Two questions hiding inside one
When someone asks how old a fossil is, they are really asking two things.
First: is this fossil older or younger than that one? Second: how many years old is it, as an actual number?
Scientists answer the first question by reading rock layers. They answer the second by measuring radioactive atoms. The two methods support each other, and neither one is a guess.
Reading the layers
Older rock sits underneath
Rock often forms in layers. Sand, mud, and volcanic ash settle, get buried, and harden. New layers land on top of older ones.
So in undisturbed rock, the layer below is older than the layer above. This is called the law of superposition. A Danish scientist named Nicolas Steno wrote it down in 1669, long before anyone knew that atoms could decay.
Superposition does not give a number of years. It gives an order. That is still powerful. It is the reason fossils in lower layers are the older fossils.
Index fossils connect distant places
Some species spread over a wide area but lived for only a short slice of time. Those species work as time markers. Geologists call them index fossils.
Ammonites, coiled relatives of squid and octopuses, are classic index fossils for the age of dinosaurs. Trilobites mark much older rock. If the same index fossil turns up in Wales and in Morocco, that is good reason to think the two rock layers formed at about the same time.
Index fossils let geologists line up rock layers across continents. Still, they only give order and matching, not years.
Radioactive clocks give the years
Some atoms are unstable. Over time they change into a different element. Nothing speeds them up or slows them down in ordinary rock: not heat, not pressure, not chemistry.
That steady change is a clock. The method is called radiometric dating.
What a half-life means
A half-life is the time it takes for half of the unstable atoms in a sample to change.
Start with 100 unstable atoms. After one half-life, about 50 are left. After two, about 25. After three, about 12 or 13.
The atoms they turn into stay in the rock. So a lab measures two things: how much of the original element remains, and how much of the new element has built up. The ratio between them tells you how many half-lives have passed.
Different elements have wildly different half-lives. That is useful, because it gives scientists a clock for every timescale.
Carbon-14: the clock for recent, once-living things
Carbon-14 is the clock most people have heard of, and it is also the most misunderstood one.
Living things take in carbon from the air and from food. A tiny fraction of that carbon is carbon-14. While an organism is alive, it keeps topping up its supply. When it dies, the topping up stops, and the carbon-14 it already holds starts running down.
Carbon-14 has a half-life of about 5,730 years. After about 50,000 years, so little is left that measuring it becomes unreliable.
This gives carbon dating two hard limits.
- It works only on material that was once alive: wood, bone, charcoal, cloth, shell, seeds.
- It works only back to roughly 50,000 years.
Why “carbon dating a dinosaur” is a mistake
People sometimes argue about evolution by saying carbon dating is unreliable for dinosaurs. That is true, and no scientist disagrees. Carbon dating is never used on dinosaurs.
Dinosaurs other than birds died out about 66 million years ago. That is more than a thousand carbon-14 half-lives. There would be nothing left to measure.
A dinosaur fossil is dated using rock layers and much slower clocks. Carbon-14 is the wrong tool, in the same way a kitchen ruler is the wrong tool for measuring the distance to another city.
Clocks for deep time
For millions and billions of years, scientists use elements that decay far more slowly.
Potassium-argon. Potassium-40 decays to argon-40 with a half-life of about 1.25 billion years. Argon is a gas, and it escapes from hot lava. Once the lava cools into rock, the argon is trapped inside. So this clock starts at the moment the rock cools.
The method dated the volcanic layers around the famous Australopithecus afarensis skeleton nicknamed Lucy, found at Hadar in Ethiopia in 1974. Those layers put her at about 3.2 million years old.
Uranium-lead. Uranium-238 decays into lead-206, through a chain of steps. Its half-life is about 4.5 billion years. A second kind, uranium-235, decays to lead-207 with a half-life of about 704 million years.
Having two uranium clocks in the same crystal is a gift. Each one can check the other.
Zircon: the crystal that keeps its own records
Uranium-lead dating works best in a tough little mineral called zircon.
When a zircon crystal grows in cooling magma, its structure accepts uranium atoms but strongly rejects lead. So a fresh zircon starts with uranium and almost no lead. Any lead found inside it later has to be decay product.
Zircon is also extremely hard and chemically stubborn. Crystals survive being weathered, buried, and recycled into new rock. Zircon grains from the Jack Hills in Western Australia have been dated to about 4.4 billion years, which makes them the oldest known pieces of the Earth.
Why we date the rock, not the fossil
Here is the part that surprises most people. Scientists usually do not date the fossil itself.
A fossil is often found in sandstone or mudstone. Those rocks are made of grains worn off older rocks somewhere else. Dating a sand grain tells you when that grain first crystallised, far away and long before, not when the animal died.
Volcanic rock is different. A lava flow or a bed of volcanic ash melts everything and then cools in one event. The clock starts then.
So the standard approach is bracketing. Find a dated volcanic layer below the fossil and another dated volcanic layer above it. The fossil must be younger than the lower layer and older than the upper one. If the ash below is 376 million years old and the ash above is 374 million years old, the fossil in between is about 375 million years old.
That is how the team searching for Tiktaalik roseae knew which Arctic rocks to dig in, and how a fish fossil earns a number like “375 million years”. Fossil evidence tells the rest of that story.
The age of the Earth
The Earth is about 4.54 billion years old.
That number does not come mainly from Earth rocks, and there is a good reason. The Earth recycles its own surface. Plate movement, mountain building, and erosion destroy old rock. The oldest rocks found so far, the Acasta Gneiss in northern Canada, are about 4.03 billion years old, and the Jack Hills zircons push back to about 4.4 billion. Those are minimum ages for the crust, not the age of the planet.
Meteorites solve the problem. Most meteorites formed at the same time as the rest of the solar system, and were never melted again. Scientists measured the lead in meteorite material, including the iron meteorite from Canyon Diablo in Arizona. The answer came out at about 4.54 billion years, with an uncertainty of less than about one percent.
Earth and the meteorites formed from the same cloud of material at close to the same time, so that age applies to the Earth too.
The cross-checks that make dating trustworthy
A clock you cannot check is only a claim. Radiometric dating can be checked, in several independent ways.
Tree rings. Many trees add one ring per year. Rings can be counted directly and matched between overlapping trunks, living and dead. Chained together, European oak and pine records reach back more than 10,000 years. Wood of a known ring-count age can then be carbon dated, and the two numbers can be compared.
Ice cores. Snow falls in yearly cycles in Greenland and Antarctica, and the buried layers can be counted like tree rings, using dust, acidity, and chemistry. The GISP2 core from Greenland was counted through tens of thousands of layers. Antarctic cores reach much further back still. Volcanic eruptions with known historical dates show up as acid spikes in the ice, exactly where the layer count says they should.
Varves. Some quiet lakes lay down a pale layer in summer and a dark layer in winter, one pair per year. These are called varves. The sediment of Lake Suigetsu in Japan preserves such layers across tens of thousands of years, and leaf fragments trapped in them have been carbon dated. The layer counts and the radiocarbon dates track each other.
Different methods, different labs. A single rock unit is often dated by more than one method, in more than one laboratory, by teams that do not want to be wrong in public. Potassium-argon and uranium-lead measure different elements with different chemistry and different equipment. When they land on the same number, coincidence stops being a sensible explanation.
What dating can and cannot do
Radiometric dating is not magic, and honest scientists say so.
- Every date comes with an error range, such as “1.8 million years, plus or minus 0.05 million”.
- A rock that was reheated can have its clock partly reset, so geologists check for that.
- A sample can be contaminated, so labs run blanks and repeats.
- The wrong method on the wrong material gives a wrong answer, which is exactly why nobody carbon-dates a dinosaur.
Knowing the limits is part of the method, not a weakness in it.
Why this matters for evolution
Evolution needs time, and deep time is not an assumption. It is a measurement, made with physics, checked against countable yearly layers, and repeated across laboratories on different continents.
Once the dates are in place, the fossil record stops being a pile of old bones and becomes an ordered history. To see what that history looks like, read fossil evidence. To see how the same story shows up in living things, read genetic evidence.