Ghana

The name is a title

formationDrafted from (eds), Fontes Historiae Africanae, Series Arabica IV (1981)

Ghana is what the Arabic writers call the king. Al-Bakri says so directly: the ruler's title is ghana, and the country is called after him. The Soninke tradition calls the polity Wagadu.

The modern republic that took the name in 1957 is eight hundred kilometres away and was never part of it, and chose the name deliberately as a claim on a West African imperial past. It is called Ghana here because everything ever written about it is written under that word, and the word is a title rather than a name: al-Bakri says plainly that ghana is what the ruler was called, and that the country's own name was something else.

Everything here was written by people who were not here

Al-Fazari, in the eighth century, knows of Ghana as the land of gold. Al-Ya'qubi and Ibn Hawqal add to it. Al-Bakri, writing at Córdoba in 1068, produces the description everything else depends on — the court, the taxes, the succession, the double capital — and al-Bakri never left al-Andalus. Al-Idrisi, working for a Norman king at Palermo a century later, largely copies him.

That is the evidentiary situation, and it is why the Arabic accounts are best read printed side by side rather than as a narrative: laid out together it becomes visible which passages are independent testimony and which are one author copying another, and for this kingdom a great deal is the second.

Nothing was written from inside. There are no Ghanaian documents, no inscriptions, no coins, and the kings were not Muslim, so the literate community in the country wrote about trade rather than about the state.

Kumbi Saleh

Al-Bakri describes a double town. One half is where the Muslims live — merchants, with twelve mosques, salaried imams and muezzins, and jurists and scholars. The other half, six miles away, is the king's, with the royal residences, the domed buildings, a grove where the priests and the idols are, and the tombs of the kings.

Excavation at Kumbi Saleh in southern Mauritania found a substantial stone-built town of about the right period, and did not find the royal half. The identification is probable and not certain, and it has been argued about for seventy years.

(eds), Fontes Historiae Africanae, Series Arabica IV (1981) · Chloé Capel · CC BY-SA 4.0

What the two halves were for

They are the whole political arrangement, and the difference between Ghana and the empires that followed it.

Mali and Songhai had Muslim dynasties. Ghana did not. What it had was a state that hosted a Muslim commercial and legal community, paid its clergy, used its literacy, employed Muslims as ministers by al-Bakri's account — and kept its own king in the other town with the grove and the idols, buried under a dome with his retainers and his weapons.

That arrangement lasted for centuries, and it worked precisely because the state carried nothing outward. It did not convert its subjects, did not send missionaries, and did not require its trading partners to be anything. It kept the two things six miles apart and let the caravans do the rest, which is a considerably more sophisticated arrangement than it is usually given credit for.

A dinar on a donkey

expansionDrafted from (eds), Fontes Historiae Africanae, Series Arabica IV (1981)

Al-Bakri's passage on Ghana's revenue is the most quoted paragraph in the early Arabic material on West Africa, and it is quoted because it is specific in a way almost nothing else about this polity is.

A dinar of gold is taken on every donkey-load of salt that enters the country, and two on every load that leaves it. Copper pays five mithqals, other goods ten. And all nuggets of gold found in the mines of the kingdom belong to the king; only the gold dust is left to the people, because if he did not take the nuggets, gold would become so plentiful as to be worthless.

That is the kingdom's entire fiscal system, set down by a geographer in Córdoba who had it from merchants: a transit duty on everything moving in either direction, and a royalty on the metal itself.

The salt goes the other way

The asymmetry in the salt duty is the trade in a sentence. One dinar to bring a load in, two to take one out.

Salt came south from the Sahara, where there was too much of it and nothing else, into a savanna that had none and needed it. Gold went north. Ghana sat between and charged in both directions, and charged more on the way out because that was the direction the gold was going.

There is a whole class of states that lived on a transit duty and nothing else — the Nabataeans on the incense road, Palmyra, Characene, the toll fortress at Derbend on the Caspian gate, Srivijaya on the Malacca strait, Trebizond at the end of the Tabriz road. They are states that owned nothing but a position, and they lasted exactly as long as the traffic did.

Ghana was not one of them, and the second half of al-Bakri's paragraph is why. It had a position and a claim on what came out of the ground behind it, which is the difference between an empire and a town on a pass.

The nugget rule outlives the state

The second half of al-Bakri's paragraph is the part that matters for the rest of this region.

All nuggets to the king, only dust to the people, and an explicit reason: to keep gold from becoming worthless. It is a monetary policy, stated as one, in the eleventh century.

Two hundred years later the mansas of Mali were running exactly the same arrangement, and two hundred years after that so was Songhai. In none of the three cases did the empire own the goldfields. The gold came out of Bambuk and later Bure, worked by people who were not the empire's subjects in any meaningful sense, and every one of these states taxed the product rather than the production — because taxing the product requires a customs post and taxing the production requires an occupation.

The instrument outlasted three empires, four capitals and six hundred years. It is the single most durable thing in the political history of the western Sudan.

Dogs in collars of gold

peakDrafted from (eds), Fontes Historiae Africanae, Series Arabica IV (1981)

Al-Bakri's description of the Ghanaian court is the only one anybody has, and its details are the kind a merchant would have brought back because they were worth repeating. The king he names as reigning in 1068, reconstructed from what merchants who had actually been there told him, was Tunka Manin - the ruler this whole description of the audience, the army and the dogs in their gold collars is actually of.

The same author gives the king two hundred thousand men, forty thousand of them archers, and describes dogs at the audience wearing collars of gold and silver.

Two hundred thousand should be read as many. Arabic geography about distant places produces round numbers of that kind routinely, and no eleventh-century West African state fielded two hundred thousand men.

What the passage is useful for is what it implies rather than what it counts. An army described as a proportion of a population, rather than as a number of men on a payroll, is a levy — and Ghana's revenue is described everywhere in customs receipts and gold royalties and nowhere in wages. The king could call people out. He was not paying them to stand about.

The sister's son

aside?What aside meansDrafted from (eds), Fontes Historiae Africanae, Series Arabica IV (1981)

Al-Bakri, writing at Córdoba in 1068 from what merchants told him, sets out how the kingdom of Ghana passed its throne, and gives the reason he was given.

The king is succeeded not by his own son but by his sister's son, because a man can be certain that his sister's child is of his own blood and cannot be certain of his wife's.

What that is actually solving

It is a serious answer to a real problem, and it is worth taking on its own terms rather than as an exotic detail.

Every hereditary system has to decide what it is tracking. A patrilineal one tracks a claim that cannot be verified: it rests on the fidelity of the ruler's wife, which nobody can guarantee and everybody can allege. A matrilineal one tracks something observable — the king's sister is his sister, and her children are demonstrably her children.

The cost is that a king does not hand the throne to his own son. That requires a court to accept that the ruler's children are not the ruler's heirs, which most dynasties find intolerable, and which produces a permanent tension between a reigning king who wants his son to succeed and a rule that says his nephew will. A good many West African successions turn on exactly that fight.

Ibn Battuta finds it, and disapproves

Three hundred years later and further east, Ibn Battuta came into the western Sudan and met the same arrangement among the merchants of Walata, and it irritated him considerably:

(eds), Fontes Historiae Africanae, Series Arabica IV (1981) · Kurt Dundy · CC BY 3.0

No one is named after his father, but after his maternal uncle; and the sister's son always succeeds to property in preference to the son: a custom I witnessed nowhere else, except among the infidel Hindoos of Malabar. But these are Mohammedans, who retain their prayers by memory, study theology, and learn the Koran by rote.

That is the whole of his objection. He can accept the practice among Hindus and cannot accept it among Muslims who know the Quran by heart, because Islamic inheritance law is explicit and this is not it.

He went on to be scandalised by the women, who did not veil, prayed alongside the men, and kept male friends whom their husbands did not mind. He remonstrated with a judge he had befriended about a young woman sitting alone with him; the judge said this is our custom, nor is there any suspicion from our being in society together; Ibn Battuta says he did not benefit by the advice and that he did not visit him again.

Where else it turns up

Ghana in the eleventh century, Mali in the fourteenth, Makuria on the Nile from the seventh to the fourteenth, Elam in the third millennium BC, and the Asante on the Akan goldfields into the twentieth century.

Five instances, on two continents, spread across four thousand years, with no possibility of contact between the earliest and the rest. The convergence is not diffusion. It is several societies independently deciding that the reliable question is who is this woman's child rather than who is this man's.

What it says about the state

One thing worth drawing out. A rule of that kind cannot be improvised, and it cannot be enforced against a determined king without an institution behind it — somebody who says whose turn it is and is obeyed.

Al-Bakri is describing an eleventh-century state with a stated constitutional rule about the succession, running in a country that had no writing of its own. The rule was known well enough that traders crossing the Sahara could repeat it accurately to a geographer in Spain who never went anywhere near it.

The conquest that did not happen

contractionDrafted from Levtzion, Ancient Ghana and Mali (1973)

For most of the twentieth century the standard account of this polity's ending was that the Almoravids took Kumbi Saleh in 1076, forced the kingdom to Islam, and broke it — and that everything afterwards is a decline into the arms of Mali.

It probably did not happen.

What the account rested on

One late source, and a habit of reading.

The claim traces to Ibn Khaldun, writing three hundred years after the event, and to a reading of earlier notices that does not survive being looked at closely. Al-Bakri, writing in 1068 — eight years before the supposed conquest — describes a functioning kingdom. The contemporary Almoravid material does not claim the conquest. Archaeology at Kumbi Saleh has produced no destruction layer of the right date.

What the earlier evidence supports is something less dramatic and more interesting: pressure, alliance, commercial competition, and a conversion of the Ghanaian ruling house to Islam somewhere in this period that appears to have been negotiated rather than imposed. The relationship between the Almoravids and Ghana looks, on the surviving evidence, like two powers with a shared interest in the same trade routes rather than a holy war.

