Brandi Lee MacDonald
A painting, as an object, consists of multiple components that, when analyzed together, have a unique story to tell about the artist, his or her practice, and the history of the piece. The supporting material, grounds, pigments, and varnishes that a painter chose to employ have the potential to reveal a great deal of information about the composition, context, and decision-making involved in the creation of a work, and their analysis contributes to our understanding of the artist’s oeuvre. Determining the chemical composition and identities of the pigments using spectrometric techniques is an essential component of the heritage scientist’s repertoire. Unsurprisingly, pigment analysis was a core component of the methods used to study paintings for The Unvarnished Truth: Exploring the Material History of Paintings. Artists’ pigments fall within one of two broad categories: inorganic and organic. Many resources exist that document histories of pigment types used over time,[i] and the essay “Histories of Selected Artists Pigments” in this volume provides a concise overview of selected types and the histories of their manufacture and use. Here we focus specifically on the analysis of inorganic mineral pigments that were used to execute the works attributed to Van Gogh, Jawlensky, and in the manner of Rubens. Four techniques were used to characterize the pigments: X-ray fluorescence (XRF), 2D macro-XRF (M-XRF), X-ray diffraction (XRD), and scanning electron microscopy-energy dispersive spectroscopy (SEM-EDS). It cannot be overstated that this research required the collaboration of individuals possessing expertise in areas of radiation physics, pigment and art history, and conservation science, and without their contributions, would not have been possible. The results of these tests provided valuable information for better understanding the works of art in question.
Analysis of Pigments: Techniques and Limitations
The material analysis of pigments is a critical component of understanding a painter’s work. A range of tools for this are available to the heritage researcher, and deciding which technique(s) to use depends on the type and resolution of data that are required to answer the research questions. Paintings are created by mixing mineral pigments with a binding medium (often an oil), which are then applied to and dried on a surface, and subsequently painted over with additional paint layers and a coating of varnish. This creates a complex series of layered materials that are challenging to isolate individually and characterize; therefore, the analytical methods applied must be carefully selected bearing this in mind. Four different techniques were used for this study, each suited for different purposes to obtain elemental and chemical data.
XRF (including portable and macro-XRF) is a technique that measures the elemental composition of the surface and near-surface of an object, in most cases covering an area of a few milimetres.2 The method involves bombarding the surface of the painting with primary X-rays powerful enough to dislodge inner K-shell electrons from their respective shell configurations. The atoms become unstable, and the outer L- and M-shell electrons replace the inner-shell vacancies until the atoms are charge satisfied. The energy that is emitted, or fluoresced, during the inner-shell ejections produces X-rays that are characteristic of those elements. Those X-rays are measured by a detector, and through a series of computations, qualitative and semi-quantitative data on the elements present are produced. (Detailed descriptions of this technique exist elsewhere.[ii]) XRF is capable of qualitatively and semi-quantitatively measuring virtually all elements in the periodic table, although low atomic elements, such as sulphur, aluminum, and sodium, have weaker fluorescence signals, are difficult to measure, and require advanced instrumentation.[iii] Many higher atomic (high-Z) elements, including barium, iron, lead, and zinc, are of interest because they are major components of prepared pigments. XRF is an important screening tool for the heritage specialist as it is nondestructive, does not require the removal of sample material, and is relatively low cost. Recent advances in available portable technologies have enabled its use for the analysis of paintings and other objects in situ, which is of significant benefit as some artworks are too large, unstable to move, or must be measured where they are housed. However, there are a number of limitations to this technique. For the purpose of examining artists’ materials, this method measures elemental concentrations, not chemical compounds, and therefore some pigment compositions can only be inferred. For some pigment types, such as zinc white (ZnO) or titanium white (TiO2), pigment identification can be a straightforward task. However, some pigment groups, such as copper carbonates or acetates[iv] or iron oxides, have heterogeneous and complex chemistries, and to determine precisely what they are requires additional complementary information, such as chemical or mineralogical data. Further compounding the XRF measurement process is the depth at which primary X-rays penetrate the surface of a painting and the attenuation of the resulting fluorescence X-rays by the target material. Primary X-rays will penetrate the surface of the object at a depth ranging anywhere from a few microns to a few millimetres, the degree of penetration being influenced by the matrix density and chemical makeup. For example, primary X-rays can penetrate samples consisting of lower Z matrices (such as bone) more deeply than those of higher Z materials (such as metal)[v]. Therefore, when analyzing pigments one must consider not only the surface of the painting but also subsurface features such as the supporting material (canvas, wood panel, cardboard), the ground applied underneath the presentation layer (lead white, chalk, or mixtures of other materials), the pigment-binding medium, the homogeneity of the pigment (one or more materials used and their degree of mixing), the manner in which the artist applied the paint (impasto, thick buildup of paint), the different attenuating effects of all of these layers combined, and the influence they have on accurate instrumental measurements. This complex series of factors is why it is nearly impossible to accurately quantify the elemental concentrations of pigments via XRF, and it is considered primarily as a qualitative and semi-quantitative method.
In addition to XRF analysis, XRD and SEM-EDS techniques were used for this project. These methods are considered micro-destructive as they often require small samples to be physically removed from the surface of a painting. The decision to remove a sample of pigment from a work of art, regardless of how inconspicuous it may be, is never taken lightly. As technologies have advanced in sensitivity, accuracy, and precision, the quantity of sample needed for analysis has dramatically decreased, and when using XRD and SEM-EDS the minimum sample sizes required are smaller than the size of a pinhead. The removal of this small amount of paint from an inconspicuous area, which is subsequently retouched by a skilled hand, is virtually indistinguishable to the naked eye. XRD is a technique that determines crystal structures of inorganic and organometallic materials. It is used for determining the mineral bases of inorganic pigments, and provides information on the presence and relative proportions of these components. A sample is mounted in the path of a beam of primary monochromatic X-rays and rotated at different angles of orientation. The X-rays that hit the target sample scatter, or diffract, in a pattern that is characteristic to the molecular crystal structure. Those diffracted X-rays are measured and compared to a library of diffraction patterns of known chemical compounds for identification. An advantage to this technique is that it is possible, with only a microscopic sample, to determine the chemical compounds present and to differentiate classes of pigments that are indistinguishable on the basis of their elemental composition, for example, copper resinates and acetates, different iron oxides, or lead-based compounds. One drawback to this technique is that because such a microscopic fragment must be removed from the painting, the small sample size runs the risk of not fully representing the range of pigments or underlayers present.
Sample preparation for SEM-EDS requires the paint sample to be embedded in a resin and ground down to produce an optically flat cross section. A focused beam of electrons is applied to the sample to produce both a high-magnification image and elemental data through the analysis of characteristic X-rays produced through the imaging process. An advantage to this technique is that it allows for the linear exploration of elements along the cross-sectional surface of a sample, making it possible to target and identify individual mineral grains (Fig. 7). This is useful for identifying multiple layers of paint as well as the different constituents of paints that are chemically heterogeneous. Our use of XRD and SEM-EDS on the portrait of Maximillian, Archduke of Austria, in the manner of Rubens, described below, yielded valuable information regarding the pigment and ground layer compositions.
