ARLINGTON HOUSE Historic Structures Report Phase II

This first portion of this report will trace the presumed structural history of the Arlington House, employing historical information on file as well as data discovered through fabric investigation. It is hoped that continued investigations in both the fabric of the house as well as document repositories will constantly improve the detail and accuracy of this sketch.

Due to limitations of time and limited documentary evidence some features or artifacts discovered during the fabric investigation will necessarily remain unexplained. With this caveat in mind, this overview of the structural history should be read in overall terms, avoiding the pitfall of dwelling too much on any specific. In general, it is felt that enough is now known about the history of the house's development to make the conjectures of this report with neither hesitation nor remorse.

The tract of land on which the Arlington House was built was purchased in 1778 by George Washington Parke Custis' father, John Parke Custis. This tract comprised about 3000 acres and was purchased from the Alexander family. John Custis moved his family into an existing frame house near the Potomac in 1779, which they called "Abingdon." [1]

Following the death of his grandmother, Martha Washington, in 1802, George Washington Parke Custis moved from Mount Vernon to the Arlington property, which he called "Mount Washington." He brought the possessions acquired from the estate of his grandmother to a four-room brick cottage near the river in the summer of 1802. [2]

It has been maintained that Custis began construction of Arlington House to provide a drier environment for his Washington relics. Supposedly the many items he had brought to his cottage near the river were rapidly deteriorating through attacks by dampness and rats. The new house on the hill was begun in 1802. [3]

English architect George Hadfield's obituary in 1826 stated that he was the designer of the Arlington House, and there is every reason to believe it. Hadfield came to the United States in 1795 to serve as superintendent for the construction of the Capitol Building in Washington, but was dismissed in 1798 after difficulties with the commission overseeing the work. After his dismissal, Hadfield remained in Washington and took scattered private commissions.

George Hadfield was a young man and was a purveyor of the then modern style — the Classical Revival. He had been a brilliant student at the Royal Academy of Arts and was privileged to win the Gold Medal and a three-year travel scholarship to Italy. While in Italy Hadfield spent time drawing restorations of a number of ancient temples. [4]

Being young, and having been trained in the most current architectural taste, he was in a difficult position to find the kind of patron who would appreciate his designs and encourage the avant-garde nature of his work.

George Washington Parke Custis was also a young man at the time, and had been brought up in what must have been one of the more sophisticated atmospheres in America. He had lived all during his childhood with George and Martha Washington, and much of this time fell during the years of the first presidency. While his foster grandfather was certainly a conservative influence on the boy, life in New York City and Philadelphia, and many of the individuals whom he came into contact with in the presidential household, provided the boy with a diverse cultural education. G. W. P. Custis' liberal tastes for the arts were undoubtedly formed early in life and later blossomed in his interests in painting and creative writing.

It is likely, therefore, that the meeting of Hadfield and Custis provided just the sort of association that each desired. Hadfield was very much in need of influential private clients of some means to commission his work; indications are that he was in particularly poor circumstances around the turn of the nineteenth century. Custis had inherited a large tract of land just across from the new capital city and was in need of an imposing family seat, placed high above the floodplain, in which to live and exhibit his relics from the Washington estate. Because of his interest in the arts and his own particular upbringing, it is reasonable to assume that Custis also wanted a house in the most modern style. It so happened that the classical style was exactly suited to the prominent site. Hadfield designed the massive portico in the Colossal Doric order like that at the Temple of Hera Argiva (Hera II) at Paestum, Italy, (see photograph #88) which had been built by Greek colonists circa 450 BC. It is also believed that Hadfield was influenced in his design for Arlington House by the Temple of Hephaestus (or Theseum) at Athens, Greece, also built circa 450 BC (see photograph #89). Not only was the chosen site ideal for Custis' intended family seat, but it was ideal as well as an exhibit of the architect's work. Hadfield remained in Washington and designed a number of public and private buildings before his untimely death in 1826. [5]

Prior to fabric investigation in 1980-81, it was believed that Custis had simply wanted a large house, hired Hadfield, and built the house between 1802 and 1819 as we see it today.

The north wing at Arlington House was to be the site of much of the structural stabilization work necessary for continued public visitation, and it was decided that fabric investigation would be carried out at the same time. The purpose of the investigation was to establish the appearance of the north wing in 1861.

A summation of the 1980-81 findings indicates that the part of the house now known as the north wing (minus the loggia) was constructed as a two-story house independent of all plans for the present house. Land surveys of the four-room brick cottage initially inhabited by Custis near the river show no house on the hills to the west.

G.W.P. Custis built the small two-story house on the hill overlooking Washington in 1802 soon after he realized the danger to the Washington relics from the dampness and rats in his cottage near the river. This small house was later modified and became the north wing of Arlington House.

The north wing has undergone extensive structural change during its history. Evidence of the two-story structure survives in the fabric of the north wing located beneath layers of plaster. The assumption of a two-story structure is based on the presence of the ledger wall believed to have once carried the joists of the first floor along the west wall of the winter kitchen and the wine cellar. Also, the archeological investigation of 1980 disclosed the fact that the original level of the land around the north wing was somewhat below its present level and that the present elevation was achieved with dirt moved during the construction of the main block. The level of the cellar floor is just below what the ground level would have been circa 1802.

