Volume 83 1974 > Volume 83, No. 3 > Archaeological discoveries on Niuatoputapu Island, Tonga, by Garth Rogers, p 308-348
                                                                                             Previous | Next   

- 308
ARCHAEOLOGICAL DISCOVERIES ON NIUATOPUTAPU ISLAND, TONGA 1

This report is in three parts: the first discusses the importance of Niuatoputapu in West Polynesia as a highly attractive settlement for the bearers of what is currently known as Lapita-style pottery; the second documents a collection of prehistoric potsherds recently discovered on Niuatoputapu Island; 2 the third describes a field survey of archaeological structures and concludes by drawing together interpretations of archaeological structures, geological features, and oral history as the basis for formulating several hypotheses. It is suggested that for the earliest period of settlement, archaeological excavation must go hand in hand with geological interpretation, and that in reconstructing the latter-day history of Niuatoputapu, the archaeological structures and the oral history of this era must both be taken into account.

Appended to this paper are separate reports by W. R. Dickinson on the analysis of tempers in Lapita pottery from Niuatoputapu and elsewhere in Tonga; and by G. K. Ward on a trace element analysis of two samples of obsidian, the first sample found in association with pottery remains on Niuatoputapu and the second an obsidian source discovered on nearby Tafahi Island. 3 Detailed descriptions of the pottery sherds and archaeological sites described in this report are to be found elsewhere. 4

- 309
PART 1: POTTERY FROM TONGA

McKern discovered pottery on Tongatapu, on 'Eua, and on some small islands in the Tongatapu lagoon in 1921, 5 but its significance was not assessed. Poulsen conducted extensive excavations of pottery-bearing deposits on Tongatapu in 1963-64 and concluded that “. . . from the evidence of Tongan and European informants, no pottery seems to be present in the Ha'apai and Vava'u groups farther north in Tonga.” 6 Groube sailed further north in 1968 and discovered pottery sites on Lifuka Island in Ha'apai, and at Falevai on Kapa Island 7 in Vava'u. Whilst acknowledging the “abundance of pottery” along the fringe of the Tongatapu lagoon, Groube's expedition to the north of the archipelago “. . . established the presence of pottery throughout the Tongan archipelago . . . but only in very small quantities.” 8 Meanwhile, Kaeppler had discovered 151 surface sherds on Tungua Island in west Ha'apai in 1967, 9 and in 1969 Davidson discovered pottery remains on at least six sites in Vava'u “. . . where surface finds were sufficiently numerous to indicate considerable excavation potential”. 10 Finally, in 1973, Green reported Lapita sherds from Mulifanua, Western Samoa. 11

The present report offers evidence which extends the range of pottery in Tonga some 292.8 kilometres north of Vava'u to the Tongan outlier, Keppel's or Niuatoputapu Island. The quantity and extent of this discovery put a major settlement of Lapita pottery bearers only 270 kilometres from Savai'i, Western Samoa, on an island which was probably inhabited by speakers of a Samoic language as late as 1616. 12 Our present state of knowledge shows a continuous distribution of Lapita pottery along a chain of islands stretching from Tongatapu to 'Upolu.

Niuatoputapu: Location and Resources

Prehistorians have not been alone in neglecting to include Niuatoputapu in discussions of West Polynesian and especially Tongan culture history; the dialect has been ignored since 1616. 13 Tongans from the principal archipelago themselves regard “the two Niuas” (Niuafo'ou and Niuatoputapu) as remote outposts; geographically, Niuatoputapu belongs rather to a group of “outliers” involving Niuafo'ou or Tin Can Island, 'Uvea or Wallis Island, and Futuna or Horne Island, and is considerably closer to any one of these and to Western Samoa than to the Tongan capital (see Figure 1).

- 310
FIGURE 1
Position of Niuatoputapu in the Pacific
- 311

The discovery of Lapita pottery suggests that Niuatoputapu should be considered of significance in the prehistory of West Polynesian settlement and of Tongan-Samoan relationships. All resources required for the manufacture of pottery, and found in Lapita-style tempers described for Tonga by Dickinson (see below), are at hand in Niuatoputapu and its neighbouring ash-cone Tafahi. Also, all natural resources required by prehistoric settlers, whether agriculturalists or not, have been present in abundance in the area. Further, there is a remarkable similarity between the ecology and topographic setting of pottery sites on Niuatoputapu and those reported by Poulsen and Groube for Tongatapu.

The precise location of Niautoputapu is 15° 57′ S, 173° 46′ W, placing it well within the southern trade-wind zone of steady east to south-east winds in winter (May-October), and less reliable but strong westerlies which blow usually from January to March.

The outline of the island is like a giant mis-shapen pear, the great round bulbous portion in the south-west and a narrow neck facing north-east: a classic morphological form of a raised coralline island with a “hook”, and a reversed image of Tongatapu. Tongatapu is about 259 square kilometres in area; Niuatoputapu is about 15 square kilometres.

In the centre of the island, along its main axis, lies an old weathered volcanic spine giving the appearance of the backbone of some giant mesozoic reptile. This restricted and steep bush-covered ridge rises to about 107 metres (350′) and is surrounded by at least two extensive flat terraces, products either of tectonic uplifts or of successive falls in sea-level. 14 The greatest portion of the land is an apron of coral uplift which has formed around the volcanic spine. This land, like the wide brim of a hat, is low-lying—even swampy—in parts and there is evidence of an old inland lagoon.

The island defies classification by the most sophisticated typologies yet devised for Pacific islands 15 combining features known as volcanic and raised limestone; on the weather coast is a fringing reef; on the lagoon coast there are apron and barrier reefs; and there is also an inland tidal lagoon and a seaward lagoon with a coral reef islet. More importantly for ancient Polynesian settlers, the island boasts a wide range of natural resources including fine-grained basalt suitable for stone artefacts; clay and sand for pottery; freshwater springs; rich volcanic soils for agriculture; an expansive lagoon providing fish and reef foods and canoe anchorage; as well as a broad strand supporting prolific coconut palms. Separated by only six and a half kilometres of open sea is the volcanic ash-cone of Tafahi (Boscawen or Cocos Island); a prominent landmark for voyagers and a source of igneous stone, of obsidian and clay, and of large trees suitable for vessels.

Hoffmeister and Alling 16 established the presence of andesite tuffs and flows in 'Eua, and Bryan, Stice and Ewart 17 extended this evidence to - 312 include Hunga, Tofua, Kao, Late and Fonualei within what Dickinson and Shutler 18 have called the “andesite arc”.

Although no comparable evidence is yet available for Niuatoputapu and Tafahi, both islands lie on the very edge of the Great Tonga Trench which plunges to 6,826 metres, dividing Niuatoputapu from Samoa. It is reasonable to assume that Niuatoputapu and Tafahi lie within the Andesite Line and are to be distinguished from Samoa and 'Uvea, which are “true” oceanic islands lacking the continental dolerite and andesite-type rocks. 19

Soil Zones

Close to the mountain, the soil is kelefatu “sticky, viscid soil”: this is rich, dark, moist, deep soil which now supports excellent crops of long yam ('ufi kahokaho, Dioscorea alata), giant dry taro (Alocasia macrorrhiza), plantains (Musa paradisiaca (normalis) and Musa paradisiaca (sapientum)), breadfruit (Artocarpus spp.), and large timber trees. Near the coast, the soils are tou'one “sandy soil”: these soils are light in texture and colour, sandy and unsuitable for garden crops, but are excellent for coconuts, village settlement, grave sites.

Between these two zones is the narrow belt of fasifasi'ifeo soils “many broken pieces of coral fingers”: these are dark loams 20-50 cm thick, containing broken igneous stones eroded down from the slopes of the central volcanic ridge and mixed with pieces of coral and shell brought up together with the underlying coarse coralline sand. This zone supports crops of sweet potato (Ipomoea batatas), but it is especially prized for producing the best crops of the shallow-rooted sweet yams ('ufi lei, Dioscorea sp.) and bulbous yams ('ufi voli, 'ufi kaumeile, etc., Dioscorea spp.). It is in and throughout this zone alone that abundant surface pottery sherds are found. Here four minor test pit excavations were made by the author in August and September 1971.

The explanation offered here for the presence of surface sherds throughout the fasifasi'ifeo soil zone is that the activities of yam-growing, involving the excavation of deep holes, has raised potsherds from their primary deposition on or just above the old beach/lagoon sand where the lapita-pottery people formerly lived 20 and has scattered them, together with waterworn shells and coral from the old lagoon floor, about the surface. The pottery sherds are not waterworn.

Hence, just as the pottery in Tongatapu is found on or below the old marine terraces in a narrow belt along the lagoon, 21 pottery in Niuatoputapu encircles the mountain in a similarly narrow belt (see Figure 2).

There are areas in Niuatoputapu where long yams are grown inland of the pottery belt, but despite the common practice of digging very deep holes to plant and harvest these crops, no pieces of coral, shell, midden or

- 313
FIGURE 2
Illustration

pottery are coterminous with them. Finally, no pottery-bearing sites in Niuatoputapu have yet been discovered without these ancient beach-line/ lagoon features though it should be conceded that such areas may exist but have not yet been disturbed by yam-planting activities.

A detailed report of the author's Niuatoputapu pottery finds follows in Part II.

- 314
PART II. POTTERY FROM NIUATOPUTAPU 22

Surface pottery sherds were first discovered in August 1971 while an abandoned yam plot was being measured; subsequent surveys established the distribution of surface sherds throughout a narrow strip of land encircling the mountain (see Figure 2).

Six hundred and sixty-one sherds were collected from the surface of 11 locations, all abandoned or contemporary yam gardens; a further 398 sherds were recovered from four small test-pit excavations.

Test Excavations

Four test pits were excavated in three sites: Lolokoka, Loto'aa, Poome'e. All yielded pottery-bearing soil profiles similar in (a) total depth (0-40 cm), (b) pottery distribution (well scattered from surface to base), (c) basal material (coral beach sand), and (d) zonal location (in the belt of intermediary fasifasi'ifeo soils between mountain and coast).

In addition to 248 potsherds recovered from the Lolokoka excavation, there were 11 baked clay objects, including a typical pottery “disc”, coral “files”, the broken poll end of a partly polished basalt adze, and a small, used, but complete shell adze (Fig. 5c).

A ground oven containing about 200 volcanic and mud-stones was unearthed at the bottom (c. 30 cm) of the cultural layers sitting directly on the underlying coarse coral sand of an old beach-line. Directly associated with this oven were numerous shells, fish- and possibly bird-bone fragments, and a rat femur. A scattering of non-waterworn shell remains was coterminous with all surface and excavated pottery; at least 10 species of shells were recovered from Lolokoka, five of which constitute major shellfish resources today.

The stratigraphy revealed by test-pits suggests that the pottery-bearing Niuatoputapuans were living and cooking on a former beach-line which eventually became covered by soils eroding down the mountain slopes. Absolutely no surface pottery remains were discovered outside this narrow belt encircling the mountain. This evidence compares remarkably with Groube's test excavations in Tongatapu, where pottery-bearing soil “. . . lay upon old beach sand containing water-worn shells and coral fingers . . . apparently . . . an ancient beach” 23; with Specht's earliest site (8) on Watom which revealed basal “. . . sterile white coral beach sand . . .” 24; and with Hedrick's Avnitare site on Malo. 25

General Description of the Pottery 26

The total collection from Niuatoputapu consisted of 1,059 pottery sherds deriving from 11 different locations. The major portion of the collection comprised body sherds (936 or 88.48 percent of the total), but - 315 82 rim sherds (7.74 percent), 20 base sherds (1.88 percent), and 21 un-identified pottery objects were recorded. Of all sherds, 37.58 percent were from excavation as against surface collections. Only 1.49 percent of all body sherds and 19.5 percent of all rim sherds were decorated. Many sherds retained evidence of the fabrication process with a variety of striations, irregular shallow grooves and score-marks.

All body sherds were measured for their average thickness, and were recorded for size by counting the number of squares each covered when placed on a grid of one centimetre squares; rim sherds were measured for their lip thickness. The range of thicknesses of plain body sherds is considerable, from 4 mm to 18 mm (see Table 1). Statistically, there is a continuum in the distribution of sherd thickness, but handling the material one receives the impression that there are a very small number of sherds of a more finely and harder textured, thin-sectioned, superior ware, contrasting with the dominant thick, coarse ware.

