Every headline that announces a bog body, a scrap of linen or a hearth "dated to" a precise year rests on a measurement that is far messier than the round number suggests. Radiocarbon dating is one of the twentieth century's most useful inventions, developed by Willard Libby at Chicago in the late 1940s, and it underpins most of what archaeology claims to know about the last fifty millennia. It is also hedged with limits that rarely survive the trip from the laboratory to the news bulletin.
The physics starts in the upper atmosphere. Cosmic rays striking nitrogen-14 produce neutrons that convert some of it into carbon-14, a radioactive isotope. That carbon-14 oxidises into carbon dioxide, mixes through the atmosphere, and is drawn into plants by photosynthesis and into animals by eating those plants. While an organism lives it keeps topping up its carbon-14 to match the atmosphere. The moment it dies the intake stops and the clock starts, because the trapped carbon-14 decays back into nitrogen at a fixed rate. The half-life is 5,730 years, meaning half of the carbon-14 in a sample is gone after that span, a quarter after two spans, and so on. Measure how much remains against the stable carbon-12, and in principle you can read off how long ago the thing stopped living.
In practice, the measuring has moved on. Early laboratories counted the faint radioactive clicks from a sample and needed grams of material. Since the 1980s most work uses accelerator mass spectrometry, which counts the carbon-14 atoms directly, needs only milligrams, and can handle a single seed or a thread from a shroud. The Oxford Radiocarbon Accelerator Unit and the Scottish Universities Environmental Research Centre at East Kilbride are the two best-known British facilities, and both quote results with an uncertainty attached rather than a bare number.
Why a radiocarbon year is not a calendar year
Libby assumed atmospheric carbon-14 had stayed constant through time. It has not. Solar activity, the Earth's magnetic field and the vast slug of old, carbon-14-free carbon pumped out by burning coal and oil since the Industrial Revolution have all shifted the baseline. Nuclear weapons testing in the 1950s and 60s roughly doubled atmospheric carbon-14, a spike that ironically now helps date very recent samples. Because the starting concentration wobbled, a raw radiocarbon measurement has to be translated into calendar time using a calibration curve.
That curve, the internationally agreed IntCal series most recently revised as IntCal20, is stitched together from things whose ages are known independently: tree rings counted back year by year, layered lake sediments, corals and cave formations. A laboratory feeds its raw figure into calibration software such as OxCal and gets back a range, not a point, often expressed at 95 per cent confidence. Where the curve is flat or kinked, a single measurement can smear across two or three centuries, which is why some periods are notoriously hard to pin down no matter how good the equipment.
Contamination is the other constant enemy. A whisker of modern rootlet, a smear of preservative from a museum shelf or fungal growth can inject fresh carbon and make a sample look far younger than it is, while old carbonate from groundwater can push it the other way. Marine organisms are a special problem: the deep ocean holds carbon that left the atmosphere centuries earlier, so shellfish, fish and anything that ate them return ages hundreds of years too old unless a reservoir correction is applied. The same effect afflicts freshwater fish and the people or animals that relied on them.
Then there is the ceiling. After about ten half-lives so little carbon-14 survives that background noise swamps the signal, which puts a practical limit near 50,000 years. Anything older, a Neanderthal site or a dinosaur bone, is out of reach and needs a different clock entirely, such as potassium-argon or uranium-series dating on the surrounding rock. Carbon dating also says nothing directly about stone tools, pottery or metal, because they never lived. Archaeologists date those by association, measuring charcoal from the same layer or organic residue clinging to a pot, and that inference is only as sound as the assumption that the two really belong together.
None of this makes radiocarbon dating unreliable. It makes it precise about a narrow question: when did this particular piece of organic material stop exchanging carbon with its surroundings? Read with its calibration range, its corrections and its ceiling in mind, the method is remarkably powerful. Stripped down to a single confident year in a headline, it promises a certainty the science never offered.