How I learned to love Autocode

By Stan Rodgers.

From 1957 I was studying for a Ph.D. in a group doing research at the Nuclear Physics Research Laboratory (NPRL) of the University of Liverpool, using the 400 Mev synchrocyclotron. My first contact with a computer was when the Maths Department, which had recently received an English Electric 'Deuce', offered an introduction to its use, which I attended. I found it all very complicated. As I recall, it was a 3-address machine, i.e. each instruction contained three addresses, operand-1, operand-2 and the address of the next instruction. The instructions were stored on a rotating magnetic drum, and one had to bear in mind that in the time it took to read an instruction and process it, the magnetic drum had moved so that the next instruction which could be read was at least 5 instructions further on. This meant that in order to have the program run at a reasonable speed, one needed to construct the program so that the 'next instruction' for each instruction was at least 5 instructions further on in the program sequence. Furthermore the computer was programmed directly in machine code! I didn't see how I could use it for my work.

When I needed to perform any serious calculations, the NPRL had a Brunsviga calculator (German), where after setting the numbers to be multiplied or divided, one turned a handle. Later there was a Facit calculator (Swedish) which was much easier to use as it had an electric motor to drive it. About 1960 I was working on some relativistic calculations, where I had to calculate the energy of particles created in a two-body interaction. It was relatively straightforward, I just had to start with the two particles going off in opposite directions at a given angle in the centre-of-momentum system, and find out how they looked in the laboratory system. This involved applying standard relativistic formulae, and I needed to do the calculations for a range of angles. I got started on this job using the Facit calculator and it was soon clear that each angle would take about half a day, so I was facing at least a week's work. While I was doing this, our Prof. (the late Sir Alec Merrison) came in and asked me what I was doing. I told him, and he then said that there was a computer in Manchester (UMIST), which was a lot easier to program, which I should try. This was the Ferranti Mercury. I contacted UMIST, and they sent me a manual for Ferranti Mercury Autocode. I studied it, then wrote a program which I keyed in on a teletypewriter to produce punched paper-tape. I sent this by post to UMIST (in those days the postal service was very efficient), and the next day got it back, showing an error in my program. I corrected that, sent it again, and it came back with my week of work done after a few minutes of computer time. That was it, I was a convert!
Later I became more ambitious and was able to write a program in Autocode which performed a Monte Carlo simulation of another reaction which would be going on in the experiment in parallel to the one we were planning to study, so as to be able to estimate the background events which we would need to allow for.

These experiences influenced the whole course of my life, in that I earned my living for 36 years working with computers for a major US computer manufacturer, during which I travelled widely throughout the world.

Stan Rodgers, February 2011.