Owing to the fragmented and worn nature of the material collected, it was not possible to physically reconstruct any vessel shapes. Consequently, the range of thickness which would occur in any one vessel is not known, and for this reason the sherd thicknesses are of dubious value for indicating the range and styles of vessels which would have constituted the suite of Niuatoputapu pottery.

Nearly all of the sherds from surface collections were well weathered and retained little evidence of their original surface or mode of manufacture owing to the friable nature of the material. The best evidence for mode of manufacture, slip, and decoration comes therefore from the 259 sherds excavated at Lolokoka and to a lesser degree from the 139 sherds excavated from Loto'aa, all of which, however, were weathered to some extent.

A number of sherds, some of them from surface collections, had been inadequately fired or fired at low temperatures, as indicated by the internal carbonisation (like black filling in a brown sandwich) seen in the cross-sections of these sherds. Many sherds have a black surface on the inside caused apparently through burning, cooking or firing, whereas no sherds reveal this phenomenon on their outside surface. The general texture of the sherds was crumbly and very porous. 27

Two sherds from Lolokoka, one excavated and one from the surface, reveal evidence of the “slab-technique” where the slabs of clay have not completely fused together. 28 Seven percent of the sherds from the excavations and some from surface finds show numerous indications of depressions on the interior of the vessel, some in the form of finger or “anvil” marks; 45 body-sherds deriving from different sources had been treated with a red slip on the surface and then brushed with some sort of fibrous material on the outside while the clay was still wet. There were no clear signs of the use of a “paddle” except for the shallow grooves discussed below.

- 316
TABLE 1
Indices of thickness in mms of body and rim sherds from Niuatoputapu
Site Body sherds         Lip of rim sherds        
  Mode Median Mean Range No. in Sample Mode Median Mean Range No. in Sample
Lolokoka (layer I) 10 9 8.65 4-15 60 7 8 10.8 6-22 20
(layer II) 7 8 7.93 4-17 144 6 7 7.33 4-12 9
(surface collection) 7 9 8.90 5-17 132 7 & 9 9 9.4 6-16 24
Loto'aa (excavation) 8 7 6.90 4-13 129 6 & 8 8 8.1 6-12 7
(surface) 13 12 12.2 4-17 93 14 13 13.3 10-17 6
Poome'e (surface & excavation) 6 7 6.9 4-12 189 6 8 8.8 6-13 7
Ha'afisi (surface) 11 12 12.1 9-18 104 13 & 19 16 16.0 13-19 2
Tu'akolo Nahau (surface) 9 9 9.1 5-13 39 9 & 7 8 8.3 7-10 6
Tokelau (surface) 6 6 5.0 4-12 21 9 9 9.0 1
Total body sherds 936 — 88.48% of total
Total rim sherds 82 — 7.74% of total
Total base sherds 20 — 1.88% of total
Pottery objects 21 — 1.9% of total
  1,059 — 100%
  • Total excavated as % of all sherds — — 37.58%
  • Total decorated body sherds as % of body sherds 1.49%
  • Total decorated rim sherds as % of rim sherds 19.5%
  • Total decorated sherds as % of all sherds 1.32%
- 317

The colour of the sherds ranged from light to dark chocolate-brown, a few were a brick-like orange and some were a burnt black, especially on the inside face.

Clay tempers from a sample of Niuatoputapu pottery have been examined by Dickinson and compared with tempers of Lapita-ware from other Tongan sites. (See Appendix 2 where Dickinson describes three distinctive temper sand types for the nine sample sherds from Niuatoputapu.)

Sherds containing high proportions of white calcareous fragments (75 percent in one Loto'aa sherd) are common throughout the entire Niuatoputapu sample, as are sherds revealing black volcanic sand and coarse volcanic rock fragments. A great many of the Niuatoputapu sherds exhibit large quantities of “glassy, pumiceous volcanic rock fragments”, some of which reach 8 mm. Small pieces of pre-baked clay (up to 6 mm) are common temper material, as are large pieces of crushed shell (up to 5 mm); occasional holes on the face or edge of sherds point up the shedding of gritty temper after firing. Black volcanic sand, obsidian, and pumice are all present on Tafahi Island; the volcanic rock and calcareous fragments may be derived from either place. On the basis of Dickinson's work and a preliminary examination of the geology of Niuatoputapu and Tafahi in 1971, it can be concluded that all ceramic material so far recovered from Niuatoputapu was manufactured in that place from local materials.

Decorated Pottery and Its Characteristics

Many sherds were so badly eroded that modification and decoration were difficult to detect. Several sherds had been modified but not decorated; these include one spout (Fig. 31) and one pottery disc, with a “countersunk” hollow in the centre of one face very similar to those already described for Tongatapu 29, Fiji 30 and Western Samoa 31. Many sherds (50 or more) exhibit irregular striations on the outside surface. They fall into three categories: (1) There are some plain body sherds with broad, irregular shallow grooves on the exterior surface similar to those reported by Poulsen for Tongatapu; 32 (2) There are sherds which have more regular striations on the exterior surface and have obviously been scraped or wiped before firing while wet. Many of these sherds have some deeper, broader incised lines across their surface as though a harder, sharper, almost toothed tool had been drawn or occasionally flicked across the wet clay. One sherd with a collar-rim and neck (Fig. 4j) has particularly pronounced sharp oblique striations on the rim and a notch on the angle formed by the rim with the neck; 33 (3) Other sherds had evidence of the coarse deeper striations being partially covered by much finer lines which run in a different direction. The suggestion here is that the vessel surface was smoothed one way with a softer, finer tool or fibre. 34

- 318
FIGURE 3
Decorated Potsherds and Spout
- 319
FIGURE 4
Selected Rimsherd Sections and Pottery Objects, Niuatoputapu
- 320

There were only 14 sherds recognisably decorated on the surface (1.32 percent of all recovered sherds): 10 incised; three with coarse (Fig. 3 g, j, k), and one with very fine (Fig. 3i) “dentate-stamping”; and only one with a band of applique (Fig. 3f).

The only sherd with three-dimensional design elements is a small piece with a thin transverse bar (Mead TB3.2) 35 added to the surface (Fig. 3f). Above (or below?) this transverse bar are two straight, oblique, incised lines meeting at the edge of the transverse bar. If these incisions were repetitive they would form a line of triangles consistent with M18.1→ Con R/E-W. 36

Four sherds (one body, one shoulder and two rim sherds) have the characteristic Lapita “dentate-stamping” 37 on their surface and warrant describing in detail. One shoulder sherd (Fig. 3i) has very fine parallel lines (GZ2) running obliquely from the sharp angle (c. 100°) of what was possibly a round-bottomed bowl similar to the sharp-angled shouldered bowls from Tongatapu and Fiji. 38

The second decorated body sherd is a large flattish piece (Fig. 3j) from the surface of the Lolokoka pit excavation with extensive dentate-stamping. The motif is built up from straight single lines (GZ2) below which are arcs (DE1a) and above which are “circles-with-legs” 39 in a continuous band round the vessel. The decoration, however, is, as Poulsen and Golson have pointed out for Tongatapu material, “carelessly executed”; the motifs have “exploded”. 40

One decorated rim-sherd (Fig. 3g) has Mead's M12.2 motif nicely executed on the collar of the straight rim with GZ2 repeated across the almost square lip of the rim. Another everted rim (Fig. 3k) has dentate-stamping on the interior surface with a Type H zone in which DE6 in single lines has been placed above GZ2 and a small part-oval single dentate line placed below it. Another slightly everted rim-sherd (Fig. 3h) with rim expanded or slightly thickened outwards is decorated with GZ4 (a deep, single, continuous incised, horizontal line) just below the expanded rim forming an upper border, Zone Type H, for a horizontal, probably continuous, series of DE3 or small circles stamped into the clay with a cylindrical stamp. The lip of this rim-sherd was decorated with a continuous series of single-line incised crosses 41, now faint with erosion. Other rim-sherds with modification to the lip include a wavy transverse notched straight rim-sherd; 42 deep-notched expanded rims; 43 and an edge-notched rim. 44

- 321

The two-dimensional decorations on sherds from Niuatoputapu are remarkably similar to those reported from Fiji, Tongatapu and Western Samoa, with exceptionally close agreement between the motif in GZ2 of a “circle-with-legs” 45 reported by Poulsen 46 for Tongatapu and by Green 47 for Mulifanua, Samoa, and also found in the Lolokoka excavation, Niuatoputapu. Other remarkable similarities are (a) between the decorated everted rim-sherd, also from Lolokoka, and the bowls with everted rims and lips recovered from Yanuca, Fiji, 48 and (b) between the pottery disc recovered from Lolokoka and almost identical discs recovered by Poulsen 49 from Tongatapu, by Birks 50 from Viti Levu and reported by Green 51 from Mulifanua. 52

The rim forms of Niuatoputapu pottery exhibit a very wide range of style, thickness and vessel form, suggesting a relationship with the earlier or decorated phases of Tongan Lapita-ware. 53

A large proportion of the 83 rim-sherds recovered are expanded asymmetrically to the exterior at the rim, some forming a very broad (up to 22 mm) flat expansion (Figs. 4c, d, e) comparable to specimens reported by Garanger 54 for Efate, New Hebrides, Birks 55 for Sigatoka, Fiji, and Poulsen 56 for Tongatapu, but without decoration. Many sherds (e.g. Fig. 4j, k) exhibit the collar-rims characteristic of Lapita-ware, one decorated sherd (Fig. 3e) being almost identical in cross-section to a sherd from Mulifanua. 57

It may be postulated that the Niuatoputapu sherds come from both decorated and plain ware; sherd shapes (as of rim, base and shoulder) suggest that the Niuatoputapu suite of pottery probably would have included the Lapita-style dish with flattish bottom, the everted open-mouthed shouldered jars and bowls, and plain, straight-rimmed simple bowls both large and heavy and small and fine.

A number of fired pottery objects include one “spout” from Matavai (surface) (Fig. 3l), and several oval sections of what might represent “handles” (Fig. 4o) and “lugs”. 58 Other less symmetrical pieces were discovered on the surface at three sites and from pit excavations at Lolokoka and Loto'aa (Fig. 4 l, m, n); 59 the suggestion is that they are portions of legs or stands or that they may have been “kiln furniture.” 60

- 322
Discussion and Significance of the Pottery

Despite the low proportion (2.94 percent) of decorated to plain sherds, the pottery from Niuatoputapu falls within the category now known as Lapita-ware. The wide range of vessel shapes and rim forms, the motifs and decorations, the methods of fabrication, firing, slip, the pottery “objects”, all relate to the general descriptions of Lapita-ware supplied by Poulsen, Specht, Golson, Hedrick, and Birks. 61 On the other hand, the colour and friable nature of the material, the temper and the proportion of decorated ware relate the collection more particularly to material from Tongatapu, Ha'apai 62, Vava'u and Western Samoa. 63 The accompanying reports by Dickinson and Ward suggest that Niuatoputapu pottery was locally produced.

The proportion and sample of decorated ware reported here should be used tentatively until full-scale archaeological excavations have been carried out. So too, the possibility suggested above of there being two kinds of ware: one fine, thin, and well executed represented by a very small proportion of the sherds; the other thick, crude and coarse; 64 but statistical and temper analysis suggests a continuum from fine thin-walled to coarse, thick ware with one local source for the raw materials.

Artefacts

The only excavated stone tools were the broken poll end of a partly polished basalt adze from Lolokoka, and a few worked basalt flakes from Poome'e. A further seven complete and 14 broken basalt adzes, and 8-10 partly polished stone flakes and points were recovered from various surface locations on the island. Of the seven complete adzes, three (Fig. 5a, d and Fig. 6d) were fully ground on all faces except the poll, and four (Fig. 6a, b, c and Fig 5e) were partly ground on two (Fig. 6a, b), three, or four faces. Of peculiar shape was the concave cutting edge of a very small adze (Fig. 5d) which had obviously been re-ground.

Nearly all of the stone adzes recovered have quadrangular cross-sections: at least four have fronts wider than the back and conform to Green and Davidson's Type IV reported for Samoa. 65

The small shell adze (Fig. 5c) recovered from the pit excavation in Lolokoka is made from the hinge part of a Tridacna shell and has been well ground on the front and bevel but mainly flaked and pecked on the sides and poll, the natural corrugations of the shell still showing on the back of the adze. This is in conformity with the 10 shell adzes found by Poulsen 66 in Tongatapu, as is the hollow ground cutting edge.

- 323
FIGURE 5
Artefacts in Stone (a, d, e) and Shell (b, c)
- 324
FIGURE 6
Basalt Adzes, Niuatoputapu
- 325

One short shell segment (Fig. 5b) from the surface of Lolokoka had three connecting drill holes which had been bored from either end and the centre of one face so that they met in the centre of the segment. No other worked shell materials were found. Several rubbed coral “files” accompanied surface and excavated material from several sites.

Lapita Economy

Although no fishing gear was recovered, food items associated with the ground oven at the very bottom of the soil horizon in Lolokoka depict the Lapita potters as a maritime people exploiting lagoon and deep-sea and supplementing their diet with bird and rat. 67 As with the pottery, one may speculate on the use to which the large ground oven (containing at least 185 oven-stones) was put.

PART III: ARCHAEOLOGICAL STRUCTURES AND VISIBLE REMAINS ON NIUATOPUTAPU

A field survey of Niuatoputapu was undertaken over a period of 12 months during 1970-71 in order to draw maps of soils, land use, communication routes and place names. The following description of archaeological monuments derives incidentally from these surveys; each site has been numbered and its local name recorded where possible. 68

McKern 69 classified Tongan (excluding Niuatoputapu) archaeological sites ethnographically, according to the linguistic categories of Tongan informants. A purely ethnographical or functional classification of the Niuatoputapu sites is not possible owing to the loss of local knowledge about many of the older sites.

Davidson 70 found McKern's ethnographic designations of 50 years earlier did not always accord with modern informants and therefore based her classification of Vava'u sites on morphology, but included ethnographic details in her descriptions. A purely morphological classification of Niuatoputapu sites would be unsatisfactory because some sites of similar form are known to differ in function, 71 while other sites of differing form share the same function. 72

Since, therefore, neither a Niuatoputapu linguistic classification nor a purely morphological classification is either wholly reliable or fully sufficient for our descriptive purposes, we shall resolve this dilemma by - 326 employing features from both systems. The classification will be based, where possible, on ethnographic evidence, and, where not possible, on regional (but not necessarily contemporaneous) complexes. Careful attention will be paid throughout this description to morphological features in an attempt to correlate them in some meaningful way to the known functional interpretations of the sites. By employing this method, the following classes are arrived at:

  • 1. Named 'esi or commemorative flat-topped platforms for chiefly persons,
  • 2. Named sia and un-named, un-faced earth-mounds,
  • 3. 1. Named langi or chiefly burial mounds faced with dressed stones (paepae).
  • 3. 2. Unnamed, square or rectangular mounds faced or partly faced with dressed stones (paepae).
  • 3. 3. “Fa'itoka” or low earth-mounds with stone edging (paepae).
  • 3. 4. “Fa'itoka” or high earth-mounds with stone edging (paepae).
  • 3. 5. Contemporary village or descent group graveyards.
  • 4. Named and unnamed stone boundaries, stone markers, former sacred stones and quarry sites.
  • 5. Fortifications, ditches, former sunken roads.
  • 6. Luo maa or food fermentation pits; vai tupu or wells.
  • 7. Post-historic abandoned settlement sites.
  • 8. Pottery-bearing non-concentrated middens.
Mounds and Platforms, Classes 1, 2, 3

Class 1 are all stone monuments; either square or rectangular, and either built of coral boulders and rubble or faced with vertical coral slabs. All have a flat-topped platform; all occur on beach vantage points from which magnificent views are afforded.

Class 2 are all earth-mounds with one notable exception, which is a slightly modified volcanic neck of stone in a swamp; they are circular, unfaced, mostly flat-topped, and occasionally are domed. All are surrounded by some sort of ditch, from which most obtain their height.

Class 3 are all earth-mounds with some form of stone facing or edging (paepae) in the form either of vertical coral slabs or of small stones placed on edge. Shape is unimportant.

Class 1

'Esi are monuments constructed to honour high-ranking persons. They are usually named after the person they are intended to commemorate and are built in a beautiful or commanding position often near the coast where the chiefly personage can sit and rest aloft in the shade of a tree. 73 Named 'esi are recorded in legend throughout the Tongan chain and can be located geographically; 74 they are still constructed and continue to be a feature of modern Tonga culture.

- 327

There are three such monuments on Niuatoputapu, all known by name to the present inhabitants. All are situated on the coast far from habitation but commanding fine sea views. Built in honour of a recent Maa'atu chief, the 'Esi 'o Panuve (Ntt. 001) is situated on the extreme north-west point of Hunganga Island and overlooks the western horizon. The outlook from the platform of the monument (about 7.3 by 6.4 metres in extent and about 2.5 metres high) is now obscured by bush, but if this were removed a fine sweep of horizon from north through west to south would be available, encompassing the bearings of 'Uvea, Niuafo'ou, Futuna, and the setting of the sun throughout its yearly course. Alokivakaloa 75 (Ntt. 002) is on the opposite, windward coast and overlooks the east. Although now partly obscured by bush, the view of the horizon from the 9 by 4.5 metre platform, about 3 metres high, would enable one to see the sun rise from the sea throughout the year.

No attempt is now made by anyone to keep or preserve these two monuments in good order and, given the local predilection for land-crabs, 76 within a few decades they may be reduced to indeterminate heaps of stone.

The 'Esi 'o Pilolevu (Ntt. 003) was built to commemorate the visit of H.R.H. Princess Pilolevu to the island in 1959 and is thus proof of a continuing tradition of and regard for stone monuments on the island. It is also a mark of the Tonga royal family's influence in the island through title to the royal estate on which the 'esi has been built. This 'esi is situated on the lagoon shore a short distance south of 'Utulongoa'a near the easternmost tip of the island It is only 10 paces or so from high-water mark, and the builder of the 'esi remarked on the facility with which a small boat or canoe could at high tide sail right up to the beach by the monument without let or hindrance from reef or rock; it is a fine sight on a clear day looking out across the lagoon from the top of 'Esi 'o Pilolevu over a long arm of white sand and green water stretching away into Matakilikili, the eastern deep-water boat passage.

Unlike any other stone structure on Niuatoputapu, 77 the 'esi comprises two tiers; both are rectangular; the upper tier is stepped back on all sides. Both tiers are faced with vertical coral slabs and in its building many coral stones were transported from the nearby monument, Mataki-'Uvea. 78 The top surfaces are covered with fine coral rubble and beach sand. The monument is not very large; its volume is only about 168 cubic metres.

McKern 79 reported two stelae of undressed basalt from the volcanic island of Niuatoputapu. He described them as “two stone monoliths . . . which, according to tradition, were erected by the culture hero, Maui. - 328 They are named Fakafafa and Tauloto”. 80 On the eastern coast of Tafahi are two volcanic necks named the Fakafafa mo e Tauloto 'a Maui.

Mata-ki-Ha'amoa and Mata-ki-'Uvea. There are two huge artificial stone monuments in the extreme eastern tip of the island of Niuatoputapu. The present inhabitants have no theories, ideas or explanations concerning these monuments (many have not even seen them), but call them simply Mata-ki-Ha'amoa (Ntt. 005) “looking-towards-Samoa” and Mata-ki-'Uvea (Ntt. 004) “looking-towards-'Uvea”. They are included here in Class I because they are the only other stone structures on Niuatoputapu. Mata-ki-'Uvea is about 27.5 metres long, 23 metres wide and about 4.5 metres high, lying NW/SE true on its long axis. It now comprises about 2,800 cubic metres of coral stones up to 60 cm but mostly 30-40 cm in diameter. It is quite flat on top with steep sides built of flattish boulders stacked like pancakes. These have been grossly disturbed by land-crab hunters, and on one side in particular by the builders of the 'Esi 'o Pilolevu. Close examination revealed pockets of light-coloured beach sand between the upper stones, suggesting that the surface had once been dressed in this manner. Mata-ki-Ha'amoa (Ntt. 005) is situated about 100 paces due east of Mata-ki-'Uvea but is obscured for a distance of more than 30 paces by fairly heavy forest and sapling growth. It measures 21.5 by 15.25 metres across the platform and over 9 metres high, and is thus only just over half the area but twice the height of its companion Mata-ki-'Uvea and of about the same cubic content (2,800 cu. m.). The surface of Mataki-Ha'amoa is covered with a layer of dark humus up to 20 cm deep in some places but shallower in others (where it has been washed down through the coral stones); otherwise, the composition of this monument is very similar to Mata-ki-'Uvea.

The true east-west alignment of the two monuments suggests an astronomical significance, whereas their current names suggest they were voyaging markers or navigational beacons. 81 The bearing from Mataki-Ha'amoa to the closest point of Savai'i, Western Samoa, is 028 degrees true (NE by N), the distance being 270 kilometres, whereas the bearing from Mata-ki-'Uvea to 'Uvea is 318 degrees true (NW), the distance being 382 kilometres.

McKern observed that “. . . no Tongan mounds are constructed entirely of stone”, 82 yet, like the numerous large stone mounds reported from Western Samoa, 83 the five stone mounds on Niuatoputapu appear to be stone and rubble throughout. Three of them (001, 002, 003) were probably built within the last 100 years, but the two huge Mata mounds (004, 005) probably belong to a distant era and a social structure radically different from that found on the island in recent times. It is calculated that Mata-ki-'Uvea and Mata-ki-Ha'amoa each contain upwards of two million stones. The builder of 'Esi 'o Pilolevu (Ntt. 003) stated that it took 100 men 15 days to construct it; by inference, therefore, it would - 329 take the same 100 men one year to construct each of the two Mata monuments.

Pending archaeological reports from 'Uvea and Futuna, the closest known relations of the two Mata sites on Niuatoputapu are the “large stone mounds” reported by Buist for Savai'i, Western Samoa. 84 Scott's description of these stone mounds is accurate for the Mata sites described here: “. . . bouldery basalt rock, rough and dry-laid, the platform top and sloping or vertical sides are recurring and measurable features . . .” 85 with the substitution of “coral” for “basalt”.

Class 2. Named Sia and Unnamed, Unfaced Earth-mounds

Stories (fananga and talanoa) associated with or referring to famous sia abound in Tongan literature; some, as reported by McKern 86 were once 'esi; an example is Siakanume in Nomuka; others, such as Sia-ko-Veiongo, a hill behind the Palace in Nuku'alofa and Sia Longo 87 near Tefisi, Vava'u, are named after famous or illustrious ancestors; one sia, Sia-fou-ki-moana on Nomuka Island, is reported by Gifford 88 to have once been used by fishermen to determine their distance from the shore, although this may not necessarily have been the original function of this mound; some sia were “a chief's place and supernormally guarded by the Tu'i Tonga's tapu”. 89 One such is the Sia-mafua'uta near Ma'ufanga, Tongatapu, described by Cook as an “etchee” ('esi) belonging to Tu'i Tonga Pau. 90 In fact, McKern's informants offered differing interpretations of this sia, some regarding it as an 'esi or chiefly resting place, others as a pigeon mound or sia-heu-lupe, and one narrative depicted it as a house-mound for Siamafua'uta, a sister of the Tu'i Tonga. Some sia — e.g. Sia-maka on Uoleva, Ha'apai and Sia-Hine-fai-tehina on Tatafa Island off 'Uiha, Ha'apai — were, according to Gifford's 91 informants, the abodes of gods or endowed with powerful supernatural significance; but the best-known and most widely recorded use of sia in Tonga was as a place where pigeons (lupe) were decoyed, this being a favourite sport and pigeons a favourite food of chiefs. Such mounds are called sia-heu-lupe. 92

Whereas 'esi is a functional term, referring to special rest places for aristocrats, sia is merely a general descriptive word which applies to many kinds of mounds; McKern 93 uses it for both 'esi and sia-heu-lupe “pigeon mounds”; Davidson 94 invariably uses the word for unfaced earthmounds “that are not recognisable as burial mounds”; on Niuatoputapu it is applied to any high unfaced earth mound known or believed not to - 330 be a grave or burial site and to one named low outcropping volcanic neck. Sia on Niuatoputapu therefore include some known house sites, some commemorative sites, and mounds of undetermined function, 95 but they are distinguished from grave sites. Sia as a general category is therefore of limited use in this or any further study of mounds in Tonga owing to the wide variety of reputed functions it denotes.

