Printer Friendly

Late Lapita occupation and its ceramic assemblage at the Sigatoka Sand Dune site, Fiji, and their place in Oceanic prehistory.

KEYWORDS: Archaeology, Lapita ceramics, Sigatoka site, Fiji

Abstract

The Sigatoka Sand Dune site on the island of Viti Levu, Fiji, has been continuously eroding over the past hall century. Large-scale excavations here in the mid-1960s documented a Late Lapita ceramic horizon, one significantly including many restorable or diagnostically complete vessels. A 1998 discovery of an intact segment of this deposit led to renewed excavation and the recovery of additional vessels. The results of this excavation are presented and used to redress long-standing problems associated with the occupation and interpretation of the ceramic assemblage. Based on radiocarbon dates and geoarchaeological analysis, it is known that the ceramics originally were deposited and rapidly buried on a back beach sand flat over a very brief period of time ca. 2500 BP (ca. 2600 cal. BP). The distribution of ceramics, the relative absence of habitation features, non-ceramic artifacts and fauna, and shoreline characteristics suggest the locale was employed as a canoe-landing site. The temporal discreteness, and the completeness of ceramic vessels have made the collection a critical assemblage for comparative analyses in Oceanic prehistory. With only limited decoration applied to the vessels, the assemblage has been interpreted as a devolutionary stage, one intermediate between the complex and highly structured decorated vessels of the initial Lapita colonists and a sequent Plainware phase where decoration has all but disappeared. Rather than a devolutionary stage, the Sigatoka pottery represents a viable industry with functional if not social importance in late Lapita society.

Introduction

Mid-1960s investigations by Lawrence Birks (1973) at the Sigatoka Sand Dune site on the Coral Coast of Viti Levu, Fiji (Figure 1), documented a unique archaeological complex for Oceanic prehistory. Eroding from the frontal face of this parabolic dune field near the Sigatoka River mouth was a series of three cultural horizons, each separated from the other by sterile layers of sand. Because each incorporated a ceramic assemblage related to different periods of the Fijian past, the site literally presented a layer-cake sequence from which detailed ceramic associations and chronology could be concisely define& For Oceanic archaeologists of the day, the earliest of these, what Birks called Level I, was of utmost importance. Level 1 ceramics were similar to types found in New Caledonia and Tonga, and they potentially were related to an ancestral population for the archipelago. And equally important, from widespread excavation across the Level 1 occupation floor, Birks was able to record a diverse range of associated vessel forms and sizes. This was not done from the normally minute-sized fragments that Oceanic archaeologists so frequently are presented with. Rather, rapid burial of the collection by wind borne sands preserved concentrations of large sherds in situ, allowing many of the vessels to be partially or fully refitted. The documentation and illustration of this assemblage serves as a landmark study for comparative ceramic analyses throughout the southwestern Pacific.

Level 1 ceramic vessels today define a late Lapita phase for Fijian prehistory. Dating to ca. 2500 BP (ca. 2600. cal BP), it occurs as a transitional assemblage between the highly decorated wares of initial Lapita colonists and a sequent Plainware phase where, presumably, decorative applications are no longer applied (Kirch 1997; Clark 1999; Best 2002). The assemblage, therefore, continues to be important for culture historical reasons as well as for shedding light on transformational processes of stylistic change in ceramic wares in Fiji. Survey of the Sigatoka Dune beach front in July 1998 discovered a newly exposed surface concentration of late Lapita ceramics just a short distance above high tide and in an area immediately west of Birks' excavation. Subsequent excavation of 85 m-" led to the recovery and refitting of a range of additional ceramic vessels as well as providing insight into the archaeological context in which they occur. In this paper we first report the results of this excavation and the ceramic assemblage that was recovered. Second, these data are used to revisit enigmatic and related issues for the late Lapita occupation at the site. And third, the long assumed position of the Sigatoka assemblage as an intermediate phase in the devolution of Lapita ceramics in Oceania is examined.

The Sigatoka Sand Dune site in context

The Sigatoka Sand Dune presently stretches westward along the Viti Levu coastline from the mouth of the Sigatoka River to Naqarai Bay for a distance of 4.8 km (Figure 1). The dune field is a massive geological feature, rising in areas to over 60 m elevation. Parent material for the dune is derived from eroded slopes of the central and upper Sigatoka River valley. Sediment is transported via the river to the ocean where east to west long-shore currents push it parallel to the delta margin. Because this flow impedes coral growth, the Sigatoka coastline has no fringing reef and is subject to heavy surf. Wave surge thus pushes the sediment plume back on shore after which southeast trade winds blow it obliquely inland to form the dune. Geomorphological and geoarchaeological studies by Dickinson et al. (1998) and de Biran (2001) indicate that extensive dune formation did not occur until after ca. 1400 BP.' It can only be hypothesized at present, but this event seems likely to correlate with enhanced erosion in the Sigatoka River valley as an expanding human population deforested bottomlands and slopes for agricultural planting (Dickinson et al. 1998:20-22).

That major development of the sand dune did not occur until after 1400 BP has obvious implications for an interpretation of prior settlement and use of the Sigatoka beachfront. The Sigatoka River is one of two principal drainage systems on the island of Viti Levu, serving historically as a pivotal 90 km long corridor for travel and trade into the interior highlands. Occurring immediately west of the divide between windward (wet) and leeward (dry) ecological zones, the river's course occurs dominantly within the latter. The Sigatoka delta plain is roughly triangular and bounded by steep hills and cliffs on both sides. Delta progradation began to occur after postglacial rise in global eustatic sea levels of ca. 7500 B.P (ibid.:3-4). Thus, at the time of first human settlement in Fiji, it was a substantial feature having built outward for a period of 4,500 years or more. On the delta surface inland of the present dune field on its eastern end are a series of low (1.5-3 in elevation) parallel sand ridges spaced 50 to 100 m apart with intervening swales (Figure 1). Because their orientation is sub-parallel to the present shoreline, in contrast to the obliquely trending dune, they are interpreted as accretionary beach ridges marking successive phases of delta growth. Over the entire period of human history at Sigatoka, the delta had but one primary channel flowing along its eastern margin. A second distributary to the west was in place about 5000 years ago but was blocked and transformed into a freshwater swamp with forest after that time (ibid.:l 4). We can but speculate on the former nature of delta ecology, particularly since different drainage patterns would facilitate different vegetation regimes. Undoubtedly it incorporated a patchwork blend of closed and open forest cover with delta margins supporting coastal plant communities (see Kirpatrick and Hassall 1981). Paleoecological and palynological investigations of the freshwater swamp have been conducted by Hope (1999) but detailed results are not yet available.

