44PG64/65: The Windmill Site

Overview

44PG64/65 is a multi-component archaeological site in Prince George County, Virginia that was once part of a 1000-acre plantation known as Flowerdew Hundred, located on the south side of the James River about 30 miles west of Jamestown. The site was occupied intermittently and intensively over millennia, with most archaeological resources relating to two significant residential occupations: a Late Woodland Virginia Indian Weyanock town, and a seventeenth century English settlement, which includes the earliest windmill discovered in North America. The somewhat unusual name for this site (two numbers divided by a slash) is due to its location between the Stone House Foundation (44PG64) site and Fortified Compound (44PG65). These three sites were originally considered one site – 44PG3 — but were later split out into three individual sites. As a result, PG64/65 has elements related to both PG64 and PG65.

Initial, limited excavations of this site were conducted under the direction of Dr. Norman Barka and Leverette Gregory in 1972-1974 by crews from the College of William and Mary and Southside Historical Sites, Inc.  Most excavations were conducted by students from the University of California Berkeley during two field seasons – 1988 and 1995. Eyde Wilen led excavations in 1988 and Andrew Beahrs in 1995, both under the direction of Dr. James Deetz.

Thousands of lithic fragments (tools, debitage, fire-cracked rock) and a variety of pre-Colonial ceramic ware types, are evidence of intermittent Archaic and early Woodland-era Indigenous occupations of the site. A circular pattern of posts believed to be part of a palisade was likely a component of the Late Woodland/Contact Period Weyanock town. The most substantial feature identified during the UC Berkeley seasons consisted of a circular pit five feet in diameter associated with insitu iron hardware hoops and brackets. The size and shape of the feature and orientation of the hardware led excavators to interpret it as the remnants of a windmill built by George Yeardley in 1621, shortly after he patented Flowerdew Hundred in 1619. Thousands of fragments of seventeenth century ceramics, tobacco pipes, roofing tiles, brick fragments, and nails, indicate the area was part of the seventeenth century residential compound but beyond serving as the spot for the windmill, it remains unclear what activities took place in this area.

In 2018, The Digital Archaeological Archive of Comparative Slavery (DAACS) was awarded a grant from the National Endowment for the Humanities for the project titled, The Origins of a Slave Society: Digitizing Flowerdew Hundred (PW-259091-18). The grant’s goals included the identification, analysis and cataloging of all the contexts, maps, and artifacts from four of the earliest sites at Flowerdew Hundred, 44PG64, 44PG65, 44PG64/65, and 44PG92. Between 2020 and 2022, DAACS staff analyzed over 12,200 artifacts recovered during William and Mary’s and Berkeley’s excavations at 44PG64/65. 46% of the assemblage is comprised of lithics and indigenous ceramics related to the Weyanoke and earlier, Woodland and Archaic indigenous occupations. The remaining 54% of the assemblage, consisting primarily of architectural and domestic artifacts, relates to the early 17^th-century English occupation.

DAACS was unable to produce a seriation-based correspondence analysis of tobacco pipe bore diameters for PG64/65; however, the site-wide Binford Dates point to an occupation in the second quarter of the seventeenth century.  The Binford Mean Date is 1636 for the site, and more information on the DAACS chronology can be found in the Chronology section below.

Ethnohistoric and documentary evidence

The settlement of the triangular tract now known as Flowerdew Hundred began thousands of years before the arrival of English colonists at Jamestown. In the Late Woodland Period (AD 900-1600) and into the Early Colonial Period (1600-1650), the land was part of the territory occupied by the Weyanock, an Algonquian-speaking Indigenous community that was loosely affiliated with the larger political entity of the Powhatan paramount chiefdom (Barbour 1986). The 1607 John Smith map displays Weyanock settlements on both sides of the James River in the vicinity of what would be patented as Flowerdew Hundred roughly a decade later. Archaeologists, anthropologists and historians have noted that the Weyanock, like many other Powhatan communities, were living in dispersed riverside towns in the decades leading up to English colonists’ arrival (Binford 1967; Gallivan 2018; Rountree 1989, Turner and Opperman 1993).

In the early 1600s, the English began driving the Weyanock from their home territory along the James River. Archaeological research indicates that English colonizers at Flowerdew exploited the same areas that the Weyanock had cultivated for hundreds of years, and many of the archaeological sites at Flowerdew contain abundant evidence of Indigenous occupations spanning thousands of years. A complete culture history of the Flowerdew Hundred tract is provided on the Flowerdew Hundred plantation page. Here we provide a review of what little is known from existing documentary sources about the earliest sites at Flowerdew that contain evidence of Indigenous, white, and Black occupations, 44PG65, and its related sites, 44PG64 and 44PG64/65.

Established in 1617, Flowerdew Hundred was one of the largest plantations granted through the Virginia Company. Only seven of the thirty-five or so plantations in the region contained holdings of over 1000 acres (Ayers 1984). Two of these, Flowerdew Hundred and Weyanoke, together comprising over 3000 acres, were the property of Sir George Yeardley (Nugent 1934). Flowerdew Hundred, located on the southside of the James River, was comprised of a 1000-acre tract. Historical documents suggest that in 1618 Yeardley purchased the property from Stanley Flowerdew, his father-in-law, who had likely started a colonial settlement at the site as early as 1617.  When Yeardley returned from England to serve as the governor of the Virginia colony in 1619, he brought 15 men (whether indentured servants or tenants is uncertain) who were tasked with cultivating tobacco at Flowerdew (Kingsbury (ed.) 1906-1935; Morgan 1975).

In late August 1619, the White Lion, an English war ship, docked at Point Comfort, Virginia. The privateer “brought not any thing but 20. and odd Negroes” (McCartney 2019). These Africans, enslaved Angolans raided from the Portuguese slave ship São João Bautista off the coast of Campeche, were the first Africans to set foot in English North America. Yeardley, and the colony’s Cape Merchant, Abraham Peirsey, purchased these men and women. At least 11 were taken to Flowerdew Hundred and their unfree labor contributed to the success of Yeardley’s economic enterprises. The successful harvesting and trading of tobacco funded Yeardley’s acquisition of more unfree laborers in 1620 and 21, which resulted in substantial tobacco exports to England and Holland. By March 1620, 68 men, five women, and four children were living at Flowerdew Hundred (McCartney 2007:62). By 1622 as many as 114 tenants, servants, and enslaved individuals may have been laboring for Yeardley’s profit, mostly at Flowerdew Hundred and Weyanoke.

While Yeardley was technically an absentee owner since he resided 30 miles downriver in Jamestown, his choice to remain in Virginia enabled him to keep tighter control over the plantation’s development and maximize his profits (Musselwhite 2019). He also invested in the development of the property in a variety of ways. A letter from the Council of Virginia to the Virginia Company of London acknowledges Yeardley’s “good example” of constructing a windmill in 1621.

That windmill can be tied to Flowerdew specifically by its presence on a deed of sale in 1624. We know it was constructed, as the remnants of this windmill were identified during archaeological investigations at 44PG64/65 by Dr. James Deetz and students from the University of California Berkeley in 1995. Additionally, documentary evidence in the form of reports and testimony submitted to the court of the Virginia Company (Kingsbury (ed) 1906 II:374–375, 383) also demonstrate that by spring of 1623 a palisaded fortification with six pieces of mounted ordinance had been erected at Flowerdew. The archaeological site 44PG65 represents the remnants of this fortified compound.

