JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 95, NO. B3, PAGES 2727-2752, MARCH 10, 1990
Recordsof GeomagneticReversalsFrom VolcanicIslandsof FrenchPolynesia
2. PaleomagneticStudyof a Flow Sequence(1.2-0.6 Ma) From the Islandof Tahiti
and Discussion
of Reversal
Models
ANNICK CHAUVIN AND PIERRICKROPERCH
1,2
ORSTOM et Laboratoire de Gdophysique
Interne, Universitdde RennesI, Rennes,France
ROBERT
A. DUNCAN
Collegeof Oceanography,OregonStateUniversity,Corvallis
A volcanic sequencealmost700 metersthick has been sampledin the Punaruuvalley on the island of
Tahiti, southerncentralPacific Ocean.Detailedpaleomagneticresultshavebeenobtainedfrom 123 sites.Three
reversalsarerecordedin thissequence.
Age determinations
(K-Ar) indicatethattheyoungestreversalcorresponds
to the Matuyama-Brunhestransition while the two other transitionslimit the Jaramillo normal polarity
subchron.An apparentR-T-R excursionhas been identified lower in the volcanic sequenceand K-Ar age
determinationsaround 1.1 Ma suggestthat it correspondsto the Cobb Mountain subchron,but no normal
paleomagnetic
directionswerediscovered.
The Matuyama-Brunhes
andthe lowerJaramillotransitionare defined
by only a few intermediatedirectionswhile many intermediatedirectionsare observedfor the upperJaramillo
transitionand the CobbMountainexcursion.We attributethesedifferencesto variationsin the rate of eruption
of the volcanic rocks.The lower Jaramillorecord is characterizedby a steepeningof the inclination at the
beginningof the reversalsuggestinga possibleaxisymmetriccontrolof the field at this stage.However, the
transitionpathfor themostdetailedrecord(upperJaramillo)is characterized
by largeloops;thispreventssimple
modelingof the transitionby low orderzonalharmonicsat all stagesof a reversal.Paleointensity
determinations
were attemptedon 48 sampleswith reliableresultsobtainedfor 26 of them.Paleointensities
for the transitional
field rangefrom 3 to 8 gT. Suchvery low field strengthswere first suggested
by the low intensityof the natural
remanentmagnetizationassociated
with intermediatedirections.An analysisof the variationof the intensityof
magnetizationwith the angulardeparturefrom the centralaxial field, includingall otheravailabledata from
Polynesia,indicatesthat (1) a paleomagnetic
directionshouldbe consideredas intermediatewhen it is morethan
30øfromtheexpectedaxialdipoledirection,and(2) theaveragetransitional
geomagnetic
field intensityis about
1/5 of its value duringstablepolaritystates.Comparisonof theseresultsobservedin Polynesiawith previous
studiesfrom Icelandsuggests
an increaseof the averageintensityof the intermediatefield with latitude.
INTRODUCTION
symmetrywere then proposedby Hoffman [1977] and Hoffman
and Fuller [1978]. However, evidence for non-axisymmetric
componentsled to 'flooding'modelswhich includedboth north[Le Mouel, 1984; Gire et al., 1986; Gubbins, 1987; Bloxham,
1988] hasprovidedsomeinsight,evenif sometimescontroversial, southand east-westcomponents[Hoffman, 1979, 1981a]. Similar
usinglow orderGausscoefficients,were proposedby
on fluid motion in the Earth's core, leading to a better approaches,
understanding
of the geomagneticdynamo.Paleomagneticstudies Williams and Fuller [ 1981]. Such models have been under active
of the behaviorof the geomagneticfield duringpolaritytransitions considerationby paleomagnetistsbecausethey have provided a
constituteanotherimportantway to constrainthe variousmodels guideline for the analysis of records of the intermediate field.
of the origin of the Earth'smagneticfield. In this paper, we will However, in orderto determinethe morphologyof the transitional
recordswith betterresolution
focus on the secondapproachin reporting some new data on field, morenumerouspaleomagnetic
are
required.
In
particular,
widely-spaced,
multiple records of
reversalsrecordedby a sequenceof volcanicflows from the island
individual reversals and successive reversals at the same site are
of Tahiti (FrenchPolynesia).
The lastpolaritytransitions(the Matuyama-Brunhes
and
Since the beginningof the studyof reversals[Sigurgeirsson, necessary.
1957; Van Zijl et al., 1962] a great number of paleomagnetic the boundariesof the JaramilloandOlduvaievents)are alreadythe
recordsof transitionshave beenobtained.The first major stepwas best documented;adding new records of these transitionswill
provide detailed understandingand constrainmodeling of the
to recognize the non-dipole characterof the transitionalfield
The studyof the secularvariationof the Earth'smagneticfield
[DagleyandLawley, 1974;Hillhouseand Cox, 1976]. Becausethe
transitionalpath lay in a restrictedsectorof longitude,Hillhouse
and Cox [ 1976] suggested
thatreversalsrecordedat one sitemight
be controlledby a standingnon-dipolefield. Models with axial
behavior of the transitional field.
D6veloppementen Coop6ration,ORSTOM, Paris,France
authors[Hoffman and Slade, 1986; Prdvot et al., 1985b] have
questionedthe reliability of suchrecords,particularlywhen they
have low sedimentation
rates.In a recentstudyof the MatuyamaBrunhesreversalrecordedat Lake Tecopa,Valet et al. [ 1988a]have
shownthat a strongoverprintfrom the following normalpolarity
SinceHoffman [1979] showedthe necessityfor additionaldata
from the southernhemisphere,severalrecordshave beenobtained
from southern low latitudes [Clement and Kent, 1984; Clement
and Kent, 1985; Hoffman, 1986] as well as from the northern
hemisphere[Clementand Kent, 1986; Liddicoat, 1982; Theyet et
1Also
atCollege
ofOceanography,
Oregon
State
University,
Corvallisal., 1985; Herrero-Bervera, 1987; Valet et al., 1988b]. However,
2Permanently
atInstitut
Franqais
deRecherche
Scientifi-que
pour
le
most of these data come from sedimentary rocks and several
Copyright1990 by the AmericanGeophysicalUnion
Papernumber89JB03452
0148-0227/90/89JB-03452505.00
2727
2728
CHAUVIN ET AL.: GEOMAGNETICREVERSALS,FRENCHPOLYNESIA,2
periodwas not removedin the early studyof Hillhouse and Cox which only one-third is subaerial.The erosionof the island has
[1976]. Becausefew studieshave usedthermaldemagnetizations carved radial valleys which enable the samplingof volcanic
[Valet et al., 1988a,b;Laj et al., 1988], the extentof overprinting sequences.
A few transitionalpaleomagneticdirectionswere first
in the other available records is uncertain.
reportedby oneof us [Duncan,1975] in thePunaruuvalley,which
Only relativechangesin paleointensitycan be obtainedfrom is one of the deepestvalleys of Tahiti Nui (Figure l a). This
sedimentaryrocks.The variationsof the magneticpropertiesin the discoverysuggested
thatsomevolcanicactivityoccurredat reversal
sedimentsare generallyrecognizedthroughthe variationsof an boundariesandthat this volcanicsequence
mightprovidedetailed
artificial remanence,such as an isothermal or an anhysteretic behavior of the transitional field.
remanentmagnetization,but this methodassumesthat the NRM
GEOLOGY AND GEOCHRONOLOGY OF VOLCANIC ROCKS FROM THE
is not altered by a multicomponent magnetization. The
PUNARUU VALLEY
normalizationtechniqueis alsouselessif geochemicalalterations
The
thick
succession
of seaward-dipping
lava flows exposedin
haveoccurredduringdiageneticprocesses
[Karlin andLevi, 1985].
the PunaruuValley was eruptedduringthe main phaseof shield
Most estimates of the time duration of reversals come from
at the Tahiti Nui volcano.Theseare typicallyblocky
sedimentary sections.Because several studies have shown a construction
significantdecreasein intensitybeforeany changesin directions, 'aa'flows with rubbly oxidizedtop surfacesandmassiveinteriors,
the estimatesvary from a few thousandyearsto severalthousand easilytracedfor hundredsof metersalongthe canyonwalls. The
years,dependingon how thetransitionboundaries
aredefined.
thicknessof the flows varies between 1 and 4 m, with a mean of 3
m. Rapiderosionduringperiodicstormsandactivequarryinghave
While sedimentsprovide only averagefield directionsand
exposedfresh outcropsurfaces.Compositionallytheserocks are
relative paleointensitychangesin the best cases,volcanicrocks
generally enable a well-determined paleofield direction and alkali basalts,but rangeconsiderablyin phenocrystcontentfrom
to aphyricvarieties.Samples
sometimesaccuratedeterminations
of the absolutepaleointensity. porphyritic(picritesandankaramites)
with
T74
prefixes
have
been
described
by Duncan [1975].
However, with volcanic recordsof reversals,the time resolution of
Complete
geochemical
and
mineralogical
analysesof the dated
the transition is generally not well establisheddue to a poor
sampleswill be reportedelsewhere.
knowledgeof therateof extrusionof the lava flows.
Samplesselectedfor K-Ar datingwere first examinedin thin
A commonfeaturein all recordsis a significantdecreasein the
section
to determinethe stateof alterationof the K-bearingphases.
paleointensityof the field during transitions,but up to now most
In
most
cases the rocks are petrographicallyfresh and wellstudieshave focussedon the directionalbehaviorduringreversals
alteration
rather than analyzing the field as a complete vector. When crystallized.Sampleswhich showedpatchygroundmass
paleointensitydata are added to the directional records, some to clays or glassymesostasiswere excluded.Samplesbearing
unexpectedcharacteristics
of the intermediatefield appear,which cumulateolivine and rare small peridotite xenoliths were also
are not shown by directional data alone. The most complete eliminated becauseof the possibilityof mantle-derivedargon
descriptionof the behaviorof the geomagneticfield, in direction cardedin xenocrysts.
After removal of any weatheredsurfaces,each sample was
and intensity,comesfrom the studyof the volcanicrecordof a
crushed
to chips(0.5 to 1.0 mm size)andultrasonicallycleanedin
Miocene agereverseto normaltransitionon the SteensMountains
distilled
water.A representative
splitof thisfractionwaspowdered
of SE Oregon[Mankinenet al., 1985,Pr•vot et al., 1985a].Very
rapidfield variations(a few thousandgammasper year) havebeen and K-concentration was determined by atomic absorption
spectrophotometric
methods.Argon wasextractedfrom the coarse
suggestedby Coe and Pr•vot [1989], but have not yet beenfully
chips
by
radio-frequency
inductiveheatingin a highvacuumglass
established.Apart from the large decreasein intensity during
transition, which is the main characteristic, Shaw [1975, 1977]
lineequipped
with38Arspike
pipette
andTi-TiO
2 getters
for
has also suggestedthat the intermediatefield may occasionally
reach some high intensity values. On the other hand, large
variations in paleointensitywithout changesin direction were
shown by Shaw [1977] and by Prdvot et al. [1985b]. We will
discusswhy somepaleointensities
may not be valid dueto several
experimentalproblems.
Volcanicrocksprovidereliableandprecisespot-readings
of the
paleomagneticfield but it is not easyto obtaina sequencethat has
recordeda detailedpolarityreversal;generallyonly partialrecords
are found [Bogue and Coe, 1984]. In this paper, we report
paleomagneticresultsobtainedfrom a volcanic sequencealmost
700 metersthick on the islandof Tahiti (FrenchPolynesia,south
removing active gases.The isotopiccompositionof argon was
thenmeasuredmassspectrometrically
usinga modelAEI MS-10S
instrumentwith computer-controlled
peakselectionanddigitaldata
acquisition
andreduction.
The agedeterminations
on 12 samplesfrom the Punaruuvalley
sectionappearin Table 1. The positionsof thesedatedsamplesare
shownin Figure lb. With the exceptionof the age of sampleT8576 which appearsto be slightly too young,the age successionis
consistentwith the stratigraphicorderof the lava flows. The ages
andmagneticpolaritiesof the rocksare alsocompatiblewith the
late Matuyama to early Brunhes portion of the
magnetostratigraphic timescale (see later discussion of
paleomagnetictransitions).This 700 m thick sectionwas erupted
between1.2 and0.6 Ma, yieldingan averageaccumulationrate of
1.2 km per million years, or approximatelyone flow every two
thousandyears. The evidence from the paleomagneticstudies,
however,indicatesthat flows were eruptedin pulseswith significanttime gaps,ratherthanat regularintervals.
central Pacific Ocean). Within this section,four transitionalzones
have
been
intermediate
found
and
two
of
them
have
recorded
various
directions.
GEOLOGICAL SETTING
The Societyarchipelago,in the southerncentralPacific Ocean,
is composedof volcanicislandswhichare alignedin a WNW-ESE
direction delineating the Pacific plate motion during the last 5
m.y. [Duncan and McDougall, 1976]. The island of Tahiti is the
largest of the Society Islands and is made of two coalesced
volcanos (Tahiti Nui and Tahiti Iti). The Tahiti Nui volcano is a
conerising6000 metersabovethe surroundingabyssalseafloor,of
PALEoMAGNETIC
SAMPLING
All the paleomagneticsampling has been done within the
Punaruuvalley andwascarriedout duringtwo field trips.The first
one was a preliminarysamplingduringwhich 150 coresfrom 25
lava flows were collected.Preliminaryresults,associatedwith the
CHAUVINET AL.: GEOMAGNETIC
REVERSALS,
FRENCHPOLYNESIA,2
•øs o
•
SOCIETY
2729
ISLANDS
• C•
• •HUAHINE
(a)
17
C
-,
.......
• • TAHITI
1._.
8øs
I
I
150
I -w
148"
I w
I
(b)
(c)
(b)
•t
•au
v
O CK
OpL
(,-,
0 TU
B.,,
•R1T
:.,.
(1)
'"'""
•
R1R
•
.•:..'
• M2B
--R1S
•R1M
R1L
R1Q
R1P
R10
• M2C
--: • M2D
•• M2E
R1N
M2F
R11
R1G
•
R1B
(R1)
M2K• PN
• TL
M2M
M2N
• TG
TI
-•
_..
CC
CB
M2J
TE
•
CD
M2H
M21
• TJ
•R1E
CF
CE
•M2G
•CA
R1K
R1J
• R1F
M2W
• U20
• ....
.:..
:';
M1E
'½•: CS
'
M1B
M2S
-
(T)
(2)
(M1)
PI
PF
TC?:;-.
CT••M1C
TB
PJ
(M2)
PC
PB
PA
(P-C)
Fig. 1. a) Map of the Societyislands(FrenchPolynesia,south-central
PacificOcean).The paleomagnetic
samplinghasbeendone
alongthe PunaruuValley in the westernpart of the islandof Tahiti (b). Flows are dippingtowardthe west.On the northsideof
the valley,the samplingwasdonealongthe river asindicatedby the 3 sectionsa, b, c. Filled circlescorrespond
to flowsbelow
(siteBKV) and abovesectionsa, b, c. Squaresare the samplingsitesof the dikes.On the southernborderof the river, numerous
crosssectionsof the hillsidehavebeendoneandare indicatedby lettersreferencedin the text. The magneticpolarity(blackfor
normal)hasbeenshownwithin eachsection.Gapsin the samplingare alsoindicatedby discontinuities
within eachsection.Stars
correspond
to the locationof thedatedsamples.
2730
CHAUVIN ET AL.: GEOMAGNETIC REVERSALS•FRENCH POLYNESIA, 2
TABLE 1. K-Ar Age Determinationsfor Basaltsfrom the PunaruuValley, Tahiti
SampleNumber
%K
Radiogenic
40Ar/g
%Radiogenic
40Ar
(x10-7cc)
Age+lc•
(Ma)
BKV-185
T85-72
0.83
0.86
1.7156
1.5745
27.9
15.2
1.19+0.02
T85-76
BKG-85
PKB- 14
1.03
2.07
1.16
1.7688
3.9120
2.2268
10.8
24.4
18.4
1.01+0.05
T85-81
1.03
1.8259
24.4
1.05+0.02
T85-93
T74-431
1.28
1.62
2.1740
2.8538
19.8
17.0
1.00+0.02
T74-434
T74-435
T74-437
T85-67
1.16
1.34
1.66
2.19
1.8886
2.0978
2.1991
2.3630
11.8
17.8
14.2
6.6
0.96+0.02
1 .O8+O.O2
1.09+0.02
1.11 +0.03
1.04+0.02
0.92+0.02
0.78+0.01
0.62+0.02
Age
calculations
based
onthefollowing
decay
and
abundance
constants:
)h•=0.58
lx10
-10
yr-1,)•[•=4.962x
10-10y
r-1.40K/K=
1.167x
10
-4mole/mole.
Jaramillo termination and the Brunhes onset, were reported by
Roperch and Chauvin [1987]. More detailedsamplingduring the
secondfield trip covereda distanceof 4 km into the valley. This
samplinginterval correspondsto a volcanic sequencealmost700
Then 4 flows with reversedirectionswere sampled:BKX on the
north side of the river and BKY, BKZ, BKZA on the south side in
an old quarrywhere the stratigraphicorderwas easyto recognize.
Two otherreverseflows PKE andPKG were sampledabout300 m
meters thick.
west of the previouslocation. The dike PKF which cutsthesetwo
A total of 788 cores have been drilled from 123 lava flows and
flows has a normal polarity. The river sites from flow B KV to
4 dikes, with an averageof 6 to 7 coresper volcanic unit. Cores PKG correspond
to the lower part of our sequence.
were obtainedwith a gasoline-powered
portablecoredrill, andthen
The secondpart of the samplingwas madealongthe southern
orientedusinga magneticcompassand, wheneverpossibleby sun wall of the valley. We tried to sampleall flows which could be
orientation.We have distributedthe samplesthroughoutthe flows, observedin the field but it was necessaryto do several small
vertically and horizontally, to minimize risks of systematic sectionsbecauseof a lack of outcropsor becauseof topographic
deviation of the paleomagnetic directions due to local block difficulties, like small waterfalls. There were no indications either
disturbances;in many cases, the location of the samples was in the geological, geochronologicalor paleomagneticdata for
determinedby accessibility.During the samplingwe useda small differencesin stratigraphybetweenthe two sidesof the river. No
field spinnermagnetometer(LETI) in orderto recognizethe polar- faultshave been observedin the valley and youngvalley-filling
ities of the flows and map the stratigraphy.The intensity of the flows, found further interior, are absenthere.
naturalremanentmagnetization(NRM) was often better than the
The sectionP-C (flows PA to PN and CA to CK) beginsnear
NRM direction in recognizing transitional zones. Because of the river, about 350 m away from the flow PKG. A reverse
viscouscomponentsfrom the presentday field, sometransitional polarity was observedfor the bottomtwo flows, and a transitional
flows showeddirectionscloseto a normal polarity but with very zone was recorded by the overlying flows (PB,PC,PE,PF). A
low magneticintensity.This informationobtainedwith a portable small baked soil, 50 cm thick, was interbeddedbetween flows PC
spinner magnetometer was more accurate than fluxgate andPE but we were unableto sampleit. No suchwell definedsoil
measurementsof orientedblocksbut, obviously,was more time was seen elsewhere in our sequence.As it does not mark the
transition from full reverse to full normal, but is interbedded
consuming.
The naturaldip of the lava flows is closeto 10 degreestoward betweentwo flows having a similar direction,this soil horizon is
the ocean.This fact allowed us to sampleeither along the river or not an indicatorof a large time gap betweentwo flows, in contrast
by vertical sections(Figure 1). The first part of the samplingwas to what we expected. A normal polarity was found above the
donealongthe river. A lava flow (BKV) with a reverseddirection transitionup to an elevationof 125 m (flows PG to PN and CA to
wasfoundat the baseof the sequence.
BetweenBKV andthe next CK).
Section M2 extends from an elevation of 30 m to 120 m above
flow (BKA), no outcropwas found.From BKA to BKG, a reverse
polarity was observed.Following this reversesuccession,there the sea level. From M2U to M2G only normal directionswere
was a smallgap in the samplingof a few tensof meters,followed observed.The top sevenflows from M2E to M2V in this section
by a transitionalzone which contained16 lava flows (from BKH have transitional directions. Section M1 is about 400 m west of
to BKT2), over a horizontal distanceof 350 m. One dike (BKM)
sectionM2 and is composedof 6 flows (M1A to M1F) with
with a reverseddirection cuts the transitionalsequenceand was intermediatedirections.The particular paleomagneticdirections
samplednearflow BKL.
recordedby the flows from the sectionsM 1 are very similar to that
observed at flows CS, CT, CU (named section 2 on Figure l)
Immediately above BKT2, we found 3 lava flows
which are situated 100 m west of section M1.
(BKS,PKB,BKW) and one dike (BKU) with reversepolarity. No
normal direction was observed in this part of the volcanic
SectionT is close by and just above flows CS, CT, CU and
sequence.No outcropwas available for the next 250 m along the starts at an elevation of 30 m; 7 flows (from TA to TG) have
river and taking into account a slope of ten degrees, this transitionaldirectionsand the upper 2 flows (TJ and TL), at an
correspondsto a vertical gap of almost40 metersin the sequence. elevationof 150 m, have a reversepolarity. We were unable to
CHAUVINET AL.: GEOMAGNETIC
REVERSALS,
FRENCHPOLYNESIA,2
2731
samplebetweenaltitudesof 150 to 250 m in this section.At 250
m (section 1, Figure 1), one flow of normal polarity (TU) was
found overlying two flows (TR and TT) with intermediate
Susceptibility
directions.
thermomagnetic
measurements
were
performed
invacuum
(10-2
The last section(R1) showeda new polarity transition.From
an elevation of 50 m above sea level to 170 m (flows R1A to
Weak field susceptibilities
were measuredusingthe Bartington
susceptibility meter, at room temperature. Strong field
Torr) or in air, using an automatic recording Curie balance.
Heating and coolingrateswere 8øC per minute, with an applied
R1S), a reversepolarity zone was observed,overlain by a small
transition zone (20 m thick) composedof 3 flows (R1T, R1U,
field of between 0.1 to 0.7 Tesla.
R1V). The intermediate direction of the flow R1V is similar to
that of flow TR which had been sampledat the top of the section
T. Above the flow R1V, the last flow PL has a normal polarity
like flow TU at the top of the section1.
Paleointensity
From this sampling,a syntheticcross-section
of the valley has
been constructed(Figure 2), which showsthe stratigraphicorder
and the geomagnetic polarities recorded. According to the
radiometricdatingandthepolaritiesobserved,we havesampledthe
Brunhes-Matuyamareversalat the top of the volcanic sequence,
followed down-sectionby the upper and lower Jaramillo event
boundariesand the Cobb Mountain event.
LABORATORY PROCEDURES
Remanence
Each core was cut in standardspecimens(2.5 cm). Remanent
magnetization was measured using a Schonstedt spinner
magnetometer.Stepwisedemagnetizationseither thermal or by
alternatingfield (AF) were performed,at least on one sampleof
eachcore,usinga Schonstedtfurnaceand AF demagnetizer.Most
of the normal and reverse samples were cleaned only by AF
demagnetizationwhile samplesfrom the transition zones were
demagnetized with thermal or AF techniques. Flows with
intermediate
directions
have lower
NRM
intensities
Jaramillo
Cobb
Mountain
(10-2Torr)inorder
tominimize
temperature-induced
alterations
of
the magneticmineralsof the samples.
As with the original Thellier method,at eachtemperaturestep
the two successiveheatingsand coolingswere performedin the
presenceof the artificial laboratoryfield, but the directionof this
field wasreversedbetweeneachheating.This methoddiffersa little
from the modified versionproposedby Coe [ 1967a,b] for which
the first heatingof eachtemperaturestepwas made in zero field
and the secondheating in the laboratory field. The artificial
laboratoryfield was appliedduringthe completetemperaturecycle
[Levi, 1975]. The appliedfields varied from 9 to 40 microteslas,in
orderto minimize differencesbetweenlaboratoryand ancientfield
strength[Konoand Tanaka, 1984].
PALEOMAGNETIC RESULTS
ilBrunhes
/
---
1959], which is more accurate than several other methods for
basalts[Coe and Grommd,1973]. Sampleswere heatedin a quartz
tube surroundedby a coil which producesthe external laboratory
magnetic field. The entire system was shielded from the
geomagneticfield by 3 layersof mu-metalcylinders.Following
Khodair and Coe [1975], the heatingswere performedin vacuum
and even
thoughthe viscousremanentmagnetization(VRM) overprintis
almostthe sameas with normal or reversemagnetizedflows, the
relativeeffect on the initial NRM is greater.AF demagnetizations
were more efficient in removing isothermal(IRM) components,
while thermaldemagnetizations
were betterin cleaningviscousremanentmagnetization(VRM) from weakly magnetizedsamples.
Primary componentswere easily identified and sometypical
demagnetization
diagramsare illustratedin Figure3.
---
Paleointensity determinations were performed using the
original Thellier's doubleheatingmethod [Thellier and Thellier,
Remanence directions
When only stepwiseAF cleaningwas usedto demagnetizea
lava flow, the meandirectionwas calculatedat the onepeak of the
alternatingfield which provided the best grouping in the data.
When boththermalandAF cleaningswere performedon the same
flow, primaryremanencedirectionswere obtainedby interpretation
of Zijderveld diagrams.For theseyounglava flows, the primary
I
i
500m
[lOOm
i
i
i
POLARITY
transitional
reversed
JARAMILLO
0.92
normal
----0.62
Matuyama
Ua
0.78
i
0.96
1.04
1.00
1
1.11
COBB
MOUNTAIN
1.09 1.01
1.08
1.19
Fig. 2. Schematic
crosssectionshowingthe magnetostratigraphy
of the volcanicsequence,
andthe corresponding
K-Ar age
determinations.
2732
CHAUVIN
ETAL.:GEOMAGNETIC
REVERSALS,
FRENCH
POLYNESIA,
2
N
N• OOøC
E
3O
2O
15
1A
NR•
530(
10mT
Down
87R1SllSB
(AF)
o•
E Up
0.5 Am -1
Down
lO
2 Am -1
20
87M1C18A
(Thermal)
87R1D22C
(Thermal)
30
E
N
I
I
I
I
I
Down
87M2H85B
o__•• $
E
N
(AF)
I
I
I
I
I
NRM
•E'Up
•5rn'I'
20
RMO••5•
N t -1 0.5Am-1 '•450
•'125øC
0.5 Am -1
87BKQ154B
• •
Down
•
•
', •
87BKH96B
•
•
0.5
Am
Down
(Thermal)
Di •wn
87R1T120B
(Thermal)
S
(AF)
Fig.3.Typical
orthogonal
diagrams
ofprogressive
thermal
(TH)oraltemating
field(AF)demagnetizations
ofnormal,
reverse
and
intermediate
samples.
Filled
circles
correspond
totheprojection
onto
thehorizontal
plane
while
open
symbols
areprojections
onto
thevertical
plane.
Asseenin these
plotsthesecondary
magnetization
is mainly
a northcomponent
suggesting
a viscous
overprinting
in therecent
Brunhes
normal
polarity
period.
Theprimary
component
isalways
welldefined
eitherwiththermal
or
AF demagnetization.
componentwas considered
a.spassingthroughthe origin. In the directionsas well as the high intensitysuggestthat the
computingmeandirectionper flow, we omittedanomalous
cores overprintof this flow is an isothermalremanentmagnetization
wheneverfield notesindicatedpossibleorientationmistakes,or (IRM) due to lighting.It wasobviousin the field that flow BKW
when their directionswere at more than 2 standarddeviationsfrom was highly weathered.Demagnetization
by AF was unableto
themeandirection
of theflow.Onlythreeflowsfailedtoprovide
a completelyremovethe secondary
component
of magnetization,
reliablepaleomagnetic
direction.In the caseof the flow M2E, although
theevolutionof theremanent
vectorsuggested
a reverse
neitherAF nor thermaldemagnetization
was able to isolatea polarityprimarydirection.On the otherhand,duringthermal
stableprimaryremanence.
This flow belongsto the upper demagnetizations
the samplesfrom this flow explodedin the
Jaramillo transition,and has a very weak intensity of furnaceat temperatures
above300øC.Thusno paleomagnetic
magnetization
with an overprintof higherintensitythanthe direction was determined for this flow.
primaryremanence.
Anotherflow fromtheJaramillotermination,
Paleomagnetic
resultsarelistedin Table2. In Figure4, flow
TI, showeda strongoverprint.Samplesfrom thisflow havean meaninclinations,declinations,
and virtualgeomagnetic
poles
unusually
highremanence
intensity
(morethan100Am-1)and (VGP) areplottedasa functionof thestratigraphic
positionin the
showa veryquickdropin intensity
for smallpeakAF values(5 to valleysection.The altitudes
reportedon therightsideof Figure4
10 mT). ForhigherAF peaks,theintensitydecrease
is lower,but have been calculatedfrom the thicknessof the lava flows, their
the directionsdo not changeup to 100 mT, which is the averagedip, their true altitudeandtheir locationin the section.The
maximumfield of ourAF demagnetizer.
The scatterobservedin stratigraphic
orderof theflowshasbeenwell observedin thefield,
CHAUVIN
ETAL.:GEOMAGNETIC
REVERSALS,
FRENCH
POLYNESIA,
2
2733
TABLE 2. PaleomagneticResults
Inc
k
358.6
2.0
232.9
230.2
235.0
224.5
226.6
176.7
185.3
-50.6
-43.6
--5.5
0.2
0.1
7.3
5.4
53.4
44.8
117
100
182.6
50.6
347
100
200
100
100
200
200
200
100
300
200
AF+TH
AF+TH
AF+TH
200
200
100
181.1
177.1
178.2
176.2
175.2
179.8
174.9
176.9
179.0
191.6
193.2
191.3
143.0
185.8
172.2
168.4
44.9
39.7
33.3
36.7
36.5
36.6
39.4
14.2
11.0
21.6
26.7
24.7
15.6
37.8
34.8
40.9
579
200
200
170.9
175.3
46.3
44.6
N.d
N.d
Site
n/N
AF/TH
Dec
TU
PL
TI'
TR-TS
R1V
R 1U
R1T
R1S
R1R
5/5
6/6
6/6
5/6
5/6
4/5
5/6
6/6
6/6
200
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
100
R1Q
5/6
R 1P
R10
R1N
R1M
R1L
R1K
R1J
RlI
R 1H
R1G
R1F
R1E
R1D
R 1C
R1B
R1A
6/6
6/6
6/6
5/5
5/5
9/11
6/6
6/6
6/6
7/8
6/6
4/5
5/7
5/6
5/6
4/5
TL
TJ
5/5
6/6
Ti
(x95
o•
Lat
Long
P
JlO
X
N
N
2.74
2.18
2.16
3.45
4.56
7.1
7.0
18.1
11.2
76.3
82.0
35.6
17.5
2.2
2.4
12.0
117.0
122.9
119.0
-34.0
-37.7
-33.1
104.6
106.4
108.6
I
I
I
0.72
0.75
0.47
2.32
N.d
N.d
3.5
4.7
6.7
4.4
4.1
2.8
2.5
3.3
2.8
2.5
2.9
1.5
3.5
3.7
5.6
3.7
11.4
5.5
2.6
3.1
2.9
131.7
128.8
159.0
167.1
161.8
167.6
172.5
178.3
174.8
174.4
175.9
172.0
161.4
158.4
165.0
167.1
167.4
142.5
172.9
173.1
167.5
-44.2
-41.9
-73.5
-80.0
-76.2
-81.2
-84.4
-88.2
-85.5
-84.8
-87.3
-83.3
-79.1
-77.8
-77.0
-76.8
-78.1
-52.7
-83.5
-82.5
-77.7
108.1
107.9
220.3
182.1
201.3
204.2
239.5
284.6
262.6
270.0
214.6
255.7
14.1
25.9
92.1
106.7
98.6
310.9
154.1
288.2
270.9
I
I
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
I
R
R
R
1.30
1.26
1.88
4.18
2.61
6.01
6.14
19.47
7.37
11.72
9.48
7.98
8.32
5.99
6.48
5.15
8.94
0.80
2.41
10.52
14.17
N.d
2.36
1.37
1.40
0.93
1.40
1.34
1.69
1.45
1.03
1.04
1.73
2.78
3.77
1.87
1.63
1.68
2.23
1.22
1.13
2.89
4.0
7.0
164.6
167.4
-77.0
-80.4
249.1
236.7
R
R
4.24
2.92
1.34
2.26
11.36
92
1.28
1.55
92
900
1055
42
702
264
101
231
691
405
724
907
311
1940
367
327
115
328
66
214
835
609
994
368
2.27
1.11
N.d
N.d
N.d
1.92
4.93
10.74
10.10
15.44
16.39
41.40
18.20
40.87
32.7
16.53
10.76
5.71
12.43
11.37
19.12
1.3
7.07
33.45
17.61
4.64
TG
TF
TE
TD
TC
TB
TA
6/6
7/7
3/5
3/4
8/8
5/8
8/8
200
200
200
200
200
200
200
133.0
134.0
297.0
299.6
301.3
312.7
337.8
74.4
72.5
42.9
42.6
41.5
39.2
35.0
249
4.2
132.3
-35.6
236.6
I
0.70
1.98
194
4.3
7.1
6.6
3.4
6.8
6.1
133.6
94.9
93.3
91.6
84.1
70.9
-37.6
15.3
17.5
19.3
28.9
46.7
239.5
153.7
154.6
154.7
159.5
178.1
I
I
I
I
I
I
0.72
0.74
0.92
1.04
1.26
1.30
1.67
2.28
1.27
1.87
3.07
2.71
M 1F
CU
M1E
M1D
CT
CS
M1C
M1B
M1A
6/6
3/5
4/5
6/6
5/5
4/5
6/6
6/7
6/7
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
18.2
37.2
17.0
51.6
66.7
53.0
56.1
55.9
69.8
26.0
14.1
44.9
59.8
50.2
57.7
62.0
51.2
54.8
261
4.2
14.3
8.2
6.1
61.1
58.6
(79.0
101.6
53.8
45.6
42.8
14.5
241.5
269.7
231.5
248.5
I
I
I
I
0.20
0.74
1.21
0.16
2.41
2.17
N.d
1.93
19.3
20.9
6.7
15.8
7.7
101.5
100.6
104.7
97.1
105.8
9.6
15.1
10.3
17.0
5.3
263.6
251.1
248.5
258.0
260.9
I
I
I
I
I
0.20
0.15
0.18
0.13
0.29
1.35
1.88
1.16
1.25
1.83
0.53
231.1
238.3
12.7
220.4
213.5
226.3
20.5
12.3
35.0
69.6
58.0
58.1
34
133.2
123.2
68.6
136.7
146.0
-40.1
-32.0
51.0
-42.6
-54.1
119.6
116.8
229.8
178.9
163.3
I
I
I
I
I
0.38
0.39
0.72
0.39
0.27
3.43
2.51
1.46
2.99
N.d
0.39
194
11.6
8.0
6.1
5.1
4.8
N.d
98
7.8
139.9
-44.7
158.1
I
0.43
N.d
N.d
N.d
N.d
401
3.8
2.3
3.8
3.3
4.6
7.9
40.2
8.2
13.8
11.6
13.1
12.9
56.1
81.2
74.7
76.9
76.3
79.3
276.8
290.6
315.1
309.0
323.6
350.8
I
N
N
N
N
N
0.53
2.57
3.58
3.97
2.80
3.92
N.d
0.96
2.81
1.37
1.07
1.82
N.d
7.6
4.7
4.1
6.1
85.3
86.3
301.6
207.0
N
N
1.73
3.53
0.59
1.36
10.57
4.0
2.9
5.1
3.1
3.9
2.2
3.4
4.2
3.8
7.5
12.9
12.7
12.8
19.6
23.5
17.6
2.5
5.8
85.3
81.4
82.4
82.4
78.0
76.3
79.5
87.2
86.3
190.0
175.7
22.6
208.3
184.5
191.7
192.3
293.8
189.0
N
N
N
N
N
N
N
N
N
9.70
3.58
2.49
4.17
4.89
4.69
3.62
3.17
3.11
2.06
0.86
0.79
1.85
1.23
1.75
0.74
0.74
1.61
16.91
M2V
M2W
M2A
M2B
M2C
M2D
6/7
5/5
6/6
6/6
6/6
5/6
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
M2E
299
354
259
128
83
75
126
123
19
20
101
19
77
92
120
172
M2F
M2G
M2H
CK
M2I
M2J
5/5
6/6
6/7
6/7
4/5
4/5
AF+TH
AF+TH
200
200
100
200
30.7
9.0
15.8
13.7
13.6
7.6
4.4
-29.8
-37.3
-34.7
-39.7
-44.0
M2K
C!
