Petrogenesis of Ultramafic Xenoliths from the 1800 Kaupulehu Flow ...

0022-3530/92 $3.00Petrogenesis of Ultramafic Xenoliths from the 1800Kaupulehu Flow, Hualalai Volcano, Hawaiiby CHEN-HONG CHEN*, DEAN C. PRESNALL, AND ROBERT J. STERNGeosciences Program, The University of Texas at Dallas, P.O. Box 830688, Richardson,Texas 75083(Received 5 August 1987; revised typescript accepted 20 June 1991)ABSTRACTThe 1800 Kaupulehu flow on Hualalai Volcano, Hawaii, contains abundant xenoliths of dunite,wehrlite, and olivine clinopyroxenite with minor gabbro, troctolite, anorthosite, and websterite. Thepetrography and mineral compositions of 41 dunite, wehrlite, and olivine clinopyTOxenite xenolithshave been studied, and clinopyroxene separates from eight of these have been analyzed for Ba, K, Rb,Sr, rare earth elements, 87 Sr/* 6 Sr, and 143 Nd/ 14 *Nd. Temperatures of equilibration obtained byolivine-spinel and pyroxene geothermometry range from 1000 to 1200 °C. Mineralogical datacombined with published fluid inclusion data indicate depths of origin in the range of 8-30 km.The rarity of orthopyroxene, the presence of Fe-rich olivine (Fo g , _ 89 ) and clinopyroxene (Fs 3 _, 2 Xand the occurrence of high TiO 2 in spinel (0-9-2-8 wt.%) and clinopyroxene (0-35—1-33 wt.%) allindicate that the xenoliths are cumulates, not residues from partial fusion. The separated clinopyroxeneshave 87 Sr/* 6 Sr (0-70348-0-70367) and '"Nd/'^Nd (O-51293-O-51299) values that are differentfrom Sr and Nd isotope ratios of Pacific abyssal basalts (< 0-7032 and > 0-5130, respectively). Also,clinopyroxenes and spinels in the xenoliths have generally higher TiO 2 contents (>O-35 and>0-91 wt.%, respectively) than their counterparts in abyssal cumulates (

164 CHEN-HONG CHEN ET AL.olivine clinopyroxenite, with gabbro, troctolite, anorthosite, websterite, and two-pyroxenegabbro being much less common. Bohrson & Clague (1988) studied the petrography andmineral compositions of some of the ultramafic xenoliths containing exsolved pyroxenes.Our study complements theirs and concentrates on the abundant dunites, wehrlites, andohvine clinopyroxenites. With one exception, sample 68KAP1, the pyroxenes in the samplesof these rocks that we have studied do not show exsolution.In his pioneering study, White (1966) interpreted ultramafic xenoliths in Hawaiian alkalicbasalts to be cumulates from basaltic magma, but he did not specify whether the magmawas of Hawaiian or mid-ocean ridge origin. Dunite, wehrlite, and olivine clinopyroxenitexenoliths from Hawaiian alkalic basalts, including those from the 1800 Kaupulehu flow,were interpreted by Jackson (1968) as being partly cumulates and partly texturally deformedrestites from partial melting of upper-mantle peridotite. On the basis of a wide andrelatively iron-rich range of olivine compositions (Fo g3 _ 90 ), Sen & Presnall (1980, 1986)argued that dunite xenoliths from Koolau Volcano, Oahu, are samples of cumulates fromKoolau tholeiites even though the dunite shows deformation textures. Jackson et al. (1981)reached a similar conclusion for deformed dunites from the 1800 Kaupulehu flow.However,the similarity of 3 He/*He values (8 x atmospheric) between Hualalai ultramafic xenolithsand mid-ocean ridge basalts (MORBs) led Kaneoka & Takaoka (1978, 1980), Kyser &Rison (1982), and Rison & Craig (1983) to conclude that these xenoliths are fragments ofMORB-type oceanic crust. Dunite and wehrlite xenoliths from Mauna Kea have beeninterpreted by Atwill & Garcia (1985) as deformed cumulates from Hawaiian alkalicmagmas. On the basis of spinel and clinopyroxene compositions, Clague (1988) concludedthat dunite and wehrlite xenoliths from Loihi Seamount were crystallized from Hawaiianalkalic magmas.To clarify the origin of the Kaupulehu ultramafic xenoliths, we have carried out a detailedmineralogical, isotopic, and geochemical study. The petrography and mineral compositionsof 46 dunite, wehrlite, and olivine clinopyroxenite xenoliths have been studied. Eightxenoliths were selected for determination of trace element concentrations (REE, K, Rb, Ba,and Sr) and isotopic (Sr and Nd) compositions of separated clinopyroxenes.SAMPLE LOCATIONThe investigated samples were collected from the xenolith beds in olivine-bearing alkalicbasalt of the 1800 Kaupulehu flow on the north slope of Hualalai Volcano (Fig. 1). Twooutcrops of the uppermost layer of xenoliths were sampled, one located ~ 120 m northeastof the Hue Hue telephone repeater station and the other ~ 120 m southeast of this station(Jackson & Clague, 1981). Samples with the prefixes A, B, and X were collected by the firstauthor. The remainder were supplied by D. A. Clague and B. Melson from the collection ofthe late E. D. Jackson, now housed at the Smithsonian Institution, Washington, DC.PETROGRAPHYThe xenoliths typically are subrounded, but some are angular. They have a meandiameter of ~ 7 cm (Jackson et a/., 1981). Most consist of varying proportions of olivine,spinel, and clinopyroxene; a few also contain small amounts of plagioclase and orthopyroxene(Table 1). Two main types of xenoliths have been identified, (1) dunite and (2) wehrliteand olivine clinopyroxenite (15-85% clinopyroxene). Several xenoliths are composites ofdunite and wehrlite. Two generations of olivine crystals are evident even in hand samples,one comprising large dark crystals and the other made up of smaller and lighter re-

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 165\/rI)\I/1PACIFIC//OCEANM*y11)[1 'I I\ \ \,\ \ < *^\\\K Kailua\ \\\"7Vi\\\\\\\iiiV )\\\i/"•i\\/rAM ^^/As)v v \\ \ v» \//(1^^,^^^ ^.4000-'"oo\\ 0 5\ \\ \\N\/ —^ HUALALAI VOLCANO\v\km6, 6156° 155°45"1800 KAUPULEHU FLOWXENOLITH BEDSFio. 1. Map of the island of Hawaii showing the 1800 Kaupulehu allcalic basaltic flow of Hualalai volcano (afterMoore et al^ 1987) and the site where the xenoliths for this study were collected.TABLE 1Modal percentages of minerals in Kaupulehu xenoliths*Sample no.ol cpx spX2X5AfX6X7X9§X10Xll§X12X14X21§X22§AlA3{AA§BlB3B4f§Dunitc98 pi999995981009790919792979692989192P122P17724238275PP31P2331411P123

166 CHEN-HONG CHEN ET AL.TABLE 1 (Continued)Sample no.olcpxHP65KAP16 (114370/17)166KAP1 (114743/1)168KAPl(114S76/l)t170K.AP2 (114970/2)75KAP5 (115032/1)75KAP6 (115032/2)75KAP7 (115032/3)**75KAP8 (115032/4)75KAP9(U5032/5)t97989396949896979811434—211WeMite and olivine clinopyroxeniteXIX3"X4X5Af§**X5§X8X17X20X23Q§B4f§6511463 (114385/13)§**65115153 (114386/29)t |)65KAP17 (114370/18)66KAP2(114743/2)§66KAP3 (114743/3)68KAP1 (114876/l)tT*71264080606675618130404021577215332970601540332539177060557943288565213122222—p—21p2——P————* Except where indicated, percentages are basedon counts of at least 7000 points.t Composite xenolith.X Present, but

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 167Schwarzmann, 1977). The most common texture is characterized by kink-banded orstrained olivine porphyroclasts (1-5 mm) in a matrix of finer-grained (0-2-0-4 mm) recrystallized,undeformed olivine grains, some of which have rounded to polygonal shapesand well-defined triple junctions. All samples except 75KAP6 contain small amounts ofclinopyroxene (augite and subcalcic augite, Table 1). It is pale green in thin section andtypically occurs as small (~ 30 /zm) interstitial grains. Clinopyroxene oikocrysts (up to1 mm across) with olivine inclusions occur in one sample (XI2), whereas clinopyroxenecrystals in samples AA, B4, and X4 are blocky and could possibly be cumulus (T. N. Irvine,pers. comm.). Opaque spinel grains of various shapes and sizes (20-400 /un) are ubiquitous.Some occur as small interstitial grains in the fine-grained matrix, some are euhedral orsubhedral grains included in olivine, and a few are enclosed in clinopyroxene. The euhedralhabit of spinel crystals in olivine indicates that they crystallized relatively early. Orthopyroxenewas found only in composite xenolith 68KAP1 and plagioclase was found in onedunite xenolith (75KAP7). Both minerals are interstitial. Inclusions of CO 2 and silicateglass are common along healed microfractures in both clinopyroxene and olivine (Roedder,1965; Kirby & Green, 1980).Wehrlite and olivine clinopyroxeniteThree types of clinopyroxene-rich xenoliths were identified by Jackson et al. (1981):(1) those that have deformed kink-banded olivine and clinopyroxene in a matrix ofrecrystallized olivine; (2) those that have kink-banded olivine but undeformed clinopyroxene;and (3) those with undeformed cumulus olivine and intercumulus clinopyroxene. InTable 1, samples XI, X3, X4, X5, X17, X20, X23, Q, B4, 65115153, and 68KAP1 are texturetype 1. Samples X5A, X8, and 66KAP2 are texture type 2. Cumulus grain shapes andpostcumulus poikilitic textures are common in types 2 and 3. Deformed olivine grains aregenerally smaller (0-5-1-0 mm) than the porphyroclastic olivine in the dunite xenoliths.Modal mineral proportions for wehrlite and clinopyroxenite xenoliths are given in Table 1.As in the dunite xenoliths, trains of CO 2 and glass inclusions are common in theclinopyroxene and deformed olivine. Rare small interstitial grains of orthopyroxene (only insample 68KAP1), spinel, and plagioclase also occur.MINERAL COMPOSITIONSOlivineOlivine compositions in the dunite xenoliths range from Fo 81 4 to Fo 894 (Fig. 2 andTable 2). There is no compositional difference between the large deformed and smallrecrystallized grains in the same sample. Olivine compositions in wehrlite and olivineclinopyroxenite xenoliths are generally more iron rich (Fo 81 _ 84 . 7 ) than those in the dunites,but the two compositional ranges overlap (Fig. 2 and Tables 2 and 3). NiO contents(0-05-0-28 wt.%) in olivines from wehrlites and clinopyroxenites also overlap those fromdunites (0-05-0-50 wt.%). Forsterite content and NiO content of olivine covary.SpinelExcept in samples X8, X9, X10, and Al, the chemical compositions of spinel grains withina single thin section are identical within analytical uncertainty (Tables 4 and 5). Despite thevariations of spinel composition in some samples, a reasonably good positive correlation

