Colossus: Face To Face With The First Electronic Computer

When the story of an invention is repeated as Received Opinion for the younger generation it is so often presented as a single one-off event, with a named inventor. Before the event there was no invention, then as if by magic it was there. That apple falling on Isaac Newton’s head, or Archimedes overflowing his bath, you’ve heard the stories. The inventor’s name will sometimes differ depending on which country you are in when you hear the story, which provides an insight into the flaws in the simple invention tales. The truth is in so many cases an invention does not have a single Eureka moment, instead the named inventor builds on the work of so many others who have gone before and is the lucky engineer or scientist whose ideas result in the magic breakthrough before anyone else’s.

The history of computing is no exception, with many steps along the path that has given us the devices we rely on for so much today. Blaise Pascal’s 17th century French mechanical calculator, Charles Babbage and Ada, Countess Lovelace’s work in 19th century Britain, Herman Hollerith’s American tabulators at the end of that century, or Konrad Zuse’s work in prewar Germany represent just a few of them.

So if we are to search for an inventor in this field we have to be a little more specific than “Who invented the first computer?”, because there are so many candidates. If we restrict the question to “Who invented the first programmable electronic digital computer?” we have a much simpler answer, because we have ample evidence of the machine in question. The Received Opinion answer is therefore “The first programmable electronic digital computer was Colossus, invented at Bletchley Park in World War Two by Alan Turing to break the Nazi Enigma codes, and it was kept secret until the 1970s”.

It’s such a temptingly perfect soundbite laden with pluck and derring-do that could so easily be taken from a 1950s Eagle comic, isn’t it. Unfortunately it contains such significant untruths as to be rendered useless. Colossus is the computer you are looking for, it was developed in World War Two and kept secret for many years afterwards, but the rest of the Received Opinion answer is false. It wasn’t invented at Bletchley, its job was not the Enigma work, and most surprisingly Alan Turing’s direct involvement was only peripheral. The real story is much more interesting.

To Bletchley, Where Miracles Happen

General Heinz Guderian overlooking the operation of an Enigma machine during the battle of France. Bundesarchiv, Bild 101I-769-0229-11A / Borchert, Erich (Eric) / CC-BY-SA 3.0, via Wikimedia Commons
General Heinz Guderian overlooking the operation of an Enigma machine during the Battle of France. Bundesarchiv, Bild 101I-769-0229-11A / Borchert, Erich (Eric) / CC-BY-SA 3.0, via Wikimedia Commons

At this point we’re going to take you to Bletchley, to the modern-day Bletchley Park site and the National Museum Of Computing which occupies one corner of it. The museum has a fascinating collection, of which two galleries are of interest to us here. The first is their Tunny gallery, which explains the context and sequence of events which led to Colossus, and the second is their Colossus gallery, which contains their fully functional replica of a MkII Colossus computer.

The most famous Nazi encoding system is the Enigma, with its portable machines resembling typewriters becoming a ubiquitous symbol of the codebreaking efforts. This was the code of German military combat units in slightly different forms by all services, and photographs show them being operated from forward positions or in mobile signals units.

Enigma was not however the only German  encoding system in use and intercepted by the Allies, and by no means the only one on which the staff at Bletchley Park were employed. The first section of the museum’s Tunny gallery explains the Lorenz cipher, which was used for secure communication at a much higher level between German high command outposts, encoding teleprinter traffic in real time. It had a superficial resemblance to Enigma in that it employed a rotor system, but instead of Enigma’s system of through-wired contacts its rotors produced a pseudo-random binary sequence that was XORed with the binary teleprinter traffic to produce an encrypted output. Also unlike Enigma the codebreakers did not have the benefit of a captured machine to study until very near the end of the war, so their only means to understand it came from intercepted messages using it.

The Fish That Shortened The War

The museum's Tunny machine
The museum’s Tunny machine.

The museum takes the visitor through the listening stations and how the frequency-shift-keyed teleprinter traffic was recorded on paper tape and hand-transcribed, before transporting them to the cryptoanalysts at Bletchley Park and their efforts to glean the workings of the system. The breakthrough came as a stroke of luck in August 1941 when an operator in Athens sent the same 4000 character message twice with the same settings on his Lorenz machine, providing the reduced odds of decryption that the Bletchley staff needed to eventually decode it. Using these two ciphertexts and the mechanics of their decoding the mathematician Bill Tutte was then given the task of deducing the operation of the machine, which by early 1942 he had completed. The resultant work was given the codename “Tunny”, after the codename for the Athens communication circuit which had provided the breakthrough. All such links took their codenames from types of fish.

At the centre of the museum’s Tunny gallery is their rebuilt Tunny machine, a British electromechanical reproduction of a Lorenz cipher machine produced by the Post Office Telephone research facility at Dollis Hill, London. It could be set with a plugboard equivalent of the Lorenz’s rotor settings and decode messages, but it still required those rotor settings to be available. The effort to automate the discovery of some of those rotor settings resulted by mid-1943 in a machine called the “Heath Robinson”, after the British cartoonist who drew intricate and complex machines performing simple tasks. If you haven’t heard of him but you are aware of Rube Goldberg, you’re on the right track.

The rebuilt Heath Robinson machine.
The rebuilt Heath Robinson machine.

The museum have recreated a Heath Robinson next to their Tunny, and like the original it keeps a pair of long punched paper tape loops under tension with a system of pulleys. One holds a ciphertext while the other has a sequence of possible settings for one set of rotors, and a set of logic derived from the Tunny machine can be fed the ciphertext automatically along with each of the rotor settings in turn. The resulting output was then used to produce collections of rotor settings that could dramatically shorten the odds for the teams of cryptoanalysts.

Once the visitor has been shown both machines in operation, the guide shows a section of tape that has been mangled by the Heath Robinson’s  mechanism, and it is explained that the machines were slow and unreliable. In particular a close synchronisation between the two tapes was essential to its operation, something they could easily lose. He then tells the story of how Alan Turing had recommended the engineer with whom he had previously collaborated on the Enigma work to the Heath Robinson team, and that this was Turing’s only direct contribution to Colossus.

Tommy Flowers was Head of the Switching Group at Dollis Hill, and it was his ideas on how the Heath Robinson’s paper tape sequences of Tunny rotor settings could be generated electronically using thyratrons that would result in the machine that became Colossus. The cypher text was still read from a punched tape, but it was fed into a programmable function could be performed electronically upon it against the thyratron-generated rotor settings. It was not yet a general purpose stored-program computer as we would know it today, but if fulfilled the description of being a programmable all-electronic digital computer.

In The Presence Of Greatness

The front of the museum's Colossus MkII.
The front of the museum’s Colossus MkII.

Walking into the museum’s Colossus gallery as one of the first groups of weekend visitors, we were lucky enough to see it being brought into life. Their Colossus is a replica of a MkII machine completed in 2007, and it stands alone in the centre of the room with the only intrusion a set of discreetly placed safety barriers to keep the public away from high voltages. There are two long parallel racks that would be close to ceiling height if they were not in a wartime hut without a flat ceiling, both studded with the thousands of octal tubes. At the far end is a paper tape reader similar to that of the Heath Robinson, close to the middle are the plugboards and switches through which the machine is programmed, and at the end closest to you is the teleprinter which records the result.

The machine is powered up slowly to reduce thermal shock and prolong the life of its tubes. Our guide told us it only needs a single digit number of tubes replacing in a typical year, which is impressive considering how many it has. Once it gets under way the slight morning chill of the room is replaced by a significant heat from all those tube filaments, and though the machine is quieter than you might expect there is a whir and cyclic clicking sound from the tape reader.

Standing in the same room as the seminal machine of your art is an interesting experience for an engineer, even when it is a replica. Some of the other visitors seemed to be there because of its association with The War rather than because of its technological significance, but it was interesting to see that we were not the only ones who had evidently wielded a soldering iron or two. It is a moment to reflect on how far we’ve come in over seven decades,  to silently praise the memory of the people who built it and — despite Colossus itself being shrouded in secrecy –praise the influence of their work on the machines that followed it.

A short video of a walk round the machine in action is below, from it if you will excuse the mobile phone video quality we hope you can get an impression of its size and complexity.

The genuine Lorenz machine at Bletchley, on loan from the Norwegian Armed Forces Museum. The codebreakers did not see one of these machines until the end of the war, which is why we have placed this picture at the bottom of our write-up.
The genuine Lorenz machine at Bletchley, on loan from the Norwegian Armed Forces Museum. The codebreakers did not see one of these machines until the end of the war, which is why we have placed this picture at the bottom of our write-up.

The National Museum Of Computing houses a fascinating collection of vintage and historic computers beyond its work on the wartime machines covered here, and is well worth a visit should you find yourself in the vicinity of Bletchley. It is operated by a charitable trust and relies on its very affordable admission fees and voluntary donations for its continued existence. Put it on your itinerary immediately!

44 thoughts on “Colossus: Face To Face With The First Electronic Computer

      1. By the way if you’re wondering why it was kept secret so long, well post WWII there were a lot of 2nd and 3rd world countries buying up ex-nazi cypher equipment, or having ex-nazis build them some. Some of them might even have been using the pre-war commercial version of the Enigma still. Anyway, it was reason enough to keep the whole of Bletchley’s achievements secret for decades.

