[Telecom] TTY 33 and 35 case and cover composition?

As an aside, AT&T (through Western Electric) owned Teletype for many years, I believe since the 1930s.

I believe the 'bodies' of Teletypes were steel with a glass insert. The model 33 had a liftable cover which I suspect was plastic.

How does one differentiate between actual Bakelite and plain plastic?

While we're on the subject, was the model 33 impact impression adjustable? That is, how many good quality carbons could be made?

Many schools used model 33s to each comp sci in the 1960s/70s. Could a school have used a spirit "ditto" master form (the kind with the heavy blue carbon sheet underneath, not mimeograph) to duplicate a terminal session? You need to press hard on those to get a clear copy, and even then the results were faint. [But before cheap Xerox machines, that was the cheapest way for short runs (under 50 copies). Long runs required the mimeograph machine which meant cutting a stencil with a typewriter; they were messy! The spirit duplicator used a fluid to copy off the backside of the master, the mimeograph bled the ink through the cut stencil.]

I got the sense that TTYs did not have a hard impact impression. We always used the pulp paper (for us, they gave us newsprint). When the TTY ribbon was too faint, I took them home and got plenty of more use of them on a regular manual typewriter.

***** Moderator's Note *****

The Model 33 case was plastic, not Bakelite, which was too prone to chipping and cracking to be used in a high-vibration environment such as a Teletype machine. Earlier machine, such as the venerable Model

15, had metal cabinets.

The Model 32/33 impression strength could be adjusted only slightly, but it was pretty high to begin with: carbon forms with up to two copies could be used without trouble. The problem with carbon forms was not the impression depth, but the "shoulder" impressions made by adjacent characters on the typewheel used by the Model 32/33. The Model 28/35, which had a type pallet with individually-struck slugs for each character, could be used for thicker forms, with more copies.

Many schools _DID_ use the Model 33 as a terminal during the start of the computer revolution, and that gave the 33 an undeserved reputation for unreliability. School administrators, who weren't expected to know any better, and system salesmen, who did know better, economized on the terminals to close sales. The Model 32/33, however, wasn't designed for service as a computer terminal, but rather for use in TELEX and TWX networks: it was rated (IIRC) for 1500 "shaft" hours before major overhaul. In TELEX or TWX service, that translated to years of reliability. Computer users, especially children in computer labs, placed much higher demands on the machines than they were designed to serve.

Bill Horne Temporary Moderator

Please put [Telecom] at the end of your subject line, or I may never see your post! Thanks!

We have a new address for email submissions: telecomdigestmoderator atsign telecom-digest.org. This is only for those who submit posts via email: if you use a newsreader or a web interface to contribute to the digest, you don't need to change anything.

The 33s used in my high school and the central computer center had continuous use during school hours but as I recall very reliable. I also recall that they had an excellent reputation. By the 1970s mini- computer makers sold the 33 as an I/O interface and I think they were extremely common in that service.

Originally they rented the machine from the phone company (about $100/ month), that machine had the built-in dial, speaker, and modem and push button control panel on the right side. (We had to get used to an all-number dial, in those days all our phone numbers were still

2L-5N). The phoneco provided service.

Later, they purchased a fleet of machines and hired a technician to service them. Instead of the built-in modem these had either an acoustical coupler or interface to a separate modem. Instead of paying $$$ to a commercial time sharing service they bought their own computer (HP-2000)

But the machines seemed to work well.

Except in the case of one kid, who had a habit of eating raisins while he was on the machine. Service was needed and the repairman (from Bell at that time), came out and dug out a large pile of raisins from the inside. A large component required replacement. He told the teacher that the company would cover the repair cost this time but not in the future.

As an aside, one rich kid rented his own Teletype for a month or so. A $100 back then would be about $500-$600 today. That kid turned into an early hacker, and today is a big shot security consultant.

The Western Union Technical Journal archives of this newsgroup, contains an article on the introduction of ASCII and the new model

1. (The model 32 was designed for Baudot use). See:
   formatting link

Clearly the 33 was designed for computer use as well as telegraphy and I suggest the designers did an excellent job especially given the very limited computer time sharing capability when they began their task. The amazing thing here we all are, 45 years later, using pretty much the protocols developed way back then. Newspaper articles in that time frame envisioned all this; that some day homes would have a terminal and connecting to supermainframes. see:

wrote in message news: snipped-for-privacy@x9g2000yqk.googlegroups.com...

Bakelite is phenol-formaldehyde and it is a thermosetting plastic, i.e. it does not melt. Another test is to touch a hot soldering iron to an unobtrusive area on the inside of the part. Bakelite will not melt but it will char releasing a very distinctive phenol odor like a burned out resistor. :-(

Ham radio operators who first used radio teletype systems usually used military surplus Model 15 and Model 19 machines from WWII. Later, Model 28 machines were used too. These were all Baudot 5-bit machines. All capital letters were used, and to get more than the 32 possible characters, two keys were dedicated to FIGS and LTRS. To type a number for example, you first pressed the FIGS key and then the key with the number on it. Then you pressed the LTRS key to go back to typing letters. Often operators would forget and garbage was printed leading to modifications like "unshift on space." At the end of a line, you pressed the return key, then usually a few LTRS keys to allow for the physical time for the print mechanism to return to the beginning of a line. Then you pressed a key for a new line (LINE FEED) unless you wanted to print over the line you just printed. Again this was often forgotten. The Model 33 was an eight bit ASCII machine. The old Baudot machines were asynchronous machines with a start and stop bit; sometimes the stop bit was one and a half times the length of the other bits. Different speeds were used, but amateur radio operators settled on the 60 speed (45.5 baud with a 22 millisecond bit).

Bill is quite correct about the weight of these old behemoths. The military Model 15 came on a HEAVYweight stand with cast iron legs. The Model 19 came with a table that seemed to weigh a ton! While in graduate school at Clemson back in the early 1970's, a math professor friend was donated a Model 19 and table. The only problem was getting it from Ft. Jackson in Columbia, SC, back to the professor's home in Clemson. The three of us had an old beat-up pickup, so off we went to Columbia. We found it easy enough. It was waiting on a loading dock - covered in Cosmoline and dirt. Some kind Sergeant was nearby and he got several recruits to load it on our truck. We tied it down to keep it from moving, a real waste of effort as that beast was so heavy it couldn't move. The truck was so weighted down that its suspension nearly bottomed out and we felt every pothole. We got back to Clemson, and tried to get it out of the truck. After much effort, the three of us and the professor got it off the truck and to his back door. THEN HE TOLD US IT NEEDED TO BE CARRIED UP THE STAIRS TO WHERE HIS HAM SHACK WAS LOCATED! So we called in more help and three more club members arrived. It took the seven of us about five hours to remove the Teletype from the table, dismantle the table, clean off the grease, lug it up the stairs piece by piece, and reassemble everything. This "junque" used metal so thick that it could take a direct hit with an M-1 and still keep clanking away. Hams refer to old vacuum tube radios and other military surplus as "Boatanchors," and believe me, the Model 19 was one.

The Wiki page on teleprinters has a partial history and some good technical information at

A booklet, "The Teletype Story," with lots of pictures may be found at Now, nearly forty years later, I can remember the fun we had with these military surplus machines, but I can also remember the pain and soreness that resulted from our crew of seven stooges trying to carry a Model 19 up those stairs! By the way, I had a Kleinschmidt teleprinter, the lightweight unit.

73 (Morse code abbreviation for Best Regards), Dr. Barry L. Ornitz WA4VZQ snipped-for-privacy@charter.net [transpose digits to reply]

Bakelite is hard, dense, and brittle, smooth on the surface but (as I recall) somewhat crumbly inside if broken. I have never seen any that wasn't dark colored. The plastic covers of the Model 33 Teletypes that I remember are uniformly gray and smooth-textured and would probably snap rather than crumble if broken.

On eBay, the antique dealers tend to say "Bakelite" for any plastic object more than five years old. Or so it seems! They don't know what they're talking about. They also say "Cold War Era" for any artifact from 1930 to 1995. And then there are people who think "the fifties" spans the whole twentieth century :)

I forgot to mention something very interestin the WU Tech Review ASCII article (see above for link). WU said it did not think ASCII would have much of a place in the WU network since it wasted three bits. Although at that time WU was very interested in serving as a data transmission carrier, choosing to avoid ASCII, in my humble opinion, condemned them to second rate status.

If you get a telegram from me "SENDING YOU 500 DOLARS", you'll deduce I meant "DOLLARS", even if the transmission line dropped an 'L'. But computers don't have that luxury, they need the protection of parity detection (one of the ASCII bits).

Further, computers require more printing characters and special control characters which Baudot had no room for.

Computers did not store data in Baudot. Converting telegraph transmissions to computer code required complicated mechanisms to insert or detect the FIGS and LTRS command codes and act accordingly. Even if a computer is EBCDIC instead of ASCII, it needs to make only a one-to-one conversion. As we know, many computers were ASCII based.

