Kilwa Sultanate

Where the wind stops

Drafted from Middleton, The Peoples of Africa (2000)formation

The Indian Ocean has a schedule. The north-east monsoon blows from November to March and carries ships from Arabia, Persia and India down the East African coast; the south-west monsoon blows from April to September and carries them home. A merchant who leaves Muscat in December must be back on the wind by April, and the distance he can reach and return from inside one year is the whole geography of what follows.

That distance ends at about Kilwa.

South of Kilwa is Sofala, sixteen hundred kilometres further down, where the gold from the Zimbabwe plateau reaches the sea. A ship from the north could not get to Sofala and back on one year's winds. So the last leg was done by coastal vessels, and whoever controlled the point where the two systems met controlled the gold.

What that produced

Not an empire. A customs house with a fleet of small boats and a sultan.

Kilwa mined no gold, produced no ivory and grew almost nothing on a small coral island. What it sold was the right to pass, and it enforced that right by owning the vessels that made the southern run.

There is a whole class of states that lived this way — Palmyra on the desert crossing, Petra on the incense road, Derbend on the pass between the mountains and the sea, Srivijaya on the straits. Kilwa's version is the purest of them, because the barrier it charged for was neither a mountain nor a strait. Derbend held a gap three kilometres wide. Kilwa held a date in the calendar.

What the town was

Coral rag and lime mortar, on an island a few hundred metres off the mainland: a mosque, a hundred-room palace on the cliff above the harbour, a mint, and houses of stone on a coast where most people lived in mud and thatch — and it mattered enormously which of the two a family lived in.

Chinese porcelain in the walls

The excavated material is the argument.

Chinese celadon and blue-and-white, Islamic glazed wares from the Gulf, glass beads from India, and enough Song and Yuan porcelain that some of it was set decoratively into the plaster of tombs and mosque niches, where it still is. Kilwan coins go the other way, turning up in the Comoros, in Oman and at Great Zimbabwe.

Zheng He's fleets reached this coast in the 1410s and 1420s, and took a giraffe home to the Ming court. An island off Tanzania and a dynasty at Beijing were, for about fifteen years, looking at each other across the same water — and nothing passed between them but goods, which is why neither remembers it as a relationship.

Sofala

Drafted from Middleton, The Peoples of Africa (2000)expansion

Sofala is a river mouth on the coast of what is now Mozambique, about sixteen hundred kilometres south of Kilwa, and for four centuries it was where the gold of the interior reached salt water.

The gold came off the Zimbabwe plateau, four hundred miles inland, where it was washed out of ponds and river mud by men working in gangs and paid for in advance with Gujarati cloth. Between that plateau and this island there was never an ambassador, a treaty or a word of writing — only metal, cloth and pots.

How a monopoly on a coastline works

Kilwa did not conquer Sofala in any sense that would show on a map. What it did was own the leg.

The mechanism is meteorological. A ship out of Arabia, Persia or India rides the north-east monsoon down the coast from November and must be back on the south-west monsoon by September. Kilwa is inside that year. Sofala is not. So a merchant from Aden who wanted Zimbabwean gold could either winter twice on a foreign coast or buy it from somebody at Kilwa who had already fetched it.

Kilwa made sure there was always somebody, and that the somebody was Kilwan. The southern run was made in smaller vessels by Kilwan crews on Kilwan account, and the town's position was not a garrison but a schedule.

That arrangement is the difference between Kilwa and the ordinary toll-taking city. A town on a pass has a position and nothing else, and can be bypassed the moment somebody finds another pass. Kilwa had a position and the boats.

What went the other way

Cloth, mostly. Indian cotton, in enormous quantity, was what the interior wanted and what gold bought; beads by the tens of thousands, which turn up in the Zimbabwe excavations; and glazed ceramics.

The exchange rates were reportedly favourable enough at the inland end that a sixteenth-century Portuguese account calls the trade an exchange of cloth for gold at rates the Portuguese could not initially believe. That is the kind of claim a new arrival makes about a market he does not understand, and it should be read as evidence about Portuguese expectations rather than about African prices.

