Ways Forward

Over 750,000 images were returned from our machine learning initiative. Now questions need to be asked of these, and relational databases established to allow for maximum functionality in searching. In other words, having produced and finessed the training data, corrected and returned errors, the 'real' medieval research can get underway. Yet the two years of the funded project were only enough to get to this preliminary stage. Information, then, is not knowledge; and data is not meaning. Our efforts to interpret the data will take some time, and we'll update the results as we go. Suffice it to say, that cursory scanning of the elaborate intials--the Litterae notabiliores--in the galleries has produced new connections between manuscripts through similarities of rare forms of H and T (plus, now, other production similarities). Previously unlocalisable manuscripts might now be assigned a potential place of origin. These galleries are open to all scholars and interested participants. Please feel free to find your own research paths using this material. We can be cited as Elaine Treharne, et al., 'Stanford Global Currents' (Stanford University, 2014--), globalcurrents.stanford.edu