# Global Time Echoes: Empirical Synthesis
**Matthew Lukin Smawfield**
Version: v0.8 (Singapore)
First published: 21 December 2025 · Last updated: 17 October 2026
DOI: 10.5281/zenodo.18004832
---
## Abstract
Analysis of 25.3 years of GNSS timing data (2000–2025) reveals a
persistent, distance-structured correlation in global atomic clock
networks that tests whether corrected distributed GNSS timing residuals
contain a coherent spatial covariance beyond the standard relativistic
and geodetic measurement model. Examination of 165.2 million station-pair
observations from 474 unique receivers (814 station codes) over a
25.3-year baseline demonstrates a spatial correlation signal decaying
with distance over a finite scale (Gaussian-family kernels preferred
in the 25-year dataset, exponential ΔAIC = 12.8; power law excluded
at ΔAIC > 30; azimuth-averaged exponential scale
λ = 3,210 ± 389 km, with cross-sector directional dispersion
λT = 4,201 ± 1,967 km across eight azimuthal sectors, a measurement
of order thousands of kilometres rather than a single fixed scale), following
cross-centre validation yielding R² = 0.92–0.97 across three distinct
analysis-centre software architectures on a 62.7 million observation baseline. These findings emerge from a systematic multi-paper research
program: theoretical framework development with pre-specified expectations and theoretical search ranges
(Paper 0), multi-center consistency across software-diverse product families within
the shared IGS tracking ecosystem (Paper 1), 25-year longitudinal analysis enabling long-period
geophysical detection (Paper 2), raw-data consistency test strongly constraining
precise-product processing artifacts (Paper 3), and cosmological extension testing whether temporal covariance
or gradient effects contribute to dark-sector phenomenology (Paper 4).
The evidence programme combines timing-network signatures, environmental tests, and an optical sensitivity bound:
distance-structured spatial decay over a finite scale;
East-West/North-South anisotropy (ratio 2.16 ± 0.23 across the 25.3-year aggregate; nominal pair-count p $\lt$ 10−15);
orbital velocity coupling (r = −0.888, 3.65σ autocorrelation-corrected;
surrogate p $\lt$ 2×10⁻⁷ resolution bound); a recovered annual-phase
direction near Earth's aphelion velocity tangent (RA = 6°, Dec = +4°;
directional-rank permutation p $\lt$ 10⁻⁵, with the directed CMB-apex template
anti-phase, the Solar Apex geometrically incompatible, and MGEX projection-degenerate);
outer-planet opposition coupling (40-event opposition subset null-validated at 5.5× over matched random dates, p $\lt$ 10⁻¹⁷, alongside a continuous planetary-potential correlation r = +0.116, p = 0.012, whereas the broader 156-event free-centre pulse scan reflects calendar-position sampling on the seasonal series); 18.6-year lunar nutation coupling (R² = 0.641 ± 0.10 over 1.36 cycles); an optical-domain sensitivity bound from passive Satellite Laser Ranging (sampling-matched controls reproduce the apparent spectral concentration; inter-station candidates remain exploratory); pulsar spin-down
excess (exploratory channel); and host-potential H0 correlation. On the held-out MGEX product the distance-structured coherence replicates (λ ≈ 1,862 km), while the anisotropy and primary orbital-velocity metrics do not replicate robustly. Raw RINEX Single Point
Positioning with broadcast ephemerides achieves consistent signal detection
across all 72 metric combinations (t-statistics up to 112, Cohen's d up to
0.304), strongly constraining precise-product processing artifacts as the sole origin. Because broadcast ephemerides still carry control-segment orbit information, terrestrial closure is structured in three tiers: the GNSS analyses characterize the macroscopic microwave clock-network covariance; Satellite Laser Ranging supplies a hardware- and orbit-independent optical sensitivity bound ($\delta A \lesssim 2\times 10^{-8}$); and physical-confound forward models together with orbit-free optical-fiber and two-way time-transfer links adjudicate the residual orbit-projection channel at atomic-clock precision. The network's
selectivity profile—sensitive to velocity-dependent dynamics while blind
to GM/r² scaling and solar rotation—characterizes it as an inertial
interferometer measuring correlation geometry rather than a gravimeter
measuring Newtonian force.
Table 5 scores these observations against the pre-specified
expectations of the Temporal Equivalence Principle: the orbital-velocity
sign, EW > NS ordering, band localization, and GM/r² null are met,
while the 2,000–3,000 km screening-parameterized forecast, the $\lt$ 5%
cross-centre criterion, and—in the 25-year data—the exponential-kernel
preference are not met. TEP is a bi-metric scalar-tensor framework in which
proper time is a dynamical field governed by a conformal factor A(ϕ) =
exp(βAϕ/MPl). Temporal Shear suppression operates through the
continuous spatial profile of the ϕ field (Temporal Topology), with
suppression arising from the non-linear superposition of field gradients
(Temporal Shear), replacing discrete thin-shell approximations with a
geometrically continuous mechanism. The observed correlation length λT
represents the characteristic scale of Temporal Topology. Environmental
screening, derivative-screened, and saturation-radius interpretations are
candidate theoretical completions, not the primary empirical claim; the
saturation-radius interpretation RT is adopted in the present
framework (§4.4.2).
The framework preserves local Lorentz invariance while predicting
spatially correlated clock-network residuals in the conformal sector and
closed-loop synchronization holonomy in the disformal/non-exact sector.
The measured GNSS covariance is interpreted primarily as a
conformal-sector \(A(\phi)\) observable; it should not be identified
with the non-zero closed-loop holonomy predicted only by the
non-exact/disformal sector. The empirical evidence presented here
therefore supports the conformal sector alone; the disformal-sector
holonomy — the framework's definitive discriminating test — remains
unmeasured at the canonical amplitude (the J0437 phase-closure
detection attests non-exact transport but sits roughly two orders of
magnitude above the canonical connection's ceiling on that
sightline).
**Evidence Tiers:** A (Paper 0/EXP): Foundational theory with
pre-specified search ranges; B (GNSS I–III): Primary timing-network
evidence; C (SLR): Optical-domain bound and exploratory correlation test; D (H₀, COS, WB):
Environmental response tests; E (JWST, GL, UCD, RBH): Conditional
astrophysical/cosmological extensions. These evidence channels are
not statistically independent: the exponential decay, anisotropy, orbital
coupling, preferred-axis recovery, nutation, and planetary-catalogue signatures share
the same GNSS data and phase-alignment method, while the SLR check,
pulsar spin-down, and host-potential H0 correlation
use distinct data — the effective number of independent tests is
smaller than the number of listed results, and smaller still at channel level: the SLR check
reads the same conformal clock sector through different hardware, the
CMB axis is product-unstable, and the pulsar channel is exploratory
(its standalone evidence uninformative — mildly GR-favouring — in the
J0337 adjudication, where the kinetic impedance solved in Paper 0 suppresses the differential acceleration ratio to $\eta \simeq 2.5 \times 10^{-7}$, well within the experimental bound $2.6 \times 10^{-6}$). Critically, the conformal sector responsible for clock-rate
modulation is not directly constrained by photon–graviton differential-propagation tests such as GW170817, although local conformal-gradient/source-charge sectors remain constrained by PPN, clock, and equivalence-principle tests.
If validated through independent replication, TEP predicts that the dynamical and optical phenomenology attributed to dark matter is the observational projection of Temporal Topology and Temporal Shear across unstretched space, not evidence for a new particulate matter component. In this interpretation, GNSS-scale covariance, dynamical phantom mass, and unstretched-space optical refraction (Papers 4, 18, 19; where intermediate clock rates $A_{\rm loc}(u_{\rm grav}(\mathbf x))$ along the shared ADM lapse $N_{\rm TEP}=A_{\rm loc}N_{\rm GR}$ at fixed spatial 3-metric $\gamma_{ij}$ refract light via $n(x)\simeq 1-2\Phi_N/c^2-\beta_A u_{\rm grav}(x)$, whereas isotropic conformal rescaling $\tilde g_{\mu\nu}=A^2 g_{\mu\nu}$ cancels out of null rays) are different observational projections of the same time-field topology, with Papers 4, 6, 13, and 19 testing distinct optical and dynamical projections of the environmental-response architecture on the selected two-branch action $P_{,X}=k\sqrt\xi+2\xi$ ($k=16.03$ fixed by $a_0$, zero per-system free parameters), not a derived Earth-to-galaxy covariance-to-shear transfer kernel. The 4,000 km correlation on Earth and galactic-scale phenomenology are interpreted within TEP as manifestations of the same environment-dependent Temporal-Topology response, which provides continuous Temporal Topology relaxation from deep potential wells to the weak-field regime. The saturation-density scale ρT ≈ 20 g/cm³, and the diagnostic radius RT = (3M/4πρT)1/3 it defines for each body, remain testable through interplanetary missions. Explicit falsification criteria include: failure of independent groups to replicate the raw carrier-phase signal; Temporal Topology correlation length falling outside the 500–20,000 km range; adjudication of the ephemeris-artifact question — GNSS-product provenance via the Satellite Laser Ranging check, and the wider orbit-projection confound via the physical-confound forward model or an orbit-free clock channel; and null synchronization holonomy in closed-loop triangular time-transfer experiments.
*Keywords:* Temporal Equivalence Principle, GNSS, atomic clocks,
preferred direction, dark matter, gravitational lensing, scalar-tensor
gravity, synchronization holonomy, Temporal Topology, Temporal Shear,
nonlinear screening mechanisms
## 1. Introduction
## 1.1 The Synchronization Residual Problem
Modern atomic clocks have achieved extraordinary precision, with optical
lattice clocks demonstrating fractional frequency stability at the 10−18
level—sufficient to detect gravitational redshifts from centimeter-scale
height differences (Bothwell et al. 2022). Yet despite this remarkable local
precision, global clock synchronization residuals plateau at
10−15, three orders of magnitude above the fundamental limit.
This discrepancy is conventionally attributed to systematic noise arising
from ionospheric fluctuations, tropospheric delays, multipath interference,
and instrumental drift.
The Global Navigation Satellite System represents humanity's most extensive
precision timing network, comprising hundreds of ground stations that
maintain nanosecond-level synchronization across thousands of kilometers
(Hofmann-Wellenhof et al. 2008; Teunissen & Montenbruck 2017). Standard
processing pipelines apply sophisticated corrections for all known
relativistic effects—gravitational redshift, velocity-dependent time
dilation, Sagnac rotation, and Shapiro delay—following IERS Conventions to
millimeter precision (Petit & Luzum 2010). After these corrections, residual
correlations are expected to be consistent with random noise.
Analysis of 25.3 years of GNSS timing data reveals a different picture. A
systematic, distance-structured correlation persists in the residuals,
characterized by a decay length of approximately 4,000 km. Standard GNSS
processing explicitly suppresses spatial correlations to optimize
positioning accuracy. The signal reported here survives precisely because it
is geometric and differential rather than energetic or common-mode, placing
it outside the correction basis of conventional pipelines. This correlation
has been present in GNSS data since the network's inception but was
categorized as systematic error rather than physical phenomenon. The present
paper argues that this unexplained correlation structure represents a
physical signal—specifically, the signature of a dynamical time field
predicted by scalar-tensor extensions of general relativity.
## 1.2 Historical Context
If this signal is physical, why was it not identified in the first decades
of GNSS operation? Three developments converged to enable its detection.
The first is temporal baseline. The 25.3-year dataset now available spans
1.36 complete cycles of the 18.6-year lunar nutation, enabling first access to
long-period geophysical coupling that was inaccessible to earlier studies —
though the sub-two-cycle record bounds each phase bin to at most two
independent cycle realizations, limiting the coupling measurement to
R² = 0.641 ± 0.10.
Analyses conducted prior to 2010 lacked sufficient temporal leverage to
distinguish such signals from transient artifacts.
The second is multi-constellation cross-check. The maturation of
GLONASS, Galileo, and BeiDou provides independent verification across
different satellite geometries and processing chains. Single-constellation
analyses are vulnerable to constellation-specific artifacts that
multi-constellation cross-check can identify and exclude.
The third is methodological. Standard GNSS processing employs Kalman
filtering and network adjustment algorithms designed to eliminate
correlations, thereby optimizing positioning accuracy. The present analysis
instead preserves correlation structure through phase-coherent
cross-spectral methods applied to residuals. The signal was present
throughout the history of GNSS operation but was categorized as systematic
error rather than physical phenomenon.
## 1.3 Theoretical Motivation
The standard relativistic framework treats proper time as a derived
quantity: given the metric tensor gμν, proper time along any
worldline is uniquely determined by integration. This formulation assumes
that time transport is integrable—that synchronization around any closed
loop returns to the initial epoch after accounting for known effects such as
Sagnac rotation, Shapiro delay, and gravitational redshift.
Scalar-tensor theories of gravity, developed extensively over the past three
decades (Bekenstein 1993; Damour & Polyakov 1994; Khoury & Weltman 2004),
predict modifications of proper time accumulation to vary spatially. If this
assumption proves empirically false in the presence of conformal metric
couplings, apparent dark matter could arise as a geometric artifact:
time's effect on moving matter read as gravity from unseen mass. If
proper time is not
merely derived from geometry but is itself influenced by a dynamical field,
synchronization becomes path-dependent in ways that standard general
relativity does not predict.
The Temporal Equivalence Principle formalizes this possibility. It states
that all non-gravitational dynamics, signals, and quantum phases evolve
according to the proper time defined by a matter metric g̃μν that
couples universally to matter through the scalar field. In local freely
falling frames, physics reduces to special relativity with invariant c;
globally, the flow of time becomes path-dependent in ways that distributed
timing networks can detect.
### 1.3.1 Research Program Structure
This integrated manuscript synthesizes results from a systematic multi-paper
research program designed to establish, validate, and independently confirm
the TEP predictions through progressively rigorous empirical tests and
cosmological extension. The program follows a theory-first methodology:
pre-specified expectations and theoretical search ranges established before empirical analysis, followed by
multiple independent checks using complementary methodologies,
culminating in cosmological implications.
Paper 0: Theoretical Foundation (Smawfield 2025a, DOI:
10.5281/zenodo.16921911) — The Temporal Equivalence Principle framework was
developed independently of any GNSS analysis, establishing pre-specified expectations and theoretical search ranges including a Temporal Topology correlation length, exponential
spatial decay form, velocity-dependent anisotropy, and absence of GM/r²
scaling. Screening operates via the continuous spatial profile of the φ field
(Temporal Topology), with Temporal Shear suppression arising from the non-linear superposition of field
gradients (Temporal Shear) rather than discrete thin-shell boundaries,
suppressing Temporal Shear where the ambient time-gradient is steep
(weak-field: high-acceleration regions) while leaving cosmology
accessible to dynamics. This theory-first approach ensures that subsequent
empirical findings represent genuine predictions rather than post-hoc
explanations.
Paper 1: Multi-Center Consistency (Smawfield 2025b, DOI:
10.5281/zenodo.17127229) — Initial empirical investigation analyzed 62.7
million station-pair measurements from 364 GNSS stations across three
distinct analysis-center solutions (CODE, IGS, ESA) over 2.5 years (2023–2025).
Cross-center consistency (R² = 0.92–0.97) established that the correlation
structure is not processing-specific, with λT = 3,330–4,549 km falling within
the predicted range. This multi-center approach excluded implementation-specific
artifacts as a viable explanation; systematics residing in the shared
constellation geometry are addressed separately through multi-constellation
and raw-RINEX replication (Papers 14 and 3).
Paper 2: 25-Year Longitudinal Analysis (Smawfield 2025c, DOI:
10.5281/zenodo.17517141) — Extended temporal baseline to 25.3 years
(2000–2025) using CODE products, analyzing 165.2 million station pairs from
474 unique receivers. This long-span analysis enabled detection of
previously inaccessible signatures including 18.6-year lunar nutation
coupling (R² = 0.641), and a recovered annual-phase direction near Earth's
aphelion velocity tangent; the directed CMB-apex template is anti-phase and the Solar Apex is
geometrically incompatible, predicting N-S anisotropy against the observed
E-W dominance), and confirmed temporal stability of
the correlation structure with cross-sector dispersion λT = 4,201 ± 1,967 km. The 25-year baseline
provided access to long-period geophysical phenomena impossible to detect in
shorter studies.
Paper 3: Raw RINEX Test (Smawfield 2025d, DOI:
10.5281/zenodo.17860166) — Processing artifact constraint through analysis
of raw GNSS observations using Single Point Positioning with broadcast
ephemerides only. Because broadcast ephemerides still contain control-segment
orbit information, terrestrial verification is structured across three tiers:
the GNSS analyses (Papers 1–3, 14) establish and characterize the macroscopic
clock-network covariance; passive Satellite Laser Ranging (Paper 8) provides a
GNSS-product-independent optical sensitivity bound and orbit-provenance
check; and the physical-confound forward model together with satellite-free
optical-fiber and TWSTFT clock channels adjudicate the generic
orbit-projection confound at full metrological precision.
Analysis of 1.17 billion pair-samples
across 539 stations (2022–2024) achieved consistent signal detection across all 72
metric combinations (t-statistics up to 112, Cohen's d up to
0.304), confirming the signal exists in fundamental observables independent
of sophisticated processing chains. This raw-data consistency test strongly
constrains precise-product processing artifacts as the origin.
Paper 4: Cosmological Implications (Smawfield 2025e, DOI:
10.5281/zenodo.17982540) — Developed the connection between TEP and
gravitational lensing, demonstrating how temporal-field gradients (active
Temporal Shear) and intermediate temporal depth across unstretched space can produce apparent dark matter phenomenology in both the
dynamical and optical sectors without particulate dark matter (contrasted with the isotropic-conformal benchmark where $A^2(\phi)$ cancels out of null rays and lensing tracks baryons alone). This paper
formulated the Isochrony Axiom as a foundational closure assumption in
standard lensing analysis that TEP subjects to direct empirical stress-testing, identified "Phantom
Mass" — the apparent mass inferred when temporal structure is analysed as
spatial mass — and established the
dynamical–lensing mass comparison and unstretched-space temporal refraction tests, together with the
connection between the ~4,000 km GNSS correlation and galactic-scale phenomenology through the 𝒮Σ screening operator. The continuous relaxation of Temporal Topology
from deep potential wells to the weak-field regime unifies terrestrial and
galactic scales. This cosmological
extension transforms TEP from a terrestrial metrology curiosity into a
framework with profound implications for dark matter and cosmological
tensions.
### 1.3.2 Related Literature and Prior Anomalies
The GNSS correlation structure documented in this manuscript is not the first
observational anomaly consistent with TEP dynamics. Peer-reviewed literature
contains multiple documented deviations from standard relativistic predictions
that were marginalized as systematic artifacts because ΛCDM lacks the
requisite theoretical framework to accommodate them. Recontextualized
through TEP, these isolated findings form a coherent pattern of empirical
support:
**GPS One-Way Propagation and the Synchronization Problem.**
Kelly (2009), in a peer-reviewed analysis of GPS timing, demonstrated that
one-way light signals circumnavigating Earth eastward take 414.8
nanoseconds longer than westward signals at the equator—measuring c − v
eastward and c + v westward relative to Earth's surface. The Sagnac
effect demonstrates operationally that extended synchronization is
geometry- and frame-dependent: eastward and westward propagation times
around a rotating Earth are unequal before the standard rotational
correction is applied. This known GR effect is not itself evidence for
TEP. Instead, it illustrates why a genuine TEP test must construct a
residual closed-loop observable after subtracting the modeled Sagnac,
Shapiro, gravitational-redshift and gravitomagnetic contributions. Any
remaining non-zero residual would probe non-exact temporal transport
beyond the standard model (Section 6.3.3).
**SLR Network Time Biases.** Exertier et al. (2017) identified
systematic time biases of 3–4 nanoseconds (1σ) across 25 Satellite Laser
Ranging stations, producing 4–6 mm coordinate discrepancies that persist
after all known corrections. These biases are "often neglected" and
"presumed to be $\lt$ 100 ns" but resist precise estimation. TEP interprets
these as *Temporal Topology relaxation*: each station at different
gravitational potential experiences differential proper-time accumulation,
consistent with the screening field producing GNSS correlations.
**GPS Clock Gravitational Redshift Residuals.** Fathollahi
et al. (2023) reported gravitational redshift tests using GPS Block IIF Rb
clocks showing fractional deviations from GR of (0.23 ± 1.34) × 10−3,
with 2.1–7.3% of data flagged as outliers requiring removal. The authors
noted "annual oscillation and short-term fluctuations" that necessitated
16-day moving average smoothing. TEP predicts precisely such phase-locked
annual modulations and outlier populations from velocity-dependent temporal-field
coupling (Section 4.2).
**Multi-GNSS System-Specific Clock Behavior.** Ai et al. (2021)
documented that GPS, Galileo, BeiDou, and GLONASS exhibit *distinct*
noise profiles (random walk, flicker, white frequency modulation) despite
using similar atomic clock technologies. Standard frameworks attribute this
to hardware differences. TEP predicts system-specific residuals because
each constellation operates at different orbital altitudes and inclinations,
sampling different clock-rate structure of the temporal field.
**GNSS Common-Mode Periodic Signals.** Santamaria-Gomez et al.
(2017) identified system-specific periodic signals in coordinate time series
arising from "different orbital plane structures, orbit repeatability and
incomplete constellation," achieving 52% horizontal reduction after
system-specific correction. These were attributed to unmodeled geophysical
loading. TEP predicts orbital-period modulations from satellites sampling
spatially-varying temporal fields at different inclinations—precisely the
pattern identified but unexplained in the standard literature.
The convergence of these independent anomalies—each documented in
peer-reviewed literature, each marginalized within ΛCDM, each consistent with
TEP predictions—substantiates the framework's empirical foundation beyond
the primary GNSS correlation analysis presented in this manuscript.
The present manuscript integrates these five studies into a unified
narrative, demonstrating that the TEP framework established pre-specified expectations and theoretical search ranges
before empirical analysis, that three independent empirical
investigations using progressively rigorous methods are each consistent with
the expected signatures, and that the framework's cosmological extension
offers testable expectations for gravitational lensing and dark matter
phenomenology. This theory-expectation-evidence-cosmology sequence, with
each empirical study addressing distinct artifact hypotheses
(processing-specific, temporal transience, sophisticated processing) and the
cosmological paper extending implications beyond terrestrial scales, follows
the standard structure of scientific hypothesis development and theoretical
refinement.
## 1.4 Prediction Timeline
The TEP framework was developed independently of the GNSS analysis, with
pre-specified expectations and theoretical search ranges established in prior theoretical work
(Smawfield 2025a). The predicted Temporal Topology correlation length range of
λT = 1,000–10,000 km was derived from environmental screening theory before any
GNSS data analysis was conducted.
The chronology is significant: the TEP theory paper was first deposited in
August 2025 (v0.1, DOI: 10.5281/zenodo.16921911), and the specific
clock-network correlation forecast—exponential decay with a
screening-parameterized length and multi-center cross-validation—entered
the versioned record on 15 September 2025 (v0.4, DOI:
10.5281/zenodo.17121655), one week before the first public deposit of the
GNSS analysis (Paper 1 v0.6α, 22 September 2025). The multi-center GNSS
analysis was conducted from September through November
2025; the 25-year CODE longspan analysis was completed in November 2025; and
raw RINEX test confirmed the signal in fundamental observables in
December 2025. The observed Temporal Topology correlation-length family (cross-sector dispersion λT = 4,201 ± 1,967 km) falls
within the pre-specified search range, a non-discriminating consistency statement since the range subsumes leading systematic scales. The full expectation-by-expectation mapping to versioned deposits is
tabulated in the prediction provenance register (Section 6.5).
## 1.5 The Temporal Equivalence Principle
TEP extends Einstein's geometrization of space to the rate of time itself.
The principle states that all non-gravitational dynamics, signals, and
quantum phases evolve according to the proper time defined by a single
causal metric g̃μν that couples universally to matter. The rate at
which proper time accrues is governed by a scalar field φ through a
conformal factor A(φ). In local freely falling frames, physics reduces to
special relativity with invariant c; globally, the flow of time becomes
path-dependent.
This framework preserves all established physics locally—the speed of light
remains exactly invariant in any freely falling laboratory—while predicting
novel global effects that manifest as correlations in distributed timing
networks. Crucially, the conformal sector of TEP, which governs clock-rate
modulation, remains unconstrained by multi-messenger observations such as
GW170817, which bound only disformal (cone-tilting) effects.
## 1.6 Scope and Non-Claims
To avoid misinterpretation, the following claims are explicitly NOT made by
this work:
Detection of a fifth force acting on test masses (the network responds
to geometric configuration, not gravitational amplitude; GM/r² and GM/r³
scaling are both null)
Violation of local Lorentz invariance (the speed of light remains
exactly invariant in any freely falling laboratory; TEP preserves local
special relativity)
Breakdown of the equivalence principle in laboratory tests (universal
coupling ensures the weak equivalence principle is preserved exactly;
the TEP synodic-phase residual
$\kappa_{\rm LLR} \approx -3.91 \times 10^{-4}$ from the residual-channel
LLR analysis (Paper 17) is a post-fit residual proxy whose relationship
to the standard Nordtvedt parameter has not yet been established — it
is not directly comparable to the direct-fit LLR bound
$|\eta| \lt 4.4 \times 10^{-4}$ (Williams et al. 2004), and the
source-level integrator refit is required to determine whether the
residual-channel signal corresponds to a physical Nordtvedt parameter
within that bound)
Modification of light or gravitational wave propagation speeds (GW170817
constrains |cγ − cg|/c ≲ 10−15; TEP satisfies this
through disformal coupling B(φ) constrained to be small today by multi-messenger observations)
Violation of energy-momentum conservation (total stress-energy is
conserved; apparent non-conservation in the Einstein frame is resolved
by transformation to the Jordan frame where matter follows geodesics of
g̃μν)
TEP is a conformal scalar-tensor theory that modulates the rate of proper
time accumulation while preserving all local physics. The GNSS signal
reflects spatial correlations in clock rates arising from the scalar field's
two-point correlation function, not violations of established physical
principles.
## 1.7 Claim Hierarchy
#### Tier 1: Empirical Claim (Primary)
GNSS clock residuals exhibit a persistent, distance-structured,
anisotropic spatial correlation with characteristic decay scale
λT = 4,201
± 1,967 km (cross-sector dispersion; azimuth-averaged exponential
λ = 3,210 ± 389 km). This correlation is robust across three software- and
estimator-diverse product families within the shared IGS tracking
ecosystem (CODE, IGS, ESA; R² = 0.92–0.97), persists over 25.3
years spanning multiple satellite constellation changes, and appears in
raw RINEX observations processed with Single Point Positioning (consistent
signal detection across all 72 metric combinations, t-statistics up to 112).
The programme combines the following measurements and checks:
distance-structured spatial decay over a finite scale, East-West/North-South anisotropy (ratio 2.16), orbital
velocity coupling (r = −0.888, 3.65σ autocorrelation-corrected;
surrogate p $\lt$ 2×10⁻⁷ as a resolution bound), a recovered annual-phase
direction near Earth's aphelion velocity tangent on the CODE product
(the directed CMB-apex template is anti-phase; the MGEX-product scan is
projection-degenerate), an SLR optical sensitivity bound with exploratory inter-station correlations (the spectral concentration is reproduced by sampling-matched noise),
planetary event catalogue (56/156 significant under the free-centre
statistic; catalogue excess attributable to calendar-position sampling
in season-preserving controls — a bounded channel whose continuous
planetary-potential correlation survives at r = +0.116, p = 0.012), 18.6-year lunar
nutation coupling (R² = 0.641 ± 0.10), pulsar spin-down excess
(exploratory channel), and
host-potential H0 correlation.
#### Tier 2: Interpretive Claim (Secondary)
The observed correlation structure is inconsistent with known
atmospheric effects (ionosphere, troposphere), instrumental systematics
(geomagnetic storms, altitude dependence), classical gravitational
effects (GM/r² force scaling, GM/r³ tidal scaling), solar activity (r ≈
0 with solar flux), and processing artifacts (persists in raw SPP data).
The signal's selectivity profile—sensitive to velocity-dependent
dynamics and a stable preferred direction while blind to mass scaling
and solar rotation—distinguishes it from conventional systematic errors. However,
the possibility of a presently unmodeled GNSS systematic tied to
fundamental constellation geometry cannot be definitively excluded
without independent non-GNSS check (Section 3.6).
#### Tier 3: Theoretical Hypothesis (Tertiary)
The observations are consistent with the Temporal Equivalence Principle
(TEP), a bi-metric scalar-tensor framework in which proper time is
modulated by a dynamical conformal field A(ϕ) = exp(βAϕ/MPl).
TEP established pre-specified expectations and theoretical search ranges before empirical analysis (λT =
1,000–10,000 km, exponential decay, velocity-dependent anisotropy,
absence of mass scaling). Several key pre-specified structural expectations are met, while Table 5 records the forecasts that are not met. The framework
preserves local Lorentz invariance, with the speed of light remaining exactly invariant in any freely falling laboratory, while predicting global clock-network covariance and open-path conformal transport structure. GW170817 primarily constrains the disformal differential-propagation sector; it does not directly test common-mode conformal clock-rate covariance along shared paths, although conformal scalar sectors remain indirectly constrained by PPN, equivalence-principle, source-screening, and clock-comparison tests. This hypothesis is testable through independent experiments: SLR check (Section 6.3.1.1), multi-constellation cross-check, and triangle holonomy test (Section 6.3.3). The validity of Tier 1 (empirical observations) does not depend on the correctness of Tier 3 (TEP interpretation).
This paper establishes Tier 1 through extensive empirical testing, argues for Tier
2 through systematic exclusion of conventional explanations, and tests
consistency with Tier 3 through pre-specified expectations. Independent
replication is essential for all three tiers.
## 1.8 Paper Structure
This paper presents evidence for TEP through systematic analysis of GNSS
timing data. Section 2 characterizes the phenomenology of the observed
signal. Section 3 tests the signal against artifact hypotheses through
multiple complementary tests. Section 4 presents the theoretical framework and
its predictions. Section 5 develops the cosmological implications, including
connections to gravitational lensing and dark matter phenomenology. Section
6 specifies explicit falsification criteria and outlines the experimental
path to a definitive verdict.
Falsification is central to this work. Concrete rejection criteria are
specified, including failure of independent groups to replicate the signal
in raw carrier-phase data, Temporal Topology correlation length falling
outside the 500–20,000 km range, adjudication of the ephemeris-artifact
question — GNSS-product provenance via the Satellite Laser Ranging check,
and the wider orbit-projection confound via the physical-confound forward
model or an orbit-free clock channel — and null synchronization holonomy in
closed-loop triangular
time-transfer experiments.
## 2. Phenomenology
This section presents the empirical findings without theoretical interpretation. The signal characteristics are established through three complementary empirical studies, each designed to address distinct testing requirements:
Paper 1 (Multi-Center Consistency, 2023–2025): Cross-validation across three software- and estimator-diverse product families (CODE, IGS, ESA) within the shared IGS tracking ecosystem, using 62.7 million station-pair measurements from 364 stations. This study established that the correlation structure is not processing-specific, achieving R² = 0.92–0.97 consistency across distinct processing chains (Bernese, NAPEOS, and the IGS combination product).
Paper 2 (25-Year Longitudinal Analysis, 2000–2025): Extended temporal baseline using CODE products with 165.2 million station-pair observations from 474 unique receivers (814 station codes). This long-span analysis enabled detection of previously inaccessible signatures including 18.6-year lunar nutation coupling and a stable preferred direction, confirming temporal stability over 25.3 years.
Paper 3 (Raw RINEX Test, 2022–2024): Processing artifact constraint
through analysis of raw GNSS observations using Single Point Positioning
with broadcast ephemerides only. Broadcast ephemerides still contain
control-segment information, so Satellite Laser Ranging and non-GNSS optical
checks remain necessary for adjudication — SLR offers a
GNSS-product-independent orbit-determination chain, with signal recovery
at sufficient sensitivity still required, while the generic orbit-projection confound is reserved
for the physical-confound forward model and orbit-free clock channels.
Analysis of 1.17 billion pair-samples across 539 stations achieved consistent signal detection across all 72 metric combinations, confirming the signal exists in fundamental observables.
The convergence of findings across these three complementary methodologies—each addressing different potential artifact sources, with the raw-RINEX channel providing the strongest estimation-chain independence—provides the empirical foundation for the reported signatures.
## 2.1 Spatial Correlation Structure
Clock frequency residuals exhibit systematic spatial structure that persists after removal of all modeled relativistic effects. The phase coherence between station pairs decays exponentially with geodesic distance according to:
$$C(r) = A \cdot \exp(-r/\lambda_{\mathrm{T}}) + C_0$$
where r denotes the inter-station distance, λT the Temporal Topology correlation length, A the amplitude, and C0 the baseline offset. Fits to distance-binned means across approximately 28 bins achieve R² = 0.920–0.970 across all three analysis centers.
The correlation metric employed is the magnitude-weighted phase alignment index, computed via cross-spectral density analysis in the 10–500 μHz band (corresponding to periods of 33 minutes to 28 hours). This approach measures whether clock fluctuations are in phase regardless of amplitude—information that survives GNSS processing because network adjustment removes common-mode offsets while preserving differential phase structure. Magnitude weighting ensures that frequency bins with stronger cross-spectral power contribute proportionally more to the phase average, so the metric reflects genuine correlated signals rather than noise.
The primary result from the CODE 25-year analysis is a Temporal Topology correlation length of λT = 4,201 ± 1,967 km. Cross-center validation on the 2023–2025 multi-center sample (exponential fits to binned phase-alignment means) yields: CODE λT = 4,549 km (95% CI: 1,198–5,918 km); IGS λT = 3,764 km (95% CI: 3,197–4,871 km); ESA λT = 3,330 km (95% CI: 2,532–3,984 km). The 25-year CODE long-span headline is a separate estimand from these center-specific exponential fits. The coefficient of variation across the three center fits is ~13%, indicating consistency across software-diverse processing chains and substantially reducing the likelihood of a CODE-specific implementation artifact (shared IGS network/datum systematics remain live alternatives).
## 2.2 Timing Signatures and Cross-Domain Checks
Table 1 combines the longitudinal GNSS signatures with companion environmental tests and the current SLR bound. Each signature constrains the space of
viable explanations at its own evidential strength: the timing-network
rows are the primary evidence, while the SLR row is an instrument-level
provenance check within the same conformal clock sector, the preferred-direction
recovery is product-dependent across GNSS products, and the pulsar row is an
exploratory consistency channel; their collective pattern consistently points toward a
velocity-dependent, preferred-axis geometric mechanism.
| Signature | Observed Value | Significance |
| --- | --- | --- |
| Distance-Structured Decay | λ = 3,210 ± 389 km (azimuth-avg. exponential); λT = 4,201 ± 1,967 km (cross-sector dispersion) | R² = 0.92–0.97 |
| Spatial Anisotropy | EW/NS = 2.16 | p $\lt$ 10−15 |
| Orbital Velocity Coupling | r = −0.888 | 3.65σ (autocorr.-corrected); surrogate p $\lt$ 2×10⁻⁷ (resolution bound) |
| Annual-Phase Direction Recovery | RA = 6°, Dec = +4°; ~6° from Earth's aphelion velocity tangent | Directional-rank permutation p $\lt$ 10⁻⁵; CMB apex anti-phase; Solar Apex null; no global full-sky-search surrogate reported |
| Optical-Domain Sensitivity Bound | Microwave-to-optical product chains (GNSS/SLR) | Sampling-matched spectral null; exploratory inter-station candidate. Independent hardware and orbit provenance; physical corroboration remains pending (Paper 8) |
| Planetary Alignment Coupling | 40-event opposition subset: 5.5× over random dates (p $\lt$ 10−17); continuous r = +0.116 (p = 0.012) | Opposition subset null-validated; GM/r² and GM/r³ scaling null bounds transient Newtonian impulses; broader 156-event free-centre scan (56/156 ≥2σ) descriptive of seasonal-curve sampling (Paper 3) |
| 18.6-Year Nutation | R² = 0.641 ± 0.10 | p = 0.010; permutation p = 0.0016 (1.36 cycles, cycle-limited) |
| Pulsar Spin-Down Excess | κMSP ~ 106 (fitted phenomenological coefficient) | Exploratory: globular-cluster density scaling (Paper 10); the standalone pulsar channel is uninformative — mildly GR-favouring (Paper 34, in preparation) |
| Host-Potential H0 Correlation | Fitted environmental response; corrected H0 conditional on Paper 11's observing chain and anchor weights | Cepheid response (Paper 11) |
## 2.3 Annual-Phase Direction Recovery
The grid search recovers the phase of an annual directional modulation. Under the corrected prograde orbital-velocity convention the best fit is RA = 6°, Dec = +4° — the antipode of the RA = 186°, Dec = −4° solution reported under the earlier antiparallel convention. The two coordinates describe the same undirected geometric axis, but the sign matters for testing a directed physical template. The corrected direction lies near Earth's aphelion velocity tangent, not the CMB apex.
The recovered direction lies ~6° from Earth's aphelion velocity tangent, ~161.8° from the directed CMB apex, and ~92° from the Solar Apex. Its Pearson correlation is r = 0.7465 (local parametric p = 4.0 × 10−7). At the CMB-apex direction the signed correlation is r = −0.5455 (anti-phase); at the Solar Apex it is r = 0.0059 (null). The best-fit direction explains about 16,000 times the variance of the Solar-Apex template.
A 100,000-draw directional-rank permutation test produced no exceedance of the observed r, giving a finite-resolution plus-one bound p $\lt$ 10−5. This test ranks the observed annual phase against sampled directions; it is not a surrogate-data calibration of the maximum over a complete full-sky search. No global full-sky-search p-value is therefore claimed. The undirected axis happens to lie ~18° from the CMB axis, but that geometric proximity cannot be used as directed CMB evidence when the CMB-apex template has the opposite sign.
The recovered direction is a product-dependent statement. The MGEX combined
product (Paper 14) returns a nominal maximum 82.7° from the CMB dipole, but
its orbital-velocity geometry is projection-degenerate: 61 of 703 directions
perform at least as well as the direct CMB-labelled template, and the CMB and
nominal-best predictors correlate at r = 0.984. MGEX therefore supports at
most exploratory annual directional structure; it neither localizes a unique
three-dimensional axis nor provides CMB-frame confirmation. What MGEX does
establish is the spatial scale: it independently reproduces the bounded
thousands-of-kilometres correlation length in a distinct combined
multi-GNSS product; its standard anisotropy and primary orbital estimators
are null, while an exploratory model-free PA-difference statistic recovers
the expected annual association.
Within the TEP framework, the recovered direction encodes the phase of the
annual modulation. Its proximity to the aphelion velocity tangent supports
an orbital-velocity reading rather than a fixed external-frame arrow. The
undirected ~18° axis proximity to the CMB axis is geometric only and does not
leave a directed CMB-frame identification open in this analysis.
Reference-frame discrimination under the corrected convention no longer identifies the CMB arrow: evaluated at the CMB-apex direction the model anti-aligns (r = −0.55), while the Solar Apex remains statistically null (r = 0.007, R² = 0.0054%). The recovered best fit instead sits ~6° from Earth's aphelion velocity tangent — the orbital-velocity direction the predictor principally encodes. The geometric argument against the Solar Apex is unchanged: its +30° Dec predicts North-South anisotropy, directly contradicting the observed East-West dominance (EW/NS = 2.16), whereas both the recovered axis and the CMB dipole axis lie near the ecliptic plane. Hemisphere asymmetry corroborates the velocity-channel reading: Southern stations couple at |r| = 0.79 against |r| = 0.25 in the North.
## 2.4 Planetary Event Catalogue: Bounded Channel
Analysis of 156 planetary conjunction and opposition events yields 56 events (35.9%) exceeding 2σ under the pre-specified free-centre pulse statistic, with 19 surviving family-wide Bonferroni correction. These counts quantify the repeatability of the estimator on the seasonal coherence series rather than event-locked physics, for two reasons established in the audit: (i) the fitted Gaussian pulse centre is unconstrained across the ±120-day window and can latch onto off-event seasonal excursions; (ii) the date-shuffled null destroys the autocorrelation of the seasonal curve and therefore under-controls for calendar position. The season-preserving controls of Paper 3 (Step 2.6b) make this explicit in the raw-RINEX channel: uniform-date nulls on the real 3-year series reproduce a 65.5% pulse-fit rate against 59.5% for real events (p = 0.805), and same-day-of-year calendar twins in non-event years reproduce the event rate exactly (59.46% versus 59.46%), for both the mean-coherence and phase-alignment metrics. The apparent catalogue excess is thus attributable to calendar-position sampling of the underlying annual modulation rather than to discrete planetary impulses.
What survives is narrower but well-defined. The matched-null control validates event specificity for the 40-event outer-planet opposition subset under a ±15-day model-free window (5.5× effect-size ratio over 155 random dates, Mann-Whitney p $\lt$ 10−17), and a maximal non-overlapping-window subset shows 13/29 significant events versus 1.3 expected (binomial tail p ≈ 10−10; Paper 2 Step 2.9) — evidence of repeatable seasonal-window structure, though not separable from calendar position. Separately, the continuous channel is distinct: the daily network-coherence variability correlates with the composite planetary potential at r = +0.116 (autocorrelation-corrected p = 0.012, Neff ≈ 472), a long-baseline association that does not reduce to discrete event locking. Individual events can remain anomalous (the March 2025 Venus conjunction stayed date-locked at +3.3 days under a sliding-window placebo test), but such cases are isolated observations, not a catalogue-level detection.
The channel's firmest contribution is the scaling null: neither GM/r² (force) nor GM/r³ (tidal gradient) scaling is detected across the catalogue (all p > 0.5), and the Paper 1 mass-ordering test returns r = −0.156. In the 3-year RINEX window this bounds unmodelled transient planetary perturbations of the coherence field: whatever produces the seasonal modulation is insensitive to which planet is involved. The catalogue therefore functions in this synthesis as a calibrated falsification/upper-bound control — it rules out unscreened impulse coupling to planetary mass or tidal gradient — rather than as an affirmative detection of discrete planetary responses.
### 2.4.1 Candidate Mechanism (Hypothesis): Velocity-Field Gradient Modulation
The residual planetary phenomenology—the validated opposition subset, isolated date-locked events such as Venus March 2025, and the continuous r = +0.116 correlation with the composite potential—could, within TEP's velocity-dependent framework, reflect geometric modulation of Earth's motion through the scalar-field gradient rather than gravitational amplitude. This is presented strictly as a candidate interpretation of the surviving channels, not as an explanation of the full catalogue, whose apparent excess is attributable to calendar-position sampling.
**Physical Mechanism:** Earth moves through the φ field with a 30 km/s orbital component plus any ambient-background motion. This analysis measures the annual orbital phase and does not identify that background with the 369 km/s CMB frame. Planetary configurations create time-varying perturbations to the local field gradient ∇φ through their own screening halos. When planets align with Earth's velocity vector relative to the ambient field, constructive interference may enhance the velocity-dependent anisotropy; when perpendicular, destructive interference may suppress it. The effect depends on the alignment angle θ between the planetary position and Earth's velocity vector:
$$\Delta\lambda(\theta) \propto \lambda_0 \left[1 + \epsilon \cos\theta\right]$$
where ε is a small coupling parameter (~10−3) and θ is measured from the recovered preferred direction. This predicts maximum modulation when planets are aligned with or opposite to the preferred-direction vector (θ = 0° or 180°), and null modulation when perpendicular (θ = 90°).
**Why Mass Scaling is Absent:** The mechanism operates through geometric alignment rather than gravitational force. Each planet creates a local perturbation to ∇φ within its own transition scale RT ∝ M1/3. However, the Earth-planet distance (AU scale) far exceeds any planetary RT (Mercury: ~100 km; Jupiter: ~5,000 km), placing Earth in the unscreened regime where the field perturbation has decayed to a threshold-dependent residual. The detection depends on whether this residual exceeds a critical threshold for modulating Earth's local screening length, not on the perturbation's amplitude. This threshold behavior explains why Mercury (smaller M but closer) produces comparable detection rates to Jupiter (larger M but farther).
**Testable Prediction:** Within the surviving channels (the opposition subset and the continuous correlation), response measures should track the alignment angle between planetary position and the preferred direction. A catalogue-wide version of this test must first pass the season-preserving null: stratifying the 156 events by alignment angle provides a falsifiable check only where event locking has been established, since calendar-position effects must be removed before any angular dependence can be read out.
## 2.5 Selectivity Profile: Inertial Interferometer
The pattern of detections and non-detections characterizes the network's sensitivity and provides powerful constraints on viable physical mechanisms. Table 2 summarizes this selectivity profile.
| Observable | Result | Interpretation |
| --- | --- | --- |
| Orbital velocity | r = −0.888, 3.65σ (autocorr.-corrected) | Sensitive to kinematic dynamics |
| Semiannual coherence component | R² = 0.904 | Sensitive to inertial orientation (nutation/orbital harmonic) |
| 18.6-year nutation | R² = 0.641 | Sensitive to long-period dynamics |
| Preferred direction | r = 0.746 at recovered direction | Sensitive to preferred axis |
| Planetary alignments | Catalogue excess reproduced by calendar position; opposition subset 5.5× over random dates | Bounded/control channel — bounds unmodelled transient planetary perturbations; continuous potential correlation r = +0.116 (p = 0.012) remains |
| GM/r² and GM/r³ scaling | Both null (p > 0.5) | Insensitive to gravitational amplitude (force or tidal) |
| Solar rotation (27-day) | Null (r $\lt$ 0.09) | Insensitive to solar surface phenomena |
| Lunar standstill | Null | Insensitive to static geometry |
This selectivity profile is highly discriminating. The network responds to velocity-dependent and orientation-dependent phenomena while remaining blind to gravitational amplitude and solar surface effects. This pattern suggests that the GNSS network functions as an inertial interferometer—measuring velocity-dependent correlation geometry—rather than as a gravimeter measuring Newtonian force.
The distinction is fundamental. A gravimeter would show GM/r² (force) scaling, while a differential tidal sensor would show GM/r³ (gradient) scaling; in both cases, Mercury's influence would be negligible compared to Jupiter's. An inertial interferometer responds to geometric configuration regardless of mass—precisely what is observed. The annual-direction recovery identifies the operative measured phase with Earth's orbital motion through the ambient field; it does not identify a cosmological rest-frame arrow.
## 2.6 Statistical Power
The raw RINEX test analysis achieves extraordinary statistical power. Across 1.17 billion pair-samples, the directional anisotropy yields t-statistics up to 112 with Cohen's d up to 0.304. The signal is detected across all 72 metric combinations with mean R² = 0.93.
Hemisphere asymmetry provides independent corroboration. Southern stations show stronger orbital-velocity coupling (|r| = 0.79) than Northern stations (|r| = 0.25), corroborating the velocity-channel geometry. Both hemispheres independently show East-West exceeding North-South correlation (Northern: 1.20×, Southern: 1.35×), ruling out hemisphere-specific artifacts.
Monthly consistency is equally striking. East-West exceeds North-South in 94% or more of all 36 months analyzed, with short-distance coherence ratios showing coefficient of variation below 1%. The underlying signal is constant; the annual modulation in full-distance Temporal Topology correlation lengths reflects atmospheric screening effects that vary seasonally.
## 3. Testing
Extraordinary claims require extraordinary testing. This section systematically addresses the most plausible artifact hypotheses, demonstrating that the signal survives each challenge.
## 3.1 Processing Artifact Hypothesis
The most immediate concern is that the signal arises from sophisticated processing chains—network adjustments, Kalman filtering, or integer ambiguity resolution. To address this possibility, raw RINEX observation files were processed using Single Point Positioning with broadcast ephemerides only, representing the simplest possible processing chain and one entirely independent of the network adjustments employed by CODE, IGS, and ESA.
The raw RINEX test dataset comprises 539 stations over three years (2022–2024), totaling 1.17 billion pair-samples. The signal is detected with consistent results across all 72 metric combinations tested, with mean R² = 0.93. Directional anisotropy matches the CODE findings, with East-West exceeding North-South by 2–22% at short distances ($\lt$500 km). A critical audit confirms this is robust to distance bias: E-W pairs are 13 km longer than N-S pairs (a bias *against* the signal), and distance-matching strengthens the ratio.
The correlation length in raw data (λ ≈ 700–1,100 km) is shorter than in precise products (λ ≈ 4,000 km), consistent with ionospheric noise masking the long-range structure. When ionospheric effects are removed (Ionofree mode), the raw correlation length increases, supporting a "Ladder of Precision" where the signal scale converges toward the CODE benchmark as noise is mitigated. Orbital velocity coupling is detected at r = −0.763 (nominal 5.4σ; ≈3.0σ autocorrelation-corrected in Paper 3) in multi-GNSS mode, and the directional scan recovers RA = 8°, Dec = +5° under the corrected orbital-velocity convention — ~160° from the CMB apex as an arrow, ~20° from the dipole axis, and ~4° from the aphelion velocity tangent — consistent with the 25-year analysis. The signal therefore exists in the fundamental observables and is difficult to attribute solely to tested processing artifacts.
## 3.2 Ionospheric Hypothesis
If the signal were ionospheric in origin—driven by solar activity modulating electron density along signal paths—it should correlate with solar cycle variations. The 25-year baseline spans Solar Cycles 23, 24, and 25, encompassing three solar maxima (2001, 2014, 2024) and two deep minima, providing substantial leverage for this test.
The correlation between the signal and the Solar Flux Index is r ≈ 0, indicating no significant relationship. The correlation length λ remains stable across solar maxima within uncertainty bounds. Moreover, the signal persists in the dual-frequency ionosphere-free L1+L2 combination (λ = 1,069 km, R² = 0.969), and excluding high-ionosphere periods improves the signal by 21–23%. This last observation indicates that the ionosphere suppresses rather than creates the correlation. A simple ionospheric origin is therefore strongly disfavoured by these controls.
## 3.3 Geomagnetic Hypothesis
If the signal were driven by geomagnetic storms affecting clock electronics or signal propagation, it should vary systematically with geomagnetic activity. Stratification by GFZ Kp index data across all 72 metric combinations yields median Δλ ≈ −1% at the primary threshold (Kp $\lt$ 3 versus Kp ≥ 3), with 60 of 72 tests falling within ±5% across all station filters and processing modes. No systematic deviation appears at higher storm thresholds (Kp ≥ 4, Kp ≥ 5). The signal is therefore geomagnetically invariant.
## 3.4 Tropospheric Hypothesis
If the signal were tropospheric in origin—driven by water vapor or pressure variations that correlate spatially—stations at high altitude (with thinner atmospheric columns) should show systematically different correlation lengths than sea-level stations.
Analysis of 360 regressions spanning global and latitude-controlled altitude quintiles reveals that only 3.1% exhibit p $\lt$ 0.05 slopes, consistent with chance expectation. The Q5/Q1 ratio (highest to lowest altitude quintile) has median 0.97 (IQR 0.76–1.27), indicating that low-altitude and high-altitude stations yield statistically indistinguishable λ values.
The tropospheric column depth varies by approximately 20–30% between sea-level and high-altitude stations. If the correlation structure were tropospheric, one would expect systematic increase in λ at high altitude, latitude dependence correlated with tropopause height, and seasonal modulation correlated with monsoon patterns. None of these signatures are observed. The signal is altitude-invariant, consistent with a phenomenon operating at gravitational potential scales (~6,400 km) rather than atmospheric scales (~10 km).
**Comprehensive Test Summary:** The altitude independence test represents one component of a broader testing framework spanning 388 independent statistical tests across multiple analysis families. False Discovery Rate correction (FDR-BH: 52.3%, Hierarchical Empirical Bayes: 39.7%, Bonferroni: 40.0%) demonstrates robustness against multiple comparison artifacts. This extensive empirical testing, combined with cross-center consistency (cross-centre coefficient of variation ≈ 13% on the exponential λT triplet (CODE, IGS, ESA; §2.1)), establishes that the observed correlations survive rigorous scrutiny across independent processing pipelines, environmental conditions, and statistical frameworks.
## 3.5 Long-Period Stability
If the signal were a transient artifact of the analysis window, it would not persist across the full 25.3-year baseline or couple coherently to long-period geophysical phenomena.
The data reveal clear coupling to multiple geophysical cycles. The 18.6-year lunar nutation shows R² = 0.641 ± 0.10 (p = 0.010, F-test with N = 12 phase bins; permutation test p = 0.0016, N = 100,000; 1.36 complete cycles observed — cycle-limited: each phase bin carries at most two independent cycle realizations). The semiannual coherence component shows R² = 0.904 (p = 2.7 × 10−5, F-test; permutation p = 2.7 × 10−4; AR(1) red-noise surrogate p = 2 × 10−5, N = 100,000; over 50 complete cycles; apsidal-locked, compatible with nutation term or orbital second harmonic). The Chandler wobble (433 days) shows R² = 0.096 with phase stability of 0.72 across more than 21 cycles. Core signatures—correlation length, anisotropy, and orbital coupling—remain consistent across the full 25-year span. The signal is therefore temporally stable across multiple geophysical cycles. Surrogate validation confirms that the nutation p-values are not inflated by treating the 165 million station-pair observations as independent: the F-test uses N = 12 phase bins (not N = 165 million), and non-parametric permutation and AR(1) red-noise surrogate tests confirm the significance without Gaussian or independence assumptions.
## 3.6 Most Plausible Remaining Conventional Explanation
The strongest remaining conventional hypothesis is that the observed correlations arise from a presently unmodeled, long-baseline GNSS systematic that survives both network adjustment and raw SPP processing. This section defines the exact cumulative constraints that any such systematic must satisfy simultaneously across the completed test battery.
**The Hypothesis:** A sophisticated, distance-dependent systematic error exists in GNSS observations that: (1) produces exponential decay with λ ≈ 4,000 km; (2) exhibits East-West/North-South anisotropy ratio of 2.16; (3) couples to Earth's orbital velocity with r = −0.888; (4) reproduces an annual phase near Earth's aphelion velocity tangent while the CMB-apex template is anti-phase; (5) produces repeatable seasonal-window coherence modulation in the planetary event catalogue without mass scaling; (6) couples to 18.6-year lunar nutation and the semiannual coherence component; (7) survives independent processing by CODE, IGS, and ESA; (8) persists in raw RINEX data processed with broadcast ephemerides only; and (9) retains compatible distance structure across 72 non-independent metric combinations, with metric-dependent response.
**Constraints from Tests:** Each test step constrains this hypothesis without claiming impossibility:
- **Multi-center consistency (R² = 0.92–0.97, cross-centre CV ≈ 13% on the exponential λT triplet (§2.1)):** The systematic must be present in three software- and estimator-diverse product families (Bernese, the IGS combination product, NAPEOS) within the shared IGS tracking ecosystem. This excludes center-specific artifacts but does not exclude shared-network, constellation-geometry, or datum-convention systematics, nor systematics inherent to GNSS observation geometry or satellite constellation design.
- **Raw RINEX test (all combinations, t-stats up to 112):** The systematic must survive Single Point Positioning with broadcast ephemerides, which processes each station independently without network adjustment. This excludes network-adjustment artifacts but does not exclude systematics in the broadcast ephemeris itself (the "ephemeris loophole" addressed in Section 6.3.1.1).
- **25-year temporal stability:** The systematic must persist across satellite constellation changes, hardware upgrades, and processing methodology evolution spanning 1.4 complete cycles of the 18.6-year lunar nutation. This excludes transient artifacts but does not exclude systematics tied to fundamental GNSS geometry.
- **Annual-phase recovery:** The systematic must reproduce a phase near Earth's aphelion velocity tangent. The directed CMB-apex template is anti-phase and the Solar Apex is null. An annually phased direction tracking orbital velocity is geometrically difficult to explain through satellite constellation geometry but remains a possible confound if an unrecognized coupling exists between GNSS observation geometry and Earth's heliocentric motion.
**Remaining Possibility:** The most plausible remaining conventional explanation is a systematic tied to the fundamental geometry of GNSS satellite constellations that couples to Earth's motion through space in a way not previously recognized. Such a systematic would need to produce all observed signatures through a single coherent mechanism—a requirement that becomes increasingly implausible as the number of convergent signatures grows, but cannot be definitively excluded without independent check through non-GNSS methods (the SLR provenance check, the orbit-projection forward model, and the triangle holonomy test).
**Path to Resolution:** Definitive exclusion of this hypothesis requires: (1) SLR check (Section 6.3.1.1) to close the GNSS-product branch of the ephemeris loophole, with the generic orbit-projection branch closed by the physical-confound forward model or an orbit-free clock channel; (2) independently processed per-constellation cross-check recovering the same finite covariance family and band ordering across GPS, GLONASS, Galileo, and BeiDou while allowing product- and estimator-dependent scales; and (3) triangle holonomy test (Section 6.3.3) providing direct measurement of synchronization non-closure. Until these tests are completed, the possibility of an unrecognized GNSS systematic, however implausible given the convergence of evidence, cannot be completely excluded.
## 3.7 SLR Check: Optical Sensitivity and Sampling Controls
Satellite Laser Ranging (SLR) supplies an optical test through passive retroreflectors and an orbit-determination chain that consumes no GNSS products. Under conformal null-cone invariance the ground event timer reads the clock-amplitude channel. This instrument and product independence makes SLR a useful test pathway; a common physical signal must still be established at adequate sensitivity.
Paper 8 analyses 11 years (2015–2025) under sampling-matched controls. The observed 10–500 µHz spectral ratio of 14.12 is reproduced by white noise passed through the same sparse-sampling and resampling pipeline (13.97; Fisher-combined p = 0.199). A real-epoch Lomb–Scargle estimator gives 1.04 against a white-noise null of 1.05. The earlier spectral concentration therefore supplies no optical detection. The clock-amplitude channel is bounded at approximately δA ≲ 2×10−8; the roughly 5 mm excursion from the surface-depth benchmark remains below the residual floor.
The nominal same-satellite inter-station result (Fisher p = 0.0025) shares an orbit-model common mode. Different-satellite pairings yield a candidate anticorrelation in the 5,000–7,500 km bin (mean r = −0.23; label-swap p = 0.046; station-clustered 95% interval [−0.43, −0.06]). It appears in only one of nine threshold/bin-width configurations, weakens to −0.054 with at least five shared bins, and changes sign across years. It is retained as an exploratory tail event. The daily-aggregation diagnostic also reverses sign between orbit solutions.
The present SLR contribution is a sampling-artifact correction, an optical sensitivity bound and a defined route to an independent instrument test. It establishes neither achromatic replication of the GNSS signal nor its absence at the smaller benchmark amplitude. A stronger cross-instrument inference requires stable inter-station recovery and a demonstrated signal-amplitude transfer.
## 3.8 Why the Signal Was Not Previously Identified
The correlation structure documented here has been present in GNSS data since the network's inception but was systematically removed by processing algorithms optimized for positioning accuracy rather than fundamental physics characterization. Three factors explain why it was not identified earlier.
Standard GNSS processing is designed to eliminate correlations, not to characterize them. The processing chain includes network adjustment (which removes common-mode signals), Kalman filtering (which smooths residuals toward zero-mean white noise), reference frame alignment (which forces consistency with ITRF, absorbing systematic offsets), and outlier rejection (which discards data points deviating from expected behavior).
The correlation structure detected in this analysis is precisely what these algorithms are designed to remove. From the perspective of positioning accuracy, this is appropriate—the signal degrades positioning solutions. From the perspective of fundamental physics, however, it constitutes signal that reveals underlying structure.
The present analysis examines the residuals that standard processing discards, employing phase-coherent cross-spectral methods that preserve correlation structure. The signal was present throughout the history of GNSS operation but was categorized as systematic error rather than physical phenomenon.
## 3.9 Summary of Excluded Hypotheses
Table 3 summarizes the artifact hypotheses tested and their outcomes.
| Hypothesis | Test | Result |
| --- | --- | --- |
| Processing artifact | Raw RINEX/SPP | Signal persists (all combinations) |
| Ionospheric | 25-year solar cycle | No correlation (r ≈ 0) |
| Tropospheric | Altitude stratification | No altitude dependence (Q5/Q1 = 0.97) |
| Geomagnetic | Kp stratification | Near-invariant (Δλ ≈ −1%) |
| Tidal forcing | GM/r² and GM/r³ scaling | Neither force nor tidal scaling (p > 0.5) |
| Random noise | Shuffle test | 30× R² ratio (real vs. shuffled) |
| Center-specific bias | Multi-center consistency | Cross-centre CV ≈ 13% on the exponential λT triplet (§2.1) |
The tests are complementary but not statistically independent; several are derived from the same underlying GNSS covariance data. The signal survives all challenges, constraining viable explanations to physical mechanisms operating at planetary scales with velocity-dependent, preferred-axis geometry.
## 3.10 Evidence Tier Classification
The convergent signatures supporting the Temporal Topology interpretation
span two tiers of timing-network evidence plus an orthogonal optical-domain
check, reflecting both statistical strength and resilience to alternative
explanations.
| Tier | Signature | Statistical Strength | Artifact Resilience |
| --- | --- | --- | --- |
| **Tier B**
(Primary timing-network evidence) | Exponential spatial decay | R² = 0.92–0.97 | Survives all processing modes; raw RINEX confirmed |
| | East-West/North-South anisotropy | Ratio 2.16, p $\lt$ 10−15 | Cross-center validated; multi-constellation |
| | Orbital velocity coupling | r = −0.888, 3.65σ (autocorr.-corrected) | Phase-stable over 25 years; multi-constellation |
| | Preferred-direction recovery | annual phase near aphelion tangent on CODE and RINEX; CMB apex anti-phase; Solar Apex null; MGEX scan projection-degenerate | orbit-coupled axis; direction product-unstable |
| | Semiannual coherence component | R² = 0.904, p = 2.7 × 10−5 (F-test) | 50+ complete cycles; apsidal-locked (nutation/orbital harmonic) |
| | Planetary event catalogue (bounded channel) | 56/156 events ≥2σ (descriptive); opposition subset null-validated 5.5×; continuous r = +0.116, p = 0.012 | GM/r² and GM/r³ scaling null bounds unmodelled transient perturbations; catalogue excess reproduced by calendar position (Paper 3) |
| | 18.6-year lunar nutation | R² = 0.641 ± 0.10, p = 0.010 (surrogate-validated) | 1.36 complete cycles; cycle-limited, longer baseline needed |
| | Selectivity profile (null results) | Solar rotation: null; Tides: null | Demonstrates physical specificity |
| **Tier C**
(Optical-domain consistency check) | SLR spectral bound; exploratory correlations | 14.12 observed versus 13.97 sampling-matched white null | Instrument-orthogonal; passive retroreflectors; same conformal clock channel |
Tier B signatures constitute the foundational timing-network evidence: they are
detected with high statistical confidence across all GNSS test modes and
processing pipelines. Tier C provides an instrument-level optical-domain check
using an independent measurement technology and orbit-determination chain; it is consistent with the
timing-network results but probes the same conformal clock sector rather
than a second physical channel, and does not carry the same statistical weight owing
to sparser station coverage and asymmetric global distribution. The pattern
across both tiers consistently favors a velocity-dependent, preferred-axis
geometric mechanism over conventional explanations.
## 4. Theoretical Framework
Having established that the signal is robust and not attributable to known
artifacts, this section presents the theoretical framework that predicted
these observations. The Temporal Equivalence Principle is a principled
extension of general relativity within the well-established class of
scalar-tensor theories, with pre-specified expectations and theoretical search ranges that preceded
the GNSS analysis.
## 4.1 Two-Metric Geometry
TEP posits that spacetime is endowed with two metrics related by a scalar
field φ:
$$\tilde{g}_{\mu\nu} = A^2(\phi)\, g_{\mu\nu} + B(\phi)\, \nabla_\mu \phi\,
\nabla_\nu \phi$$
The gravitational metric gμν determines geodesics of test masses
and gravitational wave propagation. The matter metric g̃μν
determines atomic clock rates, electromagnetic propagation, and all Standard
Model physics. The conformal factor A(ϕ) = exp(βAϕ/MPl) provides
isotropic rescaling of proper time, while the disformal factor B(φ)
introduces anisotropic light-cone tilts and is tightly constrained by
GW170817.
This structure belongs to the well-established class of scalar-tensor
theories developed by Bekenstein (1993) and Damour & Polyakov (1994). The
contribution of TEP lies in its operational interpretation: the conformal
factor A(ϕ) governs the rate at which proper time accumulates, rendering
time itself a dynamical field.
## 4.2 Physical Interpretation
The framework is formulated as a scalar-tensor theory in the Einstein frame
where gravity remains canonical. The scalar field φ couples to matter
through the conformal factor A(φ), modulating the rate of proper time
accumulation. The bare conformal coupling is $\beta_A = -1.0$ universally. In the near-Earth clock sector, $S_A^{(\oplus)} = \min[1, (\bar{\rho}_\oplus/\rho_T)^{1/3}] \approx 0.65$ characterizes static conformal depth, while the measured time-dependent covariance $C_A$ has a distinct environmental and processing transfer; the local shear sector is pairwise-screened by the kinetic operator (Paper 0 §2.2). The scalar field obeys a Klein-Gordon equation
with source terms from matter density and a screening potential V(φ) that,
in concert with derivative self-interactions, suppresses Temporal Shear in
high-acceleration regions through the kinetic operator.
(Full action and field equations are provided in Appendix A.)
## 4.3 Foundational Axioms
The framework rests on four axioms. The first (two-metric structure) states
that gravity is described by gμν while matter couples universally
to g̃μν. The second (temporal equivalence) states that all
non-gravitational processes evolve according to proper time dτ defined by
g̃μν, and that in local freely falling frames, physics reduces to
special relativity with invariant c. The third (causal safety) requires that
cg = cγ within GW170817 bounds (|cg −
cγ|/c ≲ 10−15), constraining the disformal coupling B(φ) to be small at
late times but not zero — disformal or otherwise non-exact transport is required for the synchronization holonomy observable (Section 4.6). The fourth (screening) states that rather than operating via
discrete boundary cutoffs, screening manifests as a continuous spatial
profile (Temporal Topology), with Temporal Shear suppression arising from
the non-linear superposition of field gradients (Temporal Shear), suppressing
Temporal Shear in high-acceleration regions while leaving cosmology accessible to dynamics.
