Volume 98 1989 > Volume 98, No. 3 > New information for the Ferry Berth site, Mulifanua, Western Samoa, by H. M. Leach, p 319-330
                                                                                             Previous | Next   

- 319

The history of the discovery of the Mulifanua Ferry Berth Site (Fig.1A), currently Samoa's earliest site and its only known Lapita one, was well documented 15 years ago by Green (1974a) and Jennings (1974). The presence of decorated sherds and vessel types characteristic of the Lapita culture convincingly indicated the site predated the Polynesian Plain Ware phase (300 B.C.-A.D. 200 or 300). This was subsequently supported by a date at one sigma of 950-1110 B.C. for a probable midden shell embedded with pottery sherds in the coral crust just above the pottery bearing layer (Fig. 1C).

The published result for the shell's age is 2890±80 years BP based on the old half-life value which had been corrected for what is now called the Ocean Reservoir Effect by using the New Zealand Radiocarbon Laboratory value for Oceanic surface water at 16°S. latitude (Green and Richards 1975:317; DSIR Institute of Nuclear Sciences (INS) July 1974 correspondence). Failure to fully appreciate this situation, plus a mistake in reporting the calculation of the standard error by the laboratory and additional data from the B counter mean a number of supposed secular age conversions for this date which have recently appeared in the literature are incorrect. 1 Current correspondence indicates the original conventional carbon-14 age for the NZ1958 sample determined by the DSIR INS Radiocarbon Laboratory is, in fact, 3231±250 years BP. Subsequently this was further corrected by H. Melhuish through applying an additional error multiplier for the B counter not included in the original result. This gives a conventional radiocarbon age of 3251±155 years BP. Using an updated marine correction of -50±5% based on shells from Fiji and Hawaii and Pearson and Stuiver's (1986) calibration curves, the INS Laboratory now report (October 1988 correspondence) a one- sigma age range of 3399-2779 BP, with a most probable age of 3116 BP or 1166 B.C. Other possible marine reservoir corrections can also be applied which result in most probable age values of 1043 and 977 B.C. 2 A likely true age of circa

- 320
FIGURE 1. Locality map (A), site map (B) and stratigraphic context (C) of the Ferry Berth Lapita site.
- 321

3,000 years ago for this site, as Poulsen (1987:2:Table 83) shows, accords well with dates for Lapita sites obtained throughout the Fiji-West Polynesian region.

In addition to the confirming date, some aspects of sea-level change or subsidence at the site were also further discussed (Green and Richards 1975; Jennings 1976) after the initial publications, and the ceramics examined in greater detail (Smith 1976a; Dickinson 1976; Baquie MS 1975) but little has been added to this record for over a decade. In addition to the recalculations on the age data reported above, the new information presented here concerns associated adzes and a reconstruction of the setting of the site at the time of the Lapita settlement.


The collectors who picked over the dredge tailings in 1973, chiefly Hassall, Hansen and Cowles, obtained more than 2,500 pottery sherds, of which approximately 5-7 per cent were decorated. Further collecting brought the total to 4,288 sherds (Green and Richards 1975:312). Despite this effort, only one “possible stone flake” (Green 1974a:171) and one “grooved abrading stone of fine-grained basalt” (Jennings 1974:176) were reported as occurring with the pottery. This absence of significant stone artefacts from the Mulifanua assemblage can now be remedied, some 15 years after the original site discovery. Rhys Richards of the New Zealand High Commission in Western Samoa visited the dredge tailings on several occasions in 1988 and found two stone adzes, the first in May and the second on July 2, while examining pottery sherds with Helen Leach. The material came from the section of the dredge tailings below the high-water mark but lying within the retaining wall built to hold them. It is believed that tidal action, in conjunction with the turbulence set up by the rocks in the wall, constantly reworks the upper portion of the tailings, bringing pottery to the surface and subjecting it to erosion. By removing any surface crust which may have coated the stone tools at the time of dredging, this wave action also seems to be making the stone component more visible.

The first adze (Fig.2A) is of oval cross-section with slightly flattened back, and rounded sides merging into the continuously curved front. It had been ground extensively on front and sides, but only the high points of the back showed grinding over the original flake scars. The butt end appears to have been rounded by hammer-dressing after flaking. The opposite end has been crudely flaked bifacially, possibly an attempt to re-form the cutting edge after damage or use. As a result, the original length of the adze cannot be determined. It is currently 89 mm long, 52 mm wide and 31 mm thick (maximum dimensions). The material from which it was made is a grey, coarse-grained volcanic rock, subject to step-fracturing as the platform angle steepens. The manufacturing process left sufficient traces to show that the adze was flaked bifacially from its lateral margins, and that its final oval cross-section was achieved by grinding these margins away to form a nearly continuous curve from back to front. The grinding shows a few narrow facets along the sides.

Hand-specimen examination by Dr A. Reay (Geology Department, University of Otago) suggests that the volcanic rock is a basaltic andesite, comparable to material known from the Tonga-Kermadecs area. This finding supports another feature of the

- 322
FIGURE 2. Two adzes associated with dredge tailings from the Ferry Berth Lapita site.
- 323

adze which is not characteristic of later Samoan adzes: the presence of hammer-dressing. As Buck noted (1930:355), Samoan adzes display grinding directly over the flake scars, without any hammer-dressing to remove high points. In contrast, Tongan adzes were often hammer-dressed or “bruised”, especially on the butt end (Poulsen 1987:1:163, 165, 168, 171). A brief examination of Tongan adzes at the Auckland Institute and Museum suggests that hammer-dressing is most obvious on adzes made of non-olivine basalts.

A third feature of the adze which is atypical for Samoa is its curvilinear cross-section. Although there is a tendency for the (presumed) back of the adze to be flattened, it does not fit into Green and Davidson's Type V (plano-convex in cross-section), because the Samoan Type V has a definable edge between the flat back and the curved front. In this Mulifanua adze there is a smooth transition between back, sides and front. Its cross-section is probably best described as oval with a greater curvature of the front than of the back. Such a cross-section is well represented in Poulsen's (1987) Tongan sample and would fit into his Type 2. It shares features of his Type 2a (1987:2:Fig.73 E 12) and 2b (1987:2:Fig.74 E16) but cannot be ascribed more to one subsection than another.

This artefact may well be a Tongan import but it has wider parallels which suggest that we should view it more as a typical East Lapita adze. The Birks (1968:106-7) described an adze from the Sigatoka Lapita level (No.1) which has a similar oval cross-section and measurements (width 58 mm cf. Mulifanua 52 mm, thickness 29 mm cf. Mulifanua 31 mm).

The second adze is a mid-section portion with a true plano-convex profile (Fig.2B). Its measurements are: length (incomplete) 42 mm, maximum width 38 mm, maximum thickness 22 mm. In cross-section and measurements it is similar to another Sigatoka Lapita adze (No.2) described by the Birks (1968:106-7) which has width of 35 mm and thickness of 23 mm. Additional parallels can be drawn with one of Poulsen's specimens (1987:2:Fig.74 E 18), a Type 2b from the late period of the Tongan ceramic sequence (Poulsen 1987:1:173). However, there is less likelihood that this adze was imported to Samoa, for two reasons: typologically the plano-convex form is common in early Samoan adze assemblages (Green 1974b:257; Smith 1976b:68; Hewitt 1980:136) and the rock type of this Mulifanua example is probably olivine basalt, the material in which nearly all Samoan adzes were made. Calculation of its shoulder index at 57, places it within Green's Type Va group (Green 1974c:137). In terms of manufacturing technology it shows the characteristic Samoan application of grinding directly on to flake scars. The incomplete grinding, attested by the presence of flake scars on the left side, is also typically Samoan. In comparison to the first adze, the grinding of the front and right side has resulted in much flatter surfaces.

Thus, the first two adzes to be recovered from the Mulifanua dredge tailings each reinforce the original statements made about the Ferry Berth Site. Although the larger adze reflects the interisland transport of artefacts, so typical of the Lapita period, the smaller one is typologically and geologically allied to Samoa. But, as with the pottery which was considered locally made (Dickinson 1974:179-801; 1976), this adze has close parallels in the other loci of Eastern Lapita, namely Tonga and Fiji.

- 324

In recent years considerable attention has been given to reconstructing the original environment of sites with Lapita and closely related pottery assemblages, and documenting subsequent landscape change. Reconstructions of former settings have been proposed for Lapita sites such as ECA and ECB on Eloaua Island in the Mussau Group (Kirch 1988:331 and Fig.1), TK-1 and TK-4 on Tikopia (Kirch and Yen 1982:81-125 and Figs 26, 37, 124), NT-90, 93 and 100 on Niuatoputapu (Kirch 1978:5-7 and Fig.4), and for numbers of Lapita sites around the lagoon and north coast of Tongatapu (Spennemann 1986:93 and Figs.11.5 and IV.4), both in the Tongan group. Shoreline changes have also been noted for late Polynesian Plain Ware sites in American Samoa at Aoa (Clark and Herdrich 1988:140, 173-5 and Fig.36), and on Ofu and Ta'ū in the Manu'a group (Hunt and Kirch 1988 and personal communication). About a decade ago, when it was realised that the Samoan Islands were probably slowly subsiding (Jennings 1976:7), Green suggested to a prehistory student at the University of Auckland that he attempt reconstructions of former Samoan coastlines at lower sea levels using available hydrographic maps for 'Upolu and Savai'i, plus the evidence from the Ferry Berth Lapita site (SU-MU-1), which indicated the depth of water to be removed. This exercise allowed him to redraw coastlines for sections of the islands and bays of Western Samoa. Although some of the resulting maps have since been used in teaching at Auckland, they do not give any information on the sources employed to make them or details of procedures involved in their construction. One of the most interesting, the map for the Mulifanua area, provided the inspiration for the present reconstruction, and led to a search of the map resources of the Department of Geography at the University for suitable hydrographic charts and identification of the one apparently used by the student.

The shoreline proposed here is based on a map entitled “South Pacific Ocean, Plans in Samoa”, Sheet 1339, London, and was published at the Admiralty on October 4th, 1963, under the Superintendence of Rear Admiral E. G. Irving, C.B., O.B.E., Hydrographer. One of the nine sectional maps on that sheet is of Mulifanua Harbour. The soundings are in fathoms, with those under 11 in fathoms and feet.

As can be seen in Figures 1B and 1C (after Jennings 1974:Figs 72 and 73; Green and Richards 1975:Fig.3), there was a mean sea level in 1973 about 1.5 m of water over a beach rock cemented coral crust which covered the entire turning basin and had to be blasted away in order to deepen it by dredging. The crust had an average thickness of .75 m and, just under it and in a lineal zone in one part of the turning basin, lay a black stained or muck-stained coral sand within which the Lapita sherds were concentrated (see Figure 1B). Some sherds and shells were also incorporated in the base of the beach rock crust itself. None of the sherds was water-rolled and Jennings (1974:177) considered that this, together with the fact that “wave action would also have spread, or entirely dissipated, the thick layer of organic stained midden that marked the old village site at the time of dredging”, meant that subsidence had been sudden. He did, however, concede that “a concentration of debris on a beach behind a shallow lagoon 800 m wide would not get the wave battering” he envisioned, as the force would be diminished by the reef, so that his point of rapid local subsidence was open to question. - 325 As will be seen in the reconstruction below, severe battering from wave action was, in fact, rather unlikely and, as he subsequently decided, gradual subsidence at 1.5 mm per annum is more probable (Jennings 1976:7). It seemed feasible, therefore, to trace a shoreline at a depth of approximately one fathom or 6ft (1.83 m) of water using the actual soundings on the base map. The results would probably not be very different

FIGURE 3. Reconstruction of coastline situation in vicinity of Ferry Berth Lapita site for 2.6 m change in sea-level relative to land.
- 326

using 2.4 m, 2.7 m, or 3 m, for example, except that the narrow channel to the wharf (and Lapita site) would be much shorter. Moreover, tracing a shoreline at a depth of 1.83 m below present mean sea-levels implies an actual level about 2.6 m below the present one, since at least 0.75 m of sediment has accumulated on top of the Lapita deposit in the last 3,000 years. Thus, we assume the reconstructed shoreline presented here represents a time when mean sea-level relative to adjacent land was somewhere between 2.6 and 3 m lower than today.

The reconstructed shoreline (Figure 3) is quite different from today's rather uniformly rounded coast which has resulted from the drowning of a locally flat landscape. The resulting map is similar to, but differs in detail from, that drawn by the student, which exhibited a completely enclosed lagoon embayment. However, both reconstructions suggest that the Lapita settlement was located on a former coral sand beach. Although the site was opposite a deep pass in the reef, as many Lapita sites were, it was not directly exposed to the open sea, but was situated at the inland end of a long shallow embayment (a former stream channel also responsible for the pass?) well back from this main access to the ocean itself. Moreover, two coral islets or exposed reefs protected that part of the sandy coastal beach from the direct wave action of the deeper open sea, forming a shallow lagoon between them and the shore.