Ivory, and the other cargo

Two more commodities, and the second of them is usually left out.

Ivory went north in quantity — African elephant tusks were softer and easier to carve than Indian ones and were preferred in India for that reason, which is a neat inversion of what a reader expects the trade in worked goods to look like.

And people. The East African coast supplied the Indian Ocean slave trade for something over a thousand years, north to Arabia and the Gulf and east to India, and the Zanj rebellion that convulsed southern Iraq in the ninth century — one of the largest slave revolts in recorded history — was made by East Africans shipped from this coast to drain salt marshes.

Kilwa taxed and supplied that traffic rather than running it, which is why it tends to disappear when the town's revenues are summarised. It should be said anyway.

What Sofala was worth

Enough that when the Portuguese arrived they went for it directly.

Almeida sacked Kilwa in 1505 and the fort was gone within a few years. In 1506 Portugal occupied Sofala and began buying at the source, and that — not the sack — is what ended the arrangement. A carrack coming round the Cape did not need the monsoon calendar and did not need to stop.

O Berbera and thy enchanted waves

Drafted from al-Masudi, El-Mas'udi's Historical Encyclopaedia, entitled 'Meadows of Gold and Mines of Gems', Volume I (1841)expansion

In 916 a Baghdadi scholar sailed north from an island off the East African coast to Oman in a ship owned by two brothers from Siraf on the Persian Gulf. He wrote it up decades later in a book about everything, and the passage is the earliest first-hand account of the sea route the Swahili towns were built on.

Al-Mas'udi is not describing a voyage of discovery. He is describing a commute.

The run

The island he sailed from he calls Kanbalu, a well-cultivated place whose people were Muslims but spoke the Zanj language — meaning a coast where the religion had arrived and the language had not changed, which is what the Swahili towns would be for the next thousand years.

He names his shipowners: Ahmad and Abd al-Samad, brothers of Abd al-Rahim ibn Ja'far of Siraf. And then the detail that tells you what kind of route this was:

I passed then the spot where, subsequently, the vessel of Ahmad and Abd es-Samad was wrecked, and where these two men perished with their whole crew.

He mentions another Sirafi shipowner, Muhammad ibn Zindibud, who also went down in that sea with his whole crew. Two named owners drowned on the same run, and the man recording it had sailed with one of them.

That is the Indian Ocean as it was actually worked: a predictable wind, a regular schedule, and a death rate high enough that a traveller could list the owners he had sailed with who did not come back.

The song

The sailors were Arabs of the tribe of Azd, and Mas'udi records what they sang while working the ship in heavy water — lifted up by the waves and then sinking between them:

O Berbera and Jofuni and thy enchanted waves. Jofuni, and Berbera, and their waves, as thou seest them.

It is a work chant naming two stretches of the African coast, and it is one of the oldest fragments of anything anybody sang on that ocean.

Where it reached

Mas'udi is precise about the far end. The sailors of Siraf and Oman went down this sea as far as Kanbalu and the sufala — the low country — of the Zanj, at the extremity of their country.

Sufala is a common noun in Arabic for a low-lying coast, and Mas'udi uses it that way, of a whole stretch of shoreline. Some centuries later it had become a proper name attached to one port, the outlet for the gold of the plateau inland, and the town whose trade the Kilwa sultans took over and lived on.

The word arrives in the record as a description of a coastline that Omani and Persian shipowners had been reaching for at least two hundred years before anybody in Kilwa built anything in stone.

What the passage is good for

Not for Kilwa, which did not exist yet, and not for the gold, which Mas'udi does not mention.

It is good for establishing that the system was already old. The monsoon schedule, the Sirafi and Omani shipowners, the Muslim island communities speaking an African language, the run to the far southern coast and back — all of it is in place in the tenth century, in the first-person account of a man who did the voyage and could name the drowned.

Everything the Swahili towns later were is a local elaboration on a trade that was already running past them.

