A long in situ section of the lower ocean crust: results...

A long in situ section of the lower ocean crust: results of ODPLeg 176 drilling at the Southwest Indian Ridge

Henry J.B. Dick a;*, James H. Natland b, Je¡rey C. Alt c, Wolfgang Bach c,Daniel Bideau c, Je¡rey S. Gee c, Sarah Haggas c, Jan G.H. Hertogen c,

Greg Hirth c, Paul Martin Holm c, Benoit Ildefonse c, Gerardo J. Iturrino c,Barbara E. John c, Deborah S. Kelley c, Eiichi Kikawa c, Andrew Kingdon c,

Petrus J. LeRoux c, Jinichiro Maeda c, Peter S. Meyer c, D. Jay Miller d,H. Richard Naslund c, Yao-Ling Niu c, Paul T. Robinson c, Jonathan Snow c,

Ralph A. Stephen c, Patrick W. Trimby c, Horst-Ulrich Worm c,Aaron Yoshinobu c

a Leg 176 Co-chief Scientist, Woods Hole Oceanographic Institution, Woods Hole, MA 06520, USAb Leg 176 Co-chief Scientist, Rosenstiel School of Marine and Atmospheric Science, Miami, FL 33149, USA

c Leg 176 Scienti¢c Party, c/o Ocean Drilling Program, College Station, TX 77845-9547, USAd Leg 176 Sta¡ Scientist, Ocean Drilling Program, College Station, TX 77845-9547, USA

Received 29 May 1999; accepted 8 March 2000

Abstract

Ocean Drilling Program Leg 176 deepened Hole 735B in gabbroic lower ocean crust by 1 km to 1.5 km. The sectionhas the physical properties of seismic layer 3, and a total magnetization sufficient by itself to account for the overlyinglineated sea-surface magnetic anomaly. The rocks from Hole 735B are principally olivine gabbro, with evidence for twoprincipal and many secondary intrusive events. There are innumerable late small ferrogabbro intrusions, oftenassociated with shear zones that cross-cut the olivine gabbros. The ferrogabbros dramatically increase upward in thesection. Whereas there are many small patches of ferrogabbro representing late iron- and titanium-rich melt trappedintragranularly in olivine gabbro, most late melt was redistributed prior to complete solidification by compaction anddeformation. This, rather than in situ upward differentiation of a large magma body, produced the principal igneousstratigraphy. The computed bulk composition of the hole is too evolved to mass balance mid-ocean ridge basalt back toa primary magma, and there must be a significant mass of missing primitive cumulates. These could lie either below thehole or out of the section. Possibly the gabbros were emplaced by along-axis intrusion of moderately differentiatedmelts into the near-transform environment. Alteration occurred in three stages. High-temperature granulite- toamphibolite-facies alteration is most important, coinciding with brittle^ductile deformation beneath the ridge. Minorgreenschist-facies alteration occurred under largely static conditions, likely during block uplift at the ridge transformintersection. Late post-uplift low-temperature alteration produced locally abundant smectite, often in previouslyunaltered areas. The most important features of the high- and low-temperature alteration are their respective

0012-821X / 00 / $ ^ see front matter ß 2000 Elsevier Science B.V. All rights reserved.PII: S 0 0 1 2 - 8 2 1 X ( 0 0 ) 0 0 1 0 2 - 3

* Corresponding author. E-mail: [email protected]

EPSL 5460 16-5-00 Cyaan Magenta Geel Zwart

Earth and Planetary Science Letters 179 (2000) 31^51

www.elsevier.com/locate/epsl

associations with ductile and cataclastic deformation, and an overall decrease downhole with hydrothermal alterationgenerally 9 5% in the bottom kilometer. Hole 735B provides evidence for a strongly heterogeneous lower ocean crust,and for the inherent interplay of deformation, alteration and igneous processes at slow-spreading ridges. It is strikinglydifferent from gabbros sampled from fast-spreading ridges and at most well-described ophiolite complexes. Weattribute this to the remarkable diversity of tectonic environments where crustal accretion occurs in the oceans and tothe low probability of a section of old slow-spread crust formed near a major large-offset transform being emplaced on-land compared to sections of young crust from small ocean basins. ß 2000 Elsevier Science B.V. All rights reserved.

Keywords: Leg 176; mid-ocean ridges; lower crust; gabbros; alteration; deformation

1. Introduction

Because of its inaccessibility, the nature of thelower ocean crust has largely been inferred fromremote sensing and by analogy to ophiolites.From its inception, a major goal of scienti¢cocean drilling has been to drill through the entireocean crust into the mantle to determine directlyits composition and structure. Although such adeep hole is beyond current technical capabilities,many fundamental questions can be addressed bydrilling into tectonic windows where the lowercrust is exposed. This was the goal of Ocean Drill-ing Program (ODP) Leg 118, which started Hole735B in 1987 and drilled to 504.8 m below sea-£oor (mbsf), recovering 433.3 m of gabbroic low-er crust [1] in a tectonic window at the Southwest(SW) Indian Ridge. ODP Leg 176 reoccupied thishole in October 1997 and drilled for 27 days to atotal depth of 1508 mbsf, recovering an additional866 m of gabbro. Crust formed at the ultra-slow-spreading SW Indian Ridge is believed to be onlyabout 4 km thick [2]. If pillow lavas and dikescomprise 1.5^2 km of this, the lower crustal sec-tion should be no more than 2^2.5 km thick. Hole735B is now deep enough to be regarded as rep-resentative of much of slow-spread lower oceancrust. Thus for the ¢rst time, a signi¢cant propor-tion of an all but inaccessible layer of the Earthhas been sampled in situ, and we provide here areport of our initial ¢ndings and discuss their im-plication for crustal accretion in the oceans.

The section consists chie£y of gabbro with amean density of 2.979 kg/m3 and a seismic veloc-ity of 6.777 km/s, both appropriate for oceaniclayer 3. The stratigraphy, however, varies dramat-ically with depth, exhibiting large changes in igne-ous chemistry, deformation and alteration (Fig.

1). The chemical variations are complex, and theigneous chemistry cannot be explained by simplefractionation of an upwardly di¡erentiating mag-ma chamber. Rather, the igneous, structural andalteration characteristics are the product of phys-ical processes inherent to a tectonically activeridge environment. The bulk composition ofHole 735B is too di¡erentiated to mass balanceSW Indian Ridge basalts back to the compositionof a likely primary magma without the additionof a large volume of primitive cumulates. If, asmay be the case, the crust^mantle boundary lies ashort distance below the hole, the missing cumu-lates would lie out of section. Site 735 is somedistance from the mid-point of the paleo-ridgesegment, and the gabbros could represent moder-ately di¡erentiated melts intruded down-axis fromthe segment center. The section drilled at Site 735di¡ers in many respects from those in well-knownophiolites and from gabbros sampled to date atfast-spreading ridges. These di¡erences presum-ably re£ect the tectonically diverse settings inwhich ocean crust forms, and the probabilitiesthat ocean crust of a particular provenance is em-placed on-land. Well-preserved ophiolites formedin the slow-spreading large-o¡set ridge transformintersection environment like Site 735, then, maybe rare or non-existent.

2. Methods

Hole 735B was cored continuously, with coresrecovered every 5^9.5 m. In several instances, re-covery exceeded the interval drilled, a phenomen-on referred to as polling where core fails to breako¡ during retrieval, and is recovered after drillingthe next interval. The overall high recovery (86%)


H.J.B. Dick et al. / Earth and Planetary Science Letters 179 (2000) 31^5132

compared with most basement legs (V22%)largely re£ects the massive character of the rockand possibly greater drill string stability in shal-low water. The Leg 176 cores were described inboth hand sample and thin section. Semi-quanti-tative estimates were made of alteration, igneousmodes, grain size, textures and intensity of defor-mation on a centimeter scale through all 866 m ofcore. Igneous mineralogy was point-counted onover 200 thin sections, and detailed semi-quanti-tative petrographic descriptions of rock alteration,deformation and igneous textures were made on285 thin sections. The scientists worked in disci-plinary teams, with the igneous petrologists, meta-morphic petrologists and structural geologistsworking separate overlapping shifts. Althoughfeatures were described by teams, only one indi-vidual recorded a speci¢c measurement or obser-vation throughout the core to insure the best pos-sible accuracy and precision. Samples were chosenfor shipboard XRF, physical properties and mag-netic measurements to be as representative of eachcore section as possible. A consistent attempt wasmade not to over sample small curious features ofthe core, and overall the measurements are repre-sentative of the actual material drilled. Thin sec-tions were made wherever an XRF or physicalproperties sample was taken. All the data werethen logged into spreadsheets that are presentedin electronic format in the Initial Reports [3]. Thevarious downhole plots presented here representbut a small fraction of the available data. In orderto insure consistency with earlier results, the sci-enti¢c party relogged the last 50 m of the Leg 118cores and reanalyzed some Leg 118 XRF pow-ders.

3. Tectonic setting

The SW Indian Ridge is a highly segmentedridge representing nearly the ultra-slow-spreadingend-member for crustal accretion in the oceans.Hole 735B is at Atlantis Bank, a 5 km localhigh situated along the eastern wall of the Atlan-tis II Transform. North of Atlantis Bank, the SWIndian Ridge is spreading asymmetrically 0.6 cm/yr due north and 1 cm/yr due south [6]. The hole,

at 32³43PS, 57³17³E, is in 11 Ma old crust, about18 km east of the present-day transform-slip zoneand 95 km south of the SW Indian Ridge axis. Atthe time of accretion, however, it was only about15 km from the active transform fault as therewas subsequent transtensional extension acrossthe transform due to a spreading direction change[6]. The top of Atlantis Bank, where the hole islocated, is a wave-cut platform exposing gabbro,with less than 100 m total relief over a 25 km2

area [7]. Seismic Moho is estimated to be approx-imately 5.3 km below Hole 735B [8] and is inter-preted as an alteration front rather than the igne-ous crust^mantle boundary, based on the presenceof serpentinized peridotite high on the transformwall just to the west of the hole [6].

The Atlantis Bank gabbros are interpreted ashaving been unroofed by a large low-angle de-tachment fault at the northern ridge transformintersection and then uplifted to sealevel at thenorthern inside-corner high of the paleo-SW In-dian Ridge [6]. The Bank has since subsided to itspresent 700 m depth over the last 10 Ma. Northof the SW Indian Ridge, on crust of the same ageand distance from the paleo-transform, the vol-canic carapace originally overlying the Hole735B gabbros appears to be preserved intact[6,8]. Here, a series of E^W lineated ridges closelyresemble the modern volcanic topography of theSW Indian Ridge rift valley £oor. Although thesea£oor has been mapped for 9 km to the east ofthe paleo-position of Hole 735B [9], the map areadoes not extend far enough to the east to reachthe local high de¢ning the mid-point of the vol-canic segment. Accordingly, Hole 735B lies signif-icantly west of the spreading segment center,rather than near its mid-point as originally sug-gested [6].

The exact structural position of the AtlantisBank gabbros in the lower crust is unknown.However, £uid inclusions in the upper 500 mformed 1.5^2 km below sea£oor [10]. Becausethis depth is close to thickness of dikes and pil-lows seen in many ophiolites and drilled at Hole504B, the top of Hole 735B may not be far fromthe dike^gabbro transition. The appearance oftroctolite in the lowermost 200 m, and the serpen-tinized peridotite high on the transform wall to


H.J.B. Dick et al. / Earth and Planetary Science Letters 179 (2000) 31^51 33

the west [6], could indicate that the crust^mantleboundary is close to the bottom of the hole. How-ever, whereas Mg-olivine-rich troctolitic gabbrosare abundant near the crust^mantle boundary in

some ophiolites, the lowermost Hole 735B olivinegabbros have maximum Mg/(Mg+Fe) of onlyabout 0.82. This is considerably more iron-richthan the mantle ratio (V0.90). Cumulates lying

Fig. 1. Hole 735B lithostratigraphic variations [3,4]. Left: Relative abundances of igneous rocks averaged over a moving 20 m in-terval. Rock names follow the standard IUGS classi¢cation with modi¢ers to indicate oxide abundance. Middle left : Crystal^plastic deformation intensity from zero (undeformed) to ¢ve (ultramylonite). Middle right: Amphibole veins by core percentageaveraged over a 2 m window. Right: Characteristic remanent inclinations. Mean inclination value (right gray line) is based onthe inclination only averaging technique of McFadden and Reid [5].



along a simple depositional contact with the man-tle should have Mg/(Mg+Fe) close to the mantlevalue. However, if the gabbros represent down-axis intrusion of di¡erentiated melts into the man-tle near the transform, the crust^mantle boundaryexposed on the transform wall could have virtu-ally any orientation, and mantle peridotite couldeven locally overlie the gabbro. Thus, this bound-ary cannot be simply projected subhorizontally tothe east from the transform wall to below Hole735B without evidence of its orientation andform.

4. Igneous stratigraphy

Gabbroic rocks containing 0.5% felsic veins andtwo diabase dikes (both in the upper 500 m) makeup the entire Hole 735B core. A total of 953 dis-crete igneous intervals have been described, 457 inthe lower kilometer, and there is additional sub-division possible at centimeter scales. Plagioclase(close to 60% of most rocks) and augite are theprincipal constituents of the gabbros. The maingabbro types are distinguished by variations ingrain size and abundance of olivine, magmaticoxides (ilmenite and magnetite) and orthopyrox-ene. Anorthosite and clinopyroxenite occur inrare isolated patches. The average grain sizeranges from coarse (5^15 mm) to very coarse(15^30 mm), with olivine, plagioclase and pyrox-ene generally covarying in size. Igneous contactsbetween lithologic intervals are 41% intrusive,37% gradational and 6% tectonic, with 16% notrecovered. Excluding tectonic and highly irregularcontacts, igneous contacts dip from 0³ to 90³, andaverage 36³ with no systematic downhole trends.Modal and grain size layering are locally present,with rhythmic layering at several locations. Inter-vals with obvious modal or graded layering, how-ever, constitute less than 2% of the core. Dipsrange from 0³ to 52³ and average 24³, with layerstypically lying in the plane of the crystal^plasticfoliation. A generally weak magmatic foliation,typically dipping from 20³ to 50³, is de¢ned byplagioclase laths and is present in 20% of the gab-bros. This foliation, with few exceptions, is notpresent in the layered gabbros.

Of the major cumulates expected as products ofdi¡erentiation of mid-ocean ridge basalt (MORB)(dunite, troctolite, olivine gabbro, gabbro, gab-bronorite, oxide gabbronorite and oxide gabbro),all are present except dunite (Table 2, electronicappendix). Whereas these rocks represent a gen-eral evolutionary di¡erentiation sequence, there isno systematic vertical change consistent with asingle large upwardly di¡erentiating magmabody, nor a simple evolutionary sequence suchas in layered intrusions like the Skaergaard.Rather the rock types occur in separated enclavesthat interpenetrate one another throughout thehole. While ferrogabbro is most abundant in theupper 500 m, primitive and evolved gabbros aredistributed irregularly throughout the section (Ta-ble 2, electronic appendix). Igneous intervalsrange widely in size from a few centimeters tomany meters, re£ecting the complex polygeneticcharacter of the section and the many small oxidegabbro ¢lled shear zones, intrusive microgabbrosand felsic veins. The igneous stratigraphy of Hole735B is thus characterized by extreme small-scalechemical and textural variability with multiplestages and styles of intrusion down to the handspecimen scale (e.g. Fig. 2).

The igneous stratigraphy is divided into 12 pol-ygenetic units, including modi¢cations to the Leg

Fig. 2. Shipboard XRF whole rock gabbro molecular Mg/(Mg+Fe2�) with FeO = 0.85UFe2O3 [3,4]. Filled symbols:TiO2 6 0.4 wt%, half-¢lled symbols: TiO2 from 0.4 to 1.0wt%, and open symbols: TiO2 s 1.0 wt%.



118 units [4], re£ecting changes in principal lith-ologies and various intrusive events. The productof a complex interplay of magmatic and tectonicevents, this stratigraphy provides a purely descrip-tive breakdown of the lithologic variability of thehole conceptually di¡erent than that for layeredintrusions. Some contacts are deformed, but at afew locations are simple faults, usually marked by

thin mylonites such as at the bottom of Units IIIand VIIA. Gradational contacts, characterized bythe disappearance of a distinctive mineral phase,such as orthopyroxene at the bottom of Unit IX,and abundant ilmenite and magnetite at the bot-tom of Unit IV, are common. Some unit bounda-ries are marked by the sudden appearance of anintrusive swarm, such as cross-cutting microgab-

Table 1Hole 735B igneous lithostratigraphic unitsa

Unit I: gabbronorite, 0^37.41 m, 38 intervalsSubunit IA: massive gabbronorite, 0^27.99 m, 27 intervalsSubunit IB: olivine gabbro and gabbronorite, 27.99^37.41 m, 11 intervals

Unit II: upper compound olivine gabbro, 37.41^170.22 m, 91 intervalsMinor: Intrusive : microgabbro and olivine microgabbro

Synkinematic : oxide and oxideôlivine gabbroUnit III: disseminated oxideôlivine gabbro, 170.22^223.57 m, 76 intervals

Subunit IIIA: disseminated oxideôlivine and olivine gabbro, 170.22^180.09 m, 12 intervalsSubunit IIIB: massive disseminated oxideôlivine gabbro, 180.09^209.45 m, 20 intervalsSubunit IIIC: disseminated oxideôlivine and olivine gabbro, 209.45^223.57 m, 44 intervals

Unit IV: massive oxideôlivine gabbro, 223.57^274.06 m (mylonite at upper contact), 32 intervalsUnit V: massive olivine gabbro, 274.06^382.40 m, 15 intervalsUnit VI: lower compound olivine gabbro, 382.40^536 m, 207 intervals

Subunit VIA: compound olivine gabbro, 382.40^404.01 m, 24 intervalsMinor: Intrusive : olivine microgabbronorite and olivine microgabbro

Synkinematic : olivine and oxideôlivine gabbroSubunit VIB: compound olivine, oxideôlivine and disseminated oxideôlivine gabbro, 404.01^419.28 m, 43 intervals

Minor: Intrusive : olivine and oxideôlivine microgabbroSubunit VIC: compound troctolitic and olivine gabbro, 419.28^433.77 m, 44 intervals

Minor: Synkinematic : oxideôlivine gabbro and oxideôlivine microgabbroIntrusive : troctolite and troctolitic microgabbro

Subunit VID: compound olivine and oxideôlivine gabbro, 433.77^536 m, 96 intervalsMinor: Intrusive : troctolite, diopsidic olivine and troctolitic gabbro

Synkinematic : disseminated oxideôlivine and olivine gabbroUnit VII: gabbronorite and oxide gabbronorite, 536^599 m, 35 intervals

Subunit VIIA: gabbronorite and oxide gabbronorite, 536^560 mMinor: olivine gabbro

Subunit VIIB: gabbronorite and oxide gabbronorite, 560^599 m (fault at upper contact)Unit VIII: olivine gabbro, 599^670 m, 22 intervals

Minor: oxide gabbronorite, gabbronorite and oxide gabbroUnit IX: gabbronorite and gabbro, 670^714 m, 32 intervals

Minor: oxide gabbro and oxide gabbronoriteUnit X: olivine gabbro and gabbro, 714^960 m, 116 intervals

Minor: Intrusive : microgabbro, pegmatitic intervals, rhythmic layeringRare: oxide gabbro

Unit XI: olivine gabbro, 960^1314 m, 164 intervals (shear zone at lower contact)Minor: gabbro, oxide gabbro, rhythmic layeringRare: troctolitic gabbro

Unit XII: olivine gabbro and troctolitic gabbro, 1314^1508 m, 72 intervalsMinor: Intrusive : microgabbroRare: oxide gabbro, leucogabbro

aUnit de¢nitions follow Robinson, Von Herzen et al. (1989) as modi¢ed by Dick et al. (1991), and Dick, Natland, Miller et al.(1999).



