CHAPTER 7 MATERIAL CULTURE AND GEOCHEMICAL SOURCING OF BASALT ARTIFACTS AJl.I. [Velisler; E. Conte, and P. 1X' Kircb N In the course of our two field seasons, a J ) variety of portable artifacts were recov- \ - % ered from the test excavations at several sites, and a number of adzes and flaked stone debitage were surface colected. ln addition, in 2001 we studied and photo- graphed 31 stone adzes in the collection of the Gambier Commune (housed in the MIainre of Rikitea) or in other private collections. In this chapter we report on both sets of materials, along with the results of thegeochemical analysis of sev- eral adzes and flaked stone. P( )RTABLE ARTIFACTS FRONI TEST EX(CANVATITONS Our test excavations at Rikitea QTP-3), Atiaoa (1 90-06-ATA-1), Akamaru CFP-1), Kamaka (190- 04-KAlI-2),Onemea (190-12-TAR-6), and Nenega-Iti (1 90-02-AGA-3) yielded a collection of 507 portable artifacts, as enumerated in Table 7.1. The vast majority of these consist of flakes of volcanic dikestone (N = 410, 84%Vo).' The Nenega-lti rockshelter vielded the greatest quan- titv of materials related to fishhook manufac- ture and use, including hooks, worked pearlshell, and Acropora coral files. Fi.sHHO )KS Fourteen fishhooks, mostlI consisting of incomplete or fragmentary examples (including preforms), were recovered from the excavations, nine of these at Nenega-Iti rockshelter. Table 7.2 provides a list of specimens with diagnostic measurements, and several hooks are illustrated in Figure 7.1. All hooks are of pearlshell (Pinictada ma;gaitzifera) and were presumably manufactured with files of Acroporla coral branches, which were also recovered at the sites. The hooks varn con- siderablv in size, the largest and smallest mea- surable hooks having shank lengths of 37.7 and 13.5 mm, respectively. These size differences presumablv reflect differences in prey capture strategies; the large hooks may have been in- tended for benthic fishing in the deep lagoon or on the outer reef slope, whereas the small hooks may have been more effective for angling off of rocky shelves or on reef flats. Stylisticall, most of the hooks are too fragmentarv to determine a typologv (and, of course, the sample is lim- ited), but it appears that most if not all of the hooks had recurved points. The most common head type (ine-lashing device) is a knob, although one large hook has a simple notch. To the extent 129 MATERIAL CULTURE AND GEOCHEMICAL SOURCING TABLE 7.1 Portable artifacts from test excavations. Artifact Rikitea Atiaoa Akamaru Kamaka Onemea Nenega-Iti Category TP-3 (190-06- TP-1 (190-04- (1 90-1 2-TAR-6) (190-02- ATA-1) KAM-2)| TP-1 TP-2 AGA-3) Fishhooks 1 2 1 1 9 (including fragments and preforms) Worked pearlshell 5 2 47 Whole pearlshell 1 1 Pearlshell disc 1 Bone 1 1 needles/awls Acropora coral 2 2 4 1 1 1 files Pounders 1 1 Lithic NA NA 211 37 162 flakes/blades Manuports 2 Totals 1 1 0 1 4 217 40 234 NA = material not analyzed. that this small sample allows us to make com- parisons, most of the hooks from the Nenega- Iti site appear to correspond with what Weisler and Green (2001, fig. 31.3) illustrate under the rubric of "early acute recurved point tip." It will be instructive to revisit the issue of MIangarevan fishhook typology, with appropriate comparisons to other early Eastern Polynesian assemblages, once a large collection becomes available through continued excavations at Nenega-iti and other sites. l.h)RKFI) AND UWHOLE Pl,.AR1$HIJIJ The Nenega-Iti site yielded 47 pieces of xvorked pearlshell, as well as one entire valve of Piuctada ma;^ 1 . 0 * 10_ b >7: c .7.:. - f ; T X .i.5s, i ; . 0, j :::: ' '>?a . .! - - ; .. XZ ....7. _ - :: * 7 if .S : : : ,^ . ............ E . i . .r :s v * .:;.- W 0 > ^ , .. ,"rk F: _ : : ...' f .t' . .:: j. ': . . .,.,.s1 :i: :; i ,,: ;: j: :; i.,.:S,.00007. * ::' :: ::R D,:::: : ' :x E f 'S t:'f'.':07 FIGURE 7 i1i Fishhooks from the Nenega-l from TP-3 in Rikitea. See table 7.2 for de diameter (roughlv oval in section) is the top and .65 mm at the base. It wei The second specimen was found face a few meters seaward of the I rockshelter. This is of coral and lacks In its present state it has a height of The base has a roughlv oval form (75 mm) and is polished. It weig,hs 1,05 traces of abrasion are visible. Hiroa (1938a:218- d22) (escribes ethnogr-aphicallv collecte(d pound- ers friom Niangarcva, variously made of coral or : d ;;-t coat-se vesicular basalt, and notes tlat in general these "show lack of appreciation of careful shap- ing and finish." This specimen falls xvithin the range of forms illustrated by Hiroa (1938a, figs. 13-15). Both the Onemea and Nenega-Iti sites yielded substantial quantities of flaked litlhic material, primarilv though not exclusively of dikestone. At ()nemea, there is considerable dif- ference in the densitv of lithics in TP-1 versus TP-2, which suggests the likelihood of different activity areas within the dune site. The dikestone, which makes up more than 95?)) of the flakes, presumablv derives from one ~ or more of the numerous dikes which are ex- posed along the coasts of both Taravai and Akgakaultal Island. The material at ()nemea seems to come from a single source and has a dark gray- ish color with a slightlv rough texture wvhen flaked due to its microcrystalline structure; small whit- ish phenocrvsts (0.5-2 mm in size) are present. The dikestone at Nenega-Iti is similar, but the phenocry,sts are slightl1 smaller in size (<0.25 mm), and these probablIy derive from a different source. The material tends to produce flakes that are either tabular or triangular shaped in cross sec- - .. tion, and does not yield pronounced bulbs of 2 >-- W - percussion or other typical flake 'architecture' . ... . ..i ..:..