French archaeological mission of Gedi

(Kenya, July - August 1999)

Preliminary report

By

Stéphane Pradines

Paris, 25/02/00

Don't forget to quote : Pradines© 2000

Click on the figure number or on the text to see the pictures in the French version

Introduction

In the framework of a doctorate on Swahili fortifications, we have decided to undertook new archaeological excavations on the medieval site of Gedi in Kenya. This site has been chosen for the quality of conservation of its architectural vestiges, notably of its two town walls, its important surface of more than 18 hectares and its long duration of occupation from the XI^th to the beginning of the XVII^th century. More, our logistics has been facilitated by the National Museum of Kenya infrastructure implanted on the site, recorded as National Monument since 1970.

Ruins of Gedi are at 16 Km to the South of Malindi and at 6,5Km from the sea (Fig. 1). In 1927, the site is declared protected monument. Threatened collapse monuments are restored by the Public Work Department of Kenya in 1939. Gedi is declared national park in 1948 and excavations are programmed under the direction of a British archaeologist, James Kirkman. He make a complete survey of the site, clear many buildings and excavate the great mosque (1954) and the palace (1963).

Our work has consisted to realise test pits on slovenly zones by James Kirkman, but also to make surveys and complementary plans to refine and complete our knowledge on Gedi and its fortifications (Fig. 2).

The great mosque

The most interesting contribution of this excavation is the identification of a new great mosque. This structure is found in the North-east of the site, at the exterior of the two enclosures. This sector is even carefully avoided by the two fortifications, that form at this place a reentrants angle. The excavated and planned mosque measures 26m long northern to the South, It's 6 meters more than the great mosque studied by Kirkman (Fig. 3). The results of C14 analysis, practised on wood charcoals samples, have given dates going from the XII^th to the XIV^th century.

The mihrab is completely levelled, but we have been able to find its basis and a plaster floor in the niche. A tomb was joined to the mihrab with an inscription engraved in plaster on its oriental facade. Unfortunately the or lines superior have been destroyed after the desertion of the structure. The letters are a bi-al-rafà with a ayn not terminal at the end, demonstrates that it's a ta marbuta. The skeleton inside the tomb was oriented North - South, in decubitus lateral left, the face turned to the East and withdrawn hands ahead the face.

An area of excavation has been opened inside the mosque, in front of the mihrab. Besides the different levels of grounds in plaster, the foundation of a large pillar is visible in the section (Fig. 4). To understand the organisation of bays and axes of circulation in the mosque, we have practised a test pit to the South of the prayer hall. The traces of four ground pillars allow to restore a complete plan of the mosque that comprised four bays North - South, two lateral wings and two areas of ablutions East and West.

The superior interface of the layer 453, correspondent to the foundation trench of the wall of qibla, is dated from -635±45(LY-9676), so in calibrated age from 1284 to 1405 ad. We can note that we have not found blues and whites porcelains in stratigraphic context, that confirms antiquity of all this part of the site. A skeleton (US 447) pass under the wall of qibla and therefore clearly anterior to the mosque. It belongs to the same cluster as the burial 445 that is dated from -845±45 (LY-9675), so in calibrated age from 1041 to 1278 ad. The two skeletons are oriented East - West, in decubitus lateral right, face oriented to Mecca.

The materials of superior layers is composed of Yemenites blacks and yellow, or mustard wares, dated from 1250-1350 ad. The sherds of cups, bowls and Ming great dish in celadon are very numerous, with embossed sceneries of the XIV^th century. Some celadons are more ancient and form a transition assemblage from the XIII^th century, notably some Longquan, from southern China (period Song) ; or a Dehua cup, called also Marco Polo ware, from ovens of Fujan in China, dated of the XIII^th-XIV^th centuries. For most ancient levels, only sgraffiatos allow a relative datation as monochrome sherds of blacks and brown of the XII^th-XI^th century, but also an Yueh Yueh celadon (northern Song) and fragments of hatched sgraffiatos from the XI^th century.

Stratigraphics observations, associated with C14 results and with the identification of the sherds by Mrs. Monik Kervran and Axelle Rougeulle (CNRS), allow to specify and to refine the chronology of the building (Fig. 8). The great mosque, in its final state, date of the XIV^th century, it covers an anterior building of the XIII^th century. Finally, burials 445 and 447 are from the XII^th century, they cut a primitive substrata of occupation from the XI^th century.

The necropolis

So as to confirm the datation of the exterior townwall, we have realised a test pit to the South of the city. We wanted also to verify if the defensive works preceded the exterior townwall as the fence and the ditch found by Kirkman to the North. We have not found the researched structures, it shows that the multiplication of defensive work had been made in the surroundings of the two alone entry in the exterior townwall to the North of Gedi.

The excavation has however delivered burials similarly organised (Fig. 5). It concerns a skeleton of a child cut at the foots by a pit and three skeletons of adults. Bodies are oriented feet westward, the head to the east, the face turned to Mecca. The limit of pit was slim and narrow, because the bodies were certainly enveloped in shrouds. A fine bed of white sand was deposited to the bottom of pits. The child is buried less deeply than the other individuals.

Finally, we have exhumed three skeletons in decubitus dorsal, superposed in a pit East-West. Does it concern a collective and simultaneous burial of slaves in a same trench to accompany a or the masters ? The skeleton n°317 cut the burial n°319, that belongs to the group describes higher. The skeleton n°324 is in very bad state of conservation, we have only some cranium fragments and teeth, but the pit passes under the wall of enclosure. The necropolis would be therefore anterior to the fortification of the XV^th century. According to the few available ceramic material, it would date of the end of XIV^th century.

We note that the archaeological deposit is the less powerful to this place of the city. The ceramic material is a less abundant than in the other sectors. This Muslim necropolis is therefore situated to the exterior of the townwall and at a distance of the city. It is able be associated with the mosque against the townwall, localised at some tens of meters more westward .

Internal enclosure

The interior townwall is more belated than the great external urban enclosure. Complementary statements and surveys of 1999 have allowed to determine several periods in its construction. Its elaboration continues from the middle of the XVI^th century until the beginning of the XVII^th century and the desertion of the city. Some portions of the enclosure are sometimes very crude and contain reusing elements of more ancient buildings. Destroyed structures were situated outside of the intern townwall or on its layout. This reutilization of materials has contributed to the rarity of stone vestiges between the first and the second enclosure.

The survey of the North-east quarter presents a great interest for the study of this phenomenon of urban retraction. Indeed, identified houses make a link with the mosque between the two enclosures that seemed isolated from the stone city on the plan of Kirkman. These domestic units present the same orientation East-West, North-South like the houses block cleared by Kirkman. The enclosure take the layout of the walls of internal courtyards, houses and streets of the XV^th century. Several phases are observed in this military work, exits are clogged in the second half of the XVI^th century and in the beginning of the XVII^th century. Houses are completely divided in two by this basic townwall (Fig. 2). The enclosure cut others civil structures to the South of the site and westward. This factor testifies the depopulation of the city du to political problems as the domination of Mombasa and the Portuguese presence ; or economic and ecological problems with the drainage of an arm of the river Sabaki who connected Gedi to the Mida creek on the Indian ocean and its maritime trade.

Only two doors are identified on the interior townwall : an oriental door noted and excavated by Kirkman and a western door in the axis of the first. This entry was condemned in its most recent phase, probably at the moment where others exits are also sealed in the North-east quarter.

A small defensive work localised in the Southeast angle has made the object of excavations so as to obtain chronological elements on the South of the city and on the later townwall. The quadrangular building possesses a murderess on its East flank. This later element is, in the current state of the research, unique on the Swahili townwall. It is without doubt a Portuguese influence. The very simple plan of this building in covering on the urban enclosure is a kind of low tower or shooting platform that one can find in many Swahili cites who are fortified (Fig. 6).

The walls foundation of the defensive work is not very deep. The ceramic material associate dates from the end of the XVI^th century, notably lips of cup from Haïs (Yémen), Bahla ware of Oman, a bowl of Tihama (Yemen), and white and blue porcelain from China : the sherds of great dishes with landscapes sceneries or hatched headband. The basis of the enclosure wall is at -103cm in corrected altitude, almost 2m under the current ground level. The enclosure wall has been erected on a more ancient architectural unit that we have found occupation levels characterised by plaster floors. We have the same diagram of occupation that previously : an habitat of the XV^th century covered by a later defence.

