CHAPTER 5 LAPITA ROCKSHELTERS OF ELOAUA AND MUSSAU ISLANDS (SITES EHM, EHN, EKO, EKP, AND EKQ) MARSHALL 1. WEISLER INTRODUCTION a Leitz level, stadia rod, and metric tape were used for the 1986 field season, five mapping the larger, more complex rockshelters of EKP During the 1986 field season, five and EKQ on Mussau Island. A 1 m grid was estab- rockshelter sites were excavated or lished at EKQ to facilitate collection and provenience .K(S.:tested, three on Eloaua Island and two of surface artifacts. At the Eloaua shelters a baseline in / - \ on Mussau Island, yielding evidence of 1 m divisions was oriented through the longest axis of both Lapita and post-Lapita period occu- each site. Test pits were designated consecutively, and pations. This chapter reports the results of measured 1 m2. At EKO and EKP test pits were joined these rockshelter excavations; detailed discus- to form trenches perpendicular to the long axes of the sion of faunal remains and artifacts recovered from these sites and across the dripline. sites will be presented in Volumes II and III, respec- Excavation proceeded by stratigraphic units. Within tively. these divisions, arbitrary levels or spits were excavated by trowel in 5, 10, or 20 cm increments for finer vertical GENERAL EXCAVATION METHODS control. Five mm mesh screens were used to sieve site matrix except at EHN where a 7 mm mesh was used. Investigation of each site included preparation of a de- At Site EKQ, when black, greasy, cultural layers were tailed map noting elevation contours, location of dripline encountered, making recognition of cultural material which delimited the sheltered area, and one or more difficult in the screens (especially obsidian flakes and elevational cross-sections. Grid or test pit layout, cul- charcoal), the site matrix was wet screened with sea water. tural features (e.g., stone pavements, walls, and modern After screening, cultural material was sorted into the posts), surface artifacts, and midden concentrations were following classes and bagged separately: (1) shell; (2) also plotted. On Eloaua, the smaller sites (EHM, EHN, bone; (3) obsidian; (4) pottery; and (5) charcoal includ- and EKO) were mapped by compass and tape, whereas ing macrobotanical remains. Artifacts found in situ were LAPITA ROCKSHELTERS OF ELOAUA AND MUSSAU ISLANDS 147 plotted by x, y, and z coordinates. Specimens recovered informants said was done during World War II to con- in screens were provenienced to level or, where pos- ceal a hiding place. With a constricted entrance of just sible, unit quadrant. over 4 m, the cave opens to a height of 1.4 m dropping Stratigraphic units were defined primarily by sedi- to less than 0.5 m towards the back. The rear third of ment color and texture and designated with Roman the cave has active stalactites and is home to several numerals from top to bottom. Profile descriptions fol- dozen flying foxes (Pteropus sp.). Sunlight penetrates low terminology of the U.S. Department of Agricul- the cave to - 9 m, and it is within this area that most ture Soil Survey Manual. Standardized sediment colors cultural material was found. were assigned by reference to the Munsell Soil Color Three 1 m2 test pits were excavated in 20 cm spits Charts (1975). Because EKQ is an important site with a and all sediments were dry-sieved through 5 mm screens. deep and complex stratigraphic record, sediment One unit was excavated just inside the dripline, another samples were taken for detailed lab analyses (described near the center of the lighted area, and the third to- below). wards the rear of the cave (Fig. 5.1)1. The cultural stra- Charcoal samples for radiocarbon assay were taken tum was barely discernible, consisting of dark reddish where possible, otherwise a vertical sequence of marine brown (5YR 3/4) silt with water-rounded pebble-sized shellfish was acquired from most sites. Because of the limestone, coral, and shell (Fig. 5.2). Limestone pebbles prohibitive cost of transporting field specimens back graded to cobble size near the 50 cm base of excava- to the U.S. for analysis, all marine shellfish were identi- tion. The sediment had a loose consistency and broke fied to the lowest taxonomic level in the field lab and into small peds during excavation. At the rear of the discarded, while voucher specimens were retained for cave below the stalactites, the peds were larger due to each taxon. When large quantities of undecorated ce- leaching of calcium carbonate through the profile. This ramics were recovered, such as at EKQ, a complete also contributed to the dark reddish gray (5YR 4/2) sedi- sample was taken from only one test pit. Otherwise, all ment color near the base of excavation where limestone cultural material was retained for further study. bedrock was encountered. Rootlets were common in the upper 10 cm of the deposit. No features were en- ELOAUA ISLAND EXCAVATIONS countered and sparse cultural materials were found throughout the deposit, but somewhat more frequently The S portion of Eloaua Island is dominated by a large near the lower half of the stratum. limestone escarpment, the base of which was eroded Artifacts include two obsidian flakes (mean weight, by wave action during the mid-Holocene higher sea level 1.5 g) from Lou Island sources (all sourcing assignments stand (see Chapter 2). Several incised notches, typically byM. S. Allen) and two Tridacna shell flakes (mean length, wider than they are deep, are found on the E side of the 31.25 mm; weight, 6.2 g; both with cortex, one with island and provided natural shelters for habitation. Not dorsal flake scars). Five workedAnadara shell valves have all shelters bear evidence of prehistoric occupation, a mean weight of 18.12 g and average length of 52.04 however, and test excavations at three locales are de- mm. They were all worked or showed evidence of use scribed. The three sites, EHM, EHN, and EKO form a along the dorsal margins which were straight, irregular, cluster (UTM Coordinates GU948257) at the base of a or convex. Two base sections of Trochus niloticus are prob- limestone cliff, atop a scree slope - 50 m inland from ably ring fragments and weigh 6.0 and 25.0 g, and are the shoreline, inland of a mangrove swamp that sepa- 42.29 and 67.86 mm long. Both show percussion flak- rates the locale from the lagoon and reef-protected coast. ing on one or two sides. A Tridacna-shell adz blank made from the margin of a T. maxima shell weighs 85.0 g and SITE EHM has the following measurements: length, 92.84 mm; thickness, 10.17; midpoint width, 56.41; poll width, A large solution cave 24.5 m deep and 10.4 m in maxi- 24.41; and cutting edge width,61.28. Only two undeco- mum width, EHM is the most S site of the cluster and rated, shell-tempered body sherds (mean weight, 4.5 is marked by a large Ficus tree (Fig. 5.1). The level dirt grams) were found near the base of excavation. floor of the cave covers 152 in2. Several courses of coral Some 3.2 kg of marine shellfish, identified to 27 cobbles and boulders are stacked at the entrance, which taxa, were recovered from Site EHM (Table 5.1). Sev- LAPITA ROCKSHELTERS OF ELOAUA AND MUSSAU ISLANDS FIGURE 5.1 Plan of Site EHM. Ocm 10 Coalpbble 20-Ro 11 30 - Test Pit 1 O cm - 1 10 - _ Limestone 20 a Test Pit 2 0 cm _ 1 0 20 - 30 - 40_ _ 40 L Limestone 50 cobble Root 60 Test Pit 3 FIGURE 5.2 Stratigraphic section of the E face of Units 1-3, Site EHM. enty-four percent of all shellfish by weight came from Unit 3, closest to the entrance, where there is the most available light. Gastropods totaled 38.4% of all shell- fish with Strombus maculatus, Turbo cf. setosus, and Trochus niloticus accounting for most of the weight. The domi- nant bivalves (61.65% by weight) includedAnadara sp. and Tridacna sp. Some 49 pieces of crustacea (weight, 25.0 g) were recovered from Units 1 and 3; a single piece of echinoderm weighed 3.0 g. A badly disturbed hu- man burial was located at the base of Unit 3, and only three bones could be positively identified as Homo sapi- ens, although 18 pieces of medium mammal and 20 of medium vertebrate may also be human. One fish bone concluded the bone inventory. The lack of a well-developed cultural stratigraphy, few artifacts and food remains, and no hearth features suggest that the EHM cave site was used infrequently. No chronometric dating was carried out but the ceram- ics and obsidian source tentatively suggest a post-Lapita occupation. SITE EKO This rockshelter is located 11 m W of Site EHM and is situated at the top of a scree slope. The shelter mea- sures 10 m wide and 2.5 m deep inside the dripline (19.5 . .i . . . .- .- . . .',' ,# * - .' .' . . . . . . ' , .' , .' .' .' . .' . .' I~~~~~ ~ ~~~~ ~ a, . .^ .^ . .s . '. S > . a . , , ", ' . , '. f. . . . - . - . ' . ' . " . - . . ," . . . . . ". s . * . S . ' . F ,' , # .* .' .' .' .' . ' ' s * ' '* jv *' * *. ', , '. ". ", , . * . ", s . . . ' . ^ . ' . ' . ' . ' . . . v.y~~~~~~~~ Are .of .... . . A~ ~~ ~ ~ ~~~~ .t . . . . . . . ... . ... .. A.. 'o- * ^ A 0 1 m I u G~~~~~rid t . .. * _ -- - - - - -f . . 148 .. . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . ............. ................. ............. . . . . . . . . . . . . .......... .... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . water rounded coral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ........... ........... .. . . . . . . . . . . . . ............. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ................... ........... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . .. . . .. . . . . . . . . . . . .. .. .................... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ................ . . . . . . . . . . . . . . . . . . . . . . 1 . . . . . .. .......... ......... ............ LAPITA ROCKSHELTERS OF ELOAUA AND MUSSAU ISLANDS 149 TABLE 5.1 Invertebrate fauna from sites EHM m2 of actual usable space) where the ceiling rises from and EKO. (Weight in grams). - 1.3-3.1 m in height (Fig. 5.3). A 1 by 2.5 m long trench was excavated in the middle of the site across the Molluskan Taxon Site Site dripline. Some 1.54 m3 of deposit was excavated in 20 EHM EKO cm arbitrary levels and dry-sieved through 5 mm screens, Gastropods requiring 10 person-days of labor. Trochus niloticus 155 435 Three cultural layers were identified atop a lime- Trochidae 45 80 stone bedrock base. The average thickness of the com- Turbo cf setosus 265 385 bined cultural layers was 60 cm along the trench. The Turbo marmoratus 360 top cultural layers evidenced intensive hearth and earth- Turbo operculae 30 layers* oven activity, while ash content decreased with depth Neritidae 15 30 and the volume of coral cobbles increased with boul- Cerithiidae 5 1 0 ders present in the lowest layer (Fig. 5.4). The stratigra- Cerithium nodulosum 40 145 X Strombus cf maculatus 290 455 phy of excavation Units 1 to 3, W profile, was as fol- Lambis lambis 50 995 lows: Cypraeidae 120 220 LAYE1R I: Cultural layer, gradual, wavy boundary, very Polinices spp. 2 75 ashy, silty dark gray (1OYR 4/1 to 5/1) sedi- Fusinus sp. 1 5 ment, with ash pockets, charcoal chunks, and Vasum ceramicum 40 cultural materials. Few limestone cobbles, loose Conidae 60 190 consistency, non-plastic, easy excavation with Terebridae 80 trowel, dense roots and rootlets; deposit thor- Thaididae 10 oughly reworked by crabs, and hearth and earth Stellaria solaris 50 10 oven activity. Miscellaneous gastropod taxa 5 100 LAYER II: Cultural layer, abrupt, wavy boundary, less Landsnails (by count) 5 70 ash, silty very dark gray to black (1OYR 3/1 to Bivalves 2/1) sediment, with limestone cobbles (few Arca sp. 700 water-round coral sediments < 5 cm) and cul- Anadara sp. 675 tural material; loose consistency, non-plastic; Spondylus sp. 70 140 some disturbance, but less than Layer I. Lucinidae 35 90 LAYER Im: Cultural layer, abrupt, smooth boundary, Chama sp. 110 650 much less ash, mostly subrounded limestone Cardium sp. 40 20 cobbles (some water-rounded) and boulders Fragum sp. 5 5 with a silty dark brown (lOYR 4/3) matrix, loose Tridacna maxima 230 570 consistency, non-plastic, non-sticky; large shell Tridacna gigas 90 250 midden and land snails noted. Tridacna squamosa 150 LAYER IV: Sterile, limestone bedrock. Tridacna spp. 370 1300 Hippopus hippopus 230 3640 Artifacts recovered included worked Anadara sp. Gafrarium sp. 40 5 valves (mean weight, 16.0 g), a Trochus-shell fishhook Tellinidae 10 blank (1.0 g), and three ring fragments made from Tapes sp. 5 5 Trochus. Some 37 obsidian flakes had a total weight of Mytilidae 40 20.4 g (mean, 0.55 g); 27 of these were sourced to Lou Miscellaneous bivalve taxa 45 300 Island, one to Talasea, and 9 were of uncertain origin. Ceramics were the most numerous artifact class account- Crustacea 25 51 ing for 125 specimens weighing 588 g. The assemblage Echinoderm 3was typified by undecorated, shell-tempered, body sherds. Two had volcanic temper and are noticeably thicker than the shell-tempered sherds. Only one sherd was dentate-stamped (volcanic sand temper, thickness 150 LAPITA ROCKSHELTERS OF ELOAUA AND MUSSAU ISLANDS I 10.3 mm), and five were rim sherds (4%). -- ipIinl Some 11.5 kg of shellfish were recovered with 32% comprising gastropods (mostly Lambis lambis, Strombus sp., and Trochus niloticus), and 68% bivalves including ; < / W Tridacnasp., Hippopus hippopus, andAnadara sp. One piece - -: - \ / of medium mammal bone and 80 pieces of crustacea were also inventoried (see Table 5.1). |A-- \F13T7TY1 2| As A shel sample (Turbo marroratus) from Unit 1, Layer III yielded a conventional radiocarbon age of 3200 + 70 (Beta-25669), calibrated to 1150-942 BC, thus sug- gesting a mid-Lapita age for initial occupation of the site (see Chapter 10). The EKO Site was the most intensively used of the Eloaua rockshelters and saw initial occupation - 3,000 0 1 m years ago. The relatively small size of the site and the presence of large Lapita villages a short distance away, suggests that the shelter was probably used intermit- tently by a few people at a time. The artifact assem- - . - - - l Grid N | blage, while larger than that of the surrounding shel- ters, contains a restricted set of functional classes asso- ciated with food preparation. Indeed, the upper layers are thoroughly reworked from hearth and earth-oven activity. While not a major habitation locale, the site adds definition to the Eloaua Island settlement pattern dur- ing Lapita times. SITE EHN A 1- X ~ n3 1 2 1 1 I This overhang shelter is located a short distance from A' EKO, just above the mangrove mud flats. The relatively level shelter floor measures 6.6 m wide by 2.2 m deep FIGURE 5.3 Plan of Site EKO. and slopes abruptly down towards the sea beginning 0 cm312 _~~~~~~~~~~ 0 F0-IU r raphi secton eW o - FIUR 5.4 Stratiraphi setonoiteW aefUntD13uStsEO LAPITA ROCKSHELTERS OF ELOAUA AND MUSSAU ISLANDS 151 1 m beyond the dripline (Fig. 5.5). In the center of the floor, a single 1 m2 unit was excavated in 20 cm arbitrary levels by T. L. Hunt on 3 October 1986. The unit was excavated to a depth of 70 cm bs where we encountered limestone cobbles indicating in situ, de- A composing bedrock. All sediments were dry sieved with 7 mm screens. Two layers were defined: Layer I, a gray- ish brown (10YR 5-6/2), fine, "powdery" sediment with I : :- A' angular limestone cobbles; and Layer II consisting of angular cobbles in a poorly sorted white (1OYR 8/2) . / matrix grading to decomposing limestone bedrock (Fig. 5.6). I.- . . -. . . . . The sparse artifacts included five pieces of worked I Trochus and seven obsidian flakes (six sourced to Lou . . // Grid N and one to Talasea). No bone was recovered although 4.0 kg of mollusks were dominated by bivalves of Dripline1 Tridaca,Anadara, and Hippopus. Steepslope No chronometric dates were obtained, but the - A' sparse artifacts and weakly developed cultural layer points to only intermittent use of this shelter. SUMMARY OF THE ELOAUA ROCKSHEL TERS While not major habitation locales, the cluster of | \.m rockshelters situated along the S shore of Eloaua Is- land are important components for understanding the . . . . .. overall island settlement pattern. These three sites were probably occupied or utilized by small groups for short periods of time beginning as early as 3,000 years ago. The artifact assemblages are meager, but do evidence FIGURE 5.5 Plan of Site EHN. the use of pottery for cooking, and perhaps for storage. Two volcanic-tempered sherds (one of which is den- tate-stamped) suggest at least some link with Lapita com- munities on Mussau. WorkedAnadara valves are present lets and villages during at least a portion of the Lapita whose dorsal margins are consistent with scraping soft peirod, thus provide another dimension to the overall material, perhaps tubers, breadfruit, and coconut. Ob- island settlement pattern of Eloaua Island. sidian flakes may have been used for minor cutting tasks and there is some evidence of working Trochus gastro- MUSSAU ISLAND EXCAVATIONS pods into ornaments such as arm rings. Tridacna flakes and an adz preform suggest manufacture of this im- During October and November 1986, two rockshelters portant wood-working tool. Aside from these purely were excavated in the NW part of Mussau Island, in the economic activities, the Eloaua rockshelters provided vicinity of Tanaliu Village (see Fig. 3.1). One objective ready-made shelters for persons who may simply have in choosing these sites was the possibility that they might wanted to get away from the confines or close scrutiny reveal a pre-Lapita settlement of the main island, as the of village life. Additionally, the rockshelters may have LHP had shown to be the case with rockshelters and been the locales for small-group picnics or day fishing caves in New Ireland, and has more recently been dem- sorties along this section of the coastline. The shelters, onstrated for Manus Island (Fredericksen et al. 1993). which were presumably integrated with the larger ham- Unfortunately, these sites proved-for geomorphologi- 152 LAPITA ROCKSHELTERS OF ELOAUA AND MUSSAU ISLANDS cal reasons to be made clear below-not suited to the ingHippopus hippopus, Spondylus sp., Turbo setosus, Chama preservation of such a pre-Lapita record. They did, how- sp., Codakia sp., Tellina sp., Trochus sp., Cypraea cf. ever, add considerably to our knowledge of the later maun'tiana, Cerithidae, and Neritidae. John Poli (a resi- Lapita period, especially Site EKQ. dent of Pomanai Village) related that his grandparents occupied the EKP Site and had dug earth ovens in the SITE EKIE EATULAWANA ROCKSHELTER NE corner of the shelter. Some of the surface shellfish may date to this period. The excavation crew commented About a 20 minute walk S of Pomanai Village along the that within a 15 m radius of the site, nine species of NW coast of Mussau Island, then 0.5 km inland fol- trees and shrubs were important sources of edible fruits, lowing an unnamed stream which meanders through nuts, and leaves, one provided raw material for rope, low-lying terrain with dense stands of sago (Metroxylon while other plants supplied wood for house, furniture, sp.) palm, is the limestone rockshelter of EKP at and canoe construction. Eatulawana (UTM coordinates GU836494). It