1 Institute of Mineralogy, Univ. of Hannover, Welfengarten 1, D-30167 Hannover GERMANY (f.holtz@mineralogie.uni-hannover.de)
2 Dept. of Earth and Planetary Sciences, Faculty of Science and Graduate School of Science and Technology, Kobe-University, Kobe, 657-8501 JAPAN
(hsato@kobe-u.ac.jp)
Abstract
We conducted crystallization experiments on the Unzen 1992 dacite mainly at 300 MPa and investigated phase stability relations and phase chemistry to constrain the conditions of crystallization of the Unzen 1991-1995 mixed magmas. The experimental pressure of 300 MPa corresponds to a depth of 10-11 km, which is that of the main magma chamber, determined from geodetic monitoring during the lava effusion of the 1991 to 1995 eruption of Unzen volcano. Experiments were performed between 875 and 775¡C (in 25 ¡C intervals) and variable X(H2O/(H2O+CO2)) with fO2 buffered at the Ni-NiO reaction. The starting material was dry glass prepared from the bulk rock powder of the dacite with known amount of H2O+CO2. The 1991-1995 dacite of Unze volcano show evidence of magma mixing, and a simple model of magma mixing of nearly phric high- temperature magma (1030 ¡C, SiO2=62 wt.%) and porphyritic low-temperature magma (790 ¡C, SiO2=67wt.%) is presented. We focused to constrain the crystallization condition of the low temperature magma in the present experiment. Most of the phenocryst phases of the low-temperature magma was reproduced experimentally at 300 MPa, 775-800¡C and X(H2)=0.8-1.0. There is some discrepancies between the phenocryst mineral assemblage in the low temperature magma and the experimental assemblage, i.e., quartz/and or orthopyroxene and biotite did not coexist in the experimental run products at 300 MPa, whereas they do in the low temperature phenocryst assemblage. This may be explained either by the lower pressure of crystallization of phenocrysts in the low-temperature magma than 300 MPa or by possible variation of conditions of phenocyrst crystallization in the low-temperature end member magmas in terms of temperature and X(H2O). Melt composition of the low-temperature end member magma was estimated from the composition of glass inclusions in phenocryst minerals (ca. 76-77 wt.% of SiO2), and is reproduced at temperatures of 775-800¡C, and X(H2O)=0.8-1.0 at 300 MPa. The experimentally produced plagioclases have XCa=Ca/(Ca+Na) ratio ranging from 0.77 to 0.45. XCa is higher at higher temperatures and X(H2O). The XCa ratio of natural plagioclase phenocryst ranges from 0.37-0.76, and may be explained by the variation of magmatic temperature and water activity. Al2O3 contents of amphiboles in the experimental charges are mostly 10-12 wt.per cent at 300 MPa, which is higher than those of the core of phenocryst amphibole in the 1991 Unzen dacites (7-9 wt.per cent), suggesting that the pressure of crystallization of amphibole phenocryst was at lower pressures.
1. INTRODUCTIONExperimental studies on the ejecta of volcanic eruption may elucidate the physical conditions of crystallization of phenocryst minerals in volcanic rocks. Recent eruptions whose eruption products have been studied for phase equilibrium experiments include those of 1980 Mount St.Helens (Rutherford et al., 1985; 1987; Geschwind and Rutherford, 1991), 1982 El Chichon (Luhr et al., 1990), 1991 Pinatubo (Scaillet and Evans, 1999), and 1995-1997 Soufriere Hills (Barclay et al., 1998). Concerning the 1991-1995 dome growth eruption of Unzen volcano, southwest Japan, Venetzky and Rutherford (1999) reported phase equilibrium data at 40-200 MPa under water- saturated conditions, however they briefly discussed the phase equilibrium relationship, and their data do not include the phase chemistry data. Sato et al.(1999) presented phase equilibrium and phase chemistry data on the groundmass of Unzen dacite which is mostly relevant to the groundmass crystallization processes at Unzen volcano. The present study is undertaken to constrain the magma chamber conditions through the phase equilibrium and phase chemistry data on the 1992 dacite of Unzen volcano. These experimental data are compared with the natural phenocryst assemblages and also the phase chemistry of the phenocryst and melt phases to infer the magmatic conditions and processes in the magma chamber of the 1991-1995 Unzen eruption.