The prince from Shiraz

Drafted from Middleton, The Peoples of Africa (2000)peak

The Kilwa Chronicle, written down in the sixteenth century, says the dynasty was founded by Ali ibn al-Hasan, a son of the sultan of Shiraz, who left Persia with his six brothers in seven ships and bought the island from its African ruler for enough cloth to cover it.

Dynasties from Mogadishu to the Comoros told versions of the same story. It is the Shirazi tradition, and it is almost certainly not what happened.

What the ground says instead

Horton and Middleton — an archaeologist writing with an anthropologist — put the case from the excavation. The Swahili coast's towns grow out of local Iron Age settlement, in continuous occupation, with material culture that develops in place. The language is Bantu with Arabic loanwords rather than the reverse. The stone architecture, when it appears, is a coral-rag tradition with no clear Persian antecedents. There is no colonising population in the ground.

What there is, is a coastal African society that became Muslim, became wealthy on the ocean trade, and acquired the vocabulary and the pedigree of the world it was trading with.

And the sultans went on claiming it

A false pedigree is still a pedigree, and it does the same work. The question worth asking about the Shirazi story is not whether it happened but what it was for.

What it bought was commercial standing. In an Indian Ocean where credit ran on reputation and reputation ran on descent, being the son of a Persian prince was worth having whether or not you were. A merchant in Aden deciding whether to advance goods against a cargo that would not return for a year wanted to know who he was dealing with, and "of the house of Shiraz" was an answer his own world understood.

Ethiopia's kings did the same thing with Solomon and the queen of Sheba, and for the same reason: a dynasty on the edge of a literate world acquires the genealogy that world recognises. The Portuguese, when they turned up, took the Shirazi story at face value, and de Barros printed it in 1552 from a chronicle of the kings of the city that he does not say how he came by.

Ibn Battuta

He came in 1331 and liked it, which for him was not guaranteed.

He calls Kilwa one of the most beautiful and well-constructed towns in the world, entirely built of wood — meaning, probably, the ordinary houses rather than the stone ones he would also have seen. He describes the sultan, al-Hasan ibn Sulaiman, as devoted to raiding the mainland for slaves and to giving the proceeds away in charity, and tells a story of him handing his own clothes to a petitioner.

Twenty years later the same traveller was at Mali, complaining about unveiled women and matrilineal inheritance. He is the best witness anybody has for fourteenth-century Africa, and one of very few who saw both the Indian Ocean coast and the Niger and can be read against himself.

The raiding

Ibn Battuta's own sentence has it, and it should not be read past. The sultan he admired for his charity was funding that charity with people taken off the mainland.

The East African coast supplied the Indian Ocean slave trade north to Arabia and the Gulf and east to India for something over a thousand years, before the Atlantic trade began and long after it ended. Kilwa's part in it was as a market and a tax point rather than as a raiding power, which is a distinction worth making and not a defence. The gold and the porcelain and the hundred-room palace are the visible half of what the ocean carried; this is the other half, and it moved in the same boats.

They seldom give gold

Drafted from Battuta, The Travels of Ibn Batuta (1829)peak

One traveller wrote down what Kilwa looked like at its height, and almost everything he says is at odds with what has been dug up there.

Ibn Battuta came down the coast in 1331, in the reign of the sultan who built the palace on the cliff. He gives the town four sentences.

The four sentences

He reached it from Mombasa, an island abounding in bananas, lemons and citrons, with no grain at all — everything brought in from elsewhere — and people he found religious, chaste and honest.

Then:

I set out, by sea, for the city of Kulwa, which is large, and consists of wooden houses. The greater part of the inhabitants are Zunuj of the sect of Shafi'i, of religious and peaceful habits. The king of this place, at the time I entered it, was Abu'l-Muzaffar Hasan, a person who had obtained great victories over the countries of the infidel Zunuj. He gave much away in alms. The greatest gift bestowed by the people of these countries is ivory, which is the elephant's tooth: they seldom give gold.

That is the whole of it in this translation. Then he went on to Zafar in Yemen.

What is wrong with it

Two things, and both are worth taking seriously rather than explaining away.

Wooden houses. Kilwa in 1331 had a congregational mosque of coral stone with domed and vaulted bays, and on the cliff above the harbour the hundred-room complex of Husuni Kubwa — a palace and warehouse in one, with an octagonal bathing pool cut into the coral — under construction in the reign of the very sultan Ibn Battuta names. He stayed in the town. He does not mention any of it.

They seldom give gold. Kilwa's whole position rested on controlling the outlet for the gold of the Zimbabwe plateau through Sofala. The sultan Ibn Battuta was looking at was rich on it.

How to read that

The likeliest answer to the first is that both are true. A stone town in the Swahili world meant a stone core — mosque, palace, the houses of the ruling families — surrounded by a much larger settlement of timber, mud and thatch. A visitor describing the place as consisting of wooden houses is describing most of the buildings, correctly, and passing over the few that were not.

The answer to the second is about what a guest sees. Ibn Battuta is reporting what he was given, and what visiting scholars were given on that coast was ivory. Gold moving from Sofala to the Red Sea in a sultan's account books is not a thing a guest is handed at the door.

And Lee's translation is an abridgement made in 1829 from incomplete manuscripts. Fuller versions of the Rihla give Kilwa more space and praise the town's construction. The passage as it stands here is short partly because a translator two centuries ago was working with less than the whole.

Why it is worth having anyway

Because it is the only eyewitness description of the town at its height by someone who was not Portuguese, and because of what it does record.

A sultan whose reputation with a visiting Moroccan jurist rested on holy war against the non-Muslim mainland and on the scale of his almsgiving. A population of African Muslims following the same legal school as the visitor, which is why he could be lodged and fed and consider the place unremarkable. An island to the north living entirely on imported grain. And a gift economy in which the prestigious thing to hand a distinguished stranger was a tusk.

None of that is in the archaeology, and none of the archaeology is in Ibn Battuta. The record of this town is two bodies of evidence that barely overlap, and the honest position is to hold both.

Husuni Kubwa

Drafted from Middleton, The Peoples of Africa (2000)peak

On the cliff above the harbour at Kilwa there is a building with something over a hundred rooms. It is called Husuni Kubwa — the great fort, though it was not one — and it was built in the early fourteenth century, in the reign of the sultan Ibn Battuta came to see.

It is a palace and a warehouse in the same structure. On the seaward side, a domed audience court, a set of residential rooms, and an octagonal bathing pool cut into the coral. On the landward side, a large courtyard surrounded by storerooms, entered through a gate wide enough for cargo.

It was never finished, and it seems to have been substantially abandoned within a generation of being built.

Coral

The material is the thing worth noticing, because it explains what a Swahili stone town was and why so few people lived in one.

The walls are coral rag — reef limestone quarried in blocks — set in lime mortar made by burning more coral. The decorative work is porites, a fine-grained living coral cut from the reef while soft and carved before it hardens, which is how the inscription bands and the panelled niches were made.

Building in stone on this coast meant destroying reef, burning it for lime, and employing people to do both. It was expensive, and the archaeology across the whole coast shows the same pattern: a minority of stone houses among a majority of mud and thatch, and the stone-house families are the trading families.

Horton and Middleton's argument is that the stone town is not a colony or a foreign quarter but the top of a local society — the same conclusion the Shirazi pedigree points to from the other direction.

The Great Mosque

The other building, and the older one.

The prayer hall was first built in the eleventh century and enlarged twice. The southern extension of the fourteenth century roofed a hall of sixteen bays with domes and barrel vaults on octagonal coral columns, and a good part of it is standing.

It is the earliest surviving multi-domed mosque in sub-Saharan Africa and one of the largest mosques anywhere on the East African coast, on an island a few hundred metres wide.

The money

Kilwa minted its own copper, in quantity, from about the eleventh century, and the coins are the furthest-travelled thing the town ever made.

They carry the sultan's name in Arabic on one face and a rhyming epithet on the other — al-Hasan ibn Sulaiman, and then something like trusts in the master of bounties. There is no denomination and no date. They are small change, struck for local circulation rather than for the ocean trade, which ran on gold by weight and on credit.

They turn up along the coast, in the Comoros, in Oman, and at Great Zimbabwe four hundred miles inland, where a coin of a thirteenth-century Kilwan sultan came out of the excavation and is one of the few hard links between the two ends of the gold trade. Five more were found in 1945 by an airman on a beach in the Wessel Islands off northern Australia. What that means is argued about, and the honest answer is that nobody knows.

And what none of it is

Kilwa founded no dynasty anywhere else, carried its religion nowhere, and passed its script to nobody — it had borrowed the script itself. A town that built the largest mosque on its coast, minted coin for five hundred years and put up a hundred-room palace exported none of the things by which influence is usually reckoned.

What it exported was gold it did not mine and cloth it did not weave. Venice is the same shape at the other end of the world: enormously consequential, impossible to score, because what it left behind was a way of doing business rather than a religion or an alphabet.

The ocean

Drafted from Middleton, The Peoples of Africa (2000)aside?What aside means

The Indian Ocean is the only ocean with a predictable wind that reverses, and almost everything about the world Kilwa belonged to follows from that.

The south-west monsoon blows from April to September and the north-east from November to March, and both are reliable enough that a sailing schedule could be built on them and was, for at least two thousand years. The Periplus of the Erythraean Sea, a Greek merchant's handbook of the first century, already describes the pattern and the ports.

What that produced is a trading system that worked without anybody governing it.

Nobody was in charge

This is the part that is hard to hold in mind beside the more familiar sea.

The Roman Mediterranean was a single tax area with one currency and no piracy, because one state held the whole rim. The Indian Ocean was never held by anybody. Between the eighth century and the fifteenth it carried a larger volume of long-distance trade than the Mediterranean, across four thousand kilometres, among Arabs, Persians, Gujaratis, Tamils, Malays, Chinese and Swahili, with no hegemon, no treaty system, no shared coinage and almost no naval warfare.

What held it together was Islam, credit and the calendar. A merchant from Aden and one from Kilwa shared a legal vocabulary, a set of commercial instruments and a set of dates. A dispute could be taken to a qadi in either place under recognisably the same law.

Srivijaya, on the straits between Sumatra and Malaya, was doing the same job at the eastern end of the same system and living off it in the same way.

The porcelain

Chinese ceramics are the index, because they are datable and indestructible.

The Kilwa excavations produce Yue ware, then Longquan celadon, then Jingdezhen blue-and-white, in a sequence that tracks Chinese production closely enough to date deposits by. Some of it was set into the plaster of tombs and mosque niches as decoration, where a good deal of it still is.

It reached the coast almost entirely through intermediaries. Chinese ships in the western Indian Ocean are rare before the fifteenth century and the porcelain came by way of the Gulf and India.

And then, once, the fleet

Zheng He's expeditions reached this coast in the 1410s and 1420s, and brought back tribute and a giraffe.

The giraffe came from Malindi, four hundred kilometres north of here. It was presented at the Ming court as a qilin, an auspicious beast whose appearance signals a virtuous reign, and there is a painting.

The fleets were enormous — hundreds of ships, tens of thousands of men — and they came four or five times and stopped.

From China the decision to stop is one of the great might-have-beens. From this end the visits look like an interruption rather than a turning point: no Chinese quarter was established, no Chinese trading houses stayed, and the porcelain went on arriving by the route it always had, through the Gulf and India. Both readings are accurate, which is what makes the episode worth keeping.

What the coast was, in the end

Not an empire and not a colony. A chain of independent towns — Mogadishu, Barawa, Lamu, Pate, Malindi, Mombasa, Kilwa, the Comoros, Sofala — sharing a language, a religion, an architecture and a trade, quarrelling constantly, and never once unified by any of them.

Kilwa was the largest and richest of them, which is why its name stands for the whole, and the substitution is a distortion. The Phoenician cities cause the same trouble two and a half thousand years earlier: a civilisation of ports gets told through whichever port was momentarily on top, and the thing that was actually remarkable — that thirty independent towns could share a language, a law and a trade without any of them ever ruling the others — disappears into the story of one of them.

Two thousand mithqals

Drafted from Barbosa, The Book of Duarte Barbosa (1918)contraction

The sultanate was already in trouble before the Portuguese fleet appeared, and the trouble was of the ordinary kind: somebody had taken the throne who had no business on it.

A usurper on the throne

When Vasco da Gama first came down the coast in 1498 he was told at Mozambique, by a pilot who had worked out what he wanted to hear, that there was a Christian population at Kilwa. He could not land and discovered the deception at Mombasa.

Pedro Álvares Cabral called in 1500, and da Gama came back on his second voyage in 1502. Both were received with barely concealed hostility by the man then in power — Mir Ibrahim in the Portuguese spelling, whom Dames, editing Barbosa, calls flatly a usurper.

Da Gama extracted a tribute of two thousand mithqals of gold, and from that moment Kilwa was held in Lisbon to be tributary to the king of Portugal.

That is three years before Almeida, and it is the whole shape of the collapse. A city with no army, whose ruler had no claim his own people accepted, agreed to pay an annual sum to a power that had visited twice and could not be reached.

What was still there to take

Barbosa, writing about 1518 from what he had been told and partly seen, describes what the town had been in the years just before, and his description is the best one there is.

a Moorish town with many fair houses of stone and mortar, with many windows after our fashion, very well arranged in streets, with many flat roofs. The doors are of wood, well carved, with excellent joinery. Around it are streams and orchards and fruit-gardens with many channels of sweet water.

The people, he says, some fair and some black, finely clad in rich garments of gold and silk and cotton, the women likewise, with gold and silver chains and bracelets on their legs and arms and jewelled earrings.

And then the sentence that confirms, from the other side and by a hostile witness, exactly what the sultanate lived on:

Before the King our Lord sent out his expedition to discover India the Moors of Sofala, Cuama, Angoya and Mozambique were all subject to the King of Quiloa, who was the most mighty king among them. And in this town was great plenty of gold, as no ships passed towards Sofala without first coming to this island.

A Portuguese factor with every reason to belittle the place records that before his own countrymen arrived, every ship going for the gold had to stop here first.

Ibn Battuta's town, and Barbosa's

There is a discrepancy between the two outside descriptions and it is worth stating rather than smoothing.

Ibn Battuta, who came in 1331, calls Kilwa a large city that consists of wooden houses, its people Zunuj of the Shafi'i school, of religious and peaceful habits. Barbosa, two centuries later, describes a stone town with glazed windows and carved joinery. Dames concluded the town must have developed rapidly in between.

The likelier reading, and the one the excavation supports, is that both are describing the same place from different heights: the stone houses were always a minority among a majority of mud and thatch, and a traveller notices what most of the town is made of while a factor notices what the merchants live in.

Which also means that when the Portuguese sacked Kilwa they were sacking the minority. The families in the stone houses were the ones with the gold, the warehouses and the boats, and they were the ones with something to lose.

1505

Drafted from Middleton, The Peoples of Africa (2000)end

Francisco de Almeida arrived in July 1505 with a fleet and about five hundred men. The sultan did not pay the tribute demanded; the Portuguese landed, took the town in a morning against no organised resistance, sacked it, built a fort, and installed a man of their own choosing.

They abandoned the fort within a few years — de Barros says 1507, other accounts later — and Kilwa never mattered again.

A city with no army

There is no Kilwan army in the sources, and that is a finding rather than a gap. No standing force, no military titles in the chronicle, no campaign against another coastal town, and — until the Portuguese built one — no fortification facing the sea. The excavated town has a palace, a mosque, warehouses and stone houses, and nothing designed to be defended.

For five hundred years it did not need one. Kilwa's position rested on the monsoon and on owning the boats that made the southern run, and neither of those could be taken by an enemy who arrived by sea, because an enemy who arrived by sea still needed somewhere to stop.

Then somebody arrived who did not.

What the Portuguese actually took

Not the town, in the end — they left it. What they took was the thing the town's whole position depended on.

In 1506 they occupied Sofala directly and began buying the gold at source. The transshipment monopoly Kilwa had lived on simply stopped existing, because a Portuguese carrack could round the Cape, run up to Sofala, load, and go home without stopping anywhere, and the monsoon calendar that had made Kilwa necessary did not apply to a ship coming the other way.

That is the mechanism, and it is worth stating precisely because the usual version — Portuguese guns destroy Swahili civilisation — is both too dramatic and not the point. The coast survived. Mombasa, Malindi, Pate and Lamu went on; the Omanis took Mombasa back from the Portuguese in 1698, and Zanzibar became the seat of an Omani sultanate in the nineteenth century.

What ended in 1505 and 1506 was a particular chokepoint on a particular route.

A new thing

Drafted from Barbosa, The Book of Duarte Barbosa (1918)afterlife

The Portuguese held the fort at Kilwa a few years and then pulled it down — de Barros dates the demolition to 1507 and other accounts put it later, and the disagreement matters less than the reason, which Barbosa gives in the flattest possible terms.

Afterwards he ordered that it should be pulled down, as its maintenance was of no value nor profit to him, and it was destroyed by Antonio de Saldanha.

Dames adds the judgement that the Portuguese occupation was not a success, and that the formerly great trade was ruined by their efforts to regulate it.

Both halves of that are the afterlife. Kilwa's position was destroyed, and the people who destroyed it got nothing they wanted out of the ruins.

What Mombasa got

The same treatment, a few weeks apart, and Barbosa's account of it is worth setting down because it is the fullest inventory of what a Swahili town contained.

Mombasa's king also refused to obey, "and through this arrogance he lost it". The Portuguese took it by force, killed many of his people, took many men and women captive, and left the town ruined and plundered and burnt. The booty was gold and silver — bangles, bracelets, earrings, gold beads — and a great store of copper, with other rich wares in quantity.

Barbosa records all of it in the register of a man noting a satisfactory outcome. He is not describing a catastrophe; he is describing a return.

The remedy at Sofala

And then, a few pages earlier in the same book, there is a paragraph that shows what the coast did about it, and it is the most interesting thing in Barbosa.

At Sofala, he says, they have lately begun to grow cotton in quantity and to weave it, and they make much white cloth from it. They cannot dye it — they have not got the dyes — so they buy Gujarati cloth, blue or otherwise coloured, unravel it, and weave the coloured thread up again with their own white.

so that it becomes a new thing. With this thread and their own white they make much coloured cloth, and from it they gain much gold.

Then Barbosa says why:

This they did as a remedy after they had perceived that our people were taking from them the trade of the zambucos, and that they can only obtain goods through the hands of the factors whom the King our Lord has there in his factories and forts.

An import monopoly was imposed. Within a few years the people it was imposed on had started manufacturing the import, by taking the monopolist's own product to pieces and rebuilding it into something the interior would pay gold for.

That is not the response of a destroyed civilisation. It is the response of a commercial society doing what commercial societies do when a supplier turns predatory, and it happened inside a decade.

What actually ended

Not the coast. Mombasa, Malindi, Pate and Lamu went on; the Omanis took Mombasa back in 1698 and ran the coast from Zanzibar in the nineteenth century; the language, the law and the architecture continued, and the towns are still there.

Not the trade. Gold went on leaving Sofala and cloth went on arriving; only the hands changed, and then changed again.

What ended was one arrangement: the monsoon chokepoint, and a town that had lived on it for five hundred years without ever needing a wall facing the sea.

The last object

The most widely travelled thing Kilwa ever made was its small change.

The copper coins — the sultan's name on one face, a rhyming epithet on the other, no denomination, no date — turn up along the coast, in the Comoros, in Oman, at Great Zimbabwe three thousand kilometres inland, and in 1945 on a beach in the Wessel Islands off northern Australia, where an airman found five of them. What that last find means is argued about, and the honest answer is that nobody knows.

A town that exported no religion, no script and no dynasty exported its coppers further than any of them would have gone, and mostly by accident, in other people's pockets.