Written by P. S. Garlake. Historical Monuments Commission, Salisbury (today Harare). Received : 27th May 1969
From their establishment at Sofala in 1505, the Portuguese directed their activities in south-east Africa at the exploitation of the gold resources of the interior. During the late sixteenth and seventeenth century’s agreements with the ruling Mwene Mutapas led to the opening of trading “fairs” in Rhodesia and subsequently to the formation of permanent Portuguese settlements, depending primarily on trade but concerned also with the direct mining of gold, missionary activity and administration. Of such settlements Massapa, immediately south of the Zambezi escarpment, Luanze, in the extreme north-east of Mashonaland and Dambarare, in central Mashonaland, were the most important.
The aim of the present excavations was not to test the rich documentary evidence that exists regarding Portuguese settlements in Rhodesia. The excavations were on far too small a scale to give anything more than one glimpse of a single aspect of the settlement. Rather, by excavating one of the earliest sites in Rhodesia which can be related in detail to documentary evidence, it was hoped:
firstly, to obtain a precisely dated and detailed assemblage of the indestructible goods traded by the Portuguese at this time in the interior; and secondly, to obtain material information on the indigenous Iron Age inhabitants of seventeenth-century Mashonaland and to compare it with the better known Iron Age sites of this period in Matabeleland, outside the sphere of Portuguese influence.
Without discussing or interpreting in any detail the abundant contemporary archival material bearing on Dambarare, sufficient extracts from the better known documents are quoted here to give some idea of Dambarare’s geographical position and history.
In 1631 Dambarare was listed as a fair with a church served by a Dominican priest, one of the earliest mentions of the site by name (Theal, 1898 II, p.438). In 1635, Diogo de Sousa de Maneses, a former Captain of Mozambique, speaks of the alluvial gold in “the river Manzovo….which runs from Dambarare, and reaches the river Ruenya” (Abraham, pers. comm.). In Do Estado da India, written in 1634, Pedro Barreto de Rezende states “Farther in the lands of Mocaranga (than the fort of Luanze) is the fort of Ambarare where there is a market. In this part they dig for gold, and there are gold mines” (Theal, 1898, II, p.417).
By 1667 Manoel Barreto could write “Dambarari is a noble settlement and good sized town in the heart of Mokaranga and has grown to be the centre of the conquest, with many rich inhabitants….”, one of the four “principal places where gold is obtained” (Theal, 1898, III, p.482). Very soon thereafter, Portuguese mismanagement caused a decline in trade. By the late 1670’s disease and disorder had left Dambarare almost entirely uninhabited (Axelson, 1960).
In 1684, when Caetano de Melo de Castro was Governor of Mozambique, Dambarare was chosen to become the centre for the reassertion of Portuguese influence, with the establishment of a garrison and the erection of a new fort.
Writing fourteen years later, de Almeida y Sousa described it thus: “The site of the fort constructed in the time of Caetano de Melo De Castro is on a terrace from which the entire surrounding plain was visible and where there was a most plentiful river, with excellent water flowing the whole year, called the Marauoza, a stone’s throw from the fort…” (Abraham, pers. comm.). Antonio de Conceição, writing at the same time, says: “At Dambarare we used to have a fair with its earthworks…but within there was nothing more than the Church with its Vicar and the Captain; the rest lived in the vicinity at some distance from one another…” (Abraham, pers. comm.).
After 1684 gold production started again. There was, however, a rising threat of war from the southern territories outside Portuguese influence. In November 1693 Dambarare was attacked and destroyed by Changamire Dombo. Antonio da Conceição describes how the inhabitants were killed before they could reach the safety of the fort, the church was desecrated, some burials disinterred and burnt and the priests flayed alive. Changamire went on to destroy all Portuguese settlements in Rhodesia. Thereafter the site of Dambarare was visited occasionally by traders and in 1769 appears to have been reoccupied (Schebesta, 1966). It was, however, never to regain more than a vestige of its former importance.
From documentary evidence, therefore, the site of Dambarare had long been known to lie near the headwaters of the Mazoe River. The archaeologist J.T. Bent visited the upper Mazoe Valley in 1891 after completing investigations at Zimbabwe. Throughout his work in Rhodesia, Bent was aware of archaeological evidence for the early Portuguese presence. After quoting one source on Portuguese trading posts in the lower Mazoe Valley, he writes “Further evidences of this Portuguese enterprise in Rhodesia will doubtless come to light as the Mazoe Valley is more fully explored. In the vicinity of a new mine called the Jumbo, fragments of old Delft pottery have been found, a few of which were shown to me when at Fort Salisbury. Nankin china is also reported from the same district, an indubitable proof of Portuguese presence…” (Bent, 1892). There is little doubt that the wares he describes were Portuguese, Persian or Chinese ceramics collected from the present site, though Bent neither visited the site nor related it to Dambarare. At the time of Bent’s visit, the upper Mazoe Valley was an important mining centre where prospectors were well aware of the evidence of earlier European occupation.
In the vicinity of the present site, the ruins of several prospectors’ dwellings remain. Here, mixed with the household rubbish of sixty to eighty years ago, a few fragments of much earlier Chinese porcelain can frequently be found.
During 1944 and 1945, Mrs. E. Goodall collected imported ceramics from the surface of the present site and its immediate vicinity. In 1961, Mr. D. Abraham, working from documentary and traditional sources, identified an earthwork immediately north of the present site as being an early Portuguese structure (Abraham, 1961). In December 1966, Mr. H. Light took the present author to a large mound which, in 1923, he had recognised to be the remnants of a building and near which he had unearthed a human burial.
Since 1923, however, the site had been ridged, ploughed and cultivated over many years, softening the form of the mound and completely confusing any other structures that may have existed. Many very large termite mounds in the immediate vicinity made recognition still more tentative; nevertheless, it appeared that this mound could only be of human origin, an identification made more certain by Mr. Light’s recollections. It was here that the present excavations were undertaken.
Dambarare (fig. 1) lies high on the central watershed of Rhodesia (lat. 34o 54’15” east, long. 17o 29’00” south), 26 miles north-north-west of Salisbury and immediately east of the Marodzi (or Murowodzi) River, a tributary of the Mazoe River, which it joins about 25 miles downstream. Both rivers rise close together about 10 miles south of the site. Though small at this point, both flow for almost the entire year.
The excavated site (fig. 2 and 2A, Earthwork 1) consists of a mound, 180ft long, 90ft wide and 7ft high as its centre, running almost due east-west. Vestiges of an enclosing bank, visible only as a darkening of the soil in ploughed land, we recognised north of the mound. After excavation had commenced, an infra-red aerial photograph was obtained which clearly highlights the mound and enclosure (fig. 3). (missing) The enclosure, 270ft wide from east to west, is bounded by four straight earth banks, 465ft long on the east side and 540ft long on the west side. The mound exactly divides the enclosure. Its west end is irregular in shape, possibly indicating subsidiary buildings. A second, smaller mound extends into the southern half of the enclosure from the west bank.
The site lies in a basin of light granite soils, commanding, extensive views over the Marodzi River to the west. It is drained by a permanently marshy vlei which discharges into the Marodzi. An extensive low hill mass overlooks, but by no means commands, the site from the north while the east side of the basin is bounded by a continuous range of hills.
The earthwork is the largest of at least four closely similar structures in the vicinity. A small, isolated granite kopje immediately north of the vlei bears traces of earth banks forming a rectangle which surrounds the summit, somewhat destroyed by a much later small brick house (fig. 2, Earthwork 2). A mile north of the site, the southernmost spur of the hills bear similar straight earth banks in part reverted in stone and enclosing a large mound – probably the site identified by Abraham in 1961 (fig. 2, Earthwork 3). Half a mile due west of this again, on level ground immediately above the steep banks of the Marodzi, lies a further well preserved rectangular earthwork, with outworks beyond its north-west corner. Traces of pole and daga huts are visible in the shallow, stony soil of the interior (fig. 2, Earthwork 4). All these sites yield numerous sherds of seventeenth century Chinese ceramics, and all are closely similar in plan and structure. All, moreover, are visible one from another. A mile upstream of the last site the Marodzi flows in a large bend round the foot of a commanding spur on its western bank. Here, further sherds of Chinese porcelain were recovered among the vestiges of daga floors but no earthworks were recognised (fig. 2, Earthwork 5).
The geology of the immediate area is complex. The site lies on a granite intrusion into quartzites, ironstones and schist’s, components of the Basement Complex of Rhodesia. The soils of the area vary correspondingly; as will be seen, many have contributed to the fabric of the excavated buildings. The granite contact is gold bearing.
The Marodzi contains small quantities of alluvial gold in the vicinity of the site, while “ancient workings”, of unknown date, have been described immediately below Earthwork 4 on the southern spur of the hill. These contained shafts 60ft deep which exploited a gold reef averaging 74.5 dwt. (Ferguson and Wilson, 1937). In an ironstone outcrop half a mile east of the excavated site there is another “ancient working” with very regularly cut galleries with arched roofs, quite uncharacteristic of “ancient” mining techniques. These may well have been excavated under Portuguese direction. They do not appear, however, to have struck a gold reef; iron ore may have been extracted.
Excavation of the central mound of the largest earthwork was conducted for two weeks in June, 1967 and throughout September and October, 1967 (fig. 3). (missing) A single line of 6ft. by 6ft. trenches, separated by 2ft. baulks, was laid out at right angles to the long axis of the mound, extending from the centre of the mound northwards (fig. 3A). These trenches were designated AA, A, B, C, D, E, F, G, reading from the centre of the mound to its northern perimeter. Trenches D and F were not excavated. Three trenches, each 6ft. by 3ft., designated X, Y and Z, were laid through the north wall of the enclosure, at rights angles to it (fig. 3B). The whole site had been ploughed immediately before the excavations started. This had not penetrated more than 6in. below the surface. All loose ploughed topsoil was removed prior to excavation. The ground level shown in the sections (fig. 5) can therefore be taken to be 6in. below unploughed ground level throughout.
