This dissertation has been microfilmed exactly as received 67-13,712 STICE, Gary Dennis, 1938- THE GEOLOGY AND PETROLOGY OF THE MANU'A ISLANDS, AMERICAN SAMOA. University of Hawaii, Ph.D., 1966 Geology University Microfilms, Inc., Ann Arbor, Michigan
206
Embed
STICE, Gary Dennis, 1938- THE GEOLOGY AND PETROLOGY …the Manu'a Group, was simply called Manu'a. Perhaps Ta'u is a relatively new name for the island. Friedlander (1910) states,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
This dissertation has been
microfilmed exactly as received 67-13,712
STICE, Gary Dennis, 1938-THE GEOLOGY AND PETROLOGY OF THE MANU'AISLANDS, AMERICAN SAMOA.
University of Hawaii, Ph.D., 1966Geology
University Microfilms, Inc., Ann Arbor, Michigan
THE GEOLOGY AND PETROLOGY OF THE
MANU'A ISLANDS, AMERICAN SAMOA
A DISSERTATION SUBMITTED TO THE GRADUATE SCHOOL OF THE
UNIVERSITY OF HAWAII IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF DOCTOR OF
PHILOSOPHY IN GEOLOGICAL SCIENCES
SEPTEMBER 1966
By Gary Dennis Stice
Thesis Committee:
Gordon A. Macdonald, ChairmanDoak C. CoxKost A. PankiwskyjAgatin T. AbbottJohn J. Naughton
TABLE OF CONTENTS
Page
LIST OF ILLUSTRATIONS. . • . . . . . • . . . . . . .. i
LIST OF TABLES.
ABSTRACT. . . .
INTRODUCTION
Background of the Project.Previous Investigations.Procedure of the Project .Acknowledgments .•..Geographic Description
Location and AreaHistory .•.Climate .••.Soils
GEOLOGY OF TA'U ISLAND
Nature and Distribution of Rock Types ..
ii
1
46
1011
13151926
27
Extra-caldera Volcanics . 30Intra-caldera Deposits of Ta'u. 35Pos t- ca lde ra De pos its . . • . 37Tunoa Shield Volcanics.. .....•.. 38Lua te le Shie ld Vo lcanics. . 41Post-erosional Deposits . .. .•.•. 43
Fa leas ao .Tuff Comp lex. . . . .. 43Fitiiuta Volcanics 46
Features Caused by Volcanic Activity ..Features Associated wEb Faulting.Streams and ValleysCoas ta 1 Eros ion . . .Coastal Deposition••.
PETROGRAPHY
Mic ros copy . . • • . . •• . . • . . • .
Page
73
808691949596
97100
100101102102104
107
Extra-caldera Rocks of Ta'u Shield. 107Intra-caldera Rocks of Ta'u Shield~ 112Tunoa Shie ld. . • . . 113Luatele Shield. • . 113Faleasao Tuff Compl~x .. .... . 114Fitiiuta Rocks. . . . . . • . 116Intrusive Rocks of Ta'u . . . . • . 116Extra-caldera Rocks of Ofu and Olosega. .. 117Intrusive Rocks of Ofu and Olosega. 118Intra-caldera Rocks of Ofu .....•.
Chemical Analyses
Theories on Oceanic Basalts • . . • . . •. 122Determinative Methods . . . . • . . . . .. 132Chemical Analyses of Rocks from Manu'a· 142
LIST OF ILLUSTRATIONS
Page
FIGURE 1. Location map of Samoan Archipelago. 14
FIGURE Z. Bouguer anomaly map of Ta'u Island. . . 29
FIGURE 3. Schematic diagram illustrating "gravitycollapse" and possible associated volcanicactivity. . • . . . • • . 57
FIGURE 4. Bouguer anomaly map for Ofu and 010segaIslands. . • .. .. • .. .... 78
FIGURE 5. Beach prof i 1e s on Ofu and Olos e ga Is lands. 104a
FIGURE 6. Stable joins of system Ne-Fo-SiOZ atlow and high pressures. System Ne-Fo-SiO Zat (a) 1050 0 C &nd 1 bar, and (b) 1250 0 Cand 33,000 bars . . . . . . . . . . . . . . 133
FIGURE 7. Alka1i:silica diagram of rocks from Manura 154
FIGURE 8. AFM diagram of rocks from Manu'a. . . . . . 156
PLATE 1. Geologic map of the Manu'a Islands ••••• backpocket
LIST DF TABLES
Page
TABLE 1. Percentage frequencies of w~nd.from
different directions by months atMulinu'u, Western Samoa. 20
and gnats are also present. Black lizards up to 8 inches in
length are common; snakes of the constrictor type are quite
rare. Fresh water eels up to 6 feet long and several inches
in diameter are found in the upper parts of Laufuti Stream
which cut across the high benches within the caldera on the
southern side of the island. Bats are found in several areas,
particularly in a pahoehoe tube of the Lua Tele shield less
than one-half mile west of Fitiiuta. Indigenous birds include
lupe (pigeon), fu'ia (a brown bird about the size of a mynah
with a similar "whistling" call), and tatio (a white bird that
lives in the ground but is found only at high elevations in
the uplands). Coconut crabs (utu) are quite numerous. Pigs,
25 •
chickens, and dogs comprise the domestic animals.
With the steady decline of population in Manu'a, the
"plantations" are rapidly becoming overgrown. Except for
copra, little produce is exported, even to Tutuila. The copra
production is also falling off each year. According to the
1960 census, the population of Ta'u is 1661 people, but only
about one-half of this number can be considered as permanent
residents of the island. Similarly, the 1960 census showed
the population of Ofu as 605 and of Olosega as 429, but again
only about one-half of these people can be found living on the
islands at any given time.
Soils. The soils of Manu'a are predominantly stony
latosols, or lithosols. Curry (1955) describes the same soils
in Western Samoa as weakly acid, stony, clay loams. Latosols
are formed from the weathering of basic rocks in the humid
tropics. The process results in the loss of silica, alkalies,
and alkaline earth, and a proportionate increase in the resi
dual material of the oxides of AI, Fe and Ti, as well as
primary minerals resistant to weathering such as magnetite
and ilmenite. In areas of steep slopes or young rocks, this
process of weathering is at an immature stage. Thus, much of
the soil contains fragments of relatively fresh rock. Such
soils are called stony latosols, or lithosols.
X-ray analyses of seven samples from Ta'u, five from Dfu,
and two from Dlosega indicate that very few clay minerals are
present. Nearly all of the samples contain halloysite, and
26.
over half contain gibbsite, but no kaolinite, dickite, or other
clay minerals were detected. The soils of Ofu and Olosega
have a larger portion of crystalline material than those of
Ta'u. Amorphous silica rather than halloysite occurs in the
samples taken from rngh altitudes on Ta'u, where there is more
rainfall. Hematite is also quite common, occurring in all of
the samples from Ofu and Olosega and in three from Talu.
On western Ofu and eastern Olosega lateritic soils have
developed due to deep weathering of thin-bedded lavas. These
more mature soils are probably not as fertile as the lithosols
because of leeching of potash and increasing acidity. Humus
produced from the tremendous supply of organic matter is
prevalent in all soils, especially in those of the level areas.
Fox (1962) states that lowland soils of Western Samoa compare
with the low-humic and humic latosols described from Hawaii
(Soil Survey of Hawaii, 1955), while soils developed from the
same rock types at higher elevations are similar to Hawaiian
hydro-humic la~osols.
GEOLOGY OF TA'U ISLAND
Nature and Distribution of Rock Types
Ta'u Island is composed of a major stratovolcano which was
built up on the crest of the Manu'a Ridge. The nature of the
lavas during the early shield-building stage is not known.
There may have been a long history of relatively quiet,
frequent thin lava flows from rift zones which steadily built
up a basal shield volcano, as is common in the Hawaiian
Islands. This early history is not revealed in the exposures
present on the island today. Collapse which may have been
the result of the extrusion of this large volume of lava
occurred near the summit of the volcano to form a caldera.
Following this, a period of somewhat more explosive eruptions
from cinder cones within the caldera and on its flanks con
tinued to build up the island. However, the eruptions were
not as frequent during this stage, and erosion became more
intense. Even in the lower sections exposed on Ta'u, there
are numerous local erosional unconformities, indicating a
long history of intermittent lava flows even before the for
mation of the caldera.
Two small shields built out the northeast and northwest
portions of the island. Tunoa shliad is located along the
regional rift zone. Luatele shield is part of a minor rift
zone extending down the northeastern slope of the main volcano.
These shields are composed mainly of thin-bedded olivine
basalt pahoehoe flows with dips of about 5°. The summits of
both of these shields collapsed to form central depressions
28·
that partially filled with subsequent volcanism. Small pit
craters are associated with both Tunoa and Luatele. After
the formation of these shields volcanism must have very
greatly or completely subsided, resulting in extensive
erosion. A great erosional unconformity resulted from
partial covering of this eroded surface by later flows. The
lava flows became so infrequent that an extensive sea cliff
formed around the island. Late stage lava flows sometimes
spilled over this cliff from cones on the flanks above. In
two areas, extensive late stage volcanism built large areas
of land in front of the old sea cliff. A tuff complex
approximately 1 square mile in area formed in this manner on
the northwest edge of the island. The historic village of
Fitiiuta is located on these late stage lavas, so that, even
though the lavas in this area look quite recent, they must be
at least 1500-2000 years old, i.e., prior to the settlement by
the Polynesians.
Thus, the rocks exposed On the island can be subdivided
into the following groups:
(1) Extra-caldera deposits, including:
a) pre-caldera volcanics, almost entirely lava flows,
erupted before the collapse of the main caldera
b) post-caldera deposits on the flanks of the
volcano during and after the formation of the
caldera
(2) Intra-caldera deposits, pyroclastics and lava flows
,::.71 .,,'\--
.
.....
- I
~e~i
29
~
s ,,-:-..... - .....
.~.
30.
within the main ~dera
(3) Tunoa shield volcanics
(4) Luatele shield volcanics
(5) Post-erosional deposits, including Faleasao tuff
complex and Fitiiuta volcanics
(6) Intrusive rocks
(7) Recent sedimentary deposits (alluvium, talus,
beaches, etc.)
Extra-caldera Volcanics. The pre-caldera rocks are
exposed in the high cliffs of the southern part of the island.
The summit of the island, Lata Mountain, is 3185 feet above
sea level,and represents the edge of a vertical fault scarp
1400 feet high. Assuming that the pre-caldera flows have an
average slope of 25°, their thickness measured from sea level
is 3515 feet. The base of the volcano is about 9000 feet
below sea level, giving a total section about 12,000 feet
thick. On the flanks of the volcano, because of the dense
vegetation, it is impossible to distinguish pre-caldera lavas
from post-caldera lavas that have masked much of the island.
However, the rocks cut by deep valleys on the north shore are
certainly pre-caldera lavas, although late flows may have
filled the floors in a few of these valleys at lower eleva
tions. Therefore pre-caldera lavas, since they cannot be
distinguished, are not mapped separately except in a few areas
where exposures are adequate. Otherwise the flanks of the
volcanoes are considered to be masked by post-caldera lavas,
TABLE 2. STRATIGRAPHIC SECTION OF TArU VOLCANO
Geologic Age
Recent
Formation
Sedimentarydeposits
Faleasao tuffcomplex
Fitiiutavolcanics
Thickness(in feet)
200±
400+, probab ly1100±
200+
General Description
Alluvium, beaches, and unconsolidated deposits of talus atthe base of present-day cliffs.The only consolidated materialincludes deposits of beach rockand the coral reef offshore.
At least three cones composedalmost entirely of palagonitizedtuff. Black ash, explosionbreccia and thin flows are associated. Dunite nodules, magmaticbombs, coral fragments and lavablocks, some of which containdunite inclusions, are present.
Picritic basalt and olivine basaltflows commonly containing dunitexenoliths; local cinder andspatter associated with vents.
Great Erosional Unconformity-------------------Late Pleistocene(some flows may beRecent and posterosional)
Tunoa volcanics 300±- Basalt and olivine basalt flows(aa and pahoehoe), relatively thinbedded, but difficult to distinguish from Taru volcanics. Gentlysloping shield with curvilineardepression associated with summitcollapse. Tumuli and recent surface flow features cover the floorof the depression; a few ash bedsare present locally. W
~.
Geologic Age
Late Pleistocene
Plio-pleistoceneto Recent
Formation
Luatelevolcanics
Extra-calderavolcanics
Thickness(in feet)
800.:!::
3515+
500±
General Description
Vesicular olivine basalts from asmall, gently-sloping shield, thesummit of which has collapsed toform a depression. A few pitcraters occur on the flank of thisshield.
Upper member composed of pic ritebasalts hawaiite and olivinebasalts which were extruded fromnumerous cinder cones located onthe flanks of the volcano. Generally these lavas are thick aaflows, but often they are only1_2' thick. Some inter-beddedpahoehoe flows and ash beds occur.These post-caldera volcanics forma thin cap of 0-500' over the mainpart of the volcano and are partlyequivalent to Luatele and Tunoavolcanics. Some recent lavas havecascaded over the present sea clliIand therefore must also be partlyequivalent to the post-erosionalFitiiuta lavas and the Faleasaotuff complex.
The lower membe.r is also composedof inter-bedded aa and pahoehoeflows of basalt, olivine bas~ltand picrite basalt. These precaldera lavas cannot be distinguished from the flows of the pos~
caldera cones, except in the deeplyVJ~.
Geologic Age
Middle Pleistoceneto Recent
Formation
Intra-calderavolcanics
Thickness(in feet)
600±-
General Description
eroded valleys of northern Ta'uand, of course, the caldera wallitself. Therefore these twomembers were mapped together,except in the one or two areaswhere pre-caldera lavas can berecognized with certainty.
Horizontal and nearly horizo~tal
ash beds cover most of the remaining caldera floor. The calderaapparently had not filled muchbecause some of the blocks downdropped during the collapse arenot buried. There are at leasttwo distinct benches, the upperof which is almost, if notentirely, covered by thin-beddedash deposits. The lower benchcontains three large pit cratersand several cones, some of whichare post-erosional and probablyRecent, having flowed over the seacliff that has eroded into thecaldera itself.
ww
34.
since the latter probably do cover most of the island.
Even in the pre-caldera lavas numerous local erosional
unconformities were noted, particularly at higher elevations.
