a liquid line of descent of jotunite - hypersthene monzodiorite suite

8/2/2019 A Liquid Line of Descent of Jotunite - Hypersthene Monzodiorite Suite

1/30

JOURNAL OF PETROLOGY VOLUME 39 NUMBER 3 PAGES 439468 1998

A Liquid Line of Descent of the Jotunite

(Hypersthene Monzodiorite) Suite

JACQUELINE VANDER AUWERA1, JOHN LONGHI2 ANDJEAN-CLAIR DUCHESNE1

1L.A. GEOLOGIE, PETROLOGIE ET GEOCHIMIE, UNIVERSITE DE LIEGE, B-4000 LIEGE, BELGIUM

2LAMONTDOHERTY EARTH OBSERVATORY, PALISADES, NY 10964, USA

RECEIVED JANUARY 10, 1997; REVISED TYPESCRIPT ACCEPTED SEPTEMBER 30, 1997

Proterozoic massif anorthosites are usually associated with variable INTRODUCTIONamounts of a characteristic suite of rocks ranging from a melanocratic

Proterozoic massif anorthosites are usually associatedfacies highly enriched in Fe, Ti and P ( FTP rocks) to mafic and with variable amounts of a characteristic suite of maficgranitic rocks (the jotunitecharnockite suite). Here experimental to granitic rocks. The least evolved rocks of this suite areand geochemical data on fine-grained (chilled) samples from several enriched in mafic minerals (low- and high-Ca pyroxenes,intrusions of the Rogaland Province are used to decipher their FeTi oxides, apatite), and in some cases very highpetrogenesis. Modeling of these data supports the hypothesis that concentrations of these phases give rise to melanocraticextensive fractionation of primitive jotunites can produce quartz rocks. Various names including ferrodiorite, mon-mangerites with REE concentrations in the range of jotunites, strong zonorite, jotunite, FeTiP-rich rocks ( FTP) and oxide

depletions in U, Th, Sr and Ti, and smaller to no relative depletions apatite gabbronorite have been used; however, in thisin Hf and Zr. Experimental and petrographic data indicate that study, we will refer to them by the collective term ofthe FTP rocks represent accumulations of a dense oxide jotunite (hypersthene monzodiorite). Evolved rocks of theapatitepigeonite assemblage from coexisting multisaturated jotunitic suite include mangerites (hypersthene monzonite), quartzto mangeritic liquids. The Rogaland jotuniticcharnockitic trend mangerites and charnockites (hypersthene granite). Wecorresponds to a multi-stage process of polybaric fractional crys- will refer to the suite as a whole as the jotunite suite.tallization and crystal accumulation. The early stage, in which a The origin of jotunites remains the subject of con-primitive jotunitic magma fractionates to produce an evolved jotunite, siderable debate, despite their similar textural and geo-probably took place several kilometers below the intrusion level of chemical characteristics from one anorthosite complexdikes, either in mafic chambers similar to that of the Bjerkreim to another. Several hypotheses, not mutually exclusive,Sokndal layered intrusion or in masses of crystallizing andesine have been proposed: (1) jotunites are residual liquidsanorthosite. The later stage of fractionation, which may have after anorthosite crystallization (Ashwal, 1982; Morse,involved flow differentiation, took place within the dikes themselves 1982; Wiebe, 1990; Emslie et al., 1994); (2) they are

and produced compositions ranging from evolved jotunite to mangerite the parental magmas of the andesine anorthosite suiteto quartz mangerite and charnockite. (Duchesne et al., 1974; Duchesne & Demaiffe, 1978;

Demaiffe & Hertogen, 1981); (3) they are products of

partial melting of the lower crust (Duchesne et al., 1985,

1989; Duchesne, 1990); (4) they are transitional rocks in

a comagmatic sequence from anorthosite to mangerite

(Wilmart et al., 1989; Owens et al., 1993; Duchesne &

Wilmart, 1997); (5) they are derived by fractionation of

KEY WORDS: anorthosite; experimental petrology; geochemistry; mon- mafic magmas unrelated to the anorthositic suite (Emslie,

1985); (6) they are immiscible liquids conjugate tozodiorite; Rogaland anorthosite complex

Corresponding author. Telephone:+32 4 3662253. Fax:+32 4

3662921. e-mail: [email protected] Oxford University Press 1998


2/30

JOURNAL OF PETROLOGY VOLUME 39 NUMBER 3 MARCH 1998

mangerites (Philpotts, 1981). In this paper, we present compositional variation can be explained by a processnew experimental data on two jotunite samples from the of fractional crystallization without progressive con-same dike (the Varberg dike) in the Rogaland anorthositic tamination (Wilmart et al., 1989). However, whole-rockcomplex as well as new geochemical data (major and RbSr isotopic data from other dikes such as Lomlandtrace elements) on fine-grained (chilled) jotunitic rocks do not fit tightly to isochrons and there is considerablefrom other intrusions in the Rogaland Province. We then variation in ISr from dike to dike (07040710) thatuse these data as well as published experimental and does not correlate with other geochemical parametersgeochemical data from the literature to: (1) define a liquid (Demaiffe et al., 1986; Duchesne et al., 1989), whichline of descent extending from jotunite up to quartz taken together suggest variable contamination of multiplemangerite (or acidic rocks); (2) discuss the possible origins sources. Jotunites also form small intrusions (e.g. Eiaof rocks showing extreme concentrations of FeO, TiO2 Rekefjord: Fig. 1) as well as chilled margins to theand P2O5; and (3) develop models of major and trace Hidra and Garsaknatt leuconoritic bodies (Demaiffe &element [REE (rare earth elements), Sr, U, Th, Zr, Hf,

Hertogen, 1981) and, locally, to the BjerkreimSokndalTa, Rb, Co, Ni, Cr, Sc] fractionation within the suite.

layered intrusion (Duchesne & Hertogen, 1988; Wilson

et al., 1996). Experiments on a sample from one of these

chilled margins, the Tjorn facies [sample 80123a of

Duchesne & Hertogen (1988)], have shown the near-GEOLOGICAL SETTING ANDliquidus assemblages to be plagioclase (An49)+ olivinePETROGRAPHY(Fo64) at 5 kbar and plagioclase (An47)+ low-Ca pyroxeneThe Rogaland intrusive complex of southern Norway(En66) at 7 kbar (Vander Auwera & Longhi, 1994). The(Fig. 1) (Michot, 1960; Michot & Michot, 1969) is onecompositions of most of the Rogaland jotunitic suite formof the group of Proterozoic anorthositic provinces thatcoherent trends in variation diagrams (Fig. 2), but thehave been recognized world-wide (Ashwal, 1993). Massif-least differentiated compositions (high MgO, low K2O)type anorthosites (EgersundOgna; HalandHelleren;are chilled margin samples and form a group distinctAnaSira), leuconoritic bodies (Hidra, Garsaknatt) andfrom the rest of the dike system. In the following, wea large layered intrusion (BjerkreimSokndal: BKSK)will refer to the distinctive group of chilled margin samplesoccupy most of the surface exposure. Jotunitic rocksas primitive jotunites and to the least differentiatedmainly occur in a system of dikes and small intrusionssamples of the dike trend as evolved jotunites. In this(Duchesne et al., 1985, 1989). There are also several

latter group, most samples are chilled margins to theFeTi ore bodies within the complex (Krause & Pedall,1980; Duchesne, 1998). New UPb ages obtained from dikes (75202F, 8926, 89115, 7355).zircon and baddeleyite (Scharer et al., 1996) suggest that In this study, we took special care to select samplesemplacement of this complex occurred within


3/30

VANDER AUWERA et al. PETROGENESIS OF JOTUNITE SUITE

Fig. 1. Schematicgeological mapof theRogaland anorthositiccomplex [after Michot & Michot (1969)and Bolle (1996)]. EGOG, EgersundOgna;HH, HalandHelleren; AS, AnaSira; H, Hidra; G, Garsaknatt; ER, EiaRekefjord; BKSK, BjerkreimSokndal, Ap, Apophysis. Numbers referto samples described in Table 1.

that is a general feature of massif anorthosites and relatedSAMPLE PREPARATION,rocks (see, e.g. Morse, 1982). However, the graphite

EXPERIMENTAL AND ANALYTICALcapsules imposed a relatively low oxygen fugacity in these

METHODS experiments, probably between FMQ (fayalitemagnetitequartz) 2 and FMQ 4 (Vander AuweraExperiments were carried out on two powdered rocks of

& Longhi, 1994), which inhibits magnetite stability. Tothe Varberg dike at LamontDoherty either in a standard aproximate the f(O2) of the jotunites and to determine1/2 inch piston cylinder apparatus or in a Deltech verticalthe effects of magnetite, which is a late-stage mineral inquenching furnace, following the methods described bythe primitive jotunites, on the liquid line of descent, weFram & Longhi (1992) and Vander Auwera & Longhialso performed a few 1 atm melting experiments in a(1994). One sample was from the chill margin (samplecontrolled COCO2 atmosphere. These were carried out75202F; VB); the other was from the melanocratic faciesat two different f(O2) values: NNO (nickelnickel oxide;(sample 75372; MEL). High-pressure experiments wereruns VB-16 and VB-17) and FMQ (run VB-6); f(O2) wasrun in graphite capsules at 5 kbar (runs VB-1 to VB-5,measured with a Ca-doped ZrO2 electrolyte cell. GoodVB-13 and VB-14 on sample VB, and runs MEL-1 toagreement was found between f(O2) determined directlyMEL-3 on sample MEL), the likely pressure of em-andf(O2) calculated by applying the Andersen & Lindsleyplacement of the Rogaland intrusive complex ( Jansen et(1988) model to the compositions of coexisting ilmeniteal., 1985; Vander Auwera & Longhi, 1994). Dry run

conditions are consistent with the relatively low f(H2O) and spinel produced in the experiments. To minimize

441


4/30


Fig. 2. Major element variation diagrams of the jotunitic suite. Data from fine-grained samples (chills), from the Tellnes dike (Wilmart, 1988;Wilmart et al., 1989) and from other localities [Grenville Province, Quebec: Owens et al. (1993); Laramie: Kolker & Lindsley (1989); Mitchell etal. (1996); Nain: Wiebe (1979); Emslie et al. (1994)] are shown for comparison. P corresponds to the average of four samples from the Puntervollfacies (FTP rocks) of the Lomland dike (Duchesne et al., 1985).

442


5/30


Table 1: Sample description, location and facies of jotunites

Sample UTM grid: zone 32 VLK Intrusion

89115/ch (1) 305859 Kjervall dike crosscutting the EgersundOgna

anorthosite

78211/ch (2) 407825 Eiavatn dike crosscutting the EiaRekefjord

intrusion

8925, 8926/ch (3) 315812 dike on top of the Koldal small intrusion: 8925 is

from the central part of the dike, and 8926 from

the contact

8951/ch (4) 534615 satellite to the Tellnes dike crosscutting the

AnaSira anorthosite

7355/ch (5) 399727 EiaRekefjord intrusion

91141/ch (6), 80123a/ch (7) 273992; 354983 fine-grained margin of the BjerkreimSokndal

intrusion

75182, 7519, 75202F/ch, 75202G, 249883; 249875; 244825; 244825; 244825; Varberg dike crosscutting the EgersundOgna

75 204 , 7 520 6, 7 537 2, 782 01, 244 8 25; 244 83 8; 251 79 7; 2 43 84 1 an ort hos ite

7912(8)

7838/ch, 8034/ch (9) 550624; 554627 Fidsel dike crosscutting the Apophysis

T2/ch,T221/ch,T82/ch, 7832, 7828, 514657; 455701; 517653; 523627; 472698; Tellnes dike crosscutting the AnaSira

7252 (10) 477697 anorthosite

66175, 7536, 7533, 7534 (11) 263803; 267815; 275827; 273829 Puntervold facies (melanorites) of the Lomland

dike

7234/ch (12) 589602 fine-grained margin of the Hidra leuconoritic

body

Numbers in parentheses correspond to the localities shown in Fig. 1; /ch indicates that the sample corresponds to a chill.Duchesne & Hertogen (1988). Wilmart (1988). Duchesne et al. (1974).

iron loss in highly crystalline runs, powdered samples Major element compositions of the experimental phasesare reported in Table 3. Rims and cores of feldspars andwere pressed into disks of 6 mm diameter (bonded with

polyvinyl alcohol) and loosely wrapped in Pt wire of 0254 pyroxenes occurring in the different facies of the Varbergdike have been analyzed with the Cameca SX50 of themm diameter. Run conditions and phase assemblages are

given in Table 2. CAMST (Centre dAnalyse par Microsonde pour lesSciences de la Terre, Louvain-La-Neuve, Belgium; J.Wautier analyst). Standards included natural mineralsand synthetic compounds. Accelerating voltage was set

Analytical method at 15 kV and beam current was 20 nA with 10 s countingtimes. X-ray intensities were reduced using the CamecaAfter each experiment, the charges were mounted in

PAP correction program. Results are given in Table 4.epoxy, polished and analyzed at LamontDoherty under Mass balance between the bulk composition of thethe Camebax/Micro electron microprobe equipped withstarting material and the compositions of all phasesa wavelength dispersive system. Accelerating voltage waspresent in each run has been calculated using a least-set at 15 kV and all elements were measured for 20 s atsquares multiple regression to determine phase pro-a beam current of 25 nA, except in the case of feldspars,portions and to test if Fe and Na loss occurred in the 1phosphates, and glasses, where Na and K were measuredatm experiments. Results are given in Table 2. Na lossfirst for 30 s at 5 nA. Rims of minerals were analyzedranges from 4% up to 14% whereas Fe displays a smallwith a point beam and glasses with a defocused beam ofgain in VB-16 (6%) and a significant one in VB-17 (25%).5 m to minimize alkali loss. X-ray intensities wereThe latter experiment did not equilibrate pervasively, asreduced using the Cameca PAP correction program. Athe temperature was near the solidus and the chargecombination of mineral and glass standards were usedmelted only locally. In run VB-14, the orthopyroxenefor glass analyses whereas only mineral standards were

used for plagioclase, oxides, pyroxenes, and olivines. (opx) coefficient had a negative sign. The opx coefficient

443


6/30


Table 2: Experimental conditions and products

Experiment T (C) P t (h) f(O2) Products

VB-1 1150 5 kbar 25 gl

VB-2 1120 5 kbar 26 gl94 pl6

VB-3 1100 5 kbar 36 gl96 pl4

VB-4 1080 5 kbar 47 gl82 pl17 ol0.2 il0.7 ap0.1

VB-14 1078 5 kbar 66 gl61 pl30 pig6 il3 ap0.01opx 0

VB-13 1074 5 kbar 114 gl pl ol il ap pig

VB-5 1060 5 kbar 71 gl pl pig aug il ksp qtz

VB-6 1050 1 bar 125 FMQ gl50 pl30 ol7 il3 uvsp6 phosph

VB-16 1065 1 bar 46 NNO gl47 pl28 pig4 aug4 uvsp13 phosp4

VB-17 1060 1 bar 89 NNO gl30 pl33 pig4 aug9 uvsp19 phosp5

MEL-1 1130 5 kbar 23 gl

MEL-2 1110 5 kbar 18 gl99 il1

MEL-3 1090 5 kbar 25 gl90 il4 ol4 ap2

The numbers following the phase abbreviations are the weight proportions of the relevant phases present in the experiments,calculated using a weighted least-squares minimization.

was then changed to zero and its components were (An32Ab66Or2). There seems to be a correlation betweenplagioclase composition and bulk composition, as plagio-included in the other phases. This results in an increased

proportion of the remaining phases. clase displaying the lowest anorthite content (An30 insample 75206; Table 4) is associated with the bulkThe fine-grained samples of jotunites were analyzed

for major elements and some trace elements by X-ray composition showing the lowest MgO (279%) and high-est SiO2 (4929%) content observed in the samples fromfluorescence on a CGR Lambda 2020 Spectrometer at

the University of Liege (analyst G. Bologne) (Bologne & the Varberg dike (Table 5; analysis of sample 75204 is

given instead of that of sample 75206). This observationDuchesne, 1991). The other trace elements were analyzedeither on a VG elemental PQ2 Plus inductively coupled is supported by petrographic data from the Tellnes dike,

in which there is a systematic change in the feldsparplasma-mass spectrometer at the University of Liege(Vander Auwera et al., 1998) or by neutron activation at composition associated with bulk composition: jotunites

are characterized by antiperthitic plagioclase and lesserthe Pierre Sue Laboratory (CEA, Saclay, France; analystJ. L. Joron). Major and trace element data are presented K-feldspar (microperthite), mangerites contain more K-

feldspar and mesoperthite-rimmed plagioclase, whereasin Table 5.quartz mangerites display mesoperthite (Wilmart et al.,1989).

EXPERIMENTAL AND ANALYTICAL

RESULTS Experimental results

Mineral compositions At 5 kbar and 1120C plagioclase (An49) is the soleliquidus phase of VB, the chill margin sample. OlivineThe compositions of plagioclase and pyroxenes from theVarberg dike (Fig. 3) have a limited range and are similar (Fo50), ilmenite, and apatite appear approximately to-

gether at 1080C in run VB-4. The SiO2 and P2O5to those observed in jotunites from the Tellnes dike(Wilmart, 1988). In the chill margin plagioclase is not concentrations in the 1080C liquid are 460% and

274%, respectively (Table 3); these values are consistentsignificantly zoned, with cores of An33 and mantles withan average of An34. Augite, whether as lamellae or with the apatite-saturation model of Harrison & Watson

(1984). In run VB-14 at 1078C there is a drastic increaseprimary crystals, has an intermediate mg-number (049).The low-Ca pyroxene (lpyx) is orthopyroxene (Wo12Ens35) in crystallinity and olivine is replaced by pigeonite and

orthopyroxene. At a slightly lower temperature (1074C,inverted from pigeonite (Fig. 3, Table 4). In the sampleof the melanocratic facies ( MEL), the pyroxenes are VB-13) olivine reappears as a stable phase. Only the top

half of the VB-13 charge shows signs of glass and texturalsimilar to those elsewhere in the dike, but alkali feldsparis present (An06Or89) together with plagioclase equilibration. Thus the solidus probably lies within the

444


7/30


Table

3:Com

posit

ionofex

perimen

talpro

ducts

Exp.

No.

Ph

SiO2

TiO2

Al2O3

Cr2O3

Fe2O3

FeO

MgO

MnO

CaO

K2O

Na2O

P2O5

F

Total

An,

Fo,

En

VB-1

4

gl

4726(34)

382(9)

1397(9)

1556(19)

306(8)

025(3)

770(3)

177(3)

328(6)

24(4)

9907

VB-2

4

gl

4762(31)

385(10)

1330(14)

1634(22)

316(3)

023(2)

747(11)

192(3)

295(11)

243(2)

9927

5

pl

5729(60)

015(2)

2752(47)

057(7)

006(1)

000

956(34)

078(5)

561(32)

000

10155

46

VB-3

4

gl

4751(26)

39(2)

1347(11)

1574(14)

316(3)

023(1)

757(5)

189(3)

317(9)

253(5)

9917

3

pl

5630(28)

014(5)

2789(47)

057(7)

007(2)

000

999(79)

062(8)

544(18)

000

10101

49

VB-4

4

gl

4600(52)

424(12)

1157(21)

1826(46)

366(7)

027(2)

780(11)

182(8)

280(1)

274(8)

9916

4

pl

585(14)

005(6)

2680(99)

052(9)

003(2)

000

88(12)

082(21)

588(55)

000

10136

43

5

ol

3418(19)

022(3)

005(1)

001(1)

4144(27)

2338(21)

051(2)

040(3)

000

002(1)

000

10021

50

4

il

005(2)

5593(4)

033(3)

000

000

4042(9)

420(6)

041(2)

019(5)

000

002(1)

000

10155

3

ap

000

000

002(3)

