Events · August 12, 2026

The 12 August 2026 Eclipse, Then the Perseids: the Darkest Night of the Decade

On 12 August 2026, at 20:17:12 Paris time, the Moon will cover 92.0% of the Sun’s disc as seen from the French capital. The Sun will be 7.7° above the horizon, almost due west (azimuth 283.8°). The eclipse begins at 19:22 with the Sun at 16.6°, and it does not end in the sky: it ends at the horizon, at 21:09, with the disc only 0.1° up. Two minutes later the Sun sets. One hundred and seven minutes of eclipse, the last of them played out in the haze near the horizon: you need a perfectly clear western horizon, and you need to be ready early.

Where will the 12 August 2026 eclipse be total?

Global maximum falls at 17:45:47 UT at 65.2° N, 25.2° W, off Iceland: Reykjavík sees the Sun vanish for 65 s at 17:48 local time, 24.6° above the horizon. The shadow then crosses the Atlantic and lands on Spain just as the Sun is setting there: Oviedo, 1 min 51 s of totality at 10.3°; Burgos, 1 min 47 s at 8.3°; Valencia, 1 min 06 s at 4.6°, all but resting on the rooftops.

Everywhere else the eclipse stays partial, but deep. From Paris, the Moon’s apparent diameter will be 32.5′ against 31.6′ for the Sun: it is larger than the Sun that evening, and so has more than enough to cover it entirely. What is missing is a few hundred kilometres of alignment.

Two rows deserve a careful read: Marseille shows 96.3% with the Sun at 3.0°, Berlin 84.8% at 3.5°. At those altitudes a hill, a building or a layer of haze is enough to cancel the whole thing. In Rome the computation gives 95.5% with the Sun at −1.9°: maximum happens there after sunset, which is why an obscuration map without the Sun’s altitude lies by omission.

How do you watch a solar eclipse safely?

Never look at the Sun without a filter designed for solar observation: ISO 12312-2 certified glasses for the naked eye, a full-aperture filter for any instrument. A Sun covered to 92% is still blinding. The remaining 8% is still tens of thousands of times brighter than the full Moon, and the lens of your eye concentrates it onto the retina, which has no pain receptors. The burn is painless, immediate and permanent.

Sunglasses, photographic film, smoked glass, CDs, X-ray plates, improvised filters: none of them will do. Never use binoculars or a telescope without a filter placed in front of the objective. Outside the band of totality there is no moment at which it can come off: the naked eye is only possible inside the band, and only during totality itself.

When does it actually get dark on 12 August?

After sunset, you wait. In Paris on 12 August, civil twilight ends at 21:47, nautical twilight at 22:31, and astronomical night (Sun below −18°, the only regime in which the sky background is no longer lit) does not begin until 23:23. It ends at 04:28, which makes 5 h 05 of true night.

Why do the 2026 Perseids fall in a moonless sky?

The Perseid peak falls on the same night. The stream, fed by comet 109P/Swift-Tuttle, enters the atmosphere at 59 km/s from a radiant at right ascension 48° and declination +58°, in Perseus. Its reference zenithal hourly rate is 100.

And that night the Moon is absent: 0.10% of the disc lit at 2 a.m. on 13 August, which is to say nothing. The eclipse is the new moon; it is the same body. The Moon that passes in front of the Sun at 20:17 sets at 21:11:57, one minute after it, and does not rise again that night.

We computed the lunar phase at 2 a.m. on the night of the peak for every year from 2020 to 2032: 2026 is the only year at zero. The two next best are 2023 (10.4%) and 2029 (11.4%). The worst are 2030 (99.6%) and 2022 (98.6%): a full Moon, and a meteor shower cut down to its brightest members alone.

How many meteors an hour will we see?

A zenithal hourly rate is not a count of meteors seen. The first factor is geometric: a low radiant only delivers a fraction sin(h) of its rate. From Paris we compute the radiant’s altitude hour by hour: 20.8° at 10 p.m., 29.7° at midnight, 42.4° at 2 a.m., 57.5° at 4 a.m., 65.4° at 5 a.m.

Applied to a rate of 100, that term alone caps the expectation at 36 meteors an hour at 10 p.m., 50 at midnight, 67 at 2 a.m. and 91 at 5 a.m. These are upper bounds, valid for a sky with a limiting magnitude of 6.5, that is to say a dark rural site; from a suburban garden you have to divide again, and we offer no factor for that because we cannot compute one honestly. What remains is the essential point: everything happens after one in the morning, and the best hour is the last one, before astronomical dawn on 13 August at 04:28.

What are the limits of these forecasts?

The times and obscurations are geometric. They contain no weather at all, and at 3° or 4° of altitude the weather decides everything.

Obscuration is a fraction of disc area, not a brightness. At 92%, the ambient light is still that of a hazy late afternoon: there is no darkness, no corona, no stars. The difference between 99% and 100% is not gradual, it is a difference in kind.

The Sun’s altitude is corrected for standard atmospheric refraction. Near the horizon, real refraction depends on temperature and pressure: allow a few tenths of a degree of uncertainty, which shifts the observable end of the eclipse by a few tens of seconds.

The zenithal hourly rate of 100 is an IMO reference value, not a forecast. The Perseids have had years at 150 and others at 60. The sin(h) factor is the only correction applied here; sky brightness depends on the observer’s exact location, and we treat it separately.

This article was translated from the French edition by AI, then reviewed before publication. The figures are not translated: times, altitudes, magnitudes and percentages are computed by SkyChart and are identical in both editions.

• Eclipse 2026-08-12
• Perseids