Entanglement is a quantum resource, but possessing it does not always mean we can turn it into the form we want. Distillation asks whether separated experimenters, using local operations and classical messages, can extract useful entangled pairs from a noisy shared state. This candidate studies a particularly symmetric family called Werner states. Each particle has three levels, and the experimenters are given three copies of the state. The result identifies an exact boundary. In the usual parameter convention, three-copy distillation is possible below minus one half. At minus one half, and above it, the candidate proves that three copies cannot pass the mathematical distillation test. The boundary case includes every allowed complex rank-two test, without assuming a special symmetry or restricted support. The evidence is an exact certificate. Instead of relying on a numerical search that failed to find a counterexample, the package checks seven hundred and sixteen positive semidefinite multipliers and an identity with two thousand three hundred and seventy-four coordinates. The calculations use exact arithmetic. Deliberately corrupted inputs test whether the checker detects important errors. A supplied audit supports the central argument using alternative arithmetic checks. The revised paper explains that audit's limits, adds the relevant literature, and separates optional historical construction claims from the proof that readers can replay. This remains a mathematical candidate. It has not received authenticated external specialist peer review or end-to-end proof-assistant verification. Most importantly, three copies do not mean any number of copies. The broader question remains unresolved by this work. This is Evidence Press, Three copies, the same distillation boundary, dated the fourth of October twenty twenty-six. The paper and exact evidence package are linked on the release page. This synthetic AI voice explains the research; it is not additional scientific evidence.