Pacific Science (1975), Vol. 29, No.4, p. 309-315 Printed in Great Britain

Lapita and a Lower Level in Western SamoaI "0

R. Co GREEN 2 AND HORACE G. RICHARDS3

ABSTRACT: Radiocarbon dates are presented supporting previous estimates of a 2800- to 3000-year B.P. age for a collection of Lapita pottery sherds recovered by dredging a now-submerged coastal settlement on the island of in Western . New data describing a much enlarged collection are discussed in relation to previously reported materials, and the question of possible changes of sea level as the mechanism for submergence is evaluated.

IN RECENT YEARS, Polynesian origins have in­ The recent discovery of other assemblages creasingly become identified with certain early with Eastern Lapita-style pottery in Ha', sites that range from to and Vava'u, and Niuatoputapu in the Tongan group Samoa and that contain a type ofpottery known (Davidson 1971, Kaeppler 1973, Rogers 1974), as Lapita (Figure 1). Excavations of sites in the in conjunction with the Samoan discovery, Tongan, Fijian, New Caledonian, New Hebri­ implies that early Eastern Lapita populations des, and Santa Cruz groups and on small islands were resident throughout the Fijian, Tongan, off the coasts of New Britain and and Samoan area before any distinctive develop­ have provided data leading to the definition of ment of Lapita culture in the Polynesian direc­ an associated assemblage-perhaps tion took place. The recovery ofa related plain more properly termed a "cultural complex," pottery, as in Samoa, from a site on Futuna because the array of artifacts from these sites is (B. G. Biggs, personal communication to R. Co more varied than the pottery and is not always Green, Auckland Museum no. 46332, 46371) as distinctive. Thus, it is the easily recognized indicates that in at least three island groups of decorated pottery in every site (Figure 2) that the region-Samoa, Tonga, and Futuna-local has provided the initial means of linking these developments of Polynesian-type cultures oc­ materials. curredfrom founding Lapitapopulations. How­ The history of the identification of Lapita ever, the sole island group for which we can assemblages as the ancestral cultural complex trace continued continuity of cultural develop­ from which the earliest Polynesian cultural ment to present-day is in Western assemblages were derived has been reviewed by Samoa (Green and Davidson 1974: 224), as the Golson (1971), Groube (1971), and Green last 2000 years of the Tongan sequence is but (1973). Until recently, however, archaeological little known and archaeological investigation evidence for this hypothesis has rested largely in Futuna is being undertaken only now (P. V. on the materials recovered from and Tonga. Kirch personal communication to R. Co Green). The discovery, therefore, ofsimilar materials in Thus, in the present circumstances, a Lapita site Samoa has strengthened considerably the elaims in Western Samoa that extends the sequence of made for the same antiquity ofsettlement there that island group back another 700 years to an as in the other two island groups and for the early Eastern Lapita-style pottery assemblage is development of a distinctive Polynesian cul­ of considerable importance to the documenta­ tural assemblage in Samoa itself from local tion of Polynesian origins. assemblages of the Lapita complex. Assistance in the identification and interpreta­ J Manuscript received 6 January 1975. tion ofthe mollusks was given by Dr. R. Tucker Z The UniversityofAuckland, Private Bag, Auckland, . Abbott of the Delaware Museum of Natural 3 The Academy of Natural Sciences, Nineteenth and History in Greenville, Delaware. the Parkway, Philadelphia, Pennsylvania 19103.

309 20-2 310

c. ~ - "'~ew Ireland PHOENIX ...... ~~ Ambitle GROUP ." "'~,.~ fo .. q New Britain ~,~SOLOMON IS. " . ELLICE IS.·, Ts> KELAU ~~~.S\ ~. I .:. • ~ IS. ~ ~ Reefs", ····SANTA CRUZ ~woins 'e:::..- It?t.... Nendo • Roturno Uvea BANK~~': Futuna I " S';\~i Niuofo'o Malo/~~ /'Niuatoputapu NEW ;!;..- AUSTRALIA • HEBRIDE~ ..",,.-Vovou " CI TONGA. . NEW" ,. • •"..:Haa pal CALEDONI~~ ·:·~.LOYALTY .... fongatapu r~"" ISLANDS

o 0, '.

FIGURE 1. Present distribution of sites that have been assigned to the Lapita cultural complex.