It is conceded that the reconstruction does not make allowance for coral growth on the seaward edge of the reef over the last 3,000 years, and a greater amount of sediment infilling in some of the deeper water lagoon basins and channels, so that details for the locality at the time of the Lapita settlement are lacking. However, the underwater structure today is thought to be a sufficient reflection of the former situation to yield a general view of the environmental context at the time of Lapita settlement. The site was situated on an extensive sand beach frontage along a shallow lagoon with a metre of tidal fluctuation. It is less certain whether this Lapita settlement was at the time completely on dry land as some of the evidence would imply, or whether part of it extended out into the water as well, as is now known to be the case with other Lapita sites (Kirch 1988). We are inclined to the latter view.


New information about the published C14 determination on a shell from the Ferry Berth site at Mulifanua, on 'Upolu, Western Samoa, and possible conversions of it into a secular date, confirm that a true age of circa 3,000 years ago remains a preferred interpretation. Hydrographic and stratigraphic evidence suggests that the site was originally located on a sandy beach along the shore of an arm of a sheltered lagoon. Such a context would be far more in keeping with what is known of Lapita settlements than the present underwater situation. It seems that similar reconstructions of the former Samoan coastlines could help predict suitable additional places in which to look for other early sites. Two stone adzes have now been recovered from the dredge tailings at the site. Both are of known Lapita types, one typical of an early form of adze (Type V) made in local stone, and the other an imported rock whose shape is comparable to early Tongan and Fijian adzes with oval sections not normally found in Samoan collections. They add a new dimension to the early Eastern Lapita pottery - 327 assemblage already described from the site, one which as yet has not been identified elsewhere among the pottery bearing sites of the Samoan Group, although a number of such assemblages often with adzes, are now known from Tonga, 'Uvea, Futuna, and Fiji. The exotic early Lapita-type adze from Samoa is an indication that more of these will turn up there as collections increase.


Helen Leach would like to express her sincere thanks to Rhys Richards for his hospitality during her visit to Western Samoa and for permission to bring these adzes from his collection back to New Zealand for study. Thanks are also due to Dr Tony Reay of the Geology Department, University of Otago, for the hand-specimen identification of the rock types.

Roger Green would like to thank the now-unknown undergraduate student in the Oceanic Prehistory class who initially tackled the Samoan former sea-level problem with such promising results. It is hoped its formal presentation here will now encourage others to pursue the matter further.

