Karson, J.A., Cannat, M., Miller, D.J., and Elthon, D. (Eds.), 1997Proceedings of the Ocean Drilling Program, Scientific Results, Vol. 153

21. GREENSCHIST FACIES HYDROTHERMAL ALTERATION OF OCEANIC GABBROS:A CASE STUDY OF ELEMENT MOBILITY AND REACTION PATHS1

John M. Fletcher,2 Christopher J. Stephens,3 Erich U. Petersen,4 and Lisa Skerl2

ABSTRACT

Oceanic gabbro from Leg 153 records retrograde metamorphism caused by progressive unroofing in the footwall of a brit-tle-ductile normal fault. Decreasing temperature is reflected by amphiboles that evolved from pargasitic to actinolitic composi-tions. Highest degrees of water-rock interaction are associated with greenschist facies hydrothermal alteration that producedactinolite + chlorite ± talc ± epidote ± prehnite ± pyrite ± quartz ± titanite. Alteration is strongest near microfractures and veinsthat we interpret to be the main fluid conduits.

Chlorite and actinolite make up 50% of highly altered samples. Chlorite typically occurs as zoned coronas around feltedcores of actinolite, and reaction textures show that chlorite predominately replaces Plagioclase and actinolite replaces olivine.Reaction textures and strong linear correlations in the enrichment of secondary chlorite and actinolite and depletion of primaryPlagioclase and olivine indicate that the two reactions operated simultaneously. Based on mass-balance calculations and lack ofsignificant variations in whole-rock composition, we infer that calcium required to transform olivine to actinolite was derivedfrom the alteration of Plagioclase to chlorite; whereas iron and magnesium required to transform Plagioclase to chlorite werederived from the amphibolitization of olivine.

Mass balance of hydrothermal alteration of troctolitic gabbro from Hole 922A shows that the dominant metasomatic changewas the addition of water, approximately 5110 mol/m3. The enrichment patterns of other elements are largely antithetic to pat-terns observed in submarine hydrothermal water relative to seawater. Therefore, alteration could have occurred in the root zoneof such hydrothermal systems.

INTRODUCTION METHODS

Oceanic gabbros recovered in Leg 153 are exposed in the footwallof an east-dipping, brittle-ductile, normal fault, which makes up thewestern wall of the median valley of the Mid-Atlantic Ridge, imme-diately south of the Kane Fracture zone (MARK area) (Karson,1990). Based on the metamorphic evolution documented in this studyand the microstructural evolution documented by Fletcher, Ceule-neer, et al. (this volume), we can demonstrate that the gabbroic rocksin the MARK area record progressive tectonic denudation similar tothat observed in Cordilleran metamorphic core complexes.

The most extensive hydrothermal alteration of these rocks pro-duced the greenschist facies assemblage of actinolite + chlorite ± talc± epidote ± pyrite ± quartz ± titanite. The main goals of this study areto characterize mass transfer of this extensive alteration and to inves-tigate the effects of fluid circulation on oceanic rocks that have occu-pied a wide range of crustal levels at a slow-spreading ridge. The oc-currence of these rocks in a major fault zone suggests that pressuregradients created by the opening of fractures and veins may havebeen an important driving force of fluid flow. Therefore, we comparewater:rock interactions in this setting with other well-documented,thermally driven systems.

'Karson, J.A., Cannat, M., Miller, D.J., and Elthon, D. (Eds.), 1997. Proc. ODP, Sci.Results, 153: College Station, TX (Ocean Drilling Program).

2Departamento de Geologia, Centro de Investigación Cientifica y Education Supe-rior de Ensenada, P.O. Box 434843, San Diego, CA 92143-4843, U.S.A.jfletche@cicese.mx

3 Department of Earth Sciences, The University of Queensland, Brisbane, 4072,Australia. (Present address: School of Geology, Queensland University of Technology, 2George Street, Brisbane, 4000, Australia.)

"Department of Geology and Geophysics, University of Utah, Salt Lake City, UT84112-1183, U.S.A.

Samples were collected mostly from Core 153-922A-2R, whichhad particularly high recovery. It also contains a suite of troctoliticgabbros that show a wide range of degrees of alteration (0%-60%),but only small variations in protolith rock type. This study uses threemain approaches to characterize mass transfer of hydrothermal alter-ation in the troctolitic gabbro. First, compositions and zoning patternsof phases were determined so that representative phase compositionscould be used in system reactions. Special attention was given tophase compositions near disequilibrium reaction boundaries to char-acterize diffusive mass transfer in solid phases. Second, modal min-eralogy was determined for the suite of variably altered troctoliticgabbros to define reaction paths or relative changes in abundance ofprimary and secondary phases. Third, whole-rock compositions weredetermined for the suite of variably altered troctolitic gabbros to iden-tify chemical variations resulting from metasomatism. Based on re-action textures, a set of system reactions was generated using repre-sentative phase compositions. The system reactions were combinedto produce a single reaction that fits changes in relative proportionsof primary and secondary phases, as well as changes in whole-rockcompositions.

Modal mineralogy was determined from microprobe-generatedpoint counts. This routine was largely developed during this studyand a detailed description of it follows. Peak intensities (counts/s)from wavelength dispersive spectrometry were collected for eight el-ements (K, Na, Ca, Mg, Fe, Si, Al, and Ti) over a rectangular grid ofpoints. Each grid was defined to sample the largest possible surfacearea in each sample. The number of points varied from 700 to 2000,and spacing between points was generally between 100 and 200 µm.To minimize analysis time, groups of four elements were measuredin two different passes through the grid. This eliminated the time re-

389

J.M. FLETCHER, C. STEPHENS, E.U. PETERSEN, L. SKERL

quired to change spectrometer positions at each grid point. Countingtimes were 1 s for all elements. A spot size of 1 µm and a beam cur-rent of 30 nA were used. The raw data were systematically sorted andclassified into phases using Microsoft Excel 4.0 by the following pro-cedural steps:

1. Totals for all eight intensities were calculated to sort out anom-alously low analyses that were taken on epoxy or glass.

2. Intensities were divided by molecular weight, and analyseswere normalized to total 100. This converted the analyses torelative molar proportions that closely approximated cation ra-tios expected for different phases.

3. Analyses were classified into phases by sorting the followingmolecular ratios—once classified, the analyses were sequen-tially separated from the remaining unclassified analyses.

a. The ratio (Fe + Ti)/(Si + Mg + K + Na + Ca + Al) was usedto separate opaque minerals from silicates.

b. The ratio Na/(Fe + Mg) was used to separate Plagioclasefrom amphibole, pyroxene, and olivine. Good Plagioclaseanalyses have values greater than the threshold of about 1-2.

c. The ratio Ca/(Fe + Mg) separated calcic Plagioclase fromferromagnesian minerals.

d. The ratio (Mg + Fe)/ (Al + Ca) was used to separate olivine,orthopyroxene, and talc from clinopyroxene, amphibole,and chlorite analyses. The threshold value was about 10.Among analyses with values greater than threshold valuesof 10, the ratio (Mg + Fe)/Si was used to separate olivine(= 2) from orthopyroxene (= 1), from talc (= 0.75). Amonganalyses with values less than threshold values of 10, theratio Ca/(Mg + Si + Al) was used to separate pyroxene (=1/3) from amphibole (= 1/9), from chlorite (= 0).

Analyses with molecular ratios similar to the threshold valueswere labeled unclassifiable and separated from the data set. Theseanalyses are likely to be hybrid analyses taken at grain boundaries. Itis very unlikely that any hybrid analyses were generated by errors inanalyzing the same points in the two different passes through the grid.The automatic stage-positioning system on the microprobe at theUniversity of Utah has been tested by an internal routine and is accu-rate to within 1 µm (R. Lambert, pers. coram., 1995). Additionally, itis unlikely that the population of hybrid analyses are systematicallybiased toward primary or secondary phases. Although individualgrains of secondary phases are smaller and have larger relative pro-portion of grain boundaries for a given area, they generally occur inmonomineralic clusters and pseudomorphically replace primaryphases. Approximately 2%-5% of the analyses were labeled unclas-sifiable, which should not drastically affect the geochemical model-ing in this study. Future work will include classifying these points us-ing a neural network routine.

METAMORPHIC OVERVIEW

Amphiboles from gabbroic rocks of Leg 153 show a nearly con-tinuous range in composition from high-Al pargasitic to low-Al acti-nolitic compositions (Fig. 1A). Similar compositional variations ofamphiboles are observed in gabbroic rocks at other slow-spreadingridge locations (Mével, 1987, 1988; Mével and Cannat, 1991).Changes in composition of amphiboles in the gabbroic rocks from theMARK area correspond to systematic variations in the style of coevaldeformation and degree of disequilibrium between amphibole andpreexisting phases in the rock. The style of coeval deformation wasdetermined through simple crosscutting relationships between miner-als and microfabrics. The degree of disequilibrium was determinedby smoothness of grain contacts and the presence of compositionalzoning, which requires disequilibrium according to Gibbs' phase rule(e.g., Spear, 1993).

390

0.90

0.60 .

0.30 .

0.00 j

/

'/A~\ 1 1

0.0 0.5 1.0 1.5 2.0

B 0.5

0.4 .-

0.3 . .

0.2

0.1! " 3B

0.0 0.5 1.0 1.5

Tetrahedral aluminum

2.0

Figure 1. Compositional variations of amphibole from gabbroic rocks of Leg153; n = 230. A. Tetrahedral aluminum is plotted vs. sodium + potassium. B.Tetrahedral aluminum is plotted vs. titanium.

Important populations of amphibole from gabbroic rocks of Leg153 are defined by textural criteria and compositional variations intetrahedral aluminum, alkalis, and titanium. For descriptive purposes,these populations are referred to by compositional fields shown inFigure 1. Amphiboles from Field 1 are brown and rich in aluminum,alkalis, and titanium (Fig. 1). They are found as individual grains andovergrowths on olivine and pyroxene. Contacts between brown am-phibole and preexisting grains like Plagioclase, pyroxene, and olivineare smooth, curviplanar, and show no textural evidence for disequi-librium. In dynamically recrystallized samples, amphiboles in Field1 are commonly also neoblastic, suggesting that they crystallized be-fore or during ductile deformation. In relatively undeformed samples,brown amphibole is associated with patches of anhedral Plagioclasethat have similar An compositions to euhedral cumulate Plagioclasein the samples. We interpret amphibole from Field 1 to have crystal-lized late from a gabbroic magma that was largely in equilibrium withother primary igneous phases like Plagioclase + clinopyroxene ± or-thopyroxene ± olivine.

Amphiboles in Field 2 are green-brown and have intermediatealuminum, alkali, and titanium compositions (Fig. 1). Green amphi-boles in Field 2 occur as overgrowths on preexisting pyroxene, oliv-ine, and brown amphibole, but they are also found as the outer rim ofzoned green-brown amphiboles or along cleavage planes, fractures,and inclusion trails in pyroxene. Green-brown amphiboles of Field 2are commonly associated with An35-An45 Plagioclase, which occursas neoblast mosaics and zones of alteration along the margins andfractures in preexisting Plagioclase grains with higher anorthite con-tents. Green amphibole is not commonly neoblastic, but it is associ-

GREENSCHIST FACIES HYDROTHERMAL ALTERATION

ated with recrystallization-dominated, semibrittle microfabrics, asdiscussed by Fletcher, Ceuleneer, et al. (this volume). We interpretgreen-brown amphibole in Field 2 and associated andesine to haveformed from highly evolved, silicate melts, and/or from high-temper-ature, aqueous brines that were moderately out of equilibrium withother preexisting phases, like clinopyroxene, olivine, and labradoritePlagioclase.

Amphiboles from Field 3A and 3B are low in alkalis and titanium,but show extreme variability in tetrahedral aluminum content (Fig.1). Amphiboles in Field 3A are colorless, and occur as fibrous matspseudomorphically replacing olivine, feldspar, and pyroxene. Theyare commonly associated with chlorite ± talc ± epidote ± prehnite ±pyrite ± quartz ± titanite, suggesting that alteration occurred at green-schist facies or lower conditions (Spear, 1993). The degree of dise-quilibrium between minerals of the greenschist facies paragenesesand primary igneous phases is extreme, as shown by highly irregularto sutured grain boundaries and by compositional zoning in both theprimary and secondary phases near grain boundaries. This alterationis commonly localized along fractures and veins, and can result in thecomplete replacement of olivine in highly fractured samples.

Amphiboles in Field 3B have similar grain shape and size as thebrown amphibole of Field 1, but they are nearly colorless and havemuch lower titanium contents (Fig. 1). Samples containing amphi-boles from Field 3B always show extensive hydrothermal alterationand contain a significant population of amphiboles from Field 3A.Therefore, we interpret the amphiboles in Field 3B to represent pri-mary amphiboles, which were modified in composition but not de-stroyed during later periods of strong hydrothermal alteration thatalso produced new amphiboles of Field 3A.

In summary, crosscutting relationships and reaction textures de-scribed above show that amphibole compositions evolved from Field1 to Field 3 through time. The compositional trends of decreasing Al,Ti, (Na + K), and increasing Si in amphiboles from Field 1 to 3 sug-gest that temperature and pressure decreased through time (e.g.,Spear, 1993). The hornblende-plagioclase exchange thermometer(Spear, 1980) was not used in this study because Plagioclase was notalways in equilibrium with amphibole, especially Field 3 amphi-boles. However, our proposed retrograde path is confirmed bychange in the coeval deformational fabrics from ductile microstruc-tures associated with Field 1 amphiboles, to brittle and ductile micro-structures associated with Field 2 amphiboles, to purely brittle micro-structures associated with Field 3 amphiboles (see also Fletcher, Ceu-leneer, et al., this volume). This pressure-temperature path andevolution of microstructures is similar to those observed in Cordille-ran metamorphic core complexes (e.g., Bartley et al., 1990; Fletcheret al., 1995). This is consistent with the idea that the gabbroic rocksin the MARK area also record progressive unroofing in the footwallof a brittle-ductile normal fault.

Amphiboles from Fields 3A and 3B are associated with the mostextensive hydrothermal alteration recorded by gabbroic rocks fromLeg 153. In contrast to amphiboles from Fields 1 and 2, which gen-erally make up 0%-4% of the rock, chlorite and actinolitic amphibolefrom Field 3A make up as much as 50% of highly altered samples.The main goal of this study is to characterize element mobility asso-ciated with high degrees of this greenschist facies hydrothermal alter-ation.

PHASE COMPOSITIONS AND REACTIONTEXTURES

Samples of troctolitic gabbro from Hole 922A were chosen forcharacterization of mass transfer because they display a wide rangeof degrees of alteration (0%-60%) with very little variation in pro-tolith composition. Olivine and Plagioclase typically make up 90%-95% of the primary minerals, and clinopyroxene, orthopyroxene, andbrown amphibole make up the remaining 5%-10%. Relict plagio-clase in the sample suite varies in composition from An78 to An71,which is about the same or less than the variation seen in individual

samples (Table 1). Magnesium numbers (Mg/[Mg + Fe + Mn]) of oli-vine also show little variation (0.86-0.81) but olivine compositionscould not be determined in the most highly altered samples where ithad been completely consumed (Table 1). Clinopyroxene and brownamphibole are not present in all samples, but do occur in both weaklyand highly altered samples. Magnesium numbers for clinopyroxeneranged from 0.89 to 0.85 (Table 1). Brown amphibole typically hasedenitic- to pargasitic-hornblende compositions (Table 1). Low vari-ation in primary-phase compositions allow greater consistency inmodeling reaction paths and mass transfer of hydrothermal alteration.

Felted mats of actinolite ± talc ± chlorite surrounded by coronasof chlorite form the dominant reaction texture that characterizes low-grade alteration in the troctolitic gabbro (Fig. 2). Inner contacts ofchloritic halos with actinolitic mats are generally sharp despite thenumerous, irregularly shaped grains that define them. However, outercontacts between chloritic halos and primary Plagioclase are finelysutured, suggesting that chlorite was replacing feldspar (Fig. 2). Al-teration is strongest near chlorite-filled veins. The types of vein-fill-ing minerals are strongly controlled by the host grain that the veincuts. Olivine-hosted veins contain actinolite and talc, whereas feld-spar-hosted veins contain chlorite. In weakly altered samples, veinstypically change from well-defined microstructures with discretewalls in feldspar grains to thicker diffuse zones of alteration in oliv-ine grains. In highly altered samples, chlorite-filled veins typicallycut feldspar and chloritic coronas but never the felted mats of actino-lite. Examples of partially developed actinolitic mats show that thechlorite-actinolite contact follows the primary-grain contact betweenolivine and Plagioclase (Fig. 3). More importantly, the thickness ofthe chloritic halo is directly proportional to the thickness of the adja-cent actinolitic mat. These textures demonstrate that (1) actinolitepreferentially replaced olivine, and (2) chlorite preferentially re-placed Plagioclase. The sympathetic variations in thickness of chlor-itic halos and actinolitic mats suggest that the two reactions couldhave operated simultaneously and supplied necessary reactants toeach site across the inner halo boundary. It is possible that calciumrequired to transform olivine to actinolite was derived from the alter-ation of Plagioclase to chlorite, whereas iron and magnesium re-quired to transform Plagioclase to chlorite could have been derivedfrom the replacement of olivine by amphibole.

The main difference between the spectrum of amphibole compo-sitions from troctolitic gabbro of Hole 922A and that of all gabbroicrocks from Leg 153 is that green-brown amphiboles from Field 2 arenot present (compare Figs. 1 and 4). The subset of amphiboles fromtroctolitic gabbro also shows that Field 3 A of Figure 1 can be dividedinto two distinct populations. Amphiboles from Field 3Ax are muchmore abundant and are only found as matted cores of the coronitic re-placement of olivine described above. Amphiboles in Field 3Ay areonly found in the most highly altered samples, which do not containolivine, and are seen directly replacing Plagioclase. Besides contain-ing more aluminum and alkalis, amphiboles in Field 3Ay have muchlower Mg/(Mg + Fe + Mn) ratios than those from Field 3Ax (Figs.4A, 4C). Plagioclase grains that are partially replaced by Field 3axamphibole typically have albite to oligoclase compositions. There-fore, textural evidence suggests that after olivine was completelyconsumed, amphibole with compositions that plot in Field 3Ay con-tinued to form at the expense of Plagioclase. This reaction does notappear to have operated with any other net-transfer reaction, butcould have operated with exchange reactions that modified composi-tions of existing phases.

Amphiboles in Field 3B are altered primary amphiboles, as previ-ously described. They are depleted in titanium, have lower Mg/(Mg+ Fe + Mn) ratios (Fig. 4C), and generally occur in the more highlyaltered samples. It is possible that Mg from the depletion primary am-phiboles was used in the amphibolitic replacement of Plagioclase,which created amphiboles that plot in Field 3Ay.

Coronitic chlorite displays strong and nearly continuous compo-sitional zoning (Fig. 5). Along the outer contact with Plagioclase,chlorite is higher in aluminum and has lower Mg/(Mg + Fe + Mn) ra-tios than chlorite along the inner contact with the actinolitic cores.

391

J.M. FLETCHER, C. STEPHENS, E.U. PETERSEN, L. SKERL

Table 1. Representative phase compositions for suite of variably altered troctolite.

