catalogue: 7327 events, M>=6.0, 1973-01-01 .. 2025-12-31

time measure  hourly grid  f_exp = 0.492970  (464592 points)
time measure  7-min grid   f_exp = 0.492987  (3982218 points)
  agreement |hour - fine| = 0.000017   (rule: < 0.001)  -> PASS
  departure from 0.5      = 0.007030   (rule: < 0.02)   -> PASS
time measure  full moon alone (S3): f_exp = 0.246190

declustering (Gardner-Knopoff, forward windows)
  events      7327
  mainshocks  5368  (73.3 %)
  clusters    5368
  largest cluster 33 events
  rule: fewer than 1000 mainshocks voids the study -> PASS
  symmetric variant (S4): 4664 mainshocks

N1  times drawn uniformly from the window, N = 5368
  mean E over 200 draws = +0.109 %  (sd 1.446)
  'interval excludes 0' fired 12/200 = 6.0 %  (rule: near 5 %)
  -> PASS

N2  every real mainshock time jittered by U(-365, +365) days
  mean E over 50 draws = -0.191 %  (sd 1.551)
  -> PASS

FLOOR  injected excess recovered, 5368 mainshocks, 200 simulations each
  +  2.0 %  recovered (interval excludes 0)  29.5 % of the time
  +  5.0 %  recovered (interval excludes 0)  93.5 % of the time  <- floor
  + 10.0 %  recovered (interval excludes 0) 100.0 % of the time
  + 20.0 %  recovered (interval excludes 0) 100.0 % of the time
  floor = +5 %   bar = +5 %
  -> the floor is at or below the bar: both branches live

FLOOR for S2 (M>=7.0): 663 mainshocks
  +  5.0 %  recovered  24.5 %
  + 10.0 %  recovered  70.5 %
  + 20.0 %  recovered 100.0 %
  + 40.0 %  recovered 100.0 %
  floor = +20 %  -> S2 is VOID: unfailable