• Project Manager Operational Systems
• Business Analyst - Melbourne CBD...
• Project Manager - Permanent...
• Senior Project Manager - Multi...
• ACCOUNT MANAGER

Bombs, bells and bureaucracy

Peter Davidson, Information Age

12/06/2004 15:37:02

The Luftwaffe did its best to distract a young Godfrey Lance's efforts to hone the extraordinary mathematical skills which would carry him through a doctorate and a singular career in computing.

Like a number of his contemporaries, a degree in honours mathematics led to aeronautical engineering in the decade after WW2 and a rapid embracing of computing to crunch the mountains of arithmetic involved in the quest for higher and faster fighter aircraft.

In his case, it is a career studded also with major academic appointments in Europe and the US - and an early realisation that those striving to advance computing's capabilities needed to unify in their quest, first in the British Computer Society and later as a significant catalyst in the birth of the Australian Computer Society.

But back to the War: He swotted in the cramped chaos of below ground air raid shelters while the bombs fell. On one occasion there was a near miss of his home. He finally sat his matriculation exam in the shelters among the huddled mass sheltering from the Blitz.

With a personal score of Lance 1, Luftwaffe nil, in 1946 he entered London University's Kings College to begin honours in mathematics, joining second-year honours physicists in a combined class under the tutelage of a Dr V C A Ferraro.

"Ferraro was a great character who always called the roll, perhaps because it was the first class on a Monday morning," Lance recalls. "I particularly remember two names very clearly: one was Arthur C Clarke (now Sir Arthur) who became the best known of all of us, and the other, a Mr Zobel simply because he was the last of a very long list."

His third year called for study in three "advanced subjects" from a wide choice as the London Colleges pooled their hons maths students and offered the best lecturers available to teach them. Lance chose meteorology, quantum theory and numerical methods, the last setting him on a course in computing.

"In those days we had only hand calculators like Marchant, Brunsviga and Facit machines, capable only of basic arithmetical add, subtract etc functions, with only the electrical Marchant offering relief from hand-cranking. Otherwise it was endless books of tables."

But armed with a first class honours B.Sc, he chose to go on to do a masters by examination (in 1950), adding aeronautics to quantum theory. A further two years of research, in aeronautics only, led to a Ph.D in 1952

His thesis, "Some Problems of Supersonic Dynamics" was apt at a time when high-speed aircraft design was a priority; Cold War temperatures in the 1950s were dropping and the Allies were bent on maintaining a strategic advantage over the Russians.

In the meantime, university vacations and other opportunities were spent getting real experience and with aeronautics in his sights, the Hawker Aircraft company was a natural choice, allowing him to apply burgeoning skills in one of the UK's foremost aviation outfits.

Hawker's chief designer, the redoubtable Sir Sydney Camm, lured the young Dr Godfrey Lance to his design office to work on the supersonic Hawker Hunter which Camm had designed. It first flew in July 1950 and stayed in service with air forces around the world for 50 years.

Many still fly.

Like the Hurricane, which Camm also designed to be the front line fighter in the Battle of Britain (it shot down more German aircraft than all the RAF's other types combined), the Hunter became a famous convocation of form and function, but its developmental years called for vast mathematical work.

"As a mathematician I was doing mostly flutter calculations for the Hunter's wings and control surfaces." The volume of number crunching got to the point where Camm agreed to get "one of those new fangled high-speed computers".

"We selected the Ferranti Pegasus primarily because of its ease of programming."

His work at Hawker was becoming repetitive as the search for improvements to the aircraft went on. "The difficulty was that it had been so well designed in the first place that there was little room for other than tweaks."

Water bells Any tedium was leavened by his continuing fascination with water bells, the phenomenon of water assuming a bell-shaped "skin" when pumped through a small vertical tube against a flat plate. Surface tension pulls the bottom of the sheet of water in until it forms its characteristic shape.

His interest in the phenomenon, first examined by Joseph Boussinesq in the late 19th century, and the way that pressure, density and volume of fluid in the tube would change the shape of the bell, began at Kings and would continue long after.

"The shape is primarily influenced by the difference of the air pressures inside and outside the bell, with shapes predictable by a non-linear ordinary differential equation."

The journal of the Physical Society carried a paper by G N Lance and R L Perry in August 1953 in which "A numerical method of solution of the differential equation of the bell's surface is described and the results obtained are given in the form of graphs".

The agreement between their theoretical and practical experiments verified Boussinesq's work of a century before.

A new challenge came with a position as visiting lecturer in engineering at UCLA, working with aeronautical engineer Prof John W. Miles who was internationally recognised for his work in fluid dynamics.

"UCLA had a mechanical differential analyser. It was driven by wheels and cogs and only solved differential equations - but very efficiently. Prof Miles was a prolific producer of scientific papers, probably the most productive person I've ever known, and he gave me a lot of freedom.

One result was further recognition by the Physical Society in his work in water bells, this time with an E C Deland of UCLA.

"The water bells were fun but largely unscientific as they have no real value."

Later work in France has applied water bell theory to the behaviour of fuel vapour in combustion chambers, however.

His computer work went on at UCLA including learning to program its digital computer SWAC (Standards Western Automatic Calculator), built by Harry Huskey in 1950 in its Institute for Numerical Analysis. Huskey was previously involved with the development of ENIAC and EDVAC.

"Its display and memory devices were based on cathode ray tubes (CRT), and programming it meant dealing with the individual binary digits in each word, with programs entered using 80col punched cards.

"SWAC was much faster than Pegasus, but also much more difficult to program."

Returning to England in 1955 as lecturer in mathematics at the University of Southampton, computing soon became a priority when the government offered funds to seven major universities to buy high-speed computers.

Southampton chose a Ferranti Pegasus, again because of its ease of programming, an essential given it would be used by students.

"Five of the seven chose Pegasus, the others the EE Deuce, so perhaps my choice was vindicated to that extent."

Appointed director of its computing laboratory, supported by two other lecturers and a maintenance engineer, it meant "we ran Pegasus, taught all and sundry to program it and gave lectures on numerical methods - calling the subject computer science or information technology were later ideas".

I will never forget the first programming course we gave. There were 12 people attending and their skills ranged from a 6th form schoolboy who wanted to calculate tables of logarithms, someone who wanted to run the gas company's payroll on Pegasus and our Professor of Botany Prof W.T.(Bill) Williams.

"Bill was an intriguing character and we became very close friends and colleagues; between us we wrote over 100 papers on the subject of numerical classification and it's applications to all sorts of subjects. In fact Bill joined me at DCR in Canberra in about 1966; he's the only person I know who gave his hobby in Who's Who as 'beer drinking'."

During this period he became a foundation member of the British Computer Society, formed in 1957 with Sir Maurice Wilkes as its foundation president. Lance chaired its publications committee for several years and was also a member of the Council. He was soon to bring his enthusiasm for a unified body of computer professionals to Australia.

He still holds BSC membership number 1066 and is now a Fellow of the BCS.

A stint at the UK Atomic Energy Authority at Winfrith, in Dorset, in 1960 as head of its computer branch ("we worked on reactors, not bombs") brought him into contact with a new Ferranti Mercury computer.

"Winfrith always had programs too big for Mercury, and we had to use the Atomic Weapons Research Establishment's (AWRE) IBM Stretch at Aldermaston. Cards would be flown there in the morning from Winfrith in Dorset, and print-outs of the results flown back in the evening.

"It was cumbersome and slow so we later put the data onto magnetic tape and sent it to AWRE over a dedicated landline connected to an IBM 1402 front end. It was a very primitive network but it worked and was a breakthrough."

Pearcey's foundations At the other end of the world, Trevor Pearcey had in 1962 proposed a similar "network" concept to the Council for Scientific and Industrial Research (later the CSIRO) which called for large data volumes on magnetic tape to be flown to distributed sites around Australia for batch processing.

The mag tapes would then be flown back to Canberra overnight.

Pearcey's proposal followed his pioneering work on the CSIR Mark 1 (later renamed CSIRAC) and CIRRUS, and was part of a survey of the organisation's future computing directions.

When the CIRRUS development team broke up after the home-grown minicomputer was allowed to wither as a commercial opportunity through lack of support, Pearcey joined Control Data in Minneapolis for a time, later returning to the CSIRO's computer research section.

But his networking vision was gaining support, and an international trawl for someone to head the research section, (later elevated to Divisional status,) to implement what would eventually become the hugely successful CSIRONET, saw Godfrey Lance selected in 1962, and installed in Canberra the next year.

His first task was to evaluate and select the computers to support a network which would connect Canberra initially with Sydney, Melbourne and Adelaide, and after much cogitation, a Control Data 3600 was selected for Canberra as the core of the system, to be connected to three CDC3200's in the other centres, and a fourth in Canberra.

An extract from Lance's annual report for 64-65 notes that their 3600 "has a core store of 32,768 words each of 48 bits . . . a comprehensive instruction set including single and double precision floating point arithmetic, single precision fixed point arithmetic and a full set of logical operations".

Programming languages were FORTRAN, COMPASS (assembly), ALGOL, COBOL and SORT, it had a monitor system called SCOPE, and an array of mag tape, punched card and paper tape units, line printers and graph plotters.

The selection of Control Data kit was coincidental with it also being the choice of the Commonwealth Bureau of Census and Statistics (BCS - later to become the Australian Bureau of Statistics) which was gearing up to computerise its databases.

"At the time it was rumoured that Keith Archer (BCS) and I were in collusion but this is quite untrue; we made our own independent decisions. Perhaps just great minds thinking alike ..."

The Division of Computer Research (DCR) was set up to run the network, help other divisions to exploit its capabilities efficiently and to undertake original research in computer methods.

Agencies and institutions wishing to use the computers paid 15 pounds an hour for 3200 time in Sydney, and 60 pounds an hour for the 3600 in Canberra.

(Peter Claringbold, appointed by Lance as a chief research scientist [and later to succeed him] knew as well as his boss that the network could become a valuable money spinner to fund research and worked to develop its potential. Bureaucracy and politics were to interfere however . . .)

The division moved into a new building at Black Mountain in Canberra, and a suggestion by the Chairman of the CSIRO's Executive Sir Frederick White that there should an official opening was greeted with Lance's retort that "politicians would be involved over his dead body".

Sir Frederick offered no resistance and it fell to Sir John Cockroft, 1951 Nobel Laureate for his "splitting" of the atom and at the time Chancellor of ANU, to dedicate the building which still stands as part of the CSIRO campus.

By 1966, the network had expanded well beyond Lance's original Winfrith/Aldermaston model, offering access to the 3600 from remote sites via dumb terminals, supported by an operating system for the CDC machines called DAD - Drums And Displays. DAD was written by Dr. Brian Austin, Henry Hudson and others specially for the network.

Its conversion to the electronic CSIRONET in 1968 began with experimental landline links to a DEC PDP-8, and by the early 1970s was reaching powerful tentacles across the country, served by PDP-11s.

For Lance, his tenure at DCR had been one of the longest of his career, and in 1972 he took a sabbatical, of about 18 months, as Chief Scientific Liaison Officer for the Australian Government in London, a position held by Chiefs of CSIRO divisions on a rotating basis.

Much was happening with the OECD in its early development, and this required frequent trips to Paris and the oddity of travelling with an Australian diplomatic passport while (at that time) still holding a UK passport and living in London.

But he was able to spend time with his parents in their final years, and to keep his computing hand in by using London University's new CDC 7600 machine "to practise on". (A 7600 was on order for CSIRO.)

Farting dogs In the meantime, Claringbold was minding the store in Canberra and coping with politics, and later recounted that "the worst day I ever had was when Godfrey Lance was away on sabbatical. While I was in the job of acting chief they [the bureaucrats] said: 'You can't spend the money you're getting as revenue any more. It's got to be appropriated.'

"As soon as they said that, I knew we wouldn't get the money for our research. It was just as likely to go to treating farting dogs."

(Claringbold had a Ph.D in veterinary science so his cynicism may not have been totally misplaced.)

But CSIRONET continued to thrive, and according to Claringbold, was drawing $16m in revenue, running as one of the first lights-out operations anywhere - no operators and the place locked up in darkness at night but with all the computers running.

Canberra branch ACS Lance left DCR in 1977, behind him a legacy of having established DCR and guided it through its growth years - and the formation of the Canberra branch of the ACS.

A South Australian Branch had been formed in 1961, largely comprising those working at the Weapons Research Establishment at Salisbury, and a Victorian branch in the same year drawn mostly from academia.

Pearcey had been a driving force in the Victorian branch's establishment and having returned to DCR from CSIRO's radio physics division, joined with Lance to get a branch started in Canberra.

"I recall that the branch founders held their first meeting in my office as it was a bit bigger than Trevor's, but later we had to use the lecture room."

Shortly after, the national association was formed. In 1979 Lance was made an Honorary Life Member, the fourth after John Bennett, Ashley Goldsworthy and Peter Murton.

He returned to England in 1979 to take up a professorial position at Bristol University until 1983, and later moved over to Bristol's computer science department working with research students until finally calling it a day in 1992.

Australia called him back, however, this time into retirement in Bermagui on the NSW south coast where in the afternoon of his 70s he devotes his quiet years to his wood lathe, doing genealogical digs on the Net and spending time with family and grandchildren.