The Antikythera Mechanism: How a Greek Shipwreck Revealed an Ancient Computer

On Easter Sunday in 1901, a Greek sponge diver named Elias Stadiatis surfaced from a wreck near the island of Antikythera with a story about bodies of bronze men and women on the seabed. He had stumbled, by accident, on the largest cache of classical Greek statuary ever found, a Roman-era cargo ship that had gone down sometime around 60 BCE on a route from the eastern Mediterranean to Italy. The salvage operation, conducted over the next year by the Greek navy with the assistance of the sponge divers, recovered marble statues, bronzes, glassware, jewelry, and a corroded lump of metal that nobody could quite identify. The lump was set aside in the National Archaeological Museum in Athens. It would take a century to understand what it was.

The accidental fragment

The lump revealed itself slowly. As the corrosion crust dried and split, archaeologists noticed that the interior contained gear teeth. Initial speculation in 1902 (by the archaeologist Valerios Stais, who first published on it) was that the object was an ancient astrolabe or some kind of clock. The hypothesis was dismissed by his colleagues as anachronistic; gears of that complexity were not believed to exist before the medieval period. The lump returned to obscurity.

The first systematic study came in the 1950s, when the British physicist Derek de Solla Price obtained permission to examine the fragments and X-ray them. The X-rays showed dozens of interlocking gears, of varying tooth counts, in what was clearly a deliberate mechanical structure. Price's 1959 paper in Scientific American announced that the Antikythera mechanism was a calendar computer, capable of tracking the lunar month, the solar year, and the synodic cycles of the visible planets. The reaction in the field ranged from skepticism to active dismissal. Price's interpretation was several centuries ahead of where mechanical computing was supposed to have begun.

What computed tomography revealed

The decisive evidence came in 2005. A team led by Mike Edmunds at Cardiff and Tony Freeth at the Antikythera Mechanism Research Project, using a custom-built eight-ton computed tomography scanner shipped to Athens by X-Tek Systems, produced three-dimensional images of the interior of the fragments at sub-millimeter resolution. The CT scans revealed inscriptions on hidden surfaces, gear teeth that had been invisible on flat X-rays, and the layout of dial faces that had been broken away from their pointers.

The reconstruction that emerged from the CT data is remarkable. The mechanism contains at least 30 interlocking bronze gears (37 in the most recent reconstruction), arranged in a wooden case the size of a large book. A hand crank on the side drives a series of gear trains. The front face displays the position of the sun and moon in the zodiac, the phase of the moon (via a rotating ball half-painted black), and the position of the visible planets. The back face displays two spiral dials: one tracking the 19-year Metonic cycle (which reconciles lunar and solar calendars), and one tracking the 18.03-year Saros cycle of solar and lunar eclipses, with the date and time of each predicted eclipse engraved on the spiral.

The eclipse prediction

The eclipse prediction is the part of the mechanism that surprises modern observers most. The Saros cycle was known to Babylonian astronomers from at least the 7th century BCE; eclipses recur in patterns, and a sufficiently long observational record allows future eclipses to be predicted. What is unusual about the Antikythera mechanism is that the prediction is mechanized. Turning the hand crank advances the date display, and the eclipse pointer rotates around the spiral, indicating which dates over the next several decades will see eclipses, with engraved markings indicating whether each is solar or lunar and roughly what time of day.

The mechanism does not predict eclipses by computing them from physical principles; it predicts them by mechanically advancing through a known cyclic pattern. This is the same trick used by every mechanical orrery and astronomical clock built since: encode the periodicity in gear ratios and the position of the gears tells you the answer. The Antikythera mechanism is the earliest known device to do this, by about 1,500 years.

The lunar anomaly

The most sophisticated piece of mechanism is the lunar position dial. The moon does not move at a constant rate against the stars; its motion is faster near perigee (closest approach to Earth) and slower near apogee. The variation, called the lunar anomaly, was known to ancient astronomers from observation. The Antikythera mechanism reproduces it mechanically.

The trick is a pin-and-slot mechanism between two gears mounted on slightly offset axes. As the gears rotate, the pin slides through the slot in a way that produces a non-uniform output rotation: faster at one phase, slower at another. The output gear's angular velocity varies sinusoidally over each lunar month, exactly as the actual moon's apparent motion varies. This is, as far as anyone knows, the earliest mechanical implementation of a non-uniform motion, and it is two thousand years older than the next clear example.

Who built it and why

The provenance is uncertain in detail and clear in outline. The mechanism dates to roughly 100 to 80 BCE on the basis of the inscriptions' letterforms. The cargo ship was Roman, sailing from somewhere in the eastern Mediterranean (possibly Rhodes or Pergamon), bound for Italy, when it sank around 60 BCE. The most likely manufacturing origin is Rhodes, which had a well-established astronomical tradition associated with Hipparchus (who flourished around 150 BCE and who developed the trigonometry that would have been required to calibrate the gear ratios).

The mechanism was almost certainly not unique. Cicero, writing in the first century BCE, describes a similar device built by Archimedes (in the third century BCE) that "reproduced the motions of the sun and moon and the five planets." Cicero claims to have seen Archimedes's device displayed in Rome. There are a small number of other classical references to similar instruments. The Antikythera mechanism is the only one that survives, and it survived only because it sank to the bottom of the Mediterranean and was preserved by being buried in marine sediment.

What the mechanism tells us

The cultural implication of the Antikythera mechanism is what makes it more than a curiosity. The Hellenistic Greek world, in the second and first centuries BCE, had the mathematical sophistication to model the heavens, the metallurgical skill to cut precise bronze gears, and the engineering tradition to combine the two into a working calculating device. None of this was widely known before the CT scans, because no other mechanism survived. The history of technology has a long-running tendency to assume that the past was simpler than it was, and the Antikythera mechanism is one of the strongest counterexamples.

What was lost when this tradition lapsed is harder to say. The gear-cutting craft did not disappear (there is continuity through the medieval Islamic world to the European clock-makers of the 14th century), but the integration of astronomy with computation was not picked up again at this level of sophistication until the geared astronomical clocks of the late medieval period, more than a thousand years later. The mechanism is a reminder that technological progress is not monotonic; civilizations build remarkable things and then, for reasons that are usually mundane (the death of a teacher, the loss of a workshop, an economic downturn that stops commissions), they stop.

The mechanism is currently displayed in Athens at the National Archaeological Museum, in a glass case at room temperature. The fragments have continued to yield new information; a 2021 paper used updated X-ray imaging and statistical analysis of gear ratios to propose a revised reconstruction of the planetary display, which is still partially missing. The work of understanding what the device actually computed is, a hundred and twenty-five years after Stadiatis's accidental discovery, ongoing.