Saving 1962(year) mainframe data

Thanks for your thoughts, yes I think that could work, though a bit slow for 90,000 cards. As some of the cards are printed, might need to scan them twice, once with a black background and once with white. I might use that method for the 160 column cards. For the 80 column cards, I expect to soon have a working card reader. Now I can get to the logic properly it should not take long to find out why the 'zone' is being decoded incorrectly. The 'numeric' is working fine already so I can read cards with columns of just '0' to '9' on them. One day I should even be able to read pure binary cards through the check reading brushes of the online card punch, though for protection I might remove part of the punching mechanism to stop it making more holes by mistake. I have even been toying with the idea of re-building my spare punch mechanism, putting some TTL logic in it and connecting it directly to my Mac laptop through a serial to USB converter. Might be easier than getting my IBM keypunch working without any diagrams for the relay logic. However, the question I was asking was how to encode these special characters. I currently know of two alternatives. One is to put them in pure text files with a two character sequence for the special characters. ACONIT in France have used a character starting Hex F as a marker. The lower nibble holds four of the bits row bits. This means even fully laced cards can be represented. The other is to put them in UTF8 form which has characters with rings around the numbers 10 to 15. I think they probably have the fraction characters, pound sign etc as well. On 14 Nov, 2008, at 15:01, cctalk-request at classiccmp.org wrote: