New River Village I

• Overview
• Documentary evidence
• Excavation history, procedure and methods
• Summary of research and analysis

Overview

Late-18th- to early-19th-century plat of the New River Estate (Suffolk Record Office HA178/1/55).

Archaeologists affiliated with the St. Kitts-Nevis Digital Archaeology Initiative (SKNDAI) began work at the site of the New River estate’s slave village in May 2008. Three hundred and eighty-one shovel test pits (STPs) were excavated at the New River I village between May and July. In addition, a total station and GPS unit were used to digitally map the landscape and extant structures related to the estate, which was established in the mid-1700s and was occupied through emancipation in 1834. Instrumental Neutron Activation Analysis (INAA) and petrography were conducted on a sample of the Afro-Caribbean ceramics from this village. Preliminary analysis on the archaeological data from New River I and New River II suggest that enslaved people lived at New River I from about 1750 until 1780.  The village was then moved to the area we identify as New River II. The New River II village was occupied from around 1800-1830.

The STP survey at New River are part of an international collaborative fieldwork project in the Caribbean known as The St. Kitts-Nevis Digital Archaeology Initiative. The Initiative was funded through a Transatlantic Digital Collaboration Grant from The National Endowment for the Humanities (NEH, US) and the Joint Information Systems Committee (JISC, UK).  The award was made to The Digital Archaeological Archive of Comparative Slavery (DAACS), The University of Southampton, and the International Slavery Museum, Liverpool.

The New River villages are also a key component of a multiple-year research program in the Caribbean spearheaded by DAACS known as the DAACS Caribbean Initiative (DCI). DCI’s immediate goal is to document archaeologically, through survey, excavation and collections analysis, the trajectories of change in slave lifeways on the north and south coasts of Jamaica and on the small islands of Nevis and St. Kitts during the 17th and 18th centuries. DCI’s ultimate goal is to improve our understanding of the causal forces that shaped the evolution of slave societies throughout the early-modern Atlantic World by giving researchers access to easily searchable and comparable data from archaeological sites throughout the Caribbean, Carolinas and Chesapeake.

Documentary evidence

Looking east across the New River I Village toward the New River works and the Atlantic Ocean. Photograph taken May 2008.

Located on the windward side of Nevis, the New River Estate was established in the early 1720s. The first recorded mention of the plantation was in 1794, when William Earle, planter, conveyed the plantation to Thomas Butler, a merchant who was listed as formerly of Nevis but now of Great Britain. The deed located the New River plantation in the parish of St James and noted that it was around 50 acres, bounded on the northeast with the sea, on the southwest by the lands of William Earle and Charles Williams, to the northeast with lands belonging to Elizabeth Hill, Charles Williams and the lands formerly of John Wilkinson, and to the southeast with the land of Robert Easter (Common Records 1707-1728, fols.553-4).

A listing of the militia made in 1698 listed three male heads-of-household who were named Earle, as well as their family members and the number of enslaved Africans that they owned at the time. Roger Leech, who conducted the historical research on The New River Estate for this project, concludes that this family most likely owned New River.

The origin of the name ‘New River’ has not been established. Leech theorizes that there may have been an ownership, family, investment or some other association with the New River Company established in the early 17th century to provide a new water supply to London. The naming of a nearby ghut as ‘New River Gutt’ need not mean that this was a new watercourse named as ‘New River’. The ghut might simply be named after the plantation.

Map of New River Estate, including great house, works, and the two slave villages.

By 1739 Thomas Butler lived in Camberwell, Surrey, a suburb of London on the south bank of the Thames. In his will of that year all his plantations and slaves on Nevis were left to his three sons, John, James and Duke Butler (Oliver 1912:60). John Butler was later linked by marriage to the families of Pemberton and Maynard, the latter subsequently the owners of New River. In 1745 Thomas Butler’s three sons sold or mortgaged an estate in the parish of St James, probably New River, to William Clarke of Camberwell and John Hooke of Portsmouth in Hampshire (Oliver 1912:63). By 1785 the one third share of the estate of Duke Butler, the minister of Okeford Fitzpaine in Dorset, had descended to his children, Thomas, James, William, Mary (the wife of George Ryves Hawker) and Jane (the wife of Robert Frome), who then sold their share to Thomas Coxhead, a merchant of Great Hermitage Street, St George’s, Middlesex, by then part of the west end of London (Common Records 1788-9, 121-154; the conveyance that can be viewed through a DAACS document query gives a full list of the slaves included in the sale).

