Abstract: Took Science degree at Imperial College after national service in RAF. Joined LEO after seeing an advertisement. Steve worked as a commissioning engineer first on LEO II machines including commissioning the Bull Printer on LEO II/3 and subsequently on LEO II machines. Achieved rank of Chief Engineer for LEO III. Involved in design of the abandoned LEO IV after merger with English Electric. Subsequent career from 1965 with Medical Research Council working as a design engineer.

Transcript :

LEO COMPUTERS LIMITED - Oral History Project
Interview with Steve Farrow by Martin Garthwaite 

Martin Garthwaite: So Steve…
Steve Farrow: I don't know what to do. Do I just talk or are going to ask me things?
 M. G.: Yes, I'm going to ask you questions and we’ll go from there. First, thanks very much for taking the time to speak to us today. As you know this interview is in pursuit of the Oral History Project that the LEO Computers Society is undertaking of people that worked at Lyons and their recollections of working on the LEO computers.
S. F.: Yes.
M. G.: So let's start with your family. Can you tell me where you were born and the occupation of your father and mother?
S. F.: I was born in Poole. My father was a solicitor’s clerk, my mother was a mother. That's all.
M. G.: Ah, thank you. Do you have any particular early memories of your childhood?
S. F.: Any early memories of my childhood? I remember being given a watch on my birthday and cycling excitedly along on my tricycle when I was about four and getting it out to look at it and running into a lamppost.
M. G.: Tell me about your early education, which school did you go to and whereabouts was that?
S. F.: I went to several primary schools as my parents moved around a bit. Then I ended up at the local church school in Dorset in a place called Burwood, which is near Wimborne. Then I got a scholarship, that is I passed the passed the eleven plus, as it was, and got a scholarship to go to a local grammar school where things were pretty hard. We had reports every fortnight and if we didn't do too well then we had to do extra work. I never actually did get caught for that. Then I did General School Certificate, or ‘O’ Level as it is these days, but it was School Certificate when I was fourteen, which was by these days, fairly early. As a result of that went into the sixth form and did the four sciences as it were pure maths, applied maths, chemistry and physics.  I was always interested in physics and the sort of thing. After the first year at the Secondary School, the Grammar School, we were screened either maths and physics or Latin and three other subjects, depending on how well we’d done in the first year. And I knew all six and anybody else probably would in that class. But anyway I was put into Latin. My Father sent a letter which said “Can Steven please do physics?” And the headmaster said “He will learn much sooner than he’ll need to know”’. Because no doubt he was a classics man. Anyway it turned out in the end that I did get a Physics Degree. I took it at ‘O’ Level, or whatever they called it, General School Certificate when I was fourteen and then at sixteen I took Higher School Certificate and as a result of that got a sponsored county scholarship to go to university.
M. G.: OK. Whereabouts did you go to university?
S. F.: Imperial. I didn't do too well but I went in the air force first.
M. G.:	OK.
.S. F.: When I had the choice I opted to go into the Air Force before going to university. In fact I went into the Air Force earlier than I normally would have done. I was actually eighteen but it would have been a bit older normally. I went in early because I wanted to get out early in order to go to university and that was allowed. What in fact happened was that while I was in the Air Force the Korean War happened and everybody had to do an extra six months. But those who had done what I did and went in early to get our early could get out when they should have when they were meant to. So in fact I only did eighteen months and what is more I went to Cyprus. It's a place where people spend an awful lot of money to go to and I was sent there by the Air Force and had a very responsible job. As you probably know people did a pretty awful job sometimes as National Service. But we were in charge of the radio in two places. Later they actually trained me as a ground wireless mechanic on grade B transmitters. I did some work on ground transmitters in Cyprus, but then I was also sent to the airport where we were responsible for any of the radios on any of the aeroplanes that came in. In fact there were only three of us and neither of the other two were trained in radio, they were trained in radar. I was trained in radio at least, but I was a ground wireless mechanic. An aeroplane would come in, any aeroplane, and we’d would have to try to test the radio. Anyway so that's what I did. And then I got out of the Air Force, went to Imperial. I started off all right and then got a bit disoriented with politics and girl friends and didn't do particularly well, but anyway I did get a pass degree. I didn't use it really. For the first thing I did was cooking research for the North Thames Gas Board. Is all this what you want to know?
