Williams tube

Freddie Williams and Tom Kilburn found in 1946 that the electron beam of a cathode ray tube leaves a charged spot on the phosphor, and that the charge can be sensed through a metal plate stuck to the outside of the glass. A spot is a one, no spot a zero, and since the beam can be aimed anywhere on the screen any bit is reachable immediately — unlike a delay line, where you waited for the number to come round again.

The price was leakage: the charge faded within a fraction of a second, so the contents had to be read and rewritten continuously. That is the same refresh principle dynamic RAM runs on today. One tube held roughly 1,024 to 2,048 bits. Wiring a second tube in parallel gave operators something no later memory technology has offered: a live picture of memory contents.

Manchester Baby was built in 1948 for the sole purpose of proving the tube worked as storage. The Ferranti Mark 1, IBM 701, ILLIAC and the Soviet Strela followed. The weakness was sensitivity — tubes interfered with one another, and electrical noise in the room could flip bits. Magnetic core memory made the whole approach obsolete by the mid-1950s.