THE OBSIDIAN INDUSTRY AT BUKIT TENGKORAK, SABAH, MALAYSIA

Stephen ChiaPusat Penyelidikan Arkeologi Malaysia

Universiti Sains Malaysia 11800 Penang Malaysia

INTRODUCTIOl\\

In 1994-95, archaeological research was undertaken by a joint Centre forArchaeological Re:,earch Malaysia and Muzium Sabah team at Bukit Tengkorak, aNeolithic site in Sempoma, Sabah (Figure 1). Geologically, Bukit Tengkorak fomls partof the rim of a 2 kllometre-wide volcanic crater, surrounded by numerous isolated hillsand mountains, most of them representing sites of extinct volcanoes ranging fromPliocene to Quaterr.ary in age (HD Tjia, personal communication, Kirk 1962, Lee 1970).Two seasons of excavations, over a period of 5 weeks, were carried out at two volcanicoutcrops near the summit of Bukit Tengkorak, approximately 600 feet about sea level l

. Atotal of 6-one metre trenches (G17, G19, 119, R36, S37, and T38), three in each outcrops,were excavated until the base of the undisturbed cultural deposits, about 150 em inmaximum depth (Figure 2). The excavated area covered about 10% of the total area thatcan be excavated. The top layer (0-20 cm) of trenches G17, G19, and 119 appeared to bedisturbed but the subsequent layers contained undisturbed artefacts which were excavatedin arbitrary controlled spits of 5 cm deep per spit or level. More than 6 cubic metres ofsoil was excavated and sieved through 1 mm and 0.2 mm meshes. A broad range ofarchaeological mat'C'rials were recovt:red and i~ley include large quantities of potterysherds, chert and a.gate microliths, obsidian flakes, polished adzes, a bark cloth beater,and some shell or bone artifacts. The abundant food remains, mostly marine mollusks andfish bones (as well as some terrestrial animal bones) is indicative of a maritime-orientedsociety at Bukit Tengkorak.

Five radiocarbon dates place the site between 4300 BC and possibly 50 BC. All theradiocarbon dates are listed in Table 1. Four of the samples are charcoal and one (Beta­744448) is of Anadara shells. The charcoal samples were calibrated with the StuiverINT93 Cal prograrn (1993). Out of the five samples, one (Beta-744447) is modern,confirming that the top layers (0-20cm) of trench G17 is disturbed. As such, the absolutedate for the last use of the site is still unknown but given the absence of metal artifacts,stonewares, and procelain at the top layers, the site is probably abandoned before 2000BP or around 50 BC2

• On the basis of the radiocarbon dates, soil stratigraphy, and thetemporal distribution of artifact type's, three occupational phases: Early Phase (4300-1200BC); Middle Phase (1200-900 BC); and Late Phase (900-50 BC), were defined at BukitTengkorak.

1 Test excavations were done at Bukit Tengkorak in 1987 by Bellwood (1989) & Bellwood & Koon (1989)2 Despite the lack of a well-established date for the fIrst use of metal in island Southeast Asia, a date ofabout 2,000 BP is generally accepted for the arrival of bronze and iron artifacts, particularly in Java, Bali,the Tahiud islands, and Sabah (Soejono, 1979; Bronson and Glover, 1984; Bellwood, 1985).

Table l: Radiocarbon dates from Bukit Tengkorak, Sabalr

I.zb# ConvcntionalAge (BP)

CalibratcdAgc (BC)'

Matcrial Noles

Bet*74447

Beta-74448

Beta_83783*+*

Beta-83784

Beta-83785

10t.0 + 0.9%M

3190+60

.2e40 !50

2650 t 90

5330 + aO

I l9O to 860**

1285 to-o90

980 to 745

700 to 530

434O ro3975

charcoal

Anadarashells

charcoal

charcoal

charcoal

Trcnch GlT,layer Ispit 4, l5-20cms

Trench Gl7,layer 3

spit li, 50-55cms

Trcnch Gl7,layer 4spit 15, 70-75cms

Trcnch GlT,layer 5

spir 20, 95-loocms I

Trench GlT,ltyerS I

spit 26, 125-130cms I

t**

Cal BC datcs (2 sigma, 95% probabiliry)This marine shell sample has bcen calibrated according to Stuiver & Braziunas( r ee3).'Ihis sampie was analyscd:rsirrg AMS (L.rivrencc Live,more).

rF+*

THE OBSIDIAI{ INDUSTRY

A total of 552 obsidian artifacts, weighing about 145 grams, were recovered duringour excavations at Bukit Tengkorak. The obsidian came in a vaiety of colors,translucency, luster, and texture; from very glassy, highly hanslucent, and black in colorto less glassy, totally opaque, and grey in color. They occurred mostly in the form of tinyflakes, weighing below 5.6 grams and measuring under 2 cm in maxirnum dimension(Figure 3). The obsidian assemblage v/ere classified into core fragments (3%), utilisedflakes (27%), retouched flakes (6%), and waste flakes (64%) -Figure 4, Tables 2 & 3.

The flaking technology was directed at producing very small flake tools instead ofblades. This is evident from the large number of utilised and retouched pieces. Only asrnall percentage (14%) of these utilised or retouched pieces may be coniidered blades,with their length/width ratio exceeding 2. The lack of negative blade scars on the corefragments also indicates that the cores were not originally designed to produce blades.The majority of the utilised or retouched pieces arc very small, averaging about 1.1 mmin length and about 0.3 grams in weight. It is most probable that these pieces need to behafted as they would be too small to be manipulated solely by hand. Some of theretouched flakes have secondary flake scars along the sides and ends of the flake.

f\ oO.rr.r-rr.nr- I1.5rlrtc

C,rfl

'Il-lrlr On'o|,a.'

!.Af

),ta

,o

2. a,txrurt

.rt- lqri fi?i

t tt; t.d

fr.n

hellnla taart

.fl.3rn tq

>jD

r. t;ri,

It

@'4^"@ N

.:.:: /\,.,.i':: f I ' , .

satl.l.l.l

alGer!:

-<G

- 1- - Rjo rlvokrnic srrtcr;--. Approrirrr|c toc.dor cfthcecltc afrhcAcitu&r of trrr lloe ildtolcu& lrgir

Figure l: Map of Bukit Tengkorak in Sempoma, Sabah

N

dl.

Tr"it

Figure 2: Plan of Bukit Tengkorak and the excavated trenches

The majority of the waste flakes are block-fractured flakes (7z%)while the rest wereconchoidal flakes (2s%)' The conchoidal flakes rraue closely simiiar iui, ,rri"t resses,averaging about 0'27 mm, showing consistency in flake removals. A large number of theretouched and utilised pieces (60it were made of block-fractured flakes (although theseflakes are less prefered by most tooirnakers). The low frequency of corticai flakes (4.5%)suggested that the c'res had their cortexes rernoved, probably before they were brought tothe site' The core fragments are very small, *"ururing less than 2.5 cmin maximumdiameter and weighing not more than 5.61 grams. Ali these plus the srnall amount ofobsidian (145 grams) recovered during the eicavations seem to indicate that obsidian isscarce and hence the desire to make fuil use of this exotic and valuable material.

OBSIDIATY SOURC'ES

During out 199.4-95 excavations, geological surveys were carried out in and aroundthe Semporna Peninsula but no sources of obsidian were found. since obsidian is knownto be widely exchanged or traded over long-distances during the Lapita period (1600 BC)in the southwest Pacjfic, it was decided tt ut u sourcing .toay be carried out to see if theorigins of the obsidian artifacts can be traced to un; of the 66 known sources in thesouthwest Pacific anclsome Indonesian islands which irave been chemically characterizedand catalogued by J''R. Bird and his colleagues (Bird, et al. lgTg,Duerden et al. 19g0,Bird' et al' l98la' 1981b, smith et al. 197i, Ambrose et al. l9gl, summerh ayes et al.1993)' over the last 20 years or so, research in the southwest pacific had also shown thatallocating obsidian -a*ifacts to geographical sources can be successfully done using avariety of elemental:rnalysis (smith et ar. 1977, Bird et ar L9g7a,l9glb, smith rggz,Duerden et al. 1987, (ireen l9g7).

-obsidian Analysis: A total of 30 of the 552 pieces of obsidian artifacts ercavated fromBukit Tengkorak were analysed using u n ny automated cameca MBX electronmicroprobe with wavelength dispersive X-ray splctrometers at the Department of Earthand Planetary Sciences at Harvard universiry. th"." fi"r.. were selected from obsidian

.

artifacts recovered from the undisturbed spits t2 to zo 1oo-t30 cm) in trenches Gl7 andJl9' radiocarbon dated to between 900 and 4300 BC. obsidian artifacts with differentvisible characteristics such as color, translucency, luster, and texture, that might indicatedifferent sources, were selected. This is also done in order to reduce sample bias towardselecting obsidian pieces produced from a single piece of core. Based on ull these criteria,a total of 30 samples; l5 from trench Gl7 anrl another 15 from trench Jl9, were selectedfor analysis' The elecb'on microprobe analysis was chosen as a method of choice mainlybecause it is minimalllr 6t.r*ctive (only 1mm size sample is needed) and is a relativelyfast and accurate method (r-2% accurary) for determinirrg tt e selected range of elernentswithin the required detection limits or o.ot to 0.1 *"igf,t percent, depending upon theelement and compositiori of the sample._The selected ;g" of elemenJ, ur" iru-.ly si,Al, Fe, Mg, ca, K, p, Mn, and Ba. These elements -l *orrg the most distinctiveelements that have been found to be useful in distinguishing the 66 known obsidiansources in the Pacific and its neighbouring regions (Greel and Bird r9g9:92).

Results and Discussion: The results of the electron microprobe analysis show three

different compositional groups, suggesting that the artifacts were made using obsidian

obtained from three different sources. The microprobe data are presented in Table 4. Two

dirnensional x-y cluster analysis is also used to examine the data using combinations of

two different oxides. The results reveal at least three distinct compositional groups

(Figure 5). Another noteworthy observation from this study is that each of the two main

compositional groups, A and 6, contained obsidian samples with distinct visual tlpes' In

group A, all the ousidian samples are generally very glassy, highly translucent, and black

in color while grouf B has obsidian that are less glassy, less traslucent, and grey to black

in color. The single sample in Group C is less glassy, totally opaque' and black in color'

t6

15

l3

l2

+

-ryx

o

s70 72 74

sior (%)

76

Group

OAXBOC

a\-- 14o

78

Figurc 5 : Plot of Ahor vs sior of obsidian Artefacts From Bukit Tcngkorak

The three compositional groups were then carefully compared with the database of 66

known sources in the Pacific and the neighbouring regions. Based on the weight percent

value of eight elements (Si, Al, Ti, Fe, -a, Na, K, and Mn)' each of the compositional

groups were compared with corresponding elemental weight percent values of the 66

known sources presented by Dueiden et al- (1987, Table 1) and a recent updated

catalogue on Pacific obsidian source composition provided by Dr' Roger Bird'

D

The results of the comparisons showed that compositional group A matches chemicallywith the Kutau/Bac, sub-source at Talasea in New Britain, western Melanesia. Group Bmatches the few obsidian artifacts from the Talaud islands, for which no geologicalsource has yet been identified. The single obsidian artifact representing group C appearsto match obsidian s()urces in the Admiralty Islands (Solang, Umrei, and Wekwok).

