2 of 6Image
Lyons engineers Gordon Gibbs and Ray Shawlook looking at a LEO circuit diagram
The first ever business computer, LEO I, operated by British tea-shop company J Lyons, ran its first program — to calculate the week's value of cakes, bread and pies — on 17 November, 1951.
J Lyons has been seen as an unlikely candidate to be the first business to run a commercial computer, given that it was not heavily involved with electronics manufacturing, and its primary business was serving tea-shop customers.
Nevertheless, the company had the DNA to embark on an ambitious project that would change the nature of how businesses used computers. Frank Land, who was involved with the project, told the BBC Radio 4 Today programme that J Lyons was "a smart company" that was "always looking at its business processes".
The perishable nature of many of J Lyons's products, plus its tight operating margins, led the company to consider innovative ways of tracking its business processes, according to David C Mowery of UC Berkeley.
In the 1920s, J Lyons created the 'Organisation and Methods' group, headed by Cambridge maths graduate JRM Simmons, to look at accounting and production processes, and business equipment. Simmons rose to become a board director and J Lyons' senior executive.
In 1947, Simmons sent two J Lyons executives who were interested in the business application of computers, TR Thompson and Oliver Standingford, to the US on a fact-finding mission (the pioneering computing advances made in the UK by code-breakers such as Alan Turing at Bletchley Park remained under wraps at the time due to security concerns). Thompson and Standingford learned about ENIAC, the first all-electronic computer, on a visit to the University of Pennsylvania.
Back in the UK, they also visited Cambridge University, according to the Staffordshire University Computing Futures Museum. At Cambridge, Thompson and Standingford met Maurice Wilkes, who was part of a team busy designing a computer known as EDSAC. In exchange for financial help to complete EDSAC, Wilkes agreed to help J Lyons with the design of the Lyons Electronic Office (LEO) computer, which drew on the EDSAC design.
In January 1949 engineer John Pinkerton joined J Lyons, and was involved in setting up a workshop at Cadby Hall to experiment with circuits. This picture shows Lyons engineers Gordon Gibbs and Ray Shawlook looking at a LEO circuit diagram in February 1950.
Photo credit: LEO Computers Society
LEO I filled a large room and used 5,936 valves, with another 300-400 in auxiliary equipment, according to the LSE. LEO used 64 mercury tubes — each over five feet in length — for storage, each tube weighing half a ton. The components were assembled in racks to minimise the impact of equipment failure — engineers had to replace up to 50 valves per week during times of heavy failure rates.
This picture shows the LEO I built at Cadby Hall in Hammersmith.
Photo: LEO Computers Society
LEO I control desk
LEO I had multiple input/output buffers — for punched card reader, paper tape reader, card punch and printer. The mercury tubes for LEO I held 8.75 kilobytes of data.
LEO coders programmed in two languages: Intercode, a low-level assembler language, and Clear Language for Expressing Orders (Cleo), which was like Cobol, according to the Staffordshire University Computing Futures Museum.
LEO had a speaker installed, which allowed engineers to aurally monitor certain calculations.
LEO I was initially used for evaluation, but eventually its role was extended to include functions such as payroll, inventory and production management.
This picture shows the LEO I control desk.
Photo: LEO Computers Society