Standing on the ancient workings looking towards the feira site
Standing on the ancient workings looking towards the feira site, Maramuca, Zimbabwe

Seventeenth Century Portuguese Earthworks in Rhodesia: Maramuca (Hartley/Chegutu)

Written by P. S. Garlake Historical Monuments Commission, Rhodesia


Two sites, one in the Mtoko (Mutoko) district containing two separate rectangular earthworks and the other in the Hartley (Chegutu) district with a single very similar earthwork have recently been identified and excavated. The excavations were on a very small scale intended primarily to provide, as far as possible, a dateable pottery and bead series rather than reveal details of construction or plan.

Both sites yielded glazed Chinese wares of the seventeenth century and beads of types identical to those found in the later ruin period of the Rhodesian (Zimbabwean) Iron Age, besides locally made pottery. From documentary evidence, it appears that the Mtoko (Mutoko) site can be identified with the important Portuguese trading fair of Luanze while the Hartley (Chegutu) site lies within the area known to the Portuguese as Maramuca.


In February, 1965, attention was drawn to an unsigned report dated 28th May 1947 in the Mining Commissioner’s Office in Gatooma (Kadoma), referring to a “Portuguese Fort marked on an old blueprint” and a low quartz wall retaining a terrace near the Suru Suri river which the writer identified with it. This site was rediscovered in July 1965 and found to be an earthwork almost identical to those of Mtoko (Mutoko).

The site lies on the summit of a small kopje on the west bank of the Suri Suri river, 9 miles (14.5 km’s) above its confluence with the Umfuli river, opposite one of the rare permanent pools. This is the eastern edge of the Lower Umfuli gold belt, where a granite and greenstone schist contact is marked by numerous quartz veins. Such veins are frequently mineralised and bear the gold of the belt. The kopje on which the earthwork is built is a quartz rubble, and a similar kopje 100 yards (91 m’s) to the north is pitted with shallow ancient workings. One and a half miles (2.4 m’s) north of the site, a quartz vein dams the Suri Suri river and on the eastern bank is surrounded by a rough piled wall – further indication of Iron Age exploitation of the quartz. Five and a half miles (8.9 km’s) north – east of the site is the Bay Horse mine, recently the main gold producer of the Lower Umfuli. Here, in the 1930’s, in the fill of an ancient working an ivory statuette of the Immaculate Conception, a Portuguese colonial work of the seventeenth or eighteenth century, was found, (Summers, 1950).

The site (Fig. 1) clearly bears a very close resemblance to the earthworks of Mtoko (Mutoko). Again straight banks enclose a rectangle orientated precisely north, 220 foot (67 m’s) wide and 180 foot (55 m’s) long, with bastions 20 foot (6.1 m’s) wide projecting 15 foot (4.6 m’s) from the centre of each wall. The steep eastern slope of the kopje summit has, however, necessitated further work. Thus, a further wall runs across the centre of the site with an artificial terrace retained by a revetment wall of massive quartz boulders extending east of it, doubtless to provide a level building base, for the slope below to the outer eastern wall is too steep for any building. The most important structure on the site is a rectangular brick building close to the western bastion, the top of whose southern wall stands today some 9 foot (2.7 m’s) above natural ground level. This building was externally 19 foot 8 inches (6 m’s) wide and probably 27 feet (8.2 m’s) long, divided into a series of three long, narrow rooms, each 7 foot (2.1 m’s) wide and 16 foot 4 inches (4.9 m’s) long internally – a plan identical in its basis to that of contemporary East African domestic architecture (Garlake, 1966). The floor was almost certainly raised on an artificial platform some 3 foot (90 cm’s) above ground level. Most of this however is obscured by a very large termite mound that has formed round the entire building. The wall is built of sun – dried brick of a highly calcareous, pale grey, very fine sandy loam, probably derived from antheaps on the Suri Suri alluvium. The bricks are 5¾ inches (15 cm’s) high and 10½ inches (26.5 cm’s) long and the wall 20 inches (51 cm’s) thick, with the daga mortar of the joints 1 – 1½ inches (2.5 – 4 cm’s) thick, giving four courses to 2 foot 3 inches (69 cm’s). The good state of preservation of the southern wall immediately raised some considerable doubt as to its contemporaneity with the earthwork itself but this was resolved in excavation. Beyond three separate scatters of daga there was further trace of occupation within the enclosure, and again the only indication of possible entrances was in the bastions, particularly those on the northern and eastern sides.

Maramuca plan. Maramuca, Zimbabwe

Maramuca plan. Maramuca, Zimbabwe


It was clear that, beside the brick building where excavation would have produced almost insuperable preservation problems, there was virtually no occupation deposit anywhere within or near the earthwork.

A trench on the inner face of the western terrace showed that the terrace fill consisted of the same red quartz gravel and yielded only a single sherd before natural gravel was reached, at a maximum depth of 2 foot (60 cm’s), against the inner face of the revetment wall. A trench excavated outside the western wall confirmed that there was no outer encircling ditch and sterile natural red quartz was reached 6 inches (15 cm’s) below ground level. This trench was extended inwards to the centre of the enclosure wall and again yielded no finds. This excavation did reveal that the wall was not an earth bank as anticipated but that’s its core was built of three courses of brickwork of identical size and fabric to the main building, giving a total height of only 1 foot 7½ inches (50 cm’s). Against this core, gravel was piled, sloping regularly outwards so that total thickness of brick core and gravel at the base of the wall was 4 foot (1.2 m’s) from the centre to outer edge. There is no reason to believe that the enclosure wall is not identical construction to this throughout. It confirms the contemporaneity of the brick main building with the main enclosure wall.

Outside the outer eastern wall the whole steep slope of the kopje is strewn with abundant fragments of stick daga. The scatter appears most probably to be derived from the destruction of huts, probably from the terrace within the enclosure rather than from the collapse of huts in situ. Though the steepness precluded any depth of deposit, excavation here yielded several finds for, prior to the destruction, refuse had clearly been thrown down the slope.

South of the brick building, a sherd of black glazed stoneware was found among a circular spread of daga fragments. A trench was excavated and 6 inches (15 cm’s) below the surface a thin and very broken floor of poor quality daga was revealed, immediately beyond the edge of the daga spread. This rested on 2 inches (5 cm’s) black earth over natural quartz gravel. The fill above the floor consisted almost entirely of broken daga fragments in a black humic earth and, outside the limit of the floor, rested directly on natural gravel.


Glazed Wares.

The only imported ware to occur at Hartley (Chegutu) was the later seventeenth century Chinese black glazed storage jar (Mtoko (Mutoko) Class 6). This occurred at three different areas of the site as follows:

1x large rim sherd with broken lug : surface, excavated hut.

2x sherds (1 with matt exterior glaze) : fill above floor, excavated hut.

6x sherds, 5 stratified : from Southern excavation on slope of hill outside eastern wall.

4x sherds (all with matt exterior glaze), 1 stratified : Northern excavation on slope of hill outside eastern wall.

Local Wares

The sherds from the Hartley (Chegutu) site (Fig. 4) are far too few for meaningful analysis. There was not a single decorated sherd and few rim sherds fall within the range of the Mtoko (Mutoko) assemblage and include poorly beaded and bevelled types. The only two profiles that could be reconstructed are illustrated in Fig. 4; 5, 16. They would also fit the Mtoko (Mutoko) assemblage.

Fig. 4. Local Pottery

5. Bowl with constricted opening. Coarse brown mat exterior. Humic topsoil, Bank.

16. Shouldered pot with outward bevelled rim. Brown mat. South end of the midden.


The three glass beads are identical to those of the Mtoko (Mutoko) site:

  1. 3 x 4mm. opaque Indian red oblate: beneath floor of excavated hut.
  2. 6 x 4mm. opaque white cylinders, fused together: lower level beside floor of excavated hut

From the southern excavation outside the eastern wall came six irregular, formless fragments of glass so decayed that the opaque, grey white, devitrified patina exceeds the thickness of the fresh, bright blue to grey green core, revealed on breakage. These appear to be a glass waste rather than fragments of a vessel.


