Imaging the Duke papyri

by Peter van Minnen
In the course of the first year of the project it was decided to make digital images of the papyri instead of photographs or slides. The advantages of making digital images directly from the originals seemed obvious at the time. The imaging could be done in-house, the images would preserve the dimension of depth as no photograph would, they would allow manipulation afterwards, which could serve research as well as teaching, and they could be shared with other institutions over the internet. The imaging would de done in the second year of the project by the librarian hired for the purpose in addition to making catalogue records. It was expected that this would entail an extension of the project with a third year, but this seemed also indicated by the amount of time it would take the papyrologist to deal with the remaining number of papyri.

For the imaging a flatbed color scanner was bought in September 1993. Color was deemed superior to grayscale, especially since the images were supposed to provide the archival backup to the collection. Using color would of course increase the size of the files. The scanner bought was a UMAX 800, a 24-bit scanner that allowed scanning up to 800 dpi. A digital camera was considered too expensive at the time. It was not even clear whether reliable digital cameras were already available commercially. The financial advantages of a flatbad scanner over a digital camera were more important than the disadvantages: the lack of a light source behind the papyrus to light up holes and the clumsiness of scanning upside down. Since the Duke papyri would be scanned after they were framed, however, the last-mentioned disadvantage seemed trivial.

Experience has shown that scanning through glass has another disadvantage. The effect of the glass on the registration of color by the scanner is quite noticeable. This also goes for using a digital camera on papyri framed in glass. During the first year of the scanning project the color spectrum of the scanner was not calibrated. The color of the scans was corrected manually and only approximates the colors of the originals. The second year, a special calibration tool was purchased that goes some way to correct for the glass, but it turned out to be unavoidable to manually correct the scan. Taking papyri out of the frame for scanning purposes would not be feasible for those papyri that consist of multiple fragments or are too fragile to handle upside down. Scanning unframed papyri seems impossible if it is done by an outside vendor.

Originally it was not clear whether potsherds, wooden boards and lead tablets could be scanned at all, but these three-dimensional objects posed no problem. Their surfaces are after all rather smooth. It is the fibrous structure of the papyrus itself that turned out most difficult to capture electronically. The UMAX 800 proved better at it than some other flatbed scanners, but it has been replaced in October 1995 with a Sharp JX-330, which scans much faster and records colour more accurately than the "old" UMAX 800. The angle of the light source used inside the scanner determines how the fibres will show up on the scan. The scanner reads the image in three times and corrects it as it goes along.

The computer used for the scanning is a Macintosh Quadra 800 with 1GB of memory, 32MB of RAM and a 24-bit color video card. An accelerator board (Storm) was also installed. The program used to generate the scans was Adobe PhotoShop version 2.5, later version 3.0. This seemed quite adequate for both the production of the scans and the use of the images in research and teaching. As a backup digital tape(s) are used as well as an external hard drive (DataStor), acquired in February 1995. When the project is finished the tapes with the archival scans will be kept available at Duke through the computing center. The derivatives will be stored on a server and remain universally accessible in combination with the catalogue records and other supporting materials.

Originally the color scans were made at 300 dpi, but after about one year of experimenting this was raised to 600 dpi. This would create much larger files, but the cost of storage media is dropping fast. The Advanced Papyrological Information System, an undertaking of several American universities holding substantial collections of papyri (Columbia University, Duke University, Princeton University, The University of California at Berkeley, The University of Michigan, and Yale University) has adopted 600 dpi as the standard for digital images of papyri. The reason for using 600 dpi is simple: to read original papyri papyrologists currently use microscopes that enlarge 4x, 5x, 7x or 10x; 600 dpi images can be blown up 8x on a 72 lpi screen, the kind most commonly in use today. With a digital camera a consistent 600 dpi rate can be achieved by keeping the camera at a fixed height on its stand.

It was decided early on to adjust the quality of the image through a minimum of manual operations. The color of the scans is off for reasons stated above and this needs to be corrected in the archival scans. To this end the histograms have been manipulated minimally through Adobe PhotoShop. To increase the contrast of the writing, the most important consideration for specialists, the target audience most likely to use the archival scans, the unsharp mask function has been applied consistently.

The data are saved in different formats. The archival master set consists of 600 dpi scans stored in TIFF with LZW compression. Another set of 600 dpi scans are stored with JPEG compression. These scans will be used in research and teaching at Duke and at other institutions on request. Two derivatives are made of each 600 dpi scan: a 150 dpi and a 72 dpi scan. Both are stored in GIF and made available over the internet. The relatively small file sizes should cause no problem. The 72 dpi scan will show up life-size on a viewer such as Netscape using a 72 lpi screen. The 150 dpi scan will show up twice enlarged or can be blown up to twice its size, which allows the more difficult scripts to be read with ease. But even the 150 dpi scans will not provide enough detail to deal with problem spots. This will, however, only be a problem for specialists interested in a particular papyrus, and they can always request a copy of the 600 dpi scan. It seemed more serviceable to all parties concerned (specialists and non-specialists) to have images of all Duke papyri available for quick consultation all the time.

A review of a book on the application of digital technology to ancient manuscripts such as the Dead Sea Scrolls appears in the Bryn Mawr Classical Review. A more technical description of digitizing papyri has been prepared by a team of Finnish scholars working on the carbonized papyri from Petra in Jordan. Another application of digital technology to the study of Demotic papyri has been described in a congress paper by Janet H. Johnson.

Return to the papyrus home page

Return to Duke University Special Collections Library home page

Please see our page with contact information if you have any comments or questions about the Duke Papyrus Archive.

Last updated by Peter van Minnen on 12/4/95