How Radiometric Dating Works

Radiometric dating measures how much of a radioactive isotope in a rock or fossil has decayed into a stable "daughter" isotope, then uses the isotope's known decay rate to calculate how much time has passed. It is how geologists arrived at an age of about 4.54 billion years for Earth, and how archaeologists date a piece of charcoal from an ancient campfire to within a few decades. The method works because radioactive decay is, for practical purposes, indifferent to heat, pressure, and chemistry. That indifference turns atoms into clocks.

The problem radiometric dating solves

Before the 20th century, geologists could only put rock layers in order. A layer of sandstone beneath a layer of shale was older, because sediment piles up over time and newer material sits on top. This is called relative dating, and it tells you the sequence of events but not how long ago they happened or how much time separated them.

That gap mattered. Charles Darwin needed vast stretches of time for natural selection to work, but the physicist Lord Kelvin, reasoning from how fast a molten planet should cool, gave a range of 20 million to 400 million years in the 1860s and later pushed his figure down to 20 to 40 million. His physics was sound for the model he used. The model was missing ingredients: Kelvin assumed Earth had no internal heat source, and nobody in his generation knew that radioactive decay inside the planet keeps producing heat. Radioactivity had not been discovered yet.

Henri Becquerel found it in 1896 (the story is told in the Becquerel radioactivity discovery), and Marie Curie and Pierre Curie soon isolated radium and polonium. In a 1904 lecture attended by Kelvin himself, Ernest Rutherford argued that radium supplied a new source of heat that undercut Kelvin's cooling limit, and he used the helium in a mineral he owned to estimate an age of roughly 40 million years. Rutherford is credited with proposing radiometric dating around 1905, and Rutherford in 1906 and the chemist Bertram Boltwood in 1907 pioneered dating actual minerals and rocks (Wikipedia, "Radiometric dating"). Historians add a caution: the neat tale that radioactive heat alone refuted Kelvin is a simplification, and the planet's cooling is more complicated than either man's model.

The big number arrived in 1956. Clair Patterson, using lead isotopes in several meteorites including the Canyon Diablo iron meteorite, calculated an age of 4.55 billion years, give or take 70 million, for the meteorites and, by extension, the Earth (Wikipedia, "Age of Earth"). The U.S. Geological Survey today gives 4.54 billion years with an uncertainty of less than 1 percent, still based on meteorite lead ratios, specifically the Canyon Diablo meteorite (USGS, "Age of the Earth"). The meteorite itself is a time capsule: it formed with the solar system and has not been melted and remixed the way Earth's own crust has.

Radiometric dating did more than answer "how old is this rock." It gave every historical science an absolute timeline, from evolutionary biology to paleoclimatology. Without it, you can say one fossil layer lies beneath another and is therefore older. With it, you can say how much older, and that the last non-avian dinosaurs lived in the final stretch of the Cretaceous Period, which ended about 66 million years ago.

Surprising fact: The age of Earth is not measured from Earth rocks at all. The oldest rocks that survive on Earth's surface, the Acasta Gneisses in Canada, are at least 4.03 billion years old, which is younger than the planet. Meteorites supply the 4.54-billion-year figure (USGS).

What "radioactive decay" actually means

Every atom of a given element has the same number of protons, but the number of neutrons can vary. Atoms of one element with different neutron counts are called isotopes. They behave almost identically in chemistry, but they differ in mass and, sometimes, in nuclear stability. Some isotopes last forever. Others are unstable: the nucleus holds too much energy to stay together, so it eventually transforms into a different arrangement, emitting radiation along the way.

That transformation is radioactive decay, and it comes in a few forms:

• Alpha decay: the nucleus ejects two protons and two neutrons (a helium nucleus), so the atom drops two places on the periodic table. Uranium-238 does this.
• Beta decay: a neutron turns into a proton and ejects an electron. Carbon-14 (six protons, eight neutrons) becomes nitrogen-14 (seven protons, seven neutrons) this way.
• Electron capture: the nucleus swallows one of the atom's own inner electrons, turning a proton into a neutron. Potassium-40 decays partly by this route into argon-40.

Uranium-238 does not jump to lead in one step. It passes through a chain of intermediate isotopes, including radium and radon, before ending as stable lead-206. The full journey takes eight alpha decays and six beta decays, 14 steps in all. Most of the intermediates are short-lived, so for dating purposes the chain behaves like one very slow transformation whose pace is set by the first and longest step.

You cannot predict when any single atom will decay. The moment is random, governed by quantum mechanics. Across a large population of atoms, though, the fraction that decays in a given time is remarkably steady. That is the key to everything that follows.

Half-life, explained with popcorn

The half-life of an isotope is the time it takes for half of a sample to decay. Picture a pan of popcorn kernels over heat. You cannot say when a particular kernel will pop, but you can say that half of the kernels will have popped by some predictable time, and half of the remainder by the same interval after that. Radioactive atoms follow that pattern, with one difference: the "heat" never changes, so the interval never changes.

After one half-life, 50 percent of the parent isotope remains. After two, 25 percent. After three, 12.5 percent. It never reaches zero, but after about 10 half-lives, less than one-thousandth of the original remains, which is why every isotope has a practical window of usefulness.

Why decay rates don't change with heat, pressure, or chemistry

Chemical bonds, temperature, and pressure act on an atom's outer electrons. Alpha and beta decay start deep inside the nucleus, far from anything a laboratory or a mountain range can do. Experiments that heated, compressed, and chemically transformed radioactive samples have found no meaningful change in decay rates for the isotopes used in geological dating.

There is one honest footnote. Electron capture requires an electron to be near the nucleus, so it can respond very slightly to its environment. Laboratory work on beryllium-7, the best studied case, found small shifts, on the order of a percent or less, under extreme pressure or in different chemical forms, and the exact size of the effect is still debated. Beryllium has only four electrons, so its outer electrons sit close to the nucleus, which makes it unusually sensitive. For the heavy isotopes that date rocks, such effects are far too small to matter against other sources of error.

The math behind the clock

The relationship between elapsed time and remaining parent material follows an exponential decay curve. If you start with a quantity N₀ of the parent isotope, the amount N that remains after time t is N = N₀ × (1/2)^(t / half-life).

Nobody was standing there with a detector when a granite crystallized, so N₀ is almost never known. Geochronologists instead measure what is in the sample today: the remaining parent (P) and the accumulated daughter (D). Since every daughter atom was once a parent atom, the original quantity is simply P + D. The age is then:

t = (half-life / ln 2) × ln(1 + D/P)

You do not need the logarithm to see the logic. If the sample holds equal parts parent and daughter (D/P = 1), exactly one half-life has passed. If it holds three parts daughter to one part parent (D/P = 3), two half-lives have passed.

D/P ratio	Fraction of parent left	Half-lives elapsed	Example with carbon-14 (5,730 years)
0	100%	0	0 years
1	50%	1	5,730 years
3	25%	2	11,460 years
7	12.5%	3	17,190 years
15	6.25%	4	22,920 years

A sample of granite containing equal parts uranium-238 and lead-206 produced by its decay would have crystallized about one half-life of uranium-238 ago, roughly 4.47 billion years. The accepted half-life of uranium-238 is about 4.47 billion years, and that of uranium-235 about 704 million years.

This is why radiometric dating works best on materials that started with a clean slate. Igneous rock, which cooled from molten magma or lava, is ideal. Zircon crystals are the standout example: when they form, they readily take uranium into their crystal structure but strongly reject lead (Wikipedia, "Radiometric dating"). Any lead found inside a zircon crystal therefore came almost entirely from decay after the crystal formed. The daughter count starts at essentially zero, like a stopwatch pressed at the moment of crystallization.

How do scientists know decay rates are constant?

Independent clocks cross-check each other. Tree-ring counts of known-age wood agree with carbon-14 results over thousands of years. Uranium-lead dates measured on the same zircon crystals by different laboratories decades apart agree. And different radioactive systems, built on different physics, such as potassium-argon, uranium-lead, and rubidium-strontium, give matching ages when applied to the same rock. If decay rates had drifted, each clock would drift differently and the agreement would fall apart. It does not.

Isochrons and concordia: dating without knowing the start

Sometimes a rock does start with some daughter isotope already in it, which would ruin the stopwatch picture. Geochronologists have clever workarounds.

The isochron method compares several minerals from the same rock. Each mineral contains different proportions of the parent isotope but formed together from the same melt, so they begin with the same initial ratio of daughter to a stable reference isotope. As the parent decays, the minerals spread out along a straight line on a graph. The slope of that line gives the age, and the intercept gives the original composition, so the starting amount no longer has to be assumed (Wikipedia, "Radiometric dating"). It is like timing a race by looking at how far apart the runners have spread, when you were not there to see the starting gun.

The concordia diagram, used in uranium-lead dating, exploits the two clocks inside every zircon. Uranium-235 turns into lead-207 and uranium-238 turns into lead-206 at different rates, so an undisturbed crystal gives the same age from both clocks and plots on a particular curve. If a crystal later lost some lead, it falls off the curve along a straight line that points back at the true crystallization age where it crosses the curve. A disagreement between the two clocks is not a failure. It is information about what happened to the crystal.

When does the clock actually start?

A rock does not begin keeping time when it first forms, but when it cools enough to trap the daughter isotope. This is called the closure temperature. Above it, the daughter atoms escape and the clock stays at zero. Below it, the mineral is a closed system and the count begins. Wikipedia's summary of the literature puts the closure temperature for the potassium-argon system at roughly 350 °C for mica and 500 °C for hornblende, and notes that closure temperatures differ by mineral and isotopic system.

The closest everyday picture is an hourglass that only starts to run once you turn it and the neck is narrow enough that sand cannot flow back. Heat a rock again past its closure temperature, say in a mountain-building event, and the hourglass is turned over. This is a feature, not a bug. Geologists can date the original crystallization with one mineral and the later heating with another, reading a rock's life as a sequence of events.

Willard Libby and the invention of carbon dating

The most famous radiometric method, radiocarbon dating, grew out of a discovery made in 1940, when Martin Kamen and Samuel Rubin at Berkeley first identified carbon-14. Willard Libby, a physical chemist at the University of Chicago, proposed in 1947 that the isotope could serve as a clock and demonstrated the idea soon afterward (Engineering and Technology History Wiki, Libby Develops Radiocarbon Dating). The insight is elegant. Cosmic rays striking the upper atmosphere keep converting nitrogen-14 into carbon-14, which combines with oxygen into radioactive carbon dioxide. Plants absorb that carbon dioxide through photosynthesis, and animals absorb it by eating plants, so every living organism carries roughly the same ratio of carbon-14 to stable carbon-12 as the atmosphere. When an organism dies, it stops taking in carbon. Its carbon-14 decays back into nitrogen-14 with a half-life of 5,730 years, while its carbon-12 stays put.

Measuring the ratio in a bone, a scrap of linen, or a piece of charcoal tells you how long ago the organism died. Libby received the 1960 Nobel Prize in Chemistry "for his method to use carbon-14 for age determination in archaeology, geology, geophysics, and other branches of science" (Nobel Prize). For a fuller treatment of the method itself, see Radiocarbon Dating: How Carbon-14 Tells Us the Age of Ancient Things.

Why raw radiocarbon dates need calibration

Libby's method assumed the atmosphere always held the same share of carbon-14. It does not. Cosmic ray flux, Earth's magnetic field, and ocean circulation have all shifted the ratio over time, so a raw "radiocarbon year" is not exactly a calendar year. Scientists correct for this with a calibration curve, built by measuring carbon-14 in material of independently known age.

The current standard for the Northern Hemisphere is IntCal20, published in 2020 in the journal Radiocarbon. It converts radiocarbon ages to calendar ages across 0 to 55,000 years before present. Tree rings supply the fully atmospheric record back to about 13,900 years before present, and older sections draw on floating tree-ring chronologies, lake and marine sediments, cave deposits called speleothems, and corals (Reimer et al., 2020, University of Arizona record). Marine samples need an extra reservoir correction because the ocean holds older carbon.

Human activity has added two wrinkles. Atmospheric nuclear tests in the 1950s and early 1960s roughly doubled the atmospheric carbon-14 level, and the signal has been falling since the 1963 test ban treaty (Lawrence Livermore National Laboratory, bomb pulse forensics). That "bomb pulse" turns out to be useful for dating very recent material, such as tissue formed after 1955. Burning fossil fuels, meanwhile, adds carbon with no carbon-14 left in it, which dilutes the atmospheric signal and will make future samples look older than they are.

One famous test: the Shroud of Turin

Radiocarbon dating has also been used on objects whose age was a matter of public debate. In 1988, three laboratories at Arizona, Oxford, and Zurich dated small pieces of the Shroud of Turin using accelerator mass spectrometry, alongside three control cloths of independently known age. Their 1989 Nature paper concluded the linen was medieval (Damon et al., 1989). Critics have since questioned the sampling and possible contamination, which is an instructive example of how real the contamination risk is in practice and why labs run controls.

Why can't carbon dating be used on dinosaur fossils?

Carbon-14 is the wrong tool for deep time. After about eight to ten half-lives, so little carbon-14 is left that it cannot be told apart from contamination, which is why routine radiocarbon dating works to about 50,000 years, and the best laboratories stretch toward 55,000 to 60,000. Non-avian dinosaurs died out around 66 million years ago, more than a thousand times beyond that range. In addition, most fossils have had their original bone replaced by minerals, so there is no original carbon left to measure. A scientist who wants the age of a dinosaur uses a different clock, applied to different material, as the sections below explain.

Choosing the right isotope for the job

No single isotope can date everything, because a clock's useful range is tied to its half-life. A fast-decaying isotope disappears too soon, since almost none of the parent survives after a modest span. A very slow one will not show enough change over a short span to tell a 500-year-old object from a 5,000-year-old one. A good rule of thumb is that a clock works best over a window from a small fraction of its half-life up to several half-lives. Matching the isotope to the age of the object is the first and most important decision in any radiometric study.

Method	Parent isotope	Half-life (approx.)	Typical material dated	Practical range
Radiocarbon	Carbon-14	5,730 years	Bone, wood, charcoal, shell, cloth	Up to about 50,000 years
Uranium-thorium	Uranium-234 / thorium-230	Tens to hundreds of thousands of years	Cave deposits, corals	Up to several hundred thousand years
Potassium-argon / argon-argon	Potassium-40	About 1.25 billion years	Volcanic rock, ash, micas	Wide range; best for older volcanic rock
Uranium-lead	Uranium-238 / uranium-235	4.47 billion / 704 million years	Zircon in igneous rock	From millions of years to the age of the solar system
Rubidium-strontium	Rubidium-87	About 49 billion years	Ancient igneous and metamorphic rock	Hundreds of millions to billions of years
Samarium-neodymium	Samarium-147	About 106 billion years	Ancient crustal rock, meteorites	Billions of years

Wikipedia's survey of the field notes that radioactive systems in use span half-lives from roughly 10 years (tritium) to more than 100 billion years (samarium-147), and that uranium-lead dating can deliver uncertainties below 2 million years on rocks 2.5 billion years old. Two clocks inside uranium-lead, the uranium-238 and uranium-235 chains, give a built-in cross-check that few other methods offer.

The half-lives above are measured values, not definitions. Different laboratories have published slightly different numbers for some isotopes, for example for potassium-40 and rubidium-87, and the differences matter at the level of fractions of a percent. This is one reason a geochronology paper states which decay constants it used.

How a dinosaur fossil gets dated

A fossil itself almost never contains datable radioactive material. Fossilization replaces original tissue with minerals, and sedimentary rock, the kind that usually holds fossils, is made of grains eroded from many older rocks of different ages. Dating the sediment would give you the age of the oldest grains, not the age of the burial. Paleontologists therefore use a technique called bracketing.

Volcanic ash falls in thin layers that settle quickly and carry crystals that formed at the time of the eruption. Where an ash bed sits below a fossil layer and another above it, dating the two ash beds gives a maximum and a minimum age. Suppose, purely as an illustration, that ash below a Triceratops skeleton dates to 68.2 million years and ash above it dates to 66.8 million years. The skeleton must lie within that 1.4-million-year window. Real cases are published with uncertainties on each date.

An analogy: it is like working out when a letter was written by checking the postmarks on the envelopes that arrived just before and just after it in a mailbag. The letter itself carries no date, but its neighbors do. For more on this approach, see How Paleontologists Date Dinosaur Fossils.

The ash method also pins the timing of events. In 2013, Paul Renne and colleagues used argon-argon dating on volcanic ash to show that the end-Cretaceous mass extinction and the Chicxulub asteroid impact happened within 32,000 years of each other, and that the disruption of the atmospheric carbon cycle likely lasted less than 5,000 years (Renne et al., 2013, Science 339:684, abstract record). The details of that catastrophe are in What Killed the Dinosaurs: The Chicxulub Asteroid Impact Explained.

What the oldest dates look like

Radiometric methods are not only about ordinary rocks. They also anchor the oldest chapters of Earth's and the solar system's history. The table collects benchmark ages, with the source for each.

Object or event	Approximate age	Source
Meteorites and the solar system	4.53 to 4.58 billion years	USGS
Earth (from lead in Canyon Diablo meteorite troilite)	4.54 billion years, uncertainty under 1%	USGS
Moon rocks	4.4 to 4.5 billion years	USGS
Jack Hills zircon, Western Australia	At least 4.4 billion years	University of Wisconsin-Madison, 2014
Acasta Gneisses, Canada	At least 4.03 billion years	USGS
Isua rocks, Greenland	3.7 to 3.8 billion years	USGS

The Jack Hills zircon deserves a closer look. In 2014, a team led by John Valley at the University of Wisconsin-Madison reported in Nature Geoscience that a zircon crystal from Western Australia crystallized at least 4.4 billion years ago, about 160 million years after the solar system formed. The team used atom-probe tomography to map individual lead atoms and found that the lead had gathered into tiny clusters, each under 50 atoms across, which was small enough in scale that it did not disturb the age (UW-Madison). That finding addressed a standing criticism, namely that lead might have moved around inside old zircons and spoiled the clock. For the geologic context of these ancient rocks, see How Scientists Measure the Age of Earth and How the Solar System Formed.

Where radiometric dating breaks down

No method is bulletproof, and the useful question is how each one fails.

Contamination. If a sample gains or loses parent or daughter isotopes after formation, the calculated age is off. Zircon is valued because its tightly packed crystal structure resists this far better than most minerals, but even zircon can lose lead when it is damaged by its own radiation or by later heating, which is exactly what the concordia diagram is designed to detect.

Inherited argon. In potassium-argon dating, a volcanic eruption can carry in older crystals that already contain argon-40. The sample then looks older than the eruption. Geologists check several crystals from the same layer and use the argon-argon variant, which releases argon in steps as the sample is heated and can reveal when older argon is present through the pattern of the release.

Resetting. A later heating or metamorphic event can partly or completely reset a clock. The date then reflects the second event rather than the original formation. Using more than one mineral and more than one isotopic system helps separate the events.

Undatable material. Some rocks simply do not contain a single formation event to date. Sedimentary rock built of eroded particles of many ages is the standard case, which is why ash-bracketing exists as a workaround.

Reporting uncertainty. Every credible date comes with an error range. Uncertainty comes from the measurement itself, from the decay constants, and from assumptions about the sample. A date quoted without an error range deserves skepticism, because real measurement always has one. Reporting the uncertainty honestly is a sign of rigor, not weakness.

What people get wrong about radiometric dating

It is not one technique. It is a family of methods, each suited to particular materials and time ranges. Confusing radiocarbon's ceiling of about 50,000 years with uranium-lead's reach across billions of years leads to false stories, such as the idea that scientists "failed" to carbon-date a dinosaur bone. In reality, no geochronologist would try, since carbon-14 is the wrong tool for a sample that old.

A single date is not treated as gospel. Credible labs measure many crystals, use more than one isotopic system, and publish a result with confidence only when independent methods converge. The age of the Earth rests on lead ratios from several meteorites that agree with each other, and the solar-system range of 4.53 to 4.58 billion years reflects the spread among them (USGS).

Decay rates are not assumed, they are tested. Claims that rates might have been faster in the past run into the cross-checks described above. Different clocks with different physics agree on ages, and a rate change big enough to matter would pull them apart. The few measurable exceptions, in electron-capture decay under extreme conditions, are tiny and do not touch the heavy-element clocks used for rocks.

An old date does not mean an infallible one. Radiometric ages can be wrong in individual cases, because a sample was contaminated, reset, or misinterpreted. The system is robust because those errors are found by comparison, not because errors never happen.

FAQ

Q: What is radiometric dating in simple terms? A: Radiometric dating measures the ratio of a radioactive isotope to its stable decay product in a rock, mineral, or organic sample, then uses the isotope's known half-life to calculate how much time has passed since the material formed or the organism died.

Q: How accurate is radiometric dating? A: Well-controlled uranium-lead dates on zircon can carry uncertainties of a few million years on rocks billions of years old, which is well under a few percent. Accuracy depends on avoiding contamination, on knowing the half-life precisely, and on cross-checking with independent isotopic systems.

Q: Can carbon dating be used on dinosaur bones? A: No. Carbon-14's 5,730-year half-life limits the method to about 50,000 years, and dinosaur fossils are tens of millions of years old and mostly mineral. Dinosaur fossils are dated indirectly with potassium-argon, argon-argon, or uranium-lead dating of volcanic layers above and below them.

Q: Why don't heat and pressure change decay rates? A: Alpha and beta decay originate in the atomic nucleus, while heat, pressure, and chemical bonding act on the surrounding electrons. Laboratory tests have found no meaningful change for the isotopes used in geological dating. Light electron-capture isotopes such as beryllium-7 show shifts of a fraction of a percent under extreme conditions.

Q: What is the oldest thing radiometric dating has measured? A: Meteorites give the oldest solid ages, 4.53 to 4.58 billion years, which define the age of the solar system. The oldest dated material that formed on Earth is a Jack Hills zircon of at least 4.4 billion years, according to the 2014 study led by John Valley.

Q: What does the half-life tell you about how long a clock lasts? A: A clock is most useful over a window of a small fraction of a half-life up to several half-lives. After about 10 half-lives, less than one-thousandth of the parent isotope remains, so the measurement becomes unreliable. That is why carbon-14 covers tens of thousands of years and uranium-238 covers billions.

The takeaway

Radiometric dating turns the steady decay of unstable isotopes into a calendar for events that no one was around to record. The central trick is to measure what has accumulated, a daughter isotope, relative to what is left, and let a nearly unchangeable half-life convert that ratio to time. Carbon-14 handles the last 50,000 years or so, and uranium, potassium, rubidium, and samarium clocks stretch the record across the whole history of the solar system. The method is strongest not because any one clock is perfect, but because clocks built on different elements and different physics keep agreeing, and when they disagree, the disagreement usually points to an interesting event in the rock's history.

Sources

• U.S. Geological Survey, "Age of the Earth", a basic summary of meteorite lead ratios and the age of the oldest rocks.
• Engineering and Technology History Wiki, "Libby Develops Radiocarbon Dating".
• Nobel Prize in Chemistry 1960 (NobelPrize.org).
• Reimer, P.J., et al. (2020), "The IntCal20 Northern Hemisphere Radiocarbon Age Calibration Curve (0-55 cal kBP)," Radiocarbon 62(4): 725-757 (University of Arizona record).
• Valley, J.W., et al. (2014), Nature Geoscience, summarized by the University of Wisconsin-Madison.
• Renne, P.R., et al. (2013), "Time scales of critical events around the Cretaceous-Paleogene boundary," Science 339(6120): 684-687 (record).
• Damon, P.E., et al. (1989), "Radiocarbon dating of the Shroud of Turin," Nature 337 (record).
• Wikipedia, "Radiometric dating", for the isochron, concordia, and closure-temperature summaries.

---

Read next:

• Nuclear Fusion Energy Explained — The Power of the Stars
• How the Human Brain Actually Works
• How Does DNA Work