Origin of Phenocrysts and Compositional Diversity in ... - Oxford Journals

0022-3530/94 $3.00

Origin of Phenocrysts and Compositional Diversity in Pre-Mazama Rhyodacite Lavas, Crater Lake, Oregon by SETSUYA NAKADA*, CHARLES R. BACON, AND ANNE E. GARTNER U.S. Geological Survey, 345 Middlefield Road, Menlo Park, California 94025-3591 (Received 23 October 1992; revised typescript accepted 29 March 1993)

ABSTRACT Phenocrysts in porphyritic volcanic rocks may originate in a variety of ways in addition to nucleation and growth in the matrix in which they are found. Porphyritic rhyodacite lavas that underlie the eastern half of Mount Mazama, the High Cascade andesite/dacite volcano that contains Crater Lake caldera, contain evidence that bears on the general problem of phenocryst origin. Phenocrysts in these lavas apparently formed by crystallization near the margins of a magma chamber and were admixed into convecting magma before eruption. About 20 km3 of pre-Mazama rhyodacite magma erupted during a relatively short period between ~ 400 and 500 ka; exposed pre-Mazama dacites are older and less voluminous. The rhyodacites formed as many as 40 lava domes and flows that can be assigned to three eruptive groups on the basis of composition and phenocryst content. Phenocryst abundance decreases (from 32 to 8 vol.%) and SiO2 content increases (from 68 to 73 wt.%) in the apparent order of eruption. Phenocrysts (plagioclase, orthopyroxene, augite, and Fe-Ti oxides) are commonly fragmental or form polycrystalline aggregates with interstitial glass. Discrete phenocrysts with complete euhedral outlines are rare except for small elongated crystals. The abundance of discrete phenocrysts increases with that of aggregates. The grainsize of minerals in the aggregates covers the range of discrete phenocrysts (0-2-4-2 mm). Rim compositions of phenocrysts and the range of chemical zoning are almost uniform among the three rhyodacite groups, regardless of whether crystals are discrete or in aggregates. However, a small fraction of phenocrysts, especially small elongated crystals, have different compositions: plagioclase with Fe-rich cores and augite with Wo-poor cores, both of which are characteristic of crystals in undercooled andesite enclaves in the rhyodacites. The majority of phenocrysts were derived by disintegration of polycrystalline aggregates; rare, small phenocrysts crystallized in andesitic magma similar to that represented by the andesite enclaves. The modal and chemical compositions of the rhyodacites can be explained by different degrees of admixing of crystals, represented by the aggregates, into magma having ^ 4 vol.% 'true' phenocrysts, mainly plagioclase. The aggregates may be parts of the rind formed by in situ crystallization near the wall and roof of the magma chamber. The rind was disrupted during or just before eruption, and pieces were variably disaggregated and incorporated into erupting magma. The amount of rind incorporated declined during the sequence of eruptions. Owing to vesiculation of interstitial liquid and shearing during flow, crystals in the aggregates were separated and became phenocrysts. Pre-Mazama rhyodacite was erupted dominantly as lava, as opposed to the compositionally similar rhyodacite pumice of the Holocene caldera-forming eruption of Mount Mazama, apparently because its source chamber was crystallizing inward rather than actively growing.

* Permanent address: Department of Earth and Planetary Sciences, Faculty of Science 33, Kyushu University, Hakozaki, Fukuoka, 812 Japan. [Journal of Petrology, Vol. 35, Part I, pp. 127-162, 1994]

© Oxford University Press 1994

128

SETSUYA NAKADA ET AL.

INTRODUCTION Many porphyritic volcanic rocks contain small polycrystalline aggregates (often called crystal clots or glomerocrysts) composed of the same mineral phases that appear as phenocrysts (Stewart, 1975; Garcia & Jacobson, 1979; Arculus & Wills, 1980; Scarfe & Fujii, 1987). Interstitial glass may be present between the crystals of the aggregates, which lack evidence of subsolidus crystallization that would indicate a xenocrystic origin. The crystals composing the aggregates probably grew in the same magmatic system as the true phenocrysts in the host magma. Recently, many theoretical and experimental studies have suggested that crystallization occurs at the roof and walls of a magma chamber, where temperature is lowest, whereas the interior of a magma body convects, either in its entirety or in separate layers (e.g., Brandeis & Jaupart, 1987; Langmuir, 1989). Crystals that precipitated in the boundary layer close to the country rocks adhere to the walls of the magma chamber, so that a layer of crystals plus melt (rind; Mahood, 1990) surrounds the main body of convecting magma. The polycrystalline aggregates in porphyritic lavas may represent pieces of the rind. Tait (1988), de Silva (1989), and Fichaut et al. (1989) described possible examples of rind material (crystal-rich nodule, magmatic inclusion, and plutonic xenolith) that were broken and incorporated into magmas during eruption. The concept of crystallization in the boundary layer raises simple but important questions: (1) Could phenocrysts have crystallized from liquid represented by groundmass glass of the lava, i.e., were phenocrysts in equilibrium with the liquid? (2) Did phenocrysts exist in the hypothesized convecting part of the magma body? (3) If so, how do 'true' phenocrysts differ from crystals in the rind? (4) How can fragments of the rind be identified in erupted magma? Rhyodacite lava flows erupted before construction of Mount Mazama in the High Cascades of south-central Oregon (hereafter referred to as pre-Mazama rhyodacites for brevity) are porphyritic lavas in which textural and mineral compositional relationships can be used to constrain the origin of phenocrysts and polycrystalline aggregates. These voluminous rhyodacite lava flows underlie and are exposed on the south and east edges of Mount Mazama, the stratovolcano complex in which Crater Lake caldera occurs (Williams, 1942; Bacon, 1983; Bacon & Lanphere, 1990). The oldest lavas of Mount Mazama were erupted ~400 ka. The caldera formed during the climactic eruption of Mount Mazama, dated by 14C at 6845 + 50 yr B.P. Rhyodacite from the climactic magma chamber, vented as precaldera lavas and as pumice of the climactic eruption, is compositionally similar to the pre-Mazama rhyodacites of this study. Bacon & Druitt (1988) and Druitt & Bacon (1989) reported on the geochemistry and petrology of magmas erupted from the climactic chamber, and concluded that the magma chamber grew by repeated injections of basaltic to andesitic magmas that lodged between convecting rhyodacitic magma and underlying cumulates, where they partially crystallized and liberated differentiated liquid that rose and was incorporated into the growing rhyodacitic mass. The pre-Mazama rhyodacites show that the Crater Lake area has been a locus of silicic volcanism for considerably longer than implied by the geology of the immediate vicinity of the caldera. The silicic magmas do not vary greatly in composition, yet they contrast markedly in their mode of eruption: an extensive field of lava flows vs. a catastrophic pyroclastic eruption. In this paper we present geochemical and mineralogical data for preMazama rhyodacites and their enclaves. We use these data to map the compositional diversity among the lavas and to infer processes by which the magmas formed. As the observed compositional variation appears to be intimately linked to crystal content, the discussion focuses on the origin of crystals in the rhyodacites. We conclude that most of the

