What's changed
Until now we calculated eclipse contact times—when the partial phase begins, when totality begins, when it ends—using a correct method but one that carried two simplifications. The first, less precise planetary ephemerides than are available today. The second, and more important, a fixed value for something that actually changes: exactly how much the Earth has rotated.
That second piece of data can't be predicted years in advance. It's measured and published daily, and it varies as the Earth speeds up or slows down. Until now we used a frozen number; now we always use the latest available measurement.
The same eclipse, measured better
The eclipse hasn't moved. What's moved is how we translate its geometry into a clock time. The difference is around eight seconds for 2027 and 2028, and about five for 2026 — yes, the one that has already happened changes too.
It might seem small, and in many contexts it is. But totality sometimes lasts less than a minute, and there's a decision that depends precisely on those seconds: when you can safely remove the solar filter. There ten seconds isn't a detail.
Why you might see differences from other sources
If you compare our times with those from other published sources—including official platforms and commonly used applications—you'll probably find differences of a few seconds. The usual cause isn't the geometry: on that everyone agrees with high accuracy. It's the Earth rotation value we mentioned before: many of those publications were generated years ago with an estimate we now know is out of date, and they haven't been recalculated since then.
And there's a detail worth knowing, because it invites you to do a subtraction that doesn't work: the difference in the Earth rotation value doesn't transfer directly to the contact time. The Earth rotates while the Moon's shadow advances, so the final effect depends on each eclipse's geometry. It's not the same in 2026 as in 2027 or 2028.
What remains imprecise, and we say so
There are things we don't publish with more precision than they have. The moment of maximum is one: the curve is so flat at its lowest point that different methods place it several seconds apart, even though the Sun's appearance is indistinguishable. And near the edge of the totality band, the useful question isn't how many seconds it lasts, but whether totality is visible at all.
The level of detail also varies between eclipses. For 2026 and 2027 we apply a correction for the real relief of the Moon's edge: its mountains and valleys shift the exact instant the last ray of sun disappears. For 2028 it hasn't been calculated yet, and we prefer to say so rather than present the three eclipses as equivalent.
Our goal
Our goal is to offer the most accurate eclipse time calculation you can find, and explain it completely: what data we use, what we correct and what we don't.
The technical detail, step by step, is on the methodology page.