Printer Friendly

The Swahili and the Mediterranean worlds: pottery of the late Roman period from Zanzibar.

Mortimer Wheeler famously tied together the worlds of ancient Rome and ancient India by finding Roman ceramics stratified into levels at Arikamedu, south India. Late Roman pottery from far down the East African coast new permits the same kind of matching link from the Mediterranean to a distant shore, this one in the Swahili world.


The islands of Zanzibar are situated off the Tanzanian coast, south of Ras Hafun (Hafun on FIGURE 1). The pottery under discussion has been excavated from a site on the southern part of the main island of Zanzibar, called Unguja Ukuu, where a sandy harbour is located at one of the closest points to the mainland and is the best anchorage in the southern part of the island. Until well into the 20th century it has been a place for arrival from the mainland coastline. The archaeological deposits cover about 16 ha. In the deepest areas including a mound, the cultural deposits are nearly 3 m deep. According to present knowledge, this is the oldest site occupied on the islands. Alongside enormous quantities of material of East African provenance from systematic excavations that I directed, appreciable quantities of broken imported ceramics and other exotic articles are recovered.

Some few exotic finds recovered from the earliest level are unusual on the east African coast; they include ceramics. Here, I would like to discuss only two types: Pots A and B from excavation Units L and J respectively [ILLUSTRATION FOR FIGURES 2 & 3 OMITTED]. The units are about 100 m apart. Radiocarbon determinations on charcoal samples were obtained from other units (but not from these); they confirm the primary occupation of the site spanning from the late 5th century to about 1000 AD. In TABLE 1 below are shown the earliest determinations for the site (Stuiver & Becker 1986). The pots have been excavated from the lowest levels; their date is deduced by correlation with material from these other units supported by such determinations.

Pot A (FIGURE 2)

This dish-shaped pot is small with a wall about 0.4 cm thick. The clay is medium coarse-textured and fired to orange, but the pot has an un-oxidized grey core. Inclusions consist of small sandy grits. The flat base rises in a roughly triangular shape; it juts out slightly from its low flaring wall. On the underside it has a relatively broad groove, resting the vessel on a sagging floor. It measures about 0.6 cm thick and has a diameter of 14 cm.

The rim is somewhat expansive with an outer edge slightly curving downward and gives a triangular impression. It is about 0.7 cm thick and 1.1 cm broad. The outermost diameter is 15 cm. The vessel height is about 2 cm with an internal depth of only 1.3 cm. The vessel is burnished with a matt red slip (Munsell 2.5YR 5/8) all over the surface. Some glittering flecks are probably traces of mica. Under the rim, the surface has some scraps of black irregular spots or 'pops', perhaps representing lime impurities left during washing of the clay. The black colouration may be due to dirt stain or, as Hayes (1972: 13) suggested, an effect of firing. The slip is not the only embellishment; a fine single groove impression runs on the top surface of the rim closer to the interior of the vessel. Four examples of Pot A type were identified from the excavations; one is presented here, the others are very fragmentary.


This is a carinated pot with a necked profile. The wall is thin, average thickness 0-5 cm. It derives strength from good firing. The upper part of the neck and the lower part of the body are tapered. The clay paste is reddish brown (Munsell 5YR 5/3), perhaps fired in a reducing atmosphere, and of sandy fine texture. The brownish tinge may be an effect of weathering. Few black grit inclusions can be observed, perhaps the result of unfinished cleaning of the clay. The base of this pot is not preserved.

Likewise, the complete rim profile is not preserved as the ledge is broken. The lower outer edge is intact and reveals a fine impressed groove decoration. From the section the thickness of the ledge is 0.5 cm and rim diameter may be 24 cm. An elegant ridge runs on the neck parallel with a row of punctates executed beneath it. The surface of the vessel is smoothly finished and shows no traces of a slip or glaze. Remains of carbon or the effect of fumes from burning, notable in the tapering part of the body, suggest the pot was used for cooking.


As most of the early African pottery vessels produced south of the Sahara were hand-made, evidence of the use of wheel is one of the significant criteria for foreign attribution. The two pot examples described above were produced on a wheel and unlikely to have been local imitations of the imported ceramic.

