SMITHSONIAN CONTRIBUTIONS TO THE
EARTH SCIENCES
NUMBER 4
William G. Melson,
Eugene Jarosewich,
and Charles A. Lundquist
Volcanic Eruption at
Metis Shoal, Tonga,
1967-1968:
Description and
Petrology
SMITHSONIAN INSTITUTION PRESS
CITY OF WASHINGTON
1970
ABSTRACT
Melson, William G.; Jarosewich, Eugene; and Lundquist, Charles A. Eruption of
Metis Shoal, Tonga, 1967-1968: Description and Petrology. Smithsonian Contri-
butions to the Earth Sciences, 4: 1-18, 1970.?The 1967-1968 eruption of Metis
Shoal, Tonga, was evidentially similar to the frequent shallow submarine eruptions
of the inner island arc of Tonga. The eruption began about 10 December 1967, and
an island eventually emerged; by 19 February 1968, the island had been eroded to
beneath wave base. The eruptions were characterized by explosions of steam and
ash which hurled bombs a few to several hundred feet into the air. The rocks ejected
are pumiceous dacites which, for their silica content, have unusually low alkali contents
and rare earth-element contents. The chemical characteristics of the dacite are hard
to account for by partial melting of an ocean-ridge basalt parent. The peculiar prop-
erties of the dacite appear to characterize other Tongan lavas and support the idea
that Tonga is part of a distinct petrographic province.
Official publication date is handstamped in a limited number of initial copies and is recorded in
the Institution's annual report, Smithsonian Year.
UNITED STATES GOVERNMENT PRINTING OFFICE
WASHINGTON : 1970
For sale by the Superintendent of Documents, U.S. Government Printing Office
Washington, D.C. 20402-Price 30 cents
William G. Melson,
Eugene Jarosewich,
and Charles A. Lundquist
Volcanic Eruption at
Metis Shoal, Tonga,
1967-1968:
Description and
Petrology
Introduction
In late December 1967, we were informed by the
Smithsonian Center for Short-lived Phenomena that
a volcanic eruption was under way in the Tonga
Islands. The early reports indicated that a submarine
eruption had resulted in creation of a new island. In
February 1968, Charles A. Lundquist, on a previously
planned trip to Australia, managed to detour and visit
the site. At that time, the volcanic island had been
eroded away, and only shoals remained; however,
samples were obtained by diving. The petrology of
these samples and the description of the eruption,
obtained from photographs and from other documen-
tation of the eruption obtained by correspondence and
by interviews with eye witnesses, are the subject of
this paper. Background information is summarized
on the probable relationships between volcanism and
seismicity in this island-arc environment.
Geologic Setting of Tonga
The Tonga Islands have long been known as sites
of frequent earthquakes and volcanic eruptions. The
William G. Melson and Eugene Jarosewich, Department of
Mineral Sciences, National Museum of Natural History,
Smithsonian Institution, Washington, D.C. 20560. Charles A.
Lundquist, Smithsonian Astrophysical Observatory, Cam-
bridge, Massachusetts 02138.
volcanic activity is restricted to the western islands;
the eastern islands, including Tongatapu, long in-
active volcanically, are capped by coral reefs and sel-
dom have exposures of volcanic rocks.
Figure 1 shows that the western Tonga Islands lie
on a line that, with but slight bending, can be ex-
tended to include the volcanic Kermadec Islands, and
the volcanic and hydrothermal regions of North
Island, New Zealand.
There are 10 active volcanoes in western Tonga
(Richards, 1962). Of these, only Tofua, Late, and
Fonualei have remained above sea level throughout
historic times. The remaining seven volcanoes are
submarine. Of these seven, only Falcon Island, Metis
Shoal, and Home Reef have reached above sea level
during eruptions, and, at present, even these three
have been eroded to below sea level. The most active
shoal volcano appears to be the one which periodically
builds Falcon Island. In all, nine separate eruptions
have been reported. Perhaps because of their transi-
tory existence, very little is known about the petrology
of these "jack-in-the-box" islands.
One of the few generalizations about the petrology
of the Tonga Islands was made many years ago by
Alfred Harker (1891), after examination of a suite
of Tongan rocks:
With the exception of the Falcon Island rocks, all
those [rocks] examined from the Tonga Islands ap-
1
SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
15?
20
0
25
30
35
4O?
175?
s
\ V
<
180
Vanua Levu^^?y
Vita Levu o
o
L J
O U T H
Of
A U
V r \ ?'White
? 175?
JlVall/s
Nome
? Islands
? FIJI
.a
LANDS
i
Metis She
KeTofua
Fa</'con.
Nukualofa^
iAta(P
to
z:
o
o.
o
>?
/ MERMAD?C
? ISLANDS
' P A C
C E JL N
Savaii
ISLANDS o
iuafou
,'Fanua/a/ (Amargura)
Late,
a/-:' o
0 ' TONGA
ISLANDS'onga Tonga
y7bngatabu
ylstaart)
i
/ F I c
0 50 100 200 300 400 500
STATUTE MILES
y ZEALAND
I p f NORTH ISLAND
r J ^MtKRuapehu
???~~ T\ /
EAST 175? LONGITUDE
- 15
-20
? 25
~30?
? 35?
80? WEST 175? LONGITUDE 170?
FIGURE 1.?Location of Tonga and Metis Shoal.
NUMBER 4
pear to be of submarine formation. The absence or
presence in different strata of any sensible proportion
of calcareous matter and organic remains is perhaps
related to the more or less rapid rate of accumulation
at different epochs of eruption. The volcanic material
ejected seems to have been almost exclusively of frag-
mental character, and in some cases there are indica-
tions of violent explosive action. This is quite in ac-
cord with the andesitic nature of the materials thrown
out, which are the types common in the Pacific region.
