This paper proposes a specific physical mechanism for the terminus of spacetime collapse under gravitational compression. If spacetime is a structured geometric medium — a 24-cell lattice with node spacing at the Planck length — then gravitational collapse does not produce a mathematical singularity. It produces a crystal. The 24-cell lattice, the unique regular polytope in four dimensions and the densest sphere packing in 4D, achieves a packing density of π²/16 ≈ 0. 617. The crystalline terminus density is therefore (π²/16) × ρPlanck ≈ 3. 18 × 10⁹⁶ kg/m³ — approximately 61. 7% of the Planck density. Below this density, the lattice compresses. At this density, the lattice nodes are touching and no further compression is possible. The medium bounces. This proposal is placed in direct competition with the Planck Star hypothesis (Rovelli and Vidotto, 2014), which proposed that quantum gravitational pressure halts black hole collapse at the Planck density and predicted that Fast Radio Bursts may be the signals from bouncing primordial black holes. The KishLattice Star agrees with the physical picture — collapse terminates, bounce occurs, signal is emitted — but provides a specific geometric structure for the terminus and specific harmonic predictions for the bounce signal that distinguish it from the Planck Star model. Two independent empirical handles are identified in currently available public data. First, the LIGO gravitational wave ringdown residual: after the standard quasi-normal mode ringdown is subtracted from confirmed binary merger events in the GWTC-3 catalog, the framework predicts a persistent low-amplitude oscillation whose frequency, when scalarized through the KLGHS transform, maps to a confirmed harmonic register — specifically 16/π (the kinematic primary confirmed at z = +99. 63 from 1. 8 million stellar velocity measurements) or 21/π (the structural binding register confirmed by nuclear decay and galaxy velocity dispersions). Second, the Fast Radio Burst friction model: an existing two-parameter model F (d, T) = 1/ (1 + k·d· (T₀/T) ^α) fitted to four confirmed FRB repeaters yields k = 1. 376 and α = 0. 683, within 2. 5% of the Kolmogorov turbulence exponent 2/3. The paper pre-registers a formal derivation target: α = 2/3 should emerge from the 24-cell lattice's coordination structure (8 nearest neighbours per vertex, 24 total vertices, F4 symmetry group) without free parameters. V4 correction note: the α = 2/3 derivation target described above, Pₐlphaderivation, is a separate open theoretical thread and is not part of the V4 coherence pre-registration described below. It is retained here as original V1 context and has not been advanced in V2, V3, or V4. The paper pre-registers five formal predictions with full methodology and honest if-confirmed and if-null statements: gravitational wave ringdown residual floor at a KLGHS harmonic register (Pₗigoₜerminus) ; FRB friction parameter stability across 15+ confirmed repeaters (Pfrbₜerminus) ; FRB inter-burst timing lock at KLGHS harmonic registers (Pfrbₕarmonics) ; the zero-free-parameter derivation of α = 2/3 from 24-cell lattice stiffness (Pₐlphaderivation) ; and GWTC-4 gravitational wave frequency domain confirmation of 16/π (Pₗigogwtc4). Two complete sovereign lake designs are provided with promoted file schemas and scalarization formulas ready for the KLGHS pipeline. V4 correction note: Pₗigogwtc4 as named above referred to a future, not-yet-released catalog and was never executed under that name. The LIGO test actually pre-registered and now in progress as of V4 is Pₗigoₜerminusᵥ1, a cross-spectral H1-L1 coherence analysis against the existing GWTC-3 catalog. See the V4 section of this record's attached paper for the current, accurate specification: pre-registration filed June 2026, DOI 10. 5281/zenodo. 20711135, lake build authorised, pipeline not yet executed, no coherence data ingested as of this version's timestamp. The framework connecting the predictions to the observables is KishLattice Geometric Harmonic Spectroscopy, established across ten prior volumes and 13. 7 million sovereign physical measurements spanning sub-nuclear particle masses to cosmological distance distributions. The geometric modulus kgeo = 16/π = 5. 09295817 is confirmed STRONG across 31 independent physical domains. The 24-cell's vertex count (24) is the CONTAINER constant in the chaos null modular resonance test that produces every z-score in the series. The crystal limit is calculated. The bounce frequency has harmonic structure. The signal has been in the data since September 14, 2015. All data sources (LIGO Open Science Center GWTC-3, CHIME/FRB Catalog 1) are public. All methodology is pre-registered. The prediction record is timestamped by this Zenodo publication. The convergence test is stated precisely: if the LIGO ringdown floor frequency and the FRB friction depth parameter convert to the same terminus density in common physical units, the crystal is measured by two independent instruments. V4 status note: this version adds the V2 FRB Star Probe results (Dispersion Measure converging at 25/π, z = +22. 3; inter-burst timing converging at 25/π after contamination audit and clean 403-record re-run, z = +12. 7) and the V3/V4 LIGO cross-spectral coherence pre-registration. As of this timestamp, the coherence pipeline has not been run and no coherence data has been ingested. See the attached document for full detail. V5 status note: this version sealed the dual-lane multi-detector coherence architecture (Pₗigoₘultidetectorᵥ1) after an empirical audit found Virgo's (V1) noise floor in the target band to be approximately 35x louder than LIGO's, requiring an independently derived Virgo-specific threshold rather than reuse of the H1-L1 threshold. The corrected dual-lane design was pre-registered, DOI 10. 5281/zenodo. 20766403, with the original H1-L1 threshold left unmodified as Lane 1 and Virgo corroboration added only as a non-exclusionary Lane 2 annotation. As of this version's timestamp, the corrected pipeline had not yet been executed against real merger data. [V6 status note, FINAL RESULT: the dual-lane pipeline was executed and returned a 71-of-74-event pass rate inconsistent with the pre-registered 95th-percentile threshold (the V9 single-lane run, using the identical threshold, had passed only 15 of 86). Direct code review located a multiple-comparisons (look-elsewhere) defect: the scalarized field ("frequency of maximum excess coherence") had been used both as a measurement definition and, separately, as a per-event admission gate, inflating the effective false-positive rate far above the intended 5%. A control test against pure pre-merger noise confirmed this directly: 100% false-positive rate, with no physical signal of any kind present in the test data. The architecture was corrected to remove all per-event gating from Lane 1 -- admission now depends only on data quality (H1 and L1 both present, no NaNs) -- with STRONG/null determination deferred entirely to the standard three-tier KLGHS chaos null, consistent with how every other domain in this series is evaluated. The corrected pipeline was validated clean on the same noise-only control (z = +5. 0. Per the falsification condition stated in the V3/V4/V5 pre-registration itself, this result constitutes falsification of the specific Pₗigoₜerminusᵥ1 prediction (persistent kinematic-register coherence signal at 16/pi). Combined with the V2 FRB result (strong signal at 25/pi, the electromagnetic propagation register, not at the predicted kinematic 16/pi), both pre-registered empirical handles in this paper have now reported, and neither confirms the specific 16/pi crystalline terminus mechanism as formally specified in V1 through V5. This falsifies the specific mechanism proposed in this paper. It does not falsify, and is independent of, the 25-million-record, 33+ STRONG-signal empirical base of KishLattice Geometric Harmonic Spectroscopy documented across Volumes 5 through 11, which rests on its own separate pre-registrations and chaos nulls. Full chronicle, including the look-elsewhere defect, the smoke-alarm control test, and the corrected architecture, is documented in the attached V6 paper. DOI for this version: DOI 10. 5281/zenodo. 20777524.
Kish et al. (Sat,) studied this question.