Architectural Variability in the Caddo Area of Eastern Texas

................................................................................................................................................... i Acknowledgements ................................................................................................................................. ii Table of


INTRODUCTION INTRODUCTION
The cultural and archaeological tradition commonly recognized as Caddo or southern Caddoan-language speakers developed around A.D. 800-900 in northeast Texas, southeast Oklahoma, southwest Arkansas, and northwest Louisiana. This suite of developments included the appearance of hierarchically ranked sedentary societies, evidence for extensive interregional trade, the construction of well-planned centers marked by monumental architecture, and the appearance of elaborate mortuary rituals and ceremonial practices associated with complex symbolism (Perttula and Bruseth 1998;Perttula 1996Perttula , 1998Story 1981Story , 1990. These developments are "characterized by dispersed but sedentary settlements, a horticultural to agricultural economy, and a complex socio-political structure (what Sabo [1995] calls a theocracy) denoted principally by a heterarchical network of mound centers and the differential treatment of the dead by rank, most notably in burial mound shaft tombs accompanied by elaborate grave goods" (Perttula 1996:297).
The Caddo archaeological area ( Figure 1) was home to culturally diverse groups with some shared traditions and practices (Schambach 1982:7;Story 1990). While the "Caddo" reference implies a sense of cultural unity, shared traditions, or shared identities, there is considerable evidence of variability among the groups generally subsumed under this moniker. This variability is evident first from pre-Columbian archaeological contexts as well as from the early periods of sustained interaction with Europeans beginning in the late 17 th century A.D. Despite this variability, however, there is clear evidence for shared 2 traditions and practices throughout the area. Traditions and practices associated with both prehistoric and historic Caddo architecture are the subject of this dissertation. For the late pre-Columbian to early Historic periods this study employs the chronology proposed by Story (1990:334). Table 1  Caddo peoples (Perttula 1997;Schambach 1982;Story 1990). These traditions appear to have emerged from local Woodland period traditions and practices (Perttula and Bruseth 1998;Schambach 1982aSchambach , 1982bStory 1990;Thurmond 1990) with "definite Southern Caddoan complexes…over virtually all of the Caddoan Area by A.D. 1000" (Story 1990:323). Table 1. Chronological framework (Story 1990:334 The Caddo tradition is marked by the development of a hierarchically ranked social structure that included positions of ascribed and achieved status such as political and religious leaders, aids, counselors, priests, warriors, and commoners (Sabo 1998;Smith 1995;Wyckoff and Baugh 1980). Differential treatment of the dead is indicated by elite burials in mounds and cemeteries. Such burials were occasionally accompanied by grave goods made of exotic materials such as marine shell and copper acquired through long-distance exchange, indicating expansive trade networks (Brown 1983;Perttula 1997;Story 1990).
Although diverse in nature, certain trends and patterns began to emerge throughout the area around A.D. 800 and these continued to be developed, maintained, altered, and negotiated well into historic times. Caddo groups lived in dispersed but 4 sedentary settlements that tended to be situated along the major rivers and minor streams and tributaries. These settlements ranged from isolated farmsteads and small hamlets to villages and large community centers.
Dee Ann Story (1990:334-341) has discussed four primary types of archaeological sites that characterize the post-A.D. 800-900 settlement pattern of the Southern Caddo groups. One class of sites includes limited use areas which are "special purpose camps and other areas where more ephemeral activities are carried out" (Story 1990:334). Burial loci comprise another type of site that includes burial mounds as well as family and community cemeteries. The last two site types in Story's classification, the types of sites from which most of the data used in this study come, are habitation loci, "marked by refuse, thinly distributed sheet trash as well as concentrated midden deposits, and cultural features" (Story 1990:334), as well as community/ceremonial centers. In her discussion of this final site category, Story (1990:340) states: In contrast to the rather limited occurrence of burial mounds, mounds associated with structures -capping a structure and/or providing a platform for a structureare fairly common, though not truly numerous. They appear to denote former community centers and as such to have been an integral element in most prehistoric Caddoan settlement systems.
Throughout the early years of archaeological research in the Caddo area, emphasis was primarily given to the excavation of mounds and burials. Given this history, "the most fundamental building blocks of the Caddoan settlement system -the household and community -are poorly understood" (Story 1990:336). Despite this, with continued excavations since the 1930s, there is an abundance of data available for studies of architecture and architectural space from sites throughout the Caddo area. Story 5 (1990:342) notes, "the community centers, especially, the structures they contain, are highly significant, rich in variety, and probably diverse in function".
This dissertation focuses on the nature of architecture and architectural space in the Caddo area of eastern Texas, in the southwestern portion of the Caddoan archaeological area. The study area is within the West Gulf Coastal Plain (Fenneman 1938:100-120; Story 1990) and is partly situated in the biogeographical area referred to as the Trans-Mississippi South (Schambach 1982a:7-10;1998:xi-xii). The study area is environmentally diverse and spans a variety of biotic zones including the Post Oak Savanna, the Blackland Prairie, and the Pineywoods (Gould 1962; Story 1990). The major drainages of this environmentally diverse region are the Red, Cypress, Sabine, Sulphur, Neches, and Trinity rivers.
Caddo architecture, the primary subject of this dissertation, includes structures that range in shape from round or ovoid to rectangular or square. Evidence for these wooden pole grass and cane-covered structures includes post holes or post molds, wall trenches, wattle-impressed daub fragments, hearths, pits, the occasional prepared floor, and the occasional well-preserved burned superstructure of a building that includes structural elements such as charred timbers. The early European accounts and the Caddo archaeological record indicates that there was a wide range in size, shape, form, and use of architectural space in the Caddo area. Buildings have a variety of structural attributes, including partitions and extended entranceways, and may be found isolated or associated with plazas or earthen mounds. From these and other attributes, there remain several architectural elements that provide valuable insight into the nature of Caddo architecture and architectural space. Attributes such as structure size, shape, associated features, post placement, size and depth, orientation, and the spatial and temporal relationships 6 between, as well as the geographic distribution of, different structures provide important elements for study.
This dissertation is a detailed examination of the architectural diversity found throughout the Caddo area of eastern Texas. The sites included in this study range from large multi-mound centers that have seen large-scale and long-term research, such as the George C. Davis site (41CE19), to the smaller hamlets and farmsteads that are as Story (Story 1990:336) puts it, "the most fundamental building blocks of the Caddoan settlement system". This study includes 265 structures from 31 sites throughout a small part of the broader Caddoan archaeological area, the Pineywoods, Post Oak Savanna and Blackland Prairie of eastern Texas.
The detailed portions of this study include data solely from structures identified on the basis of post hole patterns that are complete enough to provide suggestions of structure shape and size. These criteria preclude the inclusion of structures from sites where there are clear patterns of artifacts or features that suggest the presence of structures but lack identifiable post hole patterns. Structure areas from sites such as Hudnall-Pirtle (41RK4) (Bruseth 1991;Bruseth and Perttula 2006), Jamestown Mounds  (Middlebrook 1993), Tallow Grove (41NA231), Beech Ridge (41NA242), and Naconiche Creek (41NA236) , and several others, fall into this later category.
As mentioned, 265 structures are included in this study of Caddo sites from eastern Texas. While this represents a large sample of the known Caddo structures from Texas, the list is not exhaustive. These structures include previously excavated structures and those that have been identified through geophysical methods. The majority of the 265 7 structures are circular with simple entrances, although sub-square, rectangular, and subround structures are represented throughout the area. These structures range from small circular features (covering an area less than 2 m 2 ) that likely represent the footprints of granaries or storage platforms (Rogers and Perttula 2004;Perttula 2005) to large structures covering over 200 m 2 that are associated with earthen mounds.

LAYOUT OF THE DISSERTATION
With Chapter 1 as a general introduction to the research and the study area, Chapter 2 discusses some of the previous research on architecture in the Caddo area.
Building upon the foundation provided in Chapters 1 and 2, Chapter 3 is a detailed look at accounts from primarily seventeenth and eighteenth century Spanish and French missionaries and explorers as well as nineteenth century images and accounts from the Caddo area.
While the early European accounts clearly touch on the architectural diversity in the Caddo area, the variations in the architectural traditions are most evident in the archaeological record. These traditions include a wide range of variability in structure size, shape, form, construction techniques, context, and use. In order to examine Caddo architecture as evidenced in the archaeological record, Chapters 4-6 examine excavated structures from the study area. Chapter 4 provides a detailed discussion of the architectural features from the George C. Davis site (41CE19) and offers an examination of archaeogeophysical data from recent research at this major mound site. The specific details on the data from this recent work at the site is presented in Appendix A. Chapter 5 presents details on excavated Caddo structures associated with earthen mounds from eastern Texas. Chapter 6 details excavated structures from eastern Texas that are not 8 associated with earthen mounds. Chapter 7 provides a comparison of mound and nonmound architecture. Chapter 8, the concluding chapter, provides a discussion of the nature of Caddo architecture from eastern Texas.

INTRODUCTION
The production of architecture "is the deliberate creation of space made tangible, visible, and sensible" (Tilley 1994:17). Architectural space, like the broader notion of community, is dynamic and socially constituted; thus, it is dependent upon human agency, practice, and tradition. Such acts as the creation, use, and abandonment/destruction of architectural space are dependent upon human agency, practice, and tradition. Therefore, architecture reflects aspects of the broader cultural realm in which it was created (Yaeger and Canuto 2000;Lewis et al. 1998;Story 1998;Sabo 1998;Tilley 1994;Wesson 1997). In this sense, architectural space can be understood as the material expression or physical manifestation of social, political, and ideological concepts.
The cultural landscape or built environment is inscribed with affirmations of identity, ideology, and tradition, and, through its historicity it provides a context for cultural practice (Alt 2001;Bourdieu 1977;Sullivan and Rodning 2001). Along with this contextual structure, the cultural landscape of a particular people is indicative of the creation and perpetuation of a shared history and creates a sense of the continuity or change of traditions. Ephemeral or esoteric concepts related to ideology, identity, and community are made physical or tangible by the construction, destruction, control, and manipulation of both the cultural landscape and architectural space. Additionally, cultural distinctions or categories centered on such things as status, access, kinship, or group membership can be revealed in the detailed study of the cultural landscape and architectural spaces.
The Caddo archaeological record is rich with data pertaining to architecture.
These architectural data occur from the earliest to latest Caddo occupation periods throughout the region. These data suggest a range in size, shape, form, and use of various structures. Structures range from round or ovoid to rectangular. Additionally, there is strong evidence for a differentiation between "specialized" and "non-specialized" architectural space (Rogers 1982;Sabo 1998;Story 1990). The distinction between such spaces is based not only on structure location within a specific site or community but on structural form. "Specialized structures," for instance, are often identified on Caddo sites by a number of characteristics, including those structures found in association with mounds and those showing apparent signs of "deliberate destruction, capping with earth, and, often subsequent rebuilding on the resultant earthen platform" (Rogers 1982:105;Story 1990:340-341). Additionally, they may exhibit extended entranceways, are associated with special artifacts or features, and are, in some instances, buildings with unusual architectural details, such as their atypical large size, partitions, etc. (Rogers 1982:105;Story 1990:341-342).
While a comprehensive examination of architectural practices from throughout the Caddoan archaeological area has not been completed to date, there are previous studies that have examined architectural practices, variations, architectural change, and specific structural elements from the various parts of the region. This chapter provides a discussion of previous studies focusing on architecture in the Caddoan archaeological area in general and in the East Texas study area specifically.

PREVIOUS ARCHITECTURAL STUDIES IN THE CADDOAN AREA
In addition to the excavations of burial mounds, M. R. Harrington's (1920) work included excavations of several structures from sites between the Red and Ouachita rivers in Southwest Arkansas. These structures include some in mounds and some built at ground level and covered by a small earthen mound upon or after abandonment.
Harrington (1920)  Based on Caddo ethnohistoric materials available at the time, Harrington (1920) discussed three primary types of Caddo buildings. The first primary type of building was the grass-lodge of the Caddo (Figure 2), a "dome-shaped edifice of poles covered with thatch, apparently identical with the "grass house" built by the Wichita" (Harrington 1920:247). These structures have "grass thatch extending entirely to the ground" (Harrington 1920:250). A second structure type discussed by Harrington (1920:252-253) was the walled house ( Figure 3), "which must have had a wall of upright poles, five or six feet high, with canes interwoven to serve as lath, and then plastered with mud, probably mixed with Spanish moss, as were the houses of the Natchez, the whole surmounted by the domed roof of thatch." The final structure type discussed by Harrington (1920) was the earth-lodge. Harrington (1920) argued that the burned 12 structures covered by the low earthen mounds, such as the one he investigated at the Washington group, represented earth lodges similar to those built by the Pawnee.
Harrington (1920:256-258) suggested that these structures were "erected by first constructing a frame, probably in the form of a low dome, of very stout posts, upon which were placed smaller ones at right angles. These in turn were covered with brush and cane, and then finally with sedge or 'sage' grass, when the structure was ready for its heavy coating of earth."  Clarence Webb's (1940Webb's ( , 1942 analysis of the structures from the Belcher site in Caddo Parish, Louisiana, provided a detailed look at architecture from a specific site in the Red River basin. Webb's excavations identified eight separate structures in the Belcher mound. Webb briefly reviewed the architectural characteristics of the earth lodges of the Pawnee and other Plains Caddoan groups, the "fire temple" of the Natchez and Taensas, and the large council houses of the Cherokees and Creeks. Webb (1940) provided details of the Belcher structures: four of the structures were superimposed one over the other and separated by approximately 0.6 m of sand with the fifth structure built on the ground surface adjoining Mound B (Webb 1940:54). Webb's (1940) analysis of house types and their associated artifacts and features led him to six conclusions. First, the prevailing structure forms among the Southern Caddo groups were "grass lodges and lodges with wattle and daub walls and grass thatched roofs" (Webb 1940:73). Second, characteristics of the Belcher structures indicated the practices of building different types of structures and the burning of structures at the end of their use. Third, later Caddo groups built circular structures using 14 wattle and daub walls and split cane and grass roofing. Forth, the rectangular structures at the Belcher site were built by an earlier group of Caddos that produced pottery with "simpler incised straight lines," and built structures using a wall-trench technique similar to that used in Middle Mississippian construction (Webb 1940:73). Fifth, there was no evidence of earth lodges used by Southern Caddo groups. Finally, the practice of digging deep pits and burials through the floors of structures while the structures were in use is not evident (Webb 1940).
Building upon Webb's analysis of the Belcher site structures, Edwin Wilmsen (1961) provided a brief analysis of structure data and suggested a developmental sequence for Caddo architectural forms. This analysis was based largely on a comparison of structure shapes from the George C. Davis, Belcher Mound, and Spiro sites, as well as external comparisons with structures from Tchefuncte, Adena, and Huasteca sites. Basing his study on construction details, Wilmsen (1961) arrived at a structural sequence tracing architectural form from the Early Caddo period (his Gibson Phase 1 and 2) through the early colonial period (Table 2). Another detailed study of Caddo architecture was Carolyn Spock's (1977) Master's thesis. Spock's study was an analysis of the architectural remains from the George C. Davis site in Cherokee County, Texas. Spock examined the 52 excavated structures from the site and provided a classification scheme that divided the structures into two broad classes, domiciles and special function. Based partly on form and partly on location, there has long been an analytical differentiation between what are often termed "specialized" and "non-specialized" architectural spaces (Rogers 1982;Lisk 1984:74-75;Sabo 1998;Story 1990). This concept of specialized and non-specialized structures has long been used throughout the Caddoan archaeological area. Spock's (1977) most numerous class of structures, the domiciles, included those structures not associated with mounds and those generally lacking "specialized" structural attributes such as partitions and extended entrances. Spock's more elaborate classification concerned the special function structures. The specialized class of structures included those structures directly associated with mounds (although this is not the only classificatory requirement). It also included structures that were exceptionally large or small in size, unusually shaped, or characterized by unusual or distinctive interior features (Spock 1977:169). The special function structures were further subdivided into four subclasses (Spock 1977:169): • based on size (unusually small structures, less than 2 m in diamter) • mound association (on mound platforms or pre-mound structures) • non-mound structures with distinctive architectural form and/or associated features Another detailed site-specific analysis of Caddo architecture is Carolyn Good's (1982) Good's (1982:67-69) study addressed the probable duration of structures, between 10-15 years, which provided an estimate for the duration of occupation at the site of 30-80 years, based on structure overlapping. Through the analysis of the structure and burial data, Good (1982:93-95) provided a demographic analysis to estimate that the population of Deshazo ranged from 11 to 55 individuals depending upon the combinations of possibly contemporaneous structures. Good (1982:98) concluded that in terms of spatial arrangement, the site consisted of an open plaza with structures at both the north and south ends of the habitation area and a related cemetery across the creek to the south.
For the northern portion of the Caddoan area, Daniel Rogers (1982aRogers ( , 1982b provided a regional study in his examination of specialized structures from the Arkansas River Basin and Ozark Uplift. Rogers (1982a:49) defined specialized structures as "any of the variety of structures that provided a physical context for the integration of social organization beyond that of the household unit." This class of structures includes mortuaries, meeting halls, temples, elite residences, or other public buildings (Rogers 1982a(Rogers :105, 1982b. Archaeologically, Rogers defined specialized structures as those directly associated with earthen mounds, those that appeared larger than typical structures, those "that are markedly different from contemporaneous domestic dwellings" (Rogers 1982b:49), and those associated with atypical artifact assemblages or features (Rogers 1982a:105).
Focusing his study on the variability of structures geographically, functionally, and temporally, Rogers (1982aRogers ( :105, 1982b  First, the evidence suggested that the practice of covering structural remains with earthen mounds was established as early as the Fourche Maline phase, although the evidence for specialized structures during this period (n=2) was scant (Rogers 1982a(Rogers , 1982b. Like the architectural data from the Fourche Maline phase, the late Fort Coffee Phase had only two (or possibly three) structures tentatively designated as specialized in function. These structures differed from those of the other phases in that they were round and lacked interior supports. The lack of supporting contextual data concerning this rare structural form limited the interpretations that could be made concerning specialized buildings from this phase (Rogers 1982b).
Most of the data Rogers (1982b) considered was from Harlan and Spiro phase occupations. During the Harlan phase, specialized structures (n=29) were mostly square with four interior posts. They had extended wall-trench entranceways, and were relatively large. In addition to these structural features, Harlan phase specialized structures were generally burned or dismantled with the remains being covered by a small earthen mound (Rogers 1982a:87). Rogers (1982b) noted that a significant shift occurred during the Spiro phase, in that specialized structures were generally smaller than in the preceding Harlan phase and were mostly rectangular with two center posts, a pattern of construction similar to 'ordinary' structures (Rogers 1982a:107). Also in the Spiro phase, "the practice of constructing single episode mounds over building remains disappears and special buildings are found associated with platform mounds or platform-burial mound combinations" (Rogers 1982b:87). Rogers (1982b) interpreted these changes as relating to possible changes in residence rules or changes in the functions of specialized structures. As for the smaller size, the reduction in the number of specialized structure, and the possibility of functional change, he suggested that such shifts may be, "considered as one correlate of the centralization of authority and the further segregation of the elite" (Rogers 1982b:90). Rogers (1982:108) concluded his consideration of specialized structures by arguing that during the Harlan phase, specialized structures "were probably used as combined elite residences, temples, and mortuaries" while in the Spiro phase this class of buildings "served as mortuaries on platform and burial mounds; however, other possible functions are not known." In addition to these studies, there have been studies of particular architectural elements of Caddo structures. These include an analysis of the frequency and location of interior hearths (Middlebrook and Middlebrook 1996) and entrance types (Kay and Sabo 2006;Perttula 2009), studies that include identifying different kinds of structure types (Perttula and Rogers 2007;Walker 2009), and a recent study on burning of structures from southwest Arkansas (Trubitt 2010). Kay and Sabo (2006) completed a study of the structures with extended entrances from the Arkansas River Valley and western Ozark highlands. In this area, these structures are often rectangular. Structures with extended entrances are often considered specialized structures and are not as common as those with non-extended entrances. The extended entrance would set the structures apart from the rest of the buildings on a site or in the area and would seemingly provide limited access to the structure's interior. Kay and Sabo (2006:44) found that the Harlan-style charnel houses in this region had extended "entryways oriented generally to the southwest, thus aligned with the winter solstice sunset," and the direction of death. Kay and Sabo (2006:44) further suggested that these specialized structures are "an intelligible or coherent measure of widely shared death:winter symbolism and mortuary ritual." In a response to Kay and Sabo's study, Perttula (2009) examined 79 extended entranceway structures from 39 sites distributed throughout the Caddo area. Perttula (2009:36) found that there was a range of orientation of entrances, with more north or south-oriented buildings occurring in the central Caddo area; buildings oriented to the sunrise occurred in the northern, southern, and central Caddo areas; more of the southern Caddo area entrances were oriented toward the sunset, and "winter-oriented special purpose structures are most common in the northern and central areas, with summeroriented special purpose structures more important in the southern Caddo area." Perttula (2009:39) found that death symbolism was not the primary determinant in building orientation throughout all parts of the Caddo area and that there was a clear practice of orienting buildings based on "site-specific decisions on the place of specialized buildings within living communities of Caddo peoples, which were built by a series of matrilineal households to face the social community, the village, the plaza, and other important ritual places in the Caddo constructed landscape writ large." A recent study from Trubitt (2010) examined variations in Caddo architecture from southwest Arkansas. Specifically, Trubitt (2010:233) focused on sites dating from the twelfth to the seventeenth centuries A.D. and looked at "structure shape, on burning 20 at termination, and on covering with soil". Trubitt (2010:242) noted that circular structures are the Caddo architectural norm in southwest Arkansas and "there is no clear temporal shift from straight sided to circular structures (or vice versa), and mound structures of both shapes have been interpreted as "special purpose, public, or ritual buildings." Trubitt (2010:242) has suggested that there may be a significant difference between covering a structure with clean fill versus midden soil and that "burying burning structures did not always mean creating mounds." She also noted that "perhaps covering with a layer of clean earth is what distinguished the mound and off-mound locations, suggesting that burning structures in certain places was marked by ritual" (Trubitt 2010:243). In discussing the different shapes and treatment of structures, circular vs. rectangular, burned, or not burned, Trubitt (2010:243) suggested that "the variation indicates more complexity than circular = domestic versus rectangular = public/ritual, burned = special purpose, or burned = mound construction." In addition to these broader regional studies, there are several site-specific studies of Caddo architecture from throughout the Caddo area, some of which are discussed in more detail in later chapters of this dissertation. For example, for the Oak Hill Village site (41RK214), Perttula and Rogers (2007:76) Early (1988:163) concluded that the pre-mound circular structure was likely a domicile, while the rectangular structure (Feature 18) was likely a special purpose building. This rectangular building was covered by a low mound and subsequent rectangular buildings, with the earlier circular structure being covered by the mound in later construction episodes (Early 1988:163). Early (1988:162) suggested that the changes in building styles at Standridge related to "changes in function of the total site." Early (1988:163) noted that "in its initial Caddoan occupation the site served dual purposes as both a residence and a location for small scale ritual. From this beginning it evolved into a location used exclusively for special purpose activities, but on a seemingly small scale." At the Winding Stair site in the Ouachita Mountains, Early (2000:128) excavated the remains of a specialized structure dating to the 15 th century A.D. that may have served as an elite residence or "may have served as a setting for community rituals, especially those concerned with aspects of the food gathering and growing cycles." This structure was similar to the specialized structures from the Standridge mound (Early 1988). Early (2000:130) also discussed other specialized buildings from the Arkansas

22
Valley, the Little Missouri River valley, the Ouachita River drainage basin, and the Ouachita Mountains, and noted that "these 'special' buildings were focal points of dispersed Caddo communities. They conform to a widely shared and strikingly uniform set of ideas about shape and design in the built environment of socioreligious centers, and they were treated with a seemingly similar set of ritual practices when they were taken out of use." Early (2000:130) further noted that "rather than marking a cultural boundary between the Arkansas River valley and the southern Caddo area, they indicate that there was a shared set of fundamental values and practices between these two regions that support the notion of a common cultural heritage." Through these studies, it is clear that there are shared architectural practices in the Caddo area and that these shared traditions "can only strengthen the idea that peoples living in the north, central, and southern parts of the Caddo archaeological area from as early as ca. A.D. 900 were Caddo peoples with the same cultural heritage" (Perttula 2009:39 INTRODUCTION The Caddo area provides an excellent opportunity for research that utilizes both archaeological and ethnohistoric material. As Story (1990:346) has argued, we have the opportunity in the case of the Caddo to relate specific historical groups and their activities with the rich archaeological record of their pre-Columbian predecessors. This is due to the abundant and diverse archaeological record and, as Story (1990:346) notes, "because the Late prehistoric groups were relatively sedentary, were not severely displaced by intrusive Native Americans…prior to the early nineteenth century, and were visited fairly often by Europeans and Euramericans." Without integrating these diverse historical and archaeological datasets, "archaeologists can be guilty of overlooking valuable clues and the key actors of the past -the people who left the remains they study" (Story 1990:323).
With the increased use and relative success of geophysical prospection in locating buried Caddo structures and other features, the potential for such research is enhanced .
Studies of sixteenth, seventeenth, and eighteenth century European accounts and the nineteenth century images from the Caddo area have guided our perceptions concerning buildings and settlements of early colonial and late pre-Columbian Caddo groups. They have also informed us about the nature of Caddoan leadership, social organization, settlement, trade and interaction, ceremonialism, and architecture (e.g., Baugh 1998;Bolton 1908Bolton , 1987Griffith 1954;Perttula 1997;Perttula et al. 2008;Sabo 1998;Smith 1995;Swanton 1942;Wyckoff and Baugh 1980). Still, with a focus on architecture there remains a wealth of information to be gleaned from the archival and ethnographic material. As archaeologists, anthropologists, historians, and archivists continue to refine and broaden the issues and questions that can be fruitfully addressed by the archaeological and ethnohistoric record, this material continues to provide new insight into the cultural practices of Caddo peoples.
This chapter provides a detailed examination of the archival material relating to architecture and the use of structured space by Caddo groups, and builds upon previous studies that have utilized such material (Baugh 1998;Bolton 1908Bolton , 1987Griffith 1954;Perttula 1997;Perttula et al. 2008;Sabo 1998;Swanton 1942;Wyckoff and Baugh 1980

STRUCTURED SPACES
"The architectural heritage of the Hasinai provides one of the most important bridges through which the generations from the ancestors to the present offer a sketch of a people in balance and peace" (Newkumet and Meredith 1988:40

DAILY USE, COMMUNAL, AND FAMILY SPACES
This category of structures is probably the most visible archaeologically in Caddo sites and includes those spaces that were most likely accessible to the broadest range of people in the community or family as a whole. This category includes those spaces that were most likely the least restricted and conversely used or accessed on a regular or daily 27 basis. Houses, arbors and storage areas (including bins and platforms) are some of the spaces that fall into this category.

