American Mineralogist, Volume 73, pages 313-323, 1988

Origin and significance of blue coloration in quartz from Llano rhyolite (llanite),north-central Llano County, Texas

MrcHl.Br E. Zor,nNsxyPlanetary Materials Branch, SN2/NASA Johnson Space Center, Houston, Texas 77058, U.S.A.

P.q.uL J. Syr-vnsrBn*ELINASA Headquarters, Washington, D.C. 20546, U.S.A.

Jtrms B. PacnsGeology Department, Michigan Technological University, Houghton, Michigan 49931, U.S.A.

Assrn,A.cr

Blue quartz phenocrysts from the Llano rhyolite (llanite), Llano County, Texas, derivetheir coloration from Rayleigh scattering by abundant submicrometer-sized (type l) il-menite inclusions. Also included are larger, less abundant ribbon-shaped (type 2) ilmeniteinclusions lying on the rhombohedral growth surfaces of the host q\artz. These type 2inclusions produce chatoyance in certain orientations; however, they are, in general, in-dividually too large (-0.1 by I by 20 pm) to contribute to the blue color by Rayleighscattering. The total amount of ilmenite in the llanite blue quartz is calculated to be -0.02volo/0.

Llanite blue quartz and groundmass exhibit distinct trace-element crystal/liquid parti-tion coefficients that deviate from the flat patterns characteristic of other quartzlrhyolitepairs. Partition coefrcients for Hf (0.335),Zr (0.38), Cr (0.10), and Lu (0.28) are signifi-cantly greater than those for Rb (0.01), Ba (0.013), Th (0.004), La (0.003), and Tb (0.008),suggesting that a majority of the ilmenite inclusions crystallized from the llanite melt. Thisconclusion assumes that quartz equally partitions all elements except Eu, which in thellanite blue quartz as well as some colorless quartz exhibits partition coefficients with apositive anomaly (Eu/Eu* > 2.6).The trace-element data are compatible with either ofthe following scenarios: (1) both type I and type 2 ilmenite inclusions originated by crys-tallization from the llanite magma, or (2) the volumetrically major portion of the ilmeniteinclusions (probably type l) was derived by crystallization from the magma, whereas theremainder (probably only type 2) was formed by exsolution processes.

If entrapment of early crystallizing ilmenite is a generally applicable model for the originof blue quartz, two implications arise. First, calibrations of the temperature and oxygenfugacity of Fe-Ti oxide exsolution in blue quartz should not be made unless an exsolutionorigin for the inclusions is assured. Second, the dominant occurrence of blue quartz irrrocks of middle to late Proterozoic age may reflect preferential conditions that promotedearly ilmenite saturation during this time. These conditions remain largely undeterminedbut could involve particular physical parameters of magma equilibrium such as low tem-perature, high pressure, or high oxygen fugacity and/or processes of magma productionresulting in Ti-, Fe-, alkali-, and REE-rich high-silica compositions.

INrnoluc.troN

Quartz crystals exhibiting blue coloration are encoun-tered variously in granites, granodiorites, rhyolites, char-nockites, as silicic segregations within anorthosites, andin the regionally metamorphosed products of all theserocks (cf., Delong and Long, 1976;Herz and Force, 1984;Clarke, 1984; McConnell and Costello, 1984). With fewexceptions (Shaw and Flood, l98l), blue quartz is re-

* Present address: Department of Geoscience, New MexicoInstitute of Mining and Technology, Socorro, New Mexico 87801,U.S.A.

0003-004x/88/0304-03 I 3$02.00 3 I 3

stricted in occurrence to rocks ofPrecambrian, and par-ticularly middle to late Proterozoic age. The Proterozoicoccurrence of blue quartz as crystals is in contrast to bluishto gray cryptocrystalline silica, which is unconstrained inage, and with which this paper is unconcerned.

The origin of blue coloration in quartz crystals has longbeen known to result from the scattering oflight. Severalmechanisms have the potential to cause this scattering.Two less-cited mechanisms, strain deformation and Ray-leigh scattering from fluid inclusions, have been proposedas the agent for blue coloration for several occurrences(e.g., Milford granite, Emerson and Perry, 1907). How-

3t4 ZOLENSKY ET AL.: BLUE QUARTZ FROM LLANO RHYOLITE

ever, in the majority of instances, blue coloration in quartzis believed to arise by Rayleigh scattering from submi-crometer-sized solid mineral inclusions (e.g., Robertson,1885; Iddings, 1904; Jayaraman, 1939). The literatureconcerning the identity of the included phases is enor-mous, and only the more conclusive (and contradictory)studies will be summarized. The blue quartzes of theRoseland district, Virginia, have been examined by Rob-ertson (1884, 1885) and, a century later, by Nord (citedin Herz and Force, 1984), who attributed the colorationto Rayleigh scattering from submicrometer-sized crystalsof an iron-titanium oxide (probably ilmenite). In thecourse of an examination of the same material, Watsonand Beard (1917) concluded that the coloration was dueto crystals of rutile. Indeed, most early workers concluded(assumed?) that rutile was the cause of blue coloration inquartz (Robertson, 1885; Postelmann, 1937; Jayaramary1939), and this view was promulgated by later authorities(Goldschmidt, 1954; Frondel, 1962). Nevertheless, Par-ker (1962) described inclusions of tourmaline within bluequartz from the Wind River Range, Wyoming.

An unusual controversy surrounds blue quartz fromLlano County, Texas. In the initial report ofthis occur-rence, Iddings (190a) concluded that the blue colorationin this material was probably due to apatite and also pos-sibly ilmenite. Later workers, apparently misreading Id-ding's paper, stated that he reported rutile (e.g., Jayara-man, 1939). Finally, in a recent re-examination of theLlano material, Barker and Burmester (1970) concludedthat the blue coloration was due to zircon!

This lack of consensus on the included mineral oc-curred largely because the identification of microscopicto submicroscopic inclusion grains was frequently madeby visual observation, rather than by modern analyticaltechniques. However, it is also probable that all bluequartzes do not contain the same mineral inclusions.

Another uncertainty is the mechanism by which theincluded phase in blue quartz was produced. Severalworkers (Niggli and Thompson, 1979; G. L. Nord, Jr.,cited in Herz and Force, 1984, and pers. comm., 1986;J. B. Thompson, pers. comm., 1986) have speculated thatthe inclusions in blue quartzes found in metamorphosedterranes may be a high-temperature exsolution feature,implying that blue quartzes are a potential geothermom-eter-oxygen barometer for metamorphic conditions. Nig-gli and Thompson (1979) also noted that blue quartz ishighly strained in most occurrences, suggesting that thedeformation of the quartz may promote the presumedexsolution process. A magmatic origin for the inclusionsin blue quartzes is also possible-i.e., the inclusions maybe a liquidus mineral phase subsequently entrapped inthe growing quartz crystal.

Thus, for most blue quartzes, there are at least threeoutstanding questions: (l) What mineral(s) forms the in-clusions? (2) Did this mineral crystallize from a magmaticliquid or exsolve during subsolidus cooling, metamorphicheating, or deformation? (3) And why is blue quartz largelyrestricted in occurrence to Proterozoic and older rocks?

In this paper, we attempt to answer the first two of

these questions for the notable doubly terminated blue

quartz phenocrysts from the Llano rhyolite, from Llano

County, Texas, locally called "llanite" (Iddings, 1904).

Llanite is a late Proterozoic, porphyritic, hypabyssal

rhyolite intrusion within the Llano uplift of central Texas.

The igneous minerals of this rhyolite are well preserved,

although petrofabric data (Clabaugh and McGehee, 1963)

and evidence of Rb-Sr redistribution (Delong and Long,

I 976) suggest that the rock has been affected by low-grade

metamorphism. The blue quartz from this material ex-

hibits a particularly beautiful blue coloration, which is

sky-blue in the center of each crystal and darker at the

margins (Iddings, 1904). In addition, in certain orienta-

tions these crystals display an intense silver-blue chatoy-

ant flash.The results of our study of llanite may be applicable to

the origin ofblue quartz in other felsic rocks, particularly

those that have suffered only low-grade metamorphism.

In this regard, we then attempt to answer the third out-

standing question concerning the temporal association of

blue quartz, as posed above.

ExpnnrrvrBNTAL PRocEDURE

Blue quartz phenocrysts were hand-picked from a partlyweathered outcrop ofLlano rhyolite located in a roadcut on theeast side of Texas State Highway 16 (lat 3053'22" N, Iong9839'23' W), approximately 17 km north of the Llano metro-politan area. Several fresh whole-rock specimens were also col-lected. The quartz phenocrysts were ground slightly in an alu-mina shatter box; and ultrapure (>99o/o) blue quartz separatewas extracted from this mixture by hand-picking under a bin-ocular microscope. Millimeter-sized chips of llanite microclinephenocrysts and matrix material were separated from the whole-rock samples by hand. The matrix chips were examined undera binocular microscope, and those with quartz or microclinephenocryst adhesions were rejected from the separate. The mi-crocline was ground and then purified by magnetic separationand further hand-picking. A standard doubly polished petro-graphic thin section was also made from the whole-rock sample.

Several thin chips of the blue quartz phenocrysts were mount-ed directly onto Be grids for observation in a rsoI- toocx scanningtransmission electron microscope (srrrnl). Six other chips of bluequartz phenocrysts were set into epoxy and thin-sectioned usinga Porter Blum r'rr-z Ultra-Microtome and then subjected to srur,ranalysis. Unfortunately, these sections averaged approximately400 A thick, which was thicker than desired, because ofthe highhardness of quartz relative to that of the enclosing epoxy, butwere still serviceable. All rnineral identifications were verifiedusing electron diffraction (eo). The diffraction data were cali-brated by employing TlCl powder on the specimen grids and areconsidered precise to within 50/0. Additional quartz gralns weredigested in lM HF and then heated; the residue was examinedin a leor- :sc scanning electron microscope (sru) and the srru.Energy-dispersive spectroscopy (ros) analyses were obtained ofall material examined by electron microscopy using PrincetonGamma Tech thin-window and windowless detectors and ana-lyzers attached to both the srrv and sBv.

Approximately 200 mg of the blue quartz separate, 100 mg ofthe microcline separate, and 40 mg of the matrix separate were

ZOLENSKY ET AL.: BLUE QUARTZ FROM LLANO RHYOLITE 3 1 5

analyzed for Fe, Na, and trace elements by instrumental neutronactivation analysis (rNae), following the procedures of Jacobs etal. (1977\. Most of these elements have been determined to bet-ter than +50/o; the errors are <l0o/o for Rb and Th, <250lo forCr, Ba, and Cs, and <500/o for Sr, Zr, and U. Spectra were col-lected from the blue quartz in two separate Ge(Li) detectors asa check for consistency; counting times ranged from 20 to 24 h.

The FeO, (: total Fe expressed as FeO) and TiO, contents ofthe blue quartz phenocrysts were also analyzed using an auto-mated Cameca cAMEBAx scanning electron microprobe. The in-strument accelerating voltage was 15 kV, the beam current was30 nA, and the beam diameter was 20 pm. Standard Camecainstrument and data-reduction software, which corrects X-rayintensities for dead time, background, and ZAF matrix effects,was employed. Ilmenite mineral standards were used. Countingtrmes on element peak positions were 100 s, and backgroundcounts were 60 s. The precision of these analyses is no betterthan +500/0, owing to the very low Fe and Ti concentrations inthe blue quartz. Microprobe analyses of groundmass ilmeniteand magnetite were made by using the same procedures, exceptthat the electron beam was focused to about l-pm diameter andpeak-position counting times ranged from 30 to 40 s. Naturalmineral standards were used. These analyses are considered pre-cise to within +0.50/0, because of the much higher concentrationsand very low instrument drift.

Rrsur,rsPetrographic examination

Llanite is a predominantly gray porphyry with abun-dant phenocrysts of reddish microcline microperthite(Oruu,Abrn.Anor, Barker and Burmester, 1970) and blueqvarlz. The microcline phenocrysts are as much as 10mm in diameter, and the quartz phenocrysts attain halfthis size. The groundmass is holocrystalline and micro-crystalline and is composed of microcline, quartz, biotite,magnetite, ilmenite, fluorite, apatite, and zircon.

The quartz phenocrysts are doubly terminated, con-sisting of a combination of positive and negative rhombo-hedrons with no obvious prisms, yielding a quartzoid.These crystals are blue only in reflected light, being reddishin transmitted light. This indicates that the blue colorationis due to Rayleigh scattering (Nassau, 1983) from sub-micrometer-sized grains suspended within the quartz(herein called type I inclusions). Another obvious featureof llanite blue quartz (but not a general characteristic ofall blue quartz) is chatoyance, called "opalescence" byearlier llanite investigators (Iddings, 1904). This phenom-enon is manifested by a bright blue-white flash observablein certain orientations. In reflected light this chatoyanceis seen to arise from planes oforiented capillary or bladedinclusions (herein called type 2 inclusions) which, whenviewed down the quarlzc axis, intersect at angles that aremultiples of 60" (see Fig. 1). These oriented inclusions areprobably lying on rhombohedral growth surfaces of thequartz crystals. These type 2 inclusions measure approx-imately I pm across and up to 100 pm in length, althoughthe length of an individual inclusion is difficult to judge.Since the production ofblue coloration via Rayleigh scat-tering occurs only from inclusions measuring no more

than -0.3 ;rm in largest dimension (Nassau, 1983), thetype 2 inclusions are not blue coloring agents.

Electron-beam analyses

sreu analyses ofelectron-transparent flakes ofthe bluequartz revealed the presence ofubiquitous, rounded, po-lycrystalline inclusions averaging -0.06 pm in diameter(see Fig. 2). Semiquantitative EDS analyses showed thepresence of major Fe and Ti, with minor Si and trace Mn(attributed to the host quartz). ED analysis ofthese inclu-sions revealed them to be ilmenite. The most frequentlyencountered diffraction reflections observed were at3.74(19) (0r2) ,2.16( t4) (104) , and 2.24(n ) A (113) . n-menite diffraction reflections measuring 2.55(13) (ll0),2.11( l l ) (202) , r .8 ] (9) (024) , and 1.65(8) A lZt t ; wereless commonly encountered. These ilmenite grains arethe smaller type 1 inclusions noted in the optical exam-ination. srEM imaging of ultrathin sections of blue quartzproduced by the ultramicrotome (one section is shown asan inset in Fig. 2) allowed larger regions ofquartz grainsto be observed in transmission than was permitted by thepreviously described fractured samples. The only mineralinclusions observed within these ultramicrotomed quartzsections were the type I ilmenites. The spatial density ofthe type I ilmenite inclusions was observed to vary con-siderably, up to -125 inclusions per cubic micrometer.

Larger ribbon-shaped crystals of ilmenite (identified byno) were encountered during observation ofthe fracturedblue quartz, although with considerably lower frequencythan the type I ilmenite inclusions (see Fig. 3). Semi-quantitative chemistry of these inclusions is also com-patible with ilmenite. These ribbon-shaped crystals mea-sure approximately 0. I by 1 by 20 pm, although it islikely that they were severed during specimen prepara-tion. Thus these inclusions are considered to representthe type 2 inclusions noted earlier in the optical exami-natron.

Ilmenite inclusions survived HF digestion of the quartzgrains and were present in the HF residue that was ex-amined by snv. The digestion residue was shown by nosanalysis to consist only of Fe, Ti, oxygen, and trace Mnand K (the latter element is a common contaminant fromHF). The residue mass contains a few type 2 ilmeniteribbons corresponding in size to those observed duringthe srEM examination. However, the majority of the res-idue material was revealed by srernr imaging to consist ofthe type 1 ilmenite crystals. An unknown percentage ofthe small crystals in the residue may have been derivedfrom the type 2 ilmenite crystals by the digestion process;however, a volumetric dominance of type I over type 2ilmenites was also observed during sreu imaging of un-digested blue quartz. Therefore it is concluded that thebulk of the inclusions in the blue quartz consists of thetype I ilmenite inclusions, although it is difficult to esti-mate the mass ratio of type I to type 2 inclusions withpreclsron.

Thirteen electron-microprobe analyses of ten bluequartz phenocrysts have a mean FeO, (wto/o) value of

3t6 ZOLENSKY ET AL.: BLUE QUARTZ FROM LLANO RHYOLITE

Fig. l. A single llanite blue quartz crystal viewed down the c axis in reflected light. Chatoyance arises from arrays ofacicularinclusions aligned parallel to the faces ofpositive and negative rhombohedra (and the arrow in the figure) and thus intersecting atangles that are multiples of 60', which is the angle between the inclusion arrays illuminated in the two views shown in (a) and (b).Each field of view measures - 1 mm.

0.013(9) (2o) and a mean TiO, (wto/o) value of 0.030(16)(2o). These data suggest that this blue quartz is signifi-cantly enriched in Fe and Ti relative to colorless quartzes,which typically have FeO, and TiO, concentrations lessthan 0.01 wt0/0, and often 0.005 wto/o or less (Frondel,1962; Nash and Crecraft, 1985). This conclusion is moredefinitive if the analytically superior (+50/0) rNaa Fe de-terminations (described below) of the blue quartz sepa-rate (0.022 wto/o FeO) are considered. Blue quartz samplesfrom other localities have reported FeO, concentrationsof 0.001 to 0.70 wto/o and TiO, concentrations of 0.002to 0.07 wo/o (Frondel, 1962).

Microprobe analysis of groundmass ilmenite and mag-netite yields compositions near end member FeTiO, andFerOo, respectively. A relatively large pyrophanite(MnTiO.) component, up to 15 mol0/0, is present ingroundmass ilmenite. Strong partitioning of Mn into il-menite relative to coexisting magnetite is a commonlyobserved phenomenon, particularly in high-silica rhyo-litic rocks (Czamanske and Mihalik, 1972: Czamanske eta1., l98l) and in regionally metamorphosed rocks (Rum-ble, 1976). Using the techniques of Spencer and Lindsley(1981) with the modifications of Andersen and Lindsley(1985), analyses from coexisting groundmass oxide pha-ses result in consistent calculated temperatures and oxy-genfugaci t iesof 396 t 125"C and l0-32t5barsfor therecalculation technique ofAnderson (1968), Lindsley andSpencer (1982), and Stormer (1982). The high Mn con-tent of llanite groundmass ilmenites, as well as their low

apparent temperatures and oxygen fugacity values, pre-clude great confidence in these calculations (as they in-volve computational extrapolations, Rumble, 197 6; Cza-manske et al., 1977); however, these low values clearlydo not represent the original magmatic conditions.Therefore these data support earlier suggestions (Delongand Long, 197 6) thaL the llanite rhyolite was affected bya very low grade metamorphic event (subgreenschist) sub-sequent to its crystallization. However, extensive elementredistribution at the whole-rock or mineral scale is notexpected to be significant at this low metamorphic grade.

rN.Ln analyses

The concentrations of FeO,, NarO, and trace elementsin the blue quartz and microcline phenocrysts andgroundmass of the llanite rhyolite are listed in Table l.The chondrite-normalized rare-earth-element (REE) con-centrations of these phases are plotted in Figure 4. Thellanite groundmass is enriched in Hf, Th, Zr, U, and REE,particularly LREE (La and Ce have concentrations great-er than 300 times the chondritic value). A large negativeEu anomaly (Eu/Eu* : 0.15) characterizes the ground-mass (Eu* is the Eu value derived by interpolating be-tween the neighboring REE Sm and Gd). The microclinephenocrysts have 0.429 wto/o FeO, and 3.63 wto/o NarOand are enriched in Rb and Ba. The chondrite-normal-ized REE pattern of the microcline decreases from La toLu, with little or no Eu anomaly.

The blue quartz phenocrysts are depleted in all ele-

ZOLENSKY ET AL.: BLUE QUARTZ FROM LLANO RHYOLITE

Frg.2. Transmission electron micrograph of type 1 rounded inclusions in the llanite blue quartz. Two inclusions are indicatedby arrows. Inset shows a lower magnification micrograph of an ultramicrotomed llanite blue quartz grain, which is outlined by adotted line. The size and distribution ofthe dark type I ilmenite inclusions is clearly observed. The superimposed wavy interferenceis produced by chatter of the enclosing epoxy.

J L I

ments relative to llanite groundmass, but have relativelyhigh concentrations of }{f,Zr, and HREE. A negative Euanomaly (Eu/Eux : 0.38) is present. Although severalnotable compositional differences between blue and col-orless quartz are discussed later, there are also some com-positional similarities. The Ba concentration of 3.5 ppmfor the blue quartz is similar to that reported for colorlessquarlz (0.45 to 13 ppm Ba, Ghiorso et a1., 1979; Four-cade and Alldgre, l98l). The blue quartz has equal con-centrations of FeO, and NarO (0.022 wto/o). This Na con-centration falls within the range (2 to 683 ppm Na)reported for colorless quartz (Frondel, 1962; Ghiorso etal., 1979:. Nash and Crecraft, 1985). The Fe enrichmentof the blue quaftz relative to colorless quartz was notedearlier. From the Fe concentration determined by rNe.o.,it may be calculated that the blue quartz contains either0.014 or 0.024 volo/o (assumed stoichiometric, pure end-member) ilmenite, depending on whether the quartz lat-tice itself contains either a relatively high (0.0080/o) or low

(0.0010/o) (Frondel, 1962; Nash and Crecraft, 1985) con-centration of Fe, respectively.

DrscussroN AND coNcLUSroNS

Transmission electron microscopy of the llanite bluequartz revealed the presence ofubiquitous, rounded typeI inclusions of ilmenite, averaging -0.06 pm in diameter.These inclusions produce the blue color ofthis quartz bythe phenomenon of Rayleigh scattering. These type 1 in-clusions probably make up most of the foreign inclusionswithin the quartz. The intensity of blue coloration in thequartz varies directly with the number density of thesetype I ilmenite inclusions. Apatite and zircon inclusions,reported from llanite blue quartz by Iddings (1904) andBarker and Burmester (1970), respectively, and believedby them to be the cause ofboth the blue color and cha-toyance, were not encountered in the present study.

The type 2 ilmenite ribbons present in the llanite bluequartz measure -0.1 by I by 20 pm, and are therefore

3 1 8 ZOLENSKY ET AL.: BLUE QUARTZ FROM LLANO RHYOLITE

Fig. 3. Type 2 ribbonlike inclusion crystal protruding from a thin chip of llanite blue quartz. Inset is an electron-difractionpattern produced by this crystal rn one orrentatron.

in general too large to cause blue coloration by Rayleighscattering. In certain orientations, constructive reflectionfrom these inclusions is observed to produce an intensewhite flash as a result of their alignment along (probablyrhombohedral) growth planes of the enclosing quartzphenocrysts. In a section cut normal to the c axis of aquartz phenocryst, six such chatoyant reflections, sepa-rated by angles of 60o, are observed.

The trace-element data collected for the blue quartzallows discrimination between an exsolution or magmat-ic crystallization origin for the ilmenite inclusions. Thepartitioning of a trace element between the llanite meltand the crystallizing quartz phenocrysts is adequately de-scribed by a simple Nernst partition coemcient (Mclntire,1963). On the other hand, if the ilmenite formed by exso-lution during either subsolidus cooling, metamorphicheating, or deformation, the trace-element partition coef-ficients of the blue quartz would be governed solely bythe initial colorless quartz-melt equilibria. Trace-elementpartitioning between the blue quartz and the exsolvingilmenite would not afect the original bulk compositionofthe blue auartz.

The shape of the trace-element partition-coefficientpattern for blue quartz formed by an exsolution processwould be similar to that of colorless quartz. The shapesof these patterns are determined by the relative differ-ences between trace-element partition coefficients in agiven mineral. These relative differences (i.e., the shapesofthe patterns) do not vary significantly as a function oftemperature, pressure, oxygen fugacity, and bulk liquidcomposition, in contrast to the absolute values of thepartition coefficients (Philpotts, 1978).

Alternatively, if the ilmenite inclusions are early-crys-tallizing phases that have been trapped in the blue quartzcrystals growing in the llanite, the trace-element partitioncoefficients of the blue quartz would reflect both colorlessquartz-melt equilibria and ilmenite-melt equilibria. Inthis case, the shape of the trace-element partition-coeffi-cient pattern for blue quartz would reflect a combinationof the patterns for colorless quartz and ilmenite.

Mineral/melt partition coefficients for the llanite bluequartz and microcline phenocrysts were calculated by di-viding their trace-element concentrations by those of thegroundmass (Table l). This exercise assumes that the

ZOLENSKY ET AL.: BLUE QUARTZ FROM LLANO RHYOLITE

Lo Ce

319

ovN

r)T

z0v

m

UFEoz-U

=zf,

()Uz(J

oEL)

=

Sm Eu Yb Lu

Fig. 4. Chondrite-normalized (Haskin et al., 1968) REE concentrations of blue quartz phenocrysts, microcline microperthitephenocrysts, and groundmass from the llanite rhyolite.

quartz and microcline phenocrysts crystallized in equilib-rium with a melt of the same composition as the ground-mass. The microcline partition coefrcients are character-ized by high values for Rb, Sr, and Ba, a preference forLREE over HREE, and a positive Eu anomaly. Alkalifeldspar/liquid partition coemcients reported elsewhere(Mahood and Hildreth, 1983; Nash and Crecraft, 1985)are similar, but have lower absolute values for severaltrace elements including the REE.

