Selection exists, but a heritable trait is not selected: a negative result with a measured boundary

Article 032 ended with a limit: a cell that moves 0.14 cells per lifetime cannot use long-range information. That leaves a prior question, and this article answers it in two parts. First, is there anything in this world to select on at all? Yes, strongly: surviving a 20-step horizon is 0.551 in the low energy quartile against 0.686 in the high one (+0.135), and 0.560 against 0.667 for age (+0.106), and both reproduce on ten fresh seeds. Second, can a heritable trait ride that selection? No, and the boundary is cost: with the trait costing nothing the shift is -0.00002 in the rich zone and -0.0016 in the poor one, so there is no selection at all; with the trait costing something the shifts are weak and not significant (|t| between 0.60 and 1.83 on five seeds), and their direction is not established. The article is a rewrite after two independent referee rounds: v1.0 claimed the opposite and was wrong on an undocumented planting and an arithmetic error, and v2.0 still mixed the price of the trait with selection on it. Every printed number is re-derived from the deposited artifacts by a machine check (30 checks, 0 failures).