Magnetization of the La Palma Seamount Series: Implications for Seamount Paleopoles Jeff Gee, Hubert Staudigel, Lisa Tauxe, Thomas Pick
To cite this version:
Jeff Gee, Hubert Staudigel, Lisa Tauxe, Thomas Pick. Magnetization of the La Palma Seamount Series: Implications for Seamount Paleopoles. Journal of Geophysical Research, American Geophysical Union, 1993, 98767, pp.11743-11754. 10.1029/93JB00932. insu-01864349
HAL Id: insu-01864349 https://hal-insu.archives-ouvertes.fr/insu-01864349 Submitted on 29 Aug 2018
HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 98, NO. B7, PAGES 11,743-11,767, JULY 10, 1993
Magnetizationof the La Palma SeamountSeries: Implicationsfor SeamountPaleopoles
JEFFGF_11,1 HtmERT STAUDIGI•L, 2 LISA TAUX•, 3 AND THOMAS PICK
ScrippsImtitutlon of Oceanography,La Jolla, California
YVES GALLET
Institutde Physiquedu G?obede Paris
Paleopolesdetermined from seamountmagnetic anomalies constitute the major data sourcefor the Pacific apparentpolar wander path, but relatively little is known about the processesof remanence acquisitionin seamounts.Since magneticanomalies reflect both natural remanence(NRM) and the inducedfield, it is importantfirst to assesswhether the NRM is likely to representan original field directionand secondto constrainthe magnitudeof the inducedcomponent. To this end, we present paleomagneticdata from an uplifted, subaeriallyexposed section through a seamounton La Palma, Canary Islands. The PlioceneSeamount Series of La Palma comprisesa >6 km sequenceof alkalic extrusivesand intrusives which includes all lithologieslikely to be volumetricallyimportant in seamounts. The structuraltilt of the SeamountSeries allows separation of early thermalor chemicalremanence from magnetizationcomponents acquired after tilting (e.g., viscousremanence). The NRM providesa poor indicationof the originalmagnetization direction, although the characteristicmagnetization of manyLa Palma samplesis compatiblewith the originalpretilt direction.Hydrothermal alteration has resultedin the productionof Ti-poormagnetite and an increasingcontribution of hematitewith increasingdegree of alteration.More importantly,well-defined magnetization directions which deviatefrom any reasonable geomagneticdirection at La Palmacan be attributedto hydrothermalalteration in a differentpolarity than prevalentduring the originalmagnetization. Based on a comparisonof the magnitudeof low-stability componentsof magnetizationand laboratory acquisition of viscousremanence and previousestimates of the inducedmagnetization, we concludethat viscousand inducedmagnetization probably account for 15-25% of the total magnetizationof seamounts.The resultingpaleopole bias is a functionof the polarityand paleolatitudeof the seamountand rangesfrom 4ø to 16ø for Cretaceousseamounts in the Pacific.
INTRODUCTION plex anomaly patterns may be accommodatedby arbitrarily' subdividingthe seamountinto regionsof uniform magnetiza- Magnetic anomaliesassociated with seamountshave long tion [e.g., McNutt, 1986], the spatialcoherence of residuals been recognized as a source of tectonic information, and suggeststhe standardleast squaresinversion is based on an seamountpaleopoles have been particularlyuseful in r 11,743 11,744 GEEET AL.: MAGNErlZATION OF LA PALMASEAMOUNT SERIES beensuggested for large seamountswith a significantportion layeredsequence of entirely submarineextrusives and volcani- of intrusive material based on results from the La Palma clastics,three generationsof alkalic dikes/sills,and a basal SeamountSeries [Gee et al., 1989]. In addition,the presence plutonicsequence (Figure lb). Thesubmarine extrusive series of multiplepolarities is often difficultto assessfrom models. is consistentlybedded with a dip of 48o alongan azimuth For example,the ideal body solutionfor JasperSeamount of 230o [Staudigel,1981], which representsa tilt roughly doesnot requiremixed polarity [Parker, 1991] despitethe radial with respectto the presentcenter of intrusiveactivity demonstrablylong duration(>7 nay.) of volcanismwhich in the southeasternand centralportions of the caldera. The stronglysupports the presenceof both normaland reversed extrusiveseries consists predominantly of pillow lavas(50%), polarityrocks [Pringleet al., 1991]. Finally, the effects pillow fragmentbreccias, submarine scoriaceous breccias and of' progressivehydrothermal alteration (perhaps in different hyaloclastites(terminology of Fisherand Schmincke[1984]) polarity intervals)on the original magnetizationare poorly and ranges in compositionfrom alkalic basalt to mugearite understood. and rare trachyte. Three generationsof dikes and sills are In this paper we discussthe origin of the magnetization observedin the basementcomplex [Staudigel and Schmincke, in an uplifted, tilted sectionthrough a seamountexposed 1984; Staudigelet al., 1986]. The earliest generationof on La Palma, Canary Islands. The La Palma Seamount dikes (Group I) is orientedapproximately orthogonal to the Seriesprovides a uniqueopportunity to studythe magnetic bedding of the layered extrusive series. On the basis of propertiesand originof the magnetizationin seamounts.In their orientation and the similarity in both compositionand contrastto dredgingand drilling, which sample only the alteration history to the extrusive series, these dikes are outermostpart of the volcanic edifice, the exposuresof interpretedas feedersfor the extrusiveseries. Sills (Group11) the SeamountSeries provide about 4 km of stratigraphic occur within the layered extrusiveseries, parallel to bedding, relief and include a wide range of extrusivesand intrusives and also form a 2 km sheeted sill swarm which underlies the which encompassall lithologieslikely to be importantin extrusivesequence. Crosscuttingrelationships suggest that seamounts. Age constraintsfor the SeamountSeries and mostof the sillsare younger than the GroupI dikes[Staudigel the overlyingsubaerial flows allow generalcorrelation to the et al., 1986]. GroupIII dikes generallyhave steepdips, may geomagneticreversal time scale and facilitateevaluation of crosscutthe overlying Cobefta lavas and are typically less the role of multiple polarities in generatinginhomogeneous alteredthan the countryrock. These dikes are interpretedas magnetization.The tiltingof the seamountsequence (~50 ø) membersof a radial dike swarm feeding the Cobeftalavas. allows separationof pretilt thermalor chemicalcomponents Isotopicage dating suggests an agerange of 0.51 to 0.80 Ma from later viscous, chemical and thermal magnetizations for the radial dikesin the Barrancode las Angustias[Staudi- acquiredafter tilting. While the magnitudeof tilting in La gel et al., 1986]. The basalplutonic complex is exposedin Palma is probably atypical, the effect is analogousto the the Calderade Taburienteprimarily as numerous,small (10s latitudinal translation common in Pacific seamounts. Results to 100s of m) gabbro,leucogabbro and ultramaficintrusions, from La Palma indicate that most assumptionsmade in some of which are probably associatedwith the subaerial modelingseamount anomalies may be questionable,but our Cobeftalavas. Small leucocraticgabbro intrusions also occur data alsoprovide an estimateof the bias inherentin seamount within the extrusiveportion of the sequence.Staudigel and paleopoleswhich may ultimatelycontribute to a moreaccurate Schmincke [1984] refer to the extrusives, feeder dikes and Pacific apparentpolar wanderpath. sills as the SeamountSeries. For simplicity, and because such rocks are expectedto occur in seamountsas well, we GEOLOGY OF THE LA PnLMA SEAMOUNT SE•ms will also use the term to include the alkalic gabbrosof the La Palma is the northwesternmostisland of the Canary basalplutonic complex. island chain and is locatedon Jurassicoceanic crust [Hayes The SeamountSeries has been subjectedto hydrothermal and Rabinowitz, 1975]. Schmincke [1976] has summarized alteration[Hernandez-Pachecho and Fernandez-Santin,1975; the geology of the Canary Islands and a recent review Staudigel, 1981]. The upper part of the sequenceis of the geologyof La Palma is given by Staudigeland characterizedby low-temperatureassemblages of smectite, Schmincke[1984]. The oldest rocks exposedon La Palma analcime,carbonate and low-temperaturezeolites. Alteration are biostratigraphicallydated at 3-4 Ma [Staudigel,1981] increases,both in pervasivenessand extent, with stratigraphic and volcanismextends to historictimes [e.g., Middlemost, depth. Interlayeredchlorite/smectites give way to epidote 1972]. The island consistsof two major structural units. and (locally) actinolite(Figure 1). The upper stabilitylimit The Pleistocene subaerial lavas of the Coberta Series cover of analcime correspondsto a temperatureof approximately more than 90% of the island and unconformablyoverlie the 200øC [Liou, 1971]. The first appearanceof epidote Pliocenesubmarine and plutonicbasement complex [Afonso, in hydrothermal systems occurs at temperatures greater 1974]. The oldest lavas of the Coberta Series are dated than 200o to 250øC and actinolite denotes temperatures at 1.61 Ma [Abdel-Monemet al. 1972] (recalculatedusing in excess of ~300øC [Bird et al., 1984]. The silicate the decay constantsof Steigerand J•iger [1977]). The assemblagestherefore suggesta fossil geothermalgradient basementcomplex is a >6 km sectionof alkalic extrusives of approximately200øC/km. The alterationassemblages and intrusivesexposed within the Calderade Taburienteand are similar to those found in upwelling regions of present the steep-walledcanyon (Barranco de las Angustias)which day hydrothermalsystems in basaltic lavas (e.g., Iceland drains the caldera (Figure la inset). Erosion rates in the [Palmassonet al., 1979] and Reunion Island [Demangeet canyon are high (1-2 mm/yr as determinedby erosionof al., 1989]). The absenceof similar alteration in the Cobefta 1973 drillholes), providing extremely fresh exposuresalong lavas and in recent pyroclasticdeposits on the flanks of the the canyon bottom. canyon indicatesthat the metamorphismof the extrusives, The basementcomplex comprisesthree units: an approx- Group I dikes, and the sill complex was associatedwith imately 2500 m (~1800 m prior to inflationby intrusives) seamountconstruction. Later localizedhydrothermal activity, GEE El' AL: MAGNErlZATIONOF LA PALMA SEAMOUNTSERIES 11,745 17.91øW 17.90 ø 17.89 ø 17.88 ø 17.87 ø A 28.70 ø N 835-844 0 846 847 0 0 28.69 ø 0 O0 0 865 792 pro-out 28.68 ø '/ 750 calderarixn road 28.67 ø stream bed 0 1 km subaeriallavas met. isograd Fuencaliente•/ B SW NE • plutOffXcs- 600 m • ....?:' t%1%.... -400 'X/•S :•://':: % % ....::::::::::::::::::::::::: ....• .... ::..:" % % % . •s•;- ....•:•::.:' ..;::•,::.• % ....•::[•:•:-- % - 200 I i I i 0 2 4 6 km Fig. 1. Paleomagneticsample locations and schematic cross section of La PalmaSeamount Series. (a) Samplelocations andsimplified geology shown for westernmostCaldera de Taburienteand Barranco de las Angustias(see inset). Squares indicatedetailed sill samplingsites and only sitesdiscussed in the text are labeled.An additional18 sitesare widely dispersedwithin the calderato the northeastof themapped region. Plutonics, sill swarmand extrasives are indicated by hatchedpattern, grey and white, respectively. Note that southeastern margin of canyonhas not been mapped. Metamorphic isogradsmodified from Staudigel [1981] based on presentsample collection. Labels: an, analcime;pro, pumpellyite; ab, albite;ep, epidote;act, actinolite.(b) Simplifiedcross section approximately parallel to the canyonbottom. Extrasives are consistentlybedded and dip ~50 ø to SW. Crosscuttingrelationships of GroupI feederdikes (solid), Group II sills (grey), and GroupIII verticaldikes (white) and plutonicsshown schematically. as evidencedby sulfur-richhydrothermal waters, is currently be locally quite steep. For example,steep-sided (50-600 ) presentin the southeasternand centralportions of the caldera. pillow walls (10-20 m high) have been describedfrom the The tilt of the extrusive series plays a central role submergedportion of Kilauea's east rift zone [Fornari et in determining the origin of the magnetization in the al., 1978] althoughthe slopesof seamountsare typically Seamount Series. Although bedding is less obvious in <20o [e.g., Lonsdaleand Batiza, 1980; Baroneand Ryan, volcanic sequencesthan in sediments, layering in fine- 1990]. The orientationof geopetalstructures (e.g., the single grained hyaloclasfites,flattened pillow bottoms overlying drainedpillow surfacein Table 1) and the consistentbedding bedded hyaloclastites,and the upper surfacesof partially- of fine-grained (sand-sized)hyaloclastites in the Seamount drainedpillows yield a consistentattitude of 48o dip toward Series suggeststhat the presentbedding approximates the 230o (Table 1). Constructionalslopes of seamountsmay paleohorizontal. 11,746 GEE ET AL.: MAGNETIZATIONOF LA PALMA SEXMOO• SERIES TABLE 1. BeddingMeasurements for the SeamountSeries therefore yield four expected magnetizationdirections (in presentday coordinates)for comparisonwith the observed paleomagneticdata. All magnetic data in this paper are in Attitude presentday coordinatesunless otherwise noted. We refer Dip/Dip Estimated Measurement Measurement Direction Error & Type Location# to the normal(0 ø, 48ø) and reversedpolarity (180 ø, -48ø) GAD directions as the expectedposttilt directionsand the 44/228 + layeredhyaloclastite unit 2 correspondingGAD directionscorrected for the tilt of the 55/235 +++ pillow bottom baseof unit 5 SeamountSeries as the expected pretilt normal (19 ø, 10ø) and 55/232 +++ pillow bottom baseof unit 5 reversed(199 ø, -10ø) directions.We estimatethe uncertainty 58/245 ++ pillow bottom baseof unit 5 of the pretilt directionsto be --,10ø owing to the possible 54/235 ++ layeredhyaloclastite top of unit 10 errors in the structural correction. This error is similar to 40/240 ++ hyaloclastitein unit 10 the dispersionin paleomagneticdirections (--,12 ø) expected pillow interstices at La Palma based on the latitudinal dependenceof virtual 41/232 +++ pillow bottom baseof unit 11 geomagneticpole dispersionfrom paleosecularvariation [Cox, 45/228 +++ pillow bottom baseof unit 11 1970; McFaddenand McElhinny, 1984]. 45/225 +++ pillow bottom baseof unit 11 45/238 +++ pillow bottom baseof unit 11 SAMPLE COLLECTION 25/225' + layeringin pillow unit 12 Over 1200 oriented cores (-,,200 paleomagneticsites) stringerbreccia were collectedfrom the La Palma SeamountSeries during 35/190' + layeringin pillow unit 12 stringerbreccia two samplingtrips in 1985 and 1988 (Table 2). Standard 50/220 +++ pillow bottom baseof unit 12 paleomagneticcores were drilled and orientedwith a magnetic 50/240 +++ pillow bottom baseof unit 12 compass.Some cores were alsooriented with a suncompass. 58/251' ++ pillow bottom baseof unit 12 Comparisonof the two methods indicated no bias in the 53/259* ++ pillow bottom baseof unit 12 magneticcompass orientations, and the sun compassreadings 53/195' ++ hyaloclastite top of unit 13 indicatea magneticdeclination nearly identicalto the IGRF 50/225 ++ layeringin pillow baseof unit 13 declinationat the site (-10ø). More than90% of the siteswere fragmentbreccia located in the layered extrusive series and sill/dike swarm 43/206* ++ pillow bottom baseof unit 14 (Figure 1a). Approximately90 siteswere drilled in dikes/sills, 58/219' ++ pillow bottom baseof unit 14 includingthree localitieswith 10-15 dikes/sillseach. Half of 52/224 ++ layeredhyaloclastite top of unit 15 thesewere identifiedas Group II sills, primarilyon the basis 48/220 +++ pillow bottom baseof unit 15 of their attitude(mean dip/direction of dip 54ø/223ø),with 48/222 +++ pillowbottom baseof unit15 equalnumbers of samplesfrom GroupI and GroupIII dikes. 40/225 +++ partially drained unit 17 More than 60 pillow sites span the range of compositions pillow and alteration temperaturesfound in the SeamountSeries. 48/246 ++ pillow bottom baseor'unit 19 The volcaniclasticunits sampled (25 sites) were primarily 48/245 +++ layeredhyaloclastite unit 20 pillow fragmentbreccias, lapillistones and scoriaceouspillow 50/244 +++ layeredhyaloclastite unit 20 breccias. Gabbroswithin the extrusiveseries and gabbro 45/215 + - water channel screenswithin the sill complexwere sampledwhere possible. 48/220 + - water channel The remaining10% of the sitessampled gabbros and dikes 45/222 + hyaloclastitein - at 18 widely spacedsites within the caldera (not shown in ankaramiticpillows Figure la). 48/230 weightedmean USE OF THE SUM OF VECrOR * Measurementsfrom this study, remaining data from DIFFERENCE MAGNITUDES Staudigel[ 1981]. The SeamountSeries samplesexhibit both normal and # With the exceptionof the last threedata, all measurements reversedpolarity characteristicmagnetizations and most sam- are from the base of the Barrancode las Angustias. Unit ples have a low-stabilitycomponent of normalpolarity, which numbersfrom Staudigeland Schmincke[1984]. often significantlyreduces the initial remanent intensity of & Qualitativeerror estimate: +++(5ø), ++(5-10ø), +(> 10ø). reversedpolarity samples(Figure 2a). As a result, param- eters such as blocking temperatureand median destructive We have usedthe observedlayering of the extrusiveseries field, which are basedon the naturalreinanent magnetization as the basisfor our structuralcorrection. Paleomagnetic data (NRM), are biassedby polarity. To amelioratethis problem from the least alteredpillow lavas in the SeamountSeries we have used the vector difference sum (VDS) as a basis for provide independentsupport for this structural correction. calculatingmagnetic parameters such as blocking temperature. The mean characteristicmagnetization direction obtained The sum of the vectordifference magnitudes between succes- from theseten sites(1.60 , 50.3ø;tilt corrected)is statistically sive demagnetizationsteps gives an estimateof the sample's indistinguishablefrom the expectedgeocentric axial dipole original NRM, assumingthat the secondarycomponent is (GAD) directionat La Palma (0ø, 48ø). While original acquiredby reorientationof momentsin existinggrains rather paleoslopescannot be excluded, reasonableslopes (up to than net addition of magneticmaterial. This is in effect the 10ø) would have only a minor effect on our structural sum of each componentaligned to a commondirection. The correction,altering the azimuth of the rotational axis by a VDS is, of course, also artificially increasedby any noise maximumof 150 with little changein the dip. The GAD in the directional measurementand artificially reducedif an direction at La Palma and the tilt of the Seamount Series insufficient number of demagnetizationsteps were made to GEE ET AL: MAGNETIZATIONOF LA PALMA SEAMOUNTSERIES 11,747 TABLE 2. Summaryof MagneticProperties of La PalmaLithologies Number Number Low-stability Characteristic of of NRM, Koenigsberger Component, MDF, Tb, Remanence Lithology Specimens Sites A/m Ratio, Q & % VDS mT øC Directions Gabbro s 229 16 7.28 3.05 14.9 18.1 605 In extrusives/sills 65 5 8.55 2.66 22.4 21.6 610 pretilt In caldera 164 11 6.78 3.21 12.1 17.2 605 posttilt Dikes/Sills 799 88 2.46 1.88 35.5 20.6 595 GroupI (feeder) 176 22 2.14 1.79 40.3 14.1 585 pretilt GroupII (sills) 372 44 2.47 1.72 29.6 26.0 605 pretiltto posttilt GroupIll (vertical) 251 22 2.67 2.19 44.6 16.4 575 posttilt Pillow lavas # 731 66 3.87 5.29 21.7 34.0 615 No epidote 130 13 8.43 11.58 25.4 19.1 595 pretilt Ep-pm 227 18 4.87 5.62 16.9 39.6 625 pretilt Pm-act 330 29 1.89 3.00 19.7 34.0 620 pretilt >Actinolite 44 6 0.22 2.22 40.3 54.9 635 pretilt Volcaniclastics 376 25 3.11 5.45 15.0 26.1 605 Pillow fragments 240 25 4.38 6.19 13.5 23.5 605 random Hyaloclastite@ 136 36 0.86 4.15 17.1 30.5 600 pretilt Subaerial lavas 86 12.12 18.69 - - - Collectionof fully orientedsamples forms subsetof data reportedby Gee eta/. [1989]. Arithmeticmeans are used throughout.NRM is naturalremanent magnetization. Low-stability component, median destructive field (MDF), and maximumblocking temperature(Tb) all basedon vectordifference sum (VDS) as discussedin text. & Q is the ratio of the remanentto inducedmagnetization. # Subdivisionsbased on metamorphicisograds (ep, epidotein; pm, pumpellyiteout; act, actinolitein) in Figure 1. Actinolitegroup includes results from keratophyres. @ Includesinterstitial hyaloclastite material in pillowunits and pillow fragmentbreccias as well asbedded lapillistones. A w, Up 400 B 500øC 2.02.5 300 1.5 620 S / N 0.5 200 2• 4• 6• c E,Down I 2• ' 460 ' 660 Temperature(øC) Fig. 2. Illustrationof theuse of thevector difference sum (VDS). (a) Vectorendpoint diagram of thermaldemagnetization of a dike samplefrom La Palma. The dike hasa reversedpolarity characteristic magnetization (500ø-620øC) and a significantnormal polarity overprint (100ø-300øC). The low-temperaturecomponent is estimatedas 65% of the VDS usingthe techniquedescribed in the text. Solid(open) circles are projectionsof the vectoronto the horizontal(vertical) plane.(b) Thermaldecay of the NRM (solidcircles) and VDS (opencircles). Note the significantdifference between the originaland reconstructed remanent intensities. (c) Histogramof blockingtemperatures derived from the VDS. The componentremaining above 600øC (solid) is usedas an estimateof the importanceof hematitewhen this component exceeds 5% of the VDS. 11,748 GEE ET AL: MAGNETIZATIONOF LA PALMA SEAMOUNTSERIES clearly delineatethe full complexity of the remanencevectors. have been previously reported [Gee et al., 1989]. Two The effect of noise is small in most casesand is partially polaritiesare evidentin the NRM data and thesedata can be offsetby the incompleteseparation of the two componentsso arbitrarilyseparated along the E-W vertical plane and their that the VDS probablyprovides a reasonableestimate of the distributionscontoured on an equal area projection(Figure originalNRM. 3). The northernhemisphere (normal polarity) directions are The vector differencemethod also allows a quantitative skewed toward the expectedpretilt normal direction,but the estimateof the magnitudeof the variouscomponents. In the densitymaximum (11.2%) is locatednear the posttiltnormal examplein Figure2, the low-stabilitycomponent (359.4 ø, GAD direction(Figure 3a). The reverseddirections (Figure 44.6ø) is well definedbetween 100 ø and 300øC and the char- 3b) have a maximum which lies between the pretilt and acteristicmagnetization (170.0 o , -30.5o ) is isolatedbetween posttilt reverseddirections. A secondarymaximum occurs 500o and 620øC. Using the principalcomponent directions slightly to the east of the reversedGAD directionand can for these well-defined segments,the vector differencesin the largelybe attributedto samplesfrom gabbroicintrusions. The intervalfrom 300o to 500øC can be expressedas a linear expectedpretilt directionsare based on the bedding attitude combinationof the two known directionsin a least squares observedin the layeredportion of the SeamountSeries and sense[Press et al., 1986, p. 59]. A similar procedurefor thusstrictly apply only to the layeredextrusive series, Group estimatingthe magnitudeof componentsfrom demagnetiza- I dikes and the sills. Nonetheless, deviations from these tion data was suggestedby Zijderveld[1967]. We will show expecteddirections are a generalindication of the severity later that the low-stability componentpostdates tilting of the of secondarymagnetization in the sequence. Clearly, the SeamountSeries and can reasonably be attributedto a viscous original pretilt directionsfor the SeamountSeries are not well remanence(VRM) acquiredat low to moderatetemperature. representedby the NRM directionsas assumedin modeling Sincetemperature is muchmore efficientthan alternating field of seamount anomalies. treatmentfor removingVRM [Dunlop, 1973], we have used The effect of hydrothermalalteration on the NRM in- thermaldemagnetization curves exclusively for calculatingthe tensity is most clearly demonstratedin pillow lavas, where percentageof secondarymagnetization. For this studyeach complexitiesdue to differencesin cooling rate and original thermallydemagnetized specimen was typically subjectedto magneticmineralogy should be minimal (Figure 4). The 15-20 steps, sufficientfor the purposeof calculatingthe fossil hydrothermalsystem exposedon La Palma (Figure VDS. 1) is similar to presentlyactive basalt-hostedhydrothermal RESULTS systemsin Iceland [Palmasson,1979] and on ReunionIsland [Dernangeet al., 1989] and recordsa geothermalgradient Natural RemanentMagnetization of ,-,200øC/km.Samples from the least-alteredportion of The magnetic anomaly of a seamountmeasured at the the sectionhave NRM intensitiescomparable to dredgedand sea surface arises from the combined induced and remanent drilled extrusivesfrom seamounts[e.g., Kono, 1980; Gee magnetizations,without benefit of magneticcleaning. Thus, et al., 1988]. Progressivehydrothermal alteration results in the NRM intensityand directionare the most pertinentdata approximatelytwo ordersof magnitudereduction in the NRM for comparisonwith anomalymodels. The NRM intensities intensityover the temperaturerange (<2000 to >300øC) and inducedmagnetization of the La PalmaSeamount Series indicatedby the silicate assemblages. A N B N + normal polarity mode Fig. 3. Equal-areacontour plot of NRM directions(in geographiccoordinates). (a) Lower hemisphereprojection of the northernhemisphere (normal) directions. The maximumdensity (11.2 %) is locatednear the posttilt GAD (square)and the presentfield (triangle)di-rections for La Palmarather than near the expectedpretilt direction (circle). (b) Upper hemisphere projectionof •outhernhemisphere (reveraed) directions. The maximumdensity (7.2%) li• betweenthe GAD reversed direction(square) and the expected,pretilt reveraeddirection (circle). One percentisodenisty contours are shown.The antipodeis plottedfor upperhemisphere directions in Figure3a and lower hemispheredirections in Figure3b. GEE ET AL: MAONE'rrzATIONOF LA PALMA SEAMOUNT SERIES 11,749 10- a circular standard deviation of more than 20 ø have been a excluded from consideration. actinolite n=44 The range of demagnetizationbehavior in samplesfrom La Palma is illustratedin Figure 5. Approximately20% of the sampleshave univectorialbehavior, as definedby a low- stabilitycomponent < 10% of the VDS. Univectorialdirections compatible with a posttilt magnetizationare most common in gabbroic intrusions(Figure 5a) and vertical Group lIl I i I I I dikes (Figure 5b) althoughGroup II sills and a few extmsive -2 -I 0 1 2 sites also have single component posttilt magnetizations. b Directions compatiblewith a pretilt magnetizationare more 50- pmout -act in ';'":r¾.-'--'•.-."•• typical for the extrusiveportion of the SeamountSeries as 40- n= 331 r-l"•'•"":•[••'••••:•.•....-• well as many of the Group I and II feederdikes and sills. 2t0- Nearly univectorialpretilt directionsare particularlycommon in the pillow lavasof the upperportion of the SeamountSeries 2t3- (Figure 5c). Low-stabilitycomponents of variablemagnitude 10- characterizemany sitesalthough the shallowpretilt inclination can be isolated(Figures 5d and 5e). A leucocraticgabbro I i I I -2 -1 0 1 from within the extrusiveseries (Figure 50 illustratesa more complex demagnetizationbehavior but a pretilt magnetization '" 40- is also indicatedby the final shallow inclination. ep in- pm out A significant number of samples from all lithologies -½z 30 - n = 227 display characteristicmagnetization directions which deviate • _ from either the expectedpretilt or posttilt directions(Figure o 20- 6). In many suchsamples, a discrepancyis observedbetween E 10 the results of AF and thermal demagnetization(compare = - Figures6a and 6b). Note that the north and up component I ! in the thermallydemagnetized sample (520 ø-540 ø C) appears -2 -1 0 1 2 as the highestcoercivity component in the AF sample. We 30- suggestthat this inconsistencyresults from the superposition d of reversedand normal magnetizationswhich are inadequately noepidote :"• separatedby either technique.The demagnetizationdiagrams 20- n= 227 •-'-.