After many hours of CT scanning with the SkyScan 1173, we now have complete data sets for both Herculaneum scrolls at the Institut de France. From this data the entire internal structure of the papyrus can be seen: folds, breaks, creases. This exceptional look into the interior of the rolls was captured in a totally non-destructive way, leaving the rolls undisturbed for future analysis.
Through these scans, we have acquired a wealth of data. For example, each x-ray projection is a 16 bit image at 4000 x 2000 resolution. Computed Tomography requires projections of at least 180 degrees or 360 degrees for a cleaner reconstruction. By collecting an x-ray projection every 0.2 degrees, the data set ends up with 1800 projections over a 360-degree rotation. This is roughly 18 MB of data per projection times 1800 projections which covers one section of the scroll. The scrolls were divided into five zones to capture data over the whole length. Five zones, 1800 projections per zone, at 18 MB per projection gives about 170GB of raw data. This is the input to the reconstruction algorithm. Once reconstruction takes place, the 3D slices are generated: 8000 slices, each 16-bit at 4000×4000 resolution. This yields another 250GB of data for a total of 420GB. This amount of data is for one scan. The team has captured several scans of each scroll under varying parameters.
The scanning itself is just the first step; once the data is liberated from the physical handing of the scrolls themselves, it can take on a life of its own as teams push forward to analyze it and understand its significance. This kind of “virtual analysis” is a remarkable step forward, lessening the need to handle the scrolls and opening up the possibility for analysis to a wider community.
With concentration first and foremost on the safety and careful handling of the rolls, the work has been methodical and slow. This means that the secondary analysis – the search for ink – will come after the team gets back to the lab in Kentucky. The structure of the slices clearly shows that the segmentation of the visible layers into meaningful, flattened regions will be a very big challenge. The team will be working to “digitally unwrap” promising sections in the search for a way to tease out an image of the writing itself, which is the ultimate goal. It is impossible to tell at this stage if this will be successful.
For now the team celebrates the successes of the day: plenty of promising data, never-before-seen views of the internal structure of two intact rolls, and a very low impact in terms of damage and risk to the objects themselves. The scientists are happy, the scholars are hopeful, and the conservators are relieved.