The Wreck Nobody Was Looking For

In the spring of 1900, a group of Greek sponge divers were caught in a storm near a small, rocky island called Antikythera — a speck of land in the strait between Crete and the Peloponnese. They anchored and waited. When the weather broke, one diver went down to look for sponges.

He came back up talking about corpses on the seabed. And horses.

What he had found was a Roman-era shipwreck, lying at depths between 40 and 50 meters. The Greek government sent in a recovery team. Over the next year, divers hauled up an extraordinary cargo: bronze and marble statues, fine glassware, coins, amphorae, jewelry, luxury goods from across the ancient Mediterranean world.

And one other thing. A wooden box, roughly the size of a shoebox. Inside it was a lump of corroded bronze — misshapen, crumbling at the edges, breaking apart as it was lifted from the water. It looked like debris. It was catalogued, transported to the National Archaeological Museum in Athens, and largely ignored while archaeologists focused on the statues.

That lump sat in Athens for months before anyone looked at it closely. When they did, they noticed something strange inside the corroded mass. The unmistakable outline of a gear wheel.

What Was Actually Inside

The device that eventually emerged from years of careful study was unlike anything in the known archaeological record.

It was a hand-cranked mechanical calculator, built around 100 BC. The user turned a handle on the side, which set a system of interlocking bronze gear trains in motion. Those gears moved pointers across dials on both the front and back of the device, displaying astronomical and calendar information for any chosen date — past or future.

The mechanism could predict solar and lunar eclipses with precision. It tracked the irregular orbit of the Moon — modeling a mathematical theory developed by the Greek astronomer Hipparchus, which accounted for the fact that the Moon does not move at a constant speed. It modeled the 19-year Metonic cycle, which aligns the solar year with the lunar calendar. It tracked the 223-month Saros cycle, used to predict eclipses. It could pinpoint the four-year cycle of the Olympic Games — and three other Panhellenic athletic festivals.

Today, 82 fragments survive. They contain more than 30 confirmed bronze gears, some with over 200 individual teeth, each one filed by hand. The original device is estimated to have contained over 40 gears in total.

Nothing of comparable mechanical complexity would appear anywhere in the world for the next 1,400 years.

The Machine That Took a Century to Understand

The problem was not finding the mechanism. It was understanding it.

For decades after its recovery, researchers could see gears inside the fragments — but not how they connected, not how many there were, not what the inscriptions said. Early attempts at reconstruction produced incomplete models. The back section of the device was partially decoded. The front remained a mystery.

The decisive breakthrough came in 2005, when an international team of researchers brought an eight-tonne microfocus CT scanner to the National Archaeological Museum in Athens. The scanner — normally used for industrial inspection — passed X-rays through every fragment at thousands of angles, building detailed three-dimensional maps of the interior. Gear teeth that had been hidden for 2,000 years inside fused, corroded bronze were suddenly visible. Thousands of inscribed Greek characters — functioning as the original user manual — could finally be read in full.

What the scans revealed pushed the complexity further than anyone had assumed. The inscriptions described the motions of the Sun, Moon, and all five planets known in antiquity — Mercury, Venus, Mars, Jupiter, and Saturn — and specified how they were displayed on the front of the device as a working model of the Greek cosmos.

In 2021, Professor Tony Freeth and his team at University College London published a reconstruction of the front mechanism in the journal Scientific Reports. They proposed that the motions of all five planets were encoded through a system of concentric gear rings — layers of gears turning at different speeds inside each other, each representing a different planetary cycle. The Venus display alone used a 462-year planetary period — a period relation previously unknown in either Babylonian or Greek astronomical records, apparently invented specifically to make the mechanism work within the physical constraints of its bronze gears.

The UCL reconstruction was the first model that satisfied all the physical evidence and matched the inscriptions. It described a device that fused Babylonian astronomical data, mathematics from Plato’s Academy, and the mechanical traditions of the Greek island of Rhodes into a single handheld bronze machine.

Who Built It — and What Happened to the Tradition

No maker’s mark survives. No workshop records. No name.

The leading candidates trace back to Rhodes, which was a major center of astronomy and mechanics in the Hellenistic period. The astronomer Hipparchus — who catalogued thousands of stars, developed the first detailed mathematical model of the Moon’s irregular orbit, and worked on Rhodes — is the most cited intellectual ancestor of the device. The gears that model the Moon’s motion directly encode his theory.

The Roman statesman Cicero wrote that during his time on Rhodes as a young man, he had seen a similar bronze device — a planetarium that could show the positions of the Sun, Moon, and planets. Other ancient sources describe a mechanical sphere built by Archimedes. For a long time, historians treated those accounts as exaggeration. The Antikythera Mechanism suggests they were not.

The ship that carried it sank sometime between 70 and 50 BC, most likely en route from the Greek world toward Rome. Who owned it, who built it, and who was it made for — none of these questions have definitive answers.

What is certain is that after the Antikythera Mechanism, nothing like it appears in the historical record. The tradition of complex mechanical astronomy that produced it seems to have simply vanished. The first European mechanical clocks approaching its gear complexity — Richard of Wallingford’s clock at St Albans Abbey and Giovanni de’ Dondi’s astrarium in Padua — were built in the 1320s and 1360s. More than 1,400 years later.

The Greeks built a working computer. Then they lost the knowledge. And we spent 120 years figuring out what we’d found.