A Survey of Artists’ Materials: X-ray Fluorescence
For The Unvarnished Truth, we screened the nine paintings of interest using an Olympus Innov-X Delta Premium model handheld XRF. The unit has a rhodium anode X-ray tube, an SDD-type detector, and operates up to 40 kV and 0.1 mA. These devices can be configured to different modes, each applying different voltage, current, and filter combinations that optimize the fluorescence of different suites of elements.[vi] Readings were taken in 3-beam soil mode at 60 seconds each at varying currents and amperages. Focusing on testing as many different colours as possible, we targeted multiple areas of interest on each painting to obtain data on the elemental composition of the pigments used. Figure 1 shows areas of Gossaert’s portrait of a man (Cat. #) that were tested. The results indicate areas with iron, mercury, lead, copper, and calcium, and from this we infer that the painting was executed using earth pigments (Fe2O3), cuprite (Cu2O), vermilion (HgS), lead white (2PbCO3), and calcium carbonate (CaCO3). The iron oxides were likely used for browns and reds in areas of the background and in the flesh tones of the sitter. The cuprite, a red oxidized form of copper, was used in areas of the sitter’s shirtsleeve. Lead white was likely used throughout to lighten other hues and as part of the preparation ground. Calcium could represent chalk white used in the preparation ground, or bone black used to darken other hues, although chemical compound testing would be required to confirm this. Vermilion, a red-coloured mercuric sulphide, was found in trace amounts throughout. These pigments are typical of those used by a painter active during this time period.
Figure 2 shows Jawlensky’s colourful piece Murnau Landscape with Three Haystacks and the areas that were tested via XRF. By the time this work was created in the early twentieth century, the mass commercial production of synthetic pigments was widespread. Mineral pigments whose components included chromium, zinc, lead, titanium, and cadmium were ubiquitous[vii]. Table 1 summarizes the areas that were measured, the elements that were detected, and corresponding commercial pigments that the elemental results could indicate. Jawlensky was known to have used many of the commercial pigments[viii] described in Table 1; therefore, we expected to see some of their elemental components in the XRF spectra. Figures 3a–3c are spectral representations and descriptions for measurements in green, yellow-red, and violet areas.
Table 1: Summary of areas measured, elements detected, and corresponding potential commercial pigments for the analysis of Murnau Landscape with Three Haystacks, 1908–09.
|MeasurementArea and Colour||Elements Detectedvia XRF in Order of Abundance||Potential CommercialPigments in Common Use(post-1900)*|
|A: Light and darkgreen, lower area||ChromiumLeadZincTitanium
|Chrome green (Cr2O3)Viridian (Cr2O3.2H2O)Lead white (2PbCO3×Pb[OH]2)Lead chromate (PbCrO4)
Zinc white (ZnO)
Titanium white (TiO2)
Cadmium yellow (CdS)
Black iron oxide (FeO×Fe2O3)
Bone black (C + Ca3[PO4]2)
|B: Yellow and orange areas||LeadZincChromiumMercury
|Lead white (2PbCO3×Pb[OH]2)Lead chromate (PbCrO4)Zinc white (ZnO)As a component of lead chromate
Cadmium yellow (CdS)
Titanium white (TiO2)
|C: Violet area||LeadTitaniumZincChromium
|Lead white (2PbCO3×Pb[OH]2)Lead chromate (PbCrO4)Titanium white (TiO2)Zinc white (ZnO)
Chrome green (Cr2O3)
Strontium chromate (SrCrO4)
Cadmium yellow (CdS)
Black iron oxide (FeO×Fe2O3)
Umber (Fe2O3 + MnO2)
Manganese violet (NH4MnP2O7)
*In use during the time that Jawlensky was an active painter.
This suite of elements suggests that the artist was mixing a range of pigments to create a colourful palette. The chemical compositions present are consistent with those of many of the modern commercial pigments noted above, and with those described in historical documentation confirming usage by the artist.[ix] There are several challenges and limitations for precisely identifying (via XRF) which paints were used for this piece including the degree and combinations of paint mixing, the variable thickness of application and topography of the surface, and the presence of multiple chrome-derived pigments. The mixing of different paints results in variability of the proportions of different elements present. This creates a chemically complex XRF spectra that not only exhibits overlapping and unresolvable peaks (e.g., Ka and La peaks for barium and titanium) but also renders it difficult, if not impossible, to attribute specific elements to specific pigments, potentially leading to misinterpretation. For example, chrome-derived pigments—of which there are multiple occurring in this painting—could include a range of hues, from yellows to greens and reds. Where multiple chrome-derived and other similar pigments are measured simultaneously, it is not possible to attribute them to commercial pigments without the use of additional techniques that measure chemical compounds. Our analysis of the Jawlensky piece also presents an example of the challenges of interpreting overlapping peaks. We have identified the possible presence of manganese (Ka1 @ 5.8 keV; Kb1 @ 6.4 keV) in the violet pigment sections (Area C) indicating the potential use of manganese violet pigment (NH4MnP2O7); however, in the presence of significant quantities of chromium (Kb1 @ 5.9 keV) and iron (Ka1 @ 6.4 keV), the presence or quantity of manganese is virtually impossible to resolve using this technique (Fig. 3c). Furthermore, the depth at which the pigments lie from the surface has an effect on the signal that the XRF technique can detect. Because of the variable surface topography of this particular work, and the multiple, thick layers of pigment that were applied, the XRF measures a combination of pigment layers simultaneously. Of the nine paintings surveyed using this technique, Murnau Landscape with Three Haystacks has, by far, the most complicated mixture of elements and pigments to identify, and serves as an example of the chemical complexity of modern commercial pigments and the importance of careful analytical interpretation of measurements from heterogeneous materials. Further work on this piece would involve chemical or molecular spectrometric techniques, such as XRD, to determine the compounds present.
Layers of Complexity: Mapping Pigments via Macro-XRF
Using imaging techniques, such as IRR as well as neutron- and X-radiography, it is possible to visualize and examine subsurface features of a painting such as underdrawings and multiple layers of pigment.[x] X-radiography of Untitled, Still Life: Ginger Pot and Onions, attributed to Van Gogh, revealed interesting features suggesting the application of multiple layers of pigment. Figure 4 compares the film X-ray and visible light images indicating where areas of variability in contrast are evident. In the X-ray image, the areas to the immediate centre left of the green jar show an amorphous shape of white contrast. However, on the visible surface of the painting there is no corresponding change in the pigments in shape or pattern. In an X-ray image, changes in contrast such as these are attributed to mineral pigments that contain heavy elements.[xi] When we compare the X-ray image with an elemental map created using macro-XRF, it is possible to create a visual representation of the elemental compositions of the different layers.[xii]
Macro-XRF is a technique that works on the same principles as XRF, but it is capable of scanning an array of points covering an area of interest resulting in a two-dimensional elemental map of the painting. For this experiment we tested the feasibility of a coarse analysis using a low-cost, low-powered benchtop X-ray tube[xiii] and then compared the results to the handheld XRF and X-ray image data. The system used for this test was a commercial X-ray tube that focused the X-ray beam to 0.5 mm × 1.0 mm in area. A custom holder was assembled to support the painting, and the holder was mounted onto a miniature jack to allow for controlled vertical and horizontal movement. A micrometer-driven linear stage was used to manually operate the movement of the painting. Figure 5 is a schematic representation of the layout. A He-Ne laser was used to identify the location of the X-ray beam, and a semiconductor detector was used to record the X-ray signals fluoresced by the painting. A series of 60-second measurements was taken for a total of 581 points measured covering a 70 mm × 45 mm area of the painting. The peak intensities measured at each data point were identified for all elements present. This provided a spatial representation of the distribution of individual elements across the scanned region of the painting, showing changes in the intensities of a corresponding element.
Features in the 2D XRF map follow similar trends in the X-radiograph, with an angled feature visible in both modalities. This is most strikingly visible for lead (shown in Fig. 6) as well as for iron, zinc, and mercury. These features, when considered alongside the X-ray image, suggest that a different composition (although visually unresolvable) exists underneath what is visible on the surface layer. There is the potential that Van Gogh may have recycled this canvas, and it is documented that during his time in Nuenen in the Netherlands he used zinc white to re-prime previously used canvases as a way of reusing expensive materials.[xiv] Based on our handheld XRF survey of this piece, zinc was present throughout the painting. Zinc white is an inexpensive extender that dries poorly in oil[xv]and is known to result in characteristic surface cracking over time. Under high magnification, it is possible to see this cracking, and in the area of interest, we can see between those cracks to what appears to be a lighter pigment layer underneath (see Fig. 7 micrograph). While the form of this underlying image is unresolved in X-ray and IRR imaging, it is still possible to see that there are areas where whiter pigments lie underneath visible layers, and the handheld and macro-XRF elemental data support this interpretation. For this piece, perhaps Van Gogh had abandoned an earlier work, or scraped down a previous work to the point of obscurity. Others have reported that during the same months that he painted Untitled, Still Life: Ginger Pot and Onions at least another ten of his works that were analyzed using XRF and X-radiography exhibit the same pattern of overpainting and reuse.[xvi] Higher resolution techniques, such as those described by Legrand et al.,[xvii] or the combination of techniques used by Dik et al.[xviii] (portable-, macro-, and micro-XRF, infrared reflectography, XANES[xix], SEM-EDS, and synchrotron radiation XRF), could potentially resolve the underlying image of Untitled, Still Life: Ginger Pot and Onions. However, this would require overseas transportation of the work, which is, unfortunately, cost prohibitive.
SEM-EDS and XRD of Maximilian, Archduke of Austria
Our preliminary XRF survey of the portrait of Maximillian, Archduke of Austria, in the manner of Rubens, showed that elemental analysis alone was insufficient for determining all of the materials used. Testing of the area that serves as background to the sitter indicated the presence of copper-derived pigments, which one might expect to show as a brighter, vibrant green hue. However, in visible light the area appears to be a muted black-brown. Most copper-derived pigments are impossible to differentiate on the basis of elemental data alone. Because the attribution of the piece remains in question, our goal was to determine precisely which pigment was used and what the application of paint layers could reveal about the painting’s composition. To do this, we removed two tiny paint fragments for XRD and SEM-EDS analysis. Figure 8 shows the areas from which the samples were taken.
The first paint sample was submitted to McMaster’s Analytical X-ray Diffraction Facility for XRD2 rapid-phase analysis. It was mounted between two Mylar films and analyzed in air using a Bruker Mo Smart APEX2. The system is equipped with a Rigaku RU200 Cu Kα(bar) rotating anode, a Bruker Smart6000 CCD area detector, Bruker 3-circle D8 goniometer, and Göebel cross-coupled parallel focusing mirrors. The detector was calibrated using corundum powder, and three 300-second frames were collected at -20, -40, and -60 degrees 2Θ. The power setting was 50kV, 90mA, and a 0.5 mm collimator was used, and the data were collected using standard Bruker-AXS software. Figure 9 shows the spectral results, indicating that the copper-derived pigment consisted primarily of azurite with traces of chalcopyrite (CuFeS2). Chalcopyrite is a copper mineral that oxidizes to azurite, which is then prepared as a pigment through a process of grinding, washing, levigation, and sieving.[xx]
Sample 2 was embedded in epoxy resin and ground down to produce an optically flat cross section for SEM-EDS analysis. The cross section was examined in a JEOL JSM-6460LV scanning electron microscope with an Oxford Instruments INCAz-sight energy dispersive X-ray spectrometer, and a 133 eV resolution at 5.9 keV. Uncoated samples were examined in low vacuum mode (35 Pa) at 20 kV at a working distance of 10 mm. Images were taken with the JEOL BSE (back-scattered electron) detector in “shadow’” mode (Fig. 10), which combines a typical BSE image with some topographical information. The results of the X-ray spectroscopic analysis revealed that the ground layer is composed primarily of lead white. Directly above the ground layer, another layer—perhaps an initial application of tone—contains a combination of calcium, lead, and copper, and the upper three-quarters of the paint layers also contain copper. Consistent with the XRD results, these are likely azurite and chalcopyrite crystals. It is important to also note here that the cross-section image reveals that in the two layers of paint described, there is not a smooth or clean transition between them. The boundary is somewhat amorphous and disrupted in appearance, suggesting that the upper, or “presentation,” layer was applied very soon after the previous one. This leads us to believe that the background was not repainted at a later time, and in its original form may have appeared to be a green-blue hue. We suspect that the reason the painting now looks very dark is a result of successive applications of oil or resin varnishes applied after cleaning, and of the well-documented phenomenon of azurite pigments fading to dark over time. Gettens and Fitzhugh[xxi] describe how thick layers of azurite in oil become very dark green, in some cases almost black, and list other examples of this. This exploration of paint samples via XRD and SEM-EDS provided insight on the pigment composition and condition, as well as the layering of multiple applications of paint and ground.
As techniques for the visual and chemical analyses of paintings become more widely available, heritage scientists are using these tools to conduct an increasing amount of research into the techniques and materials used by painters, and the condition and life histories of works of art. When these data are combined with art historical information, these explorations have the potential to reveal significant information that would otherwise be unknown. For The Unvarnished Truth project, XRF analysis of Jawlensky’s Murnau Landscape with Three Haystacks verified the presence of a series of modern (post-1850) materials used by the artist such as titanium white, zinc white, and strontium yellow, and also highlighted the limitations of this technique for the examination of chemically complex paint layers. When combining visual modalities such as X-radiography and high-resolution microscopy with elemental mapping, it is possible to glean information on the pigments and condition, as demonstrated in our examination of Van Gogh’s Untitled, Still Life: Ginger Pot and Onions. While the image shown in the X-radiograph and the macro-XRF elemental mapping indicates the presence of an earlier work and the potential recycling of this canvas, this issue remains unresolved until higher resolution techniques can be applied. Furthermore, the analysis of paint samples removed from the portrait of Maximillian, Archduke of Austria via XRD and SEM-EDS provided valuable information regarding its composition and history including confirmation that the use of azurite as a pigment and that the portrait would have likely had a green-blue background rather than a dull black-brown one. The results from these explorations are a testament to the potential for interdisciplinary research in heritage science and what the use of these tools can contribute to our understanding of art history.
[i] Pigment classes and histories are described in further detail in the essay “Histories of Selected Artists’ Pigments” in this volume.
[ii] For extensive discussions on the physics, configurations, and limitations of XRF, see Michael Mantler and Manfred Schreiner, “X-ray Fluorescence Spectrometry in Art and Archaeology,” X-Ray Spectrometry 29 (2000): 3–17.
[iii] Michael Mantler and Manfred Schreiner, “X-ray Fluorescence Spectrometry in Art and Archaeology,” in X-Ray Spectrometry 29 (2000): 3–17.
[iv] Chris McGlinchey, “Handheld XRF for the Examination of Paintings: Proper Use and Limitations,” in Handheld XRF for Art and Archaeology, ed., Aaron Shugar and Jennifer Mass (Leuven, Belgium: Leuven University Press, 2012), 131–58.
[v] For more detailed discussion of X-ray mass attenuation coefficients for different materials of heritage interest, see Michael Mantler and Manfred Schreiner, “X-ray Fluorescence Spectrometry in Art and Archaeology,” in X-Ray Spectrometry 29 (2000): 3–17.
[vi] Nathan Goodale, David G. Bailey, George T. Jones, Catherine Prescott, Elizabeth Scholz, Nick Stagliano, Chelsea Lewis, “pXRF: A Study of Inter-instrument Performance,” in Journal of Archaeological Science 39 (2012): 875–83.
[vii] See the essay “Histories of Selected Artists’ Pigments” in this volume for detailed descriptions of pigment manufacturing history.
[viii] Roy S. Berns, Lawrence A. Talpin, Francisco H. Imai, and Ellen A. Day, “A Comparison of Small-Aperture and Image-Based Spectrophotometry of Paintings,” in Studies in Conservation 50 (2005): 253–66
[ix] Stefan Zumbuehl, Nadim C. Scherrer, Alfons Berger, and Urs Eggenberger, “Early Viridian Pigment Composition: Characterization of a Hydrated Chromium Oxide Borate Pigment,” in Studies in Conservation 54 (2009): 149–59; and Maria Jawlensky, Lucia Pieroni-Jawlensky, and Angelica Jawlensky, Alexej von Jawlensky, Catalogue Raisonné of the Oil Paintings, Vol 1, 1890–1914 (London: 1991).
[x] For detailed descriptions of imaging techniques, see the essays “Imaging Using X-rays and Neutrons” and “The Technical Art Historian’s Methods on Investigation” in this volume.
[xi] B. I. Reiner, E. L. Siegel, K. J. French, R. S. Dentry, W. T. Mazan, M. J. Maroney, “Use of Computed Radiography in the Study of an Historic Painting,” in Radiographics 17, no. 6 (1997): 1487–95; Olivier Schalm, Ana Cabal, Piet Van Espen, Nathalie Laquiére, and Patrick Storme, “Improved Radiographic Methods for the Investigation of Paintings Using Laboratory and Synchrotron X-ray Sources,” in Journal of Analytical Atomic Spectrometry 26, no. 5 (2011): 1068–77.
[xii] K. Janssens, G. Vittiglio, I. Deraedt, A. Aerts, B. Vekemans, L. Vincze, F. Wei, I. De Ryck, O. Schalm, F. Adams, A. Rindby, A. Knöcgekm, A. Simionovici, and A. Snigriev, “Use of Microscopic XRF for Non-Destructive Analysis in Art and Archaeometry,” in X-ray Spectrometry 29 (2000): 73–91.
[xiii] M. Zamburlini, “In Vivo Measurement of Bone Strontium with X-ray Fluorescence,” (PhD diss., McMaster University, 2008).
[xiv] R. Haswell, L. Carlyle, C. T. J. Mensch, and M. Geldof, “The Examination of Van Gogh’s Painting Grounds Using SEM-EDX,” in Van Gogh’s Studio Practice, ed. Marije Vellekoop, Muriel Geldof, Ella Hendriks, Leo Jansen, and Alberto de Tagle (Brussels, Belgium: Mercatorfonds, 2013), 202–15.
[xv] Ella Hendriks and Louis van Tilborgh, Vincent Van Gogh, Paintings, Volume 2: Antwerp and Paris 1885–1888 (Zwolle: Waanders and Amsterdam, the Netherlands: Van Gogh Museum, 2011), 115–16.
[xvi] L. Megens and M. Geldof, “Van Gogh’s Recycled Works,” in Van Gogh’s Studio Practice, 306–29.
[xvii] Stijn Legrand, Frederik Vanmeert, Geert Van der Snickt, Matthias Alfeld, Wout De Nolf, Joris Dik, and Koen Janssens, “Examination of Historical Paintings by State-of-the-Art Hyperspectral Imaging Methods: From Scanning Infrared Spectroscopy to Computed X-ray Laminography,” in Heritage Science 2 (May 2014): 2–13.
[xviii] Joris Dik, Koen Janssens, Geert Van der Snickt, Luuk van der Loeff, Karen Rickers, and Marube Cotte, “Visualization of a Lost Painting by Vincent van Gogh Using Synchrotron Radiation Based X-ray Fluorescence Elemental Mapping,” in Analytical Chemistry 80 (2008): 6436–42.
[xix] XANES stands for X-ray absorption near edge structure.
[xx] R. J. Gettens and E. W. Fitzhugh, “Azurite and Blue Verditer,” in Artists’ Pigments: A Handbook of their History and Characteristics, ed. A. Roy, vol. 2 (Washington, DC: National Gallery of Art, 1993), 25.
[xxi] Ibid., 27.