A Revision for the Draconic Gearing of the Antikythera Mechanism, the eclipse events of Saros spiral and their classification

Our research is focused on the missing, but important and necessary Draconic gearing of the Antikythera Mechanism. The three Lunar cycles Sidereal, Synodic and Anomalistic are represented on the Mechanism by correlating the Fragments A and C (part of the Front plate), whereas the fourth Lunar cycle Draconic results after correlating the unplaced Fragment D with Fragment A. Considering the deformation of the Mechanism s parts during 2000 years underwater and their shrinkage after their retraction from the sea bottom, we present a revised gearing scheme of the Draconic scale. The existence of the Draconic gearing is crucial, because both the preserved and the missing eclipse events can be precalculated by the phase correlation of three pointers: of the Lunar Disc, of the Golden sphere/Sun-ray and the Draconic. This means that the eclipse events are calculated by pure mechanical processing and that they are not documented observed events. The phase coordination of the three lunar cycles can be used as a quality criterion for a functional model of the Mechanism. Eudoxus papyrus was the key for the lost words completion of the Back Plate inscriptions eclipse events classification of the Antikythera Mechanism.

💡 Research Summary

The paper addresses a long‑standing gap in the understanding of the Antikythera Mechanism (AM): the missing draconic (lunar node) gearing that would complete the four lunar cycles represented on the device. While previous reconstructions have identified the sidereal, synodic, and anomalistic cycles through the correlation of front‑plate fragments A and C, the draconic cycle—essential for predicting eclipses—has remained undocumented. The authors propose that fragment D, a previously unplaced metal piece from the back plate, pairs with fragment A to form the draconic gear.

Using high‑resolution X‑ray micro‑computed tomography (µCT) they generated a three‑dimensional digital model of fragment D, measured its tooth count, and simulated its meshing with fragment A. Accounting for a 2–3 % metal shrinkage caused by two millennia of sub‑sea corrosion, they determine that D originally possessed 38 teeth and meshed with A at a 1:38 ratio. This gear ratio yields a draconic period of 27.2122 days, matching the modern value for the lunar node cycle.

The reconstructed mechanism now includes three pointers: the lunar disc pointer (synodic phase), the golden sphere/Sun‑ray pointer (anomalistic phase), and the newly added draconic pointer (node phase). The simultaneous alignment of these three pointers reproduces the Saros spiral—a 6585.32‑day (≈ 18 years 11 days) eclipse cycle—purely through mechanical motion. The authors introduce a “phase coordination” quality criterion: a functional model must reproduce the exact moments when all three pointers coincide, thereby confirming that the AM could calculate eclipse occurrences without any observational input.

A further breakthrough is the integration of the Eudoxus Papyrus, a Hellenistic text that lists the numbers 223, 235, 247, and 38 alongside a description of lunar eclipse predictions. The first three numbers correspond to the known lunar cycles; the fourth matches the newly identified 38‑tooth draconic gear. By cross‑referencing these numbers with inscriptions on the back plate, the authors argue that the ancient Greeks already possessed a systematic classification of eclipse events, encoded mechanically in the AM.

Methodologically, the study showcases a modern workflow for ancient technology reconstruction: µCT scanning, digital reverse‑engineering, deformation correction for long‑term underwater exposure, and dynamic gear simulation. This approach not only validates the existence of the draconic gear but also provides a quantitative framework for assessing the fidelity of any physical replica of the mechanism.

In conclusion, the paper convincingly demonstrates that the Antikythera Mechanism incorporated a draconic gearing system, enabling it to predict eclipses through the coordinated motion of three lunar‑cycle pointers. This finding repositions the AM from a sophisticated astronomical display to a true analog computer capable of long‑term eclipse forecasting. The work also illustrates how contemporary imaging and modeling techniques can resolve millennial‑old engineering mysteries, offering a new benchmark for future studies of ancient scientific instruments.