Pigment Analysis in Paintings
Written by Igor Lukačević, Ph.D. and Ante Matanić – University of Osijek, Croatia
As two colors in a painting can look the same to the human eye, their reflective spectral signatures can reveal differences in the visual part of electromagnetic spectrum. These differences, or lack of them, can be used to give answers looked for by art historians and collectors interested in cultural heritage.
Using a portable miniature spectrometer, diffuse reflectance spectra of the same colors in different paintings were measured and compared. If the same pigments were used on different paintings, it could be another clue that the paintings were made by the same author.
“Vukovar landscapes” is a series of oil on canvas paintings made in the mid-19th century by an unknown author. Only one of the landscapes is signed, by Joseph Franz Mücke, a royal painter of Habsburg dynasty, presumably ordered by Count Emmerich Josef Eltz as a decoration for his Manor in Vukovar, Croatia.
Of all candidate authors, Mücke is the only one who was known to live in Vukovar in the period the landscapes were made. However, certainty requires more evidence. One of the clues could be given by comparing the painting pigments used in signed and unsigned landscapes. If they are the same, then the unsigned landscape is more probably painted by the same painter. Spectroscopy analyses by UV-Vis Fiber Optic Reflectance Spectroscopy (FORS) can provide such evidence by comparing the reflectance spectra of the same colors in different paintings.
Reflectance spectra were taken using a USB2000 spectrometer (200-850 nm range) with R400-7-Vis-NIR probe and RPH-1 probe holder in diffuse mode, because the painting surfaces are not smooth and diffuse reflection is a dominant process. The WS-1 diffuse reflectance standard was used as reference. The light source was an LS-1 tungsten halogen bulb. [Editor’s note: Some of the products mentioned here are now available in next-generation versions, including the Flame-S spectrometer and HL-2000 tungsten halogen source.]
A total of 13 landscapes were analyzed (one signed landscape named “Gardens” and 12 unsigned) during their exhibition in Eltz Manor by permission from Gallery of Fine Arts in Osijek and City Museum in Vukovar, both in Croatia. As paintings were mounted on the walls during the exhibition, the portability of spectroscopy equipment was an essential feature. Some paintings were positioned at the height of over 2 meters, requiring us to use ladders. From each painting we sampled the spectrum at several points that contain the same colors as on the signed landscape.
An example of reflectance spectra comparison is given in Figure 4. It can be seen that blue pigments’ spectra on most of the landscapes agree well with the one from the signed landscape. Higher reflection in 400-450 nm region and lower reflection in 550-650 nm region characterizes all blue color pigments.
However, not all pigments’ spectra agree that well. An example is shown in Figure 5. for light green color. One can see that two spectra belonging to unsigned landscapes have higher reflection around 450 nm and even much higher after 600 nm.
Based on our results we were able to confirm that indeed some of the colors appearing on all landscapes were made using the same pigments (blue, orange). Presuming that all the painters of that era had their own source of pigments, our results give an advantage to the premise that landscapes were painted by the same author — i.e., Joseph Franz Mücke.
On the other side, some colors’ spectra did not shown agreement (greens, brown), leading us to the conclusion that they were not made using the same pigments. However, this disagreement could also be a consequence that we did not sample the areas containing the same color. Also, this disagreement could originate from mixtures of pigments, mostly additions of white or dark pigments to give a different hue. This is where other spectroscopy methods could complement our study. With Raman spectroscopy one is able to find out the molecular composition of pigments, taking these disadvantages of FORS methods out of play and bringing more certainty to the results.