Digital audio of a recorded interview with John Daines who worked on LEO II, LEO III and System 4, having joined LEO Computers in 1961.

Transcript :

This copy has had discussions between JD & RH removed, 2 Oct 2015.
Subject: John Daines.
Interviewer: Ray Hennessy.
Date: April 26th 2012

Ray Hennessy:
This interview of John Daines has been recorded by The LEO Computers Society as part of an Oral History Project to document the earliest use of electronic computers in business applications. Note that any opinions expressed are those of the interviewee and not of the Society. Copyright of this interview in recorded form and in the transcript remains with The LEO Computers Society, at 2012. John, I wonder if you’d like to introduce yourself
John Daines
Hallo, I’m John Daines, I was born in Huddersfield in 1943 so I’m just coming up to being 69 years old. When I was born my father was a sergeant instructor in the Royal Signals based in Huddersfield and before the war he’d worked in the continental travel office of the London and North Eastern Railway. Because illness had lost him 2 years at grammar school he’d had to leave school after the equivalent of O levels because he would have been 20 before he did his A levels and I think that restriction still exists; you can’t be at a school after you’re 20 years old so he’d had to go out to work. My mother had been a clerk in the Bank of England before the war and she said “We had to balance to a penny before we were allowed to go home at night”. When my father went to Catterick[1] from Huddersfield, my mother and I went back to her parents in Herne Bay, Kent where my brother was born 2 years later. After the war my father went back to the LNER but left to join the “emergency training” scheme that had been set up by the London County Council to train new teachers and he was at Leavesden[2] for 18 months. Once he started teaching he progressed rapidly and after a few years he became head of a primary school in Holloway. Later in his career he ended up as head of a very large primary school right up by Highgate on the borders of Islington. My mother stayed at home to manage the family. In 1948 we moved to Holloway, north London. 
Ray Hennessy:
Well, that’s given your background, John, what about your schooling? Did you particularly shine in any particular subjects?
John Daines
I went to Hungerford Road primary school which I suppose was a half mile walk in the days when you walked to school; I was pretty well top or second in the school. I went from the first year in the juniors straight into the third year and then into the top year but, because my birthday was at the end of the year, I was very young so did the top year twice. I then went to Woolverstone Hall for two years. It was a boarding grammar school run by the LCC and then I changed to William Ellis School at Parliament Hill in London for the next 5 years. I had done languages – Latin and French for 5 years and German for 3 but didn’t really work hard enough at school. I just didn’t enjoy learning set books and didn’t quite see the point either. At the time we did English grammar as well as the grammar of other languages and that structuring and organisation interested me. When, 10 years later, I went on a crash course in Russian, at work, that was based on speech and headphones I was more interested in how it fitted together. However, after 50 odd years I can still remember some of these languages and when I speak French I strive to get the grammar correct.
4.17
I was really interested in physics at school and at William Ellis the senior science master was Mr Nelkon, who had written many textbooks. It must have been quite lucrative because he drove to school in a maroon Daimler Conquest, which was the smaller saloon that they made. His lessons were interesting with plenty of practical work. Outside school hours I was a lab assistant, which meant coming in early in the morning to set up experiments for the various lessons that would take place during the day. In the 6 th form John Ogborn arrived from Peterhouse and was also an inspirational physics teacher. He went on to teach at Marlborough and to become very senior in the
Institute of Physics. The school was a real hot house and several people in my year went on to great things; two or three are now professors. I got 8 O levels and did maths, physics and chemistry at A-level. I got too involved with non- academic activities and failed to work very hard so ended up with 2 passes that were insufficient for a conditional place I had at LSE. In early September 1961 I went back for a third year in the 6th form and tried economics. After a couple of weeks I decided that I’d had enough.
5.42
Ray Hennessy:
Sounds reasonably familiar, similar to a lot of what I did. How then did you decide what you were going to do after school?
John Daines
There was really little career advice at school. The assumption was that people would go to university. Some people left after O-level and were seen as failures while some others left after A-level but the aim really was Oxbridge. There was a large third year in the sixth form where people would go back to do the entrance exams but if you didn’t get to Oxbridge then other universities were a target. I had no real ambition having lived in a pretty closed environment at home and having little understanding of the real world outside. I don’t remember any parental advice or guidance on careers. However, after O levels had finished in 1959, and before the end of term, (people didn’t just end the term after the exams in those days, you went back to school), Mr Nelkon took us for a day out to
Cambridge, a coach full of us. We saw two things; firstly the radio telescope (which was an interferometer rather than the large dish type) and more importantly we saw EDSAC[3] in operation. It looked to me to be an interesting job.
7.13
Ray Hennessy:
So you didn’t wish you’d gone to Cambridge at that point, or perhaps you did. What did you do, how did you follow that on?
John Daines
When I decided I’d had enough at school I looked in our Sunday paper, The Observer, for a job. The only one that didn’t want a degree was with LEO Computers, so I wrote off and got an interview. I went to Hartree House[4] and in the morning attended an aptitude test run by John Smythson. It was the one about calculating the area of a room to be painted, ignoring the door. After lunch I went for an interview as an operator. One person (I forget his name but he was the man who wrote the Hudson’s Bay billing system that worked first time! I think he was quite a legend in the company) kept asking why I didn’t want to go to university and said that really I should. I insisted that I wanted to work. After that I went to Bill Steele who managed the II/5 bureau. One thing that sticks in my mind is that, I think it must have been Dick Halford, the Chief Operator, came in to say that they couldn’t read a tape from last week and that the grandfather was also not readable. Bill exploded and left the room but returned after a while. “Now, where were we?” he said. I must have been seen as having some potential and determination to succeed because I was offered a job as an operator in the LEO II/5 bureau at £8 10s. (£8.50) per week.
On Monday, October 23rd 1961 I arrived at Hartree House and, after being welcomed by Bill, was sent
down to Cadby Hall for a medical (Dr Blood, I think it was). And on the same day I joined was Godfrey Parry who went on to be a member then secretary of the [LEO Computers] Society.
Ray Hennessy:
He was known as Geoff Parry by then, of course. 
John Daines:
On return I went into the II/5 computer room and was put on shift, I think with Bob Stevenson. It all looked really interesting and people were reaching into the card reader to take out packs of cards. I thought I’d make myself useful and reached in for a pack. Unfortunately it slipped out of my hand and went over the floor. I think it was E56D the CAV daily payroll program and about 3” thick. I was not popular, as in “who’s that ginger haired idiot”. Not an auspicious start. However, as will become apparent, here was something I was really interested in and I worked hard accordingly.
Ray Hennessy:
So, you’re at Hartree House, working on the LEO II, II/5. What sort of work were you doing, tell us something about the daily grind you had.
10.30
John Daines
First of all I’ll talk about II/5 itself. It was a LEO II. The installation consisted of the main processor which had 64 tubes of mercury delay line store (about ½ a ton of it stored in something known as “the vault” somewhere under the false floor. [Technical Note. It was LEO I that had the store under the ‘stage’ on which it was built. The LEO
II store was in a large cabinet behind the power distribution cabinet to the left of the four rack rows] it had two Ferranti paper tape readers, two ICT[6] card readers (I haven’t seen this type anywhere else; cards read row by row, not column by column), there were two Powers Samastronic printers, one or perhaps two ICT card punches and eight Decca[7] magnetic tape decks (these were in 4 cabinets). It also had two Ferranti[8] drums, one that was used all the time for program storage and another (the large / fast drum) that was used for specific jobs like the six-monthly Smiths Clocks production scheduling that was dreaded by the engineers – we’d go in to the engineers’ room and say “We’ll be needing both drums this weekend, chaps”. By the time I joined LEO the bureau operation was running pretty smoothly. In one way I was there at an early stage in the life of computers but really all the original work had been done in terms of writing systems and producing standard routines, organising operations and having relevant documentation. For example, routines and procedures that allowed rapid restarts after any job failures, routines for inputting programmes and storing them on the drum, which I think were B15 and B37. There had been a team called “Job Installation”, that was set up to make sure that when new jobs came on they were in a standardised form and people like Mike Trevanion and Richard Beckingsale had been seconded to this group. They standardised the format of Operating Instructions, generally known as “op.ins”, and dictated the quality and standards for new jobs. A job being a suite of programs to be run on a regular basis.
12.51
These days data is largely input directly via screens, storage is on very large discs and the control and decision making is incorporated into system control language. In any case the machine and storage are quite possibly in an unattended dark site. The job of operator therefore barely exists these days so it may be worth talking about some of the elements involved all those years ago.
Ray Hennessy:
Yes, I think so. Give us a flavour of the operating environment and how you performed your daily task, or even evening task.
John Daines
Jobs consisted of a sequence of programs to be run on a daily, weekly, perhaps monthly or on an ad hoc basis. They consisted of the program cards, the current input, brought forward and carried forward files (usually on magnetic tape), pre-printed stationery, operating instructions and possibly some special instructions. A key job on the day shift was Job Assembly; i.e. to assemble the work to be done for the next 24 hours. This was done using the typical punched card trays of the time. They would take 2,000 cards normally. There would be a [card] tray for each job into which went any op.ins, paper tapes that arrived from data prep together with the (original) data and a suite flowchart that required the tape numbers for that run to be inserted. So, mag tape library work was very, very important and it was critical that it was done on an accurate basis. Main files were used on what was known as a grandfather cycle so that if this run’s input (the father) couldn’t be read it would be possible to rerun the last run, that’s last week’s or yesterday’s, using the grandfather to recreate a new father. Work for the day was stacked on a trolley or trolleys and this was added to by programmers who would come up from downstairs and submit trial runs of new programs or for modifications. Unless they were of great urgency, the trials were scheduled to be done after all the operational work. There might also be some work for Eagle Star Insurance. This was a card only system that processed private company pension schemes with annual renewals; most of these were small with few members and run on an as and when basis without the usual urgency for completion. All the LEO programs were kept on punched cards with each row on the card representing the 40 binary bits in a long word (remember they are read row by (using alternate columns). A memory tube of 16 long words needed two cards, the cards each had a check total for the card and there was check total for the tube. They were also numbered and a horizontal parity bit was also used. Program initiation into the machine was by simultaneously pressing the two “card start” buttons that
read the first bootstrap card that then read in the standard card input routine that loaded the program to store or drum as appropriate.
16.31
The current input for the job would come from the customer on forms and was punched onto paper tape in Derek Jolly’s data prep section. It was punched first on a punch machine and then punched again on a verifier that read the original tape in parallel. Any differences would cause a stop and the error was corrected. There might also be some “control data” submitted on card or paper tape. This would be a small amount of data specific to the run. Some customers submitted cards that were delivered to us and required sorting. We had two ICT card sorters that were used on a regular basis. For example the daily E155 job for Durlacher, Oldham, Maudant and Godson the City stockbrokers used details of each day’s “bargains”. They sent details of the day’s bargains in one or more 2,000 card trays each day that had to be sorted first of all on whether they’d bought the shares from a broker or sold the shares to a broker and then each group had to be sorted on a 3 digit broker code and 5 digit stock code. The main job couldn’t start on the computer until this was done so there was pressure on the operator, the equipment and the mechanics who had to keep them in good fettle. High activity in the city could cause large peaks of volume, an example being the ICI / Courtaulds takeover battle when there could be 3 trays that had to be sorted. One job that wasn’t welcome was the weekly Hotpoint invoicing [run] where the cards were IBM type in the days when their cards were physically thicker than the Hollerith cards that the sorters were set up for. On top of that there was an alphabetic sort that meant pre-sorting on the top three rows 12, 11, 0 before sorting on 0-9 to get them into alphabetic sequence. This used to be a problem because with the thicker cards they would often jam or be damaged and require re-punching and help from the engineers. One week Hotpoint just didn’t arrive; it had been transferred elsewhere. Another weekly job came from Lightning Zip Fasteners in Birmingham, production control I think. That required their cards to be collected from Red Star Parcels at Paddington station on a Friday night and the use of an ICT collator that merged two sets of their input cards after we’d sorted them. Saturday afternoon involved a dash back to Paddington with the boxes so that they caught the right train.
19.23
On LEO III cards were read in as they came in and sorted on magnetic tape; I don’t remember a tape sort in the LEO II bureau.
Ray Hennessy:
John, I wonder if you could tell us a bit about the magnetic tapes because they were quite new on computers in those days and very, very few systems had them. I don’t think any commercial systems had them.
John Daines
I think that from what I’ve learned since that there was an aim to try and get some magnetic tapes  from STC I think it was in the days of LEO I but that never happened.
Ray Hennessy:
They had a ¼ inch tape that was actually supplied but never used in earnest.
John Daines
The first LEO II that had magnetic tape was, in fact, II/5 and that was used for the main files. It was ½ inch tape and it used 3600 feet reels and it was 1 mil thick. That contrasts with all the tapes that were used on LEO III and subsequent machines and other manufacturers which were 1 ½ mil tape that had 2400 feet on them. The reel sides, the sides of the reels, were open and made of relatively flimsy metal, I guess aluminium so that it wasn’t magnetic, in contrast to the later 2400 foot tapes that had rigid aluminium or plastic sides. The tape decks came from Decca and each unit consisted of a pair of decks one above the other and the top one physically behind the lower one. So, each deck had two reels, a left hand reel and a right hand reel, with heads in the middle (in between). There were four heads: when the tape moved in its normal direction there was an erase head then a read head then a write head then a read head. The tapes were driven using air bearings. So that either side of the heads there was a spinning wheel with slots in it and it was normally pressurised so that the tape sat above it and didn’t move. To move the tape it was sucked down onto the fast spinning wheel and thus rapidly accelerated. There
were vacuum reservoirs on either side of these so that only the mass of the tape was accelerated and the spool motors were only activated to feed into or empty the reservoirs as necessary. These reservoirs were about two to three feet long and went downwards which is why one deck sat behind the other so that the reservoirs of the one at the back went down behind the lower deck. This was quite advanced or very advanced stuff and I didn’t see decks like this with air bearings again until the 4453 high speed decks on System 4 in 1968.
22.43
The start of the tape and the end were defined by what was known as a window. They would be delivered with a window in but if it got damaged and you had to re-create one of these it was done by holding the tape tightly over a matchbox and removing the oxide with a solvent. Before a tape was used for the very first time it had to have Block Start Markers (BSM’s) written. This wasn’t done by the computer, it was a function of the mag tape controller. You would put the tape up and put it into “block start marker” mode. So, these markers started being written from the end of the tape back to the beginning. The tape would be run forward to the window at the end and then coming backwards it would write the block start markers. The distance between the markers was two fixed
5 length blocks plus one inter-block gap. All the mag tape blocks were 1 tube of store i.e. 16 long words. The tape was then created with these block start markers so when you actually wanted to write on it, tapes would be written starting at the beginning of the tape and it would find the first block start marker and write a block and then it would stop and then it would start up and go to the next block start marker and write the next block. And when it was going that way it would use the erase head, the write head and the read head. The read head was to do a read after write check to ensure that it had been written correctly. When it got to the end of the tape and detected the
window the direction of the tape was reversed, the erase head switched off and it continued to write and it was known as “writing round the bend”. So, this time the blocks again start at a BSM, this time using the write and the other read head to do a read after write. So, it would come back down the tape writing into the large gaps that had been left but because there was an interblock gap there was time to stop and restart the tape before writing the next block. It wasn’t unusual to see a sticky label on a deck saying “won’t write round the bend” because there were problems writing round the bend and it only happened on very large jobs that had enough data to require it.
On one occasion I remember, the erase head didn’t switch off, or we realised later that’s what had happened, so that a lengthy job wrote the first part of the tape and ran to end but with the last part (round the bend) utilising write, read after write (successful) and then, because the erase head was left on, it was then erasing all that had been written and all we had was an empty tape so that when we came to use it as an input to the next job; hey ho, we needed another rerun.
Ray Hennessy:
I seem to remember that might have happened on one of my jobs.
John Daines
26.45
Because the machine (i.e. the computer system) could be unreliable or other errors could cause failure, for whatever reason, jobs were split into what were known as “Restart Groups”, typically 20 of them. A job would run and as each group was completed a “Restart Mark” was written to the tape and a “Restart Card” containing some key totals was punched on the card punch. Should a job fail the programme was reloaded and the last restart card was fed in. The tapes would then run back to the appropriate point, the current input was realigned and the job could continue. If the tapes had been removed it was possible to rerun a group and the tapes would run forward to the right position before starting the actual program operation. It was a basic assumption that something
could or would go wrong during a run but there were sufficient facilities to ensure that it was taken in its stride. And this was an important thing; all the lessons that had been learnt from LEO I meant that facilities were built in for engineers and that operations assumed that restarts could be done. Magnetic tapes had block serial numbers and we couldn’t have done without them. The block serial number was a binary number stored in each half word of
a long word. There was a clever instruction on LEO II that would negatively add a series of long words in a tube to produce a “check total”, a negative number that when added to the positive sum of the original words produced zero. Before a block was written, long words 0 to 28 were negatively added and the block serial number was further
subtracted and the result then stored in word 30[9]. (The serial number was actually added to the negative sum so that when the block was summed on reading, the resulting number was the serial number of the block but John described the essential principle of the system whereby a parity check and a block serial number were both controlled by the use of the last long word in a block.) On input the whole block was negatively added but it should be “out of balance” by the block serial number in each of the half words: a double check. Sounds complicated but it was very elegant and effective. Operators became adept at working with the engineers to recover faulty blocks, perhaps by changing decks; or in some cases the engineers could adjust the “potentiometers” reading a track to try and get a block to read successfully. So, you could look at the store monitor and see bits creeping in and out as they adjusted it. And then if we got the correct block serial number we could accept it and carry on.
29:29
Ray Hennessy:
Before you go on, John, perhaps you could have a short word about the causes of failure. You’ve mentioned failures once or twice. What sort of failures? Was it hardware or what?
John Daines
It could be a fault on a mag tape; every so often a tape might get stretched. It could be a fault on input; although all the input data had been verified, every so often a digit that was far too large would be punched in and because the updates were driven by the current data, it would see a very large number on a current input data record and it would run the main files forward perhaps much further than they should have done, which would then cause a failure so it was necessary to stop the run, correct the input data and restart it. It could be that the paper had come off the tractors and we had lost some printout, it could be that something had jammed so there were hardware problems, there might be some mis-operation, there could be problems with media.
Ray Hennessy:
What about hardware failures, were they frequent?
John Daines
By and large the processor stayed on and ran reliably. The engineers would have it for maintenance each day. They would run test programs. The machine was built with a marginal facility which has been discussed elsewhere but they could turn the voltages up and down to try and find what the limits were and whether there were valves that were likely to be on their way out. So, I can remember Pat Blakes coming round and running the store test programs and there was a lid on top of the operator console that you could lift up and adjust the voltage margins on that. So, the processor was pretty good; I don’t think we had many store faults. It was normally mag tape and
media problems.
Ray Hennessy:
So they were basically operational difficulties rather than hardware? This must have been quite an advance over some of the earlier equipment. I think LEO I had a fairly high hardware failure rate.
John Daines
We did get failures but they were more with the peripheral equipment and the media, I think.
Ray Hennessy:
That was quite impressive at that early stage in development.
John Daines
Yes. The drum – I was thinking about it when preparing for the interview – that there was this mythical fast drum but the drum we used from day-to-day again ran pretty reliably. All the programs were stored on the drum because there was such a small main store. The card programs and sub-routines were read in and kept on the drum and then, because the programs were written in machine code, the programmer had to call overlays back in to store, so the drum was in constant use but I seem to remember those being fairly reliable.
Ray Hennessy:
As a programmer we had to provide, as a standard, restart points, I think every 20 minutes; or that was the theory. It would obviously depend on the length of the run. A 30 minute run might not produce it but it seemed to be that 
the feeling was there were frequent machine errors so there were likely to be numerous restarts in a long run of several hours. But you’re saying it was actually a lot easier than that, as far as the hardware went?
John Daines
Yes. The work had been done on the assumption that there would be problems and that, if there were problems, that there was sufficient resilience to get things going again quickly. But very often you’d do a run and at the end of it you might have a pile of restart cards you hadn’t used. On some of the tricky jobs, I think the printing could cause problems and I’ll talk about the Samastronic printers and some of the things that could happen there.
34.24
Ray Hennessy:
Tell us about the printers, because they were quite novel.
John Daines
Yes, hmm. The customer output was usually on paper which might be plain paper, which was known as music
paper; blank one side and the other side with bars going across, or perhaps pre-printed, either invoices or payslips. There could be a requirement for up to 4 copies. We had some printers that came from Powers[10] called Samastronic printers and they were a challenge both to the operators and the engineers and from what I’ve heard all other customers who used them as well. They could print 300 lines a minute across 140 columns. The paper was moved forward by tractors and the page length was defined by a “throw wheel”, a gear wheel with one tooth per 1/6 inch of paper i.e. one line-feed and a metal pin fitted through it to define the start of a page. As the paper moved, this wheel would move round. So, the standard throw wheel was 66 teeth and one pin, so that was for standard 11 inch paper. A 5 ½ inch form had 66 teeth and two pins. There was a command to “throw to head of form” so that instead if incrementally moving the tractors you said throw to head of form and it just released the tractors to go at full speed until this pin struck a stop. So that was paper movement. Above the paper there was a horizontal bar with 140 stylus points across the 14 inches; so that was a 1/10” wide character. Each of these stylus points was at the end of a Bowden cable with a solenoid at the top end, and it (the bar) oscillated above the moving paper. Above the printer there was this large print head that lifted up and there were 140 solenoids driving 140 Bowden cables each of which sat in this bar that went at right angles to the paper movement. Thus, for each 1/6 inch paper movement, or one line forward, the stylus, because it was oscillating, traced out a sine wave over the area that was 1/6” by 1/10”so that when one of the solenoids was fired it printed a dot at a known point on the sine wave. Each character was defined as a series of dots at appropriate points on the sine wave. There were 64 characters consisting of capital letters, numbers and symbols. Character design was critical as will become apparent, for example 5 and 6 had a long tail and a long top respectively so the six had a long piece coming down and then the loop at the bottom; the five came down and then a long tail. And there was also a special character known as the Lenaerts test character which had been put together by Ernest Lenaerts. That was used in printer tests and the engineers used it to set up because it used the extremes of what could be done. Now, the way that the character printing was implemented was as follows. If you imagine the sine wave being effectively stretched into a straight line and then physically represented as the circumference of a paxolin disc then on the circumference of the disc at each point on the sine wave there was a metal contact. There were 64 discs, [one per character], they were all stored one above the other in what was known as “the stack” that spun at speed and it had a set of brushes resting on each disc to detect the contacts. As it went round, if you wanted to print an “A” every contact on the
circumference (for the letter A) would be picked up by the brushes and routed to the appropriate solenoid.
Ray Hennessy:
Which might be all the solenoids?
John Daines
It might be all of them, yes. Exciting stuff, and all very noisy, especially when the covers were off! The engineers often had to “clean the stack” if things weren’t looking too good and perhaps they would get some rags and
solvent and actually clean this thing whilst it was running. They really were so noisy. In other words it was a dot matrix printer writ large; but this was in 1961, remember. Multiple copies were required and they were either by using NCR[11] forms or narrow NCR listing paper. For example Hotpoint had 5-part NCR invoices and at that time they cost five old pence each so we weren’t encouraged to have too many reruns with those as it would cost a small fortune. More usually we used carbon paper. This wasn’t carbon paper that came with the paper. We used plain paper and there was a device called a carbon creep. The paper was fed from a unit that had 5 shelves for paper and, at the top of it, there were 4 rolls of carbon paper 14” wide. For each line of paper movement the carbon reels moved by a fraction of an inch, so they really got their money’s worth out of the carbon paper. Loading paper for a job that had perhaps 1 + 2 copies involved a layer of paper off the top shelf, then a carbon coming off one of these reels, another sheet of paper from the next shelf down, another piece of carbon then another piece of paper from the next shelf down so that this sandwich would be fed through. All the creases ( i.e. the perforated joins between
the sheets) had to match correctly, and the carbon fed correctly, and the pressure brushes adjusted accordingly. These brushes were to keep the paper steady. The tractors were well ahead of the print station so that it was quite a job to maintain consistent paper speed when it moved. We could get problems of “cramping” where the paper didn’t move fast enough and at the extreme you might get “upside down”. If you’d got very heavy paper with 3 or 4 copies then if you threw to head of form, not having printed much on the previous page, the paper would be moving at such a speed that when everything stopped, and remember that this is before the tractors, if it was an upward crease it would move upwards and as it settled back again the paper could actually be moving backwards when the print occurred. So these things, you really had to watch them quite carefully depending on what you were running.
Ray Hennessy:
I suppose you got used to particular types of runs that led to this type of problem.
John Daines
That’s right. But if you ended up with some cramping or bad printing, it could be an engineering problem or whatever but that would cause a restart and you’d have to go back. If the carbon creep stopped working, if you were running jobs with copies, you had to keep checking, tear out the carbon that had been used, you had to keep checking that it hadn’t skewed and that all the columns were being printed on the bottom copy and so on. One thing that was done, for all the pre-printed stationery, there were test prints so that before you ran the job the programs would put in a test print that printed 5 5’s if there should be 5 figures in a box so that you could check a) that all the columns were there and the right number of characters were being printed and b) that the paper was lined up properly. And this was important because there were things known as plug boards on the front where you had to plug up from where the data was coming out of the machine and which columns it went to; so it was partly a check that you had the right plug boards in for the job and that there weren’t any bad contacts there.
Ray Hennessy:
That was to save the amount of data stored in the system, so you wouldn’t be storing all the spaces.
John Daines
There was also a facility on these printers to have a second sheet of paper being printed simultaneously. There was another set of tractors but these could only be driven by line feeding. These tractors were physically above the other ones so if, for example, you were running a payroll program it would have pre-printed payslips which might be 6” wide driven by the primary tractors and throw wheel whilst a report listing, which might be an A4 sized piece of paper, which might be NCR, that was printed alongside using the secondary tractors. The report would contain exceptions or group totals that often required checking by the operators to see if there were balance errors. So
that at the end of a restart group or perhaps at the end of a department in a payroll there would be some totals printed and there would be a number that should be zero to check that two ways of adding it had come together.
Ray Hennessy:
So this was an operator function at run-time?
John Daines
To check that? Yes, yes.
Ray Hennessy:
Quite a lot of responsibility then?
John Daines
Yes. Because the printed output was the key output from the bureau, quality of print was very important. And of course having a great pile of multi-part paper, it then had to be decollated. Now we didn’t have machine decollators in those days so you’d have to put it down and do it by hand.
Ray Hennessy:
Physically do that? Now what about the other output devices, there were some weren’t there, input and output devices?
John Daines
47.08
All the slow devices, that’s printers, paper tape and cards, were connected to the machine by switchable channels. So you had input channel 1 that might be used for cards or paper tape and the output channels could be printer or card punch. Each channel had a hardware buffer. So, if you put a pack of cards into a card reader and raised the key to connect it to the machine, a card was automatically read into that buffer ready for a program to read it. This gave double-buffering in that you didn’t physically have to wait for the device.
Ray Hennessy:
What do you mean by a buffer?
John Daines
It was another piece of mercury delay line store. The one for a card reader would hold all the information held on that card, which was 12 rows of 80 bits. If it was paper tape then it would be a pretty large paper tape block. If it was a line printer it was one line of print. If it was a card punch then again it was a full card
Ray Hennessy:
This was a way of getting parallel operation, really?
John Daines
It was to speed it up so that when you (i.e. the program)
said “read” you didn’t have to wait for the device to start up. So that you could do some processing so that by the time that you’d finished processing one man you’d come back for the next record and the card should probably be there because it had been fed in automatically. On later machines there weren’t hardware buffers like this
but this was probably an early implementation of using buffering. In the [same] way that cards were automatically read in, an instruction to print a line didn’t go directly to the printer, it put it into the print buffer but it didn’t actually print until the device was connected. This caused a problem oncewhen a long job, which had only printed one line, and the printer hadn’t been left open so the line was lost, that line of print was lost when the machine was reset for the next job. So that caused a lengthy rerun. A change was made to ensure that a minimum of two lines was always printed by a program. Should a buffer be empty on input or full on output when a program made a request for input or output, the machine would “howl” to notify the operator. So, our line of print that was sitting in the buffer, we would know about because the next print instruction would cause a howling noise through the loudspeaker. The machine had a loudspeaker and this was really useful for the operators, who knew the sounds that particular jobs made. And, by the way, there was a juke-box program that was sometimes run in rare slack moments. There was no output typewriter or operating system so the operator had to keep a handwritten log showing the start and end of each job, any stops that came up and the actions taken. The operating instructions would have a list of all the different [stops]. There was a stop order which was zero and the number of the stop order could be any (pair of binary) numbers between 0/0 and 63/31[12] and depending on the complexity of the job there could be a large number of stops. Now, the standard initial stop when a program was loaded was 63/31 and then you would start it running and at the end the standard halt was 63/29. All this above may seem very technical and hardware orientated but these were the days before operating systems, when the programs were written in machine code; there was no higher level language really until Intercode[13] came along on LEO III. We lived with this equipment day by day. It was a hardware orientated job because and I won’t say we were fighting it but we were coaxing and nursing it along and by and large it did the job. The amount of work we did was pretty phenomenal. When I started at Hartree House the bureau also used the British Oxygen Gases (known as BOG) machine which was LEO II/7 in Edmonton to provide additional overnight capacity. I think they changed their name to The British Oxygen Company because BOC sounded better than BOG, but it was still known as BOG. II/5 ran 24 hours a day 7 days a week. That is, it was switched on for all that time so there was a shift system, when I started, to cover 7 days, 7 evenings, 7 nights and then 7 days off. The night shift covered British Oxygen on 5 nights and it was expected that work could possibly run out at Hartree House during the latter part of the weekend. At that time, when I started, there was a fixed shift allowance of £7 15 shillings (£7.75) a week, so it was a substantial boost to my wage although less so to people with a higher basic salary or wage. One thing that was quite good about the way that Bill (Steele) ran the bureau was that, over time, as different shift arrangements were required, Bill would allow a great deal of letting people define the detail that would provide the cover. So he would say “We need to cover these shifts” and if all the operators could get together and say “well, if we work in that pattern we can do it”, and people were happy with that, then it worked well and again people felt that they were being involved.
Ray Hennessy:
Probably quite important in those days. I think that was possibly a bit of the Lyons ethos as well. Perhaps you can talk through some of the jobs you were running in those days, if you can remember them. There must have been quite a lot going through the bureau.
John Daines
There were, and when I started to think about it I’ve obviously forgotten a lot because the machine did run 24 hours a day.
53.55
Most of the jobs started with “E” which could be that they were for external customers. I think the Cadby Hall jobs were “L”, which I guess was for Lyons. A regular job was E56, a company called CAV[14], which was part of the Lucas group, I think. They were a large manufacturing company making injection equipment for diesel engines. They had a daily payroll and weekly and monthly salaries. They also sometimes did some heat transfer calculations, what Bill used to call “bubbles up a spout”. Those programs would sit there grunting and groaning and printing out occasionally. I think possibly John Parker who did a lot of mathematical stuff may have programmed those. The key programmer on the payroll when I was there was Kate Keen. She’d started in data prep and data control on LEO I at Cadby Hall and she later went on to programmer training at Radley House[15] in Ealing. She was an example of the possibilities that were open to people who started in quite mundane jobs. There were several of those people at Cadby Hall who went on and did great things later. CAV later bought a LEO III, machine III/5, and they had it on display. You could see it from the main road through Acton when it was running. Another regular job, and I mentioned it earlier, was E155 for Durlachers, the stock jobbers. They were the largest jobbers on the Stock Exchange. This job was really critical because if the previous day’s results were not available at the Stock Exchange by 9 a.m. the following morning there was absolute hell to pay. The jobbers bought shares from brokers and sold them to brokers and settlement was due the following day. A driver called at Hartree House each morning on his way to the city and if the job wasn’t finished he didn’t wait. That could mean an urgent drive across London on a very few occasions we didn’t make it. They again bought a LEO III later, III/45. E29 and a whole lot of others, we did payroll each week for 5 of the London Boroughs. We did two or three one night and the others on the following night. This involved payslips and job costing. They were the London Boroughs Joint Computer Committee and bought LEO III/4. Some jobs doing production scheduling and production control that were run at a variety of intervals were E142 for Lightning Fasteners which is the one, as I said earlier, where we used to take the cards from and to Paddington station. Every 6 months we had E131, a job for Smiths Clocks in Wishaw, Lanarkshire that required both the drums. Perkins Diesel Engines from Peterborough we did a production job for, E177 I think that was.
57.38
Now, another job of interest was E89 Hudson’s Bay Co. Twice a year they held a fur auction in London (Persian Lamb, beaver, etc) and we would get the data in and by the next morning all the invoices were ready to go. Only run twice a year and was notable really for the fact that the chap who wrote the programs, they put it on and it worked first time. A tribute a) to accurate programming and b) the LEO method that said that all programs were carefully desk checked before being used. Another monthly job was for Glyn Mills Bank, E41. They had the job of processing the pay and I’m not sure if it was for either half the Army officers and all the RAF officers or vice versa. They were early users of the bank code file. There is a standard issued file of branch codes and addresses. And
it (the program suite) produced credit transfer forms as well as the payslips. These results had to be delivered to Osterley on the Saturday afternoon. We started the job on Friday night and delivered on the Saturday afternoon. This involved a drive along the A4 and at that time involved snaking through the all the construction work because that was when the overhead sections of the future M4 (motorway) were being built. Later on there was a Standard Payroll introduced as a standard package that was sold to quite a lot of people, including the British Printing Corporation, I think. That was used by several customers. There was Eagle Star, E120, and program trials. The thing about Eagle Star and the trials was that each one of these involved a lot of job set-up for what could be a short run. So there was all the setting up work: loading tapes, cards and so on for a job that perhaps would last for a minute or two and then all the work of putting it away. So these weren’t very popular and seen as a lot of work with little to see or do and, with Eagle Star, you could run several of these schemes through which might have 10, 20 or 50 cards output and they all had to be separated, rubber banded, marked up and so on. But, I enjoyed the work so a few times I’d stay on at weekends to complete work like that.
Ray Hennessy:
So you were one of the volunteers that Bill Steele felt was reorganising his operating to your own satisfaction?
John Daines
I suppose so, yes. I’ve mentioned travel. When I started we had an Austin Cambridge and this was for taking stuff out to British Oxygen but this was replaced by a green J2 van 647 PKT with a sliding door. We used it for the nightly runs to BOG and for other ad hoc deliveries. These vehicles belonged to Normand, which was another part of the Lyons empire. When they required servicing we had to take them into Marble Arch where Normand had a garage and service centre.
1.01.25
Each machine, and all the LEO II’s, were maintained by teams of engineers and mechanics. So, we had three shifts of engineers and mechanics. Some of the chief engineers were Stan Evans, who went to Honeywell, Ray Dowson, Ken Lanham and chief mechanic was Jim Rolfe. Lots of other engineers as well and, again, it had to be a team effort. We were very dependent on them to get the work done. The machine, II/5 was really only switched off 4 times a year for bank holidays, that was Christmas, Easter, Whitsun and August Bank Holiday. This was usually quite traumatic because otherwise the machine was switched on, and was used to being switched on for 24/7. At the end of the holiday the engineers would come in especially early and if we got the machine back by 10 or 11 o’clock we counted ourselves lucky because quite often, switching it back on would cause faults to come on. If it ever went off at other times we knew it could be a big outage. It didn’t happen very often but I must have been a bit traumatised because for years after I always expected trouble if a machine was switched off. It was just in the back of my mind. Interestingly, when II/7 was transferred back to Hartree House from British Oxygen, it took a long time for that machine to get used to staying on all the time; because when it was down at Edmonton it was used to being turned
off quite frequently.
Ray Hennessy:
Well, I think that’s given us quite a good overview of the operations in general on the LEO II’s. How did you get on? Was your career developing during this time?
John Daines
Yes, I think so; I seemed to progress quite well. At that time I think they did three monthly pay reviews, probably a Lyons thing so I kept getting these pay increments of fifteen shillings (75p) a week and moved up and in the end I became a shift leader. In autumn 1963 I went to Kidsgrove to join the commissioning operators team that Bob Elmer was setting up on KDF9[16]. He’d got his main team in Minerva Road[17]. After 2 or 3 weeks I was recalled to II/5 as someone had left and they needed me back. In February 1964 I went to join Bob’s team in Minerva Road to work on LEO III.
1.04.18
I really loved working on II/5 and I could get back in the swing of operating it now, after a day or two I think. I don’t think that’s too fanciful thinking. It was interesting work and there were some good people there. I always thought Bill was a good manager and I found out later that he always took personal responsibility for the output from his operation if a customer should complain. He would say “I got it wrong”. He would then investigate and berate the staff, as only Bill could, as appropriate but he was square shouldered unlike many of the more modern slopey shouldered variety that direct criticism downwards and blame their own staff. So, we move on to when I went to what was known as “the factory” in Feb 64.
Ray Hennessy:
The factory, was that at Minerva Road, Acton?
