Some brand names become the de facto name for the generic product. Xerox, for example. Or Velcro. Teletype was a trademark, but it has come to mean just about any teleprinter communicating with another teleprinter or a computer. The actual trademark belonged to The Teletype Corporation, part of Western Electric, which was, of course, part of AT&T. But there were many other companies that made teleprinters, some of which were very influential.
The teleprinter predates the computer by quite a bit. The original impetus for their development was to reduce the need for skilled telegraph operators. In addition, they found use as crude wordprocessors, although that term wouldn’t be used for quite some time.
Telegraph
Early communication was done by making and breaking a circuit at one station to signal a buzzer or other device at a distant station. Using dots and dashes, you could efficiently send messages, but only if you were proficient at sending and receiving Morse code. Sometimes, instead of a buzzer, the receiving device would make marks on a paper — sort of like a strip recorder.
In the mid-1800s, several attempts were made to make machines that could print characters remotely. There were various schemes, but the general idea was to move a print head remotely and strike it against carbon paper to leave a letter on a blank page.
By 1874, the Frenchman Èmile Baudot created a 5-bit code to represent characters over a teleprinter line. Like some earlier systems, the code used two shift characters to select uppercase letters (LTRS) and figures (FIGS). This lets the 32 possible codes represent 26 letters, 10 digits, and a few punctuation marks. However, if the receiver missed a shift character, the message would garble badly. This was especially a problem over radio links.
Paper Tape
Donald Murray made a big improvement in 1901. Instead of directly sending characters from a keyboard to the wire, his apparatus let the operator punch a paper tape. Then a machine used the paper tape to send characters to the remote station which would punch an identical tape. That tape could go through another machine to print out the text on it. Murray rearranged the Baudot code slightly, adding things we use today, like the carriage return and the line feed.
The problem that remained was keeping the two ends of the circuit in sync. An engineer working for the Morton Salt Company solved that problem, which Edward Kleinschmidt independently improved. The basic idea had been around for a while — using a start pulse to kick off each character — but these two patents around 1919 made it work.
Patents
Instead of fighting a big patent war, the two companies, Morkrum (partly owned by the owner of Morton Salt) and Klienschmitt, merged in 1924 and produced an even better machine. This was the birth of the modern teleprinter. In fact, the company that was formed from this merger would eventually become The Teletype Corporation and was bought by AT&T in 1930 for $30 million in stock.
Some early teleprinters were page printers that typed on the page like a typewriter. Others were tape printers that spit out a tape with letters on it. Often, the tape had a gummed back so the operator could cut it into strips and stick it to a telegram form, something you may have seen in old movies.
In addition to public telegrams, there were networks of commercial stations known as Telex and TWX — precursors to modern e-mail. These networks were like a phone system for teleprinters. You’d dial a Telex number and send a message to that machine. Many teleprinters had an internal wheel that a technician could set (by breaking off tabs) to send a WRU code (who are you) in response to a query. So connecting to the Hackaday Telex and sending WRU might reply “HACKDAY.” In addition, you could ring a bell on the remote machine. So a single bell might be a normal message, but ten bells might indicate an urgent message.
Word Processing
While replacing telegraphs was an obvious use of teleprinter technology, you might wonder how people could use these as crude word processors. The key was the paper tape and a simple paper tape trick. A Baudot machine would have five possible punches on one row of the tape. You can think of it as a binary number from 00000 (no punch) to 11111 (all positions punched out). The trick is that if all positions are punched out, the reader would ignore that position and move on to the next character. They also usually had a code that would stop the reading process.
This allowed you to do a few things. First, you could punch a tape and then make many copies of the same document. If you made a mistake, you could overpunch the tape to remove any unpunched holes and “delete” characters. It was also common to use several fully punched-out characters as a leader or a trailer, which allowed you to line up two tapes and paste them together.
So, to insert something, you could identify about a dozen characters around the insert and over-punch them. Then, you’d prepare another tape that had the new text, including the characters you punched over. You’d start that tape with a leader and end it with a trailer of fully punched positions. Then, you can cut the old tape and splice the new tape’s leader and trailer over the parts you punched out in the first step. A lot of work? Yes, but it’s way better than retyping everything by hand.
Once you create your master tape, you could turn out many originals. You could even do a sort of mail merge. Suppose I have a form letter reminding you to pay your bill. The master tape would have a pause in key places. So, the operator would do something like type the date, name, and address. Then, they would press start. The tape would type “Dear ” and then read a stop code. The operator could type the name and press start again. Now, the tape would run up until a later point, and another stop code would let the operator enter the account number and press start again. The next stop might be for the balance due, and a final stop for the due date. Pretty revolutionary for the 1940s.
Really high-tech installations used two tapes, one loop with the form letter and another unlooped tape with the input data. The operator did almost nothing, and all the letters were printed automatically.
Flexowriters were known to be used to generate form letters for both the White House and the United States Congress. Combined with an autopen, the system could create letters that people would perceive as hand-typed and signed, even though they were really automatically generated. You can see a Flexowriter in action in the video below.
Handwriting Computer
Another trick was to take a tape with a header and a trailer and paste them together to form a loop. Then the printer would just print the same thing over and over. I saw a particularly odd use of this back in the 1970s.
I was in a mall. There was a booth there purporting to have a handwriting analysis computer. I wasn’t willing to spend $2 on an obvious scam, but I hovered around, trying to understand how it worked. It was oddly familiar, but I couldn’t place it. The machine was very large and had many blinking lights and spinning disks. It looked like a prop from a very cheap 1950s science fiction movie.
People would pay their money and write something on a piece of blank paper. The clerk would take that paper and place it in a slot. With the press of a button, the machine would suck the paper in and spit it out with some fortune cookie message towards the bottom of the page. It might say, “You are stronger than people realize.”
Once I realized that was the “brain” of the device, it was obvious how it worked. Hidden inside was the paper tape reader. It had a loop of tape containing some line feeds, a fortune, more line feeds, and a stop code. The whole loop might have had a dozen or so fortune cookies, each with a stop code at the end of each.
When you put the paper in the slot, it really went around the teleprinter’s platen. You press the start tape button, and the line feeds suck up the paper and advance past the writing. Then, the fortune types out on the page. The final line feeds eject the page, and then it stops, ready for the next fortune. Pretty clever, although totally fraudulent.
Death of the Teleprinter
Teleprinters couldn’t survive the “glass teletype” revolution. CRT-based terminals swept away the machines from most applications. Real wordprocessors and magnetic media wiped out the applications in wordprocessing and typesetting.
Companies like Teletype, Olivetti, and Siemens (disclosure: Hackaday is part of Supply Frame, which is part of Siemens) stopped making teleprinters֫. In today’s world, these seem impossibly old-fashioned. But in 1932, they were revolutionary, as seen in the video below.
If you noticed the similarity between most modern teleprinters and electric typewriters, you aren’t wrong. Linux will still let you log in using a hardcopy terminal.
Teletype machines hasn’t completely died out. The telecommunication device for the deaf (TDD) continued to use the Baudot system, shift garbling and all.
Fascinating. How often is that still used today? Seems like we now have better methods of converting speech to text
Teletypes for the deaf was started as a public service project by retired AT&T folks called the Telephone Pioneers. They had piles of old Teletype machines retired from service. They were big and noisy, but the deaf people didn’t mind the noisy part. With some early modems they set up a system for the deaf to call each other and communicate over the teletypes. This morphed into TDD which still uses the same Baudot encoding.
Baudot 5-Bit code (7-Bit, actually, because of start bit/stop bit) is still being used in amateur radio! As RTTY, radio teletype. Also casually being called “Ritty” (in German “Rütütü”).
It’s the ancient 45,45 Baud variant that’s most famous, which is related to 50 Baud.
Now the funny thing is, 7-Bit ASCII had been tried to establish in ham radio as early as the 1970s when microcomputers and glass terminals had arrived.
Unfortunately, the hams stuck to Baudot code somehow.
It wasn’t necessarily a technological decision, even, but about culture.
The RTTY decoders/RTTY keyboards of the late 1970s could often do CW (Western/Japanese), Baudot and ASCII.
The models from the early 80s also could do AMTOR (based on maritime SITOR) or read NAVTEX messages.
Nowadays, RTTY is most being used by weather forecast service or in ham radio contests.
There’s also a web radio, with audio streams of RTTY signals.
https://www.rtty.com/itty/index.htm
Allow me to be a bit persnickety. Baudot is not a 5-bit code nor is ASCII a 7-bit code. Baudot is a 5-level code and ASCII is a 7-level code. Each level is either a mark of a space. The length of the start determines the length of each mark and space. Only decades later in the computer age were the signals interpreted as bits. Start bits and stop bits are likewise inventions of the computer age. Start and stop are not part of the character they are synchronization signals used in asynchronous communications. In synchronous communications they don’t exist.
Start and stop bits are not inventions of the computer age, they were used in the mechanical Baudot Teletype machines. The start bit engaged the clutch to start timing the five code bits and the stop bit gave the mechanism time to type the letter or punch the tape. Teletype is by nature asynchronous as it was often driven by a human typist.
Back in the day, I learned BASIC programming on an ASR-33 connected by 300 bps modem to an HP minicomputer at the school district headquarters. A few years later, I had an ASR-33 connected by current loop to my Commodore 64 as a printer.
I still regret getting rid of both of them.
I had an ASR-33, prior to my last move about 15 years ago. TH;DM Too Heavy; Didn’t Move
Mine did move – a tiny amount every time the carriage returned. One day I left it printing unattended and the inevetable happened: it committed hara-kiri off the table.
I did that too – same configuration as well! It was the Macomb Intermediate School District (Michigan, USA) that provided the computer. First an HP, then a rather unreliable IBM, then one called a ‘Magnasyn” (sp?).
I used to play Colossal Cave Adventure on a model 33 and modem.
Great article! Some of this I had not heard before — particularly the early word processing tricks.
I learned BASIC using an ASR-33 connected to a PDP-11/34 via a 300 baud modem.
A couple of years later I found myself spending many hours typing up (punching tapes), inventory reports at a maintenance line station and then transmitting them to the head office. I think we were on a TWX network.
It’s Friden, not Frieden.
KSR = Keyboard Send/Receive
ASR = Automatic Send/Receive
Correct model designations were the number first, then the function, e.g.,
33ASR for a model 33 with paper tape punch and reader (thus the “automatic”)
I was under the (mistaken?) impression that Hewlett-Packard was the successor of Friden calculators.
I saw my dad create ASCII art for Christmas on a teletype in the 50s by rolling a picture through with the paper and typing on the image until he had punched a tape that could be sent out as a greeting.
Ahhh. Acoustic modems and paper tape. How many klingons had to die?!?!?
It doesn’t matter, they died gloriously!
Their tales are sung around campfires to this day!
I worked as a service tech for Singer in the early 70’s. Worked on Flexowriters. it was amazing to see a machine do word processing using electro-mechanical technology. A lot of the machines I worked on were generating form letters and using mail-merge technique to customize them. The machines would run 8 hrs a day, 5 days a week and beat themselves to death.
Is this where we get the word “baud”?
Yep, you bet it is. Originally it was words per minute instead of bits per second I believe.. This was changed because it’s imprecise. But when we speak of baud afaik we still compare things to the symbol rate of a typist using the specific cadence required by the machines Baudot designed. You couldn’t go too fast or too slow.
Named after Émile Baudot, an early telegraph engineer. “Baud rate” is the number of symbols sent per second and depending on the modulation method, one or many bits can comprise one baud.
That’s right. The biggest difference between Baud and Bit/s is that Baud is the raw symbol rate per second.
Bit/s as a speed rating by contrast can also include compression.
That’s why later in time modems have dropped Baud in favor of Bit/s.
By mid-90s, the use of compression in modems became a common thing.
The speed of the COM Port on PC is raw and remains valid if measured in both Bit/s and Baud rate.
Baud id defined as the rate of the shortest meaningful symbol communicated. In modern computer communications this is equivalent to bps seeing as how all symbols are the same length. When using Morse it is equivalent to dots or dits per second.
” In addition, you could ring a bell on the remote machine. So a single bell might be a normal message, but ten bells might indicate an urgent message.”
Isn’t that a plot point in the original Andromeda Strain movie?
News wires would use more and more bells to indicate more important stories, but the number could vary depending on how wound up the operator was: https://blog.ap.org/behind-the-news/ap-flashes-what-theyre-all-about
I first learned about bells and newswires from a Hunter S. Thompson book back in the late 80s. I tried looking in google books but I’m not finding it. It might have been “Fear and Loathing at Three Hundred Bits per Second”, but… nah. It wasn’t that one…
“Isn’t that a plot point in the original Andromeda Strain movie?”
Indeed! There’s the scene with the piece of paper that got stuck in the bell! 😃
I thought it was a wood chip during the construction phase.
I might be wrong, but sometimes my mind locks on to a piece of trivia and doesn’t let go.
Typo!
s/Klienschmitt/Kleinschmidt/
N.B.:
110 bps ≠ 110 baud; ⇾ X bps ≠ X baud.
Not always. With NRZ and other binary coding schemes 1 b/s = 1 bd.
and the fun with the Fridens is: they even did not standardize their own machines. I have a regular punching reading machine and a presidential, proportional machine, and they cannot read each others paper tapes. when I made a papertape converter from ascii to friden, i had to make multiple lookup tables. for people who want to punch tapes, the python 2 script from back then still resides on our hack42 github page. its called Punchconverter. for a slightly better video of its workings look at https://vimeo.com/79331917
and even electrically they were different. I have both an external reader and puncher, but they use 110v relay power, and the flexowriter itself uses 48v, so unfortunately they cannot be connected to each other without an converter for all the signals. the flexo is eight bit parallel with a strobe signal and a few extra signals coming from the internal relays and the swithes on the front.
oh and there is a slot in de front bottom where you can stick you hand in to lift the machine up vertical to reclaim some desk space and they are so heavy, you think they are bolted to the table. on the back there are two rollers attached, so when you lift the front, it can roll on its hind wheels. you absolutely need those to move the machine around.
Oh2. IT IS LOUD!!!!! imagine a room with 10 of those beasts hammering away the mailing from morning till evening.
That‘s where the name „tty“ originates, thinks it me.
UNIX is older than mainstream crts.
Linux features „line disciplines“ which are often plugged behind tty device interfaces. All those stty parameters tell a story of days long-gone.
I find it rather funny that the infastructure behind a linux console goes all the way down to emulation of hardware signals. Or the ominous control commands: Have you ever evaluated what actually happens if you press CTRL-c ?
CP/M had an “AUX” (auxiliary) device for serial port.
DOS, too. Here, AUX was an alias for COM1. Probably for compatibility with ported CP/M programs, among other things.
I think the truly awesome thing about the early teletypes is that the decoding and printing functions were entirely electromechanical – no transistors, no ICs (not even a 555 timer).
You have to admire the inventiveness of the engineers of yore.
Sure. But then there are other people like me who admire the silence and elegance of later technologies.
For example, many think that the steam locomotive is awesome!
But I just think: Meh! Heat, dust, noise. No air to breathe!
Pollution! Black clothes on the washingline everytime the train comes! Kids with athma!
People like me rather are fascinated by the electron tube and the electric tram.
Or let me put it this way, if you see a mechanical typewriter first time in action, it’s impressive!
But once you hear that noisemaker 24/7, you quickly wish for a glass terminal!
It’s just like with the steam locomotive and the tram, I think. Some want action, some want peace.
There is a value to the old mechanical typewriters: they were great for relieving frustration! Upset? Pound them keys harder! (Of course, if you’re really upset, you’re going to cut through the paper! 😲)
Oh, and their batteries never died!
I’ll spare everyone my walk down memory lane… But I did want to comment to the folks at H-A_D that I immensely enjoyed this well-researched article. Just don’t forget to wrap a heavy bath towel around your acoustic coupler if someone’s got a radio or tv turned up or someone slams a door shut, etc.! When I used my first TTY, 110 baud was where it was at. 300 was SMOKIN’! 9600 was outrageous! 14.4K or 56K were simply blue light!
Back around 1980 most companies still had telex machines buried somewhere in their head offices. If you wanted to make a big impression you would send a telex saying that you were arriving in a half hour. Nobody could figure out where the hell the message had come from or exactly who had sent it,and replying to a telex was an arcane skill , but they figured that you must be really important.