There are five named and five unnamed mounds on Niuatoputapu which are locally designated sia: Sia Manafa (Ntt. 006), Sia ko Finetengalelei (Ntt. 007), Ha'afo'ou (Ntt 008.1), Haufakalaki (Ntt. 009), Nofo'atoa (Ntt. 010), [Motu I] (Ntt. 011), [Motu II] (Ntt. 012), [Kolo] (Ntt. 014), and [Tokelau] (Ntt. 015). All except Sia Manafa, which is a stumpy volcanic rock, are raised, unfaced, earth-mounds; nearly all are circular in shape. Sia ko Fine-tenga-lelei, “Young-woman-with-beautiful-thigh” is an extensive, unfaced, circular, flat-topped mound surrounded by a ditch in the tract called Taakoto, “to-lie-down, of two (or more) persons”. It was reputedly built to commemorate the taking of a virgin by a royal chief from Tonga. Ha'afo'ou, “new-chiefly-lineage”, is reputed by the present holder of the Motu'ahala title to have been the dwelling place of his forebears, stretching back over seven generations, who were cooks and provisioners to the Maa'atu line of chiefs. The diameter across the top of the platform is about 40 metres, whereas the diameter from rim to rim of the ditch is about 55 metres. The surface of the platform is flat, covered with sandy soil, and about 1 metre above the surrounding ground level. The morphology and dimensions of this site are remarkably similar to those of Sia ko Finetengalelei. Nofo-'a-toa, lit. “Encamping of Warriors”, is a sia close to the mountain behind Falehau village: it is reputed to have been a temporary camping place of the warriors of Fuimaono (then chief of Falehau) during the battles against Maa'atu. The mound is oval, quite flat on top and is surrounded by an ill-defined ditch. It is reminiscent of other sia in Niuatoputapu in being built on a hill-slope so that on the down-hill side, and with the help of the ditch, it appears more impressive an earth-work than the relatively small effort required to build it would suggest. Hau-faka-laki, lit. “Haughty Victor”, although named, has no oral traditions to account for it. Situated close to Nofo'atoa, it is a high, circular, unfaced earth-mound, slightly domed on top, about 5 metres high on one side, and lower on the road side, where there is the suggestion of an earth bridge or access ramp. 96

Class 3: Mounds and Platforms Faced with Upright Stones

The most numerous archaeological monuments on Niuatoputapu are stone-faced mounds and platforms square, rectangular, oval or circular in shape, and 10 centimetres to 3 metres in height. A common feature is the paepae, formed of stones placed on edge to form a retaining border or wall of the structure. The paepae stones vary in size and material from small natural coral stones and igneous rocks to huge cut limestone slabs 2 m long, 1.3 m high and 18 cm thick.

- 331

The langi-type structures, the largest version of class 3, are an aristocratic form of the smaller less grand fa'itoka; both are recognised by modern-day Tongans to be burial grounds. The features of the stone facings (paepae) of classes 3.1 through 3.3 may turn out to be correlated with a continuum of gradations in rank.

Class 3.1 and 3.2 are rectangular mounds with large vertical dressed coral slab paepae. In many cases, the coral slabs are largest at one corner or side, dwindling into undressed natural coral stones at the opposite extremity.

Class 3.3 are mostly rectangular, low, flat platforms edged with undressed coral or volcanic stones set upright. Such sites are invariably designated “fa'itoka” “burial place”, although they may have been house sites.

Class 3.4 are all higher earth mounds, some of which are very old and therefore indistinct in shape, with coral stone edging around the base or occasionally around the top.

Class 3.5, contemporary village or descent group graveyards, are different in that they consist of numerous mounds of sand with occasional edging, and odd European-style headstones. Their use may extend back into prehistoric times.

Class 3.1. Langi or tombs of the Tu'i Tonga and related lineages are found throughout Tonga; 97 they, and sometimes even particular stones in them, 98 are named and still somewhat revered in Tongan custom. Their distinctive feature, well described by McKern, is the dressed rectangular slabs of limestone set on edge to retain the mound containing the chiefly tomb. There are sometimes one or as many as five rectangular courses to these monuments, each course set in on all sides like a stepped pyramid.

There are five langi-type structures in Niuatoputapu which are named, although there is no specific knowledge of who has been buried in four of them. If langi were, as McKern concluded, “. . . invariably the tombs of important members of the Tu'i Tonga ‘family’, . . . includ[ing] . . . the family of the Tamaha, Tu'i Tonga Fefine and of certain closely related chiefly families such as that of the Tu'i Ha'atakalaua . . .”, 99 the only aristocrats of Niuatoputapu fit to be buried in langi would have been members of the last line of Niuatoputapu rulers known as Maa'atu.

According to genealogical records in the Palace Records Office, Nuku'alofa, the Maa'atu line is descended from a younger brother of the Tu'ilakepa, at that time the highest ranking line of the Ha'afalefisi. The Tu'ilakepa line began with the marriage of Sinaetakala'ilangileka, sister of the Tu'itonga Fatafehi, to a Fijian chief, nicknamed “Tapu'osi”. Fatafehi was followed in succession by Kau'ulufonua, 'Uluakimata, Tu'ipulotu'ilangitu'ofefafa, Fakana'ana'a, Tu'ipulotu-'ilangitu'oteau, and Pau. - 332 Tu'itonga Pau met Cook in 1777. 100 The “Latoo” whom Cook also met at Tongatapu in 1773 was the Tu'ilakepa, Laatuunipulu'iteafua. 101 Cook concluded that he was the first man on the island; Kaeppler 102 describes the reasons for his high status.

The Latou — “head man/highest ranking person” — whom Schouten and Le Maire met in 1616 at Niuatoputapu must have been a very early Laatuu of this line as there are only six successions to the title of Tu'itonga recorded between the origin of the Tu'ilakepa line and 1773.

The Tu'ilakepa line was descended from a Tu'itonga fefine; the Maa'atus in Niuatoputapu continued to marry into high-ranking Tongan lineages. Thus, descent alone would have allowed the Maa'atus the prerogative of langi-style burial; the independent political power which the line acquired 103 would have enabled them to implement it.

The characteristics of the named langi-type structures on Niuatoputapu are as follows:

  • 1. All consist of only one platform (McKern's type D), none possessing a series of terraces.
  • 2. All have a paepae consisting at least partially of dressed rectangular limestone blocks, retaining an earth mound.
  • 3. All are rectangular in plan with the angles of the northern-eastern faces sharply demarcated.
  • 4. The largest and best-dressed stones are placed at one corner (usually NE) or on one face (usually north-east), and from there may dwindle towards the opposite end of the structure, unable to attain the consistent grandeur of the Tongatapu langi.
  • 5. A feature also reported by Davidson 104 for the Vava'u sites “Paepae Ua” at Vaihoe and “Otalafaiva” on 'Euakafa, is the presence of a slab-faced extension or slab-faced mound attached to one corner or one end of the structure.

The named langi-type structures on Niuatoputapu are:

Tu'atangiketatau (Ntt.020), Olosenga and Falepouahi (Ntt.021), Tofi'a (Ntt.022), Fa'itoka 'o Seketo'a (Ntt.023), and Fa'itoka 'o Maa'atu (Ntt.024). Two of these slab-faced burial mounds lie at the centre of the Vaipoa village cemetery beside the main coast road. The larger of these is Tu'a-tangi-ke-tatau “Commoner-trying-to-be-equal”, the other is a single structure with two names: Olosenga at the eastern end, and Fale-pou-ahi “House-of-sandalwood-posts” at the other. It is uncertain whose tombs these two structures contain, but the “Paepae 'o Olosenga mo Falepouahi” is associated in poetry with the Maa'atu line and its location close to the former seat of the Maa'atu chiefs confirms this. The langi-type structure in Tofi'a (Ntt.022) is a similar large slab-faced mound situated at the eastern - 333 end of Vaipoa village, as is the “Fa'itoka (sometimes called ‘langi’) 'o Seketo'a” (Ntt.023). Seketo'a was the putative brother of a Maa'atu; although he was of chiefly birth, he did not become a Maa'atu ruler but was killed by his brother and reappeared as the Niua shark-god, Seketo'a. 105 Some informants say that Seketo'a was buried in Tofi'a (Ntt.022), some that he rests in the Fa'itoka 'o Seketo'a (Ntt.023), others that he requested to be buried at sea. In the western corner of Hihifo village is the final resting place of the Maa'atu title-holder Sioeli Kalaekii-valu, who died in 1894. Like the Fa'itoka 'o Seketo'a, this burial mound is sometimes called a fa'itoka, sometimes a langi. Thus slab-faced mounds associated with the Maa'atu lineage are found in both Hihifo and Vaipoa. Although since the first written Constitution (1875) the Maa'atu estate has comprised only the Hihifo area, oral accounts record that the seat of the Maa'atu line was moved here from Vaipoa during the time of Sioeli Kalaekiivalu.

Class 3.2: Ve'elangi (I) (Ntt. 101.1). Between the eastern boundary of Falehau village and the stream called Me'eme'e is a slightly raised tract of land called Fatutanu. Where this ground slopes down to the eastern boundary of Falehau village, a rectangular, slab-faced burial mound has been erected and forms the centre-piece of the Falehau cemetery, Ve'elangi I, (old portion). This mound is believed to contain tombs of the chief Langi-'o-ha'a-vaka-fuhu and his lineage. Langi was a chief from Ha'apai; the origin of his lineage is not known in Niuatoputapu. This burial mound is not called a langi and it should be noted that the use of the word langi in Tonga applies not simply to a morphological class of structures but is reserved for the burial structures of chiefly persons of Tu'i Tonga descent. No matter how impressive the burial structures of other chiefs, they are not called langi.

The fonualoto “tombs” of modern-day Falehau families are situated on the lower ground below this mound, and have been designated Ve'elangi II.

McKern 106 noted a diminishing number of langi as one proceeded north from Tongatapu, recording 45 for that island, 6 for Ha'apai and only 2 for Vava'u. Davidson 107 extended this figure in Vava'u to 5 possible langi and 4 unnamed slab-faced structures of similar style. Niuatoputapu confirms McKern's observation, adding 5 named and 2 unnamed langi-type slab-faced structures some 292 kilometres north of Vava'u. This distribution is consistent with the establishment in Niuatoputapu of a line of the Ha'afalefisi lineage, which was of Tu'itonga descent, and on this evidence it may be expected that Tongan langi could be found in Samoa and 'Uvea.

Concerning the age of Tongan langi, McKern 108 concluded that they “. . . probably do not antedate the reign of Tu'itatui . . . 11th century”. Niuatoputapu “langi” are unlikely to predate the arrival of the founder of the Maa'atu lineage sometime around A.D. 1600. It was not possible, except for the last three title-holders, to locate the burial places of these - 334 high-ranking chiefs. Of the last three title-holders, only Sioeli Kalaekiivalu was buried in Niuatoputapu, on his estate at Hihifo (Ntt. 024). This was the last langi-type tomb built on Niuatoputapu. It is of similar height to earlier structures but is roughly built with crudely executed coarse unfitted blocks. The last two Maa'atu were buried at Mu'a, in Tongatapu, in chiefly graveyards, but not in langi. The other four named langi-type structures are situated in or near Vaipoa which oral history records as the seat of the Maa'atu line until the last century. Unfortunately, although the Maa'atu lineage was recorded in detail from Maa'atu Panuve by Gifford, 109 it cannot be correlated with the Vaipoa langi. Moreover, the burial sites of the earliest Maa'atus are not known and the single unnamed langi-style structure on the opposite side of the mountain at Houmafakalele (Ntt. 052) remains anomalously isolated, unfamed and forgotten.

Classes 3.3 and 3.4. Class 3.3 are low rectangular, or circular, earth platforms, whereas class 3.4 are high circular earth-mounds; both are found scattered mostly on the coastal periphery of the island; all are known locally as fa'itoka “burial sites” because they all have paepae “edging stones”. All 12 Class 3.3 sites and most of the 11 Class 3.4 sites are known locally simply by the names of the tracts in which they occur; rarely is anything known about their origins or historical relationships.

There are a considerable number of sites situated in two clusters along the liku weather coast, remote from present-day settlement. They are regarded locally as belonging to a period earlier even than oral tradition can account for and are called simply fa'itoka “graveyards”, at the liku “weather coast”. These sites are treated herein as regional (but not necessarily contemporaneous) complexes and are designated as the Atatuka and Houmafakalele complexes after the names of the areas in which they occur.

