The clearest result of the four-column matrix is that no constant deforms all domains structurally. α is active in geoid (proxy), Na, He — but inactive in H, Li, DESI. μ is active in H₂ — but inactive in CO, N₂, HCl. αₛ is inactive everywhere tested (nuclear + hadronic). αG is inactive in all three geoid bodies. The pattern is not that constants are weak — it is that they couple selectively. The coupling mechanism determines which systems respond, and that mechanism is different for each constant.

Both α (Na gaps, proxy) and μ (H₂ gaps, Tier A) produce a structural extremum at the physical value of the constant: α=1.00 is a pressure maximum for Na; μ=1.00 is a pressure maximum for H₂. In both cases β*=1.00 is not a minimum — it is a maximum, placing these systems at peak structural pressure under the actual laws of physics. The geoid α=1.00 result (TYPE III-Min, pressure minimum) is refuted by Column IV, leaving two confirmed TYPE III-Max cases at the physical constant value. Whether this is a generic principle or coincidence requires further investigation.

⁴⁸Ca (doubly-magic, Z=20, N=28) is the starkest constant-selectivity example in the corpus. Under α it is TYPE III-Fr with 17/17 marginal violations — the highest nuclear α-activity. Under αₛ it is TYPE I with Δρ̄=0.0051 and zero signal. The same physical gap geometry responds to one constant and ignores the other completely. This means the gap structure of ⁴⁸Ca contains information that is selectively legible to α-deformation but not to αₛ-deformation. The mechanism is not yet understood — it is an empirical result that the two coupling channels produce qualitatively different structural responses from the same nucleus.

²⁰⁸Pb (doubly-magic, Z=82, N=126) returns Aκ=1.0000 exact at every tested γ-value under both α (17 values) and αₛ (17 values). This is the deepest TYPE I·CALM result in the corpus — not merely no signal, but perfect admissibility at every deformation point across two independent constants. The doubly-magic character at maximum shell closure produces a qualitatively distinct inertness from ⁴⁸Ca. The contrast between ⁴⁸Ca (constant-selective excursion) and ²⁰⁸Pb (universal calm) within the same doubly-magic class is the nuclear corpus's primary structural finding.

Column I (α-proxy) classified Earth, Mars, and Moon as STRONG · TYPE III-Min with β*=1.00. All three Tier A αG runs — using physically grounded C₂₀^rot decompositions locked from published geodesy sources — confirm TYPE I across the full γ∈[0.80,1.20] sweep with Δρ̄≤0.0010 and no distinguished γ-point. The AG_DEFORMATION_PROTOCOL_v2 §10 prediction is fully verified: the true γ* for every body lies below the sweep floor (Earth γ*≈0.71, Mars γ*≈0.54, Moon γ*≈0.35). No extremum can appear within [0.80,1.20] for any body — this follows necessarily from the f_rot fractions, which range from 11% to 33.5%. The Tier A runs also produce a genuine structural reclassification for Mars: its state changes from Boundary-Stabilized (proxy run) to Weak Persistence (Tier A) — a direct consequence of applying the correct G-coupling rule rather than the α^l proxy. Column IV is fully MATRIX-GRADE across all three bodies.

Across all four constants, seven domain types, 26+ physical systems, and 1,500+ ladder assessments, the admissibility inequality inv(Pε;L) ≤ ν(Vε(L)) is not falsified. The earlier coarse H₂ minimum Aκ=0.9925 at β=1.00 is a sampling artifact — the targeted 5-point refinement confirms Aκ=1.0000 at β=1.00. An approximate-decomposition 5-point HD run appeared to show hard violations (Aκ=0.517) but is superseded by the inferred Tier A 17-point sweep returning Aκ=1.0000 throughout — those violations were a decomposition artifact. With the correct Tier A ladder, HD is TYPE I · Stable Structure. The inequality holds universally at all physical and tested non-physical constant values in the molecular domain.

HD (hydrogen deuteride, n=2000, HITRAN) returns completely different structural profiles depending on how vibrational and rotational components are separated. The 5-point approximate-decomposition sweep produced Transitional Structure, max ρ>1.0, and apparent violations — artifacts of an imprecise splitting. The full 17-point inferred Tier A sweep, using band inference (ranked energy manifolds → inferred vibrational bands → ν_vib/ν_rot separation), returns Stable Structure and Aκ=1.0000 throughout. This demonstrates that ladder classification in the molecular domain is acutely sensitive to decomposition quality. An approximate decomposition can produce spurious violations; only Tier A decomposition is admissible for classification and matrix insertion. HD classification: TYPE I. One structural curiosity survives: at β=1.00, n_unique_gaps=4924 vs ~4962 at all other β — approximately 38 gap degeneracies at the physical mass ratio, a geometric feature not visible in the STRUC-I summary statistics.