Volume 91 1982 > Volume 91, No. 2 > A revision of the Anuta sequence, by P. V. Kirch, p 245-254
A REVISION OF THE ANUTA SEQUENCE
Anuta, lying east of the Santa Cruz Islands, is one of the Polynesian Outliers situated on the fringe of Melanesia. It is one of the smallest yet most densely populated islands of the Pacific, with a land area of 0.4 km2 and a population density of 432 per km2. The population of Anuta maintained close ties with the inhabitants of Tikopia, of ethnographic fame, only 137 km to the south-west (Firth 1954). Anutan oral traditions claimed two periods of human settlement, with an autochthonous population (the apukere, ‘earth sprung’) having been supplanted some 12 generations ago by immigrants frum ‘Uea’ (presumably 'Uvea, or Wallis Island). The first linguistic study of Anutan, by Green (1971), suggested that the present population had derived from a Nuclear-Polynesian speaking source, in Western Polynesia, though a number of lexical items also appeared to reflect a period of Tongan borrowing.
As part of the first phase of the Southeast Solomons Culture History Programme (Green 1976) in 1971, Kirch and Rosendahl (1973, 1976) conducted archaeological investigations on Anuta, in a relatively deep, stratified site (SE-AN-6) immediately inland of Rotoapi Village. A large assemblage of plain ceramics, one-piece Turbo shell fishhooks, Tridacna and Cassis shell adzes, ornaments, manufacturing tools, and other artefacts was obtained. Basing their interpretations of this site upon an “essentially uniform stratigraphic regime with three major layers”, Kirch and Rosendahl (1973:98) outlined a tentative cultural sequence for the island. Supported by radiocarbon age estimates for Layer III, occupation was believed to have begun ca. 950 B.C., and to have continued unabated until perhaps A.D. 500, when the island was abandoned. A hiatus or period of non-occupation of Anuta followed, until the island was recolonised by a Polynesian-speaking population after about A.D. 1600. This recolonisation gave rise to the present-day Anuta population.
The artefactually-rich assemblage from Layer III of Site AN-6 was clearly significant for an understanding of south-west Pacific prehistory, yet it posed several problems that defied resolution in 1971. Anutan ceramics were in many respects similar to Lapita pottery, but totally lacked the characteristic dentate-stamped style of decoration, and it was possible that they represented a second ware distinct from Lapita, and overlapping with it in time. The large array of one-piece fishhooks, on the other hand, appeared as though they might be candidates for a fishing gear technology ancestral to the more developed hook arrays of - 246 Triangle Polynesia, and especially of Eastern Polynesia. Most intriguing was the apparent hiatus in the Anutan sequence, suggesting that cultural replacements and multiple origins “may be the rule rather than the exception in Outlier Polynesia” (Kirch and Rosendahl 1976:244).
The Anuta cultural sequence as presented by Kirch and Rosendahl (1973, 1976) was subsequently questioned by several authors, with the enigmatic Layer II deposits and the putative hiatus being the foci of controversy. Davidson (1975) was the first to question our interpretation of the AN-6 stratigraphy, noting that “the oven which provides the date for the upper part of this [Layer II] sequence is at the seaward end of the site, and may have little or no relationship with the cluster of fishhooks and other artefacts in Area C further inland” (1975:253). She further suggested that “if there is a hiatus in the sequence at all, it is as likely to be within Layer II” (1975:253). In another paper, Davidson extended her argument:
. . . the dated oven, which provides the upper limit for this supposedly unbroken sequence [Layers III-II] is associated with a small cluster of Layer IIa features in the seaward part of the site; most of the important fishhook collection from Layer II, however, was obtained from Layer IIb in the most inland part of the site, immediately above an oven dated to 2515 ± 90 BP (I-6274). The nature of Layer II is not satisfactorily explained . . . .
On the evidence presented, it could equally be argued that the important assemblage of pottery, fishhooks and other items from Layer III and the lower part of Layer II can be contained within a sequence of 500 years and quite possibly less in the early half of the first millennium BC, and that there is a hiatus within the perplexing Layer II deposit, before the cluster of features at the seaward end of the site which can be dated to 1390 ± 90 BP (Davidson 1974:276).
Bayard, in his review of Polynesian Outlier cultural relationships, concurred with Davidson's comments to the effect that “the evidence for a Layer I-II hiatus is not conclusive” (1976:65). Bayard brought to attention a further difficulty with Kirch and Rosendahl's stratigraphic interpretations:
While the excavators use the Layer II date of 560 A.D. to date the end of the III-II sequence, and state that the sample dated comes from the top of Layer II (pp. 97, 100), they also state elsewhere (pp. 41, 98) that the existence of the hiatus is suggested by the sterility of upper Layer II, which to me would imply that the hiatus apparently began before the sample was deposited. In fact, their section drawings make it clear that the sample in question does not come from the top of Layer II, but rather from the bottom of an oven cut from lower Layer IIa into Layer IIb (Feature 9, Figure 6b). Thus the hiatus, if present, may be represented by Layer IIa (Bayard 1976:65-66).
Davidson (1974, 1975) and Bayard (1976) further questioned the tentative correlation of the Anuta sequence with the island's traditional history of a two-phase occupation, with the autochthonous apukere superseded, after a period of abandonment, by the ancestors of the present Polynesian population. This point was also addressed by Feinberg (1976) who, however, was unable to present any new evidence for or against the sequence proposed by Kirch and Rosendahl.
It was against this backdrop of contention over the Anuta cultural sequence - 247 that, when a second phase of field work was proposed for the Southeast Solomons Programme in 1977, plans were included for further excavations at the important AN-6 site (Yen and Green 1976). The explicit purpose of such excavations was to determine the exact nature of the enigmatic Layer II deposits, and to resolve the question of a hiatus in the island's occupation sequence. As fate would have it, attempts to land on Anuta in 1977 and again in 1978 were thwarted by high seas and a breakdown in local shipping. When I finally was able to reach Anuta in late December 1978 in the mission vessel, my stay had to be limited to a mere four hours.
STRATIGRAPHIC REANALYSIS OF THE SE-AN-6 SITE
Although unable to test Davidson's and Bayard's suggestions with on-site observations and excavations, I decided to undertake a thorough review of the original field notes and sections from the 1971 excavations, paying particular attention to the internal relationships of the Layer II sub-units. In this task, I was further guided by new knowledge of the importance of shoreline processes, gained from the extensive investigation of Tikopia by Yen and myself in 1977–78 (Kirch and Yen, 1982). As a result of this review, I now believe that we made a serious error in regarding the Layer II deposits as essentially uniform. As I shall illustrate below, it appears likely that Layer II consists of at least three major stratigraphic horizons, and that a hiatus, if present, occurred between the second and third of these.
Figure 1 is a plan of part of Rotoapi Village on the southern coastal flat of Anuta, showing the locations of the various excavations that comprised Site AN-6 (cf. Kirch and Rosendahl 1973, Figs. 1, 4). As shown in Figure 1, two sets of stratigraphic sections through the AN-6 deposits have been constructed, and are plotted as transects in Figures 2 and 3. Transect A runs along a roughly inland-coastal axis, incorporating TP-28, Area A, TP-10, and TP-26. Transect B runs approximately southwest-northeast, intersecting Transect A at Area A, and incorporates TP-9, Areas B, A, D, and C, and TP-37. After plotting the stratigraphic sections in Figures 2 and 3, all original field descriptions of strata were reviewed, as well as the available data on artefact and midden distribution and density. Based upon these analyses, the correlations shown in Figures 2 and 3 were drawn; these are, of course, provisional, and should ideally be tested through further field excavations. In order to avoid confusion with the original excavation report, the stratigraphic units as re-analysed here will be termed stratigraphic zones (cf. Kirch and Yen, 1982). The horizontal distribution of these zones, to the extent that can be ascertained, has also been shown in Figure 1. The revised sequence for Site AN-6 is set out below in terms of these stratigraphic zones.
Zone E The initial human colonisation of Anuta, at ca. 950 B.C., is represented by the Zone E deposits, containing a relatively high density of ceramics, as well as other diagnostic portable artefacts (Tridacna hinge region adzes, one-piece Turbo shell fishhooks, etc.). The Zone E settlement was situated upon a dune (Zone F) surface, at a time when the shoreline was in all likelihood some 40-50 metres inland of its present position. - 248 Correlation of Zone E deposits presents no problem, as they are uniformly composed of the dark-grey midden deposits originally identified as Layer III. (For TP-37, however, I now believe that “Layer III” should be correlated with Zone D.)
Zone D This zone is represented by a lighter grey midden deposit than Zone E, but appears to represent continued occupation of the site, with no hiatus. Zone D is present in TP-28, -32, -33, -36, -37, -38, -39, and in Area D and Area C. This zone is equivalent to the Layer IIB of Area C, and is the deposit which produced the large assemblage of Turbo-shell fishhooks; it also contains the same plain ware ceramics as in Zone E, though in reduced frequency. It is quite likely that the entire time period represented by Zones E-D is no more than five or six centuries, although a time span of up to 1000 years is not precluded on available data.
Zone C Kirch and Rosendahl (1973) observed that “Layer II” in places appeared to consist of largely sterile, white dune sand. In our original field notes (Kirch, field notebook 1971–2, pp. 10, 13), an argument was, in fact, advanced that this dune sand, particularly in Area A (Layer IIB), represented dune accretion resulting from high wave or storm action. Later, we abandoned this explanation in favour of the hypothesis that
FIGURE 1- 249
Plan of part of Rotoapi Village, showing location of excavations in Site SE-AN-6. Ovals designate modern houses.
the relatively sterile deposit had accumulated through artificial mound construction (1973:41). I now believe that our original interpretation was correct, and that the sterile components of “Layer II”, grouped here as Zone C, represent a major geomorphic event: the construction, by high-energy wave action, of a dune or berm capping the seaward portions of Zones D and E. The probable direction of the storm waves that created this dune is shown in Figure 1. It also appears likely that the event represented by Zone C resulted in the aggradation of the shoreline to a position relatively close to the present one.
The event that caused the accumulation of Zone C was most likely a major cyclone, such as recur throughout the south-east Solomons-north New Hebrides area. Such a cyclone devastated Anuta in January 1972, shortly after we left the island. The high waves reputedly did extensive damage to the seawall, and washed away several houses in Rotoapi Village. Such cyclones also cause severe damage to the island's agricultural system. It is not inconceivable that a particularly severe cyclone could have so thoroughly devastated the island that the human population was exterminated as a result of starvation and inability to reconstruct the local production system. Given the intense effect of the 1952 cyclone on the much larger island of Tikopia (Firth 1959, Spillius
FIGURE 2- 250
Transect A through Site SE-AN-6, along a roughly inland-coastal axis. Refer to Figure 1 for location on transect.
1957), such a scenario for Anuta is entirely possible. Thus, if there is a hiatus in the Anuta sequence, it is likely to be represented by the Zone C-B transition.
Zone B Subsequent to the build-up of the Zone C dune and aggradation of the shoreline, and perhaps after a hiatus during which the island was unoccupied, the main locus of habitation shifted seaward, to the area of the present village. Underlying Rotoapi Village is a relatively uniform, light-grey sandy midden, up to 120 cm deep; this Zone B is devoid of ceramics, and indeed artefact densities are uniformly low. I believe that Layer IIA in Areas A and B should probably be correlated with Zone B, and may represent cookhouse activity at the inland edge of the habitation zone. If this correlation is correct, the oven in Area B dated to A.D. 580 would date the lower part of Zone B, and, possibly, reoccupation of the island following the putative hiatus.
Zone A There does not appear to be a break in the occupation sequence represented by Zones B-A. It is in Zone A, however, that certain distinctive late prehistoric artefacts (bone bead, drilled shark's tooth) appear, as well as Euro-American items in the upper 10-15 cm. Zone A is still accumulating throughout Rotoapi Village.
FIGURE 3- 251
Transect B through Site SE-AN-6, along a roughly southwest-northeast axis. Refer to Figure 1 for location of transect.
DISCUSSION OF THE SEQUENCE
Based upon the preceding stratigraphic analysis, it is now possible to construct a revision of the Anuta cultural sequence. The stratigraphic distribution of diagnostic portable artefacts for the revised Anuta sequence is given in Table 1. Initial occupation of Anuta, ca. 950 B.C., was by a population making plain ceramics, and using a variety of hinge-region Tridacna adzes, Cassis adzes, and one-piece Turbo shell fishhooks. Manufacture tools used by this initial Anutan population included abraders of coral and of echinoid spines; their ornaments en-compassed pendants, Conus shell rings, and Trochus-shell armbands. An agricultural subsistence base is indicated by the presence of Cypraea-shell food peelers, though domestic animals appear to have been absent. This initial occupation continued for a period of several centuries, during which Zones E and D accumulated.
Sometime between the mid-first millennium B.C. and A.D. 580, a major storm or cyclone resulted in the construction of a large dune deposit capping the original village site, and probably in the aggradation of the shoreline. It is very possible that the island's population was exterminated as a result of this event. Whether or not a hiatus in the occupation sequence exists, the subsequent re-establishment of the village site (as Zone B) was by a population lacking ceramics, the distinctive shell fishhooks, and the hinge-region Tridacna adzes. The accumulation of occupation Zones B and A suggests continuous occupation since about A.D. 600, with a later addition of Polynesian-speakers from 'Uvea or elsewhere in West Polynesia about A.D. 1500-1600, as reflected in the oral traditions, and in the linguistic evidence. Thus, the oral traditions relating to the demise of the autochthonous apukere, and their replacement by the present Polynesian population, probably relate to the time period represented by the Zones B-A transition.
During the decade since the Anutan excavations were originally published, considerable new information has become available concerning the prehistory of eastern Melanesia and Western Polynesia, and these data have also contributed to a fuller understanding of Anuta's position in the culture history of the south-western Pacific (cf. Kirch and Rosendahl 1973, 1976). For one, it now appears that the early Anutan ceramic assemblage should be regarded as part of the Lapitoid Ceramic Series (cf. Golson 1971, Kirch 1978, 1981), and not as a separate ceramic ware. Excavations on Tikopia, NendŎ, and the Reef Islands have now shown that Lapitoid plain ware assemblages highly similar to that of Anuta were distributed throughout this region early in the first millennium B.C. (Kirch and Yen 1982, McCoy and Cleghorn 1979, R.C. Green personal communication). The Anutan ceramics, however, are also similar to those from an as yet undated site (FU-13) on Futuna Island in Western Polynesia (Kirch 1981). In this regard, it is highly significant that a nodule of chert excavated from Layer III in the AN-6 site (Kirch and Rosendahl 1973:84, Fig. 27a) has now been shown to be of probable Futunan origin, based on X-ray fluorescence characterisation of its chemical composition (G. Ward personal communication).
The assemblage of Turbo-shell fishhooks from the AN-6 site has also now been replicated in the excavations on nearby Tikopai, where nearly identical hook forms were found in association with plain ceramics, dating to as early as 900- 252 - 253
B.C. Again, however, the picture is complicated by the discovery of several Turbo-shell hooks (with line-attachment styles identical to those from Anuta and Tikopia) in Lapita sites on the northern Tongan island of Niuatoputapu (Kirch and Dye 1979). In short, it is still difficult to assess the relative probabilities that Anuta was initially settled from one of the nearby islands of the Santa Cruz group or northern New Hebrides (Banks and Torres Islands), or from Western Polynesia, as an early “outlier”.
As we expected might be the case, the prehistoric sequence of Tikopia proved to be similar in certain respects to that of Anuta, yet the differences between the culture histories of these two islands—only a day's sail apart—are perhaps more remarkable (Kirch and Yen 1982). Firstly, there is no hiatus in the Tikopian case, but this may be a simple reflection of the larger island's greater capacity to sustain a human population, particularly following a major cyclonic disaster. More interesting is the lack of evidence in Anuta for regular trade or exchange relations with other islands in the Santa Cruz or Banks groups. For Tikopia, such evidence was forthcoming throughout the archaeological deposits, including large quantities of volcanic glass, chert and Mangaasi-style ceramics. Archaeological evidence for direct contact between Anuta and Tikopia seems to be limited to Zone A of the Anuta sequence.
As to the immediate origins of the present-day Anutan population, the further analysis of the available linguistic data by Biggs has provided a plausible hypothesis:
Anuta was probably settled from Tikopia at some stage and borrowed heavily from Tikopian. East 'Uvea was colonised by Tongans perhaps five hundred years ago and subsequently, after their language had borrowed heavily from Tonga, 'Uvean speakers came to Anuta and, in turn, their Tongan-adulterated language became donor to the language of that tiny island (1980:124-5).
The hypothesis is certainly consistent both with the oral traditions of Tikopia and Anuta, and with the archaeological evidence.
To sum up, a reanalysis of the stratigraphy of Site SE-AN-6 has indicated the following: (1) the initial phase of occupation, by a population using Lapitoid plain ware, was of considerably shorter duration than formerly thought; (2) a probable occupation hiatus is associated with the accumulation of an extensive sand dune deposit (Zone C), presumably reflecting a major cyclone; (3) reoccupation of the island, by a population lacking ceramics, occurred by A.D. 600; and (4) the advent of the present population, speakers of a Nuclear Polynesian language with Tongan borrowings, appears to date to ca. A.D. 1600, in accordance with the oral traditions.
This revision of the Anuta sequence is the outgrowth of many long discussions with my colleague Douglas Yen, during the original field work on Anuta, and at intervals since then. Auwe.- 254