168 CHEN-HONG CHEN ET AL.10nKaupulthu duniti xtnolithsKaupulehu wthrlite & clinopyroxtnitt xtnolitht20-io 15 -cnn-1 10-u.Hawaiian tholti5 -I--rm n rr-i• 1 •10^5 - Hawaiian alkalic basaltsn n90 80 70 60 50 40nMole % Fo in olivineFIG. 2 Histogram of olivine compositions in Hawaiian rocks. Data for Hawaiian tholeutic and alkalic basalts arefrom Lofgren et al. (1981). Xenolith data are from Tables 2 and 3.exists between Mg/(Mg -I- Fe 2 + ) in spinel and the forsterite content of coexisting olivine(Tables 2-5), which indicates equilibrium between spinel and coexisting olivine (Sigurdsson,1977). This correlation is similar to that observed for Mid-Atlantic Ridge basalts (Sigurdsson& Schilling, 1976), for Hawaiian basalts (Evans & Wright, 1972; Clague et al, 1980a;Lofgren et al., 1981), and in the Stillwater complex (Jackson, 1969). The only zoned spinel isin sample X2 (Table 5), which shows increasing Mg/(Mg + Fe 2 + ) from core to rim.ClinopyroxeneWithin a single thin section, compositions of clinopyroxene grains in the Hualalaiultramafic xenoliths (Tables 6 and 7) are identical within analytical uncertainty. Figure 3

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 169oo 'r- c^ ON

TABLE 2 (Continued)SiOjFeO*MgONiOCaOTotalSiFeMgNiCaTotal% Fo• TotalB339-4613-7246-740-30—100-220-9850-2861-7390006—3-01585-9Fe as FeO.B439-84151145150-210-26100-570-9960-3161-682000400073O0484-270KAP240-3912-8146-920-28009100-490-9990-2651-730000600023O0186-775KAP539-5712-4747-710-3501810O28098202591-765O0070O05301887-275KAP64074109748-5102801499-6475KAP739-78107547-61036—10O5075KAP839-9313-5246-2903301810O25Cations per 4 oxygens099802251-771000600043O0288-7099602251-776O007—3O04870099402821-7180O07O005300685-975KAP939-99141546-19022—10O55099402941-7120004—300685-36511515339-5615-72451802802110O950989O3291-683O0060006301183-765KAP16405712-4748O9023014101-50099202551-753O00500043O0887-366KAP139-6215-2845O8——99-98099503211-688——3O0584068KAP139-4314-7745-63——99-830990O3101-709——3O1084-6n3Ctiiz aozom Z fn-it--

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 171Olrs r- m »n666 -.oo\O *H -H Sc* m Q O\ •I^oom ON X *o NO Iwp — »• omoi* v£ ^ ~

TABLE 3 (Continued)SiOjFeO*MgONiOCaOTotalSiFeMgNiCaTotal% Fo• TotalB439-35151945-900-100-19100-7309830-3171-70900020005301784-3Fe as FeO.66KAP339-7116O944-43—100-230-9980-3381-665—3-00283166KAP239-8514-7045-68—100-230-99603071-701—300484-765KAP1739-4216-3644-20—93-98Cations per099503451-664—3-00582-86511515339-9315-0045-70O17—10O804 oxygens099403121-6960003—3-00683-9651146339-3016-364419—99-85099403461-666—3-00682-868KAP139-7114-7145-57—99-99099503081-702—3-00584-775KAP939-8015-0845-33—10O21099603161-692—3-00484-375KAP1039-7216-4544-07016O2010O60099803461-650O00300053-00282-7oXmZxo zo o Xtnz

X10U)0051-3026-7325-8331-7901813-30Oil99-2916-5216-9210O94001603131O1206-5593-9954-54400496-365O03731-998XW008119'34-8818-3329-300151519—99-1215-5115-34100670024027612-6714-4653-5973-95300396-975—32O00Xll0131-3021-9337-5923-3001914-690029915102414O810O17004303198-4219-6802-5113-83700527131000732O01H50oXmZofHXVi•v50OHXmC"0C rmXCr oTABLE 4Spinel compositions in dunite xenoliths (wt.%)X2(av.)X2(core)X2(rim)X6X7X9U)X9(2)SiOjTiO 2A1 2 O 3Cr,OjFeO*MnOMgOCaOTotalFe 2 O 3FeONew totalSiTiAlCrFe 3 +Fe 2 +MnMgCaTotal0-291-6715-2838-6333-750-259-84O0599-7213-98211710112O1000-4316181104793-6126075(M)725-032001932-0010351-6414-9738-3032-830331110—99-5215191916101-04O12004216021103313-9005-46800965-644—320010291 6414-8038-7935-700268-2400899-8013-5823-4810116O10104286O64106573-5536-823O0724-268002931-9950091-7233-4019-3329-9601514-46—991114-7316-7010O580028O40312-2604-7593-4534-34800406-709—32O00Oil1-3930 7025-30281501514-36_99-5313-3216-1610O86Cations per 48 oxygens00350329111576-29531554-255O0406-736—32O0200710827-7734-4319-5001215-9000498-917-4512-8099-6600230257103738-6251-7763-39200327-508001331-9990O91-3631-7529-2619-6801516-91—99-208-2212-28100020028031611-57771551-913317700397-795—32O00

174 CHEN-HONG CHEN ET AL.> O t- Tj- ON OO —o ON C-J — \c OvfN - «I ON VTVIO4 Ifo r- r- m r- »n ON6 d\ r> n TT o *i-- n r< ^- m » oo *orTOr- v> o^ o^ oo*—

11027-5528-7129-32—12-9799-6512-9417-671009402651O4087-27331214-7376-195—31-999ULT30>n Xrrt enorHc/i•npoO2Hn>c•vrmcrOTABLE 4 (Continued)70KAP2 75KAP7 65115153 68KAP1036167250227-2732-4001613-4810O3616-4317-61102O0011504019-4406-8993-9594-71300446-429—32O00. .1 _r ii :p , () () y p; () p . i yThe designations (1) and (2) for samples X9, XIO, and Al indicate single spot analyses of two extreme compositions.75KAP5 75KAP6 75KAP8 75KAP9 . _ l . i p g SiOj032021033032039033TiOj2-211102-091-921-351-49A1 2 O 3Cr 2 O 3FeO»MnOM-gOCaOTotalFe 2 O,FeONew total 24-69 3113 26-27 01514-24 Oil9912 1117 16-21 10O23 28-73 29-59 23-29 01415-51 01798-74 1079 13-58 99-82 18-84 39-47 23-81 01814-73 01799-62 1035 14-49 10O65 2303 32-38 27-65 01713-86 01099-43 12-51 16-39 10O68 19-67 36-27 28-92 0181302 01899-98 13-46 16-81 101-33 27-78 29-97 27-88 01613-86 00899-55 12-28 16-83 10O78 SiTiAlCrFe 3 +Fe 2 +MnMgCaTotal0104 0536 9-396 7-944 2-715 4-377 O041 6-851 O039 32O03 0067 0261 10711 7-397 2-569 3-591 O037 7-309 0057 31-999 0108 0516 7-288 1O240 2-557 3-977 0051 7-204 0060 32001 Cations per 48 oxygensO104 0469 8-816 8-312 3059 4-451 0047 6-708 0035 32-001 O128 0335 7-635 9-440 3-335 4-628 0051 6-388 0064 32004 0105 0357 10427 7O40 2-943 4-481 0043 6-576 O027 31-999 • Total Fe as FeO.including the zoned crystal.

176 CHEN-HONG CHEN ET AL.TABLE 5Spinel compositions in wehrlite and olivine clinopyroxenite xenoliths (wt.%)X8(1)X8(2)XI7X2375KAP10SiOjTiOjA1 2 O 3Cr 2 O 3FeO*MnOMgOCaOTotalFe 2 O,FeONew totalSiTiAlCTFe 3 +Fe 2 +MnMgCaTotal00811521-4032-7331-69017121900599-4615-4317-80101O0002702858-3318-5443-8364-91500405-999001731-9940151-4132-4919-5431-48015141899-4016-2016-90101-02Cations per 48 oxygens0047033111-9524-8203-8054-41100486-595—320090391O027-57250832-73O1812-9900599-9916-3717-99101-620124O240103596-3193-9284-79600406-169001731-992027O9143-42105328-47013160799-8014-3515-55101-230O800203151192-45931913-84300327075—320020331-84271522-9034-5001712-940159*9817-6818-59101-75O1050441102155-7774-2474-96100456-155005131-997• Total Fe as FeO.The designations (1) and (2) for sample X8 indicate single spot analyses of twoextreme compositions.shows the range of compositions. These clinopyroxenes are higher in A1 2 O 3 and Na 2 O(2-5-7-0 and 0-3-10 wt.%, respectively) than those from Koolau dunite xenoliths (21—2-7and

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 177sOO *"H ^^^ ^^^ ^^^ OO ^^ flO ^^J-66666666» i3S:i —• OO d O P*" Oi SO,66? - •• -• -•-I — m v^t Op O OO O, - . _ . , . • so ^ cn rnOO ^ O O O OO ^^ OO O O ON^H Q Q Q Q Q Q Q Q ^f fT)— OsONOsmoor--r- ooooo vi ^r-r-vi^CN — ONOO moo —

TABLE 6 (Continued)SiOjTiO,A1 2 O 3Cr 2 O 3FcO*MgOCaONa 2 OTotalSiA1 IVAl"TiCrMgFe* +a NaTotalmgEnFsWoAA50350625-730656-1116-4619-7905210O231-8430156O091O0170-0190898OT870776O0374-02682-848-310041-7Bl50721-334-521-033-8116-30211009399-741-86001400055003700300891011708290066402588-448-56-4451B349-250985-740925-5316-6419-7608899-701-817018400660027O0270915017107810063405184-34909-141-8B451-250543-750915-6416-602O52O5099-711-887011300500015002609110174O810003640228404819-242-770KAP250880864170944-7316-922O6007399-83Cations per 61-8670132O048002400270926014508100052403186-449-27-743175KAP55O810744-471074-4716-7621-25058100-15oxygens1-85901410052002000310914013708330041402887048-57-344-275KAP652-620883021054-3217-60206006110O701-9060094003500240030095001310799O043401287-95O57O42-575KAP751-400844-261014-8317-2319-8508410O261-87401260057O0230029093601470775O059402686-45O47-941-775KAP951090674-790965-5416-5819-301O099-931-87101290078O0180028O9050170075800714O2884-249-49-341-36511515351130754-261045-5815-8121-71055100021-869013200520021O03008610171O8500039402583-545-89-145-268KAP152O00353 910905-7817-4018-9906910O831-89801010067001000260947017707430049401884-35079-539-8nmz X z oowz fa*,• Total Fe as FeO.

