uc berkeley seismology lab - c ontact partial m elting of g ran...

Contact Partial Melting of Granitic CountryRock, Melt Segregation, and Re-injection asDikes into Ferrar Dolerite Sills, McMurdo DryValleys, Antarctica

TABER G. HERSUM*, BRUCE D. MARSH AND ADAM C. SIMONy

MORTON K. BLAUSTEIN DEPARTMENT OF EARTH AND PLANETARY SCIENCES, JOHNS HOPKINS UNIVERSITY,

BALTIMORE, MD 21218, USA

RECEIVED NOVEMBER 7, 2006; ACCEPTED AUGUST 10, 2007ADVANCE ACCESS PUBLICATION OCTOBER 6, 2007

Numerous, interconnected, granitic dikes (530 cm in width andhundeds of meters in length) cut Ferrar dolerite sills of theMcMurdo DryValleys, Antarctica.The source of the granitic dikesis partial melting of granitic country rock, which took place in thecrust at a depth of about 2^3 km adjacent to contacts with doleritesills. Sustained flow of doleritic magma through the sill generated apartial melting front that propagated into the granitic country rock.Granitic partial melts segregated and collected at the contact in amelt-rich, nearly crystal-free reservoir adjacent to the initial doleritechilled margin.This dolerite chilled margin was subsequently frac-tured open in the fashion of a trapdoor by the granitic melt, evacuat-ing the reservoir to form an extensive complex of granitic dikes withinthe dolerite sills. At the time of dike injection the dolerite was nearlysolidified. Unusually complete exposures allow the full physical andchemical processes of partial melting, segregation, and dike formationto be examined in great detail.The compositions of the granitic dikesand the textures of partially melted granitic wall rock suggestthat partial melting was characterized by disequilibrium mineraldissolution of dominantly quartz and alkali feldspar rather than byequilibrium melting. It is also unlikely that melting occurred underwater-saturated conditions. The protolith granite contains only! 7vol.% biotite and estimated contact temperatures of 900^9508Csuggest that melting was possible in a dry system. Granite partialmelting, under closed conditions, extended tens of meters away fromthe dolerite sill, yet melt segregation occurred only over less thanone-half a meter from the dolerite chilled margin where the degreeof partial melting was of the order of 50 vol.%. This segregationdistance is consistent with calculated length scales expected in a com-paction-driven process.We suggest that the driving force for compac-tion was differential stress generated by a combination of volume

expansion as a result of granite partial melting, contraction duringdolerite solidification, and relaxation of the overpressure drivingdolerite emplacement. On a purely chemical basis, the extent of meltsegregation necessary under fractional and batch melting to match theRb concentrations between melt and parent rock is a maximum of48 and 83 vol.% melt, respectively.

KEY WORDS: Antarctica; dike injection; disequilibrium; granitepartial melting; silicic melt segregation

I NTRODUCTIONCrustal melting by basalt injection has been proposed asa fundamental mechanism for crustal evolution (Bergantz,1989; Brown, 2006). Numerous field studies of shallow crus-tal rocks have reported melting of country rock near thecontacts of basic intrusions, mainly large dikes and stocks(e.g. Al-Rawi & Carmichael, 1967; Kovach, 1984; Kaczor,1988; Kitchen, 1989; Philpotts & Asher, 1993; Holness &Watt, 2002; Petcovic & Grunder, 2003; Holness et al.,2005). These occurrences provide important insights intothe detailed nature of the melting process, which is mostoften under transient or disequilibrium conditions whereinthe melt composition can be described by a combinationof batch and fractional melting. When the new felsic meltre-injects the adjacent basaltic heat source, the critical pro-cess of mingling and contamination of the basalt by thegranitic melt can also be examined physically and chemi-cally. However, because the most accessible examples of

*Corresponding author. Present address: Lamont^Doherty EarthObservatory, 61 Route 9W, Palisades, NY 10964, USA.Telephone: (845) 365-8662. Fax: (845) 365-8155. E-mail: [email protected] address: Department of Geoscience, University of Nevada atLasVegas, 4505Maryland Pkwy, LasVegas, NV 89154, USA.

! The Author 2007. Published by Oxford University Press. Allrights reserved. For Permissions, please e-mail: [email protected]

JOURNALOFPETROLOGY VOLUME 48 NUMBER11 PAGES 2125^2148 2007 doi:10.1093/petrology/egm054

this process are in the uppermost crust, where the countryrock, at the time of basalt intrusion, is cold, partial meltingis generally limited in spatial extent, with little to nomelt extraction or segregation and subsequent transport.At deeper levels in the crust where background tempera-tures are higher, partial melting is expected to be moreextensive and melt segregation and transport much morecommon. In areas where the crust has undergone rapidthickening (e.g. the Himalayas, the Mojave), high ambienttemperatures should favor extensive partial melting,assimilation and melt migration. These conditions mayalso be approximated in areas of the upper crust wherevoluminous basaltic magma transport and emplacementhas occurred.A pristine natural laboratory to study this process is the

extensive and massive (!10 000 km3) Ferrar dolerite sillcomplex of the McMurdo DryValleys, Antarctica, whichcontains in many areas small granitic dikes cutting adja-cent dolerite sills. The dikes are primarily in associationwith the lowermost sill, the Basement Sill, which wasemplaced in pervasive granitic basement rocks at a paleo-depth of about 3^4 km based on the thickness of overlyingunits capped by comagmatic Kirkpatrick flood basalts(discussed in the next section). In most areas the dikesare actually small dikelets about 3^8mm in width, yetthese dikelets commonly penetrate 50^100m into theadjoining dolerite sill. In light of the near-instantaneousquenching time of small dikelets (!1min), these occur-rences imply that the dolerite was at an insulatingtemperature at or above the melting range of the graniticdikelets. In Wright Valley, in the Bull Pass and EastDais regions, the granitic dikes commonly reach 20^25 cmin width and extend continuously for hundreds of meters,forming, especially in central Bull Pass, an extensive,approximately orthogonal network. The dikes are exten-sive and easy to trace; however, the critical exposures ofthe dikes with the country rock cannot generally befound, owing to poorly exposed or indistinct contact rela-tions. In the upper east wall of central Bull Pass, however,the full field relations are exposed for a complex of theselarge granitic dikes.In this paper we present field, petrographic, and geo-

chemical evidence of this unique example of extensive par-tial melting of water-undersaturated granitic country rockat the contact of a dolerite sill. Prolonged heating by thedolerite melted the adjacent granitic wall rock outwardsome tens of meters. Granitic melt was extracted from itsresidue by lateral compaction and porous melt flow andcollected into a melt-rich reservoir adjacent to the chilledmargin of the dolerite. When the dolerite sill was nearlycompletely solidified, the chilled margin fracturedopen like a trapdoor, allowing granitic melt to penetrateas dikes deep into the interior of the dolerite sill.Compaction, melt extraction, and dike propagation were

driven by the overpressure of dolerite emplacement, melt-ing, and the contraction associated with emplacement ces-sation and cooling. The completeness of the field relationsand the nature of the magmas and country rock involvedmake this natural laboratory of fundamental importanceto understanding the relationship between basaltic intru-sion and the generation of granitic magmas (sensu lato) inthe Earth’s crust.

GENERAL GEOLOGYThe McMurdo Dry Valleys (MDV) in Southern VictoriaLand are part of the Transantarctic Mountains (TAM),which divide the Antarctic continent and delineate aboundary between Archean^Proterozoic crust of the EastAntarctic craton and the Paleozoic^Mesozoic lithosphericblocks of West Antarctica (Borg et al., 1990). The oldestrocks in the MDV region of southern Victoria Land arelate Proterozoic age, multiply deformed, metamorphosed,interbedded marbles, hornsfelses, and schists of theKoettlitz group (Gunn & Warren, 1962; McKelvey &Webb, 1962). These metasediments grade into paragneissor migmatite immediately adjacent to undeformed graniticplutons [Granite Harbor Intrusives of Gunn & Warren(1962)].Up to 15 major granitic (sensu lato) plutons in the MDV

were emplaced during the late Proterozoic to earlyOrdovician (Allibone et al., 1993a, 1993b, and referencestherein). There is much similarity in lithology betweenplutons as well as varied lithology within plutons frommonzodiorite to granodiorite, biotite granodiorite to gran-ite, and monzonite to granite in progressively younger plu-tons (Allibone et al., 1993a). The youngest group of plutonsdisplays crosscutting relationships with numerous swarmsof Vanda mafic and felsic porphyry dikes (Allibone et al.,1993a). Uplift and erosion, following emplacement of theplutons, formed the Kukri Erosion Surface, a peneplainupon which the Devonian to Jurassic age BeaconSupergroup sandstones, conglomerates, siltstones, andminor coal measures were deposited.Jurassic age (!180Ma) breakup of the Gondwana

supercontinent is responsible for extensive intrusionof Ferrar dolerite sills and dikes throughout the TAMand continuing into the Karoo system of South Africa(Elliot & Fleming, 2000; Riley et al., 2006). Comagmaticpyroclastic deposits are further overlain by flood lavasof the Kirkpatrick Basalt (e.g. Elliot & Fleming,2004). Episodic exhumation and uplift of the TAMbegan in the early Cretaceous, but most uplift occurredin the early Cenozoic (Fitzgerald & Stump, 1997;Behrendt et al., 1991). The youngest rocks in the MDVare Cenozoic age McMurdo volcanics, which are alkalibasalts.

JOURNAL OF PETROLOGY VOLUME 48 NUMBER 11 NOVEMBER 2007

2126

FERRAR DOLER ITES ANDHOSTED GRANIT IC DIKESThe Ferrar dolerite sills are typically high aspect ratiosheets that vary from 100 to 500m in thickness and canextend laterally for over 100 km. At least four separatesills have been mapped, with the lowermost sill, theBasement Sill, intruded below the Kukri peneplain inPaleozoic granites, and the uppermost, hypabyssal sill, theMt. Fleming Sill, leading directly to tephra pipes near theancient paleo-surface (Gunn & Warren, 1962; Hamilton,1965). The sills are often remarkably uniform in positionand attitude, and the separation between sills is typicallyhundreds of meters and constant, although in some placesthe sills do touch one another. For example, in Bull Pass theBasement Sill climbs upwards and touches the PeneplainSill. At such contacts, the Basement Sill is always quenchedagainst the Peneplain Sill. The quenched textures arerelatively coarse, suggesting that the Basement Sill wasintruded after the Peneplain Sill, but while the latter wasstill significantly warmer than the ambient temperature.In another area, the Friis Hills, the Basement Sill actuallypushes into the Peneplain Sill, suggesting that thePeneplain Sill was still mushy during contact. It is inferredthat the general sequence of sill emplacement throughoutthe system progressed from the top downwards (Marsh,2004).The dominant lithology of the sill complex is quartz-

normative dolerite. Areally extensive orthopyroxenite, firstidentified by Gunn & Warren (1962) and Hamilton (1965),

is present primarily in the Basement Sill, and in limitedsections of the Peneplain sill, but no higher in the strati-graphic sequence. Marsh (2004) interpreted the large,modally abundant orthopyroxene (including !10 vol.%similar Cpx) to be ‘cognate’ xenocrysts or antecrystsentrained from an earlier or related phase of magmatismwithin the underlying magmatic mush column. Thesecrystals were entrained and transported as a crystal-richsuspension in ascending doleritic magma much as origi-nally hypothesized by Simkin (1967). Both orthopyroxenemode and crystal size within the Basement Sill decreasewith increasing distance away from a central locationnear Bull Pass.The central orthopyroxene zone is hereaftertermed the OPXTongue, following Marsh (2004).The spa-tial pattern of the OPX Tongue suggests that in this areathe Basement Sill was emplaced radially from a centralmagma feeder in Bull Pass (see Figs 1 and 2). Not coinci-dentally, the Basement Sill in the area of Bull Pass has thehighest areal concentration of cross-cutting granitic dikes.Granitic dikes are also observed on the south wall ofWright Valley, the Dais Intrusion, and the south wall ofVictoriaValley. The granitic dikes are numerous, intercon-nected, up to 30 cm wide, and are occasionally over 100min length. Numerous other, much thinner (2^7mm) gra-nitic dikes are found throughout the region. The graniticdikes are mostly associated with the basal section of theBasement Sill and extend upwards into the sill as faras 200m.In central^east Bull Pass, the pattern of interconnected

large granitic dikes at the outcrop scale consists of linear

N

500 m

10 kmAntarctica

Victoria ValleyBull Pass

Wright Valley

McMurdo Dry Valleys

N

N

Granitic country rock

Bull Pass

Basement Silldolerite

GDC

A B

Peneplain Sill dolerite

Dolerite feeder

(a) (b)

161° 162° 163°

77°1

5!77

°30!

77°4

5!

Fig. 1. (a) Satellite photograph of the McMurdo DryValleys of Antarctica. The dashed rectangle is the perimeter of the geological map shownin Fig. 2a. (b) Aerial photograph of Bull Pass.The contact between the Basement Sill and other rocks is traced in black.The contact between theupper lobe of the Basement Sill (dolerite feeder) and the Peneplain Sill in the upper-right hand corner of the photograph should be noted. Therectangle labeled GD is the location of the interconnected granitic dikes mapped in Fig. 3.‘A’,‘B’, and ‘C’ identify locations where granitic dikes inthe dolerite intersect granitic country rock.‘A’ is the outcrop shown in Fig. 4 and discussed in detail in the text.

HERSUM et al. DIKE INJECTION, FERRAR SILLS

2127

segments with near-orthogonal intersections (see Fig. 3).Where visible, the contact between the granitic dike rockand the surrounding dolerite is sharp both at the outcropscale and grain scale. This suggests that, at the time ofdiking, mingling or mixing between granitic melt anddoleritic residual melt was unlikely. No granitic dikes ofthe character found in the Ferrar dolerites have been dis-covered in any non-Ferrar unit in the MDV region andthese granitic dikes should not be confused with the mucholder granitic dikes andVanda felsic porphyry dikes of theGranite Harbor Intrusives (see Fig. 2).

SOURCE OF GRANIT IC DIKES INFERRAR DOLER ITE S I LLSThree locations have been found in eastern Bull Passwhere granitic dikes intersect the contact between theBasement Sill and granitic country rock (labeled ‘A’, ‘B’,and ‘C’ in Fig. 1). The best exposed example is along arise in the upper contact of a lobe of the BasementSill, hereafter referred to as the dolerite feeder, with grani-tic country rock in east Bull Pass as seen looking fromfarther west (see Fig. 2c). The contact at this location

Fig. 2. (a) Geological map of the Bull Pass area modified fromTurnbull et al. (1994). The rectangle labeled GD is the location of the intercon-nected granitic dikes mapped in Fig. 3. (b) Detailed geological map of dolerite feeder region. The circles labeled ‘A’ and ‘B’ identify locationswhere granitic dikes intersect granitic country rock. The circle labeled ‘A’ is the outcrop shown in Fig. 4 and discussed in detail in the text.The black line emanating from ‘A’ identifies the sampling transect described in the text and shown in detail in (d). The red dashed line definesthe cross-section in (d). (c) Photograph of the Basement Sill and dolerite feeder viewed towards the NE. The contact between the dolerite andgranitic country rock is traced in black. The increasing steepness of the uppermost contact from the left towards ‘A’ should be noted.(d) Schematic cross-section perpendicular to the surface trace of the contact at ‘A’ showing granitic rock sample locations referred to in thetext, tables, and other figures. The black and red sample tick marks represent unmelted and partially melted granitic country rock, respectively(see Table 1). The distances are corrected to be a projection from the ground surface onto a plane perpendicular to the contact assuming acontact dip of 458. Granitic dikes in the dolerite sills (ff) are too small to be shown at any of these scales and are not to be confused with mucholder granitic dikes of the Granite Harbor Intrusives (gr) andVanda Dike Swarms (ge).


2128

(77828"5240S,161855"3610E), labeled ‘A’ in Figs 1 and 2, strikesNNE, dips !458 to the north, and coincides with a gentleridge trending NW^SE and cresting at Mt. Orestes. Thepaleo-depth at this location is !2^3 km, or equivalent toa pressure of 60^90MPa, based on the thickness of theoverlying Beacon Supergroup and extrusive Ferrar rocksexposed farther west.At location ‘A’, a granitic dike (!30 cm in apparent thick-

ness) in the sill, with a more northerly strike, obliquelyintersects the contact between the dolerite chilled marginand the granitic country rock (see Fig. 4). The graniticdike terminates at this contact but is continuous with an!5 cm wide granophyre zone or sheet composed of thesame material that parallels the dolerite chilled margin.This granophyre zone decreases monotonically in thick-ness with increasing distance along the contact in eitherdirection from the termination point of the granitic dike.The granophyre zone is imperceptible in hand samplebeyond a distance of 10m along the contact. Outwardfrom the dolerite chilled margin and beyond the grano-phyre zone is an !1cm wide band or sheet of partiallymelted granite with a distinctly dark gray matrix con-taining tiny fragments of the dolerite chilled margin(51cm in size). This sheet is termed the partially meltedgranite breccia. Further outward, the partially meltedgranite changes abruptly to a lighter gray matrix with nofragments of dolerite chilled margin and the gray matrixdecreases in mode with increasing distance from the doler-ite chilled margin. The local granitic country rock here isthe Orestes pluton, which extends outward from the doler-ite chilled margin !200m to a change in lithology thatincludes hornblende. Vanda felsic porphyry dikes are

interspersed in the Orestes granite and because they strikemore northerly than the dolerite contact, the dikes contact,but do not cut because they are much older, the doleritefeeder towards the west (see Fig. 2b). At the terminus ofthe granitic dike and outward from the dolerite chilledmargin, the felsic porphyry dikes are located at theground surface between 9"5 and 29m and again between57 and 96m within the Orestes pluton. It should be notedthat in Fig. 2 andTable 1 these distances are corrected to bea projection from the ground surface onto a plane perpen-dicular to the contact assuming a contact dip of 458. Theolder felsic porphyry dikes are part of a large swarm ofsimilar dikes in the area of ‘A’ that range from tens of cen-timeters to 5150m in thickness. At their greatest abun-dance, the ratio of dikes to country rock is 4:1 (Alliboneet al., 1993a). There is evidence of pervasive partial meltingof both Orestes granite and Vanda felsic porphyry dikesalong the contact with the dolerite feeder.The dolerite feeder at location ‘A’ is !100m thick, con-

tains a wide, coarse OPX Tongue, and has a narrow(!0"5m), well-formed, fine-grained chilled margin. Thelower contact of the dolerite feeder is against a wedge ofgranite of the Bonney pluton that is 50m at its widest andis in contact with another lobe of the Basement Sill onits far side further down-slope. The granite wedgetapers down in thickness with proximity to central BullPass and exhibits evidence of partial melting; however,no granitic dikes have been found originating from thegranite wedge. The granitic dike at location ‘A’ cuts intothe dolerite feeder for a distance of 70m, along the surface,before intersecting another, wider granitic dike withinthe sill that strikes nearly parallel to the dolerite chilledmargin.

PETROGRAPHYThe following sections describe the various rock unitsalong the sampling traverse from the country rock towardsthe dolerite feeder shown as the black line in Fig. 2b.All distances are measured outwards from the doleritechilled margin into the country rock.The relative locationsof the various rock types at distances less than !10 cm fromthe dolerite chilled margin are shown in Fig. 4b.

Orestes graniteThe Orestes granite is a medium-grained, equigranular,hypidiomorphic rock that contains quartz (26 vol.%),alkali feldspar (40 vol.%), plagioclase (27 vol.%), and aminor amount of biotite (52"5mm and !7 vol.%) (Fig. 5).The larger biotite grains frequently host accessory phasesof euhedral to subhedral allanite (50"5mm), euhedralprismatic apatite (75 mm), and subhedral zircon (10 mm).Many biotite grains are altered to chlorite, unquantifiedclay minerals, or are reddish brown and have dustingsof ilmenite and twinned hematite grains along rims.

Latit

ude

East Bull Pass - map view

N

Dike thicknessunknown5 cm10 cm20 cm

Scree

Basement Sill dolerite

Granite dikes

Longitude

161.82 161.83 161.84 161.8577.480

77.478

77.476

77.474

77.472

Fig. 3. Map view of the surface expression of interconnected graniticdikes in the Basement Sill in east Bull Pass. Dike intersectionsare obtained from global positioning system and dike segmentsat the surface are approximately, but not necessarily always, lines asdrawn.


2129

Quartz and feldspar grains (!1"5mm) tend to form aggre-gates (!1"5mm), but each grain exhibits a unique angle ofextinction. Feldspar grains display twinning, are rarelyzoned, and share straight boundaries with quartz, whichusually displays undulatory extinction.

Partially melted Orestes granitePervasive dusting of biotite grains by hematite and lesserilmenite is common and quartz grains develop cuspateboundaries against feldspar grains but no granophyre ispresent in Orestes granite at 23m from the doleritechilled margin. Fine to coarse-grained (10^250 mm)granophyric zones are present primarily between quartzand feldspar grains in Orestes granite at 5"2m from thedolerite chilled margin. The granophyre mode increases,

minor amounts of unaltered biotite in association withgranophyre appear, quartz grain boundaries are moreembayed, and feldspars have a more developed sieve tex-ture at their margins (Fig. 5e) with distance towards thedolerite chilled margin. A sliver of partially meltedOrestes granite, which varies locally in thickness but isalways51cm, is sandwiched between the dolerite chilledmargin on the right and the granophyre zone on the left,discussed below (see Fig. 4b). This rock contains!50 vol.% (visual estimate) of granophyre with restiticcrystals of quartz, feldspar and reacted biotite, and issimilar in texture to the partially melted Orestes graniteshown on the far left of the leftmost photomicrographof Fig. 4b.

Fig. 4. (a) Photographs and schematic diagram (map view) of outcrop at location ‘A’ (see Figs. 1 and 2). These show a granitic dike emanatingfrom a zone of granophyre parallel to the contact with the dolerite chilled margin.The red dotted lines trace the contacts between the rock typeswhere visible. In the right-hand photograph, the granitic dike does not extend into the granitic wall rock. (b) Full thin-section photomicrographsobtained from a flatbed scanner with crossed polarizing films. These two sections are in sequence and collected perpendicular to the contactbetween the dolerite sill and granitic country rock. Sample locations fromTable 1 are indicated only for reference to the various rock types andare not the exact locations.


2130

Partially melted Orestes granite brecciaThere is a sharp transition to a 1cm wide zone of partiallymelted Orestes granite, hereafter referred to as partiallymelted granite breccia. Restite grains of plagioclase havea relatively less pronounced sieve texture compared withthe large feldspar grain aggregates that characterize thepartially melted Orestes granite farther away from thedolerite chilled margin. This band is also distinguishedby finer-grained (!10 mm) interstitial granophyre, whichcontains small plagioclase grains and clusters, up to severalmillimeters in size, of an equal mix of orthopyroxene(!50 mm) and anorthite-rich plagioclase grains represent-ing pieces of reacted dolerite chilled margin.These clustersare not necessarily coherent and orthopyroxene grains areobserved in the interstitial granophyre far from the clus-ters. Another population of larger orthopyroxene grains

(!200 mm) is also dispersed throughout the interstitialgranophyre.

Vanda felsic porphyryVanda felsic porphyry dikes contain !20 vol.% pheno-crysts of alkali feldspar, plagioclase, quartz, biotite, andhornblende. Feldspar grains (53"5mm) are subhedral,commonly twinned, and range from showing no to moder-ate zoning. Quartz grains (52mm) are anhedral, haveslight undulatory extinction, and commonly form aggre-gates of a few grains. Biotite and hornblende grains(52mm) often host apatite and zircon grains, are greenishbrown, altered to chlorite or clay minerals, and dustedwith opaque minerals. Groundmass is fine-grained granitecomposed of near-equal amounts of anhedral feldsparand quartz grains that are difficult to distinguish from

Table 1: XRF results for major and trace elements

Sample: TH28 TH29 TH30 TH6 TH5 TH4 TH3 A-399 TH2 TH1 A-395-4 A-395-1 A-395-3 A-393 A-395-2 A-396 A-397 A-398

Rock: gt gt pgt pge pge pgt pgt pgt pgt pgt pgt pgt pgtb gran gran gd gd gd

Distance: 68 39 23 14"1 7"1 5"2 3"4 2(?) 1"2 0"5 0"123 0"067 0"056 n.a. n.a. n.a. n.a. n.a.

