Antikythera Mechanism: What the Ancient Computer Really Could Do

In 1900, Greek sponge divers came across a shipwreck off the island of Antikythera. A year later, an unremarkable lump of corroded bronze came out of the wreck that turned out to be a gear train. For more than 120 years, researchers have argued about what this device could do. The magazine Archaeology makes the story of the wreck its cover feature in the September/October 2026 issue, and new studies from 2024 to 2026 have shifted some of the answers.

In short: the Antikythera Mechanism was a hand-operated bronze calculator that displayed moon phases, a calendar, and eclipses based on Babylonian cycles. The editors of the journal Nature called it in 2006 the oldest known analog computer. What exactly it could do is less clear than headlines suggest: whether it showed the planets, how precisely it calculated, and when it was built is still disputed today.

The Antikythera wreck is regarded as the largest and richest ancient shipwreck found so far. It sank around 65 to 60 BC and had loaded bronze and marble statues, luxury goods, and ceramics traveling from the east to Greece. It was discovered around Easter 1900: Captain Dimitrios Kontos and his sponge divers from the island of Symi were on their way to fishing grounds off North Africa when a storm drove them to Antikythera. The diver Elias Stadiatis descended to a depth of 45 meters and had himself pulled up again at once. According to tradition he reported a heap of dead people and horses, they were bronze statues.

With the Greek navy and the education ministry, the divers brought up statues, amphoras, and many other pieces. In 1901 the lump with the mechanism also came to light, probably in July. At first it was taken for an ordinary lump of bronze. Only on 17 May 1902 did the archaeologist Valerios Stais notice, in the National Archaeological Museum in Athens, that a gearwheel was embedded in the encrusted mass and legible Greek inscriptions were visible. The largest gearwheel is about 13 centimeters in diameter.

The mechanism broke into 82 fragments, of which only about a third of the original survives, including 30 corroded bronze gearwheels. It was originally about the size of a thick book, with a crank on the side. For a long time nobody could say what it was. The historian of science Derek de Solla Price had been working on the device since 1951. In 1971 he examined the fragments together with the Greek nuclear physicist Charalampos Karakalos using X-rays and gamma rays, and in 1974 he published his results under the title "Gears from the Greeks," subtitled "A Calendar Computer from ca. 80 BC." According to accounts, he recognized that a gear train of about 30 gearwheels was inside and that the back displayed cycles.

The next leap came in 2005 with the Antikythera Mechanism Research Project around Tony Freeth, Mike Edmunds, and others, which used three-dimensional X-ray tomography and surface imaging. In 2006 Nature published the decoding of the gear trains, in 2008 a second Nature paper on the dials of the back corrected an assumption: the small dial at the top was not a 76-year Callippic cycle, as had been thought, but the four-year cycle of the Olympiads. In 2021 a model of the whole device followed in Scientific Reports. The interpretation has shifted again and again over 120 years.

The breakthrough came in 2006, when a team led by Tony Freeth combined X-ray tomography and surface imaging and doubled the number of deciphered inscriptions. The result, published in Nature: the mechanism predicted lunar and solar eclipses on the basis of Babylonian cycles. Its gears also modeled the uneven speed of the Moon, according to the authors an unexpected degree of technical sophistication for the period.

On another dial ran, as Archaeology reports, a four-year cycle of athletic games. According to Paul Iversen, this includes a Rhodian festival, the Halieia, which could suggest that the mechanism was built on Rhodes. The device was hand-operated: you turned a crank, and the pointers showed the date, the Moon, and the eclipses. That it was a calculator in the analog sense is well documented by the gear trains and the inscriptions.

The front bore a large dial with two rings. The outer one showed the 365 days of the Egyptian calendar, with month names, the inner one a zodiac of 360 degrees with the twelve signs, plus Babylonian constellation names. The calendar ring could be turned: because the solar year is a quarter day longer than 365 days, it was moved back one day every four years, a leap day by hand.

Pointers showed the date and the positions of the Sun and Moon, probably also the planets, whose gears are missing. In addition there was a ball, half black and half silver, that turned and showed the phase of the Moon. From the front you could thus read what day it was, in which zodiac sign the Sun and Moon stood, and whether the Moon was waxing or waning.

The back had two large spirals and three small dials. The upper spiral was a 19-year calendar based on the Metonic cycle and had five turns: 19 solar years correspond very precisely to 235 lunar months. The month names on this calendar are also known, and they are unexpectedly of Corinthian origin. The lower spiral had four turns with 223 cells, one for each lunar month of the Saros cycle, after which eclipses recur under similar conditions, about 18 years, 11 days, and 8 hours later. In the cells stood glyphs for predicted eclipses.

A small dial, the Exeligmos, showed three Saros cycles, 54 years, and helped correct the time of day of the eclipse, because the 8 extra hours shift visibility by a third of a day. The upper small dial showed the four-year cycle of the Olympiads and the Greek games: in its four sectors stand the year number and games, the Isthmian, Olympia, Nemea, and Pythia, plus the festival Naa of Dodona and the festival Halieia of Rhodes. A device that calculated eclipses thus also tracked the sports calendar.

