FIRE and MUD Contents

Magma Mixing at Mount Pinatubo: Petrographic and Chemical Evidence from the 1991 Deposits

By John S. Pallister,1 Richard P. Hoblitt,1 Gregory P. Meeker,1 Roy J. Knight,1 and David F. Siems1

1U.S. Geological Survey.


ABSTRACT

Four principal juvenile magmatic components are present in the 1991 deposits of Mount Pinatubo: hybrid andesite (59-60 weight percent SiO2), olivine-clinopyroxene basalt (50-52 weight percent SiO2) with abundant hornblende microphenocrysts, and phenocryst-rich and phenocryst-poor dacite (both 64.5+-0.3 weight percent SiO2). The first eruptions produced hybrid andesite as lava that fed the June 7-12 dome and then as scoria in the pre-paroxysmal vertical columns and surges of June 12-15. Olivine-clinopyroxene basalt occurs only as undercooled and quenched inclusions in the June 7-12 dome. Dacitic pumice first appeared as a minor component of the vertical eruption on June 12 and increased in abundance in the subsequent 1991 eruptions. Phenocryst-rich and phenocryst-poor dacitic pumice are the primary (~85 percent) and secondary (~15 percent) juvenile components of the June 15 climactic eruption. The phenocryst-poor dacite contains abundant small (<5 micrometers) crystal fragments, most of which were produced by mechanical breakage of pheno-crysts during magma ascent.

The hybrid andesite contains a wide variety of phenocrysts: plagioclase, hornblende, cummingtonite, biotite, augite, olivine, Fe-Ti oxides, quartz, apatite, and anhydrite. Comparison of mineral compositions and textures indicates that the olivine (Fo86-89), clinopyroxene (Mg# 75-85), and many of the hornblende phenocrysts were derived from basaltic magma similar to the inclusions in the June 7-12 dome, whereas the plagioclase, cummingtonite, and anhydrite phenocrysts are from June-15 type dacitic magma. Ilmenite grains in the andesite were probably derived from the dacitic magma, as indicated by similarity in composition to those in the June 15 dacite and by the presence of magnetite rims. Olivine-melt equilibria indicate the basaltic magma was relatively primitive (Mg# 68) and hot (~1,200°C) prior to mixing. Fe-Ti-oxide thermometry and cummingtonite stability indicate that dacite was relatively cool (~780°C) and highly oxidized (nickel-nickel oxide +3 log units) prior to mixing. The magnetite-rimmed ilmenites in the andesite yield intermediate Fe-Ti oxide temperatures (~950°C) and lower oxygen fugacity (nickel-nickel oxide +2 log units) than the dacite does. Evidence of disequilibrium in the hybrid andesite includes the wide variety and varied compositions of minerals and matrix glass and the presence of magnetite rims on ilmenites, hornblende rims on olivine, and clinopyroxene rims on quartz. Preservation of these disequilibrium features indicates that little time passed between magma mixing and eruption.

Origin of the andesite by magma mixing is also indicated by major and trace element compositions of pumice and scoria samples and by variable matrix glass compositions. Mass balance indicates that the andesite represents a 64:36 mix of dacite and basalt. Phase equilibria and seismic data suggest that magma mixing took place at about 8 to 9 kilometers in depth, within a large (>50 cubic kilometers) magma reservoir beneath the volcano. Buoyant ascent of hybrid andesitic magma in early June established a conduit to the surface and thereby triggered the eruption of about 5 cubic kilometers of the relatively cool, fluid-saturated dacitic magma on June 15. Without addition of basalt, the viscous, crystal-rich dacite would probably not have erupted in 1991.

Basalt may also be the ultimate source for 20 megatonnes of SO2 that was added to the stratosphere by the June 15, 1991, eruption. We suspect that the high-level dacitic part of the >50 cubic kilometer magma reservoir was enriched in sulfur by multiple episodes of magma mixing in the past. Similar intrinsic conditions for magmas at Pinatubo and El Chichón suggest that volcanoes that give rise to sulfur-rich explosive eruptions are underlain by long-lived magma reservoirs with relatively cool, oxidized, fluid saturated, and crystal-rich upper zones that act as traps for volatiles that are supplied by basaltic magmas from greater depth.

Note to readers: Figures and tables open in separate windows. To return to the text, close the figure or table's window or bring the text window to the front.

INTRODUCTION

In a preliminary report (Pallister and others, 1992), we described evidence for mixing of basaltic and dacitic magmas shortly before the climactic eruption of Mount Pinatubo on June 15, 1991. Herein, we present more detailed petrographic and geochemical data for products of the 1991 eruptions; these data strongly confirm the magma mixing hypothesis. Not only was the mixed magma erupted, but both endmembers are also present in the 1991 deposits. We previously suggested that subordinate "crystal-poor dacite" of the June 15 eruption was produced by disequilibrium melting of phenocrysts during basalt underplating of crystal-rich dacite. In this report, we present new electron-microbeam images that show the so called crystal-poor dacite of June 15 to contain abundant micrometer-sized crystal fragments. Accordingly, we now refer to this pumice as phenocryst-poor instead of crystal-poor, and we favor a dominantly mechanical process of crystal size reduction. We follow Wilcox (1954) in placing the distinctions between phenocrysts and microphenocrysts at 0.3 mm (longest dimension) and in placing the distinctions between microphenocrysts and microlites at 0.03 mm. Analytical methods employed in this study are described in the appendix.

ERUPTIVE SEQUENCE

The 1991 eruptive sequence was reviewed by Wolfe (1992) and is presented in more detail in other chapters of this volume. In the discussion that follows, we will review only those events that are most relevant to the petrology of the deposits. The 1991 eruptions ended a 500-year hiatus in eruptive activity at Pinatubo volcano (Newhall and others, this volume). In contrast to the common silicic to mafic eruptive sequence observed at many volcanoes, the 1991 eruptions tapped andesite before dacite. The initial 1991 eruptions formed an andesitic lava dome at the head of the Maraunot River valley (fig. 1; Hoblitt, Wolfe, and others, this volume), beginning on June 7. The andesitic lava dome continued to grow until 0850 on June 12, when the first large explosive eruption destroyed the southern part of the dome and generated a vertical column that rose to more than 19 km in altitude. Ash from the column was transported mainly southwest from the volcano, and small pyroclastic flows descended the upper Maraunot River valley. The northern part of the lava dome survived the first June 12 eruption as well as vertically directed explosive eruptions of andesite beginning at 2251 on June 12, 0840 on June 13, and 1307 on June 14.

Figure 1. Distribution of 1991 pyroclastic-flow deposits from Mount Pinatubo (from W.E. Scott and others, this volume), outline of 1991 caldera (hachured), and sample localities for this study. Sample localities indicated by dots. Field photograph localities indicated by figure numbers. Small arrowhead shows perspective of oblique aerial photograph in figure 4. Areas outside of shaded region were blanketed by tephra fall.

A series of at least 13 lateral blasts occurred on June 14 and 15 (Hoblitt, Wolfe, and others, this volume). Proportions of clast types in the resulting blast deposits (fig. 2) indicate that the dominant magmatic component changed from andesite to dacite during this interval. The lateral blasts became more frequent and release of seismic energy increased exponentially during the morning hours of June 15, leading to the climactic eruption, which was under way by early afternoon. Tephra-fall deposits from the June 12-14 vertical eruptions are considerably thinner and are more areally restricted than the June 15 fall deposit (fig. 2A). Juvenile components in the June 12-14 tephra deposit are dominantly andesitic scoria. Dacite makes up less than 10 percent of juvenile clasts in the 0851 June 12 tephra, 28 percent of the 2252 June 12 tephra, and increases to 35 percent in the June 13 tephra (Hoblitt, Wolfe, and others, this volume). The June 13 deposit is recognized in proximal sections because it overlies an ash bed that was flushed by rain that fell during the night of June 12 (fig. 2B). Taken together, the June 12-14 eruptions involved at least an order-of-magnitude less magma than the June 15 eruption (Paladio-Melosantos and others, this volume). Estimates of 3.7 to 5.3 km3 (W. E. Scott and others, this volume) or possibly more (Koyaguchi, this volume) have been made for the volume of dacitic magma that was vented on June 15. The June 15 eruption resulted in collapse of a caldera 2.5 km in diameter (Wolfe and others, this volume; W.E. Scott and others, this volume), which decapitated the upper part of the Maraunot drainage and destroyed the June 7-12 lava dome.

Figure 2. A, Section of 1991 tephra-fall deposits at bridge abutment, Santo Tomas River (fig. 1). Lower, dark layer at base is organic-rich pre-1991 soil, the thin lapilli-rich layer at center is the June 12, 1991, fall deposit, which is overlain by 3-cm-thick laminated ash from the June 14-15 June blasts, which is itself overlain by the white June 15 pumiceous fall deposit. B, Dark pre-1991 soil overlain first by June 12 and 13 tephra, both rich in andesitic scoria lapilli, then by silt-to lapilli-rich blast beds of June 14-15, which contain both andesitic scoria and dacitic pumice. The blast beds are truncated and overlain by a white June 15 pyroclastic-flow deposit. The feature above and to the left of the scale is a lapilli-filled trough, possibly produced by lateral drag of an object across the surface during blast emplacement. Smaller divisions on the scale are 1 cm long.

SAMPLE LOCALITIES AND GEOLOGIC CONTEXT

Pinatubo volcano lies in the Bataan frontal arc, approximately 120 km inboard of the Manila trench and 100 km above the Wadati-Benioff zone (Defant and others, 1988; Newhall and others, this volume). The volcano was constructed atop the southern part of the broadly exposed (3,500 km2), east-dipping Zambales ophiolite complex (Evans and others, 1991). The 1991 caldera is nested within a 3x5-km caldera formed earlier in the volcano's history (fig. 1). Exploration drilling indicates that the Zambales ophiolite complex makes up the southeast wall of the older caldera and that, if present, the ophiolite must underlie more than 2 km of dacitic rocks that were drilled within the caldera (Delfin, 1983; Delfin and others, this volume).

Our samples of the June 7-12 dome were taken from prismatically jointed and scoriaceous blocks in a series of pyroclastic-flow deposits in the headwaters of the Maraunot River (figs. 1 and 3). Two pyroclastic-flow deposits are visible at the sample locality (fig. 3), and both contain blocks of dome rock. The lower deposit is correlated with the 0841 eruption of June 13 on the basis of similarity in clast-type distribution to that seen in the tephra-fall deposits (Hoblitt, Wolfe, and others, this volume). An underlying pyroclastic-flow deposit exposed upstream (and not visible in fig. 3) is correlated with the 0851 eruption of June 12; it is overlain and underlain by thin ash beds and contains breadcrust-surfaced blocks of andesitic scoria similar to lapilli in the fall deposit. The upper deposit in figure 3 contains abundant prismatically jointed dome blocks, possibly mined from the June 7-12 lava dome during caldera collapse on June 15.

Figure 3. Pyroclastic-flow de- posits in the upper Maraunot River valley. Photograph of the outcrop was taken on February 28, 1992. Prominent bouldery layer is a June 15 pyroclastic-flow deposit, rich in blocks from the June 7-12 lava dome; it is underlain by a pyroclastic-flow deposit that is correlated with the 0841 eruption of June 13. A June 12 pyroclastic-flow deposit underlies both of these deposits upstream (out of view to the right).

Tephra-fall deposits from the June 12-13 vertical eruptions are exposed downwind from the volcano, to the west and southwest. At distances of less than 10 km from the caldera, June 15 pyroclastic flows scoured to bedrock much of the west flank of the volcano (fig. 4). Consequently, proximal June 12-14 fall and blast deposits are preserved only in the lee of bedrock ridges. Our most detailed petrographic and microprobe work on the June 12 tephra was done on scoria lapilli (sample 7-1-91-1A) collected on July 1, 1991, from near the axis of the June 12 fall deposit near Mount Bagang, 16 km southwest of the caldera (fig. 1). Another scoria sample collected from the same area by A.G. Reyes is petrographically and chemically similar, as are scoria lapilli from several proximal tephra sections that we collected in 1992.

Figure 4. View up southwest flank of Mount Pinatubo showing dissected 1991 pyroclastic-flow deposits. Photograph taken March 2, 1992, from a position 5.5 km from southwest caldera rim. Banana plants grew from root stock remaining on lee sides of ridges (facing camera). Fore sides of ridges were scoured to basement by the pyroclastic flows.

We sampled pumice blocks and lapilli from the June 15 eruption at a number of localities around the volcano. Our most detailed petrographic and microprobe data are for several samples collected in June and July of 1991, shortly after the climactic eruption. Subsequent petrographic examination and bulk rock analysis of a larger suite of samples collected in 1992 indicate that our 1991 samples are representative of the principal eruptive products.

PETROGRAPHY AND MINERAL CHEMISTRY

Modal data for the 1991 magmatic samples are summarized in table 1, and petrographic descriptions are given herein. Modes were counted optically at a magnification of 400 by use of reflected and transmitted light on doubly polished thin sections. The presence of extremely fine (<1 lc mum) glass filaments, microlites or very small (<5 lc mum) crystal fragments, and low contrast between epoxy mounting medium and glass in some samples made quantitative optical analysis of groundmass phases and vesicle content difficult. As a result, the abundances of "groundmass" in table 1 include both glass and microlites (or microlite-sized broken crystal fragments), and the abundance of vesicles is approximate.

Table 1. Modal data for 1991 Pinatubo samples based on optical point counting of representative thin sections at a magnification of 400 in reflected and transmitted light (in volume percent, vesicle free).
[Groundmass consists mainly of glass in samples EW910615-1, 7-1-91-1A, and CN6791-i, and glass with abundant microlites in samples EW910615-2, PH-13d, and CN6791-h. Tr., trace amount; ( ), highly resorbed xenocrysts; dash indicates none detected; PR, phenocryst rich; PM, phenocryst moderate; PP, phenocryst poor]

Chemical variations of the principal mineral phases are shown graphically in figure 5 in terms of An (anorthite) content of plagioclase, Fo (forsterite) content of olivine, and Mg# for pyroxenes and hornblende.2 Microprobe analyses for average mineral compositions are listed in table 2.


2 Magnesium numbers (Mg#) are calculated as 100 times the molecular ratio of Mg/(Mg+Fe+2+Mn).

Figure 5. Average composition of principal minerals in samples of 1991 Pinatubo eruptive products, expressed as An content of plagioclase, Fo content of olivine, and Mg# of augite, hornblende, and bronzite. Analytical data for cummingtonite and biotite are summarized in table 2 and oxide data are given in table 5. The average compositions and +-1lc sigma range given are from table 2. Symbols: PR, phenocryst-rich dacite of June 15; PP, phenocryst-poor dacite of June 15; SC, scoria from air-fall tephra of June 12; AN, dense andesite lava from June 7-12 dome; BA, basalt inclusion from June 7-12 lava dome; m, microphenocryst or microlite; p, phenocryst; g, glomerocryst; r, reaction rim; x, xenocryst.

Table 2. Average compositions, standard deviations, and maximum and minimum values of mineral endmember components for minerals in 1991 Pinatubo samples.
[Microprobe analytical procedures given in the appendix. Avg., average; n, number of analyses averaged; SD, standard deviation; qz, quartz; hb, hornblende; cm, cummingtonite; PR, phenocryst-rich; PP, phenocryst poor; PM, phenocryst moderate. Dash indicates not determined]
Page 1 | Page 2 | Page 3 | Page 4 | Page 5 | Page 6

JUNE 7-14 HYBRID ANDESITE

Samples of the June 7-12 lava dome, taken from the Maraunot River pyroclastic-flow deposits, are remarkable both in the variety of phenocrysts present and in the development of disequilibrium textures. These andesite samples contain plagioclase, hornblende, augite, olivine, quartz, anhydrite, magnetite, ilmenite, cummingtonite, and biotite (table 1, fig. 6A). These phenocrysts are set in a pilotaxitic (microlite-charged) groundmass (fig. 6B). Andesitic scoria from tephra-fall deposits of June 12 are identical to the dome andesite samples with respect to phenocryst assemblages, mineral compositions, bulk compositions, and several reaction textures. The most distinct differences between the dome and tephra samples lie in the greater vesicularity of the scoria (table 1) and in the degree of groundmass crystallization. In contrast to the pilotaxitic texture of the dome andesite, the June 12 scoria has a glassy microvesicular groundmass.

Figure 6. Microscopic images from thin sections of June 7-12 hybrid andesite. Symbols: T, transmitted light; R, reflected light; X, cross polarized; P, plane polarized; xx mm, width of frame in millimeters. AG, augite; AHY, anhydrite; AP, apatite; GL, glass; HB, hornblende; OL, olivine; PL, plagioclase; QZ, quartz; RXN, reaction rim; VES, vesicle; MT, magnetite; IL, ilmenite; SP, chrome spinel. A, Photomicrograph showing varied mineral assemblage in June 7-12 dome andesite sample CN6791-d (T, X, 5.5 mm). B, Oscillatory and normally zoned augite (Mg# 76-86) microphenocryst set in a pilotaxitic, microlite-charged groundmass. Section is perpendicular to the c-axis (T, P, 0.34 mm). C, Fo87 olivine, rimmed by hornblende in andesitic scoria sample 7-12-91-1A from the June 12, 1991, tephra-fall deposit (T, P, 1.3 mm). D, Fo87 olivine showing euhedral prismatic crystal form and thin hornblende rims in June 7-12 lava dome sample CN6791-1d (T, P, 2.7 mm). E, Glomerocrystic clot of Fo88 olivine and Mg# 81 augite with hornblende rim in June 7-12 lava dome sample CN6791-1d (T, P, 2.7 mm). F, Resorbed low-Al-Ti hornblende with oscillatory-zoned high-Al-Ti overgrowth in June 7-12 andesite scoria sample 7-1-91-1A (T, P, 0.68 mm). Small augite inclusions decorate the margin of the resorbed core. Three TiO2 values are indicated (wt%). G, Oscillatory-zoned plagioclase with calcic rim and glass inclusions along resorption boundary, June 12 scoria sample 7-1-91-1A. Numbers give An contents at analyzed spots (T, X, 0.68 mm). H, Oscillatory zoned plagioclase with calcic rim in June 12 scoria sample 7-1-91-1A. Numbers give An contents at analyzed spots. Note wide range in composition between zoning bands (T+R, X, 0.68 mm). I, Euhedral anhydrite phenocryst with apatite inclusions in June 12 scoria sample 7-1-91-1A (T, P, 1.3 mm). J, Resorbed anhydrite microlites with apatite inclusions in June 7-12 lava dome sample CN6791-1d (T, P, 0.34 mm). K, Ilmenite grain with magnetite overgrowth on the margin in contact with groundmass glass (but not on the margin embedded in hornblende), June 12 scoria sample P22892-2A4 (T+R, P, 0.13 mm). L, Back-scattered electron image of ilmenite grain with magnetite overgrowth in June 12 scoria sample 7-1-91-1A. Bright areas indicate higher atomic number (greater iron content). Note zoning to lower Fe-ilmenite within about 10 lc mum of contact with magnetite rim. Line is 100 lc mum long; frame width is about 0.3 mm.
Figure 6A-F
Figure 6G-L

Because the dome and tephra samples share common petrographic and compositional features and are believed to represent the same magma, they are described together here. Olivine phenocrysts in the hybrid andesite are rimmed by aggregates of hornblende laths (fig. 6C). Individual hornblende laths in the rims are commonly 0.05 mm across and 0.2 mm in length, comparable in size to hornblende microphenocrysts in the groundmass of the andesite and similar in composition to the more magnesian of the hornblende phenocrysts and microphenocrysts (table 2). Two types of olivine phenocrysts are observed in the dome andesite, and both are also present in the scoria. The first type is prismatic and subhedral to euhedral, consisting of single millimeter-sized crystals with optical continuity throughout, such as seen in figure 6C and D. The second type is glomerocrystic, consisting of aggregates of smaller olivine grains or, less commonly, olivine and clinopyroxene grains (fig. 6E). Both types of olivine have the same restricted range in composition (Fo86-89; fig. 5).

Initially, we were concerned that the Pinatubo olivines might be xenocrysts from the Zambales ophiolite complex beneath the volcano, a possibility also considered by Fournelle (1991) and Fournelle and others (this volume). However, spinel inclusions within the olivines are intermediate pleonaste-chromite solid solutions with relatively high Fe2O3 and intermediate TiO2, typical of abundances in arc basalts and unlike the spinels in most ophiolitic peridotites, including the Zambales (compare Haggerty, 1976; Dick and Bullen, 1984; Evans and Hawkins, 1989). The TiO2 abundances of the spinel inclusions overlap with those in the cumulus (gabbroic) suite of the ophiolite (Evans, 1987), but high Fe2O3 distinguishes the volcanic spinels from those in the ophiolite. In addition, low nickel and high calcium abundance, subhedral prismatic form of many crystals, presence of glass inclusions, and presence of oscillatory zoned magnesian augite in the same samples indicate that the olivine is of volcanic origin. Additional proof became available in 1992 when we found basalt inclusions with identical magnesian olivine and augite phenocrysts.

Clinopyroxene is present as small (<0.5 mm), stubby euhedral prisms in the hybrid andesite. Micrometer-scale oscillatory zoning is clearly visible in transmitted light, especially when crystals are viewed down the c-axis (fig. 6B). Most crystals show normal overall zoning; Mg# ranges from as high as 88 for cores to as low as 75 for crystal edges and averages about 81. Sector-zoned augite microphenocrysts are present in some samples. The augite overlaps in other major and minor element abundances with oscillatory-zoned augites in the basalt inclusions (table 2).

Hornblende occurs as small (<1 mm) phenocrysts and microphenocrysts, larger (1-2 mm) phenocrysts, sparse xenocrysts with augite rims, and (as noted previously) in reaction rims surrounding olivine. Many phenocrysts are oscillatory zoned, and some grains have resorbed inclusion-rich cores that are overgrown by oscillatory-zoned margins with higher Ti, Al, and Mg# (fig. 6F). Glass inclusions with vapor bubbles are common in the phenocrysts. The hornblendes in the scoria are olive green to brown; those in the dome andesite are red-brown to tan, indicative of late oxidation. Dehydration rims are lacking on most of the hornblendes, so rapid ascent is suggested (compare Rutherford and Hill, in press). Compositions of hornblende phenocrysts in the andesitic scoria show a considerable range in Mg#, overlapping with those in both the andesitic lava and June 15 dacite (fig. 5). The hornblendes in both types of andesite, however, are distinct from those in the June 15 dacite in other major and minor element abundances. In particular, the analyzed phenocrysts in the andesite have significantly higher Al, Ti, Na, and K, and lower Si and Mn (table 2). In contrast, they overlap in all major and minor element abundances with hornblende phenocrysts and microphenocrysts in the basaltic inclusions. Therefore, we believe that most were derived from a mafic source similar to the basaltic inclusions.

Cummingtonite is a minor phase in the andesites. It occurs as partial rims on a few broken hornblende phenocrysts. Surprisingly little, if any, reaction of cummingtonite with the host melt is evident. Biotite occurs mainly as cores to hornblende phenocrysts and less commonly as separate small grains in the groundmass of the andesitic scoria.

Plagioclase occurs as large (1-10 mm) oscillatory-zoned, euhedral or broken phenocrysts in the hybrid andesite. Those in the dome lava tend to be larger and less commonly broken than those in the scoria. Small glass inclusions (most <30 lc mum) with vapor bubbles are common within interior regions of the phenocrysts as well as being concentrated with microlites of apatite and other phases to form dusty resorption zones near the crystal margins (fig. 6G,H). Compositions of the plagioclase phenocrysts in the andesite average about An40 and range widely (An28-64). Although the average compositions overlap with those in the June 15 dacite, phenocrysts in the andesite extend to more calcic compositions, owing to thin (10-30 lc mum) rims of calcic plagioclase (An58-65) over more sodic oscillatory-zoned interiors (typically An30-40 or An35-55) (fig. 6H). Microlites in the andesites are relatively calcic, with average compositions of about An60, similar to the calcic rims on the phenocrysts (fig. 5).

Quartz is a minor component of the hybrid andesite. It forms resorbed anhedral to subhedral grains that are typically 1 to 5 mm in diameter. These grains also contain glass inclusions with vapor bubbles, and some of the inclusions exceed 100 lc mum in diameter. Apatite is a ubiquitous microlite phase in all of the 1991 Pinatubo rocks; it occurs as inclusions in plagioclase, hornblende, and anhydrite. Apatite also forms sparse small phenocrysts and microphenocrysts, both as isolated euhedra in the groundmass and in association with anhydrite, as described below.

Anhydrite forms euhedral phenocrysts in the June 12 scoria. The anhydrite phenocrysts are up to 1 mm in length; they commonly include apatite microphenocrysts (fig. 6I) or are in growth contact with apatite phenocrysts. In contrast, anhydrite in the June 7-12 dome andesite occurs as highly resorbed anhedral masses with dusty reaction rims that contain oxide and sulfide minerals (fig. 6J). We attribute the difference to reaction of anhydrite with host melt that had cooled slowly and degassed near the surface.

Both ilmenite and magnetite occur as small (0.01-0.5 mm) grains in the hybrid andesites. Some of the magnetites in the hybrid andesite contain minute sulfide blebs (<20 lc mum), similar to those in magnetites of the basaltic inclusions. The ilmenites occur only as core regions of composite grains with magnetite borders. As is evident in figure 6K, the magnetite borders are overgrowths that nucleated and grew on ilmenite grain margins that were in contact with the hybrid melt. Where the ilmenite grain margins were embedded in silicate minerals no magnetite crystallized. Ilmenite cores of composite grains in scoria from the June 12 tephra preserve weak zonation to higher titanium within 5 to 10 lc mum of magnetite overgrowths (fig. 6L). Ilmenite cores to composite grains in the June 7-12 dome rocks contain exsolution-oxidation lamella of hematite, consistent with slower cooling of the dome samples. The compositions of the oxides, and implications for temperature and oxygen fugacity, will be discussed in a subsequent section.

BASALTIC INCLUSIONS FROM THE JUNE 7-12 DOME

Inclusions of olivine-augite basalt with abundant hornblende microphenocrysts are common in blocks of the June 7-12 dome andesite from the upper Maraunot River valley (fig. 7). These inclusions range in size and shape from relatively small (millimeter scale), irregular blebs identifiable in thin section to ellipsoidal bodies up to 40 cm across. Most macroscopic ones are ellipsoidal, have sharp margins, and are 5 to 10 cm in diameter. Basaltic inclusions with irregularly deformed margins (fig. 8A) and banded basalt-andesite scoria blocks were found in the 0851 June 12 pyroclastic-flow deposit from the upper Maraunot valley. Banded scoria lapilli with dark bands rich in olivine and clinopyroxene are also common in the June 12-14 scoria-fall deposits.

Figure 7. Boulder derived from the June 7-12 lava dome within the June 15 pyroclastic-flow deposit, upper Maraunot River valley near locality P22892-1. Arrowheads indicate a few of the largest of the relatively abundant inclusions of dark basalt. Hammer is 37 cm long.

Figure 8. A, Deformed basalt inclusion with irregular margin in andesitic scoria from June 12 pyroclastic-flow deposit. B, Oval-shaped undercooled inclusion (sample CN6791-i) in dense andesitic lava from the June 7-12 dome. Both samples are from pyroclastic-flow deposits in the upper Maraunot River valley. Labeled quartz (QZ) and hornblende (HB) are xenocrysts discussed in text.

We describe here a single 8-cm-diameter ellipsoidal basalt inclusion within dense dome andesite (sample CN6791-i, fig. 8B). This inclusion is representative of many others found in the Maraunot dome blocks. The most obvious distinguishing feature of the inclusion is the absence of the large plagioclase phenocrysts that are common in the andesitic host rock and the presence of a diktytaxitic groundmass texture.

The basalt contains 0.5- to 2-mm olivine phenocrysts and 0.1- to 0.5-mm augite phenocrysts and microphenocrysts. The larger size of the olivine crystals suggests that they were the initial liquidus phase in the basalt. Hornblende occurs as abundant microlites and as small (0.1-0.7 mm) subhedral to euhedral phenocrysts. Xenocrysts of hornblende and quartz are present in some thin sections, and both phases are rimmed by augite. Compositions of all three phenocryst phases and crystal habits of the olivine and augite are identical to those in the hybrid andesite. Olivine (Fo86-88; fig. 5, table 2) is rimmed by hornblende microphenocrysts (fig. 9A) and augite (Mg# 69-85) shows micrometer-scaled oscillatory zoning (fig. 9B). As in the andesite, the olivine occurs both as subhedral prisms and as glomerocrysts with augite. Spinel inclusions within the olivines are likewise similar in composition to those in the olivines of the hybrid andesite (table 2). These similarities leave little doubt that the basaltic inclusions represent samples of the source magma for the olivine and augite phenocrysts in the hybrid andesite.

Figure 9. Microscopic and hand-sample images of basaltic inclusions from June 7-12 lava dome. Symbols: T, transmitted light; R, reflected light; X, cross polarized; P, plane polarized; xx mm, width of frame in millimeters; AG, augite; HB, hornblende; (HB), resorbed hornblende xenocryst; OL, olivine; PL, plagioclase; QZ, quartz; VES, vesicle; MT, magnetite; GL, glass. A, Fo87 olivine phenocryst with hornblende rim in sample CN6791-i (T, P, 2.7 mm). B, Oscillatory and normally zoned augite (Mg# 70-85) microphenocryst in sample CN6791-i (T, P, 0.34 mm). C, Diktytaxitic textured basaltic inclusion CN6791-i showing acicular hornblende and plagioclase microphenocrysts, augite microphenocrysts and small phenocrysts, olivine phenocrysts with hornblende rims, and irregular vesicle space (T, X, 5.5 mm). D, Basaltic inclusion sample P22892-1a showing sparse spherical to irregular vesicles developed in porous diktytaxitic groundmass. Large white grains are quartz and feldspar xenocrysts. E, High-Al-Ti hornblende xenocryst with clinopyroxene reaction rim, basaltic inclusion sample CN6791-1i. Relict hornblende composition is similar to unreacted groundmass hornblende (table 2) (T, P, 5.5 mm). F, Margin of resorbed hornblende xenocryst from E, showing inner (Mg# 80) and outer (Mg# 66) rims of irregular, then fibrous, augite (T, P, 1.3 mm). G, Hollow, glass-inclusion-rich plagioclase microlites, interstitial glass, and void space in basaltic inclusion sample CN6791-1i. Note also hopper-shaped plagioclase microlite at center frame (T, P, 0.34 mm). H, Swallow-tailed hornblende phenocryst in basaltic inclusion CN6791-1i (T, P, 2.7 mm). I, Back-scattered electron image of mottled plagioclase and 73 wt% SiO2 matrix glass in basaltic inclusion sample CN6791-1i. Scale given by 5-lc mum bar, frame width about 100 lc mum. Numbers give An content of plagioclase. Dark pits in matrix glass locate areas analyzed (areas damaged by the electron beam). J, Skeletal magnetite grain in basaltic inclusion sample CN6791-1i (R+T, P, 0.25 mm).
Figure 9A-F
Figure 9G-J

A two-stage crystallization history is evident from the phenocrysts. Olivine and augite phenocrysts formed during an initial stage of anhydrous crystallization. That no hydrous phases were stable is indicated by reaction rims of augite on large hornblende xenocrysts (fig. 9E,F). These reaction rims are composed of two zones of augite microphenocrysts. The outer zone consists of acicular crystals (Mg# 66-68) oriented parallel to the elongation of the relict amphibole. The inner zone is composed of optically continuous and more magnesian augite (Mg# 75-79) that is intergrown with oxide minerals. Major and minor element abundances of the reaction-rim augites overlap with those of the zoned augite phenocrysts. Compositions of the resorbed cores of the amphibole xenocrysts are similar to late-crystallized hornblende microlites (table 2). If they had been included after the magma was hydrated, they would not have been resorbed, nor would augite rims have formed. Consequently, the initial phase of crystallization involved only the anhydrous phases olivine and augite.

Most of the quartz xenocrysts are relatively large (1-4 mm in diameter) and subrounded. As is the case with quartz in the andesite, these are also rimmed by anhydrous, augite-rich reaction zones, and, like the hornblende xenocrysts, they were probably also included at an early stage. Pyroxene-rich reaction boundaries form on quartz grains in static dissolution experiments when the basalt host is below its liquidus, and they effectively armor the quartz from further dissolution (Donaldson, 1985). Accordingly, it is difficult to estimate how long the quartz grains were included in the basaltic magma.

The second stage of crystallization took place after the fugacity of water built up in the melt. Hornblende crystals nucleated and grew to form borders surrounding olivine phenocrysts. Hornblende, plagioclase, and magnetite microphenocrysts and microlites crystallized to produce the diktytaxitic groundmass texture and drive the matrix melt to an evolved rhyolitic composition (74 wt% SiO2). The degree of undercooling increased during the final stage of crystallization. Rapid growth of elongate crystal faces produced glass-inclusion-rich, hopper-shaped plagioclase microlites (fig. 9G) and swallow tails on hornblende microphenocrysts (fig. 9H). Mottled plagioclase microlites, consisting of irregular zones of variable An content, grew quickly and included irregular melt channels (fig. 9I). Skeletal magnetite grains also grew during this period of strong undercooling (fig. 9J). A trace amount of magmatic copper sulfide was trapped as small (<15 lc mum) blebs, typically within or adjacent to larger magnetite grains but also within hornblende, and rarely as free grains in glass.

The diktytaxitic groundmass texture probably formed when the basalt magma was intruded into cooler dacitic to andesitic magma. This texture is characterized by abundant irregular and interconnected void space into which small crystals of plagioclase and hornblende protrude (fig. 9B,C). In addition to the irregular network of diktytaxitic voids, there are sparse 1- to 2-mm spherical vesicles (fig. 9D) that are most common within cores of the inclusions. This implies two-stage volatile exsolution. Perhaps, initial exsolution of a supercritical fluid phase created the diktytaxitic interstices, and then saturation and separation of a vapor phase formed the spherical vesicles. The diktytaxitic voids must have been pressurized for the spherical vesicles to have developed in what would otherwise have been an open medium.

The inclusions lack well-defined chilled margins, possibly because it was not possible to establish strong thermal gradients in such relatively small bodies. Alternatively, magmatic erosion may have removed chill margins. The presence of groundmass glass, undercooling textures, vesicle concentrations, deformed margins, and banded scoria provide strong evidence that the basalt was magmatic when it was included in the andesite. For comparison, similar textural evidence from a variety of mafic inclusions is reviewed by Bacon (1986) and Stimac and others (1990) as evidence for mingling of magmas.

JUNE 15 DACITIC PUMICE

Dacitic pumice from the June 15 climactic eruption can be divided into two principal types: white phenocryst-rich, and tan phenocryst-poor. There is considerable variation in phenocryst content, ranging from about 50 percent to as little as 15 percent in optical modes (table 1). This variation principally reflects smaller crystals in the tan pumice, as many small crystal fragments and microlites are difficult to discern optically and are included in the groundmass fraction in optical modes. Back-scattered electron images of pumice bubble walls reveal abundant small (<10 lc mum to <1 lc mum) crystal fragments, including plagioclase fragments with compositional zones that are truncated at grain margins (fig. 10A,B). Preliminary analysis of microprobe elemental distribution maps of phenocryst-rich and phenocryst-poor pumice sections (fig. 11A,B) suggests that the total crystal abundances of the two types approach one another when microscopic crystal fragments in the groundmass are taken into account (fig. 11; Meeker and others, 1993).

Figure 10. Back-scattered electron images of crystal-fragment-charged matrix glass filaments in phenocryst-poor dacite pumice sample PH13d from a pyroclastic-flow deposit. Image density inversely proportional to mean atomic number (lighter areas are higher atomic number). Small white grains are amphibole, lighter gray fragments are plagioclase, and darker gray material between grains is glass. A, A relatively large (15 lc mum) plagioclase fragment with arcuate internal fracture and faint composition zonation band (at lower right), surrounded by numerous small (<2 lc mum) crystal fragments. Fragment at upper edge of feldspar (at arrowhead) can be fit back onto main grain. B, Enlargement of bubble wall, which is charged with crystal fragments. Light (higher calcium) compositional zone in large plagioclase grain truncated at grain margin. Scale bar in A is 10 lc mum; bar in B, 1 lc mum.

Figure 11. Wavelength-dispersive X-ray intensity maps showing aluminum distribution in 1-cm2 thin-section areas of phenocryst-rich (A) and phenocryst-poor (B) pumice. Maps were generated by driving the microprobe stage over a 1,000x1,000 point grid with 10-lc mum node spacing and 20-ms dwell time and storing the data in a 1,000¥1,000 pixel image. Individual pixels in the image represent activation areas of ~3 lc mum2. White grains are plagioclase, dark gray grains are hornblende, gray network is matrix glass, and black areas represents epoxy-filled vesicles. Abundance of plagioclase is similar in both A and B on a vesicle-free basis, as determined from pixel counts.

Sparse, large (1-2 mm) plagioclase crystals in the phenocryst-poor pumice are rounded, with oscillatory zones that are truncated at the rounded margins (fig. 12A,B). Secondary electron images of external surfaces of these crystals are pitted, a feature we initially interpreted as solution pitting (Pallister and others, 1992). However, rounding of these large crystals could also be related to a physical process of abrasion and fragmentation, just as the larger and more equant feldspar fragments in the matrix of figure 10 have been broken into smaller and more angular fragments. But, given the evidence of extreme fragmentation seen in the groundmass, how did the large crystals survive? They could be relicts of "outsized" megacrysts (to 1 cm) that are observed in some of the phenocryst-rich pumice blocks. Alternatively, the large crystals could have been entrained at a late stage of the fragmentation process.

Figure 12. Microscopic images of plagioclase phenocrysts in June 15 dacitic pumice. Abbreviations as in figure 6. A, Large, rounded plagioclase in phenocryst-poor pumice. Normarski differential contrast illumination, showing oscillatory zoning profiles. Note termination of zoning bands at rounded crystal margins (R, X, 2.7mm). B, Rounded plagioclase crystal in phenocryst-poor pumice. Note small apatite (AP) and bubble-bearing glass inclusions (T, P, 1.3 mm). C, Euhedral, oscillatory-zoned, plagioclase phenocryst in phenocryst-rich pumice. Normarski differential contrast illumination showing zoning profiles and internal resorption discontinuities (R, X, 2.7 mm). D, Back-scattered electron image of oscillatory zoning in plagioclase crystal from phenocryst-rich pumice sample PH13a. Scale given by 10-lc mum bar; width of field about 80 lc mum.

