Wednesday, April 10, 2013

New (in 1930) electric typewriter prints whole phrases

When browsing through an old issue of Popular Mechanics (the March 1930 issue to be exact, amazing what is on the internet), the following short article drew my attention.

The text was intriguing and not quite clear to me at first reading. Did make me wonder: what did such a machine look like?

Another slightly longer article in an American newspaper of April of the same year gives more information about the machine (but still no pictures) and especially information on the rationale for such a machine:

Reading this pitch; it should be a great success! But these machines did not become ubiquitous. This had me wondering for a bit. Why is that?

The machine may have been too costly for it's benefit back in the 30-ies, but probably there is a another cause. The fact that also today we use mostly character based keyboards is a strong hint there is a fundamental flaw in the proposition of a word writing machine.

When viewing a writing machine from the information flow point of view, then I think there is a likely, fundamental reason word-writing machines are not and will not be a great success.
In the case of a regular typewriter or a modern computer keyboard, the user needs to choose over time a character from a keyspace of about 30 characters. With every decision made, the information (character) is added to the result (text).

For a word-writing machine, the user has to choose from a keyspace of 100 (50%), 400 (90%) or even 1000 words (98%) in addition to the individual characters. Assuming an average word length of 6 characters, this would allow the user to need 6 times as long to make the choice in this larger keyspace to still need the same amount of 'work' to write a text. The time ('work') needed to make the choice and enter into the machine is I suspect not a straight linear relation with the size of the keyspace. Probably there is an exponential relationship. Making a choice and a keyboard entry from a 130+ keyspace takes longer than entering 6 regular keystrokes.

The benefit of word-writing (if any) is too small to justify the cost of learning a new entry method.

Anyways, I get distracted. What did it look like...

A quick trawl through patent databases turned up a few patents with likely relevant titles. The US patent numbers 1,587,137 for a "Word Writing Machine",  1,590,998 for a "Combined Keyboard and Chart" and 1,619,691 for a "Typewriting Machine" all describe and claim such a device. These were filed for during the early 1920-ies.

From the text of the patents it seems that Mr Balston started filing patent applications for this invention already in 1914. Overall, Mr Balston was a quite prolific inventor over a broad field. His published patents range from a "Nursing Bottle" (1890) to a "Photographic Camera".

But back again to the word-'macro'-writing machine; the first patent describes the machine that has the ability to print a series of letters in one go; print a whole word in a from a single key-press. It does this with a row/rack of letter-rings, indeed much like the mileage counter on a speedometer. The hard rubber letter-rings have all printable characters on the outer diameter. The rings are rotated so as to form the word to be written and then hammered against the page. The rotation is controlled of course via the key selected on the keyboard and encoded in the mechanics.

It could be seen, in working principle, perhaps as a hugely parallel Hammond-like machine.  Or similar to the mechanism of adding machines (or cash registers) to print all digits of a number in one strike, albeit with many more characters.

Rather complex drawings, but they do show the basic arrangement:


Driving the thing was a problem to be solved, the depressing of the key does not give enough energy to make the whole-word imprint. (Adding machines have a hefty lever for that.) In the patent a clockwork spring is mentioned that would be wound up so as to have sufficient energy for the machine to work for some time. An electric motor would probably be the preferred solution...

Selecting the words was rather ingenious, as described in the newspaper article above. The diagram from the patent shows the general idea. Also it shows the complexity, there is a large choice to be made by the user for every entry:

Given the sophistication of the patents and the detail in the articles, at least one functional machine would surely have been built. Can't however find any sign of this device becoming a commercial product.

The complexity in the construction added by the 'macro' function, combined with the unusual keyboard operation probably were more costly than the cost of just writing the individual characters. Also by 1930 the form-writers were, I guess, established products, using paper-tape or pre-punched aluminium strip or sheet. These would be hard to compete against in such a specialty product for a niche market by an unproven method that would take more work.

Still curious if a photograph of this machine (or such a machine) exist?

Distracting, this internet. Interesting, but distracting...


  1. Amazing and insane.

    There have been several attempts to devise typewriters that would print syllables or words, but inevitably the keyboard becomes unwieldy, and inevitably these inventions fail.

    There were the Chinese and Japanese typewriters that required you to select from thousands of tiny character types ... but people blessed with an alphabet were not disposed to do this!

    1. Yes and yes. One of those ideas that periodically get patented...

      And yet.
      Another very elaborate, more modern one I stumbled on is 4,804,279 - close to a stenotype but words still. Hmm, they may have done it (a bit); their Velotype is at least still around: