Está en la página 1de 30

Journal of Southeast Asian Earth Sciences, Vol. I I, No. 2, pp.

135-164, 1995
Copyright 0 1995 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
0743-9w7/95 $9.50 + 0.00

Pergamon

Geology, mineralogy and magma evolution of Gunung


Slamet Volcano, Java, Indonesia
Danilo Vukadinovic*t

and Igan Sutawidjajat

*Department of Earth Sciences, Monash University,


tVolcanologica1 Survey of Indonesia, Jl. Diponegoro

Clayton, Victoria 3168 Australia, and


57, Bandung, Central Java, Indonesia

AhstracG-Gunung
Slamet, Central Java, is a large stratovolcano within the Sunda magmatic arc of
Indonesia. The volcanic edifice includes products of two large overlapping Quaternary stratocones.
Basaltic andesites and andesites, with rare basalts, dominate the western region of the comple%,
known as Slamet Tua (old); and basalts and basaltic andesites compose the eastern cone, called
Slamet Muda (young).
On the basis of stratigraphy, trace-element content, Zr/Nb, Zr/K and *Sr/*Sr ratios, Slamet lavqs
can be broadly categorized as relating to high abundance magma (HAM) and low abundance magma
(LAM) types. The Tua and Lebaksiu sequences generally comprise the LAM group, and are older,
more salic and have higher *Sr/%r ratios than those of HAM. LAM andesites contain some
amphibole, but HAM andesites do not.
Models involving randomized magma replenishment, tapping and fractionation (RTF) wetie
developed to explain the geochemical features of both LAM and HAM rock groups. The salic lavas
of the LAM suite can be generated if fractionation was dominant relative to replenishment ati
tapping in LAM magma chambers. Conversely, magma chambers with a high proportion of
replenishment and tapping relative to fractionation can produce dominantly mafic lavas, such as

those of the HAM suite.


Concave-upward heavy-rare-earth element (HREE) patterns for LAM andesites are probably due
to significant amphibole fractionation; HAM andesites display flat HREE patterns and do not requite
amphibole fractionation from parental basalts. The high TiO, contents of HAM basalts and basaltic
andesites (relative to those of average arc rocks) are due to either suppressed crystallization-or
minor accumulation-of Ti-magnetite, in conjunction with RTF processes.

Introduction
Gunung Slamet volcano, Central Java, lies about 190 km
above the northward-dipping
seismic Benioff zone
(Hamilton, 1979) and rests upon Neogene sediments of
dominantly shallow marine regressive facies (Djuri,
1975) above a 20-25 km thick crust (Hamilton, 1979).
Compared with most other arc volcanoes, Gunung
Slamet has erupted significant amounts of mafic lava (cf.
Whitford, 1975a), allowing for detailed studies on the
origin of arc magmas (Vukadinovic, 1989; Vukadinovic
and Nicholls, 1989).
Neumann van Padang (1951) presented the earliest
major-element analyses of Slamet lavas. The reconnaissance study of Whitford (1975a) contains modern analyses of Slamet rocks, i.e. basalts and basaltic andesites
(SiOZ c 56 wt%), some of which have TiOz in excess of
1.8 wt%. Whitford (1975a, b) classified Slamet as an
anomalous calcalkaline volcan-anomalous
in the
sense that 87Sr/86Sr ratios are higher in Slamet lavas
relative to those of the majority of calcalkaline lavas
from Java. He also suggested that the large abundance
of high-field-strength (HFS) elements in Slamet lavas,
relative to typical arc rocks, may be due to incorporation
of a subducted oceanic island with its underlying litho-

t Present address: Geology Department, Brandon University,


Brandon, Manitoba, Canada R7A 6A9.

sphere. Subsequent studies (Pardyanto, 1971; Aswin


et al., 1984; Sutawidjaja et al., 1985; Vukaflinovic, 1989)
discussed older andesitic rocks associated 1with Gunung
Slamet. On the basis of stratigraphy and igeochemistry,
Vukadinovic (1989) and Vukadinovic and Nicholls
(1989) broadly categorized Slamet lavas #as relating to
high-abundance magma (HAM) and low-abundance
magma (LAM) types. Compared with HAM lavas,
LAM are older and more salic and have higher 87Sr/*6Sr.
A model was developed showing that: (1) dompared with
parental HAM, parental LAM are generiCted by higher
degrees of melting within the mantle wedhe; and (2) the
degree of melting is controlled by the adount of fluids
introduced by the dehydrating, subducti g lithosphere
(Vukadinovic, 1989; Vukadinovic and N?cholls, 1989).

Gunung Slamet
Purwokerto, the largest town near Gunlung Slamet, is
located about 25 km south of the volcan4 summit. The
highest villages are about 1500 m above s$a level; above
this height, only odd footpaths through d&se vegetation
exist. Below 1500 m, the density of rdads increases
rapidly with decreasing height, providing e)xcellent access
around the base of the volcano (where, in any case, the
exposure-in
streams-is best).
The topographic
maps (1: 50,000 scale; edition
2-AMS) used during the field expeditions1 for this study

135

136

D. Vukadinovic and I. Sutawidjaja

were prepared under the direction of CINCUSARPAC


by the U.S. Army Map Service, Far East, and the 29th
Engineer Battalion. With Java so densely populated, yet
overwhelmingly rural in character, small villages abound
throughout the island. As a result, the names of some
villages are occasionally mentioned in the text that are
not shown on Fig. 1, due to a desire for clarity and
brevity. In these instances the reader is referred to the
CINCUSARPAC maps.
The geology of Gunung Slamet (Fig. 1) has been
referred to three sequences: Tua (Old), Lebaksiu and
Muda (Young). Subdivisions within each sequence have
been termed units. In most cases, the units are made up
of several lava flows. The relative ages of two of the
sequences, Tua and Muda, are well established (see
below). However, the stratigraphic position of the
Lebaksiu sequence is problematic. Also, the chemical
composition of Lebaksiu lavas is transitional between
those of the Tua and Muda sequences (Vukadinovic,
1989).
Morphology

On Java, only Gunung Semeru (3676 m) exceeds


Gunung Slamet (3428 m) in height. Gunung Slamet can
be divided broadly into two parts: the rugged, dissected,
western half of the volcano, consisting of deep valleys,
gullies, several plateaux and numerous peaks; and the
smooth, gently sloping, eastern half (see contour lines on
Fig. 1). Previous workers (van Bemmelen, 1949;
Pardyanto,
1971; Djuri, 1975; Aswin et al., 1984;
Sutawidjaja et al., 1985) have shown that the western

section is older than the eastern. Pardyanto (1971)


suggested that some of the numerous peaks in the
western region (e.g. Gunung Minggrik, Gunung
Sembung) represent eruptive centres from this period of
activity. The large plateaux of the western sector may be
due to block fracturing and faulting (van Bemmelen,
1949). Sutawidjaja er al. (1985) stated that the western
half of Slamet is composed entirely of andesitic lava and
tephra (excluding the Semaya and Waka eruptive
centres) and called it Slamet Tua. This term is retained
in the present study; however, subsequent field studies
(Vukadinovic, 1989) have found rare basaltic outcrops
within the Slamet Tua sequence. Future mapping will
undoubtedly uncover more new units.
The eastern part of the volcano is relatively young
and, as a consequence, topographically smooth due to
the relatively recent eruption of fluidal basaltic lavas;
Djuri (1975), Aswin et al. (1984), and Sutawidjaja et al.
(1985) termed it Slamet Muda. The smooth topography is disrupted on the northeastern slope by a field of
35 scoria cones studied by Sutawidjaja (1988). The cones
range in size from N 130 to 750 m in basal diameter and
from several to -250 m in height. According to the
terminology of Porter (1972), H,, = 0.25 * WC, and
WC,= 0.4. WC, for Slamet scoria cones, where H, = cone
height, W, = basal width and WC,= crater width. The
total volume of the scoria cones amounts to a mere
0.357 km3, but due to the vesicular nature of scoria, the
actual volume is even less. A K-Ar radiometric age date
on a sample of Slamet scoria gave 42 _+20 Ka (C. J.
Adams, personal communication, 1988). Most cones
have single craters, although some may have as many as

Geologic Map of Gunung Slamet Lavas

1 Keruh (ardldac.)

loo-es,

ICWO

1W.15

Cendarm(hb. emI.)

Top20

Fig. 1. Geologic map of Gunung Slamet lava and pyroclastic units. Modified from Sutawidjaja et al. (1985).

Evolution of G. Slamet Volcano, Indonesia


three. Many of the craters are breached towards the east
(downhill). Breached scoria-cone craters at Slamet
usually have a lava flow associated with the cone.
The summit of Gunung Slamet consists of four nested
craters with a spatial arrangement indicating that the
volcanic vent has migrated slightly from northeast to
southwest (N 1 km) during Muda time. Neumann van
Padang (1936) examined the possible volcanic hazards
resulting from such shifts in activity and concluded that
the northeast foot of the volcano is now best shielded
from volcanic catastrophe.
Tua sequence geology: Sirumiang mixed andesite

The Sirumiang mixed andesite, located on the western


side of the volcano at the foot of a fault scarp, is a small
dome (-0.51
km in diameter) composed of andesitic
material containing abundant basaltic enclaves up to
N 15 cm in size. The leucocratic host rock contains
abundant xenocrysts of quartz and feldspar (up to 5 mm
in diameter). All outcrops are densely overgrown, prohibiting a precise estimate of the relative proportions
of host andesite and enclaves. The origin and geochemistry of the Sirumiang mixed andesite is complexinvolving processes such as magma mixing, crustal
assimilation, and liquid-liquid diffusion (Vukadinovic,
in preparation)--and
will not be dealt with in this paper.
Tua sequence geology: Mendala basalts

The Mendala basalts are located in a confined area


west-northwest of the Slamet summit. Neither the eruption point for this rock unit nor contacts with other rock
units were found, restricting precise stratigraphic placement of the Mendala basalts, which are believed to
belong to Slamet Tua activity (Vukadinovic, 1989). In
outcrop, Mendala basalts show crude columnar jointing.
Plagioclase, pyroxene and olivine are visible in hand
specimens. No mineral alignment is evident in these
nonvesicular rocks.
Tua sequence geology : Sumbaga andesites

Sutawidjaja et al. (1985) subdivided Slamet Tua volcanic products into five units, whereas Pardyanto (1971)
split Slamet Tua into seven units using air-photo
interpretation. The field investigations and subsequent
chemical data of Vukadinovic (1989) are not in agreement with either division. The Slamet Tua andesites are
probably composed of numerous domes of viscous lava
of limited area1 extent. Due to the poor outcrops,
determination of stratigraphic/temporal
relationships
between these domes is difficult. In this study, the
andesitic rocks of this area are collectively named
Sumbaga. In general, the Sumbaga andesites are
nonvesicular,
phenocryst-rich
(N40-50%)
twopyroxene andesites with varying amounts of amphibole.
Note, however, that the Gunung Cendana, Kalipagu
and Keruh units have been distinguished on morphological and chemical grounds from Slamet Tua material
(Vukadinovic, 1989) and are discussed below.
Lebaksiu sequence geology

The term Lebaksiu was given by Aswin et al. (1984)


and Sutawidjaja et al. (1985) to the products of flank

137

eruptions on the lower southwestern slopes of Slamet.


They identified two separate eruptive centres near the
village of Semaya and Gunung Waka (Fig. 1). Both
centres have highly vesicular, thin (N 30 cm), basaltic
flows separated by thin lenses of agglutinated spatter
material. The high degree of vesicularity and low dispersal of tephra indicate a mild magmatic style of eruption
for the Lebaksiu flank eruptions. The term Lebaksiu
has been extended to include lavas with chemical composition similar to those of the Semaya and Waka basalts
(Vukadinovic, 1989).
Near the village of Siwarak on the eastern flank of the
volcano, extensive lava caves exist within Lebaksiu-type
basalts called Sirawak basalts (Vukadinovic, 1989).
The cave walls display tide marks indicating the rise and
fall of the outpouring lava, due to changing effusion
rates from the vent.
Basalts similar to those of Waka and Semaya occur on
the eastern slope above the site of the lava caves (Fig. 1).
Aswin et al. (1984) and Sutawidjaja et al. (1985) assumed
that the source of these flows is Gunung Malang, a
vent located approximately 600 m east of the summit.
The basalts contain olivine and pyroxene phenocrysts
and are relatively vesicular. Aswin et al. (1984) and
Sutawidjaja et al. (1985) called this unit Lawa Ganung
Malang and distinguished it from the Lebaksiu sequence. The Gunung Malang unit is here incorporated
in the Lebaksiu sequence solely on the basis of similar
chemical composition (Vukadinovic, 1989).
A unit of massive basalt between the village of Baturaden and the Cendana andesites is also tentatively
placed within Lebaksiu sequence on the basis of chemical parameters (Fig. 1).
Muda sequence geology

The youngest Slamet Muda volcanic productsexcluding the vent area-occur


on the northeast slopes
of the volcano, and the oldest on the southern
(Pardyanto, 1971; Aswin et al., 1984; Sutawidjaja et al.,
1985). In this study, the stratigraphy of Sutawidjaja et al.
(1985) has been adopted with slight changes.
Muda sequence geology: Baturaden basalt unit

The
Baturaden
basalt
unit occurs
on the
south-southeastern
slopes of Slamet. This unit is the
Lava Slamet 2 of Sutawidjaja et al. (1985)b who found
that Lava Slamet 2 and Lava Slamet 3 the (Banyumudal
unit of this study) sandwich a widespread scoriaceous
airfall deposit. For this reason the distinction between
the two lava units is retained.
Columnar-jointed flows that are 4-5 m thick crop up
at the village of Baturaden. Other outcrops are thick and
massive or thin and fluidal with variable vesicularity.
The source for the Baturaden unit is one of the summit
vents (Sutawidjaja et al., 1985).
Muda sequence geology: Banyumudal basalt unit

The Banyumudal basalts occur predominantly in the


northeast and east sectors of the study area. The lava
flows are generally < 3 m thick; thicker flows exist where
the underlying topography allowed the lava to accumulate. The chemical composition of these rocks, as with
the scoria cones, is similar to that of Baturaden

138

D. Vukadinovic and I. Sutawidjaja

unit rocks. The scoria cones formed during the


hiatus between Baturaden and Banyumudal activity
(Sutawidjaja et al., 1985).
Muda sequence geology : Legokmene basaltic an&site unit
In the northwest of the study area, a series of previously unmapped outcrops of basaltic-andesite lava lies
within Slamet Tua andesites. The outcrops extend northward from the eruptive centre, the Anggrung scoria cone
situated east of the town of Legokmene (Fig. 1). Geochemically, these basaltic andesites show affinities with
Slamet Muda volcanic rocks and, therefore, are classed
with them. The Legokmene lavas are generally massive
to slightly vesicular, whereas the Anggrung scoria cone
is built up of both airfall material, in which clasts range
from lapilli to bomb size, and associated surge deposits
with lapilli-sized tephra.

Other units: Keruh dacite unit

The Keruh ignimbrite occurs in the valleys of the


Keruh River system on the western slopes of the volcano. A small quarry near the village of Pengasinan
provides the best exposure. The deposit contains an
undetermined
number of ignimbrite sheets, each
N 3-7 m in thickness, with associated basal ground-surge
and co-ignimbrite ash-fall deposits. In the ignimbrites
proper, pumice clasts range from 1 to 15 cm in size and
are evenly distributed throughout a matrix of ash and
lapilli. The associated surge deposits, composed of
ash- and lapilli-sized particles, exhibit low-angle crossstratification. Unidirectional sedimentary bedforms in
surges are evidence for pyroclastic transport by a
ground-hugging, expanded, turbulent, gas-solid dispersion; this contrasts with the ignimbrite units, which
were probably transported as laminar, high-density-particle-to-gas concentrations (e.g. Cas and Wright, 1987).

Muda sequence geology: Kawah basalt unit


Other units: Cendana amphibole andesites

In 1973, magmatic activity at Slamet consisted of


relatively mild lava fountaining and the emplacement of
a ring of lava within the lowermost of Slamets summit
craters. Aswin et al. (1984) and Sutawidjaja et al. (1985)
suggested that the lava ring resulted from the explosive
disintegration of a lava dome, requiring the domeforming magma to be relatively viscous (> 10 poise;
Sutawidjaja, 1988). Indeed, in a magma undergoing
significant degassing (e.g. via fire fountaining), enough
undercooling will occur to promote rapid crystal growth,
increasing the viscosity and yield strength of the magma
(Sparks and Pinkerton, 1978). However, it is also possible that the ring was formed by a small volume of
basaltic lava rising to a height slightly above the top of
the vent. The outermost part of the lava may then have
chilled and solidified against the vent walls, leaving a
ring of basalt after the central part of the plug drained
back into the bowels of the volcano. This mechanism
avoids assuming viscosities for the Kawah basalt that are
characteristic of dacites and rhyolites. The name assigned previously to the Kawah unit was Lawa Kubah
(i.e. lava dome; Aswin et al., 1984; Sutawidjaja et al.,
1985). Since the mode of emplacement is debatable, the
name has been changed here to Kawah (crater), in
order to avoid genetic connotations.
In the Guci graben, northwest of the summit, an
unwelded scoria-flow deposit, chemically similar to that
of Kawah basal@ overlies Baturaden basaltic lava flows.
The scoria-flow deposit is -4-5 m thick and contains
normally graded, dense, scoria clasts (up to 30 cm in
diameter) and abundant charcoal in a matrix of black
lapilli and ash. Where the scoria flow overlies unconsolidated sediments, large fragments from the latter are
incorporated into the lower portion of the former. The
charcoal and rounded nature of the scoria clasts suggests
that the driving force of transport was magmatic, possibly caused by the collapse of a small eruption column
(J. V. Wright, personal communication).
Other units

The following section describes units from Gunung


Slamet that contain rocks with chemical characteristics
that are either distinct from or transitional between
those of the Tua and Muda sequences.

