The New Fire every 52 years

The Mexica calendar: 260 days of names and numbers, an 18-month year with five nameless days at its end, years named Rabbit, Reed, Flint and House — and a fire on a mountain that had to be drilled fresh every 52 years, or the world would end. ← Calendar Contraptions

Central Mexico ran two calendars side by side, like Bali and like the Maya. The tōnalpōhualli ("day count") is a 260-day wheel: every day carries a number from 1 to 13 and one of twenty names — 1 Crocodile, 2 Wind, 3 House, up to 13 Reed, then 1 Jaguar — and it never pauses. The xiuhpōhualli ("year count") is a 365-day year of eighteen twenty-day months plus five days that belong to no month at all, the dreaded nēmontēmi. The two wheels turn together, and because 260 and 365 share no factors, the full pattern takes 52 years — 18,980 days — to repeat. That is the machinery on this page. It is also the rare calendar whose reconstruction is still an open argument: the sources disagree about when the year began, about which god presided over which thirteen days, and about what day the last fire was drilled on. Where they disagree, this page shows both and marks it.

The reading

The day — tōnalpōhualli

The thirteen-day week — trecena

The year — xiuhpōhualli

The anchor: one battle

Every reading above hangs on a single dated event. The count of days is pinned where the histories agree: the day Tenochtitlan fell. Native and Spanish sources alike record 13 August 1521 (old style) as the day 1 Serpent, in a year named 3 House. From that one point, the whole 260-day count and the 52-year wheel are forced arithmetic. The other recorded dates from those years check against it — almost:

The year wheel

Years are named by a number 1–13 and one of four signs — Rabbit, Reed, Flint, House — and the names march forward one number and five signs each year, so the full wheel takes 52 years and the four thirteen-year bundles are called Rabbit, Reed, Flint and House after the year that opens them. A year is born on a day whose sign is fixed by arithmetic: Reed years begin on a Crocodile day, Flint on a Death day, House on a Monkey day, Rabbit on a Vulture day — the pattern the scholar Zelia Nuttall published in 1894, and which the anchor reproduces on its own.

The day the year is named for

The New Fire

Every 52 years — 18,980 days — the two wheels returned to their starting position, and that was a knife's edge. If the stars kept moving and no new fire was made, the Mexica expected the sun to fail and star-demons to come down and eat everyone. So on the last night of the cycle every fire in the realm was put out — hearth stones thrown away, pots smashed, the houses scrubbed — and at the top of the volcano Huixachtlan, across the lake from Tenochtitlan, a priest waited for Orion's Belt, which they called "the fire drill," to rise. Then a fire was drilled on the chest of a captive, the heart cut out to feed the flame, and runners carried the new fire to every temple and hearth. The last one the histories record was in 1507 — the year 2 Reed. When the next one fell due, the Aztec state had eleven years left.

The eighteen months and the five days

The year is eighteen months of twenty days, each opening on a Crocodile day, and then five days with no month, no name and no luck. Each month had its festival; the eyewitnesses who wrote the names down also wrote down when each month fell in our calendar — and their tables disagree with each other by 27 days, which is the single best clue to what went wrong (see the drift, below). The table shows each month's dates under this page's count for the picked year, beside the two eyewitness tables as printed.

The nameless days — nēmontēmi

The drift

Here is the flaw at the heart of the machine, and the reason the sources quarrel. The year has 365 days — every year, no exceptions. The seasons take 365.2422. Nothing is ever added back: no leap day, no rule, nothing. So the Mexica new year slides against the seasons by about one day every four years — a full thirteen days every 52-year cycle. The calendar the Spanish met in 1519 had been running for centuries; the eyewitnesses wrote their tables decades apart, after the institutions that managed the count had been destroyed. Each described the slide at a different moment, which is why Sahagún's year begins in February and Durán's in March, and why a modern count that never corrects is now more than four months out of season.

The slide, for the picked year

The book of days

The 260 days were painted in books — tonalamatl, "book of days" — laid out exactly like this: twenty columns of thirteen days, each column opened by a day numbered 1, and each thirteen-day stretch presided over by a god. The column lords are where the sources fall out worst: the two reference tables reproduced below agree on exactly one of twenty. That is not this page's error — divination manuals were never one book, and the surviving codices genuinely differ about which deity ruled which stretch. Both columns are shown, and the days they disagree on are marked.

Why the same number as the Maya?

260 × 73 = 18,980 = 52 × 365. The sacred count goes around 73 times while the year goes around 52 — that is the whole coincidence, and every 260-and-365 culture in Mesoamerica inherited it. The Maya long count treats the same 18,980-day Calendar Round as a gear in a machine that never stops and never forgets: dates roll forward through baktuns for millions of years. Central Mexico treated the same number as a life that could expire: the bundle of 52 years (xiuhmolpilli, "year bundle" — the strictly correct term is xiuhnelpilli) had to be tied shut and burned clean, and the world re-lit by hand. Same arithmetic, opposite theology.

The twenty names

Each day-sign owns a direction, cycling east, north, west, south, four at a time — the order never breaks. Wind and Rain were drawn as their gods (Ehēcatl and Tlāloc) rather than as weather. The signs also served as fate: a child born on a particular day carried that day's luck for life, which is what made the day-keepers and their books the most consulted profession in the city.

The stone that is not a calendar

The enormous "Aztec calendar stone" in Mexico City's anthropology museum — the one on every poster — shows the Five Suns cosmology and the date 13 Reed carved as a monument. It did not function as a calendar, and the encyclopedia is explicit about this: it "depicts calendrical symbols on its inner ring but did not function as an actual calendar." The real working machinery was paper, ink and record-keepers — none of which survived the fires of the conquest, which is why every reconstruction since has been an argument.

The correlations

How the Mexica count maps onto our calendar is called a correlation, and there are several. This page ships the one anchored where the evidence is strongest — Alfonso Caso's, built back from the fall of Tenochtitlan, the standard in much international scholarship. Rafael Tena's (the Mexican national institute's standard) adds a leap correction and starts the year in late February; some living practitioners re-tie the year to the equinox instead. In 1967 Caso surveyed 42 sources on when the Mexica year began: fourteen said one month, fourteen another, seven a third, and the rest scattered across five more. The encyclopedia's own live date box and its own printed year table disagree with each other to this day. When a calendar has no surviving keeper, even its friends fight about it — the honest response is to show the argument, not to hide it.

Honest limits

This page derives everything from one anchor (the fall of Tenochtitlan, 1 Serpent on 13 August 1521 old style) plus the printed 52-year table; every derived claim above is machine-checked against that anchor. That buys exactness at the anchor and honesty nowhere else: the count ships with no leap correction on purpose, so its months are now over four months out of season — that is the drift made visible, not a bug. The year-start implied by the anchor (20 February 1521 old style) matches the anchor set but is itself one of the field's open questions. Dates are whole days, no times of day, and nothing here should be read as a claim about when the Mexica actually observed any of this — the records are quoted as records, the arithmetic is labeled as arithmetic, and where they collide, both are shown.