How we know

Exhibit 01

Radiocarbon dating

Cosmic rays striking nitrogen in the air make carbon-14, and living things take it in with the rest of their carbon. When a plant or animal dies, it stops taking carbon in and its carbon-14 decays. Half is gone after 5,730 years.

Willard Libby proposed the method in 1946 and tested it on things of known age: tree rings, and timber from the funeral boat of the pharaoh Senusret III. It can now date material up to about 50,000 years old.

A lab's result is not yet a calendar date. It is reported in years “before present”, meaning before 1950, as if the air's carbon-14 had never changed and using Libby's first half-life, 5,568 years. Neither is quite true, so dates are calibrated against tree rings, whose years are known, and other records. The answer is a range: Oxford's example, a sample measured at 3000 ± 30 BP, comes out between 1375 and 1129 BC, at 95% confidence.

Try it: let a sample age

Years since it died: 5,730

50%of its carbon-14 left

Each dot: 1% of the carbon-14 it had alive

Exhibit 02

Ancient DNA

DNA starts to fall apart when its owner dies. What survives comes out in short, damaged pieces, mixed with modern DNA from anyone who touched the bone, and reading it destroys part of the sample.

It lasts best where it is very cold and dry: in permafrost and caves. The oldest human DNA read so far is from a Neanderthal about 400,000 years old; mammoth DNA from the Siberian permafrost is over a million years old. For most of human evolution our ancestors lived in the tropics and subtropics, where it rarely survives.

Where it does survive, it can reach people alive today. Europeans and Asians carry about 1 to 4% Neanderthal DNA; Melanesians and Aboriginal Australians carry about 3 to 5% Denisovan DNA.

Try it: what can a bone tell us?

Diagram: whole where DNA was read, scattered where none was recovered.

50,000 to 30,000 years old

What it told us: A girl from a kind of human no one had seen before. The Denisovans are the first ancient humans known from their genes alone; the genome was published in 2010.
What it couldn't: What she looked like. A few bones and teeth are all the Denisovans have left, too few to compare their genome properly with ours.

Source: Australian Museum, “The Denisovans”; Smithsonian Human Origins, “Ancient DNA and Neanderthals”

Exhibit 03

X-rays and CT

An X-ray sees through what can't be opened. For ten years after he was found, Ötzi was thought to have died in an accident. In 2001 an X-ray showed a flint arrowhead in his left shoulder. Then a 2 cm entry wound was found in his back: the arrow had cut an artery under the collarbone, and he bled to death within minutes.

Inside a lump of corroded bronze from the Antikythera wreck were gearwheels. X-rays of its fragments, made by Charalambos and Emily Karakalos, let the physicist Derek de Solla Price identify 30 surviving gears. In 2005, X-ray computed tomography, CT, took the 82 fragments slice by slice. It decoded the gearing at the back and found inscriptions describing the motions of the Sun, the Moon and the five planets known in antiquity.

Try it: run the scan

Reconstruction of a chest X-ray, not the 2001 image. The mark shows roughly where the arrowhead lies; we placed it.

Scan position: 30%

Exhibit 04

Reading the layers

In 1669 Nicolaus Steno set out the rule now called superposition: layers settle one on another, so each is older than the one above it and younger than the one below. He also saw that layers are laid down nearly flat, so a tilted or folded layer was moved after it formed.

Archaeologists dig by the same rule. The layers give an order, not an age in years; the years come from other methods, such as radiocarbon dates on what each layer holds.

The section here is the National Park Service's teaching example, not a real site.

Try it: dig down

GrassToday
A city parkEarly 20th century
A town1890s
An explorers' base1650s
Flood siltEarly 1500s
A hunting campAbout 2,000 years ago
River siltBefore anyone

Depth: Grass, today

Grass, today

The surface. The archaeological sites of the future are being made here.

No finds in this layer.

Exhibit 05

Reading old sources

Much of what is said about the Library of Alexandria's end comes from writers who lived long after it. Three different destructions are told: Caesar's fire in 48 or 47 BC, the sack of the Serapeum in AD 391, and the Arab conquest of 641. Each account has problems.

So we ask of every source when it was written, how far from the events, what its words actually say, and whether anything else backs it up. A late or doubtful source is flagged in our scripts, and the claim is worded to match.

The historian Roger Bagnall argues for something less dramatic: slow neglect and decay. He points to the early 270s AD, when much of the city was destroyed in fighting, as the likeliest time for library buildings to have been wrecked. Some historians give more weight to single destructions.

Try it: how far from the event?

AD 1AD 500AD 1000AD 1500

Plutarch

Cassius Dio

Ammianus Marcellinus

Edward Gibbon

Bar Hebraeus (Abu'l-Faraj)

The event it describes When it was written

Plutarch, Life of Caesar 49

About: Caesar's fire
Written: At least a century after
Says: Caesar's fire “spread from the dockyards and destroyed the great library”.
Watch for: Short on detail, and written a century or more later. Nobody knows where the library stood, so nothing confirms the fire reached it.