A portable-point X-ray flourescence (p-XRF) scanner is used on an illuminated manuscript fragment.
Photo credit: Cornell University Library

Librarians know there are many tools to classify items—subject matter, date, geographic region, and old-fashioned alphabetical order—but there is a newer method that can help classify manuscripts based on features invisible to the naked eye. X-ray fluorescence (XRF) can non-invasively find traces of chemical elements, offering clues about the origins of centuries-old paintings, manuscripts, ceramics, and stained glass. For example, researchers at the library at Saint Catherine’s Monastery, a 1,452-year-old institution at the foot of Mount Sinai, Egypt, use x-ray technology to discover once hidden text and discover lost characters from extinct languages. “Scanning x-ray fluorescence is a relatively new technique to study illuminated manuscripts,” according to Louisa Smieska, a researcher at Cornell University’s Cornell High Energy Synchroton Source (CHESS). CHESS is using XRF to examine medieval documents in Cornell’s collections. “What’s different about our work is the use of higher-energy x-rays, generated at a synchrotron source, compared to the lower-energy x-rays that have been used in laboratory-based experiments. The higher-energy x-rays gave us greater sensitivity to heavy trace elements.” CHESS has been a cross-discipline program, using technology to determine findings of interest to history, literature, and the arts. Therefore, the researchers come from both the scientific and humanities worlds. Smieska, who has a PhD in chemistry, explained, “XRF detects which chemical elements…are present in each spot we analyze. If we know which elements are present, we can often infer what the compound is—for example, the presence of mercury in a red area strongly suggests the use of vermilion, a pigment that contains mercury and sulfur. XRF works by shining x-rays onto an object and measuring the secondary x-rays that the object emits in response. These secondary x-rays have energies that are precisely specific to the elements they come from.” The presence of specific chemical elements can help researchers uncover previously unrecorded facts about both the origins of the manuscripts and their methods of creation. As she explained, “The part that is most exciting to us…is that the relative amount of barium, as well as other trace elements like iron, zinc, and antimony, in each azurite blue is not the same. We think that this detailed composition information could eventually help us identify instances where the pigments used in different leaves are related to one another—whether they may have come from the same region, or even the same workshop or the same volume.” It is possible that by using XRF, researchers could determine that two manuscripts thought previously unrelated are actually fragments of the same whole.

Library Science

From left, Ruth Mullett, Louisa Smieska, and Arthur Woll at CHESS with one of the manuscript fragments mounted to be scanned.
Photo credit: Cornell University

Ruth Mullett, a PhD candidate in medieval studies at Cornell, had been working to catalog the Cornell fragments with Laurent Ferri, then the curator of the pre-1800 collections in the Division of Rare and Manuscript Collections, when she learned of Smieska and Arthur Woll’s work with CHESS. The project intrigued Mullett. “Both Laurent and I agreed that studying these fragments using x-ray fluorescence and diffraction technologies offered tremendous potential for learning more about medieval manuscript production. So, I was involved in selecting fragments for further study according to their geographic, temporal, and decorative range. I then helped to interpret what the collected data might mean in terms of the production and use of medieval manuscripts.” Smieska also described her role: “During this research, I was a postdoctoral researcher at CHESS. My role, with Arthur, was to plan the technical details of the portable XRF survey and the scanning XRF and XRD [x-ray diffraction] measurements we made at CHESS. I was also in charge of analyzing the technical data we collected. Collaborating with Ruth and Laurent was fantastic. They were so supportive of the experiments and enthusiastic in working to understand the results!” Smierska recounted, “We started our experiments by using a portable XRF instrument to survey a group of 27 fragments in the Cornell Library Rare and Manuscript Collections, taking a few readings from different spots in each object. This survey gave us a general sense of the palette used in each leaf or fragment. Many of the blue pigments we studied were rich in copper, which suggested they were made of the blue copper mineral azurite. However, some of these copper-rich blues also contained the element barium. We were surprised to find barium in the azurite blues because we hadn’t seen this finding reported in illuminated manuscripts before. We often think of barium as associated with modern paints or, in smaller amounts, in natural clays or chalks, but not with azurite.” XRF is not the only use of x-rays that help researchers find out about these items. The CHESS researchers use, in addition to XRF, a scanning technique called x-ray diffraction (XRD). As Smieska explained, “XRD is less sensitive to small amounts of material than XRF, but it can directly identify compounds and tell the difference between compounds that contain similar elements (for example, lead white versus red lead, or copper minerals like azurite blue and malachite green). Scanning XRD allowed us to confirm the inferences we made using XRF about the major components of each pigment color. We found that many azurite blues contain small amounts of the mineral barite, or barium sulfate. Barite is a fairly common mineral, and its presence is likely related to how the azurite mineral deposits formed. We are excited because the relative amount of barium in each azurite blue is not the same, and combining this information with the amounts of other trace elements such as iron, zinc, and antimony might help with efforts to learn whether different fragments were originally related to one another.” The researchers used specific criteria for determining which works to examine in more detail. “When selecting manuscripts to take to the synchrotron, we chose fragments that yielded unusual or surprising results in the portable point XRF survey while still representing a geographical and temporal range,” explained Mullett.

Traveling treasures

In order to ensure safe transport, Cornell University Library’s conservation team prepared custom mounts for the manuscripts’ transportation and examination. The leaves were mounted, placed in a hard box, and driven the short distance between the library and the synchrotron. The researchers did not have to worry about how to protect the items during XRF because no one had to touch them. Smieska explained, “X-ray fluorescence is a noninvasive analytical method, which means we never had to touch the object surface in order to analyze it, during point analysis and at the synchrotron. The x-rays at a synchrotron are much more powerful than the laboratory point-analysis tool, so before bringing the objects to the synchrotron, we analyzed fragments of parchment from a study collection to verify that our experiment conditions were safe for the library’s pieces. The x-ray exposure we used in our experiments at CHESS were orders of magnitude smaller than maximum recommended safe levels for parchment. We also designed the experiment at CHESS to ensure that x-rays could only reach the objects while the scan was in progress—if the scan were to stop for any reason, the shutters would close and block the x-rays. Fortunately, everything went smoothly and this precaution wasn’t needed.”

More manuscripts wanted

The team published an article on their findings” in last July’s issue of Applied Physics A. Smieska concluded, “The results are encouraging, suggesting that trace element analysis could help us find links between manuscript fragments, but so far we have only examined six manuscript leaves containing azurite pigments at CHESS. We would like to see this project grow—expanding our study to additional manuscript fragments would be extremely valuable for uncovering broader trends.” Save