How we know when things happened

Three kinds of date

The first kind is documented. A contemporary record — a treaty, a newspaper, a launch log, a patent — states when something happened, and other records corroborate it. These are the dates that can carry a day and a month.

The second kind is measured. No one wrote anything down, but the material itself carries a clock: decaying isotopes, annual growth rings, layered ice. These dates arrive with a stated uncertainty, and quoting one without it is a misuse of the result.

The third kind is inferred. A date is bracketed by other things: this pottery style, that stratum, a ruler's regnal years, an eclipse mentioned in a chronicle. Inferred dates are often good to a decade or a century and almost never to a year, which is why so much ancient history is written with 'about' in front of it.

When the documents disagree with the calendar

Even a well-documented date can be ambiguous, because the calendar underneath it changed. The Gregorian reform of 1582 removed ten days and altered the leap-year rule, and adoption was staggered across centuries — Britain and its colonies in 1752, Russia in 1918.

For events in between, sources may give an Old Style or a New Style date, differing by days or even by which year it was, since the year did not always begin on 1 January. Neither is wrong. What is wrong is quietly converting one to the other without saying so.

Radiocarbon, and where it stops

Living things take in carbon-14 while alive and stop at death, after which it decays with a half-life of about 5,730 years. Measure how much is left in bone, charcoal, textile or wood and you get a time since death. It is the workhorse of archaeology.

It has two well-understood limits. After roughly 50,000 years too little carbon-14 remains to measure reliably, so anything older needs a different method. And the atmospheric level of carbon-14 has varied over time, so raw results must be converted through a calibration curve built from independently dated material — which is why a radiocarbon date is properly quoted as a range with a confidence level, not as a year.

Counting layers, and the very long clocks

Some materials lay down one countable layer a year. Tree rings give dendrochronology, where overlapping patterns from living trees and old timber extend continuous chronologies back more than ten thousand years — and, at their best, date a felling to a single year and season. Ice cores and lake varves work on the same principle over longer spans.

Past that, radiometric methods take over, each suited to a different range: potassium-argon and argon-argon for volcanic rock over hundreds of thousands to billions of years, uranium-lead on zircon crystals for the oldest material on Earth. These are how the deep-time figures are established, and they are cross-checked against each other far more often than the confident single numbers in a textbook suggest.

Why a range is the honest answer

A game that demands a single year for every event has two options when the evidence will not supply one: pick a number and pretend, or represent the uncertainty. QuoAnno does the second. Events whose evidence supports only a range store that range, and any guess inside it scores in full.

It also affects how questions are written. If an invention involved several people over a decade, the question names a specific, datable milestone rather than asking when something was 'invented'. False precision in the wording is the same error as false precision in the data.