CH
CG
M2L
M2M
M2N
M20
M2P
5/6
5/5
4/5
4/4
6/6
6/6
5/6
5/6
6/6
300
200
200
200
100
100
300
300
300
4.9
359.8
358.3
355.0
1.6
359.7
354.7
355.6
356.7
-32.6
-26.4
-25.2
-20.4
- 19.9
-19.7
-13.5
-9.4
-15.2
M2Q
5/6
300
2.9
-32.0
334
CF
7/7
200
358.6
-26.9
247
884
318
408
401
137
101
261
537
1005
174
462
385
1243
379
1.16
2.61
2.01
1.47
1.72
0.30
1.22
N.d
0.30
0.29
0.56
0.38
0.57
0.55
1.76
0.46
9.65
4.57
10.37
9.42
7.73
9.32
14.95
11.32
8.07
14.29
9.64
17.55
15.29
6.92
2734
CHAUVINETAL.:GEOMAGNETIC
REVERSALS,
FRENCHPOLYNESIA,
2
TABLE 2. (continued)
Site
n/N
AF/TH
Dec
M2R
CE
M2S
CD
CC
CB
M2T
M2U
CA
6/6
6/8
4/6
5/5
4/4
6/8
7/7
7/7
5/5
100
200
200
200
200
200
200
TH+AF
200
0.4
10.7
16.2
19.6
1.9
6.0
7.1
12.3
10.6
-27.8
-21.2
-18.6
-14.8
-27.3
-17.8
-19.0
-19.8
-29.7
299
153
2261
251
1395
140
270
1163
520
PN
PM
PJ
7/7
6/7
5/5
200
100
200
354.6
10.4
0.8
-36.5
-33.3
-40.4
Pi
6/6
200
359.3
PH
PG
PF
PE
PC
PB
PO
PA
7/7
7/7
8/8
8/8
7/7
7/7
6/6
7/7
200
200
AF+TH
AF+TH
100
200
100
300
1.9
6.4
157.8
158.6
179.4
182.6
174.4
187.4
PKG
PKE
BKZA
BKZ
BKY
BKX
4/5
5/5
9/9
4/8
6/7
11/12
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
183.8
179.3
178.7
196.2
185.2
175.4
N.d
N.d
5/7
7/7
6/6
5/5
5/6
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
155.5
156.5
332.4
358.7
342.1
48.4
47.6
27.7
45.8
44.5
BKW
PKB
BKS
BKT2
BKT1
BKR
Inc
k
cz95
o•
Lat
Long
P
Jl O
25
Q
3.9
5.4
1.9
4.8
2.5
5.7
3.7
1.8
3.4
4.8
14.8
20.1
25.1
5.0
15.7
15.0
16.8
9.5
87.0
77.7
72.3
68.4
86.6
79.8
79.5
75.9
79.7
218.1
269.1
275.3
275.1
243.0
246.1
252.5
270.1
292.3
N
N
N
N
N
N
N
N
N
3.48
5.16
7.73
9.25
3.92
3.93
5.14
5.97
7.11
2.85
2.27
2.71
2.83
1.26
1.65
1.78
1.78
1.03
4.39
8.16
10.24
11.74
11.18
8.55
10.35
12.03
24.79
102
232
413
6.0
4.4
3.8
6.0
8.8
7.9
84.3
80.1
84.6
92.7
305.0
22.8
N
N
N
7.00
5.14
4.47
2.68
2.59
3.64
9.37
7.13
4.40
-34.0
210
4.6
1.6
88.8
65.9
N
3.16
3.21
3.53
-17.3
-31.6
55.8
57.1
74.4
63.9
47.8
36.8
278
258
233
102
297
175
158
85
3.6
3.8
3.6
5.5
3.5
4.6
5.4
8.8
15.3
5.5
152.8
151.3
138.1
148.6
164.2
172.6
81.0
83.9
-62.9
-62.6
-46.9
-62.0
-77.7
-82.5
222.6
296.0
252.6
249.4
211.0
206.7
234.3
143.7
N
N
I
I
I
I
R
R
4.37
3.84
1.29
2.40
1.39
2.63
3.36
3.33
1.66
1.18
1.95
0.90
0.39
0.38
0.98
0.71
9.45
11.66
2.38
9.55
12.83
24.75
12.31
16.78
27.5
27.5
28.2
32.6
25.0
41.5
407
273
132
106
757
504
4.6
4.6
4.5
9.0
2.4
2.0
174.0
174.9
175.5
166.4
171.2
170.3
-85.2
-86.8
-87.0
-74.6
-83.2
-82.5
80.7
18.3
5.6
123.2
78.9
244.7
R
R
R
R
R
R
14.87
20.37
3.32
3.27
1.13
2.25
1.01
1.25
1.93
1.09
2.35
1.80
52.71
58.41
6.19
10.83
1.73
4.48
67
492
70
87
212
9.4
2.7
8.0
8.2
5.3
155.7
156.7
65.7
78.4
78.7
-64.8
-65.8
47.7
45.1
42.8
268.5
269.3
168.9
209.0
188.5
R
R
I
I
I
0.53
1.14
0.72
0.95
0.85
2.82
1.93
1.51
2.44
N.d
0.68
2.11
1.71
1.40
N.d
BKQ
6/6
AF+TH
339.6
60.5
31
12.3
94.5
27.9
193.4
I
0.50
1.19
1.51
PKD
BKP
BKO
5/6
7/7
7/7
AF+TH
AF+TH
AF+TH
339.9
348.1
351.4
45.8
28.7
25.5
213
992
596
5.3
1.9
2.5
80.4
62.3
58.6
41.0
55.0
57.7
186.7
190.3
194.8
I
I
I
0.56
0.53
0.51
0.97
0.87
1.63
2.07
2.18
1.12
BKN
BKL
BKK
PKC
7/7
5/7
5/5
6/6
AF+TH
AF+TH
AF+TH
AF+TH
351.2
219.7
221.3
220.8
22.6
7.3
-5.7
6.4
185
41
313
272
4.5
12.0
4.3
4.1
65.8
135.4
125.2
134.0
59.3
-48.7
-44.4
-47.5
193.5
105.4
97.9
105.5
I
I
I
I
0.29
0.62
0.92
1.38
1.48
1.88
N.d
1.96
0.70
1.19
N.d
2.53
BKJ
BKI
BKH
BKG
BKF
BKE
PKA
BKD
BKC
BKB
BKA
BKV
7/7
6/6
8/8
9/9
7/7
7/7
7/7
6/6
6/6
7/7
4/5
7/8
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
AF+TH
400
AF+TH
AF+TH
AF+TH
200
200
220.8
160.0
164.3
169.8
177.2
180.3
184.3
180.0
175.9
181.7
186.0
174.5
5.2
-40.3
-48.1
40.2
10.6
13.0
17.9
15.2
15.1
10.6
16.6
30.6
242
924
42
349
590
280
378
318
105
332
678
234
3.9
2.2
8.6
2.8
2.5
3.6
3.1
3.8
6.6
3.3
3.5
4.0
133.2
104.9
98.1
168.8
157.9
160.5
164.9
162.7
162.2
158.0
163.2
174.9
-47.2
-44.9
-40.8
-79.1
-77.3
-78.9
-80.5
-80.0
-79.2
-77.5
-79.1
-84.6
104.6
4.2
12.4
270.3
17.8
32.1
57.3
30.6
8.3
38.5
63.7
312.9
I
I
I
R
R
R
R
R
R
R
R
R
0.39
1.10
0.44
1.77
1.51
1.60
1.71
2.47
2.72
3.52
2.07
2.10
2.11
1.98
3.00
0.42
1.06
0.22
0.56
0.27
0.47
1.02
0.40
3.43
0.66
1.99
0.52
1.51
5.14
26.02
10.90
32.85
20.71
12.43
18.67
2.20
4/7
6/6
6/6
5/5
AF+TH
AF+TH
AF+TH
AF+TH
0.2
191.7
171.9
24.0
-36.5
42.2
32.8
-30.8
69
212
239
132
11.1
4.6
4.3
6.7
N.d
N.d
N.d
N.d
87.4
-77.2
-82.3
67.1
26.5
154.1
298.2
301.5
N
R
R
N
N.d
1.36
0.62
1.35
N.d
1.77
0.25
1.49
N.d
2.75
8.81
3.27
DIKE
PD
BKM
BKU
PKF
n/N,numberof samples
usedin theanalysis/total
numberof samples
collected;
AF/TH, AF+TH indicates
thatthermalandAF demagnetizations
have
beendone,whilea numberrefersto theAF peak(Oe),at whichthemeanwascalculated;
Dec andInc, meandeclination
andinclination;
k, precision
parameter
of Fisher;cz95,95%confidence
coneaboutmeandirection;
•),reversal
angle;Lat andLong,latitudeandlongitude
of VGP;P, polarityof the
flow(N'normal,
R:reverse,
I: intermediate);
JlO,geometric
mean
NRMintensity
after10mT;25,geometric
mean
susceptibility
(10-2SIunits);
Q
geometric
meanratioof J10/(35 gT*z).
CHAUVIN ET AL.: GEOMAGNETIC REVERSALS,FRENCH POLYNESIA, 2
Declination
180
i
Inclination
360
i
-90
0
i
log Jn
log Z
90
-1
0
1
2
-3
-2
i
i
i
i
i
i
i
log Qn
-1
i
-1
0
1
i
i
i
2735
V.G.P.
90S
Latitude
90N
ß
ßß..
¾.
;!,'
ß ß
- 100m
,,,--
:,...,
.:..-ß
- 300m
-600m
i
i
Fig. 4. Evolutionof differentmagneticparameters
for eachflow of the sequence
againstthe calculatedcumulativethickness.Jn,
averageNRM intensityafter 10 mT; Z, meansusceptibility;
Qn, apparentKoenigsberger
ratio(seetext).The axialnormaldipole
inclinationat thatsiteis -32.5ø (32.5ø for reverse).While the susceptibility
appearsto be equivalentthroughout
the section,large
variationsin intensityare seenandthe low valuesare alwayscorrelated
to transitionzones,identifiedon the VGP latitudeplot.
Theselow valuesarealsowell definedwith theapparentKoenigsberger
ratio.
bothalongthe river and within eachverticalsection.To matchthe These flows have purely transitionaldirections(Table 1) which
relativepositionsof the flows betweendifferentvertical sections, correspondto an angulardeparturefrom the axial dipole higher
field observations of flow attitudes were often sufficient. For
than 50ø. Their NRM intensities are also very weak, but their
example,takinginto accountthe distancebetweensectionM1 and corresponding
VGPs have a latitudehigherthan 50ø. It seemsthat
M2 (see Figure 1) and the dip of the flows toward the west, the a betterrequirementfor a directionto be consideredintermediateis
top-most flow of section M2 (flow M2V) is stratigraphically that the reversalangle (measureof the angulardeparturefrom the
beneath the bottom flow of section M1 (flow M1A). The same normal or reverseaxial dipole) would be higher than 30ø. This
argumentswere appliedto sectionsT and R1 (Figure 1) and the requirementwas alreadyproposedby Hoffman [1984] andwe will
flow TL (sectionT) is lower in the stratigraphythan flow R1A see later that, from our data, this cutoff between intermediate and
(beginningof sectionR1). This stratigraphicassignmentis in dipolardirectionsis not totally arbitrary.
goodagreementwith the resultingsequenceof magneticpolarities.
Figure 4 showsthat the thicknessof the transitionalzonesis
In two cases,only paleomagneticcorrelationshave been used. greatly variable. The upper Jaramillo transitionhas a thickness
First, section2 (CS, CT, CU) and sectionM1, which are close to around 80 m while the normal subchron Jaramillo has been
one anotherin the field, have similar paleomagneticdirections recordedonly by a sequenceof 120 m. Taking into accountthat a
which suggests that flows CS, CT, CU are in the same transition period may be 5000 years and, that the Jaramillo
stratigraphiclevel as flows M1A to M1D (CS: D=53.0, 1=57.7; subchronis 60,000 yearslong, great differencesin the extrusion
CT: D=66.7, 1=50.2; M1D: D=56.6, 1=59.8).
rate of the lava flows are then evident.
Correlationsbasedon paleomagnetic
directionsare easierin the
caseof intermediatedirectionsof the geomagneticfield, because NRM intensityand susceptibility
they are more unique. On the contrary, it is more difficult to
The stepwiseAF demagnetizations
show that generallythe
recognizeandcorrelatedirectionsduringstableperiods(reverseor
secondary
componentof magnetizationis cleanedat a peakAF of
normal) of the field. Our sectionsCA to CK and M2U to M2G
10 mT. In order to define the distribution of the intensities of the
both correspondto the Jaramillo normal subchron.Using the
altitudesof the flows, their natural dip and the paleomagnetic primary directions,we report here the NRM intensitiesafter 10
mT of AF cleaning(J10)- The NRM intensitiesare distributed
directions, we combined the two sections. However, we cannot
dismissthe possibility that the sameflow was sampledat both
sites. The stratigraphic order chosen on the basis of the
paleomagnetic
datais indicatedin Table 2.
Definingthe boundariesof transitionsin termsof directionsis
not so easy.Consideringa directionto be transitionalonly when
its corresponding
VGP falls belowa latitudeof 50ø [Sigurgeirsson,
1957], or 40ø [Wilson et al., 1972] is not alwayspossible,as can
be seenfor examplewith flows BKP, BKO and BKN (Figure 4).
between
lessthan0.1Am-1 upto30Am-1,andsamples
with
intermediate directions clearly have lower intensities of
magnetizationthannormalandreversesamples.In contrast,there
is no apparentdifferencein weak field susceptibilityfor both
populations (Figure 5); lowest values correspondto hightemperatureoxidized samples.As a result, the distributionof the
apparentKoenigsberger
ratio, (NRM intensityafterAF cleaningat
10 mT divided by the productof the bulk susceptibilityby an
2736
CHAUVINETAL.' GEOMAGNETIC
REVERSALS,
FRENCHPOLYNESIA,2
Normal
and Reversed
100 •
Intermediate
50•
0.1
I
10
JNRM (10mT)
(Am-1)
10-3
10 -2
10-1
0.1
Susceptibility
(SI)
I
10
100
JNRM (10mT)/Ji (35gT)
Fig. 5. Histograms
showing
thedistribution
of theNRM intensity(after10 mT), thesusceptibility,
andtheratioof theprevious
NRM witha calculated
induced
magnetization
in a fieldof 35 gT. Samples
havebeenseparated
intotwopopulations,
intermediate
and normal+reverse.
ambient field of 35 gT), is similar to that of the intensitiesof
magnetization.
Geometricmeanvaluesof NRM intensities
(J10), weakfield
susceptibilities
and apparentKoenigsberger
ratio calculatedper
flow arereportedin Table 2 andarerepresented
in Figure4. This
figureclearly showsthat the averageNRM intensityper flow is
lower duringtransitionalperiodsthan duringnormalor reverse
periods.This fact is particularlyclearwhenthe transitionalzoneis
well detailed. This pronounceddecreaseis also observedin the
Koenigsberger ratio, because there are no differences in
susceptibility
betweensamplesfrom the transitionsand samples
with a normalor reverseddirection.Theseresultsstronglysuggest
that the variationsobservedin intensityof magnetizationare not
mineralogically
controlledbutreflectchanges
in thegeomagnetic
field intensity.This is easily seenwhen the geometricaverage
NRM intensities,after 10 mT, are displayedversusthe reversal
anglein Figure 6. Althoughthere is considerablescatterof the
data, with a deviation of one order of magnitudebetween some
flows, the distinction between flows with intermediate directions
and thosewith stablepolarity is very clear.In a studyof a N-T-N
excursion
recorded
on lava flows
from
the island of Huahine
(Society islands), a similar distribution of the intensities of
magnetizationis observed[Roperchand Duncan, this issue].The
intensity of the geomagneticfield, during transitions,is thus
expectedto decrease
downto 10-20% of thevalueobservedduring
stablepolaritystates.
Nevertheless,we must emphasize that NRM intensity is a
parameterwhich providesonly a generaltrend of field variations,
when a largenumberof samplesand flows are used.Detailsin the
paleointensitychangesbetween flows can not be resolvedwith
NRM intensities alone. Absolute values of the geomagnetic
intensity are necessaryin order to better constrainpaleofield
variation during reversals.Unlike the sedimentaryand intrusive
rocks,lava flows recorda quasi-instantaneous
field, andthey can
provideaccurateabsolutepaleointensities.
In orderto quantifythe
relativechangein the geomagneticfield intensityreflectedby the
NRM data,somepaleointensitydeterminations
were attemptedand
will be presentedbelow.
STATISTICAL ANALYSIS OF THE DISTRIBUTION OF THE NORMAL AND
REVERSE DIRECTIONS.
A previousstudyof paleosecularvariationrecordedin volcanic
rocks from French Polynesia [Duncan, 1975] indicated the
magnitude of dispersion of the paleomagnetic data was in
agreementwith valuespredictedby secularvariationmodels,with
an angular standard deviation of the site VGP's about the
geographicalaxis of 13.8ø. This result was obtainedfor 53 sites
geographicallydistributed in 5 of the younger islands of the
volcanicchainandrepresentinga 4 m.y. time interval.In contrast,
our sitesrepresenta single volcanic sequencefrom the Punaruu
ß
valley.A statisticalanalysisof the dispersionof thepaleomagnetic
directionsmay provide someinsightto the samplingprocessof
1the magneticfield by sucha volcanicsequence.
The normalpolaritydatasetcomprises31 flowseruptedduring
0.5the Jaramillosubchron,2 during the Brunhesperiodand 2 dikes
(Table 2). As most of the data correspondto the Jaramilloperiod
they might representa maximum time interval of about60,000
years(Figure7). The setof reverseddirectionsis composedof 39
o.1
I
I
I
I
I
I
I
I
I
I
I
I
I
I
!
I
I
lava flows and 2 dikes (Figure 7) and might representa longer
90
180
o
intervalof time (approximately350,000 years).
reversal angle
First, paleomagneticdirectionsmore than 30ø away from the
Fig. 6. Variationof the meanflow intensityof magnetization(after 10 axial geocentricdipolewere excludedfrom the analysis,sincethey
mT) versusreversalangle.Intensityof magnetization
is on a logarithmic have been consideredas transitional. In some studies[McFadden
scale.
and McElhinny, 1984; McElhinny and Merrill, 1975] only VGP
CHAUVIN ET AL.: GEOMAGNETICREVERSALS,FRENCHPOLYNESIA,2
N
W
E
2737
suggests
eithera low level of the non-dipolefield activity,or an
insufficient samplingof the secularvariation possibly due to a
high rate of extrusionof the lavas flows. For the Jaramillo data
set,whetherthe flows were eruptedregularlyover 60,000 yearsor
in oneor a few shortperiodsis an unanswered
question.
For the reverseddirections,the mean VGP is indistinguishable
from the geographicalaxis at the 95% confidencelevel, but the
angularstandarddeviationremainslower thantheexpectedvalueat
thislatitude (meanVGP: 1=268.0ø,L=-88.9ø, K=52.2, (x95=3.2ø,
ST=I 1.2ø).Suchstatisticswith a moderatenumberof samples,are
very sensitiveto the selectionof the data. Adding 5 flows, with
VGP latitudebetween45-50ø,but a reversalanglehigherthan30ø,
the dispersionof the datais in betteragreementwith othermodels
(which used this cut-off level) for the secularvariation. (1=246.0ø,
L=-86.4ø,K=30.0, (x95=3.9ø,ST= 14.8ø).
The mean VGP
S
Fig. 7. Stereographicprojectionsof all normal and reversedpolarity
directionsobservedin the volcanicsequence
from Tahiti.
latitudeslower than45ø were excluded.The non-Fisherianpattern
of the directions (Figures 7a,b) is common for sites at low
latitudes,becauseof the movementof the dipole (wobble). As a
consequence,
the distributionof the VGPs associated
with the
paleomagnetic
directionsis muchmoreFisherian(Figures8 a and
b).
The mean VGP for the Jaramillo flows is: long=263.9 ø,
Lat=84.4ø,with K=87.1, ST=8.7ø and(x95=2.8ø.As wasdiscussed
duringthe description
of the sampling,two sectionscomposed
the
Jaramillo; some sitesbetweenCA to CK perhapscorrespondto
flows alreadysampledin sectionM2. Nevertheless,
evenif the ten
flows between CA to CK are taken out of the computation,the
resultis not changedsignificantly(meanVGP: 1=266.8ø L=85.0ø,
K= 83.1, (x95=3.5ø, ST=8.9ø).At the 95% confidencelevel, the
mean VGP is distinguishablefrom the geographicalaxis, and the
relativelylow valueof the angularstandarddeviationof the VGPs
(a)
for all the normal
and reversed directions
with
VGP latitudeshigher than 45ø has an angularstandarddeviation
very closeto the expectedvalue: 1=327.9ø, L=88.9ø, (x95=2.7ø,
K=34.2, ST=14.0ø. The angularstandarddeviationof the field is
usually
calculated
withthefollowing
formula:
SF2=ST2-SW2/n,
where ST is the total standarddispersionbetweensites,SW the
within site dispersionand n the averagenumber of samplesper
flow. The more inclusiveVGP populationgives a field dispersion
SF=13.9ø (SW=4.6 ø, n=5.9).
This studyof the paleosecularvariationprovidesthe following
conclusions:
The mean VGP
associated
with all the normal
and reversed
directionsis indistinguishable
from the geographicalaxis.
The results from the Jaramillo
subchron indicate either that the
Jaramillo representstoo short a period of time to average the
secular variation
at this site or that the individual
observations
(i.e., eachlava flow) are not independentin time andrepresentonly
a short time in the Jaramillo.
Finally, our estimateof the angularstandarddeviationof the
VGP is identical to the value obtainedby Duncan [1975] and is
very closeto the value given by the different models.(The most
recent (G) of McFadden et al. [1988] gives SF=13.4 ø for the
latitude of Tahiti.)
(b)
Fig. 8. Equalareaprojectionof the VGP'scorresponding
to the previousdirectionsshownon Figure7.
2738
CHAUVIN ET AL.: GEOMAGNETIC REVERSALS, FRENCH POLYNESIA, 2
CobbMountain (Figures9 and 10)
DESCRIPTION OF THE TRANSITIONS
The paleosecular
variationis not well recordedin thebottomof
The directionalpath, before (D,I) and after rotation(D',I')
followingHoffman [1984], for all the transitions
foundin our the section because several reverse flows (BKA to BKF) have
sequence,
aredisplayed
onstereoplots
(Figure9) whilethevirtual recorded statistically indistinguishable directions with low
inclinations.Only the two flows (BKV and BKG) eruptedbefore
geomagnetic
polepathsareshownin Figure10.
Normal
+
Matuyama
+
Brunhes
+
+ + • + + + + --JF+ + + +
!
!
1
Reverse
Upper
Jaramillo
+ + +++
E
Normal
N
se
Lower
+
Jaramillo
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
Normal
N
+
+
+
t..,,'
+
+
+
+
+
+
+
,,' .
',,
.
,oun.ain
f
Reverse
Fig.9. Foreachtransition,
equalareaprojections
of thedirectional
pathsbeforerotation
(left)androtated
directions
(D', I')
[Hoffman, 1984] (right).
CHAUVINET AL.: GEOMAGNETICREVERSALS,FRENCHPOLYNESIA,2
Upper Jarami!!o
2739
Matuyama-Brunhes
I
I
I
I
I
\
L
t
\
•
--R
\\
0 Site
_•_i I
I
//
/
•---•
/
i
Cobb Mountain
/
Lower Jarami!!o
Fig. 10. Evolutionof the VGPs duringeachtransitionplottedon a Mollweideprojection.Samedatasetsas in Figure9.
andafterthisgrouphavedifferentdirections.As discussed
before,
thereis a few tensof metersgap in the samplingbetweenthe last
beginning of the transition. No accurate paleointensities are
availableto confirmthisinterpretation.
reversedflow (BKG) and the next (BKH) which is intermediate.
Thus, the flow B KH might not correspond to the precise
UpperJaramillo (Figures9 and 10)
This reversalboundaryis muchmoredetailedandis recordedby
beginning of the transition but may record a direction well
23 flows.Thefirsttransitional
direction
(flowM2F) hasa very
advanced in the transition and the apparent far-sided VGP
shallow
negative
inclination,
and
shows
a
large swingof the field
movementto position about 160ø away from the site meridian
away
from
its
secular
variation
pattern
during the Jaramillo
shouldbe interpretedcautiously.After that, southwestdirections
(flows BKJ to BKL) associatedwith shallowedinclinationshave subchron.A different direction (southwest and steep positive
inclination)is observedin the threefollowing flows (M2D, M2C,
been recordedby 4 successiveflows. The VGPs associatedwith
thesedirectionscannotbe easilyclassifiedasnearor far-sided.The
following directions(flows BKN to BKT2) are north but with
positiveinclinationsbetween20 and 60ø; in the D', I' spacethey
aresituatedin the near-sidedquadrant,andare90 to 60ø awayfrom
the reversepolarity directionwhile the VGPs are closerto normal
polarity.Justabovetheseintermediatedirections,a directiononly
25ø awayfromthereversedpolaritywasrecordedby two successive
flows. Even though there are some gaps in the sampling,the
M2B). The next transitional direction (M2A) is north with a
moderatepositiveinclination,and it is followedby two southwest
directions(flow M2V, M2W). There is no gap in the samplingof
theseflows. Another transitionaldirection(eastand positive steep
inclination) has been recordedat six sites (M1A, M1B, M1 C, CS,
CT, M1D). The possibility that a gap in the sampling exists
between section M2 and section M1 cannot be ruled out. Then, a
movement
of the directions
to the north
with
a decrease
in
overlyingflows alsohavereversedpolarity.No normaldirections inclination is observed (M1E, CU, M1F). In contrast with this
were observed,sothisrecordis an apparentR-T-R excursion.K-Ar irregular behavior, flows TA to TE exhibit a smooth shift in
datingof five flowsfrom thispartof the sequence
(Table 1) yields directionwhich seemsto correspondto a better samplingin time
a weightedmean age of 1.11+0.01 Ma and indicatesthat this of the field variationor a more regularbehaviorof the field. The
end of the transitionshowssteepand southeastdirections(flows
excursion should be correlated to the Cobb Mountain
event
TF and TG). The VGPs associated with these intermediate
[Mankinen et al., 1978].
directionsare grosslynear-sidedexceptfor two (flows M2V and
LowerJaramillo(Figures9 and 1O)
M2W). All the transitional directions are associatedwith a weak
The transitionis not well sampled;4 flows (Table 2) have a NRM intensity.
The next 5 flows (TJ, TL, R1A to R1C) are reversedirections,
paleomagneticdirection which involves a steepeningin the
with NRM intensitieshigher, and characteristicof the stable
inclination and are followed above by normal polarity flows. In
contrast with what is observed for the other transitions, the NRM
polarity state. But, the next flow above (R1D) has a remanent
intensityof these4 flows cannotbe distinguished
from that of the directionwhich goesout of the typical secularvariation pattern,
underlyingflows. This fact suggests
thattheseflows recordedthe with an angulardeparturefrom the axial dipole of 37ø and low
2740
CHAUVINETAL.:GEOMAGNETIC
REVERSALS,
FRENCHPOLYNESIA,
2
intensity of magnetization.This special direction was obtained
duringthermaldemagnetizationas well as with AF cleaning,and
is well determinedas shownby sample 87R1D22c on Figure 3.
This site was clearly identifiedas a lava flow andnot a sill.
Our recordof the UpperJaramillois clearlycharacterized
by: i)
large changesbetweentwo intermediatedirections(i.e., between
the flows M2A and M2W), ii) progressivevectormovementof the
field, recordedby successive
lava flows (flows CU to TE), and iii)
relative stability in intermediatepositions(flows M1A to CT).
The questionis: are thesepropertiesrelatedto the transitionalfield
behavioror are they just an effect of the sporadicextrusionof the
lava flows?.Quick and large directionalchangescan be attributed
first to gaps in the paleomagneticsampling.We have seenthat
this is not the casehowever,for the threelargervectormovements
recorded by the flows of the top of section M2. Gaps due to
changesin the extrusionrate are moredifficult to verify.
NRM direction. The NRM
direction must be close to the mean
stable direction for the flow obtained after AF and/or thermal
demagnetization.This allowed us to reject samplescarrying
significant viscous remanences (VRM). This criterion was
particularly efficient for sampleswith a transitional direction,
whoseprimary remanentmagnetizationwas very weak, and the
overprintwas often relatively more importantthan for samples
with a non-transitional
direction.
Grain size.Multidomaingrainscausenon-idealbehaviorduring
the Thellier experiment,and a concaveup curvatureof the points
in the NRM-TRM diagram[Levi, 1977]. This effectpreventsthe
use of the low-temperaturepart of the NRM-TRM diagram to
calculatethe paleointensity,
becausethe slopeof the line wouldbe
steeperthanthoseprovidedby the ratio of the true paleointensity
to the laboratoryfield.
Remanence stability to AF demagnetization with median
demagnetizing
fields above 30 mT. was usedto indicatethat the
There is evidencethat the intensityof the Earth'smagneticfield
remanence
is
controlled
by single (SD) or pseudo-single(PSD)
during the transition was very low, as shown by the NRM
domain
particles.
These
experiments
were performedon adjacent
intensityand alsothe paleointensitydata (seenext section).With a
small magneticvector,large and suddenchangesin directionare specimensfrom the samecore.
Thermal stability. Thermal stability of our samples was
easierto producethan with a normal geomagneticfield intensity.
investigated
with strongfield thermomagneticexperiments.The
A simple model of reversals, in which the dipole term of the
degree
of
alteration
during the Js-T experimentswas estimated
sphericalharmonicexpansionis reducedto zero while the other
terms are allowed variations in 'realistic' agreement with the fromthedifferencebetweenthe Curiepointonheatingandthaton
present-non
dipolefield, givesa recordalsocharacterized
by periods cooling and also from the changein the strongfield induced
(Js)at roomtemperaturebeforeandafterheating.It
of progressive changes and periods of important and rapid magnetization
is
well
known
that volcanic rocks with a single high Curie
directionalvariationswhen the magneticfield intensityis greatly
reduced[Roperch,1987]. Similar behavioris also observedin the temperatureare goodcandidatesfor paleointensitydeterminations
becausetheir highly oxidized stategenerallypreventsa further
recent reversal models proposed by Williams et al. [1988].
Thereforethe observedfeaturesof ourrecordof the UpperJaramillo oxidation during the laboratory experiment.Js-T experiments,
transitionmight not be entirelycontrolledby the sporadicvolcanic performedon 111 samples,exhibit four typesof curves(Figure
activity, and some of them might reflect the irregular evolution 11). We assumedthat the Curie temperaturecorrespondsto the
with time of the local field directionsduring intermediatestates. temperatureof the inflection point of the Js-T curve.This method
Nevertheless,it is obviousthat we do not have a completerecord yieldsvaluesgenerallylower thanthoseobtainedusinga linearexof changesin the paleomagneticdirectionsduringthe transition,as trapolationof the Js-T curveas describedby Grommdet al. [1969].
The temperatureof the inflection point can be consideredas the
expectedwith volcanicrocks[Weekset al., 1988].
average of the broad distribution of the Curie point of the
Matuyama-Brunhes
(Figures9 and 10)
individualmagneticparticleswithin a volcanicrock [?rdvotet al.,
Beforethe Matuyama-Brunhes
boundary,a progressive
increase 1983].
The first type of Js-T curvesis the mostcommonencountered
in inclination is observedand we speculatethat this patternhas
some similarities with the Lower Jaramillo. The transition is not
(76 samples)and showsa singleferrimagneticphase,with a high
Curiepointbetween500 and565ø C. The type 1 curve
detailed and only five lava flows have recorded intermediate temperature
directions which are southwest with a shallow inclination and 50 ø
can be divided into three groups, according to the relative
discrepancy
betweenthe heatingandcoolingcurves.Samplesof
to 60ø away from the geocentricaxial dipole.
group l a are characterizedby a decayof the strongfield induced
PALEOINTENSITY DETERMINATIONS
magnetizationat room temperature,Js,after the heating,lessthan
duringthe coolingof
The aim of the Thellier method is to compare partial 10% anda decreaseof the Curie temperature
thermoremanent magnetization (PTRM) acquisition and the only a few degrees.This behaviorhasbeenfoundon 42 samples.
Group lb (13 samples)showsa decreasein Js greaterthan 10%
decreaseof the partial natural remanentmagnetization(PNRM)
with increasingtemperaturesteps.Becauseof the law of additivity after the heating and a more important drop of the Curie
temperatureduring the cooling. The evolutionof the samplesis
of the PTRM, the ratio PNRM/PTRM should be constant over
eachtemperatureinterval and equalto the ratio of the ancientfield morepronouncedwhen heatingsare madein air thanin a vacuum.
strengthand the laboratory field, provided no alteration of the The group l c (21 samples)is characterizedby an increasein the
at roomtemperature
afterheating,sincethe
magnetic minerals occurred during the heating. Only samples inducedmagnetization
whose NRM is an original TRM, with no significant secondary cooling curves cross the heating curves, typically in the
interval250-450øC.The magneticphaseproducingthe
magnetization,and showingno or little thermalinstability,enable temperature
type 1 curvecould be a titanium-richtitanomagnetite
which has
a paleointensitydetermination.
beentransformed
to a Ti-poor titanomagnetite
by hightemperature
Sampleselection
deuteric oxidation, or a primary low-Ti titanomagnetite.
To select the most suitable samples for the paleointensity Discrepancies
betweenthe groupsl a, lb and lc mightresulteither
measurements,
we usedthe following criteria:the NRM direction, from low temperature oxidation which inducesnon-reversible
the magneticmineralogyand the thermalstabilityof the magnetic behavioror minor variationsin the magneticgrainsor bulk rock
minerals.
[Mankinen et al., 1985].