SiO2 71"75 69"44 72"19 69"06 69"39 70"47 71"44 72"42 70"67 71"64 72"09 69"57 68"91 73"69 73"46 74"93 73"69 67"70

Al2O3 13"66 14"19 13"48 14"16 14"51 13"69 13"75 13"79 13"97 13"95 13"90 15"58 15"84 13"04 12"99 12"21 12"88 11"82

CaO 1"94 2"09 1"97 2"09 2"87 2"09 2"06 1"82 2"03 2"14 2"52 3"22 3"77 1"18 1"13 0"72 0"91 2"60

MgO 0"41 0"40 0"41 0"51 0"61 0"41 0"41 0"40 0"49 0"51 0"47 0"57 0"80 0"36 0"27 0"26 0"33 0"47

Na2O 2"97 3"19 3"00 3"23 3"44 3"10 3"06 3"03 3"06 3"03 3"18 3"68 3"84 2"55 2"57 2"95 3"23 3"31

K2O 4"61 4"69 4"53 4"92 4"64 4"54 4"65 4"84 4"84 4"74 3"85 3"36 2"55 6"08 5"99 6"00 5"86 3"96

Fe2O3 2"87 2"87 2"77 2"88 3"27 2"89 2"85 2"48 2"93 2"98 2"89 2"94 3"45 2"57 2"48 2"12 2"21 8"46MnO 0"04 0"04 0"04 0"05 0"05 0"04 0"04 0"03 0"04 0"04 0"04 0"04 0"05 0"03 0"03 0"02 0"02 0"10

TiO2 0"29 0"30 0"28 0"37 0"41 0"30 0"29 0"28 0"31 0"30 0"31 0"29 0"31 0"35 0"33 0"26 0"32 0"79

P2O5 0"05 0"06 0"05 0"09 0"10 0"06 0"06 0"06 0"06 0"06 0"06 0"06 0"06 0"06 0"05 0"03 0"04 0"19

Cr2O3 0"03 0"04 0"04 0"03 0"04 0"03 0"04 0"01 0"03 0"03 0"01 0"01 0"01 0"01 0"01 0"01 0"01 0"01

LOI 0"60 1"00 0"75 0"85 0"90 0"85 0"90 0"75 0"60 0"50 0"70 0"70 0"45 0"10 0"60 0"55 0"45 0"50

Sum 99"50 98"58 99"78 98"41 100"40 98"76 99"82 99"91 99"31 100"20 100"02 100"02 100"04 100"02 99"91 100"06 99"95 99"91

Rb 101 116 101 189 182 104 107 109 107 105 91 84 58 137 136 230 208 154

Sr 394 413 383 359 355 416 405 395 395 397 503 443 521 229 243 100 126 132

Y 15 12 13 27 29 14 12 12 15 16 13 12 13 14 15 28 29 51

Zr 194 216 199 233 246 210 206 186 211 214 218 223 252 104 117 132 164 199

Nb 9 11 9 16 16 9 9 11 11 10 10 9 7 8 9 12 13 16

Ba 1850 1640 1680 860 718 1860 1830 1750 1790 1770 1190 1100 1120 1380 1450 596 806 731

Cr — — — — — — — 55 — — 5 7 6 12 55 55 8 55

Co — — — — — — — 4 — — 6 5 8 4 5 6 3 15

Ni — — — — — — — 52 — — 52 52 2 52 52 52 52 7

Zn — — — — — — — 59 — — 65 66 68 51 51 26 22 67

Cu — — — — — — — 52 — — 4 5 6 2 3 4 4 47

Sample locations are shown in Figs 2 and 4. gt, Orestes granite; pgt, partially melted Orestes granite; pge, partially meltedVanda felsic porphyry; pgtb, partially melted Orestes granite breccia; gran, granophyre zone; gd, granitic dike. Thedistances (m) from the dolerite chilled margin are corrected to be a projection from the ground surface onto a planeperpendicular to the contact assuming a contact dip of 458. All Fe is reported as Fe2O3. LOI, loss on ignition; n.a. notapplicable.


2131

granophyre associated with partial melting. Therefore, noestimate of the mode of granophyre in theVanda felsic por-phyry dikes rock was attempted. Perhaps, in future work,cathodoluminescence could be used to differentiate quartzgenerations.

Granophyre zoneThe granophyre zone extends continuously into the gra-nitic dike at location ‘A’. Along the sampling traverse, thegranophyre zone is 5 cm wide and is in contact with par-tially melted Orestes granite breccia on the left and in con-tact with the thin sliver (51cm wide) of partially meltedOrestes granite attached to the dolerite chilled margin onthe right as shown in Fig. 4.The granophyre zone is a holo-crystalline, porphyritic granophyre which hosts510 vol.%large quartz, plagioclase, and alkali feldspar grainsand coherent reacted biotite in addition to much largerpolycrystalline aggregates of these same phases. Theselarge grains are texturally similar to the restite grains of

the partially melted Orestes granite that border both sidesof the granophyre zone.

Granitic dikes in Ferrar dolerite sillsAll of the granitic dike rocks observed and sampled areporphyritic and host from51 to !10 vol.% large grainsand polycrystalline aggregates in a groundmass mix offine-grained granite and granophyre. As in the granophyrezone, the large grains are texturally similar to the restitephases in the partially melted granitic country rocks. Therelative abundance of fine-grained granite (0"15^0"5mm)vs granophyre (!0"25mm) in the groundmass of eachsection varies and one can be found without the other.The most common large grains are primarily feldsparand quartz, usually 53mm in size but with rare grains47mm. Some thin sections contain minor amounts ofreacted biotite (55 vol.%), reacted hornblende (51 vol.%),and opaque minerals (51 vol.%). Alteration phases, 510vol.% in some sections, occur mostly as breakdown pro-ducts of hydrous phases and include chlorite and prehnite.

Fig. 5. (a) Unmelted Orestes granite (TH28) in plane-polarized light. (b) Cuspate quartz boundary with feldspars at onset of partial meltingof Orestes granite (TH30) in cross-polarized light. (c) Back-scattered electron image of biotite site in partially melted Orestes granite (TH29).(d) Partially melted Vanda felsic porphyry (TH6) in plane-polarized light. (e) Edge of a restitic plagioclase grain with a sieve texture in par-tially melted Orestes granite in cross-polarized light (A-395-1). (f) Stained (red, plagioclase; yellow, alkali feldspar) partially melted Orestesgranite (A-395-1) in plane-polarized light. (g) SEM image of unreacted biotite surrounded by granophyre in partially melted Orestes granite(A-395-1). (h) SEM image of contact between partially melted Orestes granite breccia (left) and granophyre zone (right) where the orangecurves sketch the perimeter of restitic plagioclase grains (A-395-3). (i) SEM image of opx replacing cpx in dolerite chilled margin.


2132

Dolerite chilled marginThe dolerite chilled margin contains a texturally andmineralogically distinctive !2"5mm wide planar zoneimmediately adjacent and parallel to the contact with thesliver of partially melted Orestes granite. This zone, here-after referred to as the dolerite reaction zone, differs fromthe rest of the dolerite chilled margin by a marked increasein grain size, the presence of biotite and orthopyroxene,and lack of clinopyroxene.

WHOLE -ROCK AND MINERALCHEMISTRYAnalytical methodsAll whole-rock chemical analyses were performed byusing X-ray fluorescence (XRF; SGS Canada Inc. MineralServices). Detection limits for major oxides are 0"01% andfor trace elements are 2 mg/g, except for Ba, which had alimit of detection of 20 mg/g. Mineral compositionswere quantified by using wavelength-dispersive spectro-metry (WDS) and X-ray maps by using energy-dispersive spectrometry (EDS) on the JEOL Superprobe8600 electron probe microanalyzer (EPMA) at JohnsHopkins University. Natural and synthetic mineral stan-dards and a ZAFcorrection scheme were used (Armstrong,1988).Abeamcurrentof 20 nA,accelerating voltageof 15 kV,and abeamdiameter of 0^10 mmwere employed.

Bulk compositionWhole-rock major and trace element chemical data fromsamples of unmelted Orestes granite, partially meltedOrestes granite, partially melted Vanda felsic porphyry,partially melted Orestes granite breccia, granophyre zone,and granitic dike rocks are presented in Table 1. All sam-ples were collected near location ‘A’ (Fig. 1). The distancecolumn refers to the distance of the sample location intothe dolerite sill from the chilled margin in a transect ortho-gonal to the strike of the dolerite chilled margin contact(i.e. the black line in Fig. 2b). It should be noted that, asbefore, these distances are corrected to be a projectionfrom the ground surface onto a plane perpendicular to thecontact assuming a contact dip of 458. All samples of par-tially melted granite are from the Orestes pluton, exceptsamples TH5 and TH6, which are from a Vanda felsicporphyry dike.Whole-rock normative compositions plotted on an

anorthiteâlbiteôrthoclase ternary and a quartzâlbiteânorthite ternary are shown in Figs 6 and 7, respectively,for unmelted Orestes granite, partially melted Orestesgranite, partially melted Vanda felsic porphyry, partiallymelted Orestes granite breccia, granophyre zone, andgranitic dike samples. A tight cluster of similar composi-tions of unmelted Orestes granite and partially meltedgranitic (Orestes and Vanda) samples from450 cm fromthe dolerite chilled margin is distinct from the composi-tional trend of decreasing orthoclase component of

AbOr

An

gran - Granophyre zone (next to dolerite chilled margin)

gd - Granitic dike

pgt & pgtb - Partially melted Orestes granite and granite breccia < 13 cm from dolerite chilled margin

pgt & pge - Partially melted Orestes granite and Vanda felsic porphyry > 50 cm from dolerite chilled margin

> 50 cm

A-395-3 (5.6 cm)

A-395-1 (6.7 cm)

A-395-4 (12.3 cm)

All distances to chilledmargin towards the left

DoleriteDolerite chilled margin

Granitic dike

PartiallymeltedOrestesgranite

Partially melted Orestes granite breccia

Granophyre zone

pgtpgt & pge

grangd

pgtb

Fig. 6. Whole-rock normative compositions on a anorthiteâlbiteôrthoclase ternary with accompanying schematic diagrams showing approx-imate sample locations at outcrop ‘A’ (see Fig. 2). Orthoclase component decreases with distance towards the chilled margin in pgt and pgtbsamples512"3 cm from the dolerite chilled margin. However, gran samples are enriched in orthoclase component with respect to gt, pgt, andpge samples450 cm from the dolerite chilled margin.


2133

partially melted Orestes granite samples beginning at leastat a distance of 12"3 cm and closer to the dolerite chilledmargin (see Fig. 6). The compositions of the granophyrezone define another tight cluster aligned with the previoustrend but are more enriched in orthoclase component.A less obvious but similar trend is shown by Fig. 7b;

however, the quartz component is relatively constant. Thegranophyre zone samples plot to the right of the water-saturated granite minimum at P(H2O)# 200MPa (Tuttle& Bowen, 1958). With decreasing An/(An$Ab) content,the eutectic composition crosses to the upper right of,but does not intersect, the granophyre zone samples.

Ab Or

Restitic feldspar

Granophyre feldspar

Whole rock norms

An

Or

Ab Or

QG

ranophyre

composition

0.21

0.36

m

ee

0

An/(An+Ab)=

Whole rock norms

Granophyrefeldspar

gran - Granophyre zone (A-393, A-395-2)pgtb - Partially melted Orestes granite breccia (A-395-3)

pgt - Partially melted Orestes granite (left of breccia)pgt - Partially melted Orestes granite (A-395-1)pgt - Partially melted Orestes granite (A-399)pgt - Partially melted Orestes granite (TH-4)gt - Unmelted Orestes granite (TH-29)

pgt - Sliver of partially melted Orestes granite

Whole rock norms

gran - Granophyre zone (next to dolerite chilled margin) gd - Granitic dike

pgt & pgtb - Partially melted Orestesgranite and granite breccia < 13cm from chilled margin

pgt & pge - Partially melted Orestesgranite and Vanda felsic porphyry> 50 cm from dolerite margin

Dolerite

Granitic dike

Partially meltedOrestes granite

Whole rock samples

Feldspar analyzed

39 5.2 ~2 0.12

Distance from doleritechilled margin (m)

Partially meltedOrestes granitebreccia

Doleritechilledmargin

0.06

Feldspar analyzed

(a)

(b)

Fig. 7. (a) Restitic and granophyre feldspar compositions in the anorthiteâlbiteôrthoclase ternary with whole-rock normative compositions.With increased degrees of granitic partial melting (less distance from dolerite chilled margin), alkali feldspar in the granophyre is enriched inorthoclase component whereas restitic alkali feldspar is enriched in albite component. Restitic plagioclase is enriched in anorthite componentwith increased degrees of granitic partial melting. (b) Whole-rock normative compositions and feldspar granophyre compositions inthe quartzâlbiteôrthoclase ternary. The line between the quartz corner and the abôr join represents all the possible compositions of thegranophyre assuming simple mixing between these two phases. Water-saturated granite solidus curves are shown for P(H2O)# 2 kbar andAn/(An$Ab) values of zero (Tuttle & Bowen, 1958), and 0"21 and 0"36 (von Platen, 1965). The circles labeled ‘m’ and ‘e’ represent minimumand eutectic melt compositions. The mean An/(An$Ab) value for partially melted Orestes granite450 cm from the dolerite chilled margin(pgt) is !0"27.


2134

Harker diagrams of major and trace elements (Fig. 8) alsoreflect the compositional trends observed in the ternarydiagrams. For reference, also shown is a least-squares linefitted between samples of partially melted Orestes graniteat512"3 cm from the contact and the granophyre zone. Formost elements, the partially melted granitic (Orestes andVanda) samples 450 cm from the contact plot near thisline. Granitic dike samples also plot near this line, particu-larly for oxides, Al2O3, CaO, and MgO.

Dolerite reaction zoneFive, 512%512 pixel X-ray maps at 100% magnificationwith a cumulative 1s count time per pixel were collectedperpendicular to the contact between the sliver of partiallymelted Orestes granite and the dolerite chilled margin.These maps include the sliver of partially melted Orestesgranite, dolerite reaction zone, and dolerite, covering a dis-tance of about 5mm. Line averages parallel to the contact,representing averages of 51 columns, for CaO and SiO2 areshown in Fig. 9. The purpose of this windowing averageis to smooth out the concentrations of a single columnaverage, which is an attempt to mask the uncertainty

associated with averaging over an area of insufficient size.The line averages demonstrate relative uniformity of oxidecontent (wt%) within zones but reveal large jumpsbetween rock types. Line averages for other oxides arerelatively constant when compared with CaO and SiO2

and are not shown.

