33.35 turns the global atomic-clock height network into a potential-scaled common-drift court: after standard gravitational-redshift subtraction, only a residual drift that scales monotonically with ΔU, co-occurs at zero lag, stays direction-consistent across atomic species and transfer links, preserves frequency-ratio stability, and reproduces across continents and pipelines can survive; under V08-compatible tighten, this is only a time-frequency network ledger under a potential external parameter, not a verdict on time ontology or universal propagation.
Use this section as a compact machine-readable EFT reference.
33.35 recasts the global atomic-clock height network from an exercise in ever finer comparison into a residual common-drift court under gravitational potential difference. The key is not that clocks are more precise, but whether anything survives standard gravitational-redshift subtraction as one cross-system pattern; under compat adjudication, the chapter is tighten.
The chapter compresses the readout into five linked tests: a potential-scaled common-drift index, zero-lag co-occurrence, cross-species and cross-link dispersion-free consistency, frequency-ratio stability, and seasonal-to-multi-year repeatability. What matters is not one drifting station, but whether residuals scale with ΔU while ratios stay stable.
The workflow must run multiple atomic species, multiple link families, and multiple continents in parallel, while keeping clock-side corrections, link-side corrections, potential and height models, and timescale alignment as separate ledgers. Feed-forward teams may only use potential, load, deformation, and space-weather masks to issue drift cards, and measurement teams may not delete links, change windows, or swap model families after unblinding.
False structure must be broken by station-label, link-label, and time-window permutations, by equipotential-surface controls, and by same-site dual-clock or same-link round-trip loops. Link thermal drift, atmospheric and ionospheric corrections, clock seasonality, and geoid or load-model error are old ledgers that have to be peeled away before a potential-scaled common-drift claim remains.
The pass line is severe: at least two atomic species, two transfer-link types, three continents, and two independent pipelines must recover a zero-lag residual drift that scales monotonically with gravitational potential difference while frequency ratios remain stable. If the effect follows dispersive or link-specific laws, depends on one season or one alignment choice, or fails on held-out units, the chapter must be falsified.
The chapter ends with only one residual ledger under a potential external parameter, not a verdict on the ontology of time, constants, or universal propagation. If the network-level common drift survives, it sharpens the first-line common-term court; if not, it goes back to engineering, correction-model, and clock-system ledgers, then forwards to the fiber-corridor audit in 33.36.