JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 95, NO. BI3, PAGES 21,483-21,502, DECEMBER 10, 1990

Geochemistryof CrustallyDerived Leucocratic Igneous Rocks From the Ulugh MuztaghArea, NorthemTibet andTheir Implications for the Formation of the

L. W. MCKENNA 1

Departmentof Earth, Atmosphericand Planetary Science,Massachusetts Institute of Technology,Cambridge

J. D. WALKER

Department of Geology, Universityof Kansas,Lawrence

Igneous rocks collectedfrom the Ulugh Muztagh, 200 km south of the northem rim of the Tibetan Plateau (36ø28'N, 87ø29'E), form intrusive and extrusivebodies whose magmas were producedby partial melting of upper-crustal,primarily pelitic, source rocks. Evidence for source composition includeshigh initial 87Sr/86Sr ratios (-0.711 to 0.713), 206pb/204pb ratios of 18.72,207pb/204pb of 15.63and 208pb/204pb of 38.73. The degree of meltingin thesource region was increased by significant heating via in situ decay of radioactive nuclides; a reasonable estimate for the heat productionrate in thesource is 3.9x 10'6 W/m3. Thecrystallization ages and cooling ages [Burchfiel et al., 1989] of the earliest intrusive rocks within the suite suggestcrustal thickening began in the northernTibetan Plateaubefore 10.5 Ma, with maximum averageunroofing rates in this part of the Tibetan Plateaufor the period between 10.5 and 4 Ma at approximately< 2 mm/yr. The Ulugh Muztagh flows are at the northernedge of a widely distributedfield of Plio-pliestocenevolcanic rocks in the north-central Tibetan Plateau. The crustally derived rocks described here are an end- member componentof a wide mixing zone of hybrid magmas; the other end-memberforms mantle- derived, potassicbasanites and tephrites exposedin the central section of the Plio-Pleistocenefield. The compositional trends in these belts strike east-west, at high angle to the N30E strike of the Plateau itself. Considerationof the chemical data and publishedgeophysical data argue that the sub- Plateau mantle is mechanically detachedfrom the overlying continental lithosphere, and that in this section of the plateau the thermal structureof the asthenosphereis not responsiblefor the formation or maintenanceof the plateau's topography.

INTRODUCFION allow inferencesto be madeon boththe compositionof the Tibetan crust and the character of Miocene to Pliocene The collision(between 40 and 50 Ma) and subsequent magmatism.Aspects of the structuralgeology in the Ulugh convergence of the Indian subcontinent and the Eurasian Muztagharea are discussedby Burchfielet al. [1989]. continentcreated both the HimalayanRange and the Tibetan The Tibetan Plateau is typically divided into three Plateau[Molnar, 1988]. Althoughrecent years have seen an structuralblocks; from north to south these are the Kunlun, enormousincrease in our knowledgeof the geologyof the Qiangtangand Lhasablocks (Figure 1; Changet al. [1986]). greater Himalayan orogen, little is still known of the The JinshaSuture, dated as end-Triassic[Harris et al., 1988], geologyof the Tibetan Plateauand its surroundingareas. is the surfacetrace of the north-dipping[Harris et al., 1988] Our limitedknowledge is derivedprimarily from teleseismic or south-dipping [Pearce and Mei, 1988] subductionzone data, a smallnumber of seismiclines, and remotesensing which separatedthe Kunlun and Qiangtangterrains prior to studiesof the area[Molnar, 1988] and, with theexception of collision. As describedby Burchfielet al. [1989], Ulugh the Royal Society-AcademicaSinica Geotraverse results, few Muztaghlies astridea seriesof ophioliticfragments which directedfield observationsare available. The samplesof may mark the western extension of the Jinsha Suture into igneousrocks discussedin this report were collectedfrom this area. The mean age of crustalmaterial in the Kunlun the Ulugh Muztagh region of the north-central Tibetan blocksome 500 km eastof UlughMuztagh, along the Royal Plateau(Figure 1) and providean opportunityto constrain SocietyGeotraverse route, is constrainedby the isotopic both the thermal structureof the Tibetan crust and mantle, composition and ages of syn-collision to post-collision and the rates of crustal unroofing within the northern granitiods exposed in the to be mid- Tibetan Plateau. In addition,analysis of thesesamples Proterozoic[Harris et al., 1988]. Neodymiummodel ages of sedimentaryrocks collectedalong the Geotraverseroute also ! Nowat the Department of Geology, University of Kansas, give mid-Proterozoicages. Accordingto the geologicalmap Lawrence of the Tibetan Plateau [Ministry of Geology and Natural Resources, 1980], country rocks of the Kunlun terrain Copyright1990 by the AmericanGeophysical Union includeCarboniferous to Permianrocks juxtaposed with units Paper 90JB01427 of Triassicand Cretaceousage. Southof the suture,upper 0148-0227/90/90JB-01427 $05.00 Triassic [Burchfiel et al., 1989] to lower to middle

21,483 21,484 MCKENNA AND WALKER: GEOC'ttEMISTRYOF LECOUCRATICIGNEOUS ROCKS

50ø/• T

USSR T T T T T T 500krn

o ß ß ß

30ø• • ß ß . ß ß

.... 2oo INDIA 75øE -L ' "• 85o • J. •5 ø 95 ø 105 ø

Fig. 1. Locationmap and regionalgeologic setting of Ulugh Muztagh ("Great Snowy Mountain"), which is situated 200 km south of the northern edge of the Tibetan Plateau, . Areas of Cenozoic magmatismare shaded,basins are stippled,strikeslip faults are shown with half arrows indicating relative movements,thrust faults are shownwith barbs on upper plate. Dashedlines with open barb are suturezones; from north to south they are the Jinsha,Banggong and Indus-ZangpoSutures, respectively. International boundariesare shownby thin lines, degreesNorth latitude and East longitudeare also shown.

Cretaceoussediments form the majorityof the pre-Cenozoicleucitites, phonolites and pyroxeneandesites; these samples rocks. The apparentdiscrepancy of the age of the suture are discussed in more detail below (see Regional (end Triassic)and that of rocks which it truncates(middle Relationships). Additional major element determinationsof Cretaceous)is probablydue to the poorlyknown ages of the a subset of these samples, along with new trace element sedimentaryunits in this largelyinaccessible region. data, were includedby Pearce and Mei [1988], who discussed The broadstructure of the TibetanPlateau as illustratedby the major and trace element chemistryof volcanicrocks geophysicaldata was recentlyreviewed by Molnar [1988]. encounteredalong the 1985 Royal Society-AcademiaSinica From his review of seismic data, Molnar concluded that the GeotraverseRoute, approximately500 km east of Ulugh depth to the Moho in the Ulugh Muztagh region of the Muztagh. north-centralTibetan Plateau(36ø28'N, 87ø29'E)is some65 Country rocks in the study area proper consist of + 5 kin. While sucha crustalthickness would generally be metamorphosed,openly folded, Triassicsedimentary rocks, consideredabnormally high, it appearsto be 5 kin, and intrudedby granitoidrocks of Mioceneage [Burchfielet al., perhaps10 km, thinnerthan the crustof surroundingareas 1989]. Theseintrusives were thoughtby theseworkers to withinthe plateau. Molnar [1988] alsoconcluded that upper be responsiblefor the andalusite-grade,regional-scale asthenospherictemperatures for this area are higher than contact metamorphismof the country rocks in the area. surroundingareas, and cautiouslysuggests temperatures at Mineralseparates of theseintrusive rocks give 40Ar/39Ar the crust-mantleboundary may be as high as 1300 K. The coolingages of 10.5 to 8.4 Ma ([Burchfielet al., 1989], see neotectonicsof this area of the plateauare dominatedby of this paper for a summaryof the geochronologicdata). north-directed thrusting of the Kunlun Shan and Tibetan The intrusive rocks are overlain, above a local Plateauover the Tarim Basinat a rate estimatedby Molnar et unconformity,by boulderconglomerate overlain in turnby a al. [1987] at 6 + 4 mm/yr. sequenceof now dissectedextrusive rocks, principally ash The earliest(and, until recently,the only) reportsof the flow tuffs and flows. These flows are dated at 4.0 + 0.1 Ma geology of the area are those of Littledale [1886], (40Ar/39Ar,biotite fusion and K-feldspar, [Burchfiel et al., Backstrom and Johanssen[1907] and Norin [1946]. 19891). Somewhatmore recently, Deng [1978] reportedthe Field relationsallow divisionof the magmaticrocks into petrologyand major elementchemistry of a numberof threegroups: (1) intrusivesamples BKSP, UBTG, 2MGR, samples of Plio-Pleistocene volcanic rocks from a transect and QTD which intrudethe metamorphosedbasement; (2) southof UlughMuztagh. These rocks included ultra potassic samples (defined below as the PotassiumPoor samples) MCKENNAAND W,M.,KER: GEOCItEMISTRY OF LECOUCRATIC IGNEOUS ROCKS 21,485 which crop out as small plugs (UM10, KSPO) and dikes in Appendix 1. The extrusive samples, with the exception (QTL) that intrude the Triassic sandstoneseast of the Ulugh of UMQP, all contain similar phenocryst assemblages of Muztagh; and (3) the capping Ulugh Muztagh extrusive quartz+sanidine+plagioclase(An20_40) +cordierite+biotite+ series(MV1B, UM1B, MV2, UMVU, UM3V, UMQP). tourmaline (rare) in microlitic to glassy groundmasses.The rocks occur as ash flow tuffs (UMVU, UM3V, MV2, and UMQP) and rhyolitic flows (UM1B, MV1B). Similar modal abundances for five of the samples suggests that Samples were taken as 1 to 3 kg blocks from outcropsor, syneruptionalcrystal sorting was not significant in the ash in the caseof samples UM10, BKSP, and UBTG, in moraine flow tuffs, except for UMQP which has no modal plagioclase and stream depositsbelow outcrops,using an ice ax as the and exceptionallyhigh K20. The intrusive samples(BKSP, sampling tool. Weathered faces were removed with a 2MGR, UBTG) are dominantly hypidiomorphic to diamond cutoff saw, and cut faces were polished with SiC to panidiomorphicgranular granites, which show some signsof remove sawmarks. All sampleswere cleaned by boiling and post crystallization strain. The typical assemblages is then ultrasoundingin deionized water, then dried and hand potassium feldspar+quartz+plagioclase +biotite :krnuscovite crushed. Whole rock powderswere made from 50 to 100 g :kkaolinite(?). Visible accessoryphases are rare, generally splits using a tungsten carbide shatterbox. limited to anhederal, turbid (xenocrystic ?) zircons, and Major and trace elements were determined by X ray prismatic, clear allanite. Samples QTL and UM10 are fluorescenceon a fully automated Rigaku X ray spectrometer quartz+plagioclase+tourmaline (minor) porphyry at the University of Kansas. Major elements were run as trondhjemites with a microlitic quartz and plagioclase fused glass beads from whole rock powders,with Li2B40 7 groundmass. Sample KSPO is a granodiorite containing flux; trace elements were run as powdered disks with coarse orthoclase phenocrysts+quartz+plagioclase with cellulose binder. Uncertainties, based on repeated analyses medium-grained biotite and muscovite. Allanite is a of standards,are 0.5-1% for major elements, 1-2% for trace common accessorymineral. elements with concentrationsgreater than 50 ppm, and 5- 10% for trace elements with concentrations less than 50 Major Elements ppm. Isotope ratios and concentrationswere determinedthrough Major element compositionsof the samplesare shown in isotope dilution techniques [Hart and Brooks, 1977]. Table 1. Silica contentsrange from 71.1 to 75.5 weight Strontium and rubidium data were collected at the percent,Fe203* (all Fe as Fe203) from 0.15 to 1.5 wt %, MassachusettsInstitute of Technology (on a 23 cm, 60ø MgO from 0.0 to 0.25 wt %, and total alkaliesfrom 7.2 to spectrometer)and Universityof Kansas(VG Sector);samples 9.6 wt %. Oxide variation diagrams for K20 and Fe203* analyzed at both institutionsgive identical results within versus SiO2 are shown in Figures2a and 2b. The majority analyticaluncertainties. Strontium data are reportedrelative of the samplesdefine a trend of approximatelyconstant K20 to NBS 987 standard87Sr/86Sr= 0.71024, and are normalized (5 wt %), and slightly decreasingFe203* contents(1.5-1 wt relativeto 86Sr/88Sr=0.1194. Precisionsare 0.1% for Rb, %), over a SiO2 of 71.0 to 74.5 wt %. (As noted above, 0.03% for Sr and <0.008% for 87Sr/86Sr. Strontiumand syneruptionalsorting has apparentlyimpoverished UMQP in rubidium blank levels are insignificantfor the analyses plagioclase, resulting in high potash.) Three samples reported. All Pb analyseswere conductedat the University (UM10, QTL and KSPO) do not follow this trend, showing of Kansas,where blank levels for Pb analysesare 150-300 low K20 and very low Fe203* (as low as 0.2 wt %). These pg. All samples are fractionation corrected; the samples are henceforth referred to as the potassium-poor fractionationfactor of 0.1 + 0.05%/amuwas determinedby samples. repeated analysesof N'BS 982 common Pb standard. Instrumental neutron activation analysis (INAA) was Trace Elements carried out at MIT; proceduresand analysistechniques are reportedby lla and Frey [1984] and Lindstrom and Korotev Trace element data are given in Table 2. All samples [1982]. High U concentrationsin the samplesrequired (except for QTL and UM10, which consistentlydiffer from corrections for La, Ce, Nd, and Sm due to interferences from the other samples) show high U, Rb, and Ba concentrations U fission products[Korotev, 1985]; correctionsare listed in and low Sr and transition metal concentrations. Rare earth Appendix2. BecauseB has an exceptionallylarge thermal element (REE) distributions(see Figure 3a) are consistently neutron capture cross section, tourmaline-bearingsamples light REE (LREE) enriched, concave upward, with large, (QTL, UM10, andQTD) mayhave received anomalously low negative Eu anomalies and constant Yb concentrations. neutron fluxes [King et al., 1988]. However, modal Trace element concentrationsof samples QTL and UM10 tourmaline is low in these samples, and elemental vary inversely to those of the other samples:Sr is high, Rb, concentrationsof Rb determinedby INAA, XRF and ID Ba and U are low, and Pb is very low. LREE abundancesin agree within uncertainties,suggesting that this problemis the potassium-poorrocks are generally lower than, and high not significantfor thesesamples. Reproducibilityof the REE (HREE) abundancesapproximately equal to, abundances analysesare approximately1-5 %; the uncertaintiesquoted in the remainder of the samples. Despite these differences, with the data includeall sourcesof errors, includingU the REE distributions of the potassium-poor samples are corrections. grossly similar to those of the other Ulugh Muztagh RESULTS samples.

