The Antikythera Mechanism
How an Ancient Greek Geared Device Predicted Eclipses
- 12 chapters
- 33m
- Ancient Civilizations
- Free · no sign-up
The book explains how the device worked through carefully designed gear trains. Chapters cover its origin in the Hellenistic world, the engineering behind its operation, and various proposed designs. It examines the accuracy of its predictions and compares it to similar devices mentioned in ancient texts.
Archaeologists and engineers will find this detailed look at ancient technology both fascinating and instructive.
Listen
-
Read this chapter
In the early 1900s, Captain Dimitrios Kontos and a team of sponge divers from Symi island found the Antikythera wreck, a Roman cargo ship lying 45 metres down off the Greek island of Antikythera. During the first expedition with the Hellenic Royal Navy, which took place between 1900 and 1901, they retrieved many artefacts including statues, pottery, glassware, jewellery, coins, and a bronze mechanism. The mechanism was pulled from the wreck in 1901, probably in July. It's unknown how it ended up aboard the ship.
All the artifacts from the wreck were moved to the National Museum of Archaeology in Athens for study. At first, the mechanism was mistaken for a simple lump of bronze and wood. The bronze had deteriorated into atacamite, a greenish compound that cracked and shrank once exposed to air after being lifted from the sea, altering the shape of its parts. For two years, it went ignored while staff concentrated on more visually striking finds like statues. No treatment was applied to the object after its removal from seawater, leading to additional structural damage.
On 17 May 1902, Valerios Stais, the archaeologist, and his cousin Spyridon Stais, a Greek politician, made a discovery that would puzzle scholars for years to come. Among the debris, they found a gear wheel embedded in stone. Stais himself believed it might be an astronomical clock, but others were skeptical. Most thought the device was too advanced for its time, calling it prochronistic. They considered it far too complex to have been built during the same period as the other objects unearthed.
Albert Rehm, a German philologist, turned his attention to the device and put forward the idea that it functioned as an astronomical calculator.
In 1951, interest in the object was rekindled when Derek J. de Solla Price, a British science historian and professor at Yale, took notice. Twenty years later, in 1971, Price worked with Greek nuclear physicist Charalampos Karakalos to take X-ray and gamma-ray images of the 82 fragments. Their findings were published in a paper by Price in 1974.
In 2012 and again in 2015, searches at the Antikythera wreck site turned up art objects and what may be a second ship linked to the one that carried the mechanism. Among the finds was a bronze disc showing the image of a bull, with four "ears" that have holes—initially thought to possibly be part of the Antikythera mechanism as a cog wheel. However, there is little evidence it was ever connected to the device. It’s more likely the disc was a decorative element on some piece of furniture.
-
Read this chapter
The Antikythera mechanism is often called the first known analogue computer. Its intricate design and craftsmanship point to it having predecessors from the Hellenistic period that have not been discovered yet. It was built using astronomical and mathematical theories developed by Greek astronomers in the second century BC, and it likely dates to either the late second century BC or the early first century BC.
In 2008, the Antikythera Mechanism Research Project suggested the device may have originated in Corinth or one of its colonies in northwest Greece or Sicily, pointing to the calendar on the Metonic Spiral as a clue. Syracuse, a Corinthian colony and birthplace of Archimedes, was considered a possible connection, though by 2017 it was determined that the calendar was of the Corinthian type but not from Syracuse. Another idea emerged from findings at the wreck site, proposing Pergamon as the origin. The city was home to the Library of Pergamum, which during the Hellenistic period ranked second only to the Library of Alexandria in importance.
The ship that carried the device was filled with vases styled in the Rhodian manner, suggesting it may have originated from Rhodes, a place known for trade and learning. There, an academy founded by the Stoic philosopher Posidonius stood as a center of astronomy and mechanical know-how. The island was home to astronomer Hipparchus, who lived roughly between 140 and 120 BC. The mechanism uses his theory about how the Moon moves, which raises the possibility he either designed it or worked on it. Some say the events marked on the Parapegma align best with latitudes between 33.3 and 37.0 degrees north, and Rhodes lies right within that range, between 35.85 and 36.50 degrees north.
In 2014, a study proposed that the device was made around 200 BC, based on reading the Saros Dial and linking it to a lunar month after the new moon of April 28, 205 BC. That theory suggested the Babylonian method of prediction matched the device’s style better than the usual Greek trigonometric approach. Then in 2017, Iversen said the original design came from Rhodes, but this specific version was adjusted for someone in Epirus, a region in northwestern Greece. Iversen believed it was built no earlier than a generation before the shipwreck, a timeline later supported by Jones.
Further exploration took place in 2014 and 2015, driven by the hope of uncovering more pieces of the mechanism. A comprehensive five-year investigation began that same year, stretching until October 2019, after which a new five-year effort commenced in May 2020.
In 2022, researchers proposed that the Antikythera Mechanism may have been initially calibrated on 23 December 178 BC, though some experts believe 204 BC is more likely. This device was so intricate that machines of similar complexity did not reappear for approximately 1,500 years. Among the earliest examples were astronomical clocks created in the fourteenth century by Richard of Wallingford and Giovanni de' Dondi.
-
Read this chapter
The Antikythera mechanism arrived at the surface as one solid, encrusted unit from the Mediterranean Sea. It soon split into three large pieces, with smaller fragments breaking off during cleaning and handling over time. Some pieces were later recovered from the ocean floor by the Cousteau expedition. Additional fragments may still be undiscovered in storage, as was the case with Fragment F, found in 2005. Of the 82 known pieces, seven are mechanically important and hold most of the device’s structure and inscriptions. Another 16 smaller parts contain partial or incomplete text.
Many of the smaller pieces recovered from the wreck yield nothing of apparent value, but a few carry inscriptions that offer key clues. Fragment 19, for instance, holds important back door markings, including one reading "... 76 years ..." which connects to the Callippic cycle. Other texts on the same fragment appear to describe what the back dials were meant to do. Beyond this notable piece, fifteen additional smaller fragments also bear traces of writing.
-
Read this chapter
On the front face of the mechanism, a fixed ring shows the ecliptic divided into twelve equal zodiacal signs, each 30 degrees across, following the Babylonian system even though the actual constellation boundaries shifted over time. Surrounding that dial is a second, rotatable ring marked with the Egyptian Sothic calendar—twelve months of thirty days each, plus five extra days. The months are labeled using Greek letters transcribing the Egyptian names. To begin operating the device, you align this Egyptian calendar ring so its months correspond to the current zodiac positions. Because the Egyptian calendar doesn’t include leap years, it slowly drifts through the zodiac, completing a full cycle roughly every 120 years.
The mechanism operated through a small hand crank, now lost, that turned a crown gear connected to the largest gear—visible on the front of fragment A as gear b1. This motion moved the date pointer on the front dial, which had to be set to the correct Egyptian calendar day. Since the year wasn't selectable, users needed either to know the current year or consult the Babylonian ephemeris tables on the back to determine it. The crank rotated the date pointer about 78 days per full turn, making it easy to align with a specific day if the device was functional. Turning the crank also set all interlocked gears in motion, calculating not just the Sun and Moon’s positions and the moon phase, but also eclipses, calendar cycles, and possibly planetary locations.
The operator had to keep track of where the spiral dial pointers were on the two large dials at the back of the device. Each pointer moved along incisions in the metal as the dials made four or five full rotations. When a pointer reached the end of the spiral, at the terminal month location, the follower attached to the pointer had to be manually reset to the opposite end of the spiral before the mechanism could continue.
-
Read this chapter
The mechanism comes in a wooden casing with a front and back door, each engraved with important information. The back door acts like an instruction manual, showing how the device worked. On one piece, fragment 19, it says "76 years, 19 years," which refers to the Callippic and Metonic cycles. It also writes "223" for the Saros cycle. Another piece, E, mentions "on the spiral subdivisions 235," pointing to the Metonic dial. These details reveal how the ancient Greeks used this complex machine to track celestial events and predict eclipses.
-
Read this chapter
The Antikythera mechanism stands out for how finely crafted its parts are, matching the complexity of 14th-century astronomical clocks. It contains at least thirty gears, with expert Michael Wright suggesting the ancient Greeks might have built even more intricate systems. Scholars debate whether the device included displays for all five planets known in antiquity. No gearing for such a feature remains, and every gear has been accounted for—except for one unexplained 63-toothed wheel labeled r1, found in fragment D.
Fragment D is a small quasi-circular constriction with an inner gear numbered 45 ("ME" in Greek) moving inside a larger hollow gear, reproducing epicyclical motion that, with a pointer, indicates Jupiter's position. The front face positioned astronomical bodies relative to the celestial sphere along the ecliptic, referencing Earth's observer position. This is irrelevant to whether positions were computed using heliocentric or geocentric views; both methods should yield same results within error factors. Ptolemy's epicyclic Solar System (c. 100–170 AD)—hundreds of years after the mechanism's apparent construction—used more epicycles and was more accurate than Copernicus's (1473–1543) until Kepler (1571–1630) introduced elliptical orbits.
Evans and colleagues proposed that showing the average locations of the five known planets would need just 17 extra gears, arranged in front of the main driving wheel and read off through separate circular dials. Freeth and Jones later built a model based on gear systems similar to those used for tracking the moon’s motion, which also allowed for the positions of the planets and the Sun’s variation. They argued this version better reflects what the Greeks were capable of, unlike Wright’s design, which adds unnecessary complexity and strain to the mechanism.
The Antikythera Mechanism's gear teeth were shaped like equilateral triangles with an average circular pitch of 1.6 millimeters and wheel thickness of 1.4 millimeters, while air gaps measured 1.2 millimeters. These teeth were likely made using hand tools, as evidenced by their unevenness. Modern imaging and X-ray technology reveal exactly how many teeth each gear had and their sizes, enabling accurate replication of the device's basic function. The major mystery remains whether planet indicators were present. A table of gears, teeth, and expected rotations follows based on work by Freeth et al. (2008) and (2012). Starting with one year per revolution for the b1 gear, all other rotations were computed from gear ratios. Some gears marked with an asterisk (*) are missing or have predecessors that are missing; these were calculated using reasonable estimates of tooth counts. Lengths in days assume a year of 365.2425 days.
Several gear ratios could align with the correct synodic periods for the planets and the Sun. The selections made appear accurate, featuring sensible numbers of teeth, yet the actual gears used remain a mystery.
-
Read this chapter
It’s very likely the mechanism included dials for the planets, since the manual mentions the complex movements of all of them. The exact setup for the planet gears remains unknown, except for the Moon, which uses a coaxial system. Fragment D, an epicycloidal design, is thought to represent either Jupiter’s motion or the Sun’s, according to Moussas (2011, 2012, 2014) and the University of Thessaloniki group. The Sun gear is driven by a hand crank connected to gear a1, which turns the large four-spoked mean Sun gear b1. That gear, along with b2 which has 64 teeth, directly powers the date pointer, possibly with a second pointer for the Sun’s true position, as discussed in Freeth’s reconstruction. All calculations here assume the b1 gear rotates exactly 360 degrees per tropical year, based only on gear ratios.
The Moon pointer follows a complex path through the mechanism, beginning at b1 and moving via c1, c2, d1, d2, e2, e5, k1, k2, e6, e1, and ending at b3. Of particular note are the gears k1 and k2, which operate face-to-face rather than meshing, with a pin in k1 sliding within a slot in k2. This setup allows the radius to vary, effectively modeling the Moon's elliptical orbit in accordance with Kepler's laws. The system averages out to a rotational period of 27.321 days, closely matching the modern lunar sidereal month of 27.321661 days. Mounted on e3, the k1/k2 gears also introduce a precessional motion with a period of 8.8826 years, nearly aligning with the Moon's actual precession of 8.85 years. The phases of the Moon are displayed through a small gear r attached to the pointer shaft, which meshes with the Sun pointer at B0. This differential action turns the gear at the synodic month period—29.53 days—accurately depicting the angle between the Sun and Moon, driving a ball visible through the dial, painted half white and half black to represent lunar phases.
The Metonic cycle on the mechanism is powered by a series of gears including b1, b2, l1, l2, m1, m2, and n1, which connect to the pointer. This pointer completes one full rotation over 6,939.5 days, closely matching the actual Metonic period of 6,939.69 days. Another gear train, the Olympiad cycle, uses the same set of gears plus n2 and o1, and it drives a pointer that turns anticlockwise. That pointer makes exactly four complete rotations, corresponding to the length of an Olympiad, which is the expected result for this system.
The Antikythera Mechanism’s Callippic train operates through a series of gears including b1, b2, l1, l2, m1, m2, n1, n3, p1, p2, and q1, which holds the pointer. Its calculated rotational period is 27,758 days, remarkably close to today’s known value of 27,758.8 days. Meanwhile, the Saros train uses the same gears b1, b2, l1, l2, m1, m3, e3, e4, f1, f2, and g1 to drive its pointer. The modeled cycle of the Saros pointer lasts 1,646.3 days across four turns of the spiral track, aligning closely with the modern figure of 1,646.33 days.
The device’s Exeligmos train operates through a series of gears including b1, b2, l1, l2, m1, m3, e3, e4, f1, f2, g1, g2, h1, h2, and i1, which carries the pointer. This pointer completes one full rotation every 19,756 days, a figure that closely matches the modern value of 19,755.96 days. Some components, like m3, n1-3, p1-2, and q1, didn’t survive the centuries, but scholars have reconstructed their roles based on the dials found on the back face. By analyzing those, they determined how the pointers functioned and proposed appropriate gear arrangements to fulfill those functions. These suggestions are now widely accepted.
-
Read this chapter
Because of the space between the mean Sun gear and the front of the case, along with the size and features of that gear, it's clear the mechanism once had more internal gearing—either lost after the shipwreck or removed before loading. This lack of evidence has led scholars to propose different solutions over time. In 2002, Michael Wright built the first working model using what was known about the device and imagined how the Ancient Greeks might have designed it. He included adjustments for both the lunar anomaly and a deeper solar anomaly, known as the "first anomaly." His version featured pointers for the true sun, Mercury, Venus, Mars, Jupiter, and Saturn, in addition to the mean sun and lunar pointers.
In 2010, Evans, Carman, and Thorndike offered a new solution to how the Antikythera Mechanism worked, differing significantly from Wright’s earlier proposal. They focused on what they saw as irregular spacing of inscriptions on the front dial, suggesting an off-centre sun indicator that would simplify the device by eliminating the need to model the solar anomaly. Instead of trying to show accurate planetary positions—something made difficult by the offset inscriptions—they proposed simple dials for each planet, displaying key events like appearances and direction changes. Their design used basic meshed gear trains and explained the previously mysterious 63-toothed gear in fragment D. In a 2012 paper, Carman, Thorndike, and Evans also explored a system using epicyclic gearing with pin and slot followers.
Freeth and Jones proposed in 2012 a compact solution for indicating planetary positions. Their design shows solar anomaly—the sun's apparent position—on a separate pointer from the date pointer, which displays the mean sun's position and date on the month dial. When properly synchronized, the front panel display resembles Wright's model, though this version has only been built as a 3-D computer simulation. The method for creating the solar anomaly uses three gears: one fixed in the center of the b1 gear and attached to the sun spindle, another on one of the spokes acting as an idle gear, and a third positioned nearby with an offset pin, arm, and slot attached to the sun spindle that creates the anomaly as the mean sun wheel turns.
The inferior planet mechanism governs the Sun, Mercury, and Venus, each featuring an epicyclic gear on b1 setting Earth year frequency. Each epicyclic meshes with a grounded frame gear, while a pin on extended gear connects to slotted bar leading to front dial pointer. These bars prevent interference among three mechanisms arranged along b1's four spokes. Configuration includes one new grounded gear, one shared gear, one gear reversing Sun's anomaly, three epicyclic gears, and three bar-pointer assemblies, totaling five gears and three slotted bars. Superior planet systems—Mars, Jupiter, and Saturn—follow lunar anomaly pattern using gears pivoted on b1 extension meshing with grounded gears, including pin and slot arrangement connecting to coaxial tube and pointer. All three fit within quadrants of b1 extension, lying flat with front dial plate. Each requires a grounded gear, driving gear, driven gear, and gear-pointer assembly, adding twelve gears total.