[FIGURE 1 OMITTED]

Pacific archaeologists have been long aware of eroding ceramics, human remains and other materials from the Sigatoka beachfront. As early as 1947, for example, Gifford (1951) designated the area as Site 20, and illustrated several surface collected ceramic specimens in his report on the archaeology of Fiji. It was Birks' project, nevertheless, that laid an interpretive foundation for all subsequent studies, and clearly outlined the chronology, complexity and importance of the Sigatoka archaeological landscape. His preliminary examination recorded "three separate cultural horizons" and it was his intention to " ... carry out on Site VL 16/1 [Sigatoka] an excavation as extensive as time and circumstance might permit, with the aim of obtaining all possible information, by means of artifactual material and otherwise, on the prehistoric occupation of the site" (Birks 1973:11). By Pacific standards then, and today, the ensuing excavation was a gargantuan effort. Over a total of 13 weeks in 1965 and 1966, Birk's teams excavated 281 "quadrants" of 10 foot by 10 foot size. This is an areal exposure of 2,610.6 m-" stretching along the Sigatoka beach for a distance in excess of 323 in (1060 ft). From these excavations, and with the geological consultation of Dickinson (1968), he associated each of the cultural horizons with a paleosol. Respectively from early to late (Levels 1-3), they included ceramics of the Sigatoka (Lapita), Navatu and Vuda phases of Fijian prehistory as then outlined by Green (1963). The excavations recovered tens of thousands of ceramic sherds, but ultimately proved enigmatic in the relative absence of faunal remains, nonceramic artifacts, and occupational features.

There have been numerous other studies at Sigatoka since Birks' excavation, most occasioned by salvage or planning needs of the site (Best 1989; Hudson 1994; Burley 1997; 2002; Burley and Shortland 1999; Marshall et al. 2000). These projects cumulatively have clarified several problems related to site composition, chronology, and settlement configuration (see Burley 2003). This has not been the case for Level 1, however. Except for the occasional collection of Late Lapita pottery sherds from the surface of limited rescue excavation (Marshall et al. 2000:105-109), additional data or insight on the occupation horizon have been non existent. In fact after the Sigatoka beach was heavily scoured by high surf during Cyclone Kina in 1993, any remaining Level 1 deposits along the dune front were assumed largely destroyed (Burley 1997:18; Dickinson et al. 1998:6). The discovery of the Level 1 surface exposure in 1998 thus afforded an opportunity not only to recover additional ceramic vessels for restoration, but to further document the structure, context, and age of these remains.

Level 1 re-exposed: the 1998 and 2000 project

Birks did not relate his grid to a permanent datum, and we can only estimate its proximate position on the dune front today from his maps and photographs. With respect to the Level 1 horizon specifically, 107 quadrants accounting for an areal extent of 994 [m.sup.2] were completed over an east/west linear distance of 152.4 m (500 ft) (Birks 1973:12). Dense concentrations of ceramics occurred on both western and eastern ends of his excavation while a central area of 36.5 m (120 ft) had "no occupational debris" and "was excavated only in selected areas, to put this question beyond doubt" (ibid.). Late Lapita surface ceramics discovered in 1998 are thought to occur immediately to the west of Birks' excavation extent (Figure 2). These were at an elevation of 1 in above and no more than 10 to 12 m distant from the extreme high-tide line (Figure 3). A Level 2 paleosol exposure was present 2.5-3 m higher in elevation and 15 in upslope and inland. Deflation of this deposit had resulted in the mixing of Level 1 and 2 sherds on the exposed surface. A 1992 survey project by Marshall et al. (2000:109) also recorded Level 1 rim forms here. (2)

[FIGURES 2-3 OMITTED]

Excavation proceeded in 1998 from the lower exposed surface into the dune face employing a provenience grid of 1 x 1 in units (Figure 2). In total 77 [m.sup.2] was completed in addition to a 0.7 x 15 m test trench extending eastward from the perimeter of the excavation block. An additional 8 [m.sup.2] was excavated in 2000, largely to complete the exposure of a feature found on the final day of the 1998 project, and to assess impact to the inland face of the site from wave attack during a 1999 tropical storm. In both field seasons, ceramics were exposed in situ, mapped, and photographed. From its surface exposure on the south, the occupation floor was found to run on a horizontal plane directly into the dune (Figure 4). On the upslope face of the excavation, it occurred at a depth of 70 cm below surface. As excavations advanced northward into the dune, Level 1 ceramics were found in situ while Level 2 sherds continued to be recovered as deflated material in drift sand surface deposits.

[FIGURE 4 OMITTED]

Birks' (1973:12-13) illustration of Level 1 stratigraphy portrays it as a distinctive layer of "grey brown sand" on average 11 inches (28 cm) thick. However, his description of the Level 1 stratum, and its characterization as a paleosol, are less definitive. There were no detectable structural differences in sediments above, below and within the stratum, and ".... the upper and lower termination of Level 1 were difficult to define exactly owing to the fine shading of colour". It was, in fact, " ... knowledge gained by experience [that] enabled excavation to proceed without serious hindrance" (ibid.:9, 12). The more recent excavations similarly lacked evidence to differentiate the Level 1 ceramic floor from sediment regimes immediately above or below. All three deposits are dark brown in colour (when moist) and without clearly defined stratigraphic breaks. Each represents a well sorted-beach sand with high contents of heavy minerals (ferromagnesian silicates and oxides) indicative of concentration by placering. Petrographic examination found no generic differences in the three, nor textural differences that could be ascribed to weathering (Dickinson 2003). In some cases in the field, it was noted that the pottery bearing layer seemed to have a slightly higher silt content and, under ideal light and moisture conditions, a very slightly darker color. As a result, the layer was estimated to be between 15 and 20 cm thick, but it could be identified securely only when ceramics were present. Level 1 is more appropriately defined as a back beach surface sand rather than a paleosol. The slightly higher silt content may have been a consequence of coastal vegetation cover and/or human presence, but evidence for pedogenesis is absent.

The excavation of the Level 1 surface recorded both scattered ceramic sherds and discrete clusters associated with individual vessels (Figure 5). The latter indicates primary discard where fragmented ceramics were left in situ near or at the place of vessel breakage. In one case two large sherds from different vessels were intentionally placed one inside the other, also suggesting a secondary and intentional deposition. Despite the concentrations, there was nothing evident in overall distribution to reflect upon activity patterning. Faunal remains, including shellfish, were absent while the non-ceramic artifact assemblage was limited to a single adze. A 1 x 1 m rock scatter in association with charcoal stained sand and charcoal flecks on the western side of the excavation block is interpreted as a concentration of heating stones that had been cleaned from an earth oven (Figure 6). Given the position of this feature, it was assumed that the earth oven occurred to the southwest, in an area destroyed by wind scour prior to 1998. Birks' interpretations of the site also were frustrated by the general absence of fauna, non-ceramic artifacts, and features. Over the extent of his 994 [m.sup2] excavation in Level 1, he recovered 15 nonceramic specimens, the majority being adzes or tools related to their production or maintenance (grinding stones) (Birks 1973:49). As for habitation features, these were limited to three small post holes and " ... three places where tires had burned, but whether they were fireplaces in the domestic sense, or simply the sites of rubbish tires, there was insufficient evidence to prove (ibid.: 14)". A general absence of fauna was assumed to be a function of preservation conditions (ibid.:15), a circumstance found not to be the case in more recent investigations elsewhere on the dune (Burley 2002).

[FIGURE 5 OMITTED]

During the 1960s excavation and the more recent one, scattered flecks of charcoal were found widely distributed throughout Level 1 deposits. From the "bottom 6 inches of the occupation layer" in one of his units, Birks collected a composite sample for a radiocarbon date, this providing a measurement of 2460 [+ or -] 90 BP (GaK 946, Birks 1973:57). Major advances in radiocarbon dating through use of accelerator mass spectrometry (AMS) now allow very small charcoal samples to be dated precisely. Two such samples in presumed association with Level 1 ceramics were submitted for AMS dating in 1998 while a third, in direct association with the stone scatter feature was submitted in 2000. The results are indistinguishable from each other, or from the earlier date obtained by Birks (Table 1). As a group, and including the date obtained by Birks, they have a radiocarbon age weighted average of 2491 [+ or -] 25 BP. Whatever nature the Level 1 occupation may be, it occurred as a single short-term event, or a series of closely timed events. The fact that the ceramics are neither trampled nor degraded from surface exposure also indicates a rapid and complete burial shortly thereafter.

Late Lapita Level 1 ceramics

Birks (1973:54) recovered a total of 48.85 kg (109.9 lbs) of ceramics from his excavation of the Level 1 floor. Because of the general condition, size and concentrated distributions of these sherds, many could be cross-mended to form near complete vessels, or at least specimens where a full range of attributes could be ascertained. Analysis and typology consequently focused at the level of the vessel, rather than the individual sherd. Birks' report presented data on a total assemblage of 87 diagnostically complete ceramic vessels. Also present were partial vessels, pot lids, pot rests and an assemblage of ceramic disks, the latter lacking a functional interpretation. The vessel form typology applied to the collection was largely functional, differentiating open-mouth jars (Type 1) from bowls (Type 2) from narrow-neck water jars (Type 3). Within each, stylistic variations were defined as sub-types, some based on minor differences with others being more substantial. To analyze the ceramics recovered in 1998 and 2000 we similarly applied Birks' typology. A complete range of diagnostic and quantitative attributes could be ascertained or projected for 22 vessels, with 20 of these corresponding to existing types (Table 2, Appendix). Nine have greater than 50% of the vessel present (Figures 7 and 8). A minimum of 47 other less complete vessels occurs in the assemblage based on comparative analyses of rim sherd groups. Two diagnostically complete pot rests were recovered in the more recent project as well.

[FIGURES 7-8 OMITTED]

Of the diagnostically complete vessels, ten are open-mouth jars of varying sizes. The largest dimension on these vessels occurs at the shoulder, and shoulder diameter normally exceeds vessel height. All specimens with rounded shoulders (Types la, lb, lc) have a straight to slightly inverted expanded rim, giving them the distinct appearance of a collar (Figure 7). Rim thickening largely has been accomplished through the addition of a separate band of clay to the outside surface. Open-mouth jars with carinated shoulders (Type I d) may have this type of rim treatment, but they also include vessels with strongly everted rims and flat to rounded lips (Figure 7). The remainder of the diagnostically complete specimens includes everted rim bowls (Type 2a), a rounded shoulder inverted rim bowl (Type 2d), a sharply carinated inverted rim bowl (Type 2e), a narrow-neck water jar (Type 3a), and two vessels that fall outside of the existing typology but are interpreted to be bowl variants (Type 2 other). One of the latter is a near complete specimen with slightly rounded base, vertical walls without shoulder, and ah expanded rim (Figure 7). The other is a flat-bottomed dish with straight vertical sides and a basal scar indicative of having legs (Figure 8). This also is the only example of a slipped vessel, it having red clay wash added to both its interior and exterior surface. The pot rests are comparable to those recovered by B irks (ibid.:125-126), including one with a fin-like form and another with knob and bar projections built on a cylindrical base (Figure 8). How exactly these stands were employed remains debated (ibid.:37; also Poulsen 1987:134; Dye 1988:187).

The Level 1 assemblage includes both plain and decorated vessels, with open-mouth jars dominating the latter. Decoration is highly restricted, almost always occurring on the lip and/or rim. Decorative applications include incised arcs, v-type excised notches as well as ovoid, vertical and diagonal impressed notches. A grouping of three finger pinches has been applied to the upper shoulder of the only narrow neck jar recovered in 1998 (Figure 8). In so far as Birks (1973:101) found an identical mark on the neck of ah open-mouth jar, this may represent a potter's mark rather than a decorative application. With one exception, decorated vessels recovered by Birks and in more recent investigations are similar both in type of application and placement. The exception, however, is a notable one. Birks (ibid.:113) recovered four carinated bowls (Type 2e) on which Eastern Lapita design motifs were applied with dentate stamps above the shoulder angle. A dentate stamp also was used to apply spaced vertical lines on an open-mouth jar with expanded rim (ibid.:89). It was this use of dentate stamping that allowed Birks and others ultimately to associate the assemblage with the Lapita cultural complex.

The Late Lapita ceramic assemblage from Sigatoka is almost entirely a hand built, slab-moulded earthenware with variability between vessels in body fabric thickness and consistency. Birks (ibid.:21) reports one potential ring-coiled open-mouth jar, while the addition of a coil to create the expanded rim form was common. Evidence for use of a paddle and anvil technique to strengthen welds is absent, but it is apparent that the potters used both soft and textured materials to smooth the vessel surface prior to firing. Firing was done in low temperature open-air hearths. Dickinson (1973) previously conducted a petrographic analysis of temper sands for 20 sherds recovered by Birks from Level 1. All but three of these contained tempers of volcaniclastic orogen type derived from local sources. The exceptions were similar, but temper composition suggested an origin from the Koroimavua group formation farther north on western Viti Levu (Dickinson 1973:69). Ten body sherds from the more recent excavations also were subjected to petrographic analysis (Dickinson 1999). Nine of these had temper sands of local derivation with point count ranges for individual constituents highly similar to modern dune sands (Table 3). The tenth sherd contained a fine-grained, pyroxene rich beach sand of unknown but exclusively volcanic derivation. Pyroxenic placer tempers are common in Fiji, but the closest place from which this sand could derive is the Navua Delta on the south coast of Viti Levu.

The Late Lapita presence at Sigatoka in local perspective

Since initial excavation in the mid-1960s, the Level 1 archaeological remains at Sigatoka have proven enigmatic for contextual interpretation. Birks recovered abundant concentrations of pottery over a considerable distance along the beachfront. However, little else was present to provide insight or explanation for this material. Architectural features are absent, hearths are limited to three putative examples, non-ceramic artifacts are highly limited in numbers, and fauna is non-existent. Birks (1973:61-63) accordingly reviewed nine potential types of sites that might account for this pattern. None proved wholly satisfactory, but he did proffer a use of the coast for "turtle-catching", where seasonal "campers" awaited "the arrival of shore-sojourning turtles" (ibid.:63). The beach ideally could have supported a nesting turtle population, yet the absence of turtle bone, butchering tools, cooking hearths, or other evidence associated with a seasonal camp seems telling. More recent excavations document the additional presence of an earth oven feature, but provide little new archaeological data to resolve the question of site function.