Petrography Sr and Pb IsotopicData The petrography of the samples are briefly described Isotopic data are shownin Table 3, and isotopecorrelation below, summarized in Table 1, and described in more detail plots for both Sr and Pb are presentedin Figures 4 and 5.

MCKENNAAND WALKER: GE(X2ttEMISTRY OFLECOU••C IGNEOUSROCKS 21,487

a ß 00 0 0

I

b

00• 0

vvvvv vv v

v vvvvv v

! uncertainty

vvvvvvv 7 7 7 7 7 7 v SiOa (wt%) Fig.2. Wholerock oxide variation diagrams for theUlugh Muztaghsamples. On thisand all subsequentdiagrams, symbols are opencircles for extrusiverocks; filled squaresfor intrusiverocks;

andfilled trianglesfor potassium-poorsamples. (a) K20 versus 1F% ß ß SiO2 (wt %) showsapproximately constant K20 overthe rangeof SiO2 for intrusiveand extrusivesamples; the potassium-poor samples(triangles) show very low K20 at relativelyhigh SiO2. SampleUMQP plots above the field. (b) Fe203*(all Fe as Fe203) versus SiO 2 though scattered, exhibits a small decrease with increasingsilica; the UlughMuztagh samples are characterizedby low transition-metalcontents, including Fe, Mg, Cr, andNi.

vv vvvv vvvvv vv

The trondhjemitic,potassium-poor samples (UM10 and QTD) have87 St/86 Sr of approximately0.7118 and 87Rb/86Sr less than 1, while the remainder of the measuredsamples have higher87Sr/86Sr ratios of 0.71555(2)to 0.71817(3)(95% uncertaintiesin the last digit quotedfor all measurementsare shown by the numbers in parentheses). Variations in Pb isotopecompositions of the Ulugh Muztagh samplesare the inverse of the Sr isotope variations (Figure 5), with potassium-poorsamples showing higher 206pb/204pb and slightly higher 208 Pb/ 204 Pb than the remainder of the samples; the latter cluster within uncertainties at 206pb/204pb= 18.73(5), 207pb/204pb= 15.63(5)and 208pb/204pb= 38.74(10). 21,488 MCCA ANDWALKER: GEOCHEMISTRY OFLECOUCRATIC IGNEOUS ROCKS

lO a

I I I I I I I I I

Element

lOOO i i i i i I I i i i i i i i o PAAS b o MV1B -_ UBTG - KSPO - - •- - UM Source

,- 10

a. 1

I I I I I I I I I I I I I I I

Element

Fig.3. Chondritenormalized [Anders and Ebihara, 1982] REE abundances forthe Ulugh Muztagh samples. Datain Table2, symbolsas in Figure2. (a) All samplesshow (La/Yb)cn greater than 10; large, negative Eu anomalies;and similarYb andLu contents.Extrusive rocks show little variation,intrusive rocks have generallylower LREE abundances than the extrasire rocks. Potassium-poor samples vary from relatively LREE-enriched(KSPO) to LREE-poor(UM10, QTL); HREEcontents for all threesamples are roughly constant.(b) RepresentativeREE abundancesfor the threesamples groups, the post-ArcheanAustralian Shalecomposite [Nance and Taylor,1976] (PAAS, open squares with solid line), and 0.3-PAAS (open squares,dotted line). A possiblemodel for the REE abundances in the Ulugh Muztagh source region is given by the 0.3-PAASline. See text for discussion. MCKENNA AND WAIXER: GEOCttEMISTRYOF LECOU•TIC IGNEOUSROCKS 21,489 21,490 MCKENNAAND WAIXER: GEOCHEMIgrRY OF LECOUCRATIC IGNEOUS ROCKS

0.720 Extrusive rocks. Field relationships of these units indicate that they were extruded approximately ß typicaluncertainty synchronously,an observation supported by the similar 0.718 ß 40Ar/39Arages (biotite and K-feldspar)for two of the samples(MV2 and UM1B) of 4.1 + 0.1 and 4.0 + 0.1 Ma [Burchfiel et al., 1989]. The major element variationsof 0.716 these rocks are similar to those producedexperimentally by meltingof pelitic schist. The K20 versusSiO 2 andA120 3

0.714

' I ' I " I ' I " I

0.712 38.9

0.710 0 10 20 30 4O 50 38.7 87 86 Rb/ Sr Fig. 4. Sr isotopecorrelation diagram for the Ulugh Muztagh samples.Symbols as in Figure2; 2o uncertaintiesare smallerthan 38.5 the symbols. Althoughthe samplesdo not define an isochron,a referenceline with a slopeequivalent to an age of 11 Ma is shown. Thepotassium-poor samples have very low 87Rb/86Sr ratios and typical indicate their initial ratios of-0.7117. The intrusive and extrusive uncertainty sampleshave similar contemporary 87Sr/86Sr ratios, despite the 38.3 , I I i I I • I differencein crystallizationtimes of at least 4-6.5 Ma [Burchfietet 18.0 19.0 20.0 21.0 al., 1989].Average initial 87Sr/86Sr ratios for the intrusive and 206 204 extrusive rocks (corrected for an age of 10.5 and 4.0 Ma, Pb/ Pb respectively)are 0.7123(7) and 0.7154(5).

