Technology & Medicine › Atomic & space age
Electronic computer (Z3, Colossus, ENIAC)
1941–1946 (Z3, Colossus, ENIAC; 'first' disputed)· Invention or discovery· Communication & computing
DebatedHistorians or scientists disagree about what happened, when, or why.
Between 1941 and 1946, machines in Germany, Britain and the US first ran long calculations automatically at electrical speed.
See it on the Technology & Medicine map →How it works
Konrad Zuse's Z3 (Berlin, 1941) used about 2,600 telephone relays; it was programmable but electromechanical, not electronic. Britain's Colossus (1944), built by Tommy Flowers, used 1,600–2,400 vacuum tubes to break German codes but did one kind of job. ENIAC (Philadelphia, 1945) used about 18,000 tubes and could be rewired for any problem. From 1948, 'stored-program' machines kept their instructions in memory alongside the data, as every computer does today.
Why it mattered
Computers turned calculation from slow human work into a cheap machine task, and now run nearly everything from banking and engineering to phones.
How we know
ENIAC's records and photos, Colossus files kept secret until the 1970s, Zuse's papers and a Z3 replica in Munich's Deutsches Museum (the original was bombed in 1943).
Still unknown or debated
Which was 'first' depends on the definition. A 1973 US court ruling credited John Atanasoff's 1942 Iowa machine with key ideas and voided ENIAC's patent.
Inside this (3)
Connections (9)
▸Ada Lovelace foresaw the general-purpose Electronic computer (Z3, Colossus, ENIAC)Mostly agreed
The connection
In her 1843 notes on Babbage's Analytical Engine, Lovelace argued that such a machine could handle any symbols, such as musical notes, not just numbers. She also set out step-by-step instructions for computing Bernoulli numbers, often called the first program. Electronic computers a century later realized her vision of a general-purpose machine.
When and how
London, 1843: her translation of Luigi Menabrea's article on the engine, with her own notes three times longer; republished in 1953 as computers emerged.
How sure are we?
Her published notes, especially Note G with the Bernoulli method, survive, and Turing quoted her in 1950. But her ideas were mostly rediscovered after the first computers were built.
Debates and open questions
How much of the program and the vision was hers rather than Babbage's is debated; some scholars stress her deep insight, others see her mainly as a gifted explainer.
▸Babbage's Difference and Analytical Engines were a century ahead of the Electronic computer (Z3, Colossus, ENIAC)Mostly agreed
The connection
Babbage's Analytical Engine had a 'mill' to calculate, a 'store' for memory, punched-card input and the ability to branch on results, the same basic layout as a modern computer. It was never built. Electronic computers rebuilt these ideas with relays and tubes instead of gears, though most of their builders had not studied him.
When and how
London, 1830s–1871 (Babbage's designs); Harvard's Howard Aiken, who built the 1944 Mark I, found Babbage's writings and saw himself as his heir.
How sure are we?
Babbage's drawings survive; London's Science Museum built a working Difference Engine No. 2 from them in 1991, proving his designs were sound.
Debates and open questions
Direct influence was small: ENIAC and Colossus engineers largely reinvented his ideas. Whether Victorian machining could have built the Analytical Engine is still argued.
▸Vacuum tube was the switch inside the Electronic computer (Z3, Colossus, ENIAC)Well documented
The connection
A vacuum tube can switch current on and off thousands of times faster than a mechanical relay, with no moving parts. Colossus used 1,600–2,400 tubes and ENIAC about 18,000 as on-off switches representing numbers. Tubes were power-hungry and failed often, so engineers ran them gently and replaced them constantly.
When and how
Iowa State College (Atanasoff–Berry Computer, 1942); Bletchley Park, England (Colossus, 1944); University of Pennsylvania, Philadelphia (ENIAC, 1945).
How we know
Machine specifications, maintenance logs and photographs survive; Tommy Flowers drew on his experience with tubes in the Post Office's telephone exchanges.
Debates and open questions
Experts doubted thousands of tubes could work reliably together; Flowers proved them wrong by never switching them off. Whether Colossus or ENIAC counts as the first electronic computer is still argued.
▸Alan Turing laid the theory of the Electronic computer (Z3, Colossus, ENIAC)Well documented
The connection
Turing's 1936 'universal machine' showed in theory that one machine could do any computable task if fed the right instructions. That idea, plus his wartime codebreaking machines, fed into postwar stored-program computers; he designed Britain's ACE (1945) and wrote programming guides for Manchester's computers.
When and how
Cambridge, 1936; Bletchley Park, 1939–45; National Physical Laboratory, London (ACE design, 1945–47); Manchester University, 1948–54.
How we know
His papers and reports survive; John von Neumann knew Turing's 1936 paper, and his 1945 stored-program design report reflects similar ideas.
Debates and open questions
How much Turing's theory actually shaped ENIAC's engineers and von Neumann's design, versus independent practical thinking, is still debated among historians.
▸Electronic computer (Z3, Colossus, ENIAC) was shrunk by the TransistorWell documented
The connection
Tube computers filled rooms, drew as much power as a small factory and broke down often. Transistors did the same switching while being smaller, cooler and more reliable. Manchester's transistor computer ran in 1953, and from 1959 IBM's transistorized 7090 and 1401 became standard machines for science and business.
When and how
University of Manchester, 1953; Bell Labs' TRADIC for the US Air Force, 1954; IBM 7090 and 1401, 1959.
How we know
Machine records show size, power use and failure rates falling sharply with transistors; IBM's 1401 sold more than 10,000 units.
Debates and open questions
Some early transistors were less reliable than tubes; how quickly the switch paid off depended on manufacturing quality, which improved fast but unevenly.
▸Electronic computer (Z3, Colossus, ENIAC) shrank into the Personal computerWell documented
The connection
The 1940s computers filled rooms, used thousands of vacuum tubes, and ENIAC cost about $487,000 in 1946 money. Transistors, chips and finally the microprocessor shrank the stored-program design that grew out of that work, a processor running programs held in memory, onto a desk. By 1977 a complete home computer cost about $600 to $1,300.
When and how
ENIAC at the University of Pennsylvania, 1946; the Altair in Albuquerque, New Mexico, 1975; the Apple II in Cupertino, California, 1977.
How we know
The architecture is continuous: PCs use the stored-program design set out in John von Neumann's 1945 report, written from the ENIAC team's work; price and size records trace the shrinking.
Debates and open questions
Whether the PC followed naturally from cheaper chips or also needed California's hobbyist and counterculture movement, as some historians argue, is debated.
▸Electronic computer (Z3, Colossus, ENIAC) ran early weapons calculations for Nuclear energy (bombs and reactors)Well documented
The connection
Nuclear weapon designs need vast calculations of how shock waves and neutrons move. Manhattan Project scientists first used rooms of human 'computers' and IBM punched-card machines. ENIAC's first major job, from December 1945, ran calculations for Edward Teller's proposed hydrogen bomb, and nuclear labs stayed leading buyers of supercomputers.
When and how
Los Alamos, 1944–45 (IBM machines and human computers); ENIAC in Philadelphia, December 1945 to early 1946, run with Los Alamos physicists Nicholas Metropolis and Stanley Frankel.
How we know
Los Alamos histories and ENIAC records document the hydrogen-bomb calculation; the Monte Carlo method, later central to computing, was developed for such problems.
Debates and open questions
How much computers shaped weapons design, versus physical insight and testing, is debated; ENIAC's early hydrogen-bomb results were in fact misleadingly optimistic.
▸Electronic computer (Z3, Colossus, ENIAC) reconstructs the images of the CT scannerWell documented
The connection
A CT scanner collects many thousands of X-ray readings, which mean nothing until a computer solves a giant puzzle: what density at each point inside would produce all of them. Hounsfield's lab prototype needed days of scanning and hours of mainframe computing for one image; faster computers cut scans to seconds and allowed 3D views.
When and how
EMI's Central Research Laboratories in Hayes, England, late 1960s; Hounsfield had earlier led the design of EMIDEC 1100, one of Britain's first all-transistor computers.
How we know
The machine is named for it, computed tomography; Hounsfield's papers describe the reconstruction calculations, and no image is possible without them.
Debates and open questions
Not disputed. Current work aims to lower radiation dose, using smarter reconstruction methods and AI to get clear images from fewer X-rays.
▸Electronic computer (Z3, Colossus, ENIAC) may be outpaced on some problems by Quantum computersMostly agreed
The connection
Electronic computers, from ENIAC on, process bits that are definitely 0 or 1, and nearly all tasks will stay with them. Quantum computers use qubits governed by quantum physics, which in theory lets them tackle a few problem types, such as factoring huge numbers or simulating molecules, vastly faster. Ordinary computers still control and check every quantum machine.
When and how
ENIAC in Philadelphia, 1945–46; Richard Feynman's 1981 talk at an MIT conference proposing quantum simulation; quantum labs at Google, IBM and others from the 2010s.
How sure are we?
Mathematical results such as Shor's 1994 algorithm show the potential speedup; real hardware has shown an advantage only on narrow, artificial tasks so far.
Debates and open questions
Whether quantum computers will beat classical ones on any commercially useful problem, and when, is debated; clever classical methods keep matching quantum claims.