Volume 95 1986 > Volume 95, No. 1 > Rethinking East Polynesian prehistory, by P. V. Kirch, p 9-40
RETHINKING EAST POLYNESIAN PREHISTORY
One result of the intensive archaeological scrutiny of Polynesia over the past three decades has seemed to be a resolution of the old “problem of Polynesian origins”. The development of an Ancestral Polynesian Society out of the Lapita Cultural Complex (Green 1979; Kirch 1984), and the subsequent dispersal of Polynesian colonisers throughout the eastern Pacific are no longer the subject of substantial arguments. Indeed, agreement on these points is so substantial that there is now an “orthodox scenario” of the settlement of Polynesia, or what Irwin (1981) has termed the “authorised version”. This scenario, which has been presented in several recent syntheses of regional prehistory (e.g., Bellwood 1978a, b; Jennings 1979; Kirch 1980), has the following main outline:
This scenario has appealed to Polynesian archaeologists because it has served reasonably well to integrate the available archaeological data, providing a framework into which the results of new excavations could - 10 be fitted. The orthodox view, however, has not been without its minor detractors (Bellwood 1970; Biggs 1972; Hunt 1979; Langdon and Tryon 1983). Most recently, in a provocative article in this Journal, Irwin (1981) challenged some of its fundamental tenets, particularly the idea that there was a 1500-year-long pause in the eastward voyaging of early Polynesian ancestors.
In recent years, I have grown increasingly suspicious of the orthodox scenario, especially as it relates to the settlement of the Hawaiian archipelago (Kirch 1974, 1975, 1984, 1985). This has prompted me to re-evaluate the archaeological evidence upon which the orthodox view was constructed, a process that has convinced me of the need for substantial revisions. This paper addresses those concerns, and focuses on the evidence for initial settlement and colonisation in Eastern Polynesia. In particular, the concept of an “Archaic East Polynesian Culture” (Sinoto 1970, 1983b) or “Early East Polynesian Culture” (Bellwood 1978a, b), its material-culture content and dating, will be reviewed in detail.
The questioning of an orthodoxy is painful. An accepted scenario—or model—such as that for the settlement of Eastern Polynesia, is an intrinsically satisfying intellectual concept, and attempts to expose questionable, underlying assumptions are not apt to be welcomed by those who have had an investment or commitment in its development. Never-theless, every model has its useful life, and when new data must be contorted to fit the paradigm, a rethinking is in order. I believe we have reached such a point with our current scenario of Eastern Polynesian settlement, and this paper is offered as a step towards constructing a new model more appropriate to the data at hand, yet one that will no doubt also see its own time.
DEVELOPMENT OF THE ORTHODOX SCENARIO
It is critical to understand the historical development of the current model within the context of the rapidly developing field of Polynesian archaeology after 1950. Before 1950, the prevailing scenario for Polynesian settlement was that summarised in popular form by Buck (1938), in which the Society Islands were regarded as the “hub of the Polynesian universe”, the primary dispersal centre from which other East Polynesian groups derived. This model relied largely on detailed trait comparisons of ethnographic-period material culture (Buck 1944; Burrows 1939), on oral traditions, and on only partially systematic linguistic comparisons (Emory 1946). The sudden advent of subsurface archaeological investigation by Emory in Hawai'i (Bonk 1961), Duff (1956) in New - 11 Zealand, the Norwegian Expedition in Easter Island (Heyerdahl and Ferdon 1961), and Shapiro and Suggs in the Marquesas (Shapiro 1958)—along with complementary excavations in Fiji, New Caledonia, Yap, and the Marianas—stimulated by the invention of radiocarbon dating, revitalised Polynesian archaeology and prehistory. The possibility of directly tracing Polynesian origins and “migration routes” through dated, material evidence stimulated field research, which in turn led to a thorough questioning of the older scenario.
Suggs (1960) was the first to attempt a complete rewriting of Polynesian prehistory based on the newly acquired subsurface archaeological data, in which his Marquesan materials played a crucial role (1961). Suggs' new synthesis stressed the following points: (1) the occupation of Eastern Polynesia “began before the Christian era and was under way by at least 200 B.C.” (1960:107); (2) the Marquesas and Society Islands were both settled at approximately the same time, around the second or third century B.C.; (3) both of these central archipelagos played important roles as “main dispersion points” for the subsequent settlement of peripheral East Polynesian islands and archipelagos. Suggs accepted Emory's early C14 date of A.D. 124 ± 120 for the settlement of Hawai'i (1960:152), and argued an approximately A.D. 1000 settlement date for New Zealand (1960:194). Suggs rejected Heyerdahl's theory for New World origins of early Easter Island culture, and maintained that Easter Island was settled about the fourth century A.D. from the Marquesas.
The orthodox scenario of East Polynesian settlement largely results from subsequent work by K. P. Emory and Y. H. Sinoto in central East Polynesia, beginning in the early 1960s—work which elaborated and substantially revised the synthesis proposed by Suggs. The Maupiti burial site (Emory and Sinoto 1964, 1965), excavated in 1962-3, provided the first direct evidence of a relatively early period of Society Island culture, with remarkable similarities to “Moa-hunter” or “Archaic” period assemblages in New Zealand. Far more important were Sinoto's excavations in the Marquesas between 1963-8, particularly at the deeply stratified Hane Dune Site (MUH 1) (Sinoto and Kellum 1965). The cultural sequence from Hane, as propounded by Sinoto (1966, 1968, 1970, 1979a), provides a fundamental underpinning for the current orthodox view of East Polynesian settlement.
The stratigraphically-complex Hane site yielded a wide variety of artefacts, including a few Polynesian Plain Ware sherds in the lowest layer; the site spans a major part of the prehistoric Marquesan sequence. In comparing the Hane materials with those reported by Suggs for Nukuhiva (particularly from the Ha'atuatua site, NHaa1), Sinoto noted - 12 discrepencies which caused him to question Suggs' proposed cultural sequence for Nukuhiva (Sinoto and Kellum 1965:37-46). A major issue was the context for two early C14 age determinations obtained by Suggs at NHaa1. Sinoto believed, on the basis of artefact type comparisons, that the lowest levels at MUH1 predated the NHaa1 site, yet the Hane C14 age determinations were surprisingly younger than those from Ha'atuatua. In 1965, Sinoto (1970, and personal communication) re-excavated at Ha'atuatua and exposed certain procedural errors in Suggs' field methods. While this cast doubt on Suggs' results, it did not necessarily invalidate the two early C14 dates. Sinoto, however, chose to rely on the contradictory C14 determinations from Hane, and revised the settlement date for the Marquesas to A.D. 700 (Sinoto 1966:296, 302; 1968). In 1970, however, after additional C14 determinations from Hane were obtained, Sinoto fixed the age of initial settlement at Hane (and, by implication, the Marquesas generally) at A.D. 300. This date for Marquesan settlement has become a key element of the orthodox scenario for Eastern Polynesian prehistory. Yet, as I hope to demonstrate, a date of A.D. 300 for Marquesan settlement has no basis in the radiocarbon corpus from early Marquesan sites.
