Scattering-side pilot (step 3, exploratory), 2026-10-15 12:07 Halifax. Fresh starts 135 to 146 (seeds 134 to 145), 10 moments each by the medium pilot's rule: 120 moments, every start counting equally. At each moment four arms run from the base run's state at T, both copies, 200 units changed and 500 restored: the untouched rerun, the raised pull (K x 1.005), arm F (c_f = 0.6) and arm F' (c_f = 0.4). Comparison A and arm U are not run (Amendment 1). A toy model; histories, not memory. An exploratory pilot can suggest, not confirm.
Bars: starts are the units; the larger of the resampled bar (400 resamples, seed 20261016) and SD/sqrt(12); a share of averages (B1, the share left after R) takes the resampled bar. 'Inside': the reading ± 2 bars within ± the margin. 'Clear': more than 3 bars from zero.

1. Comparison B (c_f = 0.6), against untouched
   B1, the gain at R as a share of the raised pull's (margin ±0.25): +0.081 ± 0.041: as expected (no meaningful change)
   B2, the faster half's own coherence at R (margin ±0.005): +0.0062 ± 0.0013: contradicts (a meaningful change)
   E-any, the share of group-moments holding a rhythm (inside ±0.05): +0.000 ± 0.000: as expected (no edge synchronization)
   E-steady (used only in the combined reading): +0.000 ± 0.000: inside ±0.05
   E-pair (described): +0.000 ± 0.000
   Combined reading: 2. A change in the faster half's coherence without a separate rhythm.

2. Comparison B' (c_f = 0.4), against untouched
   B2 (as expected: more than 3 bars above zero and at least 0.005; contradicts: inside ±0.005): +0.0542 ± 0.0018: as expected (up)
   E-pair (as expected: more than 3 bars above zero and at least 0.10; contradicts: inside ±0.05): +0.696 ± 0.037: as expected
   E-any (the same labels): +0.967 ± 0.011: as expected
   E-steady (used only in the combined reading): +1.000 ± 0.000: not inside ±0.05
   The median |Omega*| over the 399 directions holding a rhythm: 0.963 (within 0.90 to 1.10)
   B1 (described): +0.703 ± 0.079
   Combined reading: 1. The predicted counter-rotating edge rhythm.

3. Described, not labelled (arm minus untouched unless stated)
   C at R, untouched: 0.0241 ± 0.0008. The gain at R: raised pull +0.0227 ± 0.0008; arm F (0.6) +0.0018 ± 0.0009; arm F' (0.4) +0.0159 ± 0.0016.
   The extra pull the raised pull delivered over the window (the two groups' average): 0.066.
   Where any gain sits at R: the exact contributions (slower half, faster half, free turners), then each half's own coherence (slower, faster):
     raised pull: +0.0241 ± 0.0008, -0.0013 ± 0.0005, -0.0001 ± 0.0001; +0.0476 ± 0.0016, +0.0000 ± 0.0002
     arm F (0.6): +0.0002 ± 0.0007, +0.0017 ± 0.0006, -0.0001 ± 0.0001; +0.0027 ± 0.0016, +0.0062 ± 0.0013
     arm F' (0.4): -0.0007 ± 0.0021, +0.0168 ± 0.0022, -0.0002 ± 0.0002; +0.0326 ± 0.0033, +0.0542 ± 0.0018
   C by 100-unit block (the window's two, then five after R):
     raised pull: [+0.0047, +0.0149, +0.0184, +0.0132, +0.0094, +0.0065, +0.0042]
     arm F (0.6): [+0.0013, +0.0011, +0.0001, +0.0006, +0.0005, +0.0011, +0.0011]
     arm F' (0.4): [+0.0056, +0.0119, +0.0008, +0.0010, +0.0012, +0.0013, +0.0008]
   The share of the gain at R still there over (R + 200, R + 500] (it means something only if B1 shows a change): raised pull +0.30 ± 0.03; arm F (0.6) +0.51 ± 2.07; arm F' (0.4) +0.07 ± 0.05.
   Later growth over (R + 200, R + 500] (the medium pilot's shams: +0.0015 on average; the benchmark is 0.1602 x |the gain at R|): raised pull -0.0050 ± 0.0004 (benchmark 0.0036); arm F (0.6) -0.0003 ± 0.0004 (benchmark 0.0003); arm F' (0.4) +0.0002 ± 0.0006 (benchmark 0.0026).
   B2's course through the window (the faster half's own coherence, 50-unit blocks):
     raised pull: [+0.0000, +0.0000, +0.0000, +0.0001]
     arm F (0.6): [+0.0034, +0.0041, +0.0045, +0.0047]
     arm F' (0.4): [+0.0134, +0.0317, +0.0458, +0.0561]
   At R, |z2|, |z3|, |z4| and |the A-B timing|: raised pull +0.0002 ± 0.0004, -0.0009 ± 0.0005, +0.0006 ± 0.0006, -0.320 ± 0.072; arm F (0.6) +0.0002 ± 0.0007, -0.0005 ± 0.0006, +0.0006 ± 0.0006, +0.093 ± 0.043; arm F' (0.4) +0.0015 ± 0.0004, -0.0002 ± 0.0007, +0.0000 ± 0.0006, +0.017 ± 0.080.
   The rule's numbers, every arm (shares of group-moments; L, S and |Omega*| over every direction, median and largest):
     untouched: pair 0.000, any 0.000, one way only 0.000, steady 0.000; directions holding a rhythm 0; L 1.10 (1.40), S 0.0088 (0.0272), |Omega*| 0.979 (1.137)
     raised pull: pair 0.000, any 0.000, one way only 0.000, steady 0.000; E one way only +0.000 ± 0.000; directions holding a rhythm 0; L 1.10 (1.45), S 0.0090 (0.0273), |Omega*| 0.978 (1.130)
     arm F (0.6): pair 0.000, any 0.000, one way only 0.000, steady 0.000; E one way only +0.000 ± 0.000; directions holding a rhythm 0; L 1.12 (1.42), S 0.0205 (0.0522), |Omega*| 0.977 (1.081)
     arm F' (0.4): pair 0.696, any 0.967, one way only 0.271, steady 1.000; E one way only +0.271 ± 0.037; directions holding a rhythm 399; L 1.91 (2.76), S 0.1236 (0.1714), |Omega*| 0.963 (1.034)
   Each rhythm's size (medians over the directions holding one): how many of the 968 share its interval, its S, and the slower-half oscillators turning in the same interval:
     arm F' (0.4): 96, 0.1275, 19
   A and B in step (moments and directions where both groups hold a rhythm; their rates within 0.03; their steady parts within a quarter turn; both):
     arm F' (0.4): 167; 167; 144; 144
   Growth through the window and fading after R, for the directions holding a rhythm: the median S at their Omega* in each 50-unit block from T to R + 500, then the median turning rate of w in each block (signed: turning in its own direction is positive):
     arm F' (0.4): S [0.039, 0.081, 0.109, 0.128, 0.033, 0.010, 0.006, 0.005, 0.004, 0.004, 0.003, 0.003, 0.003, 0.003]
     arm F' (0.4): turning [+0.916, +0.964, +0.962, +0.963, +0.803, +0.611, +0.547, +0.549, +0.544, +0.537, +0.553, +0.521, +0.545, +0.515]
   How far each reshaping moved the rates (the faster half's median and largest |natural rate|): untouched 2.18 and 19.72; 0.6 1.69 and 12.21; 0.4 1.44 and 8.46.

4. Notes
   The untouched rerun against the base run: coherence and gap growth identical to the last digit at 120 of 120 moments (largest coherence difference 0.0e+00); its first copy's state at R identical at 120 of 120.
   Every arm on its scheduled pull and rates, and back on the original ones after R: 120 of 120 moments. Every reading present and finite, the sums adding up, the rule running: 120 of 120.
   The old-start results of step 2 aren't in this report: they're feasibility observations, never pooled with these.
   Saved: 167.7 MB in generated-artifacts/scattering-side/pilot/ (12 base runs and 120 moments).
   Compute: 11 pieces, 30.6 minutes of foreground time (7 workers at a time). Machine: INTEL(R) XEON(R) PLATINUM 8581C CPU @ 2.30GHz.
   Every piece ran under this pilot lock, plan, lock, amendment and code: yes.
   The cause test's locked files unchanged: True.
   Checks: pilot_checks.json SHA-256 c386a98f0926836cd9b06602c6802be62e3f034aa421ca16d0ac4bf2ba05220e. Go: pilot_go.json SHA-256 d55c9c0a825ac00b02b357656e408e79aa862d296f39dd42a3f0e646708e51d5.
   Moments: start 135: T = 2285, 2615, 2945, 3275, 3605, 3935, 4265, 4595, 4925, 5255; start 136: T = 2200, 2530, 2860, 3190, 3520, 3850, 4180, 4510, 4840, 5170; start 137: T = 2145, 2475, 2805, 3135, 3465, 3795, 4125, 4455, 4785, 5115; start 138: T = 2315, 2645, 2975, 3305, 3635, 3965, 4295, 4625, 4955, 5285; start 139: T = 2300, 2630, 2960, 3290, 3620, 3950, 4280, 4610, 4940, 5270; start 140: T = 2275, 2605, 2935, 3265, 3595, 3925, 4255, 4585, 4915, 5245; start 141: T = 2080, 2410, 2740, 3070, 3400, 3730, 4060, 4390, 4720, 5050; start 142: T = 2140, 2470, 2800, 3130, 3460, 3790, 4120, 4450, 4780, 5110; start 143: T = 2035, 2365, 2695, 3025, 3355, 3685, 4015, 4345, 4675, 5005; start 144: T = 2085, 2415, 2745, 3075, 3405, 3735, 4065, 4395, 4725, 5055; start 145: T = 2100, 2430, 2760, 3090, 3420, 3750, 4080, 4410, 4740, 5070; start 146: T = 2170, 2500, 2830, 3160, 3490, 3820, 4150, 4480, 4810, 5140.