The mechanism uses eight coaxial spindles of different sizes to transfer motion from gears to pointers tracking celestial events. These nested spindles manage multiple rotations simultaneously, with thirty original components involved in the process. The arrangement reflects complex understanding of astronomy and mechanics, allowing the device to predict eclipses with remarkable accuracy. This design demonstrates how ancient Greek engineers approached timekeeping and celestial prediction problems. The system relies on interlocking gears and spindles working in harmony to form a functional cosmic model. Each piece must align perfectly with others to function correctly, showing advanced knowledge of its makers. This level of engineering was unmatched for centuries, as the creators understood how to translate astronomical data into mechanical movement. Their work laid foundation for future innovations in timekeeping and astronomy.
-
Read this chapter
Investigations by Freeth and Jones show that their simulated version of the mechanism doesn’t match reality very well. In fact, the pointer for Mars can be off by as much as 38 degrees in certain spots—especially at the points where Mars appears to move backward in the sky. These errors don’t come from flaws in how the gears are built, but from the ancient Greek understanding of how planets actually move. The Greeks hadn’t yet figured out the right theory, and it wouldn’t be improved until around 160 AD, when Ptolemy published his Almagest. That work introduced a new idea called the equant, which helped explain planetary motion better. Even later, in 1609 and 1619, Kepler’s laws of planetary motion brought even greater accuracy to how we understand the cosmos.
The Antikythera Mechanism was a device created by ancient Greeks to forecast celestial events, based on the advanced astronomy of its time. It is the only known example of such remarkable engineering from antiquity, a true wonder of ancient genius. Yet despite its ambitious design and the brilliance it represents, it did not function with the precision one might hope for. It was a machine built to predict celestial movements using the complex theories of its era, a rare glimpse into a lost tradition of innovation, a masterpiece of ancient thought—but it didn't really work very well.
Freeth and Jones point out that the device’s mechanical accuracy was compromised by the way it was built. Each gear was handmade, with triangular teeth and friction between the moving parts, including the bearing surfaces. These factors likely caused a kind of looseness that would have overwhelmed the more delicate solar and lunar correction systems the designers had built in.
Recent research has shown that while the Antikythera Mechanism's engineering was impressive for its time, the original design aimed for a level of precision that went far beyond what was actually achieved in the final construction. The gear trains contain significant cumulative errors, which likely neutralized many of the carefully calculated adjustments built into the device’s framework.
While the device may have had some inaccuracies, caused by hand-made triangular teeth, its design was surprisingly advanced. The calculations and technology used to map the elliptical paths of planets and the Moon’s retrograde motion were so sophisticated that they predated the first known clocks in medieval Europe by more than a thousand years. A clockwork-type gear train, including a pin-and-slot epicyclic mechanism, made this possible. The mathematical knowledge behind it—like Archimedes’ work on pi and centers of gravity, and his early steps toward calculus—suggests the Greeks had tools beyond Babylonian algebra, allowing them to model planetary motion accurately.
The Moon mechanism on the Antikythera Mechanism uses a complex system of bronze gears, including two linked with a slightly offset axis, to show the moon’s position and phase. This design reflects an understanding of how the moon moves at varying speeds as it orbits Earth—a fact later described by Kepler’s laws. Though the ancient Greeks didn’t know the orbit was elliptical, the device still accurately models this motion, making it a remarkable achievement in ancient engineering and astronomy.
-
Read this chapter
The Antikythera Mechanism shows a level of complexity that suggests it wasn’t a one-time creation but part of a longer tradition, possibly built on knowledge passed down through generations. Such devices were often melted down because of the bronze they contained, which is why so few have survived to today. This means there may have been others like it, though we rarely see them now.
-
Read this chapter
In Cicero’s De re publica, written between 54 and 51 BC, a philosophical dialogue mentions two machines that some scholars believe were early planetariums or orreries, able to predict the movements of the Sun, Moon, and five known planets. These were built by Archimedes and brought to Rome by Marcus Claudius Marcellus after Archimedes died during the siege of Syracuse in 212 BC. Marcellus held Archimedes in high regard and kept one of the machines as a family heirloom, while the other was placed in the Temple of Virtue. In the dialogue, Cicero has Philus say that Gaius Sulpicius Gallus, consul in 166 BC and credited by Pliny the Elder with writing the first Roman book on eclipses, gave both a "learned explanation" and a working demonstration of the device.
I had often heard of this celestial globe mentioned because of Archimedes' fame. Its look, however, didn't seem all that remarkable to me. There was another version, more elegant and better known, created by the same Archimedes and later placed in the Temple of Virtue at Rome by Marcellus. But when Gallus began to describe the machine's inner workings with his deep knowledge, I realized that Archimedes must have had a genius beyond what we usually imagine. He said the solid globe was an old idea, first presented by Thales of Miletus. Then Eudoxus of Cnidus, a student of Plato, drew the stars on its surface. Years later, Aratus used those same designs in his poetry—not through astronomy, but through poetic flair. Gallus also noted that the original globe couldn't show the Sun, Moon, and planets moving in their orbits. That's where Archimedes' genius shone: he figured out how to make one revolution control many different motions at once.
When Gallus moved this globe, it demonstrated the Moon’s relationship to the Sun, with the bronze device rotating exactly as many times as there are days in the real celestial sphere. It accurately displayed the same solar eclipse shown in the sky globe, as well as the moment when the Moon enters Earth’s shadow while the Sun is aligned. The mechanism mirrored both solar and lunar eclipses.
Pappus of Alexandria, writing in the late third or early fourth century AD, said that Archimedes had composed a lost work titled On Sphere-Making, describing how to build such devices. Ancient texts tell us of several of his inventions, including one with simple drawings—a device known as an odometer. This machine was later adopted by the Romans, who used it to set mile markers, as described by Vitruvius and Heron of Alexandria, and during the reign of Emperor Commodus. The drawings in the texts looked functional, but early attempts to recreate them failed. That changed when researchers replaced the square-toothed gears shown with angled gears like those found in the Antikythera mechanism—they worked perfectly.
In the fourth and fifth centuries, writers in the Roman era referenced the device, including Lactantius, who authored Divinarum Institutionum Libri VII, and Claudian, the author of In sphaeram Archimedes. Proclus also wrote about it in his Commentary on the first book of Euclid's Elements of Geometry. If Cicero’s account is accurate, such technology was already present as early as the third century BC.
Cicero mentioned that another such device was constructed "recently" by his friend Posidonius, describing how each revolution of the mechanism produces the same motions in the Sun, Moon, and five wandering stars as those seen daily in the sky.
It’s unlikely that any of the machines referenced in ancient texts was the actual device recovered from the shipwreck. The mechanisms attributed to Archimedes and noted by Cicero were situated in Rome at least thirty years after the wreck’s estimated time, while a third device was certainly in the possession of Posidonius by that point. Scholars who have rebuilt the Antikythera mechanism agree it was far too complex to have been a one-of-a-kind invention.
In Roman Greece, other intricate metal devices have been found, including a bronze combination lock from the Augustan or Hadrianic period, discovered at Kerameikos. This lock worked using an early kind of mechanical logic: the central bolt wouldn't move unless two separate rotary dials were lined up just right. It also had a hidden bypass feature built in.
-
Read this chapter
The Antikythera mechanism wasn’t unique, pointing to a broader Greek tradition of complex mechanical devices that later influenced the Byzantine and Islamic worlds. In the Byzantine Empire, fragments of a geared calendar linked to a sundial from the fifth or sixth century have been discovered; it likely helped with timekeeping. The Islamic world preserved similar knowledge: Kitab al-Hiyal, or Book of Ingenious Devices, was commissioned by the Caliph of Baghdad in the early 9th century AD and described over a hundred machines, some possibly rooted in ancient Greek texts. Around 1000, the scientist al-Biruni outlined a geared calendar much like the Byzantine one, and a surviving 13th-century astrolabe includes a comparable clockwork mechanism. This medieval technology may have eventually reached Europe and shaped the development of mechanical clocks.
In the 11th century, a Chinese polymath named Su Song built a mechanical clock tower that did more than tell time. Among its functions, it showed the positions of stars and planets using a rotating armillary sphere. This device was part of a larger system that tracked celestial movements with remarkable precision for its era.
Read
Free to download, keep and share. For general information only — not professional medical, legal or financial advice. Please consult a qualified professional.