Shift and Rotation of Composition Trends by Magma Mixing: 1983

JOURNAL OF PETROLOGY
VOLUME 44
NUMBER 10
PAGES 1895±1916
2003
DOI: 10.1093/petrology/egg063
Shift and Rotation of Composition Trends by
Magma Mixing: 1983 Eruption at Miyakejima Volcano, Japan
TAKESHI KURITANI*, TETSUYA YOKOYAMA,
KATSURA KOBAYASHI AND EIZO NAKAMURA
THE PHEASANT MEMORIAL LABORATORY FOR GEOCHEMISTRY AND COSMOCHEMISTRY, INSTITUTE FOR
STUDY OF THE EARTH'S INTERIOR, OKAYAMA UNIVERSITY, MISASA, TOTTORI 682-0193, JAPAN
RECEIVED AUGUST 20, 2002; ACCEPTED APRIL 15, 2003
levels of ascent of the fountain in the host andesite magma.
Analysis of compositional zoning in titanomagnetite crystals
revealed that the eruption of the 1983 magmas was initiated
soon after the replenishment of the basaltic magma in the 1 kbar
magma chamber.
Pre-eruption processes are investigated for magmas erupted in
1983 from Miyake-jima volcano, which is one of the most active
volcanoes in Japan. The whole-rock compositional trends of the
eruptive products are principally smooth and linear. Magmas
erupted from some fissures have compositions that deviate from
the main linear trend. Phenocryst contents of samples displaced
from the linear compositional trends are significantly lower than
those of samples on the main trends. Anorthite-rich plagioclase
phenocrysts, present throughout the 1983 products, are too calcic
to have crystallized from the erupted magma composition, and
were derived from a basaltic magma through magma mixing.
Although the linear whole-rock composition trends favor simple
two-component magma mixing, this cannot explain the presence
of samples that deviate from the main trend. Instead, the
observed composition trends were formed by mixing of a homogeneous basaltic magma with andesitic magmas exhibiting
compositional diversity. The original linear composition trends
of the andesitic end-member magma were rotated and shifted to
the direction of the basaltic end-member magma by magma
mixing. The samples out of the main trends represent magmas
with less basaltic component than those on the trend. The density
and viscosity of the basaltic end-member magma were comparable with those of the andesitic end-member magmas. The
basaltic magma, discharged from one magma chamber at
2 kbar pressure, was injected into a magma chamber at
lower pressure occupied by the chemically zoned andesite
magma ( 1 kbar), and possibly as a fountain. To establish
the characteristic mixing trend of the 1983 magma, the basaltic
component must have been distributed systematically in the zoned
andesite magma. A requirement is that the basaltic magma
spread laterally and mixed with the andesite magma at various
Magma mixing is a fundamental process in igneous
systems, and has been investigated widely with petrological, experimental, and theoretical approaches.
There is now a consensus that magma mixing occurs
effectively in replenished magma chambers and during
eruptions (e.g. Feeley & Dungan, 1996), and recent
studies have concentrated on the physical mechanisms
and dynamics of magma mixing (e.g. Snyder & Tait,
1996). Important constraints for the mechanisms of
magma mixing are the physical properties of the two
end-member magmas. Simple liquid---liquid blending
is inhibited when temperature and viscosity contrasts
between the two magmas are large and when the proportion of mafic end-member magma is small, typically
550%, because the mafic magma is undercooled to
form isolated inclusions in felsic magma (Bacon, 1986;
Sparks & Marshall, 1986). By detailed petrological
investigations, the mechanisms of magma mixing of
*Corresponding author. Fax: ‡81-858-43-3795.
E-mail: [email protected]
Journal of Petrology 44(10) # Oxford University Press 2003; all rights
reserved
KEY WORDS: compositional trend; liquid±liquid blending; magma
chamber; magma mixing; Miyake-jima Volcano
INTRODUCTION
JOURNAL OF PETROLOGY
VOLUME 44
this situation have been studied extensively (e.g. Feeley
& Dungan, 1996; Clynne, 1999).
When the differences between the physical properties
of two end-member magmas are not large, liquid--liquid mixing is likely to occur. The dynamics of
magma mixing in recharged magma chambers of this
situation are the target of the present study. Interactions of two magmas with small contrasts of physical
properties have been well studied with laboratory
experiments and associated theoretical analyses (e.g.
Huppert & Sparks, 1980; Campbell & Turner, 1989).
Although analogue experiments are useful for elucidating some of the fundamental controls on magma
mixing, it is necessary to have evidence of postulated
mechanisms from natural observations. On the other
hand, natural observation can provide direct information on the mixing mechanism involved in individual
magmatic systems.
Miyake-jima Island is one of the most active volcanoes in Japan. The volcano has erupted about every 20
years (1962, 1983, 2000) since 1940. The 1983 lavas
and pyroclastic rocks, investigated in this study, are
basaltic andesitic in bulk composition. Previous petrological and mineralogical studies on the 1983 products
include those by Fujii et al. (1984) and Soya et al.
(1984). The pre-eruption magmatic history was, however, not discussed in detail in these studies. Miyasaka
& Nakagawa (1998) showed that phenocrysts in the
1940, 1962, and 1983 products can be divided into
three types: A-type (andesite type), B-type (basalt
type), and M-type (megacryst type). Those workers
noted that the 1940 magma contains all three types
of phenocrysts, and was produced by magma mixing.
In contrast, the 1962 and 1983 magmas contain solely
A-type phenocrysts, and formed without significant
magma mixing events.
In this paper, we present detailed petrographic,
mineralogical, and geochemical descriptions of the
1983 products to investigate the pre-eruptive history
of the magmas. Contrary to previous studies, it is
shown that the 1983 magma is a product of mixing
between homogeneous basaltic magma and heterogeneous andesitic magmas. The mechanism of magma
mixing is discussed using the characteristic mixing
trends of whole-rock compositions, estimated physical
properties of end-member magmas, and an estimated
time scale from magma mixing to eruption.
GEOLOGICAL SETTING
Miyake-jima is a volcanic island located about 200 km
south of Tokyo, Japan (Fig. 1a). Quaternary tholeiitic
basalt and andesite compose a large part of the volcano. Tsukui & Suzuki (1998) divided the formation
history of the volcano into four stages since 7000 years
NUMBER 10
OCTOBER 2003
BP: (I) inactive stage, from 7000 to 4000 years BP; (II)
caldera-forming stage, from 4000 to 2500 years BP;
(III) Oyama stage, from 2500 years BP to the 15th
century; (IV) Shinmio stage, from AD 1469 to the
present. The inactive stage is characterized by
repeated small-scale eruptions with dormancy. In the
caldera-forming stage, a caldera was formed by voluminous eruption of lapilli and scoriae. In the Oyama
stage, the caldera was filled by products from central
and lateral eruptions. Most eruptions in the Shinmio
stage occurred from lateral fissures (Tsukui & Suzuki,
1998). The volcano is still active and has erupted three
times (1963, 1983, and 2000) since 1940 (Fig. 1a).
The materials of the 1983 eruption were lavas
and pyroclastic rocks, with a volume of 001 km3
(Aramaki & Hayakawa, 1984). The eruptive fissures
were about 45 km in length and ran from the southwestern flank of the volcano (Fig. 1b). Lavas effused
from A---E craters flowed down to the west, and
those from G---K craters flowed to the SW. Eruptive
products from P---S craters were mainly scoria. The
detailed sequence of the 1983 eruption was described
by Aramaki & Hayakawa (1984).
ANALYTICAL METHODS
Whole-rock major and trace elements, and Sr, Nd,
and Pb isotopic compositions were measured at the
Pheasant Memorial Laboratory (PML), Institute for
Study of the Earth's Interior, Okayama University at
Misasa. Rock specimens were crushed by a jaw crusher
to coarse chips of 3---5 mm in diameter, and then fresh
chips were carefully hand-picked. The chips were
rinsed with deionized water in an ultrasonic bath at
least three times, and then dried at 100 C for 12 h. The
washed chips were ground using an alumina pack mill.
Concentrations of major elements, Ni, and Cr were
obtained from glass beads containing a lithium tetraborate flux (10 to 1 dilution of sample) using a Phillips
PW2400 X-ray fluorescence spectrometer (Takei,
2002). Loss on ignition (LOI) was obtained gravimetrically, and FeO content was determined by the
titration method of Yokoyama & Nakamura (2002).
Trace elements were analyzed using a Yokogawa
PMS2000 inductively coupled plasma mass spectrometer fitted with a flow injection system using
the methods of Makishima & Nakamura (1997),
Makishima et al. (1997, 1999), and Yokoyama et al.
(1999). Concentrations of B, Li, Zr, and Hf were determined by isotope dilution and those of other trace
elements by the calibration-curve method. All of the
major and trace elements analyses were duplicated for
each sample, and replicate analyses were always less
than 02 and 3---5% relative percent difference, for
major and trace elements, respectively.
1896
KURITANI et al.
MAGMA MIXING TRENDS, MIYAKE-JIMA VOLCANO
Fig. 1. (a) Index map showing the location of Miyake-jima Island and distributions of lavas and pyroclastics of the 1940, 1962, and 1983
eruptions (after Miyasaka & Nakagawa, 1998); (b) distributions of craters and lava flows of the 1983 eruption (after Aramaki & Hayakawa,
1984). Groups of craters are indicated as A---K and P---S. Encircled numbers in (b) are those listed in Table 1. The distributions of lavas are
those before the 2000 eruption. The main products of the 2000 eruption were ejected on the sea floor and from the collapsed caldera at the top
of the mountain.
The analytical procedures for chemical separation and
mass spectrometry followed Yoshikawa & Nakamura
(1993) for Sr isotope measurements, Makishima &
Nakamura (1991) for Nd, and Kuritani & Nakamura
(2002) for Pb. Mass spectrometry was carried out with
modified Finnigan MAT261 (Nakano & Nakamura,
1998) and MAT262 thermal ionization mass
spectrometers in static multi-collection mode. Normalizing factors to correct isotopic fractionation during
analysis are 86 Sr/88 Sr ˆ 01194 for Sr and 146 Nd/
144
Nd ˆ 07219 for Nd. For Pb isotope analysis,
100 ng Pb were loaded and the data were normalized
1897
JOURNAL OF PETROLOGY
VOLUME 44
NUMBER 10
OCTOBER 2003
using the measured ratio of NBS981 standard and its
recommended ratio given by Todt et al. (1996). Analytical reproducibility for natural rock samples was
0002% for 87 Sr/86 Sr, 0002% for 143 Nd/144 Nd, and
0010%, 0015%, and 0017% for 206 Pb/204 Pb, 207 Pb/
204
Pb, and 208 Pb/204 Pb, respectively.
