Colonoware is a continuing enigma in the archaeological record of southeastern North America. A hand-built, low-fired pottery whose vessel forms sometimes mimic distinctive elements of imported European ceramics, colonoware has been found as far north as Massachusetts (Sansevere Reference Sansevere, Gall and Veit2017), as far south as Florida, and as far west as Louisiana, Tennessee, and Kentucky (Figure 1). During the late seventeenth century, these locally produced, coarse earthenware vessels began to appear in domestic assemblages of enslaved households, free and enslaved Indigenous communities, and other urban and rural settings across this entire region. Although glossed as a single phenomenon, colonoware exhibits vessel-, site-, and regional-level variation that defies straightforward classification. Consequently, fundamental questions remain about who made it and how it was distributed and used, sparking ongoing and heated debates.
Colonoware examples: (a) Bowl from Yaughan 75 (38BK75); (b) partial bowl from the Pope Site (44SN180); (c) partial jar from Curriboo 245 (38BK245); (d) same vessel as 1c. Photos by Karen E. Price, Mount Vernon Preservation. (Color online)
Figure 1 Long description
The image A shows a bowl from Yaughan 75, featuring a wide, shallow design with visible cracks and a smooth surface. The image B shows a partial bowl from the Pope Site, displaying a rounded shape with a broken edge and surface imperfections. The image C shows a partial jar from Curriboo 245, characterized by a tall, narrow form with a large missing section and rough texture. The image D shows the same vessel as image C, presenting a different angle of the jar, highlighting its flat base and cracked surface.
What is clear is that colonoware is deeply contextual: it represents many traditions and serves as a material manifestation of related strategies developed in response to oppressive and violent colonial regimes (Sattes and Marcoux Reference Sattes, Marcoux, Marcoux and Sattes2024). Fashioned into bowls, jars, milk pans, cooking pots, and chamber pots, colonoware played key roles in the economic strategies, foodways, and cultural practices of those enslaved and subjugated by diverse and devastatingly effective European colonial endeavors, most notably the plantation system. In Virginia and South Carolina, in particular, hundreds of thousands of colonoware sherds have been excavated from the sites of enslaved quarters located on plantations throughout both regions. Steen (Reference Steen, Franklin and Fesler1999:100) estimates that millions of vessels may have been made in South Carolina alone. Colonoware production appears to have varied widely in scale, from the expedient production of vessels by households or individuals for their own use to manufacture and marketing on a regional scale.
Some scholars have argued that enslaved households used locally produced vessels during times when imported ceramics were not available to them for a variety of socioeconomic reasons (Galke Reference Galke2009:321; Greer Reference Greer2022:198). This is likely true in some cases. At the same time, archaeological and historical research demonstrates that enslaved and free Black households across the Southeast used the profits from a variety of economic endeavors to actively participate as consumers in the market economies of nearby towns and cities to acquire fashionable goods, including ceramics (Franklin Reference Franklin and Orser2001; Galle Reference Galle, Delle, Hauser and Armstrong2011; Heath Reference Heath and Jay B.1999; Hill Edwards Reference Hill Edwards2021; Schweninger Reference Schweninger1991). Additionally, colonoware has been identified in refuse associated with overseers’ houses and with middens and yard workspaces shared by enslaved and free laborers, as well as with wealthy planters and their families (E. Breen Reference Breen2004; Heath and Breen Reference Heath and Breen2009:7; Sattes Reference Sattes, Marcoux and Sattes2024; Veech Reference Veech1997), indicating that enslaved people were not the only users of this ware. Procured items, whether costly or not, were material expressions of economic and social strategies that varied according to labor patterns and differences in household economies (Franklin Reference Franklin2019). Yet, the persistent presence of colonoware alongside other commercial goods that presumably served the same functions leads us to ask these questions: Why did any individuals, enslaved or free, use colonoware when affordable imported ceramics and other locally produced functional equivalents were available? Was colonoware a substitute for imported or European-style ceramics, or did it play a unique role in plantation foodways? How did a potter’s choices regarding vessel paste composition, form, and decoration determine how it was used? Was colonoware morphology and function shaped by different needs in distinct regional contexts?
Just as the archives are primarily silent on the names of the men, women, and children whose labor powered colonial projects, it remains a challenge to unearth documentation of their daily activities and the objects they made and used. Unlike the detailed recordkeeping of the manufacture and marketing of ceramics imported by European colonists, locally made coarse earthenware production, marketing, and use took place largely outside the written record. The extensive fragmentation of most colonoware assemblages makes it difficult to identify diagnostic forms from individual sherds. Thus, the best evidence we have for understanding how colonoware functioned is in the minute physical attributes of archaeologically recovered sherds. Measurable variability in the attributes of handmade pottery, such as paste composition, vessel thickness and shape, decoration, and use wear, may reflect meaningful regional, cultural, or temporal differences. Standardized attribute-based comparative analyses can reveal broad trends but are also detailed enough to let us see fine-grained differences. Although earlier studies have documented important material patterning in colonoware, for the most part they have remained localized and specific, limiting synthetic interpretations.
As a first step toward determining why people chose to procure and use colonoware, we explore the role of function in determining the contexts of its use. Using data from the Digital Archaeological Archive of Comparative Slavery (DAACS; www.daacs.org), we evaluate the functional role of colonoware and how it was incorporated into domestic activities at different spatial and temporal scales. Although colonoware is traditionally defined by its imitation of European vessel forms, most assemblages are dominated by unvesselized sherds from unidentifiable forms. By shifting our analytical focus from complete vessels to individual sherds, we can interrogate a more representative proportion of assemblages to identify which sherd-level attributes show meaningful variation related to function or intended use.
To this end, we compare patterns in colonoware abundance, paste inclusion type and density, thickness, vessel morphology, surface treatments, and the presence of residue between Virginia and South Carolina. Although we consider that thickness, decoration, and form can provide clues to vessel function, we acknowledge that the accumulation of soot from use in a fire or charred encrustations’ sooting/residue is the only attribute in this analysis that provides direct evidence of function. We hypothesize that at an assemblage level, the other attributes point to the intended function of a ceramic vessel, given expected variation in utilitarian and tableware forms (Skibo Reference Skibo1992:35).
Documented differences in labor regimes, planter proximity and absenteeism, settlement patterns, and plantation products between the two regions modified the daily routines of plantation residents and their participation in the market economy (Morgan Reference Morgan1982, Reference Morgan1998; Walsh Reference Walsh2010; Wood Reference Wood1975). Prior studies have shown that households continued to use colonoware in South Carolina in large quantities into the nineteenth century, whereas colonoware consumption dropped dramatically over time in Chesapeake assemblages from both rural and urban contexts (Agha et al. Reference Agha, Isenbarger and Philips2012; Anthony Reference Anthony, Martha Zierden and Calhoun1986; Ferguson Reference Ferguson1992; Galle Reference Galle, Heath, Breen and Lee2017; Heath and Breen Reference Heath and Breen2009; Wheaton and Garrow Reference Wheaton and Garrow1985; Wheaton et al. Reference Wheaton, Friedlander and Garrow1983). We argue that functional differences in colonoware—whether it was used for cooking, processing, or serving/dining—may explain the variation between the two regions and ultimately why the colonoware tradition persisted for so long in some places (Cooper and Smith Reference Cooper and Smith2005, Reference Cooper and Smith2007). Evaluating these functional dynamics is a critical first step to developing more productive models and methods for understanding colonoware’s role in the lives of pottery producers and consumers.
Background
Defining and Contextualizing Colonoware
More than 60 years ago, Ivor Noël Hume (Reference Noël Hume1962) became the first archaeologist to identify colonoware, in archaeological assemblages from Williamsburg, Virginia. He argued that “Colono Indian” ware was primarily defined by the careful replication of European shapes in hand-built, unglazed coarse earthenware. He contended that it was produced by Indigenous groups for non-Indigenous markets, and other archaeologists adopted this definition (e.g., Binford Reference Binford1965; Deagan Reference Deagan1987:103–104; South Reference South1977; Vernon Reference Vernon1988:76). Subsequent excavations recovered abundant colonoware from enslaved domestic sites on plantations, non-elite farms, historic Native settlements, and urban lots and markets in cities like Charleston, South Carolina, and St. Augustine, Florida. As site diversity grew, so did the complexity of the colonoware assemblages found. By the 1980s, it was clear that only a minority of sherds exhibited clear elements from European ceramic forms, such as foot rings, feet, handles, marleys, and scalloped rims. Although examples of flat forms such as plates and platters do exist, bowls with flat or rounded bases and jars with rounded bases are the most common forms, with historical precedents in West Africa and the Indigenous Southeast (Ferguson Reference Ferguson and Schuyler1980, Reference Ferguson1992; Heath Reference Heath and Reinhart1996:150; King et al. Reference King, Gill, Strickland, Marcoux and Sattes2024:76).
The most abiding contention over the past three decades relates to the identity of colonoware manufacturers. Some scholars have tied colonoware’s appearance in mid-seventeenth-century Virginia to Indigenous potters’ response to European ceramic markets (Mouer et al. Reference Mouer, Hodges, Potter, Renaud, Noël Hume, Pogue, McCartney, Davidson and Singleton1999; Spivey Reference Spivey2017). The long-running Catawba project (e.g., Cranford Reference Cranford2018; Davis et al. Reference Davis, Riggs and Cranford2015; Fitts Reference Fitts2017; Plane Reference Plane2011) and studies of the Rappahannock’s (King et al. Reference King, Gill, Strickland, Marcoux and Sattes2024) and Pamunkey’s (Spivey Reference Spivey2017) engagement with the eighteenth- and nineteenth-century market economy have documented the successful production and region-wide marketing of Indigenous-made pottery that has continued into the present day. Others have argued that the rise in colonoware is correlated with increased importation of enslaved Africans in the American South (Deetz Reference Deetz1988, Reference Deetz1993; Ferguson Reference Ferguson1992). Recent research has identified certain colonoware decorations and forms that serve as material manifestations of the persistence and adaptation of African potting traditions (Sattes et al. Reference Sattes, Marcoux, Platt, Zierden and Anthony2020). It is most likely that both these arguments are true in particular contexts. Researchers have also acknowledged the difficulty in using archaeological data, and colonoware specifically, to identify ethnicity (Heath and Breen Reference Heath and Breen2009) or ethnic markers (Singleton Reference Singleton1995:133–134); instead, they have emphasized the role that hybridity and creolization played in the production and use of colonoware (Cobb and DePratter Reference Cobb and DePratter2012; Marcoux and Sattes Reference Marcoux and Sattes2024; Mouer Reference Mouer1993; Singleton and Bograd Reference Singleton, Bograd, Delle, Mrozowski and Paynter2000; Steen Reference Steen, Franklin and Fesler1999). To date, these long-standing debates about the identity of colonoware manufacturers remain unresolved because of lack of a large-scale scientific approach that relies on both standardized attribute identification and compositional analysis to identify the geographic sources of colonoware.
Regional Variation
Colonoware has been considered a shared phenomenon over the whole of the upper Southeast, including Virginia and South Carolina. However, each colony had its own trajectory of development (see Bradburn and Coombs Reference Bradburn and Coombs2011; Breen Reference Breen2001; Carr et al. Reference Carr, Morgan and Russo1988; Edelson Reference Edelson2006; Gallay Reference Gallay2002; Graham et al. Reference Graham, Hudgins, Lounsbury, Neiman and Whittenburg2007; Hatfield Reference Hatfield2004; Kulikoff Reference Kulikoff1986; Littlefield Reference Littlefield1981; Morgan Reference Morgan1975, Reference Morgan1998; Shefveland Reference Shefveland2016; Walsh Reference Walsh2010; Wood Reference Wood1975) into a “slave society” or what Berlin (Reference Berlin1998) identifies as a society fundamentally shaped by its relationship to systems of enslavement. The timing and nature of the violent displacement, resettlement, and coalescence of hundreds of thousands of Native peoples differed, as did the arrival dates and survival rates of European colonizers and enslaved Africans and African Americans. Environmental variation influenced the establishment and maintenance of plantation regimes around specific crops (tobacco and later wheat in Virginia; rice, cotton, and indigo in South Carolina), industries, and the growth and location of urban centers (Morgan Reference Morgan1998:102–124; Walsh Reference Walsh2010). We compare assemblages from Virginia and South Carolina because even though colonoware has played a fundamental role in interpreting plantation households’ activities in the quarters and workspaces excavated in these locales, we argue there is still much we do not know about the physical differences in colonoware between these two regions.
Although the daily routines and lives of enslaved laborers and plantation households in these regions shared some similarities, they were also particular to the regional ecology, demography, and specific demands of cash crops that came to dominate the plantation economies in each location. Scholars have long recognized differences in plantation settlement patterns in Virginia and South Carolina, such as varying proximity between planters and their laborers. In Virginia, smaller enslaved communities comprising a few households were dispersed across quarter farms—semi-independent agricultural production units situated away from the home plantation—whereas South Carolina plantations were characterized by larger, more nucleated villages of enslaved households. Whether enslaved workers were engaging in gang-based versus task-based labor regimes (Morgan Reference Morgan1998; Neiman Reference Neiman2008; Walsh Reference Walsh2010) also may have influenced if and how communities were producing and consuming colonoware. Understanding the impact of large-scale causal dynamics on the physical manifestation of colonoware, however, requires using a large dataset of sites from throughout these regions.
Our research was facilitated by DAACS: it was designed to enable a scientific, attribute-based approach to advance understanding of the historical development of slave-based societies with archaeological data (Galle et al. Reference Galle, Bollwerk and Neiman2019). In addition to analyzing and cataloging entire archaeological assemblages, DAACS collects fine-grained attribute data from locally made coarse earthenwares, such as colonoware at the sherd level. Equally important, DAACS data are structured so that ceramic and other artifact assemblages are organized according to archaeological provenience, thereby retaining contextual information.
Colonoware Dataset
DAACS currently makes publicly available a dataset of 46,927 colonoware sherds from 21 sites of enslavement in Virginia and South Carolina. This dataset, which includes form identifications, paste inclusions, sherd thickness measurements, surface treatments, and residue, serves as the backbone of this study (Figure 2; Table 1). An additional 19 Virginia sites published through DAACS, as well as one not published through it, are included: even though these sites lack colonoware sherds, they provide important data on the ubiquity of colonoware. We also entered colonoware attribute data from four sites into the DAACS database specifically for this study, but because of incomplete site data they are currently not publicly available through daacs.org (Table 1). We further supplemented the dataset with basic artifact counts of colonoware from 15 sites not in DAACS but available in published reports (Table 1). Our complete dataset includes data from 41 sites in Virginia and 19 sites in South Carolina with a total of 122,260 colonoware sherds.
Location of sites used in the study. Map by Karen Smith.
Figure 2 Long description
A map illustrating the locations of archaeological sites across Virginia, North Carolina and South Carolina included in the study. The map is marked with numbered circles indicating specific sites. In Virginia, sites are clustered in the eastern region. North Carolina has no marked sites, while South Carolina shows a concentration of sites along the coast. The map includes a scale in kilometers and miles and the Atlantic Ocean is labeled to the east. State boundaries are clearly delineated.
Colonoware Assemblages Included in Study.
Table 1 Long description
This table provides a comprehensive overview of 60 archaeological sites, focusing on colonoware assemblage counts, site types, regions, and time periods of occupation. Key data points include Middleburg Village in South Carolina with the highest assemblage count of 21,512 sherds, and Dean Hall with 58,985 sherds. The table also indicates whether sites were used in abundance and attribute analysesThe data is sourced from various references, and some sites have pending data availability through DAACS.org, indicating ongoing research and data compilation efforts.
1 For full citations, see Supplementary Table 1 and Supplementary Text 1.
2 A status of pending refers to sites where the field record and artifact information have been recorded but background site content development is still in progress so the site has not been made available to the public.