The Atatuka complex comprises three raised earth-mounds with paepae “stone edging”, all near the east coast where there is an old canoe passage of this name. The seven mounds comprising the Houmafakalele complex are also close to the coast about 900 metres south-east of Atatuka. Here there are five flat and one raised earth-mounds with stone edging, one of which is a langi-type structure. There is also one raised mound without edging. No local names or oral traditions concerning any of the Atatuka or Houmafakalele sites were recovered; several local people referred to the extensive mound and ditch of Atatuka II (Ntt. 041) as a kolo “fortress”; the unfaced mound at Houmafakalele (Ntt. 051) was called a sia; the remainder were called simply fa'itoka “burial mounds”.

Class 3.5; Contemporary Village Graveyards or fa'itoka. There are eight graveyards currently in use in Niuatoputapu. Falehau village has two graveyards: Ve'elangi (Ntt. 101, already described) and Taumaahina (Ntt. 102), the burial place of Fuimaono and his descent group.

Hihifo village has three graveyards in use: Peitolahi (Ntt. 103) in the centre of the village is the graveyard of the descent group of the title-holder Tupa; Nukufotu (Ntt. 104) and Taputangi (Ntt. 105) both south of and - 335 outside the village, contain members of the title-holder Telai's descent group and of the Ha'angatatupu title-holder Lapuka and his kinsmen respectively.

In the abandoned village of Matavai is a graveyard (Ntt. 106), now rarely used, containing the descendants and kinsmen of the title-holder Vivili. Kalevalio (Eng. Calvary) (Ntt.107) belongs to no particular village, being the Roman Catholic cemetery fronting the main road between Vaipoa and Hihifo villages. Haufolau (Ntt. 108), a former Roman Catholic cemetery situated on the main road near Kalevalio is now abandoned, but should be considered recent and post-contact. The main village graveyard of Vaipoa village (Ntt.109) is situated in the village beside the main coast road and actually surrounds the “langi” sites Tu'atangiketatau, Olosenga and Falepouahi.

Each of these graveyards contains a predominance of stone-lined family tombs (fonualoto), except the Roman Catholic Kalevalio cemetery where earth graves predominate. All of these graveyards except Kalevalio are considerably higher than the surrounding ground, owing to the common Tongan practice of heaping fresh sand over new or reopened graves, and most exhibit a scattering of smooth, small, black volcanic pebbles, kilikili, as decoration. Some small graveyards, especially those of descent groups, are enclosed by boundaries of paepae, with rectangular extensions to allow for more burial mounds. Several, for example, Ve'elangi, Vaipoa and Nukufotu, exhibit a mingling of aristocratic and commoner graves: in Ve'elangi the chief's tomb on the higher ground is surrounded by a paepae of small dressed limestone blocks; on the lower ground are the unfaced sand heaps of the commoners.

Of the 54 mounds described here for Niuatoputapu only 24 or 44.4 percent are named, the remainder either taking their name from the tract or area on which they are found or simply being known as “sia” or “fa'itoka”, depending on whether they have edging stones.

Class 4: Stone Boundaries, Markers, Former Sacred Stones, Quarry-sites

Makafaakimuli (Ntt. 070). Four large volcanic stones at the junction of Hala 'Uta “Inland track” and Hala Mo'unga “Mountain track” in the western corner of the tract called Avalua, are locally known as the Makafaa-ki-muli “Four-stones-?at-the-back”.

The stones have been implanted in a straight line spaced out over 6 metres. They have been partly dressed and are each roughly 50 cm high. Local informants were unable to explain them in any way: they serve now as an unnecessary boundary mark.

There are several remnants of low, dry-stone walls in tracts running up to the steep rock faces of the mountain; only three, Paamaka, Soopii, Ve'etoki (Ntt. 071, 072, 073) were recorded. Stone walls are rare on Niuatoputapu; on Tafahi, however, there are extensive stone walls on the mountain slopes.

Maka 'a Tangipaa (Ntt. 074 and 075). There are two named boulders in Vaipoa village known as the Maka'a Tangipaa “Stones of Tangipaa”: they are called Finekata “Laughing woman”, and Tukupeau “Break-water”. Finekata stands in the village allotment of that name and over- - 336 looks the lagoon, which dries for several hundred metres at low water. Tukupeau, on the other hand, stands some 500 metres seaward of Finekata in a small patch of yellow sand on the limestone reef base of the lagoon, where it is exposed at low tide but covered at high water.

Both stones are of dark, fine-grained basalt (maka hunu); Finekata is four-sided with rounded, smooth edges, 20 cm wide, 32 cm long and about 40 cm high, standing out of the ground like a thumb; Tukupeau is also four-sided with well pronounced angles, a little broader at the base (38 by 46 cm) than at the top (30 by 38 cm), and standing about 40 cm high. The surface of Finekata is smooth without incision, but Tukupeau has been engraved right round all four sides with a series of ten parallel lines about 25 mm apart. An incised line across the flat top surface of the stone is aligned due east-west as the stone stands embedded in position, but the stone itself is aligned 010 degrees on its long side.

Falehuufanga (Ntt. 076). This is a former sacred stone which, according to one informant, was once called Tu'i Niua and, according to another, once stood in Malolo-'ae-Hau, the site of the present Vaipoa cricket pitch. The stone is now placed at the back, eastern door of the Free Wesleyan Church in Vaipoa where the minister steps on it as he enters and leaves the church. It is a large hexagonal boulder of dark-grey basalt, remarkably regularly shaped, with flat top and bottom surfaces, 48.8 cm at its widest and 58.5 cm at its longest dimensions and 22 cm high. Around the sides are engraved six encircling parallel lines in the style described for the stone Tukupeau.

Quarry near Funga'ana (Ntt. 077). South-east from the two sites at Funga'ana (Ntt. 028 and 029), some 100 metres along the limestone uplift, is a place where the front face of the limestone overhang has slumped in towards the old lagoon. Here, limestone blocks have been quarried or partly quarried from the face. The blocks still in place appear similar to the larger ones in the two Funga'ana sites, but the quarry may also have provided slabs for some of the fa'itoka and langi-style structures.

Class 5: Fortifications, Ditches, Former Sunken Roads

Sites at Ataatuka (Ntt. 041), Tokelau (Ntt. 015), and possibly Haufakalaki (Ntt.009) may have been ancient earthworks associated with fortifications or defence. Two long, wide depressions without embankments were recorded in tracts called Avalua and Loto'aa, and may be former sunken roads 110, or they may be geological features. No traditions were recovered which relate specifically to any Class 5 sites.

Class 6: Luo maa “fermentation pits”

Throughout the island, but mainly on higher terraces where the soils are deeper and heavier, are deep pits now simply called luo maa “holes for fermented food”. Although the semi-anaerobic fermentation of foods is no longer practised in Niuatoputapu, these pits have been used within living memory for fermenting breadfruit, plantain, bananas, and cassava. Food - 337 storage pits must not be confused with old abandoned wells (vai tupu) although structurally the holes are often quite similar.

Class 7: Abandoned Post-historic Settlement Sites

Matavai Village (Ntt. 110). There are numerous high, oval earth-mounds, some of which are faced with stones, at the recently abandoned village of Matavai, only 10 minutes' walk from Hihifo village. Some at least of these mounds would have been the sites of buildings, such as churches or pastors', title-holders' and residents' houses before the main evacuation following the severe hurricane and epidemic in 1917-18.

Toma complex (Ntt. 111). This abandoned temporary settlement site is situated on the southern liku coast a few hundred metres south of the place called Toma. Here, old house-platforms were inhabited by people fleeing the influenza epidemic of 1918 which wrought havoc and widespread death throughout Tonga and Niuatoputapu in that year.

Class 8: Middens and Surface Pottery

There is a notable absence of middens around the present coastline except for contemporary and very recent ground-oven sites. These are hardly worth mentioning except that future archaeologists may be interested to learn that old oven-holes (ngoto'umu) are raked out, re-fuelled and lit; old stones, today mostly coral around the coast but igneous in the village, are re-used and left in situ after cooked food has been removed. It is to be expected, therefore, that an abandoned ground-oven will usually have all stones left in a neat circular concave layer seated on the ash and carbon of the fire. When fast-burning wood is used, very little ash remains between the stones.

Inland middens of non-concentrated type 111 are found in the belt of pottery-bearing middens running around the mountain. 112

SUMMARY, CONCLUSIONS, AND HYPOTHESES ARISING FROM THE REPORT

Apart from the pottery areas, there are altogether some 70 archaeological sites mentioned above: 36, or just over half, are locally believed to be burial sites; 5 are known as commemorative sites; the great majority of the remainder no longer have any special meaning or function for the present inhabitants.

The sites have been grouped into eight major categories in an attempt, where possible, to preserve local ethnographic knowledge and classification; most would fit into either McKern's 113 classification for Tonga or Davidson's 114 system for Vava'u, with the addition of a major category of stone mounds. 115

As archaeological excavation is lacking, a relative chronology cannot be reconstructed for the 80 sites described here and it is clear that the re- - 338 construction of Niuatoputapu culture history, and especially the interpretation of Lapita pottery settlement, must go hand in hand with excavation and geological interpretation.

The following summary of geological features will be used to suggest several hypotheses.

Geological Features

The area of land enclosed by the pottery belt consists of the central volcanic ridge and the rich soils of its lower slopes; this is about one seventh of the present land area. Outside the pottery belt is a great apron of low-lying land. Along the windward, south-eastern side of the ridge this apron is wide, consisting of a great expanse of coral boulders, swamp and sand-dunes, bush and scrub. Along the northern, lagoon side of the ridge, the apron is narrow and supports a profusion of coconut palms.

The central ridge is a very old volcanic basalt spine, about 107 metres above sea-level, with steep sides. High up on the mountain face in the tracts called Soopii and Ve'etoki is a 30-metre rock face where one can see laminations of scoria and bright-red tuff.

On both long faces of the ridge are at least two levels of terraces, the most pronounced being about 5 metres high with steep faces, the others being 2-3 metres high.

Immediately beneath the basalt faces of the ridge are red and black volcanic soils which extend over the terraces. Beneath the terraces is the belt of mixed fasifasi'ifeo soils marked by the Hala 'Uta and Hala Fakalava roads encircling the mountain.

Between the mixed soil zone and the northern, lagoon coast is a strip of sandy soil (tou'one). This strip is a typical beach ridge, slightly raised in the centre, along which runs the main road. The land descends seaward to the coast and inland towards the mixed soil margin. Almost at this juncture of the mixed and sandy soils is a series of depressions running parallel with the present coastline which may be (a) a former sunken road, or (b) the bed of a former restricted lagoon, or (c) a pre-flandrian sea-floor. Evidence for this depression may be seen inland of Angihoa, Ha'afisi, and Failoto, inland of the two mounds in Loto'aa, and at the eastern entrance of Hihifo village. If this depression were a former lagoon, the string of 20 or so sites along the northern coast could be sited on what was once the fringing reef of a former lagoon.

Along the southern windward side of the central ridge the mixed soil zone very gradually changes into sandy soil. Between this mixed zone and the coastline is a very wide flat low-lying apron of sand and coral boulders down the middle of which runs a low-lying swampy depression known as the Toafa “waste-land”. Towards the coast, the land rises — in places strewn with coral boulders and in places covered with a complex of old sand dunes and former beach ridges. This apron may be a recently uplifted fringing reef. The depression in the centre and the higher ground rising toward the coast are features exemplified in the modern fringing reef with its high outer perimeter.

- 339

The north-eastern arc of the island is a typical atoll “hook” with heavy sand accretion along its north-west leeward flank. It consists of an extensive plain of coral boulders lacking any soil horizon. The great expanse of the south-west bulge is similarly unsuitable for agriculture.

The site of Falehau village and the tract called Motu “Island” are raised islands of land surrounded by swamp or very flat low-lying land. Between Motu and Falehau is a stream called Me'eme'e still providing an outlet for the inland swamp called Toafa. North of Motu is another stream called Mataatolu, also providing an outlet for the swamp.

At least four volcanic necks have obtruded in an area between Falehau village and the swamp.

Pending excavation and geological interpretation, there are several hypotheses I should like to suggest:

(1) That the island was originally inhabited by bearers of Lapita pottery who discovered a small island rich in diverse resources: local materials suitable for pot and canoe manufacture; fine-grained basalt, obsidian, clay and suitable pottery tempers; volcanic soil for agriculture and trees; broad beach-lines for coconuts; lagoon and reef for sea-foods. The material so far described reveals little about the agricultural and sea-faring propensities of the Lapita people except that they made large ground ovens.

(2) That these people occupied a former beach-line now covered by the narrow belt of pottery-bearing mixed soils surrounding the mountain. This belt encircles the mountain directly below the lowest terrace (most pronounced at Faka'ahotaha on the south, Poome'e in the east and Hihifo village in the west) which could be Flandrian. That is, the Lapita people would have occupied a former beach-line sometime after 10,000 B.P. before it became covered by volcanic soil eroding down from the central ridge.

(3) That the time depth since Lapita settlement could be very great and commensurate with the earliest date proposed for Tongatapu Lapita settlement of 3090±95 B.P., or 1140 B.C. 116

(4) That the restricted distribution of Lapita pottery precludes its use in Niuatoputapu having continued into the era in which the island developed its present geographical form and extent. 117

(5) That the oral history associated with named mounds and burials does not extend back as far as the beginnings of the establishment of the Ha'a Falefisi lineage in Niuatoputapu around A.D. 1600 and that (a) these structures are post-Lapita, (b) the arrival of Lapita pottery in Niuatoputapu predates the arrival of the Ha'a Falefisi lineage in Niuatoputapu and is not associated with it as Kaeppler argues for Tungua. 118

(6) The Mata sites, the two huge stone mounds in the eastern extremity of the island, are unlike other stone mounds on the island; the lack of such - 340 mounds in the remainder of Tonga and their relative abundance in Western Samoa, suggest that these mounds are unlikely to have been built during the later era of Tongan influence begun by the Ha'a Falefisi, but could belong to an earlier “Samoic” period. They are situated on slightly raised terrain strewn with coral boulders, and like Motu and Falehau village sites may once have been isolated from the mainland by an extensive lagoon which is now toafa and swamp. While the era to which these anomalous monuments belong is still uncertain, the possibility that they were built by the Lapita voyager-traders 119 or their successors for astronomical purposes should not be overlooked. 120

It is my hope that this article will stir the interest of some archaeologist. 121

APPENDIX 1: CLASS IDENTIFICATION OF SITES ON NIUATOPUTAPU

Class 1: Named 'esi and stone mounds

  • Ntt 001 'Esi 'o Panuve
  • Ntt 002 Alokivakaloa ('Esi 'o Maa'atu)
  • Ntt 003 'Esi 'o Pilolevu
  • Ntt 004 Mataki'Uvea
  • Ntt 005 MatakiHa'amoa

Class 2: Named sia and unnamed, unfaced earth mounds

  • Ntt 006 Siamanafa
  • Ntt 007 Sia ko Finetengalelei
  • Ntt 008.1 Ha'afo'ou
  • Ntt 009 Haufakalaki
  • Ntt 010 Nofoatoa
  • Ntt 011 (Motu I)
  • Ntt 012 (Motu II)
  • Ntt 013 (Motu III)
  • Ntt 014 (Kolo)
  • Ntt 015 (Tokelau)
  • Ntt 051 (Houmafakalele II)

Class 3.1: Named langi style and chiefly burial mounds faced with dressed stones

  • Ntt 020 Tu'atangiketatau
  • Ntt 021 Olosenga and Falepouahi
  • Ntt 022 Tofi'a (Alofi'a)
  • Ntt 023 Langi 'o Seketo'a
  • Ntt 024 Langi 'o Maa'atu Sioeli Kiivalu

Class 3.2: Unnamed, square or rectangular mounds faced or partly faced with dressed stones on edge

  • Ntt 052 (Houmafakalele III)
  • Ntt 101.1 Ve'elangi I (old portion)

Class 3.3: “Fa'itoka” and low earth mounds with stone edging (paepae)

  • Ntt 025 (Angihoa)
  • Ntt 026 (Failoto)
  • Ntt 027 (Faakikava)
  • Ntt 028 (Funga'ana I)
  • Ntt 029 (Funga'ana II)
  • Ntt 030 (Hihifo I)
  • Ntt 031 (Tofi'a I)
  • Ntt 053 (Vaivai)
  • Ntt 054 (Houmafakalele IV)
  • Ntt 055 (Houmafakalele V)
  • Ntt 056 (Houmafakalele VI)
  • Ntt 057 (Houmafakalele VII)

Class 3.4: “Fa'itoka” and high earth mounds with stone edging (paepae)

  • Ntt 032 (Falehau I)
  • Ntt 033 (Falehau II)
  • Ntt 038 (Tomb of Sisipalaevae)
  • Ntt 040 (Atatuka I)
- 341
  • Ntt 034 (Falehau III)
  • Ntt 035 (Loto'aa I)
  • Ntt 036 (Loto'aa II)
  • Ntt 037 (Ve'elangi III)
  • Ntt 041 (Atatuka II)
  • Ntt 042 (Atatuka III)
  • Ntt 050 (Houmafakalele I)

Class 3.5: Contemporary Village Graveyards (fa'itoka)

  • Ntt 101.2 Ve'elangi II (new section)
  • Ntt 102 Taumaahina
  • Ntt 103 Peitolahi
  • Ntt 104 Nukufotu
  • Ntt 105 Taputangi
  • Ntt 106 (Matavai)
  • Ntt 107 Kalevalio
  • Ntt 108 Haufolau
  • Ntt 109 (Vaipoa)

Class 4: Stone Boundaries, markers, former sacred stones, quarries

  • Ntt 070 Makafaakimuli
  • Ntt 071 Paamaka
  • Ntt 072 (Soopii)
  • Ntt 073 (Ve'etoki)
  • Ntt 074 Finekata
  • Ntt 075 Tukupeau
  • Ntt 076 Falehuufanga
  • Ntt 077 (Quarry near Funga'ana)

Class 5: Fortifications, ditches, former sunken roads

  • Ntt 041 (already classified)
  • Ntt 014 (already classified)
  • Ntt 009 (already classified)
  • Ntt 078 (Avalua)
  • Ntt 079 (Loto'aa III)

Class 7: Abandoned settlement sites

  • Ntt 110 Matavai
  • Ntt 111 Toma

Unclassified sites

  • Ntt 057 (Fungamuihelu (grave))

Class 8: Pottery-yielding sites

  • Ntt 090 Lolokoka An extensive area of surface sherds in a tract just east of the eastern boundary of Vaipoa village. Site of test-pit excavation.
  • Ntt 091 Ha'afisi Surface sherds in an abandoned garden.
  • Ntt 092 Tu'akolo Nahau Surface sherds in an extensive area of the tract of this name, just west of Falehau village.
  • Ntt 093 Poome'e An extensive area of surface sherds from the Tu'akolo Nahau boundary of Poome'e eastwards to the beginning of the mountain slope. Site of two small test-pit excavations.
  • Ntt 094 Falehau village A few scattered surface sherds.
  • Ntt 095 Matavai tract Surface sherds in the tract adjoining the eastern corner of Falehau village.
  • Ntt 096 Tokelau Surface sherds on the lower side of the tract of this name near an indistinct mound.
  • Ntt 097 Faka'ahotaha Surface sherds on the mountain (inland) side of the road near the site of the 1969 Agricultural Farm.
  • Ntt 098 Lolopipi and adjoining tracts Extensive but very sparse sherds in the tracts called Lolopipi, Laafo'ou, Lolotonga, Tu'avao.
- 342
  • Ntt 099 Fanakava and Ha'angongongo Extensive but very sparse surface sherds.
  • Ntt 100 Loto'aa Extensive surface pottery between the house of the Agricultural Officer and Hala 'Uta towards the mountain. Site of pit excavation.
APPENDIX 2: SAND TEMPERS IN SHERDS FROM NIUATOPUTAPU AND ELSEWHERE IN TONGA

Sixteen sherds collected in Tonga by G. A. Rogers, 9 from sites in Niuatoputapu and 6 from sites in Ha'apai, together with one only from Tongatapu, were examined in thin section and compared with thin sections of 14 other sherds from Tonga sent by F. Shutler (5 from Tongatapu), J. Davidson (3 from Vava'u), and Y. H. Sinoto (6 from Ha'apai). Through the courtesy of the Australian National University, I was also able in 1971 to examine the thin sections of sherds from Tongatapu studied by C. Key 122. This note is a general appraisal and comparison of the sand tempers from these groups of sherds.

Tempers in all the Tongan sherds studied in thin section are similar volcanic sands of simple mineralogy except for a few which contain admixtures of calcareous sand that represents detritus from reefs. The degree of rounding varies, as does the placer concentration of heavy minerals. Pyroxene and plagioclase crystals, and volcanic rock fragments, are the dominant grain types. Quartz grains and opaque iron oxide grains are subordinate constituents. The restricted range in composition suggests that the sands were collected from related volcanic islands of island arc type. The chain of insular volcanoes along the western side of Tonga is the nearest source of volcanic detritus having a mineralogical character comparable to that of the tempers. Relative abundances of characteristic grain types are variable, and it seems likely that various kinds of sands from beaches, ravines, and pyroclastic accumulations on several islands may be represented.

Pyroxenes, with the clinopyroxene augite dominant but the orthopyroxene hypersthene also common, are the predominant ferromagnesian silicates, although minor amounts of olivine or hornblende occur in some sherds. The only feldspar is plagioclase, and it is ten or more times as abundant as quartz, which never forms more than a few percent of the temper. Volcanic rock fragments range from wholly glassy or hyalopilitic grains, in part pumiceous, that were presumably derived from felsic andesite or dacite, to more crystalline grains of intersertal or intergranular texture derived presumably from mafic andesite or basalt. Roughly 5 percent of typical tempers are opaque iron oxides. There are no clear mineralogical criteria by which to distinguish sherds from Vava'u, Ha'apai, - 343 and Tongatapu, for each group includes a similar array of tempers, but some Niuatoputapu tempers appear distinctive (see below).

Key 123 has raised the possibility that the temper sands may have been imported from outside Tonga, and suggests Fiji as a source. Although importation is difficult to disprove, I consider it unlikely, for the volcanic sand tempers are of the sort to be expected as derivative from volcanoes of the type common along the west edge of Tonga, but are unlike any sherd tempers described to date from Fiji.

Any attempt to identify specific sources of volcanic sand temper among the various volcanic islands is inherently difficult, for the lavas of the Tongan volcanoes are partly similar from island to island. According to Bryan and others, 124 the volcanic rocks of Tonga are distributed roughly as follows: (a) Fonualei: dacite with some felsic andesite; (b) Late: mafic andesite with some felsic andesite; (c) Kao: felsic andesite with some mafic andesite; (d) Tofua: felsic andesite with some dacite and mafic andesite; (e) Hunga: mafic and felsic andesite. No comparable data from Niuatoputapu or Tafahi have been reported to date, but there are no reasons to suppose that their geology or petrology is grossly different. 125

NIUATOPUTAPU TEMPERS

Three distinctive types of temper sand are present in sherds from Niuatoputapu: calcareous sand, ferromagnesian sand, and pumiceous sand. The character of all three suggests that they were probably collected locally on Niutoputapu or Tafahi, although closely similar sands might occur elsewhere in Tonga and do appear in a few sherds collected elsewhere in Tonga:

1. Calcareous sand: In one sherd (III) from Loto'aa, three-quarters of the grains are rounded calcareous fragments. The remainder is volcanic black sand similar to the ferromagnesian temper. Nearly half the grains in one sherd (VIII) with ferromagnesian temper from Ha'afisi are also calcareous. Well-rounded and well-sorted calcareous sands of this type are common as insular beach sands where fringing reefs occur.

2. Ferromagnesian sand: In one sherd each from Loto'aa (II), Ha'afisi (VIII), and Tokelau (IX), the temper is a volcanic black sand containing 85-95 percent ferromagnesian silicate mineral grains with small admixtures of calcareous grains (<5 per cent except as noted above). Plagioclase feldspar grains and partly crystalline volcanic rock fragments are minor components, amounting each to no more than 5 percent of the total grains. Among the ferromagnesian grains, the ratio of pyroxene to olivine is 9:1 or more; and among the pyroxenes, the ratio of clinopyroxene to orthopyroxene is 9:1 or more. The temper apparently differs slightly from other ferromagnesian tempers from elsewhere in Tonga, as discussed below, in the extreme proportion of ferromagnesian grains, whose abundance is less than 85 percent in other Tongan sherds, except for one from Tungua (XIII), and in the presence of subordinate olivine grains, which occur in - 344 trace amounts only in other Tongan sherds. The apparent differences between the Niuatoputapu ferromagnesian temper and other related tempers from elsewhere in Tonga are not strong, and will probably not prove to be reliable criteria.

