In spring 1900, a party of sponge divers took shelter from a violent Mediterranean storm. When the storm subsided, they dived for sponges in the local waters near the tiny island of Antikythera, between Crete and mainland Greece. By chance, they found a wreck full of ancient Greek treasures, triggering the first major underwater archaeology operation in history. Overseen by a gunboat from the Greek navy to deter looters, by early 1901 the divers had begun to recover a wonderful array of ancient Greek goods – beautiful bronze sculptures, superb glassware, jewellery, amphorae, furniture fittings, and tableware.
They also found an undistinguished lump, the size of a large dictionary, which was probably recovered because it looked green, suggesting bronze. It was not considered to be anything remarkable at the time. Now, though, it is recognised as by far the most important object of high technology ever recovered from the ancient world: an ancient Greek astronomical calculating machine, known as the Antikythera Mechanism.
Months after it was recovered, the object split apart, revealing tiny gearwheels inside, around the size of coins. It was an astonishing discovery: no one had even thought that such precision gearwheels could exist in ancient Greece. Today, only a third of the original Mechanism survives, split into 82 fragments – designated by letters A-G and numbers 1-75. It is a fiendish 3D jigsaw puzzle, all jumbled together, with incomplete and severely corroded components. Over the years, various scholars have sought to use these fragmentary elements to deduce the purpose of the machine. The latest to tackle this challenge are a multidisciplinary team of scientists, of which I am part: the University College London (UCL) Antikythera Research Team. The team was created when imaging specialist Lindsay MacDonald and materials scientist Adam Wojcik invited me to join UCL. We widened our expertise by teaming up with Myrto Georgakopoulou, an archaeometallurgist, plus two PhD students, horologist David Higgon and physicist Aris Dacanalis. Both of our students made essential contributions to our research. We have used new ideas and a close examination of all the data to challenge previous research and to create the first model that satisfies all the evidence.
An astronomical calculating machine
From the beginning, the Mechanism generated controversy, with fierce arguments about whether it was an astrolabe for tracking the stars or a navigation device. Both proved to be wrong, but uncovering the machine’s secrets would be a long and difficult detective story, peppered with major mistakes as well as surprising progress.
The first real enlightenment came from a German philologist, Albert Rehm, in the period from 1905. Buried in his unpublished research notes are some extraordinary ideas. Rehm read inscriptions on the Mechanism concerning the risings and settings of the stars as viewed from Earth, and he found key astronomical cycles, too – 19-year and 76-year cycles of the Moon and a 223-month eclipse cycle. Rehm also made the radical suggestion that the device was an astronomical calculating machine. He had the groundbreaking idea that it contained epicyclic gearing – that is, gears mounted on other gears – a level of sophistication seemingly incredible for ancient Greece. In addition, Rehm proposed that all five planets known in the ancient world (Mercury, Venus, Mars, Jupiter, and Saturn) were displayed in a ring system on the front of the Mechanism. He simply did not have enough evidence to make coherent sense of his intuitions, and Rehm’s understanding of the internal mechanical structure was entirely wrong. More than a century later, though, his astonishing ideas are at the core of the new model of the machine created by the UCL Antikythera Research Team.
Fifty years after Rehm and his struggle with inadequate data, a British physicist, Derek de Solla Price, started a 20-year odyssey of research that culminated in a famous paper Gears from the Greeks (1974). He appreciated that to understand the Mechanism, there was a pressing need for new data to guide him through the fragmentary and confusing evidence.
Much of Price’s progress was based on X-rays of the Mechanism fragments, gathered and analysed by Charalambos and Emily Karakalos. These enabled the identification of 30 surviving gears: 27 in Fragment A and one in each of Fragments B, C, and D. Almost none of the gears were complete, so they needed to estimate the all-important number of teeth on each one – essential for understanding the workings of a geared calculating machine. From these X-rays, Price made a crucial discovery that the 19-year cycle of the Moon, identified by Rehm in the inscriptions on the Mechanism, could be calculated using its gearing.
