Why Paint Analysis?©

By Susan L. Buck

    During the study of almost any old building one of the questions most frequently asked by both amateurs (homeowners) and professionals is What color was it originally? Traditionally the answer was supplied by uncovering the first paint layer through careful scraping, sanding, or chemical stripping. But, these methods carried the risk of providing incomplete information. To the naked eye a white primer layer could not always be distinguished from an early white finish layer, and varnish layers were sometimes overlooked entirely.
    Over the last decade new analytical methods adapted from the study of easel paintings and furniture have been applied to architectural paint samples. One of the most important methods involves examining the full paint sequence in cross-section under a microscope at magnifications up to 500X, using both reflected visible light and ultraviolet (UV) light sources. This type of cross-section microscopy analysis can help to distinguish between priming coats and finish coats; it can identify the presence of resinous sealing coats and varnishes (based on characteristic, identifiable autofluorescence colors in the UV); it can reveal evidence of aging and weathering; and it allows the characterization of the binding media in the various layers through the use of biological fluorescent stains.
    Cross-section information about paint and finish layer sequences, in combination with pigment identification (primarily through polarized light microscopy analysis), provides the basis for assessing the original color, surface texture, gloss level, thickness, stability and composition of historic paints. The cross-sections also provide information about the dispersion of pigments in a given layer, whether the surface of a paint layer has darkened or faded due to weathering and accumulation of dirt and debris, and how thickly and evenly a layer was applied. All of these are factors in trying to decipher the original appearance of a paint layer, and this is a topic worthy of a dissertation.

Illustration #1 Wood trim sample from the stairhall of the Gropius House contains wood substrate and three generations of white paint. In visible light it is difficult to distinguish the three white layers, but under reflected ultraviolet light the three layers appear markedly different. The first layer above the wood is brightly autofluorescent, the second generation is only faintly autofluorescent, and the uppermost layer is completely nonfluorescent in the UV. Samples were photographed with Kodak T-Max 100 at 125X. All photographs taken by Susan Buck. Click on the image to see a larger picture with more details.

    But, beyond deciphering color and appearance, the stratigraphy (or layer structure), of accumulated generations of paint and finish is a valuable source of information about alterations or additions to a structure. The best way to compare paint stratigraphies is to use photomicrographs of the cross-section samples. These are color prints photographed through a 35mm camera mounted to a UV-Visible light microscope and they allow direct comparisons of the coating histories surviving in different areas of a building. The prints can be shuffled like playing cards to line up comparable paint layers from room to room, to compare the paint histories adjacent to and on the ghost of an architectural element which has been long lost, and to identify subtle physical alterations to a space.
    In the best of circumstances, such as the direct collaboration between architectural historians and conservators, the sampling and microscopy skills of a paint analyst or paint conservator can help answer specific questions about removal, alterations, and installations of architectural elements based on comparative paint histories. In other words, the cross-sections can serve as form of paint archeology. Sometimes the paint layers can even be dated, such as when they are discovered as the earliest layer on architectural elements with known dates, or if they contain pigments only commercially available or synthesized after certain dates.


