In the 1970s, John Otto (e.g., Reference Otto and South1977, Reference Otto1984) compared ceramic vessels from the nineteenth-century homes of enslaved Africans, White overseers, and enslavers at Cannon’s Point plantation in Georgia (USA). The enslaver’s home had the highest percentage of plates (84%), followed by the overseer’s site (72%) and the enslaved households (49%). Conversely, the homes of enslaved people had the highest percentage of bowls (44%), followed by the overseer’s site (24%) and the enslaver’s household (8%). These results show a strong correlation between race, class, and ceramics: enslaved people often ate from bowls, enslavers predominantly ate off plates, and lower/middle-class White overseers fell somewhere in between.
Otto’s (Reference Otto and South1977, Reference Otto1984) research is an excellent example of how archaeologists can use ceramics to identify race, ethnicity, and/or status, and it is widely cited in case studies on other parts of the South (e.g., Andrews and Fenton Reference Andrews and Fenton2001; Wilkie Reference Wilkie2000), the Caribbean (e.g., Farnsworth Reference Farnsworth1996; Lenik Reference Lenik2009), South America (e.g., Symanski Reference Symanski2012; Symanski and Torres de Souza Reference Symanski and de Souza2007), West Africa (Gijanto Reference Gijanto and Marshall2015), and Canada (Beaudoin Reference Beaudoin2017; Beaudoin et al. Reference Beaudoin, Josephs and Rankin2010). Attempts to replicate his results on other US sites, however, show that Otto’s (Reference Otto and South1977, Reference Otto1984) findings have a smaller geographic range. Plantations in Florida, Georgia, and South Carolina generally conform to Otto’s (Reference Otto and South1977, Reference Otto1984) observations, although “too many sites show exceptions to this trend for it to be any more than suggestive” (e.g., Adams and Boling Reference Adams and Boling1989:80; Ibarrola Reference Ibarrola2022; Lewis Reference Lewis and Singleton1985; Moore Reference Moore and Singleton1985). But comparisons between enslaved and Free Black households and poor/middle-class White families in North Carolina and Virginia have not found meaningful differences in ceramic forms (Greene Reference Greene2011; Shephard Reference Shephard and Spencer-Wood1987). Research at Virginia plantations has also consistently failed to replicate Otto’s (Reference Otto and South1977, Reference Otto1984) findings (see Heath and Breen Reference Heath and Breen2009:2).
What does this difference in ceramic assemblages in Virginia and Florida, Georgia, and South Carolina mean? Does it point to similarities in the foods eaten by enslaved people and White households in Virginia? Or does it merely reflect how archaeologists categorize data, speaking more to modern assumptions about vessel use than past foodways? The current scholarship on ceramics in eighteenth- and nineteenth-century America cannot address these questions because most studies rely on the assumption that vessel form provides clear insights into vessel use (Beaudoin Reference Beaudoin2017; Greer Reference Greer2024). But vessels can be used in a variety of ways. People can drink from bowls (Ferguson Reference Ferguson1992:106) and eat from teacups (e.g., Samford Reference Samford2007:99; Wilkie Reference Wilkie2000:147). Plates can be used to serve individual portions of rice or meat or a communal pile of bread. Assessing how enslaved people used bowls and other vessels requires use-wear and/or absorbed residue analyses, which provide direct evidence about the role specific vessels played in past foodways. Until these techniques are applied to ceramics from Virginian plantations, it is impossible to determine what the relative lack of bowls from the state’s quartering sites might mean.
This article uses use-wear and absorbed residue analyses to explore why the ratio of bowls to plates seen in the lower South does not appear in Virginia and what this difference might say about enslaved foodways. My data come from three late eighteenth- to mid-nineteenth-century sites at Belle Grove plantation: an enslaved quartering site, the plantation’s manor house, and a dwelling inhabited by White employees. Although similar ratios of bowls, plates, mugs, teacups, and saucers are present at all three sites, use wear and absorbed residues indicate that enslaved people used mugs, teacups, and saucers in ways than Belle Grove’s White inhabitants did not, with enslaved people likely serving (semi-)liquid food from them. As a result, this research shows that enslaved Virginians often made choices about how to serve meals that were different from those made by people enslaved in South Carolina / Georgia, likely due to differences in rationing practices. More broadly, this study demonstrates how combining use-wear and absorbed residue analyses can expand American archaeologists’ ability to assess the role specific types of vessels played in past foodways.
Belle Grove Plantation
Belle Grove is a wheat and cattle plantation in southern Frederick County, Virginia (Figure 1). Isaac Hite Sr. acquired the land that would become Belle Grove in the mid-eighteenth century (Greer Reference Greer2022:103–104). However, no records indicate that the property was inhabited until 1783, when Hite gifted the then 195 ha (483-acre) estate to his son Isaac Hite Jr., following the younger Hite’s marriage to Nelly Madison. The newlyweds moved to the property that year, forcing 15 enslaved people to relocate with them. Belle Grove’s enslaved community grew quickly, reaching 101 people by 1810. At the same time, the Hites dramatically increased their landholdings, owning over 3,035 ha (7,500 acres) by the 1810s. This land was divided among four properties, with Belle Grove serving as the Hites’ primary residence (Greer Reference Greer2022:105–106). Although it is unknown how many people were enslaved at Belle Grove, I estimate that 30–50 people were held at the plantation.
Location of Belle Grove. Map by Erica Moses. (Color online)
In 1824, the Hites sold 60 people into the interstate slave trade. This sale, coupled with the Hites giving enslaved people to their adult children, reduced the number of people they enslaved to 44 by 1836 when Isaac Hite Jr. passed away (Greer Reference Greer2022:108). His second wife, Ann Hite, inherited Belle Grove and many of the people enslaved by the couple. Ann died in 1851, and her children managed Belle Grove until 1860, when they sold it to John and Benjamin Cooley, who ran the farm using both enslaved and free labor. The Cooleys sold the estate in 1865, and it regularly changed hands until the National Trust for Historic Preservation acquired Belle Grove in 1964.
Archaeological research has occurred at Belle Grove since 1972, with extensive excavations at the manor house, the main enslaved quartering site, and a dwelling inhabited by White families employed by the Hites. Investigations at the manor house (44FK016) occurred at various times between 1972 and 2016, and they primarily focused on identifying and sampling outbuildings and workspaces (e.g., Geier Reference Geier1995; Rockwell Reference Rockwell1974). The most extensive excavations occurred at Old Hall, where the Hites lived from 1783 to 1797. Old Hall was later inhabited by other members of the Hite family, enslaved domestic laborers, and a free Black family before being demolished around 1890 (Moses Reference Moses2022). Ceramic motifs from Old Hall are generally consistent with other areas of the manor-house grounds and distinct from the enslaved quartering site (see below), indicating that they speak more to the life of the Hites than the Black households that later lived there (Greer Reference Greer2022:197). A comprehensive catalog of artifacts from the various manor-house excavations has not been compiled, but I reanalyzed 3,004 ceramic sherds from this site in 2020–2022 (Greer Reference Greer2022), providing a representative sample of the Hites’ tea and tablewares. Sixty-two percent of these sherds were from Old Hall.
Belle Grove’s main enslaved quartering site (44FK520) is a 0.6 ha (1.5-acre) complex, approximately 150 m west of the manor-house grounds, that was continually inhabited from around 1800 to the 1850s. The quarters were identified in the 1990s but remained unexcavated until 2015–2019, when I directed the excavation of 468 shovel test pits (STPs), 114 0.9 × 0.9 m (3 × 3 ft.) units, and three smaller units (0.45 × 0.45–0.90 m [1.5 × 1.5–3.0 ft.]; Greer Reference Greer2022). These investigations identified five possible house sites and four large sheet middens. One house and three middens were sampled extensively. A total of 59,177 artifacts were recovered during the excavations, including 10,663 ceramic sherds, 44.25% of which came from the site’s middens.
The Hites regularly employed White men, many of whom served as overseers (Greer Reference Greer2022:106–107). Given that the Hites owned and managed multiple properties, it is difficult to identify where these men lived and worked. But the first, Benjamin Little, was hired when Isaac and Nelly Hite only owned Belle Grove, so he likely resided at the property (Greer Reference Greer2022:130). Archaeological investigations have identified one dwelling at Belle Grove in which White employees and their families likely lived (44FK522). This site was initially identified in the 1990s (Geier and Zienty Reference Geier and Zienty2001), and I directed the excavation of 19 STPs and 10 0.9 × 0.9 m (3 × 3 ft.) units there from 2020 to 2023, recovering 6,893 late eighteenth- to mid-nineteenth-century artifacts, including 3,378 ceramic sherds. All but four sherds came from a single midden. Ongoing comparisons of this collection with artifacts from the other Belle Grove sites and five poor White/middling households in Frederick County show that the site’s artifacts are more consistent with White sites than the enslaved quartering site (including the types of use wear on their teawares and a growing preference for serving vessels), suggesting that White individuals lived there. Given that the Hites lived elsewhere, this site was likely inhabited by White employees.
Ceramic Form
The Belle Grove sites provide an excellent comparison to Otto’s (Reference Otto and South1977, Reference Otto1984) work at Cannon’s Point, because they include the homes of enslaved people, their enslavers, and White plantation employees. A minimum vessel analysis was conducted at each site, and these data were used to compare ceramic consumption and use across Belle Grove. This is a departure from Otto’s (Reference Otto and South1977, Reference Otto1984) methods, which compared sherd counts. However, a vessel-based analysis provides a better dataset for assessing vessel forms across multiple sites (Sussman Reference Sussman and Karklins2000) and for conducting use-wear and residue analyses. The vessel form of individual sherds was recorded for the quartering and White-employee sites, and these data generally conform to findings from the vesselized data (see below). Plates were defined as wide, shallow vessels with flat bases, often with transfer-printed or edged-molded decorations, and they are distinct from platters (which are wider and often thicker) and soup plates (which are deeper). Mugs are deep vessels with relatively vertical sides, narrow rim diameters, and handles, and these were undecorated or had annular/dipt motifs on their exterior surfaces. Bowls are wider, shallower vessels than mugs, with outwardly sloping sides, that are often undecorated or have transfer-printed, sponged, or annular/dipt decorations on their exterior surfaces. Saucers are also wide, shallow vessels with outwardly sloping sides, but with narrower rim/base diameters and shorter/thinner sides than bowls. They are often hand painted or transfer printed, with motifs that predominantly occur on their interior surfaces. I use “teacups” as a catchall term for teacups (which have handles) and teabowls (which lack handles), both of which have thinner bodies than bowls/mugs and narrower rim diameters and more vertical sides than saucers, with motifs that occur predominantly on their exterior surfaces.
The 3,004 analyzed sherds from the manor house came from a minimum of 507 vessels, including 333 vessels likely used to serve food and beverages (Table 1; Supplementary Material 1). Otto’s (Reference Otto and South1977, Reference Otto1984) analysis focused predominantly on imported porcelain and refined earthenware (thin-bodied, high-fired, lead-glazed earthenwares produced in industrial potteries from the mid-eighteenth century onward). However, a small number of eighteenth-century semirefined earthenwares were recovered from the Belle Grove sites and are included in this research. Frederick County and the surrounding area also had a substantial local ceramic industry that produced tea and tablewares into the early-nineteenth century (Comstock Reference Comstock1994), several of which were recovered from the manor-house grounds. Given that these were part of the Hites’ table settings, I also included them in this analysis. Most (89.48%) manor-house vessels could be subdivided into two chronological periods—the 1780s to the 1820s (e.g., pearlware, creamware, locally made coarse earthenware; n = 165) and the 1820s to the 1860s (e.g., whiteware, yellowware, and ironstone; n = 133). The remaining 35 vessels date to the Hite occupation but could not be assigned to a chronological group.
Vessels Associated with Serving Food/Beverages.
A total of 10,663 sherds were recovered from the enslaved quartering site. The vessel form of 823 individual sherds could be determined, including 37 bowls, 44 mugs, 532 plates, 70 saucers, and 105 teacups. The minimum vessel analysis identified at least 456 vessels, 289 of which were likely used to serve food and beverages. Most (93.00%) vessels either date to the 1780s–1820s (n = 206) or the 1820s–1860s (n = 60). Research on ceramic motifs at Belle Grove shows that enslaved people likely purchased many of the earlier tea and tablewares but acquired most of their ceramics secondhand from the manor house after the 1820s (Greer Reference Greer2022:193–197).
Finally, 3,378 ceramic sherds were recovered from the White-employee site. The vessel form of 193 individual sherds could be determined, including 11 bowls, 140 plates, seven saucers, and 24 teacups. A minimum of 202 vessels were identified at the site, 130 of which were likely used to serve food and beverages. Most (91.53%) of these vessels either date to the 1780s–1820s (n = 73) or the 1820s–1860s (n = 46).
The three sites had relatively similar ceramic assemblages; the major differences were that a larger percentage of serving wares were at the manor house, whereas a larger percentage of mugs were at the quarters (Figure 2a). Sherd counts also show similarities in ceramic form between the quartering and employee sites (Figure 2b). Diachronic comparisons (Figure 2c–d) show that some of these trends changed throughout time, with enslaved people acquiring more plates and fewer bowls after the 1820s. But overall, the stark differences seen in Florida, Georgia, and South Carolina are absent at Belle Grove.
Comparison of vessel form across the Belle Grove sites. (Color online)
To restate the question raised in the introduction, what do these similarities in ceramic form mean? Do they suggest relatively homogenous foodways and table settings across Belle Grove? Or do they point to something else? Vessel form alone cannot provide concrete answers, because a vessel’s intended function does not always correlate with how it was used. Solid foods are often served on flatwares, and (semi-)liquid substances are usually served from hollowwares, but as noted earlier, there is a tremendous amount of variation within these categories. Finding meaning in the vessel data requires turning to other analyses, specifically use wear and absorbed residue.
Use Wear
Ceramics accumulate wear and tear as people use them (Griffiths Reference Griffiths1978; Schiffer and Skibo Reference Schiffer and Skibo1989; Skibo Reference Skibo1992, Reference Skibo2013). Specific wear patterns can occur consistently on certain vessel forms, indicating that people habitually used them in particular ways. Linear abrasions running horizontally across teacup interiors, for instance, could indicate the repeated act of stirring cream and/or sugar into tea. To assess how enslaved people, White employees, and enslavers used ceramic tea and tablewares, I analyzed use wear on all bowls, mugs, plates, saucers, and teacups from Belle Grove. I focused on these forms because they account for 93.35% of vessels, thereby providing a robust dataset for making comparisons between the three sites.
Comparing use wear across multiple sites requires the systematic identification and documentation of wear marks (Banducci Reference Banducci2014:189). To do this, I used a modified version of the methods presented in earlier versions of this research (Greer Reference Greer2022, Reference Greer2024). In short, all 2,726 sherds associated with the identified vessels were assessed for wear. This was done without magnification, although a 40× handheld microscopic lens was used to further document abrasive wear (Banducci Reference Banducci2014:192–194). A total of 1,037 wear marks were identified on 253 vessels (Supplementary Material 2). For each mark, I recorded the side of the vessel it was on, its location on the vessel, the mechanism that produced it, the type of trace, its shape in plan view, the amount of wear on its top edge, its orientation relative to the vessel, its length and width, and (when possible) the distance between the top and bottom of all traces to the rim and/or base of the vessel. Descriptions and examples of the terms used in this analysis are provided in Table 2 and Figure 3.
Representative images of use wear. Fresh wear in Figure 3b is an artifact of earlier residue analysis conducted on this sherd (see Greer Reference Greer2024). (Color online)
Description of Terms Used in Use-Wear Analysis.
Although postdepositional processes can leave marks on sherds that mimic use wear, these are unlikely to affect my research (Greer Reference Greer2024:646–648). Plowing and trampling are two common postdepositional processes that damage sherds, but none of the Belle Grove sites were plowed, and trampling is more likely to snap artifacts into multiple pieces than to create the surface-level chips and abrasions that are the focus of use-wear studies (Schiffer Reference Schiffer1987:189). Freeze-thaw cycles can result in spalling that appears similar to fatigue wear, especially in low-fired ceramics. However, vessels fired at over 950°C are less likely to be damaged by freeze-thaw cycles (Skibo et al. Reference Skibo, Schiffer and Reid1989), and the ceramics used in this study were predominantly refined earthenwares fired at ≥980°C (Maggetti Reference Maggetti2018:6), reducing the possibility that freeze-thaw cycles impacted my results. Furthermore, despite the same environmental conditions affecting all three sites, distinctive wear patterns consistently occur at the quartering site, which should not be possible if the identified marks are the result of freeze-thaw cycles.
Plates
Four hundred and eighty-four wear marks were observed on 68 plates from the manor house. Most (79.95%) of these traces are exterior wear (n = 198) and interior wear <3 cm from the rim (n = 189) that likely occurred while enslaved people, the Hites, or their guests handled these vessels. Of the remaining traces, 35 are associated with eating (being on or within 3 cm of the base), and 62 are potentially associated with eating (located ≥3 cm from the rim or at an indeterminate location between the rim and the base). Basal wear consisted of 13 small to medium (1–7 × 1–4 mm) and six large (8–26 × 6–11 mm) spalls on three plates, and two narrow (1 mm wide) linear abrasions on two plates—one caused by a fork or spoon, and the other by a knife (see Griffiths Reference Griffiths1978). Traces on the plates’ sides that are (potentially) associated with eating primarily consist of spalls (n = 68) on 13 plates, including 34 small to medium spalls (1–6 × 1–4 mm) and 31 large spalls (5–15 × 5–13 mm). Seven areas of abrasive wear were noted on three plates’ sides, including one patch and six ovaloid striations(2–10 mm wide).
