CHAPTER 9 SITE SURVEYS AND MISCELLANEOUS EXCAVATIONS (SITES EKL, EKU) PATRICK V. KIRCH WITH MARSHALL 1. WEISLER \ In addition to the excavations reported central plateau extensively used for swidden cultivation in Chapters 4 through 8, two minor of root crops. Today settlement is concentrated along $ excavations were carried out at sites the E-central coastline. In addition to the several sites A EKL and EKU. We also made recon- on Eloaua which were excavated (ECA, EHK, EHM, i naissance surveys throughout the off- EHN, EKO; see Chapters 4 and 5), we recorded the shore islands, and to several localities on following sites: Mussau Island, locating a total of 39 ar- chaeological sites (Fig. 9.1). This chapter Site EKA records the results of these minor excava- Coordinates: GU898278. This site, which has the top- tions and of our surveys. The latter may be of particu- onym Saumerei, is reputedly the location of a village or lar interest when renewed archaeological work is car- hamlet which was still occupied at the time the SDA ried out in Mussau in the future. The first author de- mission arrived (ca. 1930). Near the W end of Eloaua, scribes surveys and sites on the offshore islands, Mussau the site is on a sandy flat which abuts the uplifted lime- Island, and in the SE Region. The second author dis- stone escarpment. Shell midden and some oven stones cusses the NW Region. The grid coordinates given for were noted on the surface, but no pottery or obsidian surveyed sites are all in the Universal Transverse Mercator was observed. Some large fruit trees are still standing in system, Zone 55. this location. RECONNAISSANCE SURVEYS Site EKB Coordinates: GU899275. This coastal site, bearing the THE OFFSHORE ISLANDS toponym Tapakevouvou, was said by Aimalo Lavatea to have been a former village site. The shoreline has ELOAUA ISLAND eroded away (ossibly as much as 50 in), leaving a de- flated "pavement" of large midden shells (Hyotissa, As seen in Figure 9.1, Eloaua is by far the largest of the Tridacna, and other taxa) and oven stones on the reef offshore islands in the Mussau group, with an elevated flat at the foot of the beach. 188 SITE SURVEYS AND MISCELLANEOus ExCAVATIONS Cl) M C#) 4),) C E o~~~~EC 04 U~~~~~ 7C) CY) LLI Cl)~~~~~~~~~~~~~~~~~~~~D C (5~~~~~b > 7C)~~~~~~~~~4 U)~~~~~~~~~~~~~~~)L U) U)4)U) OL C) ~~~~~~~~~~~~~C)--- rCD 4) 0 b 0~~~~~~~0 ODD U)~~~~~~~~~~~~~~~~~~~U L L~~~~~~~~S ~ ~ LjU L - -C Cl) 0 _ _ _ C7D U ---v---~~~~~ ~~~~~~~~~~~~~ 7- C) SITE SURVEYS AND MISCELLANEOUS EXCAVATIONS 189 Site EKH ing pottery sherds and obsidian flakes throughout the Coordinates: GU928264. This site consists of a scatter area, but only a single incised sherd. I attempted to gain of shell midden (including Anadara, Conus, Lambis, permission to excavate here in 1985, but the landowner Chama, and Cypraea species) on the upraised limestone was away from the island, and none of the family mem- plateau, just above the steep escarpment overlooking bers locally resident were willing to give permission in the lagoon shore of Eloaua. The shell was chalky and his absence. often burnt as a result of swidden gardening. The site extends from an area of second growth into more Site EHD mature forest, and covers an area of at least 0.5 ha. I Coordinates: GU869265. Also discovered by Allen and searched for both obsidian and ceramics, but none were Specht in 1984, this is reported as an open shell midden observed. scatter, lacking pottery, - 100 m inland of Taluemalunga Village. Site EKI Coordinates: GU928264. This is a natural fissure in the Site EHE limestone cliff just seaward of Site EKH. The fissure Coordinates: GU873260. On the inland plateau of was not explored as we did not have ropes or climbing Emananus, this is a traditional habitation locality visited equipment, but informants said that there were human by Allen and Specht, whose informants said it was called skeletal materialsnwithin. Etatokatangaro. About 15 minutes walk from Taluemalunga Village, it is a shell midden scatter at which Site EK/ two small sherds were also found. Coordinates: GU942262. This is a scatter of shellfish midden, of undetermined extent, in scrubby second Site EHF growth, on the elevated plateau just above the bluff Coordinates: GU873265. About 100 m inland of which descends to the coast near a small hamlet. The Taluemalunga, this is another traditional habitation site range of shell taxa present was similar to that at EKH. visited by Allen and Specht in 1984, and reported by them to be a small shell midden scatter. Site EKK Coordinates: GU945260. A fairly extensive area of shell- Site EHG fish midden in a dark gray to black soil, covering per- Coordinates: GU871260. Like EHF, this traditional haps as much as 1 ha. Some obsidian flakes were ob- habitation locus (shell midden scatter) - 10-15 minutes served on the surface. walk inland (S) of Taluemalunga Village, is another vis- ited and reported by Allen and Specht. EMANANUS ISLAND Site EHH Emananus is the second largest of the offshore islands, Coordinates: GU875265. With the toponym Emak, this and like Eloaua has a central plateau used for shifting is the final traditional habitation locus seen by Allen and cultivation. Modern settlement is confined to a few small Specht in 1984, a shell midden scatter said to be 100 m hamlets along the lagoon shore. In addition to the exca- inland of Talemalunga Village. vation of Site EHB (see Chapter 4), we recorded the following sites: Site EKT Coordinates: GU880251. This site is a shallow cave or Site EHC rockshelter located on the island's SW coast, immedi- Coordinates: GU874266. This site was recorded by ately behind a mangrove swamp. Allen and Specht during their 1984 reconnaissance and bears the toponym Taluemalunga. Allen et al. (1984: 10) BOLIU ISLAND describe it as follows: "Throughout the village area is a series of shell mounds standing up to 150 cm above Boliu Island, which lies between Eloaua and the main the ground surface and extending for 250 m along island of Mussau, actually consists of two upraised lime- the beach and up to 40-50 m inland." They report find- stone blocks, connected by a narrow sand spit, giving 190 SITE SURVEYS AND MISCELLANEOUS EXCAVATIONS the island a dumbbell shape. The island is surrounded EKALEU ISLAND by extensive fringing reefs, and a large lagoonal basin lies to the W, a rich source of fish and invertebrates. Ekaleu is a small island situated at the NW end of the Today the island has only a single, small hamlet occu- Malle Channel, - 3.5 km from Eloaua. There is a pro- pied by one extended family group. Boliu Island was tected sandy beach on the SE where canoes can land initially reconnoitered during the 1985 field season, when readily, while the W coast has a small fringing reef. The two sites were discovered (EKE and EKF). The EKE island is low-lying and was probably an awash reef dur- Site was intensively investigated during the 1988 field ing the mid-Holocene higher sea-level stand. Ekaleu Is- season, and results are reported in Chapter 6. land was the locus of a coconut plantation established during the German colonial period, and the island is Site EKF still covered in tall coconut palms. It is also home to a Coordinates: GU948307. This open midden deposit is large population of coconut robber crabs (Birgus latro). situated on a narrow, low-lying coastal beach terrace which lies at the base of a steep slope (the upraised Site EKR limestone escarpment of Boliu Island) behind a small Coordinates: GU879308. The site is located at the SE mangrove swamp. There was a thin scatter of shell end of the island and consists of a large midden mound midden on the surface, probably brought up by land atop which sits the abandoned house and copra drier crabs, whose burrows were everywhere and have prob- of the German plantation. The mound surface is lit- ably extensively disturbed the cultural deposits. One small tered with large midden shells (Lambis, Tridacna, Anadara, Tridacna-shell adz was found on the surface, near the Strombus, Conus, and other taxa), and several Terebra-shell little inlet where we were able to land our dinghy be- and Tridacna-shell adzes were found as well. The height tween the cliff and the mangroves. of the mound (up to 3 m asl) suggests that there may be a fair depth of deposit, and the excavation potential EBOLO ISLAND seems excellent. We intended to excavate at EKR, but were unable to secure permission from the landowner. Ebolo Island was briefly reconnoitered on September 18, 1985, when we landed on the SW coast at an inlet in EBANALU ISLAND the mangroves surrounding much of the island. We walked across the island along a roughly N-S transect, Ebanalu is a small-sized island, mostly consisting of el- through a grove of coconut palms to the NE end where evated reefal limestone, with small coastal terraces on there was a copra drying shed. The terrain throughout the N and S sides. The island lies - 1 km S of Mussau the coconut plantation was fairly open, with flat, low- Island and is surrounded by extensive reefs. The up- lying calcareous sandy sediment. Given the low-lying raised plateau is extensively gardened by a small popu- nature of the terrain (maximum elevation is only 1.5 m lation residing in a hamlet on the NE tip of the island. asl), Ebolo Island was probably a reef during the mid- Holocene higher sea-level stand. Over much of this ter- Site EKV rain in the island's center we observed a low-density Coordinates: GU875338. Shell midden and artifacts scatter of shell midden, and - 10 small obsidian flakes were found to be distributed over virtually the entire (the latter were collected). The midden and flakes sug- surface of this small island, and the site number thus gest that the island was occupied at some time, most pertains to the island as a whole. Locally, the density of likely in the post-Lapita period as no sherds of any kind shell midden varies, but no areas devoid of midden were observed. Excavation potential appeared to be were seen. In some places the shell midden is remark- poor, except for a low ridge just inland of the NE ably dense, as on parts of the elevated limestone pla- point, where the shell midden appeared to be slightly teau exposed by gardening. We collected numerous more concentrated, and the possibility of some intact Tridacna-shell and Terebra-shell adzes, as well as a few stratigraphy was greater. small potsherds and obsidian flakes. On the S side of SITE SURVEYS AND MISCELLANEOUS EXCAVATIONS 191 the island we noted a substantial rockshelter in the face Site EKC, Lomakunauru Village of the limestone escarpment, with a width of - 10 m Coordinates: GU996295. This extensive site covers the and depth of 28 m. There appears to be a substantial area of Lomakunauru Village. Between the modern deposit in the shelter floor, and two Terebra-shell adzes houses, along paths, and in various modern pit expo- were found on the surface. sures, one readily observes a dark gray to black midden deposit and abundant shell midden. There are also a ENOANAULU ISLAND number of low mounds which appear to be midden dumps. I collected two Tnidacna-shell adzes, six Terebra- Enoanaulu, lying - 1.5 km SE of Ebanalu and 1.5 km shell adzes, one small potsherd, and a number of ob- off the coast of Mussau Island, consists mostly of el- sidian flakes from the surface. This site has excellent evated limestone, with a maximum elevation of 58 m excavation potential. asl. There is a small settlement on the NW tip of the island, and some shifting cultivations are cultivated on Site EKM the elevated plateau. Coordinates: GU947322. This site is situated near the mouth of Schadel Bay (on the W side) and can be ap- Site EKW proached from the sea by way of a small inlet through Coordinates: GU888326. This site was only briefly re- the dense mangrove which fringes this part of Mussau connoitered but appeared to be a fairly dense midden Island. Behind the mangroves there is a low sandy flat deposit extending from the coastal flat up onto the pla- with a small grove of coconut palms. Crab burrows in teau. There were numerous small, worn potsherds, and this flat reveal the presence of a buried midden, prob- although none with decoration were seen, the density ably of low shell density. No artifacts were seen. Ave of ceramics suggested that this could be a Lapita pe- Male said that traditional house sites were also located riod site. on the elevated limestone plateau inland of this beach. MUSSAU ISLAND Site EKN Coordinates: GU928322. As with Site EKM, this is a The majority of our efforts in the Mussau Project were slightly elevated beach terrace situated behind a man- concentrated on the offshore islands, for reasons given grove swamp along the S shore of Mussau Island. We in Chapter 1. However, we did make brief reconnais- noted extensive shell midden mounds, similar to those sance forays to the SE part of Mussau, and Weisler at the EKE Site on Boliu Island, and found Terebra-shell reconnoitered parts of the NW in conjunction with his adzes and obsidian flakes on the surface. Some of the excavation of two rockshelter sites. For the most part, midden mounds are 1 m or more in height, and 8-12 m however, Mussau Island remains archaeological terra in- in diameter. cry~~~~~~fliLi ~~~~~~The NW Region THE SE REGION (BY MARSHALL 1. WEISLER) Site EHO, Roitano Village During a stay of several weeks in the village of Tanaliu, Coordinates: HU023347. This site, which had been re- while excavating the EKQ and EKP rockshelter sites ported by Allen and Specht during their 1984 recon- (see Chapter 5), Weisler was able to reconnoiter the NW naissance, is a partly disturbed midden on the grounds region of Mussau Island for other archaeological sites. of an SDA church in Roitano Village. The church is situated on an old beach terrace ~ 150 m inland, and in Site EKZ leveling the ground for the church foundation, shell Ealingasa Cave (Coordinates: GU857491) is located 3.5 midden, obsidian flakes, and sherds of a sand-tempered, km E of Tanaliu Village, a walking distance of 3 hours thin-walled ceramic ware were exposed. We made a from Pomanai Village at the coast. The area surround- small surface collection of obsidian andpotsherds, but ing the site is called Elonatu. I was told of this site by did not excavate. several men aged in their 20s and 30s while staying in 192 SITE SURVEYS AND MISCELLANEOUS EXCAVATIONS Pomanai Village. This site proved difficult to find, and . . . . . . .... . we had to enlist the aid of an older man who remem- bered visiting the site as a youth. . . . . . . Situated at an approximate elevation of 100 m asl, the site was formed by the collapse of a limestone sink Dripline which formed a sheltered overhang (14 m wide by 2.5 A m deep) on the W side. The ceiling height is 1.6 m at the back of the shelter rising to 8.4 m at the dripline (Fig. 9.2). The N extent opens to a cave-at least 25 m Slope long-with active, dense speleothems making explora- . . ::: tion difficult. No evidence of human use was seen. The level floor of the shelter within the sinkhole was par- A A? tially cleared; an earth-oven depression and a few ma- rine shellfish (Mytilidae and Tellinidae) were recorded. A small exploratory hole excavated with a machete to 30 cm revealed a few plain, shell-tempered, body sherds. Site ELA N L KiTh A 10-minute walk from Magean Village (the most N N :: village of Mussau Island) is Etalavevanga shelter (Co- ordinates: GU832542). The landowners are Deku and ... . . . . . .. . . , . . . . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Maena Matua. There are small overhang shelters along / the length of the limestone block at the coast. Some . Cave . . shelters are 2 by 3 m in area with ceiling heights of 1.25 Dense m. The largest shelter with the most concentrated sur- speleothems A? face cultural deposit was mapped with compass and tape. The area inside the dripline measured 10.5 wide . . . . . . . (N-S) by 2.5 m deep. The ceiling height at the rear of . . . . . . the shelter is 80 cm rising to 2.4 m at the dripline (Fig. Grid N 9.3). The cultural deposit extends for at least 2.5 m out- . . . side the dripline; one obsidian flake and marine shellfish . . . includingTridacnagigas,Trochidae, Turbosp.,Cypraeasp., Drupa morum,D. grossularia, and Codakia tigerina were re- corded. , ~. . . . . . . . . . ...... Site . . . . . ..E. . . . . . . . At Ewa, E along the main path for 10 minutes from ELA is another overhang shelter measuring 12 m wide (SE-NW), - 4 m deep, with a maximum ceiling height FIGURE 9.2 Plan of Site EKZ, Mussau Island. of 3 m at the dripline (Coordinates: GU835544). The shelter is adjacent to the inland side of the main path, - 25 m from the beach. The level shelter surface was a tween Magaen and Pomanai villages. Dense vegetation dark, gray-black color with plentiful ash. Turbo sp. ap- on either side of the path, as well as time constraints, peared to be the dominant shellfish. One small pot- preclue de termining the areal extent of the site. Six sherd was collected. obsidian f lakes were collected on the path. Site EKX Site EKY An open site (Coordinates: GU832538) was found on About 4 km N of Pomanai Village are several shelters the first hill S of Magean Village on the main path be- located along the base of an upraised limestone block SITE SURVEYS AND MISCELLANEOUS EXCAVATIONS 193 150 m inland from the coast along the N side of an unnamed drainage. The area along the base of the block was examined in the early morning hours while it was still somewhat dark. One shelter measured 9 m wide and 3 m deep; another, 4 by 2 m. Shellfish and very dark ashy ground suggests intensive hearth activity. No / maps or collections were made.I MISCELLANEOUS EXCAVATIONS THE ENUSAGILA SITE (EKL) Drip::ne ..... .... ~.) / Dripline Enusagila, smallest of the coral islets comprising the Eloaua-Emananus "atoll," is situated midway between I the two larger islands along the S edge of the lagoon __,_____ (Fig. 9.1). We reconnoitered the islet in 1986 and again A :_____ A' in 1988, discovering a small shell midden eroding out of the E sand spit. The islet is low, with a maximum elevation of only - 1 m above sea level. The islet's inte- rior is relatively flat and open, shaded with a canopy of . . . . ~~~~~Edge of large trees (Pisoniagrandis) in which a population of frig- cultural ate birds (Fregata sp.) roosts,' resulting in a surface de- deposit posit of guano over the islet's interior (and a rather rank : / odor!). There are numerous coconut palms and, near . . . : . / the E end of the islet, some fruit trees (Pangium edule, I / Terminalia catappa). Some croton hedges in this area are / said to mark a traditional burial ground ("ples matmat"). The islet's N lagoon shore has been heavily colonized by mangroves, which trap fine sandy sediment resulting Grid N in a lagoonward progradation of the islet. In contrast, the S exterior shoreline is undergoing active wave erosion. o im Erosion of the seaward coast had exposed a shell midden deposit along the SE tip of the islet (UTM coordinates GU909260). The midden was dominated by large Hyotissa hyotis oysters, Chama sp., and Tridacna A A' spp., and much of the shell appeared to have been burnt. _______________________________________A A number of volcanic oven stones were observed in the eroded midden bank. John Saupa of Emananus, FIGURE 9.3 Plan of Site ELA, Mussau Island. who visited the site with us in 1986, said that there had formerly been a small group of houses located here, and he pointed out the rotted base of one wooden house post sticking out of the ground on the surface of limited test excavation would be warranted. On Sep- the narrow E sand spit. tember 30, J. Tyler returned to Enusagila with a local The absence of pottery in the eroding midden bank, crew and excavated a single 1 m2 test excavation into and Saupa's statement that there had formerly been a the highest part of the E spit, near where the shell midden small hamlet suggested that the site might be represen- was actively eroding along the S shoreline. Because the tative of the late prehistoric to early post-contact peni- deposit was essentially uniform throughout, excavation ods. In 1988, after re-visiting the islet, we decided that a was conducted in arbitrary 20-cm levels. The deposit 194 SITE SURVEYS AND MISCELLANEOUS EXCAVATIONS consisted of a brown (1OYR 5/3) calcareous sand. The sherds, obsidian, shellfish midden, and Tridacna-shell and only artifact recovered was a single fragment of bottle Terebra-shell adzes covered an area of - 2.1 ha. The site glass, 30 cm below surface. No bone, pottery, or ob- was brought to our attention by Mr. Aimalo Lavatea sidian was present. The shell midden was densely con- who had collected a handful of sherds on the surface. centrated, totaling 24.85 kg for the entire unit, and much We were particularly interested in this site because of this showed evidence of having been burned, pre- the sherds were clearly not Lapita ware, and were in sumably through cooking the mollusks on open fires. association with Terebra-shell adzes that were known only Volcanic and burned coral oven stones were also from post-Lapita sites. During the 1986 field season, T. present; these are enumerated in Table 9.1. Hunt excavated five 1 m2 test units along a single N-S A complete valve of Spondylus sp. bivalve from level transect positioned across the center of the site (i.e., the 4 was submitted for radiocarbon dating in order to area with surface artifact distribution). The transect units assess the maximum probable age of the midden de- were spaced at 10 m intervals. posit. This shell yielded a conventional age of 330 + 60 The stratigraphy was consistent across all transect BP (Beta-30695), calibrated to AD 1880-1945 at 16, or units and consisted merely of two layers: to AD 1833-1955 at 26 (based on the 20-year marine LAYER I: 0-15/28 cm bs. A black (5Y2.5/1), or- calibration curve with a AR value of 0 + 0). ganically enriched loam (A horizon) and gar- The EKL test excavation confirmed that this shell den soil, with abundant shell midden and char- midden deposit is entirely of historic (post-European coal. The boundary with Layer II is diffuse. contact) age. The site was probably little more than a LAYER II: 15/28 cm to base of excavation. A yel- small hamlet, possibly occupied by a single extended lowish brown (10YR5/4) sandy loam with an- household. The dominant shell midden suggests that gular cobbles and outcrops of decomposing the site was occupied prior to the conversion of the limestone; culturally sterile. Mussau islanders by the SDA mission in 1930, because eating shellfish was prohibited by the SDA missionaries The Layer I deposit clearly represents an in situ oc- (a ban that continues strongly in effect today). Both the cupation, but unfortunately has been somewhat mixed radiocarbon date and the single piece of historic glass and reworked due to gardening. The shallow depth (av- would be consistent with an occupation date in the early erage 25 cm) suggests that the occupation was not lengthy. colonial period, ca. AD 1900-1930. A combined sample of mammal bone (almost cer- tainly Sus scrofa) from Units 2 and 5 was submitted to THESINAKASAESITE (EKU) Beta Analytic for radiocarbon dating, and yielded a con- ventional radiocarbon age of 740 + 70 BP (Beta-25930), Sinakasae is the name of a small contemporary hamlet with a calibrated age range of AD 1222-1284 (see Chap- situated on an upraised limestone terrace on the SE end ter 10 for sample details). This confirmed our hypoth- of Mussau Island, near the larger settlement of esis that the EKU Site post-dated the Lapita period, Lomakunauru (UTM coordinates GU986314). The site and indicates an early second millennium AD age for its largely corresponds to this hamlet, and a scatter of occupation. This is consistent with evidence from the TABLE 9.1 Cultural content of the EKL test excavation. Level Mollusks Oven Stones Taxa Weight (g) Number Weight (g) 1 (0-20 cm) Hyotissa, Tridacna, Laevicardium, Neritidae 800 | 10 1 50 2 (20-40 cm) Hyotissa, Chama, Tridacna, Laevicardium, Anadara 4,000 1 74 800 3 (40-60 cm) Hyotissa, Chama, Strombus luhuanus 8,000 42 800 4 (60-80 cm) Hyotissa, Chama, Spondylus 1 2,000 9 200 5 (80-100 cm) Hyotissa 50 0 0 SITE SURVEYS AND MISCELLANEOUS EXCAVATIONS 195 EKE (Boliu Island) Site for use of Terebra-shell and and a black pyroxene temper. Decorations on some Tridacna-shell adzes at this time period, along with small sherds consist of parallel rows of simple tool-impressed quantities of dark red pottery. rectangular punctations. Aside from the Tridacna-shell The ceramics from EKU will be described in Vol- and Terebra-shell adzes already noted, the site is note- ume III. They consist of sherds with a dark red paste worthy for considerable quantities of pig bone. NOTES TO CHAPTER 9 'It was these birds that first brought Enusagila Islet to dusk one evening. Commenting on the birds to Ave Male, he our attention, when a large flock of perhaps 80 of these mag- told me that they were returning to their roost on Enusagila nificent frigates glided majestically over our Eloaua house at ('ples belong ol').