Abstract

Geochemical data for a comprehensive suite of over 700 samples of massif anorthosite and associated rocks from the Adirondack Mountains, New York, exhibit a ubiquitous mixed tholeiitic and calc-alkaline signature indicating a complex petrogenesis. The origin and relationship of massif anorthosite to associated rocks has been a major petrological problem despite decades of study. The Adirondack Mountains is one of the best areas to study these rocks because it contains one of the world's most abundant occurrences of massif anorthosite and associated rocks forming bedrock over thousands of square kilometers. The suite of rocks analyzed includes both anorthosite suite rocks, consisting of anorthosites, leucogabbros, gabbros, oxide apatite gabbronorites (OAGNs), and oxide gabbronorites (OGNs) (defined in text), and mangerite suite rocks consisting of jotunites, monzodiorites, mangerites, and charnockites. Representative major-element compositions were determined largely by X-ray fluorescence (XRF) analysis for 352 massif anorthosites and associated rocks and a variety of trace elements were determined by XRF, instrumental neutron activation analysis (INAA), and inductively coupled plasma mass spectroscopy (ICP-MS) for 296 massif anorthosites and associated rocks. All rock types show a mixture of tholeiitic and calc-alkaline compositional characteristics with major elements exhibiting a strong iron enrichment tholeiitic trend and trace elements showing a depletion of Nb and Ta characteristic of calc-alkaline rocks. Prior to this study the geochemistry of these rocks in the Adirondacks has been only poorly characterized from scattered local studies.

The anorthosite suite of rocks exhibits two distinct compositional trends. Massif anorthosites, leucogabbros, gabbros, OAGNs, OGNs, along with jotunites, separate into two distinct compositional trends on P2O5-MgO and TiO2-MgO diagrams, whereas monzodiorites, mangerites, and charnockites have only one compositional trend. Two trends in anorthosites are caused by two varieties of anorthosite: one type with a characteristic mineralogy dominated by plagioclase plus pyroxene and another type dominated by plagioclase plus oxide minerals and apatite. Mafic enclaves at some localities near the margins of anorthosite masses contain gabbro, OGN, and OAGN in close spatial association, suggesting they represent crystallization from the same or similar parental magmas at different stages of evolution.

INTRODUCTION

The Adirondack anorthosites belong to the Proterozoic massif-type anorthosites, which are coarse-grained intermediate composition plagioclase-rich plutons lacking the excellent layering and interlayered mafic rocks observed in layered-type anorthosites (Ashwal, 1993). Massif-type anorthosites occur in Proterozoic rocks on all continents (Anderson, 1969; Ashwal, 1993), but in northeastern North America are concentrated in a belt stretching from the Adirondack Mountains in New York northeast through Quebec to the coast of Labrador (Fig. 1). The vast majority of the North American Proterozoic anorthosite masses occur in the Grenville Province and Nain Province. The Adirondack Mountains are a southwestern outlier of the Grenville Province and lie within the Central Granulite terrane (Wynne-Edwards, 1972) and Allochthonous Monocyclic Belt (AMB) of Rivers et al. (1989). The Adirondack region consists of a Lowland and Highland terrane with the latter bearing similarities to the Morin terrane of Quebec (Rivers et al., 1989). Massif-type anorthosites occur in the Highland terrane and are typically associated with a variety of mafic and felsic rocks; leucogabbros, gabbros or norites, oxide-rich rocks, jotunites, mangerites, and charnockites or granites and the entire anorthosite-mangerite-charnockite-granite assemblage is referred to as the AMCG suite (Emslie and Hunt, 1990; McLelland and Whitney, 1990). The spatial relationship of the Adirondack anorthosites with other AMCG rocks is shown on the Adirondack sheet of the New York Geological Map by Isachsen and Fisher (1970).

The anorthosite-mangerite-charnockite-granite (AMCG) suite of rocks in North America is typically regarded as an association of two rock suites that are coeval but not comagmatic (Emslie et al., 1994; McLelland et al., 1994; Bickford et al., in press). The anorthosite suite is regarded as consisting of comagmatic anorthosite, leucogabbro, and gabbro, plus OGN and OAGN, while the mangerite suite is regarded as consisting of comagmatic jotunite, monzodiorite, mangerite, and charnockite, with some granite. The ubiquitous mixed tholeiitic and calc-alkaline compositional characteristics of both suites suggests that both suites are contaminated, perhaps by melting of enriched mantle or deep mafic crust, or even tongues of crustal material in the mantle (Duchesne et al., 1999). The ubiquitous presence of enriched isotopic values, such as elevated 18O values in all rocks from both suites (Valley, 2003), suggests that all AMCG rocks have a common origin early but diverged to form different suites of intrusive magmas. Hf model ages by Bickford et al. (in press) strongly support geochemical and isotopic enrichment in the two rock suits because of initial melting of enriched mantle or mafic lower crust formed during a previous 1350–1400 Ma tectonic event. Anorthosite suite magmas are typically regarded as forming from upward moving basaltic magmas that pond at a density interface where heavy mafic minerals sink while lighter plagioclase crystals float upward into escaping feldspathic gabbroic magmas that diapirically intrude overlying crustal rocks. Mangerite suite magmas are typically regarded as forming as the heat from the ponded magma melts overlying more felsic crustal rocks and intrude upward along paths similar to those of anorthosite suite magmas into shallow crustal rocks. This scenario is supported by experimental phase relation studies by Longhi et al. (1999) that indicate feldspathic gabbro magma compositions cannot be derived by partial melting of depleted mantle (DM). Isotope values are also inconsistent with melting of depleted mantle because the DePaolo (1981) DM evolution model would produce εNd(1155–1160) = +5.356 to +5.369 whereas the highest εNd(T) value observed for the ∼1155–1160 Ma Adirondack anorthosites is +4.39 (Ashwal and Wooden, 1983a). Daly and McLelland (1991) found an older crustal component in their Adirondack mangerite series εNd(T) data, and the presence of an older component is supported by study of ultramafics in the Adirondack lowlands by Chiarenzelli et al. (2010). Derivation of AMCG rocks from subcrustal lithospheric mantle (SCLM) or lower crust, present from early in Earth's history (Griffin and O'Reilly, 2007), explains both the calc-alkaline characteristics shown by AMCG rocks and their 1350 to 1400 Ma Hf model ages (Bickford et al., in press).

PREVIOUS ADIRONDACK STUDIES

The extent and distribution of massif anorthosite and associated rocks in the Adirondack Mountains were largely determined by early geologic studies summarized in the classical Geological Society of America Memoir on Adirondack Igneous Rocks and Their Metamorphism by A.F. Buddington (1939). Their distribution is best shown on a regional scale on the Adirondack sheet of the Geological Map of New York (Isachsen and Fisher, 1970). The foliation pattern formed by mafic minerals on a structural map of the 3000 km2 Adirondack anorthosite massif by Balk (1931) was interpreted to be a primary feature, suggesting the massif consists of one large intrusion bordered on the east by a few smaller intrusions (Buddington, 1939). The Adirondack anorthosite massif was divided into three separate units: the large St. Regis-Marcy massif in the west, and smaller Jay-Mt. Whiteface and Port Kent-Westport bodies in the east. Outcrop distribution by rock types in the large Marcy Massif is shown in Geology of the Saranac Quadrangle by Buddington (1953), Bedrock Geology of the St. Regis Quadrangle by Davis (1971), and several unpublished quadrangle maps. Several quadrangles have been mapped or remapped but never published. The Lake Placid quadrangle mapped by Percy Crosby (1970) and the Elizabethtown quadrangle by Matt Walton (ca. 1960) can be found as open file reports in the New York Geological Survey office. Another quadrangle map, the Santanoni quadrangle by Ollila (1984), remains unpublished. Anorthosite domes and layers northeast of the main Adirondack anorthosite massif having a larger foliated mafic component were mapped by Gasparik (1980), Whitney and Olmsted (1993) in the Au Sable Forks quadrangle, and Buddington and Whitcomb (1941) in the Willsboro quadrangle. Smaller domical anorthosite bodies south of the main Adirondack massif include the Oregon Dome in the Thirteenth Lake quadrangle (Krieger, 1937) and the Snowy Mountain Dome in the Indian Lake quadrangle. Detailed mapping of the latter by de Waard and Romey (1969) provided additional detail on its extent and petrology. The felsic mangerites and charnockites associated with massif anorthosite outcrop in several separate areas around and adjacent to the massif anorthosite plutons rather than forming a continuous band around anorthosite massifs. We have limited sampling to the central and northeastern Adirondacks (Fig. 2), although numerous smaller anorthosite plutons, layers, and lenses occur elsewhere in the Adirondack highlands.

Gravity data suggests that in three dimensions the large Adirondack anorthosite massif is a thick plate with several descending feeder pipes for carrying magma. Wavelength-filtering of gravity data indicates the Adirondack massif is actually a gravity high with deep mafics underlying an anorthosite cap (Simpson et al., 1981; Arvidson et al., 1982; Kerr, 1982). An early gravity study by Simmons (1964) used density differences between anorthosite and adjacent gneisses to define the massif as a 3.0–4.5 thick plate with two cylindrical gravity lows interpreted as possible feeder pipes descending to depths of ∼10 km. A later gravity study by Mann and Revetta (1979), shown as figure 3 in Whitney and Olmsted (1993), and Revetta (2003, personal commun), defined several additional gravity lows suggesting deeper roots beneath the anorthosite caps and providing additional evidence that even the large Marcy massif actually consists of two separate groups of multiple plutons (Fig. 3). The western Marcy group consists of two gravity lows in the St. Regis and Long Lake quadrangles and the eastern Marcy group consists of several gravity lows near Jay, Hurricane Mountain, the town of North Hudson, and Blue Ridge; additional gravity lows occur near Keeseville and Westport. The two groups of Marcy gravity lows are separated by a shallow east-west tongue of anorthosite characterized by negative gravity and another shallow tongue of anorthosite with no gravity expression, sometimes called the McKenzie lobe, extends NE from the town of Saranac Lake.

Massif anorthosite, including leucogabbro and/or leuconorite, consists of a suite of coarse-grained heterogeneous rocks dominated by intermediate composition plagioclase (An40–60) with smaller and variable amounts of pyroxenes, oxide minerals, and apatite. Rare giant high alumina orthopyroxene megacrysts with plagioclase and oxide exsolution lamellae have been described by Emslie (1975), Bohlen and Essene (1978), and Jaffe and Schumacher (1985), who interpreted them as indicative of formation at high pressures and transported to their present sites. The relative percentages of primary igneous minerals have often been altered during a later granulite facies metamorphism that also produced new minerals, mainly garnet, metamorphic clinopyroxene, amphibole, and local biotite. Anderson and Morin (1969) classified the Adirondack anorthosites as labradorite anorthosites because plagioclase may be >An50. Typically these plagioclase-rich rocks also contain clinopyroxene, orthopyroxene, ilmenite, magnetite, and apatite in general order of decreasing abundance (Ashwal, 1978, 1982). Buddington (1939) classified the Adirondack anorthosites on their plagioclase to mafic mineral content as anorthosite (>90% plagioclase and <10% mafic minerals), gabbroic anorthosite (10%–22.5% mafic minerals), anorthositic gabbro (22.5%–35% mafic minerals), and gabbro or norite (>35% mafic minerals). More recent studies of chemically analyzed Quebec anorthosites have resulted in a more detailed geochemical and mineralogical classification of the various anorthosite facies rocks into anorthosite, leucogabbro, sodic leucogabbro, gabbro, sodic gabbro, and OAGN, where OAGN stands for oxide apatite gabbronorite (Owens and Dymek, 1992) consisting of ilmenite ± magnetite, apatite, orthopyroxene ± clinopyroxene, and plagioclase (Owens and Dymek, 2001). In addition we introduce the term OGN for rocks that appear similar to OAGN but lack apatite and we employ it here. This classification is more informative for analyzed anorthosite suite rocks and we will use this classification.

The coarsely crystalline anorthosite outcrops often exhibit large mineralogical and textural variations similar to those found over an entire massif, making mapping based on mineralogical variations difficult. Most anorthosite in the western part of the Marcy massif is relatively massive, allowing Davis (1971) to map anorthosite in the St. Regis quadrangle by mafic mineral content. Crosby (1969) attempted to classify anorthosite by mineralogy by marking large areas on outcrops and counting enclosed mineral percentages, but found any simple classification difficult to map, as did Whitney and Olmsted (1993). Another feature frequently observed in outcrop consists of blocky anorthosite inclusions in anorthosite, generally blocks of massive leucocratic anorthosite within more mafic leucogabbro or leuconorite, but sometimes the opposite. This commonly observed feature was called block structure by Balk (1931), and some authors have suggested the block structures mark boundaries between adjacent anorthosite plutons.

Zircon uranium-lead dates now indicate that most Adirondack AMCG suite rocks were intruded during the interval 1165–1150 Ma following the Shawinigan orogeny from 1210 to 1160 Ma (McLelland et al., 2004; Heumann et al., 2006). Super High Resolution Ion Microscope (SHRIMP) dates indicate most anorthosites were intruded near 1150–1155 Ma ago (Bickford et al., 2002; Hamilton et al., 2004; McLelland et al., 2004) as shallow intrusions (Clechenko et al., 2002). However, intrusion times have been found to differ for the various anorthosite plutons with samples from near the Saranac Lakes and the town of Jay showing distinctly older intrusion ages of 1161 Ma and 1160 Ma, respectively (Hamilton et al., 2004; McLelland et al., 2004). Four charnockitic bodies were dated at 1176–1154 Ma and a ferrodiorite (jotunite) dike was dated at 1156 Ma by Hamilton et al. (2004). A suite of 13 anorthosite suite samples from the Marcy and Oregon Dome massifs gives an average age of 1154 ± 6 Ma (McLelland et al., 2004). Typically the anorthosites and leucogabbros are slightly younger than the essentially coeval associated felsic rocks in the AMCG suite (Hamilton et al., 2004; McLelland et al., 2004) although there is overlap. This finding is not necessarily at odds with field observations that anorthositic andesine crystals frequently occur in jotunites adjacent to anorthosite because there are sequential pulses of varying compositions. Bickford et al. (in press) and Chiarenzelli et al. (2010) have shown that the AMCG rocks formed from enriched mantle with respect to Nd and Hf.

Isotopic values are uniformly enriched in the interior of large Adirondack anorthosite massifs relative to magma derived directly from DM, but sometimes become more variable near the margins. Also εNd(T) values vary in massif anorthosites that cross tectonic boundaries (Ashwal and Wooden, 1983b; Emslie et al., 1994) and even within multiple pluton massifs (Scoates and Frost, 1996) indicating petrogenesis involved different subcrustal lithospheric mantle (SCLM) or mafic crustal rocks. The Adirondack anorthosites typically have positive εNd(1100) values (−0.70 to +4.39) indicative of a source with long-term depletion of the light rare earth elements (LREE) (Ashwal and Wooden, 1983a), as do all of the anorthosites throughout the Grenville Province (Ashwal and Wooden, 1983b; Owens et al., 1994), but not north of the Grenville Front (Ashwal and Wooden, 1985; Ashwal et al., 1986). They have initial 87Sr/86Sr ratios between 0.703 and 0.706, with a mean near 0.704 (Heath and Fairbairn, 1969; Ashwal and Wooden, 1983a). The relatively large range of initial 143Nd/144Nd and 87Sr/86Sr ratios in anorthosites is attributed to variable amounts of interaction with SCLM or crustal materials or fluids (Ashwal and Wooden, 1983a) as is the widespread enrichment in δ18O values for AMCG rocks (Morrison and Valley, 1988a). Hf model ages indicate involvement of an older (1350–1400 Ma) crust or enriched mantle source for all AMCG rocks (Bickford et al., in press), which could explain the uniform enrichment in geochemistry and isotopic values. However, the Jay Dome anorthosite mass exhibits low δ18O values near its margins in contact with calcsilicate rocks that are interpreted as resulting from interaction between magmatic anorthosite fluids and meteoric surficial fluids during shallow intrusion (<10 km) and contact metamorphism (Morrison and Valley, 1988b; Clechenko et al., 2002). Several geobarometric techniques indicate crystallization pressures in the range of 0.3–0.5 GPa for massif anorthosite (Ashwal, 1993; Vander Auwera and Longhi, 1994). Granulite facies metamorphism reached a maximum of ∼0.8 GPa and 800 °C (Bohlen et al., 1980; Bohlen et al., 1985; Spear and Markussen, 1997) in the vicinity of the massive NW lobe of the Marcy massif.

The later granulite facies Ottawan orogeny from 1020 to 1090 Ma (Heumann et al., 2006) in the Adirondacks has made separation of individual anorthosite plutons and the boundary relations of anorthosite plutons with adjacent AMCG rocks difficult to interpret. The granulite facies metamorphism in the Adirondacks was not characterized by pervasive solutions and open system behavior for most elements (Valley and O'Neil, 1984; Edwards and Essene, 1988; Morrison and Valley, 1988a; Seifert and Chadima, 1989; Cartwright et al., 1993), although late Ottawan (1020–1050 Ma) localities near the Lyon Mountain granite have hydrothermal iron-oxide deposits (McLelland et al., 2002). The significance of whole-rock data for interpreting primary igneous petrogenesis is only maintained where these fluids are absent. However, local mineralogical and textural changes caused by the metamorphism of Adirondack anorthosites, discussed by McLelland and Whitney (1977), Ashwal (1982), and Seifert and Chadima (1989), create potential problems for interpreting mineral separate geochemical data. Also, much of the difficulty in deciphering primary igneous field relations between individual anorthosite plutons and between anorthosites and adjacent rocks is due to overprinting of primary igneous textures and boundary relations by the later granulite facies metamorphism.

CLASSIFICATION OF MASSIF ANORTHOSITE AND ASSOCIATED ROCKS

The Adirondack literature, summarized by Buddington (1972), typically separates AMCG rocks into an anorthosite suite of rocks dominated by cumulus plagioclase with pyroxene and a mangerite suite of rocks dominated by two feldspars, quartz, and pyroxene. Owens and Dymek (1992) have classified the AMCG rocks using a system based largely on geochemistry, normative mineral percentages, and normative mineral compositions. One of the major advantages of using normative mineralogy is its ability to see through metamorphic changes to the modal mineralogy. Anorthosite is defined as a rock containing greater than 90% normative plagioclase and consequently is the same as in the Buddington and most other classifications. However, samples with between 70% and 90% normative feldspar are called leucogabbro if the normative plagioclase An content is greater than An40 or sodic leucogabbro if plagioclase An is less than An40. In a similar manner when samples have less than 70% normative feldspar and between roughly 45% and 52% SiO2, they are called gabbro if plagioclase An is greater than An40 and sodic gabbro when plagioclase An is less than An40. Our discussion drops the prefix sodic and, since clinopyroxene is typically more abundant than orthopyroxene, the terms gabbro and leucogabbro are used regardless of the plagioclase composition and the orthopyroxene content.

Mafic samples with cumulus oxide minerals and less than ∼45% SiO2 and high concentrations of iron, titanium, and phosphorus have been classified as OAGNs (Owens and Dymek, 1992). In this paper we have identified a new rock type we are calling oxide-rich gabbronorite (OGN). OGNs appear similar to OAGNs in having cumulus oxide minerals but they have low concentrations of phosphorus and lack the apatite characteristic of OAGNs. OAGNs have been observed along the margins of the Labrieville and St. Urbain anorthosite massifs in Quebec (Owens and Dymek, 1992), as well as along the margins of the Adirondack anorthosites (Buddington, 1939), and as dikes and sheets in the Adirondacks (Ashwal, 1982; McLelland et al., 1994). Texturally OAGNs vary from pegmatitic to fine grained, or some combination, and mineralogically they contain one or more pyroxenes, ilmenite, magnetite, apatite, with lesser plagioclase, and perhaps some K-spar and quartz. These mafic rocks rich in Fe, Ti, and P are associated with almost all massif anorthosites (Hargraves, 1962; Philpotts, 1981; Ashwal, 1982; Duchesne, 1984; Goldberg, 1984; Owens and Dymek, 1992). Their origin remains controversial with the Quebec OAGNs considered to not be comagmatic with massif anorthosites (Owens and Dymek, 1992) whereas the Adirondack OAGNs are considered to be residual magma from which the anorthosite suite rocks have separated earlier (McLelland et al., 1994). OGNs in the Adirondacks are newly described during this geochemical study and their distribution is not known at this time.

Associated mafic rocks with roughly 50% SiO2 often observed along the margins of anorthosite massifs can be separated into dark-gray gabbro with several percent MgO and gray-green jotunite with a lower MgO content. The jotunites sometimes appear to grade outward into more felsic monzodiorites, mangerites, and charnockites. In our classification charnockites contain greater than 10% normative quartz and mangerites contain less than 10% normative quartz.

SAMPLE GEOCHEMISTRY

This paper is based on the major-element analysis of 352 massif anorthosites and associated rock samples from the NE Adirondacks consisting of anorthosite, leucogabbro, gabbro, oxide-rich OAGNs and OGNs, jotunites, monzodiorites, mangerites, and charnockites (Tables 1–3). Sample analyses reported here are part of a larger collection of over 700 samples collected by a variety of individuals over several decades, including Adirondack anorthosite (AA) and Adirondack Baker Mountain (ABM) samples collected by the first author (K.E.S.), largely between 1972 and 1980, AD and ADK samples collected by the second author (R.F.D.), and many samples (prefixes AF, WB, MG, xxS, and others) collected by the third author (P.R.W.). Lew Ashwal also allowed us to analyze several of his samples. The NE Adirondack samples come from the large Marcy massif, the smaller Jay, Westport, and Keeseville domes to the east, and the smaller Snowy Mountain and Oregon anorthosite domes to the south (Fig. 3), although, because of the large number of samples, only AA, ABM, AD, and ADK plus a few other sample locations are shown in Figure 3. We have attempted to define representative compositional ranges for the various rocks by analyzing many samples of each rock type from many plutons over a large area of the central and NE Adirondacks (see the Appendix). The compositional variations within the individual anorthosite plutons have been defined in more detail by local studies reported in the many Adirondack quadrangle reports. Our sample collection was guided largely from the many existing geologic quadrangle maps, although some samples were collected from areas where no detailed geologic maps were found. Samples from the Marcy massif were collected using maps from the Saranac quadrangle (Buddington, 1953) and St. Regis quadrangle (Davis, 1971), from the Jay dome using the Au Sable Forks quadrangle map (Whitney and Olmsted, 1993), an unpublished Lake Placid quadrangle map by P. Crosby, from the Snowy Mountain and Oregon domes using the Thirteenth Lake quadrangle map (Krieger, 1937), and from the Westport and Keeseville domes using the Willsboro quadrangle map (Buddington and Whitcomb, 1941). Several samples were collected from near the top of Baker Mountain NE of the town of Saranac Lake using a detailed geologic map by Seifert (1978). Sample collecting from unmapped quadrangles involved following anorthosite outcrops into unmapped areas, searching for outcrop locations mentioned by Buddington (1939) or other geologic papers, and checking steep slopes or hills with probable outcrops using U.S. Geological Survey (USGS) topographic quadrangle maps guided roughly by the Adirondack sheet of the Geological Map of New York (Isachsen and Fisher, 1970). In many locations the only lichen-free outcrops are along highways (Fig. 4). Gravity data for the northeastern Adirondack Mountains (Mann and Revetta, 1979) indicate our samples were collected from several gravity anomalies, highs and lows, associated with anorthosite suite rocks. In addition, detailed gravity maps (F. Revetta, 2003, personal commun.) for the Saranac Lake, Lake Placid, Au Sable Forks, and Willsboro quadrangles have provided additional control on the relationship between sample location and gravity data.

ANALYTICAL TECHNIQUES

Representative major-element compositions were determined largely by X-ray fluorescence (XRF) analysis for 352 massif anorthosites and associated rocks and a variety of trace elements were determined by XRF, instrumental neutron activation analysis (INAA), and ICP-MS for 296 massif anorthosites and associated rocks. Some analyses taken from the literature were determined by wet chemistry. The major and most trace-element geochemistry of whole rock samples have been determined by XRF at Washington University, the University of Massachusetts, or McGill University. The AA, ABM, AD, and ADK samples were analyzed by R.F.D. at Washington University and trace-element geochemistry was determined for these samples and 13 of our plagioclase separates by INAA by K.E.S. at Washington University. Rare earth element (REE) analysis for samples other than the AA, ABM, AD, and ADK samples was determined by ICP-MS by P.R.W. at Union College. Only the seven REEs determined by INAA in the AA and AD samples are reported, although all 14 of the REEs were determined by ICP-MS in many whole rock samples. The AA, ABM, AD, and ADK sample powders were prepared by crushing bulk rocks with a sledgehammer and grinding in a shatter box with either a mullite or alumina chamber and disk. Major and several trace elements analyzed at Washington University were determined by XRF techniques reported by Couture et al. (1993) and Couture and Dymek (1996), whereas trace elements analyzed at Washington University were determined by INAA techniques reported by Jacobs et al. (1977) and Korotev (1987a, 1987b) using TEABAGS (Lindstrom and Korotev, 1982). Elements determined by XRF include major elements; Si, Ti, Al, Fe, Mn, Mg, Ca, Na, K, and P, along with trace elements V, Cr, Cu, Ni, Zn, Ga, Rb, Sr, Y, Co, Zr, Nb, Ba, and Pb. Elements determined by INAA include Fe, Na, La, Ce, Sm, Eu, Tb, Yb, Lu, Sc, Co, Cr, Ni, As, Sb, Th, Hf, Ta, Ba, Rb, Sr, and Cs. Reported values for elements determined by both techniques are those from the technique where they are most precisely determined: Sr, Ba, Rb, and Ni by XRF, and Cr and Co by INA analysis. USGS standards and repeated analysis of selected samples were used to assure accuracy in all analyses. Splits of the same powders were used for most XRF and INAA analyses. The “R” prefix samples from Buddington (1939) were analyzed by classical wet-chemical methods.

GEOCHEMISTRY OF MASSIF ANORTHOSITE AND ASSOCIATED ROCKS

Whole-rock geochemical data for our samples are presented in three tables; Table 1 with major elements and normative mineralogy of 352 of our analyzed samples; Table 2 with the REEs of the 201 samples with REE data; and Table 3 with up to 19 additional trace elements determined by XRF and INAA for 296 of our samples. The data reported in Tables 2 and 3 are confined to samples listed in Table 1. The usefulness of our data is extended by the isotopic data available for several of our AA-series samples (Ashwal and Wooden, 1983a; Morrison and Valley, 1988a; Morrison and Valley, 1991). The composition of the coarse-grained and heterogeneous massif anorthosites and associated leucogabbros varies largely as a function of the relative percentage of cumulus plagioclase to other minerals. Most other rocks have little or no cumulus plagioclase, although both the OAGNs and OGNs have cumulus oxide minerals and the OAGNs have cumulus apatite. Some of the less extreme OAGNs, e.g., the “Woolen Mill Gabbro” and ETS51C, are unlikely to be cumulates based on texture and field occurrence. Other associated rocks, gabbros, jotunites, monzodiorites, mangerites, and charnockites, typically have more homogenous mineral distributions and are dominantly medium-grained plagioclase, pyroxene, oxide mineral, apatite, potash feldspar, and quartz rocks with potash feldspar and quartz increasing in the more felsic mangerites and charnockites. The composition of these rocks more closely approaches either initial or derived magma compositions because they typically lack significant amounts of cumulus minerals. The later Adirondack granulite facies metamorphism is considered to be largely a closed system event not characterized by pervasive solutions and open system behavior (Valley and O'Neil, 1984; Edwards and Essene, 1988; Morrison and Valley, 1988a), although locally fluids have been active (McLelland et al., 2002), and the significance of whole-rock geochemical data for interpreting primary igneous petrogenesis should be largely maintained, although the modal mineralogy may be changed.

The major-element composition of the noncumulus rocks varies from gabbros with SiO2 between 45% and 52%, Fe2O3t between 8% and 17%, MgO between 2.5% and 7.5% to charnockites with SiO2 between 62% and 71%, Fe2O3t between 2.5% and 10%, and MgO less than 1% (Table 1). Intensive variable Mg# [Mg# = mol ratio of Mg2+/Mg2++Fe2+] relates to the evolutionary stage of the magma from which a rock crystallizes and decreases from a Mg# between 30 and 62 for the gabbros progressively downward through the jotunites, monzodiorites, and mangerites to charnockites with a Mg# between 9 and 29. Another intensive variable, %An, follows the same decreasing trend through the same rock types. Cumulus plagioclase-rich anorthosites approach plagioclase mineral compositions and both anorthosites and leucogabbros have SiO2 contents between 52% and 56% with highly variable Fe2O3t and MgO related largely to the ratio of cumulus plagioclase to other minerals. Cumulus oxide mineral OAGNs and OGNs have SiO2 contents below 45%, with Fe2O3t reaching up to 40% and MgO up to 11%, depending largely on the ratio of oxide minerals to pyroxene. The cumulus apatite OAGNs have P2O5 up to 4%. The major- and trace-element composition of massif anorthosites and associated rocks are discussed separately followed by a discussion of the modal and normative mineralogy of anorthosite, leucogabbro, gabbro, and OAGNs or OGNs.

MAJOR ELEMENTS IN MASSIF ANORTHOSITE AND ASSOCIATED ROCKS

The percentage of plagioclase versus other minerals in the anorthosites, leucogabbros, gabbros, and OAGNs or OGNs, rocks with little or no potash feldspar, can be estimated from roughly linear plots when plagioclase compatible elements, such as Na2O, Al2O3, and Sr, or plagioclase incompatible elements, such as MgO, FeO, and Sc, are plotted together. One of the most linear plots of these rocks is the Na2O-Al2O3 plot (Fig. 5) where both Na2O and Al2O3 concentrate almost entirely in plagioclase. The concentration of CaO is not helpful in estimating the relative abundance of plagioclase because it is a major constituent of both plagioclase and augite, which typically have inverse abundances, and remains nearly constant in rocks with a mineralogy dominated by plagioclase and pyroxene. The plagioclase components (Na2O and Al2O3) correlate negatively with MgO and Fe2O3t, as well as other components that concentrate in mafic and oxide mineral phases. Nearly linear plots can also be formed by elements incompatible with both plagioclase and the pyroxenes, although they are much more susceptible to scatter caused by variations in accessory mineral content.

Several attempts have been made to derive a differentiation trend for massif anorthosite and associated rocks to test for possible comagmatism of the various rock types. Buddington (1972) proposed a differentiation trend for magmas parental to these Adirondack rocks that produces increasing concentrations of MgO and FeO and decreasing concentrations of SiO2 and K2O with magmatic evolution, which is only partially consistent with a tholeiitic differentiation trend where MgO should decrease as FeO increases. A study of Adirondack anorthosites by Ashwal (1982) yielded a crystallization sequence of plagioclase first, followed by one or two pyroxenes, the oxide minerals, and finally apatite. Because of the abundance of late-crystallizing oxide minerals and apatite, Ashwal (1978, 1982), Ashwal and Seifert (1980), and McLelland et al. (1994) regarded the FeO-rich mafic dikes and mafic or ultramafic layers cutting anorthosite series rocks as late-stage mafic magmas from which cumulus plagioclase separated earlier to produce anorthosite. All of these mafic dikes and layers have an OAGN composition according to our classification. Several of the jotunite samples collected by P.R.W. are also from dikes. By contrast the mafic rocks along the margins of anorthosite masses are typically gabbros and leucogabbros without modal olivine and are regarded by Buddington (1939) as representing the parental magma from which plagioclase crystals formed and accumulated to produce anorthosite. When anorthosite suite samples are plotted on an AFM diagram only the gabbros plot in a position expected for tholeiitic, or for a few samples, slightly calc-alkaline magma (Fig. 6A). However, small amounts of cumulus plagioclase could move tholeiitic magma down into the calc-alkaline field blurring the distinction between tholeiitic and calc-alkaline. The effect of cumulus plagioclase is shown by the position of cumulus plagioclase-rich leucogabbros and anorthosites that extend downward from the gabbros. The gabbros plot as two distinct groups; the low MgO gabbros and high MgO gabbros with the possibility that additional sampling would have identified intermediate samples and they represent a continuous group. The OAGNs and OGNs show considerable iron enrichment and plot above the gabbros near the F corner suggesting they contain cumulus oxide minerals. When mangerite suite rocks without cumulus plagioclase are plotted on an AFM diagram, jotunites overlap the low MgO gabbros with monzodiorite, mangerite, and charnockite extending downward toward the A corner of the diagram in the manner expected from the differentiation of a tholeiitic magma (Fig. 6B) similar to the high- and low-MgO gabbros.

Most of the compositional spread among anorthosites and leucogabbros is caused by variable amounts of cumulus plagioclase relative to other minerals: pyroxene, oxide minerals, and apatite. One of the most useful ways to evaluate mineralogical variations is to plot MgO against other major-element oxides since MgO concentrates entirely in pyroxene, which is seldom a cumulus mineral in most of these rocks. A plot of MgO relative to TiO2 in the anorthosites, leucogabbros, gabbros, OAGNs and OGNs produces a positive correlation (Fig. 7A) with the OAGNs forming a distinct high TiO2 trend produced by a high abundance of ilmenite and a gap produced by the difference in MgO content between low and high MgO gabbros. The low MgO gabbros (MgO = 2.45%–3.06%) follow a high TiO2 trend with the OAGNs whereas the high MgO gabbros (MgO = 4.30%–7.34%) form a separate low TiO2 trend. The two MgO trends may represent anorthosites with intercumulus pyroxene and scarce oxide minerals (high MgO) and those with less pyroxene and more abundant oxides possibly trapped liquid in part (low MgO). The latter also tend to have more sodic plagioclase and thus are probably later in the differentiation sequence. The MgO-P2O5 plot (Fig. 7B) revealed that a small portion of the samples we thought were OAGNs lacked the necessary apatite and had to form a separate rock type we have called OGNs. The diagram is similar to the MgO-TiO2 plot except that the high P2O5 trend includes the OAGNs, whereas the low P2O5 trend includes the P2O5-poor OGNs. Most of the anorthosites, leucogabbros, and the high-MgO gabbros follow the low P2O5 trend because of their low apatite contents.

TRACE ELEMENTS IN MASSIF ANORTHOSITE AND ASSOCIATED ROCKS

Trace elements are far more sensitive to slight differences in the stage of magmatic evolution than major elements because trace elements often vary over several orders of magnitude during magmatic evolution, whereas major elements typically vary less than one order of magnitude. Consequently, if these rocks have formed from multiple magmas at varying stages of magmatic evolution, trace elements might show differences in rocks with similar major-element compositions. Trace elements that are readily accepted into the crystal structure of a mineral (Dxyl/liq > 1) are considered compatible with that mineral, and as crystallization continues new minerals form with structures compatible with different trace elements. Ashwal (1978, 1982) found that plagioclase was the first mineral crystallized in the Adirondack anorthosite followed by one or two pyroxenes. The only compatible trace element for plagioclase is Sr, whereas the compatible trace elements for pyroxene(s) include Cr, Ni, Sc, V, and Co. As crystallization proceeds, the oxide minerals magnetite and ilmenite eventually crystallize and trace elements Cr, Ni, V, Hf, Ta, and Zn, along with some of the LREEs, become compatible with their crystal structures. Apatite crystallizes even later in the evolutionary sequence and concentrates the REEs, which are compatible with the apatite structure. Finally in some rocks zircon crystallizes and strongly concentrates zircon-compatible elements Zr and Hf. Ilmenite can accommodate some Zr, which may be expelled during metamorphism to form metamorphic zircon as rims or separate crystals—possibly one reason for the large spread in zircon ages for anorthosites. Once a trace element becomes compatible with a crystallizing mineral, its concentration in the residual magma decreases rapidly. Thorium is the most highly incompatible element determined and must have remained in the final residual magma because of its low concentration in anorthosite and all associated rocks.

Compatible trace elements concentrate in minerals with crystal structure sites favorable for their size and charge. The two highly compatible trace elements Cr and Ni (Dxyl/liq >> 1) are not abundant in anorthosite suite rocks where they concentrate in the mafic minerals that characterize our mafic rocks. The low values of Ni in these rocks (Table 3) suggest that olivine crystallized before the emplacement of these rocks and removed some Ni. The vast majority of these rocks have <50 ppm Ni whereas primary basaltic magmas in equilibrium with mantle peridotite typically are expected to have Ni concentrations between 400 and 500 ppm similar to primary basalt sample ID1* with 432 ppm Ni (Nye and Reid, 1986). The most compatible trace element in a slightly evolved basic magma with a reduced Ni content is Cr and, to a lesser extent, Co, according to Arth (1976), Baitis and Lindstrom (1980), and Rollinson (1993). The high compatibility of Cr in early formed mafic minerals causes a rapid decrease in the Cr concentration in residual magma as the concentration of incompatible trace elements (Dxyl/liq < 1) slowly increases. Consequently, minerals and rocks crystallized from only slightly evolved parental magmas can have high concentrations of highly compatible elements, such as Cr, and low concentrations of incompatible elements. However, a few gabbro samples, such as the Paul Smiths gabbros (AA-29, AA-29A, AA-29D), have relatively high values of both Cr and Ni and two other samples (AA-63, AA-69) also have high Ni contents. Plotting Cr against SiO2 shows that gabbros and jotunites have higher Cr concentrations than any of our other rocks (Fig. 8). OAGNs and OGNs contain higher oxide mineral contents but less Cr, suggesting that both the gabbros and the jotunites were formed from less evolved magmas than the OAGNs and OGNs. The close relationship between mineralogy and elemental composition for rocks without potash feldspar is well illustrated by a plot of chondrite normalized Sm (SmN) against LaN/SmN and LaN/EuN (Fig. 9A) where all rock types lie along continuous curvilinear trends from anorthosites at low SmN values to OAGNs and OGNs at high SmN values. Both trends can be explained almost entirely by the mixing of two mineral end members, plagioclase and a generalized mafic component (Fig. 9B) that represents a mix of pyroxene, oxide minerals, and apatite of the appropriate composition. The scatter of data most likely relates to the variation in mineral compositions in these rocks.

The highly incompatible trace elements (Dxyl/liq < < 1) concentrate in residual magma as crystallization of minerals with incompatible crystal structures continues. Only very late in the crystallization sequence will trace minerals crystallize (Ashwal, 1978, 1982) that might provide compatible crystal structures for many of these elements, and some very highly incompatible elements, such as Th, may end up concentrating along grain boundaries. One such trace mineral is zircon, which forms very late in the crystallization sequence from the residual magma and has crystal structure sites compatible for Zr and Hf. A plot of Hf versus Zr (Fig. 10A) has limited scatter if sample AA-151, with a very high Zr concentration of 1511 ppm but only 10.8 ppm Hf, is not plotted. Sample AA-151 is the only exception to the linear trend between Zr and Hf and might be explained by determination of Hf by INAA in a portion of sample powder with little or no zircon, whereas determination of Zr by XRF in another portion of the powder contained more zircon. The erratic distribution of trace minerals in separate portions of even the same powder is always a potential hazard. As previously noted these data indicate that neither OAGNs nor OGNs contain much of the highly incompatible trace elements that concentrate in residual magma. Highly incompatible element Rb follows K2O (Fig. 10B) in all rock types, largely in the feldspar structure, with the highest values for both elements in the charnockites which contains the most potash feldspar. The gabbros have less K2O and Rb than other rocks without cumulus feldspars because they typically have no potash feldspar.

Early anorthositic cumulates with intercumulus pyroxene are enriched in highly compatible trace elements while later crystallizing oxide- and apatite-rich rocks have elevated incompatible trace-element contents and little of the highly compatible trace elements. Consequently when a highly compatible element such as Cr is plotted relative to an incompatible element such as Sm, a hyperbolic plot is formed that best illustrates the wide range of trace-element compositions in the anorthositic rocks (Fig. 11). Originally the Sm-Cr diagram was explained by crystallization from two distinct parental magmas by Haskin and Seifert (1981): a high REE parental magma that produced the high Sm and low Cr samples, and a low REE parental magma that produced the low Sm and high Cr samples. The actual situation must be considerably more complex since samples have been taken from several anorthosite plutons crystallized from several parental magmas. It seems more likely that much of the data spread in this diagram is caused by the formation of rocks from a diverse set of basic parental magmas crystallizing at differing stages of magmatic evolution. Rocks with cumulus plagioclase, such as anorthosites and leucogabbros, are plagioclase-rich so they have both low Sm and low Cr and concentrate at the corner of this diagram. The gabbros and jotunites form relatively continuous suites that extend along both axes showing crystallization from a wide range of evolving magmas. The low MgO gabbros plot along the Sm arm (>10 ppm Sm and low Cr) and the high MgO gabbros plot along the Cr arm (<5 ppm Sm and variable Cr) of the diagram. All other rock types, the OAGNs, OGNs, monzodiorites, mangerites, and charnockites, plot on the high Sm arm because of their enrichment in late crystallizing minerals and incompatible trace elements. In addition to the differing mineralogy of rocks from the two arms of the Sm-Cr diagram mineral separate data reveal compositional differences in minerals separated from the rocks plotting on the two arms (Seifert and Chadima, 1989). Nevertheless none of the high Sm rocks contain much Th, indicating none of our rocks contain the final residual magma that must have escaped to granites now removed by erosion.

Geochemistry of Anorthosite

Anorthosites (plagioclase > 90%) are characterized by high SiO2 (∼53–56 wt%), high Al2O3 (∼24–27 wt%), and high CaO (∼9–11 wt%), along with high Sr (∼700–800 ppm) (Table 1). Normative calculations indicate most of our anorthosites are typically quartz normative, have nearly equal Ab and An percentages, and total normative feldspar contents from 90.5 vol% to 97.2 vol%. Anorthosite Mg# values range widely from 18 to 58 and correlate positively with Cr content. A rough positive correlation between %An and Mg# for anorthosites (Fig. 12A) indicates the anorthosites have crystallized from a parental magma or magmas with a range of evolutionary stages. Perhaps the scatter in Figure 12A is caused by compositional differences among the various parental magmas. A similar scatter is observed on the K2O versus %An diagram (Fig. 12B). If variations in mineral compositions are caused largely by crystallization from multiple parental magmas crystallizing at varying stages of magmatic evolution, it should be possible to eliminate the problem by studying compositional variations among subsamples from a single large sample (Haskin et al., 1981).

Considerable REE and other trace-element data have been collected on massif anorthosite and associated rocks over the past few decades. Massif anorthosites have a composition approaching that of plagioclase, exhibiting low rare earth element (REE) concentrations, high lanthanum/ytterbium (La/Yb) ratios, and large positive europium (Eu) anomalies. A gradual enrichment in REE and other incompatible elements typically occurs as the percentage of mafic minerals increases. Typical anorthosite REE concentrations have been established for anorthosites by Green et al. (1972), Goldberg (1977), Seifert et al. (1977), Seifert (1978), Simmons and Hanson (1978), Ashwal and Seifert (1980), Haskin et al. (1981), and Goldberg (1984). However, depleted Adirondack anorthosite sample AA-194 is more similar in trace-element content to Archean anorthosites (Henderson et al., 1976; Ashwal et al., 1983). Although Ba and Sr exhibit a rough positive correlation Rb does not appear to correlate with either in the anorthosites. Rb and Ba values are highly variable in the anorthosites with two of the high Rb samples having almost the lowest Ba (AA-272 has only 178 ppm) and highest Ba (AA-40 has 502 ppm). High Rb sample AA-58 has the highest Sr concentration (873 ppm) while high Rb sample AA-272 has the lowest Sr concentration (626 ppm). But Rb contamination is a potential problem (Ashwal and Wooden, 1983a) and, on the basis of high Rb values (>20 ppm), four anorthosite samples (AA-40, AA-58, AA-272, and ETA02A) appear to be contaminated. Sample AA-40 was collected from the margin of an anorthosite mass, but sample AA-58 is within the Jay dome and sample AA-272 is within the Westport dome, both at sites well away from exposed margins of anorthosite masses. The geophysical roots below all three samples (Mann and Revetta, 1979) suggest there are no other rocks immediately below any of these three samples, but perhaps erosion has removed rocks previously above, but close, to these samples.

Geochemistry of Leucogabbro and Leuconorite

The leucogabbros and sodic leucogabbros contain between 70% and 90% normative plagioclase and are enriched in pyroxenes, oxide minerals, and sometimes apatite relative to anorthosites. Samples are classified as leucogabbros if the normative plagioclase An content is greater than An40 or sodic leucogabbro if plagioclase An is less than An40. These rocks typically occur along the margins of more massive anorthosite or within anorthosite masses where they often form heteogeneous patchs or lenses mixed with anorthosite. Many outcrops contain both anorthosite and leucogabbro in varying proportions and geometric configurations. In a manner similar to the anorthosites, the leucogabbros show evidence of crystallizing from magmas at variable stages of evolution when Mg# is plotted relative to %An (Fig. 13). The leucogabbros have silica contents between 50% and 56% similar to anorthosite, but have lower Al2O3 and CaO contents. CaO values can be influenced by scapolite alteration or contamination by thin calcite veins (Morrison and Valley, 1991). The leucogabbros display large variations in trace-element concentrations, ranging from low concentrations similar to anorthosite in less mafic varieties to higher concentrations similar to gabbros in more mafic varieties. As a result of their large compositional variation they plot along both arms of the Sm-Cr diagram (Fig. 11) outward from the anorthosites in the corner of the diagram toward gabbros. Anorthosites and leucogabbros have a similar range of Mg# values but Cr concentrations can be far higher in leucogabbros, over 200 ppm, because of higher pyroxene contents. Also Fe2O3, MgO, TiO2, Co, Ni, and Co values are higher in these more mafic rocks whereas Sr and Ba, concentrated in plagioclase, are lower. Three samples of leucogabbro (AA-20, AA-26, and AA-149) have Rb contents over 30 ppm and one sample (AA-245) has an Rb content of 61.9 ppm, along with Ba and Sr contents of 3205 ppm and 3040 ppm, and may be contaminated. The first three of these samples were collected from the margins of anorthosite masses adjacent to other rock types where contamination would be easily possible. Sample AA-245 has an unusually coarse texture suggesting late crystallization with residual fluids present. One incompatible element-enriched leucogabbro sample (AA-151) contains greatly elevated concentrations of Zr (1511 ppm), Hf (10.8 ppm), Th (4.24 ppm), La (48.8 ppm), Ce (124 ppm), and Sm (16.3 ppm) relative to all other samples and must contain zircon. Leucogabbro sample AA-148 also contains slightly elevated Zr (342 ppm) and Hf (7.49 ppm) which suggests it also contains zircon. Sample AA-151 also has very low concentrations of Cr, Co, and Ni, a low Mg#, and a high normative %Or, suggesting it formed from an evolved parental magma. Despite some isotopic contamination (Ashwal and Wooden, 1983a; Morrison and Valley, 1988a) and the location of AA-151 relatively close to the southern margin of the Marcy anorthosite massif, it has a relatively low concentration of Rb (10.1 ppm), which would seem to eliminate contamination as a major contributor to its unusually elevated incompatible element concentration.

Geochemistry of Gabbro and Norite

Adirondack mafic rocks collected largely from the mafic margins of anorthosite masses and some mafic dikes and layers studied by Ashwal (Ashwal, 1978, 1982; Ashwal and Wooden, 1983a) from the interior of anorthosite masses exhibit a wide compositional range from typical gabbro (Le Maitre, 1976) to OAGN (Owens and Dymek, 1992) or OGN. A typical igneous rock of gabbroic composition has a SiO2 content between roughly 46 and 52 wt%, TiO2 below 2 wt%, Fe2O3t below 20 wt%, and P2O5 below 1 wt% (Le Maitre, 1976). Almost all Adirondack gabbros have TiO2 contents over 2 wt%, and often over 3 wt%, whereas TiO2 contents of greater than 4 wt%, up to over 12 wt%, are typical for OAGNs and OGNs (Table 1). The two major criteria used to classify mafic samples as gabbro are normative plagioclase contents between 50% and 70% and a SiO2 content between roughly 45% and 52%. Using the SiO2 content as a major criterion causes some samples (AA-123 and ADK-948B) that would be classified as OAGNs because of their high P2O5 contents by Owens and Dymek (1992) to be listed with our gabbros. Adirondack gabbros can be subdivided into two groups of low MgO (2.45%–3.06%) gabbros and high MgO (4.30%–7.34%) gabbros on an AFM diagram (Fig. 6) and the high MgO gabbros might be considered candidates for tholeiitic parental magmas to anorthosites and leucogabbros. All of the low MgO gabbros and about half of the high MgO gabbros plot in the tholeiitic field above the I&B curve as defined by Irvine and Baragar (1971), the other half of the high MgO gabbros plot in their calc-alkaline field. All of our gabbros except one can be considered tholeiitic if the separation between the tholeiitic field and the calc-alkaline field for the Adirondack rocks is defined by another curve (Adir) placed to fit our high MgO gabbro data. The other complicating factor is that these rocks represent derivative magmas and not the true parental magmas. Any change from the original magmas would probably reduce the Mg component by settling of mafic minerals and increase the A component by reaction with crustal or subcrustal rocks. There is also the possibility that our high MgO gabbros could evolve into low MgO gabbros with increasing FeOt and decreasing MgO and that intermediate gabbros could have been revealed by additional sampling. Most of the high MgO gabbros more closely resemble leucogabbros in trace-element concentrations, whereas the low MgO gabbros more closely resemble OAGNs and jotunites in P2O5, TiO2, and La concentrations. The higher abundance of P2O5 and normative apatite, in low MgO gabbros suggests they may have some cumulus apatite, which would also account for the higher abundance of REEs in this rock type.

Gabbros have high concentrations of compatible trace elements and exhibit wide compositional variations over short distances, suggesting that complex magmatic segregation processes occurred locally during emplacement. Many of our mafic samples, including the Paul Smiths samples and the McCauley Pond samples, were collected from along or near the NE margin of the massive St. Regis anorthosite pluton. The four Paul Smiths mafic samples (AA-29, AA-29A, AA-29D, and AA-29F) were taken from a series of small outcrops along the south side of NY route 86 between the towns of Paul Smiths and Gabriels in the Saranac Lake quadrangle (Fig. 3). Compositionally the Paul Smiths mafic samples range from gabbro AA-29A (east end of outcrop) and AA-29D to sodic gabbro AA-29 to OAGN AA-29F (west end of outcrop) that plot on both arms of the Sm-Cr diagram (Fig. 11). Gabbros AA-29, AA-29A, and AA-29D have relatively high Cr concentrations between 131 ppm and 190 ppm, although only sample AA-29 also has a relatively high Ni concentration (104 ppm). The high Cr content of the high Cr-arm gabbros can be largely attributed to Cr-rich magnetite and augite. Samples AA-29A and AA-29D have compositions considered typical for high MgO gabbro and similar to sample R1017 analyzed by Buddington (1939, table 7, p. 36) and described as a norite band in anorthosite from the same location. Sample AA-29D was found to be free of crustal contamination by Ashwal and Wooden (1983a) and sample AA-29A has a similar low Rb content. Samples AA-29 and AA-29F have much higher Rb contents, 39.4 ppm and 30.3 ppm, respectively, and may be contaminated from adjacent rocks. AA-29, AA-29A, and AA-29D have similar REE patterns whereas AA-29F is greatly enriched in the REE and has a large negative Eu anomaly (Fig. 14). The calcic plagioclase (Or7.5,An43.6) from high Cr-arm gabbro AA-29D has low Rb (19.1 ppm) and Ba (458 ppm) relative to more sodic plagioclase (Or28.2,An18.9) in high Sm-arm OAGN AA-29F with 51 ppm Rb and 1580 ppm Ba. The mafic minerals follow the same trend, with the high Cr-arm mafic separate from AA-29D having 540 ppm Cr and only 2.73 ppm Sm relative to the high Sm mafic separate from AA-29F with 0.02 ppm Cr and 14.7 ppm Sm (Seifert and Chadima, 1989).

The four mafic samples (AA-121, AA-122, AA-123, and AA-127) collected from a hill approximately 1 km NE of McCauley Pond south of the Paul Smiths samples also vary widely in composition. Two of these samples (AA-121 and AA-122) have been classified as OAGNs, one sample (AA-127) is an OGN, and one (AA-123) is a gabbro. Buddington (1939, table 17, p. 66) analyzed a metagabbro (R1032, his # 69) from between McCauley Pond and Lake Colby which resembles McCauley Pond OAGN samples AA-121 and SL01. Most other mafic samples are from north and east of Lake Placid in the extreme NE portion of anorthosite exposures. Samples from the Woolen Mill and Brown Point sites, WM and BP on Figure 3, also range from gabbro to OAGN in composition, while Ashwal's widely separated dikes are all OAGNs by our classification. Two of the Brown Point samples, AD91–951C and AD91–951D, have Rb contents over 30 ppm and may be contaminated, whereas all of Ashwal's dike samples have low Rb contents and do not appear to be contaminated. Several Woolen Mill mafic rocks were collected from along a stream roughly 2 km west of Elizabethtown at a site first described by Kemp and Ruedemann (1910) on the western edge of the Westport dome. Samples MG09, ETS11, and ETS21 have a fine granular texture and sample ETS24 is coarser with a relict igneous texture. One Woolen Mill sample (AD91–948B) is similar to the Woolen Mill gabbro analyzed by Buddington (1939, table 15, analysis 64-L), but has lower MgO and higher Al2O3. The three Brown Point mafic rocks were collected from the southern margin of the Keeseville dome at the eastern edge of an anorthosite outcrop in the Willsboro quadrangle on the shore of Lake Champlain. Buddington (1939, table 7, p. 36) analyzed a coarse mafic anorthosite (R1020) from 2 km SSW of Brown Point that is more TiO2 rich than any of our Brown Point rocks. The other Adirondack mafic rocks and the mafic dikes and layers (Ashwal, 1982) were collected from widely separated parts of the Marcy massif.

Geochemistry of OAGN and OGN

Mafic rocks with an SiO2 content below ∼45% and high concentrations of iron, titanium, and phosphorus associated with massif anorthosite are called OAGNs (oxide-apatite gabbronorites), or just OGNs (oxide-rich gabbronorites) if phosphorus is low. OAGNs, as defined by Owens and Dymek (1992) and reviewed by Dymek and Owens (2001), require both abundant oxide minerals and abundant apatite. OGNs are similar to OAGNs but are oxide mineral rich and apatite poor. The OAGNs have an average P2O5 content of 2.08% with 4.75% normative apatite, whereas the OGNs have an average P2O5 content of 0.3% and a normative apatite content of 0.67% (Fig. 15), intermediate rocks appear to be absent, TiO2 is similar in both rock types. OAGNs were not initially defined to require a SiO2 content below ∼45% but we have introduced this restriction to separate them from the low MgO gabbros with high TiO2 and high P2O5 contents. Both OAGNs and OGNs are characterized by abundant oxide minerals relative to plagioclase and pyroxene and plot near the Fe corner of an AFM (A—Na2O+K2O, F— FeOt, M—MgO) diagram above the gabbros (Fig. 6) indicating they contain cumulus oxide minerals. Although our OAGN and OGN samples are frequently associated with gabbros at the margins of anorthosite masses, they have mineralogies and compositions similar to the mafic-rich dikes studied by Ashwal (1978), Ashwal and Seifert (1980), and Ashwal (1982) from the interior of Adirondack anorthosite masses. Both OAGNs and OGNs have highly variable mineralogies that produce highly variable compositions compared to all of our other rock types with SiO2 ranging between 27.16 and 45.34 wt%, TiO2 between 3.51 and 12.70 wt%, Al2O3 between 2.49 and 15.02 wt%, Fe2O3 between 16.75 and 41.17 wt%, and P2O5 between 1.36 and 3.66 wt% (Table 1). The OGNs appear similar to OAGNs but have lower P2O5 (0.04%–0.65%), REE, and other incompatible element contents and have higher Cr, Ni, and Mg# values, suggesting they crystallized earlier than the OAGNs, perhaps before apatite started to crystallize. Both OAGNs and OGNs have highly variable normative feldspar contents (9.1%–57.5%) and normative feldspar compositions (18.9–74.7%An) and both rock types may contain cumulus oxide minerals although some samples show little textural evidence for cumulus minerals. A high oxide mineral content lowers Mg# because MgO concentrates in pyroxenes whereas iron concentrates mainly in the oxide minerals. Because these mafic rocks are rich in oxide minerals, their Mg# is typically lower than the gabbros with OAGNs averaging Mg# of 30.5 and OGNs averaging Mg# of 40.2, while the gabbros have Mg# up to 48. Paul Smiths OAGN sample (AA-29F) has an Mg# of 23.4, whereas spatially associated Paul Smiths gabbro samples (AA-29A and AA-29D) have Mg#s of 45 and 47, respectively, and Paul Smiths sodic gabbro sample (AA-29) has an Mg# of 39. These differences may be attributed to gravitational mobility of heavy minerals during cooling of these mafic enclaves or pockets of magma interspersed with pockets of solids or even reintroduction of a similar magma at a different stage of magmatic evolution. Or perhaps more detailed collecting would have located intermediate varieties. Further study of these mafic enclaves by detailed textural studies accompanied by dating of textural varieties in the outcrops might distinguish between the various possibilities.

Trace-element compositions also vary widely although they are typically enriched in incompatible (D < 1) elements causing almost all of the OGNs and OAGNs, including the dikes, to plot on the high Sm arm of the Sm-Cr diagram (Fig. 11) and have high REE concentrations similar to OAGNs from the St. Urbain and Roseland anorthosites (Dymek and Owens, 2001). Apatite, or iron-rich olivine, is typically the last mineral to crystallize and it would seem that the abundance of apatite, and P2O5, indicates the OAGNs formed from more evolved magmas than did the OGNs. As expected from the high DSm value for apatite (Watson and Green, 1981), the normative apatite versus Cr plot for anorthosite suite rocks is nearly identical to the Sm-Cr plot. Despite having high incompatible element (Dxyl/liq < 1) contents none of these rocks are enriched in the highly incompatible (Dxyl/liq < < 1) elements, such as Th. The high apatite OAGNs have low Th concentrations compared to other rock types (Fig. 16), indicating that, despite their high concentration of late forming apatite, they do not contain the final residual magma component or were derived from a parent magma initially very low in Th, although such continental magmas are quite rare. But both OAGNs and OGNs are enriched in Zn (145–528 ppm) and Sc (24.2–65.1 ppm) relative to gabbroic rocks with 87.1–338 ppm Zn and 15–43 ppm Sc although OAGNs have less of the highly compatible elements Cr (0–35.3 ppm) and Ni (0–40 ppm) and OGNs have 5–74 ppm Cr and 24.2–70 ppm Ni. OAGN sample AA-29F, rich in K-spar and Rb, but with only 16.3 ppm of Cr, is only separated from gabbro samples AA-29, AA-29A, and AA-29D with high Cr concentrations (131–190 ppm) by a horizontal distance of several meters without an obvious separation boundary, suggesting these gabbros and OAGNs are comagmatic. Some studies regard OAGNs to be comagmatic with anorthosite series rocks in the Adirondacks (Ashwal, 1982; McLelland et al., 1994) and in the Laramie anorthosite complex (Goldberg, 1984). Ashwal (1982) and McLelland et al. (1994) indicate that the Adirondack OAGNs are formed during middle to late stages of magmatic evolution of the anorthosite suite based on the range of coexisting pyroxene compositions and the intrusion of OAGN dikes and layers into anorthosites. Still the comagmatism of OAGNs with other anorthosite series rocks in the various massif anorthosites has been controversial with Owens et al. (1993) and Dymek and Owens (2001) providing evidence they can form at several stages in the evolution of some Quebec anorthosites and are sometimes spatially and genetically related to mangerites and charnockites.

Geochemistry of Jotunite

Jotunites are typically gray-green to rusty, commonly fine-grained mafic rocks outcropping adjacent to and intrusive into massif anorthosite; several jotunites, e.g., ETS49, are from dikes within anorthosite, and often contain andesine crystals and/or anorthosite blocks. As a result the jotunite analyses may have an enhanced plagioclase component in their normative mineralogy. Despite some samples having a plagioclase component, the jotunites plot with gabbros (Fig. 17) and split into two distinct compositional trends with the gabbros on P2O5 diagrams. Most jotunites plot with the low MgO gabbros although a few plot with the high MgO gabbros. They are characterized by SiO2 contents similar to gabbros (46%–53%); K2O from 1% to 3%; MgO from 2% to 8%, although high MgO samples AF320, AF339, and ETS14 may contain cumulus orthopyroxene; normative quartz less than 10%; normative Or between 5% and 15%; and total mafics greater than 20%. The composition of all jotunites is roughly similar (Table 1) despite their occurrence as spatially separated patches and dikes. Most of these rocks were collected by our third author (P.R.W.) and their classification is based on field appearance and thin section evaluation as well as composition. In order to compare jotunites with other noncumulus rocks we have plotted them together. On a K2O versus SiO2 plot (Fig. 18A), gabbros, jotunites, monzodiorites, mangerites, and charnockites form a nearly linear trend. When MgO is plotted relative to SiO2 (Fig. 18B) gabbros and jotunites exhibit high MgO values relative to other noncumulus rock types that form a continuous spectrum of change at lower MgO concentrations. When all of our rocks are plotted together (Fig. 19), it is interesting to note that the jotunites plot with anorthosite, leucogabbro, gabbro, and OGN and OAGN rather than with monzodiorite, mangerite, and charnockite although they roughly occupy the intersection of these two groups of rocks.

Geochemistry of Monzodiorite, Mangerite, and Charnockite

Average SiO2 contents increase from an average value near 50% in jotunites to 54% in monzodiorites, 59% in mangerites, and 66% in charnockites, which have SiO2 contents to over 71%. In a similar manner average K2O increases to 3.3% in monzodiorites, 4.8% in mangerites, and 5.4% in charnockites. Decreasing MgO values from monzodiorite to charnockite indicates decreasing mafics just as decreasing %An and increasing %Or reveals increasing orthoclase in these rocks. Thus overall the average values of SiO2 and K2O increase while TiO2, Fe2O3t, MgO, CaO, P2O5, and Mg# decrease from monzodiorite to mangerite to charnockite. The compositional changes are relatively smooth from gabbro and monzodiorite through charnockite for elements such as CaO and K2O versus SiO2, although they were not as smooth for MgO and P2O5. Charnockites are distinguished from mangerites by having over 10% normative quartz (average 18.5%) relative to only 5.5% normative quartz in mangerites.

Spider diagrams of the felsic rocks relative to primitive mantle (P-Mantle) typically show increasing concentrations from the more compatible (right) side of the diagram to the more incompatible (left) side of the diagram with elemental concentrations varying between 20 and 200 times primitive mantle but at lower concentrations than residual rhyolite LWB-102 (Fig. 20) from the Midcontinent Rift. Large negative anomalies for Nb, Ta, Ti, and Th are common as are Sr and P in charnockites. A negative Sr anomaly is generally attributed to either a source that retains some residual plagioclase or the loss of plagioclase from a crystallizing magma prior to emplacement. Negative Ti and P anomalies have similar ramifications for oxide minerals and apatite. In many instances Ba, K, Zr, and Hf show the highest concentrations for the samples with the REE being almost as abundant. Nb and Ta always form negative anomalies but the size of the anomaly ranges widely. Samples from the Thirteenth Lake area have distinctly smaller concentration variations between elements on spider diagrams and smaller negative anomalies than rocks from other collecting areas although the same elements have negative anomalies. The Thirteenth Lake sample patterns also tend to be less steeply inclined. Samples from a large mangerite body extending west from the anorthosites and not adjacent to any outcrops of massif anorthosite do not show negative Th anomalies like mangerites from other collecting areas, but are otherwise similar. Overall the associated acidic rocks from all collecting areas have similar spider diagram patterns.

None of the associated felsic rocks are regarded as representing extreme residual magmas similar to some granites and rhyolites (Fig. 20) because of their relatively low SiO2 and trace-element concentrations. The much higher trace-element concentrations observed in high-silica igneous rocks from other continental igneous provinces, such as rhyolite sample LWB-102 from the Midcontinent Rift (Brannon, 1984), suggests that it represents a residual magma. This rhyolite has a positive εNd value of +1.04 and exhibits large negative Sr, P, and Ti anomalies compared to the Adirondack felsic rocks, indicating it represents a highly differentiated mantle magma with only limited interaction with the crust or subcrustal mantle. However, even LWB-102 has had some interaction with crustal rocks because it shows small negative Nb and Ta anomalies and a low εNd value relative to depleted mantle. Despite the interaction, the rhyolite lacks the typical negative Th anomaly characteristic of our felsic rocks. Perhaps the residual granites or rhyolites associated with Adirondack anorthosites and associated rocks have been removed by erosion.

CONCLUSIONS

All AMCG rocks in the Adirondacks have mixed tholeiitic and calc-alkaline compositional characteristics (Table 4) with major elements showing tholeiitic iron enrichment and trace elements showing HFSE Nb and Ta depletion, supporting the complex petrogenesis indicated by isotope studies. Anorthosite suite rocks follow a strong iron enrichment trend on an AFM diagram whereas mangerite suite rocks spread along the AF side of the AFM diagram. The close spatial association of gabbros with OAGNs and OGNs in mafic enclaves suggests local variations in magma versus solid pockets or reintroduction of similar magma at a different stage of magmatic evolution that might be best explored further by detailed textural studies of such outcrops. AFM plots of jotunites from the mangerite suite overlap gabbro from the anorthosite suite, exhibit two distinct compositional trends similar to anorthosites, and frequently contain inclusions of adjacent anorthosite or plagioclase megacrysts, indicating they have a more complex and variable origin than other mangerite suite rocks. Further study of compositionally distinct jotunites at specific localities might explain some of the variations.

Anorthosites, leucogabbros, gabbros, and jotunites exhibit two distinct compositional trends on TiO2-MgO (Fig. 21A) and P2O5-MgO (Fig. 21B) diagrams, low and high TiO2 and low and high P2O5. Only monzodiorite, mangerite, and charnockite plot entirely on the high TiO2 and high P2O5 trends. The split trends are more obvious on the P2O5-MgO diagram where P2O5 measures apatite content than on the TiO2-MgO diagram where TiO2 indicates ilmenite content, perhaps because apatite crystallizes later than the oxide minerals in evolving magma. The separation of rocks with cumulus oxides into high P2O5 OAGNs with apatite and low P2O5 OGNs without cumulus apatite is the basis for separating these two rock types. The rocks that split into two trends are characterized by two distinctive mineralogies, one with plagioclase associated mostly with oxide minerals and apatite and another with plagioclase associated with pyroxene and limited amounts of oxide minerals and apatite. The close spatial association of samples from the two trends at some locations suggests that samples from the two trends represent crystallization from the same or similar parental magmas at different stages of evolution rather than crystallization from geochemically distinct parental magmas.

We are greatly indebted to the New York State Museum and John Hart for allowing us to list and plot their multitude of analyzed samples of massif anorthosite and associated rocks obtained over decades by numerous individuals with the samples the authors have collected and analyzed. We thank Lew Ashwal for contributing several samples (MM-6, MM-13A, MM-14A, SA-3D, SA-5, SR-6, and SR-9) for our study and are indebted to Randy Korotev, Marilyn Lindstrom, Dave Lindstrom, and Rex Couture for help with sample preparation and accurate INAA and XRF analyses. Initial INA analyses were performed by K.E.S. at the Ames Laboratory Research Reactor with the help of Wayne Stensland and Mark Smith. The first author (K.E.S.) thanks Yngvar Isachsen for guiding initial sample collection and field identification of various rock types in the Adirondacks, Percy Crosby for several guided tours of the Lake Placid quadrangle, and Jim Olmsted for guiding sample collection in the Au Sable Forks and Willsboro quadrangles. The first author (K.E.S.) also thanks former Iowa State students Barry Gross and James Dockal for assisting in sample collection (ABM samples) and geologic mapping on Baker Mountain, and Ken Harpole for assisting in the topographic mapping of Baker Mountain. Most of the mineral separations were performed by Sarah Chadima at Iowa State University, whereas Kurt Hollocher at Union College assisted with the ICP-MS analyses. We also would like to thank Bill Kelly from the NY survey for assistance in obtaining maps, names, and data from the Adirondacks and Frank Revetta for supplying gravity maps of the sample collection area. Finally we are indebted to manuscript reviewer Jeff Chiarenzelli and an anonymous reviewer, plus comments from Mike Williams, for corrections and significant improvements to this manuscript.