Unit MBG2 mainly consists of porphyritic monzogranite that locally grades to syenogranite. The monzogranite is composed of plagioclase, K-feldspar, quartz, hornblende, and biotite in order of decreasing abundance, with minor pyroxene. The phenocryst assemblage is plagioclase, K-feldspar, and quartz (~ 40% total abundance), although a few highly altered hornblende (to chlorite) phenocrysts are evident, in which pyroxene cores were noted. Phenocrystic plagioclase forms euhedral to subhedral tabular crystals, ranging from 1 to 12 mm long, with rectangular cross sections and polysynthetic albite twinning. Glomeroporphyritic plagioclase clusters are common. Plagioclase grains are highly altered, giving them a mottled appearance under both plane and cross-polarized light. The alteration products are mainly saussurite (i.e., fine-grained aggregate of epidote group minerals, albite, prehnite, sericite, and possibly clays). K-feldspar (1– 12 mm in grain size) occurs as subhedral, most commonly rectangular, crystals, but rarely euhedral crystals are observed. Coarse patch perthite is common. K-feldspar is also very altered to sericite and clay minerals. Micrographic texture, in which quartz is intergrown in K-feldspar crystals, is present. The quartz forms small, more or less angular grains that are commonly optically continuous over the entire K-feldspar crystal. Phenocrystic quartz generally is less than 4 mm in size; it commonly occurs as subhedral crystals with partly square to hexagonal cross sections, or as anhedral, equant crystals. Groundmass (generally less than 0.5 mm in grain size) constitutes about 60% of the rock and has a similar mineralogy to the phenocryst assemblage. Accessory minerals are apatite, titanite, allanite, zircon, and opaque minerals (mainly magnetite).

The less voluminous MBG1 phase occurs on the margin or border of the MBG2 unit, and consists of porphyritic mon-zogranite (Fig. 3d) and minor equigranular to subporphyritic (containing plagioclase phenocrysts) fine-grained monzogran-ite. The porphyritic monzogranite contains two populations of subhedral to euhedral feldspar phenocrysts (1– 12 mm) with plagioclase generally more abundant than K-feldspar, constituting 30– 35% of the rock. Quartz phenocrysts (1– 4 mm) are minor (< 2%). The groundmass (generally less than 0.1 mm) includes minerals similar to the phenocrysts. Accessory minerals include apatite, zircon, titanite, and magnetite. Most magnetite in the MBG is probably secondary because anhedral magnetite crystals are intimately associated with chlorite that formed by hydrothermal alteration of pyroxene, hornblende, and biotite.

A zone of greisenized quartz breccia and veins about 1500 m long and several hundred meters wide cuts the MBG and its wall-rocks (Seelys Formation ash flow tuffs). This steeply dipping NW-trending zone (Fig. 1) is notably mineralized and contains significant gold (200 ppb up to 1 g/t Au) and silver (up to 27 g/t Ag) (see Johnson 2003). Gold is typically associated with arsenopyrite and pyrite, although the sulphide abundance is low (< 1%) and unevenly distributed in the quartz breccias and veins. These features are common in intrusion-related gold systems elsewhere (Lang et al. 2000; Lang and Baker 2001).

Units GRII and GRIII of the MPG are similar in mineralogy, and consist mainly of quartz, K-feldspar, plagioclase, and biotite. They may represent two pulses of magma injection from an inward-cooling magma chamber, as indicated by intrusive contacts (Fig. 2b). Unit GRII is mainly composed of porphyritic monzogranite, with local aplitic and seriate textural varieties. Phenocrysts (30– 40%) consist of subhedral to anhedral quartz (0.5 to 1.5 mm), anhedral K-feldspar (0.5 to 3 mm), subhedral to anhedral plagioclase (0.5 to 1 mm), and biotite (0.5 to 2 mm). The groundmass contains minerals similar to the phenocrysts, but the grain size is commonly less than 0.3 mm.

Unit GRIII mainly consists of fine- to medium-grained equigranular monzogranite, although locally porphyritic and pegmatitc varieties are present. Subhedral to anhedral quartz (30– 40%) is dominant, but euhedral quartz is also observed; grain size ranges from 0.5 to 1.5 mm. K-feldspar usually occurs as anhedral to subhedral crystals from 0.5 to 3.5 mm, which constitute 40– 50% of the rock. Plagioclase (20 to 25%) forms subhedral crystals, 1 to 1.5 mm long. Subhedral biotite flakes, 1 to 3 mm in size, contain pleochroic halos caused by radiation damage induced by zircon inclusions, form 1– 3%. Opaque minerals include ilmenite, pyrite, and pyrrhotite. Magnetite is associated with chlorite, suggesting that it formed by hydro-thermal alteration.

Accessory minerals in both GRII and GRIII are monazite, zircon, ilmenite, columbite, and Li mica. Topaz and fluorite in the GRII and GRIII are both interpreted as products of hydro-thermal alteration based on textural relations. However, a few topaz crystals that occur as inclusions in euhedral quartz in GRIII and the presence of primary topaz in an aplitic contact phase of GRII, led Sinclair and Kooiman (1990) to conclude that the MPG can be classified as topaz granite (Manning 1988; Manning and Hill 1990). Magmatic topaz also occurs in leucogranite in the vicinity of Pleasant Ridge and Kedron, as reported by Taylor et al. (1985) and Taylor (1992), who termed the Kedron granite the Bonny River granite. These granite bodies are similar in composition and age to the MPG.