-.; .-7 this is characteristic of dike rock. Figure 7.6 ti site and shows a scatterplot of length and width for a ?tails. sample of 7() flakes from Level 1 of TP-1 at the Onemea site. It can be seen that there is a strong 4() mm at tendency for flakes to be elongated (L > 2x\X), ighs 54() g. with quite a nitmber of good blades with paral- on the sur- lel sides (the line indicates the division between 'lenega-lti flakes and true blades, to the right). A fewr speci- i the head. mens show evident use-wear on one or more f 140 mm. straight-sided edges. Our impression is that these mm x 65 blade-like flakes mav have been used as expedi- 0 g. Some ent knives, scrapers, or other kinds of tools. The 132 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA a b c e _ _ _ _ _ _ _ . I_E_ FIGURE 7.2 Artifacts from the Onemea site: a, pearlshell disc; b, fishbone needle or awl; c-f, Acropora coral files. FIGURE 7.3 Worked bone object, possibly a thatching needle, from Nenega-Iti rockshelter. 133 MATERIAL CULTURE AND GEOCHEMICAL SOURCING Figure 7.4 Acropora coral files from the Nenega-Iti site. lithic assemblages at both sites do not appear to represent adze manufacture. At Nenega-Iti we recovered two waterworn, rounded basalt cobbles which are classified as "manuports" since they have a geological origin off-site. One cobble, from Layer III, is verv smooth and elongated (1 1 1 mm long, 44 mm max. w,idth) and appears to have some wvear or polish in places; it may have been used as a pol- FIGURE 7.5 f::00 0 - - Pounders from theAtiaoa(a) and Nenega-Iti (b) 103 m rockshelters. ishing, or rubbing stone. The second manuport is a small vTolcanic pebble, 4(0 mm in diameter, found near the larger cobble. MLANGAREVAN AXDS AND ADZI:S The studv of stone adzes has played a vital role in determining historical relationships be- tween Eastern Polynesian island groups (e.g., Emory 1968), and adze types also provide a measure of temporal control for relative datino- (e.g., Green and Davidson 1969:32). More re- b .... ..... T- 134 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA Dikestone Flakes, Site 190-12-6, TP-1, Level 1 60 50 40 P E F 30 20 10 0 0 10 20 30 40 50 60 70 80 90 100 lenth (mm) FIGURE 7.6 Scatterplot of length and width of flaked dikestone from TP-1 of the Onemea site (Level 1, N=70). Line shows distinction between flakes and blades. cently, the geochemistry of basalt adzes found at dated habitation sites has made it possible to assign artifacts to geologic sources and therefore reconstruct ancient patterns of interaction over time (Best et al. 1992; Weisler, ed., 1997; Weisler 1998). We collected five adzes during our survey and testing of archaeological sites in Mlangareva, and also photographed and described 31 adzes in private collections including those displayed at the Mlairie in Rikitea. SURFACE COLLECTED AXE AN) ADZES The bevel portion of what appears to be a side-hafted axe was collected from the surface of the Atiaoa rockshelter prior to excavation (Fig. 7.7). The specimen has a cutting edge width of 49.5 mm, mid-point width of 64.3 mm, and mid- point thickness of 46.1 mm. Unlike typical Polynesian adzes, the two ground bevel surfaces form a symmetrical cutting edge when viewed in cross-section (see further discussion of adzes below). This specimen has a clear hinge fracture and undoubtedl) broke during use. Although geochemicallv similar to the other local adzes, its texture in hand is more grainy and has a some- what sugary appearance. Loaned b) Tehotu Reasin of Rikitea, is a small adze of type 1 (Duff type 2C) which is rare in Eastern Polynesia, but common in West- ern Polynesia, especially, Samoa. According to Duff (1959), in Eastern Polynesia it is the most important numericall) in Mlangareva, sporadic in Pitcairn and Rapa. This specimen is just under 100 mm long, trapezoidal in section with the front narrower than the back (Fig. 7.8). It is highly polished with -75'%o polish on the front and -600/o polish on the back. The sides are almost completely ground. The smallest whole adze is a ty,pe 5A col- lected on the surface inland of the Mlairie near the recently mechanically excavated trench in a bulldozed area (Fig. 7.9). It is trapezoidal in sec- tion with the front wider than the back. The butt is slightl1 reduced, and there is an incipient tang. MATERIAL CULTURE AND GEOCHEMICAL SOURCING 0 5 cm FIGURE 7.7 Bevel portion of a side-hatted axe collected from the surface of the Atiaoa rockshelter. Aside from the butt area on the front side, the adze is well polished. This adze was sourced to the Eiao quarry in the northern Marquesas and joins a growing list of specimens that originated there (Weisler 1998). An adze fragment was found in the erosional bank of an intermittent drainage that cuts through Gatavake on the east side of Mangareva Island, designated GAT-3 (see Chapter 3), com- ing from Layer I, about 10 cm above the contact with Layer II. It is a butt and midsection frag- ment (Fig. 7.10), trapezoidal in section with the front wider than the back. It is a Hiroa (1938a) type 1 or Duff 2A. The geochemistry suggests a local source. A large axe fragment or wedge was loaned for study by Benoit Uraril who found it on the lagoon side of Rikitea village. The original speci- men broke near the midsection and shows ex- tensive reworking with thinning flakes along the butt, emanating from the poll (Fig. 7.11). The cutting edge has been reshaped, but unfinished. The surface of the Rikitea adze is black and shiny. CrS.:S'IFICATION OFA MNGAREINT AXES AND ADZES We present here the classification and metri- cal attributes of the specimens described above, along with other adzes studied in the collection of. the Rikitea Mairie, or in other private collec- tions in Mangareva. Five additional specimens studied by us are illustrated in Figures 7.12 and 7.13. The cross-sections of 25 adzes or axes in the collection of the Rikitea Mairie are shown in Figure 7.14. Hiroa (1938a:258-270) developed a classification of adzes, axes, and chisels based on the examination of 50 specimens, onl) 44 of which could be assigned to his typolog. Twenty- eight (64%o) of these were adzes, twelve axes (27'%o), and four (9'%o) could be used as either (Hiroa 1938a:261). The attributes used to define his three types of adzes are adopted below. Hiroa separated non-adze cutting tools into axes and chisels. Green (1960) developed an adze classi- 135 136 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA - - I 5cm FIGURE 7.8 Type 1 adze from Rikitea loaned for study by Tehotu Reasin. 0 5 I 1 FIGURE 7.9 Small adze of type 5A icm collected from the surface near the stratigraphic trench in Rikitea Village. This adze has been sourced to Eiao Island (Marquesas). 0 F7 - X 137 MATERIAL CULTURE AND GEOCHEMICAL SOURCING FIGURE 7.1 0 Butt and mid-section fragment of an adze from Site GAT-3 at Gatavake. FIGURE 7.1 1 Fragment of large axe or wedge loaned for study by Benoit Urarii. fication for MIangareva adzes, and we adopt that scheme here (see also Weisler and Green 2001:418-20). This scheme includes Hiroa's types 1-3 with the addition of two new types 4 and 5, described below (Weisler and Green 2001:419- 20). MIetric attributes for adzes and axes are pre- sented in Table 7.3. Type 1. This is by far the most numerous adze type in the collections studied and also among 0 10cm those reported by Weisler and Green (2001:418- 20, table 31.2, figs. 31.5-31.10). As defined by Green (1960), adzes in this type are small to large, quadrangular in section and without a tang. The base is usually slightly narrower than the face. Our specimens exhibit a greater range of cross- sectional form, possibly due to the unfinished state of some of them (see Hiroa 1938a:261-64, figs 28-32). Figure 7.14 illustrates the range of 0 5cm I m 138 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA IN A _ )'\\\ / h _~~ f . < f 'i t; ,* , 0 '\ ; / -'\ I. .\ I S A f , r J : j f ; { i . i\_ f-; X 1' ' ' -1 g _s_< _f =D , A /.... s ' '. . ? t * ' \ I ... . \ i, t : \ i '', 81 I , i -i " \ _ t. I 1 ' -SwX ^_ ,./ . ' ..-"'/- , ..y'zj, i / ./ ,/.-' / t. ,) Li corteo L]aJ........c... El corex U Polish 0 pseckin FIGURE 7.12 Adzes from Mangareva. (a) Axe from Kamaka Island, T. Reasin collection; (b) Type 1 adze from Rikitea, Mangareva, T. Reasin collection: (c) and (d) Type 1 adzes from Taravai Island, both made from dike rock, from the late J. Richeton collection. cross-sectional variability of 25 adzes that were placed in one of three groups based on similari- ties in cross-sectional shape. Type I adzes are present in each of the three groups and demon- strates that cross-sectional form is only, one of several attributes used here to group our adzes. In plan view, type I adzes expand from the poll towards the cutting edge. Poll width ranges from 16 to 50 mm (mean= 27.4 mm, n= 26) and cut- ting edge width ranges from 13 to 108 mm (mean= 51.2 mm, n= 26). Thickness at the mid- point ranges from 15 to 55 mm (mean= 26.0 mm, n= 28). Length ranges from 47 to 270 mm (mean 110 mm, n=27). Adzes of this tyNpe are most often of local origin; however, at least three specimens derived from the Eiao Island source in the MIarquesas archipelago (Weisler 1998). T)pe 2. There were no tvpe 2 adzes collected during our fieldwork. For comparative purposes we illustrate a type 2 adze (Fig. 7.1 3B) that was collected from Tenoko rnotu and reported by Weisler and Green (2001:419; see Table 1). These adzes are poorly defined at present, with one or two mesial edges occurring along the sides, while in plan the shape expands from the pointed poll towards the cutting edge (Hiroa 1 938a:264-66, figs. 33-5). Hiroa stated that "though a number of quadrangular adzes show a rounding off of the sides, especially, on the