Monthly Archives: April 2013

Teashop Technology

In my day job as a Business Development Manager at the University of Glasgow, I’m always looking for good case studies on the benefits of business-university collaboration.  One industry sector that has benefited enormously from university expertise is the computer industry.  A fascinating example of this is a project which dates back to the very start of the industry in the UK.  It involved a collaboration between catering firm J Lyons & Company and the University of Cambridge to develop the first computer specifically for business use.  Amazingly, this landmark project in the history of technology grew out of a simple requirement for the company to more accurately monitor sales of cakes throughout its chain of teashops.

The project started in May 1949 when the Lyons board took the decision to build an electronic computer to serve the company’s growing data processing needs.  This bold decision had been prompted by a fact-finding visit to the USA by two senior managers of the company, Raymond Thompson and Oliver Standingford, two years earlier in May 1947.  Despite the mundane nature of the company’s products, Lyons had a refreshingly forward-thinking attitude to business and had pioneered innovative management techniques in its quest for office efficiency.  Thompson and Standingford’s original remit had been to investigate wartime developments in office systems and equipment but this was extended to include computers following tantalising snippets of information that were beginning to appear in the press about US developments in electronic computation.

During their trip Thompson and Standingford met with Herman Goldstine, a leading figure in the US computer research community, who informed them of efforts to develop a computer taking place much closer to home at the University of Cambridge.  The Cambridge group was led by the pioneering computer scientist Maurice Wilkes.  Wilkes had been one of a handful of non-US academics invited to attend a series of lectures held at the University of Pennsylvania’s Moore School of Electrical Engineering in the summer of 1946 to disseminate the results of US government-sponsored research into electronic digital computers.  Top of the agenda was one of the Moore School’s own projects, a radical new computer design based on a stored-program architecture.  This was the breakthrough which would turn electronic computers from sequence-controlled calculators into powerful machines capable of carrying out all manner of complex tasks.  Fired up by what he had seen and heard on his visit to the US, Wilkes began working on his own version of this design during the voyage home.  The machine would be called the Electronic Delay Storage Automatic Calculator (EDSAC).

Goldstine arranged for Thompson and Standingford to be contacted by the Cambridge group on their return to the UK.  Following positive discussions, Lyons agreed to donate £3,000 in cash plus the services of an electronics technician for 1 year to support the Cambridge group in the development of EDSAC in return for advice on the company’s plans for computerisation of its data processing operation.  These extra resources helped the Cambridge group to not only catch up with but to overtake the far more experienced Moore School team in the race to construct the world’s first full-scale stored-program computer.  The fruitful relationship between Lyons and Cambridge continued beyond the initial 1-year agreement and when the company finally took the decision to build its own computer, it was obvious to all concerned that this should be based on EDSAC.  The result was the Lyons Electronic Office (LEO), the world’s first business computer.

LEOBuilt at a cost of £150,000, LEO featured a 17-bit word length and a clock speed of 526 KHz.  Like other full-scale computers of the pre-transistor era, LEO was the size of a room.  Its electronic circuits were constructed using thousands of bulky thermionic valves and its capacious 2,048-word memory was made up of 64 individual acoustic delay lines, long cylinders filled with liquid mercury that stored data in the form of sound waves.  Input and output was via punched cards and paper tape.

LEO ran its first live data processing job in November 1951, calculating the value of cakes, pies and pastries for dispatch to Lyons’ retail and wholesale outlets.  As one of the first electronic computers in the UK, LEO also attracted considerable attention from organisations interested in using the machine for scientific applications.  Lyons was able to take advantage of this situation by making LEO available to a number of these organisations on a fee paying basis in what was probably the earliest example of a computing bureau service.

LEO was conceived for internal use only but the enormous interest in the project prompted Lyons to enter the computer industry and a subsidiary company, LEO Computers Limited, was created in November 1954 to manufacture computers “for sale or hire”.  LEO Computers later merged with the Data Processing & Control Systems Division of English Electric to form English Electric Leo Computers which eventually became part of International Computers Limited (ICL).  On its formation in 1968, ICL was the largest non-US manufacturer of computers with a workforce of over 34,000 employees.

