Renaissance Magazine has an interesting article in the December, 2007 (#58) issue on how ancient books are being digitized for on-line libraries. Normally books are scanned, but the photo-like images take up lots of file space. If converted to text, the files are smaller. An additional advantage, for the researcher, is that they can then be searched.

However, converting old books to text is challenging because of the irregularity of the printing. An interesting solution makes use of the garbled letters test given to an estimated 60 million internet users a day. By incorporating snippets of a scanned book into the tests—and feeding the “translations” back to the digital library—a digital text version of the book is created.