproterozoic sedimentary S-isotope dataset to the Dalradian stratigraphy, demonstrates a mixed sedimentary and magmatic sulphur source for the mineralization. δ34S values for early molybdenite- and later gold-bearing mineralization range from −2 to +12‰, but show distinct populations related to mineralization type. Modelling of the relative input of magmatic and sedimentary sulphur into gold-bearing quartz veins with δ34S values of +12‰ indicates a maximum of 68% magmatic sulphur, and that S-rich, SEDEX-bearing, Easdale Subgroup metasedimentary rocks lying stratigraphically above the host rocks represent the only viable source of sedimentary sulphur in the Dalradian Supergroup. Consequently, the immediate host rocks were not a major source of sulphur to the mineralization, consistent with their low bulk sulphur and lack of metal enrichment. Recent structural models of the Tyndrum area suggest that Easdale Subgroup metasedimentary rocks, enriched in 34S, sulphur and metals, are repeated at depth owing to folding, and it is suggested that these are the most likely source of sedimentary sulphur, and possibly metals, for the ore fluids.

The Dalradian rocks of Scotland and Ireland have been extensively studied since the mid-nineteenth century (Murchison & Geikie 1861;Bailey & Macgregor 1912;Tilley 1925). The sequence is generally considered to be mid-Neoproterozoic (Cryogenian) to at least mid-Cambrian in age (Tanner & Sutherland 2007;Stephenson et al. 2013). However, there are only limited horizons where precise chronostratigraphy is available (Dempster et al. 2002;Rooney et al. 2011). Furthermore, correlation with other Neoproterozoic sequences is challenging, given that the Dalradian rocks have undergone polyphase deformation and regional metamorphism, locally reaching upper amphibolite grade during the Grampian Event of the Caledonian Orogeny (Oliver 2001;Stephenson et al. 2013). The Neoproterozoic Era is recognized as a crucial time in Earth history, incorporating two snowball Earth events (Hoffman et al. 1998) coupled with large fluctuations in seawater C-, O-, Sr-and S-isotope ratios (Halverson et al. 2010). The increasing understanding of Neoproterozoic events and global isotope variations (Halverson et al. 2010;Halverson & Shields-Zhou 2011) within the evolving Earth has impacted on the understanding of the Dalradian sequence and is beginning to refine the possible chronostratigraphies, particularly through the use of stable isotope stratigraphy (Thomas et al. 2004;Prave et al. 2009;Moles et al. 2014).
The Dalradian sequence also hosts the UK's largest resource of gold in Northern Ireland at Curraghinalt, and its only active metal (gold) mine at Cavanacaw (Fig. 1), with resources of 2 700 000 and 438 000 oz Au, respectively (Dalradian Resources Inc. 2012;Galantas Gold Corporation 2013). In Scotland, the Cononish deposit, near Tyndrum, hosts a JORC-compliant resource of 169 000 oz Au and 631 000 oz Ag in the combined Measured, Indicated and Inferred categories (Scotgold Resources Ltd 2012a). More widely, the Tyndrum area contains a number of gold prospects and many mineralized occurrences within a 10 km radius of Cononish (Hill et al. 2011;Tanner 2012). The origin and timing of mineralization in the Tyndrum area are subject to on-going debate (e.g. Curtis et al. 1993;Goldfarb et al. 2005).
The massive quartz veins which host the gold ores and other associated mineralization at Tyndrum are dominated by sulphides. Early limited work on the isotopic composition of the sulphides was interpreted to suggest a mixed sedimentary/magmatic sulphur source (Pattrick et al. 1983(Pattrick et al. , 1988Curtis et al. 1993). This paper presents an extensive new S-isotope dataset for sulphides from Cononish and for newly identified occurrences of a range of mineralization styles in the area, together with a suite of local host metasedimentary rocks. The global Neoproterozoic sedimentary sulphide S-isotope dataset is correlated with recent data for the Sisotope stratigraphy of the Dalradian Supergroup to constrain the S-isotope composition of the local sequence where data are absent. The potential input of sedimentary and igneous sulphur is modelled to investigate the sources of sulphur in the Tyndrum area mineralization. From this it is shown that the majority of the sulphur in the mineralization must be derived from the Dalradian sequence and the proportion of magmatic sulphur is likely to be low, but nevertheless genetically significant. Furthermore, it is demonstrated that the local host metasedimentary rocks are unlikely to have been the dominant source of sulphur for the gold vein mineralization; instead it was stratigraphically younger, but structurally underlying, Dalradian rocks which were the most likely source. This constrains a model for the source of metals and the fluid pathways for the development of mineralization.

Global context
The formation of the Rodinia supercontinent (1100-1000 Ma; Kennedy et al. 2006) and its subsequent rifting provides the setting for the early Neoproterozoic Era. Wide-scale orogenic activity was accompanied by the drawdown of biolimiting elements, such as P, Fe and C, during weathering reactions (Lenton & Watson 2004) and an associated increase in atmospheric O 2 . The onset of these conditions ultimately produced changes within the biosphere that led to the development of oceanic metazoans and a primitive land biota (Knauth & Kennedy 2009;Parnell et al. 2010). By the mid-Neoproterozoic Era the drawdown of CO 2 and oxidizing atmospheric conditions caused the onset of global episodic glaciations extending to low latitudes (Hambrey & Harland 1981;Hoffman et al. 1998) with significant associated climate fluctuations (Knauth & Kennedy 2009). The Sturtian glacial episodes are variably constrained from 746 to 663 Ma (Condon & Bowring 2011) and the Marinoan glacial episode to a period of ,10 Ma with global termination of glaciation by c. 653 Ma (Condon & Bowring 2011). The northwards drift of Laurentia relative to Gondwana at c. 570 Ma began to open the Iapetus Ocean (Cawood et al. 2001). The subsequent closure of the Iapetus Ocean caused the Grampian Orogenic Event during the mid-Palaeozoic (Soper & Hutton 1984;Pickering et al. 1988;Soper et al. 1992).

Deposition and tectonic setting
The Dalradian sequence is bounded in Scotland by the Great Glen Fault to the north and Highland Boundary Fault to the south -both crustal-scale structures (Fig. 1). The sequence was deposited along the developing east Laurentian passive margin during a period of ocean widening (Anderton 1985). The Dalradian Supergroup has a depositional history spanning the Neoproterozoic (Cryogenian) to mid-Cambrian (Tanner & Sutherland 2007;Stephenson et al. 2013) and comprises marineclastic sedimentary rocks with occasional carbonate beds and minor volcanic rocks (Stephenson et al. 2013). The oldest rocks are psammites and semi-pelitic schists deposited in an extensional basin, collectively called the Grampian Group (Figs 1 & 2). The overlying Appin Group is characterized by a limestone-pelite -quartzite assemblage deposited in a relatively stable shelf environment (Wright 1988). The Argyll and Southern Highland Groups, overlying the Appin Group, contain significant black slates and graphitic schists, and an increased incidence of mafic lavas and sills, grading upwards into coarse turbidite sequences (Harris et al. 1978;Anderton 1985), with the Argyll Group host to locally developed stratabound exhalative mineralization (Stephenson et al. 2013). The upper-Appin and lower-Argyll Group stratigraphy is absent in the Tyndrum area (Fig. 2); the missing stratigraphy is represented by the Boundary Slide (formerly termed the Iltay Boundary Slide; Bailey 1922;Hutton 1979;Roberts & Treagus 1979;Tanner 2012).

Chronostratigraphic and biostratigraphic constraints
The age of deposition is poorly constrained; a number of workers have advocated correlations within the existing age constraints ( Fig. 2; Thomas et al. 2004;Prave et al. 2009;Rooney et al. 2011). The base of the Dalradian sequence is dated at a maximum age of c. 800 Ma by the presence of Knoydartian deformation in the basement rocks, combined with no evidence of pre-Caledonian mineral ages in the Dalradian Supergroup (Noble et al. 1996). The Leny Limestone within the Southern Highland Group is dated at 510-515 Ma (Cowie et al. 1972) based on the presence of the rare Mid-Cambrian trilobite Pagetides (Pringle 1940;Cowie et al. 1972). There are two radiometric dates for sedimentation within the Dalradian: U -Pb zircon has constrained the Tayvallich Volcanics, which represent the top of the Argyll Group, to 601+ 4 Ma (Dempster et al. 2002), and a Re -Os whole rock date on the Ballachulish Slate at 659+ 9.6 Ma is interpreted to represent deposition (Rooney et al. 2011).

Deformation and metamorphism
The Dalradian package underwent polyphase deformation, the Grampian Event, as a result of the collision of Laurentia with an oceanic arc and the subsequent closure of the Iapetus ocean c. 480-465 Ma (Oliver 2001;Baxter et al. 2002;Stephenson et al. 2013). D1 is characterized by greenschist facies metamorphism and dominantly NE-SWtrending folds and ductile shears (Strachan et al. 2002). Peak metamorphism and maximum deformation occurred during continued over-thrusting recorded by D2 (Krabbendam et al. 1997;Crane et al. 2002), characterized by rotation and stacking of close to isoclinal, asymmetrical fold nappes (Strachan et al. 2002). Upright to SE-steeply dipping NE-trending structures dominate D3, reflecting decreasing intensity of deformation (Strachan et al. 2002). D4 deformation is associated with gently  plunging, NE-SW-trending upright folds, late crenulation and brittle structures recording weak deformation in the final stages of the Grampian Event. Numerous faults occur sub-parallel to the NE -SW structural trend of the Dalradian, and have undergone largely left-lateral strike-slip movement with a component of normal movement during transtension (Stephenson & Gould 1995). The change to a transpressional regime is recorded by D4 and features a component of right-lateral strike-slip movement on the major Caledonian faults (Strachan et al. 2002). Peak metamorphism in the Dalradian Supergroup exhibits significant along-strike variation with a general increase from greenschist facies in the SW Highlands to upper amphibolite in the NE of the central Highlands (Fettes et al. 1985;Harte 1988). In the Tyndrum area garnet-grade amphibolite-facies metamorphism was reached (Harte 1988

Post-tectonic magmatic activity
The Dalradian Supergroup hosts widespread posttectonic granitic intrusions ( Fig. 1) emplaced over a period of c. 25 Ma (Neilson et al. 2009;Conliffe et al. 2010). The intrusions are large granodioritegranite multiphase complexes and range in age from the Ballachulish Complex at 433 + 1.8 Ma (Re-Os molybdenite; Conliffe et al. 2010) to the Etive Complex at 408 + 0.4 Ma (U-Pb zircon; Neilson et al. 2009), both of which host mineralization (Neilson et al. 2009;Conliffe et al. 2010;. Although no outcrop of granite is observed within 10 km of the Tyndrum area mineralization, a gravity low extending from the Etive Complex into the Tyndrum area has been interpreted to represent the extent of a concealed granite body ( Fig. 3; Pattrick et al. 1988). The Dalradian Supergroup also hosts intrusive bodies characterized by a significant diorite component with minor appinite/peridotite/pyroxenite (Stephenson & Gould 1995). The Garabal Hill-Glen Fyne, Arrochar and Rubha Mor appinites, 40 km south of Tyndrum, have been dated at 426 + 4.2 to 428 + 9.8 Ma (U-Pb titanite and zircon; Rogers & Dunning 1991;Tanner 2012) and are interpreted by Tanner (2012) to be equivalent to the Sron Garbh diorite-appinite body near Tyndrum and widespread lamprophyre dykes and sills observed in the Beinn Udlaidh and Glen Orchy areas (Fig. 3).