Thoughts?

[Public replies, please]

Well... I don't think anyone uses current loop or answer back drums any more.

I worked in a Western Union Teletype refurb shop in Mahwah, NJ in the late 70's. The division was called Western Union Data Services. We would squeeze all kinds of controllers, complete with meltdown drip pans, into the pedestals of the 33's towards the end to get them to match up with "modern" equipment. At the time model 33's were with a bank to bank funds transfer network called bankwire, and [they were also used by] the tobacco industry. I still saw a couple of TWX units go out the door every couple of weeks.

Among other duties, I went to the Skokie, Il plant for model 40 printer training. I started in the disassembly room on the Model 33 side of the shop. About 95 percent of the 33's needed new plastic covers due to wear and tear or discoloration from heat or sunlight. As said in prior posts they were much lighter built than the older models. I got h*ll once from management for deviating from the official test tape verbiage.

After two years I was bored to tears of brown foxes and lazy dogs. RYRYRYRY

Steve

73 de N2UBP

***** Moderator's Note *****

Steve,

Since you worked on 33's, then I bet you know why "RY" was not used.

Semi-ultimate trivia question: what was the test sequence used in place of "RY" on Model 33's, and why?

Ultimate trivia question: what was the key combination used on the Model 33 to punch a "leader" tape, having only feed holes without any other bits punched out?

Bill Horne Temporary Moderator

Please put [Telecom] at the end of your subject line, or I may never see your post! Thanks!

We have a new address for email submissions: telecomdigestmoderator atsign telecom-digest.org. This is only for those who submit posts via email: if you use a newsreader or a web interface to contribute to the digest, you don't need to change anything.

Thanks for the information.

Regular Teletype users quickly learned about LTRS/FIGS as well as return-line feed.

Computer users often only typed return and the host computer returned the line-feed upon receipt of the data.

Often computer users prepared a punched tape of their program in off- line mode, then transmitted it to save on connect time charges. We had to include a rub-out after return to give the carriage time to return to the side, otherwise the first character of a new line would not be at the margin. (Rub out was 8 bits punched, used to overwrite tape errors)

ASCII TTYs was also asynchronous, with a start and stop bit. I believe the stop bit in ASCII was also 1.5 times the length.

***** Moderator's Note *****

Here's some more trivia:

1. What was the equivalent of the ASCII "Rubout" key on Baudot machines?
2. What was the relative length of the stop bit on Model 33 machines?

Bill Horne Temporary Moderator

Please put [Telecom] at the end of your subject line, or I may never see your post! Thanks!

We have a new address for email submissions: telecomdigestmoderator atsign telecom-digest.org. This is only for those who submit posts via email: if you use a newsreader or a web interface to contribute to the digest, you don't need to change anything.

The Model 32/33 was designed for TELEX and TWX service. When I attended the maintenance school in 1976, the instructors told us that Teletype provided a "private line" version of the machine only because some customers were still using private Teletype networks for sending orders to warehouses or to collect sales figures from dealers. They made it clear that these were light-duty machines, not to be used as printers or for keypunch. That was why they came with current-loop interfaces: it was the standard private line interface, and computer hookups needed either external RS-232-to-current-loop converters, or a modem that included a 20ma interface.

I disagree: Western Union's fear of undermining their lucrative TELEX and TWX markets did them in, since competitors leapfrogged them by offering "bring your own modem" services which could carry any kind of traffic while WU tried to keep their existng TELEX/TWX customers in a closed system.

ASCII's parity bit proved ineffective for error checking on computer data lines: it could only detect single-bit errors, but the noise encountered on voice-grade data lines was as likely to "flip" multiple bits as it was to kill just one, so data transmission networks had to combine block transmission with cyclic-redundancy checks to assure reliable transit. A noise burst that flips bits 5 and 6 would change "500 DOLLARS" to "900 DOLLARS".

The Model 33 machines I saw used as computer terminals were as likely as not to have the parity bit forced to continuous "marking": it caused premature wear on the bit 8 punch in the tape unit, but the computer programmers didn't want to hastle with parity errors and didn't think the effort was worth their time.

Computers, per se, didn't require them: peripherals did. Most of the control characters in ASCII were intended for use with automated typesetting equipment and high-speed printers, which needed form control characters, such as form-feed, to work efficiently.

Converting _telegraph_ transmissions to computer code probably required an operator who knew how to use a Morse code key and sounder ;-).

