New Antikythera mechanism analysis challenges century-old assumption

The inspiration for the titular device in last year's blockbuster, Indiana Jones and the Dial of Destiny, was an actual archaeological artifact: the Antikythera mechanism, a 2,200-year-old bronze mechanical computer. It doesn't have any mystical time-traveling powers, but the device has been the subject of fierce scientific scrutiny for many decades and is believed to have been used to predict eclipses and calculate the positions of the planets.

A new paper published in The Horological Journal found evidence, based on statistical techniques drawn from physics, particularly the study of gravitational waves, that the mechanism's calendar ring was designed to track the lunar calendar. This contradicts a century-long assumption among scholars of the mechanism that the calendar ring had 365 holes, thus tracking with a solar calendar, but is in keeping with the conclusions of a 2020 analysis.

“It’s a neat symmetry that we’ve adapted techniques we use to study the universe today to understand more about a mechanism that helped people keep track of the heavens nearly two millennia ago," said co-author Graham Woan, an astrophysicist at the University of Glasgow. “We hope that our findings about the Antikythera mechanism, although less supernaturally spectacular than those made by Indiana Jones, will help deepen our understanding of how this remarkable device was made and used by the Greeks.”

As previously reported, a Greek sponge diver named Elias Stadiatis discovered the wreck of an ancient cargo ship in 1900 off the coast of Antikythera Island in Greece. He and other divers recovered all kinds of artifacts from the ship. A year later, an archaeologist named Valerios Stais was studying what he thought was a piece of rock recovered from the shipwreck when he noticed that there was a gear wheel embedded in it. It turned out to be an ancient mechanical device. The Antikythera mechanism is now housed in the National Archaeological Museum of Athens.

In 1951, a British science historian named Derek J. de Solla Price began investigating the theoretical workings of the device. Based on X-ray and gamma-ray photographs of the fragments, Price and physicist Charalambos Karakalos published a 70-page paper in 1959 in the Transactions of the American Philosophical Society. Based on those images, they hypothesized that the mechanism had been used to calculate the motions of stars and planets—making it the first known analog computer.

Michael Wright, then curator of mechanical engineering at the Science Museum in London, made headlines back in 2002 with new, more detailed X-ray images of the device taken via linear tomography. Wright's closer analysis revealed a fixed central gear in the mechanism's main wheel, around which other moving gears could rotate. He concluded that the device was specifically designed to model "epicyclic" motion, keeping with the ancient Greek notion that celestial bodies moved in circular patterns called epicycles. (This was pre-Copernicus, so the fixed point around which they moved was believed to be the Earth.)

In 2021, an interdisciplinary team at University College London (UCL) led by mechanical engineer Tony Freeth, an honorary professor at University College London, introduced a new computational model, revealing a dazzling display of the ancient Greek cosmos. The team's efforts built on Wright's work as part of the ongoing Antikythera Mechanism Research Project, which undertook more advanced 3D X-ray imaging with the help of X-Tek Systems in the UK and Hewlett-Packard, among others. Those images revealed much more of the original Greek transcription, confirming it was an astronomical computer used to predict the positions of heavenly bodies in the sky. It's likely that the Antikythera mechanism once had 37 gears, of which 30 survive, and its front face had graduations showing the solar cycle and the zodiac, along with pointers to indicate the positions of the Sun and Moon.

Ring cycle

Freeth's team is currently building a replica mechanism, moving gears and all, using modern machinery. This latest paper relates to the work of a machinist named Chris Budiselic, who has been cataloging his yearslong efforts to re-create the device using tools from that age on his YouTube channel. In 2020, he and several collaborators published a paper on their analysis of X-ray imaging of the device—especially of the regularly spaced holes just beneath the calendar ring. It has long been unclear how many holes the mechanism might have had originally.

Budiselic et al. thought it was probably between 347 and 367 holes and favored the lower count, which would be in keeping with a lunar calendar as opposed to the 365-day Egyptian civil calendar. But the authors noted that their conclusion was still conjecture and that despite their access to high-resolution projections, the image quality was not ideal. For instance, some of the holes were filled with debris, requiring multiple measurements of some holes to reduce noise in the data.

Woan discovered Budiselic's data and attempted to create a replica of the calendar ring late last year after a colleague mentioned it to him. Intrigued, he spent a good part of the Christmas holiday thinking about what statistical techniques might be able to answer the question of how many holes the ring had likely contained. He decided to apply Bayesian analysis to calculate the number based on the positions of the surviving holes and the placement of the surviving fragments of the ring. That method suggested the ring contained either 354 or 355 holes.

Meanwhile, Woan's colleague, co-author Joseph Bayley, decided to apply a different statistical approach to the problem. A member of the university's Institute for Gravitational Research, he adapted the group's techniques for analyzing LIGO signals picked up by gravitational wave detectors, specifically Markov chain Monte Carlo and nested sampling methods. Bayley arrived at the same conclusion as Woan: either 354 or 355 holes.

The skeptics weigh in

“Previous studies had suggested that the calendar ring was likely to have tracked the lunar calendar, but the dual techniques we’ve applied in this piece of work greatly increase the likelihood that this was the case," said Bayley. “It’s given me a new appreciation for the Antikythera mechanism and the work and care that Greek craftspeople put into making it—the precision of the holes’ positioning would have required highly accurate measurement techniques and an incredibly steady hand to punch them."

But some Antikythera experts aren't buying it, including Freeth. “It’s just wrong,” Freeth told The New York Times, noting that the machinery already has an embedded lunar calendar based on the 19-year Metonic cycle that is much more precise. “Why put a second lunar calendar on the mechanism when you’ve already taken a lot of trouble to construct a lunar calendar of great accuracy and sophistication?”

Retired astrophysicist Mike Edwards, chair of the Antikythera Mechanism Research Project, was also unconvinced but thought there was no particular reason to doubt Woan et al.'s statistical findings. “The suggestion that 354 represents a lunar calendar does not seem to have any other support from within the mechanism—it is not at all clear how it would work and how it would relate to the markings on the front of the calendar ring,” he told The New York Times. “But the establishment of the count may perhaps tell us something about the level of precision in construction judged necessary and used by the mechanism’s builders.”