Global context
Isotope stratigraphy has been extensively applied to Neoproterozoic sequences, in particular the correlation of glacial deposits (Halverson & Shields-Zhou 2011). Carbon-isotope stratigraphy is a powerful tool for correlating glacial deposits between Neoproterozoic sequences where there are limited biostratigraphic constraints but significant limestone deposition. Neoproterozoic carbonate sediments are characterized by high average d 13 C carb values and large fluctuations to extremely low values, some of which correlate with glacial episodes (Halverson et al. 2010). Strontium isotopes are a useful measure of tectonic evolution and long-term climatic change and are valuable in Neoproterozoic carbonate sequences owing to the consistent increase in 87 Sr/ 86 Sr values of seawater (Halverson et al. 2007a). Since both C-and Sr-isotopes are measured in carbonate rocks, the two records can be tied together (Halverson & Shields-Zhou 2011). Neoproterozoic sedimentary rocks are also characterized by large fluctuations in both d 34 S pyrite and d 34 S sulphate , which appear to be closely related to glacial episodes Hurtgen et al. 2002;Halverson & Hurtgen 2007), although the record is not as well constrained as for C-and Sr-isotope variations. In this paper, existing data on the variation of C-and Sr-isotopes within the Dalradian Supergroup are utilized, in addition to existing radiometric dating, to establish potential age correlations for the Dalradian Supergroup. This study uses the global Neoproterozoic S-isotope record in conjunction with these age correlations to estimate d 34 S values in sulphur-poor stratigraphy. Prave et al. (2009) compared the trend in carbonate d 13 C in the Dalradian Supergroup with the global composite d 13 C curve (Halverson et al. 2005(Halverson et al. , 2007a and correlated observed excursions with key Neoproterozoic events. The metamorphic fluids affecting Dalradian carbonates during Grampian Event orogenesis are known to be carbon-poor; therefore, the isotopic composition of carbonate units is buffered (Holness & Graham 1995;Graham et al. 1997;Thomas 2000) and carbonate d 13 C values are interpreted to represent primary values. Prave et al. (2009) tentatively correlated the Ballachulish Limestone with the c. 800 Ma Bitter Springs anomaly Halverson et al. 2007b) and the Port Askaig Tillite at the base of the Argyll Group is interpreted to represent the Sturtian glacial episodes (Fig. 2). However, it Fig. 3. Simplified geology of the study area. Key structural features and sample localities are shown. Magmatic bodies shown include lamprophyre sills and dykes, appinite bodies and diorite dykes. Geology adapted from the British Geological Survey 1:50 000 scale Bedrock Geology Crianlarich and Dalmally sheets with additional detail from Tanner & Thomas (2009) and mapping in conjunction with Scotgold Resources Ltd. Gravity anomaly after Hussein & Hipkin (1981) and interpretation in Pattrick et al. (1988). should be noted that correlation of the Ballachulish Limestone with the Bitter Springs anomaly would make the formations below too old if it is accepted that the base of the Dalradian is ,800 Ma. Prave et al. (2009) (Thomas et al. 2004;McCay et al. 2006). Thomas et al. (2004) used 87 Sr/ 86 Sr values from limestones within the Dalradian Supergroup to constrain depositional age through comparison with the global 87 Sr/ 86 Sr trend (Kuznetosov 1998, given in Shields 1999Walter et al. 2000;Melezhik et al. 2001). 87 Sr/ 86 Sr values for Grampian and Appin limestones are interpreted to be well preserved and therefore to represent primary values close to contemporaneous seawater. The level of preservation in the Argyll Group is less than that observed in the Grampian and Appin Group and 87 Sr/ 86 Sr values are inferred to be less reliable. Thomas et al. (2004) conclude that 87 Sr/ 86 Sr values observed in the lower Dalradian Supergroup (Kincraig Limestone, Grampian Group; 87 Sr/ 86 Sr ¼ 0.7069-0.7074) suggest it is not older than c. 800 Ma and may be as young as 700 Ma using Kuznetsov's (1998) Neoproterozoic seawater 87 Sr/ 86 Sr curve. Comparison of the Thomas et al. (2004) data with the more recent Halverson et al. (2010) global 87 Sr/ 86 Sr curve of high-quality Sr-isotope data supports this interpretation for the age of the Dalradian; global 87 Sr/ 86 Sr does not increase to 0.7069 until 775 Ma when it decreases again before consistently increasing after 700 Ma. This suggests that the Dalradian Supergroup is not older than c. 775 Ma and may be as young as 700 Ma, in line with the geochronometric constraints that the basement is ,800 Ma (Noble et al. 1996).