M. G.: Yes, absolutely.
S. F.: The first thing I did was this cooking research. Something like what your toast looks like if you cook it in different temperatures.
M. G.: OK.
S. F.: Then I saw an advert for LEO Computers and sat an aptitude test which was based on computer programming.
M. G.: OK.
S.F.: I passed it and then went to LEO. Do you want me to carry on from there?
M. B, Yes, absolutely. Tell us about your early years at LEO. You say you sat the aptitude test and passed that, which I understand was a pretty tough test?
S. F.: Well, yes I think it was. Programming hardly existed then.
M, B.: Of course it didn’t.
S. F.: If you want to alternate the value of a number so it's alternately three and four you can test what the current value is and then load the other value. But when and that takes two or three instructions. But in fact if you subtract the current value from the sum, the answer is the other number. Less instructions.
M. G.: Yes.
S. F.: And they were looking to see whether anybody could twig that.
M. G.: OK.
S. F.: That was the sort of thing, 
S. F.: Anyway I got through that then I went on the LEO II training Course. They had just started the training school and think I was in the first batch of people who went. Everybody else learnt on the site. When I went to work on II/1 I knew more about it than anybody there, I knew more about how it actually worked. It had the algorithms for multiplication, for example or  square root. The people there could mend it but they didn't actually know how it worked. Basically you could look at the logic diagrams and trace the signals and see which terminals on the chassis were involved and look at them with an oscilloscope and find out that they looked bloody awful, you tried to find out what was wrong with that one. The other people on the course were the engineers for the first computer that was sold outside Lyons to the tobacco firm in Bristol, WD & HO Wills and it was the first time and they were having to train engineers belonging to the outside company where the computer was going. I think that I got there before the course started and I suppose it would be at that time the computers had about five thousand radio valves in them.
M. G. Yes, yes, a very a very large number.
S. F: It was important that they didn't break. And they had the idea that the it was very important that the heaters didn't break. You know what the heaters are?
M. G.: No, what are the heaters?
S. F.: You know what the valves are?
M. G.: Yes.
S. F.: It was very important the heaters were run at 6.3 volts, which is what they normally are. Well that's what they're supposed to be, I mean the factory say 6.3 volts. And in fact every unit... You know what a unit looks like? It's about two feet long and....
M. G.: Yes, very big.
S. F.: ...six inches wide.
M. G.: I've seen some photographs on the LEO website, but no I don't know about that particularly...
S. F.: But anyway, every unit which was a thing about two foot long and six inches wide. It could have up to fourteen valves in, individually plugged into the chassis. Every unit had a transformer for producing 6.3 volts.
M. G.: OK.
S. F.: But the load on the transformer depended on how many valves there were and that's what the voltage input voltage tappings on the transformers were for. They were normally 230 volts, but they had tappings between 210 and 260 volts. And you could alter the voltage on the valves by altering the tappings. And I was given the job of working out the given the change needed it so it was 6.3 volts, which it usually wasn't. So you could measure what the voltage actually was and then look to see what input tapping was and change the input tapping until you were getting 6.3 volts. Well that was the idea anyway. I was given the job of working out by hand without a computer and without a slide rule, the table for it and what can you put the tapping to? Anyway  that would go on for hours. So then I did the course. What can I say about the course? It was good. Then I went to II/2.
M. G.: What's II/2?
S. F.: II/1, sorry, I never went to II/2
M. G.: And this was the model of the LEO?
S. F: I'm sorry, what do you mean the model of LEO? The model number you mean?
M. G.: Yes. When you say 'two two', what does that mean?
S. F.: I'm sorry, LEO I was the first one which was a Chinese copy more or less of the of Cambridge University, EDSAC, (Electronic Delay Storage Automatic Calculator)  which worked in binary, as near as possible. They didn't do anything clever, they just wanted it and copied it, but they made great changes to the input and output system and added decimal and sterling conversion and reconversion, because the Cambridge computer typically was given a couple of numbers and ground away for hours and produced a single answer to do a mathematical job. Whereas what Lyons wanted to do was things with lots of input and input and output data. They wanted to calculate a payroll.
M. G.: Of course.
S. F.: So the input and output mattered a great deal. The speed of input and output mattered a great deal. And in fact a lot of the time the computer was waiting either for the next lot of input or waiting in order to build a stream for the next lot of output. It was waiting for the card reader to be ready to read a card or the printer to print something off.
M. G.: OK.
S. F.: We did in fact have two input channels and one or two output channels. They may have been more in order to have different data coming in and different printing data, not a question of parallel running. But anyway a lot of the time the computer was waiting, either for input or output. And while it was waiting it would just go round a very short programming loop. And, as you may know, there was a loudspeaker on the computer that could listen to. It was connected to particular wave form. You may have heard the recording of the computer running.
M. G.: No, I've not. But it sounds very interesting, [laughs].
S. F.: You can hear it on some disc that's available.  And it was while it was running the loudspeaker was listening to the clock, not exactly the basic clock, because that never changed, listening to the actual action being carried out, and so it would go ‘be de le de le de le de’.
M. G.: OK.
S. F.: As it did instruction after instruction. And then when it was waiting for input or output it would do the same thing over and over again and it would go ‘be de le de le de leeeee’. Now we used to call that howling on output, or howling on input. One could tell in a vague sort of a way whether a job was working properly because of the noises it was making. The programmers could never tell but the operators could. They knew what sort of noise it was supposed to make. And sometimes it didn't. I was so taken up by being a computer engineer that I used to dream of this compute, and when I was half asleep and wanting to do a pee, I used to dream I was howling on output!
M. G.: [Laughs].
S. F. I went to II/1 and the first job I did was to change the storage units. Do you know how the store works?
M. G.: I think it was a mercury delay?
S. F.: No, how the store worked, the memory?
M. G.: Oh the memory, yes. Was it a mercury tube?
S. F.: That's right. And there were sixty-four storage units, one unit to each mercury store delay tube, which we were running at radio frequencies. The actual calculating speed was 5.25 kilocycles, but the signals going down the delay lines were 12.6 megacycles. And these storage units were the things that could that. You could handle the data, putting it in and out of the delay line. And transfer it, changing the data that you wanted to. They were circulating the data as one quarter of microsecond pulses. And then returning the data to one microsecond data for computing.
M. G.: Even though you started doing the programming aptitude test, it sounds as though that you spent a lot of your time on the physical hardware aspects of the of LEO. Is that a fair assessment?
S. F. Well yes, it was only that nothing had got started properly.
M. G.: Ah, okay.
S. F.: I was asked how a logic gate worked with diodes. Does that mean anything to you?
M. G. Yes it does.
S. F.: OK. I was showed a logic gate with diodes and. they asked how that worked, and I had no idea. I was interviewed by the personnel officer, such as he was, and he told me how much money they would give me. He said ‘Well, am I making the right sort of noises?’ The whole thing was still being set up, everybody was an amateur or not far off.
 M. G.: Sure. 
S. F.: So apparently when I got to II/1, which was in part of Lyons in Hammersmith, they had just decided that the storage system wasn't made well enough. It wasn't designed properly, or could be better and a whiz-kid, who left not long after I got there, redesigned the transmitter and receiver parts of the storage units and they wanted a change to put the new ones in. The old ones were called LD. The units had numbers.
M. G.: OK.
S. F.: The units on LEO I were called LC. I don't know where the ‘C’ came from. The original units on II/1 were called LD, but now they wanted to upgrade the LD’s to LE’s which were the better designed. They tried to change one or two of them and that didn't work because they were incompatible. The old ones and new ones were incompatible, because the timings of the units were very important, very, very critical. So we were going to have to change all of them, all sixty-four LD1’s to LE1’s, and a few more for the storage units which were handling the input and output in the basic system. I was given the job of checking sixty-four plus new LE1’s to make certain every one of them worked before we put them in. So in one gigantic exercise I tried every one of them. There were three valve s in the receiver part of the thing and I tried and picked three valves in all possible positions in the three sockets in the receiver to find which valve was best. There were three by two by one different ways of doing it I suppose, anyway I collected together all sixty-four plus working, if not compatible, working LE1’s. I didn't do it because I wasn't competent at that time, but I was competent enough. Over a night and a day all sixty-four plus LD1’s were changed for LE1’s, and then all hell was broke loose when we tried to get the system working.
M. G.: So there was often down time when you made upgrades to the hardware?
S. F.: Yes and often while the programs were running. Do you know the name George Manley?
M. G.: No, I don't know that name.
S. F.: Well I’ll tell you about him in a minute, but he frequently found why the thing the computer wouldn't work and do a particular programme, I mean job.
M. G.: OK.
S. F.: And so one would redesign something. It, what's the word, I mean ‘on the hoof’. Because you suddenly discovered it wasn't as good as it might be.
M. G.: OK.
S. F.: The computer was so unreliable that the programs only ran for twenty minutes and then the total results we’d got so far were punched out on Hollerith cards, then the computer was given to us engineers to test to see if it was all right, and that test took three quarters of an hour. Then it would be handed back to the operators who would reload the part run, and off they'd go and do another twenty minute run and then it would be given back to the engineers to test.  That was the state of II/1 when I got there. 
M. G.: OK. Did you work with any other client companies of Lyons of LEO? Did you actually, as part of working at Lyons, did you get sent out to other customers to work with them, or were you just purely focused on the Lyons business?
S. F.: I'm not quite sure what you're asking. I only worked for Lyons.
M. G.: You only worked for Lyons, so I was referring to when Lyons set up this separate business to sell the computers to other companies. I wondered if you were involved in that part of the business too?
S. F.: No. That was the state when I got there. They'd made LEO I and it appeared to be a success and they were apparently cutting edge. So, I mean incredible the people who made cakes and teacakes found out what was going on in America after the war.
M. G.: Yes.
S. F.: The higher ups who looked after teacake firm decided that they would start making computers commercially, and it was an incredible decision.
M. G.: Yes.
S. F.: And then they said we want to know if our computers work “Go back and have a look at Cambridge”. Then they decided that it looked as though the future of office work was computers. They tried to get somebody to make them one but nobody would so they said ‘Well all right, we’ll make our own, we’ll copy the one in Cambridge’. And that was an incredible thing to decide to do. And then having made a success of that they then, more incredibly, decided they'd set up a separate company to make more of them and sell them.
M. G.: Yes, indeed. 
S. F.: No, I never worked for any of the other companies, but I did in fact make the computer that went to the LEO II/4. I built the one which went to Fords to handle the spare parts at Aveley and I did go with it to look after it while it got going, but that's not what you're talking about.
M. G.: That was I was getting at, if you were involved in that aspect of the business because, as you mentioned, a very amazing thing about Lyons is, number one they made an incredible decision to build a computer and then they decided that it was a business that they wanted to get into also, which is incredible for a company that sold tea and cakes.  So tell us when did you join Lyons, what year?
S. F.: ‘57.
S. F.: So anyway after I had been at II/1 to about September I suppose, and then in January I was drafted back to Minerva Road in Acton where the computer factory was. And I worked for a bit on II/3 which went to Stewarts & Lloyds at Corby. I didn't do very much on that but I got the printer system working because after the system had been designed somebody had to make it work. In fact it had to be redesigned because a major error had been made somewhere. and....
M. G.: ...it needed to be fixed.
S. F.: Yes and I found the major error when I tried to make the thing work. Then, having fixed that, I then became the commissioning engineer, the chief engineer, the man that was responsible for it,  the leader of the people building and putting together LEO II/4.
M. G.: Excellent.
S. F.: The interesting thing about the II/4, I suppose, was that it had the speed of output for what they wanted to do. The printer I commissioned on II/3 had alphabetic output code. The printers up to that stage were only Hollerith printers which only had numbers but the one on II/3, which came from France, Compagnies de Machine Bull, that could print letters. But the storage system could only hold the code for eight characters. It could print the whole alphabet but only eight characters at a time. But the printer on II/4 was a marvellous device. It stood as high as me or higher, a great big thing. And the way it printed was it had a, there was a bar which ran across the width of the paper for a hundred and thirty-two character widths. And that bar was oscillated sideways. The paper moved continuously under this horizontal bar across the paper and through the bar were a hundred and thirty-two... Do you know how the loudspeaker works?
M. G.: Yes.
S. F. OK. Well there were a hundred and thirty-two wires driven by loudspeaker coils with one loudspeaker coil per wire. And the wires were hit onto a continuous wide inked ribbon. The wires hit the ribbon at the right moment to form the character, and bearing in mind that the bar was wiggling from side to side, and the wires moved up and down, and the paper was moving continuously underneath it. Are you with it?
M. G.: Yes. Very high for those days.
S. F.: Well, yes, but the thing was one off. When the computer went to Fords an engineer who had been engaged from Powers-Samas who made the printer, went to Fords and his whole job was looking after that printer. And it was so big and so open, there was so much of it, that in the hot days one could sit down and stick your head in it and go to sleep.
M. G.: [Laughs].
S. F.: I built II/4. I don't know much else to say about it.
M. G.:. Let me ask you about your colleagues at LEO, or Lyons. What impressions did you have of them? I mean you said earlier that everything was amateurish so I'm interested in the esprit de corps that people felt about what they were doing. Did you feel you were at the leading edge of something that was going to be big?
S. F.: Are you asking about the esprit de corps?
M. G.: Yes.
S. F.: When I was building LEO III/1 I was the chief engineer. We were working twenty four hours a day you see. One of the engineers who lived next door to Brighton, and he used to come up by train. He was due to come on shift on the Sunday morning. But he was late and he apologised And the reason was that his brother-in-law had come to the house and committed suicide.
M. G.: Oh dear.
S. F.: So that made him a bit late. And that was the level of esprit de corps. We felt in those days on LEO III that if we’re half an hour late on starting that day the computer will be half an hour late in two years’ time. We really felt that what we were doing was important. As I said, we were working shifts and if the computer was broken nobody went home, they just stayed to help mend it. Lyon’s rules were overtime was paid if you worked between eight o'clock at night and eight o'clock in the morning. but if the computer broke no one would go home after one’s shift. One would stay overnight and try to help mend it and work on until the following day. And, because of Lyons’ rules, you, if you worked over twenty four hours or thirty six hours, you were still only paid overtime between eight o'clock at night and eight o'clock in the morning.
M. B: OK
S. F.: And so you'd start off on normal time, go into double time, or time and a half, and then go back to normal time. Yes, that was the esprit de corps.
M. G.: Yes.
S. F.: It was that the company, and the whole thing was so small.
M. G.: Were you married during this period of work?
S. F.: Yes, as one might imagine, it was pretty hard going.
M. G.: Yes, that was my question, how was family life with all the very long hours that you were putting in?
S. F. Well one’s wives resented it.
M. G.: Right. 
S. F.: On the other hand I'd built up so much, so many brownie points that when my wife got fatally ill they said ‘Well don't worry, come in when you can.’.
M. G.: That was very nice.
S. F.: Yes, it was. Family life was difficult, and my wife certainly resented it. On the other hand my salary kept going up and up. And there were wives , not surprisingly, who couldn't understand why one was so dedicated. Have you ever done any programming?
M. G.: Yes, I have.
S. F.: I mean you know how you can't stop.
M. G.: Yes, it's very difficult to stop.
S. F.: Yes, that's in some internationally famous biography. The man from House of Holland. When I went to the MRC (Medical Research Council) and we took him out to dinner. I was dead scared he was going to quiz me about things I couldn't answer because it was such a cloak-and-dagger project, so, so important. And all he would talk about was how he couldn't stop, leave his program, his computer.
M. G.: [Laughs].
S. F.: But you know it does get me if the cursor would be down in the bottom left hand corner and it wasn’t where I wanted it at the top.
M. G.: [Laughs].
S. F.: And you know, I go in to in to work on a Saturday morning and my wife says ‘You will write to that paper, won't you?’ And when I got into work all I could do is try to get the cursor in the right place.
M. G.: [Laughs].
S. F.: And, you know, that was a similar to the state we were in as engineers. We were absolutely fascinated what we could do. The engineers didn't program but they were fascinated about what the code and building II/4 was like having a baby. All machines had clocks and the first thing to do was to get the clock system running.
M. G.: Everything depended on an accurate clock.
S. F.: Yes. Well, there's more to it than that. The clock is more significant on LEO than it is normally. It had to be such that the time down the mercury delay tube was the correct integral number of pulse intervals. Are you with me?
M. G.: Yes.
S. F.: So the frequency had to be dynamically controlled. Recorded to the time it took the pulses to go down the delay tube. So getting the clock system working wasn't actually trivial. Anyway the first thing one would do was get the clock system going, and that is like getting the baby’s heart running, I mean beating. And then you get a card reader to read a card into the store. And it would have on it the code for twenty-two instructions. And it takes you time to get the card system working and to form the twenty-two instructions, twenty- two forty bit words.
M. G.: Yes.
S. F.: And then that would be the computer would be directed to that as a very basic programme. The bootstrap. This was a very basic test of the storage system and then you'd get that going and you can start talking to the computer. And then you start it, to see if it could add something, add things, two things together. And you'd see if it was subtracting things and then you'd go on and on like that and build it up and up and up. But it was, as I say, it was exactly like having a baby.
M. G.: [Laughs].
S. F.: And it was exciting, getting it to talk, it's heart to beat and it's exciting to talk to it and then for it to talk back to you.
M. G.: Right.
S. F.: So anyhow then I came to II/4 and went with it for a short time to Fords of Aveley to install it. And then when I came back and I went to observe Pinkerton’s Research and Development early work on a core store. I was in Commissioning and I was looking at, helping and observing the clever people of Design and Development making the core store work. Do you know how core stores work?
M. G.: Yes. 
S. F.: OK, so anyway the next thing was the thirty-two bit matrix. It came from Mullard’s, thirty-two by thirty-two matrix, a thousand and twenty- four words.
M. G.: Yes.
S. F.: Then I went to II/8. I wasn't chief engineer because Tony Morgan was the Commissioning Engineer on II/8.
M. G.: OK.
S. F.: And the store on II/8 was a core store and I got a core store working in parallel with Tony Morgan commissioning the rest of it. We had a lot of trouble getting the store to work on II/8 because it was four times as big. It was a sixty-four by sixty-four matrix and not a thirty-two by thirty- two matrix. And the noise it that came from it was significantly bigger than the level of noise we’d got out of the thirty-two by thirty-two. And so we had to re-design where necessary. There was a core switch.
M B: Yes.
S. F.: Well actually if you, having written a new word into the store, you put a lot of half currents though the matrix which are not big enough to switch the cores. The only one that does switch is the one which coincidence with an ‘x’ wire and a ‘y’ wire. But you hit the already written cores. This affects them, causing pick-up or drop-out of bits. We had to modify the store to do what is called a ‘post-write disturb’. Immediately after writing a word you would hit those already written cores with a half current to restore them.
M. G.: OK.
S. F.: That is ‘post-write disturb’, you disturb it after writing but you aren't looking to see what the outputs were. We came across this in research paper and had to do something about it. 

M. G.: Ah.
S. F.: I remember being impressed when, as a test, we took the store off one computer and put it on another one and could read what had been written on the first computer.
M. G.: Right, so that's it?
S. F.: Well, which it should have been but .it wasn't. The other thing we discovered on II/8 was the thing was temperature sensitive . The store in a box about a foot cube. We wrapped it in a heated blanket which was controlled in order to keep it at a constant temperature so the temperature then didn't affect it. That was II/8. It is relevant that the core store on II/8 had transistors in it. Up till that time everything was valves. Actually part of the system driving the store still used valves, but the transistors drove the core store. So then after II/8 I went to observe and help Deign and Development on the experimental transistorised parallel computer, not a serial computer like LEO I and LEO II. It was laid out on the great big breadboard, about six or eight foot long and, it was a forty-one bit computer. It was microprogrammed.
M. G.: OK, yes.
S. F.: That's right, so, yes, it was three things. It was microprogrammed, it was parallel and it was fully transistorised. 
M. G.: So effectively what modern a computer is today?
S. F.: I suppose so, yes. Actually I've often wondered whether there is anything in computers now that we didn't have.
M. G.: From what you've described, the core architecture, if you like, has not changed from that model. So all computers, it doesn't matter how powerful they are, are effectively built around that principal?
S F.: Yes. And the microprogramming was done by magnetic cores.
M. G.: Sure, yes.
S. F.: looked and observed what was happening and learnt about it and drifted, quite unintentionally into being the engineer who knew about the new prototype LEO III. 
M. G.: OK.
S. F.: Because I worked on the experimental computer for Design and Development I was the one who knew, vaguely, about how the new, the experiment, the prototype LEO III would work. And first, originally there was an idea, that every part of the computer of the prototype LEO III would have a dedicated engineer. There'd be one for the store, one for the input and one for the output and so on. Eventually it never came to that. It all became my responsibility. And I suppose that was in order to see what was happening. I was sent the typical microprogram. Well LEO III was one that was microprogrammed. I was sent individual instructions microprogram to look at and I found some errors in them. (When I went to the MRC my boss was going to a factory to see something I asked him if he wanted me to go with him. And he said he actually could do with a carping fault finder because that is what I am. I'm small minded and I check things many times My motto is ‘never do once what you can do twice’.) But anyway I got myself into the position where I was the final check on the microprograms that were sent for building. They always had faults in them.
M. G.: OK.
S. F.: So I became Chief Engineer on LEO III. But one or two parts of it I didn't actually know anything about until I worked on them when it was, oh a year or two later. I didn't know how every part of it worked, but I knew how most of it worked. We were working shifts, at least for half the time, to get the thing going. And I was the Chief Engineer, the one who got chased by the management.
M. G.: To make sure it was working?
S. F.: Yes. ‘Why is it taking so long to make it?’ ‘Why are you doing it so slowly?’ And, oh, there was a lot of talk during t this development about a unit called The Treasury Support Unit (TSU). They were the government people who decided whether a particular computer was good value.
M. G.: OK.
S. F.: But the important point is that LEO III was multi programmable. You could have several programs running at the same time. I don't know whether that's done these days, but because the input and output was slower than the computation it paid to be able to run other programs while it was waiting for a printer or a card reader.
M. G.: To run another job?
S. F.: It paid to get on and do something else.
M. G.: Yes. 
S. F.: It was something I think which Ferranti came up with on their Orion computer.
M. G.: Yes, I think they did.
S. F.: So we could do multiprogramming and we did a demonstration to the TSU who was the Boss. Everyone was fearful of the TSU Gods because they could say what the Government was going to buy and what it wasn't. I believe we were the first people to demonstrate a multiprogramming jobs to the TSU. But they were very suspicious because it turned out that by chance the total time to run three programmes together was less than the sum of the times when they were running individually. I mean it just turned out that way! I think that's a valid point, you know. So anyway it was the first demonstration, I think, given to the TSU of a multi programmed machine. Programmes could go wrong and halfway through the building III/1 so we added what we called a lock-out system using tags on the store and keys on the programs. If you wanted to get access to a word in the store you had to supply the relevant key to match the tag to access that word. And the idea was then that every one of the different programs running had its own key, so, one programme could not corrupt the workings of another. 
M. G.: OK.
 S. F.: The thing was covered in bells and whistles. Do you know the system, anything about the system that's marginal testing?
M. G.: Yes, I have spoken to a few people that, about the way in which the LEO was tested and it sounded very complicated and very rigorous.
S. F.: Yes if was. Your quality as  the engineer in charge of one of the LEO IIs was measured by how many of these weren't shorted out. The idea was to put a voltage on to all the sensitive points in the computer. But sometimes a particular point wouldn't accept the marginal voltage and you could sort out that.
M. G.: OK.
S. F.: And I had to take a test of how good the man running the computer was how many of these points was shorted out in order to make it run when you put margins on.
M. G.: OK. So what was the last LEO you worked on before you left for the MRC?
S. F.: The last one?
M. G.: Yes. Was it LEO III?
S. F.: No, I then commissioned the experimental computer for the one after LEO III that was being produced by Design and Development.
M. G.: OK.
S. F.: Well it never got anywhere because there was so much in-fighting when LEO and English Electric merged.
M. G.: Oh yes, I heard of that. 
S. F.: So much in-fighting as to whose computer was going to be made next that they lost so much time. But they had to Chinese copy the RCA Spectras which were IBM compatible. They gave it their own name, System 4. So I suppose that was the mistake of it. The computer that I built for Design and Development was going to be LEO IV. It never got anywhere. It was running and but it was never developed any further. So that was the last one I did. 
M. G.: And how long did you spend at Lyons?
S. F.: Until ’65 I think.
M. B.: OK, and when you left Lyons, did you carry on in computing?
S. F.: Yes. I went to work for the Medical Research Council who were always looking at cutting edge stuff. They wanted to do automatic chromosome recognition. They discovered somebody in America had made a scanner, or actually a scanner that photographed. The idea was they'd photograph the cells in which you could see the chromosomes. It photographed them on thirty-five millimetre film. It decided what they looked like and fed it into a computer. And the MRC and the Canadian government wanted to look at the chromosomes of everybody in Canada because they were worried about the background radiation from uranium or something. You know in those days we were talking about doing millions, millions of scans, looking at millions of people. The MRC was looking ahead to that and then it wasn't feasible to do it by humans because they were hoping to automate it and so the MRC bought a scanner and I went to the small four-people unit that was set up to write programs to scan these films. So that's what I did. Then I left.
M. G.: OK. And did you spend a long time at the Medical Research Council?
S. F.: Yes. I was there for thirty years.
M. G.: Oh so you effectively finished your career there?
S. F.: Yes.
M. G.: And you were always involved in computers and computing in your time at the MRC?
S. F.: Yes. Programming and designing and, well not, I mean not all computers, but designing and building bits that were useful for what we wanted to do. Well do you want this? It's not as interesting as LEO is.
M. G.: OK. The British Computer Society (BCS) was founded in 1956, so that's the year that you joined Lyons. Were you ever involved in these societies, you know, the BCS or the Institute of Electrical Engineers?
S. F.: No.
M. G.: OK. Too busy?
S. F.: Well I suppose it wasn't relevant.
M. G.: Yes.
S. F.: I was the one man who knew everything about LEOs, you know. One had to because every part of it broke every day you had to know.
M. G.: Of course.
S. F.: And it had all to be in your head. You know, how can it conceivably be. incidentally we were talking about amateurism, I mean nobody knew more about LEO II/1 when I went in because I'd been on the training course and the people who were running it didn't need to know a great deal about it. One of the best people was George Manley who later became in charge of the Commissioning Department and all sorts of things. I believe he was going to get a job in a bank and for some reason or other he couldn't start for a few months and so he got a job at Lyons on LEO and never went to the bank. He did day-release for HNC and HND. I can remember him redesigning parts. One of the great difficulties on LEO, or any computer, I suppose, is the synchronising the mechanical work of reading a card to the computer to the computer’s timing.
M. G.: OK.
S. F.: I can't remember what they did, but there were three stages controlled by three flip flops. Let's say the first flip flop was driven by the slow, the very pedestrian pulses coming out of the card reader, saying ‘I'm reading the first line’, ‘I'm reading the second line’, etc. And then there's a flip flop in the middle that you took some notice of that. And then there is a final flip flop which was all part of the synchronising with the timing system in the computer.
M. G.: OK.
S. F.: And that was sort of seen as extremely tough, difficult to make work. And I remember being very impressed by George Manley. He had redesigned and rebuilt a bit of it because it didn't work in the middle of the night, in the middle of a payroll programme for Fords. Anyway he was relatively uneducated but that didn't matter. You could do it by flair. You could either do it or you couldn't do it. You couldn't sort of get to the fault. Some people are good at tracing them. You had always to be prepared to change horses. You think you had an idea what was going wrong and you'd followed that through, and then you suddenly come across something you didn't expect, and that was the actual fault, not the one you thought it was. George Manley didn't have the Physics degree I had which I didn't need! [laughs]
M. G.: Of course.
S. F.: But do you know about the way in which four words were squeezed into the space of one in the delay system?
M. G.: Yes I think that It's to do with the time it takes to travel down the delay tube.
S. F.: I'm sorry?
M. G.: Is it the time it takes to travel down through the mercury effectively?
S. F.: Yes. But do you know that we squeezed four words into the space of one?
M. G.: So you were compressing the bits to take up less space?
S. F.: Yes. Because the system was nominally one micro second pulses, with one micro second spaces blank.
M. G.: Yes.
S. F.: But the one second pulses, which were run in the main computer, when they went into the storage system they were turned into quarter micro second pulses and four words of a quarter micro second pulses were interleaved into the space of one word of one micro second pulses.
M. G.: Okay.
S. F.: So you had to know what word you wanted to send to the store...
[Tape cuts off [01:33:39]]
[End]

Copyright
The video above may be subject to copyright of the original author and is present only on this website for non-profit educational purposes only. If you are the original copyright owner and wish to have the media removed from this website please contact us using the relevant contact email address listed on our contact page.