Multiple documents show that the second wealthiest man in the colony, Abraham Peirsey, a businessman and Cape [Head] Merchant, purchased Flowerdew Hundred and at least some of the indentured and enslaved laborers from Sir George Yeardley on October 5, 1624.  A fragment of the deed recording the sale (Flowerdew Hundred Archives), a court deposition from Temperance Yeardley (George Yeardley’s widow) attesting to the sale (1627) and a patent to the property granted to Abraham Peirsey’s eldest daughter Elizabeth (Peirsey) Stephens in 1636 (Nugent 1934:30) all confirm the transaction. While Peirsey kept his primary residence in James City, a muster taken in 1624/1625 (depending on the use of a Julian or Gregorian calendar) lists a total of 10 households, presumably tenants, on the Flowerdew property, known at the time as Peirsey’s Hundred. A total of 21 people composed these households. Peirsey also had 29 indentured servants and seven unnamed African individuals listed as laborers. Thus, out of the total of 57 people living on the property 36 were either indentured or enslaved laborers.  The plantation contained 10 dwellings, three store houses, four tobacco houses, and one windmill (Barka 1993; Deetz 1993:20-23).

One major addition to Flowerdew that is likely not called out explicitly in the Muster is a sizeable dwelling house built upon an imported siltstone foundation. The archaeological site known as 44PG64 is the footprint of this structure and nearby features. The historical documentation of the house’s construction is scant. Unlike the windmill (44PG64/65), the dwelling house (44PG64) is not referenced explicitly in the 1624 deed of sale, although only the bottom fragment of the deed has been preserved. As a result, several hypotheses exist regarding whether Peirsey or Yeardley commissioned its construction, served as its primary occupants, and even its purpose within the plantation’s operations (Hodges et al. 2011:30-32).

Upon Peirsey’s death in January of 1628 property ownership passed to his widow, Frances (his second wife), and then to his daughter, Elizabeth Stephens when Frances died.  Eight years later, Elizabeth sold part of the property, which she had repatented as Flowerdew Hundred, to William Barker, a merchant and mariner.  Archaeological data, reviewed below, suggests that the dwelling house, 44PG64, was likely abandoned by the 1650s. Upon William Barker’s death in 1655 the plantation passed to his son John Barker. When John Barker died in 1673, Flowerdew’s ownership was transferred to his two sisters, Elizabeth and Sarah.

Excavation history, methods, procedures

College of William and Mary (WM) and Southside Historical Sites Inc. 

The archaeological resources at Flowerdew Plantation were first uncovered in 1961 by Gilford Holland and Benjamin McCary, who noted a series of early English historic sites super-imposed over Late Woodland or Contact Period Native American towns or occupations based on small test pits and surface deposits (University of Virginia/VDHR Archives, WM Summary Archaeological Activity n.d.). Archaeological investigations formally began at what would later be dubbed 44PG64/65 in 1972 after Leverette Gregory and State Archaeologist Edward Heite visited Flowerdew and conducted surface survey and testing that located early 17^th century sites. Gregory successfully convinced the Flowerdew Hundred’s then landowner, David A. Harrison III, to allow archaeological investigations, which resulted in the start of fieldwork at what was then designated as 44PG3 in 1972–1973 (Gregory 1975).

In 1974, Mr. Harrison established an agreement with the College of William and Mary to fund Southside Historical Sites, Inc., an organization charged with conducting field investigations under the direction of Dr. Norman Barka. The most intensive field seasons at 44PG3 were conducted from 1974–1978 by small crews of three to seven field archaeologists. One or two laboratory and curatorial assistants washed, labeled, and bagged artifacts.

The site designation 44PG3 originally used by William and Mary encompassed a larger area that was later broken out into three separate sites — 44PG64 (the Stone House Foundation Site), 44PG64/65 (The Windmill Site) and 44PG65 (the Fortified Compound Site). During the first few years of excavation, William and Mary established a single grid over the entirety of PG3.  Once it was determined that there were at least two distinctive settlements within PG3, excavators shifted to using separate designations for the sites. The sites were also nominated for inclusion on the National Register and assigned separate labels using the Smithsonian’s trinomial numbering system.

William and Mary and Southside excavators used an “AGNU” grid, which consisted of 40-by-60 foot grid blocks labeled with a three-digit number (e.g. 190, 191, 192 etc.).  Each block was further divided into 24 10-by-10 foot excavation quadrats which were given letters in alphabetical order across rows, starting in the northwest corner of the block and ending in the southeast corner. The letters “I” and “O” were excluded to minimize confusion with the numbers 1 and 0.

When PG3 was separated into three separate sites, excavators created a list of which 40-by-60-foot blocks should be assigned to each site. Vertical control was maintained by digging according to natural stratigraphy but at times arbitrary levels were also used within natural layers. Index numbers were assigned to each separate deposit or level designated by excavators. A combination of block number, quadrat letter, and index number was used as the provenience system for tracking artifacts.  For example, artifacts labeled PG3/256A1 are from 44PG64/65, block 256, quadrat A, and level index 1 (plowzone). Excavators recorded detailed information about each unit and level in the Level Book and captured plan views and profiles in drawn maps and photographic slides. The College of William and Mary and Southside Historical Sites Inc. conducted limited testing at PG64/65. Only five 10×10 units were excavated (256A, 258C, 258K, 258L, and 258T).

University of California Berkeley (UCB)
1988 Field Season

The University of California Berkeley began conducting fieldwork at Flowerdew in the mid-1980s. The 1988 field season was managed by UCB master’s student Eyde Wilen under the direction of Dr. James Deetz and with assistance from Flowerdew Hundred Foundation staff members Taft Kiser and Ann Markell. Fieldwork began on June 8th and ran through the end of July with volunteer field crews of UCB students, local high school students, and 25 faculty members taking part in a National Endowment for the Humanities Institute. The first task was to reestablish the AGNU grid used by William and Mary. Grid points were set at 254A (using a metal rod as a datum) and 255A (which is 60 ft from 254A). The goal of the field season was to investigate whether there was any archaeological evidence of a seventeenth century occupation in this area.  Field crews spent roughly the first month conducting surface survey.  This phase was completed by July 5th.  The collection was conducted using 10×10 units.  A comprehensive list of which units were surface collected could not be located.

A plot of artifact distributions displayed three clusters.  The cluster closest to PG64 was chosen for further investigation. The next phase of excavation entailed gridding off a 50 ft area. The 10-x-10-foot excavation units were broken down into four 5-x-5ft units with A(NW), B(NE), C(SW), D(SE) designations. The first two 10-by-10s selected were 323X and 326S. Each D quad’s plowzone level was shoveled off and 100% screened until the quad was down to subsoil. The other three quads were shovel shaved and the recovery method was not recorded in field records. On July 8 the decision was made to entirely grade off the plowzone.  The first grid blocks to receive this treatment were 323 and 393 but it was not recorded which 10-by-10 units were graded.

Underneath the plowzone a variety of features were identified including a large circular stain that was interpreted as a possible barrel well (F1), a large rectangular stain (F2) and a number of postholes.  Postholes were given a separate “P” designation and numbered separately from non-posthole features.  When the season wrapped up at the end of July three features had been identified along with 23 postholes.

1995 Field season

University of California Berkeley returned to the site in 1995.  The field season was managed by master’s student Andrew Beahrs under the direction of James Deetz. The season began on June 2nd. Crews used the same AGNU grid as the 1988 crew and worked in two areas 1) a 50×50 foot area that included 393 C, D, K, J, and Q and 2) a 40-by-20-foot area that included grid squares 323(Z), 324(U, V), 393(F, M, T) and 394(A, G, N, P). Following the system used in 1988, the 10-x-10 quadrats were divided into four 5-by-5 units with A(NW), B(NE), C(SW), D(SE) designations.

The recovery methods used in 1995 also varied. The plowzone deposits in the 50-by-50-foot area uncovered earlier in the season were troweled and screened. Excavators used a different strategy for the second cluster of quadrats.  The 10-x-10s were gridded and stripped.  In the 5-x-5s, one bucket of plowzone sediment was collected and screened. This was assumed to be a 6.7% sample based on estimates that three inches of plowzone were left by the grade all (a similar strategy was employed at PG92).  After sampling the plowzone was stripped down to subsoil by shovel shaving.

Features were renumbered from 1988 because two features that were originally designated as postholes (with the P prefix — P4 and P5) were found to be burials upon further investigation. The table below summarizes the reassignments. DAACS staff used the 1995 designations for context records entered into the database. It should be noted that DAACS did not receive artifacts related to these burials and data from the burials are not included in DAACS.

Feature Reassignment:

1988	1995
F1	F1
F2	F2
F3	F3
P4	F4
P5	F4
No assignment	F5
P10	F6
P2	F7
P3	F8
P6	F9
P7	F10
P1	F11
P8	F12
P9	F13
P11	F14
P12	F15
P13	F16
P13	F16
P14	F17

Research and Analysis

A full site report for the excavations conducted at PG64/65 does not exist.  The summary of research and interpretations below was developed by DAACS and draws from original field records, a summary report by Eyde Wilen (1990), and a summary of the Windmill research by Andrew Beahrs (n.d.).

College of William and Mary and Southside Historical Sites Inc.

Excavations conducted by W&M and Southside uncovered a series of driven posts running diagonally through five units in block 258 that they interpreted as a second circular palisade.  Plan view notes and drawings show the curvature.  The rest of the palisade was identified in block 259 (see PG65 site map).  Two postholes were also identified but the units bordering those with the features (E, F, and M) were not excavated so excavators were unable to determine whether these postholes were associated with any structures or specific human activities.

University of California, Berkeley

The most substantial feature excavated at the site by UCB archaeologists was Feature 1 in 393D and E. Initially, the circular shape and large size (~4.6 ft in diameter) led excavators to believe it was a barrel well.  Further excavation uncovered multiple pieces of in situ iron hardware. The hardware included two iron hoops, one about 12 inches in diameter and a second 8 inches in diameter. The 12-inch ring was situated roughly 15 inches above the smaller ring. Excavators suggested the rings held in place a large wooden post.  Both rings also showed traces of the wooden post they had presumably encircled. Iron spikes, some 2 ¼ inches long, were also found, some of which still adhered to the rings.

These rings were surrounded by evenly spaced iron D-shaped brackets (four pairs and two single examples). The brackets were level with the upper ring. The hardware appeared to have held in place a substantial wooden frame, probably circular. Within the clamps were iron “staples”, varying in length from 8 to 12 inches, shaped upwards into points at each end, with traces of wood adhering to them. These clamps appear to have held a timber frame together by being hooked tightly around it.

Comparisons with documentation and extant early seventeenth-century windmills in England (Beahrs n.d.) found the clamps and rings’ configuration to be consistent with sunken trestle post mills that date to the late medieval period. These similarities led excavators to interpret the pit and associated hardware as the remnants of the windmill built on the property in 1621 under Yeardley’s ownership. Researchers noted the mill would have been a small one, likely supporting a sail with a span of some twelve feet, and standing no higher that fifteen feet above the ground. The quarter bars supporting the center post could have been mortised into an underground trestle which was connected in some way to a wooden frame held together with the “D” shaped clamps.

Feature 2 was the other notable discovery.  The feature was recorded as a large rectangular pit (5ft x 6ft) that bottomed out about 8 inches below the interface of plowzone and subsoil. Excavators did not interpret the pit. Excavators also uncovered three burials and one feature that contained portions of human remains but wasn’t entirely clear if it was an in situ burial (F3, 4, 5, and 25). Initial interpretations were that all the burials were Indigenous based on physical characteristics and lack of grave goods.

Collections History

Over the past fifty years, four different institutions have assumed responsibility for the curation of the Flowerdew collection. From 1971 to 2007, the archaeological assemblages were split between the Flowerdew Hundred Museum, located on the Flowerdew Hundred property, the College of William and Mary, and the Hearst (formerly Lowie) Museum at the University of California, Berkeley. It has only been in the last decade that most of the artifacts, field records, and sitemaps have been reunited at the University of Virginia. When the Harrison family sold Flowerdew Hundred in 2007, they donated the entire archaeological collection to the University of Virginia’s Harrison Institute-Small Special Collections Library. From 2007 to 2018, Karen Shriver, the collection’s curator and only permanent staff member, worked to reunite the archaeological assemblages with the smaller collections and field records held by William and Mary, UC Berkeley, and various PIs. The artifacts and field notes are in highly variable conditions. When housed at the Flowerdew Museum, a busy public program schedule, focus on field work, and general lack of funding for collections management meant that the limited staff processed the collections in a piecemeal fashion. The result is that a sizeable portion of the collection remains partially processed and cataloged. While a box inventory has been created, there is no systematized or complete itemized catalog for the entire collection. The collection’s current curator, Meg Kennedy, has been successful in locating additional archival materials at the College of William and Mary and is working to create a full inventory of the entire 400,000+ items that comprise the Flowerdew Hundred collection.

2018-2023 DAACS Analysis

In 2018, DAACS was awarded a grant from the National Endowment for the Humanities for our project titled, The Origins of a Slave Society: Digitizing Flowerdew Hundred (PW-259091-18). The grant’s goals included the identification, analysis and cataloging of all the contexts, maps, and artifacts from four of the earliest sites at Flowerdew Hundred, 44PG64, 44PG65, 44PG64/65, and 44PG92. Jillian Galle and Elizabeth Bollwerk were co-Principal Investigators on the grant, and Bollwerk directed the work based at the DAACS Lab at Monticello and a satellite lab at the University of Virginia.

For 44PG64/65, and with the help of interns from the University of Virginia, DAACS staff rehoused, identified, analyzed, and cataloged 12,268 artifacts. They also analyzed and digitized the 654 field records associated with William and Mary’s and UC Berkeley’s excavations of deposits and features at the site. DAACS Senior Archaeological Analyst Leslie Cooper compiled and digitized the numerous maps from all seasons of excavation, which are available through the images and maps section. Catherine Garcia finalized the 44PG64/65 maps. DAACS analysts responsible for identifying, photographing, cataloging and analysis of the artifacts from 44PG64/65 included Sarah Platt, Iris Puryear, Allison Mueller, Cate Garcia, and Elizabeth Bollwerk. Lindsay Bloch of Tempered Archaeological Services identified, analyzed, and cataloged the Indigenous ceramics from the site. We have relied on help of early 17th-century material culture specialists, Merry Outlaw and Bly Straube, coarse earthenware specialist Lindsay Bloch, small finds specialist Sara Rivers Cofield, and lithic specialists Dennis Blanton, Charles Cobb, and Christopher Egghart.  DAACS Diversity Interns from UVA included Shaheen Alikhan, Emily Anderson, Cindy Gwana, Adrienne Preston, Brittany Ivy, Jenna Owens, and Macie Clerkley.  All of this work was done in collaboration with the University of Virginia’s Special Collections Library, which curates the Flowerdew Collection, and we are grateful for the consistent and efficient help and continuing support of Meg Kennedy and Brenda Gunn. The site was launched on the DAACS website in December 2023. Detailed artifact and context data can be found in the Query the Database section of this website.  The physical collection (field records and artifacts) is curated at the Flowerdew Hundred Lab at the University of Virginia.

With the standardized archaeological data that resulted from this project, Elizabeth Bollwerk and Fraser Neiman developed a chronology for 44PG64/65, the detailed results of which can be found on the site’s Chronology page.  The analyses were unable to produce a statistically significant seriation-based chronology for 44PG64/65; however, the site-wide Binford Dates point to the occupation’s temporal placement as the first and second quarter of the seventeenth century.