Magma ﬂow within the Tavares pluton, northeastern Brazil ...weinberg/PDF_Papers/...MAGMA FLOW...

For permission to copy, contact Copyright Clearance Center at www.copyright.comq 2001 Geological Society of America508

GSA Bulletin; April 2001; v. 113; no. 4; p. 508–520; 13 figures.

Magma flow within the Tavares pluton, northeastern Brazil:Compositional and thermal convection

R.F. Weinberg*Nucleo de Estudos de Granitos, Department of Geology, Federal University of Pernambuco, Recife, Pernambuco, Brazil,and Institute of Earth Sciences, Uppsala University, Villavagen 16, Uppsala S-752 36, Sweden

A.N. SialR.R. PessoaNucleo de Estudos de Granitos, Department of Geology, Federal University of Pernambuco, Recife, Pernambuco, Brazil

ABSTRACT

Crystallization coupled with gravity re-moval of depleted interstitial melt haslong been recognized as a mechanism ofmagma differentiation. Similarly, heat re-leased by synplutonic basaltic magma in-trusions has long been recognized as ca-pable of driving convection in granitechambers. Direct evidence of these pro-cesses has seldom been described in gran-ites. In the Tavares pluton, we mapped anumber of melt extraction structuresfrom pores of a crystal-liquid mush (aneffectively solid magma where crystalsform an interconnected skeleton) and avariety of flow structures such as (1) me-ter-scale tear- or mushroom-shaped blobsrepresenting within-chamber diapirs; (2)meter-scale ellipsoids representing frozenthermal plumes of granite, driven by heatreleased from disrupted diorite intru-sions; and (3) ladder dikes and snailstructures representing cross sections ofseveral superposed cylindrical magmachannels (possibly feeders of diapirs andplume heads). A fundamental feature ofthe structures in the Tavares pluton isthat they are well delineated by maficschlieren developed at active channelmargins. We postulate a new model forthe origin of marginal schlieren, whichcombines shear flow sorting and melt es-cape from the flowing magma into an ef-fectively solid surrounding mush. Extrac-tion structures (representing melt

*Present address: Centre for Global Metallogeny,University of Western Australia, Nedlands, WA6907, Australia; e-mail: [email protected].

extraction from mush pores into meltpockets) and schlieren (representing re-gions where melt escaped into surround-ing mush pores) are both favored by mag-mas that form an interconnected solidframework at low crystal fractions(;50%), because these mushes are ductileand permeable. Favorable magmas arethose with a high wetting angle betweenmelt and solid (;608) and a propitiouscrystal size and shape distribution. Wepropose a model of compositional andthermal convection that accounts for alldescribed structures.

Keywords: fluid dynamics, granite, magmachambers, magmatic, structures.

INTRODUCTION

The flow of magma within granite magmachambers is fundamental for understandingpluton emplacement, growth, and magma dif-ferentiation. For this reason, magma flow pat-terns and processes have been the center ofconsiderable scientific interest. Chamber flowis controlled partly by magma inflow or out-flow and contemporaneous tectonic deforma-tion (e.g., Hutton, 1988; Ramsay, 1989; Wein-berg, 1997; Archanjo et al., 1998), and partlyby factors related to the pluton’s internal dy-namics, such as magma convection, crystalli-zation, and differentiation (e.g., Martin et al.,1987; Weinberg, 1992; Jaupart and Tait,1995). Structures defined by crystal align-ment, xenoliths, schlieren, and synmagmaticenclaves (e.g., Cloos, 1936; Barriere, 1981;Abbott, 1989; Tobisch et al., 1997; Clarke andClarke, 1998), as well as cryptic foliation de-fined by the anisotropy of magnetic suscepti-

bility (e.g., Bouchez and Diot, 1990; Archanjoet al., 1998; Cruden et al., 1999), have all beenused to define and understand magma flowand ultimately magma emplacement processesand dynamics.

Despite these efforts, mapping of magmaticflow has been hampered partly by the limita-tions of each of these flow markers, and partlyby the masking of early-formed structures bylate flow (e.g., Benn, 1994). As a result, evi-dence for several processes inferred to occurin granite plutons, such as chamber growth byaddition of small magma pulses (Deniel et al.,1987), or thermal convection (e.g., Huppertand Sparks, 1988), commonly remain unseen.Similarly, extraction of light, depleted meltfrom the interstices of a solidifying magmahas long been recognized as an importantmechanism of magma differentiation (e.g.,McCarthy and Groves, 1979), but structuresindicative of this process have seldom beendescribed in granites.

In the Tavares pluton in northeastern Brazil(Fig. 1), exceptionally well-exposed structuresprovide a wealth of information regardingmagma flow, allowing us to infer the origin ofstructures seldom described but neverthelessfundamental to our understanding of magmachamber dynamics. In this paper we describestructures related to melt extraction frompores and magma flow structures developedduring magma crystallization. We start bybriefly describing the pluton and its regionalsetting. We then describe structures related tomelt extraction from mush pores, followed bymagma flow structures (mush is an effectivelysolid magma where crystals form an intercon-nected skeleton). We propose a model for theorigin of the ubiquitous marginal schlierenand one for the development of all described

Geological Society of America Bulletin, April 2001 509

MAGMA FLOW WITHIN THE TAVARES PLUTON, BRAZIL

Fig

ure

1.G

eolo

gica

lm

apof

the

Tav

ares

plut

on.

No

maj

orin

tern

albo

unda

ryw

asfo

und.

The

ellip

ses

mar

kou

tcro

psw

here

ellip

soid

sw

ere

foun

d(f

rom

R.R

.P

esso

a’s

doct

oral

thes

is,

inpr

epar

atio

n).I

nset

show

sth

epo

siti

onof

the

Tav

ares

plut

on(b

ox)

inth

em

apof

Bra

silia

nogr

anit

icbo

dies

(in

gray

)an

dsh

ear

zone

sof

the

Bor

bore

ma

Pro

vinc

e.

510 Geological Society of America Bulletin, April 2001

WEINBERG et al.

Figure 2. Typical streaky appearance of the Tavares granite, defined by mafic schlierenand concentrations of K-feldspar megacrysts.

structures in terms of compositional and ther-mal convection.

TAVARES PLUTON

The Tavares pluton was emplaced syntecton-ically during the Brasiliano orogen, which tookplace between 650 and 550 Ma (Sial et al.,1999). The pluton is part of a group of contem-poraneous high-K calc-alkaline bodies that cropout in the Alto Pajeu terrane of the BorboremaProvince (Almeida et al., 1981), northeasternBrazil (e.g., Ferrreira et al., 1998; Sial et al.,1999). Many of the structures described herewere also found in other plutons of this group,but are generally less well developed. The plutonis 200 km2 in area and is elongated parallel tothe N60E regional foliation. It is bounded on theeast by the left-lateral N60E striking Lagoa daCruz shear zone and in the north by the dextral,east-striking Juru shear zone. The pluton in-trudes mainly orthogneisses (including mylo-nites) and related migmatites, as well as schists,quartzites, marble, and cherts of the Alto Pajeuterrane (Fig. 1).

