PECOS From Folsom to Fogelson: The Cultural Resources Inventory Survey of Pecos National Historical Park

The lithics of Pecos National Historical Park represent a largely unexploited source of information about the prehistoric inhabitants of the Upper Pecos River Valley. Few published studies have examined either flaked or ground stone from this area in any systematic fashion. While Alfred Kidder (1932) did present some descriptions of projectile points, manos, and metates, his volume focused on the rare items at Pecos Pueblo—although he suggests that he should have paid more attention to the changes in mano types (Kidder 1932:128). In her analysis of aggregation at Rowe Pueblo, Linda Cordell (1998) investigates flaked stone in regard to specific sites located by the survey and obsidian hydration. However, substantial discussions or analyses of changes through time in flaked stone or ground stone technology are not presented.

This chapter is a step toward filling the gap in our understanding of lithic use in the Upper Pecos Valley. We first present the parameters of the lithic data derived from the Pecos survey. We then attempt to address the affects of population aggregation and trade upon the organization of lithic technology using this data. We expect the affects of these processes to be particularly well reflected in patterns of lithic raw material utilization. Thus, considerable attention is given to the recognition of patterns that reflect variation in organizational strategies for the use of different flaked stone raw material classes. Though perhaps less affected by trade, changes in nonflaked lithic technology, particularly for ground stone, are expected with increasing aggregation. Changes are expected to be manifested in both ground stone tool form and intensity of use.

This chapter is divided into two main parts that address the flaked and nonflaked lithic assemblages from the Pecos survey. The flaked stone section begins with a review of available lithic raw materials followed by a general presentation of the variability in the data. Next a series of expectations are evaluated. The expectations concern changes in technological organization in regard to raw materials (including the results of an obsidian sourcing study), technology, and site types. The primary focus of the chapter is on the prehistoric and historic Puebloan occupation of the valley (A.D. 1200-1700). Earlier (Paleoindian, Archaic, and Developmental) sites were not abundant in the survey area.¹ Further, the majority of these earlier assemblages are obscured by those from later occupations that are more extensive. The data from these earlier sites are therefore not easily compared with the Puebloan data.

The structure of the nonflaked lithic section follows a pattern similar to that of the flaked stone. First, the variability in nonflaked lithic data from the survey is presented in a general summary, and then a specific artifact class, manos, is examined with regard to the affects of aggregation. This more detailed analysis is possible owing to the relatively large numbers of manos recorded during the survey. Other ground stone types were characterized by much lower frequencies and were not amenable to finer-grained analyses. In the final section of the chapter, we summarize the results from each of these subsections and draw some conclusions regarding the nature of lithic technology within the Upper Pecos Valley in general, as well as some specific affects of aggregation and trade.

The Flaked Stone Assemblage

The area of what is now Pecos National Historical Park is not unlike many other parts of the southwestern United States in that lithic raw materials are locally available but not terribly abundant nor of particularly high quality for lithic reduction.² Also like many other areas, sources of abundant, high quality raw material occur within a relatively short distance and often make a substantial contribution to local artifact assemblages. This geological situation, coupled with the substantial and intensive survey data recorded for Pecos, provides an opportunity to monitor the frequency and use of local versus nonlocal raw materials through time in order to identify changes in organizational strategies and to associate these changes with known socioeconomic trends (in particular, aggregation and trade).

Regional Sources of Lithic Raw Materials

With minor exceptions, the immediate area of Pecos National Historical Park is impoverished with respect to high quality lithic raw materials. However, the park is located at the juncture of three major physiographic provinces, and each of these provinces contains sources of lithic raw materials that were exploited by the prehistoric inhabitants of Pecos (Figure 9.1). Pecos lies in the extreme southern portion of the Rocky Mountain province, with the discontinuous mountain and valley terrain of the Basin and Range to the southwest and the expansive Great Plains to the southeast. The differing histories of geologic formation and exposure processes in these provinces have led to several sources of unique, and often easily distinguished, lithic raw materials. Some lithic sources, such as those of the Jemez Mountains, may have been exploited directly in some time periods, while trade surely played an important role in the use of other, more distant, sources.

Figure 9.1. New Mexico and the Southern Plains. (click on image for an enlargement in a new window)

The goal of this section is to briefly review the major sources of lithic raw materials that were available, through either direct procurement or trade, to the prehistoric inhabitants of the Upper Pecos Valley. The section is organized by physiographic province. For each province, major lithic raw material sources are described according to location of primary source (Figures 9.1 and 9.2), potential for secondary sources, physical description, and identification.

Figure 9.2. New Mexico and important raw material sources. (click on image for an enlargement in a new window)

Rocky Mountains

The Rocky Mountain province in New Mexico includes the Sangre de Cristo Range and the Jemez Mountains, as well as the southern portion of the San Juan Range. The Sangre de Cristo and San Juan are true ranges of the Rocky Mountains, created in the Rocky Mountain orogeny of the late Mesozoic and early Cenozoic Eras. The Jemez Mountains are the result of a series of Quaternary volcanic events culminating in explosive eruptions and caldera collapse about one million years ago.

Jemez Mountains Sources

The Jemez Mountains are a volcanic range that produced igneous flows and eruptions periodically from the Pliocene to the Middle Pleistocene. The obsidian sources of the Jemez Mountains are located 80-100 km (50-60 mi) northwest of Pecos (Figure 9.2); however, it is important to note that the Rio Grande and its tributaries that drain the Jemez Mountains (including the Chama and Jemez River systems) are potential sources for archeological obsidian from the Jemez as well. The raw material sources of the Jemez Mountains, particularly the obsidian sources, have received considerable attention due to their importance to prehistoric and protohistoric groups throughout the Southwest and Southern Plains. Although older igneous rock units are exposed, those formed in the Pleistocene are generally of the highest quality for lithic reduction. Of these, there are three varieties that appear to have been the most important for lithic raw material: El Rechuelos, Cerro Toledo, and Valle Grande (Figure 9.3). These varieties have been chemically distinguished through X-ray florescence (Baugh and Nelson 1987) and neutron activation (Glascock and Neff 1993).

Figure 9.3. Specific obsidian sources in the Jemez Mountain area. (click on image for an enlargement in a new window)

Obsidian from the El Rechuelos source is often referred to as Polvadera obsidian in reference to this peak, located northeast of the Valles Caldera, which is the main vent for the El Rechuelos locality complex (Baugh and Nelson 1987:318). Obsidian is moderately abundant as blocks and clasts around the base of this peak but is often limited in size by hydration and devitrification (Shackley and White 1998:4). Polvadera obsidian can often be distinguished in hand specimens by its smoky gray color and tiny ash inclusions (Head 1999:471-472).

The Cerro Toledo complex includes the Obsidian Ridge and Rabbit Mountain sources and is also available in several of the surrounding canyons. Rabbit Mountain may be the ultimate source of the material (Baugh and Nelson 1987:318). Obsidian Ridge/Rabbit Mountain obsidian ranges in color from black to brown to gray-green and is typically translucent, although it is occasionally opaque black. This variety of obsidian was used extensively for lithic raw material, and in particular Obsidian Ridge, a narrow ridge southeast of the Valles Caldera just west of Bandelier National Monument, shows evidence of intensive prehistoric collecting.

The Valle Grande complex consists of at least four igneous domes, the most archeologically important being Cerro del Medio on the eastern edge of the Valles Caldera (Baugh and Nelson 1987:318). Obsidian from Cerro del Medio is gray to black in color and often has a characteristic internal sheen on weathered surfaces (Vierra, in Head 1999:472). It is available in larger sizes relative to other obsidian sources in the Jemez Mountains (Baugh and Nelson 1987:319). Like that of Obsidian Ridge/Rabbit Mountain, the Cerro del Medio source was heavily utilized prehistorically. It appears to have been the most important source of obsidian for the Protohistoric period on the Southern Plains, presumably made available through trade with Puebloan groups (Baugh and Nelson 1987:319; Spielmann 1982, 1983), and is a common component of archeological assemblages as far north as northern Nebraska (Hughes and Roper 1999). Other studies have demonstrated the importance of Jemez obsidian tool stone for Folsom (LeTourneau et al. 1998), Archaic hunter/gatherers (Baugh 1997), and Puebloan groups (Cameron and Sappington 1984; Glascock and Neff 1993; Glascock et al. 1999; Harro 1998; Head 1999; Root and Harro 1993).

Extensive basalt and basaltic andesite flows are also associated with the Jemez volcanics, and range from relatively coarse and vesicular to useful fine-grained material. These igneous materials are common throughout the Jemez Mountains region, including the Caja del Rio Plateau just west of Santa Fe (Head 1999). This latter area lies as little as 35 km (22 mi) from Pecos (Figure 9.3). Basalt and andesite from this area undoubtedly were important prehistorically for the production of both flaked and ground stone tools.

Pedernal Chert

Pedernal chert occurs within silicified limestone beds of the Miocene Abiquiu Tuff (Banks 1990:67). The term "Pedernal chert" is traditional to archeological discussions and although the material is more accurately referred to as a chalcedony, the traditional use is retained here.³ One of the first descriptions of the material appears in Bryan, who describes it as easily worked, with few internal or natural flaws (Bryan 1939:17-18). The primary source for this raw material is often attributed to Cerro Pedernal, a flat-topped peak west of the town of Abiquiu, where it is available in relatively large sizes (Figure 9.2). However, lenses outcrop over much larger areas in the San Pedro Parks about 20 km (12 mi) to the west/southwest of Cerro Pedernal and are discontinuously exposed between these two areas. Quarrying activities are abundantly evident along the slopes of Cerro Pedernal and at least one locality in the San Pedro Parks (Bryan 1938). Pedernal chert occurs in secondary contexts in the Rio Chama, Rio Puerco, and Jemez River as well, with cobble-sized particles available along the Rio Chama (Banks 1990:69). It follows that Pedernal chert can be expected to occur in the alluvium of the Rio Grande as well, but whether it occurs inadequate size ranges for lithic reduction has not been reported (Phagan 1985; Warren 1979a:57). The closest primary sources of this material are located about 80 km (50 mi) northwest of Pecos, while secondary sources are at least 55 km (34 mi) away.

Pedernal chert is typically translucent and varies in color from white to gray with dendritic and amorphous discolorations ranging from black to red, blue, and yellow. The colors vary widely across small areas of the material and may all be visible within a single small specimen. Pedernal chert is of high quality for lithic reduction and tool production, but useful pieces are occasionally limited in size due to vugs and internal fractures. Pedernal chert was apparently a consistently favored lithic raw material in the northern Rio Grande area, and sources were exploited from the Paleoindian through the early Historic periods.

Other Sources in the Rocky Mountains

The Upper Pecos Valley is located at the southern tip of the Sangre de Cristo Mountains. Despite the large areal extent and complex geology of this range, only two formations are reported as containing sources of useful lithic raw material (Banks 1990:71-72). The Madera Formation outcrops extensively along the eastern edge of the Sangre de Cristos and contains cherts that vary in both color and quality. The color is mostly gray but may have red, orange, white, and yellow coloration, sometimes within the same piece. Quality in the same sample can grade from cryptocrystalline to poor quality, medium-grained material. Numerous outcrops occur about 28 km (17 mi) east of Pecos near Las Vegas, New Mexico, including a brittle opaque gray chert that is well represented among the artifacts from the survey. Secondary sources of chert from the Madera formation can be found in both the Pecos and Sapello rivers (Figure 9.2). Limestone associated with the Madera Formation is available throughout this same area and was noted as an occasional component of the Pecos flaked lithic assemblage. The Dakota Formation outcrops in the same general areas as the Madera and contains relatively coarse quartzitic chert, quartzites and sandstones that may have been useful for heavier stone tools (Banks 1990:90).

The San Juan Mountains begin just north of the Jemez Mountains and continue northward into southern Colorado (Figure 9.2). As with the Sangre de Cristos, few sources of useful lithic raw material are known relative to the extent and geological complexity of the range. Basalt (possibly with some associated obsidian), petrified wood, and several varieties of "jasper" are reported for the northern part of the range (Banks 1990:66). Raw materials fitting the descriptions of these sources are reported from archeological contexts in the northern Rio Grande Basin of Colorado (Spero and Hoefer 1999). Due both to the sparseness of useful raw materials in the San Juans and the distance of the range from Pecos, these mountains are not expected to have been an important source of lithic raw materials for the prehistoric cultures of the Upper Pecos Valley.

Between the Sangre de Cristo and San Juan Mountains just south of the New Mexico-Colorado border lie a series of volcanic domes. Basalt flows associated with these comprise the Taos Plateau through which the upper Rio Grande flows. Two domes that lie on the west side of the river, Mount San Antonio and No Agua Peak, have associated obsidian outcrops (Figure 9.2). Obsidian from these sources also occurs in a secondary context in adjacent streambeds (Findlow and Bolognese 1982). Despite the high quality and adequate sizes of obsidian from this area, it does not appear to have been widely exchanged beyond the Rio Grande Valley (Baugh and Nelson 1987; Findlow and Bolognese 1982).

Basin and Range

The Basin and Range portion of New Mexico consists of a series of north-south trending mountains separated by relatively wide valleys or basins (Figure 9.1). With a few exceptions, these are mostly fault-block mountains associated with Tertiary crustal tension (Chronic 1987). Relative to the adjacent physiographic provinces, the Basin and Range contains few important sources of lithic raw materials.

Madera and San Andres Formations

Due to the common origin for many of the mountains, an individual rock formation is often represented in many ranges. The Madera, first presented above as occurring in the Sangre de Cristos, is one such formation that includes economically useful chert. The opaque gray chert is found in nodules and lenses in the Manzano Mountains (Banks 1990:78) and also occurs in the Sandias to the north in smaller sizes, often with numerous internal planes of weakness, and in the Caballos to the south (Figure 9.2). In these areas the chert outcrops along the eastern faces of the ranges and grades from gray to pale blue.

A white and gray finely banded chert occurs within the San Andres Formation of south-central New Mexico. Like the Madera, the San Andres Formation is represented in numerous ranges including the San Andres, Zuni, and Sacramento Mountains (Figure 9.2). This chert, sometimes referred to by archeologists as "fingerprint chert," is available in relatively small sizes but is common on archeological sites in central and southern New Mexico.

The closest outcrops of the Madera Formation in the Manzanos occur about 100 km (60 mi) south of Pecos, and the San Andres chert of the Zunis and Sacramentos is over 160 km (100 mi) away. Owing to the distances of these sources, and the fact that chert of comparable quality to these is found locally in the Madera Formation of the Sangre de Cristo Mountains, these chert sources in the Basin and Range are not expected to have played an important role in the lithic industries of the Upper Pecos Valley.

The Rio Grande

One potential source of a variety of lithic raw materials that should not be overlooked is the alluvial deposits of the Rio Grande. Secondary sources of many of the raw materials described above can be expected downstream from their primary sources in both the current bedload and Pleistocene terraces associated with the river. It is not uncommon to encounter dispersed evidence of cobble testing along river terraces. These secondary sources of raw materials appear to have been important prehistorically as far north as Bandelier National Monument (Head 1999) and as far south as the Tularosa Basin, where Jemez obsidian pebbles and cobbles were extracted from the Pleistocene Santa Fe Formation (Carmichael 1986). The Rio Grande provides a closer source of some important materials (such as those of the Jemez Mountains) to the inhabitants of the Upper Pecos Valley. It is difficult, however, to evaluate the degree to which these secondary sources were exploited prehistorically.

Mount Taylor

Mount Taylor is a volcanic range located just north of Grants, New Mexico (Figure 9.2). Obsidian derived form this Tertiary volcano occurs on the mountain itself as well as on the surrounding foothills (Findlow and Bolognese 1982:56) and among the alluvium of the Rio Puerco to the southeast (Shackley 1998). Mount Taylor obsidian is typically black and opaque, occasionally with small gray flecks. Although it is not actually located within the Basin and Range province, but on the edge of the adjacent Colorado Plateau to the west, it is included in this section because secondary sources along the Rio Puerco occur in the Basin and Range and are here closest to Pecos. Mount Taylor is located 160 km (100 mi) from Pecos, with the secondary alluvial deposits occurring within 110 km (68 mi).

Great Plains

To the south and east of Pecos lie the Great Plains. The portion of the plains adjacent to New Mexico is a remnant Tertiary landform occasionally interrupted by erosional and volcanic features (Trimble 1980). Erosional features, including the Pecos, Canadian, and other smaller rivers, have served to expose underlying bedrock that includes important sources of lithic raw material. Despite relatively great distances, some of these materials are fairly common in the archeological record of Pecos National Historical Park.

Alibates Flint

Alibates flint, more accurately an agatized dolomite (Holliday 1997:245), is one of the most widely known lithic raw material sources in North America.⁴ It occurs within the Permian Quartermaster Formation as lenses and nodules in relatively massive ledges and residual blocks (Tunnell 1978:7). Outcrops of Alibates, along with associated quarries, are best known from the area now preserved as Alibates Flint Quarries National Monument but are found along the north and south sides of the Canadian River in several locations northeast of Amarillo, Texas (Figure 9.1). This material is found in secondary contexts east of the outcrops along the Canadian River (Banks 1990:91).

The Alibates material is characterized by bands, speckles, or mottles of a variety of colors including white, red, brown, purple, and gray. Multiple colors are typically represented on a single small specimen. The material is very well suited for lithic reduction but often contains large vugs with quartz crystals (Tunnell 1978:7). The prehistoric desirability of Alibates, perhaps due to its beauty as well as its technical qualities, is indicated by its recognition in archeological contexts as far away as northern Colorado (Stanford and Jodry 1988). Similarly, although it occurs more than 350 km (220 mi) east of Pecos, Alibates is not uncommon in the archeological collections from there.

Tecovas Chert

Tecovas chert, commonly referred to as Tecovas or Quitaque jasper due to its iron-stained color, occurs in the Tecovas Formation of the Triassic Dockum Group. The Tecovas formation outcrops as the lowest exposed unit on the northwestern, northern, and eastern edges of the Llano Estacado, including a portion of the Canadian River Valley near Alibates National Monument. Outcrops along the northwestern edge of the Llano Estacado occur about 160 km (100 mi) from Pecos (Figure 9.1). A lateral equivalent of the Tecovas formation (Holliday 1997:249), known as Baldy Hill, occurs across the Canadian River drainage about 15 km (9 mi) east of Folsom, New Mexico.

Both the Tecovas and Baldy Hill materials are problematic in that they can be difficult to distinguish from Alibates in hand specimens, although the former are not banded and are not typically mottled. Tecovas is typically a fairly uniform dark red to yellow color with abundant small vugs with quartz crystals. It can be distinguished from Alibates under low magnification by its greater opacity, smaller and bluish quartz crystals, and less waxy luster (Holliday 1997:249). It has also been reported that Alibates is available in larger size ranges (greater than 8 cm) than Tecovas (Banks 1990:93), although boulder-sized specimens of Tecovas have been reported (Boldurian and Cotter 1999:25).

Edwards Plateau Chert

Another widely recognized and widely used source of chert is the Lower Cretaceous limestone of the Edwards Formation. "Edwards Plateau chert" actually refers to any of several sources of chert that outcrop and were occasionally quarried (excavated) along the edges and erosional features of the Edwards Plateau in central Texas. Banks (1990:60) argues that at least some geographic distinction can be based upon color.

Edwards Plateau chert occurs in lenses and nodules, the largest of which can exceed 40 cm in diameter (Tunnell 1978:7). The material ranges from dark brown to gray to blue through white in color and is characterized by white or cream colored cortex. The chert is typically fine grained and is very well suited for lithic reduction. Sources of high quality Edwards Plateau chert occur about 480 km (298 mi) from Pecos. Secondary sources occur mainly to the east and south of the plateau and therefore do not provide any closer alternatives. Despite the distance, Edwards Plateau chert is recognized from archeological assemblages within Pecos National Historical Park.

Other Sources on the Great Plains

Two areally extensive but more heterogeneous sources of lithic raw material in the southern portion of the Great Plains deserve mention. The Ogallala Formation is a Tertiary alluvial unit that underlies all of the Southern High Plains (Holliday 1997:250). The Ogallala was formed as outwash from the Rocky Mountains to the north and west and therefore contains a wide variety of lithologically distinct sediments. Numerous varieties of quartzite occur, including a common medium-grained purple quartzite, as well as several varieties of chert ("jasper") and opal (Banks 1990:95). These materials occur in gravel to cobble size ranges and occur along all of the escarpments of the Llano Estacado.

The Dakota Formation, described above in reference to the Sangre de Cristo Mountains, discontinuously outcrops from Tucumcari, New Mexico north to Nebraska. Large amounts of a wide variety of quartzites occur in the Dakota Formation in the vicinity of northeastern New Mexico. In places the quartzite has metamorphosed to chert, and at least one such source near Fort Union National Monument is reported to show extensive evidence of prehistoric quarrying (Banks 1990:90). Both the Ogallala and Dakota Formations, and the materials within them discussed above, occur less than 100 km (60 mi) from Pecos. The great variety of materials found within them, and the difficulty in differentiating them from other materials, makes it difficult to evaluate the extent to which they are represented in archeological assemblages.

Summary

Differentiation between what can be considered local versus nonlocal raw material is typically a subjective judgement based upon assumptions of group mobility and land use. Fortunately for our purposes, the location of Pecos is such that there is a limited number of raw material sources that are immediately available on the landscape. The variously colored cherts, and possibly chalcedonies, of the Madera formation and the sandstone and quartzite of the Dakota formation outcrop in the hills and canyons of the Tecolote range within the park and the surrounding area. Primary deposits of these occur within a 30 km (19 mi) radius of any part of the park (Table 9.1) and are available as secondary deposits in most major streambeds.

The group of raw material sources considered here to be nonlocal includes those of the Jemez Mountains (including Pedernal chert), the Rio Grande, and the Ogallala gravels. These sources range from 40-100 km (25-60 mi) from the Pecos (Table 9.1) and therefore require considerable travel for transport into the valley. The remaining sources, Mount Taylor, Tecovas, Alibates, and Edwards, occur at distances greater than 160 km (100 mi) from Pecos and are considered here to be truly exotic. With the possible exception of the Paleoindian Period (Amick 1996), the transportation of lithic raw materials over this distance likely represents purposeful and planned movement of goods in the form of trade. Distances to individual sources from Pecos are presented in Table 9.1. Additional sources described above are less easily recognized in the assemblage and are typically assigned to an "Other" or "Undifferentiated" category in the following analyses.

Table 9.1. Distance to major raw material sources.

Flaked Stone Assemblage Parameters

Flaked stone artifact data were collected during the Pecos survey in order to record information relevant to both technology and function. The majority of the data was recorded in the field in an attempt to maximize sample size while minimizing both the impact to site integrity and the curation costs associated with collection. Twelve basic attributes were recorded for 21,088 items, resulting in 253,056 individual data points. A coding manual (Appendix C) was used in the field to ensure the consistent measurement of the twelve attributes.

The attributes consist of material type, raw material color, amount of cortex, platform type, condition (debitage type or tool completeness), thermal alteration (heat treatment), edge damage (use wear), technological type, functional type, and size (length, width, and thickness). Raw material type and color enable measurement of the frequency and relative diversity of stone tool resources in a given assemblage. Coupled with information regarding the natural distribution of these sources (presented above), raw material data provide a basis for inferring social interaction and landscape use among prehistoric groups. Cortex, platform type, condition, thermal alteration, and size are attributes capable of informing on lithic reduction represented by a given specimen. An understanding of the level of reduction is important for inferring general reduction strategies, including core maintenance and degree of material conservation, as well as tool manufacture, use, and maintenance. The remaining attributes, edge damage, technological type, and functional type, provide information on tool use and may reflect specific activities and tasks carried out on a site or within a given area.

Each of these attributes is presented in more detail below. Specific details concerning their measurement are available in Appendix C. Patterning in selected attributes is highlighted to provide a general description of the nature of the overall Pecos survey assemblage. While random, systematic, and judgmental sampling methods were used in the field to collect lithic data (Appendices B and C), data descriptions and analysis, unless otherwise noted, are based upon the random samples only. Finally, a simple tool classification is analytically derived based upon the attributes listed above. The analytical classification is used in the subsequent investigation of the organization of flaked lithic technology.

Raw Material Type

Material type refers to the lithic raw material from which an artifact was made. This attribute was coded in such a way that various levels of specificity were available to ensure accurate field recording. Thus one might record a material type as "Sedimentary, undifferentiated," "Chert, undifferentiated," or "Pedernal Chert" depending upon the level of confidence one had in recognition of specific type. Raw material color was recorded to help in the later identification of and distinction among types. This was especially important for varieties of chert and chalcedony.

The material type data are summarized in Table 9.2. This summary describes an assemblage that is dominated by local raw materials but has a specific, and seemingly patterned, array of nonlocal and exotic varieties. Local chert and chalcedony together make up just over 80 percent of the total sample. Quartzite, comprising 5.5 percent of the sample, is probably of mostly local origin as well. The remaining 14 percent can be considered nonlocal or exotic. Obsidian is the only nonlocal material that contributes a substantial portion, 9.1 percent, to the sample. For the Pecos survey, obsidian identification was based on visual criteria and comparison with a type collection. X-ray fluorescence characterization of a small sample of items provides specific identification of sources and is presented later in this chapter. Other nonlocal and exotic raw materials, including Pedernal chert and Alibates flint, contribute less than 1 percent each. The final 3 percent were classified as "Other" due either to an inability to confidently associate them with a specific source or a very low frequency. For example, the very few items that were classified as Edwards chert (n=22; 0.1 percent), are subsumed within this category.

Table 9.2. Flaked stone raw materials.

Size

Length, width, and thickness were recorded for each flaked lithic item. These size measurements were recorded in millimeters using calipers. The definition and technique of measuring maximum dimensions can be found in Appendix C.

Cortex

Cortex is the natural outer surface of a piece of raw material and may be primary, a reflection of the bedrock matrix in which the material formed, or secondary, a result of weathering due to sediment transport. Cortex amount and condition varies according to distance from the primary source of the raw material and the item's stage in a reduction sequence. It is increasingly realized that traditional stage typologies relating cortex amount to reduction stage are overly simplistic (Sullivan and Rozen 1985; Shott 1994). For example, experimental research indicates that flakes removed near the completion of a trajectory from generalized core to biface may still retain some cortex (Shott 1996). However, in a general sense, larger amounts of cortex can be expected to characterize the earliest flakes removed in a reduction sequence, and cortex can thus be seen as an imperfect but useful correlate of level of reduction.

