8: The first electronic Digital Computers

So far, we have been looking at the technical developments that have contributed to the possibility of computing. Many of these developments, particularly those of analogue computing (ie, the representation of real-time processes) and the development of television, have also contributed to the possibility of video art and video synthesis, electronic music and sound synthesis and various other aspects of art and technology such as interactive performance and installation work. Computing and the means of displaying its results are the prime contributors to the making of digital art in its forms of prints, animation and computer music. More recently, computing, video and the telegraph have been combined into the form we think of as internet art. In this section we will continue to cover the development of computing, and in the following chapter, its development in Australia.

Although we do not yet come to the coverage of computer art or any other of art’s electronic forms we must set up the technological basis for these electronic arts so that we can see how the technologies of the times constrained the making of art. In a sense it is the display technologies that govern what kinds of art are possible and how it will be seen, though, of course, it is impossible to get an image to a screen without some sort of device behind it producing the data that will be interpreted by that screen into an image.

In this initial period, it also important to recognise that most of the imaging was done by scientists, for scientific purposes such as data reduction and visualisation of simulations or for ascertaining that their theoretical work matched the results of their experiments. It is only as some of the people who were involved in computing more at the level of computing per se, ie, as programmers and mathematicians, that computer images in their own right began to be thought of as art.

During the Second World War, much of the scientific work both here in Australia and in the US and the UK was involved with three kinds of problems.

1. The decryption of enemy communications.
2. The production of gun firing tables which could help shoot down these incoming aircraft, and
3. Radar and the capacity to see and track incoming aircraft by electronic means.

The first of these, as such, was hidden from the world for a considerable period after the war but the people who worked on these techniques and devices came out of the secret laboratories to work in government and academic research facilities, bringing computing from its military background to the research and commercial foreground that it holds today.

During the war, Alan Turing, in the UK, had been grappling with the problem of decoding encrypted German military and high command communications. At the secret facility at Bletchley Park, north of London in the district now known as Milton Keynes, Turing had worked out methods of decryption that required various kinds of sorting and pattern recognition techniques used in the German High Command’s machine known as the Enigma and built a machine (the Bombe) to do that [Fig.1].

Following on from the kinds of concepts and statistical procedures that Turing developed a British engineer named Tommy Flowers, also working at Bletchley Park, took these ideas and, over the period from 1941 to 1943, turned them into a parallel-processing pattern-matching machine called the Colossus [Fig.2]. This machine was used to decode messages that had been enciphered on the Lorentz machine, a more sophisticated device that replace the Enigma.

It proved so successful at providing data that helped the decryption process that eventually the Telecommunications Research Establishment (TRE) produced about 10 of these large fast parallel computers during the war period. However, Colossus was not a stored-program computer, it was, more or less, an implementation of a single pattern-matching algorithm (though there is a suggestion that Turing may have spent some time experimenting with rewiring it to change its algorithm). Although the production of these machines was incredibly significant, British secrecy about the value of the decryption techniques that had been developed and the fact that the Enigma and Lorentz machines were still in foreign government and commercial use meant that it remained unknown until many years after the end of the war. Word of Colossus only started to leak out over the 1970s and in the 1980s some of the original engineers began to produce descriptions of Colossus from circuit diagrams and other data they had held onto.¹

Regarding the second type of problem taking up much of the available scientific workforce: due to the immense number of calculations required in producing artillery firing tables, the US military wanted to automate the process. These tables of numbers assisted artillery crews in the field to sight and fire guns so that they could lob a shell into the vicinity of an aircraft with such timing as to be where the aircraft was going to be by the time the shell got up there. This involved a degree of prediction that Norbert Wiener had analysed and developed equations to solve via his work on feedback. The equations could then be calculated out for large numbers of cases and printed into tables for distribution to artillery units wherever they were. The gunners could then simply look up the values for the data of any particular range and altitude, apply the numbers and fire their guns with some hope of hitting the aircraft. Thus were computers born in the US.

