Gordon Clarke

One day during the summer of 1958 I was strolling across Front Square in Trinity College Dublin after I had given some maths grinds to an engineering student. I was met there by one of the staff of the Irish Civil Service (Permanent) Building Society and handed an urgent message. My father, who was chief executive at the building society, had been given a phone number for me to ring as soon as possible. I called the number and was asked to attend immediately for an interview at the premises of Calculating & Statistical Services (C&SS) in 15 Harcourt Street. I protested that I was very informally dressed, but was told to come at once. This I did (on my Vespa Scooter!) and about 30 minutes later I was interviewed by a Mr Percival from the British Tabulating Machine (BTM) Company.

I had graduated in the previous year and taken a part time position as assistant mathematics teacher at St Columba’s College. I was still attending science lectures at the university with a view to taking a degree in that subject as well. I was particularly interested in electronics and remember putting together a transistor radio from a collection of parts. I later read that, at that time, transistors were not yet fast enough to be used in computers.

Sometime earlier I had asked the College Appointments Officer to let me know of any possible openings in the electronics industry. My call to Harcourt Street arose from this.

Mr Percival was BTM’s personnel manager. He told me that the company had installed the first computer in Ireland at the Irish Sugar Company and planned to replace technical staff seconded from the UK with Irish personnel. He offered me a position that would involve training in England for a year with a view to returning to Dublin and providing support for the Sugar Company and for other potential computer customers. Shortly afterwards I received a formal offer, stating that I should report to 17 Park Lane in London on a date in September.

BTM trainees at Moor Hall in 1958
(Photograph by Gordon Clarke)

BTM’s training involved residential courses interspersed with periods of work experience. It was standard practice that trainees should stay away from the offices where they eventually intended to work. This was to avoid the potential embarrassment of showing a lack of knowledge to customers whom they would later be advising as experts. In my case, however, they made an exception and I worked in the Dublin office with Alec Willies, the English engineer (or ‘systems advisor’ as he was known) whom I was to replace. We not only kept a watching brief on the Sugar Company installation in Thurles, but also supported other installations with wired programmable calculators and traditional punched card installations.

BTM’s basic training focused on punched card technology. This involved three sessions of 8-9 weeks at Moor Hall, a manor house in Cookham near Maidenhead. In addition I took a two week course on the HEC 555 programmable calculator and a three week course on programming the HEC 1201 computer. Both were held at Bradenham Manor – Disraeli’s former home near West Wickham in Berkshire. Ours was the first all-graduate group of trainees and we were treated very well during the courses. I recall a visit to Monkey Island on the Thames, a trip to Leoni’s ‘Quo Vadis’ a noted Italian restaurant in Soho and lunch in the Park Lane Hotel.

Gordon Clarke revisits Bradenham Manor in 2016

We also visited the original BTM factories at Letchworth and Stevenage in Hertfordshire. Not so many years earlier, during the Second World War, these had manufactured the ‘Bombes’ used by Turing and his associates in their decryption systems. In another link with that era the company leased an apartment in London which had been the wartime residence of Charles de Gaulle. It used this on occasions when it was particularly keen to impress a prospect. Indeed, we once held a private dinner there for senior officials from a government department in Dublin.

In 1959 BTM merged with Powers-Samas to form International Computers & Tabulators (ICT). I was then required to spend a further four weeks training on Powers-Samas systems at Fulwood Place in London.

After the completion of my training schedule, I was expected to spend a few months in one of the English offices. I was assigned to the department responsible for contracts to the Armed Services, based in the Liverpool Victoria Building which still stands on the west side of Southampton Row in London. I spent the first few weeks in HM Ordnance Factory at Woolwich, but was told that I would be going to Scotland to manage a project at Rosyth Naval Dockyard. This was in October 1959. There had been considerable problems in this site, leading to questions in the Commons and bad publicity for the company. Having resolved the problems initially assigned to me, to their satisfaction, the officer in charge of the IT systems requested ICT that I should be assigned for a further period to design and implement additional systems. I did get a final release date from them and was able to schedule my wedding in Dublin in May 1960. I returned to Scotland for just two weeks following a honeymoon in Italy. From then on I was based in Dublin.

ICT’s organisation in Ireland was still an agency operated by C&SS rather than a branch office (see side panel). C&SS also had a branch in Belfast. It offered two complementary bureau services: a comptometer service for the computation of stock valuations and a punched card bureau service. Both were staffed by specialist operators.

Mrs Metcalfe, the owner of C&SS, would visit the premises each Christmas to present every member of staff with a £10 note. Our chief accountant, Acheson Black, recorded this as pay and deducted tax where appropriate. In those days ￡10 was a significant amount and would have bought about 170 pints of Guinness.

