• Systems Engineer Consultant -...
• NATIONAL SALES MANAGER. WIRELESS...
• CHANNEL ACCOUNT MANAGER $ 210K...
• IT Engineer - Canberra
• C#ASP.NET Developers

Sir Maurice Wilkes, going strong at 91

Peter Davidson, Information Age

14/12/2004 11:37:01

In the days when the Cambridge Mathematical Laboratory, birthplace of the world's first stored program computer (EDSAC), was an appropriated anatomy dissecting room redolent with formalin, entry was by an otherwise nondescript green wooden door.

These days, entry to the engineering department of what is now the Cambridge Computer Laboratory, in a new high-tech building on the northern outskirts of the leafy university town, is guarded by sophisticated electronic and biometric technology.

Emeritus professor Sir Maurice Wilkes, father of EDSAC (Electronic Delay Storage Automatic Calculator), submits to its security scrutiny without demur every day as he goes to his office.

At a spry 91, mentally agile with a sense of humour that bubbles readily, he drives the few kilometres there from the modest house he bought in nearby Huntingdon in 1952.

He has been making the journey to the university since 1931 when he first read mathematics, and later to the Cavendish Laboratory as a graduate student doing research in experimental physics in 1934.

Apart from the war years involved in radar research, and for a few years in the early 80s working in industry in the US, the laboratory has been the focus of his professional life since becoming its director from 1945.

Nowadays, he continues the prolific generation of papers which has characterised his career, particularly on the life and philosophies of Charles Babbage for the Royal Society and others, gaining an international reputation in this area.

The green wooden door, like him, is something of a fixture at the lab, and carries a wealth of history with it: when the old building was demolished and the department moved eventually to its new digs in what is now the William Gates Building in 2002, the door came too.

It hangs in the building protected by a Perspex screen, missing a panel, but covered with the names of the lab's senior members from years gone by.

As each either retired or moved on, "they were shown the door", he explains and had their names engraved on it. Among them is one J M Bennett, a young Queenslander who became Sir Maurice's first research student in 1947.

John Bennett, who went on to become the first Professor of Computing at Sydney University where he built SILLIAC and was recently awarded the Pearcey Medal for 2004 (see panel), built the main control unit for EDSAC.

Bennett readily recalls joining the EDSAC team in 1947 in the reeking anatomy building, selected because it had a goods lift "capable of carrying two cadavers", and otherwise useful for carrying the tons of gear which would eventually become the world's first practical, programmable computer.

For the young Dr Maurice Wilkes, having joined the Mathematics Laboratory in 1937 under the directorship of Professor John Lennard-Jones, a lecture on differential analysers by Douglas Hartree of Manchester University brought mechanical calculators into his ken - and a re-introduction to Meccano.

Hartree, having come up with the idea of using Meccano to build Manchester's analyser to solve and plot differential equations, found its usefulness well beyond the ambitions of small boys everywhere who had bolted bits of metal together since the start of the 20th century.

It captured Wilkes' imagination, particularly as it offered a solution to an equation, which could not be expressed in terms of tabulated functions, but essential to his work in investigating ionospheric models for the propagation of long-wave radio waves.

Finding that one had been built at Cambridge, he was given charge of it by Lennard-Jones and set about enlarging it to five integrators, which meant finding more Meccano.

"We contacted Frank Hornby's Meccano factory in Liverpool, which donated all the parts we needed - a boy's dream," he recalls.

The analyser proved itself in accuracy and application, and like much of the laboratory's later technology, was made available to anyone needing it.

But it was not an end in itself, and Wilkes pushed for the building of a fully engineered analyser like the one at Manchester, and Metropolitan Vickers was commissioned to build it. Finding a home for it led to the laboratory being housed in the anatomy department.

The Meccano analyser which had performed so well eventually found its way to New Zealand in 1950 to the Seagrave Radio Research Station and later the Dominion Physical Laboratory where it was used in hydro-electric dam design.

"From memory, we got fifty pounds for it. It was a very useful scientific tool."

But before Vickers could install the new machine, war was declared and the entire building, dissecting room included, was commandeered by the Depart of Supply and the analyser diverted to ballistic calculations.

Dr Wilkes was also diverted; a congregation of scientific luminaries was soon formed to accelerate work on radar and he joined to add his knowledge of ionospherics and radio wave technology to the development of England's wartime coastal radar defences.

His was to be a busy war, serving through its entirety:

After devoting his practical engineering skills and theoretical knowledge to the detection of the Luftwaffe's predatory ambitions along England's bleak coasts with the construction of wood and wire radar posts powered by get-it-fixed-by-morning electronics, there were later some concerns of conscience as his work took him deeper into the scientific war effort.

Like a number of academics bound by war's harness, work on finding better ways of waging war, in his case precision bombing radar technologies like Oboe and H2S, raised issues. In his "Memoirs of a Computer Pioneer" (MIT Press, 1985), he speaks of visits to Germany immediately after the war to evaluate captured technology and of his seeing the havoc wrought by Allied bombing.