FIGURE 2. Design of flat-bottomed Lapita dish from the BS-RL-2 site, Reef Islands, Santa Cruz group. Lapita Pottery-GREEN AND RICHARDS 311

from earlier Lapita assemblages to those with Background ofSamoan Discovery Polynesian Plain Ware took place in Tonga, and In 1964, when only the outlines of a 2000­ only after these occurred did a more Polynesian­ year-long sequence for Western Samoa were like cultural tradition become transferred and known, it was necessary to argue that materials established in Samoa and perhaps in other belonging to the first third of the of West . sequence had not as yet been recovered. It There are now several grounds on which this was made conditional on a postulated 2000­ conclusion may be challenged, grounds quite year antiquity of settlement in East Polynesia apart from the discovery of Lapita pottery in and a hypothesis that East Polynesia came from Samoa. One is the increasing probability that West Polynesia (Green and Davidson, unpub­ the earliest sites associated with pottery in the lished preliminary report). As investigations Marquesas date to before 300 A.D. and, in fact, continued from 1965 to 1967, the requirement have not been investigated (Green 1974: 247). for a Samoan sequence longer than 2000 years The second is the probability that present was reinforced by the discovery of additional coastal settlements on the raised beach sands of sites with Polynesian Plain Ware pottery. Western Samoa date from the last 1500 years Polynesian Plain Ware may be distinguished (Green and Davidson 1974: 222-223), suggest­ from ceramics of the Lapita series by its total ing that the most likely places for earlier coastal lack of the use of the dentate stamp in decora­ settlement have not been examined. The third tion and the confinement of such decoration as is the hypothesis that some changes in Lapita does occur (less than 3 percent and in some artifacts in the Polynesian direction recorded in cases none) to the rims of vessels, In addition, the pottery, , and shell tools and orna­ there is a concentration on several simple bowl ments in Samoa are, in part, to be explained by forms ofvarious sizes and shapes at the expense the crossing of the andesite line and the need to of open-mouthed subglobular pots and an adapt (in contrast to Fiji and Tonga) to the bio­ almost complete lack of shouldered and flat­ geographically most impoverished part of the bottomed vessels of types common in Lapita Pacific (Green 1974: 267, 275). assemblages, 'This means that continuity is pro­ vided by the largely unstudied and undecorated The Ferry Berth Site Lapita vessels of bowl shape. Also the spatial distribution ofPolynesian Plain Ware is Tonga, Our original information on the first collec­ Samoa, the Marquesas, and probably Futuna, tions of pottery from the site at the Mulifanua and its time depth is approximately 4th century end of the island of Upolu and Western Samoa B.C. to 3rd century A.D., which contrasts with was limited to some 563 sherds, of which some the island Melanesian distribution of Lapita 8 percent were decorated, and to some general materials whose time depth is prior to the 4th comments on its location. More detailed records century B.C. made by Jennings (1974), in cooperation with Yet, the lack of direct evidence of earlier the project engineer, T. T. Hassall (who was in coastal assemblages more closely related to the charge of the blasting and dredging to create a Lapita examples found in Tonga and Fiji made turning basin and berth for the interisland it difficult to furnish strong support for this ferries), have revealed the precise location and position, Thus, it was possible for Groube stratigraphic context (Figure 3). All evidence (1971 : 279) to argue that current archaeological indicates that the site was an elongated settle­ examination of Samoa made unlikely the state­ ment on a former coral-sand beach located ment that the first 3rd ofSamoan was parallel to the present-day shoreline, a beach as yet unexplored, and to see no reason why which is now approximately 2.7 meters below Samoan settlement had occurred earlier than mean sea level. It would appear that refuse from 300-200 B.C. unless secure dates for East Poly­ the settlement was submerged and sealed in nesian settlement were pushed back before the fairly quickly through a rise in sea level in then-accepted datings of 400 A.D. This allowed relation to the land, so that deposits were little Groube to conclude that the initial changes disturbed and most sherds not badly eroded. 312 PACIFIC SCIENCE, Volume 29, October 1975

LAGOON MILE oEI=~~I~==:::l KM

250~ I' -- ...... _- ---I

(no scale)

MAIN SEA LEV EL

TURNING BASIN 5' WATER ~~~~~~~2~1/~11~.CEMENTED ~ CORAL CRUST ...... :~'.'. BLACK SAND ...... ­ ...... --- ZONE POTTERY POTTE RY IN BLACK SILTY / CONCEN. 15' CORAL SAND WITHIN ,/ CORAL SAND '(. ___D~SHED LINES / / ,. BASALT ...... PEBBLES

50' BOULDERS o I---l 15.2 M

SOLID BASALT

FIGURE 3. Location and geological context of Ferry Berth site, Western Samoa.