The Process
    After determining the goals of a paint analysis project and identifying the areas with the most potential to provide complete paint and finish chronologies, the paint analysis process generally begins with an on-site exploration using a micro-scalpel to make small investigatory excavations (about 1/4-inch in diameter). These excavations are then examined with a 30X monocular microscope to get a sense of the condition of the substrate (bright, fresh replacement wood and modern plasters are generally readily identifiable at this point) and the number of surviving layers.
    Samples are removed with a sharp scalpel and placed in labeled polyethylene bags. The samples should always contain a few wood fibers, or bits of plaster, brick or metal substrate, to ensure that all the layers, from the substrate at the bottom to the most recently applied coating at the top, are present. This method, adapted from the paint sampling process for easel paintings, means that samples are often as small as a pinhead, as long the layers can be removed intact. However, 20 to 30 generations of embrittled architectural paint on top of a friable substrate may require somewhat larger samples up to a 1/4-inch across to retain all the layers in proper alignment. Naturally, larger samples are taken from protected, unobtrusive locations whenever possible. The goal of sampling should be to obtain as much information as possible with the least amount of intrusion into historic building fabric. It is helpful to take several samples from the same area so that one sample can be cast for cross-section analysis and the other can be saved for other types of analytical procedures or for color matching.
    There are a variety of methods for permanently casting cross-section samples, but one easy and thrifty method of casting involves using mini ice cube trays for the molds. The small cube molds are first partially filled with polyester resin, then after curing these half-cubes form the bases for the cast samples. The half-cubes are labeled with a permanent fine point marker before the samples are set into them. This is a critical step as projects requiring hundreds of small cast samples can easily get confused or completely derailed if the sample cubes are not clearly identified. Once the samples are placed in the cubes, a second layer of casting resin is poured into the molds and the samples are left to cure overnight. After curing, the polyester resin cubes are generally ground with a belt sander to remove the excess resin, and then dry-polished with wet/dry papers and silica-embedded polishing cloths (with grits from 1500 to 12,000) to produce a flat, glassy surface on the face containing the sample.
    The polished, exposed samples can then be examined and photographed in cross-section at the magnifications that provide the most information about the full stratigraphy and about the individual layers of the most interest. It is helpful to examine the samples first in reflected visible light to get a sense of the color of each layer, but often it is the ultraviolet reflected light examination which provides the most revealing information. In fact, a paint layer history containing 24 generations of white paint can be almost impossible to sort out in visible light, but in the ultraviolet the variations in binding media and pigment composition will make it possible to identify specific paint generations based on subtle variations in autofluorescence colors. In addition, the use of biological fluorescent stains applied to the samples can also help to distinguish specific layers based on their reactions for the presence of organic components such as oils, carbohydrates, and proteins.
    This type of analysis was critical to understanding the paint layer sequences on the white-painted vertical redwood sheathing on the exterior of the 1938 Walter Gropius House in Lincoln, Massachusetts, as well as the white paints on the interior trim and clapboards used in the stairhall of the house. It also revealed that there were at least four generations of white paint of varying shades in the stairhall present on different trim elements. This assortment of whites is confusing to visitors and compromises the visual coherence of the two-story open stairhall.

Illustration #2 Gropius House wood trim stairhall sample. In ultraviolet light three white paint layers are easily distinguishable based on different autofluorescence characteristics. Photographed with Kodak T-Max 100 under reflected ultraviolet light (using a UV cube with 300 to 400 nanometers (nm.) excitation with a 420 nm. barrier filter) at 125X.
Advantages of Using Reflected Ultraviolet Light
    Paint samples have traditionally been examined solely under reflected visible light. But when viewed under this type of illumination, cross-sections which contain ground, paint and varnish may often be difficult to interpret, particularly because clear finish layers look uniformly brown or tan. It may be impossible using only visible light to distinguish between multiple varnish layers. Illumination with ultraviolet light provides considerably more information about the layers present in a sample because different organic, and some inorganic, materials autofluoresce (or glow) with characteristic colors.
    There are certain fluorescence colors which indicate the presence of specific types of materials. For example: shellac typically fluoresces orange (or yellow-orange) when exposed to ultraviolet light, while plant resin varnishes (typically amber, copal, sandarac and mastic) fluoresce bright white. Wax does not usually fluoresce; in fact, in the ultraviolet it tends to appear almost the same color as the polyester casting resin. In visible light wax appears as a somewhat translucent white layer. Paints and glaze layers which contain resins as part of the binding medium will also fluoresce under ultraviolet light at high magnifications. Other materials such as lead white, titanium white and hide glue also have a whitish or greenish autofluorescence. Modern synthetic coatings generally have lavender or bluish autofluorescence colors and appear markedly different from natural resin varnishes when examined under UV illumination.
    There are also visual indicators which show that a surface has aged, such as cracks which extend through finish layers, accumulations of dirt between layers, and sometimes a diminished fluorescence intensity, especially along the top edge of a surface which has been exposed to light and air for a long period of time. All of these subtle and not-so-subtle identifying characteristics cannot be discerned when only visible light is used for analysis.

