Mali's gold minted the Mediterranean's money — then drew an army

Before: a monetary world that could not mine its own gold

Around 800 CE, the Islamic Mediterranean ran on two metals and owned only one of them in abundance. From al-Andalus to Iraq the caliphal system priced, taxed and saved in the gold dinar and the silver dirham, and its gold came overwhelmingly from stock: Byzantine and Sasanian treasure, temple and church plate, old coin melted and re-struck. The quantities were modest. Jean Devisse, who spent two decades excavating the Saharan market town of Awdaghust, assembled the few hard mint figures that survive: Umayyad Córdoba struck about 40,000 dinars in 1009–10, perhaps 160 kilograms of gold, and the Tulunid mint of Egypt at most 100,000 dinars a year, about 400 kilograms ³. A Mediterranean gold coinage on that scale needed a new source of metal every year, and the lands around the Mediterranean did not supply it.

The Maghrib's own memory of gold was older than Islam. Timothy Garrard's metrological study showed that a trans-Saharan gold trade does not appear before the end of the third century CE; an irregular gold coinage was struck at Carthage between 296 and 311, the late Roman tax system in Africa was reorganised to collect more gold by the end of the fourth century, and the weight of Constantine's solidus, first struck in 312, became the unit in which West African gold dust was weighed — a unit that survived, he argued, in the Akan gold-weights of the nineteenth century ⁵. By the time Arab armies reached Ifrīqiya, there was already a road south. What there was not yet was a state on either side able to make that road pay on the scale it later would.

What the north had, and what it lacked

The receiving world before the gold arrived in volume had several things in place and one thing missing:

• A unified monetary law. Dinar and dirham were struck to known standards and accepted from Seville to Samarqand, so gold struck anywhere in the system could circulate everywhere in it.
• Merchant credit. Partnership contracts, bills of debt and the Mālikī law of the Maghrib allowed trade at a distance without carrying coin.
• Caravan termini. Sijilmāsa in the Tafilalt was founded in the mid-eighth century; Wargla and the Ifrīqiyan cities followed.
• The camel. The single-humped camel had been working in North Africa for centuries, a transmission this atlas follows separately.
• No primary gold. None of the Maghrib's own mines yielded gold in quantity. The metal that would make the Fatimid and Almoravid dinars famous had to come from somewhere else, and its source lay more than 1,500 kilometres south of Sijilmāsa.

That last absence is the whole story. The northern states that grew rich on gold for the next eight centuries never controlled where it came from, and for most of that time they did not know exactly where it was.

The gravel country

The gold lay in two principal fields on the upper Senegal and upper Niger, in the country of the Mande-speaking peoples. Bambuk lies between the Senegal and its tributary the Falémé; Buré lies on the upper Niger around modern Siguiri, several days' walk from the later Malian royal towns. Robyn d'Avignon's history of West African gold mining, A Ritual Geology (2022), maps further fields at Poura, Gaoua, Hiré, Yaouré, Sirba and in the Akan forest, some of which probably entered production during Mali's reign ¹. Raymond Mauny's great survey of 1961, the Tableau géographique de l'Ouest africain au Moyen Âge, put the medieval exportable output of Buré alone at about four tonnes a year, and of the Gold Coast and Ivory Coast fields at four more ⁴ — figures that, Devisse later pointed out, were extrapolated from twentieth-century production rather than from medieval evidence ³.

How the gold was won is known mainly from later observation, and the atlas flags that limit plainly. In the nineteenth- and twentieth-century goldfields that d'Avignon studies, the work was seasonal, done in the dry months between harvest and the rains, roughly January to March. Men dug circular shafts with iron picks that doubled as farming tools; women washed the gravel in nested calabashes, swirling it down to "a very fine black sand, in which is found gold," as the French traveller Raffenel recorded on the Falémé in 1846 ¹. Mining teams worked a field only with the permission of a damantigi, a mine chief, and specialists who could "see" gold sited the shafts. Whether the medieval organisation was the same in detail cannot be shown. What the medieval sources do show is that it was seasonal, decentralised and in the hands of farming communities, not of any state ¹³.

Devisse put a number on what that meant. A single pit, he estimated, yielded between 2.5 and 5 grams of gold; to export even two or three tonnes a year, mining "a dû mobiliser chaque année, en Afrique de l'Ouest, des centaines de milliers de personnes" — must have mobilised hundreds of thousands of people every year ³. The gold of the Mediterranean was a by-product of the agricultural calendar of the savanna.

Ghana: the first state on the threshold

The first West African state to profit systematically was Ghana (Wagadu), whose Soninke kings held the country between the goldfields and the desert. Al-Bakrī, the Andalusi geographer who compiled his Book of Routes and Realms in Córdoba in 1068 from merchants' reports, described a fiscal system built on the threshold ²:

• "On every donkey-load of salt when it is brought into the country their king levies one golden dinar, and two dinars when it is sent out."
• A load of copper paid five mithqāls, and a load of other goods ten.
• "The nuggets found in all the mines of his country are reserved for the king, only the gold dust being left for the people. But for this the people would accumulate gold until it lost its value."

Devisse argued that the second passage has long been misread. Al-Bakrī's Arabic opposes dhahab, refined gold of the kind a king hoards and gives, to tibr, native gold as it comes from the ground; « l'opposition faite classiquement entre pépites et poudre ne tient pas » — the classic contrast between nuggets and dust does not hold ³. Read his way, Ghana's king was not hoarding lumps: he was controlling the refining and the conversion of gold into a store of value, and leaving unprocessed gold to circulate. Either reading makes the same point. Before the Mediterranean ever saw this gold, a West African state had built a fiscal regime to manage its supply and protect its price.

The transmission: from the pit to the die

The gold travelled north in stages, and at no stage before the Sahel did a northern merchant handle it. The chain that existed by the tenth century, and that held with changing termini until the sixteenth, can be laid out simply:

The merchants in the second row of that table are the least visible and the most important. Al-Bakrī names a people called the Naghmārata who traded in native gold; Devisse suggested they may be the ancestors of the Wangara, the Mande-speaking merchant diaspora that later Arabic writers credit with the gold trade ³. The twelfth-century geographer al-Idrīsī, writing for the Norman king Roger II of Sicily in 1154, places the Wangara in the gold country south of Ghana and says, in Devisse's reading, that they could not act without the control of Ghana's ruler ³¹⁹. The pattern holds across five centuries: the goldfields were worked by farmers, the gold was moved by Mande merchants, and the states of the Sahel taxed the passage. Northern merchants met the gold only in Sahelian markets.

The sources that describe this chain were almost all written from outside it, and their distance from the gold is part of the evidence:

• Ibn Ḥawqal, a merchant-geographer of the tenth century, reports from Sijilmāsa and repeats what its merchants told him ⁷.
• Al-Bakrī, writing in Córdoba in 1068, never crossed the Strait; he compiled merchants' reports and earlier written itineraries ².
• Al-Idrīsī, at Palermo in 1154, worked from the same kind of material and is often confused about the geography of the Niger ¹⁹.
• Al-ʿUmarī, in Cairo in the 1330s, interviewed Egyptians who had dealt with Mansa Mūsā's party and informants who had lived in Mali for years ².
• Ibn Baṭṭūṭa, in 1352–53, is the only medieval Arabic writer who crossed the Sahara and wrote about Mali from inside it ².
• Al-Saʿdī, writing in Timbuktu in the 1650s, records the end of the story from the Middle Niger itself ¹⁷.

None of these writers saw a goldfield. The secrecy that the Arabic sources lament was not a curiosity of African trade; it was the Wangara's business model, and it worked.

Awdaghust, Tadmekka and the cheque

At the southern edge of the desert the market towns grew rich on the exchange. Al-Bakrī says of Awdaghust that "their transactions are in gold, and they have no silver" ². Devisse's excavations at Tegdaoust, the site identified with it, recovered Fatimid glass weights of the tenth century, gold wire drawn through draw-plates, chisel-cut gold half-ingots and a crucible fragment still holding a bead of gold ³.

The Arabic geographer Ibn Ḥawqal, writing in the tenth century, reported a written bill for 42,000 dinars owed by a merchant at Awdaghust, a sum he said he had never heard of in all his travels. Nehemia Levtzion showed in 1968 that the document was a debt between two merchants of Sijilmāsa, one of them resident in the Sahel, and that Ibn Ḥawqal probably never went to Awdaghust himself ⁷. Reread that way, the cheque says something more precise than "fabulous wealth": by the tenth century, Sijilmāsa's merchant houses kept agents at the far end of the desert and settled across it in paper.

Further east, at Essouk-Tadmekka in the Adrar des Ifoghas, Sam Nixon, Thilo Rehren and Maria Filomena Guerra identified in 2011 a set of ninth- to tenth-century clay moulds used for casting small gold blanks, the first concrete evidence of coin-making in pre-colonial West Africa. Their reading is careful: "these coins were not political statements, but were probably blank and intended to facilitate the busy early Islamic caravan trade" ⁶²¹. A Sahelian town was standardising gold to Mediterranean weight before it left the Sahel.

Sijilmāsa and the Fatimid treasury

The northern terminus was Sijilmāsa, and it was there that the gold first became a political instrument. Ibn Ḥawqal put the revenue its ruler drew from the trade at 400,000 dinars a year in the late tenth century ⁸. Yaacov Lev's study of the early Fatimids reads that figure as roughly half the dynasty's annual revenue in the mid-tenth century, alongside a massive import of high-quality sub-Saharan gold, and notes al-Maqrīzī's later claim that the caliph al-Muʿizz spent 24 million dinars on the conquest of Egypt in 969 — a figure Lev says "should perhaps not be taken literally," but which conveys the scale of resources the Fatimids commanded ⁹. The dynasty that founded Cairo did so on credit drawn, in large part, on the gravels of the upper Niger.

Peter Spufford's survey of medieval money notes a telling detail from Sicily: the gold tari struck there under the Fatimid caliphs al-Ẓāhir and al-Mustanṣir in the 1030s and 1050s "had also been of a West African fineness" before the later debasement ⁸. Gold from Buré was already the standard against which Mediterranean coins were measured.

The Almoravids take both ends of the road

In the 1050s a movement of Ṣanhāja Berbers from the western Sahara, the Almoravids (al-Murābiṭūn), seized both termini of the western route in quick succession. Al-Bakrī records that in 446 AH (1054–55) ʿAbd Allāh b. Yāsīn's forces took Awdaghust, "a flourishing locality," because its people recognised the authority of the king of Ghana; "the Almoravids violated its women and declared everything that they took there to be booty of the community" ². In the same years they took Sijilmāsa, and dinars were soon being struck there in the name of their commander Abū Bakr b. ʿUmar ¹⁰.

Riad Salih. Almoravid gold dinars, 1056–1145. National Museum of Antiquities, Algiers. CC BY 4.0 via Wikimedia Commons. · CC BY 4.0

The result was a monetary monopoly. Spufford observes that with the decline of Kairouan and Mahdia, Sijilmāsa became "the sole significant mint in North Africa for thirty years" ⁸. Ronald Messier's analyses of Almoravid dinars, published in 1974, tied their exceptionally high and stable fineness to West African metal and traced how they became the reference gold coin of the western Mediterranean ¹⁰. Christian Iberia called it the morabetino or marabotin, after the dynasty, and priced land, dowries and ransoms in it.

Mali and the tribute of the unbelievers

After Ghana's decline, the Mande state of Mali, consolidated in the thirteenth century, took over the position Ghana had held — and refined the policy. Al-ʿUmarī, the Damascene official who interviewed people in Cairo who had dealt with the Malians, recorded in the 1330s why the mansas did not conquer the goldfields ²:

"If the sultan wished he could extend his authority over them, but the kings of this kingdom have learnt by experience that as soon as one of them conquers one of the gold towns and Islam spreads and the muezzin calls to prayer there the gold there begins to decrease and then disappears, while it increases in the neighbouring heathen countries."

So, he continues, they "left the gold countries under the control of the heathen people and were content with their vassalage and the tribute imposed on them." Another Mamluk chronicler, al-Dawādārī, reported Mansa Mūsā himself saying in Cairo that if Mali conquered the gold country "it does not put forth anything" ². D'Avignon's gloss is exact: "Mali's kings shared at least one belief with gold miners: Muslim overrule made gold disappear. Gold preferred pagans" ¹.

This was not superstition alone. It was a policy that left mining communities their land chiefs, their ritual specialists and their own calendar, in exchange for tribute — a form of indirect rule that kept the supply flowing because it did not try to own it. Buré, d'Avignon shows, was multi-ethnic and non-centralised, paying tribute to the mansa while keeping its political and religious autonomy ¹. The empire that the Catalan Atlas of 1375 would call the land of "the richest and most distinguished ruler of this whole region on account of the great quantity of gold that is found in his lands" did not mine the gold itself ²⁰.

What changed and what was replaced

The gold changed the money of three civilisations: the Islamic West, Latin Christendom and Mamluk Egypt. Each took it in a different form and each, eventually, lost it in a different way.

The dinar that crossed the Strait

By the end of the twelfth century, Spufford writes, "West African gold was in use as coin in Christian Sicily and the Christian parts of Spain" ⁸. Genoese notaries of the 1180s and 1190s distinguished Sicilian tari gold from African paiola gold, the gold dust that reached Genoa through the ports of the Maghrib ⁸. When the Almoravid mint at Murcia stopped striking in 1170, Castile simply took over the coin. In 1172, Alfonso VIII "began to have his own gold morabetinos, or maravedis, struck at Toledo. They copied the format of the last Almoravid dinars, but replaced the specifically Muslim inscriptions by Christian ones, whilst still retaining their Arabic script. They too were of the West African fineness of 20¾ carats" ⁸.

A Christian king, at war with the Almohads, struck his gold in Arabic script to the fineness of metal dug by Mande farmers. León and Portugal followed. The Castilian maravedí survived as a unit of account for centuries after its gold had gone — debased into a copper coin, it was still the money of account of the Spanish empire in the early modern period.

What the gold displaced in Iberia was the absence of a Christian gold coin. Before the morabetino, Christian Spain had used Muslim gold or none; after it, gold coinage became an attribute of Christian kingship, and the shape of that coinage was Almoravid.

1252: Latin Europe returns to gold

North of the Pyrenees, Western Europe had struck its money in silver since the Carolingians. In the 1950s Robert S. Lopez gave that monetary fact a date and a title, "Back to Gold, 1252": in that year Genoa and then Florence began to strike gold coins of their own, the genovino and the fiorino ¹¹. Spufford's summary is direct: "By 1252 there was enough gold in, or passing through, Genoa and Florence for these cities to commence striking gold coins of their own… within a few months of one another" ⁸.

Sailko. Genovino d'oro (type III), from 1280. Musei di Strada Nuova – Palazzo Doria-Tursi, Genoa. CC BY-SA 3.0 via Wikimedia Commons. · CC BY-SA 3.0

Andrew Watson's 1967 study explained why the metal moved. Gold was cheaper relative to silver in North Africa than in Italy — around 6.5 to 1 at Tunis in the late twelfth century against 8 or 9 to 1 at Genoa between the 1160s and the 1250s — so Italian merchants carried silver south and gold north at a profit ¹²⁸. Spufford draws the comparison that frames this record: "It is little wonder that the gold of the western Sudan flowed to North Africa and to Muslim Spain and Sicily in much the same way as the silver of central Europe flowed to northern Italy" ⁸.

The florin became the reference coin of European international finance, the unit of papal accounts and of the great banking houses; Venice followed with the ducat in 1284 ¹³. These were Latin coins with Christian iconography — the lily of Florence, the gate of Genoa, the doge kneeling before Saint Mark. Their metal, for the first century of their existence, came substantially from the Sahel.

What the return to gold changed in Europe:

1. A two-tier money. Gold for long-distance trade, taxes, papal finance and war; silver and billon for wages and the market. The division between the money of the rich and the money of the poor was sharpened.
2. New fiscal capacity. Large sums — dowries, ransoms, mercenary contracts, papal taxes — could be carried and counted in gold.
3. A cartographic hunger. Majorcan and Catalan chartmakers began to place the "king of gold" on their maps, and the Catalan Atlas of 1375 gave him a crown, a sceptre and a gold nugget.
4. A strategic question. If the gold came from beyond the desert, could the desert be bypassed?

The coins that carried West African gold into the Mediterranean world, in the order they appeared, make the lineage visible:

Cairo, the ducat and the ashrafī

Egypt received the gold most dramatically and lost it most clearly. In 1324 Mansa Mūsā, travelling to Mecca, passed through Cairo with a train that Spufford, following the sources, says "reputedly" carried 80 to 100 camel-loads of gold, and Devisse estimated at about a tonne ⁸³. Al-ʿUmarī, writing about twelve years later, recorded the effect precisely: "Gold was at a high price in Egypt until they came in that year. The mithqal 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. The mithqal does not exceed 22 dirhams or less" ². The religious and diplomatic consequences of that journey for Mali are the subject of a separate record in this atlas; here the point is monetary. A single West African payment could move the price of gold in the largest city of the Islamic world for a decade.

Seventy years later the movement ran the other way. Jere Bacharach's study of the Mamluk gold coinage shows the Venetian ducat — itself struck from metal that had come, in part, across the Sahara — dominating the markets of the eastern Mediterranean from about 1399, until Sultan Barsbāy's ashrafī of 1425 replaced it as the principal gold coin of the Mamluk state ¹³. Spufford records that Barsbāy banned and reminted the ducat, and that the 200,000-ducat ransom of the king of Cyprus, captured by the Mamluks in 1426, "was at once reminted as ashrafi" ⁸. Egypt had passed from being the place where West African gold was spent to a place where it had to be recaptured from Venice.

What the gold displaced

The transmission's displacements were real, if less visible than its coins:

• Silver monometallism in Latin Europe, after five centuries.
• The old eastern routes — the Fatimid-era trade through Ifrīqiya and Egypt lost ground to the western road through Sijilmāsa once the Almoravids took both ends of it.
• Ghana's position as broker, overtaken in the twelfth and thirteenth centuries by Mali, whose authority reached nearer the fields.
• Local savanna stores of value — copper, cloth, cowries — which persisted for local exchange but were subordinated to gold in long-distance trade.

How much gold? The figures and their limits

No number in this record is more contested than the volume. The principal estimates, laid side by side:

Devisse's warning should be printed above every such table: « Il serait puéril d'imaginer qu'un flot d'or est sorti chaque année de l'Afrique de l'Ouest » — it would be childish to imagine that a flood of gold left West Africa every year ³. Two or three tonnes is a few dozen camel-loads. That is the paradox the record turns on: a small, steady, well-managed stream, controlled at its source by people who never saw a mint, was enough to anchor the gold money of the Mediterranean world for half a millennium.

What the cost was

The gold itself was dug by people who kept, on the best evidence, control of their land and their labour; Mali's tribute policy was built to leave them alone. The bill for this transmission was paid elsewhere along the chain — by those who dug the salt that bought the gold, by those who travelled north in the same caravans, by the towns at the termini when outsiders came for them, and finally by the Middle Niger itself.

Salt for gold: Taghāza

The principal thing the north sent south was salt, and the principal salt was cut from the dry lake beds of Taghāza, in the middle of the desert. Ibn Baṭṭūṭa crossed it in 1352 on his way to Mali and left the sharpest description in the sources: "No one lives at Taghaza except the slaves of the Massufa tribe, who dig for the salt; they subsist on dates imported from Darʿa and Sijilmasa, camels' flesh, and millet imported from the Negrolands" ². He saw the houses and mosque built of salt blocks, roofed with camel skins, and the economics behind them: "The business done at Taghaza, for all its meanness, amounts to an enormous figure in terms of hundredweights of gold-dust" ².

Taghāza is the hinge of the cost. The gold of Buré was bought, in large part, with salt dug by enslaved men in one of the most hostile workplaces in the medieval world, fed on imported food, with no settled free population at all. The sources do not tell us how many worked there, or how long they lived.

The same caravans

The camels that carried gold north also carried people. Spufford, an economic historian of coinage and not of slavery, makes the point without softening it: caravans "took with them large numbers of slaves who had been seized by the rulers of the successive savannah kingdoms of Ghana and of Mali in raids on their southern forest-dwelling neighbours" ⁸. Ibn Baṭṭūṭa returned north in 1353 with a caravan that included a trader taking 600 enslaved women across the desert ⁸².

The scale of the trans-Saharan slave trade is estimated, not counted. Paul Lovejoy's tabulation, built on Ralph Austen's census of the sources, gives ¹⁶:

These figures are not the cost of the gold alone: the slave trade had its own demand, its own law and its own history, treated in this atlas's record of Islam's arrival in Mali. But the two trades used the same roads, the same merchants, the same credit and the same camels, and the profit of one subsidised the risk of the other. Separating them neatly would misdescribe how the caravans actually ran.

Awdaghust, 1054

The first time a northern power reached for the sources, the cost was immediate. The Almoravid seizure of Awdaghust in 1054–55 was a punitive conquest of a town that recognised the king of Ghana. Al-Bakrī's report — the town's women violated, everything in it declared booty — is the earliest documented act of war in the gold trade's history ². Within two decades Awdaghust's position was in decline, and the gold trade's centre of gravity moved east, to Walata and later Timbuktu.

The Atlantic turn

From the 1440s the Portuguese tried to do by sea what the Almoravids had done by land. At Arguim, on the coast of Mauritania, Prince Henry had a fort built in 1448; the Venetian Alvise Cadamosto, who sailed for him, wrote that Arab merchants brought "slaves whom the Arabs bring from the land of the Blacks, and gold tiber" there, and that "every year the Portuguese carry away from Argin a thousand slaves" ¹⁴. In 1482 the Portuguese built São Jorge da Mina on the Gold Coast, closer to the Akan fields. Vitorino Magalhães Godinho's Os Descobrimentos e a Economia Mundial made the search for Sudanese gold one of the engines of Portuguese expansion, and the argument has held ¹⁵.

Spufford's judgement is measured: "It is not clear whether this Portuguese gold represented any increase at all in the amount of gold reaching Europe from West Africa, or merely a diversion of the means by which it came" ⁸. But the diversion had consequences of its own: it "did have a very marked adverse effect on the commerce of the whole of North Africa from Morocco eastwards to Tunis and Tripoli, and even as far as Egypt" ⁸. And the fort at Arguim that bought gold dust bought people too — the first leg of what became the Atlantic trade.

1591: the army that came for the source

The final bill was presented by the state that had suffered most from the Atlantic diversion. In 1590 the Saʿdian sultan of Morocco, Aḥmad al-Manṣūr, sent an army built around musketeers across the Sahara under Jawdar Pasha. In March 1591, near Tondibi north of Gao, they met the army of the Songhay askiya Isḥāq II — 12,500 cavalry and 30,000 infantry, according to the Timbuktu chronicler al-Saʿdī — and destroyed it with firearms ¹⁷. Lansiné Kaba's analysis of the invasion argues that its motive was economic: control of the gold trade, whose northern revenues the Atlantic diversion had cut ¹⁸.

What followed was the sack of the market rather than the capture of the mines. In Timbuktu, al-Saʿdī records, Pasha Maḥmūd arrested the scholars, his followers plundered their houses, "stripping their womenfolk," and of the wealth taken "all he sent to the sultan Mūlāy Aḥmad was 100,000 [mithqāls of] gold" — about half a tonne, Hunwick notes ¹⁷. The scholars were seized in October 1593 and marched across the desert; they entered Marrakesh on 21 May 1594 ¹⁷. Among them was Aḥmad Bābā, whose library of 1,600 volumes was confiscated ¹⁷. English merchants in Marrakesh watched the gold arrive — "thirty mules laden with gold" in 1594, thirty camel-loads of gold dust with Jawdar in 1599 ¹⁷.

In Marrakesh, the chronicler al-Ifrānī wrote, so much gold came in "that al-Manṣūr paid salaries only in nuggets of pure gold… On account of this he was named 'the Golden (al-Dhahabī)'" ¹⁷. The goldfields themselves, far to the south-west, were never taken. The Songhay state broke into fragments; the Moroccan garrisons, the Arma, became a local military caste; the trans-Saharan trade never recovered its medieval scale.

What the record cannot count

An honest cost account has to name what is missing as well as what is documented. The sources for this transmission are silent, or nearly so, on several things that matter:

• The miners' own lives. No medieval source describes the labour in Bambuk or Buré from the inside, and the ethnography that does is five or six centuries later ¹.
• The death toll at Taghāza. Ibn Baṭṭūṭa saw the enslaved salt-cutters; nobody counted them, or recorded how they were replaced ².
• Deaths on the crossing. Caravans lost animals and people to thirst and to raiding, but no medieval writer gives a figure, and modern estimates of the trans-Saharan slave trade do not separate mortality in transit ¹⁶.
• The human cost of the 1591 war beyond Timbuktu. Al-Saʿdī records the battle and the plunder of the scholars; the toll among the farming and fishing populations of the Middle Niger in the decades of conflict that followed is described only in general terms ¹⁷¹⁸.
• Where the gold of the fourteenth century actually went. The share that reached Cairo, the Maghrib, Iberia and Italy in any given decade cannot be reconstructed from mint output, as Devisse's and Spufford's very different figures show ³⁸.

These are not reasons to soften the bill. They are reasons to read the documented costs as a floor.

The bill

The gold made the coins; the coins made states on both sides of the desert. The miners who dug it, on the best evidence, were not the ones who paid. The people who paid were the salt-cutters, the captives, the merchants at the termini — and, at the end, the Middle Niger itself, conquered for gold it had never possessed in the ground.