Will the Real San Marcos Pueblo Please Stand Up: An Examination of Bias and Error in Archaeological Site Maps


 Shawn L. Penman, Ann F. Ramenofsky, Christopher Pierce, David Vaughan, and Eden A. Welker


Poster presented at the 63rd Annual Meeting of the Society for American Archaeology, March 25-29, 1998, Seattle, Wa.


Maps make up an essential element of information about the archaeological record. Although archaeologists construct a wide variety of maps at different spatial scales, site maps are most fundamental. Site maps depict the locations and arrangements of architecture, features, and artifacts at ancient settlements. We routinely use site maps to carry out important resource management and research activities such as delineating site boundaries, estimating past populations, and reconstructing the internal organization of settlements. These uses of site maps are so common, in fact, that we tend to forget that the maps are two-dimensional abstractions and interpretations of a complex three-dimensional surface, and treat them instead as objective, accurate, and reliable descriptions.


In this poster, we take advantage of the existence of three, independently produced maps of one site, San Marcos Pueblo (LA 98) located in Galisteo Basin of north central New Mexico. We use these maps, produced over a period of 82 years, to examine similarities and differences in the ways the maps depict this large, complex settlement. Further, we evaluate how different goals, methods, conditions, and perceptions affect the accuracy and precision of site maps.

The Site


San Marcos is a large pueblo located near the intersection of San Marcos Creek and New Mexico Highway 14 south of Santa Fe on a 60-acre preserve owned by the Archaeological Conservancy. Beginning in 1912, N. C. Nelson conducted the first archaeological research at San Marcos for the American Museum of Natural History. In 1915 he returned and trenched each room block and constructed the first planimetric map of the site. In the mid 1980s, Jonathan Haas and Winifred Creamer mapped the site and conducted surface collections and test excavations in the western portion of the site. Their map has not been published. Between 1988 and 1993, the University of Colorado mapped and conducted in-field artifact analyses on the eastern portion of the site. In 1997, the University of New Mexico began preliminary investigation at San Marcos which involved making a new map of the surface of the site.


The Maps


Compiling and presenting the three maps for this poster proceeded in several steps. In the spring of 1997 we discovered Nelson's original four, 16 x 20 inch map sheets of San Marcos at the American Museum of Natural History. Although different versions of Nelson's small sketch map have been published, the map reproduced for this poster is the larger, more complete, and previously unpublished map. Penman digitized Nelson's 1915 map and the University of Colorado 1996 map published in the Journal of Field Archaeology. Penman then transformed both maps to the same scale and graphic style using ARC/INFO. Data points from the 1997 University of New Mexico fieldwork were converted into a plan map using the Sokkia Map program, and then were ported to ARC/INFO for final production at the same scale and style as the other tow maps.


American Museum of Natural History (Nels C. Nelson) 1915




Current information about Nelson’s fieldwork suggests that he constructed his map of San Marcos with a surveying instrument, and possibly employed a professional surveyor. Nelson's mapping goal was to identify room blocks or architectural features at San Marcos. Photographs taken of San Marcos at the time show much less vegetation than is present currently. We presume that the absence of vegetation affected how the boundaries of room blocks and other architectural features were identified.


The University of Colorado 1993




The Colorado map was constructed in 1993 using aerial photography taken in 1989. The photographs consisted of Thermal Infrared Multispectral Scanner images and color infrared photography. The base topographic map was constructed using stereo pairs with the cultural features superimposed on the map. The Map was ground truthed by walking over the eastern section of the pueblo and the western section was compared to the Haas and Creamer’s unpublished 1990 map. Colorado's mapping project was designed to provide a modern definition of the settlement and produce a map for distributional studies of portable artifact classes. The plaza was the defining architectural element used by Colorado. We assume that the vegetation was as dense in 1989 as it was in 1997.


The University of New Mexico 1997




The New Mexico map was made in 1997 using a Sokkia Total Station. The goal of the UNM mapping project was to construct a map of San Marcos that interpreted both the cultural and natural surface topography. Outlines of mounds were interpreted using topographic variation and adobe wall visible on the surface. We were also able to map wall alignments across 5 room blocks, as well as the Church and Convento. The density of vegetation obscured surface visibility in some areas.


Map Data Analysis


Comparison of Surface Architectural Features


This graph shows the number of room blocks, kivas, and plazas identified by each mapping project. The University of New Mexico map has the lowest number of room blocks and plazas, while University of Colorado map has the largest number of all architectural features.




Comparison of the Total Area of Room Blocks and Plazas


This graph shows the variation in total area (in square meters) of room blocks and plazas.



Comparison of Feature Position


The table summarizes the absolute differences in meters in the location of various features across San Marcos. To derive these measurements, we imposed a 25m2 grid across each of the maps. We then used the grid to locate the positions of features on each map relative to each other. The point of origin for each map grid is the center of the same kiva located in the northeast quadrant of the site. This technique facilitated comparing relative position. As is apparent, there are major differences in position. For Point 6, the Nelson and Colorado maps differ by 8 meters; the Colorado and New Mexico maps differ by 69 meters; and the Nelson and New Mexico maps differ by 31 meters.
















































Discussion and Conclusions


Because site maps are fundamental to field archaeology, it is easy to proceed as if they are accurate representations of cultural and natural landscapes. We usually assume that the position, number and form of features, as well as the relationships between features are adequate for our archaeological purposes. We then proceed to use the spatial descriptions as if spatial reality has been captured. However, as we have demonstrated in the San Marcos case, site maps can, and probably do contain considerable error.


Substantial differences exist among the three maps of San Marcos Pueblo that can lead to divergent interpretations regarding the number, size, position, and spatial relationships of architectural features at the site. For example, the University of Colorado map depicts the site as composed of numerous discrete room blocks creating a more open, disconnected appearance. The University of New Mexico map, on the other hand, shows fewer, more connected room blocks suggesting a more enclosed and integrated settlement. Nelson's map lies somewhere in between the other two maps in this regard.


Which map is the most accurate description of the site? We don't know. We can identify the possible sources of error among the maps, including differences in protocols used to delineate features, different techniques used to construct the maps, differences in the condition and visibility of the surface when the maps were made, and different goals and interests of the investigators. Importantly, the lack of explicit standards for how to construct and report site maps makes it difficult to determine exactly how these sources of error may have affected each map.


Knowing the accuracy and precision of site maps matters if we are going to use these maps as sources of information about ancient settlements, and as a basis for designing further research at those settlements. The largely implicit nature of our current approaches to making site maps forces us to assume that they are both accurate and precise. This assumption in the San Marcos case is not warranted. As suggested by the long history of drawing precise and accurate profiles or floor plans, we have the tools for constructing similarly precise and accurate site maps. By applying the same rigor at the site scale and by creating and reporting site mapping conventions, and identifying the assumptions that guide the construction effort, we will come closer to the product we seek.




We would like to thank: Mark Michel and the Archaeological Conservancy for granting us permission to work at San Marcos; David Hurst Thomas and Lori Pendleton at the American Museum of Natural History for facilitating access to Nelson’s collections and archives at the museum; The Earth Data Analysis Center, University of New Mexico, for technical assistance; Dr. Dale Lightfoot, Oklahoma State University; and Elizabeth A. Bagwell and Kris Lajeskie for their assistance in the field.