-J of the pillow lavas in Figures 6c and 6d illustratethat these - anomalousdirections (NW, up and W, up) are well-defined. The inflection in the inclination curve of LP865b2 at 200øC is _ also typical of many samplesfrom the SeamountSeries. The consistentorientation of the low-stabilitycomponent parallel , i'...... [2•::.... to the present field directionat La Palma (e.g., Figure 6d) -2 -1 0 1 2 rules out sample misorientationas an explanation of these log NRM (A/in) anomalous directions. Results from a volcaniclasticunit (Figures 6e and 6f) Fig. 4. NRM intensitiesfor pillow lavasas a functionof alteration. suggestthat these aberrant directions may be the result of Note the reductionin intensityby nearly two ordersof magnitude overthe temperaturerange (<200 ø- >300øC) indicatedby the silicate alteration.The samplein Figure 6e is an amoeboidalpillow assemblages.Groupings are basedon the isogradsshown in Figure which has a pretilt normal magnetizationcompatible with 1. Actinolitesamples include two keratophyrescreens. the in situ high temperatureemplacement of the amoeboidal pillows in this unit [Staudigeland Schmincke,1984]. The water-saturatedmatrix probably never reached temperatures DemagnetizationResults > 100øCand may havebeen mechanically disturbed during Approximately1000 specimenswere subjectedto stepwise pillow emplacement.Both AF and thermaldemagnetization thermal or alternatingfield (AF) demagnetization,with a of the interstitial hyaloclastitematerial (Figure 60 yield a minimum of three samplesfrom each site. The majority of northward and up componentsimilar to that noted in other sampleswere thermally demagnetized,as this method often lithologies.These results suggest that the highglass content of providesmore reliable results, in 50øC stepsfrom 100ø to the hyaloclastitehas resulted in significantalteration while the 500øC followedby 20øC stepsuntil the directionsbecame amoeboidalfragment maintainedits original magnetization, randomor the sampleintensity fell below 1% of the NRM albeit with a significantviscous overprint. The origin of these value. The uniformity of temperaturesteps was maintained anomalousdirections will be discussedmore fully below. to facilitate comparison of blocking temperature spectra. Blockingtemperature data from the demagnetizationstudies and maximum unblocking temperatures. Componentsof also yield a first order characterizationof the magnetic magnetizationwere definedby principalcomponent analysis mineralogy.Thermal demagnetizationof isothermalreinanent [Kirschvink,1980] wheneverpossible. Specimensfor which magnetization,petrographic observations, microprobe data no stablecharacteristic magnetization could be isolatedwere and Curie temperaturedeterminations provide additional analyzedusing the great circle method described by McFadden constraintson the magneticmineralogy; however, the blocking and McElhinny[ 1988]. A smallnumber of sites(< 10%) with temperaturedata are particularly useful since this data is 11,750 GEEEl' AL.: MAGNETIZATION OFLA PALMASEAMOUNT SERIES A B W Up 300 S • ß- I 5.0 A/m NRM 560øC LP704g2 gabbro 580 / N LP806c1 dike _x• NRM '" E, Down E, Down D W, Up c LP743al W,Up pillow $• I N 8.0 A/m S NRM LP740b2 '•x E, Down -0.4 A/m dike • NRM E, Down W, Up E NRM F LP855dl gabbro W, Up 680 LP805fl -•$ pillow 450 - • 580 6OO 400oc s N 250 - 1.0 Aim 0.1 A/m E, Down E, Down Fig. 5. Representativevector endpoint diagrams (geographic coordinates) from La Palma.Nearly univectorial demag- netizationin (a) reversedpolarity gabbro and (b) normalpolarity dike. Both directions are posttilt. (c) Normalpolarity directionwhich predates tilting. (d) Shallow,normal polarity pretilt component with moderatelow-stability compo- nent.(e) Reversedpretilt direction with substantial normal polarity overprint. (f) Complexdemagnetization behavior in leucogabbrowithin the extrusive series. Solid (open) circles are projections of the vector onto the horizontal (vertical) plane. universallyavailable. We define the maximumblocking Intrusivesfrom La Palmaare dominated by Tb near600øC temperature(Tb) as the temperaturestep at which the (Table 2). Gabbroshave Tb rangingfrom 580ø-680øC, remanentintensity drops to below2% of the VDS intensity althoughthe majorityof sampleshave peak unblocking or the directionbecomes random. Peak unblockingmay, and overa verydiscrete interval (540ø-600øC) compatible with oftendoes, occur at a temperature A B LP604al LP604a2 W, Up W, Up gabbro gabbro Iloo35 • ..• s t "•5, •-1• N s580/ _ N 540øC E, Down NRM -2 A/m 200'• NRM E, Down W,Up LP847c2 pillow C D 350 W,Up LP865b2 350 pillow ' _ •••••200øC 1.0A/m NRM 500øC E, Down NRNI 660 S I N 0.3 A/m E W, Up LP792c 1 S F hyaloclastite 620 W, Up 200øC 40 20mT5 NRM 100 0.3 A/m LP792kl • •l-" - N amoeboidal NRM 0.1A/m E, Down pillow E, Down Fig. 6. Vectorendpoint diagrams illustrating anomalous remanence directions (geographic coordinates). Northward and shallowupward final direction in (a)AF-demagnetized sample corresponds to an intermediate (520ø-540øC) component in thecorresponding Co) themally demagnetizedsample. (c) Well-definedanomalous remanence direction with characteristic inflectionin inclinationprojection at ,,,200øC.(d) An extremeexample of an anomalousdirection (W, Up) with a low- stabilityoverprint similar to the presentday fieldat La Palma. (e) Amoeboidalpillow fragment maintains a shallow, pretiltdirection, whereas (f) thesurrounding hyaloclasfite shows anomalous north and up directions.Solid (open) circles areprojections of the vectoronto the horizontal(vertical) plane. commonin sills. In contrastto gabbrosamples, dikes/sills in accorawith a moreTi-rich titanomagnetite although a high often have a distributedblocking temperature spectrum which blocking temperaturephase is always present. Both Tb includessignificant unblocking at temperaturesof 250ø- and the mediandestructive field increasewith the degreeof 400øC. alteration(Table 2). Pillow samplesfrom the more altered Perhapsthe most unexpectedresult from the study of portionof the SeamountSeries may havea significantamount the SeamountSeries is the uniformly high Tb (500o- of remanenceremaining at 600øC(as high as 32%) andTb 680øC)in pillow lavas,in view of the Ti-richcomposition which extendto near the N6el temperatureof hematite(e.g., of primary titanomagnetitein rapidly-cooled,basic igneous Figure 6d). rocks (Fe3-zTizO4, with 0.50 < z < 0.85 [Carmichael, 1967; Carmichaeland Nicholls, 1967; Haggerty, 1976]). Low-StabilityComponent and lhscousRemanence Although only one third of the samplesis completely The majority of samplesfrom La Palma have a low- demagnetizedby 600øC, the majorityof samplesexhibit stabilitymagnetization component of normal polarity, the peakunblocking between 500 o and600øC, consistent with a magnitudeof which (as a percentageof the VDS) varies remanentmagnetization carried primarily by low-Ti magnetite. considerably(Figure 7). Becausethe magnitudeof the Somesamples also have significant unblocking at temperatures low-stabilitycomponent is inverselycorrelated with the VDS, 11,752 GEE ET AL: lVIAGNE•IZATIONOF LA PALMA SEAMOUNTSERIES A B ß ...... /...... Gabbros I I I I I I I I I I I I 0 20 40 60 80 0 20 40 60 80 100 C D 25- • 20- ß ' ' 20" Volcaniclastics • 15 - 15- c• 10 10- 5- 0 20 40 60 $0 100 0 20 40 60 80 Low-StabilityComponent (%) Low-StabilityComponent (%) Fig. 7. Histogramsof the percentageof the low-stabilitycomponent of magnetizationfor differentlithologies. The magnitudeof the low-stabilitycomponent is inverselyrelaled to the vectordifference sum (VDS). The magnitudeof the low-stabilitycomponent weighted by the VDS providesthe bestestimate of the low-stabilitycontribution for each lithology (Table 2). a weightedmean was calculatedto providethe best estimate at 70 mT immediatelyprior to the startof the experiment of the low-stabilitycontribution for each lithology (Table to minimize theseeffects. The remainingsamples were in: 2). Moderate low-stabilitycomponents are commonin an undemagnetizedstate and provide a comparisonbetween severalleucogabbros, although the majorityof gabbroshave NRM and AF-demagnetizedspecimens from the samecore. relatively minor secondaryoverprints. The most substantial Althoughsome samples exhibit nearly linear VRM acquisi- secondarymagnetizations (36%) are in dikes/sills from La tion with respectto log (time), the majorityshow an increase Palma. Pillow lavas typicallyhave lower percentagesof after approximately100 hours (Figure 9a). Such nonlinear secondarymagnetization than do the dikes and sills; however, behaviorhas been attributedto inhomogeneityin the grain larger low-stabilitycomponents are presentin somepillow lavas,particularly those from the morehighly altered portion N of thesequence (Table 2). Pillowfragments and hyaloclastites alsohave small secondarymagnetizations. Principalcomponent directions (maximum angular devia- tion <20ø ) for thelow-stability component of magnetization indicatethat this component was acquired after tilting (Figure 8). Apart from a smallnumber of reverseddirections (~20), which are not shown in the lower hemisphereprojection in Figure 8, the distributionis compatiblewith a recent magnetization. The more than 600 directionsin the normal modeyield a Fisherianmean of 352.3ø, 46.8ø (c•95= 1.8ø) which lies betweenthe posttiltnormal GAD directionand the presentfield directionat La Palma(351ø,40ø). This direction is compatiblewith a VRM of relativelyrecent origin, an interpretationgenerally supported by resultsfrom a laboratory viscousacquisition experiment. The potentialcontribution of viscousremanence in the La Palma sampleswas evaluatedthrough a VRM acquisition experimentin which 80 sampleswere placedin a vertical0.1 mT (4- 3%) field for 5600 hours(~8 months).The samples wereperiodically removed from thefield, measured(typically less than one minute), and returned to the field. Low- field storageand prior demagnetizationare known to affect VRM acquisitionin multidomainsamples [e.g., 7iveyand Fig. 8. Lowerhemisphere equal area projection of thelow-stability componentfrom La Palmasamples. Two percentisodensity contours Johnson,1984]. Therefore,the previouslyAF-demagnetized are shown. The GAD and presentfield directionsat La Palma are samples(61) were again subjectedto AF demagnetization indicatedby the squareand triangle, respectively. GEEET AL: ]VtAGN•IZATIONOF LA PALMASEAMOUNT SERIES 11,753 A 70- OLP747gl (1.53 A/m) • ._ _• 60- ElLP747j 1(0.47 A/m) Jpmow • ß 50- ßLP845cl (0..59A/m)'• gabbro •• ll•-• ..• 40- 2O30] 10 0 I IIII I I I I I Illl I I I I I Illl i I I I I llll I I I I III 10o 101 102 103 ElapsedTime (hours) B ...... • volcaniclastics pillow lavas dikes/sills gabbros I I I I I 20 40 60 80 100 VRM (% of NRM) c •. 80 a= 60 o oO ß-. 40 > 20 (• o Oo CDO O,• O O 20 40 60 80 Low-StabilityComponent (%) Fig. 9. Viscousremanent magnetization (VRM) in La Palmasamples. (a) RepresentativeVRM acquisitioncurves for La Palmasamples. Note the nonlinear acquisition with respect to log (time)particularly in sampleLP845. Undemagnetized sample(LP845dl) acquireslower VRM thanthe AF-demagnetizedsample (LP845cl). Samplefrom the pillow interior (LP747gl) acquiresa muchlarger VRM thanthe correspondingsample from the pillow margin(LP747jl). Magnitude of the VRM acquiredin 5600 hoursgiven in parentheses.Co) Summary of VRM at 5600 hoursin variouslithologies. The width of the line is proportionalto the numberof samples.(c) Comparisonof viscousremanence at 5600 hoursand the percentageof the low-stabilitycomponent of magnetization.The measuredviscous remanence has been corrected for the difference between the NRM and vector difference sum intensities as described in the text. 11,754 GEEET AL: MAGNETIZATIONOFLA PALMASEAMOUNT SERIES size distributionor the presenceof different mineralogies (Figure9b) allows somegeneralizations to be made which [e.g., Dunlop, 1983; Kent and Lowtie, 1977] and clearly are relevantfor the magnetizationof seamounts.Intrusives rendersextrapolation to geologicaltime scalesinappropriate. show a wide range of viscousacquisition from negligible The VRM after 5600 hours, however, gives a minimum to largerthan the correspondingNRM. Leucocraticgabbros estimateof the low-temperatureviscous contribution. In typically acquirea larger VRM than more mafic gabbros, contrastto resultsfrom syntheticsamples [e.g., 7iveyand a relationshipwhich is also supportedby the observation Johnson,1984], the undemagnetizedsamples (e.g., LP845dl of larger VRM in the more felsic portionswithin a single in Figure9a) acquirea smallerVRM thanthe correspondingintrusion. Significantvariability in VRM acquisitionwas AF-demagnetizedsample (LP845cl). This apparentinconsis- also noted within individual dikes/sills althoughno simple tencyprobably results from the presenceof a BrunhesVRM relationshipwas obvious.Although the VRM contributionin in the undemagnetizedsamples from La Palma. A portion mostpillow lavasis small (nearly2/3 have VRM <20%), of the residual VRM will be unaffected on laboratory time higher viscouscomponents do occur. The higher VRM scales,resulting in lower laboratoryVRM for the undemag- in a pillow interior(LP747gl in Figure9a) relativeto the netizedsamples. The two curves,however, should converge margin (LP747jl) suggeststhat the overall contributionof at longertime scalesand we haveused the resultsfrom the VRM from the extrusivesmay be sensitiveto such factors AF-demagnetizedsamples as the bestestimate of the viscous as average pillow size (presumablyreflecting grain size contribution. differences).Hyaloclastite samples acquire a relativelysmall The laboratoryVRM acquiredby various lithologies VRM component,however, the higherintensifies of basalt / ...... / ....., / ...