telecom@eecs.nwu.edu (TELECOM Moderator) (09/20/89)
Someone the other day asked, 'Have the public telegraph offices mostly been replaced by the 800 phone service?' The answer is yes. The 800 service along with personal terminals and accounts on Easy Link for hundreds of companies which used to rely on telex and/or TWX have virtually decimated the telegram, and the telegraph public office. I mentioned that in small (and some medium size) towns, Western Union did not actually own the facilities, but operated through an agent; someone who had other things going on as well, such as the local agent for the bus lines; now and then the local (independent and rural) telco office, etc. In larger cities such as Chicago, St. Louis, New York, Boston, Los Angeles and similar, Western Union operated their own public message centers. The Chicago office is well etched in my mind, in the 1955-60 time period. The office was 1930-ish Art Deco design. It was about the size of a typical high school basketball gymnasium, with the same sort of vaulted ceilings. In front, several small writing desks, like in a bank, where one would stand and write out the message on a form submitted to the clerk. The fountain pens were chained to the counter so they could not be stolen. A large calendar on the wall, and of course, the 'telegraph clock'. At the counter, which was marble, three or four places for clerks, although there were usually just one or two on duty at a time. Behind them, nine or ten rather large, somewhat noisy teletype machines. Usually there were two operators attending all the machines; they would walk to one, peer at the paper feeding out, look at the next machine, etc. Sitting down at a machine, these operators -- invariably men, although the counter clerks were frequently women -- would type messages and receive them. The room had a constant din from the machines. One or more was constantly typing. A small bell, driven by a <control-G> just like today would go 'bing bing bing' and the motor inside the machine would start going with a soft sort of whirring noise. Maybe two seconds later the keys would start printing and the carriage would move back and forth on the platen. As quickly as it started, it would stop. The motor would shut off, and the machine would go silent. But the starting and stopping could not be predicted. Sometimes all would be running, other times just two or three. Just as one would stop, another would start. And the men would walk around, gather up paper from each and give it to the clerks at the counter. One machine would stop, and immediatly bing! bing! another one would start. The operators seemed to be fast typists. Picking up a piece of paper from the clerk, one would sit at a machine and begin typing -- quite fast it seemed to me at the time, although I learned later they were only going about 60 baud, or some ridiculous pace by today's standards. And always with a cigarette hanging out of their lips as they sat there banging away. The clerk functioned much like a school teacher grading an essay paper. You'd take your message to the counter written out, and with a red pencil the clerk would say, "what's this word here?", "how do you spell that name?", "I can't read your writing, you will have to go back and print it over again". When satisfied, she'd count the words and looking at a chart say, "well, that's fifteen words, gimme dollar seventy five". You would pay, and she would hand the message over to the operators for typing into the network. From time to time the operators would hand over messages to the clerk, who would speak loudly, "Smith? John Smith?", and if Smith was waiting around for his message, he would come to the counter to get it. Sometimes it would be registered, meaning he signed for it, and the operator would send back a message to the other end saying delivery had been made. If Smith could not be located in the office, then a messenger -- either a twelve year old kid or a seventy five year old man -- they seemed to have no one in the middle age at all doing this work -- would take the message and set off on a bicycle to deliver the message to Smith at his office or home. Twenty four hours a day, seven days a week this went on. If money was wired to you, they would write a check payable to you, which you were free to endorse and cash right there if the clerk had enough money in the drawer, or you could cash it elsewhere. If you just hung around the place all day, talking to the people coming and going, you got quite an education and heard the life stories of everyone passing through, it seemed. Patrick Townson
johnl@esegue.segue.boston.ma.us (John R. Levine) (09/20/89)
In article <telecom-v09i0389m03@vector.dallas.tx.us> you write: >typing. A small bell, driven by a <control-G> just like today ... Smallest nit of the week -- telex machines are all Baudot five-bit code, for which there's no such thing as a control key, just letter-shift and number-shift. The bell is some number-shift key. Regards, John Levine, johnl@esegue.segue.boston.ma.us, {spdcc|ima|lotus}!esegue!johnl [Moderator's Note: Well I believe it was the 'shift - 7' now that you mention it; and of course control-G is Ascii 7. Weren't the 'number-shift' keys essentially like control keys? How did they get line feed, carriage return, ENQ (who are you?) and answerback without control codes? My handy Ascii chart here says control-E, or ASC(5) when sent polls the other end to identify itself. What do you think? PT]
npl@mozart.att.com (Nickolas Landsberg) (09/21/89)
Hello Patrick,
Hope this gets through, have had trouble in the past sending
stuff to you, but it mostly wasn't too terribly important, and, anyway
someone always had a better answer in a day or two than I did.