Parts of the aplitic to porphyritic phase of the GRII contain miarolitic cavities, comb quartz layers, and unidirectional solidification textures (UST), suggesting fluid saturation and/ or undercooling of the parental magma (Kirkham and Sinclair 1988). Micrographic intergrowths of quartz and K-feldspar are common in the granite associated with comb quartz layers, both of which reflect a significantly undercooled magma (Fenn 1986; Lentz and Fowler 1992). Detailed drilling and mapping of underground developments indicate that major Sn-Zn-Pb-Cu-In mineralization is closely associated with unit GRII (Kooiman et al. 1986; Sinclair et al. 1988; Sinclair and Kooiman 1990; McCutcheon et al. 2001; Fig. 2b).

In the North Zone at Mount Pleasant, unit GRIII clearly intrudes GRI and GRII (Fig. 2b), which is indicated by sharp contacts that in many places are marked by thin layers of UST (mainly K-feldspar) in the GRIII. Miarolitic cavities (1 mm to 2 cm in size) are locally abundant, in which very fine-grained ser-icite, chlorite, fluorite, and carbonate are present, but arseno-pyrite, molybdenite, and other metallic minerals are relatively rare (Sinclair et al. 1988). Despite well preserved fluid saturation textures in GRIII and evidence of hydrothermal activity, only minor tin mineralization is known to be associated with it (Kooiman et al. 1986; Sinclair et al. 1988; Sinclair and Kooiman 1990; McCutcheon et al. 2001). However, further exploration efforts are needed to confirm this conclusion.

Fluid inclusion studies showed that magmatic fluids were predominant in the early stage of mineralization, but more meteoric fluids were involved in late-stage ore-forming processes (Davis and Williams-Jones 1985; Samson 1990). High salinity (>30 wt.% NaCl equivalent) and high temperature (350– 490°C) fluid is interpreted to have been derived from resurgent boiling, possibly related to magmatic fluids exsolved from the cooling magmas; dilution of this early fluid by con-vecting meteoric water resulted in low- to moderate-salinity fluids that dominated the inclusion population (Davis and Williams-Jones 1985). Low temperature (180– 250°C) and high-salinity fluid may represent late residual fluid from a change in the pressure regime from dominantly lithostatic to hydrostatic conditions (Davis and Williams-Jones 1985). Clearly, hydrothermal breccias, related pervasive alteration and mineralization (silicification, chloritization, and greiseni-zation) in GRI and GRII (Kooiman et al. 1986) reflect pressure fluctuations of fluids associated with the granitic magmas in the Mount Pleasant deposits. These processes are similar to those that operated in the adjacent True Hill biotite porphy-ritic granite intrusions (Lentz et al. 1988; Lentz and McAllister 1990; Lentz 1994; Lentz and Gregoire, 1995), suggesting a comparable genesis for all Sn-W-Mo-Bi mineral deposits in the region.

Ten least-altered granite samples collected from seven drill cores at the Mount Pleasant mine and 17 samples from outcrops of the McDougall Brook Granitic Suite (see Figs. 1 and 2 for sample locations) were analyzed for major elements and selected trace elements, including rare earth elements (REE). The methods used in this study include inductively coupled plasma - optical emission spectrometry (ICP-OES), inductively coupled plasma - mass spectrometry (ICP-MS), instrumental neutron activation analysis (INAA), and X-Ray fluorescence spectrometry (XRF).

The ten samples from the MPG were systematically analyzed with ICP-OES (SiO2, TiO2, Al2O3, Fe2O3(total), MnO, CaO, MgO, Na2O, K2O, P2O5, S, As, Ba, Cd, Co, Cr, Cu, Ni, Sc, Sr, V, and Zn) and ICP-MS (Rb, Y, Zr, Nb, Cs, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, Hf, Ta, Pb, Th, U, Bi, and Sn) at the GSC-Quebec (Geological Survey of Canada-Quebec), with INAA (Au, Ag, As, Ba, Br, Ca, Co, Cr, Cs, Fe, Hf, Hg, Ir, Mo, Na, Ni, Rb, Sb, Sc, Se, Sn, Sr, Ta, Th, U, W, Zn, La, Ce, Nd, Sm, Eu, Tb, Yb, and Lu) at the Activation Labs, Ancaster, Ontario, and with XRF (SiO₂, TiO₂, Al₂O₃, Fe₂O₃(total), MnO, CaO, MgO, Na₂O, K₂O, P₂O₅, Rb, Ba, Sr, Ga, Nb, Zr, Y, Th, U, Ce, Cr, Ni, Sc, V, Cu, Pb, Zn, Cl, S, and As) on pressed pellets (Longerich 1995) at the Memorial University of Newfoundland, St. John's. The seventeen samples from the MBG were analyzed with INAA for selected trace elements including REE, also at the Activation Labs; major element data for these samples were taken from the original dataset of McCutcheon (1990a), which were determined by XRF. Certified standard samples SY2 and JG1 (Govindaraju 1994) were used to monitor data quality. Analytical precision of major elements is better than 3 % relative standard deviation (RSD), except for SiO2 (5% RSD); and for most trace elements is better than 10% RSD. Elements for which the analyzed values were beyond 5% of the recommended values in the standards were not used. The analytical results that satisfied the requirements of precision and accuracy are shown in Table 1, and are considered to best represent the chemical compositions of the samples.

Although the MBG has low silica content (SiO2 < 70 wt. %) relative to the MPG (SiO2 > 74 wt. %), both are weakly peralu-minous, with four samples from the MBG being metaluminous (Table 1). Aluminum saturation indices (A/CNK = Al₂O₃/ (CaO+Na2O+K2O), molar) range from 0.91 to 1.28 (Table 1). Harker diagrams (Figs. 5a, e and i) show that TiO2, MgO, and P2O5 decrease with increasing silica, whereas Al2O3, CaO, K2O, and Na2O show an erratic distribution; Fe2O3 and MnO (Figs. 5c and d) are somewhat less erratic. Samples from the MPG are characterized by high SiO2, (Fe2O3+FeO)/MgO ratio, and alkalis with K2O/Na2O > 1, and low CaO, P2O5, TiO2, and MgO (Table 1; Figs. 5c, e, g, and h), thus sharing many similarities with topaz rhyolite, ongonite, and topaz granite (Christiansen et al. 1983, 1984; Christiansen et al. 1986; Kontak 1990; Taylor 1992; Dostal and Chatterjee 1995; Zhu et al. 2001). The MBG and MPG seem to have two distinctive groups, but each group displays a similar variation trend (Figs. 5a-i) except for those on the plots of A/CNK versus K2O and A/CNK versus CaO (Figs. 5 j and k). These trends may indicate that magmatic evolution was controlled by a similar process such as fractional crystallization, but that minerals involved in the process may have been different in the MBG and MPG.

On the plot of SiO₂ versus K₂O (Le Maitre et al. 1989), samples from the MBG plot in the fields of high-K and sho-shonite series, whereas those from the MPG fall into the field of medium K (Fig. 6a). Using the MALI (modified alkali-lime index) against SiO₂ diagram of Frost et al. (2001), the difference in alkalinity between the MBG and MPG is clear: the former belongs to the alkalic-calcic transitional to the alkalic suite, whereas the latter is calcic-alkalic (Fig. 6b). Interestingly, all samples from the MBG are magnesian granites, defining a trend toward higher FeO^total/(FeO^total + MgO) ratios with increasing silica, but the MPG samples range from magnesian to ferroan granite (Fig. 6c), according to criteria established by Frost et al. (2001).

In terms of the Gottini Index ô (Gottini 1968) and the Rittmann Serial Index ó Rittmann 1957), the lavas of active volcanoes can be divided into two distinctive groups that reflect tectonic setting, i.e. anorogenic and orogenic, as shown in Fig. 6d. The formulations for the parameters ô (Gottini 1968) and ó (Rittmann 1957) are as follows:

-> See the list of figures

-> See the list of figures

This discrimination diagram was originally proposed by Rittmann (1973), based on plotting of Log ó versus Log ô for 1300 lava samples from active volcanoes worldwide. From mafic through intermediate to felsic igneous rocks, the ó value is a very convenient petrochemical indicator, compared to the Peacock index (Peacock 1931), to portray the alkalinity of igneous rocks. Each whole-rock analysis has a defined alkalinity value, if the Rittmann serial index ó is appropriately applied (X. Yang, unpublished data).

Most of the samples from the MBG and MPG suites fall into field B, except for six samples from the MBG (three are located in field C, and the other three fall on the boundary between the fields B and C) (Fig. 6d). Samples from the MPG have ó values of 1.5 to 2.3 (Table 1), i.e. belonging to the calc-alkaline series, except for one sample (PRL95-2-2019) that has ó value of 0.9, which may be attributed to hydrothermal alteration (i.e. chloritization). Their ô values are extremely high, ranging from 182 to 449, consistent with the character of highly differentiated rocks.

Most of the MBG samples are calc-alkaline (ó ranging from 1.8 to 3.3; Table 1), but six samples are weakly alkaline (ó ranging from 3.5 to 4.3), consistent with their high alkali contents (Fig. 6a and b). Their ô values (13 to 68; Table 1) are much lower than those of the MPG samples, indicating that they are less evolved. Notably, an apparent fractionation trend is exhibited on Figure 6c, suggesting that the MPG could be linked through differentiation of the MBG.