Mineralization in the Tyndrum area
The Cononish gold mineralization is hosted in the ,6 m wide Eas Anie structure, a complex of steeply dipping quartz veins, which cross-cuts Grampian and lower-Appin Group stratigraphic units ( Fig. 3; Treagus et al. 1999;Tanner 2012). The sulphide assemblage is dominated by pyrite, chalcopyrite and galena; gold, as electrum, is associated with galena in fractured pyrite (Earls et al. 1992). Previous Ar -Ar and K -Ar dating suggests that mineralization occurred at 410 + 14 Ma (Treagus et al. 1999), significantly after peak metamorphism and overlapping with granite magmatism, but with very large uncertainties. Recent geochronological work also suggests the age of mineralization is close to 410 Ma (Rice et al. 2012).
Previous work in the Tyndrum area noted a number of gold-bearing veins in addition to the Cononish and Beinn Udlaidh veins (Fig. 3). Halliday's Veins were first reported by Halliday (1962), with further work by Pattrick et al. (1988). The veins, hosted in north-trending structures, contain electrum and hessite (Ag 2 Te) with minor sylvanite ([AuAg] 2 Te 4 ) and petzite (Au 3 AgTe 2 ) (Pattrick et al. 1988). Electrum is the most common Au-Ag phase and is found as inclusions in galena and within fractures in pyrite (Pattrick et al. 1988). Mineralized veins at Coire Nan Sionnach were noted by Earls et al. (1992) with additional veins observed at Kilbridge by Scotgold. The Mother Reef, sub-parallel to the Tyndrum Fault, can be traced for 2 km and is interpreted to represent a series of segments at an oblique trend to the main structure; it is largely barren except where the projected line of Eas Anie meets the Mother Reef (Tanner 2012). The Tyndrum Lead Mine mineralization is hosted in NE-trending structures within the Tyndrum Fault zone as veins and vein breccias. The sulphide assemblage is dominated by galena and sphalerite with a notable absence of pyrite (Pattrick et al. 1983). Mineralization with a comparable assemblage is observed to cross-cut the goldbearing Eas Anie structure (Earls et al. 1992).
The range of data for the different mineralization styles as displayed in Figure 4 shows that different mineralization types have distinct S-isotope signatures. Sulphide d 34 S data from all mineralization types are generally positive with only rare values below 0‰. Gold mineralization is associated with d 34 S ratios in the range 20.6 to +9.8‰ with an average of 6.3‰ (+2.8‰ s n21 ) for pyrite.
Molybdenite mineralization has d 34 S values close to zero, suggesting it is magmatic in origin and is comparable with d 34 S data from deep-seated plutonic intrusions across the area (average d 34 S + 2.6 + 1.8‰; Lowry et al. 1995;Fig. 4).

Analytical methods
Samples were obtained from surface and underground outcrop and quarter drill core from Scotgold Resources Ltd's on-going exploration programme. Detailed mineralogical analysis was undertaken at the University of Leicester using a Hitachi S-3600N environmental scanning electron microscope, with an Oxford Instruments INCA 350 energy-dispersive X-ray analysis system. Whole rock geochemistry was undertaken at ALS laboratories, Ireland, as part of Scotgold's assay programme; only sulphur and gold concentrations are reported here. For sulphur, samples were dissolved in concentrated perchloric, nitric, hydrofluoric and hydrochloric acids. Sulphur was analysed by ICP-MS; the lower detection limit was 0.01%. Gold was measured by fire assay, fusing the sample with lead oxide, sodium carbonate, borax  Hall et al. (1987). LC, Late Caledonian; ba, barite; ga, galena; m, molybdenite; py, pyrite; po, pyrrhotite; s, sulphides. and silica then digesting the bead in dilute nitric and hydrochloric acid. The digested solution was analysed by atomic absorption; the lower detection limit was 0.01 ppm.
Sulphide minerals for S-isotope analysis were separated using micro-drilling from characterized sections. SO 2 was produced from sulphides by combustion with cuprous oxide for mass spectrometric analysis following the procedure of Robinson & Kusakabe (1975). In situ laser combustion analyses were undertaken on polished slabs from the Auch Estate, prepared using a method as described in Kelley & Fallick (1990) and Wagner et al. (2002). For both sulphur isotope methods mass spectrometric analysis was undertaken in a VG SIRA II gas mass spectrometer. Reproducibility, based on repeat analyses of internal and international laboratory standards (CP1, NBS 123 & IAEA S 3), was better than +0.3‰. All data are reported as d 34 S per mil (‰) relative to the Canyon Diablo Troilite standard (V-CDT).

Field relations of newly identified mineralization
Regional exploration by Scotgold since 2007 and work in this study have identified a number of additional occurrences of veins and clarified that mineralization is of a number of distinct types: (1) Additional gold-bearing quartz veins have been identified in the Glen Orchy area, at Kilbridge and the Auch Estate (Fig. 3). Steeply dipping gold-bearing quartz veins in the Glen Orchy area (up to 194.6 g/t Au and .200 g/t Ag from .1 kg grab samples; Scotgold Resources Ltd 2011a) trend east -west to SW -NE, reaching up to 1 m in width. The veins are polyphase and exhibit brecciation with host rock clasts included in the veins. Mineralized quartz veins at Kilbridge, trending 095 -125 and reaching 20 cm in width, are characterized by fine-grained pyrite concentrated in a zone, up to 3 cm wide, in the centre of the vein. Auch Estate gold-bearing quartz veins (up to 25.5 g/t Au and 14.3 g/t Ag over 1 m; Scotgold Resources Ltd 2011b) are up to 0.5 m in width and trend NE-SW to ENE -WSW. Coire a'Ghabalach veins ( Fig. 3) exhibit significant brecciation with clasts of early white quartz and altered host rock in the vein. Veins at Creag Shieleach are characterized by a simple sulphide assemblage and exhibit a lack of brecciation compared with other veins in the Auch Estate ( Fig. 3).
(2) Barren quartz veins are observed throughout the study area ( Fig. 3) and can be characterized into two sub-types: pyrite-free and pyritebearing. Pyrite-free veins are often massive and complex (e.g. Mother Vein at Cononish; Curtis et al. 1993;Treagus et al. 1999). Pyritebearing barren veins are characterized by a single quartz generation with some brecciation of altered host rock within the veins. In the Glen Orchy area all newly identified veins and breccia bodies cross-cut the metamorphic fabric and therefore post-date peak metamorphism. Cross-cutting relationships constrain gold-bearing quartz veins and pyrite-bearing barren veins to be younger than the molybdenite mineralization ( Fig. 5). In addition, gold-bearing quartz veins are observed to cross-cut the Sron Garbh appinite. There are currently no geochronological data for Glen Orchy to constrain age relationships and absolute timescale further and therefore this study assumes, in the absence of evidence otherwise, that all gold mineralization is of the same age. However, based on structural relationships, Tanner (2012) has interpreted gold mineralization at Cononish to be older than gold mineralization at Glen Orchy.

Ore petrography
Additional work at Cononish has clarified the paragenesis compared with previously published data (Pattrick et al. 1988;Earls et al. 1992;Curtis et al. 1993). Early gold (Au 1) is associated with hessite (Ag 2 Te 2 ) with minor early galena and is found in fractures in pyrite. Late gold (Au 2) is associated with galena and chalcopyrite with minor sphalerite (Spence-Jones 2013); no telluride mineralization is observed associated with late galena. Gold-bearing quartz veins in Glen Orchy, Beinn Udlaidh, Sron Garbh, Kilbridge and Coire Nan Sionnach (Fig. 3) are characterized by a sulphideassemblage dominated by pyrite and galena. Brecciation of adjacent host rock is observed with clasts altered to either K-feldspar or a chlorite-sericite assemblage. Early gold (Au 1) is associated with galena as inclusions in pyrite (Fig. 6a, b). A second phase of gold (Au 2) occurs with void-filling galena (Fig. 6d, e) often accompanied by sphalerite exhibiting chalcopyrite disease (Fig. 6c), and sporadic hessite. Gold-bearing veins in the Auch Estate (Fig. 3) have a comparable sulphide assemblage with the addition of arsenopyrite (Fig. 6h). Brecciation of the host rock and early white quartz is observed. Arsenopyrite forms syn-pyrite and electrum is hosted as inclusions within both arsenopyrite and pyrite and within fractures in pyrite. The veins at Coire a'Ghabalach ( Fig. 3) have late void-filling galena with hessite, sphalerite (Fig. 6f, g) and occasional Au-Ag tellurides and altaite (PbTe); veins at Creag Sheileach ( Fig. 3) have late galena, but no associated telluride mineralization.
Mineralization in breccia bodies is hosted in the quartz matrix and is associated with postbrecciation sericite -chlorite alteration of the clasts. Gold, in the form of Au -Ag tellurides, is found within pyrite with altaite (PbTe) and sporadic chalcopyrite and galena (Moore 2011).
Sulphides in lamprophyre sills are characterized by small (,5 mm wide) cubic pyrite; no other sulphides are observed. Mineralization has been identified in the Sron Garbh appinite-diorite body; the assemblage is dominated by pyrite and chalcopyrite with platinum group minerals associated with the appinitic portion (Graham 2013).

S-isotope results
Pyrite d 34 S values from gold-bearing quartz veins obtained for this study (n ¼ 46) show wide variation from 22 to +12‰ ( Fig. 7 Curtis et al. (1993). The majority of d 34 S data recorded for gold-bearing quartz veins at Glen Orchy, Sron Garbh and Beinn Udlaidh (Fig. 3) are within the range of previously published work from Cononish (Curtis et al. 1993;Figs 4 & 7a); d 34 S values below + 5‰ are only observed at   (Table 1) are comparable to data from the Tyndrum Lead Mine (Pattrick et al. 1983;Curtis et al. 1993).
A limited S-isotope dataset has been developed for the metasedimentary rocks in the study area (n ¼ 6; Fig. 7b; Table 1). The host rocks are generally sulphur-poor and measured d 34 S values are from a range of sulphide types: stratabound sedimentary exhalative (SEDEX) mineralized horizons (Ben Challum Quartzite), volcanogenic sulphides

Relationship between veins and host rock alteration
Host rock alteration of metasedimentary rocks extends up to 1 m from cross-cutting veins and is characterized by addition of sulphides (Fig. 8a, b) and alteration to a K-feldspar or chlorite-sericite assemblage. Lamprophyres cross-cut by gold-mineralized quartz veins show strong chlorite-sericite alteration over comparable distances and are only observed to contain sulphides in the alteration zone (Fig. 8c). In all veins d 34 S values are higher than in the surrounding altered host rock ( Fig. 8;

S-isotope fractionations between mineral pairs
Pyrite and molybdenite pairs ( Galena is consistently in disequilibrium (isotopic reversal with galena having higher values than the pyrite) over a temperature range of 250-500 8C or gives low temperatures (Table 3) with chalcopyrite and pyrite (Kajiwara & Krouse 1971;Li & Lui 2006). The S-isotope equilibrium fractionations between galena and chalcopyrite at Cononish give low temperatures that might reflect cooling during the development of the paragenesis. While the fractionation of S-isotope between arsenopyrite and pyrite is not constrained, the similar d 34 S values observed at Coire a'Ghabalach (Table 3) suggest that equilibrium was reached (Nesbitt 1988).
Overall, the S-isotope fractionations are consistent with the petrographic evidence of early pyrite and later chalcopyrite and galena.

Discussion
The wide range in S-isotope values from mineralization in the Tyndrum area is inconsistent with a single source for the sulphur in these occurrences. Curtis et al. (1993)

Sedimentary S-sources
The extensive dataset of d 34 S sulphide for Dalradian metasedimentary rocks (Appin-Southern Highland Groups) records large variations through the sequence (Figs 4 & 9;Willan & Coleman 1983;Scott et al. 1987Scott et al. , 1991Hall et al. 1988Hall et al. , 1994aLowry 1991), ranging from as low as 215‰ in the Ardrishaig Phyllite to as high as +42‰ in Bonahaven Dolomite, consistent with the large variations in d 34 S sulphide in the global Neoproterozoic record (d 34 S sulphide ¼ 230 to +50‰; Halverson et al. 2010). Sulphates have a limited occurrence in the Dalradian Supergroup and are mostly confined to the Aberfeldy deposits (Willan & Coleman 1983;Hall et al. 1991;Moles et al. 2014).
Most of the sulphide data come from the Argyll Group, with only data from the Ballachulish Slate Formation deeper in the sequence.
Stratabound sulphides in the Dalradian sequence are of three types: (a) sedimentary diagenetic sulphides which have usually undergone some recrystallization during regional metamorphism; (b) sulphide-rich but un-mineralized horizons thought to be of volcanogenic origin; and (c) syn-sedimentary mineralized hydrothermal exhalative SEDEX horizons. All three types of sulphide could be potential sources for the sulphide in the Tyndrum mineralization and all are referred to as 'sedimentary-sourced'. However, only sedimentary diagenetic sulphides that have been metamorphosed as a closed system are expected to correlate with the global sedimentary sulphide S-isotope record.
Sulphide d 34 S values within the un-mineralized Ballachulish Slate (Appin Group; Fig. 2), Bonahaven Dolomite, Ben Eagach Schist and Ardrishaig Phyllite (d 34 S ¼ 215 to +42‰; Willan & Coleman 1983;Hall et al. 1987Hall et al. , 1994aLowry 1991;Moles et al. 2014) Hall et al. 1994a) is sulphur-poor in the Tyndrum area and Lowry et al. (1995) noted a lack of contamination by external sulphur in porphyries hosted in the Ardrishaig Phyllite, indicating that it is unlikely to represent a significant sulphur source. Sulphur isotopic signatures from sulphides in the Bonahaven Dolomite are not typical of diagenetic pyrite (d 34 S ¼ +29 to +42‰; Willan & Coleman 1983;Hall et al. 1994b;Moles et al. 2014) and are probably the result of closedsystem reduction of evaporite sulphate (Hall et al. 1994b).
Sulphides within stratabound horizons of likely volcanogenic origin in the Ben Lawers and Ben Lui Schists, and in the Tayvallich Volcanics, have d 34 S sulphide values between 24 and +8‰ (Willan & Coleman 1983;Scott et al. 1987;Lowry 1991), with a single pyrite d 34 S value from this study from the Ben Lawers Schist (23.2‰; Fig. 7b; Table 2) being comparable. These sulphides are interpreted to be related to the appearance of mafic lavas and sills (now amphibolites) in the Dalradian sequence (Scott et al. 1991;Stephenson & Gould 1995).
All of the vein samples discussed in this paper are hosted in Grampian and lower-Appin Group units (Fig. 2), in particular the Meall Garbh Psammite, Beinn Udlaidh Quartzite and Leven Schists. The units are sulphur-poor with bulk sulphur less than 0.2%, except where units are cross-cut by mineralized veins (Fig. 8; Table 2). Consistent with this, Laouar (1987) noted that no sulphide mineralization is observed in granites hosted within Grampian Group units, suggesting that the host rocks of the mineralization may not have been good sulphur sources. There are no published d 34 S sulphide or d 34 Ssulphate data for these units.
To estimate the d 34 S values that might occur in trace sulphides in the Grampian and lower-Appin Group stratigraphic units the global composite sulphide S-isotope curve of Halverson et al. (2010) is superimposed onto the Dalradian stratigraphy using two age correlations suggested by previous workers (Figs 2 & 9). In both correlations the top of the Argyll Group (Tayvallich Volcanics) is fixed at 601 + 4 Ma (Dempster et al. 2002). The potential correlations are: (1) The base of the Dalradian is c. 800 Ma (Noble et al. 1996) and, following Prave et al. (2009), the Port Askaig Tillite is correlated with the Sturtian glacial episodes and the mid-Easdale Subgroup is correlated to Marinoan glacial episodes.
(2) The base of the Dalradian is c. 700 Ma based on 87 Sr/ 86 Sr variation (Thomas et al. 2004;Stephenson et al. 2013); this is consistent with the Re-Os date for the Ballachulish Slate (Rooney et al. 2011), which in turn suggests that the Port Askaig Tillite represents the Marinoan glacial episodes.
Comparison of the measured d 34 S for the Dalradian sequence with the global curve shows a good fit for correlation 1, the only outlying points being the mineralized horizons of the Ben Eagach Schist, but given that this is a SEDEX horizon with a hydrothermal component to the sulphides (Willan & Coleman 1983;Moles et al. 2014), it would not be expected to fit the global curve. The good fit of the Bonahaven Dolomite data to correlation 1 may be fortuitous as this is interpreted as being a local signature owing to closed-system reduction of evaporites, not a global signal. For correlation 2 the fit is rather poorer, but this correlation is still broadly consistent with the mostly positive d 34 S values measured through the Argyll Group. Further research is clearly required to refine correlations and distinguish local from global signals; nevertheless all existing data and both fits of the global curve indicate a significant amount of sedimentary and hydrothermal sulphide with positive d 34 S in the Easdale Subgroup stratigraphic units. Deeper in the stratigraphy, both correlations give good fits to the Ballachulish Slate data. Below this the correlations are used to provide predictions of d 34 S in the Grampian and lower-Appin host rocks. Using correlation 1 (Figs 2 & 9) the pre-Sturtian d 34 S record is limited but suggests Grampian and lower-Appin Group host rocks may have an average d 34 S value less than 0‰, although some units could be enriched in 34 S by up to 10‰. Using correlation 2 (Figs 2 & 9), the host rocks are expected to all have d 34 S ≥0‰, with values as high as +40‰ possible. This would not be consistent with observations of veins having higher d 34 S than their host rocks (Fig. 8). In either case, the lack of sulphur in the local Grampian and lower-Appin Group metasedimentary rocks (Table 2) suggests that they are unlikely sources of sulphur, although they could still potentially contribute 34 S-enriched sulphur if correlation 2 was correct. The Islay Subgroup is not present in the Tyndrum area owing to the Boundary Slide and thus could not be the source of 34 S-enriched sulphide in the Tyndrum mineralization. However, the Easdale Subgroup (Fig. 2) has varied d 34 S values (d 34 S ¼ 215 to +28‰; Figs 4 & 9), but is largely enriched in 34 S, in particular within the Ben Challum Quartzite and Ben Eagach Schist SEDEX horizons. In addition, these units are sulphur-rich (bulk S ¼ 0.39% to 2.32%; Table 2), suggesting that lithologies in this subgroup have the potential to act as a significant source of sulphur. Thus, it is proposed that the only feasible source for the sedimentary sulphur component in  (Fig. 10). It is not possible to generate a d 34 S of +12‰ by mixing involving sulphur from any other part of the stratigraphy in the area. Thus the highest d 34 S values measured in the veins place very strong constraints on the sources of sulphur and the hydrothermal pathways in the mineralizing system, with a magmatic  Fig. 9. The global composite sulphide S-isotope curve of Halverson et al. (2010) superimposed onto the Dalradian stratigraphy using the two correlations detailed in Figure 2. The global data (grey dots; grey zone ¼ line of best fit) are interpolated between tie points detailed in the text (black square and diamond; grey square and diamond) by simply scaling to stratigraphic thickness (Stephenson & Gould 1995;Stephenson et al. 2013). Solid black lines show the existing d 34 S sulphide data for the Dalradian sedimentary succession (Fig. 4) +8‰; Fig. 10). The high d 34 S values for the molybdenite mineralization compared with data from the regional magmatic complexes (Lowry et al. 1995;Conliffe et al. 2009) suggest a larger component of sedimentary-derived sulphur in the molybdenite mineralization in the Glen Orchy area. Breccia bodies or molybdenite mineralization with d 34 S values of +8‰ could have 18-43% sedimentary sulphur, sourced from the syn-sedimentary SEDEX Ben Eagach horizon or 37 -100% sedimentary sulphur sourced from the Ben Challum horizon. If it is assumed that the sulphur in all the mineralization was derived from the same sedimentary-sourced end-member, sulphur isotope values indicate that the molybdenite mineralization and gold-bearing mineralized breccia pipes have a larger magmatic component than gold-bearing quartz veins.

