In New York, the Sun sets earliest on Monday, December 7, 2026, at about 4:28 PM. The shortest day comes two weeks later. The solstice falls at 20:50 UTC on December 21 (3:50 PM EST), and on that day New York gets roughly 9 hours 15 minutes of daylight, nearly seven minutes less than on December 7. The Sun still sets later on the solstice, around 4:32 PM. The latest sunrise, about 7:20 AM, does not arrive until January 4, 2027.

So the shortest day is neither the day with the earliest evening nor the day with the darkest morning. It falls between them. The reason is the equation of time: the gap between the time a sundial shows and the time your clock shows.

Sunset Is Solar Noon Plus Half the Daylight

Each day has a midpoint, solar noon, when the Sun crosses your meridian and reaches its highest point. Sunrise and sunset sit symmetrically around it. That splits sunset time into two parts, and each part changes on its own schedule.

If solar noon stayed at the same clock time every day, sunset and sunrise would both reach their extremes on the solstice, because day length bottoms out then. Solar noon does not stay put. In New York it falls at 11:45 AM EST on December 1, 11:54 AM on December 21, and 12:01 PM on January 4. That is a 16-minute slide in five weeks, and sunrise and sunset move with it.

The day length term barely moves during this stretch. Near a solstice the Sun’s declination flattens out at its southern limit, so daylight changes by only seconds a day. In New York the daily loss is about 75 seconds on December 1, about 27 seconds on December 14, and close to zero on December 21. Over the same weeks, solar noon keeps drifting later by 23 to 30 seconds a day. Once the drift is larger than half the daily daylight loss, sunset starts moving later, even though the days are still getting shorter.

Why Solar Noon Slips About 30 Seconds a Day in December

A clock assumes the Sun moves eastward against the stars at a steady 0.9856° per day. Earth has to turn through that extra angle each day to bring the Sun back to the meridian, and every degree costs four minutes of rotation (360° in 24 hours). When the real Sun moves faster than average, solar noon arrives later by the clock. In late December, two effects push in that same direction.

The first is axial tilt. The Sun travels along the ecliptic, which is tilted 23.44° to the celestial equator, but the timing of noon depends on its motion measured along the equator. At a solstice the Sun moves parallel to the equator and sits 23.44° south of it, where the hour circles run closer together. Its effective eastward speed rises by a factor of 1/cos(23.44°), about 1.09. The second is orbital eccentricity. Earth reaches perihelion on January 3, 2027, at 02:33 UTC. Near that point, Earth’s angular speed around the Sun is about 3.4% above average, and the Sun’s apparent motion across the sky speeds up by the same proportion.

This matches the computed solar noon times: New York’s noon moves 29 to 30 seconds later per day through the week of the solstice. Taking the factors one at a time, tilt accounts for about 21 of those 30 seconds and eccentricity for about 9. That split is our own calculation from standard orbital values, not a figure from a published table. It does show that explanations which mention only Earth’s elliptical orbit leave out the larger part of the effect.

The winter solstice is the shortest day, not the earliest evening. December sunsets turn around two weeks early because solar noon drifts about 30 seconds later each day, and by early December that drift outweighs the shrinking daylight. worldtimedata

The yearly extremes of the equation of time fit the same picture. A sundial runs about 16.5 minutes ahead of mean time in early November and just over 14 minutes behind it around February 11, and the two agree around December 25 to 26. In late December the sundial is losing its autumn lead faster than at any other time of year. That is why evening light starts coming back before the solstice.

The Day the Two Drifts Cancel: New York Week by Week

Running the numbers for three December dates in New York shows the turning point. On December 1, daylight shrinks by about 75 seconds, which pulls sunset about 37 seconds earlier, while solar noon pushes it about 23 seconds later. The net result is about 14 seconds earlier. On December 7, daylight shrinks by about 54 seconds (27 on the sunset side) and noon drifts about 26. The two nearly cancel, and that is the earliest sunset. On December 14, the shrinkage is down to 27 seconds (about 13 on the sunset side) while noon drifts 29, so sunset is already moving about 15 seconds later each day.

Times were computed with the NOAA solar position equations for New York City (40.71° N), using the standard refraction allowance of 0.833°, and rounded to the nearest minute. Two things in the table stand out. First, the minimum is shallow. From December 1 to December 14, New York’s sunset stays between about 4:28 and 4:29 PM. The “earliest sunset” is really a two-week plateau, and almanacs that use a slightly different horizon model can name a date one day apart. Second, December 7 and January 4 have almost the same daylight, about 9 hours 21 minutes, yet every event on January 4 happens 13 to 14 minutes later by the clock. That gap is entirely the solar noon drift between the two dates.

The same drift explains why January mornings feel darker than December ones. From the solstice to January 4, New York gains about 10 minutes of evening light and loses about 4 minutes of morning light. Total daylight rises by roughly 6 minutes, but sunrise times keep slipping later until early January, so none of that gain reaches the morning commute before then.

Latitude Decides the Date

On any given date the solar noon drift is the same everywhere on Earth, because it comes from Earth’s orbit and tilt, not from where you stand. What changes with latitude is how fast day length changes. The farther you are from the equator, the more each small shift in the Sun’s declination stretches or shrinks your day. At high latitudes the shrinking daylight overpowers the noon drift until just before the solstice. Near the tropics daylight hardly changes, so the noon drift takes over weeks earlier.

All times are local standard time for winter 2026 to 2027, computed the same way as the New York table. From Key West to Reykjavik, the earliest sunset moves from late November to three days before the solstice, and the gap between the two extremes shrinks from about six and a half weeks to one week. North of the Arctic Circle the question no longer applies: during polar night the Sun does not rise at all, so there is no sunset to time.

At the equator the pattern stretches as far as it can go. Daylight in Singapore (1.35° N) varies by less than ten minutes over the whole year, so sunrise and sunset follow solar noon almost exactly. The earliest sunset there comes in early November and the latest sunrise around February 9, close to the dates when the sundial is furthest ahead of and furthest behind the clock. In the tropics, the “December effect” is really the equation of time spread across the whole autumn and winter.

Longitude, Time Zones and Daylight Saving Do Not Move the Date

Your position within a time zone changes the clock time of every sunset, but not the date of the earliest one. Being farther west in a zone delays solar noon by the same number of minutes every day. That constant offset shifts the whole curve up or down without moving its lowest point. Madrid is a clear example. It sits at nearly the same latitude as New York and has its earliest sunset on the same date, December 7, but at about 5:48 PM, because Spain keeps Central European Time even though Madrid lies west of Greenwich.

Daylight saving time works the same way: it is a fixed one-hour shift. In 2026, US clocks fell back on November 1 and most of Europe’s on October 25, weeks before any of the minimums in the table, so every date above falls in standard time. The fall-back change moves sunset by a full hour overnight, more than the Sun moves it in months, but that change comes from the clock, not from the sky.

June Plays the Same Trick at Less Than Half Strength

The June solstice has the same asymmetry in a milder form. In New York the earliest sunrise in 2026 came on June 14 and the latest sunset on June 27, about a week on each side, compared with two weeks in December. The tilt factor is identical at both solstices. The difference is the orbit. In early July Earth is near aphelion and at its slowest, so the eccentricity term works against the tilt instead of adding to it. By the same arithmetic as above, solar noon in late June drifts only about 13 seconds a day, less than half the December rate, and the extremes stay closer to the solstice.

In the Southern Hemisphere the December solstice is the longest day, and the same 30-second drift splits the summer extremes. Sydney’s earliest sunrise in 2026 falls on December 5, at about 5:37 AM AEDT, and its latest sunset on January 7, 2027, at about 8:10 PM. Buenos Aires, at almost the same latitude, follows nearly the same dates: earliest sunrise December 5, latest sunset January 6. Because the drift pushes every event later in both hemispheres, the “early” records (the earliest sunset in the north, the earliest sunrise in the south) arrive in early December, and the “late” records (the latest sunrise in the north, the latest sunset in the south) wait until the first days of January.