Mineral compositions were determined by a JEOL
JXA-8800 electron microprobe, located at the Institute
for Study of the Earth's Interior. An accelerating voltage of 15 kV and a beam current of 20 nA were used,
and counting time was normally 20 s. Both oxide and
natural mineral standards were used, and data were
obtained with ZAF correction.
WHOLE-ROCK COMPOSITION
Whole-rock major and trace element contents for
representative samples numbered in Fig. 1b are listed
in Table 1. Figure 2 shows a TiO2---SiO2 variation
diagram for samples of the 1983 eruption (filled
squares; this study) and those of other eruptions from
Miyake-jima volcano (Yokoyama et al., 2003). The
compositional spread of samples in the June---July
2000 eruption (SiO2 53 wt %) and that of samples
in the August 2000 eruption (SiO2 51 wt %) are also
shown (Uto et al., 2001; Geshi et al., 2002). The samples
of stages (III) and (IV), including the 1983 eruption,
are relatively rich in SiO2 and TiO2 compared with
those of the other stages. Figure 3 shows Harker variation diagrams for some major element oxides [TiO2,
Al2O3, Fe2O3 (total Fe as Fe2O3), MgO, CaO, and
Na2O] plotted against SiO2 content for samples of the
1983 eruptions. The products principally form smooth
compositional trends, but those from the A, B, and G
craters deviate slightly from trends formed by the bulk
of the samples. At given SiO2 contents, these samples
plot at lower Al2O3 and CaO contents and higher
TiO2 and Fe2O3 contents than the samples on the
main composition trends (Fig. 3). The whole-rock
compositions on the main trends exhibit a spatial
distribution: products from C---E craters have
526---529 wt % SiO2, those from J and K craters
have 529---538 wt %, and those from H and I craters
have 539---547 wt % (shown in MgO---SiO2 diagram
in Fig. 3).
Trace element concentrations of the representative
samples normalized to the values of normal-type midocean ridge basalt (N-MORB) (Sun & McDonough,
1989) are shown in Fig. 4a. Marked negative anomalies of Nb and Ta, and positive anomalies of Pb, Sr, and
Li are observed, which are characteristic of island-arc
magmas. Within the compositional variation of the
1983 samples, the more SiO2-rich samples tend to be
more enriched in incompatible trace elements (Fig. 4b).
Table 2 lists 87 Sr/86 Sr, 143 Nd/144 Nd, 206 Pb/204 Pb,
Fig. 2. Whole-rock TiO2---SiO2 variation diagram for samples from
all eruption stages of Miyake-jima volcano. The data except for those
of 1983 products (&) are from Yokoyama et al. (2003). The compositional fields of samples of the 2000 eruption are shown based on the
data of Uto et al. (2001) and Geshi et al. (2002). The dashed line
indicates the linear extrapolation of the main composition trend of
the 1983 products. (See text for details.)
207
Pb/204 Pb, and 208 Pb/204 Pb ratios of the representative 1983 samples. 87 Sr/86 Sr and 208 Pb/204 Pb are
plotted against SiO2 content in Fig. 4c and d. Although
the variation of isotopic compositions throughout the
lavas exceeds the analytical uncertainty, there is no
systematic change with the whole-rock SiO2 content.
PETROGRAPHY AND MINERALOGY
The phenocryst assemblage of the 1983 products is
plagioclase, olivine, augite, titanomagnetite, and rare
orthopyroxene. Plagioclase is the most abundant
phase, and the modal proportion of plagioclase,
olivine, and augite phenocrysts is typically 97:1:2,
independent of the whole-rock composition of the samples. Phenocryst contents of the representative samples,
which are obtained by image analysis with 5 million
pixels (per thin section) for individual samples, are
shown in Fig. 5. Standard deviations of the phenocryst
contents are given for some samples, which were
obtained by analyses of three or four thin sections for
each sample. The phenocryst contents of samples,
other than those from the A, B, and G craters, tend to
decrease with increasing whole-rock SiO2 content. At a
given SiO2 content, the products from A, B, and G
craters have lower phenocryst contents than the other
main samples. Petrographic and mineralogical features
of each mineral phase are described below.
Plagioclase
Plagioclase phenocrysts are prismatic and up to 2 mm
long. Most plagioclase phenocrysts are clear, although
1898
MAGMA MIXING TRENDS, MIYAKE-JIMA VOLCANO
KURITANI et al.
Table 1: Whole-rock compositions of the representative samples of 1983 products
Crater:
C---E
C---E
C---E
J,K
A
J,K
G
S
J,K
H,I
H,I
Number:
1
2
3
4
5
6
7
8
9
10
11
12
Sample:
myk0122
myk3043
myk3024
myk0146
myk0107
myk0149
myk0317
myk0305
myk0155
myk0142
myk0136
myk0138
53.28
1.37
53.51
1.37
53.39
1.36
53.49
1.40
53.66
1.35
53.89
1.37
53.86
1.33
54.21
1.35
54.52
1.33
54.82
1.32
55.24
1.30
15.16
3.25
15.22
3.47
15.16
4.22
14.69
3.88
15.18
4.68
14.95
3.72
15.26
5.81
15.15
2.79
15.14
3.03
15.08
2.68
14.99
2.10
9.99
0.24
9.79
0.24
9.01
0.24
9.66
0.24
8.39
0.23
9.43
0.24
7.24
0.23
10.06
0.24
9.65
0.24
9.84
0.24
10.09
0.23
4.01
9.15
4.02
9.16
4.00
9.09
4.10
8.90
3.89
8.96
3.95
8.86
3.90
8.96
3.87
8.90
3.82
8.77
3.76
8.62
3.68
8.43
2.79
0.54
2.79
0.54
2.81
0.55
2.79
0.55
2.82
0.56
2.87
0.57
2.81
0.56
2.87
0.57
2.93
0.59
2.94
0.60
3.01
0.62
0.15
0.82
0.15
0.75
0.15
0.50
0.16
0.76
0.15
0.66
0.15
0.74
0.15
0.52
0.15
0.81
0.16
0.87
0.16
0.85
0.16
0.85
99.91
52.74
100.25
52.80
99.98
52.87
99.86
99.87
99.99
52.99
53.23
53.33
100.11
53.37
100.17
53.52
100.17
53.85
100.06
54.20
99.85
54.71
26.6
8.27
27.1
7.85
26.1
7.41
26.1
6.84
26.0
8.63
28.5
8.99
27.4
8.79
25.9
7.57
24.4
7.67
24.7
7.71
22.1
7.22
7.13
18.4
7.64
7.15
18.3
8.11
6.55
18.7
7.59
7.14
7.05
7.16
7.36
7.33
7.01
7.51
7.70
19.3
8.30
18.8
8.32
19.2
8.41
19.2
8.15
18.5
8.43
17.6
8.20
19.8
8.44
20.3
9.26
Major elements (wt %)
SiO2
53.25
TiO2
Al2O3
Fe2O3
FeO
MnO
MgO
CaO
Na2O
K2O
P2O5
1.37
15.22
3.57
9.70
0.24
4.01
9.17
2.78
0.54
0.15
Total
0.88
100.00
SiO2
52.68
LOI
Trace elements (ppm)
Cr
28.9
Ni
Li
B
Rb
Sr
Y
Cs
Ba
8.66
6.85
18.2
7.43
247
33.2
0.55
199
251
34.2
0.54
198
251
35.3
0.57
203
243
35.2
0.55
200
252
38.1
0.58
214
252
35.8
0.59
210
255
38.1
0.59
210
251
35.7
0.57
205
254
37.8
0.60
216
255
38.1
0.53
216
239
35.9
0.60
215
H,I
255
39.1
0.65
235
3.48
10.2
3.58
10.4
3.60
10.6
3.46
10.4
3.74
11.0
3.79
10.8
3.75
10.8
3.64
10.7
3.78
11.1
3.74
11.3
3.83
11.3
1.72
9.61
1.67
9.71
1.81
10.17
1.69
9.67
1.91
10.5
1.87
10.3
1.86
10.4
1.81
10.1
1.85
10.5
1.86
10.6
1.89
10.4
3.25
1.20
3.26
1.19
3.31
1.23
3.32
1.17
3.57
1.29
3.40
1.25
3.48
1.25
3.31
1.22
3.42
1.24
3.40
1.31
3.49
1.26
4.29
0.81
4.29
0.82
4.43
0.85
4.26
0.80
4.70
0.87
4.38
0.84
4.61
0.87
4.41
0.83
4.62
0.86
4.53
0.86
4.61
0.86
5.23
1.19
5.37
1.20
5.53
1.22
5.32
1.18
5.79
1.29
5.51
1.23
5.76
1.23
5.52
1.23
5.59
1.26
5.71
1.28
5.75
1.26
3.19
0.50
3.25
0.50
3.35
0.52
3.26
0.50
3.54
0.55
3.40
0.53
3.37
0.53
3.35
0.51
3.45
0.55
3.43
0.53
3.46
1.33
3.64
3.46
0.51
3.38
0.51
3.58
0.52
3.46
0.51
3.76
0.54
3.56
0.52
3.66
0.53
3.55
0.51
3.60
0.55
3.72
0.54
0.55
3.68
0.57
3.87
3.70
0.34
3.27
0.34
3.37
0.35
3.30
0.34
3.61
0.37
3.57
0.37
3.47
0.37
3.19
0.37
3.40
0.38
3.53
0.38
0.54
3.53
0.57
3.76
0.20
66.8
0.21
66.9
0.21
65.1
0.20
66.5
0.21
65.0
0.22
64.6
0.22
66.5
0.22
65.8
0.23
66.7
0.23
68.0
0.38
0.23
0.42
0.24
1.88
0.53
1.88
0.54
1.89
0.53
1.91
0.57
1.92
0.56
1.94
0.57
1.92
0.55
1.99
0.59
2.05
0.57
69.3
2.10
72.3
2.15
Nb
1.86
0.54
Ta
0.04
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.59
0.05
0.57
0.05
La
Ce
Pr
Nd
Sm
Eu
Gd
Tb
Dy
Ho
Er
Tm
Yb
Lu
Pb
Th
U
Zr
Hf
Normalized content for the total weight to be 100% based on total Fe as Fe2O3.