Trench E was extended 2ft. to the west (i.e. a final area 8ft. by 6ft.) in order fully to excavate the grave shaft of Burial E1, Trench AW, 6ft. by 3ft. was an extension at right angles on the west side of Trench A at its northern end, to recover the complete skeleton of Burial AW1. It was not excavated below this burial. The very numerous burials in the lower levels clearly indicated that the site contained a cemetery. Piecemeal extensions of the trenches or removal of the baulks to recover complete burials would have complicated any future complete gridded excavation of the site. No attempt was made to extend the excavations to obtain plans of the buildings. Full excavation of the site is clearly desirable, but is beyond the present resources of the Historical Monuments Commission.
Each stratum was excavated separately wherever possible. This was clearly difficult in cases where intersecting graves of different periods contained virtually indistinguishable fills. All significant finds and all loose bones were therefore planned and, if necessary, photographed before removal. They could thus, with very few exceptions, be related to the strata from which they came. Excavation within each stratum was normally in 4in. horizontal spits. Each course of brickwork was planned before removal, while all skeletons were drawn and photographed before removal. Only Burial C1 was removed en bloc.
All deposits below the topsoil (including floors and most brickwork) were soft, slightly moist and extremely easy to dig. On exposure and drying, however, they rapidly hardened. As excavation in the lower burial levels, which contained skeletal material, fragile metalwork and the vestiges of shrouds, necessitated close examination, beads and small finds could be recovered as they appeared and sieving was not considered necessary, except in strata 8A and 13 which were sieved. Check sieving of all deposits confirmed that no beads were being missed. The methods used to recover and preserve the more fragile material has been described elsewhere (M. Garlake, 1969).
Fig. 5 illustrates the east and west sections of each trench (save for the east sections of Trenches E, G and Y) and the southern section of Trench AA. The strata, numbered in reverse order to their formation, are as follows:
1. Hard grey humic topsoil.
1A. Variegated pit fill.
2. Brown silty loam, with small, isolated fragments of buff and grey bricks, interrupted by lines of coarse sand at 2-4in. intervals.
3. Abundant, large fragments of buff and grey brick rubble in brown silty loam. Lines of coarse sand at the base of the deposit or in some cases at intervals throughout the deposit.
3A. Discoloured fine black occupation deposit (Trench G only).
4. Pure brown sandy loam with lines of coarse sand at base in some areas. Rubble fragments generally sparse and very small (sealed at the top by clay Floor A, in Trenches B and C and on the east side of Trench A).
5. Wall of buff, grey, yellow and chocolate sun-dried bricks set in a clay mortar.
5A. Wall, fabric as 5.
5B. Wall, fabric as 5.
5C. Fallen walling: the same wall as 5A.
6. Highly variegated and completely mixed grey-brown earth with stone, rubble and grey clay inclusions: the fill of grave shafts cut from immediately below Floors b or c, and sealed by them.
7. Red earth and rubble, with abundant, large fragments of broken red clay bricks: sealed by Floors b or c.
7A. As 7, but sandier, more variegated and with grey brick inclusions.
8. Wall of red clay bricks set in a clay mortar.
8A. Grey clay with stone inclusions: fill of grave shaft cut from same position as 9. but containing a much less variegated deposit.
9. As 6. Fill of grave shafts cut from immediately below Floors d or e and sealed by them.
10. Dark grey clay with discontinuous fine red quartzite gravel lenses.
11. Pure pale grey clay with quartzite gravel lines and lenses.
11A. Dark grey clay, similar to 10 but with quartzite gravel occurring more abundantly and uniformly throughout: sealed by a continuous fine red quartzite gravel and portions of Floor f.
12. As 6. Fill of grave shafts sealed by Floor f or the clays of 11.
13. Impure, fine, black earth with abundant potsherds, bone, charcoal and other finds.
13A. As 13, but paler with charcoal lenses and a high proportion of brown silt (Trench E only).
13B. Grave fill of variegated quartzite rubble, clay and brown silt (Trench E only).
14. Pure brown silty loam.
15. Pure, continuous, unweathered quartzite rubble.
16. Bedrock of decomposed red quartzite.
The undisturbed natural soil of the site is a brown silty loam, some 9-12in. thick (14), resting on a continuous rubble bed of unweathered red quartzite gravel, 7-9in. thick (15), which overlies a decomposed bedrock of red quartzite (16). The original ground level was intensively occupied before any buildings were erected. This resulted in a black discolouration of the natural soil and the accumulation of a litter of household rubbish, with no trace of buildings or building debris (13). Sherds of Chinese, Persian and European ceramics securely date this occupation to the seventeenth century and leave no doubt that the first occupiers were either Portuguese or in very close contact with the Portuguese. Stratum 13 was originally level, continuous and without variation. It occurs at the base of every trench. In two small areas, this original ground surface was sealed by a thin floor of grey clay (in Trenches A and C, not visible in section). The top of the original ground level is now 7ft. below the top of the centre of the mound while beyond the walls of buildings it is 1ft 6in. below present ground level (Trench G). All deposits above the original ground level have a human origin.
Over the original ground level (13), there are up to four or five layers of pure pale grey clay or darker grey clay with gravel (10, 11, 11A). These layers, never more than 9in. thick are separated by thin spreads of fine gravel averaging 1in. thick and not all continuous. These were occasionally capped by portions of a thin clay floor (Floor f). The precise nature and purpose of the clays and gravel spreads is unknown; they can best be explained as impervious floor fills. Virtually no finds were recovered from these deposits. The total thickness of these clays and gravels is 2ft. 6in. They are best preserved in Trenches AA and A, though vestiges remain in Trench B.
From an early stage the area was being used for burials. Four graves (12) in Trench AA (containing the burials AA 3, 4, 6, 7 and 8) are sealed by successive layers of clay and gravels. After these deposits have been laid, nine further graves (9) were dug (containing the ten burials AA 1, 2, 5; A 5, 6, 7; B 2, 3, 4, 5).
Their shafts are filled with a variegated deposit, incorporating the materials through which they were dug. Associated with the clay fills (10, 11, 11A) are walls of red clay sun-dried bricks set in a grey mortar (8 in Trenches B and C). Only portions of the foundations survive. The foundation trenches have irregular, stepped sides and were cut through the grey clay fills (11) to the top of the original ground level, in part slightly penetrating it. A thick, well made, black clay floor, e, is almost certainly the interior floor a building extending south of the red brick wall across Trenches AA and A. It seals the grey clays and burials (9). It was renewed up to nine times with thin skim coats of clay and rises to form a step at the south-west corner of Trench AA.
Away from the buildings, a discoloured silty brown earth, interspersed with charcoal lenses (13A in Trench E) marks a continued external occupation similar to that preceding it (i.e. 13). It seals a grave (13B, Burial E1).
A large grave was inserted between two red brick walls (8A in Trench C and the north end of Trench B, containing the six burials C1, 4, 5, 6, 7, 8). The grave shaft, unlike its predecessors, is filled by uniform pale grey clay. It is sealed by the very thin clay floor, d, laid after the partial destruction of the northern red brick wall, for it continues for a short way across the broken stub of the wall.
The entire red brick building was levelled to the ground. There is no indication that its destruction was violet or accidental. This resulted in the formation of a distinctive red band of broken bricks and rubble some 12-15in. thick (7). Away from the buildings, the rubble lenses out to a thickness of 4in. (Trenches E and G).
Eight burials (6) took place soon after the destruction of the first buildings (A 1, 2, 3, 4; B 1; C 2, 3; one in the west section of Trench AA, was probably destroyed by later digging). Their grave shafts contain a fill very similar to that of the earlier graves.
Renewed building activity was now initiated. This resulted in the darkened discolouration of the top of the red rubble layer at the north end of the mound (top of 7 in Trenches B and C). At the centre of the site, Floor c was laid sealing the red rubble layer and the graves it contains (top of 7 in Trenches AA and A). A second major building was now erected, represented by a wall running from south to north – down the centre of Trenches AA and A – to meet at right angles a similar wall running east-west – across the southern end of Trench B (5A). These walls are distinguished by a fabric of an entirely new character (5A, B, C). The sun-dried bricks now used were larger and thinner, predominantly sandy in texture and of various colours: buff, grey, yellow and chocolate. Their foundations cut through Floor c and the red rubble which it seals (see south section of AA). Nevertheless, Floor c was scarcely damaged and probably continued in use. In most areas the fabric of the wall entirely filled it’s very neatly cut foundation trench but in one area a rubble packing was necessary (7A in Trench B). Soon after, a portion of new walling (5) was partially demolished and Floor b (Trenches B and C) laid. It seals the red rubble outside the building for the first time, and abuts, neatly and directly, the northern face of the main wall of the major building.
Floors b and c became exposed to the weather and the surfaces outside the buildings were washed by storm water. As a result, a series of horizontal verves of coarse sand accumulated undisturbed on the external floors (base of 4 in Trenches B and C). This sand was covered by clean brown silt, identical to the natural soil of the site, though it incorporates fragments of brown brick from the building (4). A single badly decayed skeleton, A W2, composed of burial, lay within this deposit. Every floor was completely lacking in finds. Away from the buildings, however, a thin occupation deposit, very similar to the original, did accumulate over the red rubble layer (3A in Trench G).
Further alterations and additions to the main building were now undertaken. Inside, the natural silt accumulation was increased by a thick fill of the same material to give a total thickness of 18in. and Floor a laid over it (visible in the east sections of Trenches AA and A).
Outside, the silts (4) were dug away and Floor b largely destroyed, when an addition (5B) was built against the north wall of the main building. Floor a (top of 4 in Trenches B and C) was then laid some 9in. above Floor b, abutting both the latest extensions (5B) and the northern face of the original main building (5A).
The site now appears to have been finally abandoned. There are no signs of violent destruction, although the scattered and badly decayed bones and cranium of a single body, A W1, were found in stratum 3. The building disintegrated through natural processes. Clean surfaces, exposed to rain, encouraged erosion and wash evidenced by the varves of sand that accumulated over the floors and banked against the walls (base of 3). Very abundant large fragments of brickwork mark the most rapid stage of the collapse of the building (top of 3). A large section of the main wall fractured and collapsed as a block to the east (5C) lying in part directly on the last floor, a, but also on the sands accumulated over it. Thereafter disintegration proceeded more slowly. A brown silty loam, containing isolated small rubble fragments, is interrupted by successive thin horizontal lines of wash, 2-4in. apart (2). This deposit, up to 2ft. thick, is derived from the gradual complete elimination and burial of the ruined buildings over a long period of time, during which the site was completely unoccupied.
A final and comparatively recent event in the history of the site was the excavation of a trench, 6ft. deep and averaging 2ft. wide, running east-west across Trench AA (1A). It was filled in immediately after it was dug and was doubtless the work of a late nineteenth century or later treasure hunter attracted by the obvious age or alien origin of the mound. Fortunately, little damage was done, for the main burial levels were not reached. It is possible, as the east section of Trench AA shows, that one burial was destroyed.
North Wall Of The Enclosure
The stratigraphy of Trenches X, Y and Z, excavated to locate the northern enclosure wall – and outer ditch, if any – was simple (fig. 5). A soft, pure brown silty loam (14), the natural soil of the site, extended unchanged for a depth of 3ft., from the present ground level to the red quartzite rubble bed (15). The top 9in. was slightly greyer, stonier and harder than the remainder (1). Cut into this was a trench 2ft. 8in. wide and extending 2ft. 3in. below the present ground level (5). A course of bricks, of identical fabric to those of the last major building of the mound, was laid in a grey clay mortar across the entire base of the trench. These were covered by a 6in. layer of mottled red and grey clay. A further course of bricks were laid and covered in turn by a further layer of clay, reaching to within 3in. of the present surface. The alternate brick and clay courses represent the foundations of the outer wall. There was no indication of an occupation horizon associated with this wall. It must have been very close to the present ground level and destroyed by recent ploughing and ridging.
The excavations extended 11ft. outside the enclosure wall and 8ft. within it. They revealed no trace of an outer ditch or of any further structures. The greater depth of natural soil in this area is probably a result of topography: the mound stands at the crest of a slight ridge, while the northern wall is at a lower level, near the marshy land which drains the site and towards which soil and wash is carried. Four undecorated body sherds of indigenous pottery from the brown loam were the only finds from Trenches X, Y and Z.
The bricks of the first buildings were dark red clay with quartzite and grey clay inclusions. They varied in size from 10-11in. wide, 18in. long and 5-61/4in. thick, averaging 101/2x18x6in. They were set in a grey clay mortar, with joints 1-11/4in. thick, three courses to 221/4in. As only small portions of these walls survive, no certain information could be obtained on bonding patterns.
The bricks used in the second building period differed considerably from those of the first and red bricks no longer occurred. The fabrics included loose, coarse buff silt; slightly more compact and finer grey silt; yellow clay; and chocolate clay with quartz inclusions. The buff and grey fabrics differ little in consistency of texture from the natural soil of the site. These two fabrics occasionally occurred in the same brick – otherwise the material used in individual bricks was completely pure. Buff bricks formed about 50 per cent of the bricks used, grey bricks 35 per cent, yellow 10 per cent and chocolate 5 per cent. The brick size was based on a double square, averaging 103/4x22x41/4in., though individual bricks vary from 121/2x20in. to 10x23in. in size and from 4-51/2in. in thickness. They were set in a chocolate brown clay mortar, with joints 1-11/4in. thick, three courses to 17in.
The varying colours and textures of the bricks reflect the varying geology and soils near to the site. Soils of colours matching the various bricks can be seen within a quarter of a mile of the building: the red clay is derived from the quartzite’s and ironstones immediately east of the site; the buff slit from the granite soils of the site itself; the grey silt from the granite soils and vlei alluvium immediately north of the site; and the yellow and chocolate clays presumably from the Basement Complex deposits in the immediate vicinity. All bricks were sun-dried – though small areas of some bricks appear to have been slightly burnt in manufacture, forming patches of hard crust, and altering the colour of the fabric. This never penetrated more than ½ in. below the surface. Many bricks were badly cracked and had lost large portions from their corners before being used, while the very irregular size of the bricks made systematic lying impossible.
The main wall was 2ft. 11in., i.e. one and half brick lengths, thick. The courses were laid in double Flemish bond: alternate headers and stretchers were laid on each face, the header of one face meeting a stretcher on the opposite face. The gap that resulted in the centre of the wall was filled with a brick rubble packing and large quantities of mortar rather than with bats: indeed, the use of broken brick packing not only within the walls but between courses was common. Large quantities of mortar were also used to compensate not only for the irregular sizes and breakages of the bricks but also for the bad positioning of many bricks which were frequently laid at an angle to the general line of the wall. Successive courses follow the same pattern of bonding but with large and completely random variations. There was no systematic method of breaking vertical joints, although these rarely occur over more than two courses.
The quality of the fabric and construction of the walls is poor. This is compensated by the size of the bricks and the thickness of the walls, which would certainly have been capable of standing for a considerable length and height unsupported by cross walls and of carrying a roof of large span.
The foundations of the north-south wall of the main building consisted of three courses of brickwork, 3ft. 101/2in., i.e. two brick lengths thick, the bond following that of the superstructure. The foundations of the east-west wall were not thickened and appear to have been only two courses deep. This wall, however, rests on the earlier red brick wall whereas the north-south wall runs across clays and grave fills.
The finished surface of the walls consisted of a clay mortar ½ in. thick, covered in a fine, pure white wash. The latter had poor adhesion and has entirely disappeared from the wall surfaces (except under the fallen wall, i.e. Stratum 5C), being washed off and down the walls to accumulate as a thin lens on the floors at the foot of each wall.
Floors were all of thin, pure black clay, some ½ in. to 1in. thick, laid directly over natural silt, clay, gravel or brick rubble; many run across and seal grave shafts, in some instances the floors were renewed, up to nine times, with variegated washes of black, brown, red, grey and white clays, each no more than 1/8in. thick. The only case where a floor was hardened by burning occurs outside the north wall of the last major building (Floor b, Trenches B and C). There is no sign of kerbs or skirtings. One step was constructed of the same black clay as the floors lay over a fill of brown silt (south section, Trench AA).
There is no trace of stone or woodwork in the buildings – though little of the superstructures survive. The roof structure must presumably have been of timber and several iron nails were found in occupation deposits outside the building and contemporary with its erection.
From the scale of building and the continuous association of burials with the structures, it is safe to assume that both major buildings were churches, though no significant plan was obtained. It is almost certain, from several indications, not least the shape of the mound itself, that the interiors lay south of the major east-west walls (i.e. Floors e, c and finally a in Trenches AA and A). Two walls of the red brick building are 7ft. 3in. apart and the area between them is almost entirely filled by a single grave deposit (8A). These walls run east-south-east to west-north-west at an angle of 10o to the line of the later church. The east-west wall of the latter is built on the foundations of the walls of the first church which seems to indicate that the reconstruction of the church followed the earlier plan.
The plans and purposes of the first minor structure (5) of the second building period and of the later northern extension (5B) of the church are at present unintelligible, for only their foundation survive. These structures had identical fabrics and the junction between the two is not discernible: the brickwork 5B continuing directly that of 5.
Solid brickwork extends at least three courses deep over all of Trenches B and C, except for the areas of Floors a and b, 18in. wide along the west side of Trenches B and C and 18in. wide along the south side of Trench C. This brickwork forms a plinth or packing below floor level. The extensions directly abut the outer face of the church but are not keyed into it.
The enclosure wall was built of sun-dried bricks, averaging 101/2in.x20in.x41/2in. in size, somewhat smaller than those of the last major building but of the same buff and grey colours and laid in the same bond. It was 2ft. 7in., i.e. one and a half brick lengths, thick. The brick courses alternate with 6in. layers of clay mortar. Only the four courses of the foundations remain. These probably formed the base of a timber palisade and core of a small earth bank. The wall cannot have been of any height for there is no sign of any building debris in the proximity.
Thirty-one burials can be accounted for, though not all the bones of every skeleton were recovered. Interments took place throughout the history of the site and can be divided into four stratigraphic and temporal groups: firstly, those that took place before the erection or early in the occupation of the first church (five burials: stratum 12); secondly, those interred immediately beneath Floors d and e of the first church (seventeen burials: strata 9, 8A and 13B); thirdly, those interred after the demolition of the first church and before the erection of the second, which they underlay, sealed by Floor c (seven burials: stratum 6); and fourthly, two skeletons lying in and on the occupation level associated with the last church (Trench AW, strata 3 and 4).
The undisturbed skeletons lie extended full length with their arms crossed on the chest, their hands clasped at the waist or, in two cases, clasped against the head. They clearly follow Christian and not indigenous African practice. They were all orientated almost exactly east-west. Twenty-one of the bodies had their heads to the west, four otherwise normal burials had their heads to the east (the bones of the remaining skeletons were scattered). Burials were extremely crowded; there appears to have been some linear organisation – seventeen undisturbed or partially disturbed adult burials (from all strata) had either heads or toes within 6in. of the west face of the line of the trenches and extended east. Only six adult burials departed from this. This arrangement probably reflects the layout of the buildings of the site, to which the burials were related.
The grave shafts are all almost exactly 3ft. 3in. deep (maximum: 3ft. 6in., minimum: 2ft. 9in.). They were irregular in shape with sloping sides, all but one being some 1ft. 10in. wide at the centre and narrower at the ends, very little larger than the body they contained: this was well illustrated by Burial A7 where the head was bent forward as it rested against the wall of the shaft. In no case was a coffin used. The bodies of the Negroid females at least were wrapped in cotton shrouds, traces of which were preserved wherever they were in contact with copper.
All bodies were buried in individual graves, with the possible exceptions of Burials AA 6 and 7 and the bodies in the large grave between the walls of the first church (stratum 8A) where a Caucasoid male (C7), and a Negro female (C4) may have been buried together. No lasting record of any burial was preserved. Burials were not marked at the surface by any form of headstone – most were moreover sealed by floors.
No less than twelve graves were unintentionally disturbed or destroyed by later interments. This caused little concern. In five instances the finding of an earlier burial caused a halt in the digging of the grave shaft and the later body was laid immediately on top of the earlier burial – though not before some bones or even the upper half of the body had been disturbed.
In one instance, where Burial A4 struck the two earlier Burials A5 and A6, all bones above the pelvis of one body and the skull and left humerus, clavicle and scapular of the second body were displaced. They were carefully collected and reburied in two separate compact groups, probably wrapped in cloth or otherwise packaged, in their approximate original positions but immediately over the later burial. This care did not, however, prevent the bones of the two bodies becoming mixed – the mandible and one clavicle of one body being reburied with the group containing the skull, mandible and clavicles of the other. Five burials were completed destroyed by three later burials. In each case the bones were scattered haphazardly over the later body. Most of the larger bones were returned to the lower portion of the grave shaft as the grave was filled in – though vertebrae, ribs, finger and toe bones occurred throughout the grave fill. In one case disturbance took place very soon after original interment, before the body had decayed: the left tibia, fibula, foot and toe bones and the left radius, ulna and hand bones of Burial C4 remained articulated, though scattered with the rest of this skeleton over the body of Burial C5.
Furthermore, the copper ornaments of Burial C4 preserved extensive remnants of a shroud. This could not have occurred if the shroud had not been in good condition at the time of the disturbance.
Seven Caucasoid males (AA 1, 2; A 1, 2, 3, 6, 7) and two hybrid males (i.e. of mixed racial origin) (AA3; A5) are the only identified bodies among the fifteen burials within the church. Of the fourteen burials outside the church, two Negro females (B3 and 4), a hybrid female (B2) and an infant (B5) were buried first (stratum 9). A large grave (stratum 8A) inserted in this area originally contained the burial of a Caucasoid male (C6). Subsequently a Caucasoid male (C7) and a Negro female (C4) were interred, disturbed by the burial of a further Caucasoid male (C5). Finally a young child (C1) and an unidentified adult (C8) were buried in this grave. After the first church was demolished, a Negro female (B1), a child (C3) and a Caucasoid male (C2) were buried in this area (stratum 6). The only Negro male (E1) was buried well away from the buildings (stratum 13B).
Of the Caucasoid males buried within the church, at least five wore gold rings and bronze and silver medallions, aiguillettes, pins or buckles. In contrast, of the four Caucasoid males buried outside the church, only one (C5) had any grave goods – and that, a simple decorated tablet of clay.
In the arrangement of the bodies for burial, eight of the burials within the church had the arms crossed on the chest in the usual Christian manner, and only one had the hands clasped at the waist; whereas outside the church five burials of both races had the hands clasped at the waist and only one burial, an infant, had the arms crossed on the chest.
A social and possibly religious distinction, which continued unchanged throughout the history of the site, seems discernible in the locality chosen for burials. Men, predominantly Caucasoid but including two hybrids, were buried within the church, generally composed for burial in the normal Christian manner and possessing personal adornments of some quality. Outside the church, women, children and Caucasoid males whose grave goods reflect a lower status than that of the men buried within the church, and a Negro male were buried. The fact that all but one of the bodies buried outside the church had their arms uncrossed may indicate that the distinction was one of the formal religious status, e.g. they were possibly catechumens rather than communicants.
Six Hundred and fifty-five sherds of indigenous pottery were recovered. Practically all are very fragmentary and the information they yield is disappointing.
Most vessels tend to have a coarse body, with a coarse white grit tempering. In colour, 50 per cent of the sherds are brown or buff, 18 per cent grey or dark grey, 14 per cent jet-black and 10 per cent a well-fired red; the remaining 8 per cent are graphited. Only 15 per cent of the sherds are burnished; this total is made up of brown and buff, grey, jet-black and red sherds in almost equal quantities. Proportionately, therefore, far fewer of the poorly fired brown and buff sherds are burnished than the well-fired jet-black or red wares. The buff and brown sherds also tend to be noticeably thicker and coarser than the better fired material. Most of them doubtless come from utilitarian cooking and storage vessels. Of the rest, the jet-black burnished sherds (3 per cent) and the graphite burnished sherds (8 per cent) are noteworthy minor features of the assemblage.
Twenty-nine undecorated rim sherds were recovered. In all vessels, the rim receives virtually no emphasis. Of the eleven rims of bowls, nine are simple and tapered (fig. 6.1) and two have flattened, slightly inward bevelled lips (fig. 6.2, 3). There is one example of a bowl with a constricted opening. Most pots had slightly beaded, averted rims (eight examples, fig. 6.4, 5). In the larger pots this produces an almost flat lip (four examples, fig. 6.6, 7). In only two cases is a rim slightly thickened and rounded on the exterior (fig. 6.8, 9). In three cases, the rim is simple and tapered with no moulding at all (fig. 6.10). Only two sherds are of sufficient size to indicate the pot form – both have tall necks; one, roughly made, has a concave neck and sharply angled shoulder (fig. 6.7, 10).
The six decorated sherds (less than 1 per cent of the pottery recovered) are all comb-stamped. One simple, tapered pot rim bears diagonal comb-stamping to the rim itself (fig. 6.11); another, stamping immediately below the rim (fig. 6.12). One body sherd, too small and weathered to be useful, resembles these sherds and appears to have also contained diagonal comb-stamping. A fourth sherd has a wide band of rather thin, shallow, diagonal comb-stamping immediately below a very slightly rounded rim (fig. 6.13) while a body sherd exhibits similar comb-stamping (fig. 6.14). All these sherds are unburnished. One graphite burnished sherd, however, has a panel of small diagonal comb-stamped impressions above a thin painted red band (fig. 6.15).
Seven decorated sherds of indigenous pottery, found on the surface in the vicinity of the site include three with stamped decoration (fig. 6.16, 17). Two sherds bear an irregular stab decoration between grooved lines (fig. 6.18) and one a band of incised cross-hatching (fig. 6.19). One sherd, from a bowl with a constricted opening, bears a discontinuous raised horizontal band, applied below the rim, bearing incised cross-hatched decoration (fig. 6.20). This is a characteristic vessel of the Barwe Tonga of the lower Zambezi Valley. Similar pottery has been collected from the aringa of Massangano on the confluence of the Zambezi and Mazoe rivers, a site that has also yielded seventeenth century Chinese porcelain and may originally have been a seventeenth century trading post (Newitt and Garlake, 1967). This sherd thus suggests contact with people a considerable distance away from Dambarare on a known trade route. The significance of the surface finds is, however, questionable.
Forty-three per cent of the excavated sherds come from the original ground surface, and a further 35 per cent from the fill of graves penetrating to this surface and incorporating material from it. The remaining 22 per cent of the sherds are scattered sparsely but uniformly throughout the deposits: in wall fabrics (4 per cent), wall debris (14 per cent) or topsoil (3 per cent). Not a single shred rested on or can be associated with any floor.
The pottery from Luanze, a contemporary Portuguese trading post in north-eastern Mashonaland, is very closely comparable to the present assemblage (Garlake, 1967). Of the three decorated sherds recovered from excavations there, one bore comb-stamped decoration identical to that of fig. 6.16, while the other two had irregular stab decoration. Great simplicity of vessel forms, simple or slightly beaded rims, poor firing and infrequent use of burnishing or graphite burnishing are characteristic of both assemblages. Both are further distinguished by the extreme rarity of any form of decoration. Where decoration occurs, however, comb-stamping is the dominant motif. At neither site, however, is it yet possible to define the indigenous pottery adequately. Considerable trade was conducted from both sites. It can be expected therefore, that the indigenous pottery may come from more than one source and, indeed, one sherd described above appears to illustrate this clearly. There was, however, no indication that the stratified sherds did not form a homogeneous assemblage.
The assemblages bear no close resemblance to other Iron Age wares in which comb-stamping was also a characteristic motif. The assemblage has no affinities with the wares of the later Iron Age ruins in Matabeleland or southern Mashonaland or to wares from the Mutota and Kasekete ruins in the Zambezi Valley (Robinson, 1965), though the latter have frequently been described as being under direct Portuguese influence in the seventeenth century (Abraham, 1959). None of the many diagnostic features of the Ruin wares – fine finishes, heavy rims and, later, lavish polychrome decoration – occur in the present assemblage, which belongs to an altogether simpler, less sophisticated tradition.
The general appearance of the indigenous pottery from Dambarare and Luanze, in burnishing, fabric, rim forms and rarity of decoration, compares more closely with Shona pottery of the recent past. More specifically, the Korekore, the Shona tribe now inhabiting the north of Mashonaland, once the central province of the Mwene Mutapa confederacy, decorate their pottery with comb-stamping today. Their characteristic motifs are, however, narrow bands of diagonal comb-stamping between incised lines in horizontal bands or triangular patterns round the area of the shoulder of the pots. The stamping is considerably coarser and more irregular than the decoration of the Dambarare assemblage; but may well be related to or devolved from it.
Imported ceramics were collected from the immediate vicinity of the present site during 1944-45 by Mrs. E. Goodall. This is the most comprehensive assemblage of imported sherds of the Portuguese period in Rhodesia. A representative series of these sherds was examined for the present author by Mr. J. S. Kirkman and correlated with his finds from Portuguese levels at Fort Jesus, Mombasa.
From this series, the complete assemblage, 3182 sherds, is now classified. (Test excavations and surface collecting at Luanze (classified in Garlake, 1967) yielded a smaller assemblage of imported ceramics, all of which, save a single sherd (a very thin black burnished earthenware: Luanze Class 9), appear in the Goodall collection.)
1. Porcelain, blue and white glaze. Chinese, Wan Li period of the Ming dynasty, late sixteenth to early seventeenth century (Luanze, Class 1). Flat dishes, decorated in panels, are a feature of both the Dambarare and Luanze assemblages. 44 per cent.
2. Porcelain: blue and white glaze. Chinese, Transitional period, mid-seventeenth century. 6 per cent.
3. Porcelain: interior blue and white glaze, exterior chocolate glaze. Chinese, K’ang Hsi period, late seventeenth to early eighteenth century. 3.5 per cent.
4. Stoneware, thick grey body: exterior black, brown, mustard or olive glaze; interior black or olive glaze, brown wash or unglazed. Chinese storage jars, sixteenth to seventeenth centuries. 21.5 per cent.
5. Stoneware, very thick grey body; unglazed or poor, mottled, black to olive partial glaze. Chinese storage jars, late seventeenth century. 1.5 per cent.
6. Stoneware; exterior green to yellow glazed, with trailed floral decoration; interior black glaze, brown wash or unglazed. Chinese storage jars, seventeenth century (Luanze, Class 2). 1.5 per cent.
7. Earthenware, white, grey or pink body; white glaze with designs in bright blue or blue outlined in black design. Persian imitations of Chinese porcelain, late seventeenth century. 4.5 per cent.
8. Earthenware, thick buff body; exterior poor partial white glaze; interior white glaze with designs in turquoise; probably Persian, late seventeenth century. 0.5 per cent.
9. Earthenware, thick, pale buff body; exterior transparent purple, turquoise or green glaze; interior clear glaze or unglazed. Persian Jars, seventeenth or eighteenth centuries. 1 per cent.
10. Earthenware, white body; bright blue and white glaze. Portuguese imitation of Chinese porcelain, middle to late seventeenth century (Luanze, Class 8). 8.5 per cent.
11. Earthenware, white body; purple, blue and white glaze. Portuguese, middle to late seventeenth century. 1 per cent.
12. Earthenware, white body; plain white glaze. Portuguese, middle to late seventeenth century. 3 per cent.
13. Terracotta, thick body; poor yellow-green to olive glaze. Probably European, seventeenth century. (Luanze, Classes 4 and 5). 1 per cent.
14. Terracotta; unglazed or poor clear to brown glaze to interior only. Probably European, seventeenth century (Luanze, Class3). 1.5 per cent.
15. Miscellaneous: Chinese “egg and spinach” porcelain; terracotta’s with exterior grey-blue and black mottled glazes; orange glazed and blanc de Chine porcelains; Chinese olive glazed earthenware with white trailed decoration. All late seventeenth century. 1 per cent.
Of the imported ceramics, 79 per cent originate from China, 6 per cent from Persia and 15 per cent from Europe. 44 per cent of the ceramics were manufactured in the late sixteenth or early seventeenth century, 18.5 per cent in the middle of the seventeenth century, 11 per cent in the late seventeenth century and 26.5 per cent are of types manufactured at least throughout the seventeenth century.
In the present excavations, thirty-two sherds of imported ceramics were recovered from sealed positions in stratified deposits. These comprised twenty-three sherds of blue and white Wan Li porcelain from separate vessels (as 1 above); four sherds of black stoneware (as 4 above); two sherds of Persian blue and white earthenware (as 7 above) and three sherds of European terracotta (as 13 above). A fragment of flat, mottled brown glass, 1.2mm. thick, was also recovered.
Fifteen of these imported sherds (twelve Chinese porcelain, two European and one Persian earthenware) rested in or on the original ground surface of the site (stratum 13). A further eleven sherds (seven Chinese porcelain, three Chinese stoneware and one Persian earthenware) and the fragment of glass were recovered from the fills of grave shafts that had penetrated the original ground surface and so incorporated material from it. Of the remaining six sherds, a fragment of Chinese porcelain came from an early clay floor fill (stratum 10) and a fragment of European earthenware from the fallen debris of the first red brick building (stratum 7). Two sherds of Chinese porcelain were incorporated in the wall fabric of the last major building (stratum 5A) and a further two lay in the debris of this building (stratum 2) and the humic topsoil (stratum 1). It is possible that all the excavated sherds are derived from the original occupation horizon. The numerous fragments of porcelain broken and lost during this period indicate that these imports were of no special rarity or value.
The stratified imported ceramics form a small but homogeneous and diagnostic assemblage typical of a seventeenth century Portuguese site, closely comparable to the assemblages from Luanze (Garlake, 1967) and the earlier seventeenth century Portuguese levels at Fort Jesus, Mombasa (J. S. Kirkman, pers. comm.). They clearly show that the initial occupation of the site occurred in the seventeenth century. The stratified material is, however, too sparse to date the phases of the building sequence within this century. However, Kirkman has divided the ceramics from the excavations of the Portuguese deposits at Fort Jesus, Mombasa, into two periods; 1593 to c. 1632 to c. 1632 to 1698 (Kirkman, 1967).
The former contains late Ming and Transitional Chinese porcelains and blue and white Portuguese earthenware’s – all found at Dambarare; the latter contains K’ang Hsi porcelains, an increased amount of Persian imitation porcelain and blanc de Chine, famille verte, café au lait and over glaze blue Chinese porcelains – only the first two of which have been found at Dambarare (with a single blanc de Chine sherd). The Dambarare imports therefore appear to indicate that ceramics were imported after 1632 but not as late as 1698. Late seventeenth-century imports are extremely sparse in comparison with Fort Jesus. To be more precise, it would be essential to compare the two assemblages directly. This is not at present possible.
Glass and Shell Beads
The glass and shell beads recovered from the excavations are classified in Table 1. They have been divided into three groups.
Group 1 Beads are small discs from 3-6mm. in diameter with a constant bore diameter of 1.7mm. When strung, thirty-six beads cover 3cm.
Group 2 Beads are reheated glass canes: opaque oblates and cylinders from 2-4mm. in diameter and 1.5-4mm. in length. The colours tend to be dull and the shapes irregular. No bead in this group shows a patina. These beads have been established as the normal seventeenth century Portuguese trade beads. They are identical to the excavated bead assemblages from Luanze (Garlake, 1967) and compare closely with the beads of Zimbabwe Period IV (Robinson, 1961) or the later Khami series (Robinson, 1959). They also form a minor component of later Leopard’s Kopje bead assemblages (Garlake, 1968b).
Group 3 Beads are reheated glass canes: small oblates, 2.5-3mm. in diameter and 1.5mm. long, transparent or translucent in colour. They are smaller and much more regular in shape than the beads of Group 2. The colours are bright and pure, light blue and a plum so dark as to appear black completely domination the assemblage. Decay has rendered the beads porous and resulted in a thick white patina completely obscuring the original colour. These beads are broadly comparable to the small transparent oblates obtained from pre-Portuguese East African traders, which dominate Iron Age bead assemblages of the twelfth to fifteenth centuries (e.g. the later Leopard’s Kopje assemblage of Mapela: Garlake, 1968b; Zimbabwe, Period III assemblages: Robinson, 1961; and many of the Mapungubwe beads: van Riet Lowe, 1955). They are, however, by no means identical, being slightly larger, of a different quality glass and of colours not found in the earlier assemblages.
Group 4 Is a miscellany. There are beads of glass, moulded to form hexagonal prisms 8-10mm. across and 10mm. long (fig. 7.1), tabular heart-shaped beads, moulded in translucent green glass, 7mm. long, 9mm. wide and 2mm. thick, bored vertically through the centre (fig. 7.2); and tiny oblates, of opaque Indian red or transparent plum glass, identical in colour and fabric to the beads of these types in Group 3 but so small, 1mm. in diameter and 0.5mm. long, as to be barely visible to the naked eye. There are also coral beads, pale pink in colour, frequently with a blotchy white patina; they are either spherical in shape, 5-7mm. in diameter, or barrel-shaped cylinders, 2.5mm. in diameter and 4-5mm. in length. Beads of ivory, both spheres, 4mm. in diameter and segmented cylinders, 5mm. in diameter and 8mm. long formed a “rosary” hanging from a bead girdle worn by Burial B1 (fig. 7.3).
It is interesting to review the positions in which beads were found. The bead finds from stratum 13, the original ground surface of the site, represent the normal losses or discards of the occupants of the site before it assumed any specialised function or importance. Fifty-three typical beads (Group 2) were found in it, and only one transparent oblate bead (Group 3) and five shell beads (Group 1), i.e. trade beads form 90 per cent of the total. In contrast, trade beads form only 29 per cent of the glass beads and less than 2 per cent of the total beads from graves. In the four burials with beads, there were no trade beads on Burial E1, only twenty-three amongst over 11,500 beads on Burial B1, and thirty-seven amongst over 4,000 beads associated with Burial C4 though they dominated the assemblage of Burial C3.
It is clear that while the trade beads (Group 2) were those handled and lost in everyday life during the initial occupation, they provided only a minor component of the beads worn by those later buried at the site. Vice versa, the beads worn by the burials were seldom traded. It is thus not surprising that while Group 2 beads are common and characteristic finds in later Rhodesian Iron Age sites, many of the beads in Groups 3 and 4 have not before been found in Central Africa.
Glass and shell beads were worn by two Negro females (B1 and C4), by the Negro male (E1) and by one infant (C3). Intact strings and portions of strings of beads show how the beads were arranged. Shell beads formed the major component of the strings, with rare spacers of glass, coral or metal. The pelvic girdle of Burial B1 consisted of twelve strings of shell beads, broken by four small groups of glass beads and four single glass beads as spacers. Lengths of shell beads at least 12cm long were unbroken by spacers. This girdle is described in detail in Appendix II. In the beads found loose in the grave fill, stratum 8A of Trench C, but which almost certainly belonged to Burial C4, one broken string had forty-one shell beads, followed by two spherical coral spacers and then two further shell beads; copper beads were used as spacers in the same way.
It is certainly not necessary to suppose that the very restrained use of coloured beads was entirely or even predominantly due to their rarity or cost. The aesthetic effect of long lengths of uniform white shell discs broken by the single vivid points of colour provided by the rare groups of glass, coral or metal spacers must have been and indeed still is, very striking. This arrangement emphasises the colour and sparkle of the glass spacer beads in a way impossible to achieve in a variegated string of coloured beads.
No female burial was found wearing a necklace or glass or shell bead bracelets. The Negro male, Burial E1, wore a single strand necklace of copper beads with four large moulded glass beads placed together at the front. Burial C3, an infant, appears to have worn a necklace of transparent light blue and minute Indian red and plum beads. This infant also wore a single compact group, 1in. square, of Indian red trade beads, sewn to a cap or threaded in the hair.
The Dambarare beads form the most varied excavated bead assemblage yet found in Central Africa. An attempt can be made to relate the assemblage to the written inventories of Portuguese trade goods. This is hampered by the lack of published inventories for the seventeenth century, though complete and detailed inventories of the trade goods held at Sofala between 1505 and 1516 are published (da Silvo Rego and Baxter, 1962-65, IV). From these, some idea of the relative value of beads can also be obtained. The unit of currency used was the Sofala mitical of gold, approximately one-sixth of an ounce, equivalent therefore to approximately 50s. sterling. The unit of volume was the faracola, equivalent to approximately 33lb.
Due to the unpopularity of Portuguese imported beads, caracoes (“snail stone beads”), became the main item of local barter at Sofala between 1516 and 1518. Being locally manufactured, they were exempt from the Royal monopoly over the bead trade and in volume, over seven times more of these beads were traded than all forms of imported beads. They sold at a price of one faracola of beads for one mitical of gold, i.e. 1s. 6d. per lb. weight. The normal Portuguese trade beads, contas meudas (“small beads”) in black, green, blue and yellow – cost 24 to 30 miticals per faracola, i.e. between 37s. 6d. and 47s. per lb. The greatest demand was for contas de Cambaya (“Cambay beads”) which were made in red and black and worth 40-75 miticals per faracola, i.e. between 62s. 6d. and 117s. per lb. These were the beads traded extensively before the arrival of the Portuguese and whose popularity and supply they struggled to control They occurred in two sizes, the smaller the beads the greater the demand and the higher the cost. A variety of European beads of coral, crystal, pewter, jet, amber and blue Venetian glass, all larger than those described, were also imported. They formed only a small proportion of the stock, were not priced and proved impossible to trade.
The caracoes of the inventories undoubtedly correspond to the shell discs of Group 1. The contas meudas are probably represented by the irregular opaque glass beads of Group 2. It is tempting to equate the red and black contas de Cambaya with the transparent glass beads of Group 3 and more particularly with the tiny Indian red and very dark plum oblates of Group 4. Of the other beads mentioned in the inventories, only coral beads occur in the present assemblage, though pedras cristaes, of which only fifty stones are listed in the 1516 Sofala Inventory, could well describe the four large clear moulded glass beads in the necklace of Burial E1.
Metal Beads and Bangles
Copper or bronze bangles, bracelets, anklets or necklaces were worn by all the adult Negroid burials recovered, four women and a man, but were not found on any other body.
Copper or bronze wire, with a square or rectangular cross-section 0.2-0.5mm. thick, was wound to form coils from 1-3.5mm. in diameter, occasionally in the examples, over a fibre or bast core. Most frequently the coils were 2-2.5mm. in larger diameter with from 14-20 coils to 1cm. (in the classification of metalwork used in Appendix II, (missing) Inventory of Burials, this wire is Class A).
A thicker wire, 1.2mm. wide with a semi-circular cross-section, was used to make bangles, wound in a coil 3mm. in diameter with seven or eight turns to 1cm. The bangles were from 58-62mm. in diameter. The curved face of the wire was always turned towards the outer perimeter of the bangle. Such bangles are common: Burial C4 wore at least 45 (Class B).
Copper or bronze beads were most commonly formed by clipping a short length of wire over a fibre core, leaving the ends butt-jointed. They vary from 1-2mm. in width, from paper thinness to 1mm. thick and from 2-4mm. in diameter. The larger beads were commonly used to form bracelets, the beads being clipped firmly and continuously over a fibre core, always with the butt ends facing inwards and the curved face of the wire outwards so that the bracelets look very like the wire bangles just described. Such bracelets, of which nineteen were found, contain from 80 to 126 beads (Class C). In five instances, these beads were clipped over lengths of coiled wire, at regular intervals, usually 5mm. apart, so form more ornate bracelets, each with some 45 beads (Class D). Groups of from 12 to 36 of these beads also occur together, strung with shell beads.
Cast copper or bronze beads are much rarer. Bicones, 2mm. wide and 3.5mm. in diameter, were strung at 4mm. intervals on coiled wire cores in eight bracelets (Class E). In one such bracelet, three strands of coiled wire are twisted together to form the core of the bracelet. A single, spherical east copper bead, 6.5mm. in diameter, was found on the original ground surface.
The arms of one burial (C4) were adorned with 45 bangles interspersed with at least 104 coils of wire, the metalwork on each arm weighting 1lb. Burial B4 wore 36 bangles and bracelets of various types, interspersed with at least 68 coils of wire, the metalwork on each arm weighing between 1lb. and 11/4lb. Necklaces were not worn by any female. The one male (E1), however, wore a necklace of small copper beads with four large clear glass beads at the front, and coiled wire bangles on the right arm, below the right knee and at the left ankle. The Negro female burials B1 and C4 wore a great number of wire ankles. From 80 to at least 112 coils of wire (Class A) encased the legs for up to 23cm. from ankle to mid-calf, the weight of anklets on each leg reaching between 4 and 43/4lb.
One small fragment of an iron bangle was recovered (from Trench C, stratum 8A). This was of wire with a circular cross-section, 1.5mm. in diameter, wound in a coil 5mm. in diameter.
Copper, bronze, silver and gold accoutrements were found on several Caucasoid male burials (Garlake, 1968c). A gilded, bronze religious medallion (fig. 7.4; fig. 8; fig. 9) from the neck of Burial A4, bears on one side the profile of a female head wearing a crown and encircled with a halo with the inscription S. ELISABET R. LUSITANIA. On the other side is the profile of a cloaked, tonsured and haloed man carrying the figure of the Christ Child who holds an object, probably a crucifix, towards the man, with the inscription S. ANTONII…..V, with ROMA in the exergue. These represent Elizabeth, 1271-1336, wife of King Denis I of Portugal and Antony of Padua, 1195-1231, a Portuguese by birth (Thurston and Attwater, 1956). This representation of St Antony was only introduced in the late sixteenth century. The medallion must have been struck after Elizabeth’s canonisation in 1625, presumably in Rome for even her name takes the Latin form, and not the Portuguese, Isabella. This date is of little help in dating the site more closely as the burial took place late in the history of the site (stratum 6).
A tiny, heavily worn, bronze medallion was attached to a “rosary” of ivory beads hanging from the girdle of the Negro female Burial B1 (fig. 7.5; fig. 8; fig. 9). It has on one side a full length female figure, possible with a halo, standing beside an unidentified object with an inscription which appears to read G. EATIS. On the other side is a barely distinguishable full length figure standing on a crescent moon, surrounded by rays and probably crowned, probably representing the Virgin of the Assumption.
A rectangular tablet of the fine clay was found against the right humerus of the Caucasoid male Burial C5 (fig. 7.6; fig. 10). It bears an impression of the Virgin in glory, crowned, holding a sceptre, with the Christ Child in her arms and standing on a crescent moon and clouds. Below is an armorial shield apparently bearing only a single crescentic device, surmounted by a three-armed cross – the lowest arm of which is considerably shorter than the others – and supported by brackets which also support the clouds. The significance of the armorial bearings or cross is not known. This tablet was either placed loose in the grave, was contained in a pouch or pocket or was bound to the arm for there is no means of attachment on the fragile tablet itself.
Several fastenings for clothing were preserved with the burials within the Church, Three (A4, A8, AA1) wore aiguillettes on the right hip – small, hollow metal rods that formed decorative ends to the cords holding parts of a man’s clothing together (Gay, 1887; Meyerson, 1942). Two are of bronze and have clenched right fists at the tip and a red wax in their interiors (fig. 7.7, 8; fig. 11). The third is of silver, with four vertical grooves and a continuous spiral engraved round the exterior and beaded ends (fig. 7.9; fig. 11). Across the interior of the top is a small horizontal bar for the cord attachment. Burial AA2 wore an epaulette on the left shoulder: a thin copper plate curved to the shape of the shoulder with a latticed strip on the outer face to take a 1cm wide strap (fig. 7.13). At the front of the waist of this body was a silver buckle to take a strap of much the same width (fig. 7.12). This is engraved with Maltese Crosses or four-leafed rosettes within oval frames.
At the neck was the fragment of a tiny copper pin and at the breast the silver casing of another pin (fig. 7.10). The bronze casings of further pins were found on the hips of A8 (fig. 7.11) and the breast of A3, while an iron pin, 22mm. long and 0.7mm. in diameter, lay on the breast of A1. A copper hook and eye fastening was found with the disturbed bones of Burial A6 (fig. 7.14).
Rings were worn by several bodies, gold in the case of the Caucasoid burials and copper in the case of Negroes. A plain gold ring, 15mm. in diameter internally, was worn on the left hand of Burial AW2. A gold ring, 17.5mm. in diameter with nine diamond facets round the exterior was found loose amongst the scattered bones of Burial AA5 (fig. 7.15). Burial A4 wore a thin gold ring, 14.5mm. in diameter and decorated with two horizontal grooves, on the little finger of the left hand (fig. 7.16). The young Negro female Burial B4 wore five copper rings, all 17mm. in diameter, one of which is unevenly facetted and another fluted. Two of these have the ends simply butted. The Negro male Burial E1 wore a copper ring, 4mm. wide and 18mm. in diameter, crudely engraved with a series of oblique lines and the ends butt-jointed, on the fourth digit of the right hand (fig. 7.17).
There is little intrinsically remarkable about these objects. Their craftsmanship is ordinary and the designs have little individuality or artistic merit. The religious medallions were clearly mass produced to stereotypes designs. The copper rings of the Negro burials are particularly poorly made while the silver and gold rings and adornments all have a high copper content.
Almost all the metal accoutrements have retained traces of woven fabrics with which they were in contact. The best preserved examples were found on the legs of the Negro females B1 and C4. These were tightly wrapped in at least three thicknesses of cotton shroud. No selvage, seams or sewing are visible. The cotton (Gossypium sp.) has a short fibre length and was hand spun, giving a loose yarn of very variable thickness. It was hand woven in a loose, irregular plain weave – of the simplest type, “one up, one down” – with a count of 33/33 threads to the inch. It has not yet been possible to determine the species or variety of cotton used.
Close to the left ilium of Burial AA6 was a thin roundel of red wax, bearing the impression of a square seal 13mm. across. Within a plain frame is a shield, 9mm. across, bearing a coat of arms. This is not decipherable for it is not only extremely small but the wax has weathered slightly. The shield, however, is charged with an in escutcheon (5mm. across) as is the coat of arms of Portugal. The devices that appear in the border and on the inescutcheon may well represent the castles and quinas of Portugal but this is not certain. There are no supporters or crest. The seal appears originally to have adhered to a wooden base.
From the surface of Trench AA came the broken base of a finely proportioned turned lead baluster stem, possibly of a candlestick (fig. 7.18). It must originally have been set in a wooden base, for the bottom remains rough and uneven.
In the fabric of the main wall of the second church (top of stratum 5C, Trench A) was a carved schist cylinder tapering to a cylindrical head, grooved at its base (fig. 7.19). This seems very likely to have been a plumb-bob which would have been suspended by a line tied round the base of the head. It was presumably used and eventually lost during the erection of the building.
The original ground surface (stratum 13) contained many finds besides the beads and pottery already described. Animal bones and teeth, sometimes charred, were common. These certainly include cattle bones but have not been further analysed. Finds of stick-impressed daga occurred but were extremely rare.
Two large broken dolerite mullers, with completely flat, polished bases and irregular sides and tops, and a large fragment of aplite, retaining a portion of a completely regular and smooth cylindrical hole, 41/2 in. in diameter and exceeding 2in. in length – possibly a fragment of a rotary quern, were also recovered from this level. Finds of iron slag, up to 21/2 lb. in weight, were common throughout the deposits. Most were probably derived from the original surface; through strata 2 and 3, postdating the occupation of the site, also contained numerous pieces. Two probable fragments of tuyères were also recovered.
Four lengths of thin copper wire from the original ground surface, 0.2mm. in diameter and from 6 – 20cm. long, had been neatly coiled and tied into separate bundles.
That such small pieces of wire should be worth such careful attention contrasts with the extremely lavish use of similar wire in the ornaments on the Negro burials. This bears out the evidence of the glass beads that there is some economic distinction discernible between the original occupiers of the site and the Negro occupants of the graves.
A single iron arrowhead with a 4cm. long barbed head with a Z-shaped cross section and an extremely long tang (20cm.) came from the bottom of a grave fill (Burial AA3, stratum 12) but is almost certainly derived from the original ground surface (fig. 7.21).
Thirty-two iron nails were found in the trench dug through the deposits long after the site was abandoned (stratum 1A in Trench AA). All have a square shank, from 2-6mm. across at the top and malformed rectangular heads. They vary in length from 3-9.5cm., the most common being 5cm. long. These may well be of comparatively modern origin, though they may derive from an original deposit in some other area of the mound through which this trench may have cut. Eight iron nails were recovered from earlier deposits (three from Trench G, stratum 3A; three from Trenches A and G, stratum 13; and two from grave fills. Fig. 7.20). Two are only fragmentary but four complete examples have a square shank, 4mm. across and from 45-85mm. long. Two small examples have roughly circular heads; the others have the top of the shank flattened and lack heads.
Stone Age material occurred in some quantity throughout the deposits, with the main concentration in and immediately below the original ground surface. Over 100 waste flakes were found, together with one core and very occasional scrapers and utilised blades. Such finds are normal in any Iron Age excavation of any size and have little significance beyond indicating a probably disturbance by the Portuguese of a Later Stone Age deposit in the area.
Portuguese settlements in Rhodesia have now been identified at Dambarare, Luanze, Ongoe (Angwa River) and Maramuca (fig. 1). Ongoe is unexcavated; at Luanze and Maramuca small test excavations have taken place (Garlake, 1967). The excavations at Dambarare were on a more satisfactory scale but still represent only one glimpse of one aspect of a single building of which less than 2 per cent was excavated. That is was the church and that it contained numerous burials is useful as it produced considerable evidence on racial identities and personal possessions. However, a building fulfilling such a restricted function may give an unsatisfactory or incomplete overall picture of the life of the settlement and particularly of its indigenous inhabitants. This is partly corrected by the finds from the original ground surface, occupied before any buildings were erected although of very much the same date. Here there is slight evidence of a divergence between the goods owned, traded and lost by the ordinary people and those buried with the bodies in and around the church.
The choice of the site at Dambarare must obviously have been determined primarily by the proximity of gold reefs but it also offers perennial water from the Marodzi River and light, easily tilled granite soils.
The construction of every known Portuguese earthwork in Rhodesia is closely similar. Very large areas – of which the Dambarare excavated site is by far the largest – were enclosed by straight brick walls so low that they must have formed only the cores of earth banks or foundations for timber palisades. In several cases a ditch was also cut outside the walls. Inside, there was usually little more than a single brick building and a few pole and daga huts, with no trace of intensive or prolonged occupation. There is some variation in the plan of the earthworks from site to site: at Luanze and Maramuca there are bastions at the centre of each outer wall, at Ongoe two bastions are usually projected from diagonally opposite corners while, with one exception, the Dambarare enclosures lack any bastions.
Portuguese techniques of planning and construction were used in all major structures – though some subsidiary buildings were of pole and daga. All building materials were obtained from sources in the immediate vicinity of the sites, the basic building unit being a very large sun-dried brick. Floors and mortar were of clay. The Dambarare bricks were identical in size and fabric to the bricks used in the enclosure walls and buildings in the Maramuca earthwork – the only other brick structure excavated. Though considerable labour, at least semi-skilled in Portuguese building techniques, must have been employed, standards of craftsmanship were mediocre – not only was there great variation in the manufacture of bricks but they were laid with very little care. Despite the extreme simplicity of buildings, their scale is impressive.
The walls, bastions and ditches show that the necessity for protection was clearly envisaged, although effective defence against attack would have necessitated extremely large garrisons. Moreover, most earthworks are sited on level open ground and none are sufficiently close together to give mutual support. Military defence was thus clearly considered of minor importance. The close similarity of all plans clearly suggests that the enclosures normally fulfilled similar functions as completely independent units, for each is usually isolated by a considerable distance – at least half a mile – from its nearest neighbour.
Nowhere was there an interdependent community, let alone a village or town. Doubtless, as documentary evidence also suggests, each earthwork represented a separate trading or mining concern operating on some considerable scale with large numbers of people. However, occupation of the enclosures by large populations could not have been anything more than intermittent, as signs of permanent dwellings, middens or intensive occupation deposits are absent. A small permanent population may have been supplemented by short seasonal influxes of large trading groups.
There seems little doubt that the Dambarare excavations have revealed a small portion of what was probably the most important burial ground used by the Portuguese during their seventeenth century settlement of Mashonaland. Should the 98 per cent of the area still unexcavated follow the same patter, it would contain some one thousand further burials. There is no evidence from the present excavations to suggest that this will not be the case. This emphasises the danger of generalising on the material so far recovered. The Portuguese inhabitants, exclusively adult males, many of who had barely reached maturity, were usually buried within the church. The few items of clothing and possessions that remain are of little intrinsic note. No analysis of the skeletal material has taken place, beyond racial analysis of the skulls.
It is only worth remarking that while no bones exhibited lesions, almost all the mature Caucasoid males had lost many of their teeth during life. In the African burials the only sign of skeletal damage during life was a partially healed hole in the skull of the mature female burial B3, though the mature adults again had had several teeth extracted.
