October is Virginia Archaeology Month. It offers an opportunity to learn about and celebrate archaeology in the state -- including at Monticello. Before COVID the local focus made archaeological research accessible: you could visit sites and labs, chat with people making discoveries, and find out how you can participate. Those days will return!

But the local focus makes it easy to overlook that archaeology is a global discipline whose goal is to use artifacts and other physical traces of human behavior to figure out what happened in human history and why -- starting with the first appearance of stone tools 2.5 million years ago. This expansive view of archaeology is essential to our local research because models developed by archaeologists working on other times and places help solve puzzles at Monticello.

In this post, I want to explain briefly how analytical methods inspired by archaeological research on late twentieth-century campsites occupied by Aboriginal people in the Western Desert of Australia have been critical to deciphering household organization among enslaved agricultural laborers who lived at Site 6 and worked the fields on Monticello Mountain in the early nineteenth century. 

Site 6

First a bit of background about Site 6. We owe the discovery of Site 6 to our ongoing research initiative -- the Monticello Plantation Archaeological Survey. The Survey aims to locate every archaeological site on the roughly 2500 acres owned by the Foundation. This tract comprises about half of Jefferson's original land holding. The Survey is revealing a radical shift in locations of domestic sites and patterns of land use in the last decade of the eighteenth century. From 1740 to around 1790, first Peter Jefferson and then his son Thomas grew tobacco. In the early 1790's Thomas made the transition to wheat, hoping to take advantage of increasing prices in Atlantic markets. A few years later he began to grow tobacco and wheat simultaneously.

The simple hoe-based monocrop agriculture of tobacco was replaced by a far more complex, diversified regime of wheat that relied on permanently cleared fields, plows, draft animals, crop rotations, manuring, milling to sell wheat as flour, and nail making. The Plantation Survey is revealing that because of the contrasting needs of tobacco and wheat cultivation there were actually two Monticello Plantations, each with distinctive settlement patterns, labor organization, and agricultural ecology.

Site 6 was part of the second plantation. A major goal at Site 6 is to understand how the more complex labor demands of diversified agriculture affected organization of slave households and the daily lives of their members. Lacking stone or brick foundations, the log houses in which enslaved people lived left few or no subsurface archaeological traces; we must rely on studying spatial variation in artifact densities to identify distinct households. The data that allow us to accurately measure that spatial variation are counts of different artifact classes or types in 5-foot excavation quadrats scattered across the site. We use statistical methods to plot the spatial signal that underlies the raw data as a contour map.

To see how this works, consider two contour maps (Figure 1). On the left is a map of the density of all ceramics at Site 6. The small black dots are the centers of the five-foot excavation quads -- there are over 150 of them! The color gradient represents artifact density -- low values in blue, high values in yellow. There are three major high-density zones or clusters at the site spaced about 100 feet part. We think these zones represent three houses -- almost certainly log cabins. The high-density zones are areas into which the residents of each house discarded its ceramic refuse over the years that they lived at the site.

On the right is a map of nail densities. Again, we see three major clusters. Their locations match the ceramic clusters. The match offers further support for the idea that the clusters represent refuse from different houses. There is a much smaller fourth cluster in the southwest corner of the ceramics map. This may represent pots that were broken by household members as they used the adjacent spring which was their main water source.

Figure 1
Diagram, mapDescription automatically generated
Figure 1. Contour maps of artifact densities in five-foot excavation quadrats at Site 6. Blue denotes low densities and yellow denotes high densities. There are three major high-density clusters on the map for ceramics (left) and these are matched by three major clusters on the map for nails (right). Each major cluster represents a separate household. The edge of a fourth, smaller cluster appears in the southwest corner of the ceramic map, adjacent to a spring which was the water source for the site's residents. (Note: On the graph, North is up and East is to right.)

A Surprise: Variation Among Three Households

The first step in studying household organization is to determine if all the households were occupied at the same time. To do that, we assigned each of the 150 quadrats on the site to one of four groups, based on which of the clusters on the ceramic density map the quad fell into. There are three groups that correspond to households: north, southeast, southwest. A fourth group includes quadrats adjacent to the spring that was the primary water source for the site's occupants (Figure 2).

Figure 2
Diagram, mapDescription automatically generated
Figure 2. Five-foot quads are assigned to one of four groups, based on where they fall within the major clusters on the ceramic density map (Figure 1). The North, Southeast, and Southwest Groups represent the three households on the site. A fourth group includes quads adjacent to the spring that served as the water source for the site's residents.

An initially promising way to see if the three households were occupied at different times is to estimate the proportions of ceramic types in each quadrat group whose popularity is known to have changed over time. We know that creamware was fashionable in the late-eighteenth century and pearlware was popular in the early nineteenth century. So temporal differences among the households should be reflected in different proportion of pearlware relative to creamware. Figure 3 shows that there are significant differences in pearlware proportions, with the north household having much higher proportions of pearlware, followed by the southeast household, and finally the southwest household. The quads next to the spring are intermediate.

Figure 3
Diagram, mapDescription automatically generated
Figure 3. Left: Estimated proportions of pearlware, relative to creamware, vary across the north, southeast, and southwest households. The dots represent the estimated proportions, while the vertical lines represent statistical uncertainty: 95% of the time, we expect intervals like this to capture the true proportion. Right: Estimated proportions of cut nails, relative to wrought nails, show almost no variation.

Based on the ceramic evidence, it is tempting to think that the north household is later than the southeast one and the southeast one is later than the southwest one. But we should be skeptical until we can evaluate this conclusion against independent evidence that might show it is wrong. That evidence is available from nails. In the first few decades of the nineteenth century, machine-made cut nails replaced hand-made wrought nails across Virginia. If chronological differences are responsible for the variation in pearlware proportions among the three households, we should see the same pattern in in the proportions of cut nails. But a glance at Figure 3 shows that cut nail proportions are nearly identical. That raises the surprising possibility that the three households are more-or-less contemporary and that variation in pearlware has a different cause. Perhaps the household differ in levels of access to resources or wealth. Under this hypothesis, high proportions of pearlware register the ability of the north household to acquire more costly, fashionable ceramics. And we find independent support for this hypothesis in other artifacts classes. I will describe this evidence is a future blog post.

But this raises a further question. Why might the northern household have been able to accumulate greater levels of wealth than the southern ones? Our current working hypothesis is that members of the northern household enjoyed higher probabilities of continued residence at Site 6, from one season to the next, than the residents of the other two households. Continued residence at the site meant that they could count on being around to reap the rewards of investments with temporally delayed rewards. We know from documents that gardens were one of the few means that people enslaved at Monticello had to earn money. Fencing, planting, and caring for a garden make sense only if you know you will be around for the harvest. But where can we find archaeological evidence for variation among households in their member's chances on continued residence at the site to evaluate this idea?

Artifact Size Sorting in the Western Desert of Australia

This is where evidence from archaeological research with contemporary Aboriginal groups living in Australia's Western Desert becomes important. In the 1970s, archaeologists began to realize they could sharpen their understanding of the past if they knew more about the principles that govern how people dispose of trash in contemporary contexts, where they could collect data on the trash and the behavior that generated it. Jim O'Connell, recently retired from the University of Utah, did pioneering research in this field by living and working with the Alyawara, an Aboriginal people who live in Australia's Western Desert.

Figure 4
Diagram, mapDescription automatically generated
Figure 4. Left: Shaded contour map of artifact densities at a camp occupied by Alyawara foragers in the Western Desert of Australia. Archaeologist Jim O'Connell lived at the camp and mapped the artifacts that its residents threw away. The highest densities of artifacts, shown in grey, are found around the hearths (denoted by black dots) and the adjacent living spaces. Right: O'Connell's map of median artifact size. Areas with larger artifacts, shown in grey, encircle the living area of the site. Cleaning up moved the larger artifacts to the edges of the living area.

O'Connell's research has two important lessons for us. First, when people can count on residing in a place for a long period and their continuing residence requires economically important household tasks, they invest effort in cleaning up trash. But for short-term occupations people do little cleanup. Second, cleaning up is identifiable by looking at spatial patterns in the distribution of artifacts of different sizes across a site, not in raw artifact densities. Figure 4 shows an artifact density map for a long-term Alyawara camp. The black dots are hearths. Densities are highest around the hearths and we see no evidence for cleaning up. Compare that pattern to the plot of the size of the artifacts -- the grey zones have larger artifacts -- on the right panel of Figure 4. Check the distinctive donut pattern: small artifacts in the central high-density zone and large artifacts on the periphery. This size-sorted pattern emerges because when people clean up, they get the big stuff, and miss the small bits. O'Connell's conclusions have been confirmed by archeological research around the world, with diverse contemporary groups including Eskimo foragers, Maya villagers, and University of Arizona undergraduates.

Artifact Size Sorting at Site 6

Back to Site 6. If the northern household is a long-term occupation with a complex household economy, we should expect to see the distinctive size-sorted spatial pattern that betrays cleaning up. If the two southern households were comprised of a succession of short-term occupations, we expect to see little evidence for cleaning up.

Figure 5
Diagram, mapDescription automatically generated
Figure 5. Left: Map of the density of wrought and cut nails at in five-foot quads at Site 6. The density peaks mark the location of the three households. Right: Map of the proportion of "large" nails in each quad. Note that circular blue zone with low proportions of large nails falls in the same location has the northern high-density peak. It is surrounded by higher proportions of large artifacts. This is the donut pattern of size sorting that betrays cleaning up. It is unique to the northern household. The map of the proportion of large nails show no evidence for size sorting around the two southern households.

Figure 5 shows that that the Site 6 data fit these expectations. In the left panel, is a map of the density of wrought and cut nails. On the right, is a map of artifact size. To make it we computed the geometric mean of the weights of all wrought and cut nails recovered from the site. We then counter the numbers of "large" and "small" nails in each quad, based on whether they were greater than or equal to the site-wide mean or less than the site-wide mean. Finally, we mapped the proportion of large nails in each quad. The results are stunning.

On the map of the proportion of large nails, when we check the area where the north household was located, we see a circular blue zone with low proportions of large nails is surround by a yellow zone of with high proportions of large nails. This is the donut pattern of size sorting that O'Connell documented at the long-term Alyawara camp. It is the signature of cleaning up. In contrast when we look at the areas where the southeast and southwest households were located, we see no evidence of size-sorting.

Archaeological research at Site 6 gives us new insights into variation in the everyday experience of people enslaved at Monticello. We have discovered that members of the northern household were able to negotiate some degree of residential stability. We do not understand how that process worked. Greater certainty about the future made it smart to invest effort in gardens and independent household economies, maintained yards, and to signal some of their success in more stylish ceramics. On the other hand, the uncertain futures faced by residents of the southern households made investments whose payoffs lay in the future unwise. As a consequence, we see less effort in yard maintenance and fewer stylish ceramics.

This leads us to larger questions about changing character of slavery: were the less predictable futures forced on residents of the southern households a consequence of new slave labor management strategies linked to diversified wheat production? Diversification would have created increased demand for labor at spatially dispersed locations at critical times. Did Jefferson, and his overseers meet these demands by moving some workers more frequently around the landscape, resulting in greater levels of inequality among households that we see at Site 6? Do we see similar changes at other Chesapeake plantations? We do not yet have enough comparative data from other sites at Monticello or other plantations across the Chesapeake regions to answer those questions.

Acquiring those data will require more archaeological survey and more intensive household-focused fieldwork. In other words, we need more of the local archaeological research of the kind that Virginia Archaeology Month and events like it around the country celebrate. But I hope this post has also shown why making sense of those local data requires engagement with a wider orbit of archaeological method that takes us far from Jefferson-era Virginia or other early-modern slave societies. Puzzling out the historical dynamics that lie behind local data requires understanding patterns of variation across cultures that are ultimately underlain by our common humanity.