↖ Mechanikon

Methodology

How we handle reconstruction, authority, translation, and uncertainty.

Every mechanism in Mechanikon is a working reconstruction. Some of what you see is directly observed — gear teeth on a fragment, inscriptions on a stone, a day marked on an authoritative civil calendar. Some is inferred from published scholarship. Some is our own extrapolation filling a topological or arithmetic gap. We try to make these distinctions visible in the simulation itself, not only in the prose underneath.

Reconstruction vs observation

On the Antikythera mechanism, every gear in the registry carries a source tag — observed, reconstructed, or inferred — displayed in the inscription panel when you click the gear. The Calendars overlay does an equivalent thing with its historical-mode toggle: dates outside each calendar's documented historical validity (Islamic before 622, French Republican outside 1793–1805, Byzantine outside 988–1700) are still computed but marked as anachronistic.

If a reconstruction rests on a contested correlation — the GMT correlation constant 584283 for the Maya Long Count is the canonical example — we note both the consensus and the leading alternatives, and let the prose explain why we picked what we picked. We would rather be visibly uncertain than quietly wrong.

Governing authorities and living traditions

Mechanikon is a teaching tool, not a religious authority. For every living religious or civil calendar we defer to its governing body — the Chief Rabbinate of Israel for Hebrew liturgical dates, the Umm al-Qura committee for Saudi-observed Islamic dates, the Hong Kong Observatory for Chinese astronomical reckoning, the Office of the Iranian Presidency for Solar Hijri, the Imperial Household Agency for Japanese nengō era transitions, the Ethiopian Orthodox Tewahedo Church for Ethiopian observances, the Coptic Orthodox Church for Coptic ones. Each calendar's page links to its authority. Where our arithmetic produces a date and the authority publishes a different one, the authority is right. We want to hear about the disagreement so we can improve.

The same applies to dead calendars whose reckoning is still referenced by living communities — the Maya Tzolk'in is actively used by the Oxlajuj Ajpop council in the Guatemalan highlands; the Byzantine Anno Mundi still appears in some Eastern Orthodox ecclesiastical contexts. We cite the contemporary reference and avoid declaring the calendar "historical" in ways that erase the living continuity.

Translation and transliteration

Where a calendar or mechanism carries text that isn't Latin script, we transliterate conservatively and show the original script when stable Unicode rendering is possible — Arabic, Hebrew, Devanagari, Amharic, Greek, Chinese, Japanese kanji, Maya glyphs. Where scholars disagree on a reading — the inscriptions on the Antikythera front and back plates being a running example — we show the alternative readings inline rather than collapsing them into a single narrative. The Cambridge editions of Freeth et al. (2006, 2021), Anastasiou et al. (2016), and Bitsakis & Jones (2016) are the sources we return to most often.

When we disagree with ourselves

Occasionally a published algorithm and a governing authority's table diverge by a day or a month at an edge case (the classic Chinese 2022-23 leap-sui example; various Solar Hijri equinoxes). In those cases Mechanikon follows the authority's rule, not the published algorithm, and notes the divergence in the essay.

Corrections

If you spot an error — a wrong month length, a missing intercalation, a stale link to a governing body, a contested reading we should flag — please open an issue at the GitHub repository. Corrections land quickly and with thanks.