ARCHEOLOGY
AND
ARCHITECTURE

VII

The Site, its Problem, and Preliminary Tests

The preceding chapters have presented written evidence of Marlborough’s history and of the human elements that gave it life and motivation. Assembled mostly during the years following the excavations, this information was not, for the most part, available in 1956 to guide the archeological survey recounted here. Neither was there immediate evidence on the surface of the planted fields to indicate the importance and splendor of Marlborough as it existed in the 18th century.

In 1954, when Dr. Darter proposed that the Smithsonian Institution participate in making excavations, he presented a general picture of colonial events at Marlborough. He also provided photostats of the two colonial survey plats so frequently mentioned in Part I (fig. 2). From information inscribed on the 1691 plat, it was clear that a town had been laid out in that year, that it had consisted of 52 acres divided into half-acre lots, and that two undesignated acres had been set aside for a courthouse near its western boundary. It was known also that John Mercer had occupied the town in the 18th century, that he had built a mansion there, that a circular ruin of dressed lime-sandstone was the base of his windmill, and that erosion along the Potomac River bank had radically changed the shoreline since the town’s founding 263 years earlier. But nobody in 1954 could point out with any certainty the foundation of Mercer’s mansion, nor was anyone aware of the brick and the stone wall system, the two-room kitchen foundation, or the trash pits and other structures that lay beneath the surface, along with many 18th-century household artifacts. It remained for the archeologist to recover such nonperishable data from the ground.

In August 1954 Messrs. Setzler, Darter, and Watkins spent three days at Marlborough examining the site, making tests, and, in general, determining whether there was sufficient evidence to justify extended excavations. The site is located in the southeastern portion of what was known in the 17th century as Potowmack Neck (now Marlborough Point), with the Potomac River on the east and Potomac Creek on the south (map, front endpaper). It is approached from the northeast on Highway 621, which branches from Highway 608 about 2½ miles from the site. Highway 608 runs from Aquia Creek westward to the village of Brooke, situated on the Richmond, Fredericksburg, and Potomac Railroad about four miles east of the present Stafford courthouse on U.S. Route 1. Highway 621 takes a hilly, winding course through the woods until it debouches onto the flat, open peninsula of the point. The river is visible to the east, as the road travels slightly east of due south, passing an intersecting secondary road that runs west and south and then west again. The latter road ends at the southwestern extremity of the Neck, where Accokeek Creek, which meanders along the western edge of the Neck, feeds into Potomac Creek. At the point near the Potomac Creek shore where this road takes its second westerly course lies the site of the Indian village of Patawomecke, excavated between 1938 and 1940 by T. D. Stewart.

Beyond this secondary road, Highway 621 continues southward to a small thicket and clump of trees where it curves sharply to the east, its southerly course stopped by fenced-in lots of generous size (with modern houses built on them) that slope down to Potomac Creek. After the highway makes its turn, several driveways extend from it toward the creek. One of these driveways, obviously more ancient than the others, leaves the highway about 200 feet east of the clump of trees, cutting deeply through high sloping banks, where vestiges of a stone wall crop out from its western boundary (fig. 22), and ending abruptly at the water’s edge. Highway 621 continues to a dead end near the confluence of creek and river.

Some 200 feet west of the turn in the highway around the clump of trees, is a deep gully (or “gutt” in 17th-century terminology) that extends northward from Potomac Creek almost as far as the intersecting road that passes the site of the Indian village. This gully is overgrown with trees and brush, and it forms a natural barrier that divides the lower portion of the point into two parts. A few well-spaced modern houses fringe the shores of the point, while the flat land behind the houses is given over almost entirely to cultivation.

Since the two colonial land surveys were not drawn to scale, some confusion arose in 1954 as to their orientation to the surviving topographic features. However, the perimeter measurements given on the 1691 plat make it clear that the town was laid out in the southeastern section of the point, and that the “gutt” so indicated on the plat is the tree-lined gully west of the turn in the highway.