In many other respects, Sinoto accepts the cultural sequence outlined by Suggs (1961), although Sinoto uses a numbered phase sequence to distinguish his chronologically recalibrated periods from those named by Suggs. Like Suggs, however, Sinoto regards the Developmental Period (his Phase II) as a time during which voyages of colonisation originated from the Marquesas, leading to the settlement of the Hawaiian Islands, Easter Island, and certain other south-east Polynesian groups (e.g., Mangareva, Pitcairn, Henderson) (Suggs 1961:182; Sinoto 1970, 1983c).
The orthodox scenario of Eastern Polynesian settlement received its first formal expression in Emory and Sinoto's preliminary report on their work in the Marquesas and Society archipelagos (1965). Presenting their preliminary findings in the form of a diagram (1965, fig.13), Emory and Sinoto outlined a 7-stage sequence for the settlement of Eastern Polynesia. Since this graphic model has subsequently been reproduced, either exactly or with only slight modifications (see Fig.1), in a variety of syntheses of Polynesian prehistory (e.g., Sinoto 1968:117; Bellwood 1978a: fig.11.21; Jennings 1979: fig.I.1), its ordered stages may be briefly summarised: (1) colonisation of the Marquesas from West Polynesia; (2) settlement of the Societies from the Marquesas; (3) settlement of Easter Island from the Marquesas; (4) initial colonisation of New Zealand from the Society Islands; (5) secondary settlement of New- 13
A version of the orthodox scenario of East Polynesian dispersals, after Jennings (1979, fig.I.1), with permission of Harvard University Press.
Zealand from the Marquesas; (6) initial colonisation of Hawai'i from the Marquesas; and (7) secondary settlement of Hawai'i from the Societies.
It is critical that this putative model was constructed on the basis of a small number of artefact “types”, primarily from the Hane and Maupiti sites. For example, the postulation that “the earliest migration to Hawaii was most likely from the Marquesas and that contact between Hawaii and the Society Islands came later” was based solely on a generalised comparison of fishhook styles between the assemblages in question and those of the South Point sites on Hawai'i Island (Emory, Bonk, and - 14 Sinoto 1959). Similarly, the suggestion of a direct link between the Marquesas and New Zealand (stage 5 in the diagram) was based only on the presence of reel ornaments in the Hane site and New Zealand, and on their absence in the Society Islands assemblages.
The Emory-Sinoto model of 1965 (which forms the basis of the orthodox scenario) received further support from the major linguistic sub-grouping analysis of Green (1966). Green's model appeared to mesh closely with the emerging archaeological data in suggesting that the Marquesas were the location of an ancestral Proto-East Polynesian (PEP) speech community, from which the language ancestral to modern Easter Island quickly diverged. The subsequent split between Proto-Marquesic and Proto-Tahitic branches of PEP could then be correlated with the colonisation of the Society Islands from the Marquesas. Likewise, Green demonstrated that Hawaiian should be subgrouped with Marquesan, but that certain innovations uniquely shared between Hawaiian and Tahitian implied later borrowing. Thus, the two-stage colonisation of Hawai'i was also supported. Green was careful to point out that the linguistic evidence, in and of itself, could not be used to specify the locations of these various early speech communities. (He carefully noted, for example, the possibilities that Proto-Marquesic could have been spoken in Hawai'i or Mangareva.) There is no linguistic evidence, for example, to preclude either the Society Islands or the Cooks as the location of PEP. Not unreasonably, however, Green attempted to mesh his sub-grouping model with the emerging archaeological picture; since the earliest dated assemblages were from the Marquesas, it was logical to suppose that this was the PEP “homeland”.
The Emory-Sinoto model of Eastern Polynesian settlement was given fuller expression in two subsequent papers by Sinoto (1968, 1970). The artefactual evidence for the 7-stage dispersal model was succinctly laid out by Sinoto:
Shaped whale-tooth pendants of a type identical with those found in the Maupiti and Wairau Bar burials were discovered in the Hane site, but without burial association. So far in East Polynesia, there are three island groups which have yielded such whale-tooth pendants of the early period: the Marquesas, Society, and New Zealand. This order could be the route of the diffusion. Reel ornaments, which are associated with these whale-tooth pendants in New Zealand, have not been found in the Society Islands. However, from a stratigraphically later level than that of the whale-tooth pendants in the Hane site, a reel was found in association with a burial, and reels are observable in ethnological collections of Marquesan - 15 materials. Harpoon heads were found in both the Marquesas and New Zealand, but not in the Society Islands. The tattooing chisels found in the Hane site are similar to those found in New Zealand. The above evidence seems to indicate a direct link between the Marquesan and New Zealand cultures. A harpoon head and also fishhook types found by Green in Mangareva suggest a link about this time.
While studying Suggs' Nukuhiva fishhooks, we noticed a close similarity between early Marquesan fishhooks and early Hawaiian fishhooks. Fishhooks from Hane appear to have an even closer resemblance to early Hawaiian hooks, not only in the types of fishhooks present but also in the materials, in the ratio between the height of the fishhook point to that of the shank, and in the tools used in their manufacture. The late Hawaiian fishhooks are closely related to the late Tahitian fishhooks. The above is evidence for postulating that the earliest migration to Hawaii was most likely from the Marquesas and that contact between Hawaii and the Society Islands came later. . . .
At the moment, to place the Marquesas Islands as a dispersal point for East Polynesian migration would seem to be a likely determination (Sinoto 1968:116-7, emphases added).
In his 1970 paper, Sinoto reviewed the artefact sequence from the Hane site in more detail, and—based on a series of C14 age determinations— suggested that Phase I (Suggs' “Settlement Period”) dated from A.D. 300-600 and Phase II (“Developmental Period”) from A.D. 600-1300. In keeping with this revised Marquesan chronology, Sinoto then proposed that the “early dispersal to the Society Islands, Hawaii, and Easter Island probably took place between A.D. 650 and A.D. 800 . . . , and to Mangareva about A.D. 1200 . . .” (Sinoto 1970:130).
In dating Hane Phase I to A.D. 300, Sinoto shortened the Marquesan sequence of Suggs by 500 years. This clearly posed a problem with the early C14 dates obtained by Emory at the South Point sites on Hawai'i Island. (The date of A.D. 124±120 from Layer III of the H1 site had formerly been widely accepted as indicative of the age of Polynesian colonisation of Hawai'i.) If the Marquesas were not settled until A.D. 300, and if, as Suggs and Sinoto had both proposed, Hawai'i was colonised during the Marquesan Developmental Period (Phase II, A.D. 600-1300), the Hawaiian sequence would also have to be revised. The problem was exacerbated by a large series of conflicting C14 age determinations from the South Point sites. In a paper reanalysing the South Point sequence, Emory and Sinoto (1969) rejected the earliest age determinations for Site H1, and now proposed that Hawai'i had not been col- - 16 onised until about A.D. 750. This, of course, was the only interpretation of the South Point dates which would be acceptable in light of Sinoto's revision of the Marquesan sequence. I shall return to the problem of the South Point dates later in this paper.
The accidental discovery in 1972 of a partially submerged, water-logged site at Vaito'otia on Huahine in the Society group (Site ScH1-1; Sinoto and McCoy 1975) added a wealth of new archaeological data from the Society Islands which had to be incorporated into the EmorySinoto model. The initial excavations at ScH1-1 by Sinoto and McCoy yielded certain artefacts very similar to those from Phases I-II at Hane, although the C14 dating of a whale rib appeared to be too recent by comparison with the Marquesan sequence. In their preliminary report Sinoto and McCoy noted that the materials from Vaito'otia might necessitate a “re-evaluation of the hypothesis of the Marquesas Islands as the primary dispersal center in East Polynesia” (1975:184).
This first expression of reconsidering the Emory-Sinoto model of 1965 was, however, quickly dropped. In subsequent reports on the Vaito'otia-Fa'ahia excavations, Sinoto (1979b, 1983, 1983c) regarded the new Society Islands finds as direct support for the orthodox scenario. Thus, concerning the Vaito'otia-Fa'ahia assemblage, he writes:
This somewhat complex material culture was most likely brought in from the Marquesas, rather than developed in Huahine; the assemblage is nearly identical to those of the early Marquesan period, although pottery culture had already been dropped in the Marquesas. The presence of hand clubs, made and used at the Huahine site, rather than imported from New Zealand in historic times, establishes strong evidence of settlement patterns from the Marquesas to the Society Islands and then to New Zealand (1983:597).
The most recent elaboration of the 1965 Emory-Sinoto model has been the formal definition of an “Archaic East Polynesian Culture”, based on the Phase I-II assemblages from the Hane site, and on the newer ScH1-1, -2 materials. In Sinoto's words:
The diagnostic material culture assemblage of Archaic East Polynesia includes untanged adzes of various types, shaped whale-tooth pendants, one-piece fishhooks, one-piece hooks with compound shanks, trolling hooks, toggle-head harpoons, and hand clubs—called patu by the New Zealand Maoris. So far the Marquesas has yielded the earliest radiocarbon date, A.D. 300, and the pottery. Therefore, it seems logical to assume that - 17 the Marquesas was the dispersal center in East Polynesia. There is also good evidence for the movement of culture from the Marquesas to the Society Islands, and then to New Zealand (Sinoto 1981:2).
(It is interesting to note that the date of A.D. 300, originally derived from a series of rather divergent C14 age determinations from the Hane site, had by 1981 become reified as “the earliest radiocarbon date”.)
This historical account has been necessary to trace the development of the orthodox scenario for Eastern Polynesian settlement and, especially, to establish the evidence upon which it was based. To sum up, the orthodox view remains that of the 7-stage dispersal model presented by Emory and Sinoto in 1965, closely keyed to Sinoto's chronological revision of the Marquesan sequence, and with the more recent formal definition of an “Archaic East Polynesian Culture”. The key elements in this model are:
MUDDLES IN THE MODEL
The orthodox scenario, although widely accepted and frequently presented as solidly established (e.g., Jennings 1979:fig.I.1), has not been without its detractors virtually since its formal proposal by Emory and Sinoto in 1965. In a well-argued paper, Bellwood (1970) took exception to the Emory-Sinoto model, especially its claims for the Marquesas as a “primary dispersal center”. Bellwood noted that the dispersal model postulated by Emory and Sinoto was a very simple or “economical” one (in the sense of a minimal number of settlement steps), but that it could well be a result of limited archaeological activity.
. . . a large number of dispersal schemes could be formulated. Which one fits the evidence from East Polynesia? Until every island group has been examined, we are not in a position to answer this question. Isolating the primary disperśal center is one thing, but also there may be pre-primary dispersal centers and secondary dispersal centers with which we have to contend (1970:94).
Bellwood rightly observed that the environmental setting of the Marquesas favoured the discovery of early settlement sites, whereas geomorphological conditions in the Society Islands were quite the opposite; hence, a major sampling problem could well “mask” evidence of early settlement in the Societies.
Bellwood also noted the lack of a close match between the earliest artefact assemblages in the Marquesas and those in West Polynesia which were deemed to be directly ancestral.
If East Polynesia was first settled from Samoa, or even from Tonga, then we would expect the earliest eastern archaeological evidence to tie in closely with that of West Polynesia. In the author's opinion it does not yet do this. The earliest sites in the Marquesas and Society Islands, Hawaii, Easter Island, and New Zealand reveal a material culture which already has traits differentiating it from that of West Polynesia. Hence, for East Polynesia we have a polythetic assemblage whose immediate origins are not apparent in West Polynesia, although its slightly more distant origins are. The author suggests that these immediate origins lie somewhere in East Polynesia and have not yet been located with certainty (1970:96).
Davidson (1976) was troubled by this same problem, noting that “if we are to accept that Eastern Polynesia was colonized from Western Polynesia, there must have been a point at which a pioneering Eastern - 19 Polynesian culture was indistinguishable from the Western Polynesian culture of at least one island group” (1976:46). In her opinion, not only has this point not yet been archaeologically defined, but “the differences between early Eastern Polynesian cultures defined by archaeologists and any Western Polynesian cultures are still greater than any differences we can confidently identify within Western Polynesia”.
Biggs (1972), in a cautionary essay targeted at the facile use of linguistic models by archaeologists zealously searching for simple solutions to complex settlement problems, raised further objections to the Emory-Sinoto model. With regard to the current subgrouping model of Eastern Polynesian languages, Biggs pointed out that “if Tahiti is the homeland of Tahitic, and the Marquesas are the homeland of Marquesic, then . . . both Tahiti and the Marquesas must have been settled before any of the other areas of Eastern Polynesia” (1972:148). This, of course, would be impossible to reconcile with the Emory-Sinoto model. Biggs went on to observe that a simple A to B to C settlement model (such as that from Marquesas to Societies to New Zealand) was unrealistic. “Any simplistic view of Polynesian settlement passing from A to B to C in a sequence which never retraces its steps will be false” (1972:149). Rather, Biggs suggests that as our knowledge of Polynesian prehistory advances, it will be necessary to propose a theory of “multiple intra-Polynesian migration and settlement”. Linguistic data from various Polynesian islands, such as 'Uvea, Rotuma, Rurutu and Niuatoputapu, in fact attest to such multiple settlements.
Although not discussed by Biggs, the orthodox scenario poses another problem for the accepted subgrouping model of Eastern Polynesian languages. Since all Eastern Polynesian languages share certain lexical and phonological innovations in common, this implies that a unified Proto-East Polynesian speech community existed long enough before the primary split between Tahitic and Marquesic branches for such innovations to develop. The same can be said for other aspects of culture as well, for it is clear that Eastern Polynesian societies share many features which must have been developed in an ancestral community before dispersal to the various East Polynesian islands and archipelagos. These include, for example, the typical Eastern Polynesian marae concept of court, elevated ahu, and upright representations of deities or ancestral figures. It is questionable whether the Emory-Sinoto model, with the rather late settlement of the Marquesas (A.D. 300) and fairly rapid dispersal to Easter Island and the Societies, would allow sufficient time for such linguistic and cultural innovations to have developed in the ancestral East Polynesian community.- 20
Hunt (1979) also raised these points in an unpublished critique of the orthodox scenario for East Polynesian settlement. Hunt observed that the archaeological evidence from West Polynesia suggests that the ancestral Polynesian homeland was never focused on a single island; rather, a regional concept of a dispersed speech community maintained through frequent interisland contact is more reasonable (see also Green 1981 and Geraghty 1983 on the idea of regional “homelands”). The same may well have been true of early Eastern Polynesia, with a “speech community spread over more than a single island, or archipelago” (Hunt 1979:7). In this regard, the great similarities between the early Marquesan and Society Island (Vaito'otia-Fa'ahia) assemblages are noteworthy.
Another objection to the orthodox scenario centres on the difficulties of voyaging direct from Western Polynesia to the Marquesas, given the 1,800-mile distance and the generally opposing winds and currents. This objection has been raised by Biggs (1972) and others, who feel that “islands which were closest would be settled first in the upwind struggle to the east” (1972:148). Finney (1985) suggests that “anomalous westerlies” associated with E1 Nino events might have made direct voyaging from Samoa to the Marquesas possible. None the less, a chain of settlement leading through the Cook Islands and Society group to the Marquesas would be more in keeping with what we know of the process of oceanic colonisation in the south-western Pacific.
Perhaps the most serious challenges to the Emory-Sinoto model have been those posed by recent data from the Hawaiian Islands, specifically: (1) the presence of several sites with C14 age determinations predating the supposed A.D. 750 Marquesan settlement of the archipelago; and (2) the absence in these assemblages of many of the supposedly diagnostic artefact traits that define “Archaic East Polynesian Culture”. The dating of the Hawaiian sites will be discussed in detail below, but I should note here that an initial settlement of Hawai'i by perhaps the fourth century A.D. seems increasingly plausible. Such a date greatly strains the Emory-Sinoto model, in which the Marquesas themselves were just colonised by A.D. 300. The Archaic East Polynesians would no sooner have established a foothold in the Marquesas than they would have launched their canoes again for the North Pacific. Furthermore, this would be unacceptable in terms of the linguistic subgrouping model, for the reasons just noted. Equally disconcerting is the absence, in these early Hawaiian assemblages (such as the South Point, Halawa and Bellows sites) of many of the classic “Archaic East Polynesian” artefact types, such as shaped whale-tooth pendants, harpoon heads, and - 21 compound-shank fishhooks.
The questions posed by the recent Hawaiian data have not gone unnoticed by Sinoto, who remarked that the early dates “pose questions on the relation to those of the Marquesan and the Society Islands' sites” (1983c:65). One solution to the absence of the early diagnostic artefacts, of course, is simply that “the initial settlers to the Hawaiian Islands did not bring with them those . . . items” (1983c:65). Alternatively, Sinoto has suggested that Hawai'i may have been settled from the Southern Marquesas, which on present evidence lacked shaped whale-tooth pendants and harpoons (Sinoto 1979a:131). This argument is less than compelling, however, since the Southern Marquesas remain poorly explored archaeologically (and these absences may therefore simply be explained by limited sampling to date).
While such explanations can be invented to account for the absence of the diagnostic artefacts in the early Hawaiian sites, the early dates remain a serious problem. Unwilling, apparently, to accept a revision of the Marquesan chronology, Sinoto (1981) simply proposed that the recent radiocarbon evidence from Hawai'i be ignored: “it seems to me that it is more reasonably convincing to establish a sequence using these kinds of [typological] relationships than by absolute dating” (1981:5).
The most recent challenge to the orthodox scenario was the gauntlet thrown down boldly by Irwin (1981), who presents several reasons why the “discontinuous” model of Polynesian settlement (in which there was a gap of 1500 years between the settlement of Western and Eastern Polynesia) is unlikely. Irwin argues that the “absence of predicted early evidence in Eastern Polynesia can be explained in terms of (1) patchy archaeological field work and sampling error; (2) the fact that Lapita could be expected rapidly to turn aceramic” (1981:491). Irwin reminds us that a discontinuous model has once before been proven false within Polynesia, namely that of Tonga as the Polynesian homeland (Groube 1971), with a substantially later settlement of Samoa. The evidence to falsify that model—Lapita ceramics—lay submerged at Mulifanua, Upolu, until accidentally dredged to view (Green and Davidson 1974).
In short, numerous objections—both theoretical and substantive—have been raised against the orthodox scenario for Eastern Polynesian settlement. Briefly, these include: (1) the uneven nature of archaeological survey and excavation in Eastern Polynesia, and consequent problems of sampling error; (2) the substantial differences between the earliest dated East Polynesian assemblages and those known from West Polynesian sites of comparable age; (3) the problem of sufficient time for the development of shared cultural and lexical innovations in an - 22 ancestral East Polynesian community before dispersal; (4) theoretical arguments against a simple A to B to C settlement sequence; (5) radiocarbon ages from Hawaiian sites which appear unacceptably early in light of the accepted Marquesan chronology; and (6) the absence of diagnostic “Archaic East Polynesian” artefact types in the earliest Hawaiian assemblages.
The time is ripe for a careful re-evaluation of the orthodox scenario, and the evidence upon which it is based. The Emory-Sinoto model of 1965 was a reasonable set of working hypotheses based upon the rapidly emerging picture of East Polynesian prehistory at that time. Nearly two decades later, it is increasingly difficult and unsatisfying to continue to force new finds into this orthodox paradigm.
THE EVIDENCE RECONSIDERED
The Marquesas have played the pivotal role in the orthodox scenario, therefore it is appropriate to begin the process of re-evaluating the available evidence there. Three topics must be considered: (a) sampling, (b) chronology, and (c) the stratigraphic evidence for the development of early East Polynesian material culture.
Sampling. Although early sites have not been difficult to locate in the narrowly-confined Marquesan valleys, archaeological coverage of the archipelago has by no means been complete. The southern islands of Tahuata, Hivaoa, and Fatuhiva remain relatively unexplored, raising the possibility that our knowledge of early Marquesan settlement is skewed by sampling error. Sinoto cautions that “it is probable that the early part of the initial period of Marquesan culture has not yet been discovered” (1983c:65). Such an earlier stage might well be expected to fill the gap between the Hane-Ha'atuatua materials and those known from comparable West Polynesian sites, as suggested by Bellwood (1970), Davidson (1976) and others. Current field work by B. Rolett (personal communication 1985) in the southern Marquesas is aimed at this very issue, and may help to answer the question of whether earlier sites are to be found on these islands.
Chronology. Whether or not earlier sites are forthcoming, it is evident that the Marquesan chronology as proposed by Sinoto (1970) is unacceptable—a serious issue given that the dating of the Marquesan Settlement Period (Phase I) is central to the entire orthodox scenario.
In their report on the 1963-4 excavations at Hane, Sinoto and Kellum - 23 listed an initial series of C14 age determinations, all by the Gakushuin Laboratory (GaK). The earliest determination of 1100 ± 100 (GaK-529), from charcoal under the stone pavement in Layer V, was much more recent than expected given Suggs' dates from the Ha'atuatua site on Nukuhiva (Suggs 1961:20, 63). Sinoto and Kellum wrote (1965:36): “On the basis of the analysis already made it was felt that the radiocarbon dates were too recent and a rechecking was requested. Dr Kigoshi replied that a slight difference in the dating of the known-age material was observed at the time these were run”. Sinoto and Kellum, however, accepted the date of A.D. 850 for Layer V and—arguing that underlying levels VI and VII accumulated rapidly—proposed a date of A.D. 700 for initial occupation at MUH1. These arguments were reiterated in Sinoto (1966).
By 1970, when Sinoto published his major paper on the role of the Marquesas as a dispersal centre in East Polynesia, a more extensive series of C14 age determinations had been obtained from Hane, not only from the Gakushuin Laboratory, but also from the Washington State University Laboratory (WSU). According to Sinoto, this extended series of C14 determinations provided “fairly reasonal [sic] datings of the cultural sequences” (1970:113), including the assignment of the span A.D. 300-600 for Phase I. While the new C14 ages were listed in an appendix (1970:131), no analysis or discussion of these determinations was offered (nor were the C14 ages corrected for secular variation). One presumes that Sinoto believed the proposed dates for his cultural periods to have been intuitively obvious based on the list of C14 age determinations.
A close examination of the Hane C14 series, however, reveals serious inconsistencies and problems of interpretation. A total of 17 C14 age determinations were processed from the lower stratigraphic levels of MUH1 (Layers VI, VIa, and VI; no samples were dated from Layer VII). When these ages are corrected for secular variation (after Klein et al. 1982) and plotted at 95 percent confidence intervals as in Figure 2, a striking pattern emerges. Rather than sorting out by stratigraphic context, the samples group by laboratory, with the Gakushuin series distinctly younger than that processed by WSU. Indeed, all 10 Gakushuin samples are unacceptably recent for these early stratigraphic contexts, raising the likelihood that significant errors in pretreatment or laboratory method were involved. (Significantly, this is not the only instance of questionable C14 age determinations from the Gakushuin Laboratory, cf. Kirch 1975.) Considering only the WSU C14 series, it is found that these span a period from as early as the second or third century B.C. to as late as the mid-second millennium A.D. While the WSU- 24
Radiocarbon age determinations from Layers VI and V in the Hane dune site, corrected for secular variation and plotted at 95 percent confidence intervals (after Kirch 1984).
ages are more in keeping with expectations based on our knowledge of Polynesian settlement generally, they are characterised by extended standard deviations, and it is impossible to tie down a precise “date” for the initial settlement of the Hane site based on this series. Statistically, all that can be reasonably said of the Hane chronology is that Layers V and VI date to sometime between the late first millennium B.C. and the mid-first millennium A.D. Layer VII, the lowest, remains undated. While this may be unsatisfying to those desiring precise “dates”, it is the only proper manner in which to interpret this series. Certainly, Sinoto's use of A.D. 300-600 for Phase I is without foundation in the radiocarbon data from the Hane site.
From this perspective, we may reconsider the C14 ages from NHaa1 (Ha'atuatua) obtained by Suggs (1961). The two uncorrected ages of 2080±150 and 1910±180 are corrected to 405 B.C.-A.D. 220 and 385 B.C.-A.D. 450 at 95 percent confidence intervals following Klein et al. (1982). While the exact stratigraphic contexts of these samples are in doubt, the fact that they evidently do date some archaeological phenomena has never been denied (Sinoto 1970:131). Also of note is a - 25 C14 determination of 1930 ± 80 B.P. (corrected to 165 B.C.-240 A.D. at 95 percent confidence) from the Hanatukua rockshelter site (MH-3-11) on Hiva Oa; underlying layers in this site have never been dated.
In sum, when the Hane C14 series is scrutinised and compared with the available determinations from NHaa1 and MH-3-11, the available data not only do not support the proposed Marquesan settlement date of A.D. 300, they also permit the extension of Marquesan settlement back as much as 500 years, to the second century B.C. When we further consider that (a) the earliest stratigraphic contexts in both MUH1 and MH-3-11 remain undated, and (b) the sampling problems raised earlier, the possibility that the Marquesas were colonised as early as the mid-first millennium B.C. deserves a serious hearing.
The Hane Sequence. The stratigraphic evidence from Hane itself bears further scrutiny. Regrettably, Sinoto has yet to publish a detailed report on his extensive excavations at MUH1, so that we lack many basic kinds of contextual data on the vertical and horizontal distribution of artefacts and faunal remains in the Hane deposits. Until these are made available, a full analysis of the significance of the Hane sequence for Marquesan prehistory will be impossible. Nevertheless, with the limited data available (Sinoto and Kellum 1965; Sinoto 1966, 1970), some important points are discernible.
As Sinoto and Kellum (1965) and Sinoto (1966) indicate, the stratigraphy in site MUH1 was complex, as is typical in coastal dune sites throughout Oceania. The major excavations focused on Area B, the “main mound” (Sinoto 1966: fig.2), where a sequence of six strati-graphic units was uncovered. Levels I-III represent the late Expansion to Classic Periods of Marquesan prehistory, Level IV is a dune accumulation containing Expansion Period burials, and Levels V and VI represent the Settlement to Developmental Periods. Several test pits to the west of Area B exposed a yet earlier stratum, Level VII, which yielded the few ceramic sherds from this site. Seaward of the main mound, excavations in Area A revealed six stratigraphic units, with Levels II and III correlating with Level V of Area B, and Levels IV, V, and VI correlating with Level VI of Area B.
