A decade shaped by war
In 1940 the word “computer” most often meant a person, usually a clerk or mathematician who carried out calculations by hand or with a desk calculator. By 1949 a handful of machines existed that could perform long sequences of arithmetic automatically, and some could even hold their own instructions in memory. Almost every one of them was built because of, or just after, World War II.
It is worth keeping the numbers in perspective. These were one-off machines built by small teams, often for governments or universities. There was no computer industry yet, and most people never saw one.
What the war demanded
Wartime governments needed large amounts of calculation done quickly. Artillery crews needed firing tables, which show how to aim a gun to hit a target at a given distance under given conditions. Producing them by hand took enormous effort. Codebreakers needed to test huge numbers of possibilities against intercepted messages. Scientists working on projects such as aircraft design and nuclear weapons needed numerical results that people with desk calculators could not produce fast enough.
Those problems made automatic calculation worth a great deal of money and engineering talent, and that funding is the main reason the decade moved so quickly.
Relays, tubes and different approaches
Several technologies competed. Some machines used electromechanical relays, which are switches moved by electromagnets. Konrad Zuse in Germany built relay-based machines in the early 1940s, working largely in isolation from other researchers. In the United States, the Harvard Mark I, completed in 1944 with support from IBM, was a large electromechanical calculator that read its instructions from punched paper tape.
Other machines used vacuum tubes, which switch much faster than relays but are bulky and burn out. In Britain, the Colossus machines, used at Bletchley Park from about 1943 and 1944, helped break German teleprinter ciphers. They were built by a team that included the engineer Tommy Flowers, and they were designed for a specific codebreaking task rather than for general use. Their existence stayed secret for decades after the war, which is one reason older books often leave them out. Alan Turing also worked at Bletchley Park, mainly on other codebreaking methods, and his theoretical ideas about computation became influential in the following years.
Some early efforts are still debated. A small electronic calculating device built at Iowa State College around the early 1940s is credited by some historians as an important step, while others stress how limited it was. Claims about the single “first” machine depend heavily on definitions, such as whether a machine must be electronic, programmable, or general-purpose.
ENIAC and the first large electronic machines
The best-known American machine of the decade is ENIAC, built at the University of Pennsylvania. It was completed in 1945 and presented to the public in 1946. It was designed with artillery calculations in mind, but it was general enough to be used for other problems as well. It used thousands of vacuum tubes and filled a large room.
ENIAC was programmed by rewiring cables and setting switches, which could take days for a new problem. The people who did much of that work were a group of women mathematicians, who developed techniques for arranging a calculation across the machine. Their role was overlooked for many years.
Machines like this are usually grouped under the heading of early electronic computers, and they proved that large electronic calculators could work reliably enough to be useful. They also showed how awkward it was to change their programs.
The stored-program idea
That awkwardness led to one of the decade’s most important ideas. If a computer’s instructions were stored in its memory alongside its data, then changing the program would mean loading new numbers rather than rebuilding connections. A 1945 report describing a proposed machine called EDVAC, associated with John von Neumann and the ENIAC team, spread this thinking widely. The stored-program concept is often tied to von Neumann’s name, but several people contributed, and the credit is shared.
By the end of the decade, working stored-program machines had appeared in Britain. A small experimental machine at the University of Manchester ran its first program in 1948, and the EDSAC at the University of Cambridge began operating in 1949. These were modest in size, but they were the pattern for later designs.
Memory and input
Memory was a serious bottleneck. Designers tried storing data in tubes of mercury, where pulses of sound circulated in a delay line, and in specially adapted cathode ray tubes. Neither was cheap or entirely reliable. Input and output usually relied on punched cards or paper tape, which had been used in business and in telephone and telegraph work well before computers existed. Many of these machines also worked in binary numbers, though some early ones used decimal arithmetic.
What came next
At the end of the decade, the first attempts to sell computers were under way. A few companies and research groups began planning machines intended for sale rather than for a single laboratory, which led to the first commercial computers in the early 1950s. Meanwhile the transistor was announced in 1947 at Bell Labs, although it would take years before it replaced tubes in computers.
The main legacy of the 1940s is a set of ideas and examples: that electronic speed was achievable, that programs could be stored as data, and that governments and universities would pay for such machines. The next decade turned those experiments into products.