The wild bootstrap test of a 26.83–27.03-day signal in 2016–2026 Swift/BAT data from LS I +61°303 depends on block length

We tested the reported second period in LS I +61°303. We used 3,195 good daily Swift/BAT measurements at 15–50 keV from 2016-01-01 to 2026-10-19. The model includes yearly means and a 26.496-day orbital signal with two harmonics. We tested 201 periods from 26.83 to 27.03 days. A restricted wild bootstrap with 120-day time blocks and 4,999 draws gives p = 0.0616 for the maximum score in this band. The fixed 26.93-day test gives p = 0.0370 and a fitted semiamplitude of 0.620 mCrab. The band p-value is 0.1044 for 60-day blocks and 0.0134 for 240-day blocks. Thus rejection at the 5% level depends on block length. These data do not establish a stable second clock or its absence. This test cannot assess the 50–150 keV part of the earlier survey result. We also find a date mismatch in the public pointing table: 5,563 of 78,016 listed MJD values are one day later than the calendar day and mission time. All 6,846 daily dates agree with their calendar fields. Rebinning the pointings by calendar day closely matches the daily rates. We preserve the data snapshots, all affected row indices, code, simulation scores, figures and offline reproduction instructions.

LS I +61 303 · Swift BAT · X-ray binary · periodicity · wild bootstrap · time series · data quality