How Are Fossils Dated? Layers, Isotopes and Index Fossils

Sedimentary rock forms as sand, mud and other material settles in layers, and unless something has turned the sequence over, each layer is younger than the one below it. That principle, first stated by Nicolas Steno in 1669, gives the order of events without giving any dates. A fossil in a lower layer is older than one above; a layer that cuts across others is younger than all of them; a fragment of one rock embedded in another is older than its host.

Geologists used this ordering to build the whole geological timescale during the nineteenth century, long before they could attach a single number to it. The Jurassic was defined by the rocks of the Jura Mountains and the fossils in them, and it was known to lie between the Triassic and the Cretaceous decades before anyone knew it began 201 million years ago.

A layer in England cannot be laid physically alongside one in Argentina, so the ordering has to be carried from place to place, and fossils themselves do the carrying. Some species were widespread, abundant, easily recognised and short-lived, existing for only a million years or so before going extinct. Find one of these index fossils in two rock sequences on different continents and the layers containing them are the same age.

Ammonites, coiled marine shells that evolved rapidly through the age of dinosaurs, are the classic example; their species change so fast that Jurassic and Cretaceous rocks can be placed to within a fraction of a million years by the ammonites alone. Tiny fossils, such as pollen and the shells of single-celled plankton, do the same job for rock cores drilled by oil companies. None of this gives an age in years. It gives a position in a sequence that runs the same way everywhere.

Numbers come from radioactive isotopes, unstable atoms that decay into other atoms at a fixed rate unaffected by heat, pressure or chemistry. Each isotope has a half-life, the time for half of any sample to decay, and by measuring how much of the parent isotope remains against how much of its product has accumulated, the time since the clock started can be calculated. The clock starts when a mineral crystallises and locks the atoms in.

Different isotopes cover different ranges. Uranium-238 decays to lead with a half-life of 4.5 billion years and dates the oldest rocks on Earth. Potassium-40 decays to argon with a half-life of 1.3 billion years and is the workhorse for volcanic rock of any age over a hundred thousand years. Carbon-14, famous from archaeology, has a half-life of only 5,730 years and is useless beyond about 50,000 years, which is why it plays no part in dating dinosaurs, despite what television sometimes implies.

The catch is that radioactive dating works on igneous rock, which crystallised from molten material, and fossils lie in sedimentary rock, which is made of older grains that were dated already when they formed elsewhere. So a fossil is dated by bracketing. Find a layer of volcanic ash below the fossil and another above it, date both, and the fossil lies between the two numbers. Volcanic ash falls are common in the geological record and each one is an instant marker across a wide area.

The 66-million-year date for the end of the dinosaurs comes from exactly this: a thin layer of clay found worldwide, rich in iridium from the asteroid, with datable volcanic minerals close above and below it, and index fossils that disappear at the same level everywhere. Several independent methods agree to within a few tens of thousands of years, which for an event 66 million years ago is agreement to one part in a thousand.

Every method has known limits, and the limits are what make the results trustworthy. Radiometric dates carry a stated uncertainty, usually under one percent. Minerals that were heated after forming can lose their argon and read young, which is why several minerals from the same rock are dated together. Layers can be folded or faulted so that older rock ends up on top, which the index fossils reveal. The methods are chosen so that each one's weakness is another one's strength. The tools, by what they give:

• •Layer order: which is older, no numbers
• •Index fossils: the same position in the sequence anywhere in the world
• •Uranium-lead: numbers for the oldest rocks, billions of years
• •Potassium-argon: numbers for volcanic layers above and below fossils
• •Carbon-14: numbers for organic remains younger than 50,000 years only

Earth's magnetic field has flipped direction many hundreds of times, so that a compass which points north today would have pointed south during long stretches of the past. When fine grains of iron minerals settle in mud or cool in a lava flow, they line up with the field of the moment and keep that direction once the rock hardens. Measure the magnetism of a sequence of layers from bottom to top and you get a pattern of normal and reversed intervals of different lengths, a kind of barcode written into the rock.

The barcode is the same everywhere on Earth at a given moment, because the whole field flips at once, so it can match layers between continents even where no index fossils are found. The pattern was first read clearly in stripes of magnetised rock on either side of the mid-ocean ridges, and the reversals themselves have been dated by radiometric methods, the most recent one roughly 780,000 years ago. A fossil site with a distinctive run of reversals can then be slotted into the global timescale, as an independent check on the ash dates and the fossils.

The same decay clocks that date fossil-bearing layers were used to answer a bigger question: how old is the planet? Earth's own surface is constantly recycled, so its first rocks are gone. The American geochemist Clair Patterson got around this by measuring lead isotopes in a meteorite, on the reasoning that meteorites formed at the same time as the Earth from the same material and have stayed unaltered since. In 1956 he published an age of about 4.55 billion years, a figure later refinements have barely changed. The oldest minerals found on Earth, tiny zircon crystals from Western Australia, are more than four billion years old.

Patterson's work had a surprising side effect. To measure the minute amounts of lead in his samples, he had to build one of the first ultra-clean laboratories, because lead from the air, dust and equipment kept ruining his results. Tracing where all that lead came from led him to petrol, which then contained added lead compounds, and he showed that environmental lead levels had risen enormously in the industrial era. He spent years campaigning against the fuel additive in the face of fierce industry opposition, and his evidence contributed to leaded petrol being phased out across the world.

A familiar objection says the method chases its own tail: rocks are dated by the fossils in them, and fossils by the rocks they lie in. The order of the layers, though, was worked out first and independently. In the early nineteenth century William Smith, an English surveyor of canals and coal mines, noticed that the same fossils always turned up in the same sequence of strata wherever he dug, and used them to trace layers across the country. His geological map of England and Wales appeared in 1815, more than forty years before Darwin published his theory of evolution.

Smith had no theory of why fossils changed through the rock; he simply recorded that they did, and the order has been confirmed in sequence after sequence ever since. Radiometric dates arrived in the twentieth century and were slotted into this existing order, which gave them every chance to fail. If an ash layer near the base of a cliff had come out younger than one near the top, the whole framework would have been in trouble. Instead, dates from different isotopes, measured in different laboratories, keep coming out in the order the layers predict, which is the opposite of circular.

The methods do not always agree at first, and the most famous disagreement shows how the checks work. In 1972 Richard Leakey's team at Koobi Fora, east of Lake Turkana in Kenya, found a skull known by its catalogue number, KNM-ER 1470, with a surprisingly large braincase. It lay below a volcanic ash called the KBS Tuff, which had been dated by potassium-argon methods to about 2.6 million years. That made the skull older than any comparable human fossil then known, and it seemed to rewrite the early history of our own genus.

Fossils from the same beds told a different story. The palaeontologist Basil Cooke compared the teeth of fossil pigs, which evolved quickly and had been dated at other East African sites, and found that the Koobi Fora pigs matched those from deposits well under two million years old. A long and heated argument followed, until new measurements by other laboratories in the late 1970s and early 1980s settled the age of the ash at about 1.9 million years. The original samples appear to have been contaminated with older volcanic grains. The index fossils had been right, and the radiometric date was corrected.

Fossils are dated by bracketing: the order of rock layers and the index fossils in them say what is older than what, and radioactive decay in volcanic layers above and below puts numbers on the sequence. Carbon-14 covers only the last fifty thousand years; everything older is dated by uranium, potassium and the layers themselves.