Archaeology of the Guatemala Pacific Coast

Economic Interaction and Political History: A Chemical Compositional Approach
Frederick J. Bove
Hector Neff

To appear in Proceedings of the Inter-Congress Meeting of Commission 4: Data Management and Mathematical Models in Archaeology, edited by K. Kintigh & G. Cowgill, Archaeological Research Papers, Arizona State University, Tempe.

Abstract

A review of ceramic characterization by Bishop (1994) suggested that it was most often attempted on a particular class of pottery or by reference to a specific site or time period. There is, however, an increasing reliance on the use of trend data as the basis for archaeological interpretation. Innovative studies by one of us (Neff) successfully characterized the circulation of ceramics based on individual-specimen provenance by adapting standard approaches to single-specimen "groups" so that each ceramic pertains to a single-specimen "group" thus permitting the generation of individual-specimen probability surfaces. Using this methodology, we show how regional ceramic resource procurement and exchange relations that are independent of standard typological approaches can be a useful aid in the development of regional evolutionary models and in reconstructing the political history of the Pacific Coast of Guatemala during the Middle-Late Classic period.

Introduction

The research stems from our recent projects at Texas-Montana (formerly Los Chatos-Manantial) and Cotzumalguapa, two regional states on the central Escuintla, Guatemala Pacific Coast during the Middle-Terminal Classic periods (ca. A.D. 400-900/1000). We are especially interested in understanding the shifting nature of political competition and social-economic relations between the two regional polities during these periods. Available evidence suggests that the Texas-Montana core zone at its Middle Classic height of power (A.D. 400-700) probably controlled an area stretching from the Río Coyolate, including the Paryjuyu zone, to the east side of the Río Achiguate, and possibly as far north as Cristóbal and Los Cerritos-Norte. It is likely that at some point during the Middle Classic period, it either exercised dominion over, or had more intense ties with Cotzumalguapa. Beginning at about the end of the Middle Classic San Jerónimo phase (ca. A.D. 650-700) there was apparently a breakdown of the centralized political system headed by the Texas-Montana core into small competitive polities analogous to small city-states in the Terminal Classic Pantaleón phase (Bove, 1994; Bove and Medrano, in press).

Coincident with this transformation is the rise of the Cotzumalguapa state during the Late Classic period (A.D. 700-900/1000). The nuclear zone, comprised of Bilbao-El Baúl and associated El Castillo groups, is located 40 km north of Texas-Montana and centralized to a considerable extent. This zone contains an overwhelming percentage of major architecture and sculptures within a continuous settlement area. Major centers such as Palo Gordo and Los Cerritos-Norte are located at some distance but its influence in the form of monumental sculpture can be found as far as the El Salvador frontier. A potentially important factor is that the Cotzumalguapa state is largely a piedmont phenomenon in contrast to the lower coastal oriented Texas-Montana polity.

Our working hypothesis is that the two polities were politically and economically independent in the Late-Terminal Classic period. This interpretation is based on significantly different settlement patterns and architectural arrangements of contemporary site systems as well as somewhat different ceramic complexes although many attributes are shared. Another major factor is the spatial distribution of the unique Cotzumalguapa sculptural art style considered an expression of nuclear zone political ideology (Chinchilla, 1996a). There is some evidence for a single overarching domain in the Middle Classic that would have included the Cotzumalguapa nuclear zone within the Texas-Montana sphere. However, the extent to which residents of the two political units participated in similar or different economic networks are the research questions we address in this study.

Two alternative approaches are typically used. The first is to form reference groups of "knowns" by intensively sampling sources sufficient to estimate within-source variation and then to source artifacts or ("unknowns") by comparison with a series of known groups. This is a practical approach when raw materials occur in localized flows, or if potters exploited local resources preferentially. But where clays are practically ubiquitous, such as Pacific Guatemala, it is in fact difficult or impractical to obtain a large enough sample that represents the range of variation in most, or in all clay sources. A second alternative approach is to use pattern-recognition and statistical group-refinement techniques to form reference groups of "unknowns", and then to compare a series of "knowns" (i.e. source materials) to unknowns. The locations where raw materials are found within the range of group variation are most likely to contain the sources of ceramics in the reference groups. This method is the so-called standard approach to ceramic sourcing and chemical compositional analysis.

Pacific Coast Ceramic Resource Project

One of the reasons we believed that a chemical compositionally oriented project was potentially feasible for Pacific Guatemala ceramics is due to its unique geological and climatological characteristics. Although it appears superficially on maps as largely a homogenous alluvial plain it is an extremely dynamic area. The chain of quaternary volcanoes and accompanying slopes contains broadly deposited volcanic ash and other debris and are subject to tremendous annual rainfalls (Bishop, 1994). The combination of torrential rainfall and steep, locally unvegetated slopes provides abundant bed load to high-gradient, low-sinuosity braided streams discharging on the coastal plain. The resulting wide coastal strip is therefore thickly covered with fluvial sediments ranging from fine, well-stratified volcanic sands, silts, and clays to coarse, bouldery, unstratified accumulations left by torrential lahares with increasing fineness of deposits and degree of sorting toward the coast (Williams, 1960).

Bishop (1994) observed, "as a result of the sedimentary conditions this is an area that is not readily seen as conducive to a compositionally based investigation." It was believed, however, that the chemical variation presumably present in the highlands and adjacent southern slopes could show significant variation in the chemical signature of clays, sands, and ashes as erosion in the respective areas carried these materials south toward the Pacific Ocean.

Early research was undertaken of Pacific coastal ceramics for neutron activation analysis in the late 1970’s and early 1980’s initially by Frederick J. Bove, Arthur A. Demarest, and Barbara L. Stark. These analyses were carried out at Brookhaven National Laboratory (BNL) under the direction of Ronald L. Bishop (e.g. Bishop, Demarest and Sharer, 1989; Neff and Bishop, 1988; Neff, Bishop and Bove, 1989; Neff, Bishop and Harbottle, 1989; Stark et al., 1985). The preliminary results indicated that compositional subgroups should be recognized. Consequently a program of ceramic raw material sampling combined with a substantially larger ceramic sampling program was begun in 1987 with major survey conducted in 1991 and intermittently since under the direction of Neff and associates (Neff, Bove, Lou and Piechowski, 1992). These analyses were carried out at the University of Missouri Research Reactor Facility (MURR). The total MURR database now exceeds 2000 ceramic and 300 raw material analyses. The sample ranges geographically from the western Guatemala coast to the El Salvador border with additional samples from the adjacent highlands principally in the vicinity of Kaminaljuyú and the Panchoy and Sumpango valleys directly north of the central Escuintla, Guatemala coast.

It is important to emphasize the underlying premise of this research given the above conditions which is "to the extent that regional trends are characterized adequately, any particular ceramic compositional profile may be assigned to a likely zone of origin within the region, even though, in the vast majority of cases, the exact locations exploited by ancient potters will not have been sampled" (Neff, Bove, Lou and Piechowski, 1992). As observed in Figure 1, Principal Component 1 (PC1) captures a strong west-to-east trend on sampled clays and it is possible to distinguish clays at both extremes with the highest scores at the eastern end and lower on the western although an obvious continuous gradation exists. It is important to point out that the total analyzed pottery sample from central Escuintla overlaps almost entirely with analyzed clays indicating that the ancient potters did in fact use the same range of raw materials that we sampled. However, the same plot also shows that the addition of ash temper tends to make the ceramic look more "western" while sand tempers tend toward a more eastern profile. Therefore the effect of temper must also be taken into account.

The earlier research can be summarized as follows. In Pacific Guatemala where ceramic raw materials generally vary along a continuum with few discontinuities, the standard approach to chemical compositional analysis is severely limited as subgroups can only be defined by an arbitrary partition of the compositional continuum. These partitions not only represent production zones that include a number of sources as well as a number of paste recipes but the geographic meaning of the groups also shifts as group membership changes. Ultimately, the only non-arbitrary partition of the continuum is one that places every individual specimen in its own group. This approach would eliminate uncertainty in source assignments that arise from the inclusion of ceramics from multiple sources in groups obtained by arbitrary partition.

A method was developed that more realistically models the ceramic environment by adapting the standard approach to accommodate the possibility that each ceramic may be linked to its own unique or specific source. In addition the effect of temper is explicitly modeled by calculating best-fit mixtures of geographically proximate clays and tempers (or none) to compare to each individual ceramic composition. The methodology is described in detail elsewhere and is not repeated here except to state that maps representing the likelihood that a given location contains the source of a given pot are generated for each and every analyzed sample in a collection, and then all the maps for a given collection are amalgamated into a composite (Neff and Bove, 1999).

Characterizing Circulation of Ceramics–New Approach

But how can this variation be described to facilitate testing of hypotheses concerning spatial and temporal production and consumption variation? (And more importantly as archaeologists to the shifting nature of political and economic networks.) Generally the information on sources of pottery can be consolidated by defining classes of analyzed ceramics by criteria other than composition and then comparing these classes as they relate to the geographic space indicated by the source data. We believe that a major strength of our approach is that, because it assesses provenance on a specimen-by-specimen basis, it provides a monitor of ceramic resource use and circulation that is completely independent of other evidence such as is typically embodied in ceramic typologies. The resulting data are therefore independent of preconceived culture historical approaches.

To best illustrate the new approach we analyzed approximately 600 examples of serving vessels derived from a number of sites across the study region and presumed to be chronologically concordant dating generally to the Middle-Terminal Classic period (ca. A.D. 400-1000). We tested the hypothesis that geographic location determines access to serving vessel production centers and stems from the expectation that consumption location or sites should be a good predictor of similarity in access to production zones. Results of the analysis suggested, however, that Paraiso and Lirios 3 are grouped together at one extreme of principal component space, whereas Los Chatos-Manantial is grouped with Castillo-Cotzumalguapa. This finding contradicted our initial expectations that political independence would also imply a high degree of economic competition based on the working hypothesis that the two polities of Los Chatos-Manantial and Cotzumalguapa were politically and possibly economically independent (competitive) in the Late-Terminal Classic period (Bove and Neff, 1998; Neff and Bove, 1999). The ramifications of these results continue to be explored in detail and we are in the process of augmenting the individual-specimen sourcing data within the framework of further refinements in ceramic typology and chronology emerging across the region. Ultimately we hope that that our models can accommodate multiple lines of independent evidence including the incorporation of GIS methodology to prepare new derived composite map layers incorporating the chemical compositional data to compare with hypothesized political units or zones over time.

Sources Cited

Frederick J. Bove
fbove@ufm.edu.gt

Hector Neff
neffh@missouri.edu

Abstracts: 1  |  2  |  3  |  4

Previous Page  |  Main Page

Return to top of page