Mineral compositionsFeldsparPlagioclase is present in all rock units whereas alkali feld-spar is restricted to unmelted Orestes granite, partiallymelted Orestes granite and breccia, partially meltedVanda felsic porphyry, granophyre zone, and granitic dikerocks. Plagioclase occurs as a primary phase in unmeltedOrestes granite, as a primary phase in restite in partiallymelted Orestes granite, and as a reaction product formedfrom the breakdown of biotite in the presence of otherphases. Plagioclase is also in the granophyre zoneand granitic dikes as restite grains entrained frompartially melted Orestes granite and as a primary phasein the dolerite reaction zone and unreacted dolerite

Al2O3 (wt%)

13

14

15

16

69 71 73 75

CaO (wt%)

1

2

3

4

69 71 73 75

0.2

0.4

0.6

0.8

MgO (wt%)

TiO2 (wt%)

0.27

0.29

0.31

0.33

0.35

Rb (ppm)

80

120

160

200

240

Fe2O3 (wt%)

2.0

2.4

2.8

3.2

3.6Na2O (wt%)

2.4

2.8

3.2

3.6

4.0

3

4

5

6

7K2O (wt%)

Sr (ppm)

150

250

350

450

550

Partially melted Orestes granite <12.3 cm from chilled margin

Partially melted Orestes granite >50 cm from chilled margin

Granophyre zone

Granite dike

SiO2 (wt%)

69 71 73 75SiO2 (wt%)

69 71 73 75SiO2 (wt%)

69 71 73 75SiO2 (wt%)

69 71 73 75SiO2 (wt%)

69 71 73 75SiO2 (wt%)

69 71 73 75SiO2 (wt%)

SiO2 (wt%)69 71 73 75

SiO2 (wt%)

Fig. 8. Harker diagrams for selected major and trace elements. Lines represent a least-squares fit between partially melted Orestes granite (m)and granophyre (œ). All of the plotted data are fromTable 1 and do not include partially meltedVanda felsic porphyry dike samples.


2135

chilled margin. Representative feldspar compositions areshown inTable 2 in oxide weight per cent and in cations pereight oxygens.With proximity to the dolerite chilled marginrestitic plagioclase is on average more anorthitic in the par-tiallymeltedOrestes granite, but is less anorthitic thandoler-ite reaction zone and dolerite chilled margin plagioclase asseen on an anorthite^albite^orthoclase ternary ternary(Fig. 7). Restitic alkali feldspar in partially melted Orestesgranite is on average more K-rich with distance towardsthe dolerite chilledmargin.The sieve texture of the feldsparsmost probably reflects melting along cleavage planes(Philpotts & Asher, 1993, and references therein). There isno evidence for strong major element compositional varia-tion from core to rimwithin feldspar grains.Alkali feldspar is the only feldspar in the granophyre,

with the exception of two analyzed plagioclase grainsthought to be pieces broken free from sieve-textured restiticplagioclase grains. Alkali feldspar from the granophyrezone is on average more K-rich with distance towards thedolerite chilled margin (Fig. 7). Two-feldspar thermometryof restite grains yields temperatures of 635^7008C usingSOLVCALC (Wen & Nekvasil, 1994) and clearly recordsequilibration at sub-solidus temperatures.

PyroxenePyroxene is present in the partially melted Orestes granite,partially melted Vanda felsic porphyry, within pieces ofdolerite chilled margin contained in partially meltedOrestes granite breccia, and in the dolerite reaction zoneand dolerite chilled margin. Orthopyroxene occurs aseither a reaction product from biotite breakdown, a pro-duct from reactive dissolution of clinopyroxene in thedolerite chilled margin at the contact, or as xenocrysts in

the dolerite sills and, relatively less commonly, in the doler-ite chilled margin. Representative pyroxene compositionsare shown in Table 3 in oxide weight per cent and incations per six oxygens. Orthopyroxene is more Mg-richin the dolerite reaction zone than in partially meltedOrestes granite (Fig. 10). It also increases systematically inMg within the dolerite reaction zone, reaching a maxi-mum at the boundary with the unreacted dolerite chilledmargin (Fig. 11).Clinopyroxene occurs solely in the dolerite sill, dolerite

chilled margin, and within the unreacted centers of somepieces of dolerite chilled margin within the partiallymelted Orestes granite breccia. Clinopyroxene and ortho-pyroxene near the sharp contact between the dolerite reac-tion zone and unreacted dolerite chilled margin plot on anexperimentally determined tie-line (Lindsley, 1983;Andersen et al., 1993) indicating that the two pyroxeneswere in equilibrium at temperatures of 900^9508C at thecontact (Table 3).

BiotiteBiotite is present in the unmelted Orestes granite and par-tially melted granitic (Orestes andVanda) rocks, as a crys-tallization product, and in the dolerite chilled margin.Most biotite grains are partially altered to clays and givepoor EMPA totals. However, a representative biotite com-position is given inTable 4 in oxide weight per cent and incations per 11 oxygens. Magmatic magnetite and ilmeniteoccur primarily within reacted biotite; however, sub-solidus hydrothermal alteration is the likely cause for abun-dant hematite within reacted biotite. Representative Fe^Tioxide compositions are given in Table 5 in oxide weightper cent and in cations per three or four oxygens. The oxi-dation state of Fe for each analysis was calculatedby simultaneously solving linear equations for the totalnumber of cations and charge balance. Rare coexistingmagnetite and ilmenite and small grain size precludedtheir use for thermometry.

NATURE OF GRANITE PARTIALMELT ING AND REACT IONRELAT IONSH IPSA significant amount of field and laboratory work has beencarried out to understand how granitic rocks undergopartial melting (Brown & Rushmer, 2006, and referencestherein). Observations of natural partial melting haveincluded either partial melting and disaggregation of gran-ite or granodiorite xenoliths (Bacon, 1992; Green, 1994)or partial melting of granite, tonalite, gneiss, pegmatite,or metasediments at the walls of mafic dikes, plugs, sills,and feeder conduits at depths between 2 and 10 km(Al-Rawi & Carmichael, 1967; Kaczor, 1988; Kitchen,1989; Philpotts & Asher, 1993; Holness & Watt, 2002;

0

2

4

6

8

10

"1 0 1 2 3 455

60

65

70

75

80

85

Distance (mm)

Oxide wt%C

aO (

wt%

)

SiO

2 (w

t%)

CaO

SiO2

Dolerite reaction zone

Dolerite chilledmargin

Granophyrein sliver ofpartiallymeltedgranite

cpx opx + melt

Fig. 9. Variations in CaO and SiO2 across the contact between thesliver of partially melted Orestes granite and dolerite chilled margin.These variations reflect reactive dissolution of clinopyroxene and con-comitant precipitation of orthopyroxene in the dolerite reaction zone.Elemental variations are obtained from line averages over a 51column wide window of X-ray maps from EDS analysis.


2136

Petcovic & Grunder, 2003; Holness et al., 2005). The mostsalient observation in all these studies, despite varied pro-tolith compositions, is the occurrence of multiple, compo-sitionally distinct melts (as glass or granophyre)representing the products of melting reactions controlledby local mineral assemblages and not by general equili-brium minimum melting. Some of these studies haveobserved the breakdown of hydrous phases by dehydrationmelting (Patin‹ o Douce, 1999). They stress the importanceof dehydration reactions for the initiation of melting,further melt-fluxing caused by the release of water, andpermeability enhancement by crack generation owing tothe positive volume change during partial melting, albeitwith little or no melt mobilization [e.g. Philpotts & Asher(1993) and Holness & Watt (2002), respectively].During partial melting of the Orestes granite, all major

phases including plagioclase, alkali feldspar, and quartzwere consumed to varying degrees. Biotite was consumedcompletely. Stages of biotite breakdown as a result ofdehydration melting are observed in the partially meltedOrestes granite. In order of increasing degree of

breakdown and, therefore, possibly melting, these stagesinclude a change of color from greenish brown to reddishbrown, dusting of Fe^Ti oxides at biotite rims, and thecomplete breakdown of biotite to an intergrowth of Fe^Tioxides and plagioclase with orthopyroxene bordering theoriginal biotite grain boundary. This sequence is similarto that reported in other studies (e.g. Al-Rawi &Carmichael, 1967; Brearley, 1987). The absence of alkalifeldspar as a reaction product of biotite breakdown iscommon in other field (e.g. Petcovic & Grunder, 2003)and experimental studies (Naney, 1983; Johnson &Rutherford,1989; Patin‹ o Douce & Beard,1995). Alkali feld-spar may not be a stable reaction product, possibly becauseof a high water content of the initial dehydration melts andalso the low pressure of melting (Naney, 1983; Johnson &Rutherford, 1989). In this study, the sequence of biotitebreakdown is generally observed in partially meltedOrestes granite as one approaches the dolerite chilledmargin. However, evidence of more than a single stagecan be observed within single samples, particularly withdistance away from the dolerite chilled margin. Alteration

Table 2: Representative feldspar analysis in oxide weight per cent and cations per eight oxygens

Sample: 1131G1 1132G1 1158G1 1159G1 162G1 163G1 200G1 201G1 1174G2 1106G2 1114G2 981G3 982G3 639G4 141G5

WR no.: TH30 TH30 TH4 TH4 A-395-4 A-395-4 A-395-1 A-395-1 TH4 A-395-1 A-393 A-395-4 A-395-4 n.a. n.a.

Distance: 23 23 5"2 5"2 0"123 0"123 0"067 0"067 5"2 0"067 n.a. 0"123 0"123 n.a. n.a.

SiO2 62"49 66"29 62"60 65"73 61"78 65"68 66"33 57"33 66"16 66"07 66"23 58"30 57"75 50"11 55"12

Al2O3 24"19 18"54 23"62 18"60 24"56 18"52 18"43 26"56 18"60 18"61 18"33 26"00 26"68 32"94 28"05

CaO 6"07 0"10 5"50 0"09 6"33 0"04 0"15 8"70 0"15 0"15 0"10 7"70 8"18 14"92 10"90

MgO 0"00 0"00 0"00 0"00 0"02 0"00 0"00 0"00 0"01 0"00 0"00 0"00 0"00 0"02 0"12

Na2O 7"53 2"97 7"83 2"71 7"67 2"11 1"83 6"50 2"48 1"99 1"65 7"09 6"82 2"81 4"99

K2O 0"68 12"15 0"85 13"20 0"39 13"34 14"25 0"22 13"40 13"82 14"36 0"21 0"20 0"07 0"31

Fe2O3 0"10 0"00 0"18 0"06 0"21 0"13 0"10 0"16 0"08 0"09 0"08 0"82 0"82 0"54 0"50MnO 0"00 0"00 0"00 0"04 0"01 0"02 0"00 0"00 0"01 0"00 0"02 0"00 0"02 0"00 0"00

TiO2 0"00 0"05 0"01 0"00 0"02 0"06 0"05 0"02 0"05 0"02 0"04 0"00 0"03 0"04 0"05

Cr2O3 0"00 0"02 0"00 0"03 0"01 0"00 0"03 0"00 0"03 0"02 0"00 0"00 0"01 0"00 0"00

Total 101"07 100"12 100"58 100"46 100"99 99"92 101"17 99"49 100"96 100"77 100"81 100"13 100"52 101"43 100"04

Si 2"75 3"01 2"77 3"00 2"72 3"00 3"01 2"58 3"00 3"00 3"01 2"61 2"58 2"26 2"49

Al 1"25 0"99 1"23 1"00 1"28 1"00 0"99 1"41 0"99 1"00 0"98 1"37 1"41 1"75 1"49

Ca 0"29 0"01 0"26 0"00 0"30 0"00 0"01 0"42 0"01 0"01 0"01 0"37 0"39 0"72 0"53

Mg 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"01Na 0"64 0"26 0"67 0"24 0"66 0"19 0"16 0"57 0"22 0"18 0"15 0"62 0"59 0"25 0"44

K 0"04 0"70 0"05 0"77 0"02 0"78 0"82 0"01 0"78 0"80 0"83 0"01 0"01 0"00 0"02

Fe 0"00 0"00 0"01 0"00 0"01 0"01 0"00 0"01 0"00 0"00 0"00 0"03 0"03 0"02 0"02

O 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8

Total 12"97 12"98 12"98 13"01 12"98 12"98 12"99 13"00 13"00 12"99 12"98 13"01 13"02 12"99 12"99

WR no. and Distance rows correspond, respectively, to sample number and distance (m) from dolerite chilled marginin Table 1. All Fe is reported as FeO.1Restite feldspar. 2Granophyre feldspar. 3Biotite site feldspar. 4Dolerite reaction zone feldspar. 5Dolerite chilled marginfeldspar.