Though Price made great progress, he also got much wrong, and only made unresolved suggestions about the planets. When Price died in 1983, the challenge was taken up by Michael Wright, a curator of Mechanical Engineering at London’s Science Museum, who had extensive experience of studying geared devices. While Price had discovered how some of the Sun–Moon system worked, it was Wright who set about reconstructing the gearing and a display for the planets.
Here, it is helpful to pause and consider how the ancient Greeks perceived the Cosmos. Their view was (almost) entirely Earth-centred and dominated by the mistaken belief that the Sun, Moon, and planets all moved around the Earth, against a background of ‘fixed stars’. When seen from Earth, the planets appear to move against the backdrop of the stars in perplexing ways. This is even reflected in the ancient Greek origin for the modern word ‘planet’: planetai, meaning ‘wandering’. Venus, for example, is sometimes ahead of the Sun and sometimes behind when viewed from Earth. Mostly it seems to move westwards through the sky, in the same direction as the Sun, but at times Venus will stand still against the stars at a stationary point, before looping backwards towards the east and reaching another stationary point, then resuming westwards motion once more. This synodic cycle – that is, its cycle relative to the Sun – is repeated again and again. Similar motions are shared by all the planets, creating a central problem for ancient astronomers. It was the failure to appreciate that the planets move around the sun that made the planetary motions seem so inexplicable.
In the 1st millennium BC, the Babylonians discovered what are known as ‘period relations’ for the planets, which equated a whole number of synodic cycles with a whole number of years. In the case of Venus, for example, they found the period relation that the planet goes through five synodic cycles in eight years. They could then use these period relations to predict the future positions of the planets in the sky. The ancient Greeks built on this by proposing geometrical theories for explaining planetary motions. These theories were ideal for mechanising the variable motions of the planets in a geared calculating machine. It was a revolutionary idea: thanks to the machine, the outcomes of ancient Greek astronomical theories could be calculated with the simple turn of a handle.
The UCL team looked at the pioneering work by Wright. He found evidence of bearings and other structures on the Main Drive Wheel. This four-spoked gear is prominent at the front of Fragment A. It is turned by the input handle and rotates once a year, thereby setting all the other gears in motion. Wright judged that there must have been an extensive epicyclic gearing system, mounted on the Main Drive Wheel. On the basis of this evidence, he proposed that one of the main purposes of the machine was to calculate the positions of the planets, which were displayed at the front of the machine. Inspired by astronomical clocks from the Middle Ages, Wright also introduced devices known as a ‘pin-and-slotted follower’ mechanisms to his reconstruction of the Antikythera Mechanism. When used alongside the gears, these devices could be used to mimic the backward loops of the planets. With great ingenuity, he managed to construct a planetarium for the Mechanism, which tracked the date, Sun, Moon, and five planets. He thought the outputs were shown as a system of pointers on the front of the machine to indicate their positions in the Zodiac. The publication of his results in 2002 was a landmark in Antikythera research, even though multiple challenges to his model would subsequently follow.
When I first studied the Antikythera Mechanism, many questions remained unresolved. So, in 2000, I proposed new investigations of the Antikythera Mechanism to get more data to tackle the outstanding issues. After years of struggle to get all the necessary permissions, this was finally carried out in 2005 by an Anglo-Greek team of researchers in collaboration with the National Archaeological Museum in Athens.
Our work used two high-tech, non-destructive techniques: Microfocus X-ray Computed Tomography (X-ray CT) – high-resolution 3D X-rays; and Polynomial Texture Mapping (PTM) – a digital imaging technique for looking at surface features, invented by Tom Malzbender (then at Hewlett-Packard). In 2005, an eight-tonne X-ray machine from X-Tek Systems (now owned by Nikon Metrology) was coaxed into the basement of the National Archaeological Museum in Athens. Roger Hadland, then owner of X-Tek Systems, made a special prototype X-ray machine and brought a top-quality team of experts to scan all 82 fragments of the Antikythera Mechanism. The results have transformed our knowledge.