This pluton is composed of mainly high-Kcalc-alkaline granite-granodiorite with hetero-geneously distributed K-feldspar megacrysts(in general 15%–25% modally) and accessoryepidote, sphene, allanite, zircon, and only trac-es of opaque minerals. Due to heterogeneousmegacryst distribution, rock classificationvaries between granodiorite and granite (re-

ferred to hereafter as granite, sensu lato). Atypical outcrop is characterized by 0.5–1.0-m-long, mafic-rich and felsic-rich bands parallelto the regional foliation (Fig. 2), with a fewcurved, semicircular schlieren and occasionalmafic enclaves. The pluton is intruded by syn-magmatic diorite sheets and blobs, disruptedto varying extents giving rise to enclaveswarms and widespread microdioritic maficenclaves. Pillow-shaped enclaves with narrowdarker rims and K-feldspar megacryst inclu-sions, as well as the preserved stages of dis-aggregation of the mafic intrusions, attest tocontemporaneity between mafic and felsicmagmas. Despite outcrop heterogeneity, rockcomposition varies little across the pluton.

Pluton elongation and granite foliation arepreferentially parallel to the regional fabric.Except for an ;100-m-wide band near thepluton’s margins, granites are free from pen-etrative postmagmatic deformation (no visiblecrystal deformation in thin sections), suggest-ing that the dominant foliation developedwhile the magma was still partially molten.This flow foliation is a result of external tec-tonic deformation contemporaneous withmagma emplacement and cooling, and it light-ly overprints the many magmatic structures,such as the ellipsoids, described in the follow-ing. Close to the margins, and generally par-allel to them, foliation resulting from solid-state deformation becomes increasingly moreimportant, with quartz developing subgrains

and new grains, and microcline developingflame perthites.

MAGMATIC STRUCTURES

Leucogranite Pockets

Leucogranite pockets were found in a gran-ite outcrop partly quarried for gravel (outcrop51, near the town of Tavares, in Fig. 1). Here,dikes of a variety of granitic composition andtextures crosscut one another, and irregularsheets and decimetric pockets of leucograniteaplite cut a typical megacrystic granodiorite.Granitic dikes show a variety of complex in-ternal structures of uncertain origin. Here, weare mainly concerned with structures in leu-cogranites (Fig. 3). Leucogranite pockets haveirregular shapes, and at grain scale they formirregular branches that grade into the felsic in-terstitial material in the granodiorite (Fig. 3A).Some pockets have horizontal layers of maficminerals concentrated in their lower half (Fig.3A). Others have narrow (0.5–2 cm) maficlayers (melanosomes) surrounding their lowerhalf (Fig. 3B). Irregular sheets are randomlyoriented, narrow, planar, or tortuous layers ofmafic-rich granodiorite (melanosomes) ofvarying width (1–10 cm), containing narrowpockets of leucogranite (Fig. 3, C and D). Lo-cally, sheets give rise to an interconnected net-work, with leucogranite blobs at intersections(Fig. 3D). Irregular aplitic dikes without me-lanosome margins, and melanosome layerswithout internal leucosome lenses are alsocommon. In thin section, mafic-rich layers arerich in the mafic minerals typical of the sur-rounding granodiorite (hornblende 1 biotite1 epidote 1 sphene).

Many of the granitic dikes in this outcropare more evolved than the surrounding grano-diorite, and have tortuous margins and com-plex crosscutting relationships (e.g., dike acuts b, and b branches into a smaller tortuousdike that cuts a). Another leucogranite dike islinked to an irregular pocket of leucogranitethat branches into numerous fingers that dis-appear into the granodiorite; others havesharp, wedge-like endings in outcrop view.Structures in this outcrop are interpreted asrepresenting extraction of interstitial meltfrom a mush.

Schlieren and Magma Flow

Magmatic structures in the Tavares plutonare generally defined or delineated by maficschlieren. These are commonly curved, eitherindividually or in groups. Irregular structures,where schlieren bifurcate, crosscut one anoth-



er, or are joined in a variety of compositeshapes, are common (e.g., Wilshire, 1969;Barriere, 1981), but are not included in thispaper because they most likely result from acomplex history of magma flow. Here, we areconcerned with regular, geometrical structures,which are likely to result from a single pro-cess. We first describe typical magma blobsand the schlieren that delineate them, and me-ter-scale ellipsoidal structures. We then de-scribe schlieren defining snail structures andladder dikes (the arcuate schlieren of Cloos,1936) followed by a description of ubiquitousK-feldspar megacryst aggregates.

Common features of the Tavares pluton areschlieren-rimmed magma blobs, typically sev-eral in a single outcrop. These blobs are mostcommonly composed of granite of a slightlydifferent color or fabric than that of the sur-rounding granite. Occasionally these blobsmay be rich in mafic enclaves and maficschlieren, and may have irregular margins ordefine mushroom or teardrop shapes withheads from 50 cm to several meters in diam-eter (Figs. 4 and 5A. Mafic schlieren markingthe edges of magma batches are characterizedby a sharp external boundary, where maficminerals are concentrated, that grades inwardto the typical granite (Fig. 5A). Marginalschlieren often occur in groups with crosscut-ting relationships spread over a zone to 50 cmwide (Fig. 5B). These relationships indicatethat the effective margin of the flowing blobshifted with time.

Ellipsoids

Magmatic foliation and schlieren definemeter-scale two-dimensional ellipses (Fig. 6)apparently randomly distributed throughoutthe pluton (Fig. 1). These ellipses are oftenaccompanied by more irregular groups ofcurved schlieren similar to those previouslydescribed in the Ploumanac’h pluton by Bar-riere (1977, 1981). In subhorizontal exposuresthe ellipses are most commonly concentric(Fig. 6A), but they may be spiraling outward,or have two nuclei or cores, surrounded by acommon external ring foliation. Crystals aregenerally parallel to the concentric ellipticalfoliation, but tend to transpose it and becomeparallel to the dominant regional foliation,where this is at a high angle to ellipsoid fo-liation. We found five ellipses exposed onnearly vertical surfaces close to horizontal ex-posures, showing that their vertical and hori-zontal dimensions are similar. Ellipsoid 17(Fig. 7) is exposed on two vertical surfaces atright angles, and shows complex refoldingpatterns of schlieren. These observations are

indicative of the three-dimensional nature ofthe structures (ellipsoids). Many ellipsoidshave K-feldspar aggregates in their center(;20% of observed ellipsoids), and in somecases the aggregates occur as bands on the ex-ternal parts of the ellipsoids.

