Go to Table of Contents.
List of Tables
List of Illustrations
Research Objectives and Methods
Architecture and Site Layout
Population Estimates
Faunal Remains
Archaeobotanical Remains
Human Skeletal Remains
Water Control and Subsistence
Abandonment and Emigration
Appendix A

Abandonment and Emigration

by Melissa J. Churchill

Archaeologists use the term "abandonment" to describe the point at which people ceased to use a particular place. Abandonment occurs at different scales, ranging from the abandonment of structures and activity areas within a given site to the abandonment of whole sites and entire regions (Cameron 1993*1:3). Some Pueblo people object to the use of the term "abandonment" in reference to the latter scenario because they believe ancient sites to be important and integral parts of the modern Pueblo world. Ernest Vallo, Sr., a member of Crow Canyon's Native American advisory group, states that he and other traditional leaders from the pueblo of Acoma in New Mexico do not recognize the word "abandonment" and that there is no record in his village's history of people "abandoning" the Mesa Verde region. Marie Reyna, a Native American from Taos Pueblo, also in New Mexico, provides this perspective:

We as Pueblo people see the relationship between past and present as one that has been constant and continuous for many generations. The past, present, and future are not separated; they are continuous [Reyna 2000*1:par. 3].

In an effort to accommodate the Native American perspective and to clearly distinguish between the different scales of population movement, I use the term "abandonment" in this chapter to describe the cessation of use of specific buildings or areas of a site, and I use the terms "migration," "movement," and "depopulation" to describe the departure of people from an entire site or region (cf. Naranjo 1995*1:247).

Theoretical Context

In this chapter, inferences about abandonment behavior are based on material-remains models developed by Ascher (1968*1), Lightfoot (1993*1), Schiffer (1972*1, 1985*1, 1987*1), Schlanger and Wilshusen (1993*1), Stevenson (1982*1), Varien (1999*1), and Wilshusen (1986*1). When people abandon a structure, they typically make decisions about what to take and what to leave behind (Schlanger and Wilshusen 1993*1:91). Their decisions are influenced by the distance they plan to travel, the length of time they have to prepare for the move, and whether or not they plan to return (Lightfoot 1993*1:166; Schiffer 1972*1:160, 1987*1:90–91; Stevenson 1982*1:238; Varien 1999*1:113). Usable items that are left behind in the locations in which they were used are called "de facto refuse" (Schiffer 1972*1:160, 1985*1:18, 1987*1:89).

When archaeologists find little or no de facto refuse in a structure, they generally assume that usable items were removed by the inhabitants of the buildings when they left or were salvaged by others still living nearby. In these cases, it is inferred that abandonment was planned and gradual, the distance to the new habitation was short, and/or return was not anticipated (Stevenson 1982*1:255, 259; Varien 1999*1:113). In contrast, when archaeologists find large quantities of de facto refuse, they assume that the items were left behind either because abandonment was unplanned or because the distance to the next destination was great (Stevenson 1982*1:255, 259; Varien 1999*1:113). It is also possible that de facto refuse on kiva floors was left behind as ritual offerings to the structures (Susan Ryan, personal communication 2000).

De facto refuse is most commonly found on the floors or other use surfaces of buildings; however, it is also recovered from features, roof fall, and extramural surfaces. Large pieces of wood in roof fall are interpreted as de facto refuse because roof timbers were valuable items that could have been—and sometimes were—salvaged and reused for construction or fuel (Varien 1999*1:116–121).

Kiva Abandonment

In this section, I discuss the abandonment of nine kivas that were tested at Woods Canyon Pueblo. I infer the time and mode of abandonment for each kiva on the basis of floor artifact assemblages, the treatment of roofs at abandonment, and the presence or absence of refuse in structure fill. The Woods Canyon data suggest that the entire village was not depopulated at the same time but, rather, that different parts of the site were abandoned at different times.

Floor Artifact Assemblages

I evaluated the floor assemblages from the tested kivas at Woods Canyon Pueblo using a framework derived from studies that quantify floor artifacts from excavated buildings (see Lightfoot 1993*1; Schlanger and Wilshusen 1993*1). It should be noted that the studies cited involved the examination of floor assemblages from completely excavated structures, whereas I worked with assemblages from structures at Woods Canyon Pueblo that were only partly excavated (approximately 2 m2 of each kiva floor was exposed). The average floor area of Pueblo III kivas in the Mesa Verde region is 12.3 m2 (Lipe 1989*1:56), so we probably exposed an average of only about one-sixth of the floor area of each tested kiva at Woods Canyon. Furthermore, half of the test units fell in the central portions of the kivas, an area that usually does not contain many artifacts (Table 1). Therefore, the interpretability of the data for Woods Canyon kivas is limited.

Initially, I calculated the counts and weights of materials classified as de facto refuse from the floors of the tested Woods Canyon kivas, but the assemblages were too small to be used for comparative purposes. Therefore, I decided to use the counts and weights of all floor artifacts from the tested kivas. These data are compiled and presented in Table 53 in "Artifacts," and the weights are repeated in the fourth column of Table 1, this chapter. Artifacts found directly on the floor or within the 5 cm of sediment above the floor are considered to be associated with the floor, even though they could have originated from the roof or been discarded when the roof was burned or salvaged. The counts and weights of pottery sherds, chipped stone, and other artifacts were summed. A count of partial or reconstructible vessels noted during lab analysis was also included in this analysis.

Several patterns emerge when the weights of floor assemblages from kivas in different parts of the site are compared (Table 1). Floor assemblages from kivas in the canyon bottom (Structures 1-S, 2-S, 3-S, and 9-S) weigh much less than the floor assemblages from kivas in the rim complex, upper west side, and east talus slope (Structures 4-S through 8-S). In fact, the average weight of floor assemblages from the kivas in the rim complex, upper west side, and east talus slope is 33 times greater than the average weight of floor assemblages from canyon-bottom kivas. The only de facto refuse found in the tested kivas in the canyon bottom were a bone awl, a bone tube, a core, and a projectile point. In contrast, the de facto refuse in the tested kivas on the east talus slope and the upper west side included a metate, manos, an abrader, a stone disk, a stone axe, bone awls, a bone scraper, and partial pottery bowls. Also, at least 10 deceased individuals were placed on the floor of one of the talus-slope kivas (Structure 5-S) (see "Human Skeletal Remains"). Because the interment of these individuals is the last known activity to have taken place in the kiva, it is believed to have coincided with the abandonment of the structure.

There are several plausible explanations as to why so few items were present on the floors of the canyon-bottom kivas. It is possible that most of the items were taken by the occupants of the canyon bottom when they left. Transport of possessions to a new location suggests that the residents moved only a short distance, perhaps upslope, to a different part of the village. Alternatively, the residents might have given or sold items to others or chosen to not replace broken items, because they were preparing to leave (cf. Schiffer 1972*1:159, 1985*1:27, 1987*1:97); they also could have cached most of their belongings in a different location, especially if they planned to return (cf. Stevenson 1982*1:252–255). Finally, many items might have been left behind only to be removed later by others still living at the pueblo (cf. Schiffer 1985*1:27–28, 1987*1:106–120).

The people who occupied the rim, cliff, and talus slope left behind many belongings, including heavy items, which suggests that they moved relatively far away. This scenario would be consistent with the interpretation that their departure coincided with the final emigrations out of the Mesa Verde region (Ortman et al. 2000*1:128). Another possible explanation is that objects were left behind as ritual offerings. If that were so, the distance the people were planning to travel may not have been their main consideration in leaving behind certain items. In conclusion, the contrasting modes of abandonment in these different parts of the site strongly suggest that the kivas on the east talus slope, upper west side, and canyon rim were abandoned later than were the canyon-bottom kivas.

Roof Treatment

Archaeologists can infer from the treatment of pit structure roofs at abandonment the distances that people intended to travel when they ceased to occupy a structure (Lightfoot 1993*1; Stevenson 1982*1; Varien 1999*1). Specifically, the salvaging of timbers (presumably for construction or use as firewood) suggests that people were moving relatively short distances, probably within the same locality (Varien 1999*1:115, 121). In contrast, timbers were more likely to have been left in place when people moved long distances (Varien 1999*1:118). In some cases, roofs were burned or partly burned, suggesting that the residents did not intend to return (Lightfoot 1993*1:166).

Kiva roof timbers decompose through natural processes. It is uncommon, however, for all the timbers in a kiva roof to decompose so completely that there is no evidence that they ever existed (Varien 1999*1:115). Four types of pit structure roof treatments have been identified on the basis of stratigraphic analyses at various sites in the Sand Canyon locality (Varien 1999*1:115):

Roof Treatment

Stratigraphic Evidence

entire roof burned

many large, burned beams

large timbers salvaged; small timbers burned

small pieces of charcoal mixed with roof-construction sediments

all timbers left in place, unburned

rotted, unburned wood and a poorly defined stratum of roof-construction sediments

all roof timbers salvaged, unburned

roof-construction sediments with no trace of roof timbers

Two of Varien's four roof treatments were identified in eight of the nine tested kivas at Woods Canyon Pueblo (Table 1); it was not possible to infer roof treatment for the ninth kiva, Structure 6-S, because only a small portion of the roof-fall deposit was exposed. In the structures for which a determination could be made, either the roof timbers had been completely salvaged, or the primary beams had been salvaged and the remainder of the timbers had been burned. In all eight cases, roof-fall deposits rested directly on the kiva floors, which indicates that the beams were burned or salvaged soon after the structures were abandoned, before natural sediment had had time to accumulate on the floors. In Structure 5-S, the bodies of at least 10 people were placed on the floor when the structure was abandoned. The observed rotation of the bones could have occurred only after the soft tissue had decomposed but before the roof beams were salvaged and burned (collapsed) (see "Human Skeletal Remains"). Because decomposition would have taken some time, the roof beams in this structure might not have been salvaged or burned as quickly as those in the other seven kivas.

It is not known to what extent the following analysis has been affected by the fact that the excavation units in half of the tested kivas (Structures 2-S, 3-S, 5-S, and 8-S) were located near the centers of the structures, which typically contain few fallen roof beams, especially primary-support beams. Thus, it is possible that more beams would have been exposed in these four kivas had the test units been closer to the kiva walls.

Kivas With Salvaged Roofs

All roof timbers had been salvaged from the kivas in the canyon bottom (Structures 1-S, 2-S, 3-S, and 9-S); in Structure 9-S, part of the bench had been dismantled as well, most likely for use elsewhere. Roof-construction sediment was the only roofing material present in the fill of these kivas. Salvaging implies reuse nearby. This pattern is common in kivas abandoned during the Pueblo III period in the Sand Canyon locality, indicating continuity of occupation of the local area (Varien 1999*1:119).

In the case of Woods Canyon Pueblo, it appears that the village continued to be occupied after these four kivas located in the canyon bottom were abandoned. Even though a complete survey of the area surrounding Woods Canyon Pueblo has not been conducted, this site currently is thought to have been the latest inhabited site in the area (see Lipe and Ortman 2000*1). It therefore seems likely that timbers were salvaged and reused by people still living at the pueblo. It is possible that people residing in the canyon bottom salvaged the beams and moved upslope or that other villagers reused the beams from the canyon bottom to build their homes. In either case, it is clear that the beams were salvaged soon after the kivas were abandoned, because no sediment had accumulated on the floors before the roofs collapsed.

Kivas With Salvaged and Burned Roofs

In contrast to the roofs of kivas in the canyon bottom, the roofs of the tested kivas on the upper west side and east talus slope appear to have been intentionally burned after primary beams were salvaged (Structures 4-S, 5-S, 7-S, and 8-S). In all these structures, small, burned beam fragments were found in the roof-fall deposits. In one kiva (Structure 5-S), a single burned primary beam was found, making it more difficult for us to determine whether all or only part of the roof had burned (it is possible that additional burned primary beams are present in the unexcavated portion of the kiva, in which case an argument could be made that the entire roof had burned). In addition, the outlines of several burned beams were recognizable, and it is at least possible that some beams had disintegrated completely. Nevertheless, the presence of only one confirmed beam makes it more likely that most of the large timbers were salvaged, after which the remainder of the roof was burned.

In the Sand Canyon locality, the pattern of salvaging large timbers and burning smaller ones is common only in tested structures dated to the A.D. 1250–1290 time period. This is most apparent at Castle Rock Pueblo, a well-dated site whose occupation ended near the time of the final Pueblo emigrations out of the Mesa Verde region (Kuckelman 2000*1; Varien 1999*1:121). An exception to the pattern is found at Sand Canyon Pueblo, a large, late Pueblo III village. The most common roof treatment at Sand Canyon Pueblo was for entire kiva roofs to be burned, with no salvaging of timbers beforehand (Varien 1999*1:116).

It was not possible to determine the roof treatment of the one tested kiva (Structure 6-S) in the rim complex at Woods Canyon Pueblo. However, burned adobe was abundant in this kiva, which might indicate that at least part of the roof burned. The in situ beams in the rooms and the timbers lying on the modern ground surface in the rim complex indicate that wood was not salvaged but left in place, which in turn suggests that people moved a long distance. On the other hand, the wood left behind was much smaller than typical primary beams in kivas and may not have been considered as valuable.

Roof Treatment and Site Chronology

The study of kiva roof treatment adds to our understanding of the chronological history of Woods Canyon Pueblo. All the roof timbers from the canyon-bottom kivas appear to have been salvaged, whereas the roofs of tested kivas on the upper west side and east talus slope appear to have been partly burned and partly salvaged. Timbers were not salvaged from visible above-ground structures in the rim complex. The results of this analysis indicate that the canyon bottom was the first area to be occupied and the first to be abandoned. The roofs were most likely dismantled and used for construction and fuel in other parts of the village. When the kivas on the east talus slope and upper west side were abandoned, valuable primary beams appear to have been salvaged, but the remaining roof timbers were burned.

The salvaging of beams from the late kivas on the upper west side and the east talus slope implies that people were still living at the pueblo near the end of the thirteenth century. Some residents may have chosen to remain at the pueblo after others had emigrated (cf. Duff and Wilshusen 2000*1). Residents might have lived in the existing buildings and used the salvaged timbers as firewood, or they might have used the timbers to construct new buildings. If the beams were used for construction, it seems reasonable that they were used in the rim complex because this area appears on the basis of tree-ring and pottery data to have been one of the latest-occupied areas at the site (see "Chronology").

Alternatively, the inference that timbers were salvaged for reuse might be incorrect. Although it seems logical that large timbers would have been valuable materials worth recycling, it is also possible that primary timbers were removed as part of an abandonment ritual at the end of site occupation. Several researchers (Lightfoot 1994*1; Varien 1999*1; Wilshusen 1986*1) have argued that intentional burning of pit structures was part of ritual abandonment. Wilshusen (1986*1:247) makes a persuasive argument that pit structures did not accidentally catch on fire and that, in fact, it was difficult to burn pit structure roofs. Primary timbers might have been removed to facilitate burning. If this is true, it seems likely that the primary timbers would have been added to the fire after it was started and that they therefore would be present in the roof-fall deposits. An alternative explanation is that the roofs of the later kivas were burned, with large timbers not consumed in the fire being removed as part of the abandonment ritual (Scott Ortman, personal communication 2001).

Attempting to explain why timbers were burned and/or removed is beyond the scope of this chapter. Even so, it is worth considering the different kinds of behavior that may have produced the archaeological signature currently interpreted as evidence of a salvaged and burned roof. The inference that people were still living at Woods Canyon Pueblo after the tested kivas on the east talus slope and upper west side were abandoned hinges on the argument that primary beams were salvaged from these kivas for reuse nearby. If this inference is incorrect, these kivas in fact might have been abandoned during the final days at Woods Canyon Pueblo, and there may have been no residual population at the site after they were abandoned.

Finally, it is possible that the roof-fall deposits exposed during partial excavation do not adequately represent what the roof-fall deposits are like in the unexcavated portions of the kivas. If burned primary timbers are present in the unexcavated parts of the kivas on the east talus slope and upper west side, then those structures were abandoned at the end of site occupation rather than before it.

Secondary Refuse in Kiva Fill

The presence of secondary refuse—also called midden or trash—in the fill of an abandoned structure indicates continued occupation at a site (Varien 1999*1:121). Deposits in the fill of Structures 1-S and 7-S perhaps provide evidence of continued use of the site after the structures were abandoned (Table 1). A deposit of dark gray sediment containing abundant charcoal rests above roof fall in Structure 1-S. It is not typical of ashy midden deposits, but it may be cultural material that was discarded after the roof was salvaged. In Structure 7-S, which had a high density of artifacts in the fill above roof fall, it was unclear whether the artifacts had been deliberately discarded in the abandoned structure or naturally redeposited from Nonstructure 10-N, a rich midden deposit located upslope from the kiva (boulders on the north side of the kiva might have prevented the trash deposits from eroding into the kiva depression).


Data gathered during the limited testing of kivas at Woods Canyon Pueblo indicate that these structures were abandoned in different ways and at different times. First, there is a strong correlation between the type of roof treatment at abandonment and the type, quantity, and condition of items found on the floors. Kivas with completely salvaged roof beams, all of which are located in the canyon bottom, contained the fewest floor artifacts. Of the artifacts found, most were broken; the few usable (de facto) items present were small and lightweight and included bone tools and a projectile point. In contrast, kivas with partly burned and partly salvaged roofs, all of which are located on the upper west side and east talus slope, had more artifacts on their floors, and the usable items left behind were heavy and bulky, including ground-stone tools and partial and/or reconstructible vessels. The floor assemblage in the tested kiva for which roof treatment could not be determined (located in the rim complex) was similar to the assemblages found in the partly burned–partly salvaged structures. All these data suggest that some kivas (those with completely salvaged beams and little de facto refuse) were abandoned by people who moved only a short distance, perhaps to other areas of the site, whereas others (those with both burned and salvaged roofs and more de facto refuse) were abandoned by people who intended to move farther away, perhaps as part of the larger emigrations from the region as a whole. These data also generally support the conclusions drawn from the pottery and tree-ring analyses—that is, the canyon-bottom kivas were abandoned earlier than kivas in other parts of the site, and usable roof timbers and artifacts were salvaged, either by the occupants as they left the structures or by other villagers.

What is perplexing, however, is that the large, heavy roof timbers of the kivas with de facto floor refuse appear to have been partly salvaged, which suggests that they were reused nearby. De facto refuse is generally left behind by people moving far away. Either people still living at Woods Canyon Pueblo recycled roof timbers but did not salvage usable floor items, or our assumptions about kiva abandonment are incorrect. Future research should examine whether other kivas dating from the late Pueblo III period have burned and salvaged roofs with de facto floor refuse. If so, further discussions about the kinds of behavior that produce this archaeological signature are needed.

References cited | To borrow, cite, or request permission

Copyright © 2002 by Crow Canyon Archaeological Center. All rights reserved.