20.5 I

DISCUSSION typical Petrogenesisof the Ulugh Muztagh Samples 20.0 uncertainty We hypothesize below that the composition and mineralogy of the extrusive and intrusive samplegroups are consistentwith derivation by partial melting of chemically similar pelitic or psammiticsource rocks. Ideally, evidence 19.5 for such an argument would include observations and analysesof the restitic material. The Tibetan Plateau is, however, hardly an ideal place in which to conductfield 19.0 research, and such observations and samples are not available. Therefore we support our hypothesis by comparing the composition of the Ulugh Muztagh samples to the expected composition of liquids produced by partial 18.5 , I , I , I , melting of crustal material. The characteristicsof partial 0.7100 0.7125 0.7150 0.7175 0.720 melting of pelitic rocks precludes detailed petrogenetic modeling of the processbecause (1) a wide variety of phases are potentially stable in the restitc; (2) the apparent 87Sr/86 Sr partition coefficientsfor many of thesephases are poorly, if Fig. 5. Lead-leadand Pb-Sr isotope correlation diagrams for the at all, constrained, and; (3) the apparent partition Ulugh Muztagh samples,symbols as in Figure 2, typical coefficients of these phases can vary strongly with liquid uncertainties(2o) areshown. (a) 208pb/204pbversus composition in highly silicic liquids. Instead, we make 206pb/204pbplot shows the extrusive samples and all but one of quantitative arguments on the composition of the source the intrusive samples (QTD) plotting within uncertainty at area, the degree of partial melting in the sourcearea, and the 206pb/204pb= 18.72 and 208pb/204pb = 38.73. Potassium-poor samples QTD and UM10 (triangles) have considerably higher temperatureof the system during melting by comparingthe 206pb/204pbratios, due to their very low Pb contents of <2 ppm. composition of the melts to those expected through partial With sampleKSPO, the potassium-poorsamples define a trend that melting of crustal materials. We then present evidence that terminatesat the primary extrusive-intrusivegroup. (b) The plot of the potassium-poorsamples were producedby a combination 206pb/204pbversus 87Sr/86Sr shows variable 87Sr/86Sr with of partial melting of amphibolitic or tonalitic source rocks, constant206pb/204pb for boththe intrusive and extrusive rocks. along with variable addition of a potassium feldspar Thepotassium-poor samples scatter at high206pb/204pb and low componentto the liquid. Each sample group is discussedin 87Sr/86Sr.QTD appears anomalous, withhigh 206pb/204pb and turn, beginning with the extrusive rocks. lower87Sr/86Sr. MCKENNAAND WAIXER: G•MISTRY OFLECOUCRATIC IGNEOUS ROCKS 21,491 versus SiO2 variations for the extrusive samples are shown 6 superimposedupon the 10-kbar, "pelitic, fluid absent "melt compositionsof Vielzeuf and Holloway [1988] in Figure 6. ß a These compositions represent partial melts of a quartz- plagioelase(An30)-kyanite-museovite-biotite- 5 oco•OO • ' garne•staurolite+chlorite source with a pelitic bulk composition,melted over a temperaturerange of 1150 K to 4 1520 K. In these experimentsthe SiO2 contentof the melt decreasesmonotonically with increasingtemperature. Also shown in these figures are the water-saturatedhaplogranite minima of Nekvasil [1988], calculated at 5 and 10 kbar. As shown in Figure 6, the composition of the Ulugh 2 Muztagh extrusive rocks is distinctly different than compositions of haplogranite melts, as modeled by

Nekvasil's calculated compositions [Nekvasil, 1988]. As 1 noted by Nekvasil [1988, p. 979], "partial melts derived from source regions with relatively high H20 contents should cluster around the haplograniteminimum composition 0 unless the An[orthite] content of the source area is high." Because the Ulugh Muztagh extrusive rocks do not cluster about any composition,we infer that the rocks were derived either from an H20-undersaturatedor anorthite-rich source area. 17 Further insight into the chemical composition of the sourcearea is gained by comparingthe experimentalpelitic meltsof Vielzeuf and Holloway [1988] to the Ulugh Muztagh data. The Ulugh Muztagh extrusive rocks show a smooth variation in K20-A1203-SiO 2 composition space that parallels, although is not coincident with, the 15 0 ß ß o o experimentallyderived melt composition. Higher potash o ß lO kb contentsin the Ulugh Muztagh samplescould be due to small differences in anorthite contents of the source area. 14 5 kb The calculationsof Nekvasil [1988] indicatethat increasing the An content of plagioclasein the sourcefrom 30 to 50 13 , I , I , mole percentwould cause an increasein both quartz and 65 70 75 80 orthoclasecomponents in the melt. The effect of adding5 wt % quartz and 5 wt % orthoclase to one of the SiO2(wt%) experimentalmelts is illustratedby the arrow in Figure 6. The resultant change in melt compositionis sufficient to explain the differences in potash contents between the Fig. 6. Oxide variation diagrams for laboratory melting experimentalliquids of Vielzeufand Holloway[1988] and the experimentsof Vielzeuf and Holloway [1988] (shadedfield) and the Ulugh Muztagh extrusive rocks. The lower alumina contents calculated dry haplogranite minima of Nekvasil [1988] at 5 and 10 in the Ulugh Muztagh samplesrelative to the Vielzeuf and kbar. The data from Vielzeuf and Holloway [1988] representthe Holloway [1988] liquids is also partly explained by measuredcomposition of a melt producedby fluid-absentpartial increasesin the quartz and orthoclasecomponents in the melting(at 10 x 108Pa) of a pelitic-compositionsourcerock; SiO 2 melt. The sensitivity of the alumina compositionto decreaseswith increasingdegree of melting. Superimposedupon the melt data are the data for the Ulugh Muztagh samples(symbols as in pressurechanges (as shownby Nekvasil'scalculated minima) Figure2). (a) K20 versusSiO 2 illustratesthe similaritiesin bulk suggeststhat the pressurein the sourcearea may also have compositionof the experimentallydetermined melts and the Ulugh influencedthe aluminacontent of the UlughMuztagh melts. Muztagh intrusive and extrusive samples; the potassium-poor This discussionindicates that the Ulugh Muztaghmagmas samples (triangles) show no relationship to the partial melt have compositionsconsistent with derivationby partial composition. The potash composition of the Ulugh Muztagh samplesis parallel to, but greater than, the trend of the Vielzeuf and melting of a source of pelitic composition:the SiO2 Holloway data. This differenceis probablydue to higheranorthite contentsof the magmaswould then correspond, according to contentsin the Ulugh Muztagh source area, which increasesmelts the dataof Vielzeuf and Holloway[1988], to a temperaturequartz and K-feldsparcomponents. The changein melt composition and volumefraction of melt (Fv) of approximately1100- due to additionof 5 wt % quartz and K-feldsparis shownby the 1200 K and 60%, respectively. This value for F v is an arrowleading from the shadedfield. (b) The A1203versus SiO 2 for approximationonly and probablyrepresents an upper limit datafrom Vielzeufand Holloway [1988] demonstratesthe relatively low alumina contentsof the Ulugh Muztagh samplescompared to estimatefor the actual degreeof partial melting in the source. An estimateof the relativedifferences in F v for the the10 x 108Pa melts. Note the slight increase in A120 3 with decreasingSiO 2 in the Ulugh Muztagh extrusivesamples, a trend extrusive rocks can be made by assumingthat La acted which parallels that in the experimentalmelts. The differencesin completelyincompatibly during melting, in which case the alumina may be due to lower pressureof fusion in the Ulugh relative ratio in Fv is equal to the ratio of La concentrations Muztagh sourceregion. 21,492 MCKENNAAND WALKER: GEOCHEMISTRY OFLECOUCRATIC IGNEOUS ROCKS in the samples.The extremeLa ratiofor the extrusiverocks intrusiverocks range from <1040 K to 1150 K. Becausethe is 1.2, suggestingthat the variationin Fv was small. degree of partial melting within a source of pelitic Additional evidencethat the extrusiverocks were produced composition changes dramatically within this temperature by meltingof a peliticsource is providedby the Sr andPb range (10-60%, [Vielzeuf and Holloway, 1988]), it is isotoperatios of the samples. The Pb data are shown possible only to suggest a broad range of Fv, from 15 to superimposedupon Zartman and Doe's [1981] Pb 60%. Again assuming that the LREE acted completely compositiondiagrams in Figures7a and7b. All extrusiveincompatibly during melting, the relative difference in F v is samplesplot tightly within or adjacentto the "Pelagic small, <1.1. Sediments" field in both variation diagrams, and plot Lead and strontium isotope ratios of the intrusive rocks externallyto fields containing80% of measuredupper also argue for a source region of pelitic composition. crustal,lower crustal and manfie derived rocks. The measured Strontium isotope ratios for some of the intrusive samples 87Sr/86Srvalues for the extrusivesamples range from must be recalculatedto accountfor postcrystallizationdecay. 0.71636(2) to 0.71735(8); recalculatingthese ratios to Crystallization ages for these rocks are not available: correctfor postcrystallizationdecay (based on a 4.0 + 0.1 however,three samples have been dated by the40Ar/39Ar Ma agefrom 40Ar/39Ar geochronology [Burchfiel et al., method and give ages of 10.5 + 0.1 Ma (UBTG, Muscovite 1989]gives a source87Sr/86Sr of 0.7154(5).This high plateau)to 8.4 + 0.1 Ma (QTD, K-feldspartotal gas). Thus a value falls well within the "crustal"field of Faure's [1986] minimum crystallization age for the intrusive suite is 10.5 Ma, and recalculationof initial isotope ratios based on this Rb/Sr evolutiondiagram (his Figure10.4), andrequires a sourcearea with a high time-integratedRb/Sr ratio. age will representmaximum estimatesfor the initial ratios. Miller [1985, p. 674] described12 "Criteria for Age-corrected Pb isotope ratios are not considered IdentifyingPelitic Parentage of IgneousRocks," reproduced significant enough (<0.05) to require their use in the hereas Table4, whichincludes five criteriainvolving major discussion. Lead isotope data are presentedin Figure 7, elementchemistry, four involvingtrace elements, and three superimposed upon Zartman and Doe's [1981] Pb involvingisotope ratios. With the datadescribed above, compositiondiagram. With the exceptionof sampleQTD, the intrusiverocks plot, like the extrusiverocks, within or only eight of these tests are possible. Comparisonof the criteria in Table 4 with the data in Tables 1 through4 shows adjacentto thepelagic sediments field. Initial87Sr/86Sr that all pelitic parentagecriteria are met or surpassedby the ratios range from 0.7123 to 0.7133, and are higher than the extrusiverocks (with the exceptionof UMQP) collectedfrom 87Sr/86Sr mantleevolution curve of Faure [1986], Ulugh Muztagh. We concludethat althoughour knowledge suggestingthe intrusive magmasformed by melting of a of source area chemistry and the chemographyof melting source with "crustal" composition. Comparison of the relationships is imprecise, the compositionsof the Ulugh Miller's [1985] criteria in for pelitic parentagein Table 4 Muztagh extrusiverocks are consistentwith derivation as a with the data in Tables 1 through 4 show that all pelitic partial melt of a pelitic sourcearea. The degreeof partial parentage criteria except for normative corundum and melting is broadly constrainedat 50-60%, at a temperature slightly high Na20 in QTD are met or surpassedby the of 1100-1200 K. intrusiverocks collected from Ulugh Muztagh. Intrusive rocks. The geochemical,geochronological and The40Ar/39Ar cooling ages for the intrusive and extrusive petrologicaldata presentedabove suggestthat the extrusive rocks exposed at Ulugh Muztagh imply minimum rocks analyzed in this study form a genetically related, emplacement ages for these units of 10.5 and 4.0 Ma comagmaticseries. However, the relationshipbetween these respectively. Despite these age differences, the extrusive extrusive rocks, the intrusive rocks (UBTG, QTD, 2MGR, and intrusive samples plot together on the Sr isotope and BKSP) and the potassium-poorsamples (KSPO, UM10, correlation diagram in Figure 4. Correcting for their and QTL) exposed in Ulugh Muztagh is less certain. respective ages, the intrusive and extrusive rocks have Evidencefor a pelitic sourcearea for the intrusivesamples is distinctaverage initial 87Sr/86Sr ratios of 0.7123(7)and suggestedby evidencesimilar to that used for the extrusive 0.7154(5), respectively. The scatter in the initial ratios samples. The K20 and A120 3 variations for the intrusive may reflect $r isotopeheterogeneity of the sourcearea; such rocks are illustrated in Figure 6 along with the heterogeneityin leucocraticrocks formed by partial melting experimentally determinedpelitic partial melts of Vielzeuf of sedimentsis common, as discussedby, amongst others, and Holloway [1988]. The intrusive rocks have similar to Le Fort [ 1981 ]. higher SiO2 and lower A120 3 and K20 than the extrusive Potassium poor samples. The three potassium-poor rocks. By comparison to the experimentally produced samples(KSPO, UM10, and QTL) differ substantiallyfrom liquids of Vielzeuf and Holloway [1988], temperaturesof both the intrusive and extrusive rocks discussed above. The formation for the magmasrepresented by the Ulugh Muztagh most striking differences, seen in Figures 2a and 2b and

TABLE4. Criteriafor Identiftinl• Peltic Source Areas ParagenesisMajor'Element Concentration Trace Element Concentrtion Isotopic Composition (wt%) (ppm) quartz Na20 -3.5 - 4 Rb > 100 87Sr/86Sr> 0.701 A12SiO5 CaO < 2 Sr < 300 - 400 180/16O > 11-12%o + cordierite SiO2 > 65 Ba < 600 - 1000 + garnet Norm C* > 5 Rb/Ba > 0.25 + muscovite Modified from Miller [1985, Table 3]. Sr-Nd isotoperelations are not shown. * Nonnative Conmdum MCKENNAAND WALKER:GEOCHEMISTRY OF LECOUCRATICIGNEOUS ROCKS 21,493

15.9 a PelagicSeds. M UpperCrust

15.7

15.5

Lower Crust 15.3

15.1 I I I

40 ! I b Upper Crust

M Pelagic Seds.

Arc

Oceanic Volc Rocks Lower Crust

37 I I 16.5 17.5 18.5 19.5 20.5

206 204 Pb/ Pb

Fig. 7. Lead isotopedata for the UlughMuztagh samples (typical 20 uncertaintiesare shownin Figure5) superimposedupon the reservoir summary diagram of Zartman and Doe [1981]. Solid lines enclose approximately80% of all data points derived from each reservoir,including "probableaverage values" for pelagic sedimentsof Mesozoic and Cenozoic age. Range of whole rock Pb ratios for the Manaslu leucogranite(Nepalese Himalaya [Vidal et al., 1982] are shown by box labeled "M." (a) The207pb/204p versus206pb/204pb data. All samples except for QTL, UM10 (triangles) andQTD (solid square) plot within or adjacentto the Pelagic Sedimentsfield. As noted in text, correctionfor postcrystallizationdecay would probablyrelocate QTL and UM10 to withinthe Pelagic Sediment field. (b)The 208pb/204pb versus 206pb/204pbdata also plot within the Pelagic Sediments field, except for samples QTL, UM10 (triangles) and QTD (solid square). As notedin the text, correctionof QTL and UM10 ratiosfor postcrystallization decaywould relocate the samplesto withinthe PelagicSediment field.

Tables2 and 3, are the low to very low K20, Fe203, MgO, examination and repeated staining of slabs failed to show Pb and Rb contentsof the potassium-poorsamples at SiO2 any modal K-feldspar in these two samples, while the contentsequivalent to the remainderof the Ulugh Muztagh relatively K20-rich, granodioritic KSPO has 13.5 wt % samples. As noted in Appendix 1, samplesUM10 and QTL normative K-feldspar. The chemical relationship between are trondhjemites composed entirely of porphyritic these samples and the remainder of the Ulugh Muztagh plagioclase,quartz, and rare tourmalinein a microcrystalline samples is investigated in Figure 8, a Pearce-type variation groundmassof the sameminerals, with accessoryapatite and diagram [Nichols, 1988] displaying the variation of 3K/Yb very rare zircon. Although these sampleshave 0.8 and 2.6 versus Si/Yb. In this diagram, addition of componentswith wt % normarive K-feldspar, respectively, thin section a cation ratio K:Si of a:b has slope 3 a/b; thus the trend 21,494 MCKENNAAND WALKER: GEOCttEMISTRY OF LF•OUCRATIC IGNEOUS ROCKS

1.0 . [ ' I ' I ' I largestresiduals are in A1203,CaO andNa20, dueperhaps to removalof a plagioclasecomponent from the QTL or UM10 liquid. This hypothesiswas not testedbecause the resulting 0.8 model would have been overdetermined. Thesemass balance calculations require the subtractionof 12 and 13 wt % K-feldsparfrom KSPO to producethe compositionof QTL and UM10, respectively. Such a 0.6 dramaticchange in bulk compositionshould be accompanied by a parallel changein the concentrationof traceelements; Ksp, Mica 0.4 thosecompatible with K-feldsparshould be greatlydepleted in QTL andUM10 in comparisonto KSPO. Thishypothesis could be testeddirectly by analysisof K-feldsparand 0.2 •V& plagioclaseseparates for thesesamples, but becausethese Qtz, Pig dataare currently lacking we usea morequalitative approach of comparingwhole rock Rb andSr contentsto changesin 0.0 I I I I total normativeweight percent K-feldspar and plagioclase 2.4 2.6 2.8 3.0 3.2 3.4 between UM10 and QTL, as normalizedrelative to KSPO. Si/Yb BecauseRb and Sr are, relativeto one another,strongly partitionedinto K-feldsparand plagioclase,respectively Fig. 8. A Pearce-typeelement ratio diagram[Nichols, 1988] for K [Nashand Craecraft,1985], and other phases in the rocks and Si. On this diagram, trends controlledby minerals with K:Si (quartz and minor tourmaline)have small massesof these ratios of 1:3 (K-feldspar, micas) will have a slope of unity, as elements, a large fraction of the whole rock Rb and Sr shown by line labeled "Ksp". Addition or subtractionof phases with K:Si of 0 will have a slope of zero, as indicatedby the line shouldreside in the K-feldsparand plagioclasephases, labeled "Qtz, Plg." Symbolsas in Figure 2. Data for intrusive and respectively. This comparisonis shownin Table 5 and the extrusive groups fall approximately on a common line with slope results,though very qualitative,support the hypothesisthat 0.2 + 0.09 (2o); this line has a nonzero intercept at a 90% the chemicalevolution of the potassium-poorrocks was confidencelevel [seeNichols, 1988]an r 2 of 0.81,and is controlledby removal of K-feldspar. Changesin Rb significant(nonzero) above the 99% confidencelevel. Data for the contents in QTL and UM10 measured relative to KSPO are three potassium-poorsamples define a line of slope0.9 + 0.7 (2o); very similar to the relative K-feldsparcontents, while the ther 2 of 0.98is significantat the 95% confidence level, although change in relativeSr contents,though a factorof 2 greater the regressionitself is significant only at the 90% confidencelevel. than the relative plagioclaseincrease, show the samesense The 3K/Si ratio of the experimentsof Vielzeuf and Holloway [1988] (and hence the slope on the Pearce type diagram used here) for the of change. We concludethat removalof K-feldsparand SiO2 range sampledby the Ulugh Muztagh rocksaverages 0.203 + quartz from an initial melt was the dominant mechanismof 0.003, and falls to a value of 0.134 at an SiO2 of 66 vet%. Thus formationfor the potassium-poorsamples, even though we the slope of the intrusive and extrasire line in this projection is are not able to offer a testablehypothesis to explainthe consistentwith derivationby melting of a pelitic source. The 3K/Si mechanismby which this removal occurred. ratio for a granite minimum melt is 0.24, which is greaterthan that derived from the Ulugh Muztagh at greater than a 95% confidence Otherscenarios for theproduction of thepotassium-poor interval. The slope of the trend line for the potassium-poor samplescould includelate weatheringof an initially samples, although subject to great uncertainty, is consistentwith potassium-richrock or alterationof an initiallypotassium- the unit slope of the trend line expectedif removal of K-feldspar from the system was the dominant mechanism for chemical richrock by metasomaticfluids. The trendsin Figure8 differentiation. argues against the metasomatism model: such alteration wouldhave had to removenot only potassium,but also A1 and Si in the ratio expectedfor K-feldspar,an unlikely defined by addition of K-feldspar (as well as biotite or happenstance.Furthermore, there is little petrographic muscovite)has a slope of unity, while the slope for quartz evidenceof plagioclasereplacement of K-feldspar in samples and plagioclaseis zero. Figure 8 clearly illustratesthat the UM10 andQTL. Althoughlate stageweathering of the potassium-poorsample trend KSPO-QTL-UM10 could be potassium-poorsamples could explainthe odd Pb/U ratio in producedsimply by removalof K-feldspar(or mica) from a thesesamples (see below), these samples were among the liquid. A simpleleast squares regression of this trendgives freshestwe analyzed,and no petrographicevidence for a slopeof 0.9 ñ 0.7 (2s) and while the uncertaintiesare alterationis presentin the samples. clearly enormous,additional information is available to SamplesQTL andUM10 havevery high 206pb/204pb support this conclusion. ratios,but 207pb/204pband 208pb/204pb that are only A more detailed examination can be carried out through slightly higher than those of sample KSPO. The Pb ratios mass balance calculations. Whole rock contents of SiO2, of the latter sample are approximatelyequal to thoseof the A1203,CaO, Na2 ¸ andK20 in QTL andUM10 werealtered intrusive and extrusive samples collected from Ulugh by additionof orthoclase(Or) and quartz(Qz) to give the Muztagh.Part of the discrepancyin 206pb/204pb ratios best fit to the KSPO compositions.Results and assumptionswithin the potassium-poor samples is due to are given in Table 5. KSPO can be manufacturedfrom postcrystallizationdecay of 238Uin samplesUM10 and mixtures of 0.8 1QTL+0.12Or+0.07Qtz and QTL, which have extremely low Pb contents (less than 2 0.8UM10+0.13Or+0.07Qtz(weight fraction). These models ppm for both). Unfortunately, the Pb analysesfor UM10 havehigh, but not terriblyhigh Z 2 of 36 and16.3, and QTL are not sufficiently accurateto allow correctionfor respectively,with two degreesof freedom. In both casesthe this decay, and thus their initial lead isotope ratios are not MCKENNA AND WALKER: GEOCHE••Y OF LF•OUCRATIC IGNEOUSROCKS 21,495