The phenocryst-rich pumice typically has large (>1 mm) stretched vesicles. In contrast, the phenocryst-poor pumice has smaller (<1 mm), spherical vesicles and forms breadcrust-surfaced blocks. Inclusions of the white phenocryst-rich pumice in the tan phenocryst-poor pumice are relatively common, and the two types are mingled together in banded pumice blocks. Densities and relative abundances of phenocryst-rich and phenocryst-poor pumice lapilli in 21 bulk samples from June 15 pyroclastic-flow deposits are summarized in table 3. There is considerable variation in the relative abundance from site to site, possibly due to flow segregation. On average, 85 percent (by volume) of the pumice lapilli in these samples is the phenocryst-rich type and 15 percent is phenocryst-poor, similar to our previous estimate of 80 percent and 20 percent (Pallister and others, 1992). The average bulk density of the phenocryst-rich pumice is 0.98, and that of the phenocryst-poor pumice is 0.82. An upward decrease in the abundance of phenocryst-poor pumice lapilli has been noted in several sections through the tephra-fall deposits (David and others, this volume; Paladio-Melosantos and others, this volume; R.P. Hoblitt, unpub. data, 1993). This trend would normally indicate a decreasing supply of phenocryst-poor magma during the June 15, 1991, eruption. Although this may be the case, the possibility of sorting within the eruptive column has not been evaluated.

Table 3. Phenocryst pumice densities (D, in grams per cubic centimeter) and abundances of phenocryst-rich and phenocryst-poor dacitic lapilli in pyroclastic flow deposits of June 15, 1991.
[Component proportions were determined from <4-mm clast fractions by hand sorting with the aid of a binocular microscope. Densities determined by the method of Hoblitt, Wolfe, and others (this volume). n, number of lapilli]

Phenocrysts in both types of June 15 dacitic pumice are principally plagioclase and hornblende. Lesser amounts of cummingtonite, biotite, quartz, ilmenite, magnetite, apatite, and anhydrite are present (table 1). In contrast to the resorbed and broken crystals in the phenocryst-poor pumice, plagioclase in the phenocryst-rich pumice forms large (1-2 mm, rarely to 5 mm) euhedral crystals or euhedral crystals that are broken into only a few pieces. Most of these crystals show only subtle internal evidence of resorption, such as unconformities in zoning. A subordinate population of crystals shows a distinct dark resorption band that is rich in inclusions and typically located about four fifths of the way to the crystal margins.

The similarity of the average composition (about An40), and compositional range (An27-53) in plagioclase phenocrysts from the dacite to the average and range in the hybrid andesite (fig. 5, table 2) suggests that the dacitic magma was the sole source for plagioclase in the andesite. As in the andesite, the plagioclase displays complex oscillatory zonation (fig. 12C). Individual zones are recognized at 1-m scale in back-scattered electron images (fig. 12D) and variations as great as 20 mol% anorthite component (An29-50) and commonly 5 to 10 mol% An are seen between point analyses of adjacent bands or sets of bands (fig. 13). Large changes in An content sometimes, but not always, correlate with unconformities in zoning patterns. Edge compositions for two euhedral crystals in the phenocryst-rich pumice range from An34 to An37. Three analyzed microphenocrysts (<100 lc mum) in the phenocryst-rich pumice are normally zoned (An53-38).

Figure 13. Euhedral, oscillatory zoned plagioclase crystal (A), from phenocryst-rich dacitic pumice sample EW910615-1. Area of microprobe traverse B is shown in lower right quadrant. Numbers indicate percent An content at indicated spots. Point-beam analyses were used; no migration of sodium was detected.

Glass inclusions are common within the plagioclase phenocrysts; they are rhyolitic in composition and contain as much as 6.4 wt% H2O but less than 100 ppm sulfur (Westrich and Gerlach, 1992; Rutherford and Devine, this volume). The inclusions occur both as small (<10 lc mum) spheres concentrated parallel to zoning bands and as larger (to 100 lc mum) and typically more polygonal bodies that often crosscut zoning bands. Vapor bubbles are ubiquitous in the larger inclusions and are present in at least some of the smaller inclusions. A few inclusions contain a third (crystalline) phase that we have not identified. As pointed out by Westrich and Gerlach (1992), the large volume of vapor bubbles in plagioclase and other phenocrysts suggests that the dacitic magma was volatile saturated. Experimental work by Rutherford and Devine (this volume) confirms that the magma was either saturated or close to saturation with a water-rich (95 vol% H2O, 5 vol% SO2) fluid phase.

Hornblende forms euhedral or broken phenocrysts (typically 0.5-1 mm) in the phenocryst-rich dacitic pumice and forms smaller broken crystal fragments in the phenocryst-poor pumice. Euhedral crystals are rimmed by 10- to 50-lc mum-thick cummingtonite borders (fig. 14A), and some crystals show weak internal zonation, although not as pronounced as that in the higher titanium hornblendes of the andesite and basalt. The hornblende contains moderately abundant glass inclusions that range in size from <1 to 100 m and that contain high-silica rhyolite glass and vapor bubbles. Hornblende compositions are distinct from most of those in the hybrid andesite and basalt, with lower Mg#, Al, Ti, Na, and K (fig. 5, table 2). As in the andesite, the hornblende phenocrysts lack dehydration reaction rims. Lack of reaction rims on similar composition dacite from Mount St. Helens indicates ascent to the surface in a few days or less (Rutherford and Hill, 1993).

Figure 14. Photomicrographs of phenocryst-rich dacitic pumice sample EW910615-1 from a June 15 pyroclastic-flow deposit in the Sacobia River valley. Symbols: T, transmitted light; R, reflected light; X, cross polarized; P, plane polarized; xx mm, width of frame in millimeters; AHY, anhydrite; AP, apatite; CM, cummingtonite; BT, biotite; HB, hornblende; OL, olivine; BR, bronzite; MT, magnetite; SP, chrome spinel. A, Euhedral (now fractured) hornblende phenocryst with clear cummingtonite rim (T, P, 1.3 mm). B, Euhedral anhydrite(AHY)-apatite(AP) phenocryst composite. Note apatite inclusion within anhydrite (T, P, 0.68 mm). C, Fo80 olivine mantled by bronzite + magnetite and an outer rim of hornblende + biotite (T, P, 0.68 mm).

Biotite is a minor component in the dacite. It commonly occurs as cores to hornblende phenocrysts and in rare glomerocrysts with plagioclase and hornblende. Quartz forms sparse, highly fractured grains with large glass inclusions. Small (<0.3 mm) grains of both ilmenite and magnetite are common within the groundmass glass and included within the silicate phenocrysts. Apatite is ubiquitous as microlites but also forms larger (~0.1 mm) euhedral phenocrysts associated with anhydrite (fig. 14B). Anhydrite forms euhedral crystals that contain apatite inclusions and is similar in appearance to the anhydrite crystals found in the andesite scoria. Although apatite is a common inclusion in the silicate phenocrysts, anhydrite rarely occurs as an inclusion phase. Anhydrite typically forms subhedral crystals  surrounded by vesicular matrix glass. A single biotite-hornblende-plagioclase-anhydrite glomerocryst was observed in thin sections from 25 samples of June 15 dacite, and a small anhydrite inclusion was found at the contact between a hornblende phenocryst and its cummingtonite rim in a dacite pumice lapilli from a surge bed that was deposited on June 14.

Olivine and bronzite occur as sparse, highly resorbed anhedral grains. The olivine grains are bordered first by bronzite with fine-grained magnetite inclusions, and then by an outer zone composed of aggregates of hornblende and biotite microphenocrysts (fig. 14C). These olivine crystals are distinctly less magnesian than those in the scoria and basalt (fig. 5). A spinel inclusion in one of the olivines is exceedingly iron-rich (about 70 wt% total iron, table 2) and Cr- and Al-poor (11.3 wt% Cr2O3, 4.5 wt% Al2O3) and has likely undergone subsolidus alteration. We suspect that these olivine xenocrysts are the relicts of grains that were entrained during earlier (pre-1991) episodes of magma mixing.

GLASS COMPOSITIONS

Major element and volatile abundances for numerous glass inclusions and for matrix glass in the June 15 dacite are reported by Westrich and Gerlach (1992) and by Rutherford and Devine (this volume). In this report we focus on variations in the major element composition of matrix glass within our samples of (1) June 12 andesitic scoria, (2) a basalt inclusion from the June 7-12 andesitic dome, and (3) phenocryst-rich and phenocryst-poor dacitic pumice from the June 15 eruption. Average compositions and standard deviations for matrix glasses are listed in table 4 and compared graphically in figure 15. All glass analyses utilized a point electron beam; analyses were corrected for migration of alkalis, silica, and alumina by use of an iterative counting and correction procedure similar to that of Nielson and Sigurdsson (1981). Use of a point beam was required by the high vesicularity and thin bubble walls of pumice samples and by the presence of abundant microscopic crystal fragments in the phenocryst-poor pumice and microlites in the andesite. The analytical procedure is described in more detail in the appendix.

Table 4. Average compositions and standard deviations for matrix glass in 1991 Pinatubo samples.
[Analytical procedures given in the appendix. MnO less than 0.1 wt%. PR, phenocryst rich; PP, phenocryst poor; Avg., average; n, number of analyses averaged; SD, standard deviation; pl, plagioclase; hb, hornblende; cm, cummingtonite; ag, augite]

Figure 15. SiO2 variation diagrams for K2O (A) and FeO (B) for matrix glasses in 1991 eruptive products. Compositions of plagioclase and hornblende from the phenocryst-rich dacite, and bulk analyses of the 1991 dacite, andesite, and basalt (with all Fe recalculated as FeO) are shown for comparison in B. Point-glass analyses were corrected for Na, K, Al, and Si migration as described in the appendix. PR, phenocryst rich, PP, phenocryst poor.
Figure 15A
Figure 15B

Major element compositions of matrix glasses do not lie along simple mixing trends but are complicated by the effects of crystal growth and resorption following mixing. Matrix glass in the basalt is rhyolitic, with relatively high K2O (fig. 15A). The matrix melt in the basaltic inclusions must have been low originally in SiO2 but must have undergone extreme fractionation by growth of abundant hornblende, plagioclase, and magnetite microlites upon inclusion in the cooler dacite or andesite. The high K2O content of the glass is consistent with an overall relative enrichment of incompatible elements in the basalt. Dark bands within the andesitic scoria are composed of lower silica glass that is enriched in olivine and clinopyroxene phenocrysts. Consequently, we attribute much of the variation in the matrix glass of the andesite scoria to insufficient time for homogenization following mixing of rhyolitic melt from the dacitic magma and more mafic melt from the basalt.

Matrix glass in the phenocryst-poor dacite shows the widest variation in composition; that of the phenocryst-rich dacite is the most restricted and most evolved of all the 1991 samples. Because both of the dacitic types are identical in major element composition, we initially suggested that the wide range in glass compositions from the phenocryst-poor pumice was produced by disequilibrium melting of phenocrysts, a process that would shift the glass composition back toward the lower silica bulk-rock compositions (Pallister and others, 1992). Support for this hypothesis is evident in figure 15B; the less evolved glass compositions trend back toward the bulk composition of the June 15 dacite. Least-squares mass balance of the major element compositions indicates that addition of up to 22 wt% plagioclase +9 wt% hornblende +1 wt% oxides to the high-silica glass would explain the average chemical trend.

However, the compositional effects of dissolution of phenocrysts and contamination of the matrix glass by tiny phenocryst fragments has proven to be difficult to distinguish. Analysis points were located by use of high-magnification back-scattered electron images to avoid the abundant microscopic crystal fragments in the matrix glass; but even with this precaution, adjacent or underlying crystal fragments were sometimes excited and contributed to the analyses. Dispersion of the compositional trend in figure 15B toward plagioclase or hornblende is explained by excitation of adjacent or underlying fragments of these crystals, yet the overall trend back toward the bulk dacite composition would require systematic contamination by a consistent modal ratio of about 2 parts plagioclase to 1 part hornblende (22% to 9%, see above). Either we are seeing modal melting, or many areas that appear to be free of crystal fragments in our back-scattered electron images are actually composed of a well-mixed "slurry" of even smaller crystal particles and glass. Higher resolution microbeam and X-ray diffraction work is under way to evaluate this possibility. At present, although we acknowledge that crystal breakage would facilitate dissolution by increasing surface area and that some melting appears to have occurred, abundant broken crystal fragments in the phenocryst-poor pumice and the lack of higher Fe-Ti-oxide temperatures in the phenocryst-poor pumice indicate that most of the difference between the two pumice types is due to mechanical breakage of crystals (figs. 10, 11; Meeker and others, 1993).

