A ten-thousand-planet haul from TESS first light

The tally of known exoplanets used to grow in careful increments: a handful of hot Jupiters confirmed a year, a Neptune or two around a bright star, each new world announced with a press release and a photograph of the telescope that found it. A paper posted to arXiv on 20 April, from Joshua Roth, Joel Hartman, Gáspár Bakos and their colleagues at Princeton, changes the pace of that growth by an order of magnitude. Working through the full-frame images TESS took in its first observing cycle, the team has uncovered 10,091 new candidate planets, more than doubling the number of candidates the mission’s own pipelines had produced.

TESS watches swathes of the sky for the faint, repeated dips in starlight that mark a planet crossing in front of its host. Its official search has always favoured the brightest stars, where a transit is easiest to detect and follow-up with ground telescopes is practical. That makes sense, but it introduces a bias. Planet occurrence rates predict that planets should be common around fainter stars too; they were simply being overlooked because no one was willing to sift through the noisier data. The T16 project set out to correct exactly that bias.

The scale of the data is the first thing that arrests the reader. The team detrended and corrected the systematics in 83,717,157 light curves, one for each star TESS observed down to a brightness of 16th magnitude, across its primary mission. From that volume, a semi-automated, machine-learning-assisted search returned 11,554 transit candidates with orbital periods between half a day and twenty-seven days. Of those, 411 showed only a single transit and were left without orbits, 1,052 were repeats of already-known TESS candidates, and 10,091 were genuinely new.

Volume alone is not a result; a machine can manufacture false positives at speed. The authors validate the pipeline by picking one candidate host, TIC 183374187, a metal-poor thick-disk star, and measuring it with radial-velocity follow-up on the Magellan telescope. The wobble is real, and the transit signal resolves into a newly identified hot Jupiter. That single confirmation is the proof of principle: the candidates are not noise, and the faint-star population the official pipelines skipped over is populated with planets that stand up to scrutiny.

Why this matters is the census argument that underpins everything in exoplanet science. Estimates of how common Earth-like worlds are, and around which kinds of stars, are only as good as the sample they are built from. A catalogue skewed toward bright, nearby stars is a catalogue skewed a certain way, and every statistical inference drawn from it inherits that skew. By pushing the search down to fainter stars, the T16 team is filling in the part of the population the bright-star surveys could not see, and the implication, stated plainly in the paper, is that the true population of transiting planets has been sitting there all along, waiting for someone to look at the fainter data.

The honest caveats are familiar. Most of these 10,091 candidates are not confirmed planets; they are candidates, and the follow-up work to weed out eclipsing binaries and instrumental artefacts will take years. A single hot-Jupiter confirmation demonstrates the pipeline but validates only one box on a very long list. Still, the direction is unmistakable. The era in which a new exoplanet was a rare event ended some time ago, and this survey suggests the back catalogue already contains the discoveries of the next decade. The night sky has been sending its catalogue downlink all along. It just took a patient read to transcribe it.