2137

of biotite to chlorite is probably a hydrothermal phenom-enon and occurs after the partial melting event; chlorite isunstable at high temperature (Graphchikov et al., 1999).The presence of unreacted biotite (i.e. no breakdown pro-ducts) in partially melted Orestes granite with higheramounts of granophyre is inferred to be a secondary crys-tallization product during solidification of the granitic melt(see Fig. 5g).The relative contributions of plagioclase, compared with

alkali feldspar, to the melt composition, as indicated by thewhole-rock composition of the granophyre zone, is small asobserved by the anorthiteâlbiteôrthoclase ternary andquartzâlbiteânorthite ternary (Fig. 7). That is, the albitecomponent added to the melt has evidently come from thealkali feldspar and not the plagioclase. The whole-rockcompositions of the granophyre zone are depleted in both

albite and anorthite component relative to the compositionof the unmelted Orestes granite and partially meltedOrestes granite samples collected at 450 cm from thedolerite chilled margin (Fig. 6). However, the whole-rockgranophyre zone compositions do not intersect a linerepresenting simple mixing of alkali feldspar domains andquartz (Fig. 7b). This suggests either that the whole-rockcomposition of the granophyre zone does not representthe true composition of the granophyre zone because ofthe inclusion of grains of restitic plagioclase or that manyof the plagioclase domains have been subsequently hydro-thermally altered. The two plagioclase grains plotted(Fig. 7) are, as mentioned above, probably unmelted frag-ments of restitic plagioclase. Given that the An/(An$Ab)value for the host Orestes granite is !0"27, then the posi-tion of the whole-rock granophyre zone composition in

Table 3: Representative pyroxene analysis in oxide weight per cent and in cations per six oxygens

Sample: 669G1 601G2 624G3 641G4 130G5 131G5 132G5 135G5 136G5 137G5 138G5 653G5 654G5 655G5 656G5 657G5 658G5 660G5 661G5

Mineral: opx opx opx opx opx cpx opx cpx opx cpx cpx opx cpx cpx opx cpx opx opx cpx

SiO2 50"26 51"52 50"97 52"17 53"34 52"76 53"15 52"87 52"87 52"71 52"58 52"90 51"92 52"09 52"14 52"21 52"40 52"08 52"34

Al2O3 0"92 0"60 0"77 0"75 0"61 1"19 0"61 1"17 0"53 1"00 0"94 0"54 1"29 1"22 0"60 1"09 0"62 0"50 1"23

CaO 0"63 0"89 0"66 0"62 4"26 18"85 1"31 19"68 1"43 19"64 21"19 1"45 18"72 17"29 1"45 20"24 1"31 1"77 15"89

MgO 14"60 16"54 14"38 18"46 18"29 13"56 18"49 13"33 19"12 13"46 12"72 19"44 13"23 13"66 18"62 12"98 18"63 18"08 13"79

Na2O 0"04 0"02 0"00 0"03 0"06 0"19 0"03 0"23 0"01 0"23 0"25 0"03 0"21 0"17 0"00 0"20 0"00 0"01 0"18

K2O 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"01 0"01 0"00 0"00 0"00 0"00 0"00 0"00

Fe2O3 31"06 29"52 32"22 26"68 24"41 13"43 27"52 13"13 26"77 13"23 11"55 25"14 13"23 14"88 26"00 12"47 26"68 26"57 16"08

MnO 0"80 0"74 0"80 0"57 0"48 0"36 0"57 0"34 0"51 0"31 0"25 0"49 0"34 0"34 0"51 0"30 0"50 0"52 0"36

TiO2 0"17 0"13 0"20 0"19 0"25 0"42 0"24 0"48 0"24 0"42 0"33 0"27 0"42 0"42 0"32 0"41 0"25 0"24 0"38

Cr2O3 0"02 0"01 0"00 0"00 0"02 0"12 0"05 0"06 0"00 0"04 0"06 0"02 0"24 0"13 0"03 0"03 0"02 0"03 0"19

Total 98"49 99"96 99"99 99"48 101"72 100"87 101"95 101"28 101"47 101"03 99"87 100"27 99"60 100"19 99"62 99"93 100"41 99"79 100"44

Eq. pair — — — — 131G 132G — 136G — 136G 136G 654G — 656G — 658G — 661G —

T (8C) — — — — 1240& 116 906& 33 — 929& 16 — 936& 21 902& 24 957& 10 — 954& 23 — 902& 12 — 1005& 12 —

Si 1"99 1"99 1"99 1"99 1"99 1"97 1"99 1"97 1"98 1"97 1"98 1"99 1"97 1"97 1"99 1"97 1"99 1"99 1"97

Al 0"04 0"03 0"04 0"03 0"03 0"05 0"03 0"05 0"02 0"04 0"04 0"02 0"06 0"05 0"03 0"05 0"03 0"02 0"06

Ca 0"03 0"04 0"03 0"03 0"17 0"76 0"05 0"79 0"06 0"79 0"86 0"06 0"76 0"7 0"06 0"82 0"05 0"07 0"64

Mg 0"86 0"95 0"84 1"05 1"02 0"76 1"03 0"74 1"07 0"75 0"72 1"09 0"75 0"77 1"06 0"73 1"05 1"03 0"78

Na 0"00 0"00 0"00 0"00 0"00 0"01 0"00 0"02 0"00 0"02 0"02 0"00 0"02 0"01 0"00 0"01 0"00 0"00 0"01

Fe 1"03 0"95 1"05 0"85 0"76 0"42 0"86 0"41 0"84 0"41 0"36 0"79 0"42 0"47 0"83 0"39 0"85 0"85 0"51

Mn 0"03 0"02 0"03 0"02 0"02 0"01 0"02 0"01 0"02 0"01 0"01 0"02 0"01 0"01 0"02 0"01 0"02 0"02 0"01

Ti 0"01 0"00 0"01 0"01 0"01 0"01 0"01 0"01 0"01 0"01 0"01 0"01 0"01 0"01 0"01 0"01 0"01 0"01 0"01

Cr 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"00 0"01 0"00 0"00 0"00 0"00 0"00 0"01

O 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6

Total 9"99 9"99 9"98 9"99 9"99 10"00 9"99 10"00 10"00 10"00 10"00 9"99 10"00 10"00 9"99 10"00 9"99 9"99 9"99

Eq. pair, pairs of neighboring pyroxenes in the dolerite chilled margin for thermometry. All Fe is reported as FeO.1Pyroxene in granophyre of partially melted Orestes granite 12"3 cm from the dolerite chilled margin.2Pyroxene in reacted piece of dolerite chilled margin in partially melted Orestes granite breccia.3Pyroxene in granophyre of partially melted Orestes granite breccia.4Pyroxene in dolerite reaction zone in dolerite chilled margin.5Pyroxene at reaction front between dolerite reaction zone and dolerite chilled margin.


2138

Fig. 7b is shifted to the albite apex away from the expectedmelt composition between the two eutectic compositions(circles labeled ‘e’) at An/(An$Ab) values of 0"21 and0"36. This suggests that partial melting is characterizedby disequilibrium between melt and solid.The 2"5mm wide dolerite reaction zone at the outer

edge of the dolerite chilled margin was initially normalchilled dolerite. As a result of the large chemical potentialgradient at near-solidus temperatures across this contact,however, reactive dissolution of clinopyroxene with con-comitant precipitation of orthopyroxene occurred inresponse to diffusion in both directions. A similar reaction

relationship in the chilled margin of a diabase feeder dikewas reported by Philpotts & Asher (1993) and is discussedbelow. A scaling of the diffusion equation can be used toestimate the duration for the growth of this reaction zone,provided the melt produced in the dolerite as a result ofreactive dissolution is beyond the (interconnected) perco-lation threshold. If chemical transport is governedpurely by a realistic chemical diffusivity in the melt ofD!10^10m2/s, operating over a length scale (L) of 2mm,then the diffusion time scale (L2/D) is of the order of days.Alternatively, if the growth of this reaction zone is domi-nated by diffusion in the solid, with a realistic D of theorder of 10^14m2/s, the time scale is instead of the orderof 40 years. It is likely that the true duration is betweenthese two estimates; experiments under similar conditionswould clarify the appropriate diffusivity for this process.Ultimately, the growth time of this reaction zone placesan important constraint on the duration of partial meltingof granitic wall rock by the adjacent dolerite sill.

MODELS OF GRANITE PART IALMELT ING AND MELTSEGREGAT IONThe upper lobe of the Basement Sill, or dolerite feeder, wasemplaced into cool granitic country rock, producing athin, well-formed chilled margin of dolerite along the con-tact. The presence of a strongly chilled margin suggeststhat possible pre-heating of country rock as a result of theprevious emplacement of stratigraphically higher sills atthis location was negligible. Given a geothermal gradientof 258C/km, the ambient country rock temperature was

1200°C 1000°C 800°C

Dolerite

PartiallymeltedOrestesgranite

0.12

Distance fromdolerite chilledmargin (m)

0.06

Sliver ofpartiallymelted

Granitic dike

Doleritechilledmargin

PartiallymeltedOrestesgranitebreccia

Dolerite chilled marginpgtb - Partially melted Orestes

granite breccia (A-395-3)pgt - Partially melted Orestes

granite (A-395-4)pgt - Sliver of partially melted

Orestes granite

Pyroxene analyzedWo

En Fs

Fig. 10. Pyroxene compositions from the dolerite chilled margin and partially melted Orestes granite in the pyroxene quadrilateral.

0 0.5 1 1.5 2 2.5 3 3.534

36

38

40

42

44

46

48

50

52

Distance (mm)

Mg#

= 1

00%

x [M

gO/(

MgO

+FeO

)]

Dolerite reaction zone Dolerite chilledmargin

OpxCpx

Fig. 11. Mg-number of pyroxenes with distance in the dolerite chilledmargin. Distance is measured from the contact with the sliver ofpartially melted Orestes granite.


2139

about 1008C.That the dolerite feeder acted as a conduit forsustained flux of magma and was not a single magmainjection is supported by a calculated contact temperatureof !7158C (at 100MPa). This model temperature was cal-culated by solving the full heat conduction equation forcooling and solidification of an instantaneously injectedsill (e.g. Turcotte & Schubert, 1982, p. 172) assuming atemperature of 12008C for the doleritic magma based ontwo-pyroxene thermometry (Simon & Marsh, 2005). Thistemperature (i.e. 7158C) is approximately at or just abovethe water-saturated granite solidus for pressures between60 and 90MPa (Tuttle & Bowen, 1958) and is much lowerthan the temperature range of 900^9508C necessary toproduce the observed extent of melting. More probably,the dolerite feeder acted as a conduit for a sustained fluxof magma for some period of time, pinning the contacttemperature above the granite minimum, and causing a

partial melting front to propagate outward into the gra-nitic country rock. This is consistent with the preferredmodel of Petcovic & Dufek (2005) wherein they reconciledwall rock melting by invoking advective magma transportwithin a Columbia River flood basalt feeder dike.The operative granite solidus here is expected to be

hotter than the water-saturated granite minimum. A mini-mum estimate of the water content of the unmelted Orestesgranite is 0"33wt%, based on assuming complete biotitebreakdown at an initial modal content of 7"8 vol.%,obtained by detailed modal analysis (i.e. manual tracingof all biotite grains on a digitized image of a single thinsection). For comparison, partial melting experiments at50MPa involving a natural granite with an assemblage ofprimarily quartz, plagioclase, alkali feldspar, and musco-vite with a total water content of 0"58wt% yielded 5vol.% melt at 8508C and with the addition of 1wt% H2Oyielded 40 vol.% melt at 8008C (Attrill & Gibb, 2003).Manual tracing of granophyre regions in thin sectionsbetween restite phases in partially melted Orestes granite

Table 4: Representative biotite analysis in oxide weight percent and cations per 11 oxygens

Sample: 975G 978G

WR no.: A-395-4 A-395-4

Distance (m): 0"123 0"123

SiO2 36"01 36"37

Al2O3 13"15 13"19

CaO 0"12 0"05

MgO 7"86 7"54Na2O 0"15 0"14

K2O 8"04 8"59

FeO 23"18 23"27

MnO 0"25 0"30

TiO2 4"17 4"95

Cr2O3 0"00 0"02

Total 92"92 94"41

Si 2"88 2"87Al 1"24 1"23

Ca 0"01 0"00

Mg 0"92 0"87

Na 0"02 0"02

K 0"81 0"86

Fe 1"53 1"51

Mn 0"02 0"02

Ti 0"24 0"28Cr 0"00 0"00

O 11 11

Total 18"68 18"67

WR no. and Distance rows correspond, respectively, tosample number and distance from dolerite chilled marginin Table 1. All Fe is reported as FeO.

Table 5: Representative Fe^Ti oxide analysis in oxideweight per cent and in cations per three or four oxygens

Sample: 976G 1027G 1153G 1154G

WR no.: A-395-4 TH30 TH4 TH4

Distance (m): 0"123 23 5"2 5"2

Mineral: hm ilm mag mag

SiO2 0"01 0"00 0"05 0"33

Al2O3 0"37 0"03 0"12 0"44

MgO 0"00 0"03 0"00 0"00FeO 87"13 36"71 72"71 91"00

MnO 0"09 5"73 2"37 0"10

TiO2 0"63 46"11 20"30 1"00

Cr2O3 0"03 0"00 0"01 0"00

Total 88"26 88"60 95"56 92"86

Si 0"00 0"00 0"00 0"01

Al 0"01 0"00 0"01 0"02

Mg 0"00 0"00 0"00 0"00Fe2$ 0"01 0"85 1"52 1"04

Fe3$ 1"96 0"03 0"81 1"89

Mn 0"00 0"14 0"08 0"00

Ti 0"01 0"99 0"59 0"03

Cr 0"00 0"00 0"00 0"00

O 3 3 4 4

Total 4"98 4"34 6"22 6"62

All of these Fe–Ti oxide grains are in reacted biotite sites.WR no. and Distance rows correspond, respectively, tosample number and distance from dolerite chilled marginin Table 1. All Fe is reported as FeO.


2140

provides an estimate of the melt fraction at solidificationand, hence, not the degree of melting. The degree of melt-ing was most certainly greater than this estimate for par-tially melted Orestes granite at550 cm from the doleritechilled margin. At 12"3 cm, 1"2m, and 5"2m there is,respectively, 55 vol.%, 31 vol.%, and 23 vol.% granophyre.Furthermore, these melt fractions can be considered to beminimum estimates because some melt may have solidifiedas overgrowths on restite grains. Given the uncertaintyin the biotite composition and mode, and the potentialpresence of at most 0"1^0"3wt% intergranular H2O(Whitney, 1988) and water stored in grain boundaries[!0"1g of H2Om'2 estimated by Holness et al. (2005) in apelite], it is possible that melt fractions of the magnitudeobserved in the experiments of Attrill & Gibb (2003)could be generated at similar temperatures.It is clear from Figs 6 and 7 that the silicic melt forming

the granophyre zone was segregated from partially meltedOrestes granite over the interval 412"3 cm but 550 cmfrom the dolerite chilled margin. The Harker diagrams ofFig. 8 are consistent with this interpretation and are analo-gous with the restite-unmixing model of Chappell et al.(1987).The chemical relations of Fig. 6 also suggest segrega-tion of a monotonically decreasing volume of melt frompartially melted Orestes granite with increasing distancefrom the dolerite chilled margin. A simple mass-balancecalculation based on normative alkali feldspar fromwhole-rock compositions indicates that 0"45, 0"55, and0"65 mass fraction removals of granitic melt of the compo-sition of the granitic dikes from unmelted Orestes granitecan restore the compositions of partially melted Orestesgranite at distances of 12"3, 6"7, and 5"6 cm.These estimatesrepresent minimum mass fractions of melt removedbecause the partially melted Orestes granite still containsinterstitial granophyre between restite phases that isassumed to have been granitic melt.Another estimate of the mass fraction of granitic melt

removed from the partially melted Orestes granite can bemade through application of batch and fractional meltingmodels using trace element partitioning between melt andrestite. End-member fractional melting is not likely basedon the observation of granophyre within partially meltedOrestes granite, representing residual granitic melt leftafter segregation. A fractional melting model is, neverthe-less, useful for providing a lower limit on the degree ofgranitic melt removed provided that end-member batchmelting is also not likely. Based on whole-rock trace ele-ment compositions (Table 1), Rb is the least compatibleelement in this assemblage. Partition coefficients (crystal/melt) for Rb in alkali feldspar, plagioclase, quartz, andbiotite are, respectively, 0"5, 0"04, 0"04, and 4"2 (Ragland,1989). A bulk partition coefficient was calculated by usingthe results of a CIPW norm calculation on a whole-rockunmelted Orestes granite composition normalized after

adding biotite. The two melting models, using three differ-ent initial mass fractions of biotite, are plotted as a func-tion of melt mass fraction using the calculated bulkpartition coefficient for Rb in Fig. 12. Also shown inthis figure are Rb concentrations for samples A-395-4,A-395-1, and A-395-3, which are partially melted Orestesgranite collected at, respectively, 12"3, 6"7, and 5"6 cm fromthe dolerite chilled margin. Batch melting predicts 0"20,0"31, and 0"83 mass fractions of melt removed, whereas afractional melting model predicts 0"18, 0"24, and 0"48 massfractions of melt removed from partially melted Orestesgranite at, respectively,12"3, 6"7, and 5"6 cm from the doler-ite chilled margin. Except for the batch melting model at5"6 cm, both melting models predict a smaller mass frac-tion of melt removed at each location than the simplemass-balance calculation based on normative alkalifeldspar from whole-rock compositions.

MECHANICS OF GRANIT IC MELTSEGREGAT IONInterstitial pore melt will segregate from its matrix inresponse to a gradient in pore pressure. A gradient inpore pressure can be generated by body forces, such asmelt buoyancy relative to the matrix, by changes in hydro-static or deviatoric stress, and positive volume changesassociated with partial melting. In this instance, it is unli-kely, for two reasons, that melt buoyancy is responsible formelt segregation. First, the contact between the doleritechilled margin and granitic country rock dips at !458,with the granitic country rock being the hanging walland the dolerite chilled margin being the foot wall(see Fig. 2). This orientation is opposite to that required

0 0.2 0.4 0.6 0.8 10.4

0.5

0.6

0.7

0.8

0.9

1

Melt fraction

Rb s

/Rb o

BatchFractional

A-395-4 (12.3 cm)

A-395-1 (6.7 cm)

A-395-3 (5.6 cm)

8.8

All distances from chilled margin

7.8

6.38.87.86.3

wt% biotite

Fig. 12. Batch and fractional melting models using Rb concentration.Curves represent different initial weight per cent of biotite in theunmelted Orestes granite. Horizontal lines are the Rb concentrationsin partially melted Orestes granite samples collected perpendicular tothe dolerite chilled margin.


2141

for granitic melt to collect buoyantly at the contact withthe dolerite chilled margin. Second, this gravitationalinstability is unlikely to develop because the time scale formelt to buoyantly rise through partially molten graniticrock is much larger than the time scale for the dolerite sillto cool. This also prevents segregated melt that collects atthe dolerite chilled margin from buoyantly rising as asingle volume along the contact. For example, a Stokes-flow (i.e. creeping flow) scaling for the velocity ofbuoyant melt rise (not porous melt flow) in a host of par-tially molten granitic rock gives a time scale ofts# m/!rgW!1014 s, where m is the effective viscosity ofthe solid matrix, which is !1017 Pa s for Westerly granitecontaining a melt fraction of 10% (Rutter & Neumann,1995), !r is the density contrast between melt and rock(!102 kg/m3), g is gravitational acceleration, and W is thethickness of the melt layer (!1m). In contrast, a timescale for conductive cooling of a solidifying dolerite sill ofthickness L#100m is tc#L2/! !1010 s, where ! is thermaldiffusivity (10'6m2/s). That is, the ratio of the characteris-tic time for significant buoyant flow relative to that forcooling is ts/tc !104.Melt segregation caused by externally imposed changes

in hydrostatic stress is also not likely owing to lack of aplausible mechanism for causing large changes in hydro-static stress, such as introduction or release of overburden.Even if these mechanisms did occur, the length scale overwhich melt segregation has operated here is much less thanthe tectonic length scales over which changes in hydro-static stress would exist. That is, the expected gradients inhydrostatic stress are much too broad to work effectivelyover this highly restricted area.Partial melting commonly involves a local positive

volume change and, therefore, creates an elevated porepressure directed over a selected spatial region. Evidencefor reaction-driven deformation has been observed in thecontact aureoles of plutons, for example, as mostly intra-crystalline micro-cracks thought to contain melt (Holness& Watt, 2002). No evidence of cracking has been observedin the partially melted granitic (Orestes and Vanda) rocksdescribed here, possibly because of the lack of reactionswith large positive volume changes, such as muscovitebreakdown (Rushmer, 2001). Even with local pore meltpressure gradients at the sites of reaction there is no gen-eral preference to drive melt flow in the direction of thedolerite chilled margin, a condition necessary for melt seg-regation to occur. This conclusion is supported by theabsence of evidence for melt segregation in the contactaureole studied by Holness & Watt (2002). Nevertheless,elevated pore melt pressure caused by partial melting ofthe granitic rocks is probably, at least in part, responsiblefor generating the overpressure necessary to fracture thedolerite chilled margin and create space for the formationof the granitic dikes.

Changes in deviatoric stress related to both melting inthe wall rock and thermal contraction in response to doler-ite solidification, following flow cessation, are perhaps mostlikely to have generated the pore melt pressure gradientnecessary to drive melt segregation and collection into amelt-rich reservoir. Let us consider the schematic two-dimensional model shown in Fig. 13. Here it is assumedthat melt segregation takes place after partial melting andnot during melting. In the first frame, the pore melt pres-sure is at the level of the hydrostatic stress (s1#s2#s3)and the pore melt is static. A decrease in s2 as a result of,for example, a tensional strain, at the right interface causesa planar crack to be initiated in the partially melted gran-ite immediately adjacent and parallel to the dolerite chilledmargin.The tear occurs exactly where the partially meltedgranite has the least strength (i.e. highest melt fraction).Tiny slices of partially melted granite on the doleritechilled margin side of the crack remain welded to thedolerite chilled contact. The melt pore pressure in thecrack is less than in the partially melted granite. This pres-sure gradient drives melt into the crack by porous meltflow with commensurate pore space reduction, or lateralcompaction, of the partially melted granite on both sidesof the crack. With continued strain, the melt-filled crackwidens to form a staging melt reservoir for granitic dikeemplacement.If matrix deformation is assumed to occur by viscous

deformation of the crystals, then a length scale overwhich compaction occurs adjacent to an impermeableboundary is given by the compaction length, d ! [(z$4/3Z)k/m]1/2!101/2 (McKenzie, 1984). Here, z and Z are theeffective bulk and shear viscosities of the matrix [!1015 Pa sfollowing Rabinowicz & Vigneresse (2004)]. Dynamicviscosity of the granitic melt, m, is 106 Pa s and permeabil-ity, k, is 10'8m2 using the Rumpf^Gupte relation fromHersum et al. (2005) with a mean crystal size of 0"1mmand a melt fraction of 0"5. With these properties, thecompaction length scale is !3m. This compactionlength represents an upper bound because the modelassumes a partially melted host that contains aconstant melt fraction when in fact melt may be progres-sively lost and never achieve this state. However, for thissituation, permeability will decrease sharply with distanceaway from the dolerite chilled margin as the melt fractioncorrespondingly decreases. The compaction length isapproximately one order of magnitude larger than the dis-tance for which partially melted Orestes granite has acomposition that differs from the unmelted Orestesgranite, which suggests that melt can be removed bythis process.The magnitude of the deviatoric stress necessary to initi-

ate and then fill a crack by segregating melt is probablydriven by pressure reduction within the doleritic magmaupon cessation of flow in concert with contraction owing


2142

to solidification of the magma feeder itself. Each processhas a distinct characteristic strain rate that would dictateboth the mechanism of matrix deformation and alsowhether the rate of filling the granitic melt reservoir iscontrolled by the rate of porous melt flow or matrix defor-mation. A complete continuum-scale model of simulta-neous heat and mass transport is under development totest these scenarios.

EMPLACEMENT OF GRANIT ICDIKESThere are few documented natural examples of partialmelt from country rock intruding the adjacent magmaticrock that initially generated the melting. Kaczor (1988)reported that partially melted granite, up to 56% melt,around a trachyandesite plug in the Sierra Nevada,California intruded back into the plug after the plughad cooled enough to develop jointing. There is no evi-dence, however, that melt segregated from the restiteminerals or that any mixing between the granitic meltand the trachyandesite occurred (Kaczor, 1988). A fieldexample similar to the present study area was presentedby Philpotts & Asher (1993), who described multipleevents of partial melt, derived from gneiss and pegmatitecountry rock screens, intruding the Higganum diabase

dike in Connecticut. The first set of melts entering thediabase dike are preserved as felsic wisps (515 mm wide)that penetrate at most about 0"5 cm into partially solidi-fied diabase chilled margin. These wisps were subse-quently stretched and thinned by further flow of thediabasic magma. Upon complete solidification of the dia-base, a network of interconnected granophyric veins,each up to 4 cm wide, entered the diabase. These veinsare oriented perpendicular to the screens, traversing upto several meters between intervening screens. Near thechilled margins of the diabase, the veins abruptly thinand merge with networks of thinner (5100 mm) veinletsthat connect with grain boundary granophyre in the par-tially melted country rock screens (Philpotts & Asher,1993). Although this last example more closely resemblesthe Ferrar granitic dikes, the volume of intruding melt,the volume of partially melted wall rock, and the extentof intrusion are relatively minor relative to the Ferrargranitic dikes.The granophyric magma zone sandwiched between

partially melted Orestes granite and the dolerite chilledmargin, which is connected directly to granitic dike rockat location ‘A’ (Fig. 1), unequivocally identifies this zoneas the staging reservoir for the melt forming the plexusof dikes within the dolerite. Based on the volume of thegranitic dikes, it is apparent that the granophyric

#1 #1#1

#2 #2 #2

111

(a) (b) (c)

Mel

t fra

ctio

n

Mel

t fra

ctio

n

Mel

t fra

ctio

n

Granitic melt Granitic melt Granitic melt

Gra

nitic

mel

t res

ervo

ir

Dol

erite

chi

lled

mar

gin

Dol

erite

chi

lled

mar

gin

Dol

erite

chi

lled

mar

gin

Crack

Gra

nitic

mel

t res

ervo

ir

Porous Porous

Time

Partially PartiallyPartially

Fig. 13. Schematic diagram of silicic melt segregation from partially melted granite into a melt-rich reservoir. This is a map view that is per-pendicular to the contact between the partially melted granite and the dolerite chilled margin. The sequence of events is: (a) a crack initiatesparallel to the contact in the partially melted granite as a result of deviatoric stress; (b) interstitial silicic melt begins to fill the crack by porousmelt flow as a result of a lateral pressure gradient; (c) the melt reservoir thickens as more silicic melt enters accompanied by simultaneouscompaction of the partially melted granite.