"It [the bombing offensive] was, however, not the kind of work that my experience best fitted me for, and even at that time of the war I had certain unresolved ethical problems in regard to the bombing of industrial and civilian targets," he wrote.

Six decades on, discussion of the topic is gently deflected:"It's just too big a subject to discuss now."

Still, his work in night-fighter radar, and later with Coastal Command in using radar to reverse the depredations of the U-boats which harassed Allied shipping, added significant weight to the science of winning a war and a store of experience to be applied in quieter times.

Buoyed by an opportunity to "do something constructive again", he returned to the laboratory as soon as he could where he was appointed its director, turning his attention to computing, ready to unleash the ideas nurtured during six years at war.

Much had happened in computer development in those years including the building in 1943 of Colossus to decode the Enigma and other encryptions at Bletchley Park under the guidance of mathematician Max Newman.

Able to perform logical operations but not arithmetic, its development was shrouded in secrecy for three decades, stifling opportunities for further development.

But Wilkes knew that analog computing was limited, and the work done by John Mauchly and Presper Eckert to create ENIAC (Electronic Numerical Integrator and Automatic Computer) for the Department of Defence in the US loomed prominently on his horizon.

Begun under a military contract in 1943, it was completed in 1945 at the Moore School at the University of Philadelphia. Weighing 30 tons and containing 19,000 vacuum tubes, it embodied the componentry and concepts from which today's high-speed digital computers evolved.

Difficult to program and consuming 200Kw of power, it represented the future and when, out of the blue, Wilkes was invited to attend seminars conducted by its creators at the Moore School, he jumped at the chance.

Close collaboration and free information exchange with the US team was tantamount to a personal epiphany and he says that "I came back from that course feeling that I knew everything there was to know".

But even before his return he had started sketching out his plans for what would become EDSAC

However optimistic Britain felt for its future, there were tough years to weather as the war's economic miasma remained. Building EDSAC would need scarce funds.

Tea-driven recovery Help came from an unlikely source in 1947: J. Lyons and Company, operator of the chain of Lyons Corner Houses, had heard of the development of digital computing in the US and were interested in automating basic accounting work such as payroll.

Lyons had more than 3500 of the Corner House tea rooms, famed for their art-deco design and an established British institution. As a company, it had developed managerial practices far ahead of their time, recruiting graduates as management candidates direct from university among them.

They could not get what they wanted built in the US in time, and were referred to fledgling EDSAC development in Cambridge.

An offer of 3000 pounds and the provision of an assistant to work on the project for a year, largely without condition other than an understanding that Lyons would want a computer for its own use, was accepted without hesitation and serious development work could begin.

(In 1951 Lyons took delivery of its EDSAC-based computer, called LEO for Lyons Electronic Office, which it used to process its overnight production requirements for its catering business, making it the world's first computer dedicated to commercial calculations.

Adopting such unheard-of practices in the 50s as decimalising sterling for faster accounting and using microwave ovens to prepare frozen food portions, Lyons later successfully built its own commercial computers for business, eventually selling that part of its operation to English Electric.

The company had some pioneering ideas about outsourcing too: in 1954 it took over all payroll operations for the Ford Motor Company in England.)

The assistant that came with the Lyons' package was to have stayed for a year, but worked on, typical of the rapidly swelling ranks at the laboratory: John Bennett had taken up station to begin work, soon to be joined by David Wheeler who been among a small group of student "volunteers" working on construction.

By 1948 a photograph of the laboratory staff shows 19 people; EDSAC development was in full swing.

Volumes have been written about the development of EDSAC, its use of mercury delay lines for memory and vacuum tubes for logic. Wilkes is still ready to dispel the myth that EDSAC grew out of scrounged components, insisting that components were sourced new from government supply agencies and others.

A wealth of information is available on its three-year gestation until its first logged program, computing a table of squares, ran on May 6, 1949, upon which all involved headed for the local pub.

There is conjecture over whether EDSAC was the first stored memory computer, some claiming the honour for Manchester University's Small Scale Experimental Machine, nicknamed Baby. It ran a program in June, 1948.

Built by Fred Williams and Tom Kilburn, it used cathode ray tube, rather than mercury memory, and was used to validate the CRT technology. Baby evolved into the Manchester Mark 1 and later the Ferranti Mark 1.

Either way, EDSAC entered service as a practical research tool at Cambridge as soon as it was working. All these machines pre-dated their American equivalents.

Bennett moved on to Ferranti in 1950, returning to Australia in 1955 as the first professor of computing science at the Basser School at Sydney University where he led the team that built SILLIAC.

Wheeler headed for the University of Illinois, from whence came ILLIAC, forerunner of SILLIAC, returning in 1953 to write an index register for EDSAC, and along with Wilkes and Stanley Gill is credited with inventing the subroutine.

All left a significant legacy of library routines, particularly Wheeler's interpretive floating point routine, Bennett's interpreter which allowed for compact programs and his pioneering work in computation for X-ray crystallography.

At a 50th anniversary celebration of EDSAC's creation in 1949, a BBC program of the event gave the computer's vital statistics as: • 650 instructions per second • 17-bit words of memory in mercury ultrasonic delay lines • Paper tape input and teleprinter output at 6 2/3 characters per second • 3000 valves • 12kW power consumption • "Operating system" occupied 31 words of read-only memory • Occupied a room five metres by four metres • Early use to solve problems in meteorology, genetics and X-ray crystallography • First book on programming by Wilkes, Wheeler and Gill published in 1951 • Magnetic tape backing store added in 1952 • First course in Computer Science at Cambridge started in 1953, using the EDSAC. For the laboratory's director, the ensuing years were a steady progress through innovation and technical development in programs and methods.

EDSAC 2 was started in 1951 with a grant from the Nuffield Foundation, but research work continued into astrophysics, theoretical chemistry, X-ray molecular biology, atmospheric oscillations and radio astronomy on the first machine.

Over time, the load is shifted to EDSAC 2, the first full-scale micro-programmable computer, and the first bit-sliced machine, both qualities a direct result Wilkes' research.

In July 1958, EDSAC 1 was shut down and little of it remains beyond banks of valve sockets, a memory tank and the original engineering log displayed in a glass case in the foyer of the laboratory's current building.

TITAN, a modified version of Ferranti's Atlas, was to follow, and later IBM and Digital PDP-11 kit to support the Cambridge Ring project in 1974.

Retirement into private sector work In 1980, Maurice Wilkes retired to be succeeded by Roger Needham. A decision ordained more by university policy than failing energy, he accepted an offer from Gordon Bell to join Digital at its head office in the little mill town of Maynard, Massachusetts, to work as consulting engineer.

His tenure of nearly six years saw him involved in the development of Digital's high-end VAX range, and as the company's representative on Project Athena at MIT.

Started in 1983 with Digital and IBM as sponsors, Athena would see the development of a 1300-user client/server network on MIT's campus. Initially running on Berkeley 4.3 Unix, it later moved to vendor-based systems and applications and was incorporated in the campus' academic infrastructure in 1991.

Wilkes returned to his Huntingdon house and to the university as emeritus professor. He was knighted in 2000.

The Australian connections Talking in his small modern office (the first through the massive security door to the engineering department), he apologises for a failing memory of his first visit to Australia, but laughingly recalls thumbing a ride on an RAF Hastings transport to get there.

Although a civilian, he was given equivalent military rank during the war, and afterwards during his visits to Germany to speak with German scientists and evaluate captured technology, he rose to Group Captain.

He was invited to Salisbury, South Australia, by John Ovenstone in 1957 and used his status with the RAF to organise transport.

(He was accompanied by Tom Kilburn, James Wilkinson and A.S. (Sandy) Douglas - who created the world's first computer game, noughts and crosses, on EDSAC's tank CRT display.)

A protege of Trevor Pearcey, Ovenstone had studied at Cambridge under Douglas Hartree and had been sent to unravel data processing logjams at the Weapons Research Establishment in Salisbury.

WREDAC (WRE Digital Automatic Computer), made by Elliott Brothers in the UK and installed in 1955, a couple of years later automatically processed a telemetry tape from Woomera in what was believed to be a word first.

Ovenstone and others called a computer conference in June 1957 to celebrate this feat and generally to show off their new equipment. It was not the first computer conference in Australia but the biggest to that time.

Wilkes warms to the recollection: "He was a real character and we were anxious to see what he and his team had done, and also to catch up with Trevor (Pearcey) who I'd always admired.

"But the trip in a piston-engined Hastings took eight days - we went everywhere to get there, spent a few days in Australia which were wonderful, and then another eight days getting home."

The journey included a visit to Sydney to see former student John Bennett's progress with SILLIAC, and the conversation turns to questions after "young John's" health and recollections of being in Sydney.

An enquiry after Trevor Pearcey's welfare creates a pause as learns that he died in 1998. "He did wonderful work."

A change in the conversation's direction seems apt, but a question on where he thinks computing is going brings a laconic reply: "I have never been a visionary or predictor of technology; as a mathematician I have tried to take technology forward by developing opportunities as I have seen them.

"The major development in computer science, however, has been, simply, reliability. When we started, machines broke down all the time and maintenance was a continuous process for years. Transistors and other technologies changed that and now users can reasonably expect to get started and keep going."

Birth of a Society The 1957 visit was significant in that during the Conference on Automatic Computing and Data Processing, there was much international discussion about the benefits of a professional society.

Pearcey was later to write that the discussion was to lead to the formation of the British Computer Society in 1957, with Maurice Wilkes as its foundation president, and initiated the idea of an Australian computing fraternity.

That found substance in 1961 with the foundation of the Victorian and South Australian branches of the ACS with others soon to follow.

"Belonging to a professional society," Sir Maurice says,"adds a new dimension to professional values - it sets people apart

"Computer practitioners have a responsibility to their profession and membership of a professional society is its foundation."