Skip to content

Calculator languages

January 5, 2012

Kraftwerk’s song from their classic Computer World album is the soundtrack to the post:

This post is to celebrate the re-introduction of the two of HP’s “Voyager” calculators after 30 years:  the HP12C 30th Anniversary Edition and the HP15C Limited Edition.  These calculators are not so easy to buy – the HP15C has only been available for a few hours at a time on HP’s web site before selling out – apparently there is huge demand for it.  (If you do decide to buy one of these calculators, be sure to use a coupon.)

These calculators date back to the “good ol’ days” of HP – when it was a company known for super-high quality devices rather than shoddy printers and corrupt business practices.

The “Voyager” calculators are among the most beloved calculators ever released by HP.  Smaller than a modern smart phone, they featured a sideways format, an LCD display, and an elegant 39 key layout.  The design of the calculator is famously tough – one site claims that “One HP-12C was used by a zoo keeper to calculate feed mixtures. The zoo keeper dropped the calculator and it was consumed by a hippopotamus. The calculator survived the hippo’s digestive process as well as the washing that followed. “  The same site claims that the battery life was exceptionally long-lived: “The record so far is 22 years on the original set of batteries and that sample is still running!”

The HP 12C business calculator  is the most popular of the series – it has remained in production for 30 years in one form or another.  In addition to time value of money computations, this calculator offered odd period calculations (allowing interest to accrue for some arbitrary amount of time before the first payment); bond calculations; and depreciation calculations.  Why does it sell so well?  Some of the theories are based on the calculators looks and rugged construction:  “It’s good (and expensive) looking”; “Like all 10C series calculators, it has a nice solid built-like-a-brick feel that clamshell models can’t quite match”; “It has become part of the well-dressed business uniform – easily distinguished from cheap calculators due to its layout.”


Another celebrated entry in the series is the HP 15C scientific calculator.  HP produced some less capable scientific calculators (10C, 11C) as part of its Voyager line, but the 15C included features such as numerical integration, a root solver, support for complex numbers, and matrices.  Later HP calculators would add even more features, but at the cost of size, complexity, and reliability.


A more unusual entry was the HP 16c programmer’s calculator.  This entry featured extensive ability to manipulate numbers in different base systems.  Sales were reportedly poor, but this calculator has a large fan base today and good quality samples can go for several hundred dollars on eBay.


The new Limited Edition calculators mentioned above are cheaper (in real dollars) than the original releases, and they are considerably faster (reports indicate that they 30-100 times faster.)  Notably, they try to preserve almost identical executable code with the old models (which somehow hit a sweet spot in a nearly perfect design).  The calculators look good, and they handle fairly well too.

The new replacements are not actually made by HP – they are made by Kinpo Electronics under license from HP and sold under the HP name.  Still, with these models, Kinpo has attempted to create an experience as close to the original Voyagers as is possible.  One area in which that is not true is battery life.  The battery life on these models is sharply reduced – batteries on the new limited edition calculators last only a couple of months, not a couple of years.  A pity.

It is worth pausing  just to note the growing popularity of retro-tech.  Chrysler is releasing new cars for 2013 under the name “Dodge Dart.”   Commodore is releasing a modern day “Commodore 64.”  And you can still buy a Bolex H16 today – apparently there is a market for a high quality hand cranked motion film camera.


Now, what possible relevance could a post on pocket calculators have for a blog on language and translation?  More than you might think!  It turns out that each calculator comes with its own “language”, and that “language” has a lot to say about how users structure and even think about problems (yes, Sapir-Whorf really does have some application to calculator languages!)  I’ll explore that topic in a future post….

10 Comments leave one →
  1. January 5, 2012 3:14 pm

    Just to mention….

    The HP RPL languages are still heavily used. RPN is still a demanded feature in most computer calculators and among the older crowd HP like programs are run on high end phones to emulate what an HP calculator would be like today.

    As far as language I don’t think the voyager line hit the sweet spot. I used the 15C and of course the 12C but I was a 28S guy). I notice the 1 so I won’t jump to early on the discussion of languages except to say that I think RPL accomplished what it set out to do capturing the advantage of an RPL/Forth language and the advantages of a LISP.

    Call stacks are now standard in every CPU so from a performance standpoint it is very harmful to mix the call stack and the data stack the way RPL does. But the idea of what a computation language should look like lived on. A series of variables defined in terms of values and functions calling each other.

    Consider excel, and remove the GUI. Essentially it is a series of definitions
    A1 = 4
    A2 = 23
    A3 = A2 – A1

    B52 = sum(b4..b26)

    excel builds its own solution tree from those definitions. And this kind of circular solver is a major enhancement, but if we remove that as well you get something very much how RPL worked.

  2. January 5, 2012 4:17 pm

    CD: I haven’t started talking about languages yet, so I’ll defer comment on most of your remarks until later.

    But in terms of physical presentation, the HP 18/28C/28S series was kind of junky, in my opinion. The clamshell was bulky and fragile, You could not slip a 28 into a pocket the way one could with a Voyager calculator.

    Even the page you linked to indicated that battery door was fragile. The cable connecting the two halves of the clamshell was prone to wear and would eventually break. I notice that HP abandoned the clamshell design in later calculators such as the HP 48.

    There are plenty of early Voyagers that are still working today, but members of the 18/28C/28S series have not survived the decades as well.

    Another issue was that the 18/28C/28S were (I think) the first of the menu-driven HPs. By switching to a menu-driven interface, the amount that a user needed to remember grew tremendously. Of course, HP produced great manuals back in the day, but what’s the point of carrying around a pocket calculator if one needs to bring a manual with it?

    A final issue was the quality of the numerical calculations. HP had retained William Kahan (UC Berkeley) who had famously criticized the quality of numerical calculations on the HP 35 (which were often rather inaccurate). Kahan ended up designing the algorithms for the Voyager series, and they set the standard for numerical accuracy on calculators. The later 18/28/48 series fortunately kept most of Kahan’s improvements, but the Voyager series was the one that showed how to do it.

  3. January 5, 2012 5:00 pm

    Theo —

    Friend, we disagree on this one 🙂

    The 28 went in a bookbag generally. Being able to close it meant the screen and buttons were protected against heavy books and shock. The battery door was fine nothing came off the top. I treated my 28 rough for years and it worked perfectly. Using size N batteries was a pain.

    As for the quality of numerical calculation, I don’t know. That’s possible.

    As far as having to memorize, I didn’t find that at all. The 28 menu system was very logical about what was where. I never referred to either manual after I learned the 28, though learning a 28 was a few hour investment.

    Lets put it this was. On my 28s I used the built in numerical integration (romberg) and the complex number functions to create a contour integration function. I did quick and dirty multiple linear regression. That sort of thing would have been impossible on a 15.

    The only thing I didn’t love about the 28 was it couldn’t be backed up to a computer like the 48 could. In terms of the clamshell, not having it was why I never switched to the 48/49 line. That and I started using laptops which had mathematica.

  4. nzumel permalink
    January 5, 2012 7:01 pm

    I still have the HP-15C my father gave me when I graduated from high school. It has a dent in it from when I dropped it down two flights of stairs in the Electrical Engineering building at Cal my sophomore year, but nothing has ever fazed it. I think I’ve changed the battery once in the entire quarter century I’ve owned it. Still ticking.

    My husband had a 15-C and a 16-C; he still has the 15-C, but he sold the other one.
    We both have the 15-C iPhone app. It’s fun, but the real thing is better.

    I have to agree with Theophrastus about the 28-C; it was junky. The display was nice (and it did graphics, too, didn’t it?) but the case felt cheap, and the hinge was flimsy. I borrowed one for a final exam one year, and it died on me in the middle of the exam (it was relatively new, so it shouldn’t have been the battery). Had to finish the exam with a checkbook calculator.

  5. nzumel permalink
    January 5, 2012 7:10 pm

    A little off-topic, but you might enjoy a piece my husband once wrote, on
    the philosophy of engineering as expressed by the HP15-C user manual:

  6. January 5, 2012 11:18 pm

    Nina, your anecdotes reinforce my own experience and that post by John was excellent. The idea of the many different levels to the calculation (as a theoretical structure derived from experimental evidence, as a ideal of a formula, as a the representation of a formula, as the process of applying the formula to specific data, as the process of inputting the data into a calculating machine, as the idea of reading a calculating machine) is inherently fascinating.

    Drawing it back into translation, I think that there is something akin to translating going on when one converts, for example, an equation representing Newtonian mechanics into a sequence of calculator strokes.

    At another level, I think there is great similarity between the activity of “translation” and the activity of “data visualization.” Over here I have a speadsheet of data and over there I have graphical representation of that data. They both represent identical data, and either can be “translated” into the other. Yet, the graphic is interpreted as meaningful by the human observer, while the spreadsheet leads itself to ready calculation and analysis. Isn’t something analogous going on with translation among human languages?

  7. January 6, 2012 12:07 am

    Now you’ve done it. The 16mm camera has reminded me of several DVD commentaries where the directors wax nostalgic about the old little cameras. I’ll have to watch one or two now.

    I remember those HP12Cs. They were “it” for many a businessman I knew.

    All this antique computing reminds me of the TRS80s my college used about a thousand years ago. My first exposure to computer gaming.

  8. January 11, 2012 12:07 am

    I changed my mind and decided not to do a part 2 post. I started to write it up, but it just got too nerdy and I am “backed up” on posts that I want to make.

    So instead, let me just outline what I was going to say.

    I set out to write a post outlining the benefits of HP’s postfix notation, called “Reverse Polish Notation.” I did write a draft of my post, and it just became too nerdy for words. Instead of trying to make an explanation of RPN, let me instead refer to an explanation by James Redin.

    The difference between algebraic notation and reverse Polish notation is a bit like the difference between Subject Verb Object (SVO) languages such as English or Chinese and Subject Object Verb (SOV) languages such as Japanese or Latin. In the case of calculation, the RPN (= postfix = SOV) notation is particularly elegant and well suited. (For example, entering complex calculations on a calculator requires substantially fewer keystrokes and is more likely to be correct.) Once one learns to think in “reverse Polish notation,” one finds oneself actually thinking about calculation in a more elegant and faster form.

  9. January 11, 2012 12:53 am

    Let me just throw out that while this didn’t apply too much to RPL (because it was stack based) in general one of the big advantages of polish notation is the fact that it can be very light on syntax. Parsers become trivial

    Algebraic : 3+4*6 = 27 not 42. And that requires a parsing step, to convert it to 3+(4*6) but in
    RPN: 3 4 6 * + the parse is absorbed in the * before the +
    PN (+ 3 (* 4 6)) the parse is absorbed in the parentheses placement

    And the fact that you could in RPL have partial stacks which “expanded” to command sequences under evaluation along with light parsing gave HP calculators S-expressions.

    This allowed for something very close to homoiconicity in RPL programs.

    The parsing advantage you mention for calculation was somewhat valuable for humans, where it is really valuable is calculations involving manipulating programs as data and then evaluating as programs… programs that write programs. Homoiconicity allows this to go from the complex to the trivial.


  1. Airplane Reading « Multo (Ghost)

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: