Deuterium's map shows texture: the measured ensemble disperses beyond formal uncertainties with flagged structure in redshift and direction (Cooke 2014 and the wider literature), where a single global BBN event mandates one value everywhere. Systematics would be benign, varied astration demanding, and genuine primordial variation fatal.
Uniformity is structural twice over: one BBN event fixes one ratio, and smooth hierarchical assembly arranges similar processing histories along similar sightlines. The model thus predicts a flat map with thin Gaussian noise, and owns no mechanism for the texture if the texture is real.
SCT divides the map cleanly. The primordial layer is flat by registered necessity: cascade termination before one second (P40) and equilibrium BBN (P42) fix one D/H everywhere, with the Plasma Equivalence chain staking SCT against any primordial variation as firmly as ΛCDM is staked (recid 178). The texture lives entirely in the processing layer, and there SCT predicts more of it: absorbers sample a universe whose reservoirs were heterogeneous from deposition (P25), whose star formation ignited early and unevenly on the web's geography (P34), and whose first stellar generations completed fast where deposits ran dense (recid 125). Astration, deuterium's one-way destruction in stars, therefore varies sightline to sightline with the deposited environment, painting exactly the kind of redshift- and direction-correlated texture a smoothly assembled universe cannot arrange.
The registered structure of the prediction is its discriminant: dispersion should correlate with each absorber's metallicity and web position, the processed systems scattering low while the pristine floor converges on the single primordial number; and any directional structure should trace the deposited large-scale geography, not arbitrary axes. The deposit mechanics are in Paper 4, From Chaos To Collisothermal Cosmogenesis, with the cascade-termination chemistry in Paper 1, From Chaos to Convergent Foundations.
Keystone economy: P40 flattens the primordial layer, P25 textures the processing layer. The map's structure is history, not nucleosynthesis.
The kill is layered like the prediction: dispersion surviving at zero metallicity, in absorbers where astration cannot have operated, would demonstrate primordial variation and refute the Plasma Equivalence chain for both frameworks at once. Short of that, an ensemble whose scatter shows no correlation with metallicity or environment as the thirty-meter-class census grows would remove the processing-layer reading and return the texture to unexplained systematics.