It may come as a surprise to find out that computational physics — the discipline that today uses powerful supercomputers to crunch through vast amounts of data on, say, the Earth's climate — actually began in the late 1920s, some three decades before the first electronic computers were built. One of the people responsible for kick-starting this field was the physicist and mathematician Douglas Hartree, who at the time was trying to calculate atomic wavefunctions in order to determine the structural properties of atoms. While the Schrödinger equation can easily be solved for the hydrogen atom, repeating the feat for multi-electron atoms involves calculations that at the time were generally thought to be intractable. Hartree, however, developed a technique known as the self-consistent field method, which allowed these problems to be solved numerically. Unfortunately, his method was so laborious using the facilities available at the time — mechanical calculators operated by humans — that very little use was made of it until electronic computers became available.

It was not until the Second World War, however, that the development of such computers took off, when they proved invaluable for code-breaking and generating artillery firing tables. After the war, many of the scientists who had taken part in these projects were keen to use these devices for their peace-time research. Hartree himself was involved in these early applications, including advising the US military on the use of ENIAC (the first large-scale reprogrammable electronic computer) for calculating the ballistic properties of different types of ammunition.

At the time, programming computers was an esoteric art. They had to be fed instructions in "machine code" — a language that describes every single addition, subtraction and so on in the precise order required — which was tedious to write, error-prone and required very specialist knowledge. If computers were to be applied — and hence sold — more widely, then programming them had to become considerably easier, and this required a language that bore a much closer resemblance to the mathematical problems being tackled.

With this in mind, in 1954 a team of researchers at IBM led by John Backus, who died in March this year aged 82, embarked on the creation of Fortran (Formula Translation). This was the first successful "high-level" language — i.e. it used a program known as a "compiler" to translate commands describing the mathematical operations to be performed into instructions in machine code. Three years later, the first Fortran compiler became commercially available, and it was not long before physicists realized the opportunities that it offered. Since then it has evolved through many versions, each more powerful than the last, and even now Fortran is still the language of choice in many areas of physics.

In the December issue of Physics World, Peter Crouch, Clive Page and John Pelan recall the triumphs of Fortran, and discuss why it will still be used for many years to come.

To read the full version of this article — and the rest of the December issue of Physics World — please subscribe to our print edition.