The problem of whether the changes were sherds, of which 4.1 percent are decorated. movements ofland or sea level or both has not Among the Plain Ware sherds were 873 pieces been resolved. ofinterest belonging to rims, necks, shoulders, More recent collections from dredging in the and bases ofa variety ofvessels oftypical Lapita turning basin have provided an additional 4288 forms. It is probable that the percentage of Lapita Pottery-GREEN AND RICHARDS 313 decorated ware is somewhat lower than it within the past 6000 years. The other interpre­ should be, for we know that some decorated tation, held by Shepard (1963) and others, indi­ pieces were retained by various interested cates a gradual rise of the sea from the last parties. It is also prob:J.ble that the amount of glacial episode to the present, with no stands decoration in the first collection is somewhat being higher than now. high owing to an initial interest in these pieces. As will be stated below, other Lapita sites in Thus, approximately 5 to 7 percent of the pot­ the Polynesian region occur on " raised tery in the site belongs to the decorated beaches" above present sea level. This might category. seem to confirm the"Fairbridge Curve," indi­ From the style and frequency of the deco­ cating sea level both above and below that of rated sherds of pottery, from the vessel forms the present during the late Holocene. On the represented, from 14C ages for comparable other hand, it must be realized that many of pottery from Fiji and Tonga, and from 14C ages these islands are highly tectonic, and, therefore, for coral and shell overlaying the site (see it is equally possible that the higher beaches below), we can assign it to the Early Eastern with the Lapita cultural material were uplifted Lapita style of the Lapita ceramic series (Green during relatively recent times. , 1974) and place its age at circa 1000 B.C. From The presence of the cultural material below the studies of the temper of the pottery in all present sea level and covered with a coquina sites in Samoa, we can confidently say not only deposit can be explained by a lower sea level that the Ferry Berth pottery was made locally, during the Holocene rise from the glacial low but also that it was probably made on the end stand. A sea level about 3000 years ago that was of the island on which it was found (Dickinson some 2.7 meters lower than that of today is 1974). Finally, from a close analysis of tech­ logical. Furthermore, it does not disprove niques of manufacture, vessel form, and decor­ Fairbridge's contention ofa Holocene sea stand ation, we can argue with reasonable confidence 'higher than that of today. that the Lapita pottery of Samoa was ancestral to the well-dated Polynesian Plain Ware of the first few centuries B.C. to the first few centuries Molluscan Fauna A.D., before pottery manufacture in Samoa Seven species of marine mollusks have been ceased (Green 1974: 248). This latter pottery, identified from the coquina lying above the of course, is that associated with cultural cultural layer. These are: Gastropoda-Euchelus assemblages containing the range of other por­ stratus (Gmelin), Turbo sp. (operculum only), table artifacts and structural features that begin Cerithium aluco (Linnaeus), Strombus mutabilis the well-documented and continuous part of Swainson, Cypraea annulus Linnaeus, and Bulla the Samoan sequence ofthe last 2300 years. We ampula (Linnaeus); Bivalvia-Fragum fragutlJ conclude that the Ferry Berth site represents a (Linnaeus). All are living today in Samoan local assemblage of the Lapita complex at the waters and are generally found from the inter­ early end of the Samoan sequence, after which tidal line to a depth of 5 meters of water. All there a gap before reaching materials from is live in the vicinity of coral lagoons or reefs. the first few centuries B.C.

Relative Changes in Sea Level 14Carbon Dates As stated above, the explanation of the Two radiocarbon results confirming previous Samoan site being below present sea level can estimates of 2800 to 3000 years B.P. (Jennings be interpreted either by changes in sea level or 1974: 176; Green and Davidson 1974: 224) are by movements of the land. As far as Holocene now available., The samples used were pieces changes in sea level are concerned, there are from the base of the cemented coral crust or two major schools of thought. One, proposed coquina (Figure 3) which contained shells and by Fairbridge (1961), shows several stands of pieces of pottery as inclusions. One result, NZ the sea as being higher than at the present, all 1958, dates the shells included in field sample 314 PACIFIC SCIENCE, Volume 29, October 1975

TABLE 1 Green), as is the earliest site on Santa Cruz Island (Nendo) of three dated sites in the Santa RESULTS OF "CARBON TESTING OF CORAL CRUST OR COQUINA LAYER FROM SUBMERGED BEACH OFF Cruz group. Sites in are also WESTERN SAMOA associated with raised beaches, which may indi­ cate former higher stands of sea level (Frimi­ AGE IN YEARS gacci 1971, 1974), as may some of the sites in SAMPLE STANDARD BEFORE 1950 Tongatapu (Groube 1971: 297). On Niuatopu­

