Volume 111 2002 > Volume 111, No. 2 > Ritual landscape in late pre-contact Rarotonga: A brief reading, by Matthew Campbell, p 147-170
RITUAL LANDSCAPE IN LATE PRE-CONTACT RAROTONGA: A BRIEF READING
Distributional or spatial studies have a long history in archaeology. Archaeologists have often analysed settlement and spatial organisation under headings such as environment, technology and economy. Without invoking an unfashionable environmental determinism, there is no denying that these factors do function as conditions and constraints on settlement. More recently other concerns have come to prominence, under the heading of landscape.
Some have seen landscape studies as part of a historical progression in archaeology from a focus on the isolated artefact, to the site, the settlement pattern, and finally the landscape (Fisher and Thurston 1999:630, Thomas 1993:19)—in other words an ever widening contextualisation. In practice there is a lack of articulation between settlement and landscape that perhaps mirrors their respective identifications with processual and post-processual archaeologies. Landscapes tend to be analysed in terms of the sacred, the ideational or the experiential. Environmental factors rate an occasional mention, but rarely is settlement, in the old fashioned sense employed by Willey (1953), integrated into landscape. Landscapes are essentially cultural constructs arising out of the mutual interaction of environment and society, each continually forming and reforming the other. Landscape is a signifier of social mores, identity and history, but at the same time is constrained by the environment. 1
All this seems to be getting away from the idea of landscape as a distributional or spatial archaeology, but it is just this sort of analysis that I want to use to present a model of the use of ritual space on the island of Rarotonga. To do this, I use a GIS analysis in conjunction with the evidence of oral tradition to examine the function of the Ara Metua, a road that circled the island, and the function of marae. I then widen the scope to examine the role of marae in society, and how spatial conceptions in Rarotongan culture helped to define the landscape.
These spatial conceptions have little to do with environment and economy. The spatial, and to some degree temporal, scales of landscapes of this kind are indeterminate—spatially all scales tend to operate together, while the temporal scale depends on a mythical past to inform the (pre-contact) present. In these ways landscape is very unlike classic settlement pattern study, where space and time are bounded. These different spatial and temporal scales are - 148 significant factors in the lack of articulation between settlement and landscape. I do not intend to provide any resolution for this problem here, merely to point out why the distributional archaeology I present may not engage with other distributional archaeologies of Rarotonga.
Rarotonga is a typical Polynesian high island. At roughly 11 x 6km, with a maximum elevation of 653m, its topography is characterised by deeply incised valleys surrounded by a continuous coastal plain that is generally about 1km wide. Surrounding this is a fringing reef enclosing a shallow lagoon of up to 1km in width. The tapere system of land holding develops out of this concentric resource pattern. Tapere are radial land units, each centred on a valley and containing mountain, coastal plain, lagoon and reef resources. Each tapere was governed by one or more chiefly mata 'iapo, who was the (usually) senior, (usually) male member of the ngāti, or local descent group. The matakeinanga, the corporate land-holding community group, resident within the tapere, was the central political unit, subsuming the kin-based ngāti.
Rarotonga, as an island and a fairly small one at that, is spatially constrained to an unusual degree, so that environmental and economic factors play a strong role in settlement. Habitable and agricultural space are largely limited to the coastal plain and valley floors, and a pre-contact population of perhaps 8000 tended to live in a dispersed settlement pattern alongside their agricultural holdings, while remaining within their tapere (Campbell 2001). But the tapere system is as much culturally constructed as it is environmentally conditioned.
RAROTONGAN ORIGINS IN TRADITION
Around A.D.1250 (dating from genealogies) two voyaging canoes arrived on Rarotonga together. One was the Takitumu from Tahiti, the canoe of Tangi'ia Nui who was fleeing from his elder brother Tutapu. While at sea he met a canoe from Samoa captained by Karika. Tangi'ia and Karika joined forces and sailed to Rarotonga, where Tutapu caught up with his brother, but was slain.
When Tangi'ia landed at Ngatangiia harbour on the east coast he established a marae, Te Miromiro, and placed Parau-ā-To'i as purapura or guardian over it. He then made a counter-clockwise circuit of the island establishing as many as 47 marae or similar structures and placing custodians in charge of each (Tara 'Are 2000:155). These guardians were the ancestors of the chiefly and priestly lines. Tangi'ia's circuit divided up the physical landscape and established the marae system as well as the political and land tenure system based on the tapere.- 149
The Ara Metua provides a fine example of the interplay in landscape between environment and culture. This road, that in pre-contact times encircled the island along the coastal plain, and along which many marae are located, is by far the largest archaeological site on Rarotonga. It was probably paved with basalt and coral for much of its length and kerbed where habitation was most dense (Hiroa 1927:211). Parker (1974:64) suggests that it was built to link the lowland marae in a ceremonial route. The road in its current state is discontinuous, and the pre-contact road is almost completely destroyed or buried by modern roading, but the latter largely traces the course of the old road and it seems safe to assume that the physical relationships of archaeologically recorded marae to the road remain as they were in pre-contact times.
According to Rarotonga tradition, the Ara Metua was constructed by Toi, a pre-Tangi'ia ancestor (its alternative name is Te Ara Nui o Toi ‘The Great Road of Toi’) (Crocombe 1964:8), but the road as it was known in the early contact period must have been of later construction. No radiocarbon dates are known for it, but there is a date of 1530 for marae sites associated with it (Trotter 1974:146). There is no reason to think the road is not of the same antiquity as the marae.
The Ara Metua is one of the largest sites in Polynesia, and although small sections of paved road are known from elsewhere, it is also unique in Polynesia in terms of its size and elaboration. So what was the road? What did it represent, and why was it built?
We can account for its form and location by a simple environmental fact—the topography of Rarotonga constrains the road to follow the coastal plain that encircles the island. The Atlas of the South Pacific (1986) indicates that Rarotonga is the only Polynesian high island with anything like an almost continuous coastal plain (the basaltic lava flow at Black Rock—now quarried—was not very high), and so is probably the only island that could have carried such a road.
As for its function, like any road it had more than one, but it is first and foremost a road in the most pragmatic sense—an infrastructural element that facilitated the movement of people, goods and information. Its primary day-to-day function was economic. It traversed every tapere, and so may have fostered a sense of island-wide unity. The road was located away from the exposed beach ridges and close to productive lands. Bellwood (1971:149) describes it as being built along an ecotone, minimising travel to inland and coastal resources, but distance on a small island would not have been a major settlement factor. Even so it was the focus of lowland settlement. The earliest European observers of the island, from the London Missionary - 150 Society, recorded how agriculture and settlement were closely related with the road
Almost every individual having his Kaina or small farm cultivated with plantains, ti, taro, yams etc., so that the whole settlement appeared one extensive garden…. The road is a tolerable good one in most Places and shaded from the sun by the branches of the spreading trees. The land on each side of the road was cultivated all the way, and on many little farms a house was standing for the accommodation of the owner when he comes to look after his land, food, &c…. The houses of the people are on each side of the road surrounded with little gardens in which various kinds of vegetables were growing (Williams and Barff 1830).
There is a good road around the island, which the natives call the ara medua, or parent path, both sides of which are lined with bananas and mountain plantains…. The houses of the inhabitants were situated from ten to thirty yards or more from this pathway (Williams 1837:205).
The Rarotongan settlement pattern largely follows the radial pattern of the tapere, but spills along the coastal plain to occupy the same zone as, and become associated with, the Ara Metua. Both road and settlement are constrained by environment, and coincide by convenience—they will tend to follow each other. Although it is hardly possible to say which came first—settlement, the road or the marae linked by it—the establishment of such a large infrastructural element influences the future location of settlement so that the road, in the long term, does indeed become the focus of settlement.
MANA AND TAPU
However these pragmatic functions do not fully explain this enigmatic structure. A simple unpaved track could have served all these purposes, and it is only by looking at the ritual functions of the road that the reasons for its elaboration become clearer. In Polynesian, ritual is closely entangled with the concepts of mana and tapu. While expressions of mana and tapu differed throughout the region, and little is explicitly known about the tapu system of Rarotonga, some common features are clear. Tapu is a quality inherent in all people to greater or lesser degrees, depending upon their status. Unregulated contact between tapu and non-tapu resulted in pollution of the tapu and placed the polluter in great social and/or supernatural danger (Bowden 1979:53). Shore (1989:154) explains tapu as the ability to focus and channel the supernatural forces of mana. Mana had to be wrested from the gods and properly directed if it was to be efficacious and enlarge rather than diminish human life. In its active capacity, tapu was the potency of a person, place or thing, in its negative capacity, when the potency was not regulated, it denoted great danger and the forbidden.- 151
Of the rituals that directed and controlled tapu, rituals of binding are universal. On Rarotonga the missionary Charles Pitman (1833 [II]:207) describes how the carved wooden gods were wrapped in rolls of cloth:
When the god was displeased the prophets would open the door of their sacred places (Maraes), & sweep away the dust, cobwebs &c. from the floor where their god was placed & from their deity also. The prophet would then take off his robes (immense rolls of native cloth) & carefully examine it as he unfolded it. They often found the excrements of rats, their nests &c. in it, & large holes eaten by these sacrilegious intruders, which when discovered the prophet would inform the people of the cause of the anger of their deity, & give orders for fresh cloth to be made, & a new kiikii to be adzed out, as the only means of appetizing the anger of their offended god.
Taking the rolls of cloth off the gods must have been an anxious, even dangerous event for pre-Christian Rarotongans. This was a tapu removing ritual that de-sanctified the old god image (ki'iki'i), requiring the carving of a new one.
At the installation of a chief, his mana was bound up by his maro ‘loin cloth’ that concealed his genitals, the source of his generative potency. Such a ceremony is described in the Minute Books of the Cook Island Land Courts. It was performed by the Potikitaua (a priestly title) at the installation of Tinomana Mereana (incidentally a female ariki) in 1881 (Anautoa 1906 [M.B.II]:318, Taraare 1906 [M.B.II]:322). In the same case, the Potikitaua is described winding cloth around the newly built house of Makea Ariki (Urataua 1906 [M.B.II]:320). For New Zealand Māori, this latter ritual, although similar to sanctifying tapu binding ceremonies, was a de-sanctifying ritual, separating the house from divine influence and making it fit for human habitation (Smith 1974:40). It seems to have been the same on Rarotonga.
