So I found and bought "The New Student's
Reference Work", a little
encyclopedia in five volumes, published in Chicago in 1914. As it
was published before 1923, it is now in the public domain. Since
this non-Scandinavian work doesn't fit in Project Runeberg, I put it
in Wikisource.
First I scanned images (300 dpi JPEG) of all 2791 pages and uploaded
them to Wikimedia Commons, where you will find them in the category
http://commons.wikimedia.org/wiki/Category:LA2-NSRW
Then, for each book page, I created a wiki page on Wikisource
displaying the scanned image and containing the raw OCR text. If you
want to help in proofreading, use two separate browser windows to
open the enlarged image and edit the wiki text.
I don't mean this to sound particularly harsh, but I'm wondering why
we're doing this on Wikisource, when Distributed Proofreaders for
Project Gutenberg already has a well-debugged workflow for taking
texts from images to OCR to proofread to a final version. Is there an
advantage to starting our own project that does the same thing they
already do pretty well?