is the larg- Prior to excavation of EKP, the interior and scree est of several shelters at the base of an uplifted lime- slope was cleared of vegetation. A baseline, oriented stone block that trends NE-SW for - 100 m. The block 3140 MN and thus perpendicular to the dripline, was is 15-20 m high at its SW end which is near the river. A staked at 1 m intervals. Five 1 m2 units of a discontinu- scree slope fronts EKP for - 25 m and descends onto ous 7 m long trench were excavated in 10 cm arbitrary the broad swampy flat. The rockshelter measures 11 m levels during 14 person-days of work (Fig. 5.7). One wide at the entrance, 6 m deep, and the ceiling height is unit was excavated in 5 cm levels to provide finer verti- 5.4 m at the dripline. The usable level floor area is 50 m2 cal control and all sediments from 2.25 m3 were sieved (Fig. 5.7). Surface cultural material is limited to a few through 5 mm mesh. From the cultural layer, the count small undecorated sherds and marine shellfish includ- and weight of all rocks > 5 cm was inventoried to deter- mine the presence and intensity of roof-fall. A total of 268 rocks weighing 47.75 kg (mean, 178 g) was collected from four units as follows: Unit 1, 49 rocks, 3,200 g; FIGURE 5.6 Stratigraphic section of the Unit 2,74 rocks, 8,100 g; Unit 3, 88 rocks, 11,800 g; and test unit in Site EHN. Unit 4, 57 rocks, 24,650 g. Most rocks were found at Unit 4 located at the dripline where roof-fall and mate- 0 cm rial from the vertical cliff surface contributed to the deposit. Layer I, a black silty organic-rich cultural layer (7.5YR 2/0 to 10YR 3/2), was uniform throughout the length of the trench and averaged 14 cm thick. Cobble- 50 sized chunks of decomposing limestone were encoun- tered towards the base of the layer. The matrix is plas- tic, of loose consistency with dense rootlets and few East roots. The gradual boundary is wavy. An otherwise ster- ile-looking deposit, Layer II was stained from leaching O cm of the upper dark cultural layer and represents a transi- tion from Layer I to the C horizon. The sediment is very sticky, compact, silty-clay which was difficult to ex- cavate. There was virtually no cultural material and no roots. A yellowish-brown decomposing limestone bed- | p so~~~~~~5 rock is Layer r :. A stratigraphic profile is shown in Fig-l Few artifacts were recovered from the excavations but included 8 obsidian flakes with a total weight of 3.4 South g (mean weight, 0.43 g) and 63 calcareous-tempered,l l LAP/TA RoCKSHELTERS OF ELOAUA AND MUSSAU ISLANDS 153 Grid N 0 i m A' ~ ~ ~ ~ L A~~~~~~~~~~~~~~~A 7.0 6.5 6.0 5.5 5.0 4.5 4.0 3.5 3.0 I ~~~~~~~~~~~~~2.5 Driplinel A. A' 2.0 / ~~~~~~~~~~~~~~~~1.5 I 1~~~~~~~~~~~~~~~~~~~~~i.0 .. . . . . ~~~~~~~~~~~~~~~~~~~~0.5 FIGURE 5.7 Plan of Site EKP, Eatulawana rockshelter _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ plain body sherds weighing 107 g with a mean of 1.70 g. in Volume III). The stones appear unmodified and are 154 LAPITA ROCKSHELTERS OF ELOAUA AND MUSSAU ISLANDS 1 _ 2 IV| 2 I ~ Bedrock____ IV 3- FIGURE 5.8 IV ~~~~~~~Stratigraphic East Profile i I profile of the N 11 face of Units 1-5, Bedrock 111 Site EKP uting 44% of the total weight with dominant taxa such bance is evident from crab burrows, and excavation for as Turbo and Lambis lamb s.Hipopshippopus and Spondylus earth ovens and posts. sp. were the most prevalent bivalves, accounting for 56% Objectives of my research included: (1) prepara- of the assemblage. tion of a site map; (2) collection of surface artifacts The Eatulawana rockshelter (EKP) was probably plotted to 1 m2 grids; and (3) excavation of at least 2 m2 not a primary habitation locale due to the limited quan- to determine the depth and nature of the deposits and tity and restricted functional range of artifacts, low den- to obtain samples of artifacts, subsistence remains, and sities of food remains, lack of hearths and earth ovens, dating material. From 16-22 November 1986, we spent and a thin, weakly developed cultural deposit. The close a total of 47 person-days completing these objectives. proximity of the sago palm stands near the stream be- After clearing the site area of shrubs and grasses, a 27 low the site may have attracted short-term visitors to m long, E-W baseline was established roughly parallel the area for harvesting and processing sago palm pith to the dripline. Several shorter baselines perpendicular into flour. The EKP locale would have been a conve- to this axis facilitated plotting of surface artifacts and nient shelter for residing during such short periods of preparing a contour map at 25 cm intervals. time. Two 1 m2 units were excavated, one on either side of the dripline. Some 4.8 m3 of cultural and natural SITE EKQ, EPAKAPAKA ROCKSHELTER deposits were passed through 5 mm mesh sieves. The dark, "greasy" layers were water-screened with sea wa- Two km N along the coast from Pomanai Village, then ter to facilitate recovery of obsidian flakes and charred a five minute walk inland on the S side of an unnamed bones that otherwise would have been difficult to iden- drainage is the Epakapaka rockshelter (UTM coordi- tify within a black sediment matrix. nates GU833523) situated at the base of an upraised limestone block nearly 30 m high. The landowners are STRATIGRAPHY AND DEPOSITIONAL SEQUENCE OF EKQ John Poli and Denny Epeli of Pomanai Village. The shelter is 24 m wide (E-W) at the entrance, and 7.4 m The detailed description of the EKQ stratigraphy, as deep with a relatively level protected area of 88 m2 (Fig. revealed in the S profile of Unit 2, is presented below 5.9). At the dripline, the ceiling height is 2.1 m descend- (see Fig. 5.10): ing to 1.4 m at the rear of the shelter. The surface cul- LAYER I: 0-7 cm bs; probably post-occupation sur- tural deposit extends outside the shelter for _ 100 m2 face (after 1945) of relatively sterile, black (5YR and is marked by marine shellfish, potsherds, and other 2.5/12) silt; structureless; consistency: loose, artifacts. Inside, at the W end of the shelter, is a low noncoherent (dry), slightly sticky (wet); slightly pavement, 1.5 m in diameter, consisting primarily of a plastic; no roots, few rootlets; abrupt, smooth single course of water-rounded, tabular coral and lime- boundary. stone cobbles. At the rear center of the shelter are the LAYER II: 1-46 cm bs; fairly homogeneous cultural remains of three forked posts (12-15 cm in diameter) layer, no discrete features in profile, charcoal that previously supported a low sleeping platform, ac- chunks, reworked ash pockets; very dark gray cording to informants who also related that villagers (lOYR 3/1) sandy-silt, limestone pebbles and lived here during World War II. More recent site distur- cobbles (probably reworked ceiling collapse), LAP/TA RoCKSHELTERS OF ELOAUA AND MUSSAU ISLANDS 155 A ... Grid N 0 i m Al L Historic artifacts Speleothems 0.25 w.I~~~~~~~~~~~~~~~~~~~05 N ~~~~~~~~~~~~~~0.75 Dripline~ 1 .00 156 LAPITA ROCKSHELTERS OF ELOAUA AND MUSSAU ISLANDS o cm burrows; gradual, wavy boundary. LAYER III: 34-118 cm bs; cultural layer, few discrete hearths, ash, charcoal chunks, fire-altered lime- stone, very dark gray (10YR 3/2) silt, few gravel- sized sediments; structureless; weakly coherent (dry), slightly sticky (wet); slightly plastic; few 77l7,7771. |roots and rootlets; clear, wavy boundary. LAYERTV:88-146 cm bs; cultural layer dense hearths ash lenses, fire-altered I'mestone, black (5YR 2.5/1) sandy silt; structureless; weakly coher- ent (dry), sticky (wet); slightly plastic; few roots and rootlets; diffuse, smooth boundary defined by the increase of gravel towards boundary with 100 ~~~Layer V. LAYER V: 134-180 cm bs; cultural layer, charcoal