Toltec

The Toltec matter out of all proportion to what survives of them, because of what came after. When the Mexica — the Aztecs — built Tenochtitlan in the fourteenth century, they did not claim to have invented civilisation. They claimed to have inherited it, from the Toltecs, whom they described as the first people to build cities, to write, to count the calendar, to work jade and gold, and to rule legitimately. In Nahuatl, toltecatl came to mean "artisan", and toltecayotl, "Toltec-ness", meant something close to culture itself, as opposed to chichimecayotl, the barbarism of the unconverted nomad. Everything the Aztecs valued in their own state they projected backwards onto Tula.

That projection is the central problem of this entry. Almost every written source for the Toltecs was set down after the Spanish conquest, three to six centuries after the events, by people with a political interest in the answer. The archaeology is rich but stubbornly silent: Tula has produced monumental sculpture, thousands of obsidian blades and a great deal of pottery, but no readable royal inscriptions, no king list carved in stone, no chronicle. The result is a subject where the chronicle and the trench describe two different places, and where the most honest account says so at every turn.

A word on the map. This entry treats the Toltec as a phenomenon of roughly AD 900 to 1150, the conventional dating and the span given on the site card. The map itself draws the territory from AD 875 to 1187 — two and a half decades early at the start and nearly four late at the end, because the underlying dataset interpolates polygons between dated snapshots rather than tracking a frontier. Nothing in the ground suggests Toltec rule began in 875 or ended in 1187.

At a glance

What did Teotihuacan's fall leave behind?

Teotihuacan was the largest city in the Americas for half a millennium. At its height, in the fourth and fifth centuries, it housed something like 100,000 people in a grid of apartment compounds covering more than 20 square kilometres, and its obsidian workshops supplied blades to most of Mesoamerica. Then, around AD 550, the temples and palaces along the Avenue of the Dead were looted and set on fire. The Pyramid of the Feathered Serpent was stripped of its sculpture; the elite compounds were smashed and burned.

Who did it is still argued. The old story of a foreign invasion has little support: there are no burned residential barrios, no mass graves of the kind an outside sack leaves, and no military architecture. Linda Manzanilla's reading, based on excavations at the compound of Xalla and elsewhere, is that the burning was an internal uprising against a ruling group that had lost its legitimacy — the destruction is concentrated precisely on the buildings that housed priests and nobles. Whatever happened, the state did not survive it. Population fell away through the seventh century, monumental construction stopped, and by about AD 650 the city was a provincial ruin with a few thousand inhabitants living among the temples.

The century and a half that followed is what Mesoamericanists call the Epiclassic, roughly AD 600 to 900: a period without a hegemon. In the Basin of Mexico, Teotihuacan's collapse released its provinces and its trade. Xochicalco in Morelos built a fortified hilltop capital with a pyramid covered in carved panels showing feathered serpents and seated lords; Cacaxtla in Tlaxcala decorated its palaces with murals of jaguar warriors and bird-men in a style closer to the Maya than to anything in central Mexico; El Tajín on the Gulf coast grew fat on the trade that had once flowed through Teotihuacan. Further south, the Classic Veracruz culture reached its height, and at Monte Albán the Zapotec capital began its long decline.

The diagnostic pottery of this interval is a red-on-buff ware painted with geometric designs, called Coyotlatelco after a site in the Basin. It appears in post-Teotihuacan levels almost everywhere in central Mexico, and it is usually taken as the signature of new populations moving in from the north and west — the ancestors, in the traditional account, of the Chichimeca. Whether that means migration or merely a change in taste within an existing population is one of the standing arguments of Mexican archaeology; the ceramic itself does not carry a passport.

In the Tula region the pattern is clear enough. The villages that had supplied Teotihuacan with lime, maize and labour — Villagrán, Chingú and their neighbours — were abandoned around AD 650. People moved up onto hilltops and defensible slopes, and a new settlement appeared at Tula Chico, a little cluster of platforms, two ballcourts and a pyramid that was ambitious by the standards of a Coyotlatelco village. It was not a great power. It was a well-placed local centre in a fragmented landscape, on the edge of the Pachuca obsidian beds and astride the corridor running north out of the Basin. For the next two centuries that was all it needed to be.