Mineral/melt partition coemcients determined for thellanite blue quartz show that this quartz is enriched inCr, Co, Sc, HREE, Hf, Zr, and U relative to the LREE(Fig. 5A). In contrast, it is widely assumed (cf. Cullers etal., 1981) that colorless quartz does not preferentiallypartition any one trace element relative to another. How-ever, the requisite compositional data needed to confirmthis assumption have been collected only for colorlessquartz from rhyolitic lavas of the Twin Peaks volcaniccomplex, Utah (Nash and Crecraft, 1985). The Twin Peakscolorless quartz displays relatively flat REE patterns andlittle or no Sc. Hf. and U enrichment relative to the LREE(Fig. 5A). Both the llanite blue quartz and the Twin Peakscolorless quartz exhibit large positive Eu anomalies.

The contrasting partition coefficient patterns of the twoquartz types suggest that the trace-element compositionof the llanite blue quartz is not governed solely by col-orless quartz-melt equilibria. Consequently, it is doubtfulthat a majority of the ilmenite included within the bluequartz formed by exsolution.

In contrast, the partition-coefficient data are consistentwith the hypothesis that the blue quartz includes a con-tribution from ilmenite inclusions crystallized from a meltwith the composition of the llanite groundmass. Mineral/melt partition coefficients determined for ilmenite inequilibrium with basaltic liquid (Fig. 58) generally in-crease with atomic number for the REE, except for a neg-ative Eu anomaly, and are enriched in Hf and Zr relaliveto Lu, and in Sc and Co relative to La. lt is easily deter-mined from a comparison of Figures 5,A' and 5B thatmixtures of ilmenite and colorless quartz yield apparentpartition coefficients with relative values similar to thoseof the llanite blue quartz. These data suggest (l) the ma-jority of the ilmenite inclusions (probably the more per-vasive type I inclusions) are early-fractionating phasesthat have been trapped within the quartz and (2) no otherearly-crystallizing phases are present within the quartz. If

320 ZOLENSKY ET AL.: BLUE QUARTZ FROM LLANO RHYOLITE

Tnere 1. Llanite blue quartz, microcline microperthite, andoroundmass concentrations and partition coeff icients

Concentrations*Partition coefficients

Species

Llaniteground-

Blue quartz Microcline massMicro-

Quartz cline

FeO,Naro

BbSrBa

CeSmEUTbYbLU

ScCoClHIThZrU

FzuoE .roU

$ .osz9trE .orc .m5

0 022 0.4290.022 3.63

z-ov397

14440

2700.69

1 17 .9285

z+.cJ

1 . 2 13.80

14 .02.27

5.491 2 0

2.516.422.0

420c . o

0.76 30090

3 5 17600.006 0.36

0.420 47 60 . 9 1 1 1 1 . 60.108 8.730.016 2.540 029 1,370.332 2.000.064 0.287

0 048 0.6790 027 0.260 26 0.815.50 1 440 089 1 .21

160 1000 09 0.40

0.01 2 .12

0 013 6 .50.01 0.52

0.003 0.4040.003 0 3910.004 0.3590.013 2 100.008 0 3600024 0.1430.028 0.126

0 009 0j240.019 0 190.10 0 320.335 0.0880.004 0.0550.38 0.240.02 0 09

TLMENTTE/BASALT B

.m160

- In ppm, except for FeOr and Na,O, which are in wt7o.

small amounts (0.01 to 0.02 volo/o) of early-crystallizedilmenite are also trapped within the microcline pheno-crysts, their compositional contribution is masked by therelatively greater trace-element concentrations of the mi-crocline.

Unfortunately, blue quartz tace-element data cannotcompletely discriminate between the two different typesof ilmenite inclusions. The chemical data presented hereare compatible with a crystallization mode of origin forboth types of inclusions. If this scenario is correct, thenthe two types of ilmenite found within the blue quartzprobably nucleated at separate locations or times and werethen incorporated together into the growing quartz crys-tals. It seems improbable that two ilmenite types withsuch dissimilar morphologies could have nucleated to-gether. In this instance, it is also possible that the type 2ilmenite crystals nucleated onto clusters of smaller, pre-existing type I ilmenite crystals, although this possibilitywas not investigated in the present study.

It is also possible, however, that some part of eithertype of ilmenite inclusion, or one type in entirety, formedby exsolution processes. This alternate scenario is per-mitted because the exsolution process would have no ef-fect on the blue quartz bulk trace-element composition.However, since it is probable from the present trace-ele-ment study that most of the ilmenite in the llanite bluequartz crystallized from the rhyolite magma before orduring the formation of the quartz phenocrysts, if onetype of ilmenite inclusion formed in entirety by exsolu-tion, it was most likely the volumetrically inferior type 2inclusions. If the type 2 ilmenite crystals indeed formedby exsolution, it remains possible that they nucleated ontopre-existing type I ilmenite crystals.

Cr Sc Ca Nd SmGd Dy Er Yb Hf U

Co La Pr Pm Eu Tb Ho Tm Lu Zr Th

Fig. 5. Mineral/melt partition coefficients for quartz and il-menite. (A) Blue quartzlllanite rhyolite (X; this study) and col-orless quartzlTwin Peaks rhyolite (.; Nash and Crecraft, 1985).(B) Ilmenite/synthetic high-Ti mare basalt (O, Nakamura et al.,1986; a, McKay et al., 1986; v, McKay and Weill, 1976), il-menite/plagioclase-poikilitic ilmenite lunar basalt 70135 (!,Haskin and Korotev, 1977) and ilmenite/Skaergaard mafic lay-ered series (O, Paster et al., 1974). (C) Ilmenite/Twin Peaksrhyolite (1, Nash and Crecraft, 1985) and ilmenite/Sierra LaPrimavera rhyolite (O, Mahood and Hildreth, 1983), with REEpartition coeffi cients recalculated- for illustrative purposes only-assuming arbitrary amounts of apatite and allanite contamina-tion (n, Twin Peaks, 40/o apatrte,0.22o/o allanite; O, La Prima-vera, 0.040/o apatite, 0.050/o allanite), using allanite and apatitepartition data from Mahood and Hildreth (1983) and Nagasawa(1 970), respectively. Recalculation assuming contamination bychevkinite and titanite [which have REE partition coefrcientsbroadly similar to allanite (Cameron and Cameron, 1986) andapatite (Green and Pearson, 1986b), respectivelyl also yields aHREE-enriched pattern. Limited knowledge of partitioning be-havior of other trace elements in these accessory minerals pre-cludes a more complete recalculation of the observed ilmenitedata.

ZOLENSKY ET AL.: BLUE QUARTZ FROM LLANO RHYOLITE 321

The shapes of partition-coemcient patterns for ilmenitein equilibrium with rhyolitic liquid (Fig. 5C), which is themore appropriate compositional system forthe discussionof llanite crystallization, are claimed to be significantlydifferent from ilmenite/basalt partition coefficients (Ma-hood and Hildreth, 1983; Nash and Crecraft, 1985). Cam-eron and Cameron (1986) have noted, however, that such"anomalous" mineral/rhyolite liquid partition coeffi-cients may reflect trace-element-cnriched accessory-min-eral inclusions (e.g., allanite, chevkinite, apatite, or tita-nite) trapped within host minerals. Detection of theseinclusions in opaque minerals such as ilmenite is mostdifficult and allows the possibility that available ilmenite/rhyolite partition coefficients measure accessory mineralcontamination. This contention is supported by the con-sistency between the relative REE partition coefrcientsfor (l) the recalculated ilmenite-rhyolite system, assumingsmall amounts of allanite and apatite contamination (Fig.5C), (2) the ilmenite-rhyolite system inferred from theblue quartz data, and (3) the ilmenite-basalt system, de-termined experimentally.

The one characteristic of the trace-element partitioncoefficients shared by colorless and blue quartz-a posi-tive Eu anomaly-is remarkable in that it is commonlyassumed (e.g., Nash and Crecraft, 1985) that quartz doesnot preferentially incorporate Eu relative to the other rare-earth elements. Since the colorless quartz from the TwtnPeaks rhyolite has not been affected by low-grade meta-morphism and does not include ilmenite, it is unlikelythat the Eu anomaly is a metamorphic phenomenon orthe result of ilmenite contamination. The anomaly is alsounlikely to be the result of feldspar contamination of theblue quartz separate, because inclusions in or adhesionson a transparent mineral such as qvarlz arc easily iden-tifled during microscopic examination. In order to pro-duce this Eu anomaly by feldspar contamination, it wouldbe necessary for 330/o ofthe blue quartz separate (partitioncoefrcient Eu/Eux : 2.60) to consist ofthe analyzed llanitemicrocline microperthite (partition coefficient Eu/Eu* :5.83). Besides being a very evidentproportion ofthe hand-picked sample, this amount of contamination would re-quire the quartz separate to exhibit FeO, (0. l4 wto/o), NarO(l .2wto/o\, and trace-element concentrations that are muchhigher than those observed.

There are at least two crystal-chemical explanations forthe Eu anomaly, both involving the reduction of Eu3* toEu2t. The first explanation is that Eu2* may enter intosixfold coordination with oxygen and thus form a substi-tutional solid solution in quartz, analogous to the sup-posed behavior of Mg'?* (Frondel, 1962). Ditrerences inthe ionic radii of Sio* (0.40 A) and Eu2" (1.31 A) (Shannonand Prewitt, 1969) preclude direct substitution.

A second explanation involves the formation of com-plexes of Eu2* with aluminosilicate (AlrSirO! ) in mag-matic liquids (Moller and Muecke, 1984). These com-plexes may be more similar to the structure of crystallinequartz, which is typically characterized by substitution ofSi by Al and a valence compensation ion such as Na

(Dennen, 1966; Ghiorso et al., 1979),than to the equilib-rium melt. The consequent preferential partitioning of thecomplexes into liquidus quartz would result in Eu2* en-richment.

Our conclusion that the llanite blue quartz containsilmenite inclusions, of which the majority (at least) areliquidus phases, suggests that the same result may be ob-tained for blue quartz from other locales, ifthey are sub-jected to a thorough examination using modern analyticaltechniques. It also follows that extreme caution should beexercised by workers wishing to develop blue quartz as ageothermometer-oxygen barometer. For example, Nordhas speculated that iron-titanium oxide grains (probablyilmenite) in blue quartz from a leucocharnockite-anor-thosite in the Roseland district, Virginia (cited in Herzand Force, 1984; G. L. Nord, Jr., pers. comm., 1986) maybe a high-temperature exsolution feature that could becalibrated in T-J6, space. However, INAA data indicatethat the majority of the ilmenite inclusions in llanite blueqvartz are not an exsolution feature. smu imaging indi-cates that these inclusions are probably the type I ilmen-ites, which cause the blue coloration in the llanite blueqvaftz. The less abundant type 2 ilmenite inclusions arepossibly of exsolution origin. However, these latter inclu-sions are unrelated to the blue coloration of the quartzphenocrysts. If sample selection for the blue quartz geo-thermometer-oxygen barometer is made on the basis ofinclusion-produced color, workers should ensure that thecolor-producing inclusions are genetically related to theevent they are attempting to characterize.

The entrapment of liquidus ilmenite in the llanite bluequartz phenocrysts also suggests that the llanite melt wascapable of crystallizing ilmenite very early in its crystal-lization history. Early ilmenite saturation may occur pri-marily in melts extraordinarily rich in Ti and Fe. It hasalso been established, however, that ilmenite saturationis promoted by decreasing temperature. increasing pres-sure, increasing oxygen fugacity, and increasing concen-trations of silica, alkalis, and REE in equilibrium liquids(Green and Pearson, 1986a). The inset in Figure 6 showsa significant lowering of the TiO, content required forilmenite saturation by these additional conditions. Thellanite rhyolite certainly exhibits a high-silica, alkali-rich(75 wto/o SiOr, 8.0 wto/o Na, O + KrO; Goldich, 1941;Barker and Burmester, 1970), REE-enriched composition,but in the absence of quantitative compositional data forthe ilmenite inclusions, estimates of their temperature,pressure, and oxygen fugacity of crystallization are pre-cluded.

Delong and Long (1976) suggested that Fe- and Ti-rich melarhyolite dikes of the Llano uplift may be relatedgenetically to the llanite rhyolite. The melarhyolite con-tains lower concentrations of silica and higher concentra-tions of Fe and Ti relative to the llanite rhyolite (Fig. 6).Some of the melar\olite contains minor groundmass bluequarlz.In comparison to the massive granites of the Llanouplift, the melarhyolite exhibits greater Fe and somewhatgreater Ti concentrations, whereas the llanite rhyolite is

. o l k o l i n e i

i=.=Sl'.n

322

l 2

l 0

- 0 8oF

.R

E o u

o 4

0 2

0

8

6

ZOLENSKY ET AL.: BLUE QUARTZ FROM LLANO RHYOLITE

70 72

Fig. 6. Whole-rock compositions of the massive granites (.),melarhyolite dikes (!), and llanite rhyolite dike (O, whole rock;O, groundmass) from the Llano uplift (Goldich, 1941; Barnes etal., 1947; Barker and Burmester, 1970; Delong and Long, 1976).A hypothetical Ti-rich phase solubility curve for these supposedliquid or near-liquid compositions is shown by the dashed line.Insert: Experimentally determined solubility curves in TiOr-SiO,space for the Ti-rich accessory phases at 7.5 kbar in subalkaline(solid curve, 1000'C; long-dashed curve, 900'C and REE-nch;dot-dashed curve, 900 .C) and alkaline (short-dashed curve, 1000'C) compositions (Green and Pearson, 1986a). All experimentsconducted under magnetite-wtistite-buffered oxygen-fugacityconditions except for dot-dashed curve, which was under con-ditions of the hematite-masnetite bufer.

clearly enriched in both Fe and Ti at the 75 wto/o SiO,level (Fig. 6). Blue quartz has not been identified in themassive Llano granites.