The archeological investigations of 1980 uncovered impressive foundation remains in the present north wing cellar (see archeological investigation, Pousson). One of the principal features is the remains of a 1'—6" wide foundation wall (see photograph #5) running east-west, located in the approximate center of the north wing cellar floor. Investigations have shown that this foundation wall may have been built at the same time as the foundations of the west wall of the winter kitchen, but there is no bonding between them. The east end of this wall was cut off during the War Department installation of the utility trench, destroying any evidence of historic treatment. The remnant foundation of a narrower wall (13") was discovered lying perpendicular to the east-west foundation and running from the latter to the south wall of the wine cellar. This wall bonds into the east-west foundation but not into the foundation of the original south wall. There are also a number of brick features adjacent to the east-west foundation wall. These include two parallel foundation wall remnants, 1'—8" apart, each measuring 2'—8" long and 13" wide. These two short foundation walls abut the east-west primary wall near its west end. In examining the brickwork it was seen that the primary wall was built before the two secondary ones and that there was no attempt to make a tie between the lesser and greater walls The short walls also do not go as deep into the ground as the main wall, and the mortar in the small walls is very different from that in the major one. The purpose of these two small foundation walls remains unknown; as some charred remains were found on a brick-paved area between them, it is thought they may have had some supporting function for a fireplace or oven. A possible indication of the time lapse between the building date of the principal wall and the smaller perpendicular walls is given by the build-up of a crust on the main wall, also located on the wall where the smaller walls intersect it.

Adjacent to these two small foundation walls already discussed, on the north side of the main foundation wall, are two perpendicular short walls, also built against the main foundation wall but not at the same time, and unconnected to it structurally. The mortar used in this brickwork is an entirely different type from mortar used in the other walls. It resembles portland cement in hardness and texture, but may date to the historic period. [6] These foundations each measured approximately 1'—6" x 2'0" (width by depth) and were located some 6'—6" apart. A similar foundation was located to the south of the main foundation wall, 6 inches to the west of the easternmost 1'—6" x 2'—0" foundation to the north of the principal foundation. It is speculated that these small foundation pieces may have supported a fireplace above, but there is no further evidence to support the theory.

The principal remnant foundations described above indicate that the first floor of Mr. Custis' early house was divided into three rooms. The largest room occupied half of the first floor, measuring approximately 18'—4" x 21', and may have had a large cooking fireplace with a 6'—6" firebox and a small adjacent warming oven over the westernmost perpendicular foundations. The next largest room would have been located in the southeast portion of the house and measured approximately 12'—0" x 18'—4". The smallest room would have been in the southwest portion and measured approximately 7'—10" x 18'—4"; all first and second floor rooms would have had approximately 9'—0" ceilings. There could have been a fireplace as well in the southeast room if its west side was supported by the 13" perpendicular walls which divided the south part of the cellar.

No closed openings attributable to the two-story building were located during the investigation of the north wing. In all likelihood the former openings are largely incorporated into the existing ones. It is probable that both B02/D1 and B03/D1 were exterior doors to the first floor which were later cut down further into the foundation walls to serve the basement. From the rebuilding that has taken place in the basement windows, it is possible that the existing windows were once deeper than at present.

As the north wing now stands, part of the original first floor space is in the basement and part is exposed upstairs. Evidence in the existing first floor (the inner hall (north wing), Custis chamber, and Custis guest chamber) indicates that much of the original first floor was decorated in dark red paint on brown coat plaster. Some of this same scheme can still be seen on the east and south walls of the wine cellar. It can be assumed that the north and southeast rooms of the first floor were done in the red plaster. No evidence remains, however, for the walls of the southwest room except a very hard and rough grey plaster showing its natural color.

The wide foundation wall bisecting the north wing, the remains of which were found in the basement in 1980, evidently continued up through the second story of the original house. Evidence for its bonding into the west wall of the inner hall (north wing) is inconclusive.

The floor joists of the second floor of the original house rested on a ledger wall, which was located at an elevation approximately 4'—0" above the present floor. This ledger wall extended around the interior perimeter of the original two-story house. A line of broken headers located along the west walls of the school and sewing room and the inner hall (north wing) at 3'—1-1/2" (centerline) above the present floor showed that the ledger wall was once present and bonded into the principal bearing walls on the exterior. This ledger wall, cut down to 2'—5" above the present floor, still exists below the chair rail in the school and sewing room, the inner hall (north wing), and the Custis chamber closet and under 107/W1 in the Custis guest chamber. The rubble seen behind the plaster layers between the top of the chair rail and the level of the original second floor is the original interior of the wall, exposed when the ledger wall was reduced. Wall interiors were often little more than rubble as seen in the northwest corner of the school and sewing room.

Very little is known about room configuration on the second floor of the building. While plaster layers from this initial period of the house's history survive in the inner hall (north wing) the Custis chamber, and the Custis guest chamber, there is no trace of them in the school and sewing room. Much of what we do know comes from the chimney breast investigated in the inner hall (north wing).

The chimney breast shows the trace of the wall which formerly bisected the original house, the foundations for which were uncovered in the winter kitchen during the 1980 archeological investigation. Three joist pockets define the level of the original second floor. To the north of the bisecting wall on the second floor there was a room with a chair rail with cream-and-green wallpaper above it. The original finish below the chair rail has been lost. To the south of the bisecting wall was a room papered with red-and-cream leaf-patterned wallpaper. This wallpaper is also found all along the west wall of the inner hall (north wing) south of the chimney breast and on the cheek wall. It reappears as well on the cheek wall of the Custis guest chamber and along the east walls of the Custis guest chamber and the Custis chamber to 106/W1. This could indicate that this second floor room covered one-half of the house space. This red-and-cream wallpaper seems to have followed a former decorative scheme. It was applied over a dark-colored plaster in portions of the Custis guest chamber and covers the patch (in hard grey plaster which may be the hydraulic cement) of a former stud wall intersection at a point lying in the inner hall (north wing) midway between 105/D1 and 105/W1.

Little more can be speculated about the nature of the second story without further investigation. Much information was lost in the major changes which converted the north wing into a one-story building with a cellar.

No indication has been found of interior stairs in the two-story structure. Whatever may have been located on the south wall disappeared when the wall was removed. The mysterious flue in the chimney breast of the inner hall (north wing) may have served a fireplace in the northwest corner of the southwest room on the original first floor. The interior surface of the flue is plastered right out to the face of the chimney breast, suggesting that the flue originally curved through the brick bisecting wall and into a corner fireplace. The original house may have had a large central chimney on the principal interior wall possibly carrying flues from two fireplaces on each floor, and an additional chimney on the west wall carrying one flue only. The two chimneys may have co-existed at the same time.

Exterior openings on the second floor appear to have been located in the same places as the present openings. The original openings may have been widened and located lower in the walls. A thorough investigation of possible wall openings was not attempted at this time.

A feature that could be projected for the exterior of the original two-story house is a possible two-story wood porch on the west side. The presence of the small rafter pockets on the east wall of the pantry, and the regularly spaced post holes found under the brick floor of the cellar stairway hall, suggest the possibility of a makeshift wood porch of a temporary nature. Two brick footings found below grade to the north may have supported exterior stairs.

Custis was married to Mary Lee Fitzhugh in July of 1804; soon afterward he renamed his estate "Arlington." [7] It is most likely that the impending marriage had influenced him to obtain plans for his family seat.

The investigations of 1980-81 indicate that Custis probably did not have the Hadfield plans for the Arlington House when the north wing was built circa 1802, but he probably did have them before the building of the south wing in 1804.

It is felt that Cornelia Lee was describing the interior of Mrs. R.E. Lee's morning room (room 115) in the south wing when she wrote in April 13, 1804, to Mrs. Richard Bland Lee that "The house will be a very handsome building when completed. The room we were in was 24 feet square and 18 high . . . ." [8] While her measurements were not highly accurate, the actual measurements being approximately 21' x 24' with 14' ceilings, Mrs. R. E. Lee's morning room is about the only room in either wing configuration, actual or conjectural, that would have been near what she described. It is, therefore, reasonably certain that the south wing was near completion by spring of 1804 and that the plans of the whole house were available at that time for Cornelia's inspection, or at least were at a point where Custis could give an impressive verbal account of what they would be. It is thought that the south wing was built as planned and that probably no major change has occurred since its completion. However, no fabric investigation has been performed to verify this conclusion.

As has been stated, it is thought that Custis had the overall plans for the Arlington House in hand by the beginning of construction of the south wing. At the very least he knew that there would be a major center section measuring approximately 58'—6" north to south and placed the south wing accordingly. It is probable that the size of the north wing in itself dictated the size of its twin, and the overall proportions of the entire mansion. It can be speculated that Custis had Hadfield visit the site and, in interests of economy, it was decided to employ the existing building as the north wing of the main house. While it would be involved to make the necessary interior changes in the north wing to have it conform to the south, it would have been easier to convert the north wing into a compatible part of Hadfield's design rather than build an entirely new wing.

It is probable that Custis had the whole building plan formulated prior to the construction of the south wing. It was established in 1980 that the south wing's north exterior wall was a temporary partition from 1804 till the construction of the main block. The north exterior wall was probably of wood. Looking down from the north end of the south wing attic, onto the lath side of the great arch it can be seen that the south wing cheek walls, which extend out 1 foot from the east and west bearing walls and show 2'—0" on the east and west faces, were built as cheek walls intentionally when the south wing was built. The first 1'—6" from the north end is covered with stucco as it was considered part of the outside of the house, and the remaining approximately 6 inches, now seen as brick, formed the strip on either cheek face where the temporary stud wall abutted. When the main block was constructed circa 1817-19, the temporary stud wall was removed and the great arch was constructed between the east and west faces of the cheeks. The arch was built of split-wood lath and plaster on a wood frame. This arch, a major decorative feature of the house (see photograph #63) was necessary as a transition between the south wing and the main block.

