How to evaluate historical sources: primary vs secondary, bias and corroboration

TUREFF• •0

Every claim about the past rests on a source, and sources are not equally trustworthy. Historians spend much of their training learning to question documents before believing them. The same habits work for anyone reading history, checking a family story or weighing a viral "on this day" post.

Primary, secondary and tertiary sources

Primary sources were created at or close to the time by people involved: letters, diaries, court records, census returns, photographs, newspapers, maps, official reports, objects and buildings.
Secondary sources interpret primary sources after the event: history books, journal articles, documentaries.
Tertiary sources summarise secondary work: encyclopaedias, textbooks and most websites.

None of these is automatically better. A primary source can be mistaken or deliberately false, and a good secondary source may have weighed hundreds of primary sources you will never see. The categories tell you what kind of questions to ask, not whether to trust it.

Five questions for any source

Who made it? What was their position, and what did they stand to gain or lose?
When? A diary written the same evening differs from a memoir written thirty years later, when memory has blurred and reputations need protecting.
Why? Was it meant to inform, persuade, flatter a patron, win a lawsuit or simply keep accounts? Purpose shapes content.
For whom? A private letter, a speech to supporters and an official report on the same event will tell three different stories.
How do we know it is genuine? Where has it been kept, and is its history of ownership, its provenance, known?

Bias is information, not a reason to discard

Every source has a point of view. A propaganda poster is an unreliable guide to what happened on a battlefield but an excellent guide to what a government wanted people to believe. Ask what the bias tells you, and what the author could not avoid revealing. Incidental details, such as prices mentioned in passing or the furniture in a complaint about a neighbour, are often more reliable than the main argument because nobody had a reason to distort them. Also notice silence. Records were mostly made by the literate, the powerful and the state. The poor, women, enslaved people and colonised peoples are often visible only through others' eyes, so absence from the record is not absence from history.

Corroboration

One source is a lead; several independent sources are evidence.

Look for accounts that do not depend on each other. Five newspapers repeating the same news agency report count as one source, not five.
Compare different kinds of evidence: does a written account match the archaeology, the weather records or the accounts book?
Where sources disagree, ask which was better placed to know and which had more reason to shade the truth.

Forgeries and how they are caught

Two well-documented cases show the methods. In 1440 the scholar Lorenzo Valla argued that the Donation of Constantine, a document in which the emperor supposedly handed authority in the West to the Pope, could not be genuine because its Latin and its terms belonged to a much later period. Most scholars now date it to the 8th century. In 1983 the German magazine Stern published extracts from the so-called Hitler Diaries; forensic tests by German federal agencies soon showed the paper contained optical brighteners and the bindings materials that were not in use until after the Second World War. Language, materials and provenance remain the three main tests.

Practical habits

Keep a note of exactly where each fact came from, including page numbers.
Distinguish what a source says from what you infer from it.
Be most sceptical of evidence that confirms what you already hoped to find.
When a claim online has no source at all, treat it as unverified until you find one.

Which source has surprised you most, or turned out to be less reliable than it looked? Share it in the comments.