I was assigned responsibility for a number of installations, but particularly for the ICT 1201 in the Sugar Company. We had to ensure that the computer could not be blamed for any delay in its payments to farmers. It was a valve-based machine and had no conventional memory. The program was stored on a drum whose capacity was 1024 words of 40 bits. Subroutine calls had to be preceded by appropriate placement of the return address. Each instruction included the address of the next instruction, and branching instructions included two such addresses. The programmer had to take account of the latency of the drum rotation and place the next instruction in the most efficient location to save time. Large temporary results could be stored on the drum, while smaller data could be held in one of the 4×40 bit arithmetic registers. The hardware catered for multiplication. Division, however, required a programmed subroutine.

Programs were written in the Octal numeral system, where each three bits were written as a number in the range 0-7. The program was punched and then run through the system which converted the lines of code to binary. Twelve lines of code could then be compacted into a single punched card. Testing was relatively easy, as one could single step through the program and monitor the register contents on the console. Valves, however, were not as reliable as solid state circuitry. One sometimes found errors in operation that were not due to the programmer and required engineering maintenance solutions.

On one of my early visits to the Sugar Company I met Tom Winter who worked for ICT in Belfast. He had gained access to the computer in Thurles in order to test software for a new ICT 1202 at the Northern Ireland Revenue service.

Norman Frances from (I believe) the Liverpool office wrote the original programs for the Sugar Company and I recall resolving one minor problem in his code. The Dublin bureau also sent one of its most experienced and efficient operators, Brian Pardoe, to work in Thurles before the company trained its own staff as operators.

In 1962 ICT introduced its transistor-based 1300 series – a huge step up from the valve technology in the ICT 1200. It used ‘magnetic core’ memory in units of 400 words. The modules of computer hardware were delivered separately and assembled by lining them up and using a joint wrapping system to establish the connections. The cost of addressable memory was a serious constraint for computer users before the advent of integrated circuits and solid state memory.

The first ICT 1300 in Ireland was delivered to the Esso offices in Stillorgan at the end of Brewery Road. The Sugar Company also ordered one and I used to visit the Esso offices in the evenings to use their computer for program testing for the latter. I recall one evening when, just before leaving home for Stillorgan, I heard on the radio of the shooting in Dallas of JFK.

Another advance was the use of memory mapping and associative memory. This became commercially viable with the arrival of printed circuitry and cheaper memory. It was used by IBM in the System/360 series, which was launched in 1964. It had been used in earlier days by Ferranti in its Atlas (and I think Orion) systems. These were very expensive, were produced in small numbers and had been supplied mainly to UK government bodies such as the Met Service.

ICT took over Ferranti Computers in the early 1960s and brought out the 1900 series, based on Ferranti technology, to compete against IBM’s new machines. These systems had the disadvantage, however, of being based around a 6-bit character unit at a time when the 8-bit byte was becoming the de facto standard. ICT negotiated a reselling arrangement with RCA and offered its systems, as the ICT 1500 series, in situations where the byte standard was considered critical. I had to attend training courses on both series in London.

As it turned out, though, we only sold 1900s in Ireland. They were popular with the customers that we were already doing business with. Only the top tier of manufacturing and distribution companies offered realistic prospects for the ICT 1900, because the technology was still so expensive. A large number of companies continued to use unit record equipment right through the 1960s. New users generally turned to bureau services as their first step.

ICT assumed direct control of the former C&SS organisation in the early 1960s. Peter Mullett became the company’s general manager for Ireland, assisted by Michael Appleby, who had been the Powers-Samas manager in Dublin. I was told that he had previously been an RAF pilot in the Battle of Britain.

Another industry veteran, Alan Exley, had sold the first electronic stored program computer in the country to the Sugar Company. A gentleman of the old school, he was only a few years from retirement when I arrived on the scene.

This photograph of ICT’s management in Ireland was taken during a visit to Dublin by the company’s chairman and chief executive Cecil Mead in November 1965. Back row: Peter McGrath (service bureau sales manager), Bryn Davies (administration), Charlie O’Brien (service bureau operations manager), Frank Smyth (personnel manager), Joe McGee (deputy engineering manager), Atcheson Black (accounts), Jim Crossen (sales), Hugo Patterson (PR). Middle row: Charles Cooke (administration), Gordon Clarke (computer systems and support manager), Tom Corbett (engineering manager), Nick McGillycuddy (sales manager), Ewart Shaw (sales). Front row: Michael Appleby (deputy general manager), Geoffrey Elbourne (London-based divisional manager), Cecil Mead, Peter Mullett (general manager), John McBurnie (Belfast manager). (Photo source: Gordon Clarke)

The ICT staff with whom I worked closely in the early years included brothers Kevin and Charlie O’Brien. Kevin had been one of the designers of the ICT 1300 series. I reported through him until he moved to a company called Business Operations Research. Charlie O’Brien joined ICT in about 1961. He had worked as an operator for the Leo Computer Service Bureau in England and he later took charge of the bureau in the Dublin office. The company installed an ICT 1202 for this service. C&SS had always run a punched card service bureau and it was a natural progression to offer computer-based services.

Michael McMahon had been recruited just ahead of me in 1958 as a systems advisor for punched card unit record equipment . His skills in the sales area were memorable. He came from Ennis and, when the opportunity of managing a new company in Shannon arose, he moved back there from Dublin.

The bureau’s sales manager was Peter McGrath. Along with Matt Farrelly and Paddy Maguire, he left ICT and established Independent Computer Bureau Services in 1967. This, I understand, was the first independent bureau service in Ireland.

Don Dennehy was originally from Cork and returned there in the mid-1960s to manage ICT’s regional office. I recall him mentioning another Corkman, John Daly, who worked for the company in the UK, saying that it would be good if he were brought back to Dublin. This came to pass in the early 1970s when John Daly was appointed chief executive of the Irish operation.

My own job enabled me to remain involved in systems development work, while other teams worked under my direction. As more customers acquired their first computers in the mid-1960s ICT provided considerable assistance in system design and development. Most of the early applications were for accounting tasks. Companies frequently justified the cost of their systems by reducing their clerical costs. Access to additional information was a bonus.

At New Ireland, for example, we were able to innovate on the accounting systems of its Irish National Insurance subsidiary. Generally, in those days, insurance systems did basic maintenance work on policy files and managed the accounting records associated with renewals. In the case of Irish National we were able to effect further savings by setting up a successful algorithm for allocating cash automatically into agents’ accounts. By today’s standards this might seem trivial, but it was considered quite an advance at the time.

PAYE was another large value application. In the early 1960s there were companies that paid thousands of employees in cash. One of their tasks in the preparation of pay packets was coin analysis, so that the company could inform a bank how many coins and notes in each denomination were needed to pay its workforce. The old pre-decimal coinage of shillings and pence complicated things further. The advent of unit record systems, followed by the arrival of computers, transformed payroll administration and enabled employers to introduce more complicated incentive schemes.

Manual payroll routines required large numbers of clerical staff, just as manual methods called for many hands on the factory floor. I witnessed automation eliminating tedious processes in both areas during those years.

I enjoyed the challenge of providing solutions outside common accounting procedures – projects such as special production applications for Sunbeam Wolsey in Cork and route and van loading systems for the delivery of Jacob’s Biscuits. The bureau provided more opportunities for varied applications. One of its customers was Tynagh Mines, which extracted lead and zinc. It also produced some copper and small deposits of silver. Our systems used drill sample readings to assign ore grades to large blocks in the ore body. Mining engineers were then able to select those blocks which had the highest copper grades. This was particularly useful at times when the price of copper was high.

ICT developed a software package for project management services after the success of PERT in the development of the Polaris submarine. I attended a course on this package, which did not include any resource management features, and provided support to bureau customers that availed of it. I recall O’Neills and Cramptons as being early PERT users.

The punched card remained the main file medium throughout these years. Magnetic tape was sometimes used, but its cost could only be justified for very large files. In general, with cards and tapes  alike, individual records could be only accessed by running through the entire file or a selectable subset. It was not until the mid 1960s that affordable disk media became available. These provided direct access to records, but the costs were substantial. If my memory serves me right, an interchangeable cartridge disk drive with a 9Mb capacity cost around ￡12,000 in 1969. This would have been at least twice the annual salary of a top executive.

I joined the British Computer Society (BCS) in 1963 and started to attend meetings of its Belfast branch. Tom Winter, who I had met in Thurles some years earlier, was now a leading light in this branch. After a few years there were about ten other people who travelled up from Dublin and we began to see ourselves as a loosely formed society. We also held cross-border conferences at the Ballymascanlon House Hotel outside Dundalk. These meetings had a social aspect but they enabled us to keep in touch. The Irish Computer Society started in 1967 and it adopted the BCS grade structure in 1972.

Many of the country’s experienced computer systems professionals in the later 1960s were my former colleagues. Before the universities started computer science courses, the companies that manufactured the equipment provided all the training. At the start of the decade most computer professionals were employees of these suppliers. Customer organisations subsequently recruited many of them, as did the bigger accountancy firms. In the same way that these practices provided financial advice to their clients, they started to offer expertise in systems consultancy and software development.

I left ICT (or ICL as it had just become) in early 1969 and moved to Aer Lingus as a senior systems consultant. The company was setting up a systems services division, led by Maurice Foley, to sell consultancy and systems development resources. It also provided bureau services, using spare capacity on the airline’s IBM System/360s, to customers such as VHI and the Department of Posts and Telegraphs.

Shortly after I joined, Aer Lingus bought Irish Computer Bureau Services which thereafter traded as Cara Data Processing. My responsibilities changed to development manager and later I served as general manager of Cara Consulting.

Last edit: July 2017

© Gordon Clarke 2016-17