Where Antimatter Hides in the Universe

In the parity–entanglement programme spacetime begins with pure energy. Under extreme conditions, by an energetic selection subject to constraints, a single conserved check—the membrane—forms first, before any matter. The membrane is fermion parity, (−1)^F. Every charge that forms afterwards forms across it: matter on the spacetime side and its conjugate on the vacuum side, so that the global information balance ΔI = 0 holds. Because annihilation requires both members on the same side, the prior membrane is what prevents a general annihilation. The whole process is unitary: the global state stays pure, information is never lost, and ΔI = 0 holds at every step, with the antimatter in the vacuum as the side of the balance that keeps it. Antimatter is not missing: it is in the vacuum, and it must be there because ΔI = 0: from the initial state of pure energy through every change since, the total information does not change. The rules of the programme that express this—the information balance, the sum rule across the cut, global purity, the record and the image of the world held in the cross-cut correlations—all require the conjugate of every charge on our side to be held on the vacuum side. No single rule fixes the location; only their conjunction does, and each member of it is tested. This follows from the programme; the experiments reported here support each ingredient on which it rests but do not by themselves locate the antimatter, and the decisive tests of the programme's premises are the full-survey DESI dark-energy measurement and a COLTRIMS repetition of the zeptosecond birth-time experiment. We test, on superconducting quantum processors (ibm_kingston, ibm_marrakesh; nine runs), each structural ingredient, and report every measurement against its exact prediction. Exchange across the membrane moves charge without annihilating it: the bias follows P₀ cos 2θ with no fitted scale while the membrane stays at 0.985–1.000. Annihilation occurs only when both members are on the same side: a local action alone returns the vacuum with at most the vacuum weight α² of the pair (0.798 against 0.824; 0.511 against 0.500), a joint one after the partner is brought across with 0.96–0.98. The partner is the conjugate, not a copy: twenty random SU(3) tests, ten on each processor, distinguish U ⊗ U* from U ⊗ U with mean deviations of 0.033 and 0.048 from prediction. Pair creation is an entanglement event, I(A:B) = 1.204 and 1.999 bits against 1.202 and 2.000. The refund through the pair falls as √(1 − P²) on both processors, so the pair is not its own mirror. A membrane formed before the writes loses nothing; one formed after removes exactly the odd weight. Deviations in early runs were traced to time-driven parity leakage and a coherent reference phase, and corrected. The value of the baryon-to-photon ratio is not obtained here. Statements are labelled [P] proved, [K] known, [N] numerical or measured, [F] formal reading, [O] open.