7. Analogue–Digital Decisions in British Aeronautical Research

Alongside the development of these modelling technologies evolved the mathematical theory. Beginning with the work of George Cayley, applied mathematicians began to develop models of fluid flow, initiating the development of the discipline of fluid dynamics. However, the mathematical theory created equations that were difficult to solve, leading to the development of model environments whose behaviour was analogous to that of the equations (an indirect analogue).

The aerospace industry was a key agent in bringing computing out of the laboratory and into the wider world of industry. Ceruzzi explains that the aerospace industry was ‘accustomed to complex machines that needed long break-in periods’ and was therefore well suited to developing the technology. See Ceruzzi (1989) pp. 13–14. Ceruzzi is referring here principally to the digital computer, but aeronautical applications also drove the development of analogue computing.

Small (2001) p. 57.

Founded in 1909, the ARC (then called the Advisory Committee for Aeronautics) had been set up to provide the government with advice surrounding aeronautical research and was instrumental in establishing an aerodynamics department at NPL. Prior to its inception, aeronautical research was fairly ad-hoc with no overarching research direction. Ward (1962) described the establishment of the Advisory Committee for Aeronautics as marking ‘the beginning of an intensive experimental and theoretical approach to aerodynamics’ (p. 3). Until it was disbanded in the early 1970s, the council provided an interface between the closed world of government establishments and the work undertaken at universities. The ARC was structured by special interest sub-committees, on which sat security cleared academics, civil servants and government researchers. Historian Andrew Nahum described this set-up as ‘an ingenious mechanism… for peer review of secret work’ (Nahum 2002, p. 55). Of the various sub-committees within the organisational structure, two were related to important uses of computers. These were the Fluid Motion sub-committee and the Oscillation sub-committee.

This was a consequence of the centralisation of guided weapons research at RAE in 1945. See Small (2001) pp. 180–181.

Symbolising the climax of electro-mechanical analogue computing in Britain, this machine employed a mixture of electronics and hydraulic servo-mechanisms to accurately model the complexities of flight. For contemporary descriptions of this machine, see Anon. (1954), Gait (1955a, 1955b). Tridac was also the last computer of its kind to be developed in-house: during the following decades, institutions like RAE began to buy in commercially manufactured indirect analogues. Installed at various aircraft manufacturers, commercial analogue computers were used for modelling guided weapons. For instance, a large Emiac II analogue computer was installed in the factory of Armstrong Whitworth in Coventry to support their work on the Seaslug surface to air missile (EMI 1962b).

Finite element analysis divides a large system into a grid and models the interactions between adjacent cells. This is commonly used for simulating structural strain or heat flow. Like direct analogues, this approach exploits the physical structure of a problem.

Fifer (1961) pp. 770–771 (vol. 3).

Isenberg (1976) p. 514. See also Isenberg (1975a, 1975b, 1992).

The NACA was the forerunner organisation of NASA, the American space agency.

See Batchelor (1996) p. 118, n1.

Fifer (1961) p. 772.

Busemann (1960) p. 194.

See Taylor and Griffith (1917a, 1917b, 1918). Griffith went on to research brittle fracture and later designed turbojet engines at Rolls-Royce. He was the pioneer behind the ‘flying bedstead’, a prototype of vertical take-off and landing (DNB 2004).

Rubbra (1964) p. 118.

On the British side, the aerodynamics committee of the ARC were discussing the use of soap films for mapping air flow (Southwell 1922). In US aeronautics the NACA were also using them for structural investigations (Trayer and March 1930).

From a technical note by G.I. Taylor, appended to Southwell (1922) pp. 1–2.

One contemporary article estimates that during the 1940s the aircraft models used in wind tunnel investigations were costing as much as £100,000 (Bollay 1947, p. 106).

See Sect. 2.4.1.1, pp. 40–41.

Mounier-Kuhn (1989) pp. 257–258.

This lecture was an attempt to re-establish scientific links between France and Britain ‘helping…’, to quote the then Superintendent of aeronautical research at the National Physical Laboratory (NPL), ‘…to forward the understanding between England and France’. The meeting had been organised by a Lieut. Col. J. Valensi, himself a Frenchman and expert on analogue computing who had escaped from France in 1942. Fleeing to England, Valensi had shared his scientific expertise with the allied forces, ‘joining the common struggle by working at the NPL.’ After the liberation of France, he continued to work in England and became the ‘Liaison Officer for Aeronautical Research’ facilitating scientific dialogue between the two countries. Valensi would later return to an academic post in France. See introductory comments in Malavard (1947) p. 247.

Hargest (1952), Kuchemann and Redshaw (1954). While the electrolytic methods had existed on a small scale during the war based on those of Relf and Taylor, reports published after 1950 reference both the English and French work. This indicates that so-called ‘scientific mission’ to share research with France was successful.

Hartshorn (1948) p. 1.

Producing the accurate non-conductive model required special techniques. The approach taken at Saab was to create a mould using cross-sections.

Stenström (1949) p. 21.

Twigge (1993, 2002), Edgerton (1991, 2006), Nahum (2002).

Brig. Hinds as quoted in Minutes of the 1st meeting of the ARC computation panel. See ARC (1952–1953), 21st November 1952, p. 2.

Both Hollingdale and Goodwin were mathematicians by training and had been contemporaries at Cambridge University. After graduating from the mathematical tripos in 1932, Hollingdale pursued postgraduate study at Imperial College and submitted a Ph.D. thesis entitled ‘Stability and configuration of the wake behind a body moving through a fluid’ in 1936. On leaving Imperial, he joined the aerodynamics research team at Farnborough where he remained throughout the wartime period. Goodwin remained at Cambridge and obtained a Ph.D. on: ‘The quantum theory of surface phenomena’ before joining the NPL. See Anon. (1932), Hollingdale and Toothill (1970), preface, Theses (2007).

Goodwin as quoted in Minutes of the 1st meeting of the ARC computation panel. See ARC (1952–1953), Meeting 1, 21st November 1952, p. 2.

Goodwin and Hollingdale (1952).

Bollay (1947) pp. 106–107. Note the use of ‘computer’ to refer to a human computer.

The panel was established in November 1952, was re-established as the computation sub-committee in February 1954, and was disbanded in late 1958. By 1958, the network of computing expertise was sufficiently established to warrant a more ad-hoc approach to discussing computing issues. See ARC (1954–1958), Meeting 21, 17th October 1958.

Wilkes and Williams had led pioneering computer projects at the universities of Cambridge and Manchester respectively.

Hinds had served in the British Army during the Second World War, and received an OBE in the 1946 New Year’s Honours List. After the war, he began working in Whitehall. In 1957 he left his post as Director of Weapons Research to take up an appointment as Electronics Advisor for the British Transport Commission. See Anon. (1946, 1957). Wilkes oversaw much of the development of the Cambridge Edsac but had also previously used a differential analyser as part of his Ph.D.

While Hinds and Wilkes were keen on digital, others in the group had a more open-minded view to analogue–digital issues and while there was an analogue–digital debate, there were no hostilities between the two camps.

Although minuted, the statement is unattributed. ARC (1952–1953), Meeting 1, 21st November 1952, p. 2.

Kuchemann and Redshaw (1954).

ARC (1955).

Fairthorne as quoted in Malavard (1947).

Hollingdale and Diprose (1953) p. 1.

The first documented case of flutter was experienced by a bomber aircraft designed by Handley-Page during 1916. During the early half of the twentieth century, the common approach to designing flutter-free structures was through experimental test-flights. However by the late 1930s, the threat of serious accidents such as the crash of a Junkers aircraft during such an experiment resulted in engineers turning to ground-based analysis. This involved both theoretical consideration of the aerodynamics, which required the solution of complicated equations, and ground-based experiments involving wind tunnels. See Rodden (1992) pp. 223–224. Historians of computing have identified flutter as a key mathematical application of the early digital machines. See Aris (2000) p. 10, Neukom (2005) p. 17.

Ceruzzi (1989) p. 33.

Templeton (1955) p. 1.

Templeton (1955) p. 1.

ARC (1952–1953), Meeting 2, 16th December 1952, p. 3.

Firstly determining the ‘normal modes of oscillation’, and then calculating the structural and aerodynamic coefficients.

Templeton (1955) p. 6.

Alongside the modelling of the mathematical equations, RAE also developed smaller analogue computers to complete stages 1 and 2. Stage 1 was served by NOMAD—a ‘Normal Mode Analogue Computer’, and stage 2 by a combination of INCA (Integral Calculator) and MAYA (Matrix Multiplier). Thus even in the mid-1950s, RAE was developing an entire end-to-end special purpose analogue computing process (Templeton 1955, pp. 4–8).

ARC (1952–1953), Meeting 3, 23rd January 1953, p. 1.

ARC (1952–1953), Meeting 2, 16th December 1952, p. 2.

ARC (1952–1953), Meeting 2, 16th December 1952, p. 3.

See Hollingdale and Diprose (1953). Hollingdale had a broad perspective when it came to computing. With his mathematical background, he saw the merits of the digital computer. However, through working at RAE, he also understood the importance of analogue. In 1965 he co-authored a popular introduction to computers entitled Electronic Computers. In this text (and in the second edition published in 1970) the authors described both analogue and digital and emphasised their complementary roles. For instance, the authors note Hartree’s involvement with the differential analyser and the Eniac as ‘an excellent early example in avoiding narrow specialisation on either analogue or digital computers’ (Hollingdale and Toothill 1970, p. 80).

This would be relieved by the introduction of time-sharing systems, minicomputers and later personal workstations. Perhaps this explains why engineers were not fully dis-enrolled from analogue culture until the 1970s.

ARC (1952–1953), Meeting 3, 23rd January 1953, p. 3. Diprose was, at this time, employed by the RAE.

Goodwin and Hollingdale (1952).