The great houses of Chaco Canyon

The canyon runs roughly north-west to south-east through sandstone mesas in the centre of the San Juan Basin. The park covers about 137 square kilometres and holds the core great houses (Pueblo Bonito, Chetro Ketl, Pueblo del Arroyo, Pueblo Alto, Peñasco Blanco, Una Vida and others), hundreds of smaller house sites, and Fajada Butte at the canyon's south-east end⁠[2]. Pueblo Bonito covers about 3 acres, with a plan divided by a north-south wall, a great kiva on each side and more than thirty smaller kivas⁠[3].

Six road systems leave the canyon and connect it with about thirty outliers. They are wide, straight, engineered surfaces with beds about 9 metres across; the Great North Road runs straight from Pueblo Alto about 56 km north⁠[2]. They were traditionally mapped from aerial photographs and ground survey, and many that were plain in the 1980s are now nearly invisible on the ground, which is why lidar matters⁠[13].

The park has been federal land since 1907, a national historical park since 1980 and a World Heritage Site since 1987⁠[1]. In June 2023 the Interior Department withdrew federal minerals within 10 miles of the park from leasing for 20 years; in July 2026 a draft environmental assessment proposed revoking that withdrawal, and as of late August 2026 the Department was reviewing comments⁠[14]. This is context for the landscape, not part of the archaeological question.

The building sequence is fixed by tree rings: the earliest great-house rooms at Pueblo Bonito in the mid-800s, the main expansions at 1047 to 1049 and 1077 to 1082, and the last construction in the early 1100s, with many episodes falling in wet decades when food could be stored⁠[4]. Dated wood from the canyon and from Aztec Ruins is compiled in the Chaco Research Archive's tree-ring database, which the Laboratory of Tree-Ring Research has expanded from its specimen cards⁠[24].

The timber was imported. Spruce and fir beams carry strontium ratios that match the Chuska and San Mateo (Mount Taylor) mountains and not the equally close San Pedro Mountains; logs from both sources appear in the same room and year by 974, which points to stockpiling and cooperation⁠[5]. Ring-width matching of 170 timbers put 70 percent in the Chuska and Zuni mountains, with the Zunis supplying nearly all before 1020 and the Chuskas dominant by 1060⁠[6]. Isotopes have also sourced ponderosa pine, the commonest beam⁠[25].

Inequality appears early. New radiocarbon dates on Room 33 put marked social differentiation 150 to 200 years earlier than once thought⁠[17], and the macaws were arriving by the tenth century⁠[18]. In 2017 Kennett and colleagues reported that nine of the Room 33 burials share one mitochondrial genome and that nuclear DNA shows mother-daughter and grandmother-grandson links, which they read as an elite matriline lasting about 330 years⁠[8]. Indigenous scientists objected that no tribes had been consulted, that the paper described ancestors insensitively, and that NAGPRA leaves research on such remains unregulated⁠[26]. In 2025 a study initiated and controlled by Picuris Pueblo found genetic continuity between Picuris people, ancient and living, and the Pueblo Bonito burials⁠[27].

The local environment was poor and grew poorer. Pollen and plant remains show the canyon's pinyon-juniper woodland cut back from the last centuries BC and largely gone by the Bonito phase⁠[7]. Benson and Grimstead calculated that the canyon floor could grow maize for about 1,000 people at most and that small game could not supply the protein for a population of 2,300, so either food was imported or few lived there year-round⁠[10]⁠[28]⁠[29]. Great-house building ended in the early 1100s, and the system came apart from about 1140, during a drought that began about 1130⁠[4]⁠[2].

What Chaco was. A ceremonial or pilgrimage centre: great houses with few permanent residents, filled at gatherings and supplied from outside; it predicts small resident households, large stores and episodic deposits of food and goods brought in⁠[7]⁠[30]. A hereditary elite: the Room 33 matriline, the early macaws and the cacao point to a ruling house; this predicts continuity of privileged burial and control of imports across generations⁠[8]⁠[17]. A lived-in town: Lentz and colleagues argue the canyon was farmed and lived in for a millennium⁠[7]. A regional polity: Kohler, Brumbaugh and Bocinsky argue that the regional system held more people than any earlier estimate and that its rapid expansion after 1040 fits models of colonial systems, with roads and shrines as a polity's infrastructure⁠[9]. These are not exclusive; they differ on the size of the resident population and on who controlled the imports.

How many lived there. Estimates for the canyon have fallen since the 1970s; the NPS notes that research then sharply reduced earlier figures for the great houses⁠[1]. Benson's carrying-capacity work puts the self-supporting population under 1,000⁠[10]. At the regional scale the 2025 estimate goes the other way⁠[9]. The two can both be right if the canyon was a small, provisioned centre of a populous region.

What the roads were for. They moved timber and probably food: the beams came from 75 to 100 km away, and Benson's import calculation implies many thousands of foot trips⁠[5]⁠[28]. But the roads are wider and straighter than foot traffic needed, some run in parallel pairs, and modern Pueblo people associate the Great North Road with their ancestors' arrival from the north⁠[2]⁠[22]. Sofaer and others read the roads as cosmographic, laid out to directions and sky events, and the Gasco roads' solstice alignment is the strongest recent case⁠[31]⁠[22]. Kohler's reading treats them as the visible network of a polity⁠[9].

Whether the buildings encode astronomy. On Fajada Butte three slabs cast midday light on two spirals: one dagger of light crosses the centre of the large spiral at the summer solstice, two bracket it at the winter solstice, and one bisects the small spiral at the equinoxes⁠[11]. Whether people set the slabs is disputed; Newman, Mark and Vivian judged them a natural rockfall that the carvers put to use⁠[32]. Zeilik, and later Schaefer and Stamm, read the site as a sun shrine for the solstices rather than a calendar, and the lunar standstill markings Sofaer's team reported in 1982 as chance⁠[12]⁠[33]⁠[34]. In 1989 the middle slab was found to have pivoted, and the markings no longer form as they were recorded; the photographs of 1978 to 1987 and a model built from 1984 photogrammetry and a 2005 laser scan are the record⁠[35]⁠[36]. For the buildings, the Solstice Project's survey put Pueblo Bonito's main walls within a quarter of a degree of the cardinal directions and twelve of fourteen major buildings within a few degrees of solar or lunar azimuths, and found lines between buildings on the same bearings⁠[37]. Malville and Munro argue that most orientations follow an older tradition of south-facing house plans, and that the alignments between buildings to the lunar standstills are products of the topography⁠[38]. McCluskey's point that the noon shadow events at Fajada's other sites are about what chance gives is the general objection to alignment counts made without a null model⁠[39].

Why building stopped and people left. Drought: a drought of about fifty years began about 1130, as building ended, and Benson and colleagues have tested the twelfth-century migrations against tree-ring drought records⁠[2]⁠[40]. Degradation: centuries of woodcutting and erosion left the canyon unable to absorb a drought⁠[7]. Movement rather than collapse: descendant communities describe a migration, not an end, and the Picuris genomes show continuity with Pueblo Bonito⁠[1]⁠[27].

The population question can be tested with the data that exist: a room-by-room count of hearths, storage and residential features for every dated great-house room in the tree-ring database, set against the food budget, would bound the resident population⁠[24]⁠[10]. Isotope sourcing of the maize and fauna, as has been done for timber, would show how much food was imported and from where⁠[28]⁠[5].

The roads need a complete lidar map of the basin from public elevation data, with each segment's width, orientation and endpoints recorded, and a test of the alignment claims against random orientations under the same terrain⁠[41]⁠[22]. The astronomical readings need the same null-model test for great-house walls and shrine lines, decided before the data are examined, starting from the Solstice Project's published wall azimuths⁠[42]⁠[37]⁠[12]. Any further genetic work on the ancestors buried at Chaco should follow the Picuris model of tribal initiation and control⁠[27]⁠[26].