TABLE 5. Mass Balance Constraints on the Formation of the PotassiumPoor Samples Major Element Constraints ModeledCompositions SiO2 A1203 CaD Final(KSPO) 74.84 14.13 0.42 4.80 2.55 QTL 74.09 14.69 1.20 7.48 0.54 KSPOmodeled as 0.80 QTL+ 74.69 14.13 0.96 5.98 2.63 0.13Or+0.07 Qz (•2=35.7)

UM10 75.50 15.10 0.68 7.16 0.23 KSPO modeled as 0.778 UM10+ 76.09 14.27 0.53 5.57 2.51 0.138Or+0.084 Qz (Z2=9.8) Trace Element Constraints Sample Plag,wt Ksp,wt St, ppm Rb, ppm % % KSPO 41 15 54 388 QTL 67 3.2 171 49.7 Fractionalchange 1.7 0.2 3.2 0.13 UM10 62 1.4 135 21.7 Fractionalchanse 1.5 0.09 2.5 0.06 Z2 valuesin thetable are calculated assuming fractional (not wt %) uncertaintiesof 1% for SiO2 andA1203 and 10%for otheroxides. The fractional changerows in the bottomtable are calculatedby dividingthe quantityfor QTL or UM10 by the correspondingquantities in KSPO. In this sectionof the table, wt % are nonnative, not modal, percentages. well constrained. If, as suggestedabove, KSPO, QTL and in the hanging wall of thrust faults by release of fluids from UM10 are comagmatic,than the Pb isotope ratios for KSPO, footwall rocks ("flux melting") [LeFort, 1981]; and in situ which has not changed significantly since crystallization, radioactive decay and mantle heating in thickened should give the initial ratio for all three potassium-poor continental crust [Molnar et al., 1983]. The amount of melt samples:this sampleplots within the Pelagic Sedimentfield produced by "flux melting" is probably minor without the in Figure 7, although it does appear to be distinct from the input of additional heat and is not consideredfurther here. intrusiveand extrusive samples. The 87Sr/86Sri for samples Shear heating, while capableof producinglarge quantitiesof QTL and UM10 is approximately equal to their measured heat obviously requires the existenceof a fault in the source ratios due to the low Rb/Sr ratio of the rocks, the samples area, and while some Chinese workers have suggestedthrust average 0.7118(1). This ratio plots well within the faults of Cenozoicage with large displacementsmay underlie "crustally derived" field of Faure [1986], suggestingthat the Ulugh Muztagh region (B.C. Burchfiel, personal these magmas were derived from a source with crustal communication, 1988), little detailed information is composition. available to constrainheat productionby this mechanism. Heat productionin the Ulugh Muztagh sourceregion from In Situ Melting of Source Rocks in situ radioactive decay can be estimated from the A necessary (although not sufficient) requirement for a radionuclide concentration in the extrusive rocks, and their crustal sourcefor the Ulugh Muztagh magmasis that the P-T- estimated degrees of partial melting, F v. Assuming bulk X conditions within the crust were sufficient to produce a distributioncoefficients for Th and U of 0 during melting, an partial melt. A number of mechanismshave been proposed F v of 0.2 to 0.6, K20 concentrationin the sourceof 4-5 wt for in situ melting of crustal rocks, including frictional %, anda sourcedensity of 2.8 x 103kg/m 3, the calculated heating along thrust faults [LeFort, 1975]; melting of rocks heat production in the source would range from 1.7 to 4.6

TABLE 6. Source-Rock Concentrations ' ' Minimum BestEstimate " Maximun• • Concentration* U 4 10 12 7h 5 12 14 K20 2.6 3.6 4 HeatProduction t •tW/m3 1.7 3.9 4.6 HGU 4.0 9.4 11 * 'Uand"•Th concentrations inpans Per million, K20 in wt %, and are' basedon U and Th concentrationsof the Ulugh Muztaghvolcanic rocks, dividedby degreesof partial meltingof 0.2 (minimum),0.5 (bestestimate) and 0.6 (maximum), and assumeddistribution coefficients of 0 for both Th and U. , Heatproduction assuming a density of2.8x 103 kg/m 3. Heat productionrates from Turcotte and Schubert[1982] 21,496 MCKENNAAND WALKER: GEOCI•MISTRY OF LECOU•TIC IGNEOUSROCKS

!.tW/m3. Detailsof thecalculations are given in Table6. for the Ulugh Muztagh intrusive and extrusiverocks, leading While these rates of heat production are higher than those to temperature increasesof 70 to perhaps 200 K. These generallyquoted for leucogranitesource areas [e.g., Pinet and thermal perturbations can be maintained for long time Jaupart, 1987], these estimatesrepresent conservative limits scales, due to the long half-lives of the radionuclides. for the Ulugh Muztagh sourceregion: while decreasingFv Withdrawal of the radioactive nuclides by partial melting, would cause a linear decrease in the heat production rate, such as invoked for the intrusive and extrusive rocks, would increasing the density of the source or the distribution reduce (to zero, with the assumptions above) the coefficients for Th and U would increase heat production concentration of the radionuclides, effectively halting the within the source. The presence of stable zircon in the temperature increase by this mechanism. For nonzero restitc would tend to increase the apparent partition apparent partition coefficients, residual U and Th would coefficient of U in the system, and the model described remain in the source and could continue to heat the restitc. above would then underestimate the source U concentration, The pressure-temperature-compositionconditions necessary and hence the sourceheat production. The inferred Th and to producepartial melts in pelitic rocks to form magmasof K20 contents of the source region are similar to that of granitic compositionhave been discussedby a number of average shales, while our "best estimate" for the U authors [Hyndman, 1981; Thompson, 1982; Vielzeuf and concentrationis approximately 3 times that in the average Holloway, 1988]. Figure 9 illustratesthe approximateP-T post-Archeanshale [Taylor and McClennan,1985]. loci of water present and water absent liquidii for pelitic The effects of these heat production rates on crustal compositions,along with granite solidii and approximate temperaturescan be inferred from the study of Molnar et al. steady state geotherms for the Ulugh Muztagh region. [1983]. With slight modification of their results,this study M'mimumtemperatures necessary to producepartial melts are can be used to determine the thermal perturbationsof thin ~920 K if the systemis saturatedwith externallyderived horizons of uraniferous rocks. Thermal effects of discrete water and 1120 K for the more likely scenario of "fluid- horizons of radioactive material can be calculated from the absentmelting" of a systemwherein all fluid is suppliedby shear heating calculations of Molnar et al., exchangingthe dehydrationof hydrousphases such as muscoviteand biotite productAoD (heat productivityfrom radioactivedecay times [Thompson, 1982; Vielzeuf and Holloway, 1988]. The layer thickness) for c•v (resolved shear stress times steadystate geotherms shown in the diagramare constrained velocity). While these substitutionsappear ad hoc, both are by (1) the crustalthickness of 65 :l: 5 km [Molnar, 1988], suggestedby Molnar et al. [1983]. (2) TMoho <1300 K [Molnar, 1988], and (3) nominalcrustal Temperatureperturbations due to in situ radioactivedecay heatproduction rates of 0-1 x 10-6 W/m3. Thisillustration (ATr) in isolated horizons, using the parametersof Table 7, demonstratesthat temperaturesnecessary to partially melt are functions of both the thickness and volumetric heat pelitic rocks can be attained within the thickened crust of production within the layer. For heat productionrates in the Tibetan Plateau, with a wide variety of plausible heat Table 6, and a radiogeniclayer 10 km thick, ATr within this sources. horizon range from 100 to 250 K some 20 Ma after the layer For the minimumgeotherm modeled, where the assumption was created. These A Tr estimates are of course very of no heat productionwithin the crusthas beenmade, the sensitive to the assumed parameters; the uncertainty in k geothermintersects the dry liquidusat -1100 K and70 km (thermal conductivity,here assumedto be 2.1 W/m-K [Pinet depth;addition of heatby in situ decayof radioactivenuclei and Jaupart, 1987] is particularly large. Increasingk will (100-200 K) produceslocal intersectionof the liquidusat lead to lower ATr becausethe heat will be transportedfrom temperaturesand depths of lessthan 1050 K and40 km. A the source at a greater rate. Nonetheless, this procedure geothermdetermined with an average volumetric heat demonstrates that in situ radioactive decay may have productionrate of 1 x 10-6 W/m 3 in thetop 30 km of crust contributedsignificant quantities of heat to the sourceregion intersectsthe dry liquidus at 1000 K and ~30 km depth;

TABLE7. TemperaturePerturbations Time, Ma* AoD,W/m 2 10 20 30 40 z•Tr (•0 0.017 73 95 105 110 0.039 170 220 240 255 0.046 205 265 290 305

Z•Tr (30 •n)? 0.017 78 105 120 •35 0.039 183 245 285 305 0.046 220 295 340 380 D=10km; Ao-l.7 x 10-6, 3.9 x 10-6 and4.6 x 10-6 W/m3;•- 2.1W/m-K; •-- 1x 10-6 mY/s. Heat productions arecalculated from Figure 5 of Molnar et al. [1983]. *Elapsed time of decay. tTemperatureincrease (K) in the centerof a 10-km-thicklayer due solely to heatingby in situradioactive decay, given the Ao of Table6 anddepth below surfaceof 20 (upperset) and 30 km (lowerset). Otherparameters are givenabove. MCKENNA AND WALKER: G•MISTRY OF LECOUCRATIC IGNEOUS ROCKS 21,497

Temperature (K) 300 500 700 900 1100 0

-. ' --. -'q::i'•::i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:?,•' A=0"., A=I .....