-...... / Oabbros•.._2." s 3.. :'•'e : Group,_ ::.....' ß ..? t Or_oup,!, : ...... : Feederdikes ...... : .Vertical. dikes t''''''''+t''''''''+t,0 ,.. """"+ 0."'•'"...... '. :.'...... :."•...... 0 / Group,1 •.. ' ..> / GroupII •..' • ...) • ___Grou_p_11_..:'.... +/ Sites802-814 '• ...... ' 7 Sites815-834 ...... ,' •Z Sites835-844 ".'.." / sillsinextrusive, ( • 1II I I I I I+ •1 II I I I•11 I+ •1 II I I I I I+ !1oa...... ': o...... ": _ '%...... '.... -%:...... - Fig. 10. Characteristicsite mean remanence directions (geographic coordinates) for intrusivesfrom La Palma. Open (solid)symbols are upper(lower) hemisphere projections. The posttiltGAD directionsat La Palma(4-48 ø) andthe expectedpretilt normal (019 ø, 10ø) and reversed(199 ø, -10ø) are shownby the squareswith associateddispersion inferredfrom paleosecular variation studies. (a) Gabbroshave predominantly posttilt directions, (b) feederdikes for the layeredextrusive series have mostly pretilt directions, and (c) verfcalGroup III dikeshave posttilt magnetizations. (d)-(f) Sill samplingsites have a rangeof pretiltto posttiltdirections which can best be interpretedin termsof tectonictilting and reheating.Triangles indicate keratophyre screens within sill complex. GEE ET AL: MAGNETIZATIONOF LA PALMA SEAMOUNTSERIES 11,755 fragmentsin brecciaunits suggeststhat a coherentsignal may Vertical(Group IID dikeshave posttilt directions in accord be producedby viscousacquisition in the clasts. with the youngages determined isotopically and inferredfrom Comparisonof the VRM at 5600 hoursand the low-stability crosscuttingrelationships (Figure 10c). The normalpolarity componentdetermined from the thermallydemagnetized sister sampleshave a mean (357.6ø, 50.5ø; n = 10, t•a = 5.8ø) sample(both expressedas a percentageof VDS) revealsa which is indistinguishablefrom the posttiltdirection. VRM broad positive correlation (Figure 9c). A one to one data from a singlenormal polarity Group IlI dike suggests correlationbetween the magnitudeof the laboratoryVRM that the normalpolarity cannot be ascribedentirely to viscous and the low-stabilitycomponent is not to be expectedsince magnetization.The reversedpolarity sampleshave poorly the populationof grainsactivated over laboratorytime scales definedcharacteristic magnetization directions as a result of representsonly a portion of the grains which would be substantiallow-stability components (Table 2) and the site activated since the last geomagneticreversal. In addition, meansare often calculatedfrom greatcircle intersections.The a distinct low-stability componentmay be undetectablein greatcircle paths during demagnetization all passthrough the sampleswith a posttiltnormal magnetization. Several samples posttiltreversed direction and are compatiblewith a posttilt exhibit very large secondarymagnetizations but only small magnetization. to moderateviscous remanence acquisition. The maximum The characteristicmagnetization directions of sills(Group blocking temperatureof this componentin these cases is II) include pretilt, posttilt and intermediatedirections. Site incompatible with a simple low-temperatureVRM [e.g., mean directionsfor sills from within the layered extrusive Pullaiahet al., 1975]. SinceVRM acquisitionis enhanced seriesand the detailed sill samplingsite furthestfrom the at highertemperatures [e.g., Dunlop, 1973], the low-stability caldera(sites 802-814; see Figure la) are shownin Figure componentin thesesamples may representa VRM acquired 10d. Sills from within the extrusivesare dominantlyof at elevated temperatures. Both the high erosion rate in normal polarity and have directionsconsistent with a pretilt the Barrancode las Angustiasand the proximityto present magnetization.The remainingsills are almostexclusively of hydrothermalactivity in the caldera suggestthat samples, reversedpolarity and includedirections between the pretilt particularlyfrom the lower part of the SeamountSeries, may andposttilt directions. Sills from sites815-834 (Figure10e) have experiencedelevated temperatures during the Brunhes. are alsodominantly of reversedpolarity, with mostdirections Thesedata are thereforegenerally compatible with a viscous closest to the expected posttilt GAD direction. Two sills origin for the low-stability component,either at ambientor and a keratophyre(altered trachyte) screen at this locality slightly elevatedtemperatures. have shallower, apparentlypretilt reverseddirections and a small numberof posttiltnormal directions are also present. CharacteristicMagnetization Directions The sills (and a single keratophyre)sampled closest to Sixty percent of the gabbro sites from La Palma have the caldera(sites 835-844) have exclusivelymagnetizations reversed directionsconsistent with a posttilt magnetization acquired after tilting (Figure 10f). Although the sills (Figure 10a). Two gabbros(labeled 1,2 in Figure 10a) and a at this site are predominantlyof normal polarity, VRM gabbroscreen (3 in Figure10a) within the sheetedsill complex acquisition experiments eliminate viscous remanence as a have unambiguousposttilt normal polarity directions. These likely explanationfor the observedposttilt direction. two gabbro sites are in the southeastportion of the Caldera Magnetizationsin La Palma extrusiveswhich clearly de Taburiente,the locationof the presentcenter of intrusive postdatetilting are rare, as indicatedby the scarcityof activity. The low laboratoryviscous acquisition in these directionsnear the GAD normal directionin Figure 11a. samplesrules out VRM as an explanationfor their normal The origin of posttiltmagnetizations in theseextrusives can polarity, and suggeststhey are relatively recent (Brunhes) be attributed to a low-temperatureVRM in some cases intrusions.The four shallowdirections in Figure 10a are all (as inferred from large laboratoryVRM) or to chemical from siteswithin the sheetedsill complexor the extrusivesand remagnetizationas suggestedby hematiteTb. The majority are mostcompatible with a pretilt magnetization.Excluding of pillow sites are of normal polarity, but have directions thesefour sites,the meandirections in geographiccoordinates in geographiccoordinates (Figure 11a) which are generally for the reversed(179.5 ø, -50.6ø; n = 9, t•95 = 7.8ø) and to the west of the expected pretilt direction and include normal polarity gabbro sites (006.5ø, 46.5ø' n = 3, t•95 many negative inclinations. After tilt correction these = 8.1ø) are statisticallyindistinguishable from the expected normal polarity directions form a broad swath consistent posttilt directions. with great circle paths from the reversedpretilt direction Feeder dikes (Group I) for the extrusiveshave directions (Figurelib). In contrast,the reversedpolarity pillow lavas generallycompatible with a pretilt magnetization(Figure 10b). yield a tilt-correctedmean (180.6ø, -53.1ø; n = 7, t• = The reversedpolarity directions(mean 196.3ø, -14.6ø' n = 10.1ø),indistinguishable from the expectedpretilt direction. 9, a• = 6.7ø) clusterclose to the expectedpretilt direction This distributionis suggestiveof a reversedremagnetization whereasthe normal polarity directionsare distributedbetween whichpredates tilting of the SeamountSeries. Incompletely the pretilt and posttilt directions and include some shallow separatedpresent field components,on the other hand,cannot northwesterlydirections. The normal polarity directions accountfor the shallow northwesterlydirections. Normal near the posttilt direction may be viscous based on the polarity directionsalong suchremagnetization circles would high percentageof VRM (up to 105% of NRM) acquired be confinedto nearly constantinclinations of about 50ø in over 5600 hours. Contaminationby an unremovedpresent tilt-corrected coordinates. field component,however, cannot explain the anomalous Results from volcaniclastic units in the La Palma Seamount northwesterlydirections. We will demonstratebelow that Seriesgenerally support the propositionof a nonmagnetic these northwesterlydirections are probably the result of a carapaceon seamounts[e.g., Harrison, 1971; Harrison and reversedremagnetization event prior to tilting of the Seamount Ball, 1975]. The majority of pillow fragmentbreccias Series. have clastswith random magnetizationdirections, consistent 11,756 GEE ET AL.: MAGNETIZATIONOF LA PALMA SEAMOUNTSERtF3 o b ß lB ß ß ß 'e..e, 0%ß ß •eoeß o O...ø''"'*•.. •....•o.. Pillows =ß ß ..:: ß .-"• 0m eo.../ß : Geographic"...... ß 0 ...... ß ß øoo o Pillows I I I I I I I I I I I I I I + Tilt corrected o ß ß-...... ß 0 .. ..Q.... 0 "..0 ..Q... ,Q.... c d Fig. 11. Characteristicsite meanremanence directions in geographicand tilt-corrected coordinates for (a andb) pillow lavasand (c and d) volcaniclasticunits from La Palma. Symbolsas in Figure 10. Note the bandof northwesterlyand shallownormal polarity directions after tilt correctioncompared to the well clusteredreversed polarity directions. with an origin by fragmentationand redepositioninferred hyaloclastitecores and the site mean directionssuggests the from clast rounding[Staudigel and Schmincke,1984]. The higher permeabilityhyaloclastites are preferentiallyremag- interstitialhyaloclastite material (ash- or lapilli-sizedglassy netizedeven when the clastsretain their original (random) material)in pillow fragmentbreccias, however, often yields a magnetizationdirections (compare Figures 6e and 60. A consistentdirection which predatestilting. Lapillistonesalso small number of posttilt normal directionsare also evident record a pretilt magnetizationdirection althoughthe overall in Figure 11c. In many cases,this magnetizationcan be contributionof such units to seamountmagnetic anomalies ascribedto reheatingfrom an adjacentGroup lIl dike although is likely to be small. Submarinescoriaceous breccias have moderatelaboratory VRM is observedin somehyaloclastite consistentpretilt directionsin both the fragmentsand the samples. hyaloclastitematerial. These data essentiallyconfirm the DISCUSSION interpretationof Staudigeland Schmincke[1984] that the scoriaceouspillows/breccias are in situ depositswhich have The results from La Palma samples(Table 2) suggest not been displacedsubsequent to their emplacement. that remanenceacquisition processes within seamountsmay In Figures 11c and 11d we plot site means for volcani- be complex. The NRM of La Palma samplesprovides a clastic sites with coherent directions as well as directions of poor estimate of the characteristicmagnetization direction, individual hyaloclastitecores from sites with random clast indicating that preservationof a simple thermoremanent magnetizationdirections. The majority of the sites have magnetization(TRM) acquiredduring seamount construction directionsmost consistentwith a pretilt magnetization.Once is probablyrare. Viscousremanence experiments and the again,the normalpolarity directions in geographiccoordinates magnitude of the low-stability componentdocument the lie to the west and are shallowerthan the expecteddirection, importanceof later VRM. Furthermore,the uniformlyhigh suggestingan incompletelyremoved reversedcomponent of Tb and the presenceof anomalousremanence directions magnetization(Figure 11c). The consistencyof individual in the SeamountSeries may be indicative of significant GEEET AL.: MAGNgr•TION OF LA PAI3tn SEAMOUNTSERIF_.S 11,757 chemicalremagnetization. An understandingof the magnetic the productionof low-T• magnetiteand hematitein extrusives mineralogy and origin of these anomalous directions is to hydrothermalalteration processes. critical to evaluatingto what extentan originalTRM may be A samplefrom the uppermostportion of the SeamountSe- preservedwithin the SeamountSeries. ries illustratesthe dominanceof the low-Ti spinelphase even in relativelyunaltered pillows (Figures 12a-12c). Microprobe MagneticMineralogy analysesindicate a restrictedrange (z = 0.65) for the optically homogeneoustitanomagnetite in this sample and imply a We interpretthe uniformlyhigh Tb in the SeamountSeries Curietemperature of ,..,150øC[O'Reilly, 1984]. Followinga to indicatethe presenceof Ti-poor titanomagnetitein both techniquedescribed by Lowtie [ 1990], selectedsamples from intrusivesand extrusives. The presenceof nearly pure La Palmawere given an isothermalremanence (IRM) in three magnetitein intrusivesmight be expectedsimply from their orthogonaldirections in fields of 2.3, 0.4, 0.12 T and then slow cooling history,Bowever, its presencein the extrusives thermallydemagnetized. Neither the thermaldecay of the is enigmatic.The dominanceof Tl-poormagnetite is strongly NRM (Figure12a) nor of theIRM (Figure12b) in thissample supportedby the petrographicobservations and magnetic showsany indicationof unblockingtemperatures consistent data outlinedbelow. In addition,we use the percentage with the measuredtitanomagnetite composition. The decayof of remanenceremaining at 600øC as an indicatorof the the IRM is dominatedby the 0.12 T fraction,with essentially possible contributionof hematite to the remanence. The no contributionfrom the highestcoercivity axis, suggesting inferred hematite contribution is less certain, but is also the high blocking temperaturesare not the result of hematite. generallysupported by the availabledata. We attributeboth The Curietemperature curve (obtained from a verticalbalance LP750 LP846 d 186[---ø-ø-• ßNRM e-.e_e. ß NRM • o VDS e, o VDS •1o 0 I I I I I I ¾t I o I I I I I 0 200 400 600 o 200 400 600 b ©0.12T ©0.12T 3.5' 7 00.40 T 00.40 T 3.0 v2.30 T •x -- Y2.30 T z.o 1.0 0.5 0.0 0 200 400 600 0 200 400 600 4O 2O 30 lO o o o 200 400 600 o 200 400 600 Temperatt•re (øC) Temperature (øC) Fig. 12. Magneticdata illustrating the interpretationof blockingtemperature data in termsof mineralogy.Thermal decayof (a) NRM andVDS, (b) compositeIRM, and(c) Curietemperature curve for pillowsample from relatively unalteredportion of SeamountSeries indicate dominance of a low coercivity,high blocking temperature component. (d)-(f) Correspondingdata from highly altered pillow lava. A significantcontribution from hematite for the samplein Figures12d-12f is inferredfrom the highblocking temperature of the NRM andthe isothermalremanence. 