          1. Probably not:
            (1) Stalin didn’t trust the West in particular his own spies who were telling him what was happening.
            (2) The USSR was given repeated direct warnings from the UK about the German invasion, for example, and ignored them. It lost most of its airforce on the ground, avoidably.
            (3) Within the UK the risk of compromise was such that findings were not used until a plausible other source could be faked in the media to justify any successes. Any leakage of the exact mechanism would result in a change in German equipment and a major disaster for all the allies with many more lives lost.
            (4) Even so, I understand that Tunney based intercepts played a role in the Battle of Kursk.
            (5) The chances of Stalin altering his policies if told that “we have this thing that does millions of calculations and can break German codes, which is why we can give you this intelligence” {which the USSR already knew} would be zero and just provide another leakage path to degrade the secrecy.

      1. Harlan Ellison, the author brought suit against The Terminator production company Hemdale and distributor Orion Pictures for plagiarism, not the Outer Limits franchise. It was settled out of court but James Cameron emphatically denied Ellison’s allegations and was opposed to the settlement.

    1. Well, I need to disagree. More than 100 years before Herman Hollerith, Jacques de Vaucanson was trying to automate textile production with punched cards, refined about 50 years later (but still far before Hollerith) by Joseph Marie Jacquard. It’s a pity that Wikipedia’s article “Punched card” does not mention them, but given Wikipedia’s quality, I’m not surprised at all.

      1. The great thing about Wikipedia is that if you have a reliable source for the information, you can update the page and add the information yourself. Then everyone can know what you know.

    2. Well, I have to disagree. More than 100 years before Herman Hollerith, Jacques de Vaucanson was trying to automate textile production with punched cards, refined about 50 years later (but still far before Hollerith) by Joseph Marie Jacquard. It’s a pity that Wikipedia’s article “Punched card” does not mention them, but given Wikipedia’s quality, I’m not surprised at all.

  1. I was fortunate enough to visit both Bletchley Park and the National Computing Museum earlier this year, and I agree they are both fascinating places. Whilst they occupy neighbouring buildings, they are separate entities, and purchasing entry to one does not allow entry to the other, sadly. We also spent too long wondering around Bletchley and left ourselves far too little time to fully take in the computing museum. Fortunately, one of the volunteers at the museum showed us around, unlocking doors and powering up various ancient hardware as he went. If you’re in that part of the UK, they’re both well worth a visit.

    1. Indeed. I was in the Milton Keynes area for work late last year (i’m from Melbourne, Australia, so it was an opportunity i couldn’t pass up!) and it’s definitely worth a visit. There was an old dude there who gave me and another couple visiting a run down on how the Colossus worked and showed it running and it is absolutely fascinating the things they came up with what they had available to them – they were definitely ahead of their time.

      Last I heard, there’s some political issues in respect to the reasons why the Colossus being separate to the rest of Bletchley park which is unfortunate – hopefully they sort those issues out.

  2. Judging from the way Guderian is focused on what the third guy is doing, I’d say that rather than “overseeing” the decryption process, he’s very interested in the message, and is reading it letter by letter as it’s being decrypted.

    1. I’d say the opposite. The fellow at the machine is focused on a notepad and is neither looking at the machine nor leaning closer to it to see the decoded letter’s lamp, as it is being operated outdoors in broad daylight.Also he appears to have his hand ready to type the next letter.

  3. I too highly recommend visiting Bleachly park and Computor center, and the Colossus story fascinate me. I spent quite a lot of time talking to one of the operators when I visited a few years ago (since there where few visitors that day), and a friend who isnt really interested in electronics/machines/technical history where quite amused with me geeking out over it.

  4. I went to Bletchley, years ago, before it was a “thing” :-)

    For a long while, believe it or not, there was significant risk of the whole place being bulldozed and turned into a housing development (as it partially has been). When I was there (in the 90s), there was much more open space, a small airstrip and control tower, as seen in this photo — compare with Google Earth to see what’s been developed since:
    http://www.colindaylinks.com/bletchley/images/Bletchleymap.jpg

    It is a tribute to the hard work of the small group who worked to get the Park protected and repaired that anything remains for visitors today. Thankfully, some large corporations (including Google) and many individual donors stepped up to help save it. Donations are still being accepted, and are much appreciated:
    https://www.bletchleypark.org.uk/content/support/donate.rhtm

    1. Last time I was there, I didn’t see crap, just the outside of the huts, at an event in the grounds, I swear it was still under MoD then and had an element of Royal Signal Corps guarding the gate. I thought that was 90s also, but may have been late 80s. I wanted to go peek in windows but was warned off by event staff.

  5. TNMoC is well worth a visit for two reasons.

    It goes without saying that the exhibits are superb – and often working.

    More importantly, interestingly and almost uniquely, the “attendants” actually know a *lot* about the exhibits. For example, when I was there I started chatting about the first computer I used (Elliott 803, 8Kwords, 39bit words, 576us cycle time) – and the person I was talking to whipped out the original circuit diagrams (>A0 size “blueprints”) and we started discussing them.

    1. Excellent, if I was a gambler then the chances are I’d now owe my editor a fiver! :)

      There’s a simple reason it’s vertical framed, there just wasn’t enough room for me to get far enough away from it to get the whole machine’s height in a horizontal frame on my mobile phone.

  6. I got a Monopoly game “Turing Edition” from the Bletchley Park foundation to help support it. They may still be available.

    The first computer I worked on was a Stantec Zebra machine that my school was given by the British American Tobacco company when they upgraded. We had to disassemble and reassemmple it ourselves. It was a hybrid tube/semiconductor machine and we were able to run simple paper tape programs on it. The best part was putting it together and it actually working. Unfortunately it consumed around 6kw and so was expensive to run, so it would not be on all the time. That meant that the tubes would often burn out so it wasn’t used a lot. He is a link to more info: http://archive.computerhistory.org/resources/text/Standard/Stantec.Zebra.1961.102646082.pdf

  7. What’s this, no mention of the Atanasoff–Berry computer, completed in 1942? construction started all the way back in 1939, and it had been conceived of in 1937! Colossus was certainly a programable Turing complete machine, while the ABC was more restricted in function, but the ABC was certainly the first fully automatic electronic digital computer.

    It’s only mechanical component, beside Input and output devices, was a spinning drum dynamic memory, made up of numerous capacitors that were read and refreshed, and written to when required, with each drum spin. All logic was electronic, using tubes and based on the modern convention of binary arithmetic, where many other early machines used BCD arithmetic.

    Iowa State College holds the title of first electronic digital computer.

    An interesting footnote in history… The ABC had been examined by John Mauchly, chief designer of the ENIAC, in June 1941. He had filed the patent for the electronic computer in 1947, which the patent office issues in 1964. In 1967, Honeywell sued Sperry Rand to have the patents invalidated. The US District Court of Minnesota invalidated the patents on October 19, 1973, on the grounds that the ABC constituted a prior work, and that Mauchley had been given sufficient access and understanding during his visit to render Sperry Rand’s patent claims to be non-original.

    The date is significant… It was utterly overshadowed in the news by the “Saturday Night Massacre”. For those who don’t know US politics, that’s the name given to the Saturday when Nixon ordered his attorney general to fire the prosecutor that had been prosecuting him during the Watergate scandal. The AG and deputy AG both resigned, refusing to perform such an absurd act. The Solicitor General finally did it, but claiming to not have been aware of the second resignation, and almost resigning as well. The news of the event exploded, so all newspaper coverage was focused on that event. As a result, the verdict int he Honeywell vs Sperry Rand trial was virtually unknown…

    Thing is, it holds one more element of significance. The invalidation of the Sperry Rand patents for the electronic computer in 1973 paved the way to a free and open setting for the first persona computers to be created. No one had to get a license from Sperry Rand to do this. You could just build an electronic computer, and you were beholden to no one!

    the ABC was never patented, and it’s existence as first electronic computer served as the prior work to invalidate Sperry Rand’s hold over the computer market. Who knows if the personal computer revolution would have happened the same way, had Sperry Rand been breathing down the necks of the likes of Ed Roberts (inventor of the Altair), or Chuck Peddle (creator of the KIM-1) or Jobs and Wozniak (you all know them :P ).

    Something to think about!

    1. Yes, they were. But article doesn’t claim, that Bletchley Park team was first.
      Germans modified Enigma several times during WW2. At some point messages encrypted by those devices were impossible to break with polish methods and therefore huge chunk of Turing’s work was completely original.
      Teletype ciphers (also mentioned in article) were never broken (or AFAIR even examined) by polish team.
      Of course I agree – article should at least mention them.

  8. Very nice article. Thanks for writing it.

    One point worth noting about the (in)famous Tunny message: the reason it was useful in cracking Lorenz is because it was *not* an exact repetition. An exact duplicate would have provided no additional information. Instead, the operator got sloppy and re-transmitted the message with several abbreviations and other small alterations. This gave the cryptanalysts extra information, known as “depth”.

    Another interesting point: the operator re-sent the message because it was not received correctly. The receiving station sent an *uncoded* request asking for the message to be re-sent.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.