1899
4.01
12.0
2.01
11.1
3.79
1.29
4.81
0.91
6.03
JOURNAL OF PETROLOGY
VOLUME 44
NUMBER 10
OCTOBER 2003
Fig. 3. SiO2 variation diagrams for some major oxides [TiO2, Al2O3, Fe2O3 (total Fe as Fe2O3), MgO, CaO, and Na2O] in the 1983
eruptive products. In the MgO---SiO2 diagram, the craters from which magmas were derived are indicated for samples on the main linear
trend (&). Major element analyses are normalized to 100 wt % with total Fe as Fe2O3.
some with glass inclusions are also present. Some
plagioclase phenocrysts have thin sodic rims (Fig. 6a),
which have similar An contents [100 Ca/(Ca ‡ Na ‡
K)] to quenched crystals in the groundmass; these
probably crystallized during the eruption. The An
content of the region just inside the sodic rim of plagioclase phenocrysts is shown in Fig. 7. This gives the An
content of plagioclase grown at the crystallization stage
just before eruption.
Plagioclase phenocrysts can be divided into two
groups, those with high An contents (4An80; referred
to as high-An type) and those with low An contents
(5An72; low-An type). The high-An type phenocrysts
are macroscopically homogeneous with oscillatory
zoned cores surrounded by the sodic rim (Fig. 6a).
Extremely calcic cores with An90---95 are rarely present.
The An content of the oscillatory region just inside the
sodic rim ranges from 80 to 88. There is no systematic
variation with the whole-rock SiO2 content of the host
samples (Fig. 7). The MgO content of the region just
inside the rim tends to increase slightly with decreasing
An content (Fig. 8).
The low-An type phenocrysts commonly exhibit
normal zoning in An content. Some crystals of this
type show rounded margins. The size of the low-An
type crystals is small; commonly5300 mm. It is unclear
1900
KURITANI et al.
MAGMA MIXING TRENDS, MIYAKE-JIMA VOLCANO
Fig. 4. (a) MORB-normalized trace element concentrations of the 1983 products. SiO2 variation vs: (b) La, (c) 87 Sr/86 Sr, and (d) 208 Pb/204 Pb
for the 1983 products. Trace element concentrations of N-MORB are from Sun & McDonough (1989). The error bars in (b) are 5% of the
data, which is the maximum value in the measurements by ICP-MS. The error bars for the isotopic ratios are within the scales of the plot
marks. The symbols in (b)---(d) are the same as those in Fig. 3.
Table 2: Whole-rock isotopic compositions of
the representative samples of 1983 products
Sample
87
Sr/86 Sr
143
Nd/144 Nd
206
Pb/204 Pb
207
Pb/204 Pb
208
Pb/204 Pb
myk0122
070343
051309
182623
155077
380780
myk3043
070344
051307
183023
154994
380936
myk3024
070345
051308
182831
155037
380845
myk0146
070344
051307
182999
154988
380877
myk0107
070343
051308
182707
155089
380918
myk0149
070346
051307
182918
155047
380960
myk0317
070349
051307
182885
155075
381025
myk0305
070351
051308
183014
154963
380831
myk0155
070346
051309
182989
155028
380974
myk0142
070338
051308
183020
155013
380976
myk0136
070345
051307
182985
155032
380985
myk0138
070343
051307
183046
155002
380979
2SD
000001
000001
00018
00023
00065
2SD is 2 standard deviation for analyses of natural rock
samples.
Fig. 5. Phenocryst content of the representative samples plotted
against their whole-rock SiO2 contents. The phenocryst contents
were determined by image analyses with 5 120 000 pixels per thin
section for individual samples. For some samples, the standard deviations of the phenocryst contents, which are obtained by analyses of
three or four thin sections for each sample, are shown.
1901
JOURNAL OF PETROLOGY
VOLUME 44
NUMBER 10
OCTOBER 2003
Fig. 6. (a) Back-scattered electron image (BEI) of high-An plagioclase phenocryst (A group phenocryst), showing that the crystal developed
by oscillatory zoning and is rimmed with sodic composition (sodic rim); (b) BEI of crystal aggregate consisting of olivine, high-Mg number
augite, and high-An plagioclase (A group phenocrysts); (c) BEI of crystal aggregate consisting of low-Mg number augite, low-An plagioclase,
and high-Usp number titanomagnetite (B group phenocrysts); (d) BEI of crystal aggregate composed of orthopyroxene, low-Mg number
augite, high-Usp number titanomagnetite, and low-An plagioclase (B group phenocrysts), in which interstitial glass is observed. The scale bar
represents 100 mm. opx, orthopyroxene; mt, titanomagnetite; pl, plagioclase; aug, augite; gls, glass.
if a correlation exists between the An content of these
grains and the whole-rock composition of the host
samples, because they are scarce. In contrast to the
high-An type plagioclase, the MgO content of the
region inside the rim does not increase with decreasing
An content (Fig. 8). The sodic rims of the low-An type
plagioclase have similar MgO and FeO contents to
those of the high-An type plagioclase, suggesting that
the rims of this type of plagioclase also formed during
eruption. In the low-An type plagioclase, both MgO
and FeO contents of the region inside the rims are
lower than those of the sodic rims (Fig. 8).
Olivine
Most olivine phenocrysts are up to 1 mm in diameter,
and euhedral (Fig. 6b), although some with slightly
rounded margins are also present. They commonly
occur as isolated grains. In addition, some olivine phenocrysts form crystal aggregates with augite, high-An
plagioclase, and titanomagnetite (Fig. 6b); but they
are never found with low-An type plagioclase. Olivine
phenocrysts are generally homogeneous in Mg number
[100 Mg/(Mg ‡ Fe)] throughout the crystals. The
Mg number of olivine phenocrysts ranges from 68 to 70
throughout the lavas, and does not vary with the
whole-rock SiO2 content of the host samples.
Pyroxenes
Augite phenocrysts are up to 1 mm in length, and
euhedral (Fig. 6b). They occasionally have inclusions
of glass, olivine, plagioclase, and titanomagnetite.
Throughout the crystals, weak oscillatory zoning is
1902
KURITANI et al.
MAGMA MIXING TRENDS, MIYAKE-JIMA VOLCANO
Fig. 8. MgO and FeO contents of the sodic rim and the region just
inside the sodic rim for the high- and low-An type plagioclase phenorysts, plotted against the An content. FeO is total Fe as FeO.
Fig. 7. Histogram of An content of the regions just inside the sodic
rim in plagioclase phenocrysts, for samples from different craters. The
order of the histograms roughly corresponds to the order of the
whole-rock SiO2 content.
developed, and is commonly superimposed on sector
zoning. Rarely, small augite crystals (typically
5500 mm) with subhedral to euhedral shapes are
present, which commonly coexist with the low-An
plagioclase (Fig. 6c and d). Orthopyroxene crystals are
very rarely present. They have subhedral to euhedral
shapes, are 5300 mm across, and commonly form crystal clots with augite, titanomagnetite, and low-An type
plagioclase (Fig. 6d).
Augite phenocrysts can be divided into two groups
by their Mg number. The Mg numbers of crystals
coexisting with high-An plagioclase (Fig. 6b) exceed
70, whereas those of crystals with low-An plagioclase
(Fig. 6c) are 570. The Mg number of orthopyroxene
crystals at the core commonly ranges from 60 to 66.
Titanomagnetite
Titanomagnetite crystals are euhedral and up to
300 mm in size. The ulv
ospinel components [Usp
Fig. 9. Zoning profile of ulv
ospinel component in a representative
titanomagnetite crystal. The abscissa indicates the distance from the
core of the crystal. Calculated diffusion profiles are shown for
comparison. (See text for details.)
number; 100 usp/(usp ‡ mt)] of most crystals are
between 20 and 24 throughout the lavas. These crystals
commonly coexist with plagioclase of the high-An type
and high-Mg number augite phenocrysts, or occur as
inclusions in olivine, augite, and the high-An plagioclase (Fig. 6b). They are zoned at the margin of the
crystals, and the Usp number increases to 30 towards
the rim (Fig. 9). Some titanomagnetite crystals have
high Usp number of 435; these occasionally form
crystal clots with low-Mg number augite, orthopyroxene, and low-An plagioclase (Fig. 6c and d).
They are also zoned at the margin, and the Usp number decreases to about 30 towards the rim.
1903
JOURNAL OF PETROLOGY
VOLUME 44
NUMBER 10
OCTOBER 2003
reported dacite xenoliths in the 1983 products. However, these xenoliths were brought to the magmas
accidentally during eruption, and thus they are not
genetically related to the 1983 magmas (Fujii et al.,
1984).
DISCUSSION
Origin of phenocrysts
Fig. 10. Summary of the petrographic and mineralogic features of
the 1983 eruptive products. Phenocrysts can be divided into two
groups: the A group consists of high-An plagioclase, olivine, highMg number augite, and low-Usp number titanomagnetite; the B
group consists of low-An plagioclase, low-Mg number augite, orthopyroxene, and high-Usp number titanomagnetite.
Summary of petrographic and mineralogical
features of the 1983 products
Figure 10 summarizes the petrographic and mineralogical features of the 1983 samples. Phenocrysts can be
divided into two groups: crystal aggregates consisting
of high-An plagioclase, olivine, high-Mg number
augite, and low-Usp number titanomagnetite (referred
to as A group; Fig. 6a and b), and those of low-An
plagioclase, low-Mg number augite, orthopyroxene,
and high-Usp number titanomagnetite (referred to as
B group; Fig. 6c and d). The A-group phenocrysts
defined above correspond to the `A type' of Miyasaka
& Nakagawa (1998) and Amma-Miyasaka &
Nakagawa (2002); on the other hand, the B-group
phenocrysts were not classified by them. The modal
abundance of the B-group phenocrysts is always 501
vol. % throughout the lavas; therefore, most phenocrysts of the 1983 samples belong to the A group. The
compositions of the A-group phenocrysts exhibit no
systematic variation throughout the lavas (e.g. Fig. 7).
In the 1983 products, crystals that are not categorized into the A and B groups are very rarely present.
These are considered to be xenocrysts. They include
homogeneous plagioclase megacrysts with An95 and
crystal aggregates consisting of calcic plagioclase
( An95) and Mg-rich olivine (Mg number 80),
which shows cumulate structure. Fujii et al. (1984)
The products of the 1983 eruption commonly contain
3---5 vol. % phenocrysts, most of which are classified
into the A group. To clarify the pre-eruption magmatic
processes of the 1983 products, it is therefore crucial to
explain the origin of the A-group phenocrysts. In this
section, the origin of these crystals is discussed mainly
on the basis of thermodynamic equilibria between
plagioclase and silicate melt.
