applets.

Keyboard Layouts

← All apps

About Keyboard Layouts

The letters on nearly every English keyboard sit where they were put for a typewriter sold in the 1870s. This bench measures how far your fingers travel on that layout and its rivals, lets a search hunt for a better one, traces words across the keys, and tells the story, myths included.

TRAVEL: pick a passage or paste your own text. The bars give the finger travel for the same text on seven layouts; the keyboard shows where the presses land (HEAT) or which finger owns each key (FINGERS). Point at a key for its count. The ticket also gives the share of presses on the home row, how often two keys in a row need the same finger (slow and tiring), and how often the hands take turns. Your own text stays in the page: it is never sent, kept or put in a link.

FIND THE BEST: a search rearranges the 30 letter keys to lower a cost, and you watch the keys swap. DISTANCE ONLY counts finger travel and nothing else; FULLER COST also charges for two keys in a row on one finger, for staying on one hand, for the weaker ring and little fingers and for leaving the home row. The table scores QWERTY, Dvorak, Colemak, Workman and the search's find under both costs. The lesson of the tab: the best layout depends on what you measure.

WORD WALKS: type a word to see its path across the keys, the way the Stand-up Maths video measures it: the distance from key to key (19.05 mm between key centres, a repeated letter adding nothing), and the angles at its corners added up. The lists give the words of the site's word list that travel furthest per letter and the "pointiest" words, on the layout you pick.

HISTORY: how the arrangement came about, with the keyboard showing which letters each later layout moved.

THE STORY

Christopher Latham Sholes, a printer and newspaperman in Milwaukee, worked on a typing machine with Carlos Glidden and Samuel Soulé from 1867; their first patent was granted in 1868. Early versions had the letters in alphabetical order on piano-like keys. Over the next few years the arrangement kept changing, and in 1873 the gunmakers E. Remington and Sons bought the rights. Their machine, the Sholes and Glidden Type-Writer, went on sale in 1874. It typed capitals only, and its letters were close to QWERTY but not quite: M, for one, sat at the end of the middle row. The Remington No. 2 of 1878 added a shift key for small letters, and on a later version of it in the 1880s M moved down beside N and C and X swapped places, giving the layout we use. In 1893 the big typewriter makers joined in one trust, by the late 1890s QWERTY was the standard, and teleprinters and then computer terminals carried it on.

Why that order? The usual story is that Sholes spread common letter pairs apart so that neighbouring typebars, which swing up to the same spot, would not clash and jam. There is something in it: the commonest pair, TH, has its two typebars on opposite sides of the basket. The story that it was meant to slow typists down is a myth in the sense that nobody has found evidence of such an aim: no document from Sholes or Remington says so. What is documented is that some early users were telegraph operators taking down Morse: Koichi and Motoko Yasuoka (2011) trace an early machine to a telegraph college in Chicago and describe demonstrations of it receiving Morse. Their interpretation goes further: they argue those operators shaped the layout, for example by keeping S near Z and E, because in American Morse Z is easily mistaken for SE. That reading is disputed: others see too little evidence that the telegraphers decided it, and the authors themselves say the layout grew by accident among many demands.

Dvorak. August Dvorak and William Dealey designed a layout from studies of English letter frequencies and typing motions and patented it in 1936. It puts all the vowels under the left hand's home keys and the commonest consonants under the right, so about 70% of English presses land on the home row against about a third on QWERTY (this page's samples agree). A US Navy report of 1944 said typists retrained on Dvorak so fast that the cost was soon repaid. In 1990 the economists Stan Liebowitz and Stephen Margolis ("The Fable of the Keys") pointed out that Dvorak was the Navy's own expert on such studies at the time, questioned how the report was run, and cited a 1956 study for the US General Services Administration by Earle Strong that found no gain worth the retraining. Their view, that QWERTY is about as good as its rivals for speed, is itself argued over.

Since then. Shai Coleman's Colemak (2006) keeps ten letters where QWERTY has them (Q W A H Z X C V B M, so the shortcut keys stay put) and changes 17 of the 30 letter-block keys, bringing the common letters to the home row; OJ Bucao's Workman (2010) also weighs how far each finger has to stretch sideways. Martin Krzywinski's Carpalx project scores layouts with a detailed effort model (finger travel, penalties for hand, finger and row, and the path of each three-key stroke) and uses simulated annealing to find low-effort layouts; this page's FULLER COST is a much simpler cost in the same spirit, with its own weights. The AZERTY shown is the French one and the QWERTZ the German one; Belgium and Switzerland use related variants. They grew from the same typewriter keyboard with changes for their languages, and English text is not what they were made for.

HOW IT WORKS

The keyboard is a standard staggered one with the nominal key pitch of 19.05 mm (three quarters of an inch) between neighbours in a row. Each row is offset from the one above (the home row by a quarter of a key from the top row, the bottom row by a further half), so a key's neighbours on the next row up or down are not exactly 19.05 mm away: D to E is 19.6 mm, for example. UK, AZERTY and QWERTZ boards have the extra ISO keys (one left of Z, one by Enter); the US board, Dvorak, Colemak and Workman are the US kind. The fingers follow the usual touch-typing chart: two letter columns for each index finger, one for each other finger, and the little fingers take everything at the edges. On the number row, 6 is given to the right index finger; some courses teach it with the left.

Finger travel is a simplified measure, chosen to be easy to check by hand: every finger rests on its home key (A S D F and J K L ;), and each press counts the straight trip from that home key to the key and back. So E on QWERTY is 2 × 19.6 mm = 39.3 mm, and the home keys cost nothing. Real fingers don't always go home between presses, wrists move, and some reaches are harder than their length suggests; speed and comfort also depend on rhythm, rolls and practice, which no distance captures. Spaces, Enter and Shift are left out (thumbs and little fingers barely leave their places for them), a capital counts as its letter key, and curly quotes and dashes count as the plain keys people type for them. Only characters a layout types directly (one key, with or without Shift) are counted. Characters typed with a dead key, such as ê and ë on French AZERTY (^ or ¨ first, then the letter), are not modelled and are listed as not counted, as is any character a layout has no key for. So that layouts are compared on the same text, the bars and the "least" line count only the characters every layout on the ticket can type; the table below them gives the chosen layout's own counts.

Same-finger pairs and hand alternation look at each two characters in a row with no space between: a same-finger pair is two different keys on one finger (the same key twice doesn't count).

The search runs in the background (a web worker) so the page stays free. It uses simulated annealing: start from a layout, swap two keys at random, always keep a swap that lowers the cost, and sometimes keep one that raises it, less and less often as the "temperature" falls over 150,000 tries. Keeping some worse swaps early stops it settling for the first dip it finds. It rearranges only the 30 letter-block keys (the letters and ; , . /), learns its letter and pair counts from the text chosen on TRAVEL, and other keys stay where they are on a US board. A short passage gives a layout tuned to that passage. The FULLER COST weights are the page's own: travel counts 1.25 times for ring fingers and 1.5 times for little fingers, each press off the home row adds 3 mm (top row) or 6 mm (bottom and number rows), and for two presses in a row with no space between, two different keys on one finger add 30 mm, two different keys on different fingers of the same hand add 3 mm, and a repeated key or a change of hands adds nothing. Change them and the winner changes: that is the point. For DISTANCE ONLY the best is known exactly (the commonest characters on the nearest keys); the search usually reaches it, and the page checks its result against that exact minimum and says when it stops short. Many layouts tie, so different runs end differently. The table scores Dvorak, Colemak and Workman with their own keys: Dvorak's / and ' sit outside the 30-key block, and are scored where Dvorak puts them.

Word walks use the Stand-up Maths measures, worked from the figures in the video's description: distance from key centre to key centre (PAPAL is 655.7 mm, 131.1 mm a letter), and at every corner the angle between the way in and the way out, added up (a straight line through a key is 180°, turning straight back is 0°). The "longest" word has the most distance per letter; this page takes the "pointiest" to be the word whose corners add up to the smallest angle above zero (paths that only run back and forth along one line add up to exactly 0 and have no point). Ties go to fewer corners, then the shorter word. The words come from the lists the site's Word Solver uses (SCOWL, words of 3 letters or more); the video used a different list, so some of its winners differ.

FURTHER WATCHING AND READING

TEXT SHOW

Point atTap a key for its count, or a layout's name on the ticket to show it here.

REPAIR TICKET · FINGER TRAVEL

ON THIS LAYOUT