5 EXCAVATIONS AT THE TO'AGA SITE (AS-13-1) PATRICK V. KIRCH AND T. L. HUNT INTRODUCTION PRIOR Tro Th coMmEN NTrr of fte Manu'a Project in 1986, our knowledge of the archaeol- ogy and prehistory of the Manu'a Islands was lImited to a few late prehistoric architectal sites and to surface finds of stone adzes and flakes (Kikuchi 1963; Emory and Sinoto 1965; Clark 1980). No well- stratified occupation sites were known, and there was no established sequence of occupation phases extend- ing back to the Ancestral Polynesian period, as had been developed for Westem Samoa (Green and Davidson 1969, 1974). bdeed, previous archaeologi- cal reconmaissance in Manu'a had failed to locate any sites that contained prehistonc pottery and thus might date to the first millennium B.C., based on compari- sons with the Samoan sequence as defined for Upolu Island. A major goal of our 1986 reconnaissance survey of Ta'u, Olosega, and Cfiu islands was thus to determine whether the appaent absence of ceramic- bearing sites was simply an artifact of archaeological sampling, or whether such early perod sites were tnuly absent in the Manu'a Group. The surface find of a Polynesian Plain Ware sherd in Ta'u Village on the first day of our 1986 reconnaissance was followed up by test excavations, revealing the presence of subsurface occupation deposits dating to the Ancestral Polynesian period (Hunt and Kirch 1988). This discovery was sufficient to establish that the prehistory of the Manu'a Islands would in general terms parallel that of the larger and better known Westem Samoan group. Moving from Ta'u to Ofu Island, we subsequently discovered a pottery-bearing site at To'aga on the southem coast, where the Public Wo*ks Department had bulldozed a landfill dump and thereby exposed subsurface deposits. This area was within the site AS-13-1 defined by aark (1980). We canied out a limited test excavation at the To'aga landfill in 1986, eover- ing pottery and other artifacts in situ (see below). Because of the significance of the To'aga materals for elucidating Manu'a prehistory, we expanded our excavations in 1987, revealing that the site was far more extensive than originally thoughL A third season of trnsect test excavations was carried out in 1989 to assess the full area of this extensive and deeply staified site. In this chapter, we present the details of these three excavation seasons, including the specific excavadon pxcedures used, as well as the srtigraphy and depositional sequences revealed. Subsequent chapters provide detailed analyses of the ceramics, other portable artifacts, and faunal mater- als, as well as discussions of radiocarbon chronology and site geomorphology. Field Methods: General Comments Some aspects of field methodology which were constant in all field seasons may be brefly summa- rized. The prncipal sampling strategy used was that 44 The To'aga Site of "systematic trasec " (Redman 1974; Kirch and Yen 1982; Kirch 1988), with 1r-m excavation units generally spaced at 10 m intervals. We oriented these transets perpendicuarly to the coastine in order to provide geomorphic proffles across the coastal flat All trae pfeswee canied out to the reef flat and coI to mean sea leveL Usng a tlescopc level and stadia md, a srface elevato prfile was obtained for each trmuect. Observan were also made of soil vanation and vegetation associations along the trasect, as thse data indi- cated cuntland use and provided im clues as to the geomorphic history of the site. This strategy of acquiring data (surface and excava- tion) along systematic transects has the advantage of producing results that can be mapped and interpolated. Whenever possible, excavation proceeded according to natral stigraphy, although arbitray subdivisions were made within thick strata Indi- vidual excavation blocks (refened to as "spits') were designated within strata, but never cross-cut strati- graphic boundaries. Detailed rwcords were main- tained on aarz recording forms (fig. 5.1), on which horizontal feames were drawn to scale and all localized finds were plotted according to x, y, and z coo?rinates. Aftrexcavation, measuwd strat- grapic pofiles were dawn for each excavation unit, and all sta were described in terms of thickness, boundary, morphology, color, lithology, cultual conte, and other caracteristics. Color designations are from fte Munsell soil color charts (Munsell 1988). Most proffile descriptions were done by Kirch to assure consistency (in 1989, Hunt described several units after Kirch left the field to initiate a project in the CDook ls). Following the drawing and description of each profile, a senres of sediment samples were taken; sediment samples were taken from within stata, never cross-cuttng stratigraphic boundaiies. These samples were subsequently analyzed in the University of Califor- nia, Berkeley, geacaeology laboratory (see Kirch, Manning, and Tyler, chapter 7). All excavated sediment was sieved through 0.25 inch mesh screens. Although smaller mesh (particu- larly 0.125 inch mesh) would have enhanced the recovery of minute faunal remains, we opted against tfhis screening stategy in the inreests of covering a greater area duing this testing phase. Given our primary objective of detemuning t areal extent and nature of the subsurface deposits at To'aga, sufficient areal sanpling was judged to be a more important consideation than complete faunal recovery. During screening, however, we made perodic checks of the small size fraction passing thrgh the 0.25-inch mesh screens and were generally satisfied that the bulk of the faunal materals was being retained. In addition, bulk sediment samples were taken from strata in the areal excavation blocks, as well as from some of the individual test pits (see Nagaoka, chapter 13). These samples allowed us to assess the frequency of minute faunal remains. All vertebrate and invertebrate faunal materials caught in the sieves were bagged and shipped back to the laboratory for identification and analysis. Prelimi- nary sorting of the vertebrate fauna, as well as detailed identifications of the avifaunal remains, were carried out by Dr. David Steadman of the New York State Museum (see chapter 14). Furfter analysis of the non-bird vertebrates and of the invertebrate fauna was undertaken by Lisa Nagaoka of the University of Washington (see chapter 13). The numbering of stratigraphic units (distin- guished by roman numerals) is based primarily on sedimentological (rather than culturl) critera Each sedimentological-lithological unit that was deter- mined to have been deposited either as a single event, or as several events all representing the samne source and mode of deposition, was designated as a "layer." Subunits withn these sedimentological layers, including cultural occupadons, are designated by letters. Hence, an occupation episode that is wholly incorporated within a calcaeous sand beach ridge deposit, Layer III for example, might be designated as Layer HIB, with e lturally sterile, but lithologically identical sands above and below designated Layers hIA and IIIC. Thus, our layer designations emphasize fte geomorphological site formation processes rather than simply the pesence or absence of cultural materials. In describing the lithology of beach rdge depositional units, we paid special attention to ad- mixture of volcanic lithic grains with the dominant calcareous grains. When volcanic grains are present, these give the sediment a 'salt-and-pepper' appear- ance. The signficance of this 'salt-and-pepper' lithology is that it reflects the availabiity of a volcanic sediment source from which sand grains Excavation 45 MNa 'a Project, 199 8 9 '' Sieve& DIy cI4.et 0 &zt .L yI__ Mean StartZ= *_cm- PM. Sit As-i- S' Idd aW A"11111m 'r-41s0o5 Spit F ; .. 1 . . .. .; Df~~ l Depths: Surfae Ciiatum 0O .t . -P J I N End Ieve Mea En | m *cm|!r COMMENTS (Note sediment char~teristics, color; disturbances; samples taken; speclproblems): tA~8ftC-4 -tVt (J STl &SALT' (tu44AaM foft(6 -~~~~~ .~~ .)rL - $ OL~ jA B,e~ i.* S 7 .b^t u 468( mVAIO 6ete8) M}(~s 1' A LAAYE ht5RP) tc.'oAn * msScr' 'So6( -T F'S - - CliAkColF * , 74S,5 -SVAItrat &Ol CIWA I;"? AAz VAN,*w o ~~~~~~~~~~CJ ,r~u lm 1-5 .Ud'I??~W/ _ Vw--To Rr'. AA-TtO% Z Ftw&V #Ai OLA. CbCgAC -> psli,o# fw'A Nr e .v Tc: fS 5)g lb Re d Ceix Ct- 5LIRP1,^ PC"' P - VST65CmA- ShApSIOL. 7-*A7- P - 6C&AC-46 Recorder: C. mw,ae-ls P Jy P I ALJW 1 Figure 5.1 Example of the excavation recording form used in the 1987 and 1989 To'aga field seasons. This form is for level 5 of Unit 20 on Transect 9. 0 0 I x . 0 - -m L T 46 The To'aga Site were produced by wave action along the fonner shoreline. Today, fte sandy beaches fng To'aga are almost pwely calcares. Volcanic grains are presen only in a few spots, where volcanic head- lands p gh the coastal terrace and are exposed to wave action. The frequent presence of 'salt-and-pepper' lithologies in deeply buried deposits exposed by our excavation indicates a landform stage whm the coastal ten-ace was much narrower (prior to progradaton), and when substan- tially more volcanic material was available for incorportion into the sediment budget Thus, along most transects, there is a tran from deeper, earlier 'salt-and-pepper' sand lithology to pudely calcareous lithology, reflec coastalion and removal of most volcan:ic source material from the sediment system. In chapter 7 we use point counting of sediment sampes to furtxr document the precise qutities of volcanic lithic grains in strata within various excavaton units. 1986 TEST EXCAVATION TIe Toaga site (AS-13-1) was initially discov- ered duing te 1986 naissance survey, when examination of a deep bulldozer cutting made by the Public Woiks Department for a sanitary landfill distbed a buned cultual deposit contning Polynesian PaiOn Ware cermnics (Hunt and Kirch 1988:168). The bulldozercutting lies atthe edge of a fan of massive talus bouldes. It appears that the landfill site is at the inland-most location possible on this paricular section of the coastal flat A single test pit (1 m2) was placed dirty adjacent (3 m west) to the bulldozed area of the To'aga landfill. This locale had mained undistubed by the bull- dozer activities. Excavation of the test pit (designated Unit A) revealed the following major stata: Layer I: This dark (2.5 Y 2V0) mucky sand mixed with coliuvial clay is eniched with organic maer (contemporary A soil horizon). It contains coral pebbles, shell midden (mosdy Turbo spp.), and waterwom basalt pebbles and cobbles. Layer I reached a maximum depth of 60-70 cm below surface. The contactwith Layer 1 is diffuse over a 1-3 cm zone. Layer U: This layer had motted pale yellow to grayish brown (2.5 Y 7i2 & 512) calcareous sand containing coral and basalt pebbles and cobbles, shell midden (mostly Turbo spp.), and ceramics (four thickware sherds orly). Layer II rached a maximum depth of 135 cm below surface. The contact with Layer II is very diffuse. A single adiocarbon date (Beta-19742) on Turbo shell yielded a conventional radiocarbon age of 2350 ? 50 B.P. (cal 28 B.C.-A.D. 108 at one s ard deviation; see chapter 6). Layer m: No culturl material was pent in the white (10 YR 8/2) calcareous sand of this layer. Excavation of test Unit A reached a maximum depth of 160 cm below surface. Subsquent woik in 1987 revealed ta these major stratigraphic zones fit well within the geomor- phological sequence for the To'aga flat as a whole (see below). Artifacts collected frxm the bulldozed area of the To'aga landfill and from Unit A were described and illustrated by Hunt and Kirch (1988:169-76) and are included here in chapter 1 1. THE 1987 EXCAVATIONS The 1986 surface collections from the landfill site, combined with the limited results frm the test excavation, revealed the presence of early Polynesian occupation deposits in the To'aga area. Along with the discovery of pottery at Ta'u Village on Tau Island (Hunt and Kirch 1988), this was the first record of an early phase of occupation in the Marn'a Islands. With e concurence of the Amedcan Samoa Historic Preservation Officer, we therefore determined ta a major objective for a second season of fieldwork in Manu'a should be more intensive investigation of the To'aga site, with the aplicaton of subsurface systemadc transect tting in order to determine whether undistubd potteiy-bearing occupation deposits were pnt in the vicinity of te landfill site. Tibs second phase of work was caried out in 1987. Excavation Procedures Returning to the To'aga site in 1987, we focused on the coastal flat immediately northeast of the ladfill which had not been disurbed by bulldozing (fig. 5.3). As can be seen in the transect profle in figure 5.2, this flat is about 125 m wide from the base of the steep colluvial-alus slope to the Eavatio 47 Figure 5.2 Elevation profile along the 1987 excavato transect, showing the positions of the main excavato and Units 2 and 3, in relation to geomoaphic and pedologic feat. prese shoreline. Using the systematic tranct strtegy developed by Kirch for sub-surface sam- pling in similar coastal settings in Tikopia and Niuatoputapu islands (Kirch and Yen 1982; Kirch 1988), we laid out a transect baseline extending across the Toaga coastal fiat, at a location 60 m northe of the Ofu landfill. The first ree units excavated along the transect (at 0, 15, and 45 m from the base of the talus slope) revealed a complex and deep cultual stratigraphy in Unit 1, but only shallow cultual deposits overlying calcareous beach sands in the more seaward Units 2 and 3. A shovel test at 105 m fardher seaward along this transect, near the crst of the modem beach ridge, revealed a total absence of culural deposits, with only calcareous sand. These tests thus demonstated that the oldest cultunal deposits were to be found close to the base of the steep talus and volcanic cliff, and tha most of the coastal flat consisted of culturally sterile coral sands and reef detritus, which had been deposited during seaward progradation during the past 2-3 kyr B.P. Following these iniitial tansect tests, Unit 1 was expanded into a larger excavation in order to effec- tively sample the deep stratigraphic sequence, including the in-situ deposits of Polynesian Plain Ware ceramics. Units 4-9 were excavated, joining with Unit 1 to fonn a T-shaped trench as shown in figures 5.3 and 5.4. All units were dug trugh the Layer II calcareous sand deposit containing pottery, while Units l and 6 were caried deeper into under- lying Layers HI and IV (see Stratigraphy, below). The third stage of our 1987 excavation strategy was to detennine the lateral extent of the early pottery-bearing deposits southwest and northast of the main excavation, parallel to the base of the talus. Unit 10 was thus laid out 45 m southwest of the baseline transect, as close to the base of the talus as feasible (acmally set in among several massive rockfall boulders). This revealed a stigraphy similar to that in the main trench, including deeply buried cultural deposits with Polynesian Plain Ware and one fine, thin-ware sherd. Unit 1 1 was then laid out 45 m nonheast of Unit 1. Unit 11 revealed ceramic-bearing deposits, but these were tnncated by a large, deep pit (probably a late prehistodc ua'i masi or breadfruit-fermentation pit), and so an adjacent square, Unit 14, was opened to clarify the stratigraphy. In order to get as close as possible to the base of the talus, another test was laid out 15 m northwestof Unit 11, atthe footofthetalus and designated Unit 12. In Unit 12, a massive deposit of colluvium and large angular boulders had to be penetated to a depth of 1.8-1.9 m before we were able to reach a thin deposit of calcareous sand containing four thin, fine-tempered potsherds (one rim and four body sherds). These test units indicated that early, pottery-bearing cultual deposits at To'aga extend over a distance of at least 105 m northeast of the Ofu landfill, in a narrow zone at the base of the colluvial-talus slope. A final t, Unit 13, was opened in the centerof oD CL) OC 0 00E * . : Beach ca. 2000 B.P. 0 o 25m Beach ca. 350BP E 0 a. VERTICAL SCALE X4 48 The To'aga Site - 0 L.. z a) 0 l X (a 01 Lt E 0 0. I- Uf) G) C: CD c CM I- z 0 O- z 6:2 0 -4 z F4 0 0 U Ct) I- z 0 n C 4._ Co I) /.:N O.. : z .... 0 636 CM I- z I- z 0 d) '3> )& Vol .5 .8 Iz As o *s A g 'I Iza -S bO 1eI wI 0 ol: Ecavations 49 Figure 5.4 Excavation in progress in the 1987 main trench. Note the large talus boulder on the left. P. Kirch is recording the stratigraphy of the south face of Unit 1. a low pavement of waterwom pebbles (such pave- ments are called 'ili'ili in Samoan) apparently marking a later prehistoiic house floor. Our objec- tive in excavating Unit 13 was to obtain a sample of this later prehistoric midden to contrast with the older, ceramic-associated assemblage. Stratigraphy of the 1987 Main Excavation Although the stratigraphy of the 1987 excava- tions varied from unit to unit, the most complete depositional sequence was revealed in the main trench excavations (Units 1, 4-9), but was reflected as well in Units 10, 1 1, 12, and 14. Seaward of the main trnch, Units 2 and 3 displayed simpler stratigraphic profiles, resulting from the later progradation of calcareous beach sands. The westem profile of the 1987 main trench is fsown in figure 5.5, in which all of te principal depositional units are represented. The stradgraphic units follow: Layer IA: The upper 15-20 cm portion of the upper colluvium (10 YR 3/1.5) found here has been heavily reworked by gardening. Various planting pits or depressions are detectable in te secdon. Layer IB: This is a massive deposit of reddish brown colluvium (10 YR 312), very compact, with no internal lensing or bedding evident. No charcoal flecking was observed. The deposit incorporates numerous angular to subangular weathered volcanic lithic fragments in the gravel-to-pebble size range. A small, lens-shaped pocket of slightly darker soil containing charcoal flecking was noted within Layer IB (designated Feature 1) and may represent a garden bum feature. This would suggest that Layer IB accumulated gradually and that the land surface was internittendy gardened during its deposidon. Layer IC: The basal 15-20 cm of the upper collu- vium (10 YR 2/1) makes up this sm. It is somewhat darker than e overlying Layer IB, containing dispersed flecks and chunks (5-10 mm) of charcoal. The deposit is associated with an earth oven feature in Unit 9, and the charcoal (which is concentrated in a zone around the oven) appears to 50 The To'aga Site ~~~~~~~~~~~~~~~~~~~~~ 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~. I l0_, _ fi 0 ,| S *,b;''j]. - I- W t W (w\ a 1ll @ _/z1 l, - |; - @ i \ rib - t ? .!~~~~~~~~~~\1 lo. t ( f _ 1t_I - 5 -0\'' i CD~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~C -D W--< -C _'t'S : :'-\ I 0 t t X 1- ox : k l ? l < | j~~~~~~~~~~~~~~~~~~~~~~- LO 0 (/ 4 _z1l_ , 4 . i j . .E0 Z~~~~~~~~~~~~~~~~~~ ??1'm IIgN o o ? 0 Csh. I 0 - U; '~~~~~~~~~~~~~olmnSmpeiu I-~ ~~~~~~~~~~~~~ E- z Z ~~~~ Iii ~~~~~ ColumnSampleCut~~~~~~~~~~b ci c~~ c~ . Depth in Meters |a Exavations 51 have derived from oven rake outs. Some thick, coarse-tmpered pottery was also recovered from this deposit. It does not appear to have been a permanent occupation, however, and may represent a short-term or intermittent occupation such as a field shelter on a former land surface. Layer H: This unit with its subdivisions represents a period of active deposition of a calcareous beach ridge. Layer HA-1: A thin zone at the top of Layer HA, slightly darker and organically enriched (10 YR 512), this stratum contains anthrpophilic land snails and represents a phase of stabilizaton and vegetation of the Layer II sandy beach ridge. Some occupation in the vicinity is suggested by the presence of a few thick, coarse-tempered sherds and a thin scatter of marine shell midden. Layer HA: This deposit contains loose, calcareous beach sand (10 YR 7/2), not compacted or cemented and lacking culul materials. Although most of the sediment consists of calcareous materials, there is a subordinate quantity of basaltic lithic grains, giving the sand a distinct 'salt-and-pepper' appearanc. This indicates ta at the time of deposition there were exposed volcanic headlands in the vicinity of the beach, providing a source for the basaltic sand grains. Layer IIB: This principal pottery-bearing deposit represents a perod during which the surface of the actively accumulatng sandy beach ridge was occupied. Lithologically, Layer IIB is similar to HA and IIC, but with the addition of organic/cultural materials due to human occupation making it both daiker (7.5 YR 4/2) and more compacted. The deposit contains shell and bone midden, large quantities of small sea urchin spines and test frag- ments, ceramic sherds (prmarily of thick, coarse- tempered ware), and other artifacts. It also contains antrpophilic land snails. The deposit is non- concentrated and probably accumulated over a fairly brief span of time. Layer [IC: This layer, the basal component of the Layer II beach ridge consists, as with HA, of a 'salt- and-pepper' lithology with dominant calcareous grains and a subordinate quantity of volcanic lithic fragments (10 YR 5/2). Toward the base of this deposit are numerous large coral cobbles and some angular volcanic cobbles, along with branch coral fingers and coral rubble. This material indicates a relatively high energy deposidonal environmaet, such as storm activity along an exposed beach frnt. Thirteen sherds of thin, fine-tempered ware were present, although te deposit showed no evidence of being an in-situ occupation locale. Layer II: Massive silty-clay colluvium (7.5 YR 3- 4/2) with some incorporated subanguar lithic fragments makes up this layer. Occasional charcoal flecks are present, pariucularly near te top of the deposit. Nine thin, fine-tempered potsherds were also incorporated in the deposit. It appears to represent a single depositional event, resulting from a combination of mass-wasting ad fluvial spot of terigenous sediment from the colluvial slope above te site. A lower zone, designated Layer IIIB, incorporates some sand mixed frm Layer IV, presumably at the time of deposition. The presence of charcoal flecking indicates burning of this slope prior to fte deposition of fte sediment, pediaps due to gardening or oter human disturbance. Layer IV: Here one encounters a mixed deposit of fine-grained calcareous sand and reddish silt-clay (7.5 YR 5/4), apparendy culturally sterile. Layer V: This basal deposit of 'salt-and-pepper' sand (10 YR 8/2) includes marine shell and reef detritus; fairly compact, and showing the initial stages of cementation probably due to frequent ground-water wetting. The deposit represents an active beach ridge depositional environment. It yielded two sherds of thin, fine-tempered ware, which may have derived from a primary occupation locus behind the beach (which would now be bured under at least 10 m of colluvium). Excavation Units 2 and 3 The stratigraphies of Units 2 and 3, seaward of the main excavation, were essentially identical; the section for Unit 3 (fig. 5.6) is typical: Layer I: This organically enriched, silty-clay loam is presently under cultivation. Lithologically, the deposit is a mixture of calcareous sand (77%), silt, and clay-sized terigenous grains, the latter deriving from sheet wash erosion from the colluvial slope. Various planting depressions filled with loose, rewoiked soil are visible in the section. ITh deposit also contains fragments of marine shell midden from occupation in the vicinity. An earth oven feature was exposed in the west face of the unit 52 The To'aga Site UNIT 3: SOUTH UNIT 3: WEST Adz Section 0.0 0 0 4' 40 2 0.5 c 0. 0 1.0 Figure 5.6 Stratigraphic section of the south and west faces of Unit 3. "C" indicates coral, and "V" indicates volcanic. Layer II: Here one encounters loose, calcareous beach sand which is culturally sterile. This deposit is composed wholly of calcareous grains and lacks the 'salt-and-pepper' combination of calcareous and basaltic grains found in the Layer II deposit in the main excavation This absence indicates that by the time of deposition of Layer 11 in Unit 3, the coastline had prograded sufficiently to buq the volcanic headlands. Excavation Unit 10 As shown in figure 5.3, Unit 10 was located 45 m southwest of the 1987 baseline transect. The pit was situated in a small clear space between several very large boulders, at the base of the steep talus. Our aim here was to get as close to the talus as possible, in the hope of exposing deeply buried cultural deposits. This unit revealed a staigraphic sequence similar to that in the main excavation trench. The north and east faces of Unit 10 are shown in figure 5.7. The strata are descnrbed below: Layer IA: This upper pordon of tfie colluvium (5 YR 3/2) is presently gardened and tius loose and reworked. Its silty clay loam includes angular lithic gravel. The contact with Layer IB is diffuse over a 2-3 cm zone. Layer IB: This very compact deposit of clayey colluvium (5 YR 4/4) contains subangular lithic fragments. Occasional charcoal flecks are prsent There is no evidence that this zone had been re- worked by gardening, and it probably represents a single depositional event. The contact with Layer IC is diffuse over a 2-3 cm zone. Layer IC: The basal porton of the colluvium (5 YR 312), this deposit is similar to IA, and may represent an older land surface that was gardened. It is less compact than Layer IB and contains some anthropophilic land snails (e.g., Lamellaxis gracilis). Also, some disturbances penetrate down into Layer II. The contact with Layer nI is distinct. Layer HA-1: This thin, discontinuous deposit of calcareous sand is enriched with organic matter and stained gray (color 5 YR 4/1). Land snails of several genera are present, indicating the fonner presence of a stable land surface with vegetative cover. This deposit is a former paleosol fonned on the old beach ridge represented by Layer H. It reflects a phase of stability capping the earlier phase of beach ridge accumulation. This paleosol must have formed after pgration of the shoreline removed the sotve of cal os sediment that had resulted in he deposiion of the deeper Layer l deposits. The cntact with Layer IIA is diffuse and gradational over a 3-4 cm zone. I I I Excavations 53 UNIT 10: NORTH Stratigraphic section of the north and east faces of Unit 10. Layer IIA: A culturlly sterile deposit of 'salt-and- pepper' sand (7.5 YR 7/2), this stratum is predomi- nantly calcareous but incorporates some volcanic grains indicating the exposure of volcanic headlands near the site at the time of deposition. The deposit is fairly loose and was probably developed rapidly after the abandonment of the underlying Layer UB occupation. The contact with Layer IIB is very irregular but relatively sharp. Layer HIB: This is a ceramic-bearing occupation deposit (7.5 YR 5/2). The parent lithology is essentially identical to Layer IHA. This layer repre- sents a period of occupation on the beach ridge while te active deposition of sandy sediment continued, primarily through aeolian action. The deposit is UNIT 10: EAST stained grey with finely dispersed organic material but is very sparse in midden (not nearly so concen- trated as Layer IIB in the main excavation trench). The contact with Layer UC is diffuse. Layer UC: The mixture of calcareous sand and reddish clay (7.5 YR 5/4) found here prbably results from fluvial/hydraulic tnsprt of the clay from the talus upslope. It is culurally sterile. The contact with Layer IIn is sharp but iregular. Layer m: This compacted deposit of reddish brown clay-silt (5 YR 3/4) includes no larger lithic fragments, suggesting a fluvial rather than mass- wasdng mode of deposition. No charcoal flecks were observed. This deposit appears to represent a single depositonal event, perhaps associated with a * 0.0 0.5 0 at 0 0. 0 1.0 c. a 0 -1.5 2.0 54 The To'aga Site storm or cyclone. The contact with Layer IV is somewhat diffuse and mixed. Layer IV: The 'salt-and-pepper' beach sand (10 YR 812) of this layer incorporates larger coral cobbles and waterwom branch coral fingers. The layer represents a fairly exposed, high-energy beach nidge depositional event. Apart from shell and bone midden, Unit 10 yielded little artifacul material. A few small flakes of obsidian were wecovered from Layer IIA-1, and a single thinware body sherd (5.7 mm thick) with an orange (5 YR 5/6) slipped exterior came from Layer IIB. Excavation Units 11114 As shown in figure 5.3, Units 11/14 were located 45 m east of te baseline transect. This location is among massive talus boulders where the ground surface is relatively level. Excavation of Unit 11 revealed a large pit feature over the entire unit which extended from the lower colluvium (ca. 55 cm below surface), cutting down through calcare- ous sand layers to 2.25 m below surface. The fill of this pit contained 181 thickware sherds and 4 thinware sherds (2 with red-slipped exterors). Field observations indicated ta pottery came from the lower deposits adjacent to the pit and not frm the pit fill itself. As nonnal stratigraphy for this area could not be discemed with the intnasion of the deep pit, Unit l 1 was expanded into a rectangulartrench (1 by 2 m) with the excavation of Unit 14. Unit 14 exposed an undistrbed sraigraphic sequence, as follows: Layer IA: The upper 20-30 cm of colluvium found here was dark brown (10 YR 312) and had been reworked by gardening. This layer contains some charcoal flecking (probably modem) and no pottery. Layer IB: This massive lower colluvium is very poorly sorted and includes angular volcanic rock fragments with no evident bedding or lenses. No charcoal or other cultual materials were observed in this dark brown (10 YR 312) layer. Layer IC: This sm contained a dark grayish- brown (10 YR 3/1) mixture of clayey colluvium and calcareous sand with abundant midden, including shell, bone, pottery (309 sherds, with 304 [98%] thick-coarse, and 5 [2%] tiinware), and charcoal. This deposit was the source of the fill of the large pit described for Unit 1 1, but in Unit 14, it made up a discrete area between 54-75 cm below surface. This layer appears to represent a depositional event contemporaneous with Layer IC in the main trench (see above). Layer U: This calcareous sand deposit is enriched with organic matter and some eluviation of clay particles from the clay-rich Layer IC above. Layer II is brown (10 YR 5/3) and contains pottery (15 thickware sherds) and a relatively sparse amount of shell midden. This stratum is comparable to Layer II in the main area excavation. Layer Im: Calcareous sand (10 YR 7-812) with waterwom coral cobbles and unwom branch coral makes up this layer. It contains little cultural material and appears to represent a high energy depositional environment. Layer III compares in lithology and content to Layer IIC in the main trench. Layer IV: This brown (7.5 YR 4/2) mixture of silty-clay colluvium and calcareous sand has some organic matter. Few larger lithic cobbles or boulders are present, suggesting a sediment that probably represents sorting with surficial flow of the terrig- enous source, and wind (saltational) ransport of the calcareous (beach ridge) source. A small hearth comprised of waterwom coral boulders surrounding a shallow ash concentration was excavated at the surface of this layer in Unit 14. Five thinware sherds were ecovered from this stratigraphic boundary. Excavation was discontinued at the surface of Layer IV due to time constraints at the end of dte 1987 season. Excavation Unit 12 Unit 12 was located 16 m north and 45 m east of Unit I in the main trench (see figure 5.3). he unit is situated on the colluvial slope amidst huge talus boulders. Excavation of this unit (chosen to pen- etrate the subsurface deposits as close to the talus slope as possible) proved to be hazardous given te large boulders protruding from the unstable sidewalls of the pit, especially as the clayey sediment dried and became friable. The stratigraphy of Unit 12 was as follows: Layer IA: The upper colluvium (0-20/30 cm below surface) consists of an organically enriched, A horizon soil. The layer is dark brown (10 YR 3/1) Exavtions 55 and has been rewoiked by recent gardening. No cultural material was observed. Layer IB: The massive lower colluvium, dark brown (10 YR 3/2) in color, consists of very poorly sorted, large, angular rock fragments, with cobble- and boulder-sized clastics, in a clay-silt matrix. No bedding or lenses of finer-grained sediment were observed, nor was any cultural material noted. Layer IC: This is the base of the colluvial deposit, distinguished by an incmase in fine-grained sedi- ment The boundary with Layer IB is gradual, reflecfing a continuous icreae of clay and silt conte toward fte base of Layer I. This day-loam aains some calcamous sand (thoroughly intermixed) and angular volcanic rck frag , but these are re- stricted to cobble-sized. Charcoal flecks and waterwom coral gravel are present in small amounts. Small amounts of bone and shell were recovered from this layer, but pottery was absent. Layer I extends from the surface to 2 m below surface. Layer IC-1: This is a brown (10 YR 3/3) clay-loam with calcareous sand thoroughly intermixed. Layer IC- I can be distinguished by both its greater calcare- ous sand content and the presence of pottery. The pottery in this layer is predominantly thinware (25 sherds, with 2 thick-coarse and 23 thinware) and includes duee direct rims, one of them with red slip and an impressed lip. Shell and charcoal were recovered in small quantities. The deepest (2.6 m) portions of this layer had no cultural materials. The stratigraphy of Unit 12 was almost entirely colluvial in origin. The upper pordons, free of cultural materials, represent one or more rapid depositional events, i.e., mass wasting of material from the nearby steep volcanic cliffs. The clay-loam of the lowest portions of the sequence represents gradual accumulation of colluvial sheet wash during a perod of occupation. Calcareous sand entered the deposit through wind trasport (saltation) from the active beach ridge, which was then much more closely adjacent to this location. Excavation Unit 13 Located 45 m east and 37 m south from Unit 1 in the main trench, Unit 13 was near the center of a stone-faced, waterwom gravel-filled ('ili'ili) mound typical of Samoan house foundation construction. The gravel mound measures 15 by 16 m across its center and is nearly round in plan view. The mound is elevated approximately 25-30 cm above the surrounding organically enriched and stable sandy substrate. Excavation of Unit 13 revealed the following stratigraphy: Layer I: This deposit of waterwom coral gravel ('ili'ili) is mixed with a small prporton of waterwom volcanic gravel and with a finer, daick organically enriched, sandy loam matrix (7.5 YR 2/ 0). A bottle glass sherd was excavated from this layer, possibly indicating occupation of the house mound into the Historic period. Layer IB: This is a lighter colored (10 YR 5t2) sediment of the same texture as Layer I. The color difference is due to a lower percentage of organic matter. Ash, shell, and numerous sea urchin spines were mixed into the gravel paving stones. Excava- tion of the lower part (3048 cm below surface) of this stratum exposed a small pordon of a human burial cut from the floor surface (Layer I) of the house mound and extending into Layer II below. Only bones from the feet were recovered, the remainder of the burial lying outside of the excava- tion uit. Layer H: This stratum contains culrally sterile, white (10 YR 8/2). calcareous sand. Excavation was completed with the testing of this layer to 68 cm below surface. Radiocarbon Chronology of the 1987 Excavations Seven samples of charcoal and shell from the 1987 excavation units at To'aga provide the basis for a radiocarbon chronology of the deposits in the vicinity of the Ofu landfill. The 1987 dates were reported in full by Kirch, Hunt, and Tyler (1989) and are discussed in detail in chapter 6. The oldest dates are those for samples Beta- 25035 and -25673, both frm Layer V of the main trench, the basal calcareous beach sand which yielded two thin, fine-tempered sherds. Both samples consisted of unweathered marine shell which were probably deposited at approximately the same time as the sherds (details of these samples are provided in Kirch, Hunt, and Tyler 1989:11-12; see also Kirch, chapter 6). The samples yielded ovedap- ping ages at two standard deviations which indicates a calibrated time range of between 3700-3300 cal B.P. for the deposition of dtis beach sand which 56 The To'aga Site contained the thin, fine-tempered ware pottery sherds. While this age range is certainly early for human occupation in Westem Polynesia, it is not out-of-line with the earliest known dates for Lapita sites in the region (Kirch and Hunt 1988; Kirch 1988, table 48). Three samples are in direct association with the Polynesian Plain Ware assemblage (Beta-25034, -26464, and -25033), denrving from th Layer II occupation in the main trench and from its corelated deposit in Unit 10. All three samples agree well in age and indicate a calibraed time range of between 2500-1900 cal B.P. for Xt main Polynian Plain Ware occupation. A terminus post quem for the use of ceramics at the To'aga site is provided by sample Beta-26463, which derives from the base of an aceramic occupa- tion deposit which statigraphically postdates Layer n in the main trench. This sample yielded a cali- brated age of 1389-1287 cal B.P. The final sample (Beta-26465) was obtained from the base of the aceramnic house mound tested by Unit 13 and yielded a calibrated age of 1122-950 cal B.P. Since the platform is consructed on the present land surface, it is evident that the sequence of coastal progradation and of colluvial deposition in the vicinity of th main excavation had stabilized by the beginning of the second millenniUM A.D. The Depositional Sequence in the 1987 Main Trench The depositional sequence in the 1987 main trench excavation at To'aga can be summadzed as te following series of stages: Stage 1 (3700-3300 Cal B.P.): The calcareous sand beach represented by Layer V of the main trench was formed at a time when the island's shoreline was close to the base of the talus slope. Thus, volcanic headlands were exposed to active wave erosion, yielding mixed calcareous-basaltic lithology sands. Human occupation in the vicinity of the beach (presumably now buried under a considerable depth of colluvium and talus) is suggested by the presence of tin fine-tempered ware sherds. Stage 2 (> 2500 cal B.P.): The older beach ridge was buried at this time by terigenous silt-clay from upslope, with erosion initiated in part by human clearance of the vegetation with fire, indicated by charcoal flecking in the erosional deposit. Stage 3a (ca. 2500 cal B.P.): Accumulation of the calcareous beach ridge was renewed during this stage, while the shoreline was still close to the talus slope (revealed again by mixed calcareous- basaltic grain suites). Exposure to the active shoreline is indicated by the large coral cobbles and reef detritus, representing one or more storm events. Stage 3b (2500-1900 cal B.P.): Humans inhabited the active beach ridge surface (Layer IIB of the main trench) during this time, resulting in the deposition of ceramics and midden. Stage 4 (2500-1900 cal B.P.): In this stage, the beach ridge was abandoned as an occupation locus, and the accumulation of the beach ridge (Layer HA) continued. Stage 5 (ca. 1900 cal B.P.): The old beach ridge surface stabilized during this stage. This is indicated by the formation of an A soil horizon and the deposition of anhopophilic garden snails (Layer IIA-I of the main trench). The stabilization presum- ably resulted from seaward progradation of the shoreline, thus removing the immediate source of calcareous sediment. At roughly this time, the basal sands represented by Layer II in Units 2 and 3 were deposited. Stage 6 (< 1900 cal B.P.): During Stage 6 there was additional erosion and deposition of temgenous sediments from upslope in the area of the main trench. The presence of ceramics and an earth oven indicate a brief occupation event. Stage 7 (< 1900-1000 Cal B.P.): The progradation of the shoreline continued to its present position about 125 m seaward of the main excavation locus. Gradual deposition of fine-grained terigenous sediments over the surface of the newly formed coastal flat and the rewoiing of these deposits through gardening activities also occunred during this stage. Stage 8 (ca. 1000- 100 cal B.P.): Human habitation dispersed across the expanded coastal flat during Stage 8, represented by the 'ili'ili pebble pavement tested by Unit 13 and by other surface cultural features. in sum, the depositional sequence revealed in the 1987 excavations begins late in the second millennium B.C. with a coastal teriace only a few tens of meters wide, and with a reef platform substantially wider than at present. This geomorphic Excavations 57 situation fits well with evidence for a +1-2 m higher stand of the sea in the southwest Pacific at ca. 3 kyr B.P., as argued in chapter4. By about 1900 B.P., the shoreline had begun to rapidly prograde, associated with a drop in sea level to its present stand and with exposure of the reef crest leading to a higher rate of calcareous sediment production. The development of a coastal terrace more than 100 m wide was accompanied by the addition of fine terigenous sediments due to erosion and sheet-wash of the higher colluvial slopes, forming a highly prducfive zone for intensive cultivation and habitation. lhere- fore, the morphodynamic model of coastal terace fonnation outlined above in chapter 4 is closely supported by the stratigraphic and radiometric data from the 1987 excavations Swnmary f the 1987 Excavaions The program of excavations carried out in 1987 was successful in establishing the presence of undisturbed, well-stratified prehistonic occupation deposits in the To'aga area. The basal occupation layers contained Polynesian Plain Ware ceramics, including a fine, thinware pottery dating to the mid- first millennium B.C. The 1987 excavations also established a sequence of geomorphological change in the To'aga area, indicating substantial coastal progradation and consequently deep burial of the pottery-bearing deposits under recent calcareous dune ridge and colluvial slope-wash deposits. By the close of the 1987 season, it had become clear that the subsurface archaeological resources of the To'aga area-especially te deeply buried, pottery-bearing deposits-were far more extensive than originally thought. The 1987 excavation units had revealed the presence of culural deposits over an area of at least 4,000 i2, but without reaching the horizontal boundaries of these layers. Because of the great significance of the To'aga site