Q&A Wednesday: Firkovich’s Bible manuscripts, with Vince Beiler

Vince, what are you working on at the moment? 

Right now, I’m going through several hundred classmarks in the Firkovich II B collection, looking for joins between manuscripts. These are biblical manuscripts acquired by the noted Crimean Karaite collector Abraham Firkovich in the 19th century and subsequently sold to the Russian National Library in St Petersburg. While the classmarks collected by Firkovich number nearly 15,000, the II B collection is nearly one-tenth of that size, containing some 1,500 classmarks, many of which are Bibles that date to the 10th–13th centuries. 

In the II B collection, I’d estimate that well over 80% of the classmarks show an ‘Oriental,’ i.e., non-European hand. The remaining manuscripts are mostly Sephardi (Spain, North Africa). There are several Ashkenazi (French/German) manuscripts, and maybe one or two Italian manuscripts, but these are extremely rare. 

You’re looking at images, obviously, rather than the originals in St Petersburg. 

Yes, I have three screens side by side – I’m, hoping to add more soon – and I’m viewing the images either on Ktiv, the manuscript website of the National Library of Israel, or on Friedberg. The images are all in black and white. I search the classmarks by sorting according to column and line numbers, after which I separate them according to script type and general appearance. When it comes to this kind of work, a picture is always worth a thousand words. 

Several manuscripts have not been digitized, however, and countless others would benefit from an in situ examination. If COVID doesn’t create additional complications, I’m hoping to make a trip to St Petersburg in the near future. 

Why are you looking for joins? Are you trying to reassemble codices? 

I'm not trying to reassemble codices per se. I’m gathering data to compare Masoretic note collocations between codices. In brief, I enter all the notes I can find from certain reference ranges (let’s say Genesis 39–43) across the corpus into a database, and then run queries to see which manuscript note batches are the most – and least – alike with the remaining manuscripts. With enough data, it turns out that one can construct a rough manuscript stemma that does not rely on vocalization or paleography (both profitable lines of inquiry, but with their own sets of weaknesses) for manuscript categorization. Comparing Masoretic notes provides us with an external means for corroborating or calling into question analysis done using other methods. 

However, there are many manuscript irregularities in the II B collection, making my job a bit harder. At the outset, I unthinkingly assumed that one classmark equaled one manuscript. But this is NOT the case. I’ve found as many as five manuscripts in one classmark. In one instance I had three classmarks interconnected in a weird sort of love triangle. Of the c. 1500 classmarks in the II B collection, then, perhaps there are only several hundred Bibles. I’m still trying to work this out. It’s obviously important that I don’t confuse the data by wrongly attributing specific folios to the wrong codex. 

Are most of the manuscripts you’re looking at just single leaves? Or are they more significant chunks of books? 

Most classmarks are probably only two or three folios in length (the median), although if you averaged all the pages (the mean) you would probably get 20 or 30 pages per classmark. This is because the longer manuscripts sometimes contain several hundred pages. 

Can you tell us a bit more about the connections between this collection and the Cairo Genizah fragments? 

No one knows exactly from where Abraham Firkovich acquired the manuscripts that now make up the II B collection (he wouldn’t say), although he undoubtedly spent time in Egypt and Eretz Israel. You would think that personal Bibles, particularly those that appear to be little more than school exercises, would be the least likely of all Bibles to travel from their place of origin, yet we find these Bibles, highly similar in all respects, in both collections. Additionally, as the II B collection tends to have longer, i.e., less fragmentary works, it is reasonable to assume that Firkovich chose only the best, leaving the rest for Solomon Schechter and his colleagues. 

There are probably even joins between some Bible fragments in the T-S collection and the Firkovich II B collection, but to the best of my knowledge, no one has worked on this, at least not systematically. (I am, or rather I hope to, but it will take some time.) Nonetheless, it is my understanding that joins have been found between other works collected by Firkovich and the T-S collection, although I’m not familiar with the details. 

What is your next step after you have gone through the Firkovich II B manuscripts looking for joins? 

My main goal (my PhD thesis) is simply to sort these and other early Bible codices according to a stemma. In particular, I’m looking at the Masora Parva (Mp) notes in the manuscripts, as these notes are the easiest thing to compare vis à vis a database. (Masora Parva are the shorter scribal notes found in the margins of the Bibles.) There’s a certain amount of variation between these notes, and I have yet to find two Mp note recensions that match in every point. If we were to compare the codices by Mp numeral (the marginal number that comment on the main text), there is c. 80% similarity on average. In a similar manner, if we were to compare the codices by Mp string (the word or words of the main text that are being commented upon), the usual amount of similarity is around 60%. But these numbers do vary quite a bit. For example, the Aleppo Codex has an Mp note collocation that was widely copied/was widely circulated. The Mp note collocation of the Leningrad Codex is somewhat more novel, and that of Sassoon 1053 is more unusual still. In short, there is very much of an established centre and a periphery in Mp note distributions, and my aim is to categorize the 10th–12th century Bible codices accordingly. 

Is your ultimate aim to track the development of the Mp notes? 

In an ideal world, maybe. But my present goal is simply to sort the manuscripts typologically rather than to posit which codices have earlier Mp note recensions. It’s a bit like doing textual criticism. Typological sorting has broad consensus, but as soon as you posit which recension was earlier, better, or is less corrupted, there are a host of problems that are difficult to resolve.

Thanks for your time, Vince!

Vince Beiler is a PhD student at the University of Cambridge.