Mali took Arabic literacy — and a law of who could be enslaved (1324)

Before Islam: cities without kings on the Middle Niger

Around 250 BCE, on a low clay mound in the floodplain where the Bani joins the Niger, people began living year-round at a place archaeologists call Jenne-jeno. By the middle of the first millennium CE it covered some thirty-three hectares and stood at the centre of a cluster of roughly sixty-nine satellite settlements within a four-kilometre radius, each apparently specialising — iron-smelting on one mound, fishing on another, pottery, rice, cattle ¹. This was a city. It was not a capital.

What the excavations of Susan Keech McIntosh and Roderick McIntosh recovered from 1977 onwards was an urban form for which the standard Near Eastern and European models had no slot. There was no palace at Jenne-jeno. No citadel, no acropolis, no monumental temple, no royal tomb, no granary complex under central guard, no archive ¹². Roderick McIntosh's term for what the cluster represents is a self-organising landscape: a dense, occupationally differentiated urban population held together by reciprocal specialisation rather than by a ruler who could compel it. Africanist archaeology now uses Jenne-jeno as the type site for heterarchy — authority distributed laterally across many groups instead of stacked into a single hierarchy ².

That is the ground Islam would eventually arrive on, and it matters that the ground was already built. The Middle Niger in 800 CE was one of the more thoroughly urbanised river valleys in the world. Its people had imported copper from the Sahara and stone from hundreds of kilometres upriver for centuries. They had grown African rice and pearl millet for a millennium; they had smelted iron since before the common era; they had developed a floodplain economy in which fishers, herders, farmers and smiths lived in adjacent settlements and exchanged daily. What they had not produced, and would not produce on their own, was writing.

That absence is the hinge of this entire record, and it should be stated without condescension. Non-literate does not mean without records. The Mande world had extremely sophisticated instruments for storing and transmitting information across generations — they simply ran on human memory, trained from childhood and professionally maintained. What such a system cannot do is store information that outlasts the community holding it, or move information to a stranger who has no relationship with that community. Those two limits are exactly what Arabic writing removed, and the removal is the transmission's largest single consequence.

The gold was already moving

Long before the first Mande king professed Islam, the Middle Niger was inside the Islamic world's economy. West African placer gold — washed from the alluvium of Bambuk, between the Senegal and the Falémé, and from Buré on the upper Niger — was reaching North African and Egyptian mints from at least the ninth century. The Sijilmāsa and Ifrīqiyan dinars of the tenth and eleventh centuries were struck substantially on southern metal ⁷.

The archaeology at Essouk-Tadmekka makes this concrete. Sam Nixon's team recovered crucible fragments and moulds for casting blank discs of very high-purity gold — a West African town, in the ninth to eleventh centuries, standardising bullion into a form the Mediterranean monetary system could take directly ⁸. The gold trade was not something that happened to West Africa. It was something West Africans ran.

They ran it, moreover, on their own terms, and one persistent story about how deserves to be dismantled here. The "silent trade" — gold and salt exchanged by parties who never see or speak to one another, each leaving goods and withdrawing until the piles satisfy both — is reported from Herodotus through the medieval Arabic geographers to the Portuguese, and it is a travelling literary motif rather than an observed practice. P. F. de Moraes Farias's examination of the whole source chain found the accounts to be second-hand, mutually derivative and uncritically repeated across two thousand years ²⁶. What the trope actually encodes is something true and much less exotic: no Maghrebi merchant ever saw the mines. Wangara and Dyula middlemen stood between the goldfields and the caravan termini and kept the sources secret for six hundred years. That is not silence. That is commercial control, held by the people who had the gold.

What the Mande world already had

The Mande-speaking societies of the upper Niger organised themselves through instruments that a Cairo jurist would not have recognised as institutions at all, and which were nonetheless load-bearing:

• Descent groups and hereditary occupational categories — the nyamakalaw, the endogamous specialist estates of smiths, leatherworkers and bards, whose skills were understood as dangerous, inherited and non-transferable.
• Initiation associations that held knowledge in graded degrees and controlled access to it by age and ordeal rather than by literacy.
• The jeliw, the griots: hereditary professionals whose business was the past. Genealogy, precedent, praise and reproach were their property, and a ruler's legitimacy passed through their mouths ³.
• Sacral kingship, in which the ruler's body, his ancestors and the fertility of the land were a single circuit, maintained by sacrifice.
• A figurative terracotta tradition in the Inland Delta — the seated, kneeling and equestrian figures of the Jenne region, produced in quantity between roughly the eleventh and fifteenth centuries and associated with domestic and shrine contexts.

Photograph by Franko Khoury. Terracotta equestrian figure, Mali, 13th–15th century. National Museum of African Art, Smithsonian Institution. Public domain via Wikimedia Commons. · Public Domain

None of this was a religion in the sense the incoming jurists used the word. It had no book, no founder, no creed, no confessional boundary and no missionary ambition. It was, in the strict sense, a local practice — and locality was precisely the property Islam did not have.

The religion Islam met

The best single portrait of a West African polity on the eve of deep Islamization is al-Bakrī's, written in Córdoba in 1068 from merchants' reports. He describes the capital of Ghana, the Soninke state north-west of the Mande heartland, as two towns about ten kilometres apart with continuous habitation between them. One was the merchants' town, Muslim, with twelve mosques, one of them designated for Friday prayer, and a resident population of scholars, scribes and jurists. The other, al-Ghāba, was the king's: a stone-walled royal enclosure containing domed buildings, the king's audience court, and a sacred grove where the priests of the royal religion lived and were buried ⁴.

Two towns, ten kilometres, one state. That is not a transitional arrangement caught halfway to conversion. It is a stable settlement of a stable problem, and it lasted for centuries. The Muslim merchants got their law, their calendar and their Friday assembly; the king got credit, literate secretaries and diplomatic legibility; the royal cult got the grove. Al-Bakrī notes without apparent discomfort that Muslims held office in the administration — the treasurer, the interpreters — while the king himself did not pray with them ⁴. Timothy Insoll's survey of the archaeology of Islam in sub-Saharan Africa finds the same pattern in the material record across the Sahel: quarters, not conquests; two cemeteries, not one ⁵.

Al-Bakrī also preserves the moment the arrangement broke in one small polity. The king of Malal — a principality on the upper Niger that Nehemia Levtzion identified as an ancestor of the Mali state — suffered a drought. The customary sacrifices failed. A Muslim guest offered to pray, on condition that the king convert; the rain came; the king took Islam, and, in al-Bakrī's telling, destroyed the idols and expelled the sorcerers while his people remained as they were ⁴⁶. Whatever happened, the shape of the story is the shape of the whole transmission: it goes in at the top, it enters through a functional test, and for a long time it does not reach the cultivators at all.

The road from Sijilmāsa: how Islam actually crossed the Sahara

Merchants, not missionaries

Nothing about the Islamization of the western Sahel makes sense unless the camel road is put first. From the eighth century, caravans linked the Tafilalt oasis of Sijilmāsa, on the northern edge of the desert, to the market towns at its southern shore. Ibāḍī Berber traders opened the routes; Mālikī Sunni networks displaced them under and after the Almoravids. The freight in each direction was specific:

Southbound: rock salt from the Saharan pans, above all Taghāza; North African and European cloth; copper and brass; horses; dates; glass beads; paper; and books.

Northbound: gold from the Bambuk and Buré fields on the Senegal and Niger headwaters; ivory; hides; kola; and enslaved people.

The merchants were not evangelists and did not claim to be. But a caravan is a legal instrument before it is a line of camels. It runs on credit extended across a desert to a stranger who must be suable somewhere, in some law, before some judge whose ruling both parties recognise. Islamic commercial law supplied exactly that: a portable, written jurisdiction that worked identically in Fez, Sijilmāsa, Walāta and Gao ⁷. To trade at scale on this road was to be inside that law, and being inside it eventually meant being Muslim.

Tādmakka: an archaeology of the crossing

Until recently the desert crossing was known almost entirely from Arabic geography. Sam Nixon's excavations at Essouk-Tadmekka in northern Mali, published by Brill in 2017, put material evidence under the texts. Tādmakka was one of the principal southern termini from the ninth century onwards, and the dig recovered the physical apparatus of the gold trade: crucible fragments and moulds used to cast blank coin discs of very high-purity gold, alongside imported glass, Mediterranean ceramics and Arabic inscriptions on stone ⁸. This is the transmission's plumbing — a town where West African gold was standardised into a form the Mediterranean monetary system could absorb, and where Arabic writing was in daily use two and a half centuries before Mansa Mūsā was born.

Nixon's site also shows the trade's other commodity in its documentary shadow: the Arabic geographers who describe Tādmakka describe it as a source of gold, ivory and slaves in the same sentence ⁸. The two cargoes were never separable. The gold road and the slave road were one road.

The conquest that never was

The popular version of this story has a battle in it. In 1076, the telling goes, the Almoravids — the Ṣanhāja Berber movement that had swept out of the western Sahara and would shortly take al-Andalus — stormed the Ghana capital and imposed Islam by force on the western Sudan.

The Almoravids were certainly in the Sahel. Al-Bakrī, writing in 1068, describes their seizure of the caravan town of Awdaghust around 1054–55, an event he presents as a raid on a commercial rival that happened to be nominally under Ghana's authority ⁴. Beyond that, the conquest of Ghana dissolves under examination. David Conrad and Humphrey Fisher went through the evidence in two long articles in History in Africa in 1982 and 1983, taking the external Arabic sources first and the local oral traditions second. They found no source, of either kind, that unambiguously reports an Almoravid conquest of Ghana. The claim turns out to rest on a chain of late readings and modern inference rather than on any medieval statement; the Soninke oral traditions, for their part, attribute the state's disintegration to drought, not to invasion ²⁵.

This matters for the atlas more than a footnote's worth. If Islam entered West Africa at swordpoint, the story is the familiar one of imposed religion and the cost is easy to locate. It did not. The actual mechanism was slower, duller and far more consequential: three centuries of credit relationships in which the literate, legally organised party gradually made literacy and legal organisation the condition of doing business at scale — and rulers who wanted to do business at scale converted. Nothing in this record is a conquest narrative.

Two qualifications keep that honest. First, Sahelian Islamization was not always peaceful: the jihād states of the eighteenth and nineteenth centuries — Futa Jallon, Sokoto, the Massina caliphate — expanded Islam by war and enslaved those who resisted. But those are five hundred years downstream of Mansa Mūsā and belong to a different record. Second, the absence of conquest is not the absence of cost. The bill in this record is not paid in battle casualties. It is paid in a legal category.

The witnesses, and what each of them saw

Almost everything in the four centuries this record covers is known through a handful of Arabic texts. It is worth setting them out plainly, because the shape of the evidence is itself part of the story.

These texts reach modern readers through two great scholarly editions: the English Corpus of Nehemia Levtzion and J. F. P. Hopkins, and Joseph Cuoq's French Recueil des sources arabes, which gathers extracts from seventy-two Arabic authors on the bilād al-sūdān ²³. Anything written about medieval Mali in the last fifty years passes through one or both.

Note the last row and its date. For six centuries the western Sudan is described by outsiders — a Córdoban who never crossed the desert, a Damascene civil servant in Cairo, a Tangier jurist on his way home. The first substantial history of West Africa written by a West African appears in the middle of the seventeenth century. That it exists at all is a direct product of the transmission this record describes; that it took six hundred years is a measure of how slowly the technology spread below the courts and the merchant towns.

The Mali state assembles itself

Mande oral tradition dates the founding of the empire to the victory of Sunjata Keita over Sumanguru at Kirina, conventionally placed around 1235. The epic is a magnificent document and a poor chronicle: it exists in dozens of performed variants, was first published in a synthetic French prose version by Djibril Tamsir Niane in 1960 from the recitation of the jeli Mamadou Kouyaté, and is not evidence for events in the way a dated inscription is ⁹¹⁰. What can be said from the Arabic sources and the archaeology is that by the last quarter of the thirteenth century a Mande-led state controlled the goldfields, the Niger bend and the Saharan termini, and that its rulers were Muslim.

The archaeology is thinner than the fame of the empire suggests. The site conventionally identified as the capital, Niani on the Sankarani, was excavated by a Polish-Guinean mission in the 1960s and 1970s, and Sirio Canós-Donnay's survey of the field notes that the imperial-period identification has been questioned on stratigraphic grounds ¹¹. The atlas records this honestly: we know a great deal about medieval Mali's diplomacy and almost nothing certain about where its kings slept.

Mansa Mūsā's hajj, 1324

Then comes the event that made Mali visible from Damascus to Majorca. Mansa Mūsā, who had ruled since about 1312, left for Mecca in 1324, crossed Egypt, and spent in Cairo on a scale the city was still discussing more than a decade later.

The near-contemporary witness is al-ʿUmarī, a Damascene official who was in Cairo in the 1330s and compiled the testimony of people who had dealt with the Malians. He is careful, specific and, on the economics, unambiguous. Of Mūsā's spending he writes that "this man flooded Cairo with his benefactions," and of the consequence: "They exchanged gold until they depressed its value in Egypt and caused its price to fall." On the currency itself he is more precise still — the mithqāl, he says, "did not go below 25 dirhams and was generally above, but from that time its value fell and it cheapened in price and has remained cheap till now" ¹².

That last clause is the interesting one, because it is dated. Al-ʿUmarī is writing around 1337–38 and reporting that a price movement caused in 1324 had not corrected in roughly twelve years. Modern economic historians have questioned how much of Cairo's gold-price behaviour can actually be laid at Mūsā's door, and the very large caravan figures that circulate — tens of thousands of porters, twelve thousand enslaved attendants — come from later chronicles rather than from near-contemporary sources, and should be treated as literary ¹¹¹³. What survives scrutiny is smaller and more durable: a West African ruler moved enough bullion into the Mamluk capital to be remembered by the treasury, and the memory travelled.

What Mūsā brought home is the part that mattered. Not a devotion — the mansas had been Muslim for generations — but an apparatus:

1. Mālikī jurists, imported to staff a legal system the state had previously had to borrow from resident merchants.
2. An Arabic library, purchased in Cairo and Mecca — the seed of the manuscript culture that would make Timbuktu a byword.
3. Congregational mosques at Timbuktu and Gao, built under royal patronage after the return.
4. Abū Isḥāq al-Sāḥilī, an Andalusi poet from Granada, engaged at very high cost, who is documented as having worked on an audience chamber at the capital.
5. A diplomatic identity: recognition, from 1324, as a Muslim sovereign by the Mamluk chancery, and correspondence with the Marinids of Fez.

One thing he did not bring back is the thing he is most often credited with. The claim that al-Sāḥilī introduced Sudano-Sahelian mud-brick architecture to West Africa originates with the French colonial administrator-scholar Maurice Delafosse and was demolished by Suzan B. Aradeon in the Journal des Africanistes in 1989: the mosque forms of the Sahel derive from Saharan and local building traditions, North African clerics and traders had been settled in the region for centuries before al-Sāḥilī arrived, and the only project the sources actually assign him is the audience chamber — where his contribution may have been organisational rather than architectural ¹⁴. The myth is worth naming because of what it is: a nineteenth- and twentieth-century assumption that a monumental African building required a foreign designer.

What changed and what was replaced

The book arrives

The single largest consequence of the transmission is not theological. It is that West Africa acquired writing, and with writing acquired a past that could be argued about in the absence of the people who remembered it.

Before the Arabic script, Mande knowledge was held in bodies: in the jeliw's genealogies, in the graded secrets of initiation associations, in the hands of smiths. After it, knowledge could be stored, copied, sold, inherited, contested and — decisively — carried by someone with no relationship to the community that produced it. The Timbuktu manuscript corpus, tens of thousands of volumes surviving in private family libraries into the twenty-first century, is the visible end of this. Its contents are not mainly devotional: they are jurisprudence, contracts, correspondence, astronomy, medicine, grammar, arithmetic, poetry and estate inventories ¹⁵.

Anonymous manuscript, Timbuktu, Mali. World Digital Library (item 463). Public domain via Wikimedia Commons. · Public Domain

Elias Saad's social history of Timbuktu traces what this did to the shape of the city. From roughly 1400 the scholarly families — the Aqīt, the Anda Ag-Muḥammad and others — constituted a hereditary notability whose authority rested on books and teaching licences rather than on descent from a conquering lineage or on office held from a king ¹⁵. This was a genuinely new kind of West African elite. It was recruited by examination and apprenticeship rather than by birth alone; it was transregional, corresponding with Fez and Cairo; and it was, crucially, not the king's creature. Timbuktu's jurists could and did rule against the state.

A second script for a first language

Arabic writing did not stay Arabic. West African scholars adapted the script to write their own languages — the practice now called ʿajamī — producing written Fulfulde, Hausa, Wolof, Soninke, Kanuri and Mandinka. The transmission's deepest linguistic effect is therefore not that West Africans learned a foreign language but that they acquired a technology for their own, and used it for pharmacopoeias, tax records, poems and letters that had nothing to do with Arabia ⁵¹⁵.

Vocabulary followed. The Mande and Sahelian languages absorbed the Arabic terminology of law, commerce, time and literacy: words for contract, witness, inheritance, interest, book, pen, week, and the names of the days and months. A West African merchant of 1400 kept accounts on an Islamic calendar in a borrowed script, and the calendar itself was a transmission — a fixed, portable, lunar timekeeping standard replacing local agricultural reckonings that could not be synchronised across a thousand kilometres of trade route.

Law replaces the elders

The most intimate change was jurisdictional. Before the qāḍī, a dispute among Mande people was resolved inside a web of relationships: by elders whose authority came from position in a descent group, by the mediation of a jeli, by oath sworn at a lineage shrine, sometimes by ordeal. The procedure and the judgment were embedded in the community that produced them, and both parties had to go on living in it afterwards.

The imported system did something categorically different. It made the dispute portable. A written contract witnessed before a qāḍī could be enforced by a judge who had never met either party, in a town neither had visited, according to a text neither had written. That is an enormous gain in reach and an equally large loss in embeddedness, and both effects were felt at once.

The clearest instance is inheritance. Mālikī law prescribes fixed fractional shares — the Qurʾānic farāʾiḍ — allotted to specified heirs, including daughters, wives, mothers and sisters. Mande property in land and productive assets was typically held corporately by the descent group and managed rather than owned. Islamic succession is therefore not merely a different rule for the same problem; it is an individualising property regime pressed onto a corporate one, and where it took hold it fragmented estates that had previously stayed whole.

Here the gender ledger runs both ways, and the record should say so plainly rather than choosing the convenient direction. Islamic inheritance gave women a defined, enforceable, court-recoverable share of an estate — in many customary regimes women had no such claim at all. At the same time, the same body of law brought norms of seclusion, guardianship and testimony that constricted exactly the public standing Ibn Baṭṭūṭa found so shocking in 1352. Malian women gained a right in the law of property and lost ground in the law of persons. It took centuries to work through, unevenly, and it is not finished.

The jurists become an estate

Under the mansas and then under Songhay, the Muslim scholarly class became something West Africa had not had: a permanent, self-reproducing institution with an interest distinct from the ruler's. Its members drew income from teaching, from endowments, from adjudication fees and from trade — many were merchants themselves. They held a body of doctrine the king could not rewrite. And they possessed the one thing that gives an intellectual class real leverage: an external audience. A Timbuktu jurist's reputation was made in Cairo and Fez as well as at home.

Al-Saʿdī's Taʾrīkh al-Sūdān, compiled in Timbuktu in the seventeenth century and translated by John Hunwick, records the consequences when a ruler and this estate collided — most bloodily under the Songhay ruler Sunni ʿAlī, whose persecution of the Sankore scholars in the later fifteenth century the chronicle narrates as an atrocity and whose successor Askia Muḥammad restored them ¹⁶. The chronicle is a partisan document written by the scholars' own tradition, and its bias should be read openly. But the collision itself is real, and it is only possible because the transmission had created an institution capable of colliding.

What the mosque displaced

The costs of arrival begin here, and the atlas will not fence them off in a separate section.

The figurative terracotta tradition of the Inland Delta — the equestrian and kneeling figures now in museum collections around the world — thins out and ends in the fifteenth and sixteenth centuries, in the same period and the same region in which Islamic observance deepened below the elite. The causal chain cannot be proved from the ground, and honest archaeology says so: looting has destroyed most of the contexts that would have dated the decline precisely ⁵. But an iconoclastic doctrine arriving in a figurative culture, followed by the end of that culture's figurative production, is not a coincidence one gets to shrug at.

The royal funerary practice went too. Al-Bakrī describes the Ghana kings buried under a great dome with their possessions and their attendants, and matrilineal succession — the throne passing to the sister's son ⁴. The tumulus burial disappears with Islamization; so, gradually, does the sister's-son rule, replaced under the mansas by a contested but recognisably patrilineal dynastic principle. What was given up was not merely a custom. It was a theory of where authority comes from.

The jeliw survived — they survive still — but were demoted in one specific respect. The past became something a stranger could look up. A chronicle written in Timbuktu could be read in Fez by a man who had never met a Keita, and it outranked, in the courts and in the eyes of the state, the recitation of the person whose family had held that history for two hundred years.

Women, and what the elite settlement cost them

Ibn Baṭṭūṭa reached Mali in 1352 and spent roughly eight months at the court of Mansa Sulaymān, Mūsā's brother. His account is the fullest eyewitness description of the empire in existence, and it is also the record of a Maghrebi jurist's continuous irritation ¹⁷.

He admired much: the security of the roads, the horror of injustice, the memorisation of the Qurʾān by children, the punctuality of Friday prayer. What scandalised him was gender. Malian women moved unveiled in public, held property, kept male friends and companions without their husbands' supervision, and were treated by their menfolk as though none of this required comment. At Walāta he found matrilineal descent — inheritance through the sister's son — and wrote that he had encountered the like nowhere but among the Hindus of Malabar ¹⁷. He also reports, without seeing the contradiction, that the queen was named in the Friday sermon alongside the king: a woman's political standing publicly proclaimed in the central ritual of the imported religion.

That is a rare thing in the sources — a transmission caught at the exact instant of friction. Ibn Baṭṭūṭa's complaint is documentary evidence that Mande gender arrangements were still intact in 1352, and his authority as a jurist is evidence of the pressure being applied to them. Michael Gomez's African Dominion argues that it is precisely in this period, in the interaction of Islamic law with imperial expansion and an intensifying slave trade, that durable categories of gender, caste and — eventually — race were assembled in West Africa ¹⁸. What Ibn Baṭṭūṭa found startling in 1352, later Sahelian reform movements would find intolerable.

What the cost was

The law that decided who could be sold

Slavery in West Africa long predates Islam, and no honest account of this transmission says otherwise. Enslavement through war, debt and judicial penalty is attested across the region before the caravans, and the northbound trade in captives was running before the Mande elites converted.

What the transmission supplied was not the practice. It was the doctrine — and doctrine is what turns a practice into a system that can be scaled, financed, defended in court and exported.

Islamic law forbids the enslavement of Muslims. That is a protection, and for the converted populations of the Sahel it was a real one. But a rule that shields believers is, read from the other side, a rule that marks everyone else. In a Sahel where conversion had gone in at the top and had barely reached the cultivators, the legal frontier between protected and enslavable ran through the middle of the same society — along, very nearly, the line the Mali state raided.

The demand side was the Islamic Mediterranean and the Middle East, and it was continuous. Paul Lovejoy's synthesis, the standard reference on volumes, estimates roughly 4.82 million people carried north across the Sahara between 650 and 1600 ¹⁹. Ralph Austen's separate census work produces figures of a comparable order ²⁰. Both scholars stress that these are reconstructions built on fragmentary customs data and caravan reports, not counts; they should be quoted as estimates and never as ledger entries. Even discounted heavily, they describe a traffic that ran for nine centuries at a scale the Atlantic trade would later exceed but did not invent.

11 September 1353

Statistics of that size go numb. The Arabic sources occasionally supply the correction.

On 11 September 1353, Ibn Baṭṭūṭa left the copper town of Takedda, in what is now Niger, travelling north to Morocco under orders from his sultan. He notes the composition of the caravan he joined. It was carrying six hundred enslaved women to be sold in the north ¹⁷.

He does not editorialise. He is a jurist in good standing, riding home in the ordinary company of the ordinary trade, and he records the number the way one records the weather. There is no better single line in the medieval record for what the gold road actually moved, and no line that better illustrates why the transmission's bill cannot be tallied separately from its gift. The same caravan network that carried six hundred women north carried the books south.

Most of them never crossed the desert

The Saharan figures, for all their weight, measure only the exported fraction. The larger institution stayed at home.

Enslaved people in Mali and its successor states worked agricultural estates along the Niger, served in royal households, staffed the court, and formed units of the army. Ibn Baṭṭūṭa's description of Mansa Sulaymān's audiences is full of them — hundreds of attendants around the pavilion, bearing arms and insignia, in a ceremonial that displays servile manpower as a form of royal wealth ¹⁷. This is not a marginal practice at the empire's edge. It is the ordinary staffing of the central state.

Michael Gomez's argument in African Dominion is that the sorting this required — who is enslavable, who is protected, which lineages are which — is one of the mechanisms by which durable categories of caste, ethnicity and eventually race were assembled in West Africa across exactly these centuries ¹⁸. Islamic law did not create West African servitude. What it supplied was a criterion sharp enough to build a permanent status system on, and a scriptural vocabulary in which that system could be argued and defended.

The consequence has a very long tail. Descent-based servitude — hereditary status attached to lineages descended from the enslaved, with attendant restrictions on marriage, land and office — persists in parts of Mali, Mauritania and Niger into the present, and remains an active political, legal and human-rights dispute in the region. It is the oldest surviving load-bearing element of this transmission and the least discussed.

Aḥmad Bābā, and the argument he half-won

The best evidence that the enslavement question was understood as a moral problem at the time — by West Africans, in Arabic, on Islamic grounds — is the career of Aḥmad Bābā al-Timbuktī (1556–1627), the most celebrated jurist Timbuktu produced.

In 1615 he wrote the Miʿrāj al-Ṣuʿūd, a treatise answering questions about who among the peoples of the Sudan might lawfully be enslaved. It has been edited and translated by John Hunwick and Fatima Harrak ²¹. Aḥmad Bābā's central argument is a repudiation: blackness is not a cause of enslavement, the racialised reading of the curse of Ham is false, and free Muslims of the Sudan who are seized and sold are being wronged. Timothy Cleaveland's study of the treatise sets out how sharply this cut against Maghrebi practice, which routinely treated black West Africans as presumptively enslavable ²².

And then the other half. The treatise establishes the criterion as religion, and having done so it enumerates: these peoples are Muslim and may not be taken; those peoples are unbelievers and may be. The effect of the most humane statement on slavery that medieval West Africa produced was to give the trade a defensible legal test — and to leave the non-Muslim interior inside it ²¹²².

This is the record's centre of gravity. The doctrine that protected did so by defining. It is not enough to say that Islam brought literacy and law to West Africa and that slavery was a separate misfortune. The law was the instrument that sorted the enslavable from the protected, and it worked.

Costs that were not Mali's, and costs that were

Not every bill in this record was paid by West Africans, and the atlas should say which was which.

The Cairo entry is worth dwelling on for one reason: it is the only cost in this record that the receiving culture imposed on the sending one. Every other line runs the usual direction.

The bill

Mali got what it went for, and what it got was not small. A written law that a merchant in Fez and a merchant in Jenne could both invoke. An archive. A scholarly class with independent authority. A diplomatic identity that survives to this day on a Majorcan sea chart of 1375, where the mapmaker drew a crowned African king holding a gold nugget and labelled the interior of a continent Europe had never entered.

The empire itself did not last. Mali fragmented through the fifteenth century, lost the Niger bend to Songhay, and was reduced to a rump on the upper river; Songhay in turn fell to a Moroccan expedition with firearms in 1591, after which Timbuktu's scholars — Aḥmad Bābā among them — were deported north. But the transmission outlived every state that carried it. The Sahel is Muslim today at rates above ninety per cent in Mali, Senegal and Niger, and only a little below that in Guinea. The manuscripts survived, through colonial confiscation and the 2012–13 occupation of northern Mali, because families hid them.

And the other half survived too, in the form of an argument that West Africa has never stopped having: about who counts, who is protected, and who is available. That argument was not invented on the Middle Niger. It arrived by caravan, in writing, with a jurisprudence attached, and it was the price of the books.