The director of the Radiation Laboratory, Louis Ridenour, suggested as early as 1944 that the entire body of knowledge on radar accumulated in operations research during the war be gathered together in one large work. Britton Chance contacted his British colleague Frederic Williams and invited him to work on two of the volumes in the now famous series, which ultimately numbered 28. To this end, the British engineer visited the Radiation Laboratory in 1945 and 1946, and there learned about the experiments of McConnell and his co-workers in which they stored radar data on CRTs. The device did not achieve the robustness necessary for application in the field and was evidently abandoned.

The Americans had overlooked the decisive fact that by using the time gained by short-term storage for refreshing the data just read, memory could be extended indefinitely. Looking back, it is amazing how long it took to realise the fact that if one can read a record once, then that is entirely sufficient for storage, provided that what is read can be immediately rewritten in its original position. With the USAs nuclear attacks on Hiroshima and Nagasaki in August 1945, World War II came to an abrupt end, and thus also the intensive research on radar. In December 1946, a few months after Williams returned from the USA, at TRE Williams successfully stored a single bit stably on a CRT.
		Early CRT radar.

The Williams Tube came about from Sir F.C. William’s research into using Cathode Ray Tubes to store digital data for long periods of time. In November 1946, Sir William was able to show that a single CRT could be used to store a single binary digit. With the help of his Telecommunications Research Establishment (TRE) colleague Tom Kilburn, by November 1947, they were eventually able to find a way to store 2048 bits on a standard single 6 inch diameter CRT over a period of a few hours using variations on storing a set of bits.

		Binary information is stored on the insulating screen of a CRT as small areas of charge on an array of isolated dots and dashes. If the dot position is to store a 0, an electric signal would be applied to that position and that point on the screen would acquire a positive potential. Since this potential can decay over time, that dot position must be exposed to the electric signal continuously in order to maintain the charge (the first instance of regenerative memory). In order to store a 1, the electrical signal is lengthened so that the dot is extended to a dash. This, too, must be continually refreshed so that the data does not dissipate. It took between 300-400 ms to refresh a 32-40 bit word.
A C.R.T. display - 1024 digits.

A metal pick-up plate that is placed on the outside of the CRT screen is used to read the data, as it can sense the type of charge at any spot in the array. However, since the pick-up plate obscures the front of the tube (and since the tube also needs to be shielded from outside electrical interference), a separate display tube is needed in order to view the contents of the CRT storage tube. This second tube is updated synchronously with the refresh cycle of the Williams Tube. The Williams Tube had many practical advantages at the time as it was made from standard components, was cheap, compact and did not require temperature control or precisely controlled power supplies. However, the single most important advantage that the Williams Tube had over other existing memory technologies at the time was that the nature of the way that the Williams Tube stored and read data allowed for fast random access to short strings of bits.
		Dot pattern on the screen of a Williams memory tube.

		Fast random access of data is one of the factors that is important in the stored program principle as fast access to program data and program instructions between various places in memory is needed during execution time; this was not feasible through most of the other memory technologies at the time as they all implemented some form of sequential access. Kilburn knew that the most effective proof of the mechanism would be to use CRTs in a computer. So, with the help of G. C. (Geoff) Tootill, Kilburn designed and built a small computer incorporating the CRT storage. This computer 'the Baby' had a store size of 32 words, consisted of some 650 valves (vacuum tubes), was 16 feet long, and weighed half a ton. It worked successfully for the first time on June 21, 1948 and so became the world’s first functioning stored program electronic computer.
Freddie Williams in front of the Mark 1 console CRT screen.