Located 2 km west of Baturaden village, the Cendana


amphibole andesites form several steep hills of
-200-300 m relief. The morphology and limited area1
extent (~0.5 km diameter) of the individual hills
suggests that the amphibole andesites were extruded as
thick, viscous domes. The domes are located in a circular
depression, which Aswin et al. (1984) interpreted as an
old crater. Poor exposure prevented a clear assessment
of the relative age of the Cendana andesites. Compared
with the Tua sequence, the Cendana andesites have
similar trace-element ratios and contents, but lower
*Sr/*?Sr values (Vukadinovic, 1989; see below).
Other units: Kalipagu basaltic andesites

The Kalipagu basaltic andesites occur on the


south-southwest slope of the volcano and extend from
the summit down to the Cendana crater (Sutawidjaja
et al., 1985). Kalipagu rocks are grey, massive, and
generally phenocryst-rich and occasionally show flow
foliation. The Kalipagu unit also contains andesites with
chemical similarities to both the Muda and Lebaksiu
sequences (Vukadinovic, 1989). Determination of the
temporal and chemical relationship
between the
Kalipagu and other units requires further field studies.

Petrography
Slamet Tua sequence: Mendala basalts

Mendala basalts are strikingly porphyritic (h 50%


phenocrysts) compared with other basalts of the Slamet
volcanic complex. Indeed, plagioclase phenocrysts are so
strongly zoned and abundant that the rocks can be
mistaken for andesites. Most of the plagioclase phenocrysts have a concentric arrangement of internal glass
and other cryptocrystalline inclusions, which documents
the growth history of the mineral. Large (3.5 mm)
euhedral clinopyroxene phenocrysts are strongly zoned,
more than in any other samples thus far discussed.
Inclusions of plagioclase, opaque granules and minor
olivine occur in the pyroxene phenocrysts.

Evolution of G. Slamet Volcano, Indonesia


Very rare orthopyroxene phenocrysts exist in some
thin sections. Common subhedral olivine phenocrysts,
with relatively extensive iddingsitization and marginal
resorption, reach a maximum size of 1.5 mm. Rare
magnetite is always associated with other ferromagnesian minerals. The hyalopilitic groundmass consists of
plagioclase, clinopyroxene, opaque, glass and olivine.
Slamet Tua sequence:
andesites

Sumbaga an&sites and basaltic

The highly porphyritic Tua andesites contain phenocrysts (w 50 ~01%) of abundant plagioclase; common
orthopyroxene, clinopyroxene and magnetite; and rare
amphibole and apatite. Groundmass textures are either
hyalopilitic or pilotaxitic/felty. The groundmass usually
contains plagioclase, opaque granules, clinopyroxene,
cryptocrystalline components and minor orthopyroxene
and clear glass. Zoning in plagioclase phenocrysts, which
is more noticeable in andesites than in basalts, consists
of well-defined concentric patterns with slight convolutions in many rings. Plagioclase phenocryst cores are
typically resorbed and encircled by a mantle of zoned
plagioclase containing inclusions of glass and rare
opaque material and pyroxene. Subhedral bladed plagioclase is typically 3 mm in length and, on average,
comprises 3040 ~01% of the rock.
Subhedral prismatic pyroxene varies in size, attaining
maximum lengths of 4 mm, and combines with plagioclase and Ti-magnetite phenocrysts to form crystal
aggregates. Occasionally, augite jackets the prism faces
of hypersthene crystals.
Subhedral equant Ti-magnetite
microphenocrysts
(~0.5 mm dia.) are rarely embayed and generally
inclusion free (except for rare apatite) and constitute
-2 ~01% of the rock. Magnetite tends to occur in close
association with other ferromagnesian minerals, particularly as inclusions and as members of crystal aggregates.
Anhedral-subhedral
pargasitic amphibole occurs in
many of the andesite samples, sometimes as large
megacrysts that are uniformly rimmed by a very fine
aggregate of Fe-Ti oxides and clinopyroxenes. These
crystals usually contain plagioclase and clinopyroxene
inclusions and display abundant disequilibrium textures
such as embayments and reaction coronas.
Most Slamet basaltic andesites and andesites contain
small apatite crystals with distinctive optical properties
that include moderate pleochroism (with absorption
strongest in the direction of the prominent cleavage) and
interference figures that vary from biaxial (2V x 40)
negative to uniaxial negative within the same thin section. The andesites contain the greatest amount of
apatite (< 1% modal), occurring as phenocrysts surrounded by mesostasis, grains within multi-phase crystal
aggregates, and inclusions in all phenocryst phases.
Very rare olivine is present in some andesites. Olivine
is highly resorbed, alters to iddingsite, and sometimes
displays a corona of clinopyroxene and plagioclase.
Lebaksiu sequence:

Gunung Waka and Semaya basalts

Semaya rocks have a coarse intersertal texture


represented by plagioclase microlaths (ave. length =
0.25 mm) with interstices occupied by abundant clinopyroxene, opaque granules, brown glass and lesser olivine.
Phenocrysts, particularly plagioclase, tend to be larger

139

than in other basalts, reaching 5-6 mm in length. Olivine


and clinopyroxene phenocrysts are present in all samples
and have features similar to those of Muda basalts.
Nearer to source, the Gunung Waka ba@ts have a
felty groundmass of opaque oxides, plagioclase and
clinopyroxene; approximately 5 km from the $ource, the
textures-besides
being coarser grained-vary
from
intersertal to intergranular to subophitic withinsingle thin
sections. Subhedral equant ohvine phenoc sts (max.
size e 2 mm) in the Gunung Waka basalts #,how more
extensive alteration, represented by opaques rims and
internal iddingsitization, than either the Sema$ or Muda
basalts. Gunung Waka plagioclase, on the other hand, is
relatively free of internal melt inclusions and resorbed
margins. Subhedral equant clinopyroxene
henocrysts
attain widths of 2.5 mm and contain commo 1 inclusions
of plagioclase. Ti-magnetite microphenocrysts~ are absent
from both flow units of the Lebaksiu sequence..
Lebaksiu sequence : Gunung Malang basalts

All samples collected from the Gunung Malang


basalts are rich in glomeroporphyritic,
seriate olivine
and plagioclase. Plagioclase laths have a maximum
length of 3 mm, and olivine grains are usually 2 mm or
less in diameter. Clinopyroxene phenocrysts hnd crystal
aggregates are rare. Zonation, both conc@ric and
sector, is common within the pyroxene
rains. The
groundmass is intergranular, with the plagio 1lase microlite framework filled by opaque, clinopyroxene and
olivine granules.
Muda sequence : lava flows

The Slamet Muda basalts are rich in porphyritic,


seriate plagioclase and contain lesser amounts of olivine,
clinopyroxene
and Ti-magnetite.
These minerals
commonly form glomerophenocrysts
andyor crystal
aggregates.
Subhedral bladed plagioclase (max. size range from 6
to 0.5 mm) is the most abundant phenocryst phase,
averaging 20-30 vol%, and displays a flow alignment in
some rocks. Generally, plagioclase cores contain abundant melt inclusions and are surrounded by clear rims
showing marginal resorption, which is less pronounced
in rocks without clinopyroxene phenocrysts. Minor
olivine inclusions are present in the feldspar phenocrysts
of some samples. Normal concentric zoning is not as
common as in the andesites.
Olivine is a common phenocryst phase, occurring in
varying amounts in the groundmass, and has an average
diameter of 0.5 mm, yet can be as large as 4.0 mm. The
crystals are subhedral, with some larger grains displaying embayments. A thin coating of Ti-magnetite
occasionally rims some grains. Olivine phenocrysts lack
zoning and are largely devoid of inclusions-although
rare plagioclase, magnetite, spine1 and clihopyroxene
(very rare) occur.
Subhedral
prismatic
clinopyroxene
phenocrysts
(N 5 ~01%) have diameters ranging between 0.5 and
4 mm and are pale green; some rocks contain crystals
that display very faint pleochroism (pale ~green-faint
pink), indicating higher than normal Ti contents. Most
clinopyroxene phenocrysts have some concentric and
sector zoning and contain common inclusions of olivine,
plagioclase and occasional glass.

D. Vukadinovic and I. Sutawidjaja

140

Although ubiquitous in the groundmass, equant


Ti-magnetite ( m 0.25 mm) grains are variably present as
microphenocrysts (< 1 vol%), ranging in shape from
euhedral cubes to anhedral blobs. On progression from
basalt to more evolved compositions, the morphologies
of magnetite phenocrysts change from skeletal or dendritic forms-indicating
a significant degree of magma
undercooling
(e.g. quench extensions )-to
squat,
equant, well-formed crystals. This implies that magnetite
had not reached saturation in many of the more primitive rock compositions, suggesting that delayed magnetite precipitation is the cause of relatively high Ti02
contents in some basalts and basaltic andesites from the
Muda sequence. Significantly, the high-TiOz basalts of
Slamet are the only magnetite-free volcanic rocks on
Java (Whitford, 1975a).
The groundmass textures are predominantly intergranular with Ti-magnetite, clinopyroxene and olivine
granules (in decreasing order of abundance) filling the
interstices between microhtes of plagioclase (0.05 mm in
length); some rocks possess an intersertal texture with
the addition of brown glass. Orientation of the microlites
varies from strongly aligned to randomly arranged.
Muda sequence: scoria cones

Seven scoria cones were sampled, primarily on the


basis of availability of fresh scoria. Texturally, the rocks
are hyalo-ophitic with pervasive dark glass enclosing
microlites of plagioclase, granules of olivine and rare
magnetite. Porphyritic
phases include plagioclase,
olivine, and clinopyroxene-all
with the same characteristics and relationships as in Muda sequence rocks.
Occasionally, olivine forms skeletal crystals.
Muda sequence: Kawah basalt

The porphyritic
Kawah basalt samples contain
phenocrysts of olivine, clinopyroxene and plagioclase.

Although the abundance of plagioclase phenocrysts is


relatively low, groundmass plagioclase is more abundant
and coarser than in other basalts. Interstices between
plagioclase laths hold murky brown glass, olivine,
magnetite and clinopyroxene granules.
Other units: Keruh dacite

Pumice from the unwelded Keruh pyroclastic flow is


extremely vesicular. The vitrophyric texture results from
subhedral phenocrysts of plagioclase, orthopyroxene,
clinopyroxene, amphibole and opaques set in clear,
vesicular glass. Minor apatite occurs as inclusions within
amphibole and orthopyroxene. Plagioclase is usually
riddled with brown glass, although some crystals may
have anhedral inclusion-free cores or are completely
devoid of inclusions. Other phenocryst phases-excluding magnetite-may
also have brown glass inclusions,
particularly the pyroxenes.
Other units: Cendana amphibole andesites

These rocks are distinctive in having pargasitic hornblende as their only ferromagnesian phenocryst phase
(< 1 cm). Zoning within amphiboles is optically visible;
and inclusions of plagioclase, apatite and minor pyroxene (both clinopyroxene and orthopyroxene) define the
crystal growth history by forming concentric patterns
about an inclusion-free core in larger crystals. The
euhedral prismatic amphibole grains show minimal evidence of reaction with the groundmass. Subhedral tabular plagioclase phenocrysts are (N 2 mm in diameter)
very strongly zoned and contain melt inclusions. Equant
magnetite (up to 0.25 mm) is rare to common in abundance. The groundmass is characterized by abundant
equant feldspar laths, and the interstices are occupied
by orthopyroxene, minor Ti-magnetite, minor clinopyroxene, abundant clear glass, and cryptocrystalline
components.

Ab
Fig. 2. Analyses of plagioclase phenocryst interiors plotted in terms of An-A&Or (mol%). Symbols: filled
circles = Muda (including Keruh); open circles = Tua (including Cendana); filled squares = Lebaksiu; open
triangles = Sirumiang.

Evolution of G. Slamet Volcano, Indonesia

Mineralogy

due to the complexities of plagioclase melting (see


above).
Electron-microprobe
point-analysis traverses across
Slamet plagioclase phenocrysts revealed a variety of
compositional profiles, including normal, oscillatory and

A table of mineral analyses and analytical procedures


and methods may be obtained from the main author.
Plagioclase

reverse zoning

Plagioclase core compositions are plotted in terms of


anorthite (An), albite (Ab), and orthoclase (Or) endmembers (Fig. 2). Maximum An core values within the
bulk-rock MgO ranges of Fig. 2 (i.e. >6,4-6, <4 wt%)
are constant at -An,.
Or content increases smoothly
from * 1 mol% at AnsO,to about Or, at An,. Although
the ranges of plagioclase compositions are similar in
each
bulk-rock
rocks
with
category,
MgO
MgO < 4 wt% have rare plagioclase cores as albitic as
An, and many more as sodic as An,, (Fig. 2). Within
each MgO category, plagioclase core compositions of the
major rock sequences (i.e. Tua, Lebaksiu and Muda) are
similar.
The simplest explanation for the Or-rich plagioclase
analyses is that they represent partially melted plagioclase developed on crystal-melt surfaces and cracks
within the crystal. Experimental work by Tsuchiyama
and Takahashi (1983) showed that Or component is
preferentially incorporated into the liquid during plagioclase melting; some of their melts contained as much
as 10.6mol% Or component. A large number of the
Or -rich plagioclase analyses are noticeably nonstoichiometric and contain relatively exotic elements such as
PzOs and TiOz in sufficient quantity to suggest that either
some interaction between the host magma and the
plagioclase melt took place and/or that trace elements
were concentrated by melting.
Plagioclase rim analyses are generally more &-rich
than corresponding core compositions. In addition,
plagioclase inclusions within ferromagnesian
phenocrysts are consistently more A&rich than are plagioclase phenocrysts within the same rock. Inclusions
of other minerals within plagioclase also appear to
have been in equilibrium with r lore evolved melt than
that from which the host phenocryst precipitated,
suggesting that magma mixing may have taken place.
Plagioclase inclusions within plagioclases were ignored

Pyroxenes

(a) MgO> 6 wt%

Fo

._ ~_ _~*-- ~__~__---_--_

~_

(b)

--

~----~--

profiles.

Calculation of cations and endmembers for pyroxene


analyses were treated according to the method of Papike
et al. (1974).
Core compositions of Slamet pyroxenes (Fig. 3) are
typical of those from island arcs (cf. Ewart, 1976;
Cameron and Papike, 1981). Orthopyroxene cores typically have mg# (100 * Mg/Mg + Fe*+) x 7@-60. Rocks
with MgO > 6 wt% have the largest range fin orthopyroxene compositions, with mg# as high as 80. These
mg # s are compatible with orthopyroxene ~forming in
equilibrium with the coexisting Ca-rich clinopyroxenes.
The lower-mg # (60-70) orthopyroxenes are anhedral
and rimmed by clinopyroxenes with hi her mg#s
(80-85) indicating that the orthopyroxe $ es may be
xenocrystic.
Amongst
rocks with MgO < 4 wt%,
orthopyroxenes from Slamet Muda andesites are richer
in Ca relative to those from Slamet Tua andesites,
possibly due to higher temperatures of form tion for the
former. In some basaltic andesites, pigeo ,ite rims on
orthopyroxene cores are common. Others co ponentsas defined by Papike et al. (1974)---in Sla ntnet orthopyroxenes are relatively low (N 5 mol% on average), with
Al(IV) typically more abundant than Ti, $Ia or Fe+.
In Ca-Mg-Fe space, Slamet Ca-rich clinopyroxenes
concentrate near the point where diopside,l endiopside,
salite and augite fields meet (cf. Deer ei al., 1966).
With decreasing bulk-rock MgO contents, the maximum amount of Ca (relative to Mg and Fe*+) in
clinopyroxene cores decreases, and the compositions
generally shift from diopside to salite-augitie (Fig. 3): a
common phenomenon in pyroxenes from arc volcanic
rocks. Slamet clinopyroxenes contain approximately
15-20 mol% Others components. As in orthopyroxenes,
Al(IV) makes up most of the Others component in
clinopyroxenes. Muda clinopyroxenes generally have

6>MgOwt%.4

_
-

fli%fTr

141

c
_-

(c) MgO < 4 wt.96

Fa

-b

Fig. 3. Compositional variation of pyroxenes, olivines (symbols plot below the pyroxene quadrants) and
amphiboles (symbols plot in the centre of the pyroxene quadrants) in terms of Ca, Mg and Fe (at mic
proportions). Hash marks at 10% increments. Rocks are divided according to MgO wt% [(a) MgO 2 6; (b)
6 > MgO 2 4; (c) MgO < 41. Symbols as in Fig. 2.

142

D. Vukadinovic and I. Sutawidjaja


100

35

40

45

50
Mg#(host

55

60

65

rock)

Fig. 4. mg # of pyroxenes vs 100 * Mg/(Mg + Fe*+ ) of rock samples (Fe *+ = 0.85*XFe). Lines represent range
of phenocryst compositions:
thick, solid = Muda; horizontally
dashed = Lebaksiu; and diagonally
dashed = Tua (including Sirumiang). Closed symbols and x s represent clinopyroxene and orthopyroxene
inclusion compositions, respectively. Numbered curves (0.23 and 0.27) denote Kd values (Grove et al., 1982)
for clinopyroxene and orthopyroxene, respectively. Very few inclusion compositions lie above the Kd curves.
slightly higher Ti levels relative to those of other Slamet
clinopyroxenes.
Compared with phenocryst interiors, the rim and
groundmass
compositions
from Slamet pyroxenes
usually have lower mgf values. Pyroxene inclusions
occurring in other phenocryst phases are usually lower
in mg# than that of the pyroxene phenocrysts from the
same rock (Fig. 4). A similar relationship is seen for
olivine inclusions and phenocrysts. The abundance of
Others components is variable, but is generally lower
than those of the phenocrysts.
As with plagioclase, zoning profiles across Slamet
pyroxenes can vary within single samples. However,
mg# values within single pyroxene crystals do not vary
widely.

MgO < 4 wt% are from the Legokmene unit. The compositions for these olivines are approximately Fob?.
Olivine rims and groundmass usually have lower mgf
values than coexisting phenocrysts. The Fo content of
rims and groundmass can be as low as -Fo,,.
Olivine inclusions within other phenocryst phases are
usually lower in Fo component than are olivine phenocrysts from the same rock. This feature is illustrated
in Fig. 5, in which the range of Fo content for olivine
phenocrysts and inclusions is plotted against the mg#
value of the host rock.
Significant chemical zoning, in terms of mg# is rare
and subtle in olivine phenocrysts and can vary from
normal to reverse zoning within a single rock specimen.

Olivines

Oxides

The fosterite content of olivine cores ranges widely


(Fig. 3). Rocks with MgO > 6 wt% have olivines ranging
in composition from -Fows.
Rocks with 6 > MgO
wt% > 4 also display a wide range of olivine Fo contents

The method of Carmichael (1967) was used to calculate Fe,O, in both spinels and hexagonal oxides. The
predominant oxide in Slamet lavas is titanomagnetite. In
terms of Ti, Fe2+ and Fe3+, titanomagnetite phenocrysts

90

). The only analysed olivines from rocks with


(-foscrs~

80

40

45

ri

;,

..,,...a.

60

55

50
Hg%xl

65

rock)

Fig. 5. Fo contents of olivines vs 100 * Mg/(Mg + Fe*+) of rock samples (Fe*+ = 0.85*EFe). Solid symbols
represent range of phenocryst compositions; open symbols represent inclusion compositions. Circles and solid
lines = Muda; squares and lines with short dashes = Lebaksiu; triangles and lines with long dashes = Tua
(including Sirumiang enclaves). Numbered curves (0.30 and 0.40) denote Kd values.

Evolution of G. Slamet Volcano, Indonesia


from Muda lavas contain the highest ulviispinel component. Although some Muda lavas have high TiOs
levels, suggesting that the amount of ulviispinel in
magnetite is a function of bulk rock composition, many
lavas with low Ti02 contain magnetite with relatively
high amounts of ulviispinel (e.g. Keruh unit). Experimental work demonstrates that high-TiOz magnetites
can crystallize under conditions of high pressure
(3 10 kbar; Osborn et al., 1978). This implies that Muda
magmas may have undergone limited crystallization at
high pressures and may have risen and erupted quickly
enough for the preservation of the high-TiOz magnetites.
Certainly, the rarity of andesites in the Muda sequence
is in accord with short resident times at shallow crustal
levels for the parental magmas. On the other hand, the
higher &&pine1 component in Muda magnetites may
also reflect precipitation of magnetite unaccompanied by
a TiOz-bearing phase, e.g. amphibole. As a result,
interpretation of TiOz contents in Slamet magnetites
remains equivocal without further study.
Very rare ilmenites occur as inclusions in phenocryst
phases of more evolved lavas. Geothermometers
or
barometers were not applied, for it is highly unlikely that
the ilmenites and magnetites formed in equilibrium.
Chromium-bearing spinels occur in the more mafic
lavas, particularly as inclusions within Fo-rich olivine
phenocrysts, and range in composition from picotite to
chromite. The Cr-rich oxides are probably remnants
from the early stages of fractionation of mantle-derived
magmas.
Amphiboles

According to the guidelines of Leake (1978), Slamet


amphiboles are pargasites to ferroan pargasites. Amphiboles from Muda lavas are titanian pargasites (Ti > 0.25
atoms per 23 oxygen), whereas those from the Sirumiang
mixed andesite are potassian (K 2 0.25).

25 I

MgoH

Fig. 6. (a) 1000* Zr/K vs MgO wt%. (b) Zr/Rb vs MgO wt%.
Muda lavas have higher 1000+ Zr/K and Zr/Rb than Tua
lavas. N-MORB values for 1000z Zr/K and Zr/Rb are z 120
and 80, respectively (Hofmann, 1988). Symbols as in Fig. 2.

143

40A
30 a'
f

20.:

a
As25

a5167

OHat

. ..

10 *I

XNucm
AEndnw
3

8 ..

'

?? .

'0

04

MgO_
l0.8 .'

B
0

9
P

szs 0

0.6 .n
0.4 .'

0
CHmt

OS167

.
??. .

0.2 a'

-0.

??

.
.

_
1

MgOwt%

Fig. 7. (A) Zr/Nb vs MgO wt%. (B) Hf/Nb VSIMgO wt%.


Muda and Lebaksiu lavas have lower Zr/Nb and; Hf/Nb than
Tua lavas. Symbols as in Fig. 2; x = N-MORE/ (Hofmann,
1988); S167 = Keruh; S25 = Cendana.

Geochemistry
The lavas of Slamet can be subdivided on the basis of
Zr/K and Zr/Rb ratios (Fig. 6). Two groups, recognized
on the basis of these ratios, age and incompatible
trace-element abundance have been named l+w and high
abundance magmas (LAM and HAM; Vukadinovic and
Nicholls, 1989). In terms of the geological units
described above, Tua (Sirumiang, Mendala and Sumbaga), Lebaksiu and Cendana belong to the LAM
Banyumudal
and
group; and Muda (Baturaden,
Kubah), Keruh, Legokmene and Kalipagu belong to the
HAM group. In general, Muda lavas have greater Zr/K
(N 15) and Zr/Rb (N 5) ratios than those of Tua rocks
(~8 and -2, respectively), and Lebaksiu lavas are
transitional between those of Muda and Tua. However,
Cendana amphibole andesites have trace-element contents and ratios that are characteristic of Tda rocks but
have 87Sr/86Srvalues that fall within the Muda range.
Similarly, the Keruh dacite has trace-element levels and
ratios resembling those of Muda lavas but has *Sr/Sr
similar to those of Lebaksiu lavas.
Other incompatible trace-element ratios also discriminate between LAM and HAM rocks, particularly Zr/Nb
and Hf/Nb; both ratios involve elements that are relatively immobile in H,O/rock systems (e.g. Cann, 1970).
LAM rocks have Zr/Nb and Hf/Nb values of approximately 25 and 0.7, respectively (Fig. 7) whit are similar
to those of MORB (BVSP, 1981). In HA lv! lavas these
ratios are lower in value (Zr/Nb w 12; Hf/Nb -0.25)
and resemble those of enriched-MORB (cf. BVSP, 198 1;
le Roex, 1987).
Major- and trace-element variations

The variation of major and trace elements, with


respect to MgO, in Slamet rocks is illustrated in Fig. 8.
Silica is almost constant (N 50-5 1 wt%) in rocks with
MgO levels ~4.5%. Exceptions include the felsic host
the
Sirumiang
mixed
andesite
(MgO x 5,
of

D. Vukadinovic and I. Sutawidjaja

144

13
11 xf

97-i

"s"

Yx

". +

AA
c*

.A
0

II
0

?--A---

.
I

16,

4L
.'*'*'.'.'.'.'*
3
4
12

12

5
hiO

M90

Fig. S-Caption

on p. 147.

145

Evolution of G. Slamet Volcano, Indonesia

LAM
I

.I.

.I

.,

.I

.,

.
r-

600
0

500

0
0

400
i
300

?(
b..;.

200
100

t.

L
300

%-.

xx

??

X
x
*

x
*

1..
8

.,\x;xa
X
Lub)
AA
a

dX xx

:
2001
1

.I*
2

'*
3

'*'*'.'.
4
5

'.
8

5
WO

Fig. I-Continued.

Caption on p. 147.

D. Vukadinovic and I. Sutawidjaja

146

8
300 -

LAM

HAM

+ ixx
IJ

40

??
??

x2,
i,kf
+

xx
X

.,.,.,.I.,.,.

30 .i
20 -

SQ
*

10 -

400 8
300 -

*ix

""*,

x:

+
200 -

*U#!t

nla

100 -

o-

~'~'.'~'~'~'~'~

400
8
300 -

8
m

200 - t

##
A

loo -

o.aA--lA**..
12

? ?m
x

xx

12

MO

NO
Fig. 8-Continued.

Caption opposite on p. 147.

147

Evolution of G. Slamet Volcano, Indonesia


100
LAM

80 60 .
40

HAM

m
x
x

'e

x
??

X
A

&3X

140

??

60 40

*
12

'

'

'

*'

.
6

'

I
9

12

.,

??

0
m

d
7

I.
8

HAM

LAM
Lebaksiu
G. Cendana
Sumbaga
Mendala
Sirum. Host
Sirum. Enclave

.I

WJ

w&J

X
+
A

I
5

0
0

Kawah
Kalipagu
?? Keruh
+ Legokmene
x Banyumudal
A Baturaden

Fig. 8. Major elements and selected trace elements vs MgO wt% for LAM (left-hand column) and HAM
(right-hand column). Oxides in wt%, others in ppm.

SiO, x 58 wt%) and, to a lesser degree, the Kawah lavas


(MgO x 4.5, SiO 2w 53 wt%). LAM and HAM rocks
with MgO < 4.5% increase in Si02 with decreasing
MgO. The most silicic material is from the pumiceous
Keruh dacite unit (z 63 wt%).
TiO, content increases from -0.9 to 1.4 wt% with
decreasing MgO in LAM rocks with MgO > 6%; this
trend is defined mainly by the Lebaksiu sequence. The
Sirumiang felsic host has anomalously low values of
TiO, (-0.5%)
relative to its MgO level. HAM rocks
have marginally higher TiO, abundance than LAM
rocks at comparable MgO levels. HAM lavas with MgO
values between 4 and 5% have TiOz contents ranging
from N 1.2 to 1.9%; the Kawah and Banyumudal units
define the lower and upper limits of this range. Evolved
HAM rocks display a positive correlation between TiO,
and MgO.
In both LAM and HAM basaltic rocks, A&O3
increases from N 15 to 18 wt% with decreasing MgO. In
rocks with MgO < 4%, A&O, remains between N 17.5
and 20%; however, the Sirumiang felsic host has lower
Al,O, (w 16.5%). The Kalipagu andesites extend the
basaltic trend in which A&O3 increases with decreasing
MgO. Generally, A&O3 levels of Baturaden basalts are
slightly higher than those of Banyumudal lavas.
Total iron, expressed as total Fe0 (FeO*), ranges
from x 11 to 5 wt% and behaves similarly to TiO,. The
Sirumiang felsic host has lower FeO* (N 6%) than that
of the main trend formed by LAM rocks. Note that

inflections on both the FeO* and TiOz vs MgO graphs


occur at identical MgO levels for LAM (-6%)
and
HAM (-4.5%),
implying magnetite saturation and
fractionation at these levels. Also, in HAM rocks with
MgO between 4 and 5%, FeO* and TiO, are both
anomalously high-particularly
in Banyumudal lavassuggesting minor magnetite accumulation,.
CaO decreases regularly with decreasin MgO in both
LAM and HAM rocks. The Cendana and fM endala units
are slightly richer in CaO, and the Sirumiang felsic host
is poorer, compared with other LAM rocks. The CaO vs
MgO trend for HAM is tighter than that for LAM
rocks. In the HAM group, the Baturaden unit has higher
CaO levels than those of the Banyumudal unit. In HAM
rocks, the trend steepens slightly at MgO < 4 wt%.
of LAM
and HAM
lavas
Na, 0
contents
(-2.54
wt%) increase with decreasing MgO. Except for
the Lebaksiu unit, LAM lavas have slightly less Na,O
at any given MgO value compared with HAM lavas.
The Sirumiang felsic host has higher Na,O than the
trend defined by LAM rocks. In the HAM group,
Kawah rocks have higher levels of Na, 0 compared with
Baturaden lavas.
The abundance of K,O and Rb increases regularly
and similarly with decreasing MgO. Unlike Na,O, K,O
and Rb levels are similar between LAM and HAM lavas,
with ranges of m l-3 wt% and 20-80 ppm, respectively.
Within the LAM group, K and Rb levels are lower in
Cendana andesites compared with Sumbaga andesites.

148

D. Vukadinovic and I. Sutawidjaja

The less-mafic members of the Kalipagu andesites also


plot below the main trend.
Ba-MgO variation is similar to that of K and Rb, and
the range of Ba content is large (N 100-600 ppm). The
Lebaksiu basalts lie above the trend defined by the
Sumbaga andesites and Mendala basalts. In HAM
group rocks with > 4% MgO, Ba increases only slightly
with decreasing MgO; however, at ~4% MgO, Ba rises
more sharply.
Both LAM and HAM lavas generally increase in Sr
(-275-350 ppm) with decreasing MgO. The Sirumiang
host and enclave material are both significantly richer in
Sr, compared with other Slamet rocks. The Baturaden
basalts and Kalipagu andesites have higher and more
varied Sr values (between 300 and 400ppm) than do
other units with similar MgO contents.
Zr levels are considerably
lower in LAM
(N 50-l 75 ppm) than in HAM rocks (N 100-350 ppm).
In both groups, Zr content increases regularly with
decreasing MgO. The Lebaksiu basalts are marginally
richer, and the Cendana amphibole andesites poorer, in
Zr than are other LAM lavas at similar MgO contents.
Zr levels clearly distinguish Muda units from each other.
In particular, Baturaden basalts and Kalipagu andesites
are depleted in Zr, relative to Banyumudal basalts and
Legokmene andesites; this is also true for Y. Kawah
basaltic andesites, having Zr values (-200 ppm) intermediate to those of the Banyumudal and Baturaden
basalts. In the Keruh dacite unit, Zr contents rise steeply
with decreasing MgO.
Yttrium contents are relatively constant at -25 ppm
throughout the spectrum of LAM lavas. The Cendana
and some of the Sumbaga andesites are noticeably lower
in Y compared with other LAM rocks. HAM rocks are
richer in Y, which increases slowly (N 25-35 ppm) with
decreasing MgO, compared with LAM rocks.
SC and V levels decrease with decreasing MgO in both
LAM and HAM groups, with slight steepening of the
trends at ~6% and ~5% MgO for LAM and HAM
rocks, respectively. Overall levels of SC and V are similar
two
between
the
groups
(Sc z 350-10 ppm,
V x 350-50 ppm). The Kalipagu andesites are slightly
richer in SC and V compared with the Legokmene
andesites.
In both LAM and HAM groups, Cr and Ni values
decrease with decreasing MgO levels. Overall Cr contents are roughly equivalent between the two groups
(max. w 350 ppm), but Ni levels are higher in LAM lavas
(max. N 80 ppm, cf. -60 ppm in HAM). Low-MgO
Lebaksiu basalts have higher Cr and Ni abundance
compared with Mendala basalts and some Sumbaga
andesites. Baturaden basalts, particularly those with low
MgO, are richer in Cr and Ni than the other HAM
units.

Calcalkaline or tholeiitic classljication for Slamet volcanic


rocks?

The definition of the terms calcalkaline and tholeiitic has changed and diversified through time. The
problem is compounded by the frequent usage of these
terms in the literature without workers specifying what
they exactly mean by them. Peacock (1931) first proposed the term calcalkalic for rock suites with an
alkali-lime index between 56 and 61. Slamet lavas are
calcalkalic according to this classification scheme
(Fig. 9a). Peacocks alkali-lime index is seldom used
these days (except as a historical footnote); instead,
petrologists usually classify subalkalic, subductionrelated rocks as either calcalkaline or tholeiitic in nature.
Wager and Deer (1939) defined tholeiitic series as
those showing significant Fe-enrichment relative to Mg
and alkalies during differentiation (e.g. Skaergaard,
Greenland) and calcalkaline series as those lacking Feenrichment (e.g. Cascades, western U.S.A.). AFM diagrams (Fig. 9b) are usually used to distinguish between
the two series. According to the definitions set out by
Irvine and Baragar (197 1), Slamet lavas are calcalkaline
but lie close to the dividing line between the series, with
some mafic samples actually plotting in the tholeiitic
field.
Similarly, Miyashiro (1974) regarded volcanic rocks as
either calcalkaline or tholeiitic in nature on the basis of
FeO*/MgO relative to silica. Gill (1981) used this diagram to define tholeiitic rocks as those having high
FeO*/MgO relative to SiOz, regardless of the slope of
the line. On the FeO*/MgO vs SiOr diagram (Fig. SC),
Slamet rocks lie predominantly in the tholeiitic field as
defined by Gill (1981), even though the trend formed by
Slamet rocks has a shallower slope than that of the
defining line of Fig. 9c.
Alternatively, the continuum between calcalkaline and
tholeiitic rock series can be separated on the basis of
LILE vs SiOz systematics. Due to greater amounts of
data, KzO values are commonly used as representative
of LILE contents in the belief that the abundance of the
former mimics that of the latter (Gill, 1981). For
example, Jakes and Gill (1970) demonstrated a positive
correlation between La/Yb (and La) and KzO in arc
rocks. According to the K, 0 vs SiOz classification scheme
(Peccerillo and Taylor, 1976), Slamet volcanic rocks are
calcalkaline to high-K calcalkaline in character (Fig. 9d).
By considering four different methods of classification, Slamet rocks can be pigeonholed twice as calcalkaline, once as tholeiitic, and once as transitional
between calcalkaline and high-K calcalkaline. Clearly,
classification of subduction-related igneous rocks is a
moot point. The discrepancies between the different
classifications are probably due to their monitoring of

Fig. 9 (opposire). (a) Peacocks (1931) alkali-lime index applied to Slamet rocks. Slamet lavas are calcalkalic
according to this classification scheme. Circles = Na,O + K20; squares = CaO. (b) AFM diagram showing
LAM (open circles) and HAM (filled circles). Thin, dashed line with extreme FeO-enrichment is the
Skaergaard trend (Wager and Brown, 1967); thick, dashed line is the boundary between the tholeiitic (Th)
and calcalkaline fields (Ca) as defined by Irvine and Baragar (1971). (c) FeO*/MgO vs SiO, for Slamet rocks.
LAM (circles) and HAM (squares). Dividing line between tholeiitic (TH) and calcalkaline (CA) fields is from
Gill (1981). (d) K,O vs SiO, for Slamet rocks. Fields adopted from Peccerillo and Taylor (1976). Thick,
sub-horizontal lines define the following fields: I = tholeiitic; II = calcalkaline; III = high-K calcalkaline;
IV = alkaline. Thin, vertical lines define boundaries within the basalt-andesite-dacite compositional spectrum.
Symbols as in (c).

149

Evolution of G. Slamet Volcano, Indonesia

2 --

Na20

Alkali-lime
index B 60

+ K20

01

49

52

55

58

61

64

SiO2 wt.%

N&O +
K20

4-

3 a.

01..

45

-:.

*:*.

50

.
55

-:.

60

65

60

65

SiO2 wt%

hall

3
P

!
1

0
45

50

55
Si02 wt%
Fig. 9-Caption

opposite.

D. Vukadinovic and I. Sutawidjaja

150

different aspects of magma genesis. For instance,


FeO*/MgO vs SiOz tells us much about the crystal
fractionation history of a suite, particularly the role
of ferromagnesian minerals (especially magnetite; see
Osborn, 1959), but little about the nature of the source
material of the magma. On the other hand, a KzO vs
SiO, plot may indicate if a suite of rocks were derived from a LILE-enriched source, but it does not
reveal much about the nature of high-level fractionation
processes (unless a K-rich phase is involved, usually at
higher levels of SiOz, or significant crustal assimilation
has taken place). An extensive survey by Gill (1981)
indicates that the supposed inverse relationship between
Fe-enrichment and K,O contents in arc rocks (Jakes and
Gill, 1970) is . . . crude at best . . . (p. 107).
In summary, one can unambiguously state that Slamet
lavas are subduction-related, subalkalic rocks that cover
a wide compositional spectrum. One can also state
(1) that the rocks are relatively K-rich and (2) that they
show some Fe-enrichment in the mafic end of the compositional range. These are two characteristics that may
reflect the nature of different regimes: magma source (cf.
Vukadinovic and Nicholls, 1989) and magma chamber.
100

Mg016

?I

wt.%

11

100

b
0

x
*
5

4G

10

li

6>Mg0>4

111

100

wt.%

Rare earth elements

Figure 10 is a series of conventional chondritenormalized REE plots for Slamet volcanic rocks within
one of three MgO ranges (MgO 2 6 wt%; 6 > MgO >
4 wt%; MgO < 4 wt%). Across the spectrum of MgO
values, REE abundance is consistently lower in LAM
than in HAM lavas. Significant Eu anomalies
(1.05 < Eu/Eu* < 0.95) occur in Slamet volcanic rocks;
however, Eu/Eu* is generally smaller (i.e. larger downward trough) in the HAM rock group. Furthermore, as
MgO decreases from 26 to <4 wt%, the average
Eu/Eu* of HAM group lavas progressively decreases
from 0.88 to 0.82. LAM rocks have Eu/Eu* averaging
~0.94 in each of the MgO ranges. This difference between HAM and LAM implies either: (1) that plagioclase
fractionation was more effective in HAM magmas; (2)
that the oxidation state, and consequently the Eu3+/Eu2+
ratio, of LAM magmas were higher than those of HAM;
or (3) that greater amounts of fractionation occurred in
LAM and Lebaksiu magmas of a phase that accepts
Eu3+ in preference to Eu2+ (e.g. apatite), thus counteracting the effect of plagioclase fractionation. Due to the
poor analytical precision for phosphorous, this latter
possibility cannot be investigated.
LAM and HAM rocks with MgO B 6 wt% have
parallel REE patterns; e.g. average (La/Yb),, ratios for
LAM and HAM basalts are 4.1 and 4.5, respectively. In
rocks with 64 wt% MgO, LREE in LAM lavas are also
parallel to those of HAM, but HREE in the former rock
group have a slight concave-up pattern, indicating that
greater amounts of amphibole or clinopyroxene fractionation took place in LAM than in HAM magmas. The
depletion of mid-HREE in LAM lavas is most evident
in rocks with MgO < 4 wt%. HAM lavas do not have a
depletion in mid-HREE in any of the MgO groupings.
Negative Ce anomalies, which are relatively common
in arc volcanic rocks (White and Patchett, 1984; Hole et
al., 1984) occur in only two samples: S178 and S25.
Sample S104 provides the only instance of a slight
positive Ce/Ce* (> 1.05) occurring in Slamet material.
The significance of these anomalies is difficult to assess.
Hole et al. (1984) attributed the occurrence of negative
Ce anomalies in arc magmas to the participation of
subducted sediments that have strong negative Ce
anomalies. Conversely, the presence of negative Ce
anomalies in Lesser Antilles arc basalts and largely
positive ones in the fore-arc sediments (DSDP 543;
White et al., 1985) indicates that Ce decoupling
may be due to relatively oxidizing conditions in the
magma-source region (White and Patchett, 1984).
Mantle-normalized

a ,i

La

Mg014

Ce Pr

Nd

wt.%

SmEuGdTbDyHoEr

Yb

Fig. 10. Chondrite-normalized REE patterns for HAM and


LAM. Rocks are divided according to MgO wt% content.
Symbols: filled squares = HAM; open squares = LAM. Normalizing values from Taylor and McLennan (1985).

trace-element

abundance

diagrams

A notable feature of the mantle-normalized diagrams


(Fig. 11) is the behavior of Nb in LAM and HAM rocks.
As with REE, Nb, Zr, Hf, Ti and Y levels are consistently lower in LAM than in HAM lavas throughout the
MgO spectrum. As MgO decreases, Zr and Hf become
progressively enriched relative to Sm and Y; however, Ti
develops distinct negative anomalies. This suggests that
as Slamet magmas evolved, Zr and Hf became more
incompatible than did Sm and HREE, which in turn
became more incompatible than Ti. The onset of
Ti-magnetite crystallization in basaltic-andesitic and
andesitic magmas will certainly cause Ti depletion in the

151

Evolution of G. Slamet Volcano, Indonesia

Strontium levels are similar in LAM and HAM


throughout the MgO spectrum. Negative Sr anomalies
are common in HAM rocks and increase with decreasing MgO. LAM lavas generally lack strontium anomalies, suggesting that a greater proportion of plagioclase
existed in the fractionating assemblage of HAM magmas
compared with those of LAM (see Vukadinovic, 1993,
for a discussion on Sr anomalies in arc basalts).
The highly incompatible elements (e.g. Cs, Th, U)
occur in similar quantities in LAM and HkM rocks at
comparative MgO levels. Caesium shows the most scatter, possibly reflecting the difficulty involve in obtaining
precise analyses for this element via SdsMS: consequently, the Cs data should be used with caution. Ratios
such as Ba/Rb and Th/U are relatively constant throughout the whole MgO range for both LAM and HAM
groups, but Lebaksiu magmas have slightly higher
Ba/Rb than that of other Slamet rocks.

&@CJs?6wt.%

6>MgOL4wt.%

Strontium isotopes

of *Sr/%r
ratios
from
Slamet
The
range
(0.70478-0.70629) is among the widest known from a
single arc volcano. In general, 87Sr/86Srratios decrease
with decreasing age in Slamet volcanic ocks: LAM
rocks have higher 87Sr/86Sr(0.70565-0.7062 b ) compared
with those of HAM (0.70478-0.70578). This trend is
maintained by the Kawah unit, erupted in 1973, which
has the lowest Sr/%r ratio.
scatter with
*Sr/%r ratios show considerable
respect to both MgO and SiOz and correlate negatively
with IMITER ratios Zr/K and Zr/Rb (Pig. 12) the
significance of which is considered below.

Fig. 11. Mantle-normalized diagram with rocks grouped as in


Fig. 10. Symbols: HAM = filled circles; LAM = open circles;
S167 (Keruh dacite) = filled squares. Normalizing values from
Taylor and McLennan (1985). Shaded area in (c) is the range
of HAM values from (b).

Magma Evolution Process&


Introduction

liquid, the crystallization of which is in accord with petrographic and major-element observations. It is also well
documented that the crystal/liquid partition coefficients
for REE increase with increasing polymerization of
silicate melts (e.g. Watson, 1976; Ryerson and Hess,
1978; Mahood and Hildreth, 1983; Nash and Crecraft,
1985; Lesher, 1986). If increased melt polymerization
affects Hf and Zr partition coefficients less than those for
the REE, then enrichment of Zr and Hf relative to the
mid-REE can occur with progressive fractionation.

Ideas on the origin of andesites during the last 30-40


years are many and varied. Although in the 1950s
numerous workers (following Bowen, 1928) had suggested that basalts are parental to andesites via crystal
fractionation, and as such . . . the origin of andesite
magmas cannot be discussed independently 1from that of
basalt magmas (Kuno, 1968, p. 149), petrologists have
devoted much time and effort to attemptink to demonstrate a primary origin for andesites.

25

0.7045

0.705

0.7055

0.706

0.7065

67w66Y

Fig. 12. 1000 + Zr/K vs *Sr/%r. Error bars at different values of 1000 + Zr/K assuming f 10% for Zri and
K. Field enclosed by dotted line = low-*Sr/?Sr Muda lavas; dashed line = high-*Sr/Sr Muda lavas.
Symbols as in Fig. 2 and open diamonds = other units.

152

D. Vukadinovic and I. Sutawidjaja

On the basis of liquidus experiments for a range of


calcalkaline compositions (high-Al basalt through to
rhyolite), Green and Ringwood (1968) concluded that
primary andesite magma can be derived from anhydrous
melting of quartz eclogite, in turn derived from subducted basalt. This type of direct origin for andesites was
advanced primarily by Marsh and Carmichael (1974)
and Marsh (1976,1978). Subsequent advances in knowledge of equilibrium trace-element (especially REE) partitioning behavior between liquids and coexisting solid
phases (e.g. Gast, 1968; Shaw, 1970) revealed that
andesitic magmas are unlikely to have been in equilibrium with significant amounts of garnet either as a
residual phase during source melting or as part of a
fractionating assemblage (e.g. Gill, 1974, 1978; Nicholls
and Harris, 1980). To avoid this problem, Marsh and
co-workers have instead proposed that high-Al basalt
magmas can be generated from eclogite by degrees of
melting sufficient to eliminate garnet from residues and
that andesitic magmas are subsequently produced by
fractionation (e.g. Brophy and Marsh, 1986). In spite of
this, support for models of primary andesite (or, for that
matter, arc basalt) generation directly from deep eclogite
melting has been largely abandoned. The present consensus on the main role of the subducted lithosphere is
that it acts as the source of metasomatic agents that
influence the chemistry and melting behavior of the
overlying mantle wedge.
Overlying arc crust may also be involved in the
generation of andesites. Frequently, the formation of
rhyolitic magma is attributed to the melting of continental crustal material alone. Ewart et al. (1968) and Ewart
and Stipp (1968) called upon partial fusion of
greywacke-argillite
basement to produce the liquids
giving rise to the rhyolitic volcanic rocks of the North
Island, New Zealand. Blattner and Reid (1982) have
opposed this interpretation on the basis of oxygen
isotope data, concluding that the magmas in question
were originally mantle derived but underwent extensive
greywacke contamination during their ascent to the
surface. In addition, crystallization experiments by
Conrad et al. (1988) demonstrated that a peraluminous
source (such as the greywacke comprising the North
Island basement) for the metaluminous North Island
rhyolites was unlikely. Nonetheless, some examples of
rhyolites derived solely by crustal melting apparently
exist. For example, a crustal melting origin was considered on the basis of 87Sr/86Srratios obtained for the
Sumatran Toba ignimbrite (Whitford, 1975b) and for
both the rhyolites andandesites of the Padang area, West
Sumatra (Leo et al., 1980). However, such models of
simple crustal fusion are rarely invoked for the genesis
of andesites, particularly for those from intra-oceanic
arcs. Exceptions include unusual andesite occurrences
such as the peraluminous, cordierite-bearing lavas from
Ambon, Indonesia (Whitford and Jezek, 1979).
From the preceding account, it is evident that andesites are unlikely to represent purely primary magmas
from any type of source, except in unusual circumstances. Thus, ideas on andesite petrogenesis have come
full circle: most models presented in the current literature call largely upon crystal fractionation from parental
basalts (see Gill, 1981, p. 272). The aspect in which
current models diverge is in the openness of their
magma chambers. For example, what are the relative
proportions of crustal assimilant to crystallizing min-

erals involved in AFC (e.g. Briqueu and Lancelot, 1979;


DePaolo, 1981)? Does bulk assimilation occur, or do
only partial melts of the assimilant mix with the crystallizing magma (e.g. Patchett, 1980)? If the latter applies,
does the partial melt mix thoroughly with the differentiating magma, or are selective processes involved (see
Grove et al., 1988, for an example of many of the above
processes)? In addition to the above processes, are
magma chambers periodically replenished by fresh influxes of magma and tapped via eruption while continuing
to crysiallize (e.g. OHara and Matthews, 1981)? The inability to answer conclusively the above questions allows
for only semi-quantitative and equivocal modelling.
Assumptions
In view of the above discussion, the modelling of
Slamet andesitic magmas was based on the initial
premise that they are, for the most part, simple differentiates of the basalts, the origins of which were discussed
by Vukadinovic and Nicholls (1989). The major-element
trends and least squares-linear regression calculations
semi-quantitatively support crystal/liquid fractionation.
The primary evidence for the operation of magma
mixing at Gunung Slamet is the occurrence within phenocrysts of pervasive mineral inclusions with chemistry
indicating that they were precipitated from liquids more
evolved than that which produced the phenocrysts themselves. Many of the phenocrysts also show reverse zoning.
The preservation of mineral inclusions derived from
more evolved liquids and the lack of similar inclusions

0.704

I
45

50

55

60

65

sw)twt%

0.704

0.5

,
1

1.5

2.5

K2OWtlb
0.707

Fig. 13. SiOz, K,O, and U vs Sr/%r. Lack of correlation


suggests that the range of *Sr/%r ratios is not due to &ustal
assimilation. Symbols as in Fig. 12.

Evolution of G. Slamet Volcano, Indonesia


from less evolved liquids than those from which the host
phenocrysts precipitated implies that the more mafic
(and presumably hotter) endmember involved in mixing
was probably at or above its liquidus temperature. If a
magma at depth (e.g. basalt) rises adiabatically its
temperature may eventually rise above its liquidus in
P-T space. Mixing between a slightly superheated mafic
magma and a body of cooler, more salic magma would
give rise to mineral relationships, as seen in Slamet lavas.
Although *Sr/%Jr ratios for Slamet lavas span a
considerable
range, positive correlations
between
87Sr/86Sr and elements that are typically enriched in
continental crustal materials (e.g. SiO,, K,O, U) are
absent (Fig. 13). These types of positive correlations are
normally cited as evidence for assimilation of crustal
materials by magmas (e.g. Briqueu and Lancelot, 1979;
Thorpe et al., 1984; Graham and Hackett, 1987). Note
that the extremes of Sr-isotope compositions of Slamet
rocks are represented amongst the most mafic members,
with MgO > 7 wt%. Although this evidence does not
exclude the occurrence of minor crustal assimilation by
Slamet magmas, it implies that the range and trends of
isotope ratios and trace-element contents can be explained without invoking such processes. Consequently,
the evolution of Slamet magmas has been modelled
initially without invoking any AFC mechanisms.
Gerlach et al. (1988) described a similar situation with
lavas from the Puyehue-Cordon Caulle region, Chile.
Evolutionary routes for Slamet magmas
Simple crystal/liquidfractionation. Major- and minorelement variations (except MnO) for Slamet basaltbasaltic andesite-andesite
series with both coherent
87Sr/86Sr and incompatible trace element ratios (e.g.
Zr/K; Fig. 12) were modelled by crystal fractionation
using the least squares-linear regression program XLFRAC (Stormer and Nicholls, 1978). Fractionating
phases entered as input were restricted to those that are
observed as phenocrysts in either the parent or daughter
composition for each fractionating step. The phase
Table 1. XLFRAC
Parent
S61
SiOz
TiOz
Al,@
FeO*
MgO
CaO
Na,O
RzO

50.84
0.96
16.35
9.69
8.18
10.59
2.36
1.02
Zr = 0.0506
Parent
S62

SiOz
TiO,
Al,03
;$
CaO
Na,O
RIO

52.89
1.03
18.41

9.46
4.57
9.12
3.16
1.35
Cr = 0.0113

Daughter
s170
51.58
1.09
18.60
9.68
4.16
10.01
3.02
1.25
Daughter
s31
58.51
0.65
18.32
6.81
2.79
7.32
3.74
1.87

Calculated

compositions used in the modelling are the averages of


analyses obtained by electron microprobe from phenocryst cores in the parent or other similar rocks. On
occasion, phenocrysts from the daughter aomposition
were used if data from the parent were unavailable or the
mineral in question is not present in the parent.
In the literature, results from such modelling are
typically evaluated by the sum of the squares iof residuals
(W).
Modelling of Slamet rocks usually yielded
Cr2 < 0.25 and often x0.1. A further test to gauge
whether the XLFRAC results are realistic is to check if
all of the entered phases are in fact beings subtracted
from the parent magma. For example, ad XLFRAC
result requiring olivine + clinopyroxene + plagioclase
subtraction but signiJicant orthopyroxene #addition is
considered unlikely to represent the natural case (it may,
on the other hand, indicate that olivine is reacting with
liquid to form orthopyroxene). However, ,in cases in
which only small amounts of addition of a mineral were
necessary, the results were accepted so long as petrographic evidence allowed the possibility that ~the mineral
in question had just begun to crystallize. Initial crystallization may produce small grains, which-according
to
Stokes Law-may not be efficiently removed by gravitational settling, thus enabling minor accumulation to
occur while larger grains of other phases in more
advanced stages of crystallization are removed.
Major -element modelling. Tua lavas have constant
Zr/K values (N 8) but range in Sr/%r from 0.7057 to
0.7062 (Fig. 12). XLFRAC modelling was accomplished
in stepwise fashion beginning with mafic ~basalt S61
fractionating to S 170 (Mendala unit) to basaltic andesite
S62. In turn, XLFRAC was applied to S62as parental
to Sumbaga andesites S83, S9O and S3l. The resulting
solutions are excellent (Table l), with Xr* ~~0.051 in all
cases. Note that in deriving S83, S90 and S3~1from S62,
amphibole was used as a fractionating phase. The calculated proportions (w 611 wt%) exceed then abundance
of modal amphibole in these rocks. Due to significant
degassing at reduced pressures, much of th missing
amphibole could have broken down in shal fow magma

models of selected Tua lavas

Phase

Wt%

51.65
1.16
18.63
9.71
olv
7.07
4.78
cpx
9.09
10.01
Pl
1.oo
2.82
mt
-0.66
1.23
%xtls removed = 17.16
Calculated

153

Phase

58.50
0.56
18.32
cpx
6.82
opx
2.79
Pl
7.33
hb
3.79
mt
1.89
%xtls removed = 33.67

Wt%

3.21
2.30
13.42
11.14
3.60

Parent
s170
51.58
1.09
18.60
9.68
4.76
10.01
3.02
1.25
Xr =

Parent
S62
52.89
1.03
18.41
9.46
4.57
9.12
3.16
1.35
Cr =

Daughter
S62
52.89
1.03
18.41
9.46
4.57
9.12
3.16
1.35
0.0247
Daughter
s90
57.71
0.62
18.68
6.76
2.79
7.44
3.88
2.12
0.0142

Calculated

Phase

52.83
1.04
18.37
9.41
4.56
9.10
3.26
1.43
%xtls removed
Calculated

olv
cpx
Pl
mt

Wt%

0.78
4.35
8.77
1.19

= 15.09

Phase

57.72
0.65
18.68
cpx
6.76
opx
2.79
Pl
7.44
hb
3.94
mt
2.02
%xtls removed = 34.25

Wt%

4.52
4.92
15.54
5.76
3.51

D. Vukadinovic and I. Sutawidjaja

154

Table 2. XLFRAC
Parent
s154
SiOl
TiO,
A& 0,
FeO*
MgO
CaO
Na,O
R,O

50.26
1.11
16.40
10.16
7.71
10.49
2.98
0.89
Er = 0.1444
Parent
S161

Si02
TiO,
Al, 0,
FeO*
MgO
CaO
Na,O
R,O

54.20
1.10
18.79
8.75
3.56
8.38
3.70
1.51
Zr = 0.0624

Daughter
s39

Calculated

models of selected Lebaksiu lavas

Phase

Wt%

50.57
50.57
1.36
1.29
17.17
17.09
olv
10.53
10.46
cxP
6.58
6.58
PI
9.78
9.74
2.85
3.19
1.15
1.06
%xtls removed = 17.42
Daughter
s75
56.10
1.09
18.39
8.04
3.41
7.48
3.62
1.87

Calculated

Phase

56.06
1.01
18.35
cpx
oPx
8.02
PL
3.41
mt
7.49
3.85
1.84
%xtls removed = 20.02

3.29
7.28
6.85

Wt%

3.60
1.oo
13.20
2.22

chambers. This is consistent with the andesites erupting


as viscous, thick flows and the generally non-explosive
nature of Slamet andesitic volcanism. The major-element
trends in Tua intermediate lavas can be modelled without amphibole fractionation, yet the REE chemistry of
these rocks requires a fractionating phase that preferentially takes up mid-HREE. Partition coefficient REE
patterns satisfying this requirement may be found in the
literature for amphibole, clinopyroxene and apatite (e.g.
Nicholls and Harris, 1980; Watson and Green, 1981);
nevertheless, the latter two minerals are present in all
Slamet andesites, and clinopyroxene is present in nearly
all Slamet lavas, ranging from basalt to dacite. Since
amphibole is relatively abundant in Slamet Tua and
virtually absent from Muda lavas-except
in the Cendana hornblende andesite (which, importantly, is the
only low-87Sr/86Sr andesite with concave upwards
Table 3. XLFRAC
Parent
s112

Daughter
SlOl

50.45
1.31
15.72
10.36
7.70
10.51
2.88
1.06
Zr = 0.0210

51.03
1.46
16.73
10.30
5.98
9.90
3.36
1.26

Parent
Sl

Daughter
S156

SiOz
TiO,
Al, 0,
FeO*
MgO
CaO
Na,O
R,O

Si02
51.60
1.50
Ti02
A&Q
18.93
9.53
FeO*
4.28
MgO
9.46
CaO
3.30
Na20
1.39
R,O
Zr = 0.1407

55.44
1.14
18.43
8.41
3.43
7.81
3.63
1.71

Calculated

Daughter
S161

50.57
54.20
1.36
1.10
17.17
18.79
10.53
8.75
6.58
3.56
9.78
8.38
2.85
3.70
1.15
1.51
Zr = 0.0228
Parent
S161

Daughter
S167

Phase

Phase

Calculated

Phase

54.18
0.98
18.79
8.78
olv
3.55
cpx
8.39
PI
3.71
mt
1.60
%xtls removed = 29.89
Calculated

Phase

Wt%

6.66
9.09
11.35
2.79

Wt%

63.27
0.60
11.92
18.48
cpx
4.68
4.60
opx
39.89
1.39
PI
6.17
4.58
mt
3.99
2.43
%xtls removed = 62.66

54.20
63.39
0.59
1.10
18.79
18.51
4.71
8.75
1.36
3.56
4.64
8.38
4.03
3.70
2.76
1.51
Zr = 0.1410

HREE patterns)--the
simplest solution is to invoke
amphibole fractionation in the former and not in the
latter and to attribute depletion of mid-HREEs to
amphibole fractionation. Davidson et al. (1988) reached
similar conclusions regarding the cause of downbowing
of mid-HREEs in andesites from the San Pedro-Pellado
volcanic complex, Chile.
Lebaksiu basalts have narrow ranges of 87Sr/86Srand
Zr/K ratios. XLFRAC solutions for these rocks are
acceptable (none of the calculations have W > 0.25).
The solutions in Table 2 represent a stepwise transition
from basalt (S154 and S39) to basaltic andesite (S161) to
andesite (S75) and dacite (S167). Magnetite fractionation is not required in the early stages of magma
evolution (S154 to S39), but the later stages are dependent upon it to generate the low TiOz contents of the
andesites. Although they do not belong to the Lebaksiu

models of selected high-*Sr/?Sr

51.04
1.44
16.76
10.39
olv
5.97
cpx
9.92
PI
3.26
1.24
%xtls removed = 14.00
Calculated

Parent
s39

Wt%

3.49
7.50
3.01

Wt%

55.29
1.22
9.80
18.35
cpx
2.81
8.25
opx
25.04
3.46
Pl
3.77
7.72
mt
3.77
1.94
%xtls removed = 41.42

Parent
SlOl

Muda lavas

Daughter
s149

51.03
1.46
16.73
10.30
5.98
9.90
3.36
1.26

51.35
1.52
18.48
9.73
4.57
9.26
3.66
1.43
Zr = 0.0046

Parent
s29

Daughter
s25

54.32
58.28
1.18
0.57
18.68
18.87
8.80
6.67
3.64
2.54
8.45
8.06
3.50
3.55
1.45
1.44
Zr = 0.1086

Calculated

Phase

Wt%

51.31
1.53
18.47
1.02
9.73
olv
10.59
4.55
cpx
-0.56
9.28
Pf
3.71
1.42
%xtls removed = 11.61
Calculated

Phase

58.20
0.50
18.88
6.62
2.56
PI
1.89
hb
3.68
mt
1.66
%xtls removed = 21.07

Wt%

7.44
10.39
3.24

Evolution of G. Slamet Volcano, Indonesia


sequence, samples S161, S75 and S167 were modelled
from Lebaksiu parental magmas. Flat HREE patterns
indicate that amphibole played, at most, a minor role
in generating these three andesites-dacites and is in
accord with petrographic observations; consequently,
XLFRAC
calculations
were carried out without
amphibole fractionation.
XLFRAC results on high-87Sr/86Sr Muda lavas
(Fig. 12) are given in Table 3. Again, magnetite fractionation is not necessary in early stages of differentiation.
This result is consistent with petrographic observations;
a magnetite-free assemblage persists into compositions
with MgO as low as -4.5 wt% (SlOl-S149). Note
that the high Al,O, content of S149 requires minor
plagioclase accumulation in the SlOl-to-S149 step, and
that the Na,O content of S149 is too high to allow it to
be parental to other high-*Sr/@Sr Muda andesites
such as S 156 and S178; therefore, S 1 was selected as an
appropriate parental composition. Amphibole fractionation is not required to produce high-87Sr/86Sr Muda
andesites.
Deriving Cendana hornblende andesites (e.g. S25)
from Muda-type magmas has thus far been unsuccessful.
The XLFRAC example shown in Table 3 has reasonable
Cr* (~0.1); however, K,O is poorly reproduced. A
suitable parental magma to S25 (i.e. with similar *Sr/*jSr
and appropriate trace-element contents) was not found
during the course of this study.
Good overall XLFRAC results (low Zr*) were
obtained for low-87Sr/86Sr Muda lavas (Fig. 12).
Table 4 presents XLFRAC solutions showing stepwise
transitions from a mafic basalt (S104) to basalt (S117)
then branching in three directions towards a high-TiO,
basalt (S146), a low-TiO, basalt-basaltic andesite (S87),
and a basaltic andesite (S71). These results demonstrate
that the generation of a wide range of Ti02 contents in
the basalt to basaltic-andesite transition can be generated with variable magnetite crystallization. Conditions
for magnetite stability are not well understood, but it is
believed that magnetite appears on the liquidus only if the
j-0, of the magma is higher than the NNO buffer (cf. Gill,
Table 4. XLFRAC
Parent
s104

Daughter
s117

49.93
1.37
17.58
9.85
7.10
10.36
2.94
0.86
Zr = : 0.1214

SO.36
1.50
17.36
10.40
5.95
9.90
3.18
1.35

Parent
s117

Daughter
S87

30,

TiOz
A&Q
FeO*
MgO
CaO
Na,O
RIO

SiO,
TiO,
Al,03
FeO*
MgO
CaO
Na,O
RzD

50.36
1.50
17.36
10.40
5.95
9.90
3.18
1.35
Er = 0.0813

52.91
1.23
18.47
9.08
4.41
8.74
3.73
1.45

Calculated

1981, p. 197), implying that f0, values were generally


lower in primitive Muda magmas than in Tua magmas.
Trace-element modelling. Simple trace-element modelling of Slamet rocks was accomplished by utilizing the
results from XLFRAC to calculate crystal/liquid bulk
distribution coefficients (D) and proportions of residual
liquid (F) and then by applying these via the Rayleigh
Fractionation Law
C!,= C 1P-)

(Shaw, 1970).

(I)

C, and C,, represent the concentrations of element i in


the daughter and parental liquids, respectively. A range
of published mineral/melt partition coefficient values
(Kd), pertinent to the basalt-andesite
spectrum, was
used (Table 5). The following rules were used to determine the range of Kd values applicable to any particular
model: (1) if the MgO content of the daughber magma
being modelled was > 4 wt% (i.e. broadly basaltic), then
the Kd values used were those between the minimum and
median values of Table 5; (2) if the MgO content of the
daughter magma was <4 wt% (i.e. broadly andesitic),
then Kd values used were those between the median and
maximum values of Table 5.
Selected crystal fractionation results for trace elements
are presented graphically in abundance diagrams normalized to primitive-mantle values (Fig. 14). A range of
possible values is shown, illustrating the variety of results
that can be obtained by using published partitioning
coefficients. The wide range of D values for compatible
elements such as Cr and Ni did not allow for unique
interpretations, and these elements are not shown on the
graphs.
The results from this exercise are variable. Some
Slamet andesites can indeed be modelled from basalts by
crystal fractionation (Fig. 14a). However, andesites give
results in which the trace-element quantities of the
calculated liquids are consistently either too high or too
low (Fig. 14b); consequently, the generation of these
lavas requires either an alternative or a complementary
mechanism to crystal fractionation. The parallel but not
coincident trace-element
levels of calculated and

models of selected low-*7Sr/86Sr Muda lavas


Phase

50.52
1.65
17.36
10.29
olv
6.04
cpx
9.80
PI
3.21
1.15
%xtls removed = 22.66
Calculated

155

Phase

52.83
1.08
18.43
9.07
olv
4.38
cpx
8.75
Pl
3.81
mt
1.66
%xtls removed = 22.54

Wt%

Parent
Sll7

Daughter
S146

50.36
50.72
1.50
1.92
17.36
17.29
4.17
10.40
11.33
6.06
5.95
4.77
12.43
9.90
8.64
3.18
3.66
1.35
1.68
Zr = 0.0234
Wt%

Parent
s117

Daughter
s71

50.36
54.89
1.50
1.40
17.36
17.66
2.84
10.40
8.82
8.84
5.95
3.71
8.17
9.90
7.90
2.69
3.18
3.74
1.35
1.90
Zr = 0.1130

Calculated

Phase

Wt%

50.81
1.88
17.29
11.29
olv
2.28
4.73
9.97
cpx
PI
12.52
8.59
3.60
1.74
%xtls removed = 24.77
Calculated

Phase

Wr%

54.79
1.25
17.60
8.78
olv
4.43
3.68
11.67
cpx
7.91
Pl
19.26
3.96
mt
4.10
2.06
%xtls removed = 39.46

D. Vukadinovic and I. Sutawidjaja

156

Table 5. Mineral/melt partitioning coefficients used for modelling evolved compositions


Kds
olv(min)
olv(max)
plg(min)
plg(max)
cpx(min)
cpx(max)
opx(min)
opx(max)
mnt(min)
mnt(max)
amph(min)
amph(max)
ap(min)
ap(max)
Kds
olv(min)
olv(max)
plg(min)
plg(max)
cpx(min)
cpx(max)
opx(min)
opx(max)
mnt(min)
mnt(max)
amph(min)
amph(max)
ap(min)
ap(max)
Kds
olv(min)
olv(max)
plg(min)
plg(max)
cpx(min)
cpx(max)
opx(min)
opx(max)
mnt(min)
mnt(max)
amph(min)
amph(max)
ap(min)
ap(max)

cs
0.00043
0.05
0.0248
0.13
0.00035
0.64
0.00009
0.45
0.0001
0.08
0.05
0.5
0.00002
0.01
Y

Rb

Ba

Sr

Th

Zr

Hf

Nb

0.000179
0.04
0.01
0.2
0.001
0.04
0.0003
0.03
0.0001
0.47
0.05
1.9
0.00003
0.01

0.00011
0.03
0.02
0.59
0.001
0.3
0.0003
0.23
0.0001
0.4
0.08
6.4
0.01
0.03

0.000191
0.02
I.2
3.2
0.0014
0.21
0.01
0.1
0.0001
0.68
0.19
0.59
1.3
2

0.0001
0.04
0.002
0.06
0.0003
0.05
0.0002
0.22
0.008
0.44
0.005
0.15
0.46
0.46

0.0004
0.07
0.004
0.05
0.006
0.04
0.0009
0.22
0.02
0.55
0.017
0.25
0.94
1.3

0.0047
0.06
0.0094
0.03
0.05
0.36
0.02
0.22
0.02
1.7
0.08
1.789
0.005
0.636

0.0038
0.04
0.0092
0.03
0.05
0.36
0.02
0.22
0.02
1.7
0.13
1.731
0.015
0.73

0.0003
0.03
0.01
0.025
0.005
0.3
0.0005
0.35
0.4
I
0.1
1.3
0.002
0.636

Eu

Gd

DY

Ho

Yb

0.002
0.0308
0.0134
0.0454
0.195
2.2
0.0089
0.56
0.049
0.55
I.1
3.7
3.5
23

0.004
0.0468
0.0155
0.041
0.2
1.4
0.029
0.56
0.072
0.47
0.8
2.1
2.3
13.1

La

Sm

0.002
0.0308
0.01335
0.0454
0.195
2.2
0.0089
0.56
0.049
0.55
I.1
3.7
3.5
23

0.0005
0.008
0.0348
0.3017
0.02
0.4
0.0005
0.3
0.005
0.45
0.14
0.7219
2.5
11.5

SC

Cr

Ni

Zn

0.03
0.09
0.01
0.07
0.03
I8
0.025
7.2
0.11
67
6
45
0.01
0.01

0.3
34
0.01
0.08
1.9
245
2
143
1
620
0.04
90
0.048
0.2

4
58
0.01
0.25
I.5
II.7
I.1
24
1.4
77
0.5
I6
0.2
2.3

1.2
I.5
0.04
0.25
0.31
I2
2.6
4.4
3.1
I3
5
8.7
0.2
0.2

0.02
0.37
0.01
0.15
1.6
I7
0.53
7.5
0.8
3.3
6
I3
0.029
0.22

0.0019
0.0088
0.0132
0.1024
0.14
1.3
0.0028
0.43
0.009
0.55
0.8
2.6
5.5
29.3

0.0019
0.0096
0.0221
3.2
0.09
1.4
0.0036
0.42
0.007
0.42
0.83
2.95
1.3
31.5

0.0019
0.0108
0.0125
0.0665
0.18
1.7
0.0046
0.48
0.016
0.62
0.96
3.35
5
31

0.0019
0.0148
0.01 I2
0.0498
0.19
1.9
0.0072
0.56
0.038
0.58
1.15
3.7
3.7
25.6

Sources: Nagasawa (1973); Hart and Brooks (1974); Shimizu (1974); McCallum and Charette (1978); Pearce and
Norry (1979); Luhr and Carmichael (1980); Nicholls and Harris (1980); Watson (1980); Gill (1981); Villemant et al.
(1981); Watson and Green (1981); Shervais (1982); Day (1983); Irving and Frey (1984); Fujimaki et al. (1984); Ewart
and Hawkesworth (1987); Green and Pearson (1987); Watson et al. (1987); Green er al. (1989); Wyers and Barton
(1989). Kds are interchanged between Cs and Rb, Th and U, Zr and Hf, Y and Ho, and SC, V, Cr, Ni and Zn when
data is lacking for one of these elements. The minimum values are considered realistic for basalts, whereas the maximum
values are appropriate for andesite-dacites.

observed liquids (Fig. 14b) may bc quantitatively reproduced by periodic magma replenishment, tapping, and
fractionation.
Replenishment,

tapping

and fractionation

(RTF).

Steady-state RTF systematics, developed by OHara


(1977) and OHara and Matthews (1981), were expanded
upon by Cox (1988) by randomizing the amounts of
crystallization, eruption and replenishment in each cycle,
represented by the variables x, y and z respectively. Cox
found that randomized RTF can produce results from
successive lava flows that are opposite to that expected
from simple parent/daughter relationships. For example,
he was able to model two successive flows to have a

positive correlation for Zr vs Ni (incompatible vs compatible). However, if the entire sequence of flows were
plotted, the overall correlation is negative (as is
expected) but with pronounced scatter, very similar to
the Zr-Ni relationship for Slamet lavas (Fig. 15). As Cox
points out, there is no unequivocal evidence that RTF
processes take place in nature; but the idea is intuitively
acceptable, particularly in subduction environments
where there is ample proof that magmas rise and extrude
through specific points on the Earths surface for
relatively prolonged periods of time.
Calculations in this study were carried out on Microsoft EXCEL spreadsheets, which have random-number

Evolution of G. Slamet Volcano, Indonesia

157

16

Tua MgO

distrhtion

12
ave.MgO = 3.4wt%

n
0
4

E
E
P
E
5

Crystal Fractionation
s717tos71

MgOwt%

100

25
B

10

Mgo distliption
randomizeil
RTF

20

1J

15

Fig. 14. Selected results of simple crystal/liquid modelling


calculations. Stippled areas represent possible magma compositions generated by using the range of Kds as prescribed in the
text. Solid lines represent the observed daughter composition.
Normalizing values from Taylor and McLennan (1985). (A)
Sl61-S167 (cf. Table 2). (B) Sl17-S71 (cf. Table 4).

E 50
x - o.qs-0.40
y=o.10-0.15
zc2
MgO(initia/)
- 10 wt%
ave.MgtN= 3.3wt%

n
10
5
0 1
2

generator facilities. Fifty cycles of RTF were handled in


all calculations. This number of cycles was chosen simply
because larger numbers become unmanageable on
EXCELsspreadsheets when used on the Macintosh Plus
or SE@ computer. Francalanci er al. (1989) invoked
100,000 cycles to generate potassic lavas from calcalkalic
parents from Stromboli, Italy. However, their study used
equations that calculate only steady-state compositions
(OHara and Matthews, 1981) and, thus, could invoke
any number of cycles. Randomized RTF modelling
involves the calculation of the composition of the
erupted liquid from each cycle. Finally, the same set of
random numbers were used for each set of calculations
throughout the RTF modelling, allowing for meaningful
comparisons between models.
The abundance of i in the liquid at the start of each
cycle (i.e. after replenishment has taken place) is

MgOwt%

Fig. 16. (A) MgO distribution in Tua lavas. (B) M&


bution as modelled by randomized RTF.

a
distri-

determined by
cio

LciI(p)

l(p) C1

x(p)

Y(p)

cir(z)l+ Lt(p)
- Y,pl)
+ 21. (2)
(1 - X(P)

1 +

2s
20

Muda h4gOdistribution

15
n
aw.ugo = 4.7 wt%

10

a0 -

0
0

2
MgO

wt%

MgO

wt%

25
Mda
&IO distribution
via randomized
RTF
1s
07

50

x = 0.1 - 0.4

100

150

200

250

10

MgO(initiil)
= 9 wt

2rppm

Fig. 15. Ni vs Zr for Tua and Muda lavas. Symbols: closed


open squares =
squares = Tua; closed circles = Muda;
randomized RTF for Tua; open circles = randomized RTF for
Muda, solid lines = crystal fractionation
trend for Tua
(S6lSl6eS90,
DNi= 5, D, as in crystal fractionation modelling); dashed lines = crystal fractionation trend for Muda
(DNi = 5, D, as in crystal fractionation modelling).

ave.

MgO

- 5.0

wt

0
0

Fig. 17. (A) MgO distribution in Muda lava . (B) MgO


distribution as modelled by randomized $ TF.

158

D. Vukadinovic and I. Sutawidjaja


Table 6. Assemblages applied to randomized
RTF calculations

trace element (see Hanson and Langmuir, 1978). The


calculations assumed that if the MgO content of the
liquid prior to fractionation (i.e. C,) was >4 wt%, then
1
2a
2b
Stage
D MgO= 2.5; if MgO < 4, then D,, = 2. These values are
16
olv
in agreement with the average MgO levels of the frac38
60
39
plag
tionating assemblage derived by XLFRAC (Table 6) and
43
20
10
cpx
with the average MgO contents in corresponding Tua
7
opx
and
Muda rocks (Figs 16b and 17b). The value of z was
10
3
99
mnt
randomized to fall between 0 and 2 and was not varied
32
amph
2
2
from this range in any of the following calculations. C,,
Ego wt%
11.6
5.3
7.8
for MgO was set at 9 wt%-more
mafic than any
observed
sample
on
the
assumption
that
the replenishing
1, Magmas with >4 wt% MgO (both Tua
magma rarely reaches the surface (Cox, 1988). The
and Muda). 2a, Magmas with <4 wt% MgO,
variables x and y were randomized to fall within certain
amph-free (Muda). 2b, Magmas with G4 wt%
bounds until the calculated MgO distributions were
MgO, amph-bearing (Tua).
similar to those observed for Slamet Tua and Muda
(Figs 16 and 17). According to the models, Tua magma
C,,,, is the concentration of element i in the magma
chamber at the end of the previous cycle (i.e. after chambers erupted less liquid and crystallized more solid
fractionation and tapping have taken place); t(,) is the relative to Muda chambers.
Since the Tua system is dominantly andesitic, the
mass of the liquid at the start of the previous cycle; xcP)
maximum
Kd values from Table 5 were used to deterand ycPjare the mass proportions of removed solids and
mine
bulk
Ds. In the Muda RTF calculations, the
erupted liquid, respectively, from the previous cycle; C,
minimum
Kd
values were applied, for this system is
is the concentration of i in the replenishing magma; and
largely basaltic in nature. Within each set of calculations,
z is the mass proportion of the replenishing magma
two different fractionating assemblages (and, hence,
relative to the original mass of the magma chamber.
bulk Ds) were applied, dependent upon the MgO
The concentration
of i in the erupted liquid (i.e. after
content
of the liquid at the start of fractionation in
crystallization
had taken place) for any individual cycle
each cycle. If MgO > 4 wt%, then a gabbroic assemblage
was determined
by
was extracted. If MgO < 4 wt%, then an amphiboleCi,= C,(l -x)~I(Shaw, 1970).
bearing (Tua) or amphibole-free (Muda) dioritic frac(3)
tionating assemblage was used (Table 6). Apatite abundC, is the abundance
of i in the erupted liquid, equivalent
ance
in both dioritic assemblages was set at 2vol%,
in the next cycle [equation
(2)]; C, is the
to ci/@)
which is in rough agreement with the modal abundance
concentration
of i in the magma at the start of the cycle
of apatite in cognate crystal clots within the andesites.
in question [see equation (2)]; x is the mass proportion
of crystals removed relative to the mass of the liquid at
the start of the cycle; and Di is the bulk distribution
1000
Randomized RTF
A
coefficient of i in the crystallizing assemblage.
(Amphibole-bearing extract)
d
The values of x, y and z were constrained by attempting to duplicate the MgO distribution in Slamet Tua and
Muda lavas (Figs 16a and 17a) prior to handling the
trace element data. MgO was treated as a compatible
Table 7. Compositions of replenishing magmas

in randomized
calculations
Tua

RTF
1000

Muda

<

MgO
cs
Rb
Ba
Th
U
Nb
Sr
Zr
Hf
Y
La
Sm
&
DY
Ho
Yb

9
0.32
22
113
2.98
0.50
2.24
210
53
1.40
17
8.40
2.44
0.76
2.52
2.69
0.58
1.39

9
1.04
12
153
3.25
0.57
7.44
218
94
2.36
19
13.12
3.46
0.99
3.42
3.60
0.78
1.86

All values in ppm except for MgO


(wt%).

Randomized RTF
(Amphibole-free extract)

1J
a$j:f=SG=Jnrq~;Q&

29

Fig. 18. Mantle-normalized


diagrams of Slamet lavas with
4 < MgO < 5 wt% (closed circles = Muda; open circles = Tua;
open squares = Kalipagu) compared with the range of compositions calculated by randomized RTF (stippled areas) in which
(A) x = 0.25-0.40, y = 0.10-O. 15 and z = O-2 and in which the
Tua fractionating assemblage and replenishing magma are
used (cf. Tables 6 and 7); and (B) x = 0.10-0.40, y = 0.20-0.6
and z = O-2 and in which the Muda-fractionating
assemblage
and replenishing magma are used (cf. Tables 6 and 7). Normalizing values from Taylor and McLennan (1985). Note that the
two end-member calculations generally sandwich all observed
compositions.

Evolution of G. Slamet Volcano, Indonesia


1000
B

A
I

Radamized RTF
(Amphibokwbewingextract)

RandomizedRTF
(Amphibobfree extract)
3bMgo<4wt96

8fd~=qc5nE~3$3S&P~

Fig.

19. As

in Fig. 18, but for lavas with 3 < MgO < 4 wt%.

The concentration of i in the replenishing magma is


normally set constant (Cox, 1988; Table 7). In the Tua
calculations, C, for each element was arbitrarily set at
80% of the value of the trace element in S61. In the
Muda system, 80% of the minimum value of trace
elements in Muda rocks with MgO > 7 wt% was used,
again on the assumption that the replenishing magma is
less evolved than those that reach the surface.
The calculated and observed trace-element levels were
compared by taking the maximum and minimum calculated values within MgO increments of 1 wt% and
superimposing the observed trace-element contents from
rocks that fall within the specified MgO range. Selected
data are plotted on trace-element abundance diagrams
normalized by primitive-mantle estimates, with the
REEs grouped together to allow easier monitoring of
their behavior (Figs 18 and 19). The two end-member
calculations sandwich the observed lava compositions.
Another aspect investigated was the distinct relationships that Tua and Muda lavas have in Ni vs Zr plots
(Fig. 15). In the RTF models, Zr was calculated as
described above; Ni was modelled as above but with a
bulk D arbitrarily set to 5 (regardless of MgO value)
and with 125 ppm in the replenishing magma for both
the Tua and Muda systems. The randomized RTF
models not only duplicate the distinct trends in Ni vs Zr
between Tua and Muda magmas, but also mimic
the scatter that is seen between these elements (Fig. 15).
Discussion

Both crystal fractionation and RTF modelling are


generally successful in reproducing the trace-element
characteristics of Slamet lavas. However, a number of
reasons exist for favoring RTF-type processes at Slamet.
First, as previously mentioned, the mineralogical observations are not compatible with simple crystal fractionation alone, but RTF processes--in
which the
replenishing magma is relatively phenocryst free-are in
accord with the relationships between phenocrysts and
their inclusions. Second, decoupling of compatible and
incompatible trace elements, which is observed at Slamet
(Fig. 15), is easily explained by RTF processes. Although

159

simple crystal/liquid fractionation can produce the same


genera1 trends, the range of possible liquids is confined
to relatively narrow bands if DNiis constant; on the other
hand, randomized RTF approaches the natural case and
produces the geologic noise that so often accompanies
plots of two elements with markedly differlent bulk D
values. Third, the ability of randomized RilF to reproduce the MgO distributions of Slamet Tua apd Muda is
inherently appealling and worthy of considqration.
If the RTF modelling approaches what has actually
taken place at Slamet, why does the Tua system require
more fractionation and less tapping than the Muda
system? One possibility is that since Slamet Tua activity
represents the first phase of volcanism in the recent past
at Slamet, magmas rising into relatively cool crust may
undergo enhanced crystallization and become more viscous, resulting in liquids that are less eruptible. The
Muda system may then have established itsielf in crust
that was significantly warmed by Tua magmiatism, thus
promoting the eruption of dominantly baslaltic lavas.
However, Gill (198 1) maintained that ande+te-producing volcanoes usually erupt increasingly fel.$c material
with time, but there are many exceptions to this rule.
Thus, the situation at Slamet, where initial andesitic
volcanism is followed by basaltic volcanism, does not
appear to apply to other arc volcanoes and casts doubts
on the above suggestion.
Alternatively, Muda magmatic activity mag have produced pulses of replenishing magma that were separated
by relatively short periods of time compared with that of
Tua. In this scenario, Muda magma chambers would
never fractionate extensively before being recharged by
hot, mafic, fluidal magma. It is also likely that such
a magma chamber would be more eruptible than one
that was allowed to stagnate for significant lengths of
time, as may have been the case with Tua magma
chambers.
Excepting the Cendana andesite, magma chambers at
Slamet were largely open to influxes of magrtja (particularly during the period of Muda activity), but not to
significant crustal assimilation. Magma chaabers were
replenished by parental magmas with origin, similar to
that of the magma already residing in the chamber. In
other words, Tua parental magmas did ndt mix and
replenish magma chambers bearing Muda lavas, or vice
versa.

The inability to satisfactorily mode1 Cendana hornblende andesites is a reflection of their unique petrography. A suitable parent to produce the Cendaha andesite
via simple crystal/liquid fractionation is not possible
from the data available at present. Potential p&rents with
suitable isotopic values (e.g. Muda lavas with
87Sr/86Sr= 0.7045) have trace-element levels t$at are too
high to produce S25. For example, trace-eledent values
were calculated for a liquid generated by frqctionation
from S29 to S25 (Table 3), and were too high in Zr and
Y by a factor of 1.4-2.2 and 1.6-2.7, respectively.
Similarly, binary mixing models between relatively
high- and 10w-*7Sr/86Sr magmas (i.e. Tua and
Muda) fail to reproduce Zr/K ratios and abupdances of
Nb, Zr, and Y. The isotopic characteristics oti Tua lavas
render them unsuitable as parents to Cendanal andesites.
The relatively low 87Sr/86Srratios and low trabe-element
contents of the Cendana andesites prohibit deliivation by
crustal assimilation or AFC processes from any of the
Slamet lavas.

D. Vukadinovic and I. Sutawidjaja

160

The origin of the Cendana andesites can be explained


adequately only by invoking parental magmas that
have not yet been sampled, exposed or erupted. The
desired characteristics of such a parental liquid would be
to have trace-element quantities compatible with Tua
lavas but isotopic values resembling high-87Sr/*6Sr
Muda lavas. Clearly, more field work in the Cendana
area and further isotopic and trace-element determinations are required.

Summary
The date of onset of volcanism at Gunung Slamet
volcano is presently unknown. Although andesiteproducing volcanoes usually erupt increasingly felsic
material with time (Gill, 1981), the situation at Slamet
is reversed. The first period of volcanism gave rise to
the Tua sequence, composed primarily of andesitic
lavas and associated pyroclastic deposits. The second,
ongoing period of volcanic activity has been dominated
by basaltic lava flows, known as the Muda sequence. The
basaltic Lebaksiu sequence appears to be a transitional
phase between Tua and Muda rocks.
Slamet basalts are characterized by phenocrysts of
plagioclase, olivine and clinopyroxene. The appearance
of orthopyroxene phenocrysts coincides with a transition
from the intergranular/intersertal
textures of the basalts
to the pilotaxitic/hyalo-ophitic
textures of the andesites.
Clinopyroxene
and plagioclase
remain abundant
throughout the basalt-andesite spectrum. Rocks that
contain both orthopyroxene and olivine phenocrysts
usually fall into the basaltic-andesite field of Peccerillo
and Taylor (1976), and rocks that contain orthopyroxene but no olivine are generally andesites. Magnetite,
although present as a groundmass phase in virtually
all of the examined rocks, occurs as a phenocryst phase
in rocks near the basalt to basaltic-andesite transition or
in those more evolved. Amphibole occurs predominantly
in Tua rocks. Rare amphibole occurs in a few Muda
basalts and basaltic andesites and in the Keruh dacite.
On the basis of stratigraphy, incompatible traceelement abundance, Zr/Nb, Zr/K and 87Sr/86Sr,Slamet
rocks can be grouped into those corresponding to high
and low abundance magmas (HAM and LAM, respectively). Rocks from the Muda sequence belong to the
HAM group, and those from the Tua and Lebaksiu
sequences belong to LAM. The LAM suite is older,
generally more evolved, and has higher *SrlB6Srratios
than the HAM suite.
Vukadinovic and Nicholls (1989) suggested that the
source magmas for both of these groups can be derived
from mantle-wedge material that is similar in incompatible trace-element abundance to that of transitional
MORB sources and that decoupling between lithophile
elements and Sr/Sr is possible in the source region if
greater amounts of slab-derived fluid or melt promoted
greater degrees of melting of the overlying mantle wedge.
The resulting magmas would then have high values of
87Sr/86Srbut relatively low levels of lithophile elements,
due to dilution by larger degrees of melting.
However, the differences in SiOz and MgO contents
between Tua and Muda lavas cannot be accounted for
by magma-source systematics. The available geochemical data are compatible with magmatic evolution at
Slamet being driven by either simple crystal fraction-

ation or processes involving magma replenishment, fractionation and tapping (RTF). In either case, crystal
fractionation has played a significant role, especially in
generating Tua magmas. In addition, if RTF processes
were in operation then the periods of time between
replenishments within HAM magma chambers may have
been shorter compared with those of LAM, thus helping
to promote the generally less-evolved, basaltic nature of
HAM volcanism. The generation of some HAM basalts
and basaltic andesites with relatively high TiOz contents
(> 1.6 wt%) can be explained by suppressed magnetite
crystallization. In LAM (or Tua) magma chambers,
amphibole fractionation is probably responsible for the
concave-upward HREE patterns in Tua rocks with
MgO < 4 wt%.
Some rock types at Gunung Slamet (i.e. Cendana
andesite, Keruh dacite, Kalipagu andesite) do not fall
within either the LAM or HAM suite. The Cendana
andesite has trace-element levels and ratios similar to
those of LAM lavas but 87Sr/86Srsimilar to those of
HAM. A parental magma unlike any other sampled is
required for the Cendana andesites. Kalipagu andesites
and Keruh dacites can be derived from parent magmas
similar to those from the Lebaksiu sequence.
Although 87Sr/86Srratios in some Slamet rocks are
relatively high, suggesting the involvement of crustal
material, the lack of positive correlation between incompatible lithophile element levels (e.g. K and U) and
87Sr/86Srimplies that upper-level crustal contamination
was of no more than minor significance.
Acknowledgements-Australian Research Council grants
held by Dr I. A. Nicholls and Monash University
Graduate and Departmental Scholarships held by D.V. are
gratefully acknowledged. Logistical support for this project
was provided by the Volcanological Survey of Indonesia.
Reviews by Drs U. Knittel, I. Nicholls, H. Pichler, and M.
van Bergen improved the manuscript.
Drafting of the map
was aided by MS D. Gelt and Mr P. Guzina.
Drs B.

Gulson and D. J. Whitford (CSIRO, Australia) and Dr S. R.


Taylor (ANU, Australia) provided access to their respective
analytical facilities. The senior author also acknowledges the
role of the Indonesian Institute of Sciences (LIPI) in this
project.

REFERENCES
Aswin D., Sutawidjaja I. S., Sitorus K., Dana I. N. and Wahyudin D.
(1984) Geological mapping of Gunung Slamet, Central Java.
Unpub. Report II. Geological Mapping Division, Volcanological
Survey of Indonesia.
Basaltic Volcanism Study Project (BVSP) (1981) Basaltic Volcanism
on the Terrestrial Planets. Pergamon, New York.
Blattner P. and Reid F. (1982) The origin of lavas and ignimbrites of
the Taupo Volcanic Zone, New Zealand, in the light of oxygen
isotope data. Geochim. Cosmochim. Acta 44, 1417-1429.
Bowen N. L. (1928) The Evolution of the Igneous Rocks. Princeton
University Press, Princeton, NJ.
Briqueu L. and Lancelot J. R. (1979) Rb-Sr systematics and crustal
contamination models for talc-alkaline igneous rocks. Earth Planet.
Sci. Left. 43, 385-396.

Brophy J. G. and Marsh B. D. (1986) On the origin of high alumina


arc basalt and the mechanics of melt extraction. J. Petrol. 27,
763-789.

Cameron M. and Papike J. J. (1981) Structural and chemical variations


in pyroxenes. Am. Mineral. 66, l-50.
Cann, J. R. (1970) Rb, Sr, Y, Zr and Nb in some ocean-floor basaltic
rocks. Earth Planet. Sci. Lerr. 10, 7-1 I.

Evolution of G. Slame !t Volcano, Indonesia


Carmichael I. S. E. (1967) The iron-titanium oxides of salic volcanic
rocks and their associated ferromagnesian silicates. Contrib. Mineral. Petrol. 14, 3664.

Cas R. A. F. and Wright J. V. (1987) Volcanic Successions-Ancient


and Modern. Allen & Unwin, London.
Conrad W. K., Nicholls I. A. and Wall V. J. (1988) Water-saturated
and undersaturated
melting of metaluminous and peraluminous
crustal compositions at IO kb: evidence for the origin of silicic
magmas in the Taupo volcanic zone, New Zealand, and other
occurrences. J. Petrol. 29, 765-803.
Cox K. G. (1988) Numerical modelling of randomized RTF magma
chamber: a comparison with continental flood basalt sequences.
J. Petrol. 29, 681-697.
Davidson J. P., Ferguson K. M., Colucci M. T. and Dungan M. A.
(1988) The origin and evolution of magmas from the San PedroPellado volcanic complex, S. Chile: multicomponent sources
and open system evolution. Contrib. Mineral. Petrol. 100,
429-445.

Day R. A. (1983) Petrology and geochemistry of the older volcanics,


Victoria. Ph.D. Thesis, Monash University, Melbourne.
Deer W. A., Howie R. A. and Zussman J. (1966) An Introduction to
the Rock-Forming Minerals. Longman, New York.
De Paolo D. J. (1981) Trace element and isotopic effects of combined
wallrock assimilation and fractional crystallization. Earth Planet.
SC;. Lett. 53, 189-202.
Djuri M. (1975) Geologic Maps qf the Purwokerto and Tegal Quadrangles, Java. Geological Survey of Indonesia, Bandung.
Ewart A. (1976) Mineralogy and chemistry of modern erogenic
lavas-some statistics and implications. Earth Planet. Sci. Left. 31,
417432.

Ewart, A. and Hawkesworth C. J. (1987) The Pleistocene-Recent


Tonga-Kermadec arc lavas: interpretation of new isotopic and rare
earth data in terms of a depleted mantle source model. J. Petrol. 28,
495-530.

Ewart A. and Stipp J. J. (1968) Petrogenesis of the volcanic rocks of


the central North Island, New Zealand, as indicated by a study of
Sr/%r ratios, and Sr, Rb, K, U and Th abundances. Geochim.
Cosmochim. Acta 32, 6999736.

Ewart A., Taylor S. R. and Capp A. C. (1968) Trace and minor


element geochemistry of the rhyolitic volcanic rocks, central
North Island, New Zealand. Contrib. Mineral. Petrol. 18,
76-104.

Francalanci L., Manetti P. and Peccerillo A. (1989)Volcanological and


magmatological evolution of Stromboli volcano (Aeolian Islands):
the roles of fractional crystallization, magma mixing, crustal
contamination
and source heterogeneity. Bull. Volcanol. 51,
355-378.

Fujimaki H., Tatsumoto M. and Aoki K. (1984) Partition coefficients


of Hf, Zr, and REE between phenocrysts and groundmasses. Proc.
14th Lunar Planet. Sci. ConjI, B662-B672.
Gast P. W. (1968) Trace element fractionation and the origin of
tholeiitic and alkaline magma types. Geochim. Cosmochim. Acta 32,
1057-1086.

Gerlach D. C., Frey F. A., Morenoroa H. and Lopez-Escobar L.


(1988) Recent volcanism in the PuyeheCordon
Caulle region,
southern Andes, Chile (40.5s): Petrogenesis of evolved lava%
J. Petrol. 29, 333-382.

Gill J. B. (1974) Role of underthrust oceanic crust in the genesis


of a Fijian c&-alkaline suite. Contrib. Mineral. Petrol. 43, 2945.
Gill J. B. (1978) Role of trace element partition coefficients in models
of andesite genesis. Geochim. Cosmochim. Acta 42, 709724.
Gill J. B. (1981) Orogenic Andesites and Plate Tectonics. Springer,
Berlin.
Graham I. J. and Hackett W. R. (1987) Petrology of talc-alkaline lavas
from Ruapehu volcano and related vents, Taupo Volcanic Zone,
New Zealand. J. Petrol. 28, 531-567.
Green T. H. and Pearson N. J. (1987) An experimental study of Nb
and Ta partitioning between Ti-rich minerals and silicate liquids at
high pressure and temperature. Geochim. Cosmochim. Acra 51,
5562.

Green T. H. and Ringwood A. E. (1968) Genesis of the calcalkaline


igneous rock suite. Contrib. Mineral. Petrol. 18, 105-162.

161

Green T. H., Sie S. H., Ryan C. E. and Cousens D. R. (1989) Proton


microprobe-determined
partitioning of Nb, Ta, Zr, Sr and Y
between garnet, clinopyroxene and basaltic magma at high pressure
and temperature. Chem. Geol. 74, 201-216.
Grove T. L., Gerlach D. C. and Sando T. W. (1,982) Origin of
talc-alkaline series lavas at Medicine Lake Volcano by fractionation, assimilation and mixing. Conrrib. Mine&. Petrol. 80,
160-182.

Grove T. L., Kinzler R. J., Baker M. B., Donnelly-Nolan


Lesler C. E. (1988) Assimilation of granite by basaltic
Burnt Lava flow, Medicine Lake volcano, northern
decoupling of heat and mass transfer. Conrrib. MineraL

J. M. and
magma at
California:
Petrol. 99,

320-343.

Hamilton W. (1979) Tectonics of the Indonesian region. US. Geol.


Surv., Prof. Pap. 1078.
Hanson G. N. and Langmuir C. H. (1978) Modelling of major
elements in mantle-melt systems using trace element approaches.
Geochim. Cosmochim. Acta 42, 725-141.

Hart S. R. and Brooks C. (1974) Clinopyroxene-matrix partitioning


of K, Rb. Cs, Sr and Ba. Geochim. Cosmochim. Acta 38,
179991806.

Hofmann A. W. (1988) Chemical differentiation of the Earth: the


relationship between mantle, continental crust, and oceanic crust.
Earth Planet. Sri. Lett. 90, 297-3 14.

Hole M. J., Saunder A. D., Marriner G. F. and Tarney J. (1984)


Subduction of pelagic sediments: implications for the origin of
Ce-anomalous basalts from the Mariana Islands. J. Geol. Sot. Lond.
141, 452472.

Irvine T. N. and Baragar W. R. A. (1971) A guide tie the chemical


classification of the common volcanic rocks. Can. 1. Earth Sci. 8,
523-548.

Irving A. J. and Frey F. A. (1984) Trace element abundances


in megacrysts and their host basahs: Constraints on partition
coefficients and megacryst genesis. Geochim. Cosmochim. Acta 48,
1201-1221.

Jakes P. and Gill J. (1970) Rare earth elements and the island arc
tholeiitic series. Earth Planet. Sci. Lett. 9, 17-28.
Kuno H. (1968) Origin of andesite and its bearing on the island arc
structure. Bull. Volcanol. 32, 141-176.
Leake B. E. (1978) Nomenclature of amphiboles. Can. Mineral. 16,
501-520.

Leo G. W., Hedge C. E. and Marvin R. F. (1980) ,Geochemistry,


strontium isotope data, and potassium-argon ages of the andesiterhyolite association in the Padang Area, West Sumatra. J. Volcanol.
Geotherm. Res. 7, 139-156.
Le Roex A. P. (1987) Source regions of mid-ocean, ridge basalts:
evidence for enrichment processes. In Mantle Metasomatism (Edited

by M. A. Menzies and C. J. Hawkesworth), pp. 389422. Academic


Press, London.
Lesher C. E. (1986) Effects of silicate liquid composition of mineralliquid element partitioning from Soret diffusion stud&. J. Geophys.
Res. 91, 61236141.
Luhr J. F. and Carmichael I. S. E. (1980) The Colima volcanic
complex, Mexico. 1. Post-caldera andesites from Volcan Colima.
Conrrib. Mineral. Petrol. 71, 343-372.

Mahood G. and Hildreth W. (1983) Large partition coefficients for


trace elements in high-silica rhyolites. Geochim. Cosmochim. Acta
47, 11-30.
Marsh B. D. (1976) Mechanism of Benioff zone magmatism. Am.
Geophys. Union. Monogr. 19, 337-350.

Marsh B. D. (1978) On the cooling of ascending andesitie magma. Phil.


Trans. R. Sot. Lond. A288, 611-625.
Marsh B. D. and Carmichael I. S. E. (1974) Benioff zone magmatism.
Geophys. Res. 79, 11961206.
McCallum I. S. and Charette M. P. (1978) Zr and! Nb partition
coefficients: implications for the genesis of Mare basahs, KREEP,
and sea floor basal& Geochim. Cosmochim. Acta 44, 859-869.
Miyashiro A. (1974) Volcanic rock series in island arcs and active
continental margins. Am. J. Sci. 274, 321-355.
Nagasawa H. (1973) Rare earth concentrations in zircons and apatites
and their host dacites and granites. Earfh Planet. Sci. Lelf. 9,
359-364.

162

D. Vukadinovic and I. Sutawidjaja

Nash W. P. and &craft H. R. (1985) Partition coefficients for trace


elements in silicic magmas. Geochim. Cosmochim. Acta 49,
2309-2322.

Neumann van Padang M. (1936) Over de verplaattsing van de kraters


der vulkanen Slamet, Lamongan, Merapi en Semeroe. De Ingenieur
in Nederlands Indie 1, Id.
Neumann van Padang M. (1951) Catalogue of the active volcanoes of
the world, part I, Indonesia. IAVCEI.
Nicholls I. A. and Harris K. L. (1980) Experimental rare earth element
partition coefficients for garnet, clinopyroxene and amphibole
coexisting with andesitic and basaltic liquids. Geochim. Cosmochim.
.4cta 44, 287-308.

OHara M. J. (1977) Geochemical evolution during fractional crystallisation of a periodically refilled magma chamber. Nature 266,
5033507.
OHara M. J. and Mathews R. E. (1981) Geochemical evolution of an
advancing, periodically replenished, periodically trapped, continuously fractionated magma chamber. J. Geol. Sot. Land. 138,
237-277.
Osborn E. F. (1959) Role of oxygen pressure in the crystallisation
and differentiation
of basaltic magma. Am. J. Sci. 257,
609-647.
Osborn E. F., Watson E. B. and Rawson S. A. (1978) Composition of
magnetite in subalkaline volcanic rocks. Carnegie Inst. Wash.
Yearb. 78, 475481.

Papike J. J., Cameron K. L. and Baldwin K. (1974) Amphiboles and


pyroxenes: characterization of Other than Quadrilateral components and estimates of ferric iron from microprobe data. Geol.
Sot. Amer. Abstr. Progs 6, 1053-1054.

Pardyanto L. (1971) Penafsiran potret-udara daerah G. Slamet dan


sekitarnja (a commentary on air-photos of the area of Mount
Slamet). Unpublished report, Geological Survey of Indonesia,
Bandung.
Patchett P. J. (1980) Thermal effects of basalt on continental
crust and crustal contamination
of magmas. Nature 283,
559-561.
Peacock, M. A. (1931) Classification of igneous rock series. J. Geol. 39,
5467.
Pearce J. A. and Norry M. J. (1979) Petrogenetic implications of Ti,
Zr, Y and Nb variations in volcanic rocks. Conrrib. Mineral. Petrol.
69, 3347.

Trans. R. Sot. Land. A310, 675-692.

Tsuchiyama A. and Takahashi E. (1983) Melting kinetics of a plagioclase feldspar. Contrib. Mineral. Petrol. 84, 345-354.
Van Bemmelen R. W. (1949) The Geology of Indonesia, 2nd edn.
Martinus Nijhoff, The Hague.
Villemant B., Jaffrezic H., Joron J.-L. and Treuil M. (1981)
Distribution coefficients of major and traae elements; fractional
crystallization in the alkali basalt series of Chaine des Puys (Massif
Central, France). Geochim. Cosmochim. Aem 45, 1997-2016.
Vukadinovic D. (1989) The petrology and geochemistry ofsubductionrelated rocks from Gunung Slamet Valcano, Central Java.
Indonesia. Unpublished Ph.D. thesis, Monash University, Clayton.
Vukadinovic D. (1993) Are Sr enrichments in arc basalts due to
plagioclase accumulation? Geology 21, 6 I l-6 14.
Vukadinovic D. and Nicholls I. A. (1989) The petrogenesis of
island arc basalts from Gunung Slamet volcano, Indonesia: trace
element and *Sr/%r constraints. Geochim. Cosmochim. Acta 53.
2349-2363.

Wager L. R. and Brown G. M. (1967) Layered Igneous Rocks.


Freeman, San Francisco, CA.
Wager L. R. and Deer W. A. (1939) Geological investigations in
East Greenland. Part III. The petrology of the Skaergaard intrusion, Kangerdlugssuaq, East Greenland. Medd. om Grrmland 105.
t-352.
Watson E. B. (1976) Two-liquid partition coefficients: experimental
data and geochemical implications. Confrib. Mineral. Petrol. 56,
119.--134.

Watson E. B. (1980) Some experimentally determined zircon/liquid


partition coefficients for the rare earth elements. Geochim. Cosmochim. Acta 44, 8955897.

Watson E. B. and Green T. H. (1981) Apatite/liquid partition


coefficients for the rare earth elements and strontium. Earth Planet.
Sci. Lerf. 56, 405421.

Peccerillo A. and Taylor S. R. (1976) Geochemistry of Eocene


talc-alkaline volcanic rocks from the Kastamonu area, northern
Turkey. Contrib. Mineral. Petrol. 58, 63-81.
Porter S. C. (1972) Distribution, morphology and size frequency of
cinder cones on Mauna Kea volcano, Hawaii. Geol. Sot. Amer. Bull.
83, 3607-3612.

Ryerson F. J. and Hess P. C. (1978) Implications of liquid-liquid


distribution coefficients to mineral-liquid partitioning. Geochim.
Cosmochim. Acta 42, 921-932.

Shaw D. M. (1970) Trace element fractionation

Sutawidjaja I. S., Aswin D. and Sitorus, K. (1985) Geologic Map OJ


Slamet Volcano, Central Java. Volcanological Survey of Indonesia,
Bandung.
Taylor S. R. and McLennan S. M. (1985) The Continemal Crust: its
Composition and Evolurion. Blackwell, Oxford.
Thorpe R. S., Francis P. W. and OCallaghan L. (1984) Relative roles
of source composition, fractional crystallization and crustal contamination in the petrogenesis of Andean volcanic rocks. Phil.

during anatexis.

Geochim. Cosmochim. Acta 34, 237-243.

Shervais J. W. (1982) Ti-V plots and the petrogenesis of modern and


ophiolitic lavas. Earth Planet. Sci. Left 59, 101-118.
Shimizu H. (1974) An experimental study of the partitioning of K, Rb,
Cs, Sr and Ba between clinopyroxene and liquid at high pressures.
Geochim. Cosmochim. Acra 38, 178991798.

Sparks R. S. J. and Pinkerton H. (1978) Effects of degassing on


rheology of basaltic magma. Nature 276, 385-386.
Stormer J. C. Jr and Nicholls J. (1978) XLFRAC: a program for the
interactive testing of magmatic differentiation models. Camp.
Geoscl. 4, 1433159.
Sutawidjaja I. S. (1988) The cinder cones of Mount Slamet, Central
Java, Indonesia. Unpublished Diploma of Applied Science thesis,
Victoria University, Wellington.

Watson E. B., Ben Othman D., Luck J. M. and Hofmann A. W. (1987)


Partitioning of U, Pb, Cs, Yb, Hf, Re and OS between chromium
diopsidic pyroxene and haplobasaltic liquid. Chem. Geol. 62.
191.--208.

White W. M. and Patchett J. (1984) Hf-Nd-Sr isotopes and incompatible element abundances in island arcs: Implications for magma
origins and crust-mantle evolution. Earth Planet. Sci. Left. 67.
1677185.

White W. M., Dupre B. and Vidal P. (1985) Isotope and trace element
geochemistry of sediments from the Barbados Ridge-Demerara
Plain region, Atlantic Ocean. Geochim. Cosmochim. Acta 49.
187551886.

Whitford D. J. (1975a) Geochemistry and petrology of volcanic rocks


from the Sunda Arc, Indonesia. Ph.D. Thesis. Australian National
University, Canberra.
Whitford D. J. (1975b) Strontium isotopic studies of the volcanic rocks
of the Sunda arc, Indonesia, and their petrogenetic implications.
Geochim. Cosmochim. Acta 39, 128771302.

Whitford D. J. and Jezek P. A. (1979) Origin of late-Cenozoic lavas


from the Banda arc, Indonesia: Trace element and Sr Isotope
evidence. Contrib. Mineral. Petrol. 68, 141-150.
Wyers G. P. and Barton M. (1989) Polybaric evolution of talc-alkaline
magmas from Nisyros, southeastern Hellenic Arc, Greece. J. Petrol.
30, I-37.

163

Evolution of G. Slamet Volcano, Indonesia

Appendix
Selected chemical analyses of Slamet rocks
MgO > 6 wt%
HAM
Sample

SlO8

SiOz
TiOz
A&O,
FeO*
MnO
MgO
CL1
NazO
K,O
Rh
Sr
Zr
Y
SC
V
Cr
Nl
Zn
LU
Cc
PINd
Sm
El1
Gd
Tb

50.7
1.25
15.7
10.1
0.19
7.70
10.5
2.66

DY
Ho
Er
Yb
B>l
U
Th
Hf
Nb
87/86Sr

I .oo
26
279
130
27
36
262
240
57
87
n.d.
31.1
n.d.
17.3
4.33
I .24
4.74
n.d.
4.50
n.d.
2.64
2.43
191
n.d.
n.d.
n.d.
n.d.
0.70534

s112
50.3
1.30
15.7
10.3
0.19
7.67
10.5
2.87
I .06
24
273
133
24
35
260
246
57
85
16.7
37.9
4.35
19.3
4.54
1.29
4.28
0.72
4.61
0.98
2.82
2.33
212
0.87
4.08
3.07
12.6
0.70541

LAM
s104
49.7
1.36
17.5
9.80
0.19
7.07
10.3
2.93
0.86
18
361
117
27
34
277
108
53
84
16.4
37.6
4.14
18.8
4.77
1.44
5.08
0.87
4.80

1.oo
2.79
2.41
266
0.71
4.06
2.95
9.30
0.70512

72-1015

S61

49.7
I .49
17.1
10.5
0.21
6.01
10.0
3.23
1.34
29
294
I48
32
57
390
100
24
n.d.
15.7
35.9
4.53
20.5
5.07
1.45
4.88
0.83
5.11
1.08
2.83
2.61
201
0.88
4.47
3.39
12.5
0.70491

50.7
0.96
16.3
9.67
0.18
8.16
10.6
2.36
I .02
27
279
66
21
38
283
252
73
84
10.5
23.7
3.02
14.4
3.05
0.95
3.15
0.57
3.36
0.72
I .92
1.74
141
0.63
3.73
1.75
2.80
0.70623

Sl54

S36

50.1
1.11
16.4
10.1
0.19
7.69
10.5
2.97
0.89
22
287
86
23
33
283
195
69
83
9.8
22.6
2.89
12.9
3.19
1.oo
3.34
0.62
3.60
0.75
2.09
1.79
174
0.55
2.62
2.24
5.80
0.70565

50.9

1.34
16.2
10.2
0.19
6.80
9.58
3.08

1.47
36
288
112
24
32
306
232
68
89
15.3
35.0
4.13
18.5
4.50
1.27
3.97
0.66
4.04
0.87
2.38
2.26
292
1.11
5.13
2.80
6.60
0.70589

6>MgO>4wt%
HAM
Sample

Sll7

SiOz
TiO,
A&O,
FeO*
MnO
MgO
CaO
Na,O
K,O
Rb
Sr
Zr
Y
SC
V
Cr
Ni
Zn
La

50.1
1.49
17.3
10.3
0.20
5.91
9.84
3.16
1.34
32
326
177
32
34
264
118
39
96
20.2

S116
50.4

1.62
18.3
9.71
0.17
4.88
9.18
3.61

1.80
42
365
161
30
25
260
68
31
85
25.9

LAM

S146

S152

50.3
I.91
17.2
11.2
0.20
4.73
8.57
3.63
1.67
43
288
242
41
30
281
18
20
107
25.9

52.1
1.55
18.3
9.33
0.18
4.58
8.54
3.81
1.26
16
299
214
37
29
259
10
13
89
17.2

c3
52.9
I .22
18.1
9.26
0.18
4.53
8.62
3.82
1.09
28
305
185
32
24
205
6
I4
91
16.5

S164
50.7

1.29
18.4
9.45
0.17
5.29
9.61
3.46
1.36
28
346
117
24
27
268
89
46
85
16.4

S169
51.2

1.07
19.0
9.49
0.19
4.63
10.0
2.90
1.28
30
341
78
26
27
214
b.d.
11
96
12.8

D. Vukadinovic and I. Sutawidjaja

164

6>MgO>4wt%
HAM

LAM

Sample

s117

S116

S146

s152

Ce
Pr
Nd
Sm
Eu
Gd
Tb
DY
Ho
Er
Yb
Ba
U
Th
Hf
Nb
87/86Sr

45.5
5.47
23.7
5.57
1.46
5.02
0.87
5.33
1.11
3.04
2.77
278
1.06
5.73
3.72
13.4
0.70495

57.3
6.92
28.1
6.07
1.61
5.00
0.83
4.95
1.01
2.66
2.24
260
1.oo
6.03
3.50
15.0
0.70520

57.9
7.38
32.4
7.57
1.92
6.80
1.13
6.93
1.45
3.86
3.67
290
1.78
7.59
5.64
23.3
0.70510

41.7
5.46
24.6
5.95
1.77
6.16
1.02
6.67
1.39
3.83
3.42
239
0.90
3.91
4.60
17.3
0.70517

c3
39.7
4.85
21.9
5.12
1.45
5.06
0.87
5.30
1.08
3.17
2.96
183
0.79
3.90
4.47
16.9
0.70478

S164

S169

35.5
4.53
20.3
4.41
1.34
4.11
*

29.0
3.83
16.6
4.19
1.20
3.80
0.70
*

4.01
0.82
2.36
2.22
320
0.90
4.68
2.73
10.4
0.70569

0.89
2.54
2.49
178
0.99
4.76
2.38
3.50
0.70602

MgO < 4 wt%


HAM
Sample

s71

SiO,
TiO,
Al203
FeO*
MnO
MgG
CaO
Na,O
I&G
Rb
Sr
Zr
Y
SC
V
Cr
Ni
Zn
La
Ce
Pr
Nd
Sm
Eu
Gd
Tb
DY
Ho
Er
Yb
Ba
U
Th
Hf
Nb
87/86Sr

54.7
1.39
17.6
8.78
0.17
3.69
7.87
3.72
1.89
49
283
209
33
24
193
4
10
81
19.2
43.7
5.58
24.6
5.67
1.43
4.90
0.86
5.33
1.12
3.13
3.01
286
1.16
5.67
4.34
17.0
0.70522

Sl61
54.0
1.10
18.7
8.72
0.17
3.55
8.35
3.69
1.50
37
331
150
30
25
223
b.d.
5
121
19.9
44.6
5.50
23.0
5.27
1.55
5.01
0.84
5.33
1.14
3.18
2.89
305
0.98
4.71
3.80
11.9
0.70570

LAM
S178

S167

S83

s90

s31

S25

56.1
1.22
18.2
8.33
0.15
3.26
7.21
3.54
1.74
48
290
181
37
19
187
b.d.
10
81
20.0
41.9
5.77
26.3
5.85
1.53
5.31
0.95
5.60
1.16
3.11
3.03
267
1.25
6.22
4.22
14.7
0.70549

63.2
0.59
18.5
4.70
0.12
1.36
4.63
4.02
2.75
82
250
330
37
12
60
b.d.
4
70
27.6
59.5
7.15
27.6
6.13
1.27
5.34
1.oo
6.21
1.33
4.00
3.88
477
2.28
11.7
8.20
17.8
0.70578

56.1
0.73
18.3
7.73
0.17
3.80
8.13
3.29
1.64
48
316
105
19
18
177
3
11
102
16.0
34.4
3.87
16.0
3.33
0.97
3.37
0.54
*
0.63
1.91
1.95
300
1.27
6.80
2.81
3.70
0.70571

57.5
0.62
18.6
6.74
0.16
2.78
7.42
3.87
2.11
61
314
121
22
13
169
b.d.
8
83
19.1
40.4
4.43
18.2
3.80
1.15
3.51
0.56
*
0.67
1.76
2.10
385
1.36
7.57
2.76
4.50
0.70594

58.4
0.65
18.3
6.79
0.16
2.78
7.30
3.73
1.87
55
351
125
18
15
164
b.d.
7
83
18.5
37.7
4.06
16.7
3.56
1.02
3.25
0.50
2.95
0.66
1.96
2.13
387
1.67
7.76
3.23
4.50
0.70610

58.2
0.57
18.8
6.66
0.15
2.54
8.05
3.54
1.44
40
332
94
15
17
141
b.d.
7
78
12.5
26.1
3.41
14.0
3.15
0.97
2.85
0.47
3.01
0.65
1.87
2.14
346
1.19
5.30
2.60
4.20
0.70542

Oxides given in wt%, all other elements in ppm; major elements determined by electron microprobe on fused glass
chips (Monash University); Rb, Sr, Zr, Y, SC, Cr, Ni and Zn analysed via XRF (Monash University); all other
elements determined by spark-source mass spectrography (Australian National University) except for REEs in S108
and Sr in S108, S61 and S167, which were determined by isotope dilution methods (CSIRO); and Ba in S108 via
XRF methods. Sr-isotope analyses were carried out at CSIRO. *Sr/86Sr(average uncertainty +0.00006) normalized
to 0.1194 for **Sr/86Sr.n.d. = not determined; b.d. = below detection limits; * = not determined due to interferences
from LREE and Ba oxides and/or carbides.

También podría gustarte