In terms of stratigraphic levels, the majority (15 pieces or 88%) of the Talaseanobsidian come frorrr the upper levels 12-19 while most (8 pieces or 64%) of the group B(Talaud artifact group) come from the lower levels 23-26. Chronologically, this change inthe relative importance of obsidian sources with time seems to suggest that the unknownobsidian source represented by group B (Talaud artifact group) was an important obsidiansource at around 4300 BC but became less important at around 1200-900 BC. TheTalasea source, on the other hand, appears to be important from 1200 to 900 BC.

CONCLUSIONS

Our research shows that the prehistoric inhabitants of Bukit Tengkorak obtainedobsidian from three different sources at around 4300 to 900 BC to produce srnall flaketools. The majority of the obsidian artifacts, dated around 900-1200 BC, originated fromthe Kutau/Bao sub-source in Talasea, New Britain. This connection, over a distance ofabout 3500 km, not only represents the longest traded obsidian in the world for this timeperiod but shows the existence of long-distance sea trading networks which extended in awestward direction from Melanesia to as far as Borneo in Southeast Asia. The most

intriguing question arises as to the possible source of the obsidian artifacts, dated between

4300 to 1200 BC at Bukit Tengkorak, which matches a few artifacts from the Talaud

Islands but have yo: to be traced to any geological source. Since no obsidian sources are

known in Borneo, the presence of obsidian earlier than 1600 BC at Bukit Tengkorak

implies that advance seafaring trading networks could have existed in island Southeast

Asia at least 2500 )/ears prior to the Lapita period. It also opens up the possibility of an

early trading systenr as early as 4300 BC between Southeast Asia and Melanesia.

ACKNOWLEDGI'MENTS

This research fromr; part of my PhD work at Bukit Tengkorak which was supported by

the Intensification ,rf Research in Primary Areas (Universiti Sains Malaysia), Muzium

Negara, and the To;rota Foundation. I would like to thank Dato' Professor Ishak, the Vice

Chancellor of Universiti Sains Malaysia and Dato' Professor Ztnaina Majid, my

supervisor and the director of the Centre for Archaeological Research Malaysia as well as

Tuan Haji Zulkifli, the former director of Muzium Negara, Datuk Jamdin Buyong and Ms

Particia Regis, fornrer directors of Muzium Sabah. I am grateful to the following for their

assistance in the lield: Peter Koon, Peter Molijol, Osman Nassib, Affendi Rahrnat,

Jamain Musi, Johari Sibuli, Anthony Sintau, Uling Lakim, and Jeffrey Ating of Muzium

Sabah, Junaidi Rashid and Sanim Ahmad of Muzium Negara as well as my colleagues

)

o

I

t

Ivl6lLtolylolctolFl+JirlJlolEIgll-

lololo.IFl-lr!lolsll!l!lo

lll*l+,lt!

tiI$t:l!l!t:lololfl

noF

oo ooroo o

oroo oo oo oo oo

oo

ooF

ooroor

oo oo

oo

oooo o o

ooo

oor

ooo oo

roo

oo

oor

oo

oo

ooooF

o oor

ooo

@qo

{oci

loqo

qo

oqo

oqo

oci

oqo

$qo

Nqo

oc;

@qo

oci

oci

ooc,

oqo

go

oci

Nqo

o o qo

Nqo

o ocj

Noo

Nqo

qo

o qo

qo

qo

qo

6oci

o

(oqo

oqo

ooci

oqo

|oocj

(ooci

oqo

\toci

(oqo

qo

(oqo

(oqo

(oqo

(ooc;

(oqo

(ooo

l{)oci

ooct

oci ci

oqo ci

F

oooc;

oqo o o ct

ooc;

ci ciF

ctr (o

qo

(r)

oNo

ooct

Nqo

(f)qo

oqo

oqo

oqo

\toci

toct

tqo

tqo

ttqo

ooct

cDoo

oocj

oqo

oqo

tci

tf

cj oNqo

oc;

oqo

oci

ocj

oct c;

N

ct

Nqo

Nqo

o

ci

Nqo

N

cj

o \t

o

oNY

@q(I,

oq(I}

@q(f)

0lo

N0)({'

oort

c0q(I)

ori

o6ct

0l(il

0)o

N0?

oq(i,

al0lc)

(ooFi

Nod

oqo

qo

c0qo

@

\u)

(oqo

F@d

o

'rt

f?od

rod

(o@

dtod

Nod

qu)

0lo

f06d

N@

d

(oqo

(I}

{

oNt!z

0lo

oct o?

(aqo

oq(Y)

oqo

oqo

oqo

t.q(9

l.-q(')

(oori

td}rt

F'qo

oqo

()qo

oolo

t0lo

oqcl

\tqo

t0l{

(ood

Nod

ord

od

Nout

oolt

oqrt

qu)

N0lt

o0lt

o01

to q

o

()0l\t

or0o

FtF

(a NI I

C)F

@I

o o:(oI

(0 tt oI oI@I

(o

I(t'I

u)I q

o\o

o\o

ocj

o\o

oC}

ci

o\o

qo

o\o

o\o

o\o

o\o

o\o

o\o

Nqo

t

ooz

N(.1

orNct

NNc;

N0!o

Nt\o

Nc!o

NNci

Ncrl

oNC\o

(I'f!o

Nf!o

Nolo

N0lo

Nc!o

Nc\o

NNci

NNci

Nc!o

oci ct

F

o o

(!l

cict

(')qo o o o

oqo d ct q

otolo

oNolt

Nnl

glIr

N(\ Inc!

oI()C\F

Fo.|

c0c\

o0!

c! oI t

c!oc!r

$q clC\

Nq Nc\

oqo

@\oq

(r)q q (r,

\olF

qr

o Nqr

(o

\o\

6 qr

C'qo

lt1N

No

Nolo

c,qo

toc\o

rtc\o

Nolo

oqo

Nolo

Nc!o

oc!o

N$io

Nqo

tqo

onlo

Nc!o

oc\o

oolo

NNcj

(tlNct

qc

oci

c!o

qo

olo

rof\o

Nct

@F

dNc;

Nd N

ciNcj

Nci C\

oNoci

c,o

(9oN

u?

Nrtu?

Nr

(ou?

NF

(olo$i

{ooi

t.u?

N

tlo$i

qN

o1NF

u?

N

u)ni

ou?

N

$u)o,i

rooi

NqN

(r)ooir

o1N

tloni

$qo

oqtF

ot

u)ot

\t6t

oc!or

oqI

oo{

o\!f

Nq{

oqt

o.iF

(oqt

oq!t

oci

@

\o

NIU'

o(o

doC;

N

@\(o

(oq(o

N

o

(oq(o

q(I)

q,

ct

(oq(oF

u)

Gi

oq(o

1..(o

ooct

o\() o

dnd

to@qo

rtql.-

osl

u)q al€qo

()al \ @

r?Nr?

oa?

a?

N

ul o0?

N(\o

cn

ooJ

Nr N N N N o t \tF

tF

ro oF

@ oF

NN

{N

t o @ o oN

f0N

ttN

rtN

rtN

rtN

oN

(oN

u)

.goEoU,

otto-)

\to-)

oorfo1

rOoro

oo(ooo-)

rN

o

(ooo

o

Ntoo-)

$ooo-)

(v)

om

F(,

o0/)

o

o

o(oN

o

(0ooN

F

o

o\fN

1..

o

ootN

o

@No

o

of0lnNo-

o(ooo-)

oNoo-)

NF

ot.o

oN

o

nNoo-)

oooo

@c0oNo

@t(oN

o

oto

o

NloN(')

-)

oN

o

oooF.

o

otooo-t

o

Io

coq)

u o o o o o 6 6 o o 6 o 6oq)

u

a

EF{ SauTHEAST Asmt5f-\te?af*\r\ r\tr ?_\* *{