Science & Technology · The Record
NOAA office dates equal day and night to September 25, three days after the equinox
An Associated Press guide carried on September 22 told readers light and darkness split the day evenly. A National Weather Service office names a different date, and a second National Weather Service page prints day lengths of 12 hours and 6.5 to 16 minutes depending on latitude.
The Associated Press guide to the fall equinox went out under a headline that left no room for a qualifier: "Day and night get equal billing today. A guide to the fall equinox." The copy captured for this piece is the Anchorage Daily News syndication of the wire item, whose URL path dates it to September 22, 2026. The document carries no individual byline - the credit line names the wire service and no person - and its opening sentence repeats the headline's arithmetic, telling readers that the equinox is the day light and darkness split the day evenly between them.
The rest of the wire guide is a short service piece. It tells readers the equinox marks the beginning of fall in the Northern Hemisphere and of spring in the Southern, that the Sun stands directly overhead at noon on the equator, that Earth's tilt at that moment leans neither toward the Sun nor away from it, and that the word itself comes from two Latin words meaning equal and night. That is the claim half: an even split, stated as the day's defining fact, in the two places a reader is most likely to stop reading.
The federal record does not say that. The National Weather Service office in Central Illinois keeps a page called "Autumn Begins" that takes the etymology apart rather than resting on it. The page grants the Latin root and then withdraws the inference from it: "the length of day and night is not quite the same on this date." It then does something no explainer built on the word "equinox" can do - it names a different date for the thing the word promises. At that office's latitude, the page states, the date of equal day and night is September 25th this year.
The same page timestamps the equinox itself to Tuesday, September 22nd at 7:05 pm CDT, the local frame that page uses. Converting that stamp to Coordinated Universal Time pushes the moment past midnight and onto September 23 - an arithmetic conversion made by this desk, not a figure the page prints. So Tuesday and September 23 can both name the same instant, which is the second thing a single-sentence lede flattens.
The mechanism behind the gap is not exotic, and two weather service offices spell it out in near-identical wording. The Cleveland office attributes the "nearly" equal hours to "refraction of sunlight or a bending of the light's rays that causes the sun to appear above the horizon when the actual position of the sun is below the horizon." The Quad Cities office carries that same explanation in the same words. Cleveland adds the geometric half: the days become a little longer at the higher latitudes, because the Sun takes longer to rise and to set there.
That produces a pattern the word "evenly" cannot carry. The Quad Cities page publishes figures: on the equinox and for several days on either side of it, day length runs about 12 hours and six and one-half minutes at the equator, about 12 hours and 8 minutes at 30 degrees latitude, and about 12 hours and 16 minutes at 60 degrees latitude. Subtract the twelve hours and what is left is the surplus daylight - roughly 6.5 minutes at the equator, 8 minutes at 30 degrees, 16 minutes at 60 degrees. That subtraction is this desk's; the weather service prints the totals, not the differences. The imbalance is smallest at the equator and grows with distance from it, which is the reverse of a claim pitched as a uniform global event.
View the data
| Value (about min past 12h) | |
|---|---|
| Equator | 6.5 about min past 12h |
| 30 degrees latitude | 8 about min past 12h |
| 60 degrees latitude | 16 about min past 12h |
Source: Daily Pol calculation from day-length figures published by the National Weather Service, Quad Cities (NOAA), 'The Seasons (Equinoxes and Solstices) Page' · Daily Pol graphic
NASA Science carries a skywatching item, "Embracing the Equinox," credited by the site to Kat Troche. The capture available to this desk returned only fragments of it: that the equinox is the moment "the center of the Sun crosses the plane of our planet's equator," and that "the Earth's North and South Poles are simultaneously lit by sunlight." Both are statements about geometry and illumination rather than about clock hours. The captured fragments do not include any statement about day length, so this piece makes no claim about what the full post tells readers on that point.
NASA's own archive is less tidy. An image page on nasa.gov titled "The Equal Night," dated Mar 23, 2008 and built around a photograph the space shuttle Endeavour crew recorded a few weeks after the September 1994 equinox, still tells readers they will experience "nearly 12 hours of daylight and 12 hours of darkness." That sentence keeps the hedge - "nearly" - that the wire lede drops. It is the same framing at one remove: close enough to equal that the qualifier does the quiet work, and the reader carries away the round number.
The wire story does the same thing to itself. In the sixth body paragraph of the captured text, it hedges correctly, saying day and night last "almost the same amount of time," with some places picking up a few extra minutes depending on where on the planet they are. Read as analysis: the inaccuracy here is not in the reporting, it is in the compression. The headline and the first sentence assert an even split; the body, several screens down, concedes the split is approximate and location-dependent. One story holds both.
On the wire guide's other statements, the captured federal pages split. The tilt sentence is corroborated almost word for word - the Quad Cities and Cleveland pages both describe an axis tilted neither toward nor away from the Sun - and the Latin etymology matches the weather service's own gloss of aequus and nox. The autumn-and-spring framing does appear in a captured record, but in NASA's 2008 image caption, not on any weather service page read for this piece. And the statement that the Sun stands directly overhead at noon on the equator appears in no captured document other than the wire item itself, so this piece neither confirms nor disputes it.
The Central Illinois page makes its own forecast falsifiable, and the deadline is two days out: on September 25, at that office's latitude, sunrise-to-sunset and sunset-to-sunrise should come out even, and they should not have done so on the 22nd. Anyone with a published sunrise and sunset table for that latitude can settle it without a telescope.
One more thing is checkable on a date. NASA's 2008 image page is a static archive item, not a skywatching post on a publishing cycle, and pages of that kind are rarely revisited for a hedge. The prediction here: checked on December 21, 2026, nasa.gov's "The Equal Night" page will still carry the sentence about nearly 12 hours of daylight and 12 hours of darkness, unedited. The next test of the wire framing arrives with the March equinox, if the wire writes the guide again.