A few samples (30A2, 33A2, 193A; Table 1) from the MBG1 phase of the MBG show very high SiO2 (up to 77.5 wt. %) and K₂O (up to 8.41 wt.%) contents, but relatively low Na₂O (as low as 0.84 wt.%) and Al₂O₃ (as low as 11.6 wt.%) contents. These features likely reflect hydrothermal alteration (e.g., silicification, sericitization, and chloritization), as indicated by petrographic observation, and are not the result of mag-matic fractionation. Similarly, two samples (PRL95-2-2019 and NMP89-1-1759) from the GRII phase of MPG contain moderately low SiO2 and Na2O (Table 1), which also is attributed to hydrothermal alteration (e.g., chloritization). Consequently, interpretations based on alkalis must be considered with caution. As shown in Kontak and Clark (2002), unaltered and altered granitoids are distinguishable on plots of A/CNK versus K2O and A/CNK versus CaO. However, except for these partly altered samples mentioned above, no apparent alteration trends are evident on these plots (Figs. 5j & k).

Trace elements, including REE, in the MBG are distinct from those in the MPG (Table 1). The MBG samples have relatively low incompatible element contents, enriched chon-drite-normalized REE patterns [(La/Yb)_N = 3.5– 15.7] with more pronounced negative Nb anomalies, but small negative Eu anomalies (Eu/Eu* = 0.43 – 1.09), and more pronounced negative Ti, Sr, and Ba anomalies (Figs. 7 and 8). The flat "bird-wing shape" REE patterns (Fig. 7b) of the MPG samples, which are characterized by large negative Eu anomalies (Eu/Eu* = 0.014 – 0.002) and low (La/Yb)_N (ranging from 1.4 to 2.0) ratios, reflect the highly evolved nature of the MPG. These data are consistent with the major element data (e.g., high SiO₂ and alkalis, very low CaO, and high ô values). From the MBG to MPG, negative Eu anomalies become increasingly pronounced, with fractionation based on Th enrichment. The MPG samples plot in the A-type granite field, but the MBG samples fall into the I- and S-type field (Fig. 9). If they are associated with fractionation, feldspar must be one of major phases in the crystallizing assemblage (Eby 1992; Christiansen and Keith 1996). It has to be pointed out that highly evolved granites such as the MPG that display some of the geochemi-cal characteristics of A-type intrusions are not necessarily true A-type granites (Whalen et al. 1987; Eby 1990), as no fayalite or sodic hedenbergite that typify "classic" A-type granite occurs in the MPG. Furthermore, petrographic evidence shows that the MPG was saturated with water during crystallization, which also contrasts with the relatively "dry" nature of anorogenic A-type granitoids (Collins et al. 1982; Whalen et al. 1987; Eby 1990).

Relative to I-, S-, A-, and M-type granites (Collins et al. 1982; White and Chappell 1983; Whalen et al. 1987; Eby 1990, 1992), both the MBG and MPG are enriched in F (≤ 0.99 wt.%), Li (≤ 610 ppm) (see Sinclair and Kooiman 1990), Rb (≤ 1210 ppm), Cs (≤ 28 ppm), U (43 ppm), Th (≤ 50 ppm), Nb (≤ 107 ppm), Zr (≤ 580 ppm), and heavy REE (e.g., Yb ≤ 36) (Table 1), which is typical of crustal A-type felsic magmas (Fig. 10), topaz rhyolite (Christiansen et al. 1983, 1984, 1986), ongonite, and topaz granite (Kontak 1990; Taylor 1992; Zhu et al. 2001). Both the MPG and MBG suites are temporally and spatially related, i.e. part of the Mount Pleasant Caldera, and both intrusive suites formed in a within-plate tectonic setting (McCutcheon et al. 1997), but the petrogenetic relationship between them is less clear. It is noteworthy that Nb and Zr behave differently in the MBG and MPG suites. From early to late in time sequence (GRII → GRIII), Nb decreases with decreasing Zr in the MPG, whereas Nb increases with increasing Zr from MBG1 → MBG2 in the MBG (Fig. 5l).

Published isotopic data suggest that the MBG and MPG have a similar source, as they have the same åNd(t) (-0.2) and similar ä¹⁸O (7.1 and 8.2 ~ 8.6‰) values (see Taylor 1992; Whalen et al. 1996). The slightly lower ä¹⁸O value of the MBG may be attributed to meteoric water interaction (Whalen et al. 1996). These characteristics suggest that the MBG and MPG either were derived from the same parent magma or from different batches of partial melting of the same protolith. The initial ⁸⁷Sr/⁸⁶Sr ratio for the MPG is 0.7126 (Kooiman et al. 1986), which allows for a crustal source dominated by recycled crustal materials or a mantle source strongly contaminated by supracrustal rocks. The initial ⁸⁷Sr/⁸⁶Sr ratio for the Piskahegan Group is 0.7061 (Anderson 1992), requiring some input from the mantle, consistent with the bimodal nature of the caldera rocks (McCutcheon 1990a; McCutcheon et al. 1997, 2001). A mantle-derived mafic magma could cause partial melting of crustal rocks and generate the felsic magmas responsible for the MBG and MPG. Given low Sr and high initial ⁸⁷Sr/⁸⁶Sr ratio together with åNd(t) and ä¹⁸O values of the MPG rocks, the most likely scenario is that the magma originated from partial melting of juvenile materials and was subsequently contaminated by supracrustal rocks. Based on a systematic study of geochemistry, Taylor (1992) reached the same conclusion for the source material of the Pleasant Ridge topaz granite that has many chemical similarities to the MPG.