Most of the early imported pottery from archaeological excavations in East Africa comes from Persia, southern Arabia, India and the Far [TABULAR DATA FOR TABLE 1 OMITTED] East. The wheel-made red-slip wares are often suspected to have come from India, a region known for its polished red-slip wares. I have had no adequate opportunity to examine publications on the polished red-slip wares excavated in India. My appraisal of the Indian material relies on Orton's (1991) examination of an enormous quantity of ceramics from about 100 previously surveyed sites in Gujarat and catalogue of the polished red-slip wares from 12 sites. They are evenly-fired wheel-made pots produced from fine levigated clay and date from about the 1st century BC to the 5th century AD. I could identify no clear parallels with the two pots from Zanzibar.

Another region known for wheel-made red-slip ware is the Mediterranean basin. Hayes (1972) provides the standard work with examples of form, surface finish and chronology relating to the two pots. He gives the brief history of the late Roman wares. From about the end of the 1st century AD the North African pottery tradition inspired the pottery styles of the Mediterranean basin. In this great period for trade, the standard wares were generally based on coarser orange or red clay ingrained with a few particles of mica and finished with a red slip. The colour of the slip is just slightly deeper than the body-clay, a feature which readily distinguishes these wares from the glossy Roman pottery terra sigillata of the earlier period. The elegance of the pottery was in the quality of the fabric and the thinness of the walls; grooving and roulette designs were only occasionally applied for ornamentation. Hayes (1972) called these vessels 'African Red Slip wares'.

Production of the wares continued almost to the 7th century AD; most are looted bowls. Hayes could suggest a chronology, as they changed form over short periods. From about the 4th century AD, trade was revived when the foot disappeared from the bowls, displaced by a low base moulding or more commonly a slightly hollowed base. The rims are flat with a diameter range of 20-50 cm, sometimes giving an impression of a triangle. The decoration on the rim is a groove. It was for a rather short period that the group remained in favour. Hayes (1972: 13-16) indicated that in the later part of the century, the footed base generally returned to fashion. The 'Red Slip Wares' suspect to chronological problems have beaded decoration on the rim (Hayes 1980: xviii).

Pot A

The morphology, giving a pretty good parallel, suggests it is an import from the Mediterranean world. The dating of the specific features broadly corroborates the chronology of the pots in the local archaeological context. The flat base may have marginally continued for some decades while the footed base was back in fashion.

What is the exact place of origin for Pot A? The base, the slip and the rim may suggest further clues for consideration.

Among the regional variants, Pot A does not appear close in shape to any illustrated examples from the Egyptian Red Slip wares. In most illustrated varieties, the base fuses with the body without giving an angular profile at the point of convergence, the feature Hayes describes as 'oblique rising'. Pot A is associated instead with the few varieties in which a protruding base is the striking feature. In contour, the base forms an angle with the body, jutting out at the angle of convergence. Better examples may be found among the pottery of Ashmunein in Middle Egypt (Spencer et al. 1983: figure 66). They are associated in this preliminary report with Level A I IV (Group 1) and described as 'Fine ware baking pots'; the illustration is reproduced as FIGURE 4. Note in particular the base in example 5; the impression is that Pot A is of Egyptian origin.

The next attribute is the slip, for which I present the opinion of D.M. Bailey (pers. comm.) of the British Museum, as the tone of the slip in Pot A is lighter than the pottery he knows from North Africa; he thinks it probably originated from Egyptian workshops. H. Wright, who has worked with Egyptian material, shares the opinion.

The last attribute, the rim, has details which contrast with the Egyptian examples. In the Ashmunein examples, the rims are not triangular. The closest example, no. 5, has a vertical outer edge and a broader groove or somewhat concave top surface; in Pot A the edge of rim is different.

The fine fabric of Ashmunein assemblage and the coarse fabric of Pot A is another point of contrast. Disparity in fabric is not critical as the 'African Red Slip Wares' were produced in varied fabrics both in Egypt and in other provinces (Hayes 1972: 387-401). Alternatively, the contrasting features mark a caution not to rule out a possibility for the pot to have come from another place.