2143

magma zone was probably significantly wider prior tothe evacuation of melt that formed the granitic dikes.Moreover, the wide distribution of these large dikes inthe Bull Pass area suggests that the dikes originatedfrom a number of locations by the same process.Another notable aspect of this occurrence is thehigh rigidity of the dolerite chilled margin, which actedas a mechanical barrier to prevent mixing of graniticmelt with doleritic melt, allowing formation of areservoir of granitic melt from which to generate thegranitic dikes.Granitic dike emplacement clearly occurred late in the

cycle of dolerite emplacement and solidification. To makefracture-like boundaries, the dolerite had to have acquiredenough strength to allow fracturing rather than viscousflow. In partially molten rock, fracturing will occur onlyat relatively low degrees of melt fraction (!550%; e.g.Marsh, 2002). Neither the dolerite nor the granitic dikeshow chilling against one another, which suggests that thewhole system was still hot, perhaps above the solidus of thedike magma. This condition allowed the granitic magmato move without confronting the challenge of transportingsluggish viscous silicic melt through a narrow dike in coolcountry rock (Rubin, 1995). Also, unlike at the doleritechilled margin adjacent to the granophyre melt at location‘A’, there is no corresponding reaction zone in the doleriteat the contact between the dolerite and the granitic dike.The lack of a corresponding reaction zone is found only atthe contact between the dolerite and the granitic dikewithin !1m of the dolerite chilled margin and it isunknown if a reaction zone exists deeper into the interiorof the sill as outcrops of the contact are altered.Nevertheless, in the proximity of the dolerite chilledmargin, the lack of a reaction zone suggests that the empla-cement of the granitic dikes occurred after granophyricmelt began collecting at the dolerite chilled margin (andat a lower temperature).Another interesting feature of the field area is the

complete absence of granitic dikes directed away fromthe dolerite feeder into the partially melted granitic(Orestes and Vanda) rocks. This suggests that the doler-ite feeder was nearly solidified at the time of graniticdike formation and, therefore, still in the process of cool-ing. This may provide an explanation for the drivingforce and thus the direction of diking. The doleritechilled margin formed a rigid wall separating a contrac-tion on the dolerite side from an overpressure on thewall rock side. The contraction was due to !10%volume reduction associated with dolerite solidification(based on the density contrast between melt and solidrock) and the overpressure owing to excess pore meltpressure associated with melting. This approximatelydipole pressure field apparently directed all dike propa-gation into the dolerite.

IMPL ICAT ION FORCONTAMINAT ION OF BASALT ICMAGMA IN THE LOWER CRUSTThe motivation of this research is, in part, to use the obser-vations for melt generation and segregation from a shallowcrustal example to infer how similar processes operate inthe lower crust. However, there are several caveats includ-ing, for example, that the length scale of melt segregation(centimeter to meter in this study as opposed to kilometerto tens of kilometers in the deep crust) and the rheologicalbehavior (i.e. brittle vs ductile) are likely to be very differ-ent. Also, major element melt compositions in the deepcrust are less likely to be out of chemical equilibrium thanin more shallow and rapid melting environments such ascontact aureoles. Another limiting factor is the geometryof the contact. In the present study, the contact betweendolerite and granite was !458 from horizontal. Perhapsthe key to large-scale crustal anatexis (i.e. felsic meltproduction) is underplating of dolerite such that heat isadvected in a purely vertical direction from thedolerite into the overlying granite. Such a model hasbeen advocated in other field and numerical studies(see Bergantz, 1989).The field relations, essentially crustal anatexis arrested

during development, allow us to speculate on the causesof contamination of magmas in traversing continentalcrust. Here, contamination is defined as modification ofincompatible and not major element composition of theintruding magma by the host country rock. This hasbeen shown countless times in isotopic studies (e.g.Carter et al., 1979; Taylor, 1980; Stewart & DePaolo,1990; Zeng et al., 2005), but the actual physical processesby which this happens remain unconstrained. Althoughstoping is an obvious viable mechanism (it is seen to avery limited extent in these sills), contamination is evi-dently caused by much more than settling and partialdigestion of stoped blocks. Many lavas and intrusiverocks showing contamination show no evidence whatso-ever of partially digested or fragmented xenoliths, sug-gesting that there must be other, much more subtle anddifficult to recognize, common contamination processes.One such process is suggested by an extension of the pro-cess described here. This involves the reactivation of amagma feeder containing the granitic dikes generated asa consequence of a previous episode of magma flow.Plutonic magmatic systems must have a vertically exten-sive underlying plumbing style similar to that observedfor typical volcanic systems. That is, a major eruptiveepisode is generally marked by a series of eruptionevents separated by repose times. At depth this pulsativesequence manifests itself in magmatic feeders as times oflocal inflation and magma flow interspersed with periodsof deflation and partial solidification.


2144

In Figure 14a^d, we schematically show the sequenceof events involving a basaltic magma containing graniticdikes of the general nature found in the Ferrar system,which leads to crustal contamination of the basalticfeeder. The active feeder at near-solidus temperatures andcontaining silicic dikelets, generated as described herein, isreactivated by the introduction and flow of near-liquidustemperature magma from deeper in the system. The ensu-ing flow causes a melting front to propagate progressivelyoutwards towards the margins. The original basaltic rockof the feeder and the silicic dikelets are each reheated andremobilized and, with subsequent magma flow and meltback, are systematically caught up in the flow. The melteddikelets are stretched, as a result of the developing para-bolic magma velocity profile, and progressively thinneduntil chemical diffusion becomes effective in dispersingthe silicic material within the basaltic magma. That is,the characteristic time for chemical diffusion (tD) dependson the square of the dikelet half-thickness (L2) divided bythe governing chemical diffusivity (D), tD!L2/D. BecauseD is typically very small (e.g. !10'10m2/s), diffusion times

for any appreciable value of L are commonly much longerthan local solidification times. However, if the distance ofmagma flow is large, thinning becomes extreme, L isgreatly reduced, and chemical diffusion becomes highlyeffective. All physical traces of the dikelets are removed,leaving a systematic and diagnostic distribution of incom-patible element and unusual isotopic abundances. Forexample, in this process more silicic dikelets are likely toexist and be assimilated near the margins than in thecenter of the basaltic feeder, and this would produce anelevated, concave profile of 87Sr/86Sr within the feeder(see Fig. 14e). This signal would travel with the feedermagma and upon emplacement as a lava or sill wouldshow up as a diagnostic feature. Spatial signals of thisnature can be revealed through systematic analysis ofsample profiles through dikes, sills, and plutons. A concavevariation of this very nature has been found in 87Sr/86Srby Hergt et al. (1989) in a profile through the 150m thickFerrar dolerite sill at Portal Peak in the Transan-tarctic Mountains. The variation in 87Sr/86Sr is large(0"7090^0"7110; see Fig. 14f) and the process outlined above

Reactivated feeder

Silicic dikelet

Solidifiedfeeder

Country rock

87Sr/86Sr

87Sr/86Sr.70

8.71

1.70

9 .71

30

50

70

90

110

130

150

Hei

ght (

m)

(Hergt et al., 1989)

Contaminated feederBackground dolerite

Large system in deep crust

Stretching

Partialmeltingfront

Velocityprofile

Chemicaldiffusion

(a) (b) (c)

(d) (e) (f) Portal Peak, Antarctica

Fig. 14. Schematic diagram of crustal contamination of a basaltic feeder. (a) A solidified feeder hosting solidified silicic dikelets is reactivatedwith a pulse of basaltic magma. (b) With continued flow, a partial melting front propagates into the solidified feeder and the melted dikelets arecaught up in the flow. (c) The dikelets are stretched by the flow until they become sufficiently narrow that chemical diffusion homogenizes theminto the basaltic magma. (d) Considering this scenario in the deep crust with many silicic dikelets (exaggerated in the figure) over a largedistance then (e) one possible outcome is elevated isotope ratios farther along in the system with greater crustal contamination at the marginsthan in the center. (f) One example of elevated isotope ratios in a vertical profile across a dolerite sill from Antarctica.


2145

provides a realistic explanation of the variation in stron-tium isotopes. Another factor is that initial disequilibriummelts may be richer in 87Sr/86Sr than the bulk-rock com-position, as demonstrated in experiments by Knesel &Davidson (1999).With increasing depth in the crust and increasingly

hotter wall rock, this contamination process becomesincreasingly effective in all its critical aspects. Melting ismore extensive, solidification is slower, melt segregation ismore effective, the dipolar pressure field is larger, moreand larger granitic dikes are likely to form, and reactiva-tion of the feeder zone is more likely given a widerwindow of cooling time. Conversely, diking is obviouslylimited to crustal depths at which strain associated withthermal contraction is accommodated by fracture and notductile flow. Also, the process described above should notbe confused or applied to understanding the larger-scalegeneration of hybrid granitic magmas through intrusionof basaltic magmas into the deep crust.

CONCLUSIONSA natural example of silicic melt generation, segregation,and emplacement of granitic dikes into Ferrar dolerites ofthe McMurdo DryValleys, Antarctica provides an oppor-tunity to examine fully these sequential processes on a con-venient spatial scale.The granitic dikes are numerous, long(hundreds of meters), fairly thick (!10^30 cm), intercon-nected, and fine-grained. The root source of one dike iscompletely exposed at the upper contact of the BasementSill and granitic country rock. The dike emanates from athin (5 cm) melt sheet separating chilled dolerite from par-tially melted granite. Residual interstitial granophyric meltdecreases away from the contact from 55% to zero withina distance of520m. Higher than expected dolerite contacttemperatures of 900^9508C calculated using two-pyroxenethermometry suggest that the dolerite feeder acted as anopen conduit for a sustained flux of magma. As a conse-quence of this flow, the contact temperature was pinnedabove the ‘dry’ granite minimum, the most restrictive con-dition necessary to generate granitic melt. Closed-systempartial melting of granite occurred beyond 50 cm fromthe dolerite chilled margin. However, compositionalmoment balances on a anorthite^albite^orthoclase ternarybetween the alkali feldspar-enriched melt sheet and grani-tic dike whole-rock compositions are reconciled by meltssegregated from increasingly alkali feldspar-depleted par-tially melted granite at 12"3 cm and closer to the doleritechilled margin. Melting models and mass-balance calcula-tions predict a range of between 48 and 83% maximumvolumes of segregated granitic melt, but these are only esti-mates as the samples are not exclusively residuum. If gra-nitic melt segregation occurs by viscous compaction of therestitic crystal matrix, then, employing commonly usedvalues of the critical parameters, the compaction length

scale is !3m. This is an upper bound, as the compactionmodel assumes a constant melt fraction. Nevertheless, theresult is only an order of magnitude larger than the dis-tance over which the partially melted granite has a compo-sition that differs from unmelted granite. Contractionattending cessation of doleritic magma flow in addition tosolidification-induced contraction probably generateddeviatoric stresses within the partially melted zone thatinitiated crack formation at and parallel to the contact,allowing interstitial melts to flow outward into the doleritein response to a pore pressure gradient. Excess pore pres-sure within this granitic melt reservoir along the contactsubsequently tore open the brittle dolerite chilled margin,in the fashion of a trapdoor, and emplaced, essentially byevacuation, granitic dikes into the solidifying dolerite.Granite partial melting, segregation, and dike emplace-ment probably occurred within a maximum period of sev-eral tens of years, as suggested by the time estimated toproduce, by interdiffusion between the granitic melt anddolerite, a thin (2"5mm) distinctive planar dolerite reac-tion zone within the dolerite chilled margin. Reactivationof similarly injected basaltic feeders deeper in the crust,with dikelet stretching and absorption by simultaneous dif-fusion, presents a possible means of extensive and subtlecrustal contamination of basaltic magma.

