Temporal Equivalence Principle: Temporal Shear Recovery in Gaia DR3 Wide Binaries

6. Discussion: Resolving the Controversy

The wide-binary debate is often framed as a stark choice: either gravity is universally modified at low acceleration, or the observed signal is primarily artifactual. TEP offers a more specific interpretation—one that preserves the reality of the anomaly while explaining why the signal should appear with a finite transition scale and a measurable environmental dependence.

6.1 Why a Scale-Free MOND Interpretation Is Incomplete

The anomalous velocity boost is detected with high significance. The more discriminating question is whether it is adequately described as a scale-free enhancement or whether it instead follows a resolved transition in separation. The fitted screening profile is favored over a flat Newtonian description by $\Delta \chi^2 = 14{,}845$ and over a constant boost by $\Delta \chi^2 = 3{,}583$. A dedicated Newtonian orbital forward model, conditioned on the observed projected-separation distribution and marginalized over line-of-sight geometry, orbital phase, and eccentricity, produces an outer normalized median $\tilde v \simeq 1.01$ for the thermal case versus the observed $\simeq 1.37$ ($\chi^2 \simeq 14{,}531$ against the observed profile). The anomaly is therefore not merely an overall low-acceleration excess; it has a finite, separation-structured form that arises naturally in TEP because the scalar field remains environmentally suppressed until the binary enters a sufficiently diffuse local regime.

The model comparison sharpens that conclusion. A sigmoid transition is much worse than the canonical TEP exponential, while a more flexible double-exponential reduces the raw $\chi^2$ but still places the onset at a few thousand AU. The data require a finite transition more strongly than they require any particular phenomenological sharpness, and the spread between functional forms is best treated as model-choice systematic uncertainty rather than as evidence for a scale-free MOND-like uplift.

A direct MOND comparison makes this concrete. Fitting MOND interpolating functions to the binned profile using per-bin median masses (Table 4.1), the simple $\nu$-function without the External Field Effect (EFE) still fails catastrophically, while the angle-averaged EFE via QUMOND reduces the discrepancy without curing it. Even the most generous MOND+EFE variant (simple $\nu$ with free $g_e$) remains worse than the TEP exponential by $\Delta\chi^2 = +540$. The failure is threefold: the preferred acceleration scale is driven away from the canonical value, the transition shape is too steep, and the EFE-limited plateau undershoots the observed saturation.

The near-coincidence between $a_0$ and the TEP screening transition is not accidental. Smawfield (2025g) showed that the SPARC rotation-curve database yields a characteristic transition acceleration $g_{\rm TEP} \approx 5 \times 10^{-10}$ m/s$^2$, within a factor of four of $a_0$, while Smawfield (2025e) derived $a_0 \sim cH_0$ as an emergent consequence of the scalar field's strong-coupling scale. The distinction between the two frameworks is therefore not merely one of scale but of morphology: TEP produces a gradual, bounded enhancement whose amplitude is set by the local scalar-field value, whereas acceleration-dependent interpolation—even with the Galactic external field included—cannot simultaneously reproduce the transition shape, onset scale, and saturation level.

6.2 The Critical Density Scale

The canonical fit gives a transition separation of $R_s = 2{,}646 \pm 182$ AU, where the quoted error is the formal statistical fit uncertainty; when jackknife stability and model-choice freedom are included conservatively, the total uncertainty broadens to $\pm 609$ AU. For the sample median mass ($M \approx 1.2\,M_\odot$), this transition scale corresponds to an effective binary density $\rho_{eff} \approx 9.2 \times 10^{-18}$ g/cm$^3$. Although the Temporal Topology saturation density is far higher ($\rho_T \approx 20$ g/cm$^3$; Smawfield 2025g), the binary sits deep inside the screened Galactic potential. In the notation of Section 2.1, $\rho_{eff} \simeq \rho_{\rm floor} = \epsilon_{\rm env}\rho_T$, with $\epsilon_{\rm env} \sim 4.6 \times 10^{-19}$. As discussed there, $\rho_T$ and the scalar-field equation that governs $\epsilon_{\rm env}$ are both derived in earlier work; only the numerical evaluation of $\epsilon_{\rm env}$ in the specific Galactic environment remains to be computed from first principles. The observed transition density marks the point where the binary's internal Newtonian potential becomes sub-dominant to the pre-screened ambient background. The genuinely predictive test is the independent cross-check that follows.