RHYODACITE LAVAS OF CRATER LAKE, OREGON

129

crystals in the pre-Mazama rhyodacites were admixed from the crystallizing rind of a magma chamber and that this process may account for the genesis of crystal aggregates and many of the discrete phenocrysts in other porphyritic rocks as well. PRE-MAZAMA VOLCANIC ROCKS Figure 1 is a generalized geologic map of the eastern and southern parts of the Crater Lake area. The andesite and dacite composite volcano of Mount Mazama is the most prominent edifice. Pre-Mazama dacite and rhyodacite flows and domes occur within ~ 15 km of the caldera on the east and ~ 1 0 km on the south. Constructional landforms of steep-sided domes and thick lava flows generally are preserved but original pumiceous carapaces have been removed by erosion. Many landforms have been glacially modified but, with few exceptions, glaciation did not alter morphology so greatly as to obscure primary forms. Volumes of eruptive units have been conservatively estimated based on mapped areas and minimum thickness. Except in glaciated canyon walls, pre-Mazama silicic lavas are poorly exposed, forming isolated weathered outcrops protruding from a thick Holocene pyroclastic blanket on forested slopes and, especially, on summits. Lava flows from Mount Mazama, and various Holocene pyroclastic deposits, overlie the dacites and rhyodacites. The pre-Mazama rhyodacites lie on basaltic andesite flows and cinder cones in the southwest and northeast parts of the region shown in Fig. 1, and are locally overlain by andesite and basaltic andesite of several monogenetic cones (Bacon, 1990). The rhyodacites overlie and abut the earlier dacites of Dry Butte and dacites west of The Pinnacles. Pre-Mazama dacites The pre-Mazama dacites and their enclaves are described here because, unlike lavas of nearby monogenetic volcanoes (Bacon, 1990) and products of the climactic magma chamber of Mount Mazama (e.g., Bacon & Druitt, 1988), no information on them has been published and because they may be genetically related to the overlying rhyodacites. Dacite of Dry Butte Dry Butte consists of a cluster of porphyritic dacite domes (~0-3 km 3 ). The dacite contains ~30 vol.% phenocrysts of plagioclase (

-

1

1

133

9 o i

o

i

o o

10 -

68

1

1

70

72

•» • •

•• •

li

i

6

74

76

78

SiO 2 wt.% Groundmass 100,

Plagioclase 100> o Pothole B. ° S. Crater P. o Scott Ck.

o Pothole B. Q S. Crater P. o Scott Ck. • Lg. agg.

Fe-Ti oxides

Discrete phenocrysts

Aggregates

Opx+Cpx

FIG. 3. Modal compositions of pre-Mazama rhyodacites. (A) Phenocryst abundances vs. SiO 2 . Open symbols indicate whole-rock samples (WR);filledsymbols are glass separates (GL). (B) Proportions of groundmass, discrete phenocrysts, and polycrystalline aggregates. Samples 1203 is a welded breccia. (Note approximately constant ratio of discrete phenocrysts to aggregates.) (C) Proportions of phenocryst species. Proportion of plagioclase is greater in relatively crystal-poor samples; large aggregates plot on extension of trend established by rhyodacites.

which was removed by glaciation and(or) was downfaulted away. In the west wall of Sun Creek canyon, basal vitrophyre of the rhyodacite of Pothole Butte overlies upper vitrophyre of the rhyodacite south of Crater Peak (andesite of Crater Peak flowed down a valley marking the contact between the two rhyodacite groups). It is likely that the time between eruptions was short, so that the upper vitric zone of the lower lava flow was not eroded away. K-Ar age determinations (M. A. Lanphere, unpub. data, 1989) on pre-Mazama rhyodacite whole-rock samples yield 409+ 14 (la) for rhyodacite of Scott Creek (flow north of Dry Butte), and 448 + 8 ka and 468 + 9 ka for rhyodacite of Pothole Butte (flows from Pothole Butte and north of Crater Peak, respectively). However, relative ages of groups suggested by these K-Ar ages do not agree with the apparent order of eruption based on field relations. The youngest K-Ar age is within error of that of the oldest dated lava of Mount Mazama(422 + 10 ka); similar flows from the same vent (Mount Scott) overlie rhyodacite of Scott Creek. K-Ar age determinations for volcanic rocks demonstrably older than the rhyodacites are consistent with eruption of pre-Mazama rhyodacite after ~600 ka: east of Annie Creek, the rhyodacite south of Crater Peak lies on basaltic andesite flows dated at

SETSUYA NAKADA ET AL.

134

i

1

1

]

o 0

"Sot* a

0.4 -

°8DO Boo o o

a

O • i

0.2

•

«

i

1

WR GL

o • a •

•

Pothole B. S.Crater P,

Scott Ck.

400

O.

a. 200

-

100 180

200

220

240

260

280

Zr ppm

FIG. 4. TiO 2 , FeO*, and Sr vs. Zr for pre-Mazama rhyodacites (WR, open symbols) used to define chemical groups. Data for glass separates (GL, filled symbols) scatter owing to removal of zircon.

770 + 30 ka and 580+20 ka; one eruptive unit of the rhyodacite of Pothole Butte lies on andesite dated at 620 + 20 ka west of the upper reaches of Sand Creek and another flowed against a basaltic andesite cone north of Lookout Butte dated at 610 + 50 ka. Paleomagnetic determinations (D. E. Champion, unpub. data, 1983) on four sites in the rhyodacite of Scott Creek and three sites in rhyodacite of Pothole Butte (Fig. 1) show two distinct magnetic directions with tight clusters for each group (Fig. 5). The paleomagnetic results imply that each rhyodacite group erupted during a geologically short period (a few decades). We lack sufficient information, however, to be sure that all eruptive units within each group were similarly coeval. The rhyodacites are devitrified except where vitric layers are preserved at the bottoms, tops, or margins of flows. Pink oxidation and gas cavities lined with vapor-phase crystals characterize the upper parts of many flows. The rhyodacites commonly are flow banded, as defined by multiple planes of filled or collapsed gas cavities and by layers with different devitrification textures. Flow banding, as well as fine platy jointing, typically dips steeply near flow tops and margins and is subhorizontal in flow interiors. In the few localities where undercooled andesite enclaves larger than a few centimeters are observed, gas cavities are


