Readwyn 0.1.0

How volunteers turn scanned books into free ebooks

A volunteer at Distributed Proofreaders gets one scanned page and the text that OCR software read from it on a single web page, corrects the text and sends the page back. The next volunteer gets the same image with the corrected text. A page usually goes through three such proofreading rounds and two formatting rounds, and then one person joins all the pages into the finished ebook. That's where most new Project Gutenberg books come from: Gutenberg's home page names the site as its main source, and so does the site itself.

Charles Franks started it in 2000 to help Project Gutenberg, the site's history note says. It became an official Project Gutenberg site in 2002 and a separate legal entity in May 2006. On 1 November 2026 the counters on the same page showed 51,469 finished titles, 106 of them posted in October, with 638 books still in the proofreading rounds, 2,213 in their final checks before assembly, and 1,247 volunteers active in the past 30 days.

From a printed book to an ebook

The site's walkthrough for new volunteers and Project Gutenberg's own pages give this order:

  1. Copyright clearance: Project Gutenberg's copyright page asks for scans of the title page and the back of it (the verso) and promises an answer within a few days. Only a book that is out of copyright in the United States gets through.
  2. Content: a volunteer supplies the page images and the text that OCR software read from them.
  3. The rounds, each done by a different volunteer with the scan on the same screen as the text. In proofreading round 1 (P1) the raw OCR text gets corrected; P2 runs the spellcheck and compares again, and P3 looks for the small errors that two rounds missed. Formatting round 1 (F1) adds the markup, some of it set by rules for that one project, and F2 removes the formatting errors that remain.
  4. Post-processing: one volunteer assembles every page into a single e-text.
  5. “Smooth Reading”, which is optional: anyone may download the book and read it as a reader would, without the scans, and mark what looks wrong.
  6. Post-processing verification: another volunteer checks the post-processor's file before it goes to Project Gutenberg.

The rules of a page

The rules shown to P1 proofreaders keep the book as it was printed. A proofreader keeps each line ending where it falls in the scan, but joins any word that was split across two lines. Old spellings stay as printed. Each paragraph starts after a blank line with no indent, and stray spaces that OCR put around punctuation come out. If a proofreader isn't sure, the page should match the original, with a note in the form [** note] for the next volunteer or the project manager.

The walkthrough sends a new volunteer to P1 or to “Smooth Reading”, and its last page promises that at least two more experienced proofreaders check every P1 page again before it's done. Newcomers are asked to start on a project marked BEGINNERS ONLY and to do only a few pages of it, five or so; a mentor reviews those pages and sends feedback after a few days, though the sample project's comments warn that it can take a week. After 5 to 15 pages on such projects, the P1 page asks them to leave those projects to newer people and move to one marked EASY. The home page suggests a page a day but asks for no commitment beyond doing your best.

After the rounds

Once the post-processing verifier signs off, the book goes to Project Gutenberg, whose volunteer FAQ, in its advice to producers, says that most of them make the HTML version first and derive the plain text from it. Gutenberg's ebookmaker software makes the downloadable ebooks from the HTML files that volunteers submit.

The 50,000th book to come out of this process was itself a book about printing: William Savage's A Dictionary of the Art of Printing, a printer's manual published in London in 1841, which Gutenberg released on 6 December 2025. The site's blog marked the milestone the next day.