Crystallization from a Zoned Magma Sheet - Oxford Academic

Cyclicity in the Main and Upper Zones of theBushveld Complex South AfricaCrystallization from a Zoned Magma Sheet

CHRISTIAN TEGNER1 R GRANT CAWTHORN2 AND F JOHANKRUGER3

1DEPARTMENT OF EARTH SCIENCES AARHUS UNIVERSITY C F MOslashLLERS ALLE 110 DK-8000 AARHUS C DENMARK

2SCHOOL OF GEOSCIENCES UNIVERSITY OF THE WITWATERSRAND PO WITS 2050 SOUTH AFRICA

3MORUO MINERALOGICAL SERVICES PO BOX 1368 KRUGERSDORP 1740 SOUTH AFRICA

RECEIVED JANUARY 17 2006 ACCEPTED JULY 28 2006ADVANCE ACCESS PUBLICATION SEPTEMBER 6 2006

The major element composition of plagioclase pyroxene olivine

and magnetite and whole-rock 87Sr86Sr data are presented for the

uppermost 21 km of the layered mafic rocks (upper Main Zone and

Upper Zone) at Bierkraal in the western Bushveld Complex Initial87Sr86Sr ratios are near-constant (07073 plusmn 00001) for 24

samples and imply crystallization from a homogeneous magma sheet

without major magma recharge or assimilation The 2125 m thick

section investigated in drill core comprises 26 magnetitite and six

nelsonite (magnetitendashilmenitendashapatite) layers and changes up-section

from gabbronorite (An72 plagioclase Mg 74 clinopyroxene)

to magnetitendashilmenitendashapatitendashfayalite ferrodiorite (An43 Mg 5

clinopyroxene Fo1 olivine) The overall fractionation trend is

however interrupted by reversals characterized by higher An of

plagioclase higher Mg of pyroxene and olivine and higher V2O5

of magnetite In the upper half of the succession there is also the

intermittent presence of cumulus olivine and apatite These reversals

in normal fractionation trends define the bases of at least nine

major cycles We have calculated a plausible composition for the

magma from which this entire succession formed Forward

fractional crystallization modeling of this composition predicts an

initial increase in total iron near-constant SiO2 and an increasing

density of the residual magma before magnetite crystallizes After

magnetite begins to crystallize the residual magma shows a near-

constant total iron an increase in SiO2 and decrease in density We

explain the observed cyclicity by bottom crystallization Initially

magma stratification developed during crystallization of the basal

gabbronorites Once magnetite began to crystallize periodic density

inversion led to mixing with the overlying magma layer producing

mineralogical breaks between fractionation cycles The magnetitite

and nelsonite layers mainly occur within fractionation cycles not at

their bases In at least two cases crystallization of thick magnetitite

layers may have lowered the density of the basal layer of melt

dramatically and triggered the proposed density inversion resulting

in close but not perfect coincidence of mineralogical breaks and

packages of magnetitite layers

KEY WORDS layered intrusion mineral chemistry isotopes magma

convection differentiation

INTRODUCTION

The 65 km thick sequence of ultramafic and maficrocks of the Bushveld Complex is the largest layeredmafic intrusion known on Earth and it crops out in threemajor areas (limbs) in northern South Africa (Fig 1)Interpretation of gravity data suggests that the easternand western limbs are connected over at least 65 000 km2

(Cawthorn amp Webb 2001) thus the intrusion can beconsidered as a large igneous province in its own right(Coffin amp Eldholm 1994) The intrusion was emplacedinto the upper crust at about 206Ga (Buick et al 2001)in several major magma recharge events (Cawthorn ampWalraven 1998) and it hosts some of the worldrsquos largestand richest orthomagmatic metal deposits (Lee 1996Cawthorn et al 2005) The final major magma rechargeevent took place in the Main Zone (Fig 2) about 42 kmup in the stratigraphy and resulted in lateral expansion

JOURNAL OF PETROLOGY VOLUME 47 NUMBER 11 PAGES 2257ndash2279 2006 doi101093petrologyegl043

Corresponding author E-mail christiantegnergeoaudk

The Author 2006 Published by Oxford University Press All

rights reserved For Permissions please e-mail journalspermissions

oxfordjournalsorg

Dow

nloaded from httpsacadem

icoupcompetrologyarticle471122571444530 by guest on 15 July 2022

of the sheet-like magma chamber (Kruger 2005)Evidence for mixing at this level between residual andrecharged magma comes from protracted reversals inMg and An contents of pyroxene and plagioclase andchanges in initial 87Sr86Sr value Within this intervala distinct thin layer of orthopyroxenite occurs knownas the Pyroxenite Marker which is present in both theeastern (von Gruenewaldt 1970) and western limbs(Cawthorn et al 1991) The initial 87Sr86Sr composi-tion of the 21 km thick cumulate sequence above thePyroxenite Marker which comprises the upper MainZone (MZU) and the Upper Zone (UZ) (Fig 2) isconstant with an average of 07073 plusmn 00001 (2 standarderror SE) and significantly different from the under-lying 42 km of cumulates (Kruger et al 1987 Kruger1994) This has been explained by complete homo-genization between residual and added magma abovethe Pyroxenite Marker (Kruger et al 1987 Cawthornet al 1991) With an estimated volume of 140 000 km3the MZU and UZ represent the largest known sheet ofbasaltic magma emplaced into the Earthrsquos crustThis contribution aims to decipher the physical

processes of crystallization within the huge MZU andUZ magma sheet The UZ includes about 30 distinctmagnetitite and nelsonite (magnetitendashilmenitendashapatite)layers which vary from 2 to 710 cm thick and hostsworld-class deposits of V Ti and P (Cawthorn ampMolyneux 1986 von Gruenewaldt 1993 Lee 1996Cawthorn et al 2005) From bottom to top MZU andUZ evolve from gabbronorite (Mg of clinopyroxeneis 74) to iron-rich apatitendashmagnetitendashfayalite ferrodiorite(Mg of clinopyroxene is lt5) this sequence has tradi-tionally been interpreted as the result of closed-systemcrystallization without magma recharge (Wager amp Brown1968 Willemse 1969a 1969b von Gruenewaldt1973 Molyneux 1974) Several excursions from simple

100 km

26degS

25degS

24degS

28degE 29degE 30degE27degE26degE

Mafic Bushveld

Transvaal Supergroup

Felsic BushveldN

Pretoria

Lydenburg

BK1

25degS

30degS

Johannesburg

Rustenburg

BierkraalBK3

BK2

FarWestern

Limb

SouthAfrica

30degS

WesternLimb

NorthernPotgietersrus

Limb EasternLimb

SouthernBethalLimb

Fig 1 Map of the Bushveld Complex showing the location of the Bierkraal drill holes BK1 BK2 and BK3 Modified after Lundgaard et al(2006)

(1128 m)

(1662 m)

LowerZone

CriticalZone

MainZone

Upper

Zone

0

1000

2000

3000

4000

5000

60000

500

1000

MM

L

1500

2000

2500

Stra

tigra

phic

pos

ition

(m

etre

s)

Mag

netit

ite L

ayer

sN

elso

nite

Lay

ers

BushveldComplex

BierkraalDrill Core

MZL

MZU(2125 m)

UZb

UZc BK1

BK3

BK2PyroxeniteMarker

UZa(1862 m)

PyroxeniteMarker

Fig 2 Generalized stratigraphic section of the Bushveld Complex(left) and the combined Bierkraal drill cores (right) The bases ofsubzones in the Bierkraal cores delineate the lowest appearance ofcumulus magnetite (UZa) olivine (UZb) and apatite (UZc) respec-tively The correlations between the three Bierkraal drill cores are1600m depth in BK1 equals 550m depth in BK3 1420m depth inBK3 equals 200m depth in BK2 [see core logs given by Kruger et al(1987)] An inferred stratigraphic position in the Bierkraal cores iscalculated assuming the core is vertical that layering dips 24 to thenorth and that the roof contact is at 415m depth in BK1 Previouspublications on BK drill cores quote only absolute depths belowsurface To facilitate comparison with our inferred stratigraphicpositions the following conversion equations have been used BK1inferred stratigraphic position in metres frac14 (depth in BK1 ndash 415) middot cos24 BK2 inferred stratigraphic position in metres frac14 (depth in BK2 thorn1840) middot cos 24 BK3 inferred stratigraphic position in metres frac14 (depthin BK3 thorn 635) middot cos 24 MML Main Magnetitite Layer Modifiedafter Kruger et al (1987)

JOURNAL OF PETROLOGY VOLUME 47 NUMBER 11 NOVEMBER 2006

2258

Dow



up-section fractionation trends have however beendescribed and different interpretations presented Merkleamp von Gruenewaldt (1986) for example interpretedchanges in pentlandite and olivine compositions to resultfrom magma recharge and mixing Eales amp Cawthorn(1996) showed that the V2O5 content of magnetiteincreases up-section in at least one interval and notedthat this could not be explained by changes in intensiveparameters such as f O2 during fractional crystal-lization Likewise Ashwal et al (2005) documentedreversals in the Mg of pyroxene and An ofplagioclase in a drill core through the MZU and UZ inthe northern limb that were interpreted as evidence formagma recharge Many studies of the UZ have focusedon the petrogenesis of the magnetitite layers (Bateman1951 Wager amp Brown 1968 Irvine 1975 Cawthorn ampMcCarthy 1980 Klemm et al 1985 Reynolds 1985avon Gruenewaldt et al 1985 Kruger amp Smart 1987Harney et al 1990 1996 von Gruenewaldt 1993) Inter-pretations vary and will be discussed in detail belowHere we present new major element data for

plagioclase clinopyroxene olivine and orthopyroxeneV2O5 in magnetite and whole-rock Sr isotope data forthe 21 km thick section of MZU and UZ sampled in theBierkraal drill cores of the western limb (Fig 1) Withthe exception of one drill core in the northern limb(Ashwal et al 2005) there is a dearth of systematicelectron microprobe data for silicate minerals in theupper part of the Bushveld Complex The new datatogether with published P2O5 bulk-rock data for thesame drill core (Cawthorn amp Walsh 1988) demonstratepronounced cycles in mineral compositions V2O5 inmagnetite and the intermittent presence of apatiteand olivine To help explain the genesis of these cyclesand the formation of magnetitite and nelsonite layerswe developed a forward crystallization model thatpredicts the liquid line of descent the magma densityand instantaneous equilibrium mineral compositionsduring crystallization of the MZU and UZ

PETROLOGY AND GEOCHEMISTRY

OF THE MAIN AND UPPER ZONES

Zonal subdivision

Subdivisions of layered intrusions are normally basedon the appearance and disappearance of cumulusminerals and hence in principle should be identifiablein the field (Wager amp Brown 1968) Divisions based onmore sophisticated and geochemical criteria such asstratigraphic changes in Sr isotope ratios in the case ofthe Bushveld Complex may make genetic sense butare not easily implemented Here we will focus only onthe criteria for subdividing the Main and Upper Zones(Fig 2) In the upper part of the Main Zone in the

eastern limb von Gruenewaldt (1970) identified a thinorthopyroxenite layer known as the Pyroxenite MarkerBelow this level the rocks contain original pigeonitenow inverted to orthorhombic pyroxene with abundantexsolution and are referred to as the lower Main Zone(MZL) The Pyroxenite Marker and overlying rockscontain primary orthopyroxene A similar pyroxenitelayer and associated pyroxene phase changes has alsobeen identified in the western limb (Cawthorn et al1991) Several further studies have confirmed the lateralcontinuity of these mineralogical successions in bothlimbs (Mitchell et al 1998 Nex et al 1998 Lundgaardet al 2006) Through an interval of about 200m thePyroxenite Marker is also associated with a gradualupward increase in An in plagioclase and Mg inpyroxene and a change in initial 87Sr86Sr (Sr0) from07082 below to 07073 above demonstrating that thePyroxenite Marker formed as a consequence of magmarecharge and mixing (Kruger et al 1987 Cawthornet al 1991) Following the long established terminologyfor the Bushveld Complex (Wager amp Brown 1968) weto refer to the rocks immediately above the PyroxeniteMarker as the upper Main Zone (MZU) (Fig 2)Willemse (1969a 1969b) used the Main Magnetitite

Layer (Fig 2) to define the base of the Upper Zone (UZ)In contrast von Gruenewaldt (1973) suggested that theUZ was composed of four subzones with the base of thelowest subzone UZa being taken as the first appearanceof cumulus magnetite The base of subzone UZb wasplaced at the base of the Main Magnetitite Layer andthe appearance of olivine and apatite defined the basesof subzones UZc and UZd respectively We have certainreservations about these schemes The 25m thickMain Magnetitite Layer is certainly an excellent markerin the field However because the cumulus mineralogydoes not change across this layer we question whether itshould be used as a zonal or subzonal boundary Usingcumulus mineralogical criteria we suggest instead athree-fold subdivision following Wager amp Brown (1968)based on the lowest appearance of magnetite (UZa)olivine (UZb) and apatite (UZc) (Fig 2) The relativethicknesses of these subzones applied to the easternnorthern and western limbs are listed in Table 1 whichshows that UZa in the western limb as measured in theBierkraal drill core is relatively thin (200m) UZb is400ndash700m thick in all three limbs In contrast UZc iswell over 1100m thick in the western limb considerablythicker than in the eastern and northern limbs Incontrast MZU is 270m thick in the western limb butsignificantly thinner than in the east and north (Table 1Fig 2) Even this scheme is not perfect as both olivineand apatite appear intermittently above their firstappearance Their absence in several intervals withinUZb and UZc and marked reversals in mineral compo-sitions (see below) indicate that further subdivision is

TEGNER et al CYCLICITY IN BUSHVELD COMPLEX

2259

Dow



required Such subdivision has little application in thefield but is of great petrological significance We refer tothese further subdivisions as cycles

Petrography

The gabbronorites magnetite gabbros and diorites ofMZU and UZ in the eastern limb have been welldescribed previously (Wager amp Brown 1968 Willemse1969a von Gruenewaldt 1973 Molyneux 1974) andonly some important features are emphasized here asthe rocks of the western limb are extremely similarLayering is intermittently present throughout the UZThe magnetitite and nelsonite (magnetitendashilmenitendashapatite Cumulate) layers are the most conspicuous(Fig 3f) anorthosite layers are common and melano-cratic facies occur less frequently Any of the follow-ing minerals can occur as cumulus phases plagioclaseolivine clinopyroxene orthopyroxene (and invertedpigeonite) magnetite ilmenite sulphides and apatiteMinerals that are only intercumulus are biotite horn-blende quartz and potassium feldspar and appear moreabundantly toward the top of UZ Alteration is onlylocally developed and there has been no pervasivemetamorphismThe location of magnetitite and nelsonite layers in the

Bierkraal core is shown in Fig 2 and listed in Table 2Gradations exist from almost magnetite-free anorthositesto magnetitite layers with over 95 oxide Only thoselayers with greater than 50 oxides and thicker than2 cm are recorded in Fig 2 There are 26 magnetititeand six nelsonite layers with a cumulative thicknessof 204m (Table 2) Their mineralogy and textureshave been documented by Willemse (1969b) Reynolds(1985a) von Gruenewaldt et al (1985) and vonGruenewaldt (1993) Footwall and hanging-wall rocksto magnetitite layers are commonly anorthositic andlower contacts tend to be sharp whereas upper contacts

are gradational (von Gruenewaldt 1973 Molyneux1974) (Fig 3f) Anorthosites and less commonly otherplagioclase-rich rocks sometimes display a variablydeveloped planar fabric parallel to the layering(Figs 3a and b) Modal layering is occasionally developedin the interstitial phases to cumulus plagioclase (Fig 3b)Plagioclase usually has euhedral to subhedral grainshapes whereas olivine is anhedral even when it isabundant (Fig 3d) Pyroxenes vary in shape (Fig 3c)They are usually subhedral to anhedral even whenrelatively abundant In the upper parts of the UZclinopyroxene commonly displays ilmenite exsolutionand so TiO2 and FeO(total) contents from electronmicroprobe analyses should not be considered primaryApatite varies greatly in abundance and always formsprismatic grains They are commonly embedded inolivine- and magnetite-rich layers (Fig 3d) but rarely inpyroxene and plagioclase Magnetite is almost alwaysanhedral In the upper part of UZc six nelsonite layerswith up to 25 ilmenite and up to 30 apatite occur(Fig 3e Table 2) (Reynolds 1985a von Gruenewaldt1993) whereas below that level the magnetitite layers aredevoid of apatite Primary ilmenite is scarce in the lowerhalf of UZ but is ubiquitous as an exsolution phase

Samples from the Bierkraal drill core

A subset of 55 samples from the Bierkraal drill coreswas selected to obtain a systematic section of the MZU

and UZ These cores were previously investigated byCawthorn amp McCarthy (1985) Reynolds (1985b)Merkle amp von Gruenewaldt (1986) Kruger et al(1987) Cawthorn amp Walsh (1988) and von Gruenewaldt(1993) but compositional data on the silicate mineralsare few apart from the interval across the PyroxeniteMarker (Cawthorn et al 1991) The Bierkraal corematerial which was made available by the GeologicalSurvey of South Africa consists of three separate holes(BK1 BK2 and BK3) collared NE of Rustenberg in thewestern Bushveld Complex (Fig 1) Correlation betweenthe three cores has been presented (Walraven ampWolmarans 1979 Kruger et al 1987) based on cor-relation of apatite- and magnetite-rich layers The baseof the Main Magnetitite Layer appears at depths of171m and 1378m in BK2 and BK3 respectively(Kruger et al 1987 figure 2) In BK1 and BK3 thelowest occurrence of apatitendashmagnetite ferrodioritecumulates is at 1425 and 375m depth respectivelyand gives the best correlation The composite sectionshown in Fig 2 and used throughout this paper istherefore composed of the interval from 415m to1600m depth in BK1 at the top followed by the intervalfrom 550m to 1420m depth in BK3 and the intervalfrom 200m to 673m depth in BK2 at the bottom Thetrue stratigraphic position is calculated assuming that

Table 1 Stratigraphic thicknesses of subzones in the Main

and Upper Zones of the Bushveld Complex

Subzone Thickness (m)

West (1) East (2) East (3) North (4)

Upper Zone c (UZc) 1128 910 350 610

Upper Zone b (UZb) 534 740 520 390

Upper Zone a (UZa) 200 580 640 590

Main Zone upper portion (MZU) 273 700 590 mdash

Total stratigraphic thickness 2135 2930 2100 1590

(1) Bierkraal drill core this study (2) von Gruenewaldt(1973) (3) Molyneux (1974) (4) Ashwal et al (2005)


2260

Dow



the core is vertical and igneous layering dips 24 northand is reported in metres below the roof of the intrusion(located at 415m depth in BK1) The investigatedstratigraphic section between the Pyroxenite Marker andthe roof is thus 2125m thick which is slightly greater

than the standard section assumed for the western limb(2000m Eales amp Cawthorn 1996)At the top of the ferrodiorite in BK1 is a quartzite

fragment interpreted to be a xenolith and overlain bygranophyric rocks Both are intruded by granite sheets

Fig 3 Photographs showing typical textures of rocks from the Upper Zone of the Bushveld Complex The field of view in all photomicrographs is8mm middot 8mm and all sections are cut vertically to the core and so are nearly perpendicular to the layering (a) Strong planar fabric parallel tolayering displayed by cumulus plagioclase laths in UZc Clinopyroxene magnetite and olivine are anhedral but their proportions suggest that theyare cumulus phases Sample 1w5636 (b) Anorthosite in UZc In the lower part the only poikilitic phase is magnetite whereas in the upper part itis exclusively clinopyroxene Sample 1w111825 (c) Olivinendashmagnetitendashapatite gabbro dominated by subhedral plagioclase and anhedral olivineclinopyroxene and magnetite from UZc Apatite is present but rare in this photograph Sample 1w13417 (d) Pods enriched in olivine magnetiteand apatite in a more leucocratic host dominated by plagioclase from UZc Apatite is enclosed by olivine and magnetite but seldom by plagioclaseApatite is far less abundant in the plagioclase-rich areas Sample 1w53805 (e) Magnetitendashilmenitendashapatite layer (nelsonite) from UZc Apatiteforms euhedral grains whereas magnetite has polygonal grain boundaries Sample 1w11115 (f) Field photograph of a magnetitite layer showingsharp contact to underlying anorthosite and up-section decreasing abundance of magnetite and increasing abundance of euhedral plagioclase lathsFrom Magnet Heights in the eastern limb PLAG plagioclase CPX clinopyroxene MGT magnetite ILM ilmenite AP apatite


2261

Dow



The granophyric rocks are interpreted to be the originalroof rocks to the mafic sequence (Walraven 1987) Thepresence of this complex succession intersected heremeans that it is not absolutely certain that the extremedifferentiates of the mafic rocks have been preservedin this borehole However the iron-rich nature of themafic minerals (see below) suggests that not much canbe missing

Analytical methods

Mineral compositions were obtained using a JEOL8600 electron microprobe at the University of Aarhusfollowing the procedures described by Tegner et al(1999) Plagioclase was analysed using a slightlydefocused electron beam with a diameter of 10 mm tominimize the effect of Na and K volatilization Pyroxeneand olivine were analysed using a focused electron beamwith a diameter of 2mm to avoid problems of includinginclusions and exsolution lamellae Analyses of pyroxenetherefore represent subsolidus equilibrium compositionsWhen possible three points were analysed in the coresof each of three grains per sample and the reportedvalues (Table 3) are the average Analyses of anhydrousminerals with either compositional anomalies indicatingthat impurities were analysed or with an oxide sumlower than 985 wt or higher than 1015 wt wereexcluded from the average values reported in Table 3The full datasets for plagioclase clinopyroxene olivineand orthopyroxene are provided as SupplementaryDatasets 1ndash4 (available at httpwwwpetrologyoupjournalsorg)Mineral separates of magnetite were prepared as

described by Cawthorn amp McCarthy (1980) andanalysed for vanadium by X-ray fluorescence on pressedpellets Standard SARM12 was used as reference andstandard SARM38 for spiking of samples for calibrationWhole-rock powders were analysed for phosphorus alsoby X-ray fluorescence on pressed pellets and publishedpreviously by Cawthorn amp Walsh (1988)Sr isotope compositions and Sr and Rb concentra-

tions (by isotope dilution) were analysed on whole-rocksby thermal ionization mass spectrometry (TIMS) at theHugh Allsopp Laboratory of the Economic GeologyResearch Institute University of the WitwatersrandSouth Africa following the procedure described by Ealeset al (1990) The whole-rocks were crushed in a jawcrusher milled using a Siebtechnik swing mill and finelyground in an automatic agate mortar and pestle Fordissolution 01g aliquots of the powder were added toa previously spiked (84Sr and 87Rb) solution and driedin a Teflon beaker and the mixture was dissolved in amixture of distilled HF and HNO3 The solutions weredried and taken up in 6N HCl and checked for anyresidue The solution was then dried and taken up in2ml 25N HCl and loaded on an ion exchange columnand eluted and the Sr was recovered A small proportionof the dried Sr was loaded with phosphoric acidon a single outgassed Ta filament and determined byTIMS using a multicollector system The data reductionwas done during the run Rb was loaded on a doublefilament directly from the dissolved sample withoutseparation from the matrix Run temperature was con-trolled below the Sr evaporation of the side filament and

Table 2 Position and thickness of 26 magnetitite and six

nelsonite layers Bierkraal drill core western Bushveld

Complex

Sample no Subzone Stratigraphic position Thickness (cm)

1w784y UZc 3371 2

1w823y UZc 3727 10

1w885y UZc 4293 6

1w1099y UZc 6248 10

1w11126y UZc 6373 30

1w1117 UZc 6413 6

1w1206y UZc 7226 6

1w12841 UZc 7939 3

1w14498 UZc 9453 20

1w14505 UZc 9459 10

1w14513 UZc 9467 70

1w1460 UZc 9540 710 (Layer 21)

1w1465 UZc 9592 68

1w1485 UZc 9774 25

1w14882 UZc 9804 40

1w14892 UZc 9813 3

1w14921 UZc 9839 5

1w149265 UZc 9844 20

1w1494 UZc 9857 60

1w1532 UZc 10204 14

3w746 UZb 12615 80

3w927 UZb 14269 43

3w932 UZb 14315 60

3w942 UZb 14406 10

3w1247 UZa 17192 26

3w1272 UZa 17420 134

3w1294 UZa 17621 107

3w1313 UZa 17795 64

3w1315 UZa 17813 26

3w1343 UZa 18069 53

3w1368 UZa 18297 75

3w1378 UZa 18389 246 (MML)

Total thickness 2042

Layers thicker than 2 cmyNelsonite layersMML Main Magnetitite Layer


2262

Dow



Table 3 Average mineral compositions Bierkraal drill cores western Bushveld Complex

Sample Strat Zone Cycle Plagioclase Clinopyroxene Olivine Orthopyroxene Bulk-rock

no pos An (1 SD) n Mg (1 SD) n Fo (1 SD) n Mg (1 SD) n Sr0 (2 SE)

1w4222 66 UZc IX 430 (23) 9 114 (07) 6 30 (02) 6

1w4318 154 UZc IX 452 (24) 9

1w4466 289 UZc IX 428 (14) 9 49 (04) 7 13 (01) 9

1w4751 549 UZc IX 434 (24) 8

1w5047 819 UZc IX 451 (18) 9 159 (03) 6 39 (03) 9

1w534 1087 UZc IX 446 (23) 8 308 (03) 4 94 (03) 9

1w568 1398 UZc IX 500 (19) 9 364 (07) 4 104 (09) 9

1w598 1672 UZc IX 463 (06) 8 268 (07) 7 78 (03) 9

1w62585 1926 UZc IX 498 (20) 8 502 (05) 3 214 (01) 9 369 (08) 6

1w6633 2268 UZc IX 487 (16) 9

1w732 2896 UZc VIII 519 (08) 9 436 (10) 5 220 (02) 8

1w770 3243 UZc VIII 493 (04) 9 469 (10) 4 216 (03) 9

1w8086 3596 UZc VII 505 (14) 8

1w8532 4003 UZc VII 498 (05) 8 462 (05) 6 187 (04) 9

1w9086 4509 UZc VII 495 (11) 8 133 (01) 4

1w954 4924 UZc VI 498 (08) 8 357 (05) 6 148 (05) 9

1w10102 5437 UZc VI 489 (13) 7

1w10505 5805 UZc VI 501 (05) 8 362 (08) 5 159 (01) 9

1w11382 6605 UZc VI 501 (09) 9 430 (04) 6 175 (04) 6 070711 (13)

1w11586 6793 UZc VI 481 (12) 6 255 (10) 5 62 (02) 9 070739 (14)

1w11901 7085 UZc V 456 (10) 7 325 (05) 6 146 (02) 9 070720 (13)

1w12395 7528 UZc V 483 (06) 9 383 (08) 5 160 (03) 8 070717 (15)

1w13032 8112 UZc V 070726 (13)

1w13292 8351 UZc V 509 (07) 8 473 (04) 4 232 (02) 9 070714 (13)

1w14234 9212 UZc V 520 (04) 9

1w148885 9810 UZc V 550 (07) 9 578 1 449 (06) 15 070758 (16)

1w15070 9975 UZc IV 492 (08) 7 475 (13) 4 341 (07) 8

1w15507 10375 UZc IV 485 (09) 9 566 (07) 5 340 (02) 6 070726 (13)

3w540 10734 UZb IV 498 (09) 8 375 (07) 9

3w590 11190 UZb IV 535 (13) 7

3w6409 11655 UZb IV 559 (10) 9 617 (05) 6 504 (12) 6

3w715 12332 UZb III 521 (15) 9 643 (11) 8 389 (14) 9

3w742 12579 UZb III 554 (08) 9

3w780 12926 UZb III 577 (18) 9 596 (08) 9

3w810 13200 UZb III 582 (04) 8

3w885 13885 UZb II 564 (08) 9 597 (08) 5

3w938 14369 UZb II 578 (06) 8

3w1015 15073 UZb II 558 (07) 9

3w1112 15959 UZb II 563 (06) 8 641 (09) 8 442 (02) 3 561 (07) 6

3w121225 16875 UZa II 574 (08) 8

3w12678 17382 UZa II 585 (11) 9

3w12689 17392 UZa II 607 (07) 8 673 (11) 9

3w1295 17631 UZa II 599 (12) 9

3w13147 17811 UZa I 571 (09) 8 662 (07) 5 552 (05) 12

2w124 17959 UZa I 585 (05) 9

3w1360 18224 UZa I 611 (04) 9

3w138125 18418 MZU I 607 (08) 7 616 (34) 5 548 (01) 3

2w225 18882 MZU I 642 (06) 9

2w324 19786 MZU I 674 (14) 9 727 (15) 9 659 (05) 9

2w402 20499 MZU I 713 (20) 8 706 (10) 7 647 (03) 8

2w448 20919 MZU I 708 (28) 7

2w4819 21229 MZU I 720 (24) 9 740 (08) 7 676 (04) 9

2w4919 21320 MZU 723 (10) 9

2w601 22317 MZL 674 (04) 8

2w6728 22973 MZL 571 (13) 8

An frac14 100Ca(Ca thorn Na) Mg and Fo frac14 100Mg(Mg thorn Fe) all molar proportions with all iron calculated as Fe2thornSamples from drill core BK1 frac14 1w BK2 frac14 2w BK3 frac14 3w for example sample 2w402 is collected at 402m depth in BK2


2263

Dow



the 88Sr peak position was monitored during the TIMSrun No Sr was detected in any run The measured87Sr86Sr and the 2 SE (lt0011) based on the runstatistics are listed in Supplementary Dataset 5 (httpwwwpetrologyoupjournalsorg) The accuracy of themeasurements was determined by measuring theEimar amp Amend Sr standard which gave 87Sr86Sr of070800 plusmn 000002 (2 SE) and the SRM987 standardwhich gave 87Sr86Sr 071023 plusmn 000002 (2 SE) thesevalues are within error of the recommended valuesInitial 87Sr86Sr values referred to as Sr0 are calculatedand reported in Table 3 and Supplementary Dataset 5The 2 SE on Sr0 is important to judge the petrogeneticsignificance of the measured 87Sr86Sr and has beenestimated using 14 replicates of a finely ground Bushveldnorite These replicates suggest that the 2 SE onmeasured 87Sr86Sr is 0018 The 2 SE on 87Rb86Srused for age correction is considerably larger (1) anddepends on three factors sample and spike weightsand calibration errors in the spikes sample hetero-geneity and analytical error The total procedural blankvalues which were determined to be lt100 pg for bothRb and Sr are 1 of the total sample and wereignored

Mineral compositions and stratigraphicsystematics

In the Bierkraal drill core the compositions of plagioclasecores decrease systematically from An72 [An frac14 100Ca(Ca thorn Na)] at the Pyroxenite Marker to An43 at the topof UZc (Fig 4a) This up-section decline however is notcontinuous as assumed in previous studies based onfew and widely spaced samples (Wager amp Brown 1968von Gruenewaldt 1973 Molyneux 1974) Several up-section increases in An are significantly larger thananalytical error (plusmn1 SD) and are not artefacts ofcorrelation between the three cores sampled (Table 3)We refer to these stratigraphic intervals as reversals Thesections between reversals display either near-constantor up-section declining An (Fig 4a) We have usedthe reversals to higher An as one criterion tosubdivide MZU and UZ into cycles that are unrelatedto the accepted zonal subdivision (Figs 2 and 4) Thereversals typically occur over 15ndash175m of section andrange in magnitude from 3 to 6An (Table 3 Fig 4a)For comparison plagioclase changes from An57 to An72over 180m section in the reversal across the Pyrox-enite Marker (Fig 4a) (Cawthorn et al 1991) Given theaverage spacing of 40m between samples the exact

500

1000

1500

2000

2500

Pyroxenite Marker

Clinopyroxene Mg Olivine amp Opx Mg Sr0Plagioclase An

Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

10 30 50 70 10 30 50 7050 60 70 07070 07075

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es

Olivine (this study)Kruger et al(1987)

OPX (this study)

OPX (Cawthornet al 1991)

Cawthorn et al(1991)

This study


(a) (d)(c)(b)

This study This study

Olivine(published data)

subz

ones

UZc

UZb

UZa

MZU

MZL

07073plusmn00001of Krugeret al(1987)

07085

mag

netit

ite la

yers

nels

onite

laye

rs

Fig 4 Compositional variation of (a) plagioclase (An) (b) clinopyroxene (Mg) (c) olivine (Fo) and orthopyroxene (Mg) and (d) initial87Sr86Sr (Sr0) with stratigraphic position in the Bierkraal drill cores Data from Table 3 Supplementary Datasets 1ndash6 Reynolds (1985b) Merkle ampvon Gruenewaldt (1986) Kruger et al (1987) and Cawthorn et al (1991)


2264

Dow



location of cycle boundaries is only approximate Forthe section from the Pyroxenite Marker to the middleof UZc we have chosen to place the base of eachcycle immediately below the lowest sample showing amarked reversal in An In this way we have identifiedcycles IndashVI (Fig 4a Table 3) In Fig 4 a further threecycles (VIIndashIX) are shown where there is no apparentreversal in An These cycles are defined on the basis ofthe disappearance of apatite but can be explained in thesame way as cycles IndashVI (see discussion) Within somecycles (I IV and V) An declines smoothly up-sectionwith an average rate of decline of one An per 18ndash27mof section In other cycles (II VIndashVIII) the trends arelargely dispersed around a constant An (Fig 4a)The Mg [100Mg(Mg thorn Fe)] of clinopyroxene

declines from 74 at the Pyroxenite Marker to Mgcpx 5close to the top of the core (Fig 4b Table 3) and cor-relates positively with An (Fig 5) As with plagioclaseclinopyroxene displays reversals in Mgcpx acrossseveral cycle boundaries The most marked reversals inMg are between cycles IV and V (48ndash58) and betweenV and VI (26ndash43) In cycles VIndashVIII the sample spacingis too large to resolve reversals in Mgcpx but a generalup-section increase from 43 to 50 is evident In themiddle and upper portion of cycle IX Mgcpx dropsrapidly from 50 to 5 close to the top Similar to thesection across the Pyroxenite Marker (Cawthorn et al1991) the stratigraphic position of reversals in An andMg may be slightly offset (Fig 4b Table 3) The rate ofupward decline in Mgcpx is moderate in the lowercycles I and II at one Mg unit per 24 and 74mof section respectively In cycles IV and V the rate ofdecline is greater at 1Mgcpx per 11 and 9m res-pectively An important observation for the followingdiscussion of the differentiation trend is that cumulaterocks with Mgcpx and plagioclase An gt52 Mgcpx islarger than An whereas the reverse pattern is observedin the more evolved cumulates (Fig 5)The up-section variation in the Fo content [100Mg

(Mgthorn Fe)] of olivine (Fig 4c) is shown not only by our newdata (22 samples Table 3) but also previously publishedelectron microprobe data for BK1 by Reynolds (1985beight samples) Merkle amp von Gruenewaldt (1986 ninesamples) and unpublished data (seven samples) from1980 by RGC obtained using the electron microprobeat the University of Bloemfontein South Africa (Supple-mentary Dataset 6 httpwwwpetrologyoupjournalsorg) The Fo decreases from Fo44 in cycle II to Fo1 atthe top of UZc Again this up-section decrease in Focontent is interupted by reversals that coincide withreversals in Mgcpx and An (Fig 4) Some of theseincreases in Fo are relatively large for example fromFo34 to Fo52 across the boundary between cycles IVand V and from Fo6 to Fo29 between cycles V and VIWithin cycle V the olivine composition changes from

Fo54 to Fo6 over only 300m of stratigraphic sectionAs observed for the Mgcpx and An from the top ofcycle VI to the base of cycle IX the Fo is dispersedaround a slightly increasing trend Above this level Fodeclines sharply to virtually pure fayalitic compositionsat the top of UZc Figure 4c also shows that the Mgof orthopyroxene in 10 samples ranges from Mgopx 68to Mgopx 34 Although the samples are widely spacedMgopx conforms with the trends shown by Mgcpx

and Fo The FeMg exchange coefficient KD(FeMg)between orthopyroxene and clinopyroxene is relativelyconstant (13 and 14 five pairs) in cycle I but increasesto 17 (three pairs) in cycle IV This is similar to experi-mental data for FeMg exchange between coexistingpyroxenes (Toplis amp Carroll 1995) suggesting that thepyroxenes are in equilibrium

Plagioclase An across magnetitite layers

In the Bierkraal drill core some of the reversals in plagio-clase An occur in sequences with abundant magnetititelayers whereas others occur in normal leucocraticgabbronorites (Fig 4) A detailed study of plagioclasecomposition across magnetitite layers in these cores isbeing undertaken but here we refer to the relation-ship between An across magnetitite layers (Fig 6) inclosely spaced gabbronorite samples across the MainMagnetitite layer and the two subsidiary magnetititelayers (Layer 1 Layer ndash3) below and above the MainMagnetitite Layer from outcrops at Magnet Heights inthe eastern limb (Fox 1982) Although the plagioclase

020

40

60

80

0

10

20

30

40

50

60

70

80

20 30 40 50 60 70 80

cycle I

cycle II

cycle III

cycle IV

Mg = An

+apatite

+magnetite

Fractionalcrystallizationmodel

cycle V

cycle VI

cycle VII+VIII

cycle IX

Plagioclase An

Clin

opyr

oxen

e M

g

Fig 5 Mg of clinopyroxene vs An of plagioclase for the nine cyclesin MZU and UZ of the Bierkraal drill cores Data from Table 3Continuous line with tick marks shows calculated Mg of clinopyrox-ene and An of plagioclase for a forward fractionation model Datafrom Table 5 Tick marks each represent 10 crystallization Dashedlines mark the approximate onset of magnetite and apatite crystal-lization Fine continuous line shows Mg frac14 An


2265

Dow



composition varies from An51 to An57 and displays asystematic up-section decrease below magnetitite layers 1and ndash3 the An is identical within error in samplesimmediately below and above each of the threemagnetitite layers Similarly Harney et al (1996) foundno change in the An content of plagioclase below andabove two sections of the Main Magnetitite Layer Thesedata suggest that fluctuations in An content are notspecifically related to the formation of magnetitite layers

V2O5 content of magnetite

The V2O5 content of magnetite separated from 266samples is shown in Fig 7 From its first appearance as acumulus mineral near the top of cycle I to the base ofcycle IV the concentration decreases relatively uniformlyfrom about 17 to 04 (excluding two aberrant valuesin cycle III) Through the remainder of cycle IV and ineach of the subsequent cycles it shows relatively highconcentrations near the bases and rapid decreases up-section reaching close to detection limits near the top ofeach cycle However the highest concentrations do notoccur abruptly at the bases of each cycle but climb from

the low values from the top of the previous cycle to ahigh value typically several tens of metres above thereversal identified by the plagioclase composition Thusthere is no close relation between the presence ofmagnetitite layers and reversals in V content

Sr isotope compositions

The initial 87Sr86Sr (Sr0) composition of the Bierkraaldrill cores was determined previously and showed amarked shift across the Pyroxenite Marker from07085 in MZL to 07073 in MZU and UZ (Krugeret al 1987 Cawthorn et al 1991) (Fig 4) The constancyof Sr0 above the Pyroxenite Marker determined as07073 plusmn 00001 (2 SE) from the intercept of a 2066 plusmn58Ma regression line in an isochron diagram (Krugeret al 1987) was explained by complete mixing andhomogenization between residual (Sr0 frac14 07085) andrecharged (Sr0 frac14 07067) magma in proportions close to11 The near-constancy of Sr0 also implies that additionof further magma above the Pyroxenite Marker can beruled out unless it had Sr0 of 07073 (Kruger et al1987 Cawthorn et al 1991) Because our interpretationof the cycles presented here hinges on whether newmagma was added or not we have determined Srisotope compositions for eight additional samples acrossthe boundaries between cycles IV V and VI Seven ofthe eight new Sr0 determinations range from 07071 to07074 (Fig 4 Table 3) and are within error of 07073 plusmn00001 determined previously for MZU and UZ (Krugeret al 1987) Although the Sr0 (07076 plusmn 00002) ofsample 1w148885 (stratigraphic height of 980m) at thebase of cycle V is marginally higher than that of theother samples analysed (Table 3 Fig 4) the combinedSr isotope datasets suggest constancy of Sr0 in MZU andUZ The Sr0 of proposed recharge magmas to the entireBushveld Complex ranges from 07045 to 07090 butnone has compositions close to 07073 (Kruger 1994) Inthe Bethal area located SW of the eastern limb (Fig 1)subsurface mafic rocks have Sr0 of 07055 and havebeen explained as the products of crystallization fromunadulterated Upper Zone magma (Kruger 2005) Wetherefore conclude that recharge with magma withSr-isotopic composition similar to proposed Bushveldmagmas can be ruled out Therefore an internalmechanism for generation of the layered MZU and UZsequence must be sought

Phosphorus content in whole-rock samples

Figure 7a shows the variation in bulk-rock P2O5 (wt )of the Bierkraal drill cores [data from Cawthorn ampWalsh (1988)] In the lower part from cycle I to the basalpart of cycle IV apatite is not a cumulus phase andP2O5 is very low (lt010 wt ) In the middle and upper

(a)

(b)

12

6

8

10

minus2

0

2

4

6

8

10

minus2

0

2

4

minus4

Magnetitite Layer 1

Main Magnetitite Layer

Magnetitite Layer -3

An in plagioclase

Stra

tigra

phic

Hei

ght (

met

res)

Stra

tigra

phic

Hei

ght (

met

res)

51 575553

Fig 6 Compositional data for plagioclase (An) across (a) the MainMagnetitite Layer and Magnetitite Layer 1 and (b) Magnetitite Layer3 at Magnet Heights in the eastern limb of the Bushveld ComplexData from Fox (1982)


2266

Dow



part of cycle IV P2O5 is highly variable and a numberof samples contain between 1 and 10 wt (Fig 7a)Abundant cumulus apatite at this level is taken asdefining the base of UZc as discussed above Above thislevel P2O5 displays pronounced cyclicity In cycle VP2O5 is low (lt03 wt ) in the basal 50m but increasesup-section to 99 wt over only 16m (Fig 7aSupplementary Dataset 7 available at httpwwwpetrologyoupjournalsorg) Hereafter P2O5 declinessmoothly to 08 wt through 210m of section In thefollowing 56m P2O5 drops to much lower values (009ndash017 wt ) and apatite is no longer a cumulus phaseThe subsequent low-P interval (50m thick and definedby seven samples) coincides with the reversal in AnMgcpx and Fo between cycles V and VI (Fig 4)A similar pattern in P2O5 is repeated four times in theupper portion of the core (Fig 7a) For reasons discussedbelow we have placed a cycle boundary at the baseof each low-P2O5 interval as shown in Figs 4 and 7 Incycle VI the most apatite-rich rocks are nelsonite layers(Fig 3e) which exhibit extremely high contents of up

to 195 wt P2O5 (Fig 7a Supplementary Dataset 7)The associated normal leuco- and mesocratic ferrodior-ite samples contain up to 10 wt P2O5 (Cawthorn ampWalsh 1988) Another observation is that the P2O5

content of low-P2O5 intervals increases gradually up-section from lt01 wt below the appearance of apatiteto 04 wt at the top of the core (Fig 7a) A total of450 analyses of P2O5 were obtained through the UpperZone (Cawthorn amp Walsh 1988) and so the location ofbreaks is more rigorously defined than by other criteria

FRACTIONATION MODELLING

Background and assumptions

To guide the quantitative interpretation of the evolutionof the entire MainndashUpper Zone succession and the dif-ferentiation in each cycle as shown by mineral composi-tions (Figs 4 and 7b) and bulk-rock P2O5 contents(Fig 7a) we have estimated the possible compositionsof evolving residual magma and equilibrium cumulate

001 01 001 01 11 10

0

500

1000

1500

2000

2500m

agne

titite

laye

rsne

lson

ite la

yers

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es

subz

ones

UZc

UZb

UZa

MZU

MZL

Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

BK1BK3

(a) (b)

P2O5 (wt)(bulk-rock)

V2O5 (wt)(magnetite)

Fig 7 Wt P2O5 (a) of whole-rocks and V2O5 in magnetite (b) plotted against stratigraphic position in the composite section based on theBierkraal drill cores (Note logarithmic scales on the x-axes) Phosphorus data from Cawthorn amp Walsh (1988) V data are given in SupplementaryDataset 7


2267

Dow



assemblage using a forward model of closed-systemfractional crystallization of a plausible parental magmaBelow we first present the assumptions and rationaleused in estimating the parental magma for the cumulaterocks above the Pyroxenite Marker We then describethe mass-balance calculations for fractional crystalliza-tion A simple mass-balance calculation is preferredbecause thermodynamic algorithms simulating crystal-lization (eg Ariskin et al 1993 Ghiorso amp Sack 1995)cannot be constrained for evolved ferrodioritic magmassuch as those appropriate to this section of the BushveldComplexIn applying such thermodynamic algorithms specific

problems arise with selection of f O2 and H2O contentThe water content of basic magmas is probably lowHowever because calculations of up to 80 fractiona-tion are considered here the water content and itspartial pressure may become significant as demon-strated by the presence of hornblende (although not as acumulus phase) in the evolved Bushveld rocks Thestability of magnetite and its proportion crystallizing isvery strongly influenced by f O2 In the experimentalstudy by Toplis amp Carroll (1995) they presented analysesof two quenched liquids formed at the same temperature(1072C) from the same starting composition but atf O2 differing by 2 log units that contained 62 and53 wt SiO2 and 9 and 17 wt FeO (total) res-pectively In fact much of the debate about the evolution

of the Skaergaard intrusion hinges around this issue [seesummary by Tegner (1997)] In the Upper Zone of theBushveld Complex there is no independent measure ofthe prevailing fO2 or whether it remained constantThus calculations that require knowledge of f O2 arenot constrainable and we prefer to use a mass-balanceapproach that includes the mineral proportions actuallyobserved in the succession

Calculated parental magma composition

The preferred calculated parental magma compositionthat produced the succession from the PyroxeniteMarker to the top of the intrusion is given in column 6of Table 4 To obtain this estimate we have determinedthe bulk composition of the preserved cumulates andadded an estimated evolved residual component that isthought to have escaped from the intrusion (Cawthorn ampWalraven 1998) Outcrop of the western limb ofthe Bushveld Complex is poor and most studies on theMain and Upper Zones have been undertaken on theeastern limb Specifically we note that no systematicstudy of whole-rock compositions is available from thewestern limb However there are remarkable similaritiesof the entire sequence and also distinctive layers inboth limbs such as the Pyroxenite Marker the MainMagnetitite Layer (2m thick) and Magnetitite Layer21 (7m thick) and also identical initial Sr isotope ratio

Table 4 Calculation of parent magma composition at the level of the Pyroxenite Marker

Oxide (wt ) Average composition

of cumulate above

Pyroxenite Marker

Estimate of residual

melt at Skaergaard

Quenched melt A Quenched melt B Quenched melt C Calculated composition

of melt at

Pyroxenite Marker

1 2 3 4 5 6

SiO2 476 732 643 661 666 514

TiO2 12 05 20 21 04 10

Al2O3 173 133 120 134 136 166

FeO(total) 135 38 96 58 46 117

MnO 02 01 01

MgO 56 02 12 16 05 46

CaO 110 18 43 37 45 97

Na2O 27 41 36 23 39 29

K2O 04 33 18 29 18 07

P2O5 044 11 01 04

Total 999 1002 988 990 960 991

Column 1 average of all compositions (44 samples) of rocks above the level of the Pyroxenite Marker in the easternBushveld given by von Gruenewaldt (1971) plus 1 wt titanomagnetite Column 2 estimate of residual melt after 75crystallization of the Skaergaard intrusion (Hunter amp Sparks 1987) Column 3 quenched melt analysed by Toplis amp Carroll(1995) formed at 1057C Column 4 quenched melt analysed by vander Auwera amp Longhi (1994) formed at 1071C Column5 quenched melt analysed by Spulber amp Rutherford (1983) formed at 927C No value for phosphorus was given We haveincluded 01 for the purpose of this calculation Column 6 calculated melt compositon present at level of PyroxeniteMarker assuming 80 cumulate (column 1) and 20 residual melt (column 5)


2268

Dow



(Eales amp Cawthorn 1996) Hence in the absence ofsuitable data from the western limb we resort toinformation and data from the eastern limb in thefollowing discussion We have averaged all the whole-rock analyses from von Gruenewaldt (1971) for thissection in the eastern limb No analyses of magnetititelayers were included Our measurements here (Table 2)suggest that the magnetitite layers comprise 20mout of 2125m or 1 of the total thickness Hencewe have added the equivalent of 1 titanomagnetiteto this average This bulk composition is given inTable 4 column 1 The CIPW norm of this composi-tion contains 15 olivine 15 diopside and 5hypersthene Such a melt composition if totally liquidwould crystallize olivine and so is not consistent withthe observed gabbronoritic mineral assemblages inthe MZUCawthorn amp Walraven (1998) used a mass-balance

approach involving the compilation of minor and traceelement data for this entire section to suggest that therehad been loss of some evolved magma during crystal-lization The composition and proportion of this lostcomponent is impossible to quantify from the Bushveldrocks themselves The most evolved rocks found inthe intrusion are almost certainly cumulative and so donot represent melt compositions This final melt musthave been in equilibrium with olivine orthopyroxeneclinopyroxene plagioclase magnetite ilmenite andapatite We have not found any experimental data thatperfectly fit this requirement but present some analysesin Table 4 that probably bracket this compositionVander Auwera amp Longhi (1994) gave an analysis ofa melt in equilibrium with orthopyroxene pigeoniteplagioclase clinopyroxene ilmenite and magnetite at1071C Toplis amp Carroll (1995) gave an analysis of meltin equilibrium with plagioclase clinopyroxene magne-tite and ilmenite at 1057C Spulber amp Rutherford(1983) gave an analysis of melt in equilibrium witholivine clinopyroxene pigeonite plagioclase ilmeniteand magnetite at 925C which we consider to be aplausible temperature for the final residual melt for theBushveld Complex By way of comparison we includein Table 4 a calculated composition for the evolvedmagma to the Skaergaard intrusion by Hunter amp Sparks(1987) These analyses probably bracket the meltcomposition at the end of differentiation of the UpperZone in the Bushveld Complex The proportion of thismelt that has escaped is even harder to predict butCawthorn amp Walraven (1998) estimated 20 We haveadded 20 of the analysis determined by Spulber ampRutherford (1983) to the bulk cumulate composition inTable 4 as an approximation to the melt that existed atthe level of the Pyroxenite Marker The CIPW normof this composition contains 1 quartz 13 diopsideand 22 hypersthene and is expected to crystallize

orthopyroxene rather than olivine We note that thisestimate of melt composition is model-dependent butwe demonstrate that it yields an internally consistentfractionation model Our physical model presentedbelow does not depend upon the quantitative accuracyof this composition but it provides an illustration ofplausible differentiation trends

Cumulus proportions andmineral compositions

The cumulus proportions in weight per cent have beencalculated from the modal data of von Gruenewaldt(1971) for each subzone and are given in Table 5 andFig 8 The differentiation of the proposed parentalmagma at the level of the Pyroxenite Marker (column 6in Table 4) has been modelled in steps of 2 crystal-lization (Table 5 Fig 8) Gabbronorite crystallizesto produce the MZU Magnetite co-crystallization isassumed to begin at Mgcpx 67 and An 61 asobserved in this study (Fig 4) This produces cumulaterocks with modes similar to those of UZa (Table 5Fig 8) Olivine is then assumed to co-precipitate toproduce cumulate rocks similar to UZb Apatite isassumed to join the crystallizing assemblage when theP2O5 content of the evolving magma reaches 10 wt (Cawthorn amp Walsh 1988) to form cumulates equivalentto those of UZcThe Mg values of olivine ortho- and clinopyroxene

are calculated assuming a KD(FeMg) between crystaland melt that changes linearly from 03 to 04 029 to024 and 025 to 019 respectively during crystallization(Toplis amp Carroll 1995 Toplis 2005) Minor oxideabundances in calculated pyroxene compositions usedin our calculations are intermediate between those ofAtkins (1969) based on mineral separates and ourelectron microprobe data The former may includeimpurities whereas the latter are influenced by exsolu-tion effects Al2O3 contents are 2 and 1 wt for clino-pyroxene and orthopyroxene CaO is taken as occupying09 cation positions in the clinopyroxene formula and2 in orthopyroxene TiO2 contents are calculatedusing a partition coefficient of 04 for clinopyroxene Forplagioclase KD(NaCa) is assumed to change from 08to 14 during crystallization (Toplis amp Carroll 1995)Ilmenite comprises less than 10 of the oxide phase

through most of the Upper Zone and becomes a signi-ficant cumulus phase only towards the extreme top ofthis section (Reynolds 1985b) However its modalproportion is not quantified and so it has not beenincluded in these models It is qualitatively included inthe calculations in that the titanomagnetite compositionthat is extracted in these calculations increases from10 to 20 wt TiO2 from bottom to top of the UpperZone (Molyneux 1972 Reynolds 1985b)


2269

Dow



Evolution of melt composition

Over the interval from 0 to 80 crystallization(F frac14 1ndash02) the calculated residual magma evolvesfrom a slightly iron-rich tholeiitic basalt (514 wt SiO2 46 wt MgO 116 wt FeOtot and 36 wt Na2O thorn K2O) to an iron-rich dacite with 677 wt SiO2 01 wt MgO 84 wt FeOtot and 72 wt

Na2O thorn K2O (Table 5) During crystallization of rockscorresponding to MZU the forward model predicts thatthe FeOtot of the magma increases to 145 wt andSiO2 remains largely constant at 52 wt (Table 5Fig 8) In the crystallization interval where magnetitegabbronorites equivalent to UZa are produced SiO2

starts to increase slightly and FeOtot remains constant

Table 5 Calculated compositions of magma cumulate and minerals and magma density in fractional crystallization model

F 100 090 080 070 064 054 046 030 020

Mineral appearing PlOpCp Mgt Ol Ap

Magma composition (wt )

SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014

Magma density (gcm3) 268 268 269 269 270 268 266 257 245

Proportion of cumulus minerals

Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002

Mineral compositions

Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378

Cumulate bulk composition

SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112

Pl plagioclase Op orthopyroxene Cp clinopyroxene Mgt magnetite Ol olivine Ap apatite F fraction of meltremainingCalculated compositions of minerals extracted in the fractionation model are given in Supplementary Dataset 8


2270

Dow



In the crystallization interval corresponding to UZb theappearance of iron-rich olivine in the crystallizationassemblage causes an increase in melt SiO2 to 55 wt and FeO slowly decreases In the more evolved modelmagmas crystallization of apatitendashmagnetitendashilmeniteolivine gabbronorites equivalent to UZc drive SiO2 up to677 wt and FeO down to 84 wt after 80crystallization

Evolution of melt density

The density of the evolving magma is perhaps the mostimportant factor in magma chamber dynamics and hasbeen calculated following McBirney (1993) includingthe partial molar volume of phosphorus (Toplis et al1994) During crystallization of rocks of the MZUthe calculated magma density increases from 268 to270 gcm3 (Table 5 Fig 8) After magnetite starts to

26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210

Proportion of magma remaining (F)

(a) Mineral proportions assumed in cumulates

(c) Model plagioclase and augite composition in cumulates

(d) Model magma density (gcm3)

(e) Subzones in the Bushveld Complex corresponding to the model

(b) Model magma composition (wt)

plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene

orthopyroxene olivinemagnetite

apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene

Fig 8 Results of a forward incremental fractional crystallization model Data from Table 5 (see text for explanation) (a) Mineral proportionsassumed in the extracted cumulate rock (b) Major element liquid line of descent (c) Calculated Mg of clinopyroxene and An of plagioclase inequilibrium with the magma (d) Calculated magma density (e) Subzones in the Bushveld Complex corresponding to the model


2271

Dow



crystallize at the level corresponding to the base of UZmagma density decreases continuously and reaches245 gcm3 after 80 crystallization

Evolution of cumulus mineral compositions

Over the interval from 0 to 80 crystallization Mgcpx

in equilibrium with the evolving magma decreases from76 to 9 and the An of equilibrium plagioclase changesfrom 69 to 38 (Fig 8c Table 4) The first olivine tocrystallize is Fo50 and after 80 crystallization it hasevolved to Fo4 (Table 5) The slow evolution of Anrelative to Mgcpx is explained by a modest KD(NaCa)for plagioclase close to unity (08ndash14) whereas theKD(FeMg) values for the mafic phases are lower (019ndash035) and produce more dramatic changes in Mg Thisresults in a cross-over from cumulates where Mgcpx

exceeds An in the interval from 0 to 65 crystal-lization to the opposite in the more evolved magmas(Fig 8c) In other words the model predicts the typicalfractionation trend of tholeiitic intrusions towardsextremely iron-rich end-members of the mafic phaseswhereas the An of plagioclase remains relativelyelevated eg An32 in the Skaergaard intrusion (Wageramp Brown 1968 Tegner 1997) and An30ndash40 in theBushveld Complex (Wager amp Brown 1968 vonGruenewaldt 1973 Molyneux 1974 Ashwal et al2005 this study) (Fig 5)

DISCUSSION


We note a caution regarding comparison between cal-culated mineral compositions and electron microprobeanalyses (Fig 5) We have analysed the cumulus coreof plagioclase grains Even in zoned grains diffusion willbe extremely slow (Morse 1984) and so primarycompositions will be preserved However for the maficminerals post-cumulus re-equilibration will occur andzoned grains will homogenize Furthermore Mg andFe partitioning between clinopyroxene and orthopyr-oxene changes with falling temperature as first docu-mented by Kretz (1963) and applied to the pyroxenesof the Bushveld Complex by Atkins (1969) We notevariations in the KD(FeMg) values between the twopyroxenes in our data which we attribute in part to theslow cooling of the intrusion Also the clinopyroxenesin the Upper Zone display exsolution of ilmenite whichwill also cause a change in the Mg of the electronmicroprobe analysis of clinopyroxenes relative to itsprimary composition Also important is the effectof reaction with trapped liquid (Barnes 1986) Suchreaction will produce variable degrees of iron enrich-ment in the finally equilibrated pyroxenes (Lundgaard

et al 2006) All of these processes will variably influencethe analysed mafic mineral composition and hence theanalysed Mg should not be considered as rigorous anindication of evolving melt composition as the An valueof the plagioclase Finally when comparing observedMg with that calculated in the model it needs to beborne in mind that the effect of ferric iron has not beenconsidered The calculated composition for example ofclinopyroxene uses the proportion of ferrous iron onlyIn contrast the electron microprobe analysis includesferric iron as well The difference that this introducescan be demonstrated using a clinopyroxene analysis byAtkins (1969 his analysis 8) from the base of the UZ inwhich ferric iron has been determined The Mg valuecalculated using ferrous iron only gives 727 whereasif total iron is used the figure becomes 707 Hence thecalculated value (below) will always exceed that ofthe determination by electron microprobe

Magma chamber dynamics

The up-section breaks between cycles to higher Mgof pyroxene and olivine higher An of plagioclase andhigher V2O5 of magnetite and the intermittent dis-appearance of olivine and apatite (Figs 4 and 7) require amechanism in addition to the simple closed-systemfractional crystallization model often assumed for thissection of the Bushveld Complex (Wager amp Brown1968 von Gruenewaldt 1973 Molyneux 1974) In adetailed study of the Main and Upper Zones in theBellevue drill core of the northern limb Ashwal et al(2005) explained reversals in An of plagioclase andMg of pyroxene by magma recharge However asdiscussed above the constancy of Sr0 (07071ndash07074Fig 4) throughout MZU and UZ compared with thelarge spread in Sr isotope compositions measured inthe rest of the Bushveld Complex (Sr0 07045ndash07090Kruger 1994) suggests an internal mechanism forgeneration of this layered sequence We also note thatexactly the same Sr0 value was obtained for the MZU

and UZ in the eastern limb (Sharpe 1985) although hisinterpretation was different from ours No isotope dataare available for the northern limb The Sr0 data for theeastern and western limbs imply that the entire magmasheet must have been isotopically homogeneous sub-sequent to the magma mixing event that resulted in theformation of the Pyroxenite Marker (Kruger et al 1987Cawthorn et al 1991) Further the systematic up-sectionincrease in the P2O5 content of apatite-free intervalsand a similar smooth up-section decrease of V2O5 inmagnetite in high-V2O5 intervals (Fig 7) would require adelicate and unlikely balance between the compositionand proportions of residual to added magma if magmaaddition had been the cause In the following discussionwe therefore assume that MZU and UZ crystallized


2272

Dow



from a huge sheet of initially homogeneous magma morethan 2 km thick across the eastern and western andpossibly northern limbs ie over more than 65 000 km2A comparison of observed and modelled Mg of

clinopyroxene and An of plagioclase (Fig 5) indicatesthat cycle I represents about 40 crystallization of theparental magma Cycle I is 330m thick and this modelcalculation therefore suggests that it crystallized froma 800m thick magma sheet If the assumption of a21 km thick magma sheet is correct convection andfractionation in the whole vertical extent of the sheet canbe ruled out We therefore explore the possibility thatcycle I crystallized from only a portion of the stratifiedsheet-like magma chamber The assumed startingsituation with constant composition and density in themagma above the Pyroxenite Marker is shown inFig 9a We base our physical model on that initiallyproposed by Jackson (1961 fig 92) for a verticallyextensive magma chamber namely that crystallizationtook place mainly in the lower part of the chamber Hismodel began with an assumed homogeneous magmachamber which cooled at the top and became moredense but did not crystallize significantly and began toconvect The effect of the adiabatic gradient relative tothe liquidus temperature is that the liquidus is intersectedin the basal part of the chamber In the present case thecrystallization of a gabbronoritic mineral assemblageproduces a residual liquid with raised density whichwould not circulate back to the top but would pond atthe base producing a stable density profile As a result ofslower diffusion of major elements relative to heat sucha magma sheet might separate into double-diffusiveconvective layers (McBirney amp Noyes 1979) as illu-strated in a vertical slice of the Bushveld magma sheet(Fig 9b) This situation is stable as long as the densityof the residual magma increases during crystallizationThis scenario changes dramatically in the upper part ofcycle I (UZa) once magnetite began to crystallize Theforward model predicts that crystallization of magnetitegabbronorite lowers the density of the magma near thebase of the chamber Eventually the density of thisbasal magma layer becomes equal to that of the over-lying layer resulting in mixing as depicted in Fig 9cA consequence of this bottom crystallization is that thebasal magma layer becomes the most compositionallyevolved as illustrated by Mg in Fig 9b The mixingevent therefore produces a somewhat more primitivemagma composition at the crystallization front (Fig 9c)We therefore suggest that the reversal to higher Mg ofclinopyroxene and higher An of plagioclase betweencycles I and II (Fig 4) can be explained by mixingbetween the two lowermost magma layers in thechamber Magnetite gabbronorite now crystallizes tobe joined after a further small degree of fractionation byolivine Crystallization of such iron-rich cumulates drives

+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I

highlow highlowDensity Mg

stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)

Fig 9 Magma chamber model for the lower cycles of MZU and UZshowing schematically the density and composition (Mg) of magmaand cumulates in a vertical slice of the Bushveld magma sheet (a) Thisdiagram shows the starting situation with a thick homogeneousmagma sheet assumed to result from magma recharge and completemixing at the Pyroxenite Marker (b) During crystallization of MZU

gabbronorite at the bottom of the magma chamber the density of theresidual magma increases (and Mg decreases) resulting in a stabledensity profile within the magma sheet that is then likely to break intodouble-diffusive layers The illustration shows the situation at the timethe first magnetite crystallizes (c) Here the crystallization front hasadvanced by crystallizing magnetite gabbronorite (UZa) resulting in adecrease of magma Mg and a decrease in magma density Thediagram illustrates the instant when the density in the lowermost layerequals that of the overlying layer resulting in complete mixing of thetwo layers (d) This diagram illustrates the mixing event resulting in thereversal in mineral compositions between cycles II and III


2273

Dow



the residual magmas to lower density ultimately causingthe bottom two magma layers to mix terminatingcycle II The increase of plagioclase An to 58 in theupper half of cycle II (at 1437m Fig 4) indicates thepresence of at least one further cycle at this levelalthough this cannot be resolved firmly by the presentdataset The mixed magma at the base of cycle III isslightly more primitive than the parent magma to theupper half of cycle II such that it no longer crystallizesolivine but produces magnetite gabbronorite Thedisappearance of olivine is explained by this processalthough it reappears in the middle of cycle III as a resultof fractionation Further crystallization of olivine- andmagnetite-bearing assemblages continues to drive themagma to lower density resulting in periodic magmamixing events as depicted in Fig 9d Cycles IIIndashV areparticularly well-developed in the Bierkraal drill core(Fig 4) In the 300m thick cycle V for example theup-section change in An is from 55 to 46 the Mgcpx

changes from 58 to 26 and olivine changes from Fo52to Fo6 (Fig 4) A forward model calculation of fractionalcrystallization similar to the model presented above (notshown) suggests that cycle V represents about 50crystallization and therefore indicates that the magmalayer undergoing fractionation was 600m thickThe variations in V content of magnetite demonstrate

overall fractionation through the entire Upper ZoneHowever in detail the changes are extremely difficult tomodel The partition coefficient for V between clino-pyroxene and melt ranges from 1 to 3 depending uponf O2 (Toplis amp Corgne 2002) Smaller values areexpected for orthopyroxene Thus in magnetite-freegabbronorite cumulates the bulk partition coefficient willbe less than or close to unity and so small degrees offractionation will have little effect upon the V content ofthe evolving melt However for magnetite the partitioncoefficient ranges from 10 to 40 decreasing withoxidation (Toplis amp Corgne 2002) Hence the presenceof 10 or more of magnetite in the crystallizingassemblage will result in a bulk partition coefficientsignificantly greater than unity leading to a decrease inV in the melt and in subsequent magnetite Howeverquantitative modeling of the trend seen in Fig 7 is notpossible because the value of f O2 is not known andmore importantly we do not know whether it remainsconstant or responds to periods of excessive magnetiteformation (as seen in the thick layers) For example if thef O2 of the melt is reduced by removal of excessivemagnetite the partition coefficient into magnetitewould increase and the abundance of V in subsequentmagnetite might increase even though the abundanceof V in the melt might be decreasing Because ofsuch variation in the partition coefficient as a result ofchanges in f O2 it is not permissible to conclude that

an increase in V in magnetite indicates addition of lessevolved magmaThe high-field strength element phosphorus is parti-

cularly useful as a tracer of crystallization processes inmafic cumulates It is almost perfectly excluded fromsilicate minerals but is a major component of apatiteWhen apatite becomes a liquidus phase at about 1P2O5 in a basic melt (Green amp Watson 1982) thecumulate rock contains more P than the melt which istherefore gradually depleted in P (Wager 1960) BecauseP is an essential component of apatite the depletionof P in the magma results in a decrease in the amountof apatite that can crystallize at the cotectic This isdemonstrated for example in the apatite mode andP2O5 variations of the Skaergaard intrusion (Wager1963)Apatite is locally abundant in cycles IVndashIX (UZc) and

occurs intermittently throughout the stratigraphicsection (Fig 7) At the bases of cycles V and VI whichare defined by breaks in mineral compositions (Figs 4and 7b) there are 60ndash70m thick low-P2O5 intervalswhere cumulus apatite is absent (Fig 7a) Apatite mayhave been removed from the liquidus by mixing withapatite-undersaturated magma The return of abundantliquidus apatite 60ndash70m above the bases of these cycles(Fig 7a) is the result of simple fractional crystallizationdriving the magma back to apatite saturation In theupper half of the UZ we interpret low-P2O5 intervalslacking cumulus apatite as evidence for magma mixingevents (Fig 7a) We have therefore placed cycleboundaries at the levels where cumulus apatite dis-appears ie where the P2O5 content abruptly decreasesbelow 1 wt In these uppermost cycles the intervalslacking apatite become thinner upwards indicating thatapatite was only briefly removed from the liquidus Thiscould explain the lack of significant breaks in An andMgcpx at these levels (Fig 4) although it is possiblethat closer sampling could reveal subtle changes in Anand Mg In cycle IV where cumulus apatite firstappears and defines the base of UZc (Fig 4) there aretwo short high-P2O5 intervals interbedded with thelow-P2O5 succession (Fig 7a) These high-P2O5 intervalsoccur in rocks enriched in magnetite and olivine Thisassociation of apatite could possibly relate to physicalsorting of dense cumulus minerals or fluctuations in thesolubility of P in the magma (Green amp Watson 1982) buta firm explanation must await a detailed study of closelyspaced samples The smooth up-section increase in P2O5

in the low-P2O5 intervals (Fig 7a) and the smoothdecrease in V2O5 in the high-V2O5 intervals (Fig 7b)support a model involving crystallization and mixingwithin a stratified magma sheet It would appearfortuitous if recharge from an extraneous magma sourceproduced such systematic trends


2274

Dow



Formation of magnetitite andnelsonite layers

The magnetitite and nelsonite layers of the UZ containimportant orthomagmatic deposits of vanadium tita-nium and phosphorus (Lee 1996 von Gruenewaldt1993 Cawthorn et al 2005) The formation of themassive magnetitite layers (Fig 3f) has been variablyexplained as follows(1) Bateman (1951) suggested that an iron-rich

immiscible liquid might have separated and producedmagnetite-rich layers This concept has been applied tothe Bushveld Complex by Reynolds (1985a) and vonGruenewaldt (1993) We would argue that a very denselow-viscosity immiscible iron-rich liquid might beexpected to percolate downwards through the under-lying plagioclase-rich crystal mush and not produce theremarkably planar bases commonly observed (Fig 3f)Further immiscible iron-rich liquids contain only about30 total FeO (Jakobsen et al 2005) and so a furtherprocess is required to produce the near-monomineralicmagnetitite layers from such a liquid (Cawthorn et al2005)(2) The sinking and sorting of dense magnetite grains

was proposed by Wager amp Brown (1968) However theynoted that there is a significant inconsistency namelythat pyroxenes are absent in magnetite-anorthositesequences (eg Fig 3f) Had the magma simply evolvedto magnetite saturation a gravity-controlled successionought to include pyroxene between the magnetite andanorthosite layers Some additional mechanism wouldappear to be required to cause pyroxene to ceasecrystallization(3) The possibility of the formation of chromitite layers

as a result of magma addition and mixing (Irvine 1975)has been extrapolated to the formation of magnetititelayers (Harney et al 1990) although appropriate phasediagrams have not been presented Ashwal et al (2005)documented reversals in Mg of pyroxene and An ofplagioclase in a drill core through the MZU and UZ inthe northern limb which they interpreted as evidencefor magma recharge As discussed above we disagreewith this interpretation for the Bierkraal drill corethrough the western limb(4) Increase in f O2 in the magma by fluids derived

from the country rocks has been proposed as a way toinduce magnetite saturation (Klemm et al 1985 vonGruenewaldt et al 1985) The source and mechanism ofaddition of such fluid remains to be demonstratedFurthermore the lateral continuity of the magnetititelayers requires a process that can operate simultaneouslythroughout the entire magma chamber(5) By analogy with the formation of chromitite layers

as a result of pressure increase (Lipin 1993) an increasein pressure exerted on magma can induce magnetite

saturation (Cawthorn amp McCarthy 1980) Physicalprocesses causing such pressure changes need to beestablished but have been proposed by Lipin (1993) andCarr et al (1994)(6) In contrast to the crystal-settling concept

Cawthorn amp McCarthy (1980) used the cyclicity in Crcontents across massive magnetitite layers as evidencefor crystallization at the base of the magma chamberReversals in Cr content of magnetite resulted fromdiffusion from the overlying magma A slightly differentinterpretation for these data was given by Kruger ampSmart (1987) who suggested that crystallizationoccurred within a basal layer of magma that underwentperiodic mixing with overlying layers within a stratifiedchamber Similarly Harney et al (1996) interpretedchanges in SrAl2O3 of plagioclase separates takenacross the Main Magnetitite Layer as evidence of mixingas a result of the breakdown of stratified magma layerscaused by density inversion during the crystallization ofmagnetitite layersThe positions of the 32 magnetitite and nelsonite

layers (Table 2 Fig 4) relative to the cycles describedabove throw new light on their petrogenesis The basalpackage of magnetitite layers includes eight layers thatrange from 26 to 246 cm in thickness totalling 731 cmand are distributed between 1839 and 1719m in thestratigraphy (Table 2 Fig 4) Of these the 246 cm thickMain Magnetitite Layer is the lowest at 1839m Thisoccurs some 50ndash100m below the interval (1781ndash1739m)displaying a reversal in An of plagioclase betweencycles I and II (Fig 4 Table 3) Two other magnetititelayers also occur below the Reversal in An Few layersoccur within the reversal and the uppermost magnetititelayer in this package is located 20m above the top of thereversal (Tables 2 and 3) Based on much more detailedsample spacing it was shown by Fox (1982) and Harneyet al (1996) that the Main Magnetitite Layer at MagnetHeights in the eastern limb likewise is not associatedwith a reversal in An (Fig 6) In the upper part ofcycle II there is a package of three magnetitite layersbetween 1441 and 1427m totalling 113 cm in thickness(Table 2) These layers are not related to a significantreversal in mineral compositions (Fig 4) In cycle IIIan 80 cm thick magnetitite layer occurs in a sectiondisplaying rapidly declining An (Fig 4) Between1020 and 945m there is another package including12 magnetitite layers totalling 1045 cm in thickness(Table 2 Fig 4) The reversal interval between cycles IVand V is placed at 998ndash981m and shows pronouncedchanges in plagioclase clinopyroxene olivine andmagnetite compositions (Figs 4 and 7b) Similar to themagnetitite layers straddling the boundary betweencycles I and II these magnetitite layers occur belowwithin and above the regressive interval between


2275

Dow



cycles IV and V It is noteworthy that this package ofmagnetitite layers coincides with a low-P2O5 interval(Fig 7a) Of the uppermost eight oxide-rich layers sixare nelsonites and they all occur within cycles and not attheir boundaries (Fig 4) We therefore conclude thatsimple fractional crystallization led to the formation ofthe magnetitite layersThe broad coincidence of the most significant packages

of magnetitite layers with the boundaries between cyclesIndashII and IVndashV however suggests a relationship In bothcases thick magnetitite layers occur just below thereversals in mineral compositions and formed as a resultof normal fractional crystallization The crystallization ofsuch thick magnetitite layers must have lowered thedensity of the residual magma dramatically We there-fore suggest that the formation of the basal magnetititelayers in these two packages accelerated the trend ofdensity decrease and initiated magma mixing and theformation of a mineralogical reversal We furtherspeculate that the occurrence of several closely spacedmagnetitite layers in these reversal intervals resulted frommagma inhomogeneity during a single mixing eventperhaps a result of finger instability during mixingbetween two magma layers (Irvine et al 1983) In severalcases magnetitite and nelsonite layers occur within cycles(Fig 4) and are apparently not related to reversals Mostof these magnetitite layers are relatively thin and single(Table 2) We explain these magnetitite layers as theresult of normal fractional crystallizationApatite becomes a liquidus phase and coprecipitates

with magnetite in the nelsonite layers in cycles VndashIX(Figs 3e and 4) suggesting that they formed as aconsequence of fractional crystallization FendashPndashTi-richimmiscible liquids exsolved during late-stage fractionalcrystallization in the Skaergaard intrusion (Jakobsenet al 2005) and by analogy this implies that immisci-bility is also possible in the evolved stages of theBushveld Complex However we prefer to envisage asingle mechanism that produced all the magnetite-richlayers rather than appealing to liquid immiscibility forthe upper apatite-bearing layers [as suggested byReynolds (1985a) and von Gruenewaldt (1993)] and adifferent process for the lower apatite-free layers

Comparison with eastern and northernlimbs of the Bushveld Complex

The studied sections of UZ are broadly similar in thethree major limbs (von Gruenewaldt 1973 Molyneux1974 Ashwal et al 2005 this study) The thicknessesvary from 1510 to 2230m (Table 1) and they aresubdivided into three subzones defined by the firstappearance of cumulus magnetite olivine and apatiterespectively Above we have shown for the Bierkraalsection however that the presence of cumulus olivine

and apatite is intermittent (Figs 4c and 7a) and that thishas important petrogenetic implications The reportedintervals of occurrence of olivine in UZb thorn c and apatitein UZc in the eastern (von Gruenewaldt 1973) andnorthern limbs (Ashwal et al 2005) are thereforesummarized in Fig 10 This figure shows that olivineand apatite are intermittent in all three limbs Thepresence of cumulus olivine in the Bellevue core of thenorthern limb broadly corresponds to intervals showingnormal fractionation trends (Ashwal et al 2005)Furthermore the two main intervals displaying areversal in An of plagioclase and Mg of the maficsilicates in UZ occur at 720ndash640m and 420ndash330mdepth in the Bellevue core and coincide with gaps in thepresence of cumulus olivine (Ashwal et al 2005) Theinterval in UZc lacking in apatite in the Bellevue core(430ndash300m depth Ashwal et al 2005) overlaps with theinterval where olivine is absent Many of the mineralcompositional data presented above are comparablewith similar results obtained for the other limbs ofthe Bushveld Complex In the eastern limb vonGruenewaldt (1973) and Molyneux (1974) providedmineral compositional data for the Main and UpperZones although some of their data were obtained byoptical and X-ray diffraction methods rather than byelectron microprobe Also no compositional data areavailable for the clinopyroxene Ashwal et al (2005)provided electron microprobe data for a very largenumber of samples through part of the Main Zone andthe entire Upper Zone for the northern limb In generalthe order of appearance of minerals is very similar in

00

02

04

06

08

10

(a) Olivine in UZb+c (b) Apatite in UZc

west east north west east north

Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic

Fig 10 Distribution of (a) cumulus olivine in UZb thorn c and (b) apatitein UZc in the northern (Bellevue core Ashwal et al 2005) eastern(field relations von Gruenewaldt 1973) and western (this study) limbsof the Bushveld Complex


2276

Dow



all limbs although two differences exist in the northernlimb compared with the eastern and western Anorthopyroxene-rich layer with reversal in composition(ie the Pyroxenite Marker) has not been identified inthe northern limb Also Ashwal et al (2005) suggestedthat apatite appears 100m lower in the successionthan olivine but it is transitory and does not reappearfor a further 400m In all limbs considerable oscillationof mineral compositions occurs superimposed on theoverall fractionation trends The actual mineral compo-sitions at which the different phases appear are slightlydifferent although detailed comparison is hampered bythe different analytical methods used the considerablesmall-scale vertical variation in composition andirregular spacing of the data pointsThe magma chamber model developed above (Fig 9)

therefore provides a possible explanation for thecyclicity observed in all the limbs The 2ndash3m thickMain Magnetitite Layer and the 7ndash13m thick layer(called Layer 21 in the eastern limb) are unequivocallyidentifiable in the three limbs Ashwal et al (2005)identified 32 magnetitite layers in the northern limbwhich can possibly be grouped into four packages In theeastern limb only 26 layers are reported (Willemse1969b Molyneux 1974) again in four intervals butthis information is based on field observations acontinuous borehole may reveal additional magnetititelayers The grouping of the 32 layers identified in thisstudy (Fig 4 Table 2) is less definitive Further we notethe absence of layers below the Main MagnetititeLayer in our study contrasting with three and twolayers in the eastern and northern limbs respectivelyThere are also a greater number of layers (eleven) in thisstudy above the very distinctive 7m thick layer 21(Table 2) compared with the northern (three) and eastern(none) limbs We suggest that considerable similaritiesexist in the magnetitite layers between all limbs butperfect correlation is not possible Hence the number ofcycles and the proposed mechanism by which they aregenerated may not be an instantaneous chamber-wideprocess but merely an inevitable consequences ofcrystallization of magnetite at the base of a stratifiedmagma sheet

CONCLUSIONS

New mineral chemical data for plagioclase pyroxeneolivine and magnetite and whole-rock P2O5 andSr isotope data for the upper Main Zone and UpperZone in the Bierkraal drill core of the western BushveldComplex show the following features(1) Existing and new initial 87Sr86Sr values are near-

constant (07073 plusmn 00001 n frac14 22) suggesting crystal-lization from a homogeneous magma sheet withoutmajor magma recharge or assimilation

(2) The mafic rocks evolve up-section from gabbro-norite (plagioclase An72 clinopyroxene Mg 74) atthe Pyroxenite Marker to magnetitendashilmenitendashapatitendashfayalite ferrodiorite (An43 Mgcpx 5 Fo1 olivine) at theroof of the mafic complex(3) The overall fractionation trend is however

interrupted by reversals to higher An of plagioclaseMg of pyroxene and olivine V2O5 in magnetiteandor intermittent absence of cumulus apatite orolivine These reversals define at least nine majorfractionation cycles that range from 100 to 400m inthickness(4) We have estimated a plausible magma composition

that existed in the chamber at the level of the PyroxeniteMarker based on summation of rock compositionsabove that level plus the addition of postulated expelledmagma during final crystallization It has the composi-tion of a slightly quartz-normative iron-rich tholeiite(5) Forward modeling of fractional crystallization

using this composition predicts increasing FeO (total)near-constant SiO2 and increasing density of the residualmagma before magnetite crystallizes and increasingSiO2 near-constant FeO and decreasing magma densityafter magnetite crystallizes When olivine reappears as acumulus phase the FeO content and density of meltdecrease(6) We explain the observed cyclicity by crystallization

at the floor of a huge stratified magma sheet morethan 2 km thick covering at least 65 000 km2 Magmastratification with a stable density profile initiallydeveloped during crystallization of gabbronorites in theupper Main Zone from a basal layer of magma Oncemagnetite began to crystallize the magma densitydecreased and periodic density inversion led to mixingwith the overlying magma layer producing mineralogi-cal breaks between fractionation cycles(7) The investigated section includes 26 magnetitite

and six nelsonite (magnetitendashilmenitendashapatite) layers thatmainly occur within fractionation cycles In at least twocases crystallization of thick magnetitite layers may havelowered the magma density sufficiently to trigger densityinversion resulting in near-coincidence of mineralogicalbreaks and packages of magnetitite layers

ACKNOWLEDGEMENTSThis research was supported by grants from the DanishNatural Science Research Council and the NationalResearch Foundation of South Africa and the Anglo-plats Implats and Lonplats mining companies TheCouncil for Geosciences South Africa is thanked foraccess to the Bierkraal core material and permissionto publish We thank Richard Wilson Jean-ClairDuchesne Tony Morse Chris Harris and MarjorieWilson for critical comments and discussions


2277

Dow



SUPPLEMENTARY DATA

Supplementary data for this paper are available atJournal of Petrology online

REFERENCES

Ariskin A A Frenkel M Y Barmina G S amp Nielsen R L (1993)

Comagmat a Fortran program to model magma differentiation

processes Computers amp Geosciences 19 1155ndash1170

Ashwal L D Webb S J amp Knoper M W (2005) Magmatic

stratigraphy in the Bushveld northern lobe continuous geophysical

and mineralogical data from the 2950m Bellevue drillcore South

African Journal of Geology 108 199ndash232

Atkins F B (1969) Pyroxenes of the Bushveld Intrusion South Africa

Journal of Petrology 10 222ndash249

Barnes S J (1986) The effect of trapped liquid crystallization on

cumulus mineral compositions in layered intrusions Contributions to

Mineralogy and Petrology 93 524ndash531

Bateman A M (1951) The formation of late magmatic oxide ores

Economic Geology 46 404ndash426

Buick I S Maas R amp Gibson R (2001) Precise UndashPb titanite

age constraints on the emplacement of the Bushveld Complex

South Africa Journal of the Geological Society London 158 3ndash6

Carr H W Groves D I amp Cawthorn R G (1994) The importance

of synmagmatic deformation in the formation of Merensky Reef

potholes in the Bushveld Complex Economic Geology 89 1398ndash1410

Cawthorn R G amp McCarthy T S (1980) Variations in Cr content

of magnetite from the Upper Zone of the Bushveld Complexmdash

evidence for heterogeneity and convection currents in magma

chambers Earth and Planetary Science Letters 46 335ndash343

Cawthorn R G amp McCarthy T S (1985) Incompatible trace

element behavior in the Bushveld Complex Economic Geology 80

1016ndash1026

Cawthorn R G amp Molyneux T G (1986) Vanadiferous magnetite

deposits of the Bushveld Complex In Anhaeusser C R amp

Maske S (eds) Mineral Deposits of South Africa Johannesburg

Geological Society of South Africa pp 1251ndash1266

Cawthorn R G amp Walraven F (1998) Emplacement and

crystallization time for the Bushveld Complex Journal of Petrology

39 1669ndash1687

Cawthorn R G amp Walsh K L (1988) The use of phosphorus

contents in yielding estimates of the proportion of trapped liquid in

cumulates of the Upper Zone of the Bushveld Complex Mineralogical

Magazine 52 81ndash89

Cawthorn R G amp Webb S J (2001) Connectivity between western

and eastern limbs of the Bushveld Complex Tectonophysics 330

195ndash209

Cawthorn R G Meyer P S amp Kruger F J (1991) Major addition

of magma at the Pyroxenite Marker in the western Bushveld

Complex South Africa Journal of Petrology 32 739ndash763

Cawthorn R G Barnes S J Ballhaus C amp Malitch K N (2005)

Platinum-group element chromium and vanadium deposits in

mafic and ultramafic rocks Economic Geology 100th Anniversary Volume

pp 215ndash249

Coffin M F amp Eldholm O (1994) Large igneous provinces crustal

structure dimensions and external consequences Reviews of

Geophysics 32 1ndash36

Eales H V amp Cawthorn R G (1996) The Bushveld Complex

In Cawthorn R G (ed) Layered Intrusions Amsterdam Elsevier

pp 181ndash230

Eales H V de Klerk W J Butcher A R amp Kruger F J (1990)

The cyclic unit beneath the UG1 chromitite (UG1FW unit) at RPM

Union Section Platinum MinemdashRosetta Stone of the Bushveld

Upper Critical Zone Mineralogical Magazine 54 23ndash43

Fox N (1982) Variation in plagioclase compositions across magnetitite

layers in the eastern Bushveld Complex University of Cape Town

Honours thesis

Ghiorso M S amp Sack R O (1995) Chemical mass transfer in

magmatic processes IV A revised and internally consistent

thermodynamic model for the interpolation and extrapolation of

liquidndashsolid equilibria in magmatic systems at elevated tem-

peratures and pressures Contributions to Mineralogy and Petrology 119

197ndash212

Green T H amp Watson E B (1982) Crystallization of apatite in

natural magmas under high pressure hydrous conditions with

particular reference to lsquoorogenicrsquo rock series Contributions to Mineralogy

and Petrology 79 96ndash105

Harney D M W Merkle R K W amp von Gruenewaldt G (1990)

Platinum-group element behavior in the lower part of the Upper

Zone Eastern Bushveld Complexmdashimplications for the formation of

the main magnetite layer Economic Geology 85 1777ndash1789

Harney D M W von Gruenewaldt G amp Merkle R K W (1996)

The use of plagioclase composition as an indicator of magmatic

processes in the Upper Zone of the Bushveld Complex Mineralogy


Hunter R H amp Sparks R S J (1987) The differentiation of the

Skaergaard Intrusion Contributions to Mineralogy and Petrology 95

451ndash461

Irvine T N (1975) Crystallization sequences in the Muskox intrusion

and other layered intrusions 2 Origin of chromitite layers and

similar deposits of other magmatic ores Geochimica et Cosmochimica

Acta 39 991ndash1008

Irvine T N Keith D W amp Todd S G (1983) The J-M Platinumndash

Palladium Reef of the Stillwater Complex Montana II Origin by

double-diffusive convective magma mixing and implications for the

Bushveld Complex Economic Geology 78 1287ndash1334

Jackson E D (1961) Primary Tjextures and Mineral Associations in the

Ultramafic Zone of the Stillwater Complex Montana US Geological Survey

Professional Papers 358

Jakobsen J K Veksler I V Tegner C amp Brooks C K (2005)

Immiscible iron- and silica-rich melts in basalt petrogenesis

documented in the Skaergaard intrusion Geology 33 885ndash888

Klemm D D Henckel J Dehm R amp von Gruenewaldt G (1985)

The geochemistry of titanomagnetite in magnetite layers and their

host rocks of the Eastern Bushveld Complex Economic Geology 80

1075ndash1088

Kretz R (1963) Distribution of magnesium and iron between

orthopyroxene and calcic pyroxene in natural mineral assemblages

Journal of Geology 71 773ndash785

Kruger F J (1994) The Sr-isotopic stratigraphy of the western

Bushveld Complex South African Journal of Geology 97 393ndash398

Kruger F J (2005) Filling the Bushveld Complex magma

chamber lateral expansion roof and floor interaction magmatic

unconformities and the formation of giant chromitite PGE and

T-V-magnetitite deposits Mineralium Deposita 40 451ndash472

Kruger F J amp Smart R (1987) Diffusion of trace elements during

bottom crystallization of double-diffusive convection systems the

magnetitite layers of the Bushveld Complex Journal of Volcanology and

Geothermal Research 34 133ndash142

Kruger F J Cawthorn R G amp Walsh K L (1987) Strontium

isotopic evidence against magma addition in the Upper Zone of the

Bushveld Complex Earth and Planetary Science Letters 84 51ndash58

Lee C A (1996) A review of mineralizations in the Bushveld Complex

and some other layered intrusions In Cawthorn R G (ed) Layered

Intrusions Amsterdam Elsevier pp 103ndash145


2278

Dow



Lipin B R (1993) Pressure increase the formation of chromitite

layers and the development of the Ultramafic Series in the Stillwater

Complex Journal of Petrology 34 955ndash976

Lundgaard K L Tegner C Cawthorn R G Kruger F J amp

Wilson J R (1993) Trapped intercumulus liquid in the Main Zone

of the eastern Bushveld Complex South Africa Contributions to


McBirney A R (1993) Igneous Petrology 2nd edn Boston MA Jones amp

Bartlett pp 508

McBirney A R amp Noyes M N (1979) Crystallization and layering of

the Skaergaard intrusion Journal of Petrology 20 487ndash554

Merkle R K W amp von Gruenewaldt G (1986) Compositional

variation of Co-rich pentlandite relation to the evolution of the

Upper Zone of the western Bushveld Complex South Africa

Canadian Mineralogist 24 529ndash546

Mitchell A A Eales H V amp Kruger F J (1998) Magma

replenishment and the significance of poikilitic textures in the

Lower Main Zone of the western Bushveld Complex South Africa

Mineralogical Magazine 62 435ndash450

Molyneux T G (1972) X-ray data and chemical analyses of some

titanomagnetite and ilmenite samples from the Bushveld Complex

South Africa Mineralogical Magazine 48 863ndash871

Molyneux T G (1974) A geological investigation of the Bushveld

Complex in Sekhukhuneland and part of the Steelpoort valley

Transactions of the Geological Society of South Africa 77 329ndash338

Morse S A (1984) Cation diffusion in plagioclase feldspar Science 225

504ndash505

Nex P A Kinnaird J A Ingle L J Van der Vyver B A amp

Cawthorn R G (1998) A new stratigraphy for the Main Zone of

the Bushveld Complex in the Rustenburg area South African Journal

of Geology 101 215ndash223

Reynolds I M (1985a) The nature and origin of titaniferous

magnetite-rich layers in the Upper Zone of the Bushveld Complex

a review and synthesis Economic Geology 80 1089ndash1108

Reynolds I M (1985b) Contrasted mineralogy and textural

relationships in the uppermost titaniferous magnetite layers of the

Bushveld Complex in the Bierkraal area north of Rustenburg


Sharpe M R (1985) Strontium isotope evidence for preserved density

stratification in the Main Zone of the Bushveld Complex Nature 316

119ndash126

Spulber S D amp Rutherford M J (1983) The origin of rhyolite and

plagiogranite in oceanic crust an experimental study Journal of

Petrology 24 1ndash25

Tegner C (1997) Iron in plagioclase as a monitor of the differentiation

of the Skaergaard intrusion Contributions to Mineralogy and Petrology

128 45ndash51

Tegner C Robins B Reginiussen H amp Grundvig S (1999)

Assimilation of crustal xenoliths in a basaltic magma chamber Sr

and Nd isotopic constraints from the Hasvik Layered Intrusion

Norway Journal of Petrology 40 363ndash380

Toplis M J (2005) The thermodynamics of iron and magnesium

partitioning between olivine and liquid criteria for assessing and

predicting equilibrium in natural and experimental systems

Contributions to Mineralogy and Petrology 149 22ndash39

Toplis M J amp Carroll M R (1995) An experimental study of the

influence of oxygen fugacity on FendashTi oxide stability phase relations

and mineralndashmelt equilibria in ferro-basaltic systems Journal of


Toplis M J amp Corgne A (2002) An experimental study of element

partitioning between magnetite clinopyroxene and iron-bearing

silicate liquids with particular emphasis on vanadium Contributions to


Toplis M J Libourel G amp Carroll M R (1994) The role of

phosphorus in crystallization processes of basalt an experimental

study Geochimica et Cosmochimica Acta 58 797ndash810

Vander Auwera J amp Longhi J (1994) Experimental study of a

jotunite (hypersthene monzodiorite) constraints on the parent

magma composition and crystallization conditions (P T f O2) of

the BjerkreimndashSokndal layered intrusion (Norway) Contributions to


von Gruenewaldt G (1970) On the phase change orthopyroxenendash

pigeonite and the resulting textures in the Main and Upper Zones of

the Bushveld Complex in the eastern Transvaal In Visser D J L

amp von Gruenewaldt G (eds) Symposium on the Bushveld Igneous Complex

and Other Layered Intrusions Johannesburg Geological Society of

South Africa pp 67ndash73

von Gruenewaldt G (1971) A petrological and mineralogical

investigation of the rocks of the Bushveld Igneous Complex in the

TauteshoogtendashRoossenekal area of the eastern Transvaal University

of Pretoria DSc thesis

von Gruenewaldt G (1973) The Main and Upper zones of the

Bushveld Complex in the Roossenekal area Eastern Transval


von Gruenewaldt G (1993) Ilmenitendashapatite enrichments in the

Upper Zone of the Bushveld Complex a major titanium-rock

phosphate resource International Geology Review 35 987ndash1000

von Gruenewaldt G Klemm D D Henckel J amp Dehm R M

(1985) Exsolution features in titanomagnetites from massive

magnetitite layers and their host rocks of the Upper Zone eastern


Wager L R (1960) The major element variation of the layered series

of the Skaergaard intrusion and a re-estimation of the average

composition of the hidden series and of successive residual magmas


Wager L R (1963) The mechanism of adcumulus growth in the

layered series of the Skaergaard intrusion In Fisher D J

Frueh A J Hurlbert C S amp Tilley C E (eds) Symposium on

Layered Intrusions Mineralogical Society of America Special Paper 1 1ndash9

Wager L R amp Brown G M (1968) Layered Igneous Rocks London

Oliver amp Boyd pp 572

Walraven F (1987) Textural Geochemical and Genetic Aspects of the

Granophyric Rocks of the Bushveld Complex Memoirs of the Geological Survey of

South Africa 72 145 pp

Walraven F amp Wolmarans L G (1979) Stratigraphy of the upper

part of the Rustenburg Layered Suite Bushveld Complex in the

western Transvaal Annals of the Geological Survey of South Africa 13

109ndash114

Willemse J (1969a) The geology of the Bushveld Igneous Complex

the largest repository of magmatic ore deposits in the world Economic

Geology Monograph 4 1ndash22

Willemse J (1969b) The vanadiferous magnetic iron ore of the

Bushveld Igneous Complex Economic Geology Monograph 4

187ndash208


2279

Dow



of the sheet-like magma chamber (Kruger 2005)Evidence for mixing at this level between residual andrecharged magma comes from protracted reversals inMg and An contents of pyroxene and plagioclase andchanges in initial 87Sr86Sr value Within this intervala distinct thin layer of orthopyroxenite occurs knownas the Pyroxenite Marker which is present in both theeastern (von Gruenewaldt 1970) and western limbs(Cawthorn et al 1991) The initial 87Sr86Sr composi-tion of the 21 km thick cumulate sequence above thePyroxenite Marker which comprises the upper MainZone (MZU) and the Upper Zone (UZ) (Fig 2) isconstant with an average of 07073 plusmn 00001 (2 standarderror SE) and significantly different from the under-lying 42 km of cumulates (Kruger et al 1987 Kruger1994) This has been explained by complete homo-genization between residual and added magma abovethe Pyroxenite Marker (Kruger et al 1987 Cawthornet al 1991) With an estimated volume of 140 000 km3the MZU and UZ represent the largest known sheet ofbasaltic magma emplaced into the Earthrsquos crustThis contribution aims to decipher the physical

processes of crystallization within the huge MZU andUZ magma sheet The UZ includes about 30 distinctmagnetitite and nelsonite (magnetitendashilmenitendashapatite)layers which vary from 2 to 710 cm thick and hostsworld-class deposits of V Ti and P (Cawthorn ampMolyneux 1986 von Gruenewaldt 1993 Lee 1996Cawthorn et al 2005) From bottom to top MZU andUZ evolve from gabbronorite (Mg of clinopyroxeneis 74) to iron-rich apatitendashmagnetitendashfayalite ferrodiorite(Mg of clinopyroxene is lt5) this sequence has tradi-tionally been interpreted as the result of closed-systemcrystallization without magma recharge (Wager amp Brown1968 Willemse 1969a 1969b von Gruenewaldt1973 Molyneux 1974) Several excursions from simple

100 km

26degS

25degS

24degS

28degE 29degE 30degE27degE26degE

Mafic Bushveld

Transvaal Supergroup

Felsic BushveldN

Pretoria

Lydenburg

BK1

25degS

30degS

Johannesburg

Rustenburg

BierkraalBK3

BK2

FarWestern

Limb

SouthAfrica

30degS

WesternLimb

NorthernPotgietersrus

Limb EasternLimb

SouthernBethalLimb

Fig 1 Map of the Bushveld Complex showing the location of the Bierkraal drill holes BK1 BK2 and BK3 Modified after Lundgaard et al(2006)

(1128 m)

(1662 m)

LowerZone

CriticalZone

MainZone

Upper

Zone

0

1000

2000

3000

4000

5000

60000

500

1000

MM

L

1500

2000

2500

Stra

tigra

phic

pos

ition

(m

etre

s)

Mag

netit

ite L

ayer

sN

elso

nite

Lay

ers

BushveldComplex

BierkraalDrill Core

MZL

MZU(2125 m)

UZb

UZc BK1

BK3

BK2PyroxeniteMarker

UZa(1862 m)

PyroxeniteMarker

Fig 2 Generalized stratigraphic section of the Bushveld Complex(left) and the combined Bierkraal drill cores (right) The bases ofsubzones in the Bierkraal cores delineate the lowest appearance ofcumulus magnetite (UZa) olivine (UZb) and apatite (UZc) respec-tively The correlations between the three Bierkraal drill cores are1600m depth in BK1 equals 550m depth in BK3 1420m depth inBK3 equals 200m depth in BK2 [see core logs given by Kruger et al(1987)] An inferred stratigraphic position in the Bierkraal cores iscalculated assuming the core is vertical that layering dips 24 to thenorth and that the roof contact is at 415m depth in BK1 Previouspublications on BK drill cores quote only absolute depths belowsurface To facilitate comparison with our inferred stratigraphicpositions the following conversion equations have been used BK1inferred stratigraphic position in metres frac14 (depth in BK1 ndash 415) middot cos24 BK2 inferred stratigraphic position in metres frac14 (depth in BK2 thorn1840) middot cos 24 BK3 inferred stratigraphic position in metres frac14 (depthin BK3 thorn 635) middot cos 24 MML Main Magnetitite Layer Modifiedafter Kruger et al (1987)


2258

Dow







Zonal subdivision





2259

Dow




Petrography











Upper Zone c (UZc) 1128 910 350 610

Upper Zone b (UZb) 534 740 520 390

Upper Zone a (UZa) 200 580 640 590





2260

Dow








2261

Dow




Analytical methods






Complex


1w784y UZc 3371 2

1w823y UZc 3727 10

1w885y UZc 4293 6

1w1099y UZc 6248 10

1w11126y UZc 6373 30

1w1117 UZc 6413 6

1w1206y UZc 7226 6

1w12841 UZc 7939 3

1w14498 UZc 9453 20

1w14505 UZc 9459 10

1w14513 UZc 9467 70

1w1460 UZc 9540 710 (Layer 21)

1w1465 UZc 9592 68

1w1485 UZc 9774 25

1w14882 UZc 9804 40

1w14892 UZc 9813 3

1w14921 UZc 9839 5

1w149265 UZc 9844 20

1w1494 UZc 9857 60

1w1532 UZc 10204 14

3w746 UZb 12615 80

3w927 UZb 14269 43

3w932 UZb 14315 60

3w942 UZb 14406 10

3w1247 UZa 17192 26

3w1272 UZa 17420 134

3w1294 UZa 17621 107

3w1313 UZa 17795 64

3w1315 UZa 17813 26

3w1343 UZa 18069 53

3w1368 UZa 18297 75

3w1378 UZa 18389 246 (MML)




2262

Dow






1w4222 66 UZc IX 430 (23) 9 114 (07) 6 30 (02) 6

1w4318 154 UZc IX 452 (24) 9

1w4466 289 UZc IX 428 (14) 9 49 (04) 7 13 (01) 9

1w4751 549 UZc IX 434 (24) 8

1w5047 819 UZc IX 451 (18) 9 159 (03) 6 39 (03) 9

1w534 1087 UZc IX 446 (23) 8 308 (03) 4 94 (03) 9

1w568 1398 UZc IX 500 (19) 9 364 (07) 4 104 (09) 9

1w598 1672 UZc IX 463 (06) 8 268 (07) 7 78 (03) 9

1w62585 1926 UZc IX 498 (20) 8 502 (05) 3 214 (01) 9 369 (08) 6

1w6633 2268 UZc IX 487 (16) 9

1w732 2896 UZc VIII 519 (08) 9 436 (10) 5 220 (02) 8

1w770 3243 UZc VIII 493 (04) 9 469 (10) 4 216 (03) 9

1w8086 3596 UZc VII 505 (14) 8

1w8532 4003 UZc VII 498 (05) 8 462 (05) 6 187 (04) 9

1w9086 4509 UZc VII 495 (11) 8 133 (01) 4

1w954 4924 UZc VI 498 (08) 8 357 (05) 6 148 (05) 9

1w10102 5437 UZc VI 489 (13) 7

1w10505 5805 UZc VI 501 (05) 8 362 (08) 5 159 (01) 9

1w11382 6605 UZc VI 501 (09) 9 430 (04) 6 175 (04) 6 070711 (13)

1w11586 6793 UZc VI 481 (12) 6 255 (10) 5 62 (02) 9 070739 (14)

1w11901 7085 UZc V 456 (10) 7 325 (05) 6 146 (02) 9 070720 (13)

1w12395 7528 UZc V 483 (06) 9 383 (08) 5 160 (03) 8 070717 (15)

1w13032 8112 UZc V 070726 (13)

1w13292 8351 UZc V 509 (07) 8 473 (04) 4 232 (02) 9 070714 (13)

1w14234 9212 UZc V 520 (04) 9

1w148885 9810 UZc V 550 (07) 9 578 1 449 (06) 15 070758 (16)

1w15070 9975 UZc IV 492 (08) 7 475 (13) 4 341 (07) 8

1w15507 10375 UZc IV 485 (09) 9 566 (07) 5 340 (02) 6 070726 (13)

3w540 10734 UZb IV 498 (09) 8 375 (07) 9

3w590 11190 UZb IV 535 (13) 7

3w6409 11655 UZb IV 559 (10) 9 617 (05) 6 504 (12) 6

3w715 12332 UZb III 521 (15) 9 643 (11) 8 389 (14) 9

3w742 12579 UZb III 554 (08) 9

3w780 12926 UZb III 577 (18) 9 596 (08) 9

3w810 13200 UZb III 582 (04) 8

3w885 13885 UZb II 564 (08) 9 597 (08) 5

3w938 14369 UZb II 578 (06) 8

3w1015 15073 UZb II 558 (07) 9

3w1112 15959 UZb II 563 (06) 8 641 (09) 8 442 (02) 3 561 (07) 6

3w121225 16875 UZa II 574 (08) 8

3w12678 17382 UZa II 585 (11) 9

3w12689 17392 UZa II 607 (07) 8 673 (11) 9

3w1295 17631 UZa II 599 (12) 9

3w13147 17811 UZa I 571 (09) 8 662 (07) 5 552 (05) 12

2w124 17959 UZa I 585 (05) 9

3w1360 18224 UZa I 611 (04) 9

3w138125 18418 MZU I 607 (08) 7 616 (34) 5 548 (01) 3

2w225 18882 MZU I 642 (06) 9

2w324 19786 MZU I 674 (14) 9 727 (15) 9 659 (05) 9

2w402 20499 MZU I 713 (20) 8 706 (10) 7 647 (03) 8

2w448 20919 MZU I 708 (28) 7

2w4819 21229 MZU I 720 (24) 9 740 (08) 7 676 (04) 9

2w4919 21320 MZU 723 (10) 9

2w601 22317 MZL 674 (04) 8

2w6728 22973 MZL 571 (13) 8



2263

Dow






500

1000

1500

2000

2500

Pyroxenite Marker


Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

10 30 50 70 10 30 50 7050 60 70 07070 07075

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es


OPX (this study)



This study


(a) (d)(c)(b)



subz

ones

UZc

UZb

UZa

MZU

MZL


07085

mag

netit

ite la

yers

nels

onite

laye

rs



2264

Dow










020

40

60

80

0

10

20

30

40

50

60

70

80

20 30 40 50 60 70 80

cycle I

cycle II

cycle III

cycle IV

Mg = An

+apatite

+magnetite


cycle V

cycle VI

cycle VII+VIII

cycle IX

Plagioclase An

Clin

opyr

oxen

e M

g



2265

Dow











(a)

(b)

12

6

8

10

minus2

0

2

4

6

8

10

minus2

0

2

4

minus4

Magnetitite Layer 1



An in plagioclase

Stra

tigra

phic

Hei

ght (

met

res)

Stra

tigra

phic

Hei

ght (

met

res)

51 575553



2266

Dow









001 01 001 01 11 10

0

500

1000

1500

2000

2500m

agne

titite

laye

rsne

lson

ite la

yers

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es

subz

ones

UZc

UZb

UZa

MZU

MZL

Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

BK1BK3

(a) (b)





2267

Dow










of cumulate above

Pyroxenite Marker


melt at Skaergaard


of melt at

Pyroxenite Marker

1 2 3 4 5 6

SiO2 476 732 643 661 666 514

TiO2 12 05 20 21 04 10

Al2O3 173 133 120 134 136 166

FeO(total) 135 38 96 58 46 117

MnO 02 01 01

MgO 56 02 12 16 05 46

CaO 110 18 43 37 45 97

Na2O 27 41 36 23 39 29

K2O 04 33 18 29 18 07

P2O5 044 11 01 04

Total 999 1002 988 990 960 991



2268

Dow











2269

Dow








F 100 090 080 070 064 054 046 030 020



SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014



Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002


Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378


SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112



2270

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow







Zonal subdivision





2259

Dow




Petrography











Upper Zone c (UZc) 1128 910 350 610

Upper Zone b (UZb) 534 740 520 390

Upper Zone a (UZa) 200 580 640 590





2260

Dow








2261

Dow




Analytical methods






Complex


1w784y UZc 3371 2

1w823y UZc 3727 10

1w885y UZc 4293 6

1w1099y UZc 6248 10

1w11126y UZc 6373 30

1w1117 UZc 6413 6

1w1206y UZc 7226 6

1w12841 UZc 7939 3

1w14498 UZc 9453 20

1w14505 UZc 9459 10

1w14513 UZc 9467 70

1w1460 UZc 9540 710 (Layer 21)

1w1465 UZc 9592 68

1w1485 UZc 9774 25

1w14882 UZc 9804 40

1w14892 UZc 9813 3

1w14921 UZc 9839 5

1w149265 UZc 9844 20

1w1494 UZc 9857 60

1w1532 UZc 10204 14

3w746 UZb 12615 80

3w927 UZb 14269 43

3w932 UZb 14315 60

3w942 UZb 14406 10

3w1247 UZa 17192 26

3w1272 UZa 17420 134

3w1294 UZa 17621 107

3w1313 UZa 17795 64

3w1315 UZa 17813 26

3w1343 UZa 18069 53

3w1368 UZa 18297 75

3w1378 UZa 18389 246 (MML)




2262

Dow






1w4222 66 UZc IX 430 (23) 9 114 (07) 6 30 (02) 6

1w4318 154 UZc IX 452 (24) 9

1w4466 289 UZc IX 428 (14) 9 49 (04) 7 13 (01) 9

1w4751 549 UZc IX 434 (24) 8

1w5047 819 UZc IX 451 (18) 9 159 (03) 6 39 (03) 9

1w534 1087 UZc IX 446 (23) 8 308 (03) 4 94 (03) 9

1w568 1398 UZc IX 500 (19) 9 364 (07) 4 104 (09) 9

1w598 1672 UZc IX 463 (06) 8 268 (07) 7 78 (03) 9

1w62585 1926 UZc IX 498 (20) 8 502 (05) 3 214 (01) 9 369 (08) 6

1w6633 2268 UZc IX 487 (16) 9

1w732 2896 UZc VIII 519 (08) 9 436 (10) 5 220 (02) 8

1w770 3243 UZc VIII 493 (04) 9 469 (10) 4 216 (03) 9

1w8086 3596 UZc VII 505 (14) 8

1w8532 4003 UZc VII 498 (05) 8 462 (05) 6 187 (04) 9

1w9086 4509 UZc VII 495 (11) 8 133 (01) 4

1w954 4924 UZc VI 498 (08) 8 357 (05) 6 148 (05) 9

1w10102 5437 UZc VI 489 (13) 7

1w10505 5805 UZc VI 501 (05) 8 362 (08) 5 159 (01) 9

1w11382 6605 UZc VI 501 (09) 9 430 (04) 6 175 (04) 6 070711 (13)

1w11586 6793 UZc VI 481 (12) 6 255 (10) 5 62 (02) 9 070739 (14)

1w11901 7085 UZc V 456 (10) 7 325 (05) 6 146 (02) 9 070720 (13)

1w12395 7528 UZc V 483 (06) 9 383 (08) 5 160 (03) 8 070717 (15)

1w13032 8112 UZc V 070726 (13)

1w13292 8351 UZc V 509 (07) 8 473 (04) 4 232 (02) 9 070714 (13)

1w14234 9212 UZc V 520 (04) 9

1w148885 9810 UZc V 550 (07) 9 578 1 449 (06) 15 070758 (16)

1w15070 9975 UZc IV 492 (08) 7 475 (13) 4 341 (07) 8

1w15507 10375 UZc IV 485 (09) 9 566 (07) 5 340 (02) 6 070726 (13)

3w540 10734 UZb IV 498 (09) 8 375 (07) 9

3w590 11190 UZb IV 535 (13) 7

3w6409 11655 UZb IV 559 (10) 9 617 (05) 6 504 (12) 6

3w715 12332 UZb III 521 (15) 9 643 (11) 8 389 (14) 9

3w742 12579 UZb III 554 (08) 9

3w780 12926 UZb III 577 (18) 9 596 (08) 9

3w810 13200 UZb III 582 (04) 8

3w885 13885 UZb II 564 (08) 9 597 (08) 5

3w938 14369 UZb II 578 (06) 8

3w1015 15073 UZb II 558 (07) 9

3w1112 15959 UZb II 563 (06) 8 641 (09) 8 442 (02) 3 561 (07) 6

3w121225 16875 UZa II 574 (08) 8

3w12678 17382 UZa II 585 (11) 9

3w12689 17392 UZa II 607 (07) 8 673 (11) 9

3w1295 17631 UZa II 599 (12) 9

3w13147 17811 UZa I 571 (09) 8 662 (07) 5 552 (05) 12

2w124 17959 UZa I 585 (05) 9

3w1360 18224 UZa I 611 (04) 9

3w138125 18418 MZU I 607 (08) 7 616 (34) 5 548 (01) 3

2w225 18882 MZU I 642 (06) 9

2w324 19786 MZU I 674 (14) 9 727 (15) 9 659 (05) 9

2w402 20499 MZU I 713 (20) 8 706 (10) 7 647 (03) 8

2w448 20919 MZU I 708 (28) 7

2w4819 21229 MZU I 720 (24) 9 740 (08) 7 676 (04) 9

2w4919 21320 MZU 723 (10) 9

2w601 22317 MZL 674 (04) 8

2w6728 22973 MZL 571 (13) 8



2263

Dow






500

1000

1500

2000

2500

Pyroxenite Marker


Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

10 30 50 70 10 30 50 7050 60 70 07070 07075

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es


OPX (this study)



This study


(a) (d)(c)(b)



subz

ones

UZc

UZb

UZa

MZU

MZL


07085

mag

netit

ite la

yers

nels

onite

laye

rs



2264

Dow










020

40

60

80

0

10

20

30

40

50

60

70

80

20 30 40 50 60 70 80

cycle I

cycle II

cycle III

cycle IV

Mg = An

+apatite

+magnetite


cycle V

cycle VI

cycle VII+VIII

cycle IX

Plagioclase An

Clin

opyr

oxen

e M

g



2265

Dow











(a)

(b)

12

6

8

10

minus2

0

2

4

6

8

10

minus2

0

2

4

minus4

Magnetitite Layer 1



An in plagioclase

Stra

tigra

phic

Hei

ght (

met

res)

Stra

tigra

phic

Hei

ght (

met

res)

51 575553



2266

Dow









001 01 001 01 11 10

0

500

1000

1500

2000

2500m

agne

titite

laye

rsne

lson

ite la

yers

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es

subz

ones

UZc

UZb

UZa

MZU

MZL

Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

BK1BK3

(a) (b)