Core, section,interval (cm)

153-922A-1R-1,45-522R-1, 5-112R-1,65-712R-3, 25-312R-3, 71-772R-4, 20-262R-4, 100-1062R-1, 41-482R-1,65-712R-3, 25-312R-4, 20-262R-4, 100-1062R-1, 41-48IR-1,45-522R-1, 5-112R-1, 41-482R-1, 41-482R-1, 41-482R-1, 65-712R-3, 25-312R-3, 71-772R-4, 20-262R-4, 100-1062R-4, 20-262R-1.5-112R-4, 100-106IR-1,45-52

153-921E-8R-1, 91-958R-1, 91-958R-1, 91-958R-2, 45-51

153-922A-2R-1,65-712R-3, 71-772R-4, 100-1062R-4, 20-262R-3, 71-772R-3, 71-772R-3, 71-772R-3, 71-772R-3, 71-77IR-1,45-522R-1,65-712R-1, 1-8IR-1,45-52IR-1,45-52IR-1,45-522R-1,65-71IR-1,45-52

Fieure 2. Comn

Phase

ClinopyroxeneClinopyroxeneClinopyroxeneClinopyroxeneClinopyroxeneClinopyroxeneClinopyroxeneOlivineOlivineOlivineOlivineOlivineOrthopyroxenePlagioclasePlagioclasePlagioclasePlagioclasePlagioclasePlagioclasePlagioclasePlagioclasePlagioclasePlagioclaseAmphibole Field 1Amphibole Field 1Amphibole Field 1Amphibole Field 1

Amphibole Field 2Amphibole Field 2Amphibole Field 2Amphibole Field 2

Amphibole Field 3AxAmphibole Field 3AxAmphibole Field 3AxAmphibole Field 3AxAmphibole Field 3AxAmphibole Field 3AyAmphibole Field 3AyAmphibole Field 3AyAmphibole Field 3BAmphibole Field 3BAmphibole Field 3BChloriteChloriteChloriteChloriteChloriteChlorite

[ete amnhibolitizatioi

Na

0.280.360.200.350.310.350.350.010.000.010.000.010.002.052.002.452.221.922.202.092.302.351.891.671.821.972.20

0.911.21.6

0.10.20.20.30.21.21.31.42.11.92.3000000

i of ol

Ca

15.9516.1315.9415.6615.9914.8215.680.010.030.050.020.010.81

10.9111.1410.3010.7811.3410.8311.2310.3710.4011.418.378.748.218.02

7.729.227.668.03

9.248.868.69.168.938.378.128.348.718.528.720.020.040.050.040.020.05

ivine ai

Mg

10.109.96

10.4310.0110.0710.279.89

27.5627.3427.4326.5027.3919.150.010.000.030.010.020.040.000.010.000.01

10.6110.1110.769.79

7.789.227.58

10

14.113.813.813.114.19.929.99.717.727.987.83

15.915.915.816.11616

id

Fe

3.042.923.052.832.943.793.52

11.0911.4411.4714.1913.007.110.150.140.110.140.110.080.110.120.170.104.825.274.565.56

13.810.213.68.13

2.2833.833.82.568.538.568.67

10.98.849.089.078.738.588.738.688.59

• * ' " .

V

Si

24.0724.4924.9224.4024.2724.3324.1718.8119.0618.8618.6118.7625.6923.1323.0223.3623.2822.9423.2922.8923.2323.5822.4121.1920.9120.8120.12

23.01822.94522.09621.854

27.57326.48426.52926.38526.4923.35123.25722.72620.32320.17720.06213.84913.94713.93513.87813.96613.745

Representative analyses of primary phases

Al

1.461.651.211.641.671.601.610.000.000.000.010.000.61

16.6416.5716.1916.3316.7916.3416.7016.8316.7217.515.405.916.076.24

2.5892.7493.6234.399

0.0750.2240.2750.710.8853.8393.9394.3826.857.7857.7

10.31410.3710.31310.12610.31510.527

Ti

0.640.660.360.720.760.630.630.000.000.000.050.000.300.030.020.060.060.010.010.030.040.070.052.141.691.622.64

0.60.711.3

0000.100.10.10.200.10.1000000

K

0.000.010.010.020.000.000.000.000.000.020.000.010.000.040.030.060.020.020.050.040.020.040.050.230.170.280.36

0.200.20.1

000000.100.10.10.10.1000000

F

0.030.000.000.020.190.010.000.000.000.000.000.000.000.000.000.000.000.020.000.080.040.020.000.000.000.240.17

000.10.2

0.100.10.100.10000000.10000

Cl

0.020.010.020.000.010.000.010.000.030.030.000.000.000.010.000.080.040.000.020.010.000.000.000.030.010.010.01

0000

0000000.10.10.20.10.7000000

• • : ; • V : , • V " . . . ;

Mn

0.050.100.100.100.120.110.130.130.160.190.230.200.160.010.040.000.000.030.000.020.020.000.020.090.060.050.08

0.20.20.20.1

0.20.10.20.10.10.10.20.10.20.10.20.20.10.20.20.10.2

• 1

o

43.2343.8743.9943.6243.6543.4843.3342.8543.1142.9742.8343.1945,1146.3746.2546.1646.2946.4046.2946.2846.4646.8246.4142.7342.6142.6142.39

41.542.241.542.5

45.343.944.244.344.642.942.842.741.841.941.938.138.2383838.238.1

Total

98.88100.14100.2399.3899.9999.3999.32

100.46101.18101.03102.45102.5798.9599.3499.2098.8199.1899.6099.1499.4899.44

100.1799.8697.2797.3097.1997.58

98.36398.62298.78898.304

98.87896.55997.71598.03797.87298.40798.20898.37598.87297.51898.61487.9587.87587.31687.55687.81287.766

* *

'"'Mur&>i<k

An

75767174777476727278

Mg#

0.880.880.880.890.880.860.860.850.840.840.810.830.86

TetAl

1.501.561.601.77

0.730.871.031.26

0.000.100.130.190.220.860.881.021.621.691.72

1

Ti

0.390.310.290.48

0.110.130.180.24

0.000.000.000.010.000.010.010.030.010.010.01

1

Na+K

0.670.720.800.91

0.400.400.520.64

0.020.060.070.110.070.450.490.550.820.760.89

SS|

extensive development of chloritic halo. Contact between

chlorite and fibrous actinolite is the former primary con-

tact between olivine and Plagioclase. The contact is

remarkably straight, suggesting that chlorite and actino-

lite are in equilibrium. In contrast, the contact between

colorless feldspar and chlorite is sutured, evidence of dis-

equilibrium and replacement of feldspar by chlorite.

Sample 153-922A-2R-4, 20-26 cm. Field length is 2

mm. Core axis is parallel to the short axis of photograph,

and the uphole direction is toward the top of the photo-

graph.

«? ": S*

.mmJKs

392

GREENSCHIST FACIES HYDROTHERMAL ALTERATION

Figure 3. Incipient amphibolitization and developmentof achloritic halo. A. Plane-polarized light. B. Cross-polarized light. Olivine grains are more stronglyreplaced by actinolite near the blunt terminations of thegrains. The chloritic halo is thickest in areas adjacent tothe portions of olivine grains that are most stronglyreplaced by amphibole. Sample 153-922A-3R-1, 59-63cm. Field length is 4 mm. Core axis is parallel to theshort axis of photograph, and the uphole direction istoward the top of the photograph.