Investigation of the analogous cheek walls in the north wing showed an entirely different sequence of events. Examination of the brick of the cheek wall east and west faces in the inner hall (north wing) and the Custis guest chamber clearly shows that the original south wall of this wing was broken through when the north wing became a part of the completed house. Part of the south wall was removed and perhaps the brick was reused. This seems to be a further indication that the original north wing was not intended to be a part of the whole house as later constructed, or at least that plans for the house had not been formulated by 1802.

Our first indication that the history of the north wing was not as simple as had been assumed came during the stabilization of the north wing roof in late 1979. Investigation in the attic here uncovered a number of pieces of interior wood trim, some in a greenish-blue paint and some in a reddish-brown paint grained in probable imitation of mahogany, which had been variously reused to stabilize ends of studs or stiffen plaster lath. As portions of chair rail found appeared to match that extant in the school and sewing room, and as some backband was identical to that on the interior casing of 104/D1, it was assumed that this trim was once part of a decorating scheme in the north wing. Concurrent discovery of the broken end of a chimney flue on the west side of the garret added to the accruing evidence that the north wing had undergone significant change during its history. At this point recommendations were made that a fabric investigation be undertaken and the compliance process was begun.

Note was also taken in the north wing garret of changes to the roof during the life of the building. As is recorded in the Existing Conditions section, Chapter IV, the rafters once had been unseated from their bearing plates and raised. This action was necessary during the change of the roof from one with splayed eaves to one with a single angle of pitch from ridge pole to fascia board. Remains of an outrider for the splayed eaves were noted at the end of the former south hip face, the rafters and partial sheathing for which are still extant in the north wing attic (see photograph #69). It was also noticed that the structure of the south wing roof differed substantially from that of the north wing by having a more reasoned structural system. By the use of the two king post trusses at the hip apices, the south wing had been provided with a much stronger structure, and one that would have been more typically used by a designer trained in the architectural profession. On the other hand, the north wing had a system that was only marginally strong enough for its purpose. This contrast between the two roofs further reinforces the theory of Hadfield's involvement with the design of the south wing, but not the original north wing. A disturbance of the ceiling joists near the center of the north wing attic also gave further evidence to the presence of a large central chimney in the two-story north wing.

North wing fabric investigation resulted in one major conclusion: the structure was built originally as a self-contained, two-story building; it was then changed to a one-story wing of the projected complete house, and, at a later date, its floor plan was again changed. In terms of chronology, it may be that the north wing was changed from a two-story house to a one-story wing with cellar during or after the construction of the south wing. By this time (circa 1804), the Hadfield plan was in force and the north wing may have been changed from the utilitarian two-story house to the plan Hadfield had devised for it as the north wing of the projected mansion. After the south wing was completed, Custis would have had more space and may not have needed all the rooms of the two-story house which he had initially inhabited.

After the classic revival windows were installed in the north wing, the wings should have resembled each other closely. One of the major differences between the north and the south wings was the temporary wall in the north end of the south wing. From this time (circa 1804) until the building of the main block in 1817-19, there may have been some kind of very temporary connection between the two wings as there was at the United States Capitol Building across the river, also in the midst of construction at this time. There is, however, no evidence to support such a theory at this writing. It is known that both wings were stuccoed at this time and had hip roofs with probable splayed eaves.

The fabric investigation of 1980-81 uncovered very little information about the first decorative scheme and floor plan of the north wing in the period immediately following the possible 1804 remodeling. An exterior opening which may have existed at the time, other than the four arched windows and the cellar windows below them, was 104/D1, as it still is outfitted with the early backband. It is likely that the two wings were as symmetrical as possible on the exterior; it can be assumed that they were as they are today on the east facade, except that the north wing has two more cellar windows than the south. As the loggias were possibly not built until after the main block was constructed, the west facades were probably very plain with no windows and only two nearly symmetrically-placed doors on each wing, 104/D1 and 105/D1 on the north, and 115/D2 and 117/D2 on the south. On the west side of the basement, however, there may have been only one opening on the south wing (B10/D2), but probably two were on the north (B02/D1 and B03/D1). The north face of the north wing was like the south face of the south wing except for the cellar window. The failure to excavate those areas in B10A and B10B resulted in the inability to make the two wings totally symmetrical.

The initial floor plan of the north wing after 1804 is not at all clear. The questions surrounding this are further complicated by the position of the fireplaces and flues. It would be logical to assume that the large central chimney was removed completely when the north wing was transformed into a one-story wing. However, plaster layering on the chimney breast in the inner hall (north wing) clearly shows that at least the wide brick bisecting wall of the original two-story house may not have disappeared until the installation of the present, or second, north wing floor plan and decorative scheme circa 1820. Due to the changes in the level of the floors during the first change in the north wing, the fireplaces of the central chimney were evidently rebuilt to correspond to the new floor. While this operation would seem complicated, the lowering of each fireplace 5 feet on the chimney block was certainly possible and easier than rebuilding the entire chimney block.

The transformation of the original two-story house into the one-story north wing involved another major change. In addition to the reworking of the central chimney block, large arched windows had to be opened through the walls. While they incorporated some existing openings, there was still major destruction involved. This fact is partially supported by the archeological find of a large amount of brick and plaster rubble just to the east of the north wing. From the nature of this deposit and its covering, the archeologist determined that this brick rubble and plaster fragments had to have been deposited before the excavation for the main block.

