Jeff Duntemann's Contrapositive Diary Rotating Header Image

The Original Hard Disk

I know I have an original Bernoulli Box 10 MB cartridge somewhere. I saved one because I was sure that nobody would believe me when I waved this huge slab of plastic at them and said, “This holds 10 MB.” I think I was right. I won’t know until I find it. So I looked…

…and I found something else instead. See the photo above. Anybody recognize it? It’s part of one of the earliest hard drives in computer history. It’s such an old hard drive that it wasn’t even a hard drive. No. It was main memory.

I honestly don’t remember who manufactured it, though Sperry-Univac sounds familiar. (I threw out the case when we left Rochester in 1985.) I bought it at a hamfest for a couple of bucks just for curiosity’s sake; the old guy I bought it from said it was a hard disk, and I was skeptical. He was right, and it was probably fifteen years until Google allowed me to look for an explanation. I took the case apart and found a motor, a fairly heavy aluminum disk coated on one face with red iron oxide, and the assembly above, which has 160 magnetic heads in eight spiral groups. Roughly half the heads look like gray ferrite, and the other half like white ceramic. I’m guessing that the gray heads are write heads, and the white heads are read heads. (It could be the other way around.) The whole assembly is 9″ in diameter; the magnetic disk is 8″ with the outer 2 1/4″ coated with oxide.

So what we have here is a device that imposes 80 tracks on 2 1/4″ of oxide. The number of bits on each track remains a mystery. 1000? 2000? Somewhere in there, which suggests a total of 100,000 or perhaps 120,000 bits, which would provide about 16KB. This was RAM, not mass storage. The principle is basically the same as the magnetic drum memory systems that IBM sold with its vacuum tube machines like the 650. There were no moving parts other than the disk itself, and you can spin a disk faster than you can spin a drum. I’m guessing that the magnetic disk units like mine filled the (narrow) gap between magnetic drums and core memory. I’ve seen writeups indicating that magnetic disk storage was used for swap storage as late as the PDP 11/45 in 1972. The unit I have seemed a lot older than that.

One of the problems in researching a unit like this is that “hard disk” and “magnetic disk” have other, more modern meanings. So what needles may be are lost in the titanic haypile of newer technologies. If any of you know anything more about the technology send me a quick description or links, and I’ll post them in a future entry.

I had 8″ floppies in my first CP/M machine, and 5″ floppies in all my PCs until the 90s. I used SyQuest cartridge hard drives for years after my 1986-1992 romance with Bernoulli Box drives, which drove me nuts with their constant indexing of heads across the medium. (Tick…tick…tick…) The SyQuest cartridges spun fast and died young. There followed three generations of Zip drives, culminating in the monumentally awful Jaz. It was only in 2004 that I set aside moving parts in my removable storage, and began using thumb drives. I’ve had an SSD on my main system for eighteen months, and will be putting them in most of my lab machines as well, spurred by the need to get XP out of daily use here.

It’s sobering to remember: We’ve been spinning magnetic disks for what may well be sixty years now. We’re still spinning them, and we will be for another ten years or so at least. What other computer technology has been in wide use for that many years? Tape–maybe. That’s the end of my list.

Damn. That’s a lot of angular momentum. Will I miss spinning disks when they’re gona entirely?

Hah. Guess.


  1. Bill Meyer says:

    I have an old friend and colleague in California who used to talk about when he worked for a SoCal company that was doing something with “head per track” disks. This was in 1975, a year before I got my IMSAI, and I nothing of such exotica. I will have to ask him whether he remembers any of the particulars.

  2. Carrington Dixon says:

    Back in the 1970s I programmed on a Texas Instruments computer, the Advanced Scientific Computer, that had head-per-track disks. These disks were faster than the common, positioning arm disks in part because there was no seek time while the read/write head moved to the desired track. Of course, there was still rotational latency, the time it took for the start of the desired record to appear under the heads. As I recall they were big and expensive; eventually TI had to add support for convention disks as the HPT disks were so exenpsive.

  3. John Flood says:

    In the late 70’s the company I worked for had a PDP11/70 running UNIX V6 and a 1 megabyte head-per-track drive used as a swap disc. Doesn’t sound like much but with 128K core that’s a fair amount of swap space. When you’re NOT doing any graphics you can get a lot done with not much.

    1. Well, if the graphics are lousy enough, you can do those too. In 1977 I wrote a paint program for the COSMAC Elf in 256…bytes. It ran in Byte magazine, and I have a photo of Phil Foglio drawing with it. We didn’t know how bad we had it in the 70s.

      1. Lee Hart says:

        More to the point: We don’t know how GOOD we have it today!

        To write a Paint program in 256 bytes, first you must not know that it’s “supposed” to need far more. We can only do the impossible when we don’t know that it’s impossible. 🙂

      2. Tom Roderick says:

        I am pretty sure I remember seeing that program in Byte and thinking about trying to make it run on my RCA COSMAC VIP, but I sure don’t remember ever succeeding! I may still have that issue of Byte down in the shack. I kept most of the early issues and indexed and boxed them.

        1. It was called “The COSMAC Doodler,” and I wrote it for the Elf, not the VIP. I had one of those too, and had an article published in Kilobaud called “Teach Your VIP to Speak Elvish,” which was basically an expansion board with toggle switches to make it easier to toggle in “naked” Elf programs.

          I queried Popular Electronics in 1980 for an article called “The COSMAC IMP,” which was about building a little printer with a cash-register print unit and some glue logic. PE’s editor said courteously (and with some justification) that the Elf’s day had passed. I still have the IMP printer, and am still a little proud of the design work that went into it. By 1980 I had a big hulking CP/M machine and knew which way the winds were blowing.

          1. Lee Hart says:

            (…the Elf’s day had passed…) Well now, I wouldn’t say that! (an old radio show reference)

            I’m *still* selling COSMAC Elf computers (see This is a credit card sized 1802 computer, with a real front panel, serial and parallel I/O, and 32k of RAM and ROM.

            PS: Guess who that is in the illustration at the top of the web page. 🙂

          2. I wouldn’t say so either, especially since I have an unfinished Membership Card on my workbench. (I set it and all other hardware projects aside last summer when I turned my attention to this book project. It will probably be the first one I return to when the last chapter marches out the door next month.) It’s boggling how much binary code still comes easily to mind: F8 FF A2 etc. and that wonderful pair 7A and 7B.

  4. Beats me. The oldest hard drive I still have around is an original 5 MB unit in an IBM PC/XT, which also has an original Hercules Graphics Card. As far as I know, the system would still boot if I fired it up.

    I originally intended to rebuild the system on its 30th birthday to use the same case but with a Core i7 Extreme and similarly high-end video, disk, and so on, with a slot-load DVD burner behind the slot in the original 5.25″ FDD. I haven’t gotten around to doing that yet. Maybe for its 35th birthday.

  5. Stickmaker says:

    That looks like something Ronco used to sell for use in the kitchen…

Leave a Reply to Jeff Duntemann Cancel reply

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