Typically, in measuring cortex on debitage, only an item's exterior, or dorsal face, is examined. In order to effectively record information about such artifacts as tested cobbles and large choppers that do not really have an exterior face, the Pecos survey analysts included the entire surface area of an artifact in calculating the amount of cortex. Thus a piece recorded with 91-100 percent cortex has cortex on 91-100 percent of its total surface area. Few flakes have more than 50 percent cortex, but exceptions include flakes with cortical platforms and pronounced curvature, and angular debris. Cortex amount was recorded in relation to five ordinal percentage classes: 0, 1-25, 26-50, 51-90, and 91-100 percent. These classes were chosen to provide enough detail to be meaningful, while remaining general enough to facilitate reliable measurements in the field. The amount of debitage with cortex for raw materials are presented in Table 9.3.

Table 9.4 presents the breakdown of cortex amounts for all debitage. The overwhelming majority of debitage (81 percent) exhibits no evidence of cortex. Only 1.2 percent of debitage exhibit more than 50 percent cortex. Thus, very early stage reduction is poorly represented in the overall assemblage, indicating that this aspect of reduction typically took place elsewhere (perhaps at the locus of acquisition). The implications of relative differences in cortex occurrence for different raw materials are explored later in this chapter.

Table 9.3. Material types of debitage with cortex.

Table 9.4. Cortex amounts for debitage.

Platform Type

Many believe that platform attributes give the most reliable indication of the degree of reduction prevalent in a lithic assemblage, which also can help identify the range of activities the assemblage represents (Callahan 1990; Dibble 1997; Odell 1989; Whittaker 1994). Further, along with exterior flake attributes, platform morphology provides a view of a portion of the core, thus providing information concerning core state at a particular level of reduction (Teltser 1991). Platform type (technically the platform remnant remaining on a flake after it is detached from a core) was recorded for all flaked lithic items examined during the survey. The recording system allowed one or a combination (up to three) of the following types: not applicable/absent, single faceted/flat, multifaceted, abraded, cortical, retouched, crushed, collapsed, lipped, and unknown. For items lacking platform remnants, such as distal flake fragments or cores, platform type was recorded as not applicable/absent.

Table 9.5 presents the frequencies for platform types on all debitage. The largest class of the material (47 percent) had no discernible platform. Missing platforms indicate that the debitage was broken either in manufacture or by post-manufacture processes. Of the discernible platforms, the vast majority were single faceted, representing over 30 percent of the total assemblage. Only about 9 percent of the debitage exhibits two or more platform facets. These patterns, along with the frequency of collapsed platforms (almost 5 percent), are expected for debitage resulting from generalized core reduction strategies, where the production of flakes (as opposed to the shaping of the core) is the goal.

While the recording system allowed for multiple entries, only a little over 2 percent of the total pieces exhibited multiple platform types. Examining the database, a large number of the "multifaceted with other" category include lipped platforms. In experimental studies, lipped platforms are shown often to be associated with a high angle of force and a soft indentor (such as antler) (Cotterell and Kamminga 1987; Crabtree 1982). The use of a soft indentor at a high angle of force is appropriate for reducing relatively thin edged cores such as bifaces. Thus, only a small percentage of the sample fits the profile that would be expected to result from formalized (e.g., bifacial) core reduction.

Table 9.5. Platform types of flaked stone.

Debitage Type/Condition

The field recorders employed the debitage typology developed by Sullivan and Rozen (1985) for the collection of data. This formal approach provides the advantage of a systematic framework for collecting reliable (replicable) data on lithic debitage (Prentiss 1998). The framework is based upon flake completeness, and its use results in debitage categories whose proportions within an assemblage are argued to correspond to specific reduction modes. The validity of the proposed correspondence between flake completeness and mode of reduction is discussed further below.

The Sullivan and Rozen typology is implemented as follows. Debitage is categorized according to a hierarchical decision tree into one of the following debitage types: a complete flake, a broken flake, a flake fragment, or angular debris/shatter. The decision tree for debitage classification is based upon technological attributes presented in Figure 9.4. The following steps are used for debitage classification:

Figure 9.4. Flake attributes used in debitage classification.

1. Is there a single interior surface?
No=debris,
Yes=continue to question 2.

2. Is the point of applied force present?
No=flake fragment,
Yes=continue to question 3.

3. Are the flake margins intact?
No=broken flake,
Yes=complete flake.

Sullivan (1987:Figure 2) later added a step between steps 2 and 3:

2b. Is there a sheared axis of flaking?
Yes=split flake,
No=continue to question 3.

In this same article, Sullivan argued that a high incidence of split flakes corresponds to bipolar reduction. Although our recording strategy relied upon Sullivan and Rozen's (1985) original formulation of the typology and therefore omitted this "split flake" category, field analysts often included this category within the comments. It should be noted that, in order to be tallied as debitage, the lithic item could not show evidence of use. Because only about 3 percent of flakes and angular debris showed any evidence of use, the effect of their exclusion is expected to be negligible for the debitage analysis. Used flakes are discussed in a following subsection.

The Sullivan and Rozen classification scheme has had the positive effect of inspiring systematic experimentation toward defining the quantifiable relationships between debitage variability and mode of reduction (Shott 1994:78). The results of this experimentation have identified some specific validity problems with its use as an inferential tool (Amick and Mauldin 1989; Ensor and Roemer 1989). Specifically, it has been demonstrated that raw material (Amick and Mauldin 1997), flake size (Prentiss 1998), and flintknapper experience play major roles in the variability in flake completeness, regardless of reduction mode. As a data collection tool, it has been demonstrated to provide comparability among different data collectors (Prentiss 1998) but has limited utility as a tool for inference. In a well-constructed test of the reliability and validity of the Sullivan and Rozen typology, Prentiss (1998) found its fundamental limitation as a tool for inferring reduction mode to be its simplicity. Simplicity and replicability make the system very useful for data collection and organization, however, and provide a foundation upon which to incorporate additional data for forming inferences regarding lithic reduction.

Debitage class frequencies and relative percentages for the Pecos survey are presented in Table 9.6. The single largest category (41 percent) is that of "flake fragment" (medial or distal portions). Complete flakes are fairly common and comprise almost 30 percent of the sample. Broken flakes (proximal portions) and angular debris each make up less than 20 percent of the sample.

Table 9.6. Debitage classes.

Raw material appears to contribute to the overall pattern seen for debitage recorded at Pecos. An exploration of specifically how raw materials relate to flake completeness in the Pecos survey data is presented in Table 9.7. In this table raw materials are cross-tabulated with debitage class. When compared to expected debitage category proportions for each raw material type derived from Table 9.6, specific patterns become apparent. More vitreous materials such as obsidian produce relatively more flake fragments while coarser materials such as quartzite and siltstone produce comparatively more complete flakes. These results support experimentally derived arguments (Amick and Mauldin 1989; Prentiss 1998) that raw material plays an important role in patterns of flake completeness.

Table 9.7. Cross-tabulation of material type by debitage class.

Due to the limitations of the Sullivan and Rozen typology for inferring reduction mode cited above, we combine debitage category with platform characteristics to create classes for the exploration of variability in debitage as it may relate to reduction mode. Complete and broken flakes that displayed single, cortical, or split platforms, along with angular debris form a fairly internally consistent debitage class in the Pecos survey sample (Class 1). These attributes are expected to relate to generalized core reduction (Teltser 1991). Alternately, complete flakes and broken flakes with one of the following platform types: multifaceted, multifaceted with other (abraded, retouched, lipped, or crushed), collapsed platforms, or single faceted with other (abraded, retouched, lipped, or crushed) comprise another class (Class 2). These attributes are expected to be associated with formal reduction (e.g., bifacial or formal tool). Unknown/ambiguous debitage (Class 3) includes all flake fragments and angular debris without an observed platform. Table 9.8 presents the totals for each of these debitage classes. These results are comparable to those derived from the examination of platform alone, suggesting that the overall Pecos survey sample is dominated by generalized core reduction debris.

Table 9.8. Debitage classes derived from post-field analysis.

Thermal Alteration

All flaked lithic artifacts were examined for evidence of thermal alteration, which may be intentional or unintentional. Intentional thermal alteration (heat treatment) is known to increase the quality of some raw materials for lithic reduction. Although the physical process of thermal alteration is poorly understood (Luedtke 1992), it is believed that the application of heat to a siliceous material may (1) strengthen the mineral bonds through silica fusion, (2) create microscopic fractures that allow for easier fracture propagation, or (3) change the amount of water contained within the crystalline structure of the stone. The result is an increase in the ease with which a material is fractured during the flintknapping process. Sedimentary (e.g., chert, siltstone, sandstone) and possibly some metamorphic (e.g., quartzite) raw materials respond particularly well to intentional thermal alteration. Knowledge and use of this process allows flintknappers to more accurately control and more easily flake certain raw materials. The presumed goal is to alter the stone such that it is easier to reduce, either by free-hand percussion or pressure flaking.

Not all thermally altered artifacts are the result of successful or even intentional heat treatment. Unintentional thermal alteration may result from discard into a thermal feature or from natural fires. Unsuccessful heat treatment often results in flaws that make flaking more difficult to control. Potlid scars, circular conchoidal depressions, result when water within the stone is heated too quickly. A crazing pattern, a network of fine cracks resulting from heat stress, results when stone heats and cools too quickly or is raised to a temperature that exceeds the limitations of the mineral bonds. Oxidation may in some cases cause discoloration of the raw material's natural state. Differential luster is an indication that the material was properly heat treated and subsequent flaking has exposed the visually altered inner portion of the chert.

Lithic analysts noted that 809 pieces (3.8 percent) of flaked stone out of 21,088 total recorded pieces from the Pecos survey had evidence of thermal alteration. Four types of thermal alteration were recorded—crazing, potlidding, luster, and discoloration. Criteria for the identification of these attributes are presented in Appendix C. All alterations identified on a single piece were recorded. Luster was the most frequently recorded type, with crazing the next most recorded (Table 9.9).

Table 9.9. Thermal alteration for all flaked stone.

In order to address the role that heat treatment played in reduction strategies at Pecos, we examined the sizes of lithic items subjected to intentional thermal alteration. Luster is used here as an indicator of successful and intentional thermal alteration. When compared with unaltered flaked stone objects using a t-test, thermally altered stone is found to be significantly thicker (Table 9.10). That all thermally altered items tended to be large suggests that thermal alteration was attempted relatively early in the reduction process before great energetic investment was made in a particular item. Further examination of thermal alteration as it relates to aggregation and trade at Pecos is discussed later in this chapter.

Table 9.10. Student's T-comparison of thermal alteration by thickness.

Edge Damage

Edge damage for this survey is defined as damage that can be seen with either the naked eye or a 10x-power hand lens. This technique allowed the identification of microflaking, edge rounding, polish, striations, battering, and edge frosting. Up to three of these edge damage patterns in any combination were recorded for each flaked lithic item.

The interpretation of macroscopic edge damage as resulting from human use is problematic. Noncultural processes may also be responsible for edge damage patterns. Several geological (e.g., colluvial and alluvial) and biological (e.g., trampling) processes may cause damage to flaked stone objects that could be mistaken for use-wear due to human implementation. Because most of the park was once a cattle ranch, lithic artifacts are expected to be particularly susceptible to microflaking as a result of animal trampling. Presently, there is no simple way to quantify the impact of noncultural processes on artifacts. The alternative adopted here is to proceed as if all recorded edge damage represents use-wear, with the assumption that noncultural processes have an equal chance of affecting any individual item and so do not obscure patterning due to human use. Thus, as is further discussed below, all flaked lithic items exhibiting some form of edge damage are considered tools.

Table 9.11. Flaked stone use-wear.

Table 9.11 presents the recorded edge damage from the Pecos survey. Over 94 percent of all flaked stone exhibited no edge damage. Edge damage in the form of microflaking was recorded for 3.7 percent of the items and constitutes the most abundant class. Microflaking with edge rounding, edge rounding, and battering are the next most common classes, each constituting less than 1 percent of the sample. Other edge damage classes each make up less than 0.04 percent of the sample.

In our examination of the data, edge damage appeared to be more common on larger flaked lithic items. A statistical examination was carried out to evaluate the null hypothesis of no difference in size between flaked stone items exhibiting edge damage and those exhibiting none. Significantly different means for edge damaged and undamaged items were found using volume (length x width x thickness) as a measure of size (Table 9.12). Thus, large flakes and tools exhibit more evidence of use. The reasons for this are not clear, but in addition to the fact that edge damage is more likely to be identified on larger items, it seems probable that larger flaked stone items are easier both to hold and to haft.

Table 9.12. Student's T-comparison of item volume by edge damage.

Technological Type

Each flaked lithic item recorded by the survey was assigned to one of 20 technological types. Technological types are conceived of as categories that are relatively free of functional interpretation (see functional type below). Technological types consist of six fundamental categories: angular debris, core, flake, unidirectional edge/uniface, bidirectional edge/biface, and tested cobble. Each of these categories is further subdivided into more specific classes based upon thickness or directionality. An "other" category is provided for items that do not match any of the above categories. Definitions for each of these technological types can be found in Appendix C.