The 1785 indenture between Duke Butler and other provides the names and general ages of each of the 129 enslaved individuals laboring at New River. That year 21 men, 46 women, 34 boys, and 28 girls lived at New River. Several names are embedded with clues to a person’s ethnicity, age, origin or parentage. In these cases, two names are provided for a single individual, one proper name and one descriptor. For example, three men carried names that allude to their ethnicity, Indian Robin, Africa Lawrey and Cubenna Billey.

No men or women had occupations linked to their names but two women carried names suggestive of their origins in Africa or the creolized Caribbean. These women included ‘Ebbo Frankey’ and ‘Mulatto Rittah’. ‘Old Juggy’ and ‘Old Betty’ were most likely amongst the oldest enslaved women on the property. Children’s names such as ‘Young Juggy’ and ‘Little Abba’ and ‘Mulatto Jemmy’ may also provide clues to parentage and family structure at New River Plantation.

Figure 1. Plot of the 400 STP assemblages from the New River villages on the first two axes of CA. Points represent STP assemblages. Points that are close to one another have similar type frequency profiles.

One boy and one girl had names indicating they had been born on or purchased from another plantation, ‘Tom Maden’ and ‘Maria Maden’ were both probably connected to Madan’s Plantation a few miles to the north of New River in the same parish of St James.

By 1765 the estate was said to be late of Josiah Webbe of New River esq. decd. and now of Walter Nisbett and William Maynard esqs. (Common Records 1764-7, fol.167). These were probably the lands that belonged by the early nineteenth century to the Maynard family of Suffolk (Suffolk Record Office HA178). The records of the Maynard family include a plat of New River made in the late eighteenth or early nineteenth centuries, which has enabled the location of the slave village to be identified. It was located to the west of the plantation house and works. The plat also shows the southern boundary of the estate as the New River ghut, the boundary between the parishes of St James and St George. The New River estate contained by this date c.267 acres (Suffolk Record Office HA178/1/55).

Walter Maynard’s will, made in 1804, noted that he had expended ‘large sums on New River Estate, now belonging to Messrs. Lane, Son, & Fraser, under a promise … that he should become the purchaser, and … have contracted, or are about to contract, with … [his] son Walter Maynard for sale of such estate, he directs that should the sale be completed on terms in contemplation, son Walter shall have no share in the Gingerland Estate’ (Oliver 1913:340).

Excavation history, procedure and methods

Figure 2. Plot of ware types from New River of CA axes 1 and 2. Note early types on the right, late ones of the left.

New River is located in the Atlantic side of Nevis. The site of its 18th-century slave village is identified on the late-18th to early-19th-century plat discovered by Roger Leech among the Maynard Papers. In 2006, with a copy of the late-18th-century plat in hand, Neiman, Galle and Leech conducted a walking survey of the area identified on the plat as the New River slave village. A dense surface scatter of ceramics dating from the mid-18th through early 19th-centuries provided convincing evidence that we had located of the village. The estate is currently owned by Nevis’s Department of Agriculture and nearly half of the site is under vegetable cultivation. Stone field terraces, most likely constructed in the early 19th-century, continue to shape the nature of the fields. No excavations or surface collection were conducted in 2006.

With funding from JISC-NEH, archaeologists with SKNDAI returned to the New River village in May 2008 to begin STP survey. Galle, Neiman, DAACS staff and students from the University of the West Indies, Mona, excavated 222 STPs at the New River village between May 22 and June 15, 2008. Core project staff (Drs. Neiman, Leech, and Philpott) returned to Nevis on June 26. They were accompanied by a field crew of four DAACS staff and field school students from Southampton University. Between June 26 and July 9, the collaborative research team dug 161 pits to complete the shovel test pit survey at what we had come to confirm was the mid-18th-century village at New River. Archaeologists also excavated 17 pits at what was later identified as New River II, a late-18th- to mid-19th-century village located on the south side of the New River ghut.

New River I was divided into two survey areas. Area 1 consisted of fenced vegetable fields. Area I has been under cultivation for a good portion of the past two hundred years and it contained remnants of late 18th-century stone terraces for sugar fields. Area 3 consisted of an uncultivated zone of acacia scrub east of the Area I. It contains a modern shed with a poured concrete base.

Figure 3. Plot of means ceramic dates (MCD) against CA dimension-1 scores for New River STPs.

We used a total station and GPS to place a UTM grid across the site. All shovel-test-pits were placed on 6-meter centers using a total station. An alphanumeric system was established for naming STPs that combined the Area, the Transect Letter, and the STP number. Transects were labeled alphabetically across the site. STPs were numbered consecutively within each transect. As a result, STP context numbers follow this format: 13-H-01, which translates into Pit 1, on Transect H, in Area 3.

All STPs were 50 centimeters in diameter and all excavated sediment was screened through 1/4-inch mesh. In most cases, the pits were excavated to subsoil. All recovered artifacts were washed and flown to the DAACS lab at Monticello, where they were cataloged to DAACS standards. The artifacts will be returned to Nevis in 2011.

Summary of research and analysis

This section briefly outlines results from DAACS’s preliminary analysis of archaeological data recovered from the New River village sites in 2008. It includes some comparative data from the villages at the Jessups Estate, located on the leeward-side of Nevis. The analysis is intended only to evaluate the analytical potential of the data and thereby the utility of the field research and digitization design that we have employed. We cannot hope to exhaust that potential here.

Figure 4. Change over time in the proportion of Afro-Caribbean ware, relative to imported ceramics at New River I and II. The central line shows the trend. The lines on either side are 95% confidence limits.

Our initial research has focused on dating the site. We show that is it possible to accurately pinpoint the beginning and ending dates of each settlement and chart change in the intensity and location of occupation within settlements over time. With dates in hand, we then assess the extent to which it is possible to chart change over the course of the village occupations in the frequency of an artifact class that is a major focus of the project: Afro-Caribbean ware.

Our chronology relies on frequency seriation. The seriation method assumes that the relative frequencies of types, in our case ceramic ware types, in a suite of temporally successive assemblages follow lenticular or Gaussian curves. If this is right, then an ordering of undated assemblages in which type frequencies display this pattern is likely to be a chronology (Dunnell 1970). We rely on two complementary methods to estimate order: correspondence analysis (CA) and mean ceramic dates (MCDs) (Ramenofsky, Neiman, and Pierce 2009, Smith and Neiman 2007). CA converts a data matrix of type frequencies into a set on scores which estimate the positions of the assemblages on underlying axes or dimension of variation. MCD’s are weighted averages of the historically documented manufacturing date for each ware type found in an assemblage, where the weights are the relative frequencies of the types. Measuring the correlation between CA axis scores and MCDs offer an indication of whether the CA scores capture time (Ramenofsky, Neiman, and Pierce 2009).

Because artifact samples from individual STPs are small, a large proportion of variation among assemblages is the result of sampling error, which obscures any meaningful pattern. To reduce sampling error, we rely on empirical-Bayes smoothing methods (Robertson 1999). Once the ceramic assemblages from each STP were fit into a single chronological framework, we used regression splines to extract the trajectory of change in AC ware.

Excavators digging shovel test pits across Area 3 of New River I.

Our survey team excavated a total of 383 STPs at this site (New River I), covering all of its documented spatial extent. In addition, the team discovered the location of a second, later slave village site through surface reconnaissance (New River II). Seventeen STPs were excavated at New River II, to allow a preliminary assessment of its age. Complete exploration of New River II awaits additional funding.

CA of Bayes-smoothed ware type frequencies from the New River I site fits the seriation model well: a plot of the 400 STP assemblages on the first two CA axes assume the characteristic U-shape that betrays good fit to the seriation model (Figure 1). The complementary plot of the ware types in the same space shows that the earlier assemblages lie on the right of the plot, and the later ones are on the left (Figure 2). This is confirmed by plotting the MCDs for each assemblage against their CA axis-1 scores (Figure 3).

The MCDs indicate that the New River I village was occupied from about 1750 until 1780. The occupation span for New River II runs from about 1800 to 1830. This implies that the site was abandoned at emancipation. The gap between these two spans is small, and may be the expected outcome of the time-averaged character of the assemblages. Clarification of this point requires larger samples from New River II. We tentatively conclude that New River II was occupied when New River I was abandoned and that there was a single massive shift of slave housing from one site to the other.

What caused this shift? We are currently exploring several hypotheses. The first is that the change was implemented by New River’s owners as part of a larger strategy to increase the efficiency and scale of sugar production. Archaeology shows the terraces that traverse the old village site post date the slave occupation. This implies the area was put into cultivation after it was abandoned. The location of these new sugar fields adjacent to the mill and boiling house complex minimized the costs of transporting sugar to them. The new village location was more distant from the mill complex. Hence growing cane on the old village site was more efficient that growing it on the new village site.

The move may have had an unintended benefit for enslaved people because it placed them closer to a deeply eroded drainage or ghut that may have been a water source. Access to water became even less onerous with the construction of a cut-stone cistern for water storage on the site at some point during the occupation. This facility represents a considerable investment by New River’s owners. However, why they made it is not clear. One possibility is an increased concern for the well being of enslaved workers, perhaps linked to the end of the slave trade in 1807. A second hypothesis is ecological: the ghut adjacent to the new village only contains water today during storms. Its dry condition may date to the early-19th century.

Excavating STPs around tomoto plants in Area 1 of New River I, May 2008.

A second major focus of research is the role that Afro-Caribbean wares played in the lives of enslaved people on Nevis and St Kitts. These ceramics are hand made and open-fired. They were likely produced on the household level by enslaved women. The 2008 research included neutron-activation analysis and petrographic studies designed to rigorously evaluate this hypothesis. Data from these tests will be available through DAACS in September 2009.

Afro-Caribbean wares relate directly to two of the larger themes that motivate our research. First, archaeologists have often assumed that these ceramics are evidence for the conservation of African potting traditions in the Caribbean. However, an equally plausible hypothesis is that Afro-Caribbean ceramics represent a strategic reinvention of African traditions to meet challenges that were unique to life on Nevis under slavery. If these ceramics primarily represent continuity in African cultural practice, we would expect their frequency, relative to imported, specialist-produced ceramics, to decline with the passage of time.

The second question is when and if slaves on Nevis acquired the motive and means to participate in the consumer revolution, in this case by replacing household-produced pottery with European ceramics whose acquisition required cash (Carson 2003, Galle 2010, Neiman 2005). The seriation chronology developed above offers an opportunity to shed light on both these issues at New River.

We plotted the proportion of Afro-Caribbean ware in each STP against the CA axis-1 scores, which we know capture time. In order to put the result on a calendar year scale, we regressed the MCDs on the Axis 1 scores. We use the predicted MCDs to preserve the statistically desirable properties of the CA scores while re-expressing them on a rough calendar year scale. The final step is to fit a regression spline with binomial errors to the AC-Ware proportions. The fitted curve is plotted against the predicted MCDs in Figure 4.

The trend is astonishingly clear. Afro-Caribbean ceramics comprise a minority of the New River ceramic assemblages until the 1770’s. Their popularity peaks in the 1780’s and then declines until emancipation. The first half of the trend supports the idea that A-C wares are an adaptive response to conditions that enslaved Africans encountered at New River, although they may have drawn on traditional African knowledge about ceramic production. The fact that the frequency peak occurs around 1780 hints that one salient environmental factor may have been the American Revolution, which sharply curtailed shipping from Europe, Africa, and North America into the Caribbean. During this period the importation of provisions, slaves and goods into the Caribbean slowed to a trickle (O’Shaughnessy 2000).

The post-1780 decline in Afro-Caribbean wares may be linked not only to resumed importation of ceramics from England but also, perhaps after 1800, to an increase in the effort that enslaved people put into acquiring fancy ceramics. Disentangling these two factors requires larger samples of ceramics from the later village, which will make it possible to resolve whether there really are two inflection points in the downward curve, suggesting two causes.