HLOP Figure 1 (top) — conductor , ranks 0 & 1
f₀ (rank 0)
f₁ (rank 1) — antiphase
HLOP Figure 1 (bottom) — conductor , ranks 0 & 2
f₀ (rank 0)
f₂ (rank 2) — same parity, in phase
Scale collapse in y = p/N, per rank (two controls)
Disjoint windows [2500,5000] vs [5000,10000] are consistent with p/N scaling (ratio = RMS distance / sampling-noise floor F; < 3). Two controls: the reversal null (within-curve scramble) measures the test's power against a wrong scaling — it turned out limited (ratio < 3, not rejected) at these family sizes; the parity null pairs the opposite-parity rank and tests non-degeneracy (differs in rank AND range, so it is NOT a scaling null). Rank 2 is low power. F assumes independent ap (anticonservative — the murmuration IS correlation); ratios are observed empirics, not proof.