Like other technological attributes such as platform or debitage type, the frequency of technological types provides a profile of flaked stone technology in the survey sample. The frequency of technological types within both the random sample and the combined random and judgmental samples is presented in Table 9.13. The following summary is based upon the random sample. Flakes and angular debris account for over 90 percent of the 19,970 items. Biface thinning flakes are relatively rare, accounting for just over 1 percent of the sample. Alternately, bifaces themselves are the fourth most common technological type, representing 1.61 percent. This difference in the frequencies of biface thinning flakes and bifaces is somewhat unexpected, considering that the manufacture of a single biface is expected to have produced a number of flakes. However, it is recognized that bifacial reduction may proceed by other means than thinning, such as the removal of relatively short and thick flakes from both sides of a core. Cores constitute 1.8 percent of the total assemblage. Of these, multidirectional cores are by far the most common. Unidirectional and bidirectional edged items, unifaces, and cobbles each account for less than 1 percent of the random sample.

Functional Type

In contrast to technological types, functional types are inferential categories that also carry information on artifact morphology. As with technological types, every item is assigned a functional type. These categories are based upon traditional interpretations of artifact use, which are drawn variously from ethnographic analogy, experiment, and assertion. These categories are listed, along with their frequencies within the random and total samples, in Table 9.14. The identification criteria for each of the functional types are listed in (Appendix C). The frequencies of individual functional types within the random sample provide a means of understanding the range of activities represented in the Pecos survey assemblage.

Approximately 6 percent of the recorded items in the random sample show evidence of use or modification into a shape suitable for use. The most abundant functional type is that of used flake (2.7 percent), defined as a flake that has not been formally modified but shows evidence of use wear or damage. A preference for flakes over angular debris for use as expedient tools is evidenced by the relative rarity of used angular debris (0.28 percent). The most frequently recorded formed tools are projectile points, which constitute 0.8 percent of the random sample. Scrapers, when the four varieties are combined, make up 0.67 percent of the recorded items, and blanks and preforms make up 0.55 percent. Other functional types each constitute less than 0.5 percent of recorded items.

A total of 627 projectile points was recorded, constituting 2.97 percent of the total sample. All projectile points observed during survey were recorded. Functional codes were assigned to projectile points differently from other artifacts in order that information concerning size and morphology could be simultaneously collected (Appendix C). Projectile point classes and their frequencies are presented in Table 9.15. Based upon the largely Puebloan prehistoric occupation of Pecos National Historical Park, it is not surprising that the best represented categories are varieties of small notched points. Small triangular points are also well represented. Such projectile points have long been argued to be a component of bow and arrow technology (Bettinger and Eerkens 1999; Fenenga 1953) and likely relate to hunting and defense activities.

Table 9.13. Technological types.

Table 9.14. Functional types.

Table 9.15. Projectile point classes.

The patterns in the frequencies of functional types in the random sample indicate that the simplest of expedient tools (utilized flakes) are by far the most common class of tools. They are more than three times more abundant than the next largest functional class, indicating that many tasks requiring stone tools were carried out with minimal investment in tool manufacture and maintenance. Conversely, the next three most frequent functional classes are composed of relatively formal tools—projectile points, projectile point blanks and preforms, and scrapers. It thus appears that hunting was prominent among activities requiring stone tools and that greater technical investment was afforded hunting and game processing equipment. These ideas concerning technical investment and the importance of hunting are supported by more detailed investigations presented below.

Tool Classes for Analysis

The technological and functional type data classes for the Pecos Survey resulted in many different tool types. To further examine changes in tool production and use in relation to raw materials and activities (presented in the following sections), the large number of types required some reduction. Tool classes are derived through a combination of technology and inferred functionality.

The majority of flaked lithic artifacts appear to be unspecialized forms suitable for expedient use. Specialized and unspecialized tool forms are characterized by differing energy investment requirements, and their production and maintenance are expected to have been organized differently. Tool classes for analysis are constructed with energy investment in mind. Changes in the treatment of tools belonging to these classes are specifically discussed later in this chapter.

Tool categories developed for the analyses include cores, expedient tools, low-energy tools, and high-energy tools. Cores are pieces of raw material that have smaller pieces detached from them but have no functional purpose other than as a source of flakes. This class includes the four different types of cores discussed above—unidirectional cores, multidirectional cores, bidirectional cores, and undifferentiated cores—with the exception of any that show evidence of use.

Expedient tools include all debitage and cores that have evidence of use and include such functional types as used flakes and used cores. Expedient tools do not exhibit any modification for a specific use. Low-energy tools exhibit shaping for specific tasks but do not represent a large expenditure of time or energy into production. This tool category includes, for example, unifacial end scrapers, cobble tool choppers, and bidirectional edged cutting tools. High-energy tools require a relatively greater amount of time, energy, and skill to produce. These include all bifaces regardless of function—examples include small, side-notched projectile points and bifacial drills and knives. Table 9.16 presents a breakdown of each of these tool classes.

Table 9.16. Tool classes derived from post-field analysis.

The tool classes defined here are used in combination with the debitage classes defined under the debitage typology for the evaluation of the expectations that follow. All tool and debitage classes are mutually exclusive—an expedient tool cannot also be core reduction debris. While this presents problems, especially with regard to cores, exclusivity allowed for more control of the analyses presented in the following section. Single pieces would not be counted twice (for example once as cores and again as expedient tools). This enabled artifact class percentages to be used in the analyses that follow, broadening the range of appropriate statistical techniques.

The Organization of Flaked Stone Technology

The organization of lithic technology is driven by two fundamental constraints: the availability of raw material and, to a lesser degree, effective mobility (Andrefsky 1994). As each of the Puebloan periods in Pecos is characterized by a semi- to fully-sedentary mobility strategy, the focus here is upon changes in the availability of raw materials. Raw material availability is governed by geographic locations of sources and the social conditions that affect individual access to raw materials (social organization, mobility, trade relations, subsistence economy). Using our understanding of natural raw material distributions (presented in the beginning of this chapter) as a baseline provides a way in which to monitor changes in social conditions through changes in technology.

In the following sections we derive some expectations in regard to the research objectives for the Pecos survey. Using the data derived from survey and site recording in the park, we attempt to address the affects of population aggregation and trade upon the organization of lithic technology. The affects of these processes are particularly expected to be reflected in patterns of lithic raw material use and technological strategies. In the first section, a series of expectations are proposed and evaluated for technological organization and raw material use. In the following section, a series of expectations concerning technological organization and settlement patterns are proposed and evaluated. These expectations are based upon current knowledge of the prehistory of the Pecos area and our understanding of lithic technological organization and are evaluated using survey data from the random samples. Some expectations are amenable to statistical testing, while others are best examined subjectively.

Technological Organization and Raw Material

Under the simplest of conditions, the presence of specific raw materials in an archeological assemblage should be a function of distance to the source and should therefore be static through time. Deviations from this pattern are expected and require social or economic explanations. We examine the affects of changes in social organization upon raw material use in two basic ways. First, we examine the differential use of raw materials for the production of tool classes for the entire Pecos survey sample. These patterns are then monitored through time. Next, we examine changes in the use of a specific raw material category, obsidian, through time. The latter examination is facilitated by X-ray fluorescence analysis (Shackley and White 1998).

Tool Production and Raw Material

The idea that lithic raw material availability affects decisions about tool production and maintenance is fundamental to studies of technological organization (Nelson 1991). In the case of Pecos National Historical Park, many of the sites are Puebloan and semisedentary to fully sedentary. When comparing groups with a similar mobility pattern, it is the availability and quality of raw material that constrains or encourages wasteful or conservative reduction techniques (Andrefsky 1994). Earlier in this chapter raw materials were identified as being local, nonlocal, or exotic in reference to their availability. The quality and suitability of raw materials for lithic reduction were also discussed. These attributes provide a framework for the following investigations.

The relationship between raw material availability and quality in regard to conservation and tool production is modeled in Figure 9.5. Conservation for this model is defined as the employment of strategies that maximize utility through either prolonged tool use life (e.g., Cowan 1999; Kelly 1988), or intensive use of material (e.g., Parry and Kelly 1987). The latter results in minimizing the wasteful discard of usable items (e.g., flakes and large cores). This model further assumes that high-energy tools are designed for long-term use (Binford 1973, 1979) and are therefore conservative of material.

Figure 9.5. A model of conservation and tool production strategies in relation to raw material availability and quality.

According to the model, locally available raw materials are not expected to have been conserved to any great degree, and the tool classes for which they would be used are dependent upon their quality. Conversely, nonlocal materials of high quality, would be highly conserved and particularly favored for the production of high-energy tools. Nonlocal materials of low quality would not often be used, because their procurement cost probably outweighs their value for lithic reduction.

Following Andrefsky (1994:Figure 2), we derive expectations for tool classes and degree of conservation for particular raw material groups. In other words, depending on flaking quality and local availability, each raw material is expected to be used for particular types of tools. The expectations for each raw material group are as follows:

Obsidian — (nonlocal; high quality) conserved; used for high-energy tools such as projectile points. Obsidian debitage will also be used for expedient tools due to their sharp edges.

Nonlocal chert — (nonlocal; high quality) conserved; used primarily for high-energy tools.

Local chert — (local; variable quality) not conserved; used for a wide range of tools including cores, expedient tools, low-energy tools, and high-energy tools.

Local chalcedony — (local; high quality) conserved; used for high and low-energy tools. Due to higher quality compared to other local raw materials, chalcedony should be used more similarly to nonlocal chert.

Local siltstone and quartzite — (local; low quality) not conserved; used for large low-energy tools and expedient tools.

An additional set of expectations concerns changes in raw material use through time. Because local materials are always available, local material is expected to be used in similar quantities throughout the Puebloan sequence. Nonlocal materials, however, should increase during the time of greatest interaction with groups outside the Upper Pecos Valley. Ceramic, lithic, and faunal data indicate that trade relations with outside groups were the greatest during the middle to late 1400's (Kidder 1932; Spielmann 1982; Spielmann, ed. 1991). Therefore, Period 4 (A.D. 1450-1575) through the end of the Puebloan occupation of the park is expected to have visibly higher average volumes of obsidian and nonlocal cherts.

Evaluation of these expectations is based upon the comparison of relative percentages of each tool class with respect to raw materials. Figure 9.6 presents percentages of each tool type within the five main raw material groups—siltstone and quartzite, local chert, local chalcedony, obsidian, and nonlocal chert. The figure indicates that siltstone and quartzite were used for low-energy tools and cores, with roughly equal amounts of expedient and high-energy tools. As expected, local chert was used for roughly equal amounts of each tool type. The chalcedony is split between high-energy tools and expedient tools with very few low-energy tools or cores. Obsidian is highly skewed towards use in high-energy tools and very few cores but generally conforms to expectations. Nonlocal chert displays the largest departure from the expected use, with equal amounts used for expedient, low-energy, and high-energy tools.

Figure 9.6. Tool types for major raw materials, all sites.

Figure 9.7 displays these same data in a different manner, showing the relative contribution of each raw material group in each tool class. Obsidian and local chert appear to have been preferred for high-energy tools. Low energy tools and cores display a pattern that includes more use of siltstone and quartzite than expected. Expedient tools have a higher percentage of obsidian than all other classes except high-energy tools. As indicated in the expectations, this is likely due to the extremely sharp edges on obsidian debitage.

Figure 9.7. Raw materials for each tool type, all sites.

The level of concern with conserving raw material can be assessed using the actual amount of a given raw material, measured by weight or volume, recovered as a given tool class. Specimen weight was not recorded in the field; we therefore use volume (length x width x thickness) to compare raw material amounts. As an example of the relationship between raw material conservation and the volume of that raw material recovered archeologically, frequent abandonment or discard of relatively large cores of a specific raw material results in a large volume of material recovered archeologically for the core tool class and reflects a lack of concern for material conservation. At the other extreme, heavily conserved materials would only be discarded as tools and wasteful discard would be of minimal volume.

Changing patterns in raw material use and conservation through time can be examined by considering the volume of each material type and tool class by time period. Note that in the following investigation, only undated sites (with no ceramics), and Periods 2 through 6 are discussed. The frequencies of Period 0 (pre A.D. 1075) and Period 1 (A.D. 1075-1200) sites resulting from the survey are so low (n=2 for both periods) that interpretations and comparisons based upon these sites would be suspect.

Figure 9.8 displays the average volume per site of siltstone and quartzite for assemblages through time. The undated sites have the greatest average use of siltstone/quartzite. There is a trend of increased average volumes through time until Period 5 (A.D. 1575-1700). There is also an increased use of siltstone/quartzite for low-energy tools throughout the Puebloan sequence. Cores account for a considerable portion of the siltstone and quartzite in each time period, while high-energy tools consistently account for a very small portion. This pattern indicates very little in the way of material conservation.

Figure 9.8. Average volume of siltstone/quartzite per site.

Figure 9.9 displays the relation of average local chert volume to tool type through time. The overall trend is similar to siltstone/quartzite—with an even greater volume of material recovered as cores. The pattern for local chert meets our expectations particularly well. The material is used for all tool types, does not appear to have been highly conserved, and its pattern of use is stable throughout all time periods.

Figure 9.9. Average volume of local chert per site.

