Every fermion mass is the same ladder: the geometric VEV times the carrier base raised to a compiler word-length, times an O(1) residue. The word-lengths are the fixed residue matrix of the compiler — not free parameters.

The word-length matrix L = R + 6·(winding) carries only compiler numbers: its trace is N_fam², its determinant is h(D₅) = 8, its principal 2-minors are (2,3,5) with product 30 = h(E₈), and its Frobenius norm is dim E₆ = 78.

The lepton amplitudes are the rationals (16/7, 4/3, 7/6) with product 2⁵/N_fam² = 32/9, and the masses are exact φ₀-powers. Applied to the down sector the lepton law provably fails — the quark c's live on the parabolic wall.

The quark mass ratios are pure integer Plücker readouts on the derived selector stratum — no transcendental solve. The absolute amplitude reduces to one overall scale v_geo (ratios + Grand Mass Volume), the same dimensionful anchor as gravity's 1/G. The remaining ℤ₃ deck choice is since derived: the integer deck pairs the Q₊=1 line with the self-conjugate character 2, so the geometric boundary deck is the sheet-twisted class and the cusp exponential is excluded (v141) — GATE.QGEO keeps only its realisation premise, with no discrete freedom left — and that premise sits at its floor: the full Möbius D₄ of the seam curve matches the integer model parity by parity (ι = T_A exactly, δι = Σ; v146). The finite rigidity is now proven exactly [E]: μ₄ has cross-ratio 2 and a faithful Möbius D₄ stabiliser, H¹(ℙ¹∖μ₄) has rank N_fam = 3, and the eigenforms ω_k carry the μ₄ characters of weights (1,2,3) = the A₃ exponents = Spec(Q₊) — so only the seam-collar realisation stays open (v168). The Sheet Diamond carrying these operators is a discrete geometry with two axes (v218): the determinant is linear along the winding axis (A₃-driven, slope 6 = |R⁺(A₃)|) and quadratic along the sheet axis with curvatures (8, 6) = (rank E₈, |R⁺(A₃)|); the anchor-Plücker coordinates lift K→C→F in two exact steps (1,8,10) then (1,8,16) — the decuple A_Λ then the full spinor generation dim S⁺; and the characteristic-polynomial discriminants of Q,K,C,F factor as q(r)²·Disc(q) with squares (1,3,4,6) = (N_Φ,N_fam,|μ₄|,|R⁺(A₃)|) and kernels (13,48,65,105). No new numbers — it organises the existing operators more strictly. Sharper still (v410): the sheet axis V = Q·diag(0,1,1) is a binary internal compiler — its powers print the carrier spine Vⁿ·1 = (2ⁿ⁻¹, 2ⁿ, 2ⁿ⁺¹−1) = (1,2,3),(2,4,7),(4,8,15),(8,16,31), and four exact bilinear families collapse the recurring integers (6,7,9,11; 13 = Δ_Q, 27 = 1ᵀRa, 55, 56 = dim 56_E₇) into one operator's iteration. The quark ratio is then a pure V-power readout c_u/c_d = (1ᵀV⁴1)/((aᵀV1)(1ᵀV²1)) = 55/117 (v411, an exact re-encoding); the unnamed Z₂-wall corner J = M(1,−2) carries χ_J = (6,3,2), aᵀJa = 30 = h(E₈), det(I+J) = 12, det(2I+J) = 40 (v412); the sheet axis encodes the atoms as difference orders Δe₁ = 3, Δ²e₂ = 4, Δ²e₃ = 8 with anchor energy 52+11t (v413); and the center C is a resolvent portal det C = 14 = dim G₂, det(I+C) = 52 = dim F₄, det(2I+C) = 120 = |R⁺(E₈)| (v414). All [E] algebra; the binary spine is forced by Spec(V) = {0,1,2}, so the Lie-dimension readings stay [C], audit-typed.

The absolute ν-mass scale is one seesaw ratio m₃ = (y_ν v)²/(2M_R) — honestly typed as one open UV input (v272), with the NO floor Σm_ν = 0.0586 eV as the cosmological kill test. New (v481, FLAV.NUSCALE.02, CANDIDATE class like v467/v468): the y_ν = 1 probe was not the carrier normalisation — one SO(10) 16 per family with the minimal Yukawa sector (10_H / PS (1,2,2)) forces y_ν = y_t at the matching scale, collapsing the (y_ν, M_R) trade-off to M_R alone. With explicit 1-loop RG (gauge/y_t/λ up, ADKLR Weinberg-operator running down) the observed m₃ = 0.0503 eV demands M_R = 9.3×10¹³ GeV — inside the compiler's own PS window [4.2×10¹³, 2.4×10¹⁵] GeV (v249) at log_c₃(M̄/M_R) = 3.15 (y_ν = 1 gives the structureless 2.58). Honesty gate: the integer rung M_R = c₃³M̄ predicts m₃ = 0.030 eV, 40% low at 1-loop, so the ladder pin is DECLINED per the anti-numerology rule — and the named decision computation is meanwhile EXECUTED in bracketed form (v482, FLAV.NUSCALE.03): the rung needs a rescue factor ×1.670 while >3σ-generous input envelopes (m_t ±3 GeV, α_s, κ-run ±10%, PS-leg β ×[0.5,1.5]) reach at most ×1.165 combined, so the unstructured rung is EXCLUDED (not an RG artifact); the only escape, a third-generation Majorana structure factor r ≈ 1.67 sitting 0.18% from g_car/N_fam = 5/3, is recorded post-hoc and declined (no forcing mechanism). That escape is now DECIDED dead (v488, FLAV.NUSCALE.04): since the 126bar is not in the E₈ hull (v247), M_R can only come from the d=5 operator (16·16bar_H)²/Λ with singlet/45 insertion channels; every ν^c channel weight {1, 1/4, 1, −1/2, 3/8} is a {2,3}-smooth rational, so no channel combination can produce 5/3 — and the unique natural 5 of the embedding, k_Y = 5/3 (Ginsparg), is a full-multiplet trace whose direction has Y(ν^c) = 0 exactly, structurally decoupled from the Majorana operator; Clebsches are generation-blind (family-space scalars), so diag(1,1,3/5) cannot arise from group theory at all. A clean negative: the rung+5/3 rescue is a numerical coincidence without mechanism, and the one-parameter window candidate stands as the honest endpoint. The candidate band m₃ ∈ [0.002, 0.115] eV brackets the observation and DESI cuts the window from below; nothing closes and the frozen record is untouched.

Tri-bimaximal gives 1/3; the charged-lepton 1–2 misalignment is the seam ε = q(A₃)φ₀ = (3/4)φ₀, and TBM geometry gives the only previously open SM angle as a conditional derivation — 0.1% from NuFIT 6.0.

At the two branch points of the anchor-block double cover the block is rank 1, with integer kernels — at the carrier point the kernel is the democratic vector itself. Rank 1 forces the kernel image onto the antisymmetric direction (−1,1,0): up and down are the deck-odd pair, and the lepton pairing vanishes — the leptons sit on the ramification (Koide is leptonic). The anchor-forced cusp conjugation T_A (with a = e₂+e₃, the conjugation-symmetric vector) realises the same deck action, and the dictionary 'Q₊ grading = A₃ discriminant grading' is now derived (G = T_A·Σ acts integrally as the B₁⊕E decomposition on the cusp basis): the sheet question carries no separate [C] — its residual coincides with the one existing Q-geometry gate.