# Research directions The [v82 continuation](results_20261120_v82.md) proves general quartics in **163 and 179 total variables**, retaining the source arithmetic premises. The partial-trace scalar and a scaled point chart remove two field components; a positive weighted trace removes another rational coordinate from the ineffective family. The ordered global witness construction, zero-coordinate branches, full quadratic maps, and coefficient expansions are supplied. The fixed 179-variable member retains an effective unweighted height argument. Contributions originating in **notes of Gabe Conant with Astra 6.2** remain distinguished from the subsequent project work. The current unrestricted-degree counts remain **25** and **11**, and the ordinary-abc quartic remains **56**. No reduction of those counts is proved in v82. Every count includes the input. The priorities remain variable count, degree, then coefficient height; no global minimum or complete independent audit of all inherited arithmetic foundations is asserted. The [v83 exact-square checkpoint](../computation/actual_square_v83/README.md) constructs a new balanced normalization with `1+d` a rational square at the raw contact scalar. Its actual pole input is `e=4d/V²`, so the required condition is still `V²+4d` being a rational square. Exact prepared counterexamples exclude the corresponding universal genus-one and zero-parameter replacements; none is asserted to occur at the actual elliptic trace. No defining-variable count changes. The [additional research routes](research_routes_20261120.md) prove a complex-multiplication local comparison with three rational point charts and a different one-parameter sound cover. The former still needs an ordered contact construction controlling all auxiliary indices; the latter still needs actual rational realization. Neither is recorded as a new defining-variable bound. ## Immediate constructive targets 1. **Identify an external input with the new elliptic quotient.** The [prepared divisibility theorem](../papers/elliptic_divisibility/prepared_eds_divisibility_20261120_v82.tex) forces `h | j` with two rational elliptic point charts and one pole block. Every nonzero integral quotient `j/h` occurs after choosing a suitable reference. Its five-coordinate specification does not yet enforce `X=j/h`. A height argument must handle an arbitrarily large quotient; a carrier depending only on the reference leaves this uncontrolled, while including the second point can demand more contact-prime support than the reference provides. 2. **Prove rational realization at actual detector inputs.** The [genus-one criterion](../papers/pole_parameters/genus_one_sound_criterion_20261120_v82.tex) replaces the two pole parameters by one if the actual fibers have non-torsion rational points. Local solubility is now proved at every place for prepared scalar inputs. An exact prepared counterexample, including a broad shifted-abscissa menu, shows why a global argument using the actual contact and point functions is necessary. The counterexample is not asserted to be an actual detector value. The proposed condition that `1+d` be a rational square remains unproved at the actual tuple. It suffices to find actual witnesses on a single quadratic progression `cm²+b` for fixed positive squarefree `c`, fixed integer `b`, and one fixed admissible source constant, apart from finitely many integers. The new corollary proves that this restricted realization would still define every integer after substitution and finite repairs. No such progression has yet been realized. 3. **Remove the remaining general-quartic square coordinate.** The relation `tau_hat*sigma=R^4` still uses the shared `R²` register. A different scaling must keep the reference-point equations and every scalar/native-pole equation quadratic, with all zero fibers checked. The ordered witness construction is now available and need not be conjectured. 4. **Reduce the ordinary-abc quartic below 56.** A further paired-contact rechart or sharing with the point equations must remove a coordinate without adding a new cubic product. The secondary Euler identity exposes such a product but does not yet eliminate it. The [common-level scope note](../computation/variable_reductions_v82/abc/common_level_scope.md) rules out using one rank-one level to realize two distinct positive nonsquare input classes with the same prepared 2-adic class. It does not rule out a common conic parameter with different elliptic levels. A quaternion substitute currently detects only inert primes, whereas soundness needs arbitrary odd contact primes. The [v81 continuation](results_20261119_v81.md) remains the proof of the 56-variable ordinary-abc quartic and the separate ten-coordinate rational-subfield selector. The latter does not lower a variable bound for the integers over the rationals. ## Consequences of the audit of Gabe Conant’s notes with Astra 6.2 The shared norm selector and stronger odd-power packing provided the preceding 165/180 constructions and stronger local divisibility; v82 gives the further reductions described above. The six- and eight-contact weak-abc estimates give explicit variable/hypothesis tradeoffs; the eight-contact quartic family covers every fixed exponent below 3/2. Positive detector factors, square-class normalization of the intrinsic pole, and ordered removal of automatic point guards substantially lower the finite-algebra degree. These are proved changes to complete constructions. Further general variable reduction still requires a new arithmetic argument or a smaller guarded quadratic compilation. The [full-pair parity obstruction](../papers/quartic/hyperbolic_parity_20261119_v79.tex) proves failure only within its specified retained chart and termwise integral Laurent model; it does not establish optimality. The general 21 and ordinary-abc 9 targets continue to require rational realization at actual physical traces. The new diagonal Hasse and zero-ordinate obstructions concern specific universal specializations on all prepared inputs and cannot be promoted to nonexistence of 21/9 constructions. The older degree bounds cited below identify their original source architectures. Current complete bounds and all parameter qualifications are in the [main table](../README.md) and [v82 records](../data/results_20261120_v82.json). ## Rank-one curves over general number fields The [Pasten–Vidaux audit and amplified descent](../papers/number_fields/amplified_rationality_descent_20261119_v81.tex) provide a new conditional rational-subfield definition under abc over K and positive elliptic rank stable from Q to K. It handles all rational targets, both signs of odd valuation, and a fixed six-point divisor of a cubic rational map. This architecture uses the global conjugate difference t−σ(t), so it avoids the adjacent-contact obstruction below. It does not distinguish integers among rational inputs. Zywina supplies the needed stable curve for quadratic extensions; no such blanket conclusion is claimed for every higher-degree field. The [effective sum-height theorem](../papers/effective_sums/elliptic_sum_heights_over_number_fields_20261117_v63.tex) now works for every elliptic curve over every number field. Together with Zywina's rank-one existence theorem it supplies whole-domain height control on suitable curves, without retaining the original complex-multiplication model. The remaining tasks are to construct the required local integrality detector over the chosen field, prove simultaneous rational realization on that field, and encode the complete witness set. The height theorem does not identify field traces with algebraic sums and does not establish a Diophantine definition of the ring of integers. The [rank-stable contact obstruction](../negative_results/papers/number_fields/rank_stable_adjacent_contacts_20261117_v63.tex) excludes a direct use of a curve defined over the base field whose positive rank remains stable in the detector extension. Its adjacent contacts can occur at only finitely many primes. A new local architecture, or a curve defined over the extension without the stated descent property, is required. The [power-shift criterion](../papers/number_fields/power_shifts_and_finite_exceptions_20261117_v69.tex) needs a detector containing every sufficiently large positive rational integer and having bounded-height nonintegral inputs. A fixed square shift removes the finite errors without another witness coordinate. An integral basis and a field norm then give the whole ring in `dN` total coordinates and degree at most `2dD`, for field degree `d` and an original polynomial of degree `D` in `N` coordinates. The shift can satisfy prescribed congruences even over complex fields. Constructing such a detector remains the arithmetic task. The same height theorem also controls reciprocal abscissas on $E:y^2=x^3-Dx$, $D\ne0$. If $Q_i=[n_i]P+T_i$ are finite points in one cyclic group modulo torsion with $x(Q_i)\ne0$, and $M=\max(1,|n_i|)$, then effectively $$ h\!\left(\sum_i\frac1{x(Q_i)}\right) \ge \max_i h(x(Q_i))-C\log(M+2)-C'. $$ Indeed, $T_0=(0,0)$ is rational two-torsion and $x(Q+T_0)=-D/x(Q)$. The translated points remain in the same cyclic group modulo torsion and are finite under the stated nonzero-abscissa condition. Applying the sum-height theorem to them, then using $h(1/x)=h(x)$ and $|h(Dx)-h(x)|\le h(D)$ twice, gives the inequality. The constants depend only on the fixed curve, field, generator and number of summands. This is a direct corollary useful for reciprocal charts; it neither controls field-trace cancellation nor supplies detector completeness. ## Earlier quartic chart foundations The earlier **194-variable general** and **68-variable ordinary-abc** constructions provide chart foundations