The phenocrysts of the A group are principally euhedral (Fig. 6a and b), although some olivine phenocrysts
have slightly rounded margins. One possible origin is
that the phenocrysts crystallized from magmas similar
in composition to observed whole-rock compositions at
nearly equilibrium conditions. To test this hypothesis,
plagioclase---melt thermodynamic equilibrium was
examined for high-An plagioclase phenocrysts in sample myk0122 (Table 1), using thermodynamic solution
models for plagioclase (Elkins & Grove, 1990) and for
silicate melt (Ghiorso & Sack, 1995). Both equilibrium
temperature and An content of plagioclase can be
calculated independently, if pressure and melt composition are specified. Seismic studies have shown that
earthquake foci associated with the 1983 eruption
were distributed at depth from 2---3 km to 6---7 km
beneath Miyake-jima island (e.g. Miyazaki & Sawada,
1984). Given that a magma chamber was present
somewhere at depth between 2---3 km and 6---7 km,
thermodynamic equilibrium was examined for a pressure of both 1 and 2 kbar. Melt composition was calculated from whole-rock, mineral, and modal
compositions. Conversion of volume fractions of phenocryst contents to weight fractions was ignored,
because this does not affect the result. Using the compositions of augite phenocrysts in samples that were not
saturated with orthopyroxene, the minimum temperature of the melt coexisting with the A-group phenocrysts can be estimated, because of the constraint that
augite was present at a temperature above the orthopyroxene---clinopyroxene solvus. The chemical compositions of the augite phenocrysts give temperature
estimates between 1020 and 1080 C, using the method
of Lindsley (1983). A minimum temperature of 1020 C
was therefore adopted.
Because the pre-eruptive dissolved water content of
the melt is not known, it is treated as a variable.
1904
KURITANI et al.
MAGMA MIXING TRENDS, MIYAKE-JIMA VOLCANO
The thermodynamic calculations suggest that the melt
coexists with plagioclase of An68 at 1180 C for 1 kbar
pressure and An67 at 1190 C for 2 kbar pressure, if the
melt was anhydrous (0 wt % H2O). This value is much
less than the An80---88 of the high-An plagioclase
phenocrysts (Fig. 7). As the water content increases,
the An content of plagioclase in equilibrium with the
melt increases at a given pressure condition (e.g. Sisson
& Grove, 1993). At the same time, however, equilibrium temperature also falls with increasing melt
water content. As a result of the constraint of the
minimum estimated temperature of 1020 C, the melt
can coexist with plagioclase of at most An78 at 1 kbar
and An77 at 2 kbar (water content of 3 wt %). Thus,
it is concluded that the A-group phenocrysts were not
formed from magmas with the observed whole-rock
composition. This is supported by the observation
that the chemical compositions of any mineral phases
of the A group are mostly constant throughout the
lavas (e.g. Fig. 7). If they crystallized from magmas
with the observed whole-rock composition, it is
expected that the mineral compositions should exhibit
systematic variations with whole-rock composition.
Furthermore, magnetite of the A group is zoned in
Usp number at the margin of crystals (Fig. 9), which
also supports the hypothesis that the A-group phenocrysts were not present in equilibrium in the 1983
magmas.
Another possible origin of the A-group phenocrysts is
that they are xenocrysts derived from the crust,
because other magmas erupted from Miyake-jima volcano rarely contain plagioclase and olivine xenocrysts,
which are believed to have come from the crust (e.g.
Miyasaka & Nakagawa, 1998). However, these xenocrysts are characterized by large crystal sizes (41 mm)
and deformation textures (Miyasaka & Nakagawa,
1998). Furthermore, single grains of the A-group phenocrysts are principally euhedral, which suggests that
they crystallized as isolated grains from liquids. This
negates the possibility that they are fragments of disaggregated crustal materials. From these relationships,
the A-group phenocrysts are considered to have been
inherited through magma mixing.
Contrary to plagioclase of the A group, the B-group
plagioclase might have been present in equilibrium in
the 1983 magmas, judging from the relatively low An
content (5An72; Fig. 7). However, magnetite crystals
of the B group are also zoned in terms of Usp number at
the crystal margin, suggesting that the B group phenocrysts could not have been present in equilibrium in
the 1983 magmas. The zoning pattern of the B-group
magnetite is the reverse of that of the A group; the Usp
number decreases from the core to rim in the B-group
magnetite, whereas the Usp number increases from the
core to rim in the A-group magnetite (Fig. 9). This
suggests that the 1983 magmas were produced by mixing of magmas containing A-group phenocrysts with
those including B-group phenocrysts.
Magma mixing
Although the abundance of the B-group phenocrysts is
much less than that of the A-group phenocrysts, they
are present in all lavas. The B-group phenocrysts are
probably derived from magmas slightly more evolved
than those of the A group, because they lack olivine,
and have low An content plagioclase and low Mg
number augite. Thus, it is plausible that the A-group
phenocrysts crystallized from basaltic magma and
the B-group phenocrysts crystallized from andesitic
magma. Judging from the abundances of the A- and
B-group phenocrysts, the basaltic magma was crystal
rich and the andesitic magma was crystal poor, with
less than 01 vol. % crystals. In the following section,
the details of the mixing process are considered using
whole-rock compositional data.
Mixing of two homogeneous magmas
If a homogeneous mafic magma mixes with a homogeneous felsic magma, a linear whole-rock composition
trend is produced (e.g. Venezky & Rutherford, 1997;
Clynne, 1999). This mechanism might be consistent
with the linearity of the main whole-rock composition
trends (Fig. 11a) and mostly linear decrease of phenocryst content with increasing whole-rock SiO2 content
(Fig. 11b). However, samples from the A, B, and G
craters deviate from the main trends for both the
whole-rock composition and phenocryst content
(Fig. 11), which cannot be explained solely by mixing
two homogeneous magmas.
If the main whole-rock composition trend formed
by mixing of two homogeneous magmas, the basaltic
end-member magma would have a composition on
the extrapolation of the composition trend (`possible
basaltic end-member magma', Fig. 11). The andesitic
end-member magma is plotted on the extrapolated
trend at 0% phenocryst content (Fig. 11). The Agroup phenocrysts are considered to have been present
in equilibrium in the basaltic end-member magma
before the two magmas mixed. There is no evidence of
dissolution or overgrowth of A-group plagioclase
before the growth of the sodic rim (Fig. 6a). The Mg
number of olivine phenocrysts is fairly homogeneous
throughout the crystals, and the Usp number in the
inner part of magnetite crystals is also homogeneous.
This suggests that equilibration between these phases
and the surrounding melt was attained by interdiffusion, even if they had been originally zoned. The
rounded outline of some olivine phenocrysts and zoning of Usp number at the margin of magnetite crystals
1905
JOURNAL OF PETROLOGY
VOLUME 44
NUMBER 10
OCTOBER 2003
Table 3: Equilibrium conditions of
olivine---plagioclase cotectic melt for various
compositions of the possible basaltic endmember magmas, for the case of the mixing
of two homogeneous magmas
Phenocryst
SiO2
Mg number
An content
Temperature
H2O
content
content
of olivine
of plagioclase
( C)
content
5
525
698
749
1072
20
6
512
703
766
1077
20
7
500
707
783
1081
19
8
488
712
800
1084
19
9
476
716
817
1087
19
(vol. %)
Fig. 11. (a) Al2O3---SiO2 variation diagram and (b) whole-rock SiO2
content---phenocryst content diagram, showing the formation of the
main whole-rock composition trend of the 1983 samples by mixing
between a homogeneous basaltic magma (`possible basaltic endmember magma') and a homogeneous andesitic magma (&). The
andesitic end-member magma is on the linear extrapolation of the
main trend at 0% of the phenocryst content (Fig. 11b). *, the
`possible basaltic end-member magmas' for various phenocryst contents referred to in Table 3.
can be due to magma mixing. The inference that the
A-group phenocrysts were present in equilibrium in
the basaltic end-member magma indicates that the
melt phase in the end-member magma was saturated
with both olivine and plagioclase (i.e. cotectic), and
therefore the liquidus temperatures of olivine and
plagioclase in the melt were identical. Using this constraint, the Mg number of olivine and An content of
plagioclase that can equilibrate with the melt phase in
the `possible basaltic end-member magma' are calculated, and they are compared with the observed compositions of olivine and plagioclase phenocrysts of the
A group.
The magma composition and phenocryst contents on
the possible trend were calculated using average values
for samples from the C---E craters and those for samples
from the H---I craters, because these have the least and
most differentiated compositions, respectively, on the
main composition trend (Fig. 3). The melt composition for the possible basaltic end-member magma was
obtained using the calculated phenocryst contents, the
proportion of phenocryst phases, and their chemical
compositions. The proportion of plagioclase, olivine,
and augite phenocrysts (97:1:2) is based on observation. The chemical compositions of the phenocrysts
were represented by the average values of many zoning
profiles for individual mineral phases. These are: An84
for plagioclase, Mg number 69 for olivine, and Mg
number 73 for augite. The thermodynamic models for
both plagioclase---melt and olivine---melt pairs were
applied to the calculated melt phase in the `possible
basaltic end-member magma'. At a given melt composition, H2O content is still an unknown variable.
Therefore, the H2O content was varied and determined so that the calculated liquidus temperature of
olivine equals that of plagioclase. In this way, the Mg
number of olivine, An content of plagioclase, crystallization temperature, and water content in melt can be
determined uniquely for the given melt composition
and the pressure condition. The thermodynamic solution model for olivine of Hirschmann (1991) was used,
in addition to the model for plagioclase of Elkins &
Grove (1990) and for silicate melt of Ghiorso & Sack
(1995). A pressure of 2 kbar was assumed for the crystallization of the A-group phenocrysts (see below).
Calculated values for the An content of plagioclase,
Mg number of olivine, crystallization temperature,
and melt H2O content are listed in Table 3 for some
`possible basaltic end-member magma' compositions
with different phenocryst contents shown in Fig. 11.
Even when the SiO2 content of the magma is 50 wt %,
the An content of equilibrium plagioclase is 78. The
An content of plagioclase tends to increase as the composition of the `possible basaltic end-member magma'
plots further away from those of the erupted 1983
products. However, the Mg number of coexisting
olivine also increases and becomes significantly higher
1906
KURITANI et al.