Two models are commonly used to explain topaz-bearing granitic rocks: very small degrees of partial melting of continental crustal material that was subjected to an earlier episode of melting (Christiansen et al. 1986; Manning and Hill 1990) and extensive fractionation of felsic magma (Dostal and Chatterjee 1995, 2000; Zhu et al. 2001). Generally, the magmas are generated in an extensional tectonic regime (Christiansen et al. 1986; Whalen et al. 1987, 1996). Moreover, extensive liquid fractionation involving thermogravitational diffusion, liquid immiscibility, and vapour-phase transport has also been suggested to explain the petrochemical variations and the substantial enrichment of incompatible trace elements (Christiansen et al. 1983, 1984, 1986; Dostal and Chatterjee 1995, 2000; Zhu et al. 2001).

Large variations in compatible elements, such as Sr, Ba, and Eu in the MBG and MPG (Table 1, Figs. 7 and 8), rule out the possibility of variable degrees of partial melting of the same source, although small degrees of partial melting can generate high contents of incompatible elements (see Rollinson 1993; Christiansen and Keith 1996). Fractional crystallization, however, not only reduces the abundance of compatible elements, but also increases the concentrations of incompatible elements (e.g., Li, Rb, Cs, U, Th, Nb, and Ta), without any variations in the isotopic ratios if the effect of supracrustal contamination is not considered.

The petrochemical data presented in this study (Table 1; Figs. 5– 9, 11, 12) are modeled assuming that the MBG and MPG were produced by extensive low-T fractional crystallization of the same parental magma, rather than from two batch-melts that evolved separately. The extreme fractional crystallization is interpreted to have taken place mainly at depth, as indicated by two separate compositional groups, but with the same evolutionary trends for the MBG and MPG (Figs. 5, 6c & d, 9, 11, and 12). However, supracrustal contamination of selective components is also reflected in some major elements (Fig. 5) and high initial ⁸⁷Sr/⁸⁶Sr ratio, as well as trace elements such as Rb and Cs (Fig. 12; Table 1). The erratic distribution of the alkali elements is interpreted as the manifestation of contamination or alteration, whereas the tightly constrained distribution of TiO2, P2O5, and MgO is interpreted as primary eatures. No alteration trends are apparent on Figures 5j, k (cf. Kontak et al. 2002). Figure 13 illustrates the log-log plot of incompatible (Nb) versus compatible (Sr) elements for the samples from the MBG and MPG, in which the samples are mainly confined in the field between fractional crystallization (FC) and assimilation-fractional crystallization (AFC) (r = 0.5) curves. Assuming a leucogabbro (Thorne and Lentz 2002) magma as the starting point and using upper continental crust (Taylor and McLennan 1985) as contaminant, the FC and AFC paths (Fig. 13) can be constructed with the equations of Rollinson (1993), given the bulk partition coefficients (DSr and DNb) estimated by the method of Allégre et al. (1977) (see details below). The parameters used in the modeling are given in the caption of Fig. 13. General vectors produced by Rayleigh fractionation of single phases (e.g., plagioclase, K-feldspar, bio-tite, hornblende, and clinopyroxene) and their effects on melt composition are also shown for comparison. The compositions of average N-type MORB (Saunders and Tarney 1984), average OIB (Sun 1980), lower continental crust (Weaver and Tarney 1984), and upper continental crust (Taylor and McLennan 1985) are plotted on Fig. 13, which constrains the probable sources for the granitoids from the Mount Pleasant Caldera. All these data suggest that magma evolution was controlled by extreme fractional crystallization plus selective supracrustal contamination. Figure 6c indicates that fractional crystallization may have proceeded under relatively reduced conditions, as high ƒ(O2) would have resulted in crystallization of magnetite and depleted the residual magma in iron.

A simplified calculation was performed to test the probability of petrogenetic connection between the MBG and MPG through fractional crystallization. Preliminary trace-element modeling, together with major-element petrochemical evidence, indicates that the mineral assemblage involved in fractional crystallization was clinopyroxene, amphibole, pla-gioclase, K-feldspar, biotite, zircon, and apatite (Fig. 13). Some minerals rich in light REE (e.g., monazite and allanite) may also occur as separate phases, although only a very small proportion is required. The most evolved sample (NMP89-1-1849) from the MPG is equivalent to 10– 15% residual melt (for most incompatible elements, such as Nb, Ta, and HREE) generated by the most primitive sample (295) from the MBG that is assumed to represent the composition of the parental magma in terms of a fractional crystallization model (Christiansen and Keith 1996) and the estimated bulk partition coefficient (D) as described below. Inconsistency with a lower degree of frac-tionation for incompatible elements (e.g., Rb) is interpreted to imply that the enrichment of the high field strength elements, such as Nb, Ta, and HREE, may require fluid involvement. Dostal and Chatterjee (1995, 2000) showed that fluid fraction-ation could have resulted in differential enrichment of these elements to the level observed in ca. 370 Ma topaz leucogranite from South Mountain Batholith in Nova Scotia. Compared to their data, the present study indicates that fluid fractionation may have played an important role in the extreme enrichment of these elements (e.g., Nb, Ta, and HREE) in the MPG (see discussion below).

