Book review: A computer called LEO

I’ve just finished the book “A Computer Called LEO“, by Georgina Ferry, and it was a great read. I can thoroughly recommend it. It’s not new either so if you like secondhand books you can probably pick a copy up cheap. It’s a book about the early days of business computing, but don’t let that put you off.

Lyons teashops, which I don’t really remember (the last closed in 1981, when I was 8) were apparently MASSIVE before the second world war. The parent company Lyons were innovative, in their business practices and the way they approached problem solving. In 1946, two managers persuaded their boss to let them visit the USA in order to look into computers (“electronic brains”), and the manager agreed. At this stage, there were no business computers worth mentioning, indeed, there weren’t really many computers at all. They came back convinced, went to visit Wilkes in Cambridge, seconded people to Wilkes’ team, and then got on with the job of building a business-friendly version of the Cambridge computer EDSAC.

One of the interesting aspects of this book is the clarity of difference between a scientific computer and a business machine: at the very start, the plan was to build something with better data input and output (so it could do a lot of simple calculations, rather than the small number of mathematically intense calculations the Cambridge team were aiming for) and something which was reliable. Reliability came through modularity: valves are unreliable and finding one broken valve from 100s is a challenge. So the Lyons Electronic Office (LEO) had removable units each holding 28 valves, which could be swapped out in bulk, repaired, then replaced, drastically reducing the down time of the machine.

Memory was a huge problem for the entire nascent field of computing at this time, so with LEO they experimented with magenetic wire and mercury delay lines. I first came across the idea of mercury delay lines in the Neal Stephenson novel Cryptonomicon, and to be honest, until I read about their use in LEO I’d assumed they were fictitious they’re such an absurd tech.

The problems they had to solve were many and varied. How to do input? How to do storage? How to print? (There weren’t any printers at this stage).

LEO was working reliably by 1952 and was announced to the world in 1953. It managed payroll, ordering and all sorts of business function within Lyons. Others were buying time on LEO, and so Lyons decided to build more – to move into the computer business.

The second half of the book is about this, and the technical achievements of the LEO team.

You can probably guess what happened, because we’re not all typing on LEO Laptops right now, but the story is interesting and well-told.

Programming was a major issue as there was no such thing as off-the-shelf business software. There wasn’t even such a thing as a programming language, really. The first payroll software was probably written for LEO. The first ordering software, ditto. LEO programmers would spend huge amounts of time working out the business logic behind how a company worked, and then work out how that could be modelled in software taking into account the limitations of the machine (e.g. minimising the amount of new data input each run).

In the end…

However good your computer, you have to sell it. Making the first sale of LEO III in competition with rivals, principally IBM and the largest of the British companies, ICT, was a nerve-racking business for the LEO team, even though they knew they had a better product.

A Computer Called LEO, Ferry p167

All in all I’d recommend the book highly, if you’re interested in computers, computer history, or business.


  1. In the early 80s I worked in a development team with somebody who had previously worked on LEO.
    They seemed to have a lot of female developers.

  2. That takes me back. Computer history was part of the WJEC O level syllabus when I did it.

Leave a Reply

Your email address will not be published. Required fields are marked *