Volume 102 1993 > Volume 102, No. 4 > At the halls of the mountain kings. Fijian and Samoan fortifications: Comparison and analysis, by Simon Best, p 385-448
Back to Search Results                                                                                              Previous | Next   

- 385

Fortifications in Fiji and Samoa (Fig. 1) have been objects of interest to European observers since initial contact. Starting with a young seaman in Fiji in the first decade of the 19th century, through explorers, missionaries, anthropologists, enthusiastic amateurs, and archaeologists, much information has been gathered, although distributed unevenly between the two Island groups.

Figure 1. Location map.

The earliest recorded observations of Fijian forts were made by William Lockerby, in an extraordinary account of his stay in 1808, when he and six - 386 companions were marooned by their ship in Bua Bay, on the west end of Vanua Levu. Lockerby's manuscript, kept hidden in his family for over 100 years, not only describes in detail the construction of a fort and mentions others, but also recounts his involvement in storming one of these, when he was wounded three times and one of his companions killed by his side (Lockerby 1982:25-6, 52-5). Another fort is recorded by him in a set of sailing directions for sandalwood ships; this was on a small island in the Dreketi River, on the north-west coast of Vanua Levu (Lockerby, in Dodge 1972:186). This last manuscript also contains a small vocabulary, in which two types of fortifications are described: Gorobato (Korovatu) with stone defences, and Gorowey (Korowaiwai) with a water-filled ditch or moat (see also Richardson MS 1810-12).

In 1830 the shipwrecked seaman John Twyning visited the main settlement on the island of Lakeba, in the Lau Group, and described it as:

A considerable town containing a population of four or five hundred persons. It was surrounded by a deep ditch filled with water, over which four or five narrow causeways afforded entrances to the town. It was defended by several guns of various sizes from a swivel to a carronade. These guns were mounted on a platform over each of the entrances to the town across the moat and was looked upon as nearly impregnable against any native force that could be brought against it (Twyning 1850:69).

In 1831 Captain Eagelston, cruising the waters for bèche de mer and turtle-shell, recorded three fortified sites. On the island of Naigani, off the east end of Viti Levu, was a hill-fort “…situated on the most lofty part of the island, and inaccessible except by a very narrow pathway, that could be defended against all numbers” (Eagelston 1831-2:298). Another hill-fort was seen at Wailea, and a ring-ditch at Bua (Eagelston 1831-2:303, 311).

In 1840, the United States Exploring Expedition surveyed the Group, during which time six fortifications were described and another seven mentioned (Wilkes 1845:III:80, 145, 172, 176, 178, 198, 206, 211, 217, 226-7, 292, 306). These observations were distributed throughout the Group: in Central and Northern Lau, Totoya, Lomaiviti, the south and west coasts of Vanua Levu, and the north and east coasts of Viti Levu.

Although most of the population, apart from those in the centre of Viti Levu, were living on the coast by that time, often in settlements surrounded by ditch and bank fortifications and/or stockades, they either still had inland fortified positions to which they retired when threatened, or had only recently abandoned these. One of the forts referred to by Lockerby was an inland hill-fort (Lockerby 1982:34; 1972:186), while all but one of those mentioned by Wilkes was also in this category.

- 387

Wilkes' sample of sites is unusual in that, in order to take the observations on which their surveys depended, the expedition climbed to the top of many of the inland hills, and paid attention to others they did not visit. In that way, they found examples of numerous hill-forts.

On Lakeba:

Mr Totten and Dr Holmes were despatched on shore, to ascend Kendi-kendi, the highest peak of the island…The ruins of a town were found on it… This town was occupied for the purpose of defence against their enemies (Wilkes 1845:III:172).

On Munia:

The ascent proved difficult, for the path passed over steep hills and along the edges of the rocks, and it was in places so narrow that only one person could pass at a time. A few men might defend the ascent against an army. Upon the summit they found the ruins of a small village: some of the huts were, however, kept in repair, as refuge in times of danger (Wilkes 1845:III:178).

On Levuka the fort was:

situated upon a hill, and can be approached only by a narrow path along the sloping edge of a rocky ridge. At the extremity of this path is a level space of about an acre in extent, which is surrounded by a stone wall, and filled with houses. In the centre is a rock, about twenty feet high, and one hundred feet square … A house stands in the middle of the rock. This contains two Feejee drums, which, when struck, attract crowds of natives together (Wilkes 1845:III:80).

On Malaki Island, off the west end of Viti Levu:

This island is eight hundred feet high, and on the top are the remains of a fortification of stone, whose walls are four feet high, surrounded by a moat several feet deep, and ten feet wide (Wilkes 1845:III:211).

Other members of the U.S. Exploring Expedition also described forts in their journals. On Kadavu, Titian Ramsay Peale saw:

numerous traces of a former extensive population, and the remains of forts, consisting of stone walls 4 feet thick and bout the same height surrounded by a dry ditch. They are always on the crests of hills (Poesch 1961:174).

- 388

General observations were also made; off the west end of Viti Levu “…the numerous towns … perched on their eyrie cliffs, continued to meet the eye…” (Wilkes 1845:3:206).

Although Wilkes himself appeared more interested in the inland hill-forts, often neglecting to mention the defences of some of the large coastal ring-ditches (e.g. those for Tubou on Lakeba: Wilkes 1845:3:173), the accounts of other members of the expedition were more descriptive. Thus, Assistant Surgeon Holmes supplies the fact that Tubou “…was surrounded by a ditch and wall and was of considerable size, containing I should say 800 or 1000 people” (Holmes 1840: entry for May 28). Peale described the ring-ditch at Bua on June 8,1840:

It is fortified by a stockade and ditch, and has a number of very narrow and low gateways, with a kind of watchbox over each. About a quarter of a mile up the valley is another town similarly fortified. They are both surrounded by tarro patches, which are trenched and the earth thrown up between them forming a kind of causeway on which the paths wind about, the Tarro beds being a little below the surface of the water (Poesch 1961:174).

The expedition sacked the main fort on the island of Malolo, which impressed Holmes with its strength:

In fact the fortification of their town evinced no little skill in engineering. A ditch or moat, 20 feet wide and 10 deep, filled with water, surrounded the whole; within this was a regular palisade made of large timber or in some parts of living trees, about five feet asunder and connected by a thick, densely woven kind of wickerwork made of branches of trees, so strong as to defy all efforts to penetrate it and so dense as to be impossible to see with any distinctness what might be going on within. On the inside of this wall, which was about 12 feet high, was an embankment of earth thrown up completely encircling the town. This was afterwards found to be 2 ½ feet high and of very nearly the same breadth; and last of all there was a ditch of no great depth, in which the besieged might find shelter, only raising their heads occasionally to discharge their weapons through the loopholes in their walls (Holmes 1840: entry for July 26).

The early Wesleyan missionaries are another source of information on the presence and use of fortifications. Of these men, Dr R. B. Lyth was probably the most observant and prolific. In his Voyaging Journal for 1848 he notes, while passing along the Ra coast on November 18 that, at Nairara, there were “…huge overhanging craggy rocks that rise to a considerable height. At the top of these the town is built”, and for the Ra coast in general, that “The towns on the mountain sides look like so many crow's nests stuck on the rocks” (Lyth MS 1848). Four days later, still in the same area, he sees - 389

the rugged lofty cliff on whose impregnable heights stands the town of Navatu. This rocky mountain rises almost perpendicularly to a great height. A more inaccessible town could hardly be imagined. It is a town belonging to Rakiraki; which in time of war is resorted to by the Thokova chiefs (Lyth MS 1848).

On November 23, off the Tavua coast “… we passed another Feejeean stronghold situated on a lofty conical mountain with a tufty top. The town that stands on it is called Vugala” (Lyth MS 1848).

Fortifications in similar hilltop situations are recorded by the missionaries for islands in the Lau Group. For Cicia, Lyth describes three (Lyth MS 1849: entry for November 16, 1852: entries for August 4 and 9). Thomas Williams mentions two seen in 1846, on Kanacea and Vanuavatu (in Henderson 1931:364, 380), and in 1838, Cargill noted a refuge on Namuka (in Schutz, 1977:119).

In the mid-19th century other visitors to the Islands continued to find fortified sites, usually those in use at the time on the coastal flats. In 1849 Captain Erskine of H.M.S. Havannah noted two ring-ditch settlements — the above-mentioned Tubou on Lakeba, and another on Ovalau (Erskine 1853:168, 216) and, in 1856, J. D. Macdonald, assistant surgeon in H.M.S Herald, was able to name and place on a map about 100 such sites while exploring the Rewa River and its tributaries (Macdonald 1857:232).

In 1854 Captain Denham of H.M.S. Herald mapped 21 occupied settlements on the island of Gau, of which 16 were fortified. One of those on Gau was over one and a half kilometres inland, on a spur end at about the 160 metres contour, as was one of the undefended sites; the rest were on the coast (Palmer MS 1973).

In 1860 the naturalist Seeman briefly described two coastal defended sites: on the island of Nananu-i-cake on the north coast of Viti Levu: “…a town defended by a deep ditch and high earthen mounds…”, and on the south coast of Vanua Levu at the town of Solevu (Seeman 1862:223, 248). He also discovered the remains of a settlement at about the 800-metre level in the bush above Namosi, in the headwaters of the Waidina River some 25 kilometres inland on the south coast of Viti Levu (Seeman 1862:162).

In 1873 Commodore Goodenough (H.M.S. Pearl) investigated a site on the island of Wakaya: “…We went up to the highest point of the island. There is a regular fort, with double ditch, nearly circular on top … the foundation of a temple (square) and a round foundation still stand…” (Goodenough 1876:210).

By that year, most Fijians had come out from behind their moats and stockades; only the hill tribes were still fighting. On May 9,1874, the last full-blooded charge, of 1000 warriors, was fought off by Government troops at Nagusunikalou in the upper reaches of the Wainimala (Derrick 1957:244). On October 10 of that year, Fiji was ceded to Britain, and the main causes of war in the country — invasion from Tonga, internal jockeying between power centres, and heathen versus Christian — were no more.

- 390

The first anthropologists in Fiji, who bridged the time gap between the early observers and the start of archaeological research, recorded inland hill-sites and their traditions.

Hocart, a school teacher, conducted his Lauan field work in 1927, and made little mention of fortifications, stating that “…Lau did little fighting … The villages of old were very small, surrounded by a moat” (Hocart 1929:141). In another work, Hocart records that the summit of Taveuni, Nggalau Hill, know as the dying place, was once a settlement. (Hocart 1952:66).

Between August 1933 and April 1934, the anthropologist Laura Thompson visited and mapped some seven hill-forts in the Southern Lau Islands, commenting on the differences between forts on limestone ridges and those on volcanic summits. Of those on ridges she noted that walls of stone were used to supplement the natural defences and to cut off access routes (Thompson 1940:104).

What are described as temporary defences were recorded for the Namosi district by G. K. Roth, a Colonial Administrator in the Department of Fijian Affairs. These were areas of ridges marked by ditches across saddles, narrow places and any approach track. One of these was found at a place called Ngginggimatasere in 1934 during a survey by the Native Lands Commission (Roth 1953:2).


In Samoa, war and fortifications were both present when the first records occurred, although the situation there differed both from that in Fiji, and from islands in the Samoa group itself.

On December 6,1787, La Pérouse made landfall at the Manu'a group, sailing along the north coast of Ta'ū, and anchoring in the channel between that island and Olosega. He saw on Ta'ū “…habitations on the windward side of the island … the houses are built about half way up the hill, where the islanders breath a more temperate air” (La Pérouse 1799:122). The possibility that there might have been other reasons for choosing the elevated sites is discussed below.

In October 1839, Manu'a was also the first port of call for the ships of the United States Exploring Expedition, where Wilkes saw, close to the principal village on Ta'ū, “…thick stone walls, intended to all appearances for defence” (Wilkes 1945:2:66).

The Expedition sailed on to Tutuila, where no fortifications were found, with Wilkes describing the island as “…thickly settled around its shores … the only communication is by the sea-shore, the hills being too precipitous and difficult of ascent to pass over” (Wilkes 1945:2:72). During his 14-day stay at Tutuila, Wilkes did climb across the main range behind Pago Pago to visit the village of Fagasā (Wilkes 1845:2:75) and also ascended one peak, that of Matafao, west - 391 of Pago Pago, but recorded only that the air temperature was 10.1° cooler on the top than at sea-level (Wilkes 1945:2:82).

The earliest report of fortified sites in Western Samoa appears to be that of the L.M.S. missionary John Williams who, in 1832, spent three weeks on 'Upolu, mostly at the western end of the island. He describes a typical fortification, probably based on those in that area:

each party generally provides itself with a strong ola or fort. This is composed of cocoa nutt trees placed about six feet deep in the ground & standing eighteen or twenty feet high. They generally select a high mountain as the place where to erect their fort. To this they remove their property, wives & children erecting temporary huts of cocoa nutt leaves inside. These forts serve very important ends. They afford a place of refuge for them in case of a defeat preserve them from being taken by surprise, & enable them to take every favourable opportunity of sallying out upon the enemy (Williams 1984:244).

In 1838 Thomas Heath, another L.M.S. missionary, referred to a refuge on the mountain of Tafua 'Upolu, on the west end of the island, to which the settlement of Fasito'otai would retreat in times of danger (Heath 1838).

During his two-week stay on 'Upolu, Wilkes found both forts and traditions associated with them. In a general reference, he stated that:

the olos, above mentioned, were usually on the top of some high rock, or almost inaccessible mountain, where a small force could protect itself from a larger one. One of these olos, or strongholds, of the people of Aana, during the late war, was on a high perpendicular ridge, which forms the western boundary of the bay of Faleletai, and it was the scene of many a bloody contest (Wilkes 1845:2:151).

At the end of October, Wilkes cut across 'Upolu from just west of Apia, and found, on the top of the inland ridge at c.800 metres altitude, “…a clearing…in which were two mounds of earth, each about fifteen feet high, and one hundred and twenty feet in circumference; several stone walls were also seen” (Wilkes 1845:2:94, 95). Wilkes was told of a tradition that the fort was built by Tongan warriors from Vava'u, for the purpose of preventing communication between the north and south coasts. There were 60 cm diameter trees growing on the mounds, and the Tongan invasion in question was said by the missionaries to have taken place about 1760-70.

For Savai'i, where Ringgold had taken the Porpoise between October 15 and 25, and where Dr Pickering, the expedition's botanist/naturalist, was under instructions to explore inland, there was a dearth of information on fortifications. Pickering had labour problems and had to turn back after about 11 kilometres, - 392 and Wilkes wrote “…The interior of the island is rarely entered, even by natives, and has never been penetrated by strangers. The only settlements are upon the shore, along which the natives always journey, and there are no paths across it” (Wilkes 1845:2:112).

Between 1838 and 1845 the missionary John Stair journeyed inland from Apia on 'Upolu, and visited a defended site in use at the time: “…the olo, or fortress, was on the mountain side, several miles inland, and thus occupied a commanding position” (Stair 1897:99, 100).

George Turner, another L.M.S. missionary, was in Samoa for nine months in 1841-2, and then from 1843 to 1860. He recounted that all noncombatants were either taken to “…some fortified place in the bush…” or to a neutral district. Warring villages were stockaded, by posts two and a half metres long set 60 cm. in the ground (Turner 1861:298).

In 1849, Erskine saw the flat-land fort of Malietoa on the western side of Apia, where a palisade and ditch had been run across the neck of land to protect the point (Erskine 1853:75).

In the early 1880s, in the course of numerous forays into the bush on the west end of 'Upolu, William Churchward, acting British Consul in Samoa, came across several defended settlements. One of these was found in 1881 while on a trip from Suisega to Lotofaga, in Western 'Upolu:

we started up the almost perpendicular incline in front of us. On arriving at the top we found evident traces of a parapet and ditch, no doubt the work in days of old of the Tongans, who at one time held almost entire possession of Samoa, and portions of whose handiwork in fortifications and roads may be met with all over the island of Upolu (Churchward 1887:109).

In 1882, Churchward also visited a fort inland from Apia, and in 1883 noted, in the hills inland of Utumapu, east of Apia, “…a palpably artificial mound of very ancient date … surely the site of some former house of importance” (Churchward 1887:176, 282).

In the mid-1890s, the redoubtable Llewella Churchill, on her numerous hearty hikes through the bush on 'Upolu, also saw stone-faced mounds high up on the ridges (Churchill 1902:206, 277). This frontier woman, proficient with her favourite firearm, a single barrelled Winchester shotgun, once held a crowd of rebels at bay from midnight to dawn (Churchill 1902:203). They were after guns, and it was a result of their skirmishes that the last Samoan forts were built; it was these that the German anthropologist Augustin Krämer was to describe.

Krämer was in Samoa between 1897 and 1899, and had little to say on fortifications or warfare. What he did note tended to suggest that there were only minor differences between the examples occupied or built for the last stages of - 393 fighting (in Western Samoa, which Krämer photographed in use) and those of more remote times:

The construction of the forts, 'olo, is still fairly similar to that of the old time ones, and the illustrations given here, which are taken from the war of 1896, may therefore be confidently considered as typical of the old time. During the Tongan invasion, the Tongans constructed numerous forts on every side in Samoa, their stone ramparts and the highways connecting them are still shown everywhere (Krämer 1902-3:1:591).

The forts illustrated by Krämer are pretty scrappy affairs, with rough-looking stockades and a look-out platform up a tree, and appear to be fairly temporary responses to local shoot-outs (Krämer 1902-3:340, 341). A similar construction, also in Western Samoa, and dated to about 1890, is illustrated in Moors (1986:opp.53); the walls appear to be piles of wood, and the look-out post is again a tree platform. Krämer notes that William Mariner's description of a Tongan fort is applicable to a Samoan one, and that they were usually built on low mountain ridges.

Krämer also noted the village site of Siliuta on Olosega, in the Manu'a group, situated 300 metres up Piumafua (Alei) Ridge, in a position that could only have been defensive (Krämer 1902-3:1:719).

As with Fiji, the fighting in Samoa was a mix of internal rivalry, the repulsing of Tongan intruders, and strife between the followers of old and new religions. And, as in Fiji, fighting came to an end when the country came under the rule of foreign powers in 1900: for Western Samoa: Germany, and for Tutuila and Manu'a: the United States.

Sir Peter Buck, in Samoa in 1927 and 1928, was fairly noncommittal on fortifications. He did note the remains of a defensive stone wall with a tower at both ends on the sea front at Aunu'u (Buck 1930:322), and remarked that, in general,

Many villages were protected by stone walls termed 'olo. Some places of refuge also termed 'olo were situated back in the hills but owed their protective qualities to natural inaccessibility to attack. No information was obtained concerning special defences erected by the refugees (Buck 1930:609).

Buck was in Samoa for six months: two in both Manu'a and Savai'i, and about three weeks each in Tutuila and 'Upolu. His information on defended sites on 'Upolu contrasts with that of earlier observers, possibly because of his short stay there, and because of the 30 years which had elapsed since the last battles were fought.

The comments of A. Wright, a soil scientist who worked on 'Upolu in 1956, should - 394 perhaps be included here. Wright's knowledge of inland 'Upolu is probably unequalled, and he found evidence of numerous inland villages, both from soils and structural features, and the presence of “…defensible hills, many of which still show signs of the ancient earthworks and fortifications” (Wright 1963:91).


An overall view of the whole of Fiji and at least part of Samoa at the time of European contact, based on the observations and experiences of early visitors, shows that there were basic similarities between them with respect to settlement patterns and warfare. In both regions, most of the population were living on the coast, with evidence for more extensive inland settlement in the past. Both were also actively engaged in warfare, which although undoubtedly containing a symbolic aspect, and often involving treachery rather than set battles, was nonetheless deadly and resulted in considerable loss of life. Fortifications were in use in every area of Fiji, but for Samoa appeared to be restricted mainly to 'Upolu. These were found on both the coastal flats and inland hills, while on 'Upolu the remains of old forts were also to be found in the bush. Although Tongan forces were active in both countries, it is only on 'Upolu that they were said to be responsible for the remains of the inland fortifications.


The existence of impressive man-made fortifications on the rugged jungle clad peaks of both Fiji and 'Upolu in Western Samoa has, thus, been known for more than 150 years. Archaeologists, however, have usually tended to prefer the more appealing coastal sites, although in their favour it must be said that many of the latter have been exposed and threatened by modern land use, requiring salvage archaeology.

Western Samoa

Archaeological research in Western Samoa, and indeed for the group, began in late 1957 with the work of Jack Golson on 'Upolu (Golson 1969:14-20). Golson found evidence for inland fortifications, although only preliminary attempts to investigate these were carried out in the 38 days of the project. Terraces, scarps, trenches and ditches in inland situations were described in a section titled “strongpoints”. One of these sites was referred to as a fort (Mafafa), the rest as series of features that were not interpreted.

The second (and largest) archaeological programme in Western Samoa, that run by Roger Green and Janet Davidson in the early 1960s, was also somewhat enigmatic with regard to inland fortified sites. Although recognised as present in some numbers, on 'Upolu at least, many were described as consisting of single features on ridges, such as ditches and banks, i.e., SU-Lu-38 and 40 (Davidson - 395 1969a:203) and Moamoa (Davidson 1974a:202). Less common forms were described as consisting of small ditches and scarps, while only two were recorded as hilltop forts, at Luatuanu'u and Solaua (Davidson 1974b:241).

The Luatuanu'u site (SU-Lu-41) was surveyed, and the northern defences excavated. Until recently, this was the only mapped fortification for the Samoa Group, the date from which (1500±80 B.P.) has been used to suggest that Samoa was the first country east of Fiji to develop this response to warfare (Bellwood 1979:316).

The second major archaeological project in Western Samoa was that run by Jesse Jennings, carried out in 1974, 1976, and 1977. This, however, focused mainly on the excavation of coastal middens and the mapping of extremely large areas of coastal settlement, apparently undefended, which extended up the gentle inland slopes between about 50 and 115 metres above sea-level (Jennings et al. 1976: Jennings and Holmer 1980).

Between 1977 and 1992, in direct comparison with Fiji and to a lesser extent American Samoa, no archaeological work was carried out on 'Upolu and Savai'i.

American Samoa

The first archaeological research in American Samoa was that by William Kikuchi who, in 1961 and 1962, listed the known archaeological features throughout the group, including some unseen by him but recorded in local tradition (Kikuchi 1963). Observing that “… Very little in way of defensive fortifications exists in American Samoa” (Kikuchi 1963:66), he eventually recorded four, although none was visited.

The first of these was the settlement on Olosega mentioned by Krämer, and referred to above. The other three were on Tutuila, and consisted of several stone-lined trenches off the west end of the old Tafuna airport runway, a series of ditches on Alava ridge, overlooking Pago Pago, and a large ditch on the ridge inland of Leone (Kikuchi 1963:68). These last were all attributed to Tongan influences.

Janet Frost, as part of her Ph.D. field work on Tutuila in 1972, recorded what she described as two fortified settlements: on Mt Alava and at Lefutu. These were both “…on high mountain ridges that are difficult of access, and both areas show evidence that the ridges have been modified by the construction of transverse ditches to make access even more formidable” (Frost 1978:240).

These appear to be the first such recorded for American Samoa (although that on Mt Alava was probably the site reported to Kikuchi); however, the identification of these as fortifications has been questioned (Clarke 1989:144: Clarke and Herdrich 1993:159).

In 1980 Jeffrey Clarke recorded all known sites in American Samoa during the process of compiling a site inventory for the Government of American Samoa. Of the 170 sites listed, only one was described as a fortification; this was - 396 the series of stone-lined trenches on Tafuna Plain that had been reported to Kikuchi (Clarke 1981) but, again, these were not seen.

In 1985 Clarke and David Herdrich undertook an ambitious and extensive site survey in Eastern Tutuila. A large number of prehistoric inland sites were recorded, mainly terraces and mounds, but none was identified as being defensive in function (Clarke and Herdrich 1988:71).

In the same year, Helen Leach and Dan Witter relocated the adze quarry on Tutuila first recorded by Buck in 1927 (Leach and Witter 1987). This they interpreted as a fortified stone resource, based on the presence of a small discontinuous ditch on the uphill side of the highest terrace.

In 1986 Terry Hunt and Patrick Kirch undertook a site survey of the Manu'a group, finding 63 sites, but recording no fortifications (Hunt and Kirch 1988:153-83).

Also in 1986, William Ayres and D. Eisler conducted a site survey and excavations in Western Tutuila, in the Maloata district on the north-west tip of the island. The results of this project have not been published; however, evidence for a possible local basalt source is said to have been found (W. Ayres, personal communication).

In 1988, Clarke and Herdrich returned to complete their Eastern Tutuila survey (Clarke 1989), and Leach and Witter, together with the author, carried out the second stage of the Tatagamatau project (Best et al.:1989).

Between 1970 and the present, several small surveys and site specific projects, some with minor excavations, have been carried out in American Samoa, mostly as a result of road or harbour development, a proposed garbage disposal site, sewer-line projects, etc. (see Clarke and Herdrich 1993:147 for a summary, and also Best 1992a-e: Foster 1991, 1992). Although these all add to the growing body of archaeological knowledge for the area, they are in a different category from those mentioned above and which have provided most of the hard archaeological data now available for the Islands.


Archaeology began in Fiji in 1947 with the preliminary survey and excavations by Edward Gifford (1951). He visited 39 sites, all identified in local traditions, one of which was an inland fortification, Korovatu. This was a ridge fort at about the 800-metre contour and is described as having a “citadel” at the highest end, the height of which had been increased by stone walling (Gifford 1951:246, 247).

Over the next 16 years, the only archaeological work in Fiji was carried out by enthusiastic amateurs. One of these was Aubrey Parke, District Officer and later District Commissioner, who had located and recorded ring-ditch sites in the Rewa Delta, and others such as Dr Lindsey Verrier, who visited hill-forts on the smaller islands of Beqa and Ono-i-lau (Parke 1961:23). In 1960, Parke gave a - 397 paper to the Fiji Society entitled “Archaeology in Fiji” (Parke 1961:10-42), in which he brought together all the archaeological information available up to then, and also related associated oral traditions.

For the first time, Fijian archaeological sites were classified and described, with fortifications one of the 11 categories. In this section, drawing heavily on Lauan material, Parke discussed briefly the types and combinations of defences — ditches, banks, stone walls and terraces — that were associated with sites built in areas of different geology: volcanic, volcanic and limestone, and limestone.

The appointment of Bruce Palmer to the directorship of the Fiji Museum in 1963 resulted in at least four archaeological surveys in the mid-1960s, in which the recording of hill-forts was either the sole or a major part of each project.

The first of these appears to have been field work carried out in 1964 by Les Thompson on the island of Wakaya, and by Colin Smart on Kabara. Thompson was a surveyor working for the Lands Department in Suva, and he located and mapped eight fortifications in what Palmer rather delightfully describes as his off-duty “leisure work” (Palmer 1967a:19). These sites were revisited in the following year by Palmer and a Fiji Museum party, and the work extended to include another two fortifications. Two more large forts on the island were subsequently located and mapped by Fergus Clunie (personal communication).

On Kabara, Smart recorded nine fortifications, seven of them on the limestone ridges that rim the island, the other two on the volcanic hill of Delaioloi (Smart 1965:8). The ridges are described as having fairly extensive stone-faced terracing, with the highest point of the sites often containing a stone platform, and all access ways blocked with stone walls. The two sites on Delaioloi, although also terraced, were unusual for Lau in that their outer defence was an encircling free-standing stone wall.

Both projects influenced Palmer's programme for tackling the problem areas in Fijian prehistory. In an appendix to Smart's report, he defines one of the major aims as tracing the development of fortifications in Fiji, and four of the five projects outlined are concerned either wholly or in part with this (Smart 1965:28, 29).

The year 1965 saw the start of the most ambitious of these, the Sigatoka Research Programme, which set out to survey the whole 80-kilometre length of this large river valley. In the first year of the programme, 67 sites were recorded, six of which were ridge forts, and 23 of these were surveyed (Palmer 1967b:13). Another 30 traditional sites were known for the valley, and 30 sites had been identified on aerial photos. It is not stated whether there was any overlap between these.

The ridge forts were on rugged limestone peaks and ridges, employing stone terraces and walls, with some ditches, and with stone platforms on the high points. Palmer describes one of them as having 37 mounds and 20 terraces, many of which he records as being associated with “a large citadel” at one end of the site (Palmer 1967b:4).

- 398

Between 1966 and 1968, Elizabeth Hinds undertook a survey on Taveuni as part of the Fijian prehistory programme. Hinds identified five classes of fortified sites, three of which used height and slope to varying degrees (Hinds n.d.). These were:

  • 1. Ring-ditches
  • 2. Fortified volcanic cones
  • 3. Fortified ridges, promontories and villages
  • 4. Stone forts
  • 5. Single mounds

The ring-ditches were dug on flat coastal land or on gently domed hills. The single mounds appear to be a variant of these: a slightly raised area surrounded by a ditch, and also occurring on flat coastal land.

The cones, which occur on the island at all altitudes, were defended by series of concentric terraces and, in two instances, by ditches and banks. These cones appear to resemble sites recorded for Samoa: Lu-44 on 'Upolu (Davidson 1969a:187) and on Savai'i (Buist 1969:41), and may in some cases be an adaptation to certain aspects of a young volcanic environment.

The ridge forts, the most common category on the island, were fortified by transverse ditches, terraces and stone walls, some of the ditches being up to 15 metres deep. Hinds describes one of the sites (T16/2) as being representative of most such forts on Taveuni: “The whole of the village is fortified by naturally steep land, a deep ditch and stone wall, but the most heavily enclosed area is the small knoll at the head of the promontory containing the house of the chief” (Hinds n.d.:58). This knoll is the highest point in the immediate area, and contains a central large high mound, 14 x 10 metres and 2 metres high, and it is local tradition which has assigned it to the chief.

The stone forts, four in number, all had higher central areas, with platforms or mounds, and were defended by stone walls and terraces.

Hinds' work on Taveuni was followed in 1968 by that of Everett Frost, whose doctoral research involved excavating four of Hinds' inland sites, as well as three more located by himself (Frost 1974). Although no additional inland site types were described, Frost dated the emergence of forts on the island to about 700-800 B.P., with a single date, of c.2,000 B.P., from the earliest and probably undefended phase of one of these (Frost 1974:58). The later dates are single examples from three sites, but are supported by association with similar ceramic assemblages. Although subsequent work by Babcock, using Frost's material, has seriously questioned some of Frost's findings on the cultural affinities of the sites (Babcock 1976), the actual dating of the fortifications appears acceptable in the light of results from similar situations in the Lau group.

In 1967, Palmer examined Denham's Gau sites on the ground, finding a good correlation between the English captain's positions and descriptions and today's archaeological remains (Palmer 1973).

- 399

In 1975, Garth Rogers conducted a survey of archaeological sites on nine islands in the Northern Lau Group, and obtained information regarding a tenth (Lawlor 1981). One hundred and two sites were recorded, of which 19 were fortified. Ten of these were inland hill-forts, eight were coastal limestone ridge forts, and one was a coastal ring-ditch. The hill sites that are described or sketched use natural features such as steep rock faces, together with ditches, stone walls and terraces, and have a central or highest point with stone arrangements or platforms. In one instance, on the main fortified site of Delaiyatova on Munia (the fort visited by Totten and Holmes in 1840), a set of four pits also appears to have been part of the defences (Lawlor 1981: Fig. 4.0).

In three field seasons between late 1975 and 1978, in the course of the author's Ph.D. field work in Central and Southern Lau, 47 fortifications were mapped. Twelve of these were coastal ring-ditches, the remainder being fortifications of varying size either on inland volcanic hills, or on coastal limestone ridges (Best 1984:48-53 for Lakeba data: the Southern Lau material is unpublished). Although other small inland sites, on low rises, were apparently undefended, it is possible that palisading was employed in these circumstances.

In 1978-9, Rod Vickers and Charles Eyman undertook the Nadrau Archaeological Project, on the central highlands of Viti Levu, between the Nadrau Plateau and Mt Tomanivi (Vickers and Eyman MS1980). The study area was under threat from the Monasavu Hydroelectric Scheme, and 17 archaeological sites were located, two of which were excavated. Thirteen of these sites were hill-forts, defended by a combination of natural features, terraces and ditches. In general, the largest yavu or housemound was on the highest and most strongly defended part of the site, and was interpreted by the researchers to be the foundation for the men's house of the village chief. Other mounds on the high place were suggested as being the dwelling house of the chief, and the foundations for the burekalou or temple (Vickers and Eyman MS 1980:59).

Aerial site surveys, using photogrammetric techniques, were employed in three watersheds on Viti Levu by John Parry, between about 1975 and 1984. These were the Rewa Delta (Parry 1977), the Navua Delta (Parry 1981) and the Sigatoka Valley (Parry 1987). These surveys located or confirmed a large number of fortifications, mainly ring-ditches, but also some hill forts in the Sigatoka Valley.

In 1984, Geoffrey Irwin mapped and test excavated a sample of the sites in the Navua Delta, mapping a number in the Rewa Delta, and excavating more fully one site at Laucala; all these were flatland ring-ditches.

In 1985 and 1987, in the course of field work on Beqa, Andrew Crosby recorded 40 defended sites. Fourteen of these were coastal ring-ditches, the rest hill-forts, ranging from small settlements on low knolls to large and complex sites on the highest peaks and ridges (Crosby 1988:31-5).

- 400

Recent work involving fortifications includes surveys on Taveuni by both Crosby and Clarke (Crosby, personal communication), on Totoya by Clarke (personal communication), on Wakaya by Robert Rechtman (1992), on southest Viti Levu by Mara Rosenthal (1991), and on north Viti Levu by Kaplan and Rosenthal (1993).

Most of the early descriptions of Fijian fortifications, and subsequent research as outlined above, occurred on Viti Levu and Island Fiji, with little information from the second-largest landmass in the group, that of Vanua Levu. What data there are suggest that the same pattern also applies there. Coastal ring-ditch and inland ridge forts have been reported by Clunie, who recorded many large hill-forts in the Central Western area, on Delanacau mountain, and who states that “… Clearly, the interior of Vanualevu was at one time settled in much the same way as the interior of Vitilevu” (personal communication).


The archaeological work in Fiji has, thus, sampled fortifications over the whole archipelago. All site types have been mapped and described, and enough detail on the various kinds of defended sites is known for some basic rules of construction to be outlined.

Implied, if not stated, in the above research is that the shape of a fortification results from the interaction of two main factors: the immediate environment, and the culture of the builders. The environment features the topography, soils, and general setting, and the culture incorporates the desire to shape the site to some preconceived design that owes more to cultural preference than to any physical component of the landscape.

It is suggested here that there is little, if any, cultural determinant in the actual shape of each fortification. For every situation to be defended there is one optimal design of fort and, since the penalty for failing to meet this could be fatal, not just for the builder but for his whole tribe, cultural whims had no place in the plan.

The three main site types shown in Figures 2, 3 and 4, which are all on the island of Lakeba, in the Lau Group, are representative of fortifications in general throughout Fiji. The principles behind constructing these are basic and cross-cultural, indeed world wide.

If the site is on a featureless plain, one type of defence surrounds the inhabitants, usually in a circular or subcircular design. The only differences between a Boer laager in South Africa, a Rewa delta or Lakeba ring-ditch in Fiji (Fig. 2), and Avebury Henge in England, are of scale and permanence; all were constructed to protect the inhabitants and/or their gods.

The next step involves a height advantage within the outer ring. A small symmetric hill or cone provides just this situation, where an encircling ditch, or series of terraces, secures the site below the peak. Hinds' Class 2 sites on Taveuni

- 401
Figure 2. Nakorovusa; the Tubou ring-ditch site on Lakeba, Lau Islands, Fiji (kinship areas underlined).
- 402
Figure 3. Kedekede; the volcanic hillfort on Lakeba, Lau Islands, Fiji.
- 403
Figure 4. Ulunikoro; the limestone ridge fort on Lakeba, Lau Islands, Fiji
- 404

fall into this category. The man-made equivalent is the standard castle and moat, where the battlements are the hill top.

Most hills, however, are asymmetric and have spurs running up to the high point. Where the hill is low, it may still be preferable to have a circular or subcircular defence, as either the slopes between the spurs are not steep enough to act as a defence, or there will be sufficient flat area on the summit for the settlement. Many of the New Zealand Taranaki sites described by Prickett (1980) and the English hill-forts illustrated in Fox (1976) are in similar situations. The hill-forts in this class reported for the Sigatoka Valley by Parry may also be a result of these parameters (Parry 1987:109), and it may be these that Palmer is referring to when he wrote “…despite their location in hill country some upland ring-ditches should really be considered flatland varieties because they are sited on prominent benches or level areas in otherwise hilly terrain” (Palmer MS 1972). Certainly, the upland ring-ditch on the island of Nananu-i-Ra illustrated by Palmer (1969:16, see below) shows exactly this situation.

Where the hilltop is high enough, the gully heads will be sufficiently steep to act as defences in their own right (or can be sufficiently guarded by palisades), and it is only the more gradually sloping spurs that will need modifying. Kedekede, and many of the large hilltop forts in Fiji and Samoa, are examples of these. On Kedekede (Fig. 3) the remains of the ring-ditch can be seen in the ditches on two of the spurs, and in the terraces on the third. Joining the ends of these are steep gully head slopes, considered too steep to afford access, which effectively close the circle of defence around the site.

Where a hilltop is not available to the defenders, lengths of ridges or ridge ends are used. If these are in low rolling country, then elongated ring-ditches, sometimes with additional defences at one or both ends, will still be the best way to seal the site off from attack. Where the topography is rugged, transverse ditches and/or scarps across the top length of ridge, and terracing on the spurs, are blended with the naturally steep slopes on the ridge sides. The fort of Ulunikoro (Fig. 4) shows a variation of this last type, where a discrete length of ridge is steep enough to be defended almost entirely by scarps and terrace fronts.

The above are not put forward as classes or categories to which most Fijian or Pacific fortifications can be assigned, but rather to indicate how the principle behind the simplest and most obvious of all fortifications can be applied to sites which are complicated and which extend widely over the landscape.

An example of this is Ulunikoro and its wider context (Fig. 5). This fort was on one edge of a karstic limestone plateau of some 38 hectares and, in conjunction with another and smaller fortification on the other end of the plateau, used a mix of natural and artificial features to ensure that the whole area was defended. These included a row of large limestone sinkholes which, together with sets of ditches and banks at either end and the steep limestone cliffs - 405 to seaward, resulted in a line of defence stretching for two kilometres. Any weak point in this had been strengthened; an example of this is the cave at the base of the sea cliffs.

This cave had been visited by the missionary Thomas Williams on February 27, 1842, and explored by him and his party (Henderson 1931:70,71). The rock platforms at its mouth were described to Williams by the Fijians as having been erected “…by Tonguese devotees as “lounges” for the goddess “Jene vei ngere”, whose abode they suppose this cavern to be” (Henderson 1931:71). The cave runs back for some 300 metres under the plateau, connecting with the sinkhole on the edge of which the smaller fortification had been built, and is a means by which the external ring of defences around the plateau could have been breached (see Fig. 5). The stone walling is, thus, more readily interpreted as defensive in function, taking into account the whole context of the site. Other apparently unconnected features at some distance from the main site, such as the infilling by rocks of a vertical crack in a rock face (which could have been of aid to a climber) and a stone mound overlooking an access way up the cliffs, were also part of the outer ring.

Any attackers breaching this first line of defence met with a series of further obstacles. On the north side of the main fortified limestone ridge was a large walled enclosure, with a free-standing wall guarding the entrance through this. All embayments in the base and flanks of the main ridge itself were modified with stonework, from terraces 100 metres in length with three-metre-high fronts, to rock walls only a few metres long blocking off small ravines that led to the top of the site. The ends of side spurs were built up to make access even more difficult, and to provide platforms as close to the edge as possible, from which defenders could harass any intruder who had got that far.

This fortified limestone plateau can be regarded conceptually as much a ring-ditch site, with internal divisions, as any in the Rewa delta, with the series of ditches, banks, scarps and cliffs encircling the fortification. Indeed, this is exactly how it was described to Hocart by the paramount chief of Lau, who stated that the fort had “…about eight villages surrounded by one big moat” (Hocart MS n.d.:209).

Internal organisation

In many of the descriptions of Fijian hill fortifications described above is the mention of a “citadel” within the defences. This is the highest point on the site, is the most defended, and has usually been modified into a flattened level area containing one or more of the largest or tallest house-mounds on the site. In some sites, especially those along ridges, there may be more than one of these; for example, there are four major built-up points and three lower or less modified ones along the main spine of Ulunikoro (Best 1984:107, and see Figs 4 and 5).

- 406
Figure 5. The Ulunikoro defended complex. High points are marked: built-up (larger symbol), unmodified (smaller symbol).

These, as described above, are suggested or implied to be indications of status within the society. Although local oral traditions often tend to support this, it is sometimes difficult to determine whether the traditions have been influenced by the obvious physical differences among the features concerned.

- 407

Such traditions concerning the ridge forts of the upper Sigatoka Valley may be more reliable than most. These were told to Clunie by two elders, one of whom he describes as outstanding in his detail of 19th-century Sigatoka sites and warfare, and who had been schooled by men who actually fought in the 1876 wars. These informants identified mounds on the heavily defended high points of such forts as those of the chiefs and priests (Fergus Clunie, personal communication).

Status areas within coastal ring-ditches have also been identified. On Lakeba, the old ring-ditch site of Tubou, occupied between about 1790 and 1869, is well preserved, and was mapped by the author (see Fig. 2). Information is available from historic and traditional sources on the identity of people who lived in the various areas within the site, which was the house-mound (Vatuwaqa) of the paramount chief, and which were the temple mounds (e.g., Best 1984:611; Lyth MS 1850-1: general notes). Internal divisions between kinship groups are visible on the ground, as shown by the very shallow “ditch” that is present in the east half of the site. This ditch appears to separate the two main divisions of the site, as recorded by Lyth. These were Tubou, with two heads of families or mataqali, and Tai, the larger part, which was subdivided into five such districts, and which Lyth refers to as the “…royal part of the town” (Lyth MS 1850-1: general notes). There may have been other similar boundaries; on September 14, 1847, the Rev. Walter Lawry visited the town and remarked: “…This chief city of Lakemba is another Venice, intersected with water in all directions” (Lawry 1850:35). The names of the districts within the fort where the six other settlements on Lakeba took their positions in time of interisland war have also been recorded (Lyth MS 1850-1: general notes).

An attempt was made during the author's Lau field work to examine the correlation between specialised pottery and known or suspected high status areas within selected sites. This involved a surface collection of potsherds from the above-mentioned Tubou ring-ditch, and comparing this with assemblages collected from both the top and lower areas of prehistoric hillforts.

Two aspects of the ceramics were chosen. The first was the presence of two vessels associated with the ceremony of preparing and serving kava. These were the broad-rimmed bowl known as dari, in which the drink is prepared, and the narrow-necked vessel or saqa, which contains the mixing water, both of which are normally decorated. Although yaqona drinking today has a purely social aspect, there are still occasions where it possesses a definite nonsecular quality (e.g., Sahlins 1981:125, 126; Phelps-Hooper 1982:182). That the situation was different in the past is indicated by observations of early observers, (e.g., Williams 1858:144, 145; Calvert MS 1860-3; Patterson 1967:89, 90), where a strong, and possibly exclusive, religious aspect is recorded.

The second category of ceramics associated with social standing includes - 408 pots made with a temper sand that could be sourced to the Rewa area. These are taken to represent contact between the Tui Nayau, or paramount chief on Lakeba, and a chiefly area in the Rewa Delta, probably Bau. The evidence for this, in both oral traditions dating back to about 1760 and from the situation extant when the Europeans arrived in the 1830s, shows that the two chiefly lines of the areas were connected by marriage, and that Bauan women (one report suggests up to 20 at a time) were housed in the area of the chief's house at the site (Best 1984:611).

Figure 6 shows the distribution of the two categories; both are clustered closely around the paramount chief's house-mound.

Figure 6. Distribution of sherds from selected vessels at Nakorovusa, Lakeba. A; saga and dari. B: vessels with exotic quartzofeldspathic temper. Housemound shown is that of the paramount chief.

Using a 2-cell chi square test, the assemblage from this mound and its four neighbours was compared with that from the rest of the site. The difference between the two areas for the saqa and dari was significant at the 0.05 level, and for the Rewa temper at 0.01.

The only site assemblage that could be statistically compared with this was that from Ulunikoro, where the sherds from the top of the site were compared - 409 with those from around the base. Since only one dari was found on the site (this vessel shape did not appear on the island until c.400-500 B.P.), two other aspects were used: all temper that was foreign to Lau, and the decoration techniques of applique, incising, shell impressing, and end-tool impressing (all of which appear on the saqa). In both instances, the differences were significant at the 0.001 level.

There were indications from other sites, too, that the dari, at least, was more plentiful on the top area: at Kedekede, nine of the 20 dari rims from the site came from the vicinity of the two largest house-mounds on the platform; however, these were not collected as part of a total sample.

On Taveuni, a reference by Frost for site Val 16/211 describes the top as having “…concentrations of numerous sherds, mostly from wide, flat rimmed bowls called dari, which litter the surface of the levelled areas” (Frost 1974:42).

The overall distribution of pottery itself, within a flatland site at least, may also indicate areas of relative status. The tight concentration of all sherds at the Tubou ring-ditch was directly south of the chief's house-mound, some 20 metres away, perhaps an indication of the social activities concentrated around his household.

The platform often present on the highest point of the site is larger than required for the habitation area for a select individual or group. There are indications that this may represent a ceremonial area.

An observation by Phelps-Hooper on the deck layout of chiefly Fijian canoes is of interest here. Referring to the large double-hulled Fijian drua, he stresses the association between these and both chiefs and gods (see also Hornell 1936:319; Hocart 1929:128, 129). The same point was graphically made by one of the Lakeba missionaries, who wrote in his journal “…their canoes are adored as gods…” (Cargill 1832-8; entry for November 30, 1837).

The canoes transported the chief and his warriors to war, and also brought back human sacrifices and other offerings for the temple (Phelps-Hooper 1982:248). Phelps-Hooper points out that:

These double canoes were not merely efficient methods of transport, for an analysis of their form reveals that they were a mobile reproduction of the centre of the chiefly village. They had a flat deck called the rara (ceremonial ground), to one side of which was a house for the chief, built of vesi timbers and decorated with cowrie shells and elaborate coir bindings. Also on the edge of the rara was a pair of lali gongs (Phelps-Hooper 1982:248).

A further indication of status was the elevated platform above the deckhouse, where the chief could sit during fine weather; in addition, the deckhouse itself could house a miniature burekalou or god house (Fergus Clunie, personal communication).

- 410

If the chief and his warriors transported the “centre of the village” with them when engaged in warfare and other affairs of state on one element — water — then they would have been likely to do so when doing the same function on another — land. There are indications from both historic observation and archaeological survey that the chiefly centre of the hilltop fortifications may have included the rara or its equivalent.

When Hale and Sandford, from the United States Exploring Expedition, visited the hill-fort of Waitora near Levuka (see above), they foun, on the highest point of the site a rock some six metres high and 30 square, a house and a pair of lali slit drums. Almost without exception throughout Fiji, these drums are found on the edge of the rara (see Phelps-Hooper 1982:67), and an 1856 painting shows a pair beside a temple (Clunie 1977: Plate 5).

Of the Lakeba fortifications, that of Ulunikoro (Fig.4) is the most spectacular, where the two largest stone platforms each have a house-mound on one end or side of the platform itself. In the case of the largest platform, this can have been the only such mound, as the remainder is a level but uneven surface of large bare rocks with no soil covering. As will be seen below, this situation is similar for Samoan fortifications.

Summary of Fijian data

Over the past 30 years, considerable research has been carried out on Fijian fortifications, and their importance (if not their distribution) in the archaeological landscape is well understood. The various forms their defences can take are generally regarded as being, in the main, a response to their local environment, although now and again a whiff of cultural determinism can be detected in the literature (e.g., Palmer 1969). Although many researchers in Fiji have recognised the basic principles behind the internal layout of a fortified site, even to the possible significance of the highest and most heavily defended point or points (e.g., Crosby 1988), the size and complexity of some of these forts have not been appreciated.

One account of a generalised hill-fort, however, is worth reproducing here as it embodies all the principles outlined above, and stresses the dispersed nature of the defences:

By no means all Fijian forts were circular ditched ones, forts in more rugged country taking every advantage of the difficult terrain by being built along razor-back ridges and on cliffed and virtually unscaeable crags with formidable natural defences, some occupying such strong positions that even without artificial defences, a handful of determined men could defy an attacking host. On even the strongest, however, the naturally occurring defences were strengthened by scarping, or digging out natural slopes to steepen them; the excavating of war-ditches across ridges and of artificial defensive terraces, - 411 fronted by fighting fences, on steep slopes; and the erection of earthen banks, loopholed loose stone walls and fighting fences across tracks and at weak points. The narrow, steep tracks leading up to these crag or ridge strongholds were often blocked wherever they ran through a crack in the rocks by a stone wall or fighting fence, even at a considerable distance from the settlement they defended, so that an attacking army, strung out and vulnerable thanks to the rough country, had to storm barricade after barricade, and run into ambuscade after ambuscade before approaching the actual fort. In similar manner, defenders of forts near navigable rivers hindered canoe access by constructing a fence or barrage of logs across the stream (Clunie 1977:17).

Clunie based the above on both his own field work and the reports of early observers, and the situation he describes is laid out on the ground at sites like the large hill-fort of Ulunikoro.

These extended defence systems may not be restricted solely to hill-forts, but apply also to at least some of the coastal ring-ditches. An account by Cannibal Jackson describes the defences of Tokatoka in the Rewa Delta in early 1842:

I…bent my course towards Tokotoko, the place of warriors. I was surprised to see the intricate crooked paths that led round innumerable moats and ditches, so constructed as to baffle and perplex the enemy. These ditches extend at least four miles round, and beyond the suburbs of Tokotoko, and have taken, I should say, the labour of this last century to complete. One can see the remains of old ditches for seven or eight miles, and in fact all over that part of the land which is low and affords no natural defence (Erskine 1853:459).

Failure to understand the principles behind the building of a fortification is likely to have an adverse effect on any archaeological research programme that includes these structures. In the case of the smaller sites, what may be a response to local conditions can be misinterpreted as a cultural trait, and lead to spurious assumptions with regard to inter-regional comparisons. The large fortified complexes, on the other hand, may be identified only in part, or in some cases not at all, with unfortunate results for any aerial or chronological interpretation of settlement patterns, and for any functional description of the section that has been found.

An example of the first situation is provided by Palmer (1969:15-19). In that article, he illustrates two Fijian fortifications, from Nananu-i-ra island and Navolau, Rewa (Fig. 7). Both are hilltop sites; one with a ring-ditch and the other with transverse ditches across the access ridges. In his article, Palmer suggests that both fort types are culturally determined. For VL 3/3 he says: - 412

there is, however, clear evidence that the ring-ditch form was applied to a topographic situation in preference to ditch-cutting across the three separate ridges. This is a device which might legitimately be expected in such a situation. Its absence suggests a strong cultural impulse to reproduce the ring-ditch form, hence any site which does show cross-ridge cutting of ditches in a similar situation could therefore be looked upon as significant when occurring within a clearly defined concentration of ring-ditches (Palmer 1969:18).

Figure 7. Bruce Palmer's hilltop sites: VL3/3 on Nananu-i-ra, and VL13/3 at Rewa.

With regard to VL 13/3, Palmer writes:

There are known sites which suggest that with some care, a ring-ditch fortification could have been constructed at Nakovekovevou by angling the lay-out to take in more of Ridge A. On some sites with a slightly less pronounced slope the hillock has been completely ringed near its base, leaving the citadel in splendid isolation well above the defences. That the Nakovekovevou people did not do so might suggest that this site reflects a different fort-building tradition, especially as it is situated in a river valley and delta system which contains several hundred ring-ditches (Palmer 1969:18)

- 413

As is made plain in the plans and cross-sections, however, the two different defence systems are responses to the topography. If the ring-ditch of VL 3/3 was replaced with three transverse ridge ditches, there would have been no defence between them where the gently sloping gullies run up to the top.

Any attempt to construct a ring-ditch at VL 13/3 would also defy logic. There is no need for artificial earthworks at the gully heads, as the slopes are sufficiently steep enough, and including more of ridge A would place the whole site in jeopardy. This is the ridge most vulnerable to attack; the ditch has been placed in the natural, although slight, saddle before the top, and has been strengthened by an additional ditch and bank above it. Extending the site up the ridge would remove these advantages and result in the extremely difficult situation of creating a defence across land sloping uphill from the site; it would also give any raiders the advantage of attacking downhill.

In the above examples, the actual situation is exactly opposite to that which Palmer describes. Both sites were built to use advantages provided by the topography, and there is no cultural or traditional input with regard to the relationship between fort type and situation. Had the site types been reversed, as Palmer suggests they should be, then culture or tradition would have had to have been a factor; the alternative, a total misunderstanding of the problem facing them, was a luxury that the fort-builders, unlike their analysts, could not afford.

It is, of course, possible that there was a cultural reason for choosing in the first place a hill that would support a particular type of fortification. This seems unlikely in the above case, however, and a study of the local settlement pattern and topography in each district would be required before this could be determined. But one site already described above may well be an example of just this: the fortification of Ulunikoro on Lakeba and others like it in Lau (see below).

Failure to identify the fortification itself, or to merely record a part of it, is likely to happen only when the site is, like Ulunikoro, large and spaced out, and covered with tall dense vegetation. However, this is precisely the normal situation in Samoa, both for the bush and the site sizes, and examples of these very circumstances can be identified over the 35 years of archaeological work in those Islands.

Western Samoa

The pioneering site surveys and excavations of Golson, Green and Davidson in Western Samoa in the late 1950s and early 1960s contributed much of what we now know about Samoan prehistory. At the same time, many of the descriptions and illustrations of fortifications studied during the project appeared to contravene the basic principles common to similar sites in Fiji and elsewhere, as outlined above (and see Best 1992f). The main problem appeared - 414 to be that the sites were on ascending ridges or spurs, with no flattened high area or areas, and no sign of rear defences (e.g., Golson 1969:17).

In August 1992, the author visited a number of these sites; the results of this project have been described (Best 1992g, and 1993:in press). Although one of the hill-forts very briefly revisited was found to be basically as reported by Green and Davidson, further work on this site is needed. The other fortifications are described below.

Figure 8. Mafafa fort, 'Upolu, Western Samoa.
- 415

Mafafa. The fort of Mafafa, between the villages of Falelatai and Lefaga on the south-west coast of 'Upolu, was visited by Golson in 1957 and Green in 1966. A map of the site was produced by Green, and appeared in Volume I of Archaeology in Western Samoa (Golson 1969:17). This shows a flat topped hill with four main ridges leading down from the high point. Only one of these, the north trending ridge, was visited by the team, and a series of ditches, banks and terraces on this was mapped and described. This ridge was regarded as the site; the top and the other ridges were not visited and are not mentioned as being possibly associated with the defences seen.

The hill was examined during a three-hour period in 1992 and, despite being covered in a tangle of vegetation left by Hurricane Val, was found to have been modified for habitation and defence (Fig. 8). The top area had two slightly higher flattened points, and contained some low stone features. The edge of the top, in one place at least, had been built up by stonework. Only two of the ridges were examined; one was terraced, and the other terraced and ditched, the ditch 400 metres from the top. The site remains to be surveyed in detail, as many more features will undoubtably be present.

Mt. Vaea. Mt Vaea, in Apia, had also been visited by Golson and Green. This mountain rises up behind the town, which is now spreading back around it and up its slopes. The ridge that runs down north from the summit into Apia township had been recorded as having been terraced and scarped (Golson 1969:17).

The main ridges and spurs of Mt Vaea were briefly examined over two days in 1992 (Fig. 9). The two highest summits were found to have been modified for defence, and the access ridges to the mountain top terraced. The highest peak was strengthened on the north side with some large terraces and scarps, and one ditch (Fig. 10). Although the other summit had been extensively disturbed during construction of the communication mast, terracing and a small ditch were still visible. Various small knolls along the main ridge were also modified, and on the southern end of the mountain was a terraced complex which included a star mound.

A flat-topped knoll some distance below the summit, on the east side of the mountain, contains the grave of Robert Louis Stevenson, and is a popular visiting spot for locals and tourists (and at least one visiting archaeologist). This is a prehistoric house platform (see Fig. 9b), with terraces along the spur both above and below. A description of Stevenson's burial on December 4, 1894, makes reference to the flat area “…The summit is a room-sized platform, then tree-covered; it was soon cleared and the grave dug in the middle” (Furnas 1952:369).

Again, the surveyed map is preliminary only, and several weeks' work will be needed to produce a final version. At least one other hill adjacent to Mt Vaea is also fortified; this was described by Golson as having a flat area on top, defended by ditches and transverse stone walls, but with no rear defence. The - 416 contours of this small hill, which is off the south-west end of Mt Vaea, indicate that inland defences are present.

Figure 9. Mt Vaea fortified complex, Apia, 'Upolu, Western Samoa. Shaded features on contour map were surveyed; unshaded features show presence in area only.

Luatuanu'u. The fortification of Luatuanu'u was also revisited. This had been surveyed and excavated by Scott and Green in 1964 and 1965. Their map (Scott and Green 1969: opp. 205) appeared to contain some anomalies, in that the fort did not follow the same principles of construction as noted for similar sites in Fiji. The highest point on the site, which should have been level and well defended, was sloping at 8°, and the rear defences appeared to be insufficient, given their importance. These factors had led to a prediction that more of the site would be - 417 found further inland (Best 1992f:41-3).

Figure 10. Mt Vaea: main centre of defended complex.

This, however, was not the case (Fig. 11). The fortification was as surveyed by Scott and Green, but not as illustrated. A draughting error had resulted in the height difference between the two ends of the site being multiplied by a factor of 2.2, resulting in an exaggerated slope over its length. The rear defences also differ from what little was shown on the Scott and Green drawing, in that the ridge inland of the rear ditch is narrow, runs downhill at 10° for 65 metres, and falls away on each side to spectacularly steep valley heads. This situation renders it virtually impossible for attackers to carry the fort from the inland side, and it - 418 is certain that this narrow dipping ridge, not mentioned in the original report, was the main factor behind the choice of the site in the first place. The fortification as redrawn, thus, makes good sense in terms of the basic principles of hill fortifications.

Figure 11. Luatuanu'u fort, 'Upolu, Western Samoa. Inset shows original Scott and Green drawing.
American Samoa

The Western Samoa archaeological project is the standard reference work for any subsequent archaeological research in this Island group. As demonstrated above, however, the workers involved did not come to terms with the rationale behind inland fortified complexes and, thus, had some difficulty in locating and interpreting these sites. Other archaeologists in Samoa have followed the approach of Green and Davidson, and have encountered much the same phenomenon — large rambling unfamiliar inland fortifications — with much the same results.

Tatagamatau. The work by Leach and Witter on the rediscovered quarry of Tatagamatau on Tutuila is a case in point. Here, the researchers located and - 419 mapped a terraced spur that ran uphill for some 300 metres, with a small discontinuous ditch on the uphill end. Beyond this ditch, the spur continued to rise towards the main ridge above. Leach and Witter interpreted the terraced section as being the total site, describing it as a fortified quarry, which was illustrated by them in a three-dimensional sketch. Although they did locate earthworks at some distance further up the ridge, they did not consider the possibility that these were anything but either an outpost of the main quarry site, or a separate site altogether (Leach and Witter 1987:38).

Their interpretation of the site had been influenced by Green and Davidson's work in Western Samoa. Writing about the defences at the uphill end of Tatagamatau, Leach and Witter stated:

Viewed from the steep scarp on the uphill side, it presented the characteristic appearance of a ditch and bank fortification. Several others are known from the Samoan group, especially on 'Upolu. Most of these were designed to prevent access from the coast (Davidson 1974a:240). The Tataga-matau example joins the few which prevented access down the ridge from the centre of the island. In its discontinuous ditch consisting of a pit-trench combination, it has parallels at Mafafa … and Luatuanu'u (Leach and Witter 1987:38).

A survey carried out by the author, as a member of the Leach and Witter 1988 follow-up project, located and mapped a larger series of features, which were interpreted by him as a fortified complex (Fig. 12). Some objections to this have been raised by Leach and Witter, who preferred to keep their options open on the function of internal features such as the transverse stone walls, at least some of the terraces, and the large upper ditch (Best, Leach and Witter 1989:7, 9, 10). The relationship between the two main ridges and the final size of the site itself is also questioned (Leach and Witter 1990:55). Additionally, Leach has suggested recently that “…a well thought-out archaeological programme…” is needed before Tatagamatau can be shown to be a single fortification (Leach 1992:3).

As surveyed in 1988, however, Tatagamatau embodies the basic principles of hill-fort construction. The junction of three ridges had been chosen as the site centre, and the three ridges themselves were modified in order to give maximum protection to this centre. The highest point within the site is level, and contains not only the remains of at least one specialised stone structure not found elsewhere on the site, in the form of a housemound, but also an extremely large three-tiered mound in the very centre of the ridge junction. There are, undoubtedly, satellite areas of lesser importance along the two main ridges, and at least some of the terraces there would have the dual functions of habitation and defence, but the overall design of the site has one central focus.

This fortification, when reduced to its basic elements, can be directly

- 420
Figure 12. Tatagamatau fortified complex, Tutuila, American Samoa, with cross-section of central area.

compared to the very small and simple Fijian hill fort of VL13/3 described by Palmer (Fig. 7). Both are at the junction of three ridges, and the area at the junction point is both the highest and flattest on the sites. The two ridges that slope down from the central area have been strengthened by defences: the scarps and ditches in the small hill-fort are the equivalent of the terracing (and, to a lesser extent, the stone walls) strung out along the two main ridges on Tatagamatau.

The third ridge on both Tatagamatau and the small fort, unlike the other two, actually slopes down from higher country inland towards the site itself, with a - 421 natural saddle just before it rises to the high point. In both situations the saddle has been chosen as the place to position the defensive ditch (as discussed above, placing it further up the ridge, as Palmer suggested, would severely reduce its effectiveness). In addition, both sites have back-up defences for this ridge; in the case of the small site, another ditch and bank, whereas Tatagamatau has a large wrap-around terrace and the lower tiering of the mound.

Although Tatagamatau is some 30 times larger than VL 13/3, the same mental template underlies the construction of both. To suggest that the main ditch on the former does not relate to the site, or that one of the ridges may have functioned as a fortification in its own right, or that some of the terraces were a byproduct of quarrying for adze rock, is to ignore or be unaware of the basic principles behind the building of defended sites.

Fagasā. Another large adze quarry was discovered on Tutuila in early 1991 by David Herdrich, then Territorial Archaeologist on the island. The author was able to visit the site in late 1991 and, with the help of Herdrich, to map most of the quarry area. Although a contour survey of the quarry had already been carried out by C. Strecht, the finished map from this was not available at that time. As was the case with Tatagamatau, many of the features in the quarry area, such as terraces, were not associated with adze production. In addition, more terraces were seen on the ridge immediately above the site. Late in the afternoon of December 5, a very quick reconnaissance was carried out along the ridge top above the quarry, and several features were noted: terraces, mounds, and a ditch (and bank). Hurricane Val struck that evening, and it was not until July 20, 1992, that the author was able to return to Tutuila and complete mapping the quarry, and to survey the ridge and some of the spurs above.

Although it was possible at that time to survey adequately the eastern side spurs, the main ridge, between the large platform and terraces on the south end and the starmound on the north, was a tangled mass of hurricane debris. This was traversed, but none of the features noted previously could be found (and one of these was a series of terraces with two-metre-high scarps); those shown along there on the map were drawn from memory and a sketch by Herdrich made on the first visit, and are considered only approximate in detail and position. The spurs on the west side of the main ridge, additionally, have not been examined.

A part from the quarry area, the quality of the survey does not approach that of Tatagamatau, and many more features remain to be found and mapped, and those along the top ridge to be checked; however, the overall layout of the complex has been recorded.

The result was a situation comparable to that at Tatagamatau (Fig.13). A large fortified complex was strung out along the top of the main ridge and down the side spurs. The whole of this was cut off from inland by a ditch and bank, and - 422 some spectacular scarped slopes. The cliffed peak of Taumata behind these would also have been an obstacle to any attacking force from inland. At the base of one of the spurs was the quarry itself.

This stone working area also shares some features with those of Tatagamatau. The terraces on the spurs themselves are not a result of adze rock extraction or adze manufacture; as with Tatagamatau, there is as much, or more, stone working activity on the side slopes of these, where no terraces exist. On the uphill end of the quarry area, at a slope break, is a set of features resembling those described by Leach and Witter as preventing access from above. A platform has been levelled, some pits dug into this, and one access route formed between these, with possibly a second to one side. A small terrace overlooking these is also present at both sites. The only evidence of adze working above this on Fagasā (and, after Hurricane Val, pig rooting on the next series of terraces had turned over a considerable amount of ground) was a collection of small flakes within a low house platform on the next higher terrace, and these may have been recycled as 'ili'ili (personal communication, D. Herdrich).

At this site, too, the modified ridge incorporates the basic ground rules for such complexes. The highest part, at the southern end, has the largest levelled area and is also the most heavily defended (the scarps above the ditch are 10 metres high at 60°). A stone oval on the platform is also unique to this area. The ditch and bank, together with the huge scarps and Taumata peak on the main divide (the latter effectively making the inland approach to the site along the ridges extremely difficult), serve the same function as the large ditch and terrace on Tatagamatau: to protect the complex from inland attack. The remainder of the site, with its secondary peaks and terraced spurs, contains subsidiary units within series of defensive and/or habitation terraces.

Malaeimi Valley. In the course of field work and research in this valley, directly south of Taumata Peak, another fortification and some stone walling were found that illustrate both how close these sites can be one to another and also how distant are some of their associated defensive features.

This fortification runs up a narrow steep spur from the head of the valley to just below Taumata Peak (Fig. 14). The lowest defensive feature, 63 metres from the valley floor, combines a terrace and two short ditches with a central causeway. The rest of the site consists of 15 terraces, one of which is 153 metres long and 25 metres deep. The rear defences appear to be entirely natural, and consist of a c.50-metre length of crumbling knife-edge ridge, falling away each side at 60°, which runs up to sheer cliffs on the southern side of Taumata Peak. No way up these could be found and, indeed, the difficulty had in traversing the razorback was a lesson in the security of the rear defences.

- 423
Figure 13. Fagasā fortified complex, Tutuila, American Samoa, with cross-section of main area, and showing adze quarry workings.
- 424
Figure 14. Malaeimi fortified complex, Tutuila, American Samoa.

The internal organisation of this site is not obvious, as there is no clearly defined high point. At the highest place on the site is a 20 x 12 metre terrace, with a three-metre-wide extension running around to the east. Seventy metres back along the ridge is a level area 75 metres long and averaging about 10 metres wide. Either or both of these could have functioned as the centre of the site. The very large terrace half-way down the spur is also unusual.

In the course of two surveys along the valley floor, no evidence of prehistoric occupation was found. The ground is low-lying and floods after rain, and is, in

- 425
Figure 15. Spatial relationship of the Fagasā and Malaeimi fortified sites, and the Malaeimi Valley stone wall.

fact, the catchment area for much of Tutuila's water supply. At the mouth of the valley, however, the remains of a large stone wall were found, butting up to the base of the hill-slope on one side, and cut by the valley side road on the other. Although considerably disturbed, in places it was up to three metres high. The wall could not be followed across the valley, because of Second World War disturbance and plantation activities. There is a rocky gut at about the centre of the valley at this place, with some evidence of rock walling, and it is possible that the wall joined up with this.

- 426

The relationship of the last three sites can be seen in Figure 15. It is suggested that the wall was part of the overall defence system of the Malaemi valley, and would have been functionally connected to the fortification on the rear ridge just described. There would also have been fortified complexes along the top of the ridges to the east and west, as well as smaller units on spurs just above the valley floor (one of which was located during the surveys; see Best 1992e: 11-14).

Of interest is a comment that, at some time in the past, what were described as pig walls ran across the Fagasā valley mouth, approximately where the stoneworking area starts (D. Herdrich, personal communication). There is a large amount of rock walling around the village today, and it is possible that this is recycled material from similar defences that were once in place on this side too.

The proximity of the Fagasā and Malaeimi fortifications is also important. One objection to the main ditch on Tatagamatau being associated with the total complex is that there are terraces with mounds only 250 metres further up the ridge (Leach and Witter 1990:55). These two discrete sites, separated by Taumata Peak, are 380 metres apart.


This paper has put forward the hypothesis that hill-forts were built using commonsense principles related to the immediate environment, that they were internally stratified with regard to status, and that they could be extremely large and have associated features at considerable distance from the main site. What does this mean for the archaeologist?

Firstly, the interiors of both Fiji and Samoa are intricate archaeological landscapes, where the structures and hierarchy of prehistoric life are expanded and writ large on the topography. Instead of trying to unravel the layout of a coastal settlement at one point of time in its past, with the attendant problems of extensive multiple or permanent occupation, possible use as an agricultural plot, and gross disturbances from major erosional and depositional forces, the inland sites remain much as they were at the time of construction.

The large dispersed complexes that are characteristic of the rugged linear basaltic ridges of Samoa, and their limestone equivalents in Fiji, in comparison with the more compact forts on weathered volcanic hills (cf. those in Lau) can actually help the archaeologist to interpret the internal stratification between site areas. Lengths of ridge top too narrow for habitation separate raised high points that can be trimmed down to provide platforms, thus providing comparisons of height and size and, in Fiji at least, artefacts.

The large platforms which, for the Fijian sites, were suggested as having a ceremonial function, are also present on Samoan forts. In Western Samoa, the top of Mafafa (Fig. 8) is a very large level area, and the highest place on Luatuanu'u is also of some size (Fig. 11). In American Samoa, the large flattened - 427 hill on the south end of Fagasā (Fig. 13), the highest and most defended platform on the site, is of considerable size, and the flat ridge top on Tatagamatau (Fig. 12), adjacent to the highest mound, is a similar feature. These raised flat areas on the highest points of inland defended sites would appear to be equivalent to the ceremonial areas present in both Fijian and Samoan villages — the rara in Fiji, and the malae in Samoa — with both these the functional equivalent of the Polynesian marae (Hocart 1970:255, 256).

For Western Samoa, similar structures are found on the flatland coastal areas, and have been obliquely referred to as malae as, for instance, the Pulemelei mound on Savai'i, and other raised paved areas in the Islands(Scott 1969:89, 90). A reference by the missionary Platt, for February 3, 1836, somewhere a few kilometres north-west of Samatua on the west end of 'Upolu, refers to a pile of stones four to five metres high and several hundred metres long as “…a place most resembling a Tahitian marae of anything we have yet seen” (Platt MS 1836 first quoted in Davidson 1974c:205).

In a study of Samoan settlement patterns (Davidson 1969b), the possibility of demonstrating evidence for prehistoric political authority and social stratification was discussed, mainly as a response to the theories proposed by Ember (1966). The view taken in the article was decidedly negative. Green is quoted as having “…pointed out that such archaeological verification will be difficult if not impossible to obtain”, while Davidson suggests that “…it is not to be expected … that archaeology can help much in the dispute about kinship and political authority, unless in rare instances structures indicative of supralocal authority can be identified archaeologically” (Davidson 1969b:45). Davidson also says that “…the major points on which archaeological evidence can contribute are very marginal to the main lines of argument among anthropologists.” (Davidson 1969b:77).

As this paper has attempted to show, this is not the case. The archaeological evidence exists, spread over the surface of the inland hills, most of it available without the need for excavations. Archaeologists, however, will have to approach the problem of data recovery with some care.

Firstly, they must understand in detail what the architects of the inland settlements were attempting, with regard to the use of the local topography. They must think like the individual responsible for each fortification. Elsdon Best, undoubtedly the foremost authority on New Zealand fortifications, put it well when he wrote:

In studying the fortified positions constructed by the old time Maori, regularity of design must not be expected. The Maori engineer … selected a suitable site, then studied its contour lines and surroundings, until, with no help of pen or paper, he had grasped the solution of the problem, and worked out his plan. He then proceeded to mark the lines of scarp and fosse, of rampart and stockade, - 428 after which he set his men to work and superintended their labours. At certain weak places he threw in an extra stockade or fosse, or erected an elevated platform to command it. The entrance passage he laid off by narrow and tortuous ways flanked by strong defences. According to the contour of the ground, and relative levels, he devised a defence of scarp and stockade, or of fosse and parapet. He cunningly carved a hill into terraces of unequal sizes and levels, each of which became a defensive area in itself. The easiest approach to that fort possessed the strongest defences, and for an enemy to reach the summit area meant most strenuous fighting to reduce the various fortified subdivisions (Best 1975:21).

It is of considerable interest that, in a lengthy review of the above publication and of its influence over some 60 years on archaeological approaches to the subject of fortified sites in New Zealand (and to a lesser extent the Pacific), Davidson was able to state that the sections on the functional aspects of fortifications had been “…largely ignored by archaeologists…” (Davidson 1987:8).

Best apparently meant to travel to Samoa early this century and study its fortifications (Green 1967:97); had he done so, we might be at a different level of analysis by now.

Another knowledgeable observer on Maori fortifications at the same time as Best was Raymond Firth, and his description of the internal planning of such a site applies equally well to the Samoa/Fiji situation:

The sections of the fort occupied by different family groups were often partitioned off, while behind the innermost fortification on the highest ground was the citadel of the stronghold, the site of the houses of the principal chief and his relatives and the rallying point for all the people (Firth 1927:67).

Firth also recognised the importance of using slopes, even small ones: “…Sloping ground, even if only of gentle gradient, confers a great advantage both in the construction of defensive works and the repelling of an assault” (Firth 1927:68).

The effectiveness of slopes themselves as barriers was tested on a Maori fortification by Pearce (1977), who conducted elapsed time trials and weapon testing (sling stick and dart) on slopes ranging from 16° to 27°. He found that, although slopes of up to 18° could be stormed with some ease, the steepest (27°) was an obstacle in itself. This fits quite well with the Fijian and Samoan situation, where slopes of about 30° and over are not modified with earthworks. The effectiveness of the throwing weapons used by Pearce was considerably reduced from that obtained on the flat, even on the gentler slopes; the converse of this, for the defenders, means that such weapons would have been even more effective - 429 than normal. Pearce's conclusion, that most such fortifications were likely to have fallen only through tactics other than direct assault, would also apply to many of the Fijian and Samoan sites.

Failure to understand and to apply the basic principles outlined above can, as shown, result in only a portion of the site being located and described. In addition, if other activities have taken place on or near some of the features, or are a by-product of their construction, then a totally incorrect functional interpretation of an incompletely surveyed site may well result.

Survey strategies must be tailored to examine the whole of the inland landscape. Although the word “intensive” is often used to describe site surveys in Samoa (e.g., Davidson 1969a:186), there are degrees of intensity. The author has surveyed some six fortifications in Samoa, but only one of these, Tatagamatau, can be said to have been adequately covered. At that site, all side spurs, however small, were followed down, most of the flanks were traversed where the slopes permitted, and all slope angles were recorded. There is no doubt that the detail within the site, and its extent, is well understood.

It is instructive to trace the sequence through which Tatagamatau passed on its way to identification in full. The first reference appears to have been that of Kikuchi, mentioned above, who was told by informants of a large ditch in an area that was subsequently found to correspond with the north-east end of Tatagamatau (also in Best et al.: 1989:9, 10). This is most probably the upper ditch on Tatagamatau, which was classified by Kikuchi as a fortification in its own right. The next stage was the identification by Leach and Witter of the lower part of one of the south-west spurs as a fortified quarry. It was only when the total ridge landscape was intensively surveyed that these isolated features or groups of features could be shown to belong to a single large, but discrete, fortification.

The use of historical or traditional information in the search for prehistoric inland fortifications, which may be up to 900 years old, is of doubtful value. After the Western Samoa Archaeological Programme had finished, Davidson could still write “…there is no strong historical or traditional evidence to suggest that inland settlement was related to warfare, apart from modern stories about Tongan invasions” (Davidson 1969b:51). And yet, as we have seen, the hills at some stage must have been full either of hostiles,or of the fear of them. Although it makes sense for the archaeologist to be directed to known sites as a starting point in any survey (although the final size of these may differ from the local estimate), it is dangerous to make generalisations on the entire archaeological landscape based on traditions (see, for instance, Hunt and Kirch 1988:166; Green 1974:141; Davidson 1974b:241).

The use of more sophisticated techniques, such as remote sensing, is probably also of questionable value in the Samoan situation, given the vegetation cover. Even where it has achieved spectacular results, as in the Fijian aerial - 430 photogrammetric surveys by Parry, it is still uncertain just how much of the archaeological landscape on the inland hills was located. In the Sigatoka survey, Parry claims a success rate of up to about 70 per cent (Parry 1987:57, 58), but this is based on the number of sites located by Palmer, who also used aerial photos, together with local traditions (Palmer 1967:13-5). Any site which did not show up in either, and these are likely to have been of some antiquity, or sited along high rocky ridges under tall canopy forest, is absent from the equation.