How did Tula become the largest city in Mesoamerica?

Sometime between AD 850 and 900, Tula Chico burned. The fire was thorough enough to bake the plaster off the walls, and the site was abandoned. Within a generation, a new ceremonial centre was laid out 1.5 km to the south, on the limestone promontory above the river: Tula Grande. The move is the founding act of the Toltec state, and it is almost the only event in their political history that the archaeology dates with confidence.

What was built there over the next century was on a different scale from anything else in central Mexico. The city at its height covered about 16 square kilometres according to Dan Healan's synthesis, though other estimates put it nearer 14. Its population is usually given as 30,000–40,000, with a high estimate of 60,000; the surrounding hinterland, an area of some 1,000 square kilometres, held another 20,000–25,000 people. Population figures for pre-modern cities are treacherous, since they depend on how you count unexcavated apartment compounds, and at Tula the residential zones have been sampled rather than mapped. What is not in doubt is the comparison: in the tenth century no other city in Mesoamerica was close.

The urban plan was a grid, reoriented more than once, with the ceremonial core on the promontory and dense residential compounds spreading across the plain below and up the surrounding hills. Excavation at more than 22 localities has shown that these compounds were built of adobe and perishable materials, grouped around courtyards with a single altar, and walled off from the street — the same cellular arrangement found at Teotihuacan, scaled down. Between the compounds lay the marsh of El Salitre, canals and irrigated fields, and the workshops that made the city rich.

Who built it is the question the chronicles answer and the spades cannot. The traditional account, repeated by Sahagún and the Nahua annals, has two peoples arriving at Tollan: the Nonoalca, civilised immigrants from the Gulf coast who brought the cult of Quetzalcoatl and the arts, and the Tolteca-Chichimeca, hard men from the northern deserts who brought the cult of Tezcatlipoca and the warband. Their union produced the Toltec state; their quarrel destroyed it. Modern historians treat this as a charter myth of the kind Postclassic elites told about themselves everywhere, but the archaeology is at least consistent with a mixed population. The ceramics of the Tollan phase show influences from the west and north, from the Gulf coast, and from the Maya area, and the burials include people with modified skulls and dental inlays of a kind more common in the south-east. Tula was a new city in a hurry, and new cities in Mesoamerica were built by whoever turned up.

The economy that supported it was less glamorous than the legend. Tula sat on a semi-arid plain with three reliable streams; its farmers grew irrigated maize, beans, amaranth, squash and chilli, kept turkeys and dogs, and cultivated maguey on the dry slopes for fibre and for pulque. The estimated area under irrigated maize — somewhere between 3,000 and 10,000 hectares — may not have been enough to feed the population, which implies either imports or a figure for the population that is too high. What the valley did have in abundance was limestone for lime plaster, alabaster, and above all obsidian.

What did the sacred precinct look like?

The ceremonial core of Tula Grande is compact and, by Mesoamerican standards, oddly modest in bulk: the two main pyramids would fit inside the Pyramid of the Moon at Teotihuacan with room to spare. Their effect came from position and sculpture. The whole complex sits on a natural acropolis with steep drops on three sides, so that the terraces and platforms read from below as a single artificial mountain.

Pyramid B — the letter is a modern archaeological label — is the building everyone photographs. It is a five-tiered platform about 10 m high, with a stair ascending its south face, and it was once topped by a temple whose roof was carried on four colossal basalt warriors — the Atlantes — standing a little over 4 m tall. Jorge Acosta found them in 1940, face down and broken. Each figure is dressed for war: a butterfly-shaped pectoral across the chest, a sun-disc on the back, a drum headdress, sandals, and an atlatl or spear-thrower in the right hand. They are not gods and they are not portraits; they are porters, carved as warriors because in this architecture the soldier holds up the sky. The near-identical repetition of the four suggests a workshop capable of producing sculpture to a pattern — an important point about Toltec art, which is more corporate than individual.

Pyramid C, on the north-east side of the plaza, was the larger and, on Robert Cobean's reading, the more important building: the axis mundi of the sacred precinct, aligned with the main ballcourt and with the central stair of Pyramid B. It has five sloping bodies, like the Pyramid of the Moon at Teotihuacan, and Cobean interprets that resemblance as a deliberate act of antiquarianism — the Toltecs quoting the abandoned city whose prestige they had inherited.

Around the plaza run the halls that make Tula unusual. Long colonnaded rooms, open to the square, with benches running along their walls for a cumulative length of more than a kilometre, carved with processions of warriors and lords. The architecture marks a change in how Mesoamerican power was staged: from rituals performed for a few inside dark shrines to ceremonies performed for a crowd in an open square. It is the same shift visible at Chichén Itzá, and it is one reason the two cities look like cousins.

Behind Pyramid B stands the coatepantli, the "serpent wall": an L-shaped structure some 40 m long, carrying three registers of relief. The central register shows rattlesnakes devouring semi-flayed human figures — the souls of dead warriors, in the usual reading — while the upper and lower registers carry stepped fretwork and, along the top, shells cut transversely and associated with Venus. The coatepantli marked the boundary of sacred space; the Aztecs built a version of it around the Templo Mayor in Tenochtitlan two centuries later.

The precinct also had a tzompantli, or skull rack, two ballcourts, and a series of altars and benches, one of them carved with a procession of seated lords that is the closest thing Tula has to a historical document. And it had chacmools: the reclining stone figures with a bowl on the belly, which in Mesoamerican ritual received the hearts or blood of sacrificed captives. Tula's chacmools are among the earliest known, though not certainly the earliest, and their presence in the same relationship to a temple doorway at Tula and at Chichén Itzá is one of the tightest parallels between the two sites.

What the precinct does not have is writing. A handful of glyphs, a few calendrical signs, no long text, no royal name that can be read with confidence. For a city of this size at this date that is striking, and it is the single biggest reason the Toltec political order remains opaque. Either the inscriptions were on perishable material, or Tula's rulers chose not to monumentalise their own names — a choice that would fit a state whose legitimacy rested on office and cult rather than on a dynasty.

Who was Topiltzin Ce Ácatl Quetzalcoatl?

Every account of the Toltecs sooner or later arrives at a man called Ce Ácatl Topiltzin Quetzalcoatl, and every account then has to explain how much of him is a man. The name is a stack of titles. Ce Ácatl, "One Reed", is a day-name from the 260-day ritual calendar, of the kind Mesoamerican lords and priests took as a personal name. Topiltzin, "our prince" or "our revered child", is an honorific. Quetzalcoatl, "feathered serpent", is the name of a god and also the title of the priest who served him at Tollan. Whether the three belong to one person, to two, or to a god and his high priest conflated over centuries is not settled and probably never will be.

The story is best told in the Anales de Cuauhtitlán, a Nahuatl chronicle compiled around 1570 and preserved in the Codex Chimalpopoca, with variants in the Leyenda de los Soles, the Historia Tolteca-Chichimeca, Sahagún's Florentine Codex and the histories of Alva Ixtlilxóchitl and Chimalpahin. Ce Ácatl is the son of Mixcoatl, the cloud serpent, and Chimalma; he is born in the year One Reed and raised at Tollan, where he becomes a priest-king of austere habits, a reformer who forbids sacrifice and teaches the arts. Under his rule the city is a place of jade and quetzal feathers, where maize grows in colours and cotton comes already dyed.

Then the sorcerers arrive. Tezcatlipoca, the smoking mirror, and his companion Yaotl set out to destroy him. They come as old men, as a seller of green chilli, as a ballplayer, and each disguise strips away part of his authority. Tezcatlipoca shows him a mirror and he sees himself, aged and ugly, and is unmanned by the sight. He is given pulque, gets drunk, and in his drunkenness summons his sister Quetzalpetlatl; the next morning the two of them are found asleep, and the shame of it destroys his priesthood. He leaves Tollan, or is driven out, and travels east and south with a band of followers, performing miracles and leaving place names behind him at every stop. At the shore he builds a pyre and burns himself, or sails away on a raft of serpents, depending on the version; his heart rises as the morning star. In some accounts he promises to return.

The dating of all this is a trap, and it is worth explaining why. A year-name like One Reed recurs every 52 years, so "Ce Ácatl was born in One Reed and left Tollan in One Reed" is consistent with a dozen different absolute dates. The Anales correlates its years with the Christian calendar by anchoring One Reed to 1519, the year Cortés reached Mexico, which produces a birth in AD 947 and a departure in 999 in some readings, and 895/987 in others. The article by which the Anales chronology is usually reconstructed places Ce Ácatl's arrival at Tula in AD 873 and his death or departure in AD 895, and then has the Toltec capital fall 169 years later, in 1064. Archaeologists place the fall at about 1150. The chronicle and the trench are out by nearly a century, and the discrepancy cannot be repaired by adjusting a single number, because the same 52-year ambiguity infects every date in the sequence.

The most economical reading is that there was a real priest-ruler at early Tula, perhaps in the late ninth or tenth century, whose memory became the vehicle for a much older and wider myth about the feathered serpent, the fall of the proud city and the promise of return. Nothing at Tula names him. The strongest circumstantial evidence is a rock carving of the glyph One Reed on the Cerro de la Malinche near the site, and the fact that the later Nahua sources are unanimous that such a ruler existed and that his city was Tula.

One thing can be said with confidence, because it has been dismantled thoroughly over the last thirty years: Quetzalcoatl was not a white, bearded god who was mistaken for Cortés, and the story that Moctezuma surrendered because he believed the Spaniards were the returning lord is a Spanish construction, first given wide circulation by Francisco López de Gómara in 1552 and repeated by generations of historians who had read him. The Nahua accounts say nothing about a returning god arriving in 1519. Camilla Townsend's 2003 essay "Burying the White Gods" and Matthew Restall's Seven Myths of the Spanish Conquest (2003) trace how the story grew, and why it was useful: it made the conquest look like prophecy rather than politics, and it flattered the conquistadors.

What did the Toltecs trade, and who did they trade with?

Tula's wealth rested on stone. The Sierra de las Navajas, the "mountains of knives", lies about 20 km east of the city and is the largest and finest source of green obsidian in Mesoamerica — a glassy volcanic rock that fractures into blades sharper than steel. A second major source, at Ucareo-Zinapécuaro in Michoacán, supplied a grey-to-black obsidian. Control of the first and access to the second made Tula the natural successor to Teotihuacan as the distributor of sharp edges to central Mexico.

The scale of the industry is measurable. Healan's surveys recovered roughly 16,000 pieces of obsidian from the urban zone of Tula and more than 25,000 from the surrounding residential areas, and excavations at the Cruz locality uncovered a workshop that produced prismatic blades in quantity — long parallel-sided blades struck from a prepared polyhedral core, the standard cutting tool of Postclassic Mesoamerica. The waste products tell the story as clearly as the products: quantities of exhausted cores and angular debitage, the signature of production for export rather than for local use. On one estimate a very large share of the city's households were involved in some form of craft production, whether obsidian, travertine, alabaster or pottery.

The pottery is where the archaeology has had to correct itself, and the correction is instructive. The type called Mazapa Red on Brown — bowls painted inside with parallel wavy red lines, first defined by Sigvald Linné in 1934 from post-Teotihuacan levels at Teotihuacan itself — was long treated as the diagnostic of Tula's apogee. Jorge Acosta's 1940 test pits at Tula Grande produced levels in which it made up as much as 52 per cent of the non-olla sherds, and a generation of chronologies was built on that. Healan's reanalysis shows the sample was misleading: at Tula, Mazapa Red on Brown belongs to the Terminal Corral and Early Tollan phases — the beginning of the Tollan phase — and declines well before the city's Late Tollan apogee. It is a horizon marker for the beginning of the Toltec period, not its height — and the confusion between the pottery type, the ceramic complex and the period has muddied the chronology for eighty years.

Mazapa ware travelled widely, which is part of why it was over-read. It turns up from northern Mexico into Central America, and along with Tohil Plumbate — a hard, grey-green, almost glazed-looking ware made on the Soconusco coast of Chiapas and Guatemala — and the fine orange ceramics called Silho or X-Fine Orange, it defines what Richard Diehl called the Toltec horizon. Tula imported the plumbate and the fine orange; it exported obsidian, and probably salt, cotton, feathers and lime. The INAH guide to the site adds turquoise from the north, arriving along a route that ran towards what is now New Mexico, and the presence of marine shell and Pacific goods in the city shows contacts in both directions.

How far this amounted to an empire is another matter. Michael Smith and Lisa Montiel compared Tula's archaeological footprint with that of Teotihuacan and Tenochtitlan and concluded that its influence abroad was negligible by comparison: a large, complex capital with a substantial hinterland, but not the centre of an imperial system. Against that, the distribution of Pachuca green obsidian in the tenth and eleventh centuries is very wide indeed, and the same two obsidian sources supplied both Tula and Chichén Itzá — a fact that says something about the reach of the exchange network even if it says nothing about who ruled whom.

Why does Tula look like Chichén Itzá?

The resemblance is not subtle. Both cities have a Temple of the Warriors fronted by rows of carved columns; both have a Castillo-like pyramid with serpent columns at the doorway; both have chacmools at the top of the main stair; both have tzompantli, colonnaded halls around a great plaza, and atlantean warrior figures used as supports. Désiré Charnay noticed the parallel in the 1880s and wrote that the serpent columns of the Castillo were "almost an exact reproduction" of a column he had dug up at Tula.

The explanation offered for most of the twentieth century was conquest. Toltec armies, or Toltec-led merchants, were supposed to have gone east in the tenth century, taken Chichén Itzá, and installed a dynasty whose founder took the name Kukulcán — the Yucatec Maya for feathered serpent. Alfred Tozzer proposed three successive waves of Mexican intrusion; Sylvanus Morley preferred a migration in which the Itza absorbed a band of Toltec exiles and came home with their crafts; Eric Thompson made the Itza the senior partner in a commercial empire with a Toltec figurehead. All three relied on the same assumptions: that the Mexican-style buildings at Chichén were later than the Maya ones, and that Tula was the older city.

The dating did not hold. Chichén Itzá's monumental core was laid out between about AD 750 and 900, its great expansion ran from 900 to 1050, and its earliest hieroglyphic date is AD 832 while its last is 998. The city was in decline by 1100, before Mayapán rose. Tula's Tollan phase begins around AD 900. Once the ceramic chronologies of the two sites were put on a proper footing — work in which Robert Cobean and James Bey were central — the Mexican-style buildings at Chichén turned out to be contemporary with, and in some cases earlier than, their supposed models at Tula. George Kubler had already questioned the assumed priority of Tula in the 1960s; the ceramic and radiocarbon evidence then made the point in the ground.

That has not produced agreement on what did happen, only agreement on what did not. The current literature, and especially the Dumbarton Oaks volume Twin Tollans (Kowalski and Kristan-Graham, 2007), treats the two cities as participants in a shared Early Postclassic world system rather than as parent and child. Jeff Kowalski's summary is that the relationship reflects the two cities' mutual participation in wider trade networks, through which both goods and the ideological props of elite power circulated. Cynthia Kristan-Graham, analysing the colonnaded halls and sunken plazas, argues that the architectural vocabulary came from the Chichimec north and from several regions at once, and that both cities assembled it for their own purposes.

What would that contact have looked like in practice? Not an army. More likely a combination of elite marriage alliances, mercenary service, pilgrimage, and the movement of craftsmen and merchants along the Gulf coast, where El Tajín and the Totonac towns had been mixing central Mexican and Maya idioms since the Epiclassic. The feathered serpent was already a shared Mesoamerican figure; the warrior column, the skull rack and the reclining chacmool were portable symbols of a new kind of militarised kingship that appealed to rulers from Yucatán to Hidalgo. Tula and Chichén Itzá did not have to conquer each other to end up looking alike. They had to be doing the same thing at the same time, with the same borrowed kit, in the same market.

How did Tula fall?

The end is visible in the ground and legible in the chronicles, and the two do not quite agree. At Tula, the temple on Pyramid B was burned; the Palacio Quemado — the "Burnt Palace", so named by Acosta when he found abundant charcoal and the collapsed remains of burnt roofs on its floors in the 1940s — was destroyed by fire; columns and sculptures were broken and in some cases deliberately buried. Residential occupation thins out through the twelfth century, and by about 1150 the city was largely empty. The last datable construction belongs to the Late Tollan phase, and nothing follows it but squatting, looting and, later, a small Aztec-period settlement among the ruins.

The date of the catastrophe is given variously as 1150, 1168, 1179 and 1200. The Nahua annals are precise and incompatible: the Anales de Cuauhtitlán sequence puts the fall in a One Reed year that can be read as 1168 or 1179, while the reconstruction that anchors Ce Ácatl to 895 puts it at 1064. Archaeology prefers "the second half of the twelfth century", which is honest but unsatisfying.

The causes are a list, and the honest answer is that several of them were probably operating at once. The first is climate. Tree-ring reconstructions for the American south-west show a severe and prolonged drought in 1149–1167, within the lifetime of the city's last occupants; a drying trend in central Mexico in the twelfth century is visible in several lake and speleothem records. A city of 40,000 people on a semi-arid plain, dependent on irrigation and on maize surpluses from its hinterland, is exactly the kind of society a decade and a half of failed rains can break. The second is the loss of that hinterland. Tula's power rested on a network of subordinate centres and on the flow of obsidian and tribute; if the network fragmented, the city's food supply went with it.

The third is war, or the threat of it. The chronicles describe the arrival of Chichimec groups from the north, and there were fortified hilltop sites in the region, at Las Ranas and Toluquilla, though whether they answered to Tula is unknown. Whether the Chichimeca were external enemies or the same northern peoples who had been settling in the region for centuries is unclear; "Chichimec" in Nahua usage is less an ethnic label than a term for people who had not yet adopted urban ways, and it could be applied to neighbours as easily as to invaders. The fourth is internal. The division between the Nonoalca and the Tolteca-Chichimeca, between the followers of Quetzalcoatl and the followers of Tezcatlipoca, is the chronicles' own explanation, and it is at least consistent with the archaeological evidence of a violent destruction concentrated on the ceremonial centre — the same pattern Manzanilla reads at Teotihuacan.

The last ruler in the annals is Huémac, a name that may be a title. His story is the story of the city's end in miniature: he loses a ballgame wager to the rain gods, is forced to give up maize and precious stones, watches the city come apart, and dies in a cave at Cincalco near Chapultepec, either vanishing or hanging himself, depending on the version. The Toltecs disperse — to Colhuacan, to Cholula, to the Gulf coast, to Yucatán in some accounts — and the political landscape of central Mexico is rebuilt by their heirs and their refugees.

It is worth noticing that Tula did not fall alone. Chichén Itzá declined by 1100; the Classic Maya collapse was two centuries past; the whole of Mesoamerica was reorganising around smaller, more militarised, more commercially minded states. The twelfth century was a systemic rupture, not a local accident, and the Toltec were one of its casualties.

What did the Aztecs make of the Toltecs?

Nearly two centuries after Tula emptied, the Mexica arrived in the Basin of Mexico and set about acquiring a past. What they acquired was the Toltecs, and they acquired them so thoroughly that for four centuries historians took the Aztec account at face value.

The vocabulary shows how deep it went. Toltecatl meant an artisan, a skilled worker, a maker of fine things; toltecayotl meant the whole complex of arts, letters, urban life and right conduct that separated civilised people from chichimeca, the hunter-gatherers of the north. Sahagún's Book 10, the book of the people, introduces the Tolteca as the first settlers of the land, wise and learned, and says their name came from their manner of life and their works. The same book treats a "Toltec house" as a byword for excellence, and the Nahuatl texts Sahagún collected call the tolteca mirrors pierced on both sides, people who give others a face and a heart, teachers of teachers. To be Toltec, in this idiom, was not to belong to an ethnicity; it was to be accomplished.

The political use of that idiom was immediate. The Mexica traced their royal line to Colhuacan, the city that had inherited Tula's prestige after the fall, and married into its ruling house; their first king, Acamapichtli, was presented as the fruit of that alliance. Other Postclassic dynasties did the same — the K'iche' of highland Guatemala, the Itza of Petén, the rulers of Cholula and Texcoco all claimed Tollan descent or Tollan legitimacy. Alfredo López Austin gave this shared ideology a name, zuyuanidad, from the Maya word zuyuá, a term for Tollan: the belief system in which Postclassic elites claimed descent from Quetzalcoatl and inheritance of the mythical Tollan, and in which the right to rule was a matter of descent from a place rather than possession of a territory.

The material copying was equally deliberate. The Templo Mayor of Tenochtitlan had a coatepantli, chacmools, warrior reliefs and skull racks, all of them quotations of Tula. The Mexica even collected Toltec antiquities: excavation at Tula has produced Aztec-period offerings buried among the ruins, and some Toltec-style objects in the Templo Mayor deposits may have been removed from the older city.

Then there is the matter of the histories. Nahua tradition, as recorded by Sahagún's informants and by later chroniclers, held that Itzcoatl, the first ruler of the Triple Alliance, ordered the old painted books burned, on the argument that it did not become the people for all of them to be known, so that the Mexica's own account of their rise would stand alone. The story is reported at second hand and its scope is disputed — it may refer to a specific set of documents rather than a general purge — but the effect is not in doubt: the history that survives is the history the victors commissioned, and it places the Toltecs at the beginning of everything.

The irony is that the Toltec golden age the Aztecs described has almost no archaeological support. There is no evidence at Tula of the palaces roofed with jade, the cotton that grew in colour, or the universal abundance. There is a large, hard-working, defensible city with a big obsidian industry, a militarised cult, and a plaza built to hold a crowd. That is a considerable achievement. It is not the city of the legends, and the gap between the two is one of the most revealing things about how states invent their ancestors.

Was there ever a Toltec empire?

The word "empire" entered the story early and has been hard to dislodge. Nineteenth-century writers, following the chronicles and the visible parallels with Chichén Itzá, made the Toltecs a race of civilisers who had spread across Mesoamerica before the Aztecs, and the "Toltec period" became a standard slice of Mexican chronology. The identification of Tula as the Toltec capital was settled — as far as it ever is — at a round table of Mexican and foreign scholars in 1941, and confirmed by Acosta's excavations in the 1940s, which produced the Atlantes and the burnt palace. From then until the 1980s the Toltec empire was a fixture of textbooks.

The revision came from three directions. The first was chronological: once Tula's ceramics were properly sequenced, the city's apogee contracted to the tenth and early eleventh centuries, and the long Toltec dominion of the old accounts became impossible. The second was comparative: Smith and Montiel's measurement of Tula's provincial footprint found a capital with an impressive core and very little behind it. The third was art-historical: the "Toltec style" turned out to be a shared Mesoamerican vocabulary of the Early Postclassic rather than a fingerprint of one ethnic group.

Against that, Richard Diehl's notion of a Toltec horizon retains some force. There really is a package of traits — Tohil Plumbate, Silho fine orange, Mixteca-Puebla iconography, warrior columns, chacmools, skull racks — that appears across a huge area in the tenth and eleventh centuries, and Tula is as good a centre for it as any. Whether a horizon is an empire is a different question. A horizon can be produced by trade, by elite emulation, by pilgrimage, or by the simple fact that a successful visual language gets copied.

The other unresolved problem is Tollan itself. In Nahuatl the word means "place of reeds", and it was applied to any great city: Teotihuacan was Tollan, Cholula was Tollan Cholollan, Tenochtitlan was Tollan in its official rhetoric. Enrique Florescano argued that the mythical Tollan of the chronicles is really Teotihuacan, remembered across five centuries; López Austin argued that Tollan is a product of Mesoamerican religious imagination, a primordial city that never existed anywhere. If either is right, then the search for the historical Tollan is a category error, and Tula is simply the city that came closest to fitting a pre-existing template — which is not the same as being the template.

The most defensible position is the boring one. Tula was the largest and most powerful state in central Mexico between about AD 900 and 1100. Its influence was real, wide and thin: its pottery, its obsidian and its artistic vocabulary reached from the Bajío to Yucatán, but it did not govern most of the territory it influenced. It was a regional state with hegemonic ambitions, in a century when that was the most a state could be. Calling it an empire inflates it; calling it a city-state understates it; and the 6,000 square kilometres the map draws for it is a figure from the border dataset, not from any historical source.

What is left at Tula, and what is still argued?

Tula de Allende is a working town an hour and a half north of Mexico City, with a railway, an oil refinery and a tourist economy built on the ruins. The archaeological zone was declared a national park in 1981 and covers about a square kilometre — roughly the ceremonial core. The INAH estimate is that the excavated and protected area represents about 12 per cent of the ancient city. The rest lies under fields, houses, the highway and the refinery.

Visitors see Tula Grande: the plaza, Pyramid B with its Atlantes and its serpent columns, Pyramid C, the Palacio Quemado with its reliefs of processions and its Tlaloc benches, the coatepantli, the ballcourts, the tzompantli platform. The site museum, named for Acosta, holds sculpture, ceramics and human remains. Tula Chico, the earlier centre, is a grass-grown mound group a kilometre and a half to the north, unexcavated for the most part and easy to miss.

What has been lost is considerable. The city was looted in antiquity, again under the Aztecs, and again in the colonial period, when its cut stone was taken for churches and its sculptures for collectors. The Atlantes themselves were found face down and in pieces. There is no detailed archaeological map of the whole city, a gap Healan and others have repeatedly flagged, and large areas of the residential zone have never been systematically sampled.

The open questions are the ones this entry has been circling. How many people lived at Tula, and how were they fed? Population estimates still vary by a factor of two, and the carrying capacity of the valley is contested. How was the state organised — one king, a council of lineages, a confederation of the two peoples the chronicles describe? The absence of royal inscriptions is a real obstacle here, not merely a gap to be filled by analogy. What was the relationship with Chichén Itzá, and was the transmission of forms eastward, westward, or neither? When exactly did the city fall, and was the burning the cause of abandonment or a consequence of it? And who were the Toltecs, in the sense of an ethnic or linguistic identity, as opposed to a political and cultural one?

None of these is likely to be settled by a single discovery. What is changing is the toolkit: Bayesian modelling of radiocarbon sequences, isotopic analysis of human teeth to trace migration, petrography to source ceramics, and remote sensing of the unexcavated urban zone. A generation ago the Toltec were a paragraph in the textbooks between Teotihuacan and the Aztecs. They are now a test case for how Mesoamerican states worked in the century after the Classic collapse — and for how much of what we think we know about the deep past arrives through the filter of the people who came after.

Common myths about the Toltec

Timeline of the Toltec

What the map shows next

At AD 900 on the map the map shows a single large territory around Tula in a central Mexico still crowded with small polities — the Epiclassic order only just giving way. That is the moment the Tollan phase begins: Tula Grande laid out, Tula Chico in ruins behind it, and the obsidian workshops starting to supply the Basin.

At 1100 on the map the picture is at its height, with Tula's influence reaching across the Basin of Mexico and north towards the Bajío. At 1180 on the map the territory is still drawn, but the city it represents has burned and emptied; within a generation the political weight has shifted to Colhuacan, Cholula and the new Chichimec lordships from which Tenochtitlan will eventually emerge. The next entries on the map — the later Maya city-states, Tenochtitlan and the Aztec Triple Alliance — are the world the Toltecs' memory helped to build.