How this archive is built
Methodology
An archive that argues with the textbooks owes an account of its own working rules. These are ours.
The lower-bound rule
Where serious scholarly estimates disagree — and on colonial death tolls they almost always do — we publish the lower published end of the mainstream range. We do this even when higher figures are well supported. The argument we are making does not require the highest number; it requires the smallest honest one. If the most conservative published figure for the Congo Free State (about 10 million) is already monstrous, the case does not need 15 million to land.
When a single number is contested in either direction, we link the dispute and quote both sides rather than pick the higher figure.
Source hierarchy
We prefer, in this order:
- Peer-reviewed academic books and journal articles by historians, economists or political scientists.
- Primary documents — colonial-era reports, treaties, censuses, parliamentary records, court judgments, declassified state papers.
- UN, UNCTAD, World Bank, IMF and ILO data series, used with explicit notes on their limitations.
- Reputable long-form journalism (e.g. The Atlantic, Al Jazeera English, The Guardian long-reads) where it digests or extends the above.
- Wikipedia and Wikimedia, treated as starting points and as image archives — never as a primary citation.
Every chapter ends with a numbered Sources block listing the books, papers and primary documents specific to that chapter. Inline superscript citations point into that list.
Image provenance
Every photograph and artwork in this archive is drawn either from Wikimedia Commons (predominantly public-domain reproductions of colonial-era photographs) or from institutional archives credited at the caption. We do not use AI-generated imagery for historical events; the historical record does not need to be illustrated by a machine's guess.
Captions name the photographer, the date, the place and the institution holding the original where known. Where multiple sources hold a print, we link the one with the most complete provenance record. Schema.orgImageObject markup attaches the same provenance to the image programmatically, so search engines and image-indexing systems can read it directly.
What we call a colony
We use the word colony for any territory whose political and economic life was, or is, structurally controlled by an outside power for that outside power's benefit. By that test, we include the formal overseas territories that Europe and the US still hold today (Réunion, Mayotte, French Polynesia, Guam, Puerto Rico, the British Indian Ocean Territory and others) and we treat the CFA franc, the Commonwealth's residual constitutional links and the US base network as continuities rather than as innocent inheritances.
We do not pretend that this terminology is neutral. We argue for it openly throughout the archive.
Comparative history
We do not deny that other civilisations conquered, enslaved or slaughtered. The Why This Was Different chapter sets out the three axes on which European colonialism after 1492 is distinct from earlier empires (scale, racial codification, integration with industrial capital). The deaths under Genghis, Tamerlane and Assyrian conquest are part of the record. They are not a reason to let later atrocities off the hook.
Updates and corrections
Every chapter shows its last-updated date. Substantive corrections and revisions are logged publicly at the corrections page. If you find an error — a misattributed image, a wrong date, a number that has been revised in the literature since publication — please tell us and we will fix it and log the change.
What this archive is not
It is not an academic monograph. It is a documentary archive written for a general adult reader, with academic citations attached so anyone who wants to verify a claim can follow it back. It is not value-neutral: we think colonialism was a crime, that it has not ended, and that the comfortable nations of the world have organised their public memory around forgetting most of it. We say so.
In short
How does Unsilenced actually establish a figure or a claim before publishing it?
Every number on this site passes through a fixed hierarchy before it is allowed onto a page: peer-reviewed monographs and journal articles first, then primary records — censuses, mission and burial registers, shipping manifests, court judgments, parliamentary papers, declassified state files — then the findings of official commissions of inquiry (the Whitaker Report on the Herero and Nama, truth commissions, UN and ILO panels), then reputable long-form journalism that digests or extends the academic record, and only last, as a starting point and image archive rather than a citation, Wikipedia and Wikimedia. Oral history and community memory are used throughout, but as a check against archival silence rather than as a stand-alone source for a number, because the archives that would confirm or refute a survivor's account were often destroyed or never kept by the power doing the killing.
This hierarchy exists because colonial record-keeping was itself an instrument of the crimes it describes: population figures were compiled by administrations with a direct interest in undercounting forced labourers, famine dead or deported people, and many registers were never made or were deliberately burned as empires withdrew. Demographic reconstruction — comparing a pre-crisis population estimate to a post-crisis census, as Jan Vansina and later demographers did for the Congo Free State, or as the debate over Bengal 1943 and Ukraine 1932–33 requires — is therefore treated as an estimate with a stated method and a stated margin of error, never as a bare fact. Where a chapter states a single number, it is because a mainstream range has narrowed enough that a lower bound is defensible; where it has not narrowed, the chapter says so and gives the range.
Evidentiary tiers
How six kinds of evidence are weighted against each other
| Evidence type | What it is trusted for | Its known limitation |
|---|---|---|
| Peer-reviewed monograph or journal article | The headline figure or range cited on a chapter page | Can lag newer archival finds by years before revision |
| Primary administrative record | Dates, place names, quotas, orders, official correspondence | Compiled by the perpetrating power; systematically undercounts victims |
| Official commission of inquiry | Establishing that an atrocity occurred and its rough scale | Often decades delayed, politically negotiated, sometimes without prosecutorial teeth |
| Long-form investigative journalism | Recent developments, restitution disputes, named survivors | Not peer-reviewed; used only where it digests or extends academic work |
| Oral history and community memory | Detail invisible to any written record — routes, names, local toll | Not independently verifiable as a number; cross-checked, not counted alone |
| Wikipedia and Wikimedia | Locating public-domain images and as a first map of a topic | Never cited as the source for a factual or numerical claim |
The order used when sources on the same event disagree, not a ranking of importance in general.
The lower-bound rule
Why the smallest honest number, not the largest defensible one
Published estimates differ with records and reconstruction methods. For the Congo Free State, Hochschild popularised a population decline of roughly ten million; this is not a documented count of killings, and its scale and assumptions remain debated. Population change includes births and migration as well as deaths. State the source, period and what each estimate measures rather than presenting an upper bound as settled fact.
That discipline also forces a second distinction that is easy to blur: excess mortality is not the same claim as direct killing, and a chapter says which one a cited number measures. A famine death is very rarely a person shot; it is usually a person who died of disease or starvation because policy — grain requisitioning, export continuation during shortage, the closure of relief works, forced labour quotas — removed the means of survival that would otherwise have existed. Treating a famine's excess-mortality estimate as identical to a massacre's body count would flatten two different kinds of evidence and two different kinds of culpability into one number, so the two are kept visibly apart.
Contested ranges
How a live scholarly dispute is presented rather than resolved by fiat
Some ranges are not narrowing and are not going to: the population of the Americas before 1492, the scale of mortality among enslaved Africans during the Middle Passage beyond what the Trans-Atlantic Slave Trade Database can document from ship manifests, or the precise toll of famine policy in colonial India, remain genuinely disputed among specialists who have read the same archives and drawn different inferences from gaps in them. In these cases the archive does not manufacture false certainty by picking a number and presenting it as settled; it names the range, names the historians on each side of it, and explains what kind of evidence would narrow it further.
This is different from a dispute manufactured by denial literature, which typically works by inventing a methodology rather than by contesting one already in use in the peer-reviewed literature. The test applied here is simple: is the disagreement between historians using comparable evidentiary standards, or between a historian and someone rejecting the standard itself. Only the first kind is presented as an open scholarly question on these pages.
Terminology and bias
Why words like genocide, famine policy and colony are chosen deliberately, and how bias is checked
Terminology decisions are treated as claims requiring their own justification, not as neutral labels. 'Genocide' is used only where a chapter can point to evidence of intent to destroy a group in whole or in part, in the sense codified after 1948 and applied retrospectively by scholars to cases such as the Herero and Nama war of extermination; where intent is disputed or unproven, the archive uses more cautious language such as 'mass killing' or 'atrocity' and says why. 'Famine policy' rather than 'famine' is used where the historical record shows a state decision — export continuation, requisition quotas, the dismantling of relief infrastructure — materially worsened or caused a food-supply crisis, following the entitlement-failure model Sen set out for Bengal 1943 rather than treating famine as a natural disaster with no author. 'Colony' versus 'protectorate' is resolved by function, not by the label a metropole applied at the time: a nominally self-governing protectorate whose foreign policy, extraction and internal security were run from the metropole is described as a colony in substance.
Bias is controlled less by claiming neutrality, which this archive does not, than by making the controllable variables auditable: every figure is sourced to a named work a reader can find and check, every image's provenance is stated, every chapter's last-updated date is visible, and a public corrections process exists precisely so that an error of fact, as opposed to a disagreement with editorial judgment, can be found and fixed by someone other than the author. The editorial position — that this history is under-taught and its consequences ongoing — is stated openly rather than disguised as objectivity, so a reader can weigh it rather than be misled about its absence.
FAQ
Frequently asked
- Why does the archive use the lowest mainstream estimate instead of the most-cited one?
- Because the lowest published figure within a credible, peer-reviewed range is the hardest one for a good-faith critic to dismiss as exaggeration, and the argument this archive makes does not require a higher number to hold. Higher figures are noted and sourced where they exist; they are simply not the headline.
- How is an excess-mortality estimate different from a body count, and why does it matter?
- A body count records deaths directly caused by violence — a massacre, an execution, a bombardment. An excess-mortality estimate compares expected deaths under normal conditions to actual deaths during a crisis, capturing people who died of disease, starvation or exposure because a policy removed their means of survival. Both are real death tolls, but they require different evidence and support different claims about direct responsibility, so the archive never merges them into a single undifferentiated figure.
- What counts as a primary source here, and why is a colonial administration's own census treated cautiously?
- A primary source is a document produced at the time by a participant or witness — a census, a shipping manifest, a court record, a parliamentary transcript, a mission register. Colonial censuses are used, but flagged, because the administrations compiling them frequently had a direct interest in undercounting forced labourers, famine deaths or deportees, which is exactly why later demographic reconstruction by independent scholars is treated as more reliable than the original official count.
- Why does the archive sometimes present a range instead of a single number?
- Because for some events — pre-Columbian American population, the full toll of the Middle Passage beyond what shipping records document, several colonial famines — the peer-reviewed literature has not converged on a single figure and inventing false precision would misrepresent the state of the evidence. The archive states the range and names the scholars on each side rather than picking a number and calling the question closed.
- How does the archive decide whether to call an atrocity a genocide?
- It applies the post-1948 legal definition's core element — intent to destroy a national, ethnic, racial or religious group in whole or in part — as a historical question requiring evidence, not a rhetorical intensifier. Where scholars have documented that intent, as with the German colonial war against the Herero and Nama, the term is used and sourced to that scholarship. Where intent is disputed or the evidence is thinner, the archive uses a more cautious term and explains the distinction rather than defaulting to the strongest available word.
From the Archive
Wikimedia-sourced photographs and documents related to this page.
