CHAPTER 8 ELUNGUAI, AN ACERAMIC SITE ON ELOAUA ISLAND (SITE EHK) NICK ARAHO \ The Elunguai Site, designated EHK tigation of open sites of apparently non-Lapita age, I m in the PNG National Museum site in- had the opportunity to join the Project as liaison officer ventory, was first recorded during the for the PNG National Museum. After assisting Kirch 1984 reconnaissance survey byJ. Allen with the final season of excavations at Talepakemalai * \ andJ. Specht (Allen et al. 1984:10-11). (Site ECA), and aiding Tyler in initiating the EKE Site .,\ . According to their report: transect excavations on Boliu Island, I was able to spend running east from Elunguai hamlet on Eloaua is a the remainder of my stay in Eloaua testing the EHK site which spans the areas known as Elunguai, Site.2 This chapter summarizes the results obtained from Talepakengi, Talewoko, Elosaurum and Eumua. This open site is about 1 to 1.5 m above high tide level and is i 25-40 m wide. It is covered with gardens and regrowth, is admittedly limited, sufficient material was nonetheless and coconuts. The site appears to be a series of obtained to add to our understanding of the post-Lapita discontinuous shell scatters, with obsidian and highly- phase of prehistory in Eloaua Island. fragmented pieces of pottery. None of the sherds is diagnostic of Lapita. Although the pottery seen is very small and poorly preserved, test excavations may THE SETTING produce diagnostic pieces and allow the site to be placed into a context relative to the Lapita sites (Allen The EHK Site occupies two geomorphic terraces- et al. 1984:10-11).. .. separated by an escarpment of elevated coral lime- Kirch further reconnoitered EHK during the 1985 stone-immediately SE of the small hamlet called and 1986 field seasons, collecting a few shell artifacts Elunguai. The majority of the site, as revealed through from the surface of manioc gardens on the low beach exposure of the dense shell midden by gardening ac- terrace behind the mangroves.' During the 1988 Mussau tivities, is coterminous with a low beach terrace of un- Project field season, when Kirch had decided that sig- consolidated calcareous sands (Fig. 8.1). This terrace, nificant effort needed to be directed toward the inves- situated 6 m asl, has a maximum width of 70 m and 182 ELUNGUAI, AN ACERAMIC SITE ON ELOAUA ISLAND a total length of 500 m. The seaward edge of the ter- face midden distribution. After intensive survey, how- race is marked by a distinct exposure of weathered ever, we discerned that there was a slightly elevated coral, which evidently represents a former, uplifted reef "mound" in one part of the lower terrace, which was edge, probably of late Pleistocene age. Seaward of this, partly tested by our excavation units. the terrace drops abruptly down into a mangrove The upper or higher terrace to the SW lies - 13 m swamp. The inland edge of the lower terrace is marked asl. Surface shell midden was also scattered over this by the base of a steeply sloping bank with numerous terrace, although the density appeared to be significantly coral limestone outcrops, representing a yet older up- lower than on the lower terrace, and the soil lacked the lifted reef edge. characteristic dark-grey organically enriched color. How- The lower terrace is covered with a dark grey-brown ever, three small but discrete mounds were observed sandy loam, clearly anthropogenically enriched through on the upper terrace, including one on which two test the addition of shell midden, charcoal, and other debris units (E529-530N228) were excavated. of human occupation.3 This lower terrace is regularly cultivated by the inhabitants of the Eloaua hamlets, pri- EXCAVATIONS marnly for manioc (Manihot esculenta) , and coconut palms have been planted along the seaward margin of the Five 1 m' test units were excavated at EHK, three on terrace. This gardening activity had disturbed the upper the lower terrace and two on the upper terrace. The 20-30 cm of the dense midden deposit, exposing a sur- three units on the lower terrace were oriented along a face scatter of Anadara sp., Chama sp., and other mol- transect across the terrace on the E300 grid line, at N300, lusk species, as well as artifacts.4 N290, and N270 (Fig. 8.1). Units E300N300 and Although surface midden was scattered over the E300N290 were situated 10 m apart at the ends of a entire lower terrace area (covering - 1,650 m2), there low mound feature which measured - 10 x 1.5 m. Unit were no discrete features which would have aided in E300N270 lay 20 m further W along the same transect, the selection of particular areas for positioning excava- closer to the foot of the limestone escarpment. Eleva- tion units. Choosing a location for excavations was fur- tions were taken along the E300 transect, and a profile ther complicated by the mosaic of newly cleared gar- across the lower terrace is shown in Figure 8.2. On the den areas, older gardens going into fallow, and second- upper terrace, the two units were adjacent forming a 1 ary growth which partly obscured the pattern of sur- by 2 m trench, at grid location E529-530N228. FIGURE 8.1 Map of the EHK Site showing the location of excavated units. < ~~~~~~~~~~~Edge of terrace \ ~~~~~~Burial ,' \ \ *~~~~~~~~E300 N300 Lower terrace \ m>>>>\ 5E300 N290 * ~ erc Mon ihe erc \ ~~~~~~~~~~~~~~~~~~~Mound ELUNGUAI, AN ACERAMIC SITE ON ELOAUA ISLAND 183 FIGURE 8.2 Elevation profile across the lower terrace of the EHK Site, along the E300 transect. Because the cultural deposit in the lower terrace test midden and culturally sterile. This is therefore units was quite sticky and difficult to screen, it was wet the original beach terrace formation upon which sieved using water obtained from a nearby well. Sedi- human habitation initially began. In E300N300 ment from Unit E300N300 was sieved with 5 mm this sand deposit was excavated to a depth of mesh, while that from all other units was sieved with a 1.5 cm bs, when a compact deposit of coarse 7 mm mesh. Excavation proceeded in 20 cm levels, sand and numerous Acropora branch coral frag- except where natural stratigraphy was encountered. ments was exposed. In the other two (land- ward) units, bedrock consisting of weathered STRA TIGRAPHY OF THE L OWER limestone was encountered underlying Layer Ill. TERRACE UNITS To summarize the depositional sequence at EHK, A fairly consistent stratigraphy was revealed in the three there are two stratigraphically distinct occupation phases, transect units, and allowed the layers in each unit to be represented by Layers I and II. Layer II, the earlier oc- correlated with each other, as seen in Figure 8.3. The cupation, commenced on a coastal beach terrace and following layers were present: resulted in a shell midden accumulation, but without a LAYER I: Averaging 25-30 cm thick, this is a black, heavy input of charcoal or other carbon-rich organic greasy textured loam with considerable clay material. This was superseded by the Layer I occupa- content, and much organic material, including tion, which differs from Layer II in the higher shell a high density of shell midden. The black color midden density and in the enrichment of this deposit clearly derives from a great deal of finely-dis- with carbonaceous material which has given the deposit persed charcoal. This layer was present in units its black color and "greasy" texture. E300N300 and E300N290, but in Unit The only feature exposed during excavation was E300N270 it was replaced by a light brown part of a human burial in E300N290, revealed in the clay with a very low density of shell midden. SW part of the unit at 35 cm bs. The feature continued This latter clay seems to represent material which into the S wall of the unit, and only a femur and a few has eroded from the upper terrace and escarp- other post-cranial bones were exposed. At the request ment slope and been deposited through slope of the landowners, these skeletal fragments were re- wash. buried in the unit at the conclusion of the excavation. LAYER II: This is a light brown sandy loam, averag- ing 20 cm thick, and also containing shell STRATIGRAPHYOF THE UPPER midden especially in Unit E300N300 where the TERRACE UNITS density of shells was quite high. LAYER III: Also extending across all units, Layer III The two continuous test units in the upper terrace re- consists of white, calcareous sand lacking shell vealed a shallow cultural deposit, with a dark gray to 184 ELUNGUAI, AN ACERAMIC SITE ON ELOAUA ISLAND 0 0 co CO z E U) 0 ~~~~~~~~~~~ I~~~~~~~~~~~~~~~~~C 01 z I Q) o -o UL) J Ur CY Cu 0 0 U)~~~- 04 Q-~~~~~~~~~~~~~ bU) 0l) ELUNGUAI, AN ACERAMIC SITE ON ELOAUA ISLAND 185 black garden soil (heavily reworked) containing shell was recovered from the surface, and a preform of a midden, which graded rapidly into a brown clay at - 20 small Tridacna-shell adz was excavated from the top of cm bs. At - 40 cm bs the weathered limestone bedrock Layer II in E300N300. From Layer I of E300N300 was exposed. Although shell midden was present in rela- we also recovered a roughout (not ground) of a Trochus- tively high density, this deposit has been extensively re- shell armring. In Unit E300N270, Layer I, we found a worked, and no intact stratigraphy remains. thick shell ring (broken) which may have been part of a Tridacna-shell ring preform. A number of Anadara RADIOCARBON DATING antiquata shells which probably functioned as net weights were also found in both Layers I and II. Two samples of midden shell (both of Anadara From Unit E520N228 in the upper terrace, we re- antiquata) from Layers I and II of Unit E300N300 were covered a Terebra-shell adz at 10 cm bs, and a possible submitted for radiocarbon dating (see Chapter 10 for shell fishhook preform at 20 cm bs. A coral abrader sample details). The sample from Layer I (Beta-30688) was found in Unit E520N228. The two upper terrace yielded a conventional age of 1080 + 60 BP, which cali- units also yielded a total of 17Anadara antiquata valves brates to AD 940-1070 at 1a(. The sample from Layer II with rough perforations, which may have functioned as (Beta-30689) yielded a conventional age of 1560 + 70 net weights. BP, which calibrates to AD 460-620 at 1y (both calibra- tions based on AR = -320). These dates suggest that the FAUNAL MATERIALS earlier occupation phase at EHK began in the middle of the first millennium AD, while the upper occupation As noted earlier, the Layer I deposit on the lower ter- dates to the beginning of the second millennium AD. race contains dense shell midden, especially in the vicin- There is every reason to think that occupation on the ity of the low mound. Unit E300N300 had a Layer I coastal terrace at EHK was continuous over this time density of 56.7 kg/m3, and Unit E300N290 a density span, although the location of individual dwellings and of 31.5 kg/m2. The earlier Layer II deposit has a much shell midden dumps could have shifted over time. lower shell midden density of 9.32 kg/m3 in Unit E300N300, and of only 1.45 kg/m3 in E300N290. In CULTURAL CONTENT OF EHK both layers, the dominant shell taxa were Strombus lubuanus, Lambis lambis,Anadara antu uata, and Chama sp., With a total excavation area of only 5 m2, the sample of although many other taxa were also represented. cultural materials from EHK is rather limited. Nonethe- The upper terrace shell midden had an even higher less, the assemblages from Layers I and II of the lower density of - 81.5 kg/m3 as represented in Unit terrace are particularly important as they provide infor- E530N228, one of the highest shell midden densities mation on the post-Lapita period in Mussau. recorded for any of our Mussau sites. The same taxa dominated in this part of the site as on the lower ter- ARTIFACTS race. The vertebrate faunal sample from EHK totals 216 Although Allen et al. (1984:10-11) reported the pres- NISP, and is dominated by fishbones. However, there ence of "highly fragmented pieces of pottery" at EHK, is also significant representation of pig (Sus scrofa), as our excavations failed to yield additional sherds. Most well as small quantities of odontocete, bird, snake, and likely, this is a matter of sample size, with pottery being turtle; further details of the faunal assemblages will be a rare item in the EHK deposits. presented in Volume II. In Unit E300N300 we discovered a Terebra-shell adz at the base of Layer II, associated with anAnadara- CONCLUSIONS shell net sinker. This adz is significant, as it can be asso- ciated with the AD 460-620 radiocarbon date, and is Excavations at the EHK Site revealed the presence of thus the earliest Terebra-adz in the Mussau artifact as- intact stratification in a coastal habitation site with two semblages (see Volume III). Another Terebra-shell adz main phases of occupation, spanning the late first and 186 ELUNGUAI, AN ACERAMIC SITE ON ELOAUA ISLAND early second millennia AD. As such, this site adds impor- later prehistoric assemblage of Terebra-shell and Tridacna- tant data for the construction of a cultural sequence for shell adzes, and Trochus-sheli armrings in association with Mussau in the post-Lapita period. Of particular note is pig bone and high-density shell midden dumping; this the presence of a Terebra-shell adz in association with a assemblage shows close parallels with the upper strati- radiocarbon date of AD 460-620 (calibrated), the oldest graphic deposits at EKE (Boliu Island) and EKS date for this diagnostic artifact type in Mussau. The Layer (Emussau Island). I assemblage on the lower terrace also contains a typical NOTES TO CHAPTER 8 'Although Kirch extensively searched the heavily gardened site 2This project was suggested to me by P. Kirch, who had real- for further evidence of the "very small and poorly preserved" ized the potential significance of the EHK Site for the post- pottery reported by Allen et al. (n.d.), no further sherds were Lapita periods on Eloaua, and who provided funds for the recovered, nor has additional pottery resulted from my own excavation through his NSF grant. test excavations. We do not dispute the report by Allen et al. that sherds were found at the EHK Site (since similar post- The anthropogenic loam of the lower terrace, extensively uti- Lapita pottery is indeed known from other Mussau sites, such lized by the Eloaua islanders for manioc cultivation, is very as EKE and EKU), but simply comment that the failure to similar to the soil on the higher portions of the ECA Site, also obtain additional sherds suggests that pottery is an extremely used for root crop gardening; see Chapter 4. rare item in the EHK Site deposits. A large Tridacna-shell adz preform and a Terebra-shell adz were also collected on the site 4Some additional disturbance in ungardened parts of the ter- surface at the commencement of excavations in 1988. race had resulted from the burrows of land crabs.