Elliott 405 – a fascinating glimpse into vintage computing

History isn’t just about dates and the events that get memorialised in statues. Far more fascinating is the personal and the quotidian. A thumbprint left in an ordinary ceramic bowl is a more direct connection to the lived experiences of people the past than any number of crowns and sceptres.

And I think there’s an aspect of even recent history that we can easily overlook. In the story of how computers have evolved, for example, it’s too easy to get concerned only with the technology – and its occasional quaintness – and miss how the developments in data processing impacted lives.

You can still see the evidence of Big Doug’s promotion. Why waste a good nameplate?

What inspired this philosophical diversion is a package I’ve just received from my dear friend Doug Selway – artist and sage. The package contained a small archive of documents belonging to his father, who is sadly no longer with us.

SD Selway – known to most, including the many people who loved him, as Big Doug – worked for the GPO, the UK’s nationalised postal service which also provided things like the country’s telephone, telex and other communications services. He’d worked his way up, via service to his country in the Signals Corp, from being a ‘boy messenger’ to, at the end of the 1950s and into the 1960s, being the manager in charge of the London Electronic Agency for Pay and Statistics (LEAPS) computer centre, part of the Accountant General’s Department (AGD).

Elliott computers

The LEAPS operation ran not one but two National-Elliott 405 computers for managing payroll and generating statistical data.

An Elliott 405 at the London headquarters of National Cash Register (NCR).

Elliott Brothers was a highly influential firm in UK computing history. It would, through later splits and mergers, form important parts of both ICL and GEC.

The National-Elliott 405, released in 1956, was the pinnacle of the 400-series, which started with the 401. Like all computers at the time, it was a valve-driven, room-hogging monster which used 33-bit words (one bit was used for parity).

The technology looked very different from what we’re used to today. For example, immediate storage (vaguely equivalent to today’s RAM) was provided by nickel delay lines. Essentially, bits were encoded in a looped length of wire by twisting it – a process dubbed magnetostriction. The twist (representing a bit) worked its way along the wire to be read by a sensor at the far end. Delay lines had some similarity to the modern dynamic RAM (DRAM) that’s in your computer now. It constantly needed refreshing. As each twist (or none-twist) worked its way out of the end of the wire, you had to write it back to the start again, assuming you didn’t want to change that value in memory. (Another common form of delay line used audio pulses running through tubes of mercury.)

The bits flew around the wire at a rate of 600 times a second. Each wire could hold just one word of data, so we’re not talking about gigabytes here. As far as I can work out, the delay lines were used for something akin to what we would think of as stack memory today.

Unlike modern DRAM, delay lines were not random access. To read a word, you had to wait for the start of it to come around in the refresh cycle.

For more storage, you’d turn to an 8½in magnetic drum plus 19¼in magnetic disks and tape – the latter being 1,000ft reels of mag-coated 35mm film. A reel could hold around 300,000 words of data in 64-word blocks. The disks were smaller, holding either 16,384 or 32,768 words, again in 64-word sectors. And the drum was smaller still – at 4,096 words. There’s a lot more detail in an excellent, in-depth document on the Elliott 400-series computers here (PDF). And we’ll get into further details in subsequent posts when we look at the individual documents.

A couple of these machines are now preserved as museum pieces. There’s a great piece of footage showing a 405 being installed at , manufacturers of cleaning products, in Hull, 1959. It features lovely old lorries, men in white coats and more health & safety violations than you can count. The film is silent, but I’m sure you can provide your own Mr Cholmondly-Warner voiceover.

What’s in the package?

The archive is small but fascinating. There’s a brochure for the Elliott 405; a quaint eight-page booklet entitled An Introduction to Binary Arithmetic; and a small, folded cheat sheet giving stats about the 405 and a list of its opcodes – which aren’t many.

There’s also a copy of the March 1959 issue of Automation and Automatic Equipment News magazine. This includes a surprisingly long article about the LEAPS computers – surprising because they hadn’t yet been installed and also because the article is essentially a report of a meeting. Yet it contains some real gems of information – not just a few technical nuggets but also a fascinating insight into the concerns about machines taking people’s jobs and deskilling work (the meeting was attended by, among others, representatives of trades unions). This is even before the 1960s had steamrollered into view. I’ll be posting a PDF of the article in a future post.

Most interesting of all, to me, are pages of notes that appear to be Big Doug working on programming problems – getting to know the machine and how to use it. These are all handwritten. Some are labelled as ‘revision’ exercises and may actually have been exams of some sort (on one sheet he admits to having run out of time). What you can see on these sheets – many of which are custom programming sheets for the Elliott 405 and contain code – is someone learning, taking on board the new ways of thinking that came with the digital age. There’s that direct connection here with the human experience that is so often missing from more mechanistic histories of computing.

Along with the documents, the package contained one very intriguing piece of 5-bit punch tape. What’s on it? Who knows?

Pitiless clarity

Finally, there is a delightfully yellowed piece of paper on which someone (Big Doug?) has typed out a quotation. It’s from the book Electronic Business Machines edited by Joseph Harry Leveson (not ‘Leverson’, as typed) and published by Heywood & Co, London in 1959.

It says:

“Indeed, no attribute of the electronic computer is more startling than the pitiless clarity with which it exposes, and proclaims, the capabilities and shortcomings of those concerned in its use.”

Well, nothing much has changed, has it? Pitiless clarity is still the computer’s greatest power.

As I create PDFs from this material, I’ll be making them available in future posts. A big thanks to my friend Doug for entrusting me with this treasure. Stay tuned…

» LEAPS Elliott 405 page »

6 thoughts on “Elliott 405 – a fascinating glimpse into vintage computing

  1. Ed S

    I would think that the delay storage corresponds most closely with today’s registers, the drum storage with RAM, and the tape and disk storage with our filestores.

    The National Museum of Computing (on the Bletchley Park campus) has an Elliott 803 which is often seen running, and also a 903. There’s an upcoming event (Feb 2022) where you can join in and learn to program the 903.

    Peter Onion curates and demonstrates the 803 and is often there for a deep-dive chat.

    Reply
  2. Dan Johnston

    In 1967, I was studying Electrical Engineering at The University of Queensland, Australia, and obtained vacation employment from a couple of months from late November that year with the CSIRO in Sydney, who had been given a National Elliot 405 – probably the same one of which parts are now in the Powerhouse Collection. I believe they were given it without any software, and my duties there over a couple of months were to write a basic assembler for the machine, which I did. I believed this was complete, but never got to run it. I think the hardware was still being debugged, but my memory is pretty vague after that many years. Goodness knows how buggy the code is. It was written to fit in 192 words of memory. I still have the handwritten code, as well as a printout of the same., so can give you a serious attempt at writing code for this machine if you would like to see it.

    Reply
  3. Dan Johnston

    My understanding was that magnostriction did not involve twisting the wire but rather applying a magnetic field to it, which caused it to contract and this contraction passed along the wire and was picked up at the other end. (My memory of what I was told when working on the machine in 1967, supported to some extent by current Google searches.)

    Reply

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.