The claim to refuse is that the Antikythera fragments are an out-of-time computer, or alien technology with a Greek name. The National Archaeological Museum holds a Hellenistic instrument. Shipwreck finds are shown in Room 38, and the Mechanism, inventory X15087, is described as evidence of how far technology had developed at the time, the most significant scientific instrument of antiquity. "At the time" places the object in its century. The label does not put a later machine in a Roman cargo.
The museum's explainer treats the dials as named astronomical cycles. Visitors match functions to the front and the back. The Metonic cycle, dated there to 430 B.C., is 19 years or 235 lunar months, with a lunar-solar slippage of five days. The Callippic cycle, dated 370-300 B.C., is four Metonic cycles, 76 years or 940 lunar months, cutting the slippage to four days. The Saros, attributed to Babylonian observation, is 223 lunar months, about 18 years and 11 days: eclipses repeat, not at the same coordinates and not at the same hour, the shift given as eight hours. The Exeligmos, three Saros cycles or 669 lunar months, is the span that does contain a whole number of days. The page says some of these cycles served to calculate eclipses. That is an account of known periods. It is not a complete computer, and it is not a substitute for the fragments.
The 2006 Nature paper is imaging against that limit. Tony Freeth and colleagues call it a unique Greek geared device of the end of the second century B.C., discovered in 1901 in a Roman shipwreck. It displayed celestial cycles, in particular the Moon's phases and a luni-solar calendar. Functions had stayed controversial because gears and inscriptions are fragmentary. Surface imaging and X-ray tomography let them reconstruct gear function and double the deciphered inscriptions. On that reading the mechanism predicted lunar and solar eclipses from Babylonian arithmetic-progression cycles. Inscriptions support a lost mechanical display of planetary positions. Gearing realizes Hipparchos's second-century B.C. theory of the Moon's irregular motion, which they call an unexpected sophistication for the period. That is the finding. The journal page's summary adds that it can justifiably be claimed as the world's oldest known analogue computer. That sentence is the summary's. A reconstruction of function is not the machine from the nets.
The 2008 paper stays on the back dials and does not sign the object to one maker. The upper dial is a 19-year Metonic calendar in a five-turn spiral. The lower is a Saros eclipse dial, a four-turn spiral of 223 lunar months, with glyphs. All twelve month names, newly read, are Corinthian. Colonies in northwestern Greece, or Syracuse, are the contenders named, the latter suggesting a heritage back to Archimedes. Suggesting is not a signature. The upper subsidiary dial is not a 76-year Callippic dial, as earlier work had it, but the four-year Olympiad. Excluded-day months, once called implausible, are reported on this calendar. The paper reads dials and letters.
A 2014 PLOS ONE paper by Freeth carries the eclipse scheme and labels its own pictures as reconstructions. Glyphs give type and time, with index letters to solar-eclipse inscriptions. Two arithmetical models explain the scheme. The abstract says the Greeks built a machine that predicts, years ahead, eclipses and their character: direction of obscuration, magnitude, colour, the Moon's angular diameter, relation to the node, and time. The next sentence: it was not entirely accurate. Eighty-two fragments are in Athens. The exploded figure is a computer reconstruction. Front planetary gearing is called conjectural. Lunar-anomaly gearing and the back dials are called firmly established. A drawing of missing gears is not the bronze. Colour belongs to the reconstructed scheme, not to a log the original is shown to have got right.
The 2021 Scientific Reports paper says in the title that the result is a model. Discovery is again given as 1901. The machine is 82 fragments; about a third survives, including 30 corroded gearwheels. Tomography in 2005 decoded the back. The front stayed largely unresolved. Inscriptions describe the Sun, the Moon, and the five planets known then, displayed as a Greek cosmos. No earlier reconstruction, they write, matched the data. Theirs is a new model. A model of a machine two-thirds lost is not the instrument in Room 38. The pedigree they name is Babylonian cycles, mathematics tied to Plato's Academy, and Greek astronomical theory. That is the denial of an out-of-time arrival.
An out-of-time or alien computer is the claim these pages do not support. The museum calls X15087 an instrument of its time and sets Metonic, Callippic, Saros, and Exeligmos cycles against the dials. The 2006 and 2008 papers reconstruct eclipse prediction, a Saros spiral, Corinthian months, and a possible Syracusan heritage from broken gears and inscriptions, and one journal summary adds "analogue computer." The 2014 scheme includes colour and then says the result was not entirely accurate, with the front planets still conjectural. The 2021 paper is a model of a front cosmos built because most of the machine is gone. The model is not the original.
National Archaeological Museum, Collection of Metalwork: https://www.namuseum.gr/en/permanent_exhibition/erga-metallotechnias/.
National Archaeological Museum, mechanism explainer: https://antikythera-mechanism.namuseum.gr/en/what-is-the-mechanism-and-what-does-it-consist-of/.
Freeth and colleagues, Nature, 2006: https://www.nature.com/articles/nature05357.
Freeth and colleagues, Nature, 2008: https://www.nature.com/articles/nature07130.
Freeth, PLOS ONE, 2014: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0103275.
Freeth and colleagues, Scientific Reports, 2021: https://www.nature.com/articles/s41598-021-84310-w.
What Stays Contested
On these pages: X15087 in Athens, called a scientific instrument of antiquity; an explainer matching the dials to Metonic, Callippic, Saros, and Exeligmos cycles; a 2006 reconstruction of eclipse gears from fragmentary inscriptions, plus a journal line that says analogue computer; Corinthian months and a suggested Syracuse link; a 2014 scheme that includes colour and says the predictions were not entirely accurate; a 2021 model of the front, where about a third survives. Left contested: how much of any reconstruction is in the bronze that remains. No alien device is baptized, and no model is the original.
