CROSS-SECTION MICROSCOPY ANALYSIS REPORT

CROSS-SECTION MICROSCOPY ANALYSIS REPORT Gunston Hall, Mason Neck, Virginia First floor Bedchamber, Closet walls COLONIAL WILLIAMSBURG FOUNDATION WILLIAMSBURG, VIRGINIA FEBRUARY 2011 Written by: Kirsten E. Travers Graduate Fellow Winterthur / University of Delaware Program in Art Conservation Supervised by: Susan Buck, Ph.D. Conservator and Paint Analyst Williamsburg, Virginia and Edward Chappell Roberts Director of Architectural and Archaeological Research Colonial Williamsburg Foundation

Site: Gunston Hall, Mason Neck, Virginia Room: First floor bedchamber, interior of closet to right of fireplace Requested by: Edward Chappell, Roberts Director of Architectural and Archaeological Research Colonial Williamsburg Foundation Analyzed by: Kirsten Travers, Graduate Fellow Winterthur / University of Delaware Program in Art Conservation Gunston Hall, first floor bedchamber with replication verdigris finish (closet door on right) Consulted: Susan L. Buck, PhD. Conservator and Paint Analyst Date submitted: February 2011 Purpose: The goal of this project is to use cross-section microscopy techniques to obtain a better understanding of the 18th-century verdigris finish in the closet to the right side of the fireplace in the first-floor bedchamber at Gunston Hall. The bright green currently seen in this room is a replication finish based on earlier paint research conducted by Frank S. Welsh. Procedures: On October 14, 2010, Kirsten Travers, accompanied by Susan Buck (paint analyst), and Caroline Riley (Gunston Hall curator), collected two paint samples from the interior of the closet to the right of the fireplace in the first-floor bedchamber (front left room), at Gunston Hall. A monocular 30x microscope was used to examine the painted surfaces and determine the most appropriate areas for sampling. A microscalpel was used to remove the sample from the wood substrate (GHC 1), but the sample from the plaster (GHC 2) was found among debris on the floor of the closet, so its original location in unknown. The samples were labeled and stored in small Ziploc bags for transport. All samples were given the prefix GHC and numbered according to the order in which they were collected. In the laboratory, the samples were examined with a stereomicroscope under low power magnification (5x to 50x), to identify those that contained the most intact paint evidence and would therefore be the best candidates for cross-section microscopy. Uncast sample portions were retained for future examination and analysis. The best candidates were cast in resin cubes and sanded and polished to expose the cross-section surface for microscopic examination. The following samples were collected and cast for analysis: Sample # Location Samples cast Notes GHC 1 wood framing member of closet door (or, reverse of wainscot) GHC 1 GHC 1a GHC 1b wood with green (verdigris) layer and dark brown layer on top. wood with green (verdigris) layer. No brown layer seen. large green (verdigris) paint fragment with impression of wood fibers on both sides. GHC 2 painted plaster, collected from debris on closet floor GHC 2 plaster chunk (includes coarse and fine coat), with green (verdigris) layer, now black and dusty. 2

Once cast, the cross-section samples were examined and digitally photographed in reflected visible and ultraviolet light conditions at 20x to 400x magnifications. By comparing the resulting photomicrographs, finish generations could be interpreted based on physical characteristics such as color, texture, thickness, presence of dirt layers and extent of surface deterioration. Fluorochrome staining was also carried out on selected samples to characterize the types of binding media present (oils, carbohydrates, proteins). The most informative photomicrographs and their corresponding annotations, as well as comments from the author, are contained in the body of this report. All raw photomicrographs can be found in the Appendix. Results: The two samples from the closet contain a great deal of evidence from which more can be learned about the original green finish at Gunston Hall. This results indicate the finish was made with verdigris paints (Cu(C 2 H 3 O 2 ) 2 2Cu(OH) 2 ), and glazes (prepared by dissolving verdigris in heated pine resins). Due to the unstable nature of these copper-based pigments, the surface of these finishes have oxidized to a dark brown color over time, and now the interior of the closet appears almost black. This type of discoloration is characteristic for copper-based finishes (Kuhn 1986, 151). This phenomenon can be observed in the following photomicrographs, where the surfaces are very dark. In this report, the most relevant results are shown and interpreted. The paint layers in each photomicrograph have been annotated according to generation. For instance, a primer, paint layer, and varnish layer may represent one finish generation, and are all given the same number, but distinguished with letters a, b, c, d, etc., according to the order in which they were applied. All results are interpreted in the conclusion. 3

