A piece of cybernetic history returned home as a long-lost component of the Electronic Delay Storage Automatic Calculator (EDSAC), one of the first practical general purpose computers, was returned to Britain from the United States. The electronics chassis was given to the The National Museum of Computing (TNMOC) at Bletchley Park, where it will be used as part of the EDSAC reconstruction project and raises the possibility that more surviving parts may be recovered in the future.
In its heyday, EDSAC was one of the bleeding edge "electronic brains." Inspired by the ideas of computer pioneer John von Neumann, it was built in the late 1940s just after the Second World War by a team under Sir Maurice Wilkes in the Mathematical Laboratory of the University of Cambridge, and is credited by the University as the first "stored-program electronic digital computer."
UPGRADE TO NEW ATLAS PLUS
More than 1,200 New Atlas Plus subscribers directly support our journalism, and get access to our premium ad-free site and email newsletter. Join them for just US$19 a year.UPGRADE
Though Alan Turing was at Cambridge while EDSAC was being built, he didn't have a hand in its construction because he and Wilkes had fundamental disagreements about computer design, with Turing wanting simple hardware for experts while Wilkes preferred American-style reliance on hardware; making the computer what would today be called "user friendly."
In a remarkably forward-looking decision, EDSAC had a modular design, but it was hardly compact. Standing two meters (6.5 ft) high and covering an area of 20 m2 (215 ft2), it was built long before the invention of the transistor and consisted of over 3,000 thermionic valves set in chassis in 12 racks, each holding 14 chassis of various designs.
Another interesting point about EDSAC is that though it was programmed using paper tape and printed out its results on a teleprinter, it used one of the first assemblers called "Initial Orders" and instead of machine code, it was programmed symbolically. To store data, it used a mercury delay line memory, which used sound pulses running through a tank of mercury. At its most powerful, its memory had 1,024 words or the equivalent of 4 KB in modern terms. Still, it could handle 650 instructions per second, which the TNMOC says was 1,500 times faster than previous mechanical calculators.
EDSAC went online on May 6, 1949 and was used by scientists at Cambridge until it was dismantled in July 1958 with its parts being either donated to team members or sold for scrap. One part of that scrap was Chassis 1A, which was donated to the TNMOC museum by Robert Little, of Allentown, Pennsylvania after learning of the museum's project to reconstruct EDSAC.
The chassis isn't much to look at, little more than a rusted and bent steel frame with a few surviving valves still in place. Originally, it held 28 valves and, in terms of capacity, its the equivalent of a very simple logic circuit. This particular chassis was one of the Storage Regeneration Units, which fed into the mercury memory tanks.
According to TNMOC, the find is a bit on the miraculous side. Forty two chassis were built, but only three others survive with one at the Cambridge University Computer Laboratory and loaned to TNMOC, the second at the Science Museum in London, and the third at the Computer History Museum in Silicon Valley. In its distressed and corroded condition, Chassis 1A isn't usable, but it's hoped that some of the surviving valves can be incorporated into the reconstructed EDSAC.
One mystery that remains about Chassis 1A is that its provenance is unclear, but it seems to have been acquired at auction in the 1950s after EDSAC went out of service and was sold for scrap. It was purchased by Dr Robert Clark in 1969 when he lived in Cambridge. He already owned three or four EDSAC racks, which he'd planned to convert into bookshelves.
Andrew Herbert, leader of the EDSAC reconstruction project, believes that other EDSAC parts may still exist and could be stored at locations in the Cambridge area. However, any EDSAC parts in the possession of Dr Clarke before he passed away in 1984 are probably lost with the exception of Chassis 1A that passed into the hands of Robert Little.
"I regret that the probability of finding any more of Dr Clark's EDSAC parts is vanishingly small," says Little. "Dr Clark passed away in 1984. Sometime between 1969 and 1984 he relocated to a house on the outskirts of Cambridge and quite probably disposed of unneeded things then. The bulk of the machine was most likely destroyed. I have a vague recollection of Dr Clark describing how he was bidding against the local scrap metal dealer when he won the EDSAC parts at auction. Despite this, I am hopeful that those who built and worked with EDSAC kept other mementos that have been preserved intact until now."
The EDSAC reconstruction project, which began in 2011, is scheduled for completion by the end of 2015 and has already benefited from early EDSAC diagrams discovered last year.
The video below shows the opening of the EDSAC gallery at TNMOC.