Implications of structure
The S-isotope data suggest that a significant component of the sulphur in the Tyndrum veins is sourced from the Easdale Subgroup, higher in the stratigraphy. The Eas Anie structure, host to Cononish gold mineralization (Fig. 3), is not observed to cut the Boundary Slide at the current topography but is postulated to have extended across the Slide at emplacement, approximately 200 m above the mine portal (Tanner 2012). Thus here it might be possible that sulphides from the Easdale Subgroup could have been dissolved and re-precipitated in the mineralization. However, at Glen Orchy and Beinn Udlaidh, the Easdale Subgroup and higher stratigraphic units are estimated to be approximately 4 km above at the time of mineralization. If fluids were transported downwards from an enriched d 34 S source in Easdale Subgroup rocks, it might be expected that the Cononish gold mineralization would show a greater signature of this, but d 34 S values at Glen Orchy are comparable to Cononish ( Fig. 7a; Table 1). In addition, the presence of hydraulic breccia bodies and quartz-breccia veins formed from supra-lithostatic fluids (Tanner 2012) suggests a fluid pressure gradient that would preclude fluids flowing downwards during breccia formation. Furthermore, it is difficult to envisage a thermochemical gradient that could transport sulphide 4 km downwards. These considerations suggest that it was unlikely that the sulphur was derived from the overlying Easdale units.
However, consideration of the fold structures of the Tyndrum area (Tanner & Thomas 2009) has implications for possible fluid-flow pathways since all interpretations suggest that Easdale Subgroup units are likely to be repeated at depth owing to the major recumbent fold of the Beinn Chuirn Anticline (Fig. 11). The details of the likely depth at which repetition might occur depend upon the interpretation of the Boundary Slide, which represents a section of missing stratigraphy in the Tyndrum area (Figs 2 & 3). Two interpretations are proposed (Fig. 11).
(1) The Boundary Slide is a tectonic slide (Bailey 1922, Hutton 1979) that formed syn-to post-D2 (Roberts & Treagus 1979). The stratigraphy was folded around the Beinn Chuirn Anticline during D2, then removed by movement along the slide (Fig. 11a). The slide is inferred to continue to depth; the Ballachulish, Blair Atholl and Islay Subgroups and overlying stratigraphy (Fig. 2) are all interpreted to be represented on the inverted limb of the Beinn Chuirn anticline.
(2) A more recent interpretation (Tanner & Thomas 2009) is that the Boundary Slide is a pre-tectonic disconformity, and owing to D2 folding, stratigraphic units above the slide are expected to be represented at depth beneath the Glen Orchy dome ( Fig. 11b; Tanner & Thomas 2009), with the Easdale Subgroup present, but at shallower depth than in Figure 11a.
Importantly, both interpretations imply the potential presence of the Easdale Subgroup, enriched in 34 S, at depth and would allow the sedimentary S-isotope signature associated with mineralization to instead be sourced from the overturned limb of the Beinn Chuirn Anticline beneath the Grampian and lower-Appin Group rocks observed at surface. This seems a more realistic scenario and allows hotter fluids from depth (perhaps expelled by latetectonic magmatism) to carry sulphur (and metals) upwards and along major structures into the current host rocks where precipitation is most likely driven by pressure reduction from lithostatic to hydrostatic and wall rock sulphidation. In this scenario gold-bearing quartz veins in Glen Orchy and Beinn Udlaidh would be closer to the postulated source rocks, suggesting that they should contain a greater component of sedimentary-sourced sulphur than Cononish. However, the similar d 34 S values in both areas do not support this. It is proposed, that the Tyndrum fault is a key fluid pathway in Glen Cononish, allowing gold mineralization in the Eas Anie structure to source significant sedimentary sulphur (as well as magmatic sulphur) from depth despite being further above the postulated sedimentary sulphur source rocks.