MODAL MINERALOGY AND REACTION PATHS

Troctolitic gabbro from Hole 922A shows systematic variationsin modal mineralogy with degree of alteration (Table 2). Degree ofalteration is best defined by the abundance of chlorite and actinolite,which form the dominant products of alteration. Talc is not present inlarge quantities in any sample, but seems to be more abundant inweakly altered samples. As indicated by reaction textures, olivineand Plagioclase are the dominant reactant phases, and the abundanceof both decreases with higher degrees of alteration. In fact, olivine iscompletely absent in the highly altered samples. In contrast, pyrox-ene and Field 1 amphibole show no systematic variation in modalabundance with degree of alteration. Accessory phases include cal-cite, apatite, quartz, pyrite, magnetite, ilmenite, anthophyllite, andpyrophyllite.

Amphibole populations not only show strong increases in abun-dance but also systematic changes in composition with degree of al-teration. In weakly altered samples, normalized histograms of the tet-rahedral aluminum content are bimodal, and show a strong peak onpargasitic hornblende compositions (Fig. 6A). Amphibole composi-tions are also bimodal in moderately altered samples, but actinoliticcompositions dominate (Fig. 6B). In highly altered samples, the com-position histograms have a single peak on actinolitic compositions(Fig. 6C).

Changes in the molecular abundance of the dominant primary andsecondary phases define reaction paths of progressive alteration (Fig.7). Textural evidence outlined above indicates that actinolite predom-inately replaced olivine, and chlorite predominately replaced plagio-clase. However, the strong linear correlation between the molecularabundances of feldspar and olivine and of actinolite and chlorite sug-gests that the two reactions operated simultaneously in weakly andmoderately altered samples (Fig. 7). In the more advanced stages ofalteration, we interpret the reaction paths to curve (Fig. 7). Texturalevidence suggests that, after olivine is completely consumed, am-phibole with compositions that plot in Field 3Ay (Fig. 4) were pro-duced at the expense of Plagioclase. The only other reactions thatcould have accompanied this late replacement of Plagioclase by am-phibole are exchange reactions that depleted Mg content in the exist-ing amphibole population.

WHOLE-ROCK GEOCHEMISTRY

Major-element compositions of samples from the suite of variablyaltered troctolite show very little systematic variations with degree ofalteration (Table 3). The main chemical change that can be confident-ly attributed to alteration is the strong increase in material lost on ig-nition, which we interpret to be predominately H2O. Silicon, magne-

393

J.M. FLETCHER, C. STEPHENS, E.U. PETERSEN, L. SKERL

0.9 ••

0.3 ••

o.o

B 0.5

0.4 •

0.3 •

0.2 .

0.1 .

0.0 |

3Ax •

m

3Ay

•_ •

1

i

• ••*

If•

' 3B

0.00 0.5 1.0 1.5

Tetrahedral Aluminum

Figure 4. Compositional variations in amphibole from troctolitic gabbro ofHole 922A; n = 112. A. Tetrahedral aluminum is plotted vs. sodium + potas-sium. B. Tetrahedral aluminum is plotted vs. titanium. C. Tetrahedral alumi-num is plotted vs. the ratio of magnesium to (magnesium + iron + manga-nese). Amphiboles in Field 1 are primary. Amphiboles in Field 3Ax are sec-ondary and replace olivine. Amphiboles in Field 3Ay are secondary andreplace feldspar. Amphiboles in Field 3B are altered primary amphiboles.

sium, calcium, and sodium show weak depletions in highly alteredsamples. Iron shows a slightly stronger depletion in the highly alteredsamples (Table 3).

MASS BALANCE OF HYDROTHERMALALTERATION

To define element mobility of the hydrothermal alteration, bal-anced reactions must be calculated that use representative composi-

c

CUJ _+ 0.8 -

>

>

0.6 -

•

outer rim ofchlorite halos

1 1

• p • mjfm ma ••

" • T p •• • •

^ • •inner rim ofchlorite halos

H 1 1 1

Aluminum

Figure 5. Compositional variation in chlorite. Chloride halos are composi-tionally zoned and contain more Mg-rich chlorite along the inner rim.

tions of the primary phases and reproduce relative changes in (1)modal abundance of primary and secondary phases and (2) whole-rock geochemistry. Following the terminology of Thompson (1982)and Spear (1993), the chemical system of the altered troctolite is de-fined in terms of phase components and system components. Thedominant mineral-phase components of the altered troctolite systeminclude Plagioclase, olivine, chlorite, and actinolite. These phases arecomposed of seven main system components: NaO1/2, CaO, MgO,FeO, SiO2, A103/2, and H2O. Other oxide components, such as KO1/2,TiO2, and MnO, occur in minor quantities and are ignored in thechemical system. To investigate all possibilities of element mobility,the chemical system is assumed to be open to all system components.Therefore, each of the seven system components must also be phasecomponents as listed in Table 4. Mass transfer in any system with 11phase components and 7 system components is completely describedby combinations of 4 linearly independent reactions. However, thecombinatorial formula shows that there are 165 possible linear com-binations of these 4 reactions (e.g., Spear and Peacock, 1989).

Table 5 shows some of the possible reactions that convert plagio-clase to chlorite, and olivine to actinolite. Reactions 1, 2, and 3 arebalanced to conserve Mg, Fe, and Si, respectively, in the reaction ofolivine to actinolite. Reactions 4, 5, 6, and 7 are balanced to conserveSi, Al, Ca, and Na, respectively, in the reaction of feldspar to chlorite.Reaction 8 is a linear combination of reactions 3 and 5 that fits thenumber of moles of feldspar and olivine destroyed and the number ofmoles of chlorite and actinolite created in 10,000 cm3 of highly al-tered troctolite. Approximately 5 mol of actinolite and chlorite arecreated before the reaction curve flattens (Fig. 7B), and approximate-ly 40 mol of olivine are consumed for every 15 mol of feldspar (Fig.7A).

The stoichiometry of reaction 8 also crudely fits the observedchanges in major-element composition in the suite of variably alteredtroctolite (Table 6). Observed changes were calculated simply bysubtracting the average bulk composition of the four highly alteredsamples from that of the three weakly altered samples (Table 3).Modeled changes in the oxide weight abundances were calculated as-suming that major elements were transported in and out of the systemin proportions defined by the stoichiometry of reaction 8 (Table 4).In these calculations, it was also necessary to assume that the densityof highly altered troctolite is 2.85 g/cm3 (e.g., Shipboard ScientificParty, 1995). The signs of the observed and modeled changes in ox-ides are the same for all major elements except MgO (Table 6). Ad-ditionally, the magnitude of observed changes fits very well the mag-nitude of modeled changes in NaO, CaO, A12O3, and H2O. The mostsignificant differences between observed and modeled changes arefound in FeO, MgO, and SiO2. Factors that are important in produc-ing these discrepancies are discussed in the following section.

394

GREENSCHIST FACIES HYDROTHERMAL ALTERATION

Table 2. Modal mineralogy determined from microprobe-generated point counts.

Number of Modal abundance (%) of mineral phasesCore, sample,interval (cm)

153-922A-2R-3, 71-771R-1, 45-522R-4, 87-922R-4, 100-1062R-1, 41-482R-2, 46-522R-1,65-712R-2, 118-1222R-4, 20-26

classifiablepoints

1222876

104810061863917

10471128963

Olivine

0π10

1113171626

Plagioclase

535348466262636765

Clinopyroxene

002513121

Orthopyroxene

1060

3

2

Hornblende Actinolite Chlorite

234

1012140

1918161276

2

24262226149952

Talc

000002010

Accessory

101121102

40 -

30 -

20 -

:

Weak Alteration ^m

1

πri-J

10

0.2 0.6 1.0 1.4 1.8

Tetrahedral aluminum factor

Figure 6. Compositional variability among amphibole populations. A. Weakalteration, Sample 153-922A-2R-2, 118-122 cm. B. Moderate alteration,Sample 153-922A-2R-1, 65-71 cm. C. Strong alteration, Sample 153-922A-2R-3, 71-77 cm. Tetrahedral aluminum compositions were determined bycalculating structural formulas from the point-counting analyses described inthe "Methods" section, this chapter.

45 55

Feldspar (100 moles/m3)

0 2 4 6

Actinolite (100 moles/m3)

Figure 7. Reaction paths for (A) primary and (B) secondary phases. Molecu-lar abundance was determined by dividing modal abundance by molar vol-ume of the magnesium or calcium end members of each phase. Molarvolumes are listed in Robie and Waldbaum (1968).

DISCUSSION

Greenschist facies hydrothermal alteration of troctolitic rocksfrom Site 922 is dominated by two main reactions: Plagioclase tochlorite, and olivine to actinolite. Although we cannot completelyrule out the possibility that the rocks were affected by two completelydifferent episodes of alteration, several lines of evidence suggest thatboth reactions occurred simultaneously in one alteration event. Thetwo dominant alteration products, chlorite and actinolite, are alwaysspatially associated in the rock, and form coronitic replacement tex-tures. A strong linear correlation is observed in the relative abun-dance of actinolite and chlorite from different, variably altered sam-ples (Fig. 7). Additionally, this correlation is observed at a much

395

J.M. FLETCHER, C. STEPHENS, E.U. PETERSEN, L. SKERL

Table 3. Major-element compositions of variably altered troctolite.

Core, section,

interval (cm)

153-922A-2R-3, 71-77IR-1,45-522R-4,100-1062R-1, 1-82R-1, 41-482R-2, 4 6 - 5 22R-1,65-712R-2,118-1222R-4, 20-262R-3, 25-31

Alteration

HighHighHighHighModerateModerateModerateLowLowLow

SiO 2

45.7944.3444.8444.7146.8045.5844.9246.1746.3045.98

TiO2

0.190.120.220.100.160.150.050.270.460.17

A12O3

20.1220.7718.3319.8120.4419.7519.6219.8619.9019.31

Fe 2 O 3

4.625.145.784.994.485.255.455.766.625.83

Major element composition (wt%)

MnO

0.080.100.080.100.060.080.080.090.100.09

MgO

12.5913.0914.0313.9611.6014.4514.8913.6811.7615.34

CaO

10.8410.4410.7611.0211.8510.7910.8011.0411.6410.43

Na 2O

1.691.76.38.41.88.77.56

2.00

1.78

K2O

0.070.040.050.050.080.050.060.050.040.07

P 2 O 5

0.050.020.090.020.020.020.020.080.040.04

LOI

3.763.76

4.302.491.832.181.030.950.73

Total

99.8099.5899.54

100.4799.8699.7299.63

100.0399.8999.77

Note: LOI = loss on ignition.

Table 4. Phase components and system components of altered troctolite.

Phase components

OlivinePlagioclaseActinoliteChloriteNaO I / 2

CaOMgOFeOSiO 2

AlO3 / 2

H2O

NaO 1 / 2

0.000.250.070.011000000

CaO

0.000.751.840.01

0100000

System components

MgO

1.680.004.777.740010000

FeO

0.310.010.381.8200010

0

SiO 2

1.002.287.785.840000100

A1OV 2

0.001.710.274.490000010

H2O

001

0000001

smaller scale around partially replaced olivine grains (Fig. 3). Mostimportantly, a single balanced reaction reproduces reasonably wellthe changes in whole-rock geochemistry and modal mineralogy invariably altered samples.

The variation in bulk composition in the suite of samples is inter-preted to predominantly reflect (1) analytical error of the chemicalanalyses, (2) magmatic evolution of the protolith, and, of course, (3)element fluxes of hydrothermal alteration. The observed changes inbulk composition are small and near the detectable limits of the anal-yses. However, the coincidence of observed and modeled changes ofmost of the major elements that were modeled in this study suggeststhat these changes could be real variations. As presented above, sim-ilarities in primary igneous grain size, texture, and phase compositionsuggest that effects of the latter were minimized, which was an objec-tive of the sampling strategy. However, we cannot rule out the possi-bility that some of the differences between observed and modeledchanges in composition (Table 6) are the result of magmatic evolu-tion of the protolith. Because the modeled changes are only based onthe stoichiometry of Reaction 8, other alteration reactions could alsobe responsible for discrepancies between observed and modeledchanges. For example, the breakdown of titanomagnetite to titaniteand rutile is observed in other hydrothermally altered oceanic rocks(e.g., Laverne et al., 1995) and could have produce the additional ob-served depletion of FeO (Table 6). The presence of titanite in the al-tered troctolites of this study supports this interpretation.

The pattern of major-element enrichment of hydrothermal waterrelative to seawater is largely antithetic to the relative enrichments re-corded by altered troctolitic gabbros in this study (e.g., Campbell etal. 1988; Campbell and Edmund, 1989; Von Damm, 1990). Estima-tions of the magnitude of relative enrichments of some of the majorelements in hydrothermal water discharged in the TAG and MARKareas are 1.9 mmol kg~1, 19.3 mmol kg'1, 70.3 mmol kg"1, 5.3 mmolkg"1, and -52.7 mmol kg"1 for Fe, Si, Na, Ca, and Mg, respectively(Von Damm, 1990). Aluminum is typically elevated in hydrothermalwater, but magnitudes were not reported by Von Damm (1990). Of

the element enrichments modeled in the altered troctolite, aluminumis the only one that is not antithetic to enrichments in hydrothermalwater. Therefore, it is likely that the observed hydrothermal alterationof troctolitic rocks in this study occurred in the root zone of such hy-drothermal vents.

