T-time at EDSAC1 - J Bruce Forsyth

October 2015

J Bruce Forsyth (Cavendish Crystallographer 1955-61) writes:

One of the most valuable rituals for users of EDSAC were the T(est)-times when program writers, on reaching the head of the queue, could submit their latest program paper tape and have it run. This was usually of short duration, since the programs were in the course of debugging.

One frequent T-timer I remember well was Sir Edward Bullard, then the Head of the Department of Geophysics at Cambridge University. Having selected a tape from his collection, some coiled others hung around his neck, it was proffered to the operator. Whenever it crashed, which was not infrequent, he would proclaim in a loud voice:

“It’s a machine error!”

This was followed by an attempt to rejoin the queue at its head.

My wife recalls that when Sir Edward’s programs did run they were lengthy and were responsible for many a delayed supper when I eventually returned home after an early evening ‘run’.

Following my visit to see EDSAC at TNMOC, I should like to add that EDSAC and EDSAC2 were invaluable to the Cavendish crystallographers engaged in structure determination.

First the calculation of Patterson maps derived from the observed intensities of single crystal reflection intensities. Then the calculation of the structure factors from a trial model and the comparison with what had been measured (the latter task with the help of part-time 'calculating ladies'). The trial atomic positions were then moved up the slopes in a difference density map and the structure factors recalculated. The whole process was referred to as a Refinement.

The acquisition of more accurate intensity data than that obtained from photographic films and eye-estimation led to the construction of X-ray diffractometers in which film was replaced by a small-area detector scintillation detector. Both the crystal orientation and the position of the detector had to be adjusted before each reflection could be measured.

By mounting the crystal with a principal symmetry axis vertical a zone of reflections could be measured with only two angles being set: the rotation of the crystal and the position of the detector in the horizontal plane. Nevertheless, it was a tedious business crying out for automation.

It fell to Jane Brown and I in the Cavendish to design and produce MAXIM - Machine for Automatic X-ray Intensity Measurement. It was an unholy mixture of dekatrons, uniselecters, trocotrons and NOR gates with the circles encoded by homemade single turn encoders on the worm drives. Punched paper tape from EDSAC carried incremental angles from the last positions AND the absolute angle from the circle zero. Positioning was made using the increment but the sum total angle was also updated and then compared with the EDSAC tape. Error recovery was then possible by returning to the circle zero and using the absolute angle as a single increment. The measured reflection intensities were then recorded on more punched tape for return to EDSAC. It was quite a collaboration!

After I had moved on to Atomic Energy Research Establishment at Harwell to shift to polarised neutron studies, I was faced with the same diffractometer problem. The automation was to be provided by a PDP8 with two fields of 12K words each of 12 bits, but no floating point hardware. However, there was no need for paper tape, the coupling being very close!

After a few years the UK joined the Institut Laue-Langevin in Grenoble where its fine beam reactor was already operational. Part of the entry fee was that I should design a polarised beam diffractometer to be controlled by a PDP11 with hardware floating point. What a relief! Also the military had released multi-turn encoders which removed the need for incremental positioning. D3 is still operational, now equipped for general polarisation analysis.