Elias Stadiatis surfaced off the coast of Antikythera in the spring of 1900 wearing a copper diving helmet and canvas suit, and told his captain he had just seen a heap of rotting corpses on the seabed. The captain, Dimitrios Kondos, went down himself. What Stadiatis had mistaken for bodies were bronze and marble statues, scattered across the wreck of a Roman cargo ship that had sunk sometime around 60 BC while carrying looted Greek treasures back to Italy.
The divers were sponge fishermen from the island of Symi, sheltering from a storm on their way to fishing grounds off North Africa. They surfaced with statues, coins, glassware, and one unremarkable-looking lump of corroded bronze the size of a shoebox.
The lump sat in the National Archaeological Museum in Athens for another year before anyone looked at it closely. In 1902, the archaeologist Valerios Stais noticed a gear wheel embedded in what everyone had assumed was a piece of rock. That gear wheel is the reason we now talk about the Antikythera mechanism as the world’s oldest known analogue computer.
A shipwreck full of stolen art
The ship the sponge divers found was a floating warehouse of first-century BC loot. Bronze statues. Marble figures. Amphorae of wine. Glass bowls. Coins from Pergamon and Ephesus. The best current dating puts the sinking at roughly 60 BC, though the mechanism itself may have been built decades earlier — possibly as early as 205 BC.
The divers who found it were part of a long Aegean tradition. Sponge diving in the Dodecanese islands of Kalymnos and Symi had been a working profession since antiquity. By the late nineteenth century, Greek spongers from those same islands were also emigrating to Florida, where they built Tarpon Springs into a Gulf Coast sponge capital that still runs a Greek Orthodox Epiphany procession every January.
So the people who pulled the mechanism up from 45 metres of water knew exactly what a two-thousand-year-old dive looked like. They just had no way of knowing what they had grabbed.
What Stais saw in 1902
The lump had cracked while drying out on a museum shelf. Valerios Stais peered into the crack and saw a bronze gear wheel with tiny triangular teeth. Inscribed around it, in ancient Greek, were astronomical terms — the kind of language you use to describe the movement of Venus, the phases of the moon, the cycle of eclipses.
For most of the twentieth century, no one could explain it. The Romans didn’t build things like this. Nobody did. The device sat in vitrines in Athens while scholars quarrelled about whether it was an astrolabe, a planetarium, or an elaborate fake. Some assumed it must have fallen into the wreck from a later century, because the idea of Hellenistic gearing that fine was simply not accepted.
X-rays cracked it open
Everything changed in 2006, when a team using high-resolution X-ray computed tomography scanned the surviving 82 fragments. The scans revealed the interior of a device that had been invisible for two millennia: at least 30 interlocking bronze gears, some no larger than a fingernail, arranged in a nest of concentric dials driven by a single hand crank.
Turn the crank forward, and the front dial showed the sun and moon moving through the zodiac, along with the moon’s phase. Turn it further, and pointers on the back dials tracked a 19-year Metonic cycle (the interval after which lunar and solar calendars realign), an 18-year Saros cycle predicting solar and lunar eclipses, and a four-year dial marking the Panhellenic games — including the Olympics at Olympia, the Pythian games at Delphi, and the Nemean and Isthmian games.
The device was a handheld model of the cosmos as the Greeks understood it, calibrated to keep the calendar of civic and religious life in sync with the sky.
The gravitational-wave clue
The calendar ring on the front of the mechanism has small evenly spaced holes running around its edge. For most of the past century, scholars assumed there were 365 of them — one for each day of the solar year, in line with the Egyptian calendar the Greeks knew well.
In 2024, a pair of physicists at the University of Glasgow ran the numbers using statistical tools borrowed from the analysis of gravitational-wave signals — the same methods used to sift real black-hole collisions out of instrumental noise at LIGO. Applied to high-resolution images of the surviving calendar-ring fragment, the analysis suggested the ring most likely had 354 holes, matching a lunar calendar rather than a solar one.
According to research co-author Graham Woan, an astrophysicist at the University of Glasgow, the team used modern astronomical analysis techniques to study the ancient device’s calendar ring.
354 holes is the length of a 12-month lunar year. It means the person who built the mechanism was thinking in the calendar system used across much of the Greek world, not the Egyptian solar year.
