Lesson 2. How Historians Weigh Evidence

Last lesson you learned that almost everything ahead is historical reasoning — the best explanation of the evidence, not a mathematical demonstration. Now you meet the tools that do that weighing. By the end you will be able to name the historian's five questions, spot the counting error that ruins arguments on both sides, and say what the criterion of embarrassment picks out.

Historians do not have laboratories. They cannot repeat the past — you cannot run the year 30 again and watch more closely — so they reason from what the past left behind.

Four kinds of leftovers do most of the work: documents that people wrote, objects that people made and used, inscriptions carved where everyone could read them, and the testimony of people who were there.

Faced with a claim about the past, a historian works down a short list.

• How early is the source? Closer to the event usually means fewer links in the chain.
• How many independent sources are there? Independence matters more than number.
• Did the writer have a reason to lie? And does the account keep material that works against that reason?
• Does it fit what else we know? Places, officials, customs, dates.
• Is there hostile corroboration? An opponent conceding a fact is worth more than a friend asserting it.

Look hard at the last of the five. A hostile writer has little motive to strengthen the other side's case. So when one grants a point anyway, the concession is hard to explain except by its being true — strong evidence, though not a guarantee.

These are the ordinary tools of ancient history. They are applied to Homer and to Tacitus exactly as they are applied to the Gospels. Nothing in the units ahead asks for a special rule to be made for Christian sources — that request is itself a weakness, and a fair critic would say so.

Suppose five newspapers report the same story. If four of them copied the fifth, you do not have five witnesses. You have one witness and four repetitions.

This is one of the most common errors in popular argument, and it turns up on both sides of many questions. Counting sources feels like weighing evidence. It is not.

That error matters here directly. Matthew, Mark and Luke share so much material, sometimes word for word, that literary dependence between them is the standard scholarly explanation.

So an argument that counts the Gospels as four fully independent witnesses will be corrected by the first informed person who hears it — and once you have been corrected on that, everything else you said looks shakier than it is. Conceding the point up front costs you nothing and keeps the rest of your case standing.

Now guard the other flank. Dependence between documents does not make them worthless.

A source that another writer trusted enough to use is itself evidence — somebody close to the material judged it reliable. And the places where two accounts diverge are exactly where independent information may be showing through. Copying explains the overlap; it does not explain the differences.

Someone will object that the Gospel writers were believers, so of course they wrote favourably.

The observation is correct. Agree with it straight away. The real question is what follows from it — and the answer is much less than the objector assumes.

If testimony only counted when the witness was indifferent, we would have to throw away most of what we know about the ancient world. People generally write at length about what they care about.

So a historian's response to an interested witness is not dismissal. It is cross-examination.

• Does the account keep material that is awkward for the writer's own case?
• Does it name checkable people, places, offices and customs?
• Where it differs from a source it used, do the differences look like independent memory, or like tidying up?

The first of those three questions has a name: the criterion of embarrassment. It picks out details a writer would have had reason to leave out, preserved anyway.

Here is a frequently cited example. All four Gospel accounts name women as the first to find the tomb empty — in a setting where women's testimony carried little formal weight. Many scholars argue that a group inventing a persuasive story for that audience would not open with witnesses that audience could wave away.

Say a classmate tells you, "Three people told me Mr. Alvarez is quitting at the end of the year." That sounds solid — three sources. But ask the historian's first question about number: are they independent?

You ask a follow-up and learn all three heard it from the same kid, Marcus, at recess. That changes everything. You do not have three witnesses. You have one witness — Marcus — and three repetitions, just like the four newspapers that copied the fifth.

Now imagine instead that one of the three heard it from her older sister, who works in the school office and never spoke to Marcus at all. That is a different, stronger case. A genuinely separate source is worth more than five copies of the same rumour, however confident each copy sounds.

Before repeating either version, a careful listener always asks the same follow-up question first: where did you actually hear this, and did that person hear it from someone else too? That one question is doing all the work.

The independence rule works because copying cannot manufacture new information. A photocopy of a photocopy tells you nothing a single copy did not already tell you.

What the rule shows: when separate, uncoordinated sources agree, that agreement is real evidence, because neither one could have produced the match by copying the other.

What the rule does not show: it does not tell you the sources are correct. Ten independent people can all be wrong about the same event if they were all fooled the same way — by poor lighting, or a fast-moving scene. Independence rules out copying as the explanation for agreement. It does not by itself rule out a shared mistake.

That is why a historian still runs the other four questions too — how early, what motive, does it fit, is there hostile corroboration — rather than stopping once independence is confirmed. One strong test is never the whole method.

A sharp classmate might object: "But you said the Gospels aren't fully independent, since three of them share material. Doesn't that wreck the whole argument?"

That is a fair point, and this lesson already agreed with half of it: yes, Matthew, Mark and Luke share material, and denying that would be a mistake.

Here is the reply. Shared material does not make a source worthless — a document another writer trusted enough to use is itself evidence that somebody close to the material judged it reliable. And the places where accounts differ are exactly where independent information may be showing through, since copying explains the overlap but not the differences. So the objection is right that these three are not fully independent. It is wrong to conclude from that alone that nothing here counts as evidence.

Notice the shape of that reply: agree with the true part first, then correct only the part that does not follow. That is the pattern worth remembering more than any single fact in this lesson.

Say each term out loud, then use it in one sentence about this lesson.