butt, they fit better with type 1..." (1938a:264). Type 3. This type was described by Hiroa from only, four adzes and consists of thick, long adzes with a more or less triangular cross section (apex towards front), roughly finished and without a grip (1938a:266, figs. 36 and 37). One of his two illustrations of this form was described as hav- ing a cutting edge with an "axlike appearance" (1938a:fig. 36 caption), yet other specimens within this type have the asymmetrical bevel typi- cal of adzes. As discussed below, we have given a specific type designation to axes and have re- served Hiroa's type 3 for true adzes. Type 4. This is a newly defined type for MIangareva first reported by Weisler and Green (2001:419). Only one adze from our recent col- lections was tentatively assigned to this type. It is fragmentary consisting of the bevel end. Type 4 adzes are familiar Eastern Polynesian types that are tang,ed, quadrangular, and without lugs. One such adze reported by Weisler and Green (2001:438, table 31.5) was surface collected from Rikitea village bv Green in 1959 and was sourced to the Society Islands based on its geochemistry. Type 5. This is the second newly defined type first reported in Weisler and Green (2001:419- 20).Type 5 adzes are triangular to subtriangular _. _...... 139 MATERIAL CULTURE AND GEOCHEMICAL SOURCING / / 0 S C 0 10cm FIGURE 7.13 (a) Type 1 adze, from Taravai Island, made from dike rock, from the late J. Richeton collection; (b) Type 2 adze, from Tenoko (specimen exhibits water rounding); (c) Type 1 adze found by T. Reasin 15 m east of site KAM-2, Kamaka Island. II / I A C (I B 140 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA FIGURE 7.14 Cross sections of 25 adzes or axes in the collection of the Rikitea Mairie. in cross section with apex towards the back. Weisler and Green (2001: 419-420) differentiated a tanged type 5A from an untanged variant, type 5B. We have one type 5 adze in our recent col- lections which was described above (Fig. 7.9). Axes. These specimens are the second most common cutting tool, with six in our collection. They are defined primarily by having two nearly equal bevels on the blade and oftentimes are heavier than adzes. One side is flat and is hafted against the handle (as in side-hafted adzes), while the other side is rounded, forming a ridge that tapers towards the poll to facilitate lashing (Weisler and Green 2001:41 8; fig. I A). These axes range in length from 64 to 114 mm (n=4); thick- ness 15 to 46 mm (n=6); poll width 18 to 74 mm (n=5); and, cutting edge width 33 to 101 mm (n=6). Figure 7.15 summarizes the frequenc) of these adze types in our collections and in those collected by Green in 1959. Some 71.6'Vo of adzes wvere type 1. Tvpe 5 and the percentage of axes both numbered 11.3%/o. However, some adzes in the type 1 group appeared to have somewhat equal bevels on the blade and may in fact be axes. It is likely, then, that axes are the second highest frequency of cutting tools. Despite the fact that onl) one of the adzes was from a dated subsurface context (site 190- 06-GAT-3), the remaining surface collected adzes still can tell us something about chronology. The assemblage consists primarv of tyNpe 1 adzes and, in Mlangareva, these generally date to the later period of prehistorv. Green and Weisler (2000: 28, 37-8) report on two such adzes: one from the late prehistoric settlement at Tokani Bay, Akamaru Island, and another from within cul- tural deposits at Aukena (Green's site GA-1, now designated 1 90-03-AUK-4) dating to before the mid-I 8th century. This small sample does sug- gest that type 1 adzes are late in MIangareva pre- historv. It would not be surprising if adzes found on the surface, as most of the adzes in the Rikitea Mlairie collection were, date from the late pre- historic period. An adze fragment was tentatively assigned to type 4 and although it could be for- eign to Mangareva, its fragmentaryT state pre- cludes any conclusive statements. Type 5A adzes have been dated in Mlangareva to the 13th to 15th century and to the early 19th century (Weisler and Green 2001 :419). In the Marquesas, this type spans the entire culture-historical se- quence. However, it does appear that the Mlarquesas was the origin of this adze type (Weisler and Green 2001: 420). Type 1 adzes are found in New Zealand, Pitcairn, Mlarquesas, Hawaii and Rapa Nui. From recent examination of adzes in the Bishop MIuseum collection, W\eisler has also identified type 1 adzes from the atolls of Takapoto and MIakatea in the Tuamotus thus expanding the known geographic range of this adze type. Our type 1 and 5A adzes, then, show links with the Pitcairn group, the Tuamotus, I I K I 11 i I I I , _; i I.......... 141 MATERIAL CULTURE AND GEOCHEMICAL SOURCING TABLE 7.3 Mangarevan adzes. Width Illustration Length Thickness Cufting Edge Midpoint Poll Type Fig. 7.14.A 100 25 41 37 22 1 B 96 21 41 35 22 1 C 270 55 105 89 38 1 D 67 17 33 30 20 1 E 70 15 37 32 24 1 F 69 17 18 26 19 1 G 83 23 32 35 22 1 H 69 18 27 26 20 1 I 80 25 31 23 1 J 47 16 13 22 1 K 76 15 35 30 16 1 L 64 15 33 32 18 axe M 92 25 47 46 31 axe N 29 44 45 4 O 110 21 50 45 23 1 P 119 30 59 56 21 1 Q 82 25 59 53 33 1 R 89 22 55 48 29 axe S 107 21 47 43 25 1 T 144 32 67 65 20 1 U 165 44 83 71 41 1 V 78 21 43 41 29 1 W 75 20 41 37 27 1 X 87 21 40 37 1 Y 117 21 47 42 30 1 Fig. 7.12.A 114 31 57 51 33 axe B 127 30 68 63 39 1 C 128 34 66 58 24 1 D 126 30 57 56 28 1 Fig. 7.13 A 155 40 72 68 50 1 Fig. 7.13B 192 40 98 89 53 2 Fig. 7.13C 222 52 108 96 46 1 Fig. 7.8 100 22 41 39 24 1 Fig. 7.9 57 24 34 34 27 5A Fig. 7.10 18 38 27 1 Fig. 7.11 44 101 106 74 axe Fig. 7.7 46 50 64 axe All measurements to the nearest mm. Types after Weisler and Green (2001:418-420). 