Harker evidently did not examine samples of lavas,
which do make up parts of Tofua and other active
subaerial Tongan volcanoes.
Bryan (1969) drew attention to certain chemical
peculiarities of Tongan volcanic rocks?unusually low
alkalies and high iron with high silica contents. He
notes that this tendency applies not only to the da-
cites, but to the basalts from Late and Tofua as well.
This peculiarity is true of our Metis Shoal's samples,
and bears out the distinctive chemistry of Tongan
lavas.
RELATION BETWEEN VOLCANISM AND SEISMOLOGY.?
Tonga is along the junction of two of the six major
tectonic blocks of the earth's crust (Figure 2). The
contact between these blocks is normally an oceanic
trench, oceanic ridge, or a fracture zone (transform
fault zone, Figure 3). At oceanic trenches, the thick
lithosphere, estimated to be around 100 kilometers
thick, is believed to move downward. Along its upper
surface, numerous earthquakes occur, defining a nar-
row belt. Rocks of the lithosphere, as they descend,
are envisioned as becoming hot enough to undergo
partial melting. Where tensional fractures occur along
the lithosphere, melts so generated move upward, giv-
ing rise to the volcanism that characterizes most island
arcs (Figure 4).
The Tonga-Kermadec region is one of the world's
most active seismic belts (Isacks et al., 1969). Figure
^120 180 120 60
FIGURE 2.?Location of Tonga in regard to the large tectonic blocks of the new global tectonics.
Heavy lines are island arcs or arc-like features. Tensional (divergent arrows) and compressional
(convergent arrows) indicate relative movements at margins of blocks; length of arrows is roughly
proportional to rate of relative movements. Some historically active volcanoes are indicated by X.
Open circles represent earthquakes that generated tsunamis (seismic sea waves) detected at
distances of 1,000 or more kilometers from their source. The six major tectonic blocks are shown
(America, Africa, Eurasia, Antarctica, Pacific, and Australian). Modified from Isacks, Oliver, and
Sykes (1968).
SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
FIGURE 3.?Idealized block section showing probable nature of contacts between the tectonic
blocks (Figure 2). Contacts are trenches, ridges, and fracture zones (transform faults). Modified
from Isacks, Oliver, and Sykes (1968). Arrow indicates relative movements. The lithosphere
moves under at oceanic trenches and is generated by igneous activity at mid-ocean ridges in the
now widely accepted models of the new global tectonics. The contact between the lithosphere
and asthenosphere is not the contact between the crust and mantle, but is at the much greater
depth of about one-hundred kilometers.
300 100 0
Local Tension
LITHOSPHERE
Local tension
t- 4 4 4 4
ASTHENOSPHERE 1
loo
CRUST
LITHOSPHERE
4
4-
50-
4-4-4-4 4
ASTHENOSPHERE
4-4-4 4 '5^'
+ 4-4 4*?'
FIGURE 4.?Hypothetical section through an island arc. Down-going slab of lithosphere, seismic
zone (S) near surface of slab and in adjacent crust, tensional features beneath ocean deep where
slab bends abruptly and surface slab is still open are shown. Island-arc magmas move upward
along tension fractures developed near the base of the astheosphere. Modified from Isacks, Oliver,
and Sykes (1968).
5 shows epicenter locations through a portion of Tonga
during a single year. The large number of events and
their clear-cut restriction to a relatively narrow zone
(the BeniofF Zone) is characteristic of other island
arcs. It is along this zone that the upper surface of the
lithosphere is believed to be descending (Figure 4).
It is also thought that the lavas of island arcs originate
along this zone. A correlation between lava chemical
composition and the depth below the eruption site to
the Benioff Zone has been found. Specifically, the
potassium content has been found to increase with in-
creasing depth to this zone (Hatherton and Dickin-
son, 1969). At present, there is insufficient data on
Tongan lavas to adequately test for such a relation.
The presence of such a relation does indicate that the
depth of magma generation is either at the Benioff
Zone or directly related to the depth of this zone.
Course of the Eruption
On 11 December 1967 (Greenwich Time), volcanic
activity in the Metis Shoal region of the Tonga Islands
was noted by the crews of several ships and by passen-
gers on a Polynesian Airlines flight from Samoa to
NUMBER 4
Volcanic Coralline
Arc Arc
0
10 Vtri
O 200
DISTANCE FROM NIUMATE , TONGA
0.
u
a
200
400
600
-
-
-
400
A
,?i?
1
J
1
200
1
c
o
o
h
1
1
o
b
1
0
1 ? 1
A A S^8 ^?
o
1 1
200 Kn
MO ,
-
-
-
?-
-
FIGURE 5.?Location of earthquake epicenters beneath Tonga during 1965. Vertical section
through Niumate, Tonga. Triangles represent events projected from within 0-150 km south of the
section; circles represent events projected from within 0-150 km north of the section. A vertical
exaggeration of about 31:1 is used for the topographic section. Modified from Isacks, Oliver, and
Sykes (1968).
Tonga. Volcanic activity was last reported from this
region in 1876 (Phillips, 1899), and accounts exist for
still earlier events (Richards, 1962).
The United States Navy Hydrographic Office map x
of the Tonga Islands indicates three shallow areas in
the region, of which the most southerly is designated
Metis Shoal. Each area is shown with a diameter of
xMap number 2016, Tonga Islands. 3d edition, 1947.
Hydrographic Office, Washington, D.C.
about 0.5'. The approximate coordinates of these
areas are:
Metis Shoal 19?10.8'S 174?51.8'W
northwestern shoal 19?9.8'S 174?52.6'W
northeastern shoal 19?9.5'S 174?51.1'W
The United States destroyer Fox, the first ship to
sight the 1967 eruption, reported the site to be 0.7
nautical miles north of Metis Shoal, which puts the
eruption approximately at the center of the three areas
SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
in the Hydrographic Office chart. After its visit to the
site on 13 February 1968, the Soviet research vessel
Vityaz reported coordinates (Bezrukov et al., 1968) :
19?09'06"S 174?51.1'W.2
These are nearest to the coordinates of the north-
eastern shoal. The Alaimoana, en route from
Nuku'alofa, on 19 February 1968, passed near one
shoal south of the eruptive site where specimens were
collected. Hence, it appears that recent volcanic
activity was north of a more southerly shoal. No other
reports reaching us mention positions of the recent
eruption relative to other shoals, nor do we know how
many shoals might have been present at this time. On
the basis of sea discoloration, there may be some pos-
sibility that the southerly shoal, observed at a distance
from the Alaimoana, also had volcanic activity.
One observation on 11 December was from the
Tofua. Captain Peter Bennett reported:
At the closest approach of 9 miles the volcanic erup-
tion had the appearance of an incandescent island
about l/% mile long and 150 feet high, glowing cherry
red at constant intensity. Above this island a dense
pillar of steam and smoke ascended to 3000 feet and
at approximately 1 minute intervals molten lava or
boulders in parabolic arcs were observed being ejected
to a height of 1000 feet.3
His Majesty, King Taufa'ahau Tupou IV of Tonga
and his party obtained early photographic records
of the events during a charter flight to the volcano
on 14 December. At this time, rocks projected an esti-
mated 50 feet above the sea as shown in Figure 6, and
the island was some 100 to 200 yards across.4
On 14 December, two Orion Aircraft of No. 5
Squadron, Royal New Zealand Air Force, flew around
the volcano for approximately 30 minutes. Flight
Lieutenant P. K. Simpson reported:
2 P. L. Bezrukov, J. O. Murdmaa, N. G. Prokoptsev, M. A.
Repechka, Preliminary results of geological investigations at
Metis Shoal Area, Tonga Islands. 1968. Enclosure with letter
of 6 May 1968 from P. L. Bezrukov to C. A. Lundquist.
3 Letter from Captain P. Bennett, Tofua, Union Steam Ship
Company of New Zealand Limited, to C. A. Lundquist,
9 April 1968.
* Volcanic eruptions produces new "Jack-in-the-Box" island.
Article in The Chronicle, Nuku'alofa, Tonga Islands (J.
Reichelmann, editor), 15 December 1967, page 1. Also,
private communication and 8-mm motion-picture film from
J. J. Furniss, aviation adviser, Office of H. B. M. Commis-
sioner and Consul, Tonga, to R. Citron in 1968.
At the site of the eruption an island had formed and
it was variously estimated to be 20 to 40 feet high
and as much as 150 to 300 yards long. At the eastern
(upwind) extremity of the island was a crater guessed
to be 10 to 20 yards across. Its shape was circular with
dark grey lips. Inside, the grey soon changed to a dark
cherry red becoming brighter and more orange with
increasing depth. These details were observed only
momentarily when for a few seconds the smoke cleared
about the crater. . . .
Lumps of dark colored matter were being flung
several hundred feet into the air. They seemed to be
ejected at intervals of 10 to 20 seconds and they
splashed into the sea in all directions out to a distance
of 250 yards. This matter was clearly visible from
l/i to 1 mile away. Occasionally a much larger piece
(perhaps 2 or 3 feet across) would be seen to rise
to a height of 100 to 200 feet.
The sea water around the volcano was colored an
emerald green streaked with brown. This discolora-
tion extended several hundred yards upwind and 7
nautical miles on a bearing of 240? T from the island.5
Five days later, on 19 December, Captain C. H. Hill-
Willis of the motor vessel Aonui observed the island
from a distance of 4 miles. He noted: "the island was
one third of a mile long and approximately 80 feet
high. It was erupting from a single cone at intervals
of roughly 3 minutes, with steam vapour rising to some
8,000 feet." G
Particularly informative observations were provided
by Captain Thomas M. Holmes of Polynesian Airlines,
who flew over the site repeatedly. Figures 7 and 8
were taken by him on 31 December. He relates that:
The island, generally kidney-shaped at its best, lay
with its longitudinal axis approximately E?W. Steam
clouds on that day, 31 December, were blowing in a
direction of about 200 degrees true, the surface wind
was from the N.N.E.7
Concerning volcanic activity, Captain Holmes
5 Royal New Zealand Air Force report on flight to Metis
Shoal, 14 December 1967. Enclosure with letter from D.
Rishworth, New Zealand Department of Scientific and In-
dustrial Research, Geological Survey Office, to R. Citron,
8 May 1968.
0 Letter from Captain C. H. Hill-Willis of the motor vessel
Aonui Nuku'alofa, Tonga Islands, to C. A. Lundquist, 10
April 1968.
7 Letter from Captain T. M. Holmes, Polynesian Airlines
Limited, to R. Citron, 10 April 1968.
NUMBER 4
FIGURE 6.?Metis Shoal eruption. Steam and ash clouds with projecting blocks at base of cloud.
Rocks projected an estimated fifty feet above sea surface at this time. (Photograph courtesy of
Mr. J. Reichelmann.)
SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
FIGURE 7.?Metis Shoal, 31 December 1967. Island has its long axis oriented about east-west.
Steam clouds are blowing to 200? true; surface winds are blowing from approximately N.N.E.
(Photograph courtesy of Mr. T. M. Holmes.)
(1968) reported: "On Sunday December 24th, violent
eruptions were taking place. By Sunday, Decem-
ber 31st, these eruptions appeared to have reduced
considerably, although still actively steaming [Figures
7 and 8]. By Sunday, January 7th, 1968, volcanic ac-
tivity appeared to have ceased although steam was still
evident. By Sunday, 31 January 68, all activity had
ceased, volcanic and steaming." 6
Captain Holmes estimates that the island reached
a maximum size by about 7 January, and that the
island disappeared in the first or second week of
February.
On 13 February, the Soviet research vessel Vityaz
reached the site 2 (Bezrukov et al. 1968). They found
no features of continuing volcanic activity. There were
two or three low rocks above the sea level and a shoal
beneath them. They approached to within 200-300
meters of the rocks, and used a grab to obtain bottom
samples at depths from 37 to 190 meters.
The site was visited on 19 February by Charles A.
Lundquist in a small motorboat, the Alaimoana,
chartered from Nuku'alofa. Only an occasional point
of rock could be glimpsed in the towering breakers
that were present even though the sea was quite calm
(Figure 9). A linear shoal extended roughly east-west
from the breakers. The shoal increased sharply in
depth in the eastward direction and became invisible
within a few hundred meters of the breakers; westward,
the shoal increased in depth more gradually.
A dugout outrigger canoe, manned by two residents
of Matuku Island, was launched from the Alaimoana,
and samples were obtained by diving in water about
1.5 meters deep.
An immediate examination on shipboard of the
specimens revealed delicate fibers of glass exposed on
the surface, indicating only slight erosion; also, the
rocks were easily broken manually. This evidence would
seem to indicate that the rocks in their collected form
could not have been exposed very long to the environ-
ment of the shoal and must have been fractured from
larger objects shortly before their collection.
A final report by Captain Holmes describes the site
on 21 April 1968:
NUMBER 4
FIGURE 8.?Metis Shoal, 31 December 1967. Details are the same as for Figure 7, but viewed
from much closer. {Photograph courtesy of Mr. T. M. Holmes.)
The sea was calm and the island completely covered
vyithout any appearance of breaking waves. The color
looked more like yellow sand as opposed to black sand,
mentioned earlier. The water around about had a
definite yellow tinge, suggesting a sulphur content. The
coloring effect gave an impression of a discharge in
the sea. The size of the island (beneath the sea) was
roughly 100 yards across and circular in form.
We obtained no information on the volume of ejecta;
however, Murdmaa and others (1969) carried out a
dredging survey around the volcano and found very
little new debris on the sea floor. In many places, even
near the summit, there was no volcanic debris. Instead,
the area consisted of old ejecta encrusted with various
marine organisms. From this, Murdmaa and his col-
leagues (1969) concluded that the total volume of
ejecta was small.
TABLE 1.?Chronology of events, Metis Shoal, Tonga
Islands
1. Submarine eruption begins?unknown
2. Island emerges, subaerial eruptions?probably not earlier
than 10 December 1967
3. First observation and report of supermarine eruption?11
December 1967
4. Activity reached peak sometime in 31 December 1967 and
apparently ended by 7-8 January 1968
5. Maximum length, 700 meters; width, 100 meters; height,
15 meters (?50%)?reached on or about 7 January 1968
6. Erosion of island down to "rocky outcrops"?25 January
1968
7. Island only "jagged rocks and water washing across"?
1 February 1968
8. Island submerged, very high breakers on subsurface rock:
19 February 1968
9. Shoal completely beneath water with no breakers?1 April
1968
?Petrology of the Ejecta
DESCRIPTION.?The blocks, collected by diving, are
pumiceous dacites with about 43 percent glass, 37 per-
cent pore space, 9 percent plagioclase phenocrysts, 11
percent pyroxene phenocrysts (augite>hypersthene),
and 0.3 percent magnetite. In order of abundance, the
phenocrysts are bytownite, augite, hypersthene, and
10 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
FIGURE 9.?Only shoals remained at the eruption site on 19 February 1968. (Photographed by
Charles A. Lundquist.)
titanmagnetite. The phenocrysts commonly occur in
clusters (Figure 10a) and bytownite phenocrysts are up
to 5 mm long. Critical features of the glass and minerals
follow:
Glass: The glass is colorless in thin section, has a re-
fractive index of 1.502, and has the following approxi-
mate composition measured on the electron micro-
probe: SiO2=72.0?4.0 (ranges slightly from spot to
spot); Al2O3=13.7?1.0; TiO2 = 0.3?0.1; Fe=3.4?
0.2; MgO=l.l?0.1; CaO=3.4?0.2; Na2O=2.7?
0.2; and K2O = 2.1 ?0.2. The glass has a silica content
comparable to that of rhyolites, although the Na2O and
K2O contents of the glass are lower than the contents
of these elements in typical rhyolites. A separate of
glass was prepared and analyzed by traditional wet
chemical analysis (Table 2). This analysis confirmed
the compositional pecularities indicated by the micro-
probe analysis. The glass contains abundant flattened
vesicles (Figure 11) and but few microlites.
Plagioclase: The plagioclase phenocrysts are re-
markably calcic (An83-An87). One might expect oligo-
clase, and perhaps even sanidine, in a volcanic rock so
rich in silica. The pronounced oscillatory zoning in the
plagioclase reflects a remarkably small variation in com-
position. On the basis of microprobe traverses, the
maximum range on a single phenocryst is probably no
greater than 4-mole percent anorthite molecule. The
larger plagoclase phenocrysts commonly consist of com-
plex multiple intergrown crystals. Some of the smaller
phenocrysts are untwinned, or have only a single twin
plane. These might be mistaken for phenocrysts of
alkali feldspar; their high index of refraction, and com-
position given by microprobe analyses show decisively
that they are not alkali feldspar.
Augite: the augite phenocrysts are light green and
weakly pleochroic and have a composition around
En42Fsi7Wo41. Zoning is readily visible in thin section,
but compositional variations across phenocrysts are
NUMBER 4 11
FIGURE 10.?Phenocrysts of augite, hypersthene, and calcic bytownite in individual crystals and
groupings (conglomerporphyritic clots).
small, amounting to less than 0.2% Fe variation. The
augite contains around 1.3 ?0.2% A12O3, and only
traces of Na2O and TiO2.
Hypersthene: Hypersthene occurs as pleochroic
light-green to light-pink phenocrysts, and has a com-
position around En63Fs3!tWo3.o. The hypersthene is only
chromian spinel occur in the olivine phenocryst (Fig-
and only traces of other elements.
Titan-magnetite: There are octahedral opaque
phenocrysts of a magnetic spinel-group mineral con-
taining about 4.4% Ti and 2.6% Mg (only these two
elements were determined).
Olivine: One partly resorbed olivine crystal rimmed
by hypersthene granules was noted in one of four thin
sections (Figures 12 and 13). One would expect it to
be iron-rich in such a high silica rock. Surprisingly,
it is Fo93, and thus even more forsteritic than is typical
of most basaltic olivines.
Chromite(?) : Octahedra of a brown translucent
chromian spinel occur in the olivine phenocryst (Fig-
ure 13).
CLASSIFICATION.?Tongan volcanic rocks seem to fit
roughly into the basalt-andesite-rhyolite association of
Turner and Verhoogen (1960), an association com-
mon in areas of crustal deformation and most typically
developed in connection with orogenic movements
along the margins of the Pacific. Many data are avail-
able on this association because the basalt-rhyolite
series has been erupted in numerous places along the
chains of active and recently active volcanoes that form
the island arcs of the Pacific border.
This association shows the average plagioclase com-
position to be increasingly sodic along the general series
of basalt?andesite?quartz latite?rhyolite, but the
composition of plagioclase phenocrysts shows little re-
lation to the bulk chemical composition of the rocks in
which they occur; the range for rhyolitic quartz
latites being about An12 to An55.
The pumiceous Metis Shoal's rocks are difficult to
classify precisely because of their chemical and mineral-
ogic peculiarities. They are different from most dacites
in their low total alkali content, and in their mineral-
ogy. Typical dacites contain around five percent total
alkalies, and K2O typically is more abundant than
12 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
f f m ml*k
FIGURE 11.?Details of the glassy matrix. The glass was analyzed by electron probe (see text) and
by classical methods (Table II).
Na2O. For example, the type dacite (Troger, 1935, p.
75) has Na2O = 3.02 percent, and K2O = 2.51 percent,
although its silica content (66.76 percent) is quite like
that of the Metis Shoal's rocks (64.01 percent). Miner-
alogic peculiarities, which in part reflect the composi-
tional peculiarities, include the presence of forsteritic
olivine and associated chromite, and the abundance of
bytownite phenocrysts. Dacites are typically porphy-
ritic, but more sodic plagioclase phenocrysts, as well as
phenocrysts of sanidine and quartz, are common
phenocryst minerals.
RELATION TO TEKTITES.?The Metis Shoal's glass
is compositionally similar to certain tektites. Tektite
chemistry typically differs from that of most volcanic
glasses in a high silica content and low total alkali
content, a feature of the Metis Shoal's glass. Some
caution should thus be used in judging that "micro-
tektites," small glass spheres found in certain deep
sea cores, are extraterrestrial based on chemistry alone.
The Metis Shoal's glass differs, however, from tek-
tites compositionally in that Na2O is higher than K2O,
and petrographically in its pumiceous texture (Figure
11), compositional homogeneity (only slight composi-
tional differences?mainly in SiO2 were noted), and in
presence of scattered microlites of hypersthene, plagio-
clase, and possibly other minerals. Also, the porphyritic
texture of the bulk sample (glass + phenocrysts)
clearly is not a feature of tektites.
PETROGENESIS.?The Metis Shoal rocks present
some difficult problems of origin. Crustal crystal-liquid
fractionation of a basaltic liquid can generate higher
silica liquids, but, a.long with silica, the alkalies are
highly enriched in residual liquids. For example, resid-
ual glasses in a tholeiitic basalt from the prehistoric
Makaopuhi lava lake, Hawaii, have SiO2=75%-78%
and total alkali contents typically about 8.6%, with
K2O = 0.41 in the uppermost analyzed sample, and
an SiO2 content of about 50 percent in the parent
basalt (Evans and Moore, 1968). It is thus clear that
a single-stage low pressure differentiation of basalt
could not have produced the Metis Shoal's rocks.
The presence of forsteritic olivine and associated
chromite octahedra is anomalous in a rock with such
high silica. Clearly the olivine was not in equilibrium
NUMBER 4 13
TABLE 2.?Chemical composition of whole rock {column 1)
and separated glass {column 2) from Metis Shoal's 1968
eruption. These samples have been entered into the petrology
collections of the United States National Museum
{USNM 111108).
TABLE 3.?Rocks from Tonga chemically similar to Metis
Shoal's bulk docite and glass {Table 7).
SiO2
A12O3
Fe2O3
FeO
MnO
MgO
CaO
Na2O
K2O
H2O+
H2O-
TiO2
P2O,
CIPW Norm:
Q
OR
AB
AN
DI
HY
MT
IL
AP
1
64.01
12.95
1.51
5.43
0. 13
5.47
7.51
1.83
0.95
0.00
0.00
0.32
0.06
100. 17
24.52
5.61
15.49
24. 32
10.23
17.07
2. 19
0.61
0. 14
100. 18
2
73.38
12.66
1. 11
2.84
0.06
0.84
3.79
3.09
1.52
<0. 13
0.03
0.45
0.00
99.77
39. 11
8.98
26. 15
16. 19
2.21
4.64
1.61
0.85
0.03
99.77
Analyst: Eugene Jarosewich, Smithsonian Institution
with the liquid during the final phase of phenocryst
crystallization. This is shown by the jacketing of the
olivine by hypersthene granules (Figure 12). The
incongruent melting (and hence crystallization) of
hypersthene might conceivably lead to first crystal-
lization of olivine in a high silica magma, but prob-
ably not in a magma as high in silica as that of Metis
Shoal, nor, even if olivine did crystallize, would one
expect it to be so forsteritic and associated with chro-
mite. It is much more likely that the olivine and
chromite are xenocrysts, or reflect mixing and hybrid-
ization of already partially crystallized basaltic magma
with some other more siliceous magma.
At the present time, we can think of no plausible
scheme of partial melting or other processes which
SiO2
AI2O3
Fe2O3
FeO
MnO
MgO
CaO
Na2O
K2O
H2O +
H2O-
TiO2
P2O5
CIPW Norms
Q
Or
Ab
An
Di
Hy
Mt
IL
AP
H2O
1
73.50
11.90
0.76
4.84
0.63
3.80
2.81
0.75
0.46
0.58
0.09
100. 12
42.25
4.43
23. 78
17.64
0.58
8.57
1. 10
1. 10
0.20
0.46
100. 11
2
67.20
12. 13
0.52
8.47
0. 13
1.30
5.60
2.62
0.63
0.34
0.62
0. 16
99.72
31.07
3. 72
22. 17
19.48
6.38
14.28
0.75
1. 18
0.35
0. 34
99. 72
3
64.82
11.95
1.53
8. 16
0. 15
1. 19
6.54
2.64
0.85
1. 16
0.20
0.78
Tr.
99.97
27.70
5.02
22.34
18. 25
12. 28
9.32
2.22
1.48
1. 36
99.97
4
65.84
12.02
1.05
8.25
0. 13
1.29
6.22
2.66
0.77
0.84
0. 16
0.76
Tr.
99.99
28. 78
4.55
22.51
18.58
10.61
11.00
1.52
1.44
1.00
99.99
5
65.98
11.43
2.38
5.86
1.80
1. 10
5.20
2.77
1.09
0.56
0. 10
0.91
0.23
99.41
30.61
6.44
23.44
15.54
7.56
9.48
3.45
1.73
0.53
0.66
99.44
6
65.23
12.59
1.24
6.80
1.68
1.42
5.80
2.88
1.00
0.24
0. 14
0.88
0.21
100. 11
26.64
5.91
24.37
18.47
7.70
12. 70
1.80
1.67
0.49
0.34
100. 09
1. Herald Cays, light pumice, L. J. Sutherland, analyst
(Bryan, 1968).
2. Herald Cays, dark pumice, L. J. Sutherland, analyst (Bryan
1968).
3 and 4. Falcon Island, Tonga, Pumice from 1928 eruption
(Lacroix, 1939).
5 and 6. Fonualei, Tonga, dacite, (Richard, 1962).
could account for all the compositional and mineralogic
peculiarities of the Metis Shoal lava.
Dr. Norman Hubbard of the Lamont-Doharty Geo-
logical Observatory is carrying out an investigation of
the contents of the lanthanide elements in various
Tongan lavas. He performed an analysis for these
elements in the pumiceous dacite described in this
paper (Table 2, whole rock analysis). He kindly
allowed us to give the results of that analysis in this
paper. He obtained, via neutron activation analysis,
the following values (given in parts per million) :
lanthanum, 2.70; cerium, 7.31; niobium, 5.17; samar-
14 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
FIGURE 12.?Large, rounded olivine crystals (xenocryst) rimmed by hypersthene. Olivine
composition around F093.
ium, 1.56; gadolinium, 1.98; europium, 0.452; dys-
prosium, 2.51; erbium, 1.68; and ytterbium, 2.19.
Dr. Hubbard also reports a barium content of 334
parts per million.
Normally, higher silica lavas, like dacites and ande-
sites, are higher in the lanthanide elements than are
the lower silica lavas, like basalts. Remarkably, how-
ever, the Metis Shoal dacite is lower in the lanthanide
elements than most basalts, particularly those from
the ocean-ridge system. Although lower in lanthanides
than the ocean-ridge basalts, the Metis Shoal sample
has the same "primitive" lanthanide abundance ratios
compared to chondritic meteorites as have the ocean-
ridge basalts. Thus, any scheme of partial fusion or
fractional crystallization of a basaltic parent must pro-
vide for depletion of the lanthanide elements but at
the same time give high silica.
Comparisons of the Metis Shoal lava with mid-
ocean ridge basalts are appropriate. Most models of
sea-floor spreading and island arc development en-
vision the oceanic crust as moving beneath island arcs
(Figure 4). Island arcs represent the place where
oceanic crust is destroyed either by addition to conti-
nental margins or by movement downward into the
mantle. In those models where ocean-ridge basalts
are envisioned as moving downward into the mantle,
partial melting might occur as the basaltic rocks de-
scend deeper into the mantle, where they eventually
transform into eclogites. The partial melting of
oceanic-ridge basalts is thus a plausible and often con-
sidered mechanism for generating island-arc lavas,
such as those of Tonga. If partial melting of ocean-
ridge basalts is giving rise to the Tongan lavas, how-
ever, the process must involve separation of crystals
which are higher in the lanthanide elements and lower
in silica than are those of the ocean-ridge basalts. If
this were not the case, the Metis Shoal lava would not
be depleted in rare earth elements and would be
higher in silica than ocean-ridge basalts. Dr. Hubbard
(personal communication) has stated that separation
of a sizable amount of hornblende from a basaltic pa-
rent might account for the peculiarities of the Metis
Shoal sample. This process would require that the
ocean-ridge basalts undergo partial fusion at high
NUMBER 4 15
FIGURE 13.?Detail of margin of olivine xenocryst showing chromite inclusions
and hypersthene rims.
water pressures. Liquids generated under such high
water pressures, which may be on the order of a few
kilobars, would be expected to be high in dissolved
water. The absence of hydrous phases in the Metis
Shoal dacite indicates that lava was not high in water.
This may mean that the above outlined explanation
for the silica and lanthanide contents is not correct,
or that the lava lost water by much degassing prior to
the crystallization of its phenocrysts minerals, or that
ocean-ridge basalts are not the parent material of the
Metis Shoal lava.
Comparison With Eruptions at Falcon Island
The nearest historic volcanism occurred on Falcon
Island (now only shoals remain), which is about 60
miles south of Metis Shoal. An account of the eruption
of 1927 is given by Hoffmeister, Ladd, and Ailing
(1929). Beginning on 4 October 1927, and ending
sometime in the spring of 1928, the eruption built an
island about 2 miles in diameter, and more than
1.2 X1010 ft3 of fragmental volcanic debris were
ejected. No lava was erupted; the island was composed
entirely of pyroclastic material ejected as agglomerate,
a mixture of brownish-gray ash, scoria, pumice, and
blocks of solid lava. Most of the material, except on
the beaches, was poorly sorted.
No bulk nor mineral chemical analyses are available
of the erupted materials; however, Ailing (Hoffmei-
ster, et al. 1929) gives a brief petrographic descrip-
tion of the rocks:
A study of several thin sections of samples collected
on Falcon shows that the rock is a basic one of ba-
saltic affinities, its textural relations suggesting a dia-
basic pyroclastic. It is composed in large part of a
nearly opaque brown glass, the small amount of crys-
talline matter present being very fine-grained and pro-
foundly stained with oxides of iron.
Rods or lath-shaped crystals of plagioclase feldspar
are strikingly embedded in the glass. Many present a
zonal structure and when not actually concentrically
grown show an undulatory extinction. Extinction
angles point to a labradorite of bytownite composition
16 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
but it seems highly probable that the composition is not
uniform but varies. Absolutely untwinned crystals of
plagioclase are surprisingly common. These rods are
arranged in curiously grouped bunches in some por-
tions of a slide, controlled by blow holes, and in others
exhibit a crude parallelism showing distinctly a flow
structure. These portions may be interpreted as in-
clusions of volcanic matter which represent a previ-
ously formed rock, possibly a flow broken up by the
recent activity which is responsible for the reappear-
ance of the island.
Stout grains of pale green pyroxene, not far from
diopside, can be identified by exploring with the high
power. Olivine is probably present but its identification
is uncertain. Magnetite, apatite, ilmenite, are more
definitely determined. Much of the magnetite is tita-
nium bearing.
Ailing believed the rocks to be basaltic, but on the
basis of his petrographic description, there is little
justification for this view. The index of refraction of
the glass for Falcon Island was not given. This would
have been a good guide to the SiO2 content of the glass,
which in turn would allow determination of the actual
nature of the erupted materials.
The Falcon Island volcanic rocks of the 1927 erup-
tion differ from those from Metis Shoal. The large and
abundant phenocrysts in the latter obviously were not
present in equal abundance in the described Falcon
Island ejecta, nor is it likely that they would have been
missed during the sampling of the island, as they are
readily visible.
The Falcon Island eruptives of 1885 consisted in
part of pumiceous blocks, containing bytownite and
augite phenocrysts (Harker, 1891), but lacking hyper-
sthene. Harker's description of the phenocrysts shows
them to be remarkably like those from Metis Shoal:
Some of the larger (augite) crystals are so associated
with the porphyritic feldspars as to prove that they
belong to an early phase of consolidation, but the bulk
of the minerals occur in ill-shaped idiomorphic crystals
scattered through the ground.
Harker's (1891) descriptions of glassy ash from
other of the Tonga Islands commonly mention crystals
of plagioclase, clinopyroxene (presumably augite), and
rhombic pyroxene (presumably hypersthene). It thus
appears that ash and pumiceous lava blocks, such as
those erupted at Metis Shoal recently, and at Falcon
Island in 1885, have been erupted at numerous times
during the geologic development of the Tonga Islands.
Relation Between the Pumiceous Blocks and the
Nature of the Metis Shoal Eruption
The generation of pumice from lava implies release
of pressure on lava containing dissolved volcanic gases,
perhaps during ascent of the lava from depth, or due
to the sudden explosive release of pressure on a near-
surface magma chamber due to failure of the chamber's
roof. The vesiculation process, combined with injection
of the vesiculating lava into cold, water-saturated older
volcanic debris, would lead to explosive steam genera-
tion and further shattering of the injected materials.
These materials would contribute to the formation of
a pyroclastic cone, containing very little massive, un-
vesiculated rock, and no extrusive flows. Such an erup-
tive scheme appears to fit the events recorded during
the eruptions at Metis Shoal and Falcon Island.
Figure 6 shows very large, angular blocks in the sea
at the base of the eruption cloud. It seems likely that
these large, undoubtedly pumiceous blocks actually
solidified within the volcanic cone, and were then force-
fully pushed upward as solid masses during continued
injection of new pumice-forming lava at shallow depth.
Wave erosion and further shattering of these blocks
during intense steam eruptions would rapidly reduce
these blocks to fine materials, which eventually form
the surficial materials of the cone. These pyroclastic
cones are readily reduced to below sea level once the
eruption ceases. Wave erosion is the dominant process
of reduction (Hoffmeister et al., 1929) ; postulation of
some sort of volcanism-related subsidence to account
for the rapid disappearance of these islands appears
to be unnecessary.
The high silica content of the Metis Shoal volcanic
glass suggests a highly viscous lava, compared, for
example, to basaltic lava. Such high viscosity charac-
terizes volcanic rocks that typically are violently
erupted in the Pacific region, and much less frequently
form flows.
Should an unusually voluminous eruption eventually
occur, a more permanent island may form at Metis
Shoal. Petrologically it may be much like Fonualei, an
active Tongan volcanic island containing dacitic
eruptives.
NUMBER 4 17
Suggestions for Additional Studies
Further data on the Metis Shoal pumiceous blocks
are needed. High-temperature experimental studies
would help to determine the origin of the phenocrysts
in the blocks. The preliminary petrographic and micro-
probe data appear to point to derivation of the by-
tovvnite phenocrysts by direct crystallization from a
magma with the composition of the pumiceous blocks;
mixing of genetically unrelated materials seems more
likely to account for the presence of olivine. If the
phenocrysts did, in fact, crystallize from a magma of
the bulk composition of the blocks, this could be
demonstrated under experimental conditions; the pos-
sible role of water and high total pressures may com-
plicate the experimental approach. Nonetheless, one-
atmosphere anhydrous experiments would be initially
useful. These experiments would also provide some
idea of the rocks' eruption temperature. Further use-
ful data include trace elements and radiogenic isotopic
ratios. All these data are pertinent to the further char-
acterization of the eruptives of the western arc of
Tonga, and hence to the possible nature of partial melt-
ing and subsequent evolution of magmas in the upper
mantle and crust beneath an active inner-island arc.
Careful monitoring of the volcanic eruptions of
Tonga should begin. Such monitoring should yield the
rate at which materials are being emitted in the region,
the dynamics involved in shallow submarine eruptions
and in the construction and destruction of volcanic
islands. Bathymetric surveys are needed in particular
before, during, and after eruptions to at least roughly
gauge volumes of eruptions. Early detection of erup-
tions and possible locations of as yet undiscovered
active Tongan submarine volcanoes may be possible
by the acoustic means which recently located active
submarine volcanoes just north of New Zealand
(Kibblewhite, 1966). This would require new listen-
ing stations near Tonga.
Acknowledgments
We wish to thank all those who provided the infor-
mation on the eruption, and who are noted in the
referenced sources. In addition, we acknowledge the
helpful assistance provided by Mr. J. Riechelmann
and other members of the Tonga government, the
British Council in Nuku'alofa, Captain William Stan-
ley of the ship Alaimoana, and 'Aplipate and Penisi-
mani Tupouniua of Nuku'alofa. Mr. Robert Citron,
Smithsonian Center for Short-lived Phenomena, pro-
vided helpful assistance in communications with Tonga
throughout the course of the eruption.
Literature Cited
Bryan, W.
1969. Low-potash Drift Pumice from the Coral Sea.
Geological Magazine, 105(5): 431-539.
Evans, B. W., and Moore, J. G.
1968. Mineralogy as a Function of Depth in the Pre-
historic Makaopuhi Tholeiitic Lava Lake, Hawaii.
Contributions to Mineralogy and Petrology,
17: 85-115.
Harker, A.
1891. Notes on a Collection of Rocks from Tonga. Geo-
logical Magazine, 8: 257-258.
Hatherton, T., and Dickinson, W. R.
1969. The Relationship Between Andesitic Volcanism
and Seismicity in Indonesia, the Lesser Antilles,
and other Island Arcs. Journal of Geophysical Re-
search, 75: 5301-10.
Hoffmeister, J. E.; Ladd, H. S.; and Ailing, H. L.
1929. Falcon Island. American Journal of Science, 18:
461-71.
Isacks, B.; Sykes, L. R.; and Oliver, J.
1969. Focal Mechanisms of Deep and Shallow Earth-
quakes in the Tonga-Kermadec Region and the
Tectonics of Island Arcs. Bulletin of the Geological
Society of America, 80: 1443-70.
Isacks, B.; Oliver, J.; and Sykes, L. R.
1968. Seismology and the new Global Tectonics. Journal
of Geophysical Research, 73 : 5855-99.
Kibblewhite, A. C.
1966. The Acoustic Detection and Location of an Under-
water Volcano. New Zealand Journal of Science,
9: 178-99.
Lacroix, A.
1939. Composition mineralogique et chimique des lavee
des volcans des iles de l'ocean Pacifique situees
entre l'Equataur et le tropique du Capricarne, le
175? de longitude ouest et le 165? de longitude est.
Memories de I'Academie des Sciences, Paris, 63 :1?
97.
Murdmaa, I. O.; Bezruko, P. L.; Zenkevich, N. L.; Prokopt-
sev, N. G.; and Repechka, M. A.
1969. Eruption of the Submarine Volcano Metis in the
Tongan Archipelago and its Effect on Sedimenta-
tion. Akademiia Nauk, SSSR, Doklady, 185(5):
1149-52. (In Russian.)
Phillips, C.
1899. On the Volcanoes of the New Zealand-Tonga
18 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
Volcanic Zones. Transactions and Proceedings of Troger, W. E.
the New Zealand Institute, 31: 510-51. 1935. Spezielle Petrographie der Eruptivgesteine. Deuts-
Richards, J. J. chen Mineralogischen Gesellschaft, Berlin. 360
1962. Catalogue of the Active Volcanoes of the World. pages.
Part 8: Kermadec, Tonga, and Samoa. Inter- Turner, F. J., and Verhoogen, J.
national Association of Volcanology, Rome. 38 1960. Igneous and Metamorphic Petrology. McGraw-Hill
pages. Book Company. 694 pages.
U.S. GOVERNMENT PRINTING OFFICE:1970 O?381-735