EDSAC — Andrew Herbert

In recent weeks EDSAC has surprised the team by exhibiting very few thermionic valve failures and consequent power trips. There have been several days when the machine has operated for 6-7 hours without throwing up nuisance faults, which otherwise take a long time to diagnose. This has enabled the team to make significant progress with commissioning.

We are well down the path of adding the electronics to drive a paper tape reader. These were designed and constructed by Martin Evans several years ago but it only recently has been worth connecting them. Overall this went well. Martin was caught out by a logic change in the interface with main control but this was easily overcome and in the week before Christmas EDSAC was shown driving a slow mechanical reader. Work remains to be done to ensure data is correctly copied from tape and put in the store, but optimism is high that this will work as required. The photoelectric paper tape reader is standing in the gallery awaiting connection in place of the mechanical reader once we are confident in the operation of the EDSAC side of things.

The Transfer Unit is now also commissioned by James Barr and Tom Toth, in the sense that transfers from the EDSAC Master Output Bus to the Master Input Bus are routed through it. There is some question as to whether the timing is correct, but certainly data is being delayed, differentially for “long” and “short” numbers.

A faulty CRT has been replaced in the Operators’ Display Units by Les Ferguy and work done to improve the quality of the images on all the screens. These are proving useful as a diagnostic aid now the CRTs show crisp images. There is possibly some work required on the timing alignment of the display of the Accumulator contents.

The next steps in testing are moving us away from running a single order repeatedly from the engineers’ switch panel, or running simple programs that loop for ever to simple single shot programs to check the correct calculations are being carried out by the arithmetical instructions. This has shown up a number of quirks that will be the focus for the next few sessions.

First using the STOP, RESET and SINGLE EP operator buttons to control programs manually is not fully consistent. STOP always works, but a consequent RESET (i.e., resume) or SINGLE EP doesn’t always do the desired thing.

A second source of glitches arises from the single shot tests not starting with the sequence of three X (no-op) orders that characterise the loop tests – sometimes the single shot code hangs part way through. This might be related to the first problem?

A third source of glitches appears to be executing orders that reference an operand that is just a few locations ahead of the order itself – these appear to hang. It is likely that by the time the order has decided to access to operand it has missed the opportunity for an immediate coincidence and will have to wait until the next major cycle of the store.

Assuming these matters are fixed, the next goal going forwards is that Nigel Bennée and I will check all the arithmetic functions for correctness in their calculations. Following that, commissioning the Initial Orders/Main Control interface and connecting the output teleprinter remain outstanding. Then as a final step we can begin to replace our temporary “digital tank emulators” with nickel delay lines for the internal register delay lines and the main store.

2024 is the 75^th anniversary year of the original EDSAC first entering operation – it first loaded and ran a program on 6^th May 1949 and started to provide a computing service to the university in about November of that year. So the pressure is on the complete the replica...

ICL 2966 — Delwyn Holroyd

We recently experienced corruption of the (emulated) EDS80 holding the George 3 filestore. Attempting to load the Colossal Caves adventure program that normally runs as a demonstration would result in a disc error – something you don’t expect to see from the disc emulator! This was easily fixed by restoring the disc image on the emulator SD card from backup, but I decided to investigate the corruption in more detail.

I found the last few sectors on one track were corrupt, with bits set or cleared apparently at random. This affected both ID and Data fields alike, which rules out the 2966 disc controller as the source of the corruption because ID fields are never written to post formatting.

The most likely source of the trouble is the SD card itself. The card has been in operation for nearly 10 years and this kind of localized corruption is typical for NAND flash memory when the limits of the error correction code are exceeded.

Harwell Dekatron — Delwyn Holroyd

Recently some work was needed on the single-step / single-order switch on the control panel. The switch had become very stiff to operate. One of the connecting wires had broken but remained in intermittent contact with the switch, causing apparently random failure of the machine.

Normally the machine runs a “table of squares” program for demonstration purposes, but other programs do exist. One of these is a program to compute Pi using an iterative method. This program, unlike the table of squares, makes use of the divide instruction. We received reports that the machine was stopping during the divide instruction and this is currently under investigation.

We continue periodically to train up new staff members and volunteers to operate the machine.

SSEM Replica — Chris Burton

Good progress has been made in tracking down some longstanding faults which has improved stability of the system. Some unreliability of the modular high tension power supply system has led to consideration of replacing that system using modern power modules in a smaller rack. A possibility is that it could be out of sight in an adjacent room.

Authenticity would actually be improved because the original machine indeed used ex-GPO modules, in an adjacent room.

In December of last year, the Minister for Artificial Intelligence announced the opening of the government’s “Manchester Prize” in front of the Baby Computer replica. A huge thanks to Bob Geatrell, Tim Banks and Richard Leszczynski for operating Baby and answering the minister’s questions. It was a good publicity opportunity for the Baby replica!

Way back in 2020, just before lockdown, we moved one of the ICT 1301 tape decks from storage into the Large Systems Gallery at TNMOC. It was intended to facilitate recovery of both ICT 1301 and LEO tapes, thanks to the discovery of an original LEO 8-track head for the Ampex TM-4 tape transport which forms the basis of both the 1301 and LEO decks.

In 2022 Rod Brown and I managed to return the tape deck to running condition. Since then the ball has been in my court to complete the data capture system. This consists of a 16-channel ADC together with a Xilinx Zynq SoC consisting of FPGA fabric and a dual-core ARM processor. The hardware is off-the-shelf but to build a working system required FPGA interface code, a Linux kernel driver and a user space command line application which captures to WAV files stored on an SD card. The ADC is capable of up to 1M 16-bit samples per second but the SD card limits that in practice.

With the software finally complete and tested using audio input it was time to try it out for real on a 1301 tape. After a few experiments I found that a 384K 8-bit sampling format captured the flux transitions on tape with very high fidelity (the maximum flux transition rate is 22.5KHz). I captured the test tape twice for good measure.

Using WAV files makes it easy to examine and if necessary edit the captured analogue waveforms using any audio editing application. The captured data looked very clean; the image below shows an extract of five of the 10 tracks.

The next step was decoding the analogue waveform into blocks and characters. The ICT 1301 project had previously captured a few tapes when the machine still resided at Buss Farm, using an audio multi-track recorder and software written by the late Andy Keene. I was able to make use of this software to successfully decode the analogue waveform.

The 1301 tapes have an extremely robust error correction and detection code – each word on tape consists of four data bits and six error correction bits! This allows any single track error to be corrected and multiple track errors to be detected.

Only one tape block contained uncorrectable errors – looking at the analogue data it was obvious that three of the ten tracks had drop out in the same place. On two of these tracks the transitions were visible enough to be manually enlarged in the audio editor, and error correction handled the remaining track.

Having proved the process with the new capture system, the next step is to identify the most important tapes and work through them over the next few months.

Elliott 803, 903 & 920M — Terry Froggatt

I can report some progress on the 920M front. I’ve taken my non-destructive analysis of the circuitry about as far as I can, and given this better understanding, I’ve had another attempt at working out why both 920Ms have failed. I’ve known for some months that the eBay 920M (5343) behaves deterministically when it is powered up, but the RAA/TNMoC 920M (88) appeared to be non-deterministic. On both machines, the address of initial instructions is being read correctly into the M-register (store contents buffer), but I’ve now found that, on 88, it was being overwritten, just before it was written to the SCR (program counter) at location zero in store.

Before anything can be written into a core store, the existing contents have to be flushed out by a “read” cycle which ignores the store output, and 88 was failing to ignore this output. So the value being written to the SCR was a mixture of the correct value and of whatever was in the SCR when the previous test was halted, hence the non-determinism. By borrowing one module from 5343, I was able to get 88 to run some small programs, which showed a further fault, in bit 14. With this fixed, 88 was able to run XSTORE, BASIC, and the CountDown program, and I took the opportunity to record a collection of “truth” oscillograms whilst I had access to a working 920M.

Then, by borrowing three modules from 88, I was able to get 5343 to run some small programs, although somewhat unreliably, and with a possible fault in a word of each 64-word block. Dr Erik Baigar kindly posted the corresponding three modules from his cannibalized machine in Munich to my son in the Netherlands, which I collected at Christmas. Sadly 5343 did not work with these modules. By comparing pin resistances out-of-circuit, I concluded that the type-38 modules at position L04 have a capacitor across the 12v supply, which was OK on 88 but which had failed on 5343 and Erik’s machine.

Peter Williamson reports that the TNMoC 903 was very sulky, after not being used for 10 days around the Bletchley Park AI conference. It did boot up and run, but stopped on several occasions, and BASIC had to be reloaded. Test programs X3 and XSTORE ran with no problems, so it was very difficult to ascertain what was going on. As the day progressed the machine settled down again.

There is also an intermittent problem with the expansion memory PSU, where you hear the relays cut out for about one second, and the ±6V supply lamp will extinguish but returns. The unit does not need to be re-powered from the control panel “on” button after this happens. The engineers display is still not connected to the 903 but Peter has had it powered on and it seems to be OK with the replacement PSUs.

Peter Onion reports that the TNMoC 803’s 13.5V inverter (used for the positive source for the +10V supply) had started to give trouble again. This manifests itself by the +10V supply failing to come on quickly enough. After much head scratching it was found that the output filter capacitor had gone open circuit. But the symptoms were quite odd.

He believes that the inverter was designed for use in communications environments, so it has chokes in both output rails designed to keep switching noise “in”. Without the final output capacitor in play the +10V regulator was being fed from a mostly inductive source, which was causing it to oscillate. The confusing part was that the oscillations were only apparent on the output of the inverter and the input of the regulator, the +10V line only showed a very small ripple at the oscillation frequency (which was approx. six times the inverter switching rate so it clearly wasn’t coming from the inverter itself).

For now a replacement capacitor has been connected across the inverter output terminals as replacing the original part will require complete disassembly of the inverter and Peter is reluctant to do that while it is otherwise working. Removing the small ripple from the +10V supply seems to have improved the reliability of the core stores, probably because the sense amplifiers derive their bias from +10V via nothing more complicated than a potential divider and a potentiometer!

The National Museum of Computing — Aneesa Riffat

We have now submitted our application to the Arts Council to become a Nationally Styled Museum through the Accreditation Scheme. We expect to hear back in the summer with a result. This has been an important piece of work for the Museum to undertake, and it has shaped our working by professionalising procedures using the Sector standards. All areas of the business have been reviewed with many new procedures and policies in place as well as re-shaping existing procedures.

The roof restoration project will be taking place imminently which will help to care for our building externally as well as protect our collection and other assets internally.

IBM Museum — Peter Short

3279 Colour Terminal

We now have a working 3279 colour display terminal running diagnostics from a working 3174 controller. We weren’t able to get the machine working that was reported last time. The power supply checked out OK, but the analogue card is U/S. We have neither replacement nor circuit diagrams.

We retrieved the 110 volt machine from storage and with a bit of effort persuaded it into life. It is a completely different design internally and appears to have been a development model.

3174 Controller

The 3174 was in good health, and after some reading to determine how to boot it up to run diagnostics we connected it via coax cable and this –> is the result.

2378 Mono Terminal

The 3278 mono terminal has also been tested on the controller. It does now show most of the diagnostics screen, but is very faint and very jittery. This still suggests an analogue card problem, and we’ll need to see if the brightness can be boosted. Otherwise it will need to remain as a static display.

129 Card Punch/Verifier

The 129 card punch/verifier is about ready for its first power up, and has been cleaned to a level where we can put it out on display, and lubricated. Whilst it does now power up to the extent that the motor runs, the machine currently does not otherwise function. Initial measurements around the power supply reveal some out of specification voltages.

Schools Computer

We’ve been contacted by an employee interested in the 1969 Schools Computer project. He is planning to build a working replica, possibly half size and with modern screen. We’ve supplied copies of documentation we have in the archive, and facilitated a visit to Oxford University Science Library, where there is further documentation he can scan. If he succeeds with the replica he will make a second one for the museum. Our machine that is on display is an empty shell, with nothing inside.

Back Office

A long overdue tidying up of the back office workbench has finally happened. The old CE tool case contents are now hanging from magnetic strips, thus saving quite a bit of workspace.

System/390

Our recently acquired System/390 including the integrated Support Console which runs a tailored version of OS/2 has been successfully powered up. We now need to work out how to get the S/390 to do something useful.

The cooling fans are extremely noisy so we might need to find a small room in which to run demos without disturbing the whole museum! Or maybe we can find some suitable quiet-running fans for it, or some noise cancelling headsets.

This S/390 is part of the family of Multiprise servers IBM announced in the late 90s. Its prime purpose was to provide a companywide server for Small and medium enterprises, as such these machines became very popular and were adopted by many smaller corporations across the world.

Museum Websites

Christmas week saw a big increase in site visitors. The museum website saw over 3,000 page views during the week, the photo archive over 500.

Analytical Engine — Doron Swade

Work continues apace on defining which of Babbage’s Analytical Engine designs would be meaningful to build. Tim Robinson’s extensive description and analysis of Babbage’s range of designs provides the stimulus and knowledge-base for this process. Len Shustek and Tim have been digging deeper into the algorithms and mechanical design implementation of some of Babbage’s computational processes. Their experiences are both sobering and heartening. Heartening because these are the conversations we have been waiting to have. Sobering because of the complexity of what is involved in reverse engineering detailed intention and algorithmic principles from the mechanisms depicted in the drawings and their accompanying notations.

Specifically, Tim and Len have been examining Plan 27 and Plan 28a, two advanced Babbage designs, for their viability as a build target. One outcome has been Tim’s written analyses of the levels of completeness of each of these Plans. A collateral prize has been a piece of bibliographical reconstruction to inform understanding of Plan 27. The Buxton papers held at the History of Science Museum, Oxford, contain material Babbage wrote in Florence in 1841 while he was working on Plan 27. The Buxton manuscripts are unsympathetically bound: folios are out of order, contain revisions, and material in the gutters of the spine is difficult to read. Tim has revisited the images of these manuscripts, and our transcriptions of them, to reconstruct the likeliest linear account from the patchwork quilt of the primary sources. He writes that ‘the result turns out to be one of the most coherent pieces he [Babbage] ever wrote on the Engine.’

Devices Len and Tim have explored include the anticipating carriage mechanism, the method of division and the operation of the barrels used for ‘microprogramming’. The anticipating carriage was an early breakthrough for Babbage who wrote that the invention ‘produced an exhilaration of the spirits which not even [his host’s] excellent champagne could rival’. Len has been experimenting with algorithmic simulations, component-level simulations, and some 3D printing. These investigations deepen and extend understanding. They also agitate and inform an ongoing debate about trade-offs between manufacturing costs, historical fidelity and what is practically realisable in a foreseeable timescale.

Reflecting on Babbage’s failure to complete any of his engines Babbage’s son wrote: “The History of Babbage’s Calculating Machines is sufficient to damp the ardour of a dozen enthusiasts.” – Major-General H.P Babbage, 12^th September 1888.

I wonder if we can prove him wrong.

Delilah Voice Secrecy System — John Harper

We have been involved in a few internal demonstrations at Hanslope Park where Delilah successfully sent and received encrypted voice. However, to obtain reliable results the machine needs to be very carefully tuned. We have also suffered from wiring problems in the Combiner Unit where solder joints have not been reliable because too many components are crammed into one area. We have suffered from signal pickup between wires adding to the sensitive set-up issues. We attribute these issues to the very poor under-chassis wiring layout in the combiner unit. We did not fully optimise the layout when we constructed our prototype. We had nothing to go on when we first positioned the valves and other major components on the chassis.

Rather than rework the early version, which we wish to keep going and with which to continue giving demonstrations we are about to manufacture additional Combiner units. There will be an optimised component layout on new combiner chassis. In addition, there will be a version using a PCB to make the component connections.

ICT 1301 — Rob Brown

After a very long period of no Progress on the 1301 project , I am very happy to report that the biggest stumbling block to moving the project forward has started to be overcome.

Thanks to activities by Delwyn Holroyd and Martin Gillow at TNMoC (See Data Recovery report above [Ed]) and some historic software written by an original member of the students who saved the machine in the 1970s, we believe we have managed to start to recover the original software locked away on original Magnetic Tapes.

It is a priority that this software is now recovered, as the current existing emulation tools need this software to run and demonstrate how the system would have worked during its working life.

We hope to have recovered enough software by the end of this year to enable the online emulator to be used remotely as a working system. The rest of the team are very enthusiastic now we have overcome our biggest stumbling block.

Chair’s Report

I am pleased to report an energetic and successful year for the Society with active programmes on several fronts.

Following new constraints in BCS budgetary allocations we successfully weathered financial belt-tightening under the expert stewardship of our Treasurer, Arthur Dransfield, who navigated the rapids and mitigated the worst.

Significant efforts were directed to explore how we could better serve our wider membership by reviewing the Committee’s access to members’ data in full compliance with Data Protection Regulation. Discussions are ongoing with the BCS. Huge thanks to the sub-committee that worked, at times under some pressure, to craft a proposal before the summer recess.

At the end of 2022, Resurrection, the flagship journal of the Society, reached its 100^th issue. This was a milestone in an unbroken run since the founding of the Society in 1989. The event received a deserved round of applause for Dik Leatherdale, Resurrection’s editor for the last 58 issues since 2008, and for the countless contributors of articles, features and reports, who generously shared their knowledge and expertise with us.

The London Meetings Programme, our lecture/seminar series, is undiminished in the variety, quality, and popularity of its speakers. Live attendances at BCS’s London offices continue to recover post-pandemic and the hybrid platform, allowing remote attendance, continues to substantially widen domestic and international outreach. All thanks to Roger Johnson, our London Meetings Secretary who so effectively attracts expert, lively and thought-provoking speakers.

The Society is 33 years old this year and the initiatives to recover and maintain records of the Society’s history have made excellent progress. The programme to identify, locate, audit, and digitise the early video records of the Society’s lecture series, and archive these for public access, is now well under way thanks to Kevin Murrell, Dik Leatherdale and Dan Hayton. The collection is a priceless resource of contemporary thought and practice including presentations by many distinguished pioneers. The Committee has also initiated a programme to harvest still photographs, in private possession, of CCS events, restorations, Working Parties, and personages as part of the Society’s digital archive. In addition, the first of several video interviews of founding members of the Society has been completed, transcribed, edited and archived. This first is an interview with Chris Burton whose contribution to the Society’s activities and programmes is inestimable, and this initiative, under the guidance of Martin Campbell-Kelly, has provided a template for future interviews of this kind. This year marked the 25^th anniversary of the Manchester Baby replica led by Chris Burton. Congratulations to Chris and the team of volunteers at the Science and Industry Museum in Manchester, where the Baby has been demonstrated since 1998.

I am delighted to report the resumption of the Tony Sale Award after it was intermitted in 2019. The 2024 Award is in collaboration with The National Museum of Computing (TNMoC) which has secured s ponsorship and will be hosting the award ceremony. Entrants will be assessed by an independent judging panel chaired by Martin Campbell-Kelly.

This is my last report as Chair. I wish to heartily thank the Committee for their efforts over the past year. They are as committed and conscientious a team as I have anywhere encountered, and their efforts are much appreciated. It but remains to wish all the very best to my successor, Chris Rees, our new Chair. I am confident about what the future holds under his distinguished stewardship.

Chris Rees has a Master’s degree in Logic & Metaphysics and Political Economy from St. Andrews University. After 20 years in the computer industry implementing payment and trading systems and as Managing Director of Logica Software Products Ltd, he joined Deloitte Consulting as a Partner. His assignments included leading roles in major privatisation programmes in the UK, Russia and Eastern Europe. He led the Knowledge-Based Systems Centre, a unit that developed the first successful fraud detection system for Barclaycard, and deployed the technology for the major UK privatisation issues. He acted as programme director for the Gas Industry Domestic Competition Pilot and advised the Governor of the Bank of England’s Task Force on the specification of CREST, the Stock Exchange settlement system.

In October 1996, he founded Charteris with three senior industry colleagues. Drawing on his extensive experience of large scale project and programme management, he specialised in dispute resolution in IT projects. He frequently acted as an Expert Witness in IT-related disputes and led the largest specialist IT expert witness group in London. He was a Director of the company until its acquisition by the Sword Group in January 2014.

He was a founding partner of Principled Business, a firm which provided training in business ethics. He has lectured on business and professional ethics for many years, on public courses and to students at the University of Surrey.

Chris is a Fellow of the British Computer Society and a Chartered Information Technology Professional. He co-founded and was first chairman of the BCS Financial Services Special Interest Group. He was President of the BCS in 2018-2019. He took as his presidential theme The Ethics of AI. He wrote and lectured widely on the topic in the UK and overseas. He continues to research and speak on the subject with the advent of Large Language Models.

He is a Liveryman of the Worshipful Company of Information Technologists and was, for a number of years, Chairman of the Ethical and Spiritual Panel of the Company, running regular seminars on ethical topics at the House of Lords.

He teaches philosophy at the School of Philosophy and Economic Science, as well as a series of courses on Philosophy in Business. He is Treasurer of the School.

Two curious devices were connected to the Ferranti Atlas computer at Manchester University in the 1960s. One was an on-line X-Ray Diffractometer; the other was simply known as the Speech Converter. This is the story of the latter.

Modest research into optical character recognition had been carried out in the mid-1950s in Manchester. By the autumn of 1962, when the first Atlas was nearing completion, it was decided to explore a more ambitious class of pattern recognition, namely speech. Could the new supercomputer be made to respond to spoken commands? Furthermore, it would be fun if the machine could respond by uttering synthesised speech. So I, an innocent research student, was recruited to start the ball rolling.

As was customary, the first step was to simulate candidate recognition and synthesising schemes in software – meaning that realistic samples of speech waveforms had to be captured and digitised. This, in turn, required an on-line Analogue-Digital-Analogue converter – known locally as the Speech Converter. This equipment was unusual at the time (1963) because it handled DC to 10 KHz with 8-bit accuracy in order to give reasonable-quality speech. It took an 8-bit sample of an input waveform every 50 microseconds, so its potential data rate for input or output (20 Kbytes/sec) was at the time impressive. For comparison, the transfer rate for each Atlas high-performance magnetic tape decks was 48 Kbytes/sec. By the way, the idea of an 8-bit byte as a standard unit of information was not yet commonplace when Atlas was being designed. Atlas used 6-bit characters and 48-bit words.

The Speech Converter was a real-time device with a crisis time – meaning that the Atlas CPU must guarantee to service a Speech Converter’s interrupt within a critical time if continuous speech of arbitrary duration was to be accommodated. Without buffering, the critical time was very short – 50 microseconds. It was decided that the Speech Converter should therefore assemble six eight-bit samples and issue a request for a 48-bit Atlas word transfer every 300 microseconds. Double-buffering was employed, so the Speech Converter contained two 48-bit flip-flop buffer registers. The total time spent by the Atlas Supervisor (i.e.Operating System) dealing with each Speech Converter interrupt was about 40 microseconds – once higher-priority interrupts had been dealt with. Recall that the Atlas CPU shared its time between potentially many asynchronous activities.

