Gertisser Et Al 2012 Tambora

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 28

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/236858075

Processes and Timescales of Magma Genesis and Differentiation Leading to


the Great Tambora Eruption in 1815

Article in Journal of Petrology · February 2012


DOI: 10.1093/petrology/egr062

CITATIONS READS

45 1,437

6 authors, including:

Ralf Gertisser Stephen Self


Keele University University of California, Berkeley
143 PUBLICATIONS 3,686 CITATIONS 244 PUBLICATIONS 20,026 CITATIONS

SEE PROFILE SEE PROFILE

Louise Thomas P. van Calsteren


The Open University (UK) The Open University (UK)
51 PUBLICATIONS 2,161 CITATIONS 108 PUBLICATIONS 8,629 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Ralf Gertisser on 28 May 2014.

The user has requested enhancement of the downloaded file.


JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 PAGES 271^297 2012 doi:10.1093/petrology/egr062

Processes and Timescales of Magma Genesis and


Differentiation Leading to the Great Tambora
Eruption in 1815

RALF GERTISSER1*, STEPHEN SELF2, LOUISE E. THOMAS2,


HEATHER K. HANDLEY3, PETER VAN CALSTEREN2 AND
JOHN A. WOLFF4

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


1
SCHOOL OF PHYSICAL AND GEOGRAPHICAL SCIENCES, KEELE UNIVERSITY, KEELE ST5 5BG, UK
2
DEPARTMENT OF EARTH AND ENVIRONMENTAL SCIENCES, THE OPEN UNIVERSITY, WALTON HALL, MILTON
KEYNES MK7 6AA, UK
3
GEMOC, DEPARTMENT OF EARTH AND PLANETARY SCIENCES, MACQUARIE UNIVERSITY, SYDNEY, NSW 2109,
AUSTRALIA
4
SCHOOL OF EARTH AND ENVIRONMENTAL SCIENCES, WASHINGTON STATE UNIVERSITY, PULLMAN, WA 99164-2812,
USA

RECEIVED JULY 10, 2010; ACCEPTED NOVEMBER 7, 2011


ADVANCE ACCESS PUBLICATION DECEMBER 15, 2011

The cataclysmic eruption of Tambora volcano (Sumbawa, erupted in 1815. Highly calcic, corroded plagioclase crystals in
226
Indonesia) in 1815 has long been recognized as one of the largest Ra^230Th equilibrium (48000 years old) provide physical
explosive eruptions in historical time. It yielded extensive pyroclastic evidence for incorporation of ‘antecrystic’ material into the 1815
deposits from the emptying of a 30^33 km3 trachyandesite (latite)^ magma. Magma accumulation and differentiation at shallow depth
tephriphonolite (herein referred to as trachyandesite) magma body. prior to the eruption were accompanied by continuous degassing of
The parental trachybasalt magma of the trachyandesite erupted in sulphur (and other volatile species), which is thought not to have
1815 can be produced by 2% partial melting of a garnet-free, accumulated within or towards the top of the magma reservoir to
Indian-type mid-ocean ridge basalt (I-MORB)-like mantle source contribute to the volatile budget of the eruption, but to have escaped
contaminated with 3% fluids from altered oceanic crust and to the surface passively through the permeable wall rocks.
51% sedimentary material, preserving small 238U excesses in the
Tambora rocks. Magmatic differentiation from primary trachybasalt
to trachyandesite occurred during two-stage, polybaric differentiation KEY WORDS: Tambora; Sumbawa; Sunda arc; 1815 eruption; potassic
at depth(s) around the Moho and in a shallow-level crustal magma; Hf isotopes; U-series disequilibria; volatiles
magma reservoir, emplaced at a maximum depth of 7·5 km (and,
possibly, as shallow as 2·3 km). This crustal reservoir grew by
influx of basaltic trachyandesite (shoshonite) magma, which I N T RO D U C T I O N
originated predominantly by partial crystallization of primary The 10^11 April 1815 eruption of Tambora volcano
trachybasalt in the inferred deep reservoir. Subsequent magmatic (Sumbawa, Indonesia) was one of the largest explosive
differentiation dominated by fractional crystallization, magma eruptions in historical time (Self et al., 1984; Stothers, 1984;
recharge^mixing and convection over timescales of 4000^4500 Sigurdsson & Carey, 1989, 1992a). It was the main con-
years led to the trachyandesitic (and ultimately phonolitic) melts tributor to ‘the year without a summer’ in 1816, which had

 The Author 2011. Published by Oxford University Press. All


*Corresponding author. Telephone: þ44 (0)1782 733181. Fax: þ44 rights reserved. For Permissions, please e-mail: journals.permissions@
(0)1782 733172. E-mail: r.gertisser@esci.keele.ac.uk oup.com
JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 FEBRUARY 2012

a mean global temperature decrease of 1·0^1·58C T EC TON IC S ET T I NG A N D


(e.g. Rampino & Self, 1982; Stothers, 1984; Sigurdsson &
Carey, 1992a; Mann et al., 1998; Cole-Dai et al., 2009).
G E O L O G Y O F TA M B O R A
Tephra volume estimates of the eruption have varied VO L C A N O
widely between 95 and 175 km3 (Self et al., 1984; Tambora is a large shield-like alkaline volcano on
Stothers, 1984; Sigurdsson & Carey, 1989, 1992a); here we Sumbawa in the eastern part of the Sunda arc (Fig. 1),
adopt the most recent volume estimate of 107^113 km3 associated with the northward subduction of the
[30^33 km3 dense rock equivalant (DRE)] (Self et al., Indo-Australian plate beneath the Eurasian plate at
2004), making it the largest eruption of at least the past 6^7 cm a1 (Hamilton, 1979). The volcano is located at
500 years, and perhaps even the past 1000 years. the rear of the main volcanic front 180 km above the
A remarkable feature of the 1815 eruption is that it Benioff Zone, where the crust is relatively young, compos-
involved a fairly homogeneous batch of trachyandesite itionally immature and 14^17 km thick (Hamilton, 1979).
(latite)^tephriphonolite (herein referred to as trachyande- On Sumbawa, the upper portion of this crust consists of
site) magma (Self et al., 1984; Foden, 1986; Heyckendorf & Cenozoic sedimentary sequences of siliciclastic marine

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


Jung, 1992), a comparatively rare magma type in volcanic sediments and limestones, as well as volcanic and intrusive
arc settings. An understanding of how such large bodies of rocks (van Bemmelen, 1949).
potassic magmas are generated requires knowledge of the Tambora is one of several large Quaternary volcanoes
processes and timescales involved in magma genesis and on Sumbawa that have produced magma compositions
subsequent magma accumulation and differentiation in ranging from calc-alkaline andesites to highly silica-
crustal reservoirs. Such information has implications for undersaturated leucitites (Foden & Varne, 1980). The
volcanic hazard assessment and long-term eruption fore- eruptive products of Tambora are predominantly potassic,
casting at Tambora and other K-rich volcanoes located in silica-undersaturated lavas and pyroclastic rocks ranging
rear-arc (i.e. between the arc front and the back-arc) and from nepheline-normative alkali basalt to trachyandesite
other geodynamic settings. (Foden, 1986). Compositions found at Tambora are in
Despite its significance and notoriety, no recent compre- many respects similar to those of other K-rich Sunda arc
hensive geochemical and isotopic study has been con- volcanoes (Fig. 1), including Muriah and Ringgit-Beser on
ducted on the products of the 1815 Tambora eruption. Java (Edwards et al., 1991, 1994), Sangeang Api off the
Foden (1986, 1987) proposed that the 1815 eruption followed NE coast of Sumbawa (Turner et al., 2003a) and Batu
exsolution of a volatile phase during gradual, closed-system Tara to the north of Flores (Stolz et al., 1988; van Bergen
cooling and crystallization of a hydrous magma in a et al., 1992).
high-level magma chamber emplaced at a depth of at The volcano occupies most of the Sanggar Peninsula in
least 1·5^4·5 km (see Self & Wolff, 1987). At nearby north^central Sumbawa (Fig. 2). It has a sea-level diameter
Sangeang Api volcano, which produces potassic magma of 60 km and an elevation of 2850 m. The summit is now
compositions similar to those of Tambora, it has been the rim of the 6 km diameter (41000 m deep) 1815 caldera,
proposed that magmatic differentiation, inferred from although prior to 1815, Tambora was 4300 m high and
U^Th^Ra disequilibria, has occurred on timescales of capped by a steeper sloped cone. Some 20 parasitic cones
2000 years (Turner et al., 2003a). Based on similarities in are distributed over the flanks of the volcano (Self et al.,
composition and liquid line of descent to the Sangeang 1984). On the basis of present knowledge, the 1815 eruption
Api magmas, Turner et al. (2003a) proposed a crustal products are among the most evolved rock types erupted
residence time of 5000^6000 years for the 1815 Tambora in the history of the volcano (Fig. 3). The stratigraphy
magma. exposed in the walls of the 1815 caldera includes four
Here we present the results of a detailed petrological, major pre-1815 volcanic formations (Sigurdsson & Carey,
geochemical and isotopic (Sr, Nd, Hf and U-series disequi- 1992b). Barberi et al. (1987) reported a K^Ar age of 43 ka
libria) investigation of Tambora based on samples that for the lowermost series of lava flows that fill a pre-existing
span most of the volcanic succession of the 1815 eruption, caldera. These lavas are overlain by two pyroclastic units,
plus pre-1815 lava flows erupted from the central vent or the Black Sands and Brown Tuff formations. The latter
from parasitic (flank) vents (Self et al., 1984). These data consists of interbedded pyroclastic surge and fall deposits
help constrain the processes of potassic arc magma genesis and represents the latest activity prior to 1815.
and evolution and the timescales of these processes that Radiocarbon ages obtained from the lower and upper
culminated in the 1815 eruption. The major element and parts of the Brown Tuff formation suggest that it formed
volatile compositions of melt inclusions within minerals by intermittent volcanic activity between 5900 and 1210
14
and the groundmass glass of the 1815 pumices are utilized C years BP (Sigurdsson & Carey, 1992b), indicating a
to trace magmatic differentiation and degassing processes period of inactivity of at least 1000 years before the 1815
prior to (and during) the eruption. eruption.

272
GERTISSER et al. PETROGENESIS OF THE 1815 TAMBORA MAGMA

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


Fig. 1. Map of the Indonesian volcanic arc system, showing the general tectonic setting (after Hamilton, 1979) and distribution of active
volcanoes (open circles) (after Siebert et al., 2010). The enlarged map shows the location of Tambora on Sumbawa in the eastern Sunda arc,
highlighting the active (double open circles) and extinct (filled circles) potassic alkaline Sunda arc volcanoes. The former includes Tambora.

C H RO N O L O GY A N D D E P O S I T S O F (Fig. 2c), which was formed by fallout from an umbrella


cloud fed by a sustained, but partially collapsing eruption
T H E 18 15 E RU P T I O N column and co-ignimbrite ash clouds that were recycled
Eruptive activity at Tambora began in the early evening back into the main eruption column (Self et al., 2004).
of 5 April 1815 with minor and intermittent pyroclastic Caldera collapse probably occurred at the end of the
emissions including a short, high-intensity Plinian column climactic phase, and intermittent explosions continued
(Self et al., 1984; Sigurdsson & Carey, 1989, 1992a; Self into the evening of 11 April, when reports show the ash
et al., 2004). Intermittent explosive eruptions continued cloud reached as far as western Java, 1300 km from the
until the evening of 10 April, when the volcano went source (Fig. 2c). Ash fall in distal areas ceased between 14
into a climactic phase that lasted for about a day (Self and 17 April, although minor activity may have continued
et al., 1984; Stothers, 1984). At the onset of the climax, until 23 August (Stothers, 1984).
a high-intensity Plinian column produced pumice fall
deposits in proximal areas (Fig. 2a), but within an hour
column collapse occurred, generating pyroclastic flows
that reached the coast (Fig. 2b). A N A LY T I C A L T E C H N I Q U E S
The occurrence of intra-ignimbrite pumice fall layers Fifty-two samples were selected that are representative of
(Self et al., 1984, 2004) indicates that for much of most of the stratigraphic sequence of the 1815 deposits (see
Tambora’s climactic phase pyroclastic fall and flow depos- designation of samples given in Table 1 keyed to the deposit
ition was simultaneous or alternating (Fig. 2a). The most sequence in Fig. 2a). The analysed dataset also includes
voluminous 1815 eruption deposit was the distal ash fall some older lava flows from Tambora.

273
JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 FEBRUARY 2012

(a) DEPOSIT CHARACTERISTICS

Welded zone
(b)
Satonda

Moyo

Tambora Sumbawa
E. UPPER IGNIMBRITE
5-30 m Surges/cross-bedded
E
ignimbrite with interbedded
pumice fallout
Sanggar
Gulf of Saleh Peninsula
Sanggar

D
D. INTRA-IGNIMBRITE FALLOUT
Plinian pumice fallout Area covered
10 km Doro Labumbum
by 1815 ignimbrite

(c) Sulawesi
C. LOWER IGNIMBRITE Kalimantan

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


5-30 m Non-welded to incipiently
C
welded ignimbrite with 1
interbedded pumice fallout
5

20

25
Ground layer (surge) Java 50
Bali 100

B
B. LOWER PLINIAN FALLOUT Flores
0.5 m Pumice fallout (3 sub-units) Lombok Sumbawa

A A. EARLY PHREATOMAGMATIC FALLOUT 500 km Sumba


Fine ash/lapilli fallout
Soil

Fig. 2. (a) Characteristics and generalized, composite stratigraphic section of the 1815 Tambora eruptive products. (b) Sketch map of Tambora
illustrating the distribution of ignimbrite from the 1815 eruption on the Sanggar Peninsula (after Self et al., 1984). (c) Distribution of the distal
ash fall from the eruption (after Self et al., 1984). Isopach thicknesses are given in centimetres.

14
Post AD 1815
Tephri-
12 phonolite AD 1815
Pre AD 1815

10
Na2O + K2O (wt.%)

Phono-
tephrite Trachyte

Trachy- Rhyolite
8 Tephrite/
andesite
(Latite)
Basanite
Basaltic
trachyandesite
Trachybasalt
6 (K-trachybasalt)
(Shoshonite)

Dacite

4 Andesite

Basaltic
andesite
2 Picro-
Basalt
Basalt

0
35 40 45 50 55 60 65 70 75 80
SiO2 (wt.%)
Fig. 3. Total alkalis vs SiO2 diagram for the Tambora rocks. Analyses from this study are shown by filled symbols. Open symbols denote data
from other researchers (Whitford et al., 1978; Foden, 1986; Varne & Foden, 1986; Grall-Johnson, 1997; Turner & Foden, 2001). Tambora rocks
of more ‘calc-alkaline’ affinity are distinguished from the predominantly alkaline series rocks by triangles with bold outlines. All analyses are
normalized to 100 wt %, free of volatiles. Potassic rock names are added in parentheses.

274
GERTISSER et al. PETROGENESIS OF THE 1815 TAMBORA MAGMA

Table 1: Selected whole-rock major element (XRF), trace element (ICP-MS) and Sr, Nd and Hf isotopic (TIMS) data for
pumices from the 1815 eruption, as well as some older (pre-1815) lava flows

Sample: T07 T16 T25 T28 T37 T39 T52 T55 T57/58 T04 T06 T18 T35
Age: AD 1815 AD 1815 AD 1815 AD 1815 AD 1815 AD 1815 AD 1815 AD 1815 AD 1815 Pre-1815 Pre-1815 Pre-1815 Pre-1815
Unit:1 C C E B C C C D B – – – –

wt %
SiO2 54·42 56·35 55·49 55·93 55·22 56·04 55·04 56·32 56·29 52·01 51·72 47·21 48·89
TiO2 0·68 0·62 0·66 0·60 0·65 0·63 0·62 0·61 0·60 0·89 0·77 1·04 0·97
Al2O3 19·32 19·77 19·50 19·62 19·62 19·77 19·39 19·71 19·67 16·23 15·39 17·58 16·35
Fe2O3* 6·48 5·58 6·05 5·53 6·14 5·79 5·67 5·53 5·49 8·54 10·79 11·19 10·47

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


MnO 0·18 0·18 0·18 0·19 0·19 0·19 0·18 0·19 0·18 0·15 0·19 0·22 0·20
MgO 2·10 1·61 1·89 1·54 1·85 1·71 1·78 1·61 1·55 8·80 7·03 4·37 5·88
CaO 5·47 4·15 4·67 3·98 4·65 4·40 4·29 4·07 4·02 9·44 11·16 11·06 11·08
Na2O 4·76 5·29 4·63 5·26 4·83 5·23 5·58 5·23 5·27 2·95 2·42 3·75 3·03
K2O 4·78 5·77 5·53 5·73 5·37 5·62 5·48 5·74 5·77 0·90 0·85 0·92 1·86
P2O5 0·43 0·43 0·45 0·41 0·44 0·45 0·43 0·42 0·41 0·22 0·16 0·56 0·31
LOI 0·53 0·13 1·33 0·53 1·00 0·26 1·38 0·58 0·36 0·19 0·40 1·78 1·51
Total 99·15 99·88 100·38 99·32 99·96 100·09 99·84 100·01 99·61 99·94 100·08 99·68 100·55
ppm
Cs 3·56 4·47 n.d. 4·30 3·79 4·32 3·77 4·40 4·31 1·77 1·44 2·30 1·33
Li 18·9 19·8 n.d. 25·1 18·6 19·4 22·4 19·2 18·8 9·8 9·8 9·3 8·9
Rb 146·2 157·3 n.d. 175·6 134·9 150·5 153·4 154·4 151·8 22·1 25·4 130·3 34·3
Ba 1177 1227 n.d. 1292 1096 1215 1197 1199 1171 213 233 702 725
Sr 1101 989 n.d. 1048 944 1002 1024 939 919 486 341 1429 965
Mo 3·35 3·62 n.d. 3·95 2·96 3·43 3·75 3·52 3·43 0·75 0·72 0·72 0·93
Sn 1·26 1·02 n.d. 1·35 0·93 1·02 1·35 1·02 0·98 0·92 0·84 0·95 0·76
Pb 17·32 16·79 n.d. 18·84 18·85 16·23 17·39 16·55 16·12 11·23 8·18 7·86 8·17
Zr 149·8 166·5 n.d. 167·9 146·8 162·3 155·5 163·0 160·1 98·2 71·2 119·1 86·0
Nb 8·36 8·79 n.d. 9·51 7·64 9·68 8·76 8·65 8·46 6·73 3·21 5·14 4·15
Hf 3·31 3·66 n.d. 3·67 3·23 3·55 3·44 3·58 3·48 2·27 1·82 2·79 2·22
Ta 0·59 0·47 n.d. 0·50 0·40 0·76 0·47 0·46 0·44 0·43 0·20 0·51 0·23
Th 12·80 12·83 n.d. 15·26 11·44 12·52 13·24 12·55 12·28 2·96 3·52 6·92 2·02
U 3·22 3·72 n.d. 3·73 3·27 3·62 3·37 3·64 3·55 1·12 1·00 1·75 1·06
Y 27·4 27·0 n.d. 30·2 25·3 26·8 27·2 26·2 25·8 21·7 21·2 27·4 18·1
Cu 42·7 23·1 n.d. 23·7 41·9 34·7 86·8 20·2 19·8 67·8 131·3 135·0 108·6
Zn 103·5 98·2 n.d. 105·1 93·8 97·1 101·8 95·1 93·8 76·8 88·3 102·3 96·3
Ni 7·4 0·5 n.d. 0·7 1·8 2·2 15·4 0·6 0·6 154·5 45·4 13·1 27·7
Cr 15·6 1·1 n.d. 0·6 2·1 0·4 2·6 0·6 0·9 365·7 135·3 15·5 76·9
Co 14·9 10·0 n.d. 10·5 11·5 10·9 11·3 9·5 9·3 38·6 43·3 33·8 34·9
Sc 9·8 8·2 n.d. 6·1 9·0 8·5 6·2 7·6 8·3 31·8 46·8 27·5 7·2
V 153·7 104·7 n.d. 106·1 118·0 114·0 112·4 100·8 98·6 235·5 320·2 314·8 329·0
La 42·18 40·48 n.d. 48·38 36·70 40·10 43·21 39·64 38·96 12·79 11·53 35·22 29·47
Ce 71·56 74·97 n.d. 81·21 69·05 75·00 72·68 73·46 72·08 24·64 22·88 71·71 43·33
Pr 8·25 8·36 n.d. 9·14 7·68 8·34 8·25 8·16 7·99 3·09 2·90 8·70 5·05
Nd 30·21 28·58 n.d. 32·90 26·48 28·63 30·05 27·87 27·32 12·90 12·36 32·44 19·09
Sm 6·06 5·99 n.d. 6·54 5·56 6·03 5·99 5·83 5·75 3·24 3·13 7·46 4·65
Eu 1·80 1·71 n.d. 1·88 1·61 1·72 1·75 1·65 1·62 1·09 0·97 2·19 1·40
Gd 5·60 5·47 n.d. 6·00 5·13 5·57 5·52 5·36 5·23 3·54 3·34 7·02 4·46

(continued)

275
JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 FEBRUARY 2012

Table 1: Continued

Sample: T07 T16 T25 T28 T37 T39 T52 T55 T57/58 T04 T06 T18 T35
Age: AD 1815 AD 1815 AD 1815 AD 1815 AD 1815 AD 1815 AD 1815 AD 1815 AD 1815 Pre-1815 Pre-1815 Pre-1815 Pre-1815
Unit:1 C C E B C C C D B – – – –

Tb 0·79 0·75 n.d. 0·85 0·70 0·75 0·78 0·73 0·71 0·58 0·55 0·92 0·63
Dy 4·47 4·26 n.d. 4·78 3·98 4·28 4·41 4·15 4·08 3·53 3·42 4·89 3·58
Ho 0·92 0·89 n.d. 0·99 0·83 0·90 0·90 0·87 0·86 0·76 0·74 0·95 0·74
Er 2·55 2·54 n.d. 2·76 2·33 2·54 2·52 2·48 2·43 2·12 2·09 2·47 2·01
Yb 2·41 2·52 n.d. 2·71 2·30 2·50 2·46 2·47 2·42 2·01 1·98 2·09 1·82
Lu 0·38 0·39 n.d. 0·42 0·36 0·39 0·38 0·39 0·38 0·32 0·31 0·31 0·28
87
Sr/86Sr 0·704000 0·704049 n.d. 0·703992 0·704100 0·703984 0·704029 0·704021 0·703989 0·704497 0·704606 0·704151 0·704081

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


2s 0·000001 0·000004 n.d. 0·000001 0·000005 0·000001 0·000002 0·000001 0·000002 0·000013 0·000008 0·000008 0·000007
143
Nd/144Nd 0·512744 0·512809 n.d. 0·512810 0·512803 0·512808 0·512800 0·512822 0·512800 0·512911 0·512823 0·512907 0·512811
2s 0·000001 0·000010 n.d. 0·000001 0·000001 0·000001 0·000001 0·000001 0·000001 0·000010 0·000010 0·000001 0·000001
176
Hf/177Hf 0·283102 0·283125 n.d. 0·283099 0·283099 0·283169 0·283103 0·283110 0·283104 0·283137 0·283149 0·283083 0·283128
2s 0·000009 0·000018 n.d. 0·000011 0·000010 0·000065 0·000011 0·000014 0·000008 0·000008 0·000011 0·000023 0·000015

1
Keyed to Fig. 2a.
*Total iron given as Fe2O3.
n.d., not determined.

Whole-rock major element concentrations (Table 1) were unleached and leached samples. Following separation of
determined on fused discs by X-ray fluorescence (XRF) Sr and Nd using standard ion exchange techniques, Sr
spectrometry at The Open University. Reproducibility and Nd isotopic ratios were determined by thermal ioniza-
and accuracy, monitored using several international rock tion mass spectrometry (TIMS) on a Finnigan Triton
standards, are better than 1% and 5% for major elem- system at The Open University using static collection
ents. Loss on ignition (LOI) was determined by weight procedures. Sr and Nd were fractionation corrected to
86
difference on heating 1g of rock powder to 10008C for Sr/88Sr ¼ 0·1194 and 146Nd/144Nd ¼ 0·7219. The measured
1h. LOI, as reported in Table 1, is the net effect of dehydra- isotopic ratios were normalized with respect to determined
tion (loss of weight) and oxidation (gain of weight). values of the NBS 987 Sr standard [87Sr/86Sr ¼
A subset of whole-rock samples was selected for trace 0·710235  9 (2s); n ¼ 25] and the Johnson Matthey Nd
element analysis (Table 1) by inductively coupled plasma standard [143Nd/144Nd ¼ 0·511822  2 (2s); n ¼ 20] run
mass spectrometry (ICP-MS) using an Agilent 7500s during the period of these analyses. Quoted uncertainties
quadrupole mass spectrometer at The Open University. are two standard deviations. Total procedural blanks for
Calibration was based on up to six reference materials Sr and Nd were typically 51ng and 500 pg, respectively,
(BCR1, BIR1, AC, RGM1, BHVO1-2 and AGV1), using and trivial compared with the amounts of Sr and Nd
recommended element concentrations. Instrumental drift analysed.
was monitored by and corrected using an internal standard Hf isotope analysis (Table 1) was undertaken by
solution containing Be, In, Rh, Tm, Re and Bi introduced multi-collector (MC)-ICP-MS using a ThermoElectron
on-line during analysis. Residual drift in single elements Neptune system at the Arthur Holmes Isotope Geology
was corrected externally using replicate determinations Laboratory (AHIGL), University of Durham, following
of a representative sample after every 10 unknowns. sample preparation techniques and instrument operating
Precision was assessed from replicate analyses of a selected conditions described by Handley et al. (2007) and refer-
unknown sample (T4) and an appropriate reference ences therein. Instrumental mass bias was corrected for
material. Within single runs, relative precision was gener- using a 179Hf/177Hf ratio of 0·7325 and an exponential law.
ally about 1% for elements above Cs (mass 133) and Samples were analysed on two separate occasions with
1^3% for elements below this mass. average JMC 475 176Hf/177Hf values and 2s uncertainties
For Sr and Nd isotopic analysis (Table 1), whole-rock of 0·282143  5 (n ¼11, reproducibility ¼18·2 ppm) and
powders were not leached prior to dissolution as previous 0·282143  4 (n ¼ 8, reproducibility ¼13·4 ppm), respect-
experiments had shown no significant difference between ively, and lie within error of the long-term JMC 475

276
GERTISSER et al. PETROGENESIS OF THE 1815 TAMBORA MAGMA

average value of 0·282145 7 (Nowell et al., 2003). Data are Th standard gave (230Th/232Th) ¼ 1·015  0·0005 (n ¼ 2).
reported relative to an accepted JMC 475 176Hf/177Hf value Repeated analysis of the ‘in house’ uranium solution stand-
of 0·282160 (Nowell et al., 1998). Total procedural blanks of ard U456 yielded (234U/236U) ¼ 0·097495  0·000680,
573 pg Hf were determined by ICP-MS on the a 2s deviation of 0·7%, and (235U/236U) ¼ 13·272  0·029,
PerkinElmer ELAN 6000 quadrupole system at the a 2s deviation of 0·21%. Replicate analyses of the Mt
University of Durham. Lassen rock standard yielded values of 226Ra ¼1·080 
U, Th and 226Ra concentrations and isotope activity 0·026 fg g1 and 1·065  0·010 fg g1, within error of pub-
ratios of three bulk pumice samples from the 1815 deposits, lished values (Volpe et al., 1991). Reproducibility is
mineral separates (plagioclase, clinopyroxene, biotite, 1% (2s), as determined on an in-house standard. Total
titanomagnetite) and groundmass glass (Table 2) were procedural blanks for 226Ra were 50·1fg g1. Chemical
determined at The Open University using a Finnigan preparation blanks were 50 pg for Th and 70 pg for U.
MAT 262 mass spectrometer equipped with an RPQ-II The analytical error of the U/Th ratios is 0·5%.
energy filter. Samples were spiked with a mixed Decay constants (a^1) used in the calculation of activity
229
Th^236U tracer and a 228Ra tracer, respectively. ratios were 238U ¼1·551 1010, 234U ¼ 2·835 106,

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


232
Sample dissolution and column procedures for Th and U Th ¼ 4·948 1011, 230Th ¼ 9·195 106 and 226Ra ¼
used in this study follow those described in detail elsewhere 4·332 104. The reported whole-rock (226Ra/230Th)
(Turner et al., 1997; Thomas et al., 1999). Ra was extracted ratios are age corrected back to the time of eruption.
using a cation exchange column followed by separation Mineral major element compositions for several 1815
from Ba on a Sr spec resin column. During the period samples (Table 3) were determined using a CAMECA SX
of this study, multiple determinations of the Th‘U’ standard 100 electron microprobe at The Open University. Peak
(van Calsteren & Schwieters, 1995) gave 230Th/232Th ¼ counting times per element were 20 s using a 5^10 mm
6·131 0·021 106 (n ¼ 22), a 2s standard deviation of defocused beam, an acceleration voltage of 20 kV and
1%, whereas duplicate determinations of the ATHO a beam current of 20 nA. Major element analyses of

Table 2: U-series data for whole-rock samples, constituent mineral phases and groundmass glass from the 1815 Tambora
eruptive products

Type U (ppm) Th (ppm) (234U/238U) 2s (238U/232Th) 2s (230Th/232Th) 2s (230Th/238U) 2s 226


Ra (fg g1) (226Ra/230Th)

Sample T07
wr 3·35 11·11 1·002 0·003 0·914 0·008 0·902 0·001 0·987 0·001 2079·8 1·985
bio 0·29 1·03 1·010 0·005 0·854 0·004 0·811 0·001 0·950 0·001 704·0 8·036
mt 0·77 2·81 1·007 0·006 0·830 0·004 0·829 0·002 1·000 0·003 366·7 1·507
pl 0·52 2·03 1·003 0·004 0·784 0·003 0·798 0·002 1·018 0·002 203·8 1·204
gm 3·58 13·16 1·008 0·005 0·824 0·005 0·722 0·013 0·876 0·016 1165·9 1·173
Sample T39
wr 3·29 11·60 0·998 0·003 0·860 0·002 0·776 0·001 0·902 0·001 1202·8 1·527
bio 0·41 1·66 1·001 0·006 0·740 0·007 0·785 0·003 1·066 0·004 112·2 0·963
mt 0·81 3·45 1·060 0·007 0·712 0·006 0·816 0·001 1·145 0·006 306·6 1·184
pl 0·52 1·44 1·000 0·003 1·090 0·012 0·797 0·002 1·364 0·002 204·4 0·569
cpx 0·46 2·01 1·004 0·003 0·689 0·001 0·788 0·002 1·103 0·004 n.d. n.d.
gm 3·56 12·62 0·993 0·004 0·855 0·003 0·788 0·006 0·875 0·007 1149·9 1·547
Sample T57/58
wr 3·42 12·36 1·015 0·028 0·840 0·006 0·819 0·002 0·974 0·006 1219·1 1·152
bio 0·25 0·97 1·012 0·024 0·783 0·004 0·790 0·000 1·009 0·005 n.d. n.d.
mt 0·67 2·66 1·005 0·005 0·758 0·002 0·800 0·000 1·055 0·002 315·4 1·415
pl 0·63 2·43 1·003 0·006 0·791 0·001 0·808 0·003 1·021 0·004 1294·8 6·303
cpx 0·45 1·91 1·006 0·013 0·714 0·002 1·014 0·006 1·420 0·009 1683·5 8·318
gm 3·70 13·64 0·996 0·005 0·823 0·004 0·721 0·001 0·875 0·002 1025·3 0·997

wr, whole rock; bio, biotite; mt, magnetite; pl, plagioclase; cpx, clinopyroxene; gm, groundmass glass.