The earliest workable digital machines were Konrad Zuse’s Z3, a computer built with relays and developed in Germany c.1940,² and an electromechanical device built under the direction of Howard Aiken at Harvard University. This was the Automatic Sequence Controlled Calculator machine, also known as the Harvard Mark I, which was completed in 1944. Subsequently there were a Mark II and a Mark III. While digital Zuse’s machine was programmable, with input being read from paper tape, the Mark I was essentially a large relay-controlled calculator which had to be set up via patching between arithmetic modules for any new calculation, with constants stored “on a set of ten-position hand switches” and data fed in from punched paper tapes.³ During the war, the US Army was convinced by US Navy Lieutenant and mathematician Herman Goldstine to build an electronic calculator which could considerably speed up the calculation of firing tables. The Moore School of Engineering in Philadelphia was given the contract, the project was led by John Mauchly and the chief engineer was Presper Eckert. By 1946 they had built and commissioned the ENIAC computer, which was essentially a patch programmable electronic calculating machine. It had registers for the storage of currently used numbers and results but the programs for what to do with those numbers could not be stored in it, having to be set up by making a number of wired connections between different modules of the machine (essentially “patching”). The results were printed out to punched cards.⁴

The third type of problem was that of Radar. In the mid-1930s the British government recognised the need for an early warning system to guard against the possibility of an aerial attack from Europe, presumably the Germans, and began serious research into the development of Radar. By the beginning of WWII Radar was more or less operational and it, in its many forms, was able to contribute to the defence of Britain against the bombing raids of the German air force during that dark period known as the Battle of Britain.⁵

The Radar work produced a number of electrical engineers and mathematicians who were familiar with the use of what were known as “pulse techniques” for the analysis of radar signals, so that they could be displayed cleanly on the cathode ray tubes (CRT) that were adopted for this purpose. One of the primary problems here was that although very high frequency radar signals could be sent out and their reflections from flying metal objects received, they were also reflected off the surrounding topography, which would produce a great deal of “clutter” on the display screen. Since the radar signal and its echo have the form of pulses, by developing electronic techniques which allowed the storage of a radar echo pulse for a short period so that it could be compared with the next sweep, the clutter (which was mostly echoes from the stationery topography) could be subtracted out and the fast moving objects could be presented clearly.⁵ The scientists who carried out this work invented a storage device known as the acoustic delay line. This delay line consisted in a tube of monel-metal (later stainless steel) filled with mercury, having a piezo-electric transducer (a thinly-cut quartz crystal) at one end that would impress a pulse of sound (a pressure wave) into the column of mercury and another piezo-electric transducer at the opposite end which would be bent by the pressure wave and thus generate a current. By regularly pulsing the column of mercury the tube could be made to store upto a thousand pulses for a short period. By reading the pulses from the receiving end of the tube the pulses could be electronically regenerated and returned to the column of mercury, with the right timing, so that the pulses were continually regenerated and the acoustic (or mercury) delay line memory was born. Now, computers need to be able to store data and the acoustic delay line provided a form of memory so that they could do just that.⁶

In many ways it is the British research work that is of prime significance for us here. After the war, as a result of the wartime research, two major developments were made possible in computing.

1. the first stored-program computer, and

2. a new means of doing the storage.

Apart from shortening the war by probably up to two years, the primary importance of the Bletchley Park achievement was that it produced a number of mathematicians and engineers who had considerable experience in computing even if they could only talk about what they might do next. Turing went to the National Physical Laboratories and developed proposals for the PilotACE machine,⁷ which subsequently became the English Electric DEUCE, one of which was used in Australia at the then University of Technology, New South Wales (now UNSW). The University of Technology, NSW, machine was known as UTECOM and was commissioned in September 1956, but we are not going to follow it up here.

Other members of the Bletchley Park group went on to join the Mathematics Department led by Max Newman at the University of Manchester. Newman proposed the development of a computing laboratory to undertake the complementary tasks of designing the mathematical and logical circuitry, and of developing a suitable memory storage unit, for a small computer to be known as the Manchester Small Scale Experimental Machine (SSEM or the “Baby”) as a preliminary to further industrial development of computing. Turing became involved in late 1948 when he was appointed as deputy director and began to write a programming manual for the extended version of the Baby, now known as the Mark 1.⁸

The SSEM would not have been possible as a stored program machine without a storage system. Fred Williams, an electrical engineer, and Tom Kilburn, a mathematician who had been moulded into an engineer by Williams, both came to Manchester from the TRE.⁹ They had been developing a method of electrostatic storage using CRTs, and, as Newman and his mathematicians were losing interest in computing, took on the work on the SSEM as a test-bed for their electrostatic storage system. [Fig.3]

Out of the Manchester project the world’s first stored-program computer was built in Britain. As mentioned above this SSEM was a prototype. In its first version it had only 7 instructions and performed its first program in June 1948.¹⁰ It was enhanced by the addition of a further 19 instructions (to 26 in all), with the necessary expanded electronics, and a magnetic drum.¹¹ Subsequently the Mark 1 version of the machine was taken on by the Ferranti company on as a product line. They then produced the Ferranti Mark I*, the Mark II, the Atlas and the Pegasus machines over the next decade.¹²

Thus there were several independent lines of computer development followed in the UK. One was the stored program computer in the form of the Mark 1 at Manchester University which was only indirectly based on the work of Turing, who had very little to do with the actual design and building of the machine. The machines built at NPL had also been based on his ideas but were designed and built without his input. The other main computer development was at Cambridge University under the direction of Maurice Wilkes.

The Manchester machine used CRT (or electrostatic) storage memory while the NPL PilotACE and the Cambridge EDSAC used acoustic delay line memory. These two approaches to memory were maintained in the early commercial machines in the UK, the Ferranti machines being based on the Manchester approach and using electrostatic storage and the English Electric machines based on the ACE continued with the acoustic delay line type of memory. The relevance of this point becomes clear when we look at the development of machines in Australia.

The Williams-Kilburn tube type electrostatic store or memory is perhaps the prime development for the computer production of graphics but it took some years for it to be realised as the bit-map that we are so familiar with these days. The graphical potential of the CRT storage device was clearly well understood by Kilburn. In his report to the TRE of December 1947¹³ and in a paper published in the Institute of Electrical Engineers (IEE) journal,¹⁴ he showed a photograph of the CRT face which had the appropriate bits switched on so that it presented a display of the characters “2048 DIGIT STORE”.

The CRT is the storage device itself, its phosphor target holding a map of the states of the bits as memory. Since the retention of the secondary electrons in the phosphor is temporary the memory needs to be continually refreshed. It is effectively dynamic RAM. By tapping-off the regeneration circuits the memory map could be displayed on another CRT and used to monitor the state of the machine. This was, for a long time, the main use of the CRT displays which accompanied many of the early machines (including each of the first three machines in Australia).

On these early machines programming was always done in the direct code of the hardware of the machine (no assemblers, compilers or high-level languages yet) and debugging a program involved the difficult process of ascertaining where the machine had stopped or whether it was stuck in a loop. The monitoring CRT could be very helpful in diagnosing the state of the machine if a bug arose, which often happened (and may well have been either a hardware failure or a software problem).

The beginnings of computer music lie in this diagnostic function as well. Many of the early machines had a speaker attached to the machine, patched into some convenient memory address location, so that a bit could be programmed to make a noise through the speaker. If the speaker then stopped you knew the program had stopped or if it went into a repeating loop of the same sequence of sound then the machine was in a loop in the program. This was also exploited by some early users to make sounds that were simply entertaining. The most interesting sounds though, were often based on the behaviour of the program while it was engaged in processing the data of the task (see the discussion of the Leapfrog program and Jenssen’s experiments with data profiles).

It was also recognised that the CRT display could do fun things like make announcements in text characters running across it or play games and it was also possible to use the CRT display monitoring memory to produce actual useful displays of data results.

———————————————————-

Return to Contents

FOOTNOTES

1. See Hodges, Andrew (1983) Alan Turing: The Enigma, Hutchinson, 1983. [See especially pp.266-8 and 277-8 of the Vintage 1992 paperback edition.] See also Andrew Hodges’ Turing website: http://www.turing.org.uk/turing/. For information on the rebuilding of the Colossus machine see Tony Sale’s website at http://www.codesandciphers.org.uk/lorenz/colossus.htm, and Copeland, Jack, Colossus: The First Large Scale Electronic computer. http://www.colossus-computer.com/colossus1.html> ↩︎
2. Zuse’s first machine the Z1 was built between 1936 and 1938 but suffered from imprecisely machined parts and was unreliable. His first workable machine was the Z3, a relay machine completed in 1941. http://www.konrad-zuse.de/. See also Rojas, R. “Konrad Zuse’s legacy: the architecture of the Z1 and Z3”, Annals of the History of Computing, IEEE, Vol.19, No.2, Apr-Jun 1997, pp.5 – 16. ↩︎
3. Hartree, D.R. (1950) Calculating Instruments and Machines, Cambridge University Press, Cambridge, UK, p.75. See also Bloch, Richard M., “Mark I Calculator”, and Campbell, Robert V.D., “Mark II Calculator” in Aiken, Howard A. (ed.) Proceedings of a Symposium on Large-Scale Digital Computing Machinery, held at the Computation Laboratory, Harvard University, 7-10 January 1947. Harvard University Press, Cambridge, Mass. (1948). One of the original “programmers” of these machines was Grace Hopper, who was among the first women to be involved in computing since Ada Lovelace. Hopper is reputed to have found the first computer “bug”, in this case a real one – a moth. see http://www.hopper.navy.mil/grace/grace.htm. ↩︎
4. Hartree, 1950, op cit, p.81ff. See also Tabor, Lewis, P. “Brief Description and Operating Characteristics of the ENIAC” in Aiken, Howard A. (ed.) Proceedings of a Symposium on Large-Scale Digital Computing Machinery, held at the Computation Laboratory, Harvard University, 7-10 January 1947. Harvard University Press, Cambridge, Mass. (1948) ↩︎
5. For the story of Radar and its use during WWII there are numerous books. Two worth mentioning are:
   Rowe, A.P. (1948) One Story of Radar, Cambridge: Cambridge University Press. Rowe was the civilian in charge of Radar development for the TRE in Britain.
   Watson-Watt,Robert (1959) The Pulse Of Radar. The Autobiography Of Sir Robert Watson-Watt, New York: Dial Press. Watson -Watt was one of the leading scientists involved in research on the reflection of microwaves, originally from the Ionosphere in shortwave radio and then from aircraft in early warning radar. ↩︎
6. Bennett, J.M., “Autobiographical Snippets”, p.52, in Bennett, J.M., Broomham, R., Murton, P.M., Pearcey, T. and Rutledge, R.W. (1994) Computing in Australia: The Development of a Profession, Hale & Iremonger in association with the Australian Computer Society Inc., Sydney. ↩︎
7. Carpenter and Doran, (eds), (1986), “A. M. Turing’s ACE Report of 1946 and Other Papers”, Charles Babbage Institute reprint series for the History of Computing; Cambridge, Mass.: MIT Press and Tomash Publishers.. ↩︎
8. This was recognised by a number of early computer workers including Turing with the NPL machine and Wilkes and Hartree with the Cambridge EDSAC. Turing in his report “Proposal for Development in the Mathematics Division of an Automatic Computing Engine” (ACE), prepared for the National Physical Laboratory (NPL) in 1946 describes the operation and use of the delay line type of memory (Carpenter and Doron, pp.22-24). He attributes its adoption in computing machines to Presper Eckert (Carpenter and Doron, p.108) who built the ENIAC at the University of Pennsylvania over 1944-45. However Eckert did not use the delay line as a storage element in the ENIAC, using instead hand switches and paper tape, with flip-flop style registers as the immediate storage during calculation. [Page numbers refer to Carpenter, B.E. and Doron, R.W. (eds), A.M. Turing’s ACE Report of 1946 and Other Papers, MIT Pres Cambridge Mass, for the Charles Babbage Institute, 1986.] ↩︎
9. Kilburn, T. (1947) “A Storage System for use with Binary Digital Computing Machines.” Progress report to TRE issued 1st December 1947. This document was widely circulated particularly by Hartree to the US where the Williams-Kilburn method of storage was adopted for the Whirlwind project at MIT, EDVAC, and the Illinois machines (ILLIAC) based on the Princeton IAS system developed by von Neumann, which subsequently became the SILLIAC at Sydney University.
   Williams, F.C. and Kilburn, T. (1949) “A Storage System for use with Binary-Digital Computing Machines,” Proceedings of the IEE, vol.96, part 2, no.3, pp.183-202. ↩︎
10. When he arrived in Australia Trevor Pearcey adopted the delay line as the storage method for the development of the CSIRO Mk I (later known as CSIRAC). For a useful technical description of the delay line as a memory device see Sharpless, T.Kite “Mercury Delay Lines as a Memory Unit” in Aiken, Howard A. (ed.) Proceedings of a Symposium on Large-Scale Digital Computing Machinery, held at the Computation Laboratory, Harvard University, 7-10 January 1947. Harvard University Press, Cambridge, Mass. (1948), or see Wilkes, M.V., Automatic Digital Computers, Methuen, London, 1956, chapter 5. ↩︎
11. Lavington, Simon (1998) History of Manchester Computers, British Computer Society, p.17. ↩︎
12. Deane, John, (1999) The University of Manchester’s Baby: the first modern computer, Australian Computer Museum Society, Sydney, 1999, p.15. ↩︎
13. ibid, p.5. ↩︎
14. Lavington, op cit, pp.20ff. ↩︎