Shovels and Science

Shovels and science in the quest for knowledge about Soconusco’s past

Most students who have taken an introductory archaeology class appreciate that archaeology isn’t about treasure hunting and that “Indiana Jones” is at best a caricature of an archaeologist. Somewhat less well appreciated is the central role of science in creating an understanding of the past. Even people with some archaeological training tend to think of archaeology as digging holes for some unstated purpose, such as to see what stuff from the past looks like. Here my premise is that archaeology is about building verifiable knowledge about some component of the human past.

Archaeology as just defined depends on science in at least two ways. First, it uses the scientific method, which means that it is hypothesis-driven and builds knowledge through successive attempts at falsification. Second, archaeology depends on analytical techniques that are derived from physical and biological sciences; that is, it borrows techniques that extend the human ability to describe physical and chemical properties of matter beyond what we can accomplish with our visual, auditory, and tactile senses. Although one might argue (and I would agree) that use of the scientific method is the more important connection of archaeology to science, here I want to discuss the various analytical techniques we are using in our fieldwork in eastern Soconusco and how they relate to each other. Hopefully the discussion will also implicate the underlying scientific structure of our endeavor, and perhaps even some of the specific hypotheses and attempts at falsification that guide our research.

The activities of the Proyecto Arqueológico Costa del Soconusco (PACS) can be characterized in a variety of ways, as other entries in this blog have indicated. I have touched on the various ways we are collecting data in other entries, but here I want to discuss these data-collection strategies in some detail. Although I say we are “collecting data”, perhaps a better summary statement of what we are doing is “creating data”. My point is that data (observations about the world) only exist because some detector (our senses or our analytical instruments) has been used to make some kind of measurement on the physical world. 

Sampling the Archaeological Record

We are trying to develop knowledge about the past of a particular region, the lower coast of eastern Soconusco, between the Rio Suchiate (the international border between Mexico and Guatemala) and Puerto Madero. We are investigating this region’s past by sampling the archaeological record from the region and “creating data” based on this sample. Our sampling strategy is hierarchical: first, we have utilized LiDAR imagery as a guide to where archaeological deposits (mounds) are located within and around the lower coastal mangrove swamp; second, we are clearing a sample of these mounds, so that we can carry out magnetic surveys and obtain a sample of the archaeological materials on the mounds; third, we are carrying out controlled excavations on a much smaller sample of the mounds that we have visited.

LiDAR (light detection and ranging) offers the first and only available means of systematically documenting the surface archaeological features within a mangrove swamp environment. The combination of dense mangrove roots and other vegetation, sometimes deep water, and mud effectively precludes any kind of systematic pedestrian survey. But, as examples shown elsewhere in this blog make clear, the LiDAR imagery provides an incredibly detailed picture of the ground surface underneath the mangrove-forest canopy. Based on our visits to LiDAR-mapped mounds over the past couple of months, the LiDAR imagery is not only detailed, it is extremely accurate, down to the depiction of small pits excavated on the mounds by iguana hunters.

Our explorations have, however, documented a number of “false negatives,” which consist of mounds that were smaller than our initial size cutoff. Our field crews ran across these mounds in the course of accessing larger mounds, and a few minutes exploration turned up archaeological material. When we went back to the LiDAR imagery, it was clear that the imagery was accurate and that our size criterion was too conservative. Our initial estimate of slightly more than 200 mounds in and around the swamp is thus too low, perhaps by 50% or more. A reassessment of the LiDAR imagery with a less-conservative size criterion is currently under way.

So far we have documented two “false positive” mound identification based on the LiDAR imagery. One example is shown in Figure 1. Mounds GE12, GE13, and GE14 form a complex that lies at the ecotone between the sandy beach and mangrove forest, about 4 km northwest of the beach town of El Gancho. Mound GE13 is a slightly raised peninsula that extends into the mangrove swamp, while GE12 and GE14 lie just off the shore of the peninsula. During our visit to this complex on 11 February 2012 we noted that the peninsula is pure beach sand with no artifacts whatsoever. GE12 and GE14, in contrast, are clearly archaeological; along with the usual pottery recovered by surface collection, we also recovered about half of a basalt metate (maize grinding stone).

So, with the LiDAR imagery we have created extremely detailed data on the locations, sizes, and configurations of archaeological deposits within our study area. If there are non-mound archaeological deposits within the study area, they are essentially undiscoverable with current technology, as they must lie below the surface on which the mangrove trees are growing, which means they are currently below the water level of the swamp. Thus, we have close to a 100% sample of detectable archaeological deposits within our study area.

The next level of sampling entails obtaining materials that will permit an assessment of chronological and functional variation among the various archaeological contexts in our overall sample. We are doing this by finding routes to some of the mounds, clearing them, and then conducting (1) magnetometer survey; (2) surface collection; and (3) shovel testing. In a few cases of very inaccessible, small mounds or mounds on private property that could not be cleared, we have only obtained surface collections.

Surveys with the magnetometer are intended to reveal subsurface features that might have been related to pyrotechnology. By detecting the strength of the earth’s magnetic field, a magnetometer can identify locations where surface or subsurface features of the sediments enhance or reduce the strength of the field. Such anomalous locations may indicate the locations of sediments heated to high temperature, since heating will lead to realignment of iron particles in the soil so that they point to magnetic north, thus enhancing magnetic readings on one side the heated area and reducing readings on the opposite side. We have been running the magnetometer in gradiometer mode, with one sensor positioned approximately 45 cm above the other, so that gradient data (bottom minus top sensor) are obtained.

Figure 2 shows the gradiometer map from site RS3. The largest anomaly is at the far northern end of the unit, where the map shows a large magnetic anomaly centered at x=1011, y=1018. We hypothesized that this anomaly is created by the presence of intensely heated sediments in the deposits underlying this location, a hypothesis that was borne out by subsequent excavation.

When surface visibility is good and artifact density is relatively high, surface collection is an efficient means for obtaining a sample that will permit an initial assessment of the chronological placement of the underlying archaeological deposit and of some of the activities that led to the deposit’s creation. On our project, the main purpose of surface collection is to obtain a sample of ceramic fragments (sherds) that can be compared to published ceramic type descriptions from nearby sites and regions, thus permitting an initial estimate of the site’s chronological placement. Because surface visibility, disturbance, and other factors have profound effects on the apparent density of artifacts on the surface, we explicitly do not intend these collections to be used for estimating artifact density. For chronological assessment, the most informative sherds tend to be rims, bases, necks, and any sherd with incised or painted decoration; in keeping with archaeological tradition, we term these sherds “diagnostics”. Although we collect some large body sherds, we ignore many small ones, unless we are having a great deal of difficulty finding diagnostics; sometimes the thickness and surface texture of a tiny sherd can suggest a possible period of manufacture.

Another means we are using to obtain a sample of artifacts quickly and with a minimum of damage to the deposit is with “shovel tests”. These are non-controlled excavations of approximately 50 cm diameter and 60 to 100 cm depth. We generally do one shovel test on the top of the mound and one at the edge, near water level, hoping to increase the chance of hitting deposits of different time periods by sampling these two distinct depositional environments. The sediment is screened through quarter-inch mesh, and all artifacts, bones, and shells found in the screen are bagged. When completed, we measure the dimensions of the shovel test so that, unlike with the surface collections, we can use the shovel tests to estimate artifact density. We also take sediment samples from the excavation side wall for FTIR and XRF analysis, samples for luminescence dating, and samples for radiocarbon dating (if encountered). The shovel tests can’t be used to characterize site stratigraphy or change over time, and older, deeper deposits are more likely to go undetected. These tests do, however, provide a systematic sample of the later periods of the site occupation.  

Our most intensive level of sampling of the archaeological record is the controlled excavation unit. So far, we have only initiated three excavations. One of the excavations targeted the magnetic anomaly at the northern end of RS3; another targeted a similar anomaly on RS23. Currently, Richard George is excavating a location on BER19 where both the magnetometer and a later GPR survey detected an anomaly in subsurface sediments.

We excavate in arbitrary 20 cm levels, although the surface zone of heavy root disturbance is typically excavated as a single unit of up to 50 cm depth. Pottery, amorphous fired clay, lithics, shell, and bone are retrieved systematically by screening the sediment through quarter-inch mesh. Features, such as heavy sherd concentrations, areas of intensely heat-affected soil, or bone concentrations, are uncovered with trowel and brush. Two of the units have been expanded beyond the initial 1 x 2 m size in order to obtain better exposure of features. Excavation generally stops at the water table, but we excavate a bucket auger below the water table to obtain some indication of how much deeper the archaeological remains may continue.

A variety of samples are taken during and after excavation. These include charcoal for radiocarbon dating, sherds for luminescence dating, and column samples from the unit sidewall for FTIR and XRF analysis as well as fine sieving to attempt to recover small bones and charred remains. Samples of sediments from features are bagged for analysis as well.
Sampling the Paleoenvironmental Record

People leave a mark on the landscape outside of archaeological deposits, such as the mounds discussed above. By cutting and burning the forest and firing the salt- and ceramic production facilities in the swamp zone, past peoples’ activities affected the content of sediments that were accumulating in low-energy fluvial environments, such as the lagoons of the swamp zone.

We obtain our samples of the paleoenvironmental record by coring in locations that are likely to have been inundated during most of the past several thousand years. Perennially wet conditions preserve organic matter, including pollen. In past work, I have found such depositional environments near the inland margins of mangrove swamps. We use a “vibracorer”, which is basically a gasoline-driven concrete vibrator attached to 3-inch metal tubes of 3 – 6 meters length. The cores are taken in one or more drives in the same hole. After pulling the tubes containing the sediment out of the hole, the tubes are sealed and taken back to the field laboratory for sampling.
The Field Laboratory

Materials recovered during surface collection, shovel testing, excavation, and coring are returned to the PACS field laboratory for processing. Artifact collections, which mainly consist of fragmentary ceramic remains, are washed, dried, and labeled according to provenience unit by laboratory workers. Samples for luminescence and radiocarbon dating are catalogued and prepared for export. The coring tubes are opened, described, and sampled for various analyses.

One crucial task in the field laboratory is the typological study of the large collection of excavated ceramics generated by our project. Paul Burger is working with the collections from the three excavations undertaken this year, developing a preliminary taxonomy and descriptions of the identified types. Comparison with other ceramic reports for our region and nearby regions permits a more-or-less secure preliminary assessment of when the excavated deposits accumulated. We are also assessing the surface collection and shovel test pit ceramic samples for chronological placement and functional characterization.

One of the ways we are attempting to innovate on the PACS is by carrying out many “high-tech” analyses in the field laboratory, as soon after excavation as possible. Miniaturization of analytical instruments over the past 10 year or so has made this innovation a no-brainer. For instance, we have a portable x-ray fluorescence spectrometer (XRF) and a portable Fourier-transform infrared spectrometer (FTIR) in our field lab, so that we can do analyses “on the fly” that 15 years ago would have required exporting samples to a university or museum laboratory in the US. A small, high-temperature furnace gives us the ability to experiment with heat-effects on sediments. We also have a stereoscopic microscope and light source, which we are using to process and count the charcoal in our cores.

The XRF and FTIR complement each other in the analysis of sediments from archaeological excavations. XRF measures elemental concentrations rapidly and with minimal preparation of the sediment samples. FTIR characterizes the sediments at the molecular level, including identifying the presence of clays and other mineral components. If, say, the XRF analysis shows enriched calcium in a sediment sample, the corresponding FTIR spectra can be examined for the presence of pyrogenic calcite, which is produced by the complete combustion of wood.

Beyond characterizing sediments, the portable XRF also gives us the ability to characterize the elemental composition of artifacts that we have recovered. Artifact elemental characterization is particularly useful for obsidian, since elements determined by XRF easily discriminate the various Mesoamerican obsidian sources. So far, only three obsidian artifacts have been encountered on our project (the small number perhaps an indication of the functional specialization of the mangrove-zone sites); all of these artifacts have been sourced to the El Chayal source, in the highlands of Guatemala. XRF can also be used to analyze ceramic compositions; this will be important for evaluating the relative abundance of ceramics from different local and non-local sources on the sites we are investigating.

The FTIR is really a key instrument for the current project because of its ability to detect the presence of different mineral phases. In soils and sediments, clay minerals are expected to constitute a major component. When exposed to high heat, however, the clay minerals begin to lose their hydroxyl groups starting at around 500^o C, and by 900^o C vitrification begins, with the eventual formation of high-temperature silicate minerals. FTIR spectra of clays (or clay-containing sediments) heated to different temperatures thus differ systematically, and essentially provide a thermometer that monitors the heat to which the sediments were subjected. In work undertaken before our current fieldwork, Scott Bigney heated a sediment from Soconusco to temperatures ranging from 300 to 1100 degrees, and then obtained FTIR spectra from these samples; these samples form a library for assessing heat exposure of the sediments we are excavating in controlled excavations and STPs. We have also collected additional sediments from the immediate vicinity of one of our excavated mounds, and the FTIR spectra from these samples, with different heat treatments, are being added to the spectral library.

Scott Bigney is also using a procedure developed by geographer Megan Walsh to determine the concentration of charcoal at different depths within our sediment cores. Small samples of sediments are disaggregated and sieved through very fine screens. The samples are then placed into plastic Petrie dishes on which lines have been scored at regular intervals. Under a microscope, charcoal is counted by moving up one line and then moving over and counting down the next line. The total number of charcoal particles divided by the volume of sediment processed gives an estimate of charcoal concentration per cc. Eventually, when radiocarbon dates have been run on samples from the cores, we will be able to identify when the coastal plain was subjected to intense burning, which would be times when the most forest clearance was taking place.
Specialized Laboratory Analyses

While we are doing far more in the field laboratory than is common on most archaeological projects, some of the data we need to address this project’s goals have to be generated in more-specialized analytical laboratories. In particular, the chronometric analyses, which will give us information on when the deposits we have sampled accumulated, have to be undertaken in specialized radiocarbon or luminescence-dating laboratories. Isotopic analysis and analysis of pollen from the cores also has to be done in specialized laboratories; these analyses will complement the charcoal data being collected by Scott Bigney, and together they will document the history of forest clearance on the coastal plain.

All samples to be exported to the US for specialized analysis have to be entered into a catalog that lists all descriptive and contextual information. The catalog indicates the priority for analysis of each sample. For instance, samples from lower levels of the cores are a high priority, since they give a rough idea of the rate of sediment accumulation, so that we can estimate dates for the high-charcoal levels identified by Scott Bigney’s analysis. Paired luminescence and radiocarbon samples are also a high priority, since we want to validate the luminescence dates, which are still used very rarely in Mesoamerican archaeology and thus may be viewed skeptically by some archaeologists.
Some Preliminary Results

Ultimately, the data generated by this project will show how human use of the mangrove estuaries of eastern Soconusco varied in intensity and nature over time. So far, some of the broad outlines of this history are coming into focus, allowing us to refine hypotheses for future testing.

To begin with, in our area, as elsewhere on the Pacific coast east of the Isthmus of Tehuantepec, Early Formative people of the Locona and Ocos Phases lived at the inland edge of the estuaries and amassed large, tell-like habitation mounds that contain evidence of exploitation of estuarine resources. Cerro del Tigre (RS31) and El Castaño (RS19) exemplify this Early Formative focus, as does MA20, a site we have not visited but which was visited by Robert Rosenswig in 2011 (personal communication). Bermudez-1 is also most likely an Early Formative site, although we have yet to obtain a definitive surface collection or shovel test pit sample. The settlement and subsistence focus of the Early Formative in our area is consistent with observations on many other projects that have investigated Early Formative contexts in Pacific coastal Chiapas and Guatemala. Richard Lesure (2009) has recently published an excellent synthesis of current understanding of Early Formative adaptation to the estuarine environment.

So far in our area, it appears that Early Formative people abandoned the estuary prior to the Cuadros phase, around 1150 BC. This presents a contrast with the region just east of where we are working, in Guatemala, where Coe and Flannery (1967) excavated a deep Cuadros and Jocotal occupation at Salinas La Blanca. Although we have yet to study the ceramics from all of our sites carefully, so far we have noted no diagnostics of periods between the Ocos phase (Early Formative) and Crucero phase (Late and Terminal Formative). Occupations of the Middle Formative Conchas phase are well represented on the coastal plain, especially at the large Middle Formative center of La Blanca, Guatemala, which has been investigated by Michael Love (2002). It is noteworthy, therefore, that the Middle Formative does not appear in our area based on the ceramic collections examined so far. More careful typological study of the ceramic collections and chronometric dating (luminescence and radiocarbon) of excavated contexts will provide additional evidence on this inference about a hiatus in estuarine settlement.

People reoccupied the swamp during the Late Formative and Early Classic, most of the sites in our area having ceramic diagnostics of the Crucero phase, which parallel types of the Guillen immediately succeeding phases at Izapa. One especially common diagnostic is the rim flange, which comes in during the Late Formative at Izapa and persists into Early Classic times. Voorhies (1976) places this ceramic mode in the Early Classic in the Acapetahua Estuary.  If the swamp zone in our area was heavily occupied during Early Classic times, as the preliminary evidence suggests, this presents a dramatic contrast with the immediately adjacent coastal plain, where Rosenswig (2008) finds a near absence of both Late Formative and Early Classic occupation.

Late Formative and Early Classic contexts that we have investigated this year pertain to at least two distinct functional categories. Two sites, RS3 and RS23, are specialized pyro-technological features dedicated to salt and/or ceramic production. The ceramic collections at these sites are dominated by a single large, coarse ceramic type that Nance (1993) has named Porvenir Coarse. These vessels were almost certainly made on site and were very crudely finished. We believe that they were used to make salt via the sal cocida technique (in agreement with Nance 1993), but evidence of fires built inside the vessels suggests that they might also have been used as saggars in the firing of smaller ceramic vessels. Both RS3 and RS23 also have very abundant, massive wood-ash concretions that testify to the importance of combustion of fuel at these locations. Fire-reddened clay is also a major component of the midden. FTIR results also indicate that even much of the sediment that does not appear to be fire-reddened was, in fact, subject to temperatures in excess of 400^o C. Ceramics other than Porvenir Coarse are infrequent in the RS3 and RS23 deposits, and no chipped or groundstone artifacts were found, and the only faunal remains found were two fragments of oyster shell at RS3.

Less-specialized contexts seem to be present at a number of sites in the central swamp. In one of these, BER19, we excavated a 1 x 2 meter unit to a depth of 2.1 meters. Although the materials from the excavation have yet to be analyzed, there appears to be greater variety and a more even distribution of different functional classes of ceramics. This together with the presence of groundstone, obsidian, bone, and a single jade bead indicate that this context is less specialized than the pyrotechnological contexts represented by RS3 and RS23.

Non-systematic surveys in the past have documented a strong Late Classic and Postclassic presence near the mouth of the Rio Cahuacan, just northwest of the area we have been investigating this year. We may have some preliminary surface collection and STP data bearing on these earlier observations by the time our field season ends on March 10. If our previous observations are confirmed, it would seem that human use of the swamp shifted from a focus closer to the Rio Suchiate during Late Formative and Early Classic times, to a focus around the Rio Cahuacan during Late Classic and Early Postclassic times. Whether the same range of functional variation is evident at the later sites and the reasons for a shift over time in where people carried out their swamp-focused activities are topics for future investigation.

References

Coe, Michael D. and Kent V. Flannery (1967) Early Cultures and Human Ecology in South Coastal Guatemala. Smithsonian Contributions to Anthropology 3. Washington, D.C.

Lesure, Richard G. (2009), editor and principal author. Subsistence and Settlement in Early Formative Soconusco: El Varal and the Problem of Inter-Site Assemblage Variation. Cotsen Institute Press, Cotsen Institute of Archaeology at UCLA, Monograph 65. 

Nance, C. Roger (1992) Guzmán Mound: A Late Preclassic salt works on the south coast of Guatemala. Ancient Mesoamerica 3:27-46.

Rosenswig, Robert M. (2008) Prehispanic settlement in the Cuauhtémoc region of the Soconusco, Chiapas, Mexico. Journal of Field Archaeology 33:389-411.