277
JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 FEBRUARY 2012

Table 3: Mineral compositions in pumice clasts from the 1815 Tambora eruption

Mineral: pl cpx ol mt bio

Description: ph-c ph-r m ph ph ph ph

No. of analyses (n):1 1 155 (1s) 9 (1s) 98 (1s) 13 (1s) 71 (1s) 116 (1s)

wt %
SiO2 45·85 53·34 (1·41) 60·01 (1·68) 50·97 (0·72) 37·94 (0·14) 36·83 (0·34)
TiO2 0·01 0·03 (0·01) 0·24 (0·10) 0·68 (0·14) 7·47 (0·20) 5·08 (0·19)
Al2O3 33·99 28·86 (0·98) 22·35 (1·76) 3·60 (0·61) 5·27 (0·20) 15·26 (0·18)
Cr2O3 0·01 (0·01) 0·04 (0·01) 0·01 (0·01)

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


FeO* 0·54 0·57 (0·05) 1·52 (0·76) 7·90 (0·51) 25·20 (0·22) 77·22 (0·34) 12·01 (0·21)
MnO 0·43 (0·07) 1·18 (0·02) 0·78 (0·05) 0·16 (0·02)
MgO 0·02 0·06 (0·02) 0·26 (0·16) 13·88 (0·48) 35·75 (0·26) 3·65 (0·13) 16·26 (0·23)
NiO 0·00 (0·01)
BaO 0·02 0·08 (0·03) 0·41 (0·19)
SrO 0·70 0·79 (0·05) 0·44 (0·23)
CaO 17·91 11·79 (1·16) 5·28 (1·77) 22·01 (0·20) 0·22 (0·01) 0·03 (0·09)
Na2O 1·00 4·31 (0·58) 5·44 (0·35) 0·46 (0·04) 0·01 (0·01) 0·76 (0·03)
K2O 0·06 0·60 (0·15) 4·38 (1·46) 0·01 (0·01) 0·01 (0·02) 8·74 (0·11)
Total 100·10 100·43 (0·36) 100·35 (0·62) 99·95 (0·32) 100·31 (0·36) 94·43 (0·28) 95·14 (0·49)
Cations (oxygens) (8) (8) (8) (6) (4) (4) (22)
Si 2·122 2·425 (0·057) 2·732 (0·074) 1·892 (0·022) 1·001 (0·002) 5·435 (0·031)
Al 1·853 1·546 (0·057) 1·041 (0·391) 0·157 (0·027) 0·226 (0·008) 2·655 (0·024)
Ti 0·000 0·001 (0·000) 0·167 (0·473) 0·019 (0·004) 0·204 (0·006) 0·565 (0·022)
Cr 0·000 (0·000) 0·000 (0·000) 0·001 (0·000) 0·001 (0·001)
Fe3þ 2
0·058 (0·015) 1·365 (0·010)
Mg 0·002 0·004 (0·001) 0·018 (0·011) 0·767 (0·025) 1·406 (0·007) 0·198 (0·007) 3·577 (0·044)
Fe2þ 0·021 0·022 (0·002) 0·058 (0·029) 0·187 (0·013) 0·556 (0·005) 0·982 (0·009) 1·482 (0·028)
Mn 0·013 (0·002) 0·026 (0·000) 0·024 (0·002) 0·020 (0·002)
Ni 0·000 (0·000)
Ba 0·000 0·001 (0·000) 0·007 (0·003)
Sr 0·019 0·021 (0·001) 0·012 (0·006)
Ca 0·888 0·574 (0·058) 0·257 (0·086) 0·874 (0·007) 0·006 (0·000) 0·005 (0·014)
Na 0·090 0·380 (0·050) 0·480 (0·028) 0·033 (0·003) 0·001 (0·000) 0·218 (0·010)
K 0·004 0·035 (0·008) 0·255 (0·086) 0·000 (0·000) 0·000 (0·001) 1·646 (0·022)
Total 4·999 5·009 (0·057) 5·027 (0·024) 4·000 2·996 (0·002) 3·000 15·604 (0·031)

pl, plagioclase; cpx, clinopyroxene; ol, olivine; mt, Ti-magnetite; bio, biotite; ph, phenocrysts (including crystals of
apparent xenocrystic origin); ph-c, core of compositionally zoned crystals; ph-r, phenocryst rim; m, microlites.
1
All data are averages of n analyses, except for the plagioclase core composition, where only the most calcic composition
is reported. Standard deviations for single elements are shown in parentheses.
2
Fe3þ determined stoichiometrically for clinopyroxene (cpx) and Ti-magnetite (mt).
*Total iron given as FeO.

plagioclase-hosted melt inclusions (Table 4) were obtained (Table 4). Natural silicate minerals were used as primary
using a 20 kV accelerating voltage, 10 nA beam current, a standards to calibrate the instrument. Smithsonian basalt
defocused beam (10^20 mm diameter) and peak counting VG-2, USGS basalt BCR-2G and KE-12, a peralkaline
times per element between 10 and 30 s chosen to minimize obsidian from Kenya, were used as secondary, in-run
total counting times per analysis. Extended counting standards to monitor precision and accuracy during
times were used for sulphur and chlorine measurements glass analyses.

278
GERTISSER et al. PETROGENESIS OF THE 1815 TAMBORA MAGMA

Table 4: Major element and volatile contents (with 1s (e.g. Gottsmann & Dingwell, 2002). For all calculations,
a molar absorption coefficient at 3550 cm1 of
standard deviation) of melt inclusions (MI) in calcic 64  8 l mol1 cm1 was assumed (Cioni, 2000). No CO2
plagioclase cores (pl-c), more sodic plagioclase rims (and concentrations above the FTIR spectroscopy detection
compositionally similar plagioclase cores) (pl-r) and limit (100 ppm) were found in the analysed melt
inclusions.
groundmass (gm) glasses from 1815 Tambora pumices

Type1 MI (pl-c) MI (pl-r) (1s)2 gm (1s)2 R E S U LT S


Petrography and mineral chemistry
Major elements (wt %) n¼1 n ¼ 30 n ¼ 126 The major component in the 1815 pyroclastic deposits
SiO2 53·12 56·17 (0·64) 58·30 (0·49) is vesicular (30^50 vol. %), brown to black, latitic to
TiO2 0·79 0·66 (0·09) 0·53 (0·03) tephriphonolitic pumice (Fig. 3), with crystal contents
that rarely exceed 15 vol. %. The mineral assemblage

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


Al2O3 19·33 18·90 (0·36) 19·32 (0·26)
Fe2O3* 6·16 5·36 (0·37) 5·05 (0·28) includes plagioclase4clinopyroxene4magnetite, with
MnO 0·21 0·18 (0·18) 0·18 (0·02) small amounts of apatite and occasional biotite and olivine
MgO 1·97 1·52 (0·17) 1·37 (0·11)
set in a glassy or, rarely, microcrystalline groundmass.
CaO 3·97 3·27 (0·25) 3·24 (0·33)
Ranges of mineral compositions in the 1815 eruptive
products are reported in Table 3.
Na2O 5·58 5·39 (0·40) 5·43 (0·27)
Plagioclase spans a wide compositional range from An43
K2O 5·81 6·06 (0·26) 6·25 (0·31)
to An91. It occurs as two distinctive populations giving
P2O5 0·64 0·49 (0·12) 0·35 (0·04)
rise to the bimodal distribution shown in the inset in
Total 97·57 98·00 100·02
Fig. 4a. These consist of (1) small, unzoned crystals of
Volatiles (ppm) n ¼ 26 n ¼ 78
near-uniform composition (An58  6) and (2) zoned crystals
S 1255 689 (80) 290 (74)
with (in some cases resorbed) calcic cores (up to An91)
Cl 2440 1720 (169) 1511 (132)
and variably thick and finely zoned rims of An58  6
1
(Self et al., 2004). Both populations may contain melt
All data are averages of n analyses, except for melt (glass) inclusions (Fig. 4b and c). Microlites of ternary
inclusions in calcic plagioclase cores for which only the
most primitive and volatile-rich composition is reported. feldspar occur occasionally in the pumice groundmass
Standard deviations for single elements are shown in (Fig. 4a; Table 3). Augitic to salitic clinopyroxene pheno-
parentheses. crysts are typically euhedral to subhedral, with little
2
After Self et al. (2004).
*Total iron given as Fe2O3. compositional variation within and between crystals:
Wo44^47, En36^42 and Fs13^18. Euhedral biotite with a very
narrow range of compositions (Mg-number ¼ 69·4^72·3)
occurs as a minor phase in some of the 1815 Tambora
rocks. Rare unzoned and homogeneous olivine micro-
Fourier Transform infrared (FTIR) spectra of a few melt phenocrysts (Fo71^72) are also occasionally present.
inclusions and small obsidian fragments in the 1815 deposits Titanomagnetite is the ubiquitous oxide phase present in
were collected with a Thermo Nicolet Nexus FTIR spec- trace amounts; this often occurs as inclusions in plagio-
trometer coupled with a Continumm IR microscope at clase, clinopyroxene and biotite. Like clinopyroxene
The Open University. For all spectra, standard EverGlo and biotite, titanomagnetite is uniform in composition
mid-IR source optics, a Ge-on-KBr beamsplitter and with ulvo«spinel contents ranging from 22 to 25 mol %.
a liquid nitrogen-cooled MCT-A* detector (11700^ Apatite is a common accessory mineral and occurs
750 cm1) were used. Concentrations of H2O were calcu- as microphenocrysts or inclusions in plagioclase,
lated from the height of the total water (i.e. molecular clinopyroxene and titanomagnetite crystals.
H2O þ OH) peak at 3550 cm1, using the Beer^Lambert
law: H2O (wt %) ¼ 100  (MA/rde), where M is the Whole-rock major and trace element
molecular weight of H2O (18·02), A is the height of the ab- geochemistry
sorption peak, r is the sample density (g l1), d is the Representative major and trace element compositions of
sample thickness (cm) and e is the molar absorption coeffi- the 1815 Tambora products are listed in Table 1 along with
cient (l mol1 cm1). Sample thicknesses (3 mm) were analyses of some older lava flows. Our dataset is comple-
measured using a Mitutoyo Digimatic Indicator. mented by major element analyses of a few accessory or ac-
Calculated glass densities (2575 g l1) are those for the cidental lithic clasts (xenoliths) from the 1815 deposits,
average 1815 pumice clast composition (see Table 1), assum- which, for the purpose of this study, are grouped together
ing a nonlinear temperature dependence of melt volume with the other pre-1815 eruptive products of the volcano.

279
JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 FEBRUARY 2012

(a) An In the total alkali vs SiO2 diagram (Fig. 3) most of the


Tambora rocks define a mildly alkaline basalt to trachyan-
desite (latite) series, although a few older samples, marked
by a bold outline, display a more ‘calc-alkaline’ character.
pheno-
crysts
The dominant Tambora compositions are similar to those
from other potassic Sunda arc volcanoes such as
75
rims cores
Sangeang Api, but are less K-rich than those from

Number of analyses
(n = 155) (n = 157) Muriah and Batu Tara (Stolz et al., 1988; Edwards et al.,
50 1991; van Bergen et al., 1992; Turner et al., 2003a). The 1815
products range between trachyandesite (latite) and tephri-
1000
25 phonolite, exhibiting a restricted range from 55 to 57 wt
% SiO2 (Table 1).
microlites 0 Excluding some samples with low alkali contents, most
900 30 40 50 60 70 80 90 100
major element oxides show well-defined and systematic

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


An (mol%)
800 correlations with SiO2 (Fig. 5). TiO2, Fe2O3*, CaO and
700
MgO abundances show an overall decrease with increas-
ing SiO2, whereas Na2O and K2O contents increase.
Ab Or The Al2O3 and P2O5 data are more scattered, although
there is a general increase at low SiO2 contents and
nearly constant concentrations at more elevated SiO2
(b) An contents for Al2O3 and decreasing abundances for P2O5.
56-63
The trace element patterns of the 1815 Tambora volcanic
An rocks normalized to normal mid-ocean ridge basalt
77-83
(N-MORB; Sun & McDonough, 1989) have a shape that
is similar to those for the products of the typical
calc-alkaline Sunda arc volcanoes from the main volcanic
front (e.g. Merapi; Gertisser & Keller, 2003) and magmas
from subduction-related tectonic settings in general
An An (Fig. 6). They are characterized by extreme enrichment in
77-85 57-61
large ion lithophile elements (LILE), U, Th and Pb,
and to a lesser extent in light rare earth elements (LREE)
relative to the heavy rare earth elements (HREE) and
high field strength elements (HFSE). Elements such as
Nb, Ta and Ti form distinct troughs in the trace element
400 µm patterns. The pre-1815 eruptive products display trace
element patterns similar to the juvenile 1815 material,
but have distinctly lower LILE contents (Fig. 6a).
In chondrite-normalized REE diagrams (Fig. 6b),
(c)
the Tambora rocks exhibit fractionated LREE and flat,
but enriched HREE patterns. The 1815 products display
MI LREE enrichment of 150^200 times chondritic values,
and (La/Sm)N and (La/Yb)N ratios of 4·3^4·8 and
An
11·5^12·8, respectively. The middle REE (MREE) and
72-86 HREE exhibit relatively flat patterns (GdN/YbN ¼1·8^
1·9). The REE concentrations of most of the pre-1815
MI

Isothermal sections of the dry ternary feldspar solvus are calculated


using SOLVCALC (Wen & Nekvasil, 1994). (See text for discussion.)
(b) Backscattered electron image of plagioclase crystals in the 1815
An55-63
200 µm Tambora products: Ca-rich (up to An91) compositions (light grey)
are preserved in the cores of larger plagioclase crystals. The variably
thick rims of these larger crystals and smaller plagioclases (darker
Fig. 4. Feldspars in the 1815 Tambora rocks. (a) Feldspar ternary grey) in the 1815 products are more Na-rich in composition (An58  6).
(Ab^An^Or) plot (filled symbols, phenocrysts; open symbols, (c) Melt inclusions (MI) in a zoned plagioclase from the 1815
microlites) and histogram of plagioclase compositions (An mol %). eruption.

280
GERTISSER et al. PETROGENESIS OF THE 1815 TAMBORA MAGMA

Post AD 1815
1.2 25
AD 1815
Pre AD 1815
1.0 23 Stage 2
(~60% cystallisation)

Al2O3 (wt.%)
TiO2 (wt.%)
(pl > cpx > mt)
21
0.8
19
0.6
17
Stage 1
0.4 15 (~30% cystallisation)
(ol > cpx > pl > mt)

0.2 13

13 10

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


8
Fe2O3* (wt.%)

MgO (wt.%)
10
6

4
7
2

4 0

13 6

11 5
Na2O (wt.%)
CaO (wt.%)

9
4
7
3
5

3 2

1 1

8 0.7

6
P2O5 (wt.%)
K2O (wt.%)

0.5

0.3
2

0 0.1
46 50 54 58 46 50 54 58
SiO2 (wt.%) SiO2 (wt.%)
Fig. 5. Major element variation diagrams vs SiO2 (wt %) for the Tambora volcanic suite. All analyses are normalized to 100 wt %, free of
volatiles. Data sources and symbols as in Fig. 3. Arrows in the Al2O3 vs SiO2 plot illustrate a two-stage, polybaric crystallization trend
from low-Al2O3 trachybasalt to Al2O3-rich trachybasalt^basaltic trachyandesite and trachyandesite. The amount of crystallization and the
fractionating mineral assemblages (pl, plagioclase; cpx, clinopyroxene; ol, olivine; mt, Ti-magnetite) were determined by simple least-squares
mixing calculations.

281
JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 FEBRUARY 2012

1000
1815 eruption yielded similar 87Sr/86Sr and 143Nd/144Nd
AD 1815 ratios of 0·70390 and 0·51291, respectively (Turner &
Pre AD 1815
Foden, 2001). Our ranges for the entire Tambora suite are
100 87
Sr/86Sr ¼ 0·70398^0·70461 and 143Nd/144Nd ¼ 0·51274^
Rock / N-MORB

0·51291 (Table 1), similar to previously published data


(87Sr/86Sr ¼ 0·70385^0·70400; 143
Nd/144Nd ¼ 0·51281^
10 0·51283) (Whitford et al., 1978; Varne & Foden, 1986;
Grall-Johnson, 1997; Turner & Foden, 2001). There is a re-
stricted range of 176Hf/177Hf ratios of 0·28308^0·28317 for
1 both the 1815 eruptive products and the Tambora suite as
a whole (Table 1). In the 143Nd/144Nd vs 87Sr/86Sr diagram
(a) (Fig. 7a), the Tambora rocks form a cluster below the
0.1 Indian Ocean MORB field and within the Indian Ocean
Cs Ba U Ta La Pb Mo P Sm Hf Ti Tb Y Er Yb ocean-island basalt (OIB) field. Compared with other po-

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


Rb Th Nb K Ce Pr Sr Nd Zr Eu Gd Dy Ho Tm Lu
tassic Sunda arc volcanoes, the Tambora rocks are charac-
terized by the highest 143Nd/144Nd ratios and the least
250 radiogenic 87Sr/86Sr ratios (Fig. 7b). In Hf^Nd isotopic
space (Fig. 8a), the Tambora rocks have amongst the high-
100 est Hf (and Nd) isotopic ratios of all Indonesian arc vol-
canic rocks and are close to, but distinct from, the Indian
Rock / Chondrite

Ocean MORB and OIB fields. Enlargement of the


176
Hf/177Hf vs 143Nd/144Nd diagram (Fig. 8b) shows the
separate fields for Indian and Pacific mantle domains
10 (e.g. Nowell et al., 1998; Pearce et al., 1999; Chauvel &
Blichert-Toft, 2001) and the primitive character of the
Tambora suite relative to most other Sunda arc volcanoes,
including Muriah, the only other potassic Sunda arc
(b) volcano for which comparable data are available (White
1 & Patchett, 1984; Woodhead et al., 2001).
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
Fig. 6. (a) N-MORB-normalized trace element and (b)
chondrite-normalized REE patterns for the Tambora volcanic
Major element and volatile concentrations
products. A calc-alkaline basaltic andesite from Merapi volcano in in melt inclusions and groundmass glass
the central portion of the Sunda arc with a SiO2 content similar to Among the crystal phases in the 1815 pumices, only
the 1815 Tambora products (Gertisser & Keller, 2003) is shown for
comparison (grey line). The normalizing values are from Sun & plagioclase contains abundant glassy melt inclusions
McDonough (1989). (Fig. 4c). Other crystals are either free of melt inclusions
or bear inclusions that are partly crystallized. The melt in-
clusions in plagioclase tend to be round or irregular
eruptive products are lower than those of the 1815 products, in shape, with sizes ranging from 510 mm to 50 mm.
reflecting their less evolved or calc-alkaline character, Trapped in calcic plagioclase cores as well as more sodic
or both. The analysed samples display variable, but lower crystals or crystal zones, these inclusions provide a record
LREE enrichment when compared with the 1815 volcanic of the complex evolutionary history of the 1815 Tambora
products [(La/Sm)N ¼ 2·4^3·0], lack of a pronounced magma. Those trapped near the plagioclase crystal rims
Eu anomaly and HREE concentrations of 410 times were previously considered to be the most representative
chondritic values that are similar to the products from the of the pre-eruptive melt composition and used to assess
1815 eruption (Fig. 6b). The latter are also more the amount of volatiles released during the 1815 eruption
LREE-enriched than the previously mentioned Merapi (Self et al., 2004).
calc-alkaline basaltic andesite. Major element concentrations in plagioclase-hosted melt
inclusions in the 1815 products span a continuous compos-
itional range from latite to tephriphonolite (55^59 wt %
Whole-rock isotope geochemistry SiO2). The major element compositions follow and extend
The 87Sr/86Sr and 143Nd/144Nd ratios of the 1815 Tambora the trend displayed by the 1815 whole-rock compositions,
pumices are fairly constant, in the range of 0·70398^ except for TiO2 and P2O5, which are slightly higher
0·70410 and 0·51274^0·51282, respectively (Table 1). A previ- than in the whole-rocks at similar SiO2 contents (Fig. 9).
ously published Sr isotope analysis for a sample from the Groundmass glass compositions vary between 58 and

282
GERTISSER et al. PETROGENESIS OF THE 1815 TAMBORA MAGMA

0.5134
(a) Post AD 1815
0.5132 FMM AD 1815
Indian 2% 5% 10%
AOC Pre AD 1815
MORB
0.5130
143Nd/144Nd
Sunda arc (excl. Toba)
0.5128 1%
1%
Tambora Sed-3
bulk earth
0.5126 Indian OIB
1% 1%
Banda arc

0.5124 1% A

Global marine
0.5122 sediments
Sed-1, 2

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


D C
B
0.5120

bulk earth
Sed-4

0.5118
0.700 0.705 0.710 0.715
87Sr/86Sr

0.5129
Tambora

0.5128 Ringgit-Beser
Sangenges
143Nd/144Nd

0.5127 Watukubu
bulk earth
Sangeang Api

0.5126 Muriah
Batu Tara

Soromundi
0.5125 Mantle
Array

0.5124
bulk earth

(b)
0.5123
0.7035 0.7045 0.7055 0.7065 0.7075
87Sr/86Sr
Fig. 7. (a) 87Sr/86Sr vs 143Nd/144Nd diagram showing data for Tambora, Indian Ocean MORB (I-MORB) and OIB, altered oceanic
crust (AOC), volcanic rocks from the Sunda and Banda arcs, and modern marine sediments. Three deep-sea turbidites from the Java trench
(Sed-1^3) and an Indian Ocean pelagic sediment (terrigenous clay) composition (Sed-4) are highlighted (McLennan et al., 1990; Vervoort
et al., 1999). Also illustrated are mixing curves between I-MORB mantle and sediments (mixing arrays A and B) and between a fluid-modified
I-MORB mantle (FMM) and selected sediment end-members (mixing arrays C and D). The ticks along the mixing curves indicate the
per cent of sediment in the mixture. Data sources: MORB, Indian Ocean, from Hamelin & Alle'gre (1985), Ito et al. (1987), Rehka«mper &
Hofmann (1997) and Chauvel & Blichert-Toft (2001); OIB, Indian Ocean, from White & Hofmann (1982), Hamelin et al. (1985), Dosso et al.
(1988), Barling & Goldstein (1990) and Yang et al. (1998); AOC from Staudigel et al. (1995); Sunda and Banda arcs from Whitford et al. (1981),
White & Patchett (1984), Varne & Foden (1986), Stolz et al. (1988, 1990), Edwards et al. (1991, 1993, 1994), Gerbe et al. (1992), Hoogewerff et al.
(1997), Turner & Foden (2001), Woodhead et al. (2001), Gertisser & Keller (2003), Turner et al. (2003a) and Handley et al. (2007, 2008, 2010, 2011);
global marine sediments from Ben Othman et al. (1989). End-member compositions used in the mixing calculations: I-MORB source mantle:
0·1 I-MORB values from Chauvel & Blichert-Toft (2001); AOC and AOC-derived fluid: after Handley et al. (2007); FMM: 3% AOC fluid^
97% I-MORB source; sediments: as above. (b) Enlargement of the 87Sr/86Sr vs 143Nd/144Nd diagram showing data from Tambora relative to
other potassic Sunda arc volcanoes and the mantle array. Published data sources: Tambora from Varne & Foden (1986), Grall-Johnson (1997)
and Turner & Foden (2001); Batu Tara from Stolz et al. (1988, 1990) and Hoogewerff et al. (1997); Muriah from Whitford et al. (1981) and
Edwards et al. (1991); Ringgit-Beser from Edwards et al. (1993, 1994); Sangeang Api from Turner et al. (2003a); Sangenges and Watukubu from
Edwards et al. (1993); Soromundi from Varne & Foden (1986). Symbols for Tambora as in Fig. 3.

283
JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 FEBRUARY 2012

0.2834 (a) Indian MORB

Tambora
0.2832
Sunda arc

176Hf/177Hf
0.2830 Sed-3

Indian OIB
0.2828 B
A Banda arc
(Serua andesite)
0.2826
Global
marine sediments
Sed-4
0.2824 AD 1815

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


Sed-1
Pre AD 1815
Sed-2
0.2822
0.5118 0.5123 0.5128 0.5133
143Nd/144Nd

0.2833
Indian mantle Galunggung
Semeru Batur
Merapi Tambora
Krakatau Guntur
0.2831
176Hf/177Hf

Salak Ijen
Ungaran Lamongan
Sumbing

Gede
0.2829 Muriah

Papandayan

(b) Pacific mantle


0.2827
0.5124 0.5126 0.5128 0.5130
143Nd/144Nd
Fig. 8. (a) 176Hf/177Hf vs 143Nd/144Nd diagram showing data from Tambora in relation to Indian Ocean MORB (I-MORB) and OIB, volcanic
rocks from the Sunda and Banda arcs, and global marine sediments. Three deep-sea turbidites from the Java trench (Sed-1^3) and an Indian
Ocean pelagic (terrigenous) clay (Sed-4) are also shown (see Fig. 7). Mixing curves labelled A and B are bulk mixing arrays between
I-MORB mantle and selected sediment end-members. The ticks along the mixing curves indicate the per cent of sediment in the mixture.
Data sources: MORB, Indian Ocean, from Salters (1996), Nowell et al. (1998), Chauvel & Blichert-Toft (2001) and Kempton et al. (2002); OIB,
Indian Ocean, from Patchett (1983), Salters & Hart (1991) and Salters & White (1998); Sunda and Banda arcs from White & Patchett (1984),
Pearce et al. (1999), Woodhead et al. (2001) and Handley et al. (2007, 2008, 2011); global marine and highlighted regional (Indian Ocean)
sediments from Vervoort et al. (1999). End-member compositions (including data sources) used in the mixing calculations as in Fig. 7.
(b) Enlargement of the 176Hf/177Hf vs 143Nd/144Nd diagram showing data from Tambora relative to other Sunda arc volcanoes (Fig. 1).
The dashed line illustrates the boundary between Indian and Pacific mantle and is adapted from Pearce et al. (1999). Symbols for Tambora as
in Fig. 3.

60 wt % SiO2, reaching phonolitic compositions. They Melt inclusions in the more sodic plagioclase rims
overlap with the most evolved melt inclusion compositions and compositionally similar cores consistently contain
in the same sample and the 1815 products as a whole, 689  80 ppm sulphur and 1720 169 ppm chlorine.
when compared on an anhydrous basis (Fig. 9). In contrast, groundmass glasses contain lower average

284
GERTISSER et al. PETROGENESIS OF THE 1815 TAMBORA MAGMA

AD 1815 melt inclusions


1.2 AD 1815 groundmass glass 22.5
AD 1815 whole rocks
1.0 21.5

Al2O3 (wt.%)
TiO2 (wt.%)

0.8 20.5

0.6 19.5

0.4 18.5

0.2 17.5

8 3.0

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


7
Fe2O3* (wt.%)

2.5

MgO (wt.%)
6
2.0
5
1.5
4

3 1.0

6 7

6
5
Na2O (wt.%)
CaO (wt.%)

5
4
4
3
3

2 2

8 0.9

7
0.7
P2O5 (wt.%)
K2O (wt.%)

6
0.5
5
0.3
4

3 0.1
53 54 55 56 57 58 59 60 53 54 55 56 57 58 59 60
SiO2 (wt.%) SiO2 (wt.%)
Fig. 9. Major element variation diagrams vs SiO2 (wt %) for melt inclusions in calcic plagioclase cores (open diamonds) and more sodic
plagioclase rims (grey-shaded diamonds) plus groundmass glass of pumices (filled triangles) from the 1815 Tambora eruption. Small open circles
are 1815 whole-rock compositions from Fig. 5. All analyses are normalized to 100 wt %, free of volatiles.

285
JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 FEBRUARY 2012

1600
25 S (ppm) plagioclase cores and more sodic outer crystal zones range
(a) 1200
from 1·8 to 2·5 wt %, averaging 2·1 0·2 wt % (1s;
n ¼10). Obsidian fragments of similar composition to the
20 1815 pumices are essentially degassed, containing between
Number of Analyses

800
0·11 and 0·15 wt % H2O (n ¼ 3). No systematic correlation
400
between water content in the melt inclusions and host
54 55 56 57 58 59 60 plagioclase compositions has been observed.
15 SiO2 (wt.%)
Groundmass U^Th^Ra isotopic data for 1815 pumices
glass
and constituent minerals
10 Melt Whole-rock and mineral U^Th^Ra data for the 1815
inclusions
pumices are reported in Table 2. The whole-rock data
vary in the range (230Th/232Th) ¼ 0·78^0·90 and
5 (238U/232Th) ¼ 0·84^0·91, and lie within the range of pub-

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


lished data for the 1815 deposits (Gill & Williams, 1990;
Turner & Foden, 2001). 226Ra concentrations in the 1815
0 whole-rocks vary widely from 1200 to 2080 fg g1 and
(226Ra/230Th) varies from 1·15 to 1·99 (Table 2). Turner &
0 350 700 1050 1400
Foden (2001) reported a 226Ra concentration of 1279 fg g1
Sulphur (ppm) and a (226Ra/230Th) ratio of 1·20 for a single 1815 sample.
3000
Plagioclase, clinopyroxene, biotite and titanomagnetite
Cl (ppm) display both 238U and 230Th excesses, with (238U/232Th)
2500 and (230Th/232Th) ratios ranging from 0·69 to 1·09 and
2000
35 0·72 to 1·01, respectively (Table 2; Fig. 11a^c). The U and
Th contents of these minerals are much lower than in the
Number of Analyses

1500 30 whole-rock and groundmass glass. Also, with the exception


1000
of the plagioclase separate from sample T57/58, all mineral
54 55 56 57 58 59 60 25 phases are characterized by lower Ra concentrations than
SiO2 (wt.%) the whole-rock or groundmass glass. 226Ra^230Th disequi-
20 libria in the analysed mineral separates vary widely:
(226Ra/230Th) ¼ 0·57^8·32 (Table 2; Fig. 11d).
15 Groundmass glass compositions in three pumice samples
Melt show 238U excesses up to 15%, with (238U/232Th) and
10 Groundmass inclusions (230Th/232Th) ratios ranging from 0·82 to 0·86 and 0·72 to
glass
0·79, respectively. Two of the groundmass separates plot
5 close together and below the constituent mineral phases
(b) in the U^Th equiline diagram (Fig. 11b and c). The 226Ra
0 concentrations of the groundmass glasses vary from 1025
0 700 1400 2100 2800 to 1166 fg g1, with a spread of (226Ra/230Th) ratios
between 1·00 and 1·55 (Table 2; Fig. 11d).
Chlorine (ppm)
Fig. 10. Histograms of (a) sulphur and (b) chlorine concentrations in
plagioclase-hosted melt inclusions and groundmass glass in the 1815
DISCUSSION
eruptive products. The insets show sulphur and chlorine concentra- The role of crustal contamination in the
tions in melt inclusions in calcic plagioclase cores (open symbols) petrogenesis of the Tambora magmas
and more sodic plagioclase rims (grey-shaded symbols) as a function
of SiO2 content. The Tambora rocks appear to be ultimately derived from
a MORB-like mantle source (see below) and, therefore,
the slight shift in isotopic ratios from the MORB field
volatile concentrations (as a result of syn-eruptive degas- towards more ‘crustal’ values (Figs 7 and 8) may reflect
sing) of 290 74 ppm sulphur and 1511 132 ppm chlorine recycling of a crustal (subducted sediment) component
(see Self et al., 2004; Table 4). The more primitive inclusions into the mantle wedge or contamination by the arc crust
in the calcic plagioclase cores tend to be more volatile-rich, during magma ascent, or both. Owing to the younger age
with sulphur and chlorine concentrations up to 1255 and and ‘transitional to oceanic’ character of the overriding
2440 ppm, respectively (Table 4; Fig. 10). Water concentra- plate in the central and eastern portions of the Sunda
tions in a few analysed melt inclusions in both calcic arc compared with the western Sunda arc (Hamilton,

286
GERTISSER et al. PETROGENESIS OF THE 1815 TAMBORA MAGMA

1.15 e 1.15
(a) T39 ilin (c) T7 ine
u uil
Eq Eq
1.05 1.05
(230Th/232Th)

(230Th/232Th)
0.95 0.95
wr

0.85 mt 0.85 mt bio


gm pl
pl
cpx bio wr Sample: T39
0.75 f(x) = -0.0093x + 0.7992 0.75
Equiline intercept = 0.7918
Slope age: 0 a gm

0.65 0.65
0.65 0.75 0.85 0.95 1.05 1.15 0.65 0.75 0.85 0.95 1.05 1.15
238 232
( U/ Th) 238 232
( U/ Th)

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


1.15 10
e

Equil.
(b) T57/58 in (d)
uil T7
Eq T39
cpx T57/58
1.05 8 bio
cpx

(226Ra/230Th)
(230Th/232Th)

pl
0.95 6

0.85 4
mt pl
wr
bio wr
0.75 2 wr gm
gm mt
mt mt gm Equil.
pl pl wr gm
bio
0.65 0
0.65 0.75 0.85 0.95 1.05 1.15 0.5 0.7 0.9 1.1 1.3 1.5
(238U/232Th) (238U/230Th)
226 230 238 230
Fig. 11. (a^c) U^Th equiline diagrams and (d) ( Ra/ Th) vs ( U/ Th) plot for whole-rock, mineral and groundmass glass data from
three Tambora 1815 samples. The dashed line in (a) represents an isochron calculated for sample T39 that is within uncertainty of ‘zero’ age.
Mean analytical errors are smaller than the symbol size. wr, whole-rock; gm, groundmass glass; cpx, clinopyroxene; bio, biotite; mt,
Ti-magnetite; pl, plagioclase.

1979), assimilation and fractional crystallization (AFC) Harmon & Hoefs, 1995]. Whole-rock d18O values for
processes (DePaolo, 1981) in these parts of the arc have Tambora (Grall-Johnson, 1997) range between þ5·0 and
proved difficult to detect (see Turner & Foden, 2001; þ7·1ø (SMOW), although the majority of the samples lie
Gertisser & Keller, 2003; Turner et al., 2003a; Chadwick between þ5·6 and þ6·0ø (SMOW) and are indistin-
et al., 2007; Handley et al., 2007). guishable from the mantle. This means that it is unlikely
The absence of correlations between Sr, Nd and Hf that the Tambora magmas have been contaminated by
isotope ratios and indices of differentiation (e.g. SiO2; upper crustal material or, at least, that any upper crustal
Fig. 12) and the restricted ranges of these isotope ratios in component is relatively minor. By contrast, some level of
the Tambora products (with the exception of a few of magma interaction with lower crustal lithologies cannot
the older samples) suggest negligible input of isotopically be ruled out, although such a process may result in a
distinct crust during differentiation. However, as the wider than observed range in d18O values. Accordingly, it
basement of Sumbawa at the eastern edge of the Sunda is assumed that interaction of the Tambora magmas with
arc is thought to comprise relatively thin (14^17 km), the arc crust, if it occurred at all, has had a negligible
compositionally immature crust (Hamilton, 1979), the iso- effect on their geochemical and isotopic composition. It
topic composition may be similar to that of the underlying appears likely that the relatively thin eastern Sunda arc
mantle source. Therefore, assimilation of this crustal crust on which the volcano is built does minimize the likeli-
material is difficult to distinguish using radiogenic isotope hood of shallow-level crustal contamination (compared
systems alone. Upper crustal material typically possesses with the central and, in particular, the western Sunda
elevated d18O values that are distinct from mantle values arc). Thus, mantle source heterogeneity through recycling
and basaltic oceanic crust [d18O ¼ þ5·7  0·2ø (SMOW); of a crustal component into the mantle wedge beneath

287
JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 FEBRUARY 2012

Post AD 1815 MORB-type mantle source (Indian-type or I-MORB)


0.7048 AD 1815 overprinted to various degrees by slab-derived fluid and
(a) Pre AD 1815 sediment components (e.g. Stolz et al., 1990; Edwards
FC et al., 1994; Hoogewerff et al., 1997; Turner & Foden, 2001;
0.7044 Gertisser & Keller, 2003; Handley et al., 2007). The origin
87Sr/86Sr

of the less voluminous potassic alkaline Sunda arc rocks,


typically found at volcanoes situated farthest away from
the trench in the rear arc (Fig. 1), is less clear. Contrasting
0.7040
models involving MORB-type (e.g. Turner et al., 2003a)
and metasomatized, enriched (or plume-type) mantle
source components (e.g. Wheller et al., 1987; Stolz et al.,
0.7036 1990; Edwards et al., 1991, 1994; van Bergen et al., 1992),
magma interaction with phlogopite-bearing lithospheric
0.5130 or ancient sub-continental mantle (Varne, 1985; Edwards

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


(b) et al., 1991) and incongruent breakdown of amphibole
(Foden & Green, 1992) have been proposed. Moreover, it
143Nd/144Nd

0.5129 has been suggested that the potassic rear-arc volcanic


rocks reflect a higher subducted sediment contribution,
a weaker slab-derived fluid signal and a lower degree of
partial melting compared with the magmas formed along
0.5128 the main volcanic front (Whitford & Nicholls, 1976;
Hoogewerff et al., 1997; Turner & Foden, 2001).

0.5127
Mantle source characteristics
The relative immobility of the HREE and HFSE (com-
0.28322 pared with LREE and LILE) during slab dehydration
(c) (Pearce, 1983) and their comparably low abundances in
crustal rocks (Taylor & McLennan, 1985) make these
0.28317 elements (and their associated isotopic systems) useful for
176Hf/177Hf

placing constraints on the composition of the mantle


0.28312 wedge prior to addition of slab-derived components.
The HREE concentrations of the Tambora products are
10^20 times chondritic values and although the absolute
0.28307 concentrations are affected by the evolved character of
the magmas, they show unfractionated HREE patterns
(Fig. 6). This implies that garnet is not a significant residual
0.28302
46 48 50 52 54 56 58 mineral in the mantle source and that the Tambora
magmas are derived from mantle above the garnet stabil-
SiO2 (wt.%) ity field for wet peridotitic mantle. Turner & Foden (2001)
Fig. 12. Plots of (a) Sr/ Sr, (b) 143Nd/144Nd and (c) 176Hf/177Hf vs
87 86 and Turner et al. (2003a) used a plot of Tb/Yb vs La/Yb
SiO2 used to assess the role of crustal contamination beneath and partition coefficients for clinopyroxene on the mantle
Tambora. General trends are shown for fractional crystallization solidus (Blundy et al., 1998) to model melting processes
(FC) and assimilation and fractional crystallization (AFC). and the mantle source compositions of a range of Sunda
The latter can be positive or negative depending on the isotopic
composition of the (crustal) contaminant. Published Tambora data arc magmas, including those of the potassic Sangeang Api
are from Whitford et al. (1978), Varne & Foden (1986), Grall-Johnson volcano situated at approximately the same distance to
(1997) and Turner & Foden (2001). Symbols as in Fig. 3. the underlying Benioff Zone as Tambora. The uniform
La/Yb (16·0^17·8) and Tb/Yb (0·29^0·33) ratios of the
Tambora is considered to exert the prime control on the Tambora 1815 rocks (see Table 1) suggest an essentially
observed geochemical and radiogenic isotope variations garnet-free mantle source and a small degree of partial
at Tambora. melting (52%). The two lavas of more ‘calc-alkaline’
affinity have lower La/Yb (5·8^6·4) and Tb/Yb (0·28^
Magma source components and deep 0·29) ratios, which fall within the range of other Sunda
(mantle) processes arc front lavas (Turner & Foden, 2001) and imply higher
Studies of the Sunda arc have highlighted that magmas degrees of partial melting of the same source as the
erupted along the main volcanic front are derived from a 1815 rocks.

288
GERTISSER et al. PETROGENESIS OF THE 1815 TAMBORA MAGMA

Ratios of Zr/Nb of 15^23 and Ta/Nb of 0·05^0·10 (from a sediment-modified mantle source with isotopic charac-
the data in Table 1) vary relatively little within the teristics similar to those of the Tambora rocks.
Tambora suite. Ta/Nb ratios are similar to or slightly Elevated Ba/La ratios in subduction-related igneous
higher than those of MORB (both N- and I-type) and rocks when compared with MORB are widely interpreted
OIB and, therefore, in some samples non-chondritic. In to reflect fluid addition to the mantle wedge from dehydra-
contrast, Zr/Nb ratios are distinctly higher than OIB, tion of the subducting slab (e.g. Ben Othman et al., 1989;
lower than N-MORB and more closely resemble I-MORB Elliott et al., 1997). Ba/La ratios of 27^30 in the 1815
(Chauvel & Blichert-Toft, 2001; Handley et al., 2007). Tambora samples are significantly higher than those
This indicates that the mantle beneath Tambora prior to observed in I-MORB [Ba/La ¼ 7 4 (n ¼ 8); Chauvel &
the influence of subduction-related components is slightly Blichert-Toft, 2001]. As Ba is not expected to be fractionated
depleted, but otherwise similar to the source of I-MORB from La during differentiation, the Ba/La ratios in the
(Woodhead et al., 1993; Handley et al., 2007). As Hf and 1815 trachyandesite strongly imply that slab-derived fluids
Nd are thought to be relatively immobile in slab-derived have been added to the mantle source, possibly from
fluids (e.g. McCulloch & Gamble, 1991; Pearce et al., 1999; altered oceanic crust, or both altered crust and subducted

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


see also Woodhead et al., 2001), their isotopic ratios in the sediment (see Gertisser & Keller, 2003). In Fig. 13, the
Tambora rocks (Fig. 8) place further constraints on their Tambora samples lie below the mantle^bulk sediment
mantle source. Both 176Hf/177Hf and 143Nd/144Nd isotopic mixing curve. 87Sr/86Sr ratios are slightly higher than the
ratios are among the highest (i.e. the most primitive) of inferred I-MORB-like mantle source and most samples
all Sunda arc volcanic rocks (Handley et al., 2007, 2008, display higher Ba/La ratios than most subducted sediment
2011) and lie within the Hf isotopic range of I-MORB. compositions in the Sunda arc. This can be readily
The field for Tambora (as well as the Sunda arc as a explained by addition of a low (but higher than the
whole) is also, for the most part, distinct from the Indian mantle source) 87Sr/86Sr, high Ba/La fluid component, con-
OIB field in Hf^Nd space (Fig. 8). Thus, involvement of sistent with fluids derived from subducting altered oceanic
an enriched or plume component in the underlying crust (AOC), as previously suggested for other Sunda arc
mantle wedge is not apparent. Furthermore, it can be volcanoes (Turner & Foden, 2001; Turner et al., 2003a;
inferred from the above that the Tambora source appears Handley et al., 2007).
to be one of the least affected in the Sunda arc by crustal To corroborate the interpretation of fluid addition based
inputs from either the slab or arc lithosphere. Volcanic on Ba/La ratios, a more complex mixing scenario, which
rocks from Tambora and other centres on Bali and Java lie involves an additional, isotopically distinct component
above the dividing line that separates Indian and Pacific (i.e. AOC-derived fluid) in the source, is tested in Nd^Sr
mantle domains in Hf^Nd isotope space (Pearce et al., isotope space (Fig. 7). Interaction of seawater with basalt
1999; Fig. 6), implying derivation from an I-MORB-like produces significant changes in 87Sr/86Sr ratios, whereas
mantle source and corroborating conclusions based on
ratios of immobile trace elements.
0.7400
Slab-derived components Sed-2
Sed-1 Indian Ocean
marine sediments
As outlined above, the displacement of the Tambora
volcanic rocks to lower 143Nd/144Nd (and higher 87Sr/86Sr) 0.7300
ratios with respect to I-MORB (Fig. 7a) requires some
87Sr/86Sr

subducted sediment component in the magma genesis in


0.7200
addition to the dominant I-MORB mantle wedge source.
Sed-4
To test this hypothesis, mixing models between I-MORB
source mantle and selected sediments are illustrated in the 0.7100
143
Nd/144Nd vs 87Sr/86Sr and 176Hf/177Hf vs 143Nd/144Nd Sed-3
1815 Tambora pumice
OIB
diagrams (Figs 7a and 8a). Although subducted sediment
compositions vary considerably along the arc, only a few I-MORB source Fluid
0.7000
end-member compositions are available for petrogenetic 0 10 20 30 40
modelling in both Nd^Sr and Hf^Nd isotope space. A Ba/La
reasonable model fit can be obtained for Tambora by Fig. 13. Plot of 87Sr/86Sr vs Ba/La for Tambora with bulk mixing lines
two-component mixing between I-MORB source mantle between an Indian Ocean MORB-like (I-MORB) mantle source
and bulk subducted Indian Ocean sediments (e.g. mixing and selected regional sediment compositions. Average I-MORB and
array B; Figs 7a and 8a), although this depends significant- OIB data are from Chauvel & Blichert-Toft (2001) and Sun &
McDonough (1989); the sediment compositions are from McLennan
ly on the sediment end-member composition used. Small et al. (1990) and Vervoort et al. (1999) (see Figs 7a and 8a). Symbols
amounts of sediment (51%) in the mixture can produce for Tambora as in Fig. 3.

289
JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 FEBRUARY 2012

143
Nd/144Nd isotopes are essentially unaffected and remain data place further constraints on source processes.
indistinguishable from fresh MORB (Hart et al., 1974; The Tambora whole-rock data display small 238U
White & Patchett, 1984; Staudigel et al., 1995). Because of [(230Th/238U) ¼ 0·90^0·99] and 226Ra [(226Ra/230Th)  2]
the lack of isotopic data reported for altered Indian excesses (Table 2; Fig. 14). Such excesses are widespread in
Ocean MORB, data for Atlantic MORB (Staudigel et al., arc lavas and inferred to be related to the addition of
1996) have been used to evaluate the effect of adding this fluids released from the down-going slab (Turner et al.,
component to the mantle wedge beneath Tambora. 2003b, and references therein). For the Sunda arc front
I-MORB source mantle metasomatized with volcanoes, Turner & Foden (2001) noted a positive correl-
AOC-derived fluids will create a mixture with higher Sr ation between (226Ra/230Th) and (238U/230Th), an observa-
isotope ratios and similar Nd isotope ratios to I-MORB. tion that also appears to hold overall for the potassic
As the trajectory of the Tambora data crosscuts any con- rear-arc volcanoes (Fig. 14b). The extent of U^Th disequi-
ceivable two-component mixing curve produced between librium has been attributed to a constant fluid flux
the mantle source and an AOC^sediment mixture, a buffered by the Th concentration of the mantle source,
two-stage model is explored, in which an I-MORB source which is controlled by the amount of sediment added to

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


mantle is contaminated by fluids derived from AOC and the mantle wedge. As the amount of source contamination
subsequently by a sediment component. To create the by subducted sediment beneath Tambora is thought
fluid-modified mantle component 3% fluid from the
altered crust is required in the model. The mixing propor-
tions of AOC and I-MORB are highly dependent on the 1.00

ne
Sr and Nd abundances and isotopic ratios of the altered (a)

uili
Eq
crust, which are extremely variable. The fraction of fluid 0.95
Batu Tara
required is smaller if more highly altered basalt compos- Sangeang Api
itions are used and greater if a partial melt or bulk (230Th/232Th) 0.90 T07
Tambora (this study)
mixing of AOC is considered. The two-stage, Tambora (published)
three-component model produces a good fit to the 0.85
Tambora data, with only a small proportion of 51% sedi- T57/58
ment required to generate the observed isotopic values. 0.80
Such a model explains slightly better the relations dis- T39
played in Fig. 7a, although AOC is not absolutely required 0.75
by the modelling. The model proposed for Tambora is simi-
lar to that of Handley et al. (2007) for Ijen volcano in east- 0.70
ern Java. 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
Modelling of two-stage source contamination in Hf^Nd
(238U/232Th)
isotope space is hindered by the lack of Hf isotope data
for AOC. However, as Nd isotope ratios remain relatively 4
constant during hydrothermal and low-temperature
Equil.

(b)
seawater alteration of MORB, Hf isotope ratios are also
expected to be unaffected by alteration of oceanic crust 3
(226Ra/230Th)

(White & Patchett, 1984). In these circumstances, contam-


ination of the mantle source by a slab-derived fluid from
AOC would be undetectable using Hf and Nd isotopes, 2 T07
and the isotopic compositions would be the same for an
I-MORB source, with or without added AOC-derived T39
T57/58
fluids. The mixing model between an average I-MORB 1
Equil.
source and the same sediment end-members used in
Nd^Sr isotope space requires 51% addition of sediment
to explain the Tambora data (Fig. 8a). Such a value repre- 0
sents an upper limit, if a sediment partial melt (rather 0.7 0.8 0.9 1.0 1.1 1.2 1.3
than bulk sediment) is added to the mantle source (see
Turner & Foden, 2001; Gertisser & Keller, 2003; Turner (238U/230Th)
et al., 2003a; Handley et al., 2007). Fig. 14. (a) U^Th equiline diagram and (b) (226Ra/230Th) vs
(238U/230Th) plot for whole-rocks from the 1815 Tambora eruption
Source processes: constraints from U-series isotopes compared with potassic volcanic rocks from Batu Tara (Hoogewerff
et al., 1997) and Sangeang Api (Turner et al., 2003a). The published
Having identified the source components involved in Tambora analyses are taken from Gill & Williams (1990) and Turner
the petrogenesis of the Tambora magmas, the U-series & Foden (2001).

290
GERTISSER et al. PETROGENESIS OF THE 1815 TAMBORA MAGMA

to be small (see above), the 238U excesses and similar In detail, the liquid line of descent is segmented and char-
(238U/232Th) ratios in the Tambora whole-rocks are inter- acterized by an early fractionation (stage 1) of predomin-
preted to reflect primarily a small and, apparently, antly olivine and clinopyroxene, giving rise to the
constant fluid signal in the source. Minor 238U excesses at evolution from primitive, silica-undersaturated, potassic
Tambora contrast with U^Th isotope data for other and relatively Al2O3-poor trachybasalt to more differen-
potassic Sunda arc lavas from Batu Tara (Hoogewerff tiated, Al2O3-rich trachybasalt and basaltic trachyandesite
et al., 1997) and Sangeang Api (Turner et al., 2003a), which (shoshonite). Petrographic evidence (Foden, 1986) and
exhibit both relatively small 238U and more pronounced simple least-squares mixing calculations (Fig. 5) suggest
230
Th excesses (Fig. 14a). At Sangeang Api, the 230Th that small amounts of plagioclase, titanomagnetite and
excesses are thought to be related to the effects of partial biotite crystallized during this early stage, although the
melting of the mantle wedge that exceed those of fluid amount of plagioclase must have been small as Al2O3
addition in the presence of residual garnet in the source increases with SiO2. During stage 1, TiO2, Fe2O3*, MgO
(Turner et al., 2003a). and to some extent CaO are continuously depleted. An
An alternative model has been proposed for the Kurile indistinct compositional gap (Fig. 5) separates differenti-

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


rear-arc, where slab-derived, high-Th fluids are invoked to ation stage 1 from a stage 2 in which plagioclase becomes
account for 230Th excesses in alkaline lavas from Rishiri more important in the fractionating mineral assemblage
volcano situated 300 km above the subducting slab and Al2O3 remains constant with increasing SiO2 during
(Kuritani et al., 2008). At Tambora, preservation of a fluid evolution from Al2O3-rich basaltic trachyandesite to tra-
flux signal with slight 238U excesses is consistent with the chyandesite (Figs 3 and 5). The continuation of decreasing
apparent lack of residual garnet in the source and a depth trends for TiO2, Fe2O3*, MgO (with a less steep negative
to the underlying Benioff Zone (180 km) that is less than slope) and CaO indicates that clinopyroxene, titanomag-
at Rishiri and other rear-arc volcanoes, which characteris- netite and biotite continue to crystallize during stage 2.
tically exhibit 230Th excesses when the depths to the However, unlike Fe2O3*, MgO and CaO, TiO2 is
subducting slab exceeds 250 km (Turner et al., 2003b, displaced to slightly higher values among the 1815 rocks
and references therein). In Fig. 14a, the Tambora and compared with the main trend, which could be explained
some of the Batu Tara whole-rocks are displaced to higher by a decrease in the Ti content of magnetite with decreas-
(230Th/232Th) compared with the horizontal array of ing temperature. A peak in P2O5 contents (Fig. 5) shows
Sangeang Api and the Batu Tara samples that exhibit that the Tambora magmas became saturated with apatite
230
Th excesses. The latter also plot closer to the equiline at 55 wt % SiO2. Melt inclusions and groundmass glass
than the Sangeang Api samples with 238U excesses. This compositions in the 1815 eruptive products further
may indicate that the source region beneath Tambora and constrain the differentiation of the Tambora magmas. The
in the eastern portion of the Sunda arc is heterogeneous melt inclusion compositions generally follow and extend
with respect to (230Th/232Th) and suggest that the 1815 the trend displayed by the 1815 whole-rocks towards
Tambora magmas evolved from various primary magma higher SiO2 contents (Fig. 9). Groundmass glass compos-
batches that subsequently followed similar liquid lines of itions of the 1815 pumices overlap with the more SiO2-rich
descent. The heterogeneity, with respect to (230Th/232Th), melt inclusion compositions in the plagioclase rims. The
of the mantle wedge may reflect the composition of the 1815 groundmass glasses are phonolitic in composition
sediment component added to the mantle source in this (58^60 wt % SiO2) and saturated in ternary feldspar
part of the arc, although minor increases in (230Th/232Th) (anorthoclase) (Fig. 4a). They are the most evolved
may have occurred during subsequent differentiation Tambora compositions and the best representation of
processes (see below). the melt composition erupted in 1815.
In addition, the melt inclusions display the same hori-
Magma storage and shallower (crustal) zontal trend as the 1815 whole-rocks, with constant Al2O3
processes as SiO2 increases (Fig. 9). All melt inclusions trapped in
Magmatic differentiation plagioclase in the 1815 eruptive products therefore reflect
The 1815 eruption evacuated a 30^33 km3 trachyandesite stage 2 of differentiation of the Tambora magmas.
magma body from a geochemically, isotopically and min- This implies that during the differentiation from basaltic
eralogically relatively uniform reservoir. Data here and trachyandesite to trachyandesite, plagioclase of a wide
those of Foden (1986) suggest that the 1815 eruptive range of compositions (and as calcic as An91) crystallized
products are among the most evolved at Tambora (Figs 3 from these melts. Many of these calcic plagioclases are cor-
and 5). The trends observed in variation diagrams (Fig. 5) roded and occur as An-rich crystal cores that are rimmed
are consistent with the fractionation of plagioclase, by more sodic plagioclase, suggesting that they represent
clinopyroxene, titanomagnetite, small amounts of apatite, earlier formed crystals (‘antecrysts’; see Hildreth, 2001)
biotite and olivine, as well as late-stage ternary feldspar. that were subsequently entrained and partially resorbed

291
JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 FEBRUARY 2012

in magma batches replenishing the main pre-1815 crustal Volatile behaviour and magma degassing
magma reservoir. Self et al. (2004) estimated that Tambora released 53^58
Tg of SO2 in 1815 to generate 93^118 Tg of stratospheric
Magmatic intensive variables sulphuric aerosols, a value that agrees well with estimates
The increase of Al2O3 with slight enrichment in SiO2 of aerosol mass from ice core studies and proxy
during differentiation stage 1 suggests that plagioclase crys- palaeo-temperature records (for discussion, see Self et al.,
tallization was minor during this stage (Fig. 3). This may 2004). Therefore, the Tambora eruption products yield a
indicate that differentiation from trachybasalt to basaltic record of the sulphur mass released, with no ‘excess
trachyandesite possibly occurred at considerable crustal sulphur’, contrasting with other recent eruptions that were
depths near the Moho or, possibly, at sub-crustal depths fed by magmas with similar oxygen fugacities (2 log
and/or elevated water pressures (see Hildreth & units above FMQ). In such magmas, sulphur is sequestered
Moorbath, 1988). Crystallization temperatures during into the gas phase, thus depleting the melt (e.g. Scaillet
these early stages of magma evolution, inferred from et al., 2004) and accounting for discrepancies between pet-

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


liquidus temperature estimates for the most primitive rological estimates and satellite measurements of sulphur
Tambora samples using the PELE model of Boudreau gas release during eruption (e.g. Devine et al., 1984).
(1999), are of the order of 1100^11508C. We propose that Melt inclusions in the 1815 eruptive products define a
basaltic trachyandesite magmas then rose to shallower relatively continuous latite to tephriphonolite spectrum.
crustal levels, where the plagioclase-dominated crystalliza- Concentrations of sulphur in melt inclusions are highest in
tion of differentiation stage 2 led to trachyandesitic^ the least evolved compositions (1255 ppm) and decrease to
tephriphonolitic and ultimately phonolitic melts in a nearly 400 ppm in the most evolved compositions
high-level magma chamber, for which Foden (1986, 1987) (Table 4; Fig. 10). This indicates that the melt in the shal-
inferred a depth between 1·5 and 4·5 km (or possibly low, pre-1815 magma reservoir reached sulphur concentra-
more), based on various phase equilibria. Magma pressure, tions high enough for sulphur to partition into a,
estimated from the maximum measured H2O concentra- presumably, H2O- and CO2-rich fluid phase. This is
tions in melt inclusions (2·5 wt %), would be 60 MPa reflected in the compositions of ternary feldspar microlites,
(estimated using PELE; Boudreau, 1999), which corres- the positions of which relative to the dry solvus appear to
ponds to a depth of 2·3 km (assuming a crustal density indicate higher crystallization temperatures than the
of 2·7 g cm3). Such a value is very likely to be an absolute plagioclase phenocrysts (Fig. 4a). We infer that changes
minimum value, as the inclusions may not have become in H2O activity through degassing raised the liquidus tem-
closed until late during ascent or may have lost some of perature and relocated the position of the ‘two feldspar
their volatile content after entrapment (e.g. Danyushevsky plus liquid’ curve, resulting in ternary feldspar crystalliza-
et al., 2002). Application of the comparatively pressure- tion at higher apparent temperatures. Evidence for the
insensitive plagioclase^melt hygrometer of Lange et al. presence of sulphur-rich magma in the assumed
(2009) to the 1815 products, using the average plagioclase higher-level magma reservoir prior to 1815 exists in the
rim (An58) and glass compositions (Table 4) plus a form of volatile-rich melt inclusions in plagioclase crystals
temperature of 9488C (see below), yields 5·0 wt % that are interpreted to have grown at shallow depth.
H2O, corresponding to a saturation pressure of 200 MPa Sulphur degassing is thought to have accompanied
or a magma chamber depth of 7·5 km (assuming a crust- magma differentiation prior to the cataclysmic eruption
al density of 2·7 g cm3). Both more Ab-rich plagioclase in 1815. We propose that the released gas phase did not
compositions (up to An43; see above) and higher tempera- accumulate within or towards the top of the magma reser-
tures (up to 10058C; see below) yield lower calculated voir to contribute to the volatile budget of the eruption.
melt H2O concentrations and, as such, the 200 MPa Instead, sulphur degassing may have occurred continuous-
pressure (or 7·5 km depth) obtained above may be re- ly over a period of several thousand years prior to 1815
garded as a maximum value for the shallow 1815 magma (see below), with the exsolved gas phase escaping passively
storage region. QUILF geothermometry (Andersen et al., to the surface through permeable wall-rocks. At the time
1993), assuming equilibrium between ferromagnesian sili- of the eruption, the upper magma reservoir had been par-
cate phases (olivine and clinopyroxene) and titanomagne- tially degassed with respect to sulphur (45%; based on
tite in the 1815 eruptive products, indicates crystallization the average sulphur concentrations reported in Table 4)
temperatures of 977^9328C, averaging 948 188C (1s; and the melt inclusions trapped in plagioclase rims pre-
n ¼ 5) and fO2 values of 1·8^2·2 log units above the faya- serve a true record of the pre-eruptive melt volatile (sul-
lite^magnetite^quartz (FMQ) buffer. Temperatures of phur) content.
1005  88C (n ¼116) were obtained using Fe2þ^Ti In contrast to sulphur, degassing of chlorine prior to and
exchange thermometry in biotite (see Table 3) (Luhr et al., during the 1815 eruption was less effective or less complete
1984). (Fig. 10). Chlorine concentrations are only slightly higher

292
GERTISSER et al. PETROGENESIS OF THE 1815 TAMBORA MAGMA

in the less evolved melt inclusions, indicating that only 2.5


30% (based on the average chlorine concentrations
reported in Table 4) degassed prior to eruption. Melt inclu- kyr
sions in plagioclase rims (and compositionally similar 2.0 0

(226Ra/230Th)
cores) have an average of 200 ppm more chlorine than 0.5
the groundmass glass (Table 4). Using this difference and 1.0

the eruption parameters described by Self et al. (2004), the 1.5 1.5
2.0
calculated chlorine release to the atmosphere is of 3.0
4.0
the order of 14^15 Tg or half the amount of sulphur. 6.0
The degassing of chlorine prior to and during the 1815 1.0
Secular equilibrium
Tambora eruption may have implications for the behav-
iour of halogens in K-rich volcanic systems elsewhere.
0.5
Crustal processes and their timescales: constraints from 54 55 56 57

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


U-series isotopes SiO2 (wt%)
U-series data for 1815 Tambora whole-rocks, constituent 226 230
minerals and groundmass glass may be used to provide fur- Fig. 15. ( Ra/ Th) vs SiO2 (wt %) plot for 1815 Tambora
whole-rocks. The schematic arrow illustrates a time-dependent
ther insight into the pre-eruptive timing and nature of (closed-system) differentiation path from the least to the most evolved
magmatic processes. Despite the relative mineralogical, 1815 samples, assuming initial 226Ra^230Th disequilibrium similar to
geochemical and isotopic homogeneity of the 1815 eruptive the least evolved sample (T7). The published Tambora whole-rock
analysis (open circle) is taken from Turner & Foden (2001). Mean
products, the U-series whole-rock, mineral and ground- analytical errors are smaller than the symbol size. (See text for
mass glass data reveal a complex picture (Fig. 11). The sim- discussion.)
plest situation is for sample T39, where the whole-rock,
groundmass glass, plagioclase, clinopyroxene, magnetite
and biotite data fall along a horizontal isochron line
different from the cluster of the other data points may indi-
that is within uncertainty of ‘zero’ age, indicating recent
cate that clinopyroxene has been derived fairly recently
differentiation processes operating over timescales too
from a different part of the magmatic system. However, it
short to be resolved with the U^Th system (Fig. 11). Data
is also possible that the clinopyroxene (238U/232Th) ratios
for the other two samples show more complex patterns
could be affected by inclusions, such as apatite, although
and there is no evidence of any correlation between
this is not obvious petrographically.
(230Th/232Th) and (238U/232Th) that would allow mean-
Given the complexities of the Tambora mineral data in
ingful information on differentiation timescales to be
the U^Th and Ra^Th systems, meaningful information
obtained. This, as shown below, most probably results
on differentiation timescales may best be obtained from
from mixing between crystals from new and old magma
batches and entrainment of cumulate material or ‘ante- the correlation of (226Ra/230Th) with SiO2 as an index of
crysts’. The distinctly lower (230Th/232Th) ratios of the differentiation (Fig. 15). In a time-dependent model,
groundmass glass for samples T57/58 and T7 than the decreasing (226Ra/230Th) reflects the decay of 226Ra
whole-rocks and mineral phases may reflect a younger during (closed-system) differentiation from the least (bas-
melt infiltrating a somewhat older, partially solidified altic trachyandesite) to the most evolved (trachyandesite^
magma. tephriphonolite) samples from the 1815 eruption. As such,
In the 1815 magmas, (226Ra/230Th) values (Fig. 11) do not the time elapsed during differentiation can be constrained
form any coherent pattern. Several mineral phases in the by the half-life of 226Ra to be 4000^4500 years (Fig. 15).
1815 Tambora rocks have (226Ra/230Th) approaching A slightly longer crustal residence time for the 1815
secular equilibrium, indicating ages close to 8000 years Tambora magma was suggested by Turner et al. (2003a)
(Table 2; Fig. 11). Some of the mineral phases are, within based on comparison of (226Ra/230Th) disequilibria with
experimental uncertainty, in (226Ra/230Th) equilibrium data from neighbouring Sangeang Api volcano. It should
and are crystals older than 8000 years, whereas others be noted that the model presented here assumes initial
226
show evidence of 226Ra loss, possibly from recent Ra parti- Ra^230Th disequilibrium similar to that in the least
tioning into growing plagioclase crystals. Clinopyroxene evolved 1815 sample (T7). As this most probably represents
(T57/58), biotite (T7) and plagioclase (T57/58) show large a minimum value, the time elapsed, as inferred from
(226Ra/230Th) disequilibria and must have been Fig. 15, might be underestimated. Similarly, as any primi-
(226Ra/230Th) open systems until very recently, possibly as tive magma is likely to have relatively high (226Ra/230Th)
recently as the eruption age. The observations that clino- disequilibrium, injection of such magma into the differen-
pyroxene (T7) is in (226Ra/230Th) disequilibrium and has tiating magma body may buffer the rate of decrease of
(230Th/232Th) and (238U/232Th) ratios that are significantly (226Ra/230Th) such that even after several thousand years

293
JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 FEBRUARY 2012

significant (226Ra/230Th) disequilibria may be maintained trachybasaltic magma in the inferred deep reservoir.
(Hughes & Hawkesworth, 1999). This would also mean The trachyandesitic magma and residual phonolitic melts
that the inferred timescales represent minimum estimates. evolved by fractional crystallization, possibly driven by
By contrast, crustal assimilation, if it occurred, would decompression-induced volatile exsolution and heat
lead to extended apparent differentiation timescales owing exchange with the evolving magma body, magma mixing
to the incorporation of lower (226Ra/230Th) crustal mater- through additions of basaltic trachyandesite magma and
ial (assuming the assimilant is older than 8000 years), convection, which led to the development of a well-stirred,
shifting the samples closer to equilibrium. homogeneous magma body.
Results from U-series disequilibria indicate that the
above processes occurred over timescales of 4000^4500
S U M M A RY A N D C O N C L U S I O N S years. Most of the mineral phases in the 1815 eruptive prod-
A conceptual model for the magma system of Tambora ucts were not in equilibrium with the melt (groundmass)
prior to the 1815 eruption is illustrated in Fig. 16. In this at time of eruption, but formed during somewhat earlier
model, SiO2-undersaturated, potassic trachybasalt crystallization episodes (58000 years ago). Highly calcic,

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


magmas parental to the 1815 magma originate from small corroded plagioclase crystals in 226Ra^230Th equilibrium
degrees (2%) of partial melting of a garnet-free, (48000 years old) are interpreted to represent ‘antecrystic’
I-MORB-like mantle source fluxed with 3% fluids from material incorporated into the 1815 Tambora magma,
the AOC of the subducting slab and small amounts presumably through rejuvenation of crystal mushes at the
(51%) of subducted sedimentary material. These factors base of the shallow, pre-1815 magma reservoir, or entrain-
contributed to preserving the effects of fluid addition to ment of crystals during ascent of hotter magmas from
the mantle source in the Tambora rocks [(230Th/238U)51]. greater depth. Magma accumulation and differentiation
Magmatic differentiation from primary trachybasalt to at shallow depth within the crust prior to the 1815 eruption
the trachyandesite erupted from Tambora in 1815 occurred was accompanied by degassing of sulphur (and other vola-
during two-stage, polybaric differentiation at depths tile species), presumably through permeable wall-rocks.
around the Moho and in a shallow-level crustal reservoir. Such a process has implications for the assessment of the
The 30^33 km3 homogeneous, trachyandesite magma volatile budget of fairly oxidized arc magmas.
body that was emptied during the 1815 eruption appears
to have grown by influx of basaltic trachyandesite
magmas formed by partial crystallization of primary AC K N O W L E D G E M E N T S
We are grateful to Mabs Gilmour, Nick Rogers and Andy
Tindle for their help with the analytical work carried out
at The Open University (UK), and Geoff Nowell and
Chris Ottley for their assistance with the Hf isotope
analyses at the University of Durham (UK). We thank
Vicky Smith, Tod Waight, Colin Wilson and an anonymous
referee for their thoughtful and thorough reviews that
helped to clarify the ideas presented in the paper, and
John Gamble for editorial handling.

FU NDI NG
We gratefully acknowledge in-kind contributions from The
Open University and the University of Durham for the
analytical work carried out in this study. R.G. acknow-
ledges support by the European Research and Training
Network on ‘Volcano dynamics in relation to monitoring,
hazards mitigation and volcano crisis response’ (Fifth
Framework Programme; Contract number: HPRN-CT-
2000-00060).

R E F E R E NC E S
Andersen, D. J., Lindsley, D. H. & Davidson, P. M. (1993). QUILF: a
Fig. 16. A conceptual model, illustrating the petrogenetic processes Pascal program to assess equilibria among Fe^Mg^Mn^Ti oxides,
and timescales prior to the 1815 Tambora eruption. Mineral pyroxenes, olivine and quartz. Computers and Geosciences 19,
abbreviations as in Fig. 11. (See text for discussion.) 1333^1350.

294
GERTISSER et al. PETROGENESIS OF THE 1815 TAMBORA MAGMA

Barberi, F., Bigioggero, B., Boriani, A., Cattaneo, M., Cavallin, A., 1815 eruption’. Journal of Volcanology and Geothermal Research 31,
Cioni, R., Eva, C., Gelmini, F., Giorgetti, F., Iaccarino, S., 167^170.
Innocenti, F., Marinelli, G., Slejko, D. & Sudradjat, A. (1987). Foden, J. D. & Green, D. H. (1992). Possible role of amphibole in the
The island of Sumbawa: a major structural discontinuity in the origin of andesite: some experimental and natural evidence.
Indonesian arc. Bollettino della Societa' Geologica Italiana 106, 547^620. Contributions to Mineralogy and Petrology 109, 479^493.
Barling, J. & Goldstein, S. L. (1990). Extreme isotopic variations in Foden, J. D. & Varne, R. (1980). The petrology and tectonic setting of
Heard Island lavas and the nature of mantle reservoirs. Nature Quaternary^Recent volcanic centres of Lombok and Sumbawa,
348, 59^62. Sunda arc. Chemical Geology 30, 201^226.
Ben Othman, D., White, W. M. & Patchett, J. (1989). The Gerbe, M.-C., Gourgaud, A., Sigmarsson, O., Harmon, R. S.,
geochemistry of marine sediments, island arc magma genesis and Joron, J.-L. & Provost, A. (1992). Mineralogical and geochemical
crust^mantle recycling. Earth and Planetary Science Letters 94, 1^21. evolution of the 1982^1983 Galunggung eruption (Indonesia).
Blundy, J. D., Robinson, J. A. D. & Wood, B. J. (1998). Heavy REE Bulletin of Volcanology 54, 284^298.
are compatible in clinopyroxene on the spinel lherzolite solidus. Gertisser, R. & Keller, J. (2003). Trace element and Sr, Nd, Pb and
Earth and Planetary Science Letters 160, 493^504. O isotope variations in medium-K and high-K volcanic rocks
Boudreau, A. E. (1999). PELEça version of the MELTS software from Merapi Volcano, Central Java, Indonesia: evidence for the
program for the PC platform. Computers and Geosciences 25, 201^203. involvement of subducted sediments in Sunda Arc magma genesis.

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


Chadwick, J. P., Troll, V. R., Ginibre, C., Morgan, D., Gertisser, R., Journal of Petrology 44, 457^486.
Waight, T. E. & Davidson, J. P. (2007). Carbonate assimilation at Gill, J. B. & Williams, R. W. (1990). Th isotope and U-series studies of
Merapi volcano, Java, Indonesia: insights from crystal isotope subduction related volcanic rocks. Geochimica et Cosmochimica Acta
stratigraphy. Journal of Petrology 48, 1793^1812. 54, 1427^1442.
Chauvel, C. & Blichert-Toft, J. (2001). A hafnium isotope and trace Gottsmann, J. & Dingwell, D. B. (2002). Thermal expansivities of
element perspective on melting of the depleted mantle. Earth and supercooled haplobasaltic liquids. Geochimica et Cosmochimica Acta
Planetary Science Letters 190, 137^151. 66, 2231^2238.
Cioni, R. (2000). Volatile content and degassing processes in the AD79 Grall-Johnson, H. (1997). The geochemical evolution of Tambora
magma chamber at Vesuvius (Italy). Contributions to Mineralogy and volcano, Indonesia. PhD thesis, University of Rhode Island,
Petrology 140, 40^54. Kingston.
Cole-Dai, J., Ferris, D., Lanciki, A., Savarino, J., Baroni, M. & Hamelin, B. & Alle'gre, C. J. (1985). Large-scale regional units in the
Thiemens, M. H. (2009). Cold decade (AD 1810^1819) caused by depleted upper mantle revealed by an isotope study of the
Tambora (1815) and another (1809) stratospheric volcanic eruption. South-West Indian Ridge. Nature 315, 196^199.
Geophysical Research Letters 36, L22703, doi:10.1029/2009GL040882. Hamelin, B., Dupre¤, B. & Alle'gre, C. J. (1985). Pb^Sr^Nd isotopic
Danyushevsky, L. V., McNeill, A. W. & Sobolev, A. V. (2002). data of Indian Ocean Ridges: new evidence of large-scale mapping
Experimental and petrological studies of melt inclusions in pheno- of mantle heterogeneities. Earth and Planetary Science Letters 76,
crysts from mantle-derived magmas: an overview of techniques, 288^298.
advantages and complications. Chemical Geology 183, 5^24. Hamilton, W. (1979). Tectonics of the Indonesian region. US Geological
DePaolo, D. J. (1981). Trace element and isotopic effects of combined Survey Professional Papers 1078, 1^345.
wallrock assimilation and fractional crystallization. Earth and Handley, H. K., Macpherson, C. G. & Davidson, J. P. (2010).
Planetary Science Letters 53, 189^202. Geochemical and Sr^O isotopic constraints on magmatic differ-
Devine, J. D., Sigurdsson, H., Davis, A. N. & Self, S. (1984). Estimates entiation at Gede Volcanic Complex, West Java, Indonesia.
of sulphur and chlorine yield to the atmosphere from volcanic Contributions to Mineralogy and Petrology 159, 885^908.
eruptions and potential climatic effects. Journal of Geophysical Handley, H. K., Davidson, J. P., Macpherson, C. G. & Stimac, J. A.
Research 89, 6309^6325. (2008). Untangling differentiation in arc lavas: constraints from
Dosso, L., Bougault, H., Beuzart, P. & Calvez, J. Y. (1988). The unusual minor and trace element variations at Salak volcano,
geochemical structure of the South-East Indian Ridge. Earth and Indonesia. Chemical Geology 255, 360^378.
Planetary Science Letters 88, 47^59. Handley, H. K., Macpherson, C. G., Davidson, J. P., Berlo, K. &
Edwards, C. M. H., Menzies, M. A. & Thirlwall, M. F. (1991). Lowry, D. (2007). Constraining fluid and sediment contributions
Evidence from Muriah, Indonesia, for the interplay of supra- to subduction-related magmatism in Indonesia: Ijen volcanic com-
subduction zone and intraplate processes in the genesis of potassic plex. Journal of Petrology 48, 1155^1183.
alkaline magmas. Journal of Petrology 32, 555^592. Handley, H. K., Turner, S. P., Macpherson, C. G., Gertisser, R. &
Edwards, C. M. H., Morris, J. D. & Thirlwall, M. F. (1993). Davidson, J. P. (2011). Hf^Nd isotope and trace element constraints
Separating mantle from slab signatures in arc lavas using B/Be on subduction inputs at island arcs: limitations of Hf anomalies as
and radiogenic isotope systematics. Nature 362, 530^533. sediment input indicators. Earth and Planetary Science Letters 304,
Edwards, C. M. H., Menzies, M. A., Thirlwall, M. F., Morris, J. D., 212^223.
Leeman, W. P. & Harmon, R. S. (1994). The transition to potassic Harmon, R. S. & Hoefs, J. (1995). Oxygen isotope heterogeneity of
alkaline volcanism in island arcs: the Ringgit-Beser Complex, the mantle deduced from global 18O systematics of basalts from
East Java, Indonesia. Journal of Petrology 35, 1557^1595. different geotectonic settings. Contributions to Mineralogy and Petrology
Elliott, T., Plank, T., Zindler, A., White, W. & Bourdon, B. (1997). 120, 95^114.
Element transport from slab to volcanic front at the Mariana arc. Hart, S. R., Erlank, A. J. & Kable, E. J. D. (1974). Sea floor basalt
Journal of Geophysical Research 102, 14991^15019. alteration: some chemical and Sr isotopic effects. Contributions to
Foden, J. (1986). The petrology of Tambora volcano, Indonesia; a Mineralogy and Petrology 44, 219^230.
model for the 1815 eruption. Journal of Volcanology and Geothermal Heyckendorf, K. & Jung, D. (1992). Tambora volcano, Sumbawa
Research 27, 1^41. Island, Indonesia. A comparison of ancient and modern products.
Foden, J. (1987). Reply to comments by S. Self and J. A. Wolff on Mitteilungen aus dem Geologisch-Pala«ontologischen Institut der Universita«t
‘The petrology of Tambora volcano, Indonesia: a model for the Hamburg 73, 1^35.

295
JOURNAL OF PETROLOGY VOLUME 53 NUMBER 2 FEBRUARY 2012

Hildreth, W. (2001). A critical overview of silicic magmatism, Penrose provenance of mantle and subduction components in western
Conference on Longevity and Dynamics of Rhyolitic Magma Systems, Pacific arc^basin systems. Journal of Petrology 40, 1579^1611.
Mammoth, CA, 6^12 June 2001. Rampino, M. R. & Self, S. (1982). Tambora 1815, Krakatau 1883 and
Hildreth, W. & Moorbath, S. (1988). Crustal contributions to arc Agung 1963: a study of volcanic eruptions, their stratospheric
magmatism in the Andes of Central Chile. Contributions to aerosols and climatic impact. Quaternary Research 18, 127^143.
Mineralogy and Petrology 98, 455^489. Rehka«mper, M. & Hofmann, A. W. (1997). Recycled ocean crust and
Hoogewerff, J. A., van Bergen, M. J., Vroon, P. Z., Hertogen, J., sediment in Indian Ocean MORB. Earth and Planetary Science
Wordel, R., Sneyers, A., Nasution, A., Varekamp, J. C., Moens, H. Letters 147, 93^106.
L. E. & Mouchel, D. (1997). U-series, Sr^Nd^Pb isotope and Salters, V. J. M. (1996). The generation of mid-ocean ridge basalts
trace-element systematics across an active island arc^continent from the Hf and Nd isotope perspective. Earth and Planetary Science
collision zone: implications for element transfer at the slab^wedge Letters 112, 161^178.
interface. Geochimica et Cosmochimica Acta 61, 1057^1072. Salters, V. J. M. & Hart, S. R. (1991). The mantle sources of ocean
Hughes, R. D. & Hawkesworth, C. J. (1999). The effects of magma ridges, islands and arcs: the Hf-isotope connection. Earth and
replenishment processes on 238U^230Th disequilibrium. Geochimica Planetary Science Letters 104, 364^380.
et Cosmochimica Acta 63, 4101^4110. Salters, V. J. M. & White, W. M. (1998). Hf isotope constraints on
Ito, E., White, W. M. & Go«pel, C. (1987). The O, Sr, Nd and Pb isotope mantle evolution. Chemical Geology 145, 447^460.

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


geochemistry of MORB. Chemical Geology 62, 157^176. Scaillet, B., Luhr, J. F. & Carroll, M. R. (2004). Petrological and
Kempton, P. D., Pearce, J. A., Barry, T. L., Fitton, J. G., volcanological constraints of volcanic sulfur emissions to the
Langmuir, C. & Christie, D. M. (2002). Sr^Nd^Pb^Hf isotope atmosphere. In: Robock, A. & Oppenheimer, C. (eds) Volcanism
results from ODP Leg 187: evidence for mantle dynamics of and the Earth’s Atmosphere. American Geophysical Union, Geophysical
the Australian^Antarctic discordance and origin of the Indian Monograph 139, 11^40.
MORB source. Geochemistry, Geophysics, Geosystems 3, 1074, Self, S. & Wolff, J. A. (1987). Comments on ‘The petrology of Tambora
doi:10.1029/2002GC000320. volcano, Indonesia: a model for the 1815 eruption’ by J. Foden.
Kuritani, T., Yokoyama, T. & Nakamura, E. (2008). Generation of Journal of Volcanology and Geothermal Research 31, 163^166.
rear-arc magmas induced by influx of slab-derived supercritical Self, S., Rampino, M. R., Newton, M. S. & Wolff, J. A. (1984).
liquids: implications from alkali basalt lavas from Rishiri volcano, Volcanological study of the great Tambora eruption of 1815. Geology
Kurile arc. Journal of Petrology 49, 1319^1342. 12, 659^663.
Lange, R. A., Frey, H. M. & Hector, J. (2009). A thermodynamic Self, S., Gertisser, R., Thordarson, T., Rampino, M. R. & Wolff, J. A.
model for the plagioclase^liquid hygrometer/thermometer. (2004). Magma volume, volatile emissions, and stratospheric
American Mineralogist 94, 494^506. aerosols from the 1815 eruption of Tambora. Geophysical Research
Luhr, J. F., Carmichael, I. S. E. & Varekamp, J. C. (1984). The 1982 Letters 31, L20608, doi:10.1029/2004GL020925.
eruptions of El Chicho¤n Volcano, Chiapas, Mexico: Mineralogy Siebert, L., Simkin, T. & Kimberly, P. (2010). Volcanoes of the World, 3rd
and petrology of the anhydrite bearing pumices. Journal of edn. Berkeley and Los Angeles: University of California Press.
Volcanology and Geothermal Research 23, 69^108. Sigurdsson, H. & Carey, S. (1989). Plinian and co-ignimbrite tephra
Mann, M. E., Bradley, R. S. & Hughes, M. K. (1998). Global-scale fall from the 1815 eruption of Tambora volcano. Bulletin of
temperature patterns and climate forcing over the past six Volcanology 51, 243^270.
centuries. Nature 392, 779^787. Sigurdsson, H. & Carey, S. (1992a). The eruption of Tambora in 1815:
McCulloch, M. T. & Gamble, J. A. (1991). Geochemical and geodyna- environmental effects and eruption dynamics. In: Harrington, C.
mical constraints on subduction zone magmatism. Earth and R. (ed.) The Year without a Summer?: World Climate in 1816. Ottawa:
Planetary Science Letters 102, 358^374. Canadian Museum of Nature, pp. 16^45.
McLennan, S. M., Taylor, S. R., McCulloch, M. T. & Maynard, J. B. Sigurdsson, H. & Carey, S. (1992b). Eruptive history of Tambora
(1990). Geochemical and Nd^Sr isotopic composition of deep-sea volcano, Indonesia. Mitteilungen aus dem Geologisch-Pala«ontologischen
turbidites: crustal evolution and plate tectonic associations. Institut der Universita«t Hamburg 70, 187^206.
Geochimica et Cosmochimica Acta 54, 2015^2050. Staudigel, H., Plank, T., White, B. & Schmincke, H.-U. (1996).
Nowell, G. M., Kempton, P. D., Noble, S. R., Fitton, J. G., Geochemical fluxes during seafloor alteration of the basaltic
Saunders, A. D., Mahoney, J. J. & Taylor, R. N. (1998). High preci- upper oceanic crust: DSDP Sites 417 and 418 (Overview). In:
sion Hf isotope measurements of MORB and OIB by thermal ion- Bebout, G. E., Scholl, D. W., Kirby, S. H. & Platt, J. P. (eds)
isation mass spectrometry: insights into the depleted mantle. Subduction: Top to Bottom. American Geophysical Union, Geophysical
Chemical Geology 149, 211^233. Monograph 96, 19^36.
Nowell, G. M., Pearson, D. G., Ottley, C. J. & Schwieters, J. (2003). Staudigel, H., Davies, G. R., Hart, S. R., Marchant, K. M. &
Long-term performance characteristics of a plasma ionisation Smith, B. M. (1995). Large scale isotopic Sr, Nd and O isotopic
multi-collector mass spectrometer (PIMMS): the ThermoFinnigan anatomy of altered oceanic crust: DSDP/ODP sites 417/418. Earth
Neptune. In: Holland, J. G. & Tanner, S. D. (eds) Plasma Source and Planetary Science Letters 130, 169^185.
Mass SpectrometryçApplications and Emerging Technologies. Stolz, A. J., Varne, R., Davies, G. R., Wheller, G. E. & Foden, J. D.
Cambridge: The Royal Society of Chemistry, pp. 307^320. (1990). Magma source components in an arc^continent collision
Patchett, P. J. (1983). Hafnium isotope results from mid-ocean ridges zone: the Flores^Lembata sector, Sunda arc, Indonesia.
and Kerguelen. Lithos 16, 47^51. Contributions to Mineralogy and Petrology 105, 585^601.
Pearce, J. A. (1983). Role of sub-continental lithosphere in magma Stolz, A. J., Varne, R., Wheller, G. E., Foden, J. D. & Abbott, M. J.
genesis at destructive plate margins. In: Hawkesworth, C. J. & (1988). The geochemistry and petrogenesis of K-rich alkaline
Norry, M. J. (eds) Continental Basalts and Mantle Xenoliths. volcanics from the Batu Tara volcano, eastern Sunda arc.
Nantwich: Shiva, pp. 230^249. Contributions to Mineralogy and Petrology 98, 374^389.
Pearce, J. A., Kempton, P. D., Nowell, G. M. & Noble, S. R. (1999). Stothers, R. B. (1984). The great Tambora eruption in 1815 and its
Hf^Nd element and isotope perspective on the nature and aftermath. Science 224, 1191^1198.

296
GERTISSER et al. PETROGENESIS OF THE 1815 TAMBORA MAGMA

Sun, S. S. & McDonough, W. F. (1989). Chemical and isotopic systemat- global sedimentary system. Earth and Planetary Science Letters 168,
ics of oceanic basalts: implications for mantle composition and pro- 79^99.
cesses. In: Saunders, A. D. & Norry, M. J. (eds) Magmatism in the Volpe, A. M., Olivares, J. A. & Murrell, M. T. (1991). Determination
Ocean Basins. Geological Society, London, Special Publications 42, 313^345. of radium isotope ratios and abundances in geologic samples by
Taylor, S. R. & McLennan, S. M. (1985). The Continental Crust; its thermal ionisation mass spectrometry. Analytical Chemistry 63,
Composition and Evolution. Oxford: Blackwell. 913^916.
Thomas, L. E., Hawkesworth, C. J., van Calsteren, P., Turner, S. P. & Wen, S. & Nekvasil, H. (1994). SOLVCALC: an interactive graphics
Rogers, N. W. (1999). Melt generation beneath ocean islands: a U^ program package for calculating the ternary feldspar solvus and
Th^Ra isotope study from Lanzarote in the Canary Islands. for two-feldspar geothermometry. Computers and Geosciences 20,
Geochimica et Cosmochimica Acta 63, 4081^4099. 1025^1040.
Turner, S. & Foden, J. (2001). U, Th and Ra disequilibria, Sr, Nd and Wheller, G. E., Varne, R., Foden, J. D. & Abbott, M. J. (1987).
Pb isotope and trace element variations in Sunda arc lavas: Geochemistry of Quaternary volcanism in the Sunda^Banda
predominance of a subducted sediment component. Contributions to arc, Indonesia, and three-component genesis of island-arc
Mineralogy and Petrology 142, 43^57. basaltic magmas. Journal of Volcanology and Geothermal Research 32,
Turner, S., Bourdon, B. & Gill, J. (2003b). Insights into magma genesis 137^160.
at convergent margins from U-series isotopes. In: Bourdon, B., White, W. M. & Hofmann, A. W. (1982). Sr and Nd isotope

Downloaded from http://petrology.oxfordjournals.org/ at University of East Anglia on January 18, 2012


Henderson, G. M., Lundstrom, C. C. & Turner, S. P. (eds) geochemistry of oceanic basalts and mantle evolution. Nature 296,
U-Series Geochemistry. Mineralogical Society of America and Geochemical 821^825.
Society, Reviews in Mineralogy and Geochemistry 52, 255^315. White, W. M. & Patchett, J. (1984). Hf^Nd^Sr isotopes and incompat-
Turner, S., Foden, J., George, R., Evans, P., Varne, R., Elburg, M. & ible element abundances in island arcs: implications for magma
Jenner, G. (2003a). Rates and processes of potassic magma evolu- origins and crust^mantle evolution. Earth and Planetary Science
tion beneath Sangeang Api volcano, East Sunda arc. Indonesia. Letters 67, 167^185.
Journal of Petrology 44, 491^515. Whitford, D. J. & Nicholls, I. A. (1976). Potassium variation
Turner, S., Hawkesworth, C., Rogers, N., Bartlett, J., Worthington, T., in lavas across the Sunda arc in Java and Bali. In: Johnson, R. W.
Hergt, J., Pearce, J. & Smith, I. (1997). 238U^230Th disequilibria, (ed.) Volcanism in Australasia. Amsterdam: Elsevier, pp. 63^75.
magma petrogenesis, and flux rates beneath the depleted Whitford, D. J., Foden, J. D. & Varne, R. (1978). Sr isotope geochemis-
Tonga^Kermadec island arc. Geochimica et Cosmochimica Acta 61, try of calc-alkaline and alkaline lavas from the Sunda arc in
4855^4884. Lombok and Sumbawa, Indonesia. Carnegie Institution of Washington
van Bemmelen, R. W. (1949). The Geology of Indonesia. Vol. 1A: General Yearbook 77, 613^619.
Geology. The Hague: Government Printing Office. Whitford, D. J., White, W. M. & Jezek, P. A. (1981). Neodymium iso-
van Bergen, M. J., Vroon, P. Z., Varekamp, J. C. & Poorter, R. P. E. topic composition of Quaternary island arc lavas from Indonesia.
(1992). The origin of the potassic rock suite from Batu Tara Volcano Geochimica et Cosmochimica Acta 45, 989^995.
(East Sunda Arc, Indonesia). Lithos 28, 261^282. Woodhead, J., Eggins, S. & Gamble, J. (1993). High field strength
van Calsteren, P. & Schwieters, J. B. (1995). Performance of a thermal and transition element systematics in island arc and back-arc
ionisation mass spectrometer with a deceleration lens system and basin basalts: evidence for multi-phase melt extraction and a
post deceleration detector selection. International Journal of Mass depleted mantle wedge. Earth and Planetary Science Letters 114,
Spectrometry and Ion Processes 146, 119^129. 491^504.
Varne, R. (1985). Ancient subcontinental mantle: a source for K-rich Woodhead, J. D., Hergt, J. M., Davidson, J. P. & Eggins, S. M. (2001).
orogenic volcanics. Geology 13, 405^408. Hafnium isotope evidence for ‘conservative’ element mobility
Varne, R. & Foden, J. D. (1986). Geochemical and isotopic systematics during subduction zone processes. Earth and Planetary Science Letters
of eastern Sunda arc volcanics: implications for mantle sources 192, 331^346.
and mantle mixing processes. In: Wezel, F.-C. (ed.) The Origin of Yang, H. J., Frey, F. A., Weis, D., Giret, A., Pyle, D. G. & Michon, G.
Arcs. Amsterdam: Elsevier, pp. 159^189. (1998). Petrogenesis of the flood basalts forming the northern
Vervoort, J. D., Patchett, P. J., Blichert-Toft, J. & Albare'de, F. (1999). Kerguelen Archipelago: implications for the Kerguelen plume.
Relationships between Lu^Hf and Sm^Nd isotopic systems in the Journal of Petrology 39, 711^748.

297
View publication stats

You might also like