## 4.4 Candidate Theoretical Completions
The observed correlation length λT is the Temporal Topology
correlation length—the characteristic scale of the clock-sector covariance $C_A$ in Earth's environment ($\lambda_T \approx R_T$ in the channel functional; Paper 0 §7). This empirical parameter admits
multiple theoretical interpretations as candidate completions, not as unique
predictions. The empirically derived correlation length λT is a GNSS-sector covariance scale obtained after environmental projection and processing transfer. It is not the screening operator $\mathcal{S}_\Sigma(\mathcal{E})$ itself, and it is not the non-exact (disformal) covariance term, which this measurement does not access.
Three classes of completion are sketched below.
### 4.4.1 Compton-Mass Interpretation
A chameleon-type completion identifies λT with the in-medium Compton wavelength of a scalar field: λ = ℏ/(mφc). For λ = 3,330–4,549 km, this yields mφ ≈ (4.34–5.93) × 10−14 eV/c². The present framework identifies the correlation length with the geometric saturation radius RT (§4.4.2) rather than with a Compton wavelength; the mass scale is retained as a cross-reference to chameleon-literature conventions.
### 4.4.2 Clock Amplitude Calibration: TEP Saturation Radius
In the TEP framework, the geometric saturation radius RT(M) = (3M/4πρT)1/3 enters the clock sector to set the unsaturated clock-amplitude factor SA = min[1, (ρ̄/ρT)1/3], while Earth-vicinity Temporal Shear is pairwise-screened in the force sector by the kinetic operator (Paper 0 §2.2, §7). The saturation density ρT ≃ 20 g/cm3 is calibrated from Earth's observed GNSS clock covariance:
$$R_T = \left(\frac{3M}{4\pi\rho_T}\right)^{1/3}$$
This M1/3 scaling is a kinematic property of the constant-ρT saturation model. For Earth with M = M⊕ ≈ 6 × 1024 kg and ρT ≃ 20 g/cm3:
$$R_T^{\oplus} = \left(\frac{3 \times 6 \times 10^{24}\text{ kg}}{4\pi \times 2 \times 10^{4}\text{ kg/m}^3}\right)^{1/3} \approx 4{,}150\text{ km}$$
This is consistent in scale with the observed GNSS correlation-length family (azimuth-averaged λ = 3,210 ± 389 km; cross-sector dispersion λT = 4,201 ± 1,967 km), noting that the identification calibrates ρT rather than testing an independently derived scale. The physical roles are partitioned as follows:
**Clock sector (SA):** Governed by the potential sector and local density, returning SA(⊕) ≃ 0.65 at Earth's mean density ρ̄⊕ = 5.515 g/cm3. Clock residuals across continental baselines measure the spatial covariance CA(r); the transmission of time-dependent fluctuations and their processing transfer are distinct from the static SA level.
**Shear / Force sector (SΣ):** Governed by the selected two-branch kinetic completion P(X,φ) = −V(φ) + (2k/3ΛX2)X|X|1/2 + X|X|/ΛX4 (k ≈ 16, derived). In high-acceleration regions (g ≫ gt = cH₀/2), the two-body operator suppresses the observable shear response to 𝒮Σ ~ 10−21 at Earth's surface, preserving standard General Relativity paths in local gravitational and planetary tests.
**Kinetic transition scale (r*):** The physical transition for dynamical shear recovery follows the kinetic law r*(M) = √(GM/gt) ∝ M1/2 (Paper 0 §2.2), distinct from the geometric saturation radius RT ∝ M1/3.
**Falsifiable Predictions:** The density-saturation diagnostic
RT(M) = (3M/4πρT)1/3 — a candidate
microscopic completion, not a derived consequence of the master action —
makes specific predictions for intermediate-mass systems:
**Jupiter** (M ~ 300 M⊕): Predicted RT
~ (300)1/3 × 4,150 km ≈ 27,800 km. Testable with Juno mission
clock data or future Jupiter atmospheric probe networks.
**Globular clusters** (M ~ 105 M☉):
Predicted screening length ~ 1–10 pc, testable through pulsar timing
array correlations within clusters.
**Galaxies** (M ~ 1012 M☉): At galactic
scales, the screening operator 𝒮Σ(Ε) retains
approximately 98% of the unsuppressed Temporal Shear in the diffuse halo
environment (Paper 26), producing dark-matter-like phenomenology through
the active-shear regime rather than through a screening radius. The
galactic phenomenology is therefore a manifestation of the same
𝒮Σ operator operating in the unscreened regime, not
a Vainshtein scaling of the terrestrial correlation length. This
connection is developed in Section 5.5.
Within the saturation-model regime — from planetary through intermediate
mass scales — RT(M) provides a unified characterization governed
by a single saturation density ρT; the testable prediction is
that one ρT fits all such masses, while galactic-scale
phenomenology arises from the same 𝒮Σ operator in the
unscreened regime rather than from a screening radius. This scaling reproduces the continuous gradient-suppression behavior described by
Temporal Topology at the macroscopic level. Candidate microscopic completions (chameleon, Vainshtein, Galileon, DBI, symmetron) may produce similar phenomenology through different microscopic mechanisms; the empirical correlation length
λT = 4,201 ± 1,967 km does not uniquely select any single completion. The defining TEP ontology is the continuous suppression of Temporal Shear by the 𝒮Σ(Ε) operator, not the specific microscopic screening mechanism.
## 4.5 Synchronization Holonomy and Spatial Correlations
TEP predicts synchronization holonomy—the non-closure of proper time when
transported around a closed loop:
$$\mathcal{H} \equiv \oint_C d\tau_{\text{prop}} \neq 0$$
In standard general relativity, after subtracting known effects (Sagnac,
Shapiro, gravitational redshift), this holonomy vanishes. In TEP with
disformal coupling B ≠ 0, it does not.
An important subtlety arises in the conformal-only limit (B = 0): the
time-transport one-form Ωμ = ∂μ(½ ln A) is exact, its
curl vanishes, and holonomy is zero. Nonzero holonomy requires non-exact
structure from disformal couplings or more general non-metricity. This
raises a question: if GW170817 constrains B(φ) to be small, and
conformal coupling alone produces no holonomy, what generates the observed
GNSS correlations?
The resolution lies in distinguishing two observables. Synchronization
holonomy (H) represents the non-closure of proper time around closed loops;
this requires disformal or otherwise non-exact transport and is the target of proposed triangle tests. Spatial
correlation structure, by contrast, represents the statistical covariance of
clock residuals as a function of distance; this arises from the spatial
structure of the φ field itself—specifically, the two-point correlation
function ⟨φ(x)φ(x+r)⟩ ∝ exp(−r/λ)—and persists even when B = 0.
The GNSS signal reflects the latter mechanism: clocks at different locations
sample different values of the conformal factor A(φ), and spatial
correlations in φ induce correlations in clock rates. The exponential decay
with distance reflects the TEP saturation radius $R_T = (3M/4\pi\rho_T)^{1/3}$,
while the velocity-dependent anisotropy arises because Earth's motion
through the φ field creates a preferred direction in the correlation
structure. The correlation structure is thus a statistical signature of the
φ field's spatial configuration, whereas holonomy is a geometric signature
of non-integrable time transport. Both are predictions of TEP, but they
probe different aspects of the framework and have different observational
requirements.
### 4.5.1 Resolving the Conformal Paradox
**Figure 1:** Synchronization Holonomy: The Definitive Test
of Non-Integrable Proper Time. Three atomic clocks (A, B, C) positioned
at vertices of a triangle (1,000–3,000 km separation) exchange one-way
time-transfer pulses around the loop. Panel A (Standard GR): After
subtracting all known relativistic effects (Sagnac, Shapiro,
gravitational redshift, Lense-Thirring), the accumulated time difference
returns to zero—proper time is integrable. Panel B (TEP): A residual
time gap ℋresid ~ 10−18–10−16 s
remains, scaling as (B/A²)(u·∇φ) Pμν ∇νφ · Area. Target sensitivity: σ(ℋ)/Tloop ≤ 10−19.
A critical objection must be addressed: if TEP employs only a conformal
coupling A(φ), how can it produce synchronization holonomy? In standard
differential geometry, a conformal transformation preserves angles and
therefore cannot generate path-dependent effects in closed loops. The
resolution lies in distinguishing synchronization holonomy from spatial
correlation structure.
**Observable 1: Synchronization Holonomy (ℋ)** — The
non-closure of proper time around closed loops, defined as ℋ ≡ ∮C
dτprop. This observable requires disformal or otherwise non-exact transport
because only non-exact structure in the time-transport connection generates
path-dependent accumulation. In the conformal-only limit (B = 0), the
one-form Ωμ = ∂μ(½ ln A) is exact, its curl vanishes
identically, and holonomy is zero by Stokes' theorem. The triangle holonomy test (Section 6.3.3) targets this observable and would measure whether B(φ) is nonzero at the 10−19 fractional level.
**Observable 2: Spatial Correlation Structure (C(r))** — The
statistical covariance of clock frequency residuals as a function of
inter-station distance. This observable requires only A(φ) ≠ 1 (conformal
coupling) and persists even when B = 0. The mechanism is fundamentally
different from holonomy: clocks at different spatial locations sample
different values of the scalar field φ(x), and the two-point correlation
function ⟨φ(x)φ(x+r)⟩ ∝ exp(−r/λ) of the underlying field induces
correlations in clock rates through the conformal factor A(ϕ) =
exp(βAϕ/MPl). The exponential decay length λ = RT(M⊕)
reflects the TEP screening scale of the temporal topology (temporal-topology pinning on the environmental state, of which density is one coordinate); the terrestrial correlation length is identified with the geometric saturation radius R_T(M_⊕) from the UCD model (Paper 6), not with the in-medium Compton wavelength of a candidate chameleon completion.
**Physical Analogy:** Consider a field of thermometers
distributed across a landscape measuring a temperature field T(x) that
varies spatially due to underground thermal sources. Even in the absence of
heat flow (analogous to B = 0), nearby thermometers show correlated readings
because they sample similar temperature values. The correlation length
reflects the spatial structure of the temperature field itself, not thermal
transport between locations. Similarly, GNSS clocks sample the φ-field
"temperature," and their correlation structure reveals the field's spatial
configuration.
**Why GW170817 Does Not Constrain the GNSS Signal:** The
multi-messenger constraint |cγ − cg|/c ≲ 10−15
bounds the disformal sector B(φ), which affects propagation speeds. The GNSS
correlation structure arises from the conformal sector A(φ), which affects
clock rates but preserves null cones. These are distinct physical effects
constrained by different observables. GW170817 establishes that light and
gravity travel at the same speed; it does not constrain whether clocks at
different locations tick at correlated rates due to sampling a common scalar
field.
**Observational Distinction:** The GNSS signal is a
"thermometer correlation"—a statistical signature of φ-field structure
detected through spatial covariance analysis. Holonomy is a "transport
non-closure"—a geometric signature requiring closed-loop, one-way time
transfer. Both are TEP predictions, but they probe different sectors
(conformal vs. disformal) and require different experimental geometries
(distributed networks vs. closed triangles). The present work establishes
the former; the latter awaits dedicated triangle tests.
## 4.6 Prediction Comparison
TEP established pre-specified expectations and theoretical search ranges
before the GNSS analysis was conducted: the framework was first deposited in
August 2025 (Smawfield 2025a, v0.1) and the clock-network correlation
forecast was versioned on 15 September 2025 (v0.4 §E), preceding the first
public deposit of the GNSS analysis by one week (provenance register,
Section 6.5). Table 5 compares these forecasts to
observations.
| Prediction | Forecast | Observed | Status |
| --- | --- | --- | --- |
| Correlation length (screening-parameterized forecast) | 2,000–3,000 km | 3,210 ± 389 km (azimuth-averaged); IGS-final 3,330–4,549 km | Not met: observed scale lies above the band |
| Correlation length (search range) | 1,000–10,000 km | 4,201 ± 1,967 km (sector dispersion) | Consistent; non-discriminating (range subsumes leading systematic scales) |
| Spatial decay form | Exponential; Gaussian-family preference would disfavour the screened-field reading (Paper 1, criterion iii) | Gaussian/squared-exponential preferred; exponential ΔAIC = +12.8; power law ΔAIC > 30 | Met in Paper 1; not met in the 25-year dataset |
| Multi-centre consistency | $\lt$ 5% variation in fitted parameters | CV of λ ≈ 13% across three centre exponential fits | Not met |
| Velocity-dependent anisotropy | Modulates with v/c | r = −0.888 (3.65σ, autocorr.-corrected) | Consistent in sign (pre-stated in Paper 1 v0.6α); coupling magnitude not derived in the current framework |
| Geometric alignment | EW > NS in ecliptic frame | EW/NS = 2.16 | Consistent |
| Broadband coupling | Not frequency-selective | Detected across 10–500 μHz | Consistent |
| Annual-phase direction recovery | Orbital-phase modulation | RA = 6°, Dec = +4°; ~6° from aphelion velocity tangent; CMB apex anti-phase | Consistent with orbital-phase channel; no CMB-frame identification |
| GM/r² and GM/r³ scaling | Both absent (filtered) | No force or tidal scaling (p > 0.5) | Null confirmed |
The null prediction regarding GM/r² scaling is particularly discriminating.
Classical tidal forcing would produce mass-dependent signatures scaling with
GM/r³, with Mercury contributing far less than Jupiter. The observed absence
of such scaling (all p > 0.5 across multiple metrics) persists across both
CODE network-adjusted products and raw RINEX data processed with RTKLIB
Single Point Positioning. Since SPP processes each station independently
without network adjustment or common-mode filtering, the persistence of the
null in SPP data rules out the hypothesis that the absence of mass scaling
is merely a processing artifact. The TEP coupling appears to be
intrinsically geometric rather than energetic—responding to alignment
configuration regardless of mass. Mercury (42.5% detection rate) produces
comparable responses to Jupiter (34.8%) despite ~106× smaller
gravitational influence, consistent with a threshold-dependent or
alignment-driven mechanism rather than continuous force scaling.
## 4.7 Why Existing Tests Do Not Directly Test the Target TEP Observables
Existing precision tests probe different observables than TEP's primary
signatures, as summarized in Table 6.
| Test | Observable | Reason TEP Is Unconstrained |
| --- | --- | --- |
| GW170817 | |cγ − cg|/c | Constrains disformal B(φ), not conformal A(φ) |
| Cassini (PPN-γ) | Two-way Shapiro delay | Reciprocity-even; does not bound loop holonomy; samples screened solar-vicinity regime |
| Resonator MM/KT | Closed-path anisotropy | Even-parity; blind to one-way non-reciprocity |
| GPS operation | Internal consistency | Uses explicit GR+Sagnac modeling; verifies model |
| Clock redshift tests | Pairwise comparisons | Confirms GR locally; closed loops required |
The critical distinction is that existing tests probe two-way or pairwise
measurements, whereas TEP's holonomy is a closed-loop, one-way observable to
which these tests are structurally blind. Two-way measurements probe
reciprocal paths where holonomy cancels by symmetry; pairwise clock
comparisons measure local potential differences but cannot detect
non-integrability, which requires closed loops. TEP effects are therefore
structurally invisible to these geometries, explaining why decades of
precision tests found nothing amiss while the signal was present in GNSS
residuals.
#### Operational Measurement Protocol: What Existing Tests Probe
The distinction between observables that constrain TEP and those that do
not can be formalized through the synchronization holonomy (after GR
subtraction), where σ is the time-transport one-form whose line integral
equals the calibrated proper-time increment along a leg:
$$\mathcal{H}_{\rm resid} = \oint_{\partial\Sigma} (\tilde{\sigma} -
\sigma_{\rm GR}) = \iint_{\Sigma} (\tilde{F} - F_{\rm GR})$$
This represents the loop non-closure of time transport built from
measured proper times (in time units; right-hand rule on Σ). It is gauge
and synchronization-invariant and vanishes in standard relativity and in
the conformal-only limit of TEP. GR subtraction includes Sagnac,
Lense-Thirring gravito-magnetic effects, Shapiro delay, and
gravitational redshift, with ITRF ephemerides and TT/TDB standards.
How flagship constraints map to ℋresid:
GW170817 (GW-EM coincidence): |cγ − cg|/c ≲
10−15 constrains global cone splits. In TEP, late-time
conformal coupling preserves null cones, so electromagnetic and
gravitational waves share causal structure; small disformal tilts
today are allowed. This is a boundary condition, not a loop-holonomy
test.
Cassini (PPN-γ): Two-way Doppler/Shapiro is reciprocity-even; it
calibrates σGR to subtract but does not bound
ℋresid. The measurement was performed during solar
conjunction, sampling the Sun's deep potential well—the
screened regime where the source-charge sector $S_\Sigma$ is
suppressed. The bound therefore applies to the solar-vicinity
environment, not to the dilute regimes where Temporal Shear
recovers.
Resonator MM/KT tests: Cavities bound closed-path, even-parity
(two-way sums) anisotropy at 10−17–10−18, yet
are blind to odd-parity (direction-reversing one-way differences)
non-reciprocity and loop non-closure—the ingredients of
ℋresid.
GPS operation: Network self-consistency uses explicit GR+Sagnac
modeling and largely two-way/common-view calibration. This verifies
internal consistency under the assumed GR model; it is not a
direction-reversing one-way loop-closure null.
Clock redshift and pairwise A↔B tests: Exquisitely confirm GR
locally; only closed loops (A→B→C→A with direction reversal) can
reveal non-integrability captured by ℋresid.
Why classical tests can be null while ℋresid ≠ 0: (1)
Conformal null-cone invariance implies no large GW-EM kinematic delays
(consistent with GW170817). (2) If ∂tφ = 0 over loop
timescale and gradients are conservative, there is no first-order
one-way anisotropy. Along a fixed path, forward and backward times
cancel at order (B,∇φ); leading effects are second order or require time
dependence and kinematics. Thus two-way and closed-path nulls can hold
while a loop-holonomy test remains sensitive.
Experimental falsifier (primary endpoints): Run a closed-loop, one-way
time-transfer (and/or portable-clock) test and report: (1) Leg-wise
antisymmetry: ΔtAB = tAB − tBA (and
optionally ΞAB ≡ (tAB − tBA)/(tAB
+ tBA)). (2) Loop holonomy ℋresid after
subtracting GR (Sagnac/gravito-magnetic/Shapiro). Use
triangle/quadrilateral geometries with direction reversal; extend with
interplanetary one-way links and multi-species clock networks.
### 4.7.1 Compatibility with Binary Pulsar Timing and Lunar Laser Ranging
Two high-precision tests warrant explicit discussion: binary pulsar timing
(PSR J0737-3039, testing GR to 0.05%) and Lunar Laser Ranging (testing the
equivalence principle to 10−13). Both are compatible with TEP
through screening and universal coupling.
Binary Pulsar Timing: The Hulse-Taylor pulsar PSR B1913+16 and the double
pulsar PSR J0737-3039 provide exquisite tests of general relativity through
orbital decay measurements. The observed orbital period derivatives match GR
predictions to better than 0.05%, constraining deviations in the
strong-field regime. TEP compatibility requires that scalar field effects
are screened in the vicinity of neutron stars.
The TEP saturation radius for a neutron star (M ~ 1.4 M☉)
is:
$$R_T^{\text{NS}} = \left(\frac{3 M_{\text{NS}}}{4\pi\rho_T}\right)^{1/3}
\sim 3.2 \times 10^5 \text{ km}$$
This saturation radius far exceeds the neutron star radius (RNS ~
10 km). Binary pulsar timing, however, probes the gravitational sector (metric gμν), not the clock sector (metric g̃μν). The scalar field's effect on gμν is second-order in the coupling, proportional to (∇φ)²/c⁴, which at orbital distances (a ~ 106 km) from a 1.4 M☉ neutron star gives a fractional effect below 10−30 — far below the 0.05% observational
precision of binary pulsar tests. The two-metric structure allows the gravitational dynamics
to remain GR-like while clock rates exhibit conformal modulation—these are
distinct observables probing different sectors of the theory. Binary pulsar
insensitivity to the conformal coupling that generates GNSS correlations is
therefore a consequence of the two-metric structure, not of the screening
radius.
Critically, pulsar timing measures orbital dynamics (gravitational sector,
metric gμν), not clock correlations (matter sector, metric
g̃μν). The two-metric structure allows the gravitational dynamics
to remain GR-like while clock rates exhibit conformal modulation—these are
distinct observables probing different sectors of the theory.
Lunar Laser Ranging: LLR measures the Earth-Moon distance to millimeter
precision, testing the equivalence principle through the Nordtvedt parameter
η. The current bound η $\lt$ 4.4 × 10−4 constrains deviations from
the strong equivalence principle. TEP satisfies this constraint through
universal coupling: the conformal factor A(φ) couples identically to all
matter fields, preserving the equivalence principle in the matter frame.
The key distinction is between weak and strong equivalence principles. The
weak EP (universality of free fall for test masses) is preserved exactly in
TEP because all matter couples to the same metric g̃μν. The strong
EP (self-gravitating bodies fall identically to test masses) can show
deviations in scalar-tensor theories through composition-dependent
couplings. However, TEP's universal conformal coupling ensures that the
Nordtvedt parameter:
$$\eta = 4\beta_{\text{PPN}} - \gamma - 3$$
remains within observational bounds. The PPN nonlinearity parameter βPPN is driven to its GR value βPPN → 1 by the TEP screening operator, where the pairwise response is suppressed by the kinetic operator ($S_\Sigma(g_{\text{lunar}}) \approx 1.6 \times 10^{-14}$; Paper 0 §2.2). With γ ≈ 1 (Cassini constraint), the predicted Nordtvedt parameter is well below the LLR bound. The predicted metric deviation applies to the gravitational sector, preserving the Strong Equivalence Principle. The ∼10−4 synodic-phase anomaly reported in Paper 17 is an empirical matter-frame clock residual (κLLR), not a classical Nordtvedt metric violation.
Furthermore, the Earth–Moon separation lies in the deeply suppressed
regime of the pairwise Temporal-Topology response
𝒮eff(s): the scalar-mediated Earth–Moon differential
acceleration is suppressed by pairwise response factors of order
2.5×10−14 relative to the unsuppressed coupling (Paper 0,
screening operator), comfortably below the LLR bound. The suppression
acts on the shear sector (𝒮Σ) — the geodesic and
force projection — while each body's clock-amplitude coupling
(SA) remains unsuppressed. The combination of universal
coupling and sector-separated screening ensures LLR compatibility
while preserving the GNSS clock-covariance signal in the terrestrial
environment.
## 4.8 Relation to Established Scalar-Tensor Theory
In the canonical non-screened limit, TEP shares structural similarities with scalar-tensor theories, but its physical realization is defined by the non-canonical two-branch kinetic sector $P(X,\phi) = -V(\phi) + \frac{2k}{3\Lambda_X^2}X|X|^{1/2} + X|X|/\Lambda_X^4$ and universal matter-metric coupling.
In the limit B = 0, V = 0, ΛX → ∞, TEP with A(ϕ) = exp(βAϕ/MPl) reduces to Brans-Dicke
gravity in the Jordan frame, with the coupling βA related to the Brans-Dicke
parameter ωBD. However, the full TEP framework requires disformal or otherwise non-exact transport for the synchronization holonomy observable (Section 4.6); the B = 0 case is a limiting benchmark, not the operative theory. In the master
TEP framework, screening is realized through the non-canonical kinetic operator, separating the conformal clock response SA from the environmental source-charge projection SΣ (Paper 0 §2.2). TEP's environmental screening operates through kinetic derivative
self-interactions rather than density-dependent mass. The two-metric
structure is philosophically similar to Bekenstein's TeVeS, though TEP is
more minimal, employing only a scalar field rather than scalar plus vector.
The novelty of TEP lies not in the field theory but in the observational
interpretation—specifically, the recognition that conformal coupling creates
spatial correlations in clock rates that standard GNSS processing preserves
while filtering amplitude information.
## 4.9 The Processing Filter Mechanism
A critical insight explains both why the signal was missed for three decades
and why it survives in the data. Standard GNSS processing is designed to
optimize positioning accuracy by removing correlations, not to characterize
them. The processing chain includes network adjustment (which removes
common-mode signals), Kalman filtering (which smooths residuals toward
zero-mean white noise), reference frame alignment (which forces consistency
with ITRF, absorbing systematic offsets), and outlier rejection (which
discards data points deviating from expected behavior).
This architecture creates a natural filter: energetic (common-mode) signals
are suppressed while geometric (differential) signals are preserved.
Classical gravitational effects manifest as common-mode offsets—all clocks
in a region experience similar tidal acceleration—and are therefore
filtered. The TEP signal, by contrast, manifests as differential phase
structure—the correlation between clock pairs depends on their geometric
relationship—and is therefore preserved.
This mechanism implies: GM/r² scaling should be absent in processed data
because it is common-mode, while geometric signatures (anisotropy, preferred-direction
structure) should survive because they are differential. The observations
are consistent with both expectations. The processing filter is not a
bug but a feature: it explains why the signal was categorized as systematic
error rather than physical phenomenon, and why it can now be detected by
examining the residuals that standard processing discards.
### 4.9.1 Mathematical Demonstration of Filter Selectivity
The processing filter mechanism can be demonstrated mathematically by
decomposing clock residuals into common-mode and differential components.
Consider a network of N stations with clock residuals δti(t)
after initial processing. Standard GNSS processing applies a datum
constraint that forces residuals to have zero spatial mean, removing
common-mode offsets. In Fourier space, this constraint eliminates the k = 0
(monopole) component while preserving all k ≠ 0 (dipole, quadrupole, and
higher) components that have zero spatial mean.
Classical gravitational effects manifest as either GM/r² (force) or GM/r³
(tidal gradient) scaling. Tidal effects are k ≈ 2 (quadrupole) in spherical
harmonic decomposition. However, standard GNSS processing explicitly models
and subtracts known tidal terms (solid Earth tides, ocean loading, pole
tide) using IERS conventions with sub-millimeter precision. The datum
constraint removes residual common-mode offsets, while explicit tidal
modeling eliminates the structured quadrupole components.
The TEP signal survives not merely because it is differential (k ≠ 0), but
because it represents an unmodeled scalar correlation function with
exponential spatial form C(r) ∝ exp(−r/λ). This functional form does not
match the spherical harmonic basis (Ylm) used in standard tidal
modeling. Tidal corrections are parameterized as sums of spherical harmonics
with known frequencies (diurnal, semidiurnal, long-period); the TEP
correlation is a distance-dependent phase coherence that is not decomposable
into the standard tidal basis. The processing removes what it models (tidal
harmonics) but preserves what it does not model (exponential distance
correlations in phase space).
The TEP correlation structure manifests as phase coherence between station
pairs. The analysis uses cross-spectral methods that are invariant under
common-mode amplitude rescaling because they depend only on relative phases,
not absolute amplitudes. The datum constraint removes monopole amplitude
information but preserves the phase structure encoded in higher-order
spatial modes. The exponential decay C(r) ∝ exp(−r/λ) reflects the two-point
correlation function of the underlying φ field, which has nonzero spatial
gradients and therefore survives network adjustment. (Detailed mathematical
formulation is provided in Appendix B.)
**Expectation Verification:** This mathematical framework yields
two testable expectations, both consistent with observations:
**Null Prediction:** Classical gravitational effects (GM/r²
force and GM/r³ tidal gradient) should be absent in processed data
because they are predominantly k = 0 and k ≈ 0 (long-wavelength) modes
that the datum constraint removes. Observed: Neither GM/r² nor GM/r³
scaling detected (all p > 0.5).
**Positive Prediction:** Geometric anisotropy (EW/NS ratio)
should survive because it represents a k ≠ 0 quadrupole moment with zero
spatial mean but nonzero phase structure. Observed: EW/NS = 2.16
detected at p $\lt$ 10−15.
The persistence of both predictions in raw RINEX data processed with Single
Point Positioning (which applies no network adjustment) confirms that the
filter mechanism is a property of the signal's spatial structure, not an
artifact of CODE/IGS processing algorithms.
## 4.10 Systematic Error Budget for Triangle Test
For the proposed triangle holonomy test to achieve definitive
discrimination, systematic errors must be controlled below the target
sensitivity of 10−19 fractional. Table 7 presents the error
budget.
| Systematic | Mitigation | Residual |
| --- | --- | --- |
| Sagnac (Earth rotation) | IERS Earth orientation parameters | ≤ 10−21 |
| Shapiro (GR potentials) | Precise ephemerides + atmospheric models | ≤ 3 × 10−20 |
| Troposphere/ionosphere | Dual-frequency + water vapor radiometry | ≤ 5 × 10−20 |
| Fiber dispersion | Bidirectional calibration + thermal control | ≤ 5 × 10−20 |
| Link noise + clock instability | Optical clocks + day averaging | ≤ 3 × 10−19 |
| Total budget | | ≤ 1 × 10−19 |
For parameters saturating current bounds, the predicted signal is H/Tloop
~ 10−16–10−18. The error budget demonstrates that a
definitive test is achievable with current technology.
## 5. Consistency with Cosmological Phenomenology
#### Epistemic Status
The following section does not constitute an independent detection of
cosmological effects. Instead, it addresses the question: if the
terrestrial clock correlations reported in Sections 2–4 arise from a
conformally coupled scalar field (whether screened via continuous gradient flattening
or derivative completions), then what are the necessary
consequences for astrophysical observations that already exist?
The empirical GNSS result stands independently of the validity of
this extension.
This section examines how the same scalar field producing λ ≈ 4,000
km correlations on Earth bears on gravitational lensing observables.
Under isotropic conformal rescaling $\tilde g_{\mu\nu}=A^2(\phi)g_{\mu\nu}$,
where space stretches alongside time, $A^2(\phi(x))$ cancels out of null
rays, leaving only the Planck-suppressed scalar stress tensor
($\kappa_\phi\simeq6.5\times10^{-6}$ on the selected two-branch sector in
Paper 19, Step 64) and the bounded disformal transport term. Under TEP's
unstretched-space temporal metric
$\tilde g_{\mu\nu}=-A_{\rm loc}^2(u_{\rm grav}(\mathbf x))U_\mu U_\nu+A_{\rm sp}^2(u_{\rm drift}(t,\mathbf x))\Pi_{\mu\nu}$ ($U_\mu\equiv-\nabla_\mu\bar\phi/\sqrt{2\bar X}$, rescaling the shared $(3+1)$ ADM spacetime lapse $N_{\rm TEP}=A(\phi)N_{\rm GR}$ at fixed spatial 3-metric $\gamma_{ij}$ so $c_{\rm GW}=c_\gamma$ identically), intermediate
clock rates $A(\phi(x))$ along the photon path do not cancel, refracting
light by $n(x)\simeq1-2\Phi_N(x)/c^2-\beta_A u_{\rm grav}(x)$ and reproducing the
Bullet Cluster galaxy-peak convergence (Paper 4, Step 02), the SN Refsdal
Fermat delay and arrival order (Paper 19, Step 64), and the Planck CMB
lensing cosmic-web amplification (Paper 18, Step 21) from baryons and the
selected two-branch action $P_{,X}=k\sqrt\xi+2\xi$ with zero particulate
dark matter.
Within the TEP framework, the same Temporal Topology field φ that modulates
clock correlations on Earth structures the temporal response of matter
across cosmological distances. The canonical TEP transition scale
RT = (3M/4πρT)1/3, set by the
environmental screening density ρT, connects terrestrial
and astrophysical scales. The measured GNSS correlation-length family
(azimuth-averaged λ = 3,210 ± 389 km; cross-sector
dispersion λT = 4,201 ± 1,967 km) is interpreted as
a characteristic empirical scale of the clock-sector covariance $C_A$ in Earth's environment, compared with the Earth-calibrated $R_T$ scale (Paper 0 §7). At galactic scales, the distinct shear response can be active in low-acceleration environments where SΣ ≃ 0.98. The galactic ∼50 kpc phenomenology concerns that weak-field dynamical regime, not an extrapolation of the GNSS covariance length or a common screening radius. The terrestrial and galactic
observations are interpreted within the same environmental-response
architecture; Papers 4 and 13 test distinct optical and dynamical
projections of that architecture. Paper 4 identifies the
dynamical–lensing mass difference as the phantom-mass consistency test;
Paper 13 reports a SPARC-anchored wide-binary transition at
Rs ≃ 2,646 ± 182 AU. Neither supplies a derived
quantitative mapping from terrestrial $C_A$ to galactic shear.
## 5.1 GW170817 Constraints
The simultaneous detection of gravitational waves (GW170817) and gamma rays
(GRB 170817A) from a binary neutron star merger at approximately 40 Mpc is
widely cited as constraining modified gravity theories. The signals arrived
within Δt = 1.74 ± 0.05 seconds after traveling some 130 million years,
yielding:
$$\frac{|c_{\gamma} - c_g|}{c} \lesssim 10^{-15}$$
This constraint is often interpreted as ruling out theories in which light
and gravity propagate differently. However, this interpretation conflates
two distinct effects that must be carefully distinguished.
The disformal sector B(φ) tilts light cones, causing cγ ≠
cg, and is indeed tightly constrained by GW170817. The conformal
sector A(φ), by contrast, rescales proper time isotropically: photons and
gravitational waves share the same null geodesic and experience identical
time dilation. Because both messengers originated from the same coordinate
and traversed the same path through the scalar field, any conformal time
dilation is common-mode and cancels in the differential measurement.
GW170817 therefore constrains cone tilting but remains blind to clock-rate
modulation. The dark matter phenomenology in TEP arises from the Temporal
Shear acting on matter trajectories — the channel to which multi-messenger
tests and static image-plane geometry are both blind
— rather than from null-ray deflection, which the conformal sector leaves
unmodified.
This can be seen explicitly in the conformal metric limit (B = 0):
$$ds^2 = A^2(\phi) g_{\mu\nu} dx^\mu dx^\nu = 0$$
Since A(φ) is an overall prefactor, the null condition ds² = 0 is identical
for both the matter metric g̃μν (governing photons) and the
gravitational metric gμν (governing gravitational waves). Their
null cones coincide exactly: cγ = cg = c locally. The conformal factor therefore produces no direct same-path propagation delay; its observable effect is carried by proper-time and clock-rate mapping, while differential cone propagation belongs to the disformal sector.
## 5.2 The Isochrony Axiom and Observational Degeneracy
Standard gravitational lensing analysis assumes that, once geometric and
Shapiro propagation delays (the Fermat potential) are modeled, matter-clock
calibration along the sightline is exhausted by the single spacetime metric so
that static lens reconstructions treat the source as a synchronous snapshot.
TEP formalizes this single-metric temporal closure as the Isochrony Axiom and
provides a parameterized two-metric framework to empirically stress-test
whether independent scalar temporal-field gradients contribute to the
reconstructed delay and convergence surfaces.
Within the TEP framework, this axiom would generically fail in the presence
of conformal metric couplings. Photons themselves accumulate no proper time
(dτγ = 0) and conformal factors produce no additional same-path
propagation delay; what varies is the matter-clock rate. Signals emitted from
regions sampling different temporal-field values φ(x) are referenced to
matter clocks that accumulate proper time at different rates, so the measured
source-to-observer timing relation contains the standard geometric/Fermat
delay together with the source- and observer-clock mapping determined by the
temporal field — an endpoint and reconstruction contribution, not an
additional conformal photon transit delay. For an evolving source, the
resulting differential emission-time structure can survive into the
reconstructed image, creating an observational degeneracy with particulate
dark matter under standard analysis assumptions:
Consider two interpretations of the same Fermat potential surface. Under the
standard interpretation, which assumes isochrony, an Einstein ring with
apparent convergence κobs exceeding baryonic mass implies a dark
matter halo with MDM = Mobs − Mbaryons.
Under the TEP interpretation, which rejects isochrony, the observed image is
a temporal composite: photons arriving simultaneously left the source at
different epochs. For evolving sources, this temporal depth projects onto
the image plane as apparent spatial distortion—apparent mass that does not
correspond to matter. The channel is bounded: it vanishes identically for
static sources and is limited to \(\lesssim 10^{-7}\) in shear for galaxy
sources by the ratio of differential delay to source-evolution time
(Paper 4), so it cannot account for the static-source lensing excess.
The degeneracy is therefore bounded and source-dependent, not a free
reinterpretation: for static sources the two frameworks make different
predictions, with the TEP dark-sector signature living in the dynamical
sector — the dynamical–lensing mass difference — rather than in the lensed
image itself.
## 5.3 Phantom Mass
Within a two-metric framework where the Isochrony Axiom breaks down, a
mechanism for apparent dark matter emerges naturally. Phantom Mass is the
observational projection of dynamical temporal geometry: spatial variation
of the conformal time field changes the source–lens–observer temporal mapping
and the proper-time histories of the emitting regions that different image
paths sample. Standard
isochronous reconstruction maps this temporal structure onto an apparent
spatial mass distribution. Genuine cone-tilting propagation terms belong to
the disformal sector. Phantom mass has two routes: this
reconstruction channel, which requires an evolving source and vanishes
for static sources, and the dynamical channel — the Temporal Shear's
direct effect on matter trajectories — which carries the dark-sector
claim and is tested by the dynamical–lensing mass difference
(Papers 4, 19).
For an evolving source, this temporal-structure difference creates a temporal composite
image: photons arriving simultaneously at the detector left the source at
different epochs. Standard lens reconstruction, which assumes isochrony,
interprets this temporal smearing as spatial distortion—apparent mass that
does not exist.
The amplification matrix in TEP decomposes as:
$$\mathcal{A}_{ij} =
\underbrace{\mathcal{A}^{\text{GR}}_{ij}}_{\text{Baryonic}} +
\underbrace{\mathcal{A}^{(g[\phi],\text{static})}_{ij}}_{\text{bounded optical}}
+ \underbrace{\mathcal{A}^{(\text{dyn})}_{ij}}_{\text{Shear Noise}}$$
Standard analyses attribute the sum of the first two terms to total mass.
TEP identifies the second term as a bounded optical contribution — scalar
backreaction on the Einstein-frame metric ($\kappa_\phi \approx 4\times10^{-5}$,
$\sim 2\times10^{3}$ below dark-matter level) plus the disformal
null-cone tilt ($\lesssim \tfrac{1}{5}$ of the Refsdal residual under the
measured deprojection; Paper 19, Steps 61–63) — rather than
particulate matter. Phantom mass itself is not an amplification-matrix
term: it enters the dynamical sector, and the dynamical–lensing mass
difference is its observable signature.
### 5.3.1 Static versus Dynamic Time Delays
Observed strong-lens time delays (day-to-month scale, e.g., SN Refsdal
Δt ~ 1 year between images) are set by the propagated Fermat structure;
the halo-scale temporal field enters the inference not through the
measured delay itself but through the reconstructed lens model. The
distinction resolves through two complementary mechanisms.
**Mechanism 1: Static versus Dynamic Decomposition** — The
amplification matrix decomposition (Equation above) separates temporal
effects into distinct components with different observational signatures:
**Static Temporal Structure (𝒜(g[φ],static)):** The
scalar's gravitational backreaction on gμν contributes
real, structured curvature on halo scales — bounded at
κφ ∼ 10−5, about 2×103 below
dark-matter level (Paper 19, Step 61) — a smooth field on the
image plane whose Hessian an isochronous reconstruction absorbs
into the inferred convergence κ, with the mean level entering the
mass-sheet normalization. The conformal clock-transfer map
Δτ̃(θ) = c⁻¹∫(A(φ) − 1) dl is the companion reconstruction-space
field the lens model would misassign to convergence or delay
calibration; it carries no propagated photon content: null cones
are conformally invariant and the Temporal Shear one-form is exact,
so the open-path integral is endpoint-determined and the exact
static-conformal inter-image residual is 0.0 d (Paper 19, step_56).
**Source-evolution response (𝒜(dyn)):** The gradient of
the delay field ∇(Δτ̃) creates differential emission-time structure
that depends on the
ray's impact parameter and the source's evolutionary state. This is
an endpoint-sector effect — it acts on the moving source's own
clock/emission mapping (the temporal-composite response of Paper 4),
not on photon propagation, so it inherits neither the conformal
exact-one-form cancellation nor the disformal ≲ 0.2 s bound. For
transient sources with variability timescale tvar, the
dynamic contribution scales as Δτdynamic ~ ∇(Δτ̃) × vsource
× tvar, where vsource is the source's proper
motion or internal velocity field.
**Observational Consequence:** Time delays measured between
multiple images of transient events (supernovae, AGN flares) are governed
by the propagating carriers: the standard Fermat potential plus the
disformal multipath deformation, bounded by GW170817 to ≲ 0.2 s per
2 Mpc halo path — far below any measured lens delay. The halo-scale
temporal-depth field contributes no propagated delay residual; its
observable imprint enters instead through the mass reconstruction, where
the unmodeled image-plane structure is absorbed as convergence.
**Mechanism 2: Screening Locality** — The Temporal Topology field φ
varies significantly only within the region where Temporal Shear is
non-negligible, bounded roughly by the TEP transition scale
RT ~ 100 kpc around the lens galaxy, where the environmental
screening operator SΣ(ℰ) transitions from the
suppressed to the unscreened regime. Beyond this radius, Temporal Topology
flattens to its cosmological value, φ → φ0, and temporal-structure signatures cease accumulating.
The integration path for the temporal-depth scale is therefore limited to the halo
depth Lhalo ~ 2 Mpc (the physical extent of the dark matter
halo), not the full cosmological distance DL ~ Gpc to the source.
This locality ensures that the static temporal-depth scale remains far
smaller than the cosmological light-travel time DL/c ~ 109 years.
$$\Delta \tau_{\text{depth}} \sim \frac{\epsilon}{2}
\frac{L_{\text{halo}}}{c} \quad \text{(not }
\frac{D_L}{c} \sim 10^9\text{ years)}$$
**Testable Expectation for Time-Delay Cosmography:** TEP
implies a systematic offset in H0 measurements from time-delay
cosmography, but through the mass-inference channel, not the image
plane. The unmodelled temporal field enters neither the delay numerator —
the propagated conformal residual is zero — nor the reconstructed
potential: the scalar-backreaction and disformal image-plane channels are
bounded far below dark-matter level (Paper 19, Steps 56, 61, 63), so no
phantom convergence is available to inflate ΔΦ. What the temporal field
modifies is the *dynamical* input by which the mass-sheet
degeneracy may be broken: TDCOSMO-style analyses anchor the lens model
with the galaxy's stellar velocity dispersion. Paper 4 identifies a
dynamical–lensing mass difference as a test of active Temporal Shear;
Paper 19 evaluates delay-blind implied phantom fractions in several
lenses. Neither result derives a universal conversion from the
source-screening fraction $S_\Sigma$ to a stellar-dynamical mass bias
and then to a fitted $H_0$ shift. In particular, the PPN photon–source
product $\alpha_0\alpha_{\rm eff}$ does not by itself establish the
two-matter-body dynamical response at a lens aperture.
The kinematic apertures of galaxy-scale lenses probe accelerations of
order $10^{-9}$–$10^{-8}$ m/s², above the transition scale
$g_t \approx 3.4\times10^{-10}$ m/s². A time-delay prediction therefore
requires the pairwise shear response, the stellar-orbit projection and
a lens-model mass-sheet refit for each system. A numerical $H_0$ offset
cannot be inferred solely from the aperture screening factor; the
delay-blind contrasts in Paper 19 are empirical tests of the proposed
phantom-mass interpretation of the delay residual rather than a calibrated correction to H0LiCOW.
**Falsification Test:** If TEP is correct, time-delay
cosmography measurements should show systematic residuals correlated with
the acceleration at the lens's kinematic aperture relative to \(g_t\),
after controlling for halo mass, velocity dispersion and mass-model systematics. Analysis of the H0LiCOW and TDCOSMO samples
stratified by \(a_{\rm ap}/g_t\) should reveal this correlation
if the phantom-mass channel is operative; a clean null across lenses whose
apertures approach \(g_t\) would bound the channel. A first pass already
exists in Paper 19's delay-blind phantom-mass analysis (step_60): implied
phantom fractions are positive across all non-zero contrasts — λ = +8.7 ±
2.6% (Refsdal SX:S1), +9.1 ± 5.8% (Encore), +9.5 ± 4.8% (H0pe AB), with a
archived diagonal-weight mean fractional delay residual of +7.8 ± 2.0% (χ² = 2.06/3 dof; not a covariance-calibrated phantom-mass detection) —
marginally above the mundane κext floor and in the predicted
sign, though not yet σv-stratified.
## 5.4 The Variability-Mass Correlation
A critical discriminator between TEP and standard dark matter emerges from
the dynamic term in the amplification matrix. Because the stochastic shear
contribution depends on source proper motion μs through the delay
gradient, the inferred lens mass should correlate with source variability
timescale. Physical lens mass is source-independent in the standard model,
but source variability and time delays can also bias standard reconstructions.
The discriminant is therefore the residual beyond that baseline, not the
existence of a variability correlation alone (Paper 4, Regime II).
Specifically, lenses observed through rapidly variable sources (AGN,
quasars) should show systematically different mass reconstructions than the
same lenses observed through slowly evolving sources (elliptical galaxies).
The amplitude and sign require the measured source variability, ordinary
Fermat delays and reconstruction transfer; no universal percentage mass
offset is supplied by the existing temporal-composite calculation.
This test can be performed with existing data from HST strong-lens archives
and SDSS quasar lenses. A null result bounds the response at the achieved
sensitivity; exclusion requires that sensitivity to reach the specified
source-dependent prediction. A positive correlation must be tested against
source selection, ordinary delay/variability effects and reconstruction
systematics before it can discriminate TEP.
## 5.5 Cross-Scale Dynamics: Saturation vs. Kinetic Transitions
The TEP framework connects terrestrial timing phenomenology with astrophysical and galactic dynamics through two distinct, mathematically grounded scales:
**The Clock Sector (Geometric Saturation Scale RT):** For macroscopic bodies of saturation density ρT ≈ 20 g/cm3, the geometric saturation radius follows RT(M) = (3M/4πρT)1/3 ∝ M1/3. Calibrated through Earth's GNSS clock-covariance correlation length (RT(M⊕) ≈ 4,150 km; Paper 6), this scale fixes the unsaturated clock-amplitude factor SA = min[1, (ρ̄/ρT)1/3] ≈ 0.65 for Earth. It characterizes static clock-sector depth; the measured GNSS covariance CA(r) and its dynamic environmental and processing transfer are distinct from this static SA level.
**The Force / Shear Sector (Kinetic Transition Scale r*):** For geodesic kinematics, orbital dynamics, and rotation curves, the selected two-branch kinetic action P(X,φ) = −V(φ) + (2k/3ΛX2)X|X|1/2 + X|X|/ΛX4, with k = 4√2|βA|gt/a₀ ≈ 16 derived from the adopted acceleration constant, derives the two-body transition scale directly from first principles:
$$r_*(M) = \sqrt{\frac{GM}{g_t}} \propto M^{1/2}$$
where gt = cH₀/(2|βA|) ≈ 3.4 × 10−10 m/s2 with ΛX = √(MPl H₀) ≃ 1.9 meV (Paper 0 §2.2). In high-acceleration regimes (g ≫ gt), the kinetic operator suppresses the observable Temporal Shear to 𝒮Σ ≪ 10−5 (satisfying Cassini and Solar-System ephemeris bounds), while in low-acceleration regimes (g ≪ gt) the shear recovers unsuppressed (𝒮Σ → 1).
Galactic dark-matter-like phenomenology does *not* represent an M1/3 screening radius extrapolated across 15 orders of magnitude. Active Temporal Shear supplies the extra matter acceleration within the selected two-branch kinetic sector P,X = k√ξ + 2ξ of Paper 0 §7 — a single action whose quadratic branch (ξ ≫ ξ* ≈ 64, the regime of every Solar-System benchmark) supplies the bounded local enhancement and whose low-gradient branch supplies flat outer rotation curves with baryonic Tully–Fisher scaling, evaluated against SPARC (Paper 6, Steps 10–13). The field-wide-binary transition Rs ≈ 2,646 AU (Paper 13) tests the pairwise recovery profile within the same sector; it is not itself a derivation of the galactic outer force law. The corrected zero-free-parameter wide-binary joint test passes this sector at 0.4–1.7σ while excluding the incumbent unit-term baseline and the exponential-interpolation rival at ∼30σ on amplitude, and under the action-derived additive force law $g_{\rm obs}=g_N+a_\phi$ ($P_{,X}=\epsilon_{\rm reg}+k\sqrt{\xi}+2\xi$, $\epsilon_{\rm reg}=10^{-3}$, $k_*=16.03$ calibrated to $a_0$) the SPARC fit returns RMS 0.2185 dex at $\Upsilon_d=0.5$ and 0.2004 dex at $\Upsilon_d=0.35$ (matching the empirical MOND reference 0.1989 dex and outperforming the 0.2812 dex unit-term baseline by >60 mdex; phenomenological quadrature comparator $\sqrt{g_N^2+a_\phi^2}$ gives 0.1974 dex at $\Upsilon_d=0.5$ with $k_{\rm fit}=15.78$ and outer-edge BTFR median $v_{\rm obs}/v_{\rm pred}=1.0109$); uniqueness across the wider K(X) class remains open (Paper 0 §7). Table 8 distinguishes the corresponding responses.
| Physical Regime | Governing Scale | Scaling Law | Observational Channel | Typical Domain |
| --- | --- | --- | --- | --- |
| Clock Sector (Saturation) | RT = (3M/4πρT)1/3 | RT ∝ M1/3 | Static conformal-depth calibration (SA); clock covariance (CA) measured separately | Planetary bodies: Earth (RT ≈ 4,150 km) |
| Kinetic Shear Sector (Two-Body) | r* = √(GM/gt) | r* ∝ M1/2 | Geodesics, wide-binary orbital velocity excess | Wide binaries (Rs ≈ 2,646 AU at 1.24 M☉) |
| Galactic Halo Regime | Low acceleration (g ≪ gt) | Low-gradient branch of the selected two-branch sector (ξ ≪ ξ*) | Flat-tail rotation curves and BTFR from the low-gradient branch (Paper 6, Steps 10–13); dynamical phantom mass from the same action | Galactic halos (r ~ 10–50 kpc) |
This clean bifurcation separates the compact-body clock-amplitude response (SA) from the dynamical field-gradient recovery (r*), eliminating the vulnerability of treating GNSS correlation lengths as universal screening thresholds.
## 5.6 Environmental Cepheid Response and the Hubble Tension
Local distance-ladder measurements and CMB inferences yield different values of $H_0$. Paper 11 establishes that an environment-dependent Cepheid clock-calibration response resolves this discrepancy: the host-level reconstruction reaches $H_0 = 67.83 \pm 1.56$ km/s/Mpc ($0.26\sigma$ from Planck), while the prespecified ladder-matrix projection retains an identifiability-limited residual; the corrected value is conditioned on the fitted response and the observing-chain model, not a direct clock-period prediction.
### 5.6.1 Cepheids as Environmental Clocks
Cepheid pulsation periods are governed by stellar-envelope dynamics in local matter-frame proper time. A direct conformal period shift from host potential alone does not reproduce the measured environmental response after the full matter-frame observing chain is imposed: the bare core-to-disk clock-ratio contribution is bounded well below the fitted Cepheid coefficient (Paper 11, Step 61). Paper 11 instead tests a period-dependent amplitude-sector response with an inferred stellar transport weight; its absolute normalization from a common stellar and ambient-field solution remains outstanding.
### 5.6.2 Empirical Detection of the Bias
Analysis of the SH0ES Cepheid sample (Paper 11), stratified by host-galaxy inner-potential depth σ* (a TEP-independent local stellar-dispersion scale compiled from documented stellar-absorption and H I-linewidth measurements), reveals a significant positive correlation between potential depth and derived H₀ — an environmental ordering tested by Paper 11's revised amplitude-sector response, not a detection of the bare clock-ratio mechanism. Split at the sample median (σ* = 95.0 km/s), the deep-potential subsample yields H₀ = 72.64 ± 2.62 km/s/Mpc against 63.05 ± 3.06 km/s/Mpc for the shallow-potential subsample (full host-distance covariance propagated; Pearson r = 0.501, p = 0.0016; Spearman ρ = 0.553, p = 3.9 × 10⁻⁴), and the continuous endpoint response ΓX = (3.846 ± 1.653) × 10⁷ km/s/Mpc (2.33σ at the canonical σv = 250 km/s convention; 2.81σ at the σv = 150 km/s sensitivity variant) returns a positive sign in all 37 leave-one-host-out refits. The constraint on κCep is set by this environmental slope—a statistic orthogonal to the mean offset that defines the tension—and the procedure is falsifiable on the same data: a flat or opposite-sign slope admits no coefficient that reconciles the corrected value with the early-universe baseline. Two channels outside the Hubble-flow fit provide additional, differently controlled checks rather than a calibration of the absolute response. The Cepheid–TRGB differential on the 13 matched hosts—regressing μCep − μTRGB against the same environmental coordinate, independent of redshift information—returns κCep = (0.250 ± 0.313) × 10⁶ mag (0.80σ), directionally consistent with the predicted κCep > κTRGB ordering at presently low significance; and the internal radial P–L gradient within the LMC (+0.0284 ± 0.0086 mag, 3.3σ, robust to period and colour matching) propagates through the fitted transfer to the same order of response.
Under Paper 11's fitted environmental response and adopted anchor weights, the model-conditioned correction yields:
$$H_0 = 67.83 \pm 1.56 \text{ km/s/Mpc}$$
The Cepheid host-potential correlation is an empirical environmental result, while its causal normalization remains model-dependent. Paper 11 finds that the bare conformal core-to-disk clock ratio cannot supply the fitted response; the revised amplitude-sector, period-dependent carrier has an inferred transport weight but awaits an absolute stellar/ambient-field normalization. Its corrected local $H_0 = 67.83 \pm 1.56$ km/s/Mpc agrees with the companion TEP-CMB value $66.70 \pm 0.58$ km/s/Mpc at 0.68σ within the framework, not as an independent confirmation of the mechanism. The joint-ladder bookkeeping assigns the fitted response predominantly to the endpoint distance-modulus channel rather than propagation along the Hubble-flow leg (Paper 11, Appendix C.4). Host-uniform shifts can be absorbed into latent distance moduli, so the prespecified full-ladder matrix projection retains an identifiability-limited residual ($H_0 = 70.46 \pm 0.97$ km/s/Mpc, $2.80\sigma$ from Planck); the host-level reconstruction, which does not suffer that degeneracy, reaches $0.26\sigma$ statistical concordance — the route by which Paper 11 resolves the tension.
## 5.7 Observational Predictions
TEP yields testable expectations for existing and near-future
observations, summarized in Table 9.
| Observable | TEP Prediction | Dataset |
| --- | --- | --- |
| Lensed quasar time delays | Systematic residuals correlated with halo mass | H0LiCOW, TDCOSMO |
| Galaxy rotation curves | r* ∝ M1/2 fixes the kinetic transition, not outer flatness; the flat-tail/BTFR prediction is evaluated under Paper 6's low-gradient branch. Geometric saturation remains RT ∝ M1/3. | SPARC, THINGS |
| Weak lensing S8 | No universal sign fixed by the source-dependent covariance channel; the implemented HC posterior is above DES-Y3/KiDS-1000 and is a live diagnostic | DES, KiDS, HSC vs. Planck |
| Strong lens mass-variability | Source-dependent reconstruction residual; amplitude set by the measured transfer | SLACS, BELLS |
| FRB lensing residuals | Achromatic delays 0.1–10 ms | CHIME/FRB repeaters |
These predictions are falsifiable within 2–5 years using archival data and
ongoing surveys. A single null result does not exclude TEP, but a pattern of
null results across all five observables would constitute strong evidence
against the framework.
## 5.8 Connection to Cosmological Tensions
TEP offers potential resolution to persistent tensions in precision
cosmology.
Regarding the H0 tension: time-delay cosmography assumes standard
GR time transport. If TEP's temporal structure contributes to lens delays,
the inferred H0 would be biased, potentially explaining why
independent time-delay lensing measurements falsely agree with the
uncorrected distance ladder, rather than tracking the true CMB value.
Regarding the S8 tension: galaxy sources can carry additional
source-dependent reconstruction covariance absent from the CMB
source-evolution channel. Paper 4 bounds this temporal-composite
contribution far below the percent-level mean-amplitude discrepancy;
meanwhile, under the unstretched-space temporal metric
$\tilde g_{\mu\nu}=-A_{\rm loc}^2(u_{\rm grav}(\mathbf x))U_\mu U_\nu+A_{\rm sp}^2(u_{\rm drift}(t,\mathbf x))\Pi_{\mu\nu}$ ($U_\mu\equiv-\nabla_\mu\bar\phi/\sqrt{2\bar X}$), intermediate
temporal depth along the photon path amplifies baryon-only cosmic-web
and cluster-outskirt potentials by $9.4\times$–$129\times$ in
$C_\ell^{\phi\phi}$ on the selected two-branch action (Paper 18, Step 21),
while the kinetically screened cosmological branch weakens late-time growth
enhancement to $G_{\rm eff}(z{=}0)\simeq1.16\,G$ (Paper 0, Step 61).
The source-class dependence is a bounded prediction to test against
ordinary reconstruction and selection effects; it is not a demonstrated
explanation of the mean S8 discrepancy or CMB deflection.
## 5.9 Relation to ΛCDM
The standard ΛCDM cosmological model fits observational data with remarkable precision. This framework attributes gravitational lensing anomalies to cold dark matter particles and explains cosmological structure formation through hierarchical assembly of particulate halos. ΛCDM is empirically successful and theoretically well-motivated.
TEP treats ΛCDM as the successful control reconstruction, not as the final ontology. In the TEP interpretation, particulate dark matter and dark energy are not fundamental ingredients; they are effective labels produced when temporal shear, temporal topology, and screening structure are compressed into an isochronous FLRW reconstruction. The key distinction is therefore ontological, while many current static-source observables remain operationally degenerate:
- **ΛCDM interpretation:** Gravitational lensing mass reconstructions directly measure matter distribution. Excess convergence beyond baryonic mass indicates particulate dark matter with density profile ρDM(r). The cosmological constant or dark-energy sector supplies late-time acceleration. The Isochrony Axiom (single-metric temporal closure across the reconstructed image) is adopted as a foundational premise of the standard lens reconstruction.
- **TEP interpretation:** Gravitational lensing mass reconstructions measure the effective Fermat potential. While isotropic conformal rescaling $\tilde g_{\mu\nu}=A^2(\phi)g_{\mu\nu}$ cancels out of null rays (Paper 0, Appendix A2), the unstretched-space temporal metric $\tilde g_{\mu\nu}=-A_{\rm loc}^2(u_{\rm grav}(\mathbf x))U_\mu U_\nu+A_{\rm sp}^2(u_{\rm drift}(t,\mathbf x))\Pi_{\mu\nu}$ ($U_\mu\equiv-\nabla_\mu\bar\phi/\sqrt{2\bar X}$) retains intermediate clock rates $A(\phi(x))$ along every photon path across static space, refracting light by $n(x)\simeq1-2\Phi_N(x)/c^2-\beta_A u_{\rm grav}(x)$ on the selected two-branch action $P_{,X}=k\sqrt\xi+2\xi$ ($k=16.03$) and reproducing cluster convergence (Paper 4, Step 02), strong-lensing Fermat delays (Paper 19, Step 64), and CMB lensing (Paper 18, Step 21) from baryons alone, alongside the bounded emission-time temporal-composite response for evolving sources ($\lesssim10^{-7}$ in shear; Paper 4) and apparent dark energy from non-integrable temporal transport over epoch.
Both frameworks fit existing lensing data because they make identical predictions for static sources: the conformal sector supplies no propagated content, and the bounded channels — scalar backreaction and disformal transport — sit far below the lensing excess (Paper 19). The degeneracy is broken by observations that probe temporal structure or dynamical inference:
- **Lensed fast radio bursts:** ΛCDM predicts achromatic residuals (after plasma correction) consistent with zero. TEP predicts achromatic residuals at the 0.1–10 ms level arising from the disformal transport sector — the bounded non-exact channel, not conformal path accumulation (Section 5.4).
- **Variable source lensing:** ΛCDM predicts inferred lens mass is independent of source variability timescale. TEP predicts systematic correlation between inferred mass and source variability, with AGN showing higher apparent mass than elliptical galaxies for the same lens (Section 5.4).
- **CMB vs. galaxy lensing S8:** ΛCDM predicts convergence of CMB-inferred and galaxy-inferred S8 as systematics are controlled. TEP's bounded source-evolution channel predicts survey-dependent covariance, but not the observed mean offset; the implemented Paper 18 posterior lies above DES-Y3/KiDS-1000 under a CDM-containing likelihood, so the current difference is a live tension rather than an established TEP explanation.
The frameworks compete at the level of ontology. ΛCDM remains the default control model given its theoretical maturity and empirical success, but TEP denies that the dark sector is fundamental. TEP becomes distinguishable through specific tests designed to probe temporal structure in lensing observations. Independent of the cosmological extension, the terrestrial GNSS correlation structure documented in Sections 2–4 requires explanation within any framework.
**Falsification asymmetry:** Null results in the tests above would exclude the tested temporal-field replacement channel while leaving ΛCDM intact as the control reconstruction. Positive results would force a reinterpretation of dark matter phenomenology in the tested regime and would count directly against dark-sector ontology. The full TEP claim that dark matter and dark energy are non-fundamental still requires consistency across clusters, CMB lensing, structure formation, and late-time distance probes.
## 6. Falsification and Experimental Program
A theory that cannot be falsified is not science. TEP makes specific,
testable predictions with explicit null-result thresholds. This section
specifies the criteria that would exclude the framework and outlines the
experimental path to a definitive verdict.
## 6.1 Explicit Falsification Criteria
The TEP framework is falsified if any of the following observations are
confirmed:
#### Tier 1: Terrestrial Falsification (GNSS Signal)
**Independent replication failure:** Independent
research groups fail to reproduce exponential decay (λT ≈ 3,000–5,000
km) using either raw carrier-phase observables (RINEX data with
Single Point Positioning) or independent clock solutions (CLK files
from CODE, IGS, ESA, or JPL). This would exclude the entire GNSS
evidence base.
**Correlation length out of bounds:** Independently
replicated correlation length falls outside λT = 500–20,000 km range.
This would exclude the TEP prediction of λT ~ 4,000 km based on the
relaxation of Temporal Topology from Earth's potential well.
**Ephemeris loophole confirmed:** The ephemeris
question splits into two layers with different adjudicators.
(i) *GNSS-product provenance*: Satellite Laser Ranging
(SLR) residuals show no low-frequency spectral concentration or
distance-structured daily-aggregation signal when analyzed
using sampling-matched methods and demonstrated sensitivity to the predicted optical counterpart. Only a null at that sensitivity can exclude the specified common-clock transfer. The present spectral null alone does not establish a GNSS-product origin (Section 6.3.1.1).
(ii) *Generic orbit-projection confound*: a positive SLR
result cannot by itself exclude the wider hypothesis that any
satellite-range observable inherits distance-structured
residuals from its own orbit-determination chain; adjudicating
that layer requires the physical-confound forward model (the
orbit/product-error projection kernel) or an orbit-free clock
channel such as terrestrial fiber-link or transportable-clock
comparisons, which carry no satellite content at all.
#### Tier 2: Synchronization Holonomy (Direct Test)
**Null holonomy at high precision:** Closed-loop
one-way time-transfer around triangular geometry (1,000–3,000 km
legs) yields null synchronization holonomy ℋ/Tloop <
10-19 after subtracting all known relativistic effects (Sagnac,
Shapiro, gravitational redshift, Lense-Thirring). This would
constrain disformal coupling B(φ) below cosmologically interesting
levels and exclude the non-integrable proper time prediction
(Section 6.3.3).
#### Tier 3: Cosmological Falsification (Source-Dependence Channel)
**CMB-Galaxy source covariance:**
Next-generation surveys (CMB-S4 + LSST) showing no systematic
dependence on source redshift or variability would exclude the
bounded temporal-composite reconstruction channel. The current
S8 offset is a diagnostic to be confronted, not a
predicted correspondence.
**Framework Scope:** Tier 1 tests the GNSS evidence base,
Tier 2 tests the synchronization-holonomy mechanism, and Tier 3 tests
the bounded weak-lensing source-dependence channel. Failure of a single
Tier 1 criterion is sufficient to exclude the GNSS evidence base; a null
Tier 3 result constrains that channel without deciding the dynamical
sector.
## 6.2 Consequences of Failed Replication
If independent groups cannot replicate the raw RINEX signal, this would
indicate one of several possibilities: a subtle processing artifact specific
to the present pipeline that was not identified, a statistical fluctuation
in the particular dataset selection, or an unidentified systematic in the
cross-spectral analysis method.
In such a scenario, TEP would lose its primary empirical support. The
theoretical framework would remain mathematically consistent but empirically
unmotivated, and the cosmological predictions would become untestable
speculation. This outcome would represent the scientific method working as
intended.
## 6.3 Three-Tier Experimental Strategy
TEP can be confirmed or excluded through a coordinated experimental program
organized by urgency and discriminating power.
### 6.3.1 Tier 1: Immediate Replication
The most urgent priority is independent replication. If independent research
groups cannot replicate the raw RINEX signal, the GNSS evidence base is
excluded and TEP loses its primary empirical support.
Required tests include independent clock analysis using either raw
carrier-phase data (RINEX files with SPP processing) or processed clock
solutions (CLK files from independent analysis centers), multi-constellation
cross-check (verifying that the signal appears identically in GPS, GLONASS,
Galileo, and BeiDou), and cross-center consistency (confirming
CODE/IGS/ESA/JPL agreement with CV $\lt$ 20%). Both RINEX and CLK approaches
provide valid replication pathways; the key requirement is independence from
the original analysis pipeline.
These tests can be performed by any institution with GNSS processing
capability, including IGS, NIST, PTB, SYRTE, and JPL. The publicly available
data (both raw RINEX observations and processed CLK solutions) enable
independent verification of the λT ≈ 4,000 km exponential decay, and the
analysis can be completed in weeks using standard GNSS processing tools.
#### 6.3.1.1 Addressing the Ephemeris Loophole
A legitimate concern exists regarding the Single Point Positioning
test: SPP relies on broadcast ephemerides generated by control
segments using network adjustments. If the artifact originates in the
broadcast orbit determination, SPP would reproduce rather than eliminate it.
This represents a potential systematic loophole requiring independent
check.
Satellite Laser Ranging (SLR) provides the adjudicating test at the
product level, within a precisely stated scope. SLR is a purely
geometric, optical measurement independent of the GNSS microwave clock
and ephemeris loop: its satellites are passive retroreflectors, and its
orbits are determined from the laser ranges themselves — a
processing chain wholly disjoint from the GNSS broadcast and precise
products in which an ephemeris artifact would have to live. Within the
conformal channel, however, the SLR residual probes the same
endpoint-clock sector as GNSS: the ground event timer is itself a
clock, so a physical clock-rate field enters both observables at
co-located sites. The test therefore discriminates provenance rather
than sector: correlation structure present in SLR cannot originate in
the GNSS products, while structure absent at demonstrated sensitivity
supports a GNSS-product artifact.
Proposed Tier 1.5 Test: Correlate GNSS clock residuals with SLR range
residuals for co-located or nearby stations. The International Laser Ranging
Service (ILRS) operates approximately 40 SLR stations globally, with several
co-located with IGS GNSS stations. Analysis would proceed as follows: (1)
Extract SLR range residuals after precise orbit determination; (2) Compute
cross-correlation between SLR residuals and GNSS clock residuals for station
pairs as a function of distance; (3) Test whether the exponential decay
structure (λT ≈ 4,000 km) appears in SLR data.
Expected Outcome: If TEP is correct and the signal reflects physical
clock-rate modulation, SLR residuals should show distance-structured
correlations because the laser round-trip time depends on both the satellite
orbit and the ground station clock. If the signal is a GNSS-product
artifact, SLR residuals (which use independent orbit determination)
should show no correlation structure. Two qualifications bound the
verdict: the SLR measurement must reach the demonstrated sensitivity
for the GNSS-scale correlation before a null excludes the physical
reading, and an SLR-positive outcome remains degenerate with the
satellite-generic orbit-projection confound —
orbit-determination errors imprint spatially structured residuals on
any satellite-range observable. The differing orbit geometry of the
two systems (GNSS in medium Earth orbit, LAGEOS-class targets in low
Earth orbit) makes the correlation scale itself diagnostic: a common
λT across independent orbit-determination chains
indicates station-level rather than product-level origin. Final
closure of the generic confound rests with the physical-confound
forward model or the orbit-free channel above. The test can be
performed with existing ILRS data.
### 6.3.2 Tier 2: Direct Holonomy Test
The triangle holonomy test provides the most direct discrimination between
TEP and artifact hypotheses by measuring synchronization non-closure via
closed-loop, multi-leg, one-way time-transfer. The proposed geometry
involves two ground stations separated by 1,000–3,000 km plus a MEO
satellite. The target sensitivity is σ(H)/Tloop ≤ 10−19
after averaging, with a signal forecast of H/Tloop ~
10−16–10−18 for parameters saturating current bounds.
This test requires a dedicated mission.
### 6.3.3 Tier 3: Cosmological Testing
CMB-S4 combined with LSST weak lensing would test for source-dependent
kinematic noise. The discriminator is covariance with source redshift and
variability, not a universal 3–5% downward shift: the bounded Paper 4
temporal-composite channel is below the observed mean S8
discrepancy, while the implemented Paper 18 posterior lies above the
DES-Y3/KiDS-1000 central values under a CDM-containing likelihood. The
current S8 difference is therefore a live diagnostic and
falsification target, not a demonstrated TEP correspondence. Absence of
the predicted source-class covariance at Stage-IV precision would
exclude this kinematic-noise channel.
## 6.4 Path to Resolution
Table 10 summarizes the experimental program and its decision structure.
| Tier | Test | Null-Result Threshold | What It Would Exclude |
| --- | --- | --- | --- |
| Tier 1 | Raw carrier-phase replication | Signal absent | Entire GNSS evidence base |
| Multi-constellation cross-check | Different λ across constellations | Universal coupling |
| SLR ephemeris check (GNSS-product provenance) | No correlation in SLR residuals at demonstrated
sensitivity | Physical-clock origin; the generic orbit-projection
confound is excluded separately by the forward-model /
orbit-free test |
| Tier 2 | Triangle holonomy test | Null $\lt$ 10−19 | Holonomy mechanism |
| Tier 3 | CMB-S4 + LSST source covariance | No redshift/variability dependence | Temporal-composite reconstruction |
Tier 1 tests determine whether TEP survives as a viable hypothesis. If all
three Tier 1 tests confirm the signal, Tier 2 tests discriminate among
mechanisms. Tier 3 tests probe the cosmological regime. A definitive
verdict—confirmation or exclusion—is achievable through this coordinated
program.
## 6.5 Prediction Provenance Register
The pre-specification claim is anchored to the immutable, timestamped
version record of the public Zenodo deposits rather than asserted. Each
tested expectation is mapped below to the earliest versioned manuscript
section containing it, with deposit timestamps (UTC) and version DOIs that
are independently checkable through the Zenodo API. All framework-level
and pre-analysis entries precede the first public deposit of the GNSS
analysis (Paper 1 v0.6-alpha, deposited 2025-09-22); the register is
distributed as a machine-readable artifact
(`results/prediction_provenance.json`) so that the ordering can
be verified without trusting this manuscript.
| Tested expectation | Earliest versioned statement | Deposit (UTC) | Version DOI |
| --- | --- | --- | --- |
| Distance-correlated clock-network residuals (observable class;
cross-spectral estimator, 3-D distance binning, multi-center
cross-validation) | Paper 0 v0.4 (Marseille), §E Phase I | 2025-09-15 | 10.5281/zenodo.17121655 |
| Exponential spatial decay,
$C(r)=A\exp(-r/\lambda)+C_0$ | Paper 0 v0.4 (Marseille), §E Phase I; restated pre-analysis in
Paper 1 v0.6α §1.2 | 2025-09-15 | 10.5281/zenodo.17121655 |
| Correlation-length forecast, screening-parameterized
$\lambda\sim2{,}000$–$3{,}000$ km; stated search range
$\lambda=1{,}000$–$10{,}000$ km | Paper 0 v0.4 (Marseille), §E forecast; Paper 1 v0.6α §1.2
search range | 2025-09-15 / 2025-09-22 | 10.5281/zenodo.17121655; 10.5281/zenodo.17174802 |
| Multi-center parameter consistency ($\lt$ 5% variation) | Paper 0 v0.4 (Marseille), §E forecast | 2025-09-15 | 10.5281/zenodo.17121655 |
| Velocity-dependent coupling: EW/NS anisotropy anticorrelates
with Earth's orbital speed (negative sign) | Paper 1 v0.6α (Jaipur), §3.6, stated as the TEP expectation
before the reported result; framework basis in
velocity-dependent time transport, Paper 0 v0.1 | 2025-09-22 (framework 2025-08-21) | 10.5281/zenodo.17174802; 10.5281/zenodo.16921912 |
| Directional anisotropy of the correlation field | Paper 1 v0.6α (Jaipur), §3.2, carried as a critical TEP test;
the ecliptic EW>NS ordering ($2.16$) is the measured
outcome | 2025-09-22 | 10.5281/zenodo.17174802 |
| Achromatic universality (coupling not
frequency-selective) | Paper 0 v0.1 (Guyana), axiom A4 universal coupling | 2025-08-21 | 10.5281/zenodo.16921912 |
| Absence of $GM/r^2$ and $GM/r^3$ scaling in clock
correlations (selectivity profile) | Paper 0 v0.1 (Guyana), axioms A1–A4: conformal clock-sector
coupling is not a fifth-force channel | 2025-08-21 | 10.5281/zenodo.16921912 |
| Annual orbital-phase directional structure | Paper 0 v0.1+ permits a background direction; the measurement
recovers a phase near Earth's aphelion velocity tangent. It does
not identify the directed CMB rest frame. | 2025-08-21 | 10.5281/zenodo.16921912 |
Two entries are deliberately flagged as measured identifications rather
than pre-specified values: the ecliptic ordering of the anisotropy and the
preferred-axis direction. What was fixed in advance is the observable
class, the estimator, the functional decay form, the correlation-length
forecast and search range, the cross-center consistency criterion, the
sign of the orbital-velocity coupling, and the sector-level selectivity.
Of these, the screening-parameterized correlation-length forecast, the
$\lt$ 5% cross-centre criterion and the exponential-kernel criterion were not
met in full (Table 5). The discriminative weight therefore rests on the
orbital-velocity sign, band localization, the EW > NS ordering and the
sector-level selectivity.
## 7. Conclusions
This paper presents empirical evidence for a persistent, distance-structured
correlation in global atomic clock networks that probes an empirically
untested assumption of general relativity: whether distributed proper-time
rates contain a coherent spatial covariance beyond the standard relativistic model.
## 7.1 Summary of Findings
This integrated manuscript synthesizes results from a systematic research program that establishes, tests, and corroborates Temporal Equivalence Principle predictions through progressively rigorous empirical tests and cosmological extension. The convergence of evidence across seven domains—theoretical prediction (Paper 0), multi-center microwave consistency (Paper 1), 25-year longitudinal stability (Paper 2), raw RINEX test (Paper 3), optical SLR check (Paper 8), pulsar timing (Paper 10), and Cepheid environmental-response test (Paper 11)—provides an empirical foundation for TEP. The domains are not all channel-independent, and the audited strengths differ: the GNSS entries are the primary timing-network evidence; SLR supplies an optical bound and an exploratory correlation test within the same conformal clock sector; and the pulsar channel is exploratory, its standalone evidence uninformative rather than confirmatory.
The theoretical framework (Paper 0) supplied the search range and structural expectations tested by the GNSS programme. Distance structure and orbital coupling are recovered across several GNSS products and the raw-RINEX estimation chain. The optical SLR analysis (Paper 8) contributes a sampling-matched spectral null and sensitivity bound; its inter-station candidates remain exploratory. Its independent hardware and orbit provenance provide a future cross-instrument discriminator once a common signal is measured at adequate sensitivity.
Paper 11 fits an environment-dependent Cepheid response on the inner-potential coordinate σ* and reports a corrected host-level value H₀ = 67.83 ± 1.56 km s⁻¹ Mpc⁻¹ — 0.26σ from Planck — with the identifiability-limited ladder-matrix projection at 70.46 ± 0.97 km s⁻¹ Mpc⁻¹. Agreement with the companion TEP-CMB value is internal to the framework. The endpoint environmental slope ΓX = (3.846 ± 1.653) × 10⁷ km s⁻¹ Mpc⁻¹ is detected at 2.33σ with 37/37 leave-one-out sign stability, while the within-host period-coupled carrier reaches 2.28–2.79σ across anchor conventions in the joint closure. The period–density relation motivates a carrier with an inferred transport weight; its absolute response normalization remains to be derived from a common stellar and ambient-field solution. These empirical tests constrain that physical completion.
The evidence programme combines timing-network signatures, environmental tests and an optical sensitivity bound:
(1) distance-structured spatial decay over a finite scale (azimuth-averaged λ ≈ 3,210 km), (2) spatial anisotropy
(EW/NS = 2.16), (3) orbital velocity coupling (r = −0.888, 3.65σ
autocorrelation-corrected; surrogate p $\lt$ 2×10⁻⁷ as a resolution bound), (4) an
annual-phase direction near Earth's aphelion velocity tangent on the CODE
and RINEX products (CMB apex anti-phase; Solar Apex null; MGEX scan
projection-degenerate), (5) an optical SLR bound and exploratory inter-station test, (6) a planetary event catalogue functioning as a bounded control channel (catalogue excess attributable to calendar-position sampling; opposition subset null-validated; continuous potential correlation r = +0.116, p = 0.012), (7) 18.6-year nutation coupling,
(8) pulsar spin-down excess (exploratory), and (9) host-potential H₀ correlation. The
signal survives all challenges across solar cycles, geomagnetic storms,
altitude, and software-diverse processing chains—systematically excluding
conventional artifact explanations.
## 7.2 Theoretical Interpretation
These observations match the predictions of the Temporal Equivalence
Principle, a bi-metric framework in which proper time is a dynamical field.
TEP established pre-specified expectations and theoretically motivated
search ranges—including correlation length range, exponential decay,
velocity-dependent anisotropy, and absence of GM/r² (force) and GM/r³
(tidal) scaling—consistent with the observed data. The theory paper was
published in August 2025 with these predictions; the GNSS analysis was
conducted from September through December 2025.
The framework preserves local Lorentz invariance, with the speed of light remaining exactly invariant in any freely falling laboratory, while predicting global clock-network covariance and open-path conformal transport structure. GW170817 primarily constrains the disformal differential-propagation sector; it does not directly test common-mode conformal clock-rate covariance along shared paths, although conformal scalar sectors remain indirectly constrained by PPN, equivalence-principle, source-screening, and clock-comparison tests.
## 7.2a Predicted vs Fitted Coefficients across Probes
| Probe | Predicted in advance (sign/shape/ordering) | Fitted (amplitude) | Coefficient |
| --- | --- | --- | --- |
| GNSS λ (Papers 1–3, 14) | Exponential decay; EW>NS polarity; orbital-velocity sign | Decay length (metric- and product-dependent, 1,862–4,549 km on the precise-product family) | λ_T |
| Cepheid H₀ (Paper 11) | Sign of potential-depth correlation | Response amplitude | κ_Cep = (0.665 ± 0.362)×106 mag (multi-block) / (1.275 ± 0.370)×106 mag (redshift-only); theory benchmark κ_canonical = 0.96×106 |
| MSP spin-down (Paper 10) | Suppression direction vs Newtonian; field-binary null | Response amplitude | κ_MSP ~ 106 |
| LLR (Paper 17) | Synodic phase; sideband structure | κ_LLR = −3.91×10-4 | κ_LLR (Solar-System ORC) |
| Wide binaries (Paper 13) | Transition existence; environmental ordering (high-|Z| $\lt$ midplane R_s) | α_sat = 0.366; R_s = 2646 AU | α_sat, M_screen |
| Flybys (Paper 15) | Candidate clock-sector reduction artifact in clock-carrying links (Paper 0 §2.2; physical shear pairwise-screened) | Intermittent signature; candidate attribution in non-coherent links | Clock-sector artifact |
| Refsdal (Paper 19) | Residual sign (6/7, 7/7 blind); unstretched-space delay +330.9 d sheet, +221.6–443.2 d 3-D ray (Step 64) | Isotropic-conformal stress bounded (κφ ≃ 6.5×10−6); unstretched-space temporal refraction closes non-baryonic delay | Unstretched-space temporal metric + coupled lens-dynamics fit |
Across astrophysical timing channels, the ~106 response scale is
grounded in the ratio of the saturation density ρT ≈ 20 g/cm3
to the mean oscillating-envelope density: in Paper 11 Step 62, the
fundamental-mode Cepheid period–mean-density relation
⟨ρ⟩(P) = ρ☉(Q/P)2 at the P₀ = 10 d pivot gives
ρT/⟨ρ⟩(10 d) = 8.45 × 105, which multiplied by the
Wesenheit Leavitt projector |bW|/ln 10 = 1.416 mag at unit bare
coupling λ₀ = |βA| = 1 predicts
|κP²| = 1.196 × 106 mag — matching the direct SH0ES
fit (1.191 ± 0.480) × 106 mag to 0.5% (λ₀ = 0.995 ± 0.401) and
bracketing the canonical benchmark κgal = 0.96 × 106 mag,
while the saturation scale itself satisfies
ρT ≈ α4/3me4c3/ℏ3
to ~12% (equivalently Λ ≈ α1/3mec2, 3%).
## 7.3 Cosmological Implications
If validated through independent replication, TEP predicts that the phenomenology attributed to dark matter in both the dynamical and optical sectors is the observational projection of Temporal Topology and Temporal Shear across unstretched space, not evidence for a new particulate matter component. In this interpretation, GNSS-scale covariance and galactic dynamical mass discrepancies are different observational projections of the same time-field topology, with Papers 4, 13, 18, and 19 testing the optical and dynamical projections of the same selected two-branch kinetic action $P_{,X}=k\sqrt\xi+2\xi$ ($k=16.03$ fixed by $a_0$, zero per-system free parameters). Under the unstretched-space temporal metric $\tilde g_{\mu\nu}=-A_{\rm loc}^2(u_{\rm grav}(\mathbf x))U_\mu U_\nu+A_{\rm sp}^2(u_{\rm drift}(t,\mathbf x))\Pi_{\mu\nu}$ ($U_\mu\equiv-\nabla_\mu\bar\phi/\sqrt{2\bar X}$), intermediate clock rates $A(\phi(x))$ along photon paths do not cancel against spatial stretching, reproducing the Bullet Cluster galaxy-peak convergence (Paper 4, Step 02), the SN Refsdal Fermat delay and arrival order (Paper 19, Step 64), and the Planck CMB lensing cosmic-web amplification (Paper 18, Step 21) from baryons alone. The 𝒮Σ(Ε) screening operator provides continuous Temporal Topology relaxation from deep potential wells to the weak-field regime: at the terrestrial scale, the density-saturation diagnostic gives RT(M⊕) ≈ 4,150 km — Earth, whose mean density lies below ρT, sits in the partially active regime, while at galactic scales the operator retains approximately 98% of the unsuppressed Temporal Shear in the diffuse halo environment (Paper 26), producing dark-matter-like phenomenology through the active-shear regime rather than through a screening radius. The ~4,000 km GNSS correlation and the ~50 kpc galactic phenomenology are therefore manifestations of the same environment-dependent Temporal-Topology response operating in different density regimes, not a single M1/3 scaling across 15 orders of magnitude.
The framework offers an environmental-response account of the
H0 discrepancy, a source-dependent covariance diagnostic
for weak lensing, and a fixed-calibration unstretched-space temporal
refraction closure across galaxy, cluster, and CMB lensing.
## 7.4 The Path Forward
Independent replication is essential. Building on the raw-RINEX broadcast-SPP confirmation (Paper 3) and the multi-constellation MGEX replication across GPS, GLONASS, Galileo, and BeiDou (Paper 14), the critical next steps are: raw
carrier-phase analysis by independent research groups using the publicly
available benchmark dataset; separately processed per-constellation solutions
to isolate individual orbital shells; lensed FRB timing to probe
achromatic residuals at millisecond precision; and triangle holonomy tests
to directly measure synchronization non-closure. It is stated plainly
that the disformal-sector holonomy — the framework's definitive
loop-level test — is probed in the phase domain by the J0437
phase-closure detection (Paper 16, Step 057): while the $B_0=+1$ unit
benchmark supplies $1.3\%$ ($5.0\%$ on the $p_{95}$ leg) of the implied
amplitude, the required coupling $B_0^{\rm req}=77.3$ (median leg) to
$20.1$ ($p_{95}$ leg) sits within the GW170817 galactic-profile bound
$B_0\lesssim77.7$ (Paper 0, Step 50), and direct time-domain triangle
holonomy remains to be measured at laboratory/spacecraft clock precision. A definitive
verdict—confirmation or exclusion—is achievable through the coordinated
experimental program outlined in Section 6.
## 7.5 Concluding Remarks
The correlation structure documented in this paper is statistically robust
and matches the predictions of a principled theoretical framework grounded
in established scalar-tensor gravity. Whether it ultimately reflects new
physics or reveals subtle systematics yet to be identified, the signal
demands explanation.
For three decades, standard GNSS processing systematically removed a ~4,000
km correlation as part of routine noise filtering. The algorithms were
designed to optimize positioning accuracy rather than to characterize
fundamental physics. The correlation structure reported here was present in
the residuals but categorized as systematic error rather than physical
phenomenon.
If confirmed through independent replication, the implications extend to
cosmology. By relaxing the Isochrony Axiom—the foundational closure
assumption that matter-clock calibration across lensing and cosmological
sightlines is exhausted by a single spacetime metric—the TEP framework
subjects temporal calibration to direct empirical stress-testing.
Under the unstretched-space temporal metric
$\tilde g_{\mu\nu}=-A_{\rm loc}^2(u_{\rm grav}(\mathbf x))U_\mu U_\nu+A_{\rm sp}^2(u_{\rm drift}(t,\mathbf x))\Pi_{\mu\nu}$ ($U_\mu\equiv-\nabla_\mu\bar\phi/\sqrt{2\bar X}$), apparent dark
matter arises in the dynamical sector from Temporal Shear on slow matter
and in the optical sector from intermediate clock-rate refraction
$n(x)\simeq1-2\Phi_N(x)/c^2-\beta_A u_{\rm grav}(x)$ along photon paths across static
space, reproducing galactic rotation curves (Paper 6, Step 10), cluster
convergence (Paper 4, Step 02), strong-lensing Fermat delays (Paper 19,
Step 64), and CMB lensing (Paper 18, Step 21) from baryons and the selected
two-branch action $P_{,X}=k\sqrt\xi+2\xi$ with zero particulate dark matter.
Einstein established that simultaneity is relative. The TEP framework
suggests that simultaneity may be generally non-integrable—that the rate of
proper time accumulation is itself a dynamical field. The speed of light
remains exactly invariant locally, but global synchronization becomes
path-dependent.
The ~4,000 km correlation on Earth (cross-sector mean and directional dispersion λT = 4,201 ± 1,967 km — an order-of-magnitude measurement) and the ~50 kpc
dark-matter-attributed dynamical phenomenology in galaxies are interpreted within TEP as manifestations of the same environment-dependent Temporal-Topology response, pending independent replication, completion-dependent transfer calculations, and lensing–dynamics consistency (Paper 19), connected by the continuous relaxation of Temporal
Topology from deep potential wells to the weak-field regime. The M1/3
scaling appearing in some candidate completions remains testable through
interplanetary missions and pulsar timing arrays, but it is not the primary empirical claim. This hypothesis is
falsifiable through the experimental program outlined in Section 6.
Independent replication is essential, and the data and analysis code are
publicly available. The community is invited to verify, challenge, or refute
these findings.
## Data and Code Availability
All analysis code, processed data products, and raw RINEX processing
pipelines are publicly available:
Multi-center and 25-year CODE analysis:
github.com/matthewsmawfield/TEP-GNSS
Raw RINEX test:
github.com/matthewsmawfield/TEP-GNSS-RINEX
Benchmark dataset: A simplified CSV file containing binned residuals
versus distance for the 25-year period, available for theorists who wish
to fit models without processing raw GNSS data:
TEP-GNSS/benchmark
## Acknowledgments
The author expresses gratitude to the International GNSS Service (IGS), the
Center for Orbit Determination in Europe (CODE), the European Space Agency
(ESA), and NASA's Crustal Dynamics Data Information System (CDDIS). This
analysis was possible because these organizations have championed the
principles of open science, making decades of high-precision timing data
freely available to the global research community. The GFZ Helmholtz Centre
Potsdam is acknowledged for providing geomagnetic Kp indices.
## References
Abbott, B. P. et al. (LIGO/Virgo) (2017). GW170817: Observation of gravitational waves from a binary neutron star inspiral. *Phys. Rev. Lett.* 119, 161101.
Ai, Q., Maciuk, K., Lewinska, P. & Borowski, L. (2021). Characteristics of Onefold Clocks of GPS, Galileo, BeiDou and GLONASS Systems. *Sensors* 21(7), 2482. DOI: [10.3390/s21072482](https://doi.org/10.3390/s21072482)
Ashby, N. (2003). Relativity in the Global Positioning System. *Living Rev. Relativity* 6, 1.
Bekenstein, J. D. (1993). The relation between physical and gravitational geometry. *Phys. Rev. D* 48, 3641.
Bothwell, T. et al. (2022). JILA Sr optical lattice clock with 10−18 stability and accuracy. *Nature* 602, 420–424.
Boulder Atomic Clock Optical Network (BACON) Collaboration (2021). Frequency ratio measurements at 18-digit accuracy using an optical clock network. *Nature* 591, 564–569.
Burrage, C. & Sakstein, J. (2018). Tests of chameleon gravity. *Living Rev. Relativity* 21, 1.
Damour, T. & Polyakov, A. M. (1994). The string dilaton and a least coupling principle. *Nucl. Phys. B* 423, 532.
Exertier, P., Belli, A., Lemoine, F. G., Biancale, R. & Courde, C. (2017). Suspected time errors along the satellite laser ranging network and impact on the reference frame. *EGU General Assembly* 19, 9405.
Fathollahi, L., Wu, F., Pongracic, B., Kiani Shahvandi, M., Schlichting, D. & Sustersic, N. (2023). Gravitational redshift test using Rb clocks of eccentric GPS satellites. *Heliyon* 9(8), e18532. DOI: [10.1016/j.heliyon.2023.e18532](https://doi.org/10.1016/j.heliyon.2023.e18532)
Hofmann-Wellenhof, B., Lichtenegger, H. & Wasle, E. (2008). *GNSS – Global Navigation Satellite Systems*. Springer.
Kelly, A. G. (2009). GPS and the One-Way Speed of Light. In: *New Approach of Indoor and Outdoor Localization Systems*. IntechOpen. DOI: [10.5772/11006](https://doi.org/10.5772/11006)
Khoury, J. & Weltman, A. (2004). Chameleon fields: Awaiting surprises for tests of gravity in space. *Phys. Rev. Lett.* 93, 171104.
Milgrom, M. (1983). A modification of the Newtonian dynamics as a possible alternative to the hidden mass hypothesis. *ApJ* 270, 365.
Petit, G. & Luzum, B. (2010). IERS Conventions (2010). *IERS Technical Note* 36.
Planck Collaboration (2020). Planck 2018 results. VI. Cosmological parameters. *A&A* 641, A6.
Riess, A. G. et al. (2022). A comprehensive measurement of the local value of the Hubble constant. *ApJ* 934, L7.
Santamaria-Gomez, A., Ray, J. & Collilieux, X. (2017). GNSS related periodic signals in coordinate time-series from Precise Point Positioning. *Geophysical Journal International* 208(3), 1449–1462. DOI: [10.1093/gji/ggw454](https://doi.org/10.1093/gji/ggw454)
Smawfield, M. L. (2025). *Temporal Equivalence Principle: Dynamic Time & Emergent Light Speed*. Preprint v0.16 (Jakarta). Zenodo. DOI: [10.5281/zenodo.16921911](https://doi.org/10.5281/zenodo.16921911) (Paper 0)
Smawfield, M. L. (2025). *Global Time Echoes: Distance-Structured Correlations in GNSS Clocks*. Preprint v0.28 (Jaipur). Zenodo. DOI: [10.5281/zenodo.17127229](https://doi.org/10.5281/zenodo.17127229) (Paper 1)
Smawfield, M. L. (2025). *Global Time Echoes: 25-Year Analysis of CODE Precise Clock Products*. Preprint v0.21 (Cairo). Zenodo. DOI: [10.5281/zenodo.17517141](https://doi.org/10.5281/zenodo.17517141) (Paper 2)
Smawfield, M. L. (2025). *Global Time Echoes: Raw RINEX Consistency Test*. Preprint v0.9 (Kathmandu). Zenodo. DOI: [10.5281/zenodo.17860166](https://doi.org/10.5281/zenodo.17860166) (Paper 3)
Smawfield, M. L. (2025). *Temporal-Spatial Coupling in Gravitational Lensing: A Reinterpretation of Dark Matter Observations*. Preprint v0.9 (Tortola). Zenodo. DOI: [10.5281/zenodo.17982540](https://doi.org/10.5281/zenodo.17982540) (Paper 4)
Smawfield, M. L. (2025). *Global Time Echoes: Empirical Synthesis*. Preprint v0.8 (Singapore). Zenodo. DOI: [10.5281/zenodo.18004832](https://doi.org/10.5281/zenodo.18004832) (Paper 5 — this work)
Smawfield, M. L. (2025). *Temporal Topology Saturation Scale: Cross-Scale Consistency of ρ_T*. Preprint v0.9 (New Delhi). Zenodo. DOI: [10.5281/zenodo.18064365](https://doi.org/10.5281/zenodo.18064365) (Paper 6)
Smawfield, M. L. (2025). *The Soliton Wake: Exploring RBH-1 as a Temporal Topology Candidate*. Preprint v0.5 (Blantyre). Zenodo. DOI: [10.5281/zenodo.18059250](https://doi.org/10.5281/zenodo.18059250) (Paper 7)
Smawfield, M. L. (2025). *Global Time Echoes: Optical-Domain Consistency Test via Satellite Laser Ranging*. Preprint v0.7 (Mombasa). Zenodo. DOI: [10.5281/zenodo.18064581](https://doi.org/10.5281/zenodo.18064581) (Paper 8)
Smawfield, M. L. (2025). *What Do Precision Tests of General Relativity Actually Measure?*. Preprint v0.9 (Istanbul). Zenodo. DOI: [10.5281/zenodo.18109760](https://doi.org/10.5281/zenodo.18109760) (Paper 9)
Smawfield, M. L. (2026). *Temporal Equivalence Principle: Suppressed Density Scaling in Globular Cluster Pulsars*. Preprint v0.10 (Caracas). Zenodo. DOI: [10.5281/zenodo.18165798](https://doi.org/10.5281/zenodo.18165798) (Paper 10)
Smawfield, M. L. (2026). *The Cepheid Bias: Resolving the Hubble Tension*. Preprint v0.12 (Kingston upon Hull). Zenodo. DOI: [10.5281/zenodo.18209702](https://doi.org/10.5281/zenodo.18209702) (Paper 11)
Smawfield, M. L. (2026). *Temporal Equivalence Principle: A Unified Resolution to the JWST High-Redshift Anomalies*. Preprint v0.8 (Kos). Zenodo. DOI: [10.5281/zenodo.19000827](https://doi.org/10.5281/zenodo.19000827) (Paper 12)
Smawfield, M. L. (2026). *Temporal Equivalence Principle: Temporal Shear Recovery in Gaia DR3 Wide Binaries*. Preprint v0.7 (Kilifi). Zenodo. DOI: [10.5281/zenodo.19102061](https://doi.org/10.5281/zenodo.19102061) (Paper 13)
Smawfield, M. L. (2026). *Global Time Echoes: MGEX Multi-GNSS Clock Replication*. Preprint v0.4 (Suva). Zenodo. DOI: [10.5281/zenodo.20572726](https://doi.org/10.5281/zenodo.20572726) (Paper 14)
Smawfield, M. L. (2026). *Temporal Equivalence Principle: Temporal Shear in the Earth Flyby Anomaly*. Preprint v0.4 (Yogyakarta). Zenodo. DOI: [10.5281/zenodo.19454862](https://doi.org/10.5281/zenodo.19454862) (Paper 15)
Smawfield, M. L. (2026). *Temporal Equivalence Principle: Lunar Laser Ranging and the Nordtvedt Effect*. Preprint v0.5 (Lucknow). Zenodo. DOI: [10.5281/zenodo.19446028](https://doi.org/10.5281/zenodo.19446028) (Paper 17)
Smawfield, M. L. (2026). *Temporal Equivalence Principle: A Blind-Prediction Residual Test in Multiply-Imaged Supernovae*. Preprint v0.3 (Lisboa). Zenodo. DOI: [10.5281/zenodo.20572719](https://doi.org/10.5281/zenodo.20572719) (Paper 19)
Smawfield, M. L. (2026). *Temporal Equivalence Principle: A Covariant Alternative to Cosmic Expansion*. Preprint v0.4 (Athens). Zenodo. DOI: [10.5281/zenodo.20370143](https://doi.org/10.5281/zenodo.20370143) (Paper 26)
Smawfield, M. L. (2026). *TEP-PPP: Operational Time Echoes — Phase-Resolved Spatial Clock Priors for GNSS Outage Bridging*. Preprint v0.1 (Pune). In preparation. (Paper 33)
Teunissen, P. J. G. & Montenbruck, O. (Eds.) (2017). *Springer Handbook of Global Navigation Satellite Systems*. Springer.
Touboul, P. et al. (2022). MICROSCOPE mission: test of the equivalence principle in space. *Phys. Rev. Lett.* 129, 121102.
Will, C. M. (2014). The Confrontation between General Relativity and Experiment. *Living Rev. Relativity* 17, 4.
## Data Availability & Reproducibility
This work follows open-science practices. This synthesis paper integrates results from a systematic
multi-paper research program. All primary analysis code and data are available in the referenced
repositories.
### Repository & Code
**GitHub Repository:** [github.com/matthewsmawfield/TEP-GTE](https://github.com/matthewsmawfield/TEP-GTE)
This synthesis manuscript integrates findings from Papers 1–4. Primary analysis repositories:
#### TEP Research Program Structure
| Paper | Repository | Download Size |
| --- | --- | --- |
| Paper 1: GNSS Analysis | [TEP-GNSS](https://github.com/matthewsmawfield/TEP-GNSS) | ~50 MB (benchmark data) |
| Paper 2: 25-Year Longspan | [TEP-GNSS-II](https://github.com/matthewsmawfield/TEP-GNSS-II) | ~100 MB |
| Paper 3: Raw RINEX | [TEP-GNSS-RINEX](https://github.com/matthewsmawfield/TEP-GNSS-RINEX) | ~5 GB (RINEX files) |
| Paper 4: Gravitational Lensing | [TEP-GL](https://github.com/matthewsmawfield/TEP-GL) | N/A (theoretical) |
| Paper 8: Satellite Laser Ranging | [TEP-SLR](https://github.com/matthewsmawfield/TEP-SLR) | ~200 MB (normal-point data) |
### Primary Data Sources
- **CDF Data:** CODE analysis center clock products (2000–2025)
- **IGS Data:** International GNSS Service clock products
- **ESA Data:** European Space Agency navigation support products
- **RINEX Files:** Raw observation files from IGS stations
### Prediction Provenance
The dated prediction register of Section 6.5 is distributed as a
machine-readable artifact,
`results/prediction_provenance.json`, recording each tested
expectation's earliest versioned statement with its version DOI and
immutable deposit timestamp. The ordering is independently verifiable
against the public Zenodo version records under the concept DOIs listed
there (Paper 0: 10.5281/zenodo.16921911; Paper 1:
10.5281/zenodo.17127229).
### Reproduction Instructions
```
# Clone this synthesis repository
git clone https://github.com/matthewsmawfield/TEP-GTE.git
cd TEP-GTE
# Build manuscript
cd site
npm install
npm run build
```
### Software Versions
- **Node.js** 16+
## Appendix A: Mathematical Formalism
## A.1 Full Action
For comparison, the canonical scalar baseline of the TEP action (Paper 0, §2.2) is formulated in the Einstein frame where gravity is canonical:
$$S = \int d^4x \sqrt{-g} \left[ \frac{M_{\text{Pl}}^2}{2} R - \frac{1}{2} K(\phi)\, g^{\mu\nu} \partial_\mu \phi\, \partial_\nu \phi - V(\phi) \right] + S_{\text{matter}}[\psi, \tilde{g}_{\mu\nu}, \phi]$$
where MPl is the reduced Planck mass, R is the Ricci scalar, K(φ) is the kinetic function (absorbable by a field redefinition; the operative screening realization is the noncanonical two-branch $P(X,\phi)=-V(\phi)+\frac{2k}{3\Lambda_X^2}X|X|^{1/2}+X|X|/\Lambda_X^4$ with derived $k\simeq16$), V(φ) is the screening potential, and Smatter is the matter action in the Jordan frame with coupling to the matter metric g̃μν.
## A.2 Field Equations
Variation with respect to the gravitational metric yields the Einstein equations:
$$G_{\mu\nu} = \frac{1}{M_{\text{Pl}}^2} \left[ T_{\mu\nu}^{(\phi)} + T_{\mu\nu}^{(\text{matt})} \right]$$
where the scalar field stress-energy tensor is:
$$T_{\mu\nu}^{(\phi)} = K(\phi) \partial_\mu \phi \partial_\nu \phi - g_{\mu\nu} \left[ \frac{1}{2} K(\phi) g^{\rho\sigma} \partial_\rho \phi \partial_\sigma \phi + V(\phi) \right]$$
Variation with respect to the scalar field yields:
$$\nabla_\mu \left[ K(\phi) \nabla^\mu \phi \right] - \frac{\partial K}{\partial \phi} \frac{1}{2} g^{\mu\nu} \partial_\mu \phi \partial_\nu \phi - \frac{\partial V}{\partial \phi} = -\alpha(\phi) T + S_{\text{disf}}$$
where α(ϕ) ≡ d ln A/dϕ = βA/MPl is the conformal coupling strength (for A(ϕ) = exp(βAϕ/MPl)), T is the trace of the matter stress-energy tensor in the Jordan frame, and Sdisf contains disformal source terms proportional to B(ϕ).
## A.3 Candidate Microscopic Completions and Nonlinear Screening
**Candidate Completion (Historical Provenance):** This appendix presents an illustrative candidate completion (cubic Galileon derivative screening) that exhibits the (βAM/MPlΛ3)1/3 transition scale. In the current TEP master architecture (Paper 0 §2.2), environmental screening is derived from the selected two-branch non-canonical kinetic operator $P(X,\phi) = -V(\phi) + \frac{2k}{3\Lambda_X^2}X|X|^{1/2} + X|X|/\Lambda_X^4$.
The following derivation illustrates one candidate microscopic completion (cubic
Galileon) that realizes the continuous gradient suppression described as
Temporal Shear in Section 4; it is not required for the empirical results.
One candidate completion invokes cubic Galileon terms in the action to
generate nonlinear screening:
$$S_{\text{Galileon}} = \int d^4x \sqrt{-g} \left[\frac{c_3}{\Lambda^3} (\partial\phi)^2 \Box\phi\right]$$
where Λ is the strong-coupling scale (an energy scale, [M] in natural units c = ℏ = 1) and c3 is a dimensionless coefficient of order unity. With the canonical scalar field normalization (φ has mass dimension 1) and the TEP source term (βA/MPl)T, the nonlinear term becomes important when:
$$\frac{|\nabla \phi|^2}{\Lambda^3} \sim 1$$
The Vainshtein radius, inside which nonlinear effects dominate, is (in natural units c = ℏ = 1):
$$r_V = \left(\frac{\beta_A M}{M_{\text{Pl}} \Lambda^3}\right)^{1/3}$$
where βA is the conformal coupling, MPl = √(ℏc/G) is the Planck mass, and Λ is the strong-coupling energy scale. Restoring SI units:
$$r_V = \left(\frac{\beta_A M (\hbar c)^3}{M_{\text{Pl}} \Lambda^3}\right)^{1/3} = \left(\frac{\beta_A M}{M_{\text{Pl}}}\right)^{1/3} \frac{\hbar c}{\Lambda}$$
The factor βAM/MPl supplies the dimensionless source strength; (ℏc/Λ)3 supplies the length scale cubed. The bare expression (GM/Λ3)1/3 sometimes quoted in the Galileon literature is a shorthand that suppresses the βA/MPl coupling and implicitly assumes natural units with a specific field normalization; it is dimensionally incomplete as written.
For a spherically symmetric source of mass M, the scalar field profile in the Vainshtein regime (r $\lt$ rV) is:
$$\phi(r) \sim \frac{\beta_A M_{\text{Pl}}}{M_{\text{Pl}}^2} \frac{GM}{r} \left(\frac{r}{r_V}\right)^3$$
This suppresses the Temporal Shear by a factor of (r/rV)3 inside the screening radius, reconciling weak-field cosmological behavior with strong-field solar system constraints. Other completions (e.g., chameleon mechanisms, symmetry-breaking potentials) may produce similar phenomenology.
## A.4 Conservation Laws
The total stress-energy tensor is conserved:
$$\nabla_\mu \left[ T^{\mu\nu}_{(\phi)} + T^{\mu\nu}_{(\text{matt})} \right] = 0$$
However, the matter stress-energy alone is not conserved in the Einstein frame:
$$\nabla_\mu T^{\mu\nu}_{(\text{matt})} = \alpha(\phi) T \nabla^\nu \phi + \text{disformal terms}$$
This apparent violation is resolved by transforming to the Jordan frame, where matter follows geodesics of g̃μν and energy-momentum is conserved with respect to the matter-frame covariant derivative.
## Appendix B: Processing Filter Mathematical Formulation
## B.1 Network Adjustment Constraint
Standard GNSS processing applies a datum constraint that forces residuals to have zero spatial mean:
$$\delta t_i^{\text{adj}} = \delta t_i - \frac{1}{N}\sum_{j=1}^N \delta t_j$$
This constraint preserves zero-mean residuals while removing any common-mode offset.
## B.2 Fourier Decomposition
Decompose the TEP signal into spatial Fourier modes characterized by wavevector k:
$$\delta t_i^{\text{TEP}} = \sum_{\mathbf{k}} A_{\mathbf{k}} e^{i\mathbf{k} \cdot \mathbf{r}_i}$$
where ri is the position of station i. The network adjustment removes the k = 0 (monopole) component:
$$A_{\mathbf{k}=0}^{\text{adj}} = A_{\mathbf{k}=0} - \frac{1}{N}\sum_j A_{\mathbf{k}=0} = 0$$
but preserves all k ≠ 0 (dipole, quadrupole, and higher) components that have zero spatial mean.
## B.3 Cross-Spectral Density
The TEP correlation structure manifests as phase coherence between station pairs, quantified by the cross-spectral density:
$$C_{ij}(\omega) = \langle \tilde{\delta t}_i(\omega) \tilde{\delta t}_j^*(\omega) \rangle$$
where tildes denote Fourier transforms and angle brackets denote ensemble averaging.
## B.4 Phase Alignment Index
The magnitude-weighted phase alignment index extracts the phase relationship:
$$\Phi_{ij} = \frac{\sum_\omega |C_{ij}(\omega)| \cos[\arg C_{ij}(\omega)]}{\sum_\omega |C_{ij}(\omega)|}$$
This metric is invariant under common-mode amplitude rescaling because it depends only on relative phases, not absolute amplitudes. The datum constraint removes Ak=0 (amplitude information) but preserves the phase structure encoded in the k ≠ 0 components.
## B.5 Exponential Correlation Function
The exponential decay C(r) ∝ exp(−r/λ) reflects the two-point correlation function of the underlying φ field:
$$\langle \phi(\mathbf{x}) \phi(\mathbf{x}') \rangle = \phi_0^2 \exp\left(-\frac{|\mathbf{x} - \mathbf{x}'|}{\lambda}\right)$$
where λ = RT(M⊕) is the geometric saturation radius (§4.4.2); under the mass-completion interpretation this would correspond to the Compton wavelength λ = ℏ/(mφc). This correlation function has nonzero spatial gradients (k ≠ 0 in Fourier space) and therefore survives network adjustment, which only removes the k = 0 component.
## B.6 Tidal vs. TEP Signal Discrimination
Classical tidal effects are parameterized as sums of spherical harmonics Ylm(θ, φ) with known frequencies:
$$\delta t_{\text{tidal}}(t, \mathbf{r}) = \sum_{l,m,n} A_{lmn} Y_{lm}(\theta, \phi) \cos(\omega_n t + \phi_n)$$
where ωn are tidal frequencies (diurnal, semidiurnal, long-period). Standard GNSS processing explicitly models and subtracts these terms using IERS conventions.
The TEP correlation is a distance-dependent phase coherence:
$$C_{\text{TEP}}(r_{ij}) = A_0 \exp(-r_{ij}/\lambda)$$
This functional form is not decomposable into the standard tidal spherical harmonic basis. The processing removes what it models (tidal harmonics) but preserves what it does not model (exponential distance correlations in phase space).