An ancient Greek astronomical calculator, about 100 BCE. This 3D model shows a simplified gear train drawn from the surviving fragments. Move through time; the gears respond according to the known period relations.
Loading model…
Drag to orbit · Pinch to zoom · Slide the days to see the gears turn
The Antikythera Mechanism was a hand-cranked astronomical calculator. It displayed the positions of the Sun and Moon, the lunar phase, and the timing of certain astronomical cycles. It could also indicate possible eclipse months and the schedule of major Greek athletic games.
A user turned a side crank to move a date pointer. Linked gears then drove pointers on the front and back dials. The device reconciled the solar year with the lunar month and predicted repeating patterns that ancient astronomers had already recorded in arithmetic cycles.
The mechanism did not merely display mean motions. The lunar train included a pin-and-slot arrangement that reproduced the Moon’s varying speed, matching the theory associated with Hipparchus. Planetary positions were almost certainly intended as well, though those gears do not survive.
The mechanism was recovered in 1901 from a Roman-era shipwreck near the Greek island of Antikythera. Sponge divers found bronze fragments mixed with statues, pottery, glass, and other cargo from a ship that sank around 70–60 BCE. At first the pieces looked like a lump of corroded metal with gears inside.
Early study was limited by the state of the fragments. In the 1950s and 1960s Derek de Solla Price examined the largest fragment with X-rays and published a major reconstruction in 1974. His work established that the device was an astronomical calculator, though some of his gear arrangements later proved incorrect.
From the 1990s onward Michael Wright produced more accurate physical models and argued that the back dials were spirals. The decisive advance came from the Antikythera Mechanism Research Project (2005–2006). High-resolution X-ray computed tomography and surface imaging revealed hidden gears, tooth counts, and more than 3,000 characters of previously unread inscriptions. Subsequent papers by Freeth, Bitsakis, Moussas, Seiradakis, and others completed the modern consensus on the calendar, lunar, and eclipse functions.
The device is dated to roughly 100 BCE, possibly as early as the second century BCE. It was already an old and sophisticated instrument when it was lost. Nothing of comparable mechanical complexity is known from the ancient world until the medieval astronomical clocks of the fourteenth century.
Greek astronomers combined two traditions. From Babylon they took long arithmetic cycles: the synodic period (time between successive identical configurations of a planet) and the sidereal period (time to return to the same star). These “goal-year” numbers let them predict a planet’s next appearance without a physical model of the sky.
For a geometric picture they used circles. An eccentric circle places the Earth off-center. An epicycle is a small circle whose center rides on a larger deferent circle around the Earth; together they produce the observed retrograde loops. Hipparchus (2nd century BCE) refined the lunar and solar models with eccentricity and epicycles. Ptolemy (2nd century CE) later added the equant, a point offset from both the Earth and the deferent’s center, so the planet’s angular speed better matched observations while still using circular motion.
In practice an astronomer started from a known epoch, added the planet’s mean motion (degrees per day from the period relation) for the elapsed time, then applied the epicycle or eccentric correction (the “equation”) to obtain the apparent longitude. The Antikythera mechanism is a mechanical version of the same period-based arithmetic. Its surviving gears implement the calendar and lunar cycles; planetary gearing has been reconstructed from the inscriptions and the known Greek models, but those gears do not survive.
The original was made of bronze gears and plates mounted on shafts inside a wooden case. Teeth were cut by hand with a file or similar tool; pitch and tooth form vary slightly from gear to gear. The surviving fragments are heavily corroded, calcified, and incomplete. Many gears are broken, and several shafts are missing.
Modern reconstructions rest on three kinds of evidence: the physical fragments (especially Fragment A), high-resolution X-ray computed tomography that reveals internal gearing and tooth counts, and the inscriptions on the dials and door plates. Tooth-count estimates are made by fitting regular models to the surviving arcs and accounting for corrosion and missing sections.
Planetary displays for Mercury, Venus, and possibly the outer planets have been proposed on the basis of the inscriptions and the empty space inside the case. No planetary gears survive, so those trains remain hypothetical. The calendar, lunar, and eclipse functions are securely reconstructed.
The ratios below come from the Freeth et al. reconstructions (2006 and later) based on X-ray tomography of the surviving fragments. They implement the Babylonian period relations used by Greek astronomers. Many tooth counts have small uncertainties from corrosion; the functional ratios for the known trains are secure. Planetary trains are hypothesized and do not survive.
The key ratio that produces approximately 254 sidereal months in 19 years:
64/38 × 48/24 × 127/32 = 254/19 ≈ 13.36842
Modern value is ≈ 13.368267. Path: b2 (64) → c1 (38) → c2 (48) → d1 (24) → d2 (127) and further gears to the Moon pointer. A pin-and-slot device on e3/e4 adds the lunar anomaly.
The 3D model above uses simplified visual gears whose relative speeds follow these period relations. It does not yet implement the full pin-and-slot anomaly mechanism or the complete Metonic and Saros spiral gearing.
The 3D view above shows a refined gear train based on the surviving bronze fragments. Teeth are modeled with a more triangular profile closer to hand-cut ancient gears. Materials use aged bronze and brass tones. The layout includes the main drive, intermediate gears, the lunar train segment, and an additional gear for visual density. Ratios follow the known period relations. Drag to orbit; scroll or pinch to zoom. The slider is centered on today’s date and spans roughly one Metonic cycle in each direction.
The model uses a fixed J2000 epoch plus the day offset so that gear angles remain continuous and consistent as you move through past and future dates. The tropical year constant is 365.24219 days. Lunar ratios are taken from the known period relations used in the Freeth and Wright reconstructions.
Full planetary gearing and the precise pin-and-slot lunar anomaly are not modeled in this view. The original device also carried the zodiac and calendar dials on the front and the Metonic and Saros spiral dials on the back, with subsidiary dials for the Games cycle, Callippic cycle, and Exeligmos. Those dials are described in the sections above but are not rendered here.