It is possible that some combination of low-tempera-ture, high-pressure, or high-oxygen fugacity conditionsallowed early ilmenite saturation in the Ti-, Fe-, REE-,alkali-rich, high-silica llanite rhyolite melt. A hypotheticalTi-rich phase solubility curve drawn in TiOr-SiO, spaceto allow early ilmenite crystallization in the llanite rhyolite

and in some of the melarhyolites is compatible with ex-perimentally determined curves for similar compositions

at <900 "C, 1.5 kbar, and the F' of the hematite-mag-netite (HM) buffer (Fig. 6).

If this model is correct, the restriction of blue quartzlargely to middle to late Proterozoic and some older rocksmay reflect preferential production of the required highTi-Fe felsic magma compositions and their crystallizationunder low-temperature, high-pressure, or high fo2 con'ditions during this time. It is well known that a distinctiveFe- and Ti-rich suite ofA-type (rapakivi) granites, massif-type anorthosites, and mangeritic rocks is largely confinedto the middle to late Proterozoic (Emslie , 1918:, Anderson,1983). In some cases, blue quartz may be another man-ifestation of this type of magmatism.

AcxNowr-nncMENTS

We thank G. Rossman, G. L. Nord, Jr., J. B Thompson, and D Gustfor helpful comments and calculations We appreciate the helpful reviewsof this manuscript by G. McKay, D Gust, and G L. Nord., Jr D S.McKay, G McKay, and D Blanchard kindly provided access to the elec-tron-beam and rNa.A. facilities ofthe Solar System Exploration Division atthe NASA/Johnson Space Center. B Bansal graciously performed the HFdigestion. T Cobliegh is acknowledged for enlightening us as to the sig-nificance ofthis problem and ensuring that we remained sensitive to theparticular needs ofthe llanite

RnrnnnNcrs crrpoAndersen, DJ., and Lindsley, D.H. (1985) New (and final!) models for

the Ti-magnetite/ilmenite geothermometer and oxygen barometer. EOS6 6 , 4 1 6 .

Anderson, A T ( I 968) Oxidation ofthe LaBlanche Lake titaniferous mag-netite deposit, Quebec. Journal ofGeology, 76,528-541

Anderson, J L. (l 983) Proterozoic anorogenic granite plutonism ofNorthAmerica. Geological Society of America Memoir l6l, 133-154

Barker, D.S, and Burmester, RF. (1970) Leaching ofquartz from Pre-cambrian hypabyssal rhyolite poryhyry, Llano County, Texas. Contri-butions to Mineralogy and Petrology, 28, l-8.

Barnes, V.E., Dawson, R F, and Parkinson, G A (1947) Building stonesofcentral Texas. Texas University Publication 4246,198 p

Cameron, K.L , and Cameron, M. ( I 986) Whole-rock/groundmass differ-entiation trends of rare earth elements in high-stlica rhyolites. Geo-chimica et Cosmochimica Acta, 50, 759-770.

Clabaugh, S.E., and McGehee, R V. (1963) Precambrian rocks of the Llanoregion, Texas University Bureau of Economic Geology Guidebook 5,63-78.

Clarke, J.W. (1984) The core ofthe Blue Ridge anticlinorium in northernVirginia. Geological Society ofAmenca Special Paper 194,155-160.

Cullers, R L, Koch, R.J , and Bickford, M E (1981) Chemical evolutionof magmas in the Proterozoic terrane of the St. Francis Mountains,southem Missouri; 2. Trace element data. Journal ofGeophysical Re-search, 86, 10388-10401.

Czamanske, G.K, and Mihalik, P (1972) Oxidation during magmaticdifferentiation, Finnmarka complex, Oslo area, Nomay; I The opaqueoxides Journal ofPetrology, 13, 493-509.

Czamanske, G.K., Wones, D R , and Eichelberger, J.C (1977) Mineralogyand petrology of the intrusive complex of the Pliny Range, New Hamp-shire. American Journal of Science, 2'7 7, 107 3-1 123

Czamanske, G K, Ishihara, S., and Atkin, S.A (1981) Chemistry ofrock-forming minerals of the Cretaceous-Paleocene batholith in southwest-em Japan and implications for magma genesis. Journal ofGeophysicalResearch, 86, 10431-10469

Delong, S E, and Long, L.E (1976) Petrology and Rb-Sr age of Precam-brian rhyolitic dikes, Llano County, Texas Geological Society ofAmerica Bulletin. 87 . 27 5-280.

Dennen, w H (1966) Stoichiometric substitution in natural quartz. Geo-chimica et Cosmochimica Acta, 30, 1235-1241

Emerson, B K., and Perry, J.H (1907) The green schists and associaled

Hypo lhe t ico l T i - r i chphose so lub i l i t y curve

\ --F-tp_ \ \q o - - -

g t rt r t r

sMelorhyo l i te

d ikes

tr

o

=

MossiveL lono gron i tes

O L l o n i t ev r n y o l l r e

)

granites and porphyries ofRhode Island U S Geological Survey Bul-let in 3 l 1. 45-46

Emslie, R.F. (1978) Anorthosite massifs, rapakivi granites, and Late Pro-terozoic rifting of North America Precambrian Research, 7 , 6l-98

Fourcade, F , and Alldgre, C.J. (l 98 I ) Trace element behaviour in granitegenesis: A case study. The calc-alkaline plutonic association from theQuengut Complex (Pyrenees, France). Contributions to Mineralogy andPetrology, 7 6, l'77-195.

Frondel, C (1962) The system ofmineralogy: Vol III. Silica minerals,334 p Wiley, New York

Ghiorso, M S, Carmichael, I S E, and Moret, L.K. (1979) lnverredhigh-temperature quartz Unit cell parameters and properties ofthe alpha-beta inversion Contributions to Mineralogy and Petrology, 68, 307-J / J .

Goldich, S.S. (1941) Evolution ofthe central Texas granites Joumal ofGeology, 49,697-720.

Goldschmidt, V M. (1954) Geochemistry, 365 p. Clarendon Press, Ox-ford

Green, T H , and Pearson, N.J. (1986a) Ti-rich accessory phase saturationin hydrous mafic-felsic compositions at high P, Z Chemical Geology5 4 , 1 8 5 - 2 0 1 .

-(1986b) Rare-earth element partitioning between sphene and co-existing silicate liquid at high pressure and temperature. Chemical Ge-ology, 55, 105- l 19.

Haskin, L A , and Korotev, R L. (1977) Test ofa model for trace elementpartition during closed-system solidifrcation of a silicate liquid Geo-chimica et Cosmochimica Acta, 41, 921-939.

Haskin, L A., Haskin, M A , Frey, F.A., and Wildeman, T.R. (1968) Rel-ative and absolute terrestrial abundances of the rare earths In L.HAhrens, Ed, Origin and distribution ofthe elements, p 889-912 Per-gamon, New York.

Herz, N., and Force, E. (1984) Rock suites in Grenvillian terrane oftheRoseland distnct, Virginia Part I Lithologic relations Geological So-ciety ofAmerica Special Paper 194, 187-200

Iddings, J P (1904) Quartz-feldspar-porphyry (graniphyro liparose-alas-kose) from Llano, Texas. Journal ofGeology, 12,225-231

Jacobs, J.W., Korotev, R L., Blanchard, D P , and Haskin, L A (1977) Awell-tested procedure for instrumental neutron activation analysis ofsilicate rocks and minerals. Joumal of Radioanalytical Chemistry, 40,93-l | 4

Jayaraman, N (1939) The cause of colour of the blue quartzes of thecharnockites of South India and the Champion Gneiss and other re-lated rocks ofMysore. Proceedings ofthe Indian Academy ofSciences,9. 265-285.

Lindsley, D H , and Spencer, K.J. (1982) Fe-Ti oxide georhermometry:Reducing analyses of coexisting Ti-magnetite (M0 and ilmenite (Ilm)(abs) EOS, 63,471

Mahood, G., and Hildreth, W (l 983) Large partition coemcienrs for traceelements in high-silica rhyolites Geochimica et Cosmochimica Acta,47, t-30.

McConnell, K.I., and Costello, J O. ( I 984) Basement-cover rock relation-ships along the western edge ofthe Blue Ridge thrust sheet in Georgia.Geological Society ofAmenca Special Paper 194, 263-280.

Mclntire, W L. (1963) Trace elemenl partition coefficients-A review oftheory and applications to geology Geochimica et Cosmochimica Acta,27. 1209-1264

McKay, G.A, and Weill, D. F (1976) Petrogenesis of KREEP. In Pro-

5 2 3

ccedings ofthe Seventh Lunar Science Conference, p.2427-2447 Per-gamon Press, New York

McKay, G, Wagstaf, J., and Yang, S.-R (1986) Zirconium, hafnium,and rare earth element partition coefrcients for ilmenite and other min-erals in high-Ti lunar mare basalts: An experimental study. In Pro-ceedings ofthe Sixteenth Lunar and Planetary Science Conference, Part2 Journal of Geophysical Research, 91, D229-D237.

lVlollcr. P , and Muecke, G.K. (1984) Significance of europium anomaliesin silicate melts and crystal-melt equilibria: A re-evaluation Contri-butions to Mineralogy and Pelrology, 87,242-250

Nagasawa, H (1970) Rare earth concentrations in zircons and apatitesand their host dacites and granites. Earth and Planetary Science Letters,9, 359-364

Nakamura, Y , Hirokazu, F, Nakamura, N, Tatsumoto, M., McKay,G A , and Wagstaff, J (1986) Hf, Zr, and REE partition coefficientsbetween ilmenite and liquid: Implications for lunar petrogenesis InProceedings ofthe Sixteenth Lunar and Planetary Science Conference,Part 2. Journal ofGeophysical Research, 91,D239-D250.

Nash, W P., and Crecrafl, H.R (1985) Partition coefficients for trace ele-menls in silicic magmas. Geochimica et Cosmochimica 4c1a,49,2309-2322

Nassau, K (l 983) The physics and chemistry of color, p. 233-249 Wiley,New York.

Niggli, E, and Thompson, J.B. (1979) Petrogenetic significance ofblue,opalescent quartz in metamorphic rocks Memorie degli Istituti di Geo-logia e Mineralogia dell' Iniversitd di Padova, 33, 258.

Parker, R B. (1962) Blue quartz from the Wind River Range, Wyoming.American Mineralogist, 47, 120 l-1202

Paster, T.P, Schauwecker, D.S., and Haskin, LA (1974) The behaviorof some trace elements during solidification of the Skaergaard layeredseries Geochimica et Cosmochimica Acta, 38, 1549-1571 .

Philports, J.A (1978) The law ofconstant rejection. Geochimica et Cos-mochimica Acta, 42, 909-920.

Postelmann, A (1937) Die ursache der blaufarbung gesteinsbildenderquarze Neues Jahrbuch li.ir Mineralogie, Geologie und Palaontologie,12.401-440.

Robertson, R (1884) Examination ofblue quartz from Nelson Co., Va.Chemical News, 50, no 1301, 207.

-(1885) An examination of blue quartz from Nelson County, Vir-ginia The Virginias,6, 2-3.

Rumble, D (1976) Oxide minerals in metamorphic rocks. MineralogicalSociety of America Short Course Notes, 3, Rl-R24

Shannon, R.D , and Prewitt, C T (1969) Effective ionic radii in oxidesand fluorides. Acta Crystallographica, 425, 925-946

Shaw, S.E, and Flood, R.H (1981) The New England batholith, eastemAustralia: Geochemical variations in time and space. Joumal of Geo-physical Research, 86, 10530-10544

Spencer, K J , and Lindsley, D.H. (1981) A solution model for coexistingiron-titanium oxides. American Mineralogist, 66, I 189-1201

Stormer, J.C. (1982) The recalculation of multicomponent Fe-Ti oxideanalyses for geothermometry: A quasi-thermodynamic model (abs.).EOS, 63 ,471

Watson, T.L, and Beard, R.E (1917) The color of amethyst, rose, andblue varieties ofquartz. Proceedings ofthe U S. National Museum, 53,f f J - ) b J

M.rxuscrrrr REcETVED Fesnuenv 16, 198'7Mnrrrscnryr AccEprED Novprrrsrn 20. 1987

ZOLENSKY ET AL.: BLUE QUARTZ FROM LLANO RHYOLITE