Tab

le2

Res

ults

ofM

ass

Bal

ance

Cal

cula

tion

sfo

rH

ole

735B

Int.

SiO

2T

iO2

Al 2

O3

FeO

Fe 2

O3

FeO

aM

nOM

gOC

aON

a 2O

K2O

P2O

5M

g#C

a#

0-50

0m

eter

s49

349

.31.

4115

.76.

921.

978.

440.

159.

1512

.12

2.67

0.05

0.04

65.9

71.5

500-

1000

met

ers

229

51.1

0.70

16.0

6.28

1.25

7.33

0.15

9.16

12.3

62.

930.

050.

0869

.070

.010

00-1

508

met

ers

231

51.3

0.50

16.6

5.39

0.89

6.19

0.13

9.33

12.9

62.

800.

040.

0372

.971

.90-

1508

mbs

f95

350

.60.

8716

.16.

191.

377.

310.

149.

2112

.52.

800.

050.

0569

.271

.1C

alcu

late

dP

rim

ary

MO

RB

b50

.30.

9116

.4nd

nd7.

45nd

10.9

011

.42.

520.

07nd

72.3

Hol

e73

5BD

iaba

se2

49.7

1.73

14.9

7.18

2.38

9.32

0.19

7.76

11.3

03.

000.

080.

3659

.767

.5M

ost

prim

itiv

eA

IIF

.Z.

Bas

alt

51.0

1.57

15.7

ndnd

9.07

0.20

8.09

11.0

3.10

0.08

0.16

61.4

Ave

rage

Atl

anti

sII

F.Z

.B

asal

tc70

50.3

11.

9615

.25

10.3

00.

197.

2210

.39

3.25

0.17

0.22

55.5

63.9

VC

rN

iC

uZ

nR

bSr

YZ

rN

bB

ulk

dens

ity

0^50

0m

eter

s24

029

510

651

4316

313

302.

984

500^

1000

met

ers

199

119

9575

530.

616

617

441.

32.

981

1000

^150

8m

eter

s17

712

010

074

430.

717

012

352.

52.

970

0^15

08m

eter

s20

517

810

066

4616

614

36.5

2.97

8H

ole

735B

Dia

base

242

190

9161

681

157

4313

43.

7

See

elec

tron

icta

ble

for

the

full

com

puta

tion

and

data

inpu

tsfo

rth

ista

ble

incl

udin

gav

erag

ein

terv

aldi

men

sion

sby

rock

type

.M

olec

ular

rati

os:

Mg#

=10

0*M

g/(M

g+F

e)us

esal

lir

onas

ferr

ous

iron

;C

a#=

100*

Ca/

(Ca+

Na)

.aM

ost

prim

itiv

eof

72ba

salt

glas

san

alys

esfr

om14

dred

geha

uls

from

the

Atl

anti

sII

F.Z

.re

port

edby

John

son

and

Dic

k(1

992)

.bC

alcu

late

dpr

imar

yM

OR

Bco

mpo

siti

onfr

omK

inzl

eran

dG

rove

(199

3)m

ost

clos

ely

mat

chin

gth

eH

ole

735B

bulk

com

posi

tion

.L

iqui

dco

mpo

siti

onca

lcul

ated

for

isob

aric

incr

emen

tal

batc

hac

cum

ulat

ive

mel

ting

at10

kban

d10

.3%

mel

ting

ofa

`dep

lete

d'sp

inel

lher

zolit

em

odel

led

afte

rth

atof

Zin

dler

and

Har

t(1

986)

.c Se

vent

ygl

ass

aver

age

for

11dr

edge

haul

sfr

omJo

hnso

nan

dD

ick

(199

2).



bros at the top of Unit VI and abundant thinintervals of oxide gabbro at the top of Unit XIin otherwise continuous sections of coarse olivinegabbro. A summary of the igneous stratigraphy isgiven in Table 1 and a breakdown of average rockcompositions and interval thicknesses are in-cluded in the Epsl online background dataset.1

Despite this complexity, the gabbros can be di-vided into three major associations: (1) oxide-poor olivine-bearing gabbros, (2) cross-cutting mi-crogabbros, and (3) `ferrogabbros' consisting ofoxide-bearing gabbros and gabbronorites.

Coarse-grained, equigranular or vari-texturedoxide-poor (6 1%) olivine-bearing gabbros andtroctolites make up about 78% of the section.These rocks are texturally similar and exhibit acontinuous gradation in olivine content from gab-bro to troctolite. Vari-textured varieties are sim-ilar to the more abundant equigranular gabbros,but contain irregular coarse- and ¢ne-grainedequigranular gabbro patches at a variety of scales.These gabbros contain generally less than 1% lateintergranular phases such as orthopyroxene,brown amphibole, oxide minerals and apatite.The rocks are clearly cumulates, meaning thatthey crystallized and separated from magma,and do not have bulk compositions appropriatefor a melt (e.g. [11]). Chemically, the olivine-bear-ing gabbros can be separated into ¢ve units from200 to 600 m thick. These are each characterizedby an upward trend of decreasing Mg# (Mg/{Mg+Fe}), with a sharp increase in Mg# at thebeginning of each overlying unit (Fig. 2). The in-dividual chemical units de¢ned by these trendspartially overlap and each appears to be a com-posite, consisting of many smaller petrographi-cally distinct intrusive intervals. The simplest in-terpretation is that each chemical unit representssome form of cyclic intrusion that di¡erentiated insitu as the magma worked its way upwardthrough a speci¢c intrusive horizon in the section.Despite lower overall olivine contents, the upper-most two chemical units in Fig. 2 have higheraverage Mg#, lower TiO2 and Na2O, and are

on average chemically more primitive than thelower three. This suggests that the upper twochemical units and the lower three chemical unitscould, in turn, represent two major intrusivephases characterized by di¡erent parent magmas.

Small medium- to ¢ne-grained equigranular mi-crogabbros, ranging in composition from trocto-lite to oxide gabbronorite, cross-cut the coarse-grained olivine-bearing gabbros at many intervalsthroughout the core. These range from severalmeters, to little more than a centimeter thickand span nearly the full compositional range ofthe coarser olivine gabbro and ferrogabbro theyintrude (Fig. 3). Based on their composition (Ta-ble 2, EPSL online background dataset1 appendix),most microgabbros are also cumulates retaininglittle trapped melt, although some of the moredi¡erentiated varieties appear to be close to a rea-sonable magma composition (e.g. the microgab-bronorite analyzed by Hart et al. [14]). The micro-gabbros include the troctolites in Unit VI, whichare the most primitive, high-Mg rocks in the sec-tion. These cross-cut the olivine gabbros and arechemically and texturally distinct from the coars-er-grained troctolites at the bottom of the hole.Despite evidence of late reaction with Fe^Ti-richmelts, they contain Fo87 olivine [15]. Whereasmost of the Hole 735B gabbros crystallized frommoderately di¡erentiated melts, those that pro-duced the troctolites were fairly primitive [4,15].Most of the microgabbro contacts with the olivinegabbros are sharp, clearly intrusive, and in handspecimen are often highly irregular in form. Typ-ically individual grains interlock across theboundary with the ¢ner and coarser grains inter-grown, giving the contact a sutured appearance inthin section. Less abundant are simple size-gradedcontacts. Only ¢ve microgabbros in the lowerkilometer have simple ìntrusive' contacts whereindividual mineral grains are broken along thecontact by brittle fracture. These contact relation-ships suggest that in many cases, the intrudedolivine gabbro was not fully solidi¢ed at thetime of intrusion by the microgabbro. While con-tacts higher in the hole occur with shallow tomoderate dips, near the base, thin sinuous micro-gabbros intrude coarse-grained olivine gabbroalong steeply dipping to vertical intrusive con-

1 http://www.elsevier.nl/locate/epsl; mirror site: http://www.elsevier.com/locate/epsl



tacts. We interpret the latter to represent melttransport channels through the crystallizing intru-sions.

Hundreds of thin bodies of disseminated oxide(1^2 vol%) and oxide-rich (2^30 vol%) gabbronor-ite, gabbro and olivine gabbro, as well as severalmassive ferrogabbro units in the Leg 118 section,cross-cut the oxide-poor olivine gabbros. Theseferrogabbros are all distinguished from the ox-ide-poor olivine gabbros by ilmenite and magnet-ite visible in hand specimen. Although the di¡er-ence in modal oxide between disseminated oxideôlivine gabbro (1^2 vol%) and oxide-poor olivinegabbro (6 1 vol%) is small, it is petrogeneticallysigni¢cant, since the former usually also containsubstantially more sodic plagioclase and moreiron-rich pyroxenes and olivine [16,17]. In addi-tion, these ferrogabbros often contain signi¢cantquantities of hypersthene and brown amphibole.Usually, however, they are de¢cient in very latemagmatic phases such as apatite and zircon. It isclear, then, that these gabbros are also largely

cumulates, containing low percentages of residualmelt, even though they (albeit) formed at a muchlater stage of di¡erentiation than the olivine gab-bros.

The relationship between ferrogabbro and ox-ide-poor olivine gabbro is quite varied. Modallygradational contacts are common. Many othersare sharp, marked by the appearance of inter-granular oxide and a change in silicate mineralproportions (generally an increase in pyroxeneand often the disappearance of olivine) across asutured boundary of interlocking grains. In manyexamples, the oxide gabbro lies in a zone of de-formed gabbro with its contacts coinciding withthe transition from deformed to undeformed gab-bro. The most common types of ferrogabbro inthe lower kilometer are either small oxide-rich,high-temperature shear zones cross-cutting thecoarse olivine-bearing gabbros, or local unde-formed patches of oxide gabbro in olivine gabbro.In many ferrogabbros, the oxide minerals locallycross-cut microcracked mineral grains and crys-

Fig. 3. Shipboard XRF whole rock gabbro molecular Mg/(Mg+Fe) and Ca/(Ca+Na), and calculated total hole and 500 m bulkcompositions from Table 2. All iron as FeO. Also shown are average SW Indian and AmericanÂntarctic Ridge abyssal perido-tites [12], and the ¢eld of the Atlantis II Fracture Zone basalt glasses [4]. The hypothetical primary magma trend is for V5^20%melts of estimated MORB mantle source compositions using various polybaric and isobaric melting models calculated from melt-ing experiments by Kinzler and Grove [13].



tal^plastic deformation fabrics in olivine gabbro,or they lie along foliation planes, ¢lling pressureshadows around pyroxene porphyroclasts andtension gashes in the rock [11]. Many of themost massive ferrogabbros in the upper 500 mappear to have an entirely igneous fabric inhand specimen, but when examined in long coresections, local oxide segregations appear to ¢lltension gashes and fractures. Moreover, whereasplagioclase and pyroxene grain boundaries in ox-ide-poor olivine gabbros tend to interlock in com-plex ways, pyroxene grains in many massive fer-rogabbros tend to be blocky, often even rounded,suggesting that they were abraded and deformedbefore complete solidi¢cation.

Iron^titanium oxide-rich gabbros and gabbro-norites markedly diminish in volume downhole(Fig. 1). They make up 30% of the core in theupper 500 m, 12% of the core in the lower1000 m and less than 1% of the core in the lower300 m. The thickest units with the most extremeenrichment in oxides, sodium, iron and titaniumoccur above 500 mbsf (Figs. 1 and 2, Table 1).There, Unit IV, a 50 m thick section, contains20 intervals of massive oxideôlivine gabbro andmicrogabbro, averaging 10% oxide, that areinterspersed with four small intervals of unde-formed oxide-poor olivine gabbro. Overlying thissequence is Unit III, an apparently related butmore varied 53 m thick unit of disseminatedoxideôlivine gabbro. In the remainder of thehole, oxide gabbros typically form bands only afew centimeters to a few tens of centimetersthick.

A striking feature of the stratigraphy of Hole735B is a strong downward enrichment in ironand sodium from the top of Unit III throughUnit IV. In the usual case of simple di¡erentiationof a magma body, a reverse chemical trend isexpected. Units III and IV also represent a broadzone of deformation, however, with an overallcrystal^plastic foliation that £attens into a planeof high shear strain with depth. The dip of thisfoliation correlates with the downward iron andsodium enrichment [4,18]. One explanation of-fered for this unique chemical trend is that melt£ux and the extent of reaction between late Fe^Timelt and olivine gabbro protolith is controlled by

the in£uence of deformation on melt migrationalong the shear zone [4,18].

In summary, there is convincing evidence forlocal migration and intrusion of Fe^Ti-rich meltsalong shear zones throughout the olivine gabbrosduring deformation. On the other hand, there arealso many local patches or layers of ferrogabbro,particularly in the lower third of the section, withno suggestion of synmagmatic deformation. Inthese cases, it appears that late Fe^Ti-rich meltspooled locally in situ in the olivine gabbro at theend of crystallization.

Late igneous felsic veins include leucodiorite,diorite, trondhjemite and tonalite with variableamounts of dark green amphibole, quartz andbiotite. A few truly granitic veins were also recov-ered. The felsic veins represent the most di¡eren-tiated melt compositions in Hole 735B. Theyrange in size from a few mm to several centimeter,and cross-cut all other lithologies, usually ¢llingcrack networks forming net^vein complexes. Adistinctive feature of many igneous felsic veinsin the lower kilometer of the hole, evidently lesscommon in the upper 500 m, are Fe^Ti oxide-richreaction zones or halos in the adjacent olivinegabbro. This feature can be explained by localin situ di¡erentiation of ferrobasalt melt and re-action with the wall rock to produce the titano-magnetite and the leucocratic vein. Some felsicveins were deformed under granulite- to amphib-olite-facies conditions, and at the top of the sec-tion are locally transposed into the foliation planein amphibolite zones [4]. Some deformed ferro-gabbros are impregnated by felsic material, form-ing hybrids with dioritic compositions. This led tosome speculation that these represent unmixing orextreme in situ fractionation of late Fe^Ti-richmelts during transport within a shear zone.Although igneous felsic veins are minor volumet-rically, and decrease in abundance with depth,more than 200 were described in the lower kilo-meter of the hole.

5. Deformation

Macroscopic deformation in Hole 735B is gen-erally localized in discrete shear and fault zones,



with the bulk of the core (77%) being undeformed(Fig. 4). Magmatic, crystal^plastic and brittle de-formation features are locally well-developed.There is a marked downward decrease in the in-tensity of crystal^plastic deformation, and mag-matic foliations are also absent in long intervalsnear the bottom of the hole. Brittle fracture den-sity, as shown by hydrothermal veins, also de-creases downhole (e.g. Figs. 1 and 4). As in theLeg 118 section [4,11,17], some rocks were de-formed and recrystallized while still partly molten,and there is a transition between high-temperaturemetamorphic and magmatic processes [3]. Inmany instances, crystal^plastic fabrics are cross-cut by stringers of magmatic oxides and theseoxides also locally cement broken silicate grainsand ¢ll cracks. Many of the most striking zones ofcrystal^plastic deformation, however, formedunder the equivalent of granulite-facies metamor-phic conditions (s 800^1000³C), when there waslittle or no melt present. Above 500 mbsf, thedominant sense of shear is normal, whereas belowthat level, there are several zones with reversesense shear. A weak, subparallel, crystal^plasticfabric, that may record a transition from mag-matic to crystal^plastic deformation, commonlyoverprints magmatic foliations. Many of the de-formed rocks also show a continuum betweencrystal^plastic and brittle behavior. There are nar-row zones of intense cataclasis, and several faultzones, two of which coincide with Leg 118 verticalseismic pro¢le re£ectors [3].

Although igneous textures are locally obscuredby deformation, there is a strong association be-tween magmatic foliations de¢ned by feldsparlaths and zones of crystal^plastic deformation,with long intervals of the core containing neither.Crystal^plastic and magmatic fabrics also share acommon orientation (Fig. 4). Normally, becauseof rotation while drilling, core pieces have no azi-muthal orientation. The Hole 735B cores, how-ever, were oriented for splitting so that macro-scopic foliations had a common orientation.Although this automatically produces stereo plotcorrelations between unrelated foliations in thecore reference frame, the magnetic declinationsalso have a very consistent orientation (near260³ in core coordinates). Thus, assuming an orig-

inal magnetic declination of 180³, magmatic andcrystal^plastic fabrics both generally had an ap-proximately E^W strike and dip toward the ridgeaxis.

In many intervals, primary igneous textures areoverprinted by sub-solidus granulite- and amphib-olite-facies deformation. Deformation is com-monly localized in oxide-rich zones, suggestingthat these rocks may be weaker than the enclosingolivine gabbro. This could explain the associationbetween the abundance of magmatic oxides andcrystal^plastic deformation intensity in the bot-tom 1 km of the hole (Fig. 4). All the major oxideconcentration peaks, with the exception of the1250 mbsf peak, coincide with crystal^plastic de-formation intensity peaks. However, Unit IV, themost massive interval of oxide gabbro in the hole,for the most part, shows only weak macroscopicsub-solidus deformation. This suggests that theremay be more to the association of deformationand oxides than simple sub-solidus strain localiza-tion in weaker units.

6. Metamorphism and alteration

The rocks from Hole 735B preserve a complexrecord of high-temperature metamorphism, brittlefailure and hydrothermal alteration beginning atnear-solidus temperatures and continuing down tozeolite-facies conditions. Although alteration as-semblages in the Leg 176 core are broadly similarto those in the upper 500 m [4,19,20], the deeperrocks are less altered with long intervals (s 300m) having less than 5% background alteration.Secondary minerals can be divided into threemain groups (Fig. 5): (1) a high-temperaturegranulite- to amphibolite-facies assemblage; (2)lower temperature greenschist to the zeolite-faciesassemblages; and (3) very low-temperature alter-ation; principally including carbonates and claysand mixed-layer chlorite^smectites.

Brown amphibole is an accessory phase in mostLeg 176 gabbros and occurs intergranularly be-tween clinopyroxene, olivine and grains of ilmen-ite and magnetite. In olivine gabbro, it is generallyisolated, partially rimming pyroxene and olivine(6 0.1 vol%). In ferrogabbros, it is ubiquitous,





commonly exceeding 1 vol% and reaching 10% ina few samples. Brown amphibole in some samplesgrades into dark green amphibole, and whereasmost of it appears to be of late magmatic origin,some may be hydrothermal or deuteric.

In hand specimen, milky white secondary pla-gioclase is the most abundant alteration mineral,but includes both hydrothermal and dynamicallyrecrystallized igneous plagioclase. Hydrothermalplagioclase alteration, by itself, is very low, exceptnear metamorphic veins, and over large intervalsof the core amounts to 1% or less, usually con-sisting of minor amounts of actinolite or smectitealong cracks and grain boundaries. Excluding dy-

namically recrystallized plagioclase, dark greenamphibole is the most abundant alteration miner-al, forming reaction rims and local replacementsof olivine and clinopyroxene, and occurring inalteration halos around amphibole veins. Thisamphibole is found in amounts up to 20^25%above 700 mbsf, but is much less common andsparsely distributed below 750 mbsf, and all butdisappears below 1175 mbsf (Fig. 5). Olivine isthe least alteration resistant phase, and networksof irregular cracks cut even the freshest olivine.These cracks are lined with dark, opaque materi-al, likely a mixture of smectites, very ¢nely divid-ed magnetite, ¢brous amphibole and possibly talc.

Fig. 5. Distribution of the principal metamorphic phases in the lower 1000 m of Hole 735B. Curves represent running 5 m aver-ages of visually estimated mineral proportions checked against thin section modes. Secondary plagioclase includes both hydro-thermally altered plagioclase (albitized) and dynamically recrystallized igneous plagioclase.

Fig. 4. Running downhole 5 m averaged Hole 735B structural observations, including unpublished data of H. Dick for the upper500 m. Vein intensity: 0, none; 1, 6 1 per 10 cm; 2, 1^5 per 10 cm; 3, s 5 per 10 cm (n = 3245 veins). The vein intensity run-ning average is the number of veins per 31 structural intervals with the latter representing sections with near uniform veining.Stereo plots are of poles to planar features in the split core working half reference frame.6



Where alteration is more pronounced, ¢ne-grained mixtures of talc, magnesian amphiboleand ¢nely divided magnetite or pyrite replacethe grains.

Many rocks, veins and shear zones exhibit thein£uence of late magmatic hydrous £uids, and inmany cases, these zones have abundant amphi-bole. The high-temperature limit of sub-solidusalteration is represented by recrystallized gabbros,often with aggregates of equant grains with 120³triple junctions and olivine and pyroxene neo-blasts typical of granulite-facies metamorphicconditions (s 700^1000³C). Many of the shearzones acted as pathways for later hydrothermal£uids, resulting in the formation of amphiboleand clinopyroxene as deformation continued ondown into the amphibolite-facies (V450^700³C).Amphibole often partially replaces pyroxene inductily deformed rocks down to 1100 mbsf. Itforms reaction rims, tails and pressure shadowsas well as discrete grains in segregations alongfoliation planes, and its abundance is directly cor-related to that of secondary plagioclase (Fig. 5).Since the latter is mostly dynamically recrystal-lized plagioclase, this indicates a direct link be-tween high-temperature hydrothermal alterationand deformation.

Alteration at moderate temperatures corre-sponding to the greenschist-facies (V250^450³C)is sparse. Actinolite, however, may locally com-prise up to 50 vol% of the rock. It occurs almostexclusively along the margins of clinopyroxenegrains where it may extend into the crystals alongcleavage planes and into the adjacent plagioclase.Prehnite and epidote are the only calc-silicatesrecognized as vein forming minerals in handspecimen. Epidote was found in only two veins,though one of these was 12 mm wide. It alsooccurs sporadically as matrix alteration in thegabbro. Secondary quartz occurs rarely with oth-er greenschist-facies minerals and there are sixquartz veins near the bottom of the hole. Prehniteveins were found in the deepest part of the hole.Greenschist-facies assemblages are best developedin local breccia zones cemented by felsic materialin the upper 500 m. These may be local up£owzones, overprinting networks of igneous felsic

veins that formed during degassing at the end ofigneous crystallization [18,20].

Low-temperature alteration (6 300³C) includesthe local formation of abundant chlorite^smectiteminerals, carbonate veins, sul¢des and oxyhy-droxide minerals. Chlorite^smectite is the thirdmost abundant alteration mineral and occursabove 600 mbsf as orange to reddish patches par-tially or completely replacing olivine near carbo-nate and clay veins. Dark green to pale bluishgreen clay appears deeper in the hole aroundclay veins, with a pronounced gap in occurrencebetween 800 and 1100 mbsf.

On Leg 176, we described 2792 veins represent-ing 21 di¡erent assemblages. Total vein abun-dance is less than 1% as compared to 2.39% inthe upper 500 m [4]. Close to half are ¢lled withsmectite and clays (1016) and carbonate (293),with most of the latter concentrated between 500and 600 mbsf. Smectite and clay veins occurlargely from 575 to 833 m and from 1054 to1500 m and their occurrence may be related toseawater percolation in open fractures and faultsat 560 m and between 690 and 700 m. Overall, thedistribution of the veins corresponds well to thematrix alteration. The principal high-temperatureveins decrease sharply downsection (e.g. Fig. 5).Felsic veins constitute V0.45 vol%, amphibole-bearing V0.24 vol% and diopside-bearing veinsV0.04 vol%, as opposed to 0.63 vol%, 0.92vol% and 0.33 vol%, respectively, in the upper500 m [4]. Whereas felsic veins are volumetricallymost abundant, many, particularly those associ-ated with oxide-bearing gabbros, are clearly mag-matic in origin. Many of the latter have a stronghydrothermal overprint, while other felsic veinsappear entirely hydrothermal in origin. Quiteoften, the origin of a particular felsic vein is en-tirely ambiguous. Though rare below 1100 m, am-phibole veins are by number the second mostabundant type (17% of all veins). They are almostentirely monomineralic in the Leg 176 section,though the larger ones may contain feldspar. Am-phibole occasionally forms anastomosing ¢ne veinnetworks, though most veins are subvertical,around 0.5 mm wide, and may be orthogonal tothe foliation where present.



7. Rock magnetism

Cores from Hole 735B have a reversed polaritymagnetization (mean 71.4+0.3/33.1 [3]) with nosigni¢cant downhole trend (Fig. 1). This is steeperthan that expected from an axial geocentric dipole(V52³), indicating a tectonic tilt of about 19 þ 5³.The average remanent intensity of samples fromLeg 176 is V2.5 A/m, nearly identical to the esti-mated in situ intensity found during Leg 118 [21].The uniform inclination of both fresh and alteredrocks suggests relatively rapid cooling accompa-nying unroo¢ng at the rift valley wall [22]. Thus,the gabbros are an ideal source for marine mag-netic anomalies and the 1.5 km drilled can readilyaccount for the lineated marine magnetic anomalyfound over Hole 735B [6].

8. The bulk hole composition

Table 2 gives the calculated bulk compositionsof Hole 735B and its upper, middle and lower 500m sections. The calculations are shown in the Ta-ble 2 electronic appendix and were done using themeasured thicknesses of the di¡erent lithologicintervals, the shipboard XRF analyses and 452density measurements. The composition of theupper 500 m is from Dick et al. [4]. A feature ofthe bulk compositions is that they are all close tothose of various primitive to moderately di¡eren-tiated basalts. The major element composition ofmany gabbros, however, is close to that of basal-tic melts for most elements simply because thecotectic proportions of olivine, plagioclase andpyroxene have a basaltic bulk composition.Hence, most major elements vary little in multiplysaturated MORB over a large crystallizationrange until iron oxide minerals appear on theliquidus [23]. Thus, the interpretation of mostHole 735B gabbros as cumulates is based on tracerather than major elements.

The calculated 500 m bulk compositions aresigni¢cantly di¡erent, with nearly a factor of threeenrichment in TiO2, a strong increase in totaliron, a doubling of Fe2O3 and a large decreasein Mg# from 72.9 to 65.9, from the lowermostto the uppermost section. Such features are nor-

mally associated with cumulate sequences pro-duced by progressive fractional crystallization.However, molecular Ca/{Ca+Na}, which shouldcovary with Mg# if this were the case, remainsvirtually constant (Fig. 3), and highly compatibleCr more than doubles. All the other trace ele-ments, compatible and incompatible, remain vir-tually constant. These contrasting variations pre-clude simple upward progressive fractionalcrystallization of basaltic magma in a single intru-sion from explaining the overall stratigraphy.Rather, the best explanation, consistent with thestratigraphy, appears to be a hybrid origin inwhich the olivine gabbros are locally intrudedby late magmatic liquids, especially in the upperthird of the hole. The sources of these melts (ob-viously compacted out of olivine gabbro intru-sions similar to those comprising the bulk of thesection) may either lie below or laterally out ofthe Hole 735B section.

Although the total composition of the lowercrust is unknown, the bulk composition of theentire crust must be that of the average parentalmagma passing the crust^mantle boundary. Basedon the composition of abyssal dunites, this mustlie close to experimentally predicted primary meltcompositions with respect to Mg/(Mg+Fe) ratio[24,25]. This can be inferred because dunites rep-resent zones of focused melt £ow out of the shal-low mantle and therefore constrain the primarymelt composition with respect to equilibriumwith olivine [26^28]. The average composition ofthe upper crustal section (dikes and pillow lavas)originally overlying Hole 735B can be approxi-mated from the average of basalts from AtlantisII Fracture Zone and the eastern rift valley.Although most of these samples are pillow lavas,where both dikes and pillow lavas can be com-pared, such as at ODP Hole 504B and in theTroodos Ophiolite, their compositions are verysimilar [29,30]. The Atlantis II Fracture Zone ba-salts are moderately di¡erentiated, and plot farfrom likely parental melt compositions (Fig. 3).Therefore, they must have di¡erentiated withinthe crust or the shallow mantle prior to eruption.As can be seen from Fig. 3, the bulk compositionof Hole 735B itself, plotting to the left of theinferred locus of primary mantle melts, cannot



mass balance any of the Atlantis II Fracture Zonebasalts back to a reasonable primary composition.Accordingly, a considerable thickness of primitivecumulates complementary to the lavas, dikes andHole 735B gabbros is missing from the drilledsection.

