Kilwa: How a Swahili Coast City Grew

At Kilwa Kisiwani, off the southern coast of Tanzania, the available high-resolution chronology places the earliest activity in one trench in the urban centre in the ninth century. This starting point comes from stratigraphy and radiocarbon dating, not from the arrival date of a foreign fleet. It marks only the earliest phase that can currently be dated reliably in this trench; it cannot identify the island's first inhabitants.

Some earlier scholarship explained the formation of Swahili towns mainly as the work of Persian or Arab settlers who founded cities in East Africa. Later archaeology redirected attention to the coast itself: local settlements, a Bantu-language setting, earth-and-timber buildings and long-running maritime exchange all preceded the monumental stone remains. Connections across the sea mattered, but they entered a world in which people were already living and building.

Language can preserve long histories of contact and transmission. Pottery, houses and food remains can show how people used a place. Neither kind of evidence can directly assign every resident a lineage, identity or precise date. Treating “Swahili” as a people who never changed, or only as the result of outside influence, obscures how coastal societies formed differently from place to place.

The first ninth-century phase contained pottery and fish bones, with ash and small pits set into the sand, but no houses were identified. Researchers considered some pits more likely to be connected with cooking or burning, without fixing their function. The material records traces of eating, burning and disposal. It does not preserve these people's names or show how long each stay lasted.

Around 900–1000 CE, several phases of earth-and-timber construction appeared at the same location. Postholes, wall trenches and floors allow at least six buildings to be identified. Imported objects still formed only a small share of the finds. Urban life first becomes visible here in houses repeatedly built, used and taken down. Kilwa's early history enters view through this everyday construction; it does not have to wait for a stone palace to begin.

Tanzanian scholars have also proposed, on the basis of pottery types, that the island's history may extend back to the third millennium BCE. The claim should remain visible as a dispute, but it cannot yet enter the main chronology. A 2022 chronological study notes that the pottery thought to be so early has not been found in dated layers, while radiocarbon dates from the relevant excavations fall between the tenth and thirteenth centuries. Shishu therefore begins the main sequence in the ninth century. This does not deny the possibility of earlier activity; it leaves that possibility open for new dated evidence.

The city did not leap suddenly from earth-and-timber houses into an age of stone. Tenth-century layers repeatedly covered one another as houses, ash deposits and floors were repaired or buried. A single trench can reveal these changes, but it cannot provide a census of the whole city. The low proportion of imported pottery in it cannot measure the scale of trade across Kilwa. The trench is better suited to answering a smaller but important question: people lived here over time and repeatedly remade the spaces of daily life.

At the end of the tenth century, the first securely identified coral-stone building appeared in this sequence. The earliest incised pottery and a silver coin associated with Ali ibn al-Hasan were also placed in this period. Stone walls, incised pottery and coinage now become visible together, but the evidence does not show that one outside ruler founded the city in a single act. It shows new building practices entering an existing settlement; by itself, it cannot explain how power was formed.

Stone buildings could disappear as well. The early building was later demolished, its stone reused, and lime kilns subsequently worked in the same place. From the stratigraphy, researchers concluded that the stone building came before the kilns. Demolition may indicate rebuilding, reuse of material or a change in the use of a space; alone, it does not prove that the entire city was in decline. Urban history sometimes survives in standing walls and sometimes in the way walls were taken apart.

Layers dating to about 1220–1350 did not preserve monumental architecture alone. Ash, deposits and pits show that dumping, burning and the everyday handling of food or waste continued near the urban centre. Archaeological evidence can restore these actions to history, but a layer of ash cannot identify the users' social position, gender or faith.

A cooking area dating to about 1420–1480 preserved a mofa stove and ash pits. The fifteenth century here was a time not only of palaces, mosques and long-distance trade, but also of lighting fires, cooking, clearing away and using space again. We do not yet know who worked beside this stove, and one activity area cannot stand for domestic life across the city. What can be established is that ordinary actions kept their own time as urban politics changed.

Songo Mnara, nine kilometres away, offers a nearby comparison. Geophysical survey and excavation found smaller structures, open spaces and activity areas among the prominent coral-stone houses. This complicates any picture in which a “stone town” consists only of large buildings. Songo Mnara is, however, a different settlement. It can suggest what to look for at Kilwa, but it cannot speak for gaps in the evidence left by Kilwa's residents.

Kilwa's harbour is formed by channels, islands, a lagoon, reefs and anchorages. They offered shelter to vessels, while tides, coral reefs and narrow entrances also created danger. A harbour exists through the ways ships enter, leave, anchor and exchange; it is not an isolated building. The coastal landforms supplied the conditions, and repeated action made this place a port in history.

At Jiwe la Jahazi, local fishers tell of an attacking vessel turned to stone. The reef also holds distributions of basalt, sandstone and potsherds that are unusual in the local limestone environment. The story preserves a way of understanding these dangerous waters; archaeology records how materials lie on the reef. The two can be read alongside each other, but the story cannot date the stones, and the stones cannot prove every part of the story.

Researchers argue that the distribution of basalt and pottery better supports a vessel wrecking on the reef or unloading ballast in an attempt to escape. With no surviving hull, this remains an interpretation rather than an eyewitness fact. The sandstone may belong to a separate grounding, and the geological source of the basalt has not been fully established. The value of maritime archaeology lies partly in these limits: it brings a possible accident at sea into view while keeping other possibilities in the record.

At another scatter inside the harbour, local and imported pottery, animal bone and stone were found around a stone anchor; the main assemblage dates to roughly the thirteenth through sixteenth centuries. Material across the wider anchorage spans a longer period, showing repeated use of the same waters. The survey left most finds in place for preservation. Without further excavation or a ship's hull, researchers still cannot identify the scatter as the wreck of one particular vessel.

This evidence is enough to connect the harbour waters with voyages across the sea and to show that such connections involved danger and repeated use. It is not enough to prove that Kilwa controlled or monopolised Indian Ocean trade in every period. The port's significance rests on visible long-term use. Its power and returns in different centuries must be checked separately against coins, documents, buildings and evidence from other places.

UNESCO's heritage summary places occupation and architectural remains on the island between the ninth and nineteenth centuries, so the material chronology does not end neatly in 1500. This span is itself a reminder that political change, the use of objects, the repair of buildings and the abandonment of a place do not share one timetable. For a city, an important year can alter power without bringing everyone's life to a halt at the same moment.

The sixteenth-century Portuguese fort is a visible political marker, and an earlier chronology once described the period after 1500 in terms of “decline.” That word cannot explain what happened next on its own. Buildings could remain occupied or be altered, material practices might continue, and maritime routes and political rule could change at different speeds. Compressing these processes into a story in which the city vanished after conquest would once again make local history move only when outsiders arrived.

An archive preserved by the British Institute in Eastern Africa and the Archaeology Data Service catalogues a group of records as “Kilwa Kisiwani, Makutani 18th Century Building.” At minimum, this places an eighteenth-century building object in a material chain that can be revisited. The catalogue title alone cannot tell us who built it, who lived there or how it was used. Its role here is to stop the narrative from turning into a blank immediately after the sixteenth-century fort.

The ruins today stand on a living island. A study of twenty-two local residents records that people valued the cultural meaning of the site while also viewing tourism, employment, income, roads and public services as practical relationships that heritage should support. This sample cannot represent every resident, still less be projected back into the medieval period. It does show that heritage does not belong only to distant visitors and managing institutions.

The same study explains some damaging actions in the context of limited benefits, unstable employment and payments, and a lack of viable livelihoods. This does not remove individual responsibility. It refuses, however, to explain failures of preservation only by saying that residents “do not understand heritage.” When preservation restrictions and everyday livelihoods meet in the same space, the way a protection system distributes benefits and constraints also becomes part of the city's contemporary history.

UNESCO's current assessment likewise lists development and environmental pressures, together with insufficient community participation. An institution's judgement cannot substitute for residents speaking for themselves, but it does state that protection requires community involvement. Kilwa's history therefore did not stop when its ruins were formed. Who can interpret and maintain them, live from them, and decide which material will remain is creating a new layer now.

These four chapters still leave many gaps. We cannot yet write about medieval households, gender, enslaved and free workers, everyday faith or language use with the same precision as the building sequence. Something else can already be established: Kilwa grew through local residence and building, became a port through repeated use of the coast, and continued to be inhabited, remembered and managed after changes of rule. The city did not disappear because parts of its story are still missing.