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

Ceramic data from sites in Pecos National Historical Park are used to address several topics outlined in the project research design, including issues of community dynamics and culture contact (Head 1997a). This chapter focuses on three major research domains: (1) analysis of the cultural, temporal, and spatial variability in the Pecos survey ceramic assemblages; (2) the functional content of ceramic assemblages and the implications for social processes such as aggregation and the intensification of agriculture; and (3) the organization of utility and white ware ceramic production in relation to the topic of trade.

The chapter begins with a review of previous ceramic research at Pecos and briefly describes the nature of the ceramic data collected by the survey. A summary of the main ceramic types found within the park is then presented, along with a discussion of the likely cultural affiliation of the various types. To provide a culture historical context for the current research, the temporal and spatial distributions of the ceramic types are discussed. The distribution of possible trade ware is examined, and comparisons are made to ceramics found elsewhere in the northern Rio Grande region, including Rowe Pueblo, Arroyo Hondo Pueblo, and sites in the Galisteo Basin.

Ceramic Research in the Northern Rio Grande Region

Ceramic analysis occupies a central place in the history of research at Pecos, with A. V. Kidder and A. O. Shepard, pioneers in Southwestern archeology, conducting the most significant work of their careers at Pecos Pueblo. While many early Southwesternists (Judd 1922; Kidder 1915, 1916a, 1916b, 1924, 1927, 1932; Mera 1933, 1935; Morris 1939; Roberts 1935) were primarily interested in artifact description, chronology, and defining regional patterns, Shepard (1936, 1939, 1942, 1965, 1985) was the first investigator to reach beyond concerns of cultural historical reconstruction to posit questions about ceramic production and exchange. Shepard's work with ceramic petrography in pursuit of these broader objectives still sets the standard for ceramic research today. Her findings that pottery was often traded over great distances and her recognition that some communities specialized in making certain kinds of pottery, changed archeological thinking, which until then had regarded ceramic production as a purely household-based activity with each household producing ceramics to meet its own needs.

Beginning with Kidder (1916a, 1924, 1932, 1958) and Shepard (1936, 1942, 1965), research in the Upper Pecos Valley has highlighted the role of Pecos Pueblo as a center of commerce and trade in late prehistoric times, an emphasis which has continued in recent years (Habicht-Mauche 1988; Spielmann 1982, 1983, 1991b; Welker 1997). Despite this emphasis, evidence for the nature of ceramic production and trade at pueblos antecedent to Pecos Pueblo and at small sites of all time periods is virtually nonexistent. Likewise, most ceramic studies in the northern Rio Grande region have ,argely centered on the specialized production and exchange of glaze ware ceramics (Bower et al. 1986; Herhahn 1995; Motsinger 1997; Reed 1990; Staley 1990; Vint 1992; Warren 1969, 1970, 1979). Production and exchange of the earlier utility and black-on-white ware is largely unknown. The second half of the chapter addresses this problem by investigating the organization of ceramic production at the early pueblos preceding Pecos Pueblo.

Methods of Ceramic Analysis

Analysis of the Pecos ceramics was performed in the field. Sampling and recording methodologies for the project are presented in Chapter 3. Within each scatter, sherds were grouped together and classified by ware and by ceramic type and form when possible. Since construction of a site chronology (Chapter 4) was a primary goal of the survey, ceramic types were identified as the most effective means of assigning relative dates to most sites, rather than recording other attribute data such as temper, paste characteristics, or design elements. The survey recorded the vessel form for each sherd within the following categories: jar, bowl, shouldered bowl, special closed shape, canteen, soup plate, ladle, handle, pipe, miniature, other, and unknown. Sherds were also weighed in the field by groups of vessel type and form. In addition, counts of rim sherds were recorded, along with metric data on those rim arcs large enough to be measured. The curvature of rim sherds was used to estimate vessel diameter and percentage of the rim present. Rim sherds that could be refitted were measured together on the rim arc template (see Appendix D).

Finally, an experimental variable, the minimum or actual number of vessels present for each ceramic group, was assessed. This variable required analysts to visually compare the paste, surface treatment, curvature, rim forms, and decorative elements of the sherds within each type/form group and to assess the probable number of whole vessels represented by the sherds. Analysts further specified whether this figure approximated a minimum or actual number of vessels. When assemblages were small and only a few sherds of each type were present, the actual number of vessels could usually be determined with reasonable confidence. For large assemblages where this was impossible, the ceramicists made a conservative estimate of the minimum number of vessels present in each type and form class.

The resulting vessel estimates do not seem well supported, except perhaps at the smallest sites where the actual number of vessels represented on the surface could be estimated accurately. The condition of the ground cover on sites also factors into the ability to estimate numbers of surface vessels, as do other formation processes which affect the relationship between surface and subsurface materials. In the end, the estimates were considered too subjective to be of great value. The resulting data on minimum and actual numbers of vessels are reviewed briefly in the section on trends in ceramic wares.

The Ceramic Assemblage

The ceramic database for the Pecos survey is composed of information recorded for a total of 30,709 sherds from 496 site components. Despite the limitations of the field estimates of the number of vessels in each ceramic grouping, these sherds appear to represent a minimum assemblage of 7,480 total vessels, although this figure is probably conservative.

Ceramic Types

The primary unit of analysis for the survey is the ceramic type. Rather than repeat previously published descriptions of individual ceramic types, the reader is referred to Table 8.1 for the references used by the project for ceramic type descriptions, organized by ceramic ware. The Pecos Archeological Survey Ceramic Typology Field Manual (available through National Park Service, Santa Fe) contains the project coding guide and lists complete descriptions of ceramic types with details about paste, temper, surface treatment, slip, paint, and decorative attributes. Table 8.2 is a comprehensive list of the total sherd count for all ceramic types by vessel form. Chronological dates for the ceramic types—often based on tree-ring dates, but also derived from relative stratigraphic information, and in some cases, on researchers' best estimates—are provided in Table 4.1.

Table 8.1. References for ceramic type descriptions by ware.

Table 8.2. Total sherd frequencies for all ceramic types by vessel form.

Ceramic Wares

Individual ceramic types may be subsumed within the more general heading of wares. At Pecos, wares recorded by the survey are as follows: Rio Grande Gray Ware (Colton 1953), Pajarito White Ware (Cordell 1998; Habicht-Mauche 1993), Rio Grande Glaze Ware (Mera 1933), Historic polychromes and plain ware (Batkin 1987; Harlow 1973), White Mountain Red Ware (Carlson 1970), Plains/Apache (Gunnerson 1968, 1969b), and an "Other" category. Table 8.3 summarizes Shepard's (1936) petrographic conclusions regarding the source of manufacture for ceramics found at Forked Lightning Pueblo and Pecos Pueblo. Kidder and Amsden (1931) and Kidder and Shepard (1936) have published thorough descriptions of the primary ceramic wares found at Pecos; they are recapped only briefly below for the reader unfamiliar with northern Rio Grande archeology, along with the survey findings for each ware.

Table 8.3. Shepard's (1936) findings regarding the source of manufacture for ceramics found at Forked Lightning Pueblo and Pecos Pueblo.

Rio Grande Gray Ware

Pecos utility ware has received little attention aside from work by Habicht-Mauche (1987, 1988) on Southwestern-style culinary ware found on the Plains. Hence the sequence of utility ware is poorly understood, and types used for classification purposes are descriptive of surface finishes. From excavation data (Kidder 1958), we know that the earliest utility ware at Forked Lightning Pueblo is of a plain variety, followed by corrugated and indented blind corrugated ceramics. "Indented blind corrugated" is the type name favored by Kidder and still used by archeologists working in the Pecos area, although the term refers to the same surface treatment known as "smeared indented" in other parts of the northern Rio Grande region. Little temporal information exists for the utility ware types, with plain gray, corrugated, and indented blind corrugated all thought to have been produced from approximately A.D. 1100-1500, although these conventional dates are not well documented. Further clouding the picture, some whole Pecos pots in museum collections display a variety of surface treatments on different parts of a single vessel.

Utility ware at Pecos shifts to a smoothed, striated surface treatment with highly polished vessel interiors around A.D. 1500. At first striations are fairly weak, gradually becoming heavier through time (Kidder and Shepard 1936:320). Striated sherds are typically associated with Glaze V ceramics.

As seen in Figure 8.1, based on sherd counts, 30 percent of all utility ware ceramics in the Pecos database are unidentified, partly due to the substantial number of very small, eroded utility sherds found on survey. Next most prevalent are plain gray varieties (25 percent), followed by indented blind corrugated (19 per cent), and striated sherds (18 percent). Small quantities of indented corrugated, corrugated, blind corrugated, and other types are also present, but comprise less than 3 percent each of the total assemblage (Figure 8.1).

Figure 8.1. Percentages of utility ware ceramic types recorded by the survey.

Pajarito White Ware

The designation "Pajarito White Ware" is applied here to the Rio Grande white ware found at Pecos, following its usage for black-on-white ceramics by Habicht-Mauche (1993) at Arroyo Hondo Pueblo and by Cordell (1998) at Rowe Pueblo. Use of the term is not meant to imply that all of this pottery was produced on the Pajarito Plateau. Production of the various black-on-white ceramic types likely occurred in multiple locations throughout the northern Rio Grande region, as discussed in the second half of this chapter.

White ware ceramics found at the Pecos sites, in order of abundance are Santa Fe Black-on-white (27 percent), Galisteo Black-on-white (20 percent), Biscuit A and B (14 percent), and Wiyo Black-on-white (6 percent), all carbon-paint types (Figure 8.2). Approximately a third of all of the white ware sherds recorded by the survey (n=1,335) could not be classified by type. As with the Rio Grande Gray Ware, the large number of unidentifiable white ware ceramics is partly due to the very small, eroded, and exfoliated sherds found on survey at Pecos. White ware pastes—the primary diagnostic feature of the northern Rio Grande white ware ceramics—display tremendous variation, much of which cannot be pigeonholed neatly into the traditional typology. This problem and its implications for source of manufacture are addressed later in the chapter through compositional study of white ware sherds found at the pueblos preceding Pecos Pueblo.

Figure 8.2. Percentages of white ware ceramic types recorded by the survey.

The earliest ceramic type in the black-on white series, and the only type with mineral paint, Kwahe'e Black-on-white, is conspicuously scarce at Pecos. Although 15 sherds of Kwahe'e Black-on-white were recorded, the numbers are significantly less than expected, signaling absence of occupation in the Upper Pecos Valley during the A.D. 1050-1250 period of Kwahe'e Black-on-white use. Other white ware sherds found in small quantities are Pindi Black-on white (n=23), Rowe Black-on-white (n=32), Jemez Black-on-white (n=3), Sankawi'i Black-on-cream (n=12), and Chupadero Black-on-white (n=6). The small quantities of Kwahe'e Black-on-white and Jemez Black-on-white in the survey area are in keeping with Cordell's (1998) findings from Rowe Pueblo.

The lack of Rowe Black-on-white at the Pecos sites is surprising, given the park's proximity to Rowe Pueblo, the type site for this pottery. However, Rowe Black-on-white is defined primarily as a "poor-quality variant" (Habicht-Mauche 1993:29) and a "late, degenerate form" (Kidder and Amsden 1931:26) of Galisteo Black-on-white. Noted for extreme cracking of its painted surfaces, Rowe Black-on white exhibits pastes very much like Galisteo Black-on-white, making classification of sherds into this category somewhat subjective. It is possible that some Rowe Black-on-white sherds may have been unrecognized within the range of variation for Galisteo Black-on-white, but it is more likely that Rowe Black-on-white saw little use at sites outside of Rowe or Pecos pueblos.

Rio Grande Glaze Ware

Glaze paint ware, ubiquitous at Pecos, has been the subject of much study in the northern Rio Grande. The temporal sequence of changing rim styles, modified after Kidder and Kidder (1917) and Mera (1933), is well documented in Kidder's excavations at Pecos, and serves as an excellent chronological indicator for surface assemblages. Rim forms become gradually thicker over time, changing from straight and direct profiles to bulging polliwog shapes by the later part of the sequence. In the interest of direct comparability with Kidder's published work from Pecos, glaze ware designations for the project follow Kidder's original nomenclature, Glaze I-VI, rather than the Mera series Glaze A-F used in other parts of the northern Rio Grande region.

Shepard's early petrographic work on glaze ware produced eye-opening results, particularly for Kidder, who had to revise his ideas about ceramic production at Pecos. Combining stratigraphic and petrographic evidence, Shepard was able to demonstrate that the earliest glaze ware ceramics found at Pecos were almost certainly nonlocal, since they were tempered with crushed igneous rock not found in the immediate vicinity (Shepard 1936, 1942). The source of igneous rock closest to the Pecos Valley is the Galisteo Basin area, including the Cerrillos Hills, Ortiz Mountains, and San Pedro Mountains.

Shepard also observed "technically inferior" examples of Glaze I Red (A.D. 1315-1425) painted with thick, matte glaze and tempered with local sand, siltstone, and sandstone, which appeared slightly later in time (Shepard 1936:72, 73, 517). The change in raw materials and quality of the Glaze I vessels suggested to Shepard that Pecos potters copied trade ware vessels using locally available materials. In this explanation, the slightly later variant forms of Glaze I represent a period of learning to produce a new ceramic technology (Shepard 1936:520, 607). Another possible explanation may be that migrant potters versed in glaze ware technology began to manufacture glaze ware at Pecos using locally available, but unfamiliar and different raw materials.

Shepard (1936:520; 1942:181) found Glaze II through Glaze V (A.D. 1400-1700) sherds were tempered with local siltstone or stream sand, indicating a locally-made product. She contrasted the local Pecos Glaze II through IV ceramic sequence to the igneous rock-tempered glaze ware found at pueblos in the Galisteo Basin from the same time period. The Galisteo Basin pueblos appear to have been major producers of glaze ware pottery found in the northern Rio Grande region during the fifteenth century, but very little of this pottery occurs in deposits from Pecos Pueblo (Shepard 1942; Snow 1981; Warren 1969). This lack of trade ware ceramics coming into Pecos during Glaze II through Glaze V times may indicate that the Pecos community did not share a close relationship with its Galisteo neighbors. In addition, the early glaze ware types (Glaze I—Glaze IV) produced at Pecos do not appear to have been traded outside of the Upper Pecos Valley (Shepard 1942). Shepard characterized this as the time of the "greatest ceramic independence" at Pecos (Shepard 1936:520). However, later in time, some locally-made Pecos Glaze V (A.D. 1515-1700) ceramics with distinctive Pecos rim forms are found in deposits outside of the Pecos area, including Galisteo Basin sites, Rio Arriba, and near Jemez Canyon (Shepard 1942:154-155).

The pattern of local production of Pecos glaze ware pottery appears to have changed by Glaze VI times (A.D. 1625-1700) when glaze ware sherds were once again tempered almost solely with nonlocal igneous rock (Kidder and Shepard 1936). This change in raw materials indicates that the organization of ceramic production had shifted again by the seventeenth century, with the villagers of Pecos Pueblo now relying on external groups to supply them with glaze ware pottery. The reappearance of nonlocal igneous temper may signal renewed ties to the Galisteo Basin community at this time. By A.D. 1700, Rio Grande Glaze Ware pottery ceased to be produced altogether in the northern Rio Grande region, as the Spanish may have assumed control of the Cerrillos lead-ore sources by this time (Peckham 1990; Snow 1982).

Rim forms are the defining typological characteristic of the Rio Grande Glaze Ware sequence, although some surface treatments and glaze ware designs are also diagnostic. For this reason, body sherds generally cannot be classified confidently to type. Glaze body sherds were recorded by the survey as glaze-on-red, glaze on-yellow, glaze polychrome, or unknown glaze. Of the identifiable survey glaze ware sherds, 23 percent are Glaze I, 3 percent are Glaze II, 17 percent are Glaze III, 17 percent are Glaze IV, 38 percent are Glaze V, and 2 percent are Glaze VI (Figure 8.3). The Glaze II variant was never very common throughout the northern Rio Grande (Snow 1989), accounting for the low numbers seen here.

Figure 8.3. Percentages of identified glaze ware types recorded by the survey.

A distributional map of glaze ware and black-on-white sherds recorded by the survey reveals that both wares are abundant through out the park and tend to occur at the same sites, indicating reuse of locations through time (Figure 8.4). Glaze ware is the predominant ceramic type found east of the Pecos River. Presumably aggregation and increasing population led to the use of previously unexploited portions of the landscape (i.e., the northeast section of the park) by glaze ware times. An unexpected number of isolated glaze ware pot drops were recorded by the survey east of the Pecos River (n=40), also indicating a different land use pattern in this portion of the park during glaze ware times.

Figure 8.4. Distribution of black-on-white and glaze ware sherds. (click on image for an enlargement in a new window)

Historic Polychromes and Plain Ware

A new ceramic tradition emerged in the late seventeenth century. Vessels of this tradition, known as matte-paint polychromes and plain ware, are found on the survey only in small quantities outside of Pecos Pueblo. Tewa Polychrome is the most plentiful type (n=104), but smaller quantities of Tewa Black-on-red (n=2), Sakona Polychrome (n=5), Pojoaque Polychrome (n=1), and Ogapoge Polychrome (n=7), are also present. Historic Rio Grande Plain Ware found on the Pecos survey includes undecorated red (n=722), high polish undecorated red (n=91), Plain Red (n=188), Plain Black (n=469), Kapo/Tewa Black (n=30), and unknown plain polished (n=292). The high polish undecorated red sherds may be undecorated potions of polychrome vessels. Only two sherds of Los Lunas Smudged were observed.

Historic polychromes and plain ware ceramics continue the pattern of red slipping established in glaze ware ceramics, but the use of lead-based glaze is superseded by decoration with organic-based matte-paint. Potters also took advantage of their ability to control firing conditions: the ferric oxide slip on the plain ware vessels turns red when fired in an oxidizing atmosphere, and black when fired under reducing conditions (Guthe 1925). Novel design elements such as flowers and birds were also introduced on decorated vessels.

Historic Puebloan-produced polychrome and plain ware ceramics were also used by Hispanic and Anglo settlers, leading to the development of new vessel forms to suit European purposes. Still formed using a coil-and-scrape method, the new historic ceramic shapes include soup plates, candlesticks, chamber pots, and large storage vessels. Specialized ceramic "bleeding bowls" with a channeled rim segment have also been identified in collections from Pecos Pueblo (Shawn Penman, personal communication 1998). With increasing trade via the Santa Fe Trail and the eventual arrival of the railroad, historic Puebloan ceramics in the northern Rio Grande region were gradually replaced by manufactured metal containers.

Plains/Apache

Contrary to expectations, only a few, small sherds of Plains/Apache affiliation were detected on survey, although ceramic analysts maintained a close watch for non-Puebloan pottery. Figure 8.5 plots the distribution of Apache ceramics within the park. Most occur at sites very near the course of the Pecos River, a likely route of travel through the valley. Forty sherds classified as Ocate Micaceous were recognized by their thin walls, micaceous paste, and laminated texture. One sherd of Cimarron Micaceous was observed. The survey noted only one example of Potsuwi'i Incised, a possible Plains-influenced type (Harlow 1973; Wendorf 1953).

Figure 8.5. Distribution of Apache ceremaics. (click on image for an enlargement in a new window)

While Gunnerson's (1968, 1969b) excavations and numerous documentary sources (Hammond and Rey 1940; Kessell 1979; Schroeder and Matson 1965; Spielmann 1991a) support evidence for large-scale Pueblo-Plains trading at Pecos, only three sites recorded by the Pecos survey could be affiliated with Plains or Apache culture groups. Even in these instances the Plains cultural affiliation was assigned with low confidence because at least an equal amount of Puebloan material was present on site, probably indicating reuse of locations over time by different groups of people.

Two of the sites with Ocate Micaceous ceramics also have other diagnostic Plains/Apache traits. PECO 200—which dates to ca. A.D. 1475-1600—has a high proportion of obsidian and a metal projectile point. The second site (PECO 366), an extremely large artifact scatter dated to A.D. 1450-1838, includes an end scraper and a basal notched projectile point, often considered to be diagnostic Plains attributes. Aside from problems of identifying cultural affiliation using survey data, the dearth of Apache ceramics and other diagnostic materials on sites may result from a contact situation in which travelers from the Plains had little need to distinguish themselves from the local inhabitants of Pecos in terms of material culture (Head 1998).

Other Trade Ware Ceramics

Despite the role of Pecos Pueblo as a center of commerce, surprisingly few trade ware ceramics from outside the northern Rio Grande region were encountered on survey. The most common trade ware is White Mountain Red Ware (n=55) from the Cibola series (Carlson 1970), including both St. Johns Polychrome and St. Johns Black-on-red. Figure 8.6 shows the distribution of White Mountain Red Ware sherds across the park, with many found in the vicinity of Forked Lightning Pueblo. Hopi sherds, including Jeddito Black-on-yellow and Sikyatki Polychrome, account for 15 of the trade ware sherds. Pecos Pueblo was the location of the only majolica sherd recorded on survey. Earlier in the century, Kidder found other unusual trade ware at Pecos in small quantities. Table 8.4 presents a comparison of the survey findings with Kidder's excavation data on trade ware sherds originating out side of the northern Rio Grande region and found at Pecos (Kidder and Shepard 1936).

Figure 8.6. Distribution of White Mountain Red Ware sherds. (click on image for an enlargement in a new window)

Table 8.4. Comparison of survey findings and Kidder's (1936) excavation data on trade ware ceramics (originating outside of the Rio Grande region) found in the Pecos area.

Although the numbers from the survey data are small, nonlocal trade ware ceramics indicate that the Pecos community participated in broad exchange networks which at times extended far beyond the northern Rio Grande region. Trade ware ceramics are not restricted to Pecos Pueblo itself, but tend to cluster around the core monument area of the park and around Forked Lightning Pueblo (Figure 8.7).

Figure 8.7. Distibution of trade ware sherds. (click on image for an enlargement in a new window)

Vessel Form

With regard to vessel form, 15,753 (51.3 percent) of the sherds recorded by the survey are jars, while 10,668 (34.7 percent) are bowls (Table 8.5). Sherds of unknown form are the next most frequent category, totaling 3,865 (12.6 percent) items. Shouldered bowls are represented by at least 322 (1.1 percent) of the sherds, although this number is unrealistically low because this particular morphological class can usually only be distinguished by sherds which come from the shoulder or rim portions of such vessels. Other ceramic forms are present in smaller quantities, including canteens, soup plates, ladles, handles, pipes, and miniature vessels, but each of these forms comprise less than 1 percent of the total survey ceramic assemblage (Table 8.5).

Table 8.5. Total frequencies of sherds by vessel form.

Trends in Ceramic Wares

Quantities of ceramic wares were examined using three variables recorded by the survey, including sherd count, the weight in grams, and the estimated minimum numbers of vessels. Table 8.6 presents a comparison of these measures by ware class for all time periods. Two variables, sherd count and weight, provide support that utility ceramics are the most prevalent ware at Pecos (at 49.3 percent and 39.4 percent, respectively). The variation in percentages may be explained by the observation that utility sherds at Pecos tend to be smaller than decorated sherds, and thus weigh less; the mean weight of utility sherds is 4.2 grams vs. 5.8 grams for decorated sherds. This pattern is likely due to differential uselife and breakage patterns of ceramics (Mills 1989; Schiffer 1987). Estimating the minimum numbers of utility ware vessels proved much more difficult than estimating the number of the more distinctive decorated vessels, making this variable a less useful measure.

Table 8.6. Comparison of sherd counts, weights, and estimated vessels by ware.

Rio Grande Glaze Ware follows utility ceramics in abundance, although the percentage ranges from 29.5 percent of the total sherd count to 40.4 percent of the total estimated vessels. The minimum estimated vessel percentage for glaze ware is relatively higher than other wares, probably due to the ease of distinguishing sherds from different glaze ware vessels. The next most common ceramic ware recorded by the survey is Pajarito White Ware. White ware makes up 14.3 percent of the overall assemblage based on simple count (17.2 percent based on weight, and 22.2 percent based on vessel estimates). Continuing in decreasing order of abundance are Rio Grande Plain Ware (5.9 percent by count), historic polychromes (0.7 percent by count), and White Mountain Red Ware (0.2 percent by count). Percentage differences between estimation methods for these types are small.

While all three variables provide broadly similar results, estimates of the minimum number of vessels appear to be more subjective and are greatly influenced by the type of ware analyzed. Sherd count and weight provided consistently closer percentages. There is no consensus in the archeological literature about which measure is more accurate. Some studies (Solheim 1960) advocate using both sherd count and weight for the benefit of more information in the case that the two measures provide dissimilar results. Others (Glover 1972:96, cited in Orton et al. 1993) have concluded that either measure "would be quite accurate as a measure of frequency." Since sherd counts and sherd weights for the Pecos data yielded percentages that are similar, the analyses presented here rely on sherd counts for simplicity, ease in statistical manipulation, and comparability with other studies.

Ceramic Wares through Time

Ceramic wares were divided into two categories, decorated and undecorated, in order to examine changes through time (Table 8.7; Figures 8.8 and 8.9). For this analysis, decorated ware is defined as all painted ceramics. Undecorated sherds outnumber decorated sherds in all time periods except Period 4 (A.D. 1450-1575), where decorated ware accounts for 51.9 percent of the overall assemblage. Most prehistoric Puebloan ceramic assemblages are dominated by utility ware (Colton 1953; Gifford and Smith 1978; Morris 1917; Stone 1986), so it is surprising to find that decorated ware is more numerous at Pecos in Period 4. This pattern suggests that the utilitarian functions assumed for undecorated pottery (primarily cooking) were not as important as the functions associated with decorated pottery (serving or storage) during A.D. 1450-1575.

Table 8.7. Frequency of decorated and undecorated wares by time period.

Figure 8.8. Percentages of decorated and undecorated sherds by time period.

Figure 8.9. Decorated and undecorated wares by time period.

Gauthier (1982) notes a similar trend in proportions of decorated and undecorated ware for Pueblo IV sites in the northern Rio Grande. He finds that by the A.D. 1400s, decorated ware comprises 50 percent of the overall ceramic assemblage, even at field house sites. Gauthier (1982:7) suggests that the increased numbers of decorated ware at this time may be due to increased trade, redistribution, or craft specialization.

Rio Grande Gray Ware through Time

Because relatively little is known about the temporal sequence of utility ware ceramics at Pecos, a breakdown of utility ware types by Pecos time period was created (Table 8.8). Although there is some latent circularity in looking at utility ware by time period when the presence of these ceramic types was in some instances used to classify sites by time period, the shorter date ranges of decorated types had a greater effect on the ceramic assemblage date assignment than did the longer utility ware production intervals. With this caveat, utility ware types display some interesting trends.

Table 8.8. Frequency of utility ware types by time period.

Periods 1 (A.D. 1075-1200) and 2 (A.D. 1200-1325) are dominated by corrugated ware, while in Period 3 (A.D. 1325-1450), plain and corrugated ceramics, along with unknown types, each make up a third of the assemblage. The trajectory of Plain Gray utility ware peaks in Period 3, although it is possible that some of these plain sherds may be small pieces of corrugated or striated vessels without distinguishing attributes. Unknown utility ware—usually sherds too small to be classified accurately—likewise peaks in Period 3. By Period 4 (A.D. 1450-1575), plain varieties account for most of the identifiable assemblage—an unexpected finding, given that striated ware is thought to be the dominant utility ware type by A.D. 1500. Perhaps striated ceramics do not come into vogue at A.D. 1500, but appear slightly later in the sequence, or possibly they do appear around A.D. 1500, but the transition is more gradual. As expected, striated ceramics make up the largest proportion of the total utility ware assemblage in Period 5 (A.D. 1575-1700).

Ceramic Ware by Site Type

One goal of this chapter is to evaluate whether differences in assemblage composition are apparent between each of the three site types defined by the project typology: habitation, seasonal, and special-use. Based on the descriptions by Head (Chapter 5), sites defined as habitations are thought to be full-time residences with long occupations where a variety of domestic activities took place. At habitation sites, we expect to find a wide range of ceramic wares represented in large quantities. Given expectations concerning aggregation and increased trade (Head 1997a), we also expect to find the most trade ware ceramics at habitation sites.

Sites defined as seasonal are hypothesized to have been the locations of more limited activities by small groups of people, including sites that were part-time residences, field houses, or served other agriculture-related functions. Because seasonal sites are thought to have shorter durations of use and were the locations of fewer activities, we expect to find a more limited array of ceramic wares than at habitation sites. Trade ware is not expected to occur at seasonal sites.

Finally, sites defined as special-use are considered to have been the locations of small scale resource procurement or processing (Head, Chapter 5). Used only briefly, perhaps only on a single occasion, special-use sites are expected to contain a very limited range of ceramic wares. Trade ware ceramics should be absent.

Differences in site types were first explored through examination of ceramic wares found at each of the three site types (Table 8.9; Figure 8.10). The analysis began by testing the null hypothesis that there are no significant associations between site type and ceramic ware type. A chi-square test led to rejection of the null hypothesis, instead revealing significant associations between ceramic ware and site type (Table 8.9). Contributing to this pattern, Rio Grande Gray Ware was found to be the predominant ware at all three site types, which was unexpected given the differences in the site types posited above. Only at habitation sites did utility ware sherds comprise more than half of the composite assemblage, which is expected for long-duration sites. Rio Grande Glaze Ware is the second most common ware at all three site types. Glaze ceramics represent about a third of the composite assemblages of seasonal and special-use sites, and about 23 percent of the composite assemblage of the habitation sites. White ware and plain ware make up most of the remainder of the assemblage at all three site types. The trends in wares just described for habitation sites may be seen graphically in Figure 8.11, which plots the percentages of ceramic wares present at the early pueblos.

Table 8.9. Frequencies of ceramic wares by site type.

Figure 8.10. Percentages of ceramic wares by site type.

Figure 8.11. Percentages of ceramic wares recorded on the surface at the early pueblo sites.

Stronger trends are apparent in the distribution of each ware across the three site types (Table 8.9; Figure 8.10). Utility ware is most likely to be found at habitation sites. White ware and White Mountain Red Ware are more prevalent at seasonal sites. Glaze ware occurs in the highest numbers at special-use sites. Historic polychromes are most common at special-use sites and rare at habitation sites. More plain ware sherds are found at habitation and seasonal sites.

As expected, habitations are found to have a large amount of a variety of wares, reflecting many activities and a long span of use. Trade ware ceramics including White Mountain Red Ware are found at habitation sites, but contrary to expectations, they are also found in equal numbers at sites defined as seasonal and special use. Also counter to expectations, seasonal and special-use sites exhibit a variety of wares, perhaps indicating that the functions and duration of use at these sites were not as limited as expected.

Functional Analyses

In this section of the chapter, differences in site function are explored through analysis of vessel form and through evidence of how pottery was used at sites. Comparisons are made between assemblages found at the three site types. This portion of the analysis addresses whether the site types defined by the Pecos survey typology served different functions from each other, from the perspective of the ceramic evidence. Secondly, ceramic morphology is considered with regard to the Pecos chronology in order to evaluate whether ceramic functional classes and site function changed through time. Finally, drawing on both ceramic functional classes and on rim size data, vessel size is discussed, with implications for food preparation, serving, and vessel storage.

The functional classes used in this portion of the study are (1) decorated bowls, (2) decorated jars, and (3) undecorated jars. Undecorated bowls are not present in the Pecos ceramic assemblage, except in minute quantities. In the American Southwest, researchers often infer ceramic function based on vessel shape within these categories, although the relationship is not always direct. Single vessel forms were certainly used for a variety of purposes. Nevertheless, decorated bowls are generally assumed to have been used for serving food, given their relatively large and unrestricted orifice sizes, although short-term storage or mixing functions cannot be ruled out. The painted surfaces of bowls indicate that this vessel class was not used for cooking over a fire, since the paint decoration would have been obliterated by soot.

In contrast to bowls, jars have a more restricted orifice with a large volume-to-aperture ratio, as well as greater stability, making this vessel form more suitable for cooking and storage functions. For the purposes of the following analysis, undecorated jars are assumed to have been used for cooking. (The presence or absence of sooting on vessels was not a variable recorded by the survey.) This analysis also makes the assumption that decorated (painted) jars were used primarily for storage of liquid and dry goods.

Similar inferences regarding prehistoric vessel form and function have been employed in studies by Linton (1944), Henrickson and McDonald (1983), Mills (1985, 1989, 1999) and Smith (1985, 1988), based on cross-cultural ethnographic data. Bowls, as serving vessels, demonstrate the most direct relationship between ceramic morphology and use, while more variation in the relationship between form and function is present among cooking pots and storage jars (Henrickson and McDonald 1983; Mills 1985:3).

The following analyses also operate from the assumption that ceramic assemblage content is a reasonable indicator of the activities that occurred at a particular site. However, cultural formation processes, including differential breakage rates, ceramic uselife, and duration of site occupation also condition assemblage content (Mills 1989; Schiffer 1987). For this reason, differences in the distributions of the three ceramic functional classes may not directly reflect the activities performed at the different site types. One important consideration is that sites occupied for a short time (fewer than 10 years) tend to differ greatly in assemblage content, despite having similar functions (Mills 1989). Proportions of ceramic functional types in long-duration assemblages will be more stable but may look different from short-duration assemblages at sites of similar function (Mills 1989). Despite the possible influences of these factors, this study lacks the data to control for short-duration occupations, given the nature of surface artifacts and the long ceramic type ranges present at Pecos (see Chapter 4).

Functional Classes through Time

Frequency distributions of the ceramic functional classes are expected to change through time given known changes in aggregation and in population size during the Puebloan occupation of the Pecos area. Aggregation and population size are linked to the intensification of agriculture, which we would expect to see reflected in the ceramic data. For example, with aggregation we may find increased storage capacity, as seen in greater quantities or proportions of decorated jars through time. In fact, quantities of sherds in all ceramic functional classes are expected to increase through time, as more people and agricultural intensification increased the need for cooking, serving, and storage.

Turning to the data, a rejection of the null hypothesis that there is no significant association between functional class and time is warranted by a chi-square test (Table 8.10). Contributing to this result, ceramic assemblages of all time periods are dominated by undecorated jars (Table 8.10; Figure 8.12). Undecorated jars occur in the highest numbers in Period 3 (A.D. 1325-1450), before leveling off in Periods 4 (A.D. 1450-1575) and 5 (A.D. 1575-1700). Decorated bowls are most numerous in Period 3, followed by Periods 4 and 5. The sharp increase in decorated bowls between Periods 2 (A.D. 1200-1325) and 3 is most notable. For decorated jars, the distribution increases steadily between Periods 3, 4, and 5.

Table 8.10. Frequencies of ceramic functional classes by time period.

Figure 8.12. Percentages of ceramic functional classes by time period.

Looking at the proportions of the ceramic functional classes, undecorated jars approach 60 percent of the total assemblage in Periods 1 (A.D. 1075-1200) and 2 (A.D. 1200-1325), before dropping to 43-49 percent of the total assemblage in the later time periods (Figure 8.12). The proportion of decorated jars within the overall ceramic assemblage increases over time, from about 4 percent in Period 1 to 15.5 percent in Period 6 (post A.D. 1700) (Figure 8.12).

The functional interpretation for these data implies that cooking is the most vessel consumptive function overall, as seen in the numbers of undecorated jars (Table 8.10). However, serving vessels (decorated bowls) are most numerous in Period 3 (A.D. 1325-1450), followed by Periods 4 (A.D. 1450-1575) and 5 (A.D. 1575-1700). Based on quantities of decorated jar sherds alone, storage is greatest in Periods 3, 4, 5, and in particular, exhibits a dramatic increase between Periods 2 (A.D. 1200-1325) and 3. Cooking vessels (undecorated jars) are most prevalent in Period 3, although the numbers level off in Periods 4 and 5.

Revisiting the expectation that quantities of all ceramic functional classes will increase through time, we find that this holds true only for decorated jars, at least through Period 5 (A.D. 1575-1700). Increased quantities of storage vessels are in keeping with the requirements of aggregation and the intensification of agriculture. As expected, quantities of decorated bowls (serving vessels) and undecorated jars (cooking pots) show an increase from Period 2 (A.D. 1200-1325) to 3. Contrary to expectations, the numbers of both decrease in Period 4 (A.D. 1450-1575). Quantities and proportions of decorated bowls continue to decline in Periods 5 and 6 (post A.D. 1700). Numbers of undecorated jars rise again in Period 5, before tapering off in Period 6. What these unexpected trends mean in terms of aggregation at Pecos is not certain. Sizes of vessels through time are examined later in the chapter to shed additional light on this problem.

Functional Classes by Site Type

Differences between site types were further explored through examination of the frequencies of ceramic functional classes by site type (Table 8.11; Figure 8.13). Long-duration, multiple activity habitation sites are expected to show high frequencies of all three ceramic functional classes. By contrast, limited-activity, short-duration seasonal sites should show less evidence of cooking (reflected in undecorated jars) and serving (reflected in decorated bowls). Seasonal sites, as possible field houses, may be expected to have greater quantities of decorated jars for stored provisions. Lastly, lightly used special-use sites are expected to lack decorated jars for storage.

Table 8.11. Frequencies of ceramic functional classes by site type.

Figure 8.13. Percentages of ceramic functional classes by site type.

A chi-square test leads to the rejection of the null hypothesis that there is no statistically significant association between ceramic functional class and site type (Table 8.11). Supporting the significance of this result, the data show that as expected, undecorated jars make up the largest portion of the assemblage from habitation and special-use sites. At seasonal sites, however, undecorated jars and decorated bowls account for an equal proportion of the composite assemblage. Quite unexpectedly, decorated bowls are found in the highest numbers at seasonal sites (37.9 percent), followed by special-use (34.9 percent), and habitation sites (27.3 percent). Decorated jars are also most common at seasonal sites (47.2 percent), followed by special-use (35 percent) and habitation sites (17.8 percent). The low frequency of decorated jars (storage) at habitations is rather surprising. The distribution of undecorated jars looks different from the other two functional classes. Undecorated jars are most common at habitation sites (35.6 percent), followed by special-use (33.7 percent) and seasonal sites (30.8 percent).

Based on these data, cooking was the primary ceramic activity at habitation and special use sites. Cooking and serving were equally important at seasonal sites. In decreasing order of frequency, serving and storage appear to have been most prevalent at seasonal, special-use, then at habitation sites. While our expectations for cooking and serving were upheld for the habitations, quantities of storage vessels were unexpectedly lower at habitations than at special-use sites. Also contrary to expectations, seasonal and special-use sites exhibit relatively high frequencies of decorated serving bowls, a pattern which is inconsistent with the presumed short-term functions for these sites.

Habitation sites are generally thought to be the primary places where domestic activities such as cooking, serving, and storage occurred. However, based on the results of the ceramic functional analyses, it appears that all of these activities also took place at sites defined by this project as seasonal or special-use. It may also be that the inferred functions of each vessel class are too general to truly understand site function. Vessel forms may actually represent a compromise between several intended functions (Rice 1987), and assuming a decorated jar was only a storage container, for example, may be too simplistic. The low percentages of decorated jars, as seen in Table 8.11, may suggest that undecorated jars also served as storage containers. Other possible functions for ceramic vessels, such as resource or food processing, are largely ignored by the functional divisions used in this discussion.

However, it is also likely that the supposed limited-activity sites where these vessels are found were the locations of everyday domestic activities, including cooking, serving, and storage. In a similar finding, Sebastian (1983) concluded that field house assemblages primarily reflect domestic activities. The lack of significant differences in the Pecos ceramic assemblages from the different site types as defined by architecture suggests that people carried ceramic vessels from their domestic inventory at habitation sites to use at other kinds of sites.

Vessel Size

Ceramic vessel size was examined using rim diameter. The categories of undecorated jars, decorated jars, and decorated bowls were maintained for this portion of the analysis in order to control for functional differences. The assumption is made that rim diameter is a useful proxy of overall vessel size for all form classes. While Mills (1999:106-107) finds a direct linear relationship between rim diameter and vessel size for bowls, she argues that vessel height is a better indicator of jar size. Despite Mills's findings, the sherd-based analysis conducted here does not have access to height data and necessarily relies on rim diameter as a measure of size for all vessel forms.

Since ceramics were certainly used as containers (Braun 1983), vessel size can provide meaningful information about the quantities of food or liquid used prehistorically. Size of vessels may provide data on storage capacity. The size of a social group who used particular vessels may also be indicated by vessel size, particularly of cooking and serving vessels (Turner and Lofgren 1966). Diet has also been linked to the size of ceramic vessels. Blinman (1986) found that larger vessel sizes were associated with a greater dependence on agricultural products. Greater emphasis on boiling as a cooking method may also be linked to increased cooking pot size (Wandsnider 1997).

Thus, larger ceramic vessels are associated with the intensification of agriculture and a greater reliance on agricultural products. According to Mills (1999), at least three interrelated factors account for this phenomenon: changes in the organization of food preparation, increases in household size, and feasting. It is difficult to sort out which of these factors is most directly linked to vessel size, and all three likely contribute to the pan-regional trend toward increased vessel sizes through time in the American Southwest (Mills 1999).

Vessel sizes at the Pecos sites were first approached through visual inspection of the rim diameter data. Histograms of orifice diameters by functional class do not support the existence of clear-cut size classes for each vessel form. A histogram of all decorated bowl diameters indicates a unimodal distribution (Figure 8.14). The histogram of undecorated jar diameters (Figure 8.15) shows peaks at 16 cm and at 28 cm. Larger undecorated jars (those 32 cm and greater) are represented in smaller frequencies (Figure 8.15), perhaps suggestive of a possible second size class to meet the cooking requirements of larger groups of people. The mode for decorated jars is 24 cm, with smaller quantities of decorated jars in the larger than 24 cm size range (Figure 8.16).

Figure 8.14. Histogram of bowl diameters.

Figure 8.15. Histogram of undecorated jar diameters.

Figure 8.16. Histogram of decorated jar diameters.

Ceramic Vessel Size through Time

Vessel size at Pecos is expected to increase through time for all vessel classes, given greater aggregation through time. As population increased in the Pecos area, so did household size (Kidder 1958; Welker 1997). We expect to see larger cooking and serving vessels through time as an expression of this change in the commensal unit. Increased aggregation is also thought to be accompanied by agricultural intensification. Such intensification should be seen most clearly in greater storage capacity through time, although it may also be seen in cooking pot and serving bowl sizes.

The null hypothesis to be tested in this case is that the mean ranks of the rim diameters by ceramic functional class are not significantly different through time. Because the vessel size data were found to be nonnormal (through visual inspection and based on a Komolgorov-Smirnov test of normality) the nonparametric Kruskal-Wallis test is appropriate. The Kruskal-Wallis test essentially normalizes the data and compares observed mean ranks to expected mean ranks (Sokal and Rohlf 1981:429-432). The results reported at the bottom of Table 8.12 for each functional class do not allow us to reject the null hypothesis. Thus, the slight changes in rim diameter through time are not statistically significant.

Table 8.12. Mean vessel size in rim diameter by time period.

Table 8.12 and Figure 8.17 show the temporal changes in mean vessel size for each functional class. Although not a statistically meaningful finding, serving bowls do become slightly larger through time, from a 26.7 cm mean diameter in Period 2 (A.D. 1200-1325) to 28.4 cm in Period 5 (A.D. 1575-1700). Mean rim diameter of bowls drops slightly to 28 cm by Period 6 (post A.D. 1700). Unexpectedly, decorated jar size decreases during Periods 3 (A.D. 1325-1450) and 4 (A.D. 1450-1575) but then increases to 26.2 cm in Period 6. As predicted, undecorated utility jars steadily increase from 21.9 cm in Period 1 to 24.4 cm by Period 5, before decreasing in Period 6.

Figure 8.17. Mean vessel size in rim diameter by time period.

Although the results are not statistically significant, there is a pattern of slightly increased cooking jar and serving bowl size through time. However small, such an increase may be linked to a greater dependence on agricultural products and/or an increase in the size of the household unit. Slightly larger serving bowl sizes through time may be indicative of larger commensal units through time, although the 1.7 cm difference between Periods 2 (A.D. 1200-1325) and 5 (A.D. 1575-1700) is not great, given the long time span.

Decorated jars, likely to have been used for storage, display an unexpected decrease in size and hence storage capacity from Period 2 (A.D. 1200-1325) to Periods 3 (A.D. 1325-1450) and 4 (A.D. 1450-1575). However, a jump from 21.9 cm in Period 4 to 25 cm in Period 5 (A.D. 1575-1700) indicates a greater need for storage by A.D. 1575. There is also a continued increase in decorated jar diameter in Period 6 (post A.D. 1700), possibly indicating greater storage size.

Vessel Size by Site Type

Expectations for vessel size at the three different site types were also formulated. Vessel sizes are expected to be largest at the habitation sites, since these are primary residences and places where aggregation is most apparent. Larger decorated jars at habitations would indicate more storage. Larger undecorated jars and decorated bowls are also expected because habitation sites were the presumed locations of larger family groups and the likely places of group consumption, including possible feasting activities. Seasonal sites are expected to have smaller vessel sizes in all classes, because they are thought to have been used by smaller groups of people. Special-use sites are expected to have the smallest vessel sizes of the three site types.

Kruskal-Wallis tests were performed for each of the ceramic functional types to test the null hypothesis that mean ranks of the rim diameters by ceramic functional class are not significantly different at different site types. None of the results (Table 8.13) allow us to reject the null hypothesis, and we must conclude that the site types do not differ statistically in vessel size.

Table 8.13 allows us to evaluate the individual expectations for size by site type. Mean vessel size in rim diameter for decorated jars is only slightly larger at habitations than at the other site types. This does indicate slightly greater storage capacity at habitations, as expected, but the statistical comparison with mean decorated jar sizes at other site types is not significant. Mean bowl size is surprisingly larger at seasonal sites by 1 cm (28.9 cm). Unexpectedly, seasonal sites equal habitation sites in undecorated jar rim diameter. The data for special-use sites match our expectations. The mean rim diameter for decorated bowls is 26 cm, 2.9 cm smaller than at seasonal sites. Decorated bowl, decorated jar, and undecorated jar rim diameter means are smallest at special-use sites, perhaps indicating slightly less serving, storage, and cooking capacity. While suggestive, these slight changes are not statistically significant, however.

Table 8.13. Mean vessel size in rim diameter by site type.

Vessel Size by Ceramic Type

A final size analysis compared the mean rim diameter of certain ceramic types (Table 8.14). This analysis divided samples by ceramic type and vessel form. Within the utility ware category, earlier Indented Blind Corrugated types (A.D. 1000-1500) were compared to later Pecos Striated vessels (A.D. 1500-1700), which were found to have a 2.8 cm larger mean rim diameter. However, a t-test of the null hypothesis that the mean rim diameters of Indented Blind Corrugated and Striated are samples of the same population required acceptance of the null hypothesis. Therefore, while striated vessels do have a larger mean rim diameter than Indented Blind Corrugated pots, this is not a statistically significant finding.

Table 8.14. Mean rim diameter by ceramic type.

Decorated bowls, listed roughly chronologically by type in Table 8.14, were also compared by mean rim diameter. As in the previous analyses, larger vessel sizes were expected through time, given increasing aggregation, household size, and agricultural intensification. An ANOVA to test the null hypothesis that there are no significant differences between mean rim diameters of different ceramic types led to rejection of the null hypothesis. On visual examination of the data, two larger-sized ceramic types stand out: Biscuit ware bowls (A.D. 1375-1550) and Glaze V bowls (A.D. 1515-1700), both over 30 cm in mean rim diameter. The statistical results lend support to the observation that Biscuit and Glaze V mean rim diameters are larger than other ceramic types. Other white ware types and earlier glaze ware ceramics are smaller and roughly the same size as one another.

Spielmann (1998) has argued that the introduction of Glaze I Red ceramics in the Eastern Pueblos was tied to participation in a new ceremonial context involving communal feasting. She finds that larger vessel sizes for Glaze I compared to earlier black-on-white ware support this interpretation. However, based on the Pecos survey data, Glaze I bowls are not significantly larger than the earlier black-on white types (Table 8.14). They certainly represent a departure from the earlier black-on-white tradition in appearance and in ceramic technology. But in terms of size, contemporaneous Biscuit ware vessels were larger and may have been more likely to have been used in a communal context at Pecos.

The size of Glaze V ceramics also indicates a change from earlier glaze ware vessels. It is significant that large Glaze V bowls (A.D. 1515-1700) are produced locally at Pecos at a time when imported Biscuit ware was in decline. It is possible that Glaze V bowls replaced Biscuit ware as a communal serving vessel at Pecos. Compared to the earlier glaze ware vessels, Glaze V bowl sizes may signal an increase in the size of the commensal unit after A.D. 1515, either by large households or suprahousehold groups.

The Organization of Ceramic Production

This portion of the chapter examines the organization of production of utility and white ware ceramics at the early aggregated sites preceding Pecos Pueblo. Of interest is whether the pueblos manufactured their own ceramics from locally available materials, or if there is evidence for trade. The extent of trade and regional exchange during the late Coalition and early Classic periods and what this implies about social interaction are important issues.

Pecos Pueblo has long been recognized as an important trading center, yet aside from Shepard's (1936, 1942, 1965) pioneering work on ceramic sourcing of glaze ware, little material evidence exists to document Pecos' external contacts for the early time periods. The type and intensity of the relationships the Pecos community maintained with other cultural groups is pursued through the study of the organization of ceramic production and distribution.

One influential model for the interaction of social groups in the northern Rio Grande has been put forth by Habicht-Mauche (1993, 1995), based on her work at the nearby Galisteo Basin site of Arroyo Hondo Pueblo. In her view, developing ethnic groups during the Coalition period sought to express different cultural identities through locally and stylistically diverse white ware ceramics. These goods were traded between neighboring communities as part of regularized social interaction involving general reciprocity. Habicht-Mauche argues that the dramatic shift from localized production of white ware ceramics to centralized production of Rio Grande glaze ware around A.D. 1350 was a means of integrating these competing ethnic groups within a "tribal network" of social alliances. The widespread exchange of glaze ware is thought to have functioned as a stabilizing force that united competitive ethnic factions.

Although Habicht-Mauche explains the adoption of glaze ware ceramics within the context of northern Rio Grande developments, Crown (1994, 1996) maintains that it is more appropriate to consider the shift at a pan-regional scale. According to Crown, changes in the makeup of pottery production groups at this time likely relate to the widespread appearance in the Southwest of the new polychrome ceramics, which also had symbolic meaning within an emerging religious ideology. Based on design elements (Graves and Eckert 1998) and on vessel size (Spielmann 1998), other researchers find support for Crown's interpretation. They argue that the distribution of glaze ware ceramics crosscuts ethnic boundaries and economic alliances in the northern Rio Grande and may signal participation in a wider system of shared belief and ritual.

All of these researchers view the Coalition to Classic transition as characterized by a shift to specialized production of decorated ceramics in the northern Rio Grande region. But the question remains whether this was preceded by a slow, gradual change in the organization of production, or a dramatic, rapid shift. Ceramic compositional data have the potential to provide information about the organization of ceramic production and the nature of social interaction before glaze ware ceramics were introduced to the area. Only by understanding the organization of production and distribution of the earlier utility and white ware ceramics as a baseline can we fully assess the possible meanings and impacts of the shift to specialized glaze ware technology in the fourteenth century.

Prehistoric ceramics are particularly well suited to shed light on the nature of contact between the Pecos community and external groups through time. Pottery and other goods traded between groups likely served to cement reciprocal relationships between people, perhaps with trading partners who could be relied upon in times of resource stress or environmental uncertainty (Braun and Plog 1982; Ford 1972). In addition, ceramics almost certainly functioned as containers for the transport of foodstuffs or other goods between distant settings (Toll 1991). Building on Shepard's findings for the Rio Grande Glaze Ware, this research contributes to the knowledge of ceramic technology and the organization of ceramic production at Pecos.

Direct Evidence

The production, distribution, and consumption of ceramics are all part of the organization of ceramic production, but rarely do we find direct archeological evidence for any of these stages. No direct evidence for ceramic production is present at Pecos. Kidder's excavations noted isolated polishing stones and worked sherds, which could have been used in pottery manufacture. Concentrations of other items related to ceramic manufacture (Sullivan 1988), such as raw materials, wasters, high concentrations of tools such as scrapers, and unworked clay are absent. However, Guthe (1925:55) reports that worked, spongy animal bone fragments found at Pecos may have been used for smoothing. Kilns or other ceramic production facilities are unknown, despite test excavations to determine the function of four sites with burned and fire-cracked rock features, selected as possible candidates for kilns (Judy Reed, personal communication 1998). Limited magnetometer surveys at the park (including Korsmo 1983), have recognized pithouses but have failed to identify prehistoric pottery firing locations.

The only formal ceramic firing feature documented for the entire northern Rio Grande region is a pit kiln associated with Santa Fe Black-on-white ceramics in the Las Campanas area northwest of Santa Fe (Post and Lakatos 1995). Its location nearly 8 km (5 mi) from the nearest pueblo suggests that access to fuelwood was an important consideration for firing activities. Some surface evidence of pottery firing kilns at San Lazaro Pueblo in the Galisteo Basin has also been noted, although the potential areas have not been systematically excavated or published (Maxwell et al. 1994) (Orcutt, personal communication 1999).

It is possible that site-oriented research has overlooked other firing features in the northern Rio Grande area, although none have been located by this survey or other recent, intensive surveys (Cordell 1998; Powers and Orcutt 1999; Snead 1995). Alternately, firing regimes in the northern Rio Grande may have been more informal and less specialized than those represented by the Mesa Verde trench kilns (Blinman and Wilson 1993), creating a more indistinct archaeological pattern.

Indirect Evidence

At Pecos we are thus limited to examining indirect evidence for the organization of ceramic production. These measures primarily inform on the scale and intensity of ceramic production (Costin 1991). Indirect evidence available to us here includes the identification of possible production locations using compositional data. Compositional analysis is useful for distinguishing local and nonlocal ceramics in order to identify possible exchange networks. Petrography is widely used, often in combination with other characterization techniques. At a basic level, these methods determine variability in ceramic raw materials, which is often inferred to represent the size or extent of a potting group. In this way, compositional data may be used to infer the degree of craft specialization.

Producer specialization is defined as "an individual who gains part or all of their livelihood through participation in a specialized activity" (Costin 1991:4). Specialization implies production beyond the needs of an individual household for exchange (Costin 1991; Rice 1981, 1984; Stark 1995). Degree of specialization must be considered along a relative continuum. The concept of specialization is most meaningful when comparisons of production are employed—for example, between two different wares or a single ware through time (Costin 1991).

Standardization and Specialization

Standardization is linked to craft specialization, although Stark (1995) has demonstrated that the relationship is not always a direct one and may be influenced by many variables. Standardization (reduction of variability) is best considered as a process related to economic intensification (Rice 1996:178-179). In a basic sense, standardization may be regarded as a measure of the ratio of producers to consumers. Low standardization will generally indicate a low level of specialization, where many potters are involved in the production process. Higher standardization may signal greater specialization in a situation where fewer potters are present, thus introducing less variation into the finished products (Mills and Crown 1995; Rice 1984, 1991, 1996).

Standardization in compositional makeup or in various attributes of the finished products may be interpreted as evidence for craft specialization. Fewer potters will simply introduce less individual variation. In addition, because specialist potters spend more time at their craft, they possess greater skill in production, which in turn translates into less variation in the finished product (Costin 1991; Stark 1991). Routinized production and greater efficiency also contribute to the pattern of less variation (Arnold and Nieves 1992; Hagstrum 1985). Analysis of compositional data, gestures, morphological attributes, the production-step measure, the quality of design execution, and the frequency of misfired vessels have all been used as measures of standardization (Mills and Crown 1995). Of relevance to this study, Hagstrum (1985) analyzed brushstroke gestures apparent in the design elements of black-on-white ceramics from the northern Rio Grande. While based on a small sample (n=20), her results indicate these ceramics became more standardized by the Classic period.

The present research examines evidence for ceramic compositional standardization as an indicator of the relative size and makeup of the potting groups at the early pueblos. The organization of ceramic production may have been quite different for the utility ware and the white ware at Pecos. The overall goal is to distinguish local and nonlocal raw materials used in the manufacture of ceramics found at Pecos.

This study focused in part on utility ware, which has received regrettably little attention from Southwestern archeologists, despite being the most ubiquitous class of ceramics in most assemblages. Utilitarian ceramics are traditionally viewed as products of individual households, with each family group producing pottery to meet its own needs. Although often assumed to have been locally manufactured, utility ware in the Southwest was not always produced for household consumption. Various studies have documented evidence for the exchange of culinary ware (Crown 1981; Glowacki 1995; Shepard 1939; Stark and Heidke 1998; Van Keuren et al. 1997). As Shepard (1939:281) noted, "Assuming a large volume of trade in culinary ware is contrary to accepted ideas, but it seems less improbable when we consider that the making of indented corrugated pottery required a very high degree of skill, and consequently there may well have existed centers where potters specialized in the technique." Nearly 60 years later, Shepard's insight appears to have been overlooked by most archeologists (Cordell 1991).

The present study was designed to investigate change in Pecos utility ware and white ware through time, with respect to compositional at tributes and source of manufacture. The Upper Pecos Valley is an ideal location for this type of compositional and sourcing study, given the great geological variation in the surrounding region. At a basic level, the study was designed to demonstrate whether ceramics were produced locally or nonlocally. In addition, a distinctly different compositional signature for the sherds tentatively identified as Plains ceramics would strengthen our interpretation of these artifacts as trade items.

The early aggregated sites in the Upper Pecos Valley have been little investigated, and their sequences of occupation and relationship to one another are incompletely understood. Of interest to this research was whether the inhabitants of the early pueblos made their ceramics from locally available clays, or if there is evidence for nonlocal production of some of the pottery. This ceramic data has the potential to provide insight into the degree of social and economic interaction that existed between the pueblos.

Ceramic Resources in the Rio Grande Region

The geological diversity of the northern Rio Grande Valley and surrounding region permits study and sourcing of ceramic raw material resources. Important geological features (Figures 1.1 and 9.1) include the Sangre de Cristo Range, a Pre-Cambrian formation that borders the eastern Pecos Valley. Glorieta Mesa, composed of Permian sediments, separates Pecos from the Galisteo Basin to the southwest. To the northeast, exposures of Pennsylvanian limestone are prevalent in the Tecolote Range (Bezy 1981). The Pecos River and Glorieta Creek are the source of secondary deposits of cobbles of granite, pegmatite, amphibole, gneiss, mica schist, quartzite, limestone, and shale, each ground as tempering materials in locally produced prehistoric ceramics (Shepard 1936). Local sands and sandstones were also used for this purpose.

In terms of locally available raw material resources for pottery-making, the shale underlying the Pecos Pueblo mesilla is a likely source for ceramic clays (Nordby 1990:15). Pockets of clay beds are also seen in arroyos cut through the sandstones, siltstones, and shales of the Sangre de Cristo formation. Below the mesilla to the northeast of Pecos Pueblo is a possible collecting and processing work site for sandstone, which was used for temper in some Pecos Pueblo utility ware ceramics (Habicht-Mauche 1988:410). Numerous small ground cupules recorded by the Pecos survey on bedrock outcrops near the Pecos Pueblo mesilla may have also been used as ceramic temper processing areas.

Further west of the Pecos Valley (Figures 1.1 and 9.1), the Jemez Mountains and the Pajarito Plateau are the source of volcanic ash and rhyolite tuff, common ceramic tempers. Basalt, another prevalent inclusion, originates within the Rio Grande rift valley. In addition, the Ortiz Mountains are composed of igneous rock, from which andesite and other feldspathic rocks originate. Lead ore for glaze paint likely came from the Cerrillos Hills and the Ortiz Mountains, although lead ore also may be found in the mountains east of Albuquerque (Shepard 1942).

Certain ceramic clays are specific to sub areas of the Rio Grande. The Pajarito Plateau is characterized by montmorillonitic clays which originate from volcanic glass. Bentonite-like clays found in the northern part of the Pajarito Plateau have been identified as the source for Biscuit ware pottery (Shepard 1936). Along the Rio Grande, sedimentary clays are common. In the immediate vicinity of Pecos Pueblo, Permian sediments contain nonrefractory clays. Shepard (1936:576; 1965) noted that such clays are red-firing, in contrast to the buff-firing Cretaceous clays of the Galisteo Basin. The major geological exposures of clays in the Galisteo Basin occur within the Dakota Sandstone, Mancos Shale, Mesa Verde, and Chinle formations, which are not present in the Pecos Valley (Blinman and Price 1998).

Goals of the Compositional Study

The compositional study was structured to answer the following questions:

1. Is there evidence for the use of nonlocal raw materials in the manufacture of ceramics found at Pecos? Can tempers or clays be matched to raw material samples collected from potential source areas in the Pecos Valley?

2. Is temper and/or paste variability in the ceramics a function of multiple production loci? (Are many compositional groups present, or few?)

3. Is there interior intrasite variation between the early pueblos in composition of the ceramics?

4. How variable are the traditionally defined white ware types? Are Santa Fe Black-on-white and Galisteo Black-on-white compositionally homogeneous or is there more variation within types than between them?

5. Do the tentatively identified Plains ceramics have a distinctly different compositional signature, suggesting a nonlocal origin?

6. Do the pithouse ceramics match the utility ware ceramics from the later pueblos in composition, or is there evidence for manufacture in a different area?

Assumptions of Compositional Analysis

Any compositional study of ceramics requires acceptance of several assumptions. The "provenience postulate," a common archeological assumption in the analysis of compositional data, states that "identifiable...differences exist between sources of a raw material, and the analytical approach can recognize these differences" (Bishop et al. 1982:301; see also Rands and Bishop 1980:19; Rice 1987:414; Weigand et al. 1977:24). Thus, ceramic compositional data may be used to distinguish between raw material sources. A concomitant assumption of the provenience postulate is that there is less variation within a raw material source than between sources.

A significant factor that may complicate the interpretation of compositional data is that, unlike lithic artifacts, ceramics are an additive technology that may blur the "match" between raw clay sources and finished products. Potters often sift clays, add temper and water, or blend several types of clays with different properties together, all affecting the mineralogy of clays (Blackman 1992). Firing, actual use, and post depositional processes may also alter the compositional signature of pottery (Arnold et al. 1991:71; Bishop et al. 1982; Rice 1987). Clay beds within a single geological formation are often internally heterogeneous, even within a short distance. Another possibility is the depletion of clay beds used prehistorically. The weathering of clays also causes transformations to occur.

Thus, ceramic composition is affected by both natural and cultural variables. Despite these complicating factors, the provenience postulate still holds true: ceramics made in the same production areas from the same raw material sources will have a similar composition (Arnold et al. 1991; Bishop et al. 1982).

One long-standing assumption in ceramic analysis that should be applied cautiously is the "Criterion of Abundance" (Bishop et al. 1982:300-302; Rands and Bishop 1980:19-20; Rice 1987:177), which holds that the geographic vicinity with the highest occurrence of a particular ceramic type is likely the production locale for that ceramic type. The fallacy of the Criterion of Abundance principle, however, is that it assumes that most pottery was locally produced, an assumption which Shepard's work has certainly demonstrated does not always hold true.

The Sample for Compositional Analysis

The bulk of the ceramic collection for the compositional study is drawn from surface proveniences at the early pueblos (Table 8.15). Sherds were selected randomly by ware, without reference to specific ceramic type. This was done in order to sample a range of variation within each ware, rather than exclude sherds which could not be classified easily. Approximately 20 utility and 20 white ware sherds were selected for analysis from each of the early pueblos—Dick's Ruin (PECO 434), Forked Lightning (PECO 226), Loma Lothrop (PECO 227), and Arrowhead (PECO 710). A sample of 10 sherds from Black-on-White House, an early component of Pecos Pueblo (PECO 228), was also included in the analysis. In addition, five Plains sherds identified as Ocate Micaceous, found at smaller sites within the valley, were included for comparison to the Puebloan ceramics. A few utility ware sherds recovered from Gunnerson's (1968, 1969b) excavations and from Nordby and Creutz's (1993a) pithouse excavations were also included in the analysis. Table 8.15 lists the sites sampled for compositional analysis, the number of samples submitted, and the assigned specimen numbers; Figure 8.18 displays a map of the sites sampled for the compositional study.

Table 8.15. Sites sampled for compositional analysis.

Figure 8.18. Locations of sites sampled for compositional analysis. (click on image for an enlargement in a new window)

Rio Grande Gray Ware

Traditionally, utility ware at Pecos is divided into a sequence of three main types, whose production spans are little understood. These poorly dated types are known as Plain Gray (A.D. 1100-1500), Indented Blind Corrugated (A.D. 1100-1500), and Pecos Striated (A.D. 1500-1700). The sequence of utility ware types found at Forked Lightning Pueblo is mirrored at Pecos Pueblo, and these types have been noted in surface assemblages for the smaller sites within the Pecos Valley.

Aside from their relative chronological ordering, little is known about utility ware at Pecos. It is not understood whether utility ware types were manufactured locally or if some utilitarian pottery was obtained in trade from other sources in the Southwest. Early on, Shepard (1936) hypothesized that the plain and corrugated types represented distinct traditions and suggested that, due to the abrupt technological change, the first corrugated ware found at Forked Lightning may have been imported to the site.

Shepard (1936:561) also found little correspondence between types of raw materials and formation methods in Pecos utility ware pottery. She noted that "while there are important differences in pastes, they are of chronological or geographic significance and are seldom constant in the types defined by workmanship," meaning that surface treatment is independent of paste (Shepard 1936:561, emphasis mine).

Pajarito White Ware

White ware ceramics at Pecos encompass great diversity in paste color and texture, the primary classificatory criteria. After the sherds were randomly selected, they were classified to type when possible. The types of relevance to this portion of the analysis are Santa Fe Black-on white (A.D. 1175-1350 or later), Galisteo Black on-white, and Wiyo Black-on-white (both A.D. 1300-1400). While Biscuit ware sherds are present at some of the pueblos, they were not included in the compositional study because of Shepard's (1936:490) demonstration that Biscuit ware is nonlocal to Pecos. The tuff used to temper these sherds originates from the Chama and Pajarito Plateau areas. Figure 8.19 shows the scanned images of representative black-on-white sherds submitted for compositional analysis.

Figure 8.19. Representative Black-on-white sherds submitted for compositional analysis. (click on image for an enlargement in a new window)

Analytic Methods

Clay sources within the park (noted by the survey) that were potential sources of raw materials for prehistoric potters were sampled (n=10) (Figure 8.20). The clays and sherds were refired in order to examine differences in compositional makeup. Refiring small portions of sherds and the clays at a constant temperature (850 degrees Celsius) for fifteen minutes controlled for differences in original firing temperature, burned off excess soot, and allowed recognition of broad similarities and differences in paste. The next stage of analysis involved petrographic study of sherd thin sections with a low-power stereo microscope.

Figure 8.20. Clay sources sampled at Pecos National Historical Park. (click on image for an enlargement in a new window)

Results of Refiring

Munsell (1998) color values were recorded for the oxidized paste of each ceramic chip after refiring; these values were then grouped by color categories using the Munsell color name designations (Tables 8.16 and 8.17). It was immediately apparent that most of the chips refired to a light red color, with slight variation in color ranging from reddish-brown to reddish-yellow (Tables 8.16 and 8.17). Ceramic pastes that refire to a red color are indicative of clays with a high iron content (Rice 1987:343). The light red oxidized color predominates in both the utility ware and white ware samples, although the white ware sherds more often display a reddish-yellow oxidized paste. A significant finding is that 17 percent (n=12) of the white ware sherds refired to a much lighter shade (Munsell colors white, pink, and very pale brown), designated here as "buff" colors (Table 8.17). From Shepard (1936:576), we know that Permian clays from the Pecos area are nonrefractory and typically fire to a red color, in contrast to clays from the Galisteo Basin which fire to a light color. From the oxidation data for the Pecos sherds, we may infer that at least some of the white ware ceramics are constructed from nonlocal, buff-firing clays and likely originate from a different ceramic production area. No clear patterning by ceramic type is apparent in the refired colors (each type encompasses several oxidized colors), but it may be significant that all of the "buff" firing sherds (n=12) are typed as Galisteo Black-on-white.

Looking at the refiring data by site, it appears that for the utility ware sample (Table 8.16), Dick's Ruin, Loma Lothrop, and Arrowhead each have at least one unique oxidized color value, perhaps indicating the use of a unique local clay by the potters at these sites. In contrast, the oxidized colors of utility ware sherds from Forked Lightning and Pecos Pueblo are found in the pastes from the sherds at all of the other pueblos as well. Interestingly, the utility ware sample from Gunnerson's Apache House and from the possible Plains sherds refired to the same light red color as one another, suggesting use of a similar clay source. The Pecos pithouse utility ware sample is not unique but appears to match the reddish-yellow color found in other oxidized sherds from the pueblos at Pecos, a finding which may signal continuity in resource use through time from the Developmental period to the Coalition and Classic periods. However, the overall homogeneity in refired colors for the utility ware sample does not permit the definition of exact production locations.

Table 8.16. Oxidized colors of the utility ware sherds by site.

For the white ware ceramics (Table 8.17), the Pecos Pueblo sample shows less variation (more standardization) in oxidized clay color, although it must be kept in mind that the sample is small. All of the sherds from Pecos Pueblo refire to either a reddish-yellow or to a light, "buff" color. As stated above, the light-firing white ware sherds may be nonlocal. In addition to Pecos Pueblo, light-firing sherds also occur at the sites of Forked Lightning and Loma Lothrop, but not at Dick's Ruin. Most of the utility ware and white ware samples refire to the same reddish hues (Table 8.16; Table 8.17), which may result from the use of the same clays to construct both wares. However, the different appearance in paste of the two wares before refiring indicates different original firing conditions, with a reducing atmosphere for the utility ware and an oxidizing atmosphere for the white ware.

Table 8.17. Oxidized colors of the white ware sherds by site.

The raw clays collected from the Pecos area were also fired under the same conditions as the chips from the prehistoric sherds. The clays display a wide range of variability in oxidized colors (Table 8.18), although most are some variant of red. The colors of the local clays may be a match for some of the prehistoric ceramics. Surprisingly, some of the local clays fired to a pink color, close to some of the "buff" colors, which may contradict the traditional belief that Pecos clays are only red-firing. Thus, the light-firing ceramics may not necessarily have been produced outside of the local area. Other evidence is needed to address the question of source, including petrographic data on local and nonlocal ceramic tempers, as seen later in the chapter.

Table 8.18. Oxidized colors of the clay samples from the Pecos area.

Inferences derived from the oxidized colors of the ceramics must allow for the possibility that clays from different sources may refire to the same color. The iron content of clays, in particular, greatly influences the resulting oxidized colors and may obscure the patterning of clays from different production areas. The initial observations from the refiring data offered here undoubtedly are derived from somewhat subjective visual differences in ceramic paste color but serve as useful (and inexpensive) indicators of the variation that may be expected from the petrographic analysis.

Results of Petrography

A total of 209 ceramic samples, including 10 fired clays, were submitted for thin sectioning to Ray Lund, Quality Thin Sections, Tucson, Arizona. The thin sections were sent for petrographic analysis to Tammy Burns of the University of Notre Dame, Department of Civil Engineering and Geological Sciences (Burns 1999). Using a petrographic microscope, the analyst performed mineral and petrographic identification to classify the samples by temper and paste categories within each ceramic ware. One component of the temper analysis sorted the nonplastics into size groupings. In addition to mineralogical identification, analysis of the texture of the ceramic paste played a key role in sorting the samples into paste groupings.

Temper is defined here as an additive material that was used by potters to improve the workability of the raw clay and to construct vessels with desired performance characteristics. The use of temper in pottery making "counteracts shrinkage and facilitates uniform drying, thus reducing strain and lessening the risk of cracking" (Shepard 1985:25). The use of the term "temper" is confined here to those instances when material clearly appears to be added by potters, as opposed to the naturally occurring constituents of the raw clays (i.e., accessory minerals).

To account for the variation present in the samples, the final petrographic analysis produced 9 different utility ware temper groups and 16 utility ware paste groups. Likewise, 7 white ware temper groups and 8 white ware paste groups were identified. The resulting petrographic temper and paste groups of each ware are described below and defined more completely in the following tables. The individual ceramic thin section descriptions and group assignments appear in Burns (1999). The following discussion summarizes and interprets the salient findings of the technical petrographic report submitted to the National Park Service (Burns 1999).

Utility Ware Temper and Paste Categories

Tempering materials are relatively homogeneous among the utility ware samples, with most (n=93, 75.6 percent) tempered with a quartz-feldspar sand (Table 8.19; Figure 8.21). A fluvial source for the sand is indicated by weathering, grain shape, and mineralogy. Most of the other utility ware petrographic temper groups (4-9) are iterations of Utility Ware Temper Group 1 with additional minerals besides sand (Table 8.19). The slight differences in utility ware temper groups 4-9 also likely indicate variation within a fluvial source. The other petrographic utility ware temper categories include feldspars (Utility Ware Temper Group 7) and claystones (Utility Ware Temper Group 2), both of which may be found locally.

Table 8.19. Utility ware petrographic temper groups.

Figure 8.21. Total number of utility ware sherds per petrographic temper group.

Utility Ware Temper Groups

Of the nine utility ware temper groups identified through petrographic analysis, only Utility Ware Temper Group 3 (sanidine, plagioclase, and quartz) is mineralogically distinct. Utility Ware Temper Group 3 is of particular interest because the temper type contains sanidine, a form of alkali feldspar almost completely restricted to felsic volcanic rocks. In addition to sanidine, other materials present in Utility Ware Temper Group 3 are volcanic glass shards and euhedral, nonundulose quartz grains almost certainly derived from a felsic volcanic rock or deposit (Gary Smith, personal communication 2001). Sanidine is definitely nonlocal to the Pecos area (Burns 1999:5; Shepard 1936:563-564). The closest source to Pecos of sanidine-bearing volcanic rocks is the Cerrillos-La Cienega-Santa Fe area, but the overall mineralogy of the volcanic rocks found there does not line up with the mineralogy of the ceramics. Unlike the archeological samples, the volcanic rocks from the La Cienega and the Santa Fe area do not contain volcanic quartz or glass shards and so could not be the source area for Utility Ware Temper Group 3 (Gary Smith, personal communication 2001). These geological sources also exhibit abundant biotite and pyroxene, minerals which are not present in the ceramic samples.

Utility Ware Temper Group 3 most likely originates from the rhyolites of the Jemez Mountains and Pajarito Plateau, which do have abundant quartz, sanidine, and volcanic ash shards, and are a much closer match for the ceramic mineralogy of the prehistoric ceramics (Gary Smith, personal communication 2001). The Jemez rhyolites seldom exhibit mafic minerals, which are also notably absent in the ceramic samples. The lack of rhyolitic ground-mass rock fragments in the thin sections points to a source area near "one of the Quaternary ignimbrites (Bandelier Tuff) or alluvium derived from erosion of the ignimbrites" (Gary Smith, personal communication 2001). Thus, the mineralogy of this ceramic temper seems to fit best with a source in the southeast Jemez Mountains. Significantly, Utility Ware Temper Group 3 is found exclusively at Forked Lightning Pueblo (PECO 226) (n= 6), one of the earliest occupied pueblos in the Upper Pecos Valley.

The majority of the petrographic utility ware temper groups are not specific to any one site at Pecos (Figure 8.22), although Utility Ware Temper Group 3, just discussed, and Utility Ware Temper Group 6 are found only at Forked Lightning Pueblo (Table 8.20). Five utility ware sherds (17 percent) from Pecos Pueblo have temper from Utility Ware Temper Group 1, but also contain an extrusive igneous material (tuff), nonlocal to the Pecos area. The pithouse, Apache, and possible Plains samples are all tempered with Utility Ware Temper Group 1, which is in distinguishable from the temper of the Puebloan samples in this group.

Figure 8.22. Total number of utility ware sherds per petrographic temper group. (click on image for an enlargement in a new window)

Table 8.20. Summary of utility ware temper groups by site.

Utility ware clay pastes were found to be more diverse than the utility ware tempers in mineralogy and in texture (Table 8.21; Figure 8.23). Utility Ware Paste Group C is the most common (n=36; 28 percent). Four utility ware paste groups (A, B, M, and S) are distinguished by the presence and shape of mica. In these instances, there is no indication that a micaceous wash was applied as a slip, as described by Warren (1981), for some northern Rio Grande utility ware. The accessory minerals found in Utility Ware Paste Group A, in particular, are likely derived from a nonlocal high-grade metamorphic facies (Burns 1999:6). Other utility ware paste groups are distinguished by quantities of silt and sand. It is possible that some of the differences between the utility ware paste groups are due to differences in processing the same clay, which may alter the texture and appearance. Some of the utility ware paste groups (K, L, N, and P) were created to describe individual sherds with unusual paste characteristics.

Table 8.21. Utility ware petrographic paste groups.

Figure 8.23. Total number of utility ware sherds per petrographic paste group.

Figure 8.24 and Table 8.22 summarize the utility ware paste groups by site. The sherds from the unique paste groups (J, K, L, N, and P) come from different sites (Table 8.22), perhaps indicating the use of nearby local resources at each site. Utility Ware Paste Groups J and K are exclusive to Forked Lightning. Utility Ware Paste Group L is exclusive to Loma Lothrop. Utility Ware Paste Group N is exclusive to Arrowhead. Utility Ware Paste Group P is found only at Pecos Pueblo. Finally, Utility Ware Paste Group B, with mica, is unique to Gunnerson's Apache House sherds.

Table 8.22. Summary of utility ware paste groups by site.

Figure 8.24. Summary of utility ware paste groups by site. (click on image for an enlargement in a new window)

The most common paste group, Utility Ware Paste Group C, appears at five sites (Table 8.22). The majority of the sherds from Pecos Pueblo (n=24; 83 percent) are composed of Utility Ware Paste Group C, perhaps indicating greater standardization in raw material use. In contrast, the utility ware samples from the other pueblos are spread over a greater number of paste categories, indicating greater diversity in raw material composition (Figure 8.24).

As seen in a cross-tabulation of utility ware temper and paste groups (Table 8.23), most of the utility ware temper groups are not specific to any one petrographic paste grouping, with two exceptions. The exceptions are Utility Ware Temper Group 2 (mudstones), associated with Utility Ware Paste Group F, and Utility Ware Temper Group 3 (sanidine, plagioclase, and quartz), associated with Utility Ware Paste Group G. The most common utility ware temper (Utility Ware Temper Group 1), a quartz feldspar sand, appears in association with nearly all of the utility ware paste groups (Table 8.23). The unique utility ware paste groups (J, K, L, N, and P) are also associated with unique tempers (Table 8.23).

Table 8.23. Cross-tabulation of utility ware paste and temper groups.

White Ware Temper and Paste Categories

As observed petrographically, the temper of the white ware ceramics is more geologically diverse than that of the utility ware. Seven white ware temper groups are represented (Table 8.24; Figure 8.25). The white ware sherds are tempered primarily with sand (White Ware Temper Groups C, E, and F; n=24; 34 percent), vitric welded tuff/consolidated ash (White Ware Temper Group B; n=11; 16 percent), or grog (White Ware Temper Group D; n=10; 14 percent). Nonlocal, rhyolitic volcanic ash characterizes one of the white ware temper categories (White Ware Temper Group A; n=7; 10 percent). Grog (crushed sherds), a common temper (White Ware Temper Group D), is itself often tempered with nonlocal volcanic material. In some cases sand appears as temper in addition to the grog. White Ware Temper Group G, a felsitic volcanic, is represented by only one example and is likely a nonlocal sherd. Surprisingly, 25 percent (n=18) of the extremely silty black-on-white sherds are not tempered at all (Figure 8.25). Warren's (1971) brief, unpublished notes from an examination of white ware sherds at Forked Lightning Pueblo also mention the presence of a variety of tempering materials as well as the silty quality of many of the black-on-white sherds.

Table 8.24. White ware petrographic temper groups.

Figure 8.25. Total number of white ware sherds per petrographic temper group.

Summaries of the white ware temper groups by site (Figure 8.26; Table 8.25), show that most of the white ware temper groups are found at all of the pueblos, in relatively equal numbers, including at Pecos Pueblo. The unique White Ware Temper Group G (n=1) is found only at Loma Lothrop.

Table 8.25. Summary of white ware temper groups by site.

Figure 8.26. Summary of white ware temper groups by site. (click on image for an enlargement in a new window)

The eight white ware petrographic paste groups are more restricted in number than the utility ware paste groups (Table 8.26; Figure 8.27). Important differences in texture observed petrographically were key to distinguishing between the white ware pastes, which range from nonsilty to silty/sandy (Table 8.26). White Ware Paste Groups 1, 2, and 3 are most common, accounting for 83 percent of the overall variation in the white ware sample.

Table 8.26. White ware petrographic paste groups.

Figure 8.27. Total number of white ware sherds per petrographic paste group.

In terms of patterning among the sites (Figure 8.28; Table 8.27), White Ware Paste Group 4 is exclusive to Loma Lothrop, which again, may indicate the use of a unique local clay in the vicinity of the site. White Ware Paste Group 5 (with mica) is exclusive to Forked Lightning. White Ware Paste Groups 1-3 are distributed equally among the pueblo samples. The white ware sherds from Forked Lightning Pueblo and Loma Lothrop appear to have relatively more variability in petrographic paste composition than the samples from Dick's Ruin and Pecos Pueblo (Figure 8.28).

Figure 8.28. Summary of white ware paste groups by site. (click on image for an enlargement in a new window)

Table 8.27. Summary of white ware paste groups by site.

Unlike the utility ware pastes and tempers, the white ware paste groups correlate closely to the white ware temper groups (Table 8.28). For example, vitrophyric welded tuff temper (White Ware Temper Group B) is associated with a particular white ware clay (White Ware Paste Group 3; n=11). White Ware Temper Group D (n=10), grog temper, is associated with White Ware Paste Group 2. White Ware Temper Group G (n=1), a felsitic volcanic, is associated with White Ware Paste Group 6. White Ware Paste Group 4 is correlated with White Ware Temper Group F (sand). However, the most plentiful white ware paste groups (1, 2, and 3) are seen in association with several kinds of tempering materials. In addition, sherds with no added temper are found in four different paste groups, but no added temper occurs most frequently (n=15; 83 percent) with the most common white ware paste (White Ware Paste Group 1).

Table 8.28. Cross-tabulation of white ware paste and temper groups.

Between the two ceramic wares, some of the white ware pastes correspond to utility ware pastes, although the utility ware ceramics are more often tempered (Table 8.29). White Ware Paste Group 5 has abundant mica, probably signaling a nonlocal product. Significantly, this paste group is similar in composition to Utility Ware Paste Group G, which contains two examples of volcanic ash temper and six examples of sanidine temper—substantially strengthening the interpretation that these two paste groups are nonlocal to the Pecos area. No definite conclusions can be drawn about the local or nonlocal origin of the other paste groups, based on the petrographic descriptions or their associations with specific temper groups. However, these overall paste correspondences between the two wares do suggest that Pecos potters used similar clays to make some of the utility and white ware pottery.

Table 8.29. Corresponding petrographic paste groups between the utility ware and white ware ceramics.

The Raw Clays

Petrography reveals the raw clay samples (formed into briquettes, fired, and thin-sectioned) to be very distinct from one another, yet different from both the utility and white ware ceramic pastes (Burns 1999). The raw clay samples do not resemble the sherds in accessory minerals, which are nearly absent in the clays. Unlike the archeological samples, the raw clays exhibit iron oxides and calcareous marls—probably a derivative of caliche. Only Raw Clay Sample 5, while not a close match, has a silty paste suggestive of some of the clays in the prehistoric sherds. The mineralogical diversity of the raw ceramic clays is a reflection of the great variation within alluvial clay sources from the Upper Pecos Valley, although the clays were all collected within 2 km (1.2 mi) of one another. The raw clays were not tempered before firing, which may account for some of the observed differences between the raw clays and the prehistoric sherds. Another contributing factor may be that the raw clay sampling was not extensive enough for a closer match to be found to the prehistoric sherds. The mixing of clays by potters may also obscure the relationship between raw clays and finished pottery.

Addressing the Research Questions

Returning to the goals of the ceramic compositional study, the refiring and petrographic results can be used to address the original research questions presented earlier in the chapter (see pp. 275-276). These issues are concerned with distinguishing ceramic production areas, whether local or nonlocal (Questions 1 and 2), and quantifying the degree of compositional variation among the pueblos (Question 3). Describing the variation among ceramic types is also a primary goal (Questions 4-6).

Production Areas

A variety of compositional sources and production areas for the Pecos pottery is indicated by the great degree of variation in refired sherd colors and petrographic temper and paste groups within each ware category. This compositional variation probably reflects two things: the geological diversity of the northern Rio Grande region and the prevalence of trade in ceramic vessels.

Four of the utility and white ware petrographic temper groups contain nonlocal volcanics (including ash, welded tuff, and sanidine). These materials do not appear weathered, and likely originate from extensive rhyolitic welded tuff and ash deposits in the Jemez Mountains. Of all ceramic raw materials, temper is least likely to have been traded (Shepard 1985:337). While it is not outside the realm of possibility that temper was brought or traded to Pecos, cross-cultural ceramic studies indicate that tempering materials are seldom transported over 9 km (5.6 mi) (Arnold 1985:45-46). Thus, the volcanic tempered pottery is likely a result of trade with groups from the Pajarito Plateau, roughly 85 km (52.8 mi) west of the Pecos Valley, or from exchange with closer neighbors in the Galisteo Basin approximately 24 km (14.9 mi) west of the Pecos Valley, who in turn may have traded with Pajarito Plateau groups.

While constituting less direct evidence of a nonlocal source, the petrography shows that some of the grog in the white ware grog-tempered pottery found at the Pecos sites is itself tempered with volcanic rock. The presence of volcanic sherd-temper may suggest that the pottery was also traded from the Pajarito Plateau (Shepard 1985:383-384). Burns (1999:11) correctly points out that local Pecos potters could have preferentially selected trade ware (containing volcanics) for use as grog temper in their own products. Nonetheless, there is no evidence (i.e., the use of recognizable local clays with the addition of volcanic grog temper) to rule out the likelihood that these volcanic-grog-tempered ceramics are trade ware.

The three mica-rich paste groups (Utility Ware Paste Groups A, B, and M) are also likely from trade ware vessels, due to the absence of micaceous clay resources in the Pecos vicinity. The clays of Utility Ware Paste Group A could originate from the Picuris Mountains to the north, based on the presence of mica and medium to high-grade metamorphic accessory minerals (Burns 1998, 1999:13).

Variation among Sites

Most of the petrographically identified paste and temper groups appear in sherds from all of the pueblos. In particular, the sherds from Forked Lightning Pueblo (PECO 226), the earliest site, are petrographically similar to those found at Dick's Ruin (PECO 434), Loma Lothrop (PECO 227), and Arrowhead (PECO 710) in temper and paste composition. Interestingly, only samples from the earliest sites, the Pecos pithouse (PECO 207) and Forked Lightning Pueblo, have mica-rich sherds. For the utility ware sample, only Forked Lightning Pueblo has sanidine-tempered sherds. A few nonlocal utility sherds from Pecos Pueblo (n=5; 17 percent) have an extrusive igneous temper (tuft), not found at the other sites.

A comparison of the nonlocal and local utility and white ware ceramics—as defined by the presence or absence of nonlocal pastes or tempers—from each of the pueblos may be found in Figures 8.29 and 8.30. The plots show that for the utility ware sample, Forked Lightning Pueblo and Pecos Pueblo, the earliest and latest sites, are the only sites with ceramics containing nonlocal tempers (Figure 8.29). Utility ware sherds with nonlocal micaceous paste are found in small quantities at all of the pueblos except Pecos Pueblo. Nonlocal white ware is found at all of the pueblos sampled for analysis, with the prevalence of nonlocal white ware paste or tempers increasing through time (Figure 8.30), although the sample is admittedly small. If sherds tempered with volcanic-tempered sherds are considered trade ware (which I argue they should be), the total percentage of nonlocal white ware sampled from each site approaches approximately 50 percent of the total. In addition, the un-sampled Biscuit ware ceramics at each site must also be considered, because they are certainly nonlocal indicators of trade and would further increase this percentage.

Figure 8.29. Percentages of local and nonlocal utility ware ceramics at the pueblos, based on petrographic results. (click on image for an enlargement in a new window)

Figure 8.30. Percentages of local and nonlocal white ware ceramics at the pueblos, based on petrographic results. (click on image for an enlargement in a new window)

Based on these petrographic data, there is no archeological evidence to indicate that any Pecos site enjoyed preferential access to ceramic resources, whether locally produced or the result of regional exchange. Although Forked Lightning is the only pueblo site with traces of nonlocal White Mountain Red Ware, a common trade ware ceramic in the American Southwest, this could be the result of the earlier timing of the site's occupation span compared to the other pueblos. There is no indication that any one site controlled the distribution of utility or black-on-white trade ware ceramic goods coming into the Upper Pecos Valley. This pattern of almost all paste and temper groups present at all of the early pueblos in relatively equal numbers is consistent with a cooperative strategy in which trade in utility and white ware pottery was common throughout the valley.

Compared to the ceramics from the earlier pueblos, sherds sampled from Pecos Pueblo (PECO 228) exhibit less petrographic variability in temper and paste overall, especially for the utility ware sample. Only four utility ware paste groups are present at Pecos Pueblo, compared to 7-9 utility ware paste groups found at the other pueblos. Similarly, only three utility temper groups are found at Pecos Pueblo, compared to 5-6 utility ware temper groups from samples at each of the other pueblos. The same decrease in the number of compositional groupings is true of the white ware from the Pecos Pueblo sample, although the patterning is not quite as strong and may be a product of the smaller sample size for Pecos Pueblo. Pecos Pueblo has 3 white ware paste groups and 6 white ware temper groups, while the other pueblos have between 4-8 white ware paste groups and 6-8 white ware temper groups.

Such a decrease in compositional hetero geneity from the time of the early pueblos to the time of Pecos Pueblo would be expected with increasing standardization in ceramic production. This finding of decreased variation is particularly apparent in the utility ware ceramics, which, based upon the absence of nonlocal materials, appear to be predominantly locally produced. At the early pueblos, the great variation in the number of utility ware paste groups is in keeping with very low levels of standardization in ceramic composition and low productive specialization. This is the pattern we might expect if each household produced its own utility ware pottery. However, the several examples of nonlocal utility ceramics at each pueblo indicate that at least some of the vessels were not produced there and were traded or transported to the Pecos area.

With increased aggregation at Pecos Pueblo, a decreased number of practicing potters may be responsible for the pattern of less compositional variation seen in the ceramic petrographic data from that site. This may indicate that some Pecos potters produced beyond the needs of their own household, possibly in a situation of emergent craft specialization. Greater standardization in raw materials at Pecos Pueblo is apparent for the utility ware, and to a lesser extent for the white ware. However, the amount of trade ware within the white ware sample obscures the patterning.

Nonetheless, the degree of variation in white ware and utility ware temper and paste groupings at Pecos Pueblo does signal a change in compositional standardization in the ceramics from the earlier pueblos. However, the overall level of standardization in utility and white ware ceramics certainly does not match the level of standardization reflected in the later glaze ware production process documented elsewhere (Bower et al. 1986; Kidder and Shepard 1936; Motsinger 1997; Warren 1969, 1970, 1979b). Another possible explanation for the decrease in compositional variation in the utility and white ware ceramics from Pecos Pueblo is that increasing population aggregation led to greater social cohesion and a more limited "teaching framework" among potters (Burns 1999:14; Schiffer and Skibo 1987). These observations regarding possible emergent ceramic standardization are based on a relatively small sample of sherds. Other lines of evidence such as chemical characterization and measurement data from whole vessels are needed to truly understand the organization of pottery production at Pecos.

Variation between Santa Fe Black-on-white and Galisteo Black-on-white

The two predominant white ware types included in this study, Santa Fe Black-on-white and Galisteo-Black-on-white, may be generally distinguished based on temper, in conjunction with surface treatment. Many Santa Fe Black-on-white sherds are composed of local White Ware Paste Group 1 (Table 8.26) with no temper added—probably indicating a locally-made product. Other common tempers in Santa Fe Black-on-white are volcanic ash (nonlocal) and quartz/feldspar sand (Figure 8.31). Welded tuff (of nonlocal origin) and grog (crushed sherd) are less often present as temper in the Santa Fe Black-on-white sherds.

Figure 8.31. Santa Fe Black-on-white petrographic temper categories.

In contrast, Galisteo Black-on-white sherds are primarily sherd-tempered, a key classificatory criteria for defining sherds as Galisteo Black-on-white in macroscopic identification (Figure 8.32). Petrographic analysis shows that some Galisteo Black-on-white sherds also contain sand and vitrophyric welded tuff tempers. The crushed grog in the Galisteo Black-on-white sherds is itself often tempered with volcanic material.

Figure 8.32. Galisteo Black-on-white petrographic temper categories.

The petrographic results demonstrate that Santa Fe Black-on-white and Galisteo Black-on-white are compositionally heterogeneous types. The two traditional types also overlap in some temper categories. Some sherds of each type are tempered with tuff and sand (Figures 8.31 and 8.32). While Galisteo Black-on-white sherds are more often sherd tempered (60 percent), a small quantity of Santa Fe Black-on-white sherds also contain crushed sherd temper (7 percent). Significantly, fine ash temper appears only in the Santa Fe Black-on-white sherds (18 percent). Finally, the majority of the Santa Fe Black-on-white sherds sampled here (41 percent) are not tempered at all, other than with naturally occur ring components of the ceramic clay. Although Galisteo Black-on-white is often particularly considered a trade ware ceramic in the northern Rio Grande (Collins 1975; Lang 1982; Shepard 1965), the petrographic data show that at least 30 percent of the sherds in the Santa Fe Black-on white sample from the Pecos sites are also of nonlocal origin.

Comparisons to Apache and Pithouse Ceramics

Five possible Plains sherds collected by the survey were submitted for petrographic analysis to evaluate whether they could be drawn from the same population as Gunnerson's (1969b) known Apache ceramics (from collections at Pecos National Historical Park), or if they are more similar to the Puebloan utility ware sherds from the early pueblos. Sherds from Gunnerson's Apache House positively identified as Ocate Micaceous contain two of the micaceous utility ware paste groups (A and B) and are tempered with sand. Utility Ware Paste Group A is a paste with platy mica, while Utility Ware Paste Group B contains an acicular mica with more biotite. Utility ware Paste Group B is specific to Gunnerson's Plains sherds.

Two of the potential Plains survey samples are classified as Utility Ware Paste Group A, indicating a close match to the known Ocate Micaceous sherds. However, sherds from Utility Ware Paste Group A are also found at Dick's Ruin, Loma Lothrop, and Arrowhead. Based on thickness, macroscopic appearance, and surface treatment, the sherds from the pueblos are indented blind corrugated and not southern Plains ceramics. That both ceramic types are constructed from similar clays may indicate an overlap in prehistoric resource procurement zones for Puebloan and southern Plains potters.

Three other tentatively identified Plains samples are classified by the petrographic analysis as Utility Ware Paste Group S, having more acicular white mica and no accessory minerals. Utility Ware Paste Group S is not found in any other of the ceramic types or at any of the other Pecos sites of Puebloan affiliation indicating a distinctly different compositional source for these sherds. This unique paste group (Utility Ware Paste Group S) would appear to support the interpretation of these items as southern Plains/Apache ceramics.

While the pithouse ceramics (n=4) also have mica inclusions, they come from Utility Ware Paste Group M, which is distinguished by distinct accessory minerals and a lesser abundance of mica. Utility Ware Paste Group M is more accurately termed "mica-rich," rather than micaceous, following Habicht-Mauche's (1988:256) caution about labeling ceramic pastes micaceous when only very small amounts of mica are present. Given the lack of known micaceous clays from the Pecos area, Utility Ware Paste Group M is likely nonlocal. Utility Ware Paste Group M is not specific to the pithouses, however. Two sherds from Forked Lightning Pueblo also fall into this category, indicating use of a similar clay resource even though the sites are separated by at least 300 years.

Conclusions

This chapter has described the Pecos survey ceramic data and evaluated how increasing aggregation, the intensification of agriculture, and changes in regional interaction in the Pecos area affected ceramic technology through time. Given known changes in settlement aggregation and population size at Pecos, accompanying changes in ceramics were also expected. Social changes often associated with aggregation, such as an increase in household size or the intensification of agriculture, may be reflected in larger vessel sizes through time or in the need for greater storage vessels. These expectations were not always matched by the Pecos ceramic data, how ever.

The functional analysis of ceramic assemblages provides slight evidence for greater reliance on agricultural products and/or an increase in the household commensal unit, as seen in increased cooking jar and serving bowl size through time. However, these modest changes in vessel size through time are not statistically significant. The exceptions are Biscuit Ware and Glaze V ceramics, which are found to be larger than other ceramic types based on rim sherd diameter. Statistically significant evidence for increased quantities of storage vessels through time likely reflects the requirements of agricultural intensification.

The site types defined by the survey (based on architectural differences) do not differ appreciably in proportions of ceramic functional classes or in vessel size. Contrary to expectations, seasonal and special-use sites exhibit a variety of wares and functional classes, indicating that the functions and duration of use at these types of sites were not as limited as expected. The analyses indicate that overall ceramic assemblage composition at Pecos did not change substantially through time. The enduring pattern of assemblage composition through time and at all site types is suggestive of a conservative technology in which ceramics served generalized functions.

Evidence for the organization of pottery production at Pecos was investigated using refiring and petrographic data for utility and white ware ceramics, focusing on the Coalition-period pueblos. The resulting diversity of compositional groupings denotes a lack of standardization in raw material use and points to the existence of many ceramic production areas, in keeping with a pattern of unspecialized household production of ceramics. The traditionally defined ceramic types were also found to contain a wide range of variability.

A variety of compositional sources for both the utility ware and white ware are indicated, reflecting both the geological diversity of the region as well as trade in ceramic vessels. Most of the ceramics were produced locally using readily available silty clays and stream sand, although no one-to-one match could be demonstrated between the prehistoric ceramics and the local clays. However, approximately half of the white ware vessels sampled from the pueblos appear to have been trade ware. A few utility ware vessels from Forked Lightning Pueblo and from Pecos Pueblo, the earliest and latest sites in the sample, are also nonlocal. For some of the ceramic vessels, the presence of nonlocal tempering materials such as tuff, sanidine, and other volcanics indicates a source of origin near the Jemez Mountains. Compared to the earlier pueblos, the Pecos Pueblo sherds exhibit more standardization in ceramic composition, signaling greater specialization in the organization of ceramic production through time.

No one pueblo site at Pecos appears to have held preferential access to local or nonlocal ceramic resources or to have controlled the distribution of ceramic goods, implying that a cooperative economic strategy was in place prior to the main episode of aggregation at Pecos Pueblo. Significantly, the role of the Pecos community as a center of commerce and trade did not arise with the development of Pecos Pueblo but appears to have been foreshadowed in the long-distance ceramic trade relationships maintained by the earliest aggregated pueblos in the valley.