142 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA frequency (n=53) Mangarevan Adze Types 40 35 30 25 20 15 10 5 0 adze/axe type FIGURE 7.15 Frequency of adze and axe types in Mangareva. and the Mlarquesas, the closest neighbors to Mangareva where on the basis of geochemi- cal sourcing MIangyareva was the nexus of a long-distance interaction sphere (Weisler 2002). This demonstrates the complimentarv nature of tvpological and sourcing studies. Mlangarevan adze assemblages are dominated by type I (Fig. 7.15), vet it is of particular note that axes, characterized bIy their roughly ground and unfinished forms, are the second most com- mon type. In pondering the high frequenc! of axes, Hiroa (1938a:277) suggested that this heavier wood-working tool in Mangareva might be correlated to the use of rafts. Beechev xvas greeted by Mangarevans on rafts when he sailed into the lagoon in 1824, and he saw no canoes. Buck suggests that axes performed better for cutting down trees for rafts, while adzes are bet- ter suited for hollowing out logs for canoes. Based on current archaeological data, howveer, adzes co-occur with axes duringr late prehistorn, and it is unfortunate that no axes have been coHlected from dated contexts. Consequentlv, we do not know when axes first appear in Mlangare,va. Based on sourcing studies of Eastern Polvnesian adze material, lang-areva was the center of a long- distance interaction network that linked the Pitcairn group, the Mfarquesas, the Society Is- lands, and undoubtedlv the Tuamotus until some- time in the 15th century when the collapse of long-distance voyagLring may have been triggered regionally bv late prehistoric social unrest on sev- eral island groups (Weisler 2002:267-68). For Mancrareva Hiroa describes inter-tribal warfare and cannibalism as characterizing late prehistoric society. If Hiroa (1 938a:277) was right that axes are correlated to raft construction and are there- fore more numerous in late prehistorv, then we may expect that axes vould be more common after the collapse of long-distance vovaging, af- ter the 15th centurv. With the decline or end of long-distance voyaging, large ocean-gYoing canoes were no longer needed to support the chiefly prerogatives of obtaining (and then regulating at home) the distribution of exotic goods. Per- haps additional evidence for a deteriorating Mangarevan society can also be inferred from the tvpical unfinished state of axes. Unlike adzes, axes are typically made from coarse-grained rock (probably all local) and are ground mostly at the bevel leaving the rest of the tool unfinished, thus exhibiting the original xveathered rc)ck surface. Perhaps axes were considered an expedient tool only fashioned to produce inferior watercraft. GE )OCHEMICAAl ANALYSIS OF BASAL-T ARTI FACTS All five adzes collected in the field in 2001 (Fig. 7.7 to 7.11) along with 18 pieces of basalt debitage were selected for geochemical analysis. ! i I I ! -.- - I - 2 3 5 axe 1 143 MATERIAL CULTURE AND GEOCHEMICAL SOURCING The adzes were cored with a diamond-tipped drill through the poll or, in the case of broken adzes, through the exposed surface. The holes were refilled with modeling clav and the filled hole was painted wvith xvater-based paint to match the surrounding adze surface color. The samples were processed for wvavelength dispersive X-ray, fluorescence (\XTDXRF) analy- sis following procedures outlined by Johnson et al. (1999) and summarized here. The cored samples and debitage were each reduced in a hardened steel jaw crusher to small chips that were hand picked and placed into a Tema swingmill with tungsten carbide surfaces and milled for two minutes. Up to 3.5 grams of this rock powder were then xveighed into a plastic mixing jar with pure lithium tetraborate (Li,B 4 ) and, with an enclosed plastic ball, mixed for 10 minutes. The mixed powders were emptied into graphic crucibles and placed on a ceramic trav and loaded into a muffle furnace. Fusion took five minutes in a preheated furnace at 1 000?C. The crucibles were removed and cooled for 10 minutes, then loaded into a Tema swinngmill and ground for 35 seconds. The resultingglass pow- der was replaced in the graphic crucibles and refused for five minutes. Following the second fusion the cooled beads were labeled, the lower flat surface ground on 600 grit, finished brieflyi on a glass plate to removTe any metal from the grinding wheel, washed in an ultrasonic cleaner, rinsed in alcohol, and dried. The beads were then loaded in to the XRF spectrometer. The results of the \W'DXRF analyses of the fiv!e adzes and one flake are presented in Table 7.4, and a bivariate plot of the key geochiemical si(tmnatures of these artifacts is provided in Fig- ure 7.16. The flake is from site GAT-3 where a habitation pav,ement (paepae) xvas exposed in a stream cut (see Chapter 3); it is uncertain whether the flake collected from this stratigraphic con- text derived from a nearbv source, possibly from the adjacent drainage. However, this flake, the adze frag-ment also collected from the exposed section at GAT-3, and an adze from Rikitea col- lected by Benoit Urarii all share similar g,eochemi- cal properties, suggesting a common source. Two other adzes, one collected from the Rikitea shore- line by Tehotu Reasin and another from Atiaoa, are similar in major elements and most trace ele- ments to the Gatavake flakle suggesting a geolog- callv related source. That source maz be in Gatavake Vallev; but samples should be obtained from this drainage to determine the range of geochemical variability of the water-rounded rock found there. The geochemistry of secondary deposits such as this is often difficult to pin down since source rock may come from the entire length of the drainage which taps into multiple geologic formations up- stream, each with unilque geochemistr. Another adze collected from Rikitea is clearlyT from the Eiao basalt source in the Mlarquesas Islands. The Eiao geochemistry is relatively homogeneous, and ar- tifacts assigned to this source characteristically match closely. Table 7.5 documents the geochemical com- position of 18 basalt flakes and their source as- signment, where possible. Mlost of the analyzed flakes were from the site ATA-4 flake scatter Location A (see Chapter 3), while a few were from Location B, an accumulation of ash and fire-altered rock with some debitage-probably a refuse dump. The flakes from Location A are characterized by one dominant geochemical sig- nature. There are two flakes, however whose ori- gin was clearlv Tautama, Pitcairn Island, the larg- est basalt source in southeast Polvnesia (Weisler 1997:156). Two other flakes from Location 4 had a unique geochemistry (ATA-4B-2 and ATA4- SA2), and may represent material collected in- land on the taluvial slopes at Atiaoa. One of these flakes matches closely to a flake collected near the exposed trench in Rikitea, just inland from the Nlairne (Table 7.4). In his ethnography of Mangareva, Hiroa did not obtain any information about adze quarries (1938a:275) and, from what is knoxvn of the is- lands' geological formations, it is likely that no large fine-grained basalt sources exist. However, future surveys in Mlangareva should target areas 144 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA TABLE 7.4 WDXRF analytical results for Mangarevan adzes. Figure 7.9 ---- 7.10 7.11 7.8 7.7 No.: Locus Rikitea Eiao, GAT-3 GAT-3 Rikitea Rikitea Atiaoa Trench Marquesas shoreline Artifact adze flake adze axe adze axe type:.________________ Source: Eiao, Source-a Local Local Local Local Local Marquesas Mangareva Mangareva Mangareva Mangareva Mangareva Un-normalized Major Elements (Weight %): SiO2 47.50 46.95 48.51 48.11 48.42 49.34 49.27 Al203 14.89 15.23 15.440 15.28 15.25 15.05 14.83 TiO2 3.861 3.900 2.51 2.490 2.484 2.458 2.369 FeO 12.53 12.18 10.563 10.67 10.84 10.12 10.00 MnO 0.165 0.160 0.17 0.166 0.166 0.151 0.146 CaO 9.37 9.32 11.21 11.18 11.07 11.82 11.79 MgO 6.43 6.47 7.47 7.57 7.69 6.86 7.22 K20 1.00 1.00 0.69 0.65 0.67 0.65 0.65 Na2O 3.08 3.18 2.290 2.19 2.22 2.31 2.20 P205 0.511 0.540 0.33 0.308 0.309 0.311 0.302 Total 99.33 98.93 99.175 98.61 99.12 99.07 98.78 Un-normalized Trace Elements (ppm): Ni 95 100 107 108 104 103 119 cr 93 87 178 180 179 258 302 Sc 17 24 23 28 30 28 29 V 294 297 267 275 252 249 237 Ba 159 187 148 156 142 148 142 Rb 20 18 13 13 11 13 19 Sr 575 591 394 393 386 393 391 Zr 280 306 154 154 152 154 148 Y 36 37 67.0 21 21 37 73 Nb 28.0 28.0 30 28.1 29.0 27.1 26.8 Ga 24 nd 20 17 20 21 19 Cu 38 47 63 69 67 73 36 Zn 125 130 94 87 88 86 97 Pb 3 nd 3 4 5 2 1 La 31 nd 40 25 16 38 43 Ce 71 nd 75 36 61 51 40 Th 5 nd 3 4 3 1 Major elements are normalized on a volatile-free basis, with total Fe expressed as FeO. "R' denotes a duplicate bead made from the same rock powder. "t" denotes values >1 20%o of our highest standard. 145 MATERIAL CULTURE AND GEOCHEMICAL SOURCING FIGURE 7.16 Bivariate plot of SiO2 versus K20 + Na20 for adzes and flakes analyzed from Mangareva, along with Eiao (Marquesas) and Tautama (Pitcairn) sources. likel to have localized fine-grained basalt, such as the alkalic exposures identified by Brousse and Guille (1974:161, 218) on MIangareva, Akamaru and Mlakapu islands. If new in situ geologic sources are identified, it will be necessary to: (1) identify the geologic event that produced the fine-grained basalt (e.g., dike or flow); (2) collect artifacts (debitage, adze performs) and geologi- cal material to unequivocally link the artifact geochemistr) to the geological source; and (3) demonstrate the geochemical variability of the source by multiple analyses (Weisler and Sinton 1997:187-88). This should be a top priority of future research. The geochemical analsis of debitage and adzes collected during our fieldwork was aimed at our objective of better understanding social relationships xvithin the archipelago and to add to our knowledge of the Eastern Polvnesian in- teraction sphere (Weisler 1998: fig. 1; 20()(2: fig. 13.1). Because no in situ geologic sources of fine- grained basalt were located during our surveys, or during those of Weisler (1996), we decided to analyze debitage to determine the variabilit of stone used for adze production within Mangareva. Although we do not know the precisegeo/qgicsources of this adze material, it is possible to infer some form of interaction if two or more sites contain artifacts of the same geochemistry. Thus our data suggest that the same source that dominates the debitage at Atiaoa site ATA-4 was also trans- ferred to the site ATU-1A paepae at Atituiti, and from Atiaoa to Rikitea. Correlations also exist with artifacts from Gatavake and Rikitea. Rikitea, on the eastern side of Mangareva Island, is divided from Taku on the west and oceanside by a mountainous spine that follows the long axis of the island. According to Hiroa, these two villages grew into centers of influence which became the principal and competing poli- ties into which the island was divided (1938a:5- 6). Although our arclhaeological studies were conducted exclusivelyr within the Rikitea district, our geochemical analysis suggests that the vil- lages of Atiaoa, Gatavake, Atituiti, and Rikitea were linked in some manner of social interac- tion. Unfortunateb;, none of the geochemically analyzed artifacts are from dated contexts, so it 7- X ATA flake C Gatavake flake + AKU flake paepoe flake (-x X O Atiaoa source A Rikitea adze TautamaJ- 6 - UM Eiao source * Tautama source * Gatavake adze V Weisler geologic sample 9 5 - A0 -, , \Eiao + o 4 - Ai Alkalic V Tholeiitic 3 - 2 -V 46 47 48 49 50 SiO2 146 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA TABLE 7.5 Results of XRF analysis of basalt debitage from Mangareva, and of source material from Eiao (Marquesas) and Tautama (Pitcairn). ATA- ATA- ATA- ATA- ATA- ATA- ATA- ATA- Paepae ATA- RIK-2 4A-1 4A-7 4A-9 4-SA3 4-SAI 4B-1 4B-6 4A-3 Atituiti 4-SA2 flake I I I I I - -- . ~~~~~~~~~~~~~flakeI Probable Atiaoa Source Atiaoa Subsource Unnormalized Major Elements (Weight %) SiO2 49.79 49.28 48.98 49.41 49.21 48.86 49.61 49.12 49.89 49.44 49.17 AJ 203 14.77 14.60 14.54 14.66 14.58 14.41 14.56 14.72 14.64 14.65 14.98 TiO 3.060 3.058 3.063 3.075 3.005 3.020 3.043 3.041 3.059 2.886 2.988 FeO 10.36 10.66 10.67 10.75 10.94 11.36 11.10 10.48 10.83 11.14 11.49 MnO 0.167 0.160 0.167 0.169 0.165 0.173 0.165 0.166 0.155 0.180 0.169 CaO 11.34 11.22 11.37 11.30 11.39 11.23 11.35 11.56 11.29 11.36 11.07 MgO 5.99 6.13 6.10 6.04 6.17 6.31 6.11 6.03 5.95 6.39 6.35 K20 0.89 0.85 0.80 0.86 0.79 0.82 0.80 0.76 0.89 0.52 0.57 Na2O 2.71 2.65 2.66 2.73 2.66 2.60 2.59 2.63 2.64 2.30 2.44 P205 0.412 0.411 0.419 0.414 0.409 0.408 0.412 0.408 0.414 0.351 0.390 Total 99.49 99.02 98.77 99.40 99.32 99.19 99.74 98.91 99.76 99.21 99.62 Unnormalized Trace Elements (ppm): Ni 79 72 79 79 76 72 74 76 75 60 68 Cr 64 65 65 70 68 61 68 69 67 98 90 Sc 30 26 28 29 30 27 31 24 24 32 31 V 291 289 293 294 289 282 295 293 298 290 299 Ba 198 201 201 181 165 188 188 204 184 145 158 Rb 19 16 13 17 14 10 15 9 16 10 8 Sr 424 420 430 416 421 423 425 444 412 379 393 Zr 193 192 194 195 190 195 192 190 195 171 181 Y 25 24 25 25 24 26 24 25 25 25 27 Nb 37.4 37.1 37.6 37.1 36.9 38.1 37.7 37.4 38.2 32.9 339 Ga 20 20 22 22 22 20 17 20 21 21 19 Cu 92 93 87 98 96 94 87 92 92 79 96 Zn 99 100 102 100 100 97 99 100 98 104 105 Pb 2 4 0 2 1 1 0 0 1 0 2 La 29 29 29 8 13 17 21 20 10 37 12 Ce 78 58 62 61 58 54 47 58 76 52 36 Th 4 6 3 4 4 | 2 | 4 | 5 | 4 | 2 | 1 Major elements are normalized on a volatile-free basis, with total Fe expressed as FeO. a= Data from Sinton and Sinoto (1997:table 11.7). Fe203 converted here to total Fe by dividing b= US is the abbreviation for unknown source. c= Precision determined by analyzing a duplicate bead made from the same rock powder. R=repeat. 1.1113 into 13.53. is not possible to place these interactions in a face contexts, it is probable that the artifacts were temporal framework. However, because much deposited during later prehistory. If so, the spa- of our analyzed adze material derives from sur- tial distribution of these artifacts relates to so- 147 MATERIAL CULTURE AND GEOCHEMICAL SOURCING ATA- ATA- ATA-4A AKA-1 1 MAN- IGERU- IGERU- AKU AKU ATA- ATA- 4B-2 4-BH I flake flake 5 US-b 1 flake 2 flake -32 -35 4A-3 4A-3R flake flake Unknown Tautama Unknown Source Akamaru Precision- Source Source Source c in=21 -a Unnormalized Major Elements (Weight %): SiO2 49.41 49.93 50.03 50.36 50.68 48.51 49.74 49.60 50.75 51.22 49.12 49.04 AJ20, 15.14 15.57 15.30 15.32 15.45 15.44 15.02 15.15 15.27 15.65 14.72 14.62 TiO' 2.398 2.680 2.699 2.675 2.770 2.510 3.071 3.041 1 696 1.736 3.041 3.018 FeO 9.87 12.10 11.95 12.25 11.84 10.56 11.45 11.26 10.27 9.53 10.48 11.13 MnO 0.149 0.220 0.218 0.214 0.202 0.166 0.171 0.168 0.138 0 135 0.166 0.165 CoO 12.07 7.09 7.13 7.03 7.26 11.21 10.29 10.43 11.42 11.70 11.56 11.47 MgO 7.21 3.49 3.54 3.46 3.36 7.47 5.71 5.71 6.42 6.56 6.03 5.98 K20 0.65 1.99 1.99 2.04 1.96 0.69 0.93 0.88 0.53 0.45 0.76 0.76 Na2O 2.30 4.55 4.48 4.45 4.42 2.29 2.73 2.69 2.11 2.14 2.63 2.64 P205 0.315 1.250 1.251 1.237 1.282 0.326 0.451 0.424 0.188 0 204 0.408 0.404 Total 99.52 98.87 98.59 99.04 99.22 99.18 99.56 99.35 98.79 99.33 98.91 99.23 Unnormalized Trace Elements (ppm): Ni 113 1 1 3 1 107 60 55 84 92 76 74 Cr 310 8 1 0 3 178 64 64 286 293 69 67 Sc 30 14 16 15 14 23 30 33 24 23 24 29 V 245 109 131 132 121 267 287 297 208 206 293 291 Ba 141 458 453 449 435 148 206 199 87 77 204 193 Rb 16 39 35 39 38 13 16 17 20 14 9 10 Sr 405 589 587 588 594 394 402 403 284 293 444 451 Zr 150 417 386 392 381 154 213 199 99 100l 190 193 Y 21 48 47 46 47 67 26 27 17 16 25 25 Nb 28.0 89.0 67.5 69.5 67.9 29.7 39.3 36.6 17.4 17.8 37.4 37.5 Ga 17 nd 25 27 29 20 22 19 17 20 20 22 Cu 79 15 15 18 8 63 6 1 63 74 80 92 94 Zn 80 171 161 159 160 94 107 1f106 T85 85 inn 102 Pb 1 nd 5 4 4 3 2 1 2 1 0 2 La 22 nd 49 63 57 40 21 12 10 12 20 17 Ce 46 nd 115 125 122 75 66 56 29 32 58 62 Th 3 nd 5 7 7 3 3 3 2 4 5 4 cial conditions during late prehistory. To be sure, our interpretations are preliminarv at this stage, vet determining the spatial scale and temporal span of fine-grained adze material on Mlangareva Island ma) tell us something about the develop- ment of political boundaries and social groups on the island over time. External relations with islands outside the MIangareva archipelago are indicated b) an adze from Eiao, MIarquesas found at Rikitea, and by three flakes originating from Pitcairn Island (one flake from Akamaru and two from Atiaoa, Mlangareva). The straight-line distance from MIangareva to the Mlarquesas is -1,750 km and 148 ARCHAEOLOGICAL INVESTIGATIONS IN THE MANGAREVA ISLANDS, FRENCH POLYNESIA would entail at least two weeks of navigation far from the sight of any island landmarks. Indeed, it is likely that the transport of adzes from the Mlarquesas to Mangareva was not direct, but via "down-the-line" exchange via the Tuamotu ar- chipelago. Even the Pitcairn source, at -400 km, is at least four days sail distant. These are pur- poseful, long-distance movements that required planning and navigational skill, not to mention the expense of building and maintaining ocean- going watercraft. The reasons for long-distance voyaging and exchange have been canvassed be- fore (Earle 1997) with Anderson most recently suggesting that such voyages xvere limited in ex- tent (2003:173). However, with geochemical sourcing studies in Eastern Polynesia still being in their early stage (Weisler ed. 1997), it is pre- mature to attempt to assess the frequency of in- ter-archipelago interaction on a regional scale. Nonetheless, we do know that inter-archipelago interaction ended sometime during the 15th cen- tury (Weisler 2002) when the presence of im- ports is no longer detected in the archaeological records of man) Eastern Polynesian island groups. Mlangareva was part of one of the best-docu- mented prehistoric interaction spheres in Polynesia. From -A.D. 900 to 1450 there is abun- dant and varied evidence of the transfer of culti- gens, domesticated animals, oven stones, me- dium-grained basalt, and vesicular oven stones to resource-poor Henderson Island some 400 km to the east (Weisler 1997:fig 9.9). From our re- cent sourcing studies, we know that fine-grained basalt from Pitcairn Island was transferred back to Mangareva where it appears on that island as well as on Akamaru. One flake of this Pitcairn basalt had been identified previously on Aukena (Weisler and Wloodhead 1995). This is a rare ex- ample in the prehistoric record of Polvnesia of the twxo-waj transfer of commodities. While we have tentatively documented the content and diversity of this interaction sphere, and have suggested reasons for its inception as well as its demise (Weisler 1997, 2002), we now need to understand the operation of this inter- action sphere within Mlangareva. When was Mlangareva colonized and how soon after did daughter populations bud off to found new settlements to the east? Was the development of this interaction sphere a strateg) for island colonization (Kirch 1988), or did it serve to en- hance the prestige of Mlangarevan chiefs who acquired and regulated the distribution of valu- able commodities such as turtles, red feathers, and fine-grained basalt? Can we tie political events on Mlangareva, such as the development of monumental architecture, to the chan 0ing diversityN and magnitude of transferred commodi- ties? Were more prestige items brought to Mlangareva during later prehistorv, at the very time of increased status rivalry? Or, did the de- terioration of Mlangarevan society coincide with the end to inter-island voyaging, as Weisler (2002) has suggested? Further characterization and sourcing studies, chronologically tied to the sociopolitical and economic events of Mlangareva, will provide a fuller understanding of the processes of development and change within Mlangarevan society. CFIAPm:7H R 7 Exri)mr7TH 'This total does not include flaked st()nc from the Atiaoa or Kamaka sites xwhich was collected in 2000, but wvhich xwe have not been able to analyze in dctail.