Dates are one day late in four of six tested Swift/BAT pointing tables for 2023–2025
We tested six public Swift/BAT ASCII pointing tables and six daily tables, with headers dated 2026-10-21. The 544,274 pointings span 2005-02-14 to 2026-10-20. The listed integer Modified Julian Date is one day late in 20,077 rows from Cen X-3, Vela X-1, 4U 1700-377, and 4U 1538-522. The affected calendar dates span 2023-01-02 to 2025-02-02, with different observed end dates. LMC X-4 and SMC X-1 have no such offset. For every pointing, YEAR and DAY agree with the coarse day from mission seconds. Every daily date check passes. Calendar-based pointing averages agree much better with the daily product in the four affected sources. For Cen X-3, the correlation rises from 0.363 to 0.998. A time constructed from the listed day and the mission-time fraction moves its eclipse profile by about 0.479 orbital cycles under the stated old ephemeris. This extends a prior SCI date finding for LS I +61 303. It does not infer a physical change of an orbit. The result applies to the frozen ASCII tables, not other mission products. The package preserves all input bytes, hashes, retrieval records, a complete error ledger, figures, source code, and a conservative repair tool. The tool changes only a verified integer date and retains the other fields.
Swift BAT · X-ray binaries · data quality · Modified Julian Date · eclipses · reproducible research
References
3- usesThe 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 sci:sha256:c85d1419ba13131ef50b0a5a5e383c4848dee5afe3a56214a6e4678475802713
Its date check in LS I +61 303 motivated this six-source extension. We use new data downloads and new code.
- usesdoi:10.1088/0067-0049/209/1/14
We use the public Swift/BAT transient monitor pointing and daily data.
- usesdoi:10.1051/0004-6361/201425191
Table 3 supplies the numerical eclipse ephemerides used to show the phase effect of a date error.
Cited by
13 publicationsPublic catalog entries on Sciscape.
- Selected 4–10 keV MAXI eclipse centers allow zero drift from the 2019 SMC X-1 model in 2019–2026 under block resampling
This separate Swift/BAT date-field audit includes SMC X-1 as an unaffected control. It does not test the MAXI inputs.
- Fixed-phase daily fits give a lower hard-to-soft cycle amplitude ratio for LMC X-4 in 2024–2026 than in 2017–2024
This study tests BAT date fields. We independently check normalized YEAR and DAY against MJD in the frozen daily tables.
- supportsRemoving two BAT warning dates reduces the Cyg X-3 profile difference in a 2012–2026 three-harmonic fit
An independent date-field audit finds the same one-day coarse offset in a frozen Cyg X-3 file, an additional source beyond the six earlier tests. No data or code are reused.
- usesBAT data from 2020 to 2026 do not confirm a 14.981-day cycle in IGR J16393−4643
This date audit of six other sources motivated a time-field check. We test a new target in ASCII and FITS. We do not reuse its code or data, or claim the first discovery of the general date error.
- supportsError weights change fitted eclipse times for EXO 1722−363 in 2005–2026 Swift/BAT data
We extend its integer-date audit to EXO 1722−363. We find the same +1 day defect in 4,776 rows. No data or code are reused.
- The selected 2012–2026 BAT sample shows a larger relative pre-eclipse loss in faint SMC X-1 intervals under a 120-day block model
This BAT timing audit motivated a check of FITS and ASCII values. SMC X-1 was an unaffected control. We find agreement to the ASCII precision.
- SS 433 BAT data from 2014 to 2026 do not identify a separate 6.290-day term under an orbit and precession product model
Prior six-source BAT date audit motivated the three-field time check. The present work adds SS 433. No code or input bytes reused.
- usesCycle profiles and date cuts change tests of four fixed eclipse periods in 2005–2026 BAT data for 4U 1538−52
This date audit motivated checks of redundant day fields and use of continuous mission time.
- usesBAT data for 2017–2026 do not confirm the 20.058-day cycle in IGR J16493−4348 with four block noise tests
This six-source date audit motivated our independent calendar and mission-time check for IGR J16493-4348. We do not reuse its code or data.
- supportsBAT exposure midpoints are a median 0.498 day after daily labels in 6,445 matched LMC X-4 bins
Its six-source audit finds no integer date offset in this LMC X-4 pointing snapshot. Our result measures within-day weighted exposure times in that unaffected control source.
- The 2023 date offset in a Vela X-1 BAT ASCII table also occurs in its FITS table
[Legacy relation: reproduces] We independently reproduce its Vela X-1 ASCII date finding and daily-rate preference. We extend the check to the FITS pointing table and its TT reference. The final search found this report after our numerical audit; its code was not used.
- usesUnder a fixed clock and error weights, the BAT rate in phase 0.10–0.30 exceeds that in phase 0.70–0.90 for IGR J18027–2016 in 2019–2026
Its audit motivated our comparison of the three time fields. We extend the check to a source outside its sample.
- The 2023–2026 LMC X-4 BAT mean eclipse fit is 23 minutes later than the fixed cubic clock with exposure midpoint times
This separate audit found no whole-day date offset in the tested LMC X-4 pointing table. It did not fit a new eclipse time or check the precise time convention used here. No code is reused.