5
DATA RECOVERY
See the World in a Grain of Sand.
—William Blake
Before you start recovering anything, remember that data are not what you are after. Data are indeed the stuff of archaeology—stone, bone and other materials, spatial relationships, and so on—but as soon as you think “Clovis point” (you wish) or even “cortical flake,” you are entering the realm of information, the constructs you make in the production of knowledge. Knowledge about the past is the goal, something you hope will resonate throughout the archaeological community when you publish your findings.
The first steps in archaeological knowledge production are theoretical ones. You need to understand first that artifacts and features are not data: They are just artifacts and features. Data are creations you devise based on your research design (see Toolkit, volume 1) and then extract from the artifacts and features. Artifacts and features don’t speak for themselves; the archaeologist becomes their “voice.” We may like to call what we do “data recovery,” but it’s worth remembering that it’s really “data creation.” As you create data, you have to define the scope of the data with which you plan to work.
While we act at a human scale, cultural data resolve at multiple scales, from global to microscopic. You may think of an artifact, for example, as something you can hold in your hand and a landscape as the cultural environment—the background for all those archaeological sites. The problem is that the thing we call a site is an artifact on the landscape, and a projectile point is a landscape, too, a surface marked by cultural impacts (literally). We are just too big to see a projectile point that way.
So, your first decision about data recovery must refer to the original research design. What are you trying to accomplish? Do you already know the nature of the culture and activity at the site, or is it still a mystery? Is this basic data recovery, or are you testing specific hypotheses? Depending on the answer, you may need to be prepared to “recover” data ranging from the spatial relationships among artifacts and the surrounding environment to pollen grains, the smallest bits of residue from toolmaking (microdebitage), or chemical residues from bone decay or foods.
If you fail to appreciate this problem of scale, you may miss data crucial to an understanding of the phenomena you hope to study. Consider a simple example: the difference between two sites along a river, one at a confluence with a creek, and the other downstream along the bank. If hunters and gatherers saw the former place as more strategic, because it provided access to two drainages rather than one, might it have attracted more people over time than the terrace downriver with only its level well-drained ground to recommend it? If you bury your nose in a site, you may not even imagine that subtle differences in the site’s raw materials, artifacts, and activity areas might reveal differences of local or regional significance.
ARTIFACTS AND THINGS
Once you have sorted out your scales of analysis, you can turn to your data recovery strategy. As noted earlier, as soon as you start naming things, you are making interpretations, so now is the time and place to be concerned about classification. The point: You may want to collect “all” the data, but you will retrieve only the data you recognize from the tiny bit of data that survives.
Much of classification seems to be routine. You may distinguish material culture objects (artifacts and the residues of production), natural things used in cultural activity (e.g., hammerstones), and natural things indirectly affected by cultural activity (e.g., burnt soil or an unnaturally organic soil stratum). You may expect to find spatial relationships between and among cultural things. The problem is that you are inevitably imposing your own beliefs, implicitly and explicitly, on material things that do not advertise themselves at all. If the past speaks, it says what you want to hear! If you describe a scatter of artifacts as a site, for example, you are assuming a cultural relationship among these things, even though many different individuals, groups, and cultures may have stopped there for many different reasons over time. Because classification is inevitable, the only solution is to think out why you are investigating this particular place and what you hope to gain from the effort.
The issue here is the extent to which you document these artifacts and relations. On an initial reconnaissance survey, for example, you may not need to describe every piece of chert or measure spatial relationships beyond the visible extents of an artifact scatter. If you are excavating, then you will.
COLLECTION STRATEGY
After thinking about the issues discussed so far, you may worry that no amount of data recovery will ever be enough. Clearly, however, there are limits to what you can do as an archaeologist. The goal is to accomplish what you set out to do; no one expects you to recover a lost world from the meager traces left to us. To resolve this dilemma in a positive way, archaeologists develop collection strategies. What you collect depends on at least two things: what you hope to learn (your research design) and how much data you need to retrieve to learn it. Have you stumbled on the site in a reconnaissance survey? Or is this part of a hypothesis-driven study dealing with some specific archaeological problem? In most excavations, you may need to sort out occupations or activity areas, so you may want to record very carefully in three dimensions exactly where everything is located and collect a representative sample.
You may consider a systematic sampling strategy if you run the risk of being overwhelmed by data—in fact, you probably already use an informal one, if you walk in linear transects across a site. More formalized sampling methods, backed by probability theory, often involve collecting only within predetermined, randomly generated grid squares. This may be particularly effective if you have a comparative problem you want to resolve—say, studying the relative distribution of bison bone types across a meat-processing site so you can speculate on butchering patterns. You may be able to arrive at a satisfactory interpretation informally, but a systematic recovery technique will help eliminate spatial bias that might deform the results.
If you make a mistake in your collection strategy, you are wasting time, perhaps money, and adding to the ever-increasing mountain of useless data. The same caution applies in the lab. Ask yourself this when you are tempted to wave a Munsell chart at a piece of debitage so you can nail down the color: “Does this type of data have any meaning in my analysis?” If the answer is no, give it a basic description and be done with it. The only reason for making collections in the long term is to provide material for someone else to study. Let him do it!
THE WAYS AND MEANS
Excavation recovers only a tiny fraction of what might have been a habitation, a production site, or other theater of cultural activity. The vagaries of preservation mean that many tools and tool parts made of perishable materials such as plant fibers, wood, leather, and bone may break down quickly under adverse conditions.
SURFACE RECORDING AND RECOVERY
An artifact exposed on the ground surface is usually not where the maker or user left it. Time is always a culprit, working with wind or water erosion, slope movement, plowing, and other natural and cultural forces. Because you tend to learn most about people, places, and events from the spatial relationships among material things, surface finds will probably not yield as much information as artifacts in situ. Sometimes, however, surface finds might be the only data you have. It is a rare archaeologist who has not found artifacts side by side that are separated by thousands of years. Such mixed assemblages are palimpsests, a word we’ve adapted from the printing trade to describe artifact assemblages that have become jumbled because either there was no interval of soil deposition between occupations, or the soil that did separate the occupations eroded away.
If this is the situation you face, all is not lost. Temporal and functional diagnostics such as identifiable projectile point or ceramic types, scrapers, or fire-cracked rock can help you build a picture of what happened. You may also be able to map the “horizontal stratigraphy,” the spatial distribution of materials across a locality, if the artifacts may have remained more or less in the same place. This is where knowledge of the terrain and the geomorphology are invaluable. You need to be certain that a flood, sheetwash, or slope movement did not transport your artifacts from elsewhere. Finally, do not assume that all the artifacts in an assemblage derived from similar activities. Ponder this question: A toolmaker’s debitage may be identical at two different places, but do these scatters offer up the same interpretation if one of the sites is a hunting blind and the other is a staging ground for ceremonial activities?
An important advantage of a surface scatter is the fact that you can map the entire exposed assemblage, if you need to, piece by piece, and therefore quickly define its extent (see Toolkit, volume 2). Careful mapping is key to any subsequent excavation. Of course, cultural material is scattered across the landscape, so what you are really doing is isolating a discrete locus of cultural activity. In a reconnaissance survey, you may find it helpful to stick a pin flag in the ground at every artifact location or, if the finds are too dense, on the outlying edges of a scatter. Pin flags are great because they allow you to see patterning in artifact distribution, but they also call attention to the site. If you are concerned that you are advertising a site location, then use discrete markers.
Once you have surveyed the entire surface, you can map in the artifacts, features, and aspects of the physical setting. If precise spatial relationships are crucial, plot every artifact and feature with a Total Station or theodolite. With very little effort you can generate a computer-aided design (CAD) image from the locational information, giving you a precise map of the area. If you only need to map the general location of artifact clusters or features, a geographic positioning system (GPS) unit is ideal. A unit with real-time differential correction will allow you to map points, lines, and polygons (enclosed shapes) with submeter precision; one without this feature will still get you within several meters, which may be okay for a reconnaissance survey.
When you have finished mapping and taking photographs, you need to invoke your collection policy and get to work. An important point: You may be the last person to see artifacts that you do not collect, especially if you are surveying in a proposed construction area. It is therefore a good idea to have a field identification procedure in place so that you can quickly record some basic data on the artifacts observed (e.g., type of object, material, function, relative size, cultural attribution). You can then pick up the objects you plan to take to the lab for further study.
There is no point in collecting for its own sake. We already know that people have left traces of daily life on the landscape for thousands of years. We do not need any more proof. Having said that, you may want to save something because it is old and uncommon, even though you know it is not going to add significantly to your study. We cannot cling to every element of the past, however, because if we did, we’d have little room to move! Never forget that the traces you find usually relate to someone else’s ancestors. The descendants may not just want you to be respectful—they may not want you to interfere in any way at all, despite your legal rights or obligations. Sometimes you simply have to let things go. If you do, don’t get upset when collectors move in, pick up what you left, and complain that you “missed” things.
MOVING DIRT
When you need to explore below the ground surface, you have to resolve the fundamental problem: your lack of X-ray vision! Deciding where to dig takes careful observation and planning. You may have already taken the first step, assuming that the artifacts on the surface reflect the positioning of any below. As this may not be the case, depending on the site taphonomy, the next step is to devise a technique that provides you with a good general impression of the subsurface. If your project area has cutbanks, gullies, rodent burrows, or even anthills, you can learn quite a lot without digging. If you are in an agricultural area, the county soil map will outline the soil characteristics and formation processes. Or you might be fortunate to be in an area that a geomorphologist has already analyzed. If you can, conduct a magnetometer, soil resistivity, GPR, or other geophysical survey. The information you retrieve from these preliminary efforts could lead you from aimless digging for that elusive needle in the haystack to the ground truthing of potential features. At the very least, you should have shovels and a soil corer or auger, so you can explore the subsurface in the traditional way.
The goal of subsurface data recovery is not only to retrieve artifacts but to expose the traces of cultural activity remaining on what were once ground surfaces. Visualization of these buried contexts is very difficult at the best of times. Over the tens, hundreds, or thousands of years separating components, there may have been any number of impacts that added, removed, or otherwise disturbed the deposits. It is therefore extremely advantageous if you can expose a large subsurface area quickly to help inform your subsequent excavations.
At a recent University of South Dakota (USD) Archaeology Laboratory project at Fort Riley, Kansas, for example, we faced the difficult problem of interpreting the cultural resource potential of a large open field and surrounding woods at the confluence of two creeks, an area of more than fifty acres (Molyneaux et al. 2002). To complicate matters, we knew from previous excavations that there were likely to be buried soils containing cultural material several thousand years old and numerous shallow disturbances associated with the World War I barracks and the more recent firing ranges that once occupied the site.
Rather than pick away at the surface to little effect, given the size of the site and the potential depths of cultural material, we decided to excavate a number of backhoe trenches. The goal was to expose the stratification sequence and determine whether any of these strata contained cultural material. This would inform the main testing, to be done with an eight-inch power auger that could probe more than ten meters into the ground if necessary. Given the size of the area to be explored, we did not mind that several areas approximately 4 × 2 meters each were going to be destroyed by the backhoe, perhaps disturbing cultural material. It was a risk worth taking, because it would give us a very good idea of where we should conduct our less destructive tests.
As it turned out, the backhoe trenches exposed three different buried components across the project area, ranging from a near surface component of the Central Plains tradition (surface to about fifty centimeters), to Middle to Late Woodland (one-meter deep), to Late Archaic (two-meter deep) occupations below that. We were then able to conduct auger testing much more carefully and effectively than we could have without the trenching.
You may need to use heavy equipment for other reasons as well as when you must remove overburden quickly because of time constraints (i.e., a construction crew’s bulldozers are idling in the background). Mechanical stripping, using a backhoe, road patrol, or belly scraper, may expose artifact concentrations or features very quickly but at the risk of damaging portions of features and artifact scatters.
Data recovery via hand excavation is obviously a better strategy if your goals are to record material in three dimensions so you can learn more about time and space relationships. The most efficient method is shovel skimming (a.k.a. shovel scraping or shovel shining), which is removal of soil with a very sharp shovel held at an acute angle to shave off a portion of the surface. An adept shoveler can remove earth in such fine levels that very little will be missed. If you risk disturbing delicate remains or missing smaller artifacts, it is best to trowel by hand.
The choice of technique depends in part on what you expect to find and what you intend to record. Clearly, there is a difference between excavating an artifact scatter and a hearth or other intact feature. In the former case, a shovel may suffice; in the latter, very light troweling and brushing are in order.
DRY SCREENING
Whatever method you use, you always miss some artifacts, so the next step is usually screening of the soils you remove. Sifting excavated soils through a quarter- or eighth-inch screen is so simple when you are sitting in your office using your field school knowledge to create regulations for working archaeologists and so miserable when you are standing at a screen trying to break up hard lumps of earth or push damp clay through the mesh. It was a sly and cynical wag, no doubt, who gave the name “Love Slave” to one of the USD shaker screens. Screening, however, is a necessary task, because most data recovery techniques—except brushing, perhaps—miss some of the artifacts.
Carmichael and Franklin (1999) conducted a screen loss experiment at the KT site in New Mexico. Their results showed that quarter-inch screens may capture as little as 5 percent of lithic artifacts and miss many diagnostics, such as notching flakes, biface thinning flakes, initial edging flakes, and exotic raw materials. Surprised? USD archaeologists made a rather rare find in the region, a piece of obsidian from a sixty-mile highway corridor along the Floyd River in northwestern Iowa; the piece, recovered during the mechanical grade sorting of 10 percent sample bags of earth from excavation units at site 13PM407, was a piece of microdebitage of grade 5 (0.0469 inch, or 1.18 millimeters) (Molyneaux et al. 1996). The implications of the find were rather monumental in their simplicity: This chance excavation of a tiny part of a site along a large highway corridor yielded an artifact that someone chipped from a tool of material that originated far from this knoll along the Floyd River. Without screening we would never have been able to add to our knowledge of obsidian distribution and use in the region.
Basic screen design has not changed for generations. The most common form is the wooden box with a hardware cloth mesh, suspended from a tripod for long-term excavations or mounted on two legs with two handles for field surveys and short-term excavations. There are debates about which types are the most useful and easiest to use. If you work in an urban setting on large excavations and have dry earth to screen, you might consider a mechanical unit. A Gilson grade sorter with five screens will quickly reduce a pail of dirt to neatly sorted material and a tray of dust at the bottom. You might also search the websites of the industrial screening companies for innovative ideas. There are new mesh designs that work better with damp soils than the traditional square pattern; perhaps these would improve those long messy days of struggling with clay.
WATER SCREENING AND FLOTATION
Water screening is the best method for ensuring more complete recovery of cultural material, as you literally wash away the soil. It is an especially good way to retrieve small artifacts and to track artifact loss through dry screening.
A number of different water screen set-ups are in use. The most primitive method combines a high-pressure pump, a hose, and ordinary box screens. Self-contained sluice boxes and barrel systems (steel drums, fifty-five U.S. gallons or forty-five imperial gallons) pump water internally to stir up the soil. The barrel types may also double as flotation units if the light fraction can run off into a set of sieves. Do not assume that water screening makes life simple. Clay will sit in the screen unless it is physically disturbed. Pumps can be finicky. You will get wet and muddy. And you might find that there are regulations preventing the discharge of your soil back into the water.
Flotation is simpler, but don’t misunderstand: If done properly it can also be very complex. You can buy or jury-rig a plastic bucket system with an inlet at the bottom and a spout at the top that draws off the waterborne fraction into a stack of sieves, or you can just fill a pail with water, dump the dirt in, stir it around, and scoop up the material with an ordinary kitchen sieve. Another approach is to pour it into a fine-mesh screen or cheesecloth. (See Toolkit, volume 5, for details on fine-screen recovery and processing methods.)
These are wonderful methods if you have carefully thought out why you are using them. If you retrieve a lot of light or heavy fraction, you may be facing long hours at a binocular microscope carefully separating the artifacts from the debris. Make sure these data will be useful. There is no point collecting seeds and other plant material unless you can differentiate between material associated with the site occupation (s) and modern material; there is equally no point in collecting microdebitage if you are only adding the numbers to your inventory. Are you going to use microdebitage to test your recovery methods? Do you think that the presence of microdebitage might indicate the precise spot that a knapper worked or reveal areas where larger pieces of debitage are absent because tool renovation (e.g., sharpening) rather than production took place there? You would miss such potentially significant data using conventional recovery methods.
CONCLUSION
Collection strategies require careful planning, derived from the key questions in your research design. Failure to do effective planning can overwhelm you with useless data or cause you to miss important data. The truth is that most of us don’t think that carefully about our approaches to data recovery and come to regret it once we get to the lab. We need to be more deliberate in our approaches to use our time, money, and other resources effectively.
A final element in data collection, discussed in chapter 6, relates to a variety of specialized types of data collection, ranging from samples for dating to human remains.