A Diachronic Analysis of Obsidian Use at Chaco Canyon and the Influence of Social Factors on Obsidian Procurement

Abstract

Chaco Canyon in New Mexico was the center of an extensive regional cultural system. The strength of Chaco's regional interactions has been partly defined by the presence of non–local goods including obsidian. We take a diachronic look at Chaco obsidian use from AD 500–1250 using the largest sample of XRF sourced obsidian available to date and combine this with technological analyses to identify significant changes in where and how Chacoans obtained and used obsidian. In the AD 700s obsidian from the more distant Jemez Mountains began to supplant the closer Mt. Taylor obsidian. The obsidian is of roughly equal quality, suggesting this shift relates to social factors and not raw material constraints. We find more variation over time in obsidian source use and acquisition strategies than previously noted. The patterning appears to relate to regional and local cultural diversity, social and organizational heterogeneity, and the development of new exchange networks.

El Cañón del Chaco en Nuevo México fue el centro de un sistema cultural regional extensivo. La fortaleza de las interacciones regionales de Chaco ha sido parcialmente definida por la presencia de bienes no locales incluyendo la obsidiana. Con una perspectiva diacrónica, revisamos el uso de la obsidiana en Chaco desde el 500 al 1200 d.C. empleando la muestra más grande de obsidiana analizada por XRF hasta la fecha. Combinamos estos resultados con análisis tecnológicos para identificar cambios significativos en el dónde y cómo los habitantes de Chaco obtuvieron y usaron la obsidiana. En el siglo VIII, las Montañas Jémez como fuente más alejada empiezan a desplazar la obsidiana más cercana del Cerro Taylor. Los dos tipos de obsidiana son de similar calidad, lo que sugiere que el cambio se relacione con factores sociales y no con restricciones de la materia prima. También observamos mayor variación en el uso de fuentes de obsidiana y estrategias de adquisición a lo largo del tiempo, comparado con lo señalado previamente. Los patrones parecen deberse a factores sociales relacionados con la diversidad cultural regional y local, la heterogeneidad social y organizacional y el desarrollo de nuevas redes de intercambio.

KEYWORDS:

• Chaco Canyon
• chipped stone analysis
• exchange networks
• lithic technology
• obsidian procurement
• trade
• social networks
• X-ray fluorescence

X-ray fluorescence (XRF) spectrometry is commonly used to determine the geological source of obsidian artifacts, helping archaeologists address questions related to social interaction, exchange/trade networks, ethnicity, territoriality, and movement (Bayman and Shackley 1999; Dillian 2002; Arakawa et al. 2011; Dolan et al. 2017; Kocer and Ferguson 2017; Shackley 2019b). Studies of obsidian procurement often use temporal and spatial patterning in non-local obsidian use as indicators of changing social relationships between groups and social identities (Peterson et al. 1997; Jones et al. 2003; Shackley 2005; Mills et al. 2013; Carpenter et al. 2021). We contribute to this body of research by presenting a diachronic analysis of obsidian procurement at Chaco Canyon, New Mexico, spanning AD 500–1250 (all dates are AD); which includes the period when Chaco Canyon became a focal location that influenced populations across the Colorado Plateau.¹

From the tenth to twelfth centuries, Chaco Canyon was the center of a system of regional interaction and a magnet for various non-local goods. Archaeologists use these materials as indicators of the extent and strength of Chaco’s connections to its region and beyond, and of its centrality (Powers et al. 1983; Judge 1991). Maize, animal resources, ceramics, chert, obsidian, and building timbers came from within or at the edges of the San Juan Basin 90 km away (Toll 1991; Cameron 2001; Cordell et al. 2008; Grimstead et al. 2016) Guiterman et al. 2016. Turquoise, shell, copper bells, tropical birds, and cacao were acquired from more distant sources (Judd 1954; Mathien 1984; Toll 1991; Crown and Hurst 2009; Hull et al. 2014; Watson et al. 2015). Obsidian was used towards more quotidian ends and provides a different perspective on regional interactions in the Chaco World.

Despite strong interest in non-local or exotic goods and Chaco’s regional and extra-regional connections, to date there are only two studies of Chaco obsidian (Cameron and Sappington 1984; Duff et al. 2012). Both XRF studies found that people at Chaco primarily used obsidian from near Mt. Taylor (90 linear km from Chaco Canyon) and within the Jemez Mountains (130 km). The first analysis by Cameron and Sappington (1984) has been widely used to interpret social interaction and exchange networks (Powers et al. 1983:329–330; Cameron 1997:263–267; Vivian 1990:164), but some of their results were later determined to be inaccurate due to statistical errors and a lack of source data (Windes 1987c:238). Duff et al. (2012) attempt to correct earlier work by presenting new XRF results, mostly from sites dating 500–975. The authors identify a pattern of increasing use of more distant sources in the Jemez Mountains after 750 and accelerating during the Chaco Era or Pueblo II period (900–1150). However, the pattern is based on a small sample from Pueblo II sites (n = 49) and includes only a few artifacts from great houses.

Using almost 1600 artifacts from 58 sites, the largest sample of XRF-sourced obsidian artifacts from Chaco Canyon to date, we explore changes over time in where and how Chacoans acquired obsidian. By combining obsidian provenance data with technological analyses of obsidian debitage and tools, we first address two questions concerning obsidian procurement and use at Chaco: (1) is there significant variation in obsidian sources and acquisition strategies over time (Cameron 2001; Duff et al. 2012)? and (2) are common assumptions surrounding the expected products of direct or embedded procurement versus long-distance exchange networks met? We then situate Chaco obsidian procurement within regional exchange models and consider the potential influence of social factors particular to Chaco and its region.

The new XRF results and chipped stone technology data presented here fill two gaps noted in previous studies (Cameron 2001; Duff et al. 2012:2998). The most significant gap has been the lack of obsidian source provenance results for Pueblo II sites. Understanding Chaco Canyon’s relationships with its wider region during the Pueblo II period is important, since the interval spans the construction of great houses and the development, expansion, and decline of the Chaco regional system (Vivian 1990; Lekson 2006). We discuss the largest obsidian sample from Pueblo II sites within Chaco Canyon (n = 766); most are from great houses, including Pueblo Bonito. Additionally, Moore and colleagues (2020:165) note that it is difficult to relate regional obsidian procurement patterning and exchange models to Chaco due to a lack of published data on chipped stone technology. To more fully assess existing procurement models, we examine chipped stone technological attributes, artifact forms, and types from the different obsidian sources, and over time.

Our work leads to new interpretations of Chaco obsidian use and social interactions, but also confirms aspects of previous studies (Cameron 2001; Duff et al. 2012; Moore et al. 2020). We find significant temporal variation in obsidian source use and in acquisition strategies or exchange products. Further, the temporal patterning in both obsidian sources and artifact forms does not appear to relate primarily to raw material or technological factors and does not always fit expectations concerning long-distance exchange networks (Bayman and Shackley 1999:842; Arakawa et al. 2011:785; Moore et al. 2020:166). The results lead us to consider how socio-cultural dynamics from 500 to 1250 influenced obsidian procurement. We argue that the effects of social or cultural heterogeneity and discontinuities, changes in mobility, and the development of long-distance exchange networks may explain much of the obsidian source variability at Chaco.

The Chaco World

Chaco Canyon is one of the most intensely studied areas of the American Southwest (see Sebastian 1992:9–41; Plog 2015:4–14). Pueblo occupation started with two large Basketmaker III (500–750) pithouse settlements first inhabited by highly mobile agriculturalists in the late 400s/early 500s. These two sites, Shabik’eschee and 29SJ 423, may be the largest pithouse settlements on the Colorado Plateau (Wills and Windes 1989). Wills and colleagues (2012:330, 342) argue that both sites represent seasonal aggregation of households, and not static villages, and that more pithouses may be buried under alluvium.

During the Pueblo I period (750–900), occupation in the Chaco area consisted of scattered, short-lived hamlets, except for certain locations where great houses developed. These multi-storied, monumental masonry structures began to be built in and near Chaco Canyon in the late 800s (Vivian 1990:158; Lekson 2006:77; Windes 2018b). Archaeologists have identified hundreds of great houses outside of the canyon, often called outliers, that are architecturally similar to those in Chaco Canyon and are often surrounded by small houses (Kantner 2004:72–74). Many outliers were built later in the 1000s and 1100s as Chaco culture expanded over an area of about 150,000 sq km. This Chaco World, centered in the San Juan Basin, was physically and symbolically linked by 400 km of road networks, earthworks, great houses, and religious structures called great kivas (Sebastian 1992:129; Kantner 2004:75–76).

The Pueblo II period (900–1150) was a time of increasing social complexity and transformation, long-distance exchange networks, and the decline of the Chaco system beginning around 1130 (Sebastian 1992; Mills et al. 2018). While the nature of Chaco’s socio-economic and political structure is debated (Judge 1991; Earle 2001), most archaeologists acknowledge some level of social hierarchy by the Pueblo II period (Plog and Heitman 2010; Kennett et al. 2017). Some believe Chaco Canyon great house residents were elites with special access to exotic goods, who controlled and redistributed these goods to consolidate and signal their power (Neitzel 2003; Watson et al. 2015). Others propose that inhabitants of the small houses were the working class who had less access to resources (Akins 1986), and different socio-cultural, linguistic, or biological affinities than those in great houses (Vivian and Mathews 1965:108–111; Vivian 1990; Lekson 2006:93).

Jolie and Webster (2015:98) think of Chaco as a “potentially socioculturally diverse or heterogeneous community resulting from the aggregation of multiple groups of people of different ethno-linguistic, culture-historical, or enculturative backgrounds … ” This idea is supported by evidence of culturally or biologically distinct groups (Judd 1954; Akins 1986; Schillaci and Stojanowski 2003). Several factors potentially contributed to Chaco’s sociocultural diversity, including migration or mobility, social discontinuities (Windes and Van West 2021), the development of regional networks (Mills et al. 2018), and the canyon’s role as a cultural, religious, and economic center (Judge 1991; Earle 2001; Renfrew 2001). These social factors likely influenced obsidian procurement or circulation and may be reflected in obsidian source variability and richness. In socially heterogeneous contexts, different groups often have unique knowledge and social or historical connections that are reflected in the materials used (Peterson et al. 1997; Bayman and Shackley 1999; Shackley 2005, 2019b).

At its height of influence Chaco was a gathering place, drawing individuals or social groups to the canyon from areas at the edge and outside of the San Juan Basin, and perhaps as far as the Jemez Mountains and Rio Grande Valley. People came for religious events and pilgrimages (Renfrew 2001), trade fairs (Toll 1991), and great house building campaigns (Judge and Cordell 2006:204); some probably stayed. These interactions may have involved obsidian in different artifact forms (cores/debitage vs. formal tools).

At Chaco, obsidian often comprises less than 10% of chipped stone assemblages, but represents higher percentages of formal tools (Cameron 1997:561). Obtaining obsidian required knowledge of and access to source areas and/or social relationships with those with access. Some of Chaco’s obsidian comes from the southern edge of the San Juan Basin near Mt. Taylor, but a large portion originates outside of the Chaco regional system, to the east within the Jemez Mountains region (Figure 1). There are few habitation sites directly at the obsidian sources. However, there are sites near the sources in the Mt. Taylor and Jemez areas that are coeval with the Chaco sites discussed here (Vivian 1990; Wase et al. 2000; Anschuetz 2006; Anscheutz 2007:14–15; Moore et al. 2020), suggesting that territoriality, conflict, competition, and alliances likely influenced procurement.

Figure 1. Major obsidian sources and geographic or cultural regions discussed.

Note: The cultural regions are mostly used as spatial or geographic units. We recognize the regions are culturally diverse, and we only touch on certain aspects of these areas. Primary sources shown include: BR = Bearhead Rhyolite, CC = Canovas Canyon Rhyolite, CT = Cerro Toledo Rhyolite, ER = El Recheulos, MT = Mt. Taylor sources, VR = Valles Rhyolite.

The Obsidian Sample

The obsidian sample in this study derives from sites within Chaco Culture National Historical Park (Figure 2). Over half are from excavations (n = 795), and this includes almost half of the obsidian (n = 335) recovered during the National Park Service’s (NPS) Chaco Project (1973–1982). Samples also derive from surface collections during NPS surveys and on-site chipped stone analyses from 1998 to 2002 (Windes 2018a:5–7, 15). The remainder are from surface and excavation collections during NPS preservation activities (n = 63).

Figure 2. Sites and communities/areas with obsidian samples used in this study.

Ceramic analysis allows us to assign obsidian samples to four periods: Basketmaker III, Pueblo I, Pueblo II, and Pueblo III (1150–1250). These are combined into Early (500–900) and Late (900–1250) groups to analyze general trends. For excavated samples, dates are from published analyses and recent re-assessments of ceramics within excavation levels and archaeological contexts. Dates for surface collected obsidian are based on ceramic analyses by Thomas Windes or published date ranges. When possible, we placed the obsidian artifacts into temporal groups based on their association with ceramics within specific contexts. Additional sample details are provided in Supplemental Text 1.

The XRF source data presented here includes all analyses completed after the first project by Cameron and Sappington (1984), and a reanalysis of over 40% of the artifacts from their study. Cameron and Sappington's obsidian source determinations are not identified in catalog records, so direct comparison is impossible. However, we now know that all of Cameron and Sappington’s (1984) Red Hill source determinations and most of the other sources with trace amounts are incorrect (Duff et al. 2012).

Analytical Methods

X-Ray Fluorescence

XRF spectrometry is an established method of sourcing the geologic origins of archaeological obsidian (Shackley 2005, 2011). Shackley analyzed 84% of the Chaco obsidian artifacts used in this study at his Geoarchaeological XRF Spectrometry Lab. The remaining obsidian artifacts were analyzed at New Mexico Institute of Mining and Technology’s XRF Lab (Kunselman 1999). Both labs used similar equipment, and these methods are summarized by Shackley (2005:95–96, 2011; also at http://swxrflab.net/anlysis.htm). Shackley’s recent analyses (2016, 2017, 2019a) used a laboratory energy dispersive XRF spectrometer, while earlier analyses at Berkley used a wavelength dispersive XRF spectrometer (Shackley and Dillian 2000; Duff et al. 2012:2999). These analyses using the same international geological standards are statistically comparable (Shackley 2019a, 2019c; Supplemental Text 2). Supplemental Table 1 shows all XRF results by area or site and time period.