The temporal pattern is different for the local chalcedony, as shown in Figure 9.10. The volume of chalcedony rose through Period 4 (A.D. 1450-1575) and falls toward the end of the Puebloan sequence. Expedient tools account for much more volume than cores among items made of chalcedony. In addition, a higher proportion of the material is used for the production of high-energy tools. Both of these patterns indicate a higher degree of material conservation than is indicated for other locally available raw materials.

Figure 9.10. Average volume of chalcedony per site.

Raw material coming into the Pecos Valley from more distant sources increases in volume through time, at least until the end of Period 5 (A.D. 1575-1700). Figure 9.11 displays the average volumes for obsidian artifacts. Overall, there appears to be a fairly gradual increase in obsidian volume through time with a consistent emphasis upon use for high-energy and expedient tools. Between Period 3 (A.D. 1325-1450) and 5, however, there is a notable increase in average obsidian volume, along with a decrease in material conservation as indicated by the increasing proportion of tools other than high-energy tools. Cores do not represent a great portion of obsidian in any time period. A more detailed analysis of obsidian from the park is presented below.

Figure 9.11. Average volume of obsidian per site.

Lastly, Figure 9.12 shows the pattern of volume for nonlocal cherts. Prior to Period 4 (A.D. 1450-1575) there are relatively few nonlocal chert materials. The undated sites also have very little nonlocal chert. In contrast, Periods 4 and 5 exhibit a considerable increase in the volume of nonlocal cherts, indicating either increased long-distance procurement, or more likely, increased interaction with other groups in the form of trade. Examination of the survey data revealed that raw materials from the Great Plains (particularly Alibates flint) were most abundant during these periods. Unlike obsidian, which is also presumed to have been a trade item, nonlocal cherts were not used or conserved in any unique way. Like local chert material, nonlocal chert was used for all tool types and consists of a high proportion of cores and expedient tools.

Figure 9.12. Average volume of nonlocal chert per site.

Several patterns resulting from this examination of raw material and technology warrant further discussion in relation to aggregation and trade. While trade for obsidian from the Jemez Mountain region appears to have gradually increased through time, a sharp increase in average nonlocal chert volume suggests that much of the interaction with groups from other areas began in Period 4 (A.D. 1450-1575) and continued through Period 5 (A.D. 1575-1700). As maximum population in the Upper Pecos Valley appears to have been reached during Period 3 (A.D. 1325-1450) (Chapter 7), a simple population threshold would not appear to account for the abrupt increase in trade. On the other hand, slightly lower overall populations more densely aggregated at one or two pueblos, which appears to be the case for Periods 4 and 5 (Chapter 7), do appear to correspond to the increased trade activity. Thus, it appears that only a centralized population was appropriately poised for substantial trade.

Judging from the presence of specific Great Plains sources, notably Alibates flint, the increased volume of nonlocal chert in Periods 4 and 5 appears to be related to some degree with increased interaction and trade with groups from the Great Plains. This same timing of interactions between the Plains and Pueblos is identified by numerous researchers (Baugh 1991; Habicht-Mauche 1991; Spielmann 1991). Correspondingly, obsidian from Jemez sources is most abundant in these Late Prehistoric period sites on the Southern High Plains (Baugh and Nelson 1987). Because there is little evidence of Plains/Pueblo interaction in the Jemez Mountain region for this time period, it seems probable that some of the obsidian on the Great Plains was being traded through the Pecos area.

Despite some of the nonlocal cherts from the park being of truly exotic origin, their use and degree of conservation does not appear to differ from that of local cherts. Thus it does not appear that they were any more desirable than the local material. This result runs counter to our expectation that high quality nonlocal raw materials should be used differently than other raw materials. A speculative explanation for this pattern is that exotic raw materials were little more than token gifts that accompanied the trade of more valuable items such as bison products and obsidian. As such, they would be of little more economic value than locally available raw materials of comparable quality.

Pueblo Period Obsidian Use

Earlier in this chapter we established that no obsidian is locally available in the vicinity of Pecos. Thus all obsidian raw material can be considered nonlocal (Jemez sources) or exotic (Mount Taylor sources) and can be used to better understand nonlocal and exotic raw material use through time. Specifically, by incorporating X-ray fluorescence studies carried out by the University of California at Berkeley (Shackley and White 1998), we can evaluate the differential use of specific obsidian sources for individual technological classes for the Pueblo period as a whole. Further, we can examine change in Pueblo period obsidian use through time by comparing Pecos Pueblo (PECO 228) with earlier pueblos—Forked Lightning (PECO 226), Dick's (PECO 434), Shin'po (PECO 307), Loma Lothrop (PECO 227), and Arrowhead (PECO 710).

X-ray fluorescence is a technique used to characterize the elemental composition of lithic material (Shackley 1998). It is possible to identify chemical differences between visually homogenous material through comparison of specific elemental content measured in parts per-million. Individual artifacts can be traced to particular geologic sources by comparing the chemical signatures from artifacts with those from the geologic sources. A specific method of X-ray florescence, energy dispersive X-ray fluorescence (EDXRF), was used by the UC Berkeley lab (Shackley and White 1998) to characterize the obsidian from Pecos. EDXRF can be used to accomplish the same results as normal X-ray florescence without adversely affecting the artifact (Davis et al. 1998).

As discussed earlier in this chapter, the closest source area of obsidian to Pecos is in the Jemez Mountains (Baugh and Nelson 1987; Glascock et al. 1998). Other possible sources of obsidian to which the Pecos samples were compared include the Taos Plateau sources (Mt. San Antonio and No Agua Peaks), and Mt. Taylor (Grants Ridge and Horace Mesa), which are also discussed above, along with Red Hill, Gwynn Canyon, Mule Creek, Cow Canyon, and Antelope Wells (Peterson et al. 1994, 1997). The proximity of the Jemez Mountains to Pecos (see Figure 9.2) renders the use of each of these other sources less likely.

Within the Jemez Mountains, Baugh and Nelson (1987) identified seven locality complexes (Table 9.17). They sampled 16 localities within the locality complexes. The naming conventions for locality complex presented in Baugh and Nelson (1987) are used in this study. For example, rather than using "Polvadera obsidian" when reporting X-ray florescence sources, "El Rechuelos"—the name of the locality complex—will be used instead. One exception is the use of Bland Canyon as a separate complex from Cerro Toledo. This decision is based on the lab report from the UC Berkeley EDXRF lab (Shackley and White 1998). When presenting visually identified material, this analysis will use the naming conventions of the survey: Jemez Mountain obsidian, opaque Jemez Mountain obsidian, and Polvadera obsidian.

Table 9.17. Jemez Mountain obsidian sources.

Table 9.18. Energy Dispersive X-ray Fluorescence sources for tool types.

A sample of 150 obsidian artifacts was submitted for source identification. The sample included specimens from Pecos Pueblo, Loma Lothrop, Shin'po, Dick's Ruin, Forked Lightning Ruin, and Arrowhead Ruin (see Figure 1.3). Obsidian samples from each of these major pueblos in Pecos were derived mainly from collected artifacts, most of which were in the form of tools. In addition, a grab sample of 5-17 pieces of debitage from each of the above sites was collected. Some further samples were judgmentally included based upon their relevance to future research at Pecos National Historical Park or inability to be sourced based upon visual criteria. These artifacts were submitted to the EDXRF lab at UC-Berkeley. The EDXRF lab at UC-Berkeley has extensive source signatures for all known obsidian sources within the American Southwest (Shackley 1988, 1995).

Artifacts analyzed at the UC-Berkeley lab are categorized as small points/tools, large points/tools, debitage, and debitage with cortex. Small points and tools include such items as probable arrow points, drills, small scrapers, and knives but do not include used flakes. Debitage and debitage with cortex include all material removed from a tool or core that is not itself a tool, except for the case of used flakes. Most of the debitage with cortex only had a small amount of cortex on their dorsal faces.

A breakdown of tool types by EDXRF provenance is presented in Table 9.18. The Cerro Toledo (n=72), Valle Grande (n=46), and El Rechuelos (n=21) sources are the best represented in the sample. Bland Canyon (n=8), Paliza Canyon (n=1), and Grants Ridge (n=2) are present in smaller amounts. The relative abundance of each of the three major sources in the overall sample reflects their increasing distance from Pecos (Table 9.19). In other words, the geographically closest sources to the Upper Pecos Valley are the best represented in artifact assemblages, and the farthest are the least common. This represents the simplest of expected relationships—that abundance is a function of distance and indicates that differential social or economic conditions did not have a substantial effect upon the availability of one obsidian source relative to other sources of obsidian.

Table 9.19. Obsidian abundance and distance to source.

The lack of cortex on any piece of El Rechuelos debitage, even when compared to other, less numerous, obsidian sources (such as Bland Canyon) suggests that El Rechuelos obsidian came into the Pecos region in a more reduced state than any others from the Jemez Mountains and is also likely related to the greater distance of the source. Another interesting observation is that El Rechuelos is relatively less exploited as a tool stone for larger points and tools. This implies that either the material available from El Rechuelos is unsuitable in quality, availability, or not of a sufficient size to produce large points or that the makers of the larger points and tools (most of these large points are dart points and match types for Archaic and Basketmaker II periods) did not use this source as extensively.

Change in obsidian use among the largest pueblos through time is expected to reflect variations in trade patterns. The small number of artifacts submitted for X-ray florescence analysis does not allow for rigorous statistical testing of the difference between the largest pueblos in Pecos National Historical Park. Figure 9.13 displays the percentage of specific obsidian sources identified from this study for each of the large pueblos.

Figure 9.13. Obsidian sources determined by Energy Dispersive X-ray Fluorescence for pueblos.

The pueblos are arrayed along the base of the graph so as to correspond generally from earliest to latest, with considerable and unequal temporal overlap. It should be noted that there are possible later Plains components at Dick's Ruin, Loma Lothrop, and Shin'po (Head 1998) that may influence patterns in obsidian use. Early in the Puebloan sequence, at Forked Lightning Ruin (PECO 226), there are roughly equal amounts of Cerro Toledo, Valle Grande, and El Rechuelos sources. Dick's Ruin (PECO 434) has the highest percentage of El Rechuelos obsidian of all the sites (40 percent) with no Valle Grande and only a small amount of Bland Canyon. Shin'po exhibits smaller amounts of El Rechuelos and close to equal amounts of Cerro Toledo and Valle Grande. Toward the middle of the temporal sequence, Loma Lothrop (PECO 227) is distinguished by a relatively small proportion of El Rechuelos and a large percentage of material from Bland Canyon (all of the Bland Canyon sample is debitage and could have easily been detached from a single core). Arrowhead Ruin has no El Rechuelos and the only evidence of any long-distance obsidian, a drill identified as Grants Ridge obsidian. The only other piece of Grants Ridge material came from an Archaic/Developmental site. For the latter portion of the sequence, Pecos Pueblo has almost equal amounts of Cerro Toledo and Valle Grande as well as smaller amounts of El Rechuelos and Bland Canyon.

Figure 9.13 reveals several patterns in obsidian acquisition for the large pueblos. Obsidian from the Cerro Toledo source was used extensively throughout the sequence. Valle Grande obsidian is a consistent contributor to the assemblages with the exception of Dick's Ruin, where none was identified. A reduction in utilization of El Rechuelos through the Puebloan sequence is indicated. Both Forked Lightning and Dick's Ruin have relatively high percentages from this source, while all the later pueblos have very little. Two sites stand out in relation to obsidian use. Dick's Ruin exhibits a noticeable lack of Valle Grande obsidian, which is present at all others. Arrowhead lacks El Rechuelos items but is the only Puebloan site in the sample that includes Grants Ridge obsidian. These latter observations, along with the abundance of Bland Canyon obsidian at Loma Lothrop, suggest that trade may have been oriented in a more southwesterly direction from the Upper Pecos Valley during the middle portion of the sequence. This suggestion is a possible avenue for future research when more comprehensive data become available. Because these results are based upon material gathered in a nonrandom fashion, making generalizations to the broader obsidian populations on each of these sites is tenuous at best. The results do indicate that the differential use of particular obsidian sources is primarily related to the distance of those sources from Pecos. It also appears that use of the El Rechuelos locality as a source of tool stone decreased through time compared with the Valle Grande and Cerro Toledo sources. Lastly, the results indicate that, as might be expected, trade for obsidian was focused almost exclusively on groups in the Jemez Mountain area.

Technological Organization and Settlement Patterns

We also anticipate that the effects of aggregation and trade on settlement organization (e.g., centralization of activities) led to differences in the distribution of tool and debitage classes among site types through time. This is evaluated through the statistical examination of percentages of each artifact class for site types from selected time periods.

The goal of this analysis is to identify differences or similarities between site types through time using the artifact classes presented earlier in this chapter. Specific expectations are proposed for both site types and time periods. Assemblages at the level of site component are used in the analyses that follow. For a single component (as defined in the field), the percentage of the assemblage for each of the following stone classes is used: high-energy tools, low-energy tools, expedient tools, cores, class 1 debitage, class 2 debitage, and class 3 debitage (see definitions above, Tables 9.8 and 9.16).

Because multiple variables (site type, artifact class, time period) are being simultaneously evaluated in the statistical tests, the Analysis of Variance (ANOVA) is used. ANOVA is a technique that is used to identify relationships between dependent and independent variables, whether the independent variables are quantitative or qualitative in nature (Kachigan 1991). ANOVA examines the difference in the mean values of a dependent variable that are associated with different values of the independent variable. The F distribution is used to assess whether this difference can be attributed to random variation in the sample means or to significant differences between the independent variables.

Most researchers outside archeology use ANOVA where the independent variables are controlled by the design of an experiment. In the case of most archeological applications—including this analysis—controlled experimental manipulation is impossible. Such nonexperiment uses of ANOVA require the analyst to settle for the problems associated with fixed effects, especially a lack of causal linkage between the dependent and the independent variable(s) (Kachigan 1991).

This investigation is also focused on the Puebloan occupation of Pecos. Small numbers of Archaic and Developmental sites prevented a statistical analysis of those time periods. Time period 1 (A.D. 1075-1200) was also removed from the second analysis (change through time) due to the small number of sites. The first statistical evaluation serves to examine the differences among site types for all Pueblo periods (see Chapter 5). A second statistical evaluation serves to examine differences among site types for individual time periods (see Chapter 4).

Artifact Classes and Site Types

Specific expectations for the abundance of particular artifact classes for different site types are based on site type definitions, which emphasize architectural features (Chapter 5). The activities corresponding to the definitions are assumed to have taken place at each site type. Special-use sites are primarily for resource extraction and should be characterized by simple and sparse artifact assemblages that reflect the use of expedient tools and some maintenance debris from low- and high-energy tools. Because seasonal sites represent short-term domiciles, artifact assemblages are expected to reflect more tool maintenance and relatively low investment in tool production. Habitation sites represent the primary location of domestic activities, including tool production and maintenance, for most of the population, and are expected to be characterized by dense and diverse artifact scatters. Regardless of where particular tools were used, the tools themselves are expected to have most often been discarded in places where they were replaced through the production of new tools. For low- and high-energy tools these locations would most often be habitations, while at special-use sites primarily expedient tools were being produced and discarded. Thus, the following expectations for dominant artifact classes are derived for each site type:

Special-use sites — expedient tools and class 1 and class 2 debitage.

Seasonal sites — low-energy tools and class 1 and class 2 debitage.

Habitation sites — all artifact classes, with a relatively frequent occurrence of high-energy tools.

The null hypothesis to be tested is that there are no differences in artifact class frequencies for component assemblages from different site types. Rejection of the null hypothesis for any pair of site types indicates significant differences in artifact class frequencies.

The results of the ANOVA in all but one case are that the null hypothesis cannot be rejected. Thus, the analysis did not support most of the expectations for differences between site types. There were no significant differences between site types of all time periods with respect to percentages of low-energy tools, cores, class 1 debitage, class 2 debitage, and unknown/ambiguous debitage.

One expectation was supported by the analysis—more high-energy tools were found at habitation sites than at either seasonal or special-use sites (Table 9.20). Habitation sites recorded in the survey were composed of an average of 8.3 percent high-energy tools while both seasonal and special-use sites were composed of an average of 3.3 percent. This is not believed to necessarily reflect a greater frequency of activities requiring high-energy tools at habitation sites. As presented above, it more likely represents the discard of high-energy tools at a primary locus of tool replacement and maintenance.

Table 9.20. ANOVA comparison of average percentage of high energy tools from site types from all time periods.

Site Types through Time

Identifying changes in artifact classes from site types through time would allow us to discern changes in the roles of particular site types in overall settlement organization. The differences within the Puebloan time span are not anticipated to be great since the Puebloan populations in the Upper Pecos Valley were primarily sedentary. Parry and Kelly (1987) suggest that such populations would not have significant differences in overall lithic technology. However, if one accepts Kidder's (1958) interpretation of the architecture at Pecos Pueblo as defensive, the latest time periods might be expected to have a greater frequency of high-energy tools (over 75 percent of which are projectile points and preforms) for defensive purposes. Although defensive technology was not necessarily limited to high-energy tools, two human vertebrae from Pecos Pueblo with imbedded projectile points (Kidder 1932) attest to their use in human conflicts. Kidder (1958:63) suggests that the quadrangle of Pecos Pueblo was built out of concern over such conflict. The building of the quadrangle at Pecos Pueblo corresponds roughly to the beginning of Period 4 (A.D. 1450-1575) in the Pecos chronology. If conflict in the region necessitated a greater defensive posture in the valley, implements of warfare such as projectile points would be expected to increase for habitations in Period 4 relative to Period 3 (A.D. 1325-1450). Thus, the expectations for changes through time are simple:

Periods 1-3, 6 — no differences in artifact classes for site types.

Periods 4-5 — greater frequencies of high-energy tools at habitation sites.

The first expectation mirrors the null hypotheses being tested—that there are no differences in artifact class frequencies for assemblage components from different site types. Rejection of the null hypotheses would indicate that significant differences do exist. Random samples are used in the ANOVA testing.

The results of the second series of ANOVAs are the frequent inability to reject the null hypothesis. Within each time period, there were few significant differences between site types. However, some differences are indicated to be significant for a few time periods. In Period 2 (A.D. 1200-1325), the analysis indicates that habitations have significantly higher percentages of high-energy tools than either special-use or seasonal sites (Table 9.21). The analysis did not identify any significant differences for other artifact classes.

The percentage of high-energy tools is significantly different for habitations from Period 3 (A.D. 1325-1450) as well (Table 9.22). There are no significant differences between artifact classes in Period 4 (A.D. 1450-1575).

Two artifact classes have significant differences for each of the site types in Period 5 (A.D. 1575-1700)—low-energy tools and class 2 debitage. Habitations dating to Period 5 have significantly greater amounts of low-energy tools in their assemblages compared to special-use or seasonal sites (Table 9.23). A greater amount of Class 2 debitage is marginally significant (p value between .05 and .10) for special-use and seasonal sites when compared to habitations (Table 9.24). This suggests that while low-energy tools were discarded in greater numbers at habitations, tool maintenance occurred at seasonal and special-use sites. As expected, there are no significant differences between site types for Period 6 (post-A.D. 1700).

Table 9.21. ANOVA comparison of average percentage of high energy tools from Period 2 site types.

Table 9.22. ANOVA comparison of average percentage of high energy tools from Period 3 site types.

Table 9.23. ANOVA comparison of average percentage of low energy tools from Period 5 site types.

Table 9.24. ANOVA Comparison of average percentage of Class 2 debitage from Period 5 site types.

The patterns indicate that reduction relating to high-energy tool technology is concentrated at long-term habitations. This does not necessarily mean that such high-energy tools were more often produced at habitations—there are no significant differences between the debitage classes for site types in most time periods. We suggest that habitations were places where high energy tools such as projectile points and bifaces were discarded.

Following Kidder's (1958) suggestion that the quadrangle at Pecos Pueblo was built for defense, it was expected that habitations from Periods 4 and 5 would show more evidence of these high-energy tools, primarily for defense purposes. The expected increase in high-energy tools for these periods was not supported by the data. The inverse is actually indicated. Habitations in Periods 2 and 3 have significantly more of these tools than other sites.

Why should habitations from these periods have more high-energy tools? Comparison of Pecos Pueblo with Rowe Pueblo and Arroyo Hondo, two other extensively excavated Pueblos in the northern Rio Grande area, suggests a potential answer to this question (Table 9.25). Rowe Pueblo is located in the Upper Pecos Valley about 7 km (4 mi) southeast of Pecos Pueblo and was occupied from the mid 1200s through at least the late 1300s (Cordell 1998:58). Arroyo Hondo is located about 13 km (8 mi) west/northwest of Pecos Pueblo along a tributary of the Santa Fe River of the same name. Arroyo Hondo was occupied from A.D. 1300-1425.

Table 9.25. Faunal and lithic comparison of Pecos Pueblo, Rowe Pueblo, and Arroyo Hondo.

The recent Rowe report identifies a larger than expected amount of large mammal bone found at the pueblo, 34.5 percent of which was from deer (Cordell 1998:63; Mick-O'Hara 1998). This compares well with Kidder's impression that mule deer was the most abundant mammal represented in the Pecos Pueblo faunal assemblage. Kidder (1932:196) estimated that one-half to three-quarters of the mammal bone at Pecos Pueblo was mule deer. This may very well be an overestimation considering the incipient state of archeofaunal analysis in the 1930s, but nevertheless the observation indicates a conspicuous abundance of deer remains from Pecos Pueblo. In contrast to Rowe and Pecos Pueblo, only 9 percent of the identifiable sample was composed of mule deer at Arroyo Hondo (Lang and Harris 1984:46).

The pueblos also differ in regard to stone tool assemblages. The randomly sampled assemblages from all habitations recorded during this survey are composed of 8.1 percent high energy tools. The randomly sampled survey assemblage from Pecos Pueblo alone is composed of 9.8 percent high-energy tools. Roughly 3.3 percent of the total flaked stone assemblage from the pueblo of Arroyo Hondo reflects high-energy tools (derived from Phagan 1993, Table 21). Lithic assemblage data are not available for Rowe.

Taken together, the relative paucity of both mule deer remains and high-energy-stone tools recovered from Arroyo Hondo suggests that the hunting of large mammals was not a primary focus there. In contrast, the abundance of mule deer remains at Rowe Pueblo and Pecos Pueblo coupled with the higher percentage of high energy tools within Pecos National Historical Park seems to reflect a greater emphasis upon large mammal hunting. Thus, the patterns indicated for high-energy tools in the Pecos survey are more readily explained with regard to subsistence than to defense. It appears that more intensive large mammal hunting may have occurred during Periods 2 and 3 relative to Periods 4 and 5, with correspondingly greater investment in high-energy tools. If this is the case, then it would appear that the importance of large mammal hunting for the inhabitants of the Upper Pecos Valley decreased over this period of time. This proposition could be better evaluated with more detailed faunal analyses at sites within Pecos National Historical Park.

The Nonflaked Lithic Assemblage

Nonflaked lithic items are those stone artifacts that were not produced or maintained through reduction by flaking. This category includes ground, polished, and abraded stone items, as well as some that were used without intentional modification (e.g., manuports, anvils). Ground stone tools and technology are the subject of much of this section of the report due both to their abundance in the archeological record and their association with past food processing activities.

The recording system for nonflaked lithic items differed from that of flaked stone. Lithic analysts recorded all nonflaked lithic items that fell within a sample unit (Appendix C) and all others found on the site. Therefore, the ground stone database comprises all of the identifiable ground stone on the surface of sites within the park.

Material type, nonflaked lithic type, condition, and size (length, width, and thickness) were recorded for every item when possible. Length, width and thickness were recorded only for whole items or fragmentary items that fit together to make a whole. Size data were not recorded for incomplete items. For ground stone, nonflaked lithic type specified the ground stone form, cross-section shape, and number of grinding surfaces were also recorded. The recording manual (Appendix C) provides definitions for each of these attributes, including each of the ground stone types.

Nonflaked Lithic Assemblage Parameters

This section presents the summary results for the nonflaked lithic types recorded during the Pecos survey, regardless of time period or site type, with an emphasis on ground stone tools. These data serve as the baseline for the subsequent section in which specific expectations are evaluated. Summary results are presented by nonflaked lithic type—manos, metates, axes/mauls/hoes, and other nonflaked lithic materials.

Manos

Manos were assigned to several classes during field recording. Because mano morphology mirrors the morphology of the used surface on associated metates (Adams 1999), mano classes were defined in reference to probable association with certain metate types. Mano types include one-hand manos often associated with basin metates, two-hand manos associated with trough metates (with distinctive facets produced by incidental rubbing of the mano against the trough walls), and two-hand manos associated with slab metates, which are longer and narrower than trough manos and lack the distinct facets. If the type of two-hand mano could not be determined, manos were assigned two-hand mano, not further specified (nfs). If size could not be determined, the artifact was assigned to "mano, nfs." Survey crews identified a total of 510 one-hand manos; 1 two-hand mano, trough; 16 two-hand mano, slab; 48 two-hand manos, nfs; and 167 manos, nfs.

Table 9.26 presents the breakdown of mano type by raw material. For the sake of brevity, the table displays manos used with trough and slab metates collapsed into the more general "two hand mano" category. Three major material types dominate the assemblage of manos—quartzite, sandstone, and silicified sandstone. These materials account for a total of 700 of the 742 manos, or nearly 95 percent. Quartzite and silicified sandstone are primarily found as cobbles in current and remnant alluvial deposits (such as gravel bars and terraces) that comprise much of the surface deposits of this portion of the Upper Pecos Valley. Sandstone is found as exposed bedrock or boulders, though some pieces are found in the alluvial deposits as well.

Cross-section data for manos provide information that can be used to infer grinding intensity and material conservation. Graphical representations of cross-section types are presented in Figure 9.14. Using ethnographic observations of Hopi women grinding corn as well as experimental observations (Adams 1993; Bartlett 1933; Cameron 1998), archeologists have identified specific cross sections that indicate intensity of mano use. The continued use of a two-hand mano results in a progression from subrectangular (Figure 9. 14g), through wedge (Figure 9. 14d), airfoil (Figure 9. 14f), and triangular (Figure 9. 14b) cross sections. This progression is not transferable to most one-hand manos, which most likely were used in a reciprocal and/or rotary motion (Wright 1993). Adams (1993, 1999) suggests that patterns of use from experimental grinding activities with one-hand manos produce different wear patterns depending upon the motion used. Rotary motion produces wear along the edges of the mano, while reciprocal motion wears the mano in the center. Rotary motion might result in biconvex cross sections while reciprocal motion with one-hand manos would probably produce a progression similar to that for two-hand manos. The degree of grinding intensity reflected in the cross-section morphology of discarded manos indicates the level of concern for material conservation. A low level of concern for conservation would be reflected in the discard of manos characterized by relatively low intensity of use; a high level of concern would be reflected in the discard of only intensively used manos.

Figure 9.14. Cross-section classes for ground stone: a. plano-convex, b. triangular, c. biconvex, d. wedge, e. concave-convex, f. airfoil, g. subrectangular, and h. oval.

Table 9.26. Cross-tabulation of mano type by material type.

Composite mano cross-section data from all sites within Pecos are presented in Table 9.27. Cross-section data were not recorded for a small number of manos (n=15). In decreasing frequency oval, subrectangular, and plano-convex account for 497 of 727 manos (68 percent). Oval cross sections are nearly twice as frequent as the next most frequent classes. One-hand manos tend to be oval, with plano-convex, biconvex, and subrectangular being frequent as well. On the other hand, most two-hand manos are subrectangular, with others being plano convex or oval in cross section. These patterns indicate a relatively low intensity of use prior to discard or abandonment for most manos. One hand manos in particular tend to be less intensively used.

Another way of estimating grinding intensity is to examine the number of ground surfaces on each mano. Table 9.28 presents these data for each general mano type. Examination of this table indicates that the majority of manos (n=461, or 62 percent) have only one ground surface, but a noticeable number have two or more ground surfaces (n=259, or 35 percent). Within each mano type specimens with one ground surface predominate; a roughly equal ratio of one-surface to-two surface manos occurs within the one- and two-hand manos (1.47:1 and 1.46:1, respectively) but "mano, nfs" has a much higher ratio (4:1) suggesting that these ambiguous manos were less intensively used.

Table 9.27. Cross-tabulation of mano type by cross section.

Table 9.28. Number of ground surfaces by mano type.

Condition refers to the completeness of each mano as estimated by the lithic field analyst. Four condition options presented gross levels of artifact completeness and include: less than 50 percent, more than 50 percent but not complete, broken but complete (able to fit pieces together), and complete. Table 9.29 presents the totals for each of the mano types. Most of the "mano, nfs" class falls into the less than 50 percent condition (almost 80 percent). The majority of one-hand manos are complete (53 percent), while two-hand manos are evenly split between the three main classes (disregarding "broken, but complete," which has only one instance for all manos). The greater rate of breakage for two-hand manos is likely related to shape, because the greater ratios of length to thickness increase the probability of breakage under bending stress.

Table 9.29. ross-tabulation of mano type by condition.

Size measurements for each complete specimen were recorded as maximum length, width, and thickness. Table 9.30 displays the average measurements for each mano type. Comparison of the average sizes for one-hand and two-hand manos provides an opportunity to evaluate the reliability of the subjective classification. A t-test reveals that two-hand manos and one-hand manos are significantly different in length (Table 9.31) and width dimensions (Table 9.32) but not in thickness. On complete manos, total grinding area is larger for two-hand manos. Manos that could not be assigned as belonging to either the one- or two-hand variety fall between these two extremes. These results are not surprising since manos were classified as either one- or two-hand primarily based on size criteria, but they do serve to demonstrate the reliability of the field classification.

Table 9.30. Average sizes for complete manos by mano type.

Table 9.31. Student's T-comparison of mano type by length.

Table 9.32. Student's T-comparison of mano type by width.

Metates

The number of metates located and recorded within Pecos National Historical Park is rather low. Only 161 metates were identified for all Puebloan sites. Table 9.33 presents metates by material type for all Puebloan sites. A majority of these metates were made of sandstone (n=83, or 52 percent). Other frequently used materials are, in decreasing order, quartzite, orthoquartzite, and metamorphic, nfs. Other raw materials contributed less than 10 percent each. Of the metates that were complete enough to classify as to type, most were slab metates (n=46, or 29 percent). Only 29 (18 percent) were basin metates. Very few trough metates were identified (n=3, or less than 2 percent). The latter are included within the "metate, other" category in Table 9.33.

Table 9.33. Cross-tabulation of metate type by material type.

Cross-section data are generally less informative concerning intensity of use for metates than for manos. Slab metate use intensity is particularly difficult to determine due to the relatively even surface wear expected and the inability to estimate original thickness. Metate cross sections (Figure 9.14) are a result of the size of the accompanying mano and the motion with which the mano was used (Adams 1999). Therefore, assuming that two-hand manos accompanied slab metates and one-hand manos accompanied basin metates, we would expect most slab metates to be subrectangular and basin metates to be convex-concave. Table 9.34 presents the cross-tabulation of metate type by cross section. Most basin metates fit the general expectation, but a surprising number (n=8 or 21 percent) of slab metates are also convex-concave. These might represent slab metates that were used relatively intensively with a one-hand mano or a comparatively short two-hand mano. The abundance of one-hand manos (n=510 or 69 percent), in comparison with the relative dearth (n=23 or 18 percent) of basin metates with which they are expected to have been used, might indicate that one-hand manos were in fact often used with slab metates.

Table 9.34. Cross-tabulation of metate type by cross section, all sites.

Nonportable Grinding Features

In addition to portable nonflaked stone tools, 509 nonportable ground or pecked and ground features were recorded. Each of them occurs on a boulder or on exposed bedrock within a site or as an isolated occurrence (IO). The majority of these features are interpreted as processing features for subsistence items. These include 326 ground surfaces and 140 cupules and mortars. The ground surfaces vary from rectangular or square to round, oval, or indistinct in shape and may have been used as expedient metates. The cupules and mortars are round and deep relative to their circumference (typically 4-8 cm). These appear to have been created through a combination of grinding and pecking, and although their function is undemonstrated, they may represent mortars or "nutting stones," for the preparation of subsistence items or pigment. The remaining 44 features are relatively long narrow grooves traditionally interpreted as areas where ground tools (such as axes and mauls) were shaped or sharpened.

Four of the grinding features consist of grinding surfaces with smaller depressions at their bases, referred to here as complex grinding features (Figures 9.15 and 9.16). PECO 301 possesses two such features, both of which are surrounded by a wide groove that empties into the pecked and ground depression. IO 333 is analogous in form to those from PECO 301, consisting of a grinding surface surrounded by a groove leading to an underlying depression. Both the groove and the grinding surface are less well formed than those from PECO 301. IO 112 differs from these only in that no groove surrounds the grinding surface.

Figure 9.15. Idealized plan view and cross sections of a complex grinding feature.

Figure 9.16. Complex grinding feature at PECO 301 (LA 118875), feature 1-61-02, view N.

These arrangements may have served as grinding facilities analogous to a mealing bin containing a metate. Some support for this interpretation is found through a comparison of grinding surface size from these complex grinding features with those from complete metates recorded during the survey. Figure 9.17 illustrates that complex grinding feature surface sizes compare well with sizes for complete metates, falling mostly toward the larger end of the range. The size range compares especially well with that for slab metates, although statistical comparison is hindered by small sample sizes. Further support is provided by manos found near two of the features (those from PECO 301), one of which was noted to correspond in shape and size to the adjacent complex grinding surface.

Figure 9.17. Scatterplot comparing size of grinding area for complex grinding features with size of grinding area for complete metates.

Axes, Mauls, and Hoes

A relatively small number of these nonflaked stone tools were recorded. Eight ground stone axes (Figure 9.18a), 11 mauls (Figure 9.18b), and 16 hoes were identified on Puebloan sites. The only strong trend in raw material for these nonflaked lithic types is that hoes tend to be made of sandstone (n=10, or 63 percent), while mauls and axes were made from a greater variety of raw materials (Table 9.35). Kidder (1932:50) noted that fibrolite was a favorite material for what he defined as Type II axes (axes with specialized grooves) at Pecos Pueblo but that this material is not locally available. Montgomery (1963) proposed that the source for high quality fibrolite used in axes from around much of the Puebloan Southwest was the Truchas Peaks area (about 50 km [31 mi] north of Pecos Pueblo in the Sangre de Cristo Mountains). This led Montgomery, as well as Snow (1981:364), to propose that fibrolite axes were important exchange items among the northern Pueblos. Thus, it is somewhat surprising to note that a single axe of fibrolite, along with one maul, was recorded during this survey. It is difficult to ascertain the degree to which the sample is biased by previous collection in the monument, but reports of dozens of collected fibrolite axes (Kidder 1932; Montgomery 1963) suggest that such bias may be substantial and might explain the infrequency of fibrolite in the survey data.

Figure 9.18. Ground stone tools: a. axe (Cat. No. 29180), b. maul (Cat. No. 29201).

Table 9.35. Cross-tabulation of nonflaked lithic type by material type.

Axes, mauls, and hoes appear to be distributed differently throughout the park. Figure 9.19 shows the geographic distribution of all axes, mauls, and hoes recorded within the park for Puebloan sites. All sites with axes are close to major drainages and tend towards the northern sections of the park. All but one are near either the Pecos River or Glorieta Creek and one of its major tributaries, suggesting their possible use for the procurement or processing of resources that occur in riparian environments.

Figure 9.19. Hafted tools recorded by the survey. (click on image for an enlargement in a new window)

The locations of hoes presented in Figure 9.19 shows that most of these implements, traditionally assumed to be agricultural tools, are found along Glorieta Creek and one of its major tributaries. All but four appear to be very close to these drainages. These hoes are associated with a dense area of occupation within Pecos National Historical Park that may represent the greatest concentration of both population and agriculture.

The locations of mauls in the survey area reveal a contrasting pattern to both the hoes and axes. Further, mauls and hoes/axes do not co-occur on any sites (hoes and axes do co-occur on some sites). Mauls were more frequently found in upland areas between major drainages. Maul function has rarely been addressed in the archeological literature. It seems likely that they were used more for battering or crushing than for chopping, and perhaps functioned as quarrying implements for architectural stone, or as hammers for driving wedges to split wood. A lack of mauls in the excavations at Pecos Pueblo suggested to Kidder (1932) that these were more "archaic" implements, though he does report finding a number of examples at Forked Lightning Pueblo. They are not found to be particularly associated with early sites in the Pecos survey data.

Other Nonflaked Lithic Artifacts

A number of other nonflaked lithic items that do not fit into the categories described in the preceding sections were recorded during the survey. Lithic analysts recorded 202 additional nonflaked lithic artifacts in 11 different nonflaked lithic classes. These were made from a wide range of material types (Table 9.36). Recorders used the "other" category most often (n=70) and added narrative descriptions that are difficult to summarize. Most of these tools were too fragmentary for assignment to a nonflaked lithic type, but some simply did not fit into available recording classes. The latter include items such as pecking stones, shaft-straighteners, and stone pipe fragments.

Table 9.36. All other nonflaked lithic artifacts.

Manuports comprise the next most frequently recorded nonflaked lithic class presented in Table 9.36, followed closely by hammer stones. Manuports were often cobbles of raw material that were probably intended to be used in making other tools—obsidian, chert, and chalcedony for chipped stone; basalt, sandstone, and silicified sandstone for ground stone; and turquoise and petrified wood for personal adornment or some other purpose. Dense raw material was used for many hammer stones, with quartzite and chert being the most frequently encountered.

Population Aggregation and Grinding Intensity

The preceding section presents the overall patterns in the data for all nonflaked lithic materials. We can now use some of these data to address particular questions regarding the effects of population aggregation upon nonflaked lithic technology. In the research design (Chapter 1), it is hypothesized that an intensification of grinding activities, due to increased consumptive needs, occurred with the establishment of Pecos Pueblo as an aggregated population center. Based upon the relationship between grinding intensity and mano cross-section morphology presented above, the effects of aggregation should be identifiable in the Pecos survey ground stone assemblage. The prehistory of Pecos National Historical Park leads us to expect differences among both time periods and site types. After a brief review of patterning related to site types, we statistically evaluate changes in grinding intensity through time.

Grinding Intensity and Site Types

If mano cross-section morphology is related to intensity of use, the frequencies of particular cross sections should vary according to site type, which is an interpretation of the nature of site occupation based primarily upon architecture. Differences among site types are not statistically evaluated here but can be subjectively evaluated based upon graphical representations of cross-section distributions for different site types. Figure 9.20 presents a cumulative percentage graph of cross-section types from all time periods organized by site type. Subrectangular and oval cross sections are the most abundant types over all. Manos with oval cross sections constitute a greater proportion of the assemblages from seasonal and special-use sites relative to habitation sites. Those with biconvex cross sections are most common at habitations (but also common at special-use sites). These patterns probably reflect a more expedient use of available cobbles at sites that were occupied less often or for shorter periods of time. Wedge and triangular cross sections appear to be patterned according to occupational intensity, as well (although airfoil cross sections remain constant). Wedge and triangular cross sections are least common on special-use sites but increasingly common at seasonal sites and habitations. This pattern probably reflects greater reuse of ground stone tools at sites representing longer or more frequent occupation. Thus, with the assumption that our site types represent valid interpretations of occupational differences, it appears that there is a logical relationship between occupational intensity and intensity of ground stone tool use that is reflected in mano cross sections. This relationship provides a means by which we can examine change in intensity of ground stone use through time.

Figure 9.20. Cross sections of all manos by site type.

Grinding Intensity through Time

As presented above, intensification of grinding activities due to increased consumptive needs is hypothesized to have accompanied aggregation. According to this hypothesis, the social unit (presumably a household) being served by an individual grinder had increased in size. One testable implication of this hypothesis is that frequencies of mano cross-section types that are associated with intensification of grinding activity rise during the periods of greatest population aggregation. As presented earlier, wedge, triangular, and airfoil cross sections are indicative of high grinding intensity for two-hand manos while these and biconvex cross sections are indicative of high grinding intensity for one-hand manos. Maximum population aggregation in the Upper Pecos Valley is estimated to have begun during Period 4 (A.D. 1450-1575) (Chapter 7, this volume). The expectation, therefore, is that the above cross-section types will be significantly more abundant for Period 4 and later periods than for the preceding periods. The null hypothesis that there is no difference in average frequencies of cross-section types is tested for Periods 3, 4, and 5 below. Rejection of the null hypothesis supports the expectations only if changes in the most frequent cross-section types are indicative of increased grinding intensity.

Despite having presumably recorded the total population of visible manos from sites within the survey area, this assemblage represents only surface artifacts from identified sites and remains a sample of a larger population. Resampling Statistics^TM allow us to compare these subsamples derived from the mano data with a simulated distribution derived from the parameters of that same data, thus enabling us to evaluate differences within the assemblage. Resampling Statistics^TM (also known as Monte Carlo simulation) require no assumptions beyond that of assuming the sample is representative of the population. Simply summarized, through resampling it is possible to construct simulated distributions based on the parameters of the original sample. These simulated distributions provide samples against which others can be compared to assess whether the two samples are likely to have been taken from the same statistical population. Significance levels were set at a two-tailed .05 probability, with p values less than .025 indicating a significantly lower frequency of individual cross-section classes and p values greater than .975 indicating a significantly greater frequency of those cross sections.

Distributions for mano cross-section types for each Puebloan time period were constructed in the following manner. A simulated distribution was constructed based upon the observed distributions from Period 2 (A.D. 1200-1325). Temporal trends were assessed using Period 2 as the baseline dataset since there are few Period 1 (A.D. 1075-1200) sites. The observed distribution for Period 3 was then compared to the simulated distribution for Period 2. By this procedure one can ascertain the probability that Period 3 manos come from the same population as Period 2 manos. This procedure is repeated for each period, with the observed percentages for the period of interest being compared to the simulated distributions based upon the preceding period.

The results of statistical tests are summarized in Table 9.37. The observed data is presented graphically in Figure 9.21. When we examine the probabilities of getting the distribution for Period 3 (A.D. 1325-1450) based on Period 2 (A.D. 1200-1325) (shown in Table 9.37), we find that there are a significantly higher number of plano-convex cross sections and a significantly lower number of subrectangular cross sections than expected for the Period 3 assemblages. The null hypothesis is therefore rejected in regard to Period 3. Significantly greater percentages of cross-section types that are not associated with intensive grinding activity are present. Whether this supports our expectations or not depends upon statistical results for Periods 4 and 5.

Figure 9.21. Cross sections of manos from all sites from Periods 3, 4, and 5 (observed distributions).

Table 9.37. Resampling comparison of mano cross section by time periods, observed distributions displayed.

The null hypothesis must also be rejected for Period 4 (A.D. 1450-1575). Subrectangular cross sections are significantly less frequent for Period 4 in comparison with Period 3. This indicates that manos were less frequently abandoned without intensive use in Period 4 in comparison with Period 3 but does not directly support our expectations. There do appear be to somewhat greater frequencies of wedge, triangular, and airfoil cross sections for Period 4, especially at habitation sites, but the increase is not statistically significant.

In the final test, the null hypothesis must again be rejected for Period 5 (A.D. 1575-1700). There are significantly lower percentages of plano-convex cross sections than in Period 4. Again, this pattern indicates less frequent abandonment of manos without intensive use, but expected increases in intensive ground stone use as indicated by wedge, triangular, and airfoil cross sections are not evident. Due to the low number of Period 5 habitation sites, these were not included in the tested sample, so they are not shown in Figure 9.21. The exclusion of these sites might be expected to render the expectation more difficult to support with regard to Period 5. Their exclusion, however, does not affect the results for Period 4.

We expected that, given a hypothesized increase in intensity of grinding activity due to the subsistence demands of an aggregated population, mano cross-section types representing more intensive use would increase in relative abundance for sites from Period 4 and later. The statistical evaluations fail to provide any rigorous support for this expectation. Some weak support for the overall hypothesis is found in the fact that little-used manos were less frequently abandoned in Periods 4 and 5. The lack of support for the chosen implication of the hypothesis does not render the general hypothesis of population aggregation disproved. One explanation for the results might be found in the local abundance of cobbles of apparently preferred ground stone raw material. Given this situation, there would be little incentive to continue to use a particular mano once it became so worn as to impede grinding efficiency (or grinder comfort), because it could easily be replaced. Other possible explanations include a shift in disposal patterns for ground stone tools that might be expected with a change to specialized grinding facilities (however, only limited evidence for grinding facilities was found at Pecos Pueblo during excavation by Kidder [1932:71]) or that changes in grinding tool morphology do not accurately reflect aggregation so much as shifts in task specialization and the division of labor (Crown and Wills 1995:180-181).

The results from the analysis of ground stone from the Pecos survey indicate that people in Pecos did not invest a great deal of time or effort into implements for grinding. Unlike many other areas in the Southwest, Puebloan people in the Upper Pecos Valley area rarely created formally shaped grinding implements. Nor do raw materials for these tools appear to have been conserved by using the tools until their utility was exhausted. These characteristics may be a result of an available abundance of appropriate stone for grinding tools. Puebloan groups could afford to procure and abandon ground stone implements at their convenience. Subsequently, trade does not appear to have played a major role in providing material for nonflaked tools. The manos suggest that an expedient ground stone technology prevailed during the entire Puebloan occupation of the park area and that this was little changed during the time of greatest population aggregation.

Summary and Conclusions

The goals of this chapter were to summarize the flaked and nonflaked lithic data collected during the Pecos survey and to use these data to examine the affects of aggregation and trade upon lithic technological organization. Due to the low frequency of sites that could be confidently attributed to pre-Puebloan periods, the chapter focused upon Puebloan lithic technology. Taken as a whole the flaked stone data describe a very generalized lithic economy. The assemblage is dominated by local raw materials but has a specific and seemingly patterned array of nonlocal and exotic varieties. The investigation of platform and debitage types indicates that generalized core reduction debris is predominant in the overall assemblage. Further, technology was primarily expedient in nature, with low-energy and expedient tools being the most common functional types for flaked lithic artifacts.

Raw material use was mostly unspecialized, but patterns of differential use of individual raw material groups are apparent. We presented a simple model predicting the way in which strategies of use and conservation are driven by raw material quality and availability. The expectations derived from the model are mostly supported by the data for tools from the Pecos survey. Local cherts were used for all tool types and do not appear to have been conserved to any great degree. Obsidian, not available locally, was more often used for high-energy tools. Low volumes of unused material indicate that obsidian reduction strategies were designed with conservation in mind. The use of two raw material groups did not meet the predictions of the model. Local chalcedony was used conservatively and preferentially for high-energy tools. Contrary to expectations, nonlocal cherts were treated no differently than local cherts. They were made into the same array of tool classes and do not appear to have been conserved to any great degree.

Obsidian sourcing was carried out for selected artifacts using EDXRF. Examination of the results of the EDXRF analysis indicates the use of a variety of obsidian sources. The abundance in the sample of raw materials from particular obsidian sources is primarily a function of the distance of those sources from Pecos. It also appears that use of the El Rechuelos locality as a source of tool stone decreased through time compared with the Valle Grande and Cerro Toledo sources. As might be expected, obsidian procurement was focused almost exclusively on the Jemez Mountain area. Although direct procurement may have been possible, temporal patterns in the relative abundance of obsidian from individual sources suggest that trade may have played some role in access to obsidian.

Taken together, obsidian and nonlocal chert provide insight into the timing and nature of trade intensification at Pecos. More interaction with the Plains began in Period 4 (A.D. 1450-1575) based on the trends in the chipped stone raw material volumes. This is suggested by the large-scale increase in nonlocal chert during Periods 4 and 5 and the increase in average volume of obsidian in Period 4 that remains high through the end of the Pecos sequence. Because maximum regional population is estimated to have been reached in Period 3 (A.D. 1325-1450), a simple population threshold would not account for the abrupt increase in trade. It appears that only a centralized population, such as that characteristic of Periods 4 and 5, was appropriately poised for substantial trade.

An examination of the tool classes associated with site types provided a way to evaluate the degree to which the habitation sites of later puebloan periods were poised for defense. Kidder (1958) proposed that Pecos Pueblo contained defensive architectural features that indicated an increased concern with warfare in what we are here referring to as Periods 4 (A.D. 1450-1575) and 5 (A.D. 1575-1700). Contrary to expectations, lower percentages of high-energy tools (such as projectile points) were found for these periods, which implies that there was little increased production of flaked lithic defensive weaponry. If there was increased concern with defense during these time periods, it was not manifested through an increase in high-energy tools. We propose that the relatively high frequency of high-energy tools for habitation sites in general is related to the importance of hunting. We further suggest that decreases in high-energy tool frequencies through time may correspond to decreasing emphasis on large mammal hunting. Greater abundance of large mammal remains in the Upper Pecos Valley habitation sites in comparison with that for Arroyo Hondo provides further support for this interpretation.

The parameters of the nonflaked lithic data also indicate a relatively simple and expediently organized technology. We particularly emphasized the examination of ground stone tools and found that little energy investment was made into shaping or maintaining their form. One hand manos were most common and typically did not exhibit substantial morphological change from use. Both slab and basin metates were found to be common, and these also showed little sign of intensive use. Changes in mano cross-section morphology were expected to indicate an intensification of grinding activity associated with population aggregation in Periods 4 and 5. The data for manos provided little support for this expectation. A probable explanation for this, as well as for the general expedience of ground stone technology from Pecos, can be found in the local abundance of stone of appropriate size and texture for use as ground stone tools. The manos suggest an expedient ground stone technology prevailed during the entire puebloan occupation of the park area and was little changed during the time of greatest population aggregation.

In conclusion, the affects of population aggregation upon the lithic technology of the Puebloan occupants of the Upper Pecos Valley are subtle. The social changes that surely accompanied this reorganization of human population had very little effect upon ground stone technology. The most visible effect of aggregation is an increase in trade resulting in greater frequencies of nonlocal and exotic raw materials for flaked stone. With the exception of obsidian, however, these materials do not appear to have played any unique role in the overall flaked stone economy and were probably not the commodity upon which trade was focused.

Notes

1. Evidence for Paleoindian and Archaic occupation of the survey area is based upon diagnostic projectile points. The Paleoindian period is represented by a single Folsom point (IO 96). This point was located at 2,173 m (7,130 ft) in elevation on a ridge top in the Tecolote Mountains.

Three sites appear to primarily reflect Archaic occupation. These sites occur in different environmental locations, but all are on ridge slopes. The first, PECO 300, has a basalt Armijo point, a basalt knife, and a high percentage of lithic material with evidence for use-wear. Most of the debitage appears to be from core reduction of locally available gray chert. PECO 397 has one orthoquartzite Bajada point and two one-hand manos. Much of the debitage shows evidence of tool manufacture or resharpening, and several utilized flakes and formalized knives suggest that this site may represent butchering activities. PECO 694 has two San Jose points and three Basketmaker II points, with three broken dart points that could not be typed. The debitage appears to represent a variety of reduction strategies. No other associated features were found, but 13 one-hand manos and a basin metate were also recorded.

Three more probable Basketmaker sites (in addition to PECO 694 above) were found. PECO 53 and PECO 207 are multicomponent (both have Spanish components) and each includes a pithouse feature dated in the early to mid-800s (see Appendix E). Their lithic assemblages are dominated by local cherts and chalcedony but also include obsidian and possibly Alibates chert. Both PECO 53 and PECO 207 were excavated before the current survey (Nordby 1990; Nordby and Creutz 1993a). PECO 61 is also multicomponent with a possible Basketmaker component present (based upon ceramics). The Basketmaker lithic assemblage from PECO 61 could not be differentiated from the overlying Pueblo assemblage.

2. The term "lithic raw materials," unless otherwise specified, is used here to refer to raw materials for use in reduction by flaking. Sources of ground stone materials are addressed in the nonflaked lithic section.

3. What is traditionally referred to as Pedernal chert is translucent and appears to be composed of quartz in a fibrous form; thus it appears to meet both traditional archeological and petrological criteria for chacedony.

4. What is traditionally referred to as Alibates flint consists of dolomite rock within which agate, a cryptocrystalline quartz, has precipitated resulting in the characteristic banded coloration.

5. This value of 3.3 percent is based on the reported proportion of projectile points and knives from Components I and II from Arroyo Hondo. Drills and gravers are lumped into a single category in the report and thus, unlike the Pecos survey assemblages, drills (which we consider to be high-energy tools) do not contribute to the high-energy tool percentage from Arroyo Hondo (we do not consider gravers to be high-energy tools). If the drills and gravers category is included for high-energy tools from Arroyo Hondo, the value increases to 3.8 percent.