The lavas are dominantly olivine basalts with lesser amounts
of picrite-basalts and feldsophyric basalts at low elevations;
aa and pahoehoe flows are interbedded, with nearly exclusively
aa flows in the upper part of the section. The lavas at low
elevations have dips of about 15_20°, but the uppermost beds
have higher dips of 25_35°. These beds flatten out near the
summit, so that the same beds exposed in the high fault scarp
on southern Ta'u dip less than 15°. The be~s on the north side
of the caldera, then, have approximately the same attitude as
the present ground surface.
The olivine basalts generally contain phenocrysts of
olivine approximately 2-4 mm in diameter. About 1/3 of these
rocks contain plagioclase microphenocrysts 1 mm or less in
diameter. Most of the basalts contain phenocrysts or micro
phenocrysts of plagioclase. A few flows exposed in the sea
cliff on the northern shore between 'Ao'auli and Lepula con
tain plagioclase crystals up to 5 cm long. The massive
portions of these flows are only 1/2-2 feet thick, but are
separated by thick layers of clinker (up to 7 feet). They
are almost entirely plagioclase phenocrysts and therefore are
curiously friable. One 30-foot section is exposed at the b~e
of 'Ao'auli stream valley and another 60-foot section only
about 200 feet upstream. In several places higher in the
35.
section thin massive central portions of aa flows only 1-2
feet thick are associated with clinker beds up to 8 feet
thick; these flows generally have steep dips (30-35°), which
may explain the large amount of clinker. One tuff bed and a
soil horizon were found at about 1200 feet elevation in
'Ao'au1i stream valley. The section up Avate1e Stream
(Table 3) is characteristic of the extra-caldera volcanics.
Intra-caldera Deposits of Ta'u. Intra-caldera lavas
include picrite-basa1ts (both ankaramites and oceanites),
olivine basalts, hawaiites and possibly one or two flows of
mugearite. In addition, there are extensive deposits of ash
and lapi11i tuff. The flows vary from 5 feet to more than
30 feet in thickness. Horizontal and nearly horizontal ash
beds cover much of the remaining caldera floor.
The caldera was only partially filled. At least two
and possibly three normal faults formed major benches on the
caldera floor. The highest bench, at the base of the fault
scarp which rises vertically 1370 feet to the highest point
on the island, is composed of a series of approximately
horizontal pahoehoe flows. Intercalated with them is a bed
of ash 3-4 feet thick composed of individual layers less than
1 cm thick. The ash is basaltic glass. Both overlying and
underlying the ash, the flows are usually 1-4 feet thick and
are mostly vesicular olivine basalts with several oceanites.
The vesicles are filled with limonite or clay which may be
the result of alteration and deposition by rising gases and
Red vitric ash lying unconformably on an oldererosional surface and dipping 3l o N 1
A series of aa flows of olivine basalt and oceanite1-3 feet thick separated by clinker beds 2-6 feetthick, ropping approximately 15°N 25
No exposures 100
Aa flow of basalt dipping 55°N apparently pouredover fault scarp to form an angular unconformitywith underlying aa flows 5
A series of thin (1/2- 1 1/2 feet) aa flowscontaining abundant plagioclase laths up to 5 cmlong separated by clinker beds up to 7 feet thick(dip = 28°N) 50
A series of thin (1/2-11/2 feet) aa flows of basaltwith occasional olivine phenocrysts separated byclinker beds 1/2-4 feet thick (dip = 26°N) 20
Total thickness of section 226
37.
hot solutions in the vent area of th~'~ulcano, or may be
simply the result of ordinary weatheting.
Post-caldera Deposits. Post-caldera deposits cover most
of the island. The geologic map (Plate I) shows the location
of some of the vents; undoutiEdly there are many others that
were not discovered due to the dense jungle. For the same
reason, lava flows from these vents are extremely difficult
to delineate, but in many areas flows have spilled over the
old sea cliff, and apparently most of the flanks of the main
cone have been covered by these later deposits.
The contact between the post-caldera deposits and the
Tunoa shield is based on topography, since there is no petro
graphic distinction between the rocks. The contact between
post-caldera deposits and lavas of the Luatele shield was
also based on topography. Even though the Luatele lavas are
The 'Ele'elesa-Leatutia depression may, then, be the true
caldera while the larger Afuatai-Leavania-Tali'i depression
is the head of a large lands lipped area. If such a mechanism
was operative, it may have been associated with the collapse
of the summit of the dome which formed the caldera. A gravity
collapse of this nature would have probably also been associ&ed
with tectonic collapse. Perhaps it was not a simple landslide,
but was also related to normal faulting.
The sea cliff in this area may, then, be a tectonically
controlled escarpment. The numerous faults parallel to the
coastline support this conclusion. Perhaps the suggested area
of gravity collapse contains a whole series of such faults
below sea level. Soundings were not adequate to determine
this. Also, at Laufuti, there are two small normal faults
which offset some of the dikes in the dike complex; these trend
N 60° E and dip 59° S with a 5-foot seaward downdrop.
It is possible that the Laufuti dike complex, the dikes
exposed in the cliff between Papaotoma and Si'ufa'alele Points
and the vent at Latoaise Point, are related to tectonic move
ments along the sea cliff, rather than being associated with
minor rift zones and the caldera of the volcano. Gravity
collapse of such magnitude could be accompanied by volcanic
activity, if it were related to tectonic movements.
If this gravity collapse did occur, it probably was not
a single short event, but slowly took place over a long period
of time, perhaps even continuing at present. Except for the
59.
en-echelon faults perpendicular to the direction of this
proposed movement, the beds are little disturbe_d .for a dis
placement of as much as 1400 feet. Some residents of Ta'u
claim they experience r'mafui' e," or earthquakes, once every
few years. It may be that no one has ever reported these
quakes for the simple reason that no "papalagi," or foreigner,
has lived on Ta'u for any extended period of time to confirm
or deny this statement. The gravity collapse may be associated
with a large-scale deflation of the Manu'a Ridge, particularly
in this area, where it is apparently sin~ing.
There may also be some effects from movements in the Tonga
Trench which trends N-S, its northern end reaching within 100
miles to the south of Manu'a. If extended north, it would run
between Manu'a and Tutuila. This is the same area where the
Manu'a Ridge is offset to the north from the regional arch
that forms the base for the rest of the Samoan Islands to the
west. Perhaps left-lateral faults are associated with the
movements in the Tonga deep, as right-lateral faults apparently
were in California, Alaska, and other continental borderlands
of the circum-Pacific region. There seems to be some similari
ty in the depositional environment of the rocks in the Coast
Ranges during the Jurassic and that of the present continental
borderland west of the Tonga Trench.· Radiolarian cherts,
greenstone, spilites, and pods of limestone suggest that the
depositional environment for at least part of the Franciscan
formation was a deep trench with nearby island arcs.
60.
The crescent-shaped cliff bounding the Tunoa depression
suggests an original circular depression typical of calderas
formed by summit collapse. If the fault scarp is extended
seaward to form such a caldera its diameter would be approxi
mately 1-1/4 miles. The escarpment varies from 200-300 feet
in height and has an average slope of about 34°. The presence
of one or possibly two pit craters, a cinder cone on the top
edge of the cliff, another cone at the base of the cliff, and
the alignment of 3 cones parallel to the southeast rim on the
flank of the shield are evidence that the summit collapse of
the shield was associated with volcanic activity. It is
possible that the depression was not bounded by cliffs on the
southwestern side, due to the intersection of the fault scarp
and the slope of the main volcano.
No exposures were found in the sea cliff south of Talu
village to indicate there had been a buried fault scarp going
out to sea where it might be suspected. The seaward trend
of the northern part of the fau~t scarp could not be checked,
because the sea cliff has been buried by subsequent eruptions
from the Tola tuff cone. Gravity measurements (Machesky,
1965) showed no high anomalies in this area. A high Bouguer
anomaly would be expected if this were a true caldera of
Hawaiian type, but of course, this is not conclusive
(Figure 2, p. 29).
The northeast rift zone of Talu, along which the Luatele
shield, Lualaitiiti pit crater, and at least two vents at
610
Fitiiuta are perfectly aligned, extends on out at least to
5000 feet below sea level. A line of four or five extra
caldera cones just to the south also can be extended out to
sea along the same submarine ridge. The soundings are not in
enough detail to determine whether or not there are actually
two parallel submarine ridges. At any rate, subaerially at
least, the northeast rift zone seems to be comprised of two
nearly parallel rifts.
62.
Geomorphology
Features caused by Volcanic Activity. Ta'u Island was
originally a roughly circular dome which was only slightly
higher than the present summit at Lata Mountain (3185 feet).
Later volcanic activity b~i1t two shields, one each along the
northeast and northwest areas and extended the island in those
directions. Caldera collapse and possible gravity collapse
in its southern part have transformed the island to a butter
fly outline.
Numerous cones dotted the flanks of the volcano and poured
lavas down its sides. These flank eruptions were not conti
nuous in their activity, because local erosional unconformities
between the various flows are prevalent. Most of the flanks
of the volcano have uppermost beds with about fue same attitude
as the ground slope, indicating that activity of these cones
has been relatively continuous to the present, although no
historic eruptions have occurred on the island. Nevertheless,
probably in middle or late Pleistocene, there was a long
period of cessation in volcanic activity, during which an
extensive sea cliff was cut around the island.
Later flows from the cones on the flanks of the volcano
have spilled over the sea cliff, indicating a rejuvenation of
volcanic activity. Renewed activity continued to build out
the island along the northwest and northeast rifts. A tuff
complex has buried the old sea cliff in the northwest part of
the island, and two or three vents have extruded lava in front
63.
of the former sea cliff behind Fitiiuta, at the northeast
corner of the island.
Features caused by Faulting. The most striking feature
caused by faulting on Ta'u is the 1400-foot curvilinear
escarpment on the southern slope of the island. Within 150
feet from the rim there are elongated, trench-like features
which have their longest dimensions parallel to the escarp
ment. A steady breeze comes "out of the ground" from these
holes. Actually, they are merely cracks developed near the
edge of the vertical and sometimes overhanging cliff. The
breeze is caused simply by the wind hitting the large face of
the cliff and coming through the cracks. Periodically, huge
blocks must falloff and crash to the ground 1400 feet below.
The escarpment is clearly a fault scarp.
Within the depression caused by the faulting there are
numerous topographic features, such as fault scarps, pit
craters, and cinder cones, which are discussed in detail
above (p. 53). There is a curvilinear fault scarp 300-500
feet high between Afuatai and the bench at 'E1e'e1esa. A
similar fault scarp of opposite displacement separates the
other side of the bench from Leavania 300 feet above. These
two faults were probably formed during collapse of the summit
of the volcano. Both of the two blocks which may have been
downdropped from the summit contain numerous en-echelon faults.
Most of the faults on the Li'u ridge trend N 50_70° E with
5- to 10-foot displacements. The majority of the faults on
64.
the Leavania-Tali'i ridge also trend N 50° E, approximately
parallel to the sea cliff. A few faults in the latter area
have a northwest trend and are 5 to 10 feet high. These
faults form "steps" down the ridges towards the sea.
The curvilinear escarpment forming Tunoa Ridge, the
Luatele depression, and the various pit craters described
above (pp. 60-61) are all features caused by minor collapse.
Streams and Valleys. The radial drainage pattern of the
original Ta'u dome is preserved today in a general way.
Recent eruptions from extra-caldera vents have buried many of
the older stream valleys. Daly (1924, p.132) noted, "the
deepest gorge observed is about 5 meters in depth." Except
for three stream valleys on the north and a very small part
of one on the south, this statement is nearly correct for the
. entire island. Ao'auli, Matautu'ao, and Avatele Streams have
cut canyons of over 300 feet in depth. Locally, Avatele
stream bed is over 600 feet lower than the adjacent ridges.
Apparently this portion of the northern coast has been left
relatively undisturbed by lava flows from post-caldera vents.
The beds exposed in these valleys even at the higher eleva
tions are certainly pre-caldera lavas. Had there been no post
caldera activity, the entire dome would probably have been
similarly dissected.
The only perennial stream, Laufuti, is located on the
southern side of the island where the dike complex is exposed.
The surface water travels in this stream less than 0.2 mile
65.
before reaching the sea. The stream continually pours out
thousands of gallons per minute from ground water reservoirs
formed by the dike complex, which strikes approximately per
pendicular to the stream. These reservoirs are tapped by the
stream bed. The Laufuti stream drains much of the bench on
the eastern side of the down-faulted area, extending over
1-1/2 miles back into Afuatai and nearly to the base of the
precipice that rises vertically to the summit of the island
(Lata Mountain). Except for the lowermost portion that is
fed by springs at the base of the sea cliff, the stream flows
only after heavy rains. However, water stays ponded in some
of the uppermost parts of the stream, and numerous large fresh
water eels (6-7 feet long, 4-5 inches in diameter) live there.
Several small streams drain the western portion of the down
faulted area, the Leavania-Taliti ridge.
Among other important streams on Tatu not mentioned above
are Saua on the eastern side just south of Fitiiuta, Amouli
and Vaita on the western side, Faga and Patau on the northern
side, and Faleiulu on the northwestern side. The latter is
the longest stream on the island, extending over 2 miles, and
also drains the largest surface area. It begins at its source
as a radial stream on the main dome of Tatu, and then follows
down the northwestern rift zone. Upon encountering the back
slope of Tunoa shield, the stream turns to the north, running
alongside Tunoa Ridge until it reaches the sea.
None of the streams on the island are of sufficient
66.
maturity to produce alluvium finer than sand size, except for
minor amounts of soil removed as sheetwash. Most of the
stream beds are composed of large boulders 1-10 feet in
diameter. The steep gradient (usually about 25_30°) and the
large volume of water moving downstream after the frequent
heavy rains make movement of large boulders downstream
possible. The writer has witnessed an ordinary flash flood
at 'Ao'auli (all too nearly from the stream bottom itself),
during which cobbles and pebbles were moved downstream almost
as if they were made of cork. Boulders several feet in
diameter tumbled slowly seaward. Transport downstream of the
larger boulders probably is greatly facilitated by breaking
and chipping as they crash down the steep stream bed.
Coastal Erosion. During the long period of quiescence
sometime during the Pleistocene, an extensive sea cliff was
formed which can be traced completely around the island. The
height of this sea cliff is usually about 200 feet, except in
two areas. Along the northern shore the sea cliff increases
to a maximum height of 2400 feet. Offshore soundings of this
northern coast consist only of four lines running approximately
parallel to shore. The slope of the cliff in this area is
34_40°, whereas the beds dip 28_34°. The slope of the ocean
bottom offshore averages 22° down to at least 6000 feet.