Why it is still worth naming

Because the non-event shaped the field for a century.

An Almoravid conquest of Ghana in 1076 is in a great many books. It supplied a very tidy causal story — Berber jihad destroys pagan African empire, Mali picks up the pieces — and the tidiness is a good part of why it lasted so long. Nobody had to explain how an empire ends without a conquest, because a conquest had been supplied.

The dismantling is largely the work of the 1970s and after, and Levtzion is one of the people who did it.

What the story replaced is the more interesting account, and it is the one this kingdom actually needs: a state built on a position in a trade, in a country where the rains were shifting, losing its grip on its tributaries over about a century, while a nomadic power with the same commercial interests grew stronger to the north and a new one grew stronger to the south.

Sosso

endDrafted from Levtzion, Ancient Ghana and Mali (1973)

What ended Ghana was not the Almoravids and not a single event. It was a tributary that stopped being one.

Sumanguru's kingdom of Sosso took Kumbi in the first years of the thirteenth century, more than a hundred and thirty years after the conquest that did not happen. By then the empire al-Bakri described had been shrinking for a century: provinces detaching, the Saharan trade rerouting eastward to newer termini, the southern goldfields shifting from Bambuk to Bure, further from Ghana and closer to the country that would be Mali.

The version that survives

Almost everything known about Sosso comes from the oral tradition of its enemies.

Sumanguru is the sorcerer-king of the Sundiata epic, the tyrant Mali's founder overthrew at Kirina in about 1235 — a figure with an iron-bodied invulnerability and a magical balafon, defeated by an arrow tipped with a white cockspur. That is not a chronicle. It is the founding story of the state that beat him, performed by griots whose hereditary function was to perform it, and Sumanguru is in it because a founding story needs somebody to have been founded against.

Whether Sosso was a serious power that briefly held the western Sudan, or a Ghanaian successor state that Mali's tradition inflated to make its own victory worth having, cannot be settled. Both are possible and the sources for distinguishing them do not exist.

What Mali took

Not just the territory, which was in any case not the valuable part.

It took the trade position: the two directions of the salt-and-gold road, the customs posts at the desert edge, and the royalty on nuggets that al-Bakri had described in 1068. Every element of Ghana's fiscal system passed intact to Mali, and from Mali to Songhai, and went on working for four hundred more years under three dynasties and four capitals.

Which is the honest way to describe this ending. The state stopped. The thing the state had been built to do carried on without interruption, because the gold and the salt did not care who was collecting the duty.

The name

afterlifeDrafted from (eds), Fontes Historiae Africanae, Series Arabica IV (1981)

On 6 March 1957 the Gold Coast became independent and took the name Ghana.

(eds), Fontes Historiae Africanae, Series Arabica IV (1981) · British Government · Public domain

The country is eight hundred kilometres from where the kingdom was, on the Atlantic coast rather than the Sahel, and was never part of it in any sense anybody has demonstrated. The choice was deliberate and it was an argument: that the new state's ancestry ran back to an African empire that had traded gold with the Mediterranean world while Europe was in its dark ages, and that the colonial name — the Gold Coast, which described what Europeans had come for — was not going to be kept.

That is the most consequential thing this kingdom ever did, and it did it nine hundred years after it stopped existing, by having had a name worth taking.

The name it did not have

Which is the joke underneath it, and al-Bakri supplies it.

Ghana was the ruler's title, not the country's name. The kingdom was called something else — Wagadu, in the Soninke tradition — and the Arabic writers who recorded the title as though it were a place did so because that is what the merchants they questioned said when asked whose country it was.

So a twentieth-century republic named itself after a medieval empire, using a word that was never the empire's name, transmitted by geographers in Spain and Sicily who had never been anywhere near it, from traders who were reporting what they called the man in charge.

What is actually left

Very little, and none of it in the ground where the country now is.

There is a stone-built town at Kumbi Saleh in southern Mauritania, excavated in stages since the 1910s, with a mosque, a cemetery and a merchant quarter of two-storey houses. It is probably the Muslim half of al-Bakri's double capital. The royal half — the town six miles away with the sacred grove and the king's tombs — has not been found, and may not survive.

There is no Ghanaian document of any kind. No inscription, no coin with a legend, no king-list, no chronicle. Everything anybody knows was written by people who were not there, in Arabic, at Córdoba and Palermo and Cairo, out of what merchants told them.

And al-Bakri

The last thing to say about this kingdom is that it exists because of one man in Andalusia who never travelled.

Al-Bakri finished his geography in 1068 and it is the single most important text about West Africa before the sixteenth century. His account of Ghana — the double capital, the audience, the taxes on salt, the royalty on nuggets, the sister's son, the dogs in gold collars — is not corroborated by anything except itself, and it is detailed, internally consistent, and specific in exactly the way invention is not.

A kingdom left no records, was destroyed by a state that left no records, and is known because a Spanish Muslim who never crossed the Sahara wrote down carefully what the men who did told him.