The Jubilees Palimpsest Project

SBL Boston, November 20, 2017

1 The Jubilees Palimpsest Project is kind of a baby-brother of the Sinai Palimpsests Project. We’re several years behind but otherwise share the same family of technologies and personnel. The main technological innovation is the incorporation of texture imaging, which allows us to interact with digital facsimiles to gain a more complete experience of a manuscript as more than just a text container. We also have the luxury of being fully open with all our data. The Biblioteca Ambrosiana has been wonderfully supportive of making images of its artifacts freely available to scholars, and with the support of the National Endowment for the Humanities, we have been able to make all our data and tools available for free.

2 Earlier this year, four of the people on this panel, Michael Phelps, Keith Knox, Roger Easton, and myself were joined by Damianos Kasotakis, Dale Stewart, Ken Boydston, and Anthony Selvanathan for a month of data capture at the Biblioteca Ambrosiana in Milan. 3 The Ambrosiana isn’t quite as old as St. Catherine’s but it is more than four hundred years old. 4 It was founded as part of the Catholic Reformation that centralized learning in urban centers and acquired manuscripts from rural monasteries 5 such as the Bobbio monastery founded by Columbanus more than fourteen hundred years ago.

Data captured 2017

6 The manuscript that gave name and the original purpose to the Jubilees Palimpsest Project is the fifth-century copy in Latin of the book of Jubilees, originally written in Hebrew in the middle of the second century BCE. Even though Jubilees was among the more popular works found at Qumran, the fourteen or so copies are very fragmentary. Most scholarship relies on the much later copies in Ethiopic. The Latin version copied in the fifth century is not only much older, it is an independent family. 7 Our knowledge of Latin Jubilees had depended on the 1861 edition by Antonio Maria Ceriani. The condition of the manuscript has since degraded, such that subsequent scholars have not been able to make improvements on Ceriani’s edition, at least until now. 8 The Jubilees Palimpsest Project captured all 96 surviving pages of Latin Jubilees, and reconstructed the codicological structure of the original. Advanced processing of the data is still underway, but most images are available online now. Preliminary studies indicate that the images available do allow us to read letters where Ceriani did not even offer a guess. 9 More importantly, it is becoming clear that Ceriani sometimes projected his expectations onto the manuscript and offered reconstructions without qualifying them as such. It some cases it is clear that Ceriani’s reading is wrong. Going forward, it is possible and necessary to check Ceriani’s readings against the manuscript images before making an argument based on them.

10 Latin Jubilees was joined by another text attributed to Moses called the Testament or Assumption of Moses. This work is known only from the sixteen pages in the palimpsest, so Ceriani did not have an alternate edition to guide his expectations. Coincidentally, he was able to read less of it, even though the scribal properties and condition are identical. Our first effort was to read the initial three lines of the work, which might offer some hope of agreeing on a title. It seems likely that the opening lines were written with red ink to signal the beginning of a new composition in the collection of writings attributed to Moses. It is more than possible that no trace of that ink remains, or if it does we have not been able to isolate its spectral signature. 11 We also examined “Taxo,” the mysterious name, most likely gematria, of the eschatological hero of the work. The X is unusual in that the forward slash crosses low. Where the eye wants to see the upper-right stroke it is really seeing the overtext. Overall, Ceriani’s reading of X remains likely enough, but if someone thinks an A would make sense, the evidence fits that possibility as well. The problem is that neither Taxo nor Ta’o makes sense as a name or word in any relevant language, so the next step is to reevaluate the arguments for gematria with both possibilities in mind.

12 The 144 surviving pages of the palimpsest were made from 96 pages of Latin Moses and 48 pages of an Arian Commentary on Luke. (These were originally separate volumes with different quire signatures and column structure, although the collocation and paleography make it likely that Latin Moses was preserved by Arian Christians. We can only speculate as to whether Arian Christians in Italy would have interpreted the creation of angels in Jubilees as the creation of Christ, as did commentators in Ethiopia.) The Commentary on Luke is significant in its own right, as the voice of Arian Christians in their own words, rather than from their surviving opponents. It would also be useful to compare this Latin commentary on Luke from the fourth century to the anti-Arian commentary by Ambrose around the same time and place. It stands to reason that one is responding to the other, but that work remains to be done. These pages are also considerably more readable than Latin Moses. The advanced images may have been overkill, but serve as an easy introduction to navigating advanced imaging interfaces. It may be hoped that the images will fuel renewed interest in the text and the scribal features of the manuscript itself.

13 After imaging our flagship palimpsest, we still had plenty of time to sample pages from other palimpsests in the remarkable collection. Among these were ten pages of Origen’s Hexapla. Benjamin Kantor helped us identify significant pages for capture and processing. His interest is in the pronunciation of Hebrew as reflected in the Greek transliteration. Our images provided evidence for several corrections to Mercati’s edition. 14 Among these is a case where a Greek letter iota appears on the manuscript, but spectral processing indicates that it is only reagent drawn on by Mercati, not a mixture of reagent and ink as we see elsewhere on the page. This reagent is a chemical designed to enhance traces of ink in a palimpsest, at least in the short term. This is not to say Mercati intentionally falsified the evidence, but that his determination to find a trace of a letter led him to create it where there had been none.

15 We also captured twenty pages of Wulfila’s fourth century translation of Paul’s letters into Gothic. This is our earliest evidence of the Gothic language and the translation of Mediterranean religion into northern Europe. The pages we sampled also include a liturgical calendar. Hugo Mendez has been helping us but these pages are in need of further study by scholars familiar with Gothic or old Germanic languages. 16 We also sampled pages of an otherwise unknown Greek commentary on Luke, 17 and a palimpsest that remains completely unidentified. 18 We also recovered some letters in a cryptic script hidden in an illumination of Petrarch’s Vergil. As far as I know the significance of the letters has not been identified. You can see a theme here that much work remains to be done. My point is not to persuade you of my laziness, but rather to invite involvement from all of you and your students and colleagues. The project is radically open in every way. I’ll say more about that shortly.

Texture Imaging

19 First, I’d like to describe the main technological innovation from the Jubilees Palimpsest Project. As Roger Easton will explain, spectral imaging is almost as old as photography itself, long before the current generation of computationally intensive digital spectral imaging came of age earlier this century. Texture imaging has a parallel but separate history, from pre-digital efforts with raking light to computationally intensive techniques using Reflectance Transformation Imaging, or RTI. RTI began at Hewlett-Packard labs as an effort to capture real surface textures for 3D models. This original intent remains promising but unrealized. RTI took off in cultural heritage circles for imaging objects for which texture is the primary conveyor of meaning, such as cuneiform tablets, coins, and inscriptions. 20 Bruce Zuckerman and West Semitic Research pioneered use of the technology with the creation of the InscriptiFact Digital Image Library.

Like spectral imaging, texture imaging relies on the premise that the camera and object remain still for a series of controlled captures in which only the lighting changes. For spectral imaging, the wavelength of light changes. For texture imaging, the direction of light changes. For spectral imaging the many captured images are processed to create an image with highly accurate or enhanced color. 21 For texture imaging the many captured images are processed to create a dynamically relightable texture map. This basically means the user can interact with the image by manipulating a virtual flashlight. The capture techniques are complementary. Spectral imaging is concerned with color and uses diffuse light to eliminate distraction from texture. Texture imaging formerly paid no particular attention to color and utilized variability of highlights and shadows to capture the surface. At least with retrospect, the combination was not that difficult. It was inevitable once we identified objects for which we wanted to capture and enhance both the color properties, and texture properties.

