Differential analyzer

Differential analyzers were an early stage in the development of computing devices; they were analog devices - generally entirely mechanical, although some of the later ones included electrical elements. The mathematical operations they were able to perform were evaluating the integration of differential equations (hence the name); differential equations (which include feedback from the prior values of the equation, and are thus often not susceptible to closed-form solutions) are very common in scientific and engineering work.

Although they represent a significant stage in the development of computing devices, they take this spot only for temporal reasons; they were not themselves part of the evolutionary path from the machines of Charles Babbage to the later Harvard Mark I, ENIAC etc. The reason is simple; differential analyzers were analog machines (while digital machines like the others listed had fundamental advantages which led to their eventual supremacy). Differential analyzers are important, though, because they were a stage in accustoming scientists and engineers to relying on machines to handle the increasing computational needs of these fields, as they developed.

One of the first devices of this type was the Harmonic Analyzer for Tides of William Thomson (Lord Kelvin), the first of which was built in 1876; all later work in this area is built on Thomson's ground-breaking work. Starting in the late 1920's, Vannevar Bush at MIT did much pioneering work on differential analyzers (his name for them), and made them popular for technical work. (Indeed, both John Vincent Atanasoff and John Mauchly knew of his work, and were inspired by it in their own.) In fact, the differential analyzer can justly be described as the first general-purpose computing device used in scientific and engineering work.

They did serve as inspiration for similar electronic machines, analog computers, which enjoyed a bit of a vogue in the 1950's. Differential analyzers and analog computers all suffered from the same fundamental problem as the initial stage of the ENIAC; applying the machine to a new problem required re-configuring it.

One intriguing sidelight is the use of 'Meccano' (reusable mechanical engineering components, originally created as toys for older boys) to produce low-cost differential analyzers for serious work. These were actually accurate enough for the solution of many scientific problems, as well as being cheap to build. The first was created at Manchester University by Douglas Hartree and Arthur Porter in 1934. Several (one estimate is that as many as 15 were created in various locations) were built across the world up to the early 1950s, when they were replaced, first by analog computers, and then by faster digital computers.

Further reading

• J. Crank, The Differential Analyser, Longmans Green, London, 1947 - the only known book about mechanical differential analyzers
• S. Fifer, Analogue Computations, McGraw-Hill, New York, 1961 - mainly covers electrical analog computers, but has an excellent section on the differential analyzer
• D. R. Hartree, Calculating Instruments and Machines, University of Illinois, Urbana, 1953 (re-printed, MIT Press/Tomash, Cambridge/Los Angeles, 1984) - covers differential analyzers and other analog computing devices in some depth in the first section

External links

• The Differential Analyser Explained - covers the 'Meccano' version, but touches on the basic concepts
• Brian Randell, From Analytical Engine to Electronic Digital Computer: The Contributions of Ludgate, Torres, and Bush, Annals of the History of Computing, Vol. 4, No. 4, October 1982, pp. 327-341 - the work of Bush, the third subject, on differential analyzersis not really covered here, though
• Allan G. Bromley, Chapter Five: Analog Computing Devices, in William Aspray (editor), Computing Before Computers, Iowa State University Press, Ames, 1990 - covers the entire history of mechanical analog computing devices in some detail