An other test pit has been practised in the Southwest angle of the interior townwall so as to obtain dating elements and a comprehension of a structure with barred angle. The nature of occupation in this sector of the city is very different those described previously. Layers to the exterior of the fortification were composed of garbage reject : sherds, shells, bones of animals. The study of maritime faunal remains has been realised by Sirs François Meunier and Bernard Métivier of the National Museum of Natural History of Paris and Sir Jean Desse of the CRA of Sophia Antipolis. Fish remains consist of inferior pharyngiens of large Scaridae (parrot fish) ; a 5e vertebra of Epinepheliné (merou), a premaxillaries left bone and 2 vertebras of Lethrinidé (emperors). Molluscs are the maritime gastropods : Terebralia palustris, Cypraea tigris and maritime bivalve : Anadara erythraeonensis and Codakia tigerina. Bones of chickens, goats and sheep, are numerous and are often burnt or have butchery marks. The local pottery is far more abundant that in the other sectors. It concerns triangular incised wares, careened pots with triangular motives incisions or knocks of nails on the shoulder (Fig. 8). Some imported wares have been identified, notably Swatow, from southern China, dated of 1550-1650 ad., Iranians sherds of the beginning of the XVII^th century, and many flats lips from Kung beside Hormuz, manufactured at the end of the XVI^th century.

Levels of most ancient occupation are dated from the XIV^th century, by fragments of imported pottery with bluish glaze on yellow paste and blacks and blues sherds. Structures of this period consist pit holes that correspond to the basis of huts or light structures in perishable materials as granaries (Fig. 7). Walls had to be realised with mud applied on wood grid or in mud brick, because we have found a sterile hard clay layer that corresponds to the disappearance of these walls.

Town planning

The new data, brought by excavations of 1999, allow to reinterpret the urbanisation of the city of Gedi. We have a real horizontal stratigraphy, with the old town of the XI^th to the XIV^th century to the North and the recent city of the XV^th century, surrounded by an enclosure. The XIV^th century seems to be the period of maximal extension of the city. The rich imported material discovered during the excavations of the mosque militates in favour of a maximal economic prosperity at this period.

The centre of the agglomeration displaces then to the Southwest. The presence of two great mosques, localised in different places and chronologically successive, is an indication of the moving urban centre that has operated in the beginning the XV^th century. We do not know what are political, economic or religious motivations that motivate this great change. This urbanistic diagram remind the spatial organisation observed at Manda where there exists also a great mosque without datation and outside the townwall of the modern city.

At this moment, the space is managed and rationally planned. The city is enclosed by an enclosure and stone structures are multiplied. The plan of Gedi is organised on circulation axes East-West and North-South. This orthogonal layout of the streets is not an African model. That could explain the title of the first book of James Kirkman : The Arab City of Gedi. This archaeologist has only worked on ruins of the XV^th - XVI^th century and that have probably disturbed its vision the urbanisation of Gedi since it lacked all the ancient city, of African origin . In the second half of the XVI^th century, the retraction of the city ended to a nucleus protected by a second enclosure less elaborated. Gedi is finally abandoned in the beginning XVII^th century.

Conclusion and researches orientations

Contributions of the 1999 excavations are multiple. The history of fortifications allows us to understand the urbanistic evolution of the city and to ask questions on the continuity of Gedi occupation. The excavation of the necropolis and the discovery of burials in the great mosque bring new data on Swahili funeral rituals. Finally the study of this new mosque bring information's on Islamic architecture, but also on economic activities of the city during the XIV^th century.

Results of this mission demonstrate that the study of the site of Gedi is far from being ended. We do not know anything of the city from the XI^th to the XIV^th century and the organisation of excavations in the North sector seems necessary. An hydrographic analysis of the region could reveal where was found the waterways, henceforth dried, that connected Gedi to the sea. Finally, it proves urgent to establish a complete topographic survey of stone ruins localised in the interior townwall. Walls are indeed threatened by the forest and roots.

Specials thanks

At the Institutes, organisations or enterprises that have sustained materially or financially this project : the Fondation de France ; the University of Sorbonne (Paris IV) ; National Museums of Kenya (NMK) ; IFRA (French Institute for Research in Africa) ; the BIEA (British Institute in Eastern Africa) and CORSAIR.

At the administrative and scientific chiefs and all the team : Lorna et Georges Abungu, General Director of National Museums of Kenya ; Marianne Barrucand, Professor of Islamic Archaeology at Sorbonne ; Bernard Charlery de la Masselière, former director of IFRA ; Paul Lane, Director of BIEA ; Omar Athman Lali, Head of Coastal Archaeology; Abdallâh Alausy, Curator of Gedi ; Churchill Abungu ; Ceri Ashley ; Gwenaël Lemoine ; Stephen Manoa et Mohammed Mchulla.