An intrusion breccia exposed along Roaring Brook on the west face of Giant Mountain in the Adirondack Highlands (eastern USA) has been reinterpreted as a magma conduit. It lacks metasedimentary xenoliths, as previously proposed, and field, petrographic, and geochemical evidence point to an igneous origin for the vast majority of the enclaves. A mechanism involving magma mixing is proposed for the enclaves and explains their textural features (e.g., crescumulate and comb textures and mantled feldspars), compositional layering, and geochemical trends. Rock types ranging from charnockite to pyroxenite occur in the exposed section and document the variety of magmatic processes and compositional variation of magmas composing the Marcy Anorthosite massif. The bulk of the enclaves in the intrusion breccia can be explained by a batch of more felsic magma (charnockite-mangerite?) that intruded into a partially crystallized zone of anorthositic, gabbroic, and dioritic rock. This resulted in disruption, partial mixing and hybridization, and chilling of mafic magma within, and against, the more silicic melts.


In a classic study of the Mount Marcy quadrangle (New York) Kemp (1921) described an unusual and spectacular intrusion breccia exposed in the valley of Roaring Brook on Giant Mountain (Fig. 1, RB) from Roaring Brook Falls at 560 m (Fig. 2) to the water-polished slopes with ∼100% exposure at various intervals. The breccia fragments, or enclaves (Fig. 3), some of which exhibit layering, were interpreted as metasedimentary and/or metavolcanic rocks derived from an ancient basement complex (Kemp, 1921; de Waard, 1970; Jaffe et al., 1983). On 29 June 1963, torrential rains resulted in landslides that further uncovered spectacularly exposed tracts of the water and ice-polished intrusion breccia in the Roaring Brook Valley (de Waard, 1970). Several geologists, including de Waard (1970) and Jaffe et al. (1983), studied the Roaring Brook intrusion breccia and utilized it as a valuable field laboratory for petrology students. Throughout, the layered enclaves were widely regarded as xenoliths incorporated from an ancient basement complex of undetermined age. In a paper that presented evidence for multiple metamorphic events in the Adirondacks, McLelland et al. (1988a) cited the enclaves as consistent with other firmer evidence for more than one period of metamorphism in the region.

One of us (James McLelland), together with students, undertook a thorough petrographic and geochemical study of the Roaring Brook suite of rocks. Although de Waard (1970) produced a number of whole-rock analyses for the host rocks of the breccia, the enclaves were not analyzed. This was in part because the polished condition of the outcrops makes sampling of the enclaves difficult. One of us (Jonathon McLelland) backpacked a portable rock drill up the valley of the Roaring Brook and collected cores of the enclaves that yielded representative thin sections as well as whole-rock analyses (Supplemental Table 11). The results of this research clearly established that the host rocks and enclaves in the Roaring Brook intrusion breccia were all contemporaneous and could be accounted for by processes of magma mixing and commingling in a large and presumably long-lived magma conduit. Furthermore, the study concluded that none of the enclaves were metasedimentary and/or metamorphic xenoliths incorporated from a pre-Grenvillian basement terrain (Bohlen et al., 1992) into the host rocks of the anorthosite-mangerite-charnockite-granite (AMCG) suite. The observations culminating in these conclusions and reinterpretation of the exposures are presented in the following.


As shown in Figure 1, the Adirondacks are divided into two major terranes: the Adirondack Highlands terrane (AHT) and Adirondack Lowlands terrane (ALT). Rocks within the AHT have undergone two major regional orogenic events and associated high-grade metamorphism, the Shawinigan (ca. 1210–1140 Ma) and Ottawan (ca. 1090–1040 Ma). The former affected both terranes, but the latter was largely restricted to the AHT, where it attained granulite facies conditions and is thought responsible for most of the large-scale deformation of the region. The Ottawan was followed by the lower grade Rigolet orogenic event ca. 1000–980 Ma, and together the two represent the Grenvillian orogeny (Rivers, 2008).

Within the ALT, Grenvillian effects are almost exclusively restricted to minor hydrothermal activity. The two terranes are separated by the Carthage-Colton shear zone (CCSZ) (Fig. 1) that dips northwest at 40°–60°. Earlier ductile fabrics along its trace are cut by a system of late extensional shears that dropped the ALT down to the northwest and juxtaposed it against the AHT during collapse of the Grenville orogen ca. 1060–1040 Ma (Selleck et al., 2005). Prior to that event, the ALT was not contiguous with the AHT, but had evolved separately as the eastern margin of the Central Metasedimentary Belt (CMB) island arc (Chiarenzelli et al., 2010, 2015). During the Shawinigan orogeny the CMB collided with the AHT, and the CMB was metamorphosed and thrust eastward over the AHT, where it formed an orogenic lid (Rivers, 2008; McLelland et al., 2010). The lack of high-grade 1090–1050 Ma Grenvillian metamorphism in the ALT is likely related to the elevated position of the ALT in the crust at that time. At the end of the 1090–1050 Ma Ottawan orogeny the ALT was downfaulted to the northwest along the CCMZ and brought into juxtaposition with the granulite facies rocks of the AHT (Selleck et al., 2005; McLelland et al., 2010).

Most of the Adirondack Highlands are underlain by intrusive rocks. The oldest of these consists of a suite of ca. 1360–1300 Ma tonalites, trondhjemites, and granites that were rifted from the southeastern margin of Laurentia shortly after 1300 Ma. These form the basement rocks of both the AHT and Green Mountains of Vermont and are referred to as the Dysart–Mt. Holly suite (Agustsson et al., 2013). During and following the collisional phase of the Shawinigan orogeny, a great volume of igneous rock was emplaced into the highlands, and where it is represented by rocks of anorthositic, mangeritic, charnockitic, and granitic composition it is referred to as the AMCG suite and dated as 1155 ± 7 Ma (McLelland et al., 1988b, 2004, 2010). The AHT was intruded ca. 1060–1050 Ma by the Lyon Mountain Granite (LMG) that rims most of the AHT (in red in Fig. 1). This granite is coeval with the Carthage-Colton shear zone and is interpreted as having been emplaced during extensional collapse of the Grenville orogen (Selleck et al., 2005; McLelland and Selleck, 2011; McLelland et al., 2014).

The rocks discussed herein are well within the Marcy Anorthosite massif (green star in Fig. 1). All of the local units, except for a few narrow supracrustal xenolithic screens, are members of the AMCG suite, dominated by massif anorthosite. The great thickness and strength of the anorthosite buttressed the entire suite against subsequent deformation. Preservation of igneous textures and the polished nature of the exposures make the area ideal for studying relationships in the suite. We begin at the western terminus of the exposures directly above Roaring Brook Falls at an elevation of 500 m. Whole-rock chemical analyses of the principal igneous lithologies in Roaring Brook were given by de Waard (1970) and are also presented in Table 1.

Lower Elevation Exposures in Roaring Brook

At the top of Roaring Brook Falls, excellent exposures of coarse-grained, bluish-gray Marcy facies anorthosite (Fig. 4) occur and consist of ∼90% plagioclase (andesine, An48) accompanied by pyroxene. This facies composes most of the bedrock in the interior of the massif and locally grades into gabbroic anorthosite or leuconorite. In Figure 4 a large andesine megacryst is outlined with white dashes; at the top right of the figure, a mangerite dike containing clinopyroxene crystals exhibits lobate commingling contacts with the anorthosite, suggesting a coeval origin.

Approximately 15 m upstream, and parallel to the brook, a 4–6-m-wide lens-shaped dike of gray, medium-grained monzonite consisting of augite and microperthite intrudes the anorthosite. The dike displays coarse crescumulate texture at its inner contacts, suggesting undercooling against the already solidified anorthosite. The anorthosite country rock consists of 1.5–7.5 cm megacrysts of andesine-labradorite plagioclase aligned at a high angle to the stream and accompanied by ophitic to subophitic orthopyroxene and oxides. None of these minerals are deformed, and their strong alignment is due solely to magmatic flow and compaction processes.

Approximately 200 m upstream, at an elevation of 610 m and still within the anorthosite with its aligned phenocrysts, a second waterfall has a much smaller drop than Roaring Brook Falls. Here, Roaring Brook follows a trough formed by erosion of a dark orthopyroxenite dike (Fig. 5) that reaches a maximum width of 2–3 m and displays many offshoots and splays that tend to parallel the main dike body. Note in Figure 6 that, just as in the case of the monzonite dike, the dike crosscuts the aligned plagioclase flow fabric in the anorthosite and clearly postdates that event. In places the dike displays lobate contacts and mutually crosscutting relationships with the anorthositic rocks (Figs. 6 and 7), indicating a broadly coeval history. Large (meter scale) to small (centimeter scale) xenoliths of pyroxenite derived from the dike occur sporadically within the Marcy Anorthosite (Fig. 7) and confirm their coeval, but overlapping, status. As reported by de Waard (1970) and Bohlen et al. (1992), the dike continues upstream intermittently to an elevation of 800 m (a total horizontal distance of 360 m).

Chemical analyses and thin section modes of the dike (de Waard, 1970) demonstrate that it consists of 78% hypersthene (enstatite, En70), 9% plagioclase (An50), and 8.7% clinopyroxene. The hypersthene in the dike has a very high Mg/Fe ratio (∼0.85) and must have crystallized early and at high temperatures. Possible parental magmas include gabbroic anorthosite and/or leuconorite, which are the most likely candidates for early crystallization of such cumulates (de Waard, 1970). Therefore it is thought that the hypersthene dike represents an intrusion of remobilized cumulate material from norite or leuconorite that, along with minor interstitial liquid, drained down a steep fracture system into largely crystalline anorthosite and/or leuconorite (de Waard, 1970; Ashwal, 1978; Bohlen et al., 1992).

Above the falls the polished outcrops consist mainly of Marcy Anorthosite, but crosscutting gabbroic anorthosite, leuconorite, and leucogabbro facies are also present. Less common, but well represented, are dikes of fine-grained gabbroic anorthosite referred to as Whiteface Anorthosite (type locality on Whiteface Mountain). On the basis of their usual location and chemistry, these rocks are considered to be a marginal facies of the massif, but are occasionally, as in this case, found within it due to the presence of a magma conduit that channeled a wide variety of melts. There is a limited, but important, occurrence of xenolithic mats of extremely coarse grained blue-gray labradorite-andesine anorthosite containing large (decimeter scale) pyroxenes (Emslie, 1974).

These exceptionally coarse grained anorthosite xenoliths have been interpreted as deep-seated, probably subcrustal, cumulates of plagioclase that, at the high pressures at the depths of crystallization (10–12 kbar), became buoyant in the denser parental magma and floated, eventually forming rafts (Emslie, 1974). Subsequent upward intrusion of the cumulate masses tended to disrupt them, so large intact rafts are rare. However, it is not uncommon to find the remnants of the disrupted rafts; a good example, shown in Figure 8, is a few meters upstream from where the streambed levels off. Also present are pods and veins of mafic ferrodiorite (Fig. 9) that are interpreted to have been filter pressed from late differentiates of the anorthositic suite (McLelland et al., 1994).

Upper Elevation Exposures in Roaring Brook

The upper elevation exposures begin at ∼800 m and continue to ∼920 m. They are characterized by spectacular occurrences of water and ice-polished outcrops in the bed of Roaring Brook (Fig. 2). At its base, the upper level exposures consist of Marcy Anorthosite intruded by magmas best described as hybrids between gabbro and charnockite with varying degrees of commingling and mixing (for an extensive discussion, see Bohlen et al., 1992, p. 81–90 therein). One such intrusive contact is shown in Figure 10; a heterogeneous gabbroic anorthosite (unit GA, at the bottom of the photograph) intrudes an expanse of enclave-laden hybrid magmas due to mingling and mixing of mangeritic, charnockitic, and dioritic rocks. Note that the actual compositions of the enclave magmas are difficult to ascertain in the field except by sodium cobaltinitrite staining. The enclaves include not only anorthosite but a variety of other rocks that are discussed herein and that constitute the Roaring Brook intrusion breccia.

As shown in Figures 11A–11F, the breccia contains numerous enclaves of a variety of compositions and is intruded by a wide variety of magma compositions. Further upstream from the base of the exposure, a 6-m-high waterfall flows over a gabbroic body. Above this, and continuing to an elevation of 920 m, the stream flows through a wide expanse of world-class intrusion breccia exposed by ice and water polishing.

Most of the enclaves are shades of gray, and are layered and hybrid in nature (Fig. 11), but black pyroxenite to white anorthositic enclaves are not uncommon. The gray enclaves in Figure 11B are dioritic to monzodioritic and are set in a matrix of hybrid igneous rock. Note that their margins are somewhat irregular, suggesting magma commingling (see following). Most black enclaves consist of pyroxenite and range from orthopyroxenite, which dominates in the downstream exposures, to clinopyroxenite, which dominates exposures discussed here. The pyroxene crystals in the pyroxenite are interlocking and occur together with minor interstitial plagioclase. We interpret these features to reflect an origin as cumulates from mafic magmas.

Large, brecciated enclaves are shown in Figures 11A and 11B and are intruded by mangerite that displays sharp, angular contacts with the rest of the rocks, indicating that the former was solid when intruded. Several clinopyroxenite xenoliths show disruption by the mangerite, an observation consistent with a cumulate origin for the pyroxenite. Figure 11B shows a gray dioritic to monzodioritic breccia intruded by mangerite and, in this case, the brecciated fragments are far less angular and defined than those in Figure 11E. We interpret this to suggest that the diorite was not entirely solidified when intruded by the mangerite. This interpretation is supported by the relationships shown in Figures 11B–11D, where dioritic and hybridized mangeritic intrusives show embayed contacts and evidence of ductile deformation. Relationships of this sort are characteristic of magma commingling (Wiebe, 1980; Philpotts, 1990, p. 261–267 therein; Seaman and Ramsey, 1992), and we interpret the present relationships as products of that process.

The difference in color between enclaves in Figure 11B is due to variation in the abundance of mafic minerals. Ferrodiorite enclaves, often rich in oxides, tend to be darker. Dioritic to ferrodioritic rocks containing orthopyroxene are commonly referred to as jotunites. The dioritic rocks are thought to be residual liquids formed by differentiation of parental anorthositic magmas and filter pressed into the intrusion breccia (Bohlen et al., 1992). Although the quantity of such liquids is small, their compositions match the dioritic rocks in the intrusion breccia, and the genesis of such melts during anorthosite differentiation has been noted by others (e.g., Wiebe, 1994, for the Nain anorthosite complex).

As shown in Figures 11A and 11B, the dioritic magmas undergo partial disintegration that produces numerous satellite enclaves that eventually get carried off into the main hybrid mangeritic magma to form an impressive breccia, such as that in Figure 11A. Note that although some of the larger enclaves appear to be angular in form, close inspection shows many contacts consistent with magma commingling. Some enclaves appear also to be in the process of breaking apart and forming breccia-within-breccia texture, emphasizing the complex and repeating processes required to generate these relationships. Many enclaves in the Roaring Brook breccia are conspicuously layered (Fig. 11E), and some have folded or warped layering (Fig. 11F).

Figure 12 documents the enclaves and provides further evidence supporting the important role played by magmatic interactions in the Roaring Brook breccia. The enclaves in Figures 12A, 12C, and 12E are typical of those that Kemp (1921) and de Waard (1970) regarded as xenoliths of metasedimentary and metavolcanic rocks torn from an ancient, pre-Grenvillian basement by magmas of the anorthosite suite. This is particularly true of Figures 12A and 12E that display well-defined compositional layering. No chemical analyses or petrography had been undertaken on the layers until one of us (James McLelland) and students obtained drill cores of a number of samples in the late 1980s. From these samples they prepared thin sections and chemical analyses. An interesting, and unexpected, finding was that the light colored layers are syenitic to granitic in composition while the dark layers are in the range of gabbro and diorite. This is further substantiated on the outcrop and in slabs by sodium cobaltinitrite staining that distinguishes the yellow K-feldspar of the syenitic-granitic layers from the gray to white plagioclase-rich gabbro and diorite layers (Fig. 12F).

None of the layers give evidence of an origin that is metasedimentary or even metavolcanic, and they lack diagnostic minerals of a metasedimentary origin (e.g., pyroxenes of diopsidic composition, calcite and/or dolomite, graphite, scapolite). Furthermore, orthopyroxene in the mafic layers commonly forms acicular “fence-post” or “comb” (Lofgren and Donaldson, 1975) texture normal to the layering (Figs. 12B, 12D). In Figure 12A acicular fence post orthopyroxene is well developed. A fence-post arrangement is a variety of comb texture and results from undercooling and supersaturation of the mafic magma against the slightly cooler syenitic-granitic material with crystal growth taking place normal to the contact (Lofgren and Donaldson, 1975). In short, these features are the result of magmatic processes involving interaction of mafic and felsic components; none of it qualifies as xenoliths of metasedimentary and metavolcanic pre-Grenvillian basement. Figures 12C and 12D provide further confirmation of these conclusions. The enigmatic white rim (Fig. 12C) on the dioritic enclave is a chill margin within which diagnostic acicular orthopyroxene crystals are arranged in two crisscrossing directions (Fig. 12D).

Textures of this sort have been reported in the Sierra Nevada (B. McKinney, 1990, personal commun.) and are a variation on comb texture that forms during undercooling and supersaturation. Thus, all of the examples shown in Figure 12 are consistent with, and support, the interpretation of the enclaves as magmatic rocks. In this respect, note that in Figure 12E that a possible slump structure is to the upper left of the layered enclave. Given this, it is necessary to determine the magmatic processes that explain and unite all of these observations.

Unifying Mechanism

In some respects the magmatic layering in many of the enclaves is similar to flow layering, but this texture is developed in volcanic rocks when different magmas are squeezed by constrictions within a volcanic neck, and the rocks considered here are not volcanic. However, an important study provided a model for flow layering in deep-seated magmas; the model is due to experiments performed by Ottino (1989). The experiments utilized a fish tank-type container that contained an automated stirring device. In Figures 13A–13D the surface of the liquids is viewed from above the tank. A blue paint, a molasses-like liquid of known viscosity, was placed into the tank (Fig. 13A) and a red liquid of similar composition, but with a slightly different viscosity, was fed into the tank from a source at the upper left top corner of the figure. The red liquid is drawn out into layers in the blue liquid and, because of the geometrical constraints of the container, multiple folding takes place. A small, spherical mass of yellow liquid with a viscosity very similar to that of the blue liquid was also introduced into the tank and, because of the similar viscosities, the yellow sphere mixes readily with the blue matrix to yield a green liquid (Fig 13D).

Two important conclusions emerged from this experiment: (1) liquids of similar viscosities and composition can readily mix with one another, and (2) in order to mix liquids of substantially different viscosities, the assemblage must be stirred, and prior to a homogenization, a stage of interlayering is produced. The second alternative applies to, and is thought to be responsible for, the layered enclaves in the Roaring Brook breccia. The difference in viscosities is due to both composition and, and as indicated by comb layering and fence-post texture, by differences in temperature. This explanation of layered enclaves in the Roaring Brook intrusion breccia is reasonable and consistent with these observations. In nature, the stirring aspect of the experiment is provided by differential flow between different magmas ascending through a common conduit and/or mutually intrusive relations. As discussed in the following, it also accounts for the compositional range of homogeneous enclaves in the intrusion breccia as shown in (Fig. 14).

Figure 14 presents Harker variation diagrams for major elements in the Roaring Brook assemblage; plots are given for the major igneous rocks in the intrusion breccia and for a number of enclaves. Enclaves of cumulate type rocks marked by a low percentage of SiO2 and high FeOtot are represented, as are the more common dioritic enclaves. Nonenclave igneous rocks include anorthosite, jotunite, mangerite, and charnockite. Two samples of the orthopyroxenite dike are shown. It is striking that the enclave compositions define one or two tracks or relatively narrow zones away from the MgO-rich regions and toward mangeritic and charnockitic compositions. We attribute this to the fact that mangerite and charnockite are the most abundant rocks in the intrusion breccia and compositional end members. Therefore the enclaves would have the highest probability of interacting with these magmas and mixing with them. Thus the Ottino stirring and mixing model can explain both the layering in some of the enclaves as well as the hybridization of many of the more homogeneous enclaves.

Figure 15 shows two examples of mantled and reacted feldspars from the intrusion breccia. They show alkali feldspar surrounded and embayed by plagioclase (Fig. 15A) and plagioclase surrounded and embayed and replaced by alkali feldspar (Fig. 15B). These examples closely resemble field and experimental examples described by Stimac and Wark (1992) and Wark and Stimac (1992) and are explicable as the result of feldspars from charnockites having been caught up in, and reacted with, mafic magma, and plagioclase from mafic magma having been incorporated into, and reacted with, charnockite. In short, they are the result of magma commingling and mixing, and powerful proof of the interaction of magma compositions with a wide compositional range.


The valley of Roaring Brook is underlain by water-polished exposures of anorthosite and associated granitoids of an AMCG suite that provide outstanding opportunities to study the nature and evolution of these rocks. Several types of anorthositic rocks are preserved, including rafts containing giant crystals of plagioclase crystallized at depth and transported by later gabbroic and anorthositic rocks to shallower depths where the final phases of crystallization took place. These include the precipitation of layers of orthopyroxene that were mobilized to form crosscutting dikes when the cumulus pyroxenes drained down into fractures in the magmatic system. Near the upper terminus of the valley a spectacular intrusion breccia is exposed with numerous and varied enclaves set in a host that varies from mangerite to gabbroic anorthosite. For decades these enclaves had been interpreted to be xenoliths of country rock of metasedimentary and metavolcanic origin composing an early metamorphosed basement complex. However, careful study of the enclaves reveals their igneous compositions and textures, including crescumulate and comb textures. It is argued that the breccia was formed by intrusion of a new batch of magma (mangerite-charnockite?) into a partially crystallized zone of anorthositic and dioritic rock, and this resulted in disruption and chilling of more mafic magmas against more silicic melts with the production of, for example, fence-post orthopyroxenes and comb textures. The layering that is present in the enclaves is predominantly the result of magma mixing and magmatic flow. Soft lobate margins that occur on many of the inclusions, warping of layers, and partial assimilation of some xenoliths are typical of multiple injections, magma mixing, and comingling and are interpreted in this fashion. Thus, intrusion breccia actually contains few true xenoliths in a strict sense, but is the result of magma mingling and mixing, as well as chilling due to differences in magma temperatures to produce the world-class enclaves observed.

James McLelland thanks Colgate University for financial support. Chiarenzelli thanks the MacAllaster and Tory families and friends for support for his work through the Archie F. MacAllaster and Barbara Torrey MacAllaster Professorship in North Country Studies. We thank Robert Darling and an anonymous reviewer, and the editor, Raymond Russo, for their time and comments and efficient editorial handling of the manuscript.

1Supplemental Table 1. X-ray fluorescence analyses of Roaring Brook enclaves. Please visit http://dx.doi.org/10.1130/GES01260.S1 or the full-text article on www.gsapubs.org to view Supplemental Table 1.