When the Greek historian Herodotus tells us that Xerxes invaded Greece with 1,700,000 soldiers, what should we do? Believe him? Laugh? Reach for a calculator and start estimating the logistics of feeding that many people in 480 BCE? (Spoiler: it's impossible. The army would have drunk rivers dry.)

Welcome to one of the most peculiar problems in historical research: the numbers are almost always wrong. Ancient chroniclers exaggerated, medieval scribes copied carelessly, and modern propagandists massaged statistics for political ends. Yet historians can't simply throw these figures away. Instead, they've developed a clever toolkit for working with numbers that are unreliable but not useless. Let's see how they do it.

Numbers in historical sources are rarely innocent. When a Roman emperor's inscription claims he killed 50,000 enemies in a single battle, the figure isn't really about counting bodies—it's about projecting power. The number is a flex, not a fact. Historians have learned to read these inflated figures the way we might read a job applicant's resume: with appreciation for the artistry of exaggeration.

The same logic works in reverse. Defeated kings shrink their losses; persecuted minorities sometimes inflate their suffering; victorious generals balloon enemy casualties to make their wins look heroic. A medieval chronicler reporting that only 100 of his side died while 10,000 enemies fell isn't lying exactly—he's performing loyalty. The ratio (1:100) tells us about the chronicler's worldview, not the battlefield.

So the first question a historian asks isn't "Is this number correct?" but "Why was this number chosen?" Who benefits from this figure being big? Who benefits from it being small? Once you spot the political pressure shaping the digits, the source becomes useful in a different way: as evidence about belief, ideology, and audience expectations.

Takeaway

A wrong number can still tell the truth—just not the truth it pretends to. Ask who benefits from the figure before asking whether the figure is right.

Here's a useful trick from the historian's workshop: even when a specific number is fantasy, its order of magnitude might still be informative. If a source says an army had 100,000 men and the real figure was probably 30,000, that's still a different beast than an army of 3,000. We've moved from "definitely wrong" to "roughly the right neighborhood."

Historians cross-check magnitude using practical constraints. How much grain could pre-modern agriculture produce? How many wagons could a road support? How quickly could messengers travel? These physical limits act like guardrails. An army of two million can't march through a mountain pass that physically can't accommodate it—so the chronicler's claim must be wrong, but his army was probably still large by the standards of his world.

This is why historians often write things like "perhaps 20,000 to 40,000 soldiers" instead of giving a single number. It looks wishy-washy, but it's actually more honest. A range acknowledges uncertainty while still saying something meaningful. Precision and accuracy aren't the same thing—and a fuzzy answer that's roughly right is far more useful than a sharp answer that's spectacularly wrong.

Takeaway

Precision is seductive but often deceptive. A confident range built from logical constraints beats a single number plucked from a propaganda inscription.

Here's the historian's secret weapon: even when absolute numbers are garbage, relative numbers often survive intact. A medieval tax record might wildly misreport the actual population of a town, but if it consistently uses the same flawed methods across many towns, the comparisons between them remain meaningful. Town A being twice the size of Town B is probably true, even if neither figure is reliable on its own.

Ratios and proportions are remarkably robust. If a chronicler systematically inflates casualty figures, he probably inflates them on both sides. So while "50,000 dead" might be wrong, the claim that one side lost three times as many as the other could still be roughly accurate. The bias cancels out—at least partially—when you compare numbers from the same source.

This is why economic historians love price series, demographers love sex ratios, and military historians love casualty proportions. These relational figures bypass the worst distortions of historical record-keeping. You don't need to know exactly what a loaf of bread cost in 1450 to see that bread prices doubled between 1450 and 1460. The trend tells the story even when the numbers fib.

Takeaway

When in doubt, look at relationships rather than absolutes. Ratios survive what raw figures cannot, because systematic biases tend to cancel out across comparisons.

Historical numbers are almost always wrong—but that doesn't make them useless. They're evidence of mindset, magnitude, and relationship, even when they fail as literal counts. The historian's job isn't to find perfectly accurate figures (those rarely exist) but to interrogate flawed ones intelligently.

Next time you encounter a confident statistic from the past, try asking: Who chose this number, and why? What's the order of magnitude? What's it being compared to? You'll find that learning to read numbers skeptically doesn't make history less knowable—it makes it more honest.