The architecture of the Mtoko (Mutoko) earthworks clearly shows that they were never intended primarily for defence. They are sited on low, level and easily accessible ground overlooked by hills and make no use of any natural defensive feature in the vicinity, while the very long perimeters would have required an enormous garrison for adequate defence. The low banks, even if a palisade were added, are scarcely immune from storming, while the ditches are, at least in the lesser enclosure, rudimentary and shallow while their defensive function is minimised, if not negated, on the south where a terrace separates them from the bank. In the Hartley (Chegutu) site there are no ditches at all. Further, the two Mtoko (Mutoko) earthworks are not complimentary and, from their siting, could lend each other virtually no support. They clearly fulfilled identical functions but separately and independently. The interiors contained only the simplest structures of pole and daga and probably few of these. The daga was thick, massive and frequently covered with a thin, fine daga skim and wash. There was no trace of post holes or foundations and the only floor was a poor daga on a sand base. The excavations were not sufficiently extensive to recover a plan, but the plane surface of the largest daga fragments seem to indicate straight walls. Nevertheless, the basic architecture of ditch, bank and bastion has a clear military basis and, coupled with the absence of interior structures, would seem to indicate that these two earthworks were intended for secure bulk storage, that they were in fact open, unroofed warehouses. That a considerable amount of trade was conducted from them is clear from the finds of imported beads, wire and glazed wares from the small area excavated. From the very close basic similarity of the architecture at the Hartley (Chegutu) site there can be little doubt that it fulfilled an identical function; though, from its siting within a gold bearing area and so close to ancient workings, one would suspect that the occupants were not only trading but were also directly involved in gold production. Here the scarcity of imported finds suggests both a shorter occupation and a smaller quantity of trade than occurred at the Mtoko (Mutoko) site.

The variety of glazed wares affords excellent dating evidence at Mtoko (Mutoko) and gives a firm date within the seventeenth century for both sites. The Mtoko (Mutoko) site possibly spans the whole century for the Class 2 ware first occurs with Class 1 (late Ming porcelain) in the initial silting of the ditch and both occur most commonly in the early seventeenth century levels of Fort Jesus. There is no reason to suppose that they may not have been traded with the interior at that time. Late seventeenth century wares first occur in the later ditch fill at Mtoko (Mutoko), while black glazed stoneware, the only dateable import at the Hartley (Chegutu) site, though a simple ware popular for a long period, is commonest in the late seventeenth century at Fort Jesus.

All the evidence so far adduced has been purely archaeological. The readily available contemporary documentary evidence enables both sites to be identified and precisely dated, confirming the archaeological evidence. Fr. Joao dos Santos O.P., who was in the interior in 1587, says, in 1609, (Theal, VII p. 270) that at the fairs of Masapa, Luanze and Manzove: “the residents of Sena and Tete have houses called churros, where they store their merchandise and from which they sell it and send it to be sold throughout the country”. At this time Luanze already had a Dominican church, mentioned subsequently by several seventeenth century authors. The location of Luanze is described by Diogo de Couto, writing at almost the same time as dos Santos

(Theal VI p. 368). “There are three markets where the Portuguese go to trade for gold and sell their merchandise . . . the first is called Luanhé and it is about thirty five leagues from Tete to the south. It is between two rivers which when they unite are called Nausovo, and the said place is ten leagues from each of these rivers.” The old Spanish league equalled 2.63 miles (4.2 km’s), which would make Luanhé 92 miles (148 km’s) from Tete and 26 miles (41.8 km’s) from the rivers. The Mtoko (Mutoko) site lies 93 miles (149.7 km’s) from Tete in a direct line. The rivers referred to must be the Nyadire and Ruenya which both join the Mazoe. The site is 24 miles (38.6 km’s) from the Ruenya and 19 miles (30.6 km’s) from the Nyadire, distances slightly less than de Coutos’, but this must occur as the rivers are never more than 45 miles (72.4 km’s) apart. On this evidence, there can be very little doubt that the Mtoko (Mutoko) site is the site of Luanze. Abraham (1962) has made brief mention of a site at Mt. Chitomba, 13 miles (8.1 km’s) north east of the Mtoko (Mutoko) site, which he identified with Luanze. A more recent visit by the present author and Mr. Abraham yielded no evidence of occupation or structures and showed that this is almost certainly not the case.

