Back in the '30s, my grandmother typed my grandfathers, mmm, novelized account of his time with the AEF in France in WWI. 132 pages. Toward the end, the typebar characteres were getting a little clogged with ink and the paper had yellowed which made OCR more difficult. A few years ago, I scanned all of the pages into PNGs. It took some trial and error adjusting the color levels (gamma, brightness, and contrast) for Tesseract to do its thing well enough. In short, I had to retype around 10 pages, another 10 in part, and a number of errors in other pages.

On a new print, it should have few errors. It can even handle some hand-written text.

Clearly, integrating Tesseract with FairScan was not an easy process.