Manchester Transistor Computer
The Manchester transistor computer was a pioneering digital computer built at the University of Manchester, with a prototype running in 1953, that is one of the first stored-program computers to use transistors for its logic. Developed under Tom Kilburn's group, with Dick Tizard prominent in the construction, it showed in Britain that the new semiconductor devices could replace tubes in a working computer, and it led to a commercial descendant, the Metropolitan-Vickers 950.
The project came from the Manchester tradition of computer building and a wish to make machines smaller and more reliable. The university had built the Manchester Baby in 1948 and the Manchester Mark 1 afterwards, the latter commercialised by Ferranti, and Kilburn, a leading designer of that line, saw that valve-based computers would always be bulky, hot and prone to failure. Transistors were becoming available from the late 1940s, and the group set out to test them in a machine. The work is best seen as an exploration, with the aim of learning how to design with unreliable early devices rather than building a production computer.
The 1953 prototype used point-contact transistors, with a modest component count by later standards. It is generally described as using about 92 point-contact transistors and some 550 diodes, figures that vary slightly between sources, and some parts of it, such as the circuits generating the clock or driving the magnetic drum, still relied on valves. The first running date is usually given as November 1953. Point-contact transistors of the time were difficult to produce consistently and had limited life and spread in characteristics, so the logic circuits were designed to tolerate wide variations, a theme common to all early transistor machines, including the Bell Labs TRADIC which appeared in the same period.
The Metrovick 950 carried the experiment into a product. The engineering firm Metropolitan-Vickers, which had a close relationship with the university, built a commercial machine on Manchester's design from the mid-1950s, using junction transistors, and only a small number were made. It was a limited commercial success, since the Ferranti line of valve machines was established and transistor machines had yet to prove their price advantage, but it gave the transistor computer a real-world test in customer sites. The Harwell CADET of 1955, built at the Atomic Energy Research Establishment, is another British machine of the period and is sometimes credited as the first to use transistors without any valves, which shows how the "first" depends on definition.
Care is needed with claims about "firsts" and with numbers. Contemporary descriptions and later histories give different counts of transistors and diodes and different dates for the working versions, depending on whether they describe the prototype, the later full-size machine or a particular configuration. The safe statements are that Manchester demonstrated a working transistor-based stored-program computer in 1953, that Bell Labs' TRADIC and the Harwell CADET followed shortly, and that none of these machines was a direct commercial ancestor of the IBM and Control Data systems that dominated the second generation. A related misconception is that the prototype was purely transistorised, whereas the early machines mixed technologies.
Its importance is as an early demonstration and as part of Britain's early role in computing. Britain had produced the first stored-program computers and, in the early 1950s, led in several respects, and the Manchester transistor machine is part of that story, though the larger market and manufacturing scale of American firms later dominated. The idea that logic could be built from small, cool, solid-state switches is the foundation of all modern computing, and the 1953 prototype is a small, early proof, written in a few hundred hand-made components, of what is now done with billions on a chip.
Read the Manchester transistor computer as Britain's early proof that a working computer could be built from transistors: a 1953 prototype of fewer than a hundred transistors whose real legacy is the lesson, drawn before the technology was mature, that the vacuum tube's days in computing were numbered.