11,758 GEE ET AL: MAGNETIZATIONOF LA PALMA SEAMOUNTSERIES in an Ar atmosphere)is nearlyreversible with a initial Curie presenceof hematite although small percentages(<5%) temperatureabove 500øC and also shows little evidenceof a may result from noise in the demagnetizationdata. This Ti-rich titanomagnetite(Figure 12c). estimateis not intended as a representationof the volume These observationsare consistentwith the productionof fraction of hematite, however, since hematite may have low-Ti magnetite by hydrothermalalteration even at the significantunblocking below 600øC and because its saturation relativelylow temperaturesdocumented for the upperportion magnetizationis almost two orders of magnitudeless than of the Seamount Series. Rapid cooling of the pillows that of magnetite[O'Reilly, 1984]. effectivelyrules out deutericoxidation as a mechanismfor Two thirdsof pillowsamples have Tb greaterthan 600øC. producingthe pure magnetite. Similarly, the preservation With one exception(LP846 in Figure 12), however,thermal of near equilibrium(z = 0.6) titanomagnetitesin the least demagnetizationof IRM reveals no significanthigh coer- alteredparts of the SeamountSeries indicates that the Ti- civity componentwith blockingtemperatures above 600øC, poor magnetiteis probably not a primary phase in these suggestingthe remanenceis dominatedby low-Ti magnetite. alkalic lavas. Low-temperatureoxidation of titanomagnetite LP846 is from the most altered portion of the Seamount may increaseTb, however, even complete oxidation of Series (Figure l a) and has a northwestwardand up charac- titanium-rich titanomagnetitesis insufficientto produce the teristicmagnetization. Petrographic observations reveal little observedblocking temperature spectrum [e.g., O'Reilly, 1984]. magneticmaterial, and hematite is apparentlythe only phase Furthermore,the isolationof the samecomponents by AF and present.The presenceof hematite(as well as somemagnetite) thermaldemagnetization in many samplesprecludes artificial is confirmedby demagnetizationof the IRM (Figure 12e). generationof the high Tb componentduring laboratory The percentageof the IRM remainingat 600øCis similarto heating. For example, AF demagnetizationof samples that estimatedfrom the decayof the NRM (15-25% for site from site 805 (Figure 5e) recover the samepretilt reversed 846), lending credenceto the methodof estimatinghematite characteristicmagnetization direction, which is isolated at contribution. Hematite is also suggestedby the Curie tem- temperaturesof >500øC. Finally,microprobe analyses from peraturedetermination although the low magnetizationand more than 50 pillow sites documentthe productionof Ti- weight loss correctionmake the resultssomewhat ambiguous. poor titanomagnetite.Titanomagnetite compositions in the The distribution of hematite in pillow lavas from the upper part of the extrusive series are uniformly Ti-rich Seamount Series, as determined by the percentageof re- (0.40 < x < 0.68), whereas samples from more altered manenceremaining at 600øC, is also consistentwith an regionsexhibit a rangeof compositionswhich includes nearly origin during hydrothermalalteration (Figure 13). Data Ti-free magnetite[Pick and Gee, 1991]. Thus, we propose from the analcime zone are limited by the dominanceof that the uniformlyhigh Tb in the La Palmaextrusives are the volcaniclasticunits in this region; however,there is a general result of hydrothermalalteration. increasein hematite contributionwith degree of alteration. Previous authorshave recognizedthat iron titanium ox- The distribution of hematite identified in thin section also ides are sensitiveindicators of the degree of hydrothermal generallyconfirms this result. Hematite in SeamountSeries alteration. For example,Ade-Hall et al. [1971] noted a basaltsoccurs primarily as discreteprismatic grains presum- characteristicgranulation of titanomagnetiteand an increase ably of hydrothermalorigin [Ramdohr,1969]. Martitization in Curie temperatureeven in grains without microscopic of magnetite has been reported from hydrothermallyaltered evidence of alteration. Subsequent studies have con- lithologies[e.g., Hall andFisher, 1987;Fujimoto and Kikawa, firmed the latter observation(e.g., Iceland [Steinthorsson 1989], althoughthe presenceof discretehematite grains is and SveinbjOrnsdottir,1981], Hole 504B [Smithand Baner- apparentlyless common. jee, 1986]). Both experimental[e.g., Johnsonand Merrill, The magneticmineralogy of intrusivesfrom La Palma is 1973; Ryall and Ade-Hall, 1975] and observationalresults morestraightforward. Peak unblocking temperatures (540 o- [Fujimotoand Kikawa, 1989] suggestthat low-Ti magnetite 600øC) in gabbrosare consistentwith the observationof can be produced from titanomaghemitedecomposition at oxyexsolutionof ilmenite from titanomagnetitein all samples. temperaturesas low as 150øC.Furthermore, Hall andFisher Hematite occursas an oxidation productof ilmenite lamellae [1987] have documented the importance of nearly pure as well as a secondaryalteration product in cracksand may secondarymagnetite, which occurs as pseudomorphsafter account for the small percentageof remanenceremaining original titanomagnetite,in the Troodosophiolite. Our re- at 600øC. Thermal demagnetizationof the three axis IRM, sults from La Palma are compatiblewith the breakdownof however, reveals no significant amount of high coercivity titanomaghemite(titanomagnetite) to low-Ti magnetiteduring minerals such as hematite. Thus, the volumetric contribution hydrothermalalteration. Samplesfrom the upperportion of of hematite is apparentlybelow the detectionlimit of this the SeamountSeries have optically homogeneoustitanomag- technique and the remanenceis apparently dominatedby netite (0.4 < x < 0.68) but Tb indicativeof pure magnetite, low-Ti magnetite. suggestingsubmicroscopic subdivision of the original grains. Dikes and sills from La Palmaare alsodominated by pure More highly alteredbasalts often exhibit a rangeof ulv6spinel magnetite, although titanomagnetiteand hematite are also contentswhich includesnearly pure magnetite. present.The dominanceof Tb between500 o and600øC in We suggestthe percentageof the remanenceremaining dikes and sills suggeststhe ubiquitouspresence of Ti-poor at 600øC provides an indicationof the importanceof magnetite.In contrastto the gabbros,ilmenite oxyexsolution hematite. Calibration tests of our Schonstedt thermal from titanomagnetiteis relatively rare in La Palma dikes demagnetizerindicate a significantthermal gradient (~20 ø) at and sills. Rathermost grainsexhibit alterationto spheneor high temperatures;however, the sampleinterior temperatures granulationsimilar to that noted by Ade-Hall et al. [1971] apparentlyequal or exceedthe oven set point for temperatures suggestingthat the magnetizationcarried by low-Ti magnetite of 600øC and above.Thus, the percentremanence remaining resultsfrom hydrothermalalteration [e.g., Hall et al., 1987; at 600øC probably providesa reasonableestimate of the Fujimotoand Kikawa, 1989]. Substantialunblocking at much GEE ET AL: ]VIAG••TION OF LA PALMA SEAMOUNTSERIES 11,759 28.70ø N q- ß 28.69 ø q-q- q- 28.68 ø Percent Remanence at 600øC 28.67 ø ß >10% O 5-10% 0 1 km q- <5% I I I 17.91øW 17.90 ø 17.89 ø 17.88 ø Fig. 13. Hematitecontribution as inferredfrom the remanenceremaining at 600øC in pillow sitesfrom La Palma. Each point is the averageof the inferred hematitecontribution in severalsamples. Percentages < 5% may arise from noise within the demagnetizationdata. Other symbolsas in Figun; 1. lower temperaturesin many dikes/sillsis compatiblewith tilting and/or reheating. We propose that the anomalous the presenceof residualtitanium-rich titanomagnetite. This directionsare the result of overlappingnormal and reversed interpretationis confirmedby demagnetizationof the low and polarity magnetizationswhich can best be attributed to a intermediatecoercivity axes of the IRM in these samples reversed remagnetizationevent which affected much of the at temperaturesof 2500 to 400øC. Even in sampleswith SeamountSeries prior to tilting. a moderate hematite contribution, thermal demagnetization Resultsfrom a leucogabbrowithin the layered extrusives of the IRM reveals no high coercivity phase with blocking illustratehow the inflectionand shallownegative inclinations temperaturesabove 600øC. Hematite occurs most often as may be produced (Figure 14). Thermal demagnetization discreteprismatic grains but is also present as an alteration (Figure 14a) reveals three components,of which the lowest productof ilmenite. stabilitycomponent is viscoussince the sampleacquired 75% VRM in 5600 hours. Indeed,the NRM is indistinguishable Origin of AnomalousRemanence Directions from the presentfield at La Palma (Figure 14c), confirming It is evident from both the demagnetizationcurves in this component is of quite recent origin. The second Figure 6 and the distributionof site meandirections (Figures componenthas a southerlydeclination and positiveinclination, 10 and 11) that a significantportion of the sampleshave and the final two demagnetizationsteps give a direction directionswhich differ from either the expectedpretilt or of 0350, -220 in geographiccoordinates. On an equal posttilt directions at La Palma, lying outside the range area projection (Figure 14c), these three componentsare of directionspredicted from the magnitudeof paleosecular distributedalong two great circles. Removal of the lowest- variation at this latitude. The following generalizationsmay stability component gives directions along an arc away be made concerningthese anomalous directions: (1) normal from the present field. Upon further demagnetization, polarity directions show higher dispersionthan reversed the directions follow a second great circle path from the directions, (2) normal polarity directions (in geographic expectedpretilt reverseddirection towards the pretilt normal coordinates)are biassedtoward the west and include negative direction. The final componentis most compatiblewith a inclinations,and (3) many demagnetizationdiagrams show pretilt normal magnetizationalthough the negativeinclination a characteristicinflection at ~200øC (e.g., Figure6c). The of this componentprobably indicatesinadequate separation distributionof reversedsite means from the sills (Figures of the reversed and normal polarity components. These 10d-10f) is an obvious exception to the first point and is data suggestthat this leucogabbroexperienced an early probably best understoodin terms of successivetectonic remagnetizationduring a reversedpolarity interval, and that 11,760 GEE ET AL: MAGl•'T•ATION OF LA PALMA SEAMOUNTSERIES A W, Up I N 600 s I N c 52o '56O LP845c2 10mT •iD• 35O 200øC 600øC 200øC, _ •.....• ,• O.5 A/m NRM ,' ' t I ,' ', • I I NRM I E, Down I I I I II I W, Up B 3O I I I I I I -I- /I I • I • 560:'c • s-- 7 lOO ; 35oøc-t •-520oC ,' - / , • •mT ,/ - ,/ • LP845c2 / 5cl •,' /, i• LP845cl• 0.5A/m • • NRM E, Down Fig. 14. Originof anomalouscharacteristic remanence direction in leucogabbro.Vector endpoint diagrams for (a) thermal and(b) AF demagnetization.Solid (open) symbols are projectionsof the vectoronto the horizontal(vertical) plane. (c) Equal-areaprojection of demagnetizationdata. Solid (open) symbols are lower (upper) hemisphere projections. Note that vectorsfrom Figure 14alie on two greatcircle paths, an initial arc fromthe present-dayfield direction(PF) anda second greatcircle path which passes near the expectedpretilt directions (stars). AF-demagnetized sample shows a muchmore restrictedgreat circle path. this early magnetizationis, in turn, obscuredby a largeVRM expectedpretilt directions(Figure 15c). The progression which postdatestilting of the SeamountSeries. towards the pretilt normal direction indicatesremoval of a AF demagnetizationof the companionspecimen (Figure reversedpretilt magnetization. Although the characteristic 14b) also showsthree componentsof magnetization;however, magnetizationis well defined,contamination by this reversed the intermediatecomponent has a steeperinclination than in component apparently precludes isolation of the original the thermallydemagnetized specimen and the final component direction. The original magnetizationof the sample shown (015ø, -09ø) is betterdefined. The individualdirections trace in Figure 15b is almostcompletely obscured by secondary a great circle path similar to that defined by the higher magnetizationcomponents. Removal of a posttilt normal temperaturedirections from the companionsample (Figure component(similar to the GAD normaldirection) is indicated 14c). Since these two specimens are from the same by the great circle path defined by the low-temperature core, the very large differencesbetween the vector endpoint demagnetizationsteps (Figure 15c). At temperaturesabove diagramsare probably a result of the variable stabilityof the 400øC the directionsare compatiblewith a great circle componentsto AF and thermal demagnetization.Thus, the path similar to that definedby sampleLP860al. The final type of demagnetizationbehavior shown in Figure 6c can be direction(328 ø, 18ø) is the resultantof overlappingreversed producedby overlappingcomponents of oppositepolarity. and normalpretilt magnetizationswhich cannotbe effectively Demagnetizationdata from a feeder dike on the northwest separated.These results suggest that the westerlydeclinations flank of the Barranco de las Angustiasillustrate the origin of the normalpolarity feederdikes represent an originalpretilt of thesewesterly, shallow directions (Figure 15). Thermal magnetizationvariably contaminated by both a pretilt reversed demagnetizationof sample LP860al (Figure 15a) reveals and a posttilt normal polarity remagnetization.The reversed three components:an initial westwardand down component, polarity samplesare less dispersedsince both the original an intermediatedirection suggestive of reversedpolarity, and and remagnetizationdirections are of reversedpolarity and a characteristicmagnetization which approachesthe expected predatetilting of the SeamountSeries. pretilt normal direction. The directionsat temperaturesteps The mineralogicalchanges accompanying hydrothermal al- >250øC lie alonga greatcircle path whichpasses near the terationmay provide a mechanismfor producingthe aberrant GEE ET AL: MAGNETIZATIONOF LA PALMA SEAMOUNTSERIFS 11,761 A W• 0.2 Aim - N 580øC c 450øC 160ø(2. 200øC 400øC I NRM ß NRM E, Down I NRM I I I W, Up B / S N / / -400 / 560 ! ß LP860b2 ß ) LP860al •"":•LP860b2 200øC 0.4 Aim E,Down 12NRM Fig. 15. Originof shallow,northwesterly characteristic remanence directions in dike. (a)Co)Vector endpoint diagrams. Solid(open) symbols are projections of thevector onto the horizontal (vertical) plane. (c) Equalarea projection of demag- netizationdata. Filled (open)symbols are lower (upper) hemisphere projections. Note thatthe individual demagnetization stepsfor bothsamples are compatible with theinitial removalof a presentday normalmagnetization, followed by a great circlepath from the reversedpretilt to the normalpretilt direction (stars). directionsobserved in muchof the SeamountSeries. Spurious chemical remanenceparallel to the field during alteration. remanencedirections resulting from completelyoverlapping The inherentsimilarity of reversedand normallymagnetized blocking temperature spectra have been noted previously grainsproduced in this mannermight accountfor the difficulty [e.g.,Dinares-Turell and McClellan& 1991]. Suchsignificant in separationby either AF or thermaldemagnetization. overlap in blocking temperaturescan result from chemical overprinting[e.g., McClelland-Brown, 1982], and the presence Evolutionof theMagnetization of theLa Palma of coexisting magnetite and hematite may account for the Seamount Series overlapin somesamples. Tb of <600øC andlow stabilityto AF demagnetizationin other samples,however, argue against The construction and later alteration of the Seamount Se- the presenceof hematite as a universal explanationfor the ries over severalpolarity intervalshas resultedin a complex discordantdirections. Productionof nearly pure magnetite magnetizationhistory. Volcanicactivity at La Palma began during hydrothermalalteration may provide an alternative with the eruption of lavas during the Pliocene. The domi- mechanism. Smith and Banerjee [1986] reported Ti-free nantly normal polarity of the extrusiveseries, togetherwith secondarymagnetite as a precipitantfrom hydrothermalfluids biostratigraphicages of 2.9-4.0 Ma, suggestsformation pri- and also inferred the presenceof submicroscopicmagnetites marily within the Gaussnormal polarity chron (2.60-3.55 Ma in apparentlyhomogeneous hydrothermally altered titanomag- [Candeand Kent, 1992]). Pillow lavas acquireda TRM, netitesfrom DSDP Hole 504B. The latter phasehas recently presumablycarried by Ti-rich titanomagnetites,as suggested been imagedwith transmissionelectron microscopy, where it by the preservationof ulv6spinel-richtitanomagnetites and occurs in associationwith sphene,ilmenite, ulv6spinel and nearly univectorialpretilt magnetizationdirections in samples kassite[Shau et aL, 1991]. We speculatethat subdivision from the least altered part of the section. Feeder dikes of primary titanomagnetite(titanomaghemite), or possibly for the extrusiveshave pretilt directionsof both normal and precipitationof secondarymagnetite, may result in similar reversedpolarity. Crosscuttingrelationships indicate that sill populationsof fine-grainedmagnetite which may acquire a emplacement began early in the history of the seamount 11,762 GEEET AL.: MAGNETIZATION OF LA PALMASEAMOUNT SERIES [Staudigelet al., 1986]. Sills within the extrusiveshave and may have contributedto the tilting as well as to the unambiguousprefllt magnetizations,confirming that at least remagneflzaflonof earlier sills. someof the Group II intrusiveswere originallyhorizontal. Emplacementof GroupIII dikes andprobably some gabbro As the edifice shoaled, volcaniclasflc material became intrusionsoccurred significantly after the formation,alteration more prevalent althoughpillow lavas also occur high in the and tilflng of the Seamount Series. Field relationships, extrusive secflon. Amoeboidal pillows apparentlyacquired radiometricages and postflit magnetizationdirecflons are all a normal polarity TRM as suggestedby the presenceof compatiblewith the interpretationof the GroupIll dikesas a reddened,quenched clast margins and remanencedirections radial dike swarm feedingthe overlyingCoberta lavas. The nearthe expectedprefllt direction.The magnetizationof most posttilt normal polarity directionsand proximity to present clastsin pillow fragmentbreccias, however, was randomized day hydrothermalactivity in the calderasuggest that two of during redeposition. Sill intrusion continued during the the gabbrossampled represent Brunhes intrusions. Finally, shoaling stage: approximately75% of sills are younger the majority of samples from the SeamountSeries have than the feederdikes [Staudigelet al., 1986]. Althoughthe acquireda low-stability magnetizationcomponent, similar to small gabbroic intrusionssampled within the sill complex the presentfield directionat La Palma, and presumablyof and extrusives have complex magnetizations,the shallow viscousorigin. characterisflcdirections are most compatiblewith a prefllt remanence.Thus, the seamountmust have had a significant IMPLICATIONS FOR SEAMOUNT PALEOPOLES intrusivecore prior to any resolvableradial tilting. The resultspresented in the previous secflonscast doubt Emplacementof intrusives also provided the source of on the fidelity of many seamountlithologies in preserving heat for hydrothermalalteration of the Seamount Series. a record of the geomagneticfield at the time of intrusion The mineralogicaleffects of this hydrothermalalteration are or deposition. In particular, the NRM rarely provides a most apparentin the extrusives,where low-Ti magnetitewas good estimate of the characteristicmagnetization direction, produced and hematite precipitation occurred in the most often being biassedtoward the presentfield by a VRM. The altered lavas. Much of the magnetizationin extrusivesmay therefore best be termed a chemical remanence. Anomalous magnetizationdirections of extrusives and dikes and sills emplacedearly in the history of the seamountare generally remanencedirections in extrusives,as well asother lithologies, consistent with a magnetization acquired prior to tilting may alsobe attributedto hydrothermalalteration. Analysis of (Table 2); however, the distribution of the characteristic demagnetizationdata (Figures 14 and 15) indicatesthat these magnetization suggeststhe presence of an incompletely anomalousdirections are mostcompatible with an unresolved removed secondary component. Thus, even directions mixture of pretilt magnetizations. The distributionof site determinedfrom stepwisedemagnetization may not represent mean directionsfrom extrusives(Figure 11) also illustrates the originalmagnetization. We suggestthis inability to recover that the dominantalteration (remagneflzaflon) event occurred the original magnetizationdirection stems from a complex during a reversed polarity interval. Extrusiveswith an history of alteration which occurred in a different polarity original pretilt reversed magnetizationhave less dispersed statethan prevalentduring the original magnetization.Since directions, whereas the normal polarity sites are biassed the exposuresin La Palma are uniquein their completeness, toward northwesterly, shallow direcflons. The dominant we focus on processesaffecting the magneflzationwhich are reversed polarity of sills and gabbros provides additional likely to apply to other seamounts. supportfor this reversedremagneflzaflon event. Althoughthe preciseage relationshipsof the intrusionsare unknown,they may record either the short subchronswithin the Gaussor Role of Multiple Polarities may have intrudedduring the Matuyamareversed chron (but The presenceof large volumes of material with different prior to the oldestCoberta lavas at 1.61 Ma). polarities constitutesone of the more difficult problemsin Radial tilting of the SeamountSeries apparently postdates seamountanomaly modeling. Lumb et al. [1973] have hydrothermalalteration of the extrusivesand is presumably shownthat layers of alternatepolarity can lead to erroneous the result of inflation throughintrusive activity [Staudigel paleopolepositions, particularly if the two polarity statesare et al., 1986]. The emplacementof gabbroswith posttilt not antipodal. Multiple polarities might be detectedby the magnetizationdirections probably contributed to the tilting of spatialcorrelation of residualsin least squaresmodels [e.g., the SeamountSeries. In addition, sills have magnetization Parker, 1988], and the seminormminimization technique directionssimilar to both the prefllt and posttilt directions, of Par'keret al. [1987] should also detectlarge blocks of with postflit directions most prevalent in sills nearest the oppositepolarity. The more recent ideal body modeling, centerof intrusiveactivity (Figure 10) as might be expected however, suggeststhat even complex anomaliesmay be if the sills were in part responsiblefor the tilting. The compatible with a uniform direction, provided variationsin magnetizationof keratophyrescreens within the sill complex, intensityare not constrained[Parker, 1991]. however, suggeststhat reheatingand alterationmay account The duration of volcanism on ocean islands and seamounts for someof the posttiltdirections. The keratophyresrepresent impliesthat Cenozoicseamounts constructed entirely within a metamorphosedtrachyflc extrusives from the early stages singlepolarity interval will be rare. Radiometricages suggest of seamountformation [Staudigel and Schmincke,1984] and a durationof 0.5 to 1.0 rrry. for shield-buildingvolcanism thus the original magnetizationshould predate tilting. A on oceanicislands [e.g., Jacksonet al., 1972; McDougall keratophyrescreen from the intermediatesill site (Figure10e) and Schmincke, 1976]. The total duration of volcanism apparentlymaintains the originalpretilt direction,whereas a on Hawaiian Islandsmay span 5 m.y. [e.g., Clagueand keratophyrefrom the innermostsill site (Figure 100 hasbeen Dalrymple, 1987] and a similarduration has been documented reheatedafter tilting. "Sills" nearer the calderacenter may, for a Pacificseamount [Pringle et al., 1991]. Suchextended in part, have been injected along dippingplanes of weakness volcanism,together with the high reversal frequencyin the GEE ET AD: MAGNETIZATIONOF LA PALMA SEAMOUNTSERIES 11,763 Cenozoic [Lowrie and Kent, 1983], suggeststhat multiple Importanceof Viscousand InducedMagnetization polarityis likely to be a com_n•nfeature in moderateto large In the context of determiningpaleopoles from seamount seamounts. magnetic anomalies, probably the most general bias arises Resultsfrom the La PalmaSeamount Series provide support from viscousand inducedmagnetization (IM) contributions. for the presenceof multiple polaritiesin many seamounts. The percentageof the low-stability componentof magne- Extrusivesof the SeamountSeries are dominantlyof normal tization in various lithologies (Figure 7) can be used to polarity with several reversedsites in the lower part of the estimatethe bias from VRM. The low-stabilitycomponent section.Together with the biostratigraphicallydetermined age was acquiredafter tilting (Figure 8). The similarity to the of 2.9-4.0 Ma for the extrusiveseries [Staudigel, 1981], these presentfield direction at La Palma, as well as evidencefor results suggestthat most of the extrusiveserupted during alterationof the SeamountSeries prior to tilting discussed the Gauss normal polarity chron. Although the precise age above, suggeststhat the most plausible explanation for this relationshipsof the intrusionsare unknown,the presenceof componentis a VRM. The generalcorrelation between the reversedsills and gabbrosrequires that the SeamountSeries laboratory VRM and the percentageof the low-stability spana minimumof two reversedand one normalinterval, and componentalso supportsthis interpretation.The percentage it may encompassmuch of the Pliocene. In additionto the of the low-stability magnetizationdoes not coincideperfectly complexity introducedby simple superpositionof reversed with the magnitudeof VRM, since some of the secondary and normal polarity material, seamountconstruction spanning magnetizationsare probably of thermal origin and in other more than one polarity interval is apparentlya prerequisite casesVRM componentsmay not be obviousfrom the demag- for the generationof anomalousdirections from hydrothermal netizationdata. On balance,however, the parameterprobably alteration. gives a reasonableestimate of the contributionof VRM. The magneticproperties of various lithologies(Table 2) Importanceof Reheatingand HydrothermalAlteration allow an estimate of the combined contribution of VRM The emplacementof intrusiveswithin a seamountmay be and IM to seamount anomalies. In order to estimate this associatedwith either local remagnetizationby conductive contribution,we use a model lithological