In the above mentioned article, you mentioned that the speed
was around 60 baud. Well, it was actually something called "75-speed"
which indicated a maximum speed of 75 WPM, transmitted in 5-level
"baudot" encoding. After doing all the mumbo-jumbo af adding start/stop
bits, etc. this translates to an equivalent bits-per-second of 56.83
or thereabouts. A system I worked on once had to monitor transmissions
from a telco switch which also used this. An interesting sidelight is
that they could transmit the full upper-case character set, the numerics
and a goodly supply of punctuation using just 5 bits. (Actually, they
used a "switch" character to flip-flop between meanings of particular
bit patterns. The "alternate" set would continue to be used until
the "switch" character was seen again.)
Nick Landsberg
P.S. I thoroughly enjoy the TELECOM Digest. Thanks for your efforts!
[Moderator's Note: Blush. Thank you. If you have trouble getting the mail
through in the future, try one of these alternate addresses, all of which
terminate here at the Digest in Evanston, IL:
telecom@cs.bu.edu
telecom@nuacca.bitnet
attmail!ptownson (slower, but it gets here)
telecom@vector.dallas.tx.us PT]tanner@ki4pv.uucp (Dr. T. Andrews) (09/23/89)
) ... you mentioned that the [baudot tty] speed
) was around 60 baud. Well, it was actually something called "75-speed"
) which indicated a maximum speed of 75 WPM, transmitted in 5-level
) "baudot" encoding.
Some of these 5-level devices could actually be geared for
different speeds. The most common speed for the units, if you
find them today, is 45.45 baud (with 1.5 stop bits!), which works
out to ~6 chars/sec (commonly called 60 WPM).
) Actually, they used a "switch" character to flip-flop between
) meanings of particular bit patterns.
They used two characters. There was a "shift" character
(sometimes called "figs", bit pattern 0b11111) which forced the
second character set. To drop back to letters, the "unshift"
character (sometimes called "ltrs", bit pattern 0b11011) was
sent.
As a user-settable option, the machine could also be set to
"unshift-on-space", which meant that if a space (0b00100) was
received when the machine wsa in shift mode, it would drop back
into unshift mode. The user selection of the "unshift-on-space"
option is made by moving a metal bar under the vanes.
I do mean "drop" - shift mode was handled by raising the carriage
so that the other row on the type slugs would hit the platen.
The "unshift" got a nice gravity assist, and made a nice sound.
Those old model 15 and model 19 teletypes were fascinating to
watch.
...!bikini.cis.ufl.edu!ki4pv!tanner ...!bpa!cdin-1!ki4pv!tanner
or... {allegra attctc gatech!uflorida uunet!cdin-1}!ki4pv!tannerdik@cwi.nl (Dik T. Winter) (09/25/89)
> [Moderator's Note: Well I believe it was the 'shift - 7' now that you mention > it; and of course control-G is Ascii 7. No, it was shift-J. > Weren't the 'number-shift' keys > essentially like control keys? How did they get line feed, carriage return, > ENQ (who are you?) and answerback without control codes? I have here CCITT #2 which says that LF and CR are valid both shifted and unshifted and ENQ=shift-D. This gives me some remembrances. The first computer I did use had Telex typewriters as consoles. However, they where very advanced. They had a four row keyboard and would remember whether the last shift was figures or letters. So when you entered a symbol not in the current shift the apparatus whould not generate it unless you entered the correct shift first. (Yes, it was made by Siemens.) Here follows the CCITT #2 code table: Letters Figures 0 8 16 24 0 8 16 24 0 space LF E A space LF 3 - 1 T L Z W 5 ) + 2 2 CR R D J CR 4 ENQ BEL 3 O G B figures 9 undef ? figures 4 nil I S U nil 8 ' 7 5 H P Y Q undef 0 6 1 6 N C F K , : undef ( 7 M V X letters . = / letters A question is: where in this whole lot of bits figures the sprocket hole; i.e. is it 1.2.4.o.8.16 or 16.8.4.o.2.1? Next to me lies a lot of 5-level paper taper, but I cannot yet figure it. This is CCITT #2 or Baudot. There is another 5-level code: CCITT #1 which has also another name. I must have the code somewhere, but am not inclined to dig it up unless requested. There appears also to be some different Telex codes in the US. If I remember right, a 3 out of 7 code or somesuch and a 8 level code (but not ASCII). I have the codings somewhere (in the same place as CCITT #1) but do not know under what circomstances they were used. Any help is appreciated. dik t. winter, cwi, amsterdam, nederland dik@cwi.nl
morris@jade.jpl.nasa.gov (Mike Morris) (09/25/89)
