Upper Xingu Earthworks (Kuhikugu)

← Encyclopedia

Stand at the lip of one of the great ditches at the Xingu headwaters and the line of it runs away through the trees with the patience of something poured rather than dug: a curving trench two to three meters deep and more than ten across, its inner edge once banked and topped with a palisade of close-set logs. Follow it and it closes a ring. Inside the ring lay a town. Out from the town, broad and dead-straight, ran roads — some of them forty meters wide, wide enough to lose a modern avenue in — bridged where they crossed the wet ground and carried on causeways where the land sank, aiming across kilometers of forest at the next plaza, and the next. The forest that fills the rings today is not the forest that was here. It grew back over a built world.

That built world is Kuhikugu and its neighbors: a cluster of fortified pre-Columbian settlements in the upper basin of the Xingu River, in what is now the northern part of Mato Grosso state, Brazil. The towns belong to the headwater country where the southern margin of the Amazon forest gives way to the seasonally flooded gallery and the cerrado savanna — a transitional, edge-of-the-forest landscape, not the deep terra firme of the central Amazon. The people who built it were the ancestors of the Kuikuro, who live in the same country today, inside the Xingu Indigenous Park. The Kuikuro never lost the place. They guided the archaeology to it.

The form first: a town meant to be walked

The unit of this society is legible on the ground. At the heart of a settlement is a plaza — open, roughly circular, swept clean — the civic and ceremonial center around which houses ranged in an arc. From the plaza the roads radiate, and they are not paths worn by feet but engineered roadbeds: graded, curbed at the edges with low earthen berms, laid out straight across the terrain regardless of the easy line, some of them oriented on the rising and setting points of the solstice sun. Where a road met a stream it crossed on a bridge or rode a causeway over the floodable ground. Where it met the defensive perimeter it passed through a gap in the ditch-and-palisade ring that girdled the larger towns.

These rings are the most arresting feature. The ditches at the principal sites run two to three meters deep and over ten meters wide, the excavated earth thrown up on the inner face to raise a wall, the wall pierced for posts — the footprint of a timber palisade now read off the postholes. A town of this kind covered on the order of forty-five to sixty hectares, the largest reaching the scale of a hundred and fifty acres. This is not a village fence. It is fortification at urban scale, and it implies what fortification always implies: neighbors worth defending against, a population worth the labor of the wall, and a authority able to command that labor.

The galactic cluster

No town stands alone. The diagnostic Upper Xingu pattern, mapped across the headwater country by Michael Heckenberger and his Kuikuro and Brazilian collaborators, is a cluster: a single large central settlement — an exemplary ceremonial center — ringed by four major satellite towns set roughly to the cardinal directions, with smaller plaza-hamlets filling the spaces between, the whole tied together by the road network. Heckenberger named the geometry a “galactic” form of prehistoric urbanism: a bright nodal center with satellites orbiting it, the cluster functioning as a small independent polity, and the clusters in turn nested within a regional system of peer polities that shared a single material and ritual culture. Around the Ipatse and Kuhikugu nodes the surveyed network comprises well over a dozen interconnected sites; estimates for the regional population it supported at its height run into the tens of thousands — figures on the order of thirty to fifty thousand people across the integrated landscape.

Such works do not build themselves. A ditch ten meters across and kilometers long, a graded road forty meters wide, a palisade of thousands of felled and set logs — each is a measure of organized labor sustained over time, and so a measure of the social order that mustered it. The Upper Xingu polities were ranked societies led by hereditary chiefs, the plaza the stage of a public authority, the satellite towns gathered to the ceremonial center in a hierarchy legible in the geometry itself. The peer-polity arrangement — many such clusters of comparable size and culture, neither one swallowing the rest nor any single capital ruling all — is the form of a landscape densely settled by competitive equals, which is exactly the condition the encircling ditch-and-palisade answers. The defensive ring is the negative image of the regional system: a country crowded enough, and contested enough, that towns walled themselves against one another.

The point of the word urbanism is contested, and worth holding precisely. These were not the dense masonry cities of Teotihuacan on the Mexican plateau or the stone-and-adobe metropolis of the central highlands; nor the earthen monumentality of Cahokia on the Mississippi floodplain; nor the desert ground-drawings of Nazca on the Peruvian coast. The Xingu towns were low-density and green — settlements embedded in, and inseparable from, a worked countryside. The plazas and palisades sat inside a mosaic of manioc fields, fruit orchards, dammed ponds and managed forest. To call it urbanism is to insist that the relevant unit was the whole engineered landscape, not the buildings — that a city can be made of soil, water, and trees as legitimately as of stone.

The garden city as landscape

The countryside between the towns was as constructed as the towns. The staple was bitter manioc — cassava processed to leach its poison — grown in large open fields cut from the forest and, after their fertile years, allowed to return to managed regrowth rather than abandoned to wild succession. The Kuikuro distinguish their soils precisely: ordinary red earth for the manioc, and patches of deep black anthropogenic soil — Amazonian dark earth, terra preta, dense with potsherds and charcoal from centuries of habitation — for the other crops and the orchards, especially the oil-rich piqui. The dark earth is itself an artifact, a fertility manufactured by the long residence of people; the Kuikuro word for it doubles as their word for a cultivated field. Around the settlements the forest carries the fingerprints of cultivation: stands of useful and edible species in concentrations no unmanaged forest would assemble.

Water was worked as deliberately as soil. Stream channels were modified into ponds that held water and fish through the dry season; there were subterranean pits for storing manioc, enclosures the early accounts describe as turtle pens, and weirs set across the streams to take fish — fish weirs of the same kind the Kuikuro still build and use. The whole system reads as aquaculture and arboriculture braided into settlement: a way of living densely in the tropical lowlands that did not require clearing the forest away but rebuilt it into a productive, peopled garden. This is the heart of the revision that the Xingu evidence drove into the wider study of Amazonia — the demonstration, dug and mapped on the ground, that a substantial population shaped this country for centuries, that the forest standing here is in part a human creation, and that the old picture of the Amazon as a thin, fragile wilderness incapable of supporting complex society does not survive contact with the trench and the road.

When it was built, and what broke it

The dense, walled, road-linked towns belong to the late pre-Columbian centuries — the principal settlements rising and reaching their peak across roughly 1200 to 1600, with occupation of the headwater country running deeper still, more than a millennium back. This places the great phase of construction squarely in the medieval and early-modern world: the Xingu galactic clusters were being raised, walled, and linked in the same centuries that saw the medieval European cathedral and the height of the Andean and Mesoamerican states to the west and north.

What broke the system was contact. The Upper Xingu lies far inland, and no European set eyes on these towns at their height; the catastrophe arrived ahead of the colonizers themselves, carried by the epidemic diseases that ran down the river systems of the Americas after 1492 — smallpox, measles, influenza, against which there was no immunity. Across the Americas the indigenous population fell by figures conventionally estimated near ninety percent within a century and a half of contact, and the Xingu was not spared. The walled towns emptied. By the time the first sustained outside surveys reached the Upper Xingu in the twentieth century, the descendant population of the whole region — the peoples who carried the memory and the language of the builders — numbered in the hundreds. The forest came back into the plazas. The depopulation, not any architectural collapse, is why the country read as empty wilderness to outsiders: a built landscape with most of its builders gone, and the trees grown up to hide what remained.

The living heirs and the archaeology done with them

The Kuikuro are a Carib-speaking people, one nation within the multi-ethnic society of the Upper Xingu — a regional system in which Carib, Arawak, and Tupian language communities share a single ceremonial and exchange culture and a long-standing peace, intermarrying and trading across the plazas while keeping their distinct tongues. That society survives. It survives in part because, in 1961, the campaign of the Villas-Bôas brothers established the Xingu Indigenous Park, the first large indigenous reserve demarcated in Brazil and a model for those that followed — a protected territory inside which the headwater peoples have held their land and their lifeways against the advancing agricultural frontier of Mato Grosso.

The archaeology of the earthworks has been done, from the start, with the Kuikuro and not merely on their land. From 1993 onward Michael Heckenberger worked in partnership with the community; Kuikuro knowledge of where the ancestral towns lay — carried in oral memory of the named old places — pointed the survey to the buried ditches and roads, and Kuikuro leaders, among them the chief Afukaka Kuikuro, stand as named co-authors on the landmark papers. The fit between the deep-time settlement geometry and the still-living settlement form is itself a primary source: the modern Kuikuro village, with its swept plaza and arc of houses and radial paths, is the small surviving instance of the same plan that the great clusters realized at scale. The descent claim is not an inference imposed from outside; it is the continuity the people assert and the ground confirms.

For the depth of the record one turns to the published scholarship — almost all of it recent and in copyright, and best pointed to rather than paraphrased. The two papers in Science — the 2003 “Amazonia 1492” study and the 2008 study of pre-Columbian urbanism and anthropogenic landscapes — set out the cluster geometry, the dating, and the landscape evidence; Heckenberger’s book The Ecology of Power (2005) develops the long-term cultural history; and Heckenberger and Eduardo Góes Neves’ review of Amazonian archaeology places the Xingu within the wider overturning of the “pristine forest” model.

Garden-cities and the myth of the lost city

The Upper Xingu has long drawn a different kind of attention, the kind that goes looking for a buried metropolis of gold. It was into this country that the British surveyor Percy Fawcett vanished in 1925 hunting his imagined “City of Z,” and Kuhikugu has since been pressed into service as the supposed reality behind that legend. The connection is best kept loose. The dream of El Dorado — the golden king, the hidden city heavy with treasure — is a European myth that the Americas never contained, and the Xingu towns were not it: no gold, no masonry palace, no king on a throne of metal. What is here is stranger and more durable than the fantasy. It is a working civilization of earth and water and managed forest, the largest known of its kind in the southern Amazon, built to a deliberate cosmological geometry and maintained for centuries.

The Xingu builders belong to a wider pre-Columbian Amazonia whose monuments are finally being read — the mound-and-urn world of Marajó at the river’s mouth, the savanna urbanism of the Casarabe of the Llanos de Mojos, the ceramic chiefdom of the Tapajó at Santarém, the geometric ditched enclosures of the Acre geoglyphs — each a distinct answer to the question of how to live densely in the lowland tropics. And it belongs to a present in which the descendants of those answers still hold the land and the cosmologies: peoples whose worlds differ as sharply from one another as from the builders — the Yanomami of the northern border forests, the Asháninka of the western montane, the Shipibo-Conibo of the Ucayali, the Tukano cosmos of the Vaupés, the Guaraní search for a land without evil.

What the Kuikuro keep is not a ruin. The ditch the visitor stands at is the same shape as the plaza they sweep clean each morning; the road grown over with forest aims at the place where they still set their weirs. The garden-city was never a city that fell and a garden that returned — it was always both at once, a settlement that is also a worked forest, and the people who designed that braid are the people who tend it now, walking the lines their ancestors poured into the ground.

→ Related: Marajoara Culture · Casarabe Culture · Acre Geoglyphs · Santarem Tapajos Culture · Mississippian Cahokia · Nazca · Teotihuacan · Tukano Vaupes Cosmology · Land Without Evil · Yanomami · Ashaninka · Shipibo Conibo