The earliest sunset is ten days before the shortest day

Ask anyone which day has the earliest sunset and they will say the shortest day. At 40°N, that is wrong by about two weeks.

Computed with the standard solar algorithm at 40°N on a zone’s central meridian:

Date	Sunrise	Sunset	Day length
November 5	06:33	16:53	10h 20m
November 20	06:50	16:40	9h 50m
December 1	07:02	16:35	9h 33m
December 7	07:08	16:34	9h 26m
December 15	07:14	16:35	9h 21m
December 22	07:18	16:38	9h 19m
January 4	07:21	16:47	9h 25m

Earliest sunset: around December 7. Shortest day: December 22. Latest sunrise: around January 4. Three different dates for what feels like one event.

Two effects, pulling in opposite directions

The Sun’s declination controls how long it is above the horizon. Near the solstice its rate of change slows to nothing — the declination is at an extreme, so the derivative is zero. If declination were the only thing happening, day length would be flat for a week either side of December 22, and both ends of the day would move together.

But it is not the only thing. Solar noon is not at a fixed clock time; it drifts because of the equation of time, which combines two effects:

• Obliquity. The Sun moves along the ecliptic, not the equator, so its projection onto the equator does not advance uniformly — this contributes a wave with a period of half a year.
• Orbital eccentricity. Earth moves fastest at perihelion in early January, so the Sun’s apparent motion along the ecliptic speeds up then — this contributes an annual wave.

The combination reaches about +16 minutes in early November and −14 minutes in mid-February. In December it is moving fast: between December 1 and December 22, solar noon shifts from 11:49 to 11:58 — nine minutes.

So through December, sunset is being pushed later by the slowing declination change and pulled earlier by the drifting noon, and the two cancel around December 7. Sunrise gets the same drift in the same direction, but its declination-driven term is moving the other way, so its extremes land elsewhere.

Why the asymmetry

The two ends of the day respond differently because sunrise and sunset sit on opposite sides of solar noon. A shift in noon moves both, but the declination term moves them in opposite directions near the solstice. Add a sine wave to a slowly changing line and the peaks land at different phases on each side.

That is why latest sunrise (January 4) is thirteen days after the solstice while earliest sunset (December 7) is fifteen days before it, rather than both being symmetric.

Where it gets dramatic

At 40°N the effect is about two weeks and easy to miss. Go north and it dominates. At 60°N the day length changes by only a couple of minutes a week around the solstice, so the 16-minute equation-of-time swing is the only visible motion: the earliest sunset can fall in mid-December and the latest sunrise in late December, with the shortest day between them.

Above the Arctic Circle the Sun does not rise at all in December, and the question becomes meaningless in the most literal way — a fact the rise/set table prints as “does not rise” rather than a time.

The practical consequence

If you are planning anything that depends on light — a photograph, a commute, a solar panel — the solstice is the wrong date to plan around. Sunset bottoms out in early December; sunrise bottoms out in early January. Both are computable, and both are in the table.

The mechanism behind the drift is the equation of time, and the other seasonal surprise — why the equinox is not twelve hours — is a separate effect with a separate cause.

Computed at 40°N, 0° longitude. Your longitude within your time zone shifts all these times but not the dates; the dates shift by a day or two with latitude.