The Other First Computer: Konrad Zuse And The Z3

Bavarian Alps, Dec. 1945:

Since 1935, Berlin engineer Konrad Zuse has spent his entire career developing a series of automatic calculators, the first of their kind in the world: the Z1, Z2, Z3, S1, S2, and Z4. He accomplished this with a motley group of engineers, technicians, and mathematicians who were operating against all odds. With all the hardships and shortages of war and the indifference of their peers, the fact that they succeeded at all is a testament to their dedication and resourcefulness. And with the end of the war, more hardships have been piling on.

Two years ago, during the Battle of Berlin, bombers completely destroyed the Zuse family home and adjacent workshops on the Methfesselstraße, where they performed research and fabrication. All of the calculators, engineering drawings, and notes were lost in the rubble, save for the new Z4 nearing completion across the canal in another workshop on Oranienstraße. In the midst of all this, Zuse married in January of this year, but was immediately plunged into another crisis when the largest Allied air raid of the war destroyed the Oranienstraße workshop in February. They managed to rescue the Z4 from the basement, and miraculously arranged for it to be shipped out of the Berlin. Zuse, his family, and colleagues followed soon thereafter. Here and there along the escape route, they managed to complete the final assembly and testing of the Z4 — even giving a demonstration to the Aerodynamics Research Institute in Göttingen.

On arrival here in the Bavarian Alps, Zuse found a ragtag collection of refugees, including Dr Werner Von Braun and a team of 100 rocket scientists from Peenemünde. While everyone here is struggling just to stay alive and find food and shelter, Zuse is further worried with keeping his invention safe from prying eyes. Tensions have risen further upon circulation of a rumor that an SS leader, after three bottles of Cognac, let slip that his troops aren’t here to protect the scientists but to kill them all if the Americans or French approach.

In the midst of all this madness, Zuse and his wife Gisela welcomed a baby boy, and have taken up residence in a Hinterstein farmhouse. Zuse spends his time working on something called a Plankalkül, explaining that it is a mathematical language to allow people to communicate with these new machines. His other hobby is making woodblocks of the local scenery, and he plans to start a company to sell his devices once the economy recovers. There is no doubt that Konrad Zuse will soon be famous and known around the world as the father of automatic computers.

Meanwhile, Back in 2021

It was 80 years ago that Konrad Zuse demonstrated his model Z3, an appropriate occasion to look back at his life and accomplishments. But despite the musings of our imaginary correspondent above, most people outside of computer history buffs have never heard of Konrad Zuse. How is it that someone created and built the world’s first programmable computer and yet is only a footnote in history?

Konrad Zuse had trouble finding his niche in life. As a young teen he seemed destined to become an engineer based on his interests and skills, always tinkering with erector sets and inventing gadgets. But he was also a talented sketch artist, photographer, and enjoyed acting in skits and plays. During his time at the Technical University of Berlin, he changed majors twice. Initially put off by the rigid conformity of mechanical engineering, he switched to architecture, only to find it equally boring. He finally settled on civil engineering, which struck the right balance between engineer and artist.

Zuse in 1935

After completing his studies in 1935, his first job was at the Henschel Airplane Works just outside of Berlin. But he quit after only a few months, turning his attention to something he had been pondering for some time — could he build a machine to automate those boring and repetitious engineering calculations?

How Should It Work?

Zuse was working in almost complete isolation from others in the nascent field of computing. And initially, at least, he wasn’t so much focused on a general purpose computer as we think of it today. His goal was to make an automated mechanical adding machine of sorts, to handle the tedious calculations such as those required in static stress analyses. Early on, he made a key decision to use binary rather than decimal. He rightly saw that this would greatly simplify the building of such a machine, no matter what the implementation.

You might think that the choice of binary numbers would immediately make relays the most obvious choice. And while he did experiments with relays, they were ruled too unreliable. Today we take for granted the reliability and quality of relays, but Zuse was living in 1930s Germany and operating on a shoestring budget supported by his family and friends. Relays were expensive, and they were designed to switch infrequently, not run continuously at many Hertz. The relays that he could obtain were more often than not discards his friends rescued from the trash at the local telephone exchange. Along with his friend Helmut Schreyer, they also experimented with using of vacuum tubes, but realized that technology would be significantly delayed due to wartime shortages.

Zuse’s Mechanical XNOR Gate

So how did he implement a computer without relays or vacuum tubes? It was all mechanical. Zuse devised a clever arrangement of plates, flat bars, and steel pins to make a digital memory store and a floating point binary arithmetic logic unit (ALU).

Print Your Own Z1 Mechanical Logic Gate

This video explains the basic operation of the mechanical adder (don’t let the fact that it’s in German discourage you from watching it).

If you want to dig into how Zuse implemented memory and an ALU mechanically, check out [fjkraan]’s project explaining the basics and OpenSCAD 3D printer examples you can make yourself:

Z1 Specs

The memory consisted of 64 each 22-bit words ( 14-bits of mantissa and 8-bits of exponent and sign). The ALU could add, subtract, multiply, and divide. The program instructions were provided on punched tapes, and consisted of only eight operations:

  • Lu input a decimal number
  • Ld output a decimal number
  • Pr z read from address z into register r
  • Ps z store from register r into address z
  • Ls1 add two floating point numbers, R1 = R1 + R2
  • Ls2 subtract two floating point numbers, R1 = R1 – R2
  • Lm1 multiply two floating point numbers, R1 = R1 * R2
  • Li1 divide two floating point numbers, R1 = R1 / R2

The computer’s clock was equivalent to a four-phase CPU clock of modern times. But it was mechanical rather than electrical, and could be driven by either a hand crank or an electric motor. Each clock phase was one of four planar motions — imagine pushing a square plate a few centimeters North, then West, then South, then East back to the starting point. Each phase performed a basic operation, and fortunately no basic operation took more than three of these cycles.

As hinted to by the instruction set, Zuse also built converters to permit the use of human-friendly decimal numbers for input and output. Paper tape was too rare a commodity at the time, so another solution was needed. They found that blank 35 mm movie film worked just fine, and wrote their programs using a hand-held hole punch.

New Models Abound

The Z1 was finished in 1938, but was never very reliable. But it wasn’t meant to be a finished product, only a proof-of-concept prototype. Work next began on the Z2, which was planned to test whether relays could would be suitable. Zuse kept the mechanical memory for the Z2, but replaced the ALU with a relay version, but only operating on fixed-point numbers. It was completed and successfully demonstrated in 1938, whereupon Zuse received partial funding to build a successor model.

The Z3 was designed entirely with relays. It was basically an improvement on the Z1, was faster and more reliable. It was delivered and demonstrated on 12 May 1941, and then Zuse began construction of yet another computer, the Z4 mentioned in the introduction. The Z4 expanded on the Z3, adding new features like square root, punch tape output, multiple tape input units, and a conditional branch instruction. Zuse continued to use relays in the ALU, but returned to the mechanical memory because of its greater density. The Z4 was eventually put into full time service and operated from 1950 until 1959.

Analog to Digital Converter to Measure Wing Curvatures

As the Z3 project was underway, a request came from Henschel for help in the production of guided missile wings. Zuse built what are probably the first process control computers, the S1 and S2, and invented an analog to digital converter. These single-purpose computers followed the contour of newly-manufactured wings as a hundred analog-to-digital-equipped rollers measured their curvatures.

The machines would then automatically calculate the required trim for the tail surfaces and ailerons from this data, a task that previously been done by a dozen computers (people, not machines). After performing their jobs around the clock for two years, the factory was destroyed by bombing in 1944, and the S1 and S2 were either destroyed or possibly captured by the Soviets.

If you want to learn more about each model, check out this extensive website describing all the computers and much more. It was made by Konrad Zuse’s son who was born in the introduction above, Prof Horst Zuse, who became an electrical engineer, a software engineer, and an expert on computer history.

Logic and Math

Zuse and his team were able to switch back and forth between mechanical, relay, and even vacuum tube realizations of the computer components because of Zuse’s grasp on the underlying equations at work. He developed his own system of rules and notation, until his former math teacher told him he had reinvented propositional calculus, or Boolean algebra as is more commonly used today. By reducing the computer operations into equations, the design concepts were separated from the physical construction, making it easier to jump back and forth between implementations.

Zuse realized that with the framework, what we would call a computer language today, the computer could do so much more than just solve equations. He had been thinking of such a language since 1939, and in the immediate aftermath of the war, he spent much of his time formalizing what he called Plankalkül, the first high-level programming language for a computer — in English, it translates as a formal system for planning. Despite Zuse’s high hopes, Plankalkül never took off, although some say that elements of it can be found in Algol 58, albeit uncredited.


Patent Problems and Growth

Zuse made patent applications for many of his inventions, and a number of patents were issued. His patent for the computer itself, would be frustrated. He filed a patent application on 16 Jun 1941. It wasn’t until after the war in 1952 that the German Patent Office published his claims. No company in the whole industry objected, except for one — the Triumph calculator company filed opposition to the patent, later revealed to be backed by IBM. The court case lingered for years, only to ultimately be rejected in 1967 on the grounds that the invention was not worthy of a patent.

The innovation and progressiveness of the object concerned in the main application are not doubted. Yet a patent cannot be granted due to insufficient inventive merit.

In the meantime, Zuse had begun commercializing his own line of computers. There was a false start with IBM, who wanted the patents but wouldn’t let Zuse continue development. Remington Rand became an important early client, followed closely by the Swiss Federal Institute of Technology (ETH) in Zürich who took delivery of the Z4 in 1950. Business grew over the next 15 years, with a series of computers and even a plotter, the Graphomat 64. But by the mid 1960s the company was having financial troubles, and it was acquired by Siemens in 1967.


As with many historical computers that have been destroyed or sold for scrap, there have been several recreations of Zuse’s computers built over the years.

Replica of the Z1 Computer

With support from Siemens, Zuse began a project in 1984 to reconstruct the Z1. It was completed in 1989 and is now on display at the German Museum of Technology in Berlin. Prof Raúl Rojas of the Free University of Berlin has written an excellent, in-depth paper on how the Z1 and its reconstruction work. Regarding this project, Zuse once remarked:

Back then, it didn’t function very well, and in that regard, this replica is very reliable — it also doesn’t work well.

Z3 Replica by Zuse in 1961

The relay-based Z3 has been reconstructed three times. First, in 1961, Zuse himself built a replica that is still on display in the Deutsches Museum in Munich. Again in 1997, Profs Raúl Rojas and Horst Zuse built a recreation that is now on display Konrad Zuse Museum in Hünfeld, Germany. And most recently, Prof Horst Zuse built an original size reconstruction which was installed at the Konrad Zuse museum in 2010.

Zuse as a LEGO Brick Man

And speaking of replicas, this is a replica of Konrad Zuse himself, in the form of a LEGO brick man, holding a relay storage unit from the Z3 computer.

Lack of Recognition

This month, the German Patent Office wrote about Zuse and had this to say:

Zuse had the misfortune to have made his most important invention in times of war. Otherwise, the practical, entrepreneurial Zuse might be known today not only as a technology pioneer, but also as the founder of a global technology corporation of the stature of Bill Gates.

It does make you wonder why Konrad Zuse isn’t well known and why his groundbreaking inventions and computers are often ignored. Whether or not he made the first programmable computer, his contributions were undeniably significant, well ahead of their time, and deserve better recognition. Konrad Zuse’s accomplishments are best summarized by Eric Weiss, who wrote Zuse’s obituary for the IEEE in 1996 :

Zuse was a man of many talents. He was a persistent, innovative, and extraordinarily creative engineering designer and builder, a mathematical logician, the founder of several successful computing manufacturing businesses, and an impressive artist. His countrymen and the computing world will remember him chiefly for the concepts of the first Z machines but also for his entrepreneurial companies, successfully created and operated, almost without outside help under the most trying conditions.

Below the break is a 1958 television broadcast with Zuse (English subtitles are available) and some additional resources about Konrad Zuse.


To Learn More

43 thoughts on “The Other First Computer: Konrad Zuse And The Z3

    1. Inventors are often unaware of parallel developments in other places, just as most people haven’t really heard about all the historic figures that helped us develop today’s technology.
      A great enjoyable read is Konrad Zuse’s “The Computer – My Life” (1993) for the English edition or “Der Computer – Mein Lebenswerk” (2010) for the German edition. It’s a gem of a book, honestly written in a humble style, giving lots of historic background and both technical and non-technical information.
      In a documentary, Zuse’s son also pointed out that his father, Konrad, was perhaps more of a scientist than a businessman. This fact is probably valid for many tinkerers today.

      How the media intentionally pushed certain characters while totally omitting crucial figures is also visible in the example of “Who decrypted Enigma?”. Most have heard about Alan Turing, but nobody heard of Marian Rejewski and other mathematicians in Poland and France who laid most of the ground work without which Turing and his team wouldn’t have managed to decrypt Enigma in time. “Who does all the work and who gets the credit” could be a fitting headline for this article as well and possibly for a series of articles. Maybe it’s not so much about ‘who deserves the most credit’, but about getting to know all or at least some of the lesser known incredible minds who are just as important.

      Konrad Zuse is a giant. The question who invented what first seems to be closely related to a political rat race and state PR. Who invented the computer? Was it Leibniz or Babbage; later, was it Konrad Zuse or Howard Aiken?
      To quote his biography: “Another visit brought us to Harvard University in Cambridge, not far from Boston. Professor Aiken personally showed us his devices. But his interest in our work was not very keen. At Harvard they were still completely convinced that the computer was an American invention. It took a bit of convincing to get our American discussion partners to see that we did not have to hide our light under a bushel either. Aiken later recognized this, and said so (Fig. 7.1). We also had some differences of opinion on technical matters. Aiken advocated the decimal system and had developed very beautiful codes for decimal numbers using several bits. I was more a proponent of pure binary representation – in any case, at least where large scientific computers were concerned.” Letter from Howard Aiken to Konrad Zuse (1962)

      For a long time Zuse wasn’t really mentioned or even intentionally omitted, especially in American media. However, in a documentary by the American Computer History Museum, Zuse gets a few _seconds_ of coverage in the very beginning: (Computer Pioneers: Pioneer Computers Part 1)

      There is a similar debate over who invented the transistor at the same time on different continents. Who got the credit? To quote a great article by the IEEE:
      “As was true for the Bell Labs transistor, invented by John Bardeen and Walter H. Brattain in December 1947, the technology that led to the transistron emerged from wartime research on semiconductor materials, which were sorely needed in radar receivers. In the European case, it was the German radar program that spawned the invention. Both Mataré and Welker played crucial roles in this crash R&D program, working at different ends of the war-torn country.”

      Hans Camenzind, the Swiss inventor of the 555 timer, wrote a whole book about inventors and their disputes “Much Ado About Almost Nothing – Man’s Encounter with the Electron” (2007) where we can read about the battles of Galvani vs Volta and many many others.

      We’re standing on the shoulders of giants and history provides a nearly endless source of stories about mad inventors, engineers and entrepreneurs and questionable competition and corporate battles.
      In 1957, Sputnik, the first orbiter, was successfully launched by the Soviet Union which put a bit of pressure on the US to compete in the technological sector which in turn spawned a fast reaction which resulted in funding and programs.
      When Shockley hand-picked Robert Noyce and many other great talents for his new team, he didn’t expect that later the ‘traitorous eight’ would leave Shockley to start up Fairchild Semiconductor in 1957, which basically was the first big boost for Silicon Valley, as in time people left Fairchild to create own companies (the Fairchildren).

      Later, even more surprisingly, the late Noyce created another company, Intel, where without Gordon Moore, Federico Faggin and Masatoshi Shima things wouldn’t have worked out. Not much later Faggin left Intel to create his Zilog bringing forth the epic Z80. A detailed account of these thrilling developments can be found in Michael Malone’s book “The Microprocessor: A Biography” (1995).

      The movie “Pirates of Silicon Valley” outlines many corporate stealing strategies, based on the book “Fire in the Valley”. I.e. how DOS was bought from a small Seattle company or how the graphical user interface was stolen from Xerox.

      Domesticated Primates (humans) learn by copying existing stuff.

      Playfulness is a main aspect of hacker or maker culture. See “Hacking Europe – From Computer Cultures to Demoscenes” (2014).

      There is nothing new under the sun, but there are many bright engineers in every corner of the planet, how nice would it be if schools and teachers stopped actually discouraging pupils and thus sabotaging motivation & learning for life. ‘Why nerds are unpopular’ is a chapter in “Hackers and Painters” (2004) by Paul Graham.
      And how progressive could humanity be, if governments actually understood technology to sufficiently promote and support crazy inventors, hacker hippies, uptight scientists, funny makers, or other humorous weirdos that indeed have been the cause for our giant leaps in technology so far that make life easier. Surely, in an increasingly capitalist and human-loathing reality, this will become harder when everyone’s struggling to make ends meet.

      Related to Konrad Zuse, there is a short 30min documentary for people who understand German:

      Konrad Zuse built the first digital freely programmable computer, the Z1. Built the first functional program-controlled computer, the Z3. The Z3 was proven to be Turing-complete in 1998. Produced the world’s first commercial computer, the Z4. Designed the first high-level programming language, Plankalkül.

  1. “Each clock phase was one of four planar motions — imagine pushing a square plate a few centimeters North, then West, then South, then East back to the starting point.”

    Early Konami code.

    “Zuse had the misfortune to have made his most important invention in times of war. Otherwise, the practical, entrepreneurial Zuse might be known today not only as a technology pioneer, but also as the founder of a global technology corporation of the stature of Bill Gates.”


    1. Zuses “misfortune” was that our Gröfaz did not believe in stuff that does not kill in spectacular ways. No Wunderwaffe, no money. So no funding. Wernher v. B. build impressive rockets that could kill spectacular, so funding.

      OTOH Zuse worked at Henschel & Son, helped to build the Henschel Hs 293 (radio controlled bomb). He was exempted from military service and allowed to fund an engineering office with 20 staffs during wartime. His work was financed with 250.000 RM by some NS institutions. So there was recognition how important his work might be.

      He was, like W.v.B., another wheel in the fascist killing machine.

      1. It’s clear that Jan Praegert isn’t aware of the historic context related to Zuse at all. During his studies in Berlin, Konrad Zuse was part of a usual student’s society which regularly conducted small theatre plays in which the young Zuse also took part enthusiastically. In such a intellectual environment it was only natural for young artists to be critical of totalitarian leaders and their dark ambitions. In one such play they made fun of the Fuehrer _before_ the rise to power.
        In a 90s interview/documentary we can also see the old Zuse clearly expressing his disgust for Hitler. Consistency: young & old Zuse was NOT a fan of fascism.
        This anti-fascist attitude of Konrad Zuse becomes quite clear if one studies accounts of his friends and family and if one actually managed to read his biography.
        To prove this point one can quote so many excerpts from his biography:

        “After the seizure of power, many student organizations chose to disband. We in the Motiv survived the transition relati vely well, and were not shut down right away. We were obliged to give up our name, but as Kameradschaft W ilhelm Stier, it was pretty mu ch business as usual. The situation became critica l only when it was required that all Jewish members of the organization leave. Up till this tim e hardly anyone had taken notice of the fact that three of the Motivers were Jewish. I had no idea. When we wanted to disband the organizati on because of this, it was the Jewish members who volunteered and left silently, warning the others not to disband an organization so rich in tradition on their account – it was the second oldest fraternity at the University and had been in existence since 1847. But it must be said that among us there were also a few over-zealous jerks – Armleuchter, as we called them – and that after 1933 they tried , so to speak, to get us into line, working from the inside out. They were not successful. Our house remained a place for the free exchange of ideas, a place where, as we used to say, on e could still “risk a lip”.

  2. Frustrating that I didn’t know about the German Museum of Technology in Berlin, we went there on holiday a few years ago. They must have _really_bad_ advertising. The Stasi museum was excellent though, you can’t beat a bit of secret police shenanigans!

  3. I’m currently restoring a century plus old Ampico player. In these there 2 memories of 3 bits to create bass and treble of 7 intensities each. Plus one bit each to reset the memory to accept the new data. It uses timing overlap to directly change states. Pretty weird stuff from that very inventive time. One of the bits in one unit is not holding on. So basically I’m fixing a bug in a 1915-19 computer.

    I have often wondered if leather-pouch-and-valve penumatics could have been used at this time. I challenge anyone to try to model the Z1 in such air powered technology.

    1. Air-powered computer technology has certainly been done, both using mechanical valves powered pneumatically and by using 3d chambers inside a solid matrix with no moving parts. However, since air compresses easily, reliability can be a problem as compared to using water or other incompressible fluid. Pure-fluid computers with no moving parts can be expected to have excellent reliability as compared with electronic computers. (All-photonic computers are at this point an unknown.)


    “We can therefore say that, from an abstract theoretical perspective, the computing model of the Z3 is equivalent to the computing model of today’s computers. From a practical perspective, and in the way the Z3 was really programmed, it was not equivalent to modern computers.
    I am of course aware that these conclusions are curious enough to re ignite the whole discussion about the invention of the computer.”

    1. I don’t think there should be any “debate”. The “Principle” of a computer goes back to Babbage’s design for the Analytical engine. Of course he was unable to build it because of the technology. Many people knew what was needed to build a computer, what was holding them back was the lack of reliable technology. Zuse produced purely mechanical computers and relay computers but this technology was a dead end. What was really holding back computers was the lack of reliable scalable RAM. ENIAC was programable but only had ROM. The Manchester Baby had Williams Tubes for RAM which IBM licenced for its first machine but they don’t scale well and are unreliable. Mercury delay lines were more popular being used at Cambridge, NPL and at Flinders in Australia. These were also un-relaibale, none-scalable, and worse still were slower than Williams tubes.

      1. Indeed the technology of all the pioneering computers was a dead end in the long term, but many of them like the Z4 were quite useful in the short term while technology caught up. For example, that Z4 that was evacuated from Berlin, with some ongoing improvements, was in operation from 1950 until 1959, first at ETH Zurich and next at the Franco-German Institute of Research.

      2. Yes, The Analatycal Engine would have been the first computer if it ever would have been built.

        But the “Principle” of a computer was only really formailzed by Turing in 1936.

        The ENIAC in its first form was not really programmable without a serious rewiring job.

        And yes, the first problem encountered was a reliable random-access storage, the first working (but certainly too unrliable to be useful) was the Manchester Baby and its Williams tubes, which I consider as the first working computer (you may not agree with this statement!).

        Then the next problem was permanent storage, with first magnetic-core memory, magnetic tapes and disk/floppies, and now Flash memory.

        1. I got to learn a little about Germany geography while writing this article, Joshua. I made a Google Maps page with a pin in each city that was part of Zuse’s adventures during his life (the Bavarian Alps region looks like a really nice place to live, at least from a satellite view perspective). You can also find the sites where his laboratories once stood on Street View. I found it interesting that some bombed-out locations were still empty — in a large metropolis like Berlin, I assumed real estate was at a premium and that land wouldn’t stay vacant for long, certainly not decades. Maybe living in Seoul for so many years has warped my perception.

    1. The University of Erlangen-Nürnberg has restored their Z23 to working state (first Zuse transistor machine, essentially an extended version of the tube-based Z22) and there are regular demonstrations – at least when there’s no pandemic going on… I went on a field trip to Erlangen with my students some years back and they really enjoyed watching a machine running that probably existed before their parents were born. For lack of an official English web page (I think), here’s my short blog post about the trip with a number of photos:

        1. Happy to hear that you like the photos, Chris :) – I also have the original higher-resolution photos (I think those were shot using my NEX, so around 16 megapixels), so if you or anyone else is interested, just let me know. I think I should mention a license somewhere on the blog, I’m fine with CC-BY-SA-4.0.

  5. Imagine Maj3 Maj5 mechanical adder with a weighted ball in each yellow slider over a tilting green plate.
    Either of Majority or Minority blue bars might be engaged and pushed to the left by sliding green plate.
    Only two gates to compute a sum, but can’t be solved simultaneously and wanting a 90degree turn.

    Konrad’s relay adders were also amazing, since he had perfected a relay routed carry, not a ripple.
    These days we still find use for a similar circuit in the Manchester carry chain.

  6. I am imagining the miniaturisation of a cpu based on mechanical logic gates. maybe one day someone could make a MEMS version of a mechanical CPU, which runs on a winding spring (like the ones from a mechanical watch) – that would be an interesting hack!

  7. Tron: Legacy has a shout-out to him. Sam Flynn goes to the city, looking for someone named Zeus based on a reference from Quorra. However, if you turn on the captions, you find he’s not looking for “Zeus,” he’s looking for “Zuse.” A German-speaker would pronounce his name TSOO-suh but English-speakers, not knowing any better would pronounce it ZOOS, hence the confusion.

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.