THERMOBAROMETRY

FE-MG EXCHANGE EQUILIBRIA

Fe-Mg exchange between olivine and melt provides a means to estimate the Mg# and temperature of melt that coexisted with the olivines in the June 7-12 Pinatubo rocks. As previously noted, the forsterite content of olivines in the June 7-12 andesite and basalt samples overlap; they average Fo87 (table 2). There is little range in composition; the most magnesian core composition for olivine prisms is similar in the andesite and basalt samples, Fo88-89. Using the distribution coefficient for (FeO/MgO)olivine/ (FeO/MgO)melt of 0.3 (Roeder and Emslie, 1970; Ulmer, 1989), and assuming that Fe2O3/FeO=0.123, as discussed later, we calculate an average Mg# of 67 for equilibrium melts and a slightly higher value of 69 during initial crystallization of Fo88 olivine cores. The whole-rock Mg# of the least evolved basalt inclusion is 69 (see table 7), consistent with the olivine cores being early equilibrium phases and with the basalt being representative of the mafic mixing endmember for the hybrid andesites. Equilibrium fractional crystallization of the liquid line of descent for the basalt at atmospheric pressure, as determined by the CHAOS program of Neilsen (1990), predicts that Fo88 olivine would initially crystalize at 1,260°C and would remain as the only liquidus phase until 1,210°C, where An70 plagioclase would appear, to be joined by Mg# 85 clinopyroxene at 1,170°C. The predicted mineral compositions are in good agreement with the most magnesian olivine and clinopyroxene and the most calcic plagioclase analyzed in the basalt (table 2). However, the order of crystallization inferred from mineral textures (olivine and clinopyroxene preceding plagioclase) is different, probably as a result of crystallization at either high pressure (Bender and others, 1978; Grove and Kinzler, 1986) or under hydrous conditions (Spulber and Rutherford, 1983; Sisson and Grove, 1993). Higher pressure crystallization would also shift the distribution coefficient for FeO/MgO in olivine to higher values (Ulmer, 1989) but we cannot rigorously evaluate this effect because we do not know the primary oxidation state of iron in the basalt melt. Our selection of 0.123 for Fe2O3/FeO may have been fortuitous; selection of a higher ratio would require a higher distribution coefficient (and consequently higher pressure) to reproduce the Fo87-88 olivines. Presence of H2O would lower the liquidus temperature of the basalt and would also favor early crystallization of clinopyroxene (Sisson and Grove, 1993). Rutherford and others (1993) estimate that addition of 2.5 wt% H2O to the basaltic composition from Mount Pinatubo would result in an olivine-augite cotectic temperature of 1,200°C.

FE-TI OXIDES

Average compositions of coexisting ilmenite and magnetite from the 1991 Pinatubo samples, and temperatures based on the Andersen and Lindsley (1988) thermometer, are listed in table 5. Results are given for the dacite and andesite; the basalt samples contain only one oxide (magnetite). Iron distribution and mole fractions of ulvospinel and ilmenite were calculated by the method of Stormer (1983). Similar temperatures and oxygen fugacities were obtained by using the mineral projections and oxide equilibria routines in the QUILF program of Andersen and others (1993). Only homogeneous ilmenite-magnetite grain pairs that are touching or are within a few micrometers of one another in thin section were used for thermometry. Despite relatively low Mn abundances, Mg/Mn ratios for the Pinatubo oxides overlap the equilibrium envelope of Bacon and Hirschmann (1988), as seen in figure 16 (inset).

Figure 16. Oxygen fugacity - temperature diagram showing results from Fe-Ti-oxide thermometry on 1991 Pinatubo samples. Buffer curves from Rutherford (1993). Inset shows Mg/Mn distribution in Pinatubo oxides compared to equilibrium field of Bacon and Hirschmann (1988); crosses show typical analytical errors in Mg/Mn for June 15 dacite and June 12 scoria. Mag, magnetite; Ilm, ilmenite; HM, hematite-magnetite; MNO, MnO-Mn3O4; NNO, Ni-NiO; QFM, quartz-fayalite-magnetite; rxn, reaction.

Table 5. Microprobe compositions of Fe-Ti oxides, iron recalculation and formulas by the method of Stormer (1983), and calculated temperatures and oxygen fugacities for mineral pairs according to the method of Andersen and Lindsley (1988).
[Avg., average; n, number of analyses; X ulv, mole fraction ulvospinel; X ilm, mole fraction ilmenite]

The average temperature and oxygen fugacity (in log units) determined for four magnetite-ilmenite pairs in the June 15 phenocryst-rich dacitic pumice is 816+-22°C and 11.08+-0.12 respectively, indistinguishable from results for the phenocryst-poor pumice. The temperatures have been reduced by 30°C (yielding an average of 786+-22°C) prior to plotting in figure 16 to correct for overestimation of Andersen-Lindsley-Stormer temperatures at high oxygen fugacity (~3 log units above Ni-NiO) (Geschwindt and Rutherford, 1993; Rutherford and Devine, this volume).

Apparent temperatures and oxygen fugacities were also calculated for the magnetite-rimmed ilmenites of the June 12 scoria (sample 7-1-91-1A). Only grains without oxidation lamellae were used for thermometry; grains with lamellae from the more slowly cooled dome samples were not used. As described previously, the magnetite rims are overgrowths of new crystals that nucleated and grew on ilmenite grains during magma mixing. The rimmed crystals indicate changing conditions in the magma, a disequilibrium situation. The zoning of core ilmenite to higher titanium within about 10 lc mu m of the magnetite rim (fig. 6L) suggests diffusive exchange with the growing rim. Temperatures and oxygen fugacities for the pairs core versus rim and high-titanium zone versus rim yield temperatures of 961°C at log fO2=-9.1 and 930°C at log fO2=-9.7, respectively. These values are intermediate between those determined for the dacite and those inferred from olivine-melt equilibria (~1,200°C) at the reduced conditions implied by the presence of sulfides in the basalt. We can test whether the Fe-Ti-oxide temperatures and oxygen fugacities are reasonable by calculating the oxidation state of iron for the bulk sample according to the method of Sack and others (1980) and comparing the result to that obtained from direct analysis. Of the 6.1 wt% Fe2O3 total in a bulk analysis of sample 7-1-91-1A, we calculate 2.3 wt% as Fe2O3 and 3.3 wt% as FeO at T=950°C and log fO2=-9.5. This result is similar to the distribution of iron in the bulk sample that was determined analytically (see table 7), circumstantial evidence that the Fe-Ti-oxide temperatures and oxygen fugacities are representative of the mixed magma just before eruption.

CUMMINGTONITE STABILITY

The presence of cummingtonite rims on hornblende, in combination with the composition of plagioclase rims and microlites in the June 15 dacitic pumice samples, provides a means to fix the temperature and PH2O of the dacite during its final stage of crystallization prior to eruption. As is explained elsewhere (Geschwindt and Rutherford, 1993; Rutherford, 1993; Rutherford and Devine, this volume), the field of cummingtonite+hornblende (coexisting with plagioclase, magnetite, ilmenite, and melt) is limited to temperatures of <800°C and PH2O of about 100-400 MPa (1-4 kbar) in these dacitic magmas. The Fe-Ti-oxide temperature of about 780°C constrains the magma to the higher temperature range of cummingtonite stability and, with coexisting quartz and An34-40 plagioclase rims, to a more limited PH2O range of about 200-320 MPa (see fig. 2 of Rutherford, 1993, and fig. 4 of Rutherford and Devine, this volume).

BULK-ROCK CHEMISTRY

Sample localities and descriptions for analyzed samples of 1991 eruptive products are given in table 6, and major and trace element analyses are listed in table 7. Most of the analyzed samples were derived from single pumice or scoria blocks or from single basalt inclusions (as noted in table 6). Bulk-rock compositions are typical of calc-alkalic arc rocks, with relatively low TiO2 and low abundances of high-field-strength (HFS) elements (fig. 17). Pinatubo rocks are enriched in light rare earth elements (LREE) compared to those from nearby volcanoes in the Bataan arc. In this respect, the 1991 Pinatubo magmas are more like basalts from the back-arc Mount Arayat volcano to the east than to the andesites and dacites of the adjacent frontal arc Natib and Mariveles stratovolcanoes to the south (compare REE data in fig. 17 and table 7 with data of Defant and others, 1991).

Figure 17. Abundance of trace elements in 1991 Pinatubo samples normalized to midocean ridge basalt (MORB) and compared to average dacite of Mount St. Helens (MSH) of Smith and Leeman (1987). PR, phenocryst-rich; PP, phenocryst poor.

Table 6. Localities and notes for analytical samples.
[Abbreviations: p.f., pyroclastic flow; qz., quartz; xenos., xenocrysts; NF, north fork]

The dacites of Mount Pinatubo are unusually depleted with respect to incompatible element abundances. This is especially evident from the fact that the dacites have lower rare earth element abundances than the basalt inclusions. The depletion of incompatible elements in the dacites relative to the basalts indicates that the basalts cannot be parental to the dacites by major phase crystal fractionation. In addition, the parallelism of the elemental distribution patterns (fig. 17) rules out control by fractionation of trace element enriched minor phases, such as zircon, apatite, or other non-uniformly enriched REE-bearing minerals. In this respect, the dacites of Mount Pinatubo are similar to dacites from Mount St. Helens, where an origin by large-degree partial melting of mafic crustal rocks is favored by Smith and Leeman (1987).

The compositions of the phenocryst-rich and phenocryst-poor dacitic pumice types are identical at the +-1-lc sigma level for all major and trace elements (table 7). This finding is consistent with a closed-system process of crystal size reduction, such as proposed above.

Table 7. Major and trace element analyses of 1991 Pinatubo rocks.
[Oxides analyzed by X-ray fluorescence, in wt% normalized volatile free. Element symbols followed by coefficients of variation (cv, in %) were determined by instrumental neutron activation analysis, others by energy-dispersive X-ray fluorescence. LOI=loss on ignition at 925°C. Sulfur was determined by J. Curry of the U.S. Geological Survey on separate sample splits using a combustion-infrared technique. Elemental data in ppm, except S (wt%) and Au (ppb). SD, standard deviation]
Page 1 | Page 2

Three data clusters that correspond to basalt, andesite, and dacite are apparent on a silica variation diagram (fig. 18). The single data points at 62.4 wt% SiO2 correspond to a rare intermediate-composition lapillus from the tephra fall of June 13 (sample P22692-2A2). The linear trends on this diagram suggest a magma mixing origin for the andesitic bulk compositions. We tested the magma mixing hypothesis by using least-squares modeling (Wright and Doherty, 1970) of the average major element compositions of the basalt, andesite, and dacite. The model solution was then applied to the trace element data. The result: the average andesite major element and trace element composition can be produced from 36.5 wt% basalt + 63.5 wt% dacite, with a squared sum of residuals for the major elements of <0.05 (table 8). All but five of the 27 analyzed trace element abundances are within one standard deviation of the model solution, and all are within two standard deviations. The largest deviation is for K2O, which has a residual of 0.9 wt%. The model solution for the rare earth elements is graphically illustrated in figure 19.

Figure 18. SiO2 variation diagram for 1991 Pinatubo major element analyses. Samples fall into basalt (50-52 wt% SiO2), andesite (59-60 wt% SiO2), and dacite (64-65 wt% SiO2) on an anhydrous basis and display linear trends with respect to SiO2 abundance. The single sample at 62.4 wt% SiO2 is from a nonbanded, light-gray lapillus from the June 13, 1991, fall deposit.

Figure 19. Chondrite-normalized rare earth element diagram showing average patterns for 1991 Pinatubo basalt, andesite, and dacite, and pattern calculated for 36.5 wt% average basalt and 63.5 wt% average dacite (proportions derived from major-element least-squares model in table 8).

Table 8. Least-squares mixing model for hybrid andesite.
[Major element least-squares model proportions were used to calculate trace element abundances for model solution. Percentages in parentheses indicate analytical uncertainty based on average counting statistics. Residuals in bold print are outside 1lc sigma of andesite mean. SD, standard deviation; n, number of samples]

DISCUSSION AND SUMMARY

Samples from the 1991 eruptions of Mount Pinatubo provide unequivocal petrographic and chemical evidence for magma mixing involving basalt that was initially at ~1200°C and dacite at ~780°C. Eruption of the hybrid andesite magma first (on June 7-14) and preservation of disequilibrium textures indicate that mixing of basalt and dacite took place shortly before, and probably triggered, the climactic dacitic eruption on June 15, 1991. Recent experimental calibration of reequilibration rates of minerals in the mixed andesite of Mount Pinatubo provides an estimate of the time interval between mixing and eruption (Rutherford and others, 1993). The preservation of thin cummingtonite rims on hornblende, inhomogeneous glass, and magnetite rims on ilmenite suggest that mixing occurred within four days of eruption, and the lack of dehydration reaction rims on hornblende phenocrysts indicates that ascent to the surface also took place in less than four days. Consistent with these results, deep (>30 km) long-period earthquakes were recorded during two periods (May 26-28 and May 31-June 6) at Pinatubo and are related to ascent of basaltic magma to the magma reservoir within a time span of days to hours before eruption (White, this volume).

THE MAGMA RESERVOIR

Our best evidence for the site of magma mixing comes from evaluation of phase equilibria and seismic data. A high Mg# and lack of plagioclase as a liquidus phase in the basalt indicate that it originated at great depth and possibly under hydrous conditions and that it was little modified by shallow fractionation. Studies of melt and vapor inclusions in phenocrysts and phase equilibria (Westrich and Gerlach, 1992; Rutherford and Devine, this volume) indicate that the June 15 dacitic magma was saturated with a water-rich fluid phase prior to eruption. An apparent equilibration pressure of 220+-50 MPa for the dacite is calculated by Rutherford and Devine (this volume) from the Al-in-hornblende geobarometer of Johnson and Rutherford (1989). A similar result (225+-50 MPa) may be calculated from the hornblende data in table 2; however, several members of the required equilibrium assemblage are either lacking (sphene, sanidine) or resorbed (quartz). An independent pressure estimate is provided by the coexistence of cummingtonite and An34-40 plagioclase as late magmatic phases, which indicates that the June 15 dacite last equilibrated at pressures of 200-320 MPa. Comparison of natural glass compositions to those from experimental samples of June 15 dacite favors equilibration under water-saturated conditions at 780°C and further restricts the equilibration pressure to about 200 MPa (Rutherford and Devine, this volume). On the basis of geothermal exploration and drilling (Delfin, 1983; Delfin and others, this volume), we infer that the volcano is underlain by about 2 km thickness of vesicular dacite (with density of ~2.0 g cm3) and 2 km of dacitic to andesitic lavas and intrusions (~2.6 g cm3), which overlie ophiolitic rocks (~2.8 g cm). Given this rock column and assuming that PH2O=Ptotal, 200 MPa is equivalent to a depth of about 8 km beneath the pre-1991 summit (~6 km depth, relative to sea level).

Hypocenter locations for earthquakes recorded from the time the seismic net was established on May 8, 1991, until June 15 were mainly at shallow depths of <4 km. After the June 15 eruption, numerous earthquakes took place at greater depth, eventually outlining a broad envelope of seismicity that extends to depths of 20 km (fig. 20). It is tempting to take the dense dome-shaped distribution of hypocenters seen in the east-west projection of figure 20 as the outline of a large (200 km3) magma chamber at depth. This would be analogous to the approach used by Scandone and Malone (1985) to infer the shape of the magma reservoir beneath Mount St. Helens. However, in detail the picture appears to be more complex. The distribution of deep hypocenters is not symmetrical about a vertical axis through the volcano but consists of two dense dikelike or pipelike bodies (prominent in fig. 20) and one less-dense body to the north, each of which extends from about 5 to 20 km in depth. A sparse shroud of hypocenters appears to connect the pipe-like distributions, especially around the northern side of the hypocenter envelope, and outlines a virtually earthquake-free interior zone. Tomographic inversion of the seismic data shows a cylindrical low-velocity zone that is inferred to represent magma, at depths of >6 km (relative to sea level) (Mori, Eberhart-Phillips, and Harlow, this volume). This body is 4 to 5 km in diameter and has a volume of 40 to 90 km3 for the depth interval 6 to 11 km.

Figure 20. A, Speculative model for the Pinatubo pluton, as based on depth constraints from phase equilibria and projection of earthquake hypocenters for the period May 8-August 19, 1991, onto an east-west plane (B) (compare, Mori, Eberhart-Phillips, and Harlow, this volume).

We suggest that the low-velocity zone represents the source for the 1991 eruptions. The region within the envelope of hypocenters is inferred to be an arc pluton that is melt-rich only at depths of more than 8 km beneath the level of the pre-1991 summit. Much of the magma in the reservoir was "uneruptable" because of high viscosity caused by high-silica matrix melt and crystal contents that exceeded 50 percent (Marsh, 1981). Had not basaltic magma been added to the system, the dacite probably would not have erupted in 1991.

BASALT TRIGGERING OF THE JUNE 15 ERUPTION

Mixing of basalt and dacite could trigger an explosive eruption by increasing magmatic pressure as a result of addition of mass, fluid, or heat, the latter leading to convective overturn and increased volatile exsolution (Sparks and others, 1977; Williams and McBirney, 1979, p. 74-86; Cas and Wright, 1988, p. 40-41). For Pinatubo we must first explain why these two magmas, presumably with vastly different physical properties, would mix at all; then we must explain how the andesitic magma, which would normally be more dense, could reach the surface before the dacitic magma could. The problem of mixing dissimilar magmas is well known (for example, Sparks and others, 1984; Huppert and others, 1984; Sparks and Marshall, 1986). These studies point out that thorough mixing requires turbulent flow, which is favored by similar viscosities and high flow rates. Consequently, mixing models typically call on cooling of the mafic magma and heating of the felsic magma such that the densities and viscosities of the two approach one another. However, the ability to mix also depends on the relative volumes of the two components. If the volume and thermal mass of the mafic magma component is small, it tends to quench when intruded into cooler felsic magma, but when the volume of the mafic magma is relatively large (about 50%; Sparks and Marshall, 1986), mixing is possible.

We suggest the following magma mixing model for the 1991 eruptions. Basalt began to leak into a crystal-rich dacite reservoir at about 8 km in depth beneath Pinatubo, possibly starting on or before April 2, 1991, the date of the first phreatic explosions. Initial batches of basaltic magma were small and quenched in the dacite to form fluid-saturated diktytaxitic pillows, which were erupted later as inclusions in the June 7-12 dome lava. A small amount of dacite was entrained in the basalt, so hornblende and quartz xenocrysts were produced. This dacite was derived either from wallrocks at greater depth or from the dacitic magma. With continued influx, basalt accumulated at the base of the dacitic reservoir and cooled more slowly. As it cooled the basalt liberated heat to the overlying dacite and crystallized augite, magnetite, and high-titanium hornblende. By virtue of cooling and crystallization, the viscosity of the basalt increased to the point that mixing became feasible. Although much of the heat evolved to the dacite was probably transported away from the interface by convection, limited heating and disequilibrium melting of phenocrysts may also have lowered the bulk viscosity of the dacite.

We suspect that at this point either extensive fluid exsolution from the residual liquid in the basalt lowered its bulk density below that of the dacite (compare, Eichelberger, 1980; Huppert and others, 1982) and caused it to rise in a turbulent plume of basalt foam and mix, or a particularly violent new influx of basalt stirred the two components together. The latter interpretation would be consistent with ascent of a voluminous batch of basaltic magma accompanying the higher rate episode of deep, long-period earthquakes recorded by White (this volume) on May 31-June 6. In either case, mixing was exclusively a mechanical process in which phenocrysts and matrix melts were intimately mingled together to produce a commingled magma in the sense of Sparks and Marshall (1986). The lack of reequilibration of glass and minerals in the andesite indicates that ascent to the surface began almost immediately upon mixing. The fluid-saturated dacitic component was heated to ~940°C without chemical equilibration. At 220 MPa and 940°C, water has a density of 0.38 g/cm3 (Burnham and others, 1969). Consequently, only a relatively small increase in the volume percent of fluid bubbles would be required to lower the bulk density of the mixture below that of the dacite.

As the buoyant plume of mixed andesite made its way upward, it would have encountered semirigid (possibly thixotropic?) dacite crystal mush in the upper part of the magma reservoir overlain by solidified dacite porphyry, such that fracture transport would be possible. The andesitic magma continued to propagate upward and established a relatively narrow conduit to the surface, whereas, under the same pressure gradient, the more viscous dacite could not rise en masse until a larger diameter conduit was established. Some dacitic magma was entrained into the rising column of andesite, probably by viscous coupling at the base of the conduit. It was first erupted as a minor fraction of pumice in the moderate-sized June 12 vertical eruption. The conduit was enlarged as more magma was transported to the surface. A larger diameter conduit and slightly reduced load pressure (due to vesiculation at shallow levels) allowed progressively greater amounts of the viscous dacitic magma to enter and rise through the conduit, and contribute increasing volumes to the June 13-14 eruptions, and, finally, to dominate the June 15 eruption.

ORIGIN OF CRYSTAL FRAGMENTS IN THE PHENOCRYST-POOR DACITIC PUMICE

The reduction in grain size seen in the phenocryst-poor pumice of June 15 took place during ascent through the conduit system, below the fragmentation level, because crystal fragments are present in the matrices of intact pumice blocks. The apparent concentration of phenocryst-poor pumice in the lower part of the tephra-fall deposits of June 15, but the lack of pumice with the same type of texture in the blast deposits of June 14-15 (Hoblitt, Wolfe, and others, this volume; David and others, this volume), suggests that the process of crystal size reduction took place mainly during the early climactic phase of the June 15 eruption. Our electron imagery and Fe-Ti-oxide thermometry indicate that most of the reduction in grain size was brought about by mechanical fragmentation of phenocrysts rather than melting, but details of the process are unknown. Working models include (1) shear-fragmentation of phenocrysts in magma adjacent to conduit walls, (2) strain imposed by viscous coupling of matrix melt and crystals during boiling of the fluid phase (vesiculation), (3) explosion of gas-saturated inclusions in phenocrysts (Gerlach and others, this volume), and (4) shock fragmentation, possibly related to the choked-flow mechanism for long-period volcanic earthquakes of Chouet and others (1994), although probably involving a magmatic fluid phase rather than ground water. Occurrence of both textural types of dacitic pumice in deposits from previous eruptive periods suggests a common process that has occurred repeatedly during the 30,000-year history of the volcano.

PROSPECTS FOR FUTURE ERUPTIONS

The magmatic model developed above has obvious implications for the future. The hot basaltic magma added to the reservoir in 1991 could rejuvenate the magma system and lead to additional eruptions. Volcanic seismicity continued in 1992 (Ramos and others, this volume) and a new lava dome began to grow in the caldera during the summer of 1992. The dome is hybrid andesite, very similar in phenocryst assemblage, texture, and composition to the June 7-12, 1991, lava dome (Daag, Dolan, and others, this volume). Is the eruption of this second batch of mixed magma likely to trigger another large dacitic eruption? Clearly, it has not had that effect in the interval between the appearance of the 1992 dome and proofing of this paper (August 1995). We suspect that the main dacitic reservoir was partially depleted in eruptible magma by the June 15, 1991, eruption and that some unknown amount of time must pass before the volatile content can build up to an eruptible level again.

Will Pinatubo volcano erupt again? The record of past activity indicates that the answer is yes. Pinatubo has been the site of periodic eruptions of dacite and hybrid andesite over a time span of more than 30,000 years. Hydrous cummingtonite-bearing magma, compositionally and texturally similar to the 1991 dacite, has been erupted repeatedly (Newhall and others, this volume; Pallister and others, 1993). Pumice from previous episodes ranges over several wt% in SiO2, and banded pumice and mafic scoria with olivine and augite xenocrysts are present. These relations indicate that repeated mixing of dacite and basalt has taken place at the volcano. Pinatubo has produced large-volume dacitic flowage deposits and domes of hybrid andesite in the past. There is no reason to think that future eruptions will be appreciably different.

SULFUR-RICH EXPLOSIVE ERUPTIONS--THE ROLE OF BASALT AND MAGMA MIXING

The June 15, 1991, eruption, venting about 20 Mt of SO2 into the stratosphere (Bluth and others, 1992), resulted in an increase in optical depth of the atmosphere and decrease in the global mean temperature sufficient to suppress the global warming trend through the early and mid-1990's (Hansen and others, 1992). In terms of climatic effect, it is particularly important to understand the processes that lead to high-sulfur explosive eruptions and to determine the frequency of such eruptions in the past. Sources for the sulfur emissions during the June 15 eruption of Pinatubo have been attributed either to degassing of an exsolved fluid phase in the dacite (Westrich and Gerlach, 1992) or to breakdown of anhydrite during the eruption (Rutherford and Devine, 1991; this volume). But where was the sulfur before the eruption and what process led to the enrichment of the magma in sulfur in the first place?

Either the sulfur source is exotic, as in the evaporite bed hypothesis at El Chichón (Duffield and others, 1984), or it is a primary phase of the magmatic (and related hydrothermal) system. Lacking evidence for sedimentary anhydrite at Pinatubo, we tend to favor a magmatic source. The anhydrite in the pumice of Mount Pinatubo is considered as either a primary magmatic phase (Bernard and others, 1991; this volume), or it represents xenocrysts entrained from the hydrothermal system (A.R. Reyes, written commun., 1991), or both (McKibben and others, 1992; this volume). As previously noted, anhydrite occurs primarily as subhedral to euhedral crystals in the June 12 scoria and June 15 pumice. It typically includes apatite microlites and, as at El Chichón (Luhr and others, 1984), it is found in growth contact with apatite phenocrysts. Anhydrite is only rarely found in growth contact with silicate phenocrysts. These relations are consistent with growth mainly from a separate fluid phase in the magma. The euhedral character of most anhydrite crystals and the association with euhedral apatite favor magmatic crystallization over entrainment of xenocrysts, as does the high sulfur, chlorine, and fluorine content of apatite microphenocrysts (Imai and others, 1993; Pallister and others, 1993). However, the association with apatite does not rule out origin from the hydrothermal system for some crystals. Anhydrite and apatite are common in a broad region where temperatures exceed 220-300°C at depths of >1.8 km below the pre-1991 summit (Delfin, 1983; Delfin and others, this volume). This broad anhydrite-bearing region attests to the long-term evolution of sulfur-rich fluids from the Pinatubo magmatic system. Sulfur isotopic data for anhydrite crystals from the June 12, 1991, scoria are bimodal, with lc delta34S modes at +6 and +10, possibly reflecting both magmatic (+6) and hydrothermal (+10) sources (McKibben and others, 1992; this volume). Anhydrite from the June 15 dacite is unimodal at lc delta34S of +7, consistent with a dominantly magmatic source (McKibben and Eldridge, 1993).