John Daines
Yes. When I arrived there, almost in the first week I went to George Manley’s leaving do. George had been a very long serving person and was going off to Honeywell, who at that time were just setting up operations in Europe and recruiting engineers from all over the place. I had only just joined his team and he was leaving the company but he talked to me, asking about me and generally being very friendly; this was in the pub. I never forgot that and it must have been one of the reasons he did so well in life. When he rejoined the company after many years at Honeywell I bumped into him and he was just the same. Tony Morgan took over from George and I was under the wing of Allan Lyon, one of Bob Elmer’s team leaders. Minerva Road by that stage was commissioning systems that were actually built in Kidsgrove as all the wiring and board production had been transferred there. English Electric had done that. All the processors and device assemblers were delivered as individual uncovered cabinets together with peripheral units. Each system had a senior commissioning engineer whose brief was to ensure that the system was assembled (i.e. the cabinets were positioned, cabled up and connected to peripherals, etc), they were commissioned, acceptance tested, broken down and delivered to site for reassembly, re- commissioning and final acceptance testing for the customer. And there was one person in charge of that. Quite a task and they were very special people. Each one was supported by junior staff and peripheral mechanics together with key members of the maintenance team who would be the site engineers for that system. That way the field engineers would know their system before they had to maintain it. A team of commissioning operators was assigned to each system to do the acceptance testing and to help the engineers to diagnose problems. The commissioning engineers didn’t want to hear “it’s gone wrong” from the operator, he wanted to know what had gone wrong and where, so this was no
ordinary operating job. I think one other thing with having the field engineers in during commissioning, any changes or modifications were put onto that machine’s drawings, and those are the drawings that went with the machine when it was out there. An acceptance test consisted of first of all running the basic engineering tests but then followed by a series of programs running under the Master Routine, which was the operating system on LEO III. These jobs running under the master routine fell into two groups. Firstly a series of test programmes that exercised magnetic tapes, checked compatibility by writing a full tape on each deck and reading it back on that and every other deck, that tested the store, did some arithmetic or tested the line printers and some of these would be running in parallel with the other work. Secondly there was a commercial suite that had been adapted from a real job known as “Home Grown Fruits” which was run on the III/1 bureau. That did a job for some Kentish fruit growers. It had been adapted to have some checks and balances to aid fault diagnosis. There were 5 large reels of paper tape input and they all had serial numbers and possibly a check sum for each block. All the trials we did were based on the requirements of the government’s Treasury TSU or Technical Support Unit, which was a technical unit of seconded Post Office engineers who were tasked with accepting machines into government service under the contract which was known as CON84 at that time. A trial would last for several days largely dependent on the number of tape decks that dictated the size of the compatibility matrix. So, obviously, with twelve tape decks or more on a machine there would be 12 tapes to write and 144 to read back. The view was taken that if this trial was
suitable for the government then it should be suitable for all other users. Customers who wanted to use their own systems for testing were advised that their systems weren’t really designed for testing and probably were unproven. We would run our standard trials and then we’d be prepared to use their run as a demonstration to compare with their previous two runs of the program(s) they’d run twice elsewhere to produce identical results. “Yes, we will run it but you need to run it twice and give us the results.” This negotiation was usually successful once customers realised that we weren’t trying to pull the wool over their eyes and they would see their work demonstrated.
1.10.37
Under Allan Lyon’s tuition I started with Intercode, the LEO III language that was at a higher level than machine code but actually permitted the use of direct machine code because there was a block of instructions numbered 100 to 131 which corresponded to the machine instructions 0 to 31. Machine time for translation and testing programs had to begged and borrowed from the engineers who’d got a machine in an advanced state and running the Master Routine (typically whilst they were at lunch). One week I went with some other operators on a special “engineering for operators” course that had been set up and run by Ken Abbott. It explained basic logic, what basic gates, ANDs, ORs and flip- flops were, how the machine fitted together and how to use logic diagrams. LEO III and the input / output assemblers all used positive logic so that a high input produced a high output meaning that it was relatively easy to follow what was happening. All gates and flip-flops had a lead on each input and output that went from there to an engineering test point in the package handle. You’d got a printed circuit board with a blue plastic handle on the front for pulling it in and out and just behind would be a dozen or so little metal contacts that you could stick an oscilloscope probe or paper clip into. There were also two lights as well so that if there were a flip-flop or two flip-flops on a board they would show the state of each. So, following through logic with an oscilloscope and a test probe was fairly straightforward. I’m told that it was slow to do it that way i.e. the machine was a bit slower but it was really good for engineers. Now the faster machines, III/F, used Nand and Nor logic and I just couldn’t follow that. This was a really useful week and enabled sensible conversations with the engineers. Allan Lyon also introduced me to the microprograms that implemented each machine code instruction. Each machine code instruction was defined by a large flowchart which was a breakdown of the actions at the very lowest level that caused that action to function. The first machine being commissioned that I was involved with was machine 16 for Kayser Bondor, a firm that used to make stockings and ladies’ lingerie. They were at Baldock in Hertfordshire. I used machine III/17, Manchester Corporation’s machine, for some program testing; it was a machine I met
up with again many years later. I was then in the team on III/90, the first machine for the Post Office at Charles House[18]. Now, the Post Office had ordered seven LEO 326 machines, the fast machine or III/F but as an interim they had III/90 which was a very large basic LEO III. It had a large number of tape decks and a couple of printers. After that I worked with Alan Potter, who I suppose was the top and cleverest commissioning engineer, on III/23 which was the first III/F to be delivered. In fact it was a 360, which meant that it had a 6 microsecond store; the 326 had a 2 ½ microsecond store, and that was going to Dunlop’s to join their existing machine, III/3 in Fort Dunlop, Birmingham. This was a pretty traumatic time as there were still some design and transistor problems in the processor. Eventually it passed its factory trial, was delivered and re-commissioned ready for start of the (customer acceptance) trial on a Monday morning. On the Monday morning we arrived but were told that the trial couldn’t start because of an awful stench in the machine room. The floor was taken up eventually and, I’m not sure if they actually found anything but presumed that it was a dead rat or cat that had found its way through from the tyre factory which was adjacent to the main office building. The next day we started and achieved a successful test and handover to the customer. I next moved on to III/26 which was the first III/F for the Post Office, an enormous machine with 12 or perhaps more high density mag tapes and multiple printers. It took ages to get it right, again with transistor and noise problems.
Ray Hennessy:
That’s electrical noise?
John Daines
Yes. For this machine I had to double the size of the Home Grown Fruits data to 10 reels and adjusted the suite accordingly. Multiple copies of the compatibility program were required to support multiple concurrent magnetic tape channel operation.
1.16.32
During 1964 Bob Elmer left and Allan Lyon took over as chief operator. Then Allan Lyon went and I was the chief operator. I had, as key team leaders, Brian Norris and Bruce Elkington. Brian subsequently went independent and Bruce Elkington went and worked with the Post Office account for many years. There were some terrific commissioning engineers there. Alan Potter, who’d been to Cambridge and was just brilliant. Dave White, was an older engineer and as he said once, “to think I fought a war to save people like you and Brian Norris”. I first saw Dave testing boards; he’d lift each transistor out of its seating (because they were soldered in and then folded over into a seat), plug the board into an extension board, plug it in, start the test program and then go round flicking each transistor in turn until the test failed. [Technical Note. This was a problem with batches of the GET872 transistor and all machines which might have them were inspected, including those in the field, and the transisors
changed at the manufacturer’s expense] They’d know a board was faulty because they’d rattle a ruler along the handles until they could tap a board and know that there was a dry joint on it and then go on to flick the transistors. He really was a wise old bird and such a nice man. There was Dave Springle, a younger engineer and very clever. Derek Bensted, who’d started as an apprentice, Ray Boughtflower, who’d been at the original Olaf Street[19] factory way, way back, Jack Thompson, Brian Elsey. On the mechanical side were people like Brian Mills. It was really such a pleasure to work with such people. As new machines came in so did new devices. That meant that we needed new test programmes and also the master routine needed to be altered to support the devices. The Master Routine programmers had to come to us to test their changes. Among the new devices were some postal order readers for the Post Office. They were to be sited at the postal order processing centre in Chesterfield and linked by data transmission lines to a 326 in London. The documents were read online like cards. To test them they needed typically used postal orders so we were given samples of blank postal order paper to fold in particular ways, tip coffee on, carry around in pockets in the way that they would typically receive postal orders. They were then all
collected together and used for the tests, together with new ones, of course. Quite clever this, again, in 1964 or 1965 directly reading from 200 miles away postal orders that were being read in and then
either accepted or rejected over a transmission line. Other devices included the Swedish Uptime [American. I believe] card reader for the Post Office. The Post Office used to produce a 40 column round-holed card known as “trunk tickets” for every trunk call that was made, because in those days there was no STD (subscriber trunk dialling). It would read them at 2,000 per minute, selecting and either accepting or rejecting cards and it had an accepted or rejected pocket for each card. So it would read the card and there were so many milliseconds to
make the decision whether the card was good, and therefore accepted, or it whether it was positively rejected. It was a gem of a machine and worked on a simple but very effective principle. In use, in telephone billing they’d wheel in a trolley with about 26 trays typically of 2,000 cards for a telephone area so they were well used. They’d read them all in, sort them and process them. We had a special version of the ICT card reader, which was normally for 80 column square-holed cards, [for Czechoslovakia] that read their 45 column, round-holed cards which seemed to be made of compressed Bronco toilet paper and probably reflected the shortages in eastern Europe at that
time. There was also the Autolector that read bar-coded documents online for Lyons and HM Dockyards. It’s probably worth a word about Autolector and the Lector. The original Lector was an off-line device that read bar-coded documents and produced paper tape that was then read into a LEO III. They were fairly widely sold including into Eastern Europe. But then there was a requirement to read more documents, and faster, so the Autolector was used and this came from a firm called Parnell. I think on Tony Morgan’s tape he talks about it a bit and where it came from. Basically, it would read foolscap documents at a terrific speed directly and the way it was used for example at Lyons was: the delivery people would go out and take orders just by making bar marks on the sheets and they’d bring them back in, they’d be collected together, read in by the Autolector which would read, I don’t
know, I suppose perhaps 100 documents per minute; it was pretty fast. The processing was done, a (magnetic)
tape was produced and at Lyons they had a Xeronic printer so this tape would off-line print, onto the Xeronic, the new blank forms for the delivery roundsman to use next week. So the computer system produced the input documents and read the input documents in a loop that went on like that. They were used by Lyons but also in the Four Dockyards that there were at that time, which were Plymouth-Devonport, Portsmouth, Chatham and Rosyth. I think only Rosyth and Plymouth are there these days.
1.23.51
Ray Hennessy:
John, I wonder if you could give us some indication of how the work you were doing, particularly the shift work, affected your private life, your home life? I think people would be interested in how people coped with this sort of rather extreme time-keeping requirements.
John Daines
I lived at home with my parents in Elstree until I married in 1966. The time when I was on shift on LEO II I was living at home and I think it caused confusion for my parents because I’d wake up in the evening effectively wanting breakfast and there would be her evening meal that she had produced for my father because he’d come back from school. I’d stopped shifts but still worked long hours and then in 1966 I got married and moved to a 40 year old terraced house in Bushey, not the posh end at the top but the bottom end down by Watford. And that really covered my LEO experience. In January 1968 I went to Winsford[20] and we moved into a new detached house in Tarvin, 6 miles east of Chester, after Easter. I still live there after 44 years now. We had two children, a son in 1971 and daughter in 1973. I like to think that I did enough at home but really I think that my work / home balance was nearly always biased towards work. I was passionate about my job and, for example, missed a lot of my children growing up especially when they were young and I went to work before they got up and got home when they’d gone to bed. I left the children and managing the household to my wife, which is what my father had done. It can’t have helped the relationship and I was divorced in 1996. When I see how my son is involved in child rearing, I’m amazed. He takes time off work to go to school events and that sort of thing, which I never did, I’m afraid.
Ray Hennessy:
So would you say it was fairly common for people’s relationships to be damaged by their timekeeping and the stresses?
John Daines
Not consciously, I don’t think. It may have been; the effect on mine was a long time afterwards and it wasn’t just due to that. I enjoyed working shifts.
Ray Hennessy:
OK, let’s go on now. You’ve talked us through to somewhere at the end of the sixties when LEO was subsumed within ICL. What did you do after your LEO experiences?
John Daines
Basically, I never left so I ended up working for the company for 40 years. In 1966 to 67 I was at Minerva Road and System 4 was introduced and the 4/50 model was to be commissioned in Minerva Road. At that stage I had to set up sets of programmes for commissioning trials and get operators trained and myself trained. We acquired something known as the “Bank Suite” that had to have facilities incorporated to make it suitable for test purposes. During 1967 a new management arrived in the form of Vic Barnes who was the new Quality Manager for English Electric or English Electric Computers. He’d come from Rank Xerox and British Aerospace where he’d worked on Bloodhound. And his man Peter Sara who came to Minerva Road and replaced Tony Morgan whilst he was on holiday. [Historical note. He didn’t replace me but interviewed all my staff with me only being told that I needed some support. It was the beginning of the end for Minerva Road. There was a real change of management style then]. The first 4/50’s had been delivered by Christmas 1967 after much work with operating system people and engineers to get them out of the factory. By then it had been announced that 4/50 production would move to a new factory that English Electric had built at Winsford in Cheshire. This caused a great hiatus with the need to move “up north”. Vic Barnes worked hard to convince people that it would be a good move. I suppose for people living down south the thought of moving away from London is quite traumatic. Everybody says “Ah! But I couldn’t go to the theatre”: the answer is “when did you last go?” Anyway, I moved to Winsford in January 1968 and stayed in digs together with a few of the other engineers, although not in the same digs. It was a completely different environment where the factory was managed by an ex-English Electric domestic appliance machine factory manager who saw disc drives as washing machines: “Why aren’t we shipping 100 disc drives this week”? It was
also staffed largely by unionised people from the local Liverpool overspill estates although there were senior commissioning engineers still. I was involved with getting the first batch of 4/50’s out of there. In May, about six weeks after I had moved into the house, Vic Barnes asked me to go over to the Kidsgrove factory where they were having problems getting the early 4/70’s out of the factory. These were a much bigger machine than 4/50. The key machines were for the National Giro Bank, for Ministry of Social Security, UKAEA Risley and so on. Things were quite serious. I can remember Ninian Eadie, who was the salesman for Giro - a LEO person who had sold the original 326’s I think - coming into the factory one day and saying “I had lunch with the Postmaster General yesterday and he was just stressing how important it is that the National Giro opens in November.” So, no pressure there! That Postmaster General of course was John Stonehouse who disappeared some time later. One of the things I did was to get Brian Norris back, who’d been a key operator in Minerva Road, and he played a great part. During this time I worked again with the engineers (in Kidsgrove there was much more of a ‘Minerva Road type’ culture; it was almost like going home at times) and the programmers from the J operating system. There were great tensions then because they said “We can’t come and help you because we’ve got to write the next release because customers will need it”. And we said “Well, unless we can get these machines out of the factory, there won’t be any customers”. Anyway, at the end of 1968 I went into the team that looked after the J operating system, i.e. into the software department, and took with me all the experience I’d had of real life working with hardware problems. The place I moved into was quite exotic, it was just full of walking brains: really clever people with first class degrees and so on. They’d all read the manuals to see what the hardware was supposed to do and they’d written all the software on that basis, not including what could actually happen in practice. This meant that when there were hardware problems there were all sorts of software problems. So although I wasn’t as clever as them, my entirely different approach injected some different thinking and helped with many of the problems involving hardware and software interaction and this went on for a few years. I was sent out into the field on one or two notable occasions where there were difficult problems. In the early 70’s the proposed New Range came along and people were moved on to that and for a short while I went on to work on System 4 emulation working for Alan Bowers by which time I decided that I although I could do programming, the programming I did was not so much development programming as maintenance and ad hoc work and that what I really wanted to do was to work a bit more with System 4 still. In 1972 I moved to the local government sales organisation to support the 4/70’s in Edinburgh, Manchester and Coventry Corporations. I progressed on into sales support more generally, into project management, application implementations and then into Business Planning. After that I became a consultant and worked on business processes, which is where really where LEO started, I suppose. At the end I was a, quotes, “resource” that wasn’t wanted and got declared redundant after 40 years.
Ray Hennessy:
So effectively that was 40 years with just one company? [Yes, I suppose so] Quite unusual in this day
and age but very much what most of us did, I think. What else did you do in computers? Did you
join BCS (the British Computer Society)?
John Daines
Yes, I joined the BCS in 1964. I think Bruce Elkington had joined and it looked like a good thing to do.In 1968 they brought in some entry qualifications and I was an MBCS and they wanted 5 years valid
experience. I sent in what I had done and they decided that what I’d done on LEO II wasn’t counted
as valid so my 5 years’ started really in in February 1964 when I went to Minerva Road and into
commissioning and programming became involved. I stayed a member of the BCS really until I retired.
Ray Hennessy:
So when did you retire? Let’s see 40 years is that ’92?
John Daines
I was made redundant in March 2002, nearly forty and a half years. Since then I’ve kept myself busy.
I learned to ski in the early nineties and in the late nineties became a ski instructor at the North Staffs Ski Club in Kidsgrove. It operates the slope on a voluntary basis and I was also treasurer and
company secretary. One of the things we wanted to do was rebuild the slope and needed a lottery
bid for that, but that was very time-consuming but once I was made redundant I managed that lottery bid using a lot of skills acquired in my career at ICL, I must admit; in terms of project management, problem analysis and solving and I also gained a further ski instructor qualification. I still teach skiing twice a week and also when we open up especially for some work with schools. This year I’ve been away for 5 weeks skiing on snow. I’m also treasurer of the Tarvin Community Woodland Trust, a charity in our village that manages 11 acres of woodland on the edge of our village. I also do the bulk of the mowing, which is several acres, and, having gained a chainsaw qualification, I can do some other tree related work in the woodland. I also look after my 96 year old father’s paperwork and finances. Finally I’m a committee member of the LEO Computers Society and spend a lot of time trying to juggle all this as well as keeping in touch with son, daughter and grandchildren who all live on the fringes of the M25.
Ray Hennessy:
It sounds as though you are keeping quite busy. Now, just really to sum up what you feel about your time at LEO, whether you think it was really worthwhile and what have you gained out of it?
John Daines
I absolutely loved working on II/5 and whilst there I gained a good grounding in doing detailed work thoroughly. It probably helps that I’m a strong completer/finisher and a bit of an obsessive. Working in commissioning was a logical follow on and I came out of it with a set of skills and experiences that were quite unusual. They were certainly of use in J branch and much later when problem solving. Working with LEO people was just so enjoyable, the culture was good. In 40 years of work that ended without a goodbye or thank you I remember the people, many of whom were LEO people and some of them survived through into ICL where it became recognised that all that was gold was not necessarily ICT. Would I do it again? Would I go back? Certainly, it was my life.
Ray Hennessy:
John, that’s great. Thank you very much indeed. Now, I just need to summarise. This interview with John Daines has been recorded by The LEO Computers Society and the Society would like to thank John very much for his time and his reminiscences. The interview and the transcript will form part of an Oral History project which will document the early use of electronic computers in business and other applications, but particularly in business. I will repeat that any opinions expressed are those of the interviewee, that is John, and not of the Society. The copyright of this interview in recorded form and in transcript remains the property of The LEO Computers Society, 2012. Thank you very much indeed, John.
© 2012, Leo Computers Society

1 Catterick was and still is an army training camp on the Yorkshire moors – a bleak and unforgiving
place! It had a major Signals Training School.
2 Leavesden Green, Watford, was one of six emergency teacher training colleges founded by the London County Council after the Second World War. These colleges were intended to help solve some of the urgent post war problems of teacher shortages, offering a one-year course of training. They were all closed by 1951.
3 EDSAC was the acronym for an early electronic computer developed in the late 1940s at
Cambridge University on which the LEO I was based. It stands for “Electronic Delay Storage
Automatic Calculator”.
4 Hartree House was the name coined by LEO Computers Ltd for the offices the company occupied
above the Whiteleys Department Store in Queensway, London W2. It was named after Professor
Douglas Hartree FRS who developed Numerical Analysis techniques between the wars using pencil
& Paper and a Meccano-based calculator. He was involved with significant developments of digital
computer techniques in Cambridge after WW2. There were also several other small Lyons units in
the building.
5 Cadby Hall was the administrative building for Joe Lyons & Co in Hammersmith, west London. The first LEO computer was installed in that building.
6 ICT (International Computers and Tabulators) was formed by a merger of British Tabulating Machine Company (formerly Hollerith) and Powers Samas early in 1959. Some equipment was still badged as Hollerith.
7 The Decca company was primarily a recording company, manufacturing vinyl records and small tape decks for music. They expanded into computer peripherals and supplied tape units for process control and military purposes.
8 Ferranti was mainly an electrical engineering and defence electronics manufacturer that had moved into computers in the late 1940’s. They developed several early computers and were in the forefront, in the UK, of supplying drum storage devices.
9 The “long words” on LEO I & II were numbered 0, 2, 4, ... 28, 30 and the half words were each separately addressable:
0,
1,
2,
3, etc
10 See note 6, above
11 NCR paper was produced by the NCR Corporation, formerly National Cash Register Corp. In the case of the paper, which they supplied, NCR was an acronym for “No Carbon Required”. The paper was treated with chemicals so that when it was struck [by a print hammer or, in the case of the Samastronic printer, by a stylus] a copy appeared on the second and subsequent sheets. This meant that the number of copies required determined the thickness of the paper stock and was much less than stock with interleaved carbon paper. A drawback was that the print on the second and subsequent copies was subject to fading after two or three days in daylight.
12 All displays were on oscilloscope screens and the binary values in memory or register locations were represented by a dot (∙) for zero or 1. Thus 0/0 was displayed as “∙∙∙∙∙∙ ∙∙∙∙∙”, 63/31 was
“11111111111” and 63/29 was “111111111∙1”
13 “Intercode” was the name given to a translated coding system, an early assembly language.
14CAV was an early manufacturer of electrical components, started by Charles A. Vandervell and taken over by the Lucas group in 1926.
15 Radley House in Ealing, west London, was a training establishment used by LEO Computers Ltd when the need outgrew the facilities at Hartree House.
16 KDF9 was a computer developed by the English Electric team at Kidsgrove in Staffordshire. It
was primarily designed and optimised to run scientific programs and proved to be very fast for that type of work while also being able to carry out routine commercial tasks adequately.
17 “Minerva Road” in North Acton is where LEO became established as it outgrew Olaf Street (see below). It started with number 24, which was the development building by 1964. Other establishments were number 38-42 for commissioning, research and engineer training until it was moved to the “Number 1” building in Letchworth in Hertfordshire for more commissioning space and to “Number 16” for Field Engineering Services.
18 Charles House was in Kensington in West London, close to Olympia.
19 “Olaf Street” near White City, west London, was the first building used by Lyons to develop LEO I. In the mid 1950s the Engineering and Production divisions were moved to a much larger site in Minerva Road, which was nearby, in North Acton.
20 Winsford in Cheshire was the location of a new factory that had been opened by English Electric
shortly before the merger with LEO Computers Ltd.

Provenance :
Recording made by the LEO Computers Society as part of their ongoing oral history project.

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.