Two important discoveries resulted from the new X-ray CT. I established that a dial on the lower back predicted eclipses according to the 223-month cycle, which Rehm had recognised in the inscriptions. The gearing enabling this involved a disregarded 223-tooth gear at the back, which had found no role in any previous model. I then made a startling discovery about how the Mechanism calculated the variable motion of the Moon (caused by its elliptical orbit around the Earth). Wright had made an acute observation about how two of the epicyclic gears could interact to represent variable motion. Ultimately, he discarded this observation because it did not fit into his model, but I realised that this device could model the movement of the Moon. This idea also involved the 223-tooth gear – so this gear now had two essential functions. The system was utterly astonishing in its operation.
The X-ray CT delivered another crucial revelation: extensive new inscriptions were buried inside the fragments and could now be read for the first time in over 2,000 years. Critically for the UCL Antikythera Team, these texts referred to the planets on both the front and back covers. The back cover essentially acted as a user manual, describing the principles on which the Mechanism was based, which included the cycles that Rehm had observed. Even more important was a description of what the front display of the Mechanism looked like, deciphered by a professor of ancient astronomy, Alexander Jones, with whom I was working at the time. This inscription described a model of the Cosmos, with the Sun shown by a pointer, a ring system for the planets, and marker beads to indicate each planet. Now the UCL team knew that, to reconstruct the Mechanism, they must recreate this ring Cosmos system, where previous attempts had failed.
The X-ray CT of the front cover inscription disclosed extensive new information about the planets. There is a section for each planet, enumerating days between events in their synodic cycles as well as the planet’s synodic period in years. For example, Venus takes 584 days to return to the same position relative to the Sun – a number that can be read in the Front Cover Inscription. In 2016, Jones discovered the period 462 years in the Venus section of the inscription and 442 years in the Saturn section. These were astounding numbers, unknown from previous studies of ancient astronomy. Clearly, the UCL team needed to incorporate these periods into the gearing for Venus and Saturn.
All told, the UCL team faced multiple problems: we had two new periods for Venus and Saturn, but the periods for Mercury, Mars, and Jupiter were missing; gearing at the front of the mechanism was mostly lost; the space available for restoring the gearing system was very constricted; there was an unexplained gear in Fragment D with 63 teeth and an attached disc; an enigmatic block on one of the spokes of the Main Drive Wheel appeared to make no sense at all.
Identifying the missing planetary period relations was very difficult, but essential for reconstructing the lost gearing. As we have seen, ancient Babylonian astronomy in the 1st millennium BC had discovered extensive planetary cycles. Once again, Venus is a good example. The Babylonians knew of the simple period for Venus – five synodic cycles in eight years, which can be represented as (5, 8) – but they also appreciated that it was very inaccurate. From this start point, they discovered a far more accurate period relation (720, 1151). So why did the designer of the Antikythera Mechanism not use this period? The answer is simple: 1,151 is a prime number, so any gear train calculating this cycle would need a gear with 1,151 teeth, making it far too large to fit within the Antikythera Mechanism. With great ingenuity, the designer had instead found a period relation for Venus (289, 462), which can be reduced into small prime factors, allowing the cycle to be calculated using gears small enough to fit into the Mechanism. The period of 462 years for Venus came as a huge surprise, since it was previously unknown in either Babylonian or Greek astronomy.
The key question was, how had the designer discovered this period? The UCL team found an elegant process, based on ancient Greek mathematics, which combined known Babylonian periods to produce better periods. Inspired by the presence of shared gears in the surviving gear trains, the team added the key idea that this might also have been a feature of the planetary mechanisms. Using shared prime factors in the planetary periods would allow gears to be shared between mechanisms, thereby minimising the number of gears needed. Together, these ideas not only explained the newly discovered periods for Venus and Saturn, but also allowed the missing periods for the other three planets to be generated. Now these cycles could be incorporated into gear trains for all the planets.
The UCL team created compact five-gear mechanisms for Mercury and Venus. For Venus, they discovered a way of designing the gearing that exactly matched a bearing on one of the spokes of the Main Drive Wheel and also included the mysterious 63-tooth gear in Fragment D. A comparable mechanism for Mercury also fitted all the surviving evidence. For Mars, Jupiter, and Saturn, the UCL team found ingenious seven-gear mechanisms, based on the concept behind the beautiful lunar mechanism established from previous research. Using shared gears, ensured by the choices of planetary periods, these could be shoe-horned into the tight spaces available.
Earlier models had wrongly used pointers to indicate the planets – an idea that contradicts the Back Cover description of a ring system. The ancient Greek text not only specified that the planets were arranged as a ring system, but also that they had to be shown in a particular cosmological order: Earth, Moon, Mercury, Venus, Sun, Mars, Jupiter, Saturn. For the planets to be displayed on concentric rings, the gearing systems for the planets needed to output in a very particular way. For example, the rotation of Mars is calculated by gearing that output on a tube, to which is attached a ring for the Mars display. Outside the Mars tube is another tube for Jupiter and outside this another tube for Saturn. In this way, the planetary positions can be shown on a system of nested tubes and displayed as concentric rings.
There was a basic unsolved problem. Previous attempts had been unable to reconcile a ring system with the Moon phase device, which shows the various phases of the Moon on a semi-silvered ball in the centre. The phase of the Moon is determined by the difference in the positions of the Moon and Sun – to calculate this, though, the Moon phase device needed input rotations from both Moon and Sun. This was apparently impossible, because of the nested tube system. In order to appreciate this problem fully, it is necessary to be aware that just as the Mechanism calculates the variable motion of the Moon, it also calculates the variable motion of the Sun – the so-called true Sun (this is as opposed to the mean Sun, which is simply the Sun’s average position, based on a constant rotation of the Sun). The Moon output is on the central axle. The true Sun is on the third output tube from the centre, so the outputs tubes for Mercury and Venus get in the way of the true Sun rotation reaching the Moon phase device. The UCL team found an ingenious solution to this dilemma. On one of the spokes of the Main Drive Wheel there is a block, pierced for an attachment, which had never been understood. This block could transmit a second Sun rotation – the mean Sun – directly to the Moon phase device. (The loss of accuracy in using the mean Sun instead of the true Sun is negligible.) This key idea enabled a ring system of outputs for the Cosmos.
To complete the system, the team added a hypothetical Dragon Hand (a term from medieval astronomy), which is a long, double-ended pointer that indicates the Nodes of the Moon, showing when it is possible that eclipses may occur. Though no direct physical evidence for the Dragon Hand survives, its gearing explains a prominent bearing on one of the spokes of the Main Drive Wheel, and the idea thematically links both Front and Back Dials.
The UCL team set about assembling the whole system. All the gears for the Sun, planets, and the Nodes of the Moon were crammed into the constricted space defined by the pillars and attached plates. This culminates in a beautiful ring display for the Cosmos, with the added advantage that it considerably enhances the astronomical results calculated by the machine.
The positions of Sun and Moon are shown, as well as the phase of the Moon. The age of the Moon, in terms of the number of days from a new Moon, is read by the Moon pointer on the Sun ring. The team conjecture that the synodic events of the planets – such as conjunctions, stationary points, and maximum elongations – are marked on the planetary rings and indexed to the information in the front cover inscription. The Dragon Hand indicates eclipse possibilities when it is close enough to the Sun pointer.
In March 2021, the UCL Antikythera Research Team published a radical paper in Nature’s Scientific Reports, showing this new reconstruction. Ours is the first model that conforms to all the physical evidence and matches the descriptions given in the scientific inscriptions that were engraved on the Mechanism itself. This machine is an impressive tour de force of ancient Greek brilliance, which displays the Sun, Moon, and planets, allowing their future positions and events such as possible eclipses to be calculated.
As we wrote in our article: ‘Our work reveals the Antikythera Mechanism as a beautiful conception, translated by superb engineering into a device of genius. It challenges all our preconceptions about the technological capabilities of the ancient Greeks.’ Few other archaeological discoveries have created such a difficult trail of clues, such a rich history of discovery, and such an extraordinarily sophisticated and complex outcome. The Antikythera Mechanism rewrites the history of technology.
Tony Freeth discusses the Antikythera Mechanism and its history in more detail on the PastCast. You can listen to the episode here.
FURTHER INFORMATION T Freeth et al. (2021) ‘A model of the Cosmos in the ancient Greek Antikythera Mechanism’, Scientific Reports 11. Equipment kindly loaned by Nikon X-Tek Systems was used to collect the X-ray CT data.