A team of researchers, using beams of x-ray light, hope to reveal the writing locked away within 2,000-year-old charred papyri from Herculaneum.
For more than 200 years scholars have been attempting to read what remains of heavily charred scrolls rescued from the library of a luxurious villa destroyed during the eruption of Mount Vesuvius in 79 AD. At that time, Herculaneum’s buildings were buried under layers of ash, and the hot gases turned the library into an oven in which the abandoned scrolls were flash-heated and transformed into fragile rolls of carbonised papyrus resembling lumps of coal or burned logs.
Rolled Herculaneum Papyri
Excavations of the site started in 1709 but were abandoned. In 1980 around 1,800 scrolls were discovered, finding homes in the Biblioteca Nazionale in Naples, the Institut de France in Paris, the British Library in London, and Oxford’s Bodleian Library. The question for scientists now is how to read something that is illegible, and locked away in a complex matrix of tightly rolled, blackened and compacted material that is too fragile to touch, with ink that quickly fades once exposed to air.
Herculaneum
Traditional means of physically opening these scrolls result in extensive and irreparable damage to theoriginal materials. Over the centuries, these methods have included everything from immersion in water to pouring mercury through the rolls to suspension in various gases. The current common practice is to carefully cut away the hardened outer layers and then to attempt to separate and unroll the more flexible central layers. Even under the best of circumstances, much of the text will be lost through this process, and the final result will only be a fragmentary record of the original.
The first stage uses the Diamond Light Source synchrotron to probe the scrolls with beams of x-ray light. These very bright x-rays can penetrate the scrolls and map their internal structure in a completely non-invasive and non-destructive way. The x-rays pass through the scrolls and are detected on the other side to create a series of two-dimensional images that can then be stitched together to create a three-dimensional model of the scroll.
Prof Seales is director of the Digital Restoration Initiative at the University of Kentucky (US), a research program dedicated to the development of software tools that enable the recovery of fragile, unreadable texts. According to Seales,
Diamond Light Source is an absolutely crucial element in our long-term plan to reveal the writing from damaged materials, as it offers unparalleled brightness and control for the images we can create, plus access to a brain trust of scientists who understand our challenges and are eager to help us succeed. Texts from the ancient world are rare and precious, and they simply cannot be revealed through any other known process. Thanks to the opportunity to study the scrolls at Diamond Light Source, which has been made possible by the National Endowment for the Humanities and the Andrew Mellon Foundation, we are poised to take a tremendous step forward in our ability to read and visualize this material. The scan session promises to be a key moment in our quest for a reliable pathway to reading the invisible library.
MICRO COMPUTED TOMOGRAPHY, DATA ACQUISITION, ANALYSIS AND VIRTUAL UNROLLING
The second stage involves the fragments of papyrus that are the result of previous efforts to physically unroll the papyri. These fragments, which have writing on them that is visible to the naked eye, are photographed, scanned and the results fed into computers.
Micro-CT has only become commercially available in the last decade, but it is well suited to the study of objects such as papyrus scrolls. The scans provide a non-destructive method to visualize the interior of the scrolls, without the necessity of physically opening them. CT can be immensely helpful as a tool for planning and condition assessment. It shows fracture lines within the scrolls, as well as other damage that may be hidden within the object.
Micro-CT slice showing the scroll structure surrounded by the cast polyurethane container. These images are “end-on” views at about the midpoint. At higher resolution (right) the interior portions show surprising, unpredictable structure.
Complex computer algorithms (machine learning) are used to teach the computer to recognise the subtle structural differences between the inked and blackened areas (such as differences in the structure of papyrus fibres).
After the computer has been trained on these fragments, the idea is to apply it to the data from the intact scrolls collected at Diamond Light Source and, hopefully, reveal the text hidden within.
3D structure from micro CT scan. Scale shot with writing (photograph) fixes the scale expectation for the small “unwrapped” portions (15mm square), which show papyrus structure as vertical and horizontal stripes. Changes in intensity are from a multi-power, registered scan.
Given a high enough resolution scan, and the ability to segment layers within a scroll, a computer simulation can virtually unroll a scroll, revealing its original form. Assuming contrast between ink and papyrus, the ink features can then reveal the original text, without necessitating an invasive and damaging physical unrolling. Scientists have developed such software and algorithms, which have been tested on a number of proxy objects. However, the Herculaneum papyri present unique challenges to this system.
External laser-based surface scan supports a 3D proxy used to make a mold. This mold is used to create a polyurethane container to hold the scroll in a vertical orientation inside the micro-CT scanner.
We have collected the first scan of Herculaneum papyri using micro-CT, allowing visualization of the interior of these scrolls without the risk of major damage to them. These scans show the internal structure of the rolls, including fissures, fractures, and air gaps. This knowledge could inform any physical intervention or conservation plan for these papyri. Initial data analysis has been slowed by the complex nature of this internal structure. Automated segmentation for separating layers within the scrolls has been virtually impossible. Manual segmentations have yielded success in viewing small regions of the scrolls, though no ink has yet been visible.
(a) The alphabet letters from A to N (alpha to nu) as revealed by the XPCT experiment, primarily from the innermost region of the papyrus, where the individual coils are more distinguishable, are reported on lines 1 and 2; on line 3 the infrared images of the same letters from the unrolled papyrus PHerc. 1471, which was used as a reference for the writing style of the scroll PHerc.Paris. 4. Printed capital letters of the ancient Greek alphabet are given below line 3. (b) The alphabet letters revealed by the XPCT experiment from Ξ to Ω (xi to omega).
Fortunately, little to no damage was done to the scrolls by this non-destructive scanning process, thus preserving the scrolls for future study. Based on CT scans of similar objects, we know that our “virtual unrolling” method has merit, and if a procedure can be developed to increase ink contrast, another scan has the potential to reveal much hidden text.
