Spectral Inaccessibility on the Poincaré Homology Sphere

On S³/2I, every intrinsic spectral construction in the admissible class defined below can read L-function structure with arbitrary precision but cannot constrain individual L-function zeros. The reading capacity and the writing incapacity are two faces of a single structural fact, forced by the geometry, representation theory, and arithmetic of the manifold.

I. Definitions

Definition 1 (Admissible Operators). The admissible operators on $S^3/2I$ are the natural, self-adjoint elliptic operators on sections of natural bundles within the following scoped class:

• Hodge Laplacians on $p$-forms ($p = 0,1,2,3$)
• The Dirac operator and odd signature operator
• Flat connection twists of the above under the three vacua (trivial, standard, Galois)

Every natural bundle over $S^3/2I = \mathrm{SU}(2)/2I$ corresponds to a representation of $2I$. This is the admissible class considered here for natural bundles arising from representations of $2I$; operators on bundles outside this classification are outside the scope of the present theorem.

Definition 2 (Admissible Spectral Constructions). Let $\mathcal{F}$ be the class of admissible spectral constructions on $S^3/2I$: all intrinsic, basis-independent constructions generated from Definition 1 operators under finite algebraic operations, consisting of:

• The spectral zeta $Z_A(s)$, eta invariant $\eta_A(s)$, and analytic torsion $T_A$ for any admissible operator $A$
• Equivariant class averages over conjugacy classes of $2I$
• Finite additive or multiplicative combinations of the above
• Basis-independent functions of the $2I$ character table and McKay multiplicities (explicitly: symmetric functions of eigenvalues and character-weighted spectral sums)

Constructions outside $\mathcal{F}$ are outside the scope of the present theorem.

Definition 3 (Reading). A construction $F\in\mathcal{F}$ reads arithmetic if it decomposes into Dirichlet or Hecke L-values or L-functions with coefficients determined by the group-theoretic data of $2I$.

Definition 4 (Zero Constraint). A construction $F\in\mathcal{F}$ constrains zeros of a target L-function $L(s,\chi)$ only if

$F(s_0)=0 \implies L(s_0,\chi)=0.$

A relation of the form $L(s,\chi)=L(s-2,\chi)$, or $L(s,\chi_1)=L(s,\chi_2)$ for distinct characters, is a coincidence condition, not a zero constraint.

II. Main Theorem

Theorem 1 (Spectral Inaccessibility — $S^3/2I$). Let $\mathcal{F}$ be the admissible class of Definition 2. Then no $F\in\mathcal{F}$ constrains zeros of any individual Dirichlet or Hecke L-function attached to the $2I$ arithmetic data in the sense of Definition 4. Every vanishing condition on $F$ reduces to one of:

(i) a shifted-value coincidence condition (Definition 4), (ii) encoding degeneracy: multiplicative spectral structure collapses on finite groups, with spectral zeros reducing to holonomy or cyclotomic phases rather than L-function zeros, (iii) framework mismatch between spectral Dirichlet L-functions over $\mathbb{Q}$ and arithmetic Hecke L-functions over number fields, or (iv) character completeness: basis-independent content is exhausted by the class table of $2I$.

Proof: by Lemmas 1–8 in § IV.

Corollary 1. Within the admissible class $\mathcal{F}$ on $S^3/2I$, the four obstruction layers — coincidence condition, encoding degeneracy, framework mismatch, and character completeness — are exhaustive: every vanishing condition on every $F\in\mathcal{F}$ reduces to at least one of them.

Note on reading without constraining. The manifold reads the Riemann zeta function directly through the $C8$ equivariant eta:

$\eta([C8],s) = 2^{s-1}[L(s,\chi_3) - \zeta(s-1)].$

This is reading, not constraining. Isolating a zero of $\zeta(s)$ would require $L(s,\chi_3)$ to vanish simultaneously — a coincidence condition in the sense of Definition 4.

III. The Reading

The spectral geometry of $S^3/2I$ reads L-function structure completely. Each result is proved and locked. Together they motivate the question answered in § IV: whether this reading capacity can be leveraged to constrain individual zeros.

A. Torsion L-Basis

The Reidemeister torsion (analytic torsion by Cheeger-Müller) for all integer-spin irreps of $2I$ factors exactly into four Dirichlet L-function special values, with $E_8$ McKay symmetries killing 12 of 16 characters mod 60. Four characters survive, at conductors 2, 3, 5, and 5: exactly the primes dividing $\lvert 2I\rvert=120$.

Two independent derivation paths (combinatorial Reidemeister + spectral analytic torsion via Kummer/Gauss). Verified to 79 digits. The Galois pair satisfies $T^2(R_3)/T^2(R_4) = \varphi^{-4}$ (exact), with the golden ratio entering through the Legendre character of $\mathbb{Q}(\sqrt{5})$.

Selectivity: maximum (4 of 16 characters). Domain: $s=0$ only.

For half-integer irreps (R1, R2, R6, R8), the torsion values are computed by the same spectral method to equivalent precision. The difference is selectivity: 28–32 characters survive rather than 4, and the results remain as finite combinations of $L'(0,\chi)$ values without reducing to algebraic closed forms.

B. Dirac Factorization

The Dirac operator on $S^3/2I$ has eigenvalues $\pm(n+3/2)/R$, linear in $n$ with no curvature shift. Its spectral zeta factors finitely at all $s$ into three L-function layers: $L(s,\chi)$, $L(s-1,\chi)$, $L(s-2,\chi)$ for Dirichlet characters mod 120. The 120-grid carries $\phi(120)=32$ characters; of these, 28–32 survive depending on the irrep.

Selectivity: minimum (28–32 of 32 characters). Domain: all $s$.

The contrast with the torsion defines the landscape: maximum selectivity at a single point, or minimum selectivity across a full strip. § IV tests whether that reading capacity can be leveraged to constrain individual zeros.

C. Eta Character Selection

The Dirac eta invariant

$\eta_D(\sigma,s) = \sum_{n}(m_n^+ - m_n^-)(n+3/2)^{-s}$

kills 60–80% of Dirichlet characters mod 120 in the coprime sector. The killing pattern maps to the stabilizer structure of the icosahedron through $\mathrm{CRT}(8,3,5)$:

Exact rational values at $s=0$ verified by two independent paths (Hurwitz + Donnelly) to $10^{-15}$. All denominators divide $720=6!$.

D. Equivariant Eta and the C2 Closed Form

The character-table rotation transforms 9 per-irrep etas (6–12 survivors each) into 8 non-trivial conjugacy-class combinations (1–4 survivors each; the identity class contributes zero by dimension-weighted cancellation). The central element $C2=[-e]$ isolates a single L-function at all $s$:

$\eta([-e],s) = 2^{s-1}\cdot[\beta(s)-\beta(s-2)]$

where $\beta(s)=L(s,\chi_{-1})$ is the Dirichlet beta function. All equivariant etas have the universal closed form:

$\eta_g(n) = \frac{(n+2)\sin((n+1)\alpha)-(n+1)\sin((n+2)\alpha)}{\sin\alpha}$

The Tier 2 classes each retain 2 L-functions in their closed-form equivariant eta expressions. $C8$ (order 3) involves $\zeta(s)$ directly:

$\eta([C8],s) = 2^{s-1}[L(s,\chi_3)-\zeta(s-1)]$

The 2-term count is a property of the closed-form structure. The number of Dirichlet characters with nonzero projection in the coprime sector $U(120)$ differs by class and is tabulated in Lemma 3.

E. Vacuum Structure

Per-irrep twisted etas computed via fusion matrices for all three flat connections (trivial, standard, Galois). Three distinct sign patterns: the Galois twist inverts the light/heavy assignment on the McKay graph. The antisymmetric combination $\eta_\text{std}-\eta_\text{gal}=(2/5)\times\mathrm{integer}$ for every irrep, with R7 in the kernel. Equivariant class selection is invariant under vacuum twist: $\eta_\rho([g])=\chi_\rho(g)\cdot\eta([g])$.

Computational Record

IV. Lemmas 1–8

The eight lemmas defeat specific strategies for leveraging the reading capacity of § III to constrain L-function zeros. Theorem 1 follows from their conjunction. Three supporting propositions are interspersed before the lemmas they enable: Propositions 1 and 2 precede Lemma 4; Proposition 3 precedes Lemma 7.

Lemma 1 (Curvature / Pochhammer Obstruction). Strategy defeated: Extend the torsion's selectivity (4 of 16 characters) from $s=0$ to a strip containing the critical strip.

