Keyboard Layout Metrics

These are the metrics used to evaluate and compare keyboard layouts. Every alternative layout — Colemak DH, Dvorak, Graphite, Canary, and hundreds of others — is designed to improve on QWERTY by optimising some combination of these metrics. Understanding them is essential before choosing a layout, because different layouts make different tradeoffs, and without knowing what the metrics mean you cannot judge which tradeoffs suit your hands, your keyboard, and your typing style.

All examples below use QWERTY as the reference layout, since that is what most readers already know.

How distances are measured

Distances between keys are measured in U (key units), where 1U is the center-to-center distance between two horizontally adjacent keys on the same row — roughly 19.05 mm on most boards. Two keys side by side (like F and G on QWERTY) are exactly 1U apart.

Vertical row spacing is also 1U, but on traditional keyboards the rows are not stacked in a straight grid. Each row is shifted slightly to the side — this is row stagger (see row stagger explanation). Here is what QWERTY actually looks like if you measure from key center to key center:

 Q  W  E  R  T        ← top row
   A  S  D  F  G      ← home row (shifted 0.25U right of top row)
     Z  X  C  V  B    ← bottom row (shifted 0.5U right of top row)

Notice that A is not directly below Q — it’s shifted a quarter-key to the right. Z is shifted even further. The rows are like stair steps, each one nudged to the right.

Now consider the key F on the home row. If the rows were aligned in a straight grid, the key directly above F would be at the same horizontal position. But because the top row is 0.25U to the left of the home row, the key R (which is above F) is 0.25U to the left of where “directly above” would be. When you press F and then R, your finger doesn’t move straight up — it moves up 1U and left 0.25U, tracing a slight diagonal.

The actual distance of that diagonal is $\sqrt{1^2+0.25^2}\approx 1.03\text{U}$ — barely more than 1U. For bigger jumps the offset adds up: going from V (bottom row) to T (top row) crosses 2 rows vertically (2U) with 0.75U of horizontal offset, giving $\sqrt{2^2+0.75^2}\approx 2.14\text{U}$.

Columnar keyboards like the ZSA Voyager have no row stagger, so keys directly above/below each other are exactly 1U apart. The fractional distances (1.03U, 2.02U, etc.) discussed below apply only to traditional staggered keyboards.

Source

This note distils the metric definitions from Keyboard Layouts Doc (3rd Edition).

At a glance: which way is better?

There are many metrics below. Before diving in, here is the single most important thing to remember — which direction is “good” for each one:

Metric	You want it…	Why
SFB, SFS, Scissors, LSB, Redirects	Lower	These measure discomfort — same-finger strain, awkward stretches, direction reversals
Rolls, Alternation	Higher	These measure fluidity — smooth finger chains and rhythmic hand switching

Rolls and alternation trade off against each other (more of one means less of the other), so no layout maximises both. The balance between them is the core design choice — see Layout Design Philosophy.

---

SFB (Same-Finger Bigram)

A bigram is any two-letter sequence (e.g. TH, ED, IN). A same-finger bigram (SFB) occurs when two consecutive keys are typed by the same finger.

Why it matters

SFBs force a single finger to move between two keys with no rest in between. The longer the distance, the slower and more uncomfortable the motion. They are the single most important metric in layout design — virtually every alternative layout exists to reduce SFBs compared to QWERTY.

1U vs 2U SFBs

Not all SFBs are equally bad. The distance between the two keys determines severity:

• 1U SFB: the two keys are adjacent (same finger, neighbouring rows). On QWERTY, FR (left index, home to top row) is 1.03U — not exactly 1U because row stagger shifts the top row 0.25U to the left, so the real distance is $\sqrt{1^2+0.25^2}\approx 1.03\text{U}$. These are tolerable.
• 2U SFB: one key is on the top row and the other on the bottom row, so the finger must jump over the home row. On QWERTY, UN and NU (right index) are 2.02U, and CE / EC (left middle, going from home C to top E) are also 2U. These are painful and should be avoided for any common bigram.

The worst SFB on QWERTY is MY at 2.66U — the right index must leap from the bottom-row M all the way to top-row Y.

QWERTY example

Type the word decade. The letters D, E, C are all on the left middle finger. That single word contains three SFBs: DE, EC, DE — five of six letters hammered by one finger. This is why words like “decided”, “exceed”, and “receive” feel so sluggish on QWERTY.

Concrete numbers

Layout analysers like Genkey and Oxeylyzer report total SFB rate as a percentage of all bigrams. QWERTY scores roughly 6–7% SFB depending on the corpus. Well-optimised layouts achieve 0.8–1.5%.

Goal

Lower is better. Target below 1.5% total SFB, with as few 2U SFBs as possible.

---

SFS (Same-Finger Skipgram)

A skipgram is a pair of letters separated by one or more intervening characters. A same-finger skipgram (SFS) occurs when the same finger types both letters of a skipgram — they are not consecutive, but almost.

Why it matters

Even though there is an intervening keystroke by a different finger, your original finger still has to travel back and reposition. With only one letter of separation (a “skip-1-gram”), the finger barely has time to recover. The discomfort is subtler than a SFB but accumulates over thousands of words.

How it works

Take the word may on QWERTY. M and Y are both typed by the right index finger, with A (left hand) in between. That makes M_Y a skip-1-gram of 2.66U — the same enormous distance as the MY SFB, just slightly masked by the intervening A.

A deeper skipgram is less noticeable. In the word thought, T_G is a skip-3-gram (three letters separate T and G). By the time the left index needs G, it has had time to return near its home position. Skip-3-grams and beyond are barely felt.

QWERTY example

The word burn is typed by alternating both index fingers: B (left index) → U (right index) → R (left index) → N (right index). That produces B_R (2.66U) on the left index and U_N (2.02U) on the right index — two brutal SFSs in a four-letter word.

Goal

Lower is better. Optimise especially for skip-1-grams with high distance. Analysers often weight SFS distance at roughly half the penalty of SFB distance.

---

Scissors

A scissor is a bigram where one finger reaches to the top row while another finger on the same hand curls down to the bottom row, skipping the home row entirely. The name comes from the crossing, scissors-like motion of the fingers.

Why it matters

Scissors force adjacent fingers into an unnatural splay. Unlike SFBs (which burden one finger), scissors stress the tendons between two fingers. They are a leading cause of hand fatigue in long typing sessions.

Full scissors vs half scissors

• Full Scissor Bigram (FSB): the vertical separation between the two keys is two rows (top to bottom), and the finger that naturally prefers being higher is placed lower. The definition accounts for finger length: the middle finger naturally sits higher than the index, ring, or pinky, so a bigram where the middle finger is on the bottom row and a shorter finger is on the top row is a full scissor.

• Half Scissor Bigram (HSB): same directional rule, but the vertical separation is only one row (top-to-home or home-to-bottom). Half scissors are less severe than full scissors but still not ideal.

Both types also have skipgram variants (FSS — Full Scissor Skipgram, HSS — Half Scissor Skipgram) where an intervening key is pressed between the two scissoring keys, reducing the discomfort somewhat.

Finger preference

Because our arms approach the keyboard at an angle (roughly 70 degrees, not 90), each finger has a natural vertical preference:

Finger	Prefers to be…
Index	Lower (closer to bottom row)
Middle	Higher (closer to top row)
Ring	Higher than index/pinky, lower than middle
Pinky	Lower than middle/ring, higher than index

A scissor occurs when these preferences are violated.

QWERTY example

On QWERTY with standard fingering, CR is a full scissor: C is on the bottom row (left middle finger) and R is on the top row (left index). The middle finger is lower than the index — the opposite of its natural preference. Typing “across” or “screen” forces this motion repeatedly.

Subjectivity

Scissors definitions vary between analysers. Some count all top-bottom motions as scissors regardless of finger-length preference. The keysolve analyser, for example, counts all ring-middle bigrams spanning two rows as scissors regardless of direction.

Goal

Lower is better. Prioritise minimising adjacent-finger full scissors. Non-adjacent scissors and half scissors are secondary.

---

LSB (Lateral Stretch Bigram)

A lateral stretch bigram (LSB) is a bigram that forces two fingers apart horizontally, or requires a lateral wrist shift to reach from one key to the other. LSBs most commonly involve the center column — the inner column that the index finger must reach sideways to hit.

Why it matters

Lateral stretches pull your hand out of its natural resting position. On a traditional row-stagger keyboard, this is compounded by the rows being offset, which makes some horizontal distances longer than they appear. On a columnar or matrix keyboard (like the ZSA Voyager), the columns are aligned vertically, which changes which stretches are problematic.

Definition

A common threshold is:

• Adjacent-finger LSB: horizontal distance of 2U or more. This typically happens when the index finger presses a center-column key and the middle finger presses its own column key (e.g. QWERTY TE, GE, BE).
• Semi-adjacent-finger LSB: horizontal distance of 3.5U or more. This can occur between ring and index fingers on row-stagger keyboards (e.g. QWERTY BW, BS).

Like SFBs and scissors, there is also a skipgram variant: LSS (Lateral Stretch Skipgram), where the two stretching keys are separated by one intervening keystroke. For example, typing “were” on a layout where W and R are both near the center column produces a W_R lateral stretch skipgram.

Row stagger vs matrix

This is where your ZSA Voyager (a columnar split keyboard) diverges significantly from a standard keyboard:

• On row stagger: placing letters on the center column is generally less problematic on the right hand, because the top-row key sits closer to the center column on that side. Mirroring a layout can change LSB scores substantially.
• On matrix/columnar: rows are perfectly aligned, so left and right hands are symmetrical. Some LSBs that exist on row stagger disappear on matrix (e.g. QWERTY BW), but others that row-stagger hides become real (e.g. QWERTY TD, NI).

QWERTY example

On QWERTY, GE is a lateral stretch: G is on the center column (left index) and E is on the top row (left middle). The horizontal gap is roughly 2.25U on row stagger. Words like “get”, “together”, and “age” trigger this stretch constantly.

Goal

Lower is better. Either reduce center-column usage (fewer letters there) or strategically place only rare bigram partners on the center column so stretches are infrequent.

---

Rolls

A roll is a sequence of keys typed by adjacent fingers on the same hand in a single fluid motion. Rolls are measured at the trigram level (three-letter sequences) to distinguish them from redirects.

Definition

A roll (also called a 2-roll) is a trigram where two keys are pressed by one hand and the third key by the other hand. The two same-hand keys must use different fingers. For example, on QWERTY the trigram OUR has OU on the right hand (ring and right index) followed by R on the left hand — that is a roll.

A 3-roll (also called a “onehand”) is a trigram where all three keys are on the same hand and the fingers move in a single consistent direction — either all inward (toward the index finger) or all outward (toward the pinky). On QWERTY, WER is a 3-roll: W (ring) → E (middle) → R (index), all moving inward on the left hand.

Inward vs outward rolls

• Inward roll (in-roll): fingers move from pinky-side toward the index finger. Example on QWERTY: DF (middle → index). Most typists find in-rolls more comfortable because the stronger index finger “catches” the motion.
• Outward roll (out-roll): fingers move from index-side toward the pinky. Example on QWERTY: FD (index → middle). Still fluid, but slightly less natural for most hands.

Some layouts (like “Inrolly”) are specifically designed to maximise in-rolls.

Roll comfort factors

Not all rolls are equal. A roll is most comfortable when:

1. Both keys share the same row (no vertical motion).
2. When rows differ, the longer finger (middle) is on the higher row.
3. No lateral stretch is involved.
4. Strong fingers (index, middle) are used rather than weak ones (ring, pinky).
5. The fingers are adjacent (pinky-ring, ring-middle, middle-index) rather than non-adjacent.

QWERTY example

The word “the” on QWERTY is typed T (left index) → H (right index) → E (left middle). That is a roll: HE is a two-key sequence on the right+left hands bracketing a hand switch. But “the” also involves a hand change after T, so this is actually a roll trigram (1-hand, 2-hand pattern).

Well-optimised rolling layouts achieve 49–52% rolls, compared to QWERTY’s roughly 37%.

Goal

Higher is better, but there is a tradeoff: more rolls mean fewer alternations and usually more redirects. The sweet spot depends on personal preference.

---

Alternation

Alternation is a trigram where each keystroke switches hands: left → right → left, or right → left → right. All three keys are on different hands, producing a rhythmic, even cadence.

Why it matters

Alternation spreads the workload evenly between hands and produces a metronomic rhythm that many typists find comfortable and consistent. High-alternation layouts have fewer long same-hand sequences, which means fewer opportunities for redirects or awkward finger tangles. The downside is that maximising alternation necessarily reduces rolling — you cannot have both at the maximum simultaneously.

The tradeoff with rolling

This is the fundamental tension in layout design. Consider five layouts arranged from maximum alternation to maximum rolling:

Layout	Alternation	2-Rolls	3-Rolls	Redirects
Poqtea (max alt)	46.8%	35.3%	1.2%	7.5%
Graphite	40.2%	45.1%	2.6%	3.4%
Rolly	32.1%	49.0%	3.5%	6.9%
Inrolly	26.7%	52.3%	3.8%	8.2%
Seht Drai (max roll)	22.0%	48.4%	10.6%	10.6%

As alternation drops, total rolling rises — but so do redirects. No layout can maximise everything. Alternation offers a more consistent experience (rhythmic, predictable), while rolling has higher highs (words that flow beautifully) but lower lows (long same-hand sequences that tangle).

What drives alternation

Placing all vowels on one hand is the single biggest driver of alternation, because most English words alternate between consonants and vowels. Beyond that, which consonants join the vowel hand matters:

• H with the vowels maximises alternation on H-trigrams, because H almost exclusively precedes vowels (THE, THA, THI).
• T with the vowels further increases alternation (THE becomes a pure alternate: T-H-E across three hand switches).
• R and N with the vowels instead maximise rolling, because they combine with vowels in both directions (ER/RE, IN/AN).

QWERTY example

On QWERTY, AND is a pure alternate: A (left) → N (right) → D (left). QWERTY’s alternation is roughly 33% — middling, because its letter placement was not designed with this metric in mind.

Goal

Depends on preference. High alternation (~40%+) gives consistency and low redirects. High rolling (~48%+ 2-rolls) gives fluidity but more redirects. Most modern layouts aim for a balance around 35–42% alternation.

---

Redirects

A redirect is a same-hand trigram where the finger direction reverses. That is, the first two keys move inward (toward the thumb) but the third key moves outward (toward the pinky), or vice versa.

Why it matters

Redirects break the natural flow of a roll. Instead of three fingers sweeping in one direction, the hand must change direction mid-sequence. This “stutter” feels awkward, especially at speed. Redirects are the price you pay for high rolling — any layout that keeps many keys on the same hand will inevitably produce direction changes.

Weak (bad) redirects

A redirect is called weak (or “bad”) when the index finger is not involved. If all three keys are on the pinky, ring, and middle fingers, the weaker fingers must handle the direction change alone, which is noticeably harder. Layouts that place vowels on the last three fingers (pinky, ring, middle) of one hand and consonants on the index of that hand ensure that most redirects involve the index, avoiding weak redirects.

QWERTY example

On QWERTY, SAD is a redirect: S (left ring) → A (left pinky) is outward, then A → D (left middle) is inward. The direction reverses at A. This is also a weak redirect because the index finger is not involved.

Another example: ERE on a rolling layout where E, R, and E are all on the same hand would be a redirect (out-in-out), and since E appears twice it also involves a SFS.

What causes redirects

1. Vowel-consonant-vowel trigrams on the same hand (e.g. ERE, ONE, ARE).
2. Consonant-vowel-consonant trigrams on the same hand (e.g. HIN, HEN).
3. Consonant-only trigrams with mixed direction (e.g. STR where S→T is one direction and T→R reverses).
4. Punctuation trigrams (e.g. T’S).

Goal

Lower is better. QWERTY has roughly 5–6% redirects. Well-balanced layouts achieve 3–5%. High-rolling layouts can hit 8–10% — whether that is acceptable is personal preference. Minimise weak redirects especially.

---

Alt Fingering

Alt fingering means deliberately using a non-standard finger to type a key, in order to avoid a SFB. Instead of the “correct” finger hitting both keys, you reassign one key to an adjacent finger for that specific bigram.

Why it matters

Even well-optimised layouts have some residual SFBs. Alt fingering lets you eliminate them at the cost of learning a non-standard finger assignment for certain bigrams. Some layouts are explicitly designed with intended alt fingers — the SFB is placed where it can be comfortably rerouted.

How it works

On QWERTY, CE and EC are 2U SFBs on the left middle finger (C on the bottom row, E on the top row). Many advanced QWERTY typists alt-finger this by pressing C with the index finger instead of the middle, turning a painful same-finger bigram into a comfortable two-finger motion.

SFB collisions

Alt fingering can create new problems. If you press C with the index finger to avoid the EC SFB, then CT becomes a new SFB (both C and T are now on the index finger) specifically in trigrams like ECT. The best alt fingers are those that do not cause such collisions — where the rerouted finger assignment does not create a new SFB for any common trigram.

Comfort by keyboard type

On row-stagger keyboards, alt fingering is relatively comfortable because the row offsets naturally place some keys close to adjacent fingers. For example, QWERTY FR (pressing R with the middle finger instead of the index) is easy because row stagger puts R close to the middle finger’s resting position.

On matrix/columnar keyboards (like the ZSA Voyager), there are fewer comfortable alt fingers. The perfectly aligned columns mean that reaching a neighbouring finger’s key requires more lateral splay. The vertical alt fingers that felt natural on row stagger become cramped on matrix.

Sliding

Sliding is a variant of alt fingering used on flat (laptop-style) keyboards. Instead of lifting and repositioning, you slide your finger downward from one key to the next. This only works when the SFB direction is always downward and the keycaps are flat. On keyboards with sculpted or tall keycaps (including most mechanical keyboards), sliding over the gap between keys is uncomfortable.

Goal

Alt fingering is a technique, not a layout metric to minimise. Some layouts intentionally include 1–3 designed alt fingers on the index finger column, where they can be comfortably rerouted. If you are choosing a layout for a columnar keyboard, be aware that fewer alt fingers are comfortable compared to row stagger.

---

Summary comparison

Metric	Definition	Goal	QWERTY (rough)
SFB	Two consecutive keys, same finger	Lower is better	~6.5%
SFS	Same finger, one key between	Lower is better	Very high (unoptimised)
Scissors	Top↔bottom row, same hand, skipping home	Lower is better	Moderate-high
LSB	Lateral stretch between fingers	Lower is better	High (many center-column stretches)
Rolls	Adjacent-finger same-hand sequence	Higher is better	~37%
Alternation	Hand switches every keystroke	Preference	~33%
Redirects	Same-hand trigram, direction reverses	Lower is better	~5–6%
Alt fingering	Non-standard finger to dodge SFB	Technique, not a score	N/A

---

English frequency basics

All layout design starts from the same data: which letters and letter pairs appear most often in English text. The most common letters, drawn from large corpora such as Peter Norvig’s analysis of Google Books, are:

E, T, A, O, I, N, S, H, R

These nine letters account for roughly 70% of all characters typed in English. After them come L, D, C, and the vowel U.

The most common bigrams are:

Rank	Bigram	Rank	Bigram
1	TH	6	RE
2	HE	7	ON
3	IN	8	AT
4	ER	9	EN
5	AN	10	ND

Most bigrams pair a consonant with a vowel. The major consonant-only bigrams are TH, ND, ST, NT, NG, and CH.

This frequency data is why alternative layouts place E, T, A, O, I, N, S, H, R on the home row or the most accessible positions: those letters are typed the most, so they should require the least movement. It is also why all vowels typically go on one hand — since most bigrams alternate consonant-vowel, putting vowels together maximises hand alternation.

---

See also

• Touch Typing and Finger Assignment
• Layout Design Philosophy
• Colemak DH
• Keyboard Layouts MOC