Wateπrock ratios can be calculated using the above enrichmentfactors for hydrothermal water and altered troctolite (Table 7). Likeall calculations of relative volumes of wateπrock interactions, as-sumptions must be made about the flow path and composition of hy-drothermal fluids. One-dimensional drill cores contain limited infor-mation about the larger-scale three-dimensional character of hydro-thermal circulation. Therefore, wateπrock ratios were calculatedassuming the simplest possible path that could have affected theserocks: the direct inflow of seawater and outflow of hydrothermal wa-ter (Table 7). Seawater and hydrothermal water compositions wereobtained from in situ measurements by Campbell et al. (1988) andCampbell and Edmund (1989), which are summarized by Von Damm(1990). It is assumed that the difference in composition of the twotypes of water is produced by ion exchange with the troctolite duringhydrothermal alteration.

The modeled reaction 8 requires element mobility because the co-efficients of nearly all oxide-phase components are nonzero (Table5). Assuming the simple flow path and exchange scenario describedabove, wateπrock ratios vary from 0 to 36 for samples in which reac-tion 8 has gone to completion (Table 7). These ratios are within therange of values calculated for other altered portions of ophiolites andin situ oceanic crust (e.g., Gregory and Taylor, 1981; Stakes andO'Neil, 1982; Mottl, 1983; Lécuyer et al, 1990). It is also possible tocalculate wateπrock ratios assuming that all of the measured changesin bulk composition are caused by a similar exchange with hydrother-mal fluids (Table 7). These ratios are much more highly variable, andmay erroneously reflect other factors that contribute to the observedchemical variation, such as analytical error and primary magmaticvariation. We argue below that the highest estimates, based on rela-tive depletions of Fe, Si, and Ca, could be significantly lower if morecomplex circulation paths are considered. Therefore, we considerwateπrock ratios <IO to be more realistic.

Based on the metamorphic grade and the pattern of element en-richments in the troctolite, we infer that it was altered in the deeperlevels of the hydrothermal system. At shallow levels (<l km) and re-charge zones, oceanic rocks commonly show a gain of Na, and Mg,and a loss of Ca (e.g., Alt et al., 1986; Harper et al., 1988; Lécuyer etal., 1990). Oceanic rocks in discharge zones, however, are commonlyepidotized, and show a gain of Ca, ± Fe, ± Al, and a loss of Mg (e.g.,Harper et al., 1988; Nehlig, 1994). Neither of these patterns fit the al-teration observed in the troctolites. At about the base of the sheeteddike complex in the Trinity ophiolite, Lécuyer et al. (1990) documentan important reversal in the enrichment pattern of Al and Si. Belowthis level, rocks record a gain of Al and a loss of Si, which is compat-

GREENSCHIST FACIES HYDROTHERMAL ALTERATION

Table 5. Phase coefficients for balanced system reactions.

Reaction

1234

h

7

Olivine

2.81.27.8

Plagioclase

10.210.50.00.1

Actinolite

-1.0-1.0-1.0

Chlorite

-4.0-4.0-4.0-4.0

Reaction fitting product-reactant ratios (combination of 3,8 40.2 15.1 -5.2 -5.7

MgO

2.7-8.330.930.931.031.0

5)1.4

FeO

-0.5

-2.07.27.27.37.3

0.0

SiO2

4.96.5

-0.623.323.2

-0.9

NaO l / 2

0.10.10.1

-2.5-2.6

0.0

-3.4

CaO

1.81.81.8

-7.7-7.9

0.0

-1.9

A103/2

0.30.30.30.5

17.917.8

1.4

H2O

111

32323232

51.1

Note: Compositions of phases are listed in Table 4.

Table 6. Changes in oxide weight percentage. Table 7. Maximum estimates of water:rock ratio.

ObservedModeled

N a 2 O

-0.39-0.37

Chan

CaO

-0.27-0.37

ge in oxide weight percentage (g/a)

MgO

-0.180.20

FeO

-0.850.00

SiO 2

-1.23-0.19

A12O3

0.070.25

H2O

3.053.22

Notes: Observed changes are differences between average bulk composition of the four

highly altered samples and that of the three weakly altered samples (Table 3). Mod-

eled changes in the oxide weight abundances were calculated assuming that major

elements were transported in and out of the system in proportions defined by the

stoichiometry of reaction 8 (Table 4).

ible with the style of alteration observed in the troctolites in this study(Table 7). Significantly lower wateπrock ratios would be required toproduce the depletion of Fe, Si, and Ca (Table 7) if fluids that left thetroctolite system interacted with rock in a discharge zone before ex-pulsion through a hydrothermal vent.

The strong spatial association of alteration with fractures andveins suggests that they acted as conduits for fluid flow. Hydrother-mal circulation is commonly thought to be predominately driven byelevated heat flow above magma chambers (e.g., Lister, 1977,1988).However, in fault zones like that found in the MARK area, pressuregradients created by the opening of cracks and veins could be equallyimportant driving forces of fluid flow.

CONCLUSIONS

1. The nearly continuous range in amphibole compositions frompargasite to actinolite in gabbroic rocks from Leg 153 reflectsprogressive unroofing in the footwall of a brittle-ductile nor-mal fault. The degree of disequilibrium between preexistingminerals and secondary minerals increases as the contents oftitanium and aluminum of the secondary amphibole decrease.The highest degrees of alteration produced a greenschist faciesassemblage of actinolitic amphibole + chlorite ± talc ± epidote± prehnite ± pyrite ± quartz ± titanite.

2. The two dominant reactions that affected troctolitic gabbro ofHole 922A are olivine to actinolite and Plagioclase to chlorite.Coronitic reaction textures and strong linear correlations of in-creases in modal abundance of chlorite and actinolite and de-creases in modal abundances of Plagioclase and olivinesuggest that the two reactions operated simultaneously, andmutually supplied necessary reactants to the two types of alter-ation sites.

3. In the most highly altered samples, amphibole continued toform at the expense of Plagioclase after olivine was complete-ly consumed. The amphibole that replaced Plagioclase con-tains higher aluminum and alkali content than the amphibolethat replaced olivine but both contain almost no Ti. It is possi-ble that the latest change in amphibole composition was

Reaction 8Bulk composition

MgO

35

Maximum wateπrock ratio

FeO

0396.05

SiO2

560

Na2O

1010

CaO

3652

Notes: Ratios were calculated assuming direct inflow of seawater and outflow of hydro-thermal water. Seawater and hydrothermal water compositions were obtained fromin situ measurements by Campbell et al. (1988), and Campbell and Edmond (1989),which are summarized by Von Damm (1990).

caused by the change in host grain of alteration rather than bya change in pressure or temperature.

4. The dominant metasomatic change associated with greenschistfacies hydrothermal alteration was the addition of water; ap-proximately 5110 mol/m3 were added to the most highly al-tered troctolitic gabbro. This corresponds to a 3% increase inmaterial lost on ignition (LOI) that we assume to be predomi-nantly chemically bound water. Other observed and predictedchanges in major-element chemistry are small (<1%), and ar-guably would not be detectable in the suite of variably alteredsamples. It is worth noting, however, that the observed andmodeled changes of major elements in the altered troctolite arelargely antithetic to enrichment patterns observed in subma-rine hydrothermal water relative to seawater. Therefore, alter-ation could have occurred in the root zone of suchhydrothermal systems.

