AFTER THE DATA — not part of the sealed rule, not a verdict

1. the quietest weekday, with intervals on each day's share of Mon..Fri
   Mon  0.1853  [0.1820, 0.1886]
   Tue  0.2014  [0.1980, 0.2048]
   Wed  0.2031  [0.1997, 0.2066]
   Thu  0.2203  [0.2168, 0.2239]
   Fri  0.1899  [0.1866, 0.1933]
   smallest: Mon, then Fri
   Monday's share of (Mon + Fri): 0.4938 [0.4868, 0.5008]  — 0.5 would be a tie

2. the sealed bars applied to each clock (survived hi < 0.18, killed lo > 0.19)
   UTC-8.0  0.1743  [0.1710, 0.1775]  -> survived
   UTC-5.0  0.1778  [0.1746, 0.1811]  -> inconclusive
   UTC+0.0  0.1899  [0.1866, 0.1933]  -> inconclusive
   UTC+2.0  0.1991  [0.1957, 0.2026]  -> killed
   UTC+5.5  0.2243  [0.2207, 0.2280]  -> killed
   UTC+8.0  0.2359  [0.2321, 0.2397]  -> killed

3. claim B's negative control at 168 h (a week holds five working
   days no matter which day it starts on)
    24 h   control(Tue) RR 1.0500 [1.0129, 1.0886] DIRTY   |   Friday RR 0.6837 [0.6539, 0.7150]
    72 h   control(Tue) RR 1.1143 [1.0890, 1.1402] DIRTY   |   Friday RR 0.7019 [0.6808, 0.7237]
   168 h   control(Tue) RR 1.0269 [1.0076, 1.0464] clean   |   Friday RR 1.0651 [1.0462, 1.0844]

4. the 72 h follow-up rate against the number of Mon..Fri days the
   window covers (a release at day d covers days d+1, d+2, d+3)
   Mon  window holds 3 working days   rate 0.4686
   Tue  window holds 3 working days   rate 0.4923
   Wed  window holds 2 working days   rate 0.4736
   Thu  window holds 1 working days   rate 0.3900
   Fri  window holds 1 working days   rate 0.3189
   Sat  window holds 2 working days   rate 0.3596
   Sun  window holds 3 working days   rate 0.3822
   collapsed:
     1 working days in the window   0.3571   n = 21,456
     2 working days in the window   0.4422   n = 14,682
     3 working days in the window   0.4661   n = 23,817