Although technology that was undreamed of a few years ago, such as the Global Positioning System and the Geographic Information System, is now available to the archaeologist, the use of these tools to record and analyse field data will produce worthwhile results only if the parameters of the archaeological landscape are understood and employed while gathering the data itself.

Ultimately, the entire landscape must be intensively searched, and the only technique that will produce adequate results is going to be old-fashioned leg-work. Unfortunately, the apparent increase in the frequency of tropical hurricanes and the resulting fallen vegetation will make this extremely difficult; perhaps some kind of “slash and burn” archaeology would be appropriate here (see Best 1989:116), for the first stage at least. Despite the drawbacks, however, the rewards stand to be considerable.

Eastern Tutuila and Manu'a

As we have seen, the initial archaeological surveys in Western Samoa did not always locate the total site. The situation may be similar in more recent work in American Samoa. Tatagamatau has already been mentioned, and surveys in other areas of Tutuila may also have experienced the same situation.

As mentioned earlier in this paper, the research of Clarke and Herdrich in Eastern Tutuila located a large number of sites on the inland hills. Their survey strategy consisted, in the main, of walking the ridges only; side spurs and slopes (especially the higher parts of the latter) were not covered to the same extent (Clarke and Herdrich 1993:153). As Clarke rightly points out, theirs was a reconnaissance survey, and:

the aim of reconnaissance survey is to provide an overview of the distribution and types of archaeological sites found in a given area. To precisely measure and accurately map all sites and features — which would often require extensive vegetation clearing — would take a substantial amount of time and would greatly limit the area of land that could be covered (Clarke 1989:8).

As in the Western Samoa Archaeology Project surveys, the features found were given separate site numbers and treated as discrete entities. The functions of terraces found in the first season were described as either agricultural, as - 431 resting places (mālōlōga), or as refuges (Clarke and Herdrich 1988:67, 71). After the 1988 season, one of the complexes of terraces and mounds previously surveyed was reclassified as a fortified complex, and one other such complex and one smaller fortified site were found. Even then, terraces associated with the complexes are described as only “…highly unlikely to have been features primarily associated with agricultural, residential, or mālōlōga activities” (Clarke and Herdrich 1993:176).

In addition, four other defensive features were recorded: one ditch, and three mound-ditch combinations (Clarke 1989:145). The enigmatic and captivating “starmounds” played a large part in the survey, with considerable space devoted to their description and analysis both in the survey reports and later (Clarke and Herdrich 1988: Herdrich 1991).

On the basis of the two inland complexes surveyed in central and western Tutuila, and on the evidence from the resurveyed sites on 'Upolu, it is suggested here that many of the features located in eastern Tutuila by Clarke and Herdrich are part of similar large and diffuse, although recognisable, defended complexes. As an example, of the nine starmounds located by the author, all have been situated at specific points within such sites. Of the 62 described by Clarke and Herdrich, only about four or five are associated with what are described as larger sites (Clark 1989:21, 22; Clarke and Herdrich 1993:163, 164). Although starmounds are said to be the most common site in the inland hills of eastern Tutuila (Clarke and Herdrich 1993:173), an examination of the survey reports shows that terraces outnumber them by 2:1. In the central and western surveys conducted by the author, however, this figure is about 15:1 (see Best: in prep.).

Although agreeing in part with Davidson (1969b:61, 77) that prehistoric settlement in Samoa was probably more dispersed, with a larger inland component, Clarke does not see this as the case in American Samoa, suggesting the more rugged topography as the reason (Clarke 1989:137; see also Clarke and Herdrich 1993:178). It is, however, just this system of razor-back ridges and isolated peaks that was chosen in Fiji, Western Samoa and the rest of Tutuila for one kind of inland occupation, and this surely applies to the eastern districts as well. In addition, at least two of the Tutuila sites situated in this environment, and originally identified as fortified settlements by Frost (Lefutu and Mt Alava; see above), and reinterpreted as undefended by Clarke and Herdrich (1993:177), should be reconsidered in the light of this argument as either fortified sites in their own right or as possible parts of larger complexes.

Surveys in the Manu'a group have also failed to locate evidence for defended sites. On 'Ofu, however, terraces in the bush on the east side of Tia ridge, and some stone edging on at least one area on the ridge itself, found by the author in 1991, indicate that they may be present.

On Ta'ū, the feature known as the Sacred Wall of Vaovasa, on the north side - 432 of the island and above the sea -cliffs, was also visited in 1991. This had been first recorded by Clarke in 1980, and was later identified by Hunt as the front of a terrace (Clarke 1990:12). This large stone-faced terrace is about four metres high and 30 metres long; on its surface are some stone arrangements and, according to Hunt, further features occur at the back, where it meets the hill slope. Although Clarke identifies this feature as a probable mālōlōga or resting place (Clarke 1990:13), the structure is almost certainly defensive, and may also be part of a larger complex. The houses that La Pérouse saw nearly 200 years ago along this coast may have been associated with this or similar settlements.

The hills further inland on Ta'ū are apparently seldom visited today by residents; only one man was said to know the way up to the highest point; and they have certainly never been part of any intensive or even extensive archaeological survey. The question of whether fortifications exist there still has to be answered.


Although the sequence of settlement in Samoa is still to be studied in any detail, some predictions can be made, based on research results from Fiji. Firstly, there is likely to have been considerable contact between the two countries over the 3,000 years of occupation. Their ceramic sequences run in tandem until Samoa became aceramic some 2,000 years ago. The pot shapes, the rim forms, and the occasional carved paddle impressed sherd at about that time are the same for both areas (Best 1984:654). Even the use of a coarse lithic temper appears towards the end of the period, at about 2,000 B.P. (Although there is now some discussion with regard to the relationship of the pottery sequences between Fiji and American Samoa at least, I believe that future research will show that they are in fact similar.)

The settlement pattern of the Lau group, based on the distribution of 193 sites, shows two periods when the population moved inland, or at least maintained inland hilltop refuges or fortifications. These were between 2,500 and 2,100 B.P., and from 950 to 200 B.P. (Fig. 16). Although the same hilltops were often used in both periods, the first occupation appears not to have involved earthworks. This is a similar finding, in dates, pottery types and defended/undefended status, to that of Frost for Taveuni, and the situation probably holds for the larger Fijian islands as well.

It is possible the same pressures that sent the Lauans inland were also working in Samoa. If the ceramic sequences for the two groups are indeed parallel, then the potsherds scattered about on the inland hills, of Tutuila at least, point to an early use of that terrain.

The second retreat inland may also have been made at much the same time in both island groups. It would, at any rate, seem logical to suppose that these

- 433
Figure 16. Periods of inland hilltop settlement on Lakeba, Lau Islands, Fiji. Circles denote sites on volcanic hills: crosses are those on limestone.

huge artefacts were built over a discrete period in Samoa's prehistory, although it remains to be established whether this is the same as that for Fiji.

The fort of Ulunikoro in the Lau group is well dated; the pooled mean for eight dates, taken from the living terraces on the lower part of the main ridge, is 969±20 B.P. (University of Washington Radiocarbon Calibration Programme). The earliest evidence for the use of Tatagamatau as a fortification comes from three dates of c.600 B.P.; although an earlier date exists (906±157 B.P.), this may be associated with a structure that predates the fortifications themselves. Janet Frost, for the fortified site of Lefutu on Tutuila, obtained a single date of 770 B.P. (Frost 1978:241).

The other dated fortification in Samoa has already been mentioned: that of Luatuanu'u on 'Upolu. Its date of c. 1,500 B.P. comes from a sample of dispersed charcoal in bank fill, and, as the excavators point out, “…a positive association between the event (construction of the bank) and the nonartifactual sample (the charcoal included in it) cannot be shown” (Scott and Green 1969:208). An - 434 earlier date from the area, of 2,170±90 B.P., is presented as evidence that this age may, in fact, be reliable. However, this second date comes from a firepit dug into the natural beneath a terrace and starmound; other natural hollows in this layer were also filled with charcoal. These the excavator describes as “…suggesting that a burn-off might have taken place as the initial event in the first phase of the occupation sequence, followed by the use of the area before the construction of the terrace” (Peters 1969:216). Both early dates, thus, appear to be associated with the clearing of bush from relatively gentle hill slopes, and are probably associated with agriculture rather than with any occupation that included the construction of a large hill-fort.

It can no longer be tenable to describe Pacific fortifications, and those of Samoa in particular, as mere “refuges”, retreats to be used in times of emergency, with little or no attempt at permanent defences. And yet, in discussing Best's attempts to identify the “homeland” of Maori , Davidson was able to write, in 1987:

His information about Samoa was very poor, but I feel sure that if he had known what we now know about prehistoric Samoan fortifications he would probably have rejected them too, because they were primarily refuges. Buck also took this position. Other Polynesians had elements of fortification such as fences, fighting stages, and even ditches, but their settlement patterns were different and they did not have fortified villages… we have to admit that paa in New Zealand occupied a different role in the settlement pattern than did most Polynesian refuges (Davidson 1987:20).

As this work has shown, Samoa and Fiji, at least, have fortified complexes equal in size and complexity to any Maori . Their role in society would also have been similar, described by Groube for Maori as “…the centre of a more extensive settlement pattern of which it is the citadel…” (Groube 1965:52). Lyth's observation, cited above, on the social divisions within the fortified town of Tubou on Lakeba, to which the six smaller settlements on the island retired when the whole island was threatened, is an illustration of this (Lyth MS 1850-1: general notes).


Pacific fortifications are often regarded as an idea or artefact subject to the same dispersive forces that apply to other smaller objects (e.g., Green 1967:109). An objection to this on the grounds that forts are communal artefacts (Groube 1970:133) is invalid; the final form of a fort owes as much to its workers as the excavation strategy of Sir Mortimer Wheeler at Mohenjo-Daro did to the whims of his Indian labourers. On the other hand, as must be clear from this article, fort builders will come up with the same answer, given the same topographical restraints, even when separated by half a world. It is of little use to speculate on - 435 one Island group deriving the idea from another, unless it can be shown that the form of a fort, or its position in the landscape, resembles those in the other group because of a deliberate choice on the part of the builder.

Such a situation can be suggested for one type of fortification on Lakeba. These are the limestone ridge forts, and their position in the landscape, together with associated sites such as habitation and burial caves, as can be seen in Fig. 16. All these are close to the coast, a factor of the distribution of the steep limestone ridges on which they were constructed. The fortifications are linear and consist of stonework in the form of limestone scree blocks. They appear to be the first fortifications on the islands and, among the artefacts found on them — and only on them and undefended sites of the same age — are adzes which have been sourced to Tutuila, American Samoa (Best 1984:403-5, Best et al. 1992). It would seem that there was some external influence involved in the construction of this type of fortification, although the source of any such influence, and its extent, are not known.

Samoan forts are invariably described by the early observers as having been constructed by invading Tongans, based on the traditions of the Samoans themselves. There are, nevertheless, some problems with this interpretation.

'Upolu and Tutuila are very similar islands, separated by 60 kilometres of sea, and with similar traditions of Tongan influence. Only on 'Upolu did these traditions refer to Tongan involvement with fortifications, and only on 'Upolu were these fortifications found. It has become clear in the last five years, however, that Tutuila also has a considerable number of these sites. The conclusions are inescapable: the 'Upolu traditions have followed the discovery of the sites themselves and are rationalisations for the strange features that have appeared in the landscape. Moreover, all the early fortifications described above are in the western end of 'Upolu; this is both the area of easiest inland access, and the focus of initial European settlement. No forts had been reported for the more rugged eastern end of the island, nor for the whole of inhospitable Savai'i, nor, of course, for the deeply dissected landscape of Tutuila.

This is not to say that the Tongans were not involved in their construction, merely that there is no evidence from oral traditions to suggest that they were. Clues to the probable builders of the sites must come from elsewhere.

The Lau group, a mere 300 km west of Tonga, was under the domination of warriors from there at the time of European contact. At one stage, in 1859, the Tongans under Ma'afu are thought to have been within three months of conquering the entire Fiji group (Best 1984:657). This is unlikely to have been a unique event in their histories, and yet no known fortification in Lau at least is attributed to the Tongans and, indeed, most have their own traditions of local origin. An observation by Lyth for Namalata on the island of Kadavu is relevant here: - 436

Namalata consists of two towns—one inland occupied by Feejeeans, the other on the beach by Tonga-Feejeeans. The former is a well fortified Feejeean town, the latter fronts this beautiful bay and has a fine sandy beach. Being Tonguese they are not afraid of being attacked by the Feejeeans by whom they are universally feared. They have no war fence around their town. (Lyth MS Journal 1848: entry for November 3).

It would seem that, in at least some situations, the Tongans felt no need for a defended site, although Lyth records elsewhere that one of the missionaries' daughters on Lakeba fell into the moat “…surrounding the Tonga town…” (Lyth MS 1850-1: entry for April 6, 1851).

The large number of fortifications in Samoa and their apparent widespread distribution over the landscape would seem to be more a result of internal strife in which the Tongans may, at times, have joined, rather than any total domination by an outside force. This is one of the conclusions that Davidson reached, through a different route, when discussing the Tongan influence on 'Upolu. Using early missionary observations, she described the Tongan involvement in the wars of the late 18th and early-19th centuries as “meddling in conflicts in which the Samoans were engaged”, and suggested that this was probably also the situation in the more remote past (Davidson 1974b:242).

Such internally inspired conflicts often sprang, in Fiji at least, from the development and maintenance of a hierarchical social system. This paper is not concerned with the root cause or causes behind any such significant change in the scale of society, in either of the two Island groups in question. Although this is likely to have involved pressure or perceived pressure on ecological or social resources, it is the physical manifestations of such a change, and its social consequences, that are of interest here.

In the Fijian situation, three types of conflict have been described (Clunie 1977:4). The first is i valu yasa or “stealthy” war, sometimes consisting of “almost Lilliputian campaigns between near neighbours” (Clunie 1977:4). Although this was, in the main, guerilla warfare, it could result in considerable loss of life, and was often a preliminary to the next and more formal conflict. Even this type of petty skirmishing has been shown, for Fiji, to provide a system for consolidating the power of the rulers involved, in such areas as maintaining internal unity or team spirit, providing rites of passage for warriors, leaving no idle hands to make trouble, and enforcing the importance of the gods (Tippett 1968:59-66). Some of the smaller Fijian forts, and some minor skirmishes, may have resulted from such mundane reasons as the visit of the taxman. Thomas Williams, in 1846, had this to say on the island of Kanacea:

Ascended the little mountain on a point of which the town is built. It is common with the people in islands of low rank to select the most difficult part of the - 437 island and that most easily defended on which to build their town or towns. By this they avoid some of the evil resulting from the unreasonable and oppressive conduct of their chiefs who ever and anon visit them (in Henderson 1931:364).

The other two types of warfare were between chiefdoms (i valu ni tu: state wars), or confederations of these (i valu rabaraba: widespread wars) (Clunie 1977:4). Even these could be mainly showpieces, with few people killed, but the potential for catastrophic slaughter was always there.

The sparks that ignited such hostilities were many and varied. The missionary Waterhouse gives reasons that include land, women, the commission of murder, personal affront to a chief, the unlawful eating of a turtle, a tabu violation, and the refusal to part with a certain club, bird or shell (Clunie 1977:4).

