LEO Computers | 1951
J. Lyons & Co.'s pursuit of efficiency led to the development of the LEO (Lyons Electronic Office) computer, one of the most innovative technological achievements of the early 1950s. With advice from Cambridge University, Lyons embarked on building an electronic stored-program computer designed for office tasks. The rising office costs after World War II made Lyons recognize the need for automation. Over the years, office procedures had been refined, and various new ideas, such as microfilm, had been introduced. Even before the war, Lyons had been highly regarded for its efficiency not just in factories but also in office procedures and job classification. Given its policy of managing service sectors including legal, transport, laundry, box-making, central procurement, construction, architecture, food labs, tea gardens, and wine cellars, the decision to pursue building an electronic computer was not a major issue. Despite lacking a history in electronics or electromechanical engineering, they confidently undertook the design and construction themselves.
The project was sparked in 1947 when two managers, Thomas Thompson and Oliver Standingford, visited America. Upon their return, they presented a report to the Lyons board, proposing that an electronic computer could be the key to office efficiency. They suggested spending £100,000 to build a computer that would save £50,000 annually in office expenses. At the time, Cambridge University was also engaged in the EDSAC computer project, designed for academic computing, which was unsuitable for office work, but Lyons saw potential in the new technology. They did not want to remain in a passive position of merely purchasing machines from manufacturers; they felt it was important to influence the design.
Consequently, Lyons donated £3,000 to support the EDSAC project at Cambridge in exchange for the right to receive advice when needed. Maurice Wilkes, the leader of the Cambridge team, initially thought Lyons might be out of their minds but agreed with the idea. For a year, Lyons sent staff to Cambridge to learn about electronics. They also advertised for an electronic engineer in the scientific journal Nature, leading to the application of John Pinkerton, who learned about Lyons' project through Wilkes. The interview was formal, and he began work on January 17, 1949. Subsequently, several team members were hired for hardware and software design, although the term "software" was not yet common.
Over the next three years, this inexperienced team designed and built a working model named LEO. A large area at the headquarters was secured, and the computer was gradually assembled, with completion achieved three years later. Throughout the process, staff including carpenters, plumbers, metalworkers, and engineers worked on the massive assembly of valves, switches, wiring, ducts, resistors, and power supplies. Many challenges had to be overcome, as this was an unprecedented undertaking without prior experience. An operating system also had to be designed and created. Due to reliability issues, magnetic tape was initially discarded but was later reintroduced. Engineers worked closely with the software team to design "actions" related to the work being performed. All failures were recorded, and audit trails were provided. Application design was fully flowcharted, and bench tests were conducted before entering computer programs. At that time, the operational time of the machine was much more expensive than the programmers' time. Mercury delay lines were used for storage, totaling 64 lines, limited to 2,048 instructions or short numbers. The entire machine contained over 5,000 thermionic valves.
The first operation of the computer took place on September 5, 1951, with an application for bakery evaluations. Although the computer was initially unreliable, improvements were made, and it began to be used weekly. The results were a great success. However, payroll automation was one of the primary goals of John Simmons, the controller overseeing the LEO project. Between 1951 and 1953, the project team overcame many of the machine's unreliable characteristics, and by December 1953, they felt it was reliable enough for payroll processing. This significant milestone was reached on December 24, 1953. The results were astonishing: previously, an experienced clerk took eight minutes to calculate an employee's payroll, but LEO completed the task in just 1.5 seconds. This was a groundbreaking event, a quiet revolution.
News of this remarkable "electronic brain" spread throughout the industry, leading many renowned companies to approach Lyons for assistance with complex calculations that had previously seemed nearly impossible. This marked the beginning of a long-standing bureau service. Lyons also established a facility to manufacture computers for other companies. Before American computers influenced the British market, the LEO computer sold reasonably well, leading to improvements with the LEO II and LEO III models. Additionally, Lyons produced mark reading machines to enhance data entry, eliminating punch cards and paper tape. To complement these devices, they developed a high-speed output printer that projected optical and computer-generated images onto a photosensitive selenium drum. This large-scale laser printer was expensive and justified only for businesses that required substantial printing output, such as insurance companies and government agencies.
By the 1960s, American companies had come to dominate a significant portion of the British computer market. Their machines offered superior engineering, higher reliability, and, most importantly, lower prices. Other British computer manufacturers also struggled under similar American competition, prompting the UK government to support mergers of British interests to counteract imports. LEO merged with English Electric, which in turn merged with other notable companies like Marconi. Over time, British computer manufacturing began to decline.