The floor level was changed by lowering the second floor joists 5 feet. It is presumed that ledger wall joist support was abandoned, and the new joists set into pockets in the bearing walls to the east and west.

As near as can be established from the present evidence, there was a major room in the new north wing to the north of the fireplace block. From the evidence left by the chair rail and the ceiling cornice remains, it can be seen that the perimeter of the north room ran from the north edge of 106/W1 around to the north edge of the chimney breast of the inner hall (north wing). Evidently the portion of bisecting wall between the central chimney block and 106/W1 had been removed to allow for that central window. It also appears that the north half of the chimney breast in the inner hall (north wing) was enclosed and became a closet. This north room was provided with a chair rail capping the ledger wall, which had been lowered about 1' 6", hence the presence of the broken headers along the west wall uncovered in 1981. From the trim remaining on the interior of 104/D1, it can be stated with some assurance that the greenish-blue trim found in the garret belonged to this room.

The other half of the north wing was evidently another large room having a fireplace on the north side back-to-back with the one in the north room. From the spot where the chair rail began in the chimney breast, it can be speculated that there was a closet also in that location. The room was provided with a chair rail as well, attached against the brick wall with railing blocks. It is believed that the reddish-brown chair rail found in the attic was used in this room as it goes well with the simulated wood-grain wallpaper found on the portion of the wall below the chair rail channel. It is not known what finish was used above the chair rail in the south room, or on the walls above or below the chair rail in the north room.

No evidence has been found to indicate that the first floor of north wing was anything more than a two-room section after the wing's restructuring, with the possibility of some closets. While we cannot be absolutely certain of the date of this change, it is likely that it occurred circa 1804 to make it compatible with the new south wing.

The Custis family lived for approximately fifteen years in the wings. While the cellar of the main block was said to have been dug early, no building commenced until circa 1817.

We are hypothesizing that Custis had Hadfield's plans in hand by 1804. There is a possibility that Hadfield designed the wings in 1804 and the rest of the house later, or even changed an initial design. As 1804 is an early date in America for a Greek Revival house design, there may well be those who would challenge the accepted idea that the house sprang fully designed from Hadfield's workshop in 1804. As has been pointed out, however, the young architect was fresh from England and the continent and would probably have had the very latest style at his disposal. Also, the presence of the stucco on the 1804 exteriors and the provision of a temporary wall in the north end of the south wing point strongly to the consideration of the final overall design seen today. There is no way at present to place an absolute date on Hadfield's plans.

It was not long after the completion of the main block (circa 1819) that the north wing underwent substantial change. The reason for the change can be seen by examining the plans of the house as it now stands. It will be seen that the doorway connecting the main block to the north wing was placed in the north bearing wall of the main block at the only available place, between the chimney blocks. If the north wing was arranged as presumed from circa 1804-1819, it meant that from the main block one entered the north wing south room directly and passed through it to reach the north room. If these rooms were used as bedrooms as presumed, the circa 1804 plan of the north wing was inconvenient. Perhaps after the loggia was built, there was still only one actual connection between the north wing and main block, and this was 105/D2. This being the case, it is easy to see how the presence of the main block generated a need to change and adapt the plan of the north wing. A natural question to ask at this point, however, is why did Hadfield not plan the north wing circa 1804 to be compatible with the plan of the main block, if all was designed circa 1804? The answer may be simply that the family preferred to have the two large rooms at the time (if there were indeed two rooms only) in the absence of the main block, and may well have adapted the north wing circa 1804, knowing that it would be changed again once the main block was completed.

The circa 1820 change to the north wing may have included the removal of the brick bisecting wall or walls and the central fireplace block. Evidence of their removal at this time may be indicated found on the face of the chimney breast in the inner hall (north wing) where the area against which the bisecting wall abutted, and to which it appears bonded near the top, seems to be covered with one series of plaster only.

The north wing was believed to be redesigned circa 1820 to provide more than the two or three former rooms and to provide easy access to these rooms from the main block. In order to save space, the present partition wall separating the inner hall (north wing) from the Custis guest chamber at 107/D1 was placed as close to 105/D2 as possible. A new fireplace block was constructed and placed to serve two of the rooms, creating a room to the north, the school and sewing room, which balanced in size the south room of the south wing, the office and study. It is likely that the fireplace block and dividing wall between the winter kitchen and the wine cellar were both built at the same time without foundations, their base bricks having been placed only a few inches below grade. Following the building of the fireplace block and a new chimney (which can be seen in the garret to have been constructed of used brick, probably from the central fireplace block), the current partitions in use today were believed constructed. In the case of the wall between the school and sewing room and the inner hall (north wing) and the walls of the Custis chamber closet, the partitions were built right over existing chair rails. No explanation has been found of the use of the short doors at 104/D2, 104/D3, 104/D4, 107/D1, and 107/D2. They were reused; perhaps they were from the north wing when it was a two-story house. It is likely that 105/W1 was added into the wall at this time to light the inner hall (north wing), and it is possible that the Custis guest chamber was heated by a stove. It will be noted here that all the partition walls of the north wing have a single series of plaster except those in the Custis chamber, which have an additional skim coat against the south wall of the school and sewing room, which has been extensively patched. It is suggested that very little has changed in the present arrangement of the north wing since 1820, and that its current plan may be of the circa 1860 historic period.

The north wing loggia does not bond into either the north wing or the main block, and may have been built before or after completion of the main block. Being thoroughly classical in style and feeling, these loggias are felt to have been an important part of the Hadfield design. Visually they serve to tie the wings strongly to the main block of the house, giving a forceful composition where it was needed in a welding of three major elements built at disparate times. They existed in their purest classical state during the years following their construction as open arcades like those on north Italian villas of the Renaissance. Such loggias were found often on the villas of Palladio where they could serve as weather-proof connectors to utilitarian farm buildings or, as at the Arlington House, to living areas in the wings of the principal house.

It is not known why the Custis family was moved to enclose the loggias, but they needed convenient access to the main block and more usable space. It is not clear, for example, how the winter kitchen (room B02) was reached prior to the creation of the pantry (room 101), but it is likely that the movement of food from that facility to the new dining room (room 109) in the main block may have been found inconvenient. Before the building of the main block dining probably took place in the north wing in one of the principal rooms and was reached directly by an interior stairway, as evidenced by a framing box which still exists in the ceiling of the winter kitchen. If the only way to reach the winter kitchen from the main block dining room (room 109) after its construction was to descend from the north stair hall by the stairway in 110A and pass through B04, B04/D1 to the outside air before reaching the winter kitchen at B02/D1, then it is understandable why a more direct route was needed.

Once the loggias were enclosed, however, it was believed necessary to add the small north and south connecting flankers, thereby permitting one to pass directly from the main block stair halls (north stair hall, south stair hall) into each loggia, and providing an alternate route in each case for one to enter either wing. It is not known at what date the loggias were enclosed, but it can be assumed that they remained open for a few years after their construction, circa 1820. The first indication, however, that the north wing loggia may have been enclosed comes in 1837 in a letter [9] where Robert E. Lee speaks of pipe being laid at the Arlington House. If Lee was writing about the installation of the bathroom and water closet, it follows that the loggia would have been enclosed by that time. Due to the presence of the sawn lath, rather than split lath, in the interior finishing of the loggia, a later date for the enclosing seems likely. Evidence that more work than just the bathroom enclosing was taking place at the time the loggia was enclosed is found in another letter, [10] also of 1837 but three months later, in which Robert E. Lee states that "Arlington must now look beautiful, and you and mother will be engaged with your improvements...." The fact that three months have passed is indicative of more work than just the bathroom enclosing may be taking place. Use of the plural "improvements", its emphasis in being underlined, and the fact that the whole house "must now look beautiful," also indicates that the work was prolonged and must have had a significant effect on the entire house.

A small indication of the early appearance of the loggias was given by a piece of thin cavetto molding found in place near the end of a ceiling joist in the overhang of the north wing loggia shed roof. This piece, protected for many years by the plain boxed eave which presently characterizes the loggia roofs, formerly finished an exposed joist end as a decorative dentil, thereby reinforcing the concept of the loggias as strong classical elements of the house. The piece had been painted with a mixture of paint and sand in an attempt to give an appearance of stone.

Perhaps, in the years prior to its enclosure, the north wing loggia had no floor at the level of the first floor rooms, but was open with a dirt or brick floor at the basement level with stairs leading to the first floor doors. It is likely that this was the case, because it represents the traditional classical solution and because of the presence of the window B01/W4, which would have provided some significant light to B05.

An explanation for the difference in floor levels between the first floor levels of the north wing and the rest of the house is not now known. This circumstance was most likely not planned but was the consequence of a probable miscalculation on the part of the surveyor or builder when establishing the elevations of the first floor of the north and south wings. Such a mistake is understandable if the south wing was being built at the same time or before the north wing was being converted from a two-story structure to its present one-story situation. When the main block was built, its first floor elevation was chosen to be the same as that of the south wing, and a step was provided between the main block and the higher north wing.

Arlington House was pictured in a lithograph (see photograph #90) in the first edition of William Elliot's The Washington Guide in 1826. The lithograph is stylized, but does give an overall view of the completed structure. While it shows no chimneys or other details which would be of the greatest interest today, it does show the parapet roofs on the wings. It is likely that these parapets were not added until after the main block was completed as they do abut its north and south walls. While showing the actual eaves of the low-pitched hip roofs, the parapets continued the faces of the wings to a point where most of the roof was concealed. These parapets were effective in contributing another strong classical element to the house, and remained in place at least until 1853 when they were pictured in a watercolor by Benson Lossing for Harper's Weekly (see photograph #91). They also appeared prominently in a rough sketch of the house done in 1843 by Markie Williams, although they are not shown to any particular advantage (see photograph #92). Historic photographs of 1864 (see photographs #87 and #93) show that the parapets were removed and that the roofs no longer had splayed eaves. While it is known that the very earliest roof of the north wing featured splayed eaves (see photograph #69), it is not known when they disappeared. As it seems likely that these parapets would have worked well with the splayed eaves, it may be assumed that the roofs retained splayed eaves through the epoch of parapets at the Arlington House. The parapets probably caused leaks in the roof, and the parapets were removed and the roofs changed to a more conventional design once the house fell under the influence of Robert E. Lee following Custis' death in 1857. Lee, being an engineer and a practical man, may well have preferred a tight roof over the stylistic considerations of the Greek Revival. It is also likely that the money situation at the Arlington House dictated the simplest of solutions to probable roof problems.

From the historic photographs, and 1859 insurance application, there is evidence that the wing roofs were of a simple tar and gravel composition.

It is certain that the house was roofed with split cedar shingles when built. Remains of these shingles have been found in the wing attics, and their nailing pattern is extant on the old sheathing of the south hip face preserved under the north wing roof. While it would seem that these shingles would have been simply replaced when bad, it could be that the weathered shingles were coated with tar and gravel to extend their life-span.

The wing roof shingles were replaced with slate in 1885 (from Superintendent's Report for September 1929) to match that of the main block which had been slated in 1858. [11] The War Department placed balustrades on the wing roofs during restoration; they were not a successful simulation of those in the historic drawings, in addition to not being true to the mandated historic period. The balustrades failed to visually continue the faces of the wings vertically, perhaps because of the absence of the splayed eaves. Due to leaks, and the discovery of further documentation rendering them inaccurate, these balustrades were removed in 1960. The wing roofs were then outfitted with tile shingles which simulated wood shingles. The present terpolymer and pebble roof was installed in 1974.

The rough texture of the south hip on the south wing as seen in photograph #86 appears to confirm the insurance policy description. The current roofs present a texture that is too smooth to truly simulate the historic condition. The interpretation of historic wing roofs could be made more accurate when the present roofs need replacement.

Arlington House as it has evolved differs somewhat from Hadfield's original design. The enclosing of the loggias and the necessary addition of the two symmetrically-placed connecting flankers significantly changed the Hadfield intention of giving a chaste classical facade to the utilitarian back side of the house. Rather than preserving the classical feel of the composition as first built, Custis' enclosures and additions converted the design to one more mundane and two-dimensional. While adding space to the house interior, the enclosing of the loggias destroyed those features as design elements. The addition of the connecting flankers not only subtracted two arches from the composition, but threw discordant elements into the original proportional system as devised by Hadfield.

The east, or principal, facade of the house underwent change in detail only. Early sketches of the house by Williams (photograph #92) and Lossing (photograph #91) indicated that the area within the portico tympanum was scored as ashlar. It is assumed that this area, like the other exterior walls at the Arlington House, was done in stucco. The stucco was then scored as though it were stone blocks - a technique used throughout — on the exterior. This stucco treatment was evidently removed prior to the historic photographs of 1861-1865, although it had stood in place since its creation circa 1820 and was still there in 1853 when Lossing made his watercolor (photograph #91). The stucco probably had to be removed from the tympanum because it was not as weather-resistant as the stucco on the remainder of the house. This stucco in the tympanum was the only stucco on the exterior on wood lath, a fact which probably explains its relatively short life. There is stucco on lath on the inside of the entablature, but that is not exposed directly to the weather. Whether or not there was similar stucco in the tympanum of the west facade is not known. The stucco and lath of the tympanum of the east facade was replaced by horizontal tongue-and-groove planks circa 1858.

From evidence encountered in exploratory excavations on the portico, it was discovered that another portico floor in conventional brick existed several feet below the present floor level. This former floor must have been in use prior to the installation of the present hexagonal payers, known to have occurred in 1851. [12] It was also written that the portico steps were being replaced at the same time, and this rebuilding would have been logical as the level of the portico floor was raised at that time to where it exists today. It is probable that the portico steps seen in the historic photographs of 1861 and 1864 are those which were built at this time.

The marbleization at Arlington House was not noted in the 1853 watercolor by B. Lossing but was shown in the 1861 photograph by McDowell of the portico steps. The fact of its not being mentioned is unusual as the marbleization technique, while popular during the nineteenth century, was not seen universally and was nowhere seen to such an extent on an exterior as at Arlington House. In terms of fashion, however, the marbleization might be best dated to circa 1850. There was a general fashion at the time for things "Italianate", and this creation of a marble-like classical villa may well have been part of this movement to link the classical revival styles more closely with their country of origin. The claim that Custis personally did the marbleization work is at best fanciful; he was an old man by circa 1850 and was probably not capable of such a task. It may well be, however, that the idea to marbleize the house was his own, and that he directed the operation. The historic photographs suggest that the marbleization may have been somewhat crudely done; the only real detail discernable in these vestiges of early photography are the strong, grand strokes, some of which appear erratic and fail to suggest real marble. Other detail, however, that was left unrecorded by the early camera equipment, may well have been a better representation.

Additional physical research is necessary to fully interpret the details of construction of Arlington House. One of the most significant discoveries of the 1980-81 fabric investigation in the north wing was the location of a lintel matching the profile of molding used under the arches between the family parlor and the family dining room and presumably under now-removed fan lights from these arches. This lintel was found nailed between the leadlined water tank above the bath and the water closet and the west wall of the north wing loggia. This evidence would appear to indicate that the water tank was installed after the fan lights, partitions, and doors were removed from the arches. It is possible, however, that the piece was nailed in place against the tank years after the tank was installed to replace an earlier one or that it was excess material left over from other construction. Written documentation for the installation of the bathroom in the north wing is meager. A letter in 1837 proposed laying of lead pipes. The written records are inconclusive for the date when the arches between the family parlor and family dining room were opened by removal of lath plaster, louvered doors, and fan lights. It would be contrary to accepted restoration practice to change the arches without absolute proof that their present condition is not accurate to the 1861 appearance.

There is also a question surrounding the plaster molding fragment found lodged in the attempted flue hole of the Custis guest chamber. This fragment, of an approximate length of 2 inches, is identical in profile to the exterior edge of the great arch against the north wall of Mrs. R. E. Lee's morning room. No pieces of this molding are missing on the arch and if found in the flue hole they must have come from some feature destroyed at the time the flue hole was made. As the Custis guest chamber had no source of heat after its creation, it is likely that it was heated by a stove. The fact that two holes had to be opened before the main chimney flue was found would seem to indicate, however, that the flue was not opened by Custis, but either after his death by the Lees or still later, during the army occupation, when those working on the house were naturally less familiar with the location of its features. A flue hole also exists just across the chimney flue above the fireplace in the family parlor.

One of the principal contributions of the Lees to the house was a modern heating system installed in 1855. [13] Its location was confirmed by archeology in 1980. The cast-iron furnace was located toward the center of the house in B06. When the equipment was dismantled by the army, several large pieces were left in the basement and presently rest in B08. There can be seen a flat ring with a lip on one side and an operable grate approximately 1" thick which probably rested directly on the brick foundations located under the floor of B06 (see photograph #17). According to the grooves on top of the ring, there would have been another section between it and a large ring which remains in B08 as well. Between these two extant pieces was another large ring which probably held the access door for stoking the system. There was yet another large ring between the last one mentioned and another extant ring of smaller radius. Above this last ring was another missing piece from which issued the smoke pipe. All these rings were fitted with grooves which caused the whole assembly to fit together tightly. This furnace, when together, would have formed a rounded and squat iron mass which had the capability to become very hot when stoked with coal. It is thought that this cast-iron unit was further surrounded by a brick mass in which the heat was captured and from which it rose by natural convection in ducts. As noted in the Existing Conditions Section, one duct rose vertically to the center hall (room 111), and continued up the north wall to the upstairs center hall (room 201). Another duct appears to have run off horizontally through the south wall of B06 where it fed a diffuser in the floor of the white parlor. This was likely the case as three of the floor joists are replaced to the west of the center beam in the middle of the floor system of the white parlor. The diffuser could not have been directly in the center of the white parlor due to the presence of this center beam; it was probably located just to the west of the center. The flue pipe from the furnace evidently passed through the north wall of B06, where there is an unfilled patch today, and to the flue of the fireplace in B05 located under the fireplace in the family dining room. No destructive investigation was carried out in either the center hall or upstairs hall; there is still a question as to whether the vertical duct traveled on the outside of the bearing wall, as is likely, or whether it was actually inside the wall. As the north wall of the center hall and upstairs hall is a solid brick bearing wall, the latter possibility is the least likely.

The white parlor, left by Custis with only the lath in place, was plastered and fitted out with architectural trim of the period by the Lees. It is reasonable to assume that the arch over 115/D1 was planned by Hadfield as it is in the brick of the bearing wall. It is also likely that the fine composition of the two fireplaces, the central arch, and the east and west niches (see photograph #60) was also from Hadfield's plan. It is known that the Lees purchased and installed the two Victorian marble mantelpieces, [14] that Robert E. Lee requested that the room be painted white, and that the doors be grained in imitation of walnut. [15]

It is documented that the Lees supplied the morning room (room 115) with a mantelpiece similar in type to those in the white parlor [16] The baseboard profiles in both rooms are very similar, if not the same, but the other architectural trim is unlike any in the rest of the house. This trim has been shown to have been attached with early-to-mid nineteenth-century nails, and reattached later with wire nails after floor replacement during the War Department restoration, so the room was definitely not finished by the army in the early twentieth century as some have claimed. Further fabric investigation is necessary to document the origin of this room's decorative treatment.

Obviously, many questions remain unanswered which govern the return of the house, and its full interpretation, to the period of Lee tenure. Continual research is needed to locate and assimilate documentary evidence in the site historic files to apply to the continuing restoration and interpretation of the house. There are also ten major repositories of Lee and Custis documents which are monitored by the site to record additional documentary evidence as it is added to these collections. Fabric research is important to support and explain the documentary evidence available.

It is essential that the physical fabric be carefully evaluated as much of the documentary evidence has been scattered and perhaps lost. It is important that the evidence in this architectural data section and the two volume historical and architectural data section of the historic structure report be used to guarantee the accurate restoration of Arlington House.