2. Petrography of the 1991-1995 dacite and the magma mixing model
Previous studies (e.g., Nakada and Fujii, 1993; Nakamura, 1995; Nakada and Motomura, 1999) have shown that magma mixing of phyric low-temperature magma and nearly aphyric high- temperature magma took place in the formation of the 1991-1995 dacite. This is based on the petrographic evidences such as reverse zoning of hornblende and plagioclase and magnetite in the dacite. These studies reported that most of the phenocryst minerals belong to the low-temperature end member magma, and their equilibration temperature was estimated to be ca. 790 ¡C from the core compositions of iron-titanium oxide phenocrysts (Venetsky and Rutherford, 1999). These studies, however, did not reported the equilibration temperature and composition of the nearly aphyric high- temperature magma, and melt composition of the low-temperature magma. Here we describe the relevant petrography of the phenocrysts and melt inclusions and present a model of magma mixing for the 1991-1995 dacite of Unzen volcano. This is requisit for applying the experimental data to interprete the crystallization conditions of phenocyrst minerals in the mixed dacite. Petrography of the 1991-1995 dacite Phenocryst phases in the dacite includes plagioclase, quartz, hornblende, biotite, augite, orthopyroxene, magnetite, ilmenite, and apatite. Modal composition of the phenocrysts are ca. 20-30 volume percent (Nakada and Motomura, 1999). Figure 1 shows the histogram of the #Mg (=Mg/(Mg+Fe)) of mafic phenocrysts and An content of plagioclase phenocrysts in the 1991-1995 dacite of Unzen volcano, and shows bimodal composition for orthopyroxene. The #Mg number of the groundmass orthopyroxene is intermediate of the two modes of the #Mg of phenocyrst orthopyroxene, suggesting that the bimodal composition of orthopyroxene is caused by mixing of two orthopyroxene-bearing magmas. The Cr2O3 content of high #Mg orthopyroxenen and clinopyroxene are 0.2-0.5 wt.%. Chromium is refractory element and its content rapidly decreases as crystallization proceeds. Therefore, high Cr2O3 content of the high #Mg pyroxenes suggest that they crystallized from relatively primitive magmas and compose the phenocryst from high-temperature end member magma. The modal amount of pyroxenes are less than 0.1 vol.%. Other phenocrysts, i.e., low #Mg orthopyroxene, hornblende, bioitite, magnetite, ilmenite and plagioclase are derived from the low-temperature magma. It is noted that some calcic plagioclase constitute a zone in complexly zoned sodic plagioclase phenocryst, and they have low MgO (<0.02wt.%) and FeO* (0.2-0.3 wt.%). These phenocrysts contains inclusions of other minerals, suggesting coprecipitation of most of the minerals, i.e. plagioclase contains hornblende, biotite, magnetite, ilmenite, and apatite incluseions etc. These phenocrysts also contains glass inclusions often with bubbles, whose silica content is 75 to 81 wt.% (anhydrous basis), verifying that they are derived from low-temperature silicic end-member magma. Pyroxene thermometry (Anderson and Lindsley, 1989) gave ca. 1030 ¡C for the high #Mg pyroxenes, and 840 ¡C for low #Mg pyroxenes. A Magma Mixing Model of the 1991-1995 Dacite of Unzen Volcano Figure 2 illustrates our model of the magma mixing in the 1991-1995 dacite of Unzen volcano. Similar temperature versus composition diagram has already been proposed by Nakamura (1995), but we added new data constraining the temperature of the high-temperature magma. The temperature of the phyric end-member magma has been estimated by Venetzky and Rutherford (1999) to be around 790 ¡C. Their phenocryst composition are estimated from the compostion of the phenocryst minerals and their modal composition. The melt compostion is estimated from the composition of melt inclusions included in the phenocrysts of plagioclase, hornblende, and quartz. Because the melt inclusions in these phenocrysts are sufferred from melting of the host minerals due to heating by magma mixing, the original composition was restored by extrapolating the melt inclusion composition by subtracting the host mineral compositions. The SiO2 content of the melt of phyric end member magma is estimated to be 76-77 wt.