Deep time

Foundations

The number that needs defending

This course has quoted ages without apology: 13.8 billion years for the universe, 4.567 billion for the solar system, 4.4 billion for the oldest crystals, 66 million for the Deccan Traps. Those are not estimates in the loose sense, and they are not obtained by assuming how fast mud piles up. They come from clocks that were running in the rock all along, and this lesson is about how they work and why they can be trusted.

Before the clocks

The idea that Earth is old came first, and it came from looking carefully at rocks.

In the 1780s a Scottish farmer and physician named James Hutton examined a coastal outcrop where near-vertical layers of ancient rock are cut off flat and overlain by gently sloping younger layers. To make that, he reasoned, sediment must be laid down horizontally, buried, hardened, tilted on edge, raised up, worn flat by erosion, submerged again, and buried under new sediment. Every step is slow, and the whole sequence had to happen before the visible rock above it accumulated. Hutton concluded that Earth’s history showed “no vestige of a beginning, no prospect of an end”. He had no numbers, and he was right about the scale.

The nineteenth century turned this into a fight. Charles Darwin needed vast time for natural selection. Geologists agreed. And William Thomson, Lord Kelvin, the most authoritative physicist of the age, calculated how long a molten Earth would take to cool to its present state and got tens of millions of years: far too short. His physics was sound and his answer was wrong, because he did not know that the Earth contains radioactive elements generating heat continuously. The same discovery that ruined his calculation supplied the clock that settled the argument.

Why radioactive decay makes a good clock

An unstable nucleus decays at random. There is no way to say when a particular atom will go. But in any real sample there are enormous numbers of atoms, and the fraction that decays per unit time is fixed with great precision. That fixed rate is expressed as a half-life: the time for half of them to go.

The crucial property is indifference. Chemical reactions speed up when heated and change with pressure and with what else is present. Nuclear decay does not care. Heat a mineral, squeeze it, dissolve it in acid, and its decay rate stays the same, because the process happens in the nucleus, insulated from the electrons that do chemistry. A rate that nothing in ordinary geology can alter is exactly what a clock needs.

How a rock is dated

Take a mineral that incorporates a radioactive parent element when it crystallises. Over time the parent decays into a daughter, which accumulates in place. Measure how much of each is present and, knowing the half-life, you get the time since the mineral formed.

Two obvious objections arise immediately, and both have answers.

The first: how do you know how much daughter was there at the start? For some systems you do not have to assume. Zircon is the star case. When zircon crystallises it accepts uranium into its structure and strongly rejects lead, so a fresh zircon starts with essentially no lead at all. Any lead found in it later arrived by decay. Beyond that, isochron methods work on several samples at once: if they formed together but with different parent concentrations, the initial daughter amount can be solved for rather than assumed.

The second: what if parent or daughter atoms have leaked in or out? This is the real hazard, and geochronologists spend their careers on it. Zircon again helps, because it is exceptionally durable and holds its atoms tightly. And there is a built-in check: uranium comes in two isotopes that decay to two different lead isotopes at very different rates, so a single crystal contains two independent clocks. If they agree, the sample is undisturbed. If they disagree, the pattern of disagreement usually shows what happened and by how much. A method that can detect its own failures is far stronger than one that cannot.

What the clock actually records

A radiometric age is not always the age of the rock in the loose sense. It is the time since that mineral cooled below the temperature at which it stopped exchanging atoms with its surroundings, its closure temperature. Different minerals and different isotope systems close at different temperatures, from a few hundred degrees upward.

That sounds like a weakness and is a feature. Dating several systems in one rock reveals a cooling history rather than a single date: when the magma crystallised, when the mountain range it sat in was uplifted, when erosion brought it near the surface. The Himalaya of lesson 4.4 have been reconstructed this way.

Different clocks for different ages

Half-lives span an enormous range, so different isotopes suit different problems.

Carbon-14, at 5,730 years, dates once-living material back about fifty thousand years. It is made continuously in the atmosphere by cosmic rays, taken up by living things, and starts decaying at death. It is useless for rocks and for anything older than the last ice age, and it is invaluable for archaeology.

Potassium-argon and its refinement argon-argon, at 1.25 billion years, date volcanic ash layers, which is how the hominin fossils of chapter six are placed in time.

