Explore more from the Growth List archives: long-form features on the people, machines, and moments that quietly shaped how the modern world works.
Konrad Zuse finished the world’s first programmable computer on the floor of his parents’ living room in Berlin in 1938, called it the Z1, and watched it jam every few minutes because the mechanical switches kept sticking. He was working with relays his friends had fished out of the trash at the local telephone exchange, and operating in such total isolation from the rest of the nascent computing field that he had no idea Howard Aiken was building a rival machine at Harvard with IBM’s backing. Five years later, Allied bombers flattened the workshop. The Z1 was gone. So was the Z2. And so was the original Z3, the first machine that actually worked.
A civil engineer who hated arithmetic
Zuse trained as a civil engineer at the Technical University of Berlin. He took a job at the Henschel Airplane Works just outside Berlin in 1935 and quit within months. The reason was banal and consequential: he could not stand the static stress calculations the job required, page after page of repetitive arithmetic done by hand. He went home to his parents’ apartment on the Methfesselstraße and started building a machine to do the arithmetic for him.
According to Hackaday’s history of the project, Zuse made the decision early on to use binary rather than decimal, which would shape everything that followed. Binary was simpler to implement no matter what physical switching mechanism he chose, and he had not yet figured out which mechanism he could afford.
Relays from the rubbish bin
Relays were the obvious switching technology. They were also expensive, and Zuse was funded by his parents, his sister, and a few friends. The relays he could lay hands on were discards rescued from the trash at the Berlin telephone exchange. They were designed to switch infrequently, not to clatter through thousands of cycles a second inside a calculating engine. Vacuum tubes were the other option, but wartime shortages were already biting and his collaborator Helmut Schreyer concluded the tube approach would be delayed for years.
So Zuse built the Z1 mechanically. The whole thing. Memory, arithmetic unit, control. He devised an arrangement of thin metal plates, flat bars, and steel pins that physically slid past each other to represent ones and zeros. The machine stood over six feet tall, filled a large portion of the living room, and was driven by an electric motor or, when the motor failed, a hand crank.
What the Z1 actually did
The Z1 held 64 words of memory. It could add, subtract, multiply, and divide in floating-point binary. Programs were not stored in the machine. They were punched as holes into discarded 35mm movie film, which was easier to find in 1937 Berlin than blank paper tape. Zuse and Schreyer punched the holes by hand.
The clock was mechanical. Each cycle was four planar motions, imagine sliding a square plate north, then west, then south, then east, with each motion triggering one phase of the operation. No instruction took more than three of these cycles to complete. The Z1 was, in every meaningful sense, a programmable digital computer. It just barely worked. The mechanical switches got stuck. The machine ran for a few minutes at a time before jamming.
The Z3, finished four months before Pearl Harbor
Zuse built the Z2 next as a test of whether relays could handle the arithmetic, keeping the mechanical memory and swapping in a relay-based ALU operating only on fixed-point numbers. It worked well enough that the German Aircraft Research Institute provided partial funding for a successor. The Z3 was entirely relay-based, faster than the Z1, and reliable. It used roughly 2,000 telephone relays.
It was demonstrated at the German Laboratory for Aviation in Berlin on May 12, 1941. According to Inverse’s account of the unveiling, the machine was used to calculate aerodynamic flutter problems in aircraft wing design, the same kind of static stress work Zuse had quit Henschel to avoid doing by hand. Wired’s day-in-tech retrospective notes that because the work was done inside Nazi Germany and largely classified, the rest of the computing world had no idea any of it had happened.
Howard Aiken’s Mark I, backed by IBM and built at Harvard, was demonstrated in 1944. Three years after the Z3.
The bombs
The Battle of Berlin, the Allied strategic bombing offensive against the German capital, ran from November 1943 into March 1944. It was part of the wider campaign that 24/7 Wall St. catalogues as one of the largest sustained air operations in history, with waves of bombers targeting industrial centres, transport networks, and the dense residential districts that housed the workers who kept the war economy running. The Methfesselstraße was in one of those districts.
During the bombing campaign, the Zuse family home and the adjacent workshops where Zuse had built and stored the Z1, the Z2, and the original Z3 were destroyed. Everything inside was lost. The machines. The engineering drawings. The notebooks. The punched-film programs. The patent application materials. Decades of work, reduced to rubble.
The Z4, a successor machine then nearing completion, survived because Zuse had moved it across the canal to a separate workshop on Oranienstraße. In February 1945, the Oranienstraße workshop was destroyed too. Zuse and his colleagues managed to drag the Z4 out of the basement before it was lost. They loaded the Z4 onto a truck and fled south.
Hiding the computer in the Alps
The escape route ended in the Bavarian Alps, in a village called Hinterstein. Food was scarce. Shelter was scarcer.
Zuse stashed the Z4 in a farmhouse cellar and spent the final months of the war working on something he called Plankalkül, a formal notation for describing computations independent of the machine that executed them.
The patent the German government refused to grant
Zuse had filed a patent application on the computer in 1941, shortly after the Z3 demonstration. The German Patent Office sat on it through the war. It published the claims in 1952. Only one company filed opposition: the Triumph calculator company.
The case dragged on for years. In 1967, the court finally rejected the patent. The ruling acknowledged that the invention was genuinely innovative and progressive, then denied it on technical grounds. Zuse never received a patent on the programmable computer.
Why nobody outside Germany knew
Three things kept Zuse off the standard origin story of computing for decades. The first was the war: classified aviation work inside Nazi Germany did not circulate. The second was the bombing: with the original machines and documentation destroyed, there was nothing left to put on display, no working artefact for postwar journalists to photograph. The third was the parallel emergence of the Harvard Mark I, the Atanasoff–Berry Computer, Colossus at Bletchley Park, and ENIAC, all of which had institutional backers (IBM, the U.S. Navy, British intelligence, the U.S. Army) writing the history.
As an Irish Times piece on the question notes, the question of who invented the digital computer has many possible answers, with several strands converging around the late 1930s and 1940s, and no single inventor able to claim the title cleanly. Zuse’s claim, on the strict criterion of a working, programmable, automatic, binary, floating-point machine, is among the strongest. He just happened to be sitting in the wrong city when the bombers came.
The reconstruction
In 1961, working from memory and what fragments of documentation he could reassemble, Zuse rebuilt the Z3. The replica is on display at the Deutsches Museum in Munich. In 1984, with funding from Siemens, he began rebuilding the Z1 from scratch in his parents’ old style of construction, completing the work in 1989. That replica sits in the German Museum of Technology in Berlin, about a kilometre from where the original was incinerated.
Zuse died in 1995. The Z4, the only one of his wartime computers to survive, ran in commercial service at the Swiss Federal Institute of Technology in Zürich from 1950 to 1959, calculating ballistics for the Swiss Air Force, structural problems for engineers, and a flutter analysis for the wings of a Saab fighter jet. It is now in the Deutsches Museum too, a few rooms away from the Z3 replica.
If you stand close to the Z3 reconstruction you can hear it click. Two thousand relays opening and closing in sequence, the same sound that filled a room on the Methfesselstraße in May 1941, four months before the first American B-17 took off from England and two and a half years before the bombs found the address.