Moon days

The three rules

The moon is full at an instant, not on a day. At some exact moment the sun and the moon stand opposite each other along the ecliptic — a syzygy — and that moment is the same everywhere. What is not the same everywhere is the date on the wall.

1. The syzygy on your calendar. Take the instant and read it off your own clock. The full moon of 26 October 2026 happens at 04:11 UTC; in Denver that is the evening of the twenty-fifth. Same moon, earlier date.
2. The syzygy on Mysore’s calendar. Most published moon-day lists in the lineage are Mysore’s dates, copied outward. India runs five and a half hours ahead of Greenwich and eleven and a half to twelve and a half ahead of Denver, so a syzygy in the American evening is already tomorrow there. This is the single commonest reason two shalas close on different mornings.
3. The tithi at your sunrise. A pañcāṅga does not ask what time the moon was full. It asks which lunar day was running when the sun came up. The month is cut into thirty tithis of twelve degrees of elongation each, and pūrṇimā — the fifteenth — ends at the full moon. So the lunar day of the full moon is mostly the day before it. If the syzygy comes after your sunrise, this rule names the earlier date, and it does so for a reason that has nothing to do with time zones.

Rules one and two split on longitude. Rules one and three split on the hour of the syzygy relative to dawn. They are independent, so all three can name two different dates at once, and occasionally three.

Two edge cases, marked where they occur

A tithi runs a little under or a little over twenty-four hours, because the moon’s speed along its orbit is not constant. When it runs short it can open and close between two sunrises, so no day carries it at dawn at all — a kṣaya tithi, marked  †. When it runs long two consecutive sunrises can fall inside it — a vṛddhi tithi, marked  ‡. For the skipped case this page falls back to the day the tithi ends and says so; for the doubled case it takes the earlier sunrise. Both are conventions, not computations, and both are stated on the row rather than hidden.

Far enough north or south the sun may not rise or not set at all. At those latitudes the sunrise rule has nothing to stand on, and the page says that instead of inventing a dawn.

How it is computed

The syzygy comes from Meeus, Astronomical Algorithms, chapter 49 — the full correction series, not the mean lunation. Against the US Naval Observatory’s published tables, every new and full moon from 2024 to 2030 agrees to under a minute. The tithi needs the apparent longitudes of both bodies, which come from chapters 47 and 25; sunrise from chapter 15, for your own latitude and longitude. Nutation cancels in the elongation, so the tithi does not care about it.

The weakest link is the solar longitude, good to about a hundredth of a degree. The elongation opens at rather more than twelve degrees a day, so a hundredth of a degree is about seventy seconds; sunrise is good to about a minute. Between them, a tithi boundary that falls within three minutes of a sunrise is beyond what this arithmetic can resolve.

That is not hypothetical. Over the fifty-six cities named here and six years — 8,300 moon days — fifty-five fall inside that margin: Auckland on 14 January 2029 by three seconds, Paris on 26 October 2026 by fourteen. Those rows are marked ※, and on them the page says the two dates cannot be told apart rather than picking one and sounding certain. An almanac computed from the full ephemeris would settle them; this one will not pretend to.

The closest cases are also the ones the page calls skipped or doubled, which is not a coincidence: a boundary landing within seconds of a sunrise is exactly what makes a tithi ambiguous at dawn in the first place.

Where it computes from

Only the sunrise rule needs a place; the other two need a time zone alone. Search names fifty-six cities with coordinates taken from OpenStreetMap, and behind them every zone in the tz database — some four hundred — each with the coordinate that database files it under.

Those two are not the same thing, and the difference matters most exactly where this practice came from. The tz database files Asia/Kolkata at Kolkata, which is thirteen hundred kilometres and ten degrees of latitude from Mysore; sunrise there differs by the better part of half an hour in winter. So Mysore is named in its own right, as are Goa, Rishikesh and the rest. When a zone’s reference point is being used rather than a named city, the page says so, and there is a field for typing coordinates outright.

Which one should you follow?

Your shala’s. Moon days are an observance, and a room decides its own — sometimes for astronomical reasons, sometimes because everyone has agreed for years that the class is on Tuesday. A published list that differs from this page is not thereby wrong. This tool exists to show you why two calendars differ, so that a disagreement stops being a mystery, not to overrule either.

Ashtanga Yoga Denver publishes its own moon days, as most shalas do. Those are the dates that close that room.

Read more

• Moon days, and where they actually came from

  The reason everyone gives cannot be traced to Guruji. The best-documented origin is a Sanskrit college’s holiday calendar.

• The pull of the moon

  About one nanogram across your whole body. A mosquito on your arm beats it two thousand times over.

• The moon and sleep

  Twenty-five minutes of later sleep onset, biggest where there is no electric light — and an argument about why.

• The Primary Series

  The sequence, and the days it is practised on.

Everything runs in the browser. Your location stays on this device.