Pedro Barretto de Rezende described Luanze in more detail in 1634 (Theal II p. 416-417). “The fort of Luanze before mentioned, where the Portuguese hold a market is in the lands of Mocaranga, forty leagues from Tete. It is only a palisade of stakes, filled up inside with earth, allowing those within to fight under cover. The stakes are of such a nature that when they have been two or three months in the ground they take root and become trees which last many years.

“The site of this fort inside is like a large terrace, being a hundred fathoms (brassas) in circumference, where the captain resides, who is elected by the captain of Mocambique, and with him the Portuguese and Christians who may be trading in these parts.” The circumference of the large Mtoko (Mutoko) earthwork is 1184 feet (361 m’s), that of the smaller, 720 feet (219 m’s). The Portuguese brassa equals 7.21 feet.(2.2 m’s). The circumference of the smaller earthwork thus corresponds absolutely exactly with de Coutos’ measurements. The description seems to indicate that the interior of the “fort” had few structures and that the earth (actually sun dried brick) bank was of subsidiary importance to a palisade – which may explain the small size of the bank – excavation however, showed no postholes nor was there any indication of the bank being supported by any such palisade.

Luanze, Masapa (near Mt. Darwin) and later Dambarare (in the Mazoe (Mazowe) area) formed the most important trading fairs of the Portuguese for most of the seventeenth century, until the 1680’s when the reigning Changamire gradually caused their abandonment. By 1695 he had driven the Portuguese completely from the present lands of Rhodesia (Zimbabwe).

Abraham (1961a) has demonstrated convincingly, using Portuguese records and oral tradition together, that the area known to the Portuguese as Maramuca lies between the Umfuli and Umsweswe rivers, east of the Chakari river. The Hartley (Chegutu) site lies in the northern corner of this area. Manoel Barretto describes Maramuca in 1667 thus: “Maramuca is the name of a great district of kingdom of upper Mocaranga . . . This kingdom is the richest in gold known”. He goes on to give an account (Abraham 1961a) of the trading concession held there by one Goncalo Joao in the early 1660’s, whose headquarters were attacked and pillaged very shortly after by two traders whose monopoly he had superseded. Maramuca was thereafter closed to the Portuguese. According to this account Joao was unprepared and had no chuambo (i.e. fortification or defence). Abraham’s informants told him of traditions that the Portuguese lived close to Chivare Hill. This rests solely on recently recorded oral traditions and no further evidence has been found to support it. Hitherto, no early Portuguese remains, except for the Bay Horse ivory, have ever been found in this area.

The Hartley (Chegutu) site seems to provide archaeological confirmation of Abraham’s identification of the area of Maramuca and it may well be the site of Joao’s headquarters, as this appears to have been the only Portuguese settlement in Maramuca, while the date is supported by the imported stoneware. The question of the defences appears somewhat inconsistent – yet there is no doubt that the Hartley (Chegutu) site would have been easily stormed particularly if it lacked a palisade. There is certainly some indication that the huts within had been purposely destroyed.

Previous to the discovery of the Hartley (Chegutu) and Mtoko (Mutoko) sites, the only indisputable Portuguese sites known in Rhodesia (Zimbabwe) were six earthworks close together on, or near, the Angwa river 30 miles (48 km’s) north east of Sinoia (Chinhoyi). They resemble the sites described – straight rectangular earthwork banks from 160 feet (48.7 m’s) to 200 feet (61 m’s) long, the largest surrounded by a further bank and ditch from 320 feet (97.5 m’s) to 350 feet (106.7 m’s) in length. All had a single central building raised on artificial mounds, comparable to the Hartley (Chegutu) building. In place of the four projecting bastions in the centre of each face at Hartley (Chegutu) and Mtoko (Mutoko), the Angwa river earthworks seem usually to have had two bastions projecting from diagonally opposite corners of the earthwork. Thus they have the same basic plan as those of Mtoko (Mutoko) and Hartley (Chegutu) but with this important distinction. Architecturally, it thus seems safe to place them in a separate group from those of Hartley (Chegutu) and Mtoko (Mutoko), probably of a somewhat different date. Much of the archaeological evidence has been destroyed over the past seventy years and they have never been properly excavated. A single sherd of Chinese porcelain with a blue and white interior and a chocolate glazed exterior, found on the surface during a visit by the present author, is of late seventeenth century to eighteenth century date and was absent at Hartley (Chegutu) and Mtoko (Mutoko). This agrees with the tentative identification of the Angwa river site with the minor Portuguese fair of Ongo (Axelson, 1960) which enjoyed a brief period of prosperity in the early 1690’s before being abandoned before Changamire, to be one of the few sites to be reoccupied in the early eighteenth century.

Turning to the significance of the present sites in the local Iron Age context, we have in the beads a series that can be said with certainty to represent Portuguese trade goods, Beads are described by de Couto in 1634, referring specifically to the fairs in Monomatapa’s kingdom, as being, with cloth, the principle item of trade. “They also take for this trade some small beads made of potters clay, some green and others blue and yellow, with which neckets are made. . .” This description exactly tallies with the beads recovered at Mtoko (Mutoko). The first mentioned are the Indian reds, and in fact the order in which de Couto lists them virtually reflects the proportions in which they are found. The Mtoko (Mutoko) bead series is identical in colour, proportions, quality, size, opacity and preservation to Robinson’s Group II at Khami (Robinson, 1959) recovered from all levels of the excavations of the Hill Ruin, though the earlier Group I (into which the single transparent white bead and a few very small Indian red beads from Mtoko (Mutoko) would fit) is “better represented in the lower levels”.

Robinson has suggested that the Group II beads of Khami are Portuguese imports (Robinson, 1959). Though only 10 beads of this series, belonging to the corresponding Zimbabwe Period IV deposits, were recovered in 1958 Zimbabwe excavations, they are again considered as Portuguese imports. (Robinson, 1961). This origin is now proved by the beads from Mtoko (Mutoko).

Imported porcelain of the Wai Li period has been recovered from sealed deposits resting on floors adjoining two huts at the Hill Ruin, Khami (Robinson, 1959, 1961a). This is the Class I ware of Mtoko (Mutoko). A soft earthenware with blue and white glaze, recovered with this porcelain from one of these Khami huts (Robinson, 1961a), is a Glass 1 ware of Mtoko (Mutoko) (only two other imported wares were recovered from Khami – a German stoneware, possibly sixteenth century and a green glazed jar rim, probably seventeenth century Egyptian). Robinson has suggested that the imported wares may represent the loot of Changamire’s people in the campaigns of the 1690’s and in fact has given this considerable weight in dating the Khami Hill ruin to the very beginning of the eighteenth century. Using historical evidence Abraham (1961b) has questioned the date. The present archaeological evidence shows that these wares and the glass beads were being imported in quantity into the interior as trade goods at least by the mid seventeenth century. There now seems little reason to suppose that they were not able to reach Khami by that date.

Comparison of the local wares with pottery from the other Iron Age sites in Rhodesia (Zimbabwe) is difficult for, except in Inyanga (Nyanga) district, there has been no excavation of any sites of comparable date in Mashonaland.

Robinson (1965) has described pottery from two ruins in the Zambezi valley, Ruswingo wa Kasekete and Mutota’s Ruin. The former has been considered (Whitty, 1959) to be in part seventeenth century Portuguese. Characteristic features of the pottery include lavish graphite burnishing, developed rims and polychrome decoration, all of which compare more closely with Period IV Zimbabwe pottery and Khami pottery than with that from the Mtoko (Mutoko) site. The complete absence at Mtoko (Mutoko) of polychrome band and panel ware, characteristic of Khami and other similar ruins, supports the association by Robinson of this ware with the Rozwi, for seventeenth century Luanze was outside the Rozwi area. It is interesting that a sherd of an incised polychrome bowl with graphited interior was found on the surface of an Angwa River earthwork, which may again indicate that these date from the early eighteenth century after the Rozwi Changamire had occupied this area.

However, the basic pottery forms of the Mtoko (Mutoko) site do echo those of the Khami Ruin wares – particularly the tall necked pots. There the resemblance ends. The heavy rolled rims, fine finishes and lavish decorations characteristic of the Khami Ruin wares and, to a lesser extent, of the related wares from Zimbabwe and the Zambezi valley are completely absent. The Mtoko (Mutoko) pottery is all simpler and coarser. The closest comparable pottery is probably the Inyanga (Nyanga) Lowlands ware which by Summers’ dating (1958) is contemporary with the Mtoko (Mutoko) site. However, the concave necks, shouldered pots, graphite burnishing and the rarity of other burnishing – all features of the Lowland wares – make this comparison far from absolute.


I would like to thank Mr. J. S. Kirkman of Fort Jesus, Mombasa for his help in identifying the glazed wares. Mr. D. P. Abraham was of great assistance in visiting his Mt. Chitomba site and in discussing the written evidence for Luanze’s location. The Mining Commissioners of the Rhodesian (Zimbabwean) Ministry of Mines and Lands first drew the attention of the Historical Monuments Commission to both areas and this Ministry also prepared a geological report on, and assayed the samples from, the Mtoko (Mutoko) site. Mr. F. P. du Toit of the Department of Conservation and Extension investigated samples of the bricks from the Hartley (Chegutu) site.

Note: Corrections to place names and from miles to km’s I have done. (C Dunbar)


– Abraham, D. P. 1961a. Maramuca : An exercise in the combined use of Portuguese records and oral tradition. Journal of African History, 11, 2.

– Abraham, D. P. 1961b. Porcelain from the Hill Ruin, Khami. S. Afr. Arch. Bull, XVI, 63.

– Abraham, D. P. 1962. The early political history of the kingdom of the Mwene Mutapa. In : Historians in Tropical Africa. Proceedings of the Leverhulme Intercollegiate History Conference, 1960. U.C.R.N. Salisbury.

– Axelson, E. 1960. Portuguese in S.E. Africa, 1600 – 1700. Johannesburg: Witwatersrand University Press.

– Axelson, E. 1956. Some loopholed forts in the Mt. Darwin Dist. Proceedings Rhodesia Scientific Assoc., XLIV.

– Garlake, P. S. 1966. The early Islamic architecture of the East African coast. London: Oxford University Press.

– Kirkman, J. S. 1954. The arab city of Gedi. Oxford: University Press.

– Robinson, K. R. 1959. Khami Ruins. Cambridge: Cambridge University Press.

– Robinson, K. R. 1961a. Dated imports from the Khami Ruins. S. Afr. Arch. Bull., XVI, 62.

– Robinson, K. R. 1961b. Zimbabwe Pottery, Zimbabwe Beads. In Summers, R., ed. Zimbabwe excavations, 1958. Occ. Pps. Nat. Mus. S.R., III, 23A.

– Robinson, K. R. 1965. A note on Iron Age sites in the Zambezi valley. Arnoldia, I, 27.

– Summers, R. 1950, Uma reliquia Portuguesa na Rhodesia do Sul. Mocambique, 63.

– Summers, R. 1958, Inyanga. Cambridge: Cambridge University Press.

– Theal, G. 1898-1901. Records of South Eastern Africa. II, VI, VII. Cape Town: Govt. of Cape Colony.

– Whitty, A. 1959. A classification of prehistoric stone buildings in Mashonaland, S.R. S. Afr. Arch. Bull., XVI, 62.

Stay informed about Colonial Voyage
eXTReMe Tracker