Illustration #3 Colonial Williamsburg Peyton Randolph House. This sample from a second floor bedchamber reveals a thin red-brown paint layer trapped in the upper portion of the wood substrate, below a layer of white overpaint. Photographed with Kodak T-Max 100 under reflected visible light at 250X.



Illustration #4 Colonial Williamsburg Peyton Randolph House. In ultraviolet light an autofluorescent shellac sealant is revealed trapped in the wood substrate, below the first generation red-brown paint. Photographed with T-Max 100 under reflected ultraviolet light (using a UV cube with 300 to 400 nanometers (nm.) excitation with a 420 nm. barrier filter) at 250X.



Practical Applications of Cross-section Paint Analysis
    Cross-section microscopy paint analysis proved to be very effective in identifying a number of the early architectural fragments found in the basement of the Colonial Williamsburg CP. Armistead House, an 1892 Victorian house built on the site of a 1750 storehouse which was turned into a tavern in 1776. In an effort to identify the fragments, Colonial Williamsburg Architectural Historians Willie Graham and Mark R. Wenger sorted and grouped the fragments stylistically, but they hoped that paint analysis could provide a firmer connection between certain elements.
    Fortunately, a comparison of surviving paint stratigraphies revealed that an early sash, two architrave moldings, two window stops, several window pulley stiles, and an early baseboard were all painted with the same oilbound red-brown paint composed of iron oxide pigments (perhaps burnt sienna), red lead, charcoal black and calcium carbonate. More importantly, three elements the interior surface of the main door, one architrave and a window pulley stile were found to have the identical, weathered paint sequence containing seven generations of paint. This suggested that not only were these three elements all from the same building, but that they were likely all from the same room and had all been repainted each time the room was repainted.
    Cross section paint analysis was also key to identifying the paint sequences on trim elements and shingles of the 1788 dome of the 1779 Maryland State House. One critical sample provided a solid datable layer, as it was taken from on top of an area containing a signature dated 1864. A comparison of the paint sequences in this sample with the complete sequence of 29 paint generations found in a sample from an endpost block louver opening showed that the signature layer lined up with the 18th paint generation of the full 29 layers. This evidence suggested that 17 generations of paint were applied to the dome between 1788 to 1864, and that the dome was painted less frequently after 1864, with 12 generations applied from 1864 to the present. It was also possible to comparatively date other added elements based on comparisons with the sequences in these two important samples.
    Cross section microscopy can also reveal unusual craft practices. In the 1766 Ridout House in Annapolis, Maryland, samples taken from the original simulated raised plaster paneling in a first floor parlor and the best bedchamber on the second floor revealed that the original plaster was surprising coarse. Under 125X magnification, boulder-like sand particles project from the surface of the uppermost plaster layer. The first generation paint layers above the plaster in all the sampled areas were either comparatively translucent pigmented limewashes or distemper paints. This paint and plaster evidence suggests that these plaster walls were matte in appearance and noticeably roughly textured. This is a surprising discovery in a Georgian mansion house that was one of the more expensive and refined residences built in the city between 1764 to 1774.

Illustration #5 Coarse plaster finish coat used in three important rooms in the Ridout House. Signs of weathering and aging, such as dirt trapped between paint generations, cracks, and darkened surfaces, are easier to identify under reflected ultraviolet light. Photographed with Kodak T-Max 100 under reflected ultraviolet light (using a UV cube with 300 to 400 nanometers (nm.) excitation with a 420 nm. barrier filter) at 125X.