135

270 FIG. 5. Lower hemisphere plot of paleomagnetic pole positions for seven sites in pre-Mazama rhyodacites (D. E. Champion, unpub. data, 1990). Circles are a-95 confidence limits on means of measurements of at least 12 cores per site. Numbers refer to sites in Fig. 1. Results suggest that rhyodacites of Pothole Butte and Scott Creek were erupted in two brief episodes.

present in the pressure shadows at their sides. Most enclaves are ellipsoidal; subangular enclaves of undercooled andesite magma are rare. Large ( > 5 cm), ellipsoidal andesite enclaves have only been found near vent sites. ANALYTICAL METHODS At least one sample from each lava flow or dome was analyzed (Fig. 1). Most samples were ground in an alumina shatterbox, some glass separates in an agate mortar. Glass was separated from crushed vitrophyre samples using heavy liquids. Major elements were determined by X-ray fluorescence (XRF) in Lakewood, Colorado. Na 2 O and K 2 O were measured by flame photometry on approximately one-third of the samples. Rb, Sr, Y, Zr, and Ba were determined by energy-dispersive XRF in Menlo Park, California. Nb was analyzed by spectrophotometry in Reston, Virginia, and by inductively coupled plasma (ICP), after chemical separation, in Menlo Park. Instrumental neutron activation analysis (INAA) was used for Co, Cr, Cs, Hf, Sb, Ta, Th, U, Zn, Sc, and REE (Reston). Precision and accuracy of the above methods have been summarized by Bacon & Druitt (1988). The analytical results are listed in Tables 1, 2, and 3. Chemical analyses of minerals and glasses were performed using an ARL SEMQ electron microprobe equipped with nine wavelength-dispersive spectrometers. Corrections were made for interferences of Ti K a on V K a and Si K p on Sr ^.Concentrations were calculated using the corrections of Bence & Albee (1968) and Albee & Ray (1970). To prevent Na volatilization during analysis, we used an electron beam > 8 ^m in diameter and a specimen current of 20 nA on plagioclase, and >25 nm and 9 nA on glass. PETROGRAPHY Pre-Mazama rhyodacites The rhyodacites contain plagioclase, orthopyroxene, augite, and Fe-Ti oxide phenocrysts (Table 1). Amphibole phenocrysts are rare. Phenocrysts occur either as discrete crystals or in polycrystalline aggregates with or without intergranular glass (Fig. 6); both have rounded outlines against the groundmass. Many discrete phenocrysts are fragments of broken crystals. Small, elongated crystals commonly are euhedral. Apatite occurs as a common accessory mineral, but zircon is rare.

Hf Nb Rb Sc

Cs

ppm Ba Co Cr

MgO CaO Na 2 O K2O TiO 2 P2O5 MnO LOI Total

70-6 14-6 2-46 0-44 1-78 4-69 2-84 0-39 008 003 0-66 98-6

693 PB

71-5 140 2-37 0-49 1-69 4-80 2-89 0-42 011 004 0-29 98-6

7/0 PB

837 843 839 — 3-2 3-2 — 11 1-4 — 3-3 1-3 — 61 5-8 — 6-2 5-7 57 57 63 5-67 663 —

71-3 14-4 2-45 0-62 1-81 504 2-80 0-45 0-09 005 0-70 99-7

SiO° A1 2 O 3

3

534 PB

No. Unit*

859 2-2 — 1-5 5-8 6-6 58 111

70-4 14-5 2-75 0-40 1-67 519 2-72 0-40 010 007 0-60 988

711 PB

1221 PB

711 145 2-49 0-57 1-77 513 2-79 0-41 009 005 0-46 99-4

1374 PB

1220 SCP

Major elements

1203 SCP

1224 SCP

Trace 1 elements

70-4 71-8 71-2 70-6 14-4 14-5 14-9 14-5 2 41 2-74 2-72 2-66 0-71 0-77 0-77 0-56 1-84 2-12 1 76 2-13 4-80 4-94 4-83 4-86 2 81 2-76 2-63 2-68 0-42 0-42 0-42 0-43 008 008 008 009 004 004 005 005 1-71 0-75 104 1 35 99-7 100-3 100-5 100-3

1406 PB

874 834 826 846 875 844 851 — 3-4 3-8 2-6 3-2 4-3 3-9 — — — — 2-8 3-3 1-9 — 3-2 2-9 31 3-2 3-2 31 — 60 5-8 51 60 5-6 61 — 5-7 6-4 7-4 7-4 7-9 5-5 59 51 58 62 58 56 56 6-72 6-74 5-81 6-72 5-81 6-33 —

700 69-6 14-8 13-9 2-70 2-46 0-62 0-71 1-83 206 4-59 4-78 2-92 2-60 0-45 0-43 009 009 005 005 181 2-58 98-7 100-4

732 PB

813 5-3 3-8 3-5 4-8 6-4 56 6-98

67-7 14-7 2-99 0-96 2-64 419 2-94 0-45 011 005 2-90 99-6

CD212-2 SC

67-3 15-7 3-30 109 2-84 4-85 2-49 0-50 013 006 0-86 991

682 SC

832 807 — 4-4 2-2 — 11 — — 51 — 5-5 43 35 7-84 —

68-5 15-5 3-28 0-90 2-42 5-46 2-37 0-50 013 007 0-45 99-6

65/ SC

Chemical and modal analyses of representative pre-Mazama rhyodacites

TABLE 1

68-7 151 3-39 102 2-59 4-67 2-49 0-52 016 006 0-59 99-3

690 SC

792 754 51 6-7 2-8 5-5 1-5 1-3 5-1 4-8 5-5 4-9 55 50 7-40 810

69-2 14-9 309 0-68 2-25 4-68 2-57 0-45 012 006 0-75 98-8

688 SC

67-6 15-5 3-44 104 2-66 4-62 2-35 0-52 012 006 0-85 98-8

755 SC

68-4 15-4 3-52 1-21 304 506 2-27 0-54 012 006 0-85 100-5

1204 SC

786 705 808 4-8 6-6 6-6 — 6-4 6-9 10 1-8 2-9 5-2 5-3 50 5-5 5-5 50 40 49 36 7-87 8-40 907

68-6 15-5 2-94 0-79 2-46 4-87 2-33 0-45 013 006 0-69 98-8

747 SC

C-

hi

>

C

in

m

VJ

693 PB

7/0 PB

81-5 16-2 1-3 10 61 12-4

84-8 140 0-9 0-5 6-6 8-7

84-7 12-9 1-5 09 5-3 101

243 244 235 — 0-57 0-53 — 5-7 5-5 — 2-3 2-2 27 30 23 — 43 34 237 237 247 — 21 24 — — 42 — 18 21 — 3-9 4-6 0-83 0-84 — — 0-55 0-67 2-3 2-5 0-36 0-38 —

