Overview

What is this site?

This beta site is made to help folks dig just a little bit deeper into books they’re curious about.

Have you ever seen an old book and just had a hunch there’s something interesting about how it was put together? What if the color of its paper–yes paper–offered important clues?

On this site, developed by researchers at Carnegie Mellon University, you can plug in some details from online rare books repositories and examine the color profile of every page in the book.

Maybe your book was printed in two different shops? Maybe it was printed in two different eras? Maybe someone inserted some illicit sheets after it had passed the censors?

If so, chances are you’ll see those kinds of things in the data.

Give it a try.

Instructions

This tool currently works from IIIF-manifests from the following rare books repositories:

LibraryManifest Format
British Libraryhttps://api.bl.uk/metadata/iiif/ark:/[Piece 1]/[Piece 2]/manifest.json
Cambridgehttps://cudl.lib.cam.ac.uk/iiif/[Piece 1]
Folger Shakespeare Libraryhttps://luna.folger.edu/luna/servlet/iiif/m/[Piece 1]/manifest
Harry Ransom Center at UT-Austinhttps://norman.hrc.utexas.edu/notDM/objectManifest/[Piece 1]/[Piece 2]/
Harvardhttps://iiif.lib.harvard.edu/manifests/[Piece 1]
Internet Archivehttps://iiif.archivelab.org/iiif/[Piece 1]/manifest.json

IIIF-manifests for each of these libraries takes the specified form, where the part(s) in [ ] brackets are unique identifiers.

  1. Locate the unique identifier(s) for your book and input them in the tool on the "Input Manifest" page.
  2. After a short wait, you’ll see a page that randomly mashes a sample of your book’s pages together. This is normal! The purpose is to help you find a spot in the resulting image that’s just paper (no text).
  3. Use your mouse to draw a small rectangle in a region that is likely to be blank on most pages. What’ll soon happen in the background is that it’ll calculate the average color at these coordinates on every single page in book (manifest, really). So you want it to be as uniform as possible (apples to apples, etc).
  4. Click “Use this rectangle”
    View demo of rectangle selection [opens in new window]
  5. The next step is an optional filtering step. Maybe there are binding images or endpaper images you don’t want. Maybe you’ve accidentally captured a catchword or a signature mark. These will throw off your averages. You can delete or modify your rectangles one by one by clicking the delete button or the edit button on an individual image.
  6. View demo of rectangle adjustment [opens in new window]
  7. Click “View Plot” or “Extract Color Information.” The plot on the left is keyed to images on the right. Click a point and see its associated image.

  8. Explore LUV, RGB, median, and mean permutations.

  9. Download your data. The tool doesn’t currently store your history, so you won't be able to come back to this set of rectangle selections. Downloading your data is at present the only way to document your results and make them persist.

Find Something Interesting? Please share!

We love to hear about interesting examples! We’d even love to use some examples in our papers. If you find something cool, by all means let us know. You can email us at cnwarren at cmu dot edu or tag us on Twitter @chrisvvarren or @sammuellemey.

For Libraries

Want researchers to be able to use this tool for your collections? Make sure your IIIF platform has CORS enabled and let us know where we can find your manifests via the email addresses above.

Disclaimer

This tool is currently unpolished and a bit unstable. (Max wrote it while sitting up with a sleepless infant, with all of the careful planning and attention to detail that implies.) We welcome feedback about bugs or potential improvements, but are also currently limited in the time that we have available for development. If you are comfortable with React and would like to contribute, let us know!

Credits

Developed primarily by Max G’Sell with input from Samuel Lemley, Christopher Warren, and Matthew Lincoln as part of the Print & Probability research project. Intro and instructions by Chris Warren.