OrbitalHub

On 18 September 2025, the European Southern Observatory’s Very Large Telescope turned its FORS2 spectrograph toward a point of sky in Pisces that no telescope had ever measured at high resolution. The target, catalogued as TXS 