Chapter 1: Introduction, The Sand Canyon Archaeological Project: Site Testing

The Sand Canyon Archaeological Project is a case study of Puebloan settlement in southwestern Colorado. Beginning in 1983 and continuing to the present, the Crow Canyon Archaeological Center has conducted extensive archaeological investigations in an effort to better understand the ancient Puebloan communities of the mesa and canyon country west of Mesa Verde National Park. The Site Testing Program, which consisted of small-scale test excavations at 13 sites, was an essential component of the larger project and is the subject of this volume.

The Puebloans

The Puebloan peoples of the Mesa Verde region, along with those of the Chaco and Kayenta regions to the south, occupied the middle elevations of the San Juan River basin in the Four Corners area of the Southwest, where the states of Colorado, New Mexico, Arizona, and Utah intersect (Figure 1.1). These people are generally considered to be among the ancestors of present-day Pueblo Indians in Arizona and New Mexico. Puebloan emigrants are thought to have moved from the San Juan River basin into the Rio Grande Valley to the southeast and the Little Colorado River basin to the south in the late thirteenth century A.D. For decades, Southwestern archaeologists have assumed that the Puebloans were "pushed" from the San Juan River basin by unfavorable cultural or environmental trends, or a combination of both. These models assume that the population was unable to mitigate or reverse these trends. In the last decade, however, some archaeologists (Ahlstrom et al. 1995; Varien et al. 1996) have proposed that the Puebloans were "pulled" from the basin by better conditions and new forms of social organization in areas to the south. One of the goals of the Sand Canyon Archaeological Project is to further explore the "push vs. pull" issue by documenting the cultural and environmental conditions in the years leading up to abandonment.

The Sand Canyon Archaeological Project

The geographic focus of the Sand Canyon Archaeological Project is the Sand Canyon locality (Lipe 1992a:2), which is part of the McElmo drainage unit of the Northern San Juan area (Eddy et al. 1984). A locality, as defined by Willey and Phillips (1958:18), is an area larger than a settlement but smaller than a region; in terms of its practical application in the current study, the locality may be thought of as a "heuristic construct, designed to ensure that the research is conducted at a scale appropriate to investigating one or a few multisite communities" (Lipe 1992a:2). The Sand Canyon locality (Figure 1.2) is an area of approximately 200 km² surrounding Sand Canyon and Goodman Point Pueblos, two large villages dating to the final few decades before Puebloan peoples abandoned the Mesa Verde region as a whole. More than 500 prehistoric sites, large and small, and spanning thousands of years of human occupation, have been recorded in the locality.

The Crow Canyon Center's research in the Sand Canyon locality has focused on the Pueblo III period (see Kidder [1927] for a discussion of the Pecos classification system), usually defined as having lasted from A.D. 1100 to 1300. However, because the Chacoan influence in the northern Southwest does not appear to have declined until approximately A.D. 1150, we are using that date as the beginning of the period in this study.

Research Domains

Because the Sand Canyon locality is characterized by such a rich diversity of sites, because it was occupied throughout the entire prehistoric Puebloan sequence, and because the final abandonment of the study area coincided with that of the Mesa Verde region as a whole, the locality is particularly suited to research questions that have been of great interest to archaeologists throughout the Southwest for more than a century. Specifically, the Sand Canyon Archaeological Project was designed to address three research domains: (1) the abandonment of the Mesa Verde region, (2) community organization in the Sand Canyon locality in the time leading up to abandonment, and (3) the placement of the Pueblo III occupation and abandonment of the locality in broader cultural and theoretical contexts (Lipe, ed. 1992). Throughout the project, Crow Canyon archaeologists have sought to document social, cultural, and environmental changes in the study area and to compare their findings with phenomena documented elsewhere in the Puebloan Southwest and in formative societies around the world. Identifying the types of social and ideological systems that succeeded the Chacoan influence in the northern San Juan area, defining the level of sociopolitical complexity in late prehistoric Pueblo communities, and tracing the shift in settlement pattern from dispersed to aggregated communities during the A.D. 1150-1300 period are of particular interest.

The scope of research necessitated the completion of a number of instrumental studies, including environmental reconstruction and chronology building, as well as inquiries into site-formation processes, abandonment behavior (particularly as it relates to site formation), and the length and continuity of occupation at both the structure and settlement level (Lipe 1992a:5). Each study relies on a body of middle-range theory (for discussions of middle-range theory-building and its role in archaeological inquiry, see Mills [1994], Raab and Goodyear [1984], Binford [1977], Schiffer [1988], and Ebert and Kohler [1988]); one goal of the Sand Canyon Project instrumental studies is to test, and perhaps refine, the independently derived theory.

The Sand Canyon Project research design is described in greater detail elsewhere (Lipe and Bradley 1986, 1988; Lipe, ed. 1992), and the reader is referred to those sources for more information.

History and Summary of Research

Sand Canyon Project fieldwork began in 1983 and continued through 1993. The results of the first eight seasons of work (1983-1990) are summarized in Lipe (ed. 1992). Additional project-related research is reported in a number of Master's theses (Bullock 1992; Glowacki 1995; Hovezak 1992; Kelley 1996; Kenzle 1993; Kilby 1998; Mills 1987; Munro 1994; Nicklaw 1995), doctoral dissertations (Adler 1990a; Fratt 1996; Huber 1993; Muir 1998; Van West 1990; Varien 1997), books (Adler ed. 1996; Lipe and Hegmon ed. 1989; Van West 1994a; Varien 1998), journal articles (Adler 1996b; Adler and Wilshusen 1990; Bradley 1989, 1993, 1996; Glowacki et al. 1995, 1998; Lipe 1995; Mills 1993; Varien and Potter 1997; Varien and Mills 1997; Wilshusen et al. 1997), contributions to edited volumes (Adler 1989, 1994, 1996a; Driver 1996; Van West 1994b, 1996; Adler and Varien 1994; Varien et al. 1996), and dozens of meeting papers and annual reports that are archived at the Crow Canyon Archaeological Center, in Cortez, Colorado, and at the Anasazi Heritage Center, in Dolores, Colorado.

Fieldwork began in 1983 with the mapping of Sand Canyon Pueblo, a large late Pueblo III site of approximately 420 surface rooms, 90 kivas, 14 towers, and several types of public architecture, including a great kiva and a D-shaped bi-wall structure. Sand Canyon Pueblo was selected for excavation because (1) there had been little recent work at late Pueblo III settlements outside of Mesa Verde National Park, (2) large aggregated sites in the Montezuma Valley were poorly understood, and (3) the pueblo appeared to be a single-component site (Adams 1983a). Excavation has confirmed that the site is largely single component, with the main occupation dating between A.D. 1245 and 1290.

The original Sand Canyon Project research design was heavily influenced by observation of the visible remains at Sand Canyon Pueblo (Adams 1983a, 1985). The high ratio of kivas to rooms, coupled with the observation that large refuse mounds were not present on the surface, led to the proposition that the site was predominantly nonresidential and functioned as a ceremonial center for a surrounding community of dispersed hamlets (Adams 1985). To test the original hypothesis, systematic survey, intensive excavation at Sand Canyon Pueblo, and excavations at outlying sites were undertaken. This research has shown that Sand Canyon Pueblo was a village with a substantial residential population and that many of the surrounding small sites were abandoned before the construction of the large pueblo (Lipe, ed. 1992).

The research design was updated in a successful 1986 National Science Foundation proposal (Lipe and Bradley 1986) and then again in 1988 (Lipe and Bradley 1988). These documents continued to address the questions of community organization and social complexity. At the 1989 Crow Canyon research retreat in Bluff, Utah, a new emphasis--abandonment of Sand Canyon Pueblo, the Sand Canyon locality, and the Mesa Verde region as a whole--was added to the existing research design. The research design was modified once again in 1990, with a successful National Geographic Society proposal for excavation of public architecture at Sand Canyon Pueblo (Bradley and Lipe 1990).

Answering questions posed in the research design has required a multifaceted approach to data collection: site survey, environmental studies, the recording of oral histories, and both intensive and smaller-scale test excavations have yielded data essential to understanding the occupation and abandonment of the locality within a broader regional context. Each of these project components is described in the following sections.

Systematic Survey

Systematic survey began in 1985 under the direction of Carla Van West. Surveys continued in 1986 and 1987, led by Van West and Michael Adler, respectively (Van West et al. 1987; Adler 1988, 1990a, 1992c). This effort, referred to as the upper Sand Canyon survey, resulted in coverage of 2,600 ha (6,400 acres) surrounding Goodman Point and Sand Canyon Pueblos (Adler 1992c). Survey coverage in the locality expanded in 1988 when Native Cultural Services inventoried sites in part of lower Sand Canyon for the Bureau of Land Management (Gleichman and Gleichman 1989, 1992). In 1990, the Crow Canyon Center, working under its Cooperative Management Agreement with the Bureau of Land Management, surveyed an adjacent block of lower Sand and East Rock Canyons (Adler and Metcalf 1991). Together, the 1988 and 1990 surveys covered a block of 664 ha (1,640 acres) that is referred to as the lower Sand Canyon survey area (Figure 1.3). Castle Rock Pueblo is located at the southern edge of this survey area. With an estimated 13 to 16 kivas and 40 to 75 rooms, Castle Rock Pueblo is much smaller than Sand Canyon Pueblo and Goodman Point Pueblo, but it is by far the largest site in the lower Sand Canyon survey area. Figure 1.4 shows an aerial view of most of the Sand Canyon locality.