Kidding aside, I don't think the ASCII vs. Baudot conversion was that big an issue:

1. Baudot was used only on military and TELEX circuits, both of which were, by their nature, separate from the computer world. After PC's were established, Western Union did offer it's "Easylink" service, which could send and receive messages to and from the TELEX (and TWX) networks, but they controlled the conversion and customers didn't have to be concerned with it.
2. The Telnet and Tymenet services, both of which were X.25 based, offered transparent code conversion for almost all common coding systems. I had an Anderson-Jacobon 841 terminal, which used EBCD (not EBCDIC) coding, and Tymnet allowed me to connect it to timeshare systems and bulletin boards without trouble.
3. The computer revolution ramped up so quickly that most users never saw a Baudot machine, and the few who did have to deal with non-standard codes were usually geeks like me who could deal with the issue. My AJ-841 worked fine as a printer for my college papers, after I modified CP/M to do a lookup and convert from ASCII to EBCD, and changed the serial port initialization code to set the 8250 UART for six-bit words with odd parity.

I think it would be nice to have my Model 15 and my H-89 back again, but time marches on and I must use this laptop instead.

Bill P.S. If AX.25 is no longer in fashion, I don't want to know. ;-)

It was cntl-shift something, cntl-shift P? Used to use it all the time.

The tape punch had a arrow shape where the tape was torn off.

I had a bunch of rolls of paper tape with BASIC programs punched on them, but I figured there'd be no easy way to print them, so there were trashed.

I am not an expert on this history, but I have a different impression.

Yes, for the short term, the 32, 33, and 35 were intended for Telex, TWX, and private network service. But I feel that when they developed both the 33/35 and ASCII they were not thinking so much of current plain vanilla telegraphy but rather computer communications of the future. This would be both terminal-to-computer (the TTY 33/35) and computer-to-computer (ASCII).

Recall that only a year or so after the 33 was introduced Dartmouth Univ. used them as computer terminals for its pioneer time-sharing service.

Data communication--sending blocks of data to be process by machine-- began in 1940 and grew after WW II. IBM's 1950s product line included card-to-tape and tape-to-card for transmission over telegraph lines, and then later card-to-card and tape-to-tape for transmission over higher speed voice grade lines. (IBM developed its own modem but then let the Bell System handle that end; the Bell System had DataPhone devices out by 1960). Clearly, batch transmission of data from remote sites to headquarters was seen as a growing business for both computer and communiations companies by 1960 and both were developing improved technology.

Also forseen and under development were both time sharing (such as BASIC users) and remote computer inquiry and update, such as airline reservations. ASCII and the models 33 and 35 were obviously well positioned for this service.

The limited-use capability of the 33 would be fine for occassional inquiries, say from a branch office to a central computer via dial- up. In any event, for this discussion I lump the 33 and 35 together as ASCII terminals.

If all that was desired was traditional message transmission telegraphy, Baudot and earlier TTY models (eg the 28) were more than adequate. Indeed, that technology remained in active commercial service well into the 1980s for private line transmission, news reporting, etc.

My impression from reading the WU bulletins was that they sought Telex to be an open system; the bulletins described development efforts to make such interfaces possible. Their bulletins at the time (1964) had numerous articles on computer interfaces. WU wanted AT&T's TWX, but by the time AT&T finally let it go the entire concept was approaching obsolescence.

My impression is that the bulk of the WU network was actually low speed with 5-bit repeaters, despite their efforts in microwave and hopes for the future. I sense their improvements to their network were actually rather modest and AT&T simply outpaced them much faster data lines and improved technology.

As far as I know, there was no reason WU couldn't have teamed up with a computer company and offered a time sharing service using its network and terminals. But I don't believe WU, despite its advertising claims, did that.

That is true, but even one parity bit offered more checking than Baudot did. I believe the Teletype 33 could be set up to print an "*" upon receipt of a parity error. Not much, but better than nothing.

Yes, but peripherals are an inseparable part of the computer. For example, the 33 and 35 supported sprocket feed forms and forms control, I don't believe the earlier units or Baudot did so. That enabled a report Teletype to print special forms, such as a rent-a-car agreement, insurance document, airline tickets, under remote computer control.

The ASCII designers undoubtedly were providing control codes for yet to be developed peripherals. The Teletype 33/35 was a modular design to allow add-ons. By Western Union choosing not to go ASCII, they were locking their customers out of using such new devices. It would be as if a competitive PC manufacturer refused to provide any expansion slots--how popular would such a machine be?

Again, I'm not an expert on this, but it would seem that if say an airline or car rental wanted to use form-feed Teletypes to print up tickets at their counters, they could and would not use WU to carry their data because a Baudot machine couldn't print a ticket as quickly and efficiently as an ASCII machine could. (As it happened, I think IBM Selectric terminals were used instead in such applications).

If you had a computer that was ASCII, as many mini-computers were, you needed no conversion at all. In the 1960s every operation added noticeable and unwanted cost and time, especially on the mini- computers.

As I understand it, Western Union was seeking to go computer in a big way with those circuits. It had a major military contract AUTODIN as well as a major Federal non-military contract (link up social security offices). It was using compuers to do switching and services. Now, I don't know if these networks were ASCII or Baudot. However, WU's ads and tech bulletins stressed these networks as its future. My guess is that even if those networks were ASCII, WU main network was not.

After the end of WW II, WU modernized its national network. I suspect it was hesitant to scrap switchgear only ten years old to make way for newer technology. Heck, it took 25 years for the Bell System to upgrade from relay to all-electronic. The difference (as I see it) was that the Bell System could afford to upgrade enough parts to provide advanced features for enough of its customers, while WU didn't have that luxury.

It would be of great historical interest to trace WU's 1960s Federal contract life. That is, we know about the early days of service. But how long did these networks last until replaced by newer ones? Did WU provide the newer networks or if not, why not? I think that history would tell us much about WU's decline.

By the time PC's came out, WU was on its last legs, like Howard Johnson's trying to survive against fast foods. IMHO, WU missed the boat in the early 1960s. Further, when PCs first came out, they were a tiny blip in the overall world of commercial computing and business communications.

As related to WU, the "computer revolution" was well underway by 1960, and they knew and trying hard to be a part of it. Things were changing, but not so quickly back then as they were to change later. The decisions WU made between 1960 and 1969 would determine the fate of the company.

Again, I'm not an expert, but it would seem that WU's proportionate role in business communciations in 1960 was far smaller in 1970.

[public replies, please]

this is all news to me

mk5000

Alexis Fogarty: [whining] I kept her secret all though high school. I was practically her only friend at her grandmother's funeral. I've listened to all her problems. All of time. Andrea Moreno: [peeved and annoyed] Alexis, that's just what friends do. --Ghost Whisperer

There is a lot of literature out there concerning the development of ASCII and the need for a data processing character code and the problems of all the then-existing codes.

Definitely the 33 gave time sharing a big boost because an inexpensive terminal was needed.

Telex, using 50 baud and Baudot code was long-established in Europe at the time. One of W.U.'s goals was to tie into the European system, something that TWX did not attempt. W.U. was somewhat hobbled since the government had required that company to divest the international cable business. Thus to connect with Europe W.U. had to turn to RCA or WU International or other companies for connectivity.

W.U. either wanted TWX or wanted TWX to go away. It was forever an article of faith with W.U. people that AT&T had violated an agreement to stay out of the telegraph business when it introduced TWX. And W.U. argued that TWX was cream-skimming business away from their public telegram business. AT&T countered that what they had agreed to was to stay out of the public telegram business, and that TWX was in fact a different service altogether since it provided a real-time two-way conversation. W.U. used all kinds of lame arguments to try to get the government to declare there should be one national voice communication system and a separate record communication system, as if the wires cared what kind of signals they carried.

One could argue that once W.U. had declined to buy the Bell patents their goose was cooked. The telephone infrastructure provided vastly more local and long-distance circuits and bandwidth than the telegraph system could. W.U. had largely to depend on AT&T for its intra-city circuits to customers' premises, and for some of its long-haul circuits.

There was one point at which W.U. announced it was going to be in the teleprocessing business. I don't know why that never came off. Maybe companies like Tymshare and G.E. Information Services beat them at it.

Around Teletype Corp. there was a famous story of a truck parts supplier whose computer system did not incorporate any reasonableness checking on quantities. (As many systems today still lack.) A single bit error turned an order for 7 dipsticks into one for 1007 dipsticks.

The computer dutifully sent out a reply that only a hundred or so dipsticks were in stock, and they would be sent right out, and the rest of the order would be filled when they could get some more made. Fortunately some shipping clerk packing up the order asked why that dealer needed so many dipsticks, and the error was caught.

Form feed and tabulation very early showed up as customer requirements, so the Model 15 (circa 1930) had those features.

Most of the control codes were for use with switching systems and error correcting systems rather than peripheral control. Hence ASCII provided start-of-address, end-of-address, start-of-text, end-of-text, end-of-block, etc. The irony in this is that lots of customers wanted to be able to transmit pure binary data, so the control characters would happen as part of the message. The solution is to provide a single character, Data- Link-Escape, and say that the character following DLE is a control character. If DLE happens in the binary data it is transmitted as DLE-DLE and the receiver removes the extra character from the data stream.

But not as modular as it should have been. The 35 was an ASCII version of the Baudot Model 28, so it did provide for quite a few features. The 32/33 were a completely new design and were intended to not have many options as a way of holding down cost. Originally they would not have accomodated sprocket feed and the form feeding and vertical tab functions; but apparently some very important customer demanded those, as they did become available.

Sprocket feed and form feed/vertical tab usually were provided together. The need for sprocket feed was recognized early on when customers wanted to make carbon copies. Without sprocket feed it is hard to keep copies in register since the inner and outer sheets of paper move at slightly different speeds.

I don't think W.U. was particularly blind about Baudot versus ASCII. They could see that their own public message business was long past its prime. They supported Baudot for Telex because they were not in a position to make the whole world switch to a new code. They had a lot of military business based on Baudot; but when the government started requiring ASCII for everything W.U. moved in that direction. (The G.S.A. system) (Note that AUTODIN was developed before ASCII was standardized; it was based somwhat on the earlier Fieldata code that was the military's attempt at a common-language code for messages and data processing.)

Which did happen, for those us who remember a certain model DEC microvax workstation that had epoxy poured into the expansion slots to make it non-expandable.

Probably because IBM had the contract to develop the SABRE system for American Airlines, and naturally wanted to use their own terminals rather than buying them from another party.

You'd think so, yet one thing that stood in the way of getting ASCII approved was that IBM wanted a code that was easier to translate to card code. That's why we got EBCDIC.

There was a fairly short-lived business involving computers designed specifically for telegraph message switching, replacing electromechanical switching equipment and paper tape intermediate storage. These systems were largely Baudot, as they came from an era when messages went between people rather than between a person and a computer. The G.E. Datanet-30, the Collins C-3000, the IBM 7740, and some other companies had products in this area. The Bell System couldn't play that game because of the 1956 antitrust consent decree which said AT&T would not do data processing. The business of this kind of computer was fairly short-lived because the person-to-person message systems gave way to person-to-computer systems; and handling person-to-person messages was easily handled by the new systems as part of their job. G.E. had a very versatile product in the Datanet-30, which served as well for ASCII as it had for Baudot, but it grew long in the tooth and minicomputers could do what it did for a lot less money.

I agree, that would be very interesting, and I wish somebody would do it. All you learn from the business press is that W.U. had more debts than they had income. I wonder if they were shocked by having to replace the AUTODIN computer with new models rather quickly, because of the rapid evolution of computer technology.

You're entitled to your opinion, but I don't think they were looking that far ahead: integrated circuits weren't invented yet in 1963, and Teletype probably assumed that electromechanical terminals would remain viable indefinitely.

The Model 33-ASR was standard with DEC Edusystem installations as well, but the choices made by users don't reflect Teletype's design intentions any more than the choices made by truck drivers reflect those of the automotive engineers who design the trucks the drivers routinely overload. It's in the nature of capitalism that businesses get the most that they can out of machines and employees: the fact that those who sold the machines wouldn't be responsible for their maintenance probably figured into the equation as well.

I might lump Yugos and Volvos together for purposes of discussing traffic control or urban congestion issues, but I wouldn't have any inclination to purchase a Yugo if I was in the market for a vehicle that I'd be able to use for fifteen or twenty years.

The Model 35 is the ASCII version of the venerable Model 28, a machine so reliable that many are in service to the present day. That reliability is why the 1A ESS came equipped with a Model 35 as the operator's terminal: it could provide a continuous record of operator commands and alarms, while interfacing directly with the exchange's control computer.

The Model 28 _IS_ a Baudot machine. ITYMTS "Model 32 and earlier ...".

Yes, UPI and AP used Model 15's until they were made obsolete, first by dot-matrix printers and then by "online" CRT-based computers which allowed news editors to paste copy directly onto TelePrompTer or Compu=Prompt inputs, and to feed the news directly into "Pagination" newspaper preparation systems.

WU had strict prohibitions against users attaching anything to TELEX or TWX circuits: I've never heard of them opening those networks even slightly. The company's later efforts, such as Easylink, were geared to providing separate revenue streams and to attracting customers (I was one) who would not have leased a TELEX/TWX service because they didn't have enough occasion to use them.

I'm not familiar with WU's history, so I'll ask other readers to chime in on that question. I know that WU did offer dedicated data circuits in the 70's and into the 80's, but the advent of FedEx and email killed its telegram business, and high-speed fax machines were the death knell of the TELEX network.

I agree it was better than nothing for Teletype printouts on TELEX or TWX connections, which were always intended to be human-to-human conduits. However, I was speaking of computer-to-computer or computer-to-terminal-server communications, which required a more reliable error detection scheme.

AFAIK, Teletype machines were always available with sprocket feeds for forms handling: my own Model 15 came with a sprocket-feed platen when I received it after it was retired from a railroad-dispatch center.

Although the Baudot code doesn't have form-feed and other forms-management control characters, some Teletype machines were equipped with "Stunt boxes" that could interpret special sequences of letters as forms-control or other special actions. They were often used for "station select" functions, which prevented printing on machines in a network when a message was intended for a single recipient or a subgroup.

Since many PC's now come equipped with all the common functionality on one motherboard, many users are unconcerned with expansion slots. I think they were always valued more for theoretical expansion than practical need, but I digress.

IBM chose not to go ASCII: those in the mainframe world still wrestle with EBCDIC, yet IBM is a mostly-profitable business to this day. Western Union's choice not to embrace ASCII had, IMNSHO, nothing to do with its demise: the company failed to adapt to the marketplace's demand for more sophisticated tools, and WU could have provided them without abandoning Baudot in legacy uses such as TELEX. Keep in mind that TWX machines, which (at least for the "100 speed" side) were already equipped for ASCII, never played any significant role in computer data processing. Baudot is just a way of getting something done, and WU chose not to do it, i.e., the company refused to face the threat to its TELEX/TWX networks in time to reform itself for the new age.

Believe it or not, Baudot code machines are actually faster than ASCII for many "plain text" applications, which is why the military still uses them. Western Union was as well prepared to carry customer's data as Ma Bell, but chose not to: they had the people, the tools, the rights-of-way, and the expertise - just not the will.

You're right about the Selectric terminals: my Anderson-Jacobson 841 was, in fact, a Selectric typewriter which had solenoids added to control the mechanism in step with incoming data. Airlines used them to print tickets, most likely because the SABER system ran on IBM mainframes and IBM was never slow in tying its products together.

Sorry, I don't follow you here. No conversion for/from what?

[snip]

Well, in the end we're saying the same thing with different analogies: Western Union could have, but chose not to, adapt to the revolution in business practice brought about by the invention of cheap general purpose computers. It's easy to see, in retrospect, that the PC was only a gateway - but it was the mother of all apertures, through which poured an incredible, pent-up demand for connectivity, for entertainment, and for the capability to reach others who share an individual's interests.

No one ever sees it coming when it's _their_ ox that's about to be gored. C'est la vie.

Bill P.S. "Baudot" is actually a misnomer, since Baudot's code was never used for teleprinters. In fact, teleprinters like the Teletype Models 14, 15,

19, 28, and 32 use variants of "Murray" code, but it's a difference that makes no difference: Murray will forever be in the shadow of his predecessor.

Well, it's true. Of course the 33s turned out to be fine terminals for time-shared computers. Well not very fine, but they were cheap, and that's what the market needed.

TWX operated under the Bell System's "no foregin attachments" rule, so from a TWX machine all you could contact was another TWX machine. For a computer terminal you had to use Dataphone service; and the latter was charged at voice call rates whereas TWX was cheaper.

Telex was similarly a person-to-person service with everything being supplied by the telegraph company. Until W.U. was in such dire straits they they started selling off the Telex machines to raise cash.

Sorry, that doesn't (excuse the pun) compute. A single bit error would alter the total by a power of two: I'd believe that 7 turned into 1031, but not 1007. Sounds like a company legend.

I attended a computer-processing methodology class once, where they drummed into us the need for always doing a walk-through of the code design prior to coding. The instructor mentioned a case where the system had been coded to automagically order more parts any time the inventory was below the amount of an order - even if the system had already ordered more parts.

These sorts of stories morph and mutate and are wrapped around whatever organization they pass through: a Teletype salesman who needs to convince buyers that a parity bit is really important probably came up with your 7-to-1007 change, since a computer programmer would/should know better.

I didn't know that Baudot/Murray code had a form-feed character. How was it done?

Bill

Actually, shouldn't it be 1039? Even that is unlikely unless the bit flip was internal to the computer. If it was flipped coming from a terminal, one would expect that a single character would be corrupted.

***** Moderator's Note *****

I assumed that the a single bit error would be the result of digital data being transferred over a channel that had no error checking, so the closest power-of-two would be 1024, and 1024+7=1031.

If the story was about a text error, it would only make sense if the field was transmitted with zero fill, i.e., as "0007", and if the bit flip turned the first zero into a one.

As I said, these "salesman's stories" morph to fit the company that's using them, so they're best understood after having a couple of drinks.

Bill Horne Temporary Moderator

Please put [Telecom] at the end of your subject line, or I may never see your post! Thanks!

We have a new address for email submissions: telecomdigestmoderator atsign telecom-digest.org. This is only for those who submit posts via email: if you use a newsreader or a web interface to contribute to the digest, you don't need to change anything.

How soon they forget the IBM 705, 7080, et al. A one bit error could most definitely turn 0007 into 1007.

R's, John

Probaby replaced the pound symbol or something else little used in the US.

***** Moderator's Note *****

I never worked with IBM 705 or 7080 equipment. What made it possible for a one-bit error to have that effect?

Bill Horne Temporary Moderator

Please put [Telecom] at the end of your subject line, or I may never see your post! Thanks!

We have a new address for email submissions: telecomdigestmoderator atsign telecom-digest.org. This is only for those who submit posts via email: if you use a newsreader or a web interface to contribute to the digest, you don't need to change anything.

IMHO, based on Oslin's book, WU was treated unfairly when they (1) had to acquire the money-losing Postal Telegraph and (2) divest the international cable business. WU, even in the 1940s, was not a big and powerful company with unlimited resources, yet the govt acted as if it was.

Again, IMHO, WU was right and AT&T was wrong in this aspect. AT&T was a rapidly growing company, WU was not. I can't help but suspect if WU attempted significant private line voice communications on its own network (which it had every right to do and did to a very limited extent) AT&T would scream in protest. Unfortunately, WU wasn't interested in voice and much of its network couldn't handle it anyway.

AT&T did throw WU a few bones by billing telegrams directly to phone accounts and later giving deep discounts on leased lines, at least until MCI made a stink about unfair treatment.

I suspect AT&T was willing to tolerate a small WU to avoid anti-trust issues (not that it helped), the way IBM tolerated Remington Rand and predecessors in the punched card business, even helping them out on occassion, again to avoid anti-trust issues (not that it helped).

Thanks for the correction.

Well, for the first 35 years or so, that card code dependency was very important in programming, so it made sense to do so. To this day sometimes in rough mainframe programming the "overpunch" of the sign in a numeric character causes a letter to be displayed instead of the last digit.

I am speculating here, but I suspect WU's engineers and management may have been more conservative and not ready for the rapid changes in electronics that occurred in the 1960s. As such, I suspect they were shocked at the need to replace AUTODIN and not prepared to deal with it.

But in fairness to WU, many 'tech" companies found it hard to keep up and [not] make mistakes. Take IBM and the coming of computers. There's a myth that IBM's Thomas J. Watson Sr was against computers, but that's not true. He sponsored electronics research even before his son joined the company and then turned it over to his son, Tom Jr. The problem at IBM was that even the son didn't see computers coming, and most of IBM's engineers were mechanical oriented, not electronics oriented. The son had to spin the company around in the 1950s several times, one to focus on electronics, and again to switch from tubes to transistors.

Also, IBM developed advanced transistor manufacturing techniques which were very valuable, but didn't think to patent them, and other companies got the benefit.

I do want to point [out] that, unlike today, electronics were extremely expensive in the 1950s and 1960s, and just because electronics were fast did not mean a relay was replaced with a circuit card. I can well understand WU's reluctance to spend big $$ to replace their relatively new relay systems with electronics. However, given the benefit of 20/20 hindsight, I suggest they made a mistake in that an electronic faster network would've been more attractive to data communications.

As mentioned, what I don't understand is the apparent disconnect between what WU said in print and what they actually did. Their tech journal talked about computers and high speed sophisticated stuff, but my impression is the implementation of such stuff was relatively rare. I get the impression it took them a very long time to set up microwave channels.

As an aside, all companies that have been around for a long time will have to drastically re-invent themselves as times change. Some companies manage to do so and thrive, others can't and whither away.

Decimal arithmetic, of course.

R's, John