A note on the early days of computing in and related to St Andrews University – Part I
On the 4th January 1965 the University of St Andrews Computing Laboratory started a full-time computer service in St Andrews. In celebration of its Silver Jubilee in 1990, the Computing Lab issued a special supplement to its newsletter. Below is part I of the article contributed by Professor Alfred Jack Cole (1925-1997). Part II will be published on World Digital Preservation Day on Thursday.
“It has been exciting to have been involved in the development of computing, if not from the beginning then certainly from the very early days. My own active involvement began with a summer vacation consultancy from the Heriot Watt College as it was then called, in the summer of 1953, at the Royal Aircraft Establishment in Farnborough where I worked with engineers and programmers on the automatic control of machine tools. The computer being used at Farnborough at that time was an ACE (Automatic Computing Engine) which had been developed by the National Physical Laboratories Manchester Group in collaboration with Ferranti. Alan Turing (of theoretical Turing machine fame) was an active member of the design team. I learnt a very important long-term lesson in computing at that time since the work I had done in writing efficient software to enable real time control to be implemented was made redundant in a few months when the ACE was replaced by an English Electric Deuce which operated at a tenfold increase in speed.
Curiously some years before this I had played cricket in a very unsuccessful team which included a young statistician who was an enthusiastic member of the design group for the LEO computer. This was an extremely far sighted project sponsored by Joseph Lyons Tea Shops to design and manufacture their own computer for stock control and accounting. The polymath choice of design team and indeed the choice of name of the machine itself, which not only had a punning reference to the company name but also stood for Lyons Electronic Office, anticipated by nearly half a century both modern computing practice and terminology.
Early computing in Glasgow and Dundee
Having returned to Edinburgh and introduced a variety of computing courses in the Heriot Watt College, I was appointed in 1956 to a lecturership in mathematics with a special interest in computing at Queen’s College, Dundee which was still at that time part of the University of St Andrews. About this time an English Electric Deuce was installed in Glasgow University to serve the computing needs of the Scottish Universities. The offer of this machine had been turned down by Edinburgh University on the advice of Professor Aitken who was Professor of Mathematics and a mental arithmetic prodigy. At Glasgow members of staff from other universities had their first experience of liberal open shop computing outside of normal daytime working hours. Staff from Dundee booked several hours of computer time in the small hours of the morning and, supposing that the Deuce was operational when we arrived, we split the time between us and then dashed back to Dundee to have a quick breakfast and on to a nine o’clock lecture.
Dr John Iball who was a permanent Cancer Research Council funded Senior Research Fellow in Chemistry gave enthusiastic support to a scheme to purchase a computer for Dundee. Having failed to get direct funding from the University Court and with, at that time, no Computer Board or other UGC supported funds available we managed to gather sufficient funds from various University Departments, who showed their willingness by mortgaging not insignificant proportions of their class and research grants, and from Industry in Dundee, to convince the Court that computing was not just a passing whim. Standard Telephones and Cables manufactured a machine called the Stantec Zebra and came up with a much discounted offer for a valve machine which had been designed by Professor Willi van der Poel of Algol 68 fame who worked together with a remarkable computer engineer named van der May who was both blind and deaf as a result of meningitis attacks in his youth. Together they had designed a very advanced machine which was in effect microprogrammable, had a modular design based on about seven user replaceable units with associated diagnostic software and which utilised permanently wired tracks of the drum to give a bank of fast access registers.
This machine was installed in Dundee in 1961 at an eventual purchase price of £13,000 which involved a down payment of £7,000 and £1,000 a year for six years. It gave good service for a number of years and was eventually replaced by an Elliott 4130 series machine. A University Computer Committee had been set up and at the request of the St Salvators members who were Professors Allen, Copson, Dingle and Stibbs, a condition of purchase was included in the final agreement that the machine was to be permanently housed in Dundee so as not to prejudice any subsequent negotiations for a machine to be installed in the St Andrews Colleges. A number of staff and students from the St Andrews Colleges including research students of Professor Ron ( otherwise Andrew ) Mitchell, Mr Tony Davie and other mathematicians and chemists came across to use the machine. One of the early users in Dundee was Dr Killean who was at that time in the Dundee Physics Department.
Following our experience in Glasgow the overnight arrangements were on an open shop basis for people who had passed our driving test. This was successful since most people behaved responsibly with the exception of one physicist – not, I hasten to add, Dr Killean – who used to bring his screwdriver with him on overnight sessions and who was the catalyst for the introduction of daily morning diagnostic tests.
The Elliott 803 at Leicester
In 1962 I moved on to Leicester University and my good friend and colleague Dr John Rushforth took over responsibility for the Dundee machine. The machine in Leicester was a transistorised Elliott 803 with magnetic core store [memory], floating-point hardware and superb paper tape readers which had been designed originally by the EDSAC team in Cambridge. Tony Hoare of Oxford University was at that time an employee of Elliott and was responsible for writing the Algol 60 compiler. There were a large number of small and medium sized computer manufacturers at that time and a lot of very talented people were involved in their design groups. Unfortunately, the sorry tale of British Industries’ missed opportunities had started even then and most of the computer manufacturing activities of those companies were either taken over or the companies themselves went out of business to leave many of their bright ideas to be exploited in America.
Appointment to St Andrews
In 1965, to get back to the sanity of life in rural Scotland, I applied for and was appointed to a senior lectureship as Director of the Computing Laboratory in the University of St Andrews, which was about to be separated from Dundee. My spectacular promotion to Reader at 12 noon on my first day in the University was in no small way due to the advice at the interviewing committee of that well loved administrator David Devine.
The IBM 1620
The machine, already installed and run for about nine months by Dr T R Carson of the Astronomy Department, was an IBM 1620 model II with index registers, automatic floating point, core storage, disk drives, online printer and a high speed card reader/punch. An online graph plotter was added later. The disk drives which were minute in capacity by modern standards were nevertheless advanced for their period and the disk operating system which included monitor and utilities was the model for many subsequent IBM systems. The design had been based on earlier commercial systems and the whole arithmetic hardware was based on binary coded decimal digits which were ideal for accounting but were much slower than pure binary for most scientific computing applications. However, there was one very nice way in which this design feature was exploited by the compiler writers in that by using a single control card it was possible to change the length of integer and/or floating point arithmetic at compile time thereby being able to make quick empirical checks on some forms of round off error and also being able to extend the range of solution of some problems. At the assembly language level it was possible, if not very practical, to carry out arithmetic operations on integers with thousands of significant digits and reals up to 100 digits mantissa.
One serious disadvantage of the machine was that it had no hardware interrupt system which meant that it was essentially a single stream batch processor, even at the input/output level. Having said that, it was nevertheless a very reliable machine which played an important part in the development of computing in St Andrews. This was long before the days of unbundling of hardware and software and the collaboration of IBM technical staff was most welcome.
The software for the machine was largely based on Fortran roots with good Fortran II-D compilers and associated mathematical and statistical packages and algorithmic libraries. More efficient applications had to be written in assembly language which was good by the standards of that period. One Algol 60 compiler was obtained from Nancy University but despite valiant efforts from Mr Davie was never made to work completely. All of the reserved words were in French so it was amusing after much effort by Mr Davie to get a program through the lexical analysis stage, to be met by the message “End of pass one” in English. We were never sure if this was the work of a mole or just some early franglais.
It may surprise artificial intelligence experts to know that an early version of ELIZA [a program simulating a psychotherapy conversation] was available and performed in much the same way as more recent versions.”
This article will conclude later this week.