A Short Story of the Keyboard

from the November 1982 issue of Byte magazine

by Phil Lemmons

Keyboards are meant to let our fingers do the talking, but more often they make us swear aloud. Every manufacturer seems to want its keyboard to be unmistakably different from any other. The only keys that seem to be sacred and immovable are badly placed: the familiar QWERTYUIOP and its companion rows of the alphabet. The Shift and Return keys occasionally stray, and the control keys and function keys wander from one end of the keyboard to the other. Perhaps most puzzling of all, the placement of the cursor keys is not yet standardized in the most logical configuration, with the “up” key above, the “down” key below, the “left” key at the left, and the “right” key at the right. Let’s hope that Chris Rutkowski’s efforts to organize the placement of the most common control functions in sensible groupings will be a major step toward standardization.

If you think it’s hard to adjust to a new keyboard now, though, consider the situation 80 to 90 years ago. The Gay Nineties were nightmarish for office temps. They would never know what keyboard was waiting at their next assignment. Oh, the QWERTYUIOP keyboard was around, all right, but it was only one among a hundred. Almost every company that made a typewriter used a different keyboard. Typewriters of the time (and their toprow key arrangements) included the Crandall (ZPRCHMI), the American (CJPFUBL), the Hall (KBFGNIA), the Columbia (ZKPWMCR), the Morris (XVGWSLZ), and so on, ad nauseam.

Most of the early keyboards seemed to have totally random key arrangements, but a few designs represented attempts at some sort of order. Some keyboards put T, H, and E near one another, for example, on the theory that “the” has to be typed perhaps more often than any other word. Similarly, the World typewriter keyboard put A, N, and D together, and the Edison Mimeograph typewriter grouped not only A, N, and D, and T, H, and E, but also O and F and I, N, and G.

But there’s more to a keyboard than how you label the keys. Keyboards in the 1890s differed greatly in how many keys they had, how many rows of keys, and how the rows were arranged. Part of the problem was that many keyboards still lacked a shift key. Today it’s hard to conceive of a typewriter without a shift key, but the idea didn’t occur to anyone until Byron A. Brooks thought of it in 1875.

“Too complicated!” many people complained.

“Too tiring for the operator!” others insisted.

Such reactions prolonged the survival of some interesting mutant keyboards. The Caligraph, for example, had a circular keyboard and no shift key; all the small letters were grouped in the center and surrounded by all the capital letters, with no apparent correspondence between the arrangements of the two sets of letters. The Imperial Model B had three semicircular rows of 10 keys each, arranged in an arc convex to the typist — the central keys were nearest and the outer keys farthest from the fingertips (apparently it was designed for a typist whose outer fingers were longer than the inner ones); the layout included a shift key and a space bar, and the top row of keys read ZHfAYSCPG. Compared to the Imperial, the Hartford keyboard almost seemed to make sense: it had six straight rows of keys, no shift key, and a separate key for each small and capital letter, with identical rows of keys for both cases. The Yost keyboard, like the Imperial, had all the small letters arranged on the lower rows and all the capital letters in the same pattern in the upper rows; but the Yost had eight rows of keys instead of six. The Saturn keyboard had only one straight, very long row of keys. The Hammond had two semicircular rows, the Salter had three such rows, and the Kanzler had four gently arced rows.

One keyboard of the 1890s, the Ideal, seemed to make more sense than the others. The idea behind the Ideal keyboard was that more than 70 percent of all English words are made up of the letters DHIATENSOR; therefore, placing all these keys in one row should make typing more efficient. Some major companies adopted the Ideal keyboard, including the maker of the best typing machine of the day, the renowned Blickensderfer. (Blickensderfer’s engineering prowess was such that, in 1902, it was producing an electric typewriter that used a type wheel much like the modern IBM Selectric “golf ball” or the daisy wheel.) But the Ideal keyboard, despite being as sound as a Blickensderfer, lost out to an inferior competitor, our familiar QWERTYUIOP (also known as the Universal keyboard).

By 1943, when Dr. August Dvorak proposed a clearly superior keyboard, the QWERTYUIOP keyboard had become too deeply entrenched to be easily overthrown. Dvorak’s idea was to place the five vowels under the fingers of the left hand and the five most common consonants under the fingers of the right. The row of keys that resulted was AOEUIDHTNS.

Dvorak’s keyboard is not the only so-called reform keyboard. The idea behind most of the reforms is to put the most common letters in easiest reach of the strongest fingers and to put the most frequenctly combined letters under the control of opposite hands. To make typing “the” faster, for example, a keyboard might put T on the right side, H on the left side, and E on the right side of the keyboard. Note that the Dvorak keyboard doesn’t arrange these three letters in that way, which only proves that Dvorak wasn’t trying to optimize the keyboard for typing “the.”

Michael H. Adler, author of The Writing Machine (London: George Allen & Unwin, 1973), argues persuasively for a new standard keyboard that puts the 10 most common letters, ETAONIRSHD, on a single row curved in such a way that each of the 10 fingers (thumbs included) rests comfortably on one of the keys. Our thumbs now spend most of their time lolling on the space bar; Adler delegates the space bar, the shift key, and the carriage return to the feet, freeing the thumbs for a higher destiny. This somewhat piano-like arrangement should result in much faster typing. As Adler points out, “After all, . . . a pianist can comfortably handle over 1500 to 2000 keystrokes a minute (the equivalent of 300 to 400 words per minute) on a much less compact keyboard than the one described, and without trying to break world speed records, either. "

The Battle of the Numeric Pads

Many of today’s keyboards have numeric keypads — groupings of keys separate from the main alphabetic grouping — to help typists enter numbers more quickly. The numerals on the main keyboard are, of course, laid out in a single horizontal row above the QWERTYUIOP row of letters. Using the main keyboard to enter most numbers requires the use of both hands. The numberic keypad makes all the numerals available to one hand. Besides a key for each of the numerals 0 through 9, numeric keypads have a decimal-point key, a “+ " key, a “-” key, and an Enter key, but for now let’s consider only the numerals.

In the numeric (calculator) keypad the numerals are usually laid out something like this:

7 8 9
4 5 6
1 2 3
0

The usual keypad arrangement contrasts with the telephone company’s numeric pad for entering telephone numbers:

1 2 3
4 5 6
7 8 9
0

The designers of the push-button telephone considered and tested several different arrangements of the 10 numeric keys, including two vertical rows of five buttons, two horizontal rows of five buttons, and a circle. After deciding on four rows of three keys, why didn’t the designers use the traditional calculator arrangement for the numerals? Because tests established that people entered numbers more quickly and accurately with the top-to-bottom, left-to-right arrangement (perhaps because we read things in that order).

Designers of nonstandard keyboards are invited to take all these factors into account in their next designs. But a proliferation of keyboard designs would probably do more harm than good, even if most of the new designs represented an improvement on the QWERTYUIOP and calculator arrangements.

A Solution without a Standard

The programmable detached keyboard, such as those on the Victor and Epson QX-10 microcomputers, raises a new possibility: because every key on the keyboard can be programmed and the keyboard is detached, there’s no reason not to have more than one keyboard for each computer. Just unplug one keyboard, plug in another, and load the operating system that loads the correct codes for the keys. This would mean that, on one computer, Harvey could type on the Dvorak keyboard with a telephone-style numeric keypad, and Eloise could use the QWERTY layout with a calculator keypad, so long as the two were content to use the system at different times. Each person could use or edit the data entered by the other: two keyboard units would be necessary only to save the trouble of relocating the key caps. This sort of flexibility would be possible on many new systems if the manufacturers would supply utility programs to enable nonprogrammers to program the keyboard. Instead of a single standard keyboard, we would have a standard of high adaptability. ■