In light of what is missing in the archaeological record at Sigatoka, it is indeed difficult to understand or explain what is present. Recent finds of Level 1 pottery illustrate the assemblage to be distributed over a distance in excess of 500 m (Burley 2003) with identifiable concentration areas suggestive of some form of specialized activity. Since hearths for firing this pottery are absent, as is other evidence for ceramic production (wasters, unfired clay), the pottery must have been imported to these locales, not manufactured in situ. The ceramic collection as a whole, nevertheless, represents a fully mixed and diverse range of functional types, one more typical of habitation than an assemblage destined for a specialized function. Open-mouth jars of different sizes could be employed for cooking, transport of food or storage, bowls could be used for food preparation or presentation, while the narrow neck jars suggests freshwater storage and transport. Add in the pot stands, lids and pottery disks, and the assemblage includes an impressive range of ceramic wares produced by Late Lapita potters.

Geoarchaeological interpretation indicates that, whatever was going on during Level 1, it was taking place on the sandy flat of a back beach coastal strip intervening between a high energy open shoreline and the fringe of the delta plain. A storm ridge or berm capped by a low eolian dune may partially have protected this setting, as is the case on lengthy sections of the Sigatoka coast today (Figure 9) (Dickinson et al. 1998:13). However, the occurrence of a layer of beach pumice between Levels 1 and 2 illustrates that the site was still within reach of storm waves at Level 1 time (ibid.:12). The abruptness of the burial of Level 1 ceramics by wind blown sand also implies a high level of exposure to the southeast trade winds. Anyone who has spent an afternoon on the Sigatoka coast in the face of these winds and drifting sand will appreciate just how prohibitive they can be for habitation, even of the most temporary nature (see Birks 1973:12). Yet we also expect that the mouth of the Sigatoka River was an important landmark for Late Lapita peoples. It was the gateway for a travel corridor into the Sigatoka valley and interior. It provided a unique circumstance where select freshwater and marine resources could be exploited from a single locale. And while the ecology of the delta plain to its back can only be speculated upon, a mixed forest and swampland no doubt provided opportunities for natural resource exploitation that would be difficult to match in other venues. A Late Lapita settlement may not have been located on the beach-front per se, but we can hypothesize its presence somewhere in the adjacent vicinity.

[FIGURE 9 OMITTED]

The varied circumstances for Level 1 leave few potential activities that might account for such a pattern in any type of parsimonious fashion. The one we believe most likely is that of a maritime landing site, where the shoreline was intensively employed for the beaching of canoes over a short period of time. The loading of pottery and other goods on or off these vessels at different locales along the beach would lead to incidental breakage, and the fragments quickly became buried by drifting sand. As a canoe-landing beach, the Sigatoka delta margin seems a suitable locale. The absence of an off-shore reef provided access to the coast, the presence of a long sandy shoreline allowed beaching without impediment, and the prevailing trade winds and long-shore currents, once understood, facilitate landings and departures. An earlier cutbank of the Sigatoka River crosses the field behind the eastern end of the dune suggesting the river channel may have been wider and farther to the west than presently is the case (see Figure 1). The landing area thus could have bridged the open delta front surf beach with a river-flank beach of calmer persuasion. We recognize that 1-2 m high breaking surf potentially serves as an impediment, but to what degree this actually was the case requires a far greater understanding of Lapita maritime technology and expertise than presently exists. In several parts of Oceania today this type of surf is regularly faced on islands without protective fringing reefs, and we have no reason to believe that it would be an insurmountable barrier for Late Lapita voyagers.

The archaeological record, as scant as it may be for contextual data, is coincidental with, if not complementary to, our interpretation of the Sigatoka beach as a canoe-landing site during Level 1. The beaching of several canoes, or a limited number of canoes at different times could form the concentrated distributions of ceramics at dispersed locations along the beach. The few archaeological features that are present are interpretable as subsidiary activities expected to take place on shore during these activities. The small non-ceramic artifact collection dominated by adzes or adze-sharpening implements is in keeping with woodworking tasks and vessel repair. And the functional diversity of Level 1 ceramic types, particularly for storage, is highly appropriate to sea-going voyages. Birks' identification of carbonized remains adhering to the interior of several of the open-mouth jars further suggests the transport of prepared foods. Finally, and not insignificantly, the presence of ceramic vessels with tempers from both the northwest and south of Viti Levu illustrates ceramic exchange or transfer in keeping with coastal voyages and inter-community relations.

Transformations in Lapita design and Sigatoka ceramics

The distinctive suite of dentate stamped and other decorated ceramics that archaeologists define as the Lapita cultural complex demarcates the spread of Austronesian-speaking seafarers across Oceania (Kirch 1997; Spriggs 1997; Summerhayes 2000). From a staging point somewhere in the Bismarck archipelago, this group ultimately migrated and settled as far west as Tonga and Samoa in the interval 3200--2800 cal BP. Over time, and as the expansion took place, Lapita decorated ceramic wares underwent a process of "devolution" (after Best 2002:62). This is witnessed in the loss of specific types of vessels, in the increasing abstraction and simplification of decorative motifs, and in the way that these designs were then applied (see Kirch 1997; Sand 2001 ; Burley et al. 2002; Best 2002). The reason behind this transition is not easily understood, but most recent treatments of the subject relate it to larger changes in the ritual or religious realms of Lapita society (Kirch 1997; Spriggs 2002; Best 2002). In the Bismarck Archipelago, decoration on the earliest Lapita wares is elaborate, sometimes having a central frieze upon which an anthropomorphic face motif has been applied (see Spriggs 1990; Kirch 1997). Because of the latter, the pots have been associated with representations of rituals involving ancestors (Kirch 1997:143, 160). Best (2002:63), in fact, goes so far as to argue that the pots were fabricated to apply these designs, rather than the designs being configured as decoration for the pots (also Marshall et al. 2000:89). As religious practices presumably became transformed, so too did the meaning of the designs and the social values attached to the pots. This initiated a progressive process of stylization and degradation resulting ultimately in the complete disappearance of Lapita decorated wares in New Caledonia, Fiji, and the archipelagoes of western Polynesia. In the latter it also presaged the loss of pottery altogether. Few archaeologists have concrete data to track the phases by which this progression took place. Against this background, Level 1 at Sigatoka, according to Best (2002:81), is one of the very "few discrete and short-period" assemblages containing "the end of the Lapita decoration sequence". It provides a rare opportunity to document ceramic change in Lapita generally, and Fiji specifically.

The earliest Lapita period of Fiji is poorly dated (Anderson and Clark 1999), and not fully understood either for its diagnostic ceramic types or for settlement expansion across the archipelago. Some of the most complete data come from the island of Lakeba, in Lau, from which Best (2002) has developed an elaborate model of stylistic change in Lapita ceramics in keeping with its postulated role in Lapita religious beliefs. Within this model, a simplification and loss of Lapita design motifs is far more than a loss of meaning, it has additional implications for transformations in vessel form and eventually manufacturing expertise. To paraphrase Best (2002:63), since early Lapita ceramic decoration dictated the shape of the pot, particularly including flat out-turned lips and carinations as surfaces for the design, the vessel shapes became "plainer" as the designs became simplified and stylized. Thus, with the complete disappearance of the Lapita design system, there would be major and parallel changes in vessel forms, presumably including loss of carination and jars with everted rims. In Lau, Tonga and Samoa, this does appear to be the case, where only a very small number of simplified vessel types are retained in the Polynesian Plainware phase (see Green 1979; Kirch 1988; Dye 1996; Burley 1998). That this was occurring in western Fiji at Sigatoka is not so apparent (see Burley 2002, 2003).

