An eclipse is one event. It happens at one set of instants, and those instants are the same whether you read them in Varanasi, Lisbon or Auckland.
Whether you can see it is an entirely different question, and it has a different answer in each of those places.
That distinction is the whole of this page. It sounds like pedantry until you notice what rests on it.
Sutak carries a condition, and the condition is visibility
Sutak is the period before and during an eclipse in which ordinary activity is set aside. The rule that fixes its length is quoted everywhere: twelve hours before a solar eclipse, nine before a lunar one.
The rule that fixes whether it applies at all is quoted much less. Sutak is observed where the eclipse is visible. Drik Panchang states it plainly and enforces it in its data, ask its pages about an eclipse that misses your city and every sutak field reads "Not Applicable".
So a page that prints a sutak window without testing visibility has not made a rounding error. It has applied a rule while skipping the rule's own precondition.
The test is not subtle. A lunar eclipse is visible wherever the Moon is above the horizon while it is happening. If the Moon is below your horizon throughout, there is nothing to see, and the tradition says there is nothing to observe either.
Which is why this page prints altitudes
Every contact time on this page carries the altitude of the eclipsed body, in degrees, as seen from the city you selected. Positive is above the horizon. Negative is below it.
Here is what that looks like for the total lunar eclipse of 3 March 2026, read from Varanasi:
| Contact | Moon's altitude |
|---|---|
| Penumbral begins | −45.1° |
| Partial begins | −33.4° |
| Totality begins | −18.5° |
| Greatest eclipse | −12.3° |
| Totality ends | −5.9° |
| Partial ends | +9.6° |
| Penumbral ends | +23.9° |
Read it downward. The eclipse begins with the Moon forty-five degrees below the horizon, roughly underfoot. Totality happens, entirely, while the Moon is still down. The Moon rises during the closing partial phase, and an observer in Varanasi sees the last hour or so of a total lunar eclipse and no totality whatsoever.
You do not have to take my word for any of that. The altitudes are printed, and the astronomy that produces them is the same astronomy behind a sunrise time.
The partly visible case is common, and it is not rounded away
That March 2026 eclipse is not an oddity. The Moon rises or sets in the middle of a lunar eclipse often, and there are Sanskrit terms for both cases, grastodaya for rising already eclipsed, grastastamaya for setting while still eclipsed. Monier-Williams and Apte both gloss the second one. The categories are old because the situation is common.
The page reports the window that is actually watchable, found by locating the horizon crossing rather than by rounding the eclipse to visible or invisible. For 3 March 2026 at Varanasi that window opens at moonrise and closes at the last penumbral contact.
Three rules that are easy to get wrong
Only the umbral phase counts. A lunar eclipse begins when the Moon touches the penumbra, the outer shadow, where the dimming is so slight it is often invisible to the naked eye. Ritual observance runs from first to last contact with the umbra, the dark shadow. A calculator that clips the whole penumbral span to the horizon produces a window about an hour longer at each end than any Indian panchang publishes.
So a penumbral eclipse carries no sutak at all. It never reaches the umbra. This is categorical rather than a matter of how faint the eclipse is, which is worth stating because the obvious guess is wrong: the penumbral eclipse of 20 February 2027 has a penumbral magnitude of 0.91, comfortably above the roughly 0.6 at which shading becomes detectable, and it still carries none.
Sutak ends at the horizon, not at last contact. If the Moon sets while still eclipsed, the remaining phases happen where you cannot see them, and sutak ends at moonset. For London on 28 August 2026 that is forty-one minutes before the last umbral contact.
The part where the published times disagree, and why
Here the page has something to say that I have not seen anywhere else.
Sutak does not begin a flat twelve or nine hours before the eclipse. That is the figure every source quotes (including the panchangs' own explainer pages) and it is not what those same panchangs compute.
A prahar is an eighth of the day. Not three hours: the daylight is divided into four parts and the night into four more, so a prahar runs from under two and a half hours to over three and a half depending on the season and how far north you are. Sutak begins a whole number of prahars back from the start of the prahar in which the eclipse begins locally, four for a solar eclipse, three for a lunar one, and one for children, the elderly and the unwell.
That rule reproduces published windows to the minute. For 3 March 2026, Delhi comes out at 09:38 against a published 09:39, and Bengaluru at 09:33 against 09:33. For 7 September 2025 at Delhi (a case with the onset falling in a night prahar, which shifts the count by one) 12:18 against a published 12:19.
The flat nine hours would have given 12:56 for that last one. Nearly forty minutes out.
This page shows both figures and names the difference, because the round number is the one people have been told and the prahar figure is the one the almanacs use.
And a knife edge that is built into the geometry
A lunar eclipse happens at full moon. A full Moon rises within minutes of sunset, and that is what being opposite the Sun means. And sunset is itself a prahar boundary, the one dividing the last day-prahar from the first night-prahar.
So for every eclipse where the Moon rises already eclipsed, the moment the eclipse becomes locally visible sits almost exactly on a prahar boundary. Which side of it the onset falls decides sutak to within a whole prahar, close to three hours.
This is not a rare corner. It is guaranteed by the geometry, and it is a large part of why published sutak times for such eclipses disagree between publishers.
The page says when an eclipse is in that state, gives the gap to the boundary in minutes, and shows the window the other side would produce. Where it has to choose, it takes the reading the geometry supports: a full Moon rises at or after sunset, so the onset belongs to the night. Sunset here agrees with Drik Panchang to four seconds, while moonrise is convention-dependent to several minutes, and this site's moonrise sits within thirty seconds of the figure ARIES, an institute of India's Department of Science and Technology, published for Nainital that same evening.
Why a lunar eclipse is seen by half the world and a solar one by a sliver
There is an asymmetry here that explains a good deal of what the tables show, and it is not often spelled out.
A lunar eclipse is the Moon entering the Earth's shadow. The Moon is a single object, and it either is in the shadow or it is not. Anyone who can see the Moon at that moment sees the eclipse, which means everyone on the night side of the Earth: rather more than a hemisphere, since the eclipse takes hours and the Earth turns underneath it. Nothing about the appearance changes with where you stand.
A solar eclipse is the Moon's shadow falling on the Earth. That shadow is small. The dark inner cone that produces totality is at most about 270 kilometres across where it touches the ground, and it sweeps a narrow track. Step outside the track and the Sun is merely notched; step a thousand kilometres away and nothing happens at all.
So the two behave in opposite ways. Solar eclipses are the more frequent of the two, and most people alive have never seen one become total. Lunar eclipses are less frequent and far more widely seen. This is why the tables on this page treat them differently: a solar eclipse that misses your city is listed with no local contact times at all, because it genuinely has none there, while a lunar eclipse that misses you still has its global instants printed beside the note that the Moon was down.
Why eclipses arrive in pairs, about six months apart
Look at any run of eclipse dates and a pattern shows up quickly. They cluster. A solar eclipse turns up within a fortnight of a lunar one, then nothing for months, then another pair.
The reason is that the Moon's orbit is tilted about five degrees against the plane of the Earth's orbit. Most new moons pass above or below the Sun rather than across it, and most full moons pass above or below the Earth's shadow. An eclipse needs the Moon to be near one of the two points where the tilted orbit crosses that plane, and the Sun to be near the same place.
That alignment holds for a window of a few weeks, twice a year. Inside such a window a new moon gives a solar eclipse and the full moon a fortnight either side gives a lunar one, which is why they come in twos and threes. Between windows there is nothing to compute.
Those crossing points are the lunar nodes, and Indian astronomy has names for them: Rahu and Ketu. Whatever else has been built on those names, the underlying object is a geometric fact about two tilted orbits, and it is the same fact that predicts when the next eclipse season falls.
Why the ayanamsa selector does nothing here
Every other page on this site changes its answer when you switch between Lahiri, Raman and Krishnamurti. This one does not, and the reason is worth a sentence.
An eclipse is a geometric alignment of three bodies. It is not a position measured against the stars. The ayanamsa is an offset between the tropical and sidereal zodiacs, and shifting a coordinate system does not move the Moon into the Earth's shadow any earlier. So the contact times are identical under all three, and saying so out loud seemed better than leaving a selector that quietly does nothing.
Sutak, by contrast, moves a great deal, though not with the ayanamsa. It moves with the prahar arithmetic and with the moonrise convention, as above.
What the page does not compute
The path of totality. It tells you whether a solar eclipse is visible from your city and how much of the disc is covered there. It does not draw the central line or advise where to travel.
Which observances belong in the window. What may be eaten, cooked or kept during sutak, and the specific guidance given to pregnant women, vary by tradition and by family. The page computes when the window is. What fills it is for the practice you follow.
A single authoritative sutak time, in the cases where the eclipse begins near a prahar boundary. Both candidates are shown, because the choice genuinely moves the answer by three hours and different publishers make it differently. Printing one and calling it settled would be the easier thing to do and the less honest one.
What it does compute, and how you can check it
Contact times from an open-source ephemeris, with the altitude of the eclipsed body at each one. The visible sub-interval, found by locating the horizon crossing to the second. The sutak window on the prahar rule, the same window on the flat rule, and the gap between them.
The greatest-eclipse instants agree with published values to the minute, including the total solar eclipse of 8 April 2024, whose greatest point this computes at 25.3°N 104.1°W.
None of that is a claim you have to accept. Every number on the page is either an instant you can look up or an altitude you can check against where the Moon actually was.