What survives when gyro observations are omitted?

This acceleration-magnitude constraint excludes none of the candidates. All 27,468 group cases overlap under the declared assumptions. A single zero gain/offset correction remains unexcluded in every assessed group for all 24 recording/clock/receiver-error/candidate histories. No per-window calibration adjustment was needed for that necessary-constraint result.

An IMU (inertial measurement unit) records turning with gyroscopes and specific force with accelerometers. This test uses only the latter and receiver positions. It does not replace the omitted turning observations with fused attitude, assumed heading or an expected rotation field measured by the gyro.

Same recordings, broad history coverage

Every group contains the same 200 original packets under both clock hypotheses. Header elapsed durations total 552.0355 and 211.0002 seconds respectively. The discarded edges lack the required ±16-second receiver stencil; outside- interval and partial-tail groups remain listed in the inventory. The counts describe adjacent groups, not independent observations or flights.

The two recordings retain their previously selected north/south and IMU8/IMU21 coverage. Their conservative provenance groups are recorded in the manifest; physical installation identity is not inferred from configuration bytes. This is retrospective development, not untouched confirmation.

What this adds to the snapshot study

The earlier check examined only three short windows per recording and relaxed the shared-calibration dependence. This continuation covers the supported history, keeps the same packets under both clocks and tests one unchanged calibration point across all groups. Receiver estimates have a bounded central- to-continuous-average correction; they are not silently treated as exact packet averages. No scientific measurement is interpolated.

The declared error envelopes are intentionally conservative. They retain receiver-coordinate error and polynomial truncation, gravity, count resolution, clock offset, force averaging and calibration alternatives. They are sensitivity assumptions, not measured per-unit specifications or calibrated confidence limits.

What it does not establish

The common zero-calibration point passes these relaxed scalar constraints. Shared receiver perturbations, gravity and latency are not jointly solved. Thus this is neither a full feasible gyro-free trajectory nor proof that all three physical models explain the entire measurement record equally well. It supplies no percentage of evidence attributable to gyroscopes.

The complete gyro-driven reconstruction favors globe on both recordings and across all 53 selected stretches. This reduced constraint does not reproduce that discrimination. Predictive testing and measurement uncertainty remain necessary for a stronger attribution of the full method's separation.

Verification and next stage

All 3,052 calculation identities passed, with no missing or unsupported group cases. The study used 3,052 receiver operators and 9,156 force boxes in 185 seconds; it used no new fits, native integrations or original-file reads. Twenty-two focused controls qualify the new group/averaging/worker paths. The unlaunched earlier preparation and every frozen source/input remain preserved.

The contribution diagnostic can now move to prediction and covariance work, rather than another search for scalar separation under narrower chosen bounds. The earlier matched prediction study completed 10/12 fits and retained an unresolved aggregate result; its local uncertainty excluded bounded calibration and covariance-estimation uncertainty. That is the next methodological gap.

Results · Every scalar case · Common-zero checks · Frozen method · Earlier prediction study · Roadmap.