The inscriptions on the device name cycles of the planets, among them the numbers 462 for Venus and 442 for Saturn. That makes a planetary display likely. But the gears for the Sun and the planets are not preserved. The team around Freeth had to design them theoretically in 2021, in a publication in Scientific Reports in which it itself concedes that the evidence for the Sun and planets is missing.

The planetary display thus remains a reasoned assumption, not proof. The reconstructions fit the inscriptions, but they are proposals, and the parts that would settle the question have never been found. Since 2025 and 2026, further designs have appeared, for example by Jian-Liang Lin on the date display and by Aristeidis Voulgaris and colleagues in the journal Heritage on a gear on the back. They show how open the field is.

Whether the device calculated precisely in practice is disputed. The researcher Mike Edmunds judged in 2011 that the errors of the gears made it more of a teaching model than a tool for precise predictions. Esteban Szigety and Gustavo Arenas went a step further in a preprint of April 2025, that is, a paper not yet checked by peers: in a simulation, the stated errors would have caused the mechanism to jam after a third of a turn of the solar pointer (120 degrees).

The authors draw no conclusion against the construction: either the device never functioned, or the actual errors were smaller than reported, and they lean toward the second reading, because 2,000 years under water likely deformed the teeth through corrosion. That is a hint at how hard it is to infer the original from a destroyed device.

The newest hard result concerns the calendar ring. Chris Budiselic and colleagues published X-ray measurements in 2020 of the holes beneath the broken ring and estimated their number at 347 to 367. Graham Woan and Joseph Bayley of the University of Glasgow reanalyzed the same positions in July 2024 in the Horological Journal with statistical methods that come from gravitational-wave research. Their result: a ring with 360 holes is very unlikely, 365 holes, which would have matched the Egyptian calendar, also, most likely are 354 or 355, that is, a lunar year.

The ring bears Egyptian month names in Greek script, but the device calculated, according to the authors, by a lunar year. That could point to a blend of traditions. The study does not fix a single whole number. Other researchers hold to a 365-day ring, so the discussion is not over.

Who built the device we do not know, but the clues lie on the device itself. The month names belong to a Corinthian family of calendars, which alludes to Corinth and its colonies. The festival Naa belongs to Dodona in Epirus, the Halieia to Rhodes, and Paul Iversen concludes from this that the mechanism may have been built on Rhodes. The picture is mixed, and it fits the world of trade in which calendars and festivals traveled across the sea.

There are no names of inventors either, but Cicero names two: he mentions similar devices that Archimedes and Posidonius are said to have built. Whether the device of Antikythera was one of them or came from their circle is open. That it was not the only one of its kind is conceivable through Cicero but not proven, because so far only this one has been found.

The dating is open too. Nature spoke in 2006 of a construction "around the end of the second century BC." Christian Carman and James Evans proposed in 2014 on the basis of the eclipse dial 12 May 205 BC as a start date, and Paul Iversen reached the same year in 2017 from the calendar. Aristeidis Voulgaris and two colleagues argued in the journal Almagest for December 178 BC.

Both dates are the starting point of the dials and do not say when the device was finished. The wreck dates to the decades around 60 BC, there is no agreement on whether the mechanism was built shortly before the sinking or much earlier. That people knew devices of this kind is shown by Cicero: in his work on the state he describes a bronze sphere of Archimedes on which the Moon trailed the Sun by as many revolutions as it did in the sky. A much later device of this kind is a Byzantine sundial with a geared calendar from between AD 400 and 600 in London's Science Museum, which calls it the second oldest known of its kind.

The wreck is still being studied. In 2016, an international team, "Return to Antikythera," found a human skeleton on 31 August, the first remains recovered there in almost 40 years, the first from an ancient wreck in the age of DNA analysis. The hope was to determine the origin of an ancient sailor. Whether usable DNA survived is, according to published reports, still open. From 2021 to 2025 a second five-year program ran under the University of Geneva and the Greek antiquities service, which ended 125 years after the discovery.

It brought parts of the ship's hull to light, built in the "shell first" method, in which the outer skin was built first and then the inner frame, with timber of elm and oak. Added to that were the marble foot of a statue, amphoras from the island of Chios, and in 2023 a beard fragment of a statue of Herakles. In June 2024, another ancient wreck was found about 200 meters from the known wreck. The data flow into a three-dimensional model of the site.

To wrap up, the short version, so fact and story don't blur.

The Antikythera Mechanism shows how much can be contained in a single object. That's why every season in HistoryBytes is split into five categories, and this story touches all five: Archimedes as a person, the island of Antikythera as a place, the find by the sponge divers in 1900 as an event, the idea of a calculator made of gears as an idea, the mechanism itself as an object. That's how the Echoes of Olympus season builds the whole picture of a world in which science, trade, and art were closely linked.

The Antikythera Mechanism is a device that leaves more questions than it answers. We know it had gears, pointers, and inscriptions, that it showed the Moon and eclipses, and that it was loaded onto a ship that went down around 60 BC. We don't know whether it ran, who built it, or for whom. The claim that the device was too advanced for its time doesn't hold: the inscriptions are Greek, and Cicero describes a comparable device. It was the work of people who could do more than they were long given credit for.