On Tue, 2 Jun 2020 at 10:20, R.S. <r.skoru...@bremultibank.com.pl> wrote:
> As far as I know, a character set on punched cards was somehow limited, > so it is not EBCDIC or similar set of 256 characters. No, not at all. In fact in a way the opposite is true - a punched card column can contain way more than 256 values. There are 12 rows of potential holes, so 2**12 values, in theory. In practice: There are mechanical reasons to limit the number of holes punched in a column. WIth many holes the card loses strength, and also the punching action requires either one big bang, or multiple attacks on the same column - neither good for speed or relibility.. Even more practically, though, obviously that 2**12 doesn't fit in a single 8-bit byte. So it was normal for S/360 to map each column to a single byte (thus each card occupied 80 bytes when read), and the encoding used was one of two closely related ones: BDCIC or EBCDIC. These are to be seen on any S/360 Green Card. The rule is to permit only one punch in rows 1-7, so any combination in rows 12, 11, 0, 8, 9 (2**5=32) * 8 possibilities in 1-7 = 256. Some card readers supported Column Binary or Card Image mode, and in this case a card record was 160 bytes with each column mapped to the low-order 6 bits of two adjacent bytes. I think there were some other variations for this mapping. There were also at least two variants of reading hand-made marks on a card made with a pencil or pen: the older Mark Sense (marks made with a high-graphite conductive pencil and read electrically) and Optical Mark Read (OMR), read by optical reflection using a separate read station from the optical transmission one for reading holes. Some readers allowed for mixed holes and marks. Bitsavers has manuals for some of these card machines. > Was there any name for card character set? I mean something like "CP 037" or > so. Code Pages as we know them today have their roots a good deal later than punched cards. At least in the IBM mainframe world, they came from the 3270 devices, which were initially US-centric, with only upper case English (unaccented Latin) letters. Almost immediately local variants were field developed in many countries to provide characters needed. In many - maybe most - cases the character assignments clashed, and because of the 3270 addressing architecture, positions below X'40' were not available. There were also print trains (1403 or 3211) with extra or replacement characters, and these varied somewhat in their character mappings. A very good historic reference for how this developed is the 1989 SHARE "ASCII and EBCDIC Character Set and Code Issues in Systems Application Architecture" report by the ASCII / EBCDIC Character Set Task Force. To make their point, they used the short name "SHARE ÆCS Report". I have a scanned copy if you can't find it online somewhere. Parts of this became input into the design of UNICODE. Tony H. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