Bordering the clump of trees at this turn could be seen in 1954 a short outcropping of brick masonry. A few yards to the north, on the opposite side of the road, crumbled bits of sandstone, both red and gray, were concentrated in the ditch cut by a highway grader. In the fields at either side of the highway, plow furrows disclosed a considerable quantity of brick chips, 18th-century ceramics, and glass sherds.

In the field east of the clump of trees and north of the highway, opposite the steep-banked side road leading down to Potomac Creek, could be seen in a row the tops of two or three large pieces of gray stone. These stones were of the characteristic lime-sandstone once obtained from the Aquia quarries some four miles north, as well as from a long-abandoned quarry above the head of Potomac Creek. It was decided to start work at this point by investigating these stones, in preference to exploring the more obvious evidence of a house foundation at the clump of trees. This was done in the hope of finding clues to lot boundaries and the possible orientation of the survey plats. Excavation around these vertically placed stones disclosed that they rested on a foundation layer of thick slabs laid horizontally at the undisturbed soil level. Enough of this wall remained in situ to permit sighting along it toward Potomac Creek. The sight line, jumping the highway, picked up the partly overgrown stone wall that extends along the western edge of the old roadway to the creek, indicating that a continuous wall had existed prior to the present layout of the fields and before the construction of the modern highway.

The excavation along the stone wall was extended northward. At a distance of 18.5 feet from the highway the stone wall ended at a junction of two brick wall foundations, one running north in line with the stone wall and the other west at a 90° angle. These walls, each a brick and a half thick, were bonded in oystershell lime mortar. Test trenches were dug to the north and west to determine whether they were enclosure walls or house foundations. Since it was soon evident that they were the former, the next question was whether they were lot boundaries matching those on the plat. If so, it was reasoned, then a street must have run along the east side of the north-south coursing wall. Accordingly, tests were made, but no supporting evidence for this inference was found.

Nevertheless, the indications of an elaborate wall system, a probable house foundation, and a wealth of artifacts in the soil were enough to support a full-scale archeological project, the results of which would have considerable historical and architectural significance. Determining the meaning of the walls and whether they were related to the town layout or to Mercer’s plantation, learning the relationship of the plantation to the town, discovering the sites of the 1691 courthouse and Mercer’s mansion, and finding other house foundations and significant artifacts—all these were to be the objectives of the project. The problem, broadly considered, was to investigate in depth a specific locality where a 17th-century town and an 18th-century plantation had successively risen and fallen and to evaluate the evidence in the light of colonial Virginia’s evolving culture and economy. Accordingly, plans were made, a grant was obtained from the American Philosophical Society, as recounted in the introduction, and intensive work on the site was begun in 1956.

VIII

Archeological Techniques

The archeologist must adopt and, if necessary, invent the method of excavation best calculated to produce the results he desires, given the conditions of a particular site. The Marlborough site required other techniques than those conventionally employed, for instance, in excavating prehistoric American Indian sites. Moreover, because the Marlborough excavations constituted a limited exploratory survey, the grid system used customarily in colonial-site archeology was not appropriate here, and a different system had to be substituted. It was decided in 1956 to begin, as in 1954, at obvious points of visible evidence and to follow to their limits the footings of walls and buildings as they were encountered, rather than to remove all of the disturbed soil within a limited area. By itself this was a simple process, but to record accurately what was found by this method and relate the features to each other required the use mainly of an alidade and a stadia rod. Only to a limited extent were some exploratory trenches dug and careful observations made of the color and density of soil, so as to detect features such as wooden house foundations, postholes, and trash pits. Once located, such evidence had to be approached meticulously with a shaving or slicing technique, again taking careful note of soil changes in profile.

All this required the establishment of an accurate baseline and a number of control points by means of alidade and stadia-rod measurements. Then eight points for triangulation purposes in the form of iron pipes were established at intervals along the south side of the highway, east of its turn at the clump of trees, on the basis of which the accompanying maps were plotted. The full extent of the excavations is not shown in detail on these maps, particularly in connection with the walls and structures. The walls, for example, were exposed in trenches 5 feet wide. Similar trenches were dug around the house foundations as evidence of them was revealed.

IX

Wall System

DESCRIPTIONS OF EXCAVATIONS

On April 2, 1956, the junction point of the three walls found in the 1954 test was reexcavated. The bottom layer of horizontally placed stones 1.8-1.9 feet wide was found in situ, while most of the vertical stones from the second course had been broken or knocked off by repeated plowing. Construction of the highway had completely removed a section of the wall. The corner of the two brick walls was revealed to have been superimposed on the northernmost foundation block of the stone wall, thus indicating that the stone wall preceded the building of the brick ones. The upper stone block that had been removed to make room for this brick corner still lay a few feet to the east where it had been cast aside in the 18th century. This part of the stone wall, together with its continuation beyond the highway to the creek, was designated Wall A (figs. 21 and 24).

Exposure of the brick wall running westward from Wall A (designated Wall A-I) disclosed broken gaps in the brickwork, the gaps ranging from 1.8 to 3 feet in length, and the intervening stretches of intact wall, from 7.33 to 8 feet. Eight-foot spacings are normal for the settings of modern wooden fence posts, as such a fence south of the highway illustrated. It is assumed, therefore, that, following the destruction of the exposed part of the brick wall, a wooden fence was built along the same line, requiring the removal of bricks to permit the setting of fence posts (fig. 26).

Wall A-I intersected the modern highway at an acute angle, disappeared thereunder and reappeared beyond. South of the clump of trees it abutted another wall of different construction which ran continuously in the same direction for 28 feet. Because of their manner of construction, the two walls at their point of juncture were not integrated and, hence, probably were constructed at different times. The 28-foot section later proved to be the south wall of the mansion, designated as B. (This wall will be considered when that structure is described, as will another section that continued for less than 4 feet to the point where a 12-foot modern driveway crossed over it.)

To the west of the driveway another wall (B-I), still in line with Wall A-I, extended toward the “gutt.” Of this only one brick course remained, a brick and a half thick. About midway in its length were slight indications that the wall footings had been expanded for a short distance, as though for a gate; however, the crumbled condition of the brick and mortar fragments made this inference uncertain.

Near the edge of the “gutt,” 146 feet from the southwest corner of the Structure B main foundation, Wall B-I terminated in an oblique-angled corner, the other side of which was designated Wall B-II. This wall ran 384 feet in a southwesterly direction under trees and beneath a boathouse along the “gutt,” ending at the back of Potomac Creek. It was constructed of rough blocks of the fossil-imbedded marl that underlies Marlborough and crops out along the Potomac shore. Walls A, A-I, B-I, and B-II, together with the creek bank, form an enclosure measuring a little over two acres.

Returning to the point of beginning excavation, the brick wall which is extended north from stone wall A (designated as Wall A-II) was followed for a distance of 175 feet. Like Wall A-I, it was a brick and a half thick (a row of headers lying beside a row of stretchers), and was represented for a distance of 36 feet by two courses. Beyond this point for another 30 feet, a shift in the contour of the land, allowing deeper plowing in relation to the original height of the wall, had caused the second course of bricks to be knocked off. From there on, only occasional clusters of bricks remained, the evidence of the wall consisting otherwise of a thin layer of mortar and brick.

Wall A-II terminated in a corner. The other side of the corner was of the same construction and ran westerly at right angles for a total distance of 264.5 feet, passing beneath the highway (north of the turn) and stopping against the southeast corner of a structure designated E. Extending south from Structure E was an 84-foot wall (Wall E) a brick and a half thick, laid this time in Flemish bond (header-stretcher-header) in several courses.

Another east-west wall, of which only remnants were found, joined Wall E and its southern terminus. Six feet west of Wall E this fragmentary wall widened from three to four bricks in thickness in what appeared to be the foundation of a wide gate, with a heavy iron hinge-pintle in situ; beyond this it disappeared in a jumble of brickbats.

Upon completion of the wall excavations, a return was made to Wall A, where a visible feature had been observed, although not investigated. This feature was a three-sided, westward projection from Wall A, similarly built of Aquia-type stone, forming with Wall A a long, narrow enclosure. The southern east-west course of this structure meets Wall A approximately 62 feet north of the creek-side terminus of Wall A and extends 59 feet to the west. The north-south course runs 100 feet to its junction with the northern east-west segment. The latter segment is only 55 feet long, so the enclosure is not quite symmetrical. No excavations were made here. However, in line with the north cross wall of the enclosure, trenches were dug at four intervals in a futile effort to locate evidence of a boundary wall in the present orchard lying to the east of the road to the creek.

SIGNIFICANT ARTIFACTS ASSOCIATED WITH WALLS

In addition to the artifacts listed above numerous others were excavated from the trenches, although few of these have archeological value for purposes of analyzing the structures. Only the finds accompanied by depth and provenience data are significant in evaluating these structures, and in the case of the gateway few are helpful to any degree. The fragmentary bottle seal found there matches exactly a whole seal that occurs on a wine bottle described in a subsequent section. That seal is dated 1737, and thus this seal must have been similarly dated. Its presence near the lowest level suggests that the wall was in construction at the time the seal was deposited. Bottles were used for a long time, however, so the seal may have reached its final resting place years later than 1737. The Indian celt no doubt fell from the topsoil while the trench in which the wall was built was being excavated. The swingletree gear next to it probably was left there during the construction. The colter, although it appears to be of early 18th-century origin, may have been in use late in the 18th century after the wall had been removed. Since the colter is badly bent, it may have struck the top of the underground wall foundation, and, having been torn off from the plow, perhaps was left on the bricks where it fell.

John Mercer commented with exasperation in his Land Book about the unresolved discrepancies between the Buckner survey of 1691 and the missing Gregg survey of 1707 (p. 14). There are as many disparities between Buckner’s plat and the plat resulting from the Savage survey of 1731. In the latter a new row of lots is added along the western boundary, pushing the Buckner lots eastward. Where in the Buckner plat the lots and streets in the lower part of the town west of George Andrews’ lots turn westerly 1° from the indicated main axis of the town, paralleling the 30-pole fourth course of the town bounds which runs to the creek’s edge, the Savage map shows no such change. Yet Savage, in describing the courses of the survey in a written note on the plat, shows that he followed the original bounds. He does note a 4°, 10-pole error in the course along Potomac Creek, “which difference gives several Lots more than was in the old survey making one Row of Lots more than was contained therein each containing two thirds of an Acre.” This was doubtless a contrivance designed to reconcile the Gregg and Buckner surveys and also to benefit John Mercer.

In any case, it is clear that the plats themselves are both unreliable and inaccurate. What was actual was shown in the archeological survey of 1956 with its record of boundary walls and at least one street. An attempt has been made in figure 14 to give scale to the Buckner survey by superimposing the archeological map over it. There, Wall B-II, if extended north for 111 feet beyond its length of 384 feet to equal the 30 poles (495 feet) of the fourth course, would exactly touch the southwest corner of lot 21 where the fourth course began. But, in spite of this congruence, the other features of the plat are distorted and disagree with the slightly northwest-southeast basic orientation of the street and wall system. The simplest explanation might be that the layout was made on the basis of the 1707 Gregg survey. Since it was following the second Act for Ports of 1705 that the town achieved what little growth it made prior to Mercer’s occupancy, it is probable that the town’s orientation was made according to this survey.

Whether or not this is the case, the road to the creek side was fundamental to the town, and probably was built early in its history and maintained after the town itself was abandoned. We know from archeological evidence that Wall A antedates the brick walls that were connected with it. Further evaluation of the wall system in relation to the entire site will be made later. It may be concluded for now that Wall A and the road beside it represent the main axis of the town as it was laid out before Mercer’s arrival, that the stone walls were built before that event, that Wall B-II follows the fourth course somewhat according to Buckner’s plat, and that the brick walls may date as late as 1750, as some of the associated artifacts suggest.

X

Mansion Foundation

(Structure B)

DESCRIPTION OF EXCAVATIONS

With the exception of Wall A, the protruding bit of brickwork near the clump of trees (where Highway 621 makes its turn to the southeast) was the only evidence remaining above ground in 1956 of Marlborough’s past grandeur. Designated Structure B, it was plainly the remains of a cellar foundation, which the tangled thicket of vines and trees adjacent to it tended to confirm. Since its location corresponded with the initially estimated position of the courthouse, it seemed possible that the foundation might have survived from that structure.

Excavation of Structure B began accidentally when the excavators began following the westward course of Wall A-I, as described in the preceding section on the “Wall System.” Wall A-I abutted, but did not mesh with, the corner of two foundation walls, one of which ran northward and the other continued on for 28 feet in the same direction as Wall A-I. The brickwork in the 28-foot stretch of Wall A-I was laid in a step-back, buttress-type construction. At the bottom course the wall was 2.65 feet thick, diminishing upward for five successive courses to a minimum of 1.5 feet. A wall running northward—the east foundation wall—was exposed for 16 feet from the point of its junction with Wall A-I until it disappeared under the highway. It was found to have the same buttress-type construction. There was no evidence of a cellar within the area enclosed by the foundation walls south of the highway.

Excavation of the east foundation wall was resumed north of the highway, but here no buttressing was found, with evidence of a cellar visible instead. This evidence consisted of a curious complex of features, comprising remnants of two parallel cross walls only 4.5 feet apart with a brick pavement between 4.8 feet below the surface. The east wall and the cross walls had flush surfaces. The northerly cross wall was tied into the brickwork of the east wall, showing that it was built integrally with the foundation. The northerly cross wall had been knocked down, however, to within five courses on the floor level. The pavement was fitted against it.

The southerly cross wall was not tied into the brickwork of the east wall, and the pavement had been torn up next to it. Thus it was evident that this wall had been erected subsequent to the building of the foundation, that it had shortened the cellar by 4.5 feet, and that the cellar extended southward to a point beneath the highway where it was impossible to excavate. Documentary evidence to confirm this alteration will be shown below (p. 91).

Extending 12.5 feet north of the original cross wall was another cellarless section, with step-back buttressing again featuring the foundation wall. Another paved cellar was in evidence north of this, extending for 26 feet, with a final 14.25-foot cellarless portion as far as the north wall of the structure. The interior of the cellar, to the extent that inviolate trees and shrubs made it possible to determine, was filled with brickbats and debris, large portions of which were removed. Evidence, however, of construction of cross walls and of floor treatment remained concealed.

Extending westward from the cedar trees, beyond the projected 28-foot length of the north wall, was a short section of brick wall foundation, the outer surface of which was faced with slabs of red sandstone and dressed on the top with a cyma-reversa molding. The tops of the slabs were rough, but each had slots and channels for receiving iron tie bars (ill. 3) that were still in place. This wall was inset four inches to the south of the alignment of the main north foundation wall.

The northwest corner of this additional structure was hidden under the highway. Even now, however, the discerning eye can pick up the contour of a wall running parallel with the west foundation wall under the blacktop pavement. For a brief distance, between the point where the road swings eastward from it and the private driveway covers it again, excavation exposed this wall. Designated Wall C, it was 22 inches thick, entirely of brick, with no evidence remaining of red sandstone on the outside. The exterior surface was 9.5 feet beyond the west foundation wall.

At the southwest corner of the foundation, evidence matching that at the northwest corner was found. Here, again inset 4 inches from the line of the main south foundation wall, were to be seen the tops of red-sandstone slabs like those found at the north end (fig. 36), in this case with one tie rod still in place. The driveway obscured the point to which the corner of this extending structure could presumably be projected. Subsequent construction against the sandstone slabs had covered their surfaces with a rubble of brick and mortar that appeared to be the foundation for masonry steps (fig. 35). Projecting out from the southwest corner of the foundation was a rectangular red-sandstone block which appeared to be the corner of these superimposed steps. Although situated under the driveway, it was apparent by projection that Wall B-I joined the southwest corner of Wall C. It will be demonstrated from surviving records that Wall C, with its connecting sections, was the foundation of a full-length veranda.

The belief which persisted for a time that Structure B might have been the courthouse was dispelled by documentary evidence showing that it was John Mercer’s mansion.