In response to Dick Greenhaus's latest posting:

There's another option -- it's called "Hidden text PDFs". They appear to be visible scanned images on the surface with hidden OCR'd text behind them. The OCRing isn't perfect, but it's searchable, so you get perhaps 90% of the usability of a perfectly proofread document at a fraction of the cost. The human reader sees the original scan, letting her or his brain make corrections on the fly that the OCR software missed.