Unfortunately, nowhere in his various reports on the MUH1 excavation does Sinoto provide quantitative data on the distribution of cultural materials according to these primary stratigraphic units. In his presentation of the revised Marquesan sequence, Sinoto (1966, 1970, 1979a) lumps the cultural content of Levels VI and VII in Area B with IV-VI in Area A under the rubric “Phase I”, and Levels V in Area B and II and III in Area A under “Phase II”. Thus, one is led to believe that there - 26 were no significant differences in the cultural contents of the various individual stratigraphic units that comprise Phase I, and likewise for those that comprise Phase II. Presumably, the only significant differences are those between the aggregate Phase I strata and the Phase II strata. That such is indeed not the case is indicated only by the data presented graphically in Figure 3 of Sinoto (1966). This chart, which plots the presence/absence of certain artefact types by stratigraphic units (using the Area B sequence only), clearly indicates that there were, in fact, substantial and highly significant differences between the individual stratigraphic units that are otherwise lumped under the headings “Phase I” and “Phase II”. The earliest stratigraphic unit, Level VII, is indicated as containing only pottery, three types of one-piece fishhook, and trolling hook points of “West Polynesian” type (i.e., with double perforations and proximal base). It is significant that whale-tooth pendants and harpoon heads, two items which Sinoto regards as “diagnostic” of his Archaic East Polynesian Culture, do not appear until Level V of the Area B sequence. Reel ornaments, another Archaic East Polynesian diagnostic trait, do not appear until Level IV (Table 1).
While we lack the data to carry this analysis further, the implications are fundamental for the Marquesan sequence as currently constructed.
Stratigraphic Distribution of Selected Artefact Types at Hane, Uahuka 1
For one, Sinoto's Phase I, which is described in classically “normative” terms in his summaries of the Hane site (1970, 1979a), is actually composed of at least six different stratigraphic units, with highly varying cultural contents. The implication is that Phase I was actually a period of substantial change in material culture (and presumably in economic adaptation as well, cf. Kirch 1973). Furthermore, it is highly misleading to lump Phases I and II under a rubric such as “Archaic East Polynesian Culture”, since many of the purported diagnostic traits of this “culture” do not appear until the third or fourth depositional units in the Hane sequence.
When we combine this incomplete reanalysis of the Hane stratigraphic sequence with the evidence of the site's radiocarbon chronology, the probability emerges that we are not dealing with a simple two-phase sequence beginning about A.D. 300. Rather, the Hane materials may represent a much longer time span, beginning possibly in the second half of the first millennium B.C., and incorporating a complex series of cultural changes. What Sinoto has termed “Archaic East Polynesian Culture” is perhaps not truly “archaic” at all, but an intermediate phase in the development of Marquesan culture. The properly “archaic” phases, in the sense of founding stages, are those represented by Levels VII and VI, which entirely lack such distinctive items as whale-tooth pendants, reels and harpoon heads.
The content of the Hanatukua rockshelter (MH-3-11) is also remarkable in light of the above. Although no site report on this excavation is available, a brief examination of the artefacts from MH-3-11 in the Bishop Museum collection revealed that this site (with relatively early stratigraphic units) also lacks many of the purported diagnostic Archaic East Polynesian traits. What it does contain is an assemblage of pearl shell and bone one-piece fishhooks, trolling lure points, stone octopus lure sinkers, drilled dog's teeth, sea urchin and Porites coral abraders, and basalt flakes that match closely with the Levels VII and VI materials from Hane. A detailed analysis of the MH-3-11 assemblage would do much to advance our understanding of the early part of the Marquesan sequence.
A further comment must be made here on the plain ware ceramic sherds which have been recovered from a few sites, including Hane and Ha'atuatua, and which supposedly provide a direct link between the Marquesas and Western Polynesia. To date, 14 sherds have been found from four Marquesas sites: nine from Ha'atuatua recovered by Suggs (1961) and by Sinoto (1970:113); one from the Ho'oumi site on Nukuhiva (Suggs 1961:56-7); two from Hane; and two from a disturbed - 28 site in Atuona Valley, Hivaoa. (The latter two sherds, the first from the southern Marquesas, were recovered by E. Edwards in 1985, and were sent by Maeva Navarro of the Tahitian Department of Archaeology to the author for detailed analysis.) Although Dickinson and Shutler (1974) claim that three of the Ha'atuatua sherds show affinity with Fijian ceramics, and may therefore be exotic, it seems unlikely that this collection of sherds from four separate sites on three islands can all be attributed to exotic manufacture. It is more plausible to postulate, as Green (1974:256-7) has done, that these sherds are all in secondary contexts, and that “they testify both to the manufacture and use of pottery in a phase before Sinoto's Phase I and to its abandonment by A.D. 300” (emphasis added). Green further suggests that the Ha'atuatua sherds are
. . . either testimony to richer pottery-bearing sites somewhere on Ha'atuatua Bay and ones which may well be associated with Suggs' early dates of first century A.D. or before, or they are perhaps but a small selection of sherds from primary site contexts which have not yet been adequately sampled. In either case, the implication is that secure assemblages from the Settlement Phase of Marquesan prehistory associated with the use and manufacture of pottery are present and await excavation (1974:247).
I am inclined to believe that Green was quite correct in this assertion, and that primary depositional contexts containing significant quantities of ceramics may yet be revealed by further archaeological investigation.
The reanalysis of the early Marquesan material is also relevant to one of the problems with the orthodox scenario, described earlier. Bellwood (1970), Davidson (1976) and others have observed that the “Phase I” material culture described by Sinoto (1970, 1979a) is substantially different from anything known in West Polynesia, posing the question of its immediate antecedent. Yet, when we focus on the assemblage represented solely by Hane Levels VII and VI, along with that from MH-3-11, the problem diminishes; most of the artefact classes are known also from West Polynesian sites. The discovery of a site containing Polynesian Plain Ware ceramics in primary context, as suggested above, would provide the direct link between West Polynesian and East Polynesian assemblages which is currently lacking.
In short, four main points emerge from a reanalysis of the Marquesan data. (1) An initial settlement date of A.D. 300 is unacceptable. The available radiocarbon determinations do not permit a precise estimate, but certainly do not preclude an initial colonisation of the Marquesas in - 29 the late first millennium B.C. (2) The lumping of several complexly-interrelated stratigraphic units and their assemblages from the Hane site into two phases has masked substantial temporal variability. (3) Many of the traits diagnostic of “Archaic East Polynesian Culture” as defined by Sinoto do not appear until the third or fourth depositional units of the Hane sequence. They cannot therefore be properly regarded as aspects of the founding or “archaic” material culture of the Marquesas or of East Polynesia generally. (4) None of the ceramics recovered to date can be considered to be in primary deposition contexts, with the implication that yet earlier deposits containing ceramics are present and will be uncovered with further work.
The Society Islands
Before the advent of subsurface archaeology in Polynesia, the Society Islands were widely regarded as the East Polynesian “Hawaiki”, or homeland, from which more remote islands and archipelagos were settled (Buck 1938). Rich archaeological finds and associated early radiocarbon age determinations in the Marquesas have substantially altered this view. Yet, as Bellwood (1970) noted, the shift to the Marquesas as the primary dispersal centre for Eastern Polynesia may simply reflect the intensity of archaeological investigation, and therefore may be attributed to sampling error.
The earliest known sites in the Society Islands are the burial ground at Maupiti (Site Ma3; Emory and Sinoto 1964, 1965), the extensive, water-logged village site of Vaito'otia-Fa'ahia on Huahine (ScH1-1, -2; Sinoto and McCoy 1975; Sinoto 1979, 1983a, b), and the Vaihi site on Ra'iatea (Semah et al. 1978). As Sinoto notes, all of these sites contain similar material culture assemblages exhibiting all or many of the artefacts diagnostic of “Archaic East Polynesian Culture”. They would therefore appear to be related to the material culture of Level V of the Hane sequence. Site ScH1-1, -2 has the largest series of radiocarbon age determinations, with seven samples ranging in age from A.D. 805 ± 90 to 1180 ± 90. (Sinoto has not published full details for these samples, and it is not indicated whether the reported ages have been corrected for secular variation in C14.) The C14 ages for Maupiti and Vaihi fall within this range. Thus, we may conclude that by at least the 9th century A.D. the Society Islands were occupied by several communities sharing a material culture highly similar to that found in Level V at Hane. Sinoto would have us interpret these assemblages as evidence for a “migration” from the Marquesas to the Societies. (The alternative hypotheses, that these data indicate the diffusion of traits from the Societies to the Mar- - 30 quesas, or that they represent a period of frequent contacts between these two archipelagos as might be expected in a long-distance exchange relationship, seem not to have been considered.) In order to evaluate such an hypothesis, however, we need to consider the geoarchaeological context of Society Islands archaeology, and the problem of sampling error.
It may be extremely important that all three early sites described above were accidental discoveries. Unlike the Marquesas, where early occupation deposits in coastal sand dunes and rockshelters have been readily located during reconnaissance surveys, the Society Islands have proved refractory in terms of locating older sites. Two main geomorphological factors account for this situation: (1) the islands are undergoing gradual tectonic submergence; and (2) there has been substantial coastal aggradation and deposition of alluvial sediments on coastal plains and into former lagoons. Both of these processes are likely to have resulted in site burial, making the discovery of early sites through surface reconnaissance difficult. Evidence of archaeological site submergence was obtained not only at Vaito'otia-Fa'ahia and Vaihi but also at two midden deposits on Mo'orea (ScMf-2, -5; Green et al. 1967:181-2). The Society Islands are also a likely candidate for the burial of early occupation sites under alluvium resulting from human-induced burning and erosion, as argued by Spriggs (1982).
In short, on geomorphological grounds we face a serious sampling problem in Society Islands archaeology. To move beyond accidental discovery of buried sites and seek to answer the question of whether yet earlier assemblages exist in the Societies, a programme of sub-surface sampling surveys must be initiated, using an explicit geomorphological model of site formation processes. Until such a programme is carried out it is inappropriate to construct scenarios for East Polynesian settlement based on the current negative data from the Societies. Polynesian archaeologists should already have learned their lesson about negative site distribution data from the Samoan case; despite the very extensive work of the Samoan Archaeological Project (Green and Davidson 1969, 1974), the entire first one-third of the Samoan sequence remained unknown until the accidental discovery of the submerged Mulifanua Lapita site. The probability that we have yet to discover the first millennium or so of Society Islands prehistory seems to me to be quite plausible.
The Hawaiian Islands
The orthodox scenario proposed two settlement events for the - 31 Hawaiian Islands: an initial colonisation from the Marquesas about A.D. 750, followed by a second settlement from the Society Islands about A.D. 1200. Certainly, there are compelling reasons to accept a two-phase settlement of Hawai'i, particularly the linguistic arguments (Green 1966; Elbert 1982). My concern here will be solely with the dating of the initial colonisation, and with the attribution of source area.
I have elsewhere provided a thorough review of the current evidence for early Hawaiian settlements (Kirch 1985), which therefore does not need reiteration here. I shall merely note that an increasingly large number of C14 age determinations from several sites distributed throughout the archipelago (Fig.3) render a settlement date of A.D. 750 unacceptably late. A reasonably good case can be made for initial colonisation of Hawai'i by about A.D. 300-400, and sampling considerations again dictate that we leave open the possibility of yet earlier deposits.
There are two main assemblages which are presently candidates for the earliest phase of Hawaiian prehistory: (1) the Layer III assemblage from Site 018 on O'ahu Island; and (2) the Layer III assemblage from the Pu'u Ali'i sand dune site (H1) at South Point, Hawai'i Island. For reasons given in full elsewhere (Kirch 1985:82-5) the Groningen C14 age determination of 1660±60 B.P. (corrected range at 95 percent confidence level = A.D. 225-565) from Layer III at H1 can be taken as a reasonable approximation of the deposit's age; the 018 Layer III deposit has been similarly dated to approximately A.D. 323-447. Both of these sites contain similar artefact assemblages characterised by such items as reversed-triangular section adzes, a variety of one-piece fishhooks in pearl shell and bone (including incurved shank types, and double-notched line-lashing devices), porpoise-tooth pendants, a double-perforated trolling lure point (from H1), as well as sea urchin spine and coral abraders, dog's tooth ornaments, flake tools, and other items (Kirch 1985:67-88). Absent from these or any other Hawaiian sites are such supposedly “diagnostic” Archaic East Polynesian types as shaped whale-tooth pendants, lanceolate pearl-shell ornaments, reels, or harpoon heads.
These early Hawaiian assemblages pose a serious problem for the orthodox scenario of East Polynesian settlement, as Sinoto (1981, 1983c) recognises. Not only is the dating of these sites uncomfortably early in terms of Sinoto's Marquesan chronology, but the absence of so many “archaic” artefact types is truly puzzling. However, a revision of the Hane sequence along the lines suggested earlier would quickly alleviate these inconsistencies. (1) If, as suggested, the Marquesas were actually- 32
FIGURE 3.- 33
Radiocarbon age determinations from early sites in the Hawaiian archipelago, corrected for secular variation and plotted at 95 percent confidence intervals (after Kirch 1985).
settled in the later first millennium B.C., then the colonisation of Hawai'i by about A.D. 300-400 is perfectly reasonable. (2) The 018 and H1 assemblages actually compare closely with those of Levels VII and VI at Hane, which also lack the diagnostic shaped whale-tooth ornaments, harpoon heads, reels, and so forth. In short, the initial settlement of Hawai'i could have been from the Marquesas, before the development of the “archaic” material culture represented by Hane Level V.
Before constructing an alternative scenario for Hawaiian settlement, however, it would be wise to remind ourselves that the Society Islands are still a lacuna for early sites. While the Hawaiian Islands are certainly a better voyaging target from the Marquesas than from the Societies (Finney 1967, 1974), it is not impossible that undiscovered early sites in the Societies will yet prove to be a close match to the early Hawaiian material. Also, archaeological deposits on Fanning Island in the Line archipelago have yielded materials (such as trolling gear, one-piece hooks, and porpoise tooth pendants) dating to about A.D. 350-530 (A. Sinoto 1973), which compare closely to the early Hawaiian assemblages. In short, while the Marquesas remain the favoured candidate for the source of the first Polynesian colonisers to Hawai'i, we should not close the books on alternative possibilities.
