Cause test, exploratory follow-up: October 12, 2026. N = 4,000 per group, K = 1.90, eps = 0.1 EXPLORATORY: chosen after the cause test's decision was known and run on the same saved runs, so not a test. The decision stands: FAIL as registered (refuted under the tested conditions). Nothing here changes it. Read-only: the model was not run. Inputs fingerprinted before and after: unchanged. Data: the cause test's 42 new starts (19 to 60) and 420 moments; for part A also the 18 older starts (1 to 18) as a cross-check. Cells: mean ± error bar (mean in error bars); error bars from the starts, as in the cause test. Checks: this code gives the benchmark on the old 18 starts (-0.1602) and the report's S (+0.00031), Phi (0.511) and share left (0.297). A. THE UNTOUCHED RUNS (every T = 2,000, 2,005, ..., 5,500; W = growth rate over (T + 200, T + 500]) new 42 starts old 18 starts A1. The benchmark slope (growth per unit of C(T)); the benchmark is -0.1602, from the old 18 W = a + beta C(T) + c1 g1 + ... + c5 g5: beta -0.1534 ± 0.0144 (-10.7) -0.1602 ± 0.0220 (-7.3) W = a + beta C(T): beta -0.2051 ± 0.0110 (-18.7) -0.2132 ± 0.0172 (-12.4) partial correlation of C(T) with W given g1 to g5 -0.2552 ± 0.0238 -0.2654 ± 0.0337 (Test 17's own measure, -0.1589 on the new starts in the report, is a partial correlation averaged over 5-unit lags, not a slope; it is on a different scale from the benchmark and the two should not be compared.) A2. Natural persistence of a coherence difference at T (slope of the later mean C on C(T)) over (T, T + 200] +0.345 ± 0.016 (+21.8) +0.363 ± 0.015 (+24.9) over (T + 200, T + 500] +0.089 ± 0.012 (+7.6) +0.106 ± 0.019 (+5.6) the interventions, from the report: 0.511 over (T, T + 200] and 0.297 over (T + 200, T + 500] A3. Candidate carriers known at T: partial correlations with W (new 42 | old 18) feature F F with W, given g F with W, given g and C(T) C(T) with W, given g and F (nothing added: C(T) with W given g) -0.2552 ± 0.0238 | -0.2654 ± 0.0337 mean C over (T - 100, T] -0.4794 ± 0.0363 | -0.492 -0.4254 ± 0.0327 | -0.440 -0.0191 ± 0.0085 | -0.006 mean C over (T - 500, T] -0.5641 ± 0.0310 | -0.571 -0.5297 ± 0.0294 | -0.540 -0.0838 ± 0.0181 | -0.073 trend of C over (T - 100, T] +0.0224 ± 0.0122 | +0.045 +0.1044 ± 0.0124 | +0.126 -0.2730 ± 0.0238 | -0.285 alignment cos(psi_A - psi_B) at T -0.1596 ± 0.0150 | -0.173 -0.0682 ± 0.0092 | -0.084 -0.2146 ± 0.0210 | -0.222 cross term Re(z_A conj z_B) at T -0.2493 ± 0.0236 | -0.255 -0.0537 ± 0.0106 | -0.051 -0.0818 ± 0.0111 | -0.093 imbalance |R_A - R_B| at T -0.0331 ± 0.0068 | -0.041 +0.0018 ± 0.0063 | -0.000 -0.2529 ± 0.0238 | -0.263 turning rate |dpsi/dt|, (T - 25, T] +0.2603 ± 0.0233 | +0.301 +0.1967 ± 0.0205 | +0.242 -0.1944 ± 0.0203 | -0.193 every feature together +0.0011 ± 0.0085 | -0.013 A4. What comes after T: C(T)'s partial correlation with W, given g and early growth E over (T, T + 200] -0.2597 ± 0.0208 -0.2653 ± 0.0303 mean C over (T, T + 200] and (T + 200, T + 500] -0.0516 ± 0.0122 -0.0483 ± 0.0180 all three -0.0249 ± 0.0124 -0.0259 ± 0.0155 W = a + b_C C(T) + b_E E + g terms: b_C -0.1509 ± 0.0123 (-12.3) -0.1523 ± 0.0193 (-7.9) W = a + b_C C(T) + b_E E + g terms: b_E +0.2379 ± 0.0278 (+8.5) +0.2659 ± 0.0427 (+6.2) partial correlation of E with W, given g and C(T) +0.2478 ± 0.0267 +0.2771 ± 0.0443 E = a + a_C C(T) + g terms: a_C (the early link) +0.0124 ± 0.0157 (+0.8) -0.0031 ± 0.0204 (-0.2) W against the mean C over the same window, given g: slope +0.5518 ± 0.0571 (+9.7) +0.5366 ± 0.0675 (+7.9) the same as a partial correlation +0.4771 ± 0.0440 +0.4885 ± 0.0574 B. THE 420 MOMENTS, UNTOUCHED COPY AT T (W of the untouched continuation; each start's means taken out; scaled weights: W's spread per spread of the variable; error bars from redrawing the starts) C(T) + g C(T) -0.238 ± 0.051 C(T) + g + |z2|, |z3|, |z4| C(T) -0.230 ± 0.052 |z2| +0.031 ± 0.050 |z3| +0.044 ± 0.044 |z4| +0.008 ± 0.044 g + |z2|, |z3|, |z4| (no C) |z2| +0.064 ± 0.050 |z3| +0.055 ± 0.045 |z4| +0.008 ± 0.046 C(T) + g + alignment C(T) -0.208 ± 0.056 alignment -0.064 ± 0.034 C1. GROWTH AGAINST BOTH CONTROLS (per moment, then each start's average; S = (down - up) / 2, P = (up + down) / 2 - sham) immediate (T, T + 200] dense sparse pooled sparse - dense up - sham +0.00717 ± 0.00031 (+22.9) +0.00475 ± 0.00032 (+14.7) +0.00596 ± 0.00025 (+23.7) -0.00242 ± 0.00039 (-6.2) down - sham -0.00353 ± 0.00032 (-10.9) -0.00419 ± 0.00034 (-12.2) -0.00386 ± 0.00029 (-13.5) -0.00066 ± 0.00034 (-1.9) sham - none +0.00063 ± 0.00017 (+3.6) +0.00380 ± 0.00028 (+13.4) +0.00221 ± 0.00019 (+11.6) +0.00317 ± 0.00027 (+11.7) up - none +0.00780 ± 0.00033 (+23.5) +0.00855 ± 0.00033 (+26.0) +0.00818 ± 0.00030 (+27.4) +0.00075 ± 0.00028 (+2.6) down - none -0.00290 ± 0.00032 (-9.0) -0.00039 ± 0.00030 (-1.3) -0.00165 ± 0.00029 (-5.6) +0.00251 ± 0.00019 (+13.0) S -0.00535 ± 0.00019 (-28.6) -0.00447 ± 0.00019 (-23.6) -0.00491 ± 0.00016 (-29.9) +0.00088 ± 0.00018 (+4.8) P +0.00182 ± 0.00026 (+7.1) +0.00028 ± 0.00028 (+1.0) +0.00105 ± 0.00021 (+4.9) -0.00154 ± 0.00032 (-4.8) delayed (T + 200, T + 500] dense sparse pooled sparse - dense up - sham -0.00118 ± 0.00034 (-3.5) -0.00198 ± 0.00038 (-5.2) -0.00158 ± 0.00027 (-6.0) -0.00079 ± 0.00048 (-1.6) down - sham -0.00169 ± 0.00035 (-4.8) -0.00023 ± 0.00040 (-0.6) -0.00096 ± 0.00024 (-4.0) +0.00146 ± 0.00056 (+2.6) sham - none +0.00095 ± 0.00032 (+3.0) +0.00226 ± 0.00037 (+6.1) +0.00160 ± 0.00029 (+5.6) +0.00131 ± 0.00038 (+3.5) up - none -0.00023 ± 0.00041 (-0.6) +0.00028 ± 0.00042 (+0.7) +0.00002 ± 0.00038 (+0.1) +0.00052 ± 0.00032 (+1.6) down - none -0.00074 ± 0.00037 (-2.0) +0.00203 ± 0.00033 (+6.2) +0.00064 ± 0.00030 (+2.1) +0.00277 ± 0.00034 (+8.2) S -0.00025 ± 0.00018 (-1.4) +0.00087 ± 0.00020 (+4.3) +0.00031 ± 0.00015 (+2.1) +0.00113 ± 0.00024 (+4.6) P -0.00144 ± 0.00029 (-4.9) -0.00110 ± 0.00033 (-3.3) -0.00127 ± 0.00020 (-6.3) +0.00034 ± 0.00046 (+0.7) in 100-unit steps after T (pooled over the methods; dense | sparse) 0-100 100-200 200-300 300-400 400-500 up - sham +0.00893 (+28.6) +0.0096|+0.0083 +0.00300 (+8.3) +0.0048|+0.0012 -0.00042 (-1.1) +0.0005|-0.0013 -0.00145 (-3.1) -0.0008|-0.0021 -0.00287 (-7.5) -0.0032|-0.0025 down - sham -0.00409 (-11.9) -0.0039|-0.0043 -0.00363 (-9.1) -0.0032|-0.0041 -0.00269 (-5.9) -0.0034|-0.0020 -0.00059 (-1.6) -0.0013|+0.0001 +0.00040 (+1.3) -0.0004|+0.0012 sham - none +0.00127 (+8.5) +0.0003|+0.0022 +0.00315 (+10.4) +0.0009|+0.0054 +0.00251 (+6.6) +0.0012|+0.0039 +0.00158 (+3.5) +0.0009|+0.0023 +0.00072 (+1.5) +0.0008|+0.0006 S -0.00651 (-29.3) -0.0067|-0.0063 -0.00331 (-16.2) -0.0040|-0.0026 -0.00114 (-5.4) -0.0019|-0.0003 +0.00043 (+2.2) -0.0003|+0.0011 +0.00164 (+8.7) +0.0014|+0.0018 P +0.00242 (+10.0) +0.0028|+0.0020 -0.00032 (-1.0) +0.0008|-0.0014 -0.00156 (-4.2) -0.0015|-0.0016 -0.00102 (-2.7) -0.0010|-0.0010 -0.00123 (-4.2) -0.0018|-0.0007 C2. THE STATE AFTER THE CHANGE (pooled; dense | sparse means) coherence C up - sham at T: +0.01303 ± 0.00028 (+46.6) +0.01303|+0.01303 immediate: +0.00715 ± 0.00029 (+24.4) +0.00770|+0.00660 delayed: +0.00339 ± 0.00033 (+10.3) +0.00397|+0.00281 down - sham at T: -0.01303 ± 0.00028 (-46.6) -0.01303|-0.01303 immediate: -0.00617 ± 0.00034 (-18.0) -0.00571|-0.00662 delayed: -0.00434 ± 0.00043 (-10.0) -0.00432|-0.00437 sham - none at T: -0.00002 ± 0.00000 (-8.5) -0.00000|-0.00004 immediate: +0.00153 ± 0.00014 (+10.9) +0.00017|+0.00290 delayed: +0.00253 ± 0.00029 (+8.7) +0.00095|+0.00411 (up - down) / 2 at T: +0.01303 ± 0.00028 (+46.6) +0.01303|+0.01303 immediate: +0.00666 ± 0.00022 (+30.8) +0.00671|+0.00661 delayed: +0.00387 ± 0.00018 (+21.8) +0.00415|+0.00359 alignment cos(psi_A - psi_B) up - sham at T: +0.00045 ± 0.00045 (+1.0) +0.00010|+0.00081 immediate: +0.13055 ± 0.00717 (+18.2) +0.13894|+0.12217 delayed: +0.09987 ± 0.00876 (+11.4) +0.11400|+0.08574 down - sham at T: -0.00212 ± 0.00071 (-3.0) -0.00069|-0.00356 immediate: -0.18966 ± 0.00977 (-19.4) -0.19740|-0.18191 delayed: -0.14469 ± 0.01195 (-12.1) -0.15961|-0.12977 sham - none at T: +0.00003 ± 0.00033 (+0.1) +0.00007|-0.00001 immediate: +0.01265 ± 0.00348 (+3.6) -0.00202|+0.02732 delayed: +0.05131 ± 0.00806 (+6.4) +0.02199|+0.08063 (up - down) / 2 at T: +0.00129 ± 0.00048 (+2.7) +0.00040|+0.00218 immediate: +0.16011 ± 0.00506 (+31.7) +0.16817|+0.15204 delayed: +0.12228 ± 0.00513 (+23.8) +0.13681|+0.10775 cross term Re(z_A conj z_B) up - sham at T: +0.00077 ± 0.00004 (+19.5) +0.00077|+0.00077 immediate: +0.00050 ± 0.00002 (+22.2) +0.00053|+0.00046 delayed: +0.00020 ± 0.00002 (+9.6) +0.00023|+0.00017 down - sham at T: -0.00046 ± 0.00002 (-19.5) -0.00046|-0.00046 immediate: -0.00036 ± 0.00002 (-15.9) -0.00031|-0.00041 delayed: -0.00024 ± 0.00003 (-8.9) -0.00023|-0.00025 sham - none at T: -0.00000 ± 0.00000 (-5.0) -0.00000|-0.00000 immediate: +0.00010 ± 0.00001 (+10.2) +0.00001|+0.00020 delayed: +0.00015 ± 0.00002 (+7.9) +0.00005|+0.00025 (up - down) / 2 at T: +0.00061 ± 0.00003 (+19.5) +0.00062|+0.00061 immediate: +0.00043 ± 0.00002 (+27.0) +0.00042|+0.00044 delayed: +0.00022 ± 0.00001 (+20.3) +0.00023|+0.00021 Slopes in one unit, growth per unit of coherence: natural (untouched runs, given g) against induced (arms against each other: the growth difference over the coherence difference, start by start) C(T) -> delayed growth natural -0.1534 ± 0.0144 (-10.7) induced: dense +0.0200 ± 0.0137 (+1.5), sparse -0.0661 ± 0.0149 (-4.4), pooled -0.0231 ± 0.0111 (-2.1) C(T) -> early growth natural +0.0124 ± 0.0157 (+0.8) induced: dense +0.4119 ± 0.0121 (+34.0), sparse +0.3459 ± 0.0145 (+23.9), pooled +0.3789 ± 0.0111 (+34.1) C in the delayed window -> growth in it natural +0.5518 ± 0.0571 (+9.7) induced: dense +0.0187 ± 0.0473 (+0.4), sparse -0.5018 ± 0.1439 (-3.5), pooled -0.2416 ± 0.0786 (-3.1) C3. MOMENT BY MOMENT WITHIN EACH METHOD (each start's means taken out; weights ± error bars from redrawing) a. S against the up/down difference in |z2|, |z3|, |z4| right after the change, h_n = (|z_n|_down - |z_n|_up) / 2, with C(T) h2 level: dense -0.00012 ± 0.00003 (-4.3), sparse -0.00041 ± 0.00006 (-7.5) h3 level: dense +0.00001 ± 0.00002 (+0.3), sparse -0.00017 ± 0.00009 (-1.9) h4 level: dense +0.00000 ± 0.00003 (+0.1), sparse -0.00002 ± 0.00007 (-0.3) dense: h2 -0.447 ± 0.311 (scaled -0.067 ± 0.047); h3 -0.323 ± 0.436 (scaled -0.035 ± 0.048); h4 -0.131 ± 0.419 (scaled -0.019 ± 0.061); C(T) +0.051 ± 0.015 (scaled +0.138 ± 0.042) sparse: h2 -0.062 ± 0.180 (scaled -0.017 ± 0.050); h3 -0.075 ± 0.096 (scaled -0.030 ± 0.039); h4 +0.222 ± 0.129 (scaled +0.078 ± 0.045); C(T) +0.111 ± 0.018 (scaled +0.311 ± 0.050) both methods: sparse rows +0.00113 ± 0.00024 without h; +0.00104 ± 0.00025 with h2, h3, h4 b. sham - none (delayed) against the sham's own change in |z_n|, q_n = |z_n|_sham - |z_n| before, with C(T) q2 level: dense +0.00005 ± 0.00002 (+2.3), sparse +0.00145 ± 0.00013 (+11.3) q3 level: dense -0.00003 ± 0.00001 (-1.9), sparse +0.00000 ± 0.00006 (+0.0) q4 level: dense +0.00003 ± 0.00003 (+1.0), sparse +0.00017 ± 0.00009 (+1.9) dense: q2 -0.168 ± 0.749 (scaled -0.011 ± 0.050); q3 +0.629 ± 1.085 (scaled +0.034 ± 0.059); q4 +0.495 ± 0.480 (scaled +0.050 ± 0.048); C(T) +0.015 ± 0.032 (scaled +0.026 ± 0.057) sparse: q2 -0.357 ± 0.130 (scaled -0.153 ± 0.056); q3 -0.070 ± 0.280 (scaled -0.011 ± 0.044); q4 -0.185 ± 0.223 (scaled -0.047 ± 0.057); C(T) +0.076 ± 0.036 (scaled +0.113 ± 0.054) c. S against the coherence difference left in the delayed window, (C_up - C_down) / 2, with C(T) (if coherence in the window slowed growth, the weight would be positive) dense: C left (delayed) -0.5259 ± 0.0446 (scaled -0.561 ± 0.048); C(T) +0.1329 ± 0.0134 (scaled +0.360 ± 0.036) sparse: C left (delayed) -0.3255 ± 0.0362 (scaled -0.395 ± 0.044); C(T) +0.1501 ± 0.0149 (scaled +0.419 ± 0.042) both methods: C left -0.4167 ± 0.0339 d. S against the achieved change (if coherence were the cause at the benchmark's size: +0.1602) dense: +0.1118 ± 0.0292 sparse: +0.2290 ± 0.0326 both methods: +0.1704 ± 0.0263 C4. THE EARLY GROWTH AS A CARRIER (delayed effect against the sham: seen, and what the untouched runs' regression W = a + b_C C(T) + b_E E + g terms would give from each arm's own early effect) early effect delayed, seen b_E x early + b_C x change in C seen - (b_E early + b_C change) moment by moment: delayed on early dense up +0.00717 ± 0.00031 (+22.9) -0.00118 ± 0.00034 (-3.5) +0.00186 ± 0.00025 (+7.5) -0.00011 ± 0.00028 (-0.4) -0.00107 ± 0.00033 (-3.2) +0.032 ± 0.052 dense down -0.00353 ± 0.00032 (-10.9) -0.00169 ± 0.00035 (-4.8) -0.00080 ± 0.00013 (-6.0) +0.00117 ± 0.00020 (+5.8) -0.00287 ± 0.00037 (-7.7) +0.216 ± 0.078 sparse up +0.00475 ± 0.00032 (+14.7) -0.00198 ± 0.00038 (-5.2) +0.00116 ± 0.00019 (+6.2) -0.00081 ± 0.00027 (-3.0) -0.00116 ± 0.00045 (-2.6) -0.068 ± 0.054 sparse down -0.00419 ± 0.00034 (-12.2) -0.00023 ± 0.00040 (-0.6) -0.00099 ± 0.00014 (-7.1) +0.00098 ± 0.00020 (+4.9) -0.00121 ± 0.00042 (-2.8) -0.015 ± 0.055 Instrument limits (not saved, so not analysed): the higher harmonics over time and in the base runs (only |z2|, |z3|, |z4| magnitudes at T and right after the change, not their phases relative to z1); each oscillator's phase and natural rate (so no frequency-resolved or cluster measures); the gap's direction. Changes after the first run: none