We mapped the distribution of ellipsoids inpart of a particularly well-exposed outcrop(Fig. 7). The ellipses range in size from ,1 mto a maximum of 6 m in the longest direction,with typical aspect ratios of ;2 and longestaxis within 20o of the dominant flow foliation(N65E). Because of their three-dimensional na-ture, their size distribution in two dimensionsis unimportant. The granite between the ellipsesin Figure 7 is characterized by a well-definedflow foliation, and where small ellipses (,1m), curved individual schlieren, or groups ofschlieren are common.

Snail Structure and Ladder Dikes

Isolated decimetric ring schlieren are commonthroughout the pluton. Several of them may oc-cur together, defining a meter-scale structure thatwe call snail structure due to its two- dimen-sional shape (Fig. 8A). These are orderly ar-ranged rings, with diameters increasing outward(ring schlieren; see Fig. 20 in Cloos, 1936),commonly cropping out in subhorizontal surfac-es at a high angle to prevailing subvertical fo-liation. The more internal rings are younger andmay be entirely inside, outside, or crosscuttingearlier, larger rings (Fig. 8A).

Granite dikes and irregular sheets in the Ta-vares pluton commonly show complex inter-nal structures with complex banding, schlie-ren, and K-feldspar megacryst aggregates.Ladder dikes are a particular form of thesecomplex dikes formed by nested crescent-shaped dark and light layers (Reid et al.,1993). They are narrow, sinuous, magmachannels with irregular walls. There is a rangeof varieties of ladder dikes in the Tavares plu-ton, from fully developed ones with gradationbetween mafic and felsic layers (Fig. 8, B andC) similar to those described in the Sierra Ne-vada (Reid et al., 1993), to simplified versionsformed by stacked mafic schlieren (Fig. 9; seealso Trent, 1995). Like their Sierran counter-parts, fine-grained mafic layers grade intocoarser light layers. The margins of ladderdikes are irregular and cusped, and the dike issinuous, although its width remains roughlyconstant over tens of meters. The cusps havethe same curvature as the arcuate schlieren in-side the dike. The two sections across a ladderdike in Figure 9 depict its three-dimensionalgeometry. In the vertical face, lower schlierenare crosscut by those higher up, indicating up-

ward younging. At the far end of the outcrop,the youngest layer merges with the surround-ing granite. These features indicate that theladder dike is a series of stacked channel bot-toms (or side walls) with magma flowing par-allel to the schlieren planes (Fig. 9B).

K-feldspar Megacryst Aggregates

Aggregates of zoned K-feldspar mega-crysts, in granitic or aplitic matrix, are com-mon throughout the pluton. They occur (1) as10–20-cm-wide layers in the granite, (2) asirregular or rounded masses, from 10 cm toseveral meters in diameter, (3) inside oraround mafic enclaves, or (4) within ladderdikes. Aggregates inside or around mafic en-claves are ubiquitous and form rims up to 20cm wide. The cuspate margin of the enclavesagainst megacrysts indicates that the enclaveswere less competent than the megacrysts. Wealso found a 2-m-wide round mass of mega-crysts with mafic schlieren, associated with asinuous enclave swarm (disrupted dike). Oc-casionally, we found meter-scale isolated mas-ses within the granite, or several isolatedpatches defining a linear trend. In ladder dikes,K-feldspar aggregates form either curved lay-ers, parallel to mafic schlieren, or bands up to50 cm wide, close to the narrowest parts ofthe dikes (logjams in dike necks; seefollowing).

ORIGIN OF STRUCTURES

In this section we discuss the origin of thestructures we have described and propose amodel for the origin of schlieren. We suggestthat the structures described and observed sep-arately could be linked and are a result ofgravity instability.

K-feldspar Aggregates: Logjams andFilters

K-feldspar megacryst aggregates occur in avariety of structures, most of which we areunable to interpret. However, in two casesthere is evidence to suggest that they resultfrom flow necking and logjamming. In ladderdikes, aggregates occur on the younger, or up-stream, side of dike necks, as inferred fromcrosscutting relationships in the dike. We in-terpret this as indicating that there was a flowcomponent parallel to schlieren curvature, asinferred herein, and that flow necking led to amegacryst logjam, which grows by the filter-ing out of megacrysts upstream (Clarke andClarke, 1998).

Megacrysts may also form logjams and act


WEINBERG et al.

Figure 3. Leucogranite pockets. (A) Irregular pocket where melt has been extracted fromthe interstices of the granodiorite, leaving behind horizontal layers enriched in mafic min-erals. (B) Irregular aplitic pocket exposed in two roughly perpendicular surfaces, withmafic mineral concentrations (melanosome) patches irregularly distributed in the imme-diately surrounding granodiorite; scale is 10 cm long.

as filters during the breakup of diorite intru-sions. When a large diorite blob starts to breakup, megacrysts form a logjam at the entranceto the growing granite magma channel, pass-ing only megacryst-free melt to the growingchannel. In other cases, K-feldspar aggregatesmay form around mafic dike walls and en-claves as a result of the combination of K-feldspar preferential attachment to or nucle-ation and growth at the contact with the maficmagma, and disturbance of granite magmaflow by the presence of enclaves.

The mesh size of megacryst filters dependson the pressure gradient driving magma flow.A higher pressure gradient increases the meshsize, allowing larger particles through. Corre-lation between crystal size and mineralogy re-sults in change of the modal composition ofthe magma crossing the filter and could giverise to magma layering.

New Model for the Origin of MaficSchlieren

Several processes have been proposed forthe origin of granite layering and mafic schlie-ren: partial assimilation of mafic enclaves,crystal settling, shearing out of inhomogenei-ties, steep physicochemical gradients at flowmargins (e.g., Barriere, 1981) leading to pref-erential crystallization of ferromagnesian min-erals and suppression of crystallization of fel-sic minerals (Naney and Swanson, 1980) andshear sorting against an effectively rigid wall(e.g., a more viscous magma, Barriere 1981).Shearing accounts for the size and/or mineralsorting of schlieren (Bagnold effect; e.g.,Bhattacharji and Smith, 1964; Komar, 1972;Barriere, 1981), but because solid grains mi-grate away from the wall, it does not accountfor schlieren at flow rims. We postulate thatschlieren with size worting, like the ones de-scribed here, may result from the combinedeffects of shear flow and loss of interstitialmelt to the porous and/or permeable magmaticwalls (Fig. 10) and crystal filtering. This re-quires a negative pressure gradient toward thewalls, resulting from unbalanced pressures ona wider scale (beyond the local flow ob-served). However, we have not been able toshow that melt has escaped outward into thewalls, and we cannot dismiss the possibilitythat melt migrated from the flow margins intothe blob by filter pressing of a crystal-richmargin.

