Tombs, Landscape and Society in Southern Madagascar - Part V

Mike Parker Pearson

Madagascar 2000 Fieldwork in Androy

The ancient `manda´ civilisation

The first part of this season's work was devoted to the study of manda and other large settlements of the 10th-13th centuries. Ever since the discovery of Andranosoa and Mandan d'Remananga in the 1970s, these large and formerly densely occupied sites have been central to archaeological investigations in the south of Madagascar. They bear witness to a vanished civilisation which once inhabited the river systems of the south and imported Islamic sgraffiato and Chinese ceramics but has left no clear traces in oral or written history. The dating of certain of these large sites has been uncertain until now. Both Mandan d'Remananga (near Bekily) and Andaro (near Bekitro) were initially dated by their imports to the 14th-15th centuries.

Today, with more radiocarbon dates for local pottery styles and more precise dating for the imported ceramics, it is possible to redate these stone-walled enclosures to the 10th-13th centuries. We also recorded and planned another large 10th-13th century manda, enclosed by continuous earthen banks, on a tributary of the Linta near Fotadrevo. This site, known as Andasy Meriñe or Lintamandameriñe, was first visited by archaeologists over 15 years ago but no records or plans had survived. We now have large 10th-13th century stone-walled manda (earth-walled in the case of Andasy Meriñe) in the upper reaches of each of the four main river systems of the south: the Linta, the Menarandra, the Manambovo and the Mandrare.

This season we were able to complete our study of 10th-13th century manda and other large sites in the Manambovo river system. Previously we had found small stone-walled manda at Faritsoke, Bevotry and Amanda and, together with the known manda at Andaro and Andranosoa, their distribution hinted at their regular spacing of about 14 kilometres apart. The large extent of these sites compared to ordinary sites of the period and the presence of decorated bowls, footed vessels and imported ceramics suggests that these were central places, and we decided to put this idea to the test by predicting where other sites might be located and then seeing what we might find in those places. The first result came in January 1998 when Retsihisatse located a very large 10th-13th century site (but without walls) at Maienkandro on the south bank of the Lañany river, immediately to the east of its junction with the Manambovo river where we had predicted such a site in 1996. The second prediction of 1996 was fulfilled when we located a similar site at Ankotsobe on the west flank of Vohimena. Two other predicted sites were located at Manolodroe on the north flank of Vohipary and at Bebajiñe. There were four predictive failures where we looked for such sites but failed to find any of that date.

We can revise the chronology of manda and their associated ceramics to recognise a 10th-11th century phase characterised by Andaro basins (flat-bottomed pots with curved walls and wavy lugs) and an 11th-13th century phase characterised by Beropitika basins (flat-bottomed pots with straight walls and large simple lugs). The latter are found at Andranosoa, whose radiocarbon dates fall within this later period. We think that Andranosoa may have been occupied later than Andaro, Andasy Meriñe and Mandan d'Remananga. These large urban settlements of the 10th-11th century Andaro phase may have been the earliest substantial human occupation in the south, though our discoveries at the moth of the Menarandra (see below) may push evidence of human occupation back to the 7th-9th centuries AD.

From some forty finds of imported ceramics and glass at Andaro, Andasy Meriñe, Mandan d'Remananga and Maienkandro we have been able to show that the use of imported vessels in southern Madagascar was much greater than hitherto suspected. In conjunction with finds of imports at the Manambovo and Linta rivermouths, these finds suggest that the trade routes for these artefacts were not overland from northern Madagascar, as previously thought, but via sea routes to the south, presumably from the adjacent African mainland of Zimbabwe and Mozambique. This suggests that the manda civilisation of southern Madagascar had direct links with the Late Iron Age civilisation of the Swahili coast.

Excavations of a royal village visited by the French in 1649

The second stage of our programme involved the excavation of site 331 at Montefeno. This settlement site had been found in 1995 when we were struck by its impressive hillside location, by its large size relative to other adjacent sites, and by its copious quantities of 16th-17th century graphite-burnished pottery. It is the most likely candidate for the royal village of Montefeno visited by Flacourt's soldiers in 1649. Our hopes of recovering stratified sequences of cultural material were rewarded, since ploughing has only recently been employed over this 165x120 metre wide settlement site. Our aim of dating the site precisely within the 16th-17th centuries has been partially achieved. Amongst the many small finds we found a bead which appears to have been cut from a European clay pipe stem. A layer post-dating the latest graphite-burnished pottery contains a gunflint which must be dated no earlier than the end of the 17th century. Analysis of the pottery indicates that the blocked and dense decorations common on suspected 16th century sites like Trañovato and Anjampanorora are rare and confined to a few worn sherds. We conclude that Montefeno was occupied principally in the 17th century.

The house platform that we excavated in the centre of the site is much larger than any houses so far excavated in the south. It equates with the dimensions given by Robert Drury for royal houses in Androy at the beginning of the 18th century. Its associated finds of a green glass bead, a metal bracelet and a foundation deposit of cattle bones and a wild boar's tusk, together with the absence of worked bone tools from this part of the site, support this interpretation of a high-status building. We conclude that it was a royal house, possibly the residence of Andriamififarivo ('Prince of a thousand hedges') or one of his sons.

On the trail of Robert Drury: a massacre survivor of 1703

Before commencing the third stage we revisited the wreck site at Belitsaky which is probably that of the Degrave, the ship that brought Robert Drury to Madagascar. We had hoped that local lobster divers might again dive on the area where they had seen some four cannon and an anchor, to take underwater photographs of these and other artefacts. Unfortunately the sea was too rough. A useful outcome was the recognition of a trunnion marking on one of the cannon lying on the reef.

The third stage was intended to be a surface survey of a block of land to the north of the Montefeno-Anjampanorora area between Belindo and Benonoke where there are sacred forests in which many of the Tandroy roandria (kings/lords) are buried. This area was selected also because, on the basis of his directions, it is the likely location of the royal village where Drury was kept as a slave between 1703 and 1709. The area was particularly interesting because there was a complete absence of settlements characterized by 17th century graphite-burnished pottery. Instead, the earliest settlement scatters were characterised by coarse Tandroy styles with flat-topped rims and with a few comb-decorated vessels. One of these sites, at Mionjona, was formerly occupied by the royal clan and we believe it to be Drury's village. Corroboration is provided by Drury's recording that his village lay five miles to the west of Mahandrovato, which is now known as Mahatomotsy and lies six kilometres to the east (we have shown previously that Drury consistently overestimated his distances!). Other 18th century sites at Mahatomotsy (Androvabe), Añalafanjane and Ankazomanintsy may also coincide with villages mentioned by Drury. In further corroboration with Drury's account of the re-establishment of his village, there are two 18th century settlements at Mionjona with the second one lying half a kilometre to the southwest at Ankamena.

Given the possibility that we had found Drury's village, we adjusted our programme in order to carry out limited excavations. The site of Mionjona was the one selected on the basis of its 17th century comb-decorated sherds and its location close to a sacred forest within which roandria of the Andriamañare clan are buried. It had been the residence of the grandfather of the oldest man of the Andriamañare clan living nearby at Belindo-Anjankitike, and 19th century sherds were recovered from parts of the surface scatter as well as from stratified layers within one trench. The other five trenches produced ceramics with thin walls, a leathery finish and flat rims which are consonant with our interpretation of 17th century pottery. Interestingly there was only one sherd of 'red-slipped' from the excavated assemblage. Quantities of ceramics and animal bone were much smaller than those at Montefeno and we wonder whether the site was occupied briefly in the 18th century, abandoned and then reoccupied in the 19th century. Our excavations on the northwest edge of the site located two phases of defences. One survives only as three or four slight earthworks which we interpret as the adjacent lines of former cactus hedges, and thus 19th century in date. The other, lying some 10m inside this line is a series of postholes which may have formed two or possibly three rows of palisades. We suspect that the settlement was initially cut into dense primary forest, which has subsequently shrunk to the dimensions of today's sacred forest. About 10m inside the palisade lines we excavated a small house platform which was so well preserved that we could identify the placing of the doorway in the northwest corner and the hearth midway along the east side. Deposits stratified on top and underneath this floor contained flat-rimmed pottery and finds included a lump of lead, a piece of silver and iron artefacts.

Earliest settlement, extinct Elephant Birds and Swahili contacts

The fourth stage was a continuation of the coastal survey, extending westwards into Karimbola and Mahafaly. At Anavoha, on the Karimbola coast west of Cap Ste Marie, we visited the sub-fossil sites excavated in the 1940's. One of these had been disturbed recently by the digging of a well and amongst the bones of hippopotamus, archaeolemur, giant tortoise and crocodile we found two ribs which appear to have received cutmarks with iron tools. Apart from a series of 18th and 19th century settlement scatters, the only definite indication of early settlement in the area was a 14th-15th century midden settlement in the dunes associated with Aepyornis eggshells and bones of ordinary tortoise.

Ten kilometres to the west, at the mouth of the Menarandra, we found evidence of early occupation on both banks of the river. The pottery from these sites is closely comparable to Swahili Triangular Incised Ware (TIW) and Plain Ware (PW) of the 7th-10th and 11-13th centuries respectively. The TIW pottery from Enijo on the west bank makes this potentially the earliest site so far found in southern Madagascar. There were a few sherds of 10th-13th century local wares nearby on the west bank but otherwise there was no mixing of Swahili and Malagasy ceramic styles in the same contexts or sites. There were no other traces of ancient occupation until the 18th and 19th centuries when a very large village was occupied underneath and adjacent to the present Mahafaly village of Matsandry south.

The mouth of the Linta is a huge expanse of moving sand and recent landuse changes include the loss of the river due to its damming upstream. Its former channel lies to the northwest of an earlier channel. We were hoping that the placename of Androka might relate to the name of the rokh or Elephant Bird but local understandings point to it deriving from the name for a tree branch. None-the-less, we were taken to a group of middens at Ambohibola, close to the earlier palaeochannel where Aepyornis eggshell lay in association with 10th-13th century ceramics. Several of the pots had mending holes, including a piece of sgraffiato. A few sherds attested to activity in the 14th-15th, 16th and 18th centuries in this location.