In the orthodox scenario, the Cook Islands were settled from the Society group, which in fact continues the propositions originally advanced by Buck (1938, 1944). In truth, however, we know extremely little of the subsurface archaeology of the Cooks, the few modern investigations having concentrated on surface survey of marae and other structural features (Trotter 1974; Bellwood 1978). Bellwood's limited excavations at the Ngati Tiare site on Rarotonga and the Ureia site on Aitutaki have yielded the earliest stratified materials, dating to the 10th-13th centuries A.D. There is no reason to believe, however, that these sites represent a colonisation or settlement phase in the southern Cooks. The limited assemblages from these sites contain several artefacts with similarities to the Vaito'otia-Fa'ahia and Hane Layer V assemblages. Interestingly, both sites also yielded untanged, triangular-sectioned adzes of “Samoan” type, the presence of which remains to be adequately explained. In my view, until much more extensive subsurface work has been conducted throughout both the northern and southern Cook Islands, any statements on the timing and direction of initial settlement are premature.- 34
Upon first consideration, the proposal that New Zealand was settled direct from the Society Islands might appear to be one aspect of the orthodox scenario with substantial empirical support. Certainly, the materials recently recovered from the Vaito'otia-Fa'ahia site on Huahine, as well as those from the Maupiti burial site, provide a likely candidate for the type of assemblage which might have been ancestral to that of the New Zealand Archaic. Furthermore, the Huahine C14 ages are consistent with the suggested initial colonisation of New Zealand between about A.D. 800-1000.
As Davidson (1984:23) rightly cautions, however, these artefact parallels “merely show the close similarity of early New Zealand culture with contemporary culture in another part of East Polynesia”, and need not point to direct migration routes. Davidson suggests that the southern Cook Islands, the Austral chain, and even Pitcairn could have played a role in New Zealand colonisation. “It seems quite possible that these more southern islands . . . may have been more important in the settlement of New Zealand than is recognized at present” (1984:24). Given our complete ignorance of early prehistoric materials from the Australs and Pitcairn, as well as from the Cooks, we must simply conclude that it is premature to attempt to draw arrows linking New Zealand directly with the Society Islands, or with any other group for that matter.
Also questionable in the orthodox scenario is the supposed second settlement from the Marquesas (Emory and Sinoto 1965). As noted earlier, this link was proposed solely on the strength of the restricted distribution of bone reels. Recently, however, bone reels were recovered from the Vaito'otia-Fa'ahia site (Sinoto and Han 1981:23), and so the unlikely occurrence of a direct settlement voyage from the Marquesas to New Zealand can now be discarded as an explanation for what, in fact, was nothing more than sampling error.
A relatively early settlement of Easter Island is still most strongly indicated on linguistic grounds (Elbert 1953; Green 1966). (The convoluted and implausible arguments of Langdon and Tryon 1983 are rejected here; cf. Clark 1983; Green 1985.) Despite fairly extensive archaeological survey, early occupation deposits have yet to be discovered or excavated on Easter, so that we still lack a good sample of the island's settlement period material culture. Until such sites are located, arguments regarding Easter's position in East Polynesia will have to be advanced largely on the basis of controlled ethnographic and linguistic - 35 comparisons. Green (1985) neatly summarises the linguistic evidence for the breakup of Proto-Eastern Polynesian into Rapanui and Central Eastern Polynesian, and suggests that this may have occurred by the 4th or 5th century A.D.
The evidence reviewed above is sufficient to reject the orthodox scenario as an accurate or useful model of Eastern Polynesian settlement. Where does this leave us? If the orthodox scenario is to be abandoned, are we in a position to replace it with a more satisfying model? The answer appears to be “not yet”, particularly since we arguably face major sampling problems in the Society Islands, and probably also the Marquesas. Furthermore, we know almost nothing of the early phases of settlement in such archipelagos as the Cooks, Tuamotus, Australs, and Mangareva and are unable to map out a model of “dispersal patterns” that would be any more convincing than the orthodox scenario itself.
Figure 4 is an attempt to graphically collate the present evidence of island occupation sequences based on the preceding analysis. (Not included in the chart are the sequences, largely unknown, for such groups as the Cooks, Australs, Tuamotus and Mangareva.) For the Marquesas, Sinoto's Phase I is replaced by the separate stratigraphic assemblages from Hane, which are seen to span perhaps as much as 800-1000 years. The possibility of yet earlier assemblages, which would be expected to contain significant quantities of Polynesian Plain Ware in primary depositional context, remains open. In the chart, the hatched lines indicate assemblages containing artefact types diagnostic of Sinoto's “Archaic East Polynesian Culture”. As is evident, these assemblages—from the Society Islands, New Zealand and Hane V—consistently date from between A.D. 700-1100 and can be interpreted, not as an “archaic” phenomenon, but as a central Eastern Polynesian cultural complex which developed after the settlement of such remote islands as Easter and Hawai'i.
What seems obvious from a consideration of the evidence at hand, even though painful to admit, is that we have yet to adequately identify or define the earliest settlement phases throughout central Eastern Polynesia. Even in the Marquesas, where the evidence is strongest, it is doubtful that actual colonisation sites have been excavated. For the Societies and Cooks, as well as the Australs, the early phases remain wholly enigmatic.- 36
Time-space chart indicating the current state of knowledge for pre-A.D. 1200 settlement in major Polynesian archipelagos.
At this point, it seems appropriate to raise again the possibility that we should be thinking in terms of an “Eastern Polynesian homeland” as a region, rather than as a single island, or even a single archipelago. The latter error has already been committed in Western Polynesia, where Tonga was proclaimed the Polynesian homeland until a site of equal antiquity (Mulifanua) was discovered on Samoa, and more recently on Niuatoputapu and 'Uvea (and probably Futuna) as well. Thus, as Green (1981) has eloquently argued, a regional concept of adjacent Samoan, Tongan and Lau archipelagos—with an interlinked dialect chain—has come to replace the simplistic idea of a single Polynesian homeland. The orthodox scenario, in which the Marquesas represent the Eastern Polynesian homeland, may yet come to the same fate. Certainly the linguistic evidence does not run counter to the idea that there was a more extensive East Polynesian homeland region, which could well have incorporated the Society and Cook groups in addition to the Marquesas.
The major task which confronts us, however, is to test this and other alternative hypotheses of Eastern Polynesian settlement with the con- - 37 crete evidence of archaeology. If anything is clear, it is that we have been too complacent in building grand scenarios on the data from a handful of widely separated assemblages, leaving great lacunae of time and space uninvestigated. The critical gaps lie particularly in the central archipelagos—the Cooks, Societies, and the Australs. Until the early settlement phases of these islands are known, any attempts at producing a sufficient account of Eastern Polynesian dispersals will be premature. Further, a research strategy to fill these gaps will require some innovative approaches, especially in the Society Islands where the geomorphological problems of site discovery are substantial. I would gamble that the rewards will be worth the effort.
1 After Sinoto (1966).
2 Items diagnostic of “Archaic East Polynesian” as defined by Sinoto.