Fig. 1: Transit light curves of Kepler-1625 b.

Each column shows data for one of the four transits (transits 1 to 3 from Kepler and transit 4 from Hubble), respectively. The out-of-Hill-sphere parts of the Kepler-1625 b transit light curves were detrended using a sum of cosines, and the LDCs were used as free fitting parameters. Time is in units of BKJD, which is equal to BJD − 2,454,833.0 d. a–d Orange lines visualize 100 planet–moon models that were randomly drawn from the respective posterior distributions for transit 1 (a), transit 2 (b), transit 3 (c) and transit 4 (d). Planet-only models are omitted as the corresponding Bayes factor of \(2\log_\mathrm{e}({B}_{\rm{mp}})=11.2\) suggests that the planet–moon interpretation is 265 times more probable than the planet-only interpretation. The best-fitting models of a planet only and of a planet with a moon are shown with solid and dashed black lines, respectively. Grey horizontal lines labelled as ‘Kepler mean’ illustrate the mean transit depth resulting from the three transits observed with Kepler. e–h Residuals of the observed data and the best fit of the planet-only model for transit 1 (e), transit 2 (f), transit 3 (g) and transit 4 (h). Red lines denote the five-bin walking mean. i–l Residuals of the observed data and the best fit of the planet–moon model for transit 1 (i), transit 2 (j), transit 3 (k) and transit 4 (l).