Math paradoxes

The Pythagorean comma Twelve pure fifths up from C. Back to C?No: 1.36% too high.

A pure fifth multiplies the frequency by 3/2, an octave by 2. Go up from C in fifths: G, D, A… On a piano the twelfth fifth lands on C again, seven octaves up. But what if every fifth is pure? Lock in your answer, then listen to the circle of fifths, search for better cycles on a computer and read the explanation.

A pure fifth multiplies the frequency by 3/2, an octave by 2. Twelve fifths should equal seven octaves, but (3/2)¹² ≈ 129.75 while 2⁷ = 128. The circle of fifths does not close: the gap is 531,441 / 524,288, the Pythagorean comma, about a quarter of a semitone. So no piano can have all its fifths and octaves pure at once, and modern tuning narrows every fifth a tiny bit.

Make a guess
+1.36%12 fifths beat 7 octaves
129.75 / 128(3/2)¹² / 2⁷
1.4983a piano fifth instead of 1.5
The puzzle

The circle of fifths, with pure fifths

Musical notes here are just frequencies, and intervals are multipliers. You won't need anything beyond multiplying fractions. Notes are named with letters, as on keyboard apps: ♯ is a sharp.

  1. 1

    An octave: frequency × 2

    A string half as long sounds an octave higher: its frequency is doubled. It is the same note, only higher: C at 261.6 hertz and C at 523.3 hertz.

  2. 2

    A pure fifth: frequency × 3/2

    A string 2/3 as long sounds a fifth higher: the frequency is 3/2 times as high. From C that is G. The two notes together sound smooth and steady, and instruments were tuned from such fifths for centuries.

  3. 3

    Twelve fifths in a row

    From C we go up in fifths, and whenever a note climbs past the octave, we drop it an octave (halve the frequency) to stay on the same keys:

    CGDAEBF♯C♯G♯D♯A♯E♯B♯ ?
  4. 4

    The question

    On a piano B♯ is the same key as C: a fifth there is exactly 7 keys, and 12 × 7 = 84 keys is exactly 7 octaves of 12. But we tuned with pure fifths. Where have we arrived: exactly at the starting C, a little higher or a little lower?

Intuition says

"The circle of fifths is a circle: twelve steps and we are home"

In fact the circle closes only on a piano, where the fifths are narrowed a tiny bit. Pure fifths give (3/2)¹² ≈ 129.75, while seven octaves give 128. We arrive 1.36% too high, about a quarter of the step between neighbouring keys.

What we assume

  • A pure fifth is exactly 3/2, an octave exactly 2.
  • Dropping a note an octave means halving its frequency; the note stays the same note.
  • We compare the finish with the starting C after all those drops.
  • "A little" means less than the neighbouring piano key.
Round 1 · Guess

Where do 12 pure fifths lead?

Twelve pure fifths up from C, dropping octaves to stay on the keyboard. You are on C again. Pick an answer.

How does the finishing C sound next to the starting one?

A breakdown of your answer will appear here.

Hints

Hard to decide? Open the hints one at a time: the easy one first, then stronger ones.

Next: the solution

Next comes the explanation, with spoilers

The next rounds show where the circle of fifths leads: you walk it by ear, the computer tries 60 cycles of fifths, and then comes the explanation. Answer round 1 first, or go on?

You chose to always show all rounds.

Round 2 · Listen for yourself

Walk the circle of fifths by ear

Each press plays the next fifth: two notes together. The spiral shows pitch: one turn is one octave, so 12 fifths climb almost 7 turns. At the end, play the starting and finishing C together. If they differ, the sound "wobbles": the loudness swells and fades in waves, called beating. Then switch to piano fifths and compare. Sound starts only from a button; the volume is moderate.

Counters

Beating is how many times a second two notes together "wobble". A pure fifth does not beat. A piano fifth near middle C beats about once a second: barely audible, which is why a piano sounds clean in every key.

Round 3 · Computation

Is there a cycle of fifths that closes better?

The computer tries n = 1…60 pure fifths. For each n it finds the nearest number of octaves m and works out by what percentage (3/2)n misses 2m. A bar up is an overshoot, a bar down an undershoot.

error, % record: closer than every smaller n 12 fifths
Fifths
–
Nearest octaves
–
(3/2)ⁿ / 2ᵐ
–
Miss
–

Records

No bar is zero: pure fifths never close. After the first steps records are rare: 5, 12, 41, 53. After 12 fifths the miss is +1.36%, and the next better cycle has to wait until 41 fifths (−1.14%, an undershoot this time). And 53 fifths almost exactly equal 31 octaves: a miss of only 0.21%. The next record after 53 is not until 306 fifths.

Once you answer, the simulation code appears here so you can run it yourself.

Check it yourself in code

The same simulation, short and with no libraries. Run the JavaScript in your browser console (F12) or in Node; the Python needs only the standard library.

JavaScript
// How many pure fifths (×3/2) best match a whole number of octaves (×2)? We try n from 1 to 60.
let best = Infinity;
for (let n = 1; n <= 60; n++) {
  const fifths = Math.pow(1.5, n);                         // n fifths up
  const m = Math.round(Math.log2(fifths));                 // the nearest number of octaves
  const miss = fifths / Math.pow(2, m) - 1;                // the miss: + over, − under
  if (Math.abs(miss) < best) {                             // a record: closer than every smaller n
    best = Math.abs(miss);
    console.log('fifths:', n, '≈ octaves:', m, '· miss', (100 * miss).toFixed(3) + '%');
  }
}
// Exactly: 12 fifths / 7 octaves = 3^12 / 2^19
console.log('Pythagorean comma', 3 ** 12, '/', 2 ** 19, '=', (3 ** 12 / 2 ** 19).toFixed(5));
Python
# How many pure fifths (×3/2) best match a whole number of octaves (×2)? We try n from 1 to 60.
from fractions import Fraction
from math import log2

best = float("inf")
for n in range(1, 61):
    fifths = Fraction(3, 2) ** n                           # n fifths up, an exact fraction
    m = round(log2(fifths))                                # the nearest number of octaves
    miss = fifths / 2 ** m - 1                             # the miss: + over, − under
    if abs(miss) < best:                                   # a record: closer than every smaller n
        best = abs(miss)
        print("fifths:", n, "≈ octaves:", m, "· miss", f"{100 * float(miss):.3f}%")

# Exactly: 12 fifths / 7 octaves = 3^12 / 2^19
comma = Fraction(3, 2) ** 12 / 2 ** 7
print("Pythagorean comma", comma, "=", f"{float(comma):.5f}")
Round 4 · Explanations

Threes and twos never meet

The circle of fifths looks like a circle because on a piano 12 fifths of 7 keys make 84 keys, exactly 7 octaves. But a pure fifth is the multiplier 3/2, and it has a three in it that the octave lacks.

In plain wordspictures, analogies, small numbers

“In plain words” is hidden on all puzzles.

Odd and even

312 is odd, 219 is even

(3/2)¹²12 fifths
≠
2⁷7 octaves

(3/2)¹² = 3¹²/2¹². For that to equal 2⁷ we would need 3¹² = 2¹⁹. But 3¹² is a product of threes only, so it is odd, while 2¹⁹ is even. They can't be equal. The same goes for any number of fifths and octaves: no matter how many threes you multiply, you never get a power of two.

This card explains why the circle never closes, but not by how much it misses or in which direction. That is the next card.

A small error, twelve times

Every pure fifth is a little wider than a piano fifth

A piano fifth is the multiplier 1.4983, a pure one 1.5. The difference is tiny, 0.11%. But each next fifth starts from a note that is already a bit high, so the excesses add up: 0.23% after two fifths, 0.68% after six, 1.36% after twelve. Where the piano lands exactly on C, pure fifths overshoot.

Rigoroustables, charts, formulas

“Rigorous” is hidden on all puzzles.

Twelve steps

The notes of the circle: pure fifths against the piano

Each row is the previous note × 3/2, halved if it climbed past the octave. So the k-th note is 3k divided by a power of two. On a piano every fifth is 7 semitones of 21/12 ≈ 1.0595, and a note is 2(semitones)/12.

The comma

531,441 / 524,288 ≈ 1.0136

Twelve fifths are 3¹²/2¹² = 531,441 / 4,096 ≈ 129.746. Seven octaves are 128. Their ratio 3¹²/2¹⁹ = 531,441 / 524,288 ≈ 1.01364 is called the Pythagorean comma. A piano semitone is the multiplier 1.0595, or +5.95%, while the comma is +1.36%: about a quarter of a semitone. The ear easily hears such a gap when the two notes sound together.

Equal temperament

Every fifth narrowed by a twelfth of the comma

A modern piano splits the octave into 12 equal semitones, each the multiplier 21/12, the twelfth root of 2. A fifth becomes 27/12 ≈ 1.49831 instead of 1.5, 0.11% narrower. The circle of fifths now closes exactly: (27/12)¹² = 2⁷. The price is that no fifth is pure: near middle C a fifth beats about 0.89 times a second.

For formula lovers

The comma, the equal fifth and cents

Twelve pure fifths against seven octaves.

(3/2)1227=312219=531441524288≈1.01364

The equal-tempered fifth: seven semitones of 21/12.

27/12≈1.498307,1.527/12=(кома)1/12

Musicians measure intervals in cents: an octave is 1200 cents, a piano semitone 100. An interval with multiplier x is 1200 · log₂ x cents.

1200log232≈701.955,12·701.955−8400≈23.46

So a pure fifth is 1.955 cents wider than a piano fifth, and twelve of those excesses add up to 23.46 cents: the comma.

Deepergeneralisations, theory, history

“Deeper” is hidden on all puzzles.

For formula lovers

Why 12, 41 and 53: a continued fraction

n fifths ≈ m octaves means (3/2)n ≈ 2m, that is m/n ≈ log₂(3/2) = 0.5849625…, the number of octaves in a fifth. This number is irrational (for the same reason: odd ≠ even), so m/n can only approximate it.

log232=[0;1,1,2,2,3,1,5,2,23,…]
12,35,712,2441,3153,179306,…

Cutting the continued fraction short gives the best approximations: the denominators 5, 12, 41, 53, 306 are the records of round 3. 53 is so good because a large 5 follows it in the expansion: the next improvement is far away. Hence 12 keys in an octave, and a tuning with 53 equal steps.

History

From the Pythagoras legend to the twelfth root

The discovery of the 3/2 fifth is credited to Pythagoras, but that is a legend. The oldest story of blacksmiths whose hammers "sounded" in proportion appears in Nicomachus centuries later, and it is physically false: pitch does not depend on a hammer's weight. And the "Division of the Canon", ascribed to Euclid, already proves that six 9/8 tones exceed an octave, by exactly 531,441 : 524,288.

Pythagorean tuning takes 11 pure fifths, and the twelfth absorbs the whole comma, shrinking to 1.4798 instead of 1.5: it was nicknamed the "wolf" because it howls. Equal temperament with 21/12 was computed precisely by the Chinese prince Zhu Zaiyu in 1584, and, separately, in Europe by Simon Stevin around 1605 (the manuscript was printed in 1884).

Round 5 · Going further

Where to hide the comma, another comma, and the calendar

The comma can't be removed, only shared out among the fifths. Here we measure intervals in cents: an octave is 1200, a piano semitone 100, the comma 23.46.

The twelve fifths of four tunings

The maths

The deviations always add up to −23.46 cents

Let the fifths be anything, but 12 of them must make exactly 7 octaves for the keyboard to close. Then their deviations from pure always add up to minus the comma. Tunings differ only in where they hide those 23.46 cents.

Pythagorean tuning dumps it all into one wolf fifth. Equal temperament shares it evenly, −1.955 each. Andreas Werckmeister in "Musicalische Temperatur" (1691) narrowed four fifths (C–G, G–D, D–A, B–F♯) by a quarter of the comma and left eight pure. The quarter-comma meantone tuning of the 16th–17th centuries narrows 11 fifths even more, by 5.38 cents, for the sake of pure thirds (card below), and its wolf fifth comes out 35.7 cents wide.

The logic

The syntonic comma: four fifths against a pure third

The same trouble happens with fives. Four fifths from C lead to E: (3/2)⁴ / 4 = 81/64 = 1.2656. But a pure major third is 5/4 = 1.25. The gap is 81/80, about 21.5 cents, called the syntonic comma. So neither fifths with octaves nor fifths with thirds can all be pure together. A piano third (400 cents) is 13.7 cents wider than a pure one (386.3): its beating is much easier to hear than a fifth's.

The general idea: two cycles with incommensurable steps never line up exactly, because 3a ≠ 2b and 5a ≠ 2b. All you can choose is which error to tolerate, and where.

In practice

The calendar: the same comma between the day, the month and the year

The tropical year is about 365.2422 days, the synodic month 29.5306 days. Like fifths and octaves, these cycles do not divide evenly, and calendars hide the "comma" just as tunings do.

CyclesApproximationWhat is left over
Fifths and octaves12 fifths ≈ 7 octavesa comma of 1.36%; equal temperament spreads it over 12 fifths
Day and year, Julian calendar4 years ≈ 1,461 days (365.25 on average)the calendar drifts from the Sun by a day in about 128 years
Day and year, Gregorian calendar (1582)400 years = 146,097 days: 97 leap years (365.2425)an error of about a day in over 3,000 years
Month and year, the Metonic cycle19 years ≈ 235 monthsabout 2 hours per 19 years; the Hebrew calendar and the Easter computation rest on this cycle

The Metonic cycle is the calendar's "53 fifths": 235/19 is one of the best approximations to the number of months in a year, 12.368…, just as 31/53 approximates the number of octaves in a fifth.

Bonus · Fun facts

Things to tell your friends

531,441

A number from the "Division of the Canon"

The treatise "Division of the Canon", ascribed to Euclid, proves that six whole tones of 9/8 exceed an octave. (9/8)⁶ = 531,441 / 262,144, and dividing by 2 leaves the same comma, 531,441 / 524,288.

53 ≈ 31

A Chinese astronomer found a better cycle

Jing Fang, a scholar of the 1st century BC, noticed that 53 pure fifths almost equal 31 octaves. Nicholas Mercator computed the same in the 17th century, and the leftover, about 0.21%, is now called Mercator's comma.

1584

Equal temperament came from China

The Ming prince Zhu Zaiyu computed in 1584 a precise tuning in which every semitone is the multiplier 21/12 ≈ 1.059463. The Fleming Simon Stevin reached the same around 1605, but his manuscript was not printed until 1884.

1722

"Well-tempered" does not mean "equal"

The first book of Bach's "Well-Tempered Clavier", 1722: preludes and fugues in all 24 keys. For a long time it was assumed Bach meant equal temperament, but modern scholars lean towards an unequal "well" temperament, like Werckmeister's, in which every key is playable but each sounds a little different.

18 : 17

Frets on the neck

On a guitar each fret shortens the string by a factor of 21/12, so the 12th fret sits exactly halfway along the string. Vincenzo Galilei, Galileo's father, advised in 1584 placing lute frets by the ratio 18 : 17 ≈ 1.0588, very close to 1.0595.

−23.5

Why the fifth is a "wolf"

In Pythagorean tuning one fifth is 23.5 cents narrower than pure: it carries the whole comma. Its two notes together beat fast and harshly, a sound like howling, hence the name. Equal temperament has no wolf, because all its fifths are the same.

Discussion

What do you think?

Share your own explanation, an objection or a variant of the puzzle. You need a GitHub account to comment.

Is “School maths” the right level for this puzzle?

Thank you! This helps set the levels right.

Newsletter

New puzzles, explanations and discussions

I keep adding new puzzles, new ways to explain them and write-ups of the discussions they start. Leave your email and I'll write when something new is out. No spam, one-click unsubscribe.

The sign-up form will appear here soon.