Shear sorting and melt loss are most effec-tive when the walls enclosing the flow are acrystal-liquid mush. Mushes behave as a per-meable viscoplastic material (i.e., flows vis-cously at stresses higher than its yield



Figure 3. (Continued.) (C) Sheet of melanosome, associated with lenses of leucogranite,scale is 5 cm long. (D) Irregular sheets and patches of melanosome surrounding irregularpatches of leucogranite; note 5-cm-wide lens cover for scale.

strength). Magmas prone to producing schlie-ren are those that develop a relatively weaksolid framework at low crystal fraction(;50%), so as to keep a high permeability, k(a function of the cube of porosity), and duc-tility (to allow internal magma flow to devel-op). Permeable, weak mush not only favorsthe formation of marginal schlieren but alsofavors melt extraction from mush pores, as weshow here.

Ladder Dikes and Snail Structures

Reid et al. (1993) interpreted ladder dikes asresulting from crystals settling out at the bottomof slushes flowing through narrow vertical chan-nels in semisolid magma. Our observations sug-gest a different origin for the ladder dikes inTavares. First, we note that an important flowcomponent perpendicular to the curved schlierenin ladder dikes is not viable because it would

destroy curved schlieren. Furthermore, grain-sorting characteristics of well-developed ladderdikes require shearing against a rigid wall. Thus,we conclude that curved schlieren in ladderdikes (or rings in snail structures) represent thewalls of cylindrical magma paths (or the bottomof channels, as inferred from Fig. 9), and thatthe main flow direction is parallel to the walls,out of the plane of the arcuate schlieren (Figs.11 and 9). We infer that a small flow componentperpendicular to the arcuate schlieren must haveoccurred to produce curved cusps (in some dikesthe schlieren are partly disrupted and reorientedtoward parallelism with the margins) and widehomogeneous bands within dikes.

We interpret ladder dikes and snail struc-tures as related features that result from thesuperposition of sequential cylindrical magmapathways. Movement between the position ofthe magma source and the section observed,associated with intermittent magma release,

led to new channels being slightly offset fromprevious ones (Fig. 11). Snail structures resultfrom random displacement, whereas ladderdikes result from a unidirectional relative dis-placement. An alternative for ladder dikes,which could account for their length, is thatthe magma instability is not fed from a pointsource but from a linear source. Our modelexplains cross-bedding structures associatedwith ladder dikes as observed in Sierra Ne-vada (Fig. 11; Reid et al., 1993). The flow inthe channels is driven by local density inver-sions, while the unidirectional relative dis-placement between source and observationlevel results from large-scale flow within thepluton. We postulate that some of these mag-ma pulses may have fed ellipsoids and blobs.

Ellipsoids are Thermal Plume Heads

Ellipsoids differ from simple diapiric blobswith marginal schlieren. Their concentric orspiralling form and similarity between therock inside and outside them are features tobe expected from thermal plumes of hot gran-ite magma. A thermal plume is a gravitationalinstability originated from a low-density, hotthermal boundary layer that rises through liq-uid of similar composition (Campbell et al.,1989; Fig. 12). A bulbous head leads theplume, generally followed by a narrow feederconduit that links the head to the hot sourceboundary layer. The plume heads grow duringascent by entraining their surroundings (Grif-fiths, 1986) through buoyancy (heat) diffusionaway from the plume head. A thin layer ofliquid adjacent to the head is heated and be-comes unstable and stirred into the head. Grif-fiths (1986) showed that the entire hot bound-ary layer around the plume head is entrained,so that plume head buoyancy remains constantas it increases in volume and cools. The in-ternal structure of laboratory plume heads(Fig. 12) is reminiscent of the ellipsoid struc-tures and contrasts with compositional diapirs,which are unable to entrain surrounding mag-ma because of much slower chemical diffu-sion as compared to thermal diffusion.

We tested whether the ellipsoids could haveresulted simply from convection of a granitelayer within the pluton, without any externalheat source. On the basis of ellipsoid size, wefound that a layer would have to be only afew meters thick and have a steep temperaturegradient to convect. This is unlikely to existwithin a large cooling pluton (Appendix 1).

We propose two possible alternatives forthe origin of plumes: (1) they are hot graniticmagma batches released directly from thesource; or (2) they are granitic magma batches


WEINBERG et al.

Figure 4. Two magma blobs from the sameoutcrop, defined by wide heads and narrowstalks. Both were exposed on round surfac-es and were redrawn as a projection to aplane dipping ;308–408 to the viewer. Theblob on the left is formed by light coloredgranite with mafic schlieren, dioritic en-claves, and K-feldspar megacryst aggre-gates, concentrated on the left side of stemand head. Enclaves are oriented parallel tothe regional foliation, at a high angle toblob length, and overlap the boundary be-tween lighter and darker surroundinggranite. The blob on the right is internallycomposite, and the external boundary iswell delineated by graded mafic schlieren. Figure 5. (A) Magma diapir with head delineated by schlieren and tail of light colored

granite. (B) Mafic schlieren marking the boundary between an intrusive batch (on theright side) and resident magma. The boundary is characterized by a 50 cm zone of cross-cutting schlieren, indicating movement of the active boundary between the two magmas.

heated by intrusive diorite. We have no evi-dence that would favor one over the other; wehave no knowledge of local sources of hotgranite magma, nor have we found directphysical links between enclave swarms and el-lipsoids. Here, we further explore alternative2. Mafic enclave swarms are commonthroughout the pluton. In order to checkwhether they could be the local heat sourcefor the plumes, we carried out a simple studyof the effects of heat exchange between dioriteand granite (Appendix 2). We assume for sim-plicity that the crystals in the granite magmamelt immediately in response to heat diffusedfrom the mafic intrusion. First we studied theincrease in temperature and decrease in den-sity of granite heated by diorite, then we stud-ied the rise of a hot plume of granite throughcolder granite magma. We found that a smallheat input could lead to a thermal head ofgranite of 60 cm radius, lightened mainly bythe effect of crystal remelting. Such an ellip-soid could rise and grow to a radius of 3 mby entraining the surrounding, cooler magma.

Although there is little doubt, on thegrounds of physics, that thermal convectionshould occur within granite magma chambers,there has been little field evidence of this pro-cess. Barriere (1981) interpreted the curvedschlieren in Ploumanac’h as indicating turbu-lent convection. Authors before him also at-tributed curved biotite schlieren to convectioncurrents (Mayo, 1941; Wilshire, 1969), butothers attributed them to episodic shearingflow (Smith, 1975). Abbott (1989) mappedsteep cylindrical structures at Chebucto Headin Nova Scotia. These are as much 500 macross and defined by platy K-feldspar me-gacryst foliation planes. He discarded the pos-sibility that these represent frozen-in convec-tion cells and suggested that they result frominternal folding of the magma by forced flow.We interpret schlieren structures in the Tavarespluton, including structures like those mappedby Barriere (not discussed here), as a result oflocal laminar thermal or compositional con-

vection. Although we were unable to find ev-idence for chamber-wide convection, meso-scale plumes indicate that small, localized heatinput may cause substantial density difference,and drive gravitational instabilities.

