The angular generation split of the projected Dirac residue E_ is carried, on the directed Heisenberg frontier, by a chirally dressed cocycle signal _ = L\, (2/q) \, Ac^+. The raw cocycle average and the maximal-locking bound _{} are computable from the arithmetic backend, but the signal itself depends on the chiral weight L, whose origin lies in the Lorentzian spin-lift and not in the arithmetic frontier. We prove that L cannot be any function of the real cascade generator T (e): edge inversion and the residual reflection both act as T -T, so every f (T) has identical arrow-parity and -parity, whereas L must be arrow-even and -odd. The resolution is that L is the Weyl representation type, 2 versus {2}: group inversion preserves the representation (arrow-even) while complex conjugation exchanges it (-odd). In the inherited Heisenberg central-phase lift the chiral phase is the linear cyclotomic character _ (e) =q^{ Ac (e) }, with no Gauss, Maslov, or 5 component; on the nonzero-cocycle support the natural branch satisfies L (e) =sign \, q^{ Ac (e) } = sign Ac (e), giving _ = 1 and NA = _{} 0. This establishes H-orient in the inherited Heisenberg central-phase lift, not unconditionally: the only remaining obstruction is an additional spin-Galois phase outside the central character, in particular a 5 or ADE-spin factor orthogonal to q. The result is a conditional derivation of the frontier amplitude datum; it does not derive u 0, which still requires the Schur transport of the polarisation onto the J_-odd generation sector.
Jérôme Beau (Wed,) studied this question.