534 PB

732 PB

1221 PB

1374 PB 1406 PB

1220 SCP

Trace elements

1203 SCP

1224 SCP

911 7-6 0-8 0-6 4-3 4-6

82-2 161 1-4 0-4 6-3 116

86-5 116 1-4 0-6 61 7-4

89-3 9-5 0-9 0-3 5-9 4-9

84-3 13-8 1-6 0-4 4-4 11-3

76-7 20-7 1-8 0-8 11-8 11-5

75-8 21-5 1-9 0-8 8-7 15-4

80-2 17-8 1-4 0-7 5-9 140

Modal iimposition

278 226 237 240 226 242 268 266 — 0-53 0-54 0-56 0-53 0-57 0-55 0-48 — 5-3 5-6 5-3 5-6 5-8 5-9 5-2 — 21 20 2-2 2-4 2-1 2-3 2-1 27 22 32 24 25 26 24 24 — 51 42 46 39 44 36 43 220 232 231 236 245 248 218 208 — 29 21 19 19 20 21 18 — 45 41 38 38 39 41 35 — 25 17 16 18 18 18 14 — 5-7 41 3-8 40 4-1 41 3-4 — 109 0-82 0-78 0-81 0-81 0-78 0-71 — 0-57 0-58 0-60 0-64 0-93 0-57 0-51 — 2-3 3-2 2-4 2-5 2-5 2-2 2-1 0-32 0-45 0-36 0-37 0-36 0-37 0-3 —

711 PB

68-7 27-4 2-9 11 10-7 20-7

319 0-46 4-8 1-9 23 43 197 18 34 16 3-4 0-78 0-52 1-9 0-29

CD212-2 SC 682 SC

701 261 30 0-9 8-9 211

71-7 23-1 3-9 1-3 113 170

295 332 — 0-51 — 4-9 — 1-8 24 27 — 51 207 214 — 18 — 38 — 16 — 3-8 — 0-96 — 0-56 — 21 0-32 —

681 SC

690 SC

747 SC 755 SC

1204 SC

71-9 24-9 20 1-3 7-3 20-9

23-0 3-7 1-6 7-5 20-7

71-9

70-8 25-3 30 10 90 20-3

67-9 27-4 3-4 1-3 9-7 22-4

73-9 210 3-3 1-1 10-6 15 6

314 354 321 318 308 0-46 0-51 0-53 0-48 0-46 4-6 4-5 4-8 5-1 4-9 1-8 1-7 1-9 1-9 1-9 21 24 25 22 25 45 46 51 43 53 192 211 213 206 202 25 16 18 20 18 37 30 39 35 37 20 13 18 17 17 3-2 41 4-3 3-8 4-4 0-84 0-97 0-99 0-86 0-95 0-48 0-61 0-59 0-53 0-72 20 1-9 20 20 2-2 0-29 0-34 0-30 0-31 0-33

688 SC

Major elements by wavelength-dispersive XRF (WDXRF), D. F. Siems, analyst; Rb, Sr, Y,Zr, and Ba by energy-dispersive XRF (EDXRF), P. E. Bruggman, analyst; Nb by spectrophotometry •(534,681,688, 690, 693., 711,732,747,755, 1203i, 1204, and 1374), P. J.,Aruscavage, analyst; Nb by ICP (1220, 1221, 1406, and CD212-2), S. T.Pribble, analyst; rest by INAA, J. S. Mee, analyst. * PB, rhyodacite of Pothole Butte; SCP, rhyodacite south of Crater Peak; SC, rhyodacite of Scott Creek. •f Total iron as Fe 2 O 3 . | Gms., groundmass; plag., plagioclase; pyx., total pyroxene; oxides, Fe-Ti oxides; dis.ph., discrete phenocrysts; aggr., aggregates of crystals.

Plag. Pyx. Oxides Dis.ph. Aggr.

Gms.

ppm Sr Ta Th U Y Zn Zr La Ce Nd Sm Eu Tb Yb Lu

No. Unit*

T A B L E 1—continued

Z

O

m

o

m

7*

r"

TO

-i tn

o

•n

O

Vi

>

X

1,38

S E T S U Y A N A K A D A ET

AL.

FIG. 6. Photomicrographs of rhyodacites, aggregates, and andesite enclaves. (A) Rhyodacite of Pothole Butte (sample 1406). Fragmental phenocrysts and aggregates are present, whereas discrete phenocrysts with euhedral outlines are scarce. Small clot of fine crystals is an undercooled andesite enclave. Width ~ 2 cm. Plane-polarized light. (B) Rhyodacite of Scott Creek (sample 1204) showing aggregates. Width ~ 2 cm. Plane-polarized light. (C) Coarse- and fine-grained aggregates in rhyodacite of Scott Creek (SN-10). Width ~ 1-2 cm. Plane-polarized light. (D) Coarsest-grained aggregates in rhyodacite of Scott Creek (sample 682), at right, occur in a variety of sizes. Finegrained aggregate at left. Width ~ 2 cm. Plane-polarized light. (E) Coarse-grained aggregates in rhyodacite of Scott Creek [coarser-grained aggregates in (C); SN-10]. Width ~ 5 mm. Plane-polarized light. (F) Andesite enclave (sample 1638) in rhyodacite of Pothole Butte. Skeletal, fine needles of plagioclase and amphibole characterize the matrix of the andesite enclave. Width ~ 5 mm. Crossed nicols.

Modal compositions of representative rhyodacite samples were determined by point counting an area of > 5 cm 2 . Polycrystalline aggregates and andesitic enclaves with diameters > 3 mm were omitted from the point counting of thin sections, to obtain modal data representative of the bulk rock. Single phenocrysts were identified as discrete phenocrysts, even if fragmental and non-euhedral, and were distinguished from polycrystalline aggregates (hereafter simply called aggregates) (Table 1). Phenocrysts adhering to each