Proof. The scalar Laplacian on $S^3/2I$ has eigenvalues $l(l+2)=(l+1)^2-1$. The Ricci curvature shift "$-1$" generates a Pochhammer expansion:

$Z_{0,\sigma}(s)=\sum_{k\geq0}\frac{(s)_k}{k!}\cdot\widetilde{H}_k(s)$

At $s=0$: $(0)_k=0$ for $k\geq1$ collapses the tower to a single term — the mechanism that produces maximum selectivity. At general $s$: $(s)_k\neq0$, the full infinite tower activates, and no finite closed form exists. Dowker's cancellation formula confirms $F(\gamma;0)=F'(\gamma;0)=0$ only at $s=0$.

No linear combination $a\cdot Z_0+b\cdot Z_1$ with $a\neq0$ factors at general $s$: the Pochhammer tower from $Z_0$ cannot be cancelled by the unshifted $Z_1$. The Lichnerowicz formula $D^2=\nabla^*\nabla+R_\text{scalar}/4$ connects Dirac eigenvalues to scalar Laplacian eigenvalues through an additive constant shift ($3/(2R^2)$). This does not recreate the Pochhammer obstruction for the Dirac operator — the Dirac spectral zeta does factor finitely at all $s$ (§ III.B). The obstruction here is specific to the scalar/torsion route: the curvature shift blocks the extension of torsion selectivity from $s=0$ to general $s$. The Dirac route avoids this obstruction but faces a different one: insufficient character selectivity (28–32 of 32 characters survive, making zero isolation impossible).

The torsion's selectivity is a property of $s=0$, not a property that extends. The curvature that gives the Yang-Mills mass gap blocks the general-$s$ factorization for the scalar route. $\square$

Lemma 2 (Coincidence is Not Vanishing). Strategy defeated: Use the C2 equivariant eta's single-L-function selectivity at all $s$ to constrain zeros of $\beta(s)$.

Proof. The zeros of $\eta([-e],s)$ satisfy $\beta(s)=\beta(s-2)$. By Definition 4, this is a coincidence condition: it requires two values of $\beta$ to be equal at shifted arguments, not any individual value to vanish. It is therefore not a zero constraint. $\square$

Remark (non-load-bearing, confirmatory only). Numerical evaluation at the first 10 nontrivial zeros of $\beta$ confirms $\lvert\beta(s_0-2)\rvert$ is far from zero in each case (values 13.98 to 341.84 at 50-digit precision). This is consistent with but not required by the proof above, which rests entirely on Definition 4.

Lemma 3 (Oscillatory Weighting Degrades Selectivity). Strategy defeated: Couple the SU(2) character weight $\chi_j(g)$ inside the spectral sum (Type 2 equivariant eta) to achieve deeper group-theoretic constraints on the canonical Dirac eta data.

Claim: For the canonical equivariant Dirac eta invariant of $S^3/2I$ — constructed from the actual Dirac spectrum without added spectral projectors or custom level filters — Type 2 weighting never reduces Dirichlet character support below the Type 1 baseline.

Proof (C2 case — structural). Both types reduce to the same linear combination form via the McKay decomposition identity $\chi_j(g)=\sum_\sigma m(\sigma,j)\chi_\sigma(g)$. The shared outer structure conceals a critical difference in the inner kernel: $\eta_D(\sigma,s)$ uses only the diagonal McKay kernel $K_{\sigma\sigma}(s)$, while $\widetilde{\eta}_{\sigma}(s)$ uses the full off-diagonal matrix $K_{\sigma\sigma^\prime}(s)$.

For the central element $C2=[-e]$, the SU(2) character evaluates to $\chi_j(-e)=-(2j+1)$ for all half-integer $j$ — monotone and strictly negative, with no sign variation. Type 1 achieves single-character selectivity at C2 precisely because the $2I$ character table produces sign alternation between bosonic irreps ($\chi_\sigma(C2)=+\dim\sigma$) and fermionic irreps ($\chi_\sigma(C2)=-\dim\sigma$). That alternation drives the cancellations that isolate $\beta(s)$.

Type 2 at C2 is structurally blind to this mechanism. The weights $(2j+1)^2$ are strictly positive for all $j$, introducing no sign alternation and no cancellation between bosonic and fermionic contributions. Type 2 is strictly worse than Type 1 at C2. $\square$

Proof (non-central classes). For each of the 7 non-central conjugacy classes $g\in2I$, the Type 2 weight on unit residues mod 120 takes the exact form $W_{120k+r}(g)=f_r(g)\cdot(32k+B_r)$, where $f_r(g)=\sin(r\alpha)/\sin(\alpha)$ is periodic with period $2q$, and since $2q$ divides 120 for each class ($2q\in\{6,8,10,12,20\}$), restriction to residues mod 120 captures complete periods. $B_r$ is the residue-class intercept determined by the McKay recurrence. The Dirichlet character projection decomposes into two independent sums:

$S_1(g,\chi) = \sum_{r\in U(120)} f_r(g)\cdot\bar{\chi}(r)$ $S_0(g,\chi) = \sum_{r\in U(120)} B_r\cdot f_r(g)\cdot\bar{\chi}(r)$

both finite and exact over the cyclotomic field $\mathbb{Q}(\zeta_{2q})$. A character survives when at least one sum is nonzero. Exact computation over the full 32-character Dirichlet basis on $U(120)\cong U(8)\times U(3)\times U(5)$ yields:

In all 7 cases Type 2 strictly exceeds Type 1. The mechanism is the two-component structure: the slope term $32\cdot S_1(g,\chi)$ preserves the Type 1 support pattern, while the nonconstant intercept term $S_0(g,\chi)$ activates additional characters not present in Type 1. Type 2 is strictly worse than Type 1 for all non-central classes. $\square$

Proposition 1 (Operator Classification). Every admissible operator $A$ (Definition 1) on sections of a natural bundle $E_\sigma$ over $S^3/2I$ has L-function character content determined by the McKay multiplicities $\{m(\sigma,l)\}$ alone. The eigenvalue function $a(l)$ may vary freely within Definition 1 without introducing characters beyond those identified by the $2I$ character table and the $E_8$ McKay correspondence.

Proof. Step 1 — Bundle classification. Every natural bundle over $S^3/2I=\mathrm{SU}(2)/2I$ is $E_\sigma=\mathrm{SU}(2)\times_{2I}V_\sigma$ for one of the 9 irreps of $2I$. Definition 1 is exhaustive at the bundle level.

Step 2 — Schur scalar reduction. By Peter-Weyl, sections decompose as $\Gamma(E_\sigma)=\bigoplus_l V_l^{m(\sigma,l)}$. Any admissible operator commutes with $\mathrm{SU}(2)$ and acts as a scalar $a(l)$ on each block. The spectral zeta $Z_A(s)=\sum_l m(\sigma,l)\cdot(2l+1)\cdot a(l)^{-s}$.

Step 3 — Characters fixed by multiplicities. Character orthogonality mod 120 extracts Dirichlet character content from $m(\sigma,l)$ alone. The function $a(l)$ is a Mellin-type weight; it shifts poles and domains but cannot introduce or remove characters. Different operators produce different spectral zetas with the same character set. $\square$

Proposition 2 (Combination Closure). Every construction $F\in\mathcal{F}$ built from Definition 1 operators by the operations of Definition 2 has L-function character content contained within the finite character algebra determined by the $2I$ group data. No genuinely new arithmetic source enters from finite additive or multiplicative combination.

Proof. Every spectral invariant built from a Definition 1 operator decomposes into the algebra generated by character-indexed L-expressions: finite linear combinations of terms $\chi(n)\cdot f(n)$ where $\chi$ ranges over Dirichlet characters mod 120 and $f(n)$ encodes the eigenvalue structure. Under finite addition, support on a character set $S$ remains on $S$. Under finite multiplication, products may mix within the ambient character algebra mod 120, but cannot introduce characters from outside that algebra — no new arithmetic source enters beyond what the $2I$ group data and McKay multiplicities already determine. Vacuum twists leave equivariant class selection invariant: $\eta_\rho([g])=\chi_\rho(g)\cdot\eta([g])$, confirming no new character support enters through twisting. The finite character algebra over $U(120)$ is the ceiling. $\square$

Lemma 4 (Operator Completeness Inside Admissible Class). Strategy defeated: Find a different natural operator on $S^3/2I$ that avoids the obstructions of Lemmas 1–3.

Proof. By Propositions 1 and 2, every $F\in\mathcal{F}$ has character content within the set identified by the McKay multiplicities of the 9 irreps of $2I$. The following exhaust the candidates within $\mathcal{F}$:

• Hodge duality: $p$-form Laplacians for $p=2,3$ are isospectral with $p=1,0$. Only two independent spectral zetas exist, both already computed.
• Signature operator: No self-dual/anti-self-dual decomposition on 3-manifolds. Reduces to the known Dirac eta.
• Vacuum twists: Three flat connections produce three distinct per-irrep eta vectors, but equivariant class-level selection is invariant. C9 is killed by both nontrivial twists; C7 is perfectly vacuum-invariant.
• Combined invariants: Products $T^2(\sigma)\cdot\eta(\sigma,0)$ inherit factor arithmetic. No emergent L-basis from combining the two spectral invariants.

No unexplored operator remains within $\mathcal{F}$. $\square$

Lemma 5 (Finite-Product Collapse). Strategy defeated: Use multiplicative (Selberg/Ruelle) encoding of spectral data, exploiting positivity properties that additive encoding lacks.

Proof. The Ruelle zeta on $S^3/2I$ is a finite product over 3 geometric-primitive conjugacy classes (C3 order 10, C7 order 6, C9 order 4). The zeros are roots of unity (holonomy phases), not Laplacian eigenvalues. No functional equation of Selberg type exists. Integer-spin twists give $\det(I-\sigma(\gamma))=0$. The passage from sums to products is invertible for finite products and introduces no new constraints.

The Selberg zeta machinery requires Anosov (hyperbolic) dynamics. The round $S^3$ has periodic (Zoll) geodesic flow. Product-side zeros are holonomy/cyclotomic zeros determined by group elements, not by spectral parameters or L-function arguments. $\square$

Lemma 6 (Framework Disconnect). Strategy defeated: Connect spectral L-functions to the icosahedral Artin L-function via Langlands/modularity, using known automorphicity (Khare-Wintenberger) to constrain the spectral side.

Proof. The spectral L-functions are Dirichlet L-functions over $\mathbb{Q}$, arising from Fourier analysis of McKay multiplicities (modulus 120). The Artin L-function $L(s,R_1)$ is a 2-dimensional L-function attached to a Galois representation, decomposing via Brauer induction into Hecke L-functions over intermediate number fields.

Brauer decomposition: $R_1=\mathrm{Ind}(C_6,\lambda_6)+\mathrm{Ind}(C_{10},\lambda_{10})-\mathrm{Ind}(C_4,\lambda_4)$. This is virtual (negative exponent), not isolable. Known icosahedral extensions have discriminants far larger than 120, making conductor overlap numerically impossible. Shared prime support $\{2,3,5\}$ is forced by $\lvert 2I\rvert=120$ and carries no analytic content.

The spectral and arithmetic L-functions attached to $2I$ are different mathematical objects. Modularity of the Artin L-function does not constrain the spectral Dirichlet L-functions. $\square$

Proposition 3 (Kernel Reduction). Let $\sigma$ be any irrep of $2I$ and $A$ any admissible operator on sections of $E_\sigma$ over $S^3/2I$. Then: (1) $K_t^A(x,x)=\sum_{\gamma\in2I}\sigma(\gamma)\cdot k_t(d(x,\gamma x))$ with $k_t$ scalar; (2) every spectral invariant reduces to $\sum_{[g]\subset2I}\chi_\sigma(g)\cdot G_t([g])$; (3) no basis-independent content of the kernel lies outside the character table of $2I$.

Proof. The right-$\mathrm{SU}(2)$ action commutes with any natural operator, so the heat kernel depends only on geodesic distance. Combined with left-$2I$ equivariance:

$\tilde{K}_t^A(x,\gamma x) = \sigma(\gamma)\cdot k_t(d(x,\gamma x))$

Every spectral invariant is defined via the integrated diagonal trace:

$\int_M\mathrm{Tr}[K_t^A(x,x)]\,dx = \sum_{\gamma\in2I}\chi_\sigma(\gamma)\cdot G_t(\gamma)$

The matrix $\sigma(\gamma)$ collapses to the scalar $\chi_\sigma(\gamma)$ under the trace. Any basis-independent function of the residual matrix is a symmetric function of eigenvalues, expressible as characters at powers of $\gamma$, which land in known conjugacy classes. The character table is exhaustive. $\square$

Lemma 7 (Character Ceiling). Strategy defeated: Go beyond character traces to matrix coefficients, exploiting intertwining structure that the trace discards.

Proof. Three candidates:

• $2I$ representation matrices $\sigma(g)$: Factor out of spectral sums as constants. $M_{\sigma,[g]}(s)=\sigma(g)\cdot\eta_D(\sigma,s)$. No new information beyond characters.
• SU(2) Wigner $D$-matrices $D^j(g)$: Diagonal for conjugacy class representatives: $D^j(g)=\mathrm{diag}(e^{2im\alpha})$. Off-diagonal entries are basis-dependent. $\det D^j(g)=1$ identically. Any basis-independent function is expressible as characters at powers of $g$ (Proposition 3).
• Operator kernel: By Proposition 3, the kernel reduces to character data. Redundant given the character ceiling.

The character table is the complete basis-independent invariant of the group action on $S^3$. $\square$

Lemma 8 ($\Theta\leftrightarrow s$ Bridge). Strategy defeated: Construct a natural map between the phase position $\Theta$ (from the scaling law $C(\Theta)=2\sin^2(\pi\Theta)$) and the spectral parameter $s$, exploiting their shared $\mathbb{Z}_2$ symmetry.

Proof. Four independent approaches on $S^1$ (heat kernel, theta function, Poisson summation, direct decomposition) prove no such map exists. Every eigenspace on $S^1$ with anti-periodic BC is 2-dimensional (sin and cos). $C(\Theta)$ depends on choosing $\sin$ over $\cos$. The spectral zeta sees only eigenvalues and multiplicities, blind to this choice. The two $\mathbb{Z}_2$ symmetries — $C(\Theta)=C(1-\Theta)$ from spatial reflection; $\xi(s)=\xi(1-s)$ from Poisson/modular duality — arise from different mechanisms and are not related by a natural map.

On $S^3/2I$: the McKay correspondence resolves the basis ambiguity (multiplicities are canonical), but continuous geometric position drops out by Schur's lemma: the right-$\mathrm{SU}(2)$ acts transitively, forcing the twisted heat kernel to be constant on the diagonal of each fiber. $\square$

V. The Wall

The eight lemmas organize into four independent obstruction layers. Each layer would suffice alone to block a spectral Hilbert-Pólya argument on $S^3/2I$.

Proof of Theorem 1. Every $F\in\mathcal{F}$ falls under at least one obstruction layer. Lemma 1 handles the torsion selectivity strategy. Lemma 2 handles the C2 single-character strategy and establishes the coincidence/vanishing distinction of Definition 4. Lemma 3 handles character-weight oscillation; the C2 case is proved structurally, and the 7 non-central classes are proved by exact mod-120 character decomposition. Together, Lemmas 1–3 establish the coincidence condition layer. Lemma 4 (via Propositions 1 and 2) confirms the operator space is exhausted (character completeness layer). Lemma 5 handles multiplicative encoding (encoding degeneracy layer). Lemma 6 handles the Langlands/modularity route (framework mismatch layer). Lemmas 7–8 (via Proposition 3) handle matrix coefficients and the phase-to-spectral bridge (character completeness layer). Every vanishing condition on every $F\in\mathcal{F}$ reduces to at least one of (i)–(iv). $\square$

VI. Mechanism: The Curvature Duality

The proof of Theorem 1 is complete at § V. Sections VI–IX provide the structural mechanism behind the obstruction, its interpretation, scope, and physics implications.

The wall is structural, not technical. A unifying geometric mechanism is captured by a single equation.

The Ricci curvature of $S^3$ is $\mathrm{Ric}=2/R^2>0$. This positive number does two things simultaneously:

Physics (Weitzenböck bound): All gauge fluctuations on $S^3/2I$ satisfy $\lambda\geq2/R^2>0$. The mass gap exists. Every mode is massive. Matter is realized.

Arithmetic (Pochhammer obstruction): The scalar Laplacian eigenvalues shift from $(l+1)^2$ to $l(l+2)=(l+1)^2-1$. The "$-1$" generates the infinite Pochhammer tower that blocks general-$s$ factorization. The torsion's maximum selectivity is locked to $s=0$. L-function zeros are shielded.

To remove the obstruction, set $\mathrm{Ric}=0$. The eigenvalues become perfect squares. The Pochhammer tower vanishes. The spectral zeta factors at all $s$. But the mass gap also vanishes. Flat space. No particles. Nothing to observe.

Mass and spectral access to zeros are in structural opposition. The curvature that realizes one forbids the other.

This is the structural interpretation of Theorem 1, not an extension of its scope. The theorem proves inaccessibility within the admissible class $\mathcal{F}$ in the sense of Definition 4. The curvature duality is the geometric explanation for why all eight obstructions trace to the same source.

VII. Discussion

The following sections interpret Theorem 1 within the structure of $S^3/2I$ and the Mode Identity Theory (MIT) framework. They do not extend the theorem's scope beyond the admissible class $\mathcal{F}$ and the sense of Definition 4.

Hurwitz's theorem establishes that $\varphi=(1+\sqrt{5})/2$ is the hardest irrational to approximate by rationals: its continued fraction $[1;1,1,1,\ldots]$ converges as slowly as possible. In the MIT framework, this maximal irrationality stabilizes the matter wells — if $\varphi$ were rational, the sampling positions would degenerate and the wells would vanish.

The Riemann zeros encode the distribution of primes. The primes build the integers. The integers build the grid ($120=2^3\times3\times5$). The grid builds the domain. The domain builds the wells. $S^3/2I$ gives you $\varphi$ through the character field $\mathbb{Q}(\sqrt{5})$ and gives you the L-function structure through the McKay correspondence with $E_8$. The manifold does not choose between them. Its geometry requires both.

Both $\varphi$ and the zeros are stability results: the most irrational number cannot be rationalized, which is what makes it useful for positioning matter; the zeros cannot be spectrally accessed from this manifold, which is consistent with their role as the foundation of arithmetic structure. Whether zero inaccessibility is itself a stability requirement — a theorem rather than an observation — is an open question not addressed by the proof above.

Lemma 8 has a direct consequence for the MIT mass formula. The non-existence of a natural map between the phase position $\Theta$ and the spectral parameter $s$ means the fine structure of the mass formula cannot be completed by extending the spectral arithmetic. The bridge is not merely unbuilt — it is proved not to exist. The fine structure within each mass shell is therefore forced to come from representation-theoretic data directly: graph distance, Kostant polynomial, vacuum selection. This is not a limitation of the framework; it is a constraint the geometry itself imposes.

The Mirror

$S^3/2I$ is a perfect arithmetic mirror. It reflects:

• The primes $\{2,3,5\}$ dividing $\lvert 2I\rvert=120$
• The golden ratio $\varphi$ from the character field $\mathbb{Q}(\sqrt{5})$
• The L-function special values through torsion (79-digit precision)
• The Dirichlet characters through spectral decomposition
• Individual L-functions through the equivariant eta (C2: single beta function at all $s$)
• The Riemann zeta function itself through C8: $\eta([C8],s)=2^{s-1}[L(s,\chi_3)-\zeta(s-1)]$

VIII. Scope

Theorem 1 (local to $S^3/2I$): For intrinsic natural operators in the admissible class $\mathcal{F}$ on $S^3/2I$, spectral vanishing does not constrain individual L-function zeros in the sense of Definition 4.

Corollary 1 (local to $S^3/2I$): Within the admissible class $\mathcal{F}$ on $S^3/2I$, the four obstruction layers — coincidence condition, encoding degeneracy, framework mismatch, and character completeness — are exhaustive: every vanishing condition reduces to at least one of them.

Reminder on scope. The impossibility claim is in the sense of Definition 4: a construction $F$ constrains zeros only if $F(s_0)=0$ implies $L(s_0,\chi)=0$. Shifted-equality relations and cross-function coincidences do not qualify. The theorem is narrower than a casual reading of "spectral inaccessibility" might suggest, and is not a claim about all possible approaches to L-function zeros.

Remark. The curvature duality mechanism of § VI depends only on the positivity of the Ricci curvature and the resulting eigenvalue shift, not on the specific group $2I$. This suggests the inaccessibility result extends beyond $S^3/2I$.

Conjecture 1. Positive Ricci curvature with finite $\pi_1$ imposes zero-inaccessibility in the sense of Definition 4 on any compact Riemannian manifold whose spectral zeta factors into Dirichlet L-functions. $S^3/2I$ is the extremal case: $2I$ is the largest exceptional discrete subgroup of $\mathrm{SU}(2)$, $E_8$ is the largest exceptional Lie algebra, and $\lvert 2I\rvert=120$ captures the maximum arithmetic structure through the McKay correspondence.

IX. Physics Application

The spectral inaccessibility theorem is a negative result for the RH direction. It is a positive result for physics. The same L-function structure that cannot constrain zeros can and does predict physical observables.

The manifold is a perfect mirror of arithmetic, but not a tool for proving the Riemann hypothesis. Mirrors show you everything. They determine nothing. That's not a failure, it's the theorem.