We previously suggested that basaltic magma may have contributed sulfur to the dacite (Pallister and others, 1992). Matthews and others (1992) also favor this model, noting that mixing of basalt with cool dacite at shallow levels would lead to separation of a sulfur-rich vapor phase. Westrich and Gerlach (1992) discounted this possibility for the June 15 eruption because of predicted slow rates of bubble rise (from Stokes Law calculations) and the low oxygen fugacity of most basaltic magmas. We agree that these are problems for basaltic magma as the immediate source of sulfur during the June 15 eruption. However, we are more concerned with the long-term source for the sulfur, and we note that Gerlach and others (this volume) now also favor basalt as the long-term source. Because the solubility of sulfur is enhanced by high iron and calcium content of the melt (Matthews and others, 1992), basalt is a more effective transport agent for sulfur than dacite is. Assuming average abundances of 1,000 ppm sulfur in basaltic magma (Devine and others, 1984; Carroll and Rutherford, 1987), we calculate that it would require about 4 km3 of basalt to account for the 1991 emissions. Experimental data on sulfur solubility indicate that a hydrous basalt that is oxidized above the nickel-nickel oxide buffer curve (NNO, fig. 16) could have carried in excess of 1 wt% SO3 (Luhr, 1990). Consequently, if the basalt were hydrated and oxidized at depth, the required volume to balance the 1991 emissions would be lowered to less than 1 km3. In either case, these are relatively small volumes compared to the potential size and consequent longevity of the inferred magma reservoir, especially considering that many batches of basalt may have been added to the magmatic system over the >30,000-year life of the volcano.

We suggest that a large dacitic magma reservoir has been present beneath Pinatubo for an extended period of time and that basalt has repeatedly mixed with dacite in the reservoir (Newhall and others, this volume). Periodic addition and mixing of basaltic magma probably triggered many of the previous eruptions of crystal-rich dacite and hybrid andesite and also maintained a relatively elevated abundance of sulfur in resident dacitic magma within the shallow reservoir. Copper and zinc abundances are elevated in the 1991 dacite (table 7), possibly as a result of fluid transport from previous batches of basaltic magmas. Rare grains of olivine, pyroxene, and sulfide in the 1991 dacite (for example, Bernard and others, this volume; Hattori, this volume) may also record previous mixing events.

We note that the last climatically important eruption, that of El Chichón in 1980, involved sulfur-rich trachyandesite magma that was also cool (810°C +-40°; Rye and others, 1984), crystal rich (58 wt% crystals), and gas saturated and that a hydrous and oxidized basaltic parent magma is suggested to transport sulfur from depth (Luhr, 1990). Perhaps one of the characteristic features of volcanoes that give rise to sulfur-rich explosive eruptions is the presence of a long-lived magma reservoir with a relatively cool, oxidized, and crystal-rich upper zone, which acts as a trap for volatiles that are supplied by mafic magmas from greater depth.

ACKNOWLEDGMENTS

We thank Agnes Reyes of the Philippine National Oil Company for providing splits of her samples of the June 12, 1991, eruption and for her collaboration in our initial report on the 1991 eruptions. We also thank Ray Punongbayan and the staff of the Philippine Institute of Volcanology and Seismology (PHIVOLCS) for support while in the Philippines; we especially enjoyed working with Mylene Martinez, Timoteo Nillos, and Ronnie Torres of PHIVOLCS. The first author expresses his gratitude to the staff and students of the geology department of the University of the Philippines for their hospitality and interest during his visit to Manila. We thank the Philippine Air Force and the U.S. Marines for logistical and helicopter support. Finally, we thank Esperanza Soratos for her able assistance at the Pinatubo Volcano Observatory and for sharing a lively sense of humor with us during the 1992 field season. Discussions with Ken Hon, Chris Newhall, Malcolm Rutherford, and Terry Gerlach were of considerable help in interpreting the data presented here. We also appreciate the constructive and insightful manuscript reviews by Wes Hildreth, Jim Luhr, Chris Newhall, and Malcolm Rutherford.

APPENDIX

ANALYTICAL PROCEDURES

Silicate mineral and glass analyses in tables 2 and 4 were obtained on an ARL-SEMQ microprobe equipped with six scannable wavelength-dispersive crystal spectrometers. An accelerating potential of 15 kV and beam current of 10 nA were used on all samples. Natural and synthetic silicate standards were used, and off-peak background corrections were applied to standards and unknowns. Microprobe automation employed the OPUS program of Meeker and Quick (1991), and data reduction was done online by using the CITZAF routine of Armstrong (1988). On the basis of replicate analyses of secondary standards over the 4-year period 1989-92, since the new automation and data reduction software has been implemented, analytical reproducibility is estimated to be +-1-2 percent of the reported amounts for major elements and equal to or less than the standard deviation that is based on counting statistics for minor elements (typically +-10 percent of reported values in the range 0.2 to 1.0 wt%).

Oxide analyses in table 5 were obtained on a JEOL-8900 microprobe equipped with five scannable wavelength-dispersive crystal spectrometers. An accelerating potential of 15 kV and beam current of 25 nA were used. Natural and synthetic silicate and oxide standards were used, and off-peak background corrections were applied to standards and unknowns. Data reduction was by the ZAF method.

Microprobe analysis of glass filaments in the vesicular samples from Pinatubo required the use of a point electron beam. Beam damage and heating resulted in significant migration of the alkalis and silica (fig. A1). Consequently, corrections for migration of Na, K, and Si had to be applied. We used a correction procedure similar to that described by Neilsen and Sigurdsson (1981) in which count data were collected in 1-s increments and then exponential decay functions were fit to the resulting curves. This procedure was done online, and intercept values for count rate at t=0 for Na, K, and Si were used in the data reduction routine (CITZAF). Curve fits were excellent, typically yielding r2> 0.9 for Na and K such that the analytical uncertainty was affected as much by the loss in total counts (typically half the starting amount in a 20-s count window) as by curve fitting. We estimate analytical uncertainties of +-1 percent of the reported SiO2, and +-5 percent of the reported Na2O and K2O for any one analysis, based on replicate analyses of high-silica rhyolite glass standards.

Figure A1. Exponential decay curves in X-ray counts per second for Na, K, Si, and Al for high-silica rhyolite matrix glass in phenocryst-rich dacite sample EW910615-1. Accelerating potential 15 kV, beam current 10 nA. Count data corrected for background counts (-bkg).

Major element analyses reported in table 7 were determined by X-ray fluorescence methods as described by Taggart and others (1987). Trace elements were determined principally by instrumental neutron activation analysis as described by Baedecker and McKown (1987). On the basis of replicate analyses of standards (Taggart and others, 1987), analytical reproducibility of major element abundances is estimated to be better than +-0.4 percent of the reported values for SiO2 and Al2O, and for other elements better than +-2 percent of the values in the range 1-10 wt%, and better than +-6% for abundances less than 1 wt%. Coefficients of variation (100lc sigma/x overbar in percent) based on counting statistics are given for instrumental neutron activation data in table 7.

REFERENCES CITED

Andersen, D.J., and Lindsley, D.H., 1988, Internally consistent solution models for Fe-Mg-Mn-Ti oxides: Fe-Ti oxides: American Mineralogist, v. 73, p. 714-726.

Andersen, D.J., Lindsley, D.H., and Davidson, P.M., 1993, QUILF: A PASCAL program to assess equilibria among the Fe-Mg-Mn-Ti oxides, pyroxenes, olivine, and quartz: Computers and Geosciences, v. 19, p. 1333-1350.

Armstrong, J.T., 1988, Analysis of silicate and oxide minerals: Comparison of Monte Carlo, ZAF, and theta(lc rhoz) procedures: Microbeam Analysis, v. 1988, p. 239-246.

Bacon, C.R., 1986, Magmatic inclusions in silicic and intermediate volcanic rocks: Journal of Geophysical Research, v. 91, p. 6091-6112.

Bacon, C.R., and Hirschmann, M.M., 1988, Mg/Mn partitioning as a test for equilibrium between coexisting Fe-Ti oxides: American Mineralogist, v. 73, p. 57-61.

Baedecker, P.A., and McKown, D.M., 1987, Instrumental neutron activation analysis of geochemical samples, in Baedecker, P.A., ed., Methods for geochemical analysis: U.S. Geological Survey Bulletin 1770, p. H1-H14.

Bender, J.F., Hodges, F.N., and Bence, A.E., 1978, Petrogenesis of basalts from the project FAMOUS area: Experimental study from 0 to 15 kbars: Earth and Planetary Science Letters, v. 41, p. 277-302.

Bernard, A., Demaiffe, D., Mattielli, N., and Punongbayan, R.S., 1991, Anhydrite-bearing pumices from Mount Pinatubo: Further evidence for the existence of sulfur-rich silicic magmas: Nature, v. 354, p. 139-140.

Bernard, A., Knittel, U., Weber, B., Weis, D., Albrecht, A., Hattori, K., Klein, J., and Oles, D., this volume, Petrology and geochemistry of the 1991 eruption products of Mount Pinatubo.

Bluth, G.J.S., Doiron, S.D., Schnetzler, C.C., Krueger, A.J., and Walter, L.S., 1992, Global tracking of the SO2 clouds from the June, 1991 Mount Pinatubo eruptions: Geophysical Research Letters, v. 19, p. 151-154.

Burnham, C.W., Holloway, J.R., and Davis, N.F., 1969, Thermodynamic properties of water to 1000°C and 10,000 bars: Geological Society of America Special Paper 132, 96 p.

Carroll, M.R., and Rutherford, M.J., 1987, The stability of igneous anhydrite: Experimental results and implications for sulfur behavior in the 1982 El Chichón trachyandesite and other evolved magmas: Journal of Petrology, v. 28, p. 781-801.

Cas, R.A.F., and Wright, J.V., 1988, Volcanic successions, modern and ancient: London, Unwin Hyman, 528 p.

Chouet, B.A., Page, R.A., Stephens, C.D., and Lahr, J.C., 1994, Precursory swarms of long-period events at Redoubt Volcano (1989-1990), Alaska: Their origin and use as a forecasting tool: Journal of Volcanology and Geothermal Research, v. 62, p. 95-135.

Daag, A.S., Dolan, M.T., Laguerta, E.P., Meeker, G.P., Newhall, C.G., Pallister, J.S., and Solidum, R., this volume, Growth of a postclimactic lava dome at Mount Pinatubo, July-October 1992.

David, C.P.C., Dulce, R.G., Nolasco-Javier, D.D., Zamoras, L.R., Jumawan, F.T., and Newhall, C.G., this volume, Changing proportions of two pumice types from the June 15, 1991, eruption of Mount Pinatubo.

Defant, M.J., De Boer, J.Z., and Oles, D., 1988, The western central Luzon arc, the Philippines: Two arcs divided by rifting?: Tectonophysics, v. 145, p. 305-317.

Defant, M.J., Maury, R.C., Ripley, E.M., Feigensen, M.D., and Jaques, D., 1991, An example of island-arc petrogenesis: Geochemistry and petrology of the southern Luzon arc, Philippines: Journal of Petrology, v. 32, p. 455-500.

Delfin, F.G., Jr., 1983, Geology of the Mt. Pinatubo geothermal prospect: unpublished Philippine National Oil Company report, 35 p.

Delfin, F.G., Jr., Villarosa, H.G., Layugan, D.B., Clemente, V.C., Candelaria, M.R., Ruaya, J.R., this volume, Geothermal exploration of the pre-1991 Mount Pinatubo hydrothermal system.

Devine, J.D., Sigurdsson, H., and Davis, A.N., 1984, Estimates of sulfur and chlorine yield to the atmosphere from volcanic eruptions and potential climatic effects: Journal of Geophysical Research, v. 89, p. 6309-6325.

Dick, H.J.B., and Bullen, T., 1984, Chromian spinel as a petrologic indicator in abyssal and alpine-type peridotites and spatially associated lavas: Contributions to Mineralogy and Petrology, v. 86, p. 54-76.

Donaldson, C.H., 1985, The rates of dissolution of olivine, plagioclase, and quartz in a basalt melt: Mineralogical Magazine, v. 49, p. 683-693.

Duffield, W.A., Tilling, R.I., and Canul, R., 1984, Geology of El Chichón volcano, Chiapas, Mexico, Journal of Volcanology and Geothermal Research, v. 20, p. 117-132.

Eichelberger, J.C., 1980, Vesiculation of mafic magma during replenishment of silicic magma reservoirs: Nature, v. 288, p. 446-450.