RHYODACITE LAVAS O F CRATER LAKE, OREGON

139

other and large phenocrysts having phenocryst-size inclusions (>0-2 mm; e.g., orthopyroxene grains in plagioclase) were counted as aggregates. However, pyroxenes with small Fe-Ti oxide inclusions were counted as discrete phenocrysts. The ratio of discrete phenocrysts to aggregates is approximately constant (Fig. 3B). One sample (No. 1203), vitrophyre that may have formed by welding of brecciated lava, plots a little away from the main cluster of analyses in the diagram. Figure 3C shows the modal ratios among phenocrysts. Plagioclase accounts for 80-90% of the crystals. Discrete plagioclase phenocrysts are typically 0-8-2 mm long but may be as long as 4-2 mm. Plagioclase grain-size range hardly changes among the three rhyodacite groups. Relatively large discrete phenocrysts commonly are corroded, whereas small discrete phenocrysts are euhedral and elongated. Most plagioclase phenocrysts have brown glass inclusions, and are broken along cracks roughly parallel to cleavage and through glass inclusions (Fig. 6). Overgrowth rims are not found on broken edges, implying that the fragmentation occurred during or just before eruption. The fragmentation may have been caused by expansion and vesiculation of melt inclusions owing to depressurization during eruption. Alternatively, some plagioclase fragments may have been derived from polycrystalline aggregates. Pyroxene phenocrysts are commonly 0-4-0-8 mm long, but may be up to 2 mm long. Only large phenocrysts are broken. Fe-Ti oxides commonly form inclusions in, or are attached to the surface of pyroxene phenocrysts. The ratio of augite to orthopyroxene is variable. Mean abundances are 1-0 vol.% orthopyroxene and 0-3 vol.% augite. Amphibole grains (up to 1-1 mm long) rarely occur as a few discrete, fragmental phenocrysts and in aggregates ( < 0 1 vol.% per thin section). Ilmenite and titanomagnetite (hereafter called simply magnetite) occur as discrete phenocrysts and inclusions in pyroxenes. The average abundance is ~ 10 vol.%. They are commonly 0-3 mm across, but may be up to 0-8 mm across. Ilmenite grains are homogeneous in all rhyodacites examined, whereas magnetite grains are homogeneous only in glassy samples.

Large aggregates Rounded polycrystalline aggregates with diameters > 3 mm are referred to as large aggregates (Fig. 6). The size limit was chosen so that the large aggregates could be excluded from point counting. The maximum diameter observed is 3 cm. Although each large aggregate is composed of approximately uniform-sized crystals, grain size varies among the large aggregates (plagioclase is commonly 0-6-2-1 mm long). Larger-diameter aggregates apparently have finer constituent minerals (Fig. 6). The range of grain-size covers that of discrete phenocrysts: the largest plagioclase in the large aggregates (3-8 mm long; sample 682; Fig. 6) is close to the maximum size of discrete plagioclase phenocrysts (4-2 mm). Colorless or pale brown glass, or its devitrified equivalent (mainly cristobalite and feldspars), is interstitial to crystals and also forms inclusions in plagioclase of the large aggregates. Large aggregates are texturally different from andesite enclaves: (1) plagioclase occurs as closely packed, euhedral to subhedral prisms in the large aggregates, in contrast to acicular and swallowtail forms in the undercooled andesite enclaves; (2) crystals are variable in size within each andesite enclave; (3) aggregates generally lack amphibole, which is characteristically (but not always) present in andesite enclaves; and (4) andesite enclaves have comparatively high porosity (diktytaxitic texture).

140

SETSUYA N A K A D A ET

AL.

Andesite enclaves

Andesite enclaves are typically less than a few centimeters across, but up to 50 cm. They are magmatic inclusions (Bacon, 1986) and are andesite, basaltic andesite, and shoshonitic basaltic andesite (Figs. 1 and 2; Table 2). Enclaves are aphyric to porphyritic with various

TABLE 2

Chemical analyses of representative andesite enclaves from pre-Mazama rhyodacites No. Unit*

713 PB

733 PB

1638 PB

697

SC

1206 SC

1398 SC

1399 SC

1400 SC

1404 SC

55-6 181 8-57 3-32 7-38 406 103 119 0-28 013 0-64 100-3

57-4 17-9 7-15 2-89 5-47 4-15 1 56 101 0-28 012 2-25 100-2

560 16-7 6-85 3-17 703 3-91 1-44 0-92 0-26 011 3-40 99-8

55-1 190 7-58 3-30 5-87 3-68 1-26 0-95 0-23 015 2-72 99-8

415 20-3 360

598 17-4 27-0 1-7 3-4 5-5 35 16 8 512 0-30 1-9 0-8 28 88 148 16 30 17 4-8 1-22 0-71 2-4 0-36

552 18-3 31-7 1-2 31 4-1 35 ' 15-9 545 0-31 1-9 0-7 22 84 133 13 25 14 3-7 103 0-55 20 0-31

1634 SC

Major elements wt.% SiO 2 A1 2 O 3

Fe 2 O 3 t MgO CaO Na2O K2O TiO 2 P2O5 MnO LOI Total

58-5 17-2 5-96 3-35 6-70 4-34 105 0-81 0-27 012 0-65 98-9

55-8 18-2 6-40 3-86 7-07 3-94 116 0-82 0-28 009 109 98-7

58-2 180 6-24 3-52 6-52 3-95 1-25 0-76 019 010 0-99 99-7

55-3 17-8 809 3-73 7-26 3-78 104 104 0-30 013 0-63 991

58-1 18-2 6-53 3-67 7-07 419 113 0-74 018 010 0-67 100-6

52-6 17-8 7-85 406 8-01 4-46 1-32 1 30 0-72 011 0-67 98-9

Trace elements ppm

Ba Co Cr Cs Hf Nb Rb Sc Sr Ta Th

U Y Zn Zr La Ce Nd Sm Eu Tb Yb Lu

495 15-8 48-8 0-7 30 6-3 31 13-6 902 0-29 21 0-7 25 70 142 22 40 20 4-5 109 0-53 1-7 0-24

488 210 751 0-7 2-7 2-6 28 15 9 1040 0-22 20

0-5 16 74 124 17 33 17 3-6 111 0-45 1-4 0-20

546 20-4 61-5 1-2 30 4-6 36 16-5 761 0-29 2-3 0-6 25 68 143 16 34 16 3-6 107 0 51 1-4 0-22

407 22-2 38-6 0-9 2-6 31 18 21 3 618 0-27 1-2 0-6 19 85 116 14 28 16 4-1 1-23 0-38 21 0-29

387 19-5 55-5 10 21 2-2 26 17-3 654 0-2 1-4 0-7 14 73 94 9 18 10 2-7 0-87 0-39 1-2 019

0-6 2-7 5-2 20 23-8 619 0-26 1-2 0-5 25 97 122 15 30 20 4-8 1-30 0-69 2-3 0-33

492 1511 21 6 22-9 123 111 0-6 0-9 2-5 5-7 3-8 9-6 26 62 211 16-3 584 2050 0-21 0-52 1-9 4-7 0-8 1-4 31 26 110 79 123 254 14 44 33 95 16 50 4-2 90 2-24 110 0-68 0-82 2-8 1-8 041 0-24

Major elements by WDXRF, D. F. Siems, analyst; Rb, Sr, Y, Zr, and Ba by EDXRF, P. E. Bruggman, analyst; Nb by spectrophotometry (691,713, 733, and 1206), P. J. Aruscavage, analyst; Nb by ICP (1398,1399,1400,1404, 1634, and 1638), S. T. Pribble, analyst; rest by INAA, J. S. Mee, analyst. * PB, rhyodacite of Pothole Butte; SC, rhyodacite of Scott Creek. t Total iron as Fe 2 O 3 .


141

combinations of plagioclase, olivine, augite, and amphibole phenocrysts; shoshonitic basaltic andesite (highest Zr point in Fig. 2) contains only amphibole phenocrysts. Some andesite enclaves contain plagioclase phenocrysts ( ^ 4 mm) with sieved cores and clear rims that suggest that these enclaves, like many other occurrences (Bacon, 1986), themselves represent hybrid magmas. Elongated laths of groundmass plagioclase are typically 0-4-0-8 mm. Groundmass pyroxenes are anhedral and sometimes form oikocrysts (80 vol.%; K 2 O 01-0-5 wt.%). A difference of 01 wt.% K 2 O hardly changes the calculated results (0-5 wt.% in SiO 2 ). Calculated normative abundances of crystalline phases (wt.%) agree with the modal abundances of phenocrysts (Tables 1 and 8).

156

S E T S U Y A N A K A D A ET

AL.

TABLE 8

Calculated compositions ofcrystalline phases in rhyodacites and in large aggregates

25-5

_ LA WO3 —

_ An30 — —

54-3 20-4 7-3 3-6 7-9 4-9 0-2 11 0-2 01

51-2 20-6 9-5 3-7 8-4 4-5 0-2 1-9 00 01

60-8 24-6 0-3 00 61 7-8 0-4 00 00 00

— — —

— — —

—

—

No. Unit' Phen.%2 Calc.%*

1221 PB 13-5 17-8

1203 SCP 23-3 23-3

1220 SCP 24-1 22-6

1204

wt.% SiO 2 A12O3 FeO* MgO CaO Na 2 O K2O TiO 2 P2O5 MnO

55-7 21-6 6-7 2-6 6-3 5-5 0-2 10 0-3 01

55-9 21-9 6-4 2-5 60 5-8 0-2 10 0-3 0-2

55-7 214 6-6 2-6 6-3 5-8 0-2 10 0-4 0-2

ppm Ba Rb Sr Y

Zr

694 0 817 11 73

652 1 727 18 193

653 35 782 28 177

sc 261

294 25 869 16 40

C/PWnorm5 Q or ab an C di hy ol mt il ap

4-2 1-2 460 29-4 1-6 00 118 00 3-2 1-9 0-7

3-8 1-2 48-6 27-4 21 00 11-2 00 30 1-8 0-8

2-7 1-2 48-6 28-4 1-3 00 11-9 00 31 1-8 0-9

1-7 1-2 41-5 330 00 3-8 12 9 00 3-5 20 0-5

00 1-2 381 35-4 00 5-1 9-9 2-5 4-6 3-6 00

0-7 2-4 660 30-3 0-2 00 0-3 00 01 00 00

mol.% An

37-5

34-6

35-5

42-7

—

300

1

PB, rhyodacite of Pothole Butte; SCP, rhyodacite south of Crater Peak; SC, rhyodacite of Scott Creek; LA, large aggregate. 2 Phen.%, total phenocrysts in mode. 3 Modal vol.% (wt.% in parentheses) in large aggregate SN10-E2: plag = 80 (74), cpx = 51 (5-6), opx= 101 (120). Fe-Ti oxides = 5 0 (8-4). * Calculated vol.% crystals. 5 Calculated assuming Fe 3 + =0-3 x Fe.

Calulated bulk compositions of crystalline assemblages are uniformly dioritic (Table 8). They have higher A12O3 contents than the andesite enclaves (Table 2). Incompatible element abundances (Rb and Zr) should be low because the observed crystalline phases in the rhyodacites are plagioclase, pyroxenes, Fe-Ti oxides, and apatite. However, Zr contents of two samples (Nos. 1203 and 1220) are fairly high. This may be the result of zircon precipitation, as discussed above. CIPW normative compositions of crystalline assemblages are consistent with modal abundances of phenocrysts in rhyodacites, except for normative quartz and diopside. The bulk composition of a large aggregate was calculated from weight proportions of constituent crystals, based on modal analysis, and their average compositions (Table 8).


157

Although the mineral proportion used for the large aggregate (footnote to Table 8) is similar to normative compositions of the bulk crystalline assemblage in the rhyodacites, the calculated aggregate composition is poorer in SiO 2 , A12O3, and Na 2 O, and richer in FeO*, MgO, CaO, and TiO 2 . This result suggests that either this large aggregate is not strictly equivalent to the source of phenocrysts in the rhyo'dacite or that its modal analysis is not representative of large aggregates. True' phenocrysts Analysis of modal data for the pre-Mazama rhyodacites suggests the presence of a small fraction of 'true' phenocrysts that were not derived by disintegration of polycrystalline aggregates. If all crystals were derived from aggregates, a regression line for the data plotted in Fig. 3B should project through the groundmass apex. Yet, such a line intersects the discrete phenocrysts-groundmass join at ~ 4 vol.% discrete phenocrysts. This may indicate an approximate abundance of 'true' phenocrysts in the magma. If so, the degree of fragmentation (disaggregation) of the aggregates would have been similar in all three rhyodacite groups, and there would have been little crystallization during the time represented by the series of eruptions; the main difference among the rhyodacites is in the amount of admixed crystals. It should be noted that uniformity of phenocryst compositions and Fe-Ti oxide temperatures among eruptive groups, and the increase in crystal content with decreasing whole-rock SiO 2 , are consistent with derivation from a single magma chamber. Figure 3C indicates that the rhyodacites of Pothole Butte and south of Crater Peak are enriched in plagioclase, and that the two point-counted large aggregates are poorer in plagioclase than the rhyodacites. The modal compositions seem to lie astride a line connecting the large aggregates and the plagioclase apex. If'true' phenocrysts are mainly plagioclase, the modal variations could be explained by mixing of (1) crystals in proportions found in the large aggregates with (2) the plagioclase-dominated phenocryst assemblage in crystal-poor magma. Table 9 shows calculated modal compositions for mixtures with 30 and 8 vol.% crystals, the two extremes found in the rhyodacites. The calculated modal compositions are concordant with those observed (Fig. 3C); however, some samples from the rhyodacite of Pothole Butte are more enriched in plagioclase than the mixture with 8% crystals. If the crystal abundance in the large aggregates was higher than the assumed value (70%), the proportion of aggregates in the mixture would be smaller than those given in Table 9, to maintain the same crystal abundances, and the calculated proportion of plagioclase in the rock would become slightly high. When pyroxenes or Fe-Ti oxides, in addition to plagioclase, are included as 'true' phenocrysts in the magma, the proportion of plagioclase in the rock becomes smaller, keeping the mixing ratio of crystals to magma constant. Chemical compositions of whole rocks are also consistent with a model in which the crystalline phases of the rhyodacites represent crystals from the aggregates plus plagioclase phenocrysts from magma. Calculated bulk compositions of crystalline phases in the rhyodacites lie between those of large aggregates and An 30 plagioclase (Table 8; Fig. 12). In Al 2 O 3 -SiO 2 and MgO-K 2 O diagrams (Fig. 12), the bulk composition of crystals in a large aggregate falls on a line through the data for rhyodacites. Compositions of phenocryst assemblages in the rhyodacites fall within the field bounded by the straight lines representing mixtures of rhyodacitic melt, plagioclase phenocrysts, and polycrystalline aggregates. Thus, chemical variation among pre-Mazama rhyodacites is controlled by the degree of incorporation of crystals from the source of large aggregates into silicic magma containing a

158

SETSUYA N A K A D A ET

AL.

TABLE 9

Calculated compositions of mixtures of large aggregates and rhyodacitic magma Mixture with SO Mixture with 8 vol. % cryst. vol.% cryst.

Lg. Agg.*

Magma\

Agg. %t

30-2 55-2 10-9 3-7 —

960 40 00 00 —

700 24-2 4-3 1-5 39-5

92-0 7-1 0-7 0-2 61

Modal proportions^ Plagioclase Opx + Cpx Fe-Ti oxides

79-1 14-7 5-3

1000

80-8 14-4 4-9

88-9 8-3 2-8

vol.% Liquid Plagioclase Pyroxenes Fe-Ti oxides

00 00

* Average of two large aggregates. f 'True' phenocryst vol.% (4) estimated from aggregate — free phenocryst relation. { 100 x aggregates/(aggregates + magma). § Vol.% in crystal mode.

few per cent of 'true' phenocrysts, mainly plagioclase. Crystal fractionation responsible for glass compositional variation occurred before incorporation of the polycrystalline aggregates or is an artifact of sample preparation. It is not always possible to distinguish 'true' phenocrysts from admixed crystals petrographically. Phenocrysts with complete euhedral outlines (commonly 1-1-5 mm long) should be 'true' phenocrysts, except small elongated crystals (commonly < 1 mm long) that were derived from coarse-grained enclaves (e.g., sample 1206). The effect from the latter crystals on rhyodacite composition probably is negligible because of their rarity. In andesite enclaves, euhedral elongated crystals are embedded within porous glass. It is likely that the high porosity of the enclave mesostasis made it possible for euhedral crystals to disaggregate mechanically. In fact, various degrees of disaggregation of such enclaves from small blobs to discrete crystals are found in a sample from the rhyodacite of Scott Creek (No. 1204).

MAGMA CHAMBER MODEL Compositional and mineralogical continuity among the three rhyodacite groups, and the apparently brief period of emplacement, imply that they erupted from a single chamber. The long prior history of dacitic volcanism, which commonly produced lavas with abundant andesite enclaves suggestive of derivation from small volume systems, indicates substantial time available for generation of rhyodacitic magma without evidence of eruption of more primitive compositions in the area now occupied by the rhyodacites. The several andesite or basaltic andesite cones and flows with K-Ar ages of ~600 ka may reflect a major thermal pulse that led to development of the pre-Mazama rhyodacitic magma body. If the rhyodacite magma erupted from vents rooted in the top of a single magma chamber with dimensions of the lava field, the ~ 2 0 km 3 volume represents a layer with a thickness only of 50 m. If the chamber were 10 km in diameter, the erupted layer would have been 250 m thick. As each rhyodacite group has a volume 400-ka system was


161

crystallizing inward and lacked a sufficient volume of fluid magma to sustain a catastrophic eruption, just as the 25-30-ka (early climactic) chamber failed to produce a major eruption even though it erupted similar lava from vents > 8 km apart. ACKNOWLEDGEMENTS This research was undertaken while S.N. was a visiting scientist at the US Geological Survey (USGS), Menlo Park, with the financial backing of an Overseas Research Fellowship of the Ministry of Education, Science, and Culture of Japan. We thank P. E. Bruggman, M. M. Hirshmann, S. B. McKnight, C. J. Michelsen, E. B. Lougee, M. H. Price, and L. R. Stevens for technical assistance. We thank also D. E. Champion and M. A. Lanphere for permission to use their unpublished paleomagnetic data and K-Ar age determinations, respectively; R. A. Bailey, M. A. Clynne, W. Hildreth, and K. Uto for their discussion during this study; and H. Nekvasil for calculating liquidus feldspar compositions. Reviews by Hildreth, Clynne, and J. B. Lowenstern, and journal reviews by A. Ewart and an anonymous reviewer, helped us to clarify presentation of our ideas. S.N. is grateful to USGS colleagues, particularly G. K. Czamanske for his arrangement of S.N.'s work at USGS and his encouragement, and L. C. Calk for his patient and skilled supervision of the microprobe work. REFERENCES Albee, A. L., & Ray, L., 1970. Correction factors for electron microprobe microanalysis of silicates, oxides, carbonates, phosphates and sulfates. Anal. Chem. 42, 1408-14. Andersen, D. J., & Lindsley, D. H., 1988. Internally consistent solution models for Fe-Mg-Mn-Ti oxides: Part I. Fe-Ti oxides. Am. Miner. 73, 714-26. Arculus, R. J., & Wills, K. J. A., 1980. The petrology of plutonic blocks and inclusions from the Lesser Antilles island arc. J. Petrology 21, 743-99. Bacon, C. R., 1983. Eruptive history of Mount Mazama and Crater Lake caldera, Cascade Range, U.S.A. J. Volcanol. Geotherm. Res. 18, 57-115. 1985. Implications of silicic vent patterns for the presence of large crustal magma chambers. J. Geophys. Res. 90, 11 243-52. 1986. Magmatic inclusions in silicic and intermediate volcanic rocks. Ibid. 91, 6091-112. 1990. Calc-alkaline, shoshonitic, and primitive tholeiitic lavas from monogenetic volcanoes near Crater Lake, Oregon. J. Petrology 31, 135-66. 1992. Partially melted granodiorite and related rocks ejected from Crater Lake caldera, Oregon. Trans. R. Soc. Edin. Earth Sci. 83, 27-47. Adami, L. H., & Lanphere, M. A., 1989a. Direct evidence for the origin of low- 18 O magmas: quenched samples of a magma chamber's partially-fused granitoid walls, Crater Lake, Oregon. Earth Planet. Sci. Lett. 96, 199-208. Druitt, T. H. 1988. Compositional evolution of the zoned calcalkaline magma chamber of Mount Mazama, Crater Lake, Oregon. Contr. Miner. Petrol. 98, 224-56. Hildreth, W., & Druitt, T. H., 1987. Partition coefficients determined from phenocryst and glass analyses of the climatic ejecta of Mount Mazama, Oregon. U.S. Geol. Surv. Open-File Rep. 87-589, 4 pp. Hirschmann, M. M., 1988. Mg/Mn partitioning as a test for equilibrium between coexisting Fe-Ti oxides. Am. Miner. 73, 57-61. Lanphere, M. A., 1990. The geologic setting of Crater Lake, Oregon. In: Drake, E.T., Larson, G. L.,Dymond, J., & Collier, R. (eds.) Crater Lake, an Ecosystem Study. San Francisco: Pac. Div. Am. Assoc. Adv. Sci., 19-27. O'Neil, J. R., 1989ft. Strontium and oxygen isotopes in volcanic rocks near Crater Lake, Oregon, and their bearing on arc magmatism. In: Muffler, L. J. P., & Blackwell, D. D. (eds.) Geological, Geophysical, and Tectonic Setting of the Cascade Range. U.S. Geol. Surv. Open-File Rep. 89-178, 521-55. Bence, A. E., & Albee, A. L., 1968. Empirical correction factors for the electron microanalysis of silicates and oxides. J. Geol. 76, 382-403. Blundy, J. D., & Shimizu, N., 1991. Trace element evidence for plagioclase recycling in calc-alkaline magmas. Earth Planet. Sci. Lett. 102, 178-97. Wood, B. J., 1991. Crystal-chemical controls on the partitioning of Sr and Ba between plagioclase feldspar, silicate melts, and hydrothermal solutions. Geochim. Cosmochim. Ada 55, 193-209. Brandeis, G., & Jaupart, C , 1987. The kinetics of nucleation and crystal growth and sealing laws for magmatic crystallization. Contr. Miner. Petrol. 96, 24-34.

162

SETSUYA NAKADA

ETAL.

Bruggman, P. E., Bacon, C. R., Mee, J. S., Pribble, S. T., & Siems, D. F., 1989. Chemical analyses of volcanic rocks from monogenetic and shield volcanoes near Crater Lake, Oregon. U.S. Geol. Surv. Open-File Rep. 89-562,17 pp. , 1993. Chemical analyses of pre-Mazama silicic volcanic rocks, inclusions, and glass separates, Crater lake, Oregon. Ibid. 93-314, 20 pp. Chou, I-M., 1978. Calibration of oxygen buffers at elevated Pand T using the hydrogen fugacity sensor. Am. Miner. 63, 690-703. de Silva, S. L., 1989. The origin and significance of crystal rich inclusions in pumices from two Chilean ignimbrites. Geol. Mag. 126, 159-75. Druitt,T. H.,& Bacon, C. R., 1989. Petrology of the zoned calcalkaline magma chamber of Mount Mazama, Crater Lake, Oregon. Conlr. Miner. Petrol. 101, 245-59. Fichaut, M., Marcelot, G., & Clocchiatti, R., 1989. Magmatology of Mt. Pelee (Martinique, F.W.I.). II: Petrology of gabbroic and dioritic cumulates. J. Volcanol. Geotherm. Res. 38, 171-87. Garcia, M. O., & Jacobson, S. S., 1979. Crystal clots, amphibole fractionation and the evolution of calc-alkaline magmas. Conlr. Miner. Petrol. 69, 319-27. Gill, J. B., 1981. Orogenic Andesites and Plate Tectonics. New York: Springer-Verlag, 390 pp. Huebner, J. S., & Sato, M., 1970. The oxygen fugacity-temperature relationships of manganese oxide and nickel oxide buffers. Am. Miner. 55, 934-52. Langmuir, C. H., 1989. Geochemical consequences of in situ crystallization. Nature 340, 199-205. Leake, B. E., 1978. Nomenclature of amphiboles. Can. Miner. 16, 501-20. Lindsley, D. H., 1983. Pyroxene thermometry. Am. Miner. 68, 477-93. Linneman,S. R.,& Myers, J. D., 1990. Magmatic inclusions in the HolocenerhyolitesofNewberry volcano, central Oregon. J. Geophys. Res. 95, 17 677-91. Longhi, J., Walker, D., & Hays, J. F., 1976. Fe and Mg in plagioclase. Proc. Lunar Sci. Conf. 7, 1281-300. Mahood, G. A., 1990. Second reply to comment of R. S. J. Sparks, H. E. Huppert & C. J. N. Wilson on "Evidence for long residence times of rhyolitic magma in the Long Valley magmatic system: the isotopic record in the precaldera lavas of Glass Mountain". Earth Planet. Sci. Lett. 99, 395-9. Russell, J. K., & Nicholls, J., 1988. Analysis of petrologic hypotheses with Pearce element ratios. Contr. Miner. Petrol. 99, 25-35. Sato, H., 1989. Mg-Fe partitioning between plagioclase and liquid in basalts of hole 5O4B, O D P Leg 111: a study of melting at 1 atm. In: Becker, K., Sakai, H., et al. (eds.) Proceedings of the Ocean Drilling Program, Scientific Results, HI. College Station, TX: Ocean Drilling Program, 17-26. Scarfe, C. M., & Fujii, T., 1987. Petrology of crystal clots in the pumice of Mount St. Helens' March 19, 1982 eruption; significant role of Fe-Ti oxide crystallization. J. Volcanol. Geotherm. Res. 34, 1—14. Stewart, D. C , 1975. Crystal clots in calc-alkaline andesites as'breakdown products of high-Al amphiboles. Contr. Miner. Petrol. 53, 195-204. Stormer, J. C , 1983. The effects of recalculation on estimates of temperature and oxygen fugacity from analyses of multicomponent iron-titanium oxides. Am. Miner. 68, 586-94. Tait, S. R., 1988. Samples from the crystallizing boundary layer of a zoned magma chamber. Contr. Miner. Petrol. 100, 470-83. Williams, H., 1942. The geology of Crater Lake National Park, Oregon. Carnegie Inst. Wash. Publ. 540, 162 pp.