CHAUVIN
ETAL.:GEOMAGNETIC
REVERSALS,
FRENCH
POLYNESIA,
2
1.0-
•,•,.,•
•
BKI98
(V)
TG046
(V)
F=0.2
T,
lb
o
0.0
i
i
i
200
0
•
0
i
400
600
i
0
i
I
200
i
I
400
•
Tc015
(V)
F=0.1
T.
i
i
i
600
Pf43
(A)
F=0.1
T.
lc
2a
1.0-
2741
Type 3 curvesshowonly oneferrimagneticmineralwith a very
low Curie temperature,below 75øC and sometimeslower than the
room temperature.The cooling curvesdo not retracethe heating
curveswith heating performedeither in air or in vacuum. Seven
samplesproducedthiskind of curves.
Only 3 samplesgave type 4 Js-T curves. This type of curve
showssomeparamagnetic
behaviordueto thehighestappliedfield
and thesesampleshave low magneticcontentas indicatedby the
low value of the magneticforce.
As the Js-T experimentswere carried out on selectedsamples
without large secondarymagnetization,they do not representthe
true distributionof the Curie pointsover the entire flow sequence.
For the paleointensitymeasurements,
we have usedonly samples
exhibitingtype 1 Js-T curves.
Interpretationof theNRM-TRM diagrams
0.0-
i
i
i
200
i
i
400
'•
i
i
i
600
i
i
200
i
400
PKD23
(V)
i
600
CE106
(V) ti to room temperaturein the laboratoryfield. Ideally all points
F--0.2 T.
F=0,4
T.
b
2c
0.5-
0.0
0
' 2•0 ' 4•)0 ' 6•)0
PF39
(V)
i
2•
i
i
200
0
F=0.4
T.
i
400
i
600
BKE70
(V)
\•
F=0.6
T.
0.5-
0.0
i
0
200
i
i
400
i
i
600
The paleointensity data are shown in NRM-TRM diagrams
[Nagataet al., 1963] wherethe NRM (Yi) remainingat each
temperature
step(ti) is plottedagainst
theTRM (xi) acquired
from
i
0
i
200
i
I
400
600
Fig. 11. Examples of the four different types of strong field thermomagneticcurves obtainedon samplesfrom the island of Tahiti.
Analyseswere done either in a vacuum(V) or in air (A) with various
appliedfieldsfrom 0.1 to 0.6 Teslas.Arrowsindicateheatingcurves.The
unitofthespecific
magnetization
isgiven
inAm2Kg
-1.
Type 2 curveswere producedby 14 samplesand they result
from the presenceof two ferrimagneticminerals.Thesecurvescan
also be separatedinto 3 classes.In class 2a (8 samples), one
magneticphasecommonlyhas a Curie point between 100 and
200øC,while the otherphasehasa Curie temperaturehigherthan
500øC,but theheatingandcoolingcurvesarenearlyreversible.In
contrast to the previous class, class 2b (9 samples) shows
irreversibility,the coolingcurvesalwaysbeingabovethe heating
curves.Class2c curves(7 samples)are also characterizedby two
Curie points;one above500øCandthe secondlower than 100øC
and in somecaseslower than the room temperature.In all cases,
the Curie pointsof sampleswere changedonly by a few degrees
after the experiment. Bimodal Curie temperaturesamples are
commonly observedin flows where high temperaturedeuteric
oxidation has transformed only part of an originally Ti-rich
titanomagnetiteto a Ti-poor titanomagnetite
containingilmenite
lamellae. The non-reproducibilityof the heating and cooling
curvesof group2b indicatesa low temperature
oxidationandthe
lowerCuriepointmightbe a cationdeficienttitanomaghemite
but
thedisproportionation
peakhasbeenobservedin only onesample.
shouldbe on a straightline whose slope gives the ratio between
the intensity of the ancientgeomagneticfield and the laboratory
field, but generally, only one part of the diagram is linear. The
temperatureat which the secondarymagnetization,suchas VRM,
is cleaned,determinesthe lowest limit of the temperatureinterval
which can be used for paleointensitydeterminations.Because a
linear distributionof the pointson the NRM-TRM diagramis not
a proof of the absenceof chemicalor physicalchangesduringthe
Thellier experiment [Pr•vot et al., 1983], we have to check the
stability of the partial TRM acquisitioncapacity as the heating
temperature
increases(PTRM checks).Any increaseor decayin the
TRM acquisitioncapacityindicatesa physico-chemicalalteration
of the magneticminerals.
Generally, severalPTRM checkswere performedduring each
experiment using larger temperatureintervals with increasing
heating temperature. The acquisition of chemical remanent
magnetization(CRM) during the heating in the laboratoryfield
cannotalwaysbe detectedby the PTRM checks[Coe et al., 1978],
but when the laboratory field and the NRM are not parallel, a
movementof the NRM directiontoward that of the applied field
showsan alterationin the TRM spectrumdue to CRM build up.
This is easily checked on the orthogonal plots of the NRM
demagnetizationin samplecoordinates.At eachtemperaturestep,
the first heating is the most efficient in producing chemical
changes. This has been verified in measuring the bulk
susceptibilityat room temperatureafter both heatingsat eachstep.
Small changesin susceptibilityare observedafter the first heating
of eachtemperaturestep,while the secondheatingdoesnot modify
the previousvalue of the susceptibility.As the first heatingin the
true Thellier methodis made while the laboratoryfield is on, this
methodseemsto be more adequateto detectCRM acquisitionthan
the modified versionby Coe [1967a] in which the first heatingis
made in zero field. Indeed, chemical alteration, in zero field, will
not enableCRM acquisition.
The maximumtemperaturestepwhichdefinesthe linearpart of
the NRM-TRM diagramusedto calculatethe paleointensityis the
last temperature step before any evidence of PTRM capacity
changes.The linear segmentdefinedin the NRM-TRM diagram
whereoverprintandmagnetochemical
changesarenegligiblemust
containat least four to five pointsand spanno less than 15% of
the total NRM in order to be reliable [Coe et al., 1978, Pr•vot et
al., 1985b]. Specialdifficultiesin the interpretationof the NRMTRM appearwhena curvatureof the plot is observedalongall the
2742
CHAUVIN ET AL.: GEOMAGNETIC REVERS,•LS, FRENCH POLYNESIA, 2
temperaturerange,while no PRTM capacityalterationis observed.
In this case, using the first part of the curve would result in
overestimating
thepaleofieldwhile the lastpartof the curvewould
give an underestimate
of the paleointensity.
This kind of curves
can be explainedif the magneticcarriersare multidomaingrains
[Levi, 1977] or if the NRM is not a pure TRM. The last
hypothesismightbe testedby comparingthe blockingtemperature
spectrumandthe Curiepoint [Prdvotet al., 1985b].
In our experiments,a minimum of 11 temperaturestepswere
performed.The statisticalmethoddevelopedby Coe et al. [1978]
was used to obtain the slope b of the straight line, its
corresponding
relativestandard
erroro'b andthestandard
erroro(Fe)
aboutthe paleoiledintensityFe. The validity of the paleointensity
determinations
wereexpressed
usingthe qualityfactorq estimated
from the NRM fractionusedf, the gapfactorg andthe scattering
of the plot aboutthe linear segment.An indicationof the potential
for error causedby CRM acquisitionis given for eachsampleby
the factor R, as defined by Coe et al. [1984]. It representsthe
maximum error, in percent of the laboratory field, that CRM
would cause in the paleointensity estimate. Within a single
volcanicunit, the averagepaleointensitywas calculatedusingthe
weighting factor w, defined by Prdvot et al. [1985b], the
uncertainty
aboutthemeanwasexpressed
by thestandard
deviation
of the unweightedaverage.
Mountain transition.Only one experimenthas been successfulon
one flow from the Jaramillonormal polarity subchron(flow CA).
The resultsare close to the expectedvalue at this latitude, from
27.1 to 54.3 gT, andare in goodagreementwith thoseobtainedin
previousstudies[Senanayakeet al., 1982].
Twenty-one paleointensityresults from 8 distinct lava flows
were obtainedfrom 3 transitionalzones:14 paleointensityvalues
from the Cobb Mountain transition (5 flows), 2 results from the
Upper Jaramillo (1 flow) and 5 results from the MatuyamaBrunhes (2 flows). For the Cobb Mountain transition, the
paleofieldsobtainedare between4 and 8.4 pT. A lower value was
obtainedfor the UpperJaramillotransitionwith a field strengthfor
the flow M2C of 2.6 pT. The two flows from the MatuyamaBrunhestransitionalsogive very low paleointensities,from 3.4 to
3.6 pT. Good consistency
betweensamplesfrom the sameflow is
observed,exceptin the caseof the flow BKK.
DISCUSSION
The Cobb Mountain event:Excursionor short normalpolarity
event?
An apparentR-T-R excursionof the Earth's magneticfield,
dated around 1.1 Ma, is recordedin the volcanic sequencefrom
Tahiti. Severalpaleomagneticreportsfrom marine or terrestrial
sedimentsand volcanic rocks have suggestedthat a complete
Results
reversalof polarityoccurredat that time but this eventhasnot yet
Paleointensitydeterminationswere attemptedon 48 samples, been classifiedin the main polarity time scales[Mankinen and
of which 26 have given reliableresultson 11 distinctlava flows Dalrymple, 1979; Harland et al., 1982]. Therefore,a review of
availabledataappearsto be necessary.
(Table 3).
The study of marine magnetic anomalies has enabled the
Two typesof non-idealbehavioroccurredduringthe Thellier
experiments.
The first typeis represented
by samplesfor whichthe recognitionof the main featuresof the polarity sequencebut short
eventswith time durationof the orderof 10,000 yearsare difficult
evolutionin the PTRM capacityis indicatedby negativePTRM
checks.SamplePF37a is one example (Figure 12) for which a to distinguish.On half of 14 magneticprofilesfrom a surveyover
smalloverprintwascleanedat 250øC.While the first PTRM check the eastPacificRise Crest,Rea and Blakely [1975] haveidentified
a small positivemagneticanomalycenteredaround 1.1 Ma. They
at 380øC showeda decreasein the TRM acquisitioncapacity,all
the following PTRM checksindicatedan increasein the TRM arguedthat the identificationof this shortmagneticanomalyas a
capacity.Providing the PTRM checkshave been carried out at geomagneticeventwas only tentative,becausethis brief periodof
severalsteps,thiskind of behavioris easilyrecognizedandno cal- normalpolarity did not showup well in their averageprofile.
Magnetostratigraphicstudiesof marine sedimentcores and
culation of paleointensitycan be attempted.In contrast,some
samplesprovidedconcaveup paleointensitycurveswith positive Pliocene continentaldepositshave, in some cases,revealed the
PTRM checks.Similar behaviorhas been found during the study existence of a short event before the Jaramillo subchron. But
of samplesfrom the islandof Huahine[RoperchandDuncan,this discontinuoussamplingof cores as well as the time-averaging
issue]. Samples showing important non-linearity in the processin the acquisitionof the magneticremanencehave made
recognitionof shortperiodeventsdifficult. An anomalouschange
paleointensity
havebeenrejected.
More thanhalf of our selectedsampleshaveprovidedreliable in inclination before the Jaramillo normal subchron was found in
results.Most paleointensities
are determinedwith 5 to 15 points,a one marine sedimentcore from the north Pacific [Ninkovich et al.,
NRM fractiontypicallycloseto 50%, anda qualityfactorq around 1966] and transitionaldirectionswere reportedfrom two coresof
10. The highesttemperatures
usedaregenerallycloseto theCurie deep-seacalcareoussedimentsfrom the Melanesianbasin[Kawai et
point(nearor above500øC);thisindicatesa goodthermalstability al., 1977; Sueishi et al., 1979]. In this last case, the excursionhas
of the samples.The laboratoryKoenigsberger
ratio (Q1)[Prdvotet beendatedat 1.06 Ma, extrapolatedfrom the depthof the major
al., 1985b] has been calculatedfor each sample,and the average boundaries of polarity epochs and the sedimentationrate. A
value is 13.8, with a rangefrom 5 to 63. The meanQ1 is similar completereversal was suggestedin one deep-seacore from the
to the one obtainedby Prdvot et al. [1985b] in the most stable equatorialPacific [Forsterand Opdyke,1970] and four coresfrom
samplesfrom SteensMountain.Somepaleointensity
diagramsare the Southern Ocean [Watkins, 1968], from which the normal
shownin Figure 13 and resultsare summarizedin Table 3 and polarityzonewasdatedaround1.07 Ma.
Kocheguraand Zubakov[1978] summarizedthe resultsof their
Figure 14. On Figure 13, the demagnetizationdiagramof the
studiesof the marine and loessdepositsof
NRM duringthe Thellier experimentis also shown.Becausethe magnetostratigraphic
NRM is determinedwhile subtractingtwo oppositepartial TRMs the Ponto-Caspianzone of USSR, and showeda normalpolarity
at each step, the very low noise observedon the orthogonal periodjust before the Jaramillo,occurringaround 1.1 Ma, but no
informationon their paleomagneticdata were presented,making
projectionsenhances
the qualityof theseexperiments.
Four paleointensitydeterminationshave beenobtainedon two the quality of theseresultsdifficult to assess.
More recently,in the vicinity of the magnetozonecorrelatedto
flows with reversedpolarity; the flow R1N which precedesthe
Matuyama-Brunhesreversaland the flow BKB beforethe Cobb the Jaramillo, several intervals of anomalous directions of
CHAUVINET AL.:GEOMAGNETIC
REVERSALS,
FRENCHPOLYNESIA,
2
+1
+l +1
+1
+1 +l
+1
+1 +1 +l
+1
+1
+1 +1
+1
+1
+l +l +1
+1
+l +1 +1 +l
+1
+1 +l +l +l
2743
+1
+1
+l +1
+1 +1
2744
CHAUVIN ET AL.: GEOMAGNETIC REVERSALS,FRENCH POLYNESIA, 2
NRM
y
i A/m
-X
200
•__e
•5
island
Z
NRMo= 1.6
A/m
0520
87PF37A
o
I
of La R6union.
However
it should
be noted
that no
250øC informationconcemingthe polarityof the flows aboveandbelow
HLab =15 •tT
I
normal polarity have given an age about 1.08 Ma, close to the
previousone obtainedon the Cobb Mountain [Mankinen and
GrommJ,1982]. Howeverthe ageuncertainties
arequitelargefor
these two flows (about 12%). Normal polarities from volcanic
unitswith agesaround1.05 Ma havealsobeenreportedby Fleck
et al. [ 1972] in southArgentina,Abdel-Monemet al. [ 1972] on
the islandof TenerifeandMcDougalland Chamalaun[ 1966] in the
I
3
TRM
was available.Thus, the interpretationmustrely strictlyon the KAr results.A normalpolarity period,precedingthe Jaramillo,was
foundin an ashlayer and adjacentsedimentsfrom the depositsof
the Osakagroup(Japan)[Maenaka, 1979]. The ash horizonhas
given a fissiontrack age of 1.1+0.01 Ma. The use of only one
level of AF cleaningmight have not been sufficientto remove a
possiblestrongnormaloverprint.
This review shows the difficulty in identifying a short
geomagneticeventeitherin sedimentor volcanicunits.However,
thereis now strongevidencefor a worldwidegeomagnetic
feature
Fig. 12. NRM-TRM diagrams,normalizedto the initial NRM (NRM0),
showing non-ideal behavior during paleointensity determination.
Orthogonalplots in samplecoordinatesrepresentthe evolutionof the
remanentvectorduringthe experiment.HLab: intensityof the laboratory
of either a short event or an excursion around 1.1 Ma. Our data
field in gT. Sample87PF37AshowsnegativePTRM checksindicating
from the Punaruuvalley suggestthatthe field did not reacha full
thatmagneticalterationoccurredduringthe experiment.
magnetizationhave been observedin deep-seacoresfrom the
CaribbeanDSDP Site 502 [Kent and Spariosu, 1982]. The most
detailed record of an event of normal polarity preceding the
Jaramillo
normalpolarity.Nevertheless,
thedetailedrecordfrom core609b
[Clement and Kent, 1987] suggestsa very short time for the
normal period. Therefore, such a short normal polarity might
correspond
to a hiatusin thevolcanicactivityat thattime andonly
the transitional boundaries were recorded.
subchron was obtained from sediment cores from the
north Atlantic [Clementand Kent, 1986]. In this study,the event Comparisonof the transitionalfield characteristicswith other
hasbeendatedaround1.16 Ma, assuminga constantsedimentation available records
rate betweenthe depthsof the the major polaritychrons,and its
time durationwas estimatedto be 11,600 years.Nevertheless,the
normalperiodin this recordis disruptedby very shallowinclinations and the authors argue that the proximity of this large
directionalswingto theuppertransitional
zone,andtheoccurrence
within an intensity minima, suggestthat they are part of the
transitionalbehavior. Thus, the true normal period might be as
A comparisonof the intermediatestatesof the field observedat
different sites for a single reversal may provide a clue to the
understanding
of the reversalprocess.To do that,a review of the
otherrecordsis necessary.
shortas 5,000 years.
This
Some normal or intermediate directions have also been found,
The Cobb Mountain. As discussedbefore, the most detailed
recordconcerningthe Cobb Mountainevent was reportedfrom
hydraulicpistoncoredsedimentsby Clementand Kent [1987].
record
has one main
characteristic
in that
while
the
declinationexhibitsvery shorttransitionfor boththe onsetandthe
termination of the normal polarity period, large variations in
inclinationoccurbeforeandafterthedeclinationchanges.As noted
earlieran importantshallowingof theinclinationis alsoreported
in thenormalpolarityperiodasdefinedwith thedeclinationpattem
by Clementand Kent [1987]. The VGPs associatedwith these
intermediatedirectionsare almostconfinedin the meridianplane
containingthe siteandshowa well definedantisymmetry
between
in severalplaces,on volcanicflows datedaround1.1 Ma., but the
imprecisionin radiometricdatingwasoftengreatenoughto allow
the possibilitythat they cooledwithin the Jaramillosubchron.
The name Cobb Mountain was given by Mankinen et al.
[1978] to an event which, for the first time, was found in a
volcanic unit, a compositerhyolite and dacite dome, situatedin
northern California. K-Ar datings have given a mean age of
the onset and the termination.
1.12_+0.02Ma. Transitionaldirectionswere clearly recordedin 5
In the recordfrom FrenchPolynesia,we observedthreedifferent
sitesof rhyolites,while the normalpolarity was definedonly by a
directions.
Obviously,theyrepresent
onlya
strongnormaloverprintat oneparticularlocality, wherethe rock groupsof intermediate
of an excursion,or limited recordsof one or both
is completelydevitrified.Becauseno evidenceof latehydrothermal few snapshots
alteration was found on the Cobb Mountain, Mankinen et al.
transitionslimiting an undetectednormalpolarityevent.Reliable
have been obtainedfor theseflows with fields
arguedthat this devitrificationwas causedby deutericalteration paleointensities
from 4 to 8 gT. There is no evidencefor a strongintermediate
whichcouldhavestartedshortlyaftereruption.Thereforethe NRM
state.Acceptingthattheseflowsrecordan excursion,themagnetic
directionwas consideredas representinga completechangefrom
data show that the intermediatefield during the Cobb Mountain
intermediateto normalpolarityduringthe coolingof the rhyolite.
One intermediatepolarity lava flow dated by K-Ar at 1.13 had similar characteristicsto those observed during polarity
of excursionsas aborted
+0.08 Ma and six flows with a normal polarity with age from reversals,supportingthe interpretation
ß
reversals[Hoffman,198lb].
1.07 to 1.37 Ma havebeenreportedby Briden et al. [1979] in a
Lower and UpperJaramillo. Up to now, deep-seasedimentary
study of volcanic rocks from the islandsof St. Vincent and La
Guadeloupe,in the LesserAntilles.On the islandof Madeira,one corerecordsof the transitionsdelimitingthe Jaramillohave been
obtainedin the Indian Ocean [Opdykeet al., 1973, Clement and
normallymagnetizedbasaltdatedat 1.15 Ma hasbeenfoundby
Kent, 1984, 1985], in the north Atlantic Ocean [Clement and
WatkinsandAbdel-Monem[ 1971]. Usingthe new constants
in KAr datings[MankinenandDalrymple,1979], thisagebecame1.18 Kent, 1986] and Pacific Ocean [Hammondet al., 1979; HerreroMa. Two basalt flows in the Coso Range, California with a Bervera and Theyet, 1986, Theyet et al., 1985]. From volcanic
CHAUVINET AL.: GEOMAGNETIC
REVERSALS,
FRENCHPOLYNESIA,2
2745
NRM
NRM
E
EIiup NRM(•
/
350øc
•" 41o
t1A/m52
i•1•
•500
10A/m •
495øc
540øC
535øc
Down
87R1N85A
87BKI97A
(IS)
(IS)
•570øC
1
TRM
1
NRM
Down
TRM
87R1N85A
87BKI97A
NRM
300øc
E
•
•
S
NRM
N
200øC
:
:
:
:
:
:
E
:
NR
'•o••540
0.3A/
500øC
Down
500"•
530øc
,0.3A/m
NRMO,4•t
t
87BKP147A
(IS)
•'300
D•wn
550øC
O
540øC
87M2C52A
1
2
3
4
(IS)
TRM
TRM
87BKP147A
87M2C52A
NRM
E
•
S
480øC
535øc
Down
87PKC16A
1
2
3
(IS)
TRM
87PKC16A
Fig. 13.NRM-TRMdiagrams
of accurate
paleointensity
determinations
usinga largeportionof theNRM. Theorthogonal
plots
showthe evolutionof the remanentvectorin geographical(in situ) coordinates.
records,a few intermediatedirectionscome from the study of
Mankinen et al. [ 1981] on Clear Lake (Califomia).
The importantshallowingof the inclinationseenat the beginning
of the lower Jaramillo record from the Indian Ocean [Clement and
Kent, 1984] clearlydisagreeswith the observationfrom Tahiti, in
termsof a simplezonalharmonicmodel.
Herrero-Bervera and Theyet [1986] have presenteda nonsedimentation
rates,
aslowas6 mm/103
years
[Hammond
etal.,
axisymmetric
behavior of the field during the upper Jaramillo
1979]
or7.8mm/103
years
[Theyet
etal.,1985].
If theremanence
recorded
by
deep-sea sediments from a site near Hawaii
is a detrital remanentmagnetization(DRM), each sample,no
rate35 mm/103
years).
Thisrecord
ismostly
matter how closely spaced, representsan average of the (sedimentation
by oscillations
betweenreverseandnormalpolarities.
paleomagnetic
signalovera greatperiodof time;in thecaseof a characterized
PDRM (post-DRM),the transitioncanbe completelymissedor This behavior may question the possibility that antipodal
the directions observed can differ completely from the true magnetizationsare presentin the samples.Assumingthat this
Data from deep-seacoresfrom the PacificOceanare, in our
opinion,difficult to interpret,simplybecauseof the very low
configuration
of theintermediate
geomagnetic
field [Valet,1985;
Hoffmanand Slade,1986].Recordsof the lowerJaramillofrom
record is reliable
leads to the conclusion
that transitional
fields
recordedat a siteat a low northernlatitudein the samelongitudinal
terrestrialsedimentswere describedby Gurarii [ 1981], from the bandas the Societyislandshave no commonfeatureswith those
territoryof westernTurkmenia(USSR)but,sincetheremanence
is recorded at Tahiti.
The record of the Upper Jaramillo transition in deep-sea
carriedby two magneticminerals,hematiteandmagnetite,thenatureof themagneticremanence
(detritalandchemical)is unclear. sediments from the Indian Ocean shows evidence of bioturbation,
2746
CHAUVIN ET AL.: GEOMAGNETICREVERSALS,FRENCHPOLYNESIA,2
Matuyama-Brunhes. This transition is poorly defined but at
least for the limited part of the transitionrecordedin the Tahiti
section,there is convincingevidencethat the field was low (3-4
gT).
There are many sedimentaryrecordsof this reversal.A strong
dependence
of the VGP path duringthe reversalwith the longitude
of the observationsites was first pointed out by Fuller et al.
50-
[ 1979]. However, careful examination of the data convincedus that
30-
somesediments
havenot accuratelyrecordedtheweakgeomagnetic
field andthatthe geometryof the transitionalfield mightnot be as
simpleas recentlyproposedby Hoffman [ 1988]. In the caseof the
Japaneserecord [Niitsuma et al., 1971], the NRM intensitydata
are scatteredandno clearvariationof the intensitycanbe observed.
With sedimentsfrom the mid-northern and equatorial Pacific
[Clementet al., 1982], dissimilarVGP pathshave beenobtained
from two nearbycores,and an increaseddispersionin directions
--
10-
was observed
0 ' ' 3•0 ' ' dO ' ' •0 ' ' 1•)0' ' 1•0
REVERSAL
180
ANGLE
within
the transitional
zone. These observations
suggestedto Clement et al. that some sedimentologicalfactors
have probably distorted the magnetization of these sediments
characterized
byalowsedimentation
rate(7to11mm/103
years).
The accuracyof some paleomagnetictechniquesto clearly
isolatethe primaryremanentdirection(if it still existsandhasnot
geochemical
processes
[Karlin et al.,
axialnormaldipole.Paleointensities
obtainedfromtheintermediate
flows beenreplacedby subsequent
are below 10 gT.
1987]) constitutesone of the major problems.The lake Tecopa
record[Hillhouseand Cox, 1976] hasbeenlong consideredone of
which might well have actedto smooththe transition[Opdykeet the most reliable recordsof the last polarity reversal.However, a
al., 1973; Clement and Kent, 1985]. The sedimentsfrom the north
resamplingof the section and new data obtainedby thermal
Atlantic ocean [Clement and Kent, 1986], which are characterized demagnetization[Valet et al., 1988a] showsclearly that strong
byasedimentation
rateof82mm/103
years,
have
given
arecord magneticoverprintsarenotremovedby AF demagnetizations.
The
of the Upper Jaramilloin which full normalpolaritydirectionsare previouslyreportedintermediatedirectionswere the resultof a
not observedat the baseof the samplingzone, and thereforethe superimposition of reverse and normal components of
lower boundary of the transition is not defined. Moreover, this magnetization,and provide no useful information about the
record displays strong variations in intensity of magnetization reversal. Although Valet et al. [1988a] showed that AF
which are apparentlycorrelatedto changesin the lithology.
demagnetization
did not isolatea primarycomponent,
theydid not
So we are facedwith the disappointingconclusionthat no fully demonstratein a convincing manner that the intermediate
reliabledataare availablefor comparisonwith our detailedrecord directionsisolatedby thermaldemagnetization
aremeaningful.
from the Upper Jaramillo.Only one accuratepaleointensitywas
Recently,data from the equatorialAtlantic Ocean(Site 664)
obtainedfor the Upper Jaramillo transition(Table 3) showinga havebeenreportedfrom deep-seasedimentary
cores[Valetet al.,
field strengthclose to 2.6 gT, but all the NRM intensity data 1989]. In comparingthe VGP path they obtainedwith deep-sea
indicatethat the intensityof the field might havebeenparticularly core results from a site of the same longitude (Site 609), but
low during this reversal. As was already pointed out, an situatedat a mid-northernlatitude [Clementand Kent, 1986], Valet
intermediate direction associatedwith a low NRM intensity was et al. [ 1989] showedthat the two recordsprovideantipodalVGP
observed in flow R1D which follows the Jaramillo transition
paths,locatedoverAmerica(Site 609) andAsia (Site 664), a fact
termination.This directionhasan angulardepartureof almost40ø that cannot be reconciledwith a simple axial geometryof the
from the full reversedpolarity. This is not the first time that a transitionalfield. Obviously, even for this 'well' documented
brief polarity fluctuation is describedclose to that polarity reversal, a close look at the data shows that still more accurate
boundary.Clementand Kent [1986] observeda 60ø swingin the paleomagnetic
recordsare neededfor a betterdescriptionof the
inclinationjust after the upper Jaramillo transition.In the lava transitional field.
flows of Clear Lake (California) [Mankinen et al., 1981] a
The secondpart of this discussionwill includesomeof the
sequenceof intermediatepolarity lavas,apparentlyyoungerthan resultsobtainedon the islandof Huahine [Roperchand Duncan,
Fig. 14. Diagramshowingthe evolutionof the meanpaleointensities
per
flow with their standarddeviations,versusthe angulardeparturefrom the
reversed lavas above the Jaramillo termination, has been found.
The K-Ar age obtainedon one of the intermediateflows (0.90 _+
0.02 Ma) is indistinguishable
from the age of the end of the upper
Jaramillo transition,suggestingthat the brief intermediatestate
observedis part of the reversalboundary,and representswhat is
generallycalleda reboundeffect [Prdvotet al., 1985b;Laj et al,
1987, 1988]. Our data from flow RID can be interpretedin this
way. On the otherhand, somedata from Clear Lake and deep-sea
sedimentsfrom the SouthernOcean [Watkins, 1968], suggestthat
a polarity event might also have occuredat about0.83-0.85 Ma,
which could also explain the directionof flow RID. Without an
estimateof the age of this flow, we cannotchoosebetweenthese
two possibilities.
this issue).
Zonal and non-zonalcomponents
East and west intermediate directions are more numerous in the
record of the N-T-N
excursion of Huahine than in the records from
Tahiti which show smaller deviations from the north/south vertical
plane.On average,non-zonalcomponents
are asimportantasthe
componentslying within the geographicalmeridian during
transitions,as suggested
by Prdvot at al. [1985b] for the Steens
Mountainrecord.However,datafrom Polynesiamay indicatetwo
differentconfigurations
for the beginningand the middleof the
reversal.For the lower Jaramilloat Tahiti and the beginningof the
transitionfrom Huahine, there is some evidencefor axisymmetry
CHAUVIN ET AL.: GEOMAGNETICREVERSALS,FRENCHPOLYNESIA,2
2747
characterizedby a path throughhigh inclinations(Figure 15).
RoperchandDuncan[thisissue]suggest
thatthissimpleapparent
structurecanbe modeledwith a growingzonalquadrupoleof the
samesignas the decayingdipole.The evidence,from Polynesia,
for a quadrupole
field at theonsetof a reversalis in agreement
with
arithmeticmeans.Differencesbetweenarithmeticand geometric
averagesunderline the scatter in each data set. As the NRM
intensitydependson the strengthof the geomagneticfield and the
magneticmineralogyof rocks,it seemsreasonableto simply add
the predictionmade by Merrill and McFadden [1988] who have
arguedthat a reversalmay occurif the axial dipolefield is weak
andthequadrupole
family field is strong.
from the samegeographicalareaandthe volcanicrockswhich form
thesetwo islandshave similarcompositions.
The averagevalues,
within each reversal angle group, of the NRM data for Tahiti,
Huahineandfor the combineddatasetsarereportedin Table 4. In
combiningthe two datasetsthe numberof sitesusedis quitelarge
Variation in the geomagnetic
field intensity
Intensity of the NRM versus directions or VGPs. As we
discussedearlier, the NRM intensities (at 10 mT) can reflect the
averagevariationsin the geomagneticfield intensity.All the NRM
intensitydatafrom HuahineandTahiti havebeengroupedby 10ø
intervalsin reversalanglesandVGP latitudes;then,arithmeticand
geometricmean valueswere calculatedfor each group(Figures
16a, b). The standarderror bars have been reported for the
the two data sets from Huahine and Tahiti. Indeed, these data are
(220), and we assumethat short-termfield paleointensity
variationsandNRM intensityfluctuations
dueto rock magnetic
propertieshavebeenaveraged.An importantdropin intensityis
observedfor reversalanglesfrom 0øto 30ø or VGP latitudesfrom
90ø to 60ø. For higherreversalangles(i.e., lowerVGP latitudes)
the intensities stay very low, around 1 A/m or less. The same
result is obtainedusing either the data from Huahine or Tahiti,
N
N
w
E
W
E
S
S
Fig. 15. Equalareaprojections
of paleomagnetic
directionsobservedat the beginningof the N-N Huahineexcursion(a) andthe
lowerJaramillotransition(b) recordedat Tahiti.In thesetwo cases,the directions
go throughhighinclinations
leadingus to
suggest
thatthebeginningof a transitionmightbe controlledby axisymmetric
components
of the field.
5-
5-
[]
[]
','N
T
xx
Tahiti+ Huahine
Tahiti
+Huahine
,.,\
o
,,
1-
45
o
o
46
I
16
18
I 10 I
16
11
I
11
I
60
24
I
33
I
Angular Distance from Central Axial Field
I
90
56
o
o
44
'
11
'
Virtual
19
iO '
24
13
'
Geomagnetic
11
I•0 '
11
9
'
'
9O
Pole Colatitude
Fig. 16. Diagrams
showing
thevariations
of themeanNRM intensity
(after10 mT) averaged
for groupof flowsfromPolynesia,
versustheangulardeparturefromcentralaxialfield (A) or versusvirtualgeomagnetic
polelatitude(B). The reverseddirectionshave
beeninvertedandaddedto thenormalones.Thearithmetic
means(blackdots)areshownwiththeirstandard
errorbarswhileopen
circlesrefer to geometricmeans.The numberof sitesavailablein eachintervalis alsoindicated.
2748
CHAUVINET AL.: GEOMAGNETIC
REVERSALS,
FRENCHPOLYNESIA,2
TABLE 4. AverageNRM Intensities
Huahine
Tahiti
•)(ø)
10
N
Jn
or
N
Jn
4.8
2.8
1.2
0.7
0.9
0.5
1.1
0.7
0.8
20
29
34
4.9
3.7
0.97
0.43
16
12
30
40
50
11
3
11
2.6
0.8
0.7
0.94
0.77
0.13
5
15
5
60
70
80
90
7
6
10
8
0.6
0.5
0.4
0.6
0.13
0.09
0.16
0.13
4
5
14
25
Tahiti+Huahine
or
0.49
0.57
0.65
0.48
0.35
0.42
0.23
0.24
0.21
N
Jn
45
46
16
18
16
11
11
24
33
4.8
3.5
2.0
0.7
0.7
0.6
0.7
0.6
0.8
or
0.64
0.35
0.71
0.41
0.14
0.16
0.15
0.16
0.16
higher in Iceland than at a low latitude site like Polynesia;the
ratio of the transitionalfield to stableperiod intensitiesmight be
around 25-30% in Iceland and 10-20% in Polynesia. As the
intensityof the normalandreversedfield is higherin Icelandthan
in Polynesia, this indicatessignificant differences between the
averagepaleointensityduring transitionsbetweenthesetwo sites.
These data may suggesta latitudinaldependenceof intermediate
fields. It is interestingto note that one well establishedfeatureof
paleosecularvariation is the increaseof the scatterof VGPs with
increase in latitude.
While in Polynesiathe decreaseof the relative meanintensity
with decreasing VGP latitude is similar to that defined with
directions,the dropin intensitywith VGP colatitudeis particularly
N,number
ofsites;
Jn,geometric
mean
ofNRMintensities
inAm-1 smoothin Iceland. The dipole wobble is not large as seenfrom
(at 10mT),by 10øintervals
in reversal
angles(3); err, standard
error.
Polynesia and the smoothvariation observedin Iceland might
resultfrom the VGP calculationprocessbecauseof a strongernondipolefield.
giving us some confidence in the observed distribution. This
Becausethe averagefield duringperiodsjust beforeandafterthe
behaviorin the NRM intensitiessuggeststhat for reversalangles reversal may not be representativeof the average field during
higherthan30ø the geomagneticfield is purelytransitionalat this longer stable periods, such as a chron or a subchron,a bias is
latitude,and that the mean intensityof the field staysvery weak perhaps introduced in our data. Indeed, a great percentageof
duringthe entiretransitionalperiod.However,we mustemphasize measuredflows sampledtransitionalzones,especiallyin the case
that this representation
enablesonly the recognitionof average of Huahine, while the sampling in Tahiti is more regular and
field variationsand thesevaluesdo not, for example,showwhat covers an interval of time around 0.6 Ma. Our data set is about ten
wasthe maximumintensityreachedduringthe transitions.
timeslessthanthe one from Iceland,however,and this can explain
A similar studyhasbeen reportedfor a data set of more than the quite large error bars observed in some cases. Certainly,
two thousandlava flows from Iceland by Kristjansson and additional data from different sites are needed to better constrain our
McDougall [ 1982] andKristjansson[ 1985]. Comparisonof results presentobservations
and our interpretations.
from these two sites at different latitudes deserves consideration.
Absolute paleointensities during reversals. Our absolute
For thisreason,both geometricmeanvalueshavebeennormalized paleointensitydatarangefrom 2 to 8gT with an averagearound5by the intensity observedwithin the 0-10 ø interval in reversal 6 gT. More determinationsare still needed,but it seemsthat our
angles or VGP colatitudes (Figure 17). With respect to the
paleointensitiesare significantlylower than the averageintensity
deviationsfrom the axial dipoledirection,the decreasein intensity during the SteensMountainstransition,which is closeto 11 gT
observedin Polynesiais also observedin Iceland with a cut-off [Prdvot et al., 1985a]. In view of the paleointensity
level between transitionaland stable period around 30ø to 40 ø determinations, and the variations in the NRM intensities with
departurefrom the centralaxial dipole. On average,however,the angulardeparturesfrom the centralaxial dipole, it seemsthat the
intensityof the transitionalgeomagneticfield appearsto havebeen average intensity of the field during transitions might be
1-
0.5
0.5-
Polynesia
o
io
Angular Distance from Central Axial Field
o
io
Virtual Geomagnetic Pole colatitude
Fig. 17. Comparisonof the variationsof the averagegeometricmeanvaluesof intensityof magnetization
(after 10 mT) from
PolynesiaandIcelandversusthereversalangleandtheVGP latitudes.Data from IcelandareextractedfromKristjansson[1985]. In
orderto enablecomparison,
eachdatasethasbeennormalizedto theirrespectivefirst valuein the 0-10ø interval.The arithmetic
standard errors have also been normalized.
CHAUVIN ET AL.: GEOMAGNETICREVERSALS,FRENCHPOLYNESIA,2
2749
latitudinally dependent,with possibly higher values at high Implicationsfor andfrom theoreticalstudies
latitudesthan at low latitudes,in contrastto what was suggested
Recently, a new interest in the reversal process has been
by Pr•vot et al. [ 1985a].
triggeredby the developmentof modelswhich seemsufficiently
On the other hand, few data from studies of reversals in Hawaii
specificto be testedagainstthe paleomagneticdata. Combining
[BogueandCoe, 1984] andfrom threeexcursions
recordedby lava detailed studiesof the recent field and inversionmethods,Bloxham
of Oahu [Coe et al., 1984] indicatestrongerpaleointensities
than
and Gubbins[ 1985] haveproducedmapsof the radialcomponent
those observed in Polynesia. Obviously, the small amount of
of the field at the core-mantleboundary(CMB) from the years
accurate data that is available limits the discussion.Shaw [1975,
1715 to 1980. Staticfeatures,associatedwith the main standing
1977] reported high paleointensitiesduring transitions while
field generatedby the dynamo,and rapidly drifting featuresare
Lawley [1970] indicatedweak field (about5 gT) duringNeogene
observed[Gubbins and Bloxham, 1987]. The developmentof
polarity transitions recorded in Iceland, but obviously results
reverseflux patchesin the middle southernAtlantichemisphereis
obtainedby total TRM methodsare questionable
becausetheymay
possibly explained by expulsion of toroidal flux by fluid
give erroneousresultsif thermalalterationsof the samplesare not upwelling [Bloxham, 1986]. Gubbins [1987] suggestedthat the
detected.Very low intensitiesof the field, as low as 3 gT have location of these reverse flux zones would be tied to mantle
some importantimplications.With such low magneticvectors, temperature anomalies, and that their growth provides a
large variationsin direction may occur more easily than with a
mechanismfor polarity reversals.However, the large decreaseof
strongfield becausethey do not correspondto large total vector the paleointensityduring reversalsis one of the best established
changes.
characteristics
which suggestsa generalbreakdownof the dipole.
Thus,the growingof onereverseflux zonemay not systematically
providelow intensitieseverywhere.
It is well known that the irregularextrusionrate of lava flows
This model predicts also that all reversals should be similar
and the smoothingprocesses
in sedimentsand intrusionmay alter
(i.e., have the same transitional field) within the life-time of the
significantlythe recordof geomagneticreversalpaths[Weekset
mantle convectionpattern. Some sedimentaryrecordsseemsto
al., 1988]. However, we mustemphasizethatvolcanicrocksrecord
providenegativeevidencefor thismodel [Laj et al., 1988], evenif
accuratelythe magneticfield evenif providingonly incompleteand
thereare indicationsin recordsfrom Cretethatthe reversalprocess
intermittentrecordsof a reversal.On the contrary,Hoffman and
may remaininvariantoverseveralpolarityintervals[ValetandLaj,
Slade [1986] demonstratedthat smoothingeffectsin sedimentary
1984]. Different transitionsfrom Polynesiaexhibit somecommon
records may provide reversal features with no relation to the
features. In particular, southwest directions with shallow
geomagnetic field. Other problems attend interpretation of
inclination are observed in 4 out of 5 transition records, but it is
magneticfield variationsfrom paleomagneticvectors,particularly
not obviousthat suchcorrelationsare significant.
whenrecordsof low intensityfields may be damagedby any kind
Comparing his theory with available paleomagneticdata,
of local perturbations.For example, the local crustal magnetic
Gubbins[1988] arguedthat reverseflux featureshavebeenabsent
anomaly may be significant. The intensity of the magnetic
within the northernhemispheresince5 Ma. This idea of a constant
anomaly developedby an island like Tahiti remains very low
region where reversals are initiated is quite equivalent to the
comparedto the strengthof the geomagnetic
field duringnormalor
commonstartingpointfor a reversalmodelof the type suggested
reversedperiods. Giving an average remanent intensity to the
by Hoffman [1979]. Comparing our resultswith data from the
volcanic rocks between 5 to 10 A/m, a homogeneousvolcanic
Indian Ocean area would be helpful in constrainingGubbins's
cone of the size of Tahiti can develop, at its surface, a local
model. Even if there is no clear paleomagneticdata supporting
magneticfield between 0.5 to 1 gT, which are low values. The
Gubbins'sreversal scheme,there is now strongevidencethat
magnetizationinducedby the Earth'sfield is negligiblein view of
thermalandtopographic
variationsat the core-mantleboundaryare
our Koenigsbergerratios;this is particularlytrue when the field
constraining
field variations,particularlythe long-termchangesin
intensityis alreadylow. This meansthat during a transition,with
the mean reversalrate [McFadden and Merrill, 1984; Courtillot and
a geomagneticfield close to 3 gT, at least two-thirds of the
Besse, 1987].
recorded magnetic field is of internal geomagnetic origin.
Geomagnetic polarity reversals in a turbulent core were
Obviously,becausemagneticanomalieshave strongdependence
investigated
theoretically
byOlson
[1983]
using
theet2 dynamo
with the shapeof the body, short wavelengthvariations could
model.
Olson
considered
that
changes
in
the
sign
of helicity in an
exist. In the case of our data, we do not think that local effects
Alteration of the geomagneticsignal by the recordingmedium
play a major role becausesimilardirectionswere recordedseveral
hundredsmetersapart.
We do not know what the implications are of such low
paleointensitiesfor the complicated processesof remanence
acquisitionin sedimentaryrocks.One can speculatethat it would
be difficult for a sedimentto adequatelyrecorda transitionwhen
the magneticstrengthneededto align grainsbecomestoo weak.
Thus,somesedimentsmay not be ableto recordthe middlepart of
a transition but only the beginning and the end of a transition
when the field decaysor growstowardthe otherpolarity. The fact
that several records [Clement and Kent, 1984; Valet et al., 1988b]
have a shallow inclination in the middle of the reversal may
indicatethatthe sedimentaryprocessmay controlthe remanencein
some cases.A typical example might be the Lower Jararnillo
recordfrom a southernhemispherecore [Clementand Kent, 1984]
which is markedby a regularshallowingof the inclinationwhile
the noiselevel in declinationappearsto slightlyincrease.
cz
2 dynamo
canprovide
twotypesof reversals,
'component
reversals'when only the poloidalor the toroidalfield reverse,and
'full reversal'when the two field componentsreverse.Following
Olson, reversalswould be the result of the competitionbetween
two source powers of the dynamo: heat lost at the CMB and
progressivecrystallization of the solid inner core. Over a brief
interval,the contributionfrom inner core growthmay exceedthe
contribution from heat lost at the CMB, causing a transient
reversalof helicity.
The mathematicalanalysis from Olson [1983] predicts the
variation of the dipole intensity for a field reversal with a
reasonabletime duration, but it does not define the non-dipole
configurationof the field during the transition.Clement [1987]
usedthe dipoletime evolutionequationsfrom Olson'stheoryand
assumedthat the energy lost by the dipole field is partitionedto
low degree non-dipole fields [Williams and Fuller, 1981] for
modelingrecordsof two reversalsfrom both northernand southern
2750
CHAUVIN ET AL.: GEOMAGNETIC REVERSALS,FRENCH POLYNESIA, 2
hemispheres.
The fit to the datais stronglydependenton the way
the dipole energyis distributedto the other terms.The attempt
madeby Clement [1987] was not successfulfor matchingthe
intensitypatternwhichshouldbe themainparameterfor a rigorous
testof Olson'shypothesison the variationof the dipoleintensity.
The two previoustheoriesassumethat polarity reversalsresult
from disturbancesperturbingthe movementin the core which
producethe stablemain field. This hypothesisis in agreement
with the conclusionof the analysisof the geomagneticreversal
sequence
madeby McFaddenandMerrill [ 1986]whichindicates
that the triggeringeffectof reversalmightbe independent
of the
process
producingthemainfield.
Our data show,like othervolcanicrecords,that the intensityof
thegeomagnetic
field at theEarth'ssurfaceis considerably
reduced
(one fifth to one tenth) duringthe transition.It seemsreasonable
to think that the field intensityis alsoreducedin the core and that
the magneticenergy lost is transferredinto kinematic energy
[Prdvotet al., 1985a],increasingthe level of turbulencewithin the
liquid core.This effecthasbeensuggested
in orderto explainthe
large and quick directionalchangesobservedin somerecordof
polarity reversal. In particular, the 'geomagneticimpulses'
occurringduringthe SteensMountaintransitionimply a rate of
changeof thefield components
considerably
largerthanthemaximum rate calculated for the recent secularvariation [Prdvot et al.,
absolutepaleointensities
and a morecriticaljudgmenton the data
quality are neededto clearly establishsomeother featuresof
reversalsof geomagneticfield (i.e. axisymmetryat the onset,
latitudinaldependencyof the intensityof the intermediatefield,
ultrarapidfield variations),andto makefurtherprogressin the
theoreticalunderstanding
of the reversalprocess.
Acknowledgments.
Thanksaredueto LeonardChungueandtheCentre
ORSTOM of Tahiti for providing the logistical supportfor sample
collecting.We are indebtedto Dr. Talandierfor his hospitalityat the
Laboratoire
de G•ophysiquede Pamatai.Discussions
with K.A. Hoffman,
C. Laj and M. Pr•vot have beenvery helpfulat differentstagesof this
study.
REFERENCES
Abdel-Monem,A., N.D. Watkins, and P.W. Gast, Potassium-Argonages,
volcanicstratigraphyand geomagneticpolarityhistoryof the Canary
islands: Tenerife, la Palma and Hierro, Am. J. Sci., 272, 805-825,
1972.
Bloxham, J., The expulsion of magnetic flux from the Earth's core,
Geophys.J. R. Astron.Soc.,87, 669-678, 1986.
Bloxham, J., The dynamical regime of fluid flow at the core surface.
Geophys.Res. Lett., 15, 585-588. 1988.
Bloxham, J., and D. Gubbins,The secularvariationof earth'smagnetic
field, Nature, 317, 777-781, 1985.
Bogue, S.W., and R.S. Coe, Transitional paleointensitiesfrom Kauai,
Hawaii, and geomagneticreversals models. J. Geophys.Res., 89,
1985a]. Resampling of several flows and detailed thermal
10341-10354, 1984.
demagnetization
associated
with a coolingmodelof thelavaflows,
Briden, J.C., D.C. Rex,
mightindicateevenquickerdirectional
changes
of thegeomagnetic
field [Coe and Prdvot, 1989]. We havenot foundsuchbehaviorin
our volcanicsequences.
CONCLUSIONS
A.M.
Faller,
and J.F. Tromblin,
K-Ar
geochronologyand palaeomagnetism
of volcanicrocks in the lesser
Antilles island arc, Philos. Trans. R. Soc. London, Set. A, 291,485528, 1979.
Clement, B.M., Paleomagneticevidence of reversals resulting from
helicity fluctuationsin a turbulentcore,J. Geophys.Res.,92, 10629-
10638, 1987.
Four transitional zones have been sampled in a volcanic
Clement,B.M., D.V. Kent, and N. Opdyke, Brunhes-Matuyamapolarity
sequence
of 120 lava flows in onevalley of the islandof Tahiti.
transitionin three deep-seasedimentscores,Philos. Trans. R. Soc.
Accordingto theradiometricdating,theycorrespond
to the Cobb
London, Set. A, 306, 113-119, 1982.
Mountain, the lower and upper Jaramillo and the Matuyama- Clement, B.M., and D.V. Kent, A detailed record of the Lower Jaramillo
Bmnhes transitions.
polaritytransitionfrom the southernhemisphere,deep-seasediments
cores,J. Geophys.Res.,89, 1049-1058, 1984.
Paleomagneticdata indicatethat the Cobb Mountain record
Clement, B.M., and D.V. Kent, A comparisonof two sequentialgemay correspond
to a Reverseto Reverseexcursionof the Earth's
omagneticpolarity transitions(Upper Olduvai and Lower Jaramillo)
magneticfield, occurringaround1.1 Ma.
from the southernhemisphere,Phys.Earth Planet.Inter., 39, 301-313,
1985.
Combiningdatafrom the islandof Tahiti with the recordof a
Normal to Normal excursion in lavas from the island of Huahine
[Roperchand Duncan,this issue],the non-zonalcomponents
of
the field appearimportantduringthe transition.Nevertheless,
the
beginningof two transitions(the lower Jaramilloand the N-N
excursionfrom Huahine), seemsto indicate that the onset of a
reversalcouldbe moreaxisymmetric
thanthefollowingphase.An
increasein intensityof the quadrupolefield at the onsetof the
reversalmay fit the observedhighinclinationpath.
NRM intensitydatashowthat the intensityof the transitional
field, at thislatitudewasvery low. This fact is confirmedby some
reliablepaleointensity
determinations,
whichindicatea paleofield
strengtharound3 to 8 gT. ComparisonbetweenaverageNRM
intensityfrom Iceland and Polynesiaindicatea lower averaged
transitionalfield at low latitudethanat high latitude.The response
of the differentpaleomagnetic
recordersto suchlow and rapid
changingfieldsneedsto be investigated
in orderto selectthemost
reliablepaleomagnetic
recordsof thetransitional
field.
The main featuresof a polaritytransition,i.e., shortduration,
non-dipoleconfiguration
andlargedropin intensity,arenowwell
Clement, B.M. and D.V. Kent, Geomagneticpolarity transitionrecords
from five hydraulicpiston core sitesin the north Atlantic, Initial. Rep.
Deep Sea Drill. Proj., 94, 831-852, 1986.
Clement, B.M. and D.V. Kent, Short polarity intervals within the
Matutyama:Transitional field recordsfrom hydraulic piston cored
sediments from the north Atlantic, Earth Planet. Sci. Lett., 81,253264, 1987.
Coe, R.S., Paleointensities
of the Earth'smagneticfield determinedfrom
Tertiary and Quaternary rocks, J. Geophys.Res., 72, 3247-3262,
1967a.
Coe, R.S., The determinationof paleointensitiesof the Earth'smagnetic
field with emphasison mechanismswhich could cause non-ideal
behaviorin Thellier's method.J. Geomagn.Geoelectr., 19, 157-179,
1967b.
Coe, R.S., and C.S. Gromm6, A comparisonof three methodsof determininggeomagneticpaleointensities,
J. Geomagn.Geoelectr.,25,
415-435, 1973
Coe, R.S., C.S. Gromm6, and E.A. Mankinen, Geomagnetic paleointensitiesfrom radiocarbondatedlava flows on Hawai andthe question
of the Pacificnon-dipolelow, J. Geophys.Res.,83, 1740-1756, 1978.
Coe, R.S., C.S. Gromm6, and E.A. Mankinen, Geomagneticpaleointensitiesfrom excursion sequencesin lavas on Oahu, Hawaii, J.
Geophys.Res., 89, 1059-1069, 1984.
Coe, R.S., and M. Pr6vot, Evidence suggestingextremely rapid field
established.However, thesecharacteristics
providelittle evidence
variationduring a geomagneticreversal,Earth Planet. Sci. Lett., 92,
for a clearunderstanding
of theprocesses
occurringin the Earth's
292-298, 1989.
liquid core which lead to geomagnetic
reversals.Otherreversal Courtillot,
V., andJ. Besse,
Magnetic
fieldreversals,
polarwanderand
records
witha betterworldwide
coverage
of thesametransition,
core-mantle
coupling,
Science,
237,1140-1147,
1987.
CHAUVINET AL.: GEOMAGNETIC
REVERSALS,
FRENCHPOLYNESIA,2
Dagley, P., and E. Lawley, Paleomagneticevidencefor the transitional
behaviorof the geomagneticfield, Geophys.J. R. Astron. Soc.,36,
577-598, 1974
Duncan, R.A., Paleosecular variation at the Society islands, French
Polynesia,Geophys.J. R. Astron.Soc.,41,245-254, 1975.
Duncan,R. A., and I. McDougall, Linear volcanismin FrenchPolynesia
, J. Volcanol. Geotherm. Res. ,1, 197-227, 1976.
Fleck, R.J., J.H. Mercer, A.E.M. Nairn, and D.N. Peterson,Chronology
of the late Plioceneand early Pleistoceneglacial and magneticevents
in southernArgentina,Earth Planet.Sci.Lett., 16, 15-22, 1972.
Forster J.H., and N. Opdyke, Upper Miocene to recent magnetic
stratigrapyin deep-seasediments,J. Geophys.Res., 75, 4465-4473,
2751
Kent, D.V., and D.J. Spariosu,Magnetostratigraphy
of Carribbeansite
502 hydraulicpistoncores,Initial. Rep. Deep Sea Drill. Proj., 68,
419-433, 1982.
Khodair, A.A., and R.S. Coe, 1975, Determination of geomagnetic
paleointensities
in vacuum.Geophys.J. R. Astron.Soc.,42, 107115, 1975.
Kochegura,V.V., and A. Zubakov,Palaeomagnetic
time scaleof the
Ponto-Caspien Plio-Pleistocene deposits, Palaeogeogr.,
Palaeoclimatol., Palaeoecol., 23, 151-160, 1978
Kono, M., and H. Tanaka, Analysis of the Thellier's method of paleointensity determination, 1, Estimation of statistical errors, J.
Geomagn.Geoelectr..,36, 267-284, 1984.
1970.
Kristjansson,
L., Somestatistical
properties
of paleomagnetic
directions
in
Icelandiclava flows, Geophys.J. R. Astron.Soc.,80, 57-71, 1985.
Fuller, M., I. Williams, and K.A. Hoffman, Palaeomagneticrecordsof
geomagneticfield reversalsand the morphologyof the transitional Kristjansson,L., and I. Mc Dougall, Some aspectsof the late Tertiary
geomagnetic
field in Iceland,Geophys.J. R. Astron.Soc.,68, 273fields, Rev. Geophys.,17, 179-203, 1979.
294, 1982.
Gire, C., J.L. Le Mouel, and T. Madden, Motions at the core surface
Laj. C., S. Guitton,andC. Kissel,Rapidchangeandnearstationary
of the
derivedfrom SV data,Geophys.J. R. Astron.Soc.,84, 1-29, 1986
geomagnetic
field duringa polarityreversal,Nature, 330, 145-148,
Gromm6, S., T.L. Wright, and D.L. Peck, Magnetic properties and
1987
oxidation of iron-titaniumoxide minerals in Alae and Makapuki lava
Laj, C., S. Guitton,C. Kissel,andA. Mazaud,Paleomagnetic
recordsof
lakes,Hawaii, J. Geophys.Res. 74, 5277-5293, 1969.
sequential
fieldreversals
fromthesamegeographical
regionat different
Gurarii, G.Z., The Matuyama-Jaramillo
geomagneticinversionin western
epoch,J. Geophys.Res.,93, 11,655-11,666,1988.
Turkmenia,Izv., Earth Phys.,17, 212-218, 1981.
Gubbins,D., Mechanismfor geomagnetic
polarityreversals,
Nature,326,
Lawley, E.H, The intensity of the geomagneticfield in Iceland during
167-169., 1987.
Neogenepolarity transitionsand systematicdeviations,Earth Planet.
Sci. Lett., 10, 145-149, 1970.
Gubbins, D., Thermal core-mantle interactions and time-averaged paleomagneticfield, J. Geophys.Res.,93, 3413-3420, 1988.
Le Mouel, J.L., Outer-coregeostophicflow and secularvariationof Earth's
Gubbins,D., and J. Bloxham, Morphologyof the geomagneticfield and
geomagneticfield, Nature, 311,734-735, 1984.
implicationsfor the geodynamo,
Nature,325, 509-511, 1987.
Levi, S., Comparisonof two methodsof performing the Thellier exHammond, S.K., S.M. Seylo, and F. Theyer, Geomagnetic polarity
periment (or, how the Thellier method should not be done.), J.
transitions in two oriented sediment cores from the Northwest Pacific,
Geomagn.Geoelectr.,27, 245-255, 1975.
Earth Planet. Sci. Lett., 44, 167-175, 1979.
Levi, S., The effect of magnetite particle size on paleointensitydeHarland, W.B., A.V. Cox, P.G. Llewellyn, C.A.G. Pickton, A.G. Smith,
terminationsof the geomagneticfield, Phys.Earth Planet. Inter., 13,
245-259, 1977.
and R. Walters, A GeologicTime Scale;CambridgeUniversityPress,
New York, 1982.
Liddicoat, J.C, Gauss-Matuyamapolarity transition,Philos. Trans. R.
Soc. London, Ser. A, 306, 121-128, 1982.
Herrero-Bervera,E., andF. Theyer,Non-axisymmetricbehaviorof Olduvai
andJaramillopolaritytransitions
recordedin northcentralpacificdeep- Maenaka, K., Paleomagneticstudy of sedimentsaround the Komyoike
sea sediments,Nature, 322, 159-162, 1986.
volcanicashhorizonin Osakagroup,Senpokuarea, Osakaprefecture,
Herrero-Bervera,E., F. Theyer, and C.E. Hesley, Olduvai onsetpolarity
Japan,Geophys.Res.Lett., 6, 257-260, 1979.
transition:Two detailedpaleomagnetic
recordsfrom the northcentral Mankinen, E.A., J.M. Donnelly, and C.S. Gromm•, Geomagnetic
Pacific sediments,Phys.Earth Planet. Inter., 49, 325-342, 1987.
polarity eventrecordedand 1.1 Ma BP on Cobb Mountain, Clear Lake
Hillhouse, J., and A. Cox, Brunhes-Matuyamapolarity transition,Earth
volcanicfield, California, Geology,6, 653-656, 1978.
Planet. Sci. Lett., 29, 51-64, 1976.
Mankinen,E.A., andG.B, Dalrymple, Revisedgeomagneticpolaritytime
Hoffman, K.A., Polarity transitionrecordsand the geomagneticdynamo,
scalefor the interval 0-5 Ma B.P., J. Geophys.Res., 84, 615-626,
Science,4296, 1329-1332, 1977.
Hoffman, K.A., Behavior of the geodynamoduring reversal: a phenomenologicalmodel,Earth Planet. Sci.Lett., 44, 7-17, 1979.
Hoffman, K.A., Quantitativedescriptionof the geomagneticfield during
the Matuyama-Brunhes
polaritytransition,Phys.Earth Planet. Inter.
24,229-235,
1981a.
Hoffman, K.A., Palaeomagnetic excursions, aborted reversals and
transitional fields, Nature, 294, 67-69, 198 lb
Hoffman, K.A., The testingof the geomagneticreversalmodels/recent
developments,Philos. Trans.R. Soc.London,Ser. A, 306, 147-159,
1982.
Hoffman, K.A., A method for the display and analysis of transitional
paleomagnetic
data,J. Geophys.Res.,89, 6285-6292, 1984.
Hoffman, K.A., Transitional field behavior from southernhemisphere
lavas:Evidencefor two-stagereversalsof the geodynamo,
Nature,320,
228-232, 1986.
Hoffman,K.A., Ancientmagneticreversals:Cluesto the geodynamo,Sci.
Am., 258, 76-83, 1988.
Hoffman, K.A., and M. Fuller, Transitional fields configurationsand
geomagnetic
reversals,Nature,273, 715-718, 1978.
Hoffman, K.A., and S.B. Slade, Polarity transition records and the
acquisitionof remanence:A cautionarynote,Geophys.Res.Lett., 13,
483-486, 1986.
Karlin, R., and S. Levi, Geochemicaland sedimentologicalcontrolof the
magneticpropertiesof hemipelagicsediments,
J. Geophys.Res.,90,
10, 373-10,392, 1985.
Karlin, R., M. Lyle, and G.R. Heath, Authigenicmagnetiteformationin
suboxicmarine sediments,Nature, 326,490-493, 1987.
Kawai, N. T., SatoT. Sueishi,and K. Kobayashi,Paleomagnetic
studyof
deep-sea sediments from the Melanesian basin, J. Geomagn.
Geoelectr., 29, 211-223, 1977.
1979.
Mankinen, E.A., J.M. Donnelly-Nolan, C.S. Gromm•, and B.C. Hearn
JR, Paleomagnetismof the Clear Lake volcanicsand new limits of the
age of the Jaramillo normal polarity event, U.S. Geol. Surv. Prof.
Pap., 1141, 67-82, 1981.
Mankinen, E.A., and C.S., Gromm•,, Paleomagneticdata from the Coso
Range,California and the currentstatusof the Cobb Mountain normal
geomagneticevent, Geophys.Res.Lett., 9, 1279-1282, 1982.
Mankinen, E.A., M. Pr•vot, C.S. Gromm•, and R.S. Coe, The Steens
Mountain (Oregon) geomagneticpolarity transition. 1- Directional
history, durationof episodesand rock magnetism.J. Geophys.Res.,
90, 10,393-10,416, 1985.
McDougall, I., and F.H. Chamalaun,Geomagneticpolarity scaleof time,
Nature, 212, 1415-1418, 1966.
McElhinny, M.W., and R.T Merrill, Geomagneticsecularvariation over
the past5 Ma, Rev. Geophys.,13, 687-708, 1975.
McFadden,P.L., andM.W. McElhinny,A physicalmodelfor paleosecular
variation. Geophys.J. R. Astron. Soc., 78, 809-830, 1984.
McFadden, P.L., and R.T. Merrill, Lower mantle convection and geomagnetism,J. Geophys.Res., 89, 3354-3362, 1984.
McFadden, P.L., and R.T. Merrill, Geodynamoenergy sourceconstraints
from paleomagneticdata,Phys.Earth Planet. Inter., 43, 22-33, 1986.
McFadden,P.L., R.T. Merrill, and M.W. McElhinny, Dipole/Quadrupole
family modeling of paleosecularvariation, J. Geophys. Res., 93,
11,583-11,588, 1988.
Merrill, R.T., and P.L. McFadden, Secular variation and the origin of
geomagneticfield reversals,J. Geophys.Res., 93, 11,589-11,598,
1988.
Nagata, T., Y. Arai, and K. Momose, Secular variation of the
Geomagnetictotal forceduringthe last5000 years,J. Geophys.Res.,
68, 5277-5282, 1963.
2752
CHAUVIN ET AL.: GEOMAGNETICREVERSALS,FRENCHPOLYNESIA,2
Niitsuma, N.D., Paleomagneticand paleoenvironmentalstudy of sed- Theyer, F., E. Herrero-Bervera,and V. Hsu, The zonal harmonicmodel of
imentsrecordingMatuyama-Brunhesgeomagneticreversal,Tohoku
polarity transitions/a test using successivereversals,J. Geophys.
Res., 90, 1963-1982, 1985.
Univ. Sci. Rep. Geol., 43, 1-39, 1971.
Ninkovich,D., N. Opdyke,B.C Heezen,and J.H. Foster,Paleomagnetic Valet, J.P., Inversionsg6omagn6tiques
du Miocene superieuren Crete.
stratigraphy,
ratesof depositionandtephrachronology
in northPacific
Modalit6s de renversementet caracteristiques
du champde transition,
deep-seasediments,
Earth Planet.Sci.Lett., 1,476-492, 1966.
th•se, 195 pp.,Univ. d'Orsay,France,1985.
Olson,P., Geomagneticpolarityreversalsin a turbulentcore,Phys.Earth Valet, J.P., and C. Laj, Invariant and changing transitional field
Planet. Inter., 33,260-274, 1983.
configurationsin a sequenceof geomagneticreversals,Nature, 311,
552-555, 1984.
Opdyke,N.D., D.V. Kent, and W. Lowrie, Details of magneticpolarity
transitions
recordedin highdepositionrate,deep-seacore,Earth Planet. Valet, J.P., L. Tauxe, and D.R. Clark, The Matuyama-Brunhestransition
Sci. Lett.,20, 315-324, 1973.
recordedfrom the lake Tecopa sediments(California), Earth Planet.
Sci. Lett., 87, 463-472, 1988a
Pr6vot, M., E.A. Mankinen, S. Gromm6, and A. Lecaille, High paleointensities
of the geomagnetic
field from thermomagnetic
studieson Valet, J.P., C. Laj, and C.G. Langereis,Sequentialgeomagneticreversals
rift valley pillow basaltsfrom the mid-Atlantic ridge., J. Geophys.
recordedin UpperTortonianmarineclaysin westernCrete(Greece),J.
Res., 88, 2316-2326, 1983.
Geophys.Res.,93, 1131-1151, 1988b.
Pr6vot, M., E.A. Mankinen, C.S. Gromm6, and R.S. Coe, How the
Valet, J.P., L. Tauxe, andB. Clement,Equatorialand mid-latituderecords
of the lastgeomagnetic
reversalfrom the Atlantic Ocean,Earth Planet.
geomagneticfield vector reversepolarity, Nature, 316, 230-234,
1985a.
Sci. Lett., 94, 371-384, 1989.
Pr6vot, M., E.A. Mankinen, R.S. Coe, and C.S. Gromm6, The steens
Van Zijl, J.S.V., K.W.T. Graham, and A.L. Hales, The paleomagnetism
of the StormbergLavas of South Africa, 1, Evidencefor a genuine
Mountaingeomagnetic
polaritytransition2. Field intensityvariation
anddiscussion
of reversalmodels,J. Geophys.Res.,90, 10417-10448,
reversalof the earth'sfield in Triassic-Jurassic
Times., Geophys.J. R.
1985b.
Rea, D.K., and R.J. Blakely, Short-wavelengthmagneticanomaliesin a
regionof rapidseafloorspreading,
Nature,255, 126-128,1975.
Roperch,P., Comportement
du champmagn6tiqueterrestreau coursde
transitionsde polarit6,Th•se, Trav. et Doc. ORSTOM, TDM 26, 209
pp., 1987.
Roperch,P., and A. Chauvin, Transitionalgeomagneticfield behavior:
Volcanic records from French Polynesia, Geophys.Res. Lett., 14,
151-154., 1987.
Roperch,P., and R.A. Duncan, Recordsof geomagneticreversalsfrom
volcanic islandsof French polynesia, 1, Paleomagneticstudy of a
polarity transitionrecordedon a lava sequencefrom the island of
Huahine.,J. Geophys.Res., this issue.
Senanayake,W.E., M.W. McElhinny, and P.L. Mc Fadden,Comparison
betweenthe Thellier'sandShaw'spaleointensity
methodsusingbasalts
lessthan 5 million years old, J. Geomagn. Geoelectr.,34, 141-161,
Astron. Soc., 7, 169-182, 1962.
Watkins,N.D., Shortperiodgeomagnetic
eventsin deep-seasedimentary
cores, Earth Planet. Sci. Lett., 4, 341-349, 1968.
Watkins, N.D., andA. Abdel-Monem,detectionof the Gilsa geomagnetic
polarity event on the island of Madeira, Geol. Soc. Am. Bull., 82,
191-198, 1971.
Weeks,R.J., M. Fuller, andI. Williams, The effectsof recordingmedium
uponreversalrecords,Geophys.Res.Lett., 15, 1255-1258,1988.
Williams, I., and M. Fuller, Zonal harmonic models of reversal transition
fields,J. Geophys.Res.,86, 11657-11665, 1981.
Williams, I., R. Weeks, andM. Fuller, A model for transitionfield during
geomagnetic
reversals,
Nature,332, 719-720, 1988.
Wilson,R.L., P. Dagley,andA.G. McCormack,Paleomagnetic
evidence
about the sourceof the geomagneticfield, Geophys.J. R. Astron.
Soc., 28, 213-224, 1972.
1982.
Shaw, J., Strong geomagneticfields during a single Icelandic polarity
transition,Geophys.J. R. Astron. Soc.,40, 345-350, 1975
Shaw, J., Further evidence for a strong intermediatestate of the paleomagneticfield, Geophys.J. R. Astron.Soc.,48, 263-269, 1977
Sigurgeirsson,T., Directionsof magnetizationin Icelandicbasalts,Adv.
Phys., 6, 240-246, 1957.
Sueishi, T., T. Sato, N. Kawai and K. Kobayashi, Short geomagnetic
episodein the MatuyamaEpoch,Phys.Earth Planet.Inter., 19, 1-11,
A. Chauvin
andP. Roperch,
ORSTOMetLaboratoire
deG6ophysique
Interne,Universit6de RennesI, av. G6n6ralLeclerc, 35042 Rennet
France.
R.A. Duncan,Collegeof Oceanography,
OregonStateUniversity
Corvallis, OR97331.
1979.
Thellier, E., and O. Thellier, Sur l'intensit6du champmagn6tiqueterrestre
dansle pass6historiqueet g6ologique,Ann. Geophys.,15, 285-376,
1959.
(ReceivedMarch 10, 1989;
revised October 3, 1989;
accepted
November7, 1989.)