Invisible Numbers – Ted Newman and the Pilot ACE
By Oliver W Duke-Williams, on 13 June 2017
I gave a talk on ‘Coding Early Computers’ on June 10th as part of Invisible Numbers, a group show taking place as part of the E17 Art Trail; the show continues until June 18th.
The E17 Art Trail is a biennial event in Walthamstow, London; the 2017 edition is the largest so far, with over 7000 contributors, and is centred on the theme of STEAM – Science, Technology, Engineering, Art and Maths, and exploring artistic themes within STEM areas.
One part of the show is E A Newman & Pilot ACE: Turing’s Legacy; put together by historian Kirstin Sibley and illustrator Andrew Baker, it brings into the spotlight Edward (‘Ted’) Newman, who worked alongside Alan Turing and others in the development of the Automatic Computing Engine (ACE). Newman – who graduated from UCL in 1938 with a BSc in Physics – was born and grew up in Walthamstow, but is not commonly recognised (either locally or more widely) for his pioneering work in computing. This lack of recognition thus fits well with the show’s ‘Invisible’ signifier.
Kirstin and Andrew asked me to join them to describe and explain the significance of the Pilot ACE computer for a general audience, and also to provide technical advice for a series of infographics which described the ACE and Newman’s contribution. This was a good opportunity to showcase ideas and research by academics to a public audience; the talks were standing-room only, and the gallery itself recorded over 700 visitors in the show’s opening weekend.
The Pilot ACE, now on permanent display at the Science Museum, was one of the world’s first stored-program digital computers. Based on Turing’s 1946 proposed design for a computer, the ACE was arguably the world’s first ‘complete’ computer, and was the first on which processing capability was commercially exploited.
The ACE used vacuum tubes as processing elements; arranged into circuits such tubes could carry out logical functions, and fulfill the same role as transistors on a microchip in modern computers; these can be seen in the accompanying photos. The processing board was mounted on wheeled legs; as well as being one of the first computers it was also perhaps the first portable computer.
Whilst the ACE (and similar machines such as the Manchester ‘Baby’) were not the first electronic computers, earlier machines such as the Colossus and ENIAC required physical reconfiguration (using panel switches and jack connections) to adjust their programs. In contrast, a stored-program machine such as the ACE could read a program into memory from punched cards, and then run that program. To run a new program, it was simply necessary to read from a new set of pre-prepared punched cards.
The ACE (and its commercially sold derivative, the DEUCE) had a limited set of program instructions – generally speaking, the main command was simply to read a value from one part of the machine (this might be from memory, from an external source such as a punched card, from control panel switches etc) and write it to another part of the machine (again, including memory, punched cards, a set of display lamps on the operating console corresponding to binary digits, etc). Different parts of the machine had specific tasks: for example, one location in the machine, to which a value could be written, would add the received value to whatever value was currently being held and thus perform addition. Repeated addition steps allowed for multiplication to be performed. Through use of this and other parts of the machine that carried out different functions, programs to carry out a desired task could be assembled in many small steps.
Early programs run on the ACE included crystallography applications produced for Nobel laureate Dorothy Hodgkin by her research student John Rollett, and analysis of metal fatigue as a contributor to crashes of the de Havilland Comet jet airliner.
Amongst the hardware elements of the ACE I described were mercury delay lines, used as a form of memory. Like many other pieces of hardware in the history of computing, these had a brief heyday, but are now little known by the general public. Pulses of sound were generated, which then travelled down a tube of mercury (kept at a constant temperature), and could be read at other end of the tube. A pulse would be read as the binary digit ‘1’, and a lack of pulse read as a binary ‘0’. With highly engineered timing constraints, this series of pulses could be read as a binary number, with the knowledge that a given set of pulses would arrive at the end of the tube at a precisely predictable time: the delay line could thus be viewed from a programming point of view as a series of memory locations, to which values could be written and then retrieved. Pulses arriving at the end of the tube would be transmitted electrically back to the start of the tube and fed in again; thus the same binary digits would keep circulating within the tube until that memory location was re-written (or, as with other early and experimental technologies, until it stopped working properly).
Modern computers use random access memory (RAM), in which any memory location can be read (or written to) with the same time delay. In contrast, the mercury line technology enforced sequential access: if you wanted to read a particular memory location, you would need to wait for those binary digits to arrive at the end of the tube. As the ACE was developed, rotating drum memory was added. This offered larger volumes of storage, but with slower access times.
In order to make the most efficient use of the machine, programmers adopted a technique known as optimum programming. Program instructions and required data (both held in exactly the same way – as binary numbers) would, when first loaded into the machine, be written to memory locations in a very particular order: they were arranged such that whenever it was necessary to read the next instruction, or the next item of data , then that number was immediately ready to be read – in a location on the rotating drum that was just about to travel past the reading apparatus.
Amongst the infographics created by Andrew Baker for the show is one drawing comparisons between the ACE (and DEUCE) and a current iMac desktop machine, showing the extraordinary growth in speed, memory, storage and complexity of computers. As part of this, I posed a hypothetical question: suppose that you wanted to create a machine with a similar amount of memory to the iMac’s RAM, but using ACE technology (that is, with mercury lines for memory) – how much mercury would you need? The short answer (other, of course, than ‘that would be a stupid thing to do’) is: a lot of mercury.
The longer answer – in the form of an estimate of the amount of mercury required – is contained within a newspaper created for the show by Kirstin and Andrew, along with the full story of Ted Newman and the Pilot ACE.
- Newspaper download – PDF, 3.9MB
Invisible Numbers
Invisible Numbers continues until June 18th 2017 at Winns Gallery, Aveling Centre/Lloyd Park (behind William Morris Gallery), London E17 5EQ.
Gallery opening hours (during exhibition only):
Friday 9 June – Saturday 17 June, 10am – 6pm;
Sunday 18 June 10am – 4pm
Kirstin Sibley
Kirstin Sibley is a freelance project manager and local historian, with interests in the history of the E17 area of London (Walthamstow), early computing, women’s social history, the Warner Estate and 1960s fashion and photography.
Andrew Baker
Andrew Baker is a pioneer of commercial digital illustration and has worked extensively in editorial and design settings, with clients ranging from The Guardian to Nature. His work has won several awards including Gold for Editorial Illustration from the Association of Illustrators. Andrew also lectures at Middlesex University, where his research into the relationship between illustration and the written word encompasses nonsense poetry, comics and now early computer coding.