2267

Dow










of cumulate above

Pyroxenite Marker


melt at Skaergaard


of melt at

Pyroxenite Marker

1 2 3 4 5 6

SiO2 476 732 643 661 666 514

TiO2 12 05 20 21 04 10

Al2O3 173 133 120 134 136 166

FeO(total) 135 38 96 58 46 117

MnO 02 01 01

MgO 56 02 12 16 05 46

CaO 110 18 43 37 45 97

Na2O 27 41 36 23 39 29

K2O 04 33 18 29 18 07

P2O5 044 11 01 04

Total 999 1002 988 990 960 991



2268

Dow











2269

Dow








F 100 090 080 070 064 054 046 030 020



SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014



Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002


Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378


SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112



2270

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow




Petrography











Upper Zone c (UZc) 1128 910 350 610

Upper Zone b (UZb) 534 740 520 390

Upper Zone a (UZa) 200 580 640 590





2260

Dow








2261

Dow




Analytical methods






Complex


1w784y UZc 3371 2

1w823y UZc 3727 10

1w885y UZc 4293 6

1w1099y UZc 6248 10

1w11126y UZc 6373 30

1w1117 UZc 6413 6

1w1206y UZc 7226 6

1w12841 UZc 7939 3

1w14498 UZc 9453 20

1w14505 UZc 9459 10

1w14513 UZc 9467 70

1w1460 UZc 9540 710 (Layer 21)

1w1465 UZc 9592 68

1w1485 UZc 9774 25

1w14882 UZc 9804 40

1w14892 UZc 9813 3

1w14921 UZc 9839 5

1w149265 UZc 9844 20

1w1494 UZc 9857 60

1w1532 UZc 10204 14

3w746 UZb 12615 80

3w927 UZb 14269 43

3w932 UZb 14315 60

3w942 UZb 14406 10

3w1247 UZa 17192 26

3w1272 UZa 17420 134

3w1294 UZa 17621 107

3w1313 UZa 17795 64

3w1315 UZa 17813 26

3w1343 UZa 18069 53

3w1368 UZa 18297 75

3w1378 UZa 18389 246 (MML)




2262

Dow






1w4222 66 UZc IX 430 (23) 9 114 (07) 6 30 (02) 6

1w4318 154 UZc IX 452 (24) 9

1w4466 289 UZc IX 428 (14) 9 49 (04) 7 13 (01) 9

1w4751 549 UZc IX 434 (24) 8

1w5047 819 UZc IX 451 (18) 9 159 (03) 6 39 (03) 9

1w534 1087 UZc IX 446 (23) 8 308 (03) 4 94 (03) 9

1w568 1398 UZc IX 500 (19) 9 364 (07) 4 104 (09) 9

1w598 1672 UZc IX 463 (06) 8 268 (07) 7 78 (03) 9

1w62585 1926 UZc IX 498 (20) 8 502 (05) 3 214 (01) 9 369 (08) 6

1w6633 2268 UZc IX 487 (16) 9

1w732 2896 UZc VIII 519 (08) 9 436 (10) 5 220 (02) 8

1w770 3243 UZc VIII 493 (04) 9 469 (10) 4 216 (03) 9

1w8086 3596 UZc VII 505 (14) 8

1w8532 4003 UZc VII 498 (05) 8 462 (05) 6 187 (04) 9

1w9086 4509 UZc VII 495 (11) 8 133 (01) 4

1w954 4924 UZc VI 498 (08) 8 357 (05) 6 148 (05) 9

1w10102 5437 UZc VI 489 (13) 7

1w10505 5805 UZc VI 501 (05) 8 362 (08) 5 159 (01) 9

1w11382 6605 UZc VI 501 (09) 9 430 (04) 6 175 (04) 6 070711 (13)

1w11586 6793 UZc VI 481 (12) 6 255 (10) 5 62 (02) 9 070739 (14)

1w11901 7085 UZc V 456 (10) 7 325 (05) 6 146 (02) 9 070720 (13)

1w12395 7528 UZc V 483 (06) 9 383 (08) 5 160 (03) 8 070717 (15)

1w13032 8112 UZc V 070726 (13)

1w13292 8351 UZc V 509 (07) 8 473 (04) 4 232 (02) 9 070714 (13)

1w14234 9212 UZc V 520 (04) 9

1w148885 9810 UZc V 550 (07) 9 578 1 449 (06) 15 070758 (16)

1w15070 9975 UZc IV 492 (08) 7 475 (13) 4 341 (07) 8

1w15507 10375 UZc IV 485 (09) 9 566 (07) 5 340 (02) 6 070726 (13)

3w540 10734 UZb IV 498 (09) 8 375 (07) 9

3w590 11190 UZb IV 535 (13) 7

3w6409 11655 UZb IV 559 (10) 9 617 (05) 6 504 (12) 6

3w715 12332 UZb III 521 (15) 9 643 (11) 8 389 (14) 9

3w742 12579 UZb III 554 (08) 9

3w780 12926 UZb III 577 (18) 9 596 (08) 9

3w810 13200 UZb III 582 (04) 8

3w885 13885 UZb II 564 (08) 9 597 (08) 5

3w938 14369 UZb II 578 (06) 8

3w1015 15073 UZb II 558 (07) 9

3w1112 15959 UZb II 563 (06) 8 641 (09) 8 442 (02) 3 561 (07) 6

3w121225 16875 UZa II 574 (08) 8

3w12678 17382 UZa II 585 (11) 9

3w12689 17392 UZa II 607 (07) 8 673 (11) 9

3w1295 17631 UZa II 599 (12) 9

3w13147 17811 UZa I 571 (09) 8 662 (07) 5 552 (05) 12

2w124 17959 UZa I 585 (05) 9

3w1360 18224 UZa I 611 (04) 9

3w138125 18418 MZU I 607 (08) 7 616 (34) 5 548 (01) 3

2w225 18882 MZU I 642 (06) 9

2w324 19786 MZU I 674 (14) 9 727 (15) 9 659 (05) 9

2w402 20499 MZU I 713 (20) 8 706 (10) 7 647 (03) 8

2w448 20919 MZU I 708 (28) 7

2w4819 21229 MZU I 720 (24) 9 740 (08) 7 676 (04) 9

2w4919 21320 MZU 723 (10) 9

2w601 22317 MZL 674 (04) 8

2w6728 22973 MZL 571 (13) 8



2263

Dow






500

1000

1500

2000

2500

Pyroxenite Marker


Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

10 30 50 70 10 30 50 7050 60 70 07070 07075

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es


OPX (this study)



This study


(a) (d)(c)(b)



subz

ones

UZc

UZb

UZa

MZU

MZL


07085

mag

netit

ite la

yers

nels

onite

laye

rs



2264

Dow










020

40

60

80

0

10

20

30

40

50

60

70

80

20 30 40 50 60 70 80

cycle I

cycle II

cycle III

cycle IV

Mg = An

+apatite

+magnetite


cycle V

cycle VI

cycle VII+VIII

cycle IX

Plagioclase An

Clin

opyr

oxen

e M

g



2265

Dow











(a)

(b)

12

6

8

10

minus2

0

2

4

6

8

10

minus2

0

2

4

minus4

Magnetitite Layer 1



An in plagioclase

Stra

tigra

phic

Hei

ght (

met

res)

Stra

tigra

phic

Hei

ght (

met

res)

51 575553



2266

Dow









001 01 001 01 11 10

0

500

1000

1500

2000

2500m

agne

titite

laye

rsne

lson

ite la

yers

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es

subz

ones

UZc

UZb

UZa

MZU

MZL

Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

BK1BK3

(a) (b)





2267

Dow










of cumulate above

Pyroxenite Marker


melt at Skaergaard


of melt at

Pyroxenite Marker

1 2 3 4 5 6

SiO2 476 732 643 661 666 514

TiO2 12 05 20 21 04 10

Al2O3 173 133 120 134 136 166

FeO(total) 135 38 96 58 46 117

MnO 02 01 01

MgO 56 02 12 16 05 46

CaO 110 18 43 37 45 97

Na2O 27 41 36 23 39 29

K2O 04 33 18 29 18 07

P2O5 044 11 01 04

Total 999 1002 988 990 960 991



2268

Dow











2269

Dow








F 100 090 080 070 064 054 046 030 020



SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014



Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002


Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378


SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112



2270

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow








2261

Dow




Analytical methods






Complex


1w784y UZc 3371 2

1w823y UZc 3727 10

1w885y UZc 4293 6

1w1099y UZc 6248 10

1w11126y UZc 6373 30

1w1117 UZc 6413 6

1w1206y UZc 7226 6

1w12841 UZc 7939 3

1w14498 UZc 9453 20

1w14505 UZc 9459 10

1w14513 UZc 9467 70

1w1460 UZc 9540 710 (Layer 21)

1w1465 UZc 9592 68

1w1485 UZc 9774 25

1w14882 UZc 9804 40

1w14892 UZc 9813 3

1w14921 UZc 9839 5

1w149265 UZc 9844 20

1w1494 UZc 9857 60

1w1532 UZc 10204 14

3w746 UZb 12615 80

3w927 UZb 14269 43

3w932 UZb 14315 60

3w942 UZb 14406 10

3w1247 UZa 17192 26

3w1272 UZa 17420 134

3w1294 UZa 17621 107

3w1313 UZa 17795 64

3w1315 UZa 17813 26

3w1343 UZa 18069 53

3w1368 UZa 18297 75

3w1378 UZa 18389 246 (MML)




2262

Dow






1w4222 66 UZc IX 430 (23) 9 114 (07) 6 30 (02) 6

1w4318 154 UZc IX 452 (24) 9

1w4466 289 UZc IX 428 (14) 9 49 (04) 7 13 (01) 9

1w4751 549 UZc IX 434 (24) 8

1w5047 819 UZc IX 451 (18) 9 159 (03) 6 39 (03) 9

1w534 1087 UZc IX 446 (23) 8 308 (03) 4 94 (03) 9

1w568 1398 UZc IX 500 (19) 9 364 (07) 4 104 (09) 9

1w598 1672 UZc IX 463 (06) 8 268 (07) 7 78 (03) 9

1w62585 1926 UZc IX 498 (20) 8 502 (05) 3 214 (01) 9 369 (08) 6

1w6633 2268 UZc IX 487 (16) 9

1w732 2896 UZc VIII 519 (08) 9 436 (10) 5 220 (02) 8

1w770 3243 UZc VIII 493 (04) 9 469 (10) 4 216 (03) 9

1w8086 3596 UZc VII 505 (14) 8

1w8532 4003 UZc VII 498 (05) 8 462 (05) 6 187 (04) 9

1w9086 4509 UZc VII 495 (11) 8 133 (01) 4

1w954 4924 UZc VI 498 (08) 8 357 (05) 6 148 (05) 9

1w10102 5437 UZc VI 489 (13) 7

1w10505 5805 UZc VI 501 (05) 8 362 (08) 5 159 (01) 9

1w11382 6605 UZc VI 501 (09) 9 430 (04) 6 175 (04) 6 070711 (13)

1w11586 6793 UZc VI 481 (12) 6 255 (10) 5 62 (02) 9 070739 (14)

1w11901 7085 UZc V 456 (10) 7 325 (05) 6 146 (02) 9 070720 (13)

1w12395 7528 UZc V 483 (06) 9 383 (08) 5 160 (03) 8 070717 (15)

1w13032 8112 UZc V 070726 (13)

1w13292 8351 UZc V 509 (07) 8 473 (04) 4 232 (02) 9 070714 (13)

1w14234 9212 UZc V 520 (04) 9

1w148885 9810 UZc V 550 (07) 9 578 1 449 (06) 15 070758 (16)

1w15070 9975 UZc IV 492 (08) 7 475 (13) 4 341 (07) 8

1w15507 10375 UZc IV 485 (09) 9 566 (07) 5 340 (02) 6 070726 (13)

3w540 10734 UZb IV 498 (09) 8 375 (07) 9

3w590 11190 UZb IV 535 (13) 7

3w6409 11655 UZb IV 559 (10) 9 617 (05) 6 504 (12) 6

3w715 12332 UZb III 521 (15) 9 643 (11) 8 389 (14) 9

3w742 12579 UZb III 554 (08) 9

3w780 12926 UZb III 577 (18) 9 596 (08) 9

3w810 13200 UZb III 582 (04) 8

3w885 13885 UZb II 564 (08) 9 597 (08) 5

3w938 14369 UZb II 578 (06) 8

3w1015 15073 UZb II 558 (07) 9

3w1112 15959 UZb II 563 (06) 8 641 (09) 8 442 (02) 3 561 (07) 6

3w121225 16875 UZa II 574 (08) 8

3w12678 17382 UZa II 585 (11) 9

3w12689 17392 UZa II 607 (07) 8 673 (11) 9

3w1295 17631 UZa II 599 (12) 9

3w13147 17811 UZa I 571 (09) 8 662 (07) 5 552 (05) 12

2w124 17959 UZa I 585 (05) 9

3w1360 18224 UZa I 611 (04) 9

3w138125 18418 MZU I 607 (08) 7 616 (34) 5 548 (01) 3

2w225 18882 MZU I 642 (06) 9

2w324 19786 MZU I 674 (14) 9 727 (15) 9 659 (05) 9

2w402 20499 MZU I 713 (20) 8 706 (10) 7 647 (03) 8

2w448 20919 MZU I 708 (28) 7

2w4819 21229 MZU I 720 (24) 9 740 (08) 7 676 (04) 9

2w4919 21320 MZU 723 (10) 9

2w601 22317 MZL 674 (04) 8

2w6728 22973 MZL 571 (13) 8



2263

Dow






500

1000

1500

2000

2500

Pyroxenite Marker


Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

10 30 50 70 10 30 50 7050 60 70 07070 07075

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es


OPX (this study)



This study


(a) (d)(c)(b)



subz

ones

UZc

UZb

UZa

MZU

MZL


07085

mag

netit

ite la

yers

nels

onite

laye

rs



2264

Dow










020

40

60

80

0

10

20

30

40

50

60

70

80

20 30 40 50 60 70 80

cycle I

cycle II

cycle III

cycle IV

Mg = An

+apatite

+magnetite


cycle V

cycle VI

cycle VII+VIII

cycle IX

Plagioclase An

Clin

opyr

oxen

e M

g



2265

Dow











(a)

(b)

12

6

8

10

minus2

0

2

4

6

8

10

minus2

0

2

4

minus4

Magnetitite Layer 1



An in plagioclase

Stra

tigra

phic

Hei

ght (

met

res)

Stra

tigra

phic

Hei

ght (

met

res)

51 575553



2266

Dow









001 01 001 01 11 10

0

500

1000

1500

2000

2500m

agne

titite

laye

rsne

lson

ite la

yers

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es

subz

ones

UZc

UZb

UZa

MZU

MZL

Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

BK1BK3

(a) (b)





2267

Dow










of cumulate above

Pyroxenite Marker


melt at Skaergaard


of melt at

Pyroxenite Marker

1 2 3 4 5 6

SiO2 476 732 643 661 666 514

TiO2 12 05 20 21 04 10

Al2O3 173 133 120 134 136 166

FeO(total) 135 38 96 58 46 117

MnO 02 01 01

MgO 56 02 12 16 05 46

CaO 110 18 43 37 45 97

Na2O 27 41 36 23 39 29

K2O 04 33 18 29 18 07

P2O5 044 11 01 04

Total 999 1002 988 990 960 991



2268

Dow











2269

Dow








F 100 090 080 070 064 054 046 030 020



SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014



Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002


Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378


SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112



2270

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow




Analytical methods






Complex


1w784y UZc 3371 2

1w823y UZc 3727 10

1w885y UZc 4293 6

1w1099y UZc 6248 10

1w11126y UZc 6373 30

1w1117 UZc 6413 6

1w1206y UZc 7226 6

1w12841 UZc 7939 3

1w14498 UZc 9453 20

1w14505 UZc 9459 10

1w14513 UZc 9467 70

1w1460 UZc 9540 710 (Layer 21)

1w1465 UZc 9592 68

1w1485 UZc 9774 25

1w14882 UZc 9804 40

1w14892 UZc 9813 3

1w14921 UZc 9839 5

1w149265 UZc 9844 20

1w1494 UZc 9857 60

1w1532 UZc 10204 14

3w746 UZb 12615 80

3w927 UZb 14269 43

3w932 UZb 14315 60

3w942 UZb 14406 10

3w1247 UZa 17192 26

3w1272 UZa 17420 134

3w1294 UZa 17621 107

3w1313 UZa 17795 64

3w1315 UZa 17813 26

3w1343 UZa 18069 53

3w1368 UZa 18297 75

3w1378 UZa 18389 246 (MML)




2262

Dow






1w4222 66 UZc IX 430 (23) 9 114 (07) 6 30 (02) 6

1w4318 154 UZc IX 452 (24) 9

1w4466 289 UZc IX 428 (14) 9 49 (04) 7 13 (01) 9

1w4751 549 UZc IX 434 (24) 8

1w5047 819 UZc IX 451 (18) 9 159 (03) 6 39 (03) 9

1w534 1087 UZc IX 446 (23) 8 308 (03) 4 94 (03) 9

1w568 1398 UZc IX 500 (19) 9 364 (07) 4 104 (09) 9

1w598 1672 UZc IX 463 (06) 8 268 (07) 7 78 (03) 9

1w62585 1926 UZc IX 498 (20) 8 502 (05) 3 214 (01) 9 369 (08) 6

1w6633 2268 UZc IX 487 (16) 9

1w732 2896 UZc VIII 519 (08) 9 436 (10) 5 220 (02) 8

1w770 3243 UZc VIII 493 (04) 9 469 (10) 4 216 (03) 9

1w8086 3596 UZc VII 505 (14) 8

1w8532 4003 UZc VII 498 (05) 8 462 (05) 6 187 (04) 9

1w9086 4509 UZc VII 495 (11) 8 133 (01) 4

1w954 4924 UZc VI 498 (08) 8 357 (05) 6 148 (05) 9

1w10102 5437 UZc VI 489 (13) 7

1w10505 5805 UZc VI 501 (05) 8 362 (08) 5 159 (01) 9

1w11382 6605 UZc VI 501 (09) 9 430 (04) 6 175 (04) 6 070711 (13)

1w11586 6793 UZc VI 481 (12) 6 255 (10) 5 62 (02) 9 070739 (14)

1w11901 7085 UZc V 456 (10) 7 325 (05) 6 146 (02) 9 070720 (13)

1w12395 7528 UZc V 483 (06) 9 383 (08) 5 160 (03) 8 070717 (15)

1w13032 8112 UZc V 070726 (13)

1w13292 8351 UZc V 509 (07) 8 473 (04) 4 232 (02) 9 070714 (13)

1w14234 9212 UZc V 520 (04) 9

1w148885 9810 UZc V 550 (07) 9 578 1 449 (06) 15 070758 (16)

1w15070 9975 UZc IV 492 (08) 7 475 (13) 4 341 (07) 8

1w15507 10375 UZc IV 485 (09) 9 566 (07) 5 340 (02) 6 070726 (13)

3w540 10734 UZb IV 498 (09) 8 375 (07) 9

3w590 11190 UZb IV 535 (13) 7

3w6409 11655 UZb IV 559 (10) 9 617 (05) 6 504 (12) 6

3w715 12332 UZb III 521 (15) 9 643 (11) 8 389 (14) 9

3w742 12579 UZb III 554 (08) 9

3w780 12926 UZb III 577 (18) 9 596 (08) 9

3w810 13200 UZb III 582 (04) 8

3w885 13885 UZb II 564 (08) 9 597 (08) 5

3w938 14369 UZb II 578 (06) 8

3w1015 15073 UZb II 558 (07) 9

3w1112 15959 UZb II 563 (06) 8 641 (09) 8 442 (02) 3 561 (07) 6

3w121225 16875 UZa II 574 (08) 8

3w12678 17382 UZa II 585 (11) 9

3w12689 17392 UZa II 607 (07) 8 673 (11) 9

3w1295 17631 UZa II 599 (12) 9

3w13147 17811 UZa I 571 (09) 8 662 (07) 5 552 (05) 12

2w124 17959 UZa I 585 (05) 9

3w1360 18224 UZa I 611 (04) 9

3w138125 18418 MZU I 607 (08) 7 616 (34) 5 548 (01) 3

2w225 18882 MZU I 642 (06) 9

2w324 19786 MZU I 674 (14) 9 727 (15) 9 659 (05) 9

2w402 20499 MZU I 713 (20) 8 706 (10) 7 647 (03) 8

2w448 20919 MZU I 708 (28) 7

2w4819 21229 MZU I 720 (24) 9 740 (08) 7 676 (04) 9

2w4919 21320 MZU 723 (10) 9

2w601 22317 MZL 674 (04) 8

2w6728 22973 MZL 571 (13) 8



2263

Dow






500

1000

1500

2000

2500

Pyroxenite Marker


Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

10 30 50 70 10 30 50 7050 60 70 07070 07075

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es


OPX (this study)



This study


(a) (d)(c)(b)



subz

ones

UZc

UZb

UZa

MZU

MZL


07085

mag

netit

ite la

yers

nels

onite

laye

rs



2264

Dow










020

40

60

80

0

10

20

30

40

50

60

70

80

20 30 40 50 60 70 80

cycle I

cycle II

cycle III

cycle IV

Mg = An

+apatite

+magnetite


cycle V

cycle VI

cycle VII+VIII

cycle IX

Plagioclase An

Clin

opyr

oxen

e M

g



2265

Dow











(a)

(b)

12

6

8

10

minus2

0

2

4

6

8

10

minus2

0

2

4

minus4

Magnetitite Layer 1



An in plagioclase

Stra

tigra

phic

Hei

ght (

met

res)

Stra

tigra

phic

Hei

ght (

met

res)

51 575553



2266

Dow









001 01 001 01 11 10

0

500

1000

1500

2000

2500m

agne

titite

laye

rsne

lson

ite la

yers

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es

subz

ones

UZc

UZb

UZa

MZU

MZL

Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

BK1BK3

(a) (b)





2267

Dow










of cumulate above

Pyroxenite Marker


melt at Skaergaard


of melt at

Pyroxenite Marker

1 2 3 4 5 6

SiO2 476 732 643 661 666 514

TiO2 12 05 20 21 04 10

Al2O3 173 133 120 134 136 166

FeO(total) 135 38 96 58 46 117

MnO 02 01 01

MgO 56 02 12 16 05 46

CaO 110 18 43 37 45 97

Na2O 27 41 36 23 39 29

K2O 04 33 18 29 18 07

P2O5 044 11 01 04

Total 999 1002 988 990 960 991



2268

Dow











2269

Dow








F 100 090 080 070 064 054 046 030 020



SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014



Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002


Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378


SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112



2270

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow






1w4222 66 UZc IX 430 (23) 9 114 (07) 6 30 (02) 6

1w4318 154 UZc IX 452 (24) 9

1w4466 289 UZc IX 428 (14) 9 49 (04) 7 13 (01) 9

1w4751 549 UZc IX 434 (24) 8

1w5047 819 UZc IX 451 (18) 9 159 (03) 6 39 (03) 9

1w534 1087 UZc IX 446 (23) 8 308 (03) 4 94 (03) 9

1w568 1398 UZc IX 500 (19) 9 364 (07) 4 104 (09) 9

1w598 1672 UZc IX 463 (06) 8 268 (07) 7 78 (03) 9

1w62585 1926 UZc IX 498 (20) 8 502 (05) 3 214 (01) 9 369 (08) 6

1w6633 2268 UZc IX 487 (16) 9

1w732 2896 UZc VIII 519 (08) 9 436 (10) 5 220 (02) 8

1w770 3243 UZc VIII 493 (04) 9 469 (10) 4 216 (03) 9

1w8086 3596 UZc VII 505 (14) 8

1w8532 4003 UZc VII 498 (05) 8 462 (05) 6 187 (04) 9

1w9086 4509 UZc VII 495 (11) 8 133 (01) 4

1w954 4924 UZc VI 498 (08) 8 357 (05) 6 148 (05) 9

1w10102 5437 UZc VI 489 (13) 7

1w10505 5805 UZc VI 501 (05) 8 362 (08) 5 159 (01) 9

1w11382 6605 UZc VI 501 (09) 9 430 (04) 6 175 (04) 6 070711 (13)

1w11586 6793 UZc VI 481 (12) 6 255 (10) 5 62 (02) 9 070739 (14)

1w11901 7085 UZc V 456 (10) 7 325 (05) 6 146 (02) 9 070720 (13)

1w12395 7528 UZc V 483 (06) 9 383 (08) 5 160 (03) 8 070717 (15)

1w13032 8112 UZc V 070726 (13)

1w13292 8351 UZc V 509 (07) 8 473 (04) 4 232 (02) 9 070714 (13)

1w14234 9212 UZc V 520 (04) 9

1w148885 9810 UZc V 550 (07) 9 578 1 449 (06) 15 070758 (16)

1w15070 9975 UZc IV 492 (08) 7 475 (13) 4 341 (07) 8

1w15507 10375 UZc IV 485 (09) 9 566 (07) 5 340 (02) 6 070726 (13)

3w540 10734 UZb IV 498 (09) 8 375 (07) 9

3w590 11190 UZb IV 535 (13) 7

3w6409 11655 UZb IV 559 (10) 9 617 (05) 6 504 (12) 6

3w715 12332 UZb III 521 (15) 9 643 (11) 8 389 (14) 9

3w742 12579 UZb III 554 (08) 9

3w780 12926 UZb III 577 (18) 9 596 (08) 9

3w810 13200 UZb III 582 (04) 8

3w885 13885 UZb II 564 (08) 9 597 (08) 5

3w938 14369 UZb II 578 (06) 8

3w1015 15073 UZb II 558 (07) 9

3w1112 15959 UZb II 563 (06) 8 641 (09) 8 442 (02) 3 561 (07) 6

3w121225 16875 UZa II 574 (08) 8

3w12678 17382 UZa II 585 (11) 9

3w12689 17392 UZa II 607 (07) 8 673 (11) 9

3w1295 17631 UZa II 599 (12) 9

3w13147 17811 UZa I 571 (09) 8 662 (07) 5 552 (05) 12

2w124 17959 UZa I 585 (05) 9

3w1360 18224 UZa I 611 (04) 9

3w138125 18418 MZU I 607 (08) 7 616 (34) 5 548 (01) 3

2w225 18882 MZU I 642 (06) 9

2w324 19786 MZU I 674 (14) 9 727 (15) 9 659 (05) 9

2w402 20499 MZU I 713 (20) 8 706 (10) 7 647 (03) 8

2w448 20919 MZU I 708 (28) 7

2w4819 21229 MZU I 720 (24) 9 740 (08) 7 676 (04) 9

2w4919 21320 MZU 723 (10) 9

2w601 22317 MZL 674 (04) 8

2w6728 22973 MZL 571 (13) 8



2263

Dow






500

1000

1500

2000

2500

Pyroxenite Marker


Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

10 30 50 70 10 30 50 7050 60 70 07070 07075

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es


OPX (this study)



This study


(a) (d)(c)(b)



subz

ones

UZc

UZb

UZa

MZU

MZL


07085

mag

netit

ite la

yers

nels

onite

laye

rs



2264

Dow










020

40

60

80

0

10

20

30

40

50

60

70

80

20 30 40 50 60 70 80

cycle I

cycle II

cycle III

cycle IV

Mg = An

+apatite

+magnetite


cycle V

cycle VI

cycle VII+VIII

cycle IX

Plagioclase An

Clin

opyr

oxen

e M

g



2265

Dow











(a)

(b)

12

6

8

10

minus2

0

2

4

6

8

10

minus2

0

2

4

minus4

Magnetitite Layer 1



An in plagioclase

Stra

tigra

phic

Hei

ght (

met

res)

Stra

tigra

phic

Hei

ght (

met

res)

51 575553



2266

Dow









001 01 001 01 11 10

0

500

1000

1500

2000

2500m

agne

titite

laye

rsne

lson

ite la

yers

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es

subz

ones

UZc

UZb

UZa

MZU

MZL

Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

BK1BK3

(a) (b)





2267

Dow










of cumulate above

Pyroxenite Marker


melt at Skaergaard


of melt at

Pyroxenite Marker

1 2 3 4 5 6

SiO2 476 732 643 661 666 514

TiO2 12 05 20 21 04 10

Al2O3 173 133 120 134 136 166

FeO(total) 135 38 96 58 46 117

MnO 02 01 01

MgO 56 02 12 16 05 46

CaO 110 18 43 37 45 97

Na2O 27 41 36 23 39 29

K2O 04 33 18 29 18 07

P2O5 044 11 01 04

Total 999 1002 988 990 960 991



2268

Dow











2269

Dow








F 100 090 080 070 064 054 046 030 020



SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014



Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002


Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378


SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112



2270

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow






500

1000

1500

2000

2500

Pyroxenite Marker


Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

10 30 50 70 10 30 50 7050 60 70 07070 07075

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es


OPX (this study)



This study


(a) (d)(c)(b)



subz

ones

UZc

UZb

UZa

MZU

MZL


07085

mag

netit

ite la

yers

nels

onite

laye

rs



2264

Dow










020

40

60

80

0

10

20

30

40

50

60

70

80

20 30 40 50 60 70 80

cycle I

cycle II

cycle III

cycle IV

Mg = An

+apatite

+magnetite


cycle V

cycle VI

cycle VII+VIII

cycle IX

Plagioclase An

Clin

opyr

oxen

e M

g



2265

Dow











(a)

(b)

12

6

8

10

minus2

0

2

4

6

8

10

minus2

0

2

4

minus4

Magnetitite Layer 1



An in plagioclase

Stra

tigra

phic

Hei

ght (

met

res)

Stra

tigra

phic

Hei

ght (

met

res)

51 575553



2266

Dow









001 01 001 01 11 10

0

500

1000

1500

2000

2500m

agne

titite

laye

rsne

lson

ite la

yers

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es

subz

ones

UZc

UZb

UZa

MZU

MZL

Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

BK1BK3

(a) (b)





2267

Dow










of cumulate above

Pyroxenite Marker


melt at Skaergaard


of melt at

Pyroxenite Marker

1 2 3 4 5 6

SiO2 476 732 643 661 666 514

TiO2 12 05 20 21 04 10

Al2O3 173 133 120 134 136 166

FeO(total) 135 38 96 58 46 117

MnO 02 01 01

MgO 56 02 12 16 05 46

CaO 110 18 43 37 45 97

Na2O 27 41 36 23 39 29

K2O 04 33 18 29 18 07

P2O5 044 11 01 04

Total 999 1002 988 990 960 991



2268

Dow











2269

Dow








F 100 090 080 070 064 054 046 030 020



SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014



Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002


Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378


SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112



2270

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow










020

40

60

80

0

10

20

30

40

50

60

70

80

20 30 40 50 60 70 80

cycle I

cycle II

cycle III

cycle IV

Mg = An

+apatite

+magnetite


cycle V

cycle VI

cycle VII+VIII

cycle IX

Plagioclase An

Clin

opyr

oxen

e M

g



2265

Dow











(a)

(b)

12

6

8

10

minus2

0

2

4

6

8

10

minus2

0

2

4

minus4

Magnetitite Layer 1



An in plagioclase

Stra

tigra

phic

Hei

ght (

met

res)

Stra

tigra

phic

Hei

ght (

met

res)

51 575553



2266

Dow









001 01 001 01 11 10

0

500

1000

1500

2000

2500m

agne

titite

laye

rsne

lson

ite la

yers

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es

subz

ones

UZc

UZb

UZa

MZU

MZL

Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

BK1BK3

(a) (b)





2267

Dow










of cumulate above

Pyroxenite Marker


melt at Skaergaard


of melt at

Pyroxenite Marker

1 2 3 4 5 6

SiO2 476 732 643 661 666 514

TiO2 12 05 20 21 04 10

Al2O3 173 133 120 134 136 166

FeO(total) 135 38 96 58 46 117

MnO 02 01 01

MgO 56 02 12 16 05 46

CaO 110 18 43 37 45 97

Na2O 27 41 36 23 39 29

K2O 04 33 18 29 18 07

P2O5 044 11 01 04

Total 999 1002 988 990 960 991



2268

Dow











2269

Dow








F 100 090 080 070 064 054 046 030 020



SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014



Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002


Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378


SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112



2270

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow











(a)

(b)

12

6

8

10

minus2

0

2

4

6

8

10

minus2

0

2

4

minus4

Magnetitite Layer 1



An in plagioclase

Stra

tigra

phic

Hei

ght (

met

res)

Stra

tigra

phic

Hei

ght (

met

res)

51 575553



2266

Dow









001 01 001 01 11 10

0

500

1000

1500

2000

2500m

agne

titite

laye

rsne

lson

ite la

yers

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es

subz

ones

UZc

UZb

UZa

MZU

MZL

Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

BK1BK3

(a) (b)





2267

Dow










of cumulate above

Pyroxenite Marker


melt at Skaergaard


of melt at

Pyroxenite Marker

1 2 3 4 5 6

SiO2 476 732 643 661 666 514

TiO2 12 05 20 21 04 10

Al2O3 173 133 120 134 136 166

FeO(total) 135 38 96 58 46 117

MnO 02 01 01

MgO 56 02 12 16 05 46

CaO 110 18 43 37 45 97

Na2O 27 41 36 23 39 29

K2O 04 33 18 29 18 07

P2O5 044 11 01 04

Total 999 1002 988 990 960 991



2268

Dow











2269

Dow








F 100 090 080 070 064 054 046 030 020



SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014



Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002


Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378


SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112



2270

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow









001 01 001 01 11 10

0

500

1000

1500

2000

2500m

agne

titite

laye

rsne

lson

ite la

yers

I

II

III

IV

V

VI

VII

VIII

IX

cycl

es

subz

ones

UZc

UZb

UZa

MZU

MZL

Stra

tigra

phic

pos

ition

(m

etre

s be

low

roo

f)

BK1BK3

(a) (b)





2267

Dow










of cumulate above

Pyroxenite Marker


melt at Skaergaard


of melt at

Pyroxenite Marker

1 2 3 4 5 6

SiO2 476 732 643 661 666 514

TiO2 12 05 20 21 04 10

Al2O3 173 133 120 134 136 166

FeO(total) 135 38 96 58 46 117

MnO 02 01 01

MgO 56 02 12 16 05 46

CaO 110 18 43 37 45 97

Na2O 27 41 36 23 39 29

K2O 04 33 18 29 18 07

P2O5 044 11 01 04

Total 999 1002 988 990 960 991



2268

Dow











2269

Dow








F 100 090 080 070 064 054 046 030 020



SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014



Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002


Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378


SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112



2270

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow










of cumulate above

Pyroxenite Marker


melt at Skaergaard


of melt at

Pyroxenite Marker

1 2 3 4 5 6

SiO2 476 732 643 661 666 514

TiO2 12 05 20 21 04 10

Al2O3 173 133 120 134 136 166

FeO(total) 135 38 96 58 46 117

MnO 02 01 01

MgO 56 02 12 16 05 46

CaO 110 18 43 37 45 97

Na2O 27 41 36 23 39 29

K2O 04 33 18 29 18 07

P2O5 044 11 01 04

Total 999 1002 988 990 960 991



2268

Dow











2269

Dow








F 100 090 080 070 064 054 046 030 020



SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014



Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002


Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378


SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112



2270

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow











2269

Dow








F 100 090 080 070 064 054 046 030 020



SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014



Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002


Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378


SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112



2270

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow








F 100 090 080 070 064 054 046 030 020



SiO2 5140 5209 5216 5224 5229 5333 5466 6044 6772

TiO2 100 112 124 141 153 142 131 093 057

Al2O3 1660 1653 1620 1582 1555 1509 1466 1350 1210

Fe2O3 150 167 202 250 288 267 246 187 120

FeO 1020 1083 1125 1168 1191 1222 1217 1061 733

MgO 460 424 376 320 283 239 188 070 007

CaO 950 926 884 833 798 742 695 541 372

Na2O 290 303 314 325 333 348 361 392 406

K2O 070 077 086 096 104 121 140 210 308

P2O5 040 045 051 058 064 075 089 051 014



Plagioclase 058 058 058 058 057 057 057 057 057

Orthopyroxene 017 017 017 017 012 007 005 005 005

Clinopyroxene 025 025 025 025 021 018 013 013 015

Olivine 000 000 000 000 000 008 013 013 013

Magnetite 000 000 000 000 010 010 010 008 008

Apatite 000 000 000 000 000 000 003 003 002


Cpx Mg 763 746 724 692 666 631 583 389 89

Opx Mg 732 704 669 624 589 541 484 287 56

Ol Fo 496 431 233 42

Plag An 685 670 652 630 614 587 561 478 378


SiO2 5190 5200 5213 5226 4714 4607 4436 4495 4670

TiO2 009 010 011 012 211 209 206 164 163

Al2O3 1949 1934 1916 1895 1826 1781 1732 1651 1561

Fe2O3 000 000 000 000 550 550 550 440 440

FeO 520 564 617 689 817 1028 1206 1445 1813

MgO 857 825 788 737 539 553 495 278 056

CaO 1299 1280 1257 1229 1120 1029 1036 912 799

Na2O 207 217 229 244 250 269 285 341 409

K2O 013 013 013 013 013 013 013 013 013

P2O5 000 000 000 000 000 000 168 140 112



2270

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow






26

25

24

27

020304050607080910

16

12

8

4

0

02

04

06

08

10

20

0

40

60

FeO

CaO

Al2O3

SiO210







plagioclase

cross-over

MZU UZa UZb UZc

clinopyroxene


apatite

An

Na2O

5K2O

TiO2

10P2O5MgO

Mg of clinopyroxene



2271

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow








DISCUSSION








2272

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow





+mtmixing

+mt

+mtmixing

mixingcycle II

cycle I


stra

tifie

d m

agm

ast

ratif

ied

mag

ma

stra

tifie

d m

agm

aho

mog

eneo

us m

agm

a

cum

ulat

ecu

mul

ate

cum

ulat

e

(a)

(b)

(c)

(d)




2273

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow











2274

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow














2275

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow









00

02

04

06

08

10



Stra

tigra

phic

pos

ition

(no

rmal

ised

)

spor

adic



2276

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow





CONCLUSIONS











2277

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow



SUPPLEMENTARY DATA


REFERENCES



























1016ndash1026







39 1669ndash1687







195ndash209







pp 215ndash249






pp 181ndash230







Honours thesis






197ndash212















451ndash461


















1075ndash1088





















2278

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow











Bartlett pp 508


















504ndash505














119ndash126






128 45ndash51





























































109ndash114






187ndash208


2279

Dow



Crystallization from a Zoned Magma Sheet - Oxford Academic

Documents

Transcript of Crystallization from a Zoned Magma Sheet - Oxford Academic