Uranium-lead, at 704 million and 4.47 billion years for the two chains, handles the oldest material there is, and it is what dated the meteorites, the zircons and the Deccan.

Rubidium-strontium, at 48 billion years, and samarium-neodymium reach further still, and are used on whole rocks.

The tests that could have failed

Any of this could have come apart, and it has not.

Different isotope systems, with different chemistry and different half-lives, applied to the same rock, give the same answer. Radiometric dates agree with independent chronometers where they overlap: tree rings counted one by one, annual layers in ice cores, annual sediment couplets in lake beds. The sequence of dated rocks matches the order that stratigraphy established from layering alone, worked out long before anyone could date anything. And the two uranium chains inside a single zircon agree with each other.

None of these agreements was arranged. They are what you would expect if the clocks are real, and there is no reason for them if the ages are wrong.

The man who dated the Earth and cleaned the air

The age of the Earth was fixed by Clair Patterson in 1956, and getting there required something unglamorous. Lead is everywhere: in dust, in solder, in glassware, in the air of an ordinary laboratory. Patterson’s measurements kept being ruined by contamination, so he built what amounted to the first ultraclean laboratory in order to measure his samples at all. His answer, 4.55 billion years, has stood for seventy years with only small refinement.

He then noticed something about the contamination itself. Lead levels in the modern environment were vastly higher than in old ice and deep-sea sediment, and the source was tetraethyl lead in petrol. Patterson spent much of the rest of his career fighting the industry that produced it, and lost funding for doing so. Leaded petrol was eventually banned nearly everywhere. The measurement that gave us the age of the planet also, indirectly, took the lead out of the air.

Deeper

The decay equation

The number of parent atoms falls as N = N₀e^(−λt), where λ is the decay constant and the half-life is ln2/λ. In practice one measures the present parent and daughter, so the working equation is D = D₀ + N(e^(λt) − 1), where D₀ is the initial daughter. Solving for t requires knowing D₀, which is why systems with negligible D₀, such as uranium-lead in zircon, or methods that solve for it, such as isochrons, are preferred. Decay constants are known to a fraction of a percent for the important systems, and their uncertainty now limits the accuracy of the best ages.

Concordia and discordia

Because uranium-238 and uranium-235 decay to lead-206 and lead-207 at different rates, a closed system plots on a curve in lead-uranium space called the concordia. A crystal that has lost some lead, typically through radiation damage and later fluid interaction, plots off the curve on a straight line whose intersections with the concordia give both the original crystallisation age and the age of the disturbance. Chemical abrasion, in which damaged domains are removed before analysis, has reduced this problem substantially, and modern zircon ages on favourable samples reach uncertainties of a few tenths of a percent, which is what allowed the Deccan eruptions to be resolved into pulses in lesson 4.4.

Closure temperature and thermochronology

Closure temperature depends on diffusion kinetics, grain size and cooling rate. Approximate values include about 900 degrees Celsius for uranium-lead in zircon, 500 for argon-argon in hornblende, 300 for argon-argon in biotite, 240 for fission tracks in zircon, 110 for fission tracks in apatite and about 70 for uranium-thorium-helium in apatite. Applying several systems to one sample yields a cooling path, and applying them across a mountain range yields exhumation rates. This is the standard toolkit for quantifying how fast the Himalaya have been rising and eroding.

Radiocarbon calibration

Atmospheric carbon-14 production varies with solar activity and the geomagnetic field, and the reservoir has been diluted by fossil carbon since industrialisation and spiked by atmospheric nuclear testing in the 1950s and 1960s. Raw radiocarbon years therefore differ from calendar years, and are converted using calibration curves built from tree rings, corals and speleothems dated independently. The current international curves extend past 50,000 years. The bomb spike has an incidental use: it lets tissues formed since 1955 be dated to within a few years, which has been used to measure cell turnover in the human body.

Why the agreements matter

Consider what it would take for radiometric ages to be systematically wrong. Decay constants would have to have changed, in different ways for different isotopes, in exactly the manner needed to keep independent systems agreeing with each other, with tree rings, with ice layers, with stratigraphic order and with astronomical cycles recorded in sediment. No physical mechanism can do that; decay rates have been tested under extreme conditions and in the deep past by the natural fission reactor at Oklo in Gabon, which ran two billion years ago and constrains any change in nuclear parameters since.