Other important related rituals are the ritual circuits of chiefs and gods. The circuits of Lono and Kū around Hawai'i have been made famous by Sahlins' (1981) analysis of the death of Captain Cook. The Rarotongan analogue is the circuit made by Tangi'ia when he first came to the island. Not only did this circuit establish the political system, it both bound the tapu, controlling and directing the mana of the gods for human use, and, like the house-binding ritual, made the island safe for human habitation. Tangi'ia's route is permanently inscribed on the landscape in the Ara Metua. The ceremonial route linking the marae replicates the ceremonial circuit of Tangi'ia that established them, so that the Ara Metua makes the whole island a ritually sanctified landscape. The worship of the great pan-Polynesian gods took place here, where the ariki continued the tapu-controlling work of their predecessor, Tangi'ia. The Ara Metua and its associated marae are - 152 the physical aspects of a coherent pan-island ritual system, and were integral to the process of mediation between humans and the divine.
The road is not a physically imposing monument like the heiau of Hawai'i, not so overt an advertisement of power, but by virtue of its size and the labour involved in building it, it undoubtedly falls into the category of “monumental construction”. Monumental structures may arise from consensus between elites and the commoner labour force, or alternatively elite domination and the suppression of dissent (Hodder 1994:538). This begs the questions, what were the socio-political parameters that led to its construction, and what were the socio-political consequences of its being there? Does it represent political unity and consensus, or the imposition of hegemonic rule? These questions are not easily answered either from the archaeological record or from traditional histories as they are currently understood.
In 1903, the New Zealand colonial administration established a system of land courts in the Cook Islands. The oral traditions given as evidence in the courts and recorded in the minute books are an important source of ethnohistoric information for Rarotonga that seems to be reliable to at least 150 years before European contact (Campbell n.d.). From the time of Tangi'ia up until this time, there is no direct evidence either of political unity or disunity, hegemony or independence, although the establishment of the island encircling ceremonial route indicates a unified ritual system. Although mata 'iapo and matakeinanga may have been more or less independent, they were so within an island-wide system that was unified in its operation, and independence was defined in relation to other matakeinanga, rather than in opposition to them. The road cut across tapere boundaries and would have reinforced these tendencies. It also represented, and was an integral part of, a unified ritual system. Whether this political and ritual system was practiced under a dominant paramount or independent mata 'iapo would have made little difference to its operation.
The form, function and location of the road can usefully be analysed, in both settlement and landscape, with respect to at least three concepts—ritual, economy and environment. These are not either/or concepts. The road's role as marker of an ecotone and a ceremonial route are both central to the analysis of settlement and landscape presented here. The road is a source of fascination and could easily be over-analysed, which I hope I have not done, but this would be preferable to downplaying its importance.- 153
Figure 1 shows the Ara Metua and all archaeologically recorded marae. Those that Tara 'Are (2000) records as being established by Tangi'ia are highlighted. Only nine of the 47 marae established by him have so far been recorded, but the first quarter of his circuit can be traced. Not all marae, nor even all marae along the Ara Metua, were founded by Tangi'ia, and not all marae founded by him are adjacent to the road. Tara 'Are (2000:155) describes the establishment of Ārai to Tonga and “after this [Tangi'ia and his companions] went inland and made another marae and called it Paepae-tua-iva; dedicated to the god Tonga-iti, while Ta'I-vānanga was made the guardian. Again they went seaward and built a marae named Marae-koroa.” This inland:seaward movement mirrors an important spatial contrast throughout island Polynesia. Indeed landward:seaward is the basis of spatial reference and directionality in most Oceanic languages (Palmer in press). The seaward direction is also associated with high status on Rarotonga (Baltaxe 1975:82). The coast is associated with tapu, horizons and things foreign. The inland, on the other hand, is a lower status realm, associated with the non-tapu and the domestic, so that the seaward:inland (tai:uta) contrast is a status contrast.
Figure 1. The Ara Metua and recorded marae on Rarotonga, with those established by Tangi'ia highlighted. Numbers refer to their place in the sequence as given by Tara 'Are (2000).
Yamaguchi (2000:140) has noted a similar contrast between mountain marae and inland marae on Rarotonga. The coastal marae Te Mareva (RAR12) on Motutapu Island is constructed of basalt brought from the mountains, whereas the mountain marae Piako (RAR105) contains blocks of coral brought up from the coast. Bradley (2000:84) notes that raw materials may be identified with their point of origin and act as a message or mnemonic. This seems to be the case here. “The mountain marae is complete only with the additional elements of the ocean, and the ocean marae whole only when it contains an element of the mainland (or the inland). It is therefore likely that the mountain and the ocean formed an inter-complementary as well as dichotomous set in the cosmological landscape” (Yamaguchi 2000:149). Yamaguchi noted that Piako is constructed with clear views of a number of mountain peaks as well as a narrow view of the ocean, and Duff (1974:24) also noted the alignment of marae in Tupapa Valley with mountain peaks.