% (anhydrous basis). The temperature of the mixed dacite was estimated from the compositions of the groundmass iron-titanium oxides to be 880 ¡C (Venetzky and Rutherford, 1999). In Figure 2, the SiO2 content of the aphyric end member magma was obtained graphically by the intersection of the straight line connecting the groundmass and melt of phyric end member magma. This is because the groundmass of the dacite is the mixture of bulk rock of aphyric (actually slightly pyroxene-bearing) end member magma and the melt of phyric end member magma, and specific heat may not vary so much and heat of mixing is negligible (Sparks and Marshall, 1986). The aphyric end member magma may have SiO2 around 60-62 wt.%. It is noted that such nearly aphyric pyroxene-bearing andesite-dacite (SiO2=60-67) occur in the Pliocene- Pleistocene formation in the surrounding areas of Unzen volcano (Matsumoto, 1966). Therefore, such pyroxene-bearing nearly aphyric magma may have injected the phyric dacite magma chamber just before the 1991 eruption of Unzen volcano. It is also noted that bulk rock SiO2 content of the low-temperature phyric end-member magma is ca. 67 wt.%, which is fairly similar to the bulk rock composition of the dacite, although they may differ in isotopic compositions judging from the variation of isotopic compositions of 87Sr/86Sr in phenocyrsts and bulk rocks (Chen et al. 1999).
3. EXPERIMENTS
3-1 Starting Material Table 1 shows the major element composition of the starting material for the present investigation together with some relevant compositions of the Unzen dacite. The starting material was a hot block ca. 2 meters in diameter recovered from the 1992 August 13 mud flow deposit in the Mizunashi river of Unzen volcano. The temperature obtained by infrared camera recorded ca. 500 ¡C within ca. 10 cm interior of the surface of the block, suggesting that the block was just derived from the active dome lava through pryoclastic flow and subsequent mud flow. The powder of the bulk rock is melted at ca. 1600 ¡C in a Pt crusible for 5 hours under atmospheric condition. The cooled glass was pluverized and put in Au or AgPd alloy capsules with distilled water and silver acetate?. Total volatile content was kept at ca. 10 wt.% of the total charge. 3-2 Experimental method Externally heated cold seal high pressure vessels at the Institut fur Mineralogie, Univ. of Hannover were used for the high temperature and pressure processing. The bombs are made of Nickel alloy with water as pressurizing medium. The oxygen fugacity was kept at Ni-NiO buffer conditions during the run. Table 2 lists the run conditions together with the phase assemblage of the run products. Temperature of the run was regulated within 10 ¡C, and pressure was kept within 5% of the nominal pressure. The run duration was 7 days, and the charges were quenched by air after the run, allowing cooling to 300 ¡C within ca. 1 minutes. The chrages were checked by weighting and presence of water was ascertained to check leak of capsule during the run. The run products were mounted in a epoxy resin and made into polished thin sections for optical microscope examination as well as for electron-probe microanalyses. 3-3 Phase relations Phase assemblages of the experimental runs are summarized in Table 2 and Figure 3. Figure 3 shows the temperature versus X(H2O) of fluid relations at 300 MPa. Representative back- scatterred electron images are shown in Figure 4. The upper photoes in Figure 4 are the BSE images at 875 ¡C, whereas the lower two photoes are the run charges at 850 ¡C. The right hand photoes are for X(H2O)=1.0, and the left two photoes are for X(H2O)=0.8. The crystallinity of the run charges increases as temperature and X(H2O) decreases. All the experimental conditions are below liquidus conditions. Solidus line is not determined in the present experiments and drawn arbitrary in Figure 3 in harmony with the experimental data of Ebadi and Johannes (1990). Plagiolcase liquidus for water saturated (X(H2O)=1.0) condition is 860-870 ¡C. Amphibole is stable in only X(H2O)>=0.8, whereas orthopyroxene occurs only in water undersaturated conditions. Biotite is present only in the run at 775 ¡C and X(H2O)=1.0. Quartz stability is limited to lower X(H2O) conditions less than 0.8. These experimental results are crystallization experiments using glass powder as starting material, and some reversed runs using the bulk rock powder of the dacite (92081301A) with pure water will be carried out using double capsule technique for 72 hours at 875 and 800¡C. 3-4 Phase chemistry Experimental run products and some natural dacite samples were analyzed by JEOL Superprobe JXA-8900 of the Venture Bussiness Laboratory of Kobe University. The analytical conditions are 15 KV accelerating voltage and 12 nA beam current. Most of the elements were analyzed for 20 second at the X-ray peak with 10 second counting on both of the side of the wave length for background. ZAF correction was made on the background subtracted countings. For analyses of hydrous silicic glass, Na counting was limited to 4 second with broad (mostly 10 microns in diameter) beam. Some of the analytical results were shown in the Table 3. Figure 5 shows the silica content of glass in the experimental run charges. The SiO2 content of the glass at 875 ¡C, X(H2O)=1.0 is 66 wt.%, and increases up to 78 as temperature and X(H2O) decreases. The melt composition of the low-temperature magma has SiO2 content of 76 to 77, which is mostly reproduced at 775-800 ¡C, X(H2O)=0.8-1.0. Figure 6 illustrates the compositions of the melt in the experimental run products compared with the bulk rock, groundmass, and melt inclusions in phenocrysts in the 1991-1995 dacite of Unzen volcano. The experimental melt composition delineate monotonous compositional variation in the oxide-silica diagrams. Although the experimental melt composition deviate from the natural bulk rock and groundmass compositions for some elements, such as TiO2, FeO*, MgO, and Al2O3, they converge to the melt inclusion compositions in the silicic ends. Figure 7 shows the An content of plagioclase in the experimental runs. Most calcic plagioclase is An77 at 850 ¡C and X(H2O)=1.0. An content is much lower for lower temperature and lower X(H2O) run products. As shown in Figure 1, An content of most of the plagioclase is mostly in a range of 40-50, which is reproduced in the experimental run prouducts at 775-800 ¡C and X(H2O)=0.8-1.0. Trace elements such as MgO, FeO* and K2O in plagioclase are higher than the natural plagioclase composition, which may partly be due to small grain size of the plagioclase in the experimental run products (Figure 4), and may partly be due to kinetic effect of rapid growth of the experimental plagioclase under relatively large undercooled conditions. Al2O3 contents of hornblende in the experimental run products are mostly above 10 wt.%, whereas, that of hornblende phenocryst in the natural dacite is between 6 and 8 wt.%. This may also be ascribed to the kinetic disequilibrium of the experimental run products.
4. DISCUSSION
4-1 Phase relations of the low-temperature magma and its crystallization conditions As described previously, phenocryst phases in the low-temperature end-member magma consists of plagioclase, hornblende, biotite, magnetite, ilmenite, quartz, orthopyroxene and apatite. These phenocrysts often contain mineral inclusions of the other minerals expect for quartz. Qurtz phenocrysts are usually corroded in outline as are quartz crystals in the experimental charges, and include only glass inclusions. The experimental phase relations at 300 MPa in Figure 3 shows that at temperatures of 775-800 ¡C, plagioclase and iron-titanium oxides are present irrespective of the variation of X(H2O), but biotite is only stable at X(H2O)=1.0, and hornblende is stable for X(H2O)>=0.8. On the other hand, orthopyroxene and quartz are not stable at X(H2O)>=0.8. One possible interpretation of this discrepancy between the natural phenocryst mineral assemblage of the low-temperature