    Sometimes this type of cross-section microscopy analysis can reveal good craft practices, such as the use of shellac to seal a wood or plaster substrate prior to painting. Shellac is readily identifiable in the ultraviolet because of its orange autofluorescence color. The first clue that it is present as a sealant (not remnants of a first finish coat) is when it is discovered found trapped deep in the wood fibers but not present as a coherent layer on top of the wood. Or, when it shows up as a very thin, even layer on top of plaster. An ongoing study of nineteenth century painting practices in New England and New York State Shaker communities has shown that instead of shellac, the Shakers used a water soluble gum size (perhaps gum arabic or gum tragacanth) to size both wooden furniture and interior wood trim prior to painting. This was an especially relevant step as the gum size acted to seal the wood substrate and limit the penetration of the typically low viscosity paints made and used by the Shakers. This approach allowed the frugal Shakers to limit the use of the more expensive pigmented paints, as less was needed to produce an even, intensely colored paint layer above a sealed wood surface.

Illustration #6 Coarse plaster finish coat used in three important rooms in the Ridout House. The relative opacity of the six generations of paints is more readily distinguishable in visible light. Photographed with Kodak T-Max 100 under reflected visible light at 125X.





Other Analytical Methods
    The photomicrographs provide the means to identify the samples containing the most information about a given area, but this cross-section analysis technique cannot specifically identify organic and inorganic components in a paint or finish layer. Fluorescent stains (fluorochromes) for binding media components can tag oils, proteins and carbohydrates, but it is not possible with this method to find out what types of oils, carbohydrates or proteins are present When it is critical to identify a particular type of oil, resin, gum, starch, protein, or sugar material in a coating, a more complex, and more expensive analytical method called FTIR microspectroscopy can often provide that information. This type of analysis requires only a tiny amount of sample material and it produces a spectrum which can then be compared to the spectra of known materials, and can also be deciphered based on the location of specific peaks in the spectrum. FTIR analysis helped to identified the organic components in the first glossy, yellowish-tan, resinous layer on the Maryland State House dome shingles as a combination of shellac, copal, and linseed oil. This may have been applied both as a weatherproofing material and as a reflective first finish coat.
    Other analytical tools such as Scanning Electron Microscopy (SEM) with an elemental analysis function can identify specific elements present in a very tiny cross section or dispersed sample. The spectrum generated through this method provides an idea of the comparative amounts of the elements present in a given sample. For example, analysis of the first red-brown paint layer trapped in the wood fibers of many of the samples from the Colonial Williamsburg Walthoe Storehouse indicated the paint was composed of lead (red lead or white lead) and iron (iron oxide pigments), and comparatively small amounts of silica, aluminum and calcium. The silica, aluminum, and calcium may have been filler materials to bulk up the paint, or simply impurities in the paint. Larger samples can be analyzed using x-ray fluorescence (XRF) analysis to identify the inorganic components in a paint or varnish layer. In fact, painted architectural elements can be analyzed using this method if specific paint layers can be isolated.
    Reflected visible and ultraviolet light cross-section microscopy analysis can provide a tremendous amount of information about the evolution of a building at a comparatively reasonable cost. In fact, samples mailed in for analysis are often very revealing, as long as they contain a full, representative paint sequence, including substrate, and were taken from meaningful locations. The best paint studies often come from collaborations between architectural historians and paint analysts who jointly plot out the strategy for a project, and identify the best potential sampling locations based on what is already known about the building being studied. This can provide important comparisons of paint stratigraphies on original and replacement elements, and with luck, can also provide a means of dating paint sequences.
    While this is not new technology these analytical methods have been in use in paintings and furniture conservation for more than fifteen years it has been only comparatively recently adapted for use in the historic architecture field. The color photomicrographs generated through this type of cross-section microscopy provide visual references which document paint and finish histories and become part of the permanent conservation and archeological records of a building.

This article was written by Susan L. Buck and published in the Vernacular Architecture Newsletter Summer 1998, issue 76. This article is used by permission of the author and all rights are retained by the author.
Susan L. Buck is a Ph.D. candidate in the University of Delaware Ph.D. Program in Art Conservation Research. She is also a practicing conservator and paint analyst with Historic Paint and Architectural Services in Newton Centre, Massachusetts.

URL: http://resources.umwhisp.org/resources/slbuck.htm
Last modified: 1 July 1998
Send Comments/Questions To: Gary Stanton