15 0 "- -

• 10 " 0 ATr(K)200 B•e ofContinental Cr•t "'• Fig. 9. Summa• diagramof the •e•d stmctu• of •e TiPton Plateau•d dep•-temperatu•relations of •litic •d graniticliquidii, adapted• paa from Hyndman [1981], Thompson[1982] and Vielzeuf and Holloway[1988]. Tem•rature is given • Kelv•and p•ssu• • 108Pa (=kba0. •i•er l•es show l•ati• of •uminosflicatestab•ity fields (A=andalusite,K=kyanite, S=s•anite). Steadystategeothe•s we• constructedus•g cu•nt crustal•ic•esses (65 2 5 •) •d Moho t•ratures (g13• K) [Molnar, 1988],•d n•• h•t productionof0 and1 x 10'6 W• 3 (straightand cu•ed l•es labeledA=0 and A=I, respectively). •rge, shadedtriangle illustratesl•ely temperature•creases • sourceregi• of Ulugh Mu•agh magmasdue to • sireradioactive decay (see Tables 7 •d 8). •e tem•ramre increaseis a •nction of •e •ickness of •e sourcelayer; •is •c•ase is given, at a s•cdic •ickness indicatedby the veaical l•b of the triangleby •e dist•ce from the verticall•b to the hypot•euse of the •gle. D• •d wet melt•g •nes •dicate the P-T l•ii of water saturatedand water absentmelting of pelitic sourceareas [Vielzeufand Holloway,1988]. L•ely P-T field for eraplacementof •e Ulugh Mu•agh •tmsive r•ks is stippled.

addition of heat from in situ decay allows temperaturesin central Tibetan Plateau. The approximate age and type of excess of 1200 K at depths as shallow as -20 km. These volcanism is shown in detail in Figure 10, adapted from the calculations show that superposition of reasonable Geological Map of the Tibetan Plateau [Ministry of Geology geotherms for the Tibetan Plateau and temperature and Natural Resources, 1980]. Two features of interest in perturbations from radioactive decay can produce the this map are the latitudinal zonation of the chemistry of the temperaturesnecessary to partially melt even dry pelitic volcanics and the sharp, linear northern and eastern edges of source rocks and to produce large volumes of melt. The the province. This northern front strikes obliquely relative upper temperaturelimits derived in Figure 9 are compatible to the edge of the plateau (as representedby the trace of the with the temperaturerange of the experimentsof Vielzeuf Altyn Tagh Fault), but is subparallel to, and 100-50 km and Holloway [1988], demonstrating that the pressure- north of, the trace of the Jinsha Suture. Ulugh Muztagh is temperature-compositionrelationships within the source area found along the westernsection of the northernedge of the for the Ulugh Muztagh intrusive and extrusive rocks are province. sufficient to producean in situ partial melt. Deng [1978] reportedthe resultsof a reconnaissancefield trip, detailed optical petrography, and major element REGIONAL RELATIONSHIPS AND TECTONIC IMPLICATIONS chemistryof Quaternaryvolcanic rocks exposedsouth of The recent magmatismin the Ulugh Muztagh area is a part Ulugh Muztagh. He separatedthese flows into three groups: of a regionally extensive province of post-Middle Miocene the southernBamogiongzong, central Yongbohu and the vulcanism,which covers an elliptical area at least 300 km in northern Qiangbaqiansequences. As shown in Figure 10, extent north-south and 600 km east-west within the north- these units lie 200, 100, and 50 km, respectively, south of 21,498 MCKENNAAND W•: G•STRY OFLECOUCRATIC IGNEOUS ROCKS

+ + • + + -• + +30øN 82ø • 86ø 9• oLhasa 94øE zsz

Fig.10. Simplifiedgeologic map of post-EarlyMiocene volcanic rocks on the Tibetan Plateau, modified andadapted from The Geologic Map of theTibetan Plateau [Ministry of Geologyand Natural Resources, 1980]and [Coulon etal., 1986].Rock types and approximate agesare shown by shading: 1,Neogene andesiticvolcanic rocks; 2, Plioceneto Recentandesitic volcanic rocks; 3, undffferentiatedandesitic volcanicrocks; 4, potassicPliocene to Recentandesitic volcanic rocks; 5, Plioceneto Recentbasic volcanic rocks;6, potassicPliocene to Recentbasic volcanic rocks; 7, undffferentiatedbasic volcanic rocks. Unshadedareas are older, undifferentiated volcanic rocks. The central Tibetan Plateau volcanic field is shown in thecenter of the figure, dotted lines separate areas of distinct rock type. The Altyn Tagh Fault forms the northernstructural and topographic front of thePlateau; note the angular discordance between this structure andthe E-W trending northern boundary of the volcanic province and compositional subprovinces. FollowingMolnar, [1988], observe that the volcanic province islocated over an area of anomalouslyhot uppermantle; it appearsfrom the data of Deng, [1978]; and Pearce and Mei, [1988] that units 5, 6, and7 are formedby partial melting of enrichedmantle. The band of unit 2 (whichincludes sample TQ 3) between UlughMuztagh and TQ 2 representhybrid magmas formed through mixing of crustallyderived mantle melts similarto thoseexposed at Ulugh Muztagh and mantle melts similar to thoseexposed at TQ 2. Fault ornamentationas in Figure 1. the volcanicunits exposed at UlughMuztagh and lie on the trachydacitesto subalkalicrhyolites [Deng, 1978; Pearce and northernsection of the Qiangtangterrain. The southern Mei, 1988]. Theage of all of theseunits was suggested by sequenceincludes ultrapotassic rocks varying from tephrite Deng [1978]to be Quaternary,based on relationshipswith basanitesto phonolites. Typical phenocrystsinclude underlyingrocks of assumedPlio-Pleistocene age. The silica leucite, analcite,nephiline, nosean, olivine (Fo75) and andpotash concentrations of thesesamples agree reasonably aegerine-augite. The central sequenceare transitionalto well with the compositions expected from their unit calc-alkaline tephrite basanitesto trachyandesitesand identificationon the GeologicalMap of the Tibetan Plateau. dacites,while the northerngroup are typicallycalc-alkaline In that map's terminology,and in the terminologyfollowed MCKENNAAND WALKER: GEOCHEMISTRY OF LECOU••C IGNEOUSROCKS 21,499 in Figure 10, samplesTQ1, 2 and 3 are ultrapotassicbasic While we recognizethe potentialdanger of drawing rocks (unit 6 in Figure 10), ultrapotassicandesitic rocks conclusionsfrom sucha small data set, the GeologicalMap (unit 4), and intermediaterocks (unit 2), respectively. The of the TibetanPlateau suggests that the patterns seen in the variation in rock type recorded by Deng [1978] in the datadiscussed here may well be representativeof patterns in Yongbohuvolcanics may indicate that the central unit 4 the centralvolcanic province as a whole,and are certainly provincein Figure 10 includesa range of lithologiesfrom testablewhen additional data become available. These basanite to andesRe. resultsare compatiblewith a modelfor Pleistocene-Recent One samplefrom eachof the sequencesidentified by Deng magmatismalong the northern Qiangtang terrain which is [1978] was re-analyzedby Pearce and Mei [1988]. As noted dominatedby mixingof twoend-member compositions. The by Pearce and Mei [1988], the TQ samples all show silicarich end-member(represented by the UlughMuztagh extraordinaryenrichment of the REEs and the LIL elements extrusives)forms through in situpartial melting of pelitic (Rb, St, Th) indicative of derivationfrom a highly enriched rockswithin the thickenedTibetan crust. The mafic end- source area, due perhaps to assimilationof a subduction member(represented by the TQ 2, the Yongbohusequence) component [Pearce and Mei, 1988]. Despite this forms through partial melting of enriched, perhaps enrichment,all three samplesdemonstrate a consistenthigh metasomatized subcontinental mantle, and occurs, as shown field strength elements depletion [Salters and Shimizu, in Figure 10, in a band 50 km thick betweenthe southern 1988], a characteristic not atypical of subcontinental ultrapotassicand the northernhybrid provinces. These end- lithosphere. The high MgO contentsand Mg numbersfor members occur approximately 100 km apart, while the the basanite indicate that these magmas are direct mantle hybridmagma lies 50 km southof Ulugh Muztagh. Most melts, apparentlyextruded through the 65 km thicknessof likely, neither of the chemical end-membersare point the plateau. sources;we do not envisionveins of magma stretching50 Basedon the availablemajor and traceelement data for the km to mix together in the central belt. Rather, we TQ rocks,we suggestthat the volcanicrocks exposed in and hypothesizethat partial melting of the crust, and hence southof Ulugh Muztagh are related by magmamixing: the production of the silicic end-member,occurs laterally end-membersare representedby the high silica crustalmelts throughoutthe crust. To the north, unadulteratedcrustal exposedat Ulugh Muztagh and the basanitesamples of the meltsreach the surfaceowing to a lack of mantlemelt, while centralYongbohu province south of Ulugh Muztagh (Figure in the Qiangbaqian area, subequal volumes of the end 11). Mass balance calculationsindicate that the analyzed membersare present. In the Yongbohu area,an entiresuite samplefrom the northernQiangbaqian sequence (sample TQ of basaniteto rhyolite indicatesthe presenceof both hybrid 3) formedfrom approximatelyequal fractionsof the two end- liquidsand unadulteratedmantle melts. Finally,south of the members. The major element chemistry of the Yongbohubelt, a wide areaof ultra-potassicrocks represents Bamogiongzongsequence, as representedby sampleTQ I of an area of hybrid liquids stronglymodified by low-pressure Pearce and Mei [1988], apparenfiy formed by fractional fractionation,as suggestedby sampleTQ 1. crystallizationof hybrid melts. A numberof studies(see Molnar [1988] for a review) have