The teeth are triangles, and that’s a problem
For a long time, the Antikythera mechanism was described in almost worshipful terms — a precision instrument two millennia ahead of its time. A study posted in 2025 by Esteban Szigety and Gustavo Arenas at Argentina’s National University of Mar del Plata complicated that picture.
The two researchers built a computational model of the gear train and asked a simple question: given the shape and spacing of the teeth as reconstructed from the tomography scans, would this thing actually work?
The teeth are triangular. Modern gears use a curved involute shape that lets one tooth roll smoothly against the next. Triangular teeth catch and slip. Combine that with hand-cut spacing that isn’t perfectly uniform — because it was made by a person with a file, not by a CNC machine — and you get a device that, in simulation, jammed roughly 90 percent of the time before the solar pointer could complete four months of motion.
Szigety and Arenas were careful about what they were claiming. Two thousand years of seawater have replaced the original bronze with a corrosion product called atacamite — a copper chloride mineral with different physical properties from the original alloy. The gears are warped. Encrustations coat the teeth. The measurements the model depends on may be measuring damage, not design.
Szigety explained to Artnet that the mechanism’s corroded state, warped components, and encrustations make it difficult to obtain accurate measurements of the original design.
Toy, teaching tool, or working instrument?
The jamming finding revived a much older debate. If the mechanism didn’t really work, was it ever meant to? Some scholars have suggested it was a philosophical demonstration piece — a model of Hellenistic astronomy meant to impress a wealthy patron, or to teach students how the heavens fit together, rather than a field-usable calculator.
Szigety and Arenas don’t buy that. In their paper, Szigety and Arenas argue that it seems implausible someone would construct such a complex device if it didn’t function properly. More probable, they argue, is that the measurements pulled from the corroded fragments are underselling the original precision. Silicon Canals has written before about how the lump sat unrecognised in Athens for decades before anyone understood what it was, and the same shipwreck has been the subject of repeated re-examination as scanning technology has improved.
Who built it
Nobody knows the name of the maker. The inscriptions on the surviving fragments are in a Koine Greek dialect associated with the eastern Mediterranean, and the astronomical parameters encoded in the gearing match models attributed to Hipparchus, the astronomer working on Rhodes in the second century BC. Rhodes was also a known centre of mechanical engineering and a plausible port of origin for the doomed cargo ship.
Cicero, writing about a generation after the wreck went down, describes a similar device built by Archimedes at Syracuse — a bronze sphere that showed the movements of the sun, moon and five known planets. He also mentions a second instrument, made by the philosopher Posidonius on Rhodes, that displayed the same motions. If Cicero was telling the truth, the Antikythera mechanism was not one-of-a-kind. It was one surviving example of a small tradition of geared astronomical instruments that were rare, expensive, and almost entirely lost.
The 1,400-year silence
What makes the mechanism so strange is what came after it. Nothing. For roughly the next fourteen centuries, no known device combined multiple gear trains to model astronomical motion. Islamic astronomers built astrolabes with single gear pairs. Chinese engineers built water-driven clocks with escapements. But the specific idea of a hand-cranked, multi-gear computational instrument didn’t resurface in Europe until the astronomical clocks of the fourteenth century — Richard of Wallingford’s clock at St Albans Abbey, designed around 1330, and Giovanni de Dondi’s Astrarium in Padua, finished in 1364.
De Dondi’s Astrarium had seven dials showing the positions of the sun, moon and five planets. It took him sixteen years to build. It was, in effect, a Renaissance rediscovery of what a Hellenistic craftsman had already done fourteen hundred years earlier and shipped in the hold of a doomed Roman freighter.
What’s left
The 82 known fragments of the mechanism are behind glass in the National Archaeological Museum in Athens. The largest is about the size of a paperback book. The smallest pieces are chips no bigger than a coin, with gear teeth still visible under magnification.
Divers keep going back to the wreck. Expeditions in 2012, 2014, 2017 and again in the 2020s have pulled up more bronze, more marble, and human bones — at least four individuals, one of whose skull was intact enough that researchers attempted DNA extraction. The wreck is thought to hold more of the mechanism, or possibly a second one. The seabed has not given it up.
Somewhere on that ship, along with the statues and the wine jars and the coins, a Greek craftsman had packed the gears of a machine that could tell a Roman senator when the next eclipse would darken the sky over Rhodes, and when the runners would gather at Olympia. The ship sank. The craftsman’s name went with it. The gears stayed on the seabed until a sponge diver from Symi, hiding from a storm, went down to look for something else entirely.
