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

This chapter examines changes in estimated population size and aggregation in the Pecos area using the habitations (Chapter 5) and time periods (Chapter 4) identified earlier. Describing fluctuations in population size in the periods before the formation of Pecos Pueblo may provide some insight into the process by which ancestral Puebloans aggregated into one large pueblo after living a more dispersed, independent life in several different pueblos. This chapter also reviews Kidder's (1958) perspective on the development of Pecos Pueblo and Welker's (1997) reanalysis of Kidder's excavation notes and ceramics so that we have a better idea of how the pueblo evolved.

Many areas of the prehistoric Southwest experienced aggregation. A number of causes have been proposed for this phenomenon that Cordell (1996:230) describes as "socially difficult and economically inefficient." Kidder's (1924) theory is that Plains raiders created a need for defense at Pecos, hence aggregation in the Quadrangle protected by an encircling wall. Archeologists have used defense to explain aggregation and settlement changes across the Southwest. Recent research on warfare (Haas and Creamer 1996; Wilcox 1989; Wilcox and Haas 1994) indicates that concerns for defense help explain the occurrence of settlement patterns that are economically inefficient. Other suggested causes of aggregation revolve around minimizing conflict over resources (Hunter-Anderson 1979; Orcutt et al. 1990; Powers and Orcutt 1999) and encouraging cooperation or sharing to offset short-term food shortages caused by climatic perturbations (Kohler and Van West 1992). It is likely that all these causes played a role in aggregate formation in the Pecos area.

In addition to describing changes in the Pecos area, this analysis examines population fluctuations in other portions of the northern Rio Grande. Events at Pecos did not develop in a vacuum, and although it is not appropriate to reanalyze survey and excavation data from other regions, several previous studies that relate to this issue are summarized. The Upper Pecos Valley needs to be placed within an appropriate regional context so that we may better understand its place in northern Rio Grande prehistory and early history.

Reconstructing Population

The sites used in this analysis include the 10 habitation sites recorded by the Pecos survey and two pueblos on private land adjoining Pecos National Historical Park. Table 7.1 lists the 12 sites, the time periods to which they are assigned, and the estimated room counts. Rowe Ruin (Cordell 1998) and Hobson-Dressler Ruin (National Park Service 1992a), the two sites on private land, are placed in the Pecos time periods and room counts are derived from published data. Brief descriptions of Rowe and Hobson-Dressler are presented below. The habitation sites recorded by the survey are described in Chapter 5.

Rowe Ruin, approximately 1.5 km (0.9 mi) south of Pecos National Historical Park, has been the focus of professional archeological excavations at least twice in the past. Most recently, Cordell and her colleagues conducted test excavations at the site (Cordell 1998). The pueblo consists of three large quadrangular roomblocks, each arranged around its own plaza. Wall construction is limestone masonry, although earlier adobe rooms lie under one portion of the site (Cordell 1998:27). Most of the pueblo probably is two stories. Cordell does not estimate the number of rooms at the site but does provide the estimated "living area" in each roomblock (Cordell 1998:Table 1), although she does not describe how the area was calculated. The average size of the rooms excavated is ca. 7 m² in the middle roomblock and ca. 9 m² in the other two roomblocks. Using these average room sizes, Cordell's living space converts to 550 ground-floor rooms. Other estimates (National Park Service 1992a), however, are in the vicinity of 200 rooms. I chose to use the more conservative estimate of 200 ground-floor rooms.

Cordell (1998:46) believes Rowe was occupied for 100-150 years. She reconstructs the occupation sequence as follows: (1) the adobe structure is established in the 1240s or 1250s; (2) the masonry rooms are built approximately 100 years later; (3) building and remodeling continues until the 1360s and 1370s; and (4) an archeomagnetic date of 1395 from a hearth in the north quadrangle may shortly precede the abandonment of the pueblo. It appears that Rowe falls into Periods 2 (A.D. 1200-1325) and 3 (A.D. 1325-1450).

Hobson-Dressler Ruin is located about 2 km (1.2 mi) east of Arrowhead Ruin and 4.5 km (2.8 mi) northwest of Pecos Pueblo, and there is a visual link among the three (National Park Service 1992a:4). The ruin resembles a "D" with 3-4 rows of rooms radiating out from the center. Its configuration is somewhat reminiscent of Tyuonyi in Bandelier National Monument, only much smaller. There are approximately 70 masonry rooms, all apparently one story, and there is no surface evidence of plazas or kivas. The site may date to late Period 2 (A.D. 1200-1325) through the middle of Period 3 (A.D. 1325-1450) (National Park Service 1992a:8).

Using the 12 pueblos identified in the Pecos area, I constructed a population curve estimating the number of people in each time period. The curve also allows us to compare the Pecos area with other contemporaneous regions in the northern Rio Grande. Welker (1997) uses Kidder's publications and notes to break down the occupation of Pecos Pueblo into four periods and estimates population for those periods. Her estimates are reviewed here because they are based on excavation data that should be more accurate than the surface survey data from Pecos Pueblo.

Population Estimation Methods

Counting rooms and measuring floor area are the two most common techniques for estimating population at pueblos. Because of the disturbed nature of some Pecos area sites and visibility problems on others, it is difficult to measure roomblock area. The Pecos survey estimated the number of rooms at each habitation based on visible architecture, which makes calculating population fairly straight forward. Because the survey also estimated the number of stories in each pueblo, I have added upper story rooms to the estimated room counts. Except for Pecos Pueblo, none of the survey sites appears to have more than two stories. For those sites with a second story, I have increased the ground-floor room count by 0.5.

The number of rooms in use at the same time affects population estimates. Many large Southwestern pueblos are occupied over a long period of time and attain part of their size through accretion, so not all rooms are contemporaneous. This is the case at several Pecos area sites, including Forked Lightning (Kidder 1958), Rowe (Cordell 1998), and Arrowhead (Holden 1955). Early portions of Pecos Pueblo grew through haphazard abandonments and additions, leading Kidder to describe the early pueblo as "formless" (1958:59). The Quadrangle at Pecos, however, is clearly planned (Kidder 1958:63), and most rooms are occupied from 1425 until sometime in the seventeenth century (Welker 1997:Figures 28-29).

Rowe Ruin also appears to be planned, largely because the roomblocks are oriented around plazas, but in reality the outward appearance "hides a great deal of architectural building, rebuilding, and remodeling, which suggests a more complicated and longer-term development of the final community plan" (Cordell 1998:46). The fact that plaza space is "set aside," however, makes it more difficult to add rooms to the roomblocks built around a plaza, which conveys a concept of community planning (Cordell 1998:88). To escape the constraints of orienting a building around a plaza, the residents add more plaza-roomblock units.

Arrowhead Ruin also may be a planned community, although not all of it is built and occupied at the same time. Holden (1955) suggests that people from Pecos Pueblo live at Arrowhead for 75-100 years, then abandon it and return to Pecos Pueblo.

It is ideal if we know how many rooms at a site are contemporaneous, and their dates of use, but that requires excavation and the ability to date each room's construction, remodeling (if any), and final abandonment. If we can estimate the number of rooms occupied in any one period, population estimates are more accurate than assuming that all rooms are occupied. The approach to making these estimates differs from area to area depending on the kind of data available. In Bandelier National Monument, for example, data from recent excavations allowed us to estimate site use lives and occupancy rates (Kohler and Powers 1993) that could be applied to the surveyed sites (Orcutt 1999b). Without the same kind of excavation data for Pecos sites, I use a different approach to estimate the number of rooms occupied at a site in each Pecos period.

There are several steps involved in calculating how many people lived in the Pecos area. When a site is occupied in more than one time period, it is necessary to estimate how many rooms are occupied in each period. In developing the Pecos chronology, Powell (Chapter 4) generated probabilities for each site that portray the strength of a site's statistical association with each time period based on the ceramic assemblage. Because these probabilities reflect intensity of use, which is related to the number of people at a site, they are a good proxy for occupation rate. I estimate the number of occupied rooms by multiplying the total number of rooms at a site by the ceramic probability value for the site in each period of occupation. Because the information used to produce the probabilities relates directly to each individual site, the procedure is more satisfying than imposing assumptions on site use generated from archeological or ethnographic data from other regions.

The probabilities for Arrowhead (PECO 710), Forked Lightning (PECO 226), and Dick's Ruin (PECO 434) are modified based on Powell's evaluation of published or other data because the ceramic assemblages recorded by the survey are skewed by previous excavation and collection (Melissa Powell, personal communication 1998). Rowe Ruin and Hobson-Dressler are apportioned to periods based on information in Cordell (1998) and National Park Service (1992a), respectively. Table 7.2 lists the sites, probabilities, total number of rooms (taking multiple stories into account), and estimated number of occupied rooms (based on the probabilities) in each period.

Estimating Population in the Pecos Area

Figure 7.1 illustrates the number of occupied rooms through time in Periods 2-6 (A.D. 1200-1700+) based on the survey data. Beginning at around 400 rooms in Period 2, the number jumps to approximately 1,000 rooms in Period 3 (A.D. 1325-1450). In Period 4 (A.D. 1450-1575), the number of occupied rooms declines to around 450 and continues to drop slightly through Period 6 (A.D. 1700±). In Periods 4-6 (A.D. 1450-1700+), Pecos Pueblo is the only major habitation.

Figure 7.1. Estimated number of people and occupied rooms for Pecos area habitation sites.

Next, I convert occupied rooms to number of people. In the absence of excavation data to suggest how many rooms the average family used, I follow Kintigh (1985) and Orcutt (1999b) in counting two people per occupied room in all periods regardless of room function. I include Pecos Pueblo in my calculations using survey data; however, a more appropriate way to model the relationship between number of rooms and number of people at the pueblo is discussed below. Population peaks in Period 3 (A.D. 1325-1450), then declines after that (Figure 7.1). The growth rate from Period 2 (A.D. 1200-1325) to Period 3 is approximately 0.8 percent, higher than Hassan's "probable maximum growth rate for prehistoric groups" of 0.52 percent (1981:140). This suggests that people move into the Pecos area, possibly from the Plains or elsewhere in the northern Rio Grande.

Pecos Pueblo

The survey data are inadequate for an in-depth analysis of Pecos Pueblo because of the amount of excavation, surface collection, and deterioration over the years. Kidder's excavations and Welker's recent reworking of Kidder's data are far more informative (Kidder 1958; Welker 1997).

Welker (1997:108-115) uses Kidder's unpublished maps and notes to revise the occupation sequence at Pecos Pueblo. She uses stratigraphy and ceramics to define the following four time periods (1997:108-111): A.D. 1200-315, A.D. 1315-1425, A.D. 1425-1515, and A.D. 1515-1700. Her periods reflect the data for one site, thus they are different enough from those developed in Chapter 4 for the Pecos survey sites that they cannot be integrated easily into the survey scheme.

To calculate population for the first two periods (A.D. 1200-1315 and A.D. 1315-1425), Welker divides the number of rooms by two (rooms per family), and multiplies that by five (family size) (Welker 1997:114). In the two later periods, A.D. 1425-1515 and A.D. 1515-1700, the pueblo is organized into "apartments" of six rooms each, with no connecting doors from one apartment to another, even when the apartments are back-to-back in a 12-room unit (Kidder 1958:63). To calculate population in these later periods, Welker divides the number of rooms by six (rooms per family) and multiplies that by five (family size). Kidder (1958:134) and Welker (1997:86-87) both estimate family size at five people based on ranges recorded for modern pueblos. The number of rooms in each of Welker's periods increases through time while the number of people fluctuates because of the change to "apartments" in the later periods. Number of rooms and population peak in her last period, A.D. 1515-1700 (Figure 7.2).

Figure 7.2. Welker's (1997) estimates of the number of occupied rooms and population in her four time periods at Pecos Pueblo.

Neither Welker's (1997) nor Kidder's (1958) population estimates approach the 2,000 reported for 1541 when Coronado's expedition visited the pueblo (Kessell 1987:12). Welker's peak population is 800 in the A.D. 1515-1700 period, and Kidder's (1958:133) is 970 people. Kidder questions his own methodology, especially his assumption of five people per family, but it would have taken 10 people per apartment to fit 2,000 into the rooms he thinks are occupied (Kidder 1958:135). If family size is seven people, which is at the upper end of the range of 1.6 to 7.6 people per household in historic and modern pueblos (Welker 1997:86-87), population still falls well short of 2,000 (Figure 7.3).

Figure 7.3. Estimated population at Pecos Pueblo using Welker's (1997) time periods and room counts and a family size of seven people.

There are several reasons to accept the archeological estimates of population rather than the Spanish ones. The Spanish tended to exaggerate or were not very good at judging village sizes; in addition, it is possible that many of the estimates were made long after the fact (Winifred Creamer, personal communication 2000). If there were 2,000 people at Pecos Pueblo upon Coronado's arrival, it may be that members of other pueblos had gathered there to help defend Pecos Pueblo. Apaches wintering outside the pueblo walls also may have been part of the group that the Spanish encountered.

The massiveness of the building at Pecos Pueblo may have deceived the Spanish into thinking there must be many more people there. Externally, Pecos Pueblo's layout is quite different from anything else in the northern Rio Grande. Kidder views the Quadrangle as "the product of a new architectural concept which involved designing the building in advance and constructing it as a unit" (1958:63). In addition to the apartments, the pueblo has a system of corridors or galleries running all around it that are accessed through hatchways and ladders (Kidder 1958:65).

This arrangement of rooms and corridors provides better defense than earlier pueblo layouts, and Pecos Pueblo also is unique in "the rigid adherence to a transverse linear arrangement of rooms" (Kidder 1958:63). Based on his review of northern Rio Grande pueblos of comparable size and age to Pecos, Kidder concludes that many displayed transverse arrangements of rooms joined together around plazas, but often the pueblos are large, rambling, and not designed for defense like Pecos (Kidder 1958:125-127). In addition, he feels that "no site of the age and type of those... [such as San Cristobal, Paa-ko, Puye, Tyuonyi, Tschirege, Otowi, Chipiinuinge, Po-shu-ouinge, Te'ewi, Howiri, Homayo, and Posege] can be supposed to have held so long and so strictly, as did Pecos, to sections uniform in room arrangement, or to ones which invariably possessed either three or six ground-floor rooms" (Kidder 1958:127).

Welker also remarks on Pecos Pueblo's uniqueness and feels it "may be less typical than we think" (1997:118). Like Kidder, she doubts "we could say that the apartment layout of Pecos was standard for the northern Rio Grande."

Internally, the probable room functions of rooms within a six-room apartment, as interpreted by Kidder (1958:122-123) suggest that Kidder and Welker are justified in using a family size of five per apartment and that archeological estimates of population are reliable. Each six-room suite has three lower rooms, A, B, and C in Figure 7.4, that Kidder calls "cellar" rooms (1958:122). These rooms are "low-studded and usually much reduced in size by the piled 'buttressing' of all four walls, the two inner ones offering no immediate access to either light or air" (1958:122-123). The "A" rooms have access through the roof to Gallery II and initially may have been storage rooms, "but those we excavated practically always proved to have become dumps and to be choked, often almost to their low roofs, with household refuse: corn husks and cobs, squash rinds, other unidentifiable rotten organic materials, and always quantities of bones" (1958:123). Some "A" rooms are permanently sealed, possibly because of the odor (1958:123). According to Kidder, "the real purpose of the blank-walled exterior, A, rooms was of course defensive. Their roofs, all at the same level, made it possible... to go freely about the whole village. And it could be done under cover, for the second story street-like way was itself roofed" (1958:124).

The "B" and "C" rooms, on the same level as "A" rooms, "were much like A and seldom could be entered save from above. Many of them also held cellar refuse, but perhaps because living rooms were above, it consisted largely of corn husks and cobs, with few animal bones" (Kidder 1958:123). The rooms on the second and third story, which Kidder (1958:123) labels "D," "E," and "F," probably are living ("D" and "F") and living/maize storage ("E") (Figure 7.4). The "D" and "F" rooms have the best light and air circulation of any in the apartment. It appears that although one family may have six rooms, half of them at most are for living, and it is possible that the configuration of rooms is the result of the defensive architecture rather than a need to house a very large family.

Figure 7.4. Kidder's (1958:Figure 33a) reconstruction of Pecos Pueblo apartments. ©Robert S. Peabody Museum of Archaeology, Phillips Academy, Andover, Massachusetts. All rights reserved.

Population Aggregation

Throughout the prehistoric Southwest, people eventually cluster in villages, and the Pecos area is no exception. The sequence of aggregate formation in the Pecos area coincides with population increase, at least for a while. Aggregation, expressed as the number of people at individual sites, increases from Period 2 (A.D. 1200-1325) to Period 3 (A.D. 1325-1450). The sites in Period 2 average 98 people, with a median of 57 people. There is quite a range in the number of people at individual sites, from seven to 360. The mean number of people increases to 171 in Period 3, and the median jumps to 97. There is an even bigger range in the number of people—from six to 563—at individual sites in Period 3. Despite the wide range, there is a definite trend toward larger aggregates. In Period 2, there are 100 or more people at two of eight sites (25 percent), while in Period 3, there are 100 or more people at five of 12 sites (42 percent), and there are more than 200 people at four of those sites.

Pecos Pueblo is the only large residence in Periods 4-6 (A.D. 1450-1700+), so means and medians are not valid. According to the survey data, almost 900 people occupied Pecos Pueblo during Period 4 (A.D. 1450-1575). The number of occupants drops to 750 in Period 5 (A.D. 1575-1700), and declines to 455 people in Period 6 (A.D. 1700+). Welker's (1997:114-115) in-depth analysis of Pecos Pueblo using Kidder's excavation data indicates that the aggregate peaked at 800 people in the A.D. 1515-1700 period (end of survey Period 4 and all of Period 5). If Welker's data are used with a family size of seven people, which may be too large as discussed earlier, the population is more than 1,100 people (Figure 7.3). Census data for 1750 and 1790 indicate 449 and 152 people, respectively (Levine 1999:43-50, Table 2), reflecting continuous decline from disease, raids, and emigration.

Northern Rio Grande Population

What are the population dynamics in the general northern Rio Grande area during the Puebloan occupation of the Pecos area? Although survey coverage in the region is spotty, researchers have reconstructed population for the Taos, Chama, Pajarito Plateau, Jemez, Gallina, Santa Fe, and Galisteo Basin districts (Crown et al. 1996; Orcutt 1999b; Orcutt et al. 1994). Of these, only the Pajarito Plateau (Hill and Trierweiler 1986; Powers and Orcutt 1999) has been systematically surveyed.

Orcutt et al. (1994) reconstructed Classic period (A.D. 1325-1600) through early Historic period (A.D. 1600-1700) population for large pueblos in several northern Rio Grande subareas. We used the northern Rio Grande time periods, and because each of the four periods is a different length, room counts were standardized to the length of the shortest period. We used this approach so that room counts were not highest in the Middle Classic phase (A.D. 1400-1550), for example, just because more rooms may have been built and abandoned during its longer time span. Room counts were not converted to people because we simply wanted to identify gross trends in population dynamics across the larger region.

Population in each area was graphed at the beginning and end of the Early Classic phase (A.D. 1325-1400), beginning and end of the Middle Classic phase (A.D. 1400-1550), the Late Classic phase (A.D. 1550-1600), and the Historic period (A.D. 1600-1700) (Figure 7.5). The Chama and Pajarito Plateau populations peak at the end of the Early Classic, then decline through time. The Santa Fe area's population is highest at the beginning of the Early Classic, and it generally decreases from there. Population declines in the Jemez and Galisteo areas from the end of the Early Classic through the beginning of the Middle Classic. Their populations increase at the end of the Middle Classic, peak in the Late Classic, then decline in the Historic period.

Figure 7.5. Number of rooms adjusted to time period lenght for several areas in the northern Rio Grande, Early Classic through Early Hispanic period (Orcutt, et al. 1994).

The Rio Grande periods and Pecos survey periods are slightly different, but there is enough similarity to compare population trajectories. The Early Classic (A.D. 1325-1400) is roughly contemporaneous with Period 3 (A. D. 1325-1450), the Middle Classic (A.D. 1400-1550) with Period 4 (A.D. 1450-1575), and the Late Classic (A.D. 1550-1600) plus early Historic (A.D. 1600-1700) with Period 5 (A.D. 1575-1700). In the Pecos survey area, population peaks in the Early Classic, as it does in the Pajarito, Santa Fe, and Chama areas. Peaks occur in the Jemez Mountains and Galisteo Basin during the Late Classic when aggregation at Pecos Pueblo is at its height.

Arroyo Hondo Pueblo, San Marcos, and San Lazaro are three large sites used in the regional population reconstructions that have been excavated or systematically surface collected in recent years, making it possible to compare their population curves to the estimates for Pecos Pueblo. Of these, only Arroyo Hondo, in the Santa Fe area, has been the scene of major professional excavations (Schwartz 1971, 1972; Schwartz and Lang 1973).

There are two temporal components at Arroyo Hondo. The earliest has 1,000 rooms in multiple roomblocks built around several different plazas. Most of the construction took place between 1315 and 1330. Creamer (1993) suggests that when construction peaks in 1330, at least 90 percent of the Component I rooms are in use. Allowing two rooms per household and three to six people per household, she estimates a peak population of 1,300 to 1,700 people (1993:153), much higher than previous estimates for the site (Wetterstrom 1986:51). Creamer (1993:153) acknowledges that any modification of her assumptions would lower the estimates; for example, if households use three rooms rather than two, the range becomes 900 to 1,800 people. In any case, Arroyo Hondo represents a large aggregate that takes shape quickly and is fairly short lived. Most people leave the pueblo by 1340, and there is a hiatus in occupation until the 1370s. Most of the 200 Component II rooms are built between 1370 and 1390, and if 90 percent are occupied at the peak of construction, the population is between 250 and 550 people (Creamer 1993:153).

San Marcos and San Lazaro Pueblos, both in the western Galisteo Basin, are known primarily from intensive surface collection and test excavation (San Marcos) and in-field analysis of surface ceramics (San Lazaro). Welker (1997:39-45) provides details for the various excavation and survey projects at San Marcos, beginning with Nels C. Nelson's work in 1915. She identifies the following four time periods at San Marcos that are correlated with the presence of different glaze wares: A.D. 1340-1425, A.D. 1425-1515, A.D. 1515-1600, and A.D. 1600-1700.

Welker (1997:87-90) calculates population in each time period at San Marcos using Nelson's room counts for each building. She assigns roomblocks to periods based on ceramics. Because most roomblocks are at least three rooms wide and probably are two stories, she assumes that a family has six rooms, two of which are living rooms (Welker 1997:87). To arrive at numbers of people, she divides the number of rooms in a building by six (rooms per family) and multiplies that by five (family size). Welker also tries multiplying the number of ground-floor rooms by two (people per room), but she feels the estimates are too large (Welker 1997:88-89).

San Marcos's population is not small in any time period (Figure 7.6). In the A.D. 1340-1425 period, 738 people are in residence. By the A.D. 1425-1515 period, there are nearly 1,700 people at the pueblo, which is the peak of population at the site. There is a substantial decline to 729 people between 1515 and 1600, which is roughly equivalent to the original population. The population drops to 536 people in the A.D. 1600-1700 period.

Figure 7.6. Estimated population at Pecos Pueblo, San Lazaro, San Marcos, and Arroyo Hondo. Note: Points are plotted on the midpoint of periods at each site except Arroyo Hondo. Pecos Pueblo is based on Figure 7.3.

San Lazaro pueblo, a few miles south of San Marcos, is equally large and impressive. There are two major occupations at San Lazaro, one prehistoric and one historic, on opposite sides of the Arroyo del Chorro. Very little work has been done at the site by professional archeologists. Nelson excavated 60 rooms in 26 roomblocks in 1912 (Nelson 1914), and the Museum of New Mexico and Southern Illinois University held a joint testing and surface collection project there in 1988-1989 (Smiley 1993).

In 1994, I examined the ceramics associated with the prehistoric and historic roomblocks and plazas at San Lazaro in order to reconstruct the occupation sequence of the roomblocks at the site (Ware et al. 1998:19). While tabulating ceramics, I also counted the number of rooms in each roomblock based on wall alignments or extent of rubble (Ware et al. 1998:20), estimating a range of 1,600 to 1,900 ground-floor rooms for the entire site. Informal excavations at San Lazaro reveal prehistoric household suites of three rooms, although it is possible they could range up to five rooms. For my population estimates, I assume three rooms and five people per household during the prehistoric and historic periods, and I also assume that portions of both pueblos have multiple stories (Ware et al. 1998:20, 21).

Based on the decorated ceramics at the site, there are several time periods at San Lazaro (Ware et al. 1998:19, 21-27). Roomblocks are dated to the different periods and population is estimated for each period (Figure 7.6). The first period, A.D. 1300-1330, is the one of least population, estimated in the range of 250 to 400 people. The earliest portion of the site is next to the arroyo, which washed away an unknown number of rooms over the years. The A.D. 1330-1400 period is a time of much construction and appears to be the height of population, estimated to range between 1,400 and 2,000 people. In the A.D. 1400-1425 period, population decreases slightly to between 1,100 and 1,600 people. In the following period, 1425 to 1450, there are 600-700 people at San Lazaro, a 35 to 40 percent decline. Another major building episode between 1450 and 1520 results in a resurgence of population to between 1,300 and 1,600 people, very close to the earlier peak in the A.D. 1330-1400 period. Within the first quarter of the sixteenth century, San Lazaro is abandoned, so that the A.D. 1520-1600 period represents a hiatus in occupation.

A large pueblo and mission were established in 1613 on the east side of the Arroyo del Chorro, apparently under the direction of the Spanish (John Ware, personal communication 1996). The historic pueblo at San Lazaro, built around a central plaza, is large and compact. The height of the rubble suggests the original roomblocks were two to three stories. Assuming three rooms per family, five people per family, and two to three stories, I estimate that 1,300 to 2,000 people occupied the pueblo during the A.D. 1600-1690 period. If the pueblo is one to two stories and rooms were larger than the prehistoric ones, as suggested by nonprofessional excavations at the site, a more likely range is 700 to 1,200 residents (Ware et al. 1998:25-26). The pueblo was abandoned after the Pueblo Revolt of 1680 but before the reconquest of Santa Fe in 1692 (Kessell 1987).

Figure 7.6 shows the population curves for Arroyo Hondo, San Lazaro, San Marcos, and Pecos pueblos. The curve for Pecos Pueblo is based on Welker's periods and room counts but uses a family size of seven people (Figure 7.3). Population is plotted at the midpoint of each period identified at a site, except Arroyo Hondo, for which Creamer's (1993:152-153) dates for the peaks of the two components are used. The occupation of Arroyo Hondo is declining when Pecos Pueblo and San Lazaro are growing and San Marcos is established. Whether there is any relationship between the decline at Arroyo Hondo and increases elsewhere is unknown.

It appears that San Lazaro is occupied earlier than San Marcos; however, this may reflect the different beginning dates that Welker and I use for glaze wares. She cites 1340 as the earliest date for glaze manufacture in the northern Rio Grande (Welker 1997:85). I prefer the more traditional beginning date of 1315, at least at San Lazaro, because the early glaze-on-red wares are being made just after 1300 in the south and are imported into the Galisteo Basin. The first areas occupied at San Lazaro are dominated by these glaze-on-reds. Locally made glaze-on-yellow ceramics do not appear at San Lazaro until slightly later.

There are a couple of peaks and valleys in the prehistoric occupation of San Lazaro. During the first stage of growth, San Lazaro and Pecos Pueblo grow and peak in the mid- to late-fourteenth century. Both enter a period of decline that is slower and gentler at Pecos Pueblo. San Marcos is established and grows while the others decline. San Marcos continues to grow to its peak population in the late 1400s. San Lazaro experiences another growth spurt during the 1400s and reaches a second population peak just slightly lower than the earlier one. Pecos Pueblo grows steadily as the other two decline in the 1500s.

Prehistoric San Lazaro is abandoned ca. A.D. 1515-1520, probably because of a raid, which is suggested by intact assemblages of household and ceremonial goods recovered from the latest roomblocks (Ware et al. 1998). San Lazaro may be the pueblo described by Castaneda (Winship 1896:523-524) that was in ruins, with stone balls scattered around "which seemed to have been thrown by engines or catapults" (Winship 1896:524), and had been destroyed 16 years before by the Teyas. According to Nelson (1914), San Lazaro is the only pueblo in the Galisteo Basin with an abundance of round sand stone concretions ("stone balls") in the surrounding hills.

In contrast to San Lazaro, the occupations at San Marcos and Pecos continue into the Historic Period, when missions are built at both sites. The location at San Lazaro is reoccupied in the early 1600s when a group of Pueblo Indians, under Spanish direction, built a pueblo and mission across the wash from the old prehistoric pueblo (Ware et al. 1998). The population at Pecos Pueblo continues to grow in the Historic Period, although historic documents indicate that major population decline begins before the middle of the seventeenth century (Kessell 1987:170). The last occupants leave Pecos Pueblo in 1838 and move to Jemez Pueblo. San Lazaro and San Marcos are abandoned much earlier, sometime between the Pueblo Revolt in 1680 and the Reconquest in 1692.

The persistence of Pecos Pueblo through time and the lack of wide swings in population like those at San Lazaro and San Marcos may have several explanations. One may be the advantage of using excavation data to identify and quantify earlier occupations. The use of surface data at San Marcos and San Lazaro potentially accentuates the dips in population because the ceramic assemblages associated with those periods are underrepresented. It is difficult to state that Pecos Pueblo is more stable without better subsurface information from the other pueblos. If Pecos Pueblo's population is more stable, it may be because of the consistent need to regulate trade between the Plains and Rio Grande Valley.

Conclusions

According to reconstructions using the Pecos survey data, plus information from Rowe and Hobson-Dressler, the number of people living in the Pecos area peaks in Period 3 (A.D. 1325-1450). The high rate of population growth from Period 2 (A.D. 1200-1325) to Period 3 suggests that a portion of the increase is through immigration. At the same time, in the northern Rio Grande region as a whole, population grows, levels off, then declines (Crown et al. 1996; Orcutt et al. 1994:Figure 9). When the northern Rio Grande is subdivided into smaller regions, we see that many of them experience growth, then decline between 1300 and 1450 (Figure 7.5).

Population in the Pecos area declines in Periods 4 (A.D. 1450-1575) and 5 (A.D. 1575-1700), and most people are living at Pecos Pueblo. At approximately the same time, some areas in the northern Rio Grande experience population decline, while others grow and then decline. According to Welker's estimates, the number of people residing at Pecos Pueblo between 1515 and 1700 reaches a peak, and population appears to increase in the Galisteo Basin and Jemez Mountains while it declines in other areas (Figure 7.5).

Climate and Pecos Area Population

Returning specifically to the Pecos area, it might be possible to understand population fluctuations, especially changes in population aggregation, in terms of climatic events affecting agricultural productivity (Kohler and Van West 1992; Orcutt 1999b). Benedict and Orcutt (Chapter 6) devised a dryness index for the Palmer Drought Severity Index (PDSI) to characterize Periods 2-5 (A.D. 1200-1700). The index uses a scale from 0 to 2 to transform the annual PDSI values to normal/wet (0), dry (1), and very dry (2). The index emphasizes drier conditions over moister ones because we are interested in the potential affect of droughts on agricultural productivity. We calculated five-year moving averages of the index and examined the fluctuations in each period (see Chapter 6 for details).

Kohler and Van West (1992) argue that periods of high agricultural productivity and high temporal variability in production promote aggregation, which facilitates sharing to even out irregularities in productivity. Even when overall productivity is high, temporal and spatial irregularities do occur that can leave some families with a shortfall. In the long run, everyone benefits from sharing as long as there is sufficient food to go around. The dryness index and annual variability in the actual PDSI values provide a relative view of productivity in Periods 2-5 (A.D. 1200-1700).

Period 2 (A.D. 1200-1325) (Figure 6.1) is the driest, and variability is high between approximately A.D. 1240 and 1300. On a relative scale, productivity in Period 2 probably is lower and variability is higher than the other periods; on that basis, I expect lower aggregation than during Periods 3-5 (A.D. 1325-1700).

Compared to Period 2 (A.D. 1200-1325), Period 3 (A.D. 1325-1450) (Figure 6.2) is not as dry, and variability is lower. Period 4 (A.D. 1450-1575) (Figure 6.3) is similar to Period 3, but there are more very dry years and fewer very wet ones. Both periods probably are characterized by higher potential productivity than Period 2. Although variability is lower than Period 2, aggregation should be a viable strategy to survive years of low agricultural production. I expect greater aggregation in Periods 3 and 4 than Period 2.

Period 5 (A.D. 1575-1700) (Figure 6.4) is the least dry and, therefore, probably has the highest agricultural potential. Variability is low until approximately 1680. Between 1680 and 1700, a critical period encompassing the Pueblo Revolt (1680) and Reconquest (1692), variability is high. On the basis of relative potential productivity, aggregation should remain high in Period 5.

Returning to the population estimates for all known and dated habitation sites in the Upper Pecos Valley, the peak of population size in Period 3 (A.D. 1325-1450) occurs during one of the better climatic periods. Population size declines in Periods 4 (A.D. 1450-1575) and 5 (A.D. 1575-1700), despite relatively good climatic interludes.

Population aggregation conforms well to the expectations based on the PDSI. I expect aggregation to be lowest in Period 2 (A.D. 1200-1325), and to increase in Periods 3-5 (A.D. 1325-1700). To test the proposition, I identify the size of the largest aggregate (i.e., the most people at any one site) in each time period (Table 7.3). The aggregate of 360 people in Period 2 is the smallest during the periods examined, as expected. Aggregates increase in size in Periods 3 and 4 (A.D. 1450-1575), reaching a peak in the latter. By Period 4, Pecos Pueblo is the only large aggregate, as measured with survey data. According to the survey data, Pecos Pueblo also is the only large aggregate in Period 5 (A.D. 1575-1700). The Period 5 aggregate is slightly smaller than Period 4, but it is much larger than those in Periods 2 or 3.

Based on Welker's (1997) reconstruction, it appears that an aggregate forms on the Pecos mesilla during Period 3 (A.D. 1325-1450), which is climatically more favorable for agriculture than Period 2 (A.D. 1200-1325). The Quadrangle at Pecos Pueblo takes shape during Welker's Period 3, which overlaps with the late portion of survey Period 3 through mid-Period 4 (A.D. 1450-1575). The height of population and aggregation at Pecos Pueblo is in Welker's Period 4, overlapping survey Periods 4 and 5 (A.D. 1575-1700). Except for the last 20 years of Welker's Period 4, this is one of the least dry periods in the Pecos area.

Overall, the survey data for all Pecos area habitations and Welker's reconstructions of Pecos Pueblo population indicate that aggregation is greatest during periods of higher agricultural potential. The PDSI is only one small aspect of climate and does not consider any social or other climatic factors that might affect aggregation. It is likely that aggregation in the Pecos area before Pecos Pueblo is established reflects sharing (Kohler and Van West 1992). In the specific case of Pecos Pueblo, nonclimatic factors facilitating its persistence through time include its strategic location at the gateway between the Great Plains and the Rio Grande Valley, and because of that location, the ability to regulate trade between Plains Indians and the northern Rio Grande pueblos. Because Pecos Pueblo clearly is designed for defense, warfare must be a factor in aggregation. The threat is not coming from within the Upper Pecos Valley because all residents are in Pecos Pueblo, so it is originating outside the valley, probably from other northern Rio Grande populations and from the Plains.