Hello, I did not care much for that point, believing it was an easy matter. Now I realise that I simply don't know and can't prove why there are only five collisional invariants. This point is crucial to derive the Maxwell distribution from the Boltzmann equation. For sure this must have been well known 160 years ago, yet I don't have any clue. Could you relief me from this bad feeling? Thanks, Michel PS (more): In addition I wonder if there is any relation between being a "collisional invariant" and being an additive constant of motion. Being an additive constant of motion is the key point to derive the canonical ensemble.