MAGMA MIXING TRENDS, MIYAKE-JIMA VOLCANO
than the observed Mg number of 68---70. Because of the
constraint of Mg number570 of the observed olivine
phenocrysts, the A-group phenocrysts cannot have
been present in any `possible basaltic end-member
magmas', and thus such magmas were not the basaltic
end-member magma that produced the linear wholerock composition trend of the 1983 products through
magma mixing. This conclusion is strongly supported
by a TiO2---SiO2 variation diagram (Fig. 2). There are
no samples on the extrapolation of the main compositional trend of the 1983 products (shown with dashed
line). This indicates that such SiO2-poor and TiO2rich magmas cannot have been produced beneath
Miyake-jima volcano. From these relationships, the
main whole-rock composition trend is unlikely to
have been established by mixing of two homogeneous
magmas.
Mixing of heterogeneous and homogeneous magmas
Another possibility to form a linear whole-rock composition trend is mixing of homogeneous and compositionally diverse magmas (Kuritani, 2001). There is also
a possibility that both magmas were heterogeneous.
However, the basaltic component of the 1983 magmas
is suggested to have been homogeneous, because of the
limited compositional diversity of the A-group phenocrysts (e.g. Fig. 7).
The andesitic end-member magma can be shown as
a line placed at 0 vol. % phenocrysts in Fig. 12b. In
this case, it is possible to interpret that the deviation
of the phenocryst contents of the A, B, and G crater
samples from the main trend resulted from a lower
proportion of the basaltic component in these samples
than in the other main samples at given whole-rock
SiO2 contents (Fig. 12b). Separation of the A-group
phenocrysts from the magmas on the main compositional trends might also produce the A, B, and G crater
magmas. This is unlikely to have occurred, however,
because the density difference between the A-group
phenocrysts and the melt is fairly small (5100 kg/m3 )
and the time scale from magma mixing to eruption was
short (1 day), as shown below. Because the andesitic
end-member magma was mostly free of crystals, the
fraction of the basaltic component in the 1983 samples
is roughly proportional to their phenocryst contents.
Using the relation between the phenocryst contents
and the whole-rock compositions, in addition to the
information that the andesitic end-member magma
is mostly free of crystals, the andesitic end-member
magma can be expressed as a composition trend on
the Harker diagram (Fig. 12a).
In the previous section, the hypothesis that the main
whole-rock composition trend was formed by mixing of
two homogeneous magmas was rejected on the basis
Fig. 12. (a) Al2O3---SiO2 variation diagram and (b) whole-rock
SiO2 content---phenocryst content diagram, showing the formation
of the whole-rock compositional variations of the 1983 samples by
mixing between a homogeneous basaltic magma (*) and heterogeneous andesitic magmas (bold line). The linear composition trend
of the mixed magma is produced by shift of the trend of the andesitic
end-member magma to the direction of the basaltic end-member
magma. In (a), contours of the phenocryst contents estimated from
(b) are displayed; , whole-rock compositions of samples of the 1962
eruption (Yokoyama et al., 2003).
that the `possible basaltic end-member magma' cannot
equilibrate with the A-group plagioclase. This results
from the gentle slopes of the whole-rock Al2O3---SiO2
and CaO---SiO2 composition trends; the magmas on
the extrapolation of the observed composition trends
cannot have Al2O3 and CaO contents high enough to
crystallize highly calcic plagioclase. In the hypothesis
of mixing between the homogeneous and heterogeneous magmas, on the other hand, the basaltic endmember magma can have remarkably high Al2O3 and
CaO contents (Fig. 12a). In the following section, we
show that it is plausible that the A-group phenocrysts
crystallized from a possible basaltic end-member
magma predicted by this hypothesis.
Characterization of end-member magmas
Before discussing the mechanism of magma mixing, the
composition of the basaltic end-member magma is
1907
JOURNAL OF PETROLOGY
VOLUME 44
NUMBER 10
OCTOBER 2003
constrained and the physical properties of the two
end-member magmas are estimated. In the following
discussion, we refer to the `basaltic end-member' as
`BEM' and to the `andesitic end-member' as `AEM'.
It should be noted always that the BEM magma was
principally homogeneous. On the other hand, the
AEM magma was heterogeneous and formed tight
whole-rock composition trends.
Basaltic end-member magma
Estimation of the composition of the BEM magma is
not straightforward, because of the heterogeneity of the
AEM magma. However, the whole-rock compositions
of the samples from the A and B craters diverge from
the compositions of samples from the C---E craters
(Fig. 3), and the locations of the A and B craters are
close to those of the C---E craters (Fig. 1b). These
observations suggest that the magmas from the A and
B craters were closely related to those from the C---E
craters. It is therefore possible that the compositions of
the AEM magma mixed in both the A and B and the
C---E magmas were similar. In this case, the `possible
basaltic end-member magma' composition can be
shown as the area with light gray pattern in Fig. 13.
The plausible composition of the BEM magma
among the compositions of the light gray area in
Fig. 13 is constrained using the observation that it
must be in equilibrium with olivine and plagioclase of
the A group. The thermodynamic models of both the
olivine---melt and plagioclase---melt pairs are applied to
the melt phase of the `possible basaltic end-member
magma', as in the previous section. The phenocryst
content and whole-rock composition of the `possible
magma' were calculated using those of the C---E crater
samples and those of the A crater sample (myk0107;
Table 1). Because the measured phenocryst contents
show some variation in each sample (Fig. 5), calculations were performed in the range between 25 and 31
vol. % for the phenocryst content of sample myk0107
(28 03 vol. %). On the other hand, the phenocryst
contents of the C---E samples were expressed as a function of their whole-rock SiO2 content using a linear
regression. The methods for determining the crystallization temperature, melt H2O content, equilibrium
Mg number of olivine and An content of plagioclase
are similar to those described above.
Figure 13 displays an example of the result of calculations (phenocryst contents of myk0107: 28 vol. %),
showing the calculated Mg number of olivine, An content of plagioclase, and the temperature for the `possible basaltic end-member magma'. The plausible BEM
composition of this case is shown as the region with
dark gray pattern, based on the constraints of the
observed olivine composition of Mg number 68---70,
Fig. 13. Al2O3---SiO2 variation diagram, showing the `possible basaltic end-member magma' (light gray pattern) and the plausible composition of the basaltic end-member magma (dark gray pattern with
bold outline). This example is the result of calculations using the
phenocryst content of 28 vol. % for sample myk0107. The compositional area of the `possible basaltic end-member magma' is determined based on the inference that the compositions of the AEM
magma mixed in both the C---E magmas and the A---B magmas
were similar. The An content of plagioclase and the Mg number of
olivine in equilibrium with the melt phase in the `possible basaltic
end-member magma' are shown with bold and fine continuous lines,
respectively. The equilibrium temperature is also shown with fine
dashed lines. The area of the plausible composition (dark gray pattern with bold outline) is determined using the constraints that the
magmas can crystallize An80---88 plagioclase and olivine with Mg
number 68---70 at temperature higher than 1020 C.
plagioclase composition of An80---88, and the estimated
minimum temperature of 1020 C. The area of the
plausible BEM compositions considering the variation
of the phenocryst content is shown in some variation
diagrams, in which compositions of the representative
samples from Miyake-jima (Uto et al., 2001; Geshi et al.,
2002; Yokoyama et al., 2003) are plotted (Fig. 14). At a
given SiO2 content, the compositions of the plausible
BEM area are significantly lower in MgO and higher
in Al2O3 than those of most eruptive products from
Miyake-jima volcano. However, the samples of the
August 2000 eruption (SiO2 51 wt %) have relatively similar compositions to the plausible BEM composition. Considering that no magmas with such high
Al2O3 content have erupted in stage (IV) except for
the August 2000 magmas (Fig. 14), the BEM magma
might be genetically similar to the August 2000 magmas. This hypothesis is strongly supported by petrographic and mineralogical features of the August 2000
samples: they are porphyritic with 14---18 vol. %
phenocrysts, and contain olivine phenocrysts with Mg
number 68---75, plagioclase phenocrysts with An80---90,
and minor augite and titanomagnetite (Geshi et al.,
2002). A porphyritic nature is common to the BEM
magma (Fig. 12b), and mineralogical features are also
1908
KURITANI et al.
MAGMA MIXING TRENDS, MIYAKE-JIMA VOLCANO
Fig. 14. Whole-rock variation diagrams for some major oxides (Al2O3, MgO, CaO and TiO2) for samples from all eruption stages in Miyakejima volcano, along with the area of the plausible BEM composition. The data source and the legend are the same as those in Fig. 2. Large
open circle indicates the most plausible BEM composition (Table 4).
similar to those of the A-group phenocryst, although
Mg number of olivine phenocryst in the August 2000
samples is slightly higher. Because MgO content of the
plausible BEM magmas is lower than that of the
August 2000 samples, the BEM magma might have
been derivative from magmas with similar compositions to the August 2000 magmas. The most plausible
BEM composition is, therefore, assumed to be at the
point shown by the large open circle in Fig. 14
(Table 4), which can be roughly produced by separation of 7 wt % of plagioclase and 3 wt % of olivine from
the August 2000 magmas.
Unfortunately, trace elements and isotopes cannot
be used to constrain the composition of the BEM
magma, because there are no significant differences of
the whole-rock trace element concentrations and isotopic compositions between the main samples and the
samples from the A, B, and G craters (Fig. 4b, c and d).
If significant differences were present, trace element
concentrations of the A-group phenocrysts would
strongly constrain the composition of the BEM
magma, by utilizing the partition coefficients between
minerals and melt. In addition, isotopic composition
of glass inclusions in the high-An plagioclase phenocrysts might provide direct information on the BEM
composition.
Andesitic end-member magma
Using the estimated composition of the BEM magma,
the minimum variation of the AEM magma can be
calculated. The least differentiated AEM magma has a
composition along the extrapolated line from the composition of the BEM magma to the least differentiated
samples in the 1983 products (myk3019; Fig. 12;
Table 4). Similarly, the composition of the most differentiated AEM magma can be obtained using the
composition of the BEM magma and that of the most
differentiated samples in the 1983 products (myk0138;
Fig. 12; Table 4). The estimates of the compositions of
these AEM magmas are not sensitive to the uncertainty
of the estimated BEM magma composition.
The temperature and water content of the AEM
magma cannot be estimated easily because the An
content of plagioclase in equilibrium with the AEM
magma is not clear, as a result of the scarcity of the
B-group phenocrysts. Therefore, the thermodynamic
models of plagioclase---melt pairs cannot be directly
applied. For this reason, the temperature and water
content conditions are constrained using the observation that the AEM magma was saturated with plagioclase, but not olivine. The conditions are estimated for
the least and most differentiated AEM magmas, using
their compositions ( melt compositions), a pressure of
1909
JOURNAL OF PETROLOGY
VOLUME 44
NUMBER 10
OCTOBER 2003
Table 4: Calculated compositions and estimated physical properties of the least
and most differentiated andesitic end-member magmas and possible basaltic
end-member magma
Erupted magma
Andesitic end-member
least differentiated
Basaltic end-member
myk3019
myk0138
most differentiated
magma
(melt)
52.60
1.36
54.71
1.29
53.17
1.45
55.52
1.31
51.48
1.20
52.23
1.41
15.01
14.25
14.84
13.18
13.76
15.08
14.17
13.32
17.52
12.61
14.97
14.57
0.24
3.98
9.07
0.23
0.26
4.25
8.44
0.24
0.21
3.64
8.35
3.69
7.85
3.45
10.35
0.24
3.92
9.21
K2O
2.77
0.55
2.98
0.62
2.84
0.60
3.07
0.66
2.62
0.45
2.78
0.48
P2O5
0.15
0.16
0.16
0.17
0.12
0.14
Major element compositions (wt %)
SiO2
TiO2
Al2O3
Fe2O3
MnO
MgO
CaO
Na2O
Physical properties
Phenocryst contents (vol. %)
5
3
0
0
15
ÐÐ
Fraction of basaltic component (%)
33
20
0
0
100
ÐÐ
Water content (wt %)
ÐÐ
ÐÐ
1.5
52.0
Temperature ( C)
ÐÐ
ÐÐ
1080
41060
2.5
2.9
1050
ÐÐ
Density (kg/m3 )
ÐÐ
ÐÐ
2670
2610
2630
2620
Viscosity (poise)
ÐÐ
ÐÐ
1000
1600
1000
500
Total Fe as Fe2O3. Compositions are recalculated for the total weight to be 100%. The sampling locality of myk3019 is
indicated as No. 13 in Fig. 1b.
1 kbar (estimated below), and thermodynamic models
of olivine---melt and plagioclase---melt pairs. Figure 15
shows calculated liquidus temperatures of olivine and
plagioclase, along with the An content of the liquidus
plagioclase, for the least and most differentiated AEM
magmas as a function of the melt H2O content. If the
water content of the least and most differentiated AEM
magmas exceeds 16 wt % and 20 wt %, respectively,
the liquidus temperature of olivine exceeds that of
plagioclase, and olivine would appear as a crystallizing
phase, contrary to observation. Thus the temperature
and water content conditions of the least differentiated
AEM magma are estimated to be 41075 C and 516
wt %, and those of the most differentiated AEM
magma 41060 C and 520 wt %, respectively.
Sample myk0146 (the least differentiated samples
from the J and K craters) contains B-group plagioclase
with An72. This suggests that the AEM magma mixed
in the myk0146 magma can equilibrate with An472
plagioclase. Because this magma is slightly evolved
compared with the least differentiated AEM magma,
it is plausible to consider that the least differentiated
AEM magma can equilibrate with at least An72 plagioclase. The estimated temperature and water content
conditions for the least differentiated AEM magma
are, therefore, 1080 C and 15 wt %, respectively
(Fig. 15a).
Summary of the physical properties
The Al2O3 and CaO contents of the BEM magma are
much higher than those of the erupted magma
(Table 4). The MgO content of the BEM magma is
as low as 35 wt %, which is lower than that of the
AEM magma. The temperature of 1060---1080 C estimated for the AEM magma is higher than 1050 C
for the BEM magma. This mainly results from high
crystallinity of the BEM magma. The proportion of the
BEM component in the erupted 1983 magma is estimated to be 33% for the least differentiated samples
(myk3019) and 20% for the most differentiated samples (myk0138). Table 4 also lists the densities of the
end-member magmas calculated after Berman (1988)
and Lange & Carmichael (1990). The density of the
AEM magma was about 2600---2700 kg/m3 , which is
1910
KURITANI et al.
MAGMA MIXING TRENDS, MIYAKE-JIMA VOLCANO
calculated bulk viscosity is about 1000 poise because of
the high crystallinity.
Implications for magmatic processes of
end-member magmas
Basaltic end-member magma
Fig. 15. Liquidus temperatures of olivine and plagioclase, along
with the equilibrium An content of plagioclase, for (a) the least and
(b) the most differentiated andesitic end-member magmas, shown as
a function of the H2O content in melt. The temperature and the H2O
content are constrained by the observation that plagioclase was a
crystallizing phase whereas olivine was not stable in the magmas. For
the least differentiated AEM magma, the conditions are further
constrained by the presence of An72 plagioclase in the sample
myk0146.
similar to the value of 2630 kg/m3 calculated for the
BEM magma. The A-group plagioclase has density
about 2690 kg/m3 . The density of 2610 kg/m3 estimated for the most differentiated AEM magma is
mostly independent of the uncertainty of the estimations of the temperature and the melt H2O content.
This is because the effect of decreasing density by
increasing the estimated temperature is counterbalanced by the effect of increasing density by decreasing the estimated H2O content. Viscosities of the endmember magmas were calculated by the method of
Shaw (1972). The viscosity of the crystal-free AEM
magmas is about 1000---1600 poise. The estimated viscosity of the most differentiated AEM magma is again
independent of the uncertainty of the estimations of
the temperature and the H2O content. Viscosity of
crystal-bearing BEM magma was obtained by the
Einstein---Roscow equation (e.g. Marsh, 1981)
m ˆ mL …1ÿ167f†ÿ2:5
…1†
where mL is the viscosity of the interstitial melt, and f
is the crystallinity. Although the viscosity of the melt
phase of the BEM magma is low ( 500 poise), the
The BEM magma is characterized by high Al2O3 and
low MgO contents. Importantly, MgO content is estimated to have been about 35 wt %, which is significantly lower than that of the AEM magma (Table 4).
Therefore, the BEM magma was a derivative from a
primary magma probably through extensive olivine
fractionation. The MgO content of the high-An type
plagioclase increases with decreasing An content
(Fig. 8). In common magmatic differentiation processes, the MgO content of the melt phase as well as
the equilibrium An content of plagioclase tend to
decrease with falling magmatic temperature. The
negative correlation between the MgO content and
the An content can be explained by the dependence
of the partition coefficient of MgO between plagioclase
and silicate melt on the crystal chemistry. Magnesium
has been suggested to be more compatible in less Anrich compositions in plagioclase (Blundy & Wood,
1994). The depletion of the melt MgO content during
differentiation was, therefore, not so dominant as to
overcome the effect of the dependence of the partition
coefficient on crystal chemistry. This may suggest that
the extensive olivine fractionation occurred before
crystallization of the high-An type plagioclase phenocrysts. This is consistent with the Mg number of 68---70
for olivine inclusions in plagioclase phenocrysts, suggesting that crystallization of plagioclase became dominant after the melt phase was depleted in MgO. The
hypothesis of extensive olivine fractionation before
plagioclase saturation is supported by the hydrous
nature of magmas as suggested from the estimated
high water content of the BEM magma (Table 4).
Because of the presence of water in magmas, the
liquidus temperature of plagioclase is more suppressed
than that of olivine (Fig. 15, for example). In effect,
plagioclase crystallization initiates after extensive
olivine crystallization (and possibly its fractionation)
in hydrous magmas.
Andesitic end-member magma
The MgO contents of the low-An type plagioclase are
mostly constant or correlate positively with An content
(Fig. 8), contrary to the high-An type plagioclase. This
suggests that the effect of the depletion of MgO in the
melt from which plagioclase crystallized exceeds the
effect of the dependence of the partition coefficient
on crystal chemistry. Although the partition of FeO
between plagioclase and silicate melt depends strongly
1911
JOURNAL OF PETROLOGY
VOLUME 44
on oxygen fugacity (Sato, 1989; Phinney, 1992), progressive decrease of the iron content with decreasing
An content is also explained by extensive depletion of
FeO content in melt during the crystallization of
plagioclase. The positive correlations between the An
content of plagioclase and melt MgO and FeO
contents are consistent with the formation of the compositional variation of the AEM magma by fractional
crystallization. In fact, the composition trend of the
AEM magma is roughly explained by separation of
augite, plagioclase, and magnetite in the weight proportion of approximately 5:3:2. Because the density of
the most differentiated AEM magma is lower than that
of the least differentiated AEM magma (Table 4), the
AEM magma is likely to have exhibited density stratification in the magma chamber and the SiO2-poor
magma was present in the lower part of the magma
reservoir. Tsukui et al. (2001) showed that magmas
erupted in stage (IV) (AD 1469 to the present) tend to
have evolved in composition with time. This might
suggest that the compositional variation of the AEM
magma had been gradually established in the magma
chamber during the last 500 years.
The temperature of the AEM magma was about
1060---1080 C, as estimated above. On the other
hand, a crystal aggregate with An64 plagioclase in
sample myk3024 (Fig. 6d) is estimated to have formed
at about 920---950 C, based on the compositions
of augite and coexisting orthopyroxene crystals
(Lindsley, 1983). Such crystals are considered to have
been derived from low-temperature mush zones along
the wall of the AEM magma chamber, and were incorporated into the magma during the eruption.
The pressure condition of the magma chamber in
which the AEM magma evolved is estimated using
compositions of glass and plagioclase of the crystal
aggregate shown in Fig. 6d. Because the glass in the
crystal aggregate has fairly homogeneous composition
(SiO2 63---64 wt %; MgO 18---19 wt %; K2O 13---14
wt %) and is completely enclosed by crystals (Fig. 6d),
it is reasonable to consider that the glass composition
represents the composition of the interstitial melt of
the mush zones in the AEM magma chamber. The
observed An content of plagioclase crystals coexisting
with the melt is 63---64. The thermodynamic models of
plagioclase---melt pair equilibrium above are applied
with the constraint of the water solubility using
the model of Moore et al. (1998). Figure 16 displays
the equilibrium temperature and An content of plagioclase for the melt, as a function of the melt H2O content
and pressure. Because the melt is suggested to have
been in equilibrium with An63---64 plagioclase at about
920---950 C (estimated above using two pyroxenes),
the magma chamber was present at a depth corresponding to about 1 kbar pressure.
NUMBER 10
OCTOBER 2003
Fig. 16. The An content of plagioclase (continuous lines) in equilibrium with the interstitial melt of the crystal aggregate shown in
Fig. 6d, as a function of pressure and the melt H2O content. Equilibrium temperature is shown with dashed lines. The water solubility
curve (bold line) was calculated using the model of Moore et al.
(1998). Because the melt is estimated to have coexisted with
An63---64 plagioclase at temperatures of 920---950 C, the plausible
pressure conditions are shown by the area with the dark gray pattern.
