HD 139139, the "Random Transiter"

K2 Campaign 15 observed the field from August to November 2017; HD 139139 was proposed as a target by three guest-observer programmes⁠[2]. The dips were first seen in the Kepler pipeline light curve during a visual survey using the LcTools software, then analysed in a light curve rebuilt from the pixels with the K2SFF method and high-pass filtered to remove starspot modulation⁠[2]⁠[6]. In 2018 the team took spectra with the TRES spectrograph on Mount Hopkins and the 2.7-metre telescope at McDonald Observatory, and imaging to measure the neighbour's motion⁠[2]. The paper was posted in June 2019⁠[2].

The discovery paper is the only full account. Its authors include the citizen scientists who found the dips through a long-running visual survey of K2 light curves, which has also found unusual eclipsing systems and multi-planet systems that automated searches missed⁠[2]. Its stated purpose is to bring the object to wider attention and to ask for better spectra of both stars and high-precision monitoring of the fainter one⁠[2].

The record is one 87-day light curve at 30-minute cadence. Of the 28 events, all but two are 200 ± 80 ppm deep; durations range from 0.75 to 7.4 hours; the mean interval is about three days; depth and duration are not correlated⁠[2]. Box-least-squares and other period searches find no significant period⁠[2]. Tests for rolling-band artefacts, centroid shifts during dips, electronic cross-talk from bright stars and bad pixels found nothing; nearby stars on the detector do not show the dips⁠[2]. Independent reductions of the same pixels, such as the EVEREST light curve, are public alongside the pipeline and K2SFF light curves⁠[7]⁠[8]⁠[9].

The two stars are well characterised. Star A, HD 139139 itself, is a G-type star of about 5,800 K with solar metallicity, a rotation period of about 14.5 days and a gyrochronological age of 1.5 ± 0.4 billion years, at 106.7–107.8 pc from Gaia. Star B is a K5–K7 star of about 4,400 K, 3.3 arcseconds away and 2.7 magnitudes fainter in the Gaia red band, at a photometric distance of 115 ± 20 pc, with a similar radial velocity and proper motion; the pair is probably bound at about 400 au⁠[2]. Radial velocities rule out stellar or brown-dwarf companions to star A on orbits shorter than about 10 days and low-mass stars on orbits under about 100 days⁠[2]. Century-long DASCH plate photometry and five years of ASAS-SN data show nothing unusual⁠[2].

If the dips are on star A, the bodies would be about twice Earth's radius; if on star B, whose light is diluted about fifteen times by star A, they would be one to two times the size of Jupiter⁠[2]. TESS observed the field in Sector 91 in 2025, and full-frame-image light curves for two catalogue entries at the position are in MAST⁠[8]⁠[10].

Planets fail on timing. Multiple planets, planets around both stars, and a few planets with very large timing variations cannot reproduce more than about four or five of the dip times; an eccentric planet in a close binary, which would scramble transit times, still left a detectable period in simulations and is opposed by the radial velocities and the single-star spectrum; a circumbinary planet is ruled out by the lack of velocity variation⁠[2].

A disintegrating planet on an orbit of under about 5 hours, transiting only when it happened to be shedding dust, would need to be visible in only 7 per cent of more than 400 transits, near the limit at which a periodicity search would still find it; the average dip profile also lacks the asymmetry such dust tails produce⁠[2]. A swarm of dust-emitting asteroids on many orbits could give random times, but not so many dips of almost the same depth⁠[2].

Two stellar explanations remain open. The dipping seen in young stars with disks could occur on star B and be diluted, but diluted light curves of ten known dippers do not resemble this one and there is no model that explains dippers quantitatively⁠[2]. Spots larger than Earth (on A) or Jupiter (on B) that appear within half an hour, last a few hours and vanish would break no physical law, and such events could have gone unnoticed in other stars; this hypothesis is new and untested⁠[2]. Instrumental origin cannot be fully excluded with one data set⁠[2]. Other scenarios have been put forward in two research notes⁠[3]⁠[4]; a technosignature search with the Green Bank Telescope has been reported⁠[5].

A second detection is the first need. High-precision photometry of the pair over several days, resolving the two stars, would show which star dips and whether the dips still occur, which the discovery team asked for directly⁠[2]. A search of the 2025 TESS light curve for dips of the same depth and duration, with injected test dips to measure sensitivity, would show whether TESS can see them at all. If the dips recur, their colour would distinguish dust (deeper in blue), solid bodies (grey) and dark spots (dependent on spot temperature); more radial velocities would tighten the limits on hidden companions. If they never recur in a sensitive search, an undetected instrumental cause in the single K2 data set becomes more likely.