097(23)

018(13)

006(0)

5352(73)

008(6)

006(1)

4114(19)

185(51)

9788

VB-14

7

gl

4580(36)

416(8)

1087(8)

2017(22)

325(7)

032(3)

749(7)

204(9)

269(8)

284(7)

9963

7

pl

5960(72)

002(3)

2581(33)

039(11)

002(1)

000

724(31)

146(51)

565(31)

000

10019

38

8

pig

5093(55)

083(14)

11525)

001(1)

2625(77)

1641(44)

051(4)

419(56)

008(1)

10036

48

5

il

005(2)

5435(27)

032(2)

001(1)

000

4176(17)

339(8)

042(4)

011(5)

001(1)

10042

1

opx

5050

031

080

2946

1568

067

185

006

9933

47

4

ap

000

000

000

094(48)

020(21)

008(2)

5427(51)

003(1)

007(2)

4092(28)

193(55)

9844

VB-13

2

gl1

5024(11)

292(8)

1048(3)

1958(48)

185(4)

033(6)

652(5)

272(4)

284(32)

247(3)

9995

1

gl2

5101

261

1092

1715

159

032

732

320

307

328

10047

1

gl3

5518

233

1179

1558

139

022

512

404

468

173

10206

3

gl4

5050(87)

282(19)

1063(25)

1877(14)

176(15)

033(4)

679(46)

288(28)

291(26)

274(47)

10013

9

pl

5980(57)

006(7)

2559(33)

041(14)

002(2)

000

714(40)

152(46)

570(24)

000

10024

37

7

pig

4984(69)

079(7)

095(19)

001(1)

3089(91)

1270(60)

059(4)

442(62)

008(2)

10027

38

1

ol

3196

046

004

006

5439

1207

075

061

002

10036

28

5

il

008(2)

5317(38)

021(1)

002(2)

000

4363(27)

198(3)

043(5)

017(9)

002

9971

5

ap

000

000

005(3)

124(12)

023(1)

008(4)

5264(38)

006(1)

007(2)

4084(48)

215(20)

9736

VB-5

3

gl

6764(14)

077(11)

1485(20)

001(1)

449(13)

032(9)

004(2)

147(17)

617(77)

370(42)

031(5)

9977

445


8/30


Table

3:cont

inue

d

Exp.

No.

Ph

SiO2

TiO2

Al2O3

Cr2O3

Fe2O3

FeO

MgO

MnO

CaO

K2O

Na2O

P2O5

F

Total

An,

Fo,

En

VB-6

6

gl

5441(64)

254(6)

1218(11)

000

1416(31)

222(4)

025(3)

611(11)

292(15)

300(24)

186(6)

9965

11

pl

5904(13)

007(5)

2622(77)

000

065(20)

004(3)

000

827(92)

105(19)

546(55)

000

10080

43

5

ol

3449(51)

019(4)

013(10)

000

4281(26)

2164(32)

067(4)

052(10)

003(1)

10048

47

4

il

003(1)

469(37)

045(21)

002(2)

1265

3592(23)

326(10)

044(1)

014(6)

9981

6

uvsp

013(1)

2388(25)

223(3)

003(2)

2105

4881(61)

264(7)

047(4)

022(6)

9946

VB-16

7

gl

5669(20)

218(21)

1253(40)

003(2)

1102(12)

234(34)

022(4)

559(66)

283(30)

306(8)

165(32)

9814

12

pl

5712(17)

007(5)

2651(11)

000

056(22)

003(3)

000

89(12)

080(17)

500(58)

000

9897

47

8

pig

5076(43)

077(8)

125(22)

002(1)

199(13)

179(11)

058(3)

75(21)

011(4)

9885

52

1

aug

4957

096

170

005

1745

1550

048

1198

017

9786

46

5

uvsp

016(3)

1976(54)

246(6)

002(1)

2810

4408(53)

299(7)

050(2)

012(7)

001(1)

9820

6

phosp

000

000

011(9)

345(11)

375(4)

018(2)

4490(25)

023(23)

049(17)

4387(22)

000

9698

VB-17

5

gl

6694(73)

124(12)

1397(7)

571(21)

103(7)

010(2)

265(8)

413(20)

348(15)

057(4)

9982

9

pl

583(15)

011(6)

2652(97)

067(27)

004(4)

000

862(91)

092(22)

500(62)

000

10015

44

3

pig

5106(94)

061(21)

097(40)

002(2)

226(15)

173(16)

073(5)

62(25)

009(4)

9959

50

9

aug

5019(43)

088(15)

143(20)

002(1)

178(13)

141(10)

058(12)

141(21)

020(3)

9932

41

5

uvsp

037(25)

1641(31)

232(8)

002(1)

3507

4173(40)

257(12)

061(4)

029(11)

002(1)

9941

4

phosp

000

000

034(16)

359(10)

372(4)

022(2)

4549(30)

024(19)

046(13)

4356(61)

000

9762

MEL-1

4

gl

3659(16)

708(2)

734(4)

2705(8)

451(4)

034(3)

913(8)

062(6)

187(8)

440(7)

9893

MEL-2

5

gl

3695(37)

656(18)

713(15)

2719(39)

461(2)

037(2)

929(7)

057(4)

166(5)

451(8)

9885

4

il

004(1)

5453(25)

031(2)

001(1)

000

4135(31)

366(5)

036(1)

020(6)

000

001(1)

000

10047

MEL-3

6

gl

3956(21)

578(5)

814(8)

2566(28)

399(7)

035(2)

901(7)

061(7)

198(11)

391(8)

9898

5

ol

3319(16)

024(3)

003(1)

005(1)

4613(23)

1936(8)

053(3)

042(2)

000

002

000

9998

43

4

il

003(1)

5538(37)

029(2)

001(1)

000

4168(26)

334(4)

043(3)

015(2)

000

000

000

10131

3

ap

000

000

001(1)

146(40)

024(19)

005(2)

5378(41)

004(2)

004(1)

4121(51)

169(39)

9852

gl,glass;ol,olivine;pig,pigeonite;opx,orthopyroxene;pl,plagioclase;il,

ilmenite;uvsp,ulvospinel;aug,augite;ap,apatite;phosp,whitlockite.

Foreachphase,

theaverageofseveralanalyses

(No.

isthenumberofanalyses)isgivenandthestandarddeviationisinparen

theses.Fo[Mg

100/(Mg+

Fe)],

En[Mg

100/

(Mg+

Fe+

Ca)],

An[Ca

100

/(Ca+

Na+

K)]aregiveninatomicunits.

446


9/30


Table

4:Microprobeanalysesof

the

Varberg

dikefeldsparsan

dpyroxenes

Sample:

75202F

7912

7912

75202G

75182

75182

75372

75372

75206

7820

1

Phase:

Plag

Plag

FK

Plag

Plag

FK

plag

FK

Plag

Plag

No.:

6

6

2

4

8

2

8

4

7

8

Feldspars

SiO2

6106

6089

6488

6138

6134

6433

6116

6490

6181

6090

Al2O3

2462

2478

1829

2448

2443

1840

2463

1843

2416

2466

FeO

021

018

006

015

011

008

012

005

022

010

CaO

696

715

004

665

665

006

667

011

615

701

K2O

035

036

1606

035

036

1522

040

1477

032

037

Na2O

738

728

082

743

747

093

749

114

779

730

Total

10058

10064

10015

10045

10036

9901

10048

9939

10045

10034

Si

27032

26950

29957

27160

27165

29915

27075

29983

27321

27008

Al

12842

12923

09952

12763

12751

10082

12851

10031

12586

12890

Fe

00076

00065

00025

00056

00039

00030

00045

00021

00080

00038

Ca

03300

03393

00019

03153

03153

00030

03166

00056

02914

03329

Na

06333

06246

00738

06372

06416

00836

06431

01023

06673

06273

K

00201

00204

09457

00200

00205

09027

00224

08701

00181

00211

Catsum

49783

49781

50146

49704

49730

49919

49791

49813

49755

49748

%An

3356

3446

019

3242

3227

030

3223

057

2983

3393

%Ab

6440

6346

722

6553

6564

845

6549

1046

6832

6393

%Or

204

208

9260

206

210

9125

228

8897

185

215

447


10/30


Table

4:cont

inued

Sample:

75202F

75202F

7912

7912

75202G

75202G

75202G

75182

75182

75372

75372

75206

75206

78201

78201

78201

Phase:

opx

cpx

opx

cpx

opx

cpx

cpxexs.opx

cpx

opx

cpx

opx

cpx

opx

cpx

cpxexs.

No.:

4

4

4

3

7

3

1

4

3

6

4

4

4

4

2

2

Pyroxenes

SiO2

4984

5125

4969

5145

4961

5123

5098

4973

5134

4974

5113

5039

5192

5002

5114

5150

TiO2

010

014

006

016

010

021

024

010

017

012

024

007

017

011

025

023

Al2O3

059

124

058

129

071

126

133

060

116

067

131

061

119

065

133

134

Cr2O3

004

002

005

005

003

003

000

003

001

004

006

000

001

004

000

001

FeO

3793

1752

3774

1761

3714

1665

1675

3911

1828

3779

1792

3506

1536

3736

1897

1661

MgO

1169

935

1130

902

1198

946

935

1028

820

1103

879

1314

996

1165

917

959

MnO

081

039

095

040

076

041

031

084

036

082

037

092

044

082

046

034

CaO

056

2063

077

2124

080

2134

2178

087

2127

083

1981

064

2071

079

1884

2138

K2O

003

001

000

001

001

001

000

002

001

001

001

000

001

000

000

001

Na2O

001

033

000

034

000

034

034

000

033

000

034

000

037

000

037

031

Total

10161

10090

10114

10158

10114

10093

10107

10157

10113

10105

10022

10083

10025

10144

10054

10132

Si

19754

19676

19800

19653

19702

19627

19540

19840

19756

19826

19798

19849

19863

19805

19735

19631

Ti

00030

00040

00019

00046

00029

00062

00069

00031

00048

00037

00070

00020

00049

00031

00074

00067

Al

00275

00561

00270

00581

00334

00568

00602

00283

00526

00316

00597

00283

00536

00302

00607

00600

Cr

00012

00007

00016

00016

00010

00008

00000

00009

00004

00013

00018

00001

00006

00012

00001

00004

Fe

12571

05625

12575

05626

12334

05335

05368

13047

05882

12598

05803

11547

04914

12369

06126

05293

Mg

06908

05352

06712

05133

07091

05404

05341

06114

04703

06553

05072

07714

05679

06877

05275

05450

Mn

00272

00126

00321

00129

00256

00133

00100

00282

00118

00276

00120

00306

00144

00276

00150

00110

Ca

00237

08486

00327

08692

00341

08759

08943

00372

08770

00353

08220

00269

08489

00333

07785

08731

K

00015

00006

00001

00003

00006

00004

00000

00012

00005

00004

00004

00000

00005

00002

00000

00004

Na

00005

00248

00000

00255

00000

00249

00253

00000

00244

00002

00254

00004

00274

00002

00275

00230

Catsum

40081

40127

40040

40132

40101

40150

40216

39989

40055

39977

39954

39991

39957

40009

40027

40118

%En

3503

2750

3422

2639

3587

2771

2718

3130

2430

3360

2656

3950

2976

3512

2749

2799

%Fs

6376

2889

6412

2893

6240

2736

2732

6680

3039

6459

3039

5913

2575

6318

3191

2718

%Wo

120

4361

167

4469

173

4493

4551

190

4531

181

4305

137

4449

170

4060

4484

mg-no.

036

049

035

048

037

050

050

032

044

034

047

040

054

036

047

051

No.,numberofanalysespersa

mple.

exs.,cpxexsolutioninopx.

448


11/30


12/30


Fig. 3. Anorthite content of plagioclase (a) and pyroxene compositions (b) [En (MgSiO3)Fs (FeSiO3)Di (CaMgSi2O6)Hd (CaFeSi2O6)quadrilateral] in the Varberg dike (Table 4) as well as in liquidus and near-liquidus experiments (VB runs) (Table 3). Numbers in the pyroxenequadrilateral are run numbers.

charge. The few areas of glass are chemically hetero- the array of phase relations in Fig. 4, we believe that thehigh-SiO2 glass in VB-5 is primarily the result of localgeneous (Table 3), and thus represent local surface equi-

librium. Mass balance calculations show that there is a equilibrium in the experimental charge where more maficdomains having a higher solidus temperature remainedprecipitous drop in percent liquid between 1080C (82%)

and 1078C (61%), so encountering the solidus a few entirely crystalline, whereas more felsic domains havinga lower solidus temperature yielded a small amount ofdegrees lower is not surprising; however, in VB-5 at

1060C, which should have been below the solidus, a high-SiO2 melt. Therefore, as with VB-13, even thoughrun VB-5 did not achieve bulk equilibrium, the liquid issmall amount of glass with 676% SiO2 was observedadjacent to K-feldspar. An explanation of these phe- probably multiply saturated nonetheless, so we have used

the VB-5 glass composition to approximate the multiplenomena can be deduced from Fig. 4, in which liquidusphase diagrams are drawn from experimental phase saturation surface of charnockitic (high-SiO2) liquids. A

second curious point is the apparent absence of olivinecompositions. In a plagioclaseilmeniteapatite pro-jection (Fig. 4a), the compositions of liquid, pigeonite, in VB-14, despite the fact that olivine is present in

runs immediately above (VB-4) and below (VB-13) thisand olivine (plus plagioclase, ilmenite, and apatite) inVB-14 are nearly coplanar, implying a thermal divide temperature. Given that the composition of the

olivine+ plagioclase+ pigeonite (+ augite, apatite,on the olivinepigeonite (+ plagioclase, ilmenite, andapatite) liquidus boundary, which in turn promotes ex- and ilmenite) pseudo-invariant point lies close to a line

from the Pl component through the VB composition, ittensive crystallization in a narrow temperature interval.Moreover, both the plagioclaseilmeniteapatite pro- is likely that a small difference in pressure between the

two runs produced a diff

erent crystallization sequence:jection (Fig. 4a) and the wollastoniteilmeniteapatiteprojection (Fig. 4b) indicate that the VB-14 liquid com- lower pressure in VB-4 (ol, no pyx) shifted the pseudo-invariant point away from the Ol component and sta-position is also close to a second thermal divide pla-

gioclasepigeonite (+ augite, apatite, and ilmenite) bilized olivine; whereas higher pressure in VB-14 shiftedthe pseudo-invariant point toward the Ol componentand a eutectic olivine+ plagioclase+ pigeonite

(+ augite+ apatite+ ilmenite). This latter thermal di- and stabilized low-Ca pyroxene. Accordingly, olivineshould be stable at the same pressure as, but at a lowervide restricts SiO2 enrichment at 5 kbar and low f(O2)

and prevents olivine-saturated liquids from ever reaching temperature than VB-14, which is what is observed inVB-13.quartz saturation. Had magnetite been stable or the

pressure lower, it is likely that the eutectic would become At 1 atm and the NNO buffer, pigeonite, augite,plagioclase, and phosphate are stable together near thea peritectic (olivine in reaction) and that liquids would

be able to breach the augitepigeonite thermal divide. solidus with the only FeTi oxide being titanomagnetite(actually ferri-ulvospinel: uvsp58mgt42 in VB-16), whereasGiven the absence of magnetite in the various runs and

450


13/30


Fig. 4. Projections according to Longhi (1991) on the silicaolivinewollastonite plane from plagioclase, ilmenite and apatite and on the

silicaolivineplagioclase plane from wollastonite, ilmenite and apatite. In (a) and (b), 5 kbar experiments are shown; in (c) and (d), 1 barexperiments. In all diagrams except (a), the small black dots correspond to the fine-grained samples and the large stippled points to startingcompositions (VB and MEL). Low-Ca pyroxene (lpyx) composition () are from VB-13 and VB-14 in (a) and (b) and from VB-16 and VB-17in (c) and (d). In (a) and (b), VB-4 and VB-14 glasses () were used to position the ol+ plag+ lpyx boundary and the VB-5 glass providessome constraint on the sil+ lpyx+ plag (+ aug) point. In (c) and (d), the ol+ plag+ lpyx boundary has been constrained with the VB-6 ()and VB-16 () glasses whereas the VB-17 glass fixes the plag+ lpyx+ aug boundary.

at the FMQ buffer, both ilmenite and ferri-ulvospinel but not in the 1 atm experiments, suggesting that thephosphate is probably fluoroapatite in the former and(uvsp69mgt31 in VB-6) precipitate (Table 2). We presume

that olivine would be stable at higher temperatures at whitlockite in the latter. The liquidus boundaries drawnon the basis of the 1 atm experiments are shown in Fig.NNO, as it is at FMQ. Phosphate analyses (Table 3)

indicate that fluorine is present in the 5 kbar experiments 4c and d. They indicate that under moderately oxidizing

451


14/30


low-pressure conditions the thermal divide involving orthopyroxene hosts and augite exsolutions (optical de-termination). The range of calculated compositions ofaugite+ low-Ca pyroxene+ plagioclase is not stablenatural pyroxenes (4859 wt % CaO) is slightly higherand that olivine is in reaction with liquid when low-Cathan the value calculated by Duchesne (1972b) for in-pyroxene is stable. Thus olivine, which is the first mafic

verted pigeonites from BKSK (3747% CaO). Thephase to crystallize in the jotunite compositions at lowvalues indicate a pressure of emplacement between 1pressure, will be replaced by low-Ca pyroxene and pla-atm and 5 kbar (Fig. 5), the latter pressure being consistentgioclase (Fig. 4d) at a peritectic reaction, and furtherwith estimates for the crystallization of the Bjerkreimfractional crystallization will drive the liquid toward silicaSokndal intrusion (Vander Auwera & Longhi, 1994).saturation, and produce SiO2 enrichment. The liquidus

Because the chill margin composition (VB) projectstopology is consistent with petrological variations ob-well into the plagioclase field in Fig. 4b and d, it is clearserved in the Tellnes orebody and its associated dikethat the chill margin is not itself a quenched liquid at(Wilmart et al., 1989) where olivine-bearing ilmenitepressures of 5 kbar or less. In the 5 kbar near-liquidusnorite contains the minerals with the most primitiverun, VB-2, plagioclase has the composition An46, whereascompositions, and other lithologies are olivine free andthe most anorthitic plagioclase in the chill margin samplegrade continuously from jotunite to mangerite to quartzis only An36. So the chill margin is not even a simplemangerite.quenched liquid. In Fig. 4b (5 kbar) the chill margin lies

Titanomagnetite is present together with ilmenite in nearly on a line between the plagioclase component andthe matrix of the Varberg dike; however, both oxidesthe pseudo-invariant point. These relations, together withhave been exsolved and subjected to strong subsolidusthe abrupt simultaneous appearance of olivine, ilmenite,reequilibration (Duchesne, 1972a), so their compositionsand apatite in the crystallization sequence, are consistentprovide no constraint on redox conditions in the dike.with the chilled margin being a multi-saturated liquid,Accordingly, we have estimated redox conditions in thelike VB-4 or -14, enriched with 1530% plagioclase bydike on the basis of experimental FeTi oxide as-weight. It should be noted that plagioclase in run VB-semblages. At 1 atm and the FMQ buffer, two FeTi14 has an average composition of An38, which is close tooxides are stable, ilm86hem14 and uvsp69mgt31, whereas atthat observed in the dike, and which supports a multi-

1 atm and the NNO buffer, only spinel precipitatessaturated liquid composition similar to that of VB-14.

(uvsp58mgt42 in VB-16; uvsp48mgt52 in VB-17). Also, pre-Alternatively, if the jotunite magma was derived by

vious work on a primitive jotunite has shown that FMQfractionation at a higher pressure where multi-saturated

1 marks the low-f(O2) stability limit for spinel in these

liquids sustain higher plagioclase components (Vandercompositions (Vander Auwera & Longhi, 1994). Thus, Auwera & Longhi, 1994), some portion of the apparentinasmuch as both spinel (magnetite) and ilmenite are

excess plagioclase component would be due to de-present in the dike, the f(O2) was probably close to the compression, and the actual liquid composition wouldFMQ buffer during crystallization.

be displaced toward the Pl component, but not as far asNatural and experimental pyroxene compositions have

the chill margin composition (Vander Auwera & Longhi,been plotted in a Al2O3 vs CaO diagram (Fig. 5) in an 1994). So, despite the fine grain size of the chill marginattempt to glean some information about the pressure of

and the absence of phenocrysts, the material flowingcrystallization of the dike. The situation is complicated

along the margins of the dikes contained some plagioclaseby inversion and exsolution in the pyroxene; however,

crystals in suspension. The more evolved compositionsthe primary pyroxene compositions are constrained to

of the fine-grained samples with higher Qtz componentslie on mixing lines between host and lamellae. Both the

do fall close to the plagioclase+ pyroxene cotectics in1 atm and 5 kbar data fall above (higher Al2O3) the Fig. 4b and d, so it is likely that the more evolved fine-mixing lines of the natural pyroxenes in Fig. 5, indicating

grained samples within the dikes closely approximateno simple relationship. Comparison of pyroxene com- liquid compositions.positions in the two 5 kbar runs suggests that the lower

Al2O3 content observed in the natural pyroxenes may

result from continued equilibration to the solidus and intoWhole-rock analysesthe subsolidus. Even so, there is no way to discriminate

between 1 atm and 5 kbar on the basis of Al 2O3 con- The fine-grained samples from the Varberg dike andcentrations. However, the experimental data suggest that various other dikes display high total Fe as FeOt (965CaO decreases systematically with increasing pressure. 1603%),TiO2 (127% up to 462%), K2O (096% up toConsequently, we have calculated the CaO concentration 424%), and P2O5 (071% up to 259%) concentrationsof the primary pigeonite from the compositions of or- together with a modest range in SiO2 (4645% up tothopyroxenes and exsolved augites in samples 78201 6041%) (Table 5, Fig. 2). In variation diagrams (Fig. 2),

the primitive jotunites form a group distinct from theand 75202G (Table 4) and the modal proportions of

452


15/30


Fig. 5. Al2O3 and CaO contents of natural and experimentally obtained pyroxenes. For the experimental compositions, boxes correspond to 1SD. The calculated compositions of natural pyroxenes correspond to their compositions before exsolution (see text for explanation).

jotunites of the dike system, which define trends of are rather constant except for K. Duchesne et al. (1989)previously pointed out this feature and attributed it todecreasing FeOt, TiO2, P2O5 and CaO, and increasing

SiO2 and K2O, with decreasing MgO. Samples of jotu- the variability of the source of jotunites. Apart from theenormous variation in Th, the primitive jotunites havenites from other localities have been included in Fig. 2

for comparison. Among them, two [sample 738: Owens relatively featureless patterns in contrast to the highlyfractionated patterns of the evolved jotunites, mangeriteset al. (1993); sample EC90-216: Emslie et al. (1994)] are

very similar to the primitive jotunites of Rogaland and quartz mangerites. Primitive jotunites show small

depletions of Ta (not Nb) and Hf (but not Zr in all cases)whereas the others fall near or on the trend of the dikesystem. Nevertheless, a group of samples from Nain relative to the adjacent REE, whereas Ti shows a smallexcess as does P. Also, Sr may show a small depletion(Wiebe, 1979) and Grenville (Owens et al., 1993) define

a trend higher in CaO/MgO and lower in K 2O/MgO (80123a, 7234 in Fig. 7), where Eu shows no depletion(Fig. 6), or Sr may show no depletion (91141) where Euthan samples from the other localities. This probably

results mainly from different fractionation paths from shows a small excess. However, all the evolved jotunites,mangerites and quartz mangerites show prominent de-locality to locality (see discussion) and from accumulation

of plagioclase+mafics. This latter process can also ex- pletions in Sr relative to Ce and Nd, yet Eu anomaliesremain small and may even be positive in the quartzplain the dispersion observed in samples from the Laramie

complex (Mitchell et al., 1996) especially in the FeOt/ mangerites. The Sr depletions indicate extensive crys-tallization of plagioclase despite the evidence for pla-MgO, TiO2/MgO and P2O5/MgO diagrams.

La concentrations in the fine-grained samples and gioclase accumulation in some samples. The unexpectedbehavior of Eu is discussed below. Interestingly, P, whichother representative samples range from 15 ppm to 80

ppm (Table 5). The evolved jotunites, mangerites and shows small relative excesses in the primitive jotunites,shows larger excesses in the evolved jotunites and thenquartz mangerites (Fig. 6) are higher in total REE content

than the primitive jotunites, except the quartz mangerite little or no excess or depletion in the mangerites, andfinally prominent depletions in the quartz mangerites.7832 which is in the range of primitive jotunites. The

fine-grained samples display similar light REE (LREE) This pattern signals the onset of apatite crystallization asthe magma changes from jotunitic to mangeritic. Relativeenrichment [average (La/Yb)N = 9] except for one

sample [91141: (La/Yb)N = 4]. Eu anomalies are either depletions of NbTa become more pronounced, andpronounced relative depletions of Ti develop with differ-weakslightly negative or slightly positive (e.g.

91141)or absent. entiation, whereas relative depletions of Hf and Zr dimin-ish in the mangerites and even become slight enrichmentsIn a multielement diagram (Fig. 7), the primitive jotu-

nites display variable concentrations of several trace in the quartz mangerites, which are consistent with Ti-oxide crystallization, but not zircon. Th shows hugeelements, especially for Th, Rb, and REE (see also U in

Table 5), whereas their major element concentrations depletions relative to the LREE in all of the evolved

453


16/30


Fig. 6. REE patterns of the jotunitic suite [7234 (Duchesne et al., 1974); 80123a (Duchesne & Hertogen, 1988); 7252, 7828, T82, 7832 (Wilmartet al., 1989)]. REE abundances normalized to C1 chondrite of Sun & McDonough (1989).

jotunites despite its variability in the primitive jotunites, A LIQUID LINE OF DESCENT OF THEsuggesting that the ThREE relation is a characteristic

JOTUNITE SUITE AND ORIGIN OFof the parental magma,and of the three primitive jotunites

ROCKS WITH EXTREME FeTiPonly 80123a could be parental to the evolved jotunites.Primitive jotunites also show considerable range in K2O CONTENTwith limited variation in MgO (Figs 2 and 8). These

Liquid line of descent (LLD)features suggest variable contamination of the large in-

A series of samples presented in this study display petro-trusions by country rock gneisses during emplacementgraphic features typical of chilled rocks that suggest they(Hoover, 1989; Wilson et al., 1996). For all the samplesare close to liquid compositions and thus constrain athe K/Rb ratio varies from 376 to 1535 and Zr/Hf fromliquid line of descent of the jotunitic suite under dry40 to 58. These trace elements compositions are in theconditions. Nevertheless, projection of the fine-grainedrange of those previously reported by Duchesne (1990)samples in the OlPlQtz diagram (Fig. 4) seems tofor the jotunitic suite of the Rogaland Province and areindicate that some of them are enriched in plagioclase.similar to those reported in other anorthositic complexesWe have already mentioned that the fine-grained samples[e.g. Grenville Province: Emslie et al. (1994); Laramie

Complex: Kolker et al. (1990); Mitchell et al. (1996)]. define in variation diagrams (Fig. 2) two clusters of

454


17/30


Fig. 7. Multielement diagrams of the jotunitic suite [7234 (Duchesne et al., 1974); 80123a (Duchesne & Hertogen, 1988); 7252, 7828, T82,7832 (Wilmart et al., 1989)]. Abundances normalized to chondrites (Thompson, 1982).

points, one with a narrow range of intermediate MgO, Fe2+ ratio imposed by the graphite capsules; and evencorresponding to the primitive jotunites (chilled margins though the 1 atm experiments have been run at the

of large intrusions), and the other, off

set from the first, appropriate f(O2), the lower pressure increases the pro-forming linear trends ranging from evolved jotunites portion of plagioclase and olivine crystallization relativeto quartz mangerites, in which FeOt, TiO2, and CaO to pyroxene. These difficulties turn out to be relativelydecrease, whereas K2O and SiO2 increase, versus de- minor if one compares a combination of the 1 atm andcreasing MgO. These trends are repeated by samples high-pressure (5 and 7 kbar) data with the variationfrom other dikes in the Rogaland system. The com- pattern of the dikes, as illustrated in Fig. 8, where thepositions of samples from the Tellnes dike, for example, shaded areas correspond to the compositional ranges ofoverlap those from the Varberg dike and extend the the Rogaland jotunites shown in Fig. 2.trend to higher SiO2 (charnockite). Thus in Rogaland The most important feature is that the FMQ VB pointthere is an apparent discontinuity between primitive

(run VB-6) lies on or very near the dike trend in all ofjotunite and evolved jotunite, but continuous chemical

the panels (Fig. 8). The track of NNO VB is parallel tovariation from evolved jotunite, through quartz man-

the dike trend, but is offset from the dikes in the directiongerite, to charnockite.

of higher SiO2 and lower FeOt and TiO2, because of the

The experimental data obtained on the two Varberg crystallization of too high a proportion of FeTi-oxides.dike samples plus previous data on the Tjrn jotuniteThe track of 5 kbar VB follows a path of much higher(Vander Auwera & Longhi, 1994), which belongs to theFeOt and lower SiO2 than the dikes, as the reducinggroup of primitive jotunites, bring additional constraintsconditions imposed by the graphite capsules delay theon the LLD. However, some care must be taken becausecrystallization of FeTi oxides, which in turn induces anthe compositions of the experimental liquids vary byexcessive ferrous iron content in the liquid. It thus seemsequilibrium crystallization, whereas the jotunite LLDlikely that crystallization of a liquid with VB-like com-more probably results from a partially fractional crys-position at modest pressures and with f(O2) close to FMQtallization process, and also because the pressureredoxwould produce a track very close to the dike trend, andconditions of the experiments do not match those of thethat following eventual crystallization of titanomagnetitedikes. Delayed crystallization of ilmenite, the absence ofthe track would extend to high SiO2 (charnockitic) con-magnetite near the solidus, and decreased mg-number of

ferromagnesian phases all are effects of the low Fe3+/ centrations.

455


18/30


Fig. 8. Comparison of experimental data (wt % oxides) on VB (75202F), MEL (75372), and TJ (Vander Auwera & Longhi, 1994) with theRogaland jotunitic trend from Fig. 2 (shaded areas).

456


19/30


Comparison of the data on a primitive jotunite (TJ) Origin of the extreme concentrations of Fe,from a previous study (Vander Auwera & Longhi, 1994) Ti and P in some jotuniteswith data on the evolved jotunite (VB) from the present Jotunites characterized by very high concentrations ofstudy provides further constraints on their possible re-

FeO, TiO2 and P2O5 (FTP) have been recognized in alllationship. In all sets of experiments on TJ, TiO2 and anorthosite complexes (Ashwal, 1982; Owens & Dymek,FeOt increase with decreasing temperature until ilmenite 1992; Owens et al., 1993; McLelland et al., 1994). Emsliebecomes a liquidus phase. While plagioclase is the sole et al. (1994) and McLelland et al. (1994) have proposedcrystallizing phase, the initial TiO2 and FeOt tracks show that the unusually high FeTiP content is characteristican increase with decreasing temperature. The effect of of a liquid derived from the processes of anorthositeilmenite saturation is probably best illustrated by the TJ crystallization under reducing conditions. FTP-rich jotu-experiments run at 1 atm (FMQ 1 TJ). In this case, nites from Rogaland have been recognized in the Varbergthe maximum TiO2 concentration (~45 wt %) is reached and Lomland dikes. In the latter, they constitute theat ~4 wt % MgO. As temperature decreases further, Puntervoll facies, which passes progressively along strike

into jotunites, mangerites and quartz mangerites. AnTiO2 and FeOt decrease with MgO, whereas SiO2 in-

average of four analyses of the Puntervoll facies (Duchesnecreases. The tracks of TiO2, FeOt, and SiO2 for FMQ et al., 1985) plots on an extension of the trend of evolved1 TJ eventually all join the trend of the dike compositions:

jotunites at 39 wt % MgO (P in Fig. 2). In the VarbergFeOt and SiO2 at MgO lower than that of sample VB;dike, the transition between the FTP rocks and theTiO2 at MgO higher than VB. Crystallization of TJ atcommon jotunites is not observed in the field but thesehigher pressure will move the tracks of the FeOt andFTP rocks (e.g. MEL with 420% P2O5, 623% TiO2 andSiO2 closer to VB. If titanomagnetite were to crystallize2743% FeOt) plot with higher FeO and TiO2 thanafter ilmenite at 5 kbar, the 5 kbar tracks of both VBsimple extensions of the dike trend (Figs 2 and 8). Theand TJ would move closer to and follow the dike trend.chemical variation toward low-SiO2 compositions is moreP2O5 concentration in the liquid increases with decreasingnearly continuous in anorthosite complexes from thetemperature until phosphate crystallizes in experimentsGrenville Province (Owens & Dymek, 1992; McLellandon both TJ and VB. In the VB experiments, apatiteet al., 1994).begins to crystallize simultaneously with ilmenite and

Variation diagrams combining experimental and geo-olivine (VB-4): this liquid is probably very close to thechemical data (Fig. 8) indicate that the composition ofVarberg liquid composition (see above). The TJ datasample MEL cannot be simply explained by fractionation

also show that the early increase in P2O5 appears more from a primitive jotunite even under very reducing con-pronounced at high pressure. This pressure effect derivesditions [f(O2) in the 5 and 7 kbar TJ experiments liesfrom a greater proportion of pyroxene crystallizing rel-between FMQ 2 and FMQ 4]. Consequently, given

ative to olivine at higher pressure: olivine can incorporatethat field evidence indicates that sample MEL is co-

a small amount of P2O5 [more than low-Ca pyroxene: magmatic with VB, three alternatives are left: the FTP-see Vander Auwera & Longhi (1994)] and Mg decreases

rich jotunites may correspond to immiscible liquids con-more rapidly because of the higher Mg content of olivine.

jugate to the mangerites found in the same dikes (e.g.SiO2 decreases with temperature in TJ liquids at 7 kbar Lomland) or they can represent liquids more or lessbecause of co-crystallization of high-Si plagioclase and

heavily laden with FeTi oxides and apatite or possiblyorthopyroxene and a low proportion of ilmenite; whereas cumulates injected as a crystal mush (Ashwal, 1982).at 5 kbar SiO2 increases weakly because olivine is the The position of sample VB in the jotunitic differ-sole maficsilicate phase near the liquidus. K2O always entiation trend corresponds to the culmination of FeO tincreases and the residual experimental TJ liquids reach and TiO

2

concentrations in the jotunite trend and thusthe K2O content observed in dikes. it is the most likely candidate to plot within the im-

An important feature of Fig. 8 is thus the bridge made miscibility field (Roedder, 1979). However, the ex-between the field of primitive jotunites and the trend of perimental MEL liquid compositions (Fig. 8) trend towardevolved jotunites by liquids residual to TJ. At 57 kbar the array of evolved jotunite compositions with decreasingthe paths of the TJ liquids join the evolved jotunite trend temperature, i.e. SiO2 and K2O increase as MgO, FeO,close to the multi-saturated VB experimental liquids P2O5, and TiO2 decrease. Therefore the VB and MELthat most closely approximate the parental liquid of the compositions are not situated on opposite sides of someVarberg dike. Consequently, the apparent discontinuity immiscibility field. There is also no evidence (globules,between primitive and evolved jotunites shown in Fig. menisci) of two liquids of any kind (silicatesilicate or2 could result from a lack of exposures of this early silicateoxide) in any of the experiments. Plagioclase isfractionation stage. This process probably took place not a near-liquidus phase of the MEL composition,

nor is there any textural or petrographic evidence forbelow the intrusion level of the dikes.

457


20/30


immiscibility in the dike itself. Moreover, if we assume we will present a three-stage model based on previouspetrogenetic studies of other Rogaland dikes (Duchesnethat, in the Varberg dike, the melanorite represented byet al., 1985; Wilmart et al., 1989) as well as of the BKSKsample MEL and the mangeritic composition representedlayered intrusion (Duchesne, 1978). Stage 1 of the modelby the Kungland facies (Duchesne et al., 1985) correspondinvolves fractionating a primitive jotunite, similar toto conjugate immiscible liquids, the partition coefficientsthe parental magma of BKSK (TJ: sample 80123a) toof P, Zr, REE, Ba and Sr are much lower than thoseproduce an evolved jotunite; the second stage involvesmeasured by Watson (1976). We therefore conclude thatfractionating an evolved jotunite, similar to VB to produceimmiscibility is not a relevant process for the formationa mangeritic composition; and the third stage involvesof the FTP rocks, which leaves crystal accumulationfractionating a mangeritic composition to produce quartz(perhaps achieved through flow differentiation) as themangerite.only viable mechanism to produce FTP rocks.

Variation diagrams in Fig. 8 show that subtraction ofGiven that plagioclase is texturally primitivea leuconoritic assemblage from TJ drives the residual(hypidiomorphic) in the Varberg dike and is ubiquitousliquid toward the field of evolved jotunite compositions,in the cumulates (BKSK intrusion), the absence of plagio-close to the chilled margin composition of the Varbergclase near the liquidus of the melanorite MEL is adike (VB). In the different sets of experiments on TJclear indication that the rock has accumulated non-felsic(Vander Auwera & Longhi, 1994), cumulus assemblagesminerals. However, experiments performed on the MELin equilibrium with the liquid closest to VB are: 64%sample show that its liquidus temperature (1110C) isplag+ 30% low-Ca pyroxene+ 6% ilm at 7 kbar, 74%similar to that of the chilled margin (1120C) of theplag+ 20% ol+ 2% pig+ 4% ilm at 5 kbar, and 70%Varberg dike, and is only marginally higher than theplag+ 18% ol+ 12% oxides at 1 atm and FMQ 1.temperature of the likely parental liquid (~1080C; VB-These phase proportions closely match the leuconoritic14), despite its having higher FeOt and lower SiO2. If acumulate deduced by Duchesne (1978) (74% plag+ 16%rock is a chilled suspension of a single phase in a liquid,low-Ca pyroxene+ 10% ilm) from the SrCa modelingthe liquidus temperature of the rock will be higher thanof the leuconoritic stage of BKSK, except that olivine isthe temperature at the time of accumulation and thethe major ferromagnesian phase at 5 kbar and 1 atmliquidus phase is likely to crystallize over a large tem-instead of low-Ca pyroxene. Moreover, the fraction ofperature interval. However, if several phases have ac-liquid is 047 at 5 kbar and 051 at 1 atm, which is closecumulated then there is the possibility of little or noto the value (f= 047) calculated by Duchesne (1978).increase in liquidus temperature as in the case of eutectic

The phase proportions observed in the experimentalaccumulation. Multi-phase accumulation appears to be cumulates correspond to an equilibrium crystallizationthe case for the relatively low liquidus temperature ofprocess whereas those derived for BKSK are based onMEL, and the similarity of the liquidus temperatures ofa fractional crystallization process. Nevertheless, as aVB and MEL is thus partly coincidental. Nevertheless,fractional crystallization process is more relevant for thethe situation is more complex here, as only the non-felsic

jotunitic trend discussed here, we have chosen a cotecticpart of the saturating assemblageilmenite, magnetite,leuconoritic cumulate made of 74% plag+ 16% low-Caapatite, and possibly orthopyroxeneappears to havepyroxene+ 10% ilm with f= 05 for the first stage.accumulated to form the melanocratic facies. In the

To further model the fractional crystallization processprojections (Fig. 4), though, there appears to be a dis-along the jotunitic trend, there is the possibility to useplacement of the MEL composition away from liquidssimulations based either on partition coefficients (Nielsen,saturated with the dikes assemblage not toward or-1990) or on minimization of the Gibbs free energythopyroxene, but toward the olivine component. This(Ghiorso & Sack, 1995). Nevertheless, it has been showndisparity may be attributed to Fe3+ incorporated in the

that these simulations do not predict well the saturationaccumulated FeTi oxides, but treated as Fe2+

in the of FeTi oxides (Toplis & Carroll, 1996). We haveprojections. Subtraction of this Fe will drive the projectedtherefore used mass balance calculations to further modelMEL composition toward Qtz on a line parallel to thethe major element variations in the jotunitic trend. InOlQtz join, thus increasing the proportion of or-the case of fractional crystallization, the mineral phasethopyroxene relative to olivine in the apparent ac-compositions must be in equilibrium with the startingcumulated component.liquid and this will be true for a certain amount ofcrystallization after which a new set of mineral com-positions must be selected. We assume that the parent

Major element modeling magma of the Lomland dike was close to the VarbergTo model the trace element variations in the jotunitic chilled margin (VB) and to the EiaRekefjord chill (7355)trend, we must first constrain the phase proportions and, but we must point out that these compositions are prob-

ably slightly enriched in plagioclase, so that we use parenthence, the major element variation. In the following,

458


21/30


Table 6: Three-stage major element model

Calculated cumulates

Primitive jotuniteC1 = 74% plag (An43) +16% low-Ca pyrx + 10% ilm (f1 = 05, F= 05)

Evolved jotunite

C2 = 433% plag + 196% low-Ca pyrx + 85% high-Ca pyrx + 93% ilm + 113% mgt + 8% ap (f2 = 06, F= 03)

Mangerite

C3 = 469% plag + 113% low-Ca pyrx + 127% high-Ca pyrx + 28% ilm + 211% mgt + 52% ap (f3 = 067, F= 02)

Least-squares fractionation model for cumulate 2 (C2)

Mineral compositions used in fit

Evolved Mangerite Cumulate Plag Opx Cpx Ilm Mgt Apa

jotunite (An40) (mg-no. 056)(mg-no. 066)(Hem2) (Uvsp15)

SiO2 4730 5160 4055 5852 5038 5122 049 174 000

TiO2 355 241 525 000 014 042 4881 484 000

Al2O3 1370 1465 1242 2631 122 212 039 310 000

FeOt 1583 1328 1948 000 2590 1153 4471 7898 000

MgO 320 226 490 000 1862 1244 055 046 000

MnO 025 019 029 000 000 000 000 000 000

CaO 765 621 986 786 074 2034 010 011 5480

Na2O 356 396 295 650 013 065 000 000 000

K2O 170 300 035 080 000 000 000 000 000

P2O5 227 144 340 000 000 000 000 000 4170

r2 = 0086

Least-squares fractionation model for cumulate 3 (C3)

Mineral compositions used in fit

Mangerite Quartz Cumulate Plag Pig VB-16 Cpx Ilm Mgt Apa

mangerite (An30) (mg-no. 062)(mg-no. 067)(Hem0) (Uvsp31)

SiO2 5702 6571 3993 6100 5166 5106 040 122 000

TiO2 192 099 432 000 078 051 5279 1105 000

Al2O3 1416 1331 1233 2470 127 236 010 213 000

FeOt 1198 754 2191 000 2030 1151 4505 7950 000

MgO 171 065 475 000 1820 1304 131 091 000

CaO 471 256 929 610 767 2100 008 016 5679

Na2O 335 307 347 720 011 053 000 000 000

K2O 364 504 047 100 000 000 000 000 000

P2O5 105 051 259 000 000 000 000 000 4321r2 = 0057

For stage 2, the starting composition corresponds to the average of several evolved jotunites including 75202F and 7355whereas the mangerite is the average of the Kungland facies of the Lomland dike (Duchesne et al., 1985).For stage 3, the starting and final compositions are samples 7828 and 7832 of the Tellnes dike, respectively (Wilmart, 1988;Wilmart et al., 1989).

magma instead of liquid. The composition of this liquid gives Fo56 at 5 kbar (1094C). This olivine con-strains the mg-number of the pyroxenes in equilibriumparent magma is given in Table 6 (evolved jotunite).

Calculation of the virtual olivine composition, using the with that liquidand permits selection of the other minerals(plag, ilm and magnetite) in the series of BKSK cumulateFord et al. (1983) relationship, in equilibrium with that

459


22/30


assemblages. Given these compositions, it is possible to of FTP rocks such as MEL indicates that it is possiblecalculate by least-squares fitting the proportions of the to accumulate large amounts of FeTi oxides from theseminerals in the cumulate which subtracted from an magmas.evolved jotunitic liquid close to VB give a mangeriticcomposition close to that of the Lomland dike (Klunglandfacies) which is also close to that of sample 78211 (27%

Constraints from trace elementsMgO). The fitting is excellent (sum of the squared residuesWe have also modeled the abundances of various trace


23/30


Table 7: REE partition coefficients selected for the modeling

plag (1) apa,2 (2) apa,3 (3) opx (4) cpx (5) ilm (6) mgt (1)

La 013 120 145 00019 004 00023 0006

Ce 011 150 211 00035 0075 00019 0006

Pr 01 170 269 00059 0113 00016 0006

Nd 009 190 328 0013 015 00012 0006

Sm 006 200 460 0063 022 00023 0006

Eu 046 130 255 0059 02 00009 0006

Gd 0052 200 439 0069 025 0006 0006

Tb 005 190 394 011 0258 00095 0006

Dy 0048 180 348 015 0267 0013 0006

Ho 0046 168 288 02 0275 0022 0006

Er 0044 155 227 024 0283 0031 0006

Tm 0042 142 191 0315 0292 0044 0006

Yb 004 130 154 039 03 0057 0008

Lu 0038 100 138 047 03 007 0008

1, Demaiffe & Hertogen (1981); 2, see text for explanation; 3, Fujimaki (1986); 4, Dunn & Sen (1994); 5, McKay (1989) for40% Wo; 6, Nakamura et al. (1986); 2 and 3 correspond to the cumulates c2 and c3 of Table 6. Values of D not given in theliterature were extrapolated.

Table 8: Trace elements partition coefficients selected for the modeling

plag1 plag2 plag3 opx ilm apa2 apa3 cpx mgt

Sr 19 (1) 23 (1) 39 (1) 00034 (2) 14 (3) 22 (3) 009 (5)

U 034 (2) 034 (2) 034 (2) 00002 (8) 25 (12) 25 (12) 00009 (13)

Th 004 (6) 004 (6) 004 (6) 00001 (8) 009 23 19 00015 (13) 0025 (19)

Zr 0021 (2) 033 (10) 025 (14) 012 (19)

Hf 001 (7) 001 (7) 001 (7) 0004 (8) 0419 (10) 029 (15) 097 (19)

Ta 0018 (6) 0018 (6) 0018 (6) 0004 (21) 37 003 004

Rb 01 (6) 01 (6) 01 (6) 0025 (4)

Ba 038 (2 & 22) 038 (2 & 22) 038 (2 & 22) 000015 (8) 00023 (16)

Co 005 (6) 005 (6) 005 (6) 07 (8) 9 12 (4) 5 (20)

Ni 95 (8) 17 2 (17) 44 (19)

Cr 003 (7) 003 (7) 003 (7) 1 (8) 16 (11) 27 (18) 350 (11)

Sc 0015 (7) 0015 (7) 0015 (7) 2 (9) 2 (9) 4 (9) 2 (9)

(1) Duchesne (1978) and Vander Auwera et al. (1993); (2) Dunn & Sen (1994); (3) Watson & Green (1981); (4) Henderson(1982); (5) Ray et al. (1983); (6) calculated from Demaiffe & Hertogen (1981) and Duchesne et al. (1974); (7) Phinney &Morrison (1990); (8) Kennedy et al. (1993); (9) Duchesne et al. (1985); (10) McKay et al. (1986); (11) Jensen et al. (1993); (12)J. C. Duchesne, personal communication (1996); (13) Beattie (1993); (14) Johnson & Kinzler (1989); (15) Hart & Dunn (1993);(16) average of values given by Hart & Dunn (1933), Beattie (1993) and Hauri et al. (1994); (17) Steele & Lindstrom (1981);(18) average of values given by Hart & Dunn (1993) and Hauri et al. (1994); (19) Nielsen et al. (1994); (20) Horn et al. (1994);(21) Forsythe et al. (1994); (22) Duchesne & Demaiffe (1978). 1, 2 and 3 are for the three cumulates c1c3 of Table 6. Whenthe D value is not specified in a mineral, it is assumed equal to zero. Partition coefficients in italics have been calculated.

compared with the observed compositions of the different increase in the first stage and then decreases in thesubsequent stages as magnetite crystallizes. The uniformlytypes of rocks in Table 9. Sr, U, Th, Co, Ni and Cr

decrease with fractionation; Zr, Hf, Rb and Ba increase; incompatible nature of Zr and Hf suggests that zircon isnot a liquidus phase in the jotunitic suite. These resultsTa slightly increases; and, finally, Sc displays a slight

461


24/30


Table 9: Comparison between observed and calculated trace element content

Primitive jotunites Evolved jotunites Mangerites Quartz mangerites

80123a Observed Calculated Observed Calculated Observed Calculated Observed

range range range range

Sr 530 382784 400 412465 377 272310 257 128211

U 010 0113 017 032 009 0118 008 0102

Th 050 0538 097 051119 063 059067 063 066087

Zr 26200 155292 51096 89558 82185 717952 119336 12511387

Hf 650 4582 1256 5411 1910 173221 2571 285325

Ta 131 08131 201 122 278 1319 395 094162

Rb 1700 5844 3221 97134 5238 34 7663 4871

Ba 46900 469801 77187 10651602 118261 15331801 164305 18421979

Co 4900 30729 4736 347569 3376 17202 2696 72115

Ni 6000 5360 1288 66161 027 014 000Cr 2800 2850 1628 2266 000 434 000 121

Sc 1380 138259 1910 815281 1769 19211 1620 16177

For each type of rock, the composition calculated using the three-stage major element model (see text for explanation) iscompared with the range observed in the fine-grained samples. In the case of the primitive jotunites, these samples are80123a, 7234, 91141; 7020 (Duchesne et al., 1974); 200/22, 259/11 (Demaiffe & Hertogen, 1981); B90, B93, B95 (Wilson et al.,1996). For the evolved jotunites, the samples are 75202F, 7355, 89115, 8925, 8926 as well as 7252 (Wilmart et al., 1989). Forthe mangerites, the samples are 7838 and 7832 (Wilmart et al. 1989).

are similar to those obtained for the Tellnes dike (Wilmart, of contamination in the various batches of the parentaljotunite magma. Although some additional con-1988; Wilmart et al., 1989). For most elements the agree-tamination during fractionation of the Varberg and Lom-ment between observed and calculated values is veryland dikes cannot be excluded, RbSr isotopic datagood, which supports the major element modelling pro-prohibit any significant contamination of the Tellnes dikeposed above. It should be noted especially that the Srduring fractionation (Wilmart et al., 1989). The sameconcentration decreases with crystallization, whereas theconclusion can also be drawn regarding the K2O evol-LREE show a small overall increase; hence the de-ution. Small variations in the K2O content of the parentalvelopment of a pronounced Sr depletion relative tomagma batches, noted by Duchesne et al. (1989), arethe LREE in the most evolved rocksa clear sign ofamplified by fractionation of a low-K cumulate. Con-plagioclase fractionation. Continuous enrichment of Hftamination of the dikes is curious because most of theand Zr smooths out the depletions of Hf and Zr relativeoutcrop of the jotunitic dikes lies within anorthositic rocksto the middle REE (MREE) in the transition from evolvedwhich are very low in K and Rb. The unsuitability ofjotunite to quartz mangerite as observed in the rocksthe anorthosite as a source of contamination for the dikes(Fig. 7). For U and Th, our calculated values increase insuggests that the dikes intruded the anorthosites alreadystage 1 and then slowly decrease in stages 2 and 3, as

contaminated.observed in the rocks, reflecting the crystallization ofapatite in stages 2 and 3, with high partition coefficientsfor these elements (Duchesne & Wilmart, 1997). Also,

the calculated Rb concentration increases as expectedDISCUSSIONwith differentiation, but, starting with 17 ppm in 80123a,

the concentrations after stages 1 and 2 are higher than Results presented here indicate that quartz mangeritesthose measured in the evolved jotunites and the man- occurring in the vicinity of anorthositic complexes can begerites. When a starting composition lower in Rb is produced by extensive fractionation of primitive jotunites.chosen, as, for example, 580 ppm in primitive jotunite Their compositions will be dependent upon the com-91141, the calculated values in stages 2 (1787 ppm) and position of the parental jotunite, but some generalizations3 (2617 ppm) are lower than the observed ranges. The are possible: such quartz mangerites will be characterizedhighly variable Rb contents of the primitive jotunites by REE concentrations in the range of jotunites with a

weak Eu anomaly that is more positive (or less negative)(Duchesne et al., 1989) probably reflect different degrees

462


25/30


Fig. 9. Calculated REE (a) and trace elements (b) content of the evolved jotunite (L1), the mangerite (L2) and the quartz mangerite (L3) usingsample 80123a as a starting composition and the three cumulates deduced from the major element modeling (in Table 6). Partition coefficientsused in the model are given in Tables 7 and 8.

Fig. 10. Schematic representation of liquidus equilibria of ilmeniteapatitemagnetiteplagioclase saturated liquids projected onto the olivinewollastonitequartz plane. Arrows show direction of decreasing temperature; single arrows indicate cotectics; double arrows indicate reactioncurves. Shaded paths indicate lines of descent consistent with mineral assemblages as explained in text. G, Laramie and Marcy trends; Nj, Nainjotunite (ferrodiorite) trend; Ng, Nain granitoids; R, Rogaland trend (see text); I, hypothetical trend of crustal melt with intermediate composition.

than its parent; and on multi-element plots there will be of evolved compositions in the range of 5262 wt %SiO2, but a continuum in SiO2 from 62 (quartz mangerite)locally strong depletions of U, Th, Sr, and Ti with smaller

to no relative depletions of Hf and Zr. Complementary to 74% (charnockite) in the Nain Complex of Labrador.Not surprisingly, Emslie et al. (1994) ascribed their high-to the quartz mangerite will be a series of cumulates richin oxides and apatite which will form melanocratic rocks. Si compositions to partial melting of the lower crust, not

fractionation. The Nain high-Si compositions have muchRecently, Owens et al. (1993) described a broadly similarscenario for derivation of quartz mangerite from jotunite in common with the Rogaland quartz mangerites in terms

of mineralogy and trace element abundance patterns;in the Grenville Province of Quebec. These observationsdiffer in detail from those of Mitchell et al. (1996) and however, the Nain granitoids show small negative Eu

anomalies and more pronounced relative depletions ofEmslie et al. (1994). Mitchell et al. (1996) described mon-zodiorite (~jotunite) evolving to monzonite (mangerite) Sr and P, which might reflect more extensive fractionation

of plagioclase and apatite, but might also reflect partialand then monzosyenite instead of quartz monzonite(~quartz mangerite) along with the formation of com- melting of a mafic crustal source. Also, the Rogaland

quartz mangerites are found only within fractionatedplementary oxideapatite-rich rocks in the Laramie Com-plex of Wyoming. Emslie et al. (1994) observed an absence dikes or as the upper portion of a much larger jotunitic

463


26/30


intrusion such as BKSK (Duchesne & Wilmart, 1997), Fe has been removed from the Ol component to formwhereas some of the Nain granitoids comprise entire an Fe3O4 (Mgt) component. Liquid lines of descent arediscrete intrusions or substantial fractions of others. Al- represented by shaded lines and FTP trends by patternedthough the difference in field relations between Rogaland areas. Because different suites probably crystallized atand Nain may be due in part to the level of exposure, different pressures the topology of the entire diagramwe cannot exclude the possibility of deriving some quartz may not be appropriate to a single pressure or com-mangerites by partial melting of the lower crust. Indeed, position (with decreasing pressure or increasing mg-num-as discussed below, there is a wide range of pressure and ber the olivine pseudo-eutectic becomes peritectic andcomposition for which fractionation of a jotunitic magma the pyroxeneplagioclase thermal divide disappears).cannot yield quartz mangerite. However, the diagram was constructed such that the local

Our conclusions on the liquid line of descent of jotu- liquidus topologies would be appropriate. The diagramnites, as well as those of Owens et al. (1993) and Mitchell shows that the plagioclase+ low-Ca pyroxene+ augiteet al. (1996), contradict those presented by McLelland et thermal maximum is stable on the aug+ lpyx liquidusal. (1994), who modeled the compositions of PTiFe- boundary and that the Wo-rich portion of the ol+ lpyxrich mafic dikes and sheets from the contact zone of the curve is even, whereas the low-Wo portion is odd andMarcy massif in the Adirondack mountains. In variation truncates the thermal ridge that crosses the lpyx liquidusdiagrams, the trends defined by these rocks show extreme

field. This configuration allows olivine to react out ofenrichments in FeOt (up to 25%), TiO2 (up to 76%), magmas along the R curve and be replaced by pigeoniteP2O5 (36%) correlated with extreme depletion in silica (lpyx); and it also allows the fractionating magma to(down to 36% SiO2). McLelland et al. (1994) admitted eventually reach silica saturationRogaland, some ofthat these rocks are likely to represent crystal-laden liquids the Grenville intrusions (Owens et al., 1993) and parts ofbut nevertheless contended that the liquid line of descent

the Adirondacks (De Waard & Romey, 1968) are ex-followed at least in part a trend of decreasing Si and

amples. The Gcurve represents magmas in which olivineincreasing P, Ti, and Fe. However, comparison with the

crystallizes after pigeonite at the pseudo-eutectic, such asdata from this study indicates that the highest Fe, Ti, P

in the Greaser Intrusion in the Laramie Complex (Mit-and lowest Si concentrations are similar to those of

chell et al., 1996) and the Marcy trend of McLelland etRogaland sample MEL, which is demonstrably a partial

al. (1994). The Nj curve [NainEmslie et al. (1994);cumulate of a multiphase liquidus assemblage (pyroxene,Maloin Ranch plutonKolker & Lindsley (1989)] is an

apatite, ilmenite, magnetite); furthermore, experimentsexample of a trend in which the liquid lies in the

show that FeOt, TiO2, and P2O5 decrease as oxides and pyroxene+ plagioclase thermal divide, so neither olivineapatite crystallize from this composition. In projectionsnor quartz crystallizes, even after extensive fractionation.

such as Fig. 4a and b, the FTP-rich model magmasAs a result, jotunitic (ferrodioritic, monzonoritic) rocks

M6M9 from McLelland et al. (1994) (Table 1) define agrade into two-pyroxene mangerites, and subsequently

trend (not shown) pointing away from the 5 kbar pseudo-into syenites. Such rocks will show a modest increase in

eutectic toward the Ol component; whereas modelSiO2 concentration because the high-Si felsic componentsmagmas M4M6 plot close to the pseudo-eutectic.increase at the expense of the low-Si mafic componentsAs explained above, the highest apparent contents ofin the residual liquids (e.g. Longhi, 1991). Because thethe Ol component may be in part a mixture ofdecrease in the Qtz component is so small along the Gpyroxene and magnetite. The presence of trend, SiO2 will also increase weakly in the residual liquidpyroxene+ oxide+ apatite+ plagioclase in M8 fol-as it progressively forms olivine-free jotunites (ferro-lowed by the initial appearance of olivine in M9 is alsodiorites), olivine-bearing mangerites, and ol syenites. Thusconsistent with a eutectic-like pseudo-inv

a liquid line of descent of jotunite - hypersthene monzodiorite suite

Documents