During the opening ceremony Van Wijngaarden addressed the audience as follows: "*Your Excellency, here at the Mathematical Centre we have three departments: the department of Pure Mathematics, the department of Statistics and the department of Applied Mathematics. The department of Pure Mathematics thinks in abstract terms, they imagine themselves a cube in 13 dimensions. The department of Statistics is playing dice with this cube and the department of Applied Mathematics has written a program which simulates this process. If Your Excellency would be so kind to push this button then the ARRA will play dice with the cube and each score will be registered.*"

The ARRA started to generate random numbers indeed, but suddenly stopped. Van Wijngaarden reacted immediately with the remark: "*This is a highly remarkable situation: the cube is balancing on one of its corners and does not know which way to fall. If you push this button you will give the cube a little push and the ARRA will continue its computation.*" It worked, the Minister said "*That is very interesting*" and left the department of Applied Mathematics.

Source: N.C. de Troye, Memories of the Mathematical Center (translated to English)

The ARRA (*Automatische Relais Rekenmachine Amsterdam, Automatic Relais Calculating Machine Amsterdam*) was one of the first computing machines in the Netherlands, built by the Mathematical Center in Amsterdam. It was developed by Scholten and Loopstra from 1948 to 1952, both attracted to the MC by Van Wijngaarden. Dijkstra was the first programmer of the machine. Unfortunately the device never properly worked; when it was officially introduced, it broke down and never worked again.

The ARRA contained 1200 relays, connected at the back of the machine. These relays were placed in little plastic bags to prevent them from becoming dirty. Still the switch times were unreliable due to contact degeneration. The pins of the relays also had to be cleaned regularly. The computing girls often helped with this process. After the cleaning the relays had to be tested again, which put the contact pins under high voltage levels. Despite the wooden handles on the instrument used to set up the relays, this didn't seem to suffice considering the choice of vocabulary during the process.

For the flipflops in the machine double triodes were used, followed by an end tube to allow for controlling enough relays. A light bulb was used as visualization for the flipflop. With its 3 registers of 30 bits each (and thus a word-length of 30 bits), the ARRA contained 90 of these endtubes. It was discovered by Scholten and Loopstra that these could oscillate almost perfectly. It took them a while to figure out where they needed wires of roughly 2 meters to prevent this oscillation.

At first the commands were given using uniselectors from a plugboard. Drum memory made an end to this in 1951. ARRA was given instruction codes of 16 instructions, including two for multiplication and two for division - with and without rounding. The machine was able to represent binary numbers as decimal. A program was stored as one instruction per word. For input and output a (paper) punch-tape reader (magnetic tape still had to be invented) and a teleprinter (TTY) were used. A wooden kitchen table served as a control panel. To make up for the regular trips to Philips in Eindhoven Scholten and Loopstra often took home some sample parts, making the ARRA a steel chart of what the electronics industry had to offer.

Since the ARRA had no clock, every primitive operation ended with a "operation complete" signal. That signal triggered the succeeding operation. To prevent glitches such a system is even today still propagated occasionally, though Scholten hopes that such systems will be more reliable[1].

The opening of ARRA I, 1952. From left to right: Loopstra, Secretary of Education Rutten and mayor D'Ailly

The ARRA only worked once. During the demonstration, with Secretary of Education F.J. Th. Rutten and D'Ailly, mayor of Amsterdam, the developers picked a program that had the lowest chance of failure. Random numbers were to be computed. It had been rehearsed so often that the developers knew large parts of the row by heart. They were relieved that the machine produced some output. After the opening the ARRA never produced a single correct line of output again though, and it was given up as hopeless artifact towards the end of 1952.

*Computers Ontwerpen, Toen,*, 1979 .