The missing cumulates may lie below the hole,in a crustal section that remains undrilled, or theyare in the underlying mantle. They may also lieout of the section, either because of faulting orbecause they crystallized closer to the mid-pointof the paleo-ridge segment. The latter possibilityis compatible with models of focused melt £owfrom the mantle toward the mid-point of theridge, followed by lateral intrusion of di¡erenti-ated melt down-axis in the lower crust towardsHole 735B [12,31]. This is also consistent withthe small proportion of gabbro compared withbasalt and peridotite dredged from SW IndianRidge transforms, which demonstrates that ifthere is a normal gabbroic layer 3 anywhere be-neath the ridge, it must thin dramatically nearthese faults [12]. Moreover, gabbro suites sampledwell away from fracture zones at the CaymanTrough [32], DSDP Hole 334 west of the MARat 36³53PN [33] and at 7³16PE on the SW IndianRidge [34] all have average olivine compositionsgreater than Fo80. Thus, their average composi-tions, unlike that of Hole 735B, plot considerablyto the right of the primary magma trend in Fig. 3,with the potential of mass balancing MORB backto a primary magma composition. The possibilitythat the missing cumulates are in the underlyingmantle section may explain why the Moho here isdeep, without requiring it to be an alterationfront, or requiring an anomalous thickness ofgabbroic crust. However, dunites are rare fromSW Indian Ridge transforms, and at the AtlantisII Fracture Zone, in particular, there are only1.9% in 973 kg of peridotite in 16 dredge hauls[6]. This suggests that there was little melt trans-port in the shallow mantle near transforms [28]making this latter possibility unlikely.

9. Discussion

The innumerable late oxide gabbro bodies, their

upward increase in abundance, the association ofmagmatic and crystal^plastic foliations with oxideabundance, and the numerous intrusions of oxide-rich melts along small shear zones distinguish theHole 735B section from gabbros in most well-documented ophiolites or layered intrusions. De-spite the abundance of the ferrogabbros, no erup-tive equivalents (ferrobasalts) have been found.Instead, simple equilibrium relations show thatthe Atlantis II Fracture Zone basalts are comple-mentary to the olivine gabbros [18]. This samesituation has also been noted for gabbroic rocksdredged from ¢ve other SW Indian and CentralIndian Ridge fracture zones where ferrogabbrosare also anomalously abundant [29]. This is con-sistent with the supposition that the Fe^Ti-richmelts that produced the ferrogabbros were de-rived from late magmatic liquids squeezed fromthe `mesostasis' of crystallizing olivine gabbro in-trusions. Evidence for this interpretation lies inpatches of oxide gabbro in undeformed olivinegabbro deep in the hole, which likely representlocal pockets of late melt trapped within the oli-vine gabbro at the end of crystallization.

The melts producing the ferrogabbros formedby extended high-iron di¡erentiation [35,18], andcan constitute only a small residual fraction of aprimary basalt magma (10^20%) since ilmeniteand magnetite appear on the liquidus only afterabout 90% crystallization [23,36]. Moreover, sincethe oxidation state of MORBs is very low [37],simple normative calculations show that eventhe most iron- and titanium-enriched ferrobasaltliquid will produce less than 4^6% iron^titaniumoxides when solidi¢ed. Thus, the thick sequencesof massive oxide gabbro, with an average of closeto 10% modal oxide [4] and the numerous smallpatches and small shears ¢lled with oxides, arenot simple ferrobasaltic intrusions. Rather, theylikely represent cumulates from which late silicicmelt was squeezed out to leave an oxide-rich res-idue, or through which ferrobasalt melt migratedby permeable £ow, impregnating the protolithwith excess oxides as it went.

The absence of erupted ferrobasalts comple-mentary to the ferrogabbros and the similarityof the bulk hole composition to that of basaltsuggest that the intricate igneous stratigraphy of



Hole 735B is produced by essentially closed sys-tem di¡erentiation and redistribution of melt. Themechanism by which a large volume of late meltcan be compacted out of an olivine gabbro ma-trix, then aggregated and ¢nally intruded intocooler parts of the crust without eruption to thesea£oor is not clear. We believe, however, thatthis mechanism re£ects intrusion and crystalliza-tion of magmas in the dynamic environment of aslow-spreading ocean ridge where the crust is con-structed of numerous small intrusions, and asteady-state magma chamber is absent (e.g.[4,35,36,38]).

One might imagine such a process as beginningwhere an intrusion intersects an existing fault thatextends into the lower crust from the rift valleywall, or by initiation of a fault within a still par-tially molten body that is su¤ciently rigid to sup-port a shear stress. Where such faults cut the im-permeable rocks enclosing an intrusion, theaccompanying deformation, with local grainboundary sliding and cataclasis, is likely to en-hance permeability and thus create pressure gra-dients. In such a situation, with somewhere to go,late intergranular melt will be compacted out ofthe crystallizing gabbro and migrate into andalong the shear zones. As it moves into coolerrocks, it will crystallize and react with the enclos-ing gabbro. With the melt mass decreasing, ilmen-ite and magnetite will appear on the liquidus andprecipitate, providing a mechanism to enrich ox-ides locally in the gabbro by adcumulus growthfrom migrating melt. Based on other cumulatesequences, the early £ow of intergranular meltas it migrates out of the olivine gabbro will leavelittle textural imprint. As melt aggregates intonarrow zones as it migrates into the shear zones,large £uxes might locally produce magmatic foli-ations due to preferred growth of plagioclase crys-tals in the direction of £uid £ow. As melt movesalong a shear zone, however, brittle^ductile andlocal crystal^plastic deformation will accompanypermeable £ow. In the end, as the rocks solidify,and at higher levels where melt is not present,deformation would continue in the crystal^plasticregime with a transition to cataclasis at shallowerlevels. In this way, it is possible to account for the

association of oxides, crystal^plastic deformationand magmatic foliations in Hole 735B.

The absence of ferrobasalt, then, re£ects themode of melt aggregation and transport. Withno steady-state magma chamber, and with on-going deformation, late melts formed during crys-tallization of small ma¢c intrusions would eithercrystallize in situ or migrate by permeable £owalong shear zones into cooler sections of the crustwhere they would freeze without erupting to thesea£oor. In contrast, at the East Paci¢c Rise,where ferrobasalt £ows are common [39,40], anearly steady-state melt lens has been found tounderlie most of the ridge [41]. Submersible sam-pling and ODP drilling during Leg 147 at HessDeep recovered the uppermost gabbros crystal-lized beneath the rise which plausibly representthe frozen residues of such a melt lens [42]. Theseare largely ¢ne- to medium-grained gabbronoriteand olivine gabbronorite cumulates which crystal-lized from liquids ranging from ferrobasalt to fer-roandesite in composition, and contain no evi-dence of high-temperature crystal^plasticdeformation such as that found in the Hole735B section. The East Paci¢c Rise, with muchhigher rates of mantle upwelling and melt input,also has a steep thermal pro¢le, as indicated bythe melt lens which requires temperatures of1000³C or higher to be sustained just below thesheeted dikes. In this situation, since temperaturesof around 1190³C correspond to the onset of pla-gioclase crystallization at the base of the crust,late Fe^Ti-rich melts, with relatively low liquidustemperatures of 1100³C or less [43], would notcrystallize in the lower crust. Instead, they wouldtend to migrate by compaction and re-intrusion tohigher levels, aggregating to form the melt lens,from which they could erupt through dikes to thesea£oor. In Hole 735B, however, the presence ofoxide segregations and felsic veins even near thebottom of the hole demonstrates that conditionsrepeatedly dropped to near 1000³C deep in thelower crust, probably following each in£ationêruption cycle.

Melt transport during deformation can accountfor many of the features found in Hole 735B.However, melt migration along shear zones isnot the only melt migration mechanism evident



in the cores. The felsic veins and their oxide-richreaction zones clearly demonstrate fracture-con-trolled late-stage melt migration. In addition, thecumulus character of the olivine gabbros, which istypical of gabbros in many environments, by itselfrequires extensive melt migration by means ofsimple compaction, porous £ow or solution chan-neling. Thus it is likely that several di¡erent mo-dalities existed for both early and late-stage meltmigration, even though their relative importanceremains unknown.

The role of deformation in lower crustal petro-genesis in the Hole 735B gabbros continued intothe sub-solidus regime, particularly in controllingthe circulation of hydrothermal £uids and altera-tion. This is evident in the sharp downhole de-crease in amphibole abundance and coincidentdrop-o¡ in crystal^plastic deformation intensity.This suggests a strong drop in circulation ofhigh-temperature hydrothermal £uids downhole,with only very localized penetration of £uidsalong vein networks at depth. Given the relativeabundance of both high- and low-temperature al-teration, the sparse greenschist-facies alteration isquite striking. However, this is consistent withrapid cooling from amphibolite-facies conditionsto low temperatures during unroo¢ng and blockuplift at the inside-corner high [4]. The low-tem-perature oxidative alteration associated withzones of brittle fracture re£ects slower coolingwith the re-establishment of a quasi-steady-stategeotherm after uplift and unroo¢ng of the section.