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 179n ( N Oi >o-H in ffi in6 \b \b os 6• r- r- r-

sTABLE 7 (Continued)SiO 3TiO 2AljO 3Cr 2 O 3FeO*MgOCaONa 3 OTotalSiM nA1 VITiCrMgFe 1 *CaNaTotalmgEnFsWo* TotalB451 290573-460955-6017-262O49049100-111-8820118O0320016O028094401720806O0354-03384-649-18-941-9Fe as FeO.66KAP35O880504-400395-9317-5819-3205699-561-871012900620014001109640182076100404-0348415O59-639-966KAP25O80O5

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 181Fsdunitewehrllta& clinopyroxenite30Fio. 3. Clinopyroxene compositions for Hualalai lavas and Kaupulehu xenoliths plotted in theCaSiO 3 (wo)-MgSiO3(en)-FeSiO3(fs) triangle (mole percent). Clinopyroxene samples analyzed for trace elementsare labeled. Hualalai tholeiitic and alkalic basalt fields are from Bobrson & Clague (1988).TRACE ELEMENT AND ISOTOPIC COMPOSITIONSOF CLINOPYROXENESIsotopic analyses for Sr and Nd and the concentrations of K, Rb, Sr, Ba, and rare earthelements (REE), in the clinopyroxene separates from four dunite, two wehrlite, and twoolivine clinopyroxenite samples are given in Table 9 (see Appendices A and B for analyticalmethods). All the separates were acid leached after careful hand-picking to remove possiblesurface contaminants. The investigated samples were chosen to cover the entire range ofMg/Fe for clinopyroxene in the Hualalai xenoliths (Tables 6 and 7). Except for dunitesample X14, REE concentrations of clinopyroxene in the dunite xenoliths are higher thanthose in clinopyroxene from wehrlite and clinopyroxenite xenoliths (Table 9 and Fig. 5). Alleight REE patterns are convex upward (Fig. 5). Except for Sr, the concentrations of othertrace elements (K, Rb, Ba), are very low (Table 9). The clinopyroxene separates have143 Nd/ 14 *Nd ranging from 0-51293 to 0-51299 and 87 Sr/ 86 Sr ranging from 0-70348 to0-70367.EQUILIBRATION TEMPERATURESBecause of the very limited occurrence of orthopyroxene, geothermometry based on anassumption of coexisting pyroxenes must rely entirely on clinopyroxene compositions, andtherefore it yields only minimum temperatures. The method of Lindsley & Andersen (1983)gives values from 1000 to 1220 °C. On average, they are ~55°C higher than valuesdetermined by the method of Kretz (1982) and about 130°C higher than those calculated bythe procedure of Mysen (1976). One exceptional xenolith, dunite sample X2, has a relativelylow Lindsley-Anderson temperature of 915 °C. Sample 68KAP1, the only sample withcoexisting clinopyroxene and orthopyroxene, has a Lindsley-Andersen temperature of

182 CHEN-HONG CHEN ET AL.Abyssal Iherzolite& harzburgite20r-10-•1Kaupulehu/1'//xenolithsV0.2 0.4 0.6 0.8 1.0 1.2 1.4Wt. % TiO 2 in clinopyroxeneFIG. 4. Histogram of wt% TiO 2 in dinopyroxenes. Hualalai xenolitb data are from Tables 6 and 7. Abyssalperidotite data are from Clarke & Loubat (1977), Arai & Fujii (1979), Hamlyn & Bonatti (1980), Hebert et at.(1983), Dick & Bullen (1984), and Dick (1989).1185°C and a Kretz temperature of 1166°C, in good accord with values for the otherultramafic xenoliths in which orthopyroxene is absentFour olivine-spinel thermometers were tried. They give temperatures that are, onaverage, higher than those given by the Lindsley-Andersen pyroxene thermometer by 13 °C

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 183TABLE 8Orthopyroxene compositions (wt.%) in composite xenolith 68KAP1WehrlileDuniteSiOjTiOjA1 2 O 3Cr 2 O,FeO*MgOCaONa 2 OTotalSiAFAlTiCrMgFeCaNaTotalmfftEnFsWo52-950314-91025101330021-59O1310O29Cations per 6 oxygens1-86401360086000800071-576029800600009402684181-515-43154-780213-430339-873O401-34Oil10O471-91700830059000600091-586028900500007400684-682-41502-6* Total Fe as FeO.= MgxlOO/(Mg + Ft(Roeder et al., 1979), 73 °C (Fujii, 1978, with In k = 2\ 90 °C (Fabries, 1979), and 145 °C (Engi,1983, CO type diagram). Because the pyroxene temperatures are minimum values, theslightly higher olivine-spinel temperatures (except possibly the extremely high valuesobtained by the Engi procedure) are considered to be generally consistent with the pyroxenetemperatures. The large scatter of temperatures determined by the various thermometersprecludes all but the simplest conclusion that the temperatures are in or only slightly belowthe magmatic range, a conclusion consistent with the general lack of pyroxene exsolution.This interpretation implies that the observed deformation and recrystallization, especiallyin the dunites, occurred at temperatures only slightly below the solidus.DEPTH OF FORMATIONRoedder (1965) found a minimum depth for trapping of CO 2 inclusions in oh'vine fromHualalai dunites of 8-15 km (Roedder, 1965). The association of olivine, plagioclase,orthopyroxene, and clinopyroxene in one Hualalai ultramafic xenolith (68KAP1, Table 1)indicates a maximum depth of formation of ~30 km, the maximum depth for stability ofplagioclase lherzolite (Green & Hibberson, 1976; Presnall et al., 1979). These limits of~ 8-30 km correspond to lower-crust to uppermost-mantle depths in Hawaii. Because mostof the xenoliths do not have plagioclase, the maximum depth must be considered to betentative.

Trace element concentrations and isotope ratios of clinopyroxene separatesX33-934-391-630642172191-040812410071-6451070367±5O512963±2481179-6058ClinopyroxeniteX53-503-071-080391-321-29O60O4436-70131-24O3070351±80512946±1782-682-3O60oXmzXo-* oXmzTABLE 9BlDuniteWehrlittB3X7X14Q6511463REE (ppm)CeNdSmEuGdDyErYbK (ppm)RbBaSr30-3132-8110-513159-449-624-472-8527-20-121-871-912-639-532-720-962-903-321-631 3215-30-101184-48-916-672-050-732-222-2911408615-70040674-65115111-70O612122-0409507117-70040639-13-623-751-31O491-691-6607906149-80161-743-44 814-861-68O571-881-8108106226-8O080937-6B7 Sr/"Sr'"Nd/'^Nd% Fo (ol)mg (cpx)% Na 2 O (cpx)O70361±90512931±2189-488-40930-70353±50-512985±1785-984-30-88070348±70512933±2084-683-5073070351±60512945±3881-481-4058070357±60512972±2083-482-8051070360±30512965±2383182-8043

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 18510050o>£ 10oOXQ.O—-odunitewehrlite & clinopyroxeniteI I 1 i i iCe Nd SmEuGd Dy Er YbFIG. 5. Chondrite-nonnalized REE patterns for clinopyroxenes from dunite, clinopyroxenite, and wehrlitexenoliths in the Kaupulehu flow. Data are from Table 9.ORIGIN OF HUALALAI ULTRAMAFIC XENOLITHSIn discussing the origin of the Hualalai ultramafic xenoliths, we firstconsider whether thexenoliths are restites from partial melting of upper-mantle peridotite or cumulates formedby fractional crystallization of basaltic magma. We then consider whether the basalticmagma was associated with Hawaiian volcanism or had oceanic crustal (MORB) affinities.Finally, for the case of a Hawaiian affinity, we consider whether the parental magmas werealkalic or tholeiitic.

186 CHEN-HONG CHEN ET AL.Residues of partial fusionOlivine from abyssal peridotites has a higher and narrower range of forsterite content(Fo 89 . 3 _ 91 . 6 ) than olivine from the Kaupulehu ultramafic xenoliths (Prinz et al., 1976;Sinton, 1979; Hebert et al, 1983; Dick & Fisher, 1984; Dick, 1989). A peridotite fromZabargad Island in the Red Sea, which has been proposed as a sample of undepletedoceanic mantle (Bonatti et al., 1986), shows an olivine composition range of Fo 87 . 3 _ 90 . 5 .The pyrolite model mantle composition of Ringwood (1975) has an olivine composition ofFo 90 , and Carter (1970) proposed that olivines from undepleted upper mantle beneathKilbourne Hole, New Mexico, has compositions of Fo g6 _ 88 . Thus, olivine in both depletedand undepleted upper-mantle peridotite is more Mg rich than most of the olivine in theKaupulehu ultramafic xenoliths. These differences are inconsistent with a mantle residueorigin for the Kaupulehu dunites and wehrlites.A similar argument can be made on the basis of the pyroxene compositions. Mostclinopyroxene from abyssal peridotites has a higher and narrower range of 100Mg/(Mg+ Fe) (89-5-92-8, Prinz et al, 1976; Symes et al, 1977; Arai & Fujii, 1979; Sinton, 1979;Hamlyn & Bonatti, 1980; Hebert et al, 1983; Dick & Fisher, 1984; Dick, 1989), than that ofclinopyroxene from Kaupulehu ultramafic xenoliths (80-8-89-2, Table 7). Orthopyroxenefrom abyssal peridotites is normally more magnesian than En 86 (Prinz et al, 1976; Symes etal, 1977; Arai & Fujii, 1979; Sinton, 1979; Hamlyn & Bonatti, 198a, Hebert et al, 1983; Dick& Fisher, 1984; Dick, 1989), whereas the one orthopyroxene composition we have determinedfrom a Kaupulehu xenolith is En 82 (Table 6). These differences argue against amantle residue origin.Figure 4 shows that the range of TiO 2 content of clinopyroxene in abyssal peridotites isvery low (

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 187cumulates formed by crystallization of olivine and then clinopyroxene + olivine from afractionating magma. Evidence will be presented below that the xenoliths were not allcrystallized from the same magma, but this complication would not significantly disturb theoverall differences in Fe/Mg between rock types as long as the different parental magmaswere compositionally similar.A further question concerns the MORB or Hawaiian affinity of the magmas from whichthe Kaupulehu cumulates crystallized. Bohrson & Clague (1988) have pointed out that theTiO 2 content of spinels in Kaupulehu xenoliths that they studied is generally higher thantypical values (< 1 %) for spinels crystallized from MORBs. They used this difference toargue that the xenoliths did not crystallize from MORBs. Our results (Tables 4 and 5) areconsistent with theirs. Furthermore, if the clinopyroxene in the pyroxenite and wehrlitexenoliths is cumulus, as argued below, the concentration of Sr in the coexisting liquidswould be 313-703 ppm (using D, r = (M2; Arth, 1976). This is distinct from the average of127 ppm (Lofgren et al, 1981) and 80-145 ppm (Sun et al, 1979) reported for MORBs.Pacific MORBs have low 87 Sr/ 86 Sr (O-513O) (Fig. 6).Strontium and Nd isotopic ratios of clinopyroxene in the Kaupulehu ultramafic xenolithsare distinct from those of East Pacific Rise basalts but are similar to those of otherHawaiian basalts (Fig. 6). Therefore, a Hawaiian parentage for the ultramafic xenoliths isindicated. Clinopyroxene in two unusual gabbroic xenoliths from the Kaupulehu flow haslower 87 Sr/ 86 Sr (0-70285 ±6) and higher 143 Nd/ 144 Nd (0-51317±2) values (Clague & Chen,1986) than those of the ultramafic xenoliths. These two gabbroic xenoliths are distinct fromother gabbroic xenoliths in the Kaupulehu flow and were interpreted by Clague & Chen asrare fragments of oceanic crust entrapped in this flow with other xenoliths of Hawaiianorigin..5133.5132.51315130EPRHawaiian basalts- • : /.5129• duniteO wehrlite + clinopyroxenite.5128.7020 .7025 .7030 .70358 7 Sr/ 86 SrI.7040 7045FIG. 6. Comparison of Nd and Sr isotopic compositions for Kaupulehu xenoliths with those of basalts fromHawaii and the East Pacific Rise (EPR). The fields for lavas are from Staudigel et al. (1984), Roden et al. (1984), andChen & Frey (1985).

188 CHEN-HONG CHEN ET AL.Kaneoka & Takaoka (1980), Kyser & Rison (1982), and Rison & Craig (1983) haveargued, on the basis of similar 3 He/*He values, that the Hualalai xenoliths are fragments ofnormal oceanic crust. However, a very wide range of 3 He/*He values [(6-6-32) x atmospheric]is found in Hawaiian basalts (Kaneoka & Takaoka, 1980, Kyser & Rison, 1982;Kurz et al, 1983, 1987, 1990; Rison & Craig, 1983; Lupton & Garcia, 1986), and manyHawaiian alkalic basalts have 3 He/ 4 He values similar to those of MORBs. These dataindicate that similarities of 3 He/*He between Hualalai ultramafic xenoliths and MORB arenot diagnostic of a normal oceanic crustal origin.We conclude, on the basis of TiO 2 content of the spinels, calculated Sr concentrations ofcoexisting liquids, and Nd and Sr isotopic data, that the Hualalai ultramafic xenoliths arenot fragments of oceanic crustal material produced at a spreading center.Cumulates from Hualalai magmasThe only remaining explanation for the origin of the Hualalai ultramafic xenoliths is thatthey are cumulates crystallized from Hawaiian magmas. Because the vent for the Kaupulehuflow lies at least 25 km from any of the riftsextending from the summit areas of adjacentHawaiian volcanoes, the magmas from which the xenoliths crystallized are almost certainlyHualalai magmas. Hualalai has produced both tholeiitic and alkalic lavas, so a furtherquestion concerns the type of magma involved. Bohrson & Clague (1988) concluded thatthe xenoliths in the Kaupulehu flow were crystallized from tholeiitic magmas, and Sen &Presnall (1986) reached a similar conclusion for dunite xenoliths from Koolau volcano onOahu. In this section we use major element, trace element, and isotopic data along withmineralogical and phase equilibrium relationships to re-examine this issue for the xenolithsin the Kaupulehu flow.Major element dataFigures 7 and 8 show that spinels in Koolau dunite xenoliths have generally high Cr/(Cr4-Al) ratios like those in Hualalai tholeiites but widely varying Fe 3+ /(Cr-I-Al + Fe 3 + ) likethose in Hualalai alkalic basalts. Evidently, comparisons between different volcanoes canyield inconsistent results. In any case, the complete separation of the Koolau spinelcomposition field from that of spinels from xenoliths in the Kaupulehu flow suggests adifferent origin for the two suites.Fodor et al. (1975) and Lofgren et al. (1981) have shown that clinopyroxenes in Hawaiiantholeiitic basalts have lower normative wollastonite (Wo) content than those from Hawaiianalkalic basalts, and Bohrson & Clague (1988) have confirmed this relationship forHualalai lavas. As shown in Fig. 3, clinopyroxene compositions in the xenoliths rangeacross both the tholeiitic and alkalic clinopyroxene fields for Hualalai lavas, and in somecases lie outside both fields. Lavas elsewhere in Hawaii show somewhat larger clinopyroxenecomposition fields. An especially well-documented case is the transitional to alkalicEast Molokai stratigraphic section studied by Beeson (1976). He showed (see his fig.21) thatclinopyroxenes in the stratigraphically highest and most strongly alkalic lavas have aminimum CaSiO 3 proportion of ~41%, which decreases down-section to ~35% justabove the tholeiitic lavas. The minimum CaSiO 3 percentage we find for clinopyroxenes inthe Hualalai xenoliths is ~39%. Thus, comparison with the East Molokai lavas leads to theconclusion that the parental magmas for the xenoliths were alkalic to transitional. The datafrom East Molokai suggest that the fields for clinopyroxenes from Hualalai lavas may beunrealistically small because of a limited sample size.

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 1891001 1i i i1 1 1 180-Hualaiaitholeiitic basaltKoolau dunite\ ,-Abyssal basalt60 -" a peridotiteI•\ /20 -•* ^k ^F* m ^• ^^k m >• /• w1 ••i' /'••; Hualaiai alkalicbasaltKaupulehu wehrlite &clinopyroxeniteKaupulehu duniteii1 1 1i i i i100 80 60 40 20Mg * 100\Mg • F« 2+ 'spinelFio. 7. Comparison of spinel compositions (cation ratios) in Kaupulehu ultramafic xenoliths (data from Tables 4and 5) with those in Hualaiai alkalic and tholeiitic basalts (data from Bohrson & Clague, 1988), Koolau dunitexenoliths (data from Sen & Presnall, 1986), and abyssal basalt and peridotite (data from Clarke & Loubat, 1977;And & Fujii, 1979; Hamlyn & Bonatti, 1980; OTJtonnell & Presnall, 1980; Hebert et al^ 1983; Dick & Bullen, 1984;Dick, 1989).Rare earth element dataUse of clinopyroxene rare earth element (REE) data to determine tholeiitic or alkalicparentage of the xenoliths is complicated by the possibility that the clinopyroxene couldrepresent either a cumulus mineral in equilibrium with the parental magma or a crystallizedportion of the magma itself. In the latter case, the clinopyroxene could crystallize interstitiallyfrom trapped melt or as an adcumulus overgrowth on cumulus clinopyroxene. If theclinopyroxene is a cumulus mineral without postcumulus overgrowth, REE distributioncoefficients could be used to calculate REE concentrations in the parental magmas, andthese concentrations could then be compared with REE concentrations in Hualaiai alkalicand tholeiitic lavas. However, if the clinopyroxene is partly or completely the result of

190 CHEN-HONG CHEN ET AL.AlHualalaitholtiitic basaltCr• -duniteo-wthrlltt & olivine clinopyroxenlteFIG. 8. Comparison of spinel compositions (cation ratios) in Kaupulehu ultramafic xenoliths (data from Tables 4and 5) with those in Hualalai alkalic and tholeiitic basalts (data from Bohrson & Clague (1988) and Koolau dunitexenoliths (data from Sen & Presnall, 1986).postcumulus crystallization, comparison of REE concentrations would be complicated bythe possibility that part or all of the clinopyroxene would represent an unknown percentageof the magma, and its REE concentrations would be unrepresentative of the entire magma.Mineralogical and phase equilibrium data provide some evidence on the cumulus orpostcumulus origin of the clinopyroxene. On the basis of phase equilibrium relationshipsfrom 1 atm to 20 kb (Osborn & Tait, 1952; Presnall, 1966; Presnall et al., 1978), clinopyroxeneand olivine coprecipitate in proportions overwhelmingly dominated by clinopyroxene(80:20 to 90:10). Cumulates in layered intrusions such as the Duke Island complex, Alaska,and the Muskox intrusion in Canada confirm this relationship (Irvine, 1963, 1979). Thus,the very small amounts of clinopyroxene typically found in the Kaupulehu dunite xenolithsprobably crystallized from liquid trapped between the cumulus olivine grains. In contrast,the large percentages of clinopyroxene in the wehrlite and clinopyroxenite xenoliths areconsistent with a cumulus origin. In dunite samples AA, B4, and X14, the high percentage ofclinopyroxene grains (5-8%), the blocky shapes of these grains, and the relatively forsteritepoor(Fo 814 _ 84 . 2 ) olivine compositions suggest that these xenoliths represent a transitiontoward an olivine-clinopyroxene cumulus assemblage.

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 191To evaluate the bearing of the REE data on the cumulus vs. postcumulus origin of theclinopyroxene, we examine the consequences of assuming, contrary to the phase equilibriumand textural evidence presented above, that the clinopyroxene in all the xenoliths ispurely cumulus. On this assumption, distribution coefficients can be used to calculate REEpatterns for magmas that crystallized the clinopyroxenes, and these patterns can becompared with those of alkalic and tholeiitic lavas from Hualalai. For these calculations, weuse the preferred distribution coefficients (D) of Frey et al. (1978). As pointed out by Frey etal, these coefficients are near the lower limit of published values, but are preferred becauseany contamination of analyzed mineral grains would lead to high values. Also, the Frey etal. preferred values are consistent with experimentally determined values of 0-19 for D Yb(Watson et al, 1987), 0-16 for D Gd (Green et al, 1971), and 0-13 for D^ (Ray et al, 1983).Figure 9 shows that the calculated REE patterns of presumed magmas that crystallizedclinopyroxenes in the wehrlite and clinopyroxenite samples are approximately parallel tothose of Hualalai alkalic lavas (Clague et al, 19805) but are at higher concentrations.Because REE concentrations of Hawaiian tholeiitic lavas typically are lower than those ofHawaiian alkalic lavas, the match of the calculated liquids with tholeiitic lavas would beeven worse. One might argue that the presence of phenocrysts in the lavas analyzed by20010lavas (7 samples)Kaupulehu flowCe Nd Sm Eu Gd Dy Er YbFIG. 9. Calculated REE patterns for liquids in equilibrium with Kaupulehu wehrlite and clinopyroxenitexenoliths based on the preferred distribution coefficients of Frey et al. (1978). Data for Hualalai mafic alkalic lavasare seven samples from Clague et al. (1980a) and two samples from Lofgren et al. (1981).

192 CHEN-HONG CHEN ET AL.Clague et al. (19806) could raise the REE concentrations in the liquid fraction sufficiently tooverlap the concentrations of the calculated liquid compositions. In fact, the phenocrystproportions range from 5 to 15% (D. A. Clague, pers. comm.) and the REE patterns of thelavas reported by Clague et al. would be raised only an insignificant amount. The presumedmagmas that crystallized clinopyroxenes in the dunite samples are even more enriched inREE. Thus, the REE data are inconsistent with a purely cumulus origin for any of theclinopyroxenes. At least some postcumulus overgrowth must have occurred, even for thesamples with the lowest REE concentrations.Figure 5 shows that REE patterns for clinopyroxene from the wehrlite and clinopyroxenitesamples are strongly convex upward and lie at lower concentrations than those of typicalHawaiian tholeiites (Lofgren et al, 1981) and Hualalai alkalic lavas (compare Figs. 5 and 8).Therefore, although the REE data require at least some postcumulus crystallization, theyare consistent with the conclusion based on phase equilibrium and textural arguments thatthe clinopyroxene in these xenoliths is cumulus. Clinopyroxene separates from dunitesamples Bl, B3, and X7 have REE concentrations that fall generally in the range of those forHualalai alkalic lavas (compare Figs. 5 and 8) and Hawaiian tholeiitic lavas (Lofgren et al,1981), and are higher than those from the wehrlite and clinopyroxenite xenoliths (Fig. 5).This relationship is consistent with crystallization of these clinopyroxenes as a postcumulusmineral interstitial to cumulus olivine.The REE pattern for clinopyroxene from dunite sample X14 differs from those ofclinopyroxenes from the other dunite samples and shows the LREE depletion and generallylower REE concentrations typical of clinopyroxene from the wehrlite and clinopyroxenitesamples (Fig. 5). This feature is consistent with the earlier interpretation, based onpetrography and mineralogy, that this sample represents the transition between postcumuluscrystallization of clinopyroxene in the dunite xenoliths to cumulus crystallization in thewehrlite and clinopyroxenite xenoliths.One dunite xenolith, sample Bl, has clinopyroxene REE concentrations higher thanthose of typical Hawaiian tholeiites (Lofgren et al, 1981), which indicates that the parentalmagma for this xenolith was alkalic. This is the only sample in which the REE data help toclarify the alkalic or tholeiitic character of the parental magma.Isotopic dataHelium, strontium, and neodymium isotopic data exist both for xenoliths from theKaupulehu flow and for Hualalai lavas. Kurz et al. (1983) found three dredged tholeiitesfrom Hualalai to have 3 He/ 4 He values ranging from 14-4 to 17-6 times the atmosphericratio. Ten alkalic lavas, also from Hualalai, were found to have much lower values, rangingfrom 7-84 to 9-84 x atmospheric (Kurz et al, 1990). Data for five Kaupulehu xenoliths fromKurz et al. (1983), one from Rison & Craig (1983), and four from Kyser & Rison (1982) showa combined range of 3 He/*He from 8-60 to 9-60 x atmospheric, all clearly within the alkalicrange.If the xenoliths were crystallized from tholeiitic Hualalai magmas, their present alkalic3 He/ 4 He values might have been acquired by equilibration either with the lithosphere inwhich they crystallized ( 3 He/ 4 He = 8 x atmospheric) or with the host alkalic magma thattransported them to the Earth's surface, or both. To evaluate these possibilities, we use thedata of Hart (1984) for diffusion of He in olivine. Tholeiitic volcanism at Hualalai began~ 500000 years ago (Clague & Dalrymple, 1987), so we assume a conservative ageof t = 500000 years for formation of the xenoliths. If the temperature of the lithosphereis taken as 1100°C (approximately just below the solidus temperature of basalt at5-10 kb), the diffusion coefficient, D.ofHe in olivine is ~2x 10" 12 cm 2 s(Hart, 1984). Then

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 193.5131.5130-4— Dunite"I—p--Clinopyroxenite and wehrliteTholeiites(14 flows)/.5120Alkalic basalts(12 flows)Kaupulehu flow(3 samples).5128.7035 .703687 Sr/ 86 Sr.7037 .7038FIG. 10. Comparison of Nd and Sr isotopic compositions for dinopyroxenes from Kaupulehu xenoliths (Table 9)with those of tholeiitic and alkalic lavas from Hualalai Volcano and the host Kaupulehu flow. Horizontal andvertical lines show analytical uncertainties. Fields for the lavas are from Stille et at. (1986) and K. H. Park &A. Zindler (pers. corrun.) and include the uncertainties in their analyses.the characteristic transport distance for diffusion [X = s/(Dt)'] is ~ 0-2 mm (Hofmann &Hart, 1978). Thus, 3 He/ 4 He in dunite cumulates would not be significantly affected byexchange with the surrounding mantle.Helium isotope exchange with the host magma can be evaluated in a similar way. For thiscalculation, we assume a magma temperature of 1200°C (£> = 3 x 10" 10 cm 2 s) and axenolith diameter of 1-35 cm, the size range of xenoliths from the 1800 Kaupulehu flow(Clague, 1987). The time required for equilibration of 3 He/ 4 He would then range from 25years to 32 000 years, depending on the size of the xenolith. The maximum exposure time ofthe xenoliths to the host magma can be estimated from a calculation of the minimummagma flowvelocity required to counteract the settling velocity of the largest xenoliths andbring them to the Earth's surface. By this procedure, Spera (1980) found a minimum magmavelocity of 1-7 km/h for the 1800 Kaupulehu magma. If the xenoliths were entrained from amaximum depth of 30 km, the maximum magma exposure time during transport to theEarth's surface would be only ~ 18 h. Thus, essentially no helium isotope exchange wouldhave occurred between the magma and the xenoliths. We conclude that the helium isotopicratios of the xenoliths are representative of their magmatic origin and are a very strongconstraint indicating that the parental magmas were alkalic.Figure 10 shows that on an 87 Sr/ 86 Sr vs. 143 Nd/ 144 Nd plot, all the ultramafic xenolithslie in the field of alkalic Hualalai lavas. Three of the dunites (Bl, B3, and X7) have anunambiguous Hualalai alkalic signature; the rest could be derived from either alkalic or

194 CHEN-HONG CHEN ET AL.tholeiitic Hualalai magmas. None of the xenoliths have an unambiguous tholeiitic signature.The three dunites with a clear alkalic identity are also isotopically distinct from theirhost Kaupulehu lava (Fig. 10) and therefore were not crystallized from it This suggests thatthe other xenoliths with isotopic compositions analytically indistinguishable from that ofthe Kaupulehu lava also may not have crystallized from it. In particular, the one dunitexenolith (sample Bl) having isotope ratios within the Kaupulehu field has a texture thatshows deformation, which clearly indicates a history before being picked up by theKaupulehu magma.Multiple parental magmasBohrson & Clague (1988) observed two sequences of crystallization in the Kaupulehuultramafic xenoliths showing pyroxene exsolution. One order is: spinel; olivine; orthopyroxene;clinopyroxene; plagioclase. The other is the same except that the two pyroxenes appearsimultaneously. This difference suggests more than one parental magma. The ultramaficxenoliths without pyroxene exsolution studied here show a third sequence: spinel; olivine;clinopyroxene. Of particular significance is the fact that the xenoliths studied by Bohrson &Clague (1988) have a range of clinopyroxene mg values (82-4-85-5) that is completelyoverlapped by those in the wehrlite and clinopyroxenite samples we have studied(79-5-85-7). The former contain cumulus orthopyroxene; the latter do not. Therefore, morethan one parental magma is required. Finally, multiple parental magmas are indicated bythe isotopic heterogeneity of the clinopyroxenes. Figure 6 and Table 9 show that clinopyroxenein dunite X7, dunite B3, and clinopyroxenite X3 are distinct on an 87 Sr/® 6 Sr vs.i43Nd/i44Ndp l o tBohrson & Clague (1988) argued convincingly for a tholeiitic parental magma for thexenoliths with pyroxene exsolution on the basis that they contain cumulus enstatite. Thisconclusion is consistent with phase relationships both for natural tholeiite (Green &Ringwood, 1967) and for model system tholeiites (Presnall et al., 1978,1979; Sen & Presnall,1984; Presnall & Hoover, 1987; Liu & Presnall, 1990) as long as the pressure does notexceed ~11 kb. However, Bohrson & Clague (1988, p. 139) noted that these xenolithscomprise only ~1% of the total xenolith population. Dunites, wehrlites, and olivineclinopyroxenites studied here make up ~ 60% of the xenolith population (Jackson et al.,1981), and because they do not have cumulus orthopyroxene, we believe that theirparentage must be considered separately.Although we argue for the existence of multiple parental magmas, several features of thedunite, wehrlite, and olivine clinopyroxenite xenoliths support a coherent and commontype of origin that involves sequential crystallization of dunite, wehrlite, and olivineclinopyroxenite cumulates from alkalic to transition parental magmas. These features, someof which have already been discussed, are summarized as follows.(1) Olivine and clinopyroxene compositions are generally more magnesian in the dunitesthan in the wehrlites and olivine clinopyroxenites (Figs. 2 and 3), but a region of overlapoccurs. This region is marked by iron-rich dunites with blocky clinopyroxenes that arepossibly cumulus in origin. We have not found this textural feature among the moremagnesian dunite samples.(2) On a plot of modal percent clinopyroxene in dunite vs. percent forsterite in olivine, alarge scatter occurs; but the maximum percent clinopyroxene increases as the percentforsterite decreases (Tables 1 and 2). This is consistent with the introduction of smallamounts of cumulus clinopyroxene in the more iron-rich dunites.(3) Dunite sample XI4, which has very iron-rich olivine (Fo 81 . 4 ) and clinopyroxene thatappears to be cumulus, has a clinopyroxene REE plot like those for clinopyroxenes from the

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 195wehrlite and olivine clinopyroxenite samples. Thus, it appears to represent the transitionbetween more magnesian dunite cumulates and the wehrlite-clinopyroxenite suite ofcumulates.(4) Spinel compositions are similar in all the xenoliths but are generally more magnesianin the dunites than in the wehrlites and olivine clinopyroxenites (Fig. 7). This difference isconsistent with differences in the olivine and clinopyroxene compositions.(5) Spinel compositions indicate that the parental magmas were alkalic to transitional.Clinopyroxene compositions are not as definitive but are also consistent with this interpretation.(6) He and Sr isotope ratios both indicate alkalic to transitional parental magmas for thexenoliths, and the slight heterogeneity of Sr isotope ratios indicates several parentalmagmas.CONCLUDING STATEMENTThe samples from the xenolith beds of the 1800 Kaupulehu flow that we have studied arefar from an exhaustive representation of this rich and diverse accumulation of xenoliths. It isclear, both from our work and previous studies, that the magma sampled a wide range ofmaterials including cumulates from both alkalic (this study) and tholeiitic (Bohrson &Clague, 1988) magma chambers and even an occasional fragment of oceanic crust (Clague& Chen, 1986). The locality represents a superb, but scrambled, sample of the deep interiorof a mature Hawaiian volcano down to a depth of perhaps 30 km, and it deserves muchmore thorough documentation.Our conclusion regarding alkalic parental magmas stands in contrast to the conclusion ofSen & Presnall (1986) that the dunite xenoliths from Koolau volcano on Oahu have atholeiitic parentage. An interesting difference between the two dunite suites is the generallyhigher A1 2 O 3 content of clinopyroxenes in the dunites from Hualalai (2-48-7-02%) relativeto those from Koolau (1-37-2-67%), which suggests a different origin for the two suites. Thedifference could be caused by higher pressures of formation for the Kaupulehu xenoliths(O'Hara, 1967) or crystallization from more aluminous parental magmas. We have alsonoted that spinels from the Hualalai dunites have lower Cr/(Cr + Al) than those from theKoolau dunites.Extensive crystallization of olivine in the absence of pyroxenes and plagioclase has longbeen recognized to be an important part of the crystallization history of Hawaiian tholeiites(Macdonald, 1949; Powers, 1955; Wright, 1971) but it has not previously been recognized tobe important for Hawaiian alkalic magmas. Data presented here plus data from two otherHawaiian volcanoes, Loihi (Clague, 1988) and Mauna Kea (Atwill & Garcia, 1985), stronglyindicate that it should be so recognized. Experimental data (Presnall et al, 1978, 1979;Presnall & Hoover, 1987) indicate that increasing alkalinity of the magma would progressivelyreduce its capacity for crystallization of olivine alone before the crystallization ofclinopyroxene, but that this capacity still exists even for nepheline-normative magmas. Ourresults combined with data for xenoliths from other Hawaiian volcanoes (for example, Sen& Presnall, 1986; Clague, 1988; Sen, 1988) indicate that significant portions of the deepcores of Hawaiian volcanoes consist of ultramafic cumulates crystallized both from tholeiiticand from alkalic magmas.ACKNOWLEDGEMENTSThis work was supported by National Science Foundation Grants EAR-8018359, EAR-8212889, and EAR 8418685 to Presnall, and OCE-8415759 to Stern. We thank A. Zindler

196 CHEN-HONG CHEN ET AL.for providing access to his laboratory and for assistance with the Nd isotope analyses; K.-H.Park, G. Worner, A. Zindler, and D. Qague for permission to use unpublished Sr and Ndisotope data on Hualalai lavas; and D. A. Clague and W. G. Melson for facilitating access tothe Jackson xenolith collection at the Smithsonian Institution, Washington, DC. We thankD. A. Clague for numerous conversations and two very helpful reviews at different stages ofmanuscript preparation, and we thank T. N. Irvine for helpful comments regarding cumulustextures. Journal reviews by A. Basu, M. O. Garcia, T. N. Irvine, and especially F. A. Freyled to significant improvements in the finalmanuscript. This paper is Contribution 526 ofthe Geosciences Program, The University of Texas at Dallas.REFERENCESAlbee, A. L., & Ray, L., 1970. Correction factors for electron-probe microanalysis of silicates, oxides, carbonates,phosphates and sulfates. Anal. Chan. 42, 1408-14.Arai, S., & Fujii, T., 1979. Petrology of ultramafic rocks from site 395. Initial Reports of the Deep Sea DrillingProject, 45. Washington, DC: U.S. Government Printing Office, 587-94.Arth, J. G., 1976. Behavior of trace elements during magmatic process—a summary of theoretical models and theirapplication. J. Res. US. Geol. Sun. 4, 41-7.Atwill, T. M, & Garcia, M. O., 1985. Petrogenesis of ultramafic xenoliths from Mauna Kea: mantle or crust origin.Trans. Am. Geophys. Union 66, 1133.Basu, A. R., & Murthy, V. It, 1977. Ancient lithospheric lherzolite xenoliths in alkali basalt from Baja California.Earth Planet. Sci. Lett. 35, 239-46.Beeson, A. E^ 1976. Petrology, mineralogy, and geochemistry of the East Molokai volcanic series, Hawaii. U.S.Geol Surv. Prof. Paper 961.Bence, A. E., & Albee, A. L., 1968. Empirical correction factors in electron microanalysis of silicates and oxides. J.Geol. 76, 382-403.Bohrson, W. A, & Clague, D. A., 1988. Origin of ultramafic xenoliths containing exsolved pyroxenes fromHualalai Volcano, Hawaii. Contr. Miner. Petrol. 100, 139-55.Bonatti, E., Ottonello, G, & Hamlyn, P. R, 1986. Peridotites from the island of Zabargad (St. John), Red Sea:petrology and geochemistry. J. Geophys. Res. 91, 599-631.Carter, J. L., 1970. Mineralogy and chemistry of the earth's upper mantle based on the partial fusion-partialcrystallization model. Bull. Geol. Soc. Am. 88, 556-70.Chen, C.-Y., & Frey, F. A., 1985. Trace element and isotopic geochemistry of lavas from Haleakala Volcano, EastMaui, Hawaii J. Geophys. Res. 90, 8743-68.Clague, D. A, 1987. Hawaiian xenolith populations, magma supply rates, and development of magma chambers.Bull. Volcanol. 49, 577-87.1988. Petrology of ultramafic xenoliths from Loihi Seamount, Hawaii J. Petrology 29, 1161-86.— Chen, C.-H., 1986. Ocean crust xenoliths from Hualalai Volcano, Hawaii. Geol. Soc. Am. Abstr. Prog. 18, 565.Dalrymple, G. B, 1987. The Hawaiian-Emperor volcanic chain. Part I. Geologic evolution. In: Decker, R.W., Wright, T. L., & Stauffer, P. H. (eds.) Volcanism in Hawaii. US. Geol Surv. Prof. Paper 1350, 5-54.Fisk, M. R., & Bence, A. E., 1980a. Mineral chemistry of basalts from Ojin, Nintoku, and Suiko seamounts,leg 55, DSDP. Initial Reports of the Deep Sea Drilling Project, 55. Washington, DC: U.S. Government PrintingOffice, 607-37.Jackson, E. D_, & Wright, T. L_, 198Ofr. Petrology of Hualalai Volcano: implication for mantle composition.Bull. Volcanol. 43, 641-56.Clarke, D. B., & Loubat, H., 1977. Mineral analyses from the peridotite-gabbro-basalt complex at site 334, DSDPleg 37. Initial Reports of the Deep Sea Drilling Project, 37. Washington, DC: U.S. Government Printing Office,847-55.Dick, H. J. B., 1989. Abyssal peridotites, very slow spreading ridges and ocean ridge magmatism. In: Saunders,A. D., & Norry, M. J. (eds.) Magmatism in the Ocean Basins. Geol. Soc Spec. Publ. 42, 71-105.Bullen, T., 1984. Chromium spinel as a petrogenetic indicator in oceanic environments. Contr. Miner. Petrol.86, 54-76.Fisher, R. L, 1984. Mineralogical studies of the residues of mantle melting: abyssal and alpine peridotites. In;Komprobst, J. (ed.) Kimberlites II: The Mantle and Crust-Mantle Relationships. Amsterdam: Elsevier,295-308.Ehrenberg, S. N, 1982. Rare earth element geochemistry of garnet lherzolite and megacrystalline nodules fromminette of the Colorado Plateau. Earth Planet. Sci. Lett. 57, 191-210.Engi, M, 1983. Equilibria involving Al-Cr spinel: Mg-Fe exchange with olivine: experiments, thermometricanalysit, and consequences for geothermometry. Am. J. Sci. 283-A, 29-71.Evans, B. W, & Wright, T. I~, 1971 Composition of liquidus chromite from the 1959 (Kilauea lid) and 1965(Makaopuhi) eruptions of Kilauea volcano, Hawaii Am. Miner. 57, 217-30.

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 197Fabrics, J., 1979. Spinel-olivine geothermometry in peridotites from ultramafic complexes. Contr. Miner. Petrol.69, 329-36.Feigenson, M. D, Hofmann, A. W, & Spera, F. J., 1983. Case studies on the origin of basalt II. The transition fromtholeiite to alkalic volcanism on Kohala Volcano, Hawaii. Ibid. 84, 390-405.Flanagan, E. J., 1973.1972 values for international geochemical reference standards. Geochim. Cosmochim. Acta 37,1189-200.Fodor, R. V, KeiL K., & Bunch, T. E, 1975. Contributions to the mineral chemistry of Hawaiian rocks IV.Pyroxenes in rocks from Haleakala and West Maui Volcanos, Maui, Hawaii. Contr. Miner. Petrol. 50,173-95.Frey, F. A^ Green, D. H., & Roy, S. D, 1978. Integrated models of basalt petrogenesis: a study of quartz tholeiitesto olivine melilitites from South Eastern Australia utilizing geochemical and experimental petrological data.J. Petrology 19, 453-513.Fujii, T., 1978. Fe-Mg partitioning between olivine and spinel. Carnegie Inst. Wash. Yearb. 76, 563-9.Gladney, E. S., Burns, C. E^ & Roelandts, I, 1983. 1982 compilation of elemental concentrations in eleven UnitedStates Geological Survey rock standards. Geostand. Nev/slett. 7, 3-226.Green, D. H, & Hibberson, W. O, 1976. The instability of plagioclase in peridotite at high pressure. Lithos3, 209-21.Ringwood, A. E_, 1967. The genesis of basaltic magmas. Contr. Miner. Petrol. 15, 103-90.Ware, N. G., Hibberson, W. O, Major, A, & Kiss, E., 1971. Experimental petrology and petrogenesis ofApollo 12 basalts. Proc. Second Lunar Sci. Conf. 1, 601-15.Green, H. W., 1979. Trace elements in the fluid phases of the Earth's mantle. Nature 227, 465-7.Hamlyn, P. R., & Bonatti, E., 1980. Petrology of mantle-derived ultramafics from the Owen Fracture Zone,Northwest Indian Ocean: implications for the nature of the oceanic upper mantle. Earth Planet. Sci. Lett.48, 65-79.Hart, S. R, 1971. K, Rb, Cs, Sr and Ba contents and Sr isotope ratios of ocean floor basalt Phil. Trans. R. Soc.Lond. A268, 573-87.1984. He diffusion in olivine. Earth Planet. Sd. Lett. 70, 297-302.Hebert, R., Bideau, D., & Hekinian, R., 1983. Ultramafic and mafic rocks from the Garret Transform Fault near13°3O'S on the East Pacific Rise: igneous petrology. Ibid. 65, 107-25.Hofmann, A. W., & Hart, S. R., 1978. An assessment of local and regional isotopic equilibrium in the mantle. Ibid.38,44-61Irvine, T. N., 1963. Origin of the ultramafic complex at Duke Island, southeastern Alaska. Miner. Soc. Am. Spec.Paper 1, 36-45.1979. Rocks whose composition is determined by crystal accumulation and sorting. In: The Evolution of theIgneous Rocks. Fiftieth Anniversary Perspectives. Princeton: Princeton University Press, 245-306.Irving, A^ & Frey, F. A., 1984. Trace element abundances in megacrysts and their host basalts: constraints onpartition coefficients and megacryst genesis. Geochim. Cosmochim. Acta 48, 1201-21.Jackson, E. D_, 1968. The character of the lower crust and upper mantle beneath the Hawaiian Islands. Proc.XXIII Int. Geol. Congr. 1, 135-50.1969. Chemical variation in coexisting chromite and olivine in chromitite zones of the Stillwater complex.Econ. Geol. Mon. 4, 41-71.Clague, D. A., 1981. The nodule beds of 1800-1801 Kaupulehu flow, Hualalai Volcano, Hawaii. US. Geol.Surv. Misc. Field Studies Map MF-I355.Clague, D. A., Engleman, E., Friesen, W. B, & Norton, D., 1981. Xenoliths in the alkalic basalt flows fromHualalai Volcano, Hawaii. US. Geol. Surv. Open-File Rep. 81-1031.Kaneoka, I., & Takaoka, N., 1978. Excess 129 Xe and high 3 He/*He ratios in olivine phenocrysts of Kapuho lavaand xenolithic dunites from Hawaii. Earth Planet. Sci. Lett. 39, 382-6.1980. Rare gas isotopes in Hawaiian ultramafic nodules and volcanic rocks: constraint on geneticrelationship. Science 208, 1366-8.Kirby, S. H, & Green, H. W., Ill, 1980. Dunite xenoliths from Hualalai Volcano: evidence for mantle diapiric flowbeneath the island of Hawaii. Am. J. Set 280-A, 550-75.Kretz, R., 1981 Transfer and exchange equilibria in a portion of the pyroxene quadrilateral as deduced fromnatural and experimental data. Geochim. Cosmochim. Acta 46, 411-21.Kurz, M. D, Colodner, D., Trull, T. W., Moore, R. B., & O'Brien, K., 1990. Cosmic ray exposure dating with in situproduced cosmogenic 3 He: results from young Hawaiian lava flows. Earth Planet. Sci. Lett. 97, 177-89.Garcia, M. O, Frey, F. A, & O'Brien, P. A, 1987. Temporal helium isotopic variations within Hawaiianvolcanoes: basalts from Mauna Loa and Haleakala. Ibid. 51, 2905-14.Jenkins, W. J, Hart, S. R., & Clague, D. A, 1983. Helium isotopic variations in volcanic rocks from LoihiSeamount and the Island of Hawaii. Ibid. 66, 388-406.Kyser, T. K, & Rison, W., 1981 Systematic of rare gas isotopes in basic lavas and ultramafic xenoliths. J. Geophys.Res.. SI, 5611-30.Lindsley, D. H., & Andersen, D. J., 1983. A two-pyroxene thermometer. Ibid. 88, A887-9O6.Liu, T.-C, & Presnall, D. C, 1990. Liquidus phase relationships on the join anorthite-forsterite-quartz at 20 kbarwith applications to basalt petrogenesis and igneous sapphirine. Contr. Miner. Petrol. 104, 735-41Lofgren, G. G., Bence, A. E., Duke, M. B n Dungan, M. A, Green, J. C, Haggerty, S. E, Haskin, L. A^ Irving, A. J.,

198 CHEN-HONG CHEN ET AL.Lipman, P. W., Naldrett, A. J., Papike, J. J., Reid, A. M., Rhodes, J. M., Taylor, S. It, & Vaniman, D. T., 1981.Petrology and chemistry of terrestrial, lunar and meteoritic basalts. In: Basaltic Volcanism on the TerrestrialPlanets. New York: Pergamon Press, 132-60.Lupton, J. E., & Garcia, M. O., 1986. Helium isotopes in submarine basalts from the island of Hawaii. Trans. Am.Geophys. Union 67, 1271.Macdonald, G. A., 1949. Hawaiian petrographic province. Bull. Geol. Soc. Am. 60, 1541-96.Mathez, E. A, Dietrich, V. J., & Irving, A. J., 1984. The geochemistry of carbon in mantle peridotites. Geochim.Cosmochim. Acta 48, 1849-60.Merrier, J.-C. C, & Nicolas, A., 1975. Textures and fabrics of upper-mantle peridotites as illustrated by xenolithsfrom basalts. J. Petrology 16, 454-87.Moore, R. B., Clague, D. A^ Rubin, M., & Bohrson, W. A, 1987. Hualalai Volcano: a preliminary summary ofgeologic, petrologic, and geophysical data. US. Geol. Surv. Prof. Paper 1350, 571-86.Mysen, B. O, 1976. Experimental determination of some gcochcmical parameters relating to conditions ofequilibration of peridotite in the upper mantle. Am. Miner. 61, 677-83.1979. Trace-element partitioning between garnet peridotite minerals and water-rich vapor, experimental datafrom 5 to 30 kb. Am. Miner. 64, 274-87.O'Donnell, T. H., & Presnall, D. C, 1980. Chemical variations of the glass and mineral phases in basalts dredgedfrom 25°-3O°N along the Mid-Atlantic Ridge. Am. J. Sci. 280-X, 845-68.O'Hara, M. J., 1967. Mineral parageneses in ultrabasic rocks. In: Wyllie, P. J. (ed.) Ultramafic and Related Rocks.New York: John Wiley, 393-403.Osborn, E. F., & Tait, D. B., 1952. The system diopside-forsterite-anorthite. Am. J. Sci. Bowen Vol., 413-33.Pike, J. E. N, & Schwarzmann, E. C, 1977. Classification of textures in ultramafic xenoliths. J. Geol. 85, 49-61.Powers, H. A, 1955. Composition and origin of basaltic magma of the Hawaiian Islands. Geochim. Cosmochim.Acta 7, 77-107.Presnall, D. G, 1966. The join forsterite-diopside-iron oxide and its bearing on the crystallization of basaltic andultramafic magmas. Am. J. Sci. 264, 753-809.Dixon, J. R., O'Donnell, T. H., & Dixon, S. A., 1979. Generation of mid-ocean ridge tholeiites. J. Petrology20, 3-35.Dixon, S. A., Dixon, J. R., O'Donnell, T. H., Brenner, N. L, Schrock, R. L., & Dycus, D. W., 1978. Liquidusphase relations on the join diopside-forsterite-anorthite from 1 atm to 20 kbar their bearing on thegeneration and crystallization of basaltic magma. Contr. Miner. Petrol. 66, 203-20.Hoover, J. D, 1987. High pressure phase equilibrium constraints on the origin of mid-ocean ridge basalts. In:Mysen, B. O. (ed.) Magmatic Processes: Physicochemical Principles. Geochem. Soc. Spec. Publ. 1, 75-88.Prinz, M., Keil, K., Green, J. A, Reid, A. M., Bonatti, E., & Honnorez, J., 1976. Ultramafic and mafic dredgedsamples from the Equatorial Mid-Atlantic Ridge and fracture zones. J. Geophys. Res. 81, 4087-103.Ray, G. L., Shimizu, N., & Hart, S. R., 1983. An ion microprobe study of the partitioning of trace elements betweenclinopyroxene and liquid in the system diopside-albite-anorthite. Geochim. Cosmochim. Acta 47, 2131-40.Richter, D. H., & Murata, K. J., 1961. Xenoliths nodules in the 1800-1801 Kaupulehu flow of Hualalai Volcano.U.S. Geol. Surv. Prof. Paper 424B, B215-17.Ringwood, A. E., 1975. Composition and Petrology of the Earth's Mantle. New York: McGraw-Hill.Rison, W., & Craig, H., 1983. Helium isotopes and mantle volatiles in Loihi Seamount and Hawaiian Islandbasalts and xenoliths. Earth Planet. Sci. Lett. 66, 407-26.Roden, M. F., Frey, F. A., & Clague, D. A, 1984. Geochemistry of tholeiitic and alkalic lavas from the KoolauRange, Oahu, Hawaii: implications for Hawaiian volcanism. Ibid. 69, 141-58.Roedder, E, 1965. Liquid CO 2 inclusions in olivine-bearing nodules and phenocrysts from basalts. Am. Miner.50, 1746-82.Roeder, P. L, Campbell, I. M., & Jamieson, H. E., 1979. A re-evaluation of the olivine-spinel thermometry. Contr.Miner. Petrol. 68, 325-34.Schneider, M. E, & Eggler, D. H, 1986. Fluids in equilibrium with peridotite minerals: implications for mantlemetasomatism. Geochim. Cosmochim. Acta 50, 711-24.Sen, G. 1988. Petrogenesis of spinel lherzolite and pyroxenite suite xenoliths from the Koolau shield, Oahu,Hawaii: implications for petrology of the post-eruptive lithosphere beneath Oahu. Contr. Miner. Petrol.100, 61-91.Presnall, D. C, 1980. Dunite nodules from the Koolau shield, Hawaii: crystal cumulates from a tholeiiticmagma chamber. Geol. Soc. Am. Abstr. Prog. 12, 519.1984. Liquidus phase relationships on the join anorthite-forsterite-quartz at 10 kbar with applicationsto basalt petrogenesis. Contr. Miner. Petrol. 85, 404-8.1986. Petrogenesis of dunite xenoliths from Koolau Volcano, Oahu, Hawaii: implications for Hawaiianvolcanism. J. Petrology T7, 197-217.Shimizu, N., 1974. An experimental study of the partitioning of K, Rb, Cs, Sr and Ba between clinopyroxene andliquid at high pressure. Geochim. Cosmochim. Acta 38, 1789-98.Sigurdsson, H_, 1977. Spinels in leg 37 basalts and peridotites: phase chemistry and zoning. Initial Reports of theDeep Sea Drilling Project, 37. Washington, DC: U.S. Government Printing Office, 883-91.Schilling J.-G-, 1976. Spinels in Mid-Atlantic ridge basalts: chemistry and occurrence. Earth Planet. Sci. Lett.29, 7-20.

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 199Sinton, J. M, 1979. Petrology of (Alpine-type) peridotites from site 395, DSDP leg 45. Initial Reports of the DeepSea Drilling Project, 45. Washington, DC: U.S. Government Printing Office, 595-601.Spera, F. J., 1980. Aspects of magma transport. In: Hargraves, R. B. (ed.) Physics ofMagmatic Processes. Princeton:Princeton University Press, 265-323.Staudigel, H., Zindler, A^ Hart, S. R, Leslie, T., Chen, C.-Y., & Clague, D. A., 1984. Systematics of a juvenileintraplate volcano: Pb, Nd, and Sr isotope ratios of basalts from Loihi Seamount, Hawaii Earth Planet. Sci.Lett. 69, 13-29.Stern, R. J, & Bibee, L. D., 1984. Esmcralda Bank: geochemistry of an active submarine volcano in the MarianaIsland Arc. Contr. Miner. Petrol. 86, 159-69.Stille, P., Unruh, D. M, & Tatsumoto, M n 1986. Pb, Sr, Nd, and Hf isotopic constraints on the origin of Hawaiianbasalts and evidence for a unique mantle source. Geochim. Cosmochim. Acta 50, 2303-19.Sun, S. S., Nesbitt, R. W., & Sharaskin, A. Y, 1979. Geochemical characteristics of mid-ocean ridge basalts. EarthPlanet. ScL Lett. 44, 119-38.Symes, R. F., Beva, J. C, & Hutchison, R., 1977. Phase chemistry studies on gabbro and peridotite rocks from site334, DSDP leg 37. Initial Reports of the Deep Sea Drilling Project, 37. Washington, DC: U.S. GovernmentPrinting Office, 841-5.Takahashi, E., & Kushiro, L, 1983. Melting of a dry peridotite at high pressures and basalt magma genesis. Am.Miner. 68, 859-79.Watson, E. B., Othman, D. B, Luck, J. M, & Hofmann, A. W., 1987. Partitioning of U, Pb, Cs, Yb, Hf, Re and Osbetween chromian diopsidic pyroxene and haplobasaltic liquid. Chem. Geol. 62, 191-208.Wendlandt, R. F., & Harrison, W. J., 1979. Rare earth partitioning between immiscible carbonate and silicateliquids and CO 2 vapor results and implication for the formation of light rare-earth enriched rocks. Contr.Miner. Petrol. 69, 409-19.White, R. W., 1966. Ultramafic inclusions in basaltic rocks from Hawaii Ibid. 12, 245-314.Wright, T. L., 1971. Chemistry of Kilauea and Mauna Loa lava in space and time. U.S. Geol. Surv. Prof. Paper735, 40 pp.Zindler, A^ & Jagouts, E, 1988. Mantle cryptology. Geochim. Cosmochim. Acta 52, 319-32.Staudigel, H, & Batiza, R., 1984. Isotope and trace element geochemistry of young Pacific seamounts:implications for the scale of upper mantle heterogeneity. Earth Planet. Sci. Lett. 70, 175-95.Hart, S. R., Endres, R., & Goldstein, S., 1983. Nd and Sr isotopic study of a mafic layer from Rondaultramafic complex. Nature 304, 226-30.APPENDIX A: ANALYTICAL METHODSAll mineral compositions were determined at the University of Texas at Dallas with an ARL-EMXelectron microprobe equipped with a Tracor-Northern TN-2000 automation system. The acceleratingpotential was 15 kV, the beam current was 0-15-0-20/JA, and the beam diameter was < 1 //m. Thedata reduction procedure of Bence & Albee (1968) and the correction factors of Albee & Ray (1970)were used for all analyses. Each analysis listed in the tables is an average of 7-15 spot analyses.For the isotopic and trace element analyses, the clinopyroxene separates from dunite, wehrlite, andolivine clinopyroxenite xenoliths were hand-picked, crushed in an agate mortar, and sieved. The100-300 /xm size fraction was then re-picked under 40 x magnification until no further impuritiescould be seen. To avoid possible altered materials and contamination along grain boundaries andmineral surfaces, acid washing was carried out by the procedure of Zindler & Jagouts (1988), asfollows. The separates were sequentially washed ultrasonically in distilled water for 1 h, washed in hot(100 °Q 2-5 N HC1 for 30 min, washed in cold (room temperature) 5% HF for 10 min, rinsed fivetimeswith cold 2-5 N HC1, and finally rinsed fivetimes with distilled water. This procedure was establishedafter the results of a leaching experiment outlined in Appendix B. Distilled reagents were used for thewashing and other chemical procedures.The trace element and Sr isotopic analyses were performed at the University of Texas at Dallas.Both 6-in (for K, Rb, and Sr contents and 87 Sr/ 86 Sr) and 12-in (for REE and Ba contents) radius solidsourcemass spectrometers were used. Eight clinopyroxene separates and the USGS standard BCR-1were analyzed for REE, K, Rb, Sr, and Ba by isotope dilution after the methods of Hart (1971) andShimizu (1974). For each clinopyroxene analysis, 400-600 mg of acid-washed sample were used.Values of 87 Sr/ 86 Sr were normalized to 86 Sr/ 88 Sr = 0-11940 and corrected relative to 0-70800 for the E+ ASrCOa standard.Neodymium isotopic measurements were carried out at Lamont-Doherty Geological Observatoryof Columbia University, on a modified V.G. Micromass 30 mass spectrometer (Zindler el al~, 1984).Values of 143 Nd/ 144 Nd were normalized to 146 Nd/'"Nd=0-72190 and corrected relative to 0512645for the BCR-1 standard.

200 CHEN-HONG CHEN ET AL.TABLETrace element analyses of USGS standard BCR-1 basalt (ppm)Alin(2)WKRbSrBaCeNdSmEuGdDyErYb1397147-632767053-528-36-591-916-566-293-583-321413047-432968253-928-546-452-006-676-063-513-511400046-633067553-9296-61-946-66-33-593-3614000±700471 ±0-633O±5678 ±1653-7 ±0-828-7 ±0-66-58±0-17l-96±0-056-68±0-136-35 ±0-123 61 ±0093-39 ±008(1) This study, mean of three analyses except for Sr, Gd, and Yb (twoanalyses).(2) Stern & Bibee (1984).(3) Flanagan (1973).(4) Consensus values (Gladney et at, 1983).The rare earth element and other trace element contents of the BCR-1 standard (Table Al) are ingood agreement with those from other authors (Flanagan, 1973; Stern & Bibee, 1984) and theconsensus values (Gladney et al, 1983). Total blanks (in nanograms) determined at the University ofTexas at Dallas are as follows: K (63), Rb (0-1), Sr (2-6), Ba (4-5), Ce (0-5), Nd (0-3), Sm (0-05), Gd (006),Er (0-05), and Yb (007).APPENDIX B: SOURCES OF CONTAMINATION OFCLINOPYROXENE SEPARATESDespite care taken to obtain the purest possible clinopyroxene separates, certain contaminantsmight contribute to their trace element contents. Potential contaminants are grain surface materials,glass or fluid inclusions, and other mineral inclusions such as amphibole, mica, or apatite. Possiblegrain surface materials could include basaltic glass that penetrated into the clinopyroxene during orafter entrapment of the xenolith in the host magma or liquid trapped between mineral grains whencrystallization or recrystallization occurred. Also, CO 2 and silicate glass inclusions (Roedder, 1965;Kirby & Green, 1980, Mathez et al., 1984) may contain significant amounts of minor and traceelements such as K, Rb, La, Ce, and U (Green, 1979). Although Mysen (1979) and Wendlandt &Harrison (1979) suggested that CO 2 -rich fluids are an important carrier of trace elements in themantle, Schneider & Eggler (1986) argued that trace elements are not concentrated in H 2 O-CO 2fluids and instead are carried by silicate melts.Basaltic glass and surface materials along grain boundaries can be dissolved easily in 2-5 N HC1 anddilute 5% HF (Shimizu, 1974; Zindler et al^ 1983). In fact, analysis of leachates shows that even weakacids can remove K, Rb, Ba, La, Ce, and other elements that enter minerals in ultramafic rocks, andalso can change Sr and Pb isotopic compositions (Basu & Murthy, 1977; Ehrenberg, 1982; Zindler etal, 1983). To test for contamination by CO 2 and silicate glass inclusions, an acid leaching experimentwas applied to the clinopyroxene separate from wehrlite sample X3. The separate was analyzed afterinitial washing in distilled water, then after washing in HC1 and HF, and finally after crashing andwashing again in HO (Table Bl). After the first acid washing, the separated clinopyroxene grainswould be expected to be devoid of surface contamination, but CO 2 and silicate glass inclusions insidethe clinopyroxene grains would still be intact The final step of crashing and acid washing wasdesigned to evaluate REE and other trace element contents in the inclusions.The three analyses have almost identical REE contents (Table Bl). A similar result on the Lacontent in Hualalai ultramafic xenoliths was reported by Mathez et al. (1984). Except for Ce, theleachate has an REE pattern that is parallel to and below that of the clinopyroxene. The similarity of

ULTRAMAFIC XENOLITHS FROM THE KAUPULEHU FLOW 201TABLETrace element concentrations ofwehrlite X3 clinopyroxene separateBlAfterH 2 O wash*After firstacid was/itAfter secondacid wash%Firstleachate^Calculated |measuredCeNdSmEuGdDyErYbKRbSrBa3-774-361-630-632162121-0407842-601244-761-923-934-391 630642172191-0408124100845131-633-814-421-630642-232-241-0808421800840310954-713-621180451-501 3606404510723-91105-2346-151-0461004099510110999102609931031O99011301O301-240* Washed ultrasonically in distilled water for 1 h.f Washed for 30 min in hot (100 °Q 2-5 N HC1 and 10 min in cold 5% HF, rinsed withcold 2-5 N HC1 and distilled water five times.X First acid-washed clinopyroxene was pulverized and washed a second time in hot(100 °Q 2-5 N HQ for 30 min.§ Leachate after first add wash.|| Weight of pyroxene after first acid wash is 98-28% of the weight of pyroxene afterH 2 O wash. Similarly, the weight of leachate after the first acid wash is 1-72% of theweight of pyroxene after the water wash. Each calculated value is the weighted sum ofelement concentrations of cpx and leachate after the first acid wash. Measured values arefrom column 1.its REE pattern to those of the H 2 O-washed and first acid-washed clinopyroxene separates stronglysuggests that most of the REE in the leachate are from dissolved clinopyroxene. The large decrease ofK, Rb, and Ba contents from the H 2 O-washed to the first acid-washed sample is confirmed by thehigh values of these elements in the leachate, but the leachate is only slightly richer in Sr and Ce.Differences between the calculated and measured concentrations in clinopyroxene (column 5 in TableBl) are very small (0-3%) except for Ce, Rb, and Ba (4-6, 13, and 24%, respectively). A similarobservation for a leaching experiment on Ronda peridotites was reported by Zindler et al. (1983), andwas explained by an uneven distribution of contaminants. Also, the REE concentrations of clinopyroxeneafter the first and second acid washing are not measurably different.Silicate glass inclusions were probably still intact after the hot 2-5 N HC1 was used in the secondacid-washing process without the help of HF. If these silicate glass inclusions represent trappedbasaltic magmas, as suggested by Roedder (1965), their contamination effect on the trace elementcontents of clinopyroxene can be estimated from representative concentrations of trace elements inbasaltic liquids and the amount of included glass. To calculate the amount of included glass, it isassumed that K and Rb in clinopyroxene are derived totally from silicate glass inclusions. Theproportion of glass inclusions can then be calculated simply by dividing K or Rb contents fromclinopyroxene by representative concentrations in basaltic magmas.Because trace element concentrations for the host alkalic basalt are not available, a calculationinvolving the firstacid-washed clinopyroxene and a prehistoric Hualalai lava [sample 32 of Clague etal. (1980b)] is used to estimate the extent of the contamination effect. This prehistoric Hualalai lava ischosen because it has the lowest K/Ce and Rb/Ce among the reported Hualalai alkalic basalts, whichallows an estimate of the largest contamination effect. The calculation shows that the Ce contaminationin all cases is

202 CHEN-HONG CHEN ET AL.analyzed Hawaiian basalt. In this case, the maximum Ce contamination would be 5% for clinopyroxenefrom the dunite xenoliths and 18% for clinopyroxene from the wehrlite and clinopyroxenitexenoliths.Because of the extremely high REE contents observed in apatite (for example, up to 1540 ppm Ce;Irving & Frey, 1984), REE concentrations of clinopyroxene would need a significant correction if theycontain apatite inclusions. Even though a search with an optical microscope revealed no suchinclusions, it is instructive to examine some consequences of assuming that apatite is present in theanalyzed clinopyroxenes. Jackson et al. (1981) listed whole-rock P 2 O 3 concentrations along withmodal proportions of clinopyroxene for 17 Kaupulehu ultramafic xenoliths. If all the P 2 O 5 is due toapatite inclusions and if apatite inclusions are preferentially concentrated in the clinopyroxene, apositive correlation would exist between P 2 O 5 concentration and the modal amount of clinopyroxene.No such correlation exists. Thus, either apatite is preferentially concentrated in other phases oralong grain boundaries or it is uniformly distributed throughout If the abundance of apatiteinclusions in clinopyroxene is assumed to be representative of the abundance of apatite inclusions inthe xenolith as a whole (the worst case), the effect on REE concentrations can be calculated from thewhole-rock P 2 O 5 concentrations and representative REE concentrations in apatite. For this calculation,we use O01 wt.% P 2 O 5 in Kaupulehu dunite xenoliths and 002%^.% P 2 O 3 in Kaupulehuwehrlite xenoliths (Jackson et al., 1981) and 1540 ppm Ce in apatite (Irving & Frey, 1984). Themaximum contribution of Ce in apatite to the total Ce in dunite clinopyroxene would be 4%. Forcjinopyroxene-rich xenoliths, the maximum contribution would be 23%. Thus, the REE patterns inFig. 5 would not be significantly modified. Other mineral inclusions such as mica, if present, would notsignificantly affect the measured clinopyroxene REE values, because of their much lower REEcontents than those of apatite (Irving & Frey, 1984).From the above arguments, we conclude that grain surface materials contain substantial amountsof K, Rb, and Ba but only minor amounts of Sr and Ce. These surface contaminants are readilyremoved by leaching. Also, we conclude that contamination from inclusions does not seriously affectthe analytical results. Thus, we report the REE and trace element contents of clinopyroxenes obtainedafter the first acid wash.