3. Pumiceous sand: In five sherds from Niuatoputapu, the dominant grains are glassy, pumiceous volcanic rock fragments with brownish hues in thin section. Except for one sherd from Nomuka (XI), volcanic rock fragments are not so predominant in other Tongan sherds, although lesser amounts of similar glassy and pumiceous grains occur in tempers from elsewhere. In two sherds from Loto'aa (IV, V) and in one from Ha'afisi (VIIb), the pumiceous grains are the only significant grain type, forming 80 ±5 percent of the sand with plagioclase grains 10-15 percent and pyroxene grains 5-10 percent. In two other sherds, one from Loto'aa (I) and one from Pome'e (VIIa), the pumiceous grains are less prominent at 50-60 percent, whereas plagioclase is about 20 percent and pyroxene is about 25 percent. All these pumiceous sands are either only moderately sorted or moderately rounded, or both, and may be slightly reworked ash deposits. It is unlikely, however, that such sands within Tonga are unique to Niuatoputapu.

OTHER TONGAN TEMPERS

In sherds from Vava'u, Ha'apai, and Tongatapu, temper sands are composed mainly of various proportions of pyroxene grains (and minor olivine), plagioclase feldspar grains (and minor quartz), and volcanic rock fragments of varied character. Denoting pyroxene as P, plagioclass (feldspar) as F, and volcanic rock fragments as V, the relative proportions of the three main constituents can be expressed as PxFyVz, where the subscripts are frequency percentages (recalculated such that x + y + z = 100) obtained from traverse counts of 100-500 grains in thin section. By this convention, the ranges in composition of tempers in sherds studied to date are as follows: for Tongatapu, P50-75F10-30V5-20; for Ha'apai, P20-90F5-35V5-55; and for Vava'u, P25-75F10-30V10-50. Calcareous grains occur in varying proportions in some sherds from each of the three sets. The spectrum of compositional variation within each set of tempers is thus greater than the contrast in mean composition between sets, and the variations are almost wholly overlapping.

Tempers from Ha'apai include: (a) pyroxenic variants (XII from Matuku and XIII from Tungua of Rogers as well as two sherds of Kaeppler-Sinoto from Tungua): P80-90F5-10V5-10; (b) lithic variants (XIV of Rogers from Kotu as well as three sherds of Kaeppler-Sinoto from Tungua):P˜40F15-35V30-40; (c) intermediate variants (VI of Rogers from Lifuka, with 45 percent admixed calcareous grains, as well as one sherd of Kaeppler-Sinoto from Tungua): P60-70F5-15V20-30; and (d) a pumiceous variant (XI of Rogers from Nomuka): P20F25V55, where three-quarters of the volcanic rock fragments are pumiceous grains. The lithic variants are closely matched with some sherds from Vava'u (P25-35F˜25V40-50) and the intermediate variants with some sherds from Tongatapu (P50-65F20-35V15-20); similar pyroxenic variants (P˜80F10-15V5-10) occur in sherds from both Tongatapu and Vava'u.

- 345

The degree of quantitative variability indicated among tempers in sherds collected from geographically related sites effectively rules out firm interpretations of temper origins in cases like Tonga where the grain types in all the temper variants are qualitatively similar, if not essentially identical. I conclude that the tempers in the Tongan sherds are indigenous to Tonga, with origins from the volcanic islands, but that positive identification of specific temper sources is precluded by the nature of the available data.

APPENDIX 3: SOURCE OF OBSIDIAN FROM NIUATOPUTAPU SITES

The natural volcanic glass recovered during survey and excavation of the Lapita sites at Niuatoputapu exhibited similarities in hand specimen to that from the extinct island volcano Tafahi which is approximately 7 km from Niuatoputapu. In order to assess the probability that the site obsidian derived from the Tafahi source, a number of samples were submitted for trace element analysis.

Characterisation of the obsidian in terms of the proportion of four trace elements was made using X-ray fluorescence spectrographic techniques. 126 The results of the analysis are presented in tabular form in Table 2.

The results of the trace element analysis indicate that it is highly likely that the natural volcanic glass from the Lapita sites at Niuatoputapu derives from the Tafahi source.

TABLE 2
Results of trace element analysis
  accession numbers   element composition in ppm    
    Zr Mn Rb Sr
Niuatoputapu Sites NTT AS 031 18 1917 27 202
  AS 032 27 1750 08 200
  AS 034 39 1946 18 206
  AS 035 10 2318 14 200
  AS 036 26 1720 26 227
  AS 037 38 1842 20 204
LK AS 039 13 1819 10 141
  AS 040 18 1804 09 154
  AS 041 55 1776 10 219
Tafahi Source Tefitomaka 2 GT 851 49 1325 23 230
  GT 852 57 1438 32 258
Tefitomaka 3 GT 853.2 52 1145 20 239
  GT 853.3 40 1271 26 211
Tefitomaka 4 GT 854.2 66 1307 27 227
  GT 855.2 53 1431 25 236
  GT 857.2.1 62 1714 25 243
  GT 857.2.2 93 1223 30 233
  GT 858.2 58 1350 24 234
  GT 855.6 50 1331 29 226
Tafahi Source Mean   58.0 1354.6 26.1 233.4
N = 10 SEM (P=0.05)   10.2 110.2 2.6 8.7
% SEM   17.6 8.1 9.8 3.7
- 346
REFERENCES
  • BARNARD, Noel, (ed.), 1972. Early Chinese Art and its Possible Influence in the Pacific Basin. Vol. 3. New York, Intercultural Arts Press.
  • BEAGLEHOLE, John C. (ed.), 1967. The Journals of Captain Cook on his Voyages of Discovery. Vol. 3, parts 1 & 2. The Voyage of the Resolution and Discovery, 1776-1780. Cambridge University Press, for the Hakluyt Society.
  • BIGGS, Bruce, 1971. “The Languages of Polynesia,” in T. A. Sebeok (ed.), Current Trends in Linguistics, Vol. 8, Linguistics in Oceania, The Hague, Mouton, pp. 466-505.
  • BIRKS, Lawrence, 1973. Archaeological Excavations at Sigatoka Dune Site, Fiji. Suva, Bulletin of the Fiji Museum, No. 1.
  • BRYAN, W. B., G. D. STICE, and A. EWART, 1972. “Geology, Petrography and Geochemistry of the Volcanic Islands of Tonga.” Journal of Geophysical Research, 77:1566-85.
  • BUIST, A. G. 1969. “Field Archaeology on Savai'i,” in Green and Davidson (eds.), 1969:34-68.
  • CHASE, G. C., 1971. “Tectonic History of the Fiji Plateau.” Geological Society of America Bulletin, 82:3087-110.
  • DAVIDSON, Janet M., 1964. The Physical Analysis of Refuse in New Zealand Sites. Unpublished M.A. thesis. University of Auckland.
  • — 1971. “Preliminary Report on an Archaeological Survey of the Vava'u Group, Tonga,” in R. Fraser (comp.), Cook Bicentenary Expedition in the South-West Pacific. Wellington, Royal Society of New Zealand Bulletin 8, pp. 29-40.
  • DICKINSON, William R. and Richard S. SHUTLER Jr., 1968. “Insular Sand Tempers of Prehistoric Pottery from the Southwest Pacific,” in Yawata and Sinoto (eds.), 1968: 29-37
  • FAIRBRIDGE, Rhodes W., 1961 “The Melanesian Border Plateau, a Zone of Crustal Shearing in the Southwest Pacific.” Publications du Bureau Central Séismologique Internationale, Ser. A, 22: 137-49.
  • FREEMAN, O. W. (ed.), 1951. Geography of the Pacific. New York, Wiley.
  • GARANGER, José, 1971. “Incised and Applied-relief Pottery, its Chronology and Development in Southeastern Melanesia, and Extra Areal Comparisons,” in Green and Kelly (eds.), 1971:53-66.
  • GIFFORD, Edward W., 1923. Tongan Place Names. Honolulu, B.P. Bishop Museum Bulletin, 6.
  • — 1929. Tongan Society. Honolulu, B.P. Bishop Museum Bulletin, 61.
  • GOLSON, Jack, 1971. “Lapita Ware and its Transformations,” in Green and Kelly (eds.), 1971:67-76.
  • — 1972a. “Both Sides of the Wallace Line: New Guinea, Australia, Island Melanesia and Asian Prehistory,” in Barnard (ed.), 1972:533-95.
  • — 1972b. “The Pacific Islands and their Prehistoric Inhabitants”, in R. G. Ward (ed.), Man in the Pacific Islands; Essays on Geographical Change in the Pacific Islands. Oxford, The Clarendon Press, pp. 5-33.
  • GREEN, Roger C., 1972. “Revision of the Tongan Sequence.” Journal of the Polynesian Society, 81:79-86.
  • — 1973. “Lapita Pottery and the Origins of Polynesian Culture.” Australian Natural History, June: 332-7.
  • GREEN, Roger C., in press. “Pottery from the Lagoon at Mulifanua, Upolu,” Report 33 in R. C. Green and J. W. Davidson (eds.), Archaeology in Western Samoa, Vol. 2. Auckland, Auckland Institute & Museum Bulletin, 7.
  • Green, Roger C., and Janet W. DAVIDSON (eds.), 1969. Archaeology in Western Samoa, Vol. 1. Auckland, Auckland Institute and Museum Bulletin, 6.
- 347
  • GREEN, Roger C. and Marion KELLY (eds.), 1971. Studies in Oceanic Culture History, Vol. 2. Pacific Anthropological Records No. 12. Honolulu, Department of Anthropology, B.P. Bishop Museum.
  • GROUBE, Leslie M., 1971. “Tonga, Lapita Pottery, and Polynesian Origins.” Journal of the Polynesian Society, 80:278-316.
  • HEDRICK, John D., 1971. “Lapita Style Pottery from Malo Island.” Journal of the Polynesian Society, 80:5-19.
  • HOFFMEISTER, J. E., 1932. Geology of 'Eua, Tonga. Honolulu. B.P. Bishop Museum Bulletin, 96.
  • HOFFMEISTER, J. E. and H. L. ALLING, 1932. “Petrology,” in Hoffmeister 1932:39-47.
  • KAEPPLER, Adrienne L., 1971. “Eighteenth Century Tonga: New Interpretations of Tongan Society and Material Culture at the Time of Captain Cook.” Man, 6, (2):204-20.
  • — 1973. “Pottery Sherds from Tungua, Ha'apai: and Remarks on Pottery and Social Structure in Tonga.” Journal of the Polynesian Society, 82:218-22.
  • LEWIS, David, 1972. We, the Navigators; The Ancient Art of Landfinding in the Pacific. Wellington, Reed.
  • MARTIN, John (ed.), 1827. An Account of the Natives of the Tonga Islands . . 2 vols. Edinburgh, Constable. 3rd Edition.
  • MCKERN, W. C., 1929. Archaeology of Tonga. Honolulu, B.P. Bishop Museum Bulletin, 60.
  • MEAD, S. M. et al., 1973. The Lapita Pottery Style of Fiji and its Associations. Polynesian Society Memoir No. 38. Supplement to the Journal of the Polynesian Society, 82 (3-4).
  • OLIVER, Douglas, L., 1962. The Pacific Islands. Cambridge, Harvard University Press. Revised Edition.
  • PALMER, J. Bruce, 1966. “Lapita Style Potsherds from Fiji.” Journal of the Polynesian Society, 75:373-7.
  • — 1971. “Fijian Pottery Technologies: Their Relevance to Certain Problems of Southwest Pacific Pre-history,” in Green and Kelly (eds.), 1971:77-103.
  • — 1972. “Pottery in the South Pacific,” in Barnard (ed.), 1972:693-721.
  • POULSEN, Jens, 1966. “Preliminary Report on Pottery Finds in Tonga.” Asian Perspectives, 8:184-95.
  • — 1967. A Contribution to the Prehistory of the Tongan Islands. Unpublished Ph.D. Thesis. Canberra, Australian National University.
  • — 1968. “Archaeological Excavations on Tongatapu,” in Yawata and Sinoto (eds.), 1968:85-92.
  • ROGERS, Garth, 1973. Report on Archaeological Survey in Niuatoputapu Island, Tonga. Working Paper in Archaeology, 28. University of Auckland, Department of Anthropology.
  • SCLATER, J. G., J. W. HAWKINS, J. MAMMERICKX, and C. G. CHASE, 1972. “Crustal Extension between the Tonga and Lau Ridges; Petrologic and Geophysical Evidence.” Geological Society of America Bulletin, 83:505-18.
  • SCOTT, Stuart, D., 1969. “Reconnaissance and some Detailed Site Plans of Major Monuments of Savai'i,” in Green and Davidson (eds.), 1969:69-95.
  • SHEPARD, Francis P., 1968. The Earth Beneath the Sea. New York, Athenium.
  • SPECHT, Jim, 1968. “Preliminary Report of Excavations on Watom Island.” Journal of the Polynesian Society, 77:117-34.
  • SPILLIUS, Elizabeth, 1960. Tongan Society at the Time of Captain Cook's Visit. Typescript. The Palace Office, Nuku'alofa.
- 348
  • TAYAMA, Risaburo, 1952. Coral Reefs of the South Seas, Vol. II. Hydrographic Office Bulletin 11, Tokyo, Maritime Safety Agency.
  • THOMAS, William L., Jr., 1965. “The Variety of Physical Environments Among Pacific Islands,” in F. R. Fosberg (ed.), Man's Place in the Island Ecosystem, A Symposium. Honolulu, Bishop Museum Press, pp. 7-38.
  • VAN DER LINDEN, W. J. M., 1966. “Southwest Pacific Ocean,” in R. W. Fairbridge (ed.), Encyclopedia of Oceanography. New York, Rheinhold.
  • WARD, Graeme K., 1972. Obsidian and New Zealand Archaeology: A Paradigm for Sourcing Artefact Assemblages Using X-ray Fluorescence Spectography. Unpublished M.A. (Hons.) Thesis, Department of Anthropology, University of Otago.
  • YAWATA, I., and Y. H. SINOTO (eds.), 1968. Prehistoric Culture in Oceania. Honolulu, Bishop Museum Press.
  • YUNCKER, T. G., 1959. Plants of Tonga. Honolulu, B.P. Bishop Museum Bulletin 220.
1   Field research in archaeology was ancillary to investigations into Tongan social organisation as a holder of a New Zealand University Grants Committee postgraduate scholarship for study towards a Ph.D. degree from July 1969 to December 1971. I wish to thank the University of Auckland Anthropology Department for a separate grant supporting field work, and Cyril Schollum for photographic assistance. I am indebted to Lawrie Birks, Janet Davidson, and Professor Roger Green for reading and commenting on an earlier draft of this paper, and to Helen and Lawrie Birks for the opportunity to examine and handle their collection of reconstructed vessels from Sigatoka, Fiji.
2   I am grateful to the Prime Minister of Tonga, H.R.H. Prince Tu'ipelehake, for permission to remove these national treasures from Tonga for analysis; they are now being held on deposit for the Tonga Government by the Auckland Institute and Museum pending the establishment of a museum in Tonga.
3   I should like to thank Professor Dickinson of Stanford University and Graeme Ward of the University of Otago for sparing time to make these analyses, and Karel Peters of the University of Auckland for his technical skill in preparing the pottery slides.
4   Rogers 1973, available on request from the Anthropology Department, University of Auckland.
5   McKern 1929:104, 109-113. McKern's collection (1587 sherds) is deposited at the Bernice P. Bishop Museum, Honolulu, Hawaii (Kaeppler 1973, personal communication).
6   Poulsen 1968:91.
7   Davidson 1973, personal communication.
8   Groube 1971:293, see also Golson 1971:74.
9   Kaeppler 1973:218.
10   Davidson 1971:37-8.
11   Green (In Press, Report 33).
12   Biggs 1971:491. Jacob Le Maire recorded the vocabulary while anchored off Tafahi (Cocos) Island.
13   Biggs 1971:467.
14   See Thomas 1965:14-5; Shepard 1967:84-9.
15   For example, Oliver 1962:11-3; Tayama 1952, II:1-292; Thomas 1965:15-22.
16   1932:39-40.
17   1972, quoted by Dickinson, appended.
18   1968:32-4.
19   See, for example, Dickinson and Shutler 1968:33 fig. 1; Golson 1972:535, Map 1; Freeman 1951:22-3; Van der Linden 1966:848-50, especially Fig. 3; and Fairbridge 1961.
20   Groube 1971:291-2.
21   Ibid.:291.
22   The figures of pottery sherds and artefacts were drawn by Karel Peters of the University of Auckland.
23   Groube 1971:292.
24   Specht 1968:123.
25   Hedrick 1971:11.
26   I should like to thank Pamela Raspe, graduate student of the University of Auckland Anthropology Department, for her able help and co-operation in analysing, computing and describing the collection.
27   Cf. McKern 1924:115; Golson 1971:70-1, 73.
28   Reported by Poulsen (1967:177, 191) for Tongatapu; Specht (1968:128) for Watom; Hedrick (1971:14) for Malo; Green (In Press) for Western Samoa; and discussed by Palmer (1971:80, 84) and Birks (1973:21) for Fiji.
29   Poulsen 1967:271.
30   Birks 1973:40-1.
31   Green (In Press).
32   Poulsen 1967:175, and see Golson 1971:70.
33   Cf. Green (In Press) Fig. 71b.
34   Cf. Birks 1971:21.
35   The descriptive terminology which follows is taken from Mead 1973.
36   Mead 1973:22 and his Fig. 2.45.
37   See Specht 1968:129-132 for a cogent description.
38   See Birks 1973:26 and his Fig. 29 for a description.
39   Our terminology, but Mead's M12.2, ibid.:32, Fig. 2.39.
40   Poulsen 1967:367-8; and discussed by Specht 1968:131-2.
41   Cf. Hedrick 1971:15.
42   See Fig. 3a, Mead's DE 8.2 and cf. Specht 1968:128, and Garanger 1971:64, Fig. 11:5.
43   Fig. 3b, d, Mead's DE 8.1.
44   Fig. 3c, Mead's DE 8.3.
45   Mead's M12.2; our Fig. 3g.
46   1972:83, Fig. 5 K4.
47   (In Press) Fig. 71k.
48   Birks 1973, personal communication.
49   1967:271.
50   1973:40-1, especially Fig. 38:124 & 132.
51   (In Press).
52   See Birks 1973:41 for a discussion of distribution and of possible function of pottery discs.
53   Poulsen 1967:161f.
54   1971:61.
55   1973:93, vessel 86.
56   1967:124, 127-8 and Table 35.
57   Green (In Press) Fig. 71b. See also Poulson's Class 9 (1967:74); and Birks (1973: Figs. 9-21).
58   See Poulsen 1967:172-3.
59   Fig. 4L is solid; Fig. 4n is built up from a slab of clay and is partly hollow.
60   Birks 1973:38-40 and personal communication. See also Specht 1968:120 & 127 for a report of a pottery “leg” from his Vunailau site (11) on Watom.
61   Poulsen 1966:184-5; 1967:191-2, 367f; Specht 1968:127-32; Golson 1971:68-71, 73-4; Hedrick 1971:16-7; Birks 1973: passim and see 58-9, 64-7 for an excellent discussion of general distribution.
62   See Kaeppler 1973.
63   I am indebted to Janet Davidson and Roger Green for the opportunity to see materials from Vava'u and Mulifanua.
64   Cf. similar observations by Specht (1968:127-8) for Watom; Green (1969:170) for Western Samoa; see Golson (1970:69) for a contrary view.
65   1969:21, 24.
66   1967:231-4.
67   Mariner (Martin 1827, I:225) states that rats were eaten by commoners in the early nineteenth century. I should like to thank Robin Watt for generous assistance in identifying faunal remains.
68   A large-scale map giving the location of each archaeological site and the name of the tract of land on which it is situated has been deposited with the artefactual material in the Auckland Museum. The map produced here was drawn by Karel Peters from a Tonga Government base map and sketch maps made in the field. All measurements in this paper are approximate. I wish to acknowledge here my indebtedness to Tavi Maupiti of Tonga for his companionship and stimulating discussions on enigmas in Tongan archaeology and culture history.
69   1929:10 and passim.
70   1971:30 and passim.
71   For example, Ntt. 007 & Ntt 008.1, see Appendix 1.
72   For example, Ntt. 001 & Ntt 003, ibid.
73   McKern (1929:10) states; “The use of these mounds seems to have been connected with the idea, prevalent in Tonga, that the exalted social position of the chieftain should be accompanied by material exaltation, and also that social superiority involves the privilege of isolation from common contacts.”
74   Cook recorded 'esi on 'Eua (Beaglehole 1967, Vol. 3, 1:157); McKern (1929:10-19) reported 'esi for Tongatapu, Ha'apai, Vava'u; Davidson (1971:34) for Vava'u.
75   The name Alo-ki-vaka-loa is found in Tongan poetry; alo “to paddle”; vaka-loa refers to the dark lenticular clouds which gather near the horizon towards sunset representing canoes (vaka) manned by demons (nifo-loa) accompanying a royal person voyaging at sea.
76   Especially 'uu'uuBirgus latro — coconut-robber-crab”.
77   There are multi-tiered monuments both in Samoa and throughout Tonga.
78   Fuimaono Leonaitasi 1971, personal communication.
79   1929:9.
80   ibid.:66.
81   See Lewis 1972:316-22.
82   1929:8.
83   Buist 1969:39 and Scott 1969:71.
84   Buist 1969:39, 55-7.
85   Scott 1969:71.
86   1929:15.
87   Davidson 1971:34 and Fig. 2 plate 1.
88   1923:199.
89   Gifford 1929:344.
90   quoted by McKern 1929:18.
91   1929:294 and 344.
92   Described by Martin 1827 (II):222-3; McKern 1929:19-27; Gifford 1929:345-6; Davidson 1971:31, 35; (anomalously, the one siaheulupe found on Tafahi is a rectangular, stone-faced platform).
93   1929:19-27.
94   1971:34.
95   See Davidson 1971:31, 34.
96   See McKern 1929:8-9, 12, 18, 22, 23; Davidson 1971:31, 34, Plate 1, Fig. 2.
97   See McKern 1929:33. Langi and fa'itoka may describe identical structures; it is a matter of linguistic usage, langi being used only for the tombs of the Tu'i Tonga descendants.
98   Gifford 1929:78.
99   McKern 1929:33.
100   Genealogical records, Palace Records Office, Nuku'alofa; Anderson's Journal and Cook in Beaglehole 1967:880, 115.
101   Spillius 1960.
102   1971:207-8.
103   Gifford 1929:135 quoting Tu'ivakanoo, “. . . [the title of Maa'atu] belongs to Tu'i Tonga family of very olden times. He was practically king of Niuatoputapu and had nothing to do with the title on Tongatabu, but created his own chiefs.”
104   1971:31.
105   Gifford 1929:300, 312-4.
106   1929:33.
107   1971:31-2.
108   1929:75.
109   1929:284-5.
110   See McKern 1929:89; Davidson 1971:35-6.
111   See Davidson 1964:85-7.
112   These pottery-bearing middens have been discussed in Part II; they are listed and numbered in Appendix 1.
113   1929:10.
114   1971:30.
115   cf. McKern 1929:8.
116   Groube 1971:303.
117   cf. Groube 1971:306: “It is abundantly clear to the author that, unless new evidence is forthcoming, claims of persistence of decorated pottery in Tonga [i.e. Tongatapu] after about 500 B.C. cannot be sustained. The survival of plain pottery into the Christian era is also problematical but this issue must await evidence from areas outside the richness of the Tongatapu lagoon”. (See Green 1972:79 & 84 for a similar view.)
118   Kaeppler 1973:220.
119   See Green 1973:335.
120   See Lewis 1972:316-22.
121   Paulo Faka'osi has reported the discovery of the first potsherd from Tafahi Island (Tavi Maupiti, personal communication 1973).
122   e.g. Groube 1971:ft. nt. 55.
123   in Groube 1971.
124   1972.
125   e.g. Chase 1971; Sclater et. al. 1972.
126   vide Ward 1972:128ff.