Chinese New Year
When Chinese New Year falls, and how firmly
Chinese New Year is not a date anyone chose. It follows from four rules about new moons and the winter solstice, reckoned in China Standard Time — and because every rule turns on which day an instant falls on, a handful of years genuinely cannot be settled by computation. Those are named rather than guessed at.
Ten years — and how close each calculation came to a China-time midnight
| Year | New Year | Animal | Leap month | Closest call |
|---|
The four rules
A month begins on the day of a new moon, reckoned in China Standard Time — the calendar is defined for a meridian, not for wherever you happen to be. The month containing the winter solstice is always month 11. If thirteen new moons fall between one month 11 and the next, one of those months is a leap month: the first one containing no major solar term. And Chinese New Year is the first day of month 1, which is two months after month 11, or three if a leap month has come between.
That is the whole definition. Everything else — the animal, the stem, the sixty-year cycle — follows from counting. There is no table to look up and nothing to decide.
Why some years cannot be settled here
Every one of those rules turns on which calendar day an instant falls on in China. Across 1900 to 2100 there are roughly 2,600 new moons and 2,400 solar terms, so by simple arithmetic some of them land within seconds of a China-time midnight. When that happens, the day assignment is not a matter of computing more carefully — it is beyond what any finite-precision calculation can settle.
This is not a hypothetical about the far future. Nine years between 1900 and 2100 have a decision resting on a margin too fine for this engine, the closest being about six seconds. It is also why the authoritative Chinese calendar is published by an observatory rather than derived independently by everyone, and why specific years have been argued about in print.
So every answer here carries the margin that decided it, and where that margin is too small the page says the year is not settled rather than printing a date that looks as confident as the others. For what it is worth, the flagged years still agree with the published dates — but agreeing and being provable are different things, and only one of them is something to publish.
The leap month, and the year everybody argued about
A leap month has no number of its own: it repeats the number of the month before it, so a leap month following the sixth is another sixth month. Which month gets doubled is decided by the solar terms, not by a pattern, which is why it moves around — 2025 repeats the sixth month, 2028 the fifth, 2031 the third.
The interesting case is the Chinese year beginning in 2034, which carries a leap eleventh month — the first since 1645. Calendars that place leap months by habit rather than by the solar-term test get that year wrong, and several published ones did. It is the single best check on an implementation, which is why it has its own assertion in this site’s test suite.