Jurij Semkiw – CSIRAC maintenance engineer
CSIRAC was a serial machine, requiring a precise relationship between the length of the mercury delay lines and the frequency of the main clock, which in turn were dependent on temperature and supply voltages. This necessitated continual adjustment as it would only work within a limited range. This never-ending work was the responsibility of the maintenance engineers. CSIRAC users were only permitted to make elementary adjustments.
The room in which CSIRAC was housed was ventilated but not air-conditioned. To help regulate the operating temperature, air was drawn in from under the floor, passed through the machine and expelled from the building. In summer, when the temperature was in the high thirties, it (CSIRAC) had to be switched off to prevent it from overheating.
CSIRAC’s primary memory store consisted of acoustic delay lines – metal tubes, filled with mercury. The mercury would somehow become contaminated… the crystal in the transducers would become coated with oxides. When lines became inoperative, the contaminated mercury was emptied out of the line and refilled with triple-distilled mercury.
It wasn’t simply a matter of emptying and refilling the lines. A strict procedure had to be adhered to. The lines had to be disassembled, cleaned, and the oxides on the surface of the crystal had to be removed. When the lines were put back into operation they had to be individually tested and the timing adjusted.
Everyone in the Laboratory was exposed to the mercury in some way, but I had a lot of exposure to mercury metal and mercury vapour because it was my responsibility to empty and refill the lines. In those days, there was little consciousness or concern of the toxic side effects of mercury.
At the time of the closing down of CSIRAC on 24 November 1964 Ron Bowles was working on the IBM 7044, so it was left to me to disassemble it. I carefully took it apart, dismantling in such a way that it could be reassembled again if need be.