Since the dips of the beds seem to be approximately the same
as the submarine slope, there is no reason to postulate land
sliding on the order suggested for the south coast of the
67.
island. Nevertheless, landsliding on a smaller scale is an
important erosional agent in this area.
At the base of the steep slopes in this area are several
large landslides. At Faga a landslide has carried more than
8 million cubic yards of material down the steep slope. Sub
sequently the talus boulders near sea level were cemented
into a beach conglomerate (boulders and cobb~s cemented in a
calcareous sand matrix). Much smaller landslides are found
at various other localities at the base of the sea cliff.
Landsliding and subsequent removal of the talus by wave
action is one of the important agents of erosion in Manu'a and
probably in most oceanic islands with high sea cliffs.
Notches at various localities in the tuff complex about
5 feet above high tide indicate that previously there was a
higher stand of the sea. This same 5-foot bench has been
noted on many islands throughout the Pacific Ocean (Stearns,
1941), and is therefore considered to be a eustatic change,
probably brought about largely by the formation of more ice
at the poles. This bench is noted only in the tuff complex
on the northwest side of the island. Apparently this stand
of the sea was of relatively short duration, so that only the
easily eroded tuff bears evidence of its occurrence. The
present sea level, of course, is eroding away this bench,
making it difficult to estimate the relative changes in sea
level.
Coastal Deposition. A terrace 13-15 feet above sea level
68.
occurs in several areas around the island. The most extensive
of these is the one on which Ta'u village i~ built. Other
examples of this l~ to IS-foot terrace are found at Faleasao,
Faga, Saua, Tufu, Amouli, and Si'ufa'alele. The terraces on
the southern part of the island are composed of sand and coral
shingle, whereas the others are entirely sand. The layers of
coral shingle are exposed in escarpments on the backbeach.
It is probable that these terraces were formed during a
higher stand of the sea perhaps related to the +12-foot stand
recognized in Hawaii (Ruhe, Williams, and Hill, 1965).
Nevertheless, during hurricanes, waves can and do come up over
the tops of these terraces. During the hurricane of 1959,
waves destroyed a trail more than 200 feet inland at the base
of the old sea cliff near Saua. However, the villages at Ta'u
and Faleasao have not been demolished by storm waves during
historic time. It is not likely that Faleasao ever has been
under heavy wave attack, because of good protection afforded
by the coral reef and the rim of the tuff cone which nearly
encircles the village.
Most of the coastline on Ta'u consists of narrow beaches
less than 100 feet wide between mean water level and the vege
tation. During high tide the waves often wash up the foreslope
to the berm crest. Therefore, the vegetation usually does not
extend beyond the berm crest. The foreslope of the beaches is
getween 10 and 13°. Beachrock (cemented calcareous sand)
occurs in many places along these beaches. When present, it
69.
is near sea level or offshore. The beachrock in most areas
appears to be out of equilibrium with the present sea level
and is now being eroded. The slope of this beachrock is
usually somewhat less than the foreslope of the present beaches,
possibly indicating that it was formed during a higher +5-foot
(7) stand of the sea. The median grain size of the beach
material varies from medium-grained sand (0.29 mm) to gravel
(3.50 mm), whereas the beachrock is usually composed of fine
to medium-grained sand. In a few areas cobbles and boulders
are cemented in a matrix of medium to coarse sand (e.g. at
Faga and Ma'efu).
Around the island of Ta'u is a nearly continuous fringing
coral reef. Nowhere is the reef front more than 800 feet from
shore. Considering the age of the island, a more mature coral
reef would be expected. The island of Tutuila also has a
poorly-developed reef. However, soundings clearly indicate
that a barrier reef which extended over a mile offshore from
the present coast of Tutu~la has recently been submerged a
few hundred feet (see u.s. Coast and Geodetic Survey Chart
#4190). The only offshore soundings in Manu'a were completed
by the U.S. Navy in 1939(Plate 1). These soundings are
sparse, particularly for critical areas close to shore, and
their accuracy is questionable. (The islands themselves are
positioned 1.7 miles too far west on this chart.) There is
nO indication of a submerged reef around Ta'u, but the data
are too sparse and inaccurate to negate this hypothesis. The
70.
narrow reef suggests that the island has undergone some sink
ing since Pleistocene time.
Usually a white sand beach is associated with the reef.
Of course, there is no lagoonal area. The reef flat contains
patches of sand, coral and coralline algae, whereas the fore
reef is composed of prolific colonies of corals. At various
places along the reef there are channels about 15-25 feet wide
and 10-15 feet deep. The people living in the villages use
these channels as passes for their long-boats. Since there
is no anchorage in Manu'a, communication between sea-going
vessels and shore must be made by long-boat. Ta'u village,
for example, has two main channels, one at the southern end
of the village (Si'ufaga) and one at the northern end of the
village (Luma).
Water transported over the reef front by waves builds up
a hydrostatic head and runs back out to sea through these
channels. Current measurements were obtained by using a
tether ball and string. Dye
surface current measured was
was also employed. The maximum
2.8 feet/second near the main
channel at Luma during an ebbing tide. The maximum current
during flooding tide for the same area was 2.1 feet/second.
The difference is due to the greater hydrostatic head produced
at low tide between the water behind the reef and sea level.
Besides the main channels, there are numerous smaller tri
butary channels running alongshore that join the main channel
in a manner similar to a subaerial stream drainage system.
71.
Another set of small channels cut perpendicularly across
the forereef and disappear at a depth of approximately 30
feet, where they merge with the slope of the forereef.
Usually the channels are spaced about 75 feet apart and are
4-8 feet wide. Rarely do they exceed 200 feet in length.
The heads of these channels terminate in the surf zone as
cirque-like features, some of which are 15 feet deep. The
maximum depth of the channel relative to the reef surface is
usually at the head of the channel, so that the gradient of
the channel bottom is considerably less than the slope of the
forereef. The bottoms of these channels usually are covered
with a thin veneer of sand, but often are paved with coral
boulders. Strong oscillatory currents dependent in part on
wave action occur in these channels. They are therefore
referred to as "surge channels." The sides overhang due to
the more rapid growth of coral at the reef surface. Inoome
channels the overhanging corals on opposite sides of the
channel have grown together to form natural bridges.
The species of coral identified on Ta'u Island are:
Maendraeuaensis (Hoffmeister)
Acropora hyacinthus (Dana)
Pocillopora eydonxi (Milne-Edwards and Haynes)
Psammocora contigua (Esper)
Goniopora muscosa (Wells)
Porites lobata (Dana)
Favia rotumana (Gardiner)
72.
Millepora tenera (Boschma)
Acropora formosa (Dana)
Pavona frondifera (Lamarck)
Pavona divaricata (Lamarck)
Favites magna (Hoffmeister)
Goniastrea retiformis (Lamarck)
In addition, Halimeda, Porolithon, Goniolithon and other un
identifiable calcareous algae occur.
GEOLOGY OF OFU AND OLOSEGA ISLANDS
Nature and Distribution of Rock Types
Ofu and Olosega Islands are comprised of a complex
volcanic pile which was built up on the crest of the Manu'a
Ridge. At least seven Pliocene(?) cones exposed on these
islands are approximately aligned along the regional rift.
Two of these cones were built mainly by lava flows to form
coalescing shields. One is centered off the northwest coast
of Olosega near Sili, and the other is centered at Alofa on
the northern coast of Ofu. The older cones buried by lavas
from these shields include a small cinder cone at Tauga Point
on northwestern Ofu, a nearby tuff cone at the western end
of Samoli beach, a composite cone exposed in the cliff behind
Tolaga on southeastern Ofu, an explosion breccia cone with an
associated intrusive plug at Fatuaga Point on eastern Ofu,
and another tuff cone at Maga Point on the southern tip of
Olosega.
Ofu and 010sega Islands, plus the cones along the crest
of the Manu'a Ridge between Olosega and Ta'u, may have once
been physiographically similar to the present upland areas of
Upo1u and Savai'i, which consist of a double chain of cones
and craters that form the backbone of those islands (Kear and
Wood, 1959). Late stage cones along the central ridge on
Upo1u and Savai'i have poured lava over the older erosion
surfaces and across the reef to form extensive coastal areas
comprised of lowlands and swamps. The same general history
74.
applies to Tutuila, i.e., a line of cones along a regional
rift zone was modified by later volcanism (Stearns, 1944).
Apparently Ofu and Olosega have not been as active in late
stage volcanism, because there does not appear to be any ex
tensive erosional surface buried by later lavas. Also, the
islands were probably lower than Savai'i and Upolu during
most of their history, as indicated by the presence of
palagonitized tuff cones, although no included coral fragments
were found in these deposits.
Most of the total volume of the islands above the present
sea level is composed of lava flows from the two shields, both
summits of w~ich collapsed. It could not be determined
whether these coalescing shields contained two separate
calderas or a single large caldera. Offshore soundings
(Plate 1) indicate the presence of a caldera northwest of
Sili. The only other evidences of the Sili caldera are the
numerous inward dipping dikes exposed in the high cliff
behind Sili and the attitudes of the lava beds on Olosega.
The depression at A'ofa On the north coast of Ofu is
clearly the result of summit collapse of the other shield.
Vertical dikes exposed in the sea cliff behind Samo'i parallel
the fault boundary of the depression and are thereby related
to summit collapse of the shield. The lava beds on south
western Ofu dip gently away from this depression. These beds
are deeply lateritized, as are some of the beds on south-
eastern Olosega.
TABLE 4. STRATIGRAPHIC SECTION OF OFU AND OLOSEGA ISLANDS
Geologic Age
Recent
Formation
Sedimentarydeposits
Post-erosionalvolcanics
Thickness(Cn feet)
200±
255+
General Description
Unconsolidated deposits of talusat the base of present day seacliffs, alluvium, and beaches.Consolidated deposits of beachrocks and offshore coralline reefu.
The remnant of a palagonitizedtuff cone forms Nulutele andNulusilaelae Islets off westernOfu. Two or three thick flows ofhawaiite and olivine basalt mayhave filled former deeply erodedvalleys.
___________________Gre at Eros iona 1 Unconformi ty _
Pleistocene
P lio- p Ie is tocene
Intra-calderavolcanics
Ext ra- ca Ide ravolcanics
400+
2175+
Thick flows of basalt, picritebasalt and olivine basalt wereponded within the Alofa caldera.One buried cinder cone is exposedin the sea cliff.
Upper member is composed of shiel~
building pahoehoe and aa flows ofpicrite-basalt, basalt and olivinebasalt-from the Sili shield onOlosega and from the Alofa shieldon Ofu. These two coalescingshields may have shared a singlecaldera. No extensive erosional
unconformity was recognized, but~Ln
Geologic Age Formation Thickness(in feet)
1000+
General Description
the flows seem to occur as thinbedded pahoehoes in the lowestpart of the section and gradeupward into more frequent occurrences of interbedded aa flows.Thick aa flows of hawaiite andpicrite-basalt culminate thisseries. These rocks were extruded prior to the formation ofthe caldera. The thin-beddedpahoehoe flows of southwesternOfu and southeastern Olosega aredeeply weathered and lateritized.
The lower member is composed ofa series of at least five cones,which are approximately alignedalong the regional rift. TheseCOnes were formed prior to thesummit collapse of the Sili andAlofa shields, perhaps even priorto the building of the shieldsthemselves.
-....J0'\
77 .
If the boundary of the depression at Alofa were extended
seaward as a crude circle, it would have a diameter of approxi
mately 1 mile. This depression may not have been quite large
enough to be called a caldera, which by definition must have
a diameter of at least One mile, but the dimensions of its
seaward portion are not definitely known, and its average
diameter may have been considerably more than a mile. Since
this feature represents the collapse of the summit of a shieN,
it will be referred to as the A'ofa "caldera," and the shield
containing it will be called the Alofa shield. Judging from
the attitudes of lava beds and the character and position of
the dikes, the collapse of the shield centered northwest of
Sili was definitely large enough (about 1.7 miles in diameter)
to form a caldera; it will be referred to as the Sili caldera,
and the associated shield will be called the Sili shield.
By assuming that these two depressions were approximately
circular, it is estimated that they were about 0.7 mile apart.
However, no field evidence was found to confirm or deny this
supposition. No conclusive evidence for the presence of two
shields was obtained from the gravity data (Figure 4). It is
possible that, rather than two shields, there was only one
larger shield with a single elliptical caldera. The south
eastern boundary of such a caldera would be the same as that
proposed for the Sili caldera; but the southwestern boundary
would include the Alofa depression in a relationship similar
to that of Kilauea Caldera and Kilauea Iki crater on Hawaii
J
----1--·-,---\ \ .
\ \\ \\ \\\III
\\\II.
....__~~l
I f,'ile
1Gs., [:~-(~t-
FtOunc.: <-k
BOUGUER Aj\!Oi'll/\LY MAP OF opOFU ,i\j\!D OLOSEGA ISLANDS '(; #
l\iviERICAN SAMOA'-'-'CCOlOCIC Oout·:DMIYof CALDER,\ ofter G. sncE (\ ~J
' ..3-"', i1\'\\ 26:/\) b) //i \~~ ~ \ .-.--;,,~ /,...."\. I . ~.,OJ ,---\,-<;:;:,::,,~,o__"~~>'_:'" ';~,'." __ :' "..--,. ~./'
\ / '-. /' ..""'\ \ / \ ~'-:'~''::;-§'''''''--T) ........ _._J.- J II / I I "'!: / / / \/ I I~ / \I ' \\\ \
l~x"<9
- "" .(j , {)
-- So..,,~ 'OS
~Oo
Jt." \ ',,1
I ,~- t· IC/-- . -- ---- .._--_.
--...s00
79.
(Stearns and Macdonald, 1946). The dimensions of this caldera
would be about 3 miles by 1-1/2 miles. The attitudes of the
beds in the cliff on northern Ofu and the absence of shield
building lavas exposed in this area suggest that the presence
of two smaller eruptive centers is more likely. Tentatively,
then, the shield-building lavas on Olosega are mapped as part
of the Sili shield, and those on Ofu are mapped as part of the
Alofa shield.
The A'ofa caldera was partly filled with thick olivine
basalt, hawaiite, and ankaramite flows. One buried cinder
cone was found within the caldera at Sinapoto. There are
several parasitic cones along the flanks of the volcano. Some
of these cones, as well as some of the uppermost thick ankara
mite flows on the islands, may also be post-caldera deposits.
Since these units could not be separated either in the field
or petrographically, however, they were mapped together with
the earlier rocks as extra-caldera deposits. Afterwards,
there was an extensive period of volcanic quiescence during
which deep valleys were carved into the slopes, and a sea
cliff 200-400 feet high was formed around the islands. There
may also have been large-scale foundering along the southern
coasts of the islands. Possibly this could have been ini
tiated by the overburden of volcanic deposits on the earth's: crust.
Following this cessation in volcanic activity, it appears
that two or three thick hawaiite lavas flowed down a few of
the old valleys on the southwest side of Ofu at Tufu.
80.
Nu'ute1e and Nu'usi1ae1ae Islets, off the southwestern part
of Ofu, are the remnants of a Recent tuff cone.
The rocks exposed on Ofu and 010sega thus can be grouped
as follows:
(1) At least five pre-caldera cones approximately
aligned along the crest of the Manu'a Ridge
(2) Two (or possibly only one) later coalescing shields,
the summits of which collapsed
(3) Intrusive rocks
(4) Intra-caldera deposits of the Alofa caldera
(5) Post-erosional(1) flows and Nu'utele tuff cone
(6) Sedimentary deposits
Extra-caldera Volcanics. The To'agu composite cone and
the Fatuaga breccia cone are probably the oldest features
exposed on Ofu and 01osega. Unfortunately, they can be seen
only in the high cliffs on eastern Ofu. No outcrop could be
examined closely, but numerous talus blocks scattered along
the shore gave some indication of the rock types present in
the cliff above. The To'aga composite cone was built up in
large part by alternating sequences of aa flows, pahoehoe
flows, cinder and tuff. In the high vertical cliff the
exposures are disappointingly few, due to the dense vegeta
tion, but it is apparent that this composite cone was buried
by subsequent flows from another source, probably the A'ofa
shield. An ash bed representing the former surface of the
cone forms an angular unconformity (on1ap) with the later lavas.
81.
West-dipping beds are found exposed at various places
in this cliff as far west as Va'oto. Exposed in the cliff
behind Va'oto is a bed of red ash up to 20 feet thick. It
extends up the slope of the ridge to Lepu1a. A lower
impermeable tuff bed provides a spring in the cliff behind
Va'oto that is utilized for drinking water in Ofu village.
If tm thick red ash bed in this area is the same uppermost
bed of the composite cone exposed in the cliff behind Tolaga,
then this cone must have extended at least to Valoto giving
a minimum diameter at sea level of 1-1/2 miles. The summit
of the cone was probably located about 0.4 mile due east of
Tumu Mountain at approximately 700 feet above sea level. It
is not possible to determine this accurately, because the
exposures are such that one is given little better than a
two-dimensional view.
There probably was another larger cone of at least the
same height that was centered offshore between Ofu and Olosega,
about 1500 feet due east of Fatuaga Point, which is formed by
a related intrusive plug. A tuff bed exposed in the cliff on
the north side of Olosega village is probably part of the
eastern slope of this same cone, giving it a minimum diameter
at sea level of 1.8 miles. This eastern portion of the cone
has been buried by subsequent aa flows from the Sili shield.
A large percentage of the total volume of this cone is
comprised of an explosion breccia. The breccia is almost
entirely made up of fist-sized blocks of at least six distinct
82.
kinds--one vesicular pahoehoe basalt, one ankaramite, one
dike rock, one olivine basalt, and two aphanitic basalts.
The matrix makes up less than 5% of the total rock. It is
composed of very well indurated vitric ash, containing
occasional crystal fragments of olivine 1-2 mm in diameter.
Near Tauga Point at the western end of Samoli beach, an
old tuff cone has been covered by a series of about 6 aa flows
which probably were erupted from the Alofa shield to the east.
The northern half of the cone has already been eroded away by
wave action, and a horizontal flow of basalt that was ponded
within this crater now forms a bench about 15 feet above sea
level. A 10-foot thick yellow lapilli tuff bed which dips
20° away from the cone and pinches out 500 feet to the east
is probably associated with this cone. This tuff overlies a
+SO-foot thick unstratified, palagonitized tuff that is
probably related to the deposits of the Alofa shield. Table 5
is a stratigraphic section at Samoli on northern Dfu.
An adjacent but slightly later small cinder cone at Tauga
Point has also been buried by later aa flows from the Alofa
shield. Numerous ribbon and spindle bombs occur in the ash
beds of this cone.
Thus, pahoehoe and aa flows from the Alofa shield over
lapped, and eventually buried, the older coneso The pre
caldera lavas of the shield are predominantly thin-bedded
pahoehoe flows with frequent interbedded aa flows and
occasional thin beds of ash and tuff. The lava flows are
83.
TABLE 5. STRATIGRAPHIC SECTION OF EXTRA-CALDERA
VOLCANICS ON OFU UP TIA RIDGE FROM THE
WESTERN END OF SAMO'I BEACH
Top(Estimated 220-foot elevation)
Nonporphyritic, deeply weathered vesicularflows dipping 15° W, mostly aa flows 2 to 5 feetthick, possibly some pahoehoe flows also
Gray nonporphyritic aa flows of hawaiite about8 feet thick with clinker beds 2 to 3 feet thick,dipping 20° SW, one ankaramite flow with thinveinlets of almost pure olivine and augiteconcentrations (groundmass less than 5%)
Yellow palagonitized vitric lapilli tuff withpumice lapilli and containing some thin ash beds,ribbon and cow dung bombs in ash bed, tuff pinchesout 200 feet to the west
Brown unstratified palagonitized vitric lapillituff, with pumice lapilli, and rare ash layers,dipping 14° W
Deeply weathered 2- to 3-foot thick pahoehoeflows, dipping 14° NW
Talus, containing much cinder as drift
Total thickness of section
Thickness(feet)
90+
30
10
40
20
20
210+
84.
basalt or olivine basalt except for a few flows of picrite
basalt and hawaiite. Macdonald (1944) described a hawaiite
which was collected by Stearns from a talus block at the base
of the sea cliff near Tauga Point on Ofu. This talus block
must have come from the dense, thick aa flow near the top of
the sea cliff, which represents the upper portion of the
shield. The thick ankaramite and olivine basalt flows at
Tumu, the uppermost part of Ofu, are essentially horizontal
and probably represent nearly the original summit of the A'ofa
shield.
Pyroclastic pre-caldera deposits exposed on northern Ofu
include a +50-foot thick unstratified pa1agonitized tuff, a
few thin red ash beds, and a 3-foot thick breccia. Red cinder
found in the soil 700 feet due north of Tumu is probably from
an old post-caldera cinder COne in that area.
At Maga Point, on the southern tip of 01osega, is another
old tuff cone that has been buried by later flows. The northern
slope of this cone is 34°; its summit rose at least 250 feet
above the present sea level. The lower beds of this cone are
lapilli tuff with horizons locally rich in basalt blocks up
to 6 inches across. The upper 50 feet of this cone is comprised
of cinder and red ash. A 35-foot thick, dense flow of hawaiite
has ponded within the crater of Ehe"cone. A series of thin
pahoehoe flows has overridden the cone and forms Maga Point.
Interbedded aa and pahoehoe flows overlie this sequence. The
aa flows become increasingly dominant up-section. These flows
85.
are part of the main shield centered northwest of Sili.
The lowermost exposure of the pre-caldera lavas from the
Sili shield is found northeast of Sili, near Leaumasili Point.
A sequence of thin-bedded pahoehoe flows (1-2 feet thick)
with a few interbedded aa flows (up to 10 feet thick) are cut
by several thin dikes (varying from 1 to 4 feet in thickness)
that are parallel to and dip slightly towards the caldera
boundary to the west. The pahoehoe flows vary from 1 to 6
feet in thickness but are generally only 1-3 feet thick.
, These flows were erupted from vents near the center of the
shield before the formation of the caldera. They dip away
from the former summit .located northwest of Sili with atti
tudes that are usually less than 10°.
Exposures in the steep cliffs behind Sili and Olosega
villages show that the interbedded aa flows generally increase
in number and become thicker up the section, ranging from 2
or 3 feet to over 20 feet. As on Ta'u, some of the steeply
dipping aa flows (locally up to 30°) have as much as 10 feet
of clinker associated with only 1 foot of the massive central
portion of the flow. Most of the lavas on Olosega are basalts
and olivine basalts. Usually the basalts are nonporphyritic,
but· often they contain microphenocrysts of plagioclase less
than 1 mm long.
Thick aa flows comprise most of the upper 800 feet or so
of the section, which extends to the summit at Pifumafua
Mountain. The dips of these later flows are relatively steep,
86.
ranging between 15° and 20°. A series of 3 or 4 flows of
hawaiite with a total thickness of over 75 feet are the
highest flows in the section that could be closely examined.
The rock is quite fresh except along joints where some manga-
nese deposition has occurred. A few exposures were found near
the summit, at Piumafua Mountain, but the lavas are too deeply
weathered to be identified. The only other outcrops found
stratigraphically above these are thick aa flows exposed in
the cliff behind Sili.
Thus, it is possible that all of the upper 600 feet of
the cone is capped with hawaiites. There seem to be few
picrite-basalts among the lavas of this cone, whereas both
ankaramites and oceanites are abundant on Ofu. A few oceanites
and ankaramites are exposed along the eastern coast of Olosega.
One of these ankaramite flows contains a concentration of
augite phenocrysts in the highly vesicular surface of the flow.
The frothy layer is 2-3 cm thick and contains 90% augite
phenocrysts up to I cm long. Table 6 is a stratigraphic sec
tion of extra-caldera volcanics at Tafalau on eastern Olosega.
Intrusive Rocks. The intrusive rocks include numerous
dikes exposed in the cliffs of Ofu and Olosega, and one, or
possibly two, plugs on eastern Ofu at Fatuaga Point. Only one
dike was found intruding the intra-caldera volcanics. Nearly
all of the intrusive rocks are probably older than the intra
caldera volcanics. The plug at Fatuaga Point may be part of
the core of the explosion breccia cone, which is probably the
oldest feature now exposed in Manu'a. There may be a smaller,
TABLE 6~ STRATIGRAPHIC SECTION OF EXTRA-CALDERA
VOLCANICS IN THE CLIFF BEHIND TAFALAU, OLOSEGA
87.
Top Thickness(feet)
Nonporphyritic, weathered vesicular pahoehoe 90flows 1 to 7 feet thick, dipping 12° SE
Olivine basalt, moderately vesicular aa flows 305 to 7 feet thick with clinker beds 3 to 5feet thick, dipping 24° E
Olivine basalt with feldspar microlites forming 15a dense, massive flow
Nonporphyritic, vesicular aa flows 1/2 to 3 feet 20thick with clinker beds 1/2 to 7 feet thick,dipping 24° E
Moderately vesicular aa flows of olivine basalt 354 to 7 feet thick, clinker beds 3 to 4 feetthick, with much red cinder and ash, dipping 24° E
Brown palagonitized vitric crystal tuff with 5olivine and augite crystals, lying unconformablyon lower flows, tuff dipping 24° E
Minor angular unconformity due to erosion
Nonporphyritic basalt with feldspar micro litesforming dense aa flows 3 to 5 feet thick,clinker beds I-foot thick, dipping 30° E
basalt containing small laths of feldsparphenocrysts randomly oriented and rareolivine phenocrysts
Talus of blocks at base of cliff
Total thickness of section
30
40
10
10
65
350
88.
related plug about 1500 feet to the east. A concentric hill
there has the suspicious ovoid shape of an intrusive plug.
The rock cropping out on this small hill is aphanitic and
appears to be related to the dikes which form the razor-back
ridge on eastern Ofu. Perhaps the fine-grained borders of a
plug have net been eroded to reveal coarser-grained rock
comprising its center.
The Fatuaga plug represents a hypabyssal intrusion of
ankaramite magma. It was recognized and described by Daly
(1924, p. 134) as an elliptically-shaped plug with a maximum
diameter of 120 feet and a minimum diameter of 80 feet.
Actually it is considerably larger than this, probably at
least 500 feet by 300 feet. The highest Bouguer gravity
anomaly in Manu'a (more than +310 mi1liga1s) was recorded near
this plug (Machesky, 1964, see Figure 3). There is a gradation
in grain size from olivine-titanaugite gabbro in the central
portions of the plug to ankaramite near the peripheries. There
is also a gradation from a roughly ovoid plug near sea level
to a much more elongate ankaramite dike at higher elevations.
The general trend of the longest diameter of this feature
is approximately N 15° Wand vertical. At Sunu'itao, the
western edge of the plug cuts explosion breccia beds that trend
approximately N 85° Wand dip 12° toward the north. At sea
level near the top of the shark's-tooth peak at Vainu'u1ua,
this breccia appears to be trending N 30° E and dipping much
more steeply, about 30° towards the west. Some of the blocks
89.
included in the explosion breccia are identical to the in
trusive olivine gabbro, except that they are slightly finer
grained. Therefore, it seems likely that the intrusion
occurred nearly contemporaneously with the deposition of the
breccia. Included within the plug is a pod of brecciated
pahoehoe flows of vesicular basalt, which apparently was broken
off from the chamber wall and carried up in the magma chamber
during the intrusion. These flows, though brecciated, are not
the same as the heterogeneous explosion breccia which comprises
most of the associated cone.
Near sea level the plug is cut by numerous thin dikes that
strike approximately E-W and dip 55-85° toward the north.
Most of these dikes are dense and aphanitic, but some of the
thicker ones are vesicular, suggesting, along with the open
miarolitic texture of the olivine gabbro, proximity to the
surface at the time of intrusion. These dikes are also
parallel to the complex of dikes which form the backbone of
the ridge extending from Le'olo Ridge eastward to Olosega.
All of these E-W-trending dikes in turn are probably related
to the collapse that formed the Sili caldera.
The dikes forming the razor-back ridge of eastern Ofu
vary from I to 20 feet in thickness and are nearly vertical,
but some may dip steeply towards the north. The dike complex
in this area is about 400 feet wide. Most of the dikes are
dense basalts, although olivine basalts, ankaramites, and
feldspar-phyric basalts also occur. Large talus blocks of
90.
diabase lie along the shoreline north of Vainu'ulua and were
derived from the thick dikes ~t the top of the ridge.
The complex continues across Asaga Strait to the 2000-foot
cliff behind Sili village. Near Tamatupu Point, the western
most tip of Olosega, thick dikes with dips as low as 50 0 N
occur. These may be slightly curved, but this was not
definitely established, because the individual dikes could
not be traced for great enough distances. Northeast of
Piumafua the number of dikes paralleling the face of the cliff
falls off sharply. North-dipping and vertical dikes related
to this complex can be observed in the cliff behind Olosega
village. The dikes decrease markedly in number both up-section
and away from the caldera. A few apparently related dikes
trending about E-W are found as far as 3000 feet south of the
cliff behind Sili. They cut steeply across flows dipping
10_20 0 away from the caldera.
At least one sill can be seen in the cliff at Faiava.
It is horizontal and appears to be about 400 feet long and 30
feet thick. A low-angle dike near the extreme eastern side
of the sill discordantly intrudes a series of pahoehoe flows
trending about N 50 E and dipping 15° E. It appears that
this dike is the feeder dike for the sill, but this relation
ship cannot be seen clearly due to the dense vegetation.
Most of the other dikes in this portion of the cliff strike
about N 30 0 E and are vertical, indicating that the caldera
boundary swings from east-west towards the north in that area.
91.
The strike of the beds there seems to be about N 5° E, which
also indicates that the former summit of the shield was to
the west. A few of the thinner vertical dikes exposed in the
cliffs behind Si1i and Olosega villages are radial dikes which
are sometimes cut by later concentric dikes.
The A'ofa depression on northern Ofu is bounded on the
west side by a few vertical dikes that are nearly parallel to
the fault associated with the summit collapse of the shield.
Only about six of these dikes are exposed in the cliff behind
Samoli Beach; they vary from 1/2-6 feet in thickness. One of
these is a multiple dike approximately I-foot thick which is
intruded concordantly by several small dikes 3-6 inches thick.
A piece of dike rock containing dunite xenoliths was found on
the beach here. Several dikes occur in widely scattered places
along the cliffs of northern and southeastern Ofu. Usually
they are less than 4 feet thick, vertical, and approximately
parallel to the cliff face. An ankaramite dike more than 40
feet thick crops out at the top of the cliff at Mu1iolo and
Tumu, and is probably the source of nearly horizontal ankara
mite flows on Tumu. It may be related to the collapse of the
Alofa caldera.
Intra-caldera Volcanics. Since the Si1i caldera lies
offshore, it is not known whether there were deposits that
filled it. The Alofa caldera on northern Ofu has been partly
filled with lava flows and pyroclastic rocks. The four lower
most lava flows exposed are dense, nearly horizontal ponded
92.
flows of olivine basalt, three of which are 20-25 feet thick.
These are overlain by thinner (1-7 feet) interbedded aa and
pahoehoe flows of basalt and olivine basalt. One oc the upper
most flows is a dense aa flow of hawaiite that is about 8 feet
thick and dips 25° towards the northwest. However, it rests
unconformably on another similar flow that is dipping about
6° towards the north. The upper lava apparently flowed over
a local irregularity, probably a fault scarp. Less than 300
feet away a small fault was found trending in a direction
parallel to the proposed fault. The downdropped sides of
both faults are on the north. Another thick flow lower in the
section plunged over a similar small fault scarp.
An ankaramite boulder was found on the beach 1000 feet
southeast of Tafe Stream. The block probably came from one
of the thick horizontal flows exposed near the top of the sea
cliff (about 400 feet high). This indicates that some of the
later intra-caldera flows were picrite-basalts as well as
hawaiites. Table 7 is a stratigraphic section of intra-caldera
volcanics at Sinapoto.
Just west of the mouth of Sinapoto Stream the lowermost
thick flows have ponded against a cinder cone, the highest
point of which is now about 60 feet above sea level. More
than half of the cinder cone has been eroded away by the sea,
however, and its summit was originally about 150 feet above
the present sea level. This cone must have been the source of
thin (4-8 feet thick) aa lavas of olivine basalt that flowed
93.
TABLE 7. STRATIGRAPHIC SECTION OF
INTRA-CALDERA VOLCANICS AT SINAPOTO, OFU
Top(Approximately 220-foot elevation)
Dense gray aa flow of hawaiite containing abundantmicrolites of feldspar and scattered microlitesof olivine, probably flowed over a fault scarp( dip = 25 0 NW)
Clinker
Dense medium-gray aa flow identical to that above,but dipping only 60 N
Clinker
Thickness(feet)
8
2
8
2
No outcrop~ covered due to thick soil and talus cover 15+
Nonporphyritic pahoehoe flows, 1-7 feet thick 30
Olivine basalt with olivine phenocrysts 2-3 mm in 25diameter occurring as approximately horizontalvesicular pahoehoe flows 1-15 feet thick; a seriesof thin-bedded pahoehoe flows cut by a lS-footthick flow that apparently plunged down a small faultscarp which had truncated the thinner pahoehoe flows
Nonporphyritic aa flow 4
Clinker 4
Olivine basalt occurring as vesicular pahoehoe flows 3
Clinker 4
Olivine basalt occurring as a dense horizontal flow 15
Do. 25
Clinker 1
Olivine basalt forming a dense horizontal flow, 25a small spring issues from its lower contact
Olivine basalt with abundant olivine phenocrysts 20forming a dense horizontal flow (exposed 0.3 mileeast of section)
Talus
Total thickness of section
30
221+
94~
down its northwestern flank to form the present Le1ua Point.
Post-erosional Volcanics. Nu'ute1e and Nu'usi1ae1ae
Islets, off the western coast of Dfu, are the erosional rem
nants of the Nu'u 1api11i tuff cone. This cone was originally
about 4000 feet in diameter at the present sea level and was
approximately 300 feet high. The eruption must have occurred
near sea level, after an extensive period of erosion during
which a sea cliff was cut around Dfu and D1osega. The cone is
composed entirely of reddish-yellow pa1agonitized 1apilli
tuff with accidental blocks and lapi11i of basalt, plus a few
magmatic basalt 1api11i. Indiviuda1 beds vary from less than
1 inch to more than 5 feet in thickness. The slopes of the
original cone were about 30°. No coral fragments or any other
evidence of a submarine vent was found, but the eruption may
have been submarine in part.
Along the western coast of Dfu at least two flows of
aphanitic basalt over 35 feet thick in places have flowed
down the slopes of the A'ofa shield to form Nu'upu1e Rock,
just offshore from Dfu village. At sea level, just south of
Tufu Stream, one of these hawaiite flows is at least 25 feet
thick and is overlain by a 20-foot thick flow of olivine
basalt which contains a few small dunite xenoliths. It
appears that these flows poured down old, deeply eroded valleys
and possibly out over a reef. The source was probably a cinder
cone near the summit at Tumu which has since been obliterated
by erosion. Since it appears likely that these flows occurred
95.
after the formation of deep valleys and a sea cliff around
Ofu and Olosega, they are tentatively included among the post
erosional volcanics.
Noncalcareous Sedimentary Deposits. The talus and
alluvial deposits are very similar to those described on Ta'u.
A recent-appearing landslide can be seen at Pouono, behind
the southern end of Olosega village. Several fan-shaped
slides can be seen along the base of the cliff along the coast
of southern Ofu, between Va'oto and To'aga. Talus boulders
have broken off the sea cliff and tumbled across the reef flat
in many places along the northern and southern coasts of Ofu
and Olosega. At Sili, large boulders on the reef, fresh scars
in the cliff above, and Samoan tales of people being crushed
beneath these boulders indicate that this cliff is actively
eroding at the present time.
Many of the streams On southwestern Ofu and southeastern
Olosega cut through deeply lateritized thin-bedded pahoehoe(?)
flows. Therefore, the stream beds are of a somewhat different
character in that few large boulders are found, whereas the
other streams in Manu'a are made up almost entirely of boulde~.
These stream beds instead are composed of a reddish-brown,
sticky silt. Parts of Tala'isina, Top~'a, Etemuli, and
Si'umalae Streams on southeastern Olosega and the upper por
tions of Alei, Saumolia, Togalei, Malaeti'a, Tufu, and
Matasina Streams on southwestern Ofu have such stream beds.
The upper portions of Tafe and Sinapoto Streams on northern
96.
Ofu are also devoid of numerous large basalt boulders where
they traverse the relatively level floor of the caldera at
A'ofa.
Calcareous Sedimentary Deposits. Most of the beaches are
about 40 feet wide and rarely exceed 100 feet. They have an
even foreslope of 9_10° that extends right up to the vegeta
tion. The bulk of the materials making up the beaches are
gravel, pebbles, and cobbles of coralline algae and coral.
On a few beaches some volcanic fragments of basalt, ankaramite,
tuff, olivine and augite occur. The median grain size is
usually coarse sand to gravel, as on Ta'u.
Beachrock is exposed in the intertidal zone of most
beaches on Ofu and Olosega. At Olosega village it is more
than 6 feet thick. The beachrock is composed of grains some
what smaller than those making up most of the present beach,
usually medium to coarse-sized sand. Sometimes the percentage
of volcanic grains is slightly higher than that on the present
beach. Often the loose sand adjacent to the beachrock is about
the same size as the grains comprising the beachrock, being
finer grained than the rest of the beach. This is probably
because the finer-grained beachrock is being eroded (perhaps
with the aid of some process of de-cementation), and reworked
sand is deposited nearby. Also, the beachrock may act as an
obstruction to longshore currents carrying finer sand in
suspension, causing some of the load to be deposited. A few
beaches along the northern coast of Ofu, from Sinapoto to
97.
Oneonetele, and along the northeastern coast of Olosega con
tain beachrock with subrounded basalt cobbles and occasionally
sub-angular boulders that are cemented in a calcareous sand
matrix.
Major Structures
The slopes of the explosion breccia cone centered off
shore from the southeastern coast of Ofu were about 30°.
Interbedded aa and pahoehoe flows dip 20_25° from the center
of the composite cone located about 1.3 miles west-southwest
of Sunu'itao in the cliff behind Talaga. The tuff cone at
Maga Point on southern Olosega and a tuff cone and a cinder
cone at Tauga Point on northwestern Ofu also had slopes of
about 30°. These were all subsequently buried by aa and
pahoehoe flows dipping 10_20° away from the centers of the
two shields at Alofa and northwest of Sili. The later flows
on the highlands of Ofu are almost horizontal and must have
been near the summit of the Alofa shield.
A curious break in slope forms a bench-like feature at
Papausi on southeastern Olosega. Lateritized thin flows
exposed on the surface in this area dip 5_9° NE, somewhat
gentler than the present ground surface. Their approximate
strike (N 30_40° W) can be seen.w_ith the formation of ltsteps"
1-2 feet thick that presumably represent different lateritized
thin flows. A cross section drawn from the high cliff at 8ili
through Le'ala Point (cross section B-B', Plate 1) shows that
these beds most likely do not represent the natural slope of
98.
the Sili shield. There are no indications of any vents in
this area. These flows probably did come from near the summit
of the shield, but apparently their dips have been flattened
by encountering some kind of obstruction.
It seems most likely that another cone was located off
shore from Le'ala Point, whkh lies exactly along the crest of
the Manu'a Ridge. Offshore soundings are sparse, but they do
suggest the possibility of a cone in this area. A thick aa
flow of picrite-basalt forms Le'ala Point. This flow dips
about 14° SE and probably also came from the Sili shield.
It suddenly increases in thickness downslope and seems to have
ponded against an obstruction, such as the slope of an older
cone just offshore.
Less than one-half of the A'ofa caldera is now present,
the rest having disappeared below sea level. Offshore sound
ings are not in sufficient detail to determine whether the
northern half of this depression has simply been eroded away
or whether faulting or foundering was operative. The offshore
soundings suggest that at least part of the Sili caldera is
still present. Possibly erosion prior to regional sinking of
the islands and the Manu'a Ridge was largely responsible for
the disappearance of this feature, or perhaps the caldera
never did reach above present sea level. The collapse of the
Sili shield was associated with the later(7) intrusion of a
swarm of dikes approximately parallel to the caldera boundary.
The nature and distribution of this dike complex have been
99.
described above (pp. 89-91).
The Alofa caldera is bounded by a few thin dikes in the
cliff behind Samoli beach. The intrusion of dikes associated
with the summit collapse does not appear to have been nearly
as intense as those related to the formation of the Sili
caldera. However, the exposed dikes related to the Alofa cal
dera are much nearer to the surface than those associated with
the Si1i caldera. Naturally, there would be a marked decrease
upwards in the number of dikes, as is seen in the cliffs
behind Olosega and Sili villages. Nevertheless, the estimated
dimensions of these depressions suggest that the Sili caldera
was considerably larger.
In the high cliff behind Tolaga, occasional dikes can be
seen trending approximately parallel to the cliff face.
Several large normal(?) faults can be seen in this cliff, and
also in the cliff along the northern coast of Ofu near One
onete1e. The bedding in the cones has suffered large dis
placements, but there is no surface expression of the faults
such that the direction or amount of their displacement could
be measured. These faults may have been related to the collapse
that formed the Alofa caldera.
Soundings are not in enough detail to indicate the nature
of the ocean bottom in the huge embayment between southeast
Ofu and southwest 01osega. As Daly (1924) suggested, some
type of foundering has occurred in this area, but there is no
evidence for another caldera. Perhaps gravity collapse,
100.
possibly associated with tectonic displacements at depth, is
responsible for the formation of these high cliffs. Support
for this hypothesis is much more easily obtained on Taru. If
Taru were submerged to the elevation of Afuatai (1700 feet
above sea level) all that would remain would be a parabolic
escarpment about 1400 feet high. It is curious that both of
these features occur on the southern coastlines and that in
both instances the highest gravity readings were obtained
near their central axes.
Geomorphology
Features caused by Volcanic Activity. At least seven
cones, two of which are shields, form the bulk of the volcanic
pile present above sea level today as the islands of Ofu and
Olosega. The highest point on the islands after summit
collapse of the two shields was probably only a few feet
above Piumafua, the present summit of Olosega with an eleva
tion of 2095 feet above sea level. A concentration of pyro
clastic deposits within a zone between sea level and an
elevation of about 500 feet suggest that the islands were
built at an elevation approximately the same as today.
Phreatomagmatic explosions would produce a concentration of
pyroclastic material near sea level. This implies that there
has been little sinking of these islands since their forma
tion, because the pyroclastic deposits represent some of the
oldest rocks exposed on Ofu and Olosega.
If so, then the floor of the Sili caldera may never have
101.
reached above sea leve 1. Pe rhaps the northe rn boundary was
always much lower than the southern rim, so that the ocean
completely occupied the caldera, as it does the caldera of
Krakatoa Volcano in Indonesia. Wave action would then attack
only the southern rim of the caldera. Another possibility is
that only the northern portion of the caldera sank to a lower
level, leaving the southern rim exposed above sea level.
Local erosional unconformities, indicating periods of decline
in volcanic activity, were found stratigraphically low in the
pre-caldera section at Tafalau on eastern Olosega. Some of
the lavas in this area dip as steeply as 24° and lie unCOn
formably over other lava flows. Nu'upule Rock and the dense
flows forming ridges behind Ofu village appear to be late
stage flows which poured down old stream valleys that were cut
into the thin pre-caldera flows of the shield. The islets of
Nu'usilaelae and Nu'utele are remnants of a Recent tuff cone.
Features associated with Faulting. The boundaries of the
A'ofa caldera are the most obvious features caused by fault
ing. The dike complex must have also been intruded into
fractures and faults bounding the Sili caldera. The cliff
behind Sili is therefore a fault-line scarp. The northern
portion of this cliff apparently has eroded back more rapidly
because of the absence or removal of the resistant dike com
plex in that area. If the large parabolic embayment between
southeastern Ofu and southwestern Olosega was formed by
tectonic collapse, the cliffs bounding this area also are
102.
fault-line scarps.
Streams and Valleys. The stream valleys on southwestern
Ofu and southeastern Olosega extend away from the former sum
mits of the shields in a radial drainage pattern. The thin
flows in much of these areas are highly lateritized, giving
a stream bed of sticky red-brown silt rather than large
boulders. The only other streams are those within the A'ofa
caldera, which drain the intra-caldera area and empty into
the sea along the cliffed north coast. Since these islands
are both lower in elevation and smaller in area than Ta'u,
there is considerably less rainfall and resultant runoff.
Therefore, the' streams are neither as large nor as numerous
as those on Talu. There are no perennial streams on Ofu and
Olosega. The streams flow only for a few minutes to a few
hours after a downpour. The stream valleys nowhere exceed
60 feet in depth, and usually are considerably less.
Coastal Erosion. After cessation of volcanic activity on
Ofu and Olosega an extensive sea cliff 200 to 400 feet high
was carved into the island by the sea. Behind Ofu village the
cliff is only about 80 feet high due to protection from wave
attack afforded by the Nu'u tuff cone offshore. The much
higher cliffs along the northern and southern coastlines
originated by faulting and/or foundering, but they have cer
tainly been modified by marine erosion. The sea cliffs are
usually slightly higher than those on Ta'u. This may be
103.
partly due to the fact that there has been a shorter period
of time since frequent volcanic activity on Ta'u. However,
the larger amount of more easily eroded pyroclastic deposits
near sea level on Ofu and Olosega could also account for the
slightly higher cliffs.
Landsliding and subsequent removal of the talus by wave
action are important erosional processes, especially along the
high cliffs of the northern and southern coasts. The numerous
fan-shaped landslides and fresh scars in the cliffs above
indicate that this process is operative at the present time.
One of the landslides at Pouono, the southern part of Olosega
village, is so fresh in appearance that it probably occurred
since habitation by the Samoans. At the base of the cliffs
talus deposits are always present.
Regardless of whether or not there has been a·mndslide,
individual blocks work loose from the cliff face and fall
crashing to the bottom. The reef flats in front of areas
where cliffs are high enough for falling rocks to build up
sufficient momentum are littered with large talus blocks that
have rolled across the reef. A Samoan legend relates how a
young girl was killed by a large block that rolled across the
reef at Sili, where she was fishing.
On the seaward side of Nu'utele Islet a bench is present
at the same level as that which occurs in the tuff complex at
Faleasao on Ta'u. The bench has a maximum width of 30 feet
and was probably cut during a Recent +5-foot stand of the sea.
104.
As on Talu, this bench was not found developed on any hard
lava rock. The erosion of seaward-dipping beachrock on many
beaches around Ofu and 010sega may be evidence for a previous
slightly higher stand of the sea. The beachrock usually has
approximately the same fores1ope as the present beaches, about
10°. Everywhere it is exposed, the beachrock is out ofequi1~
brium with the present sea level and is being eroded away.
Coastal Deposition. The constructional bench 13-15 feet
above sea level that was found on Talu is also well developed
on Ofu and 01osega, but it is nearly 20 feet above sea level
in some places. The villages of Ofu, 01osega, and Si1i are
all built on these sandy areas. Other examples of this con
structional bench are found at Valoto, Tolaga and Mafafa on
Ofu and at Oge on 01osega. The bench is about 900 feet wide
at Valoto and over 1000 feet wide at 010sega village, but the
average width is about 300 feet. Figure 5 shows profiles
made across these benches at various places around the islands.
Usually a 9_10° fores1ope rises steadily to the crest of the
berm 15-20 feet above sea level, but sometimes there is an
intermediate berm. At Valoto and 010sega swamps have formed
between the berm crest and the talus slope at the base of the
cliff. Although storm waves during hurricanes have obliterated
a trail across one of these benches on eastern Ta'u, many of
the areas with constructional benches probably are well-enough
protected that even storm waves can not come over them. None
of the villages in Manula has been known to have been comp1etay
FIGUP£ 6--System Ne-Fo-Si02 at (a) 1050° C. and1 bar, and (b) 1250° C. and 33,000 bars.Heavy lines indicate stable joins. Liquidsin horizontally ruled area behave in anopposite way to those in cross-hatched area.Liquids in diagonally ruled area trend thesame way at both pressures. (from Yoder andTilley, 1962)
134.
contain modal hypersthene or its chemical equivalent, pigeon
ite, in the groundmass, whereas the alkali basalts never con
tain these minerals, either n"ormatively or modally. Later
Kuno (1959) noted the presence of interstitial alkali feldspar
in the alkali basalts and also claimed that the interstitial
glass of a tholeiitic rock should contain a silica mineral.
Macdonald and Katsura (1964) stated that in Hawaiian rocks
"the presence of titanian augite, of true groundmass olivine,
and of interstitial alkali feldspar appear to be positive
indications of the alkalic nature of the rock."
The reason for needing some basis of classification other
than a mineralogical one is that often there are not any
mineralogical criteria present in the rock, especially in
glassy volcanic rocks. The reaction relationship between
olivine and pigeonite or hypersthene is the best criterion,
but this relationship is rarely seen to occur in silica
saturated rocks of oceanic lavas. The presence of pigeonite
in the groundmass has been considered to be a criterion for
a tholeiitic basalt. However, as illustrated in the Samoan
lavas, the clinopyroxenes with a low 2V can actually be sub
calcic augites, being high in titanium, ferric iron or some
other element that lower the 2V of the mineral. The only
apparently reliable mineralogical criteria are the presence
of titan-augite or interstitial alkali feldspar for the
determination of a rock belonging to the alkali basalt suite.
The majority of alkali basalts do not contain these criteria,
135.
at least not so that they can readily be recognized under the
petrographic microscope.
Therefore, many petrologists have hoped to find reliable
chemical criteria for separating these two suites. The strictly
chemical scheme of classification devised by Cross, Iddings,
Pirsson and Washington is accepted by some petrologists today
as the best criterion for determining whether a volcanic rock
belongs to the tholeiitic or the alkaline basalt suite. This
normative mineral classification involves computations of
simple theoretical minerals from standard oxide chemical
analyses. All rocks containing normative hypersthene, even
though it does not appear in the mode, are considered to be
silica-saturated and therefore are tholeiitic. Those rocks
containing normative nepheline are considered to be under
saturated in silica and therefore belong to the alkali basalt
group.
Yoder and Tilley (1962) divided the basalts into five
groups according to their norm components:
(1) Tholeiite (oversaturated)--normative quartz and
hypersthene
(2) Tholeiite (saturated)--normative hypersthene
(3) Olivine tholeiite (undersaturated)--normative
hypersthene and olivine
(4) Olivine basalt--normative olivine
(5) Alkali basalt--normative nepheline and olivine
Macdonald and Katsura (1964) have pointed out the
136.
inadvisability of using the term olivine basalt in this new
classification. Olivine basalt is a widely used field term
that is applied to the majority of basaltic rocks. To re
define it to cover such a limited scope of rocks based on a
chemical rather than a mineralogical definition would only
lead to hopeless confusion. Some other name should be
selected, preserving the term "olivine basalt" in its present
sense--a basalt containing more than 5% olivine phenocrysts
(Macdonald, 1949).
Another objection to using the normative method is that
a rock containing a high amount of ferric iron relative to
ferrous iron will often contain normative hypersthene, the
'most critical mineral for the determination of a tholeiite.
This is not necessarily because of any inherent characteristic
of the magma but may be only due to oxidation of a particular
flow. According to the rules of calculation, the amount of
ferric iron oxide is matched by an equal amount of ferrous
iron oxide to produce normative hematite. Therefore, the
amount of ferric iron oxide depletes the total iron oxide
content by double that value, leaving a much lesser amount
of iron oxide to be added with magnesia and silica to form
olivine. The excess silica must then be taken up by the
formation of hypersthene.
Results of chemical analyses have often been plotted on
a variation diagram, using per cent silica as abscissa and
per cent alkalis (Na2
0 + K2
0) as ordinate. For the Hawaiian
137.
lavas plotted on such a diagram it has been shown that the
tholeiitic rocks can be separated empirically from the alkali
rocks by drawing a diagonal line through an intermediate
region where the least number of analyses are plotted
(Macdonald and Katsura, 1964, pp. 86-87). For some reason,
possibly because it approximately represents the critical
plane of silica undersaturation (Yoder and Tilley, 1962), no
rocks that are of definitely tholeiitic characteristics have
plotted above this boundary line. Conversely, all rocks of
an alkalic nature fall above the line, with the occasional
exception of a few ankaramites, which may fall below the line,
presumably because of a depletion in the salic fraction due
to corresponding enrichment in olivine and augite by crystal
settling.
There are now at least three definitions for a tholeiite
--the original mineralogical definition, the definition based
on the presence of normative hypersthene, and the definition
based on the alkali:silica ratio as related to an empirical
boundary line. One analysis of a specimen collected in June
1960 from the liquid below the crust of the lava lake which
was formed during the eruption of November l4-December 20,
1959 in Kilauea Iki crater plots just above this boundary
line, in the alkali basalt field. The rock contains micro
phenocrysts of augite with a large 2V, unresorbed olivine
phenocrysts and a high titania content. All other specimens
collected from the same lava lake appear to be tholeiites both
138.
mineralogically and chemically. However, this sample contains
one of the largest amounts of normative hypersthene (18.56%)
and one of the smallest amounts of normative olivine (6.16%)
of any of the analyzed Kilauea Iki specimens. The Fe 203
content in this sample is only 1.36% as compared with 12.99%
FeO. Therefore, the high value for normative hypersthene
cannot be attributed to a large amount of ferric iron.
Macdonald and Katsura (1961) suggest that, if this sample is
truly alkalic, alkali basalts in general can be derived from
undersaturated tholeiitic magmas. By the mineralogical
definition this rock is an alkali basalt; by the calculation
of normative minerals it is definitely tholeiitic; and accord
ing to the a1ka1i:si1ica ratio it is alkalic, although tran
sitiona1~ .
Powers (personal communication, 1965) points out that
there is considerable disagreement between the analyses of
Kilauea Iki lavas as reported by Macdonald and Katsura (1961)
and the analyses from the same eruption as reported by the
laboratory of the u.s. Geological Survey at Denver. So, not
only is there a discrepancy in the definition of the rock,
but the results of the chemical analyses themselves are in
doubt. Similarly, Macdonald and Katsura (1964) report that
32 analyzed Hawaiian rocks belong to the alkalic suite by
mineralogical definitions. Yet, only thirteen of these
samples do not contain normative hypersthene.
Recently eight samples representing the major types of
139.
volcanic rocks in Hawaii were analyzed by two major rock
analysis laboratories. Powdered samples from the same rock
were split and one was sent to each laboratory. Table 8
shows the results of these analyses. A glance at the figures
reveals large discrepancies in the results. The point is not
which laboratory is right or wrong, but it is a question of
how much emphasis should be placed on chemical analyses as a
method for determining whether a rock belongs to the tho1eiftic
or alkali basalt suite. The present methods of conducting
laboratory analyses simply are not consistent enough to regard
with any confidence, particularly when transitional rocks are
involved.
Actually the discrepancies in these analyses may not be
as critical~ they appear to be. Macdonald (personal com
munication, 1966) calculated the normative minerals for both
sets of chemical analyses (Table 8). None of these rocks are
affected in their classification by even such considerable
differences in the chemical analyses. However, transitional
rocks could be affected by differences of this magnitude. A
change in Na 20 content from 2.00% to 1.82% in a sample (#156)
from Manu'a changes the rock from an alkali basalt (contain
ing normative nepheline) to a tholeiite (containing normative
hypersthene).
It seems that the method of plotting alkalis vs. silica
used by Macdonald and Katsura (1964) is more reliable for
determining whether a rock is tholeiitic or alkalic than
140.
TABLE 8. COMPARISON OF CHEMICAL ANALYSESON THE SAME HAWAIIAN LAVAS
(Data supplied through courtesy of Dr. G. A. Macdonald)
SampleNo. HIGS 1 HIGS 2 HIGS 3 HIGS 4
.13 .37
.2 2 .26
.00
.00
.03
.11 .68
.02 .05
.01
.11
.14
.08 .31
.03 .12
.02
.01
.03
.12 .31
.02 .11
.01
.01
.03
La b...:...__U=....:S::....:G:..:S_..:..J:....:.A...::..CR:..;_I,-,--_U-,S-,G-,S_....:.J_A....:C:....:.R...::..I_--.:..U..:..S...::..G...::..S_J::....:A_C::....;.R....:.I__U_S....;.G_S_...::..J_A_C_RI
ne 0.71 oJ( 1. 42 1.14 ** Contains normative nepheline rather than normative hyper-sthene when FeO/Fe203 ratio adjusted to 3:1 keeping constantwt. % of total iron oxides.
146.
TABLE 9B. EXPLANATION OF CHEMICALLY
ANALYZED SPECIMENS FROM MANU'A
11 from Tunoa depression
A vesicular alkali olivine basalt with olivine phenocrysts approximately .5 mm in diameter, some of which arealtered to iddingsite (finely-divided goethite?). Labradorite (An 60 ) microphenocrysts occur in an intergranular groundmass of titanaugite, sub-calcic ferroaugite(?) (a few grainshave a +2V = 10°), olivine, iddingsite, andesine ( An 48),ilmenite, magnetite, apatite, and interstitial alkali feldspar.
17 from Fitiiuta, at Tiafala Point
Alkali nonporphyritic olivine basalt containing dunitexenoliths up to 2 cm long and a few microphenocrysts about2 mm across of olivine that has been altered to iddingsite.The intergranular groundmass is composed of labradorite (An57)with augite, olivine coated with iddingsite, magnetite, andinterstitial alkali feldspar. The sample was collected nearsea level, and the amygdules are composed of calcite thatprobably has been deposited in the vesicles after the lavasolidified.
44 from a post-caldera flow on the northwest flank of Ta'ushield, approximately 2500 feet south of Fogapoa
Vesicular alkali basalt, transitional to hawaiite, containing a few augite phenocrysts about 2 mm across. Microphenocrysts of olivine and augite up to .3 mm across occur,sometimes as glomeroporphyritic clots. The groundmass iscomposed of olivine, augite, magnetite, and ilmenite. Thetexture is intergranular.
62 from a post-caldera flow ponded within Lepu'e Crater
Vesicular hawaiite, transitional to alkali olivine basa~,
containing olivine phenocrysts about 1 mm in diameter. Thevery fine-grained groundmass is composed of andesine(?),olivine, augite, ilmenite, magnetite, and apatite. Theseminerals form an intergranular texture. A 1abradorite(?)xenolith about .5 mm across has been partly resorbed.
89 from intra-caldera flow on southern Ta'u
Vesicular alkali picrite-basa1t (oceanite type) containing abundant olivine phenocrysts up to 6 mm in diameter anda few augite phenocrysts up to 5 mm across in an intersertal
147.
groundmass of labradorite, ilmenite, a few magnetite grains,apatite, cryptocrystalline material and glass.
93 from a dike at Laufuti, southern Ta'u
Alkali olivine basalt comprised mainly of labradorite(An55) and brown augite with some' olivine, ilmenite, andmagnetite to form an intergranular to diabasic texture.
97 from intra-caldera flow at Papa loa loa Point on southernTa'u
Nonporphyritic hawaiite transitional to alkali olivinebasalt which contains a few olivine microphenocrysts about.3 mm in diameter and labradorite microphenocrysts about1 mm long. A hyaloophitic texture is formed by labradorite(An
52) microlites that are approximately parallel, ilmenite,
magnetite, and black glass.
107 from pre-caldera flow exposed in Avatele stream bed onnorthern Ta'u
Hawaiite composed mainly of andesine (An4S) and augitein a ratio of about 3:1. A few microphenocrysts about .5 mmacross of olivine and augite occur. Usually the olivinegrains are partly altered to iddingsite. The groundmassconsists of feldspar, colorless clinopyroxene, olivine,abundant ilmenite, some magnetite, and brown glass to forman intergranu1ar to intersertal texture.
110 from a pre-caldera lava flow exposed in ~o'auli stream bed
Fe1dspar-phyric alkali basalt,transitional to hawaiite,composed of abundant laths of labrado~ite (An65) up to 3 cmlong in an intersertal groundmass composed of ilmenite,andesine(?) microlites, olivine, apatite, and brown glass.
117 from a pre-caldera (or possibly intra-caldera) flow atAfuatai, southern Ta'u
Vesicular alkali olivine basalt, approaching a picritebasalt, which contains scattered olivine phenocrysts up to2 mm in diameter. A few glomeroporphyritic clots composedof olivine and augite grains up to 1 mm in diameter also occurin the intersertal groundmass of labradorite(?) micro1ites,abundant ilmenite, magnetite, apatite, and glass.
129 from Luatele shield
Vesicular picrite-basalt containing scattered olivinemicrophenocrysts up to 1 mm across. Some of these grainshave a birefringence in their outer zones, probably due to
148.
iron enrichment. A few glomerocrysts of olivine occur.The interserta1 groundmass is composed mainly of glass andcryptocrystalline material with labradorite(?) micro1itesand ilmenite.
131 from a post-caldera flow erupted from Salumaae Crater
Vesicular alkali picrite-basa1t (oceanite type) containing olivine phenocrysts up to 3 mm across. A few augitemicrophenocrysts approximately .2 mm in diameter are present.The intergranu1ar groundmass is composed of labradoritemicro1ites, ilmenite, augite, cryptocrystalline material,and glass.
136 from post-erosional flow forming Moso Point on westernTalu
Alkali olivine basalt, approaching a hawaiite, composedmainly of labradorite (AnS2) microlites and cryptocrystallinematerial. Olivine grains approximately .3 mm across formglomerocrysts up to 2 mm in diameter. A few augite phenocrysts up to 4 mm long occur. Sub-calcic augite microphenocrysts about .3 mm across have a low +2V and probably arehigh in titanium, but show no purplish tint. Magnetite,ilmenite, olivine, augite, and apatite form an intergranu1argroundmass texture.
156 from a flow ponded within the A'ofa caldera on northernOfu
Alkali picrite-basa1t (ankaramite type) composed ofabundant olivine and augite phenocrysts up to 4 mm across.The former are thinly coated with rims of iddingsite about.02 mm wide. A glomeroporphyritic clot of labradorite microphenocrysts approximately 1 mm long is present. Abundantmagnetite, usually less than .005 mm, and some ilmenite occur.A few euhedra1 opaque grains of magnetite are up to 1 mm indiameter. The remainder of the groundmass is composed of1a~radorite (An60) micro1ites, augite, olivine, cryptocrystalline material and glass. The texture is intergranu1ar.
196 from a pre-caldera flow of the Si1i shield, just eastof Si1i
Vesicular nonporphyritic alkali olivine basalt containing olivine microphenocrysts approximately 1 m~ long whichhave been altered to iddingsite in an intergranu1ar to hya10ophitic groundmass composed mainly of labradorite (AnS4) andtitanaugite. Ilmenite, some magnetite, apatite, and blackglass also occur.
149.
207 from post-caldera(?) flow of the Sili shield at Le'alaPoint on southeastern Olosega
Alkali picrite-basalt (oceanite type) containing bothabundant olivine phenocrysts up to 4 mm in diameter andolivine in the groundmass. The olivine has been partlyaltered to iddingsite. The remaind~r of the rock containslabradorite microlites, ilmenite, magnetite (some grains upto .3 mm), cryptocrystalline material and glass to form anintersertal to intergranular texture.
210 from a thick flow ponded within Maga Crater on thesouthern tip of Olosega
Hawaiite, transitional to olivine basalt, that containstabular olivine microphenocrysts approximately .5 mm long,which are considerably altered to iddingsite. Plagioclasemicro lites range from An52 for larger grains to An46 forsmaller grains. The remainder of the groundmass is composedmainly of magnetite (grains up to .2 mm) and ilmenite witha few zoned olivine grains up to .3 mm apatite, and interstitial alkali feldspar. The texture is intersertal tointergranular.
22lA from pre-caldera flow exposed in 'Ao'auli stream bed
Vesicular olivine picrite-basalt containing olivinephenocrysts up to 3 mm across. A few labradorite (An63)phenocrysts up to 2 mm long and very few augite phenocrystsabout 2 mm long are present. The intersertal groundmasscontains abundant magnetite, fine-grained feldspar, cryptocrystalline material and dark glass.
226 from post-caldera flow on southeast flank of Ta'u shieldnear Lalomaota
Vesicular hawaiite transitional to alkali basalt composedmainly of labradorite (An50) micro lites with a sub-parallelorientation and a few larger laths of labradorite (An55).Olivine microphenocrysts about .5 mm across and microphenoc~ysts of clinopyroxene (titanaugite?), some of which have alow +2V, are present. Abundant ilmenite, apatite, cryptocrystalline material, and brown glass form an intersertaltexture.
229 from intra-caldera flow on southern Ta'u
Hawaiite composed predominantly of andesine (An45) microlites, which have a sub-parallel orientation. A few olivinemicrophenocrysts (elongated parallel to the b-axis) are about.5 mm long and are partly altered to iddingsite. Abundant
,.u
150.
grains (approximately .01 mm across) of ilmenite and mag-
netite lie in the intergranu1ar groundmass, the remainder
of which is composed of very small amounts of biotite,
apatite, and interstitial alkali feldspar.
151.
1-2% higher than that of the corresponding Hawaiian lavas.
The rock with the lowest Ti02 content is the same one (#131)
that has the lowest alkali content. However, this sample
contains both normative olivine and abundant modal olivine
in the groundmass. Augite occurs only in the groundmass,
and the total feldspar comprises less than 30% of the rock.
It should therefore be defined as an alkali picrite-basalt.
Most of the titanium in the rocks of Manula apparently is
taken up by ilmenite and titanaugite. It is also likely
that titanium occurs in the sub-calcic augi~es with low +2V rs.
All of the analyzed spe6imens appear to belong. to the alkalic
suite according to their mineralogical characteristics, but
as discussed above (pp. 134-135) the reliable mineralogical
criter~. ~re very few, particularly when dealihg with fine
grained transitional rocks.
Normative minerals were calculated for each of the
chemical analyses of the rocks of Manura. The standard
Cross-Iddings-Pirsson-Washington method as outlined by
Johannsen (1938, vol. 1) was used in these computations. The
value in using this scheme is that it facilitates comparison
with rocks from other areas (e.g. Hawaiian rocks).
All of the samples from Manula except one (#11) contain
normative olivine. Ten of the samples contain normative
nepheline, and the other ten contain normative hypersthene.
Only four contain more than 1% normative nepheline, the
largest amount being 1.56% (#93). No modal nepheline was
152.
found to occur.
Of the ten samples containing normative hypersthene,
only two had FeO/Fe2
03
ratios greater than 3:1. However,
only four of the samples containing normative nepheline had
FeO/Fe2
03
ratios greater than 3:1 eithert Nevertheless, the
FeO/Fe2
03
ratios for eight samples containing normative hypers
thene was adjusted so that the FeO/Fe 2 03
ratio was 3:1, keep
ing the weight per cent of the total iron oxides constant.
These samp1ffiwere then recalculated for normative mineral
content. Three of the samples (#110, #207, #229) then con
tained normative nepheline rather than normative hypersthene,
leaving a total of seven samples with normative hypersthene
even though their FeO/Fe 20 3 ratios were all greater than 3:1.
These would be tholeiitic rocks according to the definition
of Yoder and Tilley (1962).
There is no correlation between the pre-caldera and
later-stage lavas in regard to their normative composition.
Some of the lavas containing normative hypersthene are post
erosional lavas, whereas others with normative hypersthene
are among the oldest rocks exposed on the islands. Also,
there is no correlation between the Ofu-01osega volcanic pile
and that of Ta'u, except that all four samples from Ofu and
010sega contain normative nepheline and less silica than any
of the rocks from Ta'u. If the calculations are considered
to be valid, the only conclusions to be drawn are that the
rocks from Ta'u Island are made up of interbedded tholeiites
153.
and alkali basalts and that no tholeiites were found to occur
on Ofu and Olosega Islands. The inadequacies of calculating
normative minerals has already been pointed out. The alkali
silica diagram with the empirical tholeiitic-alkalic boundary
line seems to be most reliable except for picrite-basalts.
What now appears to be the most serious problem is the relia
bility of chemical analyses by the present methods. Until
chemical analyses can be reproduced within a reasonable amount
of accuracy by different laboratories, no acceptable method
is of much value, at least in classifying the transitional
rocks. Perhaps the real answer to this problem is that the
two suites are transitional in nature and therefore cannot be
separated by any method.
All but one of the Samoan specimens analyzed plot above
the empirical boundary line on the alkali:silica diagram
(Figure 7) and therefore belong to the alkali basalt group.
The single "tholeiitic" basalt was a Recent(?) post-caldera
flow of picrite-basalt from Sa'umane Crater (sample #131).
This picrite-basalt, which contains both normative hypers
thene and olivine, is composed of olivine phenocrysts in a
groundmass of olivine, augite, and ilmenite. It is this
same sample that also has the lowest TiOZ
content. Because
the rock is undersaturated in silica, it is probably alkalic
rather than tholeiitic. It seems that the somewhat high
percentage of mafic minerals could have caused a decrease in
alkalies relative to silica. Therefore, all normative olivine
I ----------.------.
•
b'-
5
ol'l /", i_
~. "
i-
SRS 5 ,-~
~
2.
1~iI
..
....
.......- --- -
..
••~
.. .• . ......----......- __ ..... '- :----_X I
_- x.-------
.- - .....-
"
---.-------......-
.-----.....--,-
FIGURE 7. Alkali: _silica diagram of rocks from Manu I a J _ sho''ling the .boundary (diagonaldashed line) between the tholeiitic and alkalic fields. The picrite-basalt from Ta'uwhich plots as tholeiite is shown by an X. The same rock after subtraction of allnormative olivine is shown by an XI.
was subtracted from the rock, and the recalculated figures were
plotted on the diagram (Figure 7). The sample still falls
below the boundary line by about the same amount, although,
of course, it was displaced to the right and upwards.
The samples were plotted on another type of variation
diagram, the AFM diagram (Figure 8), in order to determine
the nature of any trends in magmatic differentiation which
could have been important in producing the various rock types
found to occur in Manula. In this triangular plot the coord~
nates are the alkalies (NaZO + KZO), total iron (FeO + Fe Z0 3 )
expressed as FeO and magnesia (MgO). The diagram clearly
illustrates the importance of olivine in the differentiation
of basaltic rocks; this mineral falls directly on the main
trend of the lavas. The occurrence of hawaiite and picrite
basalt indicates that a primitive olivine basalt magma was
just beginning to undergo differentiation by crystal settling
within the magma chamber. Apparently the magma did not bec~me
acidic enough to produce the trachytic end members of the
series, which most likely would have contained normative and
possibly some modal quartz.
The magma from which the rocks of nearby Tutuila were
derived was apparently similar to that of Manula. The mafic
rock samples fall in about the same area on the diagram as
the corresponding lavas from Manula. It seems probable that,
had there been sufficient time and/or other conditions to
__ allow complete differentiation within the magma chamber,
156.----------------_._--_._-_._---_._-----
\.
\I
oliviuG r;i7\'.
.::.L :..._.._._'I _.__ __ .\.1.-._. __.'.l\/
/ \,\\
~
\'\\
i\
FIGURE 8. A-F-M diagr28 shOWing the ~elative distriDucio:1 of hU\;!c1iite C-cI'i~~nglcs)~ picrito-bc:s:J.lt (opencirclos)? ~nd ~lkQli olivine o~sQlt (solid circ18s)';:''''-'0'(''' 1\Ii"'n'1l''''' fi. - ;-0·1-'11 "'11.~~1 i Co' (I·T .-, 0 .'. T~ 0) T;' - ;-0;-',..,1-- .:.u ......_.. J. ....... ljlo ... - - u v ...... _ <..;." _\.u__ .v -''--? . £-2 ~ - - u uc... _
i:c'on 3.8 FeO? and. IVl =1.'lgO. Also incIuded are 2..nal;y"s0sf'Y'O"71 . 1"'1':-'" i i r' (c''-'o C' ~,~ "" )\ 'C'V~'r'" r:;-.=. "-1" ':'Y1Q' 1 -i 'n r~ c o·f' u -.:1"1'" i l' ,.., y)__ ll ... .LL.:-LJ....c.___ _ -J;;) ...... ~ ., c:. c: f.. ... o ...... U t.::_ _-1-J.J.vv ..L. fi~\j ~_ t;..t.J.':'
tholeiite (solid line) and al.~~lic (dashed line) suites?" and typical olivine COL'lposition (sq~)are).
i ,L- . __ _._ .. ._ _ _ __.. . --1
157.
acidic rocks similar to those on Tutuila would have occurred
in Manu'a. Therefore these acidic rocks from Tutiila are
also included in the AFM diagram. This trend is very similar
to the trend for Hawaiian lavas of the alkali basalt series.
Iron and titanium, in general, decrease slightly with increas
ing silica content for all types of basalts and hawaiites.
However, the iron in the mugearites and trachytes from
Tutuila~ as in those from Hawaii, begins to decrease rapidly
with an increase in silica and alkalies. Perhaps this
"turnabout" in the trend is caused by an increase in the
partial pressure of oxygen, which would yield silica-enriched
differentiates rather than iron-enriched differentiates
(Osborn, 1962). The fact that this reversal in trend always
occurs at about the same point in all suites, i.e. when
mugearites are beginning to form, weakens the latter sugges
tion. It seems that the partial oxygen pressure of all magmas
yielding mugearite as differentiates is suddenly increased for
some unknown reason. Why should the partial oxygen pressure
just happen to increase at this particular point of magmatic
differentiation? Possibly some other physicochemical factors
related to the complex process of magmatic differentiation
(e.g. the effect of volatiles) are also important in this
change in differentiation trend.
REFERENCES CITED
BAILEY, K. B., CLOUGH, C~. TI., WRIGHT, loll. B., RICHEY, Jo E.,and WILSON, Gr. V~., 1924. Tertiary and post-tertiarygeology of Mull, Lock Aline, and Oban. Mem. Geo1.Survey Scotland, 445 pp.
BOWEN, N'. L., 1928. The Evolution of the Igneous Rocks.Princeton Univ. Press, Princeton, 322 pp.
BOYD, F. R'., and ENGLAND, J. V'., 1961. Melting of silicates.Carnegie Inst. Wash. Yrbk, 60, pp. 113-25.
CHAYES, FELIX, 1964. A petrographic distinction betweenCenozoic volcanics in and around the open ocean. J.Geoph. Res., vol. 69, pp. 1573-88.
DALY, R. A'., 1924. The geology of American Samoa. CarnegieInst. Wash. Pub1. no. 340, pp. 93-152.
DANA, Ji. D., 1849. Records of ~. Exploring Expeditionduring the years 1838-1842, under the command ~ CharlesWilkes, U.S'.NI;', v. 10 (Geology), National Archives,Washington, D. C., pp. 307-66.
FOX, J., and CUMBERLAND, r. B., 1962. Western Samoa.Whitcomb and Tombs, Ltd., Ghristchurch, 337 pp.
FRIEDLANDER, I., 1910. Beitrage zun der Samoa inse1in.K. Bayer. Akad. Wiss., Math-phys. K1., Munich Bd. 27,pp.358-69.
GREEN, D. Hio, and RINGWOOD, N. E., 1964. Fractionation ofbasalt magmas at high pressures. Nature, v. 201,pp • 127 6- 79 •
JOHANNSEN, ALBERT, 1938. ~ Descriptive Petrography of theIgneous Rocks. v. 1, Univ. of Chicago Press, Chicago,267 pp.
LEAR, D., and WOOD, B. L., 1959. The geology and hydrologyof West~rn Samoa. New Zealand Geo1. Survey, Bull.,no. 63, 92 pp.
KENNEDY, W. Q., 1933. Trends of differentiation in basalticmagmas. Amer. Jour. Sci., 5th series, v. 25, pp. 239-56.
KENNEDY, W. Q., and ANDERSON, E. M., 1938. Crustal layers andthe origin of magmas. Bull. Vo1cano1ogique, series II,tome III, pp. 23-82.
159.
KUNO, H., 1959. Origin of cenozoic petrographic provincesof Japan and surrounding areas. Bull. Vu1cano1ogique,series 2, v. 20, pp. 37-76.
KUNO, H., 1960. High alumina basalt. J. Pet., v. 1,pp. 121-45.
KUNO, H., YAMASAKI, K., SIDA, C., and NAGASHIMA, K., 1957.Differentiation of Hawaiian magmas. Japan J. Geol.Geog., v. 28, pp. 179-218.
MACDONALD, G., A., 1944. Petrography of the Samoan Islands.Bull. Geol. Soc. Amer., v. 55, pp. 1333-61.
MACDONALD, G. A., 1949. Hawaiian petrographic province.Bull. Geol. Soc. Amer., v. 60, pp. 1541-96.
MACDONALD, G. A., DAVIS, D. A., and COX, D. C., 1960. Geologyand groundwater resources of Kauai. Hawaii Div. Hydrog.,Bull. 13, 212 pp.
MACDONALD, G. A., and KATSURA, T., 1961. Variations in thelava of the 1959 eruption of Kilauea Iki. Pac. Sci.,vol. 15, pp. 358-69.
MACHESKY, L. F., 1965. Gravity relations in American Samoaand the Society Islands. Pac. Sci., vol. 19, pp. 367-73.
MURATA, K. J., 1960. A new method of plotting chemicalanalyses of basaltic rocks. Amer. J. Sci., v. 258-A,pp. 247-52.
OSBORN, E. F., 1962. Reaction series for subalkaline igneousrocks based on different oxygen pressure relationships.Ame r • Min., v. 47, p p • 2 11- 2 6 •
POWERS, H. A., 1935. Differentiation of Hawaiian lavas.Ame r. l. Sci., v. 255, Pp. 241- 53 •
POWERS, H. A., 1955. Composition and origin of basalticmagma of the Hawaiian Islands. Geoch. et Cosmoch. Acta.,v. 7, pp. 77-107.
RUHE, R. V., WILLIAM, J. M., and HILL, E. I., 1965. Shorelines and submarine shelves, Oahu, Hawaii~ ~ Geo1.,v. 73, pp. 485-97.
SHARP, ANDREW, 1960. Discovery of the Pacific Islands. OxfordUniv. Press, 259 pp.
Soil Survey of Hawaii, 1955, ~. Dept. Agr. Soil Survey Bull.,no. 25 (1939 series), pp. 73-9.
160.
STEARNS, H. T., 1944. Geology of the Samoan Islands. Bull.Geol. Soc. Amer., v. 55, pp. 1279-1332.
STEARNS, H. T., and MACDONALD, G. A., 1946. Geology andgroundwater resources of Hawaii. Hawaii Div. Hydrog.,Bull. 7, 344 pp.
STEARNS, H. T., and MACDONALD, G. A., 1946. Geology andgroundwater resources of the island of Hawaii. HawaiiDiv. Hydrog., Bull. 9, 363 pp.
TILLEY, C. E., 1950. Some aspects of magmatic evolution.Quart. ~. Geo1. Soc. London, v. 106, pp. 37-61.
WILLIAMS, JOHN, '1907. Missionary Enterprises in the SouthPacific. Presbyterian board of publication and Sabbathschool work, Philadelphia. 416 pp.
YODER, H. S., Jr., and TILLEY, C. R., 1962. Origin of basaltmagmas. J. Pet., v. 3, pp. 342-532.
~ft) of palagonitized vitric'ystal lapilli tuff, breccia,ld occasional horizontal lava.ows from at least three main,nes centered at Faleasao, Tota,Ld Fa' asemene Coves.,
'posits. In areas behind constructional benches m~rshE~
(-im) sometin:es occ\" ., .
, PalagonitizeJ hpilli tuff (~,n)
of Nu'u tuff cone, whi~h form Nu'utele and Nu'usilaelae Islets onwest-ern O'fu. A few qe('ent(?) i.awaiite flows (Qol) may h~ve poured downformer deep1.v eroded stream valleyson western Dfu.
Pre-caldera volcanics (Ttep)consi~t of lava flows of olivinebasalt, picrite-basalt, basalt,feldspar-phyric basalt, and hawaiite with occasional beds oftuff. Most of these rocks havebeen mantled by post-calderaflows (Qtel) of olivine basalt,picrite-basalt, basalt, and hawaiite which were erupted fromcones on the flanks of the shield.Nhere possible, late cinder cones(Qelc) and their associated lavaflows (~ell) have been mapped separately. Usually the pre-calderalavas and the post-caldera lavascannot be separated and are mappedtogether as post-caldera volcanics(~tel) except in a few areas whereit appears likely that pre-calderavolcanics (Ttep) are exposed. Theintra-c~ljera volcanics (Qtcv)consist mainly of dense flows ofolivine basalt, picrite-nasalt,basalt~ and hawaiite with numerousthin beds of vitric-crystal ash.An intra-caldera cinder cone (Qtcc)and its associated lava flows (Qtcl)of basalt are mapped separately.Other Recent flows (Qtcl) of basaltoccur. A dike complex intrudes thepre-c81dera(?) (Ttep) volcanics.
TA'UVOLCANICS
Thin lava flows (Qle)· ofolivine basalt and picritebasalt h3V~ built the .Luateleshield on the northeasternportion of the island.".The .depression w~s partly filledwith Donded olivine basaltlall8s' (Qlc).
TUNOA
Lava flows (Qnel) 0:' basaltand olivine basalt hav~ builtthe Tunoa shield on the northwestern portion of the island.The depression formed by collapse of this shield was part3y filled with volcanic deposits (Qncv) of red vitriccrystal ash~ lapilli tuff, andolivine basalt lavas, some ofwhich contain dunite xenoliths.Associated cinder cones (Qncc)are mapped where possible. Onelava flow (Qncl) of olivinebasalt plunged over the cliffsouth of Ta1u village.
Iv IV'" r-\L- 'VI"V"'I~' "'., ••••••
.UATELE
OFU RECENT VOLCANICS
..' ".
UNCONFORMITY
struc~lona~ oeocneB marlB~l~le~,,~---
(Qm) sometimes occ~~.
EROSIONAL
. Palagonitizel lapilli tuff (Qn)of ~u'u tuff cone, which form Nu'utele dhd Nu'usilaelae Islets onw~stiern o"fu. A few 'iec.en t (?) hawaiite flows (Qol) may have poured downformer deeply eroded stream valleys
~ on western Ofu.
'fTI
16''I:::0rr1O-mz-i
OFU - OLOSEGA VOLCANIC S.ERIES
-0r-
140 0IS'S ('")
rr1ZIT1
The lower member of the extra-caldera volcanicsconsists of at least five cones aligned along theregional rift zone--a cinder cone (Tetc) at TaugaPoint, a tuf! cone (Test) at the west end of Samo'iwith an associated lava flow (Tesl) ponded withinits crater, a composite cone (Tet) at Talaga withassociated red ash (Teta), a breccia cone (Tefb)with nn associated plug (Tefi) at Fatuaga ~oint,and a tuff cone at Maga Point. The upper memberof the extra-caldera volcanics consists of twocoalescinp; shields, the "A' ofa shield (Tea) andthe Sill shield (Tes), wh:ch are comprised mainlyof olivine basalt, basal4i, picri.te-basalt, andhawaiite with intercalated beds of ash, tuff, andbreccia. T~ese shields and the Fatuaga brecciacone are intruded by numerous dikes. The intracaldera volcanics consist of thick ponded flows(Teal) of oliYine basalt, hawaiite, and ankara'mite. A. buried cinder cone (Toac) is also present.
1690 421
W-, ....._---..----_......---------t "~
... '....10 ••',·' -,
~~?~.~~; , i- ..,/
--..
u- o
u---o
..", ...
.... . .
.1
EXPLANATION (
Normal fault showing de
Eroded fault scarp showapproximately located.
Buried fault scarp, apl
Contact, definitely lo(
Contact, approximately
Dip and strike of beds.
Horizontal beds.
Dike.
Intermittent stream.
Coral reef
\J OF SYMBOLS
ng downthrown CD) side.
showing downthrown (D) side,ted.
), approximately located •
.y located.
Ltely located.
beds.
m.
12°.- .. "
Ile::oz
W::Je::I-
APPROXIMATE MEANDECLINATION, DECEMBER 1963.
____________________~i'
, \
• •• • • •
OF TH E
MANU 'A. ISLAN
AMERICAN SAMCseA L E"-. ,: ~4,0 0 0
I 'h 0
Er::a:::EI MIL EIS
1000 0 1000 3000 50007000~ I I I
FEET
CONTOUR INTERVAL 200 FEET.-
•DATUM IS MEAN SEA LEVEL
SUBMARINE CONTOUR INTERVAL 60
BASE FROM U.S. GEOLOGICAL SURVEY MAGEOLOGY BY G.D. STICE AND\F:W: :,McCOYBAT HYMET RYIN T E RPRE TED B't\ G. D. STIC ESOUNDINGS ON U.S. COAST ~N~ GEODETICBOAT SHEET * 23 (1939). "" \
consist mainly of dense flows ofolivine basalt, ~icrite-baBalt,basalt, and hawaiite with numerousthin beds of vitric-crystal ash.An intra-caldera cinder cone (Qtcc)and its associated lava flows (Qtcl)of basalt are mapped separate~y.O~herRecent flows (~tcl) or ba8a~~occur. A dike complex intrudes thepre-c~ldera(?) (Tte~) volcanics.