DN> Not so. The earliest FORTRAN compilers used the BCD interchange code
DN> to hold "Hollerith" constants. The IBM S/360 used EBCDIC instead,
DN> and FORTRAN Holleriths on that architecture were (and still are)
DN> stored in EBCDIC. The Hollerith punch codes themselves were seldom
DN> used internally on any machine, and since card readers are very,
DN> very rare these days that can now be taken as "never".

     But all "EBCDIC" is is Decimal BCD Hollerith
"Expanded" into an
electronic, Hexadecimal/Binary interchange code.  What was being
interchanged if not the punch card data?  If EBCDIC was not designed to
be directly compatible, then why was the EBCDIC alphanumeric character
set left in decimal, which created the famous "hexadecimal holes".  Such
a bizarre design does not make much sense unless the designers were
trying to directly accomodate the decimal Hollerith card punching
machines and the BCD decimal Hollerith Census data....

--- ViaMAIL!/WC v1.60d