How Do Scientists Date Fossils? The Methods Explained

When you read that Homo erectus appeared around 1.9 million years ago, or that a Neanderthal site is 45,000 years old, it is fair to ask a blunt question: how do scientists date fossils that no human ever witnessed? The honest answer is that they almost never date the fossil itself. A bone or a stone tool carries no built-in date stamp. What carries a date is the rock, ash, sediment, or chemistry surrounding it — and the physics of how certain atoms and minerals change over time.

Different clocks tick at wildly different speeds, so no single method covers the whole of human evolution. A technique perfect for a 3,000-year-old hearth is useless on a 2-million-year-old skull, and vice versa. Below we walk through the main dating methods behind the ages you see across our fossil-sites map and species pages — what each one measures, how far back it reaches, and where it can go wrong.

Relative vs. absolute dating

Dating methods fall into two families. Relative dating tells you the order of events without attaching a number of years; absolute (or numerical) dating attaches an actual age in years. Both matter, and in practice they work together.

The oldest tool is stratigraphy — reading the layers of rock and sediment stacked at a site. Its founding rule is the principle of superposition: in an undisturbed sequence, deeper layers were laid down first, so lower is older and higher is younger. A skull found beneath a volcanic ash bed must be older than that ash. Layers can be folded, faulted, or eroded, so geologists read them carefully, but the principle underpins nearly every excavation.

Biostratigraphy refines this using index fossils — species that lived only during a known, short window and are common and widespread. Fossil pigs, elephants, and rodents are workhorses in East African hominin sites: if a layer contains a pig species known to exist between, say, 2.0 and 1.7 million years ago, anything in that layer falls in the same bracket. Relative methods order the story reliably, but on their own they cannot tell you whether a gap between two layers is a century or a hundred thousand years. For that you need a physical clock.

Radiocarbon dating (¹⁴C)

Radiocarbon dating is the method most people have heard of, and for the recent past it is the gold standard. Cosmic rays constantly create a tiny amount of radioactive carbon-14 in the atmosphere. Living things absorb it along with ordinary carbon while they breathe and eat, so a living organism holds a fixed ratio of ¹⁴C to stable ¹²C. When it dies, intake stops and the ¹⁴C begins to decay with a half-life of about 5,730 years — half of it is gone every 5,730 years. Measure how much is left, and you get the time since death.

Because the amount left halves so quickly, after roughly ten half-lives there is too little to measure reliably. That sets radiocarbon's practical limit at around 50,000 years. It also only works on things that were once alive: bone, charcoal, wood, shell, seeds, hair. It cannot date rock, and it cannot date a dinosaur or an early hominin millions of years old — a common misconception.

One crucial wrinkle is calibration. Atmospheric ¹⁴C has not been perfectly constant, so a raw radiocarbon age must be converted to a calendar age using curves (such as IntCal) built from tree rings, corals, and cave deposits of independently known age. This is why you see dates written as "cal BP" (calibrated years before present). Radiocarbon dates much of the later human story, including many migration and late-Neanderthal sites.

Potassium-argon and argon-argon dating

To reach the deep time of early human evolution, scientists turn to potassium-argon (K-Ar) dating and its refined successor, argon-argon (⁴⁰Ar/³⁹Ar). Potassium-40 is a naturally radioactive isotope with a half-life of about 1.25 billion years; as it decays, it produces argon-40 gas. When a volcano erupts, the intense heat drives out any existing argon, resetting the clock to zero. From that moment, argon-40 accumulates trapped in the cooled minerals. Measuring the potassium-to-argon ratio gives the eruption's age.

This is transformative for hominins because East Africa's Rift Valley is dotted with volcanoes that periodically blanketed the landscape in ash, forming datable layers called tuffs. Fossils themselves are not volcanic, but a fossil sandwiched between two dated tuffs is bracketed: older than the ash above, younger than the ash below. This is exactly how the famous ages at Olduvai Gorge and Lake Turkana were established.

The argon-argon method improves precision by irradiating the sample and analyzing single crystals with a laser, which helps flag contaminated or reworked grains. Together these methods reach from about 100,000 years back to the age of the Earth, making them the backbone of every date older than radiocarbon can touch.

Uranium-series, ESR, and luminescence (OSL/TL)

Between radiocarbon's ~50,000-year ceiling and potassium-argon's practical floor lies an awkward window — roughly 50,000 to 500,000 years — that happens to include the origin of our own species and the height of the Neanderthals. Three methods fill this gap.

Uranium-series (uranium-thorium) dating exploits the fact that uranium is water-soluble but its decay product thorium is not. When cave water deposits carbonate (stalagmites, flowstone) or when coral grows, it traps uranium but no thorium; thorium then slowly builds up from decay. This dates cave formations that seal archaeological layers, and it has been used to date the mineral crusts over Spanish cave paintings and human remains.

Electron spin resonance (ESR) measures electrons knocked loose by natural background radiation and trapped in crystal defects — for instance in tooth enamel. The longer the tooth has been buried, the more trapped electrons accumulate, so ESR can date the animal or hominin teeth directly, out to around two million years.

Luminescence dating — optically stimulated luminescence (OSL) and thermoluminescence (TL) — measures energy stored in minerals like quartz and feldspar since they were last exposed to sunlight or fire. Exposure empties the "trap"; burial refills it. OSL dates when sediment was last in sunlight (i.e., when it was buried), and TL dates when burnt flint or pottery was last heated. These methods helped confirm the ~300,000-year age of early Homo sapiens at Jebel Irhoud.

Paleomagnetism and molecular clocks

Two more tools work at very different scales. Paleomagnetism uses a global event recorded in the rocks: Earth's magnetic field has repeatedly reversed, swapping magnetic north and south. Iron-bearing minerals lock in the field's direction when rock or sediment forms, so a sequence of normal and reversed layers can be matched against the well-dated geomagnetic polarity timescale. It rarely gives a date on its own, but it powerfully cross-checks and refines ages from other methods across long sequences.

The molecular clock is different again — it dates lineages, not fossils. By counting the genetic differences between two species and estimating how fast mutations accumulate per generation, geneticists can estimate when their ancestors split. This is how we get figures like the human–chimpanzee divergence roughly 6–7 million years ago, or the Denisovan–Neanderthal split. But molecular-clock estimates carry real uncertainty: they depend on the assumed mutation rate and generation time, both of which are debated, so their error bars are wide and they are best anchored to fossil dates rather than replacing them.

Why dates sometimes change

Published fossil ages are measurements, not decrees, and they carry error ranges (the "±" you often see). It is normal, even healthy, for them to be revised. There are three main reasons.

First, recalibration: as radiocarbon calibration curves improve, older dates shift slightly to more accurate calendar ages. Second, new or better methods: a site first dated by one technique may be re-dated more precisely later. Third, contamination or reworked samples — modern carbon seeping into an old bone, or a volcanic crystal eroded from much older rock and redeposited — can throw an age off until it is caught.

A clear real-world example is Homo naledi. When this species was first announced in 2015 from South Africa's Rising Star cave, its age was genuinely unknown; some guessed it might be very ancient based on its small brain. In 2017, a combined dating effort using uranium-series, ESR, and other methods placed the fossils at only about 236,000 to 335,000 years old — surprisingly recent for such a primitive-looking hominin. That is dating working exactly as intended: replacing a guess with a measurement, and reshaping the story when the evidence demands it.