In October 1900, a party of sponge divers from the Greek island of Symi sheltered from a storm off the small island of Antikythera, between Crete and the Peloponnese. When the weather cleared, they dove to about 60 meters and found a Roman-era shipwreck. Among the bronze and marble statues they recovered over the following months was a corroded lump of metal that nobody quite knew what to do with.

The lump was deposited at the National Archaeological Museum in Athens. In 1902, the archaeologist Valerios Stais noticed gear wheels visible in cracks on its surface. He proposed that the object was some kind of astronomical instrument and was largely ignored.

For half a century, the Antikythera Mechanism — as it came to be called — sat in storage as an awkward curiosity. The conventional wisdom of the period held that ancient Greeks did not build complex geared mechanisms. The object, by this view, must be a later intrusion into the wreck.

The wreck itself was eventually dated by the amphorae and coins recovered with it. It sank between 70 and 60 BC. The mechanism, if it belonged to the ship, was the work of someone before that date.

The first serious analysis came from the British historian Derek de Solla Price, who began studying the fragments in the late 1950s and published a major paper in 1959. Price, using gamma radiography to see inside the corroded mass, identified more than thirty interlocking gears and proposed that the mechanism was a computational device for predicting astronomical phenomena.

His conclusion was so unexpected that it took another generation for the field to accept it.

His conclusion was so unexpected that it took another generation for the field to accept it.

The transformation came with the Antikythera Mechanism Research Project, founded in 2005 by an international team including Tony Freeth of University College London and Mike Edmunds of Cardiff University. The project commissioned a custom-built X-ray tomography machine, weighing eight tons and shipped to Athens, to scan the fragments at resolutions impossible by earlier methods.

The scans, published in Nature in 2006 and updated through subsequent papers, revealed inscriptions on the inner surfaces of the corroded metal — text invisible from outside the device. The text was a kind of user's manual. It described what each dial showed and how to interpret the readings.

The mechanism, it turned out, was an astronomical calculator of extraordinary sophistication. It tracked the positions of the sun and moon, predicted lunar and solar eclipses, displayed the cycle of the Olympic Games, and showed the positions of the five planets known to the ancient world. It used the Metonic cycle of 19 years to reconcile the lunar and solar calendars and the Saros cycle of 18 years to predict eclipses.

It did all of this with a system of nested gears, including a differential gear that was not reinvented in Europe until the sixteenth century.

The most recent work, published by the University College London team in 2021 and refined in subsequent papers through 2024, has reconstructed the front face of the mechanism, which had been less well preserved than the back. The new model shows a complete cosmos display, with the sun, moon, and planets moving around a central earth in accordance with the Greek epicyclic models of the period.

One puzzle the new scans helped resolve concerned the planetary mechanism. The surviving fragments contained gears whose tooth counts did not obviously correspond to known astronomical cycles. By cross-referencing inscriptions with the gear ratios, the researchers identified the cycles being modeled and accounted for nearly every gear. A small number remain ambiguous, but the gross architecture is now understood.

The question of who built the mechanism remains open. The most popular candidate is the school of Posidonius of Rhodes, a Stoic philosopher and astronomer whose work was known to Cicero. Cicero, writing about a generation after the mechanism was lost, describes a similar device built by Posidonius. Some researchers have suggested Archimedes, who is also credited with astronomical machines, but Archimedes died in 212 BC, more than a century before the wreck.

The Greek city of Pergamon and the city of Syracuse have both been proposed as possible places of manufacture, with no decisive evidence for either.

What the mechanism demonstrates, beyond its specific functions, is that ancient mechanical sophistication was greater than the surviving evidence usually suggests. The Antikythera Mechanism is unique only in that it survived. Ancient texts mention similar devices, and the design of this one is too refined to have been a first attempt. There were others, and they are presumably at the bottoms of other harbors, dissolved into the seabed.

The fragments are now displayed at the National Archaeological Museum in Athens, alongside high-resolution reconstructions. The original device is roughly the size of a shoebox, and surprisingly unimpressive at first glance. Its significance is entirely in what it could do.

It would be more than a thousand years before any comparable mechanism would be built again. Until the scans, nobody quite believed that.