end member magma and the experimental phase assemblage is that there is variation in terms of X(H2O) and/or temperature of phenocryst crystallization. That is, quartz and orthopyroxene phenocrysts crystallized from a portion of magma chamber where X(H2O) is lower than the other part of the chamber where other phenocrysts such as biotite, hornblende, plagioclase crystallized. This model is consistent with the observation of mutual mineral inclusions in the phenocrysts phases. On the other hand, it is known that quatz stability decreases as pressure increases at water saturated conditions (Johannes and Holtz, 1995; Barclay et al., 1998). It is therefore possible that the magma chamber for the crystallization of phenocrysts in the low- temperature end member magma is at slightly lower pressure than 300 MPa. Ishihara (1993) pointed out the possible presence of magma pocket at depths of 5 and 8 kilimeters from leveling data, which can be the location of the low-temperature end member magma. However, the total volume of the magma pockets at 5 and 8 km are estimated to be ca. 10^7 m^3, which are much smaller than the total erupted material of 2*10^8 m^3. More than half of the erupted magmas are derived from the low-temperature magma, and we believe that such amount of magma may reside at the main chamber at a depth of 11 km as delineated by Nishi et al. (1999). Although there is possibility that some low-temperature end member magma was derived from shallow magma pocket, we prefer a model that most of the low temperature end member magma was derived from the main magma pocket at depth around 11 km, where slight variations in X(H2O) and temperature were present. 4-2 Melt compositions Figure 6 illustrates the melt compositions of the experimental charges, and compared them with the bulk rock composition, groundmass composition, and melt inclusions in phenocyrsts. Although the experimental glass composition slightly differs from the bulk rock and groundmass compositions in low SiO2 range, they converge toward the melt inclusion composition for SiO2 contents around 75-78 wt.% at experimental conditions of 775-800 ¡C and X(H2O)=0.8-1.0. Therefore the present experiments are consistent with the crystallization of the low-temperature magma at depth of 11 km. As noted previously, the low-temperature magma may contain about 50 wt.% of phenocyrsts, which corresponds to ca. 40-45 vol. % of phenocrysts. Therefore, the low- temperature magma was a crystal mush-like stage at the time of injection of the high-temperature nearly aphyric magma from below. Slight compositional diversity for low SiO2 melt composition of the experimental melt from the natural bulk rock and groundmass composition may reflect slightly different oxygen fugacity of the experiment from the natural dacite magma. The latter oxydation state was estimated at ca. NNO+1~1.5 logarithm unit (Nakada and Motomura, 1997; Venetzky and Rutherford, 1999). On the other hand, near liquidus run products contains magnetite and amphibole, which may reduced the FeO, and TiO2 contents, and increased the Al2O3 content in the melt phase of the experimental charges. 4-3 Mineral compositions As described in theprevious section, plagioclase An content (=100*Ca/(Ca+Na)) in the experimental charges range from 43 to 77. This mostly cover the compositional range of the plagioclase phenocryst in the dacite. At temperatures of 775-800 ¡C and X(H2O)=0.8-1.0, the An content of plagioclase is 43-52, which mostly explain the main population of An content of the plagioclase phenocryst core. More calcic plagioclase phenocryst in the dacite may have crystallized from more high temperature magmas. Calcic plagioclase in natural phenocryst conform calcic zone between more sodic plaigoclase, and the calcic plagioclase has low MgO and FeO* contents (Sato 1996). On the other hand, experimentally produced plagioclase has much higher MgO (>0.1wt.%) and FeO* (0.5-1wt.%) contents compared with the core of plagioclase phenocryst. This discrepancy may al least partly be ascribed to kinetic disequilibrium of element partitioning at high growth rate. Because the starting material was dry glass + H2O+CO2, the charges are under large undercooled conditions at the early period of the runs, in which crystallization proceeded. Such a large undercooling may cause rapid growth of crystal, and inpurity elements such as MgO and FeO* in plagioclase may accumulate the boundary layer melt next to the crystal and may be incorporated in the plagioclase crystal in excess of the equilibrium values. Anyhow, the low MgO and FeO* contents in the calcic zone of plagioclase phenocryst is a problem for further investigation. Available experimental data on the diffusion of MgO and FeO* in plagioclase indicate that if the low MgO and FeO* contents in the calcic zone was due to diffusion homogenization, it will take a period of several years (calculation based on the diffusion coefficient of xxxxx, 1997). At present, we presume the calcic plagioclase in the low-temperature end member magma recorded temporal injection history of high temperature magma into the low-temperature magma chamber. Al2O3 in hornblende is often used for thermometry and barometry (Holland and Blundy, 1994; Johnson and Rutherford, 1989). Al2O3 content of hornblende in experimental charges at 300 MPa are mostly in a range of 10-12 wt.% irrespective of temperature and X(H2O), whereas that of natural hornblende phenocryst core is in a range of 6-9 wt.% (Sato et al., 1999). Some of the low pressure hornblende is slightly lower in Al2O3, although there still remains the problem of kinetic disequilibrium of element partitioning in the experiments. Because the starting material consists of dry glass and water+CO2, initial stage of the run is fairly strongly undercooled condition, in which disequilibrium element partition may take place. 5. CONCLUSIONS Followings are the major conclusions of the present study. (a) A magma mixing model of the 1991-1995 dacite of Unzen volcano is presented. Pyroxene- bearing nearly aphyric high temperature end-member magma has bulk rock composition of 62-64 wt.% of SiO2 at 1030 ¡C, whereas, the low temperature phyric end member magma is at 790 ¡C with its melt compostion of 76-77 wt.% of SiO2. (b) High pressure experiments using bulk rock composition showed the phenocyrst mineral assemblages of the low-temperature magma are mostly reproduced at 775-800¡C, X(H2O)=0.8-1.0, 300 MPa, at which pressure the main magma chamber was geodetically estimated. The melt compostion of the low-temperature magma is also reproduced at these conditions. (c) There is slight discrepancy in terms of the crystallization conditions of biotite and quartz, and orhtopyroxene, suggesting that there is variation in the physical conditions of crystallization of these minerals in the low-temperature end member magmas, or that the depth of low-temperature end member magma is slightly at shallower depth. (d) Main population of plagioclase composition is reproduced in the experiments at above described conditions for the low-temperature magma, but some calcic plagioclase (An 55-75) are not reproduced at temperatures less than 800 ¡C. The calcic plagioclase may have crystallized at higher temperatures. The inconsistency between the low MgO and FeO* contents of the calcic plagioclase phenocrysts and the high contents in the experimental plagioclase are still to be resolved. Acknowledgements Thanks are due to Prof. Wilhelm Johannes, Setsuya Nakada, Keiko Suzuki-Kamata for discussions. We appriciate the technical assistance of Dr. Jurgen Koepke, Mr. Ditter Ziegenbein, and other technical staffs at the Institut fur Mineralogie, University of Hannover. Thoughtful review on the manuscript by Setsuya Nakada, and journal reviewers are of much help to improve the contents of the present paper. References Barclay, J. Rutherford, M.J., Carroll, M.R., Murphy, M.D., Devine, J.D., Gardner, J. and Sparks, R.S.J. (1998) Experimental phase equilibria constraints on pre-eruptive storage conditions of the Soufriere Hills magma. Geophys. Res. Lett., 25, 3437-3440. Chen, H., Nakada, S., Shieh, Y-N, DePaolo, D.J. (1999) The Sr, Nd and O isotopic studies of the 1991-1995 eruption at Unzen, Japan. J. Volcanol. Geotherm. Res., 89, 243-253. Ebadi, A. and Johannes, W. (1991) Beginning of melting and composition of first melts in the system Qz-Ab-Or-H2O-CO2. Contrib. Mineral. Petrol., 106, 286-295. Holtz, F. Pichavant, M., Barbey, P. and Johannes, W. (1992) Effects of H2O on the liquidus phase relations in the haplogranite system at 2 and 5 kbar. Amer. Mineralogist, 77, 1223-1241. Hoshizumi, H. Uto, K. and Watanabe, K. (1999) Geology and eruptive history of Unzen volcano, Shimabara Peninsula, Kyushu, SW Japan. J. Volcanol. Geotherm. Res., 89, 81-94. Housh, T.B. and Luhr, J.F., (1991) Plagioclase-melt equilibria in hydrous systems. Amer. Mineralogist, 76, 477-492. Johannes, W. (1978) Melting of plagioclase in the system Ab-An-H2O and Qz-Ab-An-H2O at PH2O=5kbars: an equilibrium problem. Contrib. Mineral. Petrol., 66, 295-303. Johannes, W. and Holtz, F. (1995) Petrogenesis and experimental petrology of granitic rocks. Springer, 335p. Luhr, J.F. (1990) Experimental phase relations of water- and sulfur-saturated arc magmas and the 1982 eruptions of El Chichon volcano. J. Petrology, 31, 1071-1114. Martel, C., Pichavant, M., Holtz, F., and Scaillet, B., Bourdier, J-L., Traineau, H. (1999) Effects of fO2 and H2O on andesite phase relations between 2 and 4 kbar. J. Geophys. Res., 104, 29453-29470. Nakada, S. and Motomura, Y. (1997) Nakada, S. and Motomura, Y. (1999) Petrology of the 1991-95 eruption at Unzen: effusion pulsation and groundmass crystallization. J. Volcanol. Geotherm. Res., 89, 173-196. Nakamura, M. (1995) Continuous mixing of crystal mush and replenished magma in the ongoing Unzen eruption. Geology, 23, 807-810. Nakamura, M. (1996) Time scale of magma mixing in Unzen eruption from 1991 to 1993: Continuous mixing model of crystal-rich magma and replenished magmas. Mem. Geol. Soc. Japan, 46, 127-138. Rutherford, M.J., Sigurdsson, H. and Carey, S. (1985) The May 18, 1980 eruption of Moung St. Helens: I, Melt composition and experimental phase equilibria. J. Geophys. Res., 90, 2929-2947. Sato, H., Nakada, S., Fujii, T., Nakamura, M., and Suzuki-Kamata, K. (1999) Groundmass pargasite in the 1991-1995 dacite of Unzen volcano: phase stability experiments and volcanological implications. J. Volcanol. Geotherm. Res., 89, 197-212. Scaillet, B. and Evans, B.W. (1999) The 15 June 1991 Eruption of Mount Pinatubo. I. Phase Equilibria and Pre-eruption P-T-fO2-fH2O conditions of the dacite magma. J. Petrol., 40, 381-411. Sisson, T.W. and Grove, T.L., (1993) Experimental investigations of the role of H2O in calc- alkaline differentiation and subduction zone magmatism. Contrib. Mineral. Petrol., 113, 143-166. Sparks, R.S.J and Marshall, (1986) J. Volcanol. Geotherm. Res., Venezky, D.Y. and Rutherford, M.J. (1999) Petrology and FeTi oxide reequilibration of the 1991 Mount Unzen mixed magma. J. Volcanol. Geotherm. Res., 89, 212-230. Caption of Figures Figure 1 #Mg number of mafic phenocrysts and An content of plagioclase phenocrysts in the 1991-1995 dacites of Unzen volcano Figure 2 A magma mixing model for the 1991-1995 dacite of Unzen volcano. Figure 3 Experimental phase relationship of Unzen dacite at 300 MPa, NNO bufferred condition. Figure 4 BSE images of some of the experimental run products. Run conditions are: A, 300 MPa, 875 ¡C, X(H2O)=0.8; B, 300 MPa, 875 ¡C, X(H2O)=1.0; C, 300 MPa, 850 ¡C, X(H2O)=0.8; D, 300 MPa, 850 ¡C, X(H2O)=1.0 Figure 5 SiO2 content of melt phase in the experimental run products. Figure 6 Silica variation diagram for the experimental melt phase and natural rock composition, and melt inclusions in phenocrysts. Figure 7 An content (100*Ca/(Ca+Na)) of plagioclase in the experimental run products. Figure 8 Water saturated P-T diagram of the Unzen dacite. Data are taken from Venetsky and Rutherford (1999) and this study. Caption of Tables Table 1 Starting material and relevant melt chemistry of Unzen 1991-1995 dacite. Table 2 Experimental run conditions and mineral assemblages of the runs Table 3 Chemical compositions of the phases in the experimental run products