Tag Archives: Digital Humanities

X-Rays Expose a Hidden Medieval Library

Readers of this blog probably know that early-modern book bindings contain hidden treasure: fragments cut from medieval manuscripts, ranging from small snippets to full pages. The fragments were placed inside bindings to reinforce the bookblock and to provide support for the boards (see this post I wrote about it, and this one as well). This recycling process – plain-old slicing and dicing, really – was common practice, old-fashioned as handwritten books had become after the invention of print. In fact, medieval pages are found in as many as one in five bindings of printed books from the sixteenth and seventeenth centuries.  While the stowaways are normally hidden from our eyes, we sometimes get to meet them face to face when a binding is damaged (Fig. 1).

Fig. 1 – Leiden, University Library, 583, printed work (16th century) with medieval fragments inside (12th century) – Photo EK

But what to do with the thousands of fragments that are hidden from us in bindings that are still in pristine condition? This simple question became the drive behind the development of a method to examine fragments without removing or damaging the bindings. The method, which was presented this week, encompasses medieval book history (executed by me) and Macro X-ray Fluorescence Spectrometry (developed by Joris Dik, Delft University).This post gives you an exclusive look behind-the-scenes at how we managed to digitally leaf through invisible pages and gain access to a hidden library.

The plan

Leiden, University Library, fragment from BPL collection - Photo Julie Somers
Fig. 2 – Leiden, University Library, fragment from BPL collection – Photo Julie Somers – Source

In the spring of 2014 I was asked to write a short piece about ideas or approaches that could potentially change a scholarly discipline, even if they were not yet feasible.  I wrote about how we might be able to access a hidden medieval “library” if we were able to gain access to the thousands of manuscript fragments hidden in bindings (Fig. 2). Take their carriers– printed books whose bindings are enforced with the fragments – and give them a ride on the luggage belt at Schiphol Airport in Amsterdam. Let’s give them a bit of x-ray love, I suggested, half jokingly (here is the piece I wrote).

Then I remembered that a fellow member of The Young Academy, Joris Dik, developed an x-ray technology that enables researchers to look through paintings, in search of the earlier stages of the composition. Joris and I secured funding through The Young Academy to transport his Macro X-ray Fluorescence Spectrometry (or MA-XRF) scanner to Leiden University Library, an institution that greatly supported the unusual kind of research we were planning to undertake. We dubbed our endeavour the “Hidden Library” project and on October 4th, 2015 we started firing away at early-modern bindings in Leiden University Library.

The theory

Rembrandt, Old Man with a Beard (left) and self-portrait (right) hidden underneath
Fig. 3 – Rembrandt, Old Man with a Beard (left) and self-portrait (right) hidden underneath – Composed from images in this source

The Macro X-ray Fluorescence Spectrometry (MA-XRF) technology was developed by Joris Dik and his team at Delft University, in collaboration with industrial, academic and museum partners. The machine was designed in such a way that it could be transported to a museum in a few crates. There it would be used to show hidden paint layers in paintings by Rembrandt and other old masters. For example, in collaboration with partners in Antwerp Joris Dik successfully showed an unfinished self-portrait by Rembrandt, which was hidden underneath a painting of an old man (Fig. 3) – here is a more recent Rembrandt discovery. A thin beam of X-rays is used to scan the object, charting the presence and abundance of various elements below the surface. Theoretically, the technology ought to make it possible to show medieval inks as well, even when they are covered by a layer of parchment, paper or leather – the most common materials that hide medieval binding fragments from our eyes. But would it?

The practice

The MA-XRF-scanner developed by Joris Dik and his team at Delft University
Fig. 4 – The MA-XRF-scanner developed by Joris Dik and his team at Delft University – Photo EK

The answer to this question came very quickly after we hit the switch: yes. However, interdisciplinary research often comes with complications. Yes, we see text, but no, we could not read it. And so a series of experiments were undertaken by the team (which also included two research assistants, Anna Käyhkö and Jorien Duivenvoorden). For one thing, we learned that the distance between the head that released the x-ray beam and the fragment in the binding was key to our success (Fig. 5). A device was built that allowed us to adjust this distance between head and fragment with half a millimeter (the image was taken before it was completed).

The head of the MA-XRF scanner working at the 16th-century binding of Leiden, University Library, 617 F 19
Fig. 5 – The head of the MA-XRF scanner working at the 16th-century binding of Leiden, University Library, 617 F 19

Another variable we had to master was the length of time needed to scan the fragment. The head was moving back and forth in front of the scanner (Fig. 4), but how slow did this movement need to be? How long should the scanner scan before moving on to the next bit? When was it clear enough for me to read? Ultimately these nuts were cracked, meaning we were able to see a fragment through a binding. In fact, we managed to do it in such a way that the text was clear, legible and datable, as the following examples show (Figs. 6-8).