22 I believe there are many such objects, but my particular concern was with manuscripts. Given the rest of the session I need not spend time talking about why spectral imaging is useful for manuscripts, but let me say a bit about why texture imaging is valuable. First, texture can help us recover unreadable text. This occurs if the ink turned acidic and corroded the surface of the parchment before it was erased. We can sometimes read the outline of a letter in the corrosion of the surface of the parchment. It also helps us when texture shows that a spot is not ink, but rather a spec of material above or below the surface. The texture of the scribal scoring lines also tells us where to expect text. In the case of dry-point notation, the texture is the only trace of the text. (Dry-point notation is the medieval equivalent of using a pencil in a library book. It’s not as destructive as ink, but allows the reader to make little notes presumably for the same person to refer to later.) 23 Second, texture helps us get at the study of a manuscript as more than just a text container. It is increasingly appreciated that manuscripts are artifacts of scribal culture in their own right. Texture can give us otherwise unavailable information about how a manuscript was produced and used. The more scribal practices are systematically studied the more the data can be used to answer fundamental questions of chronological and geographic origin of an instance of a text. Last but not least, texture is fundamental to all those ineffable nuances that lead some people to assert that there is no substitute for first-hand experience. I think most of what we do with a manuscript, much more than just stare at it, has to do with texture and interactivity. 24 The way we move our heads or the light to catch clearer glimpses can largely be captured and recreated, even enhanced, with texture imaging technology. When I talk about texture closing the gap with first-hand experience, I don’t just mean as a tool for discovery in the sense of learning something that no one has known before, but the kind of discovery that might happen in a classroom or in the library. I would like to share with my students experiences of encountering ancient literature embodied in artifacts, experiences much more primary and primal than a critical edition.


25 That brings me to the next major emphasis of the Jubilees Palimpsest Project. We have worked very hard to make everything we capture and create as accessible as possible to as many people as possible. Access covers a number of points, including making the resources available, discoverable, easy to use, and legal to reuse for further research and teaching. I think of the project as changing the way we think about primary sources, such that we are less dependent on the intervention and judgment of a single editor, and get closer to the experience of literary artifacts as they were traditionally encountered.

First, when I say that absolutely everything is fully available online as we work, I say that more as a disclaimer than a boast. Our work can be thought of as an “open-beta,” meaning the work in progress is available to anyone as we are working. Not everything has a final coat of polish, but I don’t know if there will ever be a day that we say there is no more work to do here.

26 The most exciting innovation in access has to do with open standards for interoperability, especially the International Image Interoperability Framework. Interoperability is important for freeing our information from the silo of one website. The images in the repository can be accessed and reused in other contexts and combined with images from other repositories. Not only images, but annotations too can be aggregated across repositories, facilitating sideways research. That is research not focusing on a page or manuscript at a time, but on a scribal or literary feature that cuts across many manuscripts. Open standards for annotations mean humans and machines can gather disparate information for creative study. When we do construct and publish our arguments, the reader will not be asked to take our word for it. 27 Images and links will be woven into a new kind of critical edition in which an editor presents the lighting and color processing that supports the argument, and the reader can evaluate alternative views immediately.

28 Access means much more than being on the Internet and interoperable. The interface has to be easy for the user, including users who do not get excited about technological wizardry for its own sake. Interoperable standards give users a choice between interfaces. Some are easier to learn but limited in their power, while others have a steeper learning curve and more overall power. I was driven to develop a new, but still interoperable, interface based on my concern that no information should be hidden. 29 This interface, the IIIF Navigator focuses on use cases when there are many images describing a page and the student will spend considerable time on that page moving between the information available, as opposed to a “page turner” interface that makes it easy to flip through a book like a Kindle. This interface also relies on the browser and operating system to manage tabs and windows for greatest flexibility.

30 Then there is that small matter of legality. Most of us are like pirates in one way, probably only one way, in that if we find something on the Internet that we want to use in our teaching we just take it. The more established and powerful aggregators and publishers of information are more concerned with legal liability. It is important not just to have liberal access policies, but to follow clear and standardized licenses that are widely used and understood. In particular, Creative Commons makes it easy for others to understand what they can legally do with your data. In our case, the metadata, annotations, and tools we create are very open. Only if you intend commercial profit from the property of the Ambrosiana is it even necessary to ask permission, and even then I have found them to be very generous.

Opportunities to get involved

31 With my final minutes I would like to describe some opportunities for you and you students and colleagues to get involved. First, all our images are online for scholars to study and interpret. There is far more there than I or anyone I know will have time to study and publish.

32 The resources are available through the IIIF Navigator on the project website. This includes an image cube of many different lighting angles and color enhancements, so that you can focus on a region and quickly flicker between available views without losing your place. I think you’ll find no one view is sufficient for getting oriented and discovering new or different letters. The Navigator also provides links to the WebRTI images, which allow you to click on the light bulb to move the virtual light around the image in real time. This level of interactivity is good for gaining a sense of the condition of the parchment and answering questions such as whether a particular mark is ink or something else on or in the parchment. In the case of Latin Moses (including Latin Jubilees and the Testament of Moses) we have gathered various transcriptions and translations available, from Ceriani up to the new readings generated by the project. At the moment, there is not much available in the way of new readings, which is where you come in. As you and your students generate knowledge about the transcription and other features of the manuscript, I strongly urge you to contribute your discovery back to the Project or publish them as interoperable annotations on other sites.

If this seems a bit daunting now you might be interested in a scholars’ workshop planned for the first half of 2018. This will be an in-person meeting, building on the traditional medium of human interaction. In the spirit of what Keith Knox was describing, we envision an opportunity for scholars to learn what the images and software can do for them now, and also offer feedback for what could make the image processing and interface more useful to scholarship and teaching. A date and location has not been settled, but it will be announced on the project website and on my Twitter account, @thanneken.

Since this is SBL I’ll say less about the next opportunity to get involved, but if it just so happens that you or your students have advanced interest in image processing, that is not just viewing our images but creating your own, all our data is available. 33 This data archive is not for the faint of heart, but all the processing can be done using free tools such as ImageJ and the SpectralRTI_Toolkit. 34 If you are interested in the more technical side of capturing and processing your own Spectral RTI images, the NEH grant provides for training in the following year, 2018–2019.

35 Finally, I would extend an open invitation to any institution interested in collaborating on a vision for digital archaeology, particularly focusing on the many untapped collections in northern Italy. There is far more work to be done than there are people doing it, and we are by no means territorial. My vision for the future of digital archaeology, especially discovery and publication of illegible manuscripts, draws its inspiration from traditional archaeology. Archaeology was never cheap or easy, but institutions valued research and teaching made possible by sending teams abroad to gather data for further research back home. As we build sustainability, the costs will go down. Already I can look any dean in the eye and say that compared to other programs that involve new technology and international discovery, digital imaging of unreadable manuscripts is a bargain, especially for the impact on students and institutional profile. As work scales up, we will no longer be limited to imaging known palimpsests, and will have more time to focus on discovery of palimpsests that have not yet been accurately cataloged. If this vision gains momentum, future meetings of SBL will not just include more specialists excited the way scholars of Jubilees and the Testament of Moses are excited now. There will be new literature and new questions that can be asked across data repositories using emerging tools for inquiry. 36

Superscript numbers refer to slide numbers

  1. Title slide
  2. Team photo
  3. Ambrosiana facade
  4. Ambrosiana treasury
  5. Ambrosiana bindings
  6. 4Q216 and Latin Jubilees 2017
  7. Ceriani painting and Ceriani 1861
  8. Codicological structure
  9. Letters not read by Ceriani
  10. Testament Moses page and first three lines
  11. Taxo two images and paleography chart
  12. Arian Commentary on Luke and readable detail
  13. Origen, Hexapla full page
  14. Iribi with accurate color and Pseudocolor of reagent
  15. S36sup full page
  16. Greek commentary on Luke
  17. Unidentified, with readable detail
  18. Broad view and detail of Petrarch
  19. Samples of significant texture
  20. Video of InscriptiFact
  21. Second RTI video
  22. Manuscript texture, holes, scribal scoring lines, dry-point notation
  23. Full page raking
  24. Video of manuscript RTI
  25. Project website
  26. IIIF logo and Mirador index
  27. Mercati argument
  28. IIIF Navigator all resources view and navigate view
  29. IIIF Navigator four corners tabs with layers
  30. Creative Commons licenses with permissions box of website
  31. List of manifests in UCLA Mirador
  32. IIIF Navigator navigate view
  33. Data archive
  34. GitHub page with ImageJ logo
  35. Ambrosiana Treasury
  36. Bookend title slide