Ussher's equinox, by hand
Today is Ussher's creation date: the night before Sunday 23 Oct 4004 BC (Julian). By hand, JDN 259258 is a Sunday. The equinox, from Meeus's sun extrapolated 60 centuries back, is 23 Oct 21:35 TT, with the sun at perigee. The same method gets this year's equinox within half an hour. Subtract ΔT, about 30 h, and it lands on Saturday 22 Oct near sunset in Jerusalem, the moment Ussher named. Which date it was depends less on the orbit than on how much tides have slowed the Earth, and that's extrapolated. Coefficients are from memory.
Ussher's equinox, by hand
Today is 23 October. That's Archbishop Ussher's date for creation: "the entrance of the night preceding the 23rd day of October" in Julian Period year 710, which is 4004 BC. He wanted the year to begin in autumn and the first day to be a Sunday, so he took the Sunday by the autumnal equinox. Two questions: was it a Sunday, and when was the equinox?
The weekday
4004 BC is astronomical year −4003. With the standard Julian-calendar day-number formula (y = −4003 + 4800 = 797, m = 7), the date 23 Oct gives 23 + 214 + 365·797 + 199 − 32083 = JDN 259258. (259258 + 1) mod 7 = 0, so it was a Sunday. (The same formula gives Thursday for Julian 4 Oct 1582, which is right.)
The equinox
I used Meeus's low-precision sun (mean longitude L, mean anomaly M, eccentricity e, equation of centre). The coefficients are from memory. First I tested it on this year: it puts the 2026 September equinox at about 00:30 UT on 23 Sep, and USNO says 00:05. Good to well under an hour.
For noon on 23 Oct 4004 BC (T = −60.0215 centuries from 2000):
- L ≈ 179.66°, M ≈ −1.7°: the sun was almost exactly at perigee. Back then perihelion fell at the autumnal equinox.
- e ≈ 0.0192, equation of centre −0.07°, true longitude ≈ 179.60°.
- The sun moves about 1.024° a day near perigee, so it reaches 180° about 9.6 hours later.
Equinox ≈ 23 October, 21:35 Terrestrial Time.
But Terrestrial Time runs on atomic seconds, and the Earth has been slowing down since. The gap, ΔT, is extrapolated with a parabola (Morrison–Stephenson, from memory) to about 30 hours for 4000 BC. Subtract it:
- 22 October, about 15:30 UT
- about 17:50 local time in Jerusalem, 18:25 in Babylon
- sunset at an equinox is about 18:00
So on these numbers the equinox fell on Saturday evening, near sunset. That's the start of the night before the 23rd, close to the moment Ussher named.
What I'd stand behind
- In Terrestrial Time the equinox is on the 23rd, and that matches the "Sunday, October 23" that modern calculations are quoted as giving. In Earth-rotation time it's the evening of the 22nd. The 30 hours of ΔT alone carry it across a date line. So which day it fell on isn't settled by the orbit. It depends on how much tidal friction has slowed the Earth over six thousand years. That's extrapolated, because there are no eclipse records that old, and it's uncertain by hours.
- "Near sunset" is luck. "Saturday evening, give or take" is the claim.
- Ussher's own tables (probably Kepler's Rudolphine) reportedly put the equinox on Tuesday 25 October. They were about two days off, but his rule gives the same Sunday either way.
Caveats
Meeus's series is extrapolated 60 centuries past where it was fitted. Its square term alone is worth about a day, and an error in it would be a fraction of that. The coefficients and the ΔT formula are from memory. The only check against a real ephemeris is this year's equinox.
Sources: Ussher chronology (Wikipedia), USNO seasons, 2026.