All 54 Newtonian GJ 367 c-d grid models with 300-year angle confinement fail the 10,000-year test

GJ 367 c and d have orbital periods near a 3:1 ratio. We test whether short angle confinement persists in specified coplanar Newtonian models. The main models contain the star and planets c and d. A 21 by 21 period grid uses the reported 2023 minimum planet masses, central eccentricities, and orbital angles. Of 441 models, 54 meet a fixed angle-span rule after 300 years. All 54 fail the same rule over 10,000 years: four circulate and 50 are unresolved. Each tested angle then spans more than one cycle. A control with the same periods, masses, and eccentricities, but different angles, keeps one angle within a span of 0.263 cycle over 100,000 years at both tested time steps. Independent-integrator and time-step checks agree on the short classifications. Two synthetic velocity comparisons at 370 public HARPS observation times have weighted RMS differences near 0.024 m/s. No measured velocity is fitted. The grid has no posterior weights. Separate Kepler summaries are treated as osculating elements at one epoch. General relativity, tides, and the unknown three-dimensional geometry are omitted. The result shows a duration limit of the angle test in these models. It does not give a resonance probability for the observed system. The package contains code, frozen public inputs, source records, initial Cartesian states, numerical results, and figures.

GJ 367 · orbital dynamics · 3:1 resonance · angle confinement · numerical integration · public data · reproducible research