Injection of the BEM magma into the AEM
magma chamber
Either the BEM magma or the AEM magma could
have resided in the magma chamber, into which the
other type of end-member magma was injected. In the
erupted 1983 magmas, the AEM component was
dominant, at 470%. In addition, the BEM magma
did not erupt, whereas AEM magma with little BEM
component erupted. On this basis, it is likely that
the AEM magma resided in the magma chamber at
1 kbar, and then the BEM magma ascended from
another magma chamber at deeper levels and was
injected into the 1 kbar magma chamber. As noted
above, earthquake foci associated with the 1983 eruption were distributed at depths from 2---3 km to 6---7 km
beneath Miyake-jima island (e.g. Miyazaki & Sawada,
1984). Therefore, the BEM magma chamber is considered to have been present at a depth of about 6---7 km
( 2 kbar).
The composition trend of the 1962 products, shown
with crosses in Fig. 12a, is close to the estimated composition trend of the AEM magma. In the 1962 products, crystal aggregates that have similar features to
the A-group phenocrysts, in terms of mineral assemblage and compositions, are observed (Miyasaka &
Nakagawa, 1998; Amma-Miyasaka & Nakagawa,
2002). These suggest that the 1962 magmas might
also have been produced by mixing between the AEM
and the BEM magmas in different proportions from the
case of the 1983 magmas. The phenocryst contents
of 1---2 vol. % for the 1962 products (Miyasaka &
Nakagawa, 1998) are in good agreement with the
1912
KURITANI et al.
MAGMA MIXING TRENDS, MIYAKE-JIMA VOLCANO
Fig. 17. Schematic illustration of the magma chamber, showing the inferred processes of mixing between the injected basaltic magma and the
zoned andesitic magma that occupied the reservoir. It should be noted that the mixed magma preserves the heterogeneity of the host andesite.
(See text for details.)
position of the composition trend of the 1962 magmas
(Fig. 12a). If this is correct, both the 1962 and 1983
eruptions have been controlled by the discharge of the
BEM magma from the magma reservoir at 2 kbar.
Time scale from magma mixing
to eruption
Titanomagnetite crystals in the 1983 products are
zoned in terms of ulv
ospinel component (Fig. 9). The
Usp number of the crystals in the A group increases
from the core to rim and that of the B group decreases
from the core to rim. This might be caused by interdiffusion, resulting from disequilibrium of titanomagnetite with the surrounding melt after magma
mixing. By utilizing the diffusion profiles, the time
scale from the magma mixing in the magma chamber
to the solidification after the eruption can be estimated
(Nakamura, 1995).
It is assumed that titanomagnetite crystals were
homogeneous in Usp number before magma mixing.
This is supported by the flat zoning pattern of titanomagnetite inclusions in plagioclase, olivine, and augite
phenocrysts, and the fact that relatively large titanomagnetite crystals isolated in the groundmass also have
a flat zoning profile around the core. The titanomagnetite inclusions have similar composition to the
core of the isolated titanomagnetite crystals, and therefore the homogeneous composition was used as an initial
condition. The crystals are approximated to be
spherical, and are present in semi-infinite melt. The
Usp number of titanomagnetite in equilibrium with
the melt after magma mixing is assumed to be 30 (boundary condition), because the outermost margin of the
crystals commonly has composition of Usp number
30. The interdiffusion coefficient of 3 10 ÿ11 cm2 /s
was used at 1080 C after Freer & Hauptman (1978).
The diffusion equation was numerically solved by the
Crank---Nicolson finite difference scheme.
An example of the calculated diffusion profiles is
shown in Fig. 9 for several time steps. The comparison
of the calculated and actual profiles suggests that
51 day is plausible for the period from the magma
mixing to the solidification of the host magma after
eruption in this case. Analyses of many zoning profiles
(430) reveal that the time scales are always 53 days
and mostly 51 day. It is plausible to consider that the
1983 magma erupted soon after the magma mixing in
the magma chamber, although the estimated values
give a minimum time scale because the boundary condition of Usp number 30 would be attained when the
crystals were in contact with the andesitic melt during
magma mixing.
Implications for dynamics of
magma mixing
Mixing of a homogeneous mafic magma with a heterogeneous felsic magma may occur frequently in magmatic systems, because the felsic magma may be
chemically zoned before injection of mafic magma.
To form linear compositional trends by this mechanism, however, some particular conditions should be
satisfied, as discussed below. Using the physical and
chemical constraints established above, the dynamics
of magma mixing beneath Miyake-jima volcano are
considered below.
The calculated density of the injected BEM magma
is comparable with that of the AEM magma (Table 4).
The replenished BEM magma could not have
ascended in the AEM magma solely by buoyancy,
and therefore the BEM magma is likely to have entered
the andesitic magma chamber with appreciable
upward momentum, possibly as a fountain (Fig. 17a).
This is strongly supported by the estimated short time
scale from magma mixing to eruption. To produce
the linear compositional trends, the replenished BEM
magma should have been distributed systematically
in the AEM magma without disturbing the overall
compositional stratification of the AEM magma.
That is to say, the mixing ratio of the BEM to AEM
magmas should change linearly with the composition
1913
JOURNAL OF PETROLOGY
VOLUME 44
of the original AEM magma. Otherwise the resulting
mixing products show `compositional area' rather
than `compositional trend' in whole-rock variation
diagrams.
The host AEM magma may have been entrained
turbulently by the BEM magma of the ascending
fountain, in which effective mixing between the two
magmas can occur. However, this mixing mechanism
results in homogenization of the compositions of the
mixed magma, contrary to observation. Thus the
observed characteristic mixing trends were not produced solely by this mechanism. Another location of
effective magma mixing may be within the resident
andesite magma. During ascent of the fountain of the
BEM magma, the BEM component partly lost its
upward momentum and spread laterally by entrainment by the host andesite magma (Fig. 17b). This
might have occurred at various levels of ascent of the
fountain until all the BEM component was entrained
by the AEM magma (Fig. 17c). The two magmas
could have been further blended during ascent in the
conduit, preserving the vertical chemical heterogeneity. Tapping of the magma chamber occurred after the
fountain reached the upper part of the magma chamber. If the magma chamber was tapped before the
BEM component was transported to the upper level,
the original AEM magma without mixing with the
BEM component would have erupted.
Mixing processes between ascending fountains and
host liquid have been investigated by the analogue
experiments of Campbell & Turner (1989), although
the density of the injected liquid was significantly
higher than that of the host liquid. They showed
that ascending fountains partly spread laterally and
efficiently mix with the host liquid. Effective
mixing occurs when the viscosity contrast between the
two liquids is not large (5101 ---102 ; Campbell &
Turner, 1986), as is the case for the 1983 magmas
(Table 4). During lateral spreading of the mixed
liquid, the main fountain still has upward momentum
and ascends to the upper level, and then spreads
laterally. These processes are consistent with those
discussed above.
Shift and rotation of compositional trends of the
resident magmas by mixing with replenished magmas
may occur commonly if the viscosity contrast between
the two magmas is small. In the magma chamber
beneath Miyake-jima volcano, the less differentiated
magma tends to shift more than the more differentiated
magma (Fig. 12a). The manner of the shift is likely to
be controlled by the zonation style of the resident
magmas and the style of the injection of new magma,
in addition to the physical properties of the two magmas, which can be variable in individual magmatic
systems.
NUMBER 10
OCTOBER 2003
CONCLUSION
The magmatic processes involved in the production of
the 1983 eruptive products from Miyake-jima volcano
were investigated on the basis of petrology and detailed
thermodynamic analysis. Euhedral plagioclase phenocrysts in the 1983 products are too calcic to have
crystallized from the erupted magma compositions,
and thus were derived from a basaltic magma through
magma mixing. Although the whole-rock compositions
show linear trends except for some specific samples,
they cannot have been produced by simple twocomponent magma mixing. The trends were formed
by mixing of a homogeneous basaltic magma with
heterogeneous andesitic magmas exhibiting linear
composition trends. The original composition trends
of the andesitic magma were rotated and shifted to
the direction of the basaltic magma through magma
mixing.
The density and viscosity of the replenished basaltic
magma are estimated to have been mostly comparable
with those of the resident andesitic magma. The basaltic magma ascended from a magma chamber, located
at 2 kbar, and was then injected into a magma chamber at 1 kbar, possibly as a fountain. To establish the
characteristic mixing trends of whole-rock compositions, the mixing ratio of the basaltic to andesitic magmas should change linearly with the composition of the
zoned andesite magma. From this, the basaltic component is likely to have spread laterally and mixed with
the andesite magma at various levels of ascent of the
fountain.
Petrologic evidence for magma mixing, such as the
presence of disequilibrium phenocryst assemblages and
the linear composition trends of the erupted materials,
is commonly considered to suggest mixing between two
homogeneous end-component magmas. However, this
study demonstrates that the linear composition trends
could have been established by a shift of the original
trend through magma mixing. In this case, the endcomponent magmas are not always on the extrapolation of the observed linear mixing trends. The `mixing
trend' should be examined carefully so as not to
misunderstand magmatic processes.
ACKNOWLEDGEMENTS
We are grateful to Kazuhito Ozawa for critical review
of the manuscript, valuable discussions, and encouragement throughout this study. We thank Ryoji
Tanaka and all the other members of the Pheasant
Memorial Laboratory at ISEI for useful discussions.
We acknowledge H. Nagahara and M. Nakamura for
constructive discussions. H. Asada, N. Takeuchi, and
M. Tanaka are also thanked for technical assistance.
1914
KURITANI et al.
MAGMA MIXING TRENDS, MIYAKE-JIMA VOLCANO
We thank M. Amma-Miyasaka and M. Tsukui for
helpful suggestions on Miyake-jima volcano. Constructive reviews and comments by T. Feeley and
Y. Tamura significantly improved the manuscript.
R. Arculus is also thanked for editorial handling and
encouragement. This work was supported by the
Japanese Society for the Promotion of Science for
Japan Junior Scientists (T.K. and T.Y.) and the
Ministry of Education, Culture, Sports, Science and
Technology (K.K. and E.N.).
REFERENCES
Amma-Miyasaka, M. & Nakagawa, M. (2002). Origin of anorthite
and olivine megacrysts in island-arc tholeiites: petrological study
of 1940 and 1962 ejecta from Miyake-jima volcano, Izu---Mariana
arc. Journal of Volcanology and Geothermal Research 117, 263---283.
Aramaki, S. & Hayakawa, Y. (1984). Sequence and mode of
eruption of the October 3---4, 1983 eruption of Miyakejima.