Magma Extraction Structures: EnergyMinimization, Gravity-DrivenCompaction, and Contraction

Separation of melt from crystals has longbeen recognized as a diversification process ofmagmatic rocks. The extraction of viscous fel-sic melts from pores of partially molten rockshas often been used to explain cumulate tex-ture in granites (sensu lato) and to explaintheir geochemical trends (e.g., McCarthy andGroves, 1979). However, extraction structureshave seldom been described, and the processis fraught with difficulties, mostly related toslow extraction (Miller et al., 1988, p. 145;Petford, 1993; Weinberg, 1999). The struc-



Figure 6. Concentric ellipsoids (horizontal exposure); note decimetric K-feldspar aggregatein the center of B.

tures in the quarry near Tavares village presenta unique opportunity to understand how meltextraction from a solidifying magma takesplace.

In the quarry, structures were interpreted toresult from melt extraction from pores of acrystal-solid mush, similar to those developedin migmatites: leucogranite pockets are leu-cosomes extracted from mush pores, and themafic-rich residue is the melanosome. In thepluton, melt extraction took place during rocksolidification and volume decrease in a weaklyfoliated rock, in contrast to extraction struc-tures in migmatites, which result from anatex-is, generally accompanied by volume expan-sion, in strongly foliated rocks.

Details of some of the dikes exposed in thequarry and their composition suggest that theyare locally derived, and represent earlier stag-

es of extraction (possibly including crystals)from the original magma. The preserved leu-cogranite extraction structures in the quarryform a poorly developed network, which re-moved magma out of the pores of a mush, andgave rise to narrow dikes. Although we haveno independent way-up structures, the concen-tration of mafic minerals consistently below oron the lower side of leucogranite pockets sug-gests vertical way-up in a system where grav-ity played an important role in segregatingmelt from solid mafic crystals. A significantdifference to migmatites is the lack of a well-defined structural control in the extractionstructures. This is due to the lack in the Ta-vares pluton of a strong foliation and perme-ability or strength anisotropy.

In order to understand the origin of extrac-tion structures, we now consider the devel-

opment of a solid framework magma and theprocess of melt extraction from pores. Crys-tals in a cooling melt will tend to form a con-tiguous network in order to achieve minimumenergy configuration (Jurewicz and Watson,1985). The development of a solid skeleton iscontrolled by the volume fraction of solid, X,the average melt-solid wetting angle, u (gen-erally u 5 208–608, Holness, 1997), and oncrystal size and shape distribution (Miller etal., 1988). The main control on the crystalfraction X necessary to develop a solid skel-eton, Xcrit, is u. Xcrit decreases as u increases.However, for a given u, Xcrit will be largelycontrolled by crystal size and shape distribu-tion, specific to each magma. Miller et al.(1988) estimated that the crystal fraction re-quired to form a skeleton ranges widely, butcould be as low as 50%.

Once a solid skeleton is developed, meltwill tend to migrate into larger melt pocketsin order to attain a minimum energy config-uration. Jurewicz and Watson (1985) deter-mined the equilibrium melt fraction thatwould minimize the interfacial energy state intwo-dimensional partially molten crystallineaggregates. The equilibrium melt fraction de-creases with increasing u, tending to zero asu tends to 608. In oversaturated systems, i.e.,those with melt fraction above equilibrium,excess melt coalesces to form large pools thatcould be tapped by fractures or could build upinterconnected conduits (Jurewicz and Wat-son, 1985). These observations imply thatcrystal-melt systems with u approaching 608are ideal systems for the development andpreservation of extraction structures. This isbecause they develop a permeable solid skel-eton at low crystal fraction, and are furtherfrom equilibrium than systems with lower u.These features are the same as those ideal forproducing marginal schlieren as previouslyconcluded here.

We envisage two possible gravity-drivencompaction models to account for the hori-zontal mafic layers in the lower half of meltpockets or melanosome surrounding their low-er half (Figs. 3A and B): (1) settling of maficminerals in locally liquefied magma pocketsdue to melt migration; and (2) textural ripen-ing of a crystal mush, accompanied by the de-tachment and sinking of grains and upwardsegregation of melt (Miller et al., 1988). Vol-ume shrinkage and accompanying tensionalstresses may have eased segregation by cre-ating low-pressure melt sinks and favoringcrack and/or dike propagation.

Some granite dikes/sheets in this outcropseem to start out of the surrounding granodi-orite. Their tortuous shape suggests they cut


WEINBERG et al.

Figure 7. Size and distribution map of ellipsoids exposed in outcrop. Size measurementshave a large error, because elliptic foliation weakens outward (width error 61 m, lengtherror 650 cm, dashed ellipses are poorly defined structures). Ellipsoids 17 and 19 areexposed on a vertical wall. Ellipsoids smaller than 1 m long are not shown.

Figure 8. (A) Snail structure. The centers of internal ring schlieren are displaced in re-lation to the center of external ones, forming an elongated structure. Note that the moreexternal schlieren on the right tend toward a ladder dike shape.

across magma (mush?). We suggest that thedikes are also a result of a local process ofmagma extraction from the surroundinggranodiorite magma. Because of their texturalsimilarity to the granodiorite (as opposed tothe sharp contrast between leucogranite meltand granodiorite), there is less evidence forthe process of extraction. This conclusion rais-es some questions about the state of the sur-rounding magma when granitic magma wasextracted: was it still liquid magma capable ofcracking elastically; was it a mush capable ofliberating granitic melt from its pores; or wasthe flowing melt able to extract solid crystalsas well? In future work we aim to establish,on the basis of trace element and rare earthelement patterns, as well as mineral chemistry,the amount of melt that has been extractedfrom granodiorite pores, and to determinewhether melt extraction was more widespreadand involved the granitic dikes.

In summary, extraction structures devel-oped through melt expulsion from pores intomelt pockets, and schlieren developed throughmelt expulsion from blob margins into poresof surrounding magma. Both are favored bymagmas that develop a permeable and ductilesolid framework at low crystal fraction, pos-sibly as a result of high wetting angle and fa-vorable crystal size and shape distribution.Whether melt migrates out of or into poresdepends only on the sign of the pressure gra-dient, which is controlled by both large- andsmall-scale pressure balance within thechamber.

Diapirs and Compositional Convection inGranite Mush

Magma blobs are interpreted as composi-tional diapirs, driven downward or upward bycontrasting density to surrounding magma re-sulting from, for example, magma fraction-ation or disruption of dense diorite magmas.We have not attempted to determine densitydifferences between rocks inside and outsidethe blobs partly because heat and melt contentmust have played a key role in controllingdensity inversion, and partly because meltmay have been expelled out of the crystalliz-ing diapir (as proposed herein for the originof schlieren).

Compositional convection in magma mushhas been proposed based on laboratory modelsand supported by observations in mafic-ultra-mafic magma chambers (e.g., Tait and Jaupart,1992; Jaupart and Tait, 1995). We use our sep-arate observations and interpretations on leu-cogranite melt extraction, diapiric blobs, andour model for the origin of schlieren to pro-

pose a model of compositional magma con-vection in a granitic mush (Fig. 13). We stressthat extraction structures described herein arenot directly related to granite in the blobs, butare simply used here as an example of extrac-tion structures. The starting point of the pro-cess is melt segregation from pores into pock-ets and sheets. The example of the aplite dikelinked to a melt pocket suggests that diking isa possible mechanism of tapping melt pockets.

Field relations, however, suggest that meltmay have been removed from the mush poresby amplification of porosity fluctuation andthe formation of chimneys due to dissolutionand precipitation related to melt flow. In re-gions of faster melt flow, chemical and heatadvection leads to increased porosity and pos-sibly grain detachment from the solid skele-ton, either through remelting or through Ost-wald ripening (Miller et al., 1988), in a



Figure 8. (Continued). (B) Starting point of a fully developed ladder dike in the Tavarespluton. (C) Small section of a long ladder dike with cusped outlines. Cusps are continuouswith curved internal schlieren.

mechanism akin to that described by Jaupartand Tait (1995, p. 17624–17626). The posi-tion of the channels may migrate with time insearch of a wider scale equilibrium and maybe the underlying cause of ladder dikes. Onceflow is channeled, melt will rise through the

mush until it reaches its neutral buoyancy lev-el or is trapped in strong surroundings. Highmagma pressure at the diapir’s head drivesmelt out of the blob into the surroundingmush, giving rise to marginal schlieren. Theimportance of compositional convection in the

chemical evolution of the Tavares pluton re-mains undetermined, and unfortunately map-ping indicates a relatively homogeneous ex-posure level.

Why the Tavares Pluton?

We have proposed that uncommon struc-tures cropping out in the Tavares pluton are aresult of processes likely to be common tomany granites. Why are these structures notwidespread and yet are well developed in theTavares pluton? First, they may not be so un-common, but perhaps they have just beenoverlooked because they are seldom suffi-ciently well developed and/or preserved to al-low us to understand their origins. We note,for example, that ellipsoids and ladder dikesare widespread in high-K calc-alkaline plutonsof the Borborema province, but not as welldeveloped as in Tavares. The presence of sim-ilar structures in high-K granites of the SierraNevada (e.g., Cloos, 1936) suggests that theymay be best developed or preserved in gran-ites of this composition. Second, althoughconvective overturns may occur at any timebefore the magma reaches advanced stages ofsolidification, structures will only be pre-served when developed under particularly fa-vorable conditions. If they develop early, mar-ginal schlieren will not develop and overturnstructures will tend to be destroyed by laterflow and crystallization. We have argued thatmarginal schlieren, extraction structures, andconsequently ladder dikes, snail structures, di-apirs, and thermal plume heads are ideally de-veloped and preserved when a solidifyingmagma produces a weak, high-permeabilitycrystal mush. The development of this type ofmush depends on the characteristics of the sol-id framework structure developed during crys-tallization, particular to each magma. Struc-tures developed in the mush stage are alsomore likely to withstand later deformation andbe preserved until the end of crystallization.

SUMMARY AND CONCLUSIONS

Melt extraction and flow structures are par-ticularly well developed in the Tavares pluton,and rendered visible by widespread maficschlieren. These structures indicate a dynamic,viscous environment during pluton crystalli-zation, where gravity instabilities played animportant role. We describe meter-scale tear-or mushroom-shaped blobs and ellipsoids thatwere interpreted as the leading heads of com-positional and thermal instabilities, respective-ly, trailed by stems and/or feeders represented


WEINBERG et al.

Figure 9. (A) Two perpendicular sections of a ladder dike as seen in outcrop. Note thecrosscutting relationships of schlieren in the frontal, vertical face. (B) Interpretation of A.The dike is composed of a series of stacked, inclined, U-bottomed channels, younginginward (in the concave direction), as evidenced by crosscutting relationships (as shown bydetails in A). Mafic schlieren develop at the wall (bottom) of each channel, where shearingis maximum. Magma flow channeling initiates at the far end of the drawing.

Figure 11. Ladder dikes and snail structures result from the relative motion between apoint magma source and the observation level of the magma pathway so that each newpulse is displaced in relation to earlier pulses. Random relative motion between sourceand observed section leads to snail structure, whereas constant relative motion leads toladder dikes.

Figure 10. An intrusive magma flows alonga wall of magma sufficiently crystallized asto have a well-connected, permeable, crys-tal network. Crystals in the intrusive mag-ma are sorted by size as they migrate in-ward from the contact (Bagnold effect), atthe same time as the crystal-poor melt layerdeveloped near the flow margins leaks intothe porous resident magma.

Figure 12. Structure in a vertical sectionacross a thermal plume head. A horizontalsection would have the form of concentriccircles. Surrounding magma is entrainedinto the plume head by buoyancy gain dueto heat diffusion from the plume head.

by ladder dikes and snail structures. The lattertwo result from the partial superposition of thecylindrical paths of consecutive magma batch-es. K-feldspar megacrysts tend to form aggre-gates in a range of situations. In some of thesethey may behave as filters, separating mega-crysts and passing melt. These filters havevariable mesh size, depending on the magmapressure gradient.

We propose a new model for the origin ofmarginal schlieren that requires shear flowsorting against a rigid wall and melt escapeinto the pores of this same wall. We also de-scribe structures developed by melt flow in theopposite direction, from the pores of a mushy

magma into melt pockets, sheets, and chan-nels. We conclude that magmas that developan interconnected network of solid crystals atrelatively low crystal fractions (;50%), thoseof wetting angles approaching 608, and favor-able crystal size and shape distribution, areideal systems from which to extract or intrudeinterstitial melt. Depending on the sign of thepressure gradient, melt will flow into or outof a melt pocket.

We propose a general model for small-scalecompositional and thermal convection insidemushy (effectively solid) magma bodies.Compositional convection results from meltfractionation, gravity-driven segregation and

extraction, aided by magma contraction, giv-ing rise to diapirs. Thermal convection is driv-en by heat released by mafic intrusions, givingrise to thermal plumes of melt-rich graniteblobs. Although direct evidence for chamber-wide convection is lacking, small-scale insta-bilities indicate the importance of gravity in-stabilities during pluton evolution.

APPENDIX 1

ELLIPSOIDS AND A CONVECTING LAYER

If the ellipsoids represent thermal plumes result-ing from thermal convection within a granite layer,we can infer the depth of the convecting layer based



Figure 13. Compositional and thermal convection model for the development of structuresobserved in the Tavares pluton. Compositional convection develops in a magma mush(magma with a solid framework), where light unstable magma is released from pores intopockets and sheets, driven by gravity and forces resulting from the minimization of energyand pressure gradients related to magma contraction. Light melt is channeled upward,giving rise to ladder dikes and snail structures (not shown) and topped by a bulbousdiapiric head, from where melt is driven out and into the pores of the mushy surrounding,to produce schlieren. Thermal convection is driven by heat released by diorite intrusionsthat expand and melt back surrounding magma (which becomes unstable), rise, and en-train surrounding magma. In this case the surrounding magma need not be a mush, butthermal plumes developed in mushes are more likely to be preserved.

on the distance, l, between ellipsoids. The depth ofthe layer, h, can be determined from: l 5 l/h 52.828 or 2.016, depending on whether the top andbottom boundaries are free slip or no slip. For atypical distance between ellipsoid centers of l 5 10m, then h 5 425 m. For convection to occur, thedimensionless Rayleigh number, Ra, must reach acritical value Rac ; 103 (e.g., Turner, 1973) (Ra 5argDTh3/hk, where a is the coefficient of expan-sion, r is magma density, g is gravitational accel-eration, DT is the temperature difference betweentop and bottom of the convecting layer, h is theheight of the layer, h is magma viscosity, and k isthermal diffusivity.) For a 5 10–5 K–1, r 5 2500kg/m3, g 5 10 m/s2, h 5 4 m, k 5 10–6 m2/s, andh 5 104–106 Pas, Rac will be reached for a DT be-tween 0.5 and 50 8C. For large, stable magma cham-bers, such temperature differences are unlikely todevelop over such short distances without the aid ofa local source of heat, such as mafic intrusions.

APPENDIX 2

PLUME ORIGIN AND HEAT EXCHANGEBETWEEN DIORITE AND GRANITE

Modeling the evolution of a boundary layeraround an intrusive diorite is beyond the scope ofthis paper. Here we simply want to know whetherheating the granite magma by a small diorite en-clave swarm is sufficient to produce a granite plumeof the sizes observed. For this purpose we assumea linear relation between magma temperature, crys-tal fraction X, and magma density r, and study thecase where the initial magma temperature Ti is be-tween that of the magma solidus and liquidus (Ts

and Tl, respectively). More complex relations be-tween T, X, and r could have been used, but wechose the simple approach just to exemplify the pro-cesses involved.

Heat released by the diorite (H) to a surroundinggranite at initial temperature, Ti, and crystal fractionXi will raise the temperature and remelt part of thecrystals. A unit mass of granite magma to which Hhas been added will increase in temperature, DT (Tf

2 Ti; subscript f represents final) by

DT 5 H/[c 1 (L/DT )]p ls (1)

for Tf , Tl, where cp is the heat capacity (kJ/kg/8C),L is the latent heat (kJ/kg), and DTls the temperaturedifference between magma liquidus and solidus.

For Tf . Tl:

DT 5 [H 2 (T —Ti)·(L/DT )]/c .l ls p (2)

Given our initial assumptions, the decrease incrystal fraction will be directly proportional to theincrease in temperature Xf 5 Xi 2 DT/DTls for equa-tion 1, or Xf 5 0 (no solids left) for equation 2.

In order to determine the density difference be-tween heated granite magma and surrounding gran-ite magma, we assume a linear variation of densitywith temperature whereby

Dr 5 Dr DT/DT ,ls ls (3)

where Drls is the total density difference betweensolidus and liquidus (of the order of 0.1rrock).

We assume, as an example, that the diorite re-leased heat equivalent to 100 8C of sensible heat (H5 cp 3 100 8C) per unit mass of granite containing

initially ;60% crystals. Using equation 1 and typ-ical values for magmas of L 5 3 3 105 kJ/kg, DTls

5 100 8C, and cp 5 1.3 3 103 kJ/kg/ 8C, we obtainDT 5 30 8C, a loss of 30% of crystals and a de-crease in density of 0.3Drls (or Dr0.03rrock). Thedensity decrease, due to crystal melting, compareswith the much smaller effect of thermal expansivityalone of Dr 5 1.5 3 10–3rrock (for a 30 8C temper-ature increase and a 5 5.10–5 K–1). Magma viscositywill also drop, but with only minor effect on therise of the thermal plume.

On the basis of the size of exposed plume heads,the initial size of the head that left the boundarylayer may be back-calculated using the rough as-sumptions that plume growth is entirely due to en-trainment. To produce a typical ellipsoid of radius,r 5 3 m, which stopped rising when its excess tem-perature reached a subjective 1 8C, the initial plumeradius would be roughly 0.6 m, and this plumewould rise with an initial velocity, V 5 140/h or V5 1.4 3 10–4 m/s for h 5 106 Pas. The heat inputper unit mass of granite we have assumed is modest.A diorite volume similar to that of the initial felsicplume head (r 5 0.6 m), and intruded at its liquidustemperature, would be able to heat that volume ofgranite magma by releasing approximately one-thirdof its suprasolidus heat content.

ACKNOWLEDGMENTS

Weinberg thanks Barrie Clarke for introducinghim to the structures in Chebucto Head and for sug-gesting improvements to the manuscript, and ChrisTalbot for technical support and discussions. Wealso thank Mike Brown, Jonathan Miller, and DonnaWhitney for careful and helpful reviews. This workwas partly funded by a research grant from theSwedish Natural Science Research Council (NFR),and by a scholarship from the Brazilian NationalResearch Council (CNPq). We also acknowledgefunding from Programa de Apoio ao Desenvolvi-mento Cientıfico e Tecnologico/Financiadora de Es-tudos e Projetos project 65.930.619.

REFERENCES CITED

Abbott, R.N., Jr., 1989, Internal structures in part of theSouth Mountain batholith, Nova Scotia, Canada: Geo-logical Society of America Bulletin, v. 101, p.1493–1506.

Almeida, F.F.M., Hasui, Y., Brito Neves, B.B., and Fuck,R.A., 1981, Brazilian structural provinces: An intro-duction: Earth Science Reviews, v. 17, p. 1–29.

Archanjo, C.J., Macedo, J.W.P., Galindo, A.C., and Araujo,M.G.S., 1998, Brasiliano crustal extension and em-placement fabrics of the mangerite-charnockite pluton


WEINBERG et al.

of Umarizal, northeast Brazil: Precambrian Research,v. 87, p. 19–32.

Barriere, M., 1977, Deformation associated with the Plou-manac’h intrusive complex, Brittany: Geological So-ciety of London Journal, v. 134, p. 311–324.

Barriere, M., 1981, On curved laminae, graded layers, con-vection currents and dynamic crystal sorting in thePloumanac’h (Brittany) subalkaline granite: Contri-butions to Mineralogy and Petrology, v. 77, p.214–224.

Benn, K., 1994, Overprinting of magnetic fabrics in gran-ites by small strains: Numerical modelling: Tectono-physics, v. 233, p. 153–162.

Bhattacharji, S., and Smith, C.H., 1964, Flowage differen-tiation: Science, v. 145, p. 150–153.

Bouchez, J.L., and Diot, H., 1990, Nested granites in ques-tion: Contrasted emplacement kinematics of indepen-dent magmas in the Zaer pluton, Morocco: Geology,v. 18, p. 966–969.

Campbell, I.H., Griffiths, R.W., and Hill, R.I., 1989, Melt-ing in an Archaean mantle plume: Head it’s basalts,tail it’s komatiites: Nature, v. 339, p. 697–699.

Clarke, D.B., and Clarke, G.K.C., 1998, Layered granodi-orites at Chebutco Head, South Mountain batholith,Nova Scotia: Journal of Structural Geology, v. 20, p.1305–1324.

Cloos, E., 1936, Der Sierra Nevada Pluton in Californien:Neues Jahrbuch fur Mineralogie, Geologie, und Pa-leontologie, v. 76, part B,p. 355–450.

Cruden, A.R., Tobisch, O.T., and Launeau, P., 1999, Mag-netic fabric evidence for conduit-fed emplacement ofa tabular intrusion: Dinkey Creek Pluton, central Si-erra Nevada batholith, California: Journal of Geo-physical Research, v. 104, p. 10511–10530.

Deniel, C., Vidal, P., Fernandez, A., LeFort, P., and Peucat,J.-J., 1987, Isotopic study of the Manaslu granite(Himalaya, Nepal): Inferences on the age and sourceof Himalayan leucogranites: Contributions to Miner-alogy and Petrology, v. 96, p. 78–92.

Ferrreira, V.P., Sial, A.N., and Jardim de Sa, E.F., 1998,Geochemical and isotopic signatures of Proterozoicgranitoids in terranes of the Borborema structuralprovince, northeast Brazil: Journal of South AmericanEarth Sciences, v. 11, p. 439–455.

Griffiths, R.W., 1986, Thermals in extremely viscous fluids,

including the effects of temperature-dependent viscos-ity: Journal of Fluid Mechanics, v. 166, p. 115–138.

Holness, M.B., 1997, The permeability of non-deformingrock, in Holness, M.B., ed., Deformation-enhancedfluid transport in the Earth’s crust and mantle: London,Chapman and Hall, p. 9–39.

Huppert, H.E., and Sparks, R.S.J., 1988, The generation ofgranitic melts by intrusion of basalt into continentalcrust: Journal of Petrology, v. 29, p. 599–624.

Hutton, D.H., 1988, Granite emplacement mechanisms andtectonic controls: Inferences from deformation studies:Royal Society of Edinburgh Transactions, Earth Sci-ences, v. 79, p. 245–255.

Jaupart, C., and Tait, S., 1995, Dynamics of differentiationin magma reservoirs: Journal of Geophysical Re-search, v. 100, p. 17615–17636.

Jurewicz, S.R., and Watson, E.B., 1985, The distribution ofpartial melt in a granitic system: The application ofliquid phase sintering theory: Geochimica et Cosmo-chimica Acta, v. 49, p. 1109–1121.

Komar, P.D., 1972, Mechanical interactions of phenocrystsand flow differentiation of igneous dikes and sills:Geological Society of America Bulletin, v. 83, p.973–988.

Martin, D., Griffiths, R.W., and Campbell, I.H., 1987, Com-positional and thermal convection in magma cham-bers: Contributions to Mineralogy and Petrology, v.96, p. 465–475.

Mayo, E.B., 1941, Deformation in the interval Mt. Ly-ell2Mt. Whitney, California: Geological Society ofAmerica Bulletin, v. 42, p. 1002–1084.

McCarthy, T.S., and Groves, D.I., 1979, The Blue TierBatholith, northeastern Tasmania: Contributions toMineralogy and Petrology, v. 71, p. 193–209.

Miller, C.F., Watson, E.B., and Harrison, T.M., 1988, Per-spectives on the source, segregation and transport ofgranitoid magmas: Royal Society of Edinburgh Trans-actions, Earth Sciences, v. 79, p. 135–156.

Naney, M.T., and Swanson, S.E., 1980, The effect of Feand Mg on crystallization in granitic systems: Amer-ican Mineralogist, v. 65, p. 639–653.

Petford, N., 1993, Porous flow in granitoid magmas: Anassessment, in Stone, D.B., and Runcorn, S.K., eds.,Flow and creep in the Solar System: Observations,

modeling and theory: Netherlands, Kluwer AcademicPublications, p. 261–286.

Ramsay, J.G., 1989, Emplacement kinematics of a granitediapir: The Chindamora batholith, Zimbabwe: Journalof Structural Geology, v. 11, p. 191–209.

Reid, J.B., Murray, D.P., Hermes, O.D., and Steig, E.J.,1993, Fractional crystallization in granites of the Si-erra Nevada: How important is it?: Geology, v. 21, p.587–590.

Sial, A.N., Toselli, A.J., Saavedra, J., Parada, M.A., andFerreira, V.P., 1999, Emplacement, petrological andmagnetic susceptibility characteristics of diverse mag-matic epidote-bearing granitoid rocks in Brazil, Ar-gentina and Chile: Lithos, v. 46, p. 367–392.

Smith, T.E., 1975, Layered granitic rocks at ChebutcoHead, Halifax County, Nova Scotia: Canadian Journalof Earth Sciences, v. 12, p. 456–463.

Tait, S., and Jaupart, C., 1992, Compositional convectionin a reactive crystalline mush and melt differentiation:Journal of Geophysical Research, v. 97, p.6735–6756.

Tobisch, O.T., McNulty, B.A., Vernon, R.H., 1997, Micro-granitoid enclave swarms in granitic plutons, centralSierra Nevada, California: Lithos, v. 40, p. 321–339.

Trent, D.D., 1995, Front cover caption: Journal of Geolog-ical Education, v. 43, front cover and p. 14.

Turner, J.S., 1973, Buoyancy effects in fluids: Cambridge,Cambridge University Press, 367p.

Weinberg, R.F., 1992, Internal circulation in a buoyant two-fluid Newtonian sphere: Implications for composedmagmatic diapirs: Earth and Planetary Science Let-ters, v. 110, p. 77–94.

Weinberg, R.F., 1997, The disruption of a diorite magmapool by intruding granite: The Sobu body, LadakhBatholith, Indian Himalayas: Journal of Geology, v.105, p. 87–98.

Weinberg, R.F., 1999, Mesoscale pervasive melt migration:Alternative to dyking: Lithos, v. 46, p. 393–410.

Wilshire, H.G., 1969, Mineral layering in the Twin Lakesgranodiorite, Colorado, in Larsen, L.H., Prinz, M.,Manson, V., eds., Igneous and Metamorphic Petrolo-gy: Geological Society of America Memoir 115, p.235–261.

MANUSCRIPT RECEIVED BY THE SOCIETY NOVEMBER 9, 1999REVISED MANUSCRIPT RECEIVED JUNE 5, 2000MANUSCRIPT ACCEPTED AUGUST 23, 2000

Printed in the USA

Magma ﬂow within the Tavares pluton, northeastern Brazil ...weinberg/PDF_Papers/...MAGMA FLOW...

Documents

Transcript of Magma ﬂow within the Tavares pluton, northeastern Brazil ...weinberg/PDF_Papers/...MAGMA FLOW...