Evans, C.A., 1987, Oceanic magmas with alkalic characteristics; evidence from basal cumulate rocks in the Zambales ophiolite, Luzon, Philippine Islands: Geological Society of America Special Paper 215, p. 139-150.

Evans, C.A., Casteneda, G., and Franco, H., 1991, Geochemical complexities preserved in the volcanic rocks of the Zambales ophiolite, Philippines: Journal of Geophysical Research, v. 96, p. 16251-16262.

Evans, C.A., and Hawkins, J.W. Jr., 1989, Compositional heterogeneities in the upper mantle peridotites from the Zambales Range ophiolite, Luzon, Philippines: Tectonophysics, v. 168, p. 23-41.

Fournelle, J.H., 1991, Anhydrite and sulfide in pumices from the 15 June 1991 eruption of Mt. Pinatubo: Initial examination: Eos, Transactions, American Geophysical Union, v. 72, p. 68.

Fournelle, J, Carmody, R., and Daag, A.S., this volume, Anhydrite-bearing pumices from the June 15, 1991, eruption of Mount Pinatubo: Geochemistry, mineralogy, and petrology.

Gerlach, T.M., Westrich, H.R., and Symonds, R.B., this volume, Preeruption vapor in magma of the climactic Mount Pinatubo eruption: Source of the giant stratospheric sulfur dioxide cloud.

Geschwindt, C.-H., and Rutherford, M.J., 1993, Cummingtonite and the evolution of the Mount St. Helens (Washington) magma system: An experimental study: Geology, v. 20, p. 1011-1014.

Grove, T.L., and Kinzler, R.J., 1986, Petrogenesis of andesites: Annual Reviews of Earth and Planetary Science, v. 14, p. 417-454.

Haggerty, S.E., 1976, Opaque mineral oxides in terrestrial igneous rocks, in Rumble, Douglas, III, ed., Reviews in mineralogy, v. 3, Oxide minerals: Blacksburg, Va., Mineralogical Society of America, p. Hg101-Hg277.

Hansen, J., Lacis, A., Ruedy, R., and Sato, M., 1992, Potential climate impact of Mount Pinatubo eruption: Geophysical Research Letters, v. 19, p. 215-218.

Hattori, K., this volume, Occurrence and origin of sulfide and sulfate in the 1991 Mount Pinatubo eruption products.

Hoblitt, R.P., Wolfe, E.W., Scott, W.E., Couchman, M.R., Pallister, J.S., and Javier, D., this volume, The preclimactic eruptions of Mount Pinatubo, June 1991.

Huppert, H.E., Sparks, R.S.J., and Turner, J.S., 1982, Effects of volatiles on mixing in calc-alkaline magma systems: Nature, v. 297, p. 554-557.

------1984, Some effects of viscosity on the dynamics of replenished magma chambers: Journal of Geophysical Research, v. 89, p. 6857-6877.

Imai, A., Listanco, E.L., and Fujii, T., 1993, Petrologic and sulfur isotopic significance of highly oxidized and sulfur-rich magma of Mt. Pinatubo, Philippines: Geology, v. 21, p. 699-702.

Johnson, M.C., and Rutherford, M.J., 1989, Experimental calibration of the aluminum-in-hornblende geobarometer with application to Long Valley caldera (California) volcanic rocks: Geology, v. 17, p. 837-841.

Koyaguchi, T., this volume, Volume estimation of tephra-fall deposits from the June 15, 1991, eruption of Mount Pinatubo by theoretical and geological methods.

Luhr, J.F., 1990, Experimental phase relations of water- and sulfur-saturated arc magmas and the 1982 eruptions of El Chichón volcano: Journal of Petrology, v. 31, p. 1071-1114.

------1991, Volcanic shade causes cooling: Nature, v. 354, p. 104-105.

Luhr, J.F., Carmichael, I.S.E., and Varekamp, J.C., 1984, The 1982 eruptions of El Chichón volcano, Chipas, Mexico: Mineralogy and petrology of the anhydrite-bearing pumices: Journal of Volcanology and Geothermal Research, v. 23, p. 69-108.

Marsh, B.D., 1981, On the crystallinity, probability of occurrence, and rheology of lava and magma: Contributions to Mineralogy and Petrology, v. 78, p. 85-98.

Matthews, S.J., Jones, A.P., and Bristow, C.S., 1992, A simple magma-mixing model for sulfur behaviour in calc-alkaline volcanic rocks: Mineralogical evidence from Mount Pinatubo 1991 eruption: Journal of the Geological Society, London, v. 149, p. 863-866.

McKibben, M.A., and Eldridge, C.S., 1993, Sulfur isotopic systematics of the June 1991 eruptions of Mount Pinatubo: A SHRIMP ion microprobe study: Eos, Transactions, American Geophysical Union, v. 74, p. 668.

McKibben, M.A., Eldridge, C.S., and Reyes, A.G., 1992, Multiple origins of anhydrite in Mt. Pinatubo pumice: Eos, Transactions, American Geophysical Union, v. 73, p. 633-634.

------this volume, Sulfur isotopic systematics of the June 1991 Mount Pinatubo eruptions: A SHRIMP ion microprobe study.

Meeker, G.P., Pallister, J.S., and Hoblitt, R.P., 1993, Phenocryst-rich and -poor pumices from Mount Pinatubo: Eos, Transactions, American Geophysical Union, v. 74, p. 668.

Meeker, G.P., and Quick, J.E., 1991, OPUS: Old probe updated-software: Microbeam Analysis, v. 1991, p. 353.

Mori, J., Eberhart-Phillips, D., and Harlow, D.H., this volume, Three-dimensional velocity structure at Mount Pinatubo, Philippines: Resolving magma bodies and earthquake hypocenters.

Neilsen, R.L., 1990, Simulation of igneous differentiation processes, in Nicholls, J., and Russell, J.K., eds., Reviews in mineralogy, v. 27: Blacksburg, Va., Mineralogical Society of America, p. 65-105.

Newhall, C.G., Daag, A.S., Delfin, F.G., Jr., Hoblitt, R.P., McGeehin, J., Pallister, J.S., Regalado, M.T.M., Rubin, M., Tamayo, R.A., Jr., Tubianosa, B., and Umbal, J.V., this volume, Eruptive history of Mount Pinatubo.

Nielsen, C.H., and Sigurdsson, H., 1981, Quantitative methods for electron microprobe analysis of sodium in natural and synthetic glasses: American Mineralogist, v. 66, p. 547-552.

Paladio-Melosantos, M.L., Solidum, R.U., Scott, W.E., Quiambao, R.B., Umbal, J.V., Rodolfo, K.S., Tubianosa, B.S., Delos Reyes, P.J., and Ruelo, H.R., this volume, Tephra falls of the 1991 eruptions of Mount Pinatubo.

Pallister, J.S., Hoblitt, R.P., and Reyes, A.G., 1992, A basalt trigger for the 1991 eruptions of Pinatubo volcano?: Nature, v. 356, p. 426-428.

Pallister, J.S., Meeker, G.P., Newhall, C.G., and Hoblitt, R.P., 1993, 30,000 years of the "same old stuff" at Pinatubo: Eos, Transactions, American Geophysical Union: v. 74, p. 667-668.

Ramos, E.G., Laguerta, E.P., and Hamburger, M.W., this volume, Seismicity and magmatic resurgence at Mount Pinatubo in 1992.

Roeder, P.L., and Emslie, R.F., 1970, Olivine-liquid equilibrium: Contributions to Mineralogy and Petrology, v. 29, p. 275-289.

Rutherford, M.J., 1993, Experimental petrology applied to volcanic processes: Eos, Transactions, American Geophysical Union: v. 74, p. 49 and p. 55.

Rutherford, M.J., Brown, L., and Pallister, J.S., 1993, petrologic constraints on timing of magmatic processes in the 1991 Pinatubo volcanic system: Eos, Transactions, American Geophysical Union, v. 74, p. 671.

Rutherford, M.J., and Devine, J.D., 1991, Pre-eruption conditions and volatiles in the 1991 Pinatubo magmas: Transactions of the American Geophysical Union (Eos), v. 72, p. 62.

------this volume, Preeruption pressure-temperature conditions and volatiles in the 1991 dacitic magma of Mount Pinatubo.

Rutherford, M.J., and Hill, P.M., 1993, Magma ascent rates and magma mixing from amphibole breakdown: Experiments and the 1980-1986 Mount St. Helens eruptions: Journal of Geophysical Research, v. 98, p. 19667-19685.

Rye, R.O., Luhr, J.F., and Wasserman, M.D., 1984, Sulfur and oxygen isotopic systematics of the 1982 eruptions of El Chichón volcano, Chiapas, Mexico: Journal of Volcanology and Geothermal Research, v. 12, p. 109-123.

Sack, R.O., Carmichael, I.S.E., Rivers, M., and Ghiorso, M.S., 1980, Ferric-ferous equilibria in natural silicate liquids at 1 bar: Contributions to Mineralogy and Petrology, v. 75, p. 369-376.

Scandone, R., and Malone, S.D., 1985, Magma supply, magma discharge, and readjustment of the feeding system of Mount St. Helens during 1980: Journal of Volcanology and Geothermal Research, v. 23, p. 239-262.

Scott, W.E., Hoblitt, R.P., Torres, R.C., Self, S, Martinez, M.L., and Nillos, T., Jr., this volume, Pyroclastic flows of the June 15, 1991, climactic eruption of Mount Pinatubo.

Sigurdsson, H., Devine, J.D., and Davis, A.N., 1985, The petrologic estimate of volcanic degassing: Jökull, v. 35, p. 1-8.

Sisson, T.W., and Grove, T.L., 1993, Experimental investigations of the role of H2O in calc-alkaline differentiation and subduction zone magmatism: Contributions to Mineralogy and Petrology, v. 113, p. 143-166.

Smith, D.R., and Leeman, W.P., 1987, Petrogenesis of Mount St. Helens dacitic magmas: Journal of Geophysical Research, v. 92B, p. 10313-10334.

Sparks, R.J.S., Huppert, H.E., and Turner, J.S., 1984, The fluid dynamics of evolving magma chambers: Philosophical Transactions of the Royal Society of London, v. A310, p. 511-534.

Sparks, R.J.S., and Marshall, L.A., 1986, Thermal and mechanical constraints on mixing between mafic and silicic magmas: Journal of Volcanology and Geothermal Research: v. 29, p. 99-124.

Sparks, S.R.J., Sigurdsson, H., and Wilson, L., 1977, Magma mixing: A mechanism for triggering explosive eruptions: Nature, v. 267, p. 315-318.

Spulber, S.D., and Rutherford, M.J., 1983, The origin of rhyolite and plagiogranite in oceanic crust: An experimental study: Journal of Petrology, v. 24, p. 1-25.

Stimac, J.A., Pearce, T.H., Donnelly-Nolan, J.M., and Hearn, B.C., 1990, The origin and implications of undercooled andesitic inclusions in rhyolites, Clear Lake Volcanics, California: Journal of Geophysical Research, v. 95B, p. 17729-17746.

Stormer, J.C., 1983, The effects of recalculation on estimates of temperature and oxygen fugacity from analyses of multi-component iron-titanium oxides: American Mineralogist, v. 68, p. 586-594.

Taggart, J.E., Lindsay, J.R., Scott, B.A., Vivit, D.V., Bartel, A.J., Stewart, K.C., 1987, Analysis of geologic materials by X-ray fluorescence spectrometry, in Baedecker, P.A., ed., Methods for geochemical analysis: U.S. Geological Survey Bulletin 1770, p. E1-E19.

Ulmer, P., 1989, The dependence of the Fe2+-Mg cation partitioning between olivine and basaltic liquid on pressure, temperature and composition, an experimental study to 30 kbars: Contributions to Mineralogy and Petrology, v. 101, p. 261-273.

Westrich, H.R., and Gerlach, T.M., 1992, Magmatic gas source for the stratospheric SO2 cloud from the June 15, 1991 eruption of Mount Pinatubo: Geology, v. 20, p. 867-870.

White, R.A., this volume, Precursory deep long-period earthquakes at Mount Pinatubo, Philippines: Spatio-temporal link to a basalt trigger.

Wilcox, R.E., 1954, Petrology of Parícutin volcano, Mexico: U.S. Geological Survey Bulletin 965-C, p. 281-353.

Williams, H., and McBirney, A.R., 1979, Volcanology: San Francisco, Freeman, Cooper and Co., 397 p.

Wolfe, E.W., 1992, The 1991 eruptions of Mount Pinatubo, Philippines: Earthquakes and Volcanoes, v. 23, p. 5-37.

Wolfe, E.W. and Hoblitt, R.P., this volume, Overview of the eruptions.

Wright, T.L., and Doherty, P.C., 1970, A linear programming and least squares method for solving petrologic mixing problems: Geological Society of America Bulletin, v. 81, p. 1995-2008.

 

FIRE and MUD Contents

PHIVOLCS | University of Washington Press | U.S.Geological Survey

This page is <https://pubs.usgs.gov/pinatubo/pallister/>
Contact: Chris Newhall
Last updated 06.11.99