The Drift
A lunar year runs about eleven days short of a solar one. Watch what that does over time — why Passover would slide out of spring if nobody stopped it, why the Hebrew calendar yanks it back and the Islamic calendar lets it wander, and how the same machinery bears on a two-thousand-year-old argument about a Friday in April.
Hebrew calendar
Which years actually fit the crucifixion?
Every gospel puts the death on a Friday, before the Sabbath, during Pilate's prefecture (AD 26–36 — the one point all four gospels and Tacitus agree on). The gospels then disagree on which Friday: John has the death on 14 Nisan, as the Passover lambs were killed; the synoptics imply 15 Nisan. Pick a rule and search — the engine reconstructs each year's Nisan from the moon and tells you the weekday.
The calendar alone allows more than two Fridays in this window. Scholars narrow it to AD 30 and AD 33 using non-calendrical evidence — the length of Jesus's ministry and John the Baptist starting in Tiberius's 15th year. Those are historical arguments and no ephemeris adjudicates them — but they can be enumerated the same way the astronomy is. All 3,456 combinations of them →
| Year | Nisan 14 | Weekday | Notes |
|---|
Proof is not the standard, because there isn't one to meet.
Proof belongs to mathematics. History does not deal in it, and neither does science — science deals in what the evidence will bear, and revises when it bears something else. Held to the standard of proof, almost nothing about the ancient world survives, including a great deal nobody disputes. So the useful question is not what can be proved. It is what the evidence carries, and where exactly it stops.
What the evidence carries
That a man existed, was executed by crucifixion under Pontius Pilate, in Judaea, inside the ten years of his prefecture. That is attested across several independent lines — not merely several documents, most of which copy one another — and it is not seriously contested by working historians of any religious persuasion.
Where it stops
Which year. Which day. Whether the last supper was a Passover meal at all. The astronomy narrows it to a handful of candidates and cannot choose between them; the narrowing to one is done by four separate readings of texts, any one of which settles it single-handed; and no object yet dug up moves it by a day.
What this tool does and doesn't know
The cycle itself is disputed
The Metonic leap-year pattern this dial animates (years 3, 6, 8, 11, 14, 17, 19) is the modern one. The Encyclopaedia Judaica records rival intercalation sequences still in use as late as the 10th century — (1,4,6,9,12,15,17), (3,5,8,11,14,16,19), (2,5,7,10,13,16,18). Which pattern first-century Judea used is not settled. The wheel you're watching is one reconstruction among several.
Conjunction is not first light
The engine computes the astronomical new moon (conjunction) to the minute. But a month began when the crescent was first seen, one to two evenings later, depending on weather and geometry. That gap is modelled here by a crude age-at-sunset rule. Move the slider and dates shift — because the real answer genuinely shifts with the model.
The eclipse is an argument, not a proof
A partial lunar eclipse rose over Jerusalem near 6:20 pm on 3 April AD 33 — the moon already reddened — and it was the only one visible at Passover across AD 26–36. Humphreys & Waddington tie it to "the moon turned to blood" (Acts 2:20). It is suggestive and popular. It has also been contested in print for thirty-five years, on grounds of simple visibility.
The Islamic side is uncertain too
The drift you see in Islamic mode follows from abolishing intercalation (nasī') per Qur'an 9:36–37, proclaimed at the Farewell Pilgrimage. But if nasī' did mean intercalation, some early-Islamic event dates (the Hijra, Badr, Uhud) could themselves be off by one to three lunar months. Lunar dating is slippery on both sides of this instrument.