for the current records. The [translated-secant chart](../papers/quartic_existence/quartic_abc68_20261117_v65.tex) includes both reciprocal abscissas in six field coordinates, with four field quadrics. Its difference abscissa is a quadratic output; only its rational half-trace is stored. This preserves the prescribed difference carrier and the full multiplication-witness family, including tangencies, zero ordinates and every finite two-torsion difference. The imaginary output component has no other consumer and is eliminated exactly. The [194-coordinate general quartic](../papers/quartic/quartic194_20261117_v73.tex) combines three translated point charts with four polynomial cross comparisons, a cleared input numerator and a homogeneous Lucas circuit. All 168 quadratic rows and both complete coefficient expansions are independently reconstructed. A refinement of the reference progression synchronizes the two formal depths for integer witnesses. Arbitrary odd-depth extraction, the unit boundaries and all allowed zero-ordinate cases have separate proofs. The remaining quartic targets are the scalar primary ratio, the product helpers, and the transfer of this construction to the three-detector explicit source. The two excluded diagonal scaling families concern different specified factor interfaces; they impose no lower bound on these new polynomial coordinate changes. Every proposed deletion must include a complete quadratic compilation, every required guard, and a soundness and integer-completeness proof. For the ordinary-abc quartic, the [direct-elimination audit](audits/quartic68_direct_elimination_review_20261117_v65.md) explains why immediate substitutions for the two field product helpers and rational half-trace output raise the degree. Further changes of coordinates or combinations of equations remain possible. The zero-input witness and the unit proof at nonzero input must both be retained. ## From eleven toward nine The [complete derivative detector](../papers/conditional_few_variables/derivative_detector11_20261117_v63.tex) combines both heights in one signed contact argument and proves ordinary-abc completeness with one intrinsic pole block. The positive odd valuations still extract the original elliptic tests; negative odd derivative valuations are automatically harmless. The [full auxiliary-point construction](../papers/conditional_few_variables/full_auxiliary11_20261117_v71.tex) retains the root in `2E(K)` and permits the auxiliary points in all guarded affine `E(K)`. Its complete count is eleven, with degree at most 232,094,337, below $2^{28}$ and every guard included. Its cubic auxiliary charts also enlarge the sound nine-coordinate actual-trace source without changing that source's count. Rational-point arithmetic should therefore allow half-integral auxiliary indices, while retaining the integral root index and all exact contact-lifting tests. The [single actual-trace criterion](../papers/pole_parameters/single_trace_derivative_criterion_20261117_v64.tex) now verifies the complete transfer from eleven to a sound nine-coordinate source. Its sole extra equation is $Y^2=V^2+w(2-V)$, at the actual derivative trace and the existing slope. All denominator, character and terminal-root conditions are proved. The unresolved task is rational square realization for every sufficiently large positive integer at one fixed sufficiently large defining shift. The source sign involution and finitely many input factors then complete all integers without adding variables. Local squares at prepared places and selected contacts are insufficient; new even-valuation places and global realization remain uncontrolled. Another possible source uses fewer point roles, but must replace the original index-recovery argument. The [Möbius coupling obstruction](../negative_results/papers/pole_parameters/six_contact_mobius_obstruction_20261117_v63.tex) excludes the full twelve-symmetry menu for every fixed integer exponent at least four and every fixed finite exponent menu in the displayed six-contact architecture. Varying actual traces on a suitable incidence cover remains a different target. The [cofinite coupling obstruction](../negative_results/papers/pole_parameters/cofinite_mobius_coupling_obstruction_20261117_v64.tex) survives finite input repairs: local square classes and inert contacts persist on a positive integer progression. This closes that particular finite-exception escape at every obstructed coefficient shift. It does not prove an obstruction at every fixed shift. For the general target twenty-one, the derivative height proof cannot be imported without its abc hypothesis. A merger of the two general contact blocks needs an unconditional height argument under the already retained premises, together with prime realization and collision control. The original simultaneous actual-trace route remains open. For that original route it is enough to realize the two actual guarded fibers simultaneously for all but finitely many integer inputs at one fixed defining parameter. Their four free pole coordinates would then be removed. Since the retained source is sound, multiplying its polynomial by the finitely many missing integer-input factors restores completeness without adding a coordinate. A rational point on a curve with freely assigned trace, or local preparation at only the prescribed primes, does not establish this simultaneous realization premise. The [full auxiliary-point transfer](../papers/few_variables25/full_auxiliary25_20261117_v72.tex) enlarges this target by allowing the five auxiliary points in guarded affine `E(K)`, while the four root points retain integral indices. It preserves the scalar lower bound and both comparison precisions. The complete 25-coordinate source has degree at most 4,924,857,429,000,192; the corresponding 21-coordinate source still requires simultaneous actual-trace realization. Searches may now use half-integral auxiliary indices without altering the root-index height argument. The [actual-weight arithmetic reduction](../papers/pole_parameters/actual_trace_arithmetic_20261117_v70.tex) gives a fixed-tuple curve of genus at most 36 before the degree-nine contact substitution. The explicit genus-24 fixture is a smaller target for rational-point methods, with every candidate subject to the exact degree-nine equation, rational-square lift and original guards. It supplies no prepared global witness. The [actual-trace curve theorem](../negative_results/papers/pole_parameters/curves_and_abelian_families_20261117_v69.tex) gives finiteness of rational points on every fixed curve in each specified fixed-input locus, with all point roles independent. Its first fixed-point cover has genus at least eight for the derivative route and four for the general route. The [vertical-valuation theorem](../negative_results/papers/pole_parameters/vertical_and_fixed_slope_20261117_v70.tex) excludes every nonconstant rational abelian family through a prepared point, fixing the input and, for the derivative source, the root. The general statement allows all point roles to vary. In the derivative source, a fixed rational affine relation between slope and scale forces the root and slope to be constant even without a selected-prime hypothesis. The useful arithmetic target is therefore an individual guarded rational point, or points varying among different curves or on a higher-dimensional nonabelian variety. Everywhere local solubility of the derivative fiber and its boundary points do not establish global realization. The [fixed-slope incidence theorem](../negative_results/papers/rational_fibers/fixed_slope_trace_finiteness_20261117_v59.tex) proves finiteness on fixed genus-one trace curves for all independent point roles in the stated reweighted sources. Infinite propagation at a fixed slope therefore cannot complete that route. A variable slope, point-dependent auxiliary curve, untwisted conic, or individual rational witnesses remain possible. The theorem does not automatically transfer to the derivative source. ### Complete index divisibility and the remaining equality step The [elliptic index source](../papers/elliptic_divisibility/elliptic_index_poles_20261117_v60.tex) proves $h\mid j$ and has infinitely many fully prepared witnesses for each nonzero quotient. Under a fixed weak-abc exponent below two, a fixed index progression supplies the required simple primitive denominator primes. Truncating the formal logarithm amplifies the congruence between $X$ and $j/h$ to any fixed order with no new free coordinate. A sufficient missing condition is ```math h(X)+\log(1+|j/h|)\le C h(\sqrt{x(H)})+O(1), \qquad C0$; the paper supplies the necessary corrected height estimate. The source census is unchanged, with a sufficiently large ineffective fixed odd exponent. No new degree or coefficient record is assigned to the modified family. The remaining condition is a global rational square on actual detector tuples, not arbitrarily accurate local approximation. Two separate obstructions delimit that condition. The [actual-neighborhood theorem](../negative_results/papers/pole_parameters/actual_square_complement_neighborhoods_20261116_v43.tex) produces original-source independent return families with nonsquare complements throughout a fixed 23-adic neighborhood. The [normalization counterexample](../negative_results/papers/pole_parameters/square_complement_normalization_20261116_v43.tex) shows why applying a square-complement normalization directly to the final detector inputs destroys the pole test. The early scalar is changed before the detector is formed and has its own soundness proof. The [fixed-class synchronization theorem](../negative_results/papers/pole_parameters/fixed_class_synchronization_20261116_v43.tex) gives genus $1+2^r(r-1)$ for the common propagation curve when $r$ distinct positive pole scales and their twist classes are fixed. For two poles the genus is five, with an explicit $1+1+1+2$ Jacobian decomposition. Every fixed class tuple therefore gives only finitely many rational parameters. Infinitely many varying class tuples remain available. The full common-parameter variety has Kodaira dimension one when $r\ge2$, so direct Kummer-quotient theorems do not apply to it. The [literature audit](../negative_results/docs/fixed_fiber_literature_20261116_v43.md) also checks the relevant two-torsion, semistability and cohomological hypotheses; no theorem from Schinzel and finite Sha alone is claimed. The [abelian-parameter obstruction](../negative_results/papers/rational_fibers/abelian_parameter_obstruction_20261116_v44.tex) excludes a uniform rational identity on an abelian parameter space when the detector argument varies nontrivially, even if every twist class may vary. It applies both to rational square complements and to the original normalized pole equations. The argument forces an odd contact divisor and a negative odd pole valuation. It does not cover an actual detector family at a fixed comparison slope, a ramified cover, or individually chosen rational points. Local return sets still supply the required preparation and guards, but cannot establish the rational incidence themselves. At a fixed allowed slope with positive detector weight, the [fixed-argument ramification theorem](../negative_results/papers/rational_fibers/fixed_argument_complement_ramification_20261116_v44.tex) proves that the complement cover of the full independent auxiliary product is geometrically ramified. Therefore a surjective abelian coupling, including an isogeny or independent half-point doubling, cannot furnish a uniform square-complement section. The next targets are a proper coupled subvariety with a separate rational realization proof, or rational points directly on the ramified incidence cover. A common rational pole fiber without square complements remains a separate possible route. None of these obstructions proves nonexistence of the required individual actual detector tuples. The [varying-input quadric theorem](../papers/pole_parameters/varying_input_quadric_20261116_v44.tex) gives a rational parameter space for compatible coefficient inputs at any fixed nonexceptional common conic parameter. For a pole scale \(D>0\) and \(e=(\tau^2-D)/(\tau^2+3D)\), the surface with input \(b=D/(17V^2)\) is birational to $$x^2-Dy^2=e(\sigma^2+3D),\qquad V=\frac{\sigma x-Dy}{17(x-\sigma y)}.$$ The quadric has the explicit rational point \((\tau,1,\tau)\). The arithmetic target is to match this V with the native reciprocal trace while preserving the earlier source guards. This gives an exact incidence problem on which to try proper couplings or descent. The parametrization neither changes a final detector input by normalization nor proves that arbitrary actual tuples lift. Its explicit small-positive-input curve has square complement; its full two-parameter surface is the relevant broader family. The [native doubled-source norm map](../papers/pole_parameters/doubled_source_congruent_twists_20261116_v43.tex) supplies a different arithmetic handle: it constructs a nonzero-ordinate rational point on a congruent-number twist with abscissa square class $[\operatorname{Tr}(x\bar d)/2]$. Its class has not been identified with a required detector class. A useful next step is either an exact class identity or a descent argument trivializing the prescribed isogeny-Selmer elements, not only proving positive rank on an unrelated twist. This proposal shares the conic coordinate, not $e_i$ itself. Sharing $e$ requires matching input square classes; the separate [fresh-prime obstruction](../negative_results/papers/pole_parameters/fresh_prime_squareclass_obstruction_20261116_v42.tex) proves that the existing fresh-prime witnesses violate that requirement. The [common-twist theorem](../papers/pole_parameters/common_twists_20261114_v30.tex) gives genuine savings on its explicitly compatible locus, but does not put the full integer construction on that locus. A common $\tau$ allows different $e_i$ and avoids that particular necessary square-class equality. It still needs its own arithmetic existence proof. A fixed value of $\tau$ does not supply this missing result. The [fixed-parameter obstruction](../negative_results/papers/pole_parameters/fixed_pole_parameters_20261109_v17.tex) gives prepared inputs excluding $\tau=\pm1$ and $\tau=\pm17$, including arbitrarily small prepared inputs in the latter case. The proposed shared parameter must remain variable unless a new restricted completeness theorem proves otherwise. Those fixed-parameter counterexamples are global: their pole quartics are everywhere locally soluble by the preceding theorem. Their rank and square-class certificates do not constitute local obstructions. ### The stronger rational-complement target The [native complement theorem](../papers/pole_parameters/guarded_complement_reduction_20261116_v45.tex) sharpens the payoff of the square-complement route. At an actual point-stage tuple every character lies in \(K=\mathbb Q(\sqrt2)\). Prepared inputs have the form \(b_i=w_iS_i^2\), with even \(v_{17}(w_i)\). Vanishing of a complement at any character would make \(w_i/17\) a square in \(K\), contradicting its odd valuation at 17. Thus the complement norm guards are automatic. If actual simultaneous rational complements exist for every integer input, they themselves exclude the negative odd valuations needed by soundness. Both retained pole pairs can then be removed, giving the conditional counts **21 and 9**, with the existing one-coordinate encoder. This implication is proved; the native realization assertion is not. The **24-variable** common-fiber route remains a weaker sufficient target and does not require square complements. A global coupling with varying twist classes must prove one of these arithmetic incidence statements on the unmodified detector values. The same theorem constructs infinitely many prepared free inputs at a fixed common pole parameter, including square reduced denominators and all preparation depths. The missing step remains equality with the actual reciprocal traces. Neither a rational parametrization of the free-input quadric nor finite local preparation proves that equality. The [finite-class complement theorem](../papers/pole_parameters/finite_complement_classes_20261116_v46.tex) gives a weaker sufficient realization condition. Fix finite sets of levels 17m² and positive square classes supported on the source's original exceptional primes. For each actual input it suffices to make one allowed multiple of the corresponding complement a rational square. The complement norms are still units at every point-algebra character. Separate finite-algebra encodings followed by their product give the same 21/9 conditional census without a retained class-selection coordinate, even if branch radicands coincide. At the single level 17, allowing all exceptional square classes means that both actual complements need only have even valuations outside the original exceptional set. No consumed prime may be moved into that set. Its actual uniform rational realization is still missing. The explicit nonsquare-class coefficient family in the paper does not establish native trace matching. The [admissible-level theorem](../papers/pole_parameters/admissible_levels_native_traces_20261116_v47.tex) also permits finitely many fixed rational levels with nonnegative even valuations outside the exceptional set, provided every whole-algebra complement guard is proved. An exact native fixture at X=±1 and B=3 satisfies both prescribed-trace equations after fixed positive exceptional-unit rescalings, with all 128 guards checked. Its levels and rescalings are fixed constants; no completeness theorem for other inputs or sufficiently large defining B follows. The [polynomial-level local theorem](../papers/pole_parameters/polynomial_levels_local_20261116_v47.tex) supplies compensating test powers for levels cM(X)². Global realization and the corresponding height/completeness transfer are separate obligations. The [norm-compatible scalar theorem](../papers/pole_parameters/norm_compatible_scalar_20261116_v47.tex) removes the bare norm-conic obstruction in a complete prepared family. Its degree-eight input map reflects integrality and makes all four finite height-contact factors norms from Q(√10657). The new general scalar preserves the forced odd valuation at 76249, local preparation, height contacts and 25-coordinate census. The conditional sources already have uniformly soluble bare norm conics under compatible prime preparation. The concrete remaining target in either setting is $$ (m_j\operatorname{Tr}(\mathcal W_j^{-1}))^2-10657Y_j^2=4w_j/17,$$ with m_j from fixed finite admissible levels, the actual coupled detector data, and all guards. The free first coordinate of a conic point must not be substituted for the prescribed trace. The [detector-field norm scalar](../papers/pole_parameters/detector_field_norm_scalar_20261116_v52.tex) aligns this norm field with $K=\mathbb Q(\sqrt2)$ itself. Its integer elliptic label has degree four, and its scalar contact polynomial has degree sixteen. The scalar contact-height theorem is applied to the fixed rational rescaling $q=\beta/n$, while comparison tests use the integral label $\beta$. The full soundness, ordered completeness and whole-character guard transfer retain 25 variables under the existing arithmetic premises. Both prepared physical weights, and those of the unchanged conditional sources, are norms from $K$ without extra prime-pattern congruences. This supplies bare conics with class $d=2$; their actual first coordinates still need to be matched. No numerical determinant degree bound is transferred from a different scalar family. For the unchanged general scalar, the [signed finite-class theorem](../negative_results/papers/pole_parameters/finite_nonsquare_class_obstruction_20261116_v47.tex) excludes uniform realization using any finite collection consisting entirely of nontrivial square classes. Its three quartic contact fields and Chebotarev argument apply to every guarded rational auxiliary tuple, not only the prime-prepared family. Including the rational-square class or using the new norm-compatible scalar avoids this specific obstruction. ### Detector-dependent bounded levels The [bounded native norm-level theorem](../papers/pole_parameters/bounded_norm_levels_20261116_v48.tex) permits $m_i=M_iN_i/(c_i+N_i)$ and $a_i=17m_i^2$, with $N_i=\operatorname{Norm}(W_i)$. At every odd weight valuation, of either sign, the norm has nonpositive valuation and a zero-valuation norm reduces to one. The required local implication and whole-character denominator and complement guards therefore hold with no additional retained coordinate. For finitely many permitted coefficient choices, simultaneous actual realization of ```math \left(\frac{2M_iW_{i,0}}{c_i+N_i}\right)^2-d_i y_i^2=\frac{4w_i}{17},\qquad i=0,1, ``` with positive exceptional square classes $d_i$, still gives the conditional **21/9** census. The first coordinates remain the actual detector functions. For $M_i=c_i=1$, the exact relation $u_i^2-2v_i^2+z_i^2=1$ gives an additional incidence equation, with $00$ on prepared characters. This multiplier can have poles at good contacts. The normalized class $d_0=1-4w^*/17$ satisfies $16/17\le d_0<1$ and is a local unit at all relevant odd-weight primes. Its unrestricted conic has the fixed point $(1,1)$. More flexible classes $D(w^*,V^*,J^*)/E(w^*,V^*,J^*)$ are locally admissible when both polynomials have integral coefficients outside the original exceptional set and unit constant terms; their whole-character positivity and nonvanishing remain explicit obligations. Another permitted family is $(R+\kappa w^*/R)^2-4w^*/17$, with a specified whole-character unit $R$ and fixed positive exceptional-unit $\kappa>1/17$. The exact target is still simultaneous actual rational realization of ```math (m_iV_i^*)^2-d_i y_i^2=4w_i^*/17,\qquad i=0,1. ``` The [fixed-trace obstruction](../negative_results/papers/pole_parameters/fixed_trace_obstructions_20261116_v51.tex) excludes replacing both actual normalized traces by values from any finite fixed rational set for the original general scalar. In particular, the unit-trace specialization fails already at $X=4$, at the inert prime $252877$, independently of all auxiliary indices and the slope. The obstruction persists through the specified iterated native normalizations. For both conditional six-contact sources, a real scale bound excludes simultaneous unit trace at every prepared slope when $B\ge19$. The arithmetic target must therefore retain varying first coordinates, or use a separately justified modified scalar construction. The norm-compatible scalar remains available. A solution must match the actual traces in the displayed equations with all guards; a bare conic point or a free norm-one parametrization does not do so. No actual uniform realization is proved. The variable counts remain **25 and 13**, with **21/9** conditional on the displayed arithmetic assertion. The [proof and verification guide](results_20261116_v53.md) records the exact scope. ### Integral labels from one finite Pell choice The [integral Lucas-label theorem](../papers/integer_labels/pell_labels_20261115_v40.tex) supplies a different arithmetic compression. Let $a,n$ be integers with $$ |a|\ge3,\qquad c=|a-2|\ge3,\qquad n\ne0,\qquad t=(a-2)n+1. $$ If a rational $w$ satisfies $$ t^2-atw+w^2=1, $$ then monicity forces $w$ to be an integer. With $U_0(A)=0$, $U_1(A)=1$ and $U_{j+1}(A)=AU_j(A)-U_{j-1}(A)$, the sharp lower bound is $$ |t|\ge U_{c-1}(|a|)\ge(|a|-1)^{c-2}. $$ Thus a third separately integral label can be replaced by a finite quadratic choice when the first two labels are already exact integers. The equivalent square condition is $$ z^2=(a^2-4)\bigl((a-2)n+1\bigr)^2+4, $$ of total degree six and coefficient height sixteen. The same paper gives sharp witnesses and arbitrarily large solutions preserving prescribed finite congruence data of a given witness. The present source carries labels through local approximations by elliptic formal-parameter ratios. Those rational ratios are not known to be the exact integral labels, so monicity at the ratios does not prove the needed integrality. This distinction has a concrete counterexample: $a=5,n=1$ gives discriminant $340$, which is not a rational square, but for every $r\ge1$ there are rational norm-one solutions with $0