The earliest computing machines in wide use were not digital but analog. In analog representation, properties of the representational medium ape (or reflect or model) properties of the represented state-of-affairs. (In obvious contrast, the strings of binary digits employed in digital representation do not represent by means of possessing some physical property -? such as length -? whose magnitude varies in proportion to the magnitude of the property that is being represented. Analog representations form a diverse class. Some examples: the longer a line on a road AP, the longer the road that the line represents; the greater the number of clear plastic squares in an architect’s model, the greater the number of windows in the building represented; the higher the pitch of an acoustic depth meter, the shallower the water. In analog computers, numerical quantities are represented by, for example, the angle of rotation of a shaft or a difference in electrical potential.
Thus the output voltage of the machine at a time might represent the momentary speed of the object being modeled. As the case of the architect’s model makes plain, analog presentation may be discrete in nature (there is no such thing as a fractional number of windows). Among computer scientists, the term ‘analog is sometimes used narrowly, to indicate representation of one continuously-valued quantity by another (e. G. , speed by voltage).
As Brian Cantle Smith has remarked: ‘Analog’ should be a predicate on a representation whose structure corresponds to that of which it represents That continuous representations should historically have come to be called analog presumably betrays the recognition that, at the levels at which it taters to us, the world is more foundational continuous than it is discrete. (Smith , p. 271) James Thomson, brother of Lord Kelvin, invented the mechanical wheel-and-disc integrator that became the foundation of analog computation (Thomson ).
The two brothers constructed a device for computing the integral of the product of two given functions, and Kelvin described (although did not construct) general-purpose analog machines for integrating linear differential equations of any order and for solving simultaneous linear equations. Kelvin’s most successful analog computer was his tide predicting machine, which remained in use t the port of Liverpool until the sass. Mechanical analog devices based on the wheel-and-disc integrator were in use during World War I for gunnery calculations.
Following the war, the design of the integrator was considerably improved by Hannibal Ford (Ford ). Stanley Fifer reports that the first semi-automatic mechanical analog computer was built in England by the Manchester firm of Metropolitan Vickers prior to 1930 (Fifer , p. 29); however, I have so far been unable to verify this claim. In 1931, Vainer Bush, working at MIT, built the differential analyses, the first large-scale automatic general-purpose mechanical analog computer. Bush’s design was based on the wheel and disc integrator.
Soon copies of his machine were in use around the world (including, at Cambridge and Manchester Universities in England, differential analyzers built out of kit-set McCann, the once popular engineering toy). It required a skilled mechanic equipped with a lead hammer to set up Bush’s mechanical differential analyses for each new job. Subsequently, Bush and his colleagues replaced the wheel-and-disc integrators and other mechanical components by electromechanical, and finally by electronic, devices.
A differential analyses may be conceptualized as a collection of ‘black boxes’ connected together in such a way as to allow considerable feedback. Each box performs a fundamental process, for example addition, multiplication of a variable by a constant, and integration. In setting up the machine for a given task, boxes are connected together so that the desired set of fundamental processes is executed. In the case of electrical machines, this was done typically by plugging wires into sockets on a patch panel (computing machines whose function is determined in this way are referred to as ‘program-controlled’).
Since all the boxes work in parallel, an electronic differential analyses solves sets of equations very quickly. Against this has to be set the cost of massaging the problem to be solved into the form demanded by the analog machine, and of setting up the hardware to perform the desired computation. A major drawback of analog computation is the higher cost, relative to digital machines, of an increase in precision. During the sass and sass, there was considerable interest in ‘hybrid’ machines, where an analog section is controlled by and programmed via a digital section. However, such machines are now a rarity.