As it was, one did not lightly run the Speech Converter for more than a few tens of seconds at a time. On Atlas, this meant that incoming data had to be rapidly moved through Virtual Memory from core to drum to magnetic tape in real time. When I used the Speech Converter for input, the computer was usually entirely devoted to my endeavours. I remember Frank Sumner, my Ph.D. Supervisor, and sundry maintenance engineers and operators all standing around with a look of bemused amazement (or was it anxiety?) on their faces whilst this young research student tried to bring the most powerful machine in the world to its knees! I was not too popular, since most other user-jobs ceased whilst I did my stuff. I remember Frank’s wry smile as I stood there nervously counting each 3Kbyte block of data that was being written up to tape – you could hear the deck jerk forward as each block was written. He remarked that here was I, surrounded by the world’s most powerful machine, but counting on my fingers!

Once the Speech Converter’s Supervisor routines had been written in machine code, a library of eight signal-processing routines plus an initialising program were written by two research students (myself and Lynn Rosenthal) using the Compiler Special facilities in Atlas Autocode, an Algol-like language. The library, called SPP1, was stored on the same magnetic tape that contained samples of connected speech. SPP1 could be called down by applications programmers.

The SPP1 software

From the programmer’s viewpoint, the Speech Converter simply produced a very large one-dimensional array of fixed-point numbers in the range -127 to +128, representing waveform amplitude-values every 50 microseconds. If this was the result of continuous speech, the first problem was to distinguish between background noise and meaningful utterance, then to segment the speech roughly into words and finally to try and identify phonemes within words. The first six routines of SPP1 were used to get an overall picture of the total record, using statistical measurements derived experimentally. The final two routines performed detailed spectral analysis on chosen sections of the record. In all cases, the user provided each routine with several parameters, some of which related to the time-intervals of the waveform record being inspected and others of which controlled the detailed algorithms contained withing the body of a routine. Without going into too much detail, here is a summary of the eight routines in SPP1.

Applications of the Speech Converter and SPP1

My particular application was automatic speech recognition. At the time, other researchers elsewhere were mostly using analogue equipment. They typically reported accuracies of 95% when dealing with very limited vocabularies – usually the spoken digits zero to nine. Details such as any prior speaker-training or number of different people able to use their equipment were curiously omitted from the reports. In short, this was indeed the very early days of speech recognition.

My recognition scores were similarly pathetic, though I strove ab initio to deal with speaker independence by studying a wide range of male and female accents. I developed algorithms that were mathematically simple, fast, and capable of being replicated in external circuitry, thus greatly reducing the incoming data rate to the computer. But basically all my own Ph.D. thesis showed was that the recognition problem was difficult!

My ideas were later taken up by another Manchester research student, Brian Carpenter, who implemented a real-time speech recognition system for a PDP8 computer. To assess performance he used two vocabularies: one was the spoken digits zero to nine; the other was a five-word vocabulary with direct visual feedback. This gave voice control (up, down, left, right, stop) of the movement of a spot on the PDP8’s display screen. I remember some users frantically shouting “Stop!” if the spot was moving inexorably off the edge of the screen! They were ignoring Brian’s advice to ‘speak calmly’ – thus spawning a further line of investigation into so-called human factors.

In the speech synthesis area, two fellow research students, Ron Mathers and Lynn Rosenthal, had more luck. Their starting-point was a vocoder-type analogue system called PAT – (Parametric Artificial Talker) that had recently been demonstrated at the University of Edinburgh. This apparatus was a marvellous assembly of buzzers, hissers and band-pass filters, controlled by eight parameters that were periodically updated. As part of unrelated research, Ron and Lynn wrote a software simulation of PAT using eight 3-bit parameters, updated every fortieth of a second if I remember correctly. To judge the quality of the synthetically-generated speech, the data was streamed to the outside world via the Speech Converter. We stood around delighted by the sounds! Adaptations of the PAT model were investigated and the voice quality improved. Two more Ph.D.s in the bag!

At about this time we were interviewed by a puzzled BBC reporter, investigating ‘talking computers’. At the end of the chat, we were asked whether we could synthesise a short BBC announcement in a rich northern accent for their early-morning news programme. We laboured through the night, recording our sample BBC announcement spoken by one of the Electrical Engineering Department’s porters who had a distinctive Hilly Lancashire accent in the style of Fred Dibnah. We fed his voice in through the Speech Converter, analysed the waveform to produce sets of the eight arameters required by the PAT model and then fed the synthesised version of Fred’s announcement as a string of 8-bit integers representing 50-microsecond samples out through the Speech Converter. In the early hours, a courier took the audio tape to the BBC’s Manchester studio. Sure enough, over the radio next morning came not the usual posh south-of-England accent but ‘Fred’, saying something like: “The time is coming up to seven thirty and you’re listening to BBC Look North”. Several senior colleagues arrived at work that morning to describe how, unbelievable though it seemed, they swore they had heard ‘Fred’ the porter make a radio announcement!

What did the equipment look like?

The picture below shows the Speech Converter’s performance being checked by me in about 1964. The photo was taken in one corner of the crowded Atlas machine room at Manchester, where the device was installed once Tom Kilburn, everyone’s boss and inspiration, had been assured that a prototype had successfully operated in another lab. The thought of some maverick device tripping the power supply of a £2 million installation could not be contemplated! Interestingly, the prototype was built using a 50-volt Westat power unit and the steel framework of a 19” Post Office rack left over from the 1948 Baby computer project. The final product was much smarter: it used standard Atlas style of cabinetry and racking for the printed-circuit boards on which discrete components (germanium transistors in those days) were mounted. From memory, the unit in the photo measured approximately 120cm high, 55 cm wide and 45 cm deep.

Many CCS members will recognise Professor (as he then wasn’t) Simon Lavington

Looking at the photo, the Speech Converter’s front panel included:

What happened in the end?

In addition to voice studies in Manchester, the speech converter was also used briefly by the Admiralty for a classified signal-processing task. This contract was handled by Ferranti and I was not involved. My guess is that it was concerned with underwater sounds – from submarines or ship’s propellers? The Joint Speech Research Unit (an amalgamation of the speech research interests of several UK government departments) was also interested in using the Speech Converter. They came up and watched a demo but I’m not sure whether they actually used it for their speech investigations. Lowly research students were not consulted! STL Harlow did use the device and our software routines for some work on automatic speech recognition. I know they complained bitterly of the cost that Ferranti charged them!

When the Manchester Atlas was shut down in September 1971, the Speech Converter was finally switched off. It was preserved in storage, along with other Atlas bits, and certainly still existed when I left the University in 1986. It has since vanished.

Was speech research continued at Manchester? No. Ron Mathers, whose main interest lay in road traffic simulation and control, went back to New Zealand. Lynn Rosenthal, whose interest had turned to simulating neural nets, went back to America, then to Nigeria, and finally to open an underwater photography and scuba diving school on a Caribbean island. Brian Carpenter left to join CERN. By 1968 most of the efforts of the computing group, including myself and Frank Sumner, were focussed on the design of a new computer, MU5. The attraction of being able to talk to computers had faded.

Of course, nowadays what was formerly a quaint and unreliable party trick has become the robust and useful tool of automatic speech recognition. I readily admit that I am amazed and delighted every time my smart phone responds with complete accuracy to my spoken messages.

A Virtual Experience of LEO 1

Since 2018 the Leo Computers Society (LCS) has been working in partnership with the Cambridge-based Centre for Computing History (CCH) on a National Lottery-funded project to provide a long-term, centralised home for the LEO Society’s historically important collection of documentation and artefacts relating to LEO, the world’s first business computer, and to make that collection accessible to the general public.

