JAIC 1991, Volume 30, Number 2, Article 4 (pp. 163 to 177)
JAIC online
Journal of the American Institute for Conservation
JAIC 1991, Volume 30, Number 2, Article 4 (pp. 163 to 177)




THE ABOVE represents the general scheme of chemical and physical examination and chemical tests that can be applied to thick and microtomed cross sections made from paint samples. Some applications of this scheme and the results we achieved are examined below. Observations made on the cross section also are discussed, together with the analytical results. With the microtome, multiple cross sections can be generated from one paint sample, making it possible to use one sample for many tests.


A series of known pigments and binding media was applied to Masonite and aged for six months in the Conservation Analytical Laboratory's painting conservation studio using the environmental control present in the building (70 2F and 50 2% RH). By layering paints of known pigments and binding media on a Masonite board, layered cross sections were created. The test panels were used as controls to test the sensitivity and limitations of the analytical methods described above. These test samples were especially useful in testing and perfecting the sectioning system. Samples mounted in a large cylinder mold were used for thick sectioning, and samples mounted in BEEM molds were used for microtoming.

Sample A consists of: 1) Masonite sized with rabbit skin glue; 2) a ground layer of calcium carbonate and rabbit skin glue; 3) a first layer of Rowney tempera cadmium red in a protein and oil medium; and 4) an upper layer of lead white in an egg yolk medium.

Sample E consists of: 1) Masonite sized with rabbit skin glue; 2) a ground layer of lead white and rabbit skin glue; 3) a first layer of Rowney tempera ochre color in a protein and oil medium; and 4) an upper layer of zinc white with egg yolk medium.

A 15-micron microtomed cross section of sample E was prepared following the procedure described previously. This cross section was then examined using scanning electron microscopy and the EDS system. Elemental analyses confirmed the presence of the pigments. Despite the fact that the blade used for this particular cutting was not very sharp, the clear SEM image presented in figure 2 indicates that the unpolished microtomed surface is suitable for scanning electron microscopy.

Fig. 2. SEM image of 15-micron microtomed cross section of sample E. (1) a layer of zinc white with egg yolk medium; (2) a layer of Rowney tempera ochre color in a protein and oil medium; (3) a ground layer of lead white and rabbit skin glue

Another microtomed section from sample E was subjected to a fluorescence-staining test for protein, using 0.25% fluorescein isothiocyanate in ethanol. A third microtomed section then was subjected to the fluorescence staining test for oil using solution of 0.25% Rhodamine B in ethanol as the staining reagent. The staining results confirm the presence of the media applied aa described above. Under the stereomicroscope, a small portion of each layer was excavated and used as a sample for GLC identification of oils1 and the HPLC analysis of proteins.2 The ground and secondlayers were tested for protein by HPLC, and the middle layer was tested for drying oils by GLC. Figure 3 represents the chromatograms of each layer. Despite the small size of the sample retrieved from the thin-section, the methodology and instrumentation are sensitive enough for identification purposes.

Fig. 3. Chromatograms of each layer of sample E; each layer was retrieved manually under the microscope. (1) HPLC chromatogram of layer 1 in figure 2, Analysis of amino acids in egg yolk medium; the hydroxy-proline (hydroxyp) peak is from the rabbit skin glue ground layer that was retrieved by mistake; (2) GLC chromatogram of layer 2 in figure 2, analysis of fatty acids in Rowney tempera color; (3) HPLC chromatogram of layer 3 in figure 2, analysis of amino acids in rabbit skin glue.

The microtomed section from sample A was subjected to FTIR analysis by transmitted light through the microscope. The microtomed sections were thin enough to allow light to be transmitted through them. The transmission mode was chosen over reflection because existing spectra in the conservation literature are primarily transmission spectra and a large computerized data bank exists for transmission spectra; the reflection mode is still under investigation for its application in conservation (Baker et al. 1988). The Conservation Analytical Laboratory instrument consists of the Spectra-Tech IR-Plan microscope attached to a Mattson Cygnus 100 FTIR. The microscope can be operated in a visible light mode for viewing the sample, with the FTIR beam condensed to focus in on a very small spot. Spectra were collected for each paint layer in the cross section, and they confirm the presence of the material included in the preparation. These results are presented in figure 4. The FTIR spectra of the embedding material are routinely checked with the spectra from the samples to eliminate the errors caused by the contamination. The absorption bands at 1654 and 1542 cm−1 (Amide I and Amide II peaks, respectively) indicate the presence of protein (Meilunas et al. 1990), and the absorption bands at 1730 cm−1, 1235 cm−1, and 1165 cm−1 indicate drying oil.

Fig. 4a. SEM image of sample A cross section

Fig. 4b. Infrared spectrum of layer (1): Lead white in an egg yolk medium

Fig. 4c. Infrared spectrum of layer (2): Rowney tempera cadium red in a protein and oil medium

Fig. 4d. Infrared spectrum of layer (3): Calcium carbonate and rabbit skin glue ground


4.2.1 General Description of the Painting and its Condition

Thomas Dewing's America Receiving Nine Muses (1903) depicts 10 women dressed in turn-of-the-century ball gowns. America is shown in profile seated in a curule chair at the right of the painting. There are three standing figures to the left and six dancing figures on the right. Overall, light pastel colors such as pink, mauve, and gray are set in a greenish background. This work is of Dewing's last period (1900–1915). It is painted on the piano lid of the 100,000th Steinway piano, that was presented to the White House in 1905, during President Theodore Roosevelt's term of office. The piano case was very ornately gilded, designed by J. D. and J. H. Hunt. The case surface is a painted scrolled foliate border surrounding and connecting shields that represent the 13 original states. A more complete description of the painted surface and its conservation is presented elsewhere (Tsang et al. 1990).

The support for the painting is in good condition with no apparent warping, delamination, or other instability associated with the wood of which it is composed. The paint layers also are in very good condition, and it appears that the painting has never been cleaned. There are abrasions in the painting resulting from improper handling of the piano lid. The natural resin varnish layer is discolored and has marred the tonal and spatial quality of the painting. Removal and repair of the abrasions were the objective of the conservation work.

4.2.2 Analysis of the Painting

Solvent tests were conducted using cotton swabs with mild solvents commonly used for removal of natural varnish. The results of these tests indicated a slight potential paint solubility problem since the cotton swabs had a brown coloration. It was very difficult to tell from the swabs whether any of the greens had been affected. Tests of the paint layers and their binding media were then conducted to assess the effect of varnish removal on the paint underneath.

First, the cotton swab from the acetone cleaning tests was reextracted with acetone. Any material that dissolved was then cast on a potassium bromide pellet, and an FTIR spectrum was taken. This spectrum indicated the presence of a dammar-like natural resin (fig. 5). There was no indication of other material being present.

Fig. 5. FTIR spectrum of the extract from cotton swabs, used for acetone cleaning tests in removal of varnish in the greens from Thomas Dewing's America Receiving Nine Muses

Second, microtomed sections of paint sample were prepared and used for hot-stage melting points, fluorescence-staining, FTIR, and SEM analysis. Examination under the light microscope of the cross section of a sample from the green background indicated that the paint film is composed of a green over a white over a gold layer, with a white gesso layer at the bottom. The top green layer reacted positively for oil using the Rhodamine B fluorescence test.

The hot-stage melting point tests indicate that at between 80C and 90C a resin like material flows out of the top green layer and the green layer appears more broken up. At approximately 180C the green color disappears, and at 230C the white gesso layer turns brown. These color changes are unrelated to the presence of an oil and a wax or resin. The results indicate that the top green layer contains a wax or soft resin and that oil was used as the pigment binding medium. FTIR analysis of a microtomed section also indicates the possible presence of beeswax and oil in the green layer (fig 6). However, this evidence is not conclusive.

Fig. 6. FTIR Spectra of a microtomed section in the green layer from Thomas Dewing's America Receiving Nine Muses: (1) green layer from the microtomed section; (2) aged linseed oil spectrum; (3) beeswax spectrum

GLC of a portion of this section indicates the presence of oil and beeswax in the green paint layer and the presence of dammar from the varnish on top of the green paint. These results are presented as figure 7.

Fig. 7. GLC chromatogram of green layer from Thomas Dewing's America Receiving Nine Muses. A small piece of green paint with its varnish was used.

It must be noted that FTIR spectra of the extract from the cotton cleaning swab indicated only the presence of the major component—a dammarlike natural resin. The small amount of beeswax present in the green paint was not detected. FTIR is used in some conservation laboratories as a routine screening test for the safety of the solvent cleaning of pictures. Our findings indicate that the use of FTIR alone is not precise enough to provide absolute assurance. A chromatographic separation can provide more information.

The hot-stage melting point study of the microtomed cross section also indicates the presence of low-melting components such as a resin-wax in addition to the oil medium of the green paint. This study is simple and useful and can be used as an adjunct to more sophisticated techniques.

Fluorescence staining with the oil stain Rhodamine B indicates that drying oil is present in the top green layer. However, our subsequent findings indicate that oil staining alone also cannot be used as the sole source of analysis for a complex mixed media.

GLC and GC-MS of the dry paint sample and cotton swabs were used to confirm the presence of beeswax and drying oil. In this case study, the presence of beeswax was not detected by FTIR and fluorescence staining. However, it was detected by the hot-stage melting-point test and by chromatography.

A microtomed section from the green background was mounted on a graphite stub using a water-based adhesive prior to carbon coating in preparation for the SEM study. Solvent-based mounting adhesives that might affect the binding medium were avoided. A SEM photograph of the section shown in figure 8. Elemental analysis of this section by EDS shows chromium and zinc in the top layer, indicating the presence of chrome green and zinc white: lead in the second layer indicates lead white, and similarly the third layer contains gold and the fourth iron, indicating iron oxide. Lead white was present in the ground layer.

Fig. 8. SEM image of microtomed section from the green background from Thomas Dewing's America Receiving Nine Muses: (1) a top layer of chrome green and zinc white; (2) a layer of lead white; (3) a thin layer of gold; (4) a layer of red oxide; (5) a ground layer of lead white.

Copyright 1991 American Institute of Historic and Artistic Works