Bulletin of Volcanological Society of Japan 29, S24---S35 (in Japanese).
Bacon, C. R. (1986). Magmatic inclusions in silicic and intermediate
volcanic rocks. Journal of Geophysical Research 91, 6091---6112.
Berman, R. G. (1988). Internally-consistent thermodynamic data
for minerals in the system Na2O---K2O---CaO---MgO---FeO--Fe2O3---Al2O3---SiO2---TiO2---H2O---CO2. Journal of Petrology 29,
445---522.
Blundy, J. D. & Wood, B. J. (1994). Prediction of crystal---melt
partition coefficients from elastic moduli. Nature 372, 452---454.
Campbell, I. H. & Turner, J. S. (1986). The influence of viscosity on
fountains in magma chambers. Journal of Petrology 27, 1---30.
Campbell, I. H. & Turner, J. S. (1989). Fountains in magma
chambers. Journal of Petrology 30, 885---923.
Clynne, M. A. (1999). A complex magma mixing origin for rocks
erupted in 1915, Lassen Peak, California. Journal of Petrology 40,
105---132.
Elkins, L. T. & Grove, T. L. (1990). Ternary feldspar experiments
and thermodynamic models. American Mineralogist 75, 544---559.
Feeley, T. C. & Dungan, M. A. (1996). Compositional and dynamic
controls on mafic---silicic magma interactions at continental arc
volcanoes: evidence from Cord
on El Guadal, Tatara---San Pedro
Complex, Chile. Journal of Petrology 37, 1547---1577.
Freer, R. & Hauptman, Z. (1978). An experimental study of
magnetite---titanomagnetite interdiffusion. Physics of the Earth and
Planetary Interiors 16, 223---231.
Fujii, T., Aramaki, S., Fukuoka, T. & Chiba, T. (1984). Petrology
of the ejecta and lavas of the 1983 eruption of Miyake-jima.
Bulletin of Volcanological Society of Japan 29, S266---S282 (in Japanese).
Geshi, N., Shimano, T., Nagai, M. & Nakada, S. (2002). Magma
plumbing system of the 2000 eruption on Miyakejima volcano,
Japan. Bulletin of Volcanological Society of Japan 47, 419---434
(in Japanese).
Ghiorso, M. S. & Sack, R. O. (1995). Chemical mass transfer in
magmatic processes IV. A revised and internally consistent
thermodynamic model for the interpolation and extrapolation
of liquid---solid equilibria in magmatic systems at elevated
temperatures and pressures. Contributions to Mineralogy and Petrology
119, 197---212.
Hirschmann, M. (1991). Thermodynamics of multicomponent
olivines and the solution properties of (Ni,Mg,Fe)2SiO4 and
(Ca,Mg,Fe)2SiO4 olivines. American Mineralogist 76, 1232---1248.
Huppert, H. E. & Sparks, R. S. J. (1980). The fluid dynamics of a
basaltic magma chamber replenished by influx of hot, dense
ultrabasic magma. Contributions to Mineralogy and Petrology 75,
279---289.
Kuritani, T. (2001). Replenishment of a mafic magma in a zoned
felsic magma chamber beneath Rishiri Volcano, Japan. Bulletin of
Volcanology 62, 533---548.
Kuritani, T. & Nakamura, E. (2002). Precise isotope analysis of
nanogram-level Pb for natural rock samples without use of double
spikes. Chemical Geology 186, 31---43.
Lange, R. L. & Carmichael, I. S. E. (1990). Thermodynamic
properties of silicate liquids with emphasis on density, thermal
expansion and compressibility. In: Nicholls, J. & Russell, J. K.
(eds) Modern Methods of Igneous Petrology: Understanding Magmatic
Processes. Mineralogical Society of America, Reviews in Mineralogy 24,
25---64.
Lindsley, D. H. (1983). Pyroxene thermometry. American
Mineralogist 68, 477---493.
Makishima, A. & Nakamura, E. (1991). Precise measurement of
cerium isotope composition in rock samples. Chemical Geology 94,
1---11.
Makishima, A. & Nakamura, E. (1997). Suppression of matrix
effects in ICP-MS by high power operation of ICP: application to
precise determination of Rb, Sr, Y, Cs, Ba, REE, Pb, Th and U at
ng gÿ1 level in a few milligram silicate samples. Geostandards
Newsletter 21, 307---319.
Makishima, A., Nakamura, E. & Nakano, T. (1997). Determination
of boron in silicate samples by direct aspiration of sample HF
solutions into ICPMS. Analytical Chemistry 69, 3754---3759.
Makishima, A., Nakamura, E. & Nakano, T. (1999). Determination
of zirconium, niobium, hafnium and tantalum at ng gÿ1 levels in
geological materials by direct nebulization of sample HF solution
into FI-ICP-MS. Geostandards Newsletter 23, 7---20.
Marsh, B. D. (1981). On the crystallinity, probability of occurrence,
and rheology of lava and magma. Contributions to Mineralogy and
Petrology 78, 85---98.
Miyasaka, M. & Nakagawa, M. (1998). Recent magma
plumbing system beneath Miyake-jima volcano, Izu islands,
inferred from petrological study of the 1940 and 1962
ejecta. Bulletin of Volcanological Society of Japan 43, 433---455
(in Japanese).
Miyazaki, T. & Sawada, M. (1984). Seismic activity related to the
eruption of Miyakejima volcano, 1983. Bulletin of Volcanological
Society of Japan 29, S55---S67 (in Japanese).
Moore, G., Vennemann, T. & Carmichael, I. S. E. (1998). An
empirical model for the solubility of H2O in magmas to 3 kilobars.
American Mineralogist 83, 36---42.
Nakamura, M. (1995). Continuous mixing of crystal mush and
replenished magma in the ongoing Unzen eruption. Geology 23,
807---810.
Nakano, T. & Nakamura, E. (1998). Static multicollection of
Cs2BO‡
2 ions for precise boron isotope analysis with positive
thermal ionization mass spectrometry. International Journal of Mass
Spectrometry and Ion Processes 176, 13---21.
Phinney, W. C. (1992). Partition coefficients for iron between
plagioclase and basalt as a function of oxygen fugacity:
implications for Archean and lunar anorthosites. Geochimica et
Cosmochimica Acta 56, 1885---1895.
Sato, H. (1989). Mg---Fe partitioning between plagioclase and liquid
in basalts of Hole 504B, ODP Leg 111: a study of melting at 1
atm. In: Becker, K., Sakai, H. et al. (eds) Proceedings of the Ocean
Drilling Program, Scientific Results, 111. College Station, TX: Ocean
Drilling Program, pp. 17---26.
1915
JOURNAL OF PETROLOGY
VOLUME 44
Shaw, H. R. (1972). Viscosities of magmatic silicate liquids: an
empirical method of prediction. American Journal of Science 272,
870---893.
Sisson, T. W. & Grove, T. L. (1993). Experimental investigations of
the role of H2O in calc-alkaline differentiation and subduction
zone magmatism. Contributions to Mineralogy and Petrology 113,
143---166.
Snyder, D. & Tait, S. (1996). Magma mixing by convective
entrainment. Nature 379, 529---531.
Soya, T., Uto, K., Makimoto, H., Kamata, H., Okumura, K. &
Suto, S. (1984). Bulk and mineral chemistry of lavas and ejecta of
the 1983 eruption of Miyakejima Volcano. Bulletin of Volcanological
Society of Japan 29, S283---S296 (in Japanese).
Sparks, R. S. J. & Marshall, L. A. (1986). Thermal and mechanical
constraints on mixing between mafic and silicic magmas. Journal
of Volcanology and Geothermal Research 29, 99---124.
Sun, S.-S. & McDonough, W. F. (1989). Chemical and isotopic
systematics of oceanic basalts: implications for mantle composition and processes. In: Saunders, A. D. & Norry, M. J. (eds)
Magmatism in the Ocean Basins. Geological Society, London, Special
Publications 42, 313---345.
Takei, H. (2002). Development of precise analytical
techniques for major and trace element concentrations in
rock samples and their applications to the Hishikari Gold
Mine, southern Kyushu, Japan. Ph.D. thesis, Graduate
School of Natural Science and Technology, Okayama
University.
Todt, W., Cliff, R. A., Hanser, A. & Hofmann, A. W. (1996).
Evaluation of a 202 Pb---205 Pb double spike for high-precision lead
isotope analysis. Geophysical Monograph, American Geophysical Union
95, 429---437.
NUMBER 10
OCTOBER 2003
Tsukui, M. & Suzuki, Y. (1998). Eruptive history of Miyakejima
Volcano during the last 7000 years. Bulletin of Volcanological Society
of Japan 43, 149---166 (in Japanese).
Tsukui, M., Niihori, K., Kawanabe, Y. & Suzuki, Y. (2001).
Stratigraphy and formation of Miyakejima Volcano. Journal of
Geography 110, 156---167 (in Japanese).
Uto, K., Kazahaya, K., Saito, G., Itoh, J., Takada, A.,
Kawanabe, Y., Hoshizumi, H., Yamamoto, T., Miyagi, I.,
Tomiya, A., Satoh, H., Hamazaki, S. & Shinohara, H. (2001).
Magma ascending model of 2000 Miyakejima eruptions: evidence
from pyroclastics of August 18 and SO2-rich volcanic gas. Journal
of Geography 110, 257---270 (in Japanese).
Venezky, D. Y. & Rutherford, M. J. (1997). Preeruption conditions
and timing of dacite---andesite magma mixing in the 22 ka
eruption at Mount Rainier. Journal of Geophysical Research 102,
20069---20086.
Yokoyama, T. & Nakamura, E. (2002). Precise determination of
ferrous iron in silicate rocks. Geochimica et Cosmochimica Acta 66,
1085---1093.
Yokoyama, T., Makishima, A. & Nakamura, E. (1999). Evaluation
of the coprecipitation of incompatible trace elements with fluoride
during silicate rock dissolution by acid digestion. Chemical Geology
157, 175---187.
Yokoyama, T., Kobayashi, K., Kuritani, T. & Nakamura, E.
(2003). Mantle metasomatism and rapid ascent of slab components beneath island arcs: evidence from 238 U---230 Th---226 Ra
disequilibria of Miyakejima volcano, Izu arc, Japan. Journal of
Geophysical Research (in press).
Yoshikawa, M. & Nakamura, E. (1993). Precise isotope determination of trace amounts of Sr in magnesium-rich samples. Journal of
Mineralogy, Petrology, and Economic Geology 88, 548---561.
1916