GHC 1: wood door frame from interior of closet. 2b. verdigris glaze II 2a. verdigris glaze I 1c. verdigris glaze 1b. verdigris paint 1a. gray base coat 2b 2a 1c 1b 1a wood substrate wood substrate shellac sealant shellac sealant GHC 1, visible light, 200x GHC 1, UV light, 200x The orange autofluorescent material in the wood cells strongly suggests that shellac was used to seal the wood surfaces before painting. A thin, coarsely-ground grayish base coat was then applied to the wood surface (1a). This thin gray base coat was also used on the plaster wall (see page 6). This is followed by a coarsely-ground bright green paint with green, yellow, and white particles in a green-colored matrix (1b). This layer has no autofluorescence, strongly suggestive of verdigris- based paint. This paint was then coated with a verdigris-based glaze (1c), which is dark brown in visible light, resulting from oxidation of the finish. This glaze would have produced a high gloss. In fact, this gloss was still discernible when examining the uncast sample portion. Sometime later, another verdigris glaze was re-applied in two coats (2a, 2b). In generation 2b, red, yellow and orange pigments were added to the glaze. Both of these glazes are brown in visible light, suggesting age and oxidation of the finish over time. This type of discoloration is characteristic for copper-based finishes. 4

GHC 2: painted plaster fragment from interior of closet. See next page for greater detail. verdigris finish plaster finish coat coarse plaster coat (with hair) GHC 2, visible light, 20x verdigris finish (no autofluorescence) autofluorescent sealant (possibly proteinaceous sizing) plaster finish coat coarse plaster coat (with hair) GHC 2, UV light, 20x In sample GHC 2 from the plaster wall, one can see the coarse plaster layer (containing lime, sand, and hair), followed by the final plaster layer that has a higher proportion of lime, no hair, and less aggregate than the coarse layer. The bright autofluorescence at the surface of the plaster suggests that some material, possibly an animal glue, was used to seal the surface before painting to decrease porosity. The verdigris finish is discussed in greater detail on the following page. 5

GHC 2: painted plaster fragment from interior of closet. dust crack dust 1c. verdigris glaze 1c. verdigris glaze large verdigris particle 1b. verdigris paint 1b. verdigris paint 1a. gray base coat 1a. gray base coat sized plaster substrate sized plaster substrate GHC 2, UV light, 100x GHC 2, UV light, 100x After sizing the plaster, a thin, coarsely-ground grayish base coat (1a) was applied. This appears to be the same as that used on the wood (sample GHC 1), and may contain verdigris, as suggested by its lack of autofluorescence. This is followed by a coarsely-ground green paint (1b), made with large green verdigris particles as well as some yellow and white pigments. This layer also has no autofluorescence, strongly suggestive of a verdigris-based paint. This paint was then glazed with a verdigris-based glaze (1c), which has no autofluorescence and is dark brown in visible light, resulting from oxidation of the finish. Deep cracks in the surface of this layer attest to its age and deterioration. This first generation verdigris finish on the plaster appears to be the same as the first generation verdigris finish found on the wood. A thick layer of dust coats the top of the sample. 6

Fluorochrome staining results GHC 2: painted plaster fragment from interior of closet. FITC + FITC + GHC 2, B2A, 40x. Before FITC stain for protein. GHC 2, B2A, 40x. After FITC. Sample GHC 2 was stained with FITC to determine the presence of proteins. A strong reaction (bright yellow-green fluorescence) was observed in the plaster, strongly suggesting that an animal glue-based material was used to size off the plaster surface before painting. 7

Fluorochrome staining results GHC 2: painted plaster fragment from interior of closet. Detail of verdigris finish layers. GHC 2, UV, 200x. Before TTC stain for carbohydrates. GHC 2, UV, 200x. After TTC. Sample GHC 2 was stained with TTC to determine the presence of carbohydrates (starches, gums). No reactions were observed (a dark reddish-brown color), in the finish layer or the plaster substrate. DCF + GHC 2, B2A filter, 200x. Before DCF for oils. GHC 2, B2A filter, 200x. After DCF. Sample GHC 2 was stained with DCF to determine the presence of lipids (oils). A positive reaction (yellow-green fluorescence) was observed in the finish layers and substrate, suggesting they contain an oil component, although the reaction in the plaster substrate could result from oils in the finish layer that have migrated into the plaster. 8

Polarized light microscopy results GHC 2: painted plaster fragment from interior of closet. Scraping from verdigris finish layer. carbon black carbon black fracture edge transparent green copper-based glaze transparent green copper-based glaze lead white GHC 2, transmitted plane polarized light, 1000x. First generation verdigris paint. lead white GHC 2, transmitted cross polarized light, 1000x. First generation verdigris paint. A sample of the verdigris paint layer (generation 1b), was collected with a clean scalpel blade from the uncast portion of GHC 2, dispersed on a glass slide, and mounted with Cargille meltmount (refractive index 1.66) for polarized light microscopy. The sample contains white lead (very small, rounded particles that are greenish in transmitted light), carbon black (shards of an opaque, isotropic black material), and yellow earth pigments were also observed (pictured on next page). All of these pigments are suspended in a transparent, brittle green glaze (note fracture edges). This material is isotropic in cross polarized light, strongly suggesting a copper-based glaze made with oil or resin. 9

Polarized light microscopy results GHC 2: painted plaster fragment from interior of closet. Scraping from verdigris finish layer. lead white lead white yellow earth yellow earth GHC 2, transmitted plane polarized light, 1000x. First generation verdigris paint. GHC 2, transmitted plane polarized light, 1000x. First generation verdigris paint. This portion of the dispersed pigment sample exhibits a yellow earth particle (amorphous, yellow in transmitted light, isotropic), not shown on the previous page. Particles of lead white are also included. Interestingly, no verdigris particles were seen, although they were readily apparent in the cross-section. 10

Conclusions: The two samples collected for analysis reveal a great deal of information about the original 18th-century verdigris finish in the first-floor bedchamber closet at Gunston Hall. The wood elements in the closet were sealed with a shellac prior to painting, while the plaster was sized with some type of proteinaceous material, possibly an animal glue, to make the plaster less absorbent. Afterwards, both wood and plaster surfaces received the same coarsely ground verdigris-based finish. This finish is composed of three layers, beginning with a thin gray base coat over which was applied a thick layer of bright green paint. This paint appears to contain verdigris, white lead, yellow earth pigments, and carbon black in a transparent green matrix, possibly a copper resinate in oil or resin. The green paint was then glazed with a resinous, copper-based green coating that would have enhanced the depth of color and lent a high gloss to the surface. No clear protective varnish was observed over any of the finishes. Now, this finish is cracked and has discolored to dark brown at the surface, characteristic for copper-based finishes. Fluorochrome staining suggests that this finish contains a lipidic (oil), component. An instrumental technique such as FTIR (Fourier-transform infrared spectroscopy), or GC-MS (gas chromatography-mass spectrometry), could be used to identify the binding media with greater accuracy. On the plaster, only the first-generation verdigris finish was extant (albeit covered with a thick layer of dust). However, the wood surfaces were re-glazed with a verdigris-based coating once after the first generation. This could represent touch ups to areas more prone to wear (the painted wood samples were taken just below eye level, where additional abrasion would be expected). References: Gettens, R., and G. Stout. 1942. Painting materials: a short encyclopedia. New York, Dover Publications, Inc. Eastaugh, N., et. al. 2008. Pigment Compendium: a dictionary and optical microscopy of historical pigments. Oxford, Butterworth-Heinemann. Kuhn, H. Verdigris and Copper Resinate, chapter 6 in Artists Pigments: a handbook of their history and characteristics, volume 2, Ashok Roy, ed., 1984. Oxford University Press, National Gallery of Art, Washington, D.C.: 131-158. 11

Appendix A: Procedures Sample Preparation: The samples were cast in mini-cubes of Extec Polyester Clear Resin (methyl methacrylate monomer), polymerized with the recommended amount of methyl ethyl ketone peroxide catalyst. The resin was allowed to cure for 24 hours under ambient light. After cure, the individual cubes were removed from the casting tray and sanded down using a rotary sander with grits ranging from 200 600 to expose the crosssection surface. The samples were then dry polished with silica-embedded Micro-mesh Inc. cloths with grits ranging from 1500 to 12,000, lending the final cross-section surface a glassy-smooth finish. Microscopy and Documentation: The cross-section samples were examined using a Nikon Eclipse 80i microscope equipped with an EXFO X-cite 120 fluorescence illumination system fiberoptic halogen light source. Samples were examined and photographed under visible and ultraviolet light conditions (330-380 nm), at 20 to 200x magnifications. Digital images were captured using a Spot Flex digital camera with Spot Advance (version 4.6) software. All images were recorded as 12.6 MB tiff files and stored on a hard drive in a folder titled Gunston Hall Closet on Susan Buck s laboratory computer. A separate set of images will be stored on the CWF digital database, accompanied by a digital version of the final report. Information Provided by Visible and Ultraviolet Light Microscopy: When examining paint cross-sections under reflected visible and ultraviolet light conditions, a number of physical characteristics can be observed to assist with the interpretation of a paint stratigraphy. These include the number and color of layers applied to a substrate, the thickness or surface texture of layers, and pigment particle size and distribution within the paint film. Relative time periods for coatings can sometimes be assigned at this stage: for instance, pre-industrial-era paints were hand ground, lending them a coarse, uneven surface texture with large pigment particles that vary in size and shape. By contrast, more modern, industrially-prepared paints have smoother, even surfaces and machine-ground pigment particles of a consistent size and shape. Furthermore, he presence of cracks, dirt layers, or biological growth between layers can indicate presentation surfaces and/or coatings that were left exposed for an extended period of time. Under UV light conditions, the presence and type of autofluorescence colors can distinguish sealants, clear coatings, and binding media, from darker dirt or paint layers within the stratigraphy. For instance, shellacs exhibit a distinct orange-colored autofluorescence, while natural resins (such as dammar and mastic), typically fluoresce a bright white color. Oil media tends to quench autofluorescence, while most modern, synthetic paint formulations (such as latex) exhibit no fluorescence at all. Some pigments, such as verdigris, madder, and zinc white, have distinct fluorescence characteristics, as well. UV light microscopy is critical to help distinguish otherwise identical layers often found in architectural samples- such as successive varnishes, or multiple layers of unpigmented (white) limewash. 12

Binding Media Analysis using Fluorochrome staining: Fluorochrome stains adapted from the biological sciences were used to characterize the paint binding media (oils, proteins, carbohydrates), in layers within the cross-section sample. The following stains were used in this analysis: 2,7 Dichlorofluorescein (DCF): 0.02% w/v in ethanol. Fluorescent labeling reagent for lipids, particularly drying oils. One drop of stain was applied to the surface of the sample, blotted immediately, and cover-slipped with mineral spirits. The reaction was observed with the B-2A filter (450-490 nm excitation, 520 nm barrier filter). This stain exhibits a yellow-green fluorescence where lipids are present. Triphenyl tetrazolium chloride (TTC): 1.0% w/v in ethanol. Labeling reagent for carbohydrates (gums, starches, cellulosic thickeners). One drop of stain was applied to the surface of the sample, blotted dry, and allowed to sit for approximately 45 seconds before cover-slipping, (must be allowed to react with atmospheric moisture for reaction to move forward). The reaction is observed under reflected UV light conditions (330-380 nm). A dark red-brown color is seen where carbohydrates are present. Fluorescein isothiocyanate (FITC): 0.02% w/v in anhydrous acetone. Fluorescent labeling reagent for proteins. One drop of stain was applied to the surface of the sample, blotted immediately, and coverslipped with mineral spirits. The reaction was observed using the B-2A filter cube (450-490 nm excitation, 520nm barrier filter). A positive reaction is a bright yellow-green fluorescence. Pigment Identification with Polarized Light Microscopy: To collect a pigment sample for polarized light microscopy (PLM), a surgical scalpel was used to collect a small scraping from a clean, representative area of paint. The blade was then pressed and pulled across a clean glass microscope slide, dispersing the pigment particles across the surface. The pigments were then permanently embedded under a cover slip using Cargille Meltmount (refractive index 1.66). The embedded pigments were then examined in cross and plane-polarized transmitted light with the Nikon Eclipse 80i microscope at 1000x magnification (using an oil immersion objective). The observed morphologies (size, shape, agglomeration, cleavage patterns), and optical properties (including color, refractive index, extinction), were compared to reference standards as well as literature sources before making final determinations. 13

Gunston Hall First Floor Bedchamber Closet Photomicrographs GHC 1 VIS 200x GHC 1 UV 200x GHC 1a VIS 200x GHC 1a UV 200x GHC 1b VIS 100x GHC 1b UV 100x

Gunston Hall First Floor Bedchamber Closet Photomicrographs GHC 2 VIS 20x GHC 2 UV 20x GHC 2 VIS 40x GHC 2 UV 40x GHC 2 VIS 100x GHC 2 UV 100x

Gunston Hall First Floor Bedchamber Closet Photomicrographs GHC 2 VIS 200x GHC 2 UV 200x GHC 2 UV b4ttc 200x GHC 2 UV TTC 200x GHC 2 B2A b4fitc 40x GHC 2 B2A FITC 40x

Gunston Hall First Floor Bedchamber Closet Photomicrographs GHC 2 B2A b4dcf 100x GHC 2 B2A DCF 100x GHC 2 verdigris BF 1000x GHC 2 verdigris DF 1000x GHC 2 verdigris2 BF 1000x GHC 2 verdigris2 DF 1000x