Chipped Stone Artifact Analysis

The goals of the chipped stone analysis were to characterize obsidian reduction and tool technology for sourced obsidian artifacts and to identify the form in which people acquired or imported obsidian (debitage/cores vs formal tools). We sorted debitage from tools and classified tool types according to their morphology/form, attributes, and presence of edge modifications/use wear (Andrefsky 1998:75). Projectile points were classified using a descriptive typology developed for Chaco Canyon (Lekson 1997). We separated debitage into platform bearing fragments, whole flakes, and non-platform bearing fragments (Sullivan and Rozen 1985). We then conducted individual flake attribute analysis to understand raw material reduction strategies and tool production techniques. Supplemental Text 3 provides more information about the variables, the artifact type definitions used, and analysis results not included here.

We use artifact form, artifact type, and the proportion of artifacts with cortex and flakes with cortical platforms to investigate reduction and acquisition strategies. Obsidian at Chaco was part of an expedient reduction technology focused on flake production from unprepared cores, informal tools, and small bifaces/projectile points made from flake blanks. People often reduced small obsidian nodules using bipolar reduction to maximize material (Shott 1989; Andrefsky 1994), which may account for the low number of obvious cores.

To analyze artifact form and calculate debitage-to-formal-tool ratios, we group obsidian artifacts into formal tools, informal projectile points, informal tools, cores, and unmodified debitage (Andrefsky 1998:213–214). Formal tools have flaking or retouch covering over one-third of at least one face (Cameron 1997), and are mostly bifacial projectile points, bifacial knives and/or cores, or unfinished points (preforms) without haft elements (Kelly 1988). Informal projectile points are minimally retouched flakes that do not fit the definition of formal tools but have haft elements. Some of these may be considered unfinished, but most are shaped enough to have served as expedient points. Informal tools include debitage with edge damage or evidence of retouch and/or utilization (Binford 1979). Cores are raw material with flake removals and tested nodules (Andrefsky 1998:12–15). Artifact types associated with biface technology include biface thinning flakes (Frison 1968), bifaces (whole and fragments), and unfinished bifacial projectile points or preforms.

We use debitage-to-formal-tool ratios to identify obsidian procurement/use patterns and acquisition strategies. Yet, we also recognize that people likely discarded some formal tools away from habitations and many were reused. The ratios are suggestive of patterns that need further research and confirmation. For the calculated ratios, debitage counts include whole flakes, flake fragments, and utilized/retouched debitage (see Supplemental Table 2).² We assume that low debitage-to-formal-tool ratios suggest people imported or acquired more finished tools. Conversely, relatively high debitage-to-formal-tool ratios imply greater import of raw material, cores, and flakes, and on-site reduction at Chaco (Cameron 1997). Samples that may be biased due to collection methods are excluded from the ratio calculations (n = 100).³

Cortical flakes retain a portion of the weathered exterior of raw materials. We use the proportion of cortical artifacts and flakes with cortical platforms as a proxy for on-site reduction and whether obsidian nodules were reduced before or after people brought them to Chaco (Johnson 1989). We assume that relatively high percentages of cortical flakes and flakes with cortical platforms reflect more on-site raw material and core reduction, and less reduction before transport, since the weathered exterior of raw material is often removed initially (Tomka 1989).

Acquisition Strategies and Assumptions

Based on ethnographic evidence agricultural groups may travel 10–18 km for tool stone (Arakawa 2006:14; Kocer and Ferguson 2017:534). Obsidian exchange studies of agricultural groups often assume materials from more distant sources (>20 km) were acquired through informal down-the-line exchange or more formal structured exchange networks. (Moore et al. 2020; see below). However, at Chaco it is an open question whether obsidian was acquired directly from source areas or arrived through exchange networks, and the artifact forms associated with these procurement or acquisition strategies might differ.

Using the Chaco obsidian data, we examine the assumptions surrounding the expected products of direct or embedded procurement versus long-distance exchange. For instance, with direct procurement, we may expect to see more bulk material (nodules or cores) and/or evidence of raw material reduction (debitage/cortical flakes) and formal tool production at Chaco sites (Harry 1989:292; Cameron 1997:550; Peterson et al. 1997:236–238), although some material may be initially reduced at quarries for portability. With indirect procurement (either down-the-line or more formal exchange), finished tools like projectile points or bifaces are expected (Bayman and Shackley 1999:842), especially at the end of long-distance exchange networks (Moore et al. 2020). Finished tools are more portable than some obsidian raw materials, and they may be associated with male groups or hunting societies (Szuter 2000:8; Shackley 2005:164–166; Arakawa et al. 2011), ritual deposition or offerings (Cameron 2001:98) and used in ceremonies (Bayman 1995:53–54). Nodules, cores, or flakes may also be part of exchange networks. However, the literature is peppered with considerations of least-cost principles and the need for finished hunting tools or weapons (Binford 1979; Bamforth 1986; Bayman and Shackley 1999:842; Arakawa et al. 2011), leading to assumptions that long-distance obsidian exchange networks focused on light-weight tools like projectile points (Harry 1989; Peterson et al. 1997; Moore et al. 2020:166). We test these assumptions by investigating patterning in obsidian acquisition strategies.

Obsidian Technology and Sources

Before presenting the results, we briefly summarize the most common obsidian sources represented at Chaco. Shackley (2005) provides a more extensive description of Southwest obsidian sources at http://swxrflab.net/swobsrcs.htm. Most of the obsidian available within 200 km of Chaco is suitable for small tool production. The size and quality of available obsidian raw material did not constrain Chacoan chipped stone technology, which was focused mostly on small informal tools, and not bifaces, although small biface cores and tools are more common during the Basketmaker III period. The mean length of complete obsidian formal tools in the overall sample is 2.5 cm, and many are not completely flaked. Much of the Jemez obsidian from Chaco sites has spherulites and Mt. Taylor obsidian has sanidine inclusions, so quality was not a primary consideration in acquisition (Supplemental Text 4).

Mt. Taylor Obsidian

The Mt. Taylor Volcanic Field (Figure 1) contains two geochemically distinct sources, Grants Ridge and Horace/La Jara Mesa, which differ both in quality and nodule size (Shackley 2005:58–64). Grants Ridge raw material nodules are generally larger, but controlled flaking may be hindered by inclusions (Shackley 1998:1078). Horace/La Jara Mesa obsidian is a superior raw material due to its lack of phenocrysts, yet both sources were used equally at Chaco.⁴ The materials are also found in secondary contexts in the Rio San Jose, Rio Puerco, and in smaller quantities in Rio Grande alluvium, as far south as Chihuahua, Mexico (Shackley 2021). Both obsidians actively erode into Lobo Creek and the Rio San Jose, but Grants Ridge obsidian is more common in drainages. These streams drain to the south, so the primary sources near Mt. Taylor are closer to Chaco than the secondary alluvial sources.

Jemez Mountains Obsidian

The Jemez Mountains sources are the largest in the region. Of the five source materials, the finest quality obsidian for tool production is El Rechuelos Rhyolite (Shackley 2005:69). The El Rechuelos (ER) source area is at the northern edge of the Valles Caldera (Figure 1). Minor amounts erode into the Rio Grande via the Rio Chama. The two largest sources are at Cerro del Medio (Valles Rhyolite [VR]), and Rabbit Mountain (Cerro Toledo Rhyolite [CT]). Raw material nodules often exceed 30 cm, but both VR and CT can have abundant inclusions (spherulites). Very little VR erodes out of the Valles Caldera and this material was likely acquired at or near Cerro del Medio. CT obsidian can be found in Rio Grande alluvial deposits as far south as Mexico.

Bearhead Rhyolite (BR) and Canovas Canyon Rhyolite (CC) are available on the southern edge of the Jemez Mountains (Shackley et al. 2016). Nodules are small (2–5 cm) and can be found in secondary Rio Grande deposits (Shackley 2021). BR obsidian derives from Paliza Canyon and erodes into Vallecito Creek and the Jemez River. The CC obsidian source is near Bearhead Springs Peak, but material is found in drainages to the south, often with BR nodules, but rarely in Rio Grande alluvium (Shackley 2005:72–73, 2021).

Early Obsidian Use

Basketmaker III

Most of the Basketmaker III samples are from sites occupied between 500 and 650 on Chacra Mesa near Shabik’eschee (29SJ 1659) and the West Mesa near 29SJ 423 (Supplemental Table 1). Large samples are from surface collections, with smaller amounts from excavations at Shabik’eschee and 29SJ 423 (Supplemental Text 5). Obsidian is the most common non-local chipped stone material at Basketmaker III sites and it totals more than 8% of the chipped stone from Chaco Project excavations (Cameron 1997) and 5% from on-site lithic analyses (Windes 2018a). Formal tools make up 8% of our total Basketmaker III sample.

Over three-quarters of the Basketmaker III sample is Mt. Taylor (MT) obsidian (n = 547), suggesting strong connections with the southern edge of the San Juan Basin/northern Cibola region, and possibly direct procurement from MT source areas. Small amounts of Jemez obsidian at Basketmaker III sites reflect early connections with the Jemez Mountains area. Combined BR and CC obsidian comprise over half of the Jemez obsidian in the Basketmaker III sample. After the Basketmaker period, BR and CC obsidian are rarely used at Chaco, and are infrequent in regional assemblages from all time periods (Shackley 2017:4–5). VR is the second most common Jemez obsidian in the Basketmaker sample. Cow Canyon and Mule Creek (Antelope Creek) obsidian from near and along the NM/AZ border are surprisingly rare (n = 2) considering the size and extent of these sources.

The presence of CC and BR obsidian, which derives from the southern edge of the Jemez Mountains, may indicate early relationships between Chaco Basketmakers and communities near modern day Jemez Pueblo (see Anscheutz 2006:251–252). There is more CC than BR obsidian, even though they often occur together in secondary drainages, flaking qualities are similar, and BR nodules are often larger (Shackley et al. 2016).

The XRF results for excavation and surface samples from the two large sites of Shabik’eschee and 29SJ 423 appear similar, except Shabik’eschee has more Jemez obsidian. Compared to West Mesa, the Chacra Mesa sample has more Jemez obsidian than expected (16% more), suggesting stronger connections with the Jemez Mountains area.⁵ It is notable that site 29SJ 628 in Marcia’s Rincon on the canyon bottom has a higher proportion of VR obsidian than most Basketmaker sites (36%). This late Basketmaker III/early Pueblo I site was occupied into the late 700s (Newren (Truell) 2018). Pueblo I sites typically have more VR obsidian, providing evidence that VR use starts to increase in the mid-to-late 700s. Supplemental Text 5 provides more results for Basketmaker III sites.

Debitage-to-formal-tool ratios and cortex data suggest more on-site reduction of Mt. Taylor obsidian cores compared to Jemez. The debitage-to-formal-tool ratio for MT obsidian is 12 (flakes per formal tool), over twice the ratio for Jemez obsidian (5). All five of the cores in the sample are MT obsidian, and four of them have cortex. Almost half of all MT artifacts in the Basketmaker sample have cortex, and 17% of MT platform-bearing flakes are cortical.

The low debitage-to-formal-tool ratio for VR obsidian (3) shows people did not commonly reduce this material at Basketmaker III sites and imported it mostly as finished tools. In contrast, the CC (12) and BR (10.5) ratios indicate they were the only Jemez obsidian routinely acquired as raw material nodules and cores and reduced at these sites. Of the Jemez materials, CC and BR also have the highest percentage of cortical artifacts (46% to 60%) and CC flakes have a high percentage of cortical platforms (19%).

Compared to later samples, there are more biface thinning flakes and biface tools/cores in the Basketmaker III sample and it is apparent that people produced MT obsidian bifaces at the Chaco-Basketmaker sites. Roughly three-quarters of the Basketmaker III whole and fragmentary bifaces (72%) and unfinished points/preforms (79%) are MT obsidian, as are over half of the biface thinning flakes (Figure 3). VR and ER artifacts associated with on-site biface production are rare, and people imported these materials mostly as small reduced bifaces, projectile points, and flake blanks/preforms. CC is the only Jemez obsidian associated with in-canyon biface production (Bankston 2018).

Figure 3. XRF results for Basketmaker III obsidian artifacts associated with biface/projectile point production. Counts shown in boxes in all graphs.

Pueblo I

Most of the Pueblo I sample is from surface collections at sites within the South Fork Valley, 10 km south of Chaco Canyon, and excavations at 29SJ 724 (Werito’s Rincon). Sample totals from Pueblo I sites are relatively small (n = 68), since occupations were often short and artifact densities low (Cameron 1997; Windes 2018b). Obsidian use also declined during the Pueblo I period, totaling less than 1% of the chipped stone recovered from excavations (Cameron 1997) and just over 2% of the surface chipped stone at Pueblo I sites in the South Fork Valley and Pueblo Pintado areas. Formal tools comprise 20% of the total sample.

Use of MT obsidian decreases, and accounts for less than 40% of the Pueblo I obsidian. Compared to the Basketmaker III period, VR obsidian use increases from 7% to 30%. Obsidian source variability between Pueblo I sites suggests the shift away from MT obsidian was gradual and/or based on differing social relationships. Despite this variability, debitage-to-formal-tool ratios continue to reflect more on-site reduction of Mt. Taylor (11) compared to Jemez (2) obsidian. While more Jemez obsidian (VR) was used compared to Basketmaker III, Chacoans still acquired it primarily as formal tools. There is little evidence for on-site obsidian biface production during Pueblo I. Only one obsidian biface thinning flake and a few unfinished points were identified.

Late Obsidian Use

Pueblo II/III

Pueblo II samples are primarily from great house excavation assemblages dating 1000–1150, including obsidian from Pueblo Bonito (n = 454), Pueblo Alto (n = 120), and Kin Kletso (n = 68). Fewer samples are from small house sites, including surface collections from South Fork Valley (n = 7) and Pueblo Pintado sites (n = 31), excavations at 29SJ 627 (n = 10) at Marcia’s Rincon, and surface collections and excavations during NPS preservation activities at Casa Rinconada area sites (n = 73).

Moore and colleagues (2020:165) discuss the need for more XRF of Chaco obsidian from contexts dating to the 1200s for comparison with Mesa Verde or the northern San Juan. However, the excavation sample from 29SJ 633 (n = 10) is the only sourced obsidian that may date to after the 1100s or Post-Chaco era (Pueblo III). This sample derives from trash deposits within Rooms 7 and 8, which may relate to reoccupation by “Mesa Verdeans” in the late 1100s to early 1200s (Mathien 1991:116; see Supplemental Text 7).

Obsidian use appears to increase by the mid-1000s to 1100s, as Chaco reached the apex of its regional influence (Cameron 1997:604). The proportion of obsidian tools declines in the Pueblo II sample (6%; or 3% when samples biased towards tools are removed). Jemez obsidian dominates Pueblo II samples (93%). Of the Jemez obsidian, VR use increases dramatically (63%; n = 502) and the frequency of CT climbs relative to Pueblo I (from 7% to 28%). CT obsidian is more common within the great house sample (33%) compared to small houses, which have less CT than expected (6%), and slightly more ER (15%) and MT (12%) obsidian.⁶ However, this difference may be due to fewer samples from small house sites, which also appear biased towards projectile points. Small houses have more obsidian sources (8 sources; n = 120) than great houses (6 sources; n = 646), and this source richness is evident for all obsidian artifacts, including projectile points (Figures 4 and 5).

Figure 4. XRF results for all Pueblo II obsidian artifacts by site type.

Figure 5. Obsidian XRF results for projectile points from Pueblo II sites.

Pueblo II small house sites have a variety of single specimens, often projectile points, from distant or rare sources in Arizona, New Mexico, and Wyoming (debitage). Government Mountain obsidian from Arizona (350 km away) was only found at the Pueblo Bonito (n = 8) and Pueblo Alto (n = 2) great houses. Wild Horse Canyon in southwestern Utah (> 600 km) is the most distant source in the great house sample (n = 1). Obsidian Cliff, Wyoming, is the farthest source identified, and is from a small house (n = 1; 29SJ 399, Kiva 3). Two pieces of Obsidian Cliff material were also identified at Pot Creek Pueblo (LA 260), near Picuris, New Mexico (Boulanger et al. 2022). This is one of the most distant lithic materials found in the prehistoric southwest (> 1000 km from Chaco).

The decline in MT obsidian use is striking considering its proximity to Chaco, its decent quality and use during the Basketmaker III period, and the presence of Chaco-style great houses south of MT sources. Compared to the early periods, the form in which people acquired or imported MT and Jemez obsidian reverses. MT obsidian is mostly debitage in the early periods, but is commonly formal tools in the Pueblo II sample. The debitage-to-formal-tool ratio for MT obsidian is the lowest in the Pueblo II period (3), but the highest for Jemez obsidian (19). During Pueblo II, people imported more VR and CT obsidian cores and flakes. ER continues to be mostly projectile points. Below we provide a detailed analysis of results from Pueblo Bonito, Pueblo Alto, and Kin Kletso. In Supplemental Text 7 we present more results for Pueblo II/III small houses.

Pueblo Bonito and Pueblo Alto. The Pueblo Bonito samples are from the recent University of New Mexico (UNM) re-excavation of Neil Judd’s three trenches in the east and west trash mounds, first excavated in the 1920s (Crown 2016a). Judd (1954) removed few artifacts and redeposited the fill with artifacts back into the trenches (Crown 2016a:4). Ceramic analyses show the mounds accumulated over a short time period, from 1050 to 1100 (Windes 1987b:624; Crown 2016a:6). We analyzed approximately 40% of the obsidian recovered by UNM (n = 454). The samples represent 38% of the obsidian recovered from the east mound trench, 89% of the west, and 42% of the obsidian from the middle trench. All obsidian tools from the trenches were analyzed (n = 21).

Obsidian totals over 5% of the chipped stone recovered from the three re-excavated trenches in the Pueblo Bonito mounds (Wills and Okun 2016). Most of the obsidian is VR, but one-third is CT. The east trench sample has twice as much CT (40%) as the west (p = 0.01; Figure 6).⁷ The east mound may have started to accumulate slightly later than the west and was used longer (Crown 2016a:10), suggesting the use of CT increased in the late 1000s to early 1100s.

Figure 6. Obsidian XRF results by trench in the Pueblo Bonito trash mounds.

The Pueblo Alto samples are from Kiva 10 trash deposits associated with ceramics dating 1050–1150 (Cameron 1987; Windes 1987a). The results show increasing CT obsidian use around the same time as Pueblo Bonito. The Kiva 10 fill consists of three distinct units partially defined by deposition breaks that were called surfaces. The deeper excavation levels, which Windes (1987a:148, 151) describes as Units B (Levels 22–26) and C (Levels 27, 28; floor of kiva), have local Chaco series ceramics dating from about 1080 into the very early 1100s. All of the Unit B and C levels are dominated by VR obsidian, except for Level 22 which includes a possible use surface and marks the beginning of increased CT use (Figure 7). Above this surface, non-local Chuskan and Mesa Verde series ceramics, dating slightly later and up to about 1140, are more prevalent possibly representing reoccupation of the site or increased connectedness with the Mesa Verde or northern San Juan region. The frequency of CT obsidian increases notably in these upper levels, comprising 44% of the Unit A obsidian (Levels 14–22), compared to 25% in the deeper Unit B levels.

Figure 7. Obsidian XRF results by excavation levels within Kiva 10 at Pueblo Alto. “Surface” shown on graph represents “Plaza Surface 2,” a continuation of a plastered surface. Above this surface at Level 22 (480 cm) the frequency of Cerro Toledo Rhyolite increases.

Jemez obsidian from Pueblo Bonito and Pueblo Alto has the highest debitage-to-formal-tool ratios of any samples. The ratios are similar between the two sites, but with larger amounts of VR (ratios = 83 and 72) compared to CT (59 and 46). Only slightly over 2% of the obsidian from the Pueblo Bonito trenches is formal tools (Wills and Okun 2016).

Twenty percent of the obsidian debitage in the Pueblo Bonito sample has cortex, as does 17% from Pueblo Alto.⁸ There are similar proportions of CT and VR cortical artifacts in the Pueblo Bonito sample, but over 30% of VR from Pueblo Alto has cortex, compared to only 4% of CT. In the Pueblo Bonito sample, VR flakes with cortical platforms total 10% of the platform bearing flakes. In the Pueblo Alto sample 14% are cortical. CT flakes with cortical platforms constitute only 1% of platform bearing flakes in the Pueblo Bonito sample and none for Pueblo Alto.

Of the twenty-one sourced projectile points from Pueblo Bonito fourteen are Jemez obsidian (ER, CT, VR), and the rest are Mt. Taylor (n = 3), Government Mountain, AZ (n = 2), and Mule Creek (Antelope Creek), NM/AZ (n = 2). There is only one obsidian projectile point from the Pueblo Alto sample, made from VR.

Kin Kletso. The Kin Kletso samples date from the early to mid-1100s and are from excavations by Vivian and Mathews (1965). We analyzed all of the Kin Kletso obsidian in the NPS collections (n = 68). The samples are VR (60%) or CT obsidian. Most of the artifacts are debitage (93%), and over a third has cortex. No ratios were calculated since only ten artifacts have intra-site proveniences.

Summary

Our results confirm that Chacoans used different obsidian sources over time (Figure 8). There is a strong association between grouped time period and obsidian source (p = 0.00; Cramér’s V = .847).⁹ Mt. Taylor obsidian began to be replaced by Jemez materials in the 700s. Late Basketmaker III and Pueblo I samples generally have about 40–60% Jemez obsidian. By the 900s, most of the obsidian brought into Chaco was Valles Rhyolite from Cerro del Medio in the Jemez Mountains (Shackley 2005:66, 71–72). In the mid-to-late 1000s, Cerro Toledo Rhyolite use increased by 20–30%, especially at great houses. There is a minor resurgence in Mt. Taylor obsidian use in the late 1000s and 1100s, as seen at Chaco small houses and within the Mesa Verde region (Moore et al. 2020:167).

Figure 8. Obsidian XRF results by temporal group. Site 29SJ 633 data is excluded from the graphs.

The analysis of debitage-to-formal-tool ratios, tool types, and cortex data shows temporal variability in obsidian acquisition/use strategies and technological organization. Jemez obsidian was not commonly reduced at Basketmaker III or Pueblo I sites, and people imported it in small quantities, and often as Valles Rhyolite and El Rechuelos bifaces and projectile points (Figure 9; Supplemental Table 3). The exceptions are Canovas Canyon and Bearhead Rhyolite, which were acquired as small nodules or cores and reduced only at Basketmaker III and early Pueblo I sites. People procured Mt. Taylor obsidian mostly as bulk material and reduced it at these early sites. Our results confirm that this pattern reversed in the Pueblo II period (Cameron 2001; Duff et al. 2012), when Chacoans imported Jemez obsidian (Valles Rhyolite and Cerro Toledo Rhyolite) as flakes and partially reduced nodules or cores that they further reduced at sites, and Mt. Taylor obsidian as formal tools.

Figure 9. Grouped artifact types by obsidian source and temporal group.

At Pueblo II sites, Jemez obsidian is overwhelmingly debitage, and although people continued to favor Valles Rhyolite and El Rechuelos for projectile points, formal tools are rare. Chacoans may have exported or discarded some obsidian points away from habitations but compared to earlier periods it appears they imported fewer formal obsidian tools during Pueblo II. The exception continues to be El Rechuelos obsidian which occurs mostly as projectile points during all time periods. The small number of flakes with cortical platforms in the overall sample (6.2%; n = 98) indicates that people partially reduced the obsidian before they transported it to Chaco.

The small number of obvious cores is likely due to bipolar reduction to conserve obsidian during all time periods, resulting in angular debris that is difficult to classify as cores. Some Basketmaker III bifaces probably also served as cores. Only 1% (n = 18) of the overall sample are cores, and 12 of these are from Pueblo II great houses; nearly half are Valles Rhyolite obsidian from Pueblo Bonito. Sites in the Mesa Verde region and other areas of the San Juan Basin also have few obsidian cores (Arakawa et al. 2011; Moore et al. 2020), and surprisingly low percentages (2–5%) are reported at sites in Bandelier National Monument, within 20–30 km of the Jemez sources (Head 1999).

If acquisition were based only on the distance to sources or least-cost path, then use of Mt. Taylor obsidian should be consistent throughout the occupation of Chaco Canyon. The sources are 30–40 km closer to Chaco Canyon than Jemez sources. Over time, we see more distant sources being used at Chaco, which does not fit a classic distance-decay or fall-off pattern (Renfrew 1977).

Raw material quality or lack of inclusions does not seem to be the primary factor driving obsidian selection, and Mt. Taylor sources were not depleted (Shackley 2005:63). If quality were the main factor, we would expect more El Rechuelos obsidian, which is the “finest raw material for tool production in the Jemez Mountains” and the closest Jemez source to Chaco (Shackley 2005:69). Obsidian from the Horace/La Jara Mesa source at Mt. Taylor is of excellent quality, but Grants Ridge materials are overall slightly more common at Chaco and of somewhat lower quality (Shackley and Dillian 2000).

The most striking differences between the Mt. Taylor and Jemez obsidian sources is the size and availability of the material. The Jemez source areas, and the nodules available, are larger and there is more material to choose from. Valles Rhyolite and Cerro Toledo Rhyolite are quality tool-stone, yet within the Chaco assemblages, much of it has abundant inclusions (spherulites), suggesting little prospecting was done. Surely, Chacoan flintknappers considered quality important but how it was measured probably related to intended products/use. Chaco chipped stone tools show a wide range of craftsmanship, and many are not fully or finely flaked. Our impression is that lack of inclusions or flake-ability was not the primary factor in the declining use of Mt. Taylor obsidian. Interpreting changes in obsidian source use at Chaco requires a better understanding of social factors affecting obsidian procurement and exchange networks and a consideration of assumptions related to acquisition and exchange products.

Discussion

Existing models of Chaco socio-cultural history and regional exchange can be used to interpret the obsidian procurement patterning and develop new hypotheses. Below we address what we think were the most salient social factors affecting Chaco obsidian procurement. We do not have all of the data needed to know the exact social mechanisms underlying obsidian acquisition or exchange, and problems of equifinality persist. However, our data provide clues to potential avenues for future research.

The Decline of Mt. Taylor Obsidian Use

So far, explanations for the declining use of Mt. Taylor obsidian and the increase in Jemez sources have focused on the development of new exchange networks involving the northern Rio Grande Valley and the Pajarito Plateau starting in the 1100s (Duff et al. 2012; Moore et al. 2020). While this may be part of the story, it does not fully explain the decline in Mt. Taylor obsidian use, which began earlier in the 700s. We propose that two factors may have contributed: (1) changes in mobility and the reorganization of settlement and regional relationships beginning in the Pueblo I period and the strengthening of connections or formalization of exchange networks with the Jemez region that began in Basketmaker III; and (2) restricted access or reduced use of Mt. Taylor due to cultural or religious reasons, boundaries, conflict, or territoriality.

The high residential mobility of Basketmaker III groups may have provided opportunities for direct procurement of obsidian from the Mt. Taylor sources, possibly as an embedded strategy during collection of other resources. Wills and Windes (1989:359–361) propose that Basketmaker III settlement consisted of seasonal aggregation in different areas including highlands where piñon nuts and obsidian could be collected. Residential mobility decreased during the Pueblo I period with increasing reliance on agriculture, perhaps resulting in less use of the Mt. Taylor highlands and fewer opportunities for direct acquisition. Since the Mt. Taylor sources are the closest to Chaco this does not fully explain their decline after Basketmaker III. However, it does suggest a change in some relationships between Chaco Canyon and Mt. Taylor’s surrounding communities. Over time population changes near and in the northern Rio Grande Valley closer to the Jemez sources likely influenced Chaco’s acquisition of obsidian, but settlement in the Mt. Taylor area and increasing occupations along the east edge of the San Pedro Mountains and north of the Jemez Mountains in the Largo/Gallina region may have also played a role.

Mt. Taylor and the nearby obsidian source areas are at the northern edge of the Cibola or Zuni-Acoma archaeological region (Duff and Lekson 2006:316–319). Modern Pueblo people, including Keresan groups with ethnographically documented ties to Chaco, consider Mt. Taylor a place of religious significance. People at Acoma and Laguna Pueblos and their ancestors used Mt. Taylor for resources and religious activities from about 700 to the present (Wase et al. 2000; Anscheutz 2009). Horace/La Jara Mesa, which has high quality obsidian, has several trails and landscape features with religious importance (Anscheutz 2009:5–19). Ellis and Dittert (1954:19) documented that the northern boundary of traditional Acoma lands lies at Mt. Taylor, which “adjoins the area of old Chaco culture. Along this periphery a blending of traditions is seen. North of the boundary Acoma traits die out rapidly and sites showing distinct Chacoan traits are in the majority. However, the Chaco group moved out of the district by at least 1100.”

It is interesting that the earliest documented Ancestral Pueblo sites on Horace Mesa date to the 700s, right around the time the transition away from Mt. Taylor obsidian begins at Chaco. Site locations appear to be correlated with access to obsidian and piñon nuts (Wase et al. 2000). Conflict over these resources, and possibly the religious use of the area, may have restricted access. We do not have the data to demonstrate the emergence of cultural boundaries or restricted access, but negotiations of territory and use may have influenced procurement.

Obsidian from Chaco-Pueblo I sites is often evenly split between Mt. Taylor and Jemez obsidian sources. This may reflect negotiations for access to Mt. Taylor, along with strengthening connections between Chaco and Jemez or the northern Rio Grande Valley that began in the Basketmaker III period. Windes (2018b:612–613) suggests that the transition from Basketmaker III to Pueblo I settlement in Chaco Canyon represents a social discontinuity that changed interaction patterns and trade networks. The Pueblo I period was a time of profound change, population increases, and movement. New populations from the north may have entered Chaco Canyon to establish communities (Wilshusen and Ortman 1999; Lekson 2006:73–77; Windes and Van West 2021), although Mills et al. (2018) find little evidence for northern migrants during the 800s.

Great house construction began around the same time as Jemez obsidian use increased. Perhaps the changes in obsidian procurement are due to immigrants and new social connections associated with the establishment of great houses. Use of Jemez Mountains obsidian could also relate to the increasing importance of the area in Pueblo ideology and cosmology. At the very least, the shift in obsidian procurement suggests a reorganization of some relationships around the same time as a new settlement system centered on great houses emerged. Obsidian use also markedly decreases during the Pueblo I period. Population dispersal and movement may disrupt exchange networks, resulting in less obsidian import (Head 1999). This fits well with Chaco-Pueblo I settlement models (Windes 2018b; Windes and Van West 2021), and the obsidian procurement patterning.

Peregrine (2001:36–38) proposes that Chaco was a matrilineal society with matrilocal residence, and that this development occurred during the Pueblo I period (also see Kennett et al. 2017). If accurate, one expected outcome of matrilocality is that groups of men were freed up to travel longer distances to trade and/or procure resources. Men could have acquired obsidian raw material during hunting and collecting trips and projectile points may have circulated within new male networks, possibly through an association with hunting and male sodalities that cross-cut kinship (Szuter 2000:8; Arakawa et al. 2011:280–283; Arakawa 2013:284).

Social network analysis based on ceramics from Chacoan great houses and great kivas shows little overlap between ceramic and obsidian networks (Mills et al. 2018). The general patterning in non-local lithic raw material procurement at Chaco does appear to show a re-orientation of connections over time from south/north (early) to east/west (late) similar to ceramics, however there are no apparent direct ceramic connections with the Jemez area. Mills and colleagues (2018:929) show that during the 850–900 period there are strong ceramic ties between Chaco and the Rio Puerco (of the west) and Zuni areas, which are closer to Mt. Taylor obsidian sources. However, this is the time-period when Mt. Taylor obsidian use declines at Chaco. The incongruity between ceramic and obsidian networks may be explained by the association of obsidian with hunting tools and gendered procurement. Obsidian exchange may have involved primarily men (Arakawa et al. 2011; Arakawa 2013), and distant male networks may have facilitated this exchange.

Regional Obsidian Patterning, Exchange Networks, and Acquisition Strategies

The increasing use of Jemez obsidian throughout the San Juan Basin and beyond appears to relate to the development of long-distance obsidian exchange networks that differed from other networks. West of Chaco, in the Chuska Valley, there is a similar pattern of increasing Jemez obsidian use over time (Kearns 1999), suggesting shared obsidian exchange networks. In the Red Mesa Valley, south of Chaco, communities primarily use Mt. Taylor materials, but Jemez sources are also present (Duff et al. 2012:3001–3003). Once Jemez obsidian is favored by the 900s to 1000s, the dominant use of Mt. Taylor obsidian becomes confined to communities within approximately 40–50 km of the sources.

Based on changes in obsidian source proportions over time, and the prevalent artifact forms amongst the Jemez sources, we argue that the Chaco and Mesa Verde regions were not always part of the same obsidian exchange networks or that the intensity of exchange and interactions varied (Moore et al. 2020:166). Over time more use of Cerro Toledo Rhyolite obsidian is evident at both Chaco and in the Mesa Verde region, but patterning in the other Jemez sources differs.

Compared to Mesa Verde, much less El Rechuelos obsidian made it to Chaco. El Rechuelos constitutes 50% of the obsidian from Mesa Verde during early periods (600–920) (Arakawa et al. 2011), but it never reaches more than 17% at Chaco. Also, El Rechuelos totals 65% in the 1060–1140 sample from Mesa Verde, but at Chaco during this time it accounts for only 3% of the obsidian. The low frequency of El Rechuelos at Chaco and Aztec is especially striking when compared to its common use at Mesa Verde before AD 1225. This suggests weaker connections between Chaco/Aztec and the Largo/Gallina region, which is closest to the El Rechuelos source. Kocer and Ferguson (2017:550) note that a “preliminary comparison of lithic materials in Chaco and Gallina provides compelling evidence for the existence of a ‘buffer zone’ between the two groups…”¹⁰

By the Pueblo I period Valles Rhyolite use is similar between Chaco and Mesa Verde, but in the Pueblo II period almost twice as much came into Chaco (63%). The percentage of Cerro Toledo Rhyolite in the Chaco-Pueblo II sample is similar to the late Mesa Verde sample (1060–1280), but this date range includes the late 1100s and 1200s when we have few Chaco samples. Currently, the Mesa Verde-Pueblo II/III obsidian data are not temporally discrete enough for direct comparison with the late Chaco sample, and much of the Cerro Toledo Rhyolite in the Mesa Verde sample may be from the 1200s (Arakawa et al. 2011: Table 3).

The increase in Valles Rhyolite and Cerro Toledo Rhyolite use during Pueblo II may relate to Ancestral Pueblo groups moving onto the Pajarito Plateau and closer to Jemez sources, as Moore and colleagues (2020) suggest. However, Jemez obsidian (Valles Rhyolite, Canovas Canyon, and Bearhead Rhyolite) is well represented at earlier Chaco-Basketmaker III sites, suggesting connections between Chaco and Jemez or the northern Rio Grande Valley began before groups moved onto the Pajarito Plateau. Before the 1000s, Jemez obsidian may have been obtained through more informal exchange networks involving communities near and along Vallecito Creek and the Jemez River, which is closer to Chaco than the Pajarito Plateau.

Based on the great house samples, people imported more Cerro Toledo Rhyolite into Chaco starting in the mid to late 1000s. This is slightly earlier than the increase in the Mesa Verde region (Moore et al. 2020, Table 3) and may represent the addition of a new social/exchange network that was first centered on Chaco. Obsidian from sites at Aztec Ruins dating from the late 1000s to 1200s, including the Chaco-style great houses (North and West), are also mostly Cerro Toledo and Valles Rhyolite (Lori Stephens personal communication 2022; Turner 2019).

Contrary to Moore et al. (2020:162, 165), our results do not fully support the expectation that finished tools make up a larger proportion of obsidian assemblages as distance increases from sources. Our analysis of artifact form/type by obsidian source shows a more nuanced picture at Chaco. Valles Rhyolite was part of exchange networks focused mostly on projectile points, bifaces and flake preforms only during Basketmaker III and Pueblo I. The roots of increased Valles Rhyolite obsidian use during Pueblo II were established during these earlier periods. However, the acquisition strategy or commodity changed by Pueblo II to cores and flakes of Valles and Cerro Toledo Rhyolite, reflecting either a shift in desired products/use or direct procurement from source areas in the Jemez Mountains. Obsidian at the Aztec North great house is also mostly debitage (Throgmorton 2017). This is clearly different from Mesa Verde, where most of the obsidian is finished tools (Arakawa et al. 2011:790). El Rechuelos is the only obsidian that appears to be part of exchange networks focused on finished tools during the entire occupation of Chaco Canyon, but proportionally less so than in the Mesa Verde region.

Valles Rhyolite obsidian does not erode out of the Jemez Mountains in any quantity and had to be obtained close to the source area at Cerro del Medio (Figure 1). Cerro Toledo Rhyolite can be found in drainages, although the primary source areas in the Jemez Mountains are as close to Chaco as these secondary drainages (Shackley 2021). While cortex type (waterworn vs. primary nodule cortex) was not coded during all of our analyses, it was noted for the Pueblo II sample. Most of the cortex on Pueblo II obsidian is not waterworn and is from nodules collected at the primary Valles Rhyolite and Cerro Toledo Rhyolite sources, providing further evidence of direct procurement and exchange of cores/flakes.

Social and Cultural Heterogeneity in the Pueblo II Period

The obsidian procurement patterning and high source richness at some Pueblo II sites seem to support the idea that Chaco was socially and culturally diverse and incorporated groups from outside the San Juan Basin during the Pueblo II period. Social heterogeneity may be evident in the apparent dichotomy between great and small houses. Based on higher obsidian source richness, some inhabitants of small houses may have had more diverse regional and extra-regional connections than great house residents. The rare obsidian sources at small house sites possibly reflect the geographic connections of small house residents, which some see as the Chaco labor force (Vivian 1990; Sebastian 1992). The lower richness at great houses may be due to elite control of materials or more formalized connections or exchange.

The increasing use of Valles Rhyolite and Cerro Toledo Rhyolite obsidian suggests that the northern Rio Grande Valley was part of the Chaco interaction sphere at least by the 1000s. During the Pueblo II period, people from the Rio Grande Valley may have carried Jemez obsidian to Chaco during pilgrimages, given it as gifts at religious or social events, or used it as offerings in renewal ceremonies (Cameron 2001; Crown 2016b:213). Chacoans likely also acquired Valles Rhyolite and Cerro Toledo Rhyolite obsidian during hunting trips or religious pilgrimages to the Jemez Mountains (Lekson 1997). During all time periods procurement may have been embedded in other activities, including collection of turquoise, timbers, and other materials, which may account for some of the obsidian arriving to Chaco. However, social relationships still factor into embedded procurement models.

Conclusion

Our research identifies a pattern of increasing use of Jemez obsidian at Chaco that is also seen in other regions of the northern Southwest, opening up new avenues for regional analysis of social/exchange networks. Regional obsidian source patterning suggests that multiple exchange networks operated over time, perhaps reflecting shifting cultural/ethnic divisions and alliances, boundaries, and kin relations with communities closer to obsidian source areas. As populations grew in the northern Rio Grande Valley beginning in the 1000s, Jemez obsidian became part of a wider system of commodity circulation or exchange, supplanting the varied patterns of earlier periods (Duff et al. 2012), even when other obsidian sources were closer. This pattern strengthens by the 1200s, when Jemez obsidian is over-represented at sites far from the sources (Arakawa et al. 2011; Mills et al. 2013:5788). During the 1300s, more sites in northwestern New Mexico and Arizona deviate from a site-to-obsidian source distance-decay model. This may be due to the reformatting and aggregation of Ancestral Pueblo occupations following migrations out of the Four Corners region, and new social networks that reduced the cost of obsidian procurement (Mills et al. 2013:5788–5789).

The shift from the exchange of Jemez obsidian finished tools before the 1000s to mostly cores and debitage after this time suggests direct procurement and movement between portions of the San Juan Basin and northern Rio Grande Valley. However, it is also possible that obsidian acquisition became more formalized and focused on bulk material as part of a more commodity-based exchange network that favored these products for household use. Finished formal tools may have circulated within various networks including sodalities that cross-cut kinship or households and increased regional connectivity.

Provenance and technological studies of obsidian have facilitated increasingly nuanced discussions of past behavior and actions. We are hopeful that the addition of this sizable dataset from Chaco Canyon helps to enrich our understanding of obsidian procurement over time from numerous contexts at this World Heritage site and that these data further regional understandings of Pueblo land-use history and contribute to future macro-regional assessments.

Acknowledgements

The authors would like to thank Catherine Cameron for her review of an earlier version of the paper and Aztec Ruins National Monument for sharing obsidian XRF results. Special thanks to Sean Dolan and Barbara Mills for their discussions. The maps were drafted by Catherine Gilman. Spanish abstract translation by Mario Zimmermann.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Data Availability Statement

Data cited here are available as published references, and the Geoarchaeological XRF Laboratory reports are available at https://escholarship.org/. Additional data are available on request from the primary author. All artifacts can be accessed at the NPS Chaco Collections at the University of New Mexico.

Supplemental Materials

Supplemental data for this article can be accessed online at https://doi.org/10.1080/00231940.2023.2210479.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

Some of the XRF analyses were funded through Western National Parks Association grants to Windes and by National Science Foundation grants (BCS-0716333 and BCS-0905411) to Shackley.

Notes

1 Our study includes a small sample (n = 10; 29SJ 633) of obsidian from the 1200s, but most of the Pueblo II/III obsidian predates the 1200s.

2 Debitage-to-formal-tool ratios do not change significantly if only whole flakes are included. Informal projectile points are excluded from the formal tool category in the ratios.

3 Tourist finds, grab samples of tools, Archaic points, or samples with unknown collection methods were excluded from the calculations. The results are similar if these samples are included, except that the removal of Archaic points results in less Valles and El Rechuelos Rhyolite.

4 XRF analyses before 2002 were unable to separate Grants Ridge and Horace/La Jara Mesa sources for a number of samples due to a lack of source standard data from Horace Mesa (Shackley 1998). The two locations are within a few kms of each other and are combined in some of the analyses. See Supplemental Text 4 for more information on XRF results for the Mt. Taylor sources.

5 Chi-square: grouped Mt. Taylor and Jemez obsidian by Chacra and West Mesa BM III sites. (χ² = 31.443, df = 1, n = 626, p = 0.00, Cramér’s V = .224). All stats tables are in Supplemental Text 6.

6 Chi-square: great vs small house sites and Jemez obsidian sources (χ² = 83.8, df = 2, n = 728, p = 0.00; Phi = .339).

7 Possible association between east trench and CT, compared to the west (VR and CT only). Chi-square: (χ ² = 11.397, df = 1, n = 320, p = .001, Cramér’s V = .189).

8 These results differ from Wills and Okun (2016:147) who state that only 1.3% of obsidian from Pueblo Alto (Late Mix) have cortex. See Supplemental Text 5.

9 Chi-square: Early (Basketmaker III/Pueblo I) and Late (Pueblo II/Pueblo III) grouping by obsidian source. (χ² = 1109.36, df = 6, n = 1548, p = 0.00; Cramér’s V = .847).

10 The use of Government Mts. (AZ) obsidian also differs, with small amounts found at Mesa Verde early and none after 1000. Chaco shows the reverse pattern, with Government Mts. obsidian only found at Pueblo II great houses.