At a meeting held annually to reward achievements in the IT industry the Societies were awarded 2^nd place (‘Highly Commended’) in the charities category of the BCS’s UK IT awards. This was for our successful Virtual LEO l application.

The Lottery project ended formally in June this year, but the cooperation of the two organisations continues. The result to date is that the Cambridge centre now looks after many thousands of individual items on the LEO Society’s behalf.

And because the staff of J Lyons & Company and Leo Computers Limited were meticulous record keepers, recording not just decisions at every step of the journey, but the reasons for those decisions and their consequences, the museum has been able to cast the collection in a wider social context, bringing the human and social stories involved to the fore.

Virtual LEO 1

The most innovative part of the project was a digital ‘rebuild’ of the original LEO.

In May this year, a PC-based version of this was installed in the Cambridge Centre. Watching visitors of all kinds use it since then has been a real joy for us.

The virtual LEO is an historically accurate digital representation of the first LEO machine. Unlike a physical rebuild, it also points to the social context of post-war London and the Lyons company context within which LEO I was constructed. Now in a major step forward to opening up access from anywhere in the world, the virtual LEO is also available for access from the ubiquitous Apple iPad and soon from Android-based devices.

Using the virtual LEO, it is possible to ‘wander around’ the LEO room in the former Lyons’ HQ complex in Hammersmith, and view the racks of valves, hear all the peripherals and even look out of the window at the smog. Within the app, there are 35 interactive objects that can be ‘picked up’ and explored, such as a half adder and a mercury delay line.

There are 44 documents, six film snippets and many photos from the archive, accompanied by over 40,000 words of explanatory text divided into six themes. It has been purposely designed in layers so that it is perfectly possible just to walk around the room and see and hear what it was like to operate the first LEO.

Virtual LEO 1 displays the World’s first business computer in its room in J. Lyons headquarters building, Cadby Hall. Viewers of it can click on links to find out information and articles about the various parts of LEO and find out a bit about the history of the machine. The viewer can also look out of a window on either side, on one side Lyons vans can be seen in the yard, On the other side there is a view of the buildings on the other side of the road.

Virtual LEO 1 is freely available on post-2020 iPads and will soon be available on Android laptops.

Further details on the app and how to download at www.ccsoc.org/leo2.htm.

Other highlights of the collection

Over the course of the project, there were more than 50 separate deposits of LEO material made to the Cambridge museum by LCS members. There are some objects too, such as small pieces of various LEO machines including, tape reels and logic boards.

These have been catalogued, digitised and in some cases transcribed, making more than 1,600 separate items available online to all, via the CCH website. These incorporate almost 13,000 scanned pages of documentation in total. On the LCS website these sit alongside the CCH version of the LCS Leopedia database, extending that resource in a wholly searchable and search engine optimised way. Whereas Leopedia lists references to secondary sources of information on LEO in books, journals and in the media, the archive catalogue lists and makes available the primary material. The digitised LEO collection and Leopedia is at www.ccsoc.org/leo3.htm.

Finding aids are also available and are being extended all the time so that the collection can be searched easily by people who know nothing about LEO – for example by subject – see www.ccsoc.org/leo4.htm.

The archive includes reports from visits Lyons staff made to Cambridge during the development of EDSAC; photographs of the construction of LEO I, particularly those that point to the experimental nature of the early machine; a memo from TR Thompson reporting on the first fully successful and complete run of the world’s first commercial computer job, Lyons Bakery Valuations which was completed at 2:35pm on 30^th November 1951; rare BBC film footage of LEO I running the Lyons payroll and other snippets of film; over 60 oral history interviews and more than 200 written reminiscences collected during the project. And much more besides.

LEO Documentary Film

The lottery project funding also allowed for the creation of a short documentary of the LEO story. Produced by Boffin Media, the film won the Association for British Science Writers (ABSW) Video of the Year Award in July 2022.

The film is available on YouTube and the CCH website (www.ccsoc.org/leo5.htm). It was produced to coincide with the 70^th anniversary of LEO 1’s first fully successful live program run in 1951.

Obituary : Tom Hinchliffe

Tom Hinchliffe with the prototype of the ICL Series 39 SX

In October we lost Tom Hinchliffe, after a short stay in hospital.

He was chair of the NW group for 18 years from 2000 following Frank Sumner, until reduced mobility persuaded him to call it a day.

Tom was very much a local lad, living in and around Manchester his whole life. He gained a first-class honours degree in Electrical Engineering from Manchester University along with the prize for the best engineering graduate and then went to one company for the rest of his working life. That company changed its name several times around him, from Ferranti via ICT and ICL to Fujitsu.

Initially working as a design engineer on Ferranti’s Mercury, Atlas and Orion, ICT’s 1900, Tom became project manager for ICL’s 2970 and 2966 mainframes. In the early 1980s, when ICL turned to Fujitsu for their advanced technology, Tom led the technical collaborations and managed the team who designed the highly successful and powerful Estriel mainframe range. Later, as Managing Director of ICL’s High Performance Systems division he became responsible for all ICL’s large systems.

Under his leadership, the division was awarded three Queen’s Awards for Technology and to his immense pride, the UK Quality Award in 1995.

In 1995, Chris Burton made a presentation at West Gorton about the proposed “Baby” replica. Tom generously organised sponsorship from ICL, including finances and workshop resources.

Although Tom nominally retired from the day to day business of ICL in 1996, he remained on the board and was able to become more involved in a number of external interests.

He was a governor of Manchester Grammar School, director at the Science and Industry Museum and visiting professor at both the Victoria University of Manchester and UMIST. He was a member of council and court at both of those universities, albeit not at the same time, and Pro-chancellor at UMIST. When they merged in 2004, Tom was honoured to be appointed as one of the four Founders of the new Manchester University.

It wasn’t all work, though. He and his wife, Eva, met when she worked on documentation in Ferranti and had to visit the design labs to find the tea trolley.

He also leaves behind a legacy of great British mainframes.

Book Reviews – Stories of Computing Long Ago

These are two autobiographies whose authors’ IT careers started in the then-novel IT profession in the heady days of the sixties, “Bliss was it in that dawn to be alive...” – and which CCS members may therefore find particularly relevant and entertaining.

An Unexpected Development Clacton to Melbourne – The Long Way – Derrick E. Brown

Derrick started writing his book primarily for his family. It relates how he came into IT in the late 1960s as a qualified mechanical engineer, at Marconi Instruments in systems analysis. From there he joined Sainsbury’s, the most successful UK supermarket at the time, where he specified and developed the world-first barcode “Just-in-time” system for restocking supermarkets. He gave a presentation to the CCS by Zoom, “The Birth of the Barcode” in January 2023 from Melbourne, Australia, where he now lives – available on YouTube: www.ccsoc.org/ibm2.htm.

He went on to Plessey in IT management, and then became a trainer at Keith London Associates. As a director, he built up the company, which also provided consultancy services. He moved to Australia where he continued and expanded his work with the company, finally selling it. He built up a new company, principally in training. Among his clients were federal departments in Canberra.

Trains, Planes and Computers. From Executive Jet to Bus Pass – Stan Price

Stan’s book was written on the justifiable assumption that the breadth of his IT experience, unique in many respects, would be of general interest. With a degree in mechanical engineering, a brief summary of his career is as follows:

Programmer, operator, systems analyst, project leader, in the computer services department at Hawker Siddeley on the Ferranti Pegasus;

Software engineer, telecommunications engineer and manager at the Civil Aviation Authority (CAA), the civilian component of the National Air Traffic Services (NATS) where he was involved in air traffic control systems;

Systems management at Scotland Yard and Univac;

Sales at Software Sciences, a systems supplier

Stan set up his own services company, and was subsequently a teacher, lecturer, project monitor, research programme co-ordinator and expert witness, with many organisations including Sheffield University and the Alvey Programme.

Stan reminisces honestly about his career, expertise and various successes – not omitting the politics. He recounts the differences he encountered with his employers or colleagues, at one point being told that he “got up senior management’s nose” and being recommended a “Working with People” course!

Interest for CCS Members

I myself can relate to both books: I worked at Sainsbury’s as a systems analyst for 18 months in 1973-74, when Derrick was the departmental IT manager. I have served on IT Committees and had a number of speaking assignments at conferences with Stan.

Both books are self-published and would have benefited from professional proof reading and editing for consistency, structure and balance. For example, the dates of employment at a company would be useful to know. An index would have been helpful.

CCS members will certainly find the books well worth reading, whether from the perspective of IT professionals over a similar time period, perhaps identifying, comparing and contrasting from their own experiences, or for those with any other historic interest in the machines or systems.

Theatre Review – ALAN TURING A Musical Biography

To Riverside Studios in darkest Hammersmith with a few members of the Society’s great and good to experience this rather curious theatrical event. Despite the stars on the poster, most reviewers hated it. In particular the Grauniad’s Arifa Akbar complains that the song lyrics are “laden with mathematics” www.ccsoc.org/tur0.htm). This, of course, will not trouble CCS members but perhaps that is to be expected from a drama critic.

We, of course, come to the party with different eyes. So how do we experience the performance?

Well, the first thing to say is that anybody unfamiliar with Turing’s life would emerge none the wiser. But is there anybody in the UK (let alone the CCS) who isn’t up to speed with the life and works of our hero?

The playwright is at pains to mention every significant event in Turing’s life (ACE gets one rather pointless sentence) and it might have been an improvement to be more selective allowing time for more detail on what remains.

The music (recorded piano and cello + live voices) is in the musical theatre tradition but, while competently moving the story onwards, is unmemorable albeit it is sometimes redeemed by some rather beautiful harmonies. There is one joke but this is not a comedy. It tries, with some degree of success, to appeal to our emotions by depicting Turing as a troubled soul in conflict with almost everybody he comes across and that is certainly a point worth making. So then, a narrative focussed on the man, rather than on his works.

The production continues to tour the country, so there may be an opportunity to catch it in your area. Should you do so? Well if you’re a fan of musical theatre I fear you’d be disappointed. And if you want to know more about Alan Turing then you should beware, not all events depicted are true as per The Imitation Game (www.computerconservationsociety.org/imitation.htm ). But for a mildly interesting escape from an evening in front of the television, it serves its purpose.

  ★   ★   ★   ☆   ☆

50 Years ago .... From the Pages of Computer Weekly

BP to use Argus 500s for oil work: Monitoring of off-shore oil production in BP’s North Sea Forties Field With this sort of usage the company is claiming an MTBF of 5,000 hours and a minimum error rate of one in 100 million. (CW 20-27/12/1973 p8)

Hand-writing reader being studied by CCA: An investigation into the potential of a data entry system which can read hand-printed characters is being carried out by the Central Computer Agency on behalf of the Department of Health and Social Security. The system is called Grafix 1 and is manufactured by Information International Inc (III) of Los Angeles. The contract has been placed directly with III, but Manotek Ltd, of Gerrards Cross, Bucks, the company’s agent for Europe, is liaising with the CCA on the project. The CCA will test Grafix 1 by producing 2,000 forms completed by 200 different civil servants and containing about 40 different kinds of character. These will be sent to Los Angeles within the next few weeks. The results, which are expected in March, will be inspected by two government officials, one from the CCA and the other from the DHSS. A CCA spokesman stressed that the test is purely exploratory and that no specific application is yet being considered. (CW 374, 3/1/1974 p1)

Algol 68 compiler now in use; The compiler and run-time system devised at Liverpool University to allow Algol 68 programs to be written for the Computer Technology Ltd Modular One machine, is now reported to be fully operational, and in daily use at the university. The language is currently available on the Modular One in the university’s Department of Computer Science, and is operated under a multiple-access system employing 16 visual display units. As an aid to program development in Algol 68 and other languages, the department has also developed a MOP editor for use with the Modular One. The compiler runs under CTL’s own E2 executive, although, according to a university spokesman, it is written so as to be almost entirely executive-independent. Operated in compile-and-go mode, as at the university, the compiler requires 32K words of core in addition to backing storage. It is, however, said to be operable in batch mode in 24K without backing store. (CW 375, 10/1/1974 p7)

Honeywell launches data entry VDU: A new visual display unit, the VIP 7700, has been announced by Honeywell for use with both the 2000 and 6000 computer ranges. It is intended for data entry and retrieval and for interrogation. First deliveries are under way, and among the first UK customers will be the Glasgow Corporation, which has ordered 15 for use with its 6060. The VIP 7700, which is Honeywell designed and manufactured, replaces two earlier VIP models, the 775 and 785. It has all their features plus some new ones, but at a basic price of £2,160 costs considerably less; the 775 and 785 were both in the £3,000 plus range. There are three major new features. The first is the ability to display a form outline on the screen, leaving the operator to fill in variable information only. Secondly a dual magnetic tape cassette unit is supplied for storage of data from the display or from the host computer. Finally, an associated printer is supplied for hard copy of messages received, the contents of the cassette store or for output from the keyboard. Two models are available, the one printing 72 positions at 10 chps and the other 118 at 30 chps.   (CW 375 10/1/1974 p38)

Plessey aids target detection: The detection and monitoring of targets from the Royal Navy’s guided weapon destroyer, HMS Bristol, is centred on a Plessey Radar digital display system linked to a Ferranti FM 1600 configuration. The equipment forms part of the Navy’s ADA (Action Data Automation) weapon system. The Plessey system includes labelled plan displays (LPDs) with tracker-ball and keyboard, totes with tabular displays, and tactical displays which combine the facilities of the LPDs and totes. Each display console incorporates computer control and input devices. Some act directly on the display system itself, while others are used to communicate with the FM 1600 which stores the present state of control, and is programmed to formulate appropriate messages. The system has two distinct roles: firstly, to display radar information, to enable target detection and location, manual tracking and the initiation and control of automatic tracking; secondly, to display data stored in the FM 1600 to evaluate enemy threats and assign the required weapons. Radar data is received direct from sensors on board the destroyer, and some of the stored information in the FM 1600 is also derived from this source.   (CW 376 17/1/1974 p48)

Identicon launches data capture range: A complete range of automated data collection systems for stock and production control is now available in the UK from Identicon International, of Godalming, Surrey, the European marketing subsidiary of the Identicon Corporation, of Waltham, Massachusetts. The range includes both hand-held light pens and fixed-head devices for reading optical bar-coded labels and tags. The simplest device in the Identicon range is the Porta Pen data collector, which comprises a hand-held light pen and a portable magnetic tape cassette with keyboard data collection unit. This could be used for stock taking in warehouses and retail outlets. Another hand-held pen collector, the Identicon 300, is designed to interface directly with a computer, via a long cable and interface. These light pen devices read a simple black and white bar code. Identicon’s fixed-head readers can capture a variety of optical codes, including coloured codes, from tags on a wide range of different types of packages and parts passing along production lines and conveyor belts. The Identicon 100, for example, reads heavy duty reusable labels which Identicon claims can be read even if misaligned by up to 15 degrees in all three planes. (CW 377 24/1/1974 p20)

HP-65 nearest yet to hand-held computer: The nearest yet to a hand-held computer has been announced by Hewlett-Packard. The HP-65 programmable calculator has all the features of the existing HP-45 model, plus the capacity to accept pre-programmed magnetic strip cards, either supplied by H-P from a suite of standard software packages, or generated by the user using the HP-65 function keys. The HP-65 is aimed primarily at technicians in electrical engineering, surveying, medicine, statistics and mathematics. Some 200 standard programs, with up to 100 steps each are already available in these fields. These can be entered into the HP-65 in a few seconds by passing the magnetic strip through a slot feed on the side of the calculator, which is only slightly larger and different in appearance to the HP-45. The major external modification to the HP-45 is the re-assignment of the two top rows of function keys. The top five keys, designated A to E, act as labels. The parameters to which they are intended to correspond, in a specific program, are printed in the appropriate position on the face of the magnetic strip which is inserted as a guide above these keys after the program has been entered. (CW 377 24/1/1974 p56)

Pilot credit card system goes live: One of the biggest US credit card organisations, Bank Americard, part of the Bank of America, is now operating a pilot point-of-sale on-line credit authorisation system in Los Angeles. This involves 100 terminals installed at small retail outlets in the city and linked to the Bank of America’s massive Los Angeles computer centre. This is a possible future application for the Barclays Integrated Network System (BINS) being implemented in the UK by Barclays Bank. The terminals and the front end processor used at the computer centre are supplied by the Data Source Corporation of El Segundo, California, manufacturers of the System 800 credit terminal, which is designed to read embossed account numbers on credit cards. The terminals used in the Los Angeles system also read encoded data from magnetic stripes in the Bank Americard card, and this information, together with sales details entered via the terminal keyboard is transmitted to the computer centre for verification. (CW 380 14/2/1974 p8)

Microprocessor uses CMOS technology developed by RCA: A new microprocessor, said to be the first using Complementary Metal Oxide Silicon (CMOS) technology, has been developed by RCA at its New Jersey Solid State Technology Centre. The Centre’s director, Gerald Herzog, said that because of their reliability, low power requirements and simplified architecture, he expected the microprocessor chips to be used in low-cost, high volume systems, such as process control and automotive control devices, point-of-sale terminals and programmable calculators. The two-chip unit will provide systems designers with a CPU using a technology with a history of use in difficult noise, temperature and power supply conditions, he added. The chips contain a total of about 6,000 devices. The architecture is based on an array of 16 address pointers, each of which can be used as a program counter, for data storage or as data pointers, or to control an on-chip direct memory access capability. (CW 383 7/3/1974 p 18)

Airlines turning to automatic ticketing: The coming thing in airline reservation systems is automatic ticket issuing, and Air France is currently running experiments in one of its Paris offices. The airline hopes that the experiment will lead to an addition of this facility to the Apha 3 reservation network. At present the Di-An Series 8000 terminal equipment from Di-An Controls of Boston is favoured, but the Vogue 810 from Vogue Instrument Corp of New York is also under consideration. Last year the European Division of British Airways, then BEA, began trials with the Vogue equipment while the Overseas Division, then BOAC, tried out the Di-An units. In that experiment the Vogue system has now been dropped, but Di-An machines are still in use in New York and at the Victoria Air Terminal. The experiment has not progressed faster because European and Overseas Divisions are to introduce a new unified ticket in April. However, in July, a new and improved Di-An system will go on trial. This prints out the ticket in 10 to 15 seconds, compared with the 2 and a half minutes or so taken by the existing equipment. (CW 384 14/3/1974 p3)

NPL to work on EPSS interfacing protocol: Interfacing methods to link users’ computers and terminals to packet-switching exchanges will be one of the principal areas to be studied under a contract placed by the Telecommunications Division of the Post Office with the National Physical Laboratory. The contract puts on a formal basis co-operation on the Experimental Packet-Switched Service, (EPSS), which is being developed by the Post Office. The NPL has its own private packet-switched service and has for long been a leading advocate of this particular form of data transmission. A 48 kilobit link between the NPL network and the CAD Centre at Cambridge is set to go live by the middle of the year, and it is over this that the interface protocols will be tested. Other NPL work which the Post Office has already made use of includes the design of special high-speed equipment for timing and error checking of data being fed in to a packet-switched network. This latest contract at present covers about two men’s time for a year, but it is expected to be extended for two further years. Future work will involve the NPL in making sure that its proposed protocols and interfaces will be suitable for linking EPSS into an international network. Potential EPSS users, of which there are now 30, will be able to transmit and receive either packets of up to 2,040 bits of data, if they have intelligent terminals, or character-by-character data at up to 30 chps from simple terminals. The NPL is also involved in another major data network project, the European Informatics Network, or COST 11. Latest news is that tenders closed last Friday, and five bids were received, all of them from international consortia, for the design and construction of the experimental network. (CW 385 21/3/1974 p1)

Memory Typewriter for UK: A compact word processor unit designed to fill a gap in the extensive range of machines now on the market has been announced in the UK by the office products division of IBM. Known as the Memory Typewriter, the model is intended to serve the needs of companies with a workload not requiring the long term retention of texts. At its heart, the model has a 4,000 character memory which utilises the MOS/FET technology employed in the memory of some of the 370 series of mainframes. In addition, it has a 200,000 character magnetic tape storage loop, or action file, which is divided into 50 tracks. The other main element is the dual-pitch printer and keyboard used in the IBM Magnetic Card 82 word processor. Although only slightly larger than a normal electric typewriter, the Memory Typewriter embodies a variety of features not usually found in word processors. Error correction is effected by pressing the backspace key, which automatically lifts a mistyped character off the page and out of the memory, thus leaving a letter-perfect text both on the page and in the memory. When the memory is full, or the typist has completed the text, any one of 50 tracks in the action file can be selected and the text transferred by pressing the “Store” key. The transfer process takes about 7.5 seconds, including a parity check on the tape loop. Immediate access to any text in the action file is provided by the selector dial. Additions to the text are made simply by typing in appropriate words, and deletions by pressing the “Code”, key and then the “Character”, “Word”, or “Line” key. (CW 385 21/3/1974 p9)

DEC launches first microprocessor: The question which the entire minicomputer industry has been asking, “When will Digital Equipment go into microprocessors?” has been answered by DEC’s launch, in the US, of an eight-bit LSI microprocessor based on the Intel 8008 microprocessor chip. At the same time DEC has announced a minicomputer on a single circuit board. The new mini is an addition to the successful 12-bit PDP-8 range, and has been designated the PDP-8/A. The microprocessor development consists of a series of five modules known as the Microprocessor Series (MPS). These comprise a processor which can address up to 16K of memory; a semiconductor memory module which can be expanded from 1K to 4K eight bit words; a programmable read-only memory (PROM) module which can be expanded from 256 words to 4K words; an external event detection module which provides interrupt facilities; and a monitor/control panel for diagnostics and entering programs. According to the US publication, Electronic News, a basic processor plus 1K of semiconductor memory will sell for $745 in single units and $445 in 1,000 units, although these prices are unbundled. (CW 386 28/3/1974 p1)

New H-P processor to improve chemical analysis: A series of gas chromatographs built around a new design of digital processor which, it is claimed, will provide a chemical analyst with “better, faster and more complete answers than have been possible before,” has been announced by Hewlett-Packard of Slough, Bucks. To augment existing gas or liquid chromatograph systems, the company has also announced a new type of electronic integrator which employs much of the technology utilised in the gas chromatographs. Known as the 5830A, the gas chromatographs comprise two modules, a control terminal in one unit and the chromatographic system and digital processor in the other. The terminal provides interactive keyboard control of all operating parameters and has 38 function keys to set and control the operating conditions for a chromatographic run. Instructions are input to the processor via the keyboard. The processor then monitors detector output, reduces it to peak areas and retention times factors, identifies sample components and computes their concentrations. A printer/plotter unit on the terminal continuously outputs a chromatogram and, at the end of a run, a complete analytical report. (CW 386 28/3/1974 p3)