Houses
Caddo houses are often described as being round, wooden pole thatch structures. Newkumet and Meredith (1988:35)  As is apparent in the Soule photo and in the archaeological record of their ancestors, Caddo peoples constructed and used a variety of architectural forms. The open aired structure in the forefront of Figure 4 is similar to the one described in the early historic accounts of Caddo buildings. Upon arriving outside of a Cadodaquis (Kadohadacho) village along the Red River in 1687, Joutel and the French were greeted by the chief and several individuals who proceeded to carry the French party into the village upon their backs and shoulders. Following a ceremony in which a village elder washed their faces, the French were given a sign by the chief "to sit down on a kind of platform that was raised about four feet off the ground and made with wood and cane" (Foster 1998:244) Newkumet and Meredith (1988:37) note that "the arbor is a popular structure still constructed near homes." Lasting about two years, modern arbors are usually made of willow, which provides for a lightweight but sturdy structure with all four sides open to the air. They have square or rectangular roofs supported by poles made of peeled oak with the pitch of the roof extending away from the central ridge of the structure (Newkumet and Meredith 1998:36).
The first part of the nineteenth century was a time of great upheaval and change for Caddo peoples but they were able to maintain and continue much of their traditional cultural practices (Carter 1995;Smith 1995:103-104). During this period Anglo-Americans flooded into Texas along with native groups that were immigrating from the east. Despite these dramatic shifts, Caddo traditions continued. For instance, as Smith (1995:104) notes, in addition to speaking French and Spanish, the Caddoan language was still spoken, men wore their hair in a traditional manner, and "the tribes continued to live in their trademark grass houses." William Parker noted that "the Caddo, Hainai, Anadarko, Waco, and Tawakoni live in houses built of a framework of poles, in a conical shape, thatched with long prairie grass, with low doors; the fires built in the center of the lodge; the lodge, circular, about twenty-five feet in diameter and twenty high" (Swanton 1942:152 1942:153) in the 1890s stated that "they formerly lived in conical grass houses like the Wichita, but are now in log houses and generally wear citizen' dress excepting in the dance." One surviving document from the seventeenth century that has captivated the imagination of archaeologists, historians, and others interested in Caddo native history and has been at the forefront of interpretations of settlement, site layout, and architecture, 30 is the map of a Caddo settlement along the Red River from the 1691-1692 expedition led by Domingo Terán de los Rios ( Figure 6). Frank Schambach (1982a:7) and others have noted that the nineteenth century farmstead depicted in Soule's photographs "matches those in the Terán map in most details, down to and including the beehive-shaped storage platforms" and arbors. The Terán map is largely believed to be the earliest depiction of a dispersed Caddo community. The map depicts 25 clusters of buildings, 23 of which probably represent individual farmsteads or hamlets (Schambach 1993;Schambach et al. 1983;Wedel 1978). At the western edge of the map is a compound with a building atop a platform mound and an arbor near its base (Figure 7). At a distance estimated to be approximately 2.5 km east of this mound (Schambach 1993;Wedel 1978), and near the center of the settlement, is the compound labeled 'Caddi' (Figure 8). This compound is 31 the largest clearing within the settlement and is the only one lacking a storage bin (Schambach 1993;Wedel 1978   Late seventeenth and early eighteenth century accounts, such as those from Henri Joutel's journal from 1684-1687, further enhance our understanding of these images while also illustrating the diversity in architectural practices documented in the Soule photographs and in the Terán map. In his description of a recently deceased Cenis (or Hasinai) chief's hut in which lived the Provençal that had deserted La Salle during his previous journey to East Texas (Foster 1998:206), Joutel (Foster 1998:208) notes: There are normally eight or ten families in these huts, which are very large; some are 60 feet in diameter. The huts are made in a different method from those we had seen earlier. These are round, in the shape of beehives, or rather like large haystacks, being composed of the same except they are higher. They are covered with grass from the ground to the top. They make a fire in the center, the smoke going out through the top through the grass.
Joutel (Foster 1998:208) continues by providing a description of the interior of the chief's house, "once we were in the hut, which was one of the largest in the area, we were shown a place to put our packs and to sleep. These huts were much more comfortable than those we had seen before. The Indians raise their beds three feet high. They fashion them neatly with long reeds, making each bed separate with matting that forms a cradle." A May or June 1686 account by Anastasius Douay provides similar information.
Douay was among 10 Frenchmen that accompanied La Salle in his first expedition to the Hasinai. Douay describes the houses of the Cenis or Hasinai as being "fine, forty or fifty feet high, of the shape of bee-hives. Trees are planted in the ground and united above by the branches, which are covered with grass. The beds are ranged around the cabin, three or four feet from the ground; the fire is in the middle, each cabin holding two families" (Cox 1904:232; see also Swanton 1942:39, 148).
Fray Damián Massanet's 1690 account of a Nabedache chief's house matches closely the description provided by Joutel. Swanton (1942:149) suggests Massanet's description is of the same chief's house Joutel described in 1687, albeit with some differences in detail. Massanet's account states: The house is built of stakes thatched over with grass, it is about twenty varas [ca. 55.6 feet] high, is round, and has no windows, daylight entering through the door only; this door is like a room-door such as we have here. In the middle of the house is the fire, which is never extinguished by day or by night, and over the door on the inner side there is a little mound of pebbles very prettily arranged. Ranged around one half of the house, inside, are ten beds, which consist of a rug made of reeds, laid on four forked sticks. Over the rug they spread buffalo skins, on which they sleep. At the head and foot of the bed is attached another carpet forming a sort of arch, which, lined with a very brilliantly colored piece of reed matting, makes what bears some resemblance to a very pretty alcove. In the other half of the house where there are no beds, there are some shelves about two varas [5.56 feet] high, and on them are ranged large round baskets made of reeds (in which they keep their corn, nuts, acorns, beans, etc), a row of very large earthen pots like out earthen jars, these pots being used only to make the atole when there is a large crowd on the occasion of some ceremony, and six wooden mortars for pounding the corn in rainy weather, (for, when it is fair, they pound it in the courtyard). [Massanet cited in Swanton 1942:149; see also Bolton 1916Bolton :377-378, 1987 Foster (1998:237n) adds that Massanet also provided a description of a food storage area inside the house stating that "over the door on the inside there was a structure of rafters, very prettily arranged to hold jars or sacks of food items." Joutel's 1687 account of the house of a Nasoni chief along the Red River is similar to the one he provides for the Cenis. In describing the Nasoni structure, however, he offers a more detailed description of the interior: These Indians' huts are made like those of the Cenis except they are not as high.
There is a large platform above the door that is made with pieces of wood placed standing up with other pieces across, and canes are arranged and pressed close together upon which they place their ears of corn. There is another platform opposite upon which they place their barrels or casks (which are made with canes and bark) in which they put their shelled corn, beans, nuts, acorns, and other things. Under this they put their pottery. Each family has their own personal barrels; and they have their beds to the right and the left in the manner that I described earlier. These Indians also have another large platform in front of their huts that is elevated about 10 or 12 feet upon which they place their ears of corn to dry after gathering them. This they are careful to sweep every day. [Foster 1998:237] The accounts from others who visited Caddo groups in eastern Texas match closely the more detailed descriptions given above (e.g., Forrestal 1931;Padilla 1919;Swanton 1942:148-154). These accounts serve not only to confirm the observations of chroniclers such as Joutel, Massanet, and others, but also provide further details about structures, including furnishings, variety in structure size, food storage areas, and houses of other groups. Hidalgo (Hatcher 1927:56), for instance, stated that "their houses are made of grass, some of them quite large and tall. Others are medium sized and others still smaller like half an orange. In each of these many families live. They keep their corn in lofts and garrets and in big reed baskets." Hidalgo (Hatcher 1927:56) also noted that "the Indians have the doors of all their houses toward the east," which, he continued, "I heard them tell the soldiers…they did this because it never blows from that side, [but] I do not understand the mystery." From his 1767 journey through East Texas, the Marqués de Rubí describes the interior of a Tejas (Cenis) house as being "divided inside with partitions made of woven reeds, which they also use to make the bedrooms, furnished with beds that are elevated, decent, and comfortable" (Jackson and Foster 1995). Casañas, in discussing daily activities such as dining, stated that "they eat while seated on benches of wood, all one piece and not very high from the ground" (Hatcher 1927:212). Casañas also provided a description of specialized furnishings found in the houses of the caddices and other nobles. He noted that "there is a certain bench which nobody is allowed to approach except the gran xinesi himself when on a visit. In all these houses there is also a high bed like an alcove where the xinesi may sleep and rest" (Hatcher 1927:217). These and other reserved spaces will be discussed in more detail below.
Espinosa provided further details of the interior of structures among the Hasinai: in their houses they have large baskets make of heavy reeds into which they put their shelled corns and beans. In order that the weevil may not get in they cover the grain with a thick layer of ashes and then cover the baskets to keep out the rats. These Indians are so provident they make a string of the best ears of grain, leaving the shucks on , and put it up on a forked stick at a point in the house where the smoke will reach it (Hatcher 1927:156-57).
Finally, Casañas offered descriptions of houses from provinces to the north of the Hasinai, in Caddo communities along the Red River. In his reference to the houses of the Cadaudachos, the Nasitox, and others, Casañas (Hatcher 1927) noted that they "are located closer together and are well arranged and plastered." This practice among Red River Caddo groups was also seen in 1689 by Henrí de Tonti (Cox 1904), who noted that their houses are made of straw and covered with plaster. Swanton (1942:153) briefly touched on this in his discussion of architecture depicted in the Terán map, concluding that "most of the houses seem to be grass houses of the conventional type or granaries, but a few, particularly one on a mound which is presumably a temple, have what look like wattle walls." Perhaps some of the most intriguing accounts relating to Caddoan architecture concern the construction of architectural space. Espinosa offered a detailed description of the processes involved in constructing a new structure from the planning stages through completion. In this account, Espinosa's (Hatcher 1927:154-155) description of the structure itself is fairly consistent with descriptions given elsewhere: Their houses are built of wood with very long, flexible laths. Their manner of building them is as follows. Whenever the owners of a house decide to build one, they advise the captains whom, in their language, they call caddi. The latter set the day and order the overseers whom they call tammas to go around to all the houses and give notice in order that all may aid in the building. These two messengers mount their horses -of which the Texas Indians have a great number since the first entry of the Spaniards. They carry in their hands a number of little sticks so that he who receives it may take care to cut and clean a lath and bring it and put it in the hole designated from it. Another member of the household is placed in charge of a sufficient number of men to continue the work of lacing the laths together. These thongs, made of the bark of a tree, are so strong that they can not be broken between the hands however thin they may be. To the Indian women, one or two from each house, is given the duty of bringing a load of grass. This grass is coarser than the largest wheat and is used to cover the whole roof. These arrangements being made, the tammas go and sleep at the place the building is to be done. When day breaks, they call the people designated together.
At dawn, the captains arrive and take their places without putting their hands to the work other than to oversee it. At sunrise, upon the first call of the messenger, each comes running with his lath on his shoulder and puts it in the hole which he has previously dug. The laths are placed in a circle and in the middle they put a very tall pole with knots on it for climbing. Two Indians are placed on top on a cross made of two pieces of wood. Each throws out a noose and seizes a lath by the top. Working in unison they continue to tie them until they have formed a figure like a half orange.
They then cover the laths with heavy timers, all working at the same time and with such dexterity that, each working upward upon his own lath, they do not take more than an hour to finish it from bottom to top. Others came in to relieve them and cover the house with grass to a thickness of three hand breadths. They work from the bottom exactly opposite to the way the Spaniards thatch their houses. They work so dexterously that a little after midday they are finishing the hut, forming of carefully tied grass, the figure which their imagination suggests to them. The building finished, they cut the middle post off at the bottom and the building is thus left standing. During all this time, the overseer walks around with his rods made of two or three fresh, flexible branches for the purpose of hurrying the people. Even though they bring the materials they have been instructed to provide, he goes out to meet the man or woman who is late and who arrives after the work is begun. If the delinquent is a man, the overseer gives him four or five licks across the breast and, if it be a woman, he uncovers her shoulders and does the same thing. This is done without exception of persons, for even though it be his own wife or sister who is at fault, she receives her punishment. No one is offended at this but rather laughs at it. During all the time the people are working the householders are busy preparing food for everybody, having preciously provided quantities of deer meat and many pots of ground corn, which in this section of the Indies is called atole. Then they serve the food from the captains down to the smallest, in order, abundantly, and carefully, because they have earthen vessels, some large and some small, in which to serve the old and the young. This done, the crowd scatters and each goes to his own home much pleased. The difference they make in building these houses is they use more laths than usual for the captains and leading men. Consequently, their houses are very much larger. But no one, even though he be a leading captain, is excused from feeding all those who assemble. In fact, the feast is all the more abundant and more time is used in preparation so that everybody may be abundantly fed.
The significant role played by one figure not mentioned in the above quote, but mentioned elsewhere, is that of the priest or shaman. The priest or shaman was responsible for laying out the dimensions of each structure and for blessing the new house upon its completion (Griffith 1954:99;Sabo 1998:168). These actions suggest the sacredness and symbolic importance of house structures. As Sabo (1998:169) has argued, the shaman's role "indicates that the house was recognized as a (potentially) sacred place where qualities associated with the numinous realm could become manifested." A similar, but less detailed account of construction techniques has been provided by Joutel (Foster 1998:208) who stated that "they cut full-length trees as thick around as a thigh and plant them in the ground upright in a circle, joining them at the top. Then they lath (lattent) the huts and thatch them from the ground to the top." The account of house construction from Casañas is even less detailed, its significance lying in what he chose to note. Casañas (Hatcher 1927:217) emphasized the communal nature of house construction, stating, "as regards other features of their government, these Indians help each other in such a manner that if one's house and all his possessions are burned up, they all gather together, build him a new house, and furnish him whatever he needs for his subsistence and comfort. All these things they do together." This communal activity is further emphasized during descriptions of the planting season, when "they come together and plant whatever each one has to plant, according to the size of the family -beginning first at the home of the gran xinesi. There they plant only a small spot in front of the house in order that he may have something green to enjoy" (Hatcher 1927:217).
Following the planting of the plot for the xinesi, the group continued to the plots of the caddices, the other officials, and the viejos, and "in this way they continue working from the highest to the humblest" (Hatcher 1927:217).
As these accounts describe, all members of the dispersed community were called together to take active roles in the building of houses, thus signifying and reinforcing a notion of community through the creation of architectural space. Sabo (1998) has 39 examined the symbolic elements inherent in the accounts of house construction. He has argued that the communal nature of house construction emphasizes the notion that houses, and likewise households, were not viewed as independent entities, but were seen as integral components of the broader community and "stood as visible symbols of the interconnectedness of families and households comprising villages and communities" (Sabo 1998:168). In addition, he argues for the possibility that the grass figure crowning the completed house could be a clan symbol, "providing an additional representation of larger social fields within which the house and its occupants were identified" (Sabo 1998:168).

SPACES RESERVED FOR SPECIFIC ACTIVITIES AND/OR PEOPLE
Several chroniclers noted the presence of assemblies and council meetings among the Caddo groups they encountered. As Casañas noted, "if the Caddi wants to do anything, he calls the old men together, listens to each of their views, and then decides to do what he thinks best, explaining his views to some of the old men and urging agreement" (Hatcher 1927:218). In these closed council meetings there was a clear deference given to age in speaking, sitting, "and in all other courtesies that Christians are accustomed to observe" (Hatcher 1927:218).
Massanet's 1690 account states: Soon I noticed, outside the yard, opposite the door of the governor's house, another long building, in which no inmates could be seen. I asked who dwelt therein or what purpose it served, and was told that the captains were lodged in that house when the governor called them to a meeting. On the other side I saw yet another and smaller vacant house, and upon my inquiring about this one they answered that in the smaller house the pages of the captains were lodged, for theirs is a law providing that each captain shall bring his page when the governor assembles the captains, and they observe this custom. As soon as they arrive, they are lodged in that house, and for each one is laid a large, brightly colored reed mat, on which they sleep, with a bolster made of painted reeds at the head; and when they return home, each one carries with him his mat and pillow. While they attend the meeting the governor provides them with food, until he sends them home (Swanton 1942:149).
Three years before Massanet entered the Hasinai village, Joutel recorded the presence and use of assembly houses. From March 31, 1687, Joutel (Foster 1998:207) commented that "after we had smoked they led us to the chief's hut from which we proceeded quickly to where they were all in a large hut that was about a quarter of a league from there. Here they conducted their festivities and made their preparations for war. Upon our arrival, we found mats spread on the ground on which they made a sign for us to sit, and the elders also sat down themselves around us." In this meeting, a large meal was shared, tobacco was smoked, and the officials discussed their plans to go to war with the Canohatinno (Joutel 1998:207).
The following day, the French were again escorted to the assembly house where the elders were gathered. Finding the Hasinai officials seated upon reed mats, the day began with discussion and a smoke, followed by a meal and trade. The day was spent in assembly with the French securing provisions and a horse. Joutel learned from the Provençal that "the Indians had not entered their huts at all since they had built the assembly hut and that the women took care of bringing them food and the young people served them" (Foster 1998:209). Following up on this, Joutel (Foster 1998:213) noted that "these men had a large assembly hut where they prepared for war with feasts and rejoicing, and they did not return to their customary huts at all." The Hasinai with whom Joutel met, likely the canahas and elders, were officials from throughout the dispersed Hasinai community. As indicated above, these officials "live temporarily in a special house (elongated rather than circular?) when called to the Caddi's compound" (Wyckoff and Baugh 1980:243), with assemblies being held in a separate structure.
Casañas and others described various spaces that were reserved for use by specific individuals and for specific purposes. Elevated seats called tapestles were reserved for the gran xinesi and the caddices. When seated upon these seats, the officials placed their feet on a high bench, thus being fully elevated. Proclamations were made from this elevated postion and, as Casañas stated, "whatever this official says or does is carefully heeded, just as the Catholics obey the Holy Gospels. If he issues a command it is more strictly obeyed by these Indians than the Ten Commandments are observed by the Christians. Therefore, the leaders do not take their seats on this elevation except for a special ceremony" (Hatcher 1927:213).
Interestingly, elevated seats were also used during the First Fruits ceremony.
Joutel and Espinosa provided complementary accounts of this ceremony. Joutel stated that "as the corn was beginning to ripen, I watched a ceremony being conducted by one of the elders who had come to the hut. After the arrival, the women went to gather a great many ears of corn. They prepared it by parching and put it in a small basket which they carried on a ceremonial stool which is used only for that purpose and upon which no one sits. One day I wanted to sit on it and the good old woman told me that I must get up or I would die" (Joutel 1998:240).

RELIGIOUS, CEREMONIAL, AND SYMBOLIC STRUCTURES/SPACES
The seventeenth and eighteenth century accounts of the Caddo by French and Spanish chroniclers can be used to identify trends and/or patterns in the organization and use of space and offer rich details pertaining to those spaces within buildings and types of buildings themselves that are often described as "specialized" or special purpose in character (Rogers 1982a(Rogers , 1982bStory 1990). These religious or symbolic spaces were apparently prominent features in Caddo society. This section examines specific architectural and structural spaces that fall into a religious or symbolic category and what roles such spaces may have played.
Casañas provided a glimpse at otherworldly associations through the description of the supernatural or mythical foundation for a generalized architectural form. He noted that "another gross superstition they have, in which all of them believe implicitly, is that the old men made Heaven and that a woman, who sprang from an acorn, first gave them its outlines; and that it was done by placing timbers in the form of a circle and that Heaven was formed this way" (Hatcher 1927:296).
This belief in the otherworldly equivalent to earthly structures was reiterated in Casañas' discussions of death: "when a person dies, his soul -of which they are not ignorant and which in their idiom they call Cayo -goes to another house where a man guards all who are there until all are gathered together . . . when all the souls are gathered together they will enter another world to live anew" (Hatcher 1927:294). Death was accompanied by elaborate ceremonies including the shooting of arrows into the sky "to inform the master of the said house -who receives everybody" (Hatcher 1927:294). The shooting of arrows was accompanied by statements such as: "Here he comes! Make him work until we are all united" (Hatcher 1927:294).
One suggestion from Casañas for where the Caddo dead were buried was in reference to the death of a leading official, in that "the Indians go to the place of interment which is always near the house" (Hatcher 1927:297). During the ceremonies for the burial of these officials, the deceased was told "that everybody loves him dearly; that everybody is weeping for him; that he must go in peace; that he must work in that other house with the others who have gone before, until those who have begun to work shall have assembled; that he must take up his hatchet and all the rest of the things that are wrapped with him" (Hatcher 1927:298).
Casañas (Hatcher 1927:296) also noted that: If someone dies or a house burns up they say that death is angry. Therefore, they make an offering of something by hanging it on a pole in front of the house. When a house burns, they also say that the ground on which they lived, or the hill near the house, has been angered and burned the house; so they do not rebuild the home there but in another spot.
Following the death of a Xinesi all nine Hasinai tribes performed the requisite ceremonies including "placing the world in front of this door. This is done by setting up a very high pole with a large globe of grass on top. They indicate the moons by putting up some large sticks in the shape of the moon. Before these they dance ten days and nights and then they each go home" (Hatcher 1927:298).
Spaces used by religious and ritual practitioners are fundamental to the relationship and interaction between those of this world and the supernatural. Daniel Rogers (1982b:105) has argued that the "concept of cultural space relates to a variety of contexts, not the least of which is the interplay between the world of humans and the realm of the supernatural. In this context cultural space is physically represented by the buildings, grounds, and other enclosures that religious practitioners use." One example of architecture used by a religious practitioner is the temple of the xinesi or fire temple. This temple was the most sacrosanct of spaces, within which the xinesi maintained the sacred 44 and perpetual fire and transcended the gap between this world and the otherworld (Sabo 1998).
Casañas provided a detailed discussion of the practices of the xinesi and some of the religious beliefs of the Hasinai. From his descriptions, the idea of shared traditions becomes apparent, in that "all the tribes of the region have the same errors, superstitions, and ceremonies" (Hatcher 1927:290). At the core of these traditions were the beliefs associated with the two children, the caninisí, who had died in a fire (Hatcher 1927:293) and hence "live" near the house of the xinesi and are "from the other side of heaven" (Hatcher 1927:290). The caninisí act as intermediaries between the xinesi and Ayo-aymay, the "great captain" or supreme being. Through the caninisí, the xinesi offered prophecies and, made pleas to Ayo-ay-may on behalf of the Hasinai.
His communications with the caninisí began with the calling of all the groups to the temple where he "gives orders to all the caddices and the old men to come into the house where he keeps the two children. This house is very much larger than the one where he lives" (Hatcher 1927:290). Inside this house is the sacred fire, which is kept burning all day and night. The accouterments associated with his communication with the caninisí included incense, two small reed boxes (representing the children), offerings brought by the officials (some placed directly into the boxes), a rattle, and a mortar (Hatcher 1927:291). Following the ceremony, and after the officials had left, the xinesi "comes out and goes home, about a hundred paces away" (Hatcher 1927:292).
Espinosa provided further details regarding the temple of the xinesi and the related structures of the caninisí. The caninisí (cononicis in Espinosa), as described by Espinosa "are two boys or small children whom their great captain sent from the cachao ayo, or the sky, for the purpose of discussing their problems with them" (Hatcher 1927:160). The houses of the caninisí were two smaller structures and were located "about a gunshot distant" (Hatcher 1927:160) from the fire temple. The children were said to have lived in the houses until a little more than two years prior to Espinosa's visit.
"He understood through an interpreter that the caninisí were there until the Hasinai's enemies the Yojaunes (a Tonkawa band) had burned the houses. That was when the children were seen rising skyward with the smoke, and they never came down again" (Carter 1995:120-121).
Espinosa's account differed from that of Casañas in that the caninisí lived in two separate "small houses about a gunshot distant" from the temple (Hatcher 1927:160-61) rather than in the temple itself as suggested in Casañas. This separation, if an accurate depiction by Espinosa, apparently provided for additional sacred and symbolic architectural spaces associated with these supernatural figures.
Espinosa's (Hatcher 1927:160) account offered a detailed description of the temple, or "house of the great captain": They have especial superstitions in connection with fire and they worship it. There is a house set apart for this purpose where there is always a fire. They have appointed an old man whose duty it is to keep it up always. He is their chenesí or chief priest. They say that if it goes out everybody will die. This house which was rebuilt in December, 1716, is half way between the Naichas and the Ainas and is common to both people. They say it is the house of the great captain. It is large, round, and thatched, and has within it an altar made of reed mats. On the bed are three finer mats, two of them very small. To one side of the door, upon benches are other reed mats folded into a roll. In the front of the bed is a little square wooden bench, of one piece, with four feet, and slightly raised from the ground. Upon this bench there is usually tobacco and a pipe with feathers and earthenware vessels which are evidently incense burners in which they burn fat and tobacco. Their fire or bonfire is always made of four very large, heavy logs which point toward the four principal directions. The wood is brought in small and kept in a pile outside. Here the old men gather for their consultations and war dances and when they need rain for their crops. Ordinarily their prayers are vain and mere fables. The ashes from their fire continue to accumulate outside and when they bring any bones of the enemy whom they have killed, they bury them in these ashes. Near this house there are two other small houses about a gunshot distant. They call them the houses of the two cononicis.
The importance of the temple as a sacred place was further emphasized by Espinosa when he described the role of the temple as similar to that of the parish house or cathedral and referred to it as the "house of the great captain," the house of Ayo-Ay-May.
Further emphasizing the significance of this space was the role it played in housing the sacred fire from which all home fires were lit. The function played by the fire temple in providing the fire for each house further illustrates the notion of community as embodied in the cultural landscape and sacred space (Sabo 1998).
Returning to ideological and ceremonial practices, "in the last decade of the twentieth century those dances [the Turkey Dance, the War Dance, and the Drum Dance or Victory Dance], like the Caddos who perform them, descend from those watched by Joutel during the days of victory celebration in the seventeenth century" (Carter 1995:49). Dancing and singing are fundamental Caddo traditions. Dances and songs such as the cah-kit-em'-bin, the drum dance, which provides the origin story of the Hasinai, or the nuh'-ka-oa-shun, the turkey dance, which relates "the stream of events that define the history of the Hasinai people" (Newkumet and Meredith 1988:3, 102), are integral components of Caddo culture. Newkumet and Meredith (1988:3-4) state that "at the heart of Hasinai existence are the cultural traditions that carry the people through space and time. In the movement of dance and the language of song, the reality of existence is projected into the future." While most traditional architectural forms are no longer in use by the Hasinai (see Newkumet and Meredith 1988:36-37), the dance ground, ko-na-chaka-wa-ah-so, remains "the central focus of Hasinai culture and learning, as they have been for centuries." These fundamental components of Caddo culture have their roots in the past, and there are suggestions and examples of the use of such spaces in the historic and ethnohistoric record.
As for today: The Whitebead Dance Ground is a level plaza in northern Caddo county. Around the central plaza are the wooded camp sites used by the Hasinai throughout the period of the dance cycle. The most frequently used dance ground is on tribal land immediately east of the Caddo Tribal Offices near Binger. It is a plaza surrounded by arbors, public buildings, and small family cabins. Both dance grounds are level areas that have been constructed along hill sides. An unusual dance space has been enclosed as a part of the Caddo office building. It has an earthen floor, but is roofed over and has masonry walls surrounding it (Newkumet and Meredith 1988:36).
Nineteenth and twentieth century accounts and photographs document the use and importance of dance grounds for the Caddo (e.g., Parsons 1941). Ethnohistoric accounts of festivals and ceremonies from Joutel, Casañas, Espinosa, and Morfi provide evidence for dance grounds and ritual structures among Caddo groups from centuries before.
While a detailed examination of the use of space for each of the ceremonies of today and from the seventeenth, eighteenth, and nineteenth centuries is certainly warranted, it is beyond the scope of this work.
What is of significance to this study is that there is a clear indication of ritually reserved spaces such as dance grounds in the early historical accounts. At least one example indicates an open enclosure used during the September Festival. For instance, as part of the Festival, when the Pleiades were above the temple, the chief priest and a companion sat within a circle made of green reeds surrounding a large fire. Espinosa (Hatcher 1927:171-174) described the ceremony in detail, stating: From midnight on one of the Indians is stationed as a watchman or sentinel. He watches to see when the Pleiades are perpendicular -from the house. They call these stars las senates, i.e., "the women," because the devil has made them believe that these stars are people. He then informs the chief conjurer who goes in company with another conjurer to a circle made of green canes stuck in the ground where there is a big bonfire which three or four novices feed continually. The two men seated on an elevation serve as masters of ceremony. The Indians are formed, to their left, as follows, the old women in the first row, or file, behind them the married women and the young girls, and at the end the younger girls. The little girls are in front of this file. To the right there is an arbor with a bonfire under it. Three old men, dressed in the best they have, consisting of curious buffalo robes, go to this fire, each following in the footsteps of the one in front, while the women and children in the ranks begin singing. After a considerable pause, the old men again approach the circle dancing as they come…they then take hollow logs, covered on top with green branches, but the ends of them, and select eight strong Indian women, who, seated at intervals with sticks in each hand, use each the hollow log as a drum, to the accompaniment of the calabash which the old men play, and the songs of the men and women singers to the number of more than twenty. This music is for the dance in which they all engage, old women and girls, old men and boys, and little children. They dance in a circle, the men facing the women, keep time, moving only their feet.
Espinosa's account provided details concerning the use of open enclosures (the reed circle), the presence and use of arbors, and the details of the dance, held within a prepared dance ground area (Carter 1995:118-119).
Casañas described a ceremony lasting seven or eight days during which a large pole was erected in front of dancers. A fire was kept burning in front of the pole, and from the pole was hung "a portion of everything they are offering to God" (Hatcher 1927:214). In addition to this description, he provided the positions of individuals during these ceremonies, stating "nearby is a person who looks like a demon." He continued by writing that "every time a dance begins, a man steps forward as a preacher does and tells the people what they are to ask God for in the next dance," and finally, "on the last day of 49 such a meeting the caddi comes forward and encourages the men" (Hatcher 1927:214).
At these ceremonies, Casañas remarked that "we can garner in a great harvest [of souls/converts] because many tribes are gathered together in these meetings" (Hatcher 1927:214-215).
Casañas also provided an account of ceremonies honoring the caddices and the xinesi, "in celebration of the victories their ancestors have had" (Hatcher 1927:301).
Several of the tribes gathered at these ceremonies, which lasted for three days and nights, and at which there was: dancing, eating, and fun; all those who have come feast but the Xinesi. He goes without eating from three days and without sleeping for three nights. They do not let him sleep or eat. He does not even drink or rest, but he is continually stirring about from place to place as if making the sound of dancing. At these meetings and feasts certain superstitions are usually noted. The crowd that gather is very large for men, women, and children come to the feast because they are given plenty to eat (Hatcher 1927:301-302).
Casañas concluded this description by noting the preference of the caddi or xinesi to have someone "at their side whose importance they recognize, so that all the other Indians may see and know the estimation in which they, themselves, are held…to be at the side of the caddi would give him to understand that the minister is less than he and that the minister owes obedience to him" (Hatcher 1927:302).

SUMMARY
The ethnohistoric and archival records concerning the Caddo peoples that lived in the Caddo area are invaluable for examining architecture and the use of space during the early years of Caddo and European contact and interaction. These sources provide 50 information on domestic, sacred, and otherworldly spaces as well as provide insights into the cultural significance of such spaces. Our understanding of Caddo cultural practice will continue to be enhanced as we further investigate the concepts and agency behind the creation, manipulation, use, and destruction of such spaces. When combined with the rich archaeological record from the Caddo area, these sources can aid in approaching an understanding of the role that architecture and architectural practice played in the broader Caddo society. The following chapters provide details on both mound and non-mound Caddo structures from throughout the study area. 1200 (Perttula 1997;Story 1997Story , 1998Story , 2000. The site appears to have been occupied from the ninth century into the early 14 th century A.D. (Story 1997(Story , 1998(Story , 2000.  Story in 1968Story in , 1970Story in , 1977Story in , and 1978 focused on Mounds B and C, village areas, and the borrow pit. The 1978 Texas A&M University excavations (Creel 1979), and those conducted by Prewitt and Associates, Inc. in 1980 (Fields and Thurmond 1980), investigated portions of the site to the north of Weeping Mary Road. Figure   Nursery property . These data provide information on such things as structure size, shape, and location, but lack the chronological information necessary to delineate the age of individual structures or determine which structures predated which.
Nevertheless, this glimpse at structure form, intrasite distribution, and spatial patterning provides a wealth of information on intrasite spatial organization of structures and related features.
This chapter first provides a discussion of the previously excavated structures from the Davis site. This is followed by a discussion on the structures identified through magnetometer surveys at the site by archaeologists from UT-Austin, TARL, and the THC. Finally, the chapter includes a discussion of the results of the 2003 excavations at the site that were guided in large part by the findings from the magnetometer data.

PREVIOUSLY EXCAVATED STRUCTURES FROM THE GEORGE C. DAVIS SITE
Prior to 2003, 52 structures had been at least partially excavated at the site. Fortythree of these were investigated as part of the WPA work in the Mound A area and nine were identified from work elsewhere at the site. Varying in size, shape, and presumed function, these structures include those associated with two of the three earthen mounds as well as those located within the "village area" ( The pattern which can be seen at Davis shows a distribution -across the site and through time -of domiciles, small storage and/or special function structures (probably representing granaries or temporary housing for pages or dancers), larger community/special function structures (providing the meeting halls, religious centers, specialized dwellings, etc), and outside activity areas (for cooking, preparing hides or pottery, for drying food, etc). Some subsurface storage was suggested through the presence of pits.
Spock's analysis places 30 structures in the domicile category. Story (1998:40, fn. 7) notes that the two structures excavated after 1977 fit into Spock's domicile category.  (Spock 1977:167). Center posts were identified in 10 of these structures but hearths were found in only two. At least one of the center posts was associated with a central hearth.
Some of these structures were only partially exposed with between six to 65 exterior wall post holes. The exterior post holes range from 9 cm to 51 cm in diameter and from 6 cm to 73 cm deep. The mean diameter of the exterior posts for the 32 domiciles ranges from 14 cm (Feature 8) to 26 cm (Feature 51). Interior posts posed a problem in Spock's analysis given the intense rebuilding in parts of the site ( Figure 11) (Spock 1977). Spock was unable to definitely assign interior posts to 14 structures (these are marked as "?" on tables 4 and 5), although she was able to confidently assign 112 interior posts to 12 structures.
According to Spock's analysis, 20 1 Table 5 structures are classified as "special function structures" (Spock 1977:169). These structures ( ) are those "whose placement and architectural design make them unusual," they "are often exceptionally large or quite small structures, are of unusual shape or have distinctive interior features" (Spock 1977:169), and all but seven are directly associated with the mounds at the site. Spock

62
The 20 structures Spock classifies as special function structures are round (n=10), sub-round (n=3), oval (n=1), rounded oval (n=1), sub-square (n=1), square with diagonal corners (n=2), square with rounded corners (n=2), and the feature referred to as the maze, Feature 35 ( Figure 12). This structure is generally circular with protrusions on the east and west side of the structure.
The round (circular) structures range in size from 1.5 m to 18 m in diameter with a mean diameter of 8.73 m. These structures have interior areas ranging from 1.8 m 2 to 255 m 2 with a mean interior area of 85.45 m 2 . Table 5 provides the measurements and structure data for the remainder of the special function structures from the George C.
Davis site.
Only three of the previously excavated special function structures at the George  Some of these structures were only partially exposed, the exposed portions having  (Spock 1977). Spock was able to confidently assign 268 interior posts to 17 structures (see Table 4). The interior posts range from 4 cm to 70 cm in diameter and from 9 cm to 131 cm deep. The mean diameter of the interior posts from each of the 20 specialized structures ranges from 9 to 47 cm and the mean depth of these ranges from 27 to 65 cm. Center posts range in diameter from 15 to 46 cm and in depth from 12 to 84 cm. The 14 hearths range from 61 cm to 183 cm in diameter, with the largest hearths measuring 244 x 296 cm (Feature 34). Hearth depths range from 3 to 45 cm. Six of the hearths were clay-lined.
Two (Features 9 and 35), and possibly three (Feature 6), of the specialized structures have extended entrances (Spock 1977:40). The extended entrances for Features 9 and 35 measure approximately 2 m and 1.5 m long, respectively. The Feature 9 ( Figure   13) entrance was "outlined by two trenches 0.6 to 0.7 m apart. One measured 1.75 m long, the other 1.9 m long; trench depths were ca. 0.3 m," and the entrance was "paved with a layer of grayish-white, hard packed clay which measured 2.14 m long, 0.46 to 0.61 m wide and was raised 0.31 m above the structure's rim. A second layer, 9 cm thick, and consisting of compact reddish clay, overlay most of the grayish-white clay" (Spock 1977:44-47). Feature 9 was built in a 0.37 m deep pit and the floor was a 2-3 cm thick prepared clay lining that "extended across the structure and lipped up to the post trench," and extended into and lined the central hearth (Spock 1977: 44-47 (Table 5), and at least two additional features (43 and 111) have prepared floors. In addition, at least one structure (Feature 45) ( Figure 14) has a partition or exterior wall along the exterior of the structure that likely functioned as a controlled entryway (Spock 1977:100).
There is clearly a great deal of variation in the previously excavated domestic and specialized structures at the George C. Davis site. Some of this architectural diversity can also be observed in the magnetometer data. The remainder of this chapter discusses the magnetometer survey results from the site as well as the 2003 excavations that were guided by that research.

MAGNETOMETER SURVEY METHODS 2
Although ground-penetrating radar (GPR) and conductivity have both been used at the George C. Davis site, the majority of the data, and indeed the most impressive results, were obtained through magnetometry. This research was based largely on the (THC) (Bruseth and Pierson 2004;Creel et al. 2005). By the end of the summer 2003 season, a total of 173,702 m 2 (17.37 hectares) had been surveyed ( Figure 15) (Creel et al. 2005;. Following that widespread survey, the THC returned to the site in 2008 to explore part of the site at the Indian Mound Nursery, which was owned by 67 the Texas Forest Service; the property now belongs to the THC. The THC conducted a magnetometer survey of a 100 x 100 m grid on that part of the site .  THC survey resulted in the recording of at least 10 additional Caddo structures (Table 6), and at least six partial arcs that may represent additional structures. The following discussion provides the details of the UT-TARL magnetometer surveys at the site.  The UT-TARL data were collected using a dual-sensor Geometrics G858 portable cesium sensor magnetometer pulled along on a non-magnetic cart. Data collection speeds using the cart averaged 1.5 ha/day but collected as much as 2.5 ha in a single day. In order to record and later correct for the diurnal variations of the Earth's magnetic field, a stationary G-856 proton magnetometer was used to collect data at 10 second intervals.
The collection grid was tied to the existing site grid using the permanent grid markers north of Mound B and north of Weeping Mary Road. Whenever possible, data were collected in 30 x 100 m collection blocks using reference stakes set out using either a total data station (TDS) or a transit. The collection blocks were oriented north-south with 1 m interval grid lines and the data were collected in a zig-zag pattern with marks spaced at 10 m intervals. Data were collected at 0.1 second intervals with two sensors spaced 50 cm apart, 25 cm on either side of the grid line. Following this general collection strategy, approximately 2.2 million magnetometer readings were collected (Creel et al. 2005;Walker 2009).
MagMap2000 was used for downloading the data, correcting collection errors, such as missed marks, and for applying the base station correction. The data were 69 despiked in order to remove zero readings and unusually high or unusually low readings.
Following despiking, the data were imported into ArcGIS and gridded using a Kriging algorithm. Generally, grids were produced using a linear model and a fixed radius of 1 m, with a few exceptions using a fixed radius of 1.5 m. The sensor readings assigned during Kriging were then classified from low to high. This classification generally consisted of 12-15 classes. Each class was color coded to allow for easy interpretation of the divergence of mapped anomalies from the mean magnetometer reading (Creel et al. 2005). Each collection area was zero-based to a standard reading to facilitate smoothly lacing the grids together into one map (for further details regarding data processing see Creel et al. [2005]). Throughout this dissertation, the magnetometer results are displayed using a stretched gray-scale color ramp. Unless otherwise noted, stronger magnetic values are lighter in color. Most images were produced using ArcGIS.
The interpretation of the magnetometer data has been conservative as well as a collaborative effort among the researchers involved in the project (Creel et al. 2005(Creel et al. , 2008Walker 2009). When a magnetic pattern is identified that appears to correspond to either a complete or a partially recognized structure, that pattern is recorded as a feature and assigned a number. If potential features are identified within or otherwise associated with that feature, those patterns are assigned sub-feature numbers. Feature numbers have been assigned only to those anomalies most confidently recognizable as archaeological features (Creel et al. 2005). I will focus primarily on clearly defined architectural features. There are several anomalies that may represent archaeological features (Creel et al. 2005;Creel et al. 2008;Walker 2009) that will not be discussed in detail in this dissertation. It is also quite likely that additional features exist in the dataset that have not yet been identified. For a more detailed discussion of the magnetometer work itself at the George C. Davis site, as well as a use of these data to explore spatial patterning at the site, mound construction, and other topics, the reader is referred to Creel et al. (2005), Creel et al. (2008), and Walker (2009).
In 2003, excavations at the George C. Davis site were conducted in conjunction with, and were guided by, the results of the magnetometer survey. The excavations were designed to address two specific goals. One general objective was to provide the material record necessary for assessing and refining interpretations of the magnetometer data, specifically data from certain geophysical structural anomalies. The second goal of the 2003 excavations was to target specific architectural features with small excavation units.
The primary purpose of this approach was to retrieve dateable materials from specific features while minimizing direct impacts to other features. With these objectives in mind, specific features were targeted that were interpreted as representing a specific architectural form or type (see below).
The primary reference points for all excavations were the permanent grid markers north of Mound B. Using these reference points, grid coordinates were acquired from the magnetometer data for each of the features slated to be investigated in 2003. Excavation units were placed using these grid coordinates and set using a TDS. Standard archaeological recording procedures employed previously at the George C. Davis site (e.g., Story 1981Story , 1997 were followed during the course of the investigations. Excavated plow zone deposits from the one large unit excavated in 2003 (Unit 113) were dry screened through ¼-inch wire mesh with all artifacts bagged and recorded. Excavated feature fill was bagged and saved for flotation. Special samples, including radiocarbon samples, Oxidizable Carbon Ratio (OCR) soil samples, or flotation samples were assigned lot numbers and recorded on a master lot list.

RESULTS OF THE UT-TARL MAGNETOMETER SURVEY
Architectural features are the most abundant and potentially diverse feature class in the George C. Davis magnetometer data (Creel et al. 2005(Creel et al. , 2008Walker 2009). The magnetometer work conducted at the site since 2003 has substantially increased the database of known structures from the site (Creel et al. 2005(Creel et al. , 2008; Walker 2009). The majority of structures identified in the magnetometer data are roughly circular (n=60), represented primarily by the structure perimeter, and range in size from approximately 3 m to over 20 m in diameter.
Interestingly, several of the circular structures identified in the George C. Davis dataset are fairly large, with at least 17 having a diameter of 15 m or more. The large size of some of the structures identified in the magnetometer data bring to mind structures from sites such as Werner in Northwest Louisiana where an earthen mound was built over a 28 m diameter structure (Webb 1983). Table 7 provides summary data for those structures identified in the magnetometer data that are included in this study.

Circular Structures
The majority of structures (n=60) identified in the magnetometer data are circular There are additional circular anomalies with diameters exceeding 21 m, some as large as about 54 m, that may represent an open enclosure of some sort rather than a roofed structure (Creel et al. 2008;Walker 2009). This is a possibility that warrants further investigations as such features might represent open-aired structures, may demarcate ritual spaces such as dance grounds, or may be something similar to "Woodhenge" from the Cahokia site in Illinois There are at least nine circular structures that stand out as having multiple clearly identifiable interior features (Features 237,241,242,(244)(245)(246)(247)335, and 336, see Table   8). These structures offer additional architectural detail in that they are generally represented by the circular structure outline, four large patterns of magnetic returns within the interior, and a pattern of generally high magnetic returns in the center of the structure itself. These structures correspond to Walker's (2009)    interior pits, rather than post holes (Spock 1977;Story 1998).

78
The mean diameter of these structures is 11.98 m with a range from 9.5 m to 15.4 m ( Table 8) The surface area for the magnetic returns interpreted as interior support posts has been estimated from the approximate area of the measurable patterns of returns corresponding to these features. Surface area is used here given the variability in feature shape. As noted in Walker (2009), however, measurements from geophysical features should be considered as approximations given that they represent measurements of the magnetic signature of the features and not the actual features themselves. As Table 9 illustrates, the approximate area for the patterns interpreted as interior The patterns interpreted as hearths associated with these structures have surface areas ranging from 0.94 m 2 to 2.62 m 2 , with a mean surface area of 1.81 m 2 . All of the hearths that are visible are represented by patterns of high magnetic returns, given that these features have been thermally altered.

80
The remaining 51 circular structures include those with possible identifiable interior features and those without any visible interior features (Table 10)

Sub-Square Structures
There are at least three and possibly eight or more patterns of magnetic returns that have been interpreted as representing sub-square to rectangular structures (Features 250,251,252,253,254,260,281,and 347). The patterns of returns for these possible 84 structures range in size from 4.6 m x 4.8 m to 11.7 m x 11.7 m. The areas of these structures range from 22.08 m 2 to 174.37 m 2 , with a mean area of 72.92 m 2 . Figure 24 provides two clear examples of sub-square structures, Features 252 and 281, measuring 11.7 m x 11.7 m and 6.8 m x 6.8 m in size respectively. At least one sub-square structure appears to have been burned (Feature 260). Table 11 provides the dimensions of the subsquare structures.

EXCAVATIONS RESULTS 3
One

Feature 237 -Unit 113
Feature 237 was recorded from the magnetometer data as a circular structure measuring 10.6 m in diameter with four large interior posts and a central hearth (see       Figure 32) and all fill was saved for flotation. The fill directly below the hearth was red with gray clay mottling. A single sample of charred material was collected for AMS dating from below the hearth. Previously excavated hearths from the Davis site have diameters ranging from 48 to 183 cm and depths from 3 to 45 cm bs (Newell and Krieger 1949;Spock 1977).     Spock (1977:47-48) notes: there was much difficulty in identifying the posts set into the trench. Where a dark disturbed soil filled the trench, the posts were faintly visible; where red clay was used as trench fill, the posts were more apparent. Nonetheless, it was observed that the posts were for the most part set on the trench bottom, a few extending slightly below this. When seen in profile, the post stains faded out toward the surface, blending into the upper part of the trench fill.       (Spock 1977). This being said, shallow interior posts have been found at the site as well, with several structures having interior posts no deeper than 35 cm bs, and some that were quite shallower. This difference may be indicative of different uses for the various interior posts, including: support posts for different forms of structures, whether flexed pole or rigid pole structures, or supports for beds, screens, racks, or other interior furnishings or features that may not require as substantially set a post. Likewise, the difference in exterior post size, depth, and spacing likely indicates different architectural forms of the various circular buildings.

Summary of 2003 Excavations
Three Caddo structures were at least partially investigated during the 2003 fieldwork at the George C. Davis site. With the exception of one large block (Unit 113), these excavations targeted specific features identified in the site's magnetometer data.
Using small excavation units that targeted specific features, the hearth and interior posts of these structures were located, profiled, and contents sampled. As indicated by Perttula's (n.d.) analysis of the artifacts recovered from Feature 237, as well as the location, form, and associated features of each structure, Features 237, 241, and 242 fit Spock's (1977) broad classification of domestic structures. In his analysis of the small amount of artifacts recovered in 2003, Perttula (n.d.) notes that the small collection "comprises primarily domestic debris from an Early Caddo (pre-A.D. 1300) occupation (cf. Story 2000)." There is considerable variety in building forms and associated features at the George C. Davis site (Newell and Krieger 1949;Spock 1977;Story 1998). Through the addition of manual excavations, the 2003 field work provides a more refined understanding of the anomalies identified in the George C. Davis magnetometer dataset.
These results provide a better understanding of the correlation between the architectural features identified in the magnetometer data and the rich archaeological record from the site.

DATES FROM GEORGE C. DAVIS STRUCTURES
Prior to the 2003 season, the George C. Davis site had 130 radiocarbon dates that were primarily from the 1968-1970 mound and village excavations by Dee Ann Story (Perttula 1997;Story 1997Story , 1998. Table 13 provides the dates from previously excavated structures (Perttula 1997;Story 1997). With support from the Smithsonian Institution's National Museum of Natural History, the 2003 excavations of three Caddo structures at the George C. Davis site yielded five radiocarbon dates from charred material submitted to Beta Analytic, Inc. (Table 14) Sample Beta-181493 (UT sample/lot number CE19-5378-28) was collected from directly below the hearth (Feature 237.4). This sample has a measured radiocarbon age of 950 + 40 B.P., a conventional radiocarbon age of 940 + 40 B.P., and a 2 sigma calibrated age range of A.D. 1010 to 1195 (B.P. 940 to 755).
In addition to the Feature 237 samples, two samples were recovered from Feature 242 (Feature 242.2, the central hearth, and Feature 242.1, an interior post hole). Sample Beta-181495 (UT sample/lot number CE19-5380-4) has a measured radiocarbon age of 860 + 40 B.P., a conventional radiocarbon age of 860 + 40 B.P., and a 2 sigma calibrated age range of A.D. 1040-1260 (910 to 690 B.P.). Sample Beta-181496 (UT sample/lot number CE19-5380-32) has a measured radiocarbon age of 1630 + 40 B.P., a conventional radiocarbon age of 1600 + 40 B.P., and a 2 sigma calibrated age range of AD 390 to 550 (BP 1560 to 1400). This later sample was collected from the upper portion of the post hole, near where it was first identified.
One radiocarbon sample was collected from the hearth of Feature 241 (Feature 241.2). The small sample of charred material yielded a measured radiocarbon age of 1180 + 50 BP, a conventional radiocarbon age of 1170 + 50 BP, and a 2 sigma calibrated age range of AD 720 to 745 (BP 1230 to 1205) and AD 760 to 990 (BP 1190 to 960). This sample was collected from the burned clay of the hearth at the level at which the feature was first identified.

SUMMARY AND CONCLUSIONS
To date, as many as 133 features have been identified as Caddo structures at the George C. Dais site. Fifty-six of these structures (42%) have been at least partially excavated. The George C. Davis structures provide data on a range of structure forms, from sub-square to rectangular to the more commonly seen circular Caddo structures.
The approximate area of these buildings ranges from 1.80 m 2 to 346 m 2 . Presumed functions include domiciles and probable granaries or storage platforms as well as special purpose structures that were constructed on or under mounds and were likely associated with the civic and/or ceremonial leaders and ritual/communal activities that took place at the site.
In addition to details on single structures, the data obtained from excavations and geophysical surveys provide information on the spatial patterning of structures at the site.
For example, these data show evidence of intensive rebuilding of structures in several areas and include several examples of superimposed structures, a practice which is clear 107 in previous excavations at the site, particularly in the Mound A area, as well as in the magnetometer data. In addition, Walker (2009) has argued that the spatial patterning of anomalies in the magnetometer data suggests that there are multiple plazas and community areas at the site.
From archaeological and geophysical data on individual structures to data on the spatial patterning of these structures (Walker 2009), the George C. Davis site offers a sampling of the architectural diversity seen throughout one part of the Caddo area. The George C. Davis site offers specific architectural data on more structures from a single site than any other site in the region. That being said, however, the architectural patterns seen in the structures at the George C. Davis site are just a small sampling of the character of the Caddo architectural tradition. The next chapter will examine architecture from Caddo mound sites throughout eastern Texas, followed by Chapter 6, which discusses Caddo architecture from non-mound sites.

INTRODUCTION
This chapter discusses the architecture of the 50 known structures associated with earthen mounds from 13 East Texas Caddo sites ( Figure 41 and Table 15). These structures have sufficient enough architectural data to allow for their detailed analysis and comparison with other Caddo sites in the region. Several mound associated structures identified in the study have been omitted from this analysis for various reasons. In most instances, too little of the structure pattern was recorded to allow for detailed analysis. In other cases, features or artifacts related to 109 structures such as a packed floor level, a hearth and center post, burned daub, etc. were recorded, but no structure post pattern was recorded. Figure 41. Distribution of mound sites included in this study.

Caddo Area Mounds
Mounds in the Caddo area typically fall into one of three classes, burial mound, platform mound, or structural mound (sometimes referred to as sub-structural) (Brown et al. 1978;Story 1990;Vogel 2005). Burial mounds are mounds that contain one or more interments of high ranking individuals and sometimes cover and/or contain large shaft tombs. Burial mounds are found throughout the Red, Sulphur, Cypress, Sabine, and Neches drainage basins (Story 1990). Platform mounds are generally square or rectangular with one or more flat-topped stages that were used as platforms for important public buildings (Vogel 2005). Sub-structural mounds are low mounds that typically cover one or more structures, but were not used as platform mounds (Brown et al. 1978).
Schambach (1996:40) has suggested that post-A.D. 1250-1300 Caddo ceremonialism is one in which there was a shift from "individual-oriented ceremonialism to public building-oriented ceremonialism," by which he meant that important buildings were interred in the major mounds rather than important people. Whereas structural mounds in the Arkansas Basin have been viewed as capping the remains of charnel houses (i.e., the Harlan-style charnel house, see Bell 1972;Brown et al. 1978;Kay and Sabo 2006;Vogel 2005), the structural mounds in the East Texas study area likely capped specialized structures where social, political, or religious leaders "could perform sacred rites, where ritual paraphernalia could be kept, and around which members of the society could periodically congregate to reaffirm what was important and right" (Story 1990:341). Perttula (2005:39) has noted that the Late Caddo mound building practices in the Pineywoods of East Texas generally consisted of sub-structural mounds, which are those that "cap a burned circular structure that was constructed on the ground surface or in a small, shallow pit." Perttula (2005:39) further noted that there are no known platform or burial mounds in the area that date from the Late Caddo period.

MOUND-RELATED STRUCTURES
Caddo architecture is typically described as consisting of round beehive-shaped structures; however, there is considerable variation in the shape of Caddo structures throughout the Caddoan archaeological area as a whole, and within the sample of structures examined in this dissertation. Mound related structures include circular, subround (including oval structures), rectangular, and sub-square structures, many with rounded or diagonal corners, and with considerable architectural variation within these groups.
Of the 50 mound-related structures in East Texas, 38 are circular, three are subround (mainly oval or rounded oval), five are rectangular, and four are square or subsquare with rounded or diagonal corners. Table 16 provides a list of the mound-related structures included in this study and their shape and size.
Limited post hole data are available from several of the structures. In ideal cases, site files, reports, and articles provide at least minimum, maximum and average post hole diameter and depth. In most cases, however, only minimum and maximum or average post hole diameter and depth are given. Post spacing and wall to center measurements were not often provided in reports and site files. In cases where such data were not available, plan maps were used to measure the distances.

George C. Davis (41CE19)
The majority of mound-related structures within the study area are from the George C. Davis (n=16) and the Hatchel (n=13) sites. None of the other sites have more than four mound-related structures (see Table 16).
The 16 structures from the George C. Davis site range in shape from circular to subround (oval) to subsquare with rounded or diagonal corners. Chapter 4 provides a discussion of the more than 50 mound and non-mound related George C. Davis structures as categorized by Spock (1977) as either specialized structures or domiciles. The following discussion includes only the 16 structures associated with Mounds A or B from the George C. Davis site.
The mound-related structures from the George C. Davis site represent a diverse array of architectural patterns and practices of use of space. In addition to a range in structure size and shape, Spock (1977:159) notes that "some floor areas appear to have been purposefully cleaned; others are quite littered with artifacts, and a few have  (Table 17).
The seven circular structures tend to be larger, with areas ranging from 16.98 m 2 to 254.5 m 2 , with a mean area of 120.88 m 2 . The smallest of these, Feature 45a, with an area of 16.98 m 2 , was not discussed as a separate structure by Spock (1977) or Newell and Krieger (2000) and may actually be an internal feature within Feature 45. Story (1990:341) noted the presence of this structure and stated that "inside of F-45 was a miniature structure that appears to be a model of the building within that building." If that is the case, then Feature 45a could be an example of a model of the temple that was   (1997) and Spock (1977) refer to as major and minor post holes, making a distinction between the 28 posts "probably set in sloping individual trenches dug perpendicular to the wall" and the 24 posts that were "closely set in a shallow, trough-like depression and located 5 to 20 cm outside of and parallel to the major wall posts" (Spock 1977:106-109).   (   Table 19 x = 5.00 m, s = 1.9136).
There is a wide range in ratios of average distance between wall post holes and average distance to structure centers ( ). The ratio of those measurements ranges from Comparing the circular to the sub-round and sub-square structures yields the x Only one of the George C. Davis mound structures, Structure 111-the largest excavated structure at the site, and the largest previously excavated structure in this study-appears to have been burned (Spock 1977;Story 1997). Spock (1977:109-110) stated that "the prepared clay floor had been baked, and charred beams were found in the depression just outside the major post outline. In addition, several sherds from a Holly Fine Engraved bottle found in the trough on the west side of F111 were well-oxidized from burning. Other sherds from the same bottle, found in an interior post hole, F111-2, had not been affected by fire." Spock (1977:110) concluded by noting that Structure 111 appears to have been at least partially dismantled prior to burning.
= 10 cm, s = 4.73). Five of the excavated hearths were clay-lined. One of these features was described as an ash bed rather than a hearth (Spock 1977). This feature is associated with Structure 45a. Like the other central hearths, the Feature 45a ash bed covered a possible center post.
In addition to being the only excavated burned mound structure at the site and one of only two mound-related structures from George C. Davis with a prepared clay floor, Structure 111 had other interesting features. As discussed above, the excavated portion of Structure 111 had an outer wall that included 28 major post holes, with diameters ranging from 7 cm to 22 cm and an average depth of 89 cm, and a second exterior outline of 24 minor post holes, separated from the major post holes by about 10-20 cm. The minor post holes had an average diameter of 6 cm and average depth of 40 cm (Spock 1977;Story 1997:72-73). Story (1997:72) noted that the major post holes that were sectioned were set "in individual, downward sloping trenches dug perpendicular to the wall. Probably all of the major wall posts were similarly set, perhaps to facilitate erection of what must have been rather long poles." In Chapter 4, this practice was also suggested for the placement of some of the large interior post holes recorded in the magnetometer data from the George C. Davis site. As for the purpose of the outer ring of minor post holes, Story (1997:72) suggested that "since they contained smaller poles it seems probable that they relate to the wall construction and not to the roof."  as the primary mound platform (Newell andKrieger 1949, 2000:66-70), Spock (1977:75) noted that "field profiles indicate that at least two, and possibly four, levels existed." Although Spock stated that the floor levels were not distinct, she noted that based on the positioning and fill of the hearth and post holes of Feature 34, "it is apparent that Feature 122 34 is later in time that Feature 36." Interestingly, in her discussion of Feature 34, Spock (1977:76) commented that "one of the more unusual details concerning this structure is the presence of a center posthole cutting through the fill of the hearth, the only known occurrence of this sequence at the site. This would seem to indicate that the hearth was in use before the centerpost was set". This practice of post hole setting is similar to that found in the Titus phase (ca. AD 1430-1680) mound at the Pilgrim's Pride site (41CP304), where a marker post was erected through the mound and through the central hearth of the structure capped by the mound (Perttula 2005).
The green clay altar (Feature 56) was also found on this mound platform and "appears to have the same orientation -essentially east-west -as Features 34/36" (Spock 1977:76). Features 39 and 40 were located at slightly higher levels than Features 34 and 36, and Spock (1977:85) suspected that they may be later; Feature 40 may be the later of the two given the placement and fill of the hearth.  (Newell andKrieger 1949, 2000;Spock 1977:64). Spock (1977:66) notes that Feature 38 was the later of the two, "having four wall postholes which cut across the long oval pit" associated with Feature 37 (Features 37-2).
In discussing the abandonment of these structures, Spock (1977:66) indicated that they were "perhaps deliberately dismantled (no evidence of burning) the interior were cleaned and filled with 0.55 m of grayish clay mottled with yellow and green." Feature 31 was built on this fill, "raised slightly above the terrace surface by the clay filling of Features 37 and 38" and was centered directly over Features 37 and 38 (Spock 1977:63 (Perttula 2005), the large platform mound at the Hatchel site was found to have structures on seven of the 11 mound stages, including the pre-mound surface (the mound floors were designated floors A-K, with A representing the latest addition and K representing the pre-mound surface) (Jackson 2004 (Creel 1996) as well as several concentrations of post holes that did not conform to any discernable pattern. Some ambiguous or irregular structure forms were assigned by the WPA, with over 20 possible Caddo structures and associated outbuildings (arbors, drying racks, granaries, etc) represented throughout the mound (Jackson 2004).  small "pole" buildings, as opposed to large "post" buildings, those with post hole diameters averaging more than 20 cm (Lacquement 2007:55).  there is a fairly wide range in both wall post spacing and distance from structure wall to center. The relationship between these two measurements ranges from 1:6.0 to 1:13.7, 127 with an average ratio of 1:8.48. Again, the outlier is Feature 25. Table 22 provides the measurements for between the wall post holes, the distance from the wall to the center, and the relationship between the two. .14 n/a n/a 0.15 n/a n/a 0.18 5 3.66 n/a n/a 0.18 n/a n/a 0.18 7 9.14 n/a n/a 0.12 0.61 0.06 0.22 12 8.87 n/a n/a 0.18 n/a n/a 0.15 13 9.06 n/a n/a 0.15 n/a n/a 0.21 14 8.84 n/a n/a 0.15 n/a n/a 0.21 15 9.75 n/a n/a 0.18 n/a n/a 0.24 16 7.32 n/a n/a 0.18 n/a n/a 0.18 17 6.40 n/a n/a 0.  2005,2009). Finally, the Feature 25 interior partition has a gap in the post holes that likely represents a passageway through the wall. nonextended n/a n/a n/a 13 nonextended n/a n/a n/a  Only one of the structures (Feature 14) had a possible prepared floor. Feature 14 is described in UT-WPA notes as being located on a red clay deposit that may have been placed to form the structure floor. The various mound stages were clearly prepared prior to the buildings being constructed, but no clearly definable prepared floor was otherwise recorded for structures on these mound platforms.
Interior recorded as also containing 10 ceramic sherds, two mussel shells, and three animal bones.
The Hatchel site appears to have been occupied from as early as ca. A.D. 1000 to the late eighteenth century (Creel 1996;Perttula 2005). There are four radiocarbon dates from the Hatchel Mound (Perttula 1997), with additional dates from the Village deposits (Perttula 2005). As shown in Table 25, two of the dates from the mound are from separate mound floors (K and H), with the other two coming from Features 18 and 4 (Perttula 1997). Recent work in the village areas of the site (Perttula 2005) has resulted in five additional radiocarbon dates and 10 Oxidizable Carbon Ratio (OCR) dates. Perttula (2005:184) notes that these dates: 132 indicate that the Hatchel site was first settled before A.D. 1300. This is contemporaneous with the Middle Caddoan period Haley phase [A.D. 1200-1400] and lends credence to the suggestion first made by Krieger (1946) that the initial mound construction in the primary platform mound took place before the Late Caddo Texarkana phase. Other dated samples argue for the continuous Caddo settlement of the Hatchel village from at least the thirteenth century to the mid-or late seventeenth century, the same time as the construction and use of most of the platform mound, according to Creel (1996).   (Table 26) (Jelks and Tunnell 1959).    (Jelks and Tunnell 1959).
1. lower house floor zone (house 1) blackish sand with large amount of charcoal avg thickness = 0.3 ft. 2. Upper house floor zone (house 2) slightly compacted, brownish sand with lots of charcoal and a few artifacts separated from house 1 floor "by a layer of clean, sterile, yellowish sand 0.1 -0.3 ft. thick which probably was placed over the burned ruins of the first house to provide a clean floor for the second one" 3. Embankment of yellow-brown sand encircling house area max height = 2 ft. above zone IIC (humus-stained topsoil), ca. 0.6-1.1 ft. thick 4. Final addition to the mound virtually sterile layer of sand mounded over burned houses  Structure 3 had several interior post holes, including a cluster recorded in front of the entrance and a cluster found along the northeast wall, that likely represented interior 138 features such as benches, racks, storage platforms, beds, or screens (Jelks and Tunnell 1959:20). Along with the 30 irregularly spaced interior posts, four large interior posts were identified that "probably held the bases of relatively heavy roof supports" (Jelks and Tunnell 1959:20). Like the other Harroun mound structures, Structure 4 had a sand embankment built against its wall. The Structure 4 berm was smaller than the ones associated with the other structures and measured 45 cm high and 1.22 to 1.83 m in width (Jelks and Tunnell 1959:34). The floor of the structure was composed of a 6 cm think layer of hard-packed sandy clay, and "just above the house floor was a 1' [30 cm] thick layer of dark graybrown sand containing several charred poles and a large amount of charcoal, ash, and burned clay daub. Above that was the sandy fill making up the bulk of the mound" (Jelks and Tunnell 1959:34). Jelks and Tunnell (1959:30) noted that "the sparse occurrence of artifacts and other cultural refuse suggests that neither House no. 1 nor House no. 2 was an ordinary domicile. It appears likely, rather, that both were ceremonial structures of some sort. This hypothesis is supported by the fact that the houses were considered important enough to be afforded burial beneath a mound, probably after having been ceremonially cremated" (Jelks and Tunnell 1959:30). They summarized the Mound C structures by stating that: A circular house (House No. 1) was built on the south bank of the Harroun site lake, was occupied for an unknown period of time, then was burned -perhaps intentionally. After a thin layer of sand had been strewn over the burned ruins, a second, smaller house (House 2) was erected on the remains of the earlier house. House No. 2 was likewise destroyed by fire, after which the remains of both houses were buried under a mound of sand (Jelks and Tunnell 1959:31).
As noted above, the Harroun structures date to the Late Caddo Titus phase (ca. A.D. 1430-1680) (Jelks and Tunnell 1959;Perttula 2009;Thurmond 1990). At least five radiocarbon dates are available from the Harroun site (Perttula 1997). One of the dates is from Mound B, two are from Mound C, and two are from structures in Mound D (Table   27).

Dalton (41UR11)
The  The ratio of average distance between wall post holes and average distance to the center for Structure A was 1:6.08 and for Structure B the ratio was 1:5.31.
No evidence for central hearths was found in either structure, although Davis and Gibson (1960:12) noted that if the structures had central hearths, evidence for them may have been destroyed by later burning of the structures. Both structures did have center posts (Davis and Gipson 1960:12). Three center post holes were recorded and Davis and Gibson (1960:14) commented that they did not "consider it possible to correlate [a] specific center post hole with particular structures." That being said, however, their discussion of the center posts noted that the northernmost post hole was the deepest and may be related to the older of the two structures, Structure A, but that "if one uses the center posts molds as pivot points for a compass, the middle mold best matches the House A post mold ring and the southern mold best matches the House B ring" (Davis and Gibson 1960:14). Another possibility is that at least two of the post holes may relate  and Gipson (1960:18) to suggest that this "may indicate that the time between the destruction of House A and the construction of House B was relatively short, or it may simply signify that the area was cleared off before work began on House B." They continued by noting that the later is probably the case and that any debris associated with Structure A appeared to have been cleared away prior to construction of Structure B (Davis and Gibson 1960:18).
Structure B was built within the same prepared pit and also had an earthen embankment around the wall (Davis and Gipson 1960 Perttula 1997Perttula , 2009 (Davis and Gipson 1960:65). Davis and Gipson (1960:21) interpreted the Dalton structures as being ceremonial rather than simple domiciles. There was no concentration of occupational debris associated with the structures and they further noted: there is archeological and ethnohistoric evidence for the ceremonial construction and burning of houses by the Caddo, extending from Haley focus times to the late 17th century. The occasion for this practice was not always the same: at the Belcher site, in the Haley and Belcher foci, the structures were destroyed as part of the obsequies for important personages, whereas among the historic Hasinai the practice is reported as being part of the ceremonial preparations for the setting out of a war party. (Davis and Gipson 1960:21).
As discussed by Davis and Gibson (1960), the architecture from the Dalton site was similar to that from both the Harroun site, discussed above, and the Whelan site (41MR2), which consisted of four earthen mounds and a fairly extensive village area.

Whelan (41MR2)
The Whelan Site (41MR2)  Partial outlines of at least three and possibly four superimposed circular structures were identified by E. Mott Davis in Mound A (Davis 1958:26). The structures were recorded as Houses A-D in Davis (1958) and Structures A-C in Lisk (1984); Davis' House C was omitted by Lisk (1984:58) because of its poor definition. Rather than being built on deliberate mound capping zones, subsequent structures appeared to be built upon the accumulation of occupational debris (Davis 1958;Thurmond 1990:168).  (Lisk 1984:61). Davis (1958:32) suggested that perhaps: Houses C and D represent in actuality only one house with two concentric rings of post molds. In such a case the instances of one post mold intruding into another, cited earlier as evidence for two separate, sequent structures, might represent repair work on one structure which, as indicated by the accumulation of ashes and dirt on the floor, was used for a long time. Lisk's (1984: (Good 1982;Rogers and Perttula 2004). Based on that, House C may actually represent a rebuilt House D. interpretation due to the shallowness of the post holes), House A was the only building with signs of burning (Lisk 1984:62). Finally, the destruction of this structure was "accompanied by some burning and was followed by the placement of a relatively sterile earth cap to form the final, rounded surface of the mound" (Thurmond 1990:168). Davis (1958:33) mentioned "a peculiar soil feature associated with some of the post molds of House A": the soil discoloration associated with a few of these molds extended radially outward (i.e. away from the center of the structure) from the mold proper for a distance of a little over 1.5 feet [0.46 m], almost as if the posts had been placed at the inner end of a radial slot. The 'slots' were not sharply defined in every case. We were unable to determine why they were there. It is hoped that other excavations may turn up a similar phenomenon under circumstances more susceptible of interpretation (Davis 1958:33).  Mound A also had a series of at least three superimposed center post holes and central hearths (Davis 1958:30  In discussing assigning the hearths and center posts to Houses A, B, and D, Davis noted that Structure C lacked those features. Davis (1958:32) suggested that Houses C and D may represent a single structure with two concentric rings of post molds.
Given the incomplete perimeters, no entrance was found for either Houses A or B.
The cap on Mound A measured about 0.91 m. In summing up Mound A, Davis (1958:33) stated: Our minimal history of the mound is, then, as follows: a house (House D and possibly House C as part of it) was built at ground level and used for a long enough time for about a foot [30 cm] of ashes and dirt to accumulate on the floor. Then a new house, House B, was built at the higher level. Finally House A was built, and its destruction was accompanied by some burning. Eventually the cap of the mound, Zone III, was adder over the remains of House A.
These two Middle Caddo sites are located within the Harris Creek drainage in the Sabine River basin and each contains small mounds covering a single circular Caddo structure Haskins 1998, 2000).
Excavations at the Redwine site uncovered four Caddo burials and a house mound as well as two non-mound related structures (Walters and Haskins 1998:1). Later excavations at the site recovered almost 6000 artifacts and identified several features including post holes, a hearth, and two pits (Walters and Haskins 1998:4). The site is described by Walters and Haskins (1998:35)   Since a large area (4.6 m 2 ) of the structure interior was not excavated, no interior features were found (Walter and Haskins 1998:23). Situated immediately outside the structure to the west of the entrance was an "enigmatic deposit of clay covering about 50 square feet [4.6 m 2 ]" (Walter and Haskins 1998:23). Walters and Haskins (1998:25) note that while no data were available on the nature of the clay deposit, "its placement clearly suggests its function was related to the use of the special building subsequently covered by the earthen mound." Perhaps the clay deposit was part of an earthen berm around the structure.
A single radiocarbon date was obtained from the Redwine site. Charred nutshells from the hearth yielded a calibrated age range of AD 1302-1434 (calibrated at 2σ) (Walters and Haskins 1998:35).  Again, there is a discrepancy between the measurements presented in Walters and Haskins (2000), a range of 0.05 m to 0.13 m, and the measurements taken from the plan map. It is quite possible that the posthole size depicted on the plan map is inaccurate.
. The largest two post holes (0.17 m and 0.21 m) were the two located at the wall of the structure (see Figure 51). Removing those two measurements -since they were part of the structure wall -leaves a much smaller range of entrance post size, 0.08 to 0.14 (n = 29, x = 0.11 m, s = 0.0127), similar to that presented in Walters and Haskins (2000). Although there is a much tighter range and considerably smaller standard deviation, the entrance post hole mean remains almost the same. In addition to having much smaller posts, the extended entrance posts appear to 156 have been placed in groups much like some of the wall posts seen at the George C. Davis site.
Charred posts from Structure 1 at the Bryan Hardy site yielded a age range of cal AD 1345-1391 (calibrated at 1σ) and cal AD 1297-1317 (calibrated at 1σ) (Walters and Haskins 2000:4;Perttula 1998: Table 1). Walters and Haskins (2000:4) noted that "there is a 67% chance that the calibrated age of the Bryan Hardy charred post falls between AD 1297-1391." The presence of a similar unexcavated mound at the Tom Moore site (41PN149) suggested to Walters and Haskins (2000:25) that "although there are temporal and spatial differences between these sites, they share common ceramic assemblages" and "the pottery at each of the sites is consistent with Middle Caddoan period wares documented elsewhere in the middle reaches of the Sabine River basin at that time." It is quite possible that the 8 m diameter by 1 m high mound at Tom Moore site covers a burned circular structure similar to those found at the Bryan Hardy and Redwine sites. The Redwine and Bryan Hardy sites are: similar to several other Middle Caddoan sites in the general area in that they are located on landforms that had no history of previous occupations, they are considerable distances from available water, apparently are occupied year-round for short periods of time (20+ years), and have a single earthen mound covering a burned house. Whether this was the typical settlement pattern in the Middle Caddoan period is unknown, as is their association with other homesteads or larger centers (Walters and Haskins 2000:24).

Sanders (41LR2)
At the Sanders site (41LR2), 1931 excavations by Jackson (2000) (Perino 1995). Perino (1995) mentioned multiple Caddo structures, including at least two associated with five of the storage pits east of the mound, and at least three associated with the mound; however, the 1995 report only offered details on one mound-related structure.
Perino (1995:7) mentioned that there were three structures associated with the mound, but "the earliest two houses were marked by incomplete post mold patterns".   Perino 1995). Reproduced courtesy of the Friends of Northeast Texas Archaeology.
While post hole diameter and depth was not provided for this structure by Perino (1995), measurements taken from the plan map, while approximate, clearly indicate that the posts used for the entrance were generally smaller than those used for the rest of the A hearth was located inside the structure, directly across from the entrance, near one of the walls (see Figure 52). The hearth measured approximately 1.40 m in diameter.

160
The hearth had post trenches on either side of it, possibly representing a cooking platform (Perino 1995:7-10).
Several other features were found beneath the mound, including two hearths, two dog burials, two smudge pits, and the McCurtain phase (Perttula 1995) grave of a child (Perino 1995:10). Perino (1995) does not discuss whether these features were directly associated with Structure 1, other than to note that the smudge pits were located 0.60 m beyond and outside the entrance. Perino (1995:7) suggested that Structure 1 "was placed in the mound at the original ground level, or conversely it was constructed at the original ground level with soil berms placed around it." If the latter is the case, then this would be the only East Texas Caddo case of a rectangular structure associated with an earthen berm.
In his analysis of the chronological placement of the mortuary remains, Perttula (1995:68) noted that Perino's interpretations of the burials at the site being primarily of Sanders phase age "cause him to overlook the real possibility that the mound north of the spring branch was a place of burial after the Sanders phase, as well as deny the possibility that many of the pit and house features under the mound actually originated from, and cut through, the mound surface, and thus do not solely represent house and mortuary use prior to their being capped with an earth mound." Perttula (1995:73)

Roitsch (41RR16)
At least one mound structure has been recorded at the Roitsch site (41RR16, also known as the Sam Kaufman site), also located in Red River County, that had enough of a post hole pattern to be included in this study (Skinner et al. 1969:19). Structure 2 was located in the East Mound and likely dated to around the end of the Sanders phase to the early McCurtain phase (Bruseth 1998;Perttula et al. 2001). Evidence for additional mound-related structures was recorded by S. Alan Skinner in the West Mound and included signs of superimposed burned structures (Perttula et al. 2001;Skinner et al. 1969).  shaft burial that went through the floor of the structure (Perttula et al. 2001;Skinner et al. 1969:19). The shaft burial included 11 individuals, including a principal older male and female and nine other individuals (Perttula et al. 2001;Skinner et al. 1969:19). The shaft burial was dated to cal A.D. 1412-1511 (Perttula et al. 2001).
Additional mound burials were recorded by later excavations at the site during the 1991-1992 Texas Archeological Society field school (Perttula et al. 2001).

Keith (41TT11)
The Keith Site (41TT11) covers an area of 75,000+ m 2 and includes a small cemetery, an occupation area, a probable borrow pit, a midden area, and an Early to Middle Caddo earthen mound (Goldschmidt 1935b;Thurmond 1990:78, 183-185). In 1934, at the time of clearing and excavations by a UT crew under the direction of W. R.
Goldschmidt, the mound measured 73 (n-s) x 49 (e-w) x 4.6 m (height). According to Goldschmidt (1935:98), the top of the mound may have been a flat rectangular area at one time. Several mound stages were evident and Thurmond (1990:184) noted that four major construction episodes were identified with evidence of "significant time lapses" between construction events.
Given the nature of the mound, it has been suggested that there were probably structures on the original pre-mound surface and then on each of the mound platforms (Thurmond 1990:184). This would have created four sequent structures within the mound, and the possibility of a fifth atop the final mound platform (although no evidence for this was found) (Thurmond 1990:184). Despite the likelihood of multiple sequent structure, features and material remains associated with only two structures were identified, one on the third mound platform and the other on the pre-mound surface. In addition, an area of post holes, pits, ash, and charcoal off the mound was interpreted as an outdoor activity area (Goldschmidt 1935;Thurmond 1990:184).
Thurmond (1990) noted that the mound and sub-mound material culture remains corresponded to an Early Caddo occupation with additional Early Caddo debris recovered from the nearby terrace. Late Caddo debris was found on the terrace and within the vicinity and on the surface of the mound, along with several Titus phase burials (Perttula et al. 2010). One radiocarbon date from material apparently produced "prior to the construction of the second mound platform" provided a calibrated one-sigma date range of A.D. 1340-1480 (Thurmond 1990:184).
Limited information was recorded for Structure 1. Atop the center of the third mound platform, excavators identified a heavy concentration of charcoal and wattleimpressed daub (Goldschmidt 1935a(Goldschmidt , 1935bThurmond 1990:184). Despite this concentration of material, no post holes were identified. Thurmond (1990) noted that this was the remains of a burned structure destroyed immediately prior to the addition of the final mound platform and that the location of the concentration of material suggested that the structure was located at the center of the third platform. The remains of Structure 2 (originally Feature 1) and its associated features were more substantial (Figure 55). The southern portion of Structure 2 was identified on the pre-mound surface (Goldschmidt 1935:99-100;Thurmond 1990). No entranceway or central hearth were found, although less than half of the structure was exposed.
Approximately 3.6 m to the east and 1.82 m to the west of the structure were north-south running lines of post holes (Goldschmidt 1935:99-100). Within these walls, running along the interior of the east wall, was a partially exposed 1.5 m x 4 m midden deposit. In addition to this, a 3 m diameter area of post holes, pits, ash, and charcoal was located directly beyond the western wall (Thurmond 1990). Goldschmidt (1935:100) noted that "the straight lines represent portions of a rectangular enclosure can only be surmised, but it is fairly certain from historical accounts that this interpretation is correct."

Oak Hill Village (41RK214)
At least 43 structures of varying size and shape were uncovered at the Oak Hill Village site as well as a central plaza area and a small earthen mound (Perttula and Rogers 2007;Rogers and Perttula 2004 Perttula and Rogers (2007:86) is that "the combination of mound, plaza, and encircling circular residential structures, indicates the mature development of the Oak Hill Village during the middle-era." In addition, Perttula and Rogers (2007) noted "the obvious concentration of structures near the earthen mound, and the rebuilding of many structures here, seems analogous to the 'inner precincts' defined by Story (1998:26)  Architectural data on these 50 structures come from 13 sites located throughout eastern Texas. These sites include major multi-mound centers such as the George C.

INTRODUCTION
In addition to the 50 mound-related structures discussed in Chapter 5, this study includes 128 previously excavated non-mound structures as well as 87 structures recorded through magnetometer surveys. These 215 structures are from 24 sites throughout eastern Texas (Table 29 and Figure 56). These structures provide sufficient architectural data to allow for detailed analysis and comparison that is the objective of this study.  Figure 56: Distribution of sites with non-mound structures included in this study.
There are several other non-mound structures or structure areas recorded throughout the study area, but I have had to omit them from consideration for some of the same reasons discussed in Chapter 5. Namely, in most instances too little of the structure pattern was recorded to allow for detailed analysis of post hole attributes, structure size, etc. Structure areas indicated by a packed floor area, burned daub, a circular pattern of burned earth, or structure patterns that are too incomplete have also been omitted.

NON-MOUND-RELATED STRUCTURES
Of    (Table 33). With the exception of Feature 35, the remaining 38 non-mound structures have areas ranging from 1.8 m 2 to 153.9 m 2 with a mean area of 69.13 m 2 (s=38.5188). As noted in Table 34   1:13.9, or 1:6.3 to 13.9 without Features 7, 50 and 52. The range for the sub-round structures is much smaller, ranging from 1:6.3 to 1:7.3. The ratio for the sub-square structures ranges from 1:9.0 to 33.1, or from 1:9.0 to 1:15.6 with Feature 9 excluded.
When the mean post hole spacing from these three groups is compared, the    (Spock 1977). In addition, only Features 9 and 35 had prepared floors and had extended entrances. The Feature 9 floor consisted of a 2-3 cm thick clay lining that completely covered the structure floor (Spock 1977). The burning of Feature 9 is supported by the fact that the clay floor "was hard, cracked, and buckled from firing and the underlying soil was affected to a depth of 10 cm. Charred posts were found in the house fill, most aligned either parallel or perpendicular to the encircling post trench.
Within the post trench, the top 2 to 5 cm of several posts were charred" (Spock 1977:44-47).
= 21.5 cm, s = 16.2993). Two of the four hearths were clay-lined. It is possible that the lack of hearths or center posts for most of the non-mound structures may be attributed to the high concentration of rebuilding in certain areas of the site and possibly disturbance from subsequent building episodes (Spock 1977), or even to modern disturbance such as plowing of the site's archaeological deposits.
Feature 9 had a southwest-facing extended entrance consisting of two parallel trenches 0.6-0.7 m apart and extending 1.7-1.9 m away from the structure walls (Spock 186 1977). The entrance "was paved with a layer of grayish-white, hard packed clay which measured 2.14 m long, 0.46 m to 0.61 m wide and was raised 0.31 m above the structure's rim. A second layer, 9 cm thick, and consisting of compact reddish clay, overlay most of the grayish-white clay" (Spock 1977:47). Finally, Feature 9 was built in a 0.37 m deep pit. This is the only excavated semi-subterranean non-mound structure from  Spock (1977) included Features 6 and 125 as special function structures given their architectural attributes or apparent use as specialized buildings. Feature 6 is a subsquare structure with walls aligned to the cardinal directions. In addition, this structure is similar in form to Features 34, 39, and 40, which were located on Mound A platforms (Spock 1977). Feature 125, located between Mounds B and C, was included due to the apparent reuse and renewal of its central hearth. The hearth was clay-lined and had been refurbished at least three times. Spock (1977:178) commented that "the hearth with its  Table 36 provides summary data for these structures. While there were a few isolated structures at the site, most were situated in spatially distinct clusters, some of which were "composed of many dwellings that represented successive building episodes" 188 (Rogers and Perttula 2004:49). In fact, there were "as many as five episodes of house building" at the site (Rogers and Perttula 2004:49   Structure 38 measured 12 x 7.5 m, was oriented north-south, and had evidence of repair. Measuring 11.5 x 8 m, Structure 39 also showed evidence of repair. In terms of repair for both Structures 38 and 39, "some wall posts appear to have been replaced" (Rogers and Perttula 2004:78). Finally, Structure 43 measured approximately 6.5 x 9 m and was defined by portions of the northern and eastern walls (Rogers and Perttula 2004).   (Table 38).     Given the larger ratio, Rogers and Perttula (2004:81) suggested that "it may have been laid out differently or independently of all others." The Structure 2 extended entrance faced northwest, toward the plaza, and consisted of two roughly parallel shallow trenches with closely spaced upright posts (Rogers and Perttula 2004:81). The entrance trenches extended approximately 2.15 m and 2.9 m away from the structure with a total of 27 post holes, 14 in the shorter trench and 13 in the longer. Structure 2 is the only structure at the Oak Hill Village site with post trenches, further suggesting that the structure was built for a specialized purpose (Rogers and Perttula 2004;Perttula and Rogers 2007). Structure 2 dates to the Middle Village era (A.D. 1250-1350) (Rogers and Perttula 2004:86).
Most of the Oak Hill Village structure had center posts and central hearths (Rogers and Perttula 2004). Rogers and Perttula (2004:111) provided a general discussion of these associated features, noting that "the average diameter of recorded center posts at the site was approximately 35 cm, while some, such as those from Structures 1 and 7 were 50 cm in diameter." In addition, several small and large pits were recorded at the site. Many of these were located in the structure areas but were not directly noted as being associated with specific structures. Some, such as Features 85, 86, 163, and 164, were associated with Structures 2 or 10, those individual structures not located in the main domestic structure areas.
While many of the structures had interior posts, the form of Structure 5 is similar to structures at the George C. Davis site, the Earspool site, and others in eastern Texas.
Structure 5 has at least four interior support posts arranged in a squarish pattern around a center post. According to Rogers and Perttula (2004:84), Structure 5 dates to the Middle Village era (A.D. 1250-1350) (Rogers and Perttula 2004:346) and was probably built around A.D. 1230-1272.
Most of the Oak Hill Village structures date to the Middle Village era. Rogers and Perttula (2004:86) suggest that structures in Structure Groups A, B, C, D, and G were constructed during this time as well as the plaza and the small earthen mound. Rogers and Perttula (2004:86) further suggest that other structures in Structure Groups D, E, G, and H were constructed and used during the Late Village era.

Pilgrims Pride (41CP304)
The 1999  The interior smudge pits were pits with dark carbon-rich fill that extended to depths of less than 0.15 m below the scraped surface (Perttula 2005). No central hearth was recorded for Structure 1, although a central support post was recorded. Perttula (2005)  in Area VIII." The five smudge pits were recorded inside the probable structure outline, as was a center post (Perttula 2005). Unfortunately, the Structure 2 post hole pattern is fairly incomplete, a condition attributed to poor preservation as well as the probability of other shallower post holes that may not have been recorded or observed during the monitoring in Area VIII (Perttula 2005). The postholes that were recorded as part of  Perttula (2005:80-81) notes that none of the structures at the Pilgrim's Pride site appear to have been burned, and given the low density of artifacts in the post holes 203 associated with the Titus phase structures in the residential areas, the structure areas appear not to have been used heavily before the houses were built. Perttula (2005:80) also suggested that the paucity of materials in the post holes and the limited amount of trash or midden accumulation in the immediate vicinity of the structures was indicative of the fact that the structures were likely only used for a few years.

Bryan Hardy (41SM55)
In addition to the mound structure from Bryan Hardy discussed in Chapter 5, two overlapping circular structures (Houses 2 and 3) were found at the site. These two structures were located about 45 m to the north of the House mound (Walters and Haskins 2000). Unfortunately, limited information is available on the details of these two A poorly preserved burial (Burial 1) was found on the northeast side of House 2.
The burial was extended and supine with the head to the southeast and the "burial pit extended into the house between two posts" (Walters and Haskins 2000:7). Burial 1 was thought to be an infant due to "the small size of the offerings, and [the burial's] placement partially inside the house" (Walters and Haskins 2000:7).  Figure 62). As for an entrance, "there may have been 2 entrances to House 3, one consisting of a 0.9 m gap in the east wall, and another in the southwest corner" (Walters and Haskins 2000:6). The possible entrance in the southwest corner had a "very dark, heavy midden deposit" (Walters and Haskins 2000:6).

205
Charred posts from the entrance yielded date ranges (at 1 sigma) of cal A.D.

Musgano (41RK19)
The Musgano Site (41RK19) covers approximately five acres and is situated on an upland landform between Martin Creek and Dry Creek (Clark and Ivey 1974:14-41;Perttula 2004:25), two tributaries to the Sabine River. Several features were identified at the site including post holes, pits, two burned areas, midden areas, and a single 6 m diameter circular Middle Caddo structure with an approximate area of 28.27 m 2 ( Figure   63). pits (14 trash and storage pits from within the structure outline) were found at Musgano (Clark and Ivey 1974). The post holes had round bottoms and generally measured about 0.20 m in diameter (Clark and Ivey 1974). Thirteen additional post holes were found in the interior of the structure although Clark and Ivey (1974) could not confidently state that they were directly associated or contemporaneous with the structure. If they were, they likely represented interior supports and/or possible interior furnishings, stands, etc.
Several pits were located within the structure and may be associated with the use of the building. There was no evidence that the structure had a prepared floor. A burned area associated with a large pit was located near the center of the structure, probably representing a 0.40 m in diameter and 0.13 m deep central hearth that contained charcoal, bone, shell, nuts, and one flake (Clark and Ivey 1974).   In addition to the possible interior features, several post holes, pits and two midden areas were located outside of the structure perimeter. The additional post holes (see Figure 63) may represent affiliated ramadas, granaries, or elevated storage facilities.
A burned area representing a possible 0.45 x 0.25 m deep hearth was identified outside of the structure to the southwest (Clark and Ivey 1974).
About 3 m east of the structure, excavators identified more than 10 post holes and five pits associated with a relatively large midden area with lithics, animal bones, and the highest concentration of sherds in the main excavation areas (Clark and Ivey 1974:26).
Given the concentration of the midden area to the east of the structure, Clark and Ivey (1974:22-23) suggested the entrance was in this area. In addition, post hole patterns to the east of the structure suggest that the structure may have had an extended entranceway.
This is based on the argument offered for an east-facing entrance and the pattern of post holes extending to the east away from the structure perimeter. In addition, Clark and Ivey (1974:24) thought that the concentration of exterior features may "represent a granary 208 built on stilts possibly having a smoky fire under it to keep out vermin and prevent mildew."

Hines (41WD87)
Three overlapping structures, consisting of a total of 217 post holes, were recorded at the Early Caddo period Hines site (41WD87), located along Burke Creek in Wood County (Bruseth and Perttula 1981) (Figure 64). The three structures "consisted of a compacted mass of charcoal, clay daub with wattle impressions and numerous artifacts" (Bruseth and Perttula 1981:23). The burned layer of house debris was 10-20 cm below the midden layer and about 4 m in diameter (Bruseth and Perttula 1981:24).
Excavation trenches were originally focused on artifact clusters recognized in systematic surface collections at the Hines site and were then extended beyond the one defined midden area. Bruseth and Perttula (1981:22) noted that house patterns from the Deshazo site were found around a midden area and the expectations at Hines were that a similar association of midden and Caddo structures might be present there. Unlike Deshazo, however, the Hines structures were found below the midden deposit/artifact cluster and not around it (Bruseth and Perttula 1981:22).
Hines Structure 1 was a circular structure defined by an outline of 34 post holes.    (Bruseth and Perttula 1981).
As noted, the large rectangular structure was determined to probably be the most recent structure at the Hines site. Bruseth and Perttula's (1981) construction sequence was based on the evidence that interior posts from the circular structure were cut by later posts of the rectangular structures. Based on the interior supports of the large rectangular house, as well as entrance location and evidence of burning of the larger structure, it appears as though the smaller structure was built second and the larger rectangular structure last. This is an interesting and potentially unique case in East Texas where a circular Caddo structure predated the construction of rectangular structures. Bruseth and Perttula (1981:24) believed that the positioning of the structures suggested a sequential building episode:

212
The slight shift of each successive structure is similar to the pattern observed at the Deshazo site in east central Texas, where house pattern shifts were thought to be due to house rebuilding and the lesser difficulties of digging holes for new posts. If we assume a maximum of 20 years occupation per house, then we can project a 60 year occupation length for the total house sequence (Bruseth and Perttula 1981:24).
In addition to three sequent structures, a 2 m diameter basin-shaped pit (first used for storage, and then as a receptacle for trash) estimated to be 30 cm deep was identified 2 m north of Hines Structure 1 (the circular structure) (Bruseth and Perttula 1981:24-26).
There are several instances known in earlier Caddo sites where structures occurred in proximity to a large outdoor storage pit (cf. .

Spoonbill (41WD109)
The multiple component (Early, Middle, and Late Caddo) Spoonbill site (41WD109) is situated on a terrace about 2 m above the Caney Creek floodplain. Bruseth and Perttula (1981) estimated that about 25% of the site had been destroyed by construction prior to their work on the site as part of the Lake Fork Creek Reservoir project, with a large portion of the site having been previously removed for road fill.
Several features (Table 41) were found at Spoonbill including human burials, hearths, pits, and several post holes, including at least 32 post holes that defined a circular house pattern (Bruseth and Perttula 1981). Radiocarbon dates suggest that the structure dates to the Middle Caddo period (Perttula 1997).
The Spoonbill structure is an approximately 10.5 m diameter circular structure with an outline defined by 32 perimeter post holes. Figure 65   There is a discrepancy between the plan map and the report text related to the directions of features from the structure. Based on the feature descriptions and the site map (Bruseth and Perttula 1981:41), it appears as if the north arrow on the structure plan map may be reversed. If not, then the feature description locations, relative to the structure, are incorrect. I have chosen to follow the descriptions from the text and have left the plan map as it was in Bruseth and Perttula (1981  43, x = 4.12 m, s = 0.3076). The ratio of average distance between wall post holes and average distance to the center was 1:8.4 (Davis 1958:34-35). At least nine additional post holes were identified scattered around the interior of the structure, most likely reflecting the locations of interior supports. Davis (1958) also stated that given the difficulty in identifying post holes in this area of the site, many probably went unrecorded. The Structure 2 post holes were not fully excavated, and therefore no depths are available. cm in diameter (Davis 1958:38). Structure 2 was interpreted as a domicile by Davis (1958:36-38), based on the "refuse pits within the house and the abundance of cultural material in the fill." The midden accumulation in Structure 2 was interpreted to be the result of the use of the structure for waste disposal following its abandonment.

Henry M. Site (41NA60)
A A grouping of postholes immediately to the north of the structure may represent an arbor, ramada, or granary (Perttula et al. 2010). While a small gap in the partial arc of the structure may represent a west-facing entrance, Middlebrook and Perttula (2008) suggested, based on Good's (1982) analysis of the structures from the Deshazo site and the relation of platforms with historic Caddo structures, that the outdoor platform/granary/arbor may be located in front of the structure, thus possibly placing the entrance on the north side (see also Perttula et al. 2010:4). Middlebrook and Perttula (2008:20) concluded that " [G]iven that Caddo wood structures would probably only last at most 20 years before they begin to deteriorate (see Good 1982:69), available feature evidence suggests that the houses and midden deposit were created over a ca. 20-40 year period by one or two Caddo families."

Blount (41SA123)
At least one probable structure was identified among the many late prehistoric contained bone, sherds, charcoal, and "other debris" (Jelks 1965:102). This feature provides another example of a storage pit located near a Caddo structure that was later filled with trash.
Jelks described the midden as a "typical one for the region" (Jelks 1963(Jelks , 1965, and it was situated in an area with several post holes. The partial outline of only one oval or sub-round structure was recorded that measured 5.18 m long by 3.35 m wide (Jelks 1963(Jelks , 1965. A small hearth near the southern end of the structure and an interior support post were found within the structure perimeter and no entrance was recorded. Jelks (1963, 1965:101) believed that the midden "evidently accumulated around a small cluster of buildings over a period of several decades -or possibly several centuries" and "probably a sequence of different buildings stood on the spot through time, new ones being built to replace those that burned or fell into disrepair." A single calibrated 1 sigma radiocarbon date of A.D. 1405-1510 (p=0.92) was obtained from the site.

Harold Williams (41CP10)
The Harold Williams site (41CP10) consists of three areas, A-C. Area A included three burials while Area B consisted of a midden area 75 x 60 m in size, a Titus phase cemetery with 25 individuals, and a single oval structure (Turner and Smith 2002). The Area C excavations identified lithics, ceramics, charcoal fragments and 39 pieces of daub, indicating a light occupation in this part of the site (Turner and Smith 2002). The midden in Area B contained "a dense concentration of daub, probably from a burned Caddoan house, along with ceramics, lithic debris, arrowpoints, animal bone, and charred plant remains" (Turner and Smith 2002:19). Multiple structures were likely present at the site through time and Turner and Smith (2002) noted that the large quantities of daub associated with the Middle Caddo component at the site illustrates that structures were likely present during that time. Additionally, the Middle Caddo infant interments and offerings were likely "within or immediately adjacent to a house, which is a Caddoan burial custom" (Turner and Smith 2002:65). This being said, however, only a single structure, likely built during the Titus phase, was recorded in Area B (Turner and Smith 2002;Woodall 1967).
The Titus phase structure, Feature 7, measured 6.7 x 5.0 m and was associated with a concentration of daub ( Figure 67) (Thurmond 1990;Turner and Smith 2002;Woodall 1967). The circular to oval structure included at least 17 or 18 recorded post holes with average diameters of 20 cm (Woodall 1967), and from the plan map, it appears as though at least 11 of these formed part of the circular structure outline. Woodall (1967:10) stated that "unfortunately, recent erosion has cut through this portion of the site, removing any additional evidence such as a house floor or hearth." Thurmond later noted that "Woodall makes no note of artifact association, so the chronological placement of the structure is also undeterminable" (Thurmond 1990:146). However, based on their reanalysis of the recovered artifacts, Turner and Smith (2002:65) have argued that "based on the artifacts from within and adjacent to the structure, the Feature 7 house probably was built and used by Titus phase peoples."

Roitsch (41RR16)
Three rectangular structures were found at the Roitsch (Sam Kaufman) site (41RR16) during excavations by Skinner et al. (1969). Two of these structures (Structures 1 and 3) were located in the village area while Structure 2 was associated with the east mound and was discussed in Chapter 5.  (Skinner et al. 1969:19).

Roitsch
Structure 3 was defined by a layer of burned wood, bark, charcoal, and daub identified to the east of House 2 in a backhoe trench (Skinner et al. 1969:21) (Skinner et al. 1969: 21), or the remnants of the burned superstructure. Burials 3, 4, and 5 are located directly to the southwest of Structure 3, all less than 5 m from the structure entrance (Skinner et al. 1969).
As noted in Chapter 5, the construction of the East Mound and the use of  Table 42 (Bruseth 1998;Perttula 1997;Skinner et al. 1969). 1400-1650) site situated just west of the Neches River (Jackson 1936;Kleinschmidt 1982Kleinschmidt , 1984. Several features were found below the midden, including a large circular structure (Feature 3) (Jackson 1936; Kleinschmidt 1982Kleinschmidt , 1984. was not sufficient enough to convincingly argue for their association with this structure and they may have been associated with a separate occupational event (Kleinschmidt 1982).
Based on the plan map (Kleinschmidt 1982), interior, two were recorded along the structure wall (Features 5 and 6), in a location that seems to suggests they were likely not in use while the structure was standing, and two were recorded in the immediate vicinity of the structure (Features 7 and 8), but outside the structure walls (Kleinschmidt 1982). Kleinschmidt's analysis of the Saunders archeological remains suggests that this structure was built during the sixteenth century (Kleinschmidt 1982;Perttula 2004a) 225 Figure 68. Feature 3 from the A. C. Saunders site (Kleinschmidt 1982).

Spider Knoll (41DT11)
The Early to Middle Caddo period Spider Knoll site (41DT11) consisted of three distinct activity areas, including an area primarily used for trash disposal (south edge of the site), an area used for outside activities (eastern half of the site), and the north portion of the site, used for outside activities and for the construction and use of possible residences or shelters (Fields et al. 1997). Several pits, post holes, hearths, and a spatially distinct midden were found at the site.

226
The postholes likely corresponded to multiple arbors and drying racks as well as to at least two possible houses. Fields et al. (1997:59) noted that "the postholes do not form patterns that can be interpreted unquestionably as representing houses. Yet, many of the postholes can be grouped into arcs suggestive of structural walls. The two longest arcs, both encompassing semicircular arcs, account for 13 of these postholes, and it is hypothesized that these represent two houses." They further suggested that the possible structures may be open structures rather than traditional fully enclosed house (Fields et al, 1997:59).
Structure 1, the western structure, was a semi-circle consisting of six post holes with diameters ranging from 10 to 23 cm (x = 15.3, σ = 4.2) (Fields et al. 1997:59). If the arc represented part of a circular structure, then the possible structure had a maximum diameter of about 8 m with post holes spaced from 2.1 to 3.9 m apart ( Structure 2, the eastern structure, had seven post holes with diameters ranging from 9 to 21 cm ( x = 2.8, σ = 0.7) (Fields et al. 1997:59). A pit was located within the Structure 1 outline.
x = 15.9, σ = 3.5) (Fields et al. 1997:59). Based on the arc of post holes, this possible circular structure could have had a maximum diameter of about 10 m, with post holes spaced from 2 to 3.7 m apart ( A hearth and two pits were situated near the center of Structure 2, possibly representing a central hearth. Fields et al. (1997:59) noted that "the hearth central to the eastern arc may well be associated with this possible structure given the correspondence between the corrected radiocarbon assays from this hearth and one of the postholes in the arc (1065 + 55 and 1095 + 55 B.P., respectively)." In addition to the central hearth and pits, a second hearth and small pit were situated along the structure's eastern side, a small pit was along the structure's northern edge, and an additional pit was located to the x = 2.7, σ = 0.7) (Fields et al. 1997:59).
227 southwest of the structure. "The hearth along the eastern edge probably is not associated given its radiocarbon assay of 850 + 90 B.P., and it may represent use of this area for outside activities associated with occupation of the western structure" (Fields et al. 1997:59).
The Spider Knoll structures may be similar to the structures found at the Sanders site (see Chapter 5), although those from the Sanders site were situated on a mound platform. Fields et al. (1997:61) summarized the possible Spider Knoll structures as follows: The incomplete post hole patterns, small post hole size, and wide spacing between the post holes indicate that, if the arcs in the western subarea do represent structures, these were not the sorts of substantial, permanently occupied houses typically found at Caddoan residential sites. Rather, they may have been more akin to large arbors or sun/wind screens that were used seasonally.

Deshazo (41NA27)
The Deshazo site (41NA27) was divided by Story (1982:35)  Each of the fully exposed structures had evidence of a center post, an exterior wall with between 20-24 posts, and a diameter ranging from 9.0-12.2 m. No prepared floors were found associated with any of the structures; in fact, excavations were unable to determine floor levels for individual structures due largely to past plowing and erosion at the site (Good 1982:53).
Structures 1, 2, and 3 were identified in excavation unit 1 in Area D along with two burials (Burials 11 and 12) and three hearths (Hearths 1, 2, and 3). All three structures were circular with diameters of 12.2 m (Structure 1), 9.4 m (Structure 2), and 9.8 m (Structure 3). Each structure had a center post and the post hole patterns of each closely overlapped the other structures.
Although the hearths were located within Structures 1, 2, and 3, Good (1982:71-72) was unable to associate the hearths with individual houses. Despite this, Good (1982:73) suggested that Hearth 1 postdated the construction of Structure 1, although the relationship between Hearth 1 and the other two structures was not determined. Given its location, Hearth 2 was probably associated with either Structure 1 or Structure 2 and would have been located between the exterior wall and the center post (Good 1982:76).
Finally, Hearth 3 was located outside of the exterior walls of the three structures. Hearth 3 had a post hole beneath it, however, although no associated structure was identified (Good 1982:76-78). In addition to the three hearths, Burials 11 and 12 probably "represent juvenile interments in shallow, basin-shaped graves associated with structures" (Good 1982:88). Like the other features from Unit 1, it was difficult to associate the burials with individual structures. Both burials 11 and 12 were found in shallow, oval, basin-shaped pits near the wall of Structure 3, presumably buried within or under the floor of the structure (Good 1982:90).
Three overlapping circular structures (Structures 4-6) were identified north of Structures 1-3 in Units 3 and 11. More than 113 features were identified in this area that corresponded to the three structures as well as additional outlying features interpreted as possible ramadas or storage platforms. Structures 4-6 measured 9.6 m, 9.6 m, and 11.9 m in diameter, respectively. No prepared floors, hearths, or midden areas were found in association with these structures (Good 1982).
Structure 7 measured 9.0 m in diameter and was identified southeast of structures 1-3. This structure had several associated interior features, including an "arc of disturbances northwest of the center post lying about 1.5 m inside the arc of the exterior support posts, and the cluster of disturbances southwest of the center post close to the wall of the structure" (Good 1982:55). These interior features likely represent support posts or supports for interior shelves, racks or furnishings (Good 1982).
A single structure was identified on the same side of the creek as the Historic Caddo cemetery. Structure 8 consisted of an arc of five post holes (Good 1982:55). The structure was not further excavated due to disturbances prior to the 1976 excavations.
Based on the arc of five post holes, Structure 8 had an estimated diameter of 9.2 m.
Structure 9 consisted of 10 post holes forming an arc that represented approximately half of a circular structure. Structure 9 was approximately 9 m in diameter with an area of approximately 63.59 m 2 . No internal structural features were identified in the limited excavations that only exposed a portion of the outer wall and not the interior of the structure. Burial 13, "an apparent juvenile burial," was "located against the southeast wall of the structure" (Good 1982:55).
Summary post hole data for the wall post holes associated with each structure are provided in Table 43 Good (1982:63-64) provided measurements for distances between wall post holes and from the center posts to the wall (Table 44). The wall post holes for the nine structures had a mean spacing ranging from 1.197 to 1.71 m, a much larger spacing than that seen with structures from other sites. The mean distances from the center to the wall posts for the nine Deshazo structures ranged from 4.6 m to 6.1 m. While the mean spacing between posts at the Deshazo site was larger than that seen at other East Texas Caddo sites, the 231 ratio of the average distances between exterior post holes and average distance to center post holes was actually smaller than that seen in most other structures in this study. n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a 9 10 0.14 0.40 0.25 0.0749 0.26 0.48 0.38 0.058 n/a n/a Good (1982) explains the similarity between the ratios of these distances by arguing that in building the structures, the center was first established, followed by establishing the radius of the structure, using that measurement for the placement of all the wall post holes. Good (1982:64) suggests that "the distance between the exterior postholes was established at a little better than 1/3 the radius, regardless of the length of the radius". Good (1982:64) does not point towards the use of a standard unit of measurement, but suggests that the process was repeated for each structure independently. Clearly, as Good (1982:64) argues, the Caddo builders recognized the "spacing and number of the exterior supports required for structural integrity in relation to the size of the structure. Having this understanding would allow shamans to designate precisely the number of support posts required for a particular structure" when they sent 232 the tammas to order members of the community to prepare poles for the construction of building. This in turn suggests that "initial measurements were made and spacing determined well in advance of actual construction" (Good 1982:65). At Deshazo, it is clear that structures were rebuilt using the same location in many instances. The sequence of structures in each unit was contemporaneous and structure replacement "was determined by the rate of deterioration of the structure materials" (Good 1982:69) rather than increases in population. Based on records from the Texas Forest Service, the support timbers for the Deshazo structures would have remained sound for at least 10 years and possibly up to 30 years, with an average lifespan of around 20 years (Good 1982:69).

New Hope (41FK107)
A single structure and evidence for associated activity areas were recorded at the New Hope site (41FK107) in Franklin County (Figure 70), in the Big Cypress Creek basin. Although only nine post holes were recorded, Nelson and Perttula (2006) suggested that the circular structure measured approximately 8. Radiocarbon dates were obtained from charred hickory nutshells (Carya sp.) from the hearth fill (Nelson and Perttula 2006). These produced a 2 sigma date of cal. AD 1280-1420 (Nelson and Perttula 2006:30). Based on the radiocarbon date from the hearth as well as the predominantly Middle Caddo ceramic assemblage recovered at the site (Nelson and Perttula 2006:31-33), Nelson and Perttula (2006) concluded that the  (Nelson and Perttula 2006:30). Reproduced courtesy of the Friends of Northeast Texas Archaeology.

Walter Bell 41SB50
At least two and likely three circular structures were excavated at the Walter Bell site (41SB50) prior to the construction of Lake Sam Rayburn on the Angelina River.

235
Structures 1 and 2 were each represented by a ring of post holes indicating that they were built of individually set vertical posts (Jelks 1965). Structure 3 was identified as an arc of posts within the Structure 2 outline and was interpreted by Jelks (1965) as a separate structure. Both Structures 1 and 2 were interpreted as having been of wattle-anddaub construction based on the recovery of wattle-impressed daub fragments (Jelks 1965).  Jelks (1965:56) noted that a gap on the west side may have been indicative of a non-extended entrance that was 2.1 m wide, although it may be a gap in identified post holes, given that two other gaps were also present on the southeast and 236 northwest sides that measured 1.46 and 1.2 m wide, respectively. For Jelks (1965:56), it was difficult to discern the exact number of posts that made up the perimeter of the structure, but it probably had between 30 and 40 posts. Seven post holes were identified on the structure interior, but no hearth was found. The floor of the structure was likely in the midden level immediately above where the structure was first identified, "probably not far from the modern surface of the ground" (Jelks 1965:56). This explains the shallowness of the perimeter poles since none were deeper than 0.15 m below the dark midden zone (Zone 2) that extended to about 0.15-0.30 m bs (Jelks 1965).  (Jelks 1965:58). No definite entrance was identified and the posts tended to be spaced from 0.30 to 0.61 m apart with the exception of three large gaps, one on the northeast side and two adjacent gaps on the northwest side, measuring about 1.5 m, 1.68 m and 1.68 m wide respectively (Jelks 1965:58). No prepared floor was found, but "the floor of House 2 was probably at or above the modern surface of the ground as no hard-packed zone or other definite indication of a floor was found. Perhaps the original floor was disrupted by the plow" (Jelks 1965:58). The structure had at least two large interior posts, each measuring about 0.31 m in diameter, one of which was a center post. An arc of six post holes (Structure 3) cut across the southwest portion of Structure 2. "The molds were comparable in size, depth and spacing to those of Houses 1 and 2, therefore they probably represent a third house" (Jelks 1965:60). Based on the arc, Structure 3 probably measured 12.2 m or more in diameter.
The semi-flexed burial of a middle-aged male, Burial 4, was located on the east side of Structure 2, "at a spot where the wall of House 3 passed over one end of the grave" (Jelks 1965:59, 64-65). It was not possible to ascertain "whether the burial was earlier or later than the house or if the association between the two was accidental or intentional" (Jelks 1965:59).

Rookery Ridge (41UR133)
Two Late Caddo Titus phase circular structures were found at the Rookery Ridge   (Parsons, n.d.).
Structure 1 appears to have had a 0.4-0.5 m high earthen berm around at least a portion of the structure (Parsons, personal communication 2009), similar to those found associated with mound structures at Bryan Hardy and Morse Mounds site (Middlebrook, personal communication 2009;Parsons, personal communication 2009;Walters and Haskins 2000). Although a berm is not shown in the plan map (see Figure 73), and was not clearly identified during the excavations, Parsons (personal communication 2009) argues that the placement of Burial 1a suggests that a berm was in place when the child was interred. If Structure 1 at Rookery Ridge was associated with an earthen mound, in addition to a berm, evidence of the mound has long since been plowed away or eroded (Parsons, personal communication 2009).
Structure 1 had several associated features, including multiple smudge pits, possible cooking pits, and other pits of undetermined use. There were at least three and possibly four smudge pits within the structure as well as three additional hearths (Parsons, personal communication 2009). Most of these features are located opposite the entrance, in the southern half of the structure (see Figure 73). Parsons (personal communication 2009) suggests that the area between the central hearth and the entrance may have been kept cleared to allow for easy passage in this area and possibly to allow for ceremonies, meetings, etc., related to its possible specialized function. Interestingly, the structure was only partially burned as indicated by Feature 10, an area of burned structural members and charcoal covering an area of approximately 6.5 m 2 , and an area that corresponded with that portion of the structure with most of the smudge pits (with the exception of Feature 11) (Parsons, personal communication 2009).
Perhaps the fire was accidental and not necessarily associated with the deliberate destruction of the building (Parsons, personal communication 2009).
Two child burials (Burials 1a and 1b), estimated to have been about 18 to 24 months of age at death, were located immediately to the west of Structure 1, approximately 50-60 cm from the structure wall (Parsons 2002).  These observations, according to Parsons (2009 personal communication) fit those recorded in Spanish accounts that indicate that "the houses of important men were "better made through having more rib-poles than others" (Bolton 1987:113). This argument 243 lends further support to Parsons' suggestion that Structure 1 was a specialized structure while Structure 2 was perhaps a domicile for a non-elite member of the community.

Lang Pasture Site (41AN38)
Two structures and two possible ramadas or arbors were recorded at the Lang associated with Structure 2 fits with the size difference between the two structures, in that "Structure 2 is 32% larger in diameter than Structure 1, and presumably needed larger poles to form the taller peaked roof" (Perttula et al. 2009:5-15). Furthermore, the wide range in post hole size and depth, based on the coefficient of variation values (21.66 and 18.34, respectively), suggested that poles were chosen less selectively and that they pointed to "somewhat haphazard construction efforts" associated with Structure 2 when contrasted with Structure 1 (Perttula et al. 2009:5-15). The "deliberate selection of larger posts associated with Structure 2 and its surrounding outdoor activity areas raises the possibility that timbers available for construction at the time of the Structure 2 occupation were naturally larger, and that by the time of the Structure 1 occupation (which presumably came slightly later), the larger timber had been culled, at least to the extent where it affected timber selection choices in the construction of Structure 1 and its associated Structure 3." Given that the post hole pattern for Structure 2 is so incomplete, I have not measured the distances between wall posts or the distances from the wall to the center. Perttula et al. (2009:5-15), however, noted that "the Structure 2  located within the outdoor activity areas (Perttula et al. 2009). Pits included large storage pits, most ranging from 132 x 70 cm to 118 x 148 cm in length and width, with depths ranging from 38 to 94 cm bs (Perttula et al. 2009); small storage pits, with diameters less than 80 cm and depths as shallow as 30 cm bs; cooking pits; and four smudge pits. The largest pit, Feature 76, measured 360 x 210 cm in length and width, with a depth of 135 cm bs. With an estimated volume of 7.18 m 2 , this pit could have held enough maize "to meet the dietary requirements of 8-9 families for a year, suggesting that it may have been a shared multi-household storage feature. This also suggests that the Lang Pasture site was occupied by multiple Caddo households" (Perttula et al. 2009). This large pit feature was also used for a child or juvenile burial after its use as a storage pit, with the burial pit being dug into the pit floor.
Other features from the Lang Pasture site include a midden deposit about 10 m from the excavated area and a family household cemetery with at least nine burials.
Artifact distributions at the site, especially ceramic vessel sherds and pipe fragments, indicate that the outdoor activity areas played an important role in the daily lives of the Caddo inhabitants (Perttula et al. 2009). Most ceramic vessel sherds and pipe fragments were found discarded in the activity areas. The low densities of sherds within the structure walls suggests that they were swept to gather up broken vessels and vessel fragments that had been previously broken" (Perttula et al. 2009:5-61). The cleaning of the Structure 1 floor could also help explain why so few artifacts accumulated in the post hole fill (Perttula et al. 2009).
The Lang Pasture site dates to the late 14th to early 15th century A.D. (Perttula et al. 2009). Perttula et al. (2009:5-59) suggests that the features from the Lang Pasture site represent a household compound in which "individual structures were grouped together 247 into a small compound arrayed around an open courtyard, with outdoor activity areas and ancillary facilities (i.e., ramadas, granaries, or other sorts of elevated platforms) and shared trash/midden disposal areas." Interestingly, Perttula et al. (2009:5-59) concludes that the high number of large storage pits at the site "suggests that above-ground granaries were not at use at the Lang Pasture site during the prehistoric Caddo occupation."

Hatchel (41BW3)
The Based on recent work at the Hatchel site, Perttula (2005) identified at least five different village areas to the south, southeast, and east of the Hatchel platform mound.
While no new structures have yet been recorded in these areas, it is likely that "these villages areas represent different individual compounds or sets of related compounds of Caddo families and/or extended families that contained grass-covered house structure, above-ground granaries, and outdoor ramadas or arbors" (Perttula 2005:3), much like those seen in the 1691 Terán map, and found elsewhere at the Hatchel site Perttula et al. 2008).
Post hole patterns for at least three structures (F-1, F-1a, and F-2) were recorded in the Hatchel Village excavations of Village Plot 1. F-1 represented at least two overlapping circular structures, the first measuring approximately 9.5 m in diameter (F-1a), with an area of approximately 70.85 m 2 , and the second measuring approximately 7.14 m in diameter (F-1) (recorded by the WPA as measuring approximately 7.3 m by 6.6 m), with an area of approximately 40.02 m 2 . The smaller structure was recorded as F-1 by the WPA, who did not recognize the larger arc of post holes as a structure, with the larger structure (F-1a identified by Perttula [2005]). The exterior wall of neither structure was well defined (Perttula 2005). The smaller structure was represented by as many as 20 or 21 exterior wall post holes while the larger structure had 30 to 34 wall post holes. Several interior post holes were recorded, as were three larger support posts or possible pits (Perttula 2005). Unfortunately, however, it was not clear which interior features may have been associated with either structure (Perttula 2005). There was no clear entrance for either structure, although the WPA plan suggested the presence of a north-facing extended entrance for the smaller structure. As Perttula (2005:186) noted, however, "the post hole pattern of the entranceway in plan maps and photographs is not very compelling." The larger structure likely predated the smaller structure and the "smaller version of F-1 probably represents a rebuilding of the larger structure in approximately the same 249 place." (Perttula 2005:186) The larger structure did clearly predate Feature 2, "as the walls of the larger structure cut across the central part of this smaller structure" (Perttula 2005:186). Whereas F-1 probably represented a domicile, given its small size and direct association with F-1, F-2 was likely a granary (Perttula 2005), similar to those recognized at the Oak Hill Village (Perttula and Rogers 2007;Rogers and Perttula 2004) and George C. Davis sites.
As discussed in Chapter 5, the Hatchel site was occupied from as early as ca. A.D.
1000 to as late as the mid-eighteenth century (Creel 1996;Perttula 2005). While there are no dates from the specific village structures discussed above, in addition to the four radiocarbon dates from the Hatchel Mound (Perttula 1997) discussed earlier, five radiocarbon dates and 30 OCR dates have been obtained from Village areas of the site (Perttula 1997(Perttula , 2005. As noted by Perttula (2005:184-185), the dates from Area II of the village provide "solid evidence for the settlement of the Hatchel site before ca. A.D.   = 3.76 m, s = 0.049). The ratio of average distance between wall post holes and average distance to the center was 1:6.9.

Hurricane Hill (41HP106)
The Hurricane Hill Site on the South Sulphur river is divided into three areas, the North, South, and Southwest rises (Perttula 1999). Several midden areas, multiple burials, and over 225 features, including pits, hearths, and post holes, were excavated at the site.  77) (Perttula 1999(Perttula , 2004.  (Perttula 1999, Figures 6-21 and 6-22). Reproduced courtesy of the Friends of Northeast Texas Archaeology.
Features recorded on the South Rise included 98 post holes or probable post holes, 23 pits, two partly overlapping hearths (central hearths of Structures A and B), and burials, including three human and one dog burial (Fields et al. 1997;Perttula 1999). In addition, there were three middens that were likely associated with superimposed Structures A and B, and one midden to the west of Structure C (Fields et al. 1997;Perttula 1999).
While complete wall patterns were not evident for the South Rise structures, enough of a pattern was evident for Perttula (1999) to define the later Structure A as measuring 6 x 7.5 m with an east facing non-extended entrance and the earlier, and larger, Structure B as measuring 7.1 x 7.6 m with a possible south-facing extended entrance. Perttula (1999:106) noted that "the entrance was marked by two parallel stains or trenches," and that "although the eastern trench lacks postholes, the fact that the western trench has postholes at either end strongly suggest that both trenches did have wooden posts in them to make a covered entranceway." Large interior support posts were present in each structure and were arranged in a somewhat square pattern similar in form to the four support post structures found at the George C. Davis site and other East Texas Caddo sites (Creel et al. 2005;Fields et al. 1998;Perttula 1999;Perttula and Sherman 2009). In addition, the partially overlapping hearths suggested that the structures were used sequentially.
Structure C was recorded as a partly exposed 6 m diameter circular structure. Perttula (1999:110) thought that Structure C may represent an arbor or ramada rather than a domicile. center post (situated below Feature 5B, the Structure B hearth) had a diameter of 39 cm (Perttula 1999).
In addition to interior support posts, both Structures A and B had superimposed central hearths (Features 5A and 5B). The Structure A hearth (Feature 5A) was the later of the two hearths and extended from 16-28 cm bs (Perttula 1999). Feature 5B extended from 31-42 cm bs (Perttula 1999). An archaeomagnetic date of A.D. 1300 was obtained from the Structure A hearth basin walls (Perttula 1999).
There was also a 15 cm thick pit associated with the interior of Structure B (Feature 12) and two pits (Features 30 and 38) were associated with the interior of Structure A. Additional large pits were situated outside of the structures (Perttula 1999). Two partly exposed rectangular structures (Structures D and E) were recorded on the North Rise. These appeared "to be of Early Caddoan affiliations, based on the archaeological content of the soil zones and midden deposits in proximity to the posthole patterns" (Perttula 1999:94). Structure D, the later of these two Early Caddo structures, was partly delineated by 15 outer wall post holes (Perttula 1999:96) Structure D was interpreted as a rectangular structure oriented with magnetic north (Perttula 1999:96). The only non-post hole feature identified in the interior of the structure was a 16 cm deep basin-shaped rock hearth containing large amounts of firecracked rock (Perttula 1996:94-96). Rather than being primarily a cooking feature, Perttula (1999:94) suggested that the rock hearth "was probably used more as a source of radiating heat." 261 Structure E was delineated by 10 post holes corresponding to nine wall post holes and one feature representing either a center post or a small pit (Perttula 1999:98). The exposed portions of the structure measured 5 m along one wall and slightly less along an intersecting wall. Based on the exposed wall segments, Perttula (1999:98) (Perttula 1999). Reproduced courtesy of the Friends of Northeast Texas Archaeology.

Hill Farm (41BW169)
The Hill Farm Site (41BW169) is a Late Caddo site occupied from approximately A.D. 1500 into the late seventeenth century Walker 2009). The site is considered part of the surrounding Hatchel site, which is a component of the larger Hatchel-Mitchell-Moores site complex occupied from approximately the 13 th century A.D. into the eighteenth century (Perttula 2005;Perttula et al. 2008;Walker 2009).
Archaeogeophysical investigations at the Hill Farm site revealed multiple Caddo structures and associated features as well as several relic meander scars of the nearby Red River, the later helping in planning field investigations as well as assisting with the interpretation of the site's place within the broader Hatchel village (Grealy and Conyers 2008;Perttula et al. 2008;Sundermeyer et al. 2008;Walker 2009 (Schultz and Walker 2006;Perttula et al. 2008;Walker 2009;Walker and Schultz 2008). For a more detailed discussion of the collection methods, data processing, and archaeogeophysical interpretations, the reader is referred to Perttula et al. (2008), Walker (2009), and Walker and Schultz (2008).
At least 10 and possibly 11 circular Caddo structures were identified in the magnetometer data from the Hill Farm site (Table 46) Schultz and Walker 2006;Walker 2009;Walker and Schultz 2008). As Walker (2009:135-154) discusses, while none of the Hill Farm structures have been ground truthed through excavations, associated shovel test data as well as the broader understanding of the nature of the archaeological remains from the Hatchel site lends support to the use of the Hill Farm dataset as primary archaeological data in assessing not only types of features and architectural styles associated with the site, but the broader organization of space and settlement dynamics at a sixteenth to seventeenth century A.D. Nasoni Caddo Village (see also Perttula et al. 2008). 103.82 y y n/a n/a n/a n/a n/a 2 c 15.00 176.63 y n/a n/a n/a n/a n/a 3 c 7.00 38.47 n/a n/a n/a n/a n/a feature is large di-pole; burned structure or large borrow pit 4 c 14.5-17.5 165.05-240.41 y n/a n/a n/a n/a n/a 5 c 12.00 113.04 n/a e n/a n/a nw patterns of returns w/in structure may be small storage pits 6 c 10.00 78.50 y n/a n/a n/a n/a n/a 7 c 8.00 50.24 y n/a n/a n/a n/a n/a 8 c 10.00 78.50 y n/a n/a n/a n/a n/a possible structure 9 c 10.00 78.50 y y n/a e n/a n/a se structure appears to have been burned 11 c 10.00 78.50 y y n/a n/a n/a n/a n/a structure appears to have been burned 13 c 8.00 50.24 n/a n/a n/a n/a n/a possible structure The Hill Farm structures have diameter ranging from 7 m to 17.5 m, with a mean diameter of 10.82 m. These structures likely represent residential or special purpose buildings, along with at least one possible granary (Feature 3) Schultz and Walker 2006;Walker 2009)   and 11 appear to have been burned Walker 2009;Walker and Schultz 2008). Feature 9 has a southeast-facing extended entrance that, as Perttula et al. (2008) and Walker (2009) note, may have been oriented to face towards Area A. Figure   80 provides an image of the magnetometer results from Area B while Figure 81 shows the interpretation of the features. 268 In addition to the two structures, two linear patterns of magnetic returns, Features 10 and 12, were recorded in Area B Walker 2009). Given their locations on either side of Feature 11, Perttula et al. (2008) and Walker (2009)    The research conducted at the Hill Farm Site is an excellent example of the potential contribution to Caddo archaeology, and archaeology as a whole, that can occur when archaeogeophysical data can be used as primary data Walker 2009). The domestic deposits at the Hill Farm site appear to date to the seventeenth century A.D. Caddo occupation in this part of the larger Hatchel village . In fact, Perttula et al. (2008:102)

Foggy Fork (41NA235)
In discussing house construction at five sites investigated as part of the Lake Naconiche project in the Attoyac Bayou basin,  stated: there is evidence for wood structures at all five Lake Naconiche sites, although because of poor preservation, the difficulty in identifying post hole stains in sandy soil, and structure rebuilding, the structure post hole patterns were not always clearly defined. The structures were built primarily with oak posts or poles well set in the ground (especially the center post and any main support posts), and were likely covered with thatch, sticks, and certain grasses; the general absence of daub suggests that the Lake Naconiche Caddo structures did not have a clay covering over the thatch. As best as can be discerned, the structures are circular in shape, with "the outer walls [following] very regular circles that must have been scribed before actual construction, perhaps by using a peg tethered by a rope to a pole" (Story 1995:239).
While there are clearly structure areas at the Tallow   Six interior post holes were recorded, three of which measured about 18.17 + 3.87 cm in diameter . These interior post holes may represent either interior structure supports or possibly supports for small interior screens or dividers   .
Three post holes, measuring 18 + 1.33 cm in diameter, found immediately to the south and east of Structure 1 were likely used for outdoor screens or racks . A 1 m diameter hearth and a basin-shaped pit measuring 1.1 m in diameter were recorded well outside of Structure 1 and a small midden deposit was recorded to the south of the structure (see Figure 84).

Size and Shape
The majority of both non-mound and mound-related Caddo structures are circular; however, sub-round, sub-square, and rectangular structures occur in the area as well.
While I have classified the structures in this study as falling into one of four general shapes, it should be noted that there is considerable variation within each of these shape categories. For instance, some of the sub-square structures are nearly rectangular in form, while others appear nearly sub-round.

276
The    (Creel et al. 2008;Hatcher 1927;Walker 2009), or possibly fences similar to the large, 25 m, fence found at the Hardman site encircling living areas (Early 1993). Alternatively these features could represent defined spaces such as dance-grounds or other types of ritual space. However, it is entirely possible that these large features recorded in the George C. Davis magnetometer data represent large domiciles, meeting or assembly houses, or other types of structures.    generally less than 3 m (Rogers and Perttula 2004;Spock 1977), and are associated with other buildings. Figure 97: Histogram comparing interior areas of mound and non-mound related structures by shape.

Summary
The majority of Caddo structures in eastern Texas, whether associated with earthen mounds or not, are circular (n = 223). These structures range in size from 1.  Creel et al. (2008) and Walker (2009), some of these exceptionally large structures (relative to the average structure size for the area) 292 may represent open enclosures rather than roofed buildings. Alternatively, these larger structures may have served as public buildings or assembly houses. The large circular structure from the Werner site measured approximately 24.4-27.4 m in diameter (Webb 1983), approximately 3.5 to 6.5 m in diameter larger than the largest of the George C.
Davis magnetometer structures, and about 5.5 to 9.5 m larger than any of the previously excavated structures in this study. The Werner structure was interpreted by Webb (1983:221) as being "an immense lodge, presumably ceremonial; the destruction of which was memorialized (Caddoan fashion) by the covering mound." The smaller inner circle of the Werner structure was interpreted as being a ring of roof supports.

Wall Trench, Single Set Posts, and Composite
Only six of the 213 (

Entrances
The majority of non-mound structures included in this study have non-extended entrances. Only 14 to 18 of the non-mound structures (6.5-8.4%) have extended entrances. Four of the extended entrance structures were recorded in the magnetometer surveys at Hill Farm and the George C. Davis site (Table 49). approximately 1.9 m, with a mean width of 1.02 m. Of the 15 extended entrance mound structures, one faces northeast, one faces south, five face southeast, four face southwest, three face to the west, and one faces east.
As discussed in Chapter 3, Hidalgo (Hatcher 1927) and others noted that the Caddo in historic times placed entrances to their houses on the east or west. In the archaeological dataset, only 22% face either east or west (see Table 50) These data suggest that the direction these structure entrances faced was largely dependent upon other features on the landscape such as the village plaza, earthen mound, or possibly even a mound at another site (Perttula 2009), as well as varying beliefs about the significance of directions in the Caddo cosmos (Kay and Sabo 2006;Perttula 2009). For example, most of the Hatchel structures with extended entrances faced in the general direction of the village, rather than to the west or east.
Construction technique for non-mound entrances varies from single set posts to post trenches. Of the 18 non-mound entrances, 33% were constructed using post trenches, 33% were constructed without post trenches and the remainder (identified in the remote sensing) unknown given that they have not been excavated. Mean trench width for nonmound structures with entrance trenches is 1.12 m, while the mean width for the singleset post entrances is 0.81 m. Mean length for the two categories is similar as well, with post trench entrances extending to a mean length of 2.7 m and single set post entrances extending to a mean length of 2.45 m. In addition, some of the non-mound entrances have special treatment, such as the Feature 9 entrance from the George C. Davis site, where the entrance was "paved with a layer of grayish-white, hard packed clay which measured 2.14 m long, 0.46 m to 0.61 m wide and was raised 0.31 m above the structure's rim. A second layer, 9 cm thick, and consisting of compact reddish clay, overlay most of the grayish-white clay" (Spock 1977:47). In addition, the Earspool Structure 1 entrance was slightly sloped, indicating that in order to enter the structure one would have walked downward from the ground surface into the structure (Perttula and Sherman 2009). With the exception of structures with extended entrances it is difficult to determine with absolute certainty where a structure entrance is located in most cases.
Non-extended entrances could be discerned from gaps in wall patterns or from locations of associated middens or activity areas. As Good (1982) noted at Deshazo, and as is evident in Soule's photo discussed in Chapter 3, non-extended entrances may not always be as easily identifiable as a gap in the structure wall. One of the structures in Soule's photo shows the entrance "raised off the ground, perhaps a foot, with the stepover silled.

302
The stepover appears to consist of a row of short supports topped by the sill" (Good 1982:62). Figure 101 shows Mean exterior wall post spacing for mound-related structures indicates a tendency for posts to be slightly more closely set. The mean ranges from 0.26 m (Structure 2 from the Keith site) to 0.90 m (Structure 1 from Sanders). As discussed above, however, the Sanders structure appears to be a different type than the other mound-related structures, referred to by Jackson (2000) as a lean-to, or an open-air structure, perhaps similar to the Spider Knoll structures described by Fields et al. (1997). The range for mean exterior wall spacing for mound-related structures excluding the Sanders structure is slightly smaller, ranging from approximately 0.26 m to 0.86 m (Hatchel Structure 5).

Summary
The majority of non-mound wall posts in the dataset tend to be spaced between 0.40 m and 0.70 m, with 70 (68%) of the 103 structures with these measurements having mean post spacing in this range (Table 51) Figure   105). The ratio of average distance between exterior post holes and from the exterior post holes to the center of the structure ranges from 1:1.14 (Feature 7 from the George C.

Mean Distance Between Wall Posts
Davis site) to 1:33.12 (Feature 9 from the George C. Davis site) for non-mound structures (Table 52).
x = 3.97, s = 1.58) (Figure 106). The ratio of average distance between exterior post holes and from the exterior post holes to the center of the structure ranges from 1:2.05 (Hatchel Structure 5) to 1:18.15 (Feature 45   308 from the George C. Davis site) for mound-related structures (Table 53). Figures 107 and   108 are graphical representations of this measurement by structure shape for the nonmound and mound structures. Figures 109 and 110 are scatterplots of post hole spacing plotted with distance from wall posts to structure center. In addition, Figure 111 is a figure of the ratio for all the structures in this study and Figure 112 is a comparison of the ratios for the mound and non-mound structures.   Figure 106. Comparison of mean distance from wall post holes to center of structure by structure shape for mound-related structures. Table 53. Ratio of the mean distance between wall post holes and from the wall post holes to the structure center for mound-related structures. c=circular; ss=sub-square; sr=sub-round; r=rectangular. The ratio of average distance between exterior post holes and from the exterior post holes to the center of the structure was obtained for 117 structures. These structures are all circular, sub-round, or sub-square. The ratio ranges from 1:1.14 to 1:33.12, although the large ratio is from Structure 9 at the George C. Davis site and is considerably larger than the ratio for the remaining structures. For Early to Middle Caddo period structures, the ratio ranges from 1:1.14 to 1:33.  (Perttula and Sherman 2009), and may be a useful tool for examining architecture within a specific community.

Pits and Berms
Three non-mound structures were built within pits and at least one had evidence Gipson 1960). Each of these structures also had an extended entrance. These entrances, as noted by Jelks and Tunnell (1959) Brennan (2007:77) suggested "the excavated fill is then piled against the walls, functioning structurally to strengthen them in a manner similar to daub." Either way, a sand embankment around the structure would have served to give the structure a semi-subterranean appearance, thus further separating these specialized structures from the domiciles and other non-mound structures in a village or mound center. Parsons (2009 personal communication) has also suggested that a berm placed around a structure may be seen as an incipient mound. As such, berms may be symbolically related to mounds and the structures associated with them may have been afforded the same symbolic significance as mound structures. If berms can be considered incipient mounds, then upon entering a structure surrounded by an earthen berm, one would have been entering the mound itself.

Prepared Floors
Prepared clay floors were recorded at only four of the non-mound structures included in this study (Table 54). Three of these structures were also built within pits.
Features 9 and 35 from the George C. Davis site and Structures 1 and 3 from the Earspool site all had evidence for clay floors. In addition to a clay floor, Earspool Structure 3 also had an original sandy loam floor while Structure 1 had a later sandy loam floor (Perttula and Sherman 2009). It is possible that prepared floors were once present at some of the other structures, but they have been disturbed or destroyed to the extent that they were not recorded (e.g., Jelks 1965). Walters and Haskins (2000) note that the floor of the Bryan Hardy structure was compacted, possibly representing a prepared floor.

Summary
Prepared floors were noted at only 11 (6%) of the previously excavated structures in this study. Three of the four non-mound structures with prepared floors were built within pits, suggesting a specialized function for these structures. Most of the prepared floors were hard-packed clay and measured between 1-4 cm thick. It is possible that other structures may have once had prepared floors that were too heavily disturbed to be recorded at the time of excavation.

REBUILDING AND REUSE
Rebuilding and reuse of structures are seen at Caddo mound sites as well as small hamlets and farmsteads. Structures in some areas of sites appear to have been used for decades with structures being repaired and rebuilt as needed. The practice of rebuilding in the same general location indicates that the Caddo took an active role in creating a longterm sense of place and community. This connectedness with a specific location in a village or community applied to both mound and non-mound structures with sites such as Oak Hill Village, Deshazo, and George C. Davis showing intensive rebuilding of nonmound structures in the same general location as previous structures.
In discussing mound-related structures, Story (1990:341) notes "the most ubiquitous, thus single best attribute for defining this type structure, is their apparent deliberate destruction -by dismantling and/or burning -capping with earth, and, often subsequent rebuilding on the resultant earthen platform." Story (1990:341) continues: "these structural mounds are the tangible remains of a predetermined ritual cycle.
Minimally, this cycle consisted of the following sequence: construction, use, destruction, and final capping (sealing off) with earth." By following such a cycle, the Caddo political elite or ritual practitioners were apparently taping into the power and authority of the past, as well as integrating the power of the ancestors into their own time and place by appropriating and reusing sacred spaces or other spaces of power. The continued construction and use of specialized buildings atop the floors of earthen mounds, as well as the close vertical alignment of many of those structures with ones that came before, suggests a continuity of traditions related to the construction, use, and destruction of sacred spaces.
At least 17 and possibly as many as 34 or more of the 215 non-mound structures (7.8-15.8%) show evidence of being destroyed by burning. In addition, at least 14 and possibly 24 of the 50 mound-related structures (28-48%) included in this study were burned and subsequently capped by an earthen mound. In some instances, however, such as that of Dalton Structures A and B, the burned debris of the preceding structure was cleared away and the subsequent structure was built directly on the floor area of the previous building. At Whelan, subsequent structures in Mound A were built on an accumulation of soil and ash, rather than on intentionally placed mound "floor" or caps (Thurmond 1990;Davis 1958). Following the destruction of the last structure, Structure A, the mound at the Whelan site was capped, sealing the sacred spaces within and creating a landmark that, in some instances may later be marked by marker posts (e.g.,

Perttula and Sherman 2009).
On the more practical side, there is abundant evidence that structures were repaired and wall posts were replaced as needed. Generally, repair is evident by closely spaced wall posts or wall posts placed adjacent to earlier wall posts, such as those at Oak Hill Village, Hatchel, Lang Pasture, Earspool, and others, as well as overlapping posts such as that seen at the Henry M. site, Oak Hill Village, and others.
In addition to the repair of damaged or rotting members, structures were clearly often replaced or rebuilt in the same general location as a preceding structure and it appears in many cases that recycled materials were used where possible. This practice, as seen at Oak Hill Village, Deshazo, Hines, as well as other sites, results, at least in some instances, in subsequent structures being smaller than the earlier buildings. As Rogers and Perttula (2004:49) suggested for many of the overlapping structures at the Oak Hill Village site, "in many instances the newer dwelling was placed adjacent to the first and closely paralleled it in the number, spacing, and size of the wall posts used, suggesting the reuse of existing materials. Often the two houses differed in size, further suggesting that the rebuilt house contained the recycled wall posts of the former, which may have rotted at ground level. As a consequence of using shorter posts, the second structure became smaller in diameter." The village structures at the Hatchel site follow this rebuilding pattern, and Perttula (2005:186) pointed out that the later smaller Feature 1 "probably represents a rebuilding of the larger structure in approximately the same place," suggesting also that recycled materials may have been used. In addition to the benefits of reusing wall posts, the Caddo builders would have the added benefit of possibly being able to reuse existing post holes or other features.
In her analysis of the Deshazo structures, Good (1982:69) noted that the frequency of rebuilding structures was largely determined by the need to replace rotting supports or other structure materials. Based on her analysis, Good (1982)  Deshazo, as well as smaller sites such as Bryan Hardy. The practice of rebuilding indicates a preference of place. This preference may be in proximity to places of power such as the inner precincts at George C. Davis (Story 1997(Story , 1998, proximity to 329 communal spaces such as the plaza at Oak Hill Village (Rogers and Perttula 2004) or those identified at the George C. Davis site by Walker (2009). In addition, this practice links a village to its past and solidifies the place of a structure's inhabitants as members of the community and as one of "Us", as opposed to those whose structures lack that link to the village's shared past who may be seen as "Not Us" or even "Other" (Helms 1998).
One method of perpetuating this shared identity and linking a community or village to the past is through the practice of marker posts, as seen at the Pilgrims Pride site and the Earspool site (Perttula 2005;Perttula and Sherman 2009). The marker posts placed directly through the burned structural remains and central hearth of Earspool Structure 1, an earlier Titus phase structure, marked the earlier structure and its hearth, a significant feature that housed the structure's fire that was traditionally lit from a shared source. Perttula and Sherman (2009)  The practice of rebuilding indicates a preference of place as well as a continuity of traditions. This practice links a village to its past. In much the same way that the communal nature of house building helped to bind the community, building special purpose structures atop earthen mounds over the remains of earlier special purpose 330 buildings further illustrates how these mound-related structures and the organized spaces they created are themselves "objects of sacred display," much like the earthen mounds themselves or other ritual paraphernalia used in those special structures (Knight 1986;Sabo 1998;Story 1990).
The creation of sacred spaces through the burial or rebuilding of certain specialized structures (whether or not they were on mounds) seems to have become prevalent after A.D. 1250 to A.D. 1300. Schambach (1996:40) has noted that at that time Caddo ceremonialism appears to have shifted from "individual-oriented ceremonialism to public building-oriented ceremonialism." Conversely, Perttula (1997, 2004:403) notes "the Titus phase community cemeteries appear to have replaced the use of mounds for community ceremonial and religious functions by about the 1550s." This would seem to suggest that the period of "public-building oriented ceremonialism" suggested by Schambach (1996) may have had its heyday between around A.D. 1300 to the early to mid-sixteenth century, at least in parts of Caddo East Texas. Type 3, sub-round structures (Figure 122).

STRUCTURE FORMS/TYPES
At the Hill Farm site, Walker (2009) places the structures into two architectural forms, circular structures with extended entrances and those without ( Figure 123).
Finally, at the Oak Hill Village site, Perttula and Rogers (2007:76) (2009), there have been at least 10 complete or partial Type 1 non-mound structures identified in the magnetometer data from the Davis site (Table 56) ( Figure 124). These structures correspond in form with at least three excavated structures from the site, Features 31, 42, and 125 ( Figure 125) (Newell and Krieger 1949;Spock 1977).
While the Davis site certainly has the largest number of non-mound Type 1 structures, this distinct form shows up at other East Texas Caddo sites. Type 1 nonmound structures occur at Hurricane Hill, Earspool, Hines, and Oak Hill Village ( Figure   126). Type 1 structures may occur at other sites such as Lang Pasture where possible interior support posts were recorded but less than half of a structure had been excavated.     (Creel et al. 2008:188;Walker 2009). Figure 128 provides a histogram of non-mound Type 2 structures by area.

341
There are at least 20 previously excavated Type 2 mound structures, and at least two possible Type 2 mound structures recorded through the magnetometer survey at the George C. Davis site (Table 57)

Type 3
Type 3 structures are circular in plan with extended entrances. There are 10 Type 3 non-mound structures from seven different sites (Table 58). These structures have diameters ranging from 5.6 m to 12.0 m, with areas ranging from 24.63 m 2 to 113.04 m 2 .
The difference between the Type 2 and Type 3 structures is the presence of the extended entrance, as the latter is often interpreted to indicate a special purpose structure (Kay and Sabo 2006;Perttula 2009;Rogers 1982aRogers , 1982bSpock 1977). Four of the Type 3 structures are from the George C. Davis and Hill Farm sites, and were recorded through magnetometer surveys; out of the more than 110 non-mound structures recorded at the George C. Davis site, only two (1.8%) fall into the Type 3 category.
Type 3 structures tend to be associated with mounds. Twelve (24%) of the 50 mound structures included in this study fit into this category of structure (Table 59). The Type 3 mound structures have diameters ranging from 3.66 m to 9.75 m, with areas ranging from 10.51 m 2 to 74.72 m 2 . While extended entrances are often thought of as a feature typically associated with mound structures (Kay and Sabo 2006;Perttula 2009;Rogers 1982aRogers , 1982bSpock 1977), they only occur in 26% of the mound structures in this study.

Type 4
Type 4 structures are small circular structures with diameters of approximately 3 m or less. Approximately 3 meters was chosen for the cutoff based on work from other Caddo sites that has identified smaller structures, those measuring approximately 3 m or less, as granaries or elevated storage platforms (Rogers and Perttula 2004;Perttula 2005;Spock 1977). There are 11 Type 4 non-mound structures included in this study from four sites ( Table 60). The Type 4 structures range in size from 1.5-3.45 m in diameter with interior areas ranging from 1.8 m 2 to 9.35 m 2 . In discussing what are interpreted as above-ground granaries from the Oak Hill Village site, Perttula and Rogers (2007:78) noted that these structures have no identifiable entrance, "instead, entrance to the structures was probably by a ladder to an elevated platform." In addition to the possible use of small circular structures as granaries, Spock (1977:169) provided additional possible functions for small structures, and noted that "the ceremonial or community nature of small structures were described by Espinosa (Hatcher 1927:160) as homes for the cononicis; by Hidalgo (Hatcher 1927:52) as huts for dancers during feasts; and by Massanet (in Swanton 1942:149) as lodging for the pages of the captains."

Type 5
There are four Type 5 non-mound structures. These structures are all sub-round in shape, have no extended entrance, and have interior areas ranging from approximately 17.35 m 2 to 95 m 2 (Table 61). Walker (2009:107; see also Spock 1977) describes subround structures as those that have truncated or rounded corners. All but one of the Type 5 structures come from the George C. Davis site. The remaining sub-round structure is from the Blount site. All of the Type 5 non-mound structures have non-extended entrances.
There are four Type 5 mound structures, all from the George C. Davis site. These structures have interior areas ranging from approximately 36.3 m 2 to 61.5 m 2 (Table 62).
One of the Type 5 structures has wall trenches.  use, communal, and family spaces (including houses, arbors, granaries, etc), spaces reserved for particular activities and/or people (i.e. assembly houses and the reserved seats for the xinesi), and religious, ceremonial, and symbolic structures or spaces (i.e the temple, the houses of the coninisi, etc). The first category of structures are nonspecialized or non-special purpose spaces while the last two, particularly the third category, can be considered specialized or special purpose spaces.
In archaeological terms, Spock (1977) had classified structures from the George C. Davis site as either domiciles or special function structures. Spock's (1977:169) special function structures included those structures directly associated with mounds, structures that were exceptionally large or small in size, those with unusual shapes, and/or those characterized by unusual or distinctive interior features. Rogers (1982b:49) classified specialized structures as "any of the variety of structures that provided a physical context for the integration of social organization beyond that of the household unit." Rogers (1982Rogers ( :105, 1982b further included a range of architectural spaces, including mortuaries, meeting halls, temples, elite residences, or other public buildings, as specialized structures. These architectural spaces would include those directly associated with earthen mounds, those that are larger than typical structures, those "that are markedly different from contemporaneous domestic dwellings" (Rogers 1982b:49), and those associated with atypical artifact assemblages or features (Rogers 1982a:105).
Extended entranceway structures are also generally considered to be specialized structures (Kay and Sabo 2006;Perttula 2009;Walker 2009), possibly representing buildings associated with ritual practitioners or the elite.
Given these considerations, all of the Type 3 and Type 8 structures on East Texas Caddo sites can be categorized as specialized structures, as would Feature 9 from Davis, 354 which is a Type 6 structure. In addition, based on size, Type 4 structures can also be considered specialized structures, as they likely represent raised platforms or granaries.
The unusually large structures, those with interior areas exceeding 150 m 2 can be classified as specialized structures. These structures may represent open enclosures, assembly houses, community spaces, or other buildings that could have accommodated large numbers of people or certain kinds of activities requiring considerable amounts of space, such as dances or feasting. Structures 6 (Type 6) and 125 (Type 1) from the George C. Davis site were classified as specialized structures by Spock (1977:177-179) due to their architecture or associated features and are so classified here . Specialized structures (Structures 1 and 3) from Earspool, both Type 2 structures, were each built in pits. Table 67 provides a list of the 56 non-mound specialized structures identified in this study.
Structures associated with mounds in the Caddo area have been thought of as special purpose structures (e.g., Perttula 2009;Rogers 1982aRogers , 1982bSchambach 1996;Spock 1977;Story 1990). Accordingly, all of those structures (of whatever structure type) associated with either platform mounds or sub-structural mounds in this study are considered special purpose structures. In reference to Caddo platform and structural mounds, Story (1990:340) noted that "while it is the earthen mound that today stands out, it surely is the associated structures (as well as possibly related features such as plazas) and the activities/events that took place in them that are important." Furthermore, Schambach's (1996:40) assertion of the shift from "individual-oriented ceremonialism to public-building ceremonialism" in the area indicates the importance placed on these kinds of architectural spaces. Caddo mounds have been examined as expressions of political ideology and, as Sabo (1998) has argued, mound construction expressed material symbolism through a number of ways. Sabo (cited in Story 1998) noted that certain aspects of mound construction and use are "physical expressions of underlying principles that guided Caddoan religious and social symbolism." Similarly, Rogers (1982b:105) has pointed out that the "concept of cultural space relates to a variety of contexts, not the least of which is the interplay between the world of humans and the realm of the supernatural. In this context cultural space is physically represented by the buildings, grounds, and other enclosures that religious practitioners use." The socio-political/ideological significance of earthen mounds is bolstered by Knight's (1986:678) argument that platform mounds associated with Mississippian traditions should themselves be included in the triad of iconic families of sacra. This recognition of earthen mounds as symbolically significant "objects of sacred display" (Knight 1986:678) places the specialized structures associated with mounds and those that apparently were used for political or religious purposes within the realm of social, political, or ideological symbolism as well.
Furthermore, the superposition of structures often documented on earthen mounds is a clear expression of the significance of location and place in the construction of these important or "specialized" spaces. Such superposition would act as a means of harkening back and tying into the past and previous use of space, possibly offering a means of legitimizing power.
Parsons (personal communication 2009) notes that extended entrances (an architectural attribute that limits access), the act of ritually destroying and rebuilding (signifying a pattern of death and renewal of ritual or otherwise specialized spaces), and the act of covering with an earthen mound or incorporating an earthen berm around a 357 structure may be seen as an interlocking complex. Although not all of these attributes are associated with each of the specialized structures in Caddo East Texas, such a complex would identify special purpose structures on the landscape, setting such structures apart, and setting the rules for how such structures are to be treated.
The European accounts provide ample evidence for the symbolic significance of Caddo architecture, primarily in the role played by domiciles themselves. While those structures categorized as "non-specialized" structures may not play what is traditionally thought of as "specialized" role, such as those buildings used or lived in by the elite or ritual practitioners, Caddo domiciles played a significant role in maintaining the cohesiveness of the community and grounding individuals and families within the larger group. Cus and Raharijaona (2000:101) have noted that "while there exists a 'plurality of meaning' to a house within a given culture, one recurrent symbolic theme that has been noted cross-culturally is that the arrangement of the domestic structure often reproduces the map of the cosmos, the order of the world." This practice is clear among the Caddo where concepts such as who was considered "Us," "Not Us," and "Other" (Helms 1998) (i.e., "member of a specific community or village", "not a member of a specific community or village", and "supernatural, ancestral, otherworldly") as well as distinctions or categories centered on such things as status, access, kinship, or group membership are surely manifest in the cultural landscape and structured environment (Helms 1998;Locock 1994;Tilley 1994;Sullivan and Rodning 2001;Yeager and Canuto 2000). The symbolism inherent in house construction illustrates concepts and actions promoting group membership as well as the role of house as metaphor for the social group, and also illustrates the symbolism inherent in Caddo architectural space and the practices associated with its creation, use, and destruction (Cus and Raharijaona 2000;358 Donley 1987;Sabo 1998;. As Sabo (1998:168) has argued, "it seems clear enough that houses (or, more specifically, the households represented by the houses) were regarded as constituent elements of matrilineally organized communities." The similarity between common domestic structures and the residences of the elite as well as the control of the structuring of space (e.g., scheduling construction, structure size, and layout organized by political and religious figures) may be reflective of what Cus and Raharijaona (2000:101) note as a desirable action by political or ritual elites "to gain an ideological foothold in local knowledge and symbols to guarantee intelligibility, to facilitate the assimilation of its order by average citizens, and to argue for its legitimacy." Furthermore, as discussed by Tilley (1994:17), "features of the settings of social interaction may constitute 'disciplinary' spaces through which knowledge is controlled or acquired in a highly structured manner. The ability to control access to and manipulate particular settings for action is a fundamental feature of the operation of power as domination." These points are illustrated in the accounts of specialized structures situated near the house of a Caddo chief. These structures were limited to use by particular individuals, some for ritual practitioners, some for visiting dignitaries. By limiting access to, or the use of certain spaces to, ritual practitioners, for instance, those spaces, such as the temple, and the activities associated with them, can be instilled with ideological or supernatural significance. Additionally, these specialized spaces were limited to use at specific times and for specific occasions. The temple and related houses of the cononices are evidence for specialized, disciplinary spaces with limited access, specialized functions, and symbolic significance, through which the rules of society are brought 359 forth, and in which interactions between this world and the otherworld take place (Sabo 1998;. Architectural spaces may become significant as social, political, or ideological symbols signifying meanings not intrinsic to or identifiable in a structure itself. Once the cultural association of architectural spaces transcends the ordinary and takes on the role of signifier, such spaces can act to mediate the interactions and the relationships between the individual and the broader social realm, reflecting broader cultural beliefs or practices. Symbolic meanings can be assigned to architectural spaces through several means, including practices associated with given structures, such as the use of spaces by individuals or groups to which significance has been assigned, through associations with supernatural or religious spaces, or through the locations of structures. Sabo (1998:171) has argued "since kinship was traced through the mother's line, the house-as-icon may be regarded as a symbolic representation of the matrilineage," while "since leadership offices were inherited through the male line, temples and ceremonial centers-as-icons symbolically represented patrilineal jural descent." The emphasis on an otherworldly house of the dead to which all will go as well as the supernatural origin for patterns of spatial organization, and the construction of heaven in the manner and shape of earthly structures, provides a sound foundation for the examination of sacred and symbolic spaces. In addition, the European accounts of the Caddo suggest the symbolism inherent in the temple as icon (Sabo 1998) in its representation of the heavens in the form of the house of the great captain; in housing the sacred and perpetual fire and the accouterments necessary for, and associated with, communication between this and the otherworld; and in serving as a sort of vessel in which the xinesi transcends that gap and communicates with the caninisí (Sabo 1998;360 Schultz 2004). The iconic significance of this architectural space is further stressed in the accounts of the caddi's residence, within which was kept a small replica of the temple itself. This recreation of sacred and symbolic space within the home of the caddices associated these officials and their compounds with the power inherent in the temple space, thus bringing that power into their compounds and providing these officials with a direct association with the otherworld.
The function played by the fire temple in providing the fire for each house in a community or village further illustrates the notion that a community is embodied in the cultural landscape and architecture of the Caddo. Sabo (1998:169) has suggested that "the lighting of house fires with embers from the fire temple can be interpreted as a symbolic representation of sacred connections that subsumed individual households within a large community." The spiritual capital gained by elites, specifically the Gran Xinesi, through their ability to access spaces of power, such as the temple and the houses of the caninisi, necessitated a strong symbolic association with these architectural spaces. The recreation of this sacred and symbolic space within the homes of the caddices further illustrates the significance of architectural spaces as icons and the symbolic significance associated with specialized or iconic spaces such as a temple structure, itself possibly a metaphor for otherworldly space.
The broader social and symbolic roles played by Caddo architecture suggest that physical structures themselves and the organized spaces they create played active roles as "objects of sacred display" (Knight 1986:678). In addition, by being a correlate to the house structure in the otherworld, a Caddo house may be seen as having a close relationship with the cosmos, functioning to locate the Caddo occupants at the focal point of cosmic power and linking together every Caddo household (Sabo 1998).
These conclusions suggest that the broad range of architectural spaces used by the Caddo in prehistoric and early historic times, including daily use, communal, and family spaces; spaces reserved for particular activities and/or people; and religious, ceremonial, and symbolic structures/spaces; all played significant roles in those Caddo societies.
Among the Caddo of eastern Texas, there was a direct relationship between cultural identity, power and ideological authority, and architecture. These spaces all worked together to provide the Caddo with reflections of their cultural identity as well as physical links to the past and the otherworld. For the Caddo, identity and community cohesiveness appears to have been encoded in architecture (Sabo 1998), which acted to reify ideas of community and otherness.

REGIONAL PERSPECTIVE
This dissertation has shown that, while Caddo architecture includes various forms, the dominant structure form throughout the range of Caddo time periods in eastern Texas is circular. This is true for both mound and non-mound contexts. Trubitt (2010) recently noted that, while straight-sided structures are found in southwest Arkansas, the norm, particularly for non-mound structures, is circular. For Webb (1940), the dominant structure shape appeared to be circular with the rectangular structures at the Belcher site being built by an earlier Caddo group. In addition, Webb (1940:73)  Webb's assessment. This suggests that in non-mound contexts at least, circular structures 362 are the norm for the southern Caddo area, with square to rectangular structures being the anomaly in many instances or primarily being associated with mounds.
Many of the mound-related "specialized" structures in portions of southwest Arkansas are square or rectangular (Early 2000). For the Ouachita River Basin, at the northern reaches of the southern Caddo area, Early (2000:130) noted that "straight-walled structures are well-documented phenomena at Caddo mound sites in the greater Ouachita River basin." At the Winding Stair site, Early (2000) excavated a rectangular structure measuring approximately 6.2 x 4.5 m. Early (2000:129) noted "in looking at how the Winding Stair structure, both its general form and the fragmentary superstructure remains, compares with architecture at other sites in the Little Missouri River valley and elsewhere in the Ouachita drainage basin, it seems that rather than being a regional anomaly, it is consistent with general trends in "special" architecture throughout this portion of the greater Caddo area." Many of the mounds excavated by Harrington (1920) in southwest Arkansas had square to rectangular structures as did Mound A excavated by Schambach at the Ferguson site in the Little Missouri drainage. Rectangular structures were also recorded at the Standridge and Adair sites in the Ouachita Mountains (Early 1988(Early , 2000. At the Caddo Valley site on the Caddo River, Trubitt (2010)  Arkansas dating from about A.D. 1200 to 1500, surveys recorded eight rectangular structures at the base of one of the mounds (Lockhart 2007). McKinnon (2008) recorded both circular and square to rectangular structures at the Battle Mound site located along the Red River in southwest Arkansas.
For eastern Oklahoma and the northern Caddo area, the common structure form appears to be square to rectangular, many with two to four interior support posts, similar to the pattern of posts seen in the Type 1 structures from eastern Texas. There is an apparent shift in architectural practices in the Arkansas Basin during the Harlan to Spiro phases. Structures during the Harlan Phase were generally square with four large interior support posts while the later Spiro Phase structures are small and are rectangular with two interior support posts (Rogers 1982a). There is another possible shift during the Fort Coffee phase (A.D. 1450-1650) from predominantly rectangular structures to circular buildings. In fact, Rogers (1982a:107) noted that "round structures are unusual in the Arkansas Basin and it is possible that they represent a very late component rather than specialized buildings in a Fort Coffee Phase context." As is clear from research throughout the Caddoan archaeological area, there is considerable architectural variation and apparent regional architectural traditions. Future studies should explore this variation and look into regional practices and variations in the architectural traditions of the different Caddo groups.

AVENUES FOR FURTHER RESEARCH
This dissertation has provided a detailed look at 265 Caddo structures from 31 sites throughout eastern Texas. One-hundred seventy six of these have been at least 364 partially excavated with the rest (n=89) recorded as part of the magnetometer surveys at the George C. Davis and Hill Farm sites.
As noted, this dissertation has focused on a portion of the broader Caddoan archaeological area. Future studies of Caddo architecture should broaden the scope to include data from Caddo sites in Louisiana, Arkansas, and Oklahoma. Including research on Caddo architecture from archaeological sites from throughout the Caddoan archaeological area will provide a large-scale perspective on architectural variability and practices. In discussing the variety of architectural styles throughout the Caddo area, Early (2000:129) noted "an important future avenue to explore is the development trajectories and regional variation within this far-flung tradition." Future research guided by exploring the different architectural traditions from the Caddoan archaeological area will provide a range of possibilities for future research. For instance, are there architectural practices that are distinct to a particular Caddo group or community? Given that architecture appears to have had such a strong association with the broader community and played such a significant role in group identity, could the architectural diversity seen at sites such as Oak Hill Village and George C. Davis, be indicative of co-residence of different Caddo or even non-Caddo individuals or groups?
Can traits associated with the different architectural traditions in the Caddo area be used to help identify different Caddo or non-Caddo groups? Could these different forms at the same site, rather than simply being related to different functions, be related to different cultural identities of the inhabitants? Are the architectural anomalies at specific sites or in specific areas due to different Caddo or non-Caddo peoples moving into the area or taking up residence among another group and bringing their own, very different architectural traditions with them?

365
The use of daub versus thatch is another potential research issue yet to be addressed. Is the practice of plastering structure walls a trait that is specific to certain Caddo groups and not others? After all, Casañas (Hatcher 1927) and others (Cox 1904;Swanton 1942) noted that some Caddo groups, those to the north of the Hasinai for example, particularly the Red River Caddo groups, had the tradition of plastering their buildings. If plastered structures are found in the Hasinai area, for example, could they possibly represent settlers or other inhabitants from northern Caddo groups?
This dissertation includes only a small portion of the architectural data that is available from throughout the larger Caddoan archaeological area. Additional avenues for research include looking at the archaeological evidence related to the use of space within structures (i.e., how the interiors of different types of structures were laid out and what that evidence may reveal about use of space, or how the archaeological evidence corresponds to the European account of structure interiors). An additional avenue for research is to construct and examine architectural models and reconstructions of Caddo structures. Such a study could provide evidence that could be used to address the function, layout, design, purpose, etc., of structures seen archaeologically.
In addition to the structures recorded via magnetometer surveys, this dissertation included a detailed look at only those previously excavated structures with well-defined posthole patterns. Future studies should include less well defined structures and structure areas that were excluded from this study. Doing so would expand the corpus of possible buildings considerably and will undoubtedly provide additional data pertaining to the construction, use, and destruction of architectural spaces in the Caddo world.
Finally, Chapter 3 briefly touched on modern Caddo perspectives on traditional architecture. An additional avenue for future research is to explore modern Caddo 366 perspectives on traditional architecture and how Caddo architecture, including its design, meaning, and function, has changed over the course of the last one to two centuries.
These are but a few potential research questions that may be explored as we gain a better understanding of the role of architecture in the Caddo world and the architectural traditions that were active in the region from pre-Columbian into historic times.