TIC 219329329.01

51,810 brightness measurements from six sectors, stacked on the 3.4816-day period. The dip is 0.97% deep and lasts 2.09 ± 0.05 hours.

The fitted model is a planet-size body crossing near the edge of the star (impact parameter 0.87 ± 0.02), which is why the dip has sloping sides and a short flat bottom.

Blue points are 10-minute averages with their uncertainties; grey dots are a sample of individual 200-second measurements. Photometry is my own 3×3 pixel aperture on the full-frame images.

Each panel is one 27-day TESS sector. The dips line up with the predicted transit times (vertical lines) in all six, across three years.

30-minute averages of the detrended light curve. Gaps are where TESS was downlinking data or the pipeline flagged the images.

A box-shaped-dip search of the 78 continuous days in sectors 102–104 finds one signal: 3.4816 days, at a signal-to-noise of 59.

The smaller peaks are the same signal at multiples and fractions of the period, as expected for a real periodic dip.

Subtracting in-transit images from out-of-transit images leaves a single point of lost light, on the target. Its centre is within about 1″ of the star in five of six sectors.

Sector 102 is 7.8 ± 4.1″ off (1.9σ). There the wing of a bright star 263″ away crosses the corner of the image; its own "dip" is a single event, not periodic.

The dip has the same fractional depth whether the light is summed over 4, 9 or 25 pixels: 9753 ± 121, 9784 ± 123 and 9752 ± 156 ppm.

If the dip came from a neighbour, a wider aperture would take in more of that star and the dip would deepen.

Changing the detrending window from 0.5 to 3 days moves the depth by about 1%. The official TESS-SPOC light curve and my own photometry agree to within a few percent.

The QLP pipeline's light curves run about 10% shallow, and its small-aperture product is far shallower and changes from sector to sector (3100–7200 ppm). My own 2×2 pixel photometry does not show this, so it is probably a normalisation effect in that product. This has not been confirmed.

The duration and shape of a transit measure the density of the star being crossed. With no prior knowledge of the star, the fit returns 0.61–0.73 times the Sun's density. Catalogues give 0.56–0.65 for this star.

A dip on a smaller, hidden star would return a density several times higher.

Odd-numbered transits are 9745 ± 170 ppm deep and even-numbered ones 9561 ± 173 ppm: 0.8σ apart.

Two unequal stars at twice the period would give alternating deep and shallow eclipses.

Half an orbit after the transit, the brightness changes by −196 ± 130 ppm: no eclipse. The 3σ upper limit is about 390 ppm.

A companion star would be hidden behind the primary at that moment and its light would drop out. The limit rules out a companion that shines; it does not rule out a very cool one.

A massive companion stretches the star, which then brightens and fades twice per orbit. The measured amplitude is 5 ± 54 ppm. The 3σ limit of 167 ppm caps the companion at 0.083 solar masses, or 87 Jupiter masses.

This excludes ordinary stellar companions. It does not separate a planet from a brown dwarf or from a star right at the bottom of the main sequence. The once-per-orbit terms are dominated by instrument systematics and were not used.

The period was fitted on sectors 68–104 only. The six transits in sector 105, not used in that fit, arrive −0.7 ± 1.3 minutes from the predicted times.

All 34 individually measured transits fit a constant period (χ² = 38.7 for 32 degrees of freedom). Sector 105 time stamps carry an extra systematic uncertainty of up to 1.5 minutes.

With the transits masked, a second search finds no other periodic dip. The star shows no brightness variation at the orbital period or at half of it.

A first pass flagged a 13.5-day signal. It consists of two events at the very start of data segments, where the pipeline light curve has ramps; with one day trimmed next to each gap, nothing remains above a detection statistic of 6. The official sector 68 light curve is flat to about 200 ppm.

The star has no entry in SIMBAD and is not a TESS Object of Interest. No variable-star or eclipsing-binary catalogue checked lists it. Gaia finds no orbital solution and no double image, but its astrometric quality number (RUWE) is 1.27, slightly above the value typical of single stars.

A search of every VizieR catalogue at once failed with server errors, so only the catalogues listed were checked.

The false-positive probability is about 6% (median 0.057 on both light curves). The probability that a known neighbouring star is responsible is 0.

Statistical validation needs a false-positive probability below 1.5%. This candidate falls in the "likely planet" class. All of the remaining false-positive weight sits in scenarios with an unseen companion star bound to the target, which high-resolution imaging would test. TRICERATOPS does not model brown dwarfs.

Scenario probabilities

Mean of six runs of 10 million samples each on my own photometry. The pipeline light curve gives the same ranking.

Run-to-run spread

Each dot is one run. Runs with 1 million samples scattered from 0.005 to 0.47 and were discarded as not converged; they are in the data file.