Most of these 60 sites are interpreted as domestic sites of enslavement associated with 19 different plantations. These include quarters for enslaved laborers, overseers, and yard workspaces / refuse middens. Five sites—the Accotink site, which included an overseer’s house and enslaved quarter; Pope site, a middling farmstead with a small population of enslaved laborers; Silver Bluff, a fortified town; Catherine Brown Cowpen, a dwelling, smokehouse, and cowpen; and Fort Frederick, a short-lived British military fort—did not serve exclusively as quarters or workspaces for enslaved people but provide comparative data. Indigenous production and use of colonoware are integral to this ceramic tradition’s broader history; however, the available dataset constrains our current analysis to examining colonoware’s function on domestic sites of enslavement. Overall, the sites are predominantly rural. Together, these sites have occupations whose mean ceramic dates (MCDs) span 1670–1840 (see www.daacs.org for individual site chronologies). Although Virginia sites dominate the first two quarters of the eighteenth century, most South Carolina and Virginia sites represented in the dataset have overlapping occupations dating to around 1760–1840.
Measuring Colonoware Abundance
Although colonoware was first described in the Virginia Tidewater (Nöel Hume Reference Noël Hume1962), archaeologists working in the Southeast have long focused on the remarkable frequency of colonoware in South Carolina, particularly in contrast to the amount recovered from Virginia plantations (Ferguson Reference Ferguson1992:9). However, high abundance values for ceramics can be influenced by any number of factors related to manufacture, use, and disposal patterns (Rice Reference Rice2015:216–219), and these factors can be difficult to untangle for inter-site comparisons. Some archaeologists have interpreted high frequencies of colonoware in an assemblage as an indication of on-site manufacture (Wheaton et al. Reference Wheaton, Friedlander and Garrow1983:237–239), using the rationale that pots were being expediently replaced when needed. Higher overall abundance may also be a result of vessel function. Cooking pots, for example, have a shorter use life than storage vessels because of more consistent exposure to thermal and mechanical stresses (Shott Reference Shott1996). Other prior research has interpreted higher frequencies of colonoware as evidence of the presence of low-status or enslaved individuals who lacked market access (e.g., Galke Reference Galke2009; Hacker and Adams Reference Hacker and Adams1993:114). Conversely, others view lower colonoware abundance as an indicator of acculturation (Wheaton et al. Reference Wheaton, Friedlander and Garrow1983:334–336), with proximity to markets and the low cost of European ceramics driving a consumer shift away from colonoware (Trinkley et al. Reference Trinkley, Hacker and Adams1995:205–206).
Underpinning all these interpretations is the assumption that when plantation households and enslaved laborers had access to imported ceramics and iron cooking pots, they no longer needed to use colonoware. If acquiring colonoware was a consumer choice inversely related to the ability to participate in the broader market for more costly imported goods, we would then expect an inverse relationship between the use and discard rates of colonoware and imported ceramics in contemporary assemblages. This pattern should be found not only in South Carolina but also in any region in which those same market forces were at work and access to markets was comparable. But how do we measure variation in use and discard rates using archaeological data or infer consumer choice, given the complexity of ceramic acquisition and discard processes?
The standard relative frequency equation for a given ceramic type
\begin{align*}
{\text{Relative Frequency for Ceramic Type}}\,i = & {\text{ Count for Ceramic Type}}\,i\nonumber\\
\qquad\quad & /{\text{Sum of Counts for All Ceramic Types}}
\end{align*}is problematic for two reasons. The first is the spurious negative correlation between the relative frequency estimates among ceramic types because frequency sums to 1 (100% of an assemblage). Thus, as counts of one ware type increase, the relative frequency of others decreases (Aitchison Reference Aitchison1981; Pearson Reference Pearson1897). Second, if we expect relative frequencies for a type to scale with its discard rates, then we rely on the implicit assumption that the discard rates for all the other types that contribute to total counts (denominator) are constant. However, it seems likely that cultural processes, such as function, availability, and occupation span, affecting discard rates for colonoware will affect rates for the other ware types as well. In fact, these kinds of linked effects are precisely what we want to measure: For example, do acquisition, use, and discard rates for colonoware decline in time or across space as rates for imported ceramic types increase?
Galle (Reference Galle, Delle, Hauser and Armstrong2011, Reference Galle, Heath, Breen and Lee2017) argues for a more robust relative frequency calculation called an abundance index (AI): it uses a denominator value for a second artifact class whose discard rate may more reasonably be assumed to remain consistent across the assemblages being compared, thereby allowing for a more accurate estimate of discard variation. We compute the AI by dividing the count for a given ceramic type by the sum of its count and the count for the second (nonceramic) artifact type:
\begin{align*}
{\text{AI for Ceramic Type}}\,i = & {\text{ Count for Ceramic Type}}\,i /{\text{ }}( {\text{Count for Ceramic Type}}\,i \nonumber\\
\qquad & + {\text{ Count for Artifact Type}}\,j)
\end{align*}Galle (Reference Galle, Heath, Breen and Lee2017) identifies wine bottle glass as a reasonable option for that denominator type. This consistency in discard rates is expected, given the use and reuse of wine bottles as storage containers for liquid, and it is empirically supported by the ubiquity of wine bottle glass across sites and excavation contexts within them. If wine bottle discard rates are relatively constant, increases or decreases in the AI values should scale with increases or decreases in ceramic discard rates. Galle’s (Reference Galle, Heath, Breen and Lee2017) application of the AI revealed that although some sites in her DAACS sample had high discard rates for colonoware and other sites had very low rates, all sites had high AI values for refined and imported ceramics, implying high discard rates of these wares.
Generating abundance indices on our expanded dataset, we saw similar trends to those identified by Galle (Reference Galle, Heath, Breen and Lee2017). Colonoware AI values for Virginia sites, with a few exceptions, were significantly lower than colonoware AI values for South Carolina sites (Figure 3a). In contrast, the AI values of imported refined wares, identified here as refined earthenwares (Figure 3b), refined stonewares (Figure 3d), and porcelains (Figure 3f), were greater for both Virginia and South Carolina sites dating to the last half of the eighteenth century. In fact, the only consistent difference in ceramic AI values between South Carolina and Virginia sites is for colonoware.
Abundance index scores for different ceramic categories by site with 95% binomial confidence limits plotted against blueMCDs (Bates et al. Reference Bates, Galle, Neiman, Hauser and Wallman2020) to evaluate potential temporal trends. The abundance index was calculated using wine bottle glass (WBG) as the constant.
Figure 3 Long description
The image contains six graphs labeled a to f, each showing abundance index (AI) scores using wine bottle glass (WBG) as a constant, plotted against blueMCDs. Graph a shows Colonoware, graph b shows Refined Earthenwares, graph c shows Coarse Earthenwares, graph d shows Refined Stonewares, graph e shows Coarse Stonewares and graph f shows Porcelain. Each graph includes data points for South Carolina and Virginia, represented by different symbols. The x-axis is labeled blueMCD and the y-axis is labeled AI (WBG).
Figure 3b also illustrates a sharp increase in the abundance of refined wares over time in both South Carolina and Virginia, a trend observed on sites across the American South and the Caribbean. Scholars have linked the increase in refined earthenwares to the consumer revolution and rising demand for and the availability of nonessential goods (e.g., T. Breen Reference Breen2004; Carson et al. Reference Carson, Ronald and Peter J.1994; Galle Reference Galle2010, Reference Galle, Delle, Hauser and Armstrong2011, Reference Galle, Heath, Breen and Lee2017). If the discard rate of colonoware were tied to the availability of alternatives such as refined earthenwares, we would expect an inverse relationship in the abundance indices; however, that was not observed in our analysis.
Our colonoware abundance analysis shows that the use of these locally produced wares persisted in the eighteenth and nineteenth centuries in both regions, despite the introduction and increasing availability of refined earthenwares, stonewares, and other types of coarse earthenwares. Consistently lower discard rates in Virginia compared to South Carolina suggest that plantation consumers in Virginia did not need or want as much colonoware, or it was not as readily available as it was in South Carolina. However, there is a suggestive relationship between isolated enslaved households in Virginia and higher colonoware discard. At the Pope site in a remote part of Tidewater Virginia (Reinhart et al. Reference Reinhart, Ackermann, Davis and White1987), the higher discard rate of colonoware associated with two domestic earthfast structures may stem from proximity to a local production community or may indicate a lack of market engagement to access European ceramics by those laboring at the farmstead. The Pope site is located within the ancestral home territory of the Nottoway Indian Tribe of Virginia and Cheroenhaka (Nottoway) Indian Tribe, who were known to produce colonoware (Binford Reference Binford1965) and may have offered a local source for necessary household wares. The inter-site abundance variability suggests that factors such as the location and structure of a site and its adjacency to other households and communities affected formal market participation and deserve more attention in future research.
