A Compositional Characterization of the Z Z -relation via Closure and Additivity Aleksa Joksimović Abstract This paper gives a structural explanation for the Z Z -relation by modelling pitch-class sets as complete weighted graphs and encoding their interval content in a composition of n n via an additivity rule. We introduce the realization number R ⁡ ( μ , n ) R(\mu,n) — the count of T / I T/I -inequivalent compositions producing the same interval multiset μ \mu — and show that the Z Z -relation is precisely the condition R ⁡ ( μ , n ) ≥ 2 R(\mu,n)\geq 2 . We prove that the Z Z -relation cannot occur at cardinality k = 3 k=3 in any ℤ n \mathbb{Z}_{n} , construct an explicit primitive Z Z -related pair at k = 4 k=4 for every n n divisible by 4 4 with n ≥ 8 n\geq 8 , and establish a Scaling Theorem showing that any Z Z -pair in ℤ m \mathbb{Z}_{m} propagates to ℤ d ​ m \mathbb{Z}_{dm} for all d ≥ 1 d\geq 1 . All results are verified computationally for ℤ 12 \mathbb{Z}_{12} and ℤ 19 \mathbb{Z}_{19} . 1 Introduction Take two pitch-class sets and compute their interval vectors. If the vectors match but the sets are not related by transposition or inversion, the sets are called Z -related. This has been known since Forte’s 1973 classification [ 1 ] , and there are exactly 23 such pairs in 12-tone equal temperament. What has never been satisfactorily explained is why they exist — what it is about these particular sets that forces two different configurations to produce the same interval counts. The standard tools work in the frequency domain. The interval vector is equivalent to the squared magnitudes of the discrete Fourier transform [ 2 ] , and this observation underpins the theory of homometric sets [ 3 , 4 ] . These approaches characterize the Z -relation precisely but do not explain its mechanism. This paper provides a structural explanation. A pitch-class set in \mathbb{Z}_{n} written in ascending order as 0=p_{0}i is: \mathrm{ic}(p_{i},p_{j})=\min\!\left(\sum_{\ell=i+1}^{j}s_{\ell},\;n-\sum_{\ell=i+1}^{j}s_{\ell}\right). Non-adjacent interval classes are called diagonals and are each determined by a single addition and a single min. This is the additivity rule : the interval class between any two pitch classes equals the interval class of the sum of the steps between them. The collection of all \binom{k}{2} interval classes of P is the interval multiset \mu(P) , equivalently the interval vector written as a multiset. Since every interval class is determined by the composition (s_{1},\ldots,s_{k}) , the entire interval structure of P is encoded in its composition of n . Figure 1 shows all six pitch class graphs realizing the interval multiset \{3,4,5\} in \mathbb{Z}_{12} . These form the complete T/I orbit of \{0,3,7\} : transposition and inversion act as vertex relabelings preserving every edge weight, so all six graphs are isomorphic and the T/I class is represented by the single unlabeled triangle in Figure 2 . 0 3 7 3 4 5 0 4 7 4 3 5 0 5 8 5 3 4 0 4 9 4 5 3 0 3 8 3 5 4 0 5 9 5 4 3 Figure 1: All six realizations of the interval multiset \{3,4,5\} in \mathbb{Z}_{12} , forming the complete T/I orbit of \{0,3,7\} . 3 4 5 Figure 2: Canonical weighted triangle representing the T/I class of \{0,3,7\} with interval multiset \{3,4,5\} . 3 The Realization Number Definition 3.1 (Realization) . A realization of an interval multiset \mu in \mathbb{Z}_{n} is a composition (s_{1},\ldots,s_{k}) of n into k positive parts whose interval multiset, computed via the additivity rule, equals \mu . Since cyclic rotations of a composition correspond to transpositions of the underlying pitch-class set and reversals correspond to inversions, two realizations are equivalent if one can be obtained from the other by a cyclic rotation or a reversal of the step sequence. Definition 3.2 (Realization number) . The realization number R(\mu,n) is the number of distinct realizations of \mu in \mathbb{Z}_{n} , counted up to equivalence. Proposition 3.3 . Two pitch-class sets P_{1},P_{2}\subseteq\mathbb{Z}_{n} are Z -related if and only if R(\mu,n)\geq 2 , where \mu=\mu(P_{1})=\mu(P_{2}) . Proof. Z -related means same interval multiset and not T/I equivalent. The first condition is \mu(P_{1})=\mu(P_{2})=\mu . The second is that their compositions are not equivalent under cyclic rotation or reversal. Together these give R(\mu,n)\geq 2 . ∎ An interval multiset with R(\mu,n)=1 is reconstructible : its interval vector uniquely determines the set up to T/I equivalence. An interval multiset with R(\mu,n)\geq 2 is non-reconstructible : two or more T/I -inequivalent sets share it. 4 Z -relation Cannot Occur at Cardinality Three Proposition 4.1 . For any n\geq 3 and any interval multiset \mu arising from a trichord in \mathbb{Z}_{n} , we have R(\mu,n)=1 . Proof. Let P\subseteq\mathbb{Z}_{n} be a trichord with p_{0}=0 and steps (s_{1},s_{2},s_{3}) . By the additivity rule, \mu(P)=\{\{\mathrm{ic}(s_{1}),\,\mathrm{ic}(s_{2}),\,\mathrm{ic}(s_{1}+s_{2})\}\} . Since s_{3}=n-s_{1}-s_{2} , the third step contributes \mathrm{ic}(s_{3})=\mathrm{ic}(s_{1}+s_{2}) , already determined by s_{1} and s_{2} . Suppose (t_{1},t_{2},t_{3}) is a second realization of \mu , inequivalent under cyclic rotation and reversal. Then the two three-element multisets of interval classes are equal. Any two trichord compositions with the same interval class multiset must assign the same values to \mathrm{ic}(s_{1}),\mathrm{ic}(s_{2}),\mathrm{ic}(s_{3}) in some order. The six permutations of three elements correspond exactly to the three cyclic rotations and three reversals of the composition, i.e. to transpositions and inversions of the underlying set. Therefore any two realizations at k=3 are equivalent. ∎ Corollary 4.2 . The minimum cardinality at which Z -relation can occur in any \mathbb{Z}_{n} is k=4 . Proof. By Proposition 4.1 , R(\mu,n)=1 for all k\leq 3 . Theorem 6.3 in Section 6 exhibits R(\mu,n)=2 at k=4 for all n divisible by 4 with n\geq 8 . ∎ 5 Computational Verification 5.1 \mathbb{Z}_{12} Exhaustive enumeration of all compositions of 12 into k positive parts for k=3,\ldots,9 , grouped by interval multiset and T/I class, yields: k T/I classes Interval multisets R(\mu,12)\geq 2 3 12 12 0 4 29 28 1 5 38 35 3 6 50 35 15 7 38 35 3 8 29 28 1 9 12 12 0 Each non-reconstructible multiset contributes exactly one Z -related pair. Summing: 1+3+15+3+1=23 Z -related pairs, matching the known count [ 1 ] exactly. No multiset has R(\mu,12)\geq 3 , confirming that \mathbb{Z}_{12} admits no Z -groups of size greater than two. The distribution 0,1,3,15,3,1,0 is palindromic (a consequence of the complement theorem, k\leftrightarrow n-k ) and peaks at k=n/2=6 , where the composition simplex is largest and the partial sum structure admits the most inequivalent realizations. Consistent with Corollary 4.2 , no Z -relation occurs at k=3 . 5.2 \mathbb{Z}_{19} The same algorithm in \mathbb{Z}_{19} for k=3,\ldots,7 yields: k T/I classes Interval multisets R(\mu,19)\geq 2 3 30 30 0 4 120 120 0 5 324 324 0 6 756 735 21 7 1368 1311 57 No Z -relation occurs at k\leq 5 in \mathbb{Z}_{19} , a stronger result than in \mathbb{Z}_{12} where a Z -pair already appears at k=4 . The first Z -relations appear at k=6 . A representative Z -related pair at k=6 : P_{1}=\{0,1,2,3,6,10\},\quad P_{2}=\{0,1,2,4,5,11\}, with steps (1,1,1,3,4,9) and (1,1,2,1,6,8) respectively, both summing to 19 and producing the same interval multiset under the additivity rule. The enumeration algorithm was implemented with the assistance of Claude Sonnet 4.6 (Anthropic); all results were independently verified against the known count of 23 Z -related pairs in \mathbb{Z}_{12} [ 1 ] . 6 The Scaling Theorem and the General k=4 Construction The two results in this section are complementary. The Scaling Theorem shows that Z -relation propagates upward through the divisibility lattice: any Z -pair in \mathbb{Z}_{m} generates an infinite family of Z -pairs in \mathbb{Z}_{dm} for all d\geq 1 . The General k=4 Construction shows that primitive Z -pairs — those not arising as scalings — exist at every n divisible by 4 and gives them all explicitly. 6.1 The Scaling Theorem Theorem 6.1 (Scaling) . Let d\geq 1 and m\geq 3 . If (P_{1},P_{2}) is a Z -related pair in \mathbb{Z}_{m} , then dP_{1}=\{dp\bmod dm:p\in P_{1}\},\quad dP_{2}=\{dp\bmod dm:p\in P_{2}\} is a Z -related pair in \mathbb{Z}_{dm} . Proof. Same interval multiset. For any p,q\in\mathbb{Z}_{m} : \mathrm{ic}(dp,dq,dm)=\min\bigl(d|p-q|_{m},\,dm-d|p-q|_{m}\bigr)=d\cdot\mathrm{ic}(p,q,m). Therefore \mu(dP_{i})=d\cdot\mu(P_{i}) for i=1,2 , and since \mu(P_{1})=\mu(P_{2}) , we have \mu(dP_{1})=\mu(dP_{2}) . T/I inequivalence. Suppose dP_{1}\sim_{T/I}dP_{2} in \mathbb{Z}_{dm} . Then there exists t\in\mathbb{Z}_{dm} such that \{dp+t:p\in P_{1}\}\equiv\{dq:q\in P_{2}\}\pmod{dm} or the inversion analog. Since all elements of dP_{1} and dP_{2} are multiples of d , we need t\equiv 0\pmod{d} , so t=ds for some s\in\mathbb{Z}_{m} . Then P_{1}+s\equiv P_{2}\pmod{m} , giving P_{1}\sim_{T}P_{2} in \mathbb{Z}_{m} , a contradiction. The inversion case is identical. ∎ Corollary 6.2 (Inheritance) . If m\mid n and \mathbb{Z}_{m} has a Z -related pair at cardinality k , then \mathbb{Z}_{n} has a Z -related pair at cardinality k . Proof. Set d=n/m and apply Theorem 6.1 . ∎ The Scaling Theorem partitions all Z -pairs into derived pairs (scalings of a smaller Z -pair) and primitive pairs (not arising as any scaling). The inheritance corollary provides a sufficient but not necessary condition for Z -relation in \mathbb{Z}_{n} : a Z -pair may exist at a cardinality smaller than any proper divisor can supply, via a primitive pair arising from the intrinsic arithmetic of \mathbb{Z}_{n} itself. The case n=20 illustrates this: k_{\min}(10)=5 , so inheritance from \mathbb{Z}_{10} predicts no Z -pair below k=5 , yet k_{\min}(20)=4 , achieved by the primitive pair from Theorem 6.3 with m=10 and a=1 . 6.2 The General k=4 Construction The following construction was found independently by Althuis and Göbel [ 7 ] , who state it as Property 1 of their memorandum and verify it by a direct interval check. We reproduce it here within the realization number framework, give a structural proof identifying the step exchange responsible for the coincidence of interval multisets, and derive the primitivity condition (Corollary 6.5 ) that does not appear in their work. The Scaling Theorem (Theorem 6.1 ) is a distinct operation from their Observation 3: whereas they multiply all elements of a set by a fixed integer within the same \mathbb{Z}_{n} , we scale both the elements and the modulus, mapping P\subset\mathbb{Z}_{m} to dP\subset\mathbb{Z}_{dm} . Theorem 6.3 (General k=4 construction) . Let n be divisible by 4 and set m=n/2 . For any integer a with 1\leq a0 ), and \tfrac{m}{2}+a\neq m (as a4 for all primes p\neq 13 ? 3. Closed-form criterion for Z -relation. Is there an arithmetic condition on \mu and n predicting R(\mu,n)\geq 2 without enumeration? This is equivalent to the Turnpike problem in the cyclic discrete setting [ 6 ] , open since the 1970s.