  • Baquie, Barry, MS 1975. Report on Mulifanua Lagoon Pottery, Upolu. Class research report in possession of R. C. Green, University of Auckland.
  • Birks, L. and H., 1968. Adzes from Excavations at Sigatoka, Fiji. Archaeology and Physical Anthropology in Oceania, 3:105-15.
  • Buck, P. H., 1930. Samoan Material Culture. Honolulu, Bernice P. Bishop Museum, Bulletin 75.
  • Clark, J. T. and D.J. Herdrich, 1988. The Eastern Tutuila Archaeological Project 1986, Final Report. Prepared for Government of American Samoa Office of Historic Preservation, Department of Parks and Recreation, Pago Pago, American Samoa.
  • Dickinson, W. R., 1974. Temper Sands in Sherds from Mulifanua and Comparison with Similar Tempers at Vailele and Sasoa'a (Falefa), in Green and Davidson 1974 (q.v.). pp.179-80.
- 328
  • —— 1976. Mineralogy and Petrology of Sand Tempers in Sherds from the Ferry Berth Site, Paradise Site, and Jane's Camp, in Jennings et al. 1976 (q.v.). pp.99-103.
  • Green R. C., 1974a. Pottery from the Lagoon at Mulifanua, Upolu, in Green and Davidson 1974 (q.v.) pp.170-5.
  • —— 1974b. A Review of the Portable Artifacts from Western Samoa, in Green and Davidson 1974 (q.v.) pp.245-76.
  • —— 1974c. Excavation of the Prehistoric Occupations of Su-Sa-3, in Green and Davidson 1974 (q.v.). pp.108-54.
  • —— and J. M. Davidson, 1974. Archaeology in Western Samoa. Volume II. Auckland Institute and Museum, Bulletin 7.
  • —— and H. G. Richards, 1975. Lapita Pottery and a Lower Sea Level in Western Samoa. Pacific Science, 29:309-15.
  • Hewitt, N. J., 1980. Adzes and Basalt Artifacts, in J. D. Jennings and R. N. Holmer (eds), Archaeological Excavations in Western Samoa. Honolulu, Bernice P. Bishop Museum, Pacific Anthropological Records 32. pp.132-42.
  • Hunt, T. L. and P. V. Kirch 1988. An Archaeological Survey of the Manu'a Islands, American Samoa. Journal of the Polynesian Society, 97:153-83.
  • Jennings, J. D., 1974. The Ferry Berth Site, Mulifanua District, Upolu, in Green and Davidson 1974 (q.v.). pp.176-8.
  • —— 1976. Introduction, in Jennings et al. 1976 (q.v.) pp.1-9.
  • —— R. N. Holmer, J. C. Janetski and H. L. Smith, 1976. Excavations on Upolu, Western Samoa. Pacific Anthropological Records 25. Honolulu, Bernice P. Bishop Museum.
  • Kirch, P.V., 1978. The Lapitoid Period in Western Polynesia: Excavations and Survey in Niuatoputapu, Tonga. Journal of Field Archaeology, 5:1-13.
  • —— 1988. The Talepakemalai Lapita Site and Oceanic Prehistory. National Geographic Research, 4:328-42.
  • —— and D. E. Yen, 1982. Tikopia: The Prehistory and Ecology of a Polynesian Outlier. Honolulu, Bernice P. Bishop Museum, Bulletin 238.
  • —— and T. L. Hunt, 1988. The Spatial and Temporal Boundaries of Lapita, in P. V. Kirch and T. L. Hunt (eds), Archaeology of the Lapita Cultural Complex: a critical review. Seattle, Thomas Burke Memorial Washington State Museum, Research Report No. 5.
  • Poulsen, J., 1987. Early Tongan Prehistory: The Lapita Period on Tongatapu and its Relationships. 2 Vols. Terra Australis 12. Canberra, Department of Prehistory, Research School of Pacific Studies, Australian National University.
  • Pearson, G. W. and M. Stuiver, 1986. High-precision Calibration of the Radiocarbon Time Scale, 500-2500 BC. Radiocarbon, 28:839-62.
  • Smith, H. L., 1976a. The Principal Components Analysis of Samoan Ceramics, in Jennings et al. 1976 (q.v.). pp.83-95.
  • —— 1976b. Jane's Camp (Su-F1-1), in Jennings et al. 1976 (q.v.). pp.61-71.
  • Spennemann, D. H. R., 1986. Archaeological Fieldwork in Tonga 1985-86. Tongan Dark Ages Research Programme, Report No. 7. Canberra, Department of Prehistory, Research School of Pacific Studies, Australian National University.
- 329
  • Stuiver, M., G. W. Pearson and T. Braziunas, 1986. Radiocarbon Age Calibration of Marine Samples Back to 9000 cal yr B.P. Radiocarbon, 28:980-1021.
- 330

Wage, Trade, and Exchange in Melanesia

A Manus Society in the Modern State

James G. Carrier and Achsah H. Carrier

“Detailed, well documented, and logically argued.”

—Miriam Kahn, University of Washington

The authors' findings challenge the idea that colonial and Western-oriented encroachment leads to the decay of village societies or to their adopting Western values and practices. $35.00

The Heart of the Pearl Shell

The Mythological Dimension of Foi Sociality

James F. Weiner

This innovative study locates Foi social meaning in the re-creation and attempted solution of the moral dilemmas that are crystallized in mythology and other poetic usages.

Studies in Melanesian Anthropology, 5 $40.00


At bookstores or call toll-free 1-800-822-6657. Visa & MasterCard only.

University of California Press Berkeley 94720

1   Poulsen (1987:Table 83), following Polach's advice (Appendix 3, 1987:270-72), published the environmentally corrected New Zealand laboratory result of 2890±80 BP as equating to the calendrical age at one standard deviation of 1120 to 1255 B.C,. following the methods of Pearson and Stuiver (1986). The estimate published by Poulsen appears several hundred years too great, while the two of Kirch and Hunt (1988:22, Table 23) for 788 to 555 B.C. and 741 to 404 B.C. are certainly several hundred years too young. The latter have the error that old and new half-life values for the one sample have been given as separate dates and then calculated as if they were not already corrected for the Ocean Reservoir Effect. The results, therefore, do not meet with their assumptions (1) and (3).
2   Using the △R value for this area of ocean of 45±30 as recommended by Stuiver, Pearson and Braziunas (1986), 3251±150 would translate to 1280 B.C. to 840 B.C., given a onesigma date range. Using the more general mid-Pacific value of 100±24, as recommended by Kirch and Hunt (1988:23), the comparable result would be 1210 B.C. to 800 B.C. It would, therefore, appear that a true age of circa 3,000 years ago is as good an estimate for the age of the Ferry Berth Lapita site as is available to us at present.