for Samoan prehistory, it was clearly essential to establish the full areal extent of the subsurface archaeological features at this extensive and well-staified site. A third phase of excavations was therefore required, applying the systematic tansect strategy of subsur- face testing over the whole extent of the southeem coastal plain on Ofu Island, extending from te Ofu landfill site and adjacent 1987 excavations eastward towards Fa'ala'aga. Potentially, this entire coastal strip neady 2 km long and varying between 50 to 100 m wide, might contain subsurface archaeologi- cal deposits; this could only be determined drugh subsurface testing. Such a testing prgrmn was proposed to the Historic Preservation Office of American Sanoa in 1988 and implemented by a third field phase of the Manu'a Project in 1989. THE 1989 TRANSECT EXCAVATIONS 1989 Excavation Procedures The prmary objectives of the 1989 field season at To'aga included-mong other things-the areal definition of the fuill extent of subsurface archaeo- logical deposits witin the coastal terrace and the furter elaboration of a morphodynamic model of site fonnation processes. These goals dictated the continued use of a systemadc tract sanpling strategy, with excavation transects spaced along the full extent of fte coastal terrace (as far to the north- east as Fa'ala'aga). In order to maintain accurate spatial control on the test units to be excavated along these ransects, our first task was to establish a baseline. This was staked out at 100 m intervals, following the course of the dirt road that paallels the shoreline trughout te To'aga area. A zero or origin point was established where this baseline bisected tX 1987 trmnsect (which was designated Transect 1); from this 0-point, the baseline extended 300 m to the southwest, and 1500 m to the norheast. Each 100-m interval along the baseline was identi- fied as a potential transect-intersection point for subsurface testing, and tansects were labeled fron 1 to 19. Transects 2-3 extended southwest from the origin point (Transect 1), while tansects 4-19 extended to the norheasL Choice of specific trans for test excavation was dictated by several factors, including permission from local landowners, locations of gardens (we did not wish to disturb active Alocasia garden sites), presence of 'sacred' sites (in particular, the Tui Ofu tomb complex), as well as te usual constraints of time and budget. Over the course of the field season, we were able to excavate test units along six of the potential sampling tansects: Transects 3 (200 m W), 5 (100 m E), 7 (300 m E), 9 (500 m E), 11 (700 m E), and 17 (1300 m E). The locations of these transects, and their relationship to te baseline, am 58 The To'aga Site diagrammed in figure 5.8. (Note that in this figure the scales for the baseline and te trinsects are different.) These trants thus cover a total area extending 1.5 kn from southwest to nonheast, and in every case across te lateral extent of the coastal terrace. Prior to the commencement of excavation, each selected traect was cleared of obscuring vegeta- tion, and a transe baseline staked out at 10-m intervals. A continuous elevation profile along the Utnsect was then surveyed using a telescopic level and stadia rod. These surveyed proffles provided valuable infonnation on the geomorphic structure of the coastal terrace. In order to ascenain the relation- ships between surface and subsurface geomorphic features and modem sea level, each tansect profile was also surveyed down to the shoreline, and out onto the reef fiat. A total of 16 -iM2 units were excavated along the six tranwcs listed above. Virtually all units extended to at least 1.5 m below surface, and some units reached depths as great as 2.9 m before cultur- ally sterile sediments were exposed. Thus fte total excavated volume was approximately 32 m3. Excavation units were numbered sequentially in the order that they were iniidated, beginning with Unit 15 and ending with Unit 30. In the following pages, the 1989 excavations are described by trasect, beginning with Transect 3 at the southwestem end of the coastal terrace and progressing towards the norheast to Transect 17. Transect 3 Transect 3 is situated 200 m to the west of the 1987 excavation area (Transect 1); between Transect 3 and Transect 1 lies the bulldozed landfill site where pottery was first discovered during the 1986 reconnaissance survey. The surveyed elevation profile of Transect 3 is shown in figure 5.9, with the location of the single st, Unit 27, excavated along this transect. The ground surface along Transect 3 is composed of clayey colluvium with numerous volcanic cobbles and boulders scattered over te surface. The vegetation consists predominantly of cultivated bananas. Excavation Unit 27 This test excavation is situated 42 m inland of the road. The profile of the east face is shown in figure 5.10. The following strata were identified: Layer I: This dari brown colluvial clay contains angular and subangular basalt cobbles. Charcoal flecks were present thughout in low frequency. Layer I: An orange-brown-stained calcareous sand makes up this layer. The lower part of this stm contains many coral and basalt pebbles and cobbles. The distribution of ese larger clastics suggests a storm or other high-energy depositional event, followed by a more gradual accumulation of the finer-grained sands in the upper portion of the deposit. Layer MIIA: The brown-to-tan calcareous sand in this statum exhibits some "peppenrng" of volcanic grains. The straum also contains many angular and subangular basalt cobbles and boulders, shell, and coral rubble. Layer hIB: This layer is composed of brown sandy clay with angular to subangular basalt cobbles, rounded coral cobbles, and shell. The clay compo- nent of this sand is higher than in Layer lIA, although the boundary separating the two strata is diffuse and unclear. Layer IIIB produced the bulk of the cultural material in Unit 27, between about 175- 214 cm below surface. This cultual materal included more than 6 kg of shell midden, 120 fish bones, and some turtle (see Nagaoka, chapter 13). The lower part of Layer EB was sterile, and excava- tion was discontinued at 225 cm below surface. Unit 27 yielded a variety of prehistorc arifacts. iMese include pottery, pieces of obsidian (probably natural), basalt (from Layers H and IIA), and an adz butt fragment from Layer IRA. Layer I did not contain any artifacts. A total of 95 sherds was recovered, with the following strafigraphic distribu- tion: Layer 11, 5 thickware sherds; Layer IIA, 80 sherds (5 [23%] thinware, 57 [71%] thickware); Layer HIB, 10 sherds (3 [309] thinware, 7 [70%] thickware). No radiocarbon dates were processed from Unit 27. On the basis of the ceramics recovered, how- ever, the occupation in Layer HsB could date to avationh 59 q~~~~~~~~~~~~~~~Y LL~~~~~~~~~~~~~~~~~~~~~~~~~ 0 ~~~~~0- LL~~~~~~~~i -l) I * U, S ~~~~ gioesuleji 4 \%* %%N%~~~j<(I) 1; ISUSJ1L V~~~~~~~~~~~~~~~~~~ > LL co ) S 0) ~ ~ ~ ~ ~ ~ ~ E0 0~~ E M0 0~~~~~~i 60 The To'aga Site w 0. 4 0 0 W cnD 2co -l C CM 0) CD~~C) C C~~~~~ 0 w- 0 CO 12 -3 -4 -5 0 10 20 m TRAN -92 A 0 CD) 0 Q. x w t' LI. LI. w [SECT 3 Figure 5.9 Elevation profile along Transect 3, showing the location of Unit 27 in relation to geomorphic features. approximately 1000 B.C. Only a single occupation phase is indicated (in Layers IIIA and IIIB), fol- lowed by fte accumulation of calcareous sands, mixed with increasing quantities of termgenous, clay- silt sediment in the higher parts of the stratigraphic column. The concentration of large clastics in the lower part of Layer II is probably the result of a high-energy storm event. In the most recent deposi- tional phase, the locality has been capped by 60-70 cm of colluvium. Transect 5 Transect 5 lies 100 m east of the 1987 baseline which incorporates the main excavation trench (Units 1, 4-9) and Units 2 and 3. In 1989, Transect 5 was largely in second growth vegetation, with some Alocasia and bananas near the talus slope. The elevation profile along Traect 5 is shown in figure 5.11. Excavations along Transect 5 commenced with three units, designated Units 15, 16, and 17, spaced 10 m apart After these units had been completed and their statigraphic proffles correlated, it became apparent that the most inland test, Unit 15, had exposed a deeply buried, thin cultnal deposit containing a few sherds of thin, orange-slipped pottery. Seeking to sample more of this early deposit, another excavation (Unit 28) was opened 10 m farther inland, where the talus slope began to rise steeply. However, in this pit the massive upper colluvial deposit extended filly 2.4 m deep, making exposure of the underlying calcareous sandy strata that contained cultual materials exceedingly difficult. (Nonetheless, we did manage to carry the excavation down to 3.6 m, a risky matter in an un- reinforced lxI m pit!) We therefore retmned to the Unit 15 excavation, and expanded it with 1I-m extensions north (Unit 29) and west (Unit 30). The excavation of these extensions was complicated by the partial collapse of the expanded pit walls at one point. Notwithstanding this setback, we carried te excavation of these extensions down to 2.5 m, thereby exposing a larger sample of the early culural deposits. Our persistence in seeking an enlarged sample of this deeply buried material was justified subsequently by the msults of 14C dating of a charcoal sample from Layer nI of Unit 28. The result of 3257-2879 cal B.P. is the oldest date for unques- tionably in-situ cultural materal obtained from the To'aga site. Indeed, this sanple is penecontempora- neous with the submerged Lapita site of Mulifanua on Upolu Island. Excavation Unit 17 Unit 17 was the most seaward excavation along Trnsect S, situated 56 m inland from the baseline Excavations 61 UNIT 27: EAST FACE ' " ~~II 0 m A i - ( 9 ~~~Sherds .,,O o D '): { ok ) :? 0.0 0.5 1.0 0 0 .' * 0 C a 0. 1.5 2.0 L Figure 5.10 Stratigraphic profile of the east face of Unit 27. along the road. A detailed profile of te west face is shown in figure 5.12, and an 'exploded' profile of the entire unit in figure 5.13. This unit was of particular note because of the exposure of a sequence of gravel house floor pavements ('iliill) in Layer III, associated with several post molds and a hearth. The stratigraphy as wecorded on the west face was as follows: Layer I: This 'greasy' midden deposit of sandy clay loam contains much dispersed coral 'iliili gravel and shll middenm It has a very dark gray color (10 YR 3/1). Ihe contact with Layer f is gradational over 2-3 an. Layer fl: A compact deposit of 'ili'ili gravel occurs here in an ashy, brown (10 YR 5/3), sandy matix. This appears to be a tick, artificialy laid house floor deposit and contains well-preserved shell midden. Four post molds visible in te exploded section (fig. 5.13) are filled with Layer II material and were cut thrugh the undedying Layer III pavements. In addition, a hearth visible in the east and south profiles lies at the contact of Layers II and Ill. Layer m: This thick deposit is made up of dutee successive ili'ili pavements, laid directly over each other. The paving matenals range from small waterwom coral pebbles (ca 0.5-1 cm diameter) down to very coarse coral sand (1-2 mm diamneter). There are discontinuous bands of black ash and fine charcoal trughout the deposit as well, probably deriving frm hearth rake-out events. The lowest pavement (Layer HIC) lies dircly on Layer IVA, with a very sharp, abrupt, flat contac All of the pavements were pale brown (10 YR 7/3). The individual sub-components were: IIA, faidty clean coral gravel; HIB, coral gravel mixed with sand and ashy beds; and HIC, coarse coral sand and gravel in fine beds. These ree gravel deposits presumably represent successive repavements of a house floor. Buck (1930:68) describes the traditional Samoan practice of "levelling off the upper surface of the platform within the house and covering it with small stones ('iitiI). ... .he larger coral gravel was picked out on the beach and carried up in baskets. It often took some time to get a sufficient quantity. A Samoan woman pickdng over the coral gravel might remind one of a woman of a higher culture selecting a carpet." Layer IVA: This deposit consists of strong brown (7.5 YR 5/6) calcareous sand with a mixture of fine volcanic silt-clay. The admixture of silt-clay suggests surficial alluvial deposition of fine-grained sediment emanating from one of the colluvial fans inland, perhaps after a torrential rain. Some charcoal flecking is also present Layer IVB: The grayish brown (10 YR 5/2) calcareous sand of this deposit has small dispersed charcoal flecks. The deposit had been distubed by the deposition of Layer IVA, and is therefore discontinuous. It represents an old paleosol horizon on top of the beach ridge following stabilization. The contact with Layer V is iregular and diffuse. Layer V: This is a massive, structureless deposit of 62 The To'aga Site 6 E 4 C 0 -W 2 co 0- (0- O -4 I 2 w cn 0c -J0 (0) CO) -j Lll I co)I L( I') a _: D D LL w w -t 0 20 40m & a . . a TRANSECT 5 Figure 5.11 Elevation profile along Transect 5, showing the locations of Units 28, 15, 16, and 17 in relation to geomophic features. cD C* 0) c .) Cu (0) co 0) qm fine to medium-graired calcareous sand (white, 10 YR 812) with a minor component of volcanic lithic fragments. It is cultally sterile. The contact with Layer VI is gradational. Layer VI: This layer contains coarse-grained calcareous sand (white, 10 YR 812) with a greater frequency of volcanic lithic grins. The deposit also incorporates two distinct bands of fist-sized coral cobbles. Layer VII: Composed of fine-grained sand with a distinctly 'salt-and-pepper' mixture of calcareous and volcanic grains, this layer maiked te base of the excavation which was reached at 225 cm below surface. Unit 17 did not produce a large quantity of cultural materials. One bone of Sus scrofa from Layer I was the only identifiable vertebrate faunal specimen. Shell midden was present in Layers I-Ill, with the highest concentration in Layer III (see chapter 13 for details). Unit 17 yielded a single fishhook fragment (the point of a jabbing hook) from Layer II. No pottery was recovered from this unit, suggesting deposition of dtese strata after the abandornent of pottery production in Manu'a. Excavation Unit 16 Unit 16 was positioned 10 m farther inland along Transect 5 from Unit 17. The straigraphic profile of the west face, shown in figure 5.14, follows: Layer IA: This is the A horizon-garden soil. On the surface one finds much leaf litter with numerous anthropophilic snails in the genera Subulina, Lamellaxis, Succinea, and Pleuropoma. Layer IB: The very dark gray (10 YR 3/1) sandy loam here is a cultural midden deposit. Parent materal consists primarily of calcareous sand, with a minor component of fine clay-silt. The dar1k gray color results from the inclusion of much finely dispersed charcoal and other organic material. The layer contains much shell midden (9.4 kg) and some dispersed 'ili'ili gravel. The contact with Layer IC is gradational. Layer IC: The very daik grayish brown (10 YR 3/ 2) calcareous sand in this stratum has some fine charoal flecking. The layer appears to represent an Exavataons 63 UNIT 17: WEST FACE * :. I X . - ; ; 0~~~~~~~~~~~ f ...','0,.- WBWA . Branch Coral * VI. Coral CoWes . ' ' * *. . * VII..~~~~*.. 0.0 0.5 0 1.0 c ._ a 0 1.5 2.0 extremely non-concentrated culturl deposit. It yielded a few sherds during excavation, and one sherd was visible in situ in the west wail during profile recording (see figure 5.14). Several volcanic fire-altered oven stones were also noted in this layer. Except for its slightly darker color, this deposit is lithologically hardly distinguishable from Layer II. It probably represents the old beach slope fronting either the Layer IIB or Layer UD occupadons in Unit 15. The lower contact with Layer IV is gradational and unclear. Layer IV: The white (10 YR 8/2) fine grained calcareous sand in this deposit contains volcanic lithic grains ('salt-and-pepper' sand). Unit 16 produced considerable culual material, including 741 fish bones, as well as rat, turtle, and bird bones. Shell midden was most heavily concen- trated in Layer I (9.4 kg), with lesser quantities in the deeper layers. Full details of the faunal analysis are presented in chapter 13. Unit 16 yielded a worked Tridacna shell (possibly an adz preform) and a shell scraper from Layer IB. Artifacts from Layer III include a shell bracelet fragment, a Turbo shell fishhook fragrnent, obsidian (including a red and black banded speci- men), and basalt flakes. In all, forty-four sherds was recovered from Unit 16. These include twelve thickware sherds in Layers I and H, one of which (in Layer I) has a parallel-rib, paddle-impressed exterior surface. Layer III yielded twelve (37%) thinware sherds and twenty (63%) thickware sherds. Figure 5.12 Stratigraphic section of the west face of Unit 17. old stable surface (paleosol A horizon) of the former beach ridge. The surface is somewhat disturbed and rewoiked by the overlying Layer IB occupation. The contact with Layer II is irnegular, and root casts penetat from Layer IC into Layer II, indicative of vegetation on the former stable surface. Layer II: This white (10 YR 8/2), massive, stuc- ureless deposit of very fine-grained calcareous sand incorporates some fine volcanic lithic fragments. At the base of Layer II is a zone of fist- to head-sized, waterwom coral cobbles indicative of a high-energy stonn event. The contact with Layer Ill is very unclear and gradational. Layer m: This is a very pale brown (10 YR 7/3), Excavation Units 15129130 As indicated above, Unit 15 was excavated along with Units 16 and 17 on Trnsect 5. When a deeply buried, thin cultural deposit containing thin, red-slipped pottery was encountered, we decided to expand Unit 15 by excavating two adjacent squares, designated Units 29 and 30. The straigraphy described below, and shown in.figure 5.15, is that of the west face of Unit 15. Layer IA: The upper 10 cm of Layer I is composed of dark reddish brown (2.5 YR 3/4) colluvium. This upper zone is slightly humic, with an A soil horizon at the surface. The deposit is a poorly sorted, very rubbly mixture of volcanic sand and clay, full of angular to subangular gravel and small cobbles. This deposit is the tongue of a colluvial fan emanating 64 The To'aga Site UNIT 17 w N E * Postmolds ~ ~ ~ ~ Po tm ld VI V. Figure 5.13 Exploded stratigraphic section of all faces of Unit 17, showing postmolds and hearth associated with Layer II. from a small hanging valley above the Le'olo cliff. Layer IA grades into IB. Layer IB: Although almost identical to Layer IA, this layer lacks the organic, humic component. The lower 5-10 cm of IB is noticeably more clayey, with a lower frequency of larger clastics. Layer II: This 'greasy' midden deposit is a very dark gray (5 YR 3/1). The parent material is prima- rily calcareous sand with some silt-clay admixture. The dark color and greasy textu result from the incorporation of much finely dispersed ash, charcoal, and other organic material. The deposit contains abundant fist-sized, volcanic, fire-altered oven stones. Shell midden is also present (3 kg) but is somewhat chalky and chemically degraded. The contact with Layer IB is straight and abrupt; that with Layer lIIA-I is irregular, varying from abrupt to gradational. The high carbon content and the presence of oven stones suggest that this layer represents a cookhouse activity area. Layer IIA-1: The reddish-yellow (7.5 YR 6/6) calcareous sand of this deposit has pockets of pinkish-gray sand (7.5 YR 6/2). In addition to considerable mottling, various irregulanties ad disturbaces appantly resulted from the Layer I occupation on an old, stabilized beach ridge surface represented by this layer. Root casts extend from IIIA-l into iIIA. The contact with Layer IIIA is shaip but highly irregular. This layer is interpreted as a formerly vegetated, stabilized dune surface (paleosol). Layer IIA: This sterile deposit of white (10 YR 8/ 2) calcareous sand is well-sorted with a medium- sized (2 phi) mode and has a minor admixture of volcanic lithic grains ('salt-and-pepper'). The contact with Layer IUB is very sharp but highly irregular ('wiggly'). Layer IUB: This pale brown (10 YR 6/3), very non-concentrated occupation deposit is indicated primarily by its slightly darker color and the S 0.0 0.5 co 0 0 0 1.0 c c 1.5 2.0 I I Excavations 65 UNIT 16: WEST FACE ...S ,.IB: Voicanlci Boule IC . . .U * * II- I Root Mold Coral Cobbles (9 Vocanic Oven Stones * IV* 6 . . Iv. Water Table r 0.0 0.5 1.0 to 0 ._ a 0 a 1.5 2.0 2.5 Figure 5.14 Stratigraphic section of the west face of Unit 16. Black squares indicate sediment sampling locations. Thinware Layer II Layer IIIA Layer IIIB Layer IIID 3 5% 8 24% 7 13% 19 38% presence of some marine shell midden. The deposit is medium- to coarse-grained, and less well-sorted than Layer IIIA. The lower contact with Layer IIIC is irregular and slightly more gradational than the top contact. Lithologically, this is the same as Layer IIIA. Layer mc: The white (10 YR 8(2). calcareous sand of this stratum is essentially identical to Layer IIIA but contains a higher frequency of coarse grains (I phi). Layer IUD: In Unit 15, this deposit appeared only discontinuously, as pockets in the west face of the section. These were pale brown (10 YR 6/ 3), very similar to the Layer IIIB deposit. In the east face of Unit 15, the zone was continuous. This is the deposit that yielded thin, fine-tem- pered, red-slipped pottery, together with thickware, during excavation. Layer IV: The coarse- to very coarse-grained, white (10 YR 8(2) calcareous sand that is found here is poorly sorted, having a considerable mixture of volcanic lithic grains ('salt-and- pepper' appearance) and some small volcanic pebbles. The deposit also contains consider- able quantities of branch coral fingers (water rolled) and coral pebbles. The presence of these larger grained sediments and the lack of sorting suggest that this was a relatively high en- ergy, exposed beach depositional environ- ment. Excavation of Units 15(29/30 yielded an array of artifacts that include two shell fishhook tabs (Layers II and IIID), cut pearlshell (Layer II), a shell bracelet fragment (Layer IIIB), an unfinished fishhook (Layer IIIB), and numerous flakes of basalt and obsidian (Layers II, IIIB, and IIID). A total of 200 sherds was recovered from these units. The distribution of these sherds by layer was as follows: Thickware 58 26 48 31 95% 76% 87% 62% Total 61 34 55 50 66 The To'aga Site UNIT 15: WEST FACE .Trkdacna i L ,'Oven Stones . 0 - , U .I' A.- _L~fB> U/ *: mc. ! j .~ ~ ~ ~~ ~ ~ ~~~ .--- 0 v .IV 0 M. Figure 5.15 Stratigraphic section of the west face of Unit 15. Black squares indicate sediment sampling locations. Two thickware direct rims from Layer HIA have impressed lips, which appear to have been made with a carved paddle. Red slip is resticted to thinware (2 sherds only) from Layer II). As the frequencies above indicate, there is a generl trend toward increasing relative frequency of thickware in proporton to thinware over time in this unit. Excavation Unit 28 Unit 28 was excavated 10 m farther inland from Unit 15 in an effort to trace the deep Layer IIID occupation in Unit 15, inland under te steeply nsing colluvial fan. Clearly, the depth of colluvial overburden which would have to be removed was substantial, making this the most inland position in which excavation could be attempted witout the use of heavy machinery. The stratigraphy described below is that of the norh face. Layer I: This massive deposit of dark reddish- brown (25 YR 312) clayey colluvium is between 2.2-2.4 m thick and contains numerous, large, angular volcanic cobbles and boulders. No bedding or lenses were present, although clay content increases toward the base of the stratum. Free of cultural material, this colluvial deposit represents one or more events of mass wasting (landslides) from the adjacent volcanic cliffs. Layer HA: This discontinuous zone (pockets) of calcareous sand has been discolored (7.5 YR 6/6) by eluviation of clay from the colluvium above (Layer I). A small amount of cultural material was recov- ered, including pottery, shell and bone midden, and charcoal. Layer TIB: The white (10 YR 812) calcareous sand in this stratum has waterwom coral cobbles and a few coral boulders. This layer contains a minor admixture of volcanic sand grains ('salt-and-pepper' lithology) as well as cultural material. Layer IC: The pale brown (10 YR 6/3) calcareous sand of Layer HC is similar to IIB in origin, but is distinguished by a darker color explained by te addition of organic matter and some clay, resulting from human occupation. Its relatively abundant cultural material includes midden ad pottery (see below). Layer IIC of Unit 28 is comparable to Layer IlD of Units 15129/30. Layer mI: This layer is a white (10 YR 812). coarse- grained calcareous sand with abundant waterwom coral cobbles and fresh (unwom) brnch coral fingers. Layer III yielded less cultural material than IIC and clearly is culurally sterile at its lower depths (3.3-3.4 m below surface). Artifacts from Unit 28 came primarily from Layer IIC. These include worked shell, basalt flakes, obsidian flakes, an echinoid-spine abrader, and duee Cypraea shell dorsa found artificially nested to- gether, which may be parts of an octopus lure apparatus (see Kirch, chapter 1 1). A total of 127 sherds was recovered, with the following stratigraphic distribution: v -k . Q , Excavatio 67 Thinware Layer IIA Layer IIB Layer IIC 7 30% 51 49% Thickware The decline in thinware in proportion to thickware over time is notable, as is the overll decrease in density of ceramics. Included in the above sherd counts are two thinware rims from Layer IIB, and nine thinware and five thickware rms from UC. One of the thickware rims has an impressed lip. Eight thinware sherds, all from Layer IIC, carry a red or orange slip. Since the ceramic assemblage from Layer HB dates to 1308-930 cal B.C. (Beta-35601), Layer UC must be of equal or possibly greater antiquity. This suggests that in the Manu'a Islands, te ceramic trnsition from Early Eastem Lapita (calTying dentate-stamped decoration) to thin, plainware must have occurred relatively rapidly at the end of the second millennium B.C. Radiocarbon Dates from Transect S Three 14C age deterninations were obtained from Transect 5 samples. Ihe oldest date (Beta- 35601) comes from charcoal at the Layer IVIll interface in Unit 28, associated with the early, thin, red-slipped pottery. This sanple has an age of 3257- 2879 cal B.P. at one standard deviation. A sample of marine shell midden (Turbo setosus) from a cookhouse activity zone, Layer II in Unit 15, is associated with thick, coarse-tempered pottery. This yielded an age of 1631-1477 cal B.P. at one standard deviation (Beta-35924). The third sample (Beta- 35600) was obtained from the Layer Ill 'ili'ili pavement in Unit 17. This sample, which had no ceramics associated, yielded an age of 1256-1007 cal B.P. at one standard deviation. Further details of these radiocarbon dates and their implications are provided in chapter 6. Transect5: Sunmary of the Depositional Sequence The correlation diagram of the Transect 5 excavation units (fig. 5.16) also shows the relation- ship of these units to contemporary shoreline and sea level features. The general depositional sequence along Transect 5 can be reconstructed as follows: Stage 1 (> 3200-2800 cal B.P.): A narrow coastal bench at the base of the steep talus was fomied. The active shoreline at this time would have been in the vicinity of Units 16-17, considerably inland of the modem shoreline. The 'salt-and-pepper' lithology of the beach ridge sediments indicates exposure of volcanic headlands along the coasline, providing a source of volcanic lithic grains. In addition, the presence of larger clastics (coral cobbles, branch coral fingers) indicates a fairly high energy shore- line. Stage 2 (ca. 3200-2800 cal B.P.): Humans began to occupy the narrw bench formed during Stage 1, resulting in non-concentrated midden deposits that contain thin, fine-tempered, orange- or red-slipped pottery in Units 28 and 15129/30. The main area of occupation was probably farther inland from Unit 28, and thus is now deeply bured under talus rockfall and colluvium. The deposits exposed in Units 28 and 15/29/30 appear to represent the seaward periphery of such an occupation, down the slope of the former beach ridge toward the old shoreline. Archaeological exposure of the putative main occupation zone would require the use of heavy machinery, since as much as 5-15 m overburden of boulder talus and colluvium would probably have to be removed. Stage 3 (ca. 2800-2000 cal B.P.?): Deposition of calcareous sands onto the beach ridge continued, with significant seaward progradation of the shore- line occuning late in this phase which resulted in the deposition of the basal deposits in Units 16 and 17. During this stage, a second occupation phase resulted in the midden deposit of Layer IIIB in Units 15129/30. Stage 4 (ca. 2000-1600 cal B.P. ?): A stabilized land surface formed during these four centuries over the now wider and prograded coastal terrace, marked by the paleosol horizon (Layer IIIA-I in Units 15/29/30; Layer IC in Unit 16; Layer IVB in Unit 17). Stage SA (ca. 1600-1400 cal B.P.): The stabilized coastal terrace in the vicinity of Units 16 and 15/29/ 30 was occupied during this terminal phase of ceramic manufacture and use on Ofu Island. The 1 16 52 100% 70% 51% Total 1 23 103 68 The To'aga Site UNIT 15 UNIT 16 UNIT 17 0.0 - 0.5 1.0 1.5 e 0 0 2 2.0 *r 2.5 3.0 3.5 Figure 5.16 Stratigraphic correlations between Units 15, 16, and 17 on Transect 5. Layer II deposit in Units 15129/30 is interpreted as a cookhouse activity area, while the contemporaneous Layer lB in Unit 16 is a concentrated midden. Stage SB (ca. 1300-1000 cal B.P.): Acermic occupation on the coastal terrace in the vicinity of Unit 17 resulted in the construction of a low house mound fonned by several successive gravel ('ili'ili) pavements. Stage 6 (< 1000 cal B.P.): A tongue of clay-silt colluvium was deposited out onto the coastal ten-ace, probably due to increased up-slope forest clearance, agriculural acdvity, and subsequent erosion. At this time the coastal terrace was used for tree-cropping and shiffing cultivadon, continuing into te present era. The stratigraphic sequence along Trnsect 5 and its corelation to modem sea level, as well as the key topographic features of the contemporary coastline (reef flat, active beach ridge) also provide key evidence for testing the morphodynamic model developed in chapter 4. As indicated in figure 5.16, the oldest cultural deposit at TransectS is the Layer IIID occupation in Unit 15, yielding thin, fine- tempered, orange-slipped pottery, between 180-200 cm below surface. In Unit 28, this occupation appeared at the base of Layer II, between 290-300 cm below surface, and was 14C dated to 3257-2879 cal B.P. When corelated to modem sea level, this occupation zone lies at virtually the same elevation as mean high tide and about 1.8 m above the reef flat. These elevation relationships provide incontro- vertible evidence for tectonic subsidence, as the in situ cultural matenials in Units 15 and 28 were cleatly deposited on a narrow terrace or beach ridge that must have been at least 1 m, and more likely 2 m, above te sea level at 3 kyr B.P. Given a +1-2 m high sea level stand at 3 kyr B.P., this means ftat fte To'aga site has undergone between 2-3 m of tectonic subsidence over fte past three thousand years, as suggested by the morphodynamic model developed in chapter 4. In other words, if we retrodict the elevation of te early beach-ridge occupation at 3 kyr B.P. to +3 m (based on our hypothesized subsid- ence rate of 1 m/kyr), this would put the occupation surface between 1-2 m above the 3 kyr B.P. sea level (which itself was +1-2 m above the present level). The altemative hypothesis-that there was no tectonic subsidenc-would require the deposition of the occupation deposit under water, a physical impossibility given the sedimentological evidence. In short, the Transect 5 stratigraphic profile strongly confilrs several key elements of the morpho- dynamic model of coastal terrace fonration devel- oped in chapter 4. Excavation 69 Transect 7 Transect 7 is situated 300 m east of the 1987 excavation area. The coastal terrace here is fairly narrow, only about 65 m from the crest of the active beach ridge to the base of the steep talus slope. The elevation profile along Transect 7 is shown in figure 5.17 (a vegetation transect is given in chapter 2, fig. 2.5). The area was primarily in banana gardens under coconut and breadfruit trees in 1989. Two tests were excavated along this transect, 10 m aparL Unit 18 was located at the base of the talus slope, while Unit 19 was situated on the flat depres- sion in the center of the arboricultural zone. Unit 18 was excavated to 1.4 m below surface, at which depth the presence of large basalt boulders forced us to tenninate the excavation. Unit 19, on the other hand, was taken to a depth of 2.2 m, and a thin, non- concentrated cultural deposit was exposed near the base. Excavadon Unit 18 The following strata were discemible in Unit 18: Layer IA: This massive deposit of colluvial clay has numerous inclusions of subangular to angular gravel and pebbles. It is weak red to dusky red (2.5 YR 3-412) and very compact and sticky. Layer IB: This layer contains the same material as Layer IA, but with an admixture of snall quantities of calcareous sand grains. Some charcoal flecking (arge chunks) was observed in a 10 cm thick zone between Layers IA and lB. The boundary between IA and IB is gradational over this zone of charcoal flecking. The contact with underlying Layer II is diffuse and inregular, over a 2-3 cm zone. Layer II: Fie-grained calcareous sand with some clay admixture due to eluviation from Layer IB makes up Layer l. This pale brown (10 YR 6/3) layer incorpomtes massive volcanic cobbles and boudders which forced a temiination of the excavation. Excavation Unit 19 Unit 19 was positioned 45 m inland of the coastal road, in a relatively level area, sunrounded by a few large talus boulders. The stratigraphy of the west face is described below and is illustrated in figure 5.18: Figure 5.17 Elevation profile along Transect 7, showing the locations of Units 18 and 19 in relation to geomorphic features. 6 w 0. 4 - co CIO D E 2 < 2 O Cu 111 0 c 0 -2 -m -4 -6J c w W a Co CC m a) I-1 w w cr. TRANSECT 7 0 10 20m I co IV,, 0) 9- :L-l 4-0 c I c :3 70 The To'aga Site UNIT 19: WEST FACE . K InIY .~~IB Turtio ~ G Tutx VI Figure 5.18 Stratigraphic section of the west face of Unit 19. Black squares indicate sediment sampling locations. Layer IA: This dark reddish brown (5 YR 3/2), poorly sorted colluvial deposit has some subangular volcanic pebbles, but the general absence of gravel suggests deposidon of the sand to clay-sized par- ticles by sheet wash at the margin of a colluvial fan. Layer IA grades into IB over a 5 cm zone. Layer IB: This strum contains reddish-brown (5 YR 5/3), fine clay mixed with calcareous sand. The calcareous sand component increases with depth. The contact with Layer II is sharp and only slightly irregular. Layer U: The very pale brown (10 YR 7/3), very fine-grained calcareous sand of this layer is cultur- ally stenle. This deposit appears to represent a faidy long period of beach ridge stability, probably under vegetation. (The very pale brown color is virtually identical to that of beach ridge deposits under Pannus and Hibiscus just inland of the modem beach.) The contact with Layer III is diffuse and irregular. Layer IH: This stratulm is composed of pale brown (10 YR 6/3), medium- to fine-grained calcareous sand which is well-sorted with scattered coral and volcanic pebbles and cobbles. Layer III is barely distinguishable from Layers n and IVA. The contact with Layer IVA is very irregular. Layer IIn appears to represent another phase of beach ridge stability. Layer IVA: The white (10 YR 8/2), medium- to fine-grained calcareous sand of this layer is well- sorted, lacks larger inclusions, and is cultually sterile. Layer IVB: This layer with its very pale brown (10 YR 7/3), medium- to fine-grained calcareous sand, probably represents a period of beach ridge stability. Layer V: The color of this cultural deposit ranges from dark grayish-brown (10 YR 4/2) to grayish- brown (10 YR 512). This non-concentrated midden consists of medium- to fine-grained calcareous sand stained with charcoal and ash. There are consider- able quantities of dispersed charcoal flecks and charred wood fragments, some quite large (5-10 cm diameter). A Turbo shell fishhook was excavated from Layer V at 164 cm below surface. No pottery was present The contacts with Layer IVB and with VI are both quite irregular but fairly distict. Layer VI: This layer contains white (10 YR 8/2), coarse- to fine-grained calcareous sand and is cultally sterile. Excavation was terminated at 220 cm below surface. An auger was then used to core to a depth of 385 cm below surface, at which point the water table was encountered. No cultual materials were observed in the augured sediments. Unit 19 was notable also for the presence of a large pit, sectioned in the east and south faces of the square, as shown in figure 5.19. The pit, which was cut from Layer I down trugh Layers II and HI into Layer IV, has a mixed fill incorporating Layer IB sediment. The pit is straight-sided, with a flat base. There is no evidence of burning or use as an oven pit, nor was the feature filled with midden or trash. In shape and size, the pit is consistent with Iua'i masi, subterranea silos for the fernentation and Excavation 71 UNIT 19 E Figure 5.19 Stratigraphic section of the east and south faces of Unit 19, showing the large pit cut from Layer I into Layers II-IV. storage of breadfruit paste or masi (Buck 1930:132- 33; Kirch 1984:132-35; Cox 1980). This function is also suggested by the presence of large volcanic cobbles in the pit fill; such cobbles are used to cap lua'i masi pits after filling with breadfruit. The To'aga area is dotted with shallow circular depres- sions, ranging from 1-2 m in diameter, which our infonmants described as fonner masi pits. Unit 19 produced a modest faunal sample (see Nagaoka, chapter 13), and a Turbo shell fishhook from Layer V. No other arfifacts or pottery were present. Although charcoal and Turbo shell samples were collected from Layer V for radiocarbon dating, these have not been processed due to budgetary limitations. We are thus uncertain wheter te absence of pottery in Layer V reflects sampling enror, or whether this deposit post-dates the cessation of pottery use on Ofu (i.e., after about 1500 cal B.P.). Transect 9 Transect 9 was positioned 500 m east of the baseline origin point. Inland of the road, the coastal tenrace here was in dense banana plantations under breadfruit and coconut. No surface archaeological features were evident, other thwa a filled-in lua'i masi pit (1.5 m diameter, 30 cm deep depression) between Units 21 and 22. The elevation profile along Trnsect 9 is shown in figure 5.20. We initially laid out three units for subsurface sampling along Transect 9: Units 20, 21, and 22, situated at 10 m intervals. Unit 20 was posidoned at the base ofthe talus slope, with Units 21 and 22 on the flat between the talus and the road. Unit 20 penentated a thick series of cultual deposits between about 1-2.6 m below surface, yielding an aray of thin, fine-tempered ceramics (some with notched iims) and thickware, as well as fistooks, onaman, and odter antifacts. We therefore decided t expand this test in order to enlarge our sample of the compara- tively rare cultual materals recovered from these deposits. Unit 23 was dterefore opened adjacent to Unit 20 on the north (inland) side (fig. 5.21). Thus a total of 4 m2 was excavated along Transect 9. S 0.0 0 L. 0 0 -1.0 *E 0 -2.0 I I I 72 The To'aga Site Figure 5.20 Elevation profile along Transect 9, showing the locations of Units 20/23, 21, and 22 in relation to geomorphic features. Excavation Units 20/23 The stratigraphy of this 1x2 m excavation was recorded along the west and north faces, as depicted in figure 5.22, and described below: Layer IA: This stratum consists of dusky red (2.5 YR 312) colluvium; the upper 10 cm is humic garden soil. The very poorly sorted, compact, sticky, very plastic sand-to-clay material has abundant subangular volcanic gravel- to cobble-sized inclu- sions. The contact with Layer IB is gradational. Layer IB: This dark reddish-brown (5 YR 3/2) massive colluvial deposit is poorly sorted and has abundant larger volcanic clastics. Calcareous sand grains and occasional charcoal flecks are dispersed th}oughout. The deposit also contains a low fre- quency of dispersed, chalky shell midden (due to chemical decomposition in the humic soil matrix) and occasional waterwom MiMi'ili coral pebbles. These cultural materials suggest some low intensity use of the area during the perod of colluvial deposi- tion. This deposition was presumably gradual and incremental, and not a single event. The contact with Layer IC is gradational over a 2 cm zone. Layer IC: The dark reddish-brown (5 YR 3/3), compact, sticky silty-clay of this stratum lacks larger pebble-sized inclusions. The clay appears to repre- sent a single depositional event, presumably result- ing from sheet wash (fluvial mode of deposition) following heavy rains. The contact with Layer IIA is abrupt and slightly irregular. Layer IHA: This stratum is composed of dark reddish-brown to reddish-brown (5 YR 34/4), coarse- to fine-grained, poody sorted, calcareous sand mixed with fine clay-silt and is marked by abundant charcoal flecking. This appears to repre- sent a paleosol or fonmer stable surface of the beach ridge. The contact with Layer IIB is gradational. Layer UB: This layer contains yellowish-red (5 YR 5/6) to reddish-yellow (7.5 YR 7/6) well-sorted, medium- to coarse-grained, structureless calcareous sand mixed with reddish clay and is culturally sterile. The clay content decreases with depth, with the base of the layer consisting of nearly pure, medium- grained sand. The contact with Layer IIIA is abrupt and irregular. Layer MA: This cultural midden deposit is very daik gray to very dark grayish-brown (10 YR 3/1-2) and has a sandy loam texture. The dark coloration of the calcareous sand matrix results from a high carbon and organic content. This layer was associ- ated with a large earth oven feature, from which a 14C sample was obtained (see below). The oven and adjacent sediment contained heavy concentrations (more than 2 kg) of the spines of the large slate- pencil sea urchin (Heterocentrotus mamnillatus) and 6 0 C 1C cO - 2 2 0c C I-0 \lc - 2 -L co L 'IW -4 -6 rI,T A A F-J w w Cc u 1U 20m 11AtIobrikl :i XTOVrP a Excavations 73 Figure 5.21 Expansion of Unit 20 into Unit 23; removal of the thick colluvial deposit in progress. of fire-altered volcanic oven stones. This stratum clearly represents a cookhouse or food preparation activity area The contact with Layer IIIB is grada- tional. Layer IUB: In this layer, a dark brown to brown (10 YR 4/3), thick, cultural midden deposit of calcareous sandy loam incorporates numerous, large, subangular, volcanic cobbles. Near the top of the layer is a distinct lens of compact grayish-white ash, about 1 cm thick. This ash lens was discontinuous over the lx 2m excavation unit. The ash lens is capped by 1-2 cm of clean white beach sand, probably an artificially deposited house floor. Layer ilB contned a heavy concentration of shell and bone midden, ceramics, and other portable artifacts. The contact with Layer IIIC is gradational over 5-10 cm. Layer mC: Subangular volcanic pebbles and cobbles are contained in the light yellowish-brown (10 YR 6/4) calcareous sand midden of this stratum. The matrix consists of coarse- to medium-grained calcareous sand. This layer is essentially a lower facies of Layer IIIB, distinguished primarily by a lower concentration of midden and organic staining. The contact with Layer IV is gradational. Layer IV: The white (10 YR 8/2), coarse- to medium-grained, 'salt-and-pepper' sand (calcareous grains with an admixture of volcanic liftic grains) of Layer IV contains abundant large clastics consisting of branch coral fingers, and waterwom volcanic pebbles and gravel. It is culturally sterile. An earth oven was exposed in Layer IIIA of Unit 23, as noted above, and is illustrated in figure 74 The To'aga Site UNIT 20 w UNIT 23 w UNIT 23 N IA X I I 1 1 1 v I 1 I I I I IN I LBE K' K K'~~~~~~~Ah en v~~~~~~ *1 C3~~~~ Stratigraphic section of the west and north faces of Units 20f23. Black squares indicate sediment sampling locations. "C" indicates coral; "V" indicates volcanic. 5.23. The oven consisted of a circular pit, with a diameter in excess of 80 cm, and a depth of 38 cm from the rim to the base. The oven pit fill consisted of fire-cracked volcanic oven stones (fist-sized) with a few larger stones at the base, charcoal, and ash. Both the oven fill and the sunTounding midden contained large quantities of echinodenm spines and Thinware Layer IIB Layer IIIA Layer IIIB Layer IIIC 5 8 50 32 10% 8% 21% 28% test fragments, particularly of the slate-pencil sea urchin (Heterocentrotus mammllats), and several smaller echinoid species. Units 20/23 yielded a large sample of pottery and a number of noteworthy portable artifacts. A total of 499 sherds was recovered, with the following stratigraphic distribution: Thickware 45 87 191 81 90% 92% 79% 72% Total 50 95 241 113 0.0 0 0 0 1.0 c 2.0 .2.0 Figure 5.22 _ Excavations 75 Figure 5.23 View into Units 20/23, showing circular pit or postmold features filled with Layer IIB sedi- ment, exposed at the contact with Layer IIIC. These frequencies confinn a decrease in thinware relative to thickware also noted from other tmnsects (see above). The sherd assemblage from Units 20/23 includes one Wtckware rim from Layer IIB, six thickware rims and one thinware rim from Layer hIA, twelve thickware and six thinware rims from Layer HIB, and five thinware rims from Layer HIC. None of the thickware rims have impressed lips, and all were recovered from Layer IIIB. Of 499 sherds, 15 are red-slipped (on thinware only) and were recovered from Layers IIIA, IIIB, and IIIC. One small thinware body sherd from Layer IIIC has incised lines on the exterior, but no pattem is discemible. Several Turbo-shell fishhooks were recovered from Layer III, including one with a distinctly bent shank similar to early Eastem Polynesian fonns (see Kirch, chapter 1 1 for further discussion). The butt of a plano-convex section adz (Samoan Type V) was excavated from Layer IIIB. Layer IIIC produced an unfinished ring of Tridacna shell which evidently broke in half during the process of manufacture. At the contact of Layers IIIB and IIIC, we also recov- ered an abrading stone designed for the manufacture of small Conus-shell beads. All of these finds are further described and illustrated in chapter 11. Excavation Unit 21 Unit 21 was positioned 10 m seaward of Units 20/23, at essentially the saime elevation above sea level. The profile of the south face is illustrated in 76 The To'aga Site figure 5.24, and described below: Layer IA: This dusky red to weak red (2.5 YR 34/ 2) massive colluvial deposit is pnimanly clay with some large volcanic cobbles and talus boulders and some smaller subangular volcanic inclusions. It is very sticky and plastic. Small quantities of chalky shell midden are dispersed thrughout the deposit. The contact with Layer IB is very gradational over 10-15 cm. Layer IB: In this mixture of fine clay and calcare- ous sand, the sand content increases with depth. Some shell midden was noted near the top of the deposit, but the layer becomes culturally sterle near the bottom. There are many discretely dispersed flecks and chunks of wood charcoal, probably deposited with the clay, and suggestive of forest clearance/burning up slope. The contact with Layer IIA is quite irregular and somewhat diffuse. Layer HA: This stratum is composed of very pale brown (10 YR 7/3) calcareous sand. There is a minor lithological component of volcanic lithic grains. Layer UB: The sandy midden deposit of Layer IIB is grayish-brown to dark grayish-brown (10 YR 4-5/ 2). Contacts with both Layers IIA and IIC are irregular (wavy) but fairly distinct. This deposit contains large quantities of Heterocentrotus mamillas spines, some fire-altered volcanic stones, shell midden, bone, and ceramic sherds. This layer appears to correlate with the thick Layer III midden deposit in Units 20/23, being a seaward 'pinching out' of this occupation zone in the direc- tion of the former beach slope. Layer IIC: The white (10 YR 8/2), fine-grained, calcareous sand of this stratum has a minor compo- nent of volcanic lithic grains. The layer is culturally sterile and represents a beach slope depositional environment. The Layer IIB midden yielded a large faunal assemblage, including 13.1 kg of marine shell, 1,803 fish bones (including a large number of spines of the puffer fish Diodon hystrix), rat, bird, and turtle (see Nagaoka, chapter 13 for further details). Unit 21 yielded several artifacts of worked shell and a shell omnament (see Kirch, chapter 11), all from Layer IIB. Pottery from IIB includes three (15%) thinware and 17 (85%) thickware sherds. No rims, decorated sherds, or red-slipped sherds were present. UNIT 21: S FACE I7IhIA nic Slab j l Turb /+ I - ItE 2 i/+ IA Volcanic f~~~/, :7| ImA . . H A . .ER. Oven Stones 0 * A2 uc . Coral .~~~~ ,:~ . 0.0 -0.5 1.0 to I._ 0 0. 0 -1.5 -2.0 2.5 K- Figure 5.24 Stratigraphic section of the south face of Unit 21. Black squares indicate sediment sampling locations. Excavation Unit 22 Unit 22 was positioned 10 m further seaward from Unit 21, and at a slightly higher ground Excavation 77 elevation (on the inland slope of the present beach ridge). The noith face of Unit 22 is illustated in figure 5.25 and described below: Layer I: The very daik grayish-brown (10 YR 3/2), sandy loam of this sm has a 'greasy' texture. The deposit consists primarily of calcareous sand with a minor component of fine silt-clay. Dispersed trughout is chalky shell midden and some 'ili'ili gravel. The deposit becomes daiker toward the base, with increasing charcoal and ash content. The contact with Layer II is diffuse and irregular. Layer U: Considerable coral and basalt waterwom 'ili'ili gravel is dispersed throughout this brown (7.5 YR 5/4) sandy midden. There is some charcoal flecking, especially in the bottom of a shallow pit feature at the base of the deposit. The contact with Layer III is sharp but irregular. Layer m: The white (10 YR 8/2), medium-grained, calcareous beach sand of this stratum with its minor component of volcanic lithic grains is culturally sterile. UNIT 22: N FACE . 1..r. rt.r~F-.-- nic ni 16 V c olcani V~~~~ Tur6 0 .SCharcoal Concentration O) . . * . >.1.. . . ~ ~ . . . b Charcoal mI I 0.0 No ceramics were present in this unit. Radiocarbon Dates from Transect 9 Three 14C age detenminations were obtained from Transect 9, all on samples excavated from Unit 23. Charcoal from the large earth oven in Layer IIA (Beta-35602) yielded an age range of 2845-2612 cal B.P. at one standard deviation. This sample was in direct association with a ceramic assemblage composed of 8% thinware and 92% tickware. A second sample (Beta-35603) of dispersed charcoal flecks from d sratigraphically older Layer IIB cultual deposit yielded an age range of 2917-2382 Cal B.P. at one stadard deviation. A large single valve of Tridacna maxima shell (Beta-35604) from Layer IIlB was also dated, yielding an age range of 2444-2289 cal B.P. at one standard deviation. All dte samples from Unit 23 ovetlap at one standard deviation, suggesting that the deposition of Layers IIIB and IIIA occurred fairly rapidly, possibly over a span of only one or two hundred years. Furter details on all samples are provided in chapter 6. Swnmary of the Transect 9 Depositional Sequence Figure 5.26 is a correlation diagram of the strata exposed in the Transect 9 excavation units. This diagram also shows the elevational relationships between these strata and the modem sea level and reef-shoreline features. Evidence from the 0.5 Transect 9 excavation data suggests the following * depositional sequence: Stage 1 (> 2800 Cal B.P.): A narrow coastal terrace % or bench with a mixed calcareous-volcanic lithic .= sand lithology was fonned. At this time the active shoreline was probably in the vicinity of Units 21- 22. 1.0 Stage 2 (ca 2800-2300 cal B.P.): The occupaion of the nanow coastal terrace formed in Stage 1 resulted in the accumulation of the thick pottery-bearing Layer III strata in Units 20/23. The Layer IIB deposit in Unit 21 represents a seaward diminution of this occupation down the former beach slope. The shoreline itself was presumably no more thm 1.5 10-20 m funher seaward of Unit 21 at this stage. Stage 3 (ca 2300-1900 cal B.P. ?): The rapid progradation of the coastal terrace was initiated by rth face an increase in the biogenic/calcareous sediment Figure 5.25 Stratigraphic section of the no of Unit 22. - 78 The To'aga Site Figure 5.26 Sigraphic correlations between Units 20/23, 21, and 22 shown in relation to a schematic elevation profile of Transect 9. The horizontal dimension along this transect is schematic and arbitrary, while the vertical dimension is expressed in meters above and below mean high water. budget. Ihis resulted in the sterile calcareous Layer UB sand deposit in Units 20q23 (conrelated with Layer HA in Unit 21 and Layer HI in Unit 22), capping the earlier pottery-bearing occupation. During this stage, the shoreline prograded substan- tially, and active accumulation of calcareous sedi- ments in the vicinity of Units 20-21-22 subsequently ceased. Stage 4 (ca 1900 cal B.P. ?): Following progradation of the shoreline, the coastal terrace in the vicinity of the excavation units was stabilized, with the surface covered by vegetation. This is indicated by the stable paleosol surface represented by Layer HA in Units 20123. Stage 5 (< 1900 cal B.P.): At this time colluvial clays and larger volcanic clasfics were deposited onto the coastal tenrace. The presence of charcoal in these clays suggests forest cleane and buming on the talus slope inland of the site. These human activities thus initiated increased rates of erosion. Stage 6 (< 1900 cal B.P.): The mixed colluvial- calcareous soil of the coastal terrace was used for subsistence gardening and for dispersed habitation during Stage 6. In sum, the depositional sequence of Transect 9 closely replicates that of the 1987 excavation locality, and that of Transect 5, as described above. The elevational data from Transect 9 also fully support the morphodynamic model outlined in chapter 4, and argued above for Transect 5. As can be seen in the correlation diagram (fig. 5.26), the Stage 2 occupation is no more than I m above the modem high water mark, and some 3.5 m below the modem beach ridge. Given a + 1-2 m higher sea level stand at 2-3 kyr B.P., the Stage 2 habitations would have been awash if they were not at a higher elevation than at present. Retrodicting a subsidence rate of about 1 m/kyr puts these habitation surfaces at 1-2 m above the then sea level, which also fits the lithological-sedimentological data (see Kirch, Manning, and Tyler, chapter 7). Thus, the incorpo- ration into our morphodynamic model of a tectoni- cally induced subsidence rate for Ofu Island of ca. I m/kyr is supported by the Tansect 9 stratigraphic data. Transect 11 Transect 1 1 lies 700 m east of the baseline origin-point (the 1987 excavation locality). Here the coastal tenrace is quite broad and flat, neatly 100 m wide from te present shoreline to the base of the talus slope. The elevation profile is depicted in figure 5.27. Only a single excavation, Unit 26, was located here, some 60 m inland from the baseline. The soil along Transect I I is more sandy than at Transects 9 or 5, lacking the colluvial clay compo- nent except very close to the talus slope. The vegetation consists of tree crops (coconut and CN4 &fi CNJ ci ,Higher Berm 2 Ceramic Zone - o 2 Ceramic Zone Post- 1900 BP Progradation -1 v e t Approx. Shoreline -2 ? at 3000 BP -3 I Vegetation Line 0 v- Excavation 79 Figure 5.27 Elevation profile along Transect 11, showing the location of Unit 26 in relation to geomorphic features. breadfruit) with bananas, but aroids are absent, reflecting tfie sandy, edaphic conditions. The stratigraphy of the west face of Unit 26, shown in figure 5.28, follows: Layer IA: The dark grayish-brown (10 YR 3/2) sandy loam in this layer is culturally sterile. A thick root mass was present in the upper 5 cm. This zone contains abundant shells of several terestrial molluscs, including Subulina, Lamellaxis, and Pleuropoma. The contact with Layer IB is grada- tional over 3-4 cm. Layer IB: This very dark gray (10 YR 3/1), non- concentrated midden deposit is lithologically identical to Layer IA. There is some charcoal flecking, and the texture is slightly 'greasier' than Layer IA due to carbon and organic material. Some fist-sized, volcanic fire-altered stones were noted. The contact with Layer IHA is iregular but distinct Layer HA: This stratum contains a white (10 YR 8/ 2), massive, stnuctureless deposit of cultually sterile calcareous, medium-grained sand. There is an irregular, diffuse contact with Layer IIB. Layer JIB: The pale brown (10 YR 6/3), faintly darker calcareous sand in this zone is apparently an old, stabilized A horizon paleosol within the beach ridge depositional sequence represented by Layer II. Layer UC: The white (10 YR 8/2), coarse-grained, culturally sterile sand of this layer has a 'salt-and- pepper' lithology (mixture of calcareous and volca- nic grains). Little cultural material was recovered durng the excavation of Unit 26. Layer lB, the major source, yielded slightly more than 2 kg of shell midden. No pottery was present in any of the layers. One broken Turbo shell fishhook was recovered at the Layer IB/ IIA contact. No radiocarbon dates were obtained from this unit. The Layer IIB deposit may possibly correlate with the phase of progradation and subse- quent coastal terrace stability at ca. 1900 cal B.P. identified fiom other trnsects. This could not be verfied in the absence of radiometric dates, how- ever. If a ceramic period occupation were ever present in the vicinity of Transect 1 1, it would have had to have been inland of Unit 26, and therefore now deeply buried beneath the steep talus rockfall. Transect 17 This was the most easterly of our 1989 excava- tion transects, situated 1300 m east of the baseline origin-point. Transect 17 mns just slightly west of the low gap over the island's central ridge at Fa'ala'aga. At this easterly end of the To'aga coastal terrace, the reef flat is quite narrow; thus the coast- line is exposed to greater wave energy, particularly during storms and cyclones. The beach fronting Transect 17 consists of coarse sand and gravel, rather tha the finer-grined sands typical of locations farther to the west Also,.the modem beach sedi- ments at Fa'ala'aga contain a high percentage of w 3 -a I - 4 co 0 0 O-1- m 0 -2 I 0 CO -3 I -4 w 0 10 20m 'C ) cc m U. w TRANSECT 11 80 The To'aga Site UNIT 26: W FACE Figure 5.28 Straigraphic section of the west fact of Unit 26. Black squares indicate sediment sampling locations. volcanic clastics, derived from the exposed volcanic headland at the eastem tip of Ofu (a control sample of this modem beach sediment is described in chapter 7). These particular geomorphological features are likewise reflected in te litology of te two units, 24 and 25, excavated along Trnsect 17. The elevation profile along the trnset is shown in figure 5.29, indicating the positions of the two excavated pits just seaward of the talus slope. Excavation Unit 24 Unit 24 was situated 1 14 m inland of the baseline (along the coastal road) and about 10 m from te base of tX talus slope. The talus here consists entirely of large subangular boulders, without a colluvial clay component. The ground surface at Unit 24 was sandy loam with a cover of coconut and breadfruit. The stratigraphy of the north face of Unit 24, shown in figure 5.30, follows: Layer IA: This stratum contains dark brown (10 YR 3/3) sandy loam (A honzon) developed under Hibiscus tiliaceus, banana, breadfiuit, and other vegetation. The parent material consists wholly of coarse- to medium-grained calcareous sand. The contact with Layer IB is irregular and slightly diffuse. Layer IB: Light yellowish-brown (10 YR 6/4), coarse- to medium-grained, calcareous sand with a very minor volcanic component makes up this stratum. Also, there is slight humic staining. The contact with Layer IC is gradational. Layer IC: This white (10 YR 812). very coarse- to coare-grained sand contains lenses of waterwom coral, volcanic gravel, and waterwom branch coral fingers. The sand is dominantly calcareous but with a strong secondary mode of volcanic lithic grains. TMe contact with Layer ID is fairly sharp and regular. Layer ID: Coral, gravel, and rubble with waterwom branch coral fragments, in a matrix of coarse-grained sand make up this stratum which appears to repre- sent a single, high-energy depositional event. The contact with Layer IE is sharp and distinct. Layer IE: The coarse- to medium-grained 'salt-and- pepper' sand of this layer has scattered waterwom coral and volcanic gravel- to cobble-sized clastics. Layer IF: This stratum contains volcanic, lithic, coarse-grained sand with a slightly dominant calcareous mode. The deposit incorporates large cobbles of waterwom basalt and cobble-sized coral head shingles. It represents a very high-energy beach, with constant input of volcanic source materials to the sediment budget. The depositional sequence in Unit 24 reveals a gradual transition from a very high-energy, exposed beach directly at the base of the talus slope, then to a lower-energy beach resulting from progradation, and finally to the formation of a stable coastal tenrace ......1 . .1.. . 1 1* . ... * ~ ~ ~ . . . . . 0 _ . X e.: :: . :.~~~(9 ...: .. 110.-. Excavatio, 81 Figure 5.29 Elevation profile along Transect 17, showing the locations of Units 24 and 25 in relation to geomor- phic features. UNIT 24: W FACE *IA . Root Mold LB '. . 0 O ~~~0 ~E o . IC . 0.0 with soil development under vegetation. Three ceranic body sherds (1 thickware, 2 thinware) were recovered from Layer LB of Unit 24. One of these is particularly unusual in that it is thick, coame- tempered pottery yet with a very clear red-slipped exterior surface. While thickware is obviously contemporaneous with early thinware assemblages, 0.5 ilufing those witi red slipping, this is the only 0 thickware sherd from the To'aga site with a red slip. 0 2 a 4._ 9 Figure 5.30 St secto of dte west fae of Unit 24. Black squares indicate sediment sampling lbcatio. "4C" indicae coal; "V" i vocanic. Excavation Unit 25 Unit 25 was positioned 10 m seaward of Unit 24. The stratigraphy of the north face, shown in figure 5.31, is: Layer IA: The very dark grayish-brown (10 YR 3/ 2) sandy loam of this stratum has some subangular, fist-sized, volcanic stone inclusions (not fire altered). The contact with Layer LB is gradational over 2-4 cm. Layer IB: The black (10 YR 2/1) sandy loam of this layer is a very 'greasy' textued cultural deposit of calcareous sand with a heavy carbon and organic materal component Shell midden and fire-altered volcanic stones are present. The contact with Layer IHA is fairly disdnct and slightly inregular. Layer IC: The daik grayish-brown (10 YR 4/2) to very dark grayish-brown (10 YR 312) sandy loam of this deposit appears to be identical to Layer IA. It probably represents a fairly long period of coastal W-0 cor 0- 6 :30 e 4 6.2 W U- 0 U-. 0 -2 W o -4 I- -Jw a: 0 20 40m TRANSECT 17 82 The To'aga Site UNIT 25: N FACE 0.0 IA 4 * TB - :o.: 1.5 Base:of Excavation at2.1 ma. of U 25 Bl;/ squares Coral Graveli * and Branch Coral - * , . O * o ? / ? %'0'' ~~~~1.5 ***Base of Excavation at 2. 10 m Figure 5.31 Sltlga_ secd of th ot fac of Unit 25. Black squaresidct sediment sampling loations. terrace stability under vegetation, with gradual accumulation. The contact with Layer IIA is distinct and slightily irregular. Layer HlA: This layer contains pale brown (10 YR 6/3), coarse-graied sand with a marked 'salt-and- pepper' appearance due to a heavy mixture of volcanic grains. In the northeast comer of the unit there is a facies of sand matrix incorporafing water- rolled coral and volcanic gravelpebbles. The contact with Layer IEB is sharp but slightly irregular. Layer JIB: The white (10 YR 812), coarse-grained, 'salt-and-pepper' sand of this stratum has a lens of water-rolled coral rubble and gravel (incorporating branch coral fingers) at about 145 cm below surface. The heavy carbon content of Layer IB is sugges- tive of cookhouse activity. This interpretation is also supported by the presence of fire-altenred volcanic oven stones. No pottery was present in Unit 25. SUMMARY AND CONCLUSIONS A detailed discussion of the results of our duee seasons of excavations at the To'aga site is best defened to te concluding chapter of this volume (see Kirch and Hunt, chapter 15). Here we confine ourselves to a brief summary of several key points: 1. Areal extent qf the To'aga site. One of our pnmary objectives during the 1989 excavation season was to determine the full horizontal extent of the subsurface cultural deposits at To'aga. Through the use of systematic tnsect sampling, we have been able to ascertain that buiied cultural deposits dating as early as ca. 3000 cal B.P. extend contnu- ously from Transect 3 to Transect 9, a distance of some 700 m. Between Transects 9 and Ilthe density of cultual materials declines significantly, but some culturl materals were present in Unit 26 along Trauect 1 1. Transt 17 also yielded a very low density of culual materials. We were restricted by the local landowners from testing the 600-m gap between Transects 11 and 17 because of the pres- ence of the sacred-and rather feard-Tui Ofu tomb and other surface cultural remains in this area. Based on our transect results, we can now state ta subsurface archaeological deposits are present thrughout the entire To'aga coastal flat as far as Fa'ala'aga, but that the main pottety-bearing, concen- trated midden deposits are restricted to the area between Transects 3 and 9. The bured deposits are concentrated in a narrow band, extending no more than about 30 m seaward from the present base of the talus slope. Shovel tests in vanous localities in the present beach ridge (on both sides of the road) revealed no archaeological materials. We are also certain that cultual deposits extend inland under the steep talus for an unknown extent. The great thickness of e talus and colluvium, however, prevents testing of these deposits without the aid of heavy machinery which was not available to us. It would probably require fte removal of as much as 15 m of massive talus-olluvial overburden to expose some of these buried occupation deposits. 2. Testing o.f the morphodynnic model. The morphodynamic model of coastal terrace fonnation outlined in chapter 4 has been substantively sup- ported by our stratigraphic and radiometric data from the various tansect excavations. In particular, we have been able to confirm that Ofu Island has been Excavations 83 undergoing subsidence at a rate of 1-2 m/kyr during the late Holocene. Rapid progradation and fonna- tion of the coastal terrace occurred during the period 1900-1000 cal B.P., largely due to an increase in the marine-biogenic sediment budget. Thus, occupation deposits dating earlier than 2000 cal B.P. are con- fined to a very narrow, fonner bench or beach ridge situated at the base of the talus slope. 3. The O.fu cultural sequence. The cultural materials recovered from our varous transect excavations also provide the basis for outlining a cultural sequence for Ofu Island. This sequence commences at about 3000 cal B.P. with colonization of the island by makers of a ceramic complex including both thin and thickware vessels. While no classic dentate-stamped Lapita pottery has been recovered at To'aga yet, the early ceramic assem- blage from AS-13-1 fits well within the Lapitoid ceramic series (see Kirch 1988). Other portable artifacts, such as Type V basalt adzes and shell annbands, are also similar to Early Eastem Lapita assemblages from Tonga and Fiji. Over the period from 3000-1900 cal B.P., the To'aga site was continuously occupied, and the ceramic assemblages reveal a gradual shift in dominance from thinware to thickware. At around 1900 cal B.P., a rapid progradation of the shoreline commenced, resulting in the construction of the present coastal terrace. Also, the deposition of terrigenous sediments onto the coastal terrace increased, creating a prime zone for horticultural and arboricultural activities. During the past two millennia, the To'aga coastal terrace has been heavily utilized for such subsistence pursuits, although habitations dispersed over the terrace continued to be occupied, evidenced by such features as the 'ili'ili house mound tested by Unit 13. In chapter 15 we discuss this sequence in greater detail, and compare it with the sequence defined for Westem Samoa. REFERENCES CITED Buck, P. H. [Te Rangi Hiroa] 1930. Samoan Mate- rial Culture. Honolulu: Bemice P. Bishop Museum Bulletin 75. Clark, J. T. 1980. Historic preservation in American Samoa: Program evaluation and archaeological site inventory. Report prepared for the Amen- can Samoa Government. Anthrpology Deparment, Bernice P. Bishop Museum, Honolulu. Cox, P. A. 1980. Masi and tanu 'eli: Ancient Polynesian tehnologies for the prservation and concealment of food. Pacfic Tropical Botanical Garden Bulletin 10:81-93. Emory, K. P., and Y. H. Sinoto 1965. Preliminary report on the archaeological investigations in Polynesia. Report prepared for the National Science Foundation. Anthropology Depart- ment, Bemice P. Bishop Museum, Honolulu. Green, R. C., and J. M. Davidson, eds. 1969. Archaeology in Western Samoa, Vol. 1. Bulletin of the Auckdand Insfitute and Museum. . 1974. Archaeology in Western Samoa, Vol. 11. Bulletin of the Auckland Institute and Museum. Hunt, T. L., and P. V. Kirch 1988. An archaeological survey of the Manu'a Islands, American Samoa. Journal of the Polynesian Society 97:153-83. Kikuchi, W. K. 1963. Archaeological surface ruins in American Samoa. Unpublished M.A. thesis, University of Hawaii, Honolulu. Kirch, P. V. 1984. The Evolution of the Polynesian Chiefdoms. Cambridge: Cambridge University Press. 1988. Niuatoputapu: The Prehistory of a Polynesian Chiefdom. Thomas Burke Memo- ral Washington State Museum Monograph No. 5. Seattle. Kirch, P. V., T. L. Hunt, and J. Tyler 1989. A radiocarbon sequence from the To'aga site, Ofu Island, American Samnoa. Radiocarbon 31:7- 13. Kirch, P. V., and D. E. Yen 1982. Tikopia: The Prehistory and Ecology of a Polynesian Outlier. Honolulu.: Bemice P. Bishop Mu- seum Bulletin 238. Munsell Soil Color Charts 1988. Munsell Color, MacBeth Division of Kollmorgen Instruments Corporation. Baltimore. Redman, C. L. 1974. Archaeological Sampling Strategies. Reading, Mass.: Addison-Wesley Module in Antwpology, No. 55.