Eighty-six wear marks were observed on 42 plates from the quartering site. Most (75.58%) are exterior wear (n = 30) and interior wear <3 cm from the rim (n = 35). Of the remaining traces, 11 are associated with eating, and 10 are potentially associated with eating (see above). Basal wear consisted of six larger spalls (5–7 × 4–10 mm) on five plates and five narrow linear abrasions on three plates—three from a fork or spoon and two from a knife. Two of the three abraded plates (2.109 and 3.001) have decorative motifs that are not found at the manor house (Greer Reference Greer2022:193–197), suggesting that enslaved people likely purchased these vessels and, by extension, that they used utensils when eating from them. Traces on the plates’ sides that are (potentially) associated with eating consist of fatigue wear on five plates—including four small to medium spalls (2–5 × 1–3 mm) and one large spall—and three ovaloid/irregular abrasive striations on one plate.
Fifty wear marks were observed on 19 plates from the White-employee site, most (86%) of which are exterior wear (n = 18) and interior wear <3 cm from the rim (n = 25). The remaining seven traces include a medium-sized (4 × 3 mm) fatigue spall associated with eating on one plate, and six small to medium (1–4 × 1–6 mm) spalls on the base of another plate—although I could not determine whether the latter were on the vessel’s interior or exterior.
Comparing wear on plates can be difficult. Their bases provide the best evidence for how they were used, but bases are often undecorated, making them difficult to vesselize. Differences between the enslaved quartering site and the White sites, however, are apparent in the limited data at hand. Specifically, no interior small to medium fatigue wear was observed on basal sherds from the quarter, whereas these accounted for 75% of spalls from the manor house, suggesting that enslaved households and the Hites used plates in different ways.
Bowls
Forty-five wear marks were observed on 11 bowls from the manor house. Most (88.88%) are exterior wear (n = 26) and interior wear <2 cm from the rim (n = 14). The remaining wear consists of nine spalls on two bowls, including four small to medium (1–4 × 1–3 mm) spalls on the vessels’sides.
Twenty-three wear marks were identified on 10 bowls from the quartering site. Most (56.52%) are exterior wear (n = 12) and interior wear <2 cm from the rim (n = 1). The remaining wear includes three small to medium (1–3 × 1 mm), one large (7 × 5 mm), and one indeterminately sized spalls on five bowls. One small (2 × 1 mm), three medium (3–5 × 1–3 mm), and one indeterminately sized abrasions were oriented horizontally along three bowls’ sides.
Ninety wear marks were identified on five bowls from the White-employee site. Most (72.22%) are exterior wear (n = 38) and interior wear <2 cm from the rim (n = 27). Only one vessel had wear from eating, including a highly worn abrasive patch, 23 small to medium (1–3 × 1–3 mm) spalls, and one large (8 × 4 mm) fatigue spall.
Differences can be seen in the frequency and type of wear marks on the bowls. Most (54.5%) bowls from the enslaved site had wear marks from eating, half of which had horizontal abrasive striations. By contrast, 15.78% of bowls from the manor house and 11.11% from the White-employee site had wear from eating, and only one area of abrasion was identified—although it was a patch rather than a horizontal striation. This suggests that enslaved people used bowls in ways that were more likely to leave wear marks in general and abrasive striations in particular.
Mugs
Seventeen wear marks were identified on three mugs from the manor house. Most (64.70%) are exterior wear (n = 10) and interior wear <1 cm from the rim (n = 1). Two mugs had interior wear that resulted from the ways they were used. One had a large (12 mm long) fatigue spall, whereas the other had a large (8 mm long) fatigue spall and five medium abrasive striations (5–6 × 4–5 mm) whose orientation could not be determined.
Thirty wear marks were identified on six mugs from the enslaved quartering site. Most (76.66%) are exterior wear (n = 20) and interior wear <1 cm from the rim (n = 3). Four mugs had interior wear that likely resulted from the way these vessels were used. Two large (3–4 × 2–3 mm) spalls were identified on one mug, and seven horizontal abrasive striations were identified on three mugs, including two large (10–12 × 5 mm) striations (see Figure 3b). No mugs were recovered from the White-employee site.
Abrasive striations were the most common wear on the Belle Grove mugs. However, these traces were larger on vessels from the quartering site. Longer abrasive striations, such as those seen at the quarter, are the result of prolonged pressure applied to a vessel’s sides, an action that is more consistent with eating than preparing beverages (Greer Reference Greer2024:652), which suggests that enslaved people ate (semi-)liquid foods from their mugs, whereas the Hites did not.
Saucers
Thirty-six wear marks were identified on 14 saucers from the manor house. Most (66.66%) are exterior wear (n = 15) and interior wear <1 cm from the rim (n = 9). The remaining 12 traces are located on eight saucers. One saucer had a small (2 × 1 mm) fatigue spall on its base. Spalls were observed on the sides of five saucers, including three small to medium (2–4 × 2 mm) and five large (7–13 × 4–8 mm) spalls. Two saucers had a total of three small to medium (3–4 × 1–3 mm) abrasive striations, the orientation of which could not be determined.
Fifty-seven wear marks were identified on 17 saucers from the enslaved quartering site. Just under half (47.36%) are exterior wear (n = 20) and interior wear <1 cm from the rim (n = 7). The other 30 traces were located on 10 saucers. Five saucers had spalls on their bases, including three small (2 × 1 mm), three medium (4–5 × 3–4 mm), and one large (6 × 5 mm). Three saucers had medium to large spalls on their sides (4–6 × 2–3 mm). But the most common wear was abrasive striations, with 19 marks occurring on five saucers. Four abrasions on three saucers were oriented horizontally, including two medium striations (3–6 × 2–4 mm). Eight small to medium (2–5 × 1–2 mm) striations on three saucers were oriented vertically. An additional striation was oriented diagonally, and the orientation of the remaining six could not be determined.
Twenty wear marks were identified on 10 saucers from the White-employee site. Just over half (55%) are exterior wear (n = 8) and interior wear <1 cm from the rim (n = 3). The remaining nine traces were located on six saucers. Five spalls, including two small (1–2 × 1 mm) and two large (10–13 × 9 mm) spalls, were observed on the bases of three saucers. Two small (2 × 1–2 mm) and two large (6 × 5 mm) spalls were also recorded on the sides of three saucers.
Relatively consistent sets of small to large spalls were observed on the sides of saucers from all three sites. But the abrasive wear on vessels from the quartering site sets these saucers apart. Whereas two saucers from the manor house had abrasive striations, five vessels from the quarters had these traces. Nine of the 13 abrasions from the enslaved site whose orientation could be determined were vertical or diagonal, whereas no vertical or diagonal abrasions were seen at the other sites. Such traces are unlikely to occur when using saucers for their intended purpose of serving/consuming beverages (Greer Reference Greer2024:650–651). Instead, they are probably from enslaved people scooping (semi-)liquid foods upward along the saucers’ sides, a practice not seen at the White sites. Basal wear was also more common at the White-employee site than the manor house (possibly from placing teacups onto the saucers), potentially suggesting ways of using saucers that differed between these sites.
Teacups
Thirty wear marks were identified on 19 teacups from the manor house. Most (80%) are exterior wear (n = 17) and interior wear <1 cm from the rim (n = 9). The remaining four traces are small spalls (1–2 × 1 mm) located on the sides of three teacups. Compared to the saucers, the manor-house teacups are less likely to have interior wear >1 cm below the rim (16.66% vs. 7.46%). Fatigue wear dominated both vessel types (75% of traces on saucers and 83.33% on teacups), but the spalls were generally smaller on the teacups. Basal wear was noted on one of the eight saucers with base sherds but was absent among the five teacups with base sherds.
Thirty-one wear marks were identified on 16 teacups from the quartering site. Just under half (48.38%) are exterior wear (n = 13) and interior wear <1 cm from the rim (n = 2). The remaining 16 traces are located on eight teacups. Five spalls were identified on the sides of four teacups, although dimensions could only be recorded for one (4 × 2 mm). Eleven abrasive striations were identified on five teacups, seven of which were small (1–2 × 1 mm), and two were large (9–12 × 7 mm). All eight striations whose orientation could be determined were horizontal. Compared to the saucers, the quarters’ teacups are less likely to have interior wear >1 cm below the rim (25.64% vs. 16.00%). Abrasive wear dominated both vessel types (63.33% of traces on saucers and 68.75% on teacups). Striations on teacups, however, were generally smaller and more likely to be horizontal. Basal wear was also noted on five of the 12 saucers with base sherds but was absent among the three teacups with base sherds.
Thirty-seven wear marks were identified on 13 teacups from the White-employee site. Most (70.27%) are exterior wear (n = 20) and interior wear <1 cm from the rim (n = 6). The remaining 11 traces are located on the sides of six teacups. Five vessels have fatigue wear, including seven small (1–2 × 1–2 mm) and two large (5–7 × 4–5 mm) spalls. One vessel had a large (8 mm wide) medium abrasive striation whose orientation could not be determined. Compared to the saucers, teacups from the White-employee site are less likely to have interior wear >1 cm below the rim (37.50% vs. 18.18%). Fatigue wear dominated both vessel types (all traces on saucers and 90.90% on teacups), but teacups generally had smaller spalls. Basal wear was also noted on three of five saucers with base sherds but was absent among the three teacups with base sherds.
Teacups from the White sites had relatively consistent sets of wear marks that were distinct from saucers. This suggests that Belle Grove’s White inhabitants used teacups in ways that differed from the ways they used their saucers—and that this practice was common at both sites. Compared to the White sites, the quarters’ teawares are remarkably similar, with the primary difference being the size and orientation of abrasions, although some of these traces are large enough to suggest that enslaved people served food from teacups (Greer Reference Greer2024:651). It should be noted that enslaved people could have also consumed beverages from the same teacups from which they served food—especially if the latter use did not create much wear. This points to White and enslaved households using teacups in the different ways, with enslaved people using at least some teacups in the same ways they used saucers (i.e., for serving food).
Absorbed Residue
Use wear shows that enslaved and White households at Belle Grove used ceramics in different ways. This, however, only tells half the story. Understanding how these vessels were used also requires absorbed residue analysis (e.g., Fanti et al. Reference Fanti, Drieu, Mazuy, Blasco, Lugliè and Regert2018; Forte et al. Reference Forte, Cesaro and Medeghini2018). As people use ceramics, lipids from food and other substances can absorb into their paste, where they can survive for thousands of years (Evershed Reference Evershed2008; Reber Reference Reber2022). Absorbed residue analysis is rarely conducted on glazed vessels, such as the tea and tablewares discussed here (Pecci et al. Reference Pecci, Degl’Innocenti, Giorgi, Ángel Cau Ontiveros, Cantini, Potrony, Alós and Miriello2016:879–880), given that glazes provide a protective barrier that prevents residues from seeping into pastes. But use wear can create gaps in the glaze in which substances can become absorbed. To assess the type of foods/beverages served in the Belle Grove vessels, I analyzed 51 samples from 50 vessels using gas chromatography-mass spectrometry (GC-MS) at the University of Bristol (Supplementary Material 3).
To prepare the samples, I used a solvent-washed Dremel tool to clean all portions of the paste that had been exposed through use wear and removed any glaze adjacent to the wear mark (see Greer et al. [Reference Greer, Cramp, Mason, Schempp and Ali2026] for a more detailed description of these methods). Between 0.5 g and 2 g of the cleaned paste was crushed to a fine powder using a solvent-washed mortar and pestle, and lipids were extracted using acidified methanol (see Correa-Ascencio and Evershed Reference Correa-Ascencio and Evershed2014). This involved placing the powdered sample into a furnaced culture tube along with an internal standard (n-tetratriacontane at ∼1 mg per mL concentration) and 5 mL of H2SO4/MeOH 4%. I vortex-mixed the tubes and heated them at 70°C for one hour before transferring the contents to a furnaced test tube and centrifuging it for 10 minutes at 2,500 rpm. I then transferred the solvents from the test tube into a second culture tube along with 2 mL of dichloromethane-extracted double-distilled water. To remove additional lipids from the powdered sample, hexane was added to the first culture tube three times (3 mL the first two times and 2 mL the third), vortex mixed, transferred to the second culture tube, vortex mixed again, transferred to a furnaced 3.5 mL vial, and blown down under N2 until ∼3 mL remained. I derivatized aliquots of each sample by adding 20 μL N,O-bis(trimethylsilyl)trifluoroacetamide containing 1% trimethylchlorosilane, heating the sample at 70°C for one hour, and blowing it down under N2 to dryness before adding 100 μL of hexane.
Samples (1 µl) were screened and quantified using an Agilent Technologies 7890 gas chromatograph fitted with a flame ionization detector and nonpolar fused silica capillary column (Agilent DB-1ht, 100% dimethylpolysiloxane, 15 m × 0.32 mm × 0.1 µm film thickness). The oven temperature comprised a two-minute isothermal at 50°C, ramping to 350°C at 10°C/minute, followed by a 10-minute isothermal. Data were acquired and processed using ChemStation. Structural identification of compounds was performed using an Agilent Technologies 6890N plus network gas chromatograph coupled to an HP Agilent 5973 mass spectrometer. The temperature program comprised an isothermal at 50°C, ramping to 300°C at 10°C/minute, followed by an 8-minute isothermal. The MS was operated in EI mode at 70eV, with a GC interface temperature of 300°C and source temperature of 300°C. Data were acquired in full scan mode from m/z 50–650 Daltons. Data were processed using Xcalibur software and identified through comparison to NIST mass spectral library.
Twenty-six samples yielded significant concentrations of historic lipids (>5 μg g−1). Concentrations ranged from 7 to 496 μg g−1, with a mean of 105 μg g−1. Samples from the quartering site generally had larger concentrations than samples from the White sites (mean of 141 μg g−1 and 73 μg g−1, respectively)—a trend that is discussed in detail elsewhere (Greer et al. Reference Greer, Cramp, Mason, Schempp and Ali2026). A total of 325 compounds have been identified in the GC-MS data (Supplementary Material 4). Before continuing, I should make three quick notes. First, all samples contained palmitic acid (C16:0) and stearic acid (C18:0). Plants have more palmitic acid, whereas animal fats have more stearic acid (Evershed Reference Evershed2008). Therefore, the ratio of stearic to palmitic acids provides insights into the origins of the recovered lipids. Second, unsaturated fatty acids are rarely recovered from archaeological samples, given that they are more likely to degrade than saturated fatty acids, so their presence in archaeological samples is often indicative of contamination. Unsaturated fatty acids were common in the Belle Grove data, with oleic acid (C18:1) in all samples, erucic acid (C22:1) in 92% of samples, and palmitoleic acid (C16:1) in 81% of samples. The widespread presence of unsaturated fatty acids has been seen in other historic-era ceramics (e.g., Briggs et al. Reference Briggs, Cooper, Craig, Heron, Lucquin, Mercedes Martínez Milantchi and Samson2023; Pecci et al. Reference Pecci, Degl’Innocenti, Giorgi, Ángel Cau Ontiveros, Cantini, Potrony, Alós and Miriello2016; Poulain et al. Reference Poulain, Baeten, De Clercq and De Vos2016; Trinkley et al. Reference Trinkley, Hacker and Southerland2003:135; also see Greer et al. Reference Greer, Cramp, Mason, Schempp and Ali2026), likely because these compounds have not had time to degrade fully. Therefore, the unsaturated fatty acids in the Belle Grove ceramics are most likely related to historic foods/beverages. Third, cholesterol was identified in 31% of samples. Heating animal fats in ceramic vessels quickly degrades cholesterol (Hammann et al. Reference Hammann, Cramp, Whittle and Evershed2018), so this sterol is often assumed to be the result of contamination—especially when it co-occurs with squalene (Evershed Reference Evershed1993:88–90; Whelton et al. Reference Whelton, Hammann, Cramp, Dunne, Roffet-Salque and Evershed2021:2–3). However, squalene was not present in any Belle Grove sample, and cast-iron pots and pans were widely used in Virginia, strongly suggesting that the identified cholesterol is historic (also see Greer et al. Reference Greer, Cramp, Mason, Schempp and Ali2026).
Plates
Of the 11 plate samples, four were empty (<5 μg g−1) and six yielded significant concentrations of historic lipids (Table 3). Four plates from the quarters and manor house appear to be relatively similar, with low stearic-to-palmitic-acid ratios, long chain (LC; C12:0 to C20:0) fatty acids and (in all but one case) fatty alcohols, palmitoleic acid, erucic acid, and a relatively high ratio of oleic to stearic acids (Figure 4a). This suggests that enslaved people and the Hites ate similar foods from plates at times, even if they often handled the vessels differently (as indicated by the use wear). The remaining two plates, both from the manor house, are different from one another and from the other manor-house samples. The large amount of variation at the manor house suggests that the Hites and their guests consumed a variety of foods from plates.
Representative chromatograms of samples discussed in the text. (Color online)
Results of Residue Analysis.
a Palmitic (C16:0) and stearic acids (C18:0) not included in the long chain (LC) fatty acid column.
Bowls
Of the nine bowl samples, four were empty, and two—one each from the White-employee and quartering sites—yielded significant concentrations of historic lipids. All empty samples came from the White sites, suggesting that low-lipid foods (possibly corn; see Reber et al. Reference Reber, Dudd, van der Merwe and Evershed2004) were often served in these vessels. A relatively large stearic-to-palmitic-acid indicates that meaty dishes, probably stews/soups, were served from the employee site’s bowl (Figure 4b). The sample from the quarters has a high baseline and should be interpreted with caution, but it did have a low stearic-to-palmitic-acid ratio, suggesting that it was not used to serve meaty dishes (Figure 4c). Therefore, enslaved people and White employees likely served different foods from at least some of their bowls.
Mugs
Of the five mug samples, three yielded significant concentrations of historic lipids, with two from the manor house and one from the quartering site. All three samples appeared very similar, with LC and very long chain (VLC) fatty acids, palmitoleic and erucic acid, LC fatty alcohols, and low stearic-to-palmitic-acid ratios (Figure 4d). However, the use-wear data indicated that enslaved people ate food from this mug (see Figure 3b for an image of the wear associated with this sample), whereas the Hites likely consumed beverages from their mugs. Either the foods eaten from the mug had similar lipid compositions to the beverages consumed by the Hites or similar beverages were served in all three vessels, and enslaved people only ate low-lipid foods (corn, etc.) from their mug, which might be masked by the other compounds present in this sample.
Teawares
Of the 26 teaware samples, seven were empty and 15 yielded significant concentrations of historic lipids. I identified three lipid profiles—or collections of lipids that regularly co-occur. The first was seen in two teacups and a saucer at the White-employee site, and it consists of LC fatty alcohols, erucic acid, and moderate quantities of oleic acid, although this profile should be interpreted with caution given that several samples had high baselines (Figure 4e). The second was seen in a teacup and saucer from the manor house and a teacup from the quarters, and it consists of LC and VLC fatty acids, palmitoleic and erucic acids, and LC fatty alcohols (Figure 4f). Both profiles were only identified in teawares, suggesting that they are from substances that were only consumed from this type of vessel. Use-wear data did not indicate that White households served food from teawares, so these profiles are probably from hot beverages. The different profiles seen at the employee site may indicate that different hot beverages were consumed there (tea vs. coffee, etc.). Three of the six White sites’ uncontaminated saucers had significant lipid concentrations, two of which belong to the same lipid profiles as the sites’ teacups. This might be the result of “saucering,” or serving hot beverages in saucers, which was a common early to mid-nineteenth-century practice (Shadel Reference Shadel1971).
The third profile was identified in five saucers and one teacup from the quartering site, and it consists of LC and (with one exception) VLC fatty acids, palmitoleic and erucic acids, elevated levels of oleic acid, LC and VLC fatty alcohols, and cholesterol (Figure 4g). Most of these samples have stearic-to-palmitic-acid ratios of <0.30, which is often indicative of contamination. However, most methods of assessing contamination have been developed for low-fired, unglazed ceramics, not higher-fired, lead-glazed refined earthenwares (Evershed Reference Evershed1993; Whelton et al. Reference Whelton, Hammann, Cramp, Dunne, Roffet-Salque and Evershed2021). This is important when considering stearic-to-palmitic-acid ratios, because palmitic acid is more water soluble than stearic acid and therefore more likely to leech out from sherds after deposition (Whelton et al. Reference Whelton, Hammann, Cramp, Dunne, Roffet-Salque and Evershed2021:15). Both the presence of the glaze and the fact that my samples are higher fired (and therefore less porous; see Greer et al. Reference Greer, Cramp, Mason, Schempp and Ali2026) should reduce postdepositional leeching of palmitic acid. As a result, lower stearic-to-palmitic-acid ratios would be expected in these samples than in low-fired, unglazed ceramics. Three other factors suggest that this profile represents historic lipids. First, it is present in one sample with no evidence of contamination (Figure 4h), so the profile is not limited to samples with problematic stearic-to-palmitic-acid ratios. Second, this profile was only seen at the quarters and only in the site’s teawares. These samples were not handled in the same ways that other ceramics were, so it is unlikely that only the quarters’ teawares would have this specific contamination signature. Finally, this profile also aligns with the use-wear data, which suggested that saucers and some teacups from the quarters were used differently from the way teawares from the White sites were.
Although this lipid profile should be approached with caution, it suggests that enslaved people used their saucers and some of their teacups in ways that were different from those used by members of the White households, which, again, the use-wear analysis also demonstrated. Initial research on residues from one saucer (reanalyzed here as BGQ-R10) suggested that enslaved people ate cruciferous vegetables from these vessels (Greer Reference Greer2024). Lipids from the saucers/teacups are consistent with cruciferous vegetables, but the higher chain hydrocarbons needed to confirm the presence of leaf waxes are not recoverable using acidified methanol (e.g., Briggs et al. Reference Briggs, Cooper, Craig, Heron, Lucquin, Mercedes Martínez Milantchi and Samson2023; Charters et al. Reference Charters, Evershed, Quye, Blinkhorn and Reeves1997; Dunne et al. Reference Dunne, Chapman, Blinkhorn and Evershed2019; Komaitis and Melissari-Panagiotou Reference Komaitis and Melissari-Panagiotou1990). If additional testing confirms that this profile is from cruciferous vegetables, its cholesterol might be from small amounts of meat added to the greens, which was a common practice among enslaved Virginians (e.g., Franklin Reference Franklin and Orser2001:97).
Discussion
Susan Pollock (Reference Pollock, Robb and Pauketat2013:148) uses the concept of “space of action” (or Handlungsraum), which maps out the “limits on and possibilities for action,” to theorize how vessel form imposes social and material boundaries that constrain the ways people use them. In ancient Mesopotamia, Pollock (Reference Pollock, Robb and Pauketat2013) argues, increasingly specialized ceramic forms restricted people’s choices about how to use vessels, either due to social pressure to handle them in “appropriate” ways or because vessels’ physical attributes (rim diameter, etc.) made certain gestures difficult if not impossible. Applying this to Belle Grove, the White sites shared a similar conception of the space of action afforded by bowls, mugs, plates, saucers, and teacups—one that conformed to the social expectations among White Americans that bowls and plates should be used for eating and that saucers and teacups should be used for drinking. Enslaved people broadly conformed to this space of action for bowls, plates, and some teacups, given that use-wear and residue analyses indicate that they generally used these vessels in the same way as White households (eating from bowls and plates and drinking from some teacups). When it came to saucers, other teacups, and (at least some) mugs, enslaved people pushed beyond the social limits of this space of action to create new ways of using vessels defined by their physical attributes—in this case, their ability to hold (semi-)liquid food. Enslaved people may have used some vessels in both ways, with the possibility for action defined as much by what was being served as by predefined conceptions of how vessels should be used (also see Greer Reference Greer2024).
Detailed use-wear and residue analyses have not been conducted at other enslaved sites in the United States, but data on vessel form from 39 enslaved homesites in Virginia and South Carolina / Georgia suggest that the patterns at Belle Grove might be part of a broader trend (Figure 5). Bowls were always less common in Virginia, but at the end of the eighteenth century, enslaved Virginians began acquiring fewer bowls and more teawares. If these trends are related, it might suggest that enslaved people began expanding the socially accepted space of action for teawares around the turn of the nineteenth century. A similar decrease in bowls can be seen in South Carolina and Georgia, but it was not accompanied by the same increase in teawares, suggesting that the space of action associated with teawares retained their preestablished boundaries further south.
Comparison of vessel form in Virginia and South Carolina / Georgia. Ratios of bowls to plates and teawares are used here rather than overall percentages due to inconsistencies in how this data is presented in the various sources. Data and sources are presented in Supplementary Material 5. (Color online)
Differences in food rations issued to enslaved people might explain the growing gulf between these spaces of action in South Carolina / Georgia and Virginia. Pre-1840 rations in South Carolina / Georgia often included rice in the fall/winter and legumes in the spring/summer, both of which are best served in bowls (Greer Reference Greer2025). Corn was the primary starch issued to enslaved Virginians, who often used it to make bread (i.e., ashcakes/hoecakes) that was dipped into soups/stews (Franklin Reference Franklin and Orser2001; Greer Reference Greer2024:656, Reference Greer2025). This way of serving meals might not require larger vessels, such as bowls, if adequate servings of soup/stew can be placed within smaller saucers or teacups. By the 1790s, teawares cost less than bowls (Greer Reference Greer2024:658–660), potentially creating a financial incentive for enslaved Virginians to eat from teawares instead of bowls. Dismal Town, Virginia (44SK70; Goode Reference Goode2018), which has the highest ratio of bowls to plates in my Virginia data and the lowest ratio of teawares to bowls among late eighteenth- to nineteenth-century sites, provides additional evidence for this interpretation. Dismal Town is the only Virginia site in my dataset from a region where rice was grown and regularly issued as rations, further demonstrating a connection between rice rations and higher ratios of bowls.
Conclusion
One reason why bowls are rarely recovered from enslaved quartering sites in Virginia might be that enslaved Virginians began serving food from some mugs, saucers, and teacups around the turn of the nineteenth century, expanding the space of action for how these vessels were used to better accommodate enslaved foodways within the state. These results suggest that Otto’s (Reference Otto and South1977, Reference Otto1984) trends are predominantly seen in South Carolina and Georgia because they were largely tied to the specific rationing systems seen in these states. My analysis, however, focused on one nineteenth-century plantation, so my results may speak more to idiosyncratic practices at Belle Grove than broader trends throughout Virginia. Assessing this requires use-wear and residue analyses to be conducted at other quartering sites in Virginia, South Carolina, and Georgia. Although doing so is far beyond the scope of this article, my work provides a set of methods that can be used for this larger study and preliminary comparative data for assessing how future results might fit into trends in ceramic consumption practices in the two regions.
When assessing the role specific types of vessels played in past foodways, American archaeologists often rely solely on vessels’ formal characteristics (shape, thickness, etc.). Use wear is applied in some cases (e.g., Griffiths Reference Griffiths1978; Lesure Reference Lesure1998), but absorbed lipids are rarely used to study mundane aspects of cooking and serving meals (e.g., Fiorani et al. Reference Fiorani, Fundurulić, Manhita, Salvi and Dias2025; Reber et al. Reference Reber, Baumann, Monaghan and Myers2015). Few American studies combine these three lines of evidence. Yet, as shown above, combining these analyses provides more nuanced understandings of past foodways than any one or two of these techniques. The White sites show that the Hites and their employees used ceramic vessels in similar ways and that these uses broadly conform to modern assumptions about how bowls, plates, and teawares should be used—although the shared practice of saucering would have been impossible to identify without the absorbed residue analysis. Data from the enslaved quartering site, however, more clearly demonstrate the advantages of this combined approach. Previous studies have argued that enslaved people served food from teawares, although the only evidence presented to support these arguments consisted of the similar shape of these vessels and that of small bowls used to serve sauces and relishes throughout West Africa (e.g., Samford Reference Samford2007:99; Wilkie Reference Wilkie2000:147). Use-wear and residue analysis allowed my research to test this argument, showing that whereas saucers and teacups were used to serve food, teacups were also used to consume hot beverages. This work also shows that food was likely served in some mugs. In short, use-wear and residue analyses allow archaeologists to identify alternative ways of using ceramics that either do not align with modern assumptions about vessel use or are unprovable by studies that just use vessel form, and I hope that this study leads others to adopt these techniques in their own research.
Acknowledgments
I would like to thank Lucy Cramp, Brandi MacDonald, Rob Mason, Cavit Schempp, and Erica Moses for their assistance in various portions of this research. The two anonymous reviewers provided excellent feedback that greatly improved the manuscript. Joseph Wager provided the Spanish-language abstract.
Funding Statement
This research was supported by the National Science Foundation under Grant 2204085 and the Carter G. Woodson Institute for African-American and African Studies at the University of Virginia through their Pre-Doctoral Fellowship Program.
Data Availability Statement
Raw data and supplementary tables are available at https://doi.org/10.6084/m9.figshare.30813791.
Competing Interests
The author declares none.
Supplementary Material
For supplemental material accompanying this article, visit https://doi.org/10.1017/aaq.2026.10199.
Supplementary Material 1. List of vessels from the Belle Grove sites (table).
Supplementary Material 2. Results of use-wear analysis (table).
Supplementary Material 3. GC-MS sample register (table).
Supplementary Material 4. Lipids identified in GC-MS samples with significant concentrations of historic lipids (table).
Supplementary Material 5. Comparative ceramic data for Figure 5 (table).
Open access
