Appearance
1.1 — How We Know: Reading Rock and Time
On a cold Scottish shore in 1788, a farmer and physician named James Hutton took two friends out in a boat to look at a cliff. What they saw at Siccar Point was this: a set of rock layers standing almost vertically, sliced off flat at the top, and lying across that cut surface a second set of layers, nearly horizontal, running out to sea.
Hutton's companion John Playfair wrote afterwards that "the mind seemed to grow giddy by looking so far into the abyss of time." He had understood what the picture demanded. Those bottom layers had to be laid down flat as sediment on a sea floor, because that is the only way layers form. Then they had to be buried, hardened into rock, tilted until they stood on end, raised above the sea, and worn flat by weather. Then the sea had to return and lay a completely new set of layers over the cut. Then all of that had to be lifted up again to make the cliff he was standing on.
Not one of those steps is fast. Sediment settles in millimetres per year. Weather grinds rock down in fractions of a millimetre per year. To get that one picture you need an amount of time that has no relationship to human experience, and in 1788 the accepted age of the Earth in Britain was about six thousand years, calculated from biblical genealogies by Archbishop Ussher a century earlier.
Hutton had no way to put a number on it. What he had was the argument, and the argument is the foundation of this entire volume: you can read the past out of the physical evidence it left behind, if you understand the processes that made the evidence. This page is about how that reading is done — for rock, for time, for climate, and eventually for human history — because a claim in this book is only worth as much as the method that produced it.
The first tool: things pile up in order
The oldest idea in geology is almost insultingly simple. A layer of sediment lies on top of older layers and underneath younger ones. Nicolas Steno wrote this down in 1669 and it is called the law of superposition. Nobody has ever found a reason to doubt it, because the alternative would require sand to settle underneath rock that is already there.
Two companions come with it. Layers start out roughly horizontal, because that is what loose sediment does under gravity in water — so a layer standing vertically has been tilted since, and something must have tilted it. And a layer, or a crack, that cuts through another one is younger than the thing it cuts, because you cannot cut what does not yet exist. A vein of quartz running across five layers postdates all five.
With those three rules you can order the events at any one cliff. What you cannot yet do is compare two cliffs a thousand kilometres apart, and you cannot attach a date to anything.
The second tool: fossils are clocks made of animals
The man who solved the first problem was a canal surveyor. William Smith spent the 1790s digging channels across England for coal barges, which meant cutting through the same rock layers over and over in different counties. He noticed that each layer carried its own particular set of shells, and — this is the crucial part — the sets always appeared in the same order, no matter where in England he dug.
That gives you a matching tool. If a layer in Somerset holds the same distinctive fossils as a layer in Yorkshire, the two layers formed at about the same time, even though there is no continuous rock between them. Smith published the first geological map of a whole country in 1815, and the principle is now called faunal succession.
The reason it works is evolution, though Smith did not know that and Darwin's book was still forty-four years away. Species appear, spread, and go extinct, and once a species is extinct it never returns. So the presence of a particular species in a rock is a stamp saying "this formed within the window when that animal was alive". The most useful stamps come from index fossils — species that were widespread, common, easy to identify, and short-lived as species. Some trilobites, ammonites and microscopic sea creatures called foraminifera work beautifully; a clam that survived unchanged for a hundred million years tells you almost nothing.
This is how the geological calendar — Cambrian, Jurassic, Cretaceous and the rest — was assembled during the nineteenth century, entirely by fossils and layer order. Every division of that calendar was named and put in sequence before anybody could date a single one of them in years. Chapter 1.10 works through the calendar itself.
The third tool: atoms that keep time
The dating problem was solved by the discovery of radioactivity, and it was solved almost immediately once physicists realised what they had.
Some atoms are unstable. A radioactive atom will, at some unpredictable moment, throw off part of itself and become a different element. You cannot predict when any single atom will do this. But in a lump of rock containing quadrillions of them, the proportion that decays in a given time is fixed, unchangeable, and specific to that isotope. That fixed proportion is expressed as a half-life: the time it takes for half of whatever is present to decay.
Write N_0 for how many radioactive atoms ("parent" atoms) the rock started with, N for how many are left now, t for the time elapsed and t_{1/2} for the half-life. Then
N = N_0 \left(\tfrac{1}{2}\right)^{t/t_{1/2}}
Read aloud, that is: the number left equals the number you started with, times one half raised to the power of how many half-lives have gone by. One half-life, half remains. Two half-lives, a quarter. Three, an eighth.
The trick that makes this usable is that the parent does not vanish — it becomes something. Uranium-238 becomes lead-206. Potassium-40 becomes argon-40. So the rock contains both a shrinking stock of parent atoms and a growing stock of daughter atoms, and the ratio between them is a clock reading. Rearranging the equation to solve for the age gives
t = t_{1/2}\,\frac{\ln\!\left(1 + \dfrac{D}{N}\right)}{\ln 2}
where D is the number of daughter atoms measured and N the number of parent atoms still there. In words: measure how much daughter has accumulated relative to remaining parent, take the logarithm of one plus that ratio, divide by the logarithm of two, and multiply by the half-life. The \ln is the natural logarithm, which is the function that answers "what power do I raise the number 2.718 to in order to get this?" — the only thing you need from it here is that it undoes an exponential, which is exactly what decay is.
Different clocks for different jobs, because a clock is only useful over a range comparable to its half-life:
| Clock | Half-life | Good for |
|---|---|---|
| Carbon-14 | 5,730 years | Wood, bone, cloth up to ~50,000 years |
| Potassium-40 → argon-40 | 1.25 billion years | Volcanic ash, lava; dating fossil beds |
| Uranium-238 → lead-206 | 4.47 billion years | The oldest rocks and minerals |
| Uranium-235 → lead-207 | 704 million years | Runs alongside U-238 as a cross-check |
Carbon-14 deserves a sentence of its own because it is the one that dates human history. It is made continuously in the upper atmosphere when cosmic rays hit nitrogen, so living things take it up while they are alive and stop the moment they die. The clock starts at death. After about ten half-lives — 57,000 years — there is too little left to measure, which is why carbon dating cannot touch dinosaurs and why anyone who claims a carbon date for one is telling you something is wrong with their source.
Why should you believe any of it? Because the clocks are checked against each other and against things whose age is known independently.
- Uranium-lead runs two clocks in one mineral. A grain of zircon contains both U-238 and U-235, decaying at rates that differ by a factor of six, into two different lead isotopes. If the grain has leaked atoms or been reheated, the two clocks disagree and the sample is thrown out. When they agree, the agreement is not a coincidence you can arrange.
- Carbon-14 was calibrated against tree rings, which are a physical count of years you can do with your eyes, going back over 12,000 years in continuous overlapping sequences of oak and pine. The calibration also revealed that atmospheric carbon-14 is not perfectly constant, which is why raw radiocarbon dates are corrected before use.
- Decay rates do not vary. They have been tested under pressure, at high and low temperature, in strong magnetic and electric fields, and in chemical combinations. The rate is set by the physics of the nucleus, which is untouched by anything happening to the electrons around it — the only things that chemistry, heat and pressure can reach.
The oldest rock formations found so far date to about 4.0 billion years, in northern Canada. The oldest individual mineral grains — zircons from the Jack Hills in Western Australia, worn out of older rock and recycled into a younger sandstone — date to about 4.4 billion years. That single number is the anchor for the next chapter.
The fourth tool: layered archives that are not rock
Rock is not the only thing that stacks up in order.
Ice. Snow falling on Antarctica and Greenland never melts. It compacts under each year's new fall into a layer of ice, and trapped in that ice are actual bubbles of the atmosphere of the year it fell. Drill a core, count the layers, and you have a direct sample of ancient air — not an estimate of it, a sample. The Vostok and EPICA cores in Antarctica reach back over 800,000 years, and Chapter 2.8 uses them.
Tree rings. One ring per year, thick in a good year and thin in a bad one. Because the pattern of good and bad years is shared across a region, a living tree's rings can be matched to the outer rings of an older dead log, and that log's inner rings to an older one still, chaining back thousands of years. This is how a wooden beam in an old building can be dated to a specific year, sometimes to a season.
Lake and sea-floor mud. Many lakes deposit a pale layer in summer and a dark one in winter, giving countable annual pairs called varves. Deep-sea mud accumulates far more slowly but far more steadily, and the shells in it carry a chemical record of the temperature of the water they grew in.