Example 1: fragment underneath paper
Fig. 6 – Leiden, University Library, 603 G 11 (16th-century): 15th-century fragment visible underneath paper cover – Photo Anna Käyhkö
Example 2: fragment inside a parchment binding
Fig. 7 – Leiden, University Library, 617 F 19 (16th century): 15th-century fragment seen inside a parchment binding – Photo Anna Käyhkö
Example 3: large fragment inside parchment binding
Fig. 8 – Leiden, University Library, 180 E 18: large fragment inside parchment binding – Photo Anna Käyhkö

We ultimately scanned around twenty different early printed books. This seems a modest amount, but the main aim of the Hidden Library project was to discover if we could indeed expose bindings to x-rays and reveal the medieval fragments inside. In that respect the project was a success.

Challenges for the future
Before we start thinking that this method will enable us, starting right now, to trace thousands of new fragments, we are forced to take a reality check. First of all, the second variable – of how long a binding needed to be scanned in order to reveal its hidden treasure – remains a practical road block between science and unveiling a medieval source that has never been tapped into systematically. The images you see in Fig. 6-8 were each produced with over 24 hours of scanning time. A shorter period makes fragments visible, but not legible, as seen in Fig. 9 (which shows the same fragment as in Fig. 6, yet exposed significantly shorter).

Leiden, University Library, 603 G 11
Fig. 9 – Leiden, University Library, 603 G 11, short exposure

Secondly, another problem is to distinguish between the text on the front and back of the fragment. Depending on the composition of the ink, it may or may not be possible to separate the two sides of a leaf. Iron, for example, will be returned from both sides of a leaf, producing a peculiar image like the one seen in Fig. 10. One has to look for elements that only show the side closest to the beam, which in this case is calcium. Consequently, this particular fragment is seen in its most optimal form when only calcium is shown, as is the case in Fig. 6.

Leiden, University Library, 603 G 11, iron
Fig. 10 – Leiden, University Library, 603 G 11, iron returns text from both sides of leaf at same time

So yes, the new technique is amazing in that it shows us fragments – medieval text – that we could otherwise never see because they are hidden behind a layer of parchment or paper. However, the method is not yet perfect in that it comes with challenges that need to be overcome, of which the long exposure time is the most significant. To stay within the analogy of this post, while the door to a virtual medieval library has been opened by the MA-XRF technique, we need to find a way to enter and take a look around as quickly and efficiently as we would in a real library.

The interdisciplinary research introduced here was executed by Joris Dik (Delft University) and Erik Kwakkel (Leiden University) in close collaboration with the University Library Leiden. Fundamental in our research was the help of the library’s conservator, Karin Scheper (who is, by the way, a guest blogger for medievalbooks). Two research assistants worked with us for the duration of the project: Anna Käyhkö and Jorien Duivenvoorden. The project was financed by De Jonge Akademie (The Young Academy), a branch of the Koninklijke Nederlandse Academie van Wetenschappen (The Royal Dutch Academy of Arts and Sciences). Two formal publications, providing a more detailed description of the research and its results, are planned. Various national Dutch news outlets have paid attention to the project (newspaper, radio [start at 9.50 min] and television [start at 20 sec.]).

My Week of Lecturing in Oxford

It is the evening of Thursday 27 February, 2014, and at the moment I am sitting in The White Horse being stared at by Inspector Morse, who frequented this pub back in the day – and who seems to have left a portrait behind every time he did. When I look out the window I can see the Bodleian Library, that treasure trove of medieval books. For the past week this is where I have been: Oxford, more precisely Corpus Christi College. Invited to be the 2014 E.A. Lowe Lecturer in Palaeography, I came to the college for a week to give lectures on medieval script. I learned a lot over the past week, both from the audience’s responses and through discussions with colleagues, both inside and outside Corpus Christi College – a most welcoming and hospitable community. This blog has everything to do with medieval manuscripts, but it is more personal than usual, if you forgive me: I thought I’d give a sense of what it entails, doing lectures in a historical place like this. So here is my week of being part of a stimulating academic world.

Corpus Christi College, Oxford (pic: my own)
Corpus Christi College, Oxford (pic: my own)

I arrived in Oxford last Thursday afternoon, just in time to hear @WillNoel do the annual McKenzie Lecture in bibliography in St Cross College, titled “Bibliography in Bits”. Will, Twitter celebrity par excellence, is a well-known proponent of Open Access and making digital data available for all to download and use. (“For free!” he would add to that in a loud voice.) He talked about digital surrogates of medieval books: how a manuscript’s digital representation is its own entity, how it exists – is – in its own right. He showed how a digital manuscript is a resource that teaches us things the material object itself may not reveal. His lecture (and the dinner that followed) formed a great start of my week here. It would also be the last thing I would do – until now, sitting in this pub – without a certain amount of pressure.

That pressure was not just generated by the venue, but also by the fact that the data at the heart of my lectures had just been harvested – with the indispensable help of my Research Assistant, RV. The paint of my lectures still wet, much of my free time was filled with going through my data and deducing how to expand the scope of my papers with their help. My three lectures aimed to show how the major book script of the Early Middle Ages (Caroline Minuscule, in use from c. 800 to c. 1100) morphed into the major book script of the Later Middle Ages (Littera Textualis, or Gothic script, used from c. 1200 to c. 1600). It’s a great topic because the century in between the two is filled with experimentation by scribes, of mixing older and newer features, and of fights for dominance between opposing letter shapes. My approach was threefold. First, finding a way to describe in objective terms what the actual difference is between the letter shapes in the two scripts. Second, registering in a database whether scribes in different ages and geographical areas preferred the Caroline or Gothic presentation of a letter. Third, translating this data into graphs that provide insight into the transformation from the one script into the other. Each step of my research came with challenges and limitations, but also with opportunities to advance our knowledge.

Title slide of my first lecture
Title slide of my first lecture

So, here we are, on the Friday: my first lecture. It focuses on how the transitional script from the Long Twelfth Century evolved over time. The theater is filling up nicely (about 75 people have come) and I start to do my thing. I discuss the method for about twenty minutes and then dive into the paleographical depths of the complex hybrid script. Highlighting differences between letter shapes I begin to carve out an objectified description of the road between Caroline and Gothic, focusing not on the overall impression but on hard, measurable features. I challenge the traditional temporal boundaries of the two scripts and feel brave enough to query, at the end of the lecture, whether we ought to perhaps abolish our notion of Caroline and Gothic being different scripts. Would it not be better to regard them as different expressions of the same writing system? My data certainly backed up this provocative idea. After the lecture a dinner was organized in the founder’s room of Corpus Christi Corpus: a great end of my first performance.

Then came the weekend, which I spent with family just outside Oxford – walks, pubs, and a newspaper on Sunday. On Monday I did an extra-curricular masterclass for the Centre for the Study of the Book, which runs under auspices of The Bodleian Libraries (here). It had been arranged only a few weeks earlier, within half an hour, and entirely through Twitter messages between me and @DanielWakelin1 – medieval books are so modern! In a seminar room filled with graduate students, faculty and nine medieval manuscripts I talked about two unusual book types: the elegantly tall and narrow holsterbook, and the off-cut manuscript, which is made from recycled strips of parchment. The students had picked out a selection of specimens from the Bodleian Library, adding to the class’s hands-on character. In the afternoon the Fellow Librarian of Merton College gave me a private tour through their medieval library, which is Britain’s oldest surviving library designed for use by scholars (I blogged about it here).

Medieval Library at Merton College (pic: my own)
Medieval Library at Merton College (pic: my own)

Then came Tuesday and my second Lowe Lecture. Having done temporal development last Friday, it made sense to focus on the enormous regional variation in the transformation from Caroline to Gothic. When you place a manuscript from Germany next to one from France you can sense that they are not equally advanced, but with a new tool I developed I could support such intuitive verdicts with a number. I introduced the notion of Gothic Weight, which measures the “Gothicness” of a script written between 1075 and 1225. It gives a value to each of the thirty or so features I track in my database: 2 points for Gothic, 0 points for Caroline and 1 point if the script trait in question is presented in a mixed form – meaning that the scribe uses both Caroline and Gothic on the same page. When the points are added up, you end up with one number representing how advanced or behind an individual scribe was with respect to his handwriting. It means you can compare a German and a French manuscript objectively, as well as comparing Germany and France as a whole (by taking the average Gothic Weight), or even map the increase of Gothiness over time within European regions.

Sculpture of a scholar, Bodleian Library (Pic: my own)
Sculpture of a scholar, Bodleian Library (Pic: my own)

Wednesday and Thursday were filled with working hard on the third paper, in which I aimed to show how Gothic script evolved not over time or in different regions, but as a novelty that was passed on to new generations of scribes. How did idiosyncrasy turn into new norm? My database did not provide an answer to this question, but using my data in combination with broader questions I could explore the very difficult question of a script’s dissemination – albeit without providing a definitive answer. The paper mainly focused on training, as I figured this was the moment when a new script feature had the chance to jump to a new user – a novice in a monastery being trained to write for the first time. In fact, such transmission would depend on the person in charge of training novices: if he was modern and advanced, then likely so would his pupils be. The main part of the paper therefore examined two cases where teacher and pupil were found on the same page: the first prompting the latter, monitoring progress, and correcting mistakes. It is exciting research given that you study, in a sense, the “homework” of a medieval scribe.

After the lecture I was invited for dinner in St Edmunds Hall and after that I retreated to the White Horse, where I am currently writing this blog. The bell for the last round has just rung, so I must be off – away from this pub and, tomorrow, from Oxford. One last time I will open the medieval gate of Corpus Christi College with my electronic key chain – a contrast that strikes me as a suitable parallel for my week of studying digitally the handwriting of medieval scribes.

More information – The full abstracts of my Lowe Lectures are found here. More about my approach to medieval script in this blog about kissing letters. Download this free book if you want to read an article I wrote on mapping script objectively.

Note: This post was originally written for, and posted on, my project blog MedievalFragments.