How did Alfred Butts determine the Scrabble tile distribution?

Alfred Butts counted letter frequencies by hand from the front pages of The New York Times in the 1930s. He tallied how often each letter appeared, then scaled those frequencies to a 100-tile set. He adjusted some counts for gameplay balance — reducing very common letters and ensuring rare letters like Q, Z, X, and J each appeared exactly once.

Why are there 12 E tiles in Scrabble?

E is the most common letter in English, appearing in roughly 12-13% of all text. Butts scaled this to 12 tiles in a 100-tile set, making E represent 12% of the bag. This high count ensures players can form common words easily, but also means E tiles pile up on racks.

Does Scrabble tile distribution exactly match English letter frequency?

Not exactly. Butts made deliberate deviations for gameplay balance. S appears in 6.3% of English text but only has 4 tiles (4%) in Scrabble — because S is too powerful as a hook letter. Similarly, T and N are slightly underrepresented. The distribution is inspired by English frequency but tuned for game balance.

Solver Quiz Anagram Rack WOTD 60s Magic Squares Wordrus

Scrabble Tile Distribution Explained: Why 12 E's and 1 Z?

June 14, 2026 9 min read Word Finder

Every Scrabble game uses exactly 100 tiles with a precisely calibrated distribution. This isn't random — it's the result of one man counting letters on newspaper front pages in 1930s New York. Alfred Butts' distribution has survived nearly a century with only minor tweaks, proving his hand-counted frequency analysis was remarkably accurate. Here's the full story behind the numbers.

The Complete Distribution

Tile	Count	Points	Tile	Count	Points
A	9	1	N	6	1
B	2	3	O	8	1
C	2	3	P	2	3
D	4	2	Q	1	10
E	12	1	R	6	1
F	2	4	S	4	1
G	3	2	T	6	1
H	2	4	U	4	1
I	9	1	V	2	4
J	1	8	W	2	4
K	1	5	X	1	8
L	4	1	Y	2	4
M	2	3	Z	1	10

100

Total tiles

Vowels (A,E,I,O,U)

Consonants

Blanks (0 pts)

Alfred Butts' Methodology

💡 Historical Context

In 1938, unemployed architect Alfred Mosher Butts counted letter frequencies from front pages of The New York Times. He tallied thousands of letters by hand, calculated percentages, and scaled them to a 100-tile set. He then adjusted for gameplay — reducing overpowered letters and ensuring rare tiles remained exciting but not game-breaking.

🧩 Butts' Design Process

Count letter frequencies from newspaper text (thousands of samples)

Scale percentages to a 100-tile set (12.7% E → 12 tiles)

Assign point values inversely to frequency (rare = high points)

Adjust for gameplay: reduce S count (too powerful), limit blanks to 2

English Frequency vs Scrabble Distribution

The deviations between actual English text frequency and Scrabble tile counts reveal deliberate game design choices.

Letter	English %	Scrabble %	Deviation	Reason
E	12.7%	12%	≈ Match	Accurate
T	9.1%	6%	−3%	Reduced: too easy
S	6.3%	4%	−2.3%	Nerfed: too powerful as hook
A	8.2%	9%	≈ Match	Accurate
I	7.0%	9%	+2%	Boosted: vowel balance
N	6.7%	6%	≈ Match	Accurate
Z	0.07%	1%	+0.93%	Minimum 1 for excitement

⚠️ The S Nerf

S appears in 6.3% of English text, suggesting 6 tiles. But Butts reduced it to 4 because S is the most powerful letter in Scrabble — it pluralizes nearly any word, creating easy hooks. With 6 S tiles, the game would lose strategic depth. The 4-tile limit makes each S a precious resource.

How Distribution Creates Game Balance

✓ Balanced Draws

42% vowels, 56% consonants ensures most 7-tile draws have a playable ratio. Only ~8% of initial draws are unplayable without exchange.

✓ Risk-Reward Tiles

High-value tiles (Q, Z, X, J) appear once each. Drawing one is exciting and rewarding, but holding it too long is punishing. Creates strategic tension.

The Point Value System

Point values are inversely related to frequency — rare letters are worth more. But it's not a perfect inverse; Butts tuned values for gameplay feel.

1 ptA,E,I,L,N,O,R,S,T,U 2 ptsD,G 3 ptsB,C,M,P 4 ptsF,H,V,W,Y 5 ptsK 8 ptsJ,X 10 ptsQ,Z

🎯 Summary

Scrabble's tile distribution is a masterpiece of game design — rooted in real English letter frequency but tuned with deliberate adjustments for strategic depth. The S nerf, the blank limit, and the single-tile high-value letters all create tension between risk and reward. After 85+ years, the distribution remains virtually unchanged because it works.

🔤 See real-time tile probability as you play — instant results, no signup

Open Word Finder →

Advanced Rack Management

Elite Scrabble players think in terms of rack leave — the tiles remaining after each play. A good rack leave balances vowels and consonants (ideally 3-4 consonants and 2-3 vowels), retains common letters like S, E, R, T, and avoids duplicates. Every word you play should be evaluated not just by its score but by what it leaves you to work with next turn.

The concept of synergy applies to Scrabble tiles just as it does to other games. Certain tile combinations work together beautifully — for example, holding ING or TION or NESS makes it easier to find 7-letter bingos. Conversely, holding double or triple copies of the same letter severely limits your options and often forces an exchange.

Tile tracking becomes increasingly important as the game progresses. By the midgame, you should have a rough idea of which high-value tiles remain in the bag. In the endgame (fewer than 7 tiles left in the bag), precise tracking allows you to deduce your opponent's exact rack and play accordingly — blocking their best plays while maximising your own remaining scoring opportunities.

Consider the concept of turnover — how many tiles you draw from the bag each turn. Playing 5-6 tiles per turn cycles through the bag faster, giving you more chances to draw premium tiles. Conversely, playing single-tile or two-tile words slows your turnover and reduces your overall access to the tile distribution across the full game.

← Back to all articles