Written records, coins and pottery — the historian's version of the same logic. A coin hoard cannot have been buried before the date on its newest coin. A pottery style, like an index fossil, has a period when it was made and after which it was not.
When these independent archives agree, that agreement is the evidence. A volcanic eruption can appear as an ash layer in a lake bed, a sulphuric acid spike in Greenland ice, a frost-damaged ring in Californian pines, and a line in a Chinese chronicle about a dimmed sun. Four systems with nothing in common except that they were all there.
The fifth tool: chemistry as a fingerprint
A great deal of what this volume states about the deep past comes from ratios of isotopes — atoms of the same element with different weights, which behave identically in chemistry but differently in physics.
Oxygen isotopes are a thermometer. Water made with the lighter oxygen-16 evaporates a little more readily than water made with the heavier oxygen-18. In a cold world, a lot of that light water is locked up in ice sheets, so the ocean left behind is enriched in the heavy kind, and sea creatures build their shells from it. Measuring the oxygen-18 to oxygen-16 ratio in a fossil shell therefore tells you, with calibration, how cold the world was when that animal lived.
Carbon isotopes track life. Living things prefer the lighter carbon-12 when they build tissue, so anything biological is depleted in carbon-13. A sudden shift in the carbon-13 ratio in the rock record means the global carbon cycle was violently disturbed, which is why such a shift is one of the fingerprints of a mass extinction.
Strontium tells you where somebody grew up. The strontium isotope ratio in bedrock varies from region to region, it passes into the local water and plants, and it is incorporated into teeth while they form in childhood and then never changes. Test the tooth enamel of a skeleton and you can often say whether that person was born where they were buried. This is how archaeologists showed that some of the people buried near Stonehenge came from far away.
What honest uncertainty looks like
Every method has a range where it is reliable and edges where it is not, and this volume states which is which.
A date has an error bar and the error bar is part of the date. "3.66 \pm 0.02 million years" means the evidence constrains it to a window, not a moment. Quoting the middle number alone as though it were exact is one of the commonest ways popular writing misleads.
Absence of evidence is weak evidence. Fossilisation is extraordinarily rare — it needs rapid burial in the right chemistry, and then the rock must survive being melted, crushed or eroded for millions of years, and then somebody must find it. Soft-bodied creatures, forest animals and small populations are all systematically under-represented. "The earliest known X" almost always means the earliest found, and it moves whenever somebody digs a new hole.
Written history has the opposite problem: too much evidence, all of it interested. A rock has no motive. A royal inscription has nothing but motive. Chapter 16.4 is entirely about how to handle that, but the short version applies from here on — ask who wrote it, when, for whom, what they gained, and whether anything independent corroborates them.
And a good method tells you when it fails. The two uranium clocks disagreeing, radiocarbon needing tree-ring correction, oxygen ratios being thrown off by local salinity: these are not embarrassments hidden in footnotes, they are how the field found and fixed its own errors. A claim that has never been checked against anything is the one to distrust.
Where this shows up in your life
The gap between what your intuition can hold and what the evidence says is not a small gap here, and it is worth setting your scale before the next chapter.
Compress the whole history of the Earth into one calendar year, starting at midnight on 1 January. The planet forms in the first seconds. The first single-celled life appears around late February. For the next seven months, nothing but microbes. Complex animals show up in mid-November. Dinosaurs walk in around 13 December and are gone by the evening of 26 December. The first recognisable humans appear at about 8 p.m. on 31 December. All of agriculture, every city, every empire, every war and every book in this volume happens in the last thirty seconds before midnight. Your own life is the final tenth of a second.
That is not a poetic flourish, it is arithmetic on the dates in this chapter, and it is the reason the next few pages spend their time on rock rather than on kings.
What the next page covers
We now have the tools: layer order, fossil matching, radioactive clocks, ice and tree archives, and isotope chemistry. The oldest mineral grain reads 4.4 billion years, which means the planet those grains formed in was already there. Chapter 1.2 goes back further than any rock — to a disc of dust around a young star, the process that turned that dust into a planet in a few tens of millions of years, and the collision that gave the Earth its Moon, its tilt, and eventually its tides.