Early days at the ISC

Notes from David McGregor (former Deputy Director of the ISC)

Received: February 2015

I was born in Liverpool in 1938, educated at Liverpool College and Liverpool University where graduated in 1960 with a BSc(Hons) degree in Geology. Following this, I was a research student for three years using geophysical methods to investigate geological structures in Aberdeenshire. This was primarily done by a gravity survey. I used an English Electric DEUCE computer for reduction and analysis of the gravity data.

In the autumn of 1963 I saw an advertisement in the Guardian for a job at the International Seismological Research Centre (ISRC) – as it was called at that time. I applied and came up for an interview with Pat Willmore, Donald Michie and Peter Felgett. I was offered a job and began my employment at the start of 1964. The first two days of January were national holidays in Scotland. Pat Willmore did come to Edinburgh Royal Observatory on the second of January to see if I had turned up for work in case I was unaware that it was a holiday in Scotland.

I was the first ISC employee. A statistician, of Indian nationality, also joined, possibly at the same time, but left fairly soon afterwards to take up a lectureship at Reading University.

One of the first tasks was to set up the data collection system. It had been decided to base the data collection on punched cards. Stations and networks contributing data were encouraged to send the data in a computer readable form. Data was to be accepted on half inch seven track IBM magnetic tapes (nine track came later), Hollerith punched cards, mark sense cards, coding forms issued by the ISC and printed station bulletins.

Two types of Hollerith cards were printed. A red card for the first P/PKP and S/SKS onset times with two columns for a log A/T value. A blue card for the times of other supplementary phases. The ISS in Kew had had some experience of using mark sense cards and had not found them very satisfactory. Mark sense cards had the data recorded using a pencil containing very soft lead. The marked cards were then put through an IBM reproducing punch machine which detected the pencil marks and punched a corresponding hole in the card, so that they became punched cards. These card punches were controlled by wired panels. There would be a selection of wired panels and the appropriate one would be put onto the card punch machine for the particular task.

The printed bulletins and coding forms were punched onto cards using IBM026 card punch machines. The cards were then sorted using an IBM card sorting machine that sorted a chosen column into 13 bins. The cards were then written to magnetic tape files and merged with data that we already had on magnetic tape. The input to the programs which ran the location procedures was a time sorted set of card images in which previously determined epicentral estimates and station first arrival times were assembled in time order but with no link between the epicentres and the station data.

At this time Edinburgh University had very limited computing facilities. There was a desk top PDP computer in the department of Artificial Intelligence which we never used. There was also a paper reader connected to a paper tape punch in Manchester. I can remember using this. The paper tape was prepared on teletypes in the Royal Observatory, taken into the centre of Edinburgh, where the paper tape reader was located. The paper tape was read in Edinburgh and a copy punched in Manchester. The paper tape, containing a program in Atlas Autocode, was read into the Atlas computer in Manchester and a paper tape was created with the results. The output tape was then read in Manchester and punched in Edinburgh. The user collected the paper tape and read it on a teletype to print the results.

The Animal Breeding Research Organisation moved out of 6 South Oswald Road in the summer of 1964 and the University of Edinburgh made the premises available to Dr Willmore’s groups. The IGS seismology group and the ISC moved down from the Royal Observatory at the top of Blackford Hill.

I was the first person to move into South Oswald Road as I had been banned from the Royal Observatory. My wife and I had visited Aberdeen one weekend at the end of May and while we were there it became apparent that there was an outbreak of typhoid and the number of cases was increasing rapidly. Dr Bruck, the Astronomer Royal, discovered that I had been in Aberdeen and instructed me to keep away from the Observatory for the next few weeks. The invitation to his daughter’s wedding was also rescinded. I had already checked with the Medical Officer of Health, who advised that I was not a risk to others as long as I took normal hygiene precautions. My wife was also instructed to not turn up for her job as a teacher. Another Royal Observatory employee who had driven through Aberdeen and went to see Dr Bruck, hoping to be banned, but was told to carry on.

Annette Fluendy joined in the late summer of 1964.

We got an allocation of time on the Atlas computer at the Atlas Laboratory near Harwell and needed to find somewhere locally where we could copy the punched cards to magnetic tape and prepare them for processing. Although there was an IBM bureau in Glasgow, which we did use occasionally, we needed a more local facility if we were going to be able to process the data. We looked around Edinburgh to see where we could find card processing machinery and/or a computer that we might be able to use. Scottish Omnibuses allowed us to use their card handling equipment and the Bank of Scotland allowed us to use their IBM 1401 (later IBM 1460) computer between 08:00 and 09:00 on weekday mornings. The bank were really very generous to us as we only paid for the machine clock time in contrast to the IBM bureau where we were charged for the time that we spent in the room.

The IBM 1460 computer had six tape decks and a small memory. In order to sort a file, the tape was loaded onto one drive and scratch tapes were put on the other five drives. The sort/merge program would then be read into the card reader with the appropriate controls. The program would then sort the input file onto three of the working tapes, we would then replace the input reel by a scratch tape and the data would be moved back and forth between the drives until a sorted file was produced. I remember that on one occasion we realised that the output file had lost 10% of the data and we rang the bank to warn them of the problem.

Once in South Oswald Road, we acquired our own card punches, a card sorter, a reproducing punch and a tabulator. We still needed to prepare the input tape and to convert the output tape which came back from the Atlas computer to punched cards. The punched cards were then printed on the tabulator. The resulting printed pages were then the source material for the printing company to make the ISC publications.

The program which processed the time sorted input tape was written in Atlas Autocode. This attempted to take the input file of previously known epicentre determinations and the phase readings to produce a bulletin without intervention. Included in this program was a section to locate new events from phase arrivals unassociated with any known events. One ISC bulletin was published using this method - ‘Preliminary determinations for 1964 January’. It was unsatisfactory.

It was after this that Ed Arnold, who had been working for his PhD under Sir Harold Jeffreys, came to work for the ISC. A new approach was adopted based on the work that Ed had been doing in Cambridge. Programming was done in FORTRAN. No attempt was made to find new events in the data. The early development of these programs was done at the National Engineering Laboratory at East Kilbride, then on a UNIVAC computer at a bureau in Birmingham. Later the work was transferred to the IBM360/195 at the Rutherford Laboratory in Chilton, Oxfordshire.

The following sequence was used to create the monthly bulletin.

  • Prepare a month’s data onto a magnetic tape in Edinburgh at the Bank of Scotland.
  • GROUP – Associate epicentral estimates and phase data
  • AUTO-REVISE – The location program in automatic mode
  • PI – Anatoli Levshin’s program to identify supplementary phases
  • Inspect the results and prepare a file to disassociate erroneously associated phase data and relocate selected events sometimes with restrictions on depth etc.
  • SEMI-REVISE – The location program with edits
  • Inspect the results and rerun SEMI-REVISE again if needed
  • PRINT – Prepare the printed listing for publication.
  • The output file would then be transferred from magnetic tape to punched cards in Edinburgh and printed on the tabulator creating copy for the printer.

It was only when George Gibowicz came to the ISC that new events were was added. Initially this was done by inspection of the unassociated data stream, later the program used by the NEIC to locate new events was incorporated into ISC procedures.

The early ISC Catalogues and Bibliographies were also produced using the Bank of Scotland computer and the card handling equipment in South Oswald Road.

It was also while we were using the bureau in Birmingham that the ‘Shannon’ tapes were created. The company in Birmingham were looking for work for their punching bureau in Shannon, Ireland and suggested that we could get some ISS bulletins punched onto cards. They got to work on this and soon asked for more ISS bulletins. It was soon obvious that the ISC funds would be soon be exhausted and a stop had to be called.