JAIC 1993, Volume 32, Number 2, Article 5 (pp. 141 to 152)
JAIC online
Journal of the American Institute for Conservation
JAIC 1993, Volume 32, Number 2, Article 5 (pp. 141 to 152)




The inherent difficulties in determining the original colors and finishes of heavily damaged and overpainted architectural surfaces have been understood for many years (Johnston and Feller 1967; Miller and Phillips 1976; Johnston-Feller and Bailie 1982). The often unorthodox use of media and colorants, the use of inexpensive or poor-quality materials, blanching and fading from ultraviolet exposure, “bleed-through” of media from layers of overpaint and decades in darkness under overpaint, all contribute to darkening, yellowing, alterations of tone, and sometimes total mechanical failure of the finishes. To address these adverse conditions, a multistep process of examination, analysis, and synthesis is critical to the definition and interpretation of architectural finishes.

Seven principal steps can generally be anticipated in order to determine, with a good degree of certainty, the original appearance of polychromy and finishes: (1) archival research that might provide information on the original design, aesthetic objectives, materials, and techniques of the finishes; (2) sampling and microscopical examination of full stratigraphies from every major element, followed by initial color matches to a standardized system, such as Munsell, or possibly chromometric quantification using spectrophotometry; (3) exposure of the original surfaces by mechanical or chemical means and initial matches to a standardized system; (4) qualitative and quantitative analyses of colorants and media to support identification of the intended color and gloss; (5) development of an initial palette based on these data; (6) probable adjustment of the colors of the initial palette; and (7) a mock-up (e.g., a model panel) of the final palette and finishes on a large wall area in situ under final lighting conditions. The process must be repeated for the glazes and scumbles. All seven steps were required in the U.S. Custom House.


Hundreds of mounted cross sections from throughout the building were examined using reflected light microscopy, despite the inherent difficulty posed by the very thin, friable, and discontinuous state of the degraded and overpainted original finishes (fig. 4).4 A consistent stratigraphy was evident: (1) the plaster support of the walls; (2) an assumed size on the plaster; (3) a buff-toned primer; (4) the base paints (restricted to five colors); and (5) at least 10 layers of overpaint.

Fig. 4. Cross section of the thinly applied stratum of light gray paint, indicated by arrow, on the buff-toned primer. (65x)

The five base colors of Garnsey's palette initially were identified descriptively as mauve, apple green, cream yellow, light gray-green, and dark gray-green. However, when matched to Munsell colors and then made up as an initial palette, a distinctly “muddy” tonality was perceived in all five base colors. Alteration of the original base colors and known surface finishes, such as glazes, was posited, prompting additional research, examination, and analysis.


Several documents written by Garnsey provided insights into his technical expertise and highly developed aesthetic vision.5 In his proposal entitled “United States Custom House, New York City, Description of Decorative Painting,” dated May 5, 1911, Garnsey wrote:

The great vaulted ceiling [of the Main Hall] is to be covered with canvas and the principal ornamental designs and borders decorating it are to be executed in flat relief [e.g., pastiglia]. Then they are to be gilded with gold leaf and lacquered and toned to an antique patina; after which the backgrounds of the ornament are to be emphasized with color, bringing the painted decorations into harmony with the marble columns and panels beneath.

Garnsey also described the ceilings of the elevator lobbies and the gallery:

The ceilings of the Elevator Halls (e.g., lobbies) and the Corridor [of the Gallery] are to be decorated in harmony with the vaulted ceiling [e.g., of the Main Hall], on a background of gray-blue or violet-gray. The wall decorations in these spaces are to be painted on canvas, mounted on the wall, and are to correspond in scale and style with the character of the vaulted ceiling.

In a letter of March 19, 1912, to Gilbert, Garnsey proposed the decoration of the main stairwells:

The surfaces to be decorated include the plaster surfaces of the ceilings, walls and staircases of the northeast, from the First and Seventh Floors inclusive. The painting is to include the soffit of the beam bounding the stairwell of each floor level.

The general scheme is to carry the color of the marble wainscot through the ceiling beams and styles and rails of the wall panels, modifying this by gray-yellow tones in certain fillets and flat channels around the panels. The ceiling panels to be in warm grey, and the wall panels a gray-violet stippled over with gray, to the same general effect as the wall panels of the Gallery of the Third Floor … .

The wall panels … to be covered with canvas before painting and to be finished after painting with a flat varnish protective coat.

These documents indicate that Garnsey intended the architectural polychromy to chromatically reference the distinctive tones of the multicolored marbles and breccias used throughout the building. This hypothesis was confirmed in part when the unusual mauve and apple-green base tones were exposed on the wall panels of the gallery. They matched the mauve and apple-green base tones that had survived intact on the vaults of the main hall. These mauve and green paints do, indeed, correspond to the colors of the unusual variegated marble panels of the main hall.

Thus, it was posited that the predominant color used throughout the Custom House, initially identified as a “light gray-green” in cross sections and exposures, originally was a true “stone” gray that referenced the gray-toned marble wainscot used throughout the building—exactly as Garnsey had described. This hypothesis was supported when analyses of the colorants identified only lead white and a small amount of carbonaceous material, the constituents of a light gray. The surviving original ornament of the ceilings of the third-floor elevator lobbies were also executed in two shades of “cool” gray, providing further support.