The Level 1 assemblage at Sigatoka can be related to the Lapita ceramic series without question. Most notable in this respect are the four carinated bowls with dentate stamped motifs above the shoulder. Equally important, however, is the fact that each of the vessel forms incorporated in Birk's typology similarly occurs in early Lapita assemblages elsewhere, some with decoration and some without (see Green 1979; Kirch 1988; Best 2002; Burley et al. 2002). Expanded rims, rim notching, and other applications also are present earlier, often applied to vessel types all but identical to those of Level 1. And in all of these respects, a direct and uninterrupted cultural continuity from Lapita can be concluded. The principal and critical difference between Level 1 and what we know of the earlier Lapita ceramic series is the almost complete abandonment of the highly structured decorative system in which design motifs were applied with dentate stamp, roulette, or incision. In point of fact, the four small carinated bowls recovered at Sigatoka by Birks may represent the last vestiges of this decoration as it was being applied in Fiji (also see Best 2002:81)(3). Abandonment of the Lapita decorative design system in Fiji further occasioned the loss of a small number of highly specific vessel forms upon which it was most complexly applied. Included here are a flat bottom bowl with everted sides and flaring rim, a wide mouth bowl or dish, also with flaring rim and shallow rounded base, and a ring-footed vessel form (see Mead el al. 1975; Green 1979; Marshall et al. 2000; Best 2002). These may have been employed as presentation vessels, and their disappearance well supports interpretations related to changing rituals or social protocols. The data, unfortunately, are insufficient to measure how gradual or abrupt this transformation may have been.

With its continuity in many of the earlier vessel form styles and at least some of the earlier decorative elements, the Sigatoka assemblage presents an intriguing transitory phase between early Lapita and later Plainware ceramics in Oceanic prehistory. In this we must admit that the complete abandonment of traditional Lapita design and structure is difficult to explain or understand without reverting to epiphenomenal or metaphysical inference. To engage in such interpretive discourse requires a thorough documentation and understanding of what carne before, what elements were lost, and how the latter conveyed meaning within the Lapita cultural system. In this we can but defer to colleagues who are better informed by their data (i.e. Kirch 1997; Spriggs 2002; Best 2002). What we can say with certainty about Lapita ceramic transformation on western Viti Levu is that, even without complex decorative designs, the Sigatoka assemblage represents a highly viable industry, with skilled potters who took pride in their work (see Figures 7 and 8). It does not appear as some later phase in a devolutionary spiral ultimately leading toward the abandonment of ceramics, as Best (2002:62-65) has proposed. Vessel types closely match long-established and acceptable cultural templates, including forms with carinations and everted rims. Pot lids, a variety of pot rests, and limited new vessel forms, including a flat-bottomed type with legs and slip, bespeak of ceramic innovation, not degradation. And most importantly in all of these respects, recent excavations of Fijian Plainware ceramics in the Level 2 deposit at Sigatoka illustrate a maintenance and expansion of ceramic diversity over the ensuing millennium (Burley 2002). This notably includes the retention of some Level 1 vessel forms as well as a continued use of rim notching, carination and other features.

The Sigatoka Level 1 ceramic assemblage has become a unique resource for Oceanic prehistorians. It includes a range of diagnostically complete and restorable vessels. It provides detailed insight into Lapita vessel forms and nuances, and it serves as a highly visual catalogue for comparative purpose. Comparison to late Lapita and sequent Plainware assemblages in other archipelagoes of Melanesia and western Polynesia do find overlap in vessel forms (see Green 1974; Best 1984; Poulsen 1987; Kirch 1988; Bedford 2000), as anticipated given a common ancestry for the ceramic industry. At the same time, there are few of these later assemblages that have the full range of types or the diversity of types that continues to persist at Sigatoka. We must necessarily conclude, then, that the Sigatoka assemblage is not a pan-Oceanic phase through which Lapita ceramic change was destined to pass. Rather, it is more likely a localized transformation that was subject to economic and social processes acting on one community of potters in western Viti Levu. The abandonment of Lapita decoration ultimately may have resulted from changing religious beliefs, but the Sigatoka ceramic assemblage continued to serve important functional and social roles in that community. Not insignificantly those roles appear to have continued onward into the historic era with villages in the Sigatoka Valley being important centres for ceramic production in western Fiji (Palmer et al. 1968).

Conclusion

In reporting upon our excavations at Sigatoka in 1998 and 2000, the objectives of this paper have been three-fold. First the intent was to present a review of the project and its recoveries. Importantly this included the collection of an additional 22 ceramic vessels and two pot stands that, in a diagnostic sense, are complete. While most could be related to the vessel form typology derived from the earlier analysis of Birks, two new types also were recovered. The second objective was to use these data to reassess the nature and existing interpretations for the Level 1 occupation. Despite an abundance of pottery scattered over a considerable distance along the shoreline, non-ceramic artifacts, fauna and archaeological features indicative of habitation were all but non-existent in earlier work. Again this was found to be the case in the more recent investigations. In consideration of this context, and given the diversity of the ceramic assemblage per se, we have interpreted the site as a canoe-landing beach. Ceramic vessels were broken during loading or unloading activities and then quickly became buried in blowing beach sand. Finally, the third goal was to assess the veracity of Sigatoka ceramic forms as a transitional phase between early Lapita decorated wares and later Plainware ones. In this latter respect, they have been proposed by Best as a stage in the devolution of the Lapita ceramic industry as a whole. The abandonment of Lapita decorative design in Fiji does parallel changes elsewhere in Melanesia and western Polynesia. Yet at the same time, ceramics continued to play an important functional role in late Lapita society in western Viti Levu, as witnessed in the retention of vessel form diversity, innovation, and technological expertise. On a different trajectory from many other areas, we accordingly propose it to be a local or regional phase without necessary consequence for the interpretation of ceramic change beyond western Fiji.

Despite the more recent excavations at Sigatoka and our reinterpretation of different aspects of the Level 1 occupation, there continue to be many unresolved issues. If the site in fact is a canoe-landing locale as we propose, why then was it such a short-term event? Is the Sigatoka Level 1 collection representative of the full range of ceramic variability in a Late Lapita phase, or is it a specialized assemblage linked to transport, storage and use on sea-going voyages? To what extent is the ware produced by one community of potters at Sigatoka representative of other communities of potters on western Viti Levu, or Fiji more generally? And above all else, we must query the location of a village to which this beachfront activity and assemblage can be related. Without that knowledge, a substantiation of the site as a canoe-landing beach remains all but impossible. Unfortunately, yet opportunistically, the various geological and other processes that led to our discovery of the 1998 exposure of Level 1 continue unabated. Now monitored by archaeologists with the Fijian government, we can but hope that future discoveries at this quite remarkable site will provide the remaining data needed to fill in Late Lapita prehistory at the mouth of the Sigatoka River.

Appendix A Sigatoka Level 1 Ceramic Vessel Dimensions and Typological Descriptions

Vessel Types 1a, 1b, 1e--Subglobular jar forms with everted rims, rounded shoulders and well defined neck. Birks (1973: 21) classification into Types a, b and c emphasize minor variations that are not quantitatively defined. Expanded (collar-like) rim forms predominate in the Birks' assemblage and here.