Figure 2 shows the viewsheds 2 of three marae in the Tupapa Valley traditionally established by Tangi'ia. Only two of the marae, Koroa and Ārai to Tonga, have sea views, and in fact the viewshed of Paepae Tuaiva is considerably more restricted. However all three retain views of four major peaks—Ikurangi, Te Manga, Te Atukura and Oroenga. In fact, as Figure 2 shows, the actual area of valley floor and lowland from which all four of the peaks can be seen is restricted to the valley mouth and coastal plain to the north, and the area around Paepae Tuaiva, indicating that the view to mountain peaks was one of the factors influencing where marae were sited. Moving east around the coastal plain, Murivai is again sited with a view to the same four peaks. Further south, Anga Takurua and Ta'u mā Keva can only see three of the four (though Anga views other peaks). Neither of these marae can view Te Manga because they are located in the shadow of Oroenga—the two peaks and the two marae describe a direct line. Even though they do not view Te Manga (‘The Mountain’ that is the highest point on the island), it seems the mountain is still an influential factor in their location. This analysis reinforces the notion that connections with mountain peaks, associated with the heavens, were an important factor in marae construction. 3
Figure 4 contrasts the individual viewsheds of three marae, one a mountain marae (RAR105: Marae Piako, a marae of Pa Ariki), one a coastal marae (RAR24: Marae Pūkuru Va'anui, established by Tangi'ia, and the kōutu of Pa Ariki) and a valley marae (RAR157). The orientation of the viewsheds associated with each location is clear. Pūkuru Va'anui is part of the public ritual circuit, physically and ritually related to the Ara Metua. Like the marae in Tupapa Valley it has extensive views of the sea as well as to prominent mountain peaks. Piako, located far up the Turangi Valley, has- 155
Figure 2. Combined viewsheds of three marae in the Tupapa Valley traditionally established by Tangi'ia. The area from which the four peaks (Ikurangi, Te Manga, Te Atukura and Oroenga) can be seen has been simplified for ease of display.
Figure 3. View of Te Manga from Ārai to Tonga. Te Atukura is the smaller twin peak to the left of and behind Te Manga.
Figure 4. Contrasting viewsheds of two marae associated with the Pa Ariki title—RAR105: Marae Piako in the mountains, and RAR24: Marae Pūkuru Va'anui on the coast—and also RAR157: a marae deep within a small valley at Kiikii.
been located just high enough on a ridge spur to have a narrow view of the ocean as well as mountain views, as described by Yamaguchi (2000:149). This marae was not part of the public ritual of the Ara Metua, but would have been associated with the private community ritual of Pa Ariki and the community of Turangi Tapere. This community ritual would have intersected with the public ritual at Pūkuru Va'anui. Piako has clear views of almost the entire head of the valley/tapere, so a defensive or lookout role for this site is also a possibility. There is little overlap between the viewsheds of Pūkuru Va'anui and Piako, but they are complementary; between them they view most of the valley.
RAR 157 has a view of neither peaks nor sea. Domestic sites further down Kiikii Valley have good views of Ikurangi and Oroenga at least, though not of the sea. There is in fact a low hill at the mouth of the valley that bends the Kiikii Stream sharply to the west as it enters the coastal plain. This hill is less than 20m high and so does not show up on any contour map, with the result that it does not form part of the viewshed analysis. Even so it still excludes a sea view for the entire valley floor. Although the viewshed of RAR157 is limited, it is still a fairly elaborate structure (Campbell 2000:53). - 157 Ritual here would have been largely divorced from the public ritual circuit. It would have been private, associated with the community.
The Land Court records mention a marae called Toronae in Kiikii Valley, which I have suggested is probably the same as RAR157 (Campbell 2000:51). Pakitoa (1908 [M.B.V]:20) tells the court that “my ancestor Taumatatau was at Kiikii, Toronae was his marae”. Uritaua (1912 [M.B.V]:141) gives a little more historical detail:
The fight was at Kiikii…. The tribe were in the valley at Toronae and did not join in the fight. Kapo their chief warrior stood in the entrance of the valley and the tribe were behind him. Kapo would not let them join in the fight. But when he heard Rupe was killed he let them go out of the valley.
The description implies that Toronae was located in a defensible position in the valley and was easily sealed from the coastal plain. RAR 157 matches these requirements. It is in an almost hidden location, difficult to find and very private. I would hesitate to ascribe my own, late 20th century, subjective experiences of the site to pre-contact Rarotongans, but my own observation of its physical location, the viewshed analysis and the traditional history of the Land Court description all reinforce each other.
This is not to imply that there is a clear-cut equation between the founding of a marae by Tangi'ia, its associated status and its viewshed. On the one hand, Ta'u mā Keva, a marae of Tangi'ia's, views only three peaks while marae like Te Mareva and Vaerota view seven and eight respectively. This is because they are located on the seaward edge of the coastal plain, in Te Mareva's case on a sand cay in the lagoon, and so their views are less impeded. On the other hand, Ta'u mā Keva seems to have been precisely positioned so as not to see Te Manga. What is of importance is not how many peaks can be viewed from any marae, but which peaks.
It is at this point that the limitations of the analysis become apparent. To begin with, neither this analysis nor the oral traditions are able to ascribe greater or lesser status, or stronger or weaker community associations, to any mountain peaks. The GIS is unable to predict which peaks will be important, and yet the analysis has employed the tools of predictive modelling. It will be clear enough that I have used these tools to explain why marae are located where they are, rather than to predict where they might be found. A predictive model would indicate that marae would be found in the blank space in Figure 2, southeast of Anga Takurua. There are two possible reasons why no marae are recorded here. One, because the data set, already small, is incomplete. As Figure 1 indicates, as many as five of Tangi'ia's marae between Pukuru Va'anui and Ta'u mā Keva (numbers 8–11 and 13) have yet to be recorded, so perhaps there are marae in the - 158 blank space. The other reason is that views to mountain peaks are only one of many factors influencing the location of marae. Many of these other factors will be entirely contingent, and so cannot be modelled predictively.
