Hold feasibility (prelock_checks_plan.md, part 2): old starts 13 to 18, ten moments each, +-25%
made 2026-10-13 10:59 Halifax; not a locked plan; no new start; no gap and no growth measured

Reading: given below, with the revised exactness check (adopted after the preregistered one failed in the calibration; prelock_amendment.md)
Base runs, as preregistered (each reproduces its saved run exactly): met (6 of 6)
   first difference from the saved run, by start (gap; coherence): 13: none, none; 14: none, none; 15: none, none; 16: none, none; 17: none, none; 18: none, none
Base runs, revised (not the preregistered check): each agrees to the last digit with the cause test's unchanged base-run code run in one go on the same machine (coherence at full precision and every kept state), in one numerical environment throughout: met (6 of 6 agree; one environment: True)
   the environment: INTEL(R) XEON(R) PLATINUM 8581C CPU @ 2.30GHz, Python 3.11.14, glibc 2.40, numpy 2.4.3 (X86_V3, X86_V4, AVX512_ICL, AVX512_SPR); one boot id in every full record: True
   bitwise agreement on one machine shows reproducibility, not numerical accuracy (prelock_amendment.md)
Seen before this check (prelock_amendment.md; dryruns/feasibility_dryrun.txt): the dry run's holds on old start 1 at T = 600, outside these starts and moments: 500-unit fractions up 0.976, down 0.962, sham +0.218, one up block at 0.15; 100-unit fractions 0.927 and 0.907, sham +0.031.

500-unit hold (101 kicks): NOT FEASIBLE
  F1 kicks on target: 50 of 60 moments carried out, fewest in a start 7 (needs at most 3 left out, at least 5 per start): NOT met
     left out: start 13, moment 0: kick 51 at t = 1960 outside its band: sham A at 1.0882
     left out: start 13, moment 6: kick 86 at t = 4235: down B infeasible
     left out: start 15, moment 9: kick 91 at t = 5250: down A infeasible
     left out: start 16, moment 5: kick 32 at t = 3675: up B infeasible, down B infeasible
     left out: start 16, moment 7: kick 42 at t = 4425: down B infeasible
     left out: start 17, moment 0: kick 88 at t = 1965: up A infeasible, down A infeasible
  F2 history achieved: both fractions in 0.75-1.25 in 82% of moments (needs 90%); mean up 0.965 (SD 0.066), down 0.893 (SD 0.126) (each needs 0.85-1.15): NOT met
  F3 held throughout: every 25-unit block in 0.5-1.5 in 8% of moments (needs 90%); blocks ranged up -0.74 to 2.77, down -0.67 to 1.71: NOT met
  F4 sham (reported apart): effort against up and down 0.762 to 0.847 (needs 0.8-1.2): NOT met; mean history fraction +0.680 (needs within +-0.25): NOT met
  Diagnostics:
    history change dM = 0.00632 (SD 0.00177) = 1.15 natural SDs of the 500-unit mean (planned 0.00648, 1.18 SDs)
    time-integrated history H x (M - M_none): up +3.249, down -3.072, sham +2.0566
    just before each kick, ratio to untouched minus target: up +0.1307 (spread 0.8410), down +0.1590 (spread 0.6821), sham +0.3895 (spread 1.1570)
    ratio to untouched over the hold, 5th to 95th percentile: up 0.693 to 2.141, down 0.387 to 1.519, sham 0.543 to 3.088
    effort, sum of squared moves over the hold: up 216.29, down 109.85, sham 131.49; against one pulse: up 470.54x, down 238.71x
    first kick's share: up 0.004, down 0.008; median maintenance kick: up 1.1406, down 0.6054; largest single move: up 0.118, down 0.095, sham 0.075
    sum of |a| (both groups): up 3.547, down 2.547; net signed push: up +0.185, down -0.457
    coherence left after release, per unit of the planned change (down's sign turned): early up +0.968, down +0.405, sham +1.953; later up +0.351, down -0.387, sham +1.133

100-unit hold (21 kicks): NOT FEASIBLE
  F1 kicks on target: 52 of 60 moments carried out, fewest in a start 6 (needs at most 3 left out, at least 5 per start): NOT met
     left out: start 13, moment 0: kick 8 at t = 2145: down B infeasible
     left out: start 14, moment 8: kick 9 at t = 5025: down B infeasible
     left out: start 15, moment 9: kick 11 at t = 5250: down A infeasible
     left out: start 16, moment 0: kick 3 at t = 2180: down A infeasible
     left out: start 17, moment 0: kick 8 at t = 1965: down A infeasible
     left out: start 17, moment 5: kick 14 at t = 3745: down B infeasible
  F2 history achieved: both fractions in 0.75-1.25 in 81% of moments (needs 90%); mean up 0.954 (SD 0.107), down 0.874 (SD 0.151) (each needs 0.85-1.15): NOT met
  F3 held throughout: every 25-unit block in 0.5-1.5 in 81% of moments (needs 90%); blocks ranged up 0.29 to 1.48, down 0.22 to 1.29: NOT met
  F4 sham (reported apart): effort against up and down 0.763 to 0.966 (needs 0.8-1.2): NOT met; mean history fraction -0.017 (needs within +-0.25): met
  Diagnostics:
    history change dM = 0.00626 (SD 0.00214) = 0.89 natural SDs of the 100-unit mean (planned 0.00655, 0.93 SDs)
    time-integrated history H x (M - M_none): up +0.646, down -0.605, sham -0.0103
    just before each kick, ratio to untouched minus target: up +0.0389 (spread 0.3381), down +0.0744 (spread 0.2974), sham +0.0022 (spread 0.1127)
    ratio to untouched over the hold, 5th to 95th percentile: up 0.959 to 1.679, down 0.506 to 1.144, sham 0.858 to 1.144
    effort, sum of squared moves over the hold: up 11.27, down 7.85, sham 8.34; against one pulse: up 21.79x, down 16.72x
    first kick's share: up 0.088, down 0.119; median maintenance kick: up 0.2876, down 0.2326; largest single move: up 0.067, down 0.044, sham 0.048
    sum of |a| (both groups): up 0.371, down 0.320; net signed push: up +0.059, down -0.119
    coherence left after release, per unit of the planned change (down's sign turned): early up +0.788, down +0.624, sham +0.010; later up +0.434, down +0.196, sham +0.125

Pulse at +-25%: 60 of 60 carried out
  sum of squared moves: 0.92 (each arm the same)
  coherence left after release, per unit of the planned change: early up +0.534, down +0.491, sham +0.004; later up +0.676, down +0.390, sham +0.203

Pieces: 14; parts matching the run log: 126 of 126
The cause test's locked files unchanged: True; its folder and the exploratory folder unchanged: True