ACKNOWLEDGEMENTSWe thank MikeWeiss, Amanda Charrier, and participantsof the 2005 NSF-funded ‘Magmatic Field LaboratoryWorkshop’ for field assistance. Special thanks go to thePHI pilots and the McMurdo support staff for their dedi-cated, high-quality help.We thank Ken Livi for assistancewith microprobe analysis, andJohn Ferry for comments onan initial draft of the manuscript. George Bergantz,Alberto Patin‹ o Douce, and an anonymous reviewer arethanked for constructive reviews. Editorial assistance anda review from MarjorieWilson are gratefully appreciated.This work is supported by NSF Grants OPP 0229306 andOPP 0440718 to Johns Hopkins University.

REFERENCESAllibone, A. H., Cox, S. C., Graham, I. J., Smillie, R. W.,

Johnstone, R. D., Ellery, S. G. & Palmer, K. (1993a). Granitoids ofthe Dry Valleys area, southern Victoria Land: plutons, field rela-tionships, and isotopic dating. New Zealand Journal of GeophysicalResearch 36, 281^297.

Allibone, A. H., Cox, S. C. & Smillie, R.W. (1993b). Granitoids of theDryValleys area, southernVictoria Land: geochemistry and evolu-tion along the early Paleozoic Antarctic Craton margin. NewZealandJournal of Geophysical Research 36, 299^316.

Al-Rawi, Y. & Carmichael, I. S. E. (1967). Natural fusion of granite.American Mineralogist 52, 1806^1814.

Andersen, D. J., Lindsley, D. H. & Davidson, P. M. (1993). QUILF:a PASCAL program to assess equilibria among Fe^Mg^Ti oxides,


2146

pyroxenes, olivine, and quartz. Computers and Geosciences 19,1333^1350.

Armstrong, J. T. (1988). Quantitative analysis of silicate and oxideminerals: Comparison of Monte Carlo, ZAFand phi^rho^z proce-dures. In: Newbury, D. E. (ed.) Microbeam Analysisç1988. SanFrancisco, CA: San Francisco Press, pp. 239^246.

Attrill, P. G. & Gibb, F. G. F. (2003). Partial melting and recrystalliza-tion of granite and their application to deep disposal of radioactivewaste. Lithos 67, 103^117.

Bacon, C. R. (1992). Partially melted granodiorite and related rocksejected from Crater Lake caldera, Oregon.Transactions of the RoyalSociety of Edinburgh 83, 27^47.

Behrendt, J. C., LeMasurier, W. E., Cooper, A. K., Tessensohn, F.,Trehu, A. & Damaske, D. (1991). Geophysical studies of the WestAntarctic Rift System.Tectonics 10, 1257^1273.

Bergantz, G. W. (1989). Underplating and partial meltingçimplica-tions for melt generation and extraction. Science 245, 1093^1095.

Borg, S. G., DePaolo, D. J. & Smith, B. M. (1990). Isotopic structureand tectonics of the central Transantarctic Mountains. Journal ofGeophysical Research 95, 6647^6667.

Brearley, A. J. (1987). A natural example of the disequilibriumbreakdown of biotite at high temperature: TEM observations andcomparison with experimental kinetic data. Mineralogical Magazine51, 93^106.

Brown, M. (2006). Melt extraction from lower continental crust oforogens: the field evidence. In: Brown, M. & Rushmer, T. (eds)Evolution and Differentiation of the Continental Crust. New York:Cambridge University Press, pp. 331^383.

Brown, M. & Rushmer, T. (2006). Introduction. In: Brown, M. &Rushmer, T. (eds) Evolution and Differentiation of the Continental Crust.NewYork: Cambridge University Press, pp. 1^20.

Carter, S. R., Evensen, N. M., Hamilton, P. J. & O’Nions, R. K.(1979). Basaltic magma sources during the opening of the NorthAtlantic. Nature 281, 28^30.

Chappell, B.W.,White, A. J. R. & Wyborn, D. (1987).The importanceof residual source material (restite) in granitic petrogenesis. Journalof Petrology 28, 1111^1138.

Elliot, D. H. & Fleming, T. H. (2000). Weddell triple junction:The principal focus of Ferrar and Karoo magmatism during initialbreakup of Gondwana. Geology 28(6), 539^542.

Elliot, D. H. & Fleming, T. H. (2004). Occurrence and dispersalof magmas in the Jurassic Ferrar Large Igneous Province,Antarctica. Gondwana Research 7, 223^237.

Fitzgerald, P. G. & Stump, E. (1997). Cretaceous and Cenozoic episo-dic denudation of the Transantarctic Mountains, Antarctica;new constraints from apatite fission track thermochronology inthe Scott Glacier region. Journal of Geophysical Research 102,7747^7765.

Graphchikov, A. A., Konilov, A. N. & Clemens, J. D. (1999). Biotitedehydration, partial melting, and fluid composition: Experimentsin the system KAlO2^FeO^MgO^SiO2^H2O^CO2. AmericanMineralogist 84, 15^26.

Green, N. L. (1994). Mechanism for middle to upper crustal contami-nation: Evidence from continental-margin magmas. Geology 22,231^234.

Gunn, B. M. & Warren, G. (1962). Geology of Victoria Land between theMawson and Mulock Glaciers, Antarctica. New Zealand Geological SurveyBulletin71, , 157 pp.

Hamilton, W. (1965). Diabase Sheets of the Taylor Glacier Region VictoriaLand, Antarctica. US Geological Survey, Professional Papers 456-B.

Hergt, J. M., Chappell, B.W., Faure, G. & Mensing,T. M. (1989). Thegeochemistry of Jurassic dolerites from Portal Peak, Antarctica.Contributions to Mineralogy and Petrology 102, 298^305.

Hersum, T. G., Hilpert, M. & Marsh, B. D. (2005). Permeability andmelt flow in simulated and natural partially molten basalticmagmas. Earth and Planetary Science Letters 237, 798^814.

Holness, M. B. & Watt, G. R. (2002). The aureole of the Traigh Bhanna Sgurra Sill, Isle of Mull: reaction-driven microcracking duringpyrometamorphism. Journal of Petrology 43, 511^534.

Holness, M. B., Dane, K., Sides, R., Richardson, C. & Caddick, M.(2005). Melting and melt segregation in the aureole if theGlenmore Plug, Ardnamurchan. Journal of Metamorphic Petrology23, 29^43.

Johnson, M. C. & Rutherford, M. J. (1989). Experimentally deter-mined conditions in the Fish CanyonTuff, Colorado, magma cham-ber. Journal of Petrology 30, 711^737.

Kaczor, S. M. (1988). Disequilibrium melting of granite at the contactwith a basic plug: a geochemical and petrographic study. Journal ofGeology 96, 61^78.

Kitchen, D. E. (1989). The disequilibrium partial melting and assimi-lation of Caledonian granite byTertiary basalt at Barnesmore, Co.Donegal. Geological Magazine 126, 397^405.

Knesel, K. M. & Davidson, J. P. (1999). Sr isotope systematics duringmelt segregation by intrusion of basalt into continental crust.Contributions to Mineralogy and Petrology 136, 285^295.

Kovach, L. A. (1984). Dynamics of magma ascent through the SierraNevada, California. PhD dissertation, Johns Hopkins University,Baltimore, MD, 287 pp.

Lindsley, D. H. (1983). Pyroxene thermometry. American Mineralogist68, 477^493.

Marsh, B. D. (2002). On bimodal differentiation by solidification frontinstability in basaltic magmas, I: Basic mechanics. Geochimica etCosmochimica Acta 66, 2211^2229.

Marsh, B. D. (2004). A magmatic mush column Rosetta Stone:The McMurdo Dry Valleys of Antarctica. EOS Transactions,American Geophysical Union 47, 497^502.

McKelvey, B. C. & Webb, P. N. (1962). Geological investigations insouthern Victoria Land, Antarctica. New Zealand Journal ofGeophysical Research 5, 143^162.

McKenzie, D. (1984). The generation and compaction of partiallymolten rocks. Journal of Petrology 25, 713^765.

Naney, M. T. (1983). Phase equilibria of rock-forming ferromagnesiansilicates in granitic systems. American Journal of Science 283,993^1033.

Patin‹ o Douce, A. E. (1999). What do experiments tell us about therelative contributions of crust and mantle to the origin of graniticmagmas? In: Castro, A., Fernandez, C. & Vigneresse, J. L. (eds)Understanding Granites. Integrating New and Classical Techniques.Geological Society, London, Special Publications 158, 55^75.

Patin‹ o Douce, A. E. & Beard, J. S. (1995). Dehydration-melting of bio-tite gneiss and quartz amphibolite from 3 to 15 kbar. Journal ofPetrology 36, 707^738.

Petcovic, H. L. & Dufek, J. D. (2005). Modeling magma flow and cool-ing in dikes: Implications for emplacement of Columbia Riverflood basalts. Journal of Geophysical Research 110, B10201, doi:10.1029/2004JB003432.

Petcovic, H. L. & Grunder, A. L. (2003). Textural and thermal historyof partial melting in tonalitic wallrock at the margin of a basaltdike, Wallowa Mountains, Oregon. Journal of Petrology 44,2287^2312.

Philpotts, A. R. & Asher, P. M. (1993).Wallrock melting and reactioneffects along the Higganum diabase dike in Connecticut: contami-nation of a continental flood basalt feeder. Journal of Petrology 34,1029^1058.

Rabinowicz, M. & Vigneresse, J. L. (2004). Melt segregation undercompaction and shear channeling: Application to granitic melt


2147

segregation in a continental crust. Journal of Geophysical Research 109,B04407, doi:10.1029/2002JB002372.

Ragland, P. C. (1989). Basic Analytical Petrology. New York: OxfordUniversity Press.

Riley, T. R., Curtis, M. L., Leat, P. T., Watkeys, M. K., Duncan, R.A., Millar, I. L. & Owens, W. H. (2006). Overlap of Karoo andFerrar magma types in KwaZulu Natal, South Africa. Journal ofPetrology 45, 541^566.

Rubin, A. M. (1995). Getting granite dikes out of the source region.Journal of Geophysical Research 100, 5911^5390.

Rushmer, T. (2001). Volume change during partial melting reactions:implications for melt extraction, melt geochemistry and crustalrheology.Tectonophysics 342, 389^405.

Rutter, E. H. & Neumann, D. H. K. (1995). Experimental deformationof partially moltenWesterly granite under fluid-absent conditions,with implications for the extraction of granitic magmas. Journal ofGeophysical Research 100, 15697^15715.

Simkin, T. (1967). Flow differentiation in the picritic sills of NorthSkye. In: Wyllie, P. J. (ed.) Ultramafic and Related Rocks. NewYork:JohnWiley, pp. 64^69.

Simon, A. C. & Marsh, B. (2005). Pyroxene thermometry in the OPXTongue of the Basement Sill, McMurdo Dry Valleys, Antarctica.EOS Transactions, American Geophysical Union 86(52Fall MeetingSupplement, Abstract V13H-08.

Stewart, B. W. & DePaolo, D. J. (1990). Isotopic studies of processesin mafic magma chambers; The Skaergaard Intrusion, EastGreenland. Contributions to Mineralogy and Petrology 104, 125^141.

Taylor, H. P. (1980). The effects of assimilation of country rocks bymagmas on 18O/16O and 87Sr/86Sr systematics in igneous rocks.Earth and Planetary Science Letters 47, 2483^2499.

Turcotte, D. L. & Schubert, G. (1982). Geodynamics Applications ofContinuum Physics to Geological Problems. NewYork: JohnWiley.

Turnbull, I. M., Allibone, A. H., Forsyth, P. J. & Heron, D.W. (1994).Geology of the Bull Pass^St. Johns Range Area, SouthernVictoriaLand, Antarctica. Institute of Geological & Nuclear SciencesGeological Map 14. Lower Hutt, New Zealand: Institute ofGeological & Nuclear Sciences Limited.

Tuttle, O. F. & Bowen, N. L. (1958). Origin of granite in the light ofexperimental studies in the system NaAlSi3O8^KAlSi3O8^SiO2^H2O. Geological Society of America Bulletin 104, 1^153.

von Platen, H. (1965). Experimental anatexis and genesis of migma-tites. In: Pitcher, W. S. & Flinn, G. W. (eds) Controls ofMetamorphism; a Symposium. Edinburgh: Oliver & Boyd, pp. 203^216.

Wen, S. & Nekvasil, H. (1994). SOLVCALC: an interactivegraphics program for calculating the ternary feldspar solvus andfor two-feldspar geothermometry. Computers and Geosciences 20,1025^1040.

Whitney, J. A. (1988). The origin of granite: The role and source ofwater in the evolution of granitic magmas. Geological Society ofAmerica Bulletin 100, 1886^1897.

Zeng, L., Asimow, P. D. & Saleeby, J. B. (2005). Coupling anatecticreactions and dissolution of accessory phases and the Sr and Ndisotope systematics of anatectic melts from a metasedimentarysource. Geochimica et Cosmochimica Acta 69, 3671^3682.


2148

uc berkeley seismology lab - c ontact partial m elting of g ran...

Documents