Magnetic core memory was a critical innovation in the development of computers. Per https://en.wikipedia.org/wiki/Magnetic-core_memory it was the most advanced form of memory from about the mid-fifties to early seventies.

For development of a semi-steampunk setting, I'm thinking about how much earlier it could've been developed. (Setting is otherwise largely similar to our own world in relevant aspects of history, geography, economics etc. I'm just trying to figure a way for it to have early computers powered by e.g. electromechanical relays.)

What are the prerequisites? Clearly it needs electricity and a fair degree of proficiency in processing electrical signals, as well as the ability to mass produce identical parts. What other difficulties are there? Could a well-equipped R&D group that stumbled onto the right path, develop practical core memory at the 1900 tech level, say?

  • 2
    $\begingroup$ So... I thought that steampunk was specifically about a world in which we didn't have the Maxwell equations and electricity as an energy transmission and utilisation method didn't exist. As such, if you have electrical memory, by definition you don't have steampunk. If you want a steampunk computer, look up China Mieville's Perdido Street Station in which he describes computers and automata that operate via thousands of miniature steam valves that act in a similar manner to transistors to create logic gates. But magnetic core memory (IMO) can't be reconciled with steampunk. $\endgroup$ – Tim B II Apr 17 '19 at 23:48
  • 2
    $\begingroup$ The concept of a ferrite ring getting magnetized and demagnetized could be stumbled upon in 1900. However, there would not be a path forward to a functioning memory without a number of other advances in electronics. $\endgroup$ – Alexander Apr 17 '19 at 23:54
  • 2
    $\begingroup$ Pushing back technology is no small thing. Read this meta question and its answers for more insight. It's doubtful that MCM could be pushed back more than 15 years. $\endgroup$ – JBH Apr 17 '19 at 23:55
  • 2
    $\begingroup$ semi. Love it, increasing accuracy by reducing precision. :) +1 from me, it's a good question, sorry about the nit pick. $\endgroup$ – Tim B II Apr 18 '19 at 0:02
  • 1
    $\begingroup$ @rwallace core memory is easy to manufacture, but what circuitry would you utilize to actually use it? $\endgroup$ – Alexander Apr 19 '19 at 19:15

This is outrageous, and the more I think about it, the more wondrous it gets

For all intent and purpose, the electromechanical relay was invented by Joseph Henry in the late 1820s. If that name sounds familiar, it should. The SI unit for magnetic inductance was named the Henry in his honor. But what's amazing is that electromagnetic relays were not used for computing for a century! Heck, you didn't see them in telegraph machines until the 1860s.

But! What if enamel-coated wire could be manufactured to a smaller gauge and more economically!

It was the height of the Industrial Revolution! Factories were popping up everywhere! If Hollywood can be believed, Dr. James Moriarty is poised to take over the world! And all someone had to do was draw that magic line between the analytical functionality of Babbage's mechanical wonder — and a relay that would take up a fraction of the volume.

Boom! Electromechanical computing in the mid 1800s.

But, why is that important?

Because the most common reason any technological advance took place when it did is that there was a reason to look for it. This is really important! In many cases, it's not the technology part of the tree that's causing an advancement to take place when it did, it was the fact that a need for it finally arose. Necessity is quite often the mother of invention.

And if we have serious computing going on in the 1800s, we have the need for memory in the 1800s (not just registers, but serious computational memory on the order of whole Kilobytes!) Electomagnetic relay computing would do it, but it's also important to understand why.


Babbage's mechanical wonders had memory. Persistent memory. If you stopped turning the proverbial crank, all the gears stopped where they were and whatever had been stored would be remembered — forever. Why would anyone want memory that would, eventually, degrade?

Because you can only get so much out of spinning gears. Relays are fast! Lightening fast! And the best way to take advantage of that speed is to have memory that's just as fast as the relays are!


The necessary technological advances that would bring about magnetic core memory already existed in the mid 1800s. Electricity and magnetism. What lacked was a reason to even think about the need for this marvelous thing called "memory." And that's what we've provided by bringing electromagnetic computing to the fore a century before it did.

But even this requires a reason. Maybe. That's the wonderful thing about stories. You don't really need a reason for anything. But you might want to consider why you need faster computation in the mid 1800s. It's not like they were putting people into orbit (seriously, go watch the movie Hidden Figures or read the history of those amazing ladies. NASA didn't need computers as we know them today for quite some time, because human computers were fast enough. It wasn't until they needed the data faster that they started replacing the people...). What in your story needs fast computing? If you answer that, you've justified electromagnetic relay computers, which justifies researching memory, which brings MCM into the fore a century early.

In my comments I mention that MCM likely couldn't have been brought to light more than about 15 years before it did. This is because of this dependency on the need for speed. MCM was dependent on fast computation. Bring about an early reason for fast computation and you speed up a lot of things.

  • $\begingroup$ Artillery firing tables were an important early application of computers. Would they not provide an application of fast computation in the nineteenth century? $\endgroup$ – rwallace Apr 18 '19 at 0:39
  • $\begingroup$ @rwallace, no. Tabular data is trivial, and you only need to calculate it once and then print it in books. If you think about it, it was an early application for computers - and yet computer tech didn't evolve any faster than it did. I'm afraid a better reason must be found (which would make a great bit o' window dressing for your story). $\endgroup$ – JBH Apr 18 '19 at 1:45
  • 1
    $\begingroup$ I can think about only one sector that could use fast computation in the 19th century: Government. If China were leading the industrial revolution, that would bring a huge populational boom to an alredy large population, the data processing needs of such a society could exceed the capabilities of factories filled with people doing calculations (like the Napoleon's census used). Combine that with telegraphs for fast data exchanges and you may have an application and need for fast computation. $\endgroup$ – Geronimo Apr 18 '19 at 17:43
  • 1
    $\begingroup$ One application that needs to perform firing calculations fast, in real-time, is a fire-control computer. Historically, mechanical computers were able to make these fast enough by machining a three-dimensional solid on a lathe that functioned as a sort of lookup table. $\endgroup$ – Davislor Apr 19 '19 at 7:30
  • 2
    $\begingroup$ And the classic supercomputer arms race, ever since the invention of the electronic computer, has been cryptography. $\endgroup$ – Davislor Apr 19 '19 at 7:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.