The
LEO Computer
Among all
the technological achievements developed
by J. Lyons Co. in their quest for
efficiency, the development of their LEO
(Lyons Electronic Office) computer in the
early 1950s must rank as one of the most
innovative. Single-handedly, and with
advice given by Cambridge University, J.
Lyons & Co. embarked upon one of the
most ambitious DIY projects of the
century; the building of an electronic,
stored program computer designed
specifically to undertake any clerical
task. The upward trend in office costs
following the end of the Second World War
made Lyons realise that some form of
automation was essential if these costs
were to be brought under control. For many
years their office procedures had been
honed and adapted and many novel ideas had
been introduced (micro film for example).
In the pre-war years Lyons had developed
an enviable reputation not only factory
efficiency, but also for office procedures
and job classification. Since it had been
the policy for Lyons to control their own
service departments (legal, transport,
laundries, box making, central buying,
works department, architects, food
laboratory, tea estates, wine cellars,
etc.) it was not such a big deal for them
to embark on the building of an electronic
computer, even though they had no history
of electronics or electro-mechanical
engineering. There was nothing available
to them at this time which met their needs
and so with their usual self-assurance
they set about designing and building one
themselves.
The
catalyst came in 1947, following a trip to
America by Thomas Thompson and Oliver
Standingford, two managers with wide
experience of clerical procedures. On
their return they produced a report for
the Lyons board which basically said that
electronic computers hold the key to
office efficiency and for £100,000
Lyons could build one themselves which
would show a saving in office expenditure
of £50,000 per year. At this time
Cambridge University were involved in
their own computer project, EDSAC. This
was designed for academic calculations and
quite inappropriate for office work but
Lyons did see the potential in the new
technology. However, they did not want to
play a passive role merely keeping in
touch and in due course acquiring machines
as they came available from manufacturers.
In this way they could not influence
machine design and this they felt was
essential if the problem of commercial
clerical automation was to be sold
successfully.
Instead
Lyons donated £3,000 to Cambridge, to
help in their EDSAC project, on the
understanding that Cambridge would give
them advice when needed. Maurice Wilkes,
who headed the Cambridge team, thought
Lyons had taken leave of their senses but
was happy to go along with the idea. Lyons
seconded a man to the Cambridge team for a
year during which time he learned a great
deal about electronics. Meanwhile Lyons
advertised for an electronic engineer in
the scientific journal Nature and it was
answered by John Pinkerton. He had
recently obtain his Phd at Cambridge and
through Wilkes, had learned of this
extraordinary project which Lyons had
planned. The interview was a formality and
Pinkerton started work on 17 January 1949.
Several other team members were recruited
for both hardware and software design
although this terminology was not yet in
use.
During the
next three years this inexperienced team,
designed and built a working model which
was dubbed LEO. A large area at head
office was vacated and the computer
gradually assembled piece by piece until
the whole came together as a working
machine three years later. During the
process carpenters, plumbers, sheet metal
and engineers staff toiled on the huge
assembly of valves, switches, wires,
ducting, resistors and power supplies in a
well organised undertaking. Many, many
difficulties had to be overcome. This had
never been done before and so there was no
experience to draw on; the operating
system even had to be designed and
written. Magnetic tape was introduced and
discarded because it was not reliable. It
was later introduced again. Engineers
worked in close co-operation with the
software team to design 'actions' relevant
to the work which was to be performed.
Every failure was logged and recorded to
provide an audit trail. Application design
was fully flow-charted and bench tested
long before it even got to a computer
program. In those days machine time was
far more expensive that programmers time.
Mercury delay lines were used for storage,
64 tubes in total, with a limited capacity
of 2,048 orders or short numbers. The
complete machine used over 5,000
thermionic valves.
The first
operational run of the computer took place
on 5 September 1951 when an application
known as Bakeries Valuations was
performed. It was nursed through a pretty
unreliable machine but from then on it was
run each week as improvements were made.
It was a resounding success. However,
payroll automation had been one of the
main objectives of John Simmons, the Lyons
Controller who was responsible for the
whole LEO project. Between 1951 and 1953
the project team began to overcome many of
the machine's unreliable quirks so that by
December 1953 it was felt reliable enough
to undertake payroll, a task which had to
be performed to time because staff had to
be paid and in those days this meant
weekly pay. This milestone came on 24
December 1953. The results were
astounding. The task of calculating a
employees pay, until now, took an
experienced clerk 8 minutes. LEO had done
the job in 1.5 seconds. It was a
watershed, a quiet revolution.
News of
this fantastic 'electronic brain'
circulated through industry and many
famous companies commissioned Lyons to
undertake a range of tasks on their
computer; tasks which had been almost
impossible to conduct previously because
of the complex calculations needed. Hence
a bureau service started which continued
for many years. Lyons also set up a
manufacturing facility to build computers
for other companies. Until the American
computers began to have an impact on the
UK the LEO computers sold moderately well
and the models were improved on LEO II and
LEO III. Lyons also built mark-reading
machines to improve data input and
dispensed with punch card and paper tape.
Complimenting these devices was a fast
output printer using optical and computer
generated images which were projected on
to a light sensitive selenium drum. A
large laser printer. They were incredibly
expensive and could only be justified in
businesses with large printed output such
as insurance companies and government
departments.
By the
1960s the Americans had captured much of
the UK computer market. Their machines
were better engineered, more reliable and
above all less expensive. With other
British computer manufacturers suffering
from the same American onslaught the
British government supported the merger of
British interests to counteract the
imports. LEO merged with English-Electric
and they in turn merged with other famous
companies such as Marconi. In time British
computer manufacture faded away.
This
remarkable story is told in, LEO - The
First Business Computer. ISBN
0-9521651-0-4
Hasler
Publishing, 4 Luckley Wood, Wokingham,
Berkshire. RG41 2EW. Email:
[email protected]
Hardback
270 pages, 83 illustrations, biographies,
14 appendices. Price £18.95 plus
postage overseas.
|