distributionfor a heating or hydrothermal alteration. Conductive heating large Hawaiian-type seamountand a second model based from dike emplacement is an inefficient mechanism for on Jasper,an intermediate-size(570 kma) seamountin the remagnetizingcountry rock. Samplestaken near dike contacts easternPacific. On the basisof seismicand gravitydata [Hill often exhibit low-stabilitycomponents of magnetizationwith and Zucca, 1987] the model Hawaiian seamount includes unblocking.temperatures higher than expectedfor a low- 14% gabbros,29% dikes/sillsand roughlyequal proportions temperatureVRM [e.g., Pullaiah et al., 1975]. The extent of pillow lavas (22%) and volcaniclastics(24%), and 11% of the reheatingindicated by thesepartial thermal remanence subaerialflows [Gee et al., 1989]. In applying the data componentsis limited to lengthscales approximately equal to from Table 2 to this lithologicalmodel we have made the the width of the dike. Buchanand Schwarz[1987] reported conservativeassumptions tha• subaerial flows have negligible a similar result in country rock adjacentto a 38 m dike in VRM and that the pillow lavas have intensitiessimilar to the the NorthwestTerritories. Complete directional reorientation leastaltered samples and a VRM equalto that determinedfor is confined to a smaller fraction of the dike width [e.g., all pillows. The resultinginduced (17% [Gee et al., 1989]) Ade-Hall et al., 1972]. The averagedike width (<1 m) in andviscous (10%) magnetizationssuggest that approximately the SeamountSeries suggests that the volume of countryrock one quarter of the total magnetizationof large seamounts remagnetizedby this mechanism is probably insignificant might have a presentfield direction. except where the dike/sill density is high. The similarity Geophysicalstudies of JasperSeamount [Hildebrand et al, of magnetizationdirections in adjacentsills and the posttilt 1989; Hammeret al., 1991] suggestthat smallerseamounts magnetizationof a keratophyrescreen from the innermostsill have porportionatelymore extmsive material. We have samplinglocality (Figure 10f) suggestthat the basalpart of constructeda model lithological distributionbased on the the sill swarm has experiencedthis type of reheating. observed seismic velocities in a vertical section through In contrast,hydrothermal alteration associated with intru- Jasper[Hildebrand et al., 1989]and the flexuraldepression of sive activity can be a very effective mechanismfor remag- the crustto a level of ~2 below the seafloor(P. T. C. Harmher netization[e.g., Rice et al., 1980]. If seamountconstruction et al., JasperSeamount structure: Seafloor seismic refraction occurswithin a singlepolarity interval the directionalchanges tomography,submitted to Journal of GeophysicalResearch, associatedwith suchintrusive activity would be insignificant, 1993). The bulk of the seamount has seismic velocities althoughthe mineralogyand bulk magneticproperties would <4.5 krrds which we infer to consistprimarily of pillows still be affected. The results from La Palma, however, indi- (65%), volcaniclasticmaterial (30%) with a small amount catethat intrusiveactivity spanning different polarity intervals of feeder dikes (5%). The presenceof higher velocities may producemagnetization directions which are dramatically (>4.5 krrds) beneath the flank of the seamountis thought differentfrom any expectedgeomagnetic direction (Figures 6 to reflectthe presenceof a dike complexwith ~40% dikes and 11). The similarityof the La Palma hydrothermalsystem [Hildebrand et al., 1989; P. T. C. Harmher et al., submitted to that on Reunion[Demange et al., 1989] suggeststhat such mansuscript,1993]. We have used the velocity contours alterationis probablya com_n•nfeature in seamountsand as an estimate of the percentageof dikes, increasingfrom islandswith a significantproportion of intrusives.Although 40% at 4.5 knds to 70% at 6.5 knds. The resulting hydrothermalalteration is generallyconfined to the early model is compatible with the geophysicalconstraints and historyof the seamount,Rice et al. [1980] havedocumented comprises53% pillows and 18% volcaniclasticmaterial, with a similarremagnetization process in the BermudaSeamount less than one third intrusives. The data in Table 2 suggest where the intrusive activity postdatesseamount construction a similar contributionof viscous (16%) and induced (11%) by more than50 rn•y. magnetizationas inferredfrom the Hawaiian type model. 11,764 GEE ET AL.: 1VIAGNE•IZATIONOF LA PALMA SEAMOUNTSERIES Although the contributionof VRM and IM deducedfrom of the paleolatitudeerror is dependenton the polarity of this study of the magnetic properties of seamount rocks the seamount. This asymmetry arises from the reduced is gratifyingly similar to the 25% inferred from modeling magnetizationin reversedpolarity seamountsdue to a normal of seamountmagnetic anomalies [Hildebrand and Staudigel, polarity overprint,resulting in a proportionatelygreater angu- 1986; Bryan and Gordon, 1991], the estimatesgiven above lar deviationfrom the samemagnitude overprint. Paleopoles are subjectto severaluncertainties. Our estimateof VRM from reversedseamounts are far-sidedby an amount,-,50% and IM does not account for the reduced contribution due greaterthan the correspondingnear-sided bias of a normally to the greater depth to the intrusive sources,although the magnetizedseamount. Second, the paleolatitudeerror is also increased temperature at depth may partially offset this dependenton the original paleolatitudeof the seamount.For effect. In addition,significantly higher Koenigsbergerratios example,a reversedseamount formed at 20øS and now at (ratio of remanent to induced magnetization,Q) have also 5øN has a paleopolewhich is far-sidedby 14ø if IM and been reported from dredged seamountsamples [e.g., Gee VRM accountfor 25% of the seamountmagnetic anomaly. et al., 1988], indicating a smaller IM contributionfrom A reversedseamount constructed at 20øN andsubject to the the outermostpart of seamounts. Seamountsamples from samelatitudinal transport (25 ø) wouldyield a pole position drilling, however, yield intensitiesand Q ratios similar to only 7.5ø far-sided.Therefore, the bias introducedby vis- thosefrom La Palma samples[e.g., Kono, 1980; Marshall, cousand inducedmagnetization is significantfor older Pacific 1978], suggestingthat the very high Q ratios from dredged seamounts.Fixed hotspot models suggest ,-,40 ø of northward materialprobably are not representativeof a large volume of motion for the Pacific plate since the late Cretaceous[e.g. the seamountedifice. The IM impliedby the Q ratio provides Engebretsonet al., 1985]. We note that the majority of a mini.mumestimate of the total magnetizationparallel to the Cretaceouspaleopoles used to constructthe Pacificapparent ambientfield, as the remanentcontribution may also include polar wander path are from seamountspresently located at a viscouscomponent. The overall contributionof VRM and low latitudes. Many of these seamountswere presumably IM may be slightlylower than 25%, but the percentageof the constructedat low southernlatitudes and therefore might low-stabilitycomponent in extrusives(15%-22%) suggestsa havepaleopoles in errorby up to 10ø to 16ø for normaland minimumcontribution of 15% evenin seamountsdominantly reversedpolarity seamounts,respectively. composedof pillows and volcaniclasticmaterial. A combined viscous and induced contribution of 15% to CONCLUSIONS 25% representsa significantpotential error in the determina- The La Palma SeamountSeries constitutes a singlesample tion of paleolatitudes(Figure 16). The bias in paleolatitude suiterepresehting theinterior of a seamount,butin the absence in this figure is basedon the simplestcase of coplanar of significantdrilling penetrationinto seamountsthese results vectorsin which a percentageof the original magnetization provide the best opportunity to evaluate the assumptions is reorientedparallel to the ambientfield. In additionto the inherent in determiningpaleopoles from seamountmagnetic far-sided or near-sidedpole positionsgenerated by present anomalies.The fundamentalassumption of suchmodeling is field overprints[e.g. Sager, 1987], severalobservations may that the anomaly is dominatedby a reinanentmagnetization be made from this simplifiedmodel. First, the magnitude which preservesan accuraterecord of the geomagneticfield -20 i I I i I I I 0 5 10 15 20 25 30 35 40 NorthwardTransport (degrees) Fig. 16. Potentialbias in seamountpaleopoles resulting from viscousand inducedmagnetization. Paleolatitude error applicableto Pacificplate is shownfor normaland reversedpolarity seamounts constructed at 20øN (areasbetween dashedlines) and 20 ø S (shadedareas) as a functionof northwardtransport. The rangeof the potentialpaleolatitude error in eachof the four casescorresponds to a combinedviscous and inducedcontribution ranging from 15% to 25%. Note that bias is dependentnot only on latitudinaldisplacement but is alsoa functionof seamountpolarity and paleolatitude. Paleolatitudeerror calculatedfrom simplecoplanar vector model with percentageof originalremanence reoriented into ambient field direction. GEE ET AL: MAGNET•ATION OF LA PALMA SEAMOUNTSERIES 11,765 during seamountconstruction. Despite the high Q ratiosoften seamountsconstructed within a single polarity interval may reportedfrom dredged/drilledseamount samples [e.g., Kono, providethe bestopportunity for determiningpalcopoles. 1980; Gee et al., 1988], a significantIM is expectedfrom The effectsof hydrothermalalteration noted in La Palma lithologieswithin the seamountinterior [Gee et al., 1989]. samples suggestthat caution should be excercisedin inter- The NRM directionsof La Palma samplesalso give a poor preting resultsfrom similarly altered lavas. Hydrothermal representationof the characteristicmagnetization direction. alteration may offer an alternative explanation for similar The distributionof NRM directionsis stronglybiassed toward anomalous remanence directions often encountered in drilled the presentfield direction, rather than reflectingthe original samplesfrom the oceancrust and attributedto tectonictilting. pretilt directionas requiredby anomalymodeling. The mean The potentialseverity of this problemis limited, however, direction(352.30 , 46.80) of the low-stabilitycomponent of by the lower temperaturesprevalent in the extrusiveportion magnetization,together with laboratory viscous acquisition of the crust. For example,the geothermalgradient at DSDP experiments, suggeststhis component can reasonably be Hole 504B resultsin temperatures• 100øC only at the base ascribed to a recent VRM. The magnitude of the low- of the extrusivelayer and deeper[Becker et al., 1989]. The stability componentand the average Q ratios of La Palma bulk of the extrusiveswould presumablynot be subjectto the samples imply a combined IM and VRM contributionof alteration temperaturessufficient to cause the mineralogical 15%-25%, with the lesservalue probablyapplicable for changesnoted in extrusivesfrom La Palma except in cases smaller seamountsdominated by extrusive material. The where sedimentcover is especiallythick. bias introduced by viscous and induced magnetizationis dependentnot only on the amount of latitudinal translation, Acknowledgments.We thankJ. L. G. Afonso,C. Alba, A. P. Mar- but is also a function of the polarity and paleolatitudeof tinez, U. Tams and the Weinrichfamily for logisticalsupport and seamount formation. L. Stokkingand E. Barg for assistingin the field wore We thank M. Collins and J. Collins for help in measuringsamples. We are The inducedand viscous contribution to seamountmagneti- gratefulto R. Enkin for providinga greatcircle analysis program. We zation derivedfrom our studysuggests several approaches for are indebtedto P. Schiffmanfor his contributionsto unravelingthe evaluatingthe reliability of individual seamountpaleopoles alterationhistory of the La Palmasamples as well as providinguse of as well as for possiblyimproving the Pacific apparentpolar an electronmicroprobe. Special thanks are extendedto D. V. Kent and E. L. Wintererfor many helpfuldiscussions. M. T•vey, J. Hall, and wanderpath. The relative contributionof IM and VRM is C. Raymondprovided thorough reviews which greatly improved the dependenton lithology, with greater contributionsassociated text. The Keck Foundationkindly providedsupport for the Scripps with intrusive material. As suggestedpreviously [Gee et PalcomagneticLaboratory. This projecthas been supportedby the al., 1989], the averagedensity might be used to identify National GeographicSociety and NSF grant OCE90-00182. J. Gee was partiallysupported by an NSF graduatefellowship. seamountswith a high proportionof intrusivesand therefore a potentiallylarger bias due to viscousand inducedmagneti- zation. In addition,comparison of solutionsfrom the various R•NCES modeling techniquesprovides an indicationof the error for an individual palcopole determination. For example, the Abdel-Monem,A., N. D. Watkins,and P. W. Gast, Potassium-argon ages,volcanic stratigraphy, and geomagnetic polarity history of the ideal body, seminormand least-squaressolutions for Darwin CanaryIslands: Tenerife, La Palma,and Hiefro, Ant J. $ci., 272, Guyot, a large seamountformed in the early Cretaceous, 805-825, 1972. differ by more than 200 [Parker, 1991]. Someestimate of Ade-Hall, J. M., P. Dagley, R. L. Wilson, A. Evans, A. Riding, the potentialpaleopole error may alsobe derivedby modeling P. J. Smith,R. Skelhorne,and T. Sloan, A palaeomagneticstudy of the Mull regionaldyke swarm, Geophys.J. 'R. Astron. $oc., the seamountanomaly after removal of an estimatedVRM 27, 517-545, 1972. and IM contributionparallel to the ambientfield. The over- Ade-Hall,J. M., H. C. Palmer,and T. P. Hubbard,The magneticand all contributionof 15%-25% may be appropriatefor many opaquepetrological response of basaltsto regionalhydrothermal seamounts,although additional geophysical constraints on the alteration, Geophys.J. R. Astron. $oc., 24, 137-174, 1971. Afonso,A., Geologicalsketch and historicvolcanoes in La Palma, internal structure of the seamount are obviously desirable Canaryislands, Estud. Geol., TeneguiaVol., 7-13, 1974. to constructa geologically reasonablemodel. Finally, we Barone,A.M., andW. B. E Bryan, Singleplume model for asyn- suggestthat palcopolesmodified in this way may provide a chronousformation of the Lamont Seamountsand adjacentEast more realisticestimate of plate motion. The asymmetryof the Pacific Rise terrains, J. Geophys. Res., 95, 10,801-10,827, 1990. palcolatitudebias shouldbe consideredwhen combiningsuch Becker,K., et al., Drilling deepinto youngoceanic crust, Hole 504B, correctedpalcopoles from seamountsof different polarity. CostaRica Rift, Rev. Geophys.,27, 79-102, 1989. Our data from La Palma also indicate that preservation Bird, D. K., P. Schiffman, W. Elders, and A. E. Williams, Calc- of the original magnetizationdirection may be affected by silicatemineralizafion in activegeothermal systems, Econ. Geol., hydrothermalalteration, which results in the productionof 79, 671-695, 1984. Bryan, P., and R. G. Gordon, Determiningthe averagecontribu- 1ow-Ti magnetiteand an increasingcontribution of hematite tion of viscousoverprints to the characteristicmagnetization of Pa- to the remanence. In addition to significant changes cific seamounts(abstrac0, Eos Trans.A GU, 72(44) Fall Meeting in the bulk magnetic properties, magnetization directions suppt., 125, 1991. which deviate significantlyfrom any reasonablegeomagnetic Buchan, K. L., and E. J. Schwarz, Determination of the maximum direction at La Palma can be attributedto hydrothermal temperatureprofile acrossdyke contactsusing remanentmagne- tizationand its applications,in Mafic Dyke Swarms,edited by alteration in a different polarity than prevalent during the H. C. Halls andW. E Fahrig, Geol. Assoc.Can. Spec.Pap., 34, original magnetization.Seamount construction over multiple 221-227, 1987. polarity intervals,when coupledwith hydrothermalalteration, Cannichael, I. S. E., The iron-titanium oxides of salic volcanic rocks may therefore result in additional complicationsbesides and their associatedferromagnesian silicates, Contrib. Mineral. Petrol., 14, 36-64, 1967. the simple superpositionof rocks of different polarity. Cannichael,I. S. E., andJ. Nicholls,Iron-titanium oxides and oxygen These results suggest that small seamounts, with little fugacifiesin volcanicrocks, J. Geophys.Res., 72, 4665-4687, intrusivematerial and possiblyless hydrothermal activity, and 1967. 11,766 GEEET AL.: MAGNETIZATION OF LA PALMASEAMOUNT SERIES Clague,D. A., andG. B. Dalrymple,The Hawaiian-Emperor volcanic Hill, D. P., and J. J. Zucca, Geophysicalconstraints on the structure chain,part I, Geologicevolution, in VolcanisminHawaii, U.S. of Kilauea and Mauna Loa volcanoesand someimplications for Geol. Surv.Prof. Pap., 1350, 5-54, 1987. seismomagmaticprocesses, U.S. Geol. Surv.Prof. Pap., 1350, Cox, A., Latitudedependence of the angulardispersion of the ge- 903-917, 1987. omagneticfield, Geophys.J. R. Astron. Soc., 20, 253-269, Jackson,E. D., E. A. Silver,and G. B. Dalrymple,Hawaiian-Emperor 1970. Chain and its relation to Cenozoic circum-Pacifictectonics, Geol. Demange,J., P. Chovelon,and P. Puvilland, Geothermalmodel of Soc. Am. Bull., 83, 601-618, 1972. the SalazieCirque (Reunion Island): Volcanic and structural impli- Johnson,H. P., and R. T. Merrill, Low-temperatureoxidation of a cations, J. Volcanol.Geotherrn. Res., 36, 153-176, 1989. titanomagnetiteand the implicationsfor paleomagnetism,J. Geo- Dinares-Turell,J., and E. McClelland, A cautionarytale for palaeo- phys. Res., 78, 4938-4949, 1973. magnetists:A spuriousapparent single component remanence due Kent, D. V., and W. Lowrie, VRM studiesin leg 37 igneousrocks, to overlapof blocking-temperaturespectra of two components, Initial Rep. Deep SeaDrill. Proj., 37, 525-529, 1977. Geophys.Res. Lett., 18, 1297-1300, 1991. Kirschvink,J. L., The least-squaresline and planeand the analysis Dunlop,D. J., Theoryof themagnetic viscosity of lunarand terrestrial of paleomagneticdata, Geophys.J. R.Astron. Soc., 62, 699-718, rocks, Rev. Geophys.,11, 855-901, 1973. 1980. Dunlop,D. J., Viscousmagnetization of 0.04-100•m magnetites, Kono,M., Magneticproperties of DSDP leg 55 basalts,Initial Rep. Geophys.J. R. Astron.Soc., 74, 667-687, 1983. Deep Sea Drill Proj., 55, 723-736, 1980. Engebretson,D.C., A. Cox, andR. G. Gordon, Relativemotions Liou, J. G., Analcimeequilibria, Lithos, 4, 389-402, 1971. betweenoceanic and continentalplates in the PacificBasin, Spec. Lonsdale,P., and R. Batiza, Hyaloclastiteand lava flowson young Pap. Geol.Soc. Am., 206, 59 pp., 1985. seamountsexamined with a submersible, Geol. Soc. Am. Bull., Fisher,R. V., and H.-U. Schmincke, PyroclasticRocks, 472 pp., 91, 545-554, 1980. Springer-Verlag,New York, 1984. Lowtie, W., Identificationof ferromagneticminerals in a rock by Fornail, D. J., A. Malahoff, and B.C. Heezen, Volcanic structure coercivityand unblockingtemperature properties, Geophys.Res. of the crestof PunaRidge, Hawaii: Geophysicalimplications of Lett., 17, 159-162, 1990. submarinevolcanic terrain, Geol. Soc. Am. Bull., 89, 605-616, Lowrie, W., and D. V. Kent, Magnetic reversalfrequency since the 1978. Late Cretaceous, Earth Planet. Sci. Lett., 62, 305-313, 1983. Lumb, J. T., M.P. Hochstein,and D. J. Woodward, Interpreta- Francheteau,J., C. G. A. Harrison,J. G. $clater, and M. L. Richards, tion of magneticmeasurements in the Cook Islands, south-west Magnetizationof Pacificseamounts: A preliminarypolar curve for Pacific Ocean, in The WesternPacific: Island Arcs, Marginal the northeasternPacific, J. Geophys.Res., 75, 2035-2061, 1970. Seasand Geochemistry,edited by P. J. Coleman,pp. 79-101, Fujimoto,K., andE. Kikawa, Alterationof titanomagnetitesand its Crane, Russak, New York, 1973. relatedmagnetic properties in the Noya geothermalarea, Central Marshall,M., The magneticproperties of someDSDP basaltsfrom Kyushu,Japan, J. Geomagn.Geoelectr., 41, 39-64, 1989. the North Pacificand inferences for Pacificplate tectonics, J. Geo- Gee, J., L. Tauxe,J. A. Hildebrand,H. Staudigel,and P. Lonsdale, phys. Res., 83, 289-308, 1978. Nonuniformmagnetization of JasperSeamount, J. Geophys.Res., McClelland-Brown,E., Discriminationof TRM and CRM by 93, 12,159-12,175, 1988. blocking-temperaturespectrum analysis, Phys. Earth Planet. In- Gee, J., H. Staudigel,and L. Tauxe, Contributionof inducedmag- ter., 30, 405-414, 1982. netizationto magnetizationof seamounts,Nature, 342, 170-173, 1989. McDougall, I., and H.-U. $chmincke, Geochronologyof Gran Ca- naria, Canary Islands:Age of shieldbuilding volcanism and other Gordon,R. G., Testfor biasin paleomagneticallydetermined pale- magmaticphases, Bull. Volcanol., 40, 1-21, 1976. olatitudesfrom Pacificplate Deep Sea Drilling Projectsediments, McFadden,P. L., and M. W. McElhinny, A physicalmodel for J. Geophys.Res., 95, 8397-8404, 1990. palaeosecularvariation, Geophys.J. R. Astron. Soc., 78, 809- Gordon,R. G., and A. Cox, Calculatingpalaeomagnetic poles for 830, 1984. oceanicplates, Geophys. J. R. Astron.Soc., 63, 619-640, 1980. McFadden,P. L., and M. W. McElhinny, The combinedanalysis of Haggetty,$. E., Oxidationof opaquemineral oxides in basalts,in remagnetizationcircles and direct observations in palaeomagnetism, Oxide Minerals, Mineral. Soc. Am. Short CourseNotes, edited by Earth Planet. Sci. Lett., 87, 161-172, 1988. D. Rumble,III, pp. 101-140, MineralogicalSociety of America, McNutt, M., Nonuniform magnetizationof seamounts:A least Washington,D.C., 1976. squaresapproach, J. Geophys.Res., 91, 3686-3700, 1986. Hall, J. M., and B. E. Fisher, The characteristicsand significanceof Middlemost,E. A. K., Evolutionof La Palma,Canary Archipelago, secondarymagnetite in a profilethrough the dike component of the Contrib. Mineral. Petrol., 28, 272-279, 1972. Troodos,Cyprus, ophiolite, Can. J. EarthSci., 24, 2141-2159, O'Reilly, W., Rockand Mineral Magnetism, 220 pp., Chapmanand 1987. Hall, New York, 1984. Hall, J. M., C. Walls, M. Williamson,and B.-X. Wang, Depth trends Palmasson,G., S. Arnorson, I. B. Freidleifsson, H. Kristmannsdottir, in magneticproperties in an areaof prolongedcold seawater draw- K. Saemundsson,V. Stefansson,B. Steingrisson,and J. Tomas- downin uppermostTroodos-type oceanic crust, Can. J. EarthSci., son, The Iceland crust: evidence from drillhole data on structure 24, 941-952, 1987. and processes,in Deep Drilling Resultsin the Atlantic Ocean: Hammer, P. T. C., J. A. Hildebrand, and R. L. Parker, Gravity in- Ocean Crust, Maurice Ewing Ser., vol. 2, edited by M. Tal- versionusing seminorm minimization: Density modeling of Jasper wani, C. G. A. Harrison,and D. E. Hayes, pp. 43-65, AGU, Seamount,Geophysics, 56, 68-79, 1991. Washington,D.C., 1979. Harrison,C. G. A., A seamountwith a nonmagnetictop, Geophysics, Parker,R. L., A statisticaltheory of seamountmagnetism, J. Geo- 36, 349-357, 1971. phys. Res., 93, 3105-3115, 1988. Harrison,C. G. A., and M. M. Ball, Geophysicalobservations on an Parker,R. L., A theoryof ideal bodiesfor seamountmagnetism, J. exposedseamount in the Afar Depression,Bull. Volcanol.,38, Geophys.Res., 96, 16,101-16,112, 1991. 26-43, 1975. Parker,R. L., L. Shure,and J. A. Hildebrand, The applicationof Hayes,D. E., and P. D. Rabinowitz, Mesozoicmagnetic lineations inversetheory to seamountmagnetism, Rev. Geophys.,25, 17- andthe magneticquiet zone off NorthWest Africa, EarthPlanet. 40, 1987. Sci. Lett., 28, 105-115, 1975. Pick, T. S., and J. S. Gee, Magnetic petrologyof hydrothermally Hernandez-Pacheco,A., and S. Fernandez-Santin,The submarinefor- alteredpillow basaltsfrom the La PalmaSeamount Series (abstract), mationsof the Calderade Taburientein La Palma(Canary Islands) Eos Trans.AGU, 72(44) Fall Meetingsuppl., 144, 1991. andtheir metasomaticprocesses, IAVCEI Symposium,Santiago de Press,W. H., B. P. Flaunery,S. A. Teukolsky,and W. T. Vetterling, Chile, Bull Volcanol.Spec. VoL, 1975. NumericalRecipes: The Art of ScientificComputing, 702 pp., Hildebrand,J. A., and H. Staudigel,Seamount magnetic polarity and CambridgeUniversity Press, New York, 1986. Cretaceousvolcanism of thePacific Basin, Geology,14, 456-458, Pringle, M. S., H. Staudigel,and J. Gee, JasperSeamount: Seven 1986. million yearsof volcanism, Geology, 19, 364-368, 1991. Hildebrand, J. A., L. M. Dorman, P. T. C. Hammer, A. E. Schreiner, Pullaiah,G., E. Irving, K. L. Buchan,and D. J. Dunlop, Magnetiza- and B. D. Cornuelle, Seismictomography of JasperSeamount, tion changescaused by burial and uplift, Earth Planet. Sci. Lett., Geophys.Res. Lett., 16, 1355-1358, 1989. 28, 133-143, 1975. GEE ET AL.: MAGNET•ATION OF LA PALMA SEAMOUNT SERIES 11,767 Ramdohr,P., The Ore Minerals and their Intergrowths, 1207 pp., Steiger, R. H., and E. Jager, Subcommissionon geochronology, Pergamon,New York, 1969. Earth Planet. Sci. Lett., 36, 359-362, 1977. Rice, P. D., J. M. Hall, and N. D. Opdyke, Deep Drill 1972: A Steinthorsson,S., and A. E. Sveinbj6rnsdottir,Opaque minerals in paleomagneticstudy of the BermudaSeamount, Can. J. Earth geothermalwell no. 7, Krafia, northernIceland, J. Volcanol. Sci., 17, 232-243, 1980. Geotherm. Res., 10, 245-261, 1981. Ryall, P. J. C., and J. M. Hall, Laboratoryalteration of titanomag- Txvey,M., and H. P. Johnson, The characterizationof viscousre- netitesin submarinepillow lavas, Can. J. Earth Sci., 16, 496-504, manentmagnetization in large and small magnetiteparticles, J. 1978. Geophys.Res., 89, 543-552, 1984. Sager,W. W., Late Eoceneand Maastdchtianpaleomagnetic poles Vacquier,V., A machinemethod for computingthe magnitudeand for the Pacific plate: Implicationsfor the validity of seamountpa- the directionof magnetizationof a uniformlymagnetized body from leomagneticdata, Tectonophysics,144, 301-314, 1987. its shapeand a magneticsurvey, in Proceedingsof BenedumEarth Schmincke,H. -U., The geologyof the CanaryIslands, in Biogeog- MagnetismSymposium, pp. 123-137, University of Pittsburgh raphy and Ecologyin the CanaryIslands, edited by G. Kunkel, Press, Pittsburgh,Pa., 1962. pp. 67-184, W. Junk, The Hague, 1976. Zijderveld, J. D. A., A.C. demagnetizationof rocks: analysisof re- Shau, Y.-H., D. R. Peacor,and E. J. Essene, Single-domain-sized sults,in Methodsin Palaeomagnetism,edited by D. W. Collinson, magnetite(Uv0) after titanomagnetitein hydrothermallyaltered K. M. Creer, and S. K. Runcorn,pp. 254-286, Elsevier, New oceanicbasalt (abstract), Eos Trans.AGU, 72(17) SpringMeeting York, 1967. suppl., 141, 1991. Smith, G. M., and S. K. Banerjee, Magnetic structureof the upper kilometer of the marine crust at Deep Sea Drilling Project hole Y. Gallet, Institutde Physiquedu Globe de Paris,4 Place Jussieu, 504B, easternPacific Ocean, J. Geophys. Res., 91, 10,337- 75252 Paris Cedex 05, France. 10,354, 1986. J. Gee, Lamont-DohertyEarth Observatory,Palisades, New York, 10964. Staudigel,H., Der basaleKomplex von La Palma, submarinevulka- nischeProzesse, Petrologie, Geochemie und sekundareProzesse im T. Pick andL. Tauxe,Scripps Institution of Oceanography,La Jolla, CA 92093-0220. herausgehobenen,submafinen Teil einer ozeanischenInsel, Ph.D. thesis, 357 pp., Ruhr Univ., Bochum,Germany, 1981. H. Staudigel,Vrije UniversiteitAmsterdam, Fac. der Aardweten- Staudigel,H., and H.-U. Schmincke,The PlioceneSeamount Series schappen,De Boelelaan1085, 1081 HV Amsterdam,Netherlands. of LaPalma/CanaryIslands, J. Geophys.Res., 89, 11,195-11,215, 1984. Staudigel,H., G. Feraud,and G. Giannedni, The historyof intrusive (ReceivedJuly 10, 1992; activity on the islandof La Palma (CanaryIslands), J. Volcanol. revised March 11, 1993; Geotherm. Res., 27, 299-322, 1986. acceptedApril 9, 1993.)