Another limitation, and a common limitation of landscape studies in general, is that there is no notion of the processes by which the landscape has been created and changed through time. The analysis is heavily dependent on the works of Tara 'Are (2000), written in the 1860s, and the records of the Land Courts, recorded from 1903 on. While these clearly and reliably refer to the late pre-contact period, they are nonetheless limited to that time. Tangi'ia's story is a justification of the political system and social self-image of the time, set in quasi-historical terms. Marae, the Ara Metua and the ritual system associated with them evolved together, as did the story of Tangi'ia that legitimises the system, but the story itself is timeless (at least as far as Western historical notions of time are concerned), and this evolutionary process is invisible.
THE ROLE OF MARAE
One implication of this discussion is that different marae had different roles in the community and in public ritual. The Proto Polynesian term *malaqe is glossed in Biggs and Clark (n.d.) as ‘open, cleared space used as a meeting-place or ceremonial place’. Although the marae concept and its architectural expression have been elaborated differently throughout Eastern Polynesia, the meeting/ceremonial concept remains central. On Rarotonga they are characterised by their diversity (Campbell 2000:64), but not all societies exhibit variation in marae morphology. In the Cook Islands, Yamaguchi (2000:178) demonstrates that Mangaian marae are very homogenous in their morphology and topographical location, correlated with the highly unified and formalised political and religious system on the island in late prehistory. Yet, marae on Tongareva, in the Northern Cook group, are also homogenous in their morphology, even though the atoll lacked a unified political system. Tongarevan marae functioned as territorial markers, and so they shared a uniform system of territorial signs expressed in a uniform morphology (Yamaguchi 2000:224). Rarotongan marae were more cosmological in nature, and Yamaguchi (2000:225) proposes that the variation in morphology reflects variation in ritual space. The degree and nature of variation in marae morphology in different societies depends largely on contingent factors.
Although Mangaian marae lack significant variation and are probably culturally conservative in form, retaining many older features (Kirch and Green 2001:251), the activities associated with individual marae were nonetheless varied. In fact some were “entirely secular” (Hiroa 1934:174). - 159 As Endicott (2000:124) points out, no pre-contact ritual would or could have been entirely secular, but variation and specialisation in the activities that took place at marae was indeed the case. Some marae were the domain of individual descent or corporate groups, while others were used for ceremonies affecting the entire island. Other activities listed by Hiroa include places where chiefs were installed, meeting places for chiefs and places where ceremonial drums were beaten.
This was also the case for Rarotonga. Most tapere contain more than one marae, and in many of the larger tapere these marae seem to have been distributed through similar sets of topographical locations, that is to say on beach ridges, along the Ara Metua, in valleys and far inland (Yamaguchi 2000:138). This indicates an extension of the mountain marae:ocean marae, or tai:uta, dichotomy, with a complete ritual system within the tapere requiring marae with complementary functions, indicated by their complementary topographical locations. One of these marae would have functioned as the matakeinanga share in the public ritual of the Ara Metua. This complementary functionality also represents an articulation between cosmological and political ritual. There was no separation of church and state in pre-gospel Polynesia, and all marae, the Ara Metua and the ritual associated with them would have had political as well as religious meaning (Yamaguchi 2000:163).
One particular kind of marae with a specific function was the kōutu. Kōutu were defined by Savage (1980:119) as “the seat or the royal court of a reigning ariki or high chief…. [Where] the ariki usually, or mainly, resided… certain koutus had one or more maraes formed or laid out within its confines”. This definition indicates a separation of site type between kōutu and marae, although archaeologists have often referred to kōutu as marae (Campbell 2000:50). The Land Court records add little to Savage's description, but they do confirm that kōutu were a special type of marae associated with particular rituals of power and chiefly consecration.
Pukuru Vaanui was the court of royalty (koutu) of Pa and Kainuku, a very sacred place. Maroariki means the binding on of the sacred girdle. The children of arikis when born were all placed in this koutu…. When an ariki is appointed he is taken to the koutu and there installed…. Each mataiapo has his koutu where he was installed (Uritaua 1912 [M.B.V]:135).
Pouara is my tapere, that made me a mataiapo. I am a mataiapo there. My koutu is Areporia. I made it! There are two big maraes in Pouara, yours and mine. Te Aia made his, you made yours (Te Aia 1907 [M.B.III]:235).- 160
Other examples from the records also show that different marae served specific purposes (though doubtless any marae was associated with more than just one particular activity). Toronae has already been discussed, and Vaerota is another example:
They agreed to act and fought Tamaariki and his people who were driven out and from that time he had no power. His power went to the mataiapos of Takitimu. And the two arikis. They all returned home and met at Vaerota and it was decided to hold this land as conquerors. So they went and seized it each from Teaio to Maoate. Each had his own piece (Te Rei 1904 [M.B.I]:115).
Vaerota was evidently a meeting place of chiefs, and Maretu (writing in the 1880s of the pre-Christian Rarotonga of his youth) associates Vaerota with cannibalistic rituals in time of war (Maretu 1985:42). He also refers to Vaerota as a kōutu (Maretu 1985:53).