johnl@esegue.segue.boston.ma.us (John R. Levine) writes: >In article <telecom-v09i0389m03@vector.dallas.tx.us> you write: >>typing. A small bell, driven by a <control-G> just like today ... >Smallest nit of the week -- telex machines are all Baudot five-bit code, >for which there's no such thing as a control key, just letter-shift and >number-shift. The bell is some number-shift key. >[Moderator's Note: Well I believe it was the 'shift - 7' now that you mention >it; and of course control-G is Ascii 7. Weren't the 'number-shift' keys >essentially like control keys? How did they get line feed, carriage return, >ENQ (who are you?) and answerback without control codes? My handy Ascii >chart here says control-E, or ASC(5) when sent polls the other end to This isn't ascii! 7-bit ascii has 128 combinations, we have 5 bits with 32 combinations. Like they say in Oregon: "Things are different here!" >identify itself. What do you think? PT] Here's the map of the baudot / 3-row / pick your name.. 1 2 3 4 5 6 7 8 9 0 The upper case is the same in all character sets Q W E R T Y U I O P for this row of the keyboard. - $ ! & ' ( ) A S D F G H J K L " / : ; ? , . FIGS Z X C V B N M LTRS LF (blank) (space bar) There is actually four different US character sets, plus the international set. There is the "Military", the "Weather", the "TWX" and the "Telex" set. LTRS Int'l Mil TWX TELEX WX A - - - - up arrow S ' Bell Bell ' Bell D currency $ $ WRU upper right arrow F DV* ! 1/4 $ right arrow G DV* & & & lower right arrow H DV* STOP DV* # down arrow J Bell ' , Bell lower left arrow K ( ( 1/2 ( left arrow L ) ) 3/4 ) upper left arrow Z + " " " + X / / / / / C : : WRU : circle V = ; 3/8 ; circle with a vertical bar inside B ? ? 5/8 ? circle with a + inside N , , DV* , circle (my chart dupes shift-C) M . . . . . DV* means "Domestic Variation" My first machine was purchased from Southern Pacific, and had the "TWX" character set - commonly knowin in ham circles as the "stock market" set. Fortunately the Model 15 and 19 printers had a type basket with soldered-on type pallets, and I was able to purchase replacement pallets (for 30 cents apiece!) and solder them on. Changing the shift-J bell to a "S-Bell" required disassembling the machine and replaciing the decoding arm. Interestingly, breaking off a tab allowed printing a bell character (like the <Ma Bell logo) on the page, in addition to sounding the signal. The 15 and 19 had a bell and cam-operated hammer, the 28 closed a contact which powered a solenoid to ring the bell. WRU was a contact closure. The hardware varied with the vintage/user, and could be a paper tape reader with a loop of tape, a toothed drum with 32 rows of 5 teeth, that were broken out to create character codes, or a bank of diodes and a rotary stepper switch. STOP was a contact opening or closure, depending on the model. It actually turned the machine off, if the option was wired in. I saw exactly one machine where it was used. Shift-H and shift-N in the TWX set varied with the user. I've seen a shift-h with the UK currency "L" on it, and with the double-dot-U (Scandinavian character). As to speeds, there were/are 8 "standards": Pulses Data Stop per Pulse Pulse Char WPM Baud Ms Ms Use 7.42 61.33 45.45 22 31 US Bell system 7.0 65 45.5 22 22 US Western Union 7.5 66.67 50 20 30 ITU standard - Europe 7.42 67.33 50 20 28.4 US Military, for interoperation with allies 7.0 71.43 50 20 20 Old CCIT European standard 7.42 76.67 56.88 17.57 25 US "75 WPM" all commercial & military 7.42 100.0 74.2 13.47 19.18 US "100 WPM" all commercial & military 7.0 100.0 74.2 13.47 13.47 US military very limited use The first two are compatible, the three 50-baud are compatible and the two 100-wpm are compatible. As a sidelight, I have seen more than one 60-wpm two-fingered typists. Also, the model 32 machine was available with a 3-row keyboard as diagramed above, and a 4-row that had the numbers on the 1st row, and the QWER... on the second. A mechanical lock that was triggered by the FIGS and LTRS keys would lock out one or the other row. I actually saw a implementation of a baudot BASIC running on an old 8080-based home computer in 1976-1977. The gentleman had hacked the character I/O drivers in a copy of a ASCII basic (the old Microsoft BASIC) to do table lookup so that a 5-level TTY could be used. He was using the military character set, and things like "<" and ">" were translated into strings like fortrans ".LT." and ".GT." - he had implemented .NE. for <>, etc. I was amazed that it could be done, and flabbergasted that he did it in a 8k system (note that 8k was the total RAM - which held the system, the I/O lookup and the BASIC interpreter!). I wish I had saved a copy of the I/O driver listing. Mike Morris UUCP: Morris@Jade.JPL.NASA.gov ICBM: 34.12 N, 118.02 W #Include quote.cute.standard PSTN: 818-447-7052 #Include disclaimer.standard cat flames.all > /dev/null
tanner@ki4pv.uucp (Dr. T. Andrews) (09/29/89)
) [Moderator's Note: Well I believe it was the 'shift - 7' now that
) you mention it;
There is no "shift-7" in baudot code. The figure '7' is on the
"shifted" character set (same character code as the letter 'U',
0b00111). The had the same code as the letter 'J' (0b01011).
) Weren't the 'number-shift' keys essentially like control keys?
Presuming that you mean the "shift" (FIGS) and "unshift" ("LTRS")
keys, no, they weren't like control keys. They differed in two
important ways. First, they sent character codes. Secondly,
they had no printing effect themselves, but caused later
characters to be printed from a possibly different charset.
) How did they get line feed, carriage return,
There are baudot codes for CR and LF. These, interestingly
enough, are effective in both shifted and unshifted modes.
) ENQ (who are you?)
You typed "WHO DAT<FIGS>?<LTRS><CR><LF>" and stopped typing.
) and answerback without control codes?
The other guy typed "IT<FIGS>'<LTRS>S ME.<CR><LF>" in response.
He might have had a tape reader in which case he could have this
text prepared for your editfication; it would save him typing it
again.
) My handy Ascii chart here says control-E, or ASC(5)
It should be noted that ASCII is not a 5-level code. When
discussing the 5-level devices, refer not to your ASCII chart but
to a BAUDOT chart instead.
) What do you think? PT]
I think that you haven't worked on many 5-level devices, to be
honest. These slow devices (noisy, too) are pretty much
unrelated to any modern computing needs. I only ever heard of
one general-purpose computer system which purported to support
the things, and never actually saw it done.
Please see the [Radio Amateur's Handbook] for information on
the BAUDOT code. You might also consult a maint manual for a
model 15/model 19 teleprinter.
The original wiring on these devices is somewhat baroque.
Considering that very little of it is actually needed (a common
maint trick is to remove the old wiring, add the 10 or so wires
needed in the whole device, and try to figure out what to do with
the many, many feet of old wire. It has been thought that
perhaps the added wiring was for the military - whether to
confuse the enemy, or just to raise the price, has not been
determined.
The characters in the FIGS set, by the way, were somewhat
variable depending on the particular unit in hand. On some, you
could get motor stop instead of one of the printing characters (I
think that you might have been able to configure which you got).
On others, you might not have the prime ('). Other units had
weather symbols instead of some of the figures.
...!bikini.cis.ufl.edu!ki4pv!tanner ...!bpa!cdin-1!ki4pv!tanner
or... {allegra attctc gatech!uflorida uunet!cdin-1}!ki4pv!tannerandyb@coat.com (Andy Behrens) (10/03/89)
Dik T. Winter <dik@cwi.nl> writes: > Here follows the CCITT #2 code table: > > Letters Figures > 0 8 16 24 > +-------------------------------------------------------- > 0 | space LF E A ... > 1 | T L Z W ... > 2 | CR R D J ... > 3 | O G B figures ... > 4 | nil I S U ... > 5 | H P Y Q ... > 6 | N C F K ... > 7 | M V X letters ... At first there seems to be no pattern to the way the letters are assigned, but if you arrange them according to the number of bits that are set in each character, it becomes clear. The most common letters of the English alphabet (ETAIONSHRDL....) and the word separators (space, return, line feed) have the codes with the fewest number of marking bits. Question 1. Does anyone know why this is so? My theory: If the teletype machines are driven with a current-loop interface, this arrangement of codes minimizes the power that needs to be transmitted. Question 2. Why was "Z" grouped with the most common letters? ============ ============ ============ ============ ============ ============ no bits 1 bit 2 bits 3 bits 4 bits 5 bits ============ ============ ============ ============ ============ ============ 00000 blank 10000 E 11000 A 01110 C 11110 K 11111 ltrs 01000 lf 01100 I 10110 F 11101 Q 00100 space 00110 N 01101 P 11011 figs 00010 ret 00011 O 11010 J 10111 X 00001 T 10100 S 10101 Y 01111 V 01010 R 01011 G 00101 H 11100 U 10010 D 11001 W 01001 L 10011 B 10001 Z 00111 M ============ ============ ============ ============ ============ ============ Live justly, love gently, walk humbly. Andy Behrens andyb@coat.com uucp: {harvard,rutgers,decvax}!dartvax!coat!andyb RFD 1, Box 116, East Thetford, Vt. 05043 (802) 649-1258 Burlington Coat, PO Box 729, Lebanon, N.H. 03766 (603) 448-5000