Tibetan Plateau Tarim Basin TQ2 UM TQ1 TQ3

Fig. 11. A sketch of the sourcesand mixing areas of magmas in the central Tibetan province, modified from Molnar, [1988]. The areas of mantle upwelling are shown by arrows; approximatelocations of Ulugh Muztagh, TQ 3 and TQ 2 are also shown. Silicic magmasare producedthrough crustal melting of pelitic or perhapsamphibolitic material over a wide region of the plateau (N-dipping hatching), while manfie-derived melts occur only southof the area of active upwelling (S-dippinghatching). The resultinghybrid magmas can cover wide areas of the central plateau, but the northern terminusof the marie magmas should indicate the northernterminus of the upwelling. If so, data for the Ulugh Muztagh extmsive rocks indicatethe length of the boundarybetween eroded and unerodedlithosphere is of the order of 100 km. 21,500 MCKENNA AND WALKER: GE•MISTRY OF LECOUCRATICIGNEOUS R• demonstratedthe existence of a high impedancezone in the the time-integrated unroofing rate for the Ulugh Muztagh upper mantle below the north-central Tibetan Plateau, the region is less than 1.25-2 mm/yr. Interestingly,Molnar et same area occupied by the volcanic province. Molnar al. [1987] estimated that thrust faults along the northern [1988] hypothesized that this anomaly was due to the edge of the Tibetan Plateau have vertical slip rates of 1-2 presenceof relatively hot, perhaps partially melted, mantle mm/yr, in good agreementwith the crustal thickeningrates material driven convectively against the base of the Tibetan estimatedby examination of the Ulugh Muztagh igneous continental lithosphere. This thinning of the mantle rocks. lithosphere beneath the central Plateau has produced the lateral thermal gradients resolved by teleseismic studies. CONCLUSIONS Molnar also suggestedthat the approximate correspondence of the volcanic province with the thermal anomaly was The granites, granodiorites,rhyolites, and trondhjemites evidence of a causal relationship. collected from the Ulugh Muztagh region of the northern The large-scale mixing observed in the volcanic rocks Tibetan Plateau are leucocratic, potassic, primarily south of Ulugh Muztagh confirms this model and allows peraluminousrocks which formed from crustally derived some detail to be added to it. The geochemicalgradient partial melts. These melts were producedat depthsof 20-40 between the central Yongbohu sequence and the crustally km by heating of the source region through in situ derived melts is causedby the thermal gradient imposedby radiogenic decay and mantle heat flux into thick, and hence mantle convection. The mantle-derived melts form south of insulative, continental crust. The igneous rocks currently the line at which hot, superadiabaticmaterial is juxtaposed exposedat Ulugh Muztagh form both extrusive and intrusive with continental lithosphere. North of this line, mantle bodies; regional contact metamorphicrelationships suggest temperaturesare too low, or lithosphericthicknesses too that the intrusiverocks were emplacedat pressuresless than 4 x 108 Pa. high, to allow formation of melts or their transportto the surface. If true, the distribution of volcanic rocks on the Burchfielet al. [1989]have reported 40Ar/39Ar data that plateau suggest that the width of the thinned zone of documentsa minimum age difference between the intrusive asthenosphereis approximatelyequal to the width of the and extrusive samplesof 4-6.5 Ma. The younger rhyolitic extrusive rocks have major element abundances, trace mixing area defined by the Ulugh Muztagh and Yongbohu 87 86 sequences,or approximately100 km. This would require elementabundances, and St/ Sr and206pb/204pb ratios significantlateral thermal and mechanicalgradients in the similar to those of the intrusive rocks. Although the mantle below the plateau. extrusives appear to be significantly younger than the If the northward limit of Cenozoic volcanism does follow intrusives,the chemical data suggestthat both the extrusive and the intrusive rocks considered here were derived from the subcrustal thermal anomalies, the angular discordance between this anomaly and the topographicand structural same, or very similar, source rocks of pelitic composition. front of the Tibetan Plateauhas importantimplications. The This history of crustally derived magmatism requires topographicrelief (from the westernmostexposure to the maintenanceor episodic attainment of super-soliduscrustal easternmostexposure of mafic melts, see Figure 10) along temperaturefor time scales of 5 to 10 million years. The the northern limit of volcanism is over 2 km. If the depth of emplacementof the intrusiverocks and the inferred northern terminus of the mafic rocks maps the thermal depth of the source rocks suggests that rates of crustal anomaly within the mantle, the subcrustalthermal structure thickening and unroofing in the north-central Tibetan is clearly not affecting the formation of the plateau. In Plateau may be approximately equal, at 1-2 mm/yr. These otherwords, the mantle-crustsystem in centralAsia appears rates are similar to those estimatedby Moltmr et al. [1987] to be decoupled,with the mechanismsof crustalthickening from ground level reconnaissanceobservations and suggest operating independently, but synchronously,with the that the north central Tibetan Plateau may be in a steady mechanismsaffecting the mantle. If a link between the state condition, with the rates of crustal thickening mantle thermal structure and the creation and maintenance of approximatelyequal to the rate of unroofing. the plateau's elevation does exists, it appears to be Considerationof the extrusivesamples at Ulugh Muztagh operatingon time scales longer than that recordedin the and other Pliocene to recent volcanic rocks south of Ulugh upperCenozoic volcanic rocks present in the centralplateau Muztagh suggeststhat large-scale mixing of crustal melts that are examined here. and mantle derived melts is occurring in the north central The time scale and rate of thickeningin this part of the Tibetan Plateau. The east-westtrending compositional zones plateaucan be estimated from the 40Ar/39Ar geochronology are due to similarly trending thermal gradientsin the upper resultsof Burchfielet al. [1989], alongwith the estimatefor mantle below the plateau. The oblique angle between the maximumdepth of emplacementfor the intrusiverocks (4 x subplateauthermal structure and the Tibetan Plateau itself 108Pa, or approximately10-12 km). Theminimum age of indicates that the mechanisms of crustal thickening may crystallizationfor the intrusiverocks is providedby the operate independently of mechanisms controlling the 10.5 Ma muscoviteplateau age for UBTG; the Sr isotopic thermal and mechanical structure of the mantle. data presentedabove suggest possible crystallization ages of APPENDIX 1 11-12Ma. The intrusivesuite was exposed at the surfaceby 4.0 Ma, as the well-dated extrusive rocks overlie a boulder MV2, UMVU and UM3V are rhyolitic tuffs characterizedby glomerophyricplagioclase (An20_25) and sanidine,phenocrystic conglomerate which contains abundant clasts of the intrusive quartz (up to 7 mm dianemr)and biotite, and subhederalto anhederal rocks. Thus the maximum averageunroofing rate between cordieritein a microliticgroundmass. CordierRe is surroundedby a 10-12 Ma and 4 Ma is 10-12 km/6-8 Ma=1.25-2 mm/a. The felty, colorlessrim 10-30 •.m thick, and locally has altered to actualunroofing rate may well have exceededthis average pinite. Zircon includedin cordieritecrystals are anhederal,broken rate for shorttime intervals;our estimatemerely statesthat grains,with no mandes. Sanidineand biotite separatesfrom MV2 MCKENNAAND WAIXER: GEOCHEMISTRYOF LECOUCRATIC IGNEOUS ROCKS 21,501