The African burials, interred outside but in close proximity to the church, are of greater interest. They were almost certainly people of considerable economic and social status, who may well have belonged to the ruling group. Although they had abandoned indigenous burial customs and were almost certainly Christian, their adornments show considerable affinities with those worn by their Rozwi counterparts inhabiting the stone buildings of Matabeleland (Butua), outside Portuguese influence. The quality of metal bangles or anklets worn by the adult females was considerable: one young girl wore ornamental metalwork weighing over 11lb. Isolated examples of closely similar ornaments made with the same techniques have been found at numerous late Rhodesian Iron Age ruins (i.e. Khami, Dhlo Dhlo, Zimbabwe: Randall-Maclver, 1906; Chibvumani, Matendere, Zimbabwe: Caton-Thompson, 1931; Khami; Robinson, 1959). Anklets and bracelets worn by some Venda women in the Limpopo Valley today are also closely similar: they include bangles formed by clipping beads over a coiled wire core. The closest parallel, however, comes from two Negro female burials excavated by Caton-Thompson within the Dhlo Dhlo ruins, associated with early eighteenth century imported ceramics (Caton-Thompson, 1931). Both wore coiled wire anklets and one had coiled wire bangles on one arm. Otherwise, the Dhlo Dhlo burials contained only typical Portuguese trade beads (cf. Class 2 beads, above): a much greater variety of imported trinkets, especially beads, was available to those living under direct Portuguese influence.
The commoner glazed ceramics and trade beads imported by the Portuguese during the seventeenth century occur in quantity at Dambarare. They are closely related to finds from the major Matabeleland Iron Age ruins of Khami and Dhlo Dhlo. The dating evidence provided by these ceramics has been fully discussed elsewhere (Garlake, 1968a).
The indigenous pottery used at Dambarare and Luanze during the seventeenth century is distinct from contemporary wares from other areas of Rhodesia. Though this pottery cannot yet be adequately defined, its typological origins may lie in the stamp-decorated wares of the Early Iron Age. The modern pottery of the Korekore tribes of the extreme north of Mashonaland may also be related to it.
Certain anomalies between the evidence provided by these excavations and documentary evidence or previous conjecture remain to be discussed. Though the building excavated is clearly part of the church of Dambarare, no sign has yet been discerned of the damage caused by Changamire Dombo in 1693, when, as described by Conceição, he destroyed the church furniture and disinterred some burials. Such evidence may still appear. However, it is not certain that Changamire fired or demolished the building and these are the only actions that would definitely leave their trace in the archaeological record. The scattered and badly decayed bones of Burial AW1 (at the base of a deposit formed at the final abandonment of the site) may have been disinterred at this time. Though Conceição states that the church was within the fort there is no indication that the excavated earthwork contains any specifically military installations.
Earthwork 4 (fig. 2) on the banks of the Marodzi River is more likely to have been the fort of the settlement, for it appears to be more defensively sited and to have a defensive outwork, while de Almeida y Sousa states explicitly that the fort was “a stone’s throw” from the river. Earthwork 4 has little sign of buildings within it and is most unlikely to have contained a church.
The Kasekete and Mutota ruins in Zambezi Valley are poorly constructed Iron Age schist enclosures, the former containing a rectangular loop holed building of stone and mortar. This has been suggested as the early seventeenth century Zimbabwe, of the Mwene Mutapa Mavura II which contained a small Portuguese garrison (Abraham, 1959). While there can be little doubt that the rectangular building is alien in origin, no Portuguese trade goods have been recovered from it.
The indigenous pottery collected from the surface of both ruins, has nothing in common with Dambarare and Luanze wares but is related to the ceramics of the Rozwi, as found at Khami (Robinson, 1965). It is improbable, therefore, that these ruins are contemporary with or related to Dambarare. They may possibly date from the eighteenth century, when the Portuguese was able to re-establish themselves to a limited extent as some sites in northern Mashonaland. A small square enclosure at Makaha, near the Ruenya River, two walls of which are of stone set in a clay mortar, has also been ascribed on architectural grounds to the early Portuguese (Whitty, 1959). There is as yet no archaeological evidence to support this.
There is clear stratigraphic evidence in the Dambarare excavations of two distinct periods. The first was of considerable duration; marked by a steady evolution of the structures and continuous burials, culminating in the erection of the first church (the red brick building: strata 13 to 8). This was demolished and a considerable number of burials took place after its demolition (strata 7 and 6). Thereafter, there was a second period, probably of comparatively short duration, during which considerable, if intermittent, building work took place. The church was entirely rebuilt and a series of minor alterations and additions were made to it (strata 5 and 4). The great majority of the imports recovered belong to the first half of the seventeenth century. It is tempting to equate this sequence with the historical evidence that Dambarare enjoyed considerable prosperity for a long period between at least 1631 and 1675, that it was then depopulated by disease and almost entirely abandoned between 1675 and 1684, when it was re-established and the fort at least was rebuilt. This short period of intense activity ended in the final destruction of 1693. The archaeological dating evidence is, however, at present not sufficiently precise to confirm this correlation.
The mound that was excavated was shown to me by Mr. H. Light of Bulawayo who, following a newspaper report on excavations at Luanze, recognised the significance of a site he had seen over forty years ago. Mr. and Mrs. J. Collcutt gave permission to excavate and generous co-operation, interest and hospitality. The initial excavation team was organised and largely staffed by the History Department of the University College of Rhodesia. The 1967 Ranche House School of Archaeology held a short training dig at the site. Mrs. J. Andrew, Mrs. E. Knopf, Mr. J. MacKenzie and Mr. E. Mandaza further assisted in the excavations. My wife undertook the recovery and preservation of the many fragile finds. The cranial material was analysed by Dr. H. de Villiers of the Anatomy Department of the University of the Witwatersrand Medical School. Her report is included as Appendix I. Mrs. E. Goodall of the Queen Victoria Museum, Salisbury, allowed me to examine her surface collection of imported ceramics from this site. Mr. J. Kirkman of Fort Jesus, Mombasa, identified a representative series of these and read and criticised a first draft of this paper. Dr. J. Thompson of the Forensic Science Laboratory of the British South Africa Police examined the material of the shrouds. I have had most useful discussions on Portuguese documentary material with Mr. D. Abraham of the University College of Rhodesia. Fig 2 is based on Crown Copyright topographic material by permission of the Surveyor General of Rhodesia. I am most grateful to them all.
A preliminary report on the excavations was presented at the 66th Annual Congress of the South African Association for the Advancement of Science at Lourenco Marques in July 1968.
Abraham, D.P. (1959). The Monomatapa Dynasty. Nada, 36.
Abraham, D.P. (1961). Maramuca: an exercise in the combined use of Portuguese records and oral tradition. J. Afr. Hist, 2, 2.
Axelson, E. (1960). Portuguese in south-east Africa 1600-1700. Johannesburg: Witwatersrand University Press.
Bent, J.T. (1892). The ruined cities of Mashonaland. London: Longmans, Green.
Caton-Thompson, G. (1931). The Zimbabwe culture. Oxford: Clarendon Press.
Ferguson, J.C. and Wilson T.H. (1937). The geology of the country around the Jumbo mine. Mazoe District. Salisbury: Geological Survey Office.
Garlake, M. (1969). Recovery of fragile objects from an excavation. S.Afr. archaeol. Bull., 24, 94.
Garlake P.S. (1967). Seventeenth century Portuguese earthworks in Rhodesia, S.Afr. archaeol. Bull., 21, 157.
Garlake, P.S. (1968a). The value of imported ceramics in the dating and interpretation of the Rhodesian Iron Age. J.Afr. Hist., 9, 1.
Garlake, P.S. (1968b). Test excavations at Mapela Hill near the Shashi River, Rhodesia. Arnoldia, 3, 34.
Garlake, P.S. (1968c). Some early Portuguese relics from Dambarare, Rhodesia. Rhod. Sci. News, 2, 12.
Gay, V. (1887). Glossaire archeologique du moyen age et de la renaissance. Paris: Librarie de la Societe Bibliographique.
Kirkman, J.S. (1967). Some conclusions from excavations on the coast of Kenya, 1948-1966. Paper delivered at the Nairobi Conference on “East Africa and the Orient”, 1967.
Meyerson, A. (1942). Bindiker, nålremmer och ägiljetter. Liv Rust Kammáren. Stockholm.
Newitt, M.D.D., and Garlake, P.S. (1967). The “aringa” at Massangano. J.Afr, Hist., 8, 1.
Randall-Maclver, D. (1906). Mediaeval Rhodesia. London: Macmillan.
Robinson, K.R. (1959). Khami Ruins. Cambridge: University Press.
Robinson, K.R. (1965). A note on Iron Age sites in the Zambezi valley and on the escarpment in the Sipolilo District, Southern Rhodesia. Arnoldia, 1, 27.
Schebesta, D. (1966). Portugals konquistamission in sudost-afrika. St Augustin: Steyler.
Silvo Rego, A.da, and Baxter, T.W., eds (1962-65). Documents on the Portuguese in Mozambique and Central Africa, Vols, I-IV. Lisbon: Centro de Estudos Historicos Ultramarinos.
Theal, G. (1898-1903). Records of south-eastern Africa, Vols. I-IX. Cape Town: Government of Cape Colony.
Thurston, H., and Attwater, D., eds (1956). Butler’s Lives of the Saints. London: Burns and Oates.
Lowe, C.van Riet (1955). The glass beads of Mapungubwe. Pretoria: Archaeological Survey.
Whitty, A. (1959). A classification of prehistoric stone buildings in Mashonaland, Southern Rhodesia, S.Afr. Archaeol. Bull., 14, 54.