Vessel 1 (Figure 7)--Rim is expanded, straight to slightly inverted and with incised arc decoration on exterior surface. Rim interior diameter = 22.5 cm, rim exterior diameter = 23.9 cm, shoulder diameter = 29.9 cm, shoulder to rim height = 11 cm, vessel height = 17 cm.

Vessel 2 (Figure 7)--Rim is expanded, slightly inverted and with impressed notch decoration on lower exterior edge. Rim interior diameter = 16 cm, rim exterior diameter = 17.8 cm, shoulder diameter = 22 cm, shoulder to rim height = 8 cm, vessel height = 15.5 cm.

Vessel 19--Rim is expanded, straight to slightly inverted and with impressed notch decoration on lower exterior edge. Rim interior diameter = 18.6 cm projected (pr), rim exterior diameter = 20.2 cm (pr), shoulder diameter = 22.6 cm (pr), shoulder to rim height = 8.2 cm, vessel height = 15.6 cm (pr).

Vessel 22--Rim is expanded, straight to slightly inverted and with impressed notch decoration on lower exterior edge. Rim interior diameter = 14.8 cm (pr), rim exterior diameter = 16.2 cm (pr), shoulder diameter = 16.4 cm (pr), shoulder to rim height = 6.4 cm, vessel height = 10.8 cm (pr).

Vessel 23--Rim is expanded, straight to slightly inverted and undecorated. Rim interior diameter = 15.0 cm (pr), rim exterior diameter = 16.2 cm (pr), shoulder diameter = 19 cm (pr), shoulder to rim height = 5.6 cm, vessel height = 11 cm (pr).

Vessel Type 1d--Everted rim jars with well defined carinated shoulders.

Vessel 3 (Figure 7)--Rim is everted and without decoration . Rim interior diameter = 15 cm (pr), rim exterior diameter = 18.8 cm (pr), shoulder diameter = 20.2 cm (pr), shoulder carination to rim height = 4.8 cm, vessel height = 10 cm.

Vessel 4 (Figure 7)--Rim is everted and with vertical incision decoration on exterior lip. Rim interior diameter = 20.4 cm (pr), rim exterior diameter = 26.4 cm (pr), shoulder diameter = 24.8 cm (pr), shoulder carination to rim height = 8.5 cm, vessel height = 15.2 cm (pr.).

Vessel 8 (Figure 7)--Rim is everted and without decoration. Rim interior diameter = 20 cm, rim exterior diameter = 25.5 cm, shoulder diameter = 26 cm (pr), shoulder carination to rim height = 8.2 cm, vessel height = 18.4 cm.

Vessel 14--Rim is everted and without decoration. Rim interior diameter = 17.2 cm (pr), rim exterior diameter = 19.6 cm (pr), shoulder diameter = 18 cm (pr), shoulder carination to rim height = 4.4 cm, vessel height = 10 cm (pr).

Vessel 20--Rim is expanded, straight to slightly inverted and with impressed notches on upper lip. Rim interior diameter = 11.2 cm (pr), rim exterior diameter = 12 cm (pr), shoulder diameter = 13.6 cm (pr), shoulder carination to rim height = 5.6 cm, vessel height = 10 cm (pr)

Vessel Type 2a--Shallow everted rim bowls ranging from cup size to platter sizes.

Vessel 9 (Figure 8)--Interior rim diameter = 36 cm, vessel height = 8.9 cm. Vessel is undecorated.

Vessel 10--Interior rim diameter = 27 cm (pr), vessel height = 4.8 cm (pr). Vessel is undecorated

Vessel 13--Interior rim diameter = 30 cm (pr), vessel height = 5 cm (pr). Vessel is undecorated

Vessel 16--Interior rim diameter = 19.6 cm (pr), vessel height = 6 cm (pr). Vessel is undecorated

Vessel 17 (Figure 8)--Interior rim diameter = 12.8 cm (pr), vessel height = 4.6 cm. Vessel is undecorated

Vessel 18--Interior rim diameter = 27 cm (pr), vessel height = 6 cm (pr). Vessel is undecorated

Vessel 24--Interior rim diameter = 37 cm (pr), vessel height = 10 cm (pr). Vessel has impressed notches on exterior lip surface.

Vessel Type 2d--Deep inverted rim globular bowl

Vessel 12--Interior rim diameter = 24.4 cm (pr), shoulder diameter = 30.8 cm (pr), vessel height = 14.2 cm (pr). Vessel is undecorated.

Vessel Type 2e--Bowl with carinated shoulders.

Vessel 11 (Figure 8)--Interior rim diameter = 24.4 cm, carinated shoulder diameter = 30.8 cm, vessel height = 14.2 cm. Vessel is undecorated.

Vessel Type 2 other--Bowls that are undefined in the Birks Typology. Vessel 5 might alternatively be classified as a jar form.

Vessel 5 (Figure 7)--Rim is expanded, slightly everted and with impressed notches on lower exterior edge. Rim interior diameter = 21.6 cm, rim exterior diameter = 23 cm, vessel height = 17.9 cm, maximum basal diameter = 20 cm.

Vessel 6 (Figure 8)--Flat bottomed bowl with a minimum of three feet. Maximum vessel base diameter = 32 cm (pr), foot diameter at base join = 6 cm (pr).

Vessel Type 3a--Vessels with narrow orifice, well-defined constricted neck, ovoid body and without handles.

Vessel 7 (Figure 8)--Rim is slightly everted. Interior rim diameter = 8.1 cm, shoulder diameter = 24 cm (pr), shoulder to rim height = 11.9 cm, vessel height = 21.4 cm. The vessel is decorated with three adjoining finger impressions on the shoulder.

Ceramic Pot Stands--

Non Vessel 25 (Figure 8)--Birks Non Vessel Type C (subconical with concave projections), Maximum base diameter = 17 cm (pr), stand height = 16.9 cm.

Non Vessel 26 (Figure 8)--Birks Non Vessel Type B (subconical with solid projections), Maximum base diameter = 15.1 cm, stand height = 14.6 cm.

Acknowledgements

Funding for fieldwork and analysis was provided by Simon Fraser University IESS (1998, 2000), Simon Fraser University/Social Sciences and Humanities Research Council Program (1998), and National Geographic Scientific, Exploration and Research Committee (2000). The 1998 and 2000 excavations were undertaken as one component of a field school in archaeological field techniques offered by Simon Fraser University in cooperation with the Fiji Museum and the National Trust for Fiji. We thank the myriad of students, volunteers and local fieldworkers who worked at Sigatoka in both project years. Fiji Museum archaeologists Sepeti Matararaba and Jone Balenaivalu were instrumental participants in fieldwork, while then Sand Dune National Park warden Tui Komaimua provided logistical support. Robert Shortland served as field supervisor in 1998, assisted in the original refitting of ceramic vessels, and drew most of the vessel form profiles included here. Lori White served as lab director in 2000 and was responsible for the final restoration of the pots and their return to the Fiji Museum for exhibition. Alice Story and Jessi Witt are gratefully acknowledged for their expertise in preparing the publication figures while Greg Ehlers took the photographs that are integrated into Figures 7 and 8. To the many unnamed others who also helped with this project, we sincerely offer our thanks.