This model of fractional crystallization coupled with fluid fractionation plus supracrustal contamination proposed in this study (Fig. 14) differs from the previous model by McCutcheon (1990a), wherein he suggested that two different magmas (i.e. two stages of partial melting of lower crust) evolved separately in high-level chambers by AFC processes to give rise to the different phases. McLeod (1990) proposed that highly evolved granites like the MPG may be derived from partial melting of a F-rich source (crust) at depth, which subsequently fractionated repeatedly in a higher level subvolcanic chamber. In that model, the origin of the MBG and its relation to the Mount Pleasant granites were not discussed.

In terms of the Rayleigh fractionation law, the partition coefficient D of a hygromagmatophile element such as U can be assumed to be zero in order to evaluate geochemical behaviour of the other elements in a fractionating magmatic system. An equation for this relation is deduced based on this assumption (Allégre et al. 1977):

-> See the list of figures

where C_L and C_o denote the concentration of an element of interest in residual melt and parental melt, respectively, D is the bulk partition coefficient of the element between the crystallizing mineral assemblage and the residual melt, C*_L and C*_o represent the content of a hygromagmatophile element in the residual and parental melts, respectively, whose D* is supposed to equal to zero. Therefore, plotting ln C_L against lnC*_L gives a straight line, given D remains constant during fractional crystallization (Allégre et al. 1977; Christiansen et al. 1983; Fowler and Jensen 1989). Using least-square regression, the slope (m) of the regressed line can be obtained (Fig. 12). Hence, the bulk partition coefficient (D) can be determined for that element (i.e. m = 1 – D).

The D calculation results indicate that Au (1.14) together with Sr (1.81), Ba (2.73), Hf (1.27), Zr (1.53), La (1.20), Ce (1.12), Pr (1.31), Eu (3.29), Sc (1.65), Zn (1.21), and Pb (1.24) were compatible elements during magmatic evolution, suggesting that there is less chance to find Au mineralization in the evolved MPG. Thus, it is better to look for gold within and/or around the early MBG. Based on their D values, Rb (0.34), Cs (0.80), Ga (0.70), Ta (0.16), Nb (0.22), Th (0.56), Y (0.22), Sm (0.83), Gd (0.6), Yb (0.21), Lu (0.23), Sn (-0.18), W (0.96), Cu (0.52), and Mo (0.60) behaved as incompatible elements during magmatic evolution.

The technique of Allégre et al. (1977) applied to granitoid systems, such as those in the Mount Pleasant Caldera, may not be interpreted easily. Firstly, the mobility of U in such subvolca-nic systems must be considered during late- to post-magmatic hydrothermal alteration processes, although a large proportion of U is hosted in zircon in the MBG and MPG. Incompatibility of Th in a granitoid system is fairly similar to that of U, but Th is considered to be immobile during alteration processes (see Fowler and Jensen 1989; Jenner 1996). Therefore, Th is a better fractionation index than U. However, using Th as the frac-tionation index to conduct the same modeling as U described above gave the same tendency of compatibility for all elements listed above. Secondly, the fact that some elements such as Sn, Sb, Cu, and Au (see Jenner 1996) would partition into volatile phases during crystallization of the magma, must be also considered in the trace-element simulation. This consideration suggests that a significant fraction of the metals enters volatile phases that reduce their abundance in the degassed magmas. Thus, compatibility of these elements modeled in this empirical fashion (Allégre et al. 1977) may not really reflect their behaviour in the crystal-melt system alone. It is noticeable that the bulk partition coefficient of Sn is negative (-0.18). This number itself is meaningless, although it is smaller than 1. Thirdly, alteration may elevate or decrease some elements in altered granitoids, which also influences the modeling results. All these effects must be carefully evaluated before a conclusive determination of D for a metal during evolution of a granitic magma system as described here can be achieved. Then, these data for estimated metal bulk partition coefficients can be applied to assess the mineralization potential.

It is notable that the MPG has very low K/Rb ratios (26– 56), requiring fluid involvement in formation of the rocks (Shaw 1970; Dostal and Chatterjee 1995, 2000). Conversely, the MBG has high K/Rb ratios (172– 376), which are typical for magmatic processes. Nb/Ta ratios are lower for the MPG (2.7– 7.6) compared to the MBG (7.7– 28.6), also requiring fluid involvement in the late-stage granites. Intriguingly, the MPG has relatively constant Zr/Hf ratios (12.8-17.1), whereas the MBG shows remarkably higher Zr/Hf ratios (24.3– 48.0), suggesting that Zr/Hf ratios in the residual melts were controlled mainly by fractional crystallization of zircon (Linnen and Keppler 2002), or by fluid fractionation (Dostal and Chatterjee 1995, 2000). The Th/U ratios are lower in the MPG (1.1– 1.8) and relatively high in the MBG (3.0– 7.1), suggesting that U increases more rapidly than Th, consistent with the above estimation of the bulk partition coefficient (D). In other words, U was more incompatible than Th in the magma system. It is known that Nb and Ta are strongly affected by the presence of volatiles and can be considered with Li, B, F, Rb, and Sn as volatile-associated elements (Manning and Hill 1990). In contrast, Zr and Hf are considered to be more refractory, in view of the low solubility of their accessory minerals in peraluminous granitic melts (Watson 1979; Watson and Harrison 1983). P, Ti, Sr, and Ba mainly behave as compatible elements in the granitic system.

Liquidus temperatures of granitic rocks can be estimated from accessory mineral saturation temperatures (Watson and Harrison 1983; Harrison and Watson 1984; Montel 1993). These temperatures are capable of reflecting the mechanism of magmatic evolution. In a fractional crystallization sequence, the liquidus temperatures of successive intrusive phases in a granitoid complex decrease from early to late stage crystallization. On the other hand, the liquidus temperatures of a second stage melt would be higher than a first-stage melt during partial melting, but once formed both evolve by FC related to decreasing temperatures.

Monazite-saturation geothermometry (Montel 1993) yields temperatures as high as 825– 950°C for the MPG, similar to those of the MBG (834– 950°C). Abnormally high estimated temperatures for the MPG, assuming it formed through fractional crystallization discussed above, are attributed to LREE contents (Table 1 and Fig. 7) similar to those of the MBG. Possibly other unidentified minerals host a significant part of the LREE in the bulk rock. If this portion of LREE were removed from the calculation, then the estimated temperatures would have been lowered in terms of the monazite-saturation geothermometry. However, zircon-saturation geothermom-etry (Watson and Harrison 1983) produces different results for the MPG (725– 789°C) and MBG (769– 933°C). Apatite-saturation geothermometry (Harrison and Watson 1984) gives results similar to those from zircon-saturation geother-mometry for the MPG (712– 791°C) and MBG (833– 963°C). Decrease of the estimated liquidus temperatures (Fig. 15) from the MBG to MPG is consistent with dropping temperatures in a fractionating magma sequence.

High F concentration (3500– 9900 ppm,) together with high Li (up to 610 ppm) (see Sinclair and Kooiman 1990) and high alkalis in the water-saturated MPG could remarkably reduce solidus temperatures to as low as 525°C, as shown by experimental work on F-rich (1.2 wt.%) topaz-bearing granitic melts (Webster et al. 1987). The large temperature range for crystallization (950– 525°C) allows protracted extreme fractionation of granitic melts such as the MPG, resulting in unusual enrichment of incompatible elements such as F, Li, Rb, Cs, Y, Nb, Ta, Ga, Sn, and W (Fig. 14).

The geochemical features of granitoids may reflect their tectonic environment to some extent (Pearce et al. 1984). Various geochemical discrimination diagrams for tectono-magmatic classification have been proposed to distinguish tectonic settings of magmatic rocks in terms of their chemistry (see Rollinson 1993).

Geochemical discrimination diagrams suggest that the MBG could have formed in volcanic-arc to within-plate tectonic settings, whereas the MPG samples plot entirely in the within-plate field (Figs. 11a and 11b). Figure 11d yields a similar result. As shown above, extensive fractionation plus fluid involvement have strongly enriched high field strength elements, which may shift the MPG samples to the within-plate field (Figs. 11b & d). Supracrustal contamination also affects the discriminant result (Fig. 11a). Therefore, caution must be exercised when using these diagrams to discriminate tectonic setting. These classification diagrams are suspect (cf. Barker et al. 1992), because they characterize the tectono-magmatic environment under which the granite protolith formed, not necessarily the environment prevailing during granite emplacement (Whalen et al. 1996; Christiansen and Keith 1996). Thus, the trace element signatures of granitoids directly reflect their protolith, melting, and crystallization histories, rather than their tectonic settings. Granitoids from the Mount Pleasant Caldera appear to be crustal A2-type (Fig. 11c), inferred to have been derived from continental crust or underplated crust that has been through a cycle of continent-continent collision or island-arc magmatism (Eby 1990, 1992). Particularly, the MPG is similar in geochemistry to topaz rhyolite (Christiansen et al. 1983, 1984; Christiansen et al. 1986), ongonite, and topaz granite (Kontak 1990; Taylor 1992; Dostal and Chatterjee 1995; Zhu et al. 2001), and also to the Late Devonian Mount Douglas Granite in the eastern part of the Saint George Batholith (Lentz et al. 1988; McLeod et al. 1988; McLeod 1990).

During Late Devonian, the studied region was in the early stages of basin extension (McCutcheon et al. 1990a) followed the Acadian Orogeny, consistent with the discriminant results in terms of Fig. 6d and the multicationic diagram of Batchelor and Bowden (1985) (not shown).

The highly evolved MPG is associated with granophile element mineralization and characterized by high fluorine, lithium (see Sinclair and Kooiman 1990), and alkali concentrations, which enhance the low-T fractionation. The presence of Li, B, and F have a significant effect on phase equilibria in the water-saturated granite system (Q-Ab-Or-H2O), reducing liquidus and solidus temperatures and affecting minimum melt compositions (Manning 1981; Webster et al. 1987; Manning and Hill 1990) and decreasing viscosity of the melt (Baker and Vaillancourt 1995), thus promoting the duration of crystallization to low temperatures. On the basis of F alone, the shift in minimum position toward the Ab apex in the Q-Ab-Or system has been taken as evidence for origin of topaz granite, and other F-rich granitic rocks, through fractional crystallization of biotite granite magma (Manning 1982). F enrichment generally leads to a concomitant enrichment in Sn, Nb, and Ta, which may possibly be directly complexed with F in the melt (Manning and Hill 1990; Dostal and Chatterjee 1995). A strongly depolymerized F-rich melt is more capable of hosting incompatible elements relative to polymerized volatile-poor granite (Watson 1979; Watson and Harrison 1983; Linnen and Keppler 1997), irrespective of complexing with F (Manning and Hill 1990). Based on the estimated D value of Sn, the late GRIII appears to have high potential for producing Sn mineralization.

Geochemically, the less differentiated MBG is broadly comparable to the granitoids associated with intrusion-related gold systems elsewhere (McCoy et al. 1997; Thompson et al. 1999; Lang et al. 2000; Lang and Baker 2001; Yang et al. 2002a, b). Granitoid intrusions related to gold mineralization are mostly metaluminous, subalkalic (calc-alkalic) to weakly alkalic with intermediate to felsic compositions, forming simple to multiphase complexes (Lang et al. 2000). Locally, peraluminous intrusions are present, but they are ascribed to late-stage frac-tionation rather than different magmas with distinctive sources (Duncan et al. 1998). According to the summary of Lang et al. (2000), the main-stage, subalkalic (calc-alkalic) intrusions have silica contents between 62 and 72 wt. % SiO2, and late differentiates reach 78 wt. % SiO2. REE patterns display LREE enrichment relative to HREE, with La and Yb concentrations about 100 and 10 times chondrite values, respectively. Eu anomalies are absent to slightly negative. Most intrusions are characterized as reduced I-type granitoids.

Most of the analyzed samples from the MGB and MPG (Table 1) have low Au contents, although the MBG samples appear to contain higher gold (up to 8 ppb). Compared to continental crust (2.5 ppb) and normal sedimentary rocks, such as greywacke (4.8 ppb) and shale (3 ppb) (see Lentz 2003), Au does not appear to be elevated significantly in the granitoids from the Mount Pleasant Caldera. Regionally, the granitoids in the area also contain relatively low Au compared to continental crust (McLeod 1990; Whalen 1993). Available studies of other intrusion-related Au deposits have shown that the content of Au in unaltered granitoids is not significantly different from those of barren granitoid rocks (McCoy et al. 1997). These data suggest that background Au abundance in the granitoids is not an essential parameter controlling associated gold mineralization. This observation is in accord with the conclusion by Tilling et al. (1973) that it is impossible to relate gold abundance in igneous rocks and gold mineralization. The geochemical behaviour of gold during the magmatic-hydro-thermal evolution of granitoid systems may be a key factor responsible for gold mineralization, which is controlled by the physiochemical conditions and composition of the magmas. Local geological setting (e.g. structure and lithology) and interaction between the intrusions and country rocks affect the magmatic-hydrothermal systems, which ultimately places controls on the emplacement of gold deposits.

Gold is not expected to concentrate in the residual melt during protracted fractionation processes, because it occurs as a compatible metal. Gold is preferentially partitioned into sulphides in equilibrium with a reduced magma, based on empirical observation (McCoy et al. 1997) and experimental studies (Jugo et al. 1999; Kesler et al. 2002). However, early crystallization of magnetite did not appear to severely influence the ability of a calc-alkaline granitic magma to yield a gold-rich ore fluid, according to experimental study of Au partitioning between magnetite and granitic magma (Simon et al. 2003). Consequently, any sulphide saturation in magma systems must greatly deplete Au in the silicate melts. Inverno (1991) reported much higher Au values in his granite samples from the Mount Pleasant Mine, e.g., Au in the GRII ranging from 8 to 85 ppb (average = 28.7 ± 2.6 ppb, n = 6), and the GRIII ranging from < 5 to 27 ppb (average = 13.6 ± 2.5 ppb, n = 6) (see his appendix 7). These high Au contents may be ascribed to hydrothermal alteration, although the Au concentrations decrease from the GRII to GRIII. Our greisenized granite samples taken from True Hill also show higher Au contents (up to 100 ppb), suggesting that late-stage hydrothermal fluids could scavenge gold that is incorporated in early immiscible sulphides in granitoids (Yang and Lentz 2003). At the present stage, it is not easy to evaluate timing of sulphide saturation in felsic magmatic systems, in particular in granite intrusions that were usually influenced by subsolidus processes. These processes may alter the primary sulphides on one hand, and form hydrothermal sulphides on the other. To date, few convincing methods exist for distinguishing primary and hydrothermal sulphides in granitoids, although studies on petrography, chemical, and sulphur isotopic compositions of the sulphides could provide some clues as to their origin. More work needs to be done to better understand the origin of intrusion-related Au ore deposits.