ACKNOWLEDGMENTS

This study would not have been possible without discussions andadvice from Dan Barnet, John Bowman, Mathilde Cannat, and JohnBartley. Ray Lambert and Bill Nash provided many hours to help de-sign the point-counting procedure. The paper significantly benefitedfrom careful reviews by Christine Laverne, Jonathan Snow, MathildeCannat, and Jeff Karson. Guillermo Rubio and Gabriel Rendon pro-vided many hours of creative technical support to process samplesused in this study.

REFERENCES

Alt, J.C., Honnorez, J., Laveme, C, and Emmermann, R., 1986. Hydrother-mal alteration of a 1 km section through the upper oceanic crust, DeepSea Drilling Project Hole 504B: mineralogy, chemistry, and evolution ofseawater-basalt interactions. J. Geophys. Res., 91:10309-10335.

Bartley, J.M., Fletcher, J.M., and Glazner, A.F., 1990. Tertiary extension andcontraction of lower plate rocks in the Central Mojave MetamorphicCore Complex, southern California. Tectonics, 9:521-534.

Campbell, A.C., and Edmond, J.M., 1989. Halide systematics of submarinehydrothermal vents. Nature, 342:168-170.

397

J.M. FLETCHER, C. STEPHENS, E.U. PETERSEN, L. SKERL

Campbell, A.C., Palmer, M.R., Klinkhammer, G.P., Bowers, T.S., Edmond,J.M., Lawrence, J.R., Casey, J.F., Thompson, G., Humphris, S., Rona,P.A., and Karson, J.A., 1988. Chemistry of hot springs on the Mid-Atlan-tic Ridge. Nature, 335:514-519.

Fletcher, J.M., Bartley, J.M., Martin, M.W., Glazner, A.F., and Walker, J.D.,1995. Large-magnitude continental extension: an example from the cen-tral Mojave metamorphic core complex. Geol. Soc. Am. Bull., 107:1468-1483.

Gregory, R.T., and Taylor, H.P., Jr., 1981. An oxygen isotope profile in asection of Cretaceous oceanic crust, Samail ophiolite, Oman: evidencefor δ18O-buffering of the oceans by deep (>5 km) seawater-hydrothermalcirculation at mid-ocean ridges. J. Geophys. Res., 86:2737-2755.

Harper, G.D., Bowman, J.R., and Kuhns, R.J., 1988. A field, chemical, andstable isotope study of subseafloor metamorphism of the Josephine Ophi-olite, California-Oregon. J. Geophys. Res., 93:4625-4656.

Karson, J.A., 1990. Seafloor spreading on the Mid-Atlantic Ridge: implica-tions for the structure of ophiolites and oceanic lithosphere produced inslow-spreading environments. In Malpas, J., Moores, E.M., Panayiotou,A., and Xenophontos, C. (Eds.), Ophiolites: Oceanic Crustal Analogues:Proc. Symp. "Troodos 1987": Nicosia, Cyprus (Minist. Agric. Nat.Resour.), 547-555.

Laverne, C, Vanko, D.A., Tartarotti, P., and Alt, J.C., 1995. Chemistry andgeothermometry of secondary minerals from the deep sheeted dike com-plex, Hole 504B. In Erzinger, J., Becker, K., Dick, H.J.B., and Stokking,L.B. (Eds.), Proc. ODP, Sci. Results, 137/140: College Station, TX(Ocean Drilling Program), 167-190.

Lécuyer, C , Brouxel, M., and Albaréde, F, 1990. Elemental fluxes duringhydrothermal alteration of Trinity Ophiolite (California, U.S.A.) by sea-water. Chern. Geol., 89:87-115.

Lister, C.R.B., 1977. Qualitative models of spreading-center processesincluding hydrothermal penetration. Tectonophysics, 37:203-218.

, 1980. Heat flow and hydrothermal circulation. Annu. Rev. EarthPlanet. Sci., 8:95-117.

Mével, C, 1987. Evolution of oceanic gabbros from DSDP Leg 82: influenceof the fluid phase on metamorphic crystallizations. Earth Planet. Sci.Lett., 83:67-79.

, 1988. Metamorphism of oceanic layer 3, Gorringe Bank, easternAtlantic. Contrib. Mineral. Petrol., 100:496-509.

Mével, C , and Cannat, M., 1991. Lithospheric stretching and hydrothermalprocesses in oceanic gabbros from slow-spreading ridges. In Peters, T,Nicolas, A., and Coleman, R.J. (Eds.), Ophiolite Genesis and Evolutionof the Oceanic Lithosphere. Petrol. Struct. Geol., 5:293-312.

Mottl, MJ., 1983. Metabasalts, axial hot springs, and the structure of hydro-thermal systems at mid-ocean ridges. Geol. Soc. Am. Bull., 94:161-180.

Nehlig, P., Juteau, T., Bendel, V., and Cotten, J., 1994. The root zone of oce-anic hydrothermal systems: constraints from the Samail ophiolite(Oman). J. Geophys. Res., 99:4703-4713.

Robie, R.A., and Waldbaum, D.R., 1968. Thermodynamic Properties of Min-erals and Related Substances at 298.15°K (25.0°C) and One Atmosphere(1.013 Bars) Pressure and at Higher Temperatures. U.S. Geol. Surv.Bull., 125.

Shipboard Scientific Party, 1995. Site 922. In Cannat, M., Karson, J.A.,Miller, D.J., et al., Proc. ODP, Init. Repts., 153: College Station, TX(Ocean Drilling Program), 179-215.

Spear, F.S., 1980. NaSiCaAl exchange equilibrium between Plagioclase andamphibole: an empirical model. Contrib. Mineral. Petrol, 72:33-41.

, 1993. Metamorphic Phase Equilibria and Pressure-Temperature-Time Paths: Washington (Mineral. Soc. Am.), Monogr. 1.

Spear, F.S., and Peacock, S.M., 1989. Metamorphic Pressure-Temperature-Time Paths. Am. Geophys. Union Short Course in Geology, 7.

Stakes, D.S., and O'Neil, J.R., 1982. Mineralogy and stable isotopegeochemistry of hydrothermally altered oceanic rocks. Earth Planet. Sci.Lett, 57:285-304.

Thompson, J.B., 1982. Reaction space: an algebraic and geometric approach.In Ferry, J.M. (Ed.), Characterization of Metamorphism through Miner-alogical Phase Equilibria. Mineral. Soc. Am., Rev. Mineral., 10:33-51.

Von Damm, K.L., 1990. Seafloor hydrothermal activity: black smoker chem-istry and chimneys. Annu. Rev. Earth Planet. Sci., 18:173-204.

Date of initial receipt: 15 August 1995Date of acceptance: 6 February 1996Ms 153SR-036