Frontier

Pushing the precision

The limiting uncertainty on the best uranium-lead ages is now the decay constants themselves, which are being re-measured. Better precision matters for questions where events must be ordered within tens of thousands of years across millions of years of elapsed time, such as whether Deccan volcanism preceded or followed the Chicxulub impact.

Dating the undatable

Many rocks contain no mineral suitable for the standard methods. Advances in dating individual mineral grains in place, in dating carbonates and clays directly, and in luminescence methods for sediment deposition are extending coverage to rock types and time ranges that were previously beyond reach.

Transcript and shot list

Target runtime about 9 minutes. Built as: establish that the Earth is old, show how the clocks work, answer the two obvious objections properly, then show the cross-checks nobody arranged. Shot types: ANIM is rendered from code with Manim, AI is a generated shot from Higgsfield with the shared style reference, REAL is agency imagery with credit, CAM is the narrator.

[00:00] S01 · ANIM · A list of the ages quoted so far in the course: 13.8 billion, 4.567 billion, 4.4 billion, 66 million; each with a question mark that then resolves.

Every number this course has given you came from a clock that was running inside a rock the whole time. Thirteen point eight billion years for the universe. Four point five six seven billion for the solar system. Four point four billion for the oldest crystals. Sixty-six million for the Deccan Traps. Those aren’t estimates in the loose sense, and nobody worked them out by assuming how fast mud piles up. And by the end of this you’ll know exactly why you cannot radiocarbon-date a dinosaur.

[00:40] S02 · REAL · Siccar Point, Scotland: near-vertical grey beds truncated and overlain by gently dipping red sandstone (photograph; confirm licence and credit).

The idea that Earth is old came first, and it came from looking at a rock like this. In the 1780s a Scottish farmer and physician, James Hutton, examined an outcrop where near-vertical layers are cut off flat and overlain by gently sloping younger ones. To make that, he reasoned, sediment must be laid down flat, buried, hardened, tilted on end, raised up, worn flat by erosion, submerged again, and buried under new sediment. Every step is slow. And all of it had to happen before the rock on top even started. He had no numbers, and he was right about the scale.

[01:27] S03 · ANIM · Kelvin’s cooling calculation giving tens of millions of years; Darwin and the geologists needing far more; then radioactivity discovered, generating heat and breaking the calculation.

The nineteenth century turned this into a fight. Darwin needed vast time for natural selection. Geologists agreed. And Lord Kelvin, the most authoritative physicist alive, calculated how long a molten Earth would take to cool and got tens of millions of years. Far too short. His physics was sound. His answer was wrong, because nobody knew the Earth contains radioactive elements making heat continuously. And the very discovery that ruined his calculation handed over the clock that settled the argument.

[02:04] S04 · ANIM · A population of unstable nuclei halving at regular intervals; then the same sample heated, squeezed, dissolved in acid, with the decay rate unchanged.

Here’s why decay makes a good clock. An individual nucleus decays at random, and you can never say when. But in any real sample there are astronomical numbers of atoms, and the fraction going per unit time is fixed with great precision. That’s the half-life. And the crucial thing is indifference. Chemical reactions speed up when you heat them. Nuclear decay doesn’t care. Heat a mineral, squeeze it, dissolve it in acid: the rate is unchanged, because it happens in the nucleus, insulated from the electrons that do chemistry. A rate nothing in geology can alter is exactly what a clock needs.

[02:51] S05 · ANIM · A zircon crystallising: uranium accepted into the lattice, lead rejected; over time lead accumulates in place; a ratio is measured.

So: take a mineral that grabs a radioactive parent when it crystallises. The parent decays into a daughter, which piles up in place. Measure both, know the half-life, get the time since it formed. Two obvious objections, and both have answers.

[03:10] S06 · ANIM · Objection one: how much daughter at the start? Zircon starting with essentially zero lead; then an isochron plot solving for the initial amount from several samples.

First: how do you know how much daughter was there at the start? For some systems you don’t have to assume. Zircon is the star case: when it crystallises it takes in uranium and strongly rejects lead, so a fresh zircon starts with essentially none. Any lead in it arrived by decay. And for other systems, isochron methods use several samples that formed together with different parent concentrations, and solve for the starting amount instead of assuming it.

[03:46] S07 · ANIM · Objection two: leakage. Two uranium chains in one crystal plotted on a concordia curve; a disturbed grain falls off the curve on a line whose intercepts reveal both the original age and the disturbance.

