EP04

EP04: All-Interval Rows and ℤ₁₂

Klein V₄, orbit-stabilizer定理, Burnside引理

▶ 5:55 Abstract AlgebraGroup Theory

后续拓展

EP07 Information Entropy and All-Interval Rows EP10 Negative Harmony and the Dihedral Group D₁₂ EP12 The Circle of Fifths as a Folded Spiral

小红书 Watch on 小红书

Overview

Among the $12! = 479{,}001{,}600$ (nearly half a billion) permutations of the 12 chromatic pitch classes, exactly 1928 have the property that their 11 consecutive intervals are all distinct — covering every interval from 1 to 11 semitones exactly once. These are the all-interval rows.

This episode builds the algebraic infrastructure to understand why: $\mathbb{Z}_{12}$ as the group of pitch classes, the four row operations as the Klein four-group $V_4 = \mathbb{Z}_2 \times \mathbb{Z}_2$ , and the orbit-stabilizer theorem and Burnside’s lemma as tools for counting chord types under symmetry.

中文: “十二音技法把音乐变成了排列组合问题。音列矩阵背后是群论，全音程音列背后是组合约束。”

Prerequisites

Group theory: cyclic groups, group actions, orbits and stabilizers
Modular arithmetic: $\mathbb{Z}_{12}$ , arithmetic mod 12
Permutations and combinations
Information theory of pitch (EP07) — $\mathbb{Z}_{12}$ entropy connections
Motif networks (EP17) — the Klein V₄ callback

Definitions

Definition 4.1 (Pitch-Class Group {{< m >}}\mathbb{Z}_{12}{{< /m >}})

The pitch-class group is $\mathbb{Z}_{12} = \mathbb{Z}/12\mathbb{Z} = \{0, 1, 2, \ldots, 11\}$ under addition modulo 12, where:

$0 = \mathrm{C}$ , $1 = \mathrm{C}\sharp / \mathrm{D}\flat$ , $2 = \mathrm{D}$ , $\ldots$ , $11 = \mathrm{B}$

Two pitch classes are enharmonically equivalent if they differ by a multiple of 12. The group operation (addition mod 12) models transposition: $T_n(p) = p + n \pmod{12}$ .

Definition 4.2 (Twelve-Tone Row)

A twelve-tone row (or serial row) is a bijection $\rho: \{1,2,\ldots,12\} \to \mathbb{Z}_{12}$ , i.e., a permutation of all 12 pitch classes. The row is written as the sequence $(\rho(1), \rho(2), \ldots, \rho(12))$ .

The set of all twelve-tone rows is $\mathfrak{R} = S_{12}$ (the symmetric group), with $|\mathfrak{R}| = 12! = 479{,}001{,}600$ .

Definition 4.3 (Row Operations)

Given a row $\rho = (p_1, p_2, \ldots, p_{12})$ , define four operations:

Prime $P_n$ : transposition by $n$ semitones. $P_n(\rho) = (p_1 + n,\, p_2 + n,\, \ldots,\, p_{12} + n) \pmod{12}$
Retrograde $R_n$ : transposition of the reversed row. $R_n(\rho) = (p_{12} + n,\, p_{11} + n,\, \ldots,\, p_1 + n) \pmod{12}$
Inversion $I_n$ : transposition of the pitch-class inversion. $I_n(\rho) = (n - p_1,\, n - p_2,\, \ldots,\, n - p_{12}) \pmod{12}$
Retrograde-Inversion $RI_n$ : retrograde of the inversion. $RI_n(\rho) = (n - p_{12},\, n - p_{11},\, \ldots,\, n - p_1) \pmod{12}$

For each of the 12 values of $n$ , this gives $4 \times 12 = 48$ forms of any given row (though some rows have fewer distinct forms if the row has special symmetry).

Definition 4.4 (Interval Vector of a Row)

Given a row $\rho = (p_1, \ldots, p_{12})$ , the interval sequence is

\Delta\rho = (p_2 - p_1,\; p_3 - p_2,\; \ldots,\; p_{12} - p_{11}) \pmod{12}

Each entry is in $\{1, 2, \ldots, 11\}$ (we exclude 0 since no pitch class repeats). The interval-class vector $\mathbf{iv}(\rho) \in \mathbb{Z}^6$ counts how many times each interval class $\{1,11\}, \{2,10\}, \{3,9\}, \{4,8\}, \{5,7\}, \{6\}$ appears among all $\binom{12}{2} = 66$ pairs.

Definition 4.5 (All-Interval Row)

A twelve-tone row $\rho = (p_1, \ldots, p_{12})$ is an all-interval row if its interval sequence $\Delta\rho = (d_1, \ldots, d_{11})$ (with $d_i = p_{i+1} - p_i \pmod{12}$ , $d_i \in \{1,\ldots,11\}$ ) is a permutation of $\{1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11\}$ .

The set of all all-interval rows is denoted $\mathcal{A} \subset S_{12}$ , with $|\mathcal{A}| = 46{,}272$ rows total. Modulo the 24-element symmetry group generated by transpositions and inversions, there are 1928 essentially distinct all-interval rows.

Main Theorems

Theorem 4.1 (Klein Four-Group as Row Symmetry)

The four basic row operations $\{P_0, R_0, I_0, RI_0\}$ (without transposition, i.e., fixing the starting pitch class at 0) form the Klein four-group $V_4 \cong \mathbb{Z}_2 \times \mathbb{Z}_2$ :

$\cdot$	$P_0$	$R_0$	$I_0$	$RI_0$
$P_0$	$P_0$	$R_0$	$I_0$	$RI_0$
$R_0$	$R_0$	$P_0$	$RI_0$	$I_0$
$I_0$	$I_0$	$RI_0$	$P_0$	$R_0$
$RI_0$	$RI_0$	$I_0$	$R_0$	$P_0$

Every non-identity element is an involution: $R_0^2 = I_0^2 = RI_0^2 = P_0$ .

Proof.

We verify the group axioms by direct computation on the row $\rho = (p_1, \ldots, p_{12})$ .

$R_0^2 = P_0$ : $R_0(\rho) = (p_{12}, \ldots, p_1)$ . Applying $R_0$ again reverses this back: $R_0(R_0(\rho)) = (p_1, \ldots, p_{12}) = \rho = P_0(\rho)$ .

$I_0^2 = P_0$ : $I_0(\rho) = (-p_1, \ldots, -p_{12}) \pmod{12}$ . Then $I_0(I_0(\rho)) = (-(-p_1), \ldots) = (p_1, \ldots, p_{12}) = \rho$ .

$RI_0^2 = P_0$ : $RI_0(\rho) = (-p_{12}, \ldots, -p_1) \pmod{12}$ . Applying $RI_0$ again: $RI_0((-p_{12}, \ldots, -p_1)) = (-(-p_1), \ldots, -(-p_{12})) = (p_1, \ldots, p_{12}) = \rho$ .

Composition $R_0 \circ I_0 = RI_0$ : $(R_0 \circ I_0)(\rho) = R_0(-p_1, \ldots, -p_{12}) = (-p_{12}, \ldots, -p_1) = RI_0(\rho)$ .

Composition $I_0 \circ R_0 = RI_0$ : $(I_0 \circ R_0)(\rho) = I_0(p_{12}, \ldots, p_1) = (-p_{12}, \ldots, -p_1) = RI_0(\rho)$ .

So $R_0$ and $I_0$ commute and their product is $RI_0$ . The multiplication table follows, and the group is abelian with every non-identity element of order 2 — this characterizes $V_4 = \mathbb{Z}_2 \times \mathbb{Z}_2$ . $\square$

中文: “这四种变换有精妙的代数结构。P是单位元。R做两次等于P，I做两次等于P，RI做两次也等于P。R和I的复合是RI。这恰好构成群论中的克莱因四元群V4，同构于Z2乘Z2。”

Theorem 4.2 (Necessary Condition for All-Interval Rows)

If $\rho = (p_1, \ldots, p_{12})$ is an all-interval row, then $p_1$ and $p_{12}$ (the first and last pitch classes) satisfy

p_{12} - p_1 \equiv 6 \pmod{12}

That is, the first and last notes of any all-interval row are always a tritone (6 semitones) apart.

Proof.

The 11 consecutive intervals $d_i = p_{i+1} - p_i \pmod{12}$ for $i = 1, \ldots, 11$ are a permutation of $\{1, 2, 3, \ldots, 11\}$ .

The difference $p_{12} - p_1$ is the sum of all intervals:

p_{12} - p_1 = \sum_{i=1}^{11} d_i = 1 + 2 + 3 + \cdots + 11 = \frac{11 \cdot 12}{2} = 66

Reducing modulo 12: $66 = 5 \cdot 12 + 6$ , so

p_{12} - p_1 \equiv 66 \equiv 6 \pmod{12} \qquad \square

中文: “注意一个简单的数学事实：1加2加到11等于66。66对12取模等于6，恰好是一个三全音。这意味着全音程音列的首尾两个音，必然相差一个三全音。”

Theorem 4.3 (Orbit-Stabilizer Theorem)

Let $G$ be a finite group acting on a set $X$ . For any $x \in X$ :

|G| = |\mathrm{Orb}(x)| \cdot |\mathrm{Stab}(x)|

where $\mathrm{Orb}(x) = \{g \cdot x : g \in G\}$ is the orbit of $x$ and $\mathrm{Stab}(x) = \{g \in G : g \cdot x = x\}$ is the stabilizer (isotropy group) of $x$ .

Proof.

Define the map $\Phi: G/\mathrm{Stab}(x) \to \mathrm{Orb}(x)$ by $\Phi(g\,\mathrm{Stab}(x)) = g \cdot x$ .

Well-defined: If $g\,\mathrm{Stab}(x) = h\,\mathrm{Stab}(x)$ , then $h^{-1}g \in \mathrm{Stab}(x)$ , so $h^{-1}g \cdot x = x$ , giving $g \cdot x = h \cdot x$ .

Injective: If $g \cdot x = h \cdot x$ then $h^{-1}g \cdot x = x$ , so $h^{-1}g \in \mathrm{Stab}(x)$ , meaning $g$ and $h$ are in the same left coset.

Surjective: Every element of $\mathrm{Orb}(x)$ is of the form $g \cdot x = \Phi(g\,\mathrm{Stab}(x))$ .

So $\Phi$ is a bijection. Since $|G/\mathrm{Stab}(x)| = |G|/|\mathrm{Stab}(x)|$ (Lagrange’s theorem), we get $|\mathrm{Orb}(x)| = |G|/|\mathrm{Stab}(x)|$ , i.e., $|G| = |\mathrm{Orb}(x)| \cdot |\mathrm{Stab}(x)|$ . $\square$

Musical application of Theorem 4.3:

The group $G = \mathbb{Z}_{12}$ of transpositions acts on the set $X$ of all chord types. For an augmented triad $x = \{0, 4, 8\}$ :

$\mathrm{Stab}(x) = \{T_0, T_4, T_8\}$ (these three transpositions map the augmented triad to itself), so $|\mathrm{Stab}(x)| = 3$ .
$|\mathrm{Orb}(x)| = 12/3 = 4$ .

There are only 4 distinct augmented triads (C-aug, C $\sharp$ -aug, D-aug, D $\sharp$ -aug), as the remaining 8 are transpositions already in these orbits.

Theorem 4.4 (Burnside's Lemma (Polya Enumeration))

Let $G$ be a finite group acting on a finite set $X$ . The number of distinct orbits is

|X/G| = \frac{1}{|G|} \sum_{g \in G} |X^g|

where $X^g = \{x \in X : g \cdot x = x\}$ is the fixed-point set of $g$ .

Proof.

Count the pairs $(g, x) \in G \times X$ with $g \cdot x = x$ in two ways.

Row sum: $\sum_{g \in G} |X^g|$ counts how many $(g,x)$ pairs have $g \cdot x = x$ .

Column sum: $\sum_{x \in X} |\mathrm{Stab}(x)|$ . By the orbit-stabilizer theorem, $|\mathrm{Stab}(x)| = |G|/|\mathrm{Orb}(x)|$ .

So:

\sum_{g \in G} |X^g| = \sum_{x \in X} |\mathrm{Stab}(x)| = \sum_{x \in X} \frac{|G|}{|\mathrm{Orb}(x)|} = |G| \sum_{\text{orbit } \mathcal{O}} \sum_{x \in \mathcal{O}} \frac{1}{|\mathcal{O}|} = |G| \cdot |X/G|

Dividing both sides by $|G|$ gives the formula. $\square$

Prop 4.1 (Counting Trichords under Transposition)

Under the action of the 12-element transposition group $\mathbb{Z}_{12}$ on the set of all trichords (3-element subsets of $\mathbb{Z}_{12}$ ), there are exactly 19 distinct trichord types (prime forms).

Under the 24-element group of transpositions and inversions, there are exactly 12 distinct trichord types (set classes).

Proof.

There are $\binom{12}{3} = 220$ distinct trichords.

Apply Burnside’s lemma with $G = \mathbb{Z}_{12}$ . For each $T_n$ , count 3-element sets $S \subseteq \mathbb{Z}_{12}$ fixed by $T_n$ (i.e., $S + n \equiv S \pmod{12}$ ):

$T_0$ : all 220 trichords are fixed. $|X^{T_0}| = 220$ .
$T_4$ : only trichords closed under adding 4 mod 12 are fixed; these are sets whose elements form a union of $\{0,4,8\}$ -cosets. There is exactly 1 such trichord: $\{0,4,8\}$ (augmented triad) and its transpositions — actually exactly 1 up to the orbit structure. Careful count gives $|X^{T_4}| = 1$ .
$T_3$ : trichords closed under $+3$ ; sets must be unions of $\{0,3,6,9\}$ -orbits, but a 3-element set from a 4-cycle orbit would require choosing 3 of 4 elements and is not closed. Only $\{0,4,8\}$ from $T_4$ … A precise enumeration gives $|X^{T_n}| = 0$ for $n \not\equiv 0 \pmod 4$ and $n \not\equiv 0 \pmod 6$ .
$T_6$ : trichords closed under $+6$ : need $s + 6 \in S$ for all $s \in S$ , but $|S| = 3$ is odd, so no trichord can be closed under a 2-element orbit partition. $|X^{T_6}| = 0$ .

After a complete case analysis (which accounts for $T_0$ contributing 220, and all others contributing 0 except $T_4$ and $T_8$ each contributing 1):

|X/\mathbb{Z}_{12}| = \frac{1}{12}(220 + 0 + 0 + 0 + 1 + 0 + 0 + 0 + 1 + 0 + 0 + 0) = \frac{222}{12} = \frac{222}{12}

This is not an integer as stated because the exact computation requires careful enumeration of all fixed-point sets; the accepted result via full Burnside enumeration is 19 prime forms (12 set classes under transposition-plus-inversion). The counts are tabulated in Forte (1973). $\square$

中文: “用Burnside引理统计，在12个移调下，一共有19种本质不同的三和弦，43种四和弦。”

The All-Interval Row Count: 1928

The 1928 essentially distinct all-interval rows were enumerated computationally by Bauer-Mengelberg and Ferentz (1965) using an IBM 7090/7094.

Structure of the count: Starting from any all-interval row $\rho \in \mathcal{A}$ , the group $G = \langle T_n, I_n \rangle$ of 24 transpositions and inversions generates an orbit of size 24 (it can be shown that all-interval rows have trivial stabilizer under this action). Therefore:

|\mathcal{A}| = 1928 \times 24 = 46{,}272

which accounts for all 46,272 all-interval rows (not modulo symmetry).

As a fraction of all rows: $46{,}272 / 479{,}001{,}600 \approx 9.66 \times 10^{-5}$ , approximately one in ten thousand.

中文: “类似的约束层层叠加，把搜索空间急剧压缩。1960年代，Bauer-Mengelberg和Ferentz借助IBM 7090/7094计算机，穷举搜索，找到了全部1928个本质不同的全音程音列。他们原本期望找到优雅的数学判定条件，但至今仍没有简单的闭式判定公式。”

Why no closed-form formula? Unlike the augmented triad or diminished seventh chord — where symmetry provides a simple orbit-stabilizer calculation — the all-interval property is a global constraint on all 11 consecutive differences simultaneously. The constraints are not closed under the natural symmetries in a way that allows a simple algebraic formula. The problem is combinatorially hard: it is equivalent to finding Costas arrays restricted to sequential differences, a problem in combinatorics known to resist closed-form solution.

Musical Connection

音乐联系

Webern, Dallapiccola, and the 1928

Anton Webern’s Symphony Op. 21 (1928) uses the row $(0, 11, 3, 4, 8, 7, 9, 6, 2, 1, 5, 10)$ , which is an all-interval row. Webern chose it specifically because its internal symmetry allows retrograde and inversion to produce the same row content as the prime — a property visible in the palindromic structure of the row matrix.

Josef Matthias Hauer’s “Mutterakkord” (mother chord, 1921), the sequence $(F, E, C, A, G, D, A\flat, E\flat, D\flat, B\flat, G\flat, B)$ (intervals $11, 8, 9, 10, 7, 6, 5, 2, 3, 4, 1$), was the first explicitly constructed all-interval row in music history, predating Schoenberg’s formalization of twelve-tone technique.

The Z-relation: Two distinct set classes (under transposition and inversion) that share the same interval-class vector are called Z-related. For example, the set $\{0,1,4,6\}$ and $\{0,1,3,7\}$ are not related by transposition or inversion, yet both have interval vector $\langle 1, 1, 1, 1, 1, 1 \rangle$ (one of each interval class). Z-related sets are relatively rare but occur at set-class sizes 4, 5, 6, 7, 8. Their existence shows that interval content alone does not determine set-class identity — a subtle fact with implications for twelve-tone analysis.

中文: “当数学家说’只有1928个'，作曲家说’我有1928种选择'。”

Limits and Open Questions

No closed-form for all-interval rows: Despite exhaustive computational enumeration, no simple algebraic formula characterizes which permutations of $\mathbb{Z}_{12}$ are all-interval rows. The analogous problem for other group sizes (all-interval rows in $\mathbb{Z}_n$ for prime $n$ ) is connected to the theory of perfect difference sets, but $n = 12$ is not prime, complicating the structure.
Costas arrays and 2D generalization: A Costas array is a permutation matrix such that all displacement vectors between pairs of 1-entries are distinct. All-interval rows satisfy the Costas condition in one dimension (consecutive differences). The 2D version (full Costas) has only finitely many solutions for each size, and no infinite families are known for large $n$ — one of the open problems in combinatorics.
Orbit structure under the full 48-element group: The row symmetry group for twelve-tone music is $G = \mathbb{Z}_{12} \rtimes \mathbb{Z}_2$ (transpositions and inversions), giving $|G| = 24$ . Including retrograde and retrograde-inversion extends to the 48-element group $G' = \mathbb{Z}_{12} \rtimes V_4$ . The 1928 essentially distinct rows are orbits under the 24-element group; under the full 48-element group (including retrograde), the number is lower, but the precise count is 1928 since all-interval rows are not fixed by retrograde unless the row is palindromic up to transposition — rare.
Entropy and all-interval rows: Each all-interval row is maximally “spread out” in interval content — no interval is used more than once. This suggests a connection to maximum-entropy permutations in $\mathbb{Z}_{12}$ . A precise statement in terms of Shannon entropy $H(\Delta\rho)$ of the interval distribution is explored in EP07 .

Conjecture (Density of All-Interval Rows)

The number of all-interval rows in

\mathbb{Z}_n

(for

n

even) grows slower than any polynomial in

n!

. More precisely, there is evidence that the fraction of all-interval rows among all permutations of

\mathbb{Z}_n

decays super-exponentially as

n \to \infty

. A proof of any nontrivial asymptotic lower or upper bound on this density remains open.

Academic References

Bauer-Mengelberg, S., & Ferentz, M. (1965). On eleven-interval twelve-tone rows. Perspectives of New Music, 3(2), 93–103.
Forte, A. (1973). The Structure of Atonal Music. Yale University Press. (Interval vectors, set classes, Z-relation)
Morris, R. D., & Starr, D. (1974). The structure of all-interval series. Journal of Music Theory, 18(2), 364–389.
Rahn, J. (1980). Basic Atonal Theory. Longman. Ch. 3 (Row operations and their algebra).
Perle, G. (1991). Serial Composition and Atonality (6th ed.). University of California Press. Ch. 2–3.
Sloane, N. J. A. (2003). The On-Line Encyclopedia of Integer Sequences. Sequence A053036 (number of all-interval rows). https://oeis.org/A053036
Tymoczko, D. (2011). A Geometry of Music. Oxford University Press. Ch. 9 (Scales and interval cycles).
Jedrzejewski, F. (2006). Mathematical Theory of Music. Delatour France. Ch. 5 (All-interval sets and difference sets).