Pegasus the Early Seminal Computer

predominant.

Nicholas and the Elliott 401

The Elliott Research Laboratory was set up in 1946 as part of the effort to capitalize on British technological developments during World War 2. It was attached to the long-established company Elliott Brothers (London) Ltd which had a reputation for highquality electrical and mechanical instruments. Its Director was John F Coales, an inspiring leader. He placed emphasis on his Laboratory being multi-disciplinary.

One of the departments, led by William S Elliott* (Bill) was developing digital electronic circuits, especially those potentially applicable to computers. Annex No 1 gives the original paper discussing Elliott’s work on assembling digital electronic circuits into modules, each carrying out a specific logical function. The electronic circuits themselves were the responsibility of Charles Owen.

At that time the most crucial factor in the design of any digital computer was the technique to be used for the high-speed store. Elliott had read in an American journal that nickel possesses magnetostrictive properties. It is not known whether that article proposed, or Bill Elliott conceived, the idea of exploiting this effect to make a digital delay-line storage unit. In this transducers were placed at each end of a nickel wire and arranged with a circuit to inject a pulse at one end, with a valve amplifier to return the pulse from the transducer at the receiving end to the driver transducer, thus giving a recirculating digital delay-line. Annex No 3 gives an original document discussing a later version of this type of delayline store.

This design had advantages over the mercury delay-line store with piezo crystals used in EDSAC at Cambridge University and in the Pilot ACE at the National Physical Laboratory. It permitted the transducers and the ending of the delay-line to be independent, it minimized reflections at the ends, it avoided the task of machining the ends of a long tube parallel, and the problem of obtaining mercury in much larger quantities than for any previous purpose.

Nicholas* came out of a contract placed in late 1951 by the Royal Aircraft Establishment that involved much calculation. It was valued at £20,000. It was initially thought that the project would be done using hand calculators of the Munro or Brunsviga type. However a mathematician, Bruce Bambrough, realized that the sum could be better used to develop a computer utilizing the logical circuits established in Elliott’s department of the Laboratory together with a storage system based on nickel delay-lines.

So the three-man team of Ed Hersom, Charles Owen and Bruce Bambrough set about designing the computer. The electronic circuits were to be the second generation of those devised by Charles Owen. These are described in Owen’s document ‘Three standard circuits for digital computers’ Elliott Research Laboratories Report no. 301, 2nd Sept. 1952. However, at a crucial moment in the Nicholas project Owen fell ill with mumps, and was off work for three weeks. During this time he devised the complete logic design for the computer, together with the order code. His logic design, on a single large sheet of paper, is still in existence.

What became known as the Nicholas computer thus comprised a test bed for nickel delay-line storage together with a practical machine to carry out the work of the Government contract. It went on to provide a useful computing facility for eight years.

George and Ruth Felton worked on the programming of Nicholas. They report that it was user-friendly from the programmer’s point of view. Ruth Felton wrote the programmers’ manual and fortunately a copy of it still exists. At that time matrix mathematics had begun to be applied to practical problems, and George Felton wrote an embryonic matrix interpretive scheme to assist their use in Nicholas. A particular applications of this was by Peter Hunt of the de Havilland Aircraft company, for the design of the Comet aircraft, the first civilian jet-engined ’plane. Because of the very limited storage of Nicholas, he had to idealize the whole aircraft into only ten elements.

The National Research Development Corporation (NRDC) was set up to exploit, on a national scale, the British technological advances arising from World War 2. It was funded by the tax-payer, but was expected to recover its costs from royalties on patents e.g. the cathode-ray tube store invented by Professor Williams at Manchester University, and by pay-back from sales of sponsored projects.

By the early-1950s its Director was Lord Halsbury. He had great foresight, and was supported by a tiny group of very talented men. It must be presumed that Bill Elliott, supported by John Coales, persuaded him to sponsor a real general-purpose computer based on the techniques established in Nicholas. A contract was placed, and it was given the Laboratory cost-collection number of 401.

Years later, Bill Elliott’s wife Betty told me she would sometimes find Bill had been working obsessively all through the night, and the floor was littered with discarded papers. My surmise is that he was experimenting on paper with the selection and arrangement of basic logical elements in order to arrive at groupings that would provide the minimum number of types of standardized packages. These might have been for the 401, or when optimized for the later Pegasus as described in Glyn Emery’s formal paper reproduced in Annex No 6.

For the 401 Bill Elliott assigned the responsibility of specifying the logical connections of the Charles Owen packages to Hugh Devonald. He also carried out the necessary basic programming of the 401.

Apart from the nickel delay-line storage, the most crucial component of the Elliott 401 was the magnetic drum store. Its importance lay in the provision of a clock track to control the timing of signals throughout the computer. A crucial feature of the circuits devised by Charles Owen was that within each digit-period, typically 3 microsec long, the signals were strobed or gated by a shorter, sharply defined signal derived from the drum. This permitted substantial tolerance against distortion of the waveforms associated with the functioning of the machine.

The drum also provided a backing-store of a really useful capacity.

The Elliott Laboratory was well equipped to provide this drum, with a highly competent mechanical department, accustomed to precision work. The importance of the drum is evidenced by the fact that it still exists as an exhibit in the Science Museum, London.

As the work progressed Bill Elliott conceived the ambition to display the machine at the Physical Society Exhibition, held annually in London. During W.W.2 and subsequently, this was the paramount occasion for displaying novel technological developments. This spurred on his obsessional manner of working, which he communicated to the tiny group of men assisting him.

In the event, he achieved that ambition. It created a real stir, for at that date very few people had even seen a digital computer, and it was the first time a practical digital computer had been transported and got into working condition for such an occasion.

It will be recounted in the next chapter that at about that time John Coales found it necessary to resign as the Director of the Elliott Laboratory, and Bill Elliott (and myself) chose to do the same. Both Coales and Elliott moved to Cambridge University. The NRDC, who owned the 401 computer, decided that it should go with them. Accordingly, it was transferred to Cambridge. Bernard Swann’s History of the Ferranti Computer Department* records that Strachey did a considerable amount of redesign work on it there, employed by the NRDC. It is apparent that Lord Halsbury combined his insight that computers would be important for British industry with an appreciation that Christopher Strachey and Bill Elliott as a pair were of exceptional value in achieving that objective. It must be assumed that Strachey was carrying out programming studies to investigate features that would be desirable in a computer, and discussing with Elliott the practical implementation of them, using the logical design facilities offered by the packages carrying the circuits devised by Charles Owen.

In support of this, Hugh Devonald who had been deeply involved in the design and construction of the 401, was engaged by NRDC to work for two or three days a week to complete the development of it and to make sure it was bought up to a high degree of reliability.

At that time it was not appreciated that the hardware of a computer had to be complemented by a substantial amount of software. What is most probable is that when the machine was at Cambridge Christopher Strachey assisted by Donald Gillies did a significant amount of software experimentation and development on the 401.

After some time at Cambridge, the NRDC decided to donate it to the Rothampstead Agricultural Research Station. For a substantial number of years it did much useful work there, especially on the statistical analyses of experimental results.

__________________________

*The similarity of name between the company an the individual is merely a coincidence.

*A fuller account of Nicholas and its creation is given by Ed Hersom in Computer Resurrection issue 27, Spring 2002.

*Written some years after ICT had taken over the commercial business of the Department, but never published.

The Migration from Elliotts to Ferranti

This chapter recounts and reflects the intense personal loyalties that lay at the heart of the vitality of the early computer scene.

In addition to what has already been told about the Elliott Research Laboratory some more needs to be written. Having been set up in 1946, by the early 1950s it had grown to over 450 persons. One of the qualities of its founder John Coales was to establish a direct and personal relationship with the more senior members of staff, who regarded him with great respect. While in practice almost independent, at least at first, it was beneficial to be attached to the long-established and highly regarded company Elliott Brothers (London) Ltd.

During 1947 Leon Bagrit became managing director of the whole company. He was an outstanding entrepreneur, with a conviction that extensive automation including computers would be introduced into industry. He also had much financial skill. For all technical decisions he relied implicitly on his colleague Dr Ross.*

During 1952 there were signs that the difference