Examples of two of these classes of war were recorded in Lakeba by the Wesleyan missionaries. An intervillage battle between some of the smaller settlements on the north coast of the island, caused by one side eating the sacred coconuts of the other, resulted in 10 people being killed and many other individuals wounded; included in these were two chiefs, one from each side. Permission to continue the fighting was sought from the paramount chief at Tubou, and this was granted, with the order that any subsequent dead be brought to “…the principal town…in order that they may be baked and eaten, on account of the famine which exists in the land” (Calvert 1838-48: entry for September 26, 1839).

Slightly different was the battle between the principal town of Tubou itself and that of Waciwaci on the north coast: this was a quarrel between the paramount chief's son, Vuetesau, and the Waciwaci chief Zepheniah Lua, and resulted in one man being shot (Lyth MS 1845-8: entry for August 13). Lua was, in fact, an enemy of the paramount chief himself, and at some time plotted to kill him (Lyth MS 1850-1: general notes).

Less than a year later, these two adversaries, faced with a threat from a greater power from outside the island, were on the same side, and indeed ready to share the same fortification. As Lyth records:

We sat conversing…on the subject of war from Bau, the subject that at present engages all the attention of this people…they [Waciwaci] are strengthening themselves to be ready for war, & Zepheniah has acceded to the kings request for him & his people to remove into the Tubou fortress. (Lyth MS 1845-48: entry for June 12, 1846).
A similar threat to Lakeba from Bau in 1842 had developed because a large canoe intended as tribute had not been finished (Calvert 1838-48: entry for October 22, 1842).

- 438

Even though the types of warfare described above may not have resulted in many casualties, the person of the chief was not only not exempt from danger but was often a prime target. Although in an office whose holders are regarded as having near-divine qualities, and engaged at the time in an activity seen as almost sacramental (drinking yaqona), the first Tui Nayau was, nevertheless, clubbed down through treachery (Reid 1977:14). This was not an isolated instance; Lyth states that“…several of the former kings of Lakeba, the Tui Nayaus of their day, were variously killed” (1850-1: general notes). One of these was cut down by a Tongan chief for failing to maintain the food supply to his people.

The construction and layout of large fortifications on the rugged ridges and peaks of inland hills of both Fiji and Samoa would have been neither entirely symbolic nor a result merely of grim necessity. They were examples of public works, reflecting the status of the chief involved, and yet designed and constructed with the greatest care, as the life — or, at the very least, the way of life — of everyone concerned depended on the efficiency of the design.

It is suggested in this paper that the inland archaeological landscape of Samoa is the result of a internally driven change in the social structure of the Islands. This is unlikely to have been primarily due to any mere onset or escalation of hostilities themselves. Warfare and some kind of defensive reaction, such as that of the Vuna people, who, when threatened with an attack by Bau, said “…our legs are our fence” (Hunt MS 1839-41: entry for October 29, 1840) were probably present in the islands before the earthworks under discussion, as they were for Fiji, at least in the Lau group. The stimulus behind the large, well-constructed and labour-intensive fortified complexes is more a result of some change in the structure of Samoan society. Although forces external to Samoa may have been involved, their influence would have been tangential to the main theme.


Since fortifications in at least some of the high rugged vegetation covered islands described above appear to be underrepresented in the archaeological record, one can speculate whether the same situation exists in similar situations elsewhere.

The Tongans have appeared throughout this paper as being involved, in traditions at least, with fortifications in both Fiji and Samoa. Since it is suggested that the development of fortified sites in these two island groups may have followed the same sequence, the role of fortifications in the Tongan chain, situated virtually between the two, is of interest. Over 40 have been recorded: 34 from Tongatapu (Spennemann 1989:477-84), and the remainder from the Vava'u and Ha'apai Groups (McKern 1929:80-9; Davidson 1971:35), and one from 'Ata (Anderson 1978:6). These are all encircling ditch and bank sites, on - 439 flat or slightly raised land, and all but one are said to belong to the last phase of civil warfare in the Islands; the exception, Mu'a, may be some 600 years old (Spennemann 1989:481). Spennemann suggests that sites of equal or earlier age may have been destroyed by erosion and gardening.

No hill-forts have yet been recorded, even in the higher islands where they might have been expected to occur (see Davidson 1971:35 for Vava'u). However, an observation by the Rev. Nathaniel Turner, in October 1840, on the island of Mago, suggests that they may be present given the appropriate topography: “…The fortification wh [sic] they had in time of war was on the top of a hill, and most difficult of ascent. It was thought impregnable. There are still remaining marks of 3 fortifications” (Turner 1839-40:63). He also added the interesting comment that “…Finau the king of Vavau, whose authority extended to the Haabies also, put some Feejians on this island to defend it from the many hostile attacks from Tongataboo” (Turner MS 1839-40:67).

Although the full range of Tongan fortification types, and their dates, still remains to be investigated, upland areas may repay investigation, since the flatland examples appear to have been largely disturbed or destroyed.

For the Cook Islands, the high broken interior of Rarotonga has apparently not been systematically surveyed by archaeologists. One terraced fort is traditionally known from there, on Mt Piako (Duff 1974:24; Trotter 1974:81), but this was not visited by these authors, and the inland hilltops have not been covered in more recent work on the island (e.g., Walter 1990).

The inland archaeological landscape of some of the high islands in the Hawaiian Group, more the domain of hardy pig-hunters than archaeologists, may also contain more features associated with defended occupation than at present realised. On Kaua'i the modified flat tops of hills in defendable positions, even below the cloud line, may fit in this category (e.g., the three platforms and possible terrace of Kaukaopua in Hanalei valley), whereas features such as the ridge notches described in Emory (1924:75) and McAllister (1933:35 and Pl. 11 A) for Lanai and Oahu, and reported elsewhere (Green 1967:103), may be part of larger complexes.

Even in the green and well-studied hills of New Zealand, it is certain that the larger social units have not always been recognised as such on the ground, given the same topographical situation of long linear broken ridges.


In this paper, fortifications from Fiji have been compared with those of Samoa, not in any cultural dimension, but by an approach that examines both the defensive structure and function of the sites themselves, and which sets out some basic parameters by which such sites can be studied. The data from one small island in the Lau Group, Fiji, have been used as a basis to suggest the presence - 440 and form of fortified complexes in Samoa. Examples of Samoan complexes have been located and surveyed, and others resurveyed, and predictions made concerning the archaeological landscape of these islands.

It is suggested that, due to the constraints of the topography, the hierarchical social relationships that existed within the societies concerned are reflected in the physical layout of their hilltop fortifications.

To find, record and interpret this information is a task of some magnitude. The data, however, are unlikely to be recovered by any other approach, and should throw light on aspects of society hitherto not obtained by archaeologists working in these Island groups.


The data used in this project have been collected over a number of years. For the early part of this, I wish to thank again all those who contributed to this area of research in my Ph.D. thesis, especially Amaleki Tagici, my survey partner on Lakeba.

I also wish to acknowledge my debt to the late Garth Rogers, who helped both in the field and also during subsequent research.

The Department of Anthropology, University of Auckland, provided the institutional support for the 1991-2 survey permit required by the Western Samoan Government. The Department of Conservation, Auckland, also gave assistance at various stages.

I owe thanks to Roger Green, who influenced the project throughout.

In American Samoa, I am especially grateful to David Herdrich, for help and advice in the field.

The Historic Preservation Office, Department of Parks and Recreation, Pago Pago, kindly met accommodation expenses incurred while mapping the quarry area of Fagasā, and has always provided equipment, information and assistance during this and other projects.

For help in obtaining permission to survey the head of the Malaeimi valley and the Fagasā ridge, I thank David Herdrich, Mr Lagafua of Samoan Affairs, and the Rev. Fa'atui Laolagi (who also lent me his pickup truck).

For his unfailing good humour while being dragged through hurricane-felled bush, I thank Danny Tosh.

In Western Samoa, I should like to thank the following: Dr Kilifoti Eteuati, Secretary to the Government, Prime Minister's Department, for permission to undertake the site survey; Dr Fanaafi le Tagaloa, National University of Samoa, for support during the project; Mulipola Apelu, for his help in getting permission to visit the sites from the villages concerned; Sam Sesega and Roger Cornforth, of the Department of Lands and Environment, for arranging transport to and from Mafafa; Harry Brown, for contacts.

Funding for all field work after 1988 (except for the Fagasā quarry part of the fort) has been at the author's own expense.

I should like to thank Sue Bulmer, Geoff Irwin and Fergus Clunie for reading and commenting on the paper. Fergus also generously provided unpublished references and information from his own field work. The responsibility for the finished product is, however, mine.

- 441
  • Anderson, A. J., 1978. Archaeological Explorations on 'Ata Island, Tonga. Lau-Tonga 1977, Bulletin of the Royal Society of New Zealand, 17:1-21.
  • Babcock, T. F., 1976. An Analysis and Reanalysis of Taveuni Pottery. Unpublished M.A. thesis, Eastern New Mexico University.
  • Bellwood, P., 1979. Man's Conquest of the Pacific. New York: Oxford University Press.
  • Best, E., 1975. The Pa Maori. Reprint of 1927 edition. Bulletin 6. Wellington: Dominion Museum.
  • Best, S. B., 1984. Lakeba: The Prehistory of a Fijian Island. University Microfilms Inc. (1989) Ann Arbor, Michigan.
  • —— 1989. Archaeology of the Lapita Cultural Complex: a Critical Review. Archaeology in Oceania, 24:116.
  • —— 1992a. Archaeological Reconnaissance Survey and Excavation, Ofu-Olosega Highway Link. Report prepared for Hastings Associates Inc., Pago Pago, American Samoa.
  • —— 1992b. The Ta'u Road Archaeological Project, Phase 2, Part 1. Survey and Evaluation of Selected Locations for Proposed Drainage Works. Report prepared for Pacific International Engineering Ltd., Pago Pago, American Samoa.
  • —— 1992c. The Tafuna Plain Sewer System, Phase I, Archaeological Investigation of the Surveyed Route. Report Prepared for the American Samoa Power Authority, Pago Pago, American Samoa.
  • —— 1992d. The Aunu'u Island Sewer Project, Archaeological Survey and Excavation. Report prepared for the American Samoa Power Authority, Pago Pago, American Samoa.
  • —— 1992e. The Malaeimi Archaeological Survey, Tutuila, American Samoa. Report Prepared for the South Pacific International Christian Center, Inc., Pago Pago, American Samoa.
  • —— 1992f. Fortifications in Fiji and Samoa: Comparisons and Predictions. Archaeology in New Zealand, 35:40-44.
  • —— 1992g. Prehistoric Fortifications in West Samoa. A Re-examination of Selected Sites. Report Prepared for the Government of Western Samoa, Apia.
  • —— In press. A Short Site Survey in Western Samoa; Preliminary Results. Archaeology in New Zealand.
  • —— in prep. Samoan Starmounds; a Reinterpretation.
  • ——, H. Leach and D. Witter, 1989. Report on the Second Phase of Fieldwork at the Tataga-Matau Site, American Samoa, July-August 1988. Dunedin: Department of Anthropology, University of Otago.
  • ——, P. Sheppard, R. C. Green, and R. J. Parker, 1992. Necromancing the Stone: Archaeologists and Adzes in Samoa. Journal of the Polynesian Society, 101:45-85.
  • Buck, P. H. (Te Rangi Hiroa), 1930. Samoan Material Culture. Bulletin 75. Honolulu: B. P. Bishop Museum.
  • Buist, A. G., 1969. Field Archaeology on Savai'i, in Green and Davidson (eds), Archaeology in Western Samoa, Volume I. Report 3, Bulletin 6. Auckland:
- 442
  • Auckland Institute and Museum, pp.34-68.
  • Calvert, J., MS 1838-48. The Journal of the Rev. James Calvert, March 15, 1838 - July 1848. MF 37, FM4/1426. Adelaide: Barr Smith Library.
  • —— MS 1860-3. The Journal of the Rev. James Calvert, December 12, 1860 - January 7, 1863. Habits, Manners and Customs. MF 37, FM4/1426. Adelaide: Barr Smith Library.
  • Cargill, D., MS 1832-8. The Journal of Rev. D. Cargill. Sydney: Mitchell Library.
  • Churchill, L. P., 1902. Samoa Uma, Where Life is Different. New York: Forest and Stream.
  • Churchward, W. B., 1887. My Consulate in Samoa. London: Richard Bentley and Sons.
  • Clarke, J. T., 1981. Archaeology in American Samoa, in A Zonal Coastal Management Atlas of American Samoa. Department of Planning Office; American Samoa Government, and Department of Geography. Honolulu: University of Hawai'i.
  • —— 1989. The Eastern Tutuila Archaeological Project: 1988 Final Report. Unpublished Report for the Government of American Samoa. American Samoa: Office of Historic Preservation, Department of Parks and Recreation, Pago Pago.
  • —— 1990. The Ta'u Road Archaeological Project, Phase 1 Survey and Test Excavation. Report prepared for G.M. Meredith and Associates, Western Samoa. American Samoa: Office of Historic Preservation, Department of Parks and Recreation, Pago Pago.
  • —— and D.J. Herdrich, 1988. The Eastern Tutuila Archaeological Project: 1986 Final Report. Unpublished Report for the Government of American Samoa. American Samoa: Office of Historic Preservation, Department of Parks and Recreation, Pago Pago.
  • —— 1993. Prehistoric Settlement System in Eastern Tutuila, American Samoa. Journal of the Polynesian Society, 102:147-85.
  • Clunie, F., 1977. Fijian Weapons and Warfare. Bulletin 2, Suva: Fiji Museum.
  • Crosby, A., 1988. Beqa: Archaeology, Structure and History in Fiji. Unpublished M.A. thesis, University of Auckland.
  • Davidson, J. M., 1969a. Survey of Sites and Analysis of Associated Artifacts, Luatuanu'u, in Green and Davidson (eds) Archaeology in Western Samoa, Volume I, Report 12, Bulletin 6. Auckland: Auckland Institute and Museum, pp.185-204.
  • —— 1969b. Settlement Patterns in Samoa before 1840. Journal of the Polynesian Society, 78:44-82.
  • —— 1971. Preliminary Report on an Archaeological Survey of the Vava'u Group, Tonga, in R. Fraser (ed.). Cook Bicentenary Expedition in the South-West Pacific. Bulletin of the Royal Society of New Zealand, 8:29-40.
  • —— 1974a. Site Surveys on 'Upolu, in Green and Davidson (eds), Archaeology in Western Samoa, Volume IIReport 36. Bulletin 7. Auckland: Auckland Institute and Museum, pp. 181-204.
  • ——, 1974b. Samoan Structural Remains and Settlement Patterns, in Green and Davidson (eds), Archaeology in Western Samoa, Volume II. Report 39. Bulletin 7. Auckland: Auckland Institute and Museum, pp. 225-44.
- 443
  • —— 1974c. Specialised Sites in the Upper Falefa Valley, in Green and Davidson (eds) Archaeology in Western Samoa, Volume II. Report 37. Bulletin 7. Auckland: Auckland Institute and Museum, pp. 205-9.
  • —— 1987. The Paa Maaori Revisited. Journal of the Polynesian Society, 96:7-26.
  • Derrick, R. A., 1957. A History of Fiji. Suva: Government Press.
  • Dodge, E. S., 1972. A William Lockerby Manuscript in the Peabody Museum of Salem. Journal of Pacific History, 7:182-8.
  • Duff, R., 1974. Physiography of Rarotonga, In Trotter (ed.), Prehistory of the Southern Cook Islands. Bulletin 6. Christchurch: Canterbury Museum, pp. 22-7.
  • Eagelston, R. MS 1831. Log of the Peru. Copy of typescript. Adelaide: Barr Smith Library.
  • Ember, M., 1966. Samoan Kinship and Political Structure: An Archaeological Test to Decide Between the Two Alternative Reconstructions. American Anthropologist, 68:163-8.
  • Emory, K. P., 1924. The Island of Lanai: A Survey of Native Culture. Bulletin 12.Honolulu: B. P. Bishop Museum.
  • Erskine, J. E., 1853. Journal of a Cruise among the Islands of the Western Pacific. London: John Murray.
  • Firth, R., 1927. Maori Hill-forts. Antiquity, 1:66-78.
  • Foster, R., 1991. Faleafao Relief Harbor Archaeological Survey, American Samoa. Report prepared for the Department of Public Works, American Samoa Government, Pago Pago, American Samoa.
  • ——, 1992. Archaeological Monitoring of the Faleafao Relief Harbor Construction Project, Stage 1. Report prepared for McConnell Dowell, Tafuna, Pago Pago, American Samoa.
  • Fox, A., 1976. Prehistoric Maori Fortifications in the North Island of New Zealand. Monograph No. 6., New Zealand Archaeological Association. Auckland: Longman Paul.
  • Frost, E. L., 1974. Archaeological Excavations of Fortified Sites on Taveuni, Fiji. Asian and Pacific Archaeology Series, No. 6. Honolulu: University of Hawai'i.
  • Frost, J. O., 1978. Archaeological Investigations on Tutuila Island, American Samoa. Unpublished Ph.D. thesis, University of Oregon.
  • Furnas, J. C., 1952. Voyage to Windward. London: Faber.
  • Gifford, E. W., 1951. Archaeological Excavations in Fiji. Anthropological Records, 13:189-288. Berkely and Los Angeles: University of California Press.
  • Golson, J., 1969. Preliminary Research: Archaeology in Western Samoa, 1957, in Green and Davidson (eds), Archaeology in Western Samoa, Volume I, Report 1. Bulletin 6. Auckland: Auckland Institute and Museum, pp. 14-20.
  • Goodenough, J. G., 1876. Journal of Commodore Goodenough, R.N., C.B., C.M.G., During his Last Command as Senior Officer in the Australian Station, 1873-1875. V. H. Goodenough (ed.). London: King.
  • Green, R. C, 1967. Fortifications in Other Parts of Tropical Polynesia. New Zealand Archaeological Newsletter, 10:96-113.
  • —— 1974. Excavation of the Prehistoric Occupations of SU-Sa-3, in Green and Davidson (eds), Archaeology in Western Samoa, Volume II, Report 29. Bulletin
- 444
  • 7. Auckland: Auckland Institute and Museum, pp.108-53.
  • Groube, L. M., 1965. Settlement Patterns in New Zealand Prehistory. Occasional Papers in Archaeology 1. Otago: Anthropology Department, University of Otago.
  • —— 1970. The Origin and Development of Earthwork Fortification in the Pacific. Studies in Oceanic Culture History, 1:133-64. Honolulu: Department of Anthropology, B.P. Bishop Museum.
  • Heath, T., MS 1838. The War of A'ana. A Samoan Tale. L.M.S. South Sea Letters, Box 11, Folder 8, Jacket A. Microfilm in University of Auckland Library.
  • Henderson, G. C, 1931. The Journal of Thomas Williams, Missionary in Fiji 1840-53. Volumes I and II. Sydney, Angus and Robertson.
  • Herdrich, D. J., 1991. Towards an Understanding of Samoan Starmounds. Journal of the Polynesian Society, 100:381-435.
  • Hinds, E. S., MS n.d. Prehistoric Sites of Taveuni: A Preliminary Survey. Suva: Fiji Museum.
  • Hocart, A. M., 1929. Lau Islands, Fiji. Bulletin 75. Honolulu: B. P. Bishop Museum.
  • —— 1952. The Northern States of Fiji. Royal Anthropological Institute Occasional Publication No. 11.
  • —— MS n.d. The Windward Islands of Fiji. Wellington: Alexander Turnbull Library.
  • Holmes, S., 1840. Journal of a cruise in the U.S. Brig Porpoise, C. Ringgold, Esq., Commander. Exploring Expedition commanded by Charles Wilkes, Esq. Volumes I and II. New Haven, Connecticut: Beinecke Library, Yale University.
  • Hornell, J., 1936. The Canoes of Polynesia, Fiji and Micronesia. Special Publication 27. Honolulu: B.P. Bishop Museum.
  • Hunt, J., MS 1839-41. The Journal of John Hunt. Volume I, January 1839 to August 1841. MF FM4/1431. Adelaide: Barr Smith Library.
  • Hunt, T. L., and P. V. Kirch, 1988. An Archaeological Survey of the Manu'a Islands. Journal of the Polynesian Society, 97:153-83.
  • Jennings, J. D., R. N. Holmer, J. C. Janetski, and H. L. Smith, 1976. Excavations on 'Upolu, Western Samoa. Pacific Anthropological Records No. 25. Honolulu: Department of Anthropology, B. P. Bishop Museum.
  • —— 1980. Archaeological Excavations in Western Samoa. Pacific Anthropological Records No. 32. Honolulu: Department of Anthropology, B.P. Bishop Museum.
  • Kaplan, M. and M. Rosenthal, 1993. Battlements, Temples and the Landscape of Tuka; the Archaeological Record of a Cultural Transformation in 19th-Century Fiji. Journal of the Polynesian Society, 102:121-45.
  • Kikuchi, W. K., 1963. Archaeological Surface Ruins in American Samoa. Unpublished M.A. thesis. Honolulu: University of Hawai'i.
  • Krämer, A., 1902-3. Die Samoa-Inseln. Stuttgart: Schweizerbartsche Verlag.
  • —— 1978. The Samoan Islands. [An English translation of the 1902-3 edition by T. E. Verhaaren, Palo Alto].
  • La Pérouse, J. -F. G. de., 1799. A Voyage Round the World, Performed in the Years 1785, 1786, 1787, and 1788, by the Boussole and Astrolabe. Volume II. London:
- 445
  • Hamilton.
  • Lawlor, I., 1981. Report on an Archaeological Survey of Northern Lau, Fiji. Working Papers in Anthropology, Archaeology, Linguistics and Maori Studies No. 60. Auckland: Department of Anthropology, University of Auckland.
  • Lawry, W., 1850. Friendly and Feejee Islands: A Missionary Visit to Various Stations in the South Seas, in the year MDCCXLV11. 2nd edition. London: John Mason.
  • Leach, H. M., 1992. Letter to the Editor, Archaeology in New Zealand 35:129-30.
  • Leach, H. M., and D. C. Witter, 1987. Tataga-Matau “Rediscovered”. New Zealand Journal of Archaeology, 9:33-54.
  • —— 1990. Further Investigations at the Tataga-matau Site, American Samoa. New Zealand Journal of Archaeology, 12:51-83.
  • Lockerby, W., 1982. The Journal of William Lockerby, Sandalwood Trader in the Feejee Islands, 1808-1809. Reprint of 1925 Hakluyt Society edition.(eds) E. Imthurn and L. C. Wharton. Suva: Fiji Times and Herald.
  • Lyth, R. B., MS 1848-52. Voyaging Journals 1-4. (B537). Sydney: Mitchell Library.
  • —— MS 1845-48. Journal. (B535). Sydney: Mitchell Library.
  • —— MS 1848-50. Journal. (B536). Sydney: Mitchell Library.
  • —— MS 1850-1. Daybook and Journal. (B539). Sydney: Mitchell Library.
  • Macdonald, J. D., 1857. Proceedings of the Expedition for the Exploration of the Rewa River and its Tributaries in Na Viti Levu, Fiji Islands. Journal of the Royal Geographical Society, 17:232-68.
  • McAllister, J. T., 1933. Archaeology of Oahu. Bulletin 104. Honolulu: B. P. Bishop Museum.
  • McKern, W. C, 1929. Archaeology of Tonga. Bulletin 60. Honolulu: B.P. Bishop Museum.
  • Moors, H. J., 1986. Some Recollections of Early Samoa. Apia: Western Samoa Historical and Cultural Trust.
  • Palmer, B., 1967a. Archaeological Sites of Wakaya Island. Records of the Fiji Museum 1. Suva.
  • —— 1967b. Sigatoka Research Project — Preliminary Report. New Zealand Archaeological Association Newsletter, 10:2-15.
  • —— 1969. Fortified Sites on Ridge-Junctions, Fiji. New Zealand Archaeological Association Newsletter 12:15-19.
  • —— MS 1972. Fijian Warfare. Auckland: Auckland Institute and Museum Library.]
  • —— MS 1973. Mapping Fijian Fortifications on Gau Island. Auckland: Auckland Institute and Museum Library. [
  • Parke, A. L., 1961. Archaeology in Fiji. Transactions and Proceedings of the Fiji Society, 8:10-42.
  • Parry, J. T., 1977. Ring-ditch Fortifications in the Rewa Delta, Fiji: Air Photo Interpretation and Analysis. Bulletin 3. Suva: Fiji Museum.
  • —— 1981. Ring-ditch Fortifications II: Ring-ditch Fortifications in the Navua Delta, Fiji: Air Photo Interpretation and Analysis. Bulletin 7. Suva: Fiji Museum.
  • —— 1987. The Sigatoka Valley — Pathway into Prehistory. Bulletin 9. Suva: Fiji Museum.
- 446
  • Patterson, S., 1967. Narrative of the Adventures and Sufferings of Samuel Patterson. Facsimile of 1817 edition. Washington: Ye Galleon Press, Fairfield.
  • Pearce, P., 1977. Simulated Aspects of Prehistoric Attacks on One Tree Hill. New Zealand Archaeological Association Newsletter, 20:18-24.
  • Peters, K. M., 1969. Excavation of a Star Mound and Earthen Terrace at SU-LU-53, in Green and Davidson (eds), Archaeology in Western Samoa, Volume I, Bulletin 6. Auckland: Auckland Institute and Museum, pp. 210-21.
  • Phelps-Hooper, S. J., 1982. A Study of Valuables in the Chiefdom of Lau, Fiji. Unpublished Ph.D. thesis, University of Cambridge.
  • Platt, G., MS 1835-6. Journal: Raiatea to Hervey and Samoa Groups. London Missionary Society South Seas Letters and Journals, 110, Box 8.
  • Poesch, J., 1961. Titian Ramsey Peale and his Journals of the Wilkes Expedition. Philadelphia: American Philosophical Society.
  • Prickett, N., 1980. Maori Fortifications of the Omata and Oakura Districts, Taranaki. Recordings of the Auckland Institute and Museum, 17:1-48.
  • Rechtman, R., 1992. The Evolution of Socio-political Complexity in the Fiji Islands. Unpublished Ph.D. Thesis, University of California at Los Angeles.
  • Reid, A. C., 1977. The Fruit of the Rewa. Journal of Pacific History, 12:1-24.
  • Richardson, W. P., MS 1810-12. Fegee Vocabulary. (P.M.B. 225). Salem: Peabody Museum.
  • Rosenthal, M., 1991. Realms and Ritual: The Form and Rise of Civitas and Urbs in Southeastern Viti Levu. Unpublished Ph.D. Thesis, University of Chicago.
  • Roth, G. K., 1953. Fijian Way of Life. Melbourne: Oxford University Press.
  • Sahlins, M. D., 1981. The Stranger-King, or Dumezil among the Fijians. Journal of Pacific History, 16:107-32.
  • Schutz, A. J., 1977. The Diaries and Correspondence of David Cargill, 1832-1843. Canberra: Australian National University Press.
  • Scott, S. D., and R. C. Green, 1969. Investigation of SU-LU-41, a Large Inland Fortification, in Green and Davidson (eds), Archaeology in Western Samoa, Volume I. Report 13. Bulletin 6. Auckland: Auckland Institute and Museum, pp. 205-8.
  • Seeman, B. C., 1862. Viti: an account of a Government Mission to the Vitian or Fijian Islands in the years 1860-61. Cambridge: Macmillan.
  • Smart, C. D., 1965. Preliminary Report on Archaeological Field-work in Kabara, Southern Lau Islands, Fiji. October to December, 1964. Unpublished Report. Suva: Fiji Museum.
  • Spennemann, D. H. R., 1989. 'ata 'a Tonga mo 'ata 'o Tonga: early and later prehistory of the Tongan Islands. Unpublished Ph.D. thesis. Australian National University.
  • Stair, J. B., 1897. Old Samoa. London: Religious Tract Society.
  • Tippett, A. R., 1968. Fijian Material Culture. Bulletin 232, Honolulu: B. P. Bishop Museum.
  • Thompson, L., 1940. Southern Lau, Fiji: an Ethnography. Bulletin 162. Honolulu: B. P. Bishop Museum.
  • Trotter, M. M., 1974. List of Rarotongan Archaeological Sites, in M. M. Trotter (ed.),
- 447
  • Prehistory of the Southern Cook Islands. Bulletin 6. Christchurch: Canterbury Museum, pp. 80-3.
  • Turner, G., 1861. Nineteen Years in Polynesia. London: Snow.
  • Turner, P., MS 1839-40. Journal of the Rev. P. Turner. Volume 6 (B305). Sydney: Mitchell Library.
  • Twyning, J. P., 1850. Shipwreck and Adventures of John P. Twyning Among the South Sea Islanders: Giving an Account of their Feasts, Massacres etc, etc, with the Certificates of Wesleyan Missionaries who Lived in the islands. London: Dean.
  • Vickers, J. R., and C. E. Eyman, MS 1980. The 1979 Nadrau Archaeology Project, Viti Levu, Fiji. [Suva: Fiji Museum.]
  • Walter, R., 1990. The Southern Cook Islands in Eastern Polynesian Prehistory. Unpublished Ph.D. thesis, University of Auckland.
  • Wilkes, C., 1845. Narrative of the United States Exploring Expedition, During the Years 1838, 1839, 1840, 1841, 1842. Volumes II and III. Philadelphia: Lea and Blanchard.
  • Williams, J., 1984. The Samoan Journals of John Williams, 1830 and 1832. R. M. Moyle (ed.). Pacific History Series No. 11. Canberra: Australian National University Press.
  • Williams, T., 1858. Fiji and the Fijians, Volume I. London: Heylin.
  • Wright, A. C. S., 1963. Soils and Land Use of Western Samoa. Soil Bureau Bulletin 22. Wellington: Government Printer.
- 448
For the First Time in 100 Years
Compiled by Dorothy Brown

Founded in 1892 by a small band of amateur enthusiasts who considered it their duty to record the life and language of Maori and other Polynesian peoples, the Polynesian Society and its Journal grew in status and influence, and is now the longest-serving academic journal devoted to the scholarly study of the peoples of the Pacific.

Celebrating a century of published research, the Centennial Index documents the achievements and aspirations of a wide variety of professional and amateur scholars. The Index is a completely new work, and acknowledges by name the many Pacific Islander co-workers whose recorded or translated oral traditions were originally published under the name of the collector.

Quite simply, this is an indispensable and unrivalled reference work for all Pacific studies.

Memoir No. 50. Brown, Dorothy. Centennial Index 1892-1991. 279 pp., 1993. Price $30.00. (US$17.50). Postage $2.00 (N.Z.), $3.85 (overseas). Note: NZ residents add GST to above price.


This publication is available to Members of the Polynesian Society (at a 20 per cent discount on the listed price) and nonmembers (at the listed price) from the Asst Secretary, Polynesian Society, Department of Maori Studies, The University of Auckland, Private Bag 92019 Auckland, New Zealand.