The characteristic screening scale $R_s \approx 2{,}646 \pm 182$ AU is not merely a curve-fitting parameter; it represents the first kinematic detection of the Galactic screening floor. As shown in Section 2.1, the TEP characteristic acceleration $g_{\rm TEP} \approx 5 \times 10^{-10}$ m/s$^2$ (Smawfield 2025g) predicts a transition scale $R_s^{\rm pred} \approx 3{,}929$ AU for the transition-bin mass scale. This order-of-magnitude agreement (factor 1.48) is notable given that $g_{\rm TEP}$ is derived from rotation curves of distant galaxies. Including the Galactic external field as an additional screening floor ($\eta=2$) further reduces the discrepancy to a factor of 1.05 ($R_s^{\rm pred} \approx 2{,}778$ AU). The wide-binary anomaly is therefore quantitatively consistent with the cross-scale screening scale of the TEP conformal sector.

6.3 Sensitivity to the Mass-Luminosity Relation

The environmental ordering depends critically on the stellar mass estimates. If a solar-metallicity Mass-Luminosity Relation (MLR) is applied uniformly, the masses of metal-poor high-$|Z|$ stars are systematically overestimated, the Keplerian baseline $v_c = \sqrt{GM/s}$ is inflated, and the inferred velocity ratio $\tilde{v}$ in the high-$|Z|$ population is suppressed. The color-dependent MLR correction (Section 3.2) removes this bias by adjusting masses according to each star's offset from the disk color-magnitude ridge, but it now does so without circular beta calibration: spectroscopic metallicities are used if cached, otherwise a conservative external prior is adopted and stress-tested. The corrected analysis recovers the expected ordering: the high-$|Z|$ transition radius is smaller than the midplane value for both the full sample ($R_s = 4{,}662 \pm 196$ AU versus $7{,}131 \pm 1{,}341$ AU) and the solar-track control ($R_s = 4{,}145 \pm 276$ AU versus $6{,}856 \pm 920$ AU; $p < 10^{-4}$ for the full sample and $p < 10^{-3}$ for the solar track). That the solar-track control—which uses an empirical mass calibration with no color-dependent correction at all—independently recovers the same ordering makes the result difficult to dismiss as an MLR artifact.

Supplemental controls narrow the most common non-TEP interpretations still further. If the observed profile were driven mainly by one demographic subset, the transition would collapse under stratification by mass ratio, primary mass, or observational geometry. Instead, the signal persists across all four stellar-population half-samples—each remaining strongly preferred over a constant boost—and is still recovered in a stricter radial-velocity consistency subset of 6,117 systems, though with broader uncertainty owing to the much smaller sample size. Conversely, when catalog-level labels are scrambled, either globally or within distance quartiles, none of $10{,}000$ valid realizations reproduces the observed screening preference ($p < 10^{-4}$). The data indicate that the Gaia DR3 anomaly is structured, population-robust, and difficult to reduce to a single calibration artifact.

6.4 Limitations and Outlook

Several caveats deserve explicit acknowledgment. First, the reduced $\chi^2_{\nu} = 5.1$ for the canonical diagonal fit indicates that the diagonal error model does not capture the full scatter. A residual autocorrelation analysis identifies significant lag-1 correlation ($\rho_1 = 0.491$, $z = 2.14$, Durbin–Watson $= 0.97$), consistent with spatially correlated substructure or distance-correlated selection effects. Two covariance-aware refits address this directly: an AR(1) GLS model ($\chi^2_{\nu} = 0.97$) and a Gaussian Process with a squared-exponential kernel in $\log_{10}(s)$ space ($\chi^2_{\nu} = 1.01$). The GP is preferred by marginal log-likelihood and finds short-range correlated scatter ($\sigma_f = 0.015$, $\ell = 0.18$ dex). Under both covariance models all model-comparison conclusions are preserved, and parameters remain within the existing systematic uncertainty budget.

A direct investigation of the physical origin of this scatter finds no compelling single-population explanation. The standardized residuals show only weak rank correlation with median heliocentric distance, distance spread, and $\tilde{v}$ kurtosis, while the lag-1 residual autocorrelation rises from $\rho_1 = 0.09$ in the nearest distance quartile to $\rho_1 = 0.55$ in the most distant quartile. The correlated structure therefore appears concentrated in the distant subsample, consistent with a smooth observational selection effect rather than a localized astrophysical contaminant.