Pot B

With Pot B, close similarities may be found among a set of pottery also published in Spencer et al. (1983). A total of 17 pottery types (Spencer et al. 1983: A-Q, figure 62) are illustrated which Bailey entitled 'More groups of Late Roman pottery'. Group I, described as 'Fine cooking-pot fabrics', is reproduced as FIGURE 5. The characteristic colours range between brown and red. Some examples have cores fired in a reducing atmosphere. The fabrics contain mica, the rim is occasionally painted with white spots, and the neck is decorated with fine grooves (Spencer et al. 1983: 42). This description of form, fabric and function almost parallels Pot B under discussion, see specifically example no. 3 (formerly no. 11.1). The white paint, possibly a peculiarity of Middle Egypt, cannot be seen in Pot B. The incised decoration with a row of punctates, observed in Pot B, is not uncommon in these wares; the group under discussion offers examples but such decoration is more profound with examples illustrated in Groups E and H. In a second opinion communicated, Bailey provisionally dates these Ashmunein 'cooking-pot fabrics' to c. AD 400-550. Again the fabric, form and chronology of Pot B concur with Egyptian material.


Attributes of form, fabric, surface finish and chronology make a good match between the two types of pottery found in Zanzibar and the examples from the Mediterranean world referred from the publications. At least with Pot B, the evidence indicating that it was used for cooking appears to correspond with its function in the land of origin. Both types of pottery could be prestige items used by a few members of a local elite in the beginning of the town settlement in east Africa. They were produced in the Mediterranean world probably from the 5th century and remained current until the early 6th century AD. Their presence in East Africa along with other uniquely early material from the same site suggests a possible trade with the classical world in which the earliest settled community on Zanzibar may have participated. The rarity of the examples in the lower archaeological levels and their absence from succeeding levels may suggest this early pattern of trade was diminishing. Written sources suggest that Roman merchants did not participate during this time. The late Roman pottery possibly reached south of Ras Hafun with Axumite or Persian traders. About this time, they were in control of the trade in the western part of the Indian Ocean and carried out overseas journeys (Hourani 1951: 40). Links to Egyptian workshops may suggest connections with the Red Sea ports rather than direct with other parts of the Mediterranean.

Chittick (1986) excavated some imported pottery at Ras Hafun sites on the Somali coast dated from the 1st century BC to the 5th century AD. Within the broader region of east Africa, this is the first decisive evidence for that early pattern of trade. The material from the Hafun Main Site, 2nd-5th centuries AD, is discussed by Wright & Smith (1990). They also acknowledge the trading contact between the region and the classical Mediterranean world, but maintain it was indirect, through middlemen in Arabia or the Red Sea. The scenario suggested in textual sources is that the Roman subjects were substantially involved only in the first two centuries AD: later the trade declined due to political instability in many areas of the Roman empire. Revival began from the 4th century; most merchandise of the Eastern world was largely under the control of Levantines, South Arabians and Axumites (Sidebotham 1991: 34; Curtin 1984: 90-96).

Economic recession affecting the trade may have been experienced again in the 5th century, making difficult the production or preservation of documentary records explaining events going on in East Africa. South of Ras Ha:fun, as far down the coast as Mozambique, there is evidence for external trade dating earlier than the 8th century (Sinclair 1991). However, accounts of research have not reported finds from regions under the Roman Empire. The discovery of archaeological deposits of the late 5th century or so on Zanzibar, yielding material that links with the ancient pattern of trade, is significant. Hardly anything is known of towns during this period, and perhaps the reason for the paucity of widespread evidence of such trade in East Africa may imply invisibility of the proper sites rather than absence of the trade itself. One possible explanation may be that they are submerged; some evidence suggests the sea was rising over about 2000 years until c. AD 1000 (Morner 1992: 262). At local level, the assertion is generally supported by oral tradition (Freeman-Granville 1962: 299).

The East African coast is generally inhabited by Swahili people. They hosted the foreign merchants to trade in their settlements. Scholars debate their origin in the attempt to associate the wider question of how the early towns which hosted early trade were formed. One opinion says they were Bantu groups still dominating the present cultural setting. Another asserts they originated from pastoral communities ethnically affiliated to Cushitic speakers in the Lama archipelago and only invited the Bantu farming and iron-using communities to settle in their towns as collaborators (Allen 1983). Archaeological evidence of the earliest occupation in this northern part of the coast, attested from the 8th century AD, is used to suggest that it is about this time the pastoral elite, possibly thwarted by drought, left their ranks by adopting sedentism and became the Swahili merchants establishing the towns. Settlements like Unguja Ukuu were dependent successor states. The general hypothesis is already under heavy criticism (Chami 1994). The hint given in this article is just the beginning of the discussion on the major development in the southern part of the coast prior to this period and serves as a further caution. More material concerning the early pattern of trade and cultural development will be published in a special treatise based on the site.

Acknowledgements. I would like to thank all colleagues who helped me with encouragement and comments: in the Archaeology Department Uppsala University, P. Sinclair, F. Herschend, A. Grenberger (illustrations), O. Lindman (photography); L. Johansson and G. Nordqvist (Classical Archaeology and Culture Department); H. Wright (Michigan University): D.M. Bailey (British Museum), who also granted permission to reproduce his illustrations; and other reviewers.


ALLEN, J. DE V. 1983. Swahili origins. London: James Curry.

BEGLEY, V. & R.D. DE PUMA (ed.). 1991. Rome and India. Madison (WI): Wisconsin University Press.

CHAMI, F. 1994. The Tanzanian coast in the first millennium AD. Ph.D thesis, Uppsala University.

CURTIN, P.D. 1984. Cross-cultural trade in world history. Cambridge: Cambridge University Press.

CHITTICK, H.N. 1986. An archaeological reconnaissance in the Horn: the British Somali Expedition, 1975, Azania 11: 117-34.

FREEMAN-GRANVILLE, G.S.P. 1962. The East African coast: select documents from the 1st to the earlier 19th century. Oxford: Clarendon Press.

HAYES, J.W. 1972. Late Roman pottery. London: British School at Rome.

1980. A supplement to the Late Roman pottery. London: British School at Rome.

HOURANI, G.F. 1951. Arab sea-faring in the Indian Ocean in the ancient and early medieval times. Princeton (NJ): Princeton University Press.

MORNER, N. 1992. Ocean circulation, sea level changes and east African coastal settlements, in P. Sinclair & A. Juma (ed.), Urban origins in Eastern Africa: proceedings of 1991 Workshop in Zanzibar: 258-66. Stockholm: Swedish Central Board of National Antiquities.

ORTON, N.P. 1991. Red polished ware in Gurarat: a catalogue of twelve sites, in Begley & de Puma (ed.): 46-81.

SIDEBOTHAM, E. 1991. Maritime trade routes between the Mediterranean basin and the East via the northern end of the Red Sea, in Begley & de Puma (ed.): 12-38.

SINCLAIR, P.J. 1991. Archaeology in eastern Africa: an overview of current chronological issues, Journal of African History 32: 179-219.

SMITH, M.C. & H.T. WRIGHT. 1990. Notes on the classical maritime site: the ceramic from Ras Hafun, Somalia, in Proceedings of the 1989 Madagascar Workshop: 106-114. Stockholm: Swedish Central Board of National Antiquities. Urban Origins in Eastern Africa Project, Paper 4.

SPENCER, A.J., D.M. BAILEY & A. BURNET. 1983. Ashmunein 1982. London: British Museum. Occasional paper 46.

STUIVER, M. & M. BECKER. 1986. High precision decadal calibration of the radiocarbon time scale, AD 1950-2500 BC, Radiocarbon 28: 863-910.
COPYRIGHT 1996 Antiquity Publications, Ltd.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1996 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Juma, Abdurahman M.
Date:Mar 1, 1996
Previous Article:Pattern in the Epipalaeolithic of the Levant: debate after Neeley & Barton.
Next Article:A mutilated human skull from Roman St Albans, Hertfordshire, England.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters