Hum, the OCR of that book is not perfect, I admit. BUT : that book, it was my copy of the book I scanned that book in 2001, these where my first experiments with OCR I did the error of scanning the book with 'replacements' which can create artifacts. The technique now is to simply scan the book, do the OCR after and acrobat will put a layer of more or less recognizable text in shadow on the back, it is hard to understand why a clean copy of a scan at 600 will generate errors but it does, usually 90 % of the words are recognized which is quite sufficient most of the time but with mathematics most programs won't be able to disassemble formulas properly, that problem is still not resolved... that book is available on a directory of mine on http://www.plouffe.fr/simon/math/ under ConcreteMathematics. That copy of the book you mention is probably the result of robots gathering information here and there, I know these robots they are always at my site (30000 hits a day), scrubbing the directories for information, especially Googlebots. by the way , type <control-D> when in acrobat and look for the signature. I know this is foolish to put copyright material on the internet directly available for robots, this is intended. I am patiently waiting that one day one of those editors will knock at my door. I did that with about 300 books, finally my HP scanner just broke but I had finish my collection of books. These books are available on the <dark internet> for about 10+ years and directly on my site for 4 years... Best regards, Simon Plouffe