The close association of oxide-rich and oxide-poor gabbro at Hole 735B and the importance ofdeformation-related processes in the hyper- andsub-solidus regimes are also seen in gabbrosfrom other slow-spreading ridges, including theCayman Trough [44], and Indian Ocean [17] andAtlantic [45,46] fracture zones. Thus Hole 735Bappears representative of the lower crust sampledat slow-spreading ridges, at least that formed nearlarge fracture zones. Hess Deep East Paci¢c Risegabbros, and those in ophiolites believed to havealso formed at fast-spreading ridges (e.g. Oman),on the other hand, lack intimately intercalatedferrogabbro and olivine gabbro, extensive crys-tal^plastic fabrics, and have a di¡erent patternof alteration [47,48]. They contain a microscopic

amphibole vein assemblage responsible for perva-sive (10^50%) alteration of gabbros at tempera-tures of 600^700³C under static conditions inthe near-axis regime. Moreover, the lack of well-de¢ned igneous layering at Hole 735B contrastswith layering in the lower two-thirds of theOman section and the report of layered gabbrosat Hess Deep [49]. These di¡erences, then, sup-port the long-standing inference that accretionaryprocesses in the lower crust are highly sensitive tospreading rate.

Hole 735B in many ways is unlike gabbro se-quences in any well-described major ophiolite.Several features stand out. Massive layered gab-bros and wehrlitic intrusions prominent in manyophiolites such as Oman and Troodos are miss-ing. These could be present beneath the hole orout of section; however, they are also absent inthe many gabbro dredge hauls from Atlantis IIFracture Zone [6]. Some features of the Hole735B section, including the sparcity of layeringand numerous small intrusive units, are found inophiolites believed to be formed in slow-spreadingenvironments, such as the Trinity or Josephine.However, several of the major features of Hole735B have not been described in these ophiolites.These include (1) apparent synkinematic igneousdi¡erentiation resulting in intercalations of oxideand olivine gabbros at all scales, (2) the concen-tration of ferrogabbros near the top of the sec-tion, and (3) the sharp downward decrease of theintensity of crystal^plastic and brittle^ductile de-formation. Whereas the Bay of Islands ophiolitehas associated high-temperature deformation andoxide gabbro intrusion [50,51], it also has massivelayering and wehrlites. The Ligurian ophiolitegabbros have some similarities to Hole 735B, no-tably extensive crystal^plastic deformation andcoarse grain size [52]. However, they lack anyconnection to a sheeted dike or pillow sequence,and were intruded through old continental mantleduring continental breakup [53]. Together withthe well-known arc-related geochemical a¤nities[54], then, most ophiolites are not good analogsfor Hole 735B. Although an exact on-land coun-terpart may not exist, the relatively small LizardOphiolite [55,56] may be a reasonable match;careful study of ophiolites may reveal more. The



scarcity of comparable ophiolites, however, is notsurprising, since obduction of ocean crust fromold, cold slow-spread lithosphere is far less likelythan obduction of crust from hot young litho-sphere formed in small basins, fore-arcs andback-arc ocean environments.

Acknowledgements

We thank the JOIDES Resolution crew, in par-ticular Captain Ed Oonk. Ed retired followingLeg 176, but like most of the crew and many ofthe scientists was also on Leg 118. His good senseand humor were greatly appreciated through ourmany ups and downs, and all the scientists of theOcean Drilling Program will miss him. The ODPtechnical sta¡ was magni¢cent. USSSP providedsupport for manuscript preparation and for addi-tional post-cruise analysis to the co-chief scientistsincluding USSSP Grant F000650/418925-BA102.We also thank Don Elthon and Sherman Bloomeras well as one anonymous reviewer for their help-ful comments.[CL]

References

[1] P.T. Robinson, R. Von Herzen et al. (Eds.), Proc. ODP,Initial Reports, Ocean Drilling Program, College Station,TX, 1989, 826 pp.

[2] J.W. Bown, R.S. White, Variation with spreading rate ofoceanic crustal thickness and geochemistry, Earth Planet.Sci. Lett. 121 (1994) 435^439.

[3] H.J.B. Dick, J.H. Natland, D.J. Miller et al. (Eds.), Proc.ODP, Initial Reports 176 [CD-ROM], 1999.

[4] H.J.B. Dick, P.S. Meyer, S. Bloomer, S. Kirby, D. Stakesand C. Mawer, Lithostratigraphic evolution of an in situsection of oceanic layer 3, in: R.P. Von Herzen, P.T.Robinson et al. (Eds.), Proc. ODP, Sci. Results 118,Ocean Drilling Program, College Station, TX, 1991, pp.439^540.

[5] P.L. McFadden, A.B. Reid, Analysis of paleomagneticinclination data, Geophys. J. R. Astron. Soc. 98 (1982)16033^16052.

[6] H.J.B. Dick, H. Schouten, P.S. Meyer, D.G. Gallo, H.Bergh, R. Tyce, P. Patriat, K.T.M. Johnson, J. Snowand A. Fisher, Tectonic evolution of the Atlantis II Frac-ture Zone, in: R.P. Von Herzen, P.T. Robinson et al.(Eds.), Proc. ODP, Sci. Results 118, Ocean Drilling Pro-gram, College Station, TX, 1991, pp. 359^398.

[7] H.J.B. Dick, C.J. MacLeod, P.T. Robinson, S. Allerton

and M. Tivey, A high resolution bathymetric map of At-lantis Bank, in: H.J.B. Dick, J. Natland, J. Miller et al.(Eds.), Proc. ODP, Initial Reports 176 [CD-ROM], BackEnvelope Map, Ocean Drilling Program, College Station,TX, 1999.

[8] M.R. Muller, C.J. Robinson, T.A. Minshull, R.S. South-west Indian Ridge?, Earth Planet. Sci. Lett. 148 (1^2)(1997) 137^151.

[9] H.J.B. Dick, H. Schouten, P.S. Meyer, D.G. Gallo, H.Berg, R. Tyce, P. Patriat, K. Johnson, J. Snow and A.Fisher, Bathymetric map of the Atlantis II Fracture Zone,Southwest Indian Ridge, Proc. ODP, Sci. Results 118,1991.

[10] D.A. Vanko, D.S. Stakes, Fluids in oceanic Layer 3: evi-dence from veined rocks, Hole 735B, Southwest IndianRidge, Proc. ODP Sci. Results 118 (1991) 181^215.

[11] J.H. Natland, P.S. Meyer, H.J.B. Dick and S.H. Bloomer,Magmatic oxides and sul¢des in gabbroic rocks fromODP Hole 735B and the later development of the liquidline of descent, in: R.P. Von Herzen, P.T. Robinson et al.(Eds.), Proc. ODP, Sci. Results 118, Ocean Drilling Pro-gram, College Station, TX, 1991, pp. 75^112.

[12] H.J.B. Dick, Abyssal peridotites, very slow spreadingridges and ocean ridge magmatism, in: A.D. Saundersand M.J. Norry (Eds.), Magmatism in the Ocean Basins,Geological Society Special Publication No. 42, 1989, pp.71^105.

[13] R.J. Kinzler, T.L. Grove, Corrections and further discus-sion of the primary magmas of mid-ocean ridge basalts,1 and 2, J. Geophys. Res. 98 (1993) 22339^22347.

[14] S.R. Hart, J. Blusztajn, H.J.B. Dick, P.S. Meyer, K.Muehlenbachs, The ¢ngerprint of seawater circulation ina 500-meter section of ocean crust gabbros, Geochem.Cosmochem. Acta 63 (1999) 4059^4080.

[15] L. Angeloni and H.J.B. Dick, Troctolitic gabbros fromHole 735B, Atlantis II Fracture Zone, Southwest IndianRidge, AGU Fall Meeting, 1990.

[16] K. Ozawa, P.S. Meyer, S.H. Bloomer, Mineralogy andtextures of iron^titanium oxide gabbros and associatedolivine gabbros from Hole 735B, Proc. ODP Sci. Results118 (1991) 41^73.

[17] S.H. Bloomer, J.H. Natland, P.S. Meyer and H.J.B. Dick,Textural and mineralogical variations in gabbroic rocksfrom Hole 735B, in: R.P. Von Herzen, P.T. Robinson etal. (Eds.), Proc. ODP 118, Ocean Drilling Program, Col-lege Station, TX, 1991, pp. 21^40.

[18] H.J.B. Dick, P.T. Robinson and P.S. Meyer, The plutonicfoundation of a slow-spreading ridge, in: R.A. Duncan etal. (Eds.), Synthesis of Results from Scienti¢c Drilling inthe Indian Ocean: Geophysical Mon. 70, American Geo-physical Union, 1992, pp. 1^39.

[19] P.T. Robinson, H.J.B. Dick and R. Von Herzen, Meta-morphism and alteration in oceanic layer 3: Hole 735B,in: R.P. Von Herzen, P.T. Robinson and et al. (Eds.),Proc. ODP, Sci. Results 118, Ocean Drilling Program,College Station, TX, 1991, pp. 541^552.

[20] D. Stakes, C. Mevel, M. Cannat and T. Chaput, Meta-



morphic stratigraphy of Hole 735B, in: R.P.a.R. VonHerzen (Ed.), Proc. ODP, Sci. Results 118, Ocean DrillingProgram, College Station, TX, 1991, pp. 153^180.

[21] J.E. Pariso, P.H. Johnson, Do lower crustal rocks recordreversals of the Earth's magnetic ¢eld? Magnetic petrol-ogy of oceanic gabbros from ocean drilling program Hole735B, J. Geophys. Res. 98 (1993) 16013^16032.

[22] E. Kikawa, J.E. Pariso, Magnetic properties of gabbrosfrom Hole 735B, Southwest Indian Ridge, Proc. ODP Sci.Results 118 (1991) 285^307.

[23] D.A. Clague, F.A. Frey, G. Thompson, S. Rindge, Minorand trace element geochemistry of volcanic rocks dredgedfrom the Galapagos Spreading Center Role of crystalfractionation and mantle heterogeneity, J. Geophys. Res.86 (1981) 9469^9482.

[24] H.J.B. Dick, T. Bullen, Chromian spinel as a petrogeneticindicator in abyssal and alpine-type peridotites and spa-tially associated lavas, Contrib. Mineral. Petrol. 86 (1984)54^76.

[25] H.J.B. Dick and J.H. Natland, Late stage melt evolutionand transport in the shallow mantle beneath the EastPaci¢c Rise, in: K. Gillis, C. Mevel, J. Allan et al.(Eds.), Proc. ODP, Sci. Results 147, Ocean Drilling Pro-gram, Texas A and M University, College Station, TX,1996, pp. 103^134.

[26] J.E. Quick, The origin and signi¢cance of large, tabulardunite bodies in the Trinity Peridotite, Northern Califor-nia, Contrib. Mineral. Petrol. 78 (1981) 413^422.

[27] A. Nicolas, A melt extraction model based on structuralstudies in mantle peridotites, J. Petrol. 27 (1986) 999^1022.

[28] P.B. Kelemen, N. Shimizu, V.J.M. Salters, Extraction ofmid-ocean-ridge basalt from the upwelling mantle by fo-cused £ow of melt in dunite channels, Nature 375 (1995)747^753.

[29] W.R.A. Baragar, M.B. Lambert, N. Baglow and I.L. Gib-son, The sheeted dike zone in the Troodos Ophiolite, in:J. Malpas, E.M. Moores, A. Panayiotou and C. Zeno-phontos (Eds.), Ophiolites: Ocean Crust Analogues,1990, pp. 37^52.

[30] W. Bach, J. Erzinger, J.C. Alt and A.H. Teagle, Chemis-try of the lower sheeted dike complex, Hole 504B (Leg148): In£uence of magmatic di¡erentiation and hydro-thermal alteration, in: J.C. Alt, H. Kinoshita, L.B. Stok-king and P.J. Michael (Eds.), Proc. ODP 148, OceanDrilling Program, College Station, TX, 1996.

[31] S.H. Bloomer, J.H. Natland and R.L. Fisher, Mineralrelationships in gabbroic rocks from fracture zones ofIndian Ocean ridges: evidence for extensive fractionation,parental diversity and boundary-layer recrystallization,in: A.D. Saunders and M.J. Norry (Eds.), Magmatismin the Ocean Basins, Geological Society Special Publica-tion No. 42, 1989, pp. 107^124.

[32] D. Elthon, Petrology of Gabbroic Rocks from the Mid-Cayman Rise Spreading Center, J. Geophys. Res. 92(1987) 658^682.

[33] F.N. Hodges, J.J. Papike, DSDP Site 334 Magmatic cu-

mulates from oceanic layer 3, J. Geophys. Res. 81 (1976)4135^4151.

[34] P.S. Meyer, H.J.B. Dick, G. Thompson, Cumulate gab-bros from the Southwest Indian Ridge, 54³S^7³16PE im-plications for magmatic processes at a slow spreadingridge, Contrib. Mineral. Petrol. 103 (1989) 44^63.

[35] R. Hebert, M. Constantin and P.T. Robinson, Primarymineralogy of Leg 118 gabbroic rocks and their place inthe spectrum of oceanic ma¢c igneous rocks, in: R.P. VonHerzen, P.T. Robinson et al. (Eds.), Proc. ODP, Sci. Re-sults 118, Ocean Drilling Program, College Station, TX,1991, pp. 3^20.

[36] T.C. Juster, T.L. Grove, M.R. Per¢t, Experimental con-straints on the generation of Fe^Ti basalts, andesites andrhyodacites at the Galapagos Spreading Center, 85³W and95³W, J. Geophys. Res. 94 (1989) 9215^9247.

[37] D.M. Christie, I.S.E. Carmichael and C.H. Langmuir,Oxidation states of mid-ocean ridge basalt glasses, 79,1989, pp. 397^411.

[38] E.G. Nisbet and C.M.R. Fowler, The Mid-Atlantic Ridgeat 37 and 45³N: some geophysical and petrological con-straints, Geophys. J. R. Astron. Soc. 54, 631^660.

[39] J.H. Natland, W.G. Melson, Compositions of basalticglasses from the East Paci¢c Rise and Siqueiros FractureZone, Near 9³N, Init. Rep. Deep Sea Drill. Proj. LIV(1980) 705^723.

[40] J.M. Sinton, R.S. Detrick, Mid-Ocean Ridge magmachambers, J. Geophys. Res. 97 (1992) 197^216.

[41] R.S. Detrick, P. Buhl, E. Vera, J. Mutter, J. Orcutt, J.Madsen and T. Brocher, Multi-channel seismic imaging ofa crustal magma chamber along the East Paci¢c Rise, 326,1987, pp. 35^41.

[42] J.H. Natland and H.J.B. Dick, Melt migration throughhigh-level gabbroic cumulates of the East Paci¢c Rise atHess Deep: the origin of magma lenses and the deepcrustal structure of fast-spreading ridges, in: C. Mevel,K.M. Gillis, J.F. Allan and P.S. Meyer (Eds.), Proc.ODP 147, Ocean Drilling Program, College Station, TX,1996, pp. 21^58.

[43] S.E. DeLong, C. Chatelain, Trace element constraints onaccessory-phase saturation in evolved MORB magma,Earth Planet. Sci. Lett. 101 (1990) 206^215.

[44] D. Elthon, Petrology of gabbroic rocks from the Mid-Cayman Rise Spreading Center, J. Geophys. Res. 92(1987) 658^682.

[45] A. Miyashiro, F. Shido, M. Ewing, Diversity and originof abyssal tholeiite from the Mid-Atlantic Ridge near 24³and 30³ north latitude, Contrib. Mineral. Petrol. 23 (1969)38^52.

[46] M. Cannat, J.A. Karson, D.J. Miller et al., Initial Re-ports, Ocean Drilling Program, College Station, TX,1995, 798 pp.

[47] C.E. Manning and C.J. MacLeod, Fracture-controlledmetamorphism of Hess Deep gabbros, Site 894: con-straints on the roots of mid-ocean-ridge hydrothermalsystems at fast-spreading centers, in: C. Mevel, K.M. Gil-lis, J.F. Allan and P.S. Meyer (Eds.), Proc. ODP 147,



Ocean Drilling Program, College Station, TX, 1996, pp.189^212.

[48] C.E. Manning, P.E. Weston and K.I. Mahon, Rapidhigh-temperature metamorphism of East Paci¢c Rise gab-bros from Hess Deep, Earth Planet. Sci. Lett., 1996.

[49] R. Hekinian, D. Bideau, J. Francheteau, J.L. Cheminee,R. Armijo, P. Lonsdale, N. Blum, Petrology of the EastPaci¢c Rise crust and upper mantle exposed in Hess DeepEastern Equatorial Paci¢c, J. Geophys. Res. 98 (1993)8069^8094.

[50] A. Berclaz, J.H. Bedard, R. Hebert, V. Varfalvy, B.Melanc°on, K.R. Yao, L. El Bilali, Tectonic evolution ofthe plutonic section of the Bay of Islands Ophiolite, EOSTrans. Am. Geophys. Union 75 (44) (1994) 650^651.

[51] J.F. Casey, J.F. Dewey, P.J. Fox, J.A. Karson and E.Rosencrantz, Heterogeneous nature of oceanic crust andupper mantle: a perspective from the Bay of Islandsophiolite complex, in: C. Emiliani (Ed.), The Sea, TheOceanic Lithosphere, 7, John Wiley and Sons, Inc.,1983, pp. 305^338.

[52] R. Tribuzio, M. Tiepolo, R. Vannucci, P. Bottazzi, Trace

element distribution within the olivine bearing gabbrosfrom the Northern Apennine ophiolites (Italy) evidencefor post-cumulus crystallization in MOR-type gabbroicrocks, Contrib. Mineral. Petrol. 134 (1999) 123^133.

[53] E. Rampone, A.W. Ho¡man, I. Raczek, Isotopic con-trasts within the Internal Liguride ophiolite (N. Italy)the lack of a genetic mantle-crust link, Earth Planet.Sci. Lett. 163 (1998) 175^189.

[54] J.A. Pearce, S.J. Lippard and S. Roberts, Characteristicsand tectonic signi¢cance of supra-subduction zone ophio-lites, in: B.P. Kokelaar and M. Howells (Eds.), MarginalBasin Geology, Geological Society Spec. Pub. 16, 1984,pp. 77^94.

[55] L. Hopkinson, S. Roberts, Ridge axis deformation andcoeval melt migration within layer 3 gabbros evidencefrom the Lizard Complex, UK, Contrib. Mineral. Petrol.121 (1995) 126^138.

[56] G.A. Kirby, Layered gabbros in the Eastern Lizard,Cornwall, and their signi¢cance, Geol. Mag. 115 (1978)199^204.



Date post:	25-May-2020
Category:	Documents
Upload:	others
View:	0 times
Download:	0 times

A long in situ section of the lower ocean crust: results...

Documents