APPENDIX 2: REE Corrections From U Fission

Extrusives Intrusives Potasslum-poor Element MV1B UM1B MV2 UM3V UMVU UMQP QID BKSP UBTG 2MGR KSPO QIL UM10 ls 1.03 1.03 0.72 0.69 0.63 0.47 0.33 0.57 0.46 0.3 0.54 0.25 0.27 oLa* 0.24 0.2 0.1 0.15 0.1 0.06 0.04 0.12 0.1 0.05 0.07 0.03 0.04 Ce 7.02 7.02 4.91 4.7 4.27 3.19 2.24 3.89 3.13 2.08 3.67 1.73 1.86 oCe 1.62 1.35 0.68 1.03 0.7 0.43 0.3 0.84 0.65 0.32 0.49 0.22 0.24 Ni 5.46 5.46 3.82 3.65 3.32 2.48 1.74 3.02 2.44 1.62 2.86 1.34 1.45 oNd 1.4 1.18 0.6 0.89 0.62 0.39 0.27 0.73 0.56 0.29 0.44 0.19 0.22 Sm 0.1 0.1 0.07 0.07 0.06 0.05 0.03 0.06 0.05 0.03 0.05 0.03 0.03 •Sm 0.04 0.03 0.02 0.02 0.02 0.01 0.01 0.02 0.02 0.01 0.02 0.01 0.01 Valuesin the tableare the concentrations(in partsper million) of elementsproduced in the samplesby inducedU fission[Korotev and Kindstrom,1985]. Thesevalues were subtracted from the INAA resultsto givethe true concentrations of the elementsreported in Table2. * Fully propagated2s uncertaintiesin the corrections,in partsper million. wereanalysed for39Ar/40Ar. The sanidine hada plateauage of 4.0 The isotropicgroundmass consists of quartzand plagioclasewith _+0.1 Ma (72% of gas in threesteps, total gas age was 4.1 Ma), and diameters less than 30-50 mm. In UM10, faces of tourmaline and biotite gave a total gas age of 4.2 _+0.1 Ma. quartzare roughon the scaleof 30-50ram,aparently due to reaction UMQP is a porpyritic rhyolite with quartz, sanidine, cordieritc of the grainswith the groundmass,while the facesof plagioclase and biotite phenoctystsin a microlitic groundmass. Tourmaline is grainsare quitesharp. In QTL, the plagiodasegrains display this present in trace quantities,and althoughtypically of small size is roughtexture, while the quartz and tourmaline faces are sharp. Trace clearly a phenocrysticphase. The isotropic groundmassconsists of quantitiesof zircon (cloudy,anhederal) are presentwithin the tourmaline. 10-50 •tm diameter quartz+plagioclase+sanidine(70%), opaques KSPO is a rathercoarse grained granodiorite with phenocrystsof (20%) and biotite (10%) alkali feldspar(2-15mm in length),quartz and plagioclasewith MV1B and UM1B are glassy rholitic flows with porphyritic to interstitial,fine-grained biotite and muscovite. Lathsof allanitc locally glomerophyricplagioclase and porphyrititc sanidine,quartz, occurboth as inclusions in otherphases and as interstitialgrains. cordieritc and biotite in a glassy matrix. Plagioclase is locally In the formerenvironment, the grainsare stubyand anhederal, while fluid-inclusion rich, individual laths are up to 4-5 mm in length. the interstitialgrains are long, thin, optically clear laths upto Cordieritc grains are typically smaller in size than the plagioclases, and some display overgrowths of cordieritc around central, zoned 5001•min length. grains. A sanidine separate from MV1B had a 4.0 _+ 0.1 Ma Acknowledgments. We would like to thank Drew Coleman for 39Ar/40Arplateau age (five steps, 100% of the gas). XRF preparation;S. R. Hart for the use of the Geoalchemylab at BKSP and UBTG are coarsegrained, panidiomorphic granites with MIT; W. R. Van Schmus and M. E. B ickford for consultationat the phenocrystsof perthitic, poikolitic alkali feldspar and quartz with XRF and mass-specfacilities at the University of Kansas; and F. smaller phenocrysts of plagioclase (An25.45) and biotite in a Frey for use of, and P. Ila for her help in all aspectsof, the INAA medium grained, equigranulargroundmass of quartz, alkali feldspar facility at MIT. Ye Hongzhuankindly translatedmap legends and and biotite. Zircon inclusionsin biotite included clear to very dark tables for us: his assistancewas crucial to our understandingof ideas colored grains all less than 50 mm long. Clear grainsare euhederal presentedhere. Kevin Furlong wrote the program"Geotherm" used and show no radiation haloes, dark grains are anhederal and are in producing Figure 9. B. Nelson and an anonymousreviewer surroundedby radiation-damagehaloes. Three mineral separateswere providedexcellent reviews of this manuscript;B.C. Burchfiel, P. Le analyzedby39Ar/40Ar. Muscovites gave a 10.5+ 0.1Ma plateau Fort, K. Burke and V. Salters read early versions of the manuscript (78% of the gas in two steps), biotite a total gas age of 10.1 + 0.1 and suggested improvements to it. Although they may not Ma and sanidine displayed a minimum age of 9.1 Ma with a total necessairily agree with any of the ideas presented herein, we gas age of 10.2 + 0.1 Ma. appreciate their efforts. The paper was typeset by Elizabeth 2MGR is a medium grained hypidiomorphicgranular, two-mica- Spizman. Financial supportprovided by the Student ResearchFund granite. Quartz grains typically show undulatory extinction, Committee of the Department of Earth Atmospheric, and Planetary muscovite locally shows very minor kinking. Mucovite has very Sciencesat MIT (L.W.M.), the National Science Foundation (EAR few inclusions of any sort except for rare zircon, while biotites 8805125 to S. R. Hart, EAR 8517889 to W. R. Van Schmus) and contain abundant(a few volume percent) opaqueinclusions. Zircons the Shell Oil Company (J.D.W.). occurs as both inclusionsin other phasesand as interstitial grains; in both environments the zircons are irregular, anhederal and optically cloudy. A biotite separate from this sample gave an 39Ar/40Arage of 10.0+ 0.1Ma (totalgas), while a potassiumAnders, E., and M. Ebihara, Solar-system abundances of the feldspar separatedisplayed a 9.8 + 0.1 Ma total gas age with a elements, Geochim. Cosmochim. Acta, 46, 2363-2380, 1982. minimum of 8.9 Ma. Backstrom, H., and H. Johanssen,Geology, in Scientific Results of QTD consistsof medium grained,porphyrititc quartz, orthorase a Journey in Central Asia 1899-1902, editedby S. Hedin, Vol. 6, and plagioclase in a fine grained, microlitic groundmass. pt. 2, Stockholm, 1907. Tourmaline is present in trace quantities - these grains generally Burchild, B.C., P. Molnar, Z. Zhao, K. Liang, S. Wang, M. Huang, contain abundant inclusions of zircons. Locally poikiolititc and J. Sutter, Geology of the Ulugh Muztagh Area, Northern sanidinegrains are less than 5 mm in length and containquartz and Tibet, Earth Planet. Sci. Lett., 94, 57-70, 1989. very rarely allanitc inclusions. A potassium feldspar separate Chang,C., et al., Preliminaryconclusions of the Royal Societyand analysedby39Ar/40Ar returned a total gas age of 8.4 + 0.1Ma with Academia Sinica 1985 Geotraverse of Tibet, Nature, 323, 501- 507, 1986. a minimum age of 8.1 Ma. Coulon, C., H. Maluski, C. Bollinger,and S. Wang, Mesozoicand UM10 and QTL are unusual rocks. Both samples consist of phenocrysticquartz and plagioclase,typically 5-10 mm in length, Cenozoicvolcanic rocks from central and southern Tibet: 39Ar- in a fine grained microcrystalinegroundmass. Minor tourmaline 40At dating,petrological characteristics and geodynamical (schod) occurs as pheocrysticlaths less than 4-5 mm in length. significance,Earth Planet. Sci. Lett., 79, 281-302, 1986. 21,502 MCKENNAAND WALKER: GEOCItEMISTRY OFLECOUCRATIC IGNEOUS ROCKS

Deng, W., A preliminarystudy on the petrologyand petrochemistry crustal evolution, I, Australian post-Archeansedimentary rocks, of the Quaternary volcanic rocks of northern Tibet autonomous Geochim. Cosrnochim.Acta, 40, 1539-1551, 1976. region (in Chinese), Acta. Geol. Sinica, 52, 148-162, 1978. Nash, W. P., and H. R. Craecraft, Partition coefficients for trace Faure, G., Principles of Isotope Geology, 587, pp., John Wiley, elements in silicic magmas, Geochim. Cosmochim. Acta, 49, New York, 1986. 2309-2322, 1985. Harris,N. B. W., R. Xu, C. L. Lewis,and C. J'in,Plutonic rocks of Nekvasil, H., Calculation of equilibrium crystallizationpaths of the 1985Tibet Geotraverse,Lhasa to Golmud,Philos. Trans. R. compositionallysimple hydrous felsic melts, Am. Mineral., 73, Soc.London Ser. A, 327, 145-168, 1988. 956-965, 1988. Hart,S. R., andC. Brooks,The geochemistry andevolution of early Nichols, J., The statistics of Pearce element diagrams and the Precambrianmantle, Contrib. Mineral. Petrol., 61, 109-128, Chayes closure problem, Contrib. Mineral. Petrol., 99, 11-24, 1977. 1988. Hyndman,D. W., Controlson sourceand depth of emplacementof Norin, E., Geological Explorations in Western Tibet, in Reports graniticmagmas, Geology, 9, 244-249, 1981. from the scientific Expedition to the NorthwesternProvinces of •a, P., andF. Frey,Utilization of neutronactivation analysis in the China under the leadership of Dr. Sven Hedin, Publ. 29 (III), studyof geologicmaterials, Atomkernenerg. Kerntech., 44 S, Geology 7, Tryckeri Aktiebolaget, Thule, 205 pp., Stockholm, 710-716, 1984. 1946. King, R. W., R. W. Kerrich,and R. Dadar,REE distributionsin Pearce, J. A., and H. Mei, Volcanic rocks of the 1985 Geotraverse: tourmaline:An INAA techniqueinvolving pretreatment by B Lhasa to Golmud, Philos. Trans. R. Soc. London Ser. A, 327, volatilization,Am. Mineral., 73, 424-431, 1988. 169-201, 1988. Korotev,R. L. and D. J. Lindstrom,Interferences from fissionof Pinet, C., and C. Jaupart, A thermal model for the distributionin 235Uin I•AA of rocks,Trans. Amer. Nuclear Soc., 49, 177-178, spaceand time of the Himalayangranites, Earth Planet. Sci. Lett., 1985. 84, 87-99, 1987. Le Fort,P., Himalayas: The collided range, Present knowledge of the Salters,V., and N. Shimizu, World-wide occurrenceof HFSE-depleted continentalare, Am. J. Sci., 275, 1-44, 1975. mantle, Geochim. Cosmochim. Acta, 52, 2177-2182, 1988. Le Fort, P., Manasluleucogranite: A collisionsignature of the Taylor, S. R., and S. M. McClennan,The ContinentalCrust: Its Himalaya,A modelfor its genesis and emplacement, J. Geophys. Compositionand Evolution,312, pp., BlackwellScientific, Res., 86, 10545-10568, 1981. Boston, Mass., 1985. Lindstrom,D. J., andR. L. Korotev,Teabags: Computer programs Thompson, A. B., Dehydrationmelting of Pelitic rocksand the for instrumentalneutron activation analysis, J. Radioanal.Chem., generationof H20-undersaturatedgranitic liquids, Am. J. Sci., 70, 439-458, 1982. 282, 1567-1595, 1982. Littledale,S. G. R., A journeyacross Tibet from north to southand Turcotte,D. L., andG. Schubert,Geodynamics Applications of westto Ladak,Geogr. J., 7, 453-465, 1886. ContinuumPhysics to GeologicalProblems, John Wiley & Sons, Miller,C., Arestrongly peraluminous magmas derived from pelitic New York, 450 pp. sedimentarysources?, J. Geol.,93, 673-689, 1985. Vidal,P., A. Cocherie,and P. Le Fort,Geochemical investigations Ministryof Geologyand Natural Resources, Geological map of the of the originof the Manasluleucogranite (Himalaya, Nepal), Qinghai-Xizang(Tibet) Plateau (1:1.5 Million scale), Ministry of Geochim.Cosmochim. Acta, 46, 2279-2292, 1982. Geologyand Natural Resources, Beijing, 1980. Vielzeuf,D., andJ. R. Holloway,Experimental determination of the Molnar,P., A reviewof geophysicalconstraints on the deep fluid-absentmelting relations in the pelitic system,Contrib. structureof theTibetan Plateau, the Himalaya, and the Karakorum Mineral. Petrol., 98, 257-276, 1988. andtheir tectonic implications, Philos. Trans. R. Soc.London, Zartman,R. E., and B. R. Doe, Plumbotectonics--Themodel, Ser. A, 326, 33-88, 1988. Tectonophysics,75, 135-162, 1981. Molnar,P., W.-P. Chen, and E. Padovani,Calculated temperatures in overthrustterrains and possiblecombinations of heat sources responsiblefor thetertiary granites in thegreater Himalaya, J. L.W. McKenna, J. D. Walker, Dept. of Geology,Univ. Kansas, Geophys.Res., 88, 6415-6429, 1983. Lawrence, KS 60045. Molnar,P., B.C. Burchfiel,L. K'uangyi,and Z. Ziyun,Geomorphic evidencefor activefaulting in the AltynTagh and northern Tibet andqualitative estimates of its contributionto the convergence of (ReceivedSeptember 13, 1989; India andEurasia, Geology, 15, 249-253, 1987. revised3anuary 17, 1990; Nance,W. B., andS. R. Taylor,Rare Earth element patterns and accepted3anuary 23, 1990)