Examples like this abound in the Land Court records but it would be pointless to multiply them when there is no real evidence of how these different functions and different marae were tied together into systems of ritual.
The opposition coast:mountain is of a different kind to the tai:uta opposition. It is not one of status since both mountain and coast are signifiers of tapu and status. The opposition, from this analysis, represents at least in part a contrast between public and private (or corporate). Mountain marae are located deep within the tapere, away from the public ceremonial route of the Ara Metua and so are the focus of community, rather than pan-island, ritual. There is a movement from the coast, which was associated with tapu and status, but public, to the low-status zone of valley floors with their domestic associations, back to a high-status zone in the mountains, associated with tapu and the heavens, and with private ritual. That the three spaces may be conceptually linked is indicated by the use of both local and imported (from elsewhere on the island) materials, that is basalt and coral. It is notable that, in contrast to the usual archaeological definition of a tapere as being centred on a valley—an agricultural and settlement zone—the definition invariably given in the Land Courts was “from the mountain to the sea”—from one high status zone to another.
Hanson and Hanson (1983:21) note the gender basis for structural oppositions in New Zealand Maori culture, a situation common throughout East Polynesia. Male, tapu, active is opposed to female, non-tapu, passive. These associations can get very complex, and will only be touched on here. The male/seaward:female/inland opposition, and the male/mountain - 161 (heavens):female/valley (earth) oppositions seem clear enough, but the gender associations of the mountain:coast opposition remain obscure. These relationships are diagrammed in Figure 5. 4 This diagram implies that mountains and coast represent two related structural transformations of domestic space (cf. Leach 1996:40). Less directly it also implies that the coastal plain requires teasing apart into a domestic settlement zone and the high-status zone of road, marae and horizon. Since lowland zones of settlement and ritual overlap in physical space, this teasing apart perhaps only takes place in conceptual or social space. I do not intend to explore this notion beyond presenting it visually—for one thing structural analysis is almost as unfashionable as environmental determinism, but mostly because I suspect that any further speculation at this stage would yield only meaningless complexity.
POINT FIELD SPACE
Binary structural oppositions are not the anthropologists only tool for examining the Polynesian conception of space. The usual binary spatial opposition given is the tai:uta one, but it will be clear that seaward or landward are entirely dependent on where the observer stands. From Marae Piako, far inland, virtually all is tai. This is a slightly trivial example perhaps, but consider that the marae along the Ara Metua will be oriented to the road. Marae inland of the road will be oriented seaward, and vice versa. A - 162 simple binary tai:uta opposition is not sufficient to account for the complexities that arise from even a simple application of the concept to the archaeological evidence.
Lehman and Herdrich (n.d.) propose that underlying this familiar spatial conception is a “point field model”. This model is opposed to the “container” model, where space is seen as bounded and any point is defined relative to boundaries. In the point field model fields of space extend outward like rays from a point. Boundaries are defined in relation to points, and occur in the spaces where fields overlap.
This conception of space would seem to apply throughout Polynesia, although Lehman and Herdrich have only systematically worked it out for Samoa. Gill (1876:20), for instance, describes the Mangaian conception of creation as emanating from a primordial point. The point field model is strongly reflected in Samoan language, and this is paralleled throughout Polynesian languages. In Samoan the word mata means both ‘eyes’ and a ‘point’. The dictionaries are ambivalent regarding the second meaning in Rarotongan (Buse and Taringa 1995, Savage 1980) but an examination of the usage of mata as a prefix confirms it. The name of tapere of Matavera translates literally as ‘Hot eyes’, but is better considered as a hot (or potent) point, or source. 5 A matavai is a the dam across the stream that directs the water to the taro terraces, that is, the head or source of the water. Seen in this light, when Akanoa (1907 [M.B.III]:251) says: “Taiaruru is the marae and belongs to Akanoa. Kuruai is ours. It is the head of Akanoa's land”, the concept of the head of the land takes on new meaning. The land section Kuruai and Marae Taiaruru are the point, the potent source, from which Akanoa;'s place, both in society and on the landscape, flows. This potency is the mana bound and controlled by the tapu of Akanoa and Taiaruru. The greater the tapu and mana, the greater the point field space, or territory that falls under Akanoa's mana.
In Samoa boundaries, social and cadastral, are always negotiable and subject to change. Even today boundaries are often overlapping and subject to constant dispute (Lehman and Herdrich n.d.), and it is always anticipated that they will eventually shift (Herdrich and Clark n.d.). The boundaries between point fields change as the power relations between the relevant points change. On Rarotonga there is less evidence of this. Certainly boundaries were subject to occasional dispute in the Land Court, but it is not really clear to what degree this represents the contending mana of the disputants. The court employed its own surveyor who imposed a “container” model on Rarotongan land holding, and boundaries are rarely mentioned after a few early cases (Campbell 2001:155). They seem to have been regarded as ancient, even if contending litigants claimed different ancient - 163 boundaries. The reason for this may lie in the tapere system. 6 Tapere boundaries follow the ridge-lines between major valley systems, by nature immovable, and smaller sections are similarly often centred on tributary valleys. These boundaries may have been seen as essentially set. Williams (1837:204) noted the “rows of superb chestnut trees (inocarpus) planted at equal distances, and stretching from the mountain's base to the sea”, which marked permanent boundaries on the coastal plain. It is unlikely that boundaries had become less mutable during the missionary period, or because of the influence of the Court, which was always limited. The Minute Books lack evidence for the point field model strongly affecting land tenure, and this may have been the Rarotongan norm.
