November 21, 2021

Telecom Miscellany #1

Morse wasn’t first. The Morse system was significant–it was the earliest telegraph system to be widely deployed in much of the world, the first international standard and the direct predecessor to technologies still in use–but there were earlier telegraphs.

The term “telegraph” was first applied to optical telegraphy, which involved visual signals being transmitted over long distances using mechanical apparatuses on hilltops who viewed each other using telescopes. This is related to many other means of visual communication–semaphore flags, maritime signal flags, smoke signals, “one if by land and two if by sea”, the daily ball drop at the Greenwich observatory, the annual ball drop at Times Square–of which there are too many to write about here.

The first electric telegraph to see commercial deployment was developed by William Forthergill Cooke and Charles Wheatstone in Britain in the 1830s. The sender would turn handles at one end, creating a signal that would rotate needles on the display at the other end. It was marketed to railroads to be used for internal signaling purposes and was first deployed on the Great Western Railway in 1838, a full six years before Morse.

The initial system had five needles, two of which could be rotated at once to point to a letter on the display board. This meant that messages could be read and written without training in a specialized code, but as it turned out the cost of the additional wiring and equipment was greater than the cost of training telegraph operators, so within a few years it was replaced by a two-needle system, and shortly thereafter a one-needle system, in which there was no display board and a series of rotations encoded a letter.

With permission from the railway, Cooke opened the telegraph line to the public in 1843, a year before “what hath God wrought” introduced the telegraph to America. The initial line ran between central London and Slough, near Windsor Castle, making it well placed for political communication at a time when the royal family was still very relevant. The first line’s widely publicized use to apprehend a murder suspect in 1845 increased public interest in telegraphy and Cooke incorporated the Electric Telegraph Company to build out a nationwide network later that year.

(It can be argued that Cooke was running the London-Slough line as a publicity project rather than a business endeavor, and the Morse syndicate was in business earlier, first under contract to the US Post Office and then in their own companies’ names. I’m not sure if this distinction is meaningful and am happy to give this one to the Brits.)

The Electric and its main rival, the British & Irish Magnetic Telegraph Company (no relation to Morse’s “Magnetic” company that ran the Washington-Baltimore line) used needle telegraphy on the British domestic networks until they were nationalized in 1870. The state monopoly gradually converted to Morse code, by then the international standard, over the following decades.

There were two Morse codes. The original code Morse used in 1844 had, in addition to dots and dashes, a slight pause shorter than the pause between letters that was part of some letters’ encodings, as well as a long dash for L and an extra long dash for zero. Although Morse’s name is on the code, it was mainly developed by his assistant Alfred Vail. (Some have questioned whether Morse, a portrait painter with no engineering training, invented much of anything at all, or whether he was just a Steve Jobs type visionary figurehead. The truth is probably somewhere in between.)

What we know as Morse code, with three dashes for O instead of dot-pause-dot, was developed by Friedrich Clemens Gerke and used on the German telegraph network from its inception in 1848. This simplified operation at the cost of making it less efficient, as typified by a common letter like O getting stuck with a long code like dash-dash-dash. That was a number five in the old code, but to make room for relocated letters, numbers and punctuation marks had to be completely redone. “German Morse” became “Continental Morse” as neighboring countries adopted it, and then “International Morse” when it was included in the first International Telegraph Union standard in 1865. The old system became known as “American Morse.”

American Morse was used on the domestic wired telegraph systems in the US and Canada. International Morse was used almost everywhere else: domestic systems in most other countries, underwater cables connecting the continents, and radiotelegraphy worldwide. Radio Morse stuck around a good while longer than wired Morse did, so that’s why even in America today we call the distress signal “SOS” and not “S5S.”

Early regulation. In 1848 the New York State Legislature enacted a law (chapter 265 of 1848) permitting “any number of persons” to incorporate a telegraph company by filing a certificate with the secretary of state. Any such telegraph company would then have the right to place its telegraph lines along any public right-of-way in the state, without needing any further license or franchise from any state or local authority.

This seems an extraordinarily broad grant to anyone willing to pay to construct these lines. Undoubtedly the legislature wanted to encourage the development of this new technology, and was heavily, shall we say, lobbied to grant strong privileges to the Morse syndicate, in particular their New York, Albany and Buffalo Telegraph Company which had been operating in the state, unincorporated and without a special franchise, since 1846. But unlike Maryland, which offered similar privileges to any telegraph company incorporated by act of the legislature (e.g., the Magnetic Telegraph Company, incorporated by chapter 39 of 1846-47), New York offered it to all. In 1851, the company that would become Western Union was chartered under this law by affiliates of Morse’s rival O’Reilly; and in 1885, the Bell System would use this law to charter AT&T. (There were no laws applicable to telephone companies at the time, which may be why AT&T and other early phone companies included “telegraph” in their names even if they didn’t do any telegraphy.)

Other provisions of the law required telegraph companies to transmit messages in order of receipt and to interchange messages with connecting lines on reasonable terms. Almost two centuries later and we’re still dealing with these issues regarding ISPs.

Early information theory. As the first telecommunications network, and one with a very low bandwidth, the telegraph was the first to deal with many issues:

Addressing. Every message sent on a telegraph line would propagate to both ends and reach every other station on the line. To address an individual station, operators developed one- or two-letter call signs to alert the station they wanted to reach. This was the ancestor of the three- and four-letter call signs used in radiotelegraphy, which in turn led to the use of call signs in the same namespace by broadcasters in some countries. (And the FCC takes it even further, issuing licenses starting with “K” or “W” for things like mobile phone networks which never use them for identification purposes.)

The “last mile” of telegraphy was delivery by mail or by messenger, and accordingly physical mailing addresses were typically used. Businesses who dealt with high volumes of communication could register a short one-word address to make sending and receiving that much more efficient. This was the birth of a modern brand name: the City Bank of New York used the telegraph address “Citibank” because an “I” was shorter in Morse than a “Y.” It became their brand, and then their official name.

Error detection and correction. ITU-T Recommendation F.1, the most recent standard for the telegram service, specifies how operators can request retransmission of potentially garbled messages. If the word count doesn’t match the length of the received message, the receiving operator sends back the first letter of each word received so the sender can correct the discrepancy. It also recommends internal redundancy for important parts of messages, like “FIFTEEN HUNDRED (1500)” or “DO NOT, REPEAT, DO NOT.”

Data compression. Some have remarked upon the resemblance between cryptic abbreviations used by telegraph operators and “txt spk” that was popular on SMS before everyone got smartphones with autocorrect.

But also, telegrams were expensive and they charged by the word. People looked for whatever efficiencies they could find to save money. Leaving out unnecessary words like “the” was an easy solution, as was writing “STOP” instead of a period if the company charged extra for punctuation. In the late 19th century some companies started publishing code books to reduce a long message to a few short nonsense words. When ITU authorized charging extra for “cipher messages” that did not consist of known words in certain allowed languages, the code books moved to encoding phrases as foreign (or obscure English) words. So for example in the 1901 Western Union Telegraphic Code, the word “Dimetrorum” encodes the sentence “Unless you make an exception.”

The code books were public so this “encoding” was not a form of encryption. Truly encrypted messages, as sent in diplomatic services, could be sent at the higher rate for ciphers.

Quality of service. As I mentioned above, New York mandated that no messages be prioritized over others way back in 1848. As the system matured, different classes of messages were gradually developed. ITU F.1 now lists eight priorities, which messages involving the safety of life at highest, and “letter telegrams” to be delivered by mail at lowest.

The telex machine is kept so clean. The later forms of telegraphy didn’t leave as much of a cultural footprint as the Morse telegraph key. I’ll try to summarize as best as I can what came later.

Printing telegraphs, i.e., telegraphs that directly produced printed messages without having to be decoded by an operator, were considered desirable from the very beginning. In Britain, Wheatstone invented a printing system after what proved to be an acrimonious split from Cooke; in America, Royal E. House patented a printing telegraph in 1846. Other systems followed. The Gold and Stock Telegraph Company, a Western Union affiliate, pioneered stock tickers in the 1870s. (Gold and Stock is also known for two other things: they operated Western Union’s telephones in competition with the Bell System before Bell won his patent case, and they deployed a burglar alarm network that was later merged into American District Telegraph, now just known as ADT.)

The breakthrough in usability for printing telegraphs was Donald Murray’s use of a typewriter keyboard to punch a message on paper tape, which could then be transmitted by telegraph (it was a fixed-length Baudot style code though incompatible with Baudot’s original) and decoded on a printer at the other end. This took some time to catch on, as the equipment was expensive and transmission speed was about the same as a skilled operator on a Morse key. Western Union tried rolling out Murray teleprinters during an operators’ strike in 1907 but it didn’t catch on at the time. But by the 1920s teleprinters were cheap and reliable enough that they started displacing Morse code.

(Side note: “teleprinter” is a generic term. The trademark “Teletype” belonging to the AT&T subsidiary of the same name is much better known.)

(Another side note: Western Union may not have upgraded its last Morse code lines to teleprinters until the 1960s. Hard to track down a reliable source on this. But it would mark the end of the “American Morse” variant after over a century of continuous use.)

The next step was telex, a public switched network that let teleprinters directly dial each other, rather than be tied to a single line and require manual forwarding from line to line. This rolled out in the 1930s, in the US as the AT&T Teletypewriter Exchange (TWX). By the 1960s Western Union had its own telex service, and would acquire AT&T TWX and rebrand it as “Telex II.”

In the 1950s AT&T developed an early modem to encode data as audible sound and send teleprinter messages over telephone lines. Initally meant for telex, it soon saw additional use for connecting teleprinter terminals to computer systems. Then in the 1960s a more advanced 7-bit code for teleprinters and computers began to replace the 5-bit Baudot-Murray code: ASCII. Which of course is the basis of Unicode and still used today.

Telex machines connected to computers, computers connected to other computers, so why not send telegrams over computers? And indeed ITU F.1 suggests that telegrams can be sent as messages over public data systems. One of many applications for ITU’s network protocol suite, developed in the 1970s, is to encapsulate a telex message as an X.400 email, which can then be sent over an X.25 network and read at the other end, either electronically or on paper. The plan was to smoothly transition telegrams, a century-old technology, into the modern age, and what were then still mostly state-owned telecom monopolies would lead the way…

But it didn’t happen. TCP/IP won out over X.25 as the standard for networking. (A good thing too - X.25’s reputation is precisely the kind of bureaucratic cumbersome protocol you’d expect from people who charge by the word. And have you seen what X.400 email addresses look like? SMTP may be a trash fire but at least it’s somewhat readable.) And by that point telegrams had faded into irrelevance. Rather than being integrated into email, they were forgotten entirely.

The end. Long-distance telephony made the telegram a niche product, still useful for certain business transactions requiring written proof that couldn’t be provided in a phone call. In the 1980s fax machines took up most of that niche, for the price of a phone call, which cost much less than a telegram. By that point pretty much the only thing you could do with a telegram that you couldn’t do more cheaply elsewhere was send money. And why bother maintaining a general-purpose communication network for that?

So Western Union went about a major financial restructuring in which practically all its assets were sold off leaving only a payment network. MCI bought the fiber-optic network, Hughes took over the Westar satellites (how did the perpetually cash-strapped Western Union launch satellites?), TelecomUSA acquired the long-distance telephone services (and was promptly bought up by MCI) and the email and telex services, branded as “EasyLink”, went to AT&T in 1990. FCC filings related to this last sale, which ended Western Union’s status as a telecommunications company, indicate that EasyLink ran over leased lines belonging to TelecomUSA, so I take it that means MCI ended up with WU’s copper wire network too. And also that AT&T moved EasyLink to its own physical network as soon as it could.

Up until 2006, you could still go to a Western Union office and send a telegram over, I assume, the EasyLink network. These were “letter telegrams” only and usage had fallen to a few dozen a week when WU finally discontinued them. Shortly after that, a company calling itself iTelegram claimed to have taken over the telegram service from Western Union. They appear to be legitimate enough but I’m not sure what if anything they got from Western Union or how they’re transmitting messages. (Wikipedia claims that iTelegram bought the telegraph network in 1991, but I can’t find any sign that they existed prior to 2006. All evidence I can find points to iTelegram being a subsidiary of Telegrams Canada, owned by one Colin Stone of Toronto, who similarly “took over” the former CNCP telegram service when its then-owner AT&T shut it down in 2002.)

What of EasyLink? It spun off from AT&T, changed owners a few times, and was acquired by OpenText in 2012. The last annual report by EasyLink as an independent company shows it as being in the EDI business with telex a small and declining part of their business. OpenText’s annual reports and their website don’t mention telex at all.

Other countries’ telex/telegraph systems have gradually shut down. The ITU list of telex agencies hasn’t been updated since 2011 and still lists EasyLink as an AT&T subsidiary, which it wasn’t by then. Reading the ITU bulletins, there are a few announcements of a Swiss company called UTS buying systems from former state monopolies like BT and NTT, and they still seem to be in business. A few countries have simply closed their systems with no replacement.

Radio telex still exists and is used for communicating with ships at sea. Satellite connections are more modern but also high-latency and expensive; given the low bandwidth in the frequency bands that carry out to sea, it’s unlikely that any ground-based data link more featureful than radio telex will happen. Wired telex used to have direct interfaces with radio telex and this was its main remaining use. Maybe it still is. As usage becomes scarcer, it becomes harder to find current information on how it’s still used.

And of course ASCII, with its odd control codes meant for Teletype machines, is still around. Since this post is in ASCII, maybe this can be considered a form of telegraphy? Maybe the internet can? Hell I could probably run a service that prints emails and sends them via snail mail, cheaper than iTelegram and just as fast, but would that be a telegram? And if not, iTelegram is probably sending their messages over the internet (maybe encapsulated X.25 to be authentic) so is that a telegram?

I was hoping to find a definitive ending, or perhaps a limping continuation. Instead all I find is Eliot’s “not with a bang but with a whimper.”

© 2020-21 Bronx River Software

Powered by Hugo & Kiss.