◬ MIMZY · instrument № 00 · the first computer, recovered

The Antikythera Mechanism

A bronze hand-cranked computer built in Greece around 100 BC, lost in a shipwreck for two thousand years. Turn the crank, or pick a date — and watch it compute the sky: the sun and moon on the zodiac, the moon's phase, the Metonic calendar, and — from the Saros cycle — real eclipse-season prediction.

⊙ EDUCATIONAL & SIMULATION — runs on real mean-element astronomy (J2000)

Date

Crank

Speed4 d/frame

The Sky, computed

Date shown

—

Sun — ecliptic longitude

—

Sun — in the zodiac

—

Moon — ecliptic longitude

—

Moon — in the zodiac

—

Moon phase

—

Illuminated

—

Days from a new moon

—

Saros eclipse predictor—

The Calendar Cycles

Metonic — month of 235 (19 years)—

Saros — month of 223 (18y 11d)—

Callippic — year of 76—

Olympiad — year of the 4-year Games cycle—

The Gearwork it modelled (and this sim reproduces)

235 : 19	The Metonic cycle — 235 synodic months almost exactly equal 19 solar years. The calendar dial is a 5-turn spiral of 235 cells.
254 : 19	The lunar train — 254 sidereal months per 19 years (235 + 19). Cut into bronze with a 127-tooth gear (half of 254), one of the device's most famous surviving wheels.
223	The Saros cycle — 223 synodic months ≈ 18 years 11⅓ days. After one Saros, eclipses repeat. The Saros dial is a 4-turn spiral of 223 cells, carrying glyphs for predicted solar (☉) and lunar (☾) eclipses.
3 × 223	The Exeligmos — three Saros (54 years) brings eclipses back to the same hour of day; the small dial added +0, +8, or +16 hours.
pin & slot	A pin-and-slot on two stacked gears made the moon speed up and slow down — modelling Hipparchus's lunar anomaly. The first known mechanism for a varying (non-uniform) motion.

How real is this? The cycles and gear ratios above are the genuine, attested mechanism (Freeth et al., Nature 2006 & 2008; Wright's reconstructions). The sky in the panels is computed live from standard low-precision mean elements (J2000) — the same kind of mean-motion model the bronze gears embodied, accurate to about a degree over centuries. The eclipse light turns on inside the real ecliptic limits (a solar eclipse is possible when a new moon falls within ~16°–18° of a lunar node; a lunar eclipse within ~14° of a full moon at a node) — so it flags true eclipse seasons, exactly as the Saros dial did. It catches the central eclipses and may slip a degree or two at the edges, just as a mean model must — checked against the 2017 Great American total solar (dead on the node) and the 2025 total lunar; it is a teaching instrument, not an ephemeris.