AIC's 46th Annual Meeting

Material analysis was crucial in treating Portrait of an Old Woman by Maarten de Vos (National Gallery of Art, Washington). During varnish removal the extent of overpaint became apparent; non-invasive and invasive analytical methods were used to determine its composition and distribution. Micro-sample analysis of the background and sitter’s hat revealed non-original materials: a discolored drying-oil layer (characterized by FTIR and GCMS); at least two layers of oil-based overpaint covering the hat; and at least three layers of oil-based overpaint covering the background. Stratigraphy revealed in cross sections guided decisions regarding treatment in these areas. The non-original oil layer was key to successful overpaint removal, providing a barrier between original and overpaint. More challenging was determining the extent of overpaint on the sitter’s black garment. Microscopic visual examination of the paint surface showed clear evidence of overpaint: a coarse-textured dark paint layer traversing cracks and damage in the underlying paint. A cross-section taken from the garment revealed two dark paint layers without intervening varnish or oil layer. The upper layer (the coarse dark overpaint noted above) was rich in smalt as determined by PLM and SEM-EDX (Si, Co, As, Ni identified). This layer also contains earth pigments (Fe) and small amounts of lead white (Pb). The lower layer did not contain smalt and had larger amounts of lead white and earths with traces of umber (Pb, Fe, Mn). To determine the extent of the dark, smalt-rich overpaint compared to the original paint, X-ray fluorescence imaging spectroscopy was performed. The co-localization of cobalt, arsenic, and nickel in the XRF maps indicated the presence of smalt across the garment. Smalt original to the painting was also present on the right side of the background. However, interestingly, the ratio of nickel to cobalt showed the smalt used in the background had a higher Ni content compared to that found in the garment, suggesting two different sources of smalt were used. XRF maps of Co, As and Ni have distributions that relate to the surface design of the garment; however, XRF maps of Pb, Fe, and Mn show a different design that may relate more to the lower, original paint layer identified in the cross-section. The inclusion of smalt in the overpaint, rare after the seventeenth century, suggests it was an early intervention. Subsequently, tests were undertaken to remove the overpaint from the garment. It was challenging, however, to see a clear separation between the overpaint and the original layer, and it was ultimately decided that full removal imparted too much risk. The dark overpaint was reduced slightly in some areas, and any discontinuities between overpaint and exposed original paint were compensated during retouching. The treatment of Portrait of a Woman offers an example of the important role analytical and imaging techniques play before and during treatment in identifying original versus non-original materials and making informed treatment decisions. By the same token, this project highlights the humbling physical limitations of treatment options that conservators often encounter despite having a thorough understanding of materials.