Settlement-pattern studies that draw on survey data allow us to estimate prehistoric population levels, assess the distribution of sites relative to the natural resources in the locality, study the spatial boundaries of communities, and map the locations of community integrative facilities, all of which are important in addressing the questions raised in the Sand Canyon Project research design. Four hundred twenty-nine sites were recorded in the upper Sand Canyon survey area; 99 were documented in the lower Sand Canyon survey area (75 new and 24 previously recorded sites). The sites include Archaic sites and Puebloan sites dating to each of the Pecos classification periods (Basketmaker, Pueblo I, Pueblo II, and Pueblo III). The culture history in both survey areas is similar. There is evidence of a substantial Basketmaker III (A.D. 600-725) occupation; Pueblo I (A.D. 725-900) and early Pueblo II (A.D. 900-980) sites are present, but sparsely represented relative to the other periods. Population apparently increased in the middle of the Pueblo II period (A.D. 980-1060) and continued to grow throughout the late Pueblo II (A.D. 1060-1150), early Pueblo III (A.D. 1150-1225), and late Pueblo III (A.D. 1225-1300) periods. Population estimates range from 5 to 12 people/km² in the mid-A.D. 900s to a maximum of 30 to 50 people/km² in the late A.D. 1200s.

The distribution of sites relative to the important natural resources in the locality was documented by Adler (1990a). Sites dating to the Basketmaker III through the early Pueblo III periods are located on or near what are today the best agricultural soils; late Pueblo III settlement shifted to the canyon rims and to the talus slopes, alcoves, and benches inside the canyons. Many late Pueblo III sites, including Goodman Point and Sand Canyon Pueblos, were built around large springs. Castle Rock Pueblo is located on a sloping exposure of slickrock dissected by a drainage channel across which multiple small dams had been constructed.

Studies of the spatial distribution of communities have focused on the Pueblo II-Pueblo III period (Adler 1990a, 1990c, 1992a, 1992c, 1994, 1996a; Adler and Varien 1994). Middle-range theory provides a functional definition of communities as decision-making groups that define and validate social access to important productive resources, for example, land, water, and information (Adler 1990a, 1990c). Cross-cultural data are used to develop criteria for recognizing communities in nonstratified societies: population parameters, the size and distribution of integrative facilities, and land-tenure systems are all factors to be considered (Adler 1990a; Adler and Varien 1994; Adler and Wilshusen 1990; Kosse 1996). In the Sand Canyon locality study, communities were defined using both the distribution and clustering of residential sites and the distribution and location of public architecture, for example, great kivas (Adler and Varien 1994).

Two large clusters of sites, with a sparsely settled area between them, are present in the Sand Canyon locality. The two clusters can be recognized from the middle of the Pueblo II period through the early Pueblo III period. Each had its own integrative facility--either a great house or a great kiva--during those periods (Adler 1994). In the approximate center of each cluster is a large aggregated settlement constructed in the A.D. 1200s (Adler and Varien 1994). The two clusters are interpreted as two distinct communities, called the Sand Canyon and Goodman Point communities, after the large sites at their centers. Two main changes took place in the form and location of the Sand Canyon and Goodman Point communities between the middle Pueblo II and late Pueblo III periods. Earlier in their histories, the communities had consisted of small, dispersed residential farmsteads located on the deep, mesa-top soils. During the late Pueblo III period, however, the two communities aggregated into the large villages of Sand Canyon Pueblo and Goodman Point Pueblo, which are located on the canyon rims and talus slopes at the head of Sand and Goodman Canyons. A third community may have been present in lower Sand Canyon, centered around Castle Rock Pueblo (Lipe 1992b:123).

Cross-cultural and survey data are used to understand the shift in community spatial organization from dispersed to aggregated settlement. Models have been proposed that emphasize the level of agricultural intensification, the size and distribution of households and coresidential groups, and access to primary resources, which is directly related to the development of land-tenure systems (Adler 1990a, 1990c, 1992b, 1996b; Varien 1997). In the case of the Sand Canyon locality, this shift from dispersed to aggregated settlement does not appear to have been abrupt. Instead it appears to have been a gradual process that began in the middle Pueblo II period and culminated in the late Pueblo III period.

Between A.D. 980 and 1225, sites became larger. The Pueblo II sites in the locality have small, six- to nine-room roomblocks and a single pithouse or kiva. By the early Pueblo III period, roomblocks averaged more than 13 rooms and often had more than one kiva (Adler 1990a, 1992c; Adler and Varien 1994). This change represents a shift in the size of the group with assured access to primary resources, probably from single-household groups (small sites) to multiple-household, coresidential groups (larger sites).

Sites also became more tightly clustered between A.D. 980 and 1225. By the early Pueblo III period, many individual sites consisted of multiple roomblock clusters. These clusters reflect a trend toward more-aggregated settlement during the early Pueblo III period as compared with the middle and late Pueblo II settlement patterns. These early Pueblo III aggregates, however, are not as large or as nucleated as the three late Pueblo III settlements in the Sand Canyon locality (Sand Canyon, Goodman Point, and Castle Rock Pueblos). The formation of multiple roomblock clusters reflects yet another change toward larger resource-access groups. Adler (1990a, 1992b) argues that this growth in the size of groups with assured access to resources was associated with the change from a low level of agricultural intensity and low levels of resource scarcity to a moderate level of agricultural intensity and moderate levels of resource scarcity.

Sand Canyon Pueblo, Castle Rock Pueblo, and Goodman Point Pueblo are the largest and most aggregated sites in the Sand Canyon locality. The main occupations of Sand Canyon and Castle Rock Pueblos are tree-ring dated to after A.D. 1240. Goodman Point Pueblo probably dates to the later A.D. 1200s, but this is inferred on the basis of pottery types, site layout, and site location, rather than on tree-ring dates, and is therefore more speculative. By the late Pueblo III period, most people in the Sand Canyon locality were probably living in Sand Canyon Pueblo, Goodman Point Pueblo, or Castle Rock Pueblo, although a few small outlying residential sites were still occupied as well (Adler 1992c; Adler and Varien 1994).

To summarize, survey data indicate that there were two communities in the upper Sand Canyon survey area, and possibly a third in the lower Sand Canyon survey area. The settlement pattern of these communities changed dramatically through time, but there were also significant social organizational continuities. There was continuity in the locations of the residential site clusters through time. There was also continuity in the locations of the various types of public buildings in the center of the Sand Canyon and Goodman Point communities (Adler 1994, 1996a; Adler and Varien 1994). During the tenth and eleventh centuries, these included a great kiva in each of the two community clusters. A great house, the Casa Negra site, was constructed in the Sand Canyon community, probably in the late eleventh or early twelfth centuries, and there may have been a great house in the Goodman Point community during this period as well. A prehistoric road links the areas where each communities' integrative facilities are located (Adler 1990a, 1994; Hayes 1981:63).

Several types of public architecture were incorporated into Sand Canyon and Goodman Point Pueblos during the thirteenth century. Both sites have great kivas. Sand Canyon Pueblo has a D-shaped, bi-wall structure that encloses two kivas, and Goodman Point Pueblo has an above-ground, multiwall structure that appears to enclose multiple kivas. Both sites have plazas, isolated towers, and enclosing walls that might have functioned as socially integrative architecture. These spatial and organizational continuities argue for the existence of two distinct communities in the upper Sand Canyon survey area for a period of at least three centuries. Finally, there was an increase in the size of residential sites and site clusters over time. This is interpreted as reflecting a change in the size of resource-access groups, which in turn is associated with increased population density and greater competition for productive resources. The multiple roomblock clusters that housed the resource-access groups early in the Pueblo III period are comparable in size to the architectural blocks that make up the late Pueblo III aggregated sites of Sand Canyon Pueblo, Castle Rock Pueblo, and Goodman Point Pueblo. Communities appear to have incorporated their smaller multiple-roomblock coresidential habitations into the structure of the larger aggregated sites (Adler 1990a, 1992c, 1994).

Environmental Archaeology

The Environmental Archaeology Program was developed as a result of the 1989 Bluff research retreat. The decision to focus on the problem of abandonment demanded a more systematic and intensive investigation of environmental data than had been undertaken to that point. Consequently, Karen Adams was hired as Director of Environmental Archaeology later that year. The goals of the new program were to (1) reconstruct past environments, (2) model agricultural productivity through time, (3) analyze faunal and floral material recovered from the excavated sites, (4) provide physical anthropological and chemical constituent analysis of human bone, (5) devise modern studies to measure the productive value of a series of key native plants, and (6) implement studies designed to provide an understanding of the formation of floral and faunal assemblages at archaeological sites (Adams 1992).

Past environments are reconstructed at the regional level on the basis of tree-ring analysis, pollen studies, and fluvial records. A study of the fluvial history of McElmo Creek indicates that a braided, aggrading stream dropped up to 3 m of sediment in Pueblo II times and that fans continued to be built during the Pueblo III period (Force and Howell 1997). Aggradation may have ceased late in the Pueblo III period, but additional studies are needed to determine when arroyo cutting began. Force and Howell (1997) conclude that entrenchment may have resulted in local abandonments, but it was unlikely to have caused the sudden, rapid abandonment of an entire valley, much less of the entire Mesa Verde region. Local paleoenvironmental reconstruction depends on studies of packrat (Neotoma) middens, as well as on the faunal and floral remains recovered from the sites. Studies of floral remains indicate that there are no major differences between present and past plant communities and that people did not have to travel long distances to obtain the plant materials found in sites (Adams 1992).

