Computer Weekly is marking its 50th anniversary this year with a series of articles celebrating 50 years of British technology innovation. In this article, we look back at the years before 1966, when Computer Weekly was first published.
The story of modern computing is tied intricately to wartime technology. In fact, one of the seminal papers, First draft of a report on the EDVAC, which set in concrete the definition of a modern computer, was written by mathematician John von Neumann, who worked on a way to process the vast numbers of calculations for the Manhattan project in Los Alamos, which was needed to design an atomic bomb.
In the UK, putting aside Colossus, Bletchley Park’s ingenious cypher-breaking machine that was bound for many years by the Official Secrets Act, there was plenty coming out of wartime technology. “A lot of wartime tech put us on the road to computing,” says computer scientist Andrew Herbert, a trustee at The National Museum of Computing. Much of this work came from British wartime effort on radar.
As Simon Lavington, author and digital archivist at the Computer Conservation Society, explains: “During World War Two, high-speed pulse electronics had been developed to a high state of expertise for radar.”
Many of those working on radar had studied maths or physics degrees before the Second World War, and went back to their universities afterwards, which enabled the technology developed in wartime to be utilised in academia.
The big issue was how to design memory. “They didn’t know how to store data economically,” says Andrew Herbert, leader of the EDSAC reconstruction project at The National Museum of Computing. “For the first decade of computing, memory was the Achilles heel. It takes five to six valves to store just one bit.”
The story of computer memory
Given that it takes eight bits to make a single byte, the gigabytes of memory on modern smartphones would have seemed impossible in the 1940s and 1950s.
Read more about the history of British computing
- The National Museum of Computing marks the anniversary of Colossus, the computer that broke coded messages from Hitler to his generals in World War Two.
- Official secrecy has almost certainly stopped Bill Tutte and Tommy Flowers from receiving due recognition as pioneers of the modern computer.
- Leo: How did a purveyor of tea and cakes claim a place at the cutting edge of office technology?
- The Computer Conservation Society (CCS) has commissioned a working replica of EDSAC, the world’s first operational programmable computer.
A computer, as originally defined in 1935 by the British mathematician Alan Turing – famed for cracking the German Enigma code – and later refined in the US by von Neumann in 1945, comprises an arithmetic logic unit (ALU) to run a series of calculations fetched from memory on data held temporarily in an accumulator. When a program runs or is executed, a series of machine instructions are fetched from the memory store sequentially. Each runs a simple mathematical operation on data held temporarily in the ALU – like add, subtract, multiply or divide.
Lavington says the early pioneers could only afford to use valves to make the accumulator and a few working registers that were required to perform these mathematical operations. What they really needed was a low-cost method to build the computer’s memory, needed to store the program.
Towards the end of the war, the problem the radar engineers faced was one of permanent echo cancellation. “They needed to cancel out the permanent echoes from hillsides and the landscape,” says Lavington.
Basically, once the landscape has been removed from a radar image, anything that remains is a moving object, such as a warship. Radar used mercury delay lines to store blips representing these moving objects. In early computer memory designs, transducers were used to send and receive a sound blip transmitted through a tube filled with mercury.
Sound travels through mercury relatively quicky – at 1,450 metres per second, compared with 343.2 metres per second in air – and the delay between the transmission of the blip and receving it enabled the memory engineers to send a stream of blips, representing binary data, from one transducer to another.
Another technique the early computer pioneers used was the persistence present on a phosphor radar screen, which could be used as a form of computer memory to store programs. In both cases, the blip – or data – would need to be refreshed continually, just like DRAM in modern computers needs refreshing today, otherwise the data it held would quickly fade away.
With memory came the first stored program computer. The two pioneering machines were the Manchester Baby and EDSAC in Cambridge. For Lavington, the Manchester Baby represents the very first working demo of a computer, and therefore has great significance, not only to the British computer industry, but also globally.
Cambridge University had a mathematical laboratory that housed mechanical computing devices for use by the rest of the university.
“Maurice Wilkes, director of the Cambridge Maths Lab [pictured below with EDSAC], heard about von Neumann’s ideas and realised Cambridge needed a computer,” says Herbert. “Wilkes went to a conference at Princeton where all the pioneers met, and on the days sailing back on the Queen Mary ocean liner he designed EDSAC – a machine that was big enough and reliable enough to do useful calculations for Cambridge scientists.”
With a clock speed of 500KHz, it was a relatively slow machine, says Herbert, even by the standards of the 1940s. By 6 May 1949, the team at Cambridge had a fully working machine that could print out squares of numbers.
For Herbert, the second calculation the EDSAC ran, printing out the prime numbers, showed the full capability of a programmed machine to both do calculations and take decisions based on the results.
“By 1951, the Maths Lab was running a computing service. Wilkes, along with David Wheeler and fellow computer scientist Stanley Gill, had written re-usable subroutines for input, output and mathematical functions. These were fed on paper tape into EDSAC as required, along with the user’s program,” he says. According to Herbert, the computer scientists at the Cambridge Mathematical Laboratory invented the modern concept of a software library.
The era of business software began with J Lyons & Co, Britain’s largest catering company, famous for making tea, cakes and ice cream, along with running teashops, corner houses and hotels and doing the catering at Buckingham Palace garden parties.
In 1947, Oliver Standingford and Raymond Thompson of Lyons went to the US to find out about automation and were told of the pioneering work that was taking place in Cambridge. But on visiting Cambridge they were told EDSAC was two years away. The Lyons board agreed to help fund EDSAC. Once the machine had proved it could work, Lyons went about building its own machine, Leo, which went live in 1951.
“The reason Lyons got going was because it was the biggest catering company in the UK and decided automation was necessary – this is how Leo got started,” says Peter Byford, chairman of the Leo Computer Society.
According to Byford, a lot of the basic structures of modern business computing were developed in the 1950s and 1960s. “David Caminer from Lyons was the first system analyst,” he says.
Caminer went to work for Lyons at its Cadby Hall headquarters in Hammersmith, London, in 1936, with no idea that his later work would be intrinsically linked with the development of British business computing. One of Caminer’s innovations was around optimising the business processes at Lyons.
As Byford explains: “All the tea shops put in their orders on Saturday. Caminer went round to every tea shop and made changes they needed every Saturday night.” This allowed Lyons to have the bulk of the goods needed by the tea shops pre-ordered, while enabling each tea shop to tweak their individual stock orders.
Frank Land, an emeritus professor in the Department of Information Systems at the London School of Economics (LSE), is regarded as the UK’s first information systems professor. The young Land worked as a programmer on the Leo in 1953.
“We knew everything we did had never been done before,” he says. “We didn’t realise we were transforming everything and we shared our work in software.”
Among the challenges the Leo programmers faced, he says, was the size of the memory store on the original computers and trying to accomplish the calculations for programs such as an inventory system, given the time it took to read one punch card.
There was an emphasis on software quality, Land explains: “Human time was much less valuable than computer time. What we had on the computer was already in the right so two people had to see everything before it went onto the computer.”
When Land started at Lyons, the tea company was expanding the amount of automation it wanted to do. “We very quickly recruited six more people, and when Leo Computer Company was formed in 1955 we had 35 people,” he says.
From the very beginning, the Leo programmers used a sophisticated assembly code for programming.
Among the ideas Leo adopted from Cambridge was microcode. “You could program an instruction set tuned to business computing such as to convert decimal to imperial or run checksums on the punch cards,” says Land.
While the Cambridge Maths Lab developed mathematical subroutines, Land says the Leo programmers developed their own routines. “One of our programmers created a floating point subroutine,” he says.
The Leo quickly became a popular computer for business. Many of Britain’s best-known companies, including household names such as Dunlop, Imperial Tobacco, BP, Phoenix Insurance, Heinz of baked bean fame, the Ford Motor Company, Durlachers, the Post Office, Royal Dockyards, as well as government offices and local authorities, acquired Leo Computers or used Leo services.
British Rail used a Leo computer in 1955/56 to calculate station-to-station distances, which it needed to accurately charge for rail fares. Leo was also the first machine to run weather modelling at the Met Office.
The programmers needed speed and agility, just like today. “In 1963/64 we were producing tax tables for the Treasury. On Budget day, we would wait for the courier to arrive with the tax changes. We had to print out the tax tables for the government very quickly,” says Land. Many of the changes involved changing parameter values that could be read in by the tax program. But there was some recoding.
The great tragedy
In the 1940s, 1950s and early 1960s, there was a huge amount of innovation coming out of the nascent British IT industry. Along with Leo there were Elliot Brothers, Ferranti and British Tabulating Machines.
In 1962, Manchester University, Ferranti and Plessey jointly developed Atlas, the world’s first supercomputer. Among the innovations in the Atlas was the Atlas Supervisor, widely regarded as the world’s first operating system.
But clearly no story on the early history of British computing is complete without at least a passing note on Colossus and the work of Alan Turing and Tommy Flowers at Bletchley Park. Unfortunately, as Lavington notes: “Since all of Bletchley Park was secret, how could it ever have had any influence on the British computer industry. It did not have any effect. There was no connection.”
A brief history of Colossus
Colossus was a 1,500-valve machine designed by Post Office engineer Tommy Flowers at Bletchley Park to enable the code breakers to crack the cypher used in the communications between Hitler and his generals during World War Two.
The machine is based on the work of Bill Tutte, who reverse engineered the Germans’ Lorenz SZ742 cypher machine by analysing samples of encrypted messages to work out how they had been encoded. The work led to the development of Tunny, a machine which emulated Lorenz.
Reading 5,000 characters per second (faster than anything ever produced commercially), Colossus’s job was to crack the original settings the Lorenz operators used to send their communications. It performed 5,000 algorithmic tests every second. With the right settings, the encrypted message could be read in plain German.
For Land, the gap between military and business was “a great tragedy for British computing”, whereas in the US, “it was all integrated and they benefited from the link with the military”.
Under the Harold Wilson government, minister of technology Tony Benn began the initiative to create a British computer industry with the formation of ICL in 1968, which came about through the merger of International Computers and Tabulators (ICT) with English Electric Computers, which itself had acquired Elliot, Leo and Marconi. However, sadly for Land, the defence computing sector in the form of Marconi and Plessey were kept out of the merger.