There is more to the Declaration of Independence than “We hold these truths to be self-evident: that all men are created equal.” Further down the parchment, the future Americans list reasons why the king of Great Britain is an unfair and unsuited ruler for America. At one point, Thomas Jefferson wrote, “He has constrained our fellow citizens taken captive on the high seas to bear arms against their country, to become the executioners of their friends & brethren, or to fall themselves by their hands.” However, there is a smudge—rather than a cross out, which is common throughout the rest of the document—by the word citizen.
Heritage science researchers recently used a variety of instruments typically found in the laboratory to discover that Jefferson actually wrote fellow “subjects” before smudging the ink to replace the word with “citizens.”
“It is a fairly distinctive change given what was happening at the time,” said Fenella France, Chief, Preservation Research and Testing Division at the Library of Congress, in an oral session at Pittcon 2020.
France emphasized the multi-disciplinary nature of heritage science research, with collaboration from often disparate fields needed to undertake basic and applied research. In physical, chemical, optical and aging laboratories across the world, researchers rely on hyperspectral and multispectral imaging, FTIR, SEM, particle analysis, XRF, FORS, XRD and more to uncover secrets of the past and enable preservation of our world cultural history.
Hidden Medicine for Gladiators
Galen of Pergamon was a physician, surgeon and philosopher in the Roman Empire from 129 to 210 AD. Tending to gladiators and emperors alike, Galen was one of the most influential physicians of the time, and his views went on to influence Western medical science. One of his many works, “On the Mixtures and Powers of Simple Drugs,” was an important pharmaceutical text; however, the 6th-century manuscript was erased and written over with hymns sometime in the 11th century, creating a palimpsest, as it’s called.
“The palimpsest’s importance lies in the fact that Galen wrote in Greek. His medicine has been influential for over 1,000 years, especially in the Arab world. The question is, ‘How did these Arab writings come about? Were they directly translated from Greek, or did they come through Syria, translated through Syriac?’ Being able to read the Syriac translation of Galen’s work, we might shed light on how the information was transmitted from the Greek/Roman empire to the rest of the world,” said Uwe Bergmann, a distinguished staff scientist at the Stanford PULSE Institute at SLAC, during a session at Pittcon 2020.
To make a palimpsest, ancient peoples used lemon juice, milk or wine to wash out the old writing, cut the parchment in two, rotated it by 90 degrees, and then wrote over it.
“That’s why the original text and new text are in different directions on palimpsests,” explained Bergmann. “In the case of Galen’s work, the original text is vertical and the new text is at a 90-degree angle.”
After failed attempts to use multispectral imaging to see the important writing underneath the hymn, Bergmann proposed using XRF, which he had previously leveraged to decipher hidden mathematical theories in a copy of a work by Archimedes.
Using XRF, synchrotron X-rays knock out electrons close to the nuclei of metal atoms, and the holes are filled with outer electrons, resulting in characteristic X-ray fluorescence. However, XRF can be time-consuming and results in an immense amount of data. For example, a single scan of the Galen palimpsest takes about 10 hours and leaves vast amounts of nuanced data, so the researchers turned to machine learning.
William Sellers, an expert in data processing and the director of the University of Manchester’s zoology department, designed an algorithm that enhances the XRF signal. Rather than focusing on just one location, Sellers taught his algorithm to recognize different parts of the manuscript—including the undertext, partial text and the overtext—and measure spectra there.
“He trained the images to emphasize what is in this part of the spectra, which is the text,” said Bergmann. “Incredibly, after several iterations, the image did enhance.”
As shown in the photo to the left, this kind of data mining results in colorized images of the layers of text and parchment, giving scientists a much sharper image to analyze.
While the researchers are still processing the data and results, Bergmann was excited about the doors opened by the possibility of using XRF enhanced by machine learning for future projects.
“Next, we are thinking of going to Google next door to see if we can use non-linear algorithms,” he joked.
Top photo: credit Andy Freeberg/SLAC National Accelerator Laboratory. In-text photo: A sample page scanned with X-rays showing the never-before-seen under text (green), especially visible in the left margin. Credit: University of Manchester/SLAC National Accelerator Laboratory