Metal source rocks
The S-isotope data cannot confirm, nor exclude, a magmatic input to the mineralization, although the spread of data down to approximately 0‰ could be interpreted as supporting evidence and, together with the likely age that correlates with the intrusion of the granites, suggests it is probable. Thus a magmatic source for the gold and other metals is possible. However, it is proposed that a significant proportion of the sulphur in the mineralization must have originated from Easdale Subgroup lithologies at depth, in particular the SEDEX Ben Challum or Ben Eagach horizons. While there is no requirement that metals and sulphur are derived from the same source, and this is frequently not the case in hydrothermal gold systems (Goldfarb et al. 1991(Goldfarb et al. , 2001, the S-isotope data demonstrate that mineralizing fluids originated from, or passed through, these units and hence it is reasonable to consider whether they could also be potential metal sources. This suggestion has some credence since, along with base metals, gold is known to be concentrated in shale-hosted SEDEX mineralization (e.g. Cooke et al. 2000;Alchin & Moore 2005). Furthermore, Willan (1996) demonstrated that the Ben Eagach Schist is regionally enriched by hydrothermal activity in Bi, Sb, As, Mo, Ni and Ba, while in the section between the Tyndrum Fault and the Ericht -Laidon Fault to the NW containing the study area, this unit has elevated Mo, Sb and Bi and isolated occurrences of strongly anomalous Cu, Zn and Pb. Unfortunately gold was not analysed in this study and no gold grains have been observed to date in the Aberfeldy deposits (N. Moles pers. comm.), but nevertheless it seems that SEDEX horizons in the Ben Eagach Schist represent feasible sources for at least some or all of the base metals in the veins, in particular lead, which is notably enriched as abundant galena.
Alternatively, gold and other metals could have been pre-concentrated from the sedimentaryvolcanic pile during formation of volcanogenic exhalative horizons, such as in the Ben Lawers Schist, which could subsequently produce gold-rich fluids (e.g. Hodgson et al. 1993;Mernagh & Bierlien 2008). Some support for this comes from the observation by Moles (1985) of a gold inclusion within chalcocite -bornite in a sample of the Ben Lawers Schist.
A third option could be that the gold and other metals are sourced from carbonaceous pyritic metasedimentary rocks in the Dalradian, such as parts of the Ben Eagach Schist, since these may also concentrate gold along with other metals Fig. 11. Cross-sections through the Tyndrum area. Sulphur is interpreted to be sourced from stratigraphic units enriched in 34 S at depth, dependent upon the interpretation of the Boundary Slide: (a) interpretation after Roberts & Treagus (1979), the Boundary Slide is a tectonic slide formed syn-to post-D2; (b) interpretation after recent mapping at Cononish (Tanner & Thomas 2009). Diagram after Tanner & Thomas (2009). The Tyndrum Fault and many of the major structures in the area run sub-parallel to the strike of the section and much of the fluid flow is therefore likely to be parallel to the plane of the section. Vertical scale approximate only.
including Ag, Zn, Mo and Cu (Large et al. 2011), as postulated for Dalradian rocks by Plant et al. (1997). Pitcairn et al. (2006) show that even quartzofeldspathic turbiditic metasedimentary rocks, which may not have anomalous metal contents, have the potential to release elements including Au, Ag, As, Sb, Hg, Mo and W during metamorphism. Tomkins (2012) argues that pyritic organic-rich sedimentary rocks deposited after the second Great Oxidation Event (635 -510 Ma) have the potential to be a better source of gold (and molybdenum) than sedimentary rocks deposited earlier, owing to the increase in gold solubility in a more oxidized ocean. In both correlations (Fig. 9) sections of the Dalradian stratigraphy are deposited after 635 Ma (Marinoan) and therefore may have been enriched in gold and molybdenum at deposition.
From the discussion above it is clear that carbonaceous pyritic metasedimentary rocks could have been a potential source for the molybdenum in the earliest veins in the area. Compared with the later gold-mineralization, the generally lower d 34 S value of the molybdenite mineralization indicates either that a larger proportion of magmatic sulphur, or that the sulphur is from a distinct metasedimentary source. For example, average d 34 S for diagenetic sulphides in the Ben Eagach Schist is +5.6‰ (Willan & Coleman 1983;Moles et al. 2014), very close to the average value measured for the sulphides in the molybdenite mineralization (Fig. 7b). Thus it is possible that both the sulphur and the molybdenum in the early molybdenite mineralization could be wholly derived from parts of the Ben Eagach Schist. A potentially different source of sulphur and metals (and by inference fluid pathway) is conceivable for this mineralization given the distinct timing and nature of this mineralization compared with the later gold mineralization.
Overall there are a number of reasons why the Easdale Subgroup rocks represent the likely source of sulphur and potentially gold and other associated metals in the gold veins and thus place important constraints on the fluid pathways to the mineralized systems. Further work is required to confirm the source of gold and other metals but the presence of Easdale Subgroup units at depth may be an important criterion determining the prospectivity of the Dalradian Supergroup as a whole.

Conclusions
This work provides clear evidence that gold and other metal mineralization hosted in the Tyndrum area, Scotland, has a mixed magmatic and sedimentary source of sulphur. The identification of the individual units in the Dalradian Supergroup that have the potential to be the main source of sedimentary sulphur has been possible through thorough sampling of mineralization and host rock Dalradian units, and correlation with the global Neoproterozoic S-isotope record for the units that are poorly exposed or there is a lack of data for. Key conclusions can be summarized as: ( 1)  Lowry and D. Craw are thanked for constructive discussion on the evolution of the Dalradian Supergroup and comments on an earlier version of the manuscript. The manuscript benefitted from the suggestions for improvement from N. Moles and M. Smith, who are thanked for their positive and constructive reviews, and P. Lusty is thanked for his editorial assistance.