Second: what if atoms have leaked in or out? That’s the real hazard, and geochronologists spend careers on it. Zircon helps, because it’s exceptionally tough. But here’s the better answer: uranium comes in two isotopes that decay to two different lead isotopes at very different rates. So a single crystal contains two independent clocks. If they agree, the sample is undisturbed. If they disagree, the pattern usually shows what happened and by how much. A method that can detect its own failures is far stronger than one that can’t.

[04:27] S08 · ANIM · Several minerals in one rock closing at different temperatures: zircon at ~900 °C, biotite at ~300, apatite at ~70; a cooling path emerges; the path applied to a Himalayan sample.

One subtlety worth knowing. A radiometric age isn’t always the age of the rock in the loose sense. It’s the time since that mineral cooled below the temperature where it stopped swapping atoms with its surroundings. Different minerals close at different temperatures. That sounds like a weakness and it’s a feature: date several systems in one rock and you get a cooling history, not a date. When the magma crystallised. When the mountains rose. When erosion brought it near the surface. That’s how the uplift of the Himalaya has been reconstructed.

[05:09] S09 · ANIM · A ladder of half-lives: carbon-14 at 5,730 years for archaeology; potassium-argon at 1.25 billion for volcanic ash and hominin fossils; uranium-lead for the oldest rock; rubidium-strontium at 48 billion.

Different clocks for different jobs. Carbon-14, five thousand seven hundred and thirty years, dates once-living material back about fifty thousand. Useless for rocks, invaluable for archaeology. Potassium-argon, one and a quarter billion years, dates volcanic ash, which is how the hominin fossils in chapter six get their ages. Uranium-lead handles the oldest material there is. And that’s the answer to a common question: no, you cannot radiocarbon-date a dinosaur. After about ten half-lives there’s nothing left to measure, and dinosaurs are sixty-six million years old.

[05:48] S10 · ANIM · Cross-checks converging: different isotope systems on one rock; tree rings; ice-core layers; varves; stratigraphic order; the two uranium chains.

Any of this could have fallen apart, and it hasn’t. Different isotope systems, with different chemistry and different half-lives, give the same answer on the same rock. Radiometric dates agree with tree rings counted one by one, with annual layers in ice cores, with annual mud couplets in lake beds. The order of dated rocks matches the sequence stratigraphers worked out from layering alone, long before anyone could date anything. And the two uranium chains inside a single crystal agree with each other. None of that was arranged. It’s what you’d expect if the clocks are real, and there’s no reason for it if they aren’t.

[06:37] S11 · AI · A 1950s ultraclean laboratory; a researcher in whites; then a graph of environmental lead rising through the twentieth century and falling after regulation.

One last story. The age of the Earth was fixed by Clair Patterson in 1956, and getting there needed something unglamorous. Lead is everywhere: in dust, in solder, in glassware, in ordinary laboratory air. His measurements kept being ruined by contamination, so he built what amounted to the first ultraclean laboratory just to measure his samples. His answer, four point five five billion years, has stood for seventy years. And then he noticed something about the contamination itself: lead in the modern environment was vastly higher than in old ice and deep-sea mud, and the source was leaded petrol. He spent much of the rest of his career fighting the industry that made it, and lost funding for it. Leaded petrol is now banned almost everywhere. The measurement that gave us the age of the planet also, indirectly, took the lead out of the air.

[07:44] S12 · ANIM · Labels: ESTABLISHED for radiometric dating and the cross-checks; ACTIVE RESEARCH for pushing precision and dating difficult rocks.

The scorecard. That radioactive decay is a reliable clock, that the methods cross-check, and that the resulting ages are sound: established, and about as firmly as anything in this course. Squeezing the precision further, which now runs up against how well the decay constants themselves are known, and dating rocks that contain no convenient mineral: active research.

[08:11] S13 · ANIM · A single zircon a tenth of a millimetre across; a caption: 4.4 billion years, measured to a few million; the calendar returns.

So when this course says four point five six seven billion years, that’s not a guess. It’s a measurement, from a crystal you could lose in the creases of your palm. Next lesson: all of it together. Rock, water, air and life, working as one machine.

Shot list

One AI shot, eleven code-rendered shots, one photograph. Estimated Higgsfield use: 2 to 4 generations including retries.