Volume 98 1989 > Volume 98, No. 4 > A ceramic sherd from Ma'uke in the Southern Cook Islands, by R. Walter, p 465-470
                                                                                             Previous | Next   

- 465

During excavations at the Anai'o site on the island of Ma'uke in 1987 a single potsherd (sample number AN 501) was recovered from a dated stratigraphic context. A second sherd (sample number AN 700) was part of a surface collection recovered from the site several months later. 1

A sand temper analysis was carried out on sherd AN 501 with results indicating a possible source for the temper material in the island arcs fringing the south-west Pacific basin. We believe that the most probable specific source lies in the Tongan island group, most likely on Tongatapu itself. On the basis of microscopic examination of temper inclusions, AN 700 is argued to be of similar origin. These findings have important implications for Cook Island prehistory and in regard to estimates of the upper time span of the Tongan ceramic tradition. The results of radiocarbon analysis place the AN 501 sherd within a 14th century A.D. occupation period (see Table 1).

Island Setting

Ma'uke is a raised coral reef island of approximately 1800 ha lying about 245 km north-east of Rarotonga. It has a central volcanic rock core rising to a height of 30 m and fringed by a raised coral reef platform (makatea). The central core consists of deeply weathered basaltic rocks dissected by a number of streams which flow into depressions at the intersection of the central volcanics and the makatea limestone. These depressions are infilled with basaltic alluvium and are used for growing taro. The water eventually drains out to sea in underground streams beneath the makatea.

The coastal soils are derived from water-laid and wind-blown coral sands and gravel. The inland soils derive from weathered basaltic alluvium in the interior lowlands, and from weathered basalt in the interior uplands (Wilson 1982:8-11).

The Anai'o Site

The Anai'o site is a prehistoric settlement site located on the west coast of Ma'uke approximately 1 km south of the main coastal village of Kimiangatau. The site is located on a storm beach ridge adjacent to the inland reef beds. The total area of the site is estimated to be approximately 2,000 m2 (Walter 1987:239).

Two cultural horizons were found at Anai'o. Occupation 1 (Layer 4), which was the layer from which sherd AN 501 derived, lay at an average depth of about 65 cm and represented a village settlement constructed on a cleared beach ridge. The excavated areas of the settlement exposed a number of dwelling structures, cooking houses, storage pits and manufacturing floors. These latter features included areas associated with the working of pearl-shell as well as stone flaking floors. Pig and dog - 466 bones as well as the bones of inshore and offshore fish species were found in the midden associated with the cooking features. Artefacts associated with this occupation included one-piece and two-piece hooks made of pearl-shell, shell ornaments, spear points and vegetable scrapers. Adze types included quadrangular, reversed triangular, lenticular and triangular types.

Occupation 2 (Layer 2) was separated from the first occupation horizon by about 10cm of wind-and wave-borne sands. The second occupation contained few features but the artefacts were similar to those recovered in Occupation 1.

Four samples of Turbo setosus marine shell from Occupation 1 were submitted to the D.S.I.R., Institute for Nuclear Sciences, Wellington, for radiocarbon analysis. The conventional radiocarbon ages were corrected for the marine reservoir effect taking account of regional ocean variation (delta-R) following Stuiver, Pearson and Braziunas (1986). We have used a delta-R value of 45 ± 30 which is the value determined for the Society Islands by Stuiver et al. 1986 (Table 1 and Fig 10B) and which we consider is the most appropriate estimate to apply in the Southern Cook Islands (see Table 1).

TABLE 1 14C Determinations from Occupation 1, Anai'o
Lab. No. Conventional Age. delta 13 Calibrated Age Range 1
N.Z. 6939 1075 ± 48 B.P. 2.6 ± 0.1 1337-1435 A.D.
N.Z. 6960 1015 ± 35 B.P. 3.9 ± 0.1 1360-1434 A.D.
N.Z. 6984 1026 ± 24 B.P. 2.9 ± 0.1 1348-1424 A.D.
N.Z. 6943 1055 ± 58 B.P. 2.5 ± 0.1 1307-1422 A.D.
1   At one standard deviation.
The Sample

The AN 501 sherd was recovered from Occupation 1 (Layer 4) in excavation Area 250 at a depth of 90 cm. This area of the occupation surface was located outside the eastern wall of a large structure and contained a number of storage pits and portable artefacts. These artefacts included several complete fishhooks and one hook blank, some coral grinders (probably used for the maintenance of shell fishing gear) a pearl-shell neck ornament and some slingstones.

The sherd sample (AN 501) was 35 mm by 29 mm with a width of 4 mm. Sample AN 700, which was collected from the surface of the site, was 18 by 13 mm with a width of 4 mm. Both sherds were slightly concave/convex and formed part of the body of the pot vessel(s). They were undecorated course-grained sherds of an orange/brown colour with black temper inclusions. On both sherds the concave side was of a darker colour.

- 467
Temper Composition

The sand temper is moderately well sorted, subangular to subrounded volcanic sand rich in pyroxene mineral grains, and was probably collected from a beach placer containing a concentrate of dark heavy minerals along the shoreline of a volcanic island (Dickinson and Shutler 1971:194; 1979:Fig. 3). No grains of calcareous reef detritus are present, and their absence makes a source anywhere within the Cook Islands unlikely. Certainly, unweathered black sand like that of the sherd temper could not have been collected on Ma'uke.

Frequency counts of grain types in two thin sections prepared by Ruperto Laniz of Stanford University from slices obtained from the edges of the sherd were made using a petrographic microscope. Results are given in Table 2. Pyroxene mineral grains of volcanic origin are dominant and the other grain types are also compatable with derivation from volcanic rocks.

TABLE 2 Frequency Counts of Grain Types of Sand Temper in AN 501 Sherd (A and B are two separate thin sections)
Grain Type % (A) 2 % (B) 3 Mean
Plagioclase feldspar mineral grains 6 6 6
Ferromagnesian pyroxene mineral grains 73 69 71
Opaque iron oxide mineral grains 8 10 9
Microcrystalline volcanic rock fragments 11 13 12
Quartzose grains (both mono-crystalline and polycrystalline) 2 2 2
  100 100 100
2   400 grains counted
3   600 grains counted
Temper Origin

The volcanic rock fragments in the sand temper are chiefly microlitic varieties typical of intermediate (andesitic to dacitic) volcanics and felsitic varieties typical of originally glassy (but now devitrified) silicic volcanics. Such volcanic rocks are common among the island arcs (Solomons, Vanuatu, Fiji, Tonga) that fringe the south-west periphery of the Pacific basin, but are rare or absent among the intraoceanic archipelagos (Samoa, Cooks, Societies, Marquesas) within the Pacific basin proper.

Phenocrysts of pyroxene and plagioclase similar to the mineral grains in the sand temper are characteristic of volcanic rocks from both tectonic settings, but the quartzose materials that are present in small amounts occur sparingly within the island arcs only and are not seen at all within the intraoceanic archipelagos. Special care was taken during the petrographic analysis to obtain unequivocal uniaxial interference - 468 figures diagnostic of quartz rather than feldspar. On this basis, there are clear grounds to infer that the temper sand from AN 501 was collected somewhere along the trend of the fringing island arcs, and not within the Pacific basin proper.

General Provenance Conclusions

The AN 501 sherd temper is exotic to the island where it was found and almost certainly exotic to the entire Cook Group where all exposed volcanic rocks are mafic oceanic basalts or basanites and their alkalic differentiates (Wood and Hay 1970; Wood 1978a,b). By contrast, the sherd temper was apparently derived from an island arc along the periphery of the Pacific basin (i.e., andesitic arc temper class of Dickinson and Shutler 1968, 1971). The nearest potential source is Tonga, 1,750 km due west of Ma'uke, while other potential sources lie much farther away.

Pyroxenic variants of Tongan tempers have compositions closely comparable to the temper of AN 501 (Dickinson 1974). Features in common with the pyroxenic tempers of Tonga include (a) predominance of clinopyroxene grains; (b) subequal proportions of plagioclase grains, volcanic rock fragments, and opaque iron oxides; and (c) a couple of per cent of quartz grains (Dickinson and Shutler 1979). No other pyroxenic tempers observed to date in any Pacific Island collections match the AN 501 sherd temper composition as closely as do those from Tonga.

Specific Temper Comparisons

Indigenous Tongan tempers include pyroxenic, feldsparthic, and lithic variants among which relative proportions of ferromagnesian, quartzo-feldsparthic, and lithic sand grains vary, and some also contain admixtures of calcareous reef detritus (Dickinson 1974). The AN 501 temper is microscopically indistinguishable from some of the pyroxenic variants that lack admixtures of reef detritus. There is not a close similarity to Niuatoputapu materials because the volcanic rock fragments in sherds from that locality are dominantly microvesicular pumiceous fragments that are not present in the AN 501 sherd. Rock fragments in all sherds from Tongatapu, Ha'apai and Vava'u generally display the same range of microcrystalline and partly glassy lithic fragments with variable pilotaxitic, hyalopilitic, and felsitic textures that are seen in the array of lithic fragments in AN 501.

Available collections from Tonga are not extensive enough to warrant designation of a specific likely provenance for the AN 501 sherd. However, the overall texture of the temper most closely resembles tempers in selected sherds from Tongatapu, and the grain types in the two cases seem qualitatively identical as well. Although no single Tongatapu temper studied has a composition quantitatively identical to an AN 501 temper, the range of compositions of available Tongatapu tempers constitutes a spectrum within which the AN 501 temper appears to fit.


The Anai'o sherds are not the only sherds recovered in the Cook Islands. In 1987 a larger sherd measuring 73 mm by 56 mm and with a thickness of 11 mm was recovered from the site of Vairakaia on the island of Atiu, 40 km west of Ma'uke (Altonn 1988). This sherd was soft in texture and of a light brown colour. The temper - 469 material was dissimilar to the Anai'o sherd on hand examination and showed no clear similarities with Tongan ware. A date for the Vairakaia sherd has not yet been determined.

It has been pointed out elsewhere that the earliest excavated sites in the Southern Cook Group (sites of the 10th to 14th century) all show evidence of the importation of exotic resources (Walter in press). The sources of some of these goods certainly lie elsewhere in the Southern Group, and the sources of others may lie to the east, in what is now French Polynesia. The discovery of potsherds with an assumed origin in Tonga adds a new dimension to the possible extent of outside contact in the group during this period. This is further reinforced by the discovery of the Atiu sherd, which is possibly Melanesian in origin (Altonn 1988).

The role that Tonga played in Cook Island prehistory bears particular attention in view of the recent work by Katayama. His osteometric analysis of prehistoric Mangaian skeletons revealed that the closest affinities with Mangaia were with material from the Tonga/Samoa region. This led him to propose that there may have been frequent contact between the Cook Islands and Tonga in ancient times (in press). It would now be appropriate to consider the possibility that the cultural gap between eastern and western Polynesia has not always been well maintained and that irregular links between these two areas existed as late as the first centuries of this millennium.

Finally, however, there exists a discrepancy between the date of the AN 501 sherd and the assumed date of the final phases of Tongan pottery manufacture. It has been suggested that pottery manufacture ceased in Tonga in the first few centuries A.D. (Davidson 1979:94; Dye 1987:260). However, the Anai'o sherd postdates this period by 800 years at least. On this basis we believe that the possibility that Tongan pottery manufacture lasted several centuries longer than our present estimates deserves consideration. As this paper was going to press, Kirch reported the discovery of late ceramic ware on Niuatoputapu (personal communication). This material dates to around 800 A.D., which fits more comfortably with the dating of the Anai'o site, although there is still an apparent gap of 400 years or more to be considered.


The field work on Ma'uke was made possible by a grant received from the Wenner-Gren Foundation for Anthropological Research (Grant No. 4814). We are also grateful for the comments received from Prof. Roger Green and Prof. Pat Kirch.

  • Altonn, H., 1988. Broken Pottery Adds Weight to the Legends of Polynesia. Honolulu Star Bulletin, June 29.
  • Davidson, J. M. 1979. Samoa and Tonga, in J. D. Jennings (ed), The Prehistory of Polynesia. Canberra: Australian National University. pp.82-109.
  • Dickinson, W. R. 1974. Sand Tempers in sherds from Niuatoputapu and elsewhere in
- 470
  • Tonga. Journal of the Polynesian Society, 83:342-5.
  • ——and Richard Shutler, Jr, 1968. Insular Sand Tempers of Prehistoric Pots from the Southwest Pacific, in L. Yawata and Y. H. Sinoto (eds), Prehistoric Culture in Oceania. Honolulu: Bishop Museum Press. pp.29-37.
  • ——1971. Temper Sands in Prehistoric Pottery of the Pacific Islands. Archaeology and Physical Anthropology in Oceania, 6:191-203.
  • ——1979. Petrography of Sand Tempers in Pacific Islands Potsherds. Geological Society of America Bulletin 90, Part I (summary), 993-5 and Part II (microfiche), 1644-701.
  • Dye, T. S., MS 1987. Social and Cultural Change in the Prehistory of the Ancestral Polynesian Homeland. Unpublished Ph.D. thesis, Yale University.
  • Katayama, Kazumichi, in press. Biological Affinity between the Southern Cook Islanders and New Zealand Maoris, and its Implications for the Settlement of New Zealand. Submitted for publication in D. G. Sutton (ed.), The Origin of the First New Zealanders.
  • Stuiver, M., G. W. Pearson, and T. Braziunas, 1986. Radiocarbon Age Calibration of Marine Samples back to 9000 cal BP, in M. Stuiver and R. S. Kra (eds), International 14C Conference, 12th Proc: Radiocarbon, 28:980-1021.
  • Walter, R. K., 1987. Preliminary Excavation Report, Anai'o, Ma'uke, Cook Islands, 1987. New Zealand Archaeological Association Newsletter, 30:239-48.
  • ——1989. An Archaeological Fishhook Assemblage from the Southern Cook Islands. Man and Culture in Oceania, 56:67-77.
  • ——in press. The Cook Island-New Zealand Connection, in D. G. Sutton (ed.), The Origin of the First New Zealanders.
  • Wilson, A. D., 1982. Soils of Mauke, Cook Islands. New Zealand Soil Survey Report 52. New Zealand Soil Bureau. Wellington, D.S.I.R.
  • Wood, B. L., and R. F. Hay, 1970. Geology of the Cook Islands. New Zealand Geological Survey Bulletin, 82.
  • Wood, C. P., 1978a Petrology of Aitutaki, Cook Islands New Zealand Journal of Geology and Geophysics, 21:761-5.
  • —— 1978b. Petrology of Atiu and Mangaia, Cook Islands. New Zealand Journal of Geology and Geophysics, 21:767-71.
1   This sherd was recovered from a disturbed area of the site by Melinda Allen of the Department of Anthropology, University of Washington.