NZ 1958A old Tt 2890±80 tapu, all Lapita sites are confined to a former NZ 1958B newTt 2980±80· raised beach level that encircles the rising center NZ 1959A old Tt 2170±70 of the island, whereas none are found in the NZ 1959B newTt 2230±70 exposed former lagoon flats later used for occu­ pation (Rogers 1974). Thus, archaeological evi­ NOTE: Corrections supplied by the New Zealand dence of Lapita materials from a wide area of Radiocarbon Dating Laboratory. geologically fairly diverse island types is associ­ ated with relative changes between land and sea SU/Cf18/1; the other, NZ 1959, the coral that level from 1300 B.C. to about 500 B.C. In this formed the cementing agent. Results in the context, the presence of a Lapita site in Samoa form as they are now reported by the New from a now submerged situation is sufficiently Zealand Radiocarbon Dating Laboratory are interesting to warrant further geological shown in Table 1. As requested by the labora­ investigation. tory, I discussed the first result with J. C. Schofield of the New Zealand Geological Sur­ vey, Otara, because the laboratory was worried LITERATURE CITED about the possibility of recrystallization in the DAVIDSON, J. M. 1971. Preliminary report on coral during or after cementation. Basing his an archaeological survey of the Vava'u opinion on other work, Schofield correctly pre­ group, Tonga. Pages 29-40 in R. Fraser, dicted that the coral result would prove to be compiler. Cook Bicentenary Expedition in too young and would have to be checked the south-west Pacific. Bull. R. Soc. N.Z. 8. against the shell inclusions. He also believed DICKINSON, W. R. 1974. Temper sands in that the coral crust would have formed fairly sherds from Mulifauna and comparison with rapidly, so that no great interval in time need similar tempers at Vailele and Sasoa'a have elapsed between abandonment of the site (Falefa). Pages 179-180 in R. C. Green and and the formation ofthe protecting coral crust. J. M. Davidson, eds. Archaeology in Western This fits with the generally good state ofpres­ Samoa. Vol. 2. Bull. Auckland Inst. Mus. 7. ervation of most of the pottery sherds and the FAIRBRIDGE, R. W. 1961. Eustatic charges in shells. Since the age of the shell inclusions con­ sea level. Pages 99-185 in L. H. Ahrens, F. forms with expections, both in relation to the Press, K. Rankama, and S. K. Runcorn, eds. coral date and to previous estimates for pottery Physics and chemistry of the earth. Vol. 4. assemblages of this type, we accept it as a Pergamon Press, New York. v+317 pp. reasonable estimate. FRIMIGACCI, D. 1971. Fouilles archeologiques a Vatcha (pres de Vao), tIe des Pins. Etud. Melanesiennes 21-25: 23-42. Other Lapita Sites ---. 1974. Les deux niveaux a poterie du Other Lapita sites, a few of them of slightly site de Vatcha. J. Soc. Oceanistes 42-43: 25­ greater antiquity, are known from Tonga, Fiji, 70. New Caledonia, the , and the GOLSON, J. 1971. Lapita ware and its transfor­ Santa Cruz group. Some are associated with mations. Pages 67-76 in R. C. Green and M. changes between land and sea levels. Thus, Kelly, eds. Studies in Oceanic culture his­ early sites on Malo in the New Hebrides are on tory. Vol. 2. Pacific Anthropological Records. a raised beach level inland of the present coast No. 12. Bernice P. Bishop Museum, Depart­ (J. Hedrick, personal communication to R. C. ment of Anthropology, Honolulu. Lapita Pottery-GREEN AND RICHARDS 315

GREEN, R. C. 1973. Lapita pottery and the ori­ JENNINGS, J. D. 1974. The Ferry Berth site, gins of . Aust. Nat. Hist. Mulifanua district, Upolu. Pages 176-178 in 17(10): 332-337. R. C. Green and J. M. Davidson, eds. Ar­ ---. 1974. A review of portable artifacts chaeology in Western Samoa. Vol. 2. Bull. from Western Samoa. Pages 245-275 in R. C. Auckland lnst. Mus. 7. Green and J. M. Davidson, eds. Archaeology KAEPPLER, A. L. 1973. Pottery sherds from in Western Samoa. Vol. 2. Bull. Auckland Tungua, Ha'apai: and remarks on pottery lnst. Mus. 7. and social structure in Tonga. J. Polynesian GREEN, R. c., and J. M. DAVIDSON. 1974. A Soc. 82(2): 218-222. radiocarbon and stratigraphic sequence for ROGERS, G. 1974. Archaeological discoveries on Samoa. Pages 212-224 in R. C. Green and Niuatoputapu Island, Tonga. J. Polynesian J. M. Davidson, eds. Archaeology in Western Soc. 83(3): 308-348. Samoa. Vol. 2. Bull. Auckland lnst. Mus. 7. SHEPARD, F. P. 1963. Thirty-five thousand years GROUBE, L. M. 1971. Tonga, Lapita potteryand of sea level. Pages 1-10 in Thomas Clements, Polynesian origins. J. Polynesian Soc. 80(3): ed. Essays in marine geology in honor of 278-316. K. O. Emery. University ofSouthern Califor­ nia Press, Los Angeles. 201 pp.