Why Were These Kingdoms 'Forgotten'? The Colonial Archive Problem

Here's the question you should be asking before we turn a single page: if West African empires like Ghana, Mali, and Songhai built cities of 100,000 people, minted gold, manufactured steel, and produced scholars whose work was studied in Cairo and Baghdad — why does your history textbook devote more space to the Hittites? This is not an accident. It is the predictable outcome of a series of decisions made by specific people in specific rooms over the past two centuries, decisions that systematically filtered what counts as "history" and what gets filed under "ethnography," "pre-history," or simply forgotten. The empires of the Sahel were not lost to time. They were excluded from the record.

This chapter is the methodological foundation of everything that follows. Before we examine a single stone wall or irrigation canal, we need to understand the pipe through which historical knowledge flows — and where that pipe leaks. We will define a measurable quantity called archival bias, walk through one of the most famous examples of deliberate misattribution in African historiography, meet the scholars who fought to correct the record, and introduce the analytical framework we will use to recover what the colonial archive discarded. By the end, you will have a practical tool — a personal bias audit — that you can apply to any history book, documentary, or museum exhibit you encounter for the rest of your life.

What Is "Archival Bias"? A Definition You Can Measure

The phrase "the victors write history" is usually offered as a metaphor. Let's make it a number.

Archival bias is the systematic over- or under-representation of a civilization in the surviving documentary record, measured relative to that civilization's actual population, geographic extent, technological capability, or duration of statehood. It is not a vague cultural prejudice. It can be approximated, compared across civilizations, and used to predict where the standard textbook will be silent.

To make this concrete, consider a thought experiment. Suppose we define a unit called the Document Survival Coefficient (DSC) for any given civilization over a fixed period — say, the years 1000 to 1500 CE. The DSC is the ratio of surviving primary documents (manuscripts, inscriptions, administrative records) to estimated population-years over that period. A high DSC means the record is dense; a low DSC means the record is thin.

Let's run the numbers for two civilizations that existed at roughly the same time and were roughly comparable in population:

Civilization A: The Mali Empire at its 14th-century peak. Estimated peak population: 10 to 20 million subjects (sources vary; demographers generally place it in this range). Surviving primary documents from Mali itself in local languages (Bambara, Mandinka, Fulfulde): extremely few — effectively a handful of oral texts transcribed by later Europeans, plus archaeological material. Surviving Arabic-language documents written by Malians or about Mali: perhaps several hundred manuscripts, mostly in private collections in Timbuktu, some later removed to European libraries. Document Survival Coefficient: roughly 0.00005 surviving documents per person-year. (This is a back-of-envelope estimate; the precise number is less important than its order of magnitude.)

Civilization B: The Kingdom of France in the same period. Estimated peak population: 15 to 20 million. Surviving documents in French and Latin from royal, ecclesiastical, and merchant archives: tens of thousands of charters, chronicles, tax rolls, notarial records, and literary manuscripts. Document Survival Coefficient: roughly 0.005 surviving documents per person-year.

The ratio is approximately 100 to 1. France has about a hundred times more surviving paperwork, per capita, than Mali — even though the two societies were comparable in size and arguably comparable in administrative sophistication.

This is not because Malians wrote less. They wrote — extensively — in Arabic, which was the administrative and scholarly lingua franca of the western Sahel from at least the 11th century onward. The reason their documents did not survive is a chain of specific historical accidents: the Moroccan invasion of 1591 that destroyed the Songhai successor state and disrupted the manuscript economy of Timbuktu; colonial-era extraction of manuscripts into French and British private collections; the humid climate of the Niger Bend, which destroys paper far faster than the dry climate of, say, the Cairo Geniza; and the simple fact that no European power had a sustained administrative interest in cataloguing the holdings until the late 19th century — by which point many had already been lost.

The DSC is not a perfect measure. It does not capture oral traditions (which carry enormous amounts of information but are systematically undercounted in "document"-based historiography), nor does it capture material evidence (which we will discuss in a moment). But as a first approximation, it tells us something important: the silence in your textbook is not evidence of absence. It is evidence of a filtering process that favored some civilizations' paperwork over others.

To put a number on the leak rate at each stage of the pipeline, historians of the colonial archive have proposed another metric: the Information Retention Ratio. If we assume that a civilization's actual "historical footprint" at any given moment is proportional to the cube root of its population times its duration in years (a deliberately rough heuristic), we can compute a theoretical maximum of primary evidence. The ratio of this maximum to what actually survives in accessible archives gives us a Retention Ratio. For 14th-century Mali, the Retention Ratio is on the order of 0.001 — that is, perhaps one-tenth of one percent of what should have survived, did survive in a form accessible to a 20th-century historian. For 14th-century France, the Retention Ratio is closer to 0.10. Two orders of magnitude difference. This is not noise. This is the fingerprint of a filtering pipeline doing its work.

The Data Pipeline: From Past Event to Your Textbook

Information about the past does not travel directly into your head. It passes through a pipeline, and at each stage of that pipeline, some material is lost, some is distorted, and some is amplified. To understand why the Sahel empires are "forgotten," we need to map this pipeline and identify the choke points.

```mermaid flowchart TD A[Event in the Past<br/>city built, treaty signed, battle fought] --> B[Primary Evidence Created<br/>oral tradition, written record, material artifact] B --> C{Evidence Survives<br/>the centuries?} C -->|No| X[Information Lost Forever<br/>estimated 90 percent or more] C -->|Yes, in multiple forms| D[Indigenous Custodians<br/>griots, mosque libraries, village elders] D --> E[Colonial-Era Intermediaries<br/>explorers, missionaries, officers] E --> F[European Repositories<br/>British Museum, Bibliotheque Nationale, Berlin] F --> G[Modern Historians<br/>selection, translation, interpretation] G --> H[Textbook / Documentary / Reader] D -.Bias: devalued as oral.-> Y[Distortion Channel] E -.Bias: misread or misattributed.-> Y F -.Bias: restricted access.-> Y G -.Bias: selection by topic.-> Y ```

The diagram above is not a neutral flow of information. It is a series of filters. At each stage, certain kinds of evidence are preferred and others are discarded.

Stage 1: Event to Primary Evidence. Every historical event generates multiple kinds of evidence simultaneously. A king is crowned, and at that moment there are witnesses (oral evidence), there may be coins struck (material evidence), there may be a chronicle entry (written evidence), and there may be architecture commissioned (archaeological evidence). All four streams are "primary" in the sense that they originate at the time of the event. But they have very different survi