So are there any lessons that businesses today can learn from the success of the LEO project?  Here are three I can think of:-

  1. Play to each other’s strengths.  Lyons took a lead from the Cambridge group on the design of the hardware for LEO.  This allowed the company to focus its efforts on developing the application software, a task which made the best use of in-house expertise in office systems.  The company also subcontracted much of the construction work to other firms.
  2. Don’t be tempted to reinvent the wheel.  The LEO developers did not try to create the most advanced machine possible.  LEO’s conservative design was closely based on EDSAC which itself was closely based on the Moore School’s design.  Proven components and subsystems were chosen over newer technologies.  This combination of modest design goals and the use of readily available technology facilitated speed to market.
  3. Seconding staff can be one of the most effective methods of transferring knowledge from one organisation to another.  The technician chosen for the secondment, Ernest Lenaerts, quickly became a valued member of the Cambridge group.  When Lenaerts later returned to the company, the knowledge he had gained transferred back with him, making the job of building LEO much easier.

If you want to read more about LEO, I can recommend the book A Computer Called LEO by Georgina Ferry.  Also worth reading is Memoirs of a Computer Pioneer by Maurice V Wilkes.  My own book will also cover the development of EDSAC in Chapter 5 and LEO in Chapter 6.

Advertisements

The Military Origins of Electronic Computers

The BBC News web site carried an article last week about the tragic death of a British soldier in Afghanistan.  What made this story different from the depressingly constant stream of similar reports was that his death was caused by a data error.  He was killed by a smoke shell fired from a British artillery weapon which fell short of its target due to an error in the data used to compensate for the effects of weather conditions on the shell’s flight.  The computer system which would normally provide this data automatically was not working so the gunners had to input the data manually.  Unfortunately, they used the wrong data, with tragic consequences.

Ironically, one of the driving forces behind the development of electronic computers was the need for accurate artillery firing tables.  In the days before fully computerised fire control systems, these tables provided the settings that gunners require to compensate for the effects of external factors (such as wind speed and direction, air temperature and the rotation of the Earth) on the trajectory of an artillery shell.  However, the preparation of such tables was very labour intensive, with a single trajectory taking two full working days to calculate using a mechanical desk calculator, and the increased demand for them during World War II prompted the US Army to fund the development of calculating machinery which could speed up this process.  This resulted in the development of the first general-purpose electronic digital calculator, the Electronic Numerical Integrator and Computer (ENIAC).

ENIAC

ENIAC was developed at the University of Pennsylvania’s Moore School of Electrical Engineering in Philadelphia.  The completed machine weighed over 30 tonnes and its cabinets lined the walls of a room 50 feet long by 30 feet wide.  Although ENIAC’s main purpose was to perform calculations for artillery firing tables, it also possessed the flexibility to cope with a wider range of computational problems.  The insights which the ENIAC project team gained through providing this extra flexibility led directly to the development of the stored-program computers that we use today.

The two men who led the development of ENIAC, John Mauchly and Presper Eckert, went on to create the UNIVAC I, one of the first computers to reach the market.  UNIVAC I was designed primarily for business data processing but its origins, and the origins of all stored-program electronic digital computers, can be traced back to the US Army’s need for artillery firing tables during World War II.

A Wealth of Information

One of the great challenges in writing a non-fiction book is gathering the primary source material.  Since I began writing 10 years ago, the quantity and quality of information available on computers and the computer industry via the Web has increased enormously.  I’m fortunate to have access to the massive collection of over 2.5 million books and journals in the University of Glasgow Library through my day job but in most cases I can now find the information I need simply by typing some carefully chosen keywords into a search engine and clicking on a few links.  It’s amazing to think that only 20 years ago much of the material used in the preparation of this book would have been extremely difficult or in some cases impossible to find.  Now anyone with a smartphone and an Internet connection can access it online.

I plan to create a comprehensive list of available online resources at some point in the future.  In the meantime, here are a few I’ve found to be particularly useful:-