Historical context

The history of the development of the original first Manchester computer has been well told.3-5 Only a summary will be given here to place the replica project in context.

From the perspective of the UK, World War II caused an upsurge in electronics know-how as a result of developments in radar and communications. Much of the radar work was done at the deliberately misleadingly named Telecommunications Research Establishment (TRE). There, F. (Freddie) C. Williams and his small team were the experts in electronic circuit design. Williams was invited to visit the Massachusetts Institute of Technology in 1945 and 1946 to edit and contribute to part of the Radiation Laboratory series of books (known popularly as “the five-foot shelf”) on radar techniques. While in the US, he viewed experiments in which researchers attempted to store radar images using CRTs, work with which he would have felt entirely at home. He also saw the ENIAC and heard about the research toward an electronic memory system for that machine. He returned to the UK after his second trip with an idea as to how to use a CRT as a calculating machine’s memory device.

By the end of 1946, Williams and his colleagues had successfully stored one digit and had filed a provisional patent for their method. At this point Williams was appointed to the Chair of Electro-technics (now Electrical Engineering) at the University of Manchester, apparently with the blessing of TRE, which wanted him to continue the CRT memory research. He was accompanied by one of his colleagues, Tom Kilburn, a mathematician-turned-engineer who was on loan from TRE. In a late-Victorian-era university laboratory originally called the Magnetism Room, they developed the storage system, with Kilburn doing much of the hands-on technical work. By late 1947, they had a working highspeed electronic memory storing 2,048 bits, and Kilburn issued a widely read report fully explaining the technique (see the sidebar, “The Principle of CRT Storage”).

Although their work in the electrical engineering department was completely independent of outside influence, Williams and Kilburn were nevertheless encouraged by the Mathematics Department’s Max Newman, who secretly knew about the Colossus code-breaker machine and was familiar with the work of Alan Turing. Newman wanted a computer in his department and could see that the Williams–Kilburn work might lead to one. Newman’s receipt of funds to establish a Royal Society Computing Machine Laboratory was not used by Williams (who was fully funded by the university and TRE), though he later appropriated the name as a tag for the Magnetism Room.

Although the CRT storage system was working in late 1947, it had not been proven to work “in the hurly-burly of computing,” as Kilburn put it in one of our conversations. He needed a stringent, dynamic tester for it and resolved on a very simple computer as the appropriate test bed. With the help of Geoff Tootill, who had also joined the team from TRE, the three men designed and built the necessary and barely sufficient equipment to make a general-purpose computer out of their storage system. (For functional structure details, see the “Functional Structure of the Baby” sidebar). They did this work from late 1947 into the first half of 1948, and by June 1948 the machine was complete but not yet working. Tootill has characterized the period as being like a wartime radar crash program: intense activity to get the job done without too much regard for aesthetic niceties. At about 11:15 a.m. on Monday, 21 June, a program written by Kilburn executed correctly, the first time that a stored program worked in an all-electronic computer.

Over the next few months, the machine was extensively enhanced both in capacity and functionality, and what was originally known as the Small-Scale Experimental Machine (SSEM), and informally as “The Baby,” gradually evolved into the Manchester Mark 1 computer. 6 Figure 1 shows a familiar photograph of the machine as it was in about April 1949. The size of the team doubled when postgraduate researcher Alec Robinson and two research students, “Dai” Edwards and “Tommy” Thomas, joined in late summer 1948. This evolution has been succinctly described by Lavington3 and by Napper.7

The Baby machine had already vanished by late 1948 in the sense that it had been heavily modified and enhanced into the prototype of the University of Manchester Mark 1 computer, and it was this enlarged machine that performed useful work through 1949 and 1950. All the equipment was scrapped at the end of 1950, ready for the delivery of the properly engineered version, the Ferranti Mark 1. Just one of the original bare steel racks has been preserved at the university, though its provenance seems to be undocumented.

Some of the early programs that ran on the SSEM in summer 1948 are described by Shelburne and Burton; in particular, there is a description of the best estimate of the first program’s code as retrospectively reconstructed by Kilburn and Tootill.8 The occasion when that program first ran correctly is regarded as a key date in computer history, and it was the defining moment for inspiring the replica building project.