Overall, the persistence of colonoware use and discard, particularly in South Carolina, demonstrates that refined earthenwares and other imported ceramics were not direct functional equivalents for colonoware. These trends suggest that market access was not the limiting factor in South Carolina. If imported refined earthenwares and utilitarian wares remained accessible to enslaved laborers in both regions, then a different causal factor must explain higher colonoware discard in South Carolina in the early nineteenth century.
Although the majority of assemblages in our study were excavated from enslaved quarters on larger or middling plantations, generally other site types fit within the same regional trends. For example, in South Carolina, Fort Frederick, Silver Bluff, and Catherine Brown Cowpen, a fort, trading post, and a backcountry farmstead, respectively, all had high colonoware discard rates that were within the range found on South Carolina plantations, indicating that the demand and availability for colonoware extended to a variety of communities laboring under different circumstances. Notably, the South Grove Midden at Mount Vernon in Virginia and the South Flanker well from Drayton Hall in South Carolina contained refuse associated with the planter’s household, plantation operations such as a dairy, and housing for enslaved people. The presence of colonoware milk pans alongside European- and Euro-American-made milk pans in the South Grove Midden assemblage (Breen Reference Breen2013) suggests that the plantation dairy may have had demands for milk pans above what could be fulfilled through Euro-American marketing channels. In contrast, the prevalence of bowl forms in the Drayton Hall refuse is likely tied to their functional versatility (Sattes Reference Sattes, Marcoux and Sattes2024). These patterns emphasize that colonoware consumption was a pragmatic choice at various times for households of all social classes.
We argue that these trends suggest that consumers in South Carolina continued to use colonoware because it met a distinctly different need that could not be filled by imported, industrially produced ceramics. It was also likely a need that was not as pressing to enslaved consumers in Virginia: the low discard of colonoware in that region indicates that the majority of enslaved laborers were able to fulfill their daily food preparation, consumption, and storage needs using industrially produced ceramic types or nonceramic pots. This raises the question of how and why colonoware was being used in South Carolina, if not as a substitute for imported wares.Footnote 1
Understanding Colonoware Function
To begin to answer this question, we conducted a series of sherd-based analyses to identify functional differences in colonoware from a subset of 21 domestic sites for which we had detailed colonoware attribute data. Determining functional differences among locally made coarse earthenwares is complex and often ambiguous. However, aspects of function can be inferred through measurable material correlates, such as inclusion type and density, thickness, form, surface treatment, and residue. Correlations among attributes can help us gain confidence in how people may have used these vessels.Footnote 2
Trade catalogs, merchant and manufacturing records, and kiln site remains relating to the manufacture and marketing of industrially manufactured imported ceramics demonstrate that European ceramics were produced in standard forms and sizes intended for specific functions (Beaudry et al. Reference Beaudry, Long, Miller, Neiman and Stone1983; Bloch Reference Bloch2016; Griffin Reference Griffin2005; Romaine Reference Romaine1960). No such archival record exists for colonoware manufacture. Instead, research on the performance and use characteristics of handmade ceramic vessels can identify diagnostic physical attributes of a vessel’s specific function, such as cooking, food service, and storage.
For decades, archaeologists cataloging and analyzing colonoware at DAACS have recorded these key attributes, as specified in the DAACS cataloging protocols (Digital Archaeological Archive of Comparative Slavery 2018:31–48). As a result, we have access to standardized data at the sherd level on paste inclusions, sherd thickness, vessel form, surface treatment, and the presence of residue/sooting, among other features. Our review of these diagnostic features is organized following the chaîne opératoire or pottery production sequence (e.g., Gosselain Reference Gosselain2000; Marcoux et al. Reference Marcoux, Sattes and Sherard2023; Rye Reference Rye1981) as we consider how functional needs may have determined the material choices of potters.
Paste Inclusions
Inclusion type and density are early-stage production decisions intimately tied to vessel function. Whether intentionally added as temper or naturally present within the clay, the type, shape, and density of aplastic inclusions affect the performance characteristics of vessels. Specific inclusions may increase efficiency in conducting heat or increase resistance to thermal and mechanical shock (Braun Reference Braun, Moore and Keene1983:122–123; Bronitsky and Hamer Reference Bronitsky and Hamer1986; Rice Reference Rice2015:328–330; Skibo Reference Skibo2013:36–41; Steponaitis Reference Steponaitis1984). Inclusions also contribute to mechanical strength, allowing for larger vessel size. However, little attention has been paid to functional characteristics of the different tempering types or materials selected for colonoware.
Typological analyses of colonoware have reported a variety of aplastic inclusions within the clay matrix of colonoware from Virginia and South Carolina, such as shell, quartz sand (<0.5 mm diameter), quartz pebbles or grit (>0.5 mm diameter), and mica (Anthony Reference Anthony, Martha Zierden and Calhoun1986; Brilliant Reference Brilliant2011; Heath Reference Heath and Reinhart1996:167–168; Melcher Reference Melcher2009; Veech Reference Veech1997). The most common inclusions in colonoware from both regions are quartz grains of various sizes (aka sand/grit) and mica. Both are naturally occurring throughout the study regions and are often natural constituents of clay formations. Generally, it is difficult to determine whether these aplastics are natural inclusions or added tempers. However, crushed shell, found in some colonoware vessels, most often presents as an intentionally prepared temper. It is likely a continuation of Indigenous Mississippian and Late Woodland potting traditions in the region (Henry Reference Henry1980; Neiman Reference Neiman, Franklin and Fesler1999). Pamunkey Indian potters in Virginia, among others, are known to have used burned and crushed shell as part of their production process (Speck Reference Speck1928; Spivey Reference Spivey2017). This temper type is far more common in Virginia, making up 15.2% (n = 125) of colonoware in the Virginia dataset but only <1% in South Carolina (n = 8).
The proportion of aplastics to clay also affects a vessel’s thermal properties, with higher proportions shown to better mitigate thermal shock (Bronitsky and Hamer Reference Bronitsky and Hamer1986; Tite et al. Reference Tite, Kilikoglou and Vekinis2001). Virginia colonoware tends to have lower inclusion / temper density overall (Figure 4a). Vessels in South Carolina are more likely to have abundant quartz inclusions, constituting greater than 15% of the paste. There are exceptions, however. For example, the Fort Frederick, Silver Bluff, and the Smith Plantation assemblages show much higher proportions of sherds with very low temper density. This variation could reflect the different activities and living patterns pertinent to these site types versus those more common to plantation life. The fort, where soldiers consumed provisioned rations, would have required more serving and tablewares, whereas the trading post likely served as a hub for commercial transactions (i.e., the exchange of wares made for market). These sites could have had similar functional requirements for serving and tablewares, in contrast with plantation domestic settlements where cooking would have been more common.
(a) Proportions of inclusion density categories by project; (b) proportions of inclusion density categories by inclusion type groupings. Inclusion density measurements are as follows: Very Low Density = < 5%; Low Density = ≥ 5% and < 7.5%; Medium Density = ≥ 7.5% and <15%; High Density = ≥ 15%.
Figure 4 Long description
The image contains two bar graphs. The first graph on the left shows proportions of inclusion density categories by project, divided into South Carolina and Virginia. The x-axis lists various projects, while the y-axis represents proportions. The second graph on the right displays proportions of inclusion density categories by inclusion type, also divided into South Carolina and Virginia. The x-axis lists different inclusion types and the y-axis represents proportions. Both graphs categorize inclusion density into very low, low, medium and high density.
A comparison of inclusion type groupings with density categories by state reinforces that low inclusion density is more common in Virginia colonoware but occurs among all different combinations of inclusion types (Figure 4b). Among South Carolina sherds, high density is more common but again is not necessarily strongly tied to inclusion type. Although there are some exceptions, the differences in density proportions between the regions largely align with our expectations that South Carolina colonoware may indicate more utilitarian functions, whereas that in Virginia was more suited to serving and consuming.
Vessel Thickness
Vessel thickness is influenced by numerous factors, including the working properties of the raw materials, a potter’s skill, and the chosen manufacturing technique. However, the intended function of the vessel is a key determinant of vessel thickness as well. Variations in vessel thickness directly affect a vessel’s ability to conduct heat and withstand mechanical stress and thermal shock (Braun Reference Braun, Moore and Keene1983:118; Müller et al. Reference Müller, Hein, Kilikoglou and Day2013; Rice Reference Rice2015:330). Thinner walls usually have higher thermal conductivity (Braun Reference Braun, Moore and Keene1983:118), and thicker walls are less prone to breakage. Thus, we may expect cooking vessels to be somewhat thinner on average than vessels suited for food storage and bulk processing. Thin vessels are also expected for tablewares, mimicking industrially made European ceramics. Assemblages composed of vessels with a range of thicknesses are more likely to indicate a pottery tool kit with multiple functional needs; conversely, an assemblage limited to only thin or only thick vessels may represent more narrow usage.
Mean sherd thickness is notably consistent across South Carolina sites, with individual assemblages exhibiting substantial internal variation but the same general spread (Figure 5). In contrast, mean sherd thickness across Virginia sites is highly variable, with some sites dominated by thin vessels and others by thick ones. Although this trend may be exacerbated by smaller sample sizes in Virginia, it suggests that some assemblages are dominated by utilitarian vessels and others by tablewares. Conversely, South Carolina assemblages may more consistently mix these two functions. In addition, South Carolina assemblages are thicker on average (mean = 6.36 mm) than those from Virginia (mean = 5.5 mm), indicating consistently higher frequencies of thicker vessels and thus a wider range of vessels intended for cooking, storage, or processing. In contrast, the thinner sherds more consistently found on Virginia sites could be tied to more limited uses as tablewares or for other functions where durability was not the primary concern. This interregional pattern in assemblage-level sherd thickness mirrors the data on whole colonoware vessels presented by Ferguson (Reference Ferguson1992:52). We also note that the sites in Figure 5 are ordered by time, and there is no apparent temporal trend in sherd thickness in either region.
Comparison of median sherd thickness by site. Sites within each region are ordered by blueMCD. Black horizontal line represents 1.80, the dataset-wide mean that equates to ∼ 6.36 mm. Red horizontal lines (Virginia, 1.7; South Carolina, 1.8) represent mean for each state dataset, respectively. The range of measurements on the y axis are 1 = Natural log of ∼ 3 mm; 2 = Natural log of ∼7.5 mm; and 3 = Natural log of ∼20 mm. (Color online)
Figure 5 Long description
Box plots display the log of colonoware sherd thickness in millimeters for sites in South Carolina and Virginia. The x-axis is labeled 'Project (Ordered by blueMCD)' and the y-axis is labeled 'Log of Colonoware Sherd Thickness (mm)'. South Carolina sites show a consistent spread around the mean, marked by a red horizontal line at 1.8. Virginia sites exhibit more variability, with some sites having thicker or thinner sherds. Each box plot represents a different site, with individual data points showing the range of thickness values for each assemblage.
Vessel thickness varies by completeness; that is, which part of the vessel is measured. Bases are typically thicker than bodies in many hand-built forms, and rims may be thick or thin depending on form. To better account for this potential variation based on vessel completeness, we considered body thickness and rim thickness for bowl and jar forms separately. On average, colonoware bowl rims were thinner than body sherds in both regions. In South Carolina, bowl rims had a mean thickness of 5.54 mm and a body mean of 7.15 mm. In Virginia, bowl rim mean thickness was 5.61 mm and body mean 7.05 mm. However, Virginia assemblages exhibit far more site-based variation in rim thickness compared to South Carolina, which maintains the overall thickness pattern (Supplementary Figure 1). This variation in Virginia assemblages mirrors the pattern found in overall thickness (Figure 5) and serves as an additional indicator that specific functional requirements (tableware vs. utilitarian) may be driving use at each site.
Vessel thickness also varies with inclusion types. Shell-tempered sherds from Virginia are significantly thinner than other inclusion types (mean = 5.29 mm; p < 0.01), suggesting that the addition of shell may have been reserved for certain types of vessels that functioned as tablewares—those used primarily for serving and eating (Figure 6a). It is worth noting that shell-tempered pottery is also efficient for cooking, because the thermal expansion of shell matches the expansion of clay, which reduces vessel stress (Rye Reference Rye1981:33). Quartz, in contrast, has a higher coefficient of thermal expansion, which can cause stress unless the particle size is fine (Bronitsky and Hamer Reference Bronitsky and Hamer1986:98) or the body is porous enough to allow expansion (Rice Reference Rice2015:329–330).
(a) Comparison of median sherd thickness by inclusion type and state. Horizontal line represents 1.80, the dataset-wide mean, which equates to ∼ 6.36 mm; (b) median sherd thickness by inclusion density and state. Inclusion density measurements are as follows: Not Applicable = < 5%; Low Density = > 5% < 7.5%; Medium Density = > 7.5% < 15%; and High Density >15%.
Figure 6 Long description
The image contains two box plots. The first plot, labeled 'a', shows the log of colonnware thickness by inclusion type for South Carolina and Virginia. The x-axis is labeled 'Inclusion Type' and the y-axis is labeled 'Log of Colonoware Thickness (mm)'. Various inclusion types are plotted with corresponding box plots. The second plot, labeled 'b', displays the log of colonoware thickness by inclusion density for the same regions. The x-axis is labeled 'Inclusion Density' and the y-axis is the same as the first plot. Different inclusion densities are represented with box plots, showing variations in thickness across the two states.
When examining the relationship between inclusion density and vessel thickness, we also observed significant differences in the average thickness of density categories in both regions (Figure 6b). Colonoware with a low density of inclusions tends to be thinner (mean = 5.64 mm for SC; 5.37 mm for VA), whereas those sherds with medium (mean = 6.02 mm for SC; 6.72 mm for VA) and high densities (mean = 6.61 mm for SC; 7.1 mm for VA) are thicker. Prior research has shown that higher densities of aplastics correlate with lower rates of breakage due to thermal shock (Braun Reference Braun, Moore and Keene1983; Tite et al. Reference Tite, Kilikoglou and Vekinis2001). We suggest that these patterns could be related to the functionality of different vessels, with high-inclusion-density vessels serving more utilitarian purposes that are correlated with higher rates of thermal and mechanical stress (processing, cooking), whereas thinner, lower-density varieties functioned as eating or serving vessels, which are not likely subject to the same stressors. We acknowledge that thinner, shell-tempered sherds would have high rates of thermal conductivity and therefore would have been ideal for cooking. However, we hypothesize that potters working in technical traditions featuring shell temper may have been prioritizing tableware forms in Virginia and that potters in both regions may have been producing thicker walls less prone to breakage for utilitarian vessels.
Form
The thickness of a vessel is also related to its planned form. Morphology—the shape of a ceramic vessel—has played a key role in the classification of colonoware for decades. Here we focus on another vessel attribute that can tell us about function: accessibility (Rice Reference Rice2015:419–421). Accessibility has to do with how easily the contents of a vessel can be retrieved. Ethnographic research on southeastern historic Native American groups and Kalinga communities of the Philippines suggests that the frequency with which a vessel’s contents need to be accessed and the degree to which its contents need to be contained are two important variables that affect a vessel’s form (Braun Reference Braun1980:172; Henrickson and McDonald Reference Henrickson and McDonald1983; Skibo Reference Skibo2013). Vessels with open or unrestricted orifices such as simple or everted-rimmed bowls are selected when frequent access is the overarching concern (Boudreaux Reference Boudreaux2010; Hally Reference Hally1983; Rice Reference Rice2015:419–420; Skibo Reference Skibo1992, Reference Skibo2013; Wilson and Rodning Reference Wilson and Rodning2002:32), whereas vessels with more restricted openings (e.g., narrow rim diameters, presence of a neck) are used when containment of contents is important. The reduction of the evaporation of cooking liquids is one example. Restricted forms include globular jars with everted rims, bottles, and bowls with inverted rims (Skibo Reference Skibo2013).
Whether a vessel is hollow or flat can also provide clues to its function. Flat forms like plates and platters prioritize access over containment. Only four flat vessels—all plates—were identified in the dataset.Footnote 3 Hollow forms predominate the colonoware assemblages used in this analysis. Bowls and jars are the most common hollow forms (Figure 7), together comprising 61% and 26%, respectively, of all sherds with a specific form identification.Footnote 4 Bowls are defined as vessels with a maximum body height that is no greater than their maximum diameter. Jars are defined as vessels that are taller than their maximum diameter (Beaudry et al. Reference Beaudry, Long, Miller, Neiman and Stone1983; Rice Reference Rice2015:236). Because of the functional variation, we focused on distinguishing bowl and jar vessels within the dataset and on characteristics such as orifice diameter that relate to vessel classification as open or restricted. As expected, given distinct functional requirements, bowls and jars tend to have distinct rim diameters (Figure 8). On average, the orifices of open bowls have larger diameters (>140 mm) than jar orifices, which are more restricted (<140 mm; Figure 8). We assigned hollow sherds originally entered as “unidentifiable” in form with measurable rim diameters of >140 mm to the bowl category and <140 mm to a jar category for use in these analyses.
Comparison of the most common identifiable forms within the dataset. Rare forms with sample size, n < 30 (e.g., cup, chamber pot, jug, milk pan, and plate) are grouped into an “Other” category. Unidentifiable forms, and instances where form was not recorded in the legacy data are not included here. Sites within each region are ordered by blueMCD along the x-axis (Bates et al. Reference Bates, Galle, Neiman, Hauser and Wallman2020).
Figure 7 Long description
A bar chart comparing proportions of identifiable colonoware vessel forms across projects in South Carolina and Virginia. The x-axis is labeled 'Project (Ordered by blueMCD)' and the y-axis is labeled 'Proportion'. Each bar represents a project, with sections indicating different vessel forms: bowl, jar, other, unid. tableware and unid. utilitarian. The chart shows varying proportions of these forms across different projects in both states, with some projects having a higher proportion of bowls or jars, while others have more unid. tableware or utilitarian forms. The legend on the right identifies the color coding for each vessel form category.
Log rim diameter of sherds identified as bowls versus jars.
Figure 8 Long description
A box plot comparing the log rim diameters in millimeters of bowls and jars. The x-axis is labeled 'Form' with categories 'Bowl' and 'Jar'. The y-axis is labeled 'log(Rim Diameter) (mm)'. Each category shows a distribution of data points with a box plot indicating the median, quartiles and outliers. The plot is limited to sherds with 'Colono Completeness = Rim' and a rim diameter measurement. Related data is from link.
Colonoware bowls dominate the assemblages in both Virginia and South Carolina; however, there are regional differences in the abundance and size of the patterning of vessel form. In South Carolina, jars were recovered from nearly all sites, but only four jar sherds (<1%) were recovered from Virginia sites in this study (two from Pope, one from Wingos, and one from South Grove). Given that the basic form of bowl or jar can be modified for any number of functions, it was important to determine whether Virginia and South Carolina consumers were using colonoware bowls and jars in the same way. In her study of South Carolina colonoware, Corey Sattes (Reference Sattes, Marcoux and Sattes2024:Figure 8.2) found that most assemblages had a bimodal distribution of bowl orifice diameter, indicating at least two common size classes for bowls. Although we did not see the bimodal pattern in our data for either region (Figure 8; Supplementary Figure 2), Sattes’s (Reference Sattes, Marcoux and Sattes2024) findings suggest that certain size classes existed that could have performed different roles. The lack of bimodal patterning in our data could be due to differences in the granularity of the comparison. Sattes (Reference Sattes, Marcoux and Sattes2024) separated bowls by site and temporal phase, but the larger scope of our analysis and the smaller sample sizes of sherds in Virginia assemblages limited our ability to break down the data to such a level. We included the same data analyzed by Sattes (Reference Sattes, Marcoux and Sattes2024) in our study, adding to a dataset that spans a broader temporal, geographic, and functional range; therefore, the lack of bimodal distribution of orifice diameter when aggregated at the regional level suggests that this characteristic may not be a significant marker of all South Carolina colonoware. Further analysis of variation in orifice diameter in a tighter chronological framework is needed to understand how this pattern may be related to temporal or spatial dynamics.
Surface Treatment
Once a vessel has been shaped, modifications to the surface also impart certain functional qualities to the resulting form. Surface treatments can have both functional and aesthetic purposes depending on a vessel’s intended uses (Schiffer et al. Reference Schiffer, Skibo, Boelke, Neupert and Aronson1994; Skibo Reference Skibo2013:47–50). We define surface treatment as an alteration applied to the surface of the vessel as part of the production process. Alterations may include paddle or textile impressions reflecting the technical choices of forming, where paddles were used to compress and form the clay, or incisions, punctations, or notchings that typically served a decorative purpose. Vessels for serving tend to be more highly decorated than utilitarian wares for food preparation and storage.
Given the low numbers of decorated sherds in the dataset, here we focus on burnishing, the most common decorative surface treatment.Footnote 5 Burnishing consists of buffing a vessel with a stone or other smooth tool when the vessel is unfired but dry. The process compacts and aligns the clay particles, resulting in a dense and glossy surface. Burnishing a vessel’s surfaces will decrease its permeability (Rice Reference Rice2015:317), at the same time increasing heating effectiveness by influencing heat transference through the vessel wall (Schiffer Reference Schiffer1990; Schiffer et al. Reference Schiffer, Skibo, Boelke, Neupert and Aronson1994). Moreover, burnishing serves an aesthetic purpose and may be added primarily for decorative value, sometimes appearing to imitate lead-glazed industrially made vessels. Burnishing lacks strong antecedent Indigenous traditions in the region encompassing Virginia; thus, it may primarily be seen as a postcontact phenomenon in that region (Egloff and Clem Reference Egloff and Clem2016). In South Carolina, burnishing is a minor constituent in precontact St. Catherines, Savannah, and Lamar series pottery assemblages (Anderson Reference Anderson1996; DePratter Reference DePratter1991).
In general, burnishing and decorating pots indicate greater time investment in production (Rice Reference Rice2015:440) and the incorporation of aesthetic value. Given that burnishing potentially serves both utilitarian and aesthetic purposes, we analyzed its presence/absence alongside vessel thickness to better understand its relationship to intended function. Thinner burnished sherds typically suggest a focus on aesthetics, whereas thicker burnished sherds indicate a more utilitarian purpose. Burnished sherds are found in both South Carolina and Virginia assemblages, but they make up a much larger proportion of samples from Virginia (n = 494, 23% in VA vs. n =1913, 4% in SC). In South Carolina, burnished sherds tend to be slightly thinner than unburnished sherds (mean = 6.17 mm and 6.42 mm, respectively), but the difference is much greater in Virginia (mean burnished = 5.36 mm, mean unburnished = 6.25 mm; Figure 9). These patterns suggest that in South Carolina colonoware was more likely to serve a utilitarian function or one less reliant on the emulation of glazed European wares.
Boxplot showing variation in thickness for burnished and unburnished sherds in Virginia and South Carolina.
Figure 9 Long description
A boxplot comparing the log of colonoware thickness in millimeters for burnished and unburnished sherds in South Carolina and Virginia. The left section shows data for South Carolina, with two boxplots for burnished and unburnished sherds. The right section shows data for Virginia, also with two boxplots for burnished and unburnished sherds. Each boxplot displays the distribution of thickness, with median lines, quartiles and outliers marked by individual points. The x-axis is labeled 'Burnishing' and the y-axis is labeled 'Log of Colonoware Thickness (mm)'.
Residue
Finally, we consider the most direct material evidence of a vessel’s functionality. The presence and location of carbon-rich substances on a vessel’s surface, such as burned food residue or soot, not only indicate that a vessel was used with fire but also provide direct evidence that it was used for cooking (Hally Reference Hally1983:7–10; Rice Reference Rice2015:428–429). DAACS cataloging protocols distinguish this type of carbon deposition from fire-clouding, smudging, and postdepositional burning based on the presence of discrete areas of flaky, crusty, or shiny drips or deposits, typically on vessel exteriors. We acknowledge that our recording protocols for residue do not account for subsurface absorptions that lack surface carbon and, in that sense, is a narrower use of the term. The location of residue on a vessel also provides information about how that vessel was oriented to the heat source. For example, residue or sooting on the shoulder or rim of a vessel suggests it was propped against rocks and wood directly in a fire, whereas base sooting indicates suspension over a fire (Briggs Reference Briggs2016; Hally Reference Hally1983).
Sooting was not present on the majority of colonoware sherds. More South Carolina colonoware sherds exhibited sooting overall than Virginia (n =1,708 for SC; n = 48 for VA), but it is notable that the percentage of colonoware with sooting/residue is similar (3% for SC; 2% for VA). A look at individual assemblages, however, shows an uneven distribution of sooted sherds across South Carolina assemblages, with 11% of sherds from Old Dorchester representing the highest proportion of sooted sherds on a single site (Figure 10). The higher proportions of residue/soot in some South Carolina assemblages indicate a greater use of colonoware for cooking. This pattern is reinforced by the lack of jar forms in Virginia, the most common earthenware cooking form. Notably however, both jar and bowl forms had residue/sooting (see also Ferguson Reference Ferguson1992:Appendix 3, Tables 2 and 3). The presence of sooting was highly variable by site, particularly for bowls, which may indicate specialized use cases on specific plantations. The overall similarity across regions in percentages may reflect preservation issues of sooting in general and reinforces the need for additional lines of evidence when examining function.
Proportion of colonoware sherds with residue with error bars, ordered within state by blueMCD.
Figure 10 Long description
The x-axis is labeled 'Project (Ordered by blueMCD)' and the y-axis is labeled 'Proportion of Residue'. The plot is divided into two sections: South Carolina on the left and Virginia on the right. Each section contains multiple points representing different projects. The points show varying proportions of residue, with error bars indicating variability. The graph highlights differences in residue proportions across these projects in both states.
A comparison of sooting presence paired with inclusion type and density shows some interesting trends. Notably, only two shell-tempered sherds in the dataset—one from Middleburg Village in South Carolina and one from Richneck Quarter in Virginia—exhibited evidence of sooting. With the caveat that most sherds in the dataset lack residue, the limited examples of shell-tempered vessels with residue suggest that the primary function of the shell temper was not related to efficiency for cooking, despite the studies noted earlier that showed the strong relationship between shell temper and high thermal conductivity. Inclusion density, thickness, and residue are also not correlated, because the thickness of sherds with residue remains relatively consistent in South Carolina (thicker on average), no matter the inclusion density, and sherds with residue and low density in Virginia are very thin (Supplementary Figure 3). This pattern continues to show more co-variation between the two regions than within them. It also aligns with our expectations that colonoware was used more for utilitarian functions in South Carolina, where we see thicker vessels with higher densities of inclusions that would increase the vessel’s thermal shock resistance (Tite et al. Reference Tite, Kilikoglou and Vekinis2001), as well as more instances of residue or soot.
Discussion
One of the most basic functional distinctions in pottery vessels is between utilitarian vessels intended for food storage, bulk processing, or cooking, and tableware vessels meant for consuming and serving food. Whereas utilitarian vessels for storage and processing tend to be thicker, sturdier, and lack decoration, tablewares are often thinner, more finely potted, have less temper or inclusions, and are more likely to be decorated (Rice Reference Rice2015:415, 424). Previous investigations exploring functional variation suggested that enslaved people produced and used colonoware as a way to replicate and maintain African-based cooking styles, diet preferences, and folk traditions (Deetz Reference Deetz and Singleton1999:89; Ferguson Reference Ferguson and Singleton1999; Trinkley et al. Reference Trinkley and Hacker2003), as well as ritual or medicinal practices from their lives before enslavement (Ferguson Reference Ferguson1992, Reference Ferguson and Singleton1999, Reference Ferguson2007; Ferguson and Goldberg Reference Ferguson and Goldberg2019; Joseph Reference Joseph2007; Ogata Reference Ogata1995; Singleton and Bograd Reference Singleton, Bograd, Delle, Mrozowski and Paynter2000). Their cooking styles emphasized low and slow cooking of stew-based meals and cooking multiple smaller dishes rather than one-pot meals. However, the ability to evaluate these hypotheses has been limited by a lack of engagement with detailed attribute data of colonoware on a broad geographic scale.
Our study, the first interregional, attribute-driven analysis of colonoware in the southeastern United States, explored whether there is evidence that functionality was shaped by distinct regional contexts. The results of our analysis quantitatively reinforce some well-understood elements of colonoware—that it has a much greater abundance in South Carolina than in Virginia—while adding important dimensions to our understanding of the possible factors that influenced this pattern. The results of our initial analysis using abundance indices demonstrate that although the differences in colonoware abundance between the two regions persists through time, the discard of imported ceramics on sites of enslavement on the same sites was uncorrelated. This is the first significant indicator that colonoware occupied a functionally different role in South Carolina than in Virginia.
Regional differences in other functional attributes—paste inclusion densities, sherd thickness, vessel form, rim diameter, surface treatments, and sooting—all point to the relatively heavy reliance on colonoware for cooking and food preparation and storage in South Carolina. In Virginia, ceramic utilitarian vessels serving these needs were less common overall and occurred in larger quantities at a subset of sites (South Grove and Wingos). Notably the distribution of jars was isolated to South Carolina (apart from four sherds in Virginia). Narrow-necked jars could have served a variety of functions including storage, serving or cooking, and their restricted necks would have been a useful morphological trait for slow cooking and easily pouring the contents. Although bowls were the most commonly identified form in both regions, those from a larger proportion of sites from South Carolina had direct evidence of use for cooking. This is consistent with Ferguson’s (Reference Ferguson1992) suggestion that bowls in South Carolina were used for cooking side dishes or sauces and medicines. The majority of bowls with residue/sooting from Virginia sites were associated with just two assemblages, South Grove and Pope.
Previous researchers have distinguished types of colonoware in South Carolina according to whether they were “market” wares or for local/household use (e.g., Cranford Reference Cranford, Marcoux and Sattes2024; Joseph Reference Joseph2004, Reference Joseph, Marcoux and Sattes2024; Sattes Reference Sattes, Marcoux and Sattes2024; Zierden et al. Reference Zierden, Anthony, Platt, Marcoux and Sattes2024). Market wares are presumed to be thinner and more likely to be decorated or in European forms. In contrast, nonmarket wares are defined as thicker and more utilitarian. From a functional perspective, it may be that market wares were made to appeal to the broad market that included Euro-American consumers, whereas local products in South Carolina were made to fulfill culturally specific culinary and medicinal functions related to African lifeways. The assemblages in our sample appear to be mixtures of both market and nonmarket wares, with the latter dominating.
Overall, sherds in Virginia are thinner on average than sherds in South Carolina, and burnishing is much more common. For bowls in particular, burnished bowl sherds from Virginia sites were significantly thinner than unburnished ones. Together these attributes suggest that within Virginia there was greater standardization of production methods that emphasized thin bodies and lustrous surfaces. In terms of inclusions and tempering practices, some Virginia colonoware assemblages are distinguished by the presence of shell tempering, which is also correlated with thinner vessels.
Although shell-tempered vessels are quite functional for cooking, typically Virginia vessels lack sooting. The general absence of sooting suggests that shell tempering is a distinct phenomenon within Virginia colonoware that may be more indicative of a particular production community or tradition than imparting functionality. A similar relationship between sherds that lack visible inclusions and exhibit thinner vessel walls may also be related to production traditions versus intended use. This could explain why we lack evidence to support that cooking vessels are thinner, despite the potential for better thermal conduction. Further studies are needed to clarify what role tempering (or lack thereof) played in Virginia colonoware production and use.
Rim diameters of bowls from Virginia assemblages also tend to be larger, pointing to the prioritization of easy access to contents for serving/consumption as opposed to containment for cooking purposes. Although we did not compare paste color across regions in this study, we also note that Virginia colonowares tend to be pale-bodied, without smudging (e.g., Binford Reference Binford1965; Noël Hume Reference Noël Hume1962), which also mimics a European aesthetic more closely. These findings align with prior research arguing that colonoware recovered from Virginia sites, especially those located in Chesapeake, tended to imitate European tableware forms, rather than cooking forms (Ferguson Reference Ferguson1992:52; Henry Reference Henry1980; Noël Hume Reference Noël Hume1962), prioritizing aesthetics and accessibility over sturdiness. Overall, the patterning of colonoware from Virginia sites aligns more with what we would expect of tableware: vessels with wider/larger orifices, more thinly potted with higher rates of lustrous surfaces, lower densities of temper, and less evidence of exterior residue.
Conclusions
Our focus in this article has been the systematic description of variation in the abundance, material, and form of colonoware in assemblages in Virginia and South Carolina and its relationship to functionality. Given that each region has some internal consistency in patterning, it is worth briefly considering what economic and social factors may have served as the causal mechanisms of this patterning within its most common context—the enslaved household. Enslaved people in Virginia and South Carolina were typically allotted rations such as cornmeal and salted meat but also relied on foraging, gardening, fishing, and other methods to obtain necessary sustenance. Although historic documentation of enslaved individuals’ cooking activities is limited, archaeological and documentary evidence suggest meals that could be cooked and eaten quickly, such as breads, and slow-cooking and low-supervision dishes like stews were preferred (Yentsch Reference Yentsch2008:14–15). Recipes from the time describe the slow cooking of stews in earthenware vessels (e.g., Randolph Reference Randolph1838), which was said to result in tastier meals than cooking in iron pots (Ferguson Reference Ferguson1992:90).
The amount of time available for food procurement and preparation was generally limited in both regions. Nevertheless, key factors structuring colonoware’s role in food preparation and consumption practices were likely the different labor management systems and settlement patterns that dominated in each region (Morgan Reference Morgan1998). In South Carolina the enslaved communities in our study were housed in quarters on small farms and or in large villages on plantations that operated task-based labor systems for rice and indigo cultivation. This system, although a grueling and violent method of labor extraction, would potentially enable enslaved individuals to use time outside mandated cultivation tasks to procure and slow cook combinations of rice, vegetables and meat. Additionally, the more common pattern of larger enslaved households on South Carolina plantations living in closer proximity to each other may have enabled family members to share cooking duties.
In contrast, enslaved individuals in Virginia laboring on middling and larger plantations tended to experience a higher degree of residential isolation on widely dispersed quarter farms and were subject to gang-based labor systems for tobacco cultivation for much of the seventeenth and eighteenth centuries (Walsh Reference Walsh1993:173). Additionally, shifts to other crops, such as wheat cultivation, in the late eighteenth and nineteenth centuries still required rigorous management, with little free time allocated to non-subsistence related tasks. The greater oversight from overseers, longer workday common to this labor system, and smaller family groupings could mean that enslaved individuals’ time budgets in the Chesapeake were more constrained. As a result, they may have been forced to rely more on food and meals that could be made in provisioned iron pots or on griddles and to rely less on colonoware-based cooking methods.
These varying conditions could have encouraged enslaved people to make different choices in pursuit of their own and their families’ well-being. Intraregional comparative analysis may provide a better way to identify the functional role of colonoware within regions. For example, variation within each region could be further investigated by site type category (e.g., rural enslaved households, trading posts, taverns, mixed midden deposits from manor houses etc.). Moving forward, approaching Virginia colonoware and South Carolina colonoware as separate but related phenomena will make it possible to account for specific causal mechanisms while retaining the broader framework.
Another important causal mechanism that needs to be better understood is the influence of different production traditions and their material correlates in colonoware. We are using the results of this study to inform ongoing research into the production, marketing, and use of colonoware across the upper Southeast, targeting samples identified via attributes for elemental analysis and other specialized analyses (e.g., Crane Reference Crane1993; Marcoux et al. Reference Marcoux, Sattes and Sherard2023). This team of authors, along with additional collaborators, are currently engaged in a large-scale project that is applying a communities-of-practice approach to determine what co-occurring attributes of colonoware—for example, elemental composition, paste types, vessel morphology, and decoration—are material correlates of learning traditions and shared contexts of consumption and use. We are examining these processes using a dataset of approximately 182,263 colonoware sherds and a compositional analysis of 608 samples from 40 sites to characterize colonoware traditions across the upper Southeast (Virginia, North Carolina, and South Carolina) from the seventeenth to nineteenth centuries. Critically, North Carolina colonowares are included in the sample, which will help us further tease apart the relationships between different colonial projects and labor regimes and their impacts on colonoware production and use. Finally, incorporating samples from sites located within and adjacent to urban centers, such as Williamsburg and Charleston, will enable us to examine market access and exchange and its impact on colonoware production. Situating our synthesis of elemental and attribute data within geographic and temporal frameworks will allow us to identify potting communities and tie them to the social conditions of production, marketing, and use under which colonoware variants were created, retained, or disappeared. Furthermore, tracing variations between the location of production and the location of deposition offers a way to reconstruct past networks of potters and consumers within informal regional economies.
Colonoware played a complex economic role in colonial systems. Despite its complicated origins, this pottery was central to the economic activities, food preparation, and cultural traditions of enslaved and subjugated peoples who endured the devastating colonial expansion across the southeastern United States from the seventeenth century onward. As a result, colonoware sherds provide some of the best evidence for identifying the active roles these communities played in negotiating the economic and social systems of early America. By reexamining important legacy collections, we conducted an analysis of the largest and most detailed standardized dataset not only on colonoware but also on low-fired coarse earthenwares in the field of historical archaeology.
The results of our reanalysis improve our understanding of colonoware’s role in social and economic dynamics of early America and serve as a model for archaeologists working on ceramic production in different regions and periods. We have demonstrated how synthesis can increase the analytical potential of individual collections and permits comparative analyses at a much larger scale than previously possible. We hope that this research demonstrates how systematic analytical frameworks reveal overarching regional trends and fine-grained variation among sites and assemblages and the ways in which these patterns offer valuable clues to the underlying historical processes responsible for them. We encourage others to build on this foundation by evaluating our results, developing new questions, and finding new answers in the data we have provided.
Acknowledgments
We thank our collaborators who have provided suggestions and feedback on this study, the excavators who carefully documented these sites and assemblages, the analysts who created the data, and the stewards of all the collections utilized herein. We thank Cate Garcia and Nicolas Delsol for their assistance with the Spanish abstract. We also thank four anonymous reviewers for their thoughtful feedback, which has improved this work.
Funding Statement
This research received no specific grant funding from any funding agency, commercial, or not-for-profit sectors.
Data Availability Statement
The data and contextual information from DAACS sites reported in this article may be downloaded from DAACS.org. Code and full dataset used for the analysis are available through the Open Science Framework repository (https://osf.io/g2jux/; see also Supplementary Code 1).
Competing Interests
The authors declare none.
CRediT Statement
Elizabeth A. Bollwerk: conceptualization, methodology, formal analysis, writing — original draft & editing, visualization. Karen Y. Smith: conceptualization, methodology, formal analysis, writing — original draft & editing, visualization. Lindsay C. Bloch: conceptualization, formal analysis, writing — review & editing. Jillian E. Galle: conceptualization, methodology, writing — review & editing, resources, project administration. Leslie S. Cooper: conceptualization, investigation, writing — original draft. Corey A. H. Sattes: methodology, writing — review & editing. Fraser D. Neiman: conceptualization, methodology, software, resources, writing — review & editing.
Supplementary Material
The supplementary material for this article can be found at https://doi.org/10.1017/aaq.2026.10198.
Supplementary Code 1. Dataset and R Code for All Reported Analysis.
Supplementary Figure 1. Bowl and jar rim sherd thickness by site and state.
Supplementary Figure 2. Density histograms illustrating rim diameter sizes for bowl and jar forms in Virginia and South Carolina.
Supplementary Figure 3. Log thickness of sherds with residue, by inclusion density category.
Supplementary Text 1. References Reporting Colonoware in Table 1.
Supplementary Table 1. Complete version of Table 1 including References column.
Open access
