[GreenKeys] Model 29 ASR

[Jim Haynes]

When I was living in the Chicago area I got a couple of machines by way
of Michigan Bell, but I don't remember how it was arranged.  Probably
Bert Prall had something to do with it.

One of these, which I still have, was the Delta Airlines version of the
28ASR which I have written about several times in the past.  The other
I don't remember if I got the whole ASR or just the cabinet, but it was
one of the "pink" cabinets like Model 29s were made in.  I converted
that to an 8-level machine using Model 35 parts.  I never did like the
Model 35 cabinet anyway, and this one I just repainted dark brown and
put the M35 stuff inside.  (As they say of certain artists, this is when
I was in my Dark Brown phase.)

The Model 29 as we know it was not EBCDIC; it was 6-level IBM BCD code,
a forerunner of EBCDIC to be sure, but only 6-level and upper case only.
As you have no doubt read it was originally developed as a 6-level
up/low case machine to replace the Model 20 in Teletypesetter service,
but apprently nobody wanted to buy any.  (Marketing goof on Teletype's
part, I guess.  By then hot metal typesetting and Teletypesetter were
on the way out.)  So it was reworked a little with the IBM BCD code for
internal Bell System use in data processing.  It was officially called
the Model 28 Integrated Data Processing set.  Bell didn't want any of them
to escape into the world at large.  One reason I'm told is that it would
seem they were playing favorites with IBM and ignoring the other computer
companies which had slightly different versions of the BCD code.  That
reason seems bogus, because it would be trivial to modify it for some
other BCD code.  The more substantial reason was that development of
ASCII was under way and Bell didn't want anything that would compete with
ASCII to get out there; they wanted to insure that ASCII would be the one
standard.  In the meantime the military had Fieldata and I believe that
was what got used in AUTODIN, though no Fieldata teleprinters were
released - AUTODIN came with an electronic box that could convert Baudot
or BCD to its own transmission code.

EBCDIC was a late-comer in the code development.  During the development
of ASCII, IBM kept putting up objections that were unacceptable to the
other companies participating in the standard development.  The main point
of contention is that Hollerith punched-card code has some gaps in the
alphabet, with other characters in the gaps.  BCD does also because it
is really just a simple translation from Hollerith.  And IBM considered
this so important that they wanted to require any new standard code to
retain the property.  In this they were wrong, since 512 bytes of memory
would provide conversion tables between any two 8-level codes and there
is no real need for a conversion that can be done by trivially simple
logic.  But IBM has always had a penchant for solving the problems of
the previous generation of equipment, like they say the generals are
always fighting the previous war.  The good guys insisted there should be
no breaks in the alphabet, so that sorting would not be upset by the
funny characters in the middle.  So ASCII had to have an unbroken series
of binary numbers representing A through Z.

IBM surprised everyone by suddenly withdrawing their objections to ASCII,
and the standard was adopted.  Then IBM announced their new System/360
computer line and EBCDIC.  EBCDIC has the breaks in the alphabet that
they wanted.  I assume IBM thought they could use their dominant marketing
position to establish EBCDIC as a de facto standard and that ASCII would
then wither away, with only some legacy services such as TWX using it.
Since the U.S. government required ASCII in all its computer systems
thenceforth, IBM put an ASCII mode bit into System/360, which made some
very minor change to the way the system operated, and which was so
unimportant that it was dropped from System/370.

I believe what turned the tide in favor of ASCII was the Model 33.  Not
in its intended role as a TWX terminal, but in the totally new role of
a computer remote terminal and minicomputer console device.  The
minicomputer and time-sharing businesses had no need for punched card
input; and there was nothing to compete with the Model 33 in DataPhone
service as a remote terminal, nor hardwired to the computer as a local
console.  What we now call the mainframe computer companies really had
no means of attaching cheap start-stop terminals to their systems; they
preferred schemes in which a single terminal adapter polled multiple
terminals on the same line for messages.