The reported angle quantiles for TOI-1472 b and TOI-1648 b do not agree with their eccentricity components under either 180 degree convention

We test three planet solutions in Tables 5 and 6 of Carleo et al., DOI 10.1093/mnras/staf1958. The tables list two eccentricity components and a derived periastron angle. Their footnotes define the errors by the 16th and 84th posterior percentiles. Three such 68 percent intervals must have at least 4 percent common probability if they come from one joint posterior. We test whether any point in the three intervals obeys the polar angle relation. We allow both zero and 180 degree angle shifts. We also expand endpoints for rounding of every median and error width. TOI-1472 b and TOI-1648 b have no common solution under either shift. Their minimum angle gaps are 11.58 degrees and 19.73 degrees. Both conflicts remain with a larger rounding allowance of one full printed unit per number. TOI-1472 c has overlap after a 180 degree shift and gives no conflict by this test. An independent ray and rectangle calculation agrees with the angle bounds. Synthetic posterior controls also pass. The result assumes the written component definitions and a shared posterior with the stated quantile coverage. It does not identify a cause or give corrected orbit parameters. We release the proof, source transcriptions, dated archive discovery screen, provenance, code, results, and figures.

TOI-1472 · TOI-1648 · periastron · eccentricity · posterior quantiles · catalog consistency · union bound · reproducible astronomy