In 7-level papertape codes not all punches are data. There are additional
punches that ensure that data is read correctly. This is done in two
ways, parity (the number of punches in a row is odd) or number (the
number of punches in a row is always the same). In most cases a row with
no punches or with all punches on was ignored when it violated the
checking pattern. As far as I know two of the first kind are used and two
of the second kind:
Odd parity, bit 7 is parity bit
Bit 7 is the "100" bit. It will not be shown in the following code
displays. I know of the following codes in this class:
RCA 501 (Bizmac)
The Bizmac code was fairly sane. The reason for the placement of the
digits in the middle of a row is simplification of BCD arithmetic.
LGP-30 Flexowriter
The LPG used this code. The arrangement is fairly strange, and I have
no idea for the reasons, except the placement of the letter L. Because
the lowercase form serves a dual function as digit 1 in this code, it's
placement is where one would normally expect that digit.
RPC-9000
A precursor of the following code. It still lacks uppercase letters.
It is strange that there is no symbol on the uppercase of 8.
RPC-4000
A suprisingly reasonable and complete code.
Alwac III Flexowriter
This code has strange allocations for the letters, although they make
slight sense when you consider card codes.
1103A Typewriter
This (and the following) are clearly based on the
CCITT #2 5-level code.
Note that the digit 1 (and his small equivalent) have no allocation.
Flexowriter 7-level
This code is clearly base on the previous one, but now the digit 1
(and his small equivalent) is included.
Soroban
Again a renewal of the previous codes. Amongst others the small
digits disappeared and are replaced by other symbols.
Odd parity, bit 5 is parity bit
Bit 5 is the "020" bit. It will not be shown in the following code
displays. Actually of course, any bit could be said to be the parity
bit, but the coding makes it clear that bit 5 is the one added to
the basic code. I know of the following codes in this class:
Hidac 101
A small code with a curious placement of the letters.
Correspondence
The code used in some Telex operations in the US. An extremely
strange arrangement of letters and digits can be found here.
PTTC/6 (PTTC/BCD)
One of the papertape standards from IBM. This one was to match
the BCD card codes, as can be seen by the arrangement of the
letters.
PTTC/8 (PTTC/EBCD)
This is a followup, intended to match EBCDIC. Curiously, there
are fewer symbols in this map!
MC Flexowriter
The code used at our institute for the preparation of Algol 60
code for the computer. It was fairly customized for that purpose,
but the basis was a matching of the letters with the letters in
the BCD card codes.
PDP-1
A variation of the previous scheme, used with the PDP-1, but with fewer,
and different, symbols.
Exactly 3 bits
I know of two codes that used a 3 out of 7 code, code tables are given in
full, incorrect entries are hatched:
Moore 3 out of 7
It is clear that a 3 out of 7 code is not particularly dense. This coding
has 35 possible code positions out of a possible 128. As with 5-level
codes the traditional LetterShift and FigureShift take their place to
accomodate all letters and digits plus a few assorted symbols.
CCITT #3 (3 out of 7)
This is the third CCITT standard code, the second is a
five bit code, the fourth is a
six bit code. This code is based on
the Moore code given above.
Exactly 4 bits
I know of only one 4 out of 7 code, the code table gives incorrect entries
hatched:
CCIR 476 (4 out of 7)
The CCIR 4 out of 7 code is based on CCITT #2.
In almost all cases you get this code by augmenting the original 5 bit
CCITT code. A CCITT code with 2 bits has a one bit prepended and a one bit
appended, a code with 3 bits has a one bit prepended and a zero bit
appended and a code with 4 bits has a zero bit prepended and appended.
You may look for the exceptions. Note, due to the way the codes are
presented here, you should bit-reverse the codes in CCITT #2.