It may be of limited relevance to Rarotongan conceptions of physical space, but the point field model continues to work as a conception of Rarotongan social space. Social structure in pre-contact Rarotonga was notably fluid and variable (Campbell n.d.) and this can be explained as a consequence of changing social boundaries. Mana can be seen as a point field concept and as the mana of contending individuals or communities waxes and wanes, so the social and status boundaries between them change. If this is the case, it begs the question of just how limited the model as a conception of physical space is for Rarotonga. If landscapes can be usefully analysed with reference to mana and tapu, point field concepts, then the relative immutability of boundaries on Rarotonga is a contingent exception to a probably general Polynesian concept.
The landscape is a product of environmental and cultural factors, of the adaptation of one to the other. Settlement is constrained by environment, but humans also modify their environment. The coastal zone of public ritual along the Ara Metua coincides with the highly productive zone of lowland settlement. Inland of this zone are lower status, private, community ritual sites, occupying the zone of inland settlement and production. Further inland still are the high status mountain marae. Thus a concentric zonation of soils and productive resources (which includes the outermost zones of ocean, reef and lagoon) is matched by a concentric zonation of settlement and by a concentric status zonation of marae. This is hardly likely to be a coincidence. Overlying this is the radial pattern of the tapere. That the two spatial patterns together are a diagram of tapu binding and controlling the radiating emanations of mana is possibly no coincidence either. Certainly radiating valleys and concentric ecological zones existed on Rarotonga before humans did, but landscapes are found as well as formed.- 164
I have presented a model of the ritual use of space on Rarotonga and sought insights that can be incorporated into a more comprehensive model of spatial organisation. Settlement and landscape are both spatial approaches to archaeological interpretation, but there the similarity seems to end. Their articulation is not as straightforward a matter as some have proposed. The landscape analysis undertaken here has not grown naturally out of a settlement analysis, and has not provided a wider context for it. They deal with the same things, but they approach those things from quite different viewpoints. It is not my intention here to attempt a solution to this knotty theoretical and methodological problem, merely to point out that we are not currently able to integrate the ritual and conceptual aspects of landscape into a settlement pattern study. In this respect I have not done what I set out to do, but perhaps this brings the problem into clearer focus.
Monuments like marae and the Ara Metua are extensions of, or links to, the sacred landscape. Mountains and marae are identified one with the other. This sort of identification is at the heart of the landscape concept, but this will not account for all identifications of either mountains or marae. Multiple readings of the same sign, be it architectural monument, landscape element or the relationships between them, are possible, preferable, even unavoidable. Such readings would, in pre-contact times, have been complexly interrelated in ways that gave rise to further readings (Bradley 2000:145). Most of this is not now recoverable, hence the subtitle of this paper.
The landscapes presented here have benefited from access to the records of the Land Court and the published account of Tangi'ia's story. Landscape is an arena for political and social action. The formation and perception of landscape is a continual process, often carried out in contests over rights and access to land and resources (both physical and conceptual) (Snead and Preucel 1999:173). Thus the Land Courts are part of this process of recreation, or creating anew the landscape, reconfiguring a historically constituted landscape in a time of great social change in order to deal with the outcomes of that change. The process, in this instance, was mediated by an external power, the imposition of court rulings. The Land Court can be seen as an administrative attempt to appropriate the process of landscape formation. In order to effectively administer the land, it had to suppress the Rarotongans' capacity to speak for and about their own landscape. Its failure to achieve its aims stems from its inability to appropriate the discourse of landscape creation. It had some limited success, for instance, in imposing a European cadastre on Rarotongan land holding, but this was largely cosmetic. Rarotongans maintained their own discourse, which constitutes the basis of the ethnographic and oral historic material of the Minute Books. Equally, - 165 the writing (in the 1860s) and especially the eventual publication (from 1918) of Tara 'Are's version of the Tangi'ia story is another, though more subtle, example of landscape recreation. Tara 'Are's version has become the received version. Though others exist both in print (e.g., Maretu 1985, Te Aia 1893, Williams 1837) and in the Land Court records (e.g., Te Pa 1905 [M.B.II]:53), Tara 'Are's version has gained the weight of authority. It constrains the mythic worldview of modern Rarotongans as much as it constrains my analysis.
The Rarotongan worldview was inscribed on the landscape, reinforcing social continuity, identity and status. All landscapes, by definition, incorporate the worldview of their creators/inhabitors in this fashion, but this analysis has benefited form the assistance of the ethnohistoric record, without which it could not have proceeded as it did. On its own a landscape/viewshed analysis could not predict the status of marae or which lowland marae were established by Tangi'ia, though it would still expose some general patterns. Another advantage of using oral tradition is that it gives direction, suggesting lines of inquiry that might not otherwise be obvious, such as the Tangi'ia story making more explicit the tapu binding function of the Ara Metua. Landscape analysis benefits from the use of ethnohistory not just in specific instances but also in general methodological terms. Landscape seems to provide a fertile meeting ground for archaeology and ethnohistory. I believe these sorts of archaeological analyses (explorations might be an equally applicable term) will become increasingly important and appropriate in Oceania, where the ethnohistoric record is so rich.
This paper has developed out of my doctoral research at the Archaeology Department, University of Sydney. Thanks are due to my supervisors, Peter White and Ian Johnson. Robin Torrence, Roland Fletcher and Tracey Ireland are just the first three of many who discussed with me some of the ideas that ended up in this paper at various times over various cups of coffee and glasses of red wine. On Rarotonga Ngatuaine Maui and the late Kauraka Kauraka in particular have earned my gratitude for enabling my work. Dave Herdrich and Bill Palmer generously supplied me with copies of their unpublished manuscripts. Jacqui Craig and Simon Holdaway have read and commented on earlier drafts of this paper.
- 166 Page of endnotes- 167
Note: References to evidence from the Rarotongan Land Court Records take the form: Witness, year. Land section name(s) and number(s). [Minute book number]: pages of case (may be discontinuous).
1 The term landscape derives from Western art traditions, and ultimately implies a proprietorial oversight of the land, an abstraction of nature from culture (or vice versa) and the resulting cultural representation of nature (Lemaire 1997, Olwig 1993). In non-Western societies this opposition of nature and culture may not exist, so that the social scientist's concepts are not directly relevant to the societies under study. It seems impossible to escape the use of visual metaphors in describing landscape, even when talking about the not visible. The same vision that gazes on a painted or painterly landscape is the possessive vision that gazes on imperial and colonial domains—the Enlightenment or Western gaze (Bender 1999) that revealed the South Pacific to the oversight of missionaries and colonial administrators. The scholarly gaze on its subject is a descendant of that vision. Salmond (1982) points out that these metaphors arise out of the Western system and conception of knowledge, and that other knowledge systems use different metaphors. The visual metaphor must be used with care, and any attempt to use the landscape concept to retrieve some putative non-Western nature/culture unity (rather than use it as an analytical tool within the Western tradition) will ultimately fail, 'since the concept derives from the nature:culture dichotomy.
2 By viewshed is meant the area that can be viewed from the marae. Murivai and Pureora could have been included in the analysis, but have been omitted in order to avoid cluttering the map. If the reader finds the map already too cluttered they might simply note that where viewsheds overlap the symbology will be relatively darker, therefore dark equals three overlapping viewsheds, light equals only one.
3 Some points about the limitations of viewshed analyses, and this analysis in particular, should be made here. The analyses were carried out in the Spatial Analyst module of Arc View 3.1 on a computerised digital elevation model (DEM), which in this case is derived from a 20m contour topographic map (NZMS 272/8/6). This is a rather coarse baseline, and viewshed accuracy is limited by the definition of the base data. Microtopography will not register at this level, and a 3m hill can block a view just as well as a 300m one. Viewsheds are based on the view from ground level, taking no account of the height of any possible, but currently unknown, superstructures on the marae, or even the height of a human, both of which increase the view, and taking no account of a view being blocked by vegetation. Large i'i (Inocarpus fagifer) and 'utu (Barringtonia asiatica) are today associated with many marae, limiting their viewsheds. This may also have been the case in prehistory—Gill (1876:302) describes sacred trees on many marae of Mangaia. The view, both today and in pre-contact times, may have been more conceptual than real. Viewsheds are taken from a single point, whereas marae cover an area. This is particularly the case in the Tupapa Valley mouth, where the marae around Ārai te Tonga virtually abut each other. Parts of both Koroa and Murivai (Figure 2) probably lie outside the area that views all four peaks. There is no substitute for standing at these sites and observing the actual view in person, as Yamaguchi (2000:140) has done. However the GIS analysis replicated Yamaguchi's observations for Marae Piako, and the other viewsheds may be confidently assumed to obtain a similar accuracy. Finally the peaks I have plotted and labelled are derived from the same topographic map, and were chosen by the cartographer for topographic as much as social reasons, though it is hard to think that such imposing landscape elements as the four peaks mentioned above would not have been equally imposing to pre-contact Rarotongans.
4 A couple of interesting implications of this analysis of status zones on Rarotonga are worth mentioning. In 1997 my fieldwork programme concentrated on archaeological survey of the valley floors, initially recording the taro terraces. These are conceptually the lowest status zone, even though they are a focus of settlement and contain a number of elaborate marae and paepae. I am by no means the first archaeologists in Polynesia to place an emphasis on this low status zone. Secondly the Ara Tapu, the coast road built in missionary times, is everywhere seaward of the old Ara Metua, so not only did the missionaries replicate Tangi'ia's tapu controlling circuit of the island, they played a trump card by putting their circuit in a higher status location. They probably did not realise they were doing so, since the Ara Tapu links the sacred sites of the new order (the churches) which were built on the coast. The cultural prejudices of the Europeans would have been a factor in initial site location. They preferred to be near the harbours of Avarua and Ngatangiia, and purposely underlined the break between the old and the new by establishing their settlements away from those already in existence inland.
5 Maretu (1985:36) refers to Tangi'ia's attempts to cook his defeated brother Tutapu at Matavera, and Marjorie Crocombe, translator and editor of Maretu's manuscript, translates Matavera as ‘Hot face’. The two explanations are by no means exclusive of each other.
6 An alternative is that the Court so comprehensively failed to understand the prevailing point field model of space that the two sides—Rarotongans and judge—gave up talking past each other and the Court imposed the new system by fiat, but the few cases that deal with boundaries do not support this interpretation.