Second, the null-control and environmental permutation tests use $10{,}000$ realizations each, giving a $p$-value resolution floor of $10^{-4}$. No realization in either ensemble reproduced the observed screening preference, so the significance statements are limited primarily by the permutation-grid resolution rather than by an assumed parametric error model.

Third, an injection-recovery test validates that the pipeline recovers a known screening signal and does not hallucinate one when none is present. Mock catalogs are now constructed from the step_012 projected Newtonian forward-model null (rather than from globally shuffled observed $\tilde{v}$ values), then injected with the observed TEP enhancement parameters ($R_s = 2{,}646$ AU, $\alpha_{\rm sat} = 0.366$). This directly tests recovery against an orbital-population null that includes projection, phase-mixing, and eccentricity structure.

Fourth, dedicated triple-contamination forward models directly test the claim that unresolved hierarchies can mimic the observed profile. At $10\%$ residual contamination the predicted outer-bin median reaches only $\tilde{v} \approx 0.994$ versus the observed $\tilde{v} \approx 1.372$, and even at $50\%$ contamination it reaches only $\tilde{v} \approx 0.968$. The Pittordis-faithful population model performs no better, leaving a deficit of at least 0.38 in the outer-bin median. The median statistic is too robust for any plausible triple fraction to reproduce the observed enhancement.

Fifth, the observable $\tilde{v}$ is a projected quantity. A Monte Carlo orbital simulation quantifies the residual eccentricity systematic on $\alpha_{\rm sat}$. Relative to the thermal case, the uniform distribution recovers about 14% higher $\alpha_{\rm sat}$ and the most super-thermal case about -5% lower, while the realistic interval remains bounded between 8% and 0%. The transition radius remains stable to about 4% across the full sweep, so the eccentricity systematic does not propagate materially into $R_s$ or the model-comparison hierarchy.

Sixth, distance selection and mass calibration remain obvious places to look for failure modes. Both distance halves independently favor a finite transition, with the nearer subsample yielding $R_s = 3{,}369 \pm 415$ AU. Likewise, four alternative MLR prescriptions preserve the environmental ordering, with midplane $R_s$ spanning $6{,}911$ to $8{,}582$ AU and high-$|Z|$ $R_s$ spanning $4{,}156$ to $7{,}215$ AU. The result is therefore robust to the reasonable range of calibration choices already explored.

Seventh, the MOND comparison in Table 4.1 is already nontrivial, but it is not exhaustive. Marginalizing over full within-bin mass distributions or testing alternative EFE geometries may reduce the discrepancy somewhat. Even so, the present best MOND+EFE variant remains worse than TEP by $\Delta\chi^2 = +540$, so any viable rescue must overcome simultaneous failures in acceleration scale, transition shape, and outer-plateau amplitude.

Eighth, several quantities imported from other TEP papers—the Temporal Topology saturation density $\rho_T \approx 20$ g/cm$^3$, the characteristic acceleration $g_{\rm TEP} \approx 5 \times 10^{-10}$ m/s$^2$ (Smawfield 2025g), and the emergent MOND-scale derivation $a_0 \sim cH_0$ (Smawfield 2025e)—are drawn from preprints not yet independently peer-reviewed. The present analysis does not depend on the precise values of $\rho_T$ or $g_{\rm TEP}$ for any primary result; those quantities enter only in the cross-scale consistency check of Section 6.2. Readers should regard that check as a promising but provisional link pending independent verification.

Ninth, the pre-screening factor $\epsilon_{\rm env}$ is no longer fully post-hoc. Adopting the chameleon completion as a tractable benchmark (TEP itself does not commit to a specific microscopic mechanism; Smawfield 2025a, §A4), the chameleon scaling relation predicts the two-bin environmental ratio within the quoted errors, and the inferred potential index from the two-bin fit is $n = 1.4 \pm 0.7$. A finer stratification into 5 equal-count $|Z|$ bins, each containing 68{,}263 systems, yields transition radii declining from $R_s = 8{,}340 \pm 1{,}910$ AU at median $|Z| = 22$ pc to $R_s = 3{,}924 \pm 229$ AU at median $|Z| = 347$ pc. The corresponding log-linear exponent is $0.45 \pm 0.14$, implying $n = 1.22 \pm 0.69$. Although the five-bin rank correlation alone is not individually decisive, the overall downward trend remains robust and sharpens the case for TEP screening.

Tenth, the environmental comparison in Section 5.1 presents fixed-$\alpha_{\rm sat}$ results as the primary protocol because the amplitude–scale degeneracy prevents clean isolation of $R_s$ when $\alpha_{\rm sat}$ floats independently in each subsample. A joint profile-likelihood fit eliminates this concern: both subsamples are modeled simultaneously with a single shared $\alpha_{\rm sat}$ and separate transition radii, yielding $R_s = 4{,}912 \pm 904$ AU (midplane) versus $R_s = 3{,}447 \pm 355$ AU (high-$|Z|$) with $\alpha_{\rm sat} = 0.332 \pm 0.020$. A likelihood ratio test gives $\Delta\chi^2 = 60.3$ ($p = 8.1 \times 10^{-15}$), and the ordering is preserved in 100% of bootstrap resamples.

Eleventh, at the shortest separations unresolved spectroscopic binaries could in principle deflate the inner normalization and inflate the apparent outer enhancement. The normalization sensitivity sweep tests this directly by refitting the canonical model under every contiguous baseline window that remains plausibly within the screened core. Across those screened-core windows the recovered $R_s$ ranges from $2{,}106$ to $3{,}647$ AU, a spread of only 41% around the fiducial value. The transition scale is therefore stable while the apparent saturation amplitude responds to the normalization, which is the opposite of the pattern expected from a baseline-deflation artifact.

Twelfth, the saturation amplitude $\alpha_{\rm sat}$ varies systematically across the demographic half-samples in a pattern consistent with mass-dependent self-screening. The high-primary-mass subset yields $\alpha_{\rm sat} = 0.237$, the full sample $0.366$, and the low-primary-mass subset $\alpha_{\rm sat} = 0.509$ ($\Delta\chi^2 = 2{,}234$ versus a constant boost for the strongest split). A quantitative self-screening model $\alpha_{\rm sat}(M) = \alpha_0\,\exp(-M/M_{\rm screen})$ fits these three data points with $\chi^2 = 0.89$ for one degree of freedom, yielding a wide-binary-regime bare coupling $\alpha_0 = 1.75 \pm 0.22$ and a self-screening mass scale $M_{\rm screen} = 0.45 \pm 0.03\,M_\odot$. This $\alpha_0$ is the locally active coupling at wide-binary scales, where the source-charge suppression factor $\mathcal S_\Sigma(\mathcal E)$ has only partially suppressed the underlying Temporal Shear; it is not directly comparable to either the Cassini PPN bound $\alpha_0 \lesssim 3\times10^{-3}$ in the deeply screened Solar System (Smawfield 2025a, Section 7) or the compact-object regime indicator $\alpha_{\rm eff} \sim 10^6$ extracted from millisecond-pulsar spin-down (Smawfield 2025b), because TEP explicitly predicts that each environment samples a different effective coupling. The relevant cross-scale check is qualitative: the same conformal-sector physics that screens the Temporal Shear in dense regimes and recovers it in low-density regimes underlies all three measurements. Notably, the self-screening model is itself a continuous exponential function of mass—there is no thin-shell step function or discrete screening boundary—consistent with the Temporal Topology framework in which self-screening operates via continuous flattening of the field profile rather than a discrete shell transition.

6.5 Systematic Controls Summary

The following table collects the robustness checks performed across the analysis pipeline. Each row tests whether the primary conclusions—a finite screening transition and environmental ordering—survive a specific perturbation to the data selection, error model, mass calibration, or binning procedure.

No single control eliminates every conceivable systematic, but the breadth of this matrix—spanning data selection, error modeling, mass calibration, binning, projection, contamination, and theoretical prediction—makes it difficult to construct a single astrophysical or methodological scenario that simultaneously survives all tests while mimicking both the transition morphology and the environmental ordering.

6.6 Distinguishing TEP from Generic Scalar Screening

The chameleon-completion benchmark in Section 5.2 demonstrates that, within a tractable microscopic realization of TEP screening, the observed environmental modulation is consistent with a standard Ratra–Peebles potential. A fair question is therefore: what distinguishes TEP from any other chameleon or symmetron model? TEP itself does not commit to a chameleon (or any other) microscopic completion (Smawfield 2025a, §A4, §7); chameleon, Vainshtein, Galileon, DBI, and symmetron mechanisms are all admissible candidates, with the defining ontology being continuous suppression of the Temporal Shear $\Sigma_\mu = \nabla_\mu \ln A$ by source-charge and environmental state. Four features are specific to TEP rather than generic to the broader scalar-screening family:

1. The cross-scale anchor. The transition acceleration $g_N(R_s) \approx 1.1 \times 10^{-9}$ m/s$^2$ at the observed screening radius is within a factor of two of the independently measured $g_{\rm TEP} \approx 5 \times 10^{-10}$ m/s$^2$ from SPARC rotation curves (Smawfield 2025g), with no parameter adjustment. Generic scalar-screening models (chameleon, symmetron, Vainshtein in their stand-alone forms) have no independent prediction for the absolute screening scale in wide binaries; in TEP this scale is fixed by the Temporal Topology saturation density $\rho_T$ and the SPARC-derived $g_{\rm TEP}$, regardless of which microscopic completion realizes the suppression.
2. The mass-dependent saturation amplitude. The monotonic progression $\alpha_{\rm sat} = 0.237$ (high mass) $\to$ $0.366$ (full sample) $\to$ $0.509$ (low mass) is a natural consequence of TEP's conformal coupling, where self-screening by the binary's internal potential attenuates the velocity-profile amplitude. Standard chameleon models predict mass-dependent screening radii but not a specific amplitude hierarchy tied to an independently constrained coupling scale.
3. The emergent $a_0$. TEP derives $a_0 \sim cH_0$ as a consequence of the scalar field's strong-coupling scale (Smawfield 2025e), explaining the near-coincidence between $a_0$ and the TEP screening transition without fine-tuning. Chameleon and symmetron models do not generically produce this coincidence.
4. Continuous Temporal Topology and the proper-time signature. Standard chameleon thin-shell models (Khoury & Weltman 2004) predict a discrete shell boundary with a step-like transition from screened to unscreened regions, yielding a profile with a geometric $R_s/s$ prefactor, and treat the resulting scalar gradient purely as an extra fifth force on matter. TEP is not committed to thin-shell matching: its defining ontology is continuous Temporal Topology (Smawfield 2025a, §7), where screening operates via the smooth spatial profile of $\ln A(\phi)$, the locally observable Temporal Shear $\Sigma_\mu = \nabla_\mu \ln A$ is suppressed continuously in dense environments rather than gated by a shell boundary, and the same conformal factor $A(\phi)$ that sources the kinematic enhancement also rescales matter-frame proper time as $\mathrm{d}\tau/\mathrm{d}t = A(\phi)$. The kinematic and clock-rate signatures are therefore tied together in TEP, whereas a generic chameleon predicts only the kinematic effect. This predicts a smoother transition than thin-shell models—specifically, a pure exponential approach to saturation rather than the thin-shell form $f(s) = 1 - (R_s/s)\,e^{-m_{\rm bg}(s-R_s)}$. The data already favor this prediction: the sigmoid model, which represents a sharper step-like onset of the kind that discrete thin-shell boundaries would produce, is rejected at $\Delta\chi^2 = +131.5$ relative to the TEP exponential (Table 4.1). This morphological distinction is not available to generic chameleon models, which share the thin-shell functional form.

These distinctions are currently anchored to quantities from the wider TEP series that are not yet independently peer-reviewed (Section 6.4, Eighth). The most decisive future test is the predicted $R_s(R_{\rm gc},\,|Z|)$ map: TEP predicts a specific density-dependent gradient calibrated by the Temporal Topology saturation density $\rho_T$ and the cross-scale acceleration $g_{\rm TEP}$, whereas a stand-alone chameleon (or other) screening model would require fitting both the potential index and the response normalization as free parameters. Gaia DR4, with its expanded volume and improved astrometry, could enable the finer $|Z|$ stratification needed to distinguish these scenarios. An anisotropy test—comparing the screening transition along and perpendicular to the Galactic disk—would provide a further discriminator, since the TEP screening floor is set by the local baryonic density field through the source-charge suppression of the Temporal Shear, rather than by a uniform acceleration threshold.