I think my OCR has more problem with "her" kitties' pawprints than with the coffee stains. Accuracy seems to be inversely proportional to the length of the document scanned. Something like Number of Errors is proportional to the 1.3 power of number of characters.

I have learned that when scanning newspaper clippings it's almost an absolute necessity to put a BLACK backing paper behind the page when you scan, to cut the bleedthrough from the other side; but one of the cats has a taste for my black backup sheets, and keeps chewing on them, so there's never one handy when I need it.

Most printer paper is thick/opaque enough to get a clean scan without backup. I do run into the problem of my OCR trying to read multi-column text as graphics, so I usually Photoshop the scans to single column if it's anything large.

Now is this thread drift?

John