Bar code reders for goods routeing: Automatic routeing of goods in warehouses and factories are the main applications for a range of laser beam-based bar code reading equipment now available in the UK from Techex of Poole, Dorset. Known as the Decodertran, the range is manufactured by Computer Identics Inc of Westwood, Massachusetts and is already being used in the UK at the Peterborough warehouse of Freemans Ltd, the mail order group. Goods being controlled by Decodertran are affixed with either permanent high quality reflective labels or paper labels that can be produced on a modified line printer. These are printed with binary coded decimal black and white bar code, 2.6 inches long by 1.25 inches high with zeros and ones represented by bars of varying thickness. In practice, the label is affixed vertically to the item and read from top to bottom by the scanner. The fixed-head laser reader, located alongside a goods conveyor, is linked to a hardwired processor that decodes the label data and actuates the relevant electro-mechanical routeing devices. (CW 387 4/4/1974 p3)

PoS numbering system study by grocers: According to John Kimball, chairman of the Safeways supermarket group, the present prohibitively high cost of point-of-sale terminal hardware will be one of the major factors to be taken into account in a feasibility study into the possible introduction of a Standard Product Numbering System which has now been formally announced in London by the Institute of Grocery Distribution. Mr Kimball is chairman of the SPNS steering group which has appointed consultants, McKinsey and Co, to carry out a 10-week study of the UK grocery trade, which will involve collecting information and opinions from retailers, wholesalers and manufacturers for processing with McKinsey-developed software using, probably, the Honeywell dialup time sharing network. Cost will be the vital factor, and objections in this direction among grocery outlets are likely to involve factors such as high capital cost of terminals and tag readers, depreciation, and computer maintenance and supervision costs. On the manufacturing side, objections will include product code pre-printing problems involving factors such as printing tolerances and new quality control equipment and procedures. Benefits to retail outlets are expected to include speeding throughput at the checkouts, which in turn should increase sales turnover while reducing the number of checkout staff required and, probably, the error rate in ringing up items. (CW 387 4/4/1974 p9)

1903T aids prevention of road accidents: One of the main applications of Berkshire County Council’s ICL 1903T installed at its offices in Shire Hall, Reading, is in helping to prevent road accidents in the county by highlighting sites where accidents are increasing, and pinpointing the reasons why. The team responsible for capturing data for this invaluable project is a unit within the County Surveyor’s Department, based at Kennet House, Reading. The Accident Investigation and Prevention team extracts relevant information from a police report taken at the scene of the accident, examines conditions on stretches of road which are indicated to be high risk areas and generally investigates why accidents occur at various points. Data is fed directly to the 96K 1903T through a Kode terminal in the team’s offices. The actual software involved includes programs developed by the Buckinghamshire and Oxfordshire county councils in 1966 for similar work. (CW 388 11/4/1974 p3)

Digitising Ordnance Survey Maps: With some large local authorities making up to 500 copies a day of Ordnance Survey maps of their areas, there are very pressing reasons for making the whole system of map production as efficient as possible. At present, almost all such maps are printed from transparencies, but two years ago the Ordnance Survey at Southampton began the immense task of digitising the large scale 1,250 and 2,500 series – almost a quarter of a million maps in total. Now the Survey is coming up to the 1,000 mark, but five more Ferranti Freescan digitisers are on order to operate alongside the nine already engaged on this work. Enlargements of the maps, in the form of negatives, are digitised onto magnetic tape, which is then transferred to the Survey’s ICL 1904, for validating, editing, addition of text, and storage The final maps are output for plotting offline on a flatbed Ferranti master plotter, which prepares a positive which with minor alterations can then be used as a lithographic master. (CW 388 11/4/1974 p10)

Babbage gives high level boost to 4080: The tendency among mainframe manufacturers to write systems software for their newer machines in high-level languages, has now been extended to the minicomputer area, with the advent of the Babbage language for the GEC 4080. The language, conceived as the lowest level of coding for the 4080, is termed a “high-level assembler”. While permitting such advanced facilities as the definition of vectors (one-dimensional arrays) and the accessing of their individual elements, Babbage also has low-level capabilities, for example, direct user access to registers. The language is, moreover, specifically oriented to the architecture of the 4080 hardware. Some valuable ideas for the development of the language were provided by Prof Hoare of Belfast University, who has collaborated to a great extent with Prof Niklaus Wirth of Zurich. Babbage hence shows some affinities with Prof Wirth’s languages, PL/360 and Pascal. The use of a high-level language simplifies the writing and modification of software, usually at the expense of some efficiency. GEC claims, however, that a Babbage program, even when written by a programmer knowing nothing of the 4080 machine code, can display 75 per cent of the efficiency of a program written directly in machine code. This, it is said, can increase to 100 per cent when the programmer is acquainted with machine and compiler properties. (CW 389 18/4/1974 p2)

Honeywell announces new range of computer: In what must surely be one of the most important mainframe announcements of recent years, Honeywell has introduced Series 60, a range of nine computers developed in Europe and the US, which together cover the entire spectrum from small business systems to large-scale multiprocessor configurations. In thus unveiling the shape of its systems for the next decade, Honeywell has firmly underlined the importance of the user in design-thinking, and this may eventually be recognised as the most significant fourth-generation characteristic. Steve Jerritts, managing director of Honeywell UK, has outlined the thinking behind this strategy. “Eighty or more per cent of our world-wide revenue now comes from existing users. Consequently we are bringing together users of our; present diverse product lines into the single mainstream of Series 60. ” In the short term this complete compatibility between the old and the new is likely to win the company plenty of loyalty. In the long term this will be strengthened by the user oriented facets of Series 60 design, such as detailed maintenance and reliability features, security mechanisms and a host of smaller details such as the carefully-designed consoles. From a technical viewpoint there are no great surprises in the Series 60 specification, though many f eatures usually associated only with very large systems are available throughout the range. The most interesting development is the extensive use of firmware, or read-only memory programmed to perform functions traditionally incorporated in software. (CW 390 25/4/1974 p1)

Forthcoming Events

Members and others are welcome to attend CCS Seminars but these are also available via Zoom.

London Seminar Programme

London meetings take place at the BCS — 25 Copthall Avenue Moorgate EC2R 7BP starting at 14:30. The venue is near the corner of Copthall Avenue and London Wall, a three minute walk from Moorgate Station and five from Bank.

You should use the BCS event booking service to reserve an in-person place at CCS London. Go to www.computerconservationsociety.org/lecture.htm . The service must be used for remote attendance.

For queries about meetings please contact the CCS meetings secretary, Roger Johnson at

North West Group contact details

Computer Conservation Society

Aims and Objectives

The Computer Conservation Society (CCS) is a co-operative venture between BCS, The Chartered Institute for IT; the Science Museum in London; the Museum of Science and Industry (MSI) in Manchester; The National Museum of Computing in Bletchley and the Bletchley Park Trust.

The CCS was constituted in September 1989 as a Specialist Group of the British Computer Society (BCS). It thus is covered by the Royal Charter and charitable status of BCS.

The aims of the Computer Conservation Society (“Society”) are:

• To promote the conservation, restoration and reconstruction of historic computing systems and to identify existing computing systems which may need to be archived in the future;
• To develop awareness of the importance of historic computing systems;
• To develop expertise in the conservation, restoration and reconstruction of historic computing systems;
• To represent the interests of the Society with other bodies;
• To promote the study of historic computing systems, their use and the history of the computer industry;
• To publish information of relevance to these objectives for the information of Society members and the wider public.

Membership is open to anyone interested in computer conservation and the history of computing.

The CCS is funded and supported by a grant from BCS and donations. Some charges may be made for publications and attendance at seminars and conferences.

There are a number of active Projects on specific computer restorations and early computer technologies and software. Younger people are especially encouraged to take part in order to achieve skills transfer.