The most ambitious environmental study completed to date is Van West's use of tree-ring and soils data to model paleoclimate and agricultural productivity (Van West 1990, 1994a, 1994b, 1996; Van West and Lipe 1992). The study area for her work is a 1,470-km² area in the Montezuma Valley that includes the Sand Canyon locality. She used tree-ring data to reconstruct Palmer Drought Severity Indices (PDSI) and Soil Conservation Service (SCS) maps to categorize soils into 4-ha cells. The PDSI values and soils information were combined to estimate potential maize and bean yields under a dryland farming regime. Geographic Information System (GIS) technology was used to integrate, quantify, and visually display these data. Van West models climate for the A.D. 900-1970 period and agricultural productivity for the A.D. 900-1300 period.

Van West concludes that the potential maize yield in her study area was always adequate to support a population density of 21 persons/km², even in the driest periods. This conclusion assumes that there was mobility and that access to productive land was not restricted. Fluctuations in rainfall and soil moisture, then, do not appear to be the sole and sufficient cause of the total abandonment of the Mesa Verde region. It is possible that environmental factors other than rainfall and soil moisture contributed to abandonment. These factors might include scarcity of potable water, wood depletion, soil-nutrient depletion, animal-protein deficiency, and cooling temperatures and resulting growing seasons that were too short to produce maize.

Analyses of floral, faunal, and human skeletal remains further contribute to our understanding of the environmental factors that may have led to the abandonment of the region. In addition, modern studies have been designed to evaluate the potential yields of productive resources and the human impact on these resources. These studies include cultivation of modern experimental gardens, assessment of the quality and availability of fuelwood, evaluation of the availability of construction timbers in pinyon-juniper woodlands, and the documentation of the recovery of plant communities after a fire (Adams 1992).

Oral History

The area around Sand Canyon Pueblo and Goodman Point Pueblo is referred to today as the Goodman Point community. Modern farming in this area is entirely dryland farming; there is no irrigation. In 1989 and 1990, the Crow Canyon Center, with financial assistance from the Colorado Endowment for the Humanities and the Ballantine Family Charitable Fund, conducted an oral history project to document the homesteading of this area that occurred most intensively between 1911 and 1925 (Connolly 1992). The oral history project had several objectives: (1) understanding the criteria and techniques farmers used in selecting land for farming, (2) documenting the location of early homesteads and the identities of the families who occupied them, (3) learning about crop selection and successes and failures in farming, (4) identifying local water sources, (5) documenting the impact of modern farming on archaeological sites, and (6) recording current residents' perceptions of archaeology. This project was carried out between 1989 and 1990. Fifteen people were interviewed, eight of whom were children of the first homesteaders.

Historic farming in the Goodman Point community has taken place for just over 80 years. Much of the land was cleared in the 1911-1925 period, but dryland fields continue to be expanded today. The most productive lands were thought to be the north-facing slopes or hollows, presumably because they retain more soil moisture. A variety of crops were raised, including corn, but--with the introduction of tractors in the late 1920s--pinto beans became the most important crop. Crop yields have varied over the years, but 1951 was the only year of crop failure. Winter snows and summer rains were cited as the most important factors in producing good crop yields.

The people interviewed identified approximately 100 sites in the study area. For the most part, these sites are rubble mounds and therefore date to the Pueblo II and/or Pueblo III periods. Among the features described were an ancient road remembered by everyone and two more roads that several people recalled. Twenty-five sites were documented as having been removed during the clearing of land for agricultural purposes.

Intensive Excavations

Intensive excavation, in contrast to site testing, includes the excavation of whole structures or major portions of structures. Intensive excavations were conducted at two sites, Sand Canyon Pueblo (5MT765) and Green Lizard (5MT3901), located about 1 km apart in the central portion of the locality.

Sand Canyon Pueblo. Sand Canyon Pueblo is the largest site in the Sand Canyon locality: approximately 420 surface rooms, 90 kivas, 14 towers, an enclosed plaza, a D-shaped bi-wall structure, a great kiva, and various additional structures and features were identified during mapping (Figure 1.5). Most of the site is built inside an enclosing wall, which wraps around the head of a canyon and is thought to be one of the first constructions at the site (Bradley 1992b, 1993). The drainage divides the pueblo into east and west halves, and near its head is a spring that would have provided the inhabitants with a reliable source of water.

Intensive excavations at Sand Canyon Pueblo produced data relevant to issues raised in the overall Sand Canyon Project research design. Hundreds of tree-ring samples yielded dates that indicate that the majority of Sand Canyon Pueblo was built and occupied between A.D. 1245 and 1290, immediately before abandonment of the region as a whole. The site therefore is ideal for the study of abandonment at the site, locality, and regional levels. In addition, the results of excavation allowed us to evaluate functional differences between structures and areas within Sand Canyon Pueblo, as well as possible differences between Sand Canyon Pueblo and other contemporaneous sites in the locality and region. And finally, Sand Canyon Pueblo data, along with information gathered at other sites, were used to address questions of change in community organization (particularly the change from dispersed to aggregated communities) and degree of social complexity in the region and greater Puebloan Southwest.

The initial research design for intensive excavation at Sand Canyon Pueblo was shaped by the observation of visible surface remains, which indicated that the site was composed of 14 spatially discrete architectural blocks distinguishable on the basis of their various room-to-kiva ratios. An excavation strategy was devised to sample this variation in the composition of architectural blocks (Bradley 1992b, 1993). Architectural blocks were divided into kiva-dominated blocks (fewer than 5 rooms per kiva; N = 7), architectural blocks with standard room-to-kiva ratios (5 to 15 rooms per kiva; N = 6), and a single room-dominated block (1 kiva to 30 rooms). Architectural blocks with standard room-to-kiva ratios are most common on the east side of the site, whereas kiva-dominated blocks are most common on the west side of the pueblo. The room-dominated block, as well as the public buildings and public space, cluster on the west side of the pueblo.

Six kiva units (a kiva along with its associated rooms, courtyard, and midden) were selected for excavation because they appeared on the basis of surface evidence to constitute a sample of the types of architectural blocks present at the site ("judgmental" sample). Three kiva-dominated blocks (Blocks 100, 200, and 500), two architectural blocks with standard room-to-kiva ratios (Blocks 1000 and 1200), and the single room-dominated block (Block 300) made up this sample, which was excavated between 1984 and 1989 (Adams 1985, 1986; Bradley 1986, 1987, 1988a, 1992b, 1993; Kleidon and Bradley 1989). In addition, a stratified random sampling technique was used to select excavation units in the areas of the site where no architecture was apparent. During the 1991-1992 field seasons, excavations focused on the D-shaped structure and the great kiva (Bradley and Lipe 1990; Bradley and Churchill 1995).

The Sand Canyon Pueblo excavations produced a number of results (Bradley 1992a, 1992b, 1993). Excavations indicate that estimates of the number and types of structures made on the basis of surface remains are reasonably accurate. The majority of the site formed quickly, in the A.D. 1250s-1270s, and appears to have been built according to a general plan that began with communal construction of the site-enclosing wall. Labor estimates indicate that this wall could have been constructed in two months by 36 people, 5 percent of an estimated site population of 725 people (Bradley 1992b). Inside the wall, the site is divided into a number of use zones that correspond to the locations of architectural blocks, open spaces, and community integrative facilities.

Several excavated kivas and their associated rooms are interpreted as having specialized uses. These include Kivas 102, 107, 108, and 208, each of which is in a kiva-dominated architectural block located in the west half of the pueblo. The kiva suites associated with Kivas 501, 1004, and 1206 are interpreted as habitations. Kivas 1004 and 1206 are in architectural blocks with standard room-to-kiva ratios located on the east side of the site. Kiva 501 is associated with a standard number of rooms, but this kiva suite is surrounded by kivas with few or no associated rooms, making Block 500 a kiva-dominated architectural block. This architectural block is located on the west side of the pueblo, between the D-shaped structure and the great kiva. These habitations grew by accretion during their use. Excavation in the room-dominated Block 300 did not conclusively determine its function; it is possible that this area was designed as a community or suprahousehold storage area and was later converted to a habitation.

The Green Lizard Site. The Green Lizard site, located approximately 1 km down-canyon from Sand Canyon Pueblo, has two kivas and an E-shaped roomblock containing approximately 20 rooms. Like Sand Canyon Pueblo, Green Lizard is located adjacent to a spring. Intensive excavations at the Green Lizard site were conducted during the 1987 and 1988 field seasons (Huber 1989, 1993; Huber and Bloomer 1988; Huber and Lipe 1992). Green Lizard has two kivas, approximately 20 surface rooms, an associated midden, and various extramural areas (Figure 1.6). The kiva suite in the west half of the site was excavated, along with a stratified random sample of the associated midden and extramural areas (Huber 1993; Huber and Lipe 1992). This sampling strategy was designed to provide data comparable to the data gathered during the intensive excavation of portions of Sand Canyon Pueblo. Our goals were to address questions of chronology and structure and site function, as well as to test the various models of community organization outlined in the Sand Canyon Project research design (Lipe and Bradley 1986, 1988).

The occupation of Green Lizard is dated on the basis of pottery types, radiocarbon dates, and a few tree-ring dates. The presence of small amounts of Mancos Black-on-white pottery indicates that there was a Pueblo II occupation between A.D. 1000 and 1150. Most of the pottery dates to the Pueblo III period, and the architectural facilities and depth of the midden indicate that the site was used as a year-round habitation. The ratio of Mesa Verde Black-on-white to McElmo Black-on-white pottery is high, suggesting occupation sometime in the A.D. 1200s. Two radiocarbon samples, one from material associated with the kiva surface and another from the midden, yielded dates of A.D. 1258 and 1259 ± 40. These dates also place the Pueblo III occupation in the A.D. 1200s. The latest tree-ring date, a noncutting date of A.D. 1233, supports this interpretation and, together with the high percentage of Mesa Verde Black-on-white pottery, suggests that the site was abandoned after A.D. 1250.

Excavation of Green Lizard uncovered a kiva that is mostly earth-walled, but with masonry pilasters and masonry in the southern recess. Both masonry and jacal rooms were found in the roomblock located north and west of the kiva. Masonry rooms include Structures 9 and 2 through 7, which date to the Pueblo III occupation of the site. Wall abutments and a plugged doorway in Structure 3 indicate that the masonry roomblock grew by accretion. Structure 12 is a masonry room found beneath the floors of Structures 3 and 4. Small quantities of Mancos Black-on-white pottery were found in association with Structure 12, indicating that the room probably dates to the late Pueblo II period. Three jacal rooms (Structures 8, 10, and 13) were also found. Their construction may date to the Pueblo II period, although Pueblo III pottery refuse was found in them.

The Green Lizard midden has a maximum depth of 1.2 m. The depth of the midden and the evidence for accretional growth of the roomblock suggest a relatively long Pueblo III occupation. Comparative studies of Sand Canyon Pueblo and Green Lizard architecture, artifacts, and ecofacts indicate that there were subtle social and functional differences between the two sites, but the differences are not great enough to support a strongly hierarchical model of community organization.

Test Excavations

Test excavations at selected sites were undertaken in an effort to bridge the gap between intensive survey and intensive excavation. Survey had documented the presence of a Pueblo III community, and intensive excavations had provided abundant information about two sites, Sand Canyon Pueblo and Green Lizard. To better understand the relationships between the many smaller sites in the community and between the smaller sites and Sand Canyon Pueblo, additional excavation was needed. The Site Testing Program was designed to maximize the collection of information while minimizing the physical impact to the sites themselves. The program is discussed in detail in the following section.

The Site Testing Program

The Site Testing Program was designed as a multiple-year study, including four seasons of fieldwork (1988-1991). Testing Program fieldwork resulted in the collection of comparable information for 13 sites (Figure 1.7). These sites were tested with stratified random samples complemented by a minimum number of judgmentally located test pits (Varien 1990b, 1991; Kuckelman et al. 1991). Ten of the sites are located in the upper Sand Canyon survey area in the Sand Canyon community; three are located in the lower Sand Canyon survey area in the lower Sand Canyon community. Figure 1.8 shows a view of the Sand Canyon drainage, looking south.

Eleven of the 13 tested sites are interpreted as habitation sites. The other two sites lack substantial roomblocks, have shallow middens, and may represent minimal habitations, special-function, or limited-use sites. Sites were selected for testing because surface remains indicated that their main occupations dated between A.D. 1150 and 1300. Excavation revealed that several of the sites were also used as habitations during the A.D. 1050-1100 and A.D. 650-700 periods as well.

History of Research

Planning for Site Testing Program research took place during the winter of 1987-1988. The Testing Program research design was influenced by the most current Sand Canyon Project research design (Lipe and Bradley 1986) and by the preliminary results of other Sand Canyon Project fieldwork. The fourth season of excavation at Sand Canyon Pueblo had just been completed, and the dating of this site to the mid- to late A.D. 1200s was firmly established. The first season of excavation at Green Lizard made it clear that burned structures with accompanying large assemblages of tree-ring samples and floor-associated artifacts were not going to be found at that site. Therefore, firmly dating the small sites surrounding Sand Canyon Pueblo was identified as a key objective for the Testing Program.

The second and final season of the upper Sand Canyon survey was completed in 1987. The research goals of this survey included identifying community boundaries, documenting the distribution of sites, and investigating the shift from dispersed to aggregated communities. Crow Canyon survey crews had mapped Goodman Point Pueblo, completed the large block survey surrounding it and Sand Canyon Pueblo, and recorded the early great kiva in the Goodman Point cluster. In addition, they had documented the area of sparser settlement between the site cluster around the head of Goodman Canyon and the site cluster around the head of Sand Canyon. The results of the Crow Canyon survey confirmed a settlement pattern noted during other surveys in the region--that is, Mesa Verde phase (A.D. 1180-1300) settlement had a wide topographic distribution, extending from the mesa tops to the canyon rims, talus slopes, and benches inside the canyons (Adler 1988; Fetterman and Honeycutt 1987; Neily 1983; Hayes 1964). Gaining a better understanding of community boundaries, of the timing of the move from mesa tops to canyons, and the timing of the shift from dispersed to aggregated settlement was also a priority for the Testing Program.

Refining chronology was one of the instrumental studies critical to addressing the higher-level questions of community organization; understanding the interaction between sites requires establishing their contemporaneity or lack of contemporaneity. The results of fieldwork demonstrated that Sand Canyon Pueblo was not occupied for the full duration of the Mesa Verde phase and that the site was probably occupied only during the last 25 to 50 years of that phase. The paucity of datable tree-ring samples from Green Lizard made it difficult to firmly establish the contemporaneity of that site and Sand Canyon Pueblo. The wide range of topographic settings in which Mesa Verde-phase sites were found suggested that there was a high degree of residential mobility within communities during this period. In short, tree-ring dates were needed to establish which of the many sites with Mesa Verde Black-on-white pottery recorded by survey were contemporaneous with Sand Canyon Pueblo.

It was also clear that tree-ring dates alone would not provide sufficient data to resolve the chronological questions posed by the Sand Canyon Project research design. The brief occupation of Sand Canyon Pueblo and the potentially high residential mobility among inhabitants of the surrounding sites meant that it would be important to develop methods to determine the length of occupation of sites.

Excavations at Sand Canyon Pueblo and Green Lizard also indicated that the abandonment of the structures at these two sites differed. Understanding abandonment behavior and its effects on assemblage formation would be critical to the comparative analyses between these two sites. It was seen that further research on structure- and site-abandonment processes and how these affect assemblage formation was yet another important study.

The emphasis on community organization in the 1986 research design required studies of the scale, differentiation, integration, and intensity of community organization (Lipe and Bradley 1986:19; Lipe 1992a:4-5). It was clear that comparative samples of artifacts and ecofacts were needed for these studies. Interesting differences had already emerged in comparisons of Sand Canyon Pueblo and Green Lizard, but the intensive excavation of kiva suites took two field seasons to complete. Intensive excavation, because it is so time consuming, would limit the number of sites investigated for comparative analyses. A different strategy was needed, one that would allow us to collect data from a larger number of sites. Interpretations based on evidence from many sites would be more secure than interpretations based on the comparison of just a few sites.

The mesa-top sites, located on or near what are today the best agricultural soils, became the first priority for testing. Four mesa-top sites (Figure 1.7) were tested during the 1988 field season: Lillian's Site (5MT3936), Roy's Ruin (5MT3930), Shorlene's Site (5MT3918), and Troy's Tower (5MT3951). In 1989, excavation at Troy's Tower was completed. The testing of sites located on the talus slope also began in 1989 with the excavation of Catherine's Site (5MT3967) and Stanton's Site (5MT10508). Catherine's Site is located on a relatively level bench inside the canyon, the same setting as the Green Lizard site. Stanton's Site is located at the base of a cliff, on top of the talus slope directly above Catherine's Site. Two more talus-slope sites were excavated in 1990: Lester's Site (5MT10246) and Lookout House (5MT10459). The testing of sites in lower Sand Canyon also began in 1990. Mad Dog Tower (5MT181), Saddlehorn Hamlet (5MT262), and Castle Rock Pueblo (5MT1825) were tested during this field season. Testing at Lookout House and Castle Rock Pueblo was completed in 1991, the final season of fieldwork. In 1991, it was decided to expand the time depth of the Testing Program sample. Sites selected in 1991 had more McElmo Black-on-white pottery on the surface, possibly dating their occupations to the mid- to late A.D. 1100s, just earlier than the previously tested sites. Two of these sites, G and G Hamlet (5MT11338) and Kenzie Dawn Hamlet (5MT5152), were tested to complete the Site Testing Program sample.

Research Domains

Chronology

The primary objective of the Site Testing Program was to refine the Pueblo III chronology. The first step in building this microchronology was to recover datable tree-ring samples from the tested sites. Every piece of wood thought to have more than 20 rings was submitted for tree-ring dating analysis. Pottery assemblages from tree-ring-dated sites could then be used to develop a seriation. Our goal was to be able to recognize 20- to 25-year periods on the basis of changes in pottery attributes. The seriation could then be used to date sites from which datable tree-ring samples were not collected.

Archaeomagnetic samples were collected from every possible context that could have plausibly yielded results. Absolute dates obtained for archaeomagnetic samples are not extremely precise, but they do provide an absolute date for some sites from which no datable tree-ring samples were collected. On sites with just a few tree-ring dates, archaeomagnetic dates can help us evaluate whether the tree-ring samples were from dead or reused wood and therefore yielded misleading results. We also examined the orientation of each archaeomagnetic sample to develop a relative chronology and to assess the contemporaneity of samples. Taken together, the various dating techniques were used to make the strongest argument possible for the chronological placement of sites and to argue for or against the contemporaneity of site occupations. This improved understanding of the chronological relationships between sites provides the groundwork for studies of community development, organization, and interaction.

Length and Season of Site Occupation

An important goal of the Site Testing Program was to determine the length and season of occupation of the tested sites. On multiple-component sites, this requires determining if the occupation was continuous or intermittent, that is, separated by periods of abandonment. The estimates of the length of site occupation complement the chronological studies and provide the most accurate assessment possible of site contemporaneity. These studies are needed to refine survey-based momentary population estimates. In addition, the residential mobility of Sand Canyon locality households can be evaluated using the estimates of the length and season of site occupation. Residential mobility strategies have important implications for community integration and organization.

The method used to estimate the length and continuity of habitation site occupation is based on the simple assumption that there is a relationship between the number of people who occupy a site, the length of time that they occupy the site, and the amount of discarded material that accumulates (Varien 1990a). The length of occupation can be estimated if the population size and the total discard can be estimated. This requires determining the rate at which certain activities result in the discard and accumulation of material. Testing Program research approaches these accumulation-rate studies by estimating the total weight of cooking-pot sherds on a site. Strong archaeological cases are used to develop estimates of annual accumulation rates for cooking-pot sherds, and experimental and ethnoarchaeological studies are used to evaluate these estimates. Understanding the rate at which cooking-pot sherds accumulate helps us better understand the accumulation of other types of materials on sites. It should be stressed that the goal of these studies is not to determine the "typical" length of site occupation, but to document the range of variation in the length of site occupation.

Artifact and ecofact assemblage composition studies are used to interpret whether occupations are seasonal or year-round. Abandonment and reoccupation are also assessed through stratigraphic analysis of abandoned pit structures and kivas. Finally, an analysis of artifact density in pit structure and kiva fills is used to determine whether sites were subject to limited use after they were abandoned as habitations.

Site-Formation Processes

Site-formation processes include the depositional events, both cultural and natural, that occurred during the occupation and after the abandonment of a site. Discard behavior is a cultural formation process that is critical to estimating the length of site occupation. The behavior that accompanied the abandonment of structures and sites is another cultural process of particular interest.

More than seven centuries have passed since the tested sites were abandoned. During that time, the character of these sites has been altered by natural and cultural processes. Studies of these processes include geoarchaeological projects that focus on the abandonment and filling of pit structures, studies of hill-slope erosion, and studies that evaluate the effects of modern land use on prehistoric sites.

Abandonment also conditions assemblage formation (Cameron and Tomka 1993). Determining what happened to structure roofs and examining de facto artifact assemblages on floors are key to understanding abandonment. Different circumstances of structure abandonment have been documented at Sand Canyon Pueblo, but it is clear that much of the site was abandoned at the same time that the entire Mesa Verde region was abandoned (Bradley 1993). Comparative data on abandonment strategies from the tested sites expand our understanding of how structures and sites are abandoned and allow us to examine these processes at the larger social scale of the community. By comparing the abandonment of the tested sites with the abandonment of Sand Canyon Pueblo, we can better understand the processes by which populations aggregated and dispersed, and how and why the Pueblo peoples left the Mesa Verde region by A.D. 1300. Interpretations of site abandonment rely on accurate dating, detailed recording of site stratigraphy, examination of artifact density in the fill of abandoned structures, and assessment of floor assemblages. Theory and method developed by Lightfoot (1993), Schlanger and Wilshusen (1993), and Reid (1973) serve as models for these analyses.

Paleoenvironmental Reconstruction

Understanding the paleoenvironment is critical if we are to fully understand how and why community organization changed and why the Mesa Verde region was abandoned. This does not mean that environmental change was the prime mover, but simply acknowledges that the environment provides an important context for life in the Mesa Verde region. Evaluating paleoenvironments requires reconstructing past environmental conditions, documenting the spatial and temporal variability in those conditions, and assessing the human impact on the environment. Paleoenvironmental reconstruction allows us to ask how environmental change and resource depletion affected the inhabitants of the Mesa Verde region. Pollen and macrobotanical data are the primary data sets used to address these questions.

Community Organization and Change

Adler (1990a) has argued that communities are more than territorial entities and instead should be viewed as having a decision-making capability above the level of the primary economic units (for example, households or coresidential groups). Lipe (1992b:121) follows Murdock and Wilson (1972) and calls these "first-order" or "face-to-face" communities to emphasize that they are made up of individuals who interact on a regular basis and are therefore relatively small in terms of population size and geographic extent. Communities, like households, coresidential groups, and regional systems, are one of the basic analytical units employed in Southwestern archaeology (Wills and Leonard 1994). Lipe (1992a:4) argues that communities will be best understood by comparative analyses that focus on several dimensions of community organization. These include scale, differentiation, integration, and intensity.

Studies of length and continuity of site occupation allow us to estimate population size, which in turn improves our understanding of the demographic scale of the community. In addition, clarifying community boundaries by careful analysis of sites in both upper and lower Sand Canyon addresses the question of the geographic scale of the community.

A better understanding of community differentiation can be achieved by determining site function. Differentiation includes both horizontal differentiation, which refers to the degree to which a community is made up of functionally redundant or functionally specialized parts, and vertical differentiation, which refers to whether the components of a community exhibit any social ranking (Lipe 1992a:4). Site function can be inferred from the types of artifacts, ecofacts, architecture, and features present at each of the sites. Because inferences about site function are made on the basis of comparative analyses of these material remains, obtaining representative, unbiased samples of adequate size is critical.

Integration is related to differentiation in that it examines the interdependence of the structural units that make up the community; integration can be evaluated by examining how ideology is reinforced through ritual and through flows of information, material, energy, or people among the structural units that make up the community (Lipe 1992a:5). Comparative studies that examine architecture, artifact assemblage composition, and faunal data are used to examine community integration.

Lipe (1992a:5) defines intensity as "the amounts of population, material, information, or energy use per unit area or per capita." Reconstructing paleodemography and subsistence economies is critical to evaluating intensity. Macrobotanical and faunal data sets are particularly important in this regard.

Change in community organization is also reflected in changing settlement patterns. Refining the chronological relationship between sites and estimating site use life allow us to look at settlement patterns within increasingly refined time periods. This chronological control enables us to examine in greater detail two important changes in settlement patterns that occurred in the locality during the Pueblo III period. First, there was a shift from dispersed to aggregated community settlement patterns and an accompanying change in site size from smaller to larger. Second, there was a shift from settlement located primarily on the mesa tops to settlement located primarily in, and adjacent to, the canyons. These settlement-pattern studies address the question of whether Sand Canyon Pueblo was a center for a surrounding contemporaneously occupied dispersed community, or whether an earlier dispersed community was abandoned in favor of a more aggregated settlement, Sand Canyon Pueblo.

Data Requirements

Addressing the research domains listed above requires specific types of data from specific contexts. Sampling, both in terms of selecting sites for testing and the testing methods used to obtain collections from the sites, was an important consideration as we designed the Testing Program field research.

Site Selection

Sites were selected for testing from the pool of sites inventoried during the Sand Canyon Project surveys. These surveys indicated that there were two, and possibly three, settlement clusters with associated public architecture that may have represented first-order communities (Adler and Varien 1994; Lipe, ed. 1992). As defined by Adler and Varien (1994), these include the Sand Canyon and Goodman Point communities in the upper survey area; a third community, the Castle Rock community, may have existed in lower Sand Canyon (Adler and Metcalf 1991; Gleichman and Gleichman 1992).

Sites selected for testing are located in both the upper (10 sites) and lower (three sites) Sand Canyon survey areas. The three sites from the lower survey area include two small sites, as well as the largest site in this area, Castle Rock Pueblo. The sites selected for testing had to meet several criteria.

The emphasis on chronology was the highest consideration in selecting sites. The need for tree-ring samples biased site selection toward sites with architecture. The temporal emphasis on community organization in the period immediately before abandonment dictated selection of sites on which Mesa Verde Black-on-white pottery was the predominant decorated white ware. In the final field season, sites with McElmo Black-on-white and Mancos Black-on-white pottery were selected to increase the time depth of the sample.

Site location was also an important criterion. Several archaeologists in the Mesa Verde region have suggested that there was a move from the mesa tops to locations in, or adjacent to, the canyon during the A.D. 1200s (Fetterman and Honeycutt 1987; Neily 1983). This variation in site location is seen in the Sand Canyon locality, as documented during survey. Sites from a variety of locations were selected for testing in order to address questions about the shift in site location.

Identifying possible special-function or seasonally occupied sites was also a goal of the community organization studies. The variation in site size, site layout, and surface artifact density documented by survey might indicate functional differences between sites. The Testing Program was designed to sample some of the variation in these attributes to see if this variation represents differences in site function, season of occupation, or duration of occupation.

Other considerations influencing site selection included the degree of historic disturbance, the presence or absence of multiple components, whether or not we could secure permission to work on private land, and logistical and safety concerns for our program participants. In general, we tried to select single-component sites with a minimum of historic disturbance. However, every mesa-top site surveyed had suffered some recent vandalism, and sites that appeared on the basis of surface evidence to be single component often turned out to have more than one component when excavated. Some landowners would not permit excavation of sites on their land. Finally, we tried to select sites that were reasonably accessible, although a 30-minute hike over rough terrain was required to gain access to some sites.

These conditions made it impossible to select sites randomly from the population of survey sites. Instead, sites were selected judgmentally to test the variation documented in the survey records. The following site-location categories were designated: (1) mesa-top sites that were on or near what are today the best agricultural soils, (2) talus-slope sites that are located just off the mesa tops at the base of the Dakota Sandstone cliffs (these include sites located on relatively level benches below the upper canyon rims), and (3) sites in lower Sand Canyon.

Sites with architecture that might produce tree-ring samples fell into two categories: (1) sites containing roomblocks, kivas, and deep middens that were interpreted by survey crews as habitations, and (2) sites with a kiva and a tower, but no roomblocks and only shallow middens. Within the first category, sites ranged from those with only one kiva to those with 13 to 16 kivas. There was also variation in surface artifact density, site layout (including the presence or absence of a tower), and orientation of architectural features from site to site. Table 1.1 summarizes this information on the tested sites. More than one component is listed only when architecture from each component was encountered; not listed are components identified only by the presence of pottery. The information summarized on this table is for the latest component identified.

Multiple sites were selected from each topographic category because we anticipated conducting comparative studies. It was thought that three sites in each site-location category would enable us to test for similarities and differences within and between categories. More mesa-top sites were selected because, when we expanded the Testing Program time depth, the earlier sites were found on the mesa tops.

Troy's Tower and Mad Dog Tower are the two sites that may or may not represent year-round occupations. The absence of roomblocks on these sites indicates less investment in long-term storage facilities, and smaller middens suggest that the occupations were of shorter duration than those of some of the other tested sites. Perhaps these sites had specialized functions or were seasonally occupied. On the other hand, there was probably a wide range of household composition, and these sites might represent the minimum facilities associated with the habitation of a small household.

Sampling Strategy

The excavation sampling strategy at sites selected for testing had to meet a variety of requirements. Recovering tree-ring specimens, the highest priority, required sampling in architectural units. To address the difficult questions of the length and continuity of occupation, we turned to recent work on accumulation rates (Kohler and Blinman 1987; Pauketat 1989; Varien and Mills 1997). These studies made clear the need to devise a sampling strategy that would allow estimates of the total amount of material discarded at sites and the size of the groups discarding the material. Unbiased, representative samples were needed to determine site function and season of occupation. Structure- and site-abandonment processes were of interest; these processes can be documented by examining the stratigraphy in architectural units. Finally, the sampling strategy had to accomplish our research goals with a minimum impact on the sites.

These diverse needs were met by using a stratified random sampling technique. Each site was divided into sampling strata on the basis of observable surface remains. The sampling-stratum boundaries were drawn to enclose areas that on the surface appeared to be internally similar. That is, each sampling stratum isolated areas where we hoped the features, depth of deposits, and number of artifacts recovered would be similar from one excavation unit to the next.

The sampling design ensured that test pits would fall in specific contexts that would provide the data needed to answer the Testing Program research questions. Separate sampling strata were created for the surface architecture and pit structures, which are the most likely places to collect tree-ring samples, and for the midden, where we find the greatest number of discarded artifacts. Additional sampling strata included courtyard, inner-periphery, and one or more outer-periphery areas. In this manner, a sample of artifacts from each area of the site was obtained, a sample that could be used to estimate the total number of items that would have been recovered had we excavated the entire site. The degree to which the sample is representative of this target population can also be measured.

To select sampling units for excavation, the total number of sampling units in each sampling stratum was calculated by numbering each 1-×-1-m sampling unit sequentially. A computer program was written to randomly order the sampling units in each sampling stratum. Excavation began with the first square on the list of randomly ordered sampling units and continued down the list until the desired number of sampling units in each stratum had been excavated. The number of sampling units excavated in each sampling stratum excavated as a part of the Site Testing Program ranged from two to 12.

The surface-architecture stratum at each tested site generally encompassed areas of rubble, usually appearing as low mounds, interpreted to be the remains of collapsed surface structures. The pit structure stratum was usually defined by the presence of shallow depressions. However, on talus-slope sites, pit structures seldom leave depressions. On these sites, locations of pit structures were detected by the absence of rubble, flatter topography, proximity to surface architecture, and occasionally by an observable segment of curved retaining wall. The courtyard sampling stratum was the area surrounding the pit structure depression and bounded by the surface architecture and inner periphery. We did not create courtyard sampling strata on sites at which clear roomblocks were absent. The inner-periphery sampling stratum was defined as the area surrounding the major cultural units (surface architecture, pit structures, courtyard, and midden) at the site. On mesa-top sites, the inner periphery was characterized by a low topographic rise and a surface-artifact density higher than in any other portion of the site except the midden. The midden sampling stratum encompassed the area of heaviest artifact density and gray brown-stained soil; at mesa-top sites, the midden formed low mounds. The outer-periphery sampling stratum extended from the inner periphery to the edge of the surface artifact scatter. More than one outer-periphery stratum was designated on most sites.

Data from the stratified random sample address the research questions of the Site Testing Program. Contexts that most often contain tree-ring specimens were sampled by creating sampling strata for surface structures and pit structures. The stratigraphy exposed in architectural units permitted an assessment of abandonment processes. Test pits in structures enabled us to sample both artifactual and ecofactual floor assemblages, which contribute to determining site function and season(s) of occupation.

Stratified random sampling also provides the data necessary to estimate total discard on sites. These estimates are the basis for accumulation-rate studies, which in turn enable us to estimate the length of site occupation. By creating depositionally homogeneous sampling strata, we reduced the variance between sampling units and ensured greater precision in total discard estimates.

The stratified random sample resulted in the collection of unbiased data for intersite comparisons by testing a wide range of contexts on the site. A random sample does not guarantee representative samples, but it quantifies the degree to which the samples are representative of an entire site. This is important for site-function studies that compare variation between site assemblages.

The stratified random sampling technique achieved our research objectives while resulting in the least possible damage to cultural deposits--only about one percent of each site was actually excavated during the Site Testing Program. Details of each site's sample vary in terms of the number of sampling strata defined and the number of test pits (sampling units) in each stratum. The specifics of each sample are outlined in the following chapters, which contain individual site descriptions. Similarities among the samples are described below in the section on field methods.

Field Methods

Fieldwork began by transit mapping of surface remains at each site. Sampling stratum boundaries--drawn primarily on the basis of observable architectural features, variation in surface artifact densities, and soil color--were then added to the maps. The outer limit of the artifact scatter became the boundary of the site.

A site datum was established for mapping and for locating excavation units. This datum was arbitrarily designated the 100 north/100 east (100N/100E) grid point, which, by Crow Canyon convention, is the southwest corner of the 100N/100E grid square. This datum point is marked by rebar set in concrete. At least one additional grid point was set in concrete to ensure that one grid line could be reestablished in the future.

The grids at the mesa-top sites, and the grids at Catherine's Site, Saddlehorn Hamlet, and Mad Dog Tower, were oriented along visible wall alignments. Talus-slope grids were oriented to the cliff face and to the contours of the talus slope. The grid at Castle Rock Pueblo was oriented parallel to the butte and the contours, and the grid at Troy's Tower was oriented to magnetic north (13.3° east of true north). Grid orientation was adjusted to better coincide with, rather than cut across, room orientations and major topographic features. The specific orientation of each grid is reported in the following chapters.

A vertical datum--designated Datum A--was also established. Datum A was arbitrarily assigned an elevation of 100 m and was set on a high point at each site. The top of each rebar set in concrete was shot in as an additional permanent vertical datum point.

Sampling units at each site consisted of 1-×-1-m grid squares. These units provided an efficient means of sampling the small sites; that is, they provided an expedient method of achieving a widely distributed sample with little impact to the site. However, the small size of these units also caused problems: (1) the units often did not provide stratigraphic exposures large enough to interpret complex deposits; (2) they exposed such a limited horizontal area that it was sometimes difficult to interpret a feature or structure; (3) the need to excavate these units all the way down to undisturbed sterile sediments sometimes resulted in a unit being excavated to an unsafe depth (usually within a pit structure); and (4) when only a portion of a unit fell inside a structure, the sampling unit was further subdivided, and excavating such a small area to any great depth became logistically impossible. When any of these problems occurred, one or more judgmental excavation units were placed adjacent to the original unit in order to provide greater exposure or to provide a larger or safer work area.

Although shovels were used to remove some postabandonment pit structure fill, trowels and small, hand-held picks were the primary excavation tools. All sediment from randomly selected units was screened through one-quarter-inch mesh. This was also true for almost all judgmentally located test pits. Crow Canyon participants wrote notes on excavation forms at the end of each day. In addition, the professional staff kept a separate set of notes on every provenience investigated. For every vertical and horizontal subdivision within an excavated unit, a description of the exact provenience, excavation method, and archaeological context was recorded in both narrative and computer-coded formats. Study units, features, and point-located (individually mapped) artifacts were inventoried in separate catalogs, and specialized forms were used to record field observations and interpretations of specific features and architectural elements.

The Crow Canyon Center uses a hierarchical proveniencing system. Study units are the largest horizontal subdivisions within sites. They may be cultural--for example, a room--or they may be arbitrarily designated spaces defined for administrative purposes. Study units are numbered sequentially for each type of study unit assigned at the site. The first structure discovered is designated Structure 1, followed by Structure 2, and so on. Study units are usually subdivided horizontally. For instance, a structure might be tested through the excavation of a 1-×-1-m grid unit, and a feature within the 1-×-1-m unit might be horizontally subdivided, for example, into east and west halves. The finest horizontal subdivision is the point-location (PL), that is, the exact location on the site grid of an individual artifact or sample.

On all the tested sites, an arbitrary unit number was assigned to every sampling stratum; therefore, the terms "sampling stratum" and "arbitrary unit" are synonymous. When cultural study units were found within arbitrary units, the cultural study unit received its own study unit designation. For example, a room found within Arbitrary Unit 1 would have been designated Structure 1, provided it was the first structure found on the site.

Natural strata were used as vertical subdivisions when they could be distinguished. Exceptionally thick strata (more than 40 cm thick in pit structures and more than 20 cm thick in other study units) were subdivided into arbitrary levels. When natural stratigraphy could not be distinguished, arbitrary levels were used. Excavation in the random sampling units was terminated when undisturbed sterile sediments were exposed.

Subtle stratigraphic breaks can escape observation during excavation but become visible when a profile is cleaned and sprayed with water in preparation for drawing and photographic recording. Therefore, stratum numbers assigned during excavation often do not correspond to the stratum numbers assigned on the field profile drawing. These differences were noted on the field profile drawings.

Standard references were used for soil profile and sediment descriptions. Soil colors were described using Munsell Soil Color Charts. Standard terms for describing soil horizons, profiles, and sediments followed those found in Chapter 4 of the Soil Survey Manual (Soil Conservation Service 1981). Use of standard terms was especially important in consistently describing the texture of the sediments, the inclusions within strata, and the boundaries between strata. In addition to the objective description of each stratum, an interpretation of the depositional processes was made. These interpretations were recorded for every provenience by assigning fill assemblage position (FAP) and fill assemblage type (FAT) codes.

Multiple surfaces within a single study unit were assigned surface numbers sequentially. Materials found in direct contact with a well-defined surface were provenienced with that surface and were assigned point-location numbers. Materials found within 5 cm above the surface also were provenienced with the surface, but they were kept separate from the items in direct contact by the assignment of a separate provenience-designation number. Items found in situ in the 5 cm of sediment above the floor were assigned point-location numbers; items discovered during screening were not. Isolating the 0- to 5-cm layer was important because the origin of the artifacts contained within it often was ambiguous. Items in this layer could have been associated with the surface, particularly if debris and sediment had been allowed to accumulate during occupation; it is also possible that various cultural and natural processes caused upward displacement of artifacts that originally were in contact with the surface. On the other hand, materials in the 0- to 5-cm layer might have been deposited with sediments that accumulated after abandonment and therefore may not have been associated with the use of the surface at all. By keeping these materials separate from those in direct contact, we maintained a higher degree of control over the integrity of surfaces, which are among the most interpretively important contexts on any site.

Environmental Setting

Climate

Climate is critical to agricultural success in the Sand Canyon locality. Among the most important climatic variables are the amount of precipitation, the number of frost-free days, and the number of growing degree days (GDD). GDD is a measure of summer warmth (see Petersen [1987a:218-222] for the method used to calculate GDD), which is an important factor in crop maturation (Petersen 1987a, 1987b; Petersen and Clay 1987). Siemer (1977:20) cites 2500 GDD as necessary to mature corn. The length of the growing season is another important factor to consider when rating the agricultural suitability of an area: the adequate frost-free season for corn production is about 110 to 130 days (Hack 1942:23; Petersen 1987a:63).

The Yellow Jacket weather station (elevation 6860 ft) and the Cortez weather station (elevation 6210 ft) are closest to the Sand Canyon locality, and they record conditions similar to the high and low ends of the Sand Canyon locality study area. Table 1.2 summarizes climatic data for these two weather stations.

Petersen (1987b:225-230) discusses how climatic factors--temperature and precipitation--affect dryland farm production, and he looks specifically at how successful production can occur with GDDs below 2500. Petersen summarizes data that show that corn can be grown with 2500 GDD in areas receiving 18 inches of precipitation annually, but he then goes on to document high corn yields in cooler, dryer areas where GDD is as low as 1600 and precipitation as low as 13 inches. He finds a negative correlation between modern crop yields in the Four Corners area and GDD until he limits GDD to years below 2000 GDD. Considering only these cooler years, a correlation is obtained in which temperature explains 52 percent of the variation in crop yields.

High corn yields at GDDs well below Siemer's limit of 2500 GDD may be due in part to high elevation and low humidity. These conditions create soil and plant temperatures higher than surrounding air temperatures. Petersen further believes that soil moisture is a key to understanding dryland farm production, and that temperature (GDD) is critical in understanding the mechanics of soil moisture. Winter snows may wet soils to field capacity, but cool summers are necessary for the retention of moisture during the growing season. During cool summers, GDD is a good predictor of yield per acre, and therefore GDD represents an inverse measure of soil moisture. Petersen estimates that when summers are as cool as 1700 GDD, there is enough winter snow moisture retention through the growing season to meet the needs of corn plants. Higher GDDs produce increased evapotranspiration. As GDD increases, successful crop yields require summer rains to offset the loss of soil moisture due to warmer days. Cool summers and summer precipitation work together--afternoon showers not only provide soil moisture, but reduce air temperatures during the warmest time of the day, lowering evapotranspiration.

To summarize, in areas where the growing season is long enough, the combination of winter snows, cool summers, and summer rains are the conditions most favorable to dryland farming. Deep soil profiles that store soil moisture from winter snows are critical. The Sand Canyon locality lies within an area defined by Petersen as having this favorable mix of climatic conditions (Petersen 1987b:222, Figure 14.4).

Topography and Geology

The sites investigated during the Site Testing Program are located on the McElmo Dome, a structural uplift located north of Sleeping Ute Mountain. The sites investigated on the mesa tops surrounding Sand Canyon Pueblo are located at or near the crest of the McElmo Dome, at approximately 7000 ft. Many of the tested sites are located near the head of Sand Canyon, the main canyon that runs to the south, where it empties into the more substantial McElmo Creek. Tested sites to the north of Sand Canyon Pueblo are located near the heads of smaller tributary canyons, which run to the north and empty into Yellow Jacket Canyon. The tested sites in the lower McElmo drainage are located adjacent to Sleeping Ute Mountain, near where Sand Creek drains into McElmo Creek. The elevation at the confluence of Sand and McElmo Creeks is approximately 5500 ft.

The McElmo Dome is a nearly circular uplift containing five anticlines that enlarge the dome area to a region approximately 20 mi east-west by 10 mi north-south. The Sand Canyon locality lies entirely within this dome area. Although the origin of the McElmo Dome is uncertain, its proximity to the Ute Dome, a laccolith, suggests that it, too, may be underlain by an igneous mass. According to geologists, this uplifting may have occurred during the late Cretaceous or early Tertiary periods (65 million years ago), as it did elsewhere on the Colorado Plateau (Ekren and Houser 1965:6). Another possibility for the origin of the McElmo Dome is the movement of salt far beneath the modern surface (Ekren and Houser 1965:51-52).

At the crest of the McElmo Dome are the gently rolling highland plateaus near Sand Canyon Pueblo. Elevations of this highland range from 6600 ft to nearly 7200 ft. Deeply eroded canyons incise the dome, draining southwest-northeast into Yellow Jacket Creek and northeast-southwest into McElmo Creek. The canyons are the products of thousands of years of wind and water erosion. These relatively short drainages (Sand Creek is only 4.4 mi long) expose approximately 1,500 vertical ft of underlying geologic strata from their source in the higher plateau country to their confluence with McElmo Creek at elevations of 5500 to 5600 ft. This topographic setting creates many microenvironmental zones within a small area.

One of these microenvironmental zones is found in the lower drainages of the Sand Canyon locality. The incising action of the southwest-trending tributary canyons have formed a "canyon within a canyon" topography, consisting of an inner gorge, a lower--or first--terrace, an upper cliff face, and an upper--or second--terrace. Mad Dog Tower and Saddlehorn Hamlet are both located on the lower terrace on opposite sides of Sand Creek. Deeply incised side drainages flow into the inner gorge of Sand Creek from the east and the west. In the south faces of these minor tributaries, Puebloan campsites and permanent habitations are often found in natural alcoves formed by the chemical and physical weathering action of spring water. The most reliable water sources for these sites appear to have been natural seeps emitting from sandstone exposures (Gleichman and Gleichman 1989:69).

McElmo Creek, which defines the southern edge of the Sand Canyon locality, originates in the broad Montezuma Valley to the east, flows generally westward, and empties into the San Juan River in southeastern Utah. The walls of McElmo Canyon are as high as 1,500 ft on the north side; slopes average 10 degrees. McElmo Creek carries a substantial amount of sediment from a variety of sources and deposits rich alluvium along its banks. This alluvium today supports small but productive fruit orchards, farms, and cattle ranches.

Because the lower elevations along McElmo Creek result in warmer average summer temperatures, irrigation is necessary today. Prior to modern water-diversion practices, McElmo Creek was an intermittent drainage with reliable stream flow only during the winter-spring runoff and after summer flash floods in the highlands to the northeast (Van West et al. 1987). Today the creek flows permanently only because irrigation water diverted from the Dolores River runs into the McElmo drainage.

McElmo Creek is now incised as much as 9 m below its floodplain. Changes in the depositional regime of this drainage are dated by the archaeological remains found throughout the Holocene stratigraphy exposed in the sides of arroyos cut during the historic period (Force and Howell 1997). Force and Howell document an aggrading hydrological system during Basketmaker III times (A.D. 600-725); an erosional unconformity, or entrenchment, during Pueblo I times (A.D. 725-900); and a return to aggrading conditions and a braided main channel during Pueblo II and Pueblo III times (A.D. 900-1300).

Force and Howell examine the relationship between hydrological processes and ancient settlement in some detail. They view erosion and deposition as diachronic processes, not as abrupt, synchronic events. In their study, they estimate that stream entrenchment migrated upstream at a rate of about 20 m per year during the erosional cycle between A.D. 725 and 900, and they believe that settlement moved upstream ahead of this entrenchment. The aggrading conditions that prevailed between approximately A.D. 900 and 1300 ended with another period of entrenchment; it is unclear if the onset of this entrenchment coincided with the final years of Puebloan occupation of the McElmo area. Regardless, Force and Howell conclude that entrenchment may have resulted in local abandonments, but it probably did not cause the sudden, rapid abandonment of the entire valley or region (Force and Howell 1997).

Geologic strata exposed in the northern tributaries of McElmo Creek reveal 100 million years of sedimentation dating from the late Triassic/Jurassic through the Middle Cretaceous. In descending order, the formations are the Dakota Sandstone, Burro Canyon, Morrison, Junction Creek Sandstone, Summerville, Entrada, and Navajo Formations. Also exposed adjacent to Sleeping Ute Mountain near McElmo Creek are igneous intrusions into these sedimentary layers. These laccoliths, sills, and dikes formed as a result of the more recent Tertiary uplift of Ute Mountain. The Dakota Sandstone caps the McElmo Dome and includes tan to gray sandstones, gray to black mudstones, and a yellow brown to yellow gray sandstone or conglomerate. The Dakota Sandstone fractures naturally into tabular pieces and was the stone most commonly used for building on the tested sites. Coarser sandstones and conglomerates found at the base of this formation are suitable for making grinding tools, and relatively fine grained orthoquartzites are suitable for making chipped (flaked) tools. In addition, the Dakota Sandstone is a good aquifer, and many springs develop within the formation where the fine sandstones rest on less permeable mudstones.

Beneath the Dakota is the Burro Canyon Formation. The Burro Canyon Formation ranges from 30 to 200 ft thick and consists of a green, predominantly nonbentonitic mudstone interbedded with lenses of conglomerate and conglomeratic sandstone (Ekren and Houser 1965:18). The top of the Burro Canyon Formation is an erosional disconformity (Ekren and Houser 1965:18); the base is sometimes interbedded with the Brushy Basin Member of the Morrison Formation (Ekren and Houser 1965:19). As a result, distinguishing the contact between the Dakota Sandstone, Burro Canyon, and Morrison Formations within the Sand Canyon locality can be difficult.

The Morrison Formation lies beneath the Burro Canyon Formation. The Morrison Formation is exposed within the canyons that drain toward McElmo and Yellow Jacket Creeks. Four members make up the Morrison Formation. These members intertongue, producing variability that includes light gray to tan sandstones, red to green siltstone and clay stone, and a bentonitic mudstone that is mainly green, but is subordinately red, purple, and gray (Ekren and Houser 1965:13). The uppermost member, the Brushy Basin Member, is a bentonitic mudstone that contains interbeds of thin, resistant, fine-grained silicified sandstone and siltstone Ekren and Houser 1965:15). These were the most commonly used raw materials for the chipped-stone tools found at the tested sites. Clay sources for pottery manufacture occur in both the Dakota and Morrison Formations.

Beneath the Morrison Formation lies the Junction Creek Sandstone. Averaging 280 ft in thickness (Ekren and Houser 1965:12), the Junction Creek Sandstone forms a conspicuous cliff in the McElmo Canyon area. Below the Junction Creek Sandstone is the Summerville Formation, a bench-forming sandstone that ranges from 125 to 150 ft thick. Chert beds located in both the Junction Creek and Summerville Formations may have provided additional raw material for tool production.

Beneath the Summerville Formation is the Slick Rock Member of the Entrada Formation. The Slick Rock Member weathers into nearly vertical cliffs 70 to 80 ft in height. It consists of pale brown, very fine grained sandstone. Beneath this is the Dewey Bridge Member of the Entrada Formation, which is 25 to 35 ft thick. The Dewey Bridge Member is a brick-red, argillaceous and silty, very fine grained sandstone that forms the bench over the Navajo Sandstone. Several prehistoric cliff dwellings are located in alcoves that form at the transition between the Dewey Bridge Member and the Slick Rock Member of the Entrada Formation. Saddlehorn Hamlet is found in such a location.

The top of the Navajo Sandstone, the last formation exposed in the project area, forms the bedrock base of Castle Rock Pueblo. Above and below that site, a broad expanse of slickrock surface dips slightly (approximately 5 to 10 degrees) toward the McElmo floodplain to the south. The Navajo Formation is a pale red brown, fine-grained sandstone. Alcoves used by the Puebloans occur within the Navajo Sandstone, and stone from this formation was available for use as building material. Although the base of the Navajo Formation is not exposed in this area, the underlying Kayenta Formation is believed to be not far beneath Sand Creek (Ekren and Houser 1965:7).

The igneous exposures near McElmo Creek resulted from intrusions of magma within Sleeping Ute Mountain that cooled and were later uncovered by erosion. The magma formed a series of igneous rock types that range from microgabbro to quartz monzonite (Ekren and Houser 1965:1). These igneous rocks were used by the Puebloans as raw material for ground-stone tools (such as hammerstones and peckingstones), chipped-stone tools, and especially for pottery temper (Van West et al. 1987:15).

Soils

Soils vary within the project area, and they were an important consideration in selecting sites for testing. The deepest soils are found on the level mesa tops, where dryland farming is practiced today. The Soil Conservation Service classifies most mesa-top soils as Witt loam, a deep, well-drained silty loess, with a slope of 1 to 12 percent. Gladel Series soils often border the Witt soils in the Sand Canyon locality; Gladel soils are relatively shallow, somewhat excessively drained soils located on slopes of 3 to 25 percent. The Gladel stony fine sandy loam in the Sand Canyon locality lies on the margins of the mesa tops and is less suitable for dryland farming than is the Witt loam.

The soils associated with the talus-slope sites are in the Batterson-Gladel rock outcrop complex. These shallow soils occur on 3 to 25 percent slopes, consist of loamy sand to sand, and are well to somewhat excessively drained. They form in the calcareous residuum of the Dakota Sandstone Formation. Depth to bedrock ranges between 25 and 51 cm in typical Batterson-Gladel complex soils; however, excavations at the talus-slope sites tested by Crow Canyon indicate that deeper pockets of soil (more than 1 m deep) do occur. Stone alignments interpreted as prehistoric agricultural terracing features are present at Lester's Site, Catherine's Site, and possibly Lookout House. Sediment accumulated behind these alignments, which further increased the soil depth and suggests that the Puebloans were farming these soils.

Dolores Archaeological Project soil scientists who ranked soils for their suitability for agriculture on the basis of present-day use rated Witt Series soils as among the best (Leonhardy and Clay 1985:153). Moisture from winter snow remains in the deep profiles of the Witt loam, making them especially important for dryland farming. Gladel soils are shallow and rocky, and rangeland is the most common use today. Small areas of Gladel soils may be somewhat deep, and therefore possibly suitable for agriculture, but low moisture retention limits their agricultural potential (Leonhardy and Clay 1985:153). Other soils located near the tested sites include Ackmen loam, a fine, silty, deep, well-drained soil found in swales and drainages; the Cahona and Bowdish-Cahona complex soils that develop in the deep loess on mesa tops; and Limon Series soils that develop in fine-textured alluvium.

Flora and Fauna

Diverse physiography in the Sand Canyon locality supports equally diverse biotic resources. Partial lists of the most commonly occurring flora and fauna are presented in Table 1.3 and Table 1.4. Mesa tops have pinyon-juniper woodland with sagebrush, rabbitbrush, and many species of herbs and grasses. In general, sagebrush and rabbitbrush are dominant on the mesa tops, where soils are the deepest. Pinyon-juniper woodlands become more dominant toward the edges of the mesas and on talus slopes, where the soils are shallower. The understory includes Gambel oak, serviceberry, squawapple, bitterbrush, cliffrose, four-wing saltbush, and mountain mahogany. Also present in the area around Sand Canyon Pueblo are greasewood, wolfberry, chokecherry, Mormon tea, snowberry, lemonadeberry, and dogbane.

Plants found in the canyon bottoms and in the areas around springs include cottonwood, willow, cattail, currant, and wild rose. Canyon vegetation also includes box elder, hackberry, and single-leaf ash. Lower elevations are characterized by more xeric conditions, and consequently, yucca (both wide and narrow leaf) and several species of cactus (hedgehog, prickly pear, and cholla) become more common.

Many animal species are present in the Sand Canyon locality. Common species include desert cottontail, black-tailed jackrabbit, squirrel, pocket gopher, prairie dog, marmot, porcupine, woodrat, and mouse. Mule deer is the only artiodactyl represented. Occasional animals include coyotes, foxes, raccoons, badgers, and skunks. Bobcats, mountain lions, and black bears are present but rarely seen. Birds in the area include bald eagles, golden eagles, several species of hawk, falcons, and turkey vultures. Ravens, magpies, jays, red-wing blackbirds, doves, wrens, swallows, swifts, hummingbirds, meadowlarks, and buntings are also common. Less common, but present, in the Sand Canyon locality are tanagers, bluebirds, ducks, and teals. Reptilian and amphibian fauna include numerous species of lizards, snakes, and toads.

The above lists of plants and animals are not exhaustive for the Sand Canyon locality. The most recent and systematic survey of modern vegetation in the Mesa Verde region appears in Bye (1985) and Benz (1985), and the most complete survey of faunal species appears in Neusius (1985).

About this Publication

This volume presents detailed descriptions and interpretations of the 13 sites excavated during the Site Testing Program. Individual chapters for each site describe the specific excavation strategies employed and the stratigraphic and architectural details recorded during excavation. A chapter on material culture presents the results of pottery, stone, bone tool, and shell analysis, and additional chapters provide descriptions and interpretations of the botanical, zoological, and human skeletal remains. Synthetic chapters present a chronological summary of the tested sites, as well as discussions of settlement pattern, architecture, and abandonment on a regional scale. The final chapter summarizes the contributions of the Site Testing Program to the larger Sand Canyon Archaeological Project.

The Sand Canyon Archaeological Project: Site Testing

The Sand Canyon Archaeological Project

Site Testing

Edited by Mark D. Varien

1

Introduction