CHAPTER 6 THE BOLIU SITE (SITE EKE) PATRICK V. KIRCH 40 On September 18, 1985 while return- ever, and it was not until the 1988 season that the Boliu E ing by boat from a trip to Palakau, I Site was finally investigated. As discussed in Chapter 1, , landed on Boliu Island and conducted one major objective for this final season was the exca- ' a surface reconnaissance, finding two vation of a site with potential to elucidate the post- archaeological sites (EKE and EKF). Lapita occupation period in Mussau. My initial choice / \ . The EKF Site, situated on the island's N was a large shell midden on Ekaleu Island (Site EKR, \'~i/~ perimeter behind a small mangrove see Chapter 9), but I could not obtain permission to swamp, did not appear promising for ex- work there. The EKE Site on Boliu seemed an equally cavation, being heavily bioturbated by land good choice, and Jason Tyler was assigned the task of crabs (see Chapter 9). In contrast, the extensive EKE directing this excavation. Tyler began work at EKE on Site situated on the narrow neck of low-lying land in October 16, 1988 and worked there continuously until the island's center presented an attractive prospect for November 24. I was present during the initial period excavations. A series of large, undulating shell midden of site mapping and opening up of the first transect mounds (up to 1.5 m high) suggested relatively deep units, and followed the progress of Tyler's excavations cultural deposits, and on their surfaces we found sev- until I left Mussau on October 29. Tyler also received eral Terebra-shell adzes, a Trochus-shell armband frag- considerable assistance from Nick Araho of the PNG ment, some obsidian flakes, and two small potsherds National Museum during the first two weeks of exca- of a dark reddish color with a dark mineral temper. vations at EKE. Shell midden littered the surface, with large Anadara, Tyler, who was to have written up the EKE Site Tridacna, Strombus, and Trochus shells being especially report under my supervision, prepared the final site common. maps and a preliminary stratigraphic interpretation, but The landowner, Joseph Kuiavua, whose house stood failed to complete a final written report. It has thus fallen near the site, was receptive to the idea of excavations, to me to author this chapter, using Tyler's and Araho's We were unable to carry out such work in 1985, how- field notes. Fortunately, the notes and records were well THE BOLIU SITE 163 maintained throughout the excavation, and the task has tively level plateaux, the latter with a thin soil veneer been fairly straightforward, aided by the fact that I was intensively cultivated for taro, yams, and manioc. A single able to observe the first completed transect pits and large Syzigium fruit tree stands sentinel on top of the their stratigraphy prior to leaving Mussau in 1988. NE block, and a low density midden scatter was ob- served near it, which was also test excavated (see be- THE SETTING low). The N coastline is rimmed with a small mangrove swamp, and the EKF Site lies between these mangroves Boliu Island, one of the smaller offshore islands with a and the limestone escarpment. surface area of - 1 kiM2, lies between Eloaua and Mussau Sea cliffs surround most of the SW uplifted block, islands. Boliu is surrounded by extensive reefs and and a typical incised notch marking the mid-Holocene seagrass beds, with a large lagoonal basin to the W, + 1.5 m sea level stand is clearly visible; a cross-section making it an excellent base for fishing and shellfish gath- of this was measured and is illustrated in Chapter 2 (see ering. The island is shaped rather like a barbell (Fig. 6.1), Fig. 2.16). This high stand played a significant role in the with blocks of upraised reef limestone at the SW and island's recent geomorphological history, for the low- NE ends, connected by a low-lying spit of unconsoli- lying sandy spit that connects the two uplifted limestone dated calcareous sands (on which the EKE Site is situ- blocks would have been tidally awash. The fall in rela- ated). The SW limestone block has a maximum eleva- tive sea level to its modern stand resulted in deposition tion of 20 m, and the NE block 26 m asl. Both of of the calcareous sands and silts that make up the spit, these elevated blocks have steep escarpments and rela- accreting simultaneously from the SW and NE, and even- FIGURE 6.1 Map of Boliu Island, showing the general location 9 of sites EKE and EKF _ _EKE _t |~~~~~~~~~~K MALL C&dV: 26 mas E u s a u <0ttE -EQ I 20 m as i : { sln 164 THE BOLIU SITE tually joining the two separate islets together. stone escarpment at the N perimeter of the mapped The EKE Site covers most of the low-lying sand area; a similar steep rise occurs just S of the mapped spit between the two uplifted blocks, as shown in Fig- area as well. As can be seen from the topographic con- ure 6.1. Tyler made a detailed surface topographic map figuration, if one walks E from the island's present W of much of this area, with a contour interval of 10 cm, shoreline following roughly the N250 transect line, one shown here in Figure 6.2. From this map a three di- first crosses a low sandy flat (- 0.2-0.4 m asl) covered in mensional block diagram projection was computer-gen- typical strand vegetation (Wesserchimidia, Scaeola, Panda- erated, here reproduced as Figure 6.3. The map and nus, Casuanrna), -40 m wide. There is then an abrupt rise block diagram clearly show the edge of the steep lime- in elevation from - 0.4 to - 1.2 m asl. On top of this FIGURE 6.2 Topographic map of the EKE Site area, with contour intervals at 10 cm. Grid is oriented N-S. -375m 0.6 14 ., -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~- - --- 0.4 1 0 350 0 18 04 ~~~~~~~~~~~~~~~~~~~~~~-325 0.6 300 1.6 0'6 12 275 1.4 0.4 02~~~~~~~~~~~~~~~~~~~~~~~5 08 1.0 1.2 1.2 -225 200 1.6~ ~ ~~~0 1.0 175 THE BOLIU SITE 165 FIGURE 6.3 Three-dimensional block diagram representation of the surface topography of the EKE Site, with vertical exaggeration x 10. rise the ground surface changes dramatically from loose drop from 1.2 m back down to 0.4 m asl, and an- sand to a grayish-black loam, obviously rich in charcoal other sandy flat largely devoid of surface cultural mate- and other organic materials, and littered with large num- rials. This low-lying sandy flat continues until one reaches bers of gastropods and bivalves, many showing typical the island's E shoreline another 120 m distant. breakage patterns due to meat extraction. It is thus evi- dent that one is now standing on a deposit which is at EXCAVATIONS least partly anthropogenic, made up in large measure of shell midden and other cultural materials. During EXCAVATION PROCEDURES our initial reconnaissance, we found several Terebra-shell adzes, obsidian flakes, and other artifacts in this area. Based on our close examination of the EKE Site area Although much of this elevated ridge is nearly level, while laying out a metric grid system in mid-October there are also three or four discrete mounds which rise 1988, it was clear that the site was both extensive (with above the surrounding terrain (Fig. 6.4); these can be an area of - 2 ha) and probably complexly stratified. clearly discerned on the topographic map, with grid Large areal excavations were not within the scope of centers at roughly E115N375, E160N335, E150N285, our project, and the best approach to gain an overall and E200N175. These mounds appeared to have been assessment of the site's stratigraphy and range of cul- culturally deposited, probably as midden dumps, an hy- tural deposits appeared be to apply the transect testing pothesis later confirmed by our excavations in two of method used at Site ECA. We realized that there would them. Continuing this "walking tour" along the N250 be a potential drawback in that it might not be possible line, at 130 m from our starting point on the island's to precisely correlate stratigraphic zones from unit to W coast, we reach another major topographic break, a unit (especially as we had a spacing interval of 25 m 166 THE BOLIU SITE FIGURE 6.4 View of the central part of the EKE Site, showing the low, undulating mounds under coconut palms and other economic trees. between units), but felt that sampling more of the total the original transect unit, to create a lx2 m excavation site area was more important than concentrating on one block. This was done to increase the sample size of or two smaller sample areas. cultural materials and to clarify aspects of stratigraphy With this approach and objectives in mind, a grid was at these locations. Along the N200 (E-W) transect line, laid out over the site after the low brush growing under we excavated three units between EB 00 and E150, and coconut, Inocarpus, Syzigium, and other large trees was a fourth unit at E325. After these transect units were cleared by hand with bush knives. The grid was oriented completed, an additional 1 m2 test was excavated at so that it was aligned with the main axis of the sandy spit E1 17N372, in the top of another evident shell midden which makes up the central part of Boliu Island (Fig. 6.5). mound situatedE ofJoseph Kuiavua's house, an area where Stakes were placed at 25 m intervals, and surface eleva- Terebra-shell adzes had been noted on the surface. Finally, tions were taken with a Leitz level at every 10 m, using the another test pit was dug on top of the elevated limestone high tide line on the island's W coast as a datum.' We plateau N of the site, where a small midden deposit chose the E200 line as the primary transect for initial was observed in the vicinity of a large Syzigium fruit excavations, as this ran N-S down the center of the tree. In all, a total of 19 m2 was excavated at the EKE - 1.2 m high "midden flat" that made up the site, and Site. Excavation, screening, and recording procedures also bisected the highest of the discrete mound fea- followed standard Mussau Project methodology, as tures, with a high point of 1.7 m asl in the vicinity of described in Chapter 3. E200N175. A second, E-W cross-cutting transect was chosen along the N200 line. Elevation profiles along THE E200 TRNSECT these two transect lines are shown in Figure 6.6. Test units (each 1 in2) were excavated at 25-in in- The stratigraphy and cultural content of the EKE Site is tervals along the E200 transect from N350 to N 100, as best described by proceeding along the E200 transect shown in Figure 6.5. In two locations (at N305 and from S to N, as the deposits can be grouped into roughly N 174-175) a second unit was opened up adjacent to four horizontal zones or sets of facies. THE BOLIU SITE 167 375 m 1.4 Ouse 0.4 1. 035 N35035 0. 1.4 0.4 -325 .20W\t\ 1.2 'N300E200 0.6 0 N305El199 4 N305 -300 .616/ .~ 0.6 :; V X ;::f:; V V :;a:;Xt;:E:;0: : \ 1 squarem1.2 N275 - 275 0 U 0.4 N250 ~~~~~250 0.2 1.2~ ~ ~~~~~~~~~2 E100\ El. 25 N200 Transect (i) E325 1.0~~~~~~~~~~~~2 : :Xa: ::0 : :; ;: : :tu;0 f :: ; ;0:$40 00: :;;0: ti :0010 cm contour interval 100 FIGURE 6.5 Plan of te EKE Site showing: the main N-S and E-W transects, location oftest excavation units, and other features.175 200225250 and~~~~~~~~~~~~~~~~~~~~~~~~~~ ote cmeatures.teva 168 THE BOLIU SITE 30- 28- 26- 24- E _ 22- ? 20- 8) 18- \E+ = xcavation 18\ units 16 Vertical scale: Xl 0 0 .0 c 14- 12- 2 CU 10] Limestone 01 + + loJ1- escarpment \ escarpmen E86 100 150 200 250 300 350 365 8 N200 E-W Transect 6- 4- 2- 0 N500 450 400 350 300 250 200 150 100 E200 N-S Transect FIGURE 6.6 Elevation profiles along the N-S (E200) and E-W (N200) transect lines at Site EKE. Vertical scale exaggerated x 10. DEPOSITS AT THE S END OF THE TRANSECT is calcareous sand tempered, with some red slip, in many respects similar to the Lapita plainware assemblage from The S part of the transect covers the area from N100 Area A at Site ECA. to N150 along the E200 grid line, at the base of the limestone escarpment rising up to the island's SW up- THE SEQUENCE AT UNITS E200N I 74-175 raised block. In the vicinity of E199N100, limestone outcropping was visible on the surface, and the total At E200N175, we were now atop the largest of the depth of soil was only 30-40 cm, when cemented reef shell midden mounds with a surface elevation of - 1.7 limestone was exposed in the base of the excavation. m asl, and some 30 cm higher than the surface elevation Shell midden was sparse, and the only artifact recov- at E200N150. This excavation, which penetrated to a ered was an armring fragment. Twenty-five m to the total depth of 200 cm bs, yielded a more complex se- N, the concentration of shell midden began to increase, quence than the three units to the S, and to gain a larger with a few worn sherds and some obsidian, and the sample and clarify the stratigraphic sequence, Unit depth of deposit increased to 65 cm bs. In the N174E200 was opened upiadjacent to the first test. In E200N 150 test, a significant concentration of calcare- brief, we discovered that the deposits containing cal- ous-sand tempered, reddish plainware sherds began to careous-sand tempered Lapita plainware- which had appear at ~60cm bs, and continued until near the base been exposed in E200N 150 to the S-continued be- of the excavation at 140 cm bs. These ceramics did not neath the high shell midden mound, which had later include any decorated sherds, but most of the material capped them with 120 cm of post-Lapita shell midden THE BOLIU SITE 169 deposition. Thus two separate occupation phases were tested to 200 cm bs, and proved to be com- present at the E200N174-175 locale. pletely sterile below 180 cm. Strat%iraphy In depositional terms, two major episodes are evi- The stratigraphic profile of the W face of Unit denced. During the earlier period, represented by Lay- E200N175 is shown in Figure 6.7, and a view of the ers II and III, this area was an open beach foreshore, completed excavation in Figure 6.8. The main deposi- and the presence of abundant venerid bivalves in Layer tional units were as follows: m indicates that it was deposited subtidally (these shells Layer IA. 0-70 cm bs. Very dark brown (1OYR2/ were naturally deposited, not midden material). A Lapita 2) loam with much shell midden, with a period occupation was associated with this depositional "greasy" texture due to the high concentration phase, likely involving one or more stilt-house struc- of fine charcoal and organic materials charac- tures as no features indicative of an on-ground occupa- terize this first layer. Within this layer at - 20 cm tion were exposed (and, the area would have been tid- bs, was a thick bed of concentrated, large shells, ally awash). The interface (Layer IC) containing some probably a single dumping event. clay content between Layers II and IB seems to repre- Layer IB. 70-120/130 cm bs. The transition to this sent a period of landscape stability when no deposition lower part of the shell midden deposit is gradual took place, and is likely to postdate the drop in sea level with no sharp break. In this lower zone the from the mid-Holocene + - 1.5 m stand, and thus the midden was very dark grayish brown in color progradation of the shoreline N of this area. The in- (10YR3/2), and not as "greasy" as in the upper corporation of a significant quantity of calcareous pot- component, but still contained large midden tery in this zone could well have resulted from the bur- shells. rowing activities of terrestrial crabs. After some time, Layer IC. At the base of the deposit, just above the deposition at this locus began again, now in the form interface with Layer II, was a zone of light of shell midden dumping and occupation in the imme- brown to gray sediment containing some clay diate vicinity (given the high charcoal content and pres- mixed with calcareous sand. This transitional ence of oven stones). Deposition was fairly rapid, and zone is likely to represent an old soil surface (A led to a vertical accumulation of more than 1 m of horizon), which may have been vegetated and shell midden deposit. stable for some time prior to the deposit of Layer I. Although in the profile this zone ap- Cultural Content pears to be only 5-10 cm thick, the excavation Table 6.1 summarizes the stratigraphic distribution of notes indicate that it spanned as much as 20 cm some of the main categories of cultural materials in the in some parts of the unit. Calcareous-sand tem- EKE Site. As can be seen, important changes are asso- pered pottery first appears in this deposit. ciated with the transition from Layers 11 to IB. Calcare- Layer II. 130-150 cm bs. Coarse-grained calcare- ous-sand tempered pottery, present in significant quan- ous sand, white (1OYR8/2) in color, with very tities in Layers Il and II, as well as in the transitional IC little shell but a significant quantity of reddish, deposit, are absent in Layers IB and IA. On the other calcareous-sand tempered plainware ceramic hand, significant numbers of burnt coral oven stones sherds occur in this layer. There were pockets occur in throughout Layer I, indicative of cooking ac- of heavily concreted (CaCO) sand distributed tivities in the vicinity. The differential concentrations of throughout the layer. shell midden are also notable. Layer III. 150-180 cm bs. Fine-grained calcareous sand, white (1OYR8/1) in color makes up this Radiocarbon Dating layer, with small Veneridae bivalves included. Four samples, all consisting of midden shell (Tridacna Ceramics and other cultural materials were and Hippopus valves) were submitted from Unit scarce, and largely confined to the upper part E200N175 for radiocarbon dating (see Chapter 10 for of the deposit, but did include a single dentate- sample details). The upper two samples, from Layer IA stamped Lapita sherd. This sand deposit was (Beta-30690) above the thick lens of shells and from 170 THE BOLIU SITE 1 4C Dates Level 910 f70 BP 1 Oc 2 Very dark brown 870 + 60 BP 3 50 4 Tridacna 5 IB 1 OYR3/2 _ 0 Very dark grayish brown 1 140 +60BsP 6_ Oven stones Interfa-ce I C 7 8 mConcretions 1 OYR8/2_ l | White < _ 150 3420 + 70 BP 9 (Lapita Sherds) - III 1 0 |1 OYR8/1 - ~White_ 101 1 OYR8/1 1 2 White_ _ - 200 Unexcavated 0 50 100cm FIGURE 6.7 West face of Unit E200N1 75 at the EKE Site. Sediment samples See text for discussion. Layer IB (Beta-3069 1), yielded conventional 14C ages of the stratigraphic sequence in this area accumulated of 910 + 70 and 870 + 60 BP, while a sample from quite rapidly, probably within the period from cal AD Layer IC (Beta-30692) yielded a slightly older age of 900-1500 (see Chapter 10). A single sample of Hippopus 1140 + 60 BP. These three dates suggest that the entire shell from Layer II (Beta-30693), associated with the shell midden mound which makes up the upper half calcareous-sand tempered Lapita plainware, yielded a THE BOLIU SITE 171 FIGURE 6.8 View of EKE Unit E200N1 75 after excavation. ._ conventional age of 3420 + 70 BP (1416-1263 cal BC), from the underlying Layer II, and deposited them on a entirely consistent with our interpretation of a Lapita stable land surface. period occupation. These four dates also confirm our stratigraphic interpretation of a long depositional hia- UNIT E200N200 tus associated with the Layer IC interface, and heighten the likelihood that the presence of calcareous-sand tem- Moving N along the E200 transect line the next unit pered sherds in this zone reflects a long period of land- was situated on the slope of the same large midden crab burrowing which would have brought sherds up mound sampled by E200N174-175. Not surprisingly, TABLE 6.1 Cultural content of Unit E200N1 75. Category Layer IA/B Layer IC Layer II Layer III Ceramic Sherds 6 74 47 20 Flakes 5 10 4 - Shell Adzes 3 - - Worked Trochus (no.) 12 12 - 9 Total Bone (NISP) 125 38 16 2 Pig Bone (NISP) 8--- Oven Stones 50 28 -- 172 THE BOLIU SITE then, the stratigraphy was somewhat similar here to that was building at this time, it may have been that the locus found 25 m to the S. In E200N200 the upper 40-50 cm of occupation had shifted from the S part of the transect bs consisted of a black to dark brown, "greasy" tex- to this area, remaining on the prograding shoreline. tured shell midden (Layer I), which then changed to a brown sand and clay loam mixture (Layer II) to a depth DEPOSITS AT THE N END OF THE TRANSECT of - 100 cm bs. Terebra-shell and Tridacna-shell adzes, along with small quantities of a dark red, non-calcare- The final part of the E200 transect lies from N305 to ous tempered ceramic ware were recovered from these N350, an area where the ground surface rises slightly as layers. A sample of Laevicardium midden shell (Beta- one approaches the base of the steep limestone escarp- 30694) from the base of Layer II was radiocarbon dated, ment. In these test units, the deposits were shallow, and yielding a conventional age of 540 + 70 BP (cal AD 1689- consisted of a thin upper midden deposit (darker in 1910), suggesting that this entire deposit may have ac- color, but not so concentrated as in units to the S), over- cumulated rapidly, quite possibly as "sluff" or wash lying gray to white calcareous sands containing Lapita down the slope of the midden mound. plainware ceramics. In Unit E200N305 a dentate- Afteratransitionalzoneoflightbrownsand(Layer stamped Lapita sherd was also found. In Unit III), the profile changed first to white, coarse-grained N350E200, which lay at the base of the steep scarp, the calcareous sand (Layer IV), followed by white fine- deposit was no more than 10-40 cm thick, overlying grained sand containing numerous venerid bivalve shells cemented reefal limestone; only one worn sherd and (Layer V), to a total overall depth of 200 cm bs. These three obsidian flakes were found in this unit. In essential lower deposits (Layers IV and V) thus correlate with respects then, the N part of the E200 transect mirrors the lower layers in E200N174-175, and like the latter, the sequence in the far S part of the transect. also contained quantities of calcareous-sand tempered Lapita plainware. THE N200 TRANSECT DEPOSITS FROM N225 TO N275 The N200 transect cross-cut the E200 transect on an E- W axis, joining at Unit E200N200 (Fig. 6.2). The Units N225, N250, and N275 along the E200 transect westernmost unit, E100N200, located near the present are completely off the high shell midden mound, on shoreline on the low sandy flat, proved to be com- the relatively level flat which is - 1.2 m asl (see Fig. 6.2). pletely sterile. Moving E, E125N200 was also on this In these three units, the stratigraphic sequence was es- flat, just below the 1.2 m rise, and yielded only a mini- sentially identical, with only minor variations, consisting mal amount of cultural material, including one armring of: (1) a relatively thin upper deposit of dark gray to fragment. black, "greasy" textured shell midden (typically only 20- Unit E150N200 was situated on the slope itself, at 25 cm thick), with Terebra-shell adzes and a few non- - 1 m asl. This was excavated to the water table at 100- calcareous sand tempered sherds; (2) overlying grayish- 120 cm bs, but no ceramics were recovered. Rather, the brown to white calcareous sands containing calcareous- upper deposits consisted of shell midden, and produced sand tempered, reddish plainware, and also including a a Triacna-shell adz and a worked pearl shell disc. Thus, few sherds with incised decorations. Again, this sequence this part of the EKE Site seems never to have had a is consistent with an interpretation of an earlier Lapita Lapita period occupation. period occupation, followed by a much later occupa- The final unit along the N200 transect was tion characterized by heavy shell midden dumping. In E325N200, near the island's W shoreline, again on a this area, however, the presence of incised sherds in the low-lying sandy flat. Here there seems to have been some lower deposits suggests that the Lapita occupation was occupation, as both a Terebra-shell adz and a Tridacna- towards the late end of the Lapita period, perhaps com- shell adz were recovered, along with four small sherds parable in time to the Area C: deposits at Site EGA. If, of uncertain type. The water table was reached at only as we suspect, the shoreline was rapidly prograding and 40 cm bs, and the deposits may well have been mixed the sand spit connecting the two parts of Boliu Island through land-crab burrowing. THE BOLIU SITE 173 THE SEQUENCEATE! 7N327(TPI) in excavating the site was to obtain a good sample of the post-Lapita period in Mussau, we were only partially suc- After the transect excavations were completed, an ad- cessful, for the Boliu sequence appears to have a lengthy ditional 1 m2 test unit was excavated at grid Unit hiatus between these two occupation phases. Nonetheless, El 17N372, into the top of another shell midden mound a good sample of material culture and faunal materials on which Terebra-shell adzes had been found. The exca- dating to the mid-second millennium AD was obtained, as vation reached a total depth of 180 cm bs, at which well as additional insights to the Lapita period. cemented reefal limestone was exposed. The stratigra- Initial human occupation-or at least use-of Boliu phy consisted of two components, which graded into Island dates to - 1400-1200 cal BC, when the Lapita each other: (1) an upper deposit (Layer I) from 0-80 plainware and a few dentate-stamped sherds were depos- cm bs, of black (10YR2/1), "greasy" textured shell ited in subtidal sandy deposits at both the N and S ends midden, much like that in the upper deposits at of the E200 transect. At this time we would infer the E200N174-175; overlying (2) a very dark grayish brown + 1.5 m higher sea level to have been in existence, so (10YR3/2) loam (Layer II) in which the quantities of that the sandy spit now connecting the two upraised shell midden decreased significantly with depth. limestone blocks making up the island had not yet been Layer I produced seven Terebra-shell adzes and a deposited. Boliu was thus two islets, with a narrow chan- large quantity of shell midden. The only potsherd in the nel or tidal flat between them. The deposits of Lapita unit was recovered from the interface of Layers I and sherds, obsidian, and other cultural materials at the N II (80 cm bs), and is a body sherd with unique punctate and S ends of the transect seem to imply the presence decorations (see Volume I). Another Terebra-shell adz of two small Lapita sites, each just off their respective was found near the base of the excavation, at 160 cm escarpments, facing each other across a channel per- bs. Although no radiocarbon dates were run for this haps 75 m wide. The quantity of ceramics and other unit, it seems likely that the entire deposit is relatively cultural materials, however, is hardly sufficient to sug- late, certainly within the last thousand years. The period gest that these were permanently occupied hamlets, and of deposition probably corresponds to the upper part the dominance of plainware with only a handful of deco- of the shell midden deposit in E200N174-175. rated sherds, also contrasts strikingly with the situation at even the smaller ECB and EHB Lapita sites (see Chap- TESTEXCAVATION ON THE ter 4). It seems likely that these were not permanently UPRAISED PZATEAU (TP2) occupied settlements, but temporary or intermittently utilized sites, perhaps fishing camps,2 and/or houses While reconnoitering Boliu Island, Tyler observed a small where persons planting tree crops or gardens on the midden scatter on top of the limestone plateau N of limestone plateaux of Boliu stayed temporarily. What- the EKE Site, in the vicinity of a large Syzigium fruit ever their function(s), these minor Lapita sites add an- tree. A 1 m2 test unit was placed - 8 m from the base other variant to the range of Lapita settlement types of the tree. The deposit consisted of a reddish brown evidenced in Near Oceania. soil overlying limestone which was reached at 30-40 cm This kind of relatively ephemeral or temporary use bs. Cultural material was confined to the upper 20 cm, of the area between the two uplifted limestone blocks and consisted of a substantial amount of shell midden of Boliu probably continued for some time, even as (27 kg), 81 oven stones (2.8 kg), a small quantity of the sea level dropped and the sand spit began to accu- bone, and four small, eroded sherds with non-calcare- mulate along the central portion of the E200 transect, ous temper. The deposit was not dated, but is clearly as evidenced by the deposition of plainware in this part post-Lapita in age. of the transect, including some incised sherds compa- rable to those from Area C of ECA, or from the EKQ SUMMARY rockshelter site. Eventually, perhaps by early in the first millennium ADr, the locus seems to have been completely The EKE Site has two major phases of occupation, abandoned for habitation. Then, by 700 BP but possi- one of Lapita age and cultural affinity, the other dating bly earlier, renewed occupation commenced, this time to the second millennium AD. Given that our objective associated with shell midden dumping and the accumu- 174 THE BOLIU SITE lation of several large shell midden mounds. The mate- well as significant quantities of pig bone (Nevermann 1933; nal culture associated with this phase is not unlike that eth- see Chapter 2). Also noteworthy are small quantities of nographically documented for the German colonial pe- non-calcareous tempered hard-fired red pottery, pre- nod, with Terebra-shell adzes and Trochus-shell armrings, as sumably trade ware (possibly of Admiralty Islands origin). NOTES TO CHAPTER 6 1 Tyler recorded the tidal range along this shoreline over several 2Indeed, given that the channel between these two Lapita sites days, finding that the maximal range was 0.5 m. at the N and S ends of the transect would have been tidally awash, an effective fishing strategy might have been to lay a seine net across this channel.