However, initial over-reliance on archival documents delayed recognition of the complexity of some of the finishes throughout the Custom House, especially those of the blue panels of the main hall and the blue groin vaults of the gallery. In 1990, visual examination of the blue panels, studies of cross sections, FTIR analysis, and analysis by fluorescent dye staining indicated a perplexing stratigraphy: (1) lead white preparation on the canvas; (2) pebble-textured blue base-paint composed of ultramarine and lead white paints; (3) application of one, perhaps two, tones of tinted glazes; and (4) a blanched protein-carbohydrate stratum overall.6

A 1915 description by Garnsey of a very similar decorative scheme in the library of the City Art Museum, St. Louis (now the St. Louis Art Museum) suggested the same materials and techniques of execution that characterize the decoration of the main hall of the U.S. Custom House (Garnsey 1915). Garnsey's 1915 description seemed especially applicable to the blue panels, in which the blanched stratum apparently was a pastelike material, intended to function as a matte-textured protective coating that could be removed easily with soap and water as it became soiled in urban environments. Garnsey wrote of his decorative finishes in St. Louis:

The ornament, which was executed in a material resembling the Italian “gesso” was applied upon a canvas ground and delicately modelled before it hardened [e.g., pastiglia]. It was then gilded [with] gold leaf and its background was solidly painted in blue relieved by accents of violet, green and red in the cartouches, wreaths and masks. Finally, the entire wall surface was glazed with transparent colors and scumbled with semi-opaque pigment in order to suggest a little of the patina of the Renaissance chapel walls.

I have attempted to combine certain essentials of good decorative painting in this commission—to achieve adequate design as a whole, to clothe that design in sonorous color and to add an agreeable texture and surface finish to the wall.

It is not an attempt at story-telling, nor are the ideas and personifications intended to serve as texts for literary exercises. The figures are merely incidental to the design as a whole and are relatively of no more importance than any of the rest of the ornament of which they are a part.

The design is my own and while it is reminiscent of certain Italian Renaissance schemes, I know of none that might be called its legitimate ancestor. The Borgia apartments in the Vatican are decorated with compositions of figures and ornament by Pinturicchio, painted in about the color key I have employed. Much of his architectural detail has been slightly modelled in relief and little dots in relief are carried over skies, trees, etc., and then gilded. All of which carried the richness of gilding through the entire design; therefore, I should say that here was the prototype of my scheme, although my design is quite different from Pinturicchio's.

The processes of execution are as follows: The walls were covered with canvas, then primed with white lead and linseed oil. Upon this ground the relief patterns were laid with stencils, after which portions were modelled into slightly high relief, to catch the glint of light falling from above. Then the whole was painted in orange-red upon which ground the gold-size was applied and then all the ornament was gilded with gold leaf. The background was then picked out in ultramarine blue. Then the surfaces were glazed with transparent colors to lower the brilliancy of the gold and to add a deeper richness to the under wall colors. Further pickings out in rich color followed and then the whole surface was scumbled over with semi-opaque glazes rubbed into and around the ornament. The figures were painted on a thin canvas and cut to fit into niches and framings, then mounted on the wall with a mixture of white lead and varnish. Finally, all was again covered with a protective coat of paste which makes the surface flat and permits the cleaning of the walls by washing off and re-pasting when smoke and grime have unduly blackened the painting. I have taken this last into some account and I have not lowered the general tone as much as I should have done in a city where the atmosphere is less opaque.


From 1991 to 92, analysis and reanalysis of several samples of surviving original finishes and heavily overpainted finishes were conducted.

Analysis of samples of the blue panels of the main hall indicated a technique of execution that differs from Garnsey's 1915 description of apparently similar finishes in St. Louis. In summary, the protein-carbohydrate material appears to have been worked into the blue base paint wet into wet rather than applied as a continuous surface coating. Thus, in a strictly sequential order, the toned glazes—composed of an oil medium and a resin medium—were the final application, but they were also worked wet into wet into the protein-carbohydrate material with a patterned tool (fig. 5).

Fig. 5. Cross section from a blue panel of the main hall, revealing the multimedia composition of these finishes. Arrow indicates Gamsey's decorative stratum. (65x)

Two researchers concluded independently that some finishes were executed as wet into wet multimedia composites, enhanced with tinted glazes in many areas.7 From these data, an intended marbleized effect was posited on some elements. Other finishes almost certainly were executed in emulsion paints, while others were standard oil paints. Analyses by fluorescent dye stains provided the following data:

SAMPLE 1. Light gray-green paint, south stairwell. The entire layer stained positively for oil with rhodamine B (RHOB) uniformly; the bottom half reacted slightly with eosin isothiocyanate (EITC), a general protein stain. When stained with antimony pentachloride, the upper half did not react, while the lower half of the layer strongly reacted, indicating a resinous component in that portion of the film. This pattern of staining strongly indicates a phase separation on drying in the binding materials, such as an emulsion paint system. There was an unusual staining with antimony pentachloride on the surface of the layer, indicating a resinous component such as a varnish or glaze layer.

SAMPLE 2. Mauve paint, wall panel, south stairwell. This paint stained positively for oil, carbohydrate, and protein (using RHOB, triphenyl tetrazolium chloride, and EITC) and must be considered an emulsion type of paint.

SAMPLE 3. Dark gray-green paint, pilaster, south hall. The original surface appears to have presented a marbleized effect because there are three layers applied almost wet into wet. Layers 2 and 4 are colored the same and stain positively for oil and a resinous component, while layer 3, the actual dark gray-green, appeared to be only oil.

SAMPLE 4. Mauve paint, rosette on pilaster, second floor. This paint stained positively only for oil.

SAMPLE 5. Light gray-green paint, outer border, main stairwell. This paint stained for oil, carbohydrates, and trace amounts of proteins, indicating an emulsion. Slight staining with antimony pentachloride, observed near the surface, indicates a varnish or glaze.

Copyright 1993 American Institute for Conservation of Historic and Artistic Works