References

Anderson, A. and Clark, G. 1999. The age of Lapita settlement in Fiji. Archaeology in Oceania 34:31-39.

Bedford, S. 2000. Pieces of the Vanuatu Puzzle: Archaeology of the North, South and Central. Unpublished PhD Dissertation, Australian National University, Canberra.

Best, S. 1984. Lakeba: The Prehistory of a Fijian Island. Unpublished PhD Dissertation, University of Auckland, Auckland.

Best, S. 1989. The Sigatoka Dune Burials. Unpublished report on file, Fiji Museum, Suva.

Best, S. 2002, Lapita: A View From the East. New Zealand Archaeological Association Monograph 24, Auckland.

Birks, L. 1973. Archaeological Excavations at Sigatoka Dune Site, Fiji. Fiji Museum, Bulletin No. 1, Suva.

Burley, D. V. 1997. Archaeological Research Sigatoka Dune National Park, June 1996. Unpublished report on file, Fiji Museum, Suva.

Burley, D. V. 1998. Archaeology and the Tongan past; 2850150 BP. Journal of World Prehistory 12:337-392.

Burley, D. V. 2002. Archaeology of the Sigatoka Sand Dunes National Park, Report on the 2000 Field Season. Unpublished report on file, Fiji Museum, Suva.

Burley, D. V. 2003. Dynamic landscapes and episodic occupations: Archaeological interpretation and implications in the prehistory of the Sigatoka Sand Dunes. In C. Sand (ed), Pacific Archaeology: Assessments and Prospects, pp. 327-335, Le Cahiers de l'Archeologie en Nouvelle Caledonie 15, Noumea.

Burley, D. V. and Shortland, R. 1999. Report on 1998 Field Work Activities, Sigatoka Dunes National Park, Viti Levu, Fiji. Unpublished report on file, Fiji Museum, Suva.

Burley, D. V., Storey, A. and Witt, J. 2002. On the definition and implications of eastern Lapita ceramics in Tonga. In S. Bedford, C. Sand and D. Burley (eds), Fifty Years in the Field: Essays in Honour and Celebration of Richard Shutler Jr's Archaeological Career, pp. 213-226. New Zealand Archaeological Association Monograph 25, Auckland.

Clark, G. 1999. Post-Lapita Fiji: Cultural Transformation in the Mid-sequence. Unpublished PhD Dissertation, Australian National University, Canberra.

de Biran, A. 2001. The Holocene Geomorphic Evolution of the Sigatoka Delta, Viti Levu, Fiji Islands. Unpublished PhD Dissertation, University of the South Pacific, Suva.

Dickinson, W. R. 1968. Singatoka Sand Dunes, Viti Levu (Fiji). Sedimentary Geology 2: 115-124.

Dickinson, W. R. 1973. Appendix 1: Sand temper in prehistoric potsherds from the Sigatoka dunes, Viti Levu, Fiji. In L. Birks, Archaeological Excavations at Sigatoka Dune Site, Fiji, pp. 69-73. Fiji Museum, Bulletin No. 1, Suva.

Dickinson, W. R. 1999. Petrography of Sand Tempers in Selected Lapita Sherds from the Sigatoka Dune Site, Southwest Coast of Viti Levu in Fiji. Petrographic Report WRD-189, Unpublished ms on file with the author, Tucson.

Dickinson, W. R. 2003. Petrography of subsurface sand, Sigatoka Dune site, Viti Levu, Fiji. Petrographic Report WRD-226, Unpublished ms on file with the author, Tucson.

Dickinson, W. R., Burley, D. V., Nunn, P. D., Anderson, A., Hope, G., de Biran, A., Burke, C. and Matararaba, S. 1998. Geomorphic and archaeological landscapes of the Sigatoka Sand Dune site, Viti Levu, Fiji. Asian Perspectives 37:1-31.

Dye, T. 1988. Social and Cultural Change in the Prehistory of the Ancestral Polynesian Homeland. Unpublished Ph.D. Dissertation, Yale University, New Haven.

Dye, T. 1996. Early eastern to Polynesian plainware at Tongatapu and Lifuka: An exploratory data analysis and comparison. In J. M. Davidson, G. Irwin, B. F. Leach, A. Pawley and D. Brown (eds), Oceanic Culture History: Essays in Honour of Roger Green, pp. 461-473. New Zealand Journal of Archaeology, Special Publication, Auckland.

Gifford, E. W. 1951. Archaeological excavations in Fiji. Anthropological Records 13(3): 189-288.

Green, R. C. 1963. A suggested revision of the Fijian sequence. Journal of the Polynesian Society 72: 235-253.

Green, R. C. 1974. A review of portable artifacts from Western Samoa. In R, C. Green and J. M. Davidson (eds), Archaeology in Western Samoa, Vol. 2. pp. 245-275.

Auckland Institute and Museum Bulletin 7, Auckland. Green, R. C. 1979. Lapita. In J. D. Jennings (ed), The Prehistory of Polynesia. pp. 27-60. Harvard University Press, Cambridge.

Hope, G. 1999. Holocene vegetation histories in the western Pacific--alternative records of human impact. In J.C. Galipaud and I. Lilley (eds), The Pacific from 5000 to 2000 BP: Colonisations and Transformations, pp. 387-404. Editions de L'Institut de Recherche pour le Developpement, Collection Colloques et Seminiaires, Paris.

Hudson, E. 1994. Sigatoka Dune Site Archaeological Rescue Project 1993. Unpublished report on file, Fiji Museum, Suva.

Kirch, P. V. 1988. Niuatoputapu: The Prehistory of a Polynesian Chiefdom. Thomas Burke Memorial Washington State Museum, Monograph No. 5, Seattle.

Kirch, P. V. 1997. The Lapita Peoples, Ancestors of the Oceanic World. Blackwell, Oxford.

Kirpatrick, J. B. and Hassall, D. C. 1981. Vegetation of the Sigatoka Sand Dunes, Fiji. New Zealand Journal of Botany 9:285-297.

Marshall, Y., Crosby, A. Matararaba, S. and Wood, S. 2000. Sigatoka: The Shifting Sands of Fijian Prehistory. Oxbow Books, University of Southampton, Department of Archaeology Monograph 1, Oxford.

Mead, S., Birks, L., Birks, H. and Shaw, E. 1975. Lapita Pottery Style of Fiji and its Associations. Polynesian Society Memoir No. 38, Wellington.

Palmer, J., Shaw, E., Dickinson, E. and Sykes, M. 1968. Pottery making in Sigatoka, Fiji. Records of the Fiji Museum Vol. 1, No. 3, Suva..

Petchey, E 1995. The Archaeology of Kudon: Archaeological Analysis of Lapita Ceramics from Mulifanua and Sigatoka, Fiji. Unpublished MA Thesis, University of Auckland, Auckland.

Poulsen, J. 1987. Early Tongan Prehistory, 2 vols. Terra Australis 12, Canberra.

Sand, C. 2001. Evolutions in the Lapita cultural complex: A view from the southern Lapita province. Archaeology in Oceania 36:65-76.

Spriggs, M. 1990. The changing face of Lapita: Transformation of a design. In M. Spriggs (ed.) Lapita Design, Form and Composition, pp. 83-122, Occasional Papers in Prehistory No. 19, Australian National University, Canberra.

Spriggs, M. 1997. The Island Melanesians. Blackwell, Oxford.

Spriggs, M. 2002. They've grown accustomed to your face. In S. Bedford, C. Sand and D. Burley (eds), Fifty Years in the Field: Essays in Honour and Celebration of Richard Shutler Jr's Archaeological Career. pp. 51-57. New Zealand Archaeological Association Monograph 25, Auckland.

Summerhayes, G. 2000. Lapita Interaction. Terra Australis 15, Australian National University, Canberra.
Table 1. Radiocarbon dates in association with Late Level I ceramics at
the Sigatoka Sand Dune Site, Viti Levu, Fiji.

Sample
Number       Radiocarbon Age       [[delta].sup.13]C

Gak 946      2460 [+ or -] 60 BP
CAMS 48565   2470 [+ or -] 50 BP         -23.7
CAMS 48566   2490 [+ or -] 50 BP         -21.4
CAMS 68196   2510 [+ or -] 40 BP         -24.3

Sample
Number       Reference                       Comment

Gak 946      Birks (1973, 57)
CAMS 48565   Burley & Shortland (1999, 44)   Pooled sample of charcoal
CAMS 48566   Burley & Shortland (1999, 44)   AMS charcoal date
CAMS 68196   Burley (2002, 11)               AMS charcoal date
                                             taken from oven stone
                                             feature

Table 2. Vessel forms and other ceramic specimens recovered from
excavation of the Late Lapita Level 1 floor at the Sigatoka Sand Dunes
in 1965/66 and as reported upon herein. The typology was designed for
the earlier excavvation and is fully defined in Birks (1973).
Diagnostic refers to vessels where all diagnostic features including
dimensions could be determined. Because differences between Types 1a,
1b and 1c are subtle and difficult to define, they were combined as a
single type. Type 2 other refers to recently excavated vessel forms
that could not be related to the existing typology.

                         1965/66     1965/66   1998/2000
                        diagnostic   partial   diagnostic

Open Mouth Jars
Type la/b/c                 40          0          5
Type 1d                      5          0          5
Type l undefined             0         25          0

Bowls
Type 2a                     21          0          7
Type 2b                      2          0          0
Type 2c                      1          0          0
Type 2d                      3          0          1
Type 2e                      4          0          1
Type 2 other                 0          0          2

Narrow Neck Water Jar
Type 3a                      8          0           1
Type 3b                      1          0           0
Type 3c                      1          0           0
Type 3d                      1          0           0

Pot Rests
Type a                       8          0           0
Type b                       2          0           1
Type c                       2          0           1
Type d                       2          0           0
Pot Lids                     2          0           0

Ceramic Disks               13          0           0
Total                      116         25          24

                        1998/2000     Total
                         partial

Open Mouth Jars
Type la/b/c                  0          45
Type 1d                      0          10
Type l undefined            26          51

Bowls
Type 2a                     16          44
Type 2b                      0           2
Type 2c                      3           4
Type 2d                      2           6
Type 2e                      0           5
Type 2 other                 0           2

Narrow Neck Water Jar
Type 3a                      0           9
Type 3b                      0           1
Type 3c                      0           1
Type 3d                      0           1

Pot Rests
Type a                       0           8
Type b                       0           3
Type c                       0           3
Type d                       0           2
Pot Lids                     0           2

Ceramic Disks                0          13
Total                       47         212

Table 3. Mean grain type composition with standard deviation for
tempers in nine late Lapita sherds recovered in 1998 compared to nine
Sigatoka dune sand samples analysed by volumetric point count in
Dickinson (1968). Absolute ranges are included in parenthesis. All the
ranges overlap, and comparative means for each constituent lie within
one standard deviation except for plagioclase. The difference in
plagioclase may be due to operator variance resulting from the 30 years
between sample counts. The similarities are sufficient to identify the
sherds as having local temper types.

                 Sherd Temper               Modern Dune
Grain Type       Sands                      Sands

quartz           17 [+ or -] 4    [9-23]    20 [+ or -] 5    [12-28]
plagioclase      22 [+ or -] 6    [13-31]   12 [+ or -] 3    [7-17]
K-feldspar       2 [+ or -] l     [1-4]     2 [+ or -] 1     [1-3]
total quartzo/
  feldspathic    41 [+ or -] 10   [23-54]   34 [+ or -] 7    [20-48]

clinopyroxene    20 [+ or -] 6    [10-31]   24 [+ or -] 7    [14-37]
hornblende       7 [+ or -] 2     [5-12]    5 [+ or -] 2     [3-9]
epidote          3 [+ or -] 3     [1-11]    3 [+ or -] 1     [2-4]
opague iron
  oxide          10 [+ or -] 4    [3-19]    9 [+ or -] 4     [3-15]
total ferro-
  magnesian      39 [+ or -] 10   [27-61]   41 [+ or -] 10   [24-60]

total poly-
  crystalline
lithic           20 [+ or -] 4    [16-31]   24 [+ or -] 3    [19-30]


(1) In a recent review of archaeological research at Sigatoka, Marshall et al. (2000:14) alternatively propose that a large section of the dune had formed by 3000 BP based on their interpretation of palaeosol distributions and associated archaeological data. Additional archaeological survey in 2000 and 2002, more recent geomorphological field research by Dickinson in 2002, and the conclusions of a several year study on dune geomorphology by de Biran in 2001 have found no evidence to support such a claim.

(2) The Marshall et al. survey recorded these sherds at 640-650 m along their baseline. Their positioning of Birks' excavation grid is approximately 100 m further west than the one plotted here (see Marshall et al. 2000:36). In this context the Level I exposure recorded by them and reexamined in 1998 would be inland of Birks' original excavation as opposed to being adjacent to its western boundary.

(3) We have no concrete data to question the association of dentate stamped bowls with other vessels recovered from Level 1. The motifs, nevertheless, seem more complex than would be expected if Best's assertions on simplification were true. An earlier Lapita component with dentate stamped jars and bowls has been recorded at the site (Hudson 1994; Petchey 1995; Burley 2003), and it is conceivable that Birks' dentate stamped vessels relate to this component, not the Late Lapita phase of Level 1. In this respect, Birks notes that weathering had "removed much of the surface on at least one side" of several of these pieces (Birks' 1973, 27) potentially indicating an exposure on the beach surface before Level 1 materials were deposited and became buried.

DVB: Dept of Archaeology, Simon Fraser University, Burnaby BC, Canada V5A IS6, burley@sfu.ca; WRD: Dept of Geosciences, University of Arizona, Tucson AZ, USA 85701
COPYRIGHT 2004 Blackwell Publishing Limited, a company of John Wiley & Sons, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2004 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Burley, David V.; Dickinson, William R.
Publication:Archaeology in Oceania
Geographic Code:80OCE
Date:Apr 1, 2004
Words:9806
Previous Article:Kimberley points and colonial preference: new insights into the chronology of pressure flaked point forms from the southeast Kimberley, Western...
Next Article:Radiocarbon dates from the Ulong site in Palau and implications for Western Micronesian prehistory.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters