Weighing Testimony: Marc Bloch and the Historian's Craft

How a historian weighs witnesses who may be sincere and still wrong, how false reports grow from what people already believe, and why an origin is not an explanation, after Marc Bloch's The Historian's Craft and his 1921 essay on the false news of the war.

In September 1917 a French raiding party on the Chemin des Dames brought back a German sentry, an older reservist from Bremen. The rumour that followed made him a shopkeeper from Braisne, a town a few kilometres away, who had come back in enemy uniform and could only be a spy. Marc Bloch, a historian serving in that sector, traced the story to a misheard name: Brême, French for Bremen, heard as Braisne. He published the case in 1921, in an essay on the false news of the war. The same historian later wrote The Historian's Craft. Read together, the two teach how to weigh a witness.

A book about how the past is known

The Historian's Craft opens with a boy asking his historian father what history is for (unverified against an authorised edition). Bloch died in 1944, and the French original, Apologie pour l'histoire ou Métier d'historien, appeared after his death, in 1949. Its chapters move from "History, Men, and Time" to "Historical Observation", with sections on evidence and how evidence is transmitted. Next comes "Historical Criticism", with sections on the history of the critical method, the pursuit of fraud and error, and a logic of the critical method. The book ends with historical analysis and causation. The Level that assigns the book asks for the chapters on observation and criticism; this page gives the questions to carry into them.

There is no good witness

In 1921 Bloch called the psychology of testimony a very young science, barely more than twenty years old, and credited it with helping historical criticism. Its lesson, in his words: "Il n'y a pas de bon témoin ; il n'y a guère de déposition exacte en toutes ses parties." In English: there is no good witness, and hardly any testimony is exact in every part.

That does not make witnesses worthless. Bloch went straight on to the real question: on which points does a sincere witness, one who believes he is telling the truth, deserve belief? He called it infinitely delicate, with no answer that can be fixed in advance. Each case has to be judged. Established as Bloch's position; the experimental work on memory and testimony since 1921 is not reviewed here.

How a false report is born

The Braisne story shows the mechanism. The mishearing was an accident. The spy was not. Belief in German espionage had become, Bloch wrote, a kind of dogma with almost no unbelievers. Only the first incident was chance; the story took off because imaginations were already prepared. His general rule: "Une fausse nouvelle naît toujours de représentations collectives qui préexistent à sa naissance." A false report is always born of collective beliefs that existed before it. He called it a mirror in which a group sees its own features.

He found the same pattern in a larger case. The Belgian sociologist Fernand van Langenhove had studied the legends of Belgian francs-tireurs, civilian snipers, among German troops. Holes left in house fronts for plasterers' and painters' scaffolding were read as loopholes for snipers. A soldier who expected snipers saw them in the walls.

Bloch also named the conditions that weaken judgment. "On croit aisément ce que l'on a besoin de croire": people readily believe what they need to believe. Methodical doubt, he wrote, is usually a sign of good mental health, so strong emotion and fatigue erode it. Censorship pushed news back into word of mouth, and the field kitchens became the "agora" of the trenches, where rumours were traded.

A routine for weighing two accounts

For each point on which two accounts disagree, ask:

Was the witness placed to see it, or repeating someone else?
How soon was it written down, and through how many mouths did it pass?
What did the witness expect or fear beforehand?
Do independent witnesses agree, and are they really independent?
Would the error, if it is one, fit a belief the witness's group already held?

This routine is the site's own condensation of Bloch's cases, not a list he gives: Provisional. The first question is the one Observe, Define, Represent trains: the soldiers heard a sound; the spy was an interpretation. Peirce argued that clinging to a belief breaks down when you meet people who think differently (Peirce on the Fixation of Belief). In the trenches, Bloch found almost no one who did.

Origins are not causes

Causes in history raise a different trap, which Bloch called the idol of origins. The word "origins" can mean a beginning or a cause, and the two meanings blur. In popular usage, he says, an origin is a beginning that explains, and worse, a complete explanation (paraphrase; unverified against an authorised edition). Knowing when something began does not show why it happened or why it lasted. In the same chapter he argues that understanding needs a range of comparison, and that the present helps explain the past as the past explains the present.

History cannot run an experiment on the past. A claim that one event caused another is a counterfactual, "had this not happened, that would not have", argued by comparison with other cases. The page on Cause and Effect sets out what such a claim must assume.

The arts behind it

Weighing testimony is old work for the trivium. Grammar establishes what a document actually says. Dialectic sets witness against witness. Rhetoric notices how a story was shaped to persuade, and to whom. Modern inquiry adds the question of transmission: how the report travelled to you. Lateral reading (Checking Sources) is the same discipline applied to the web. So is an outage report at work: Technical Writing as Reasoning asks its writer to keep observation and inference apart, because the next reader has to weigh it as testimony. In the site's five acts (its own synthesis, not a classical scheme), this page trains examining reasons and checking against evidence.

Where this could be wrong

The strongest objection. "There is no good witness" can harden into blanket suspicion, and blanket suspicion is as lazy as blanket trust. The routine also assumes you can tell which witnesses are independent, which is often the hardest fact to establish.

The best reply. Bloch's own question rules out blanket doubt: the task is to find the points on which a sincere witness deserves belief, not to discard witnesses. Where independence cannot be established, say so, and weigh the agreement less.

What would settle it. Test the routine on events whose truth was later fixed by documents or physical evidence, and see whether readers using it pick the more reliable account more often than readers without it. No such test is cited here, so the routine stays Provisional.

Try this

Compare two accounts of the same event, such as two newspapers or two memoirs. Judge which is more reliable on three points, giving the reason for each judgment from the routine above.
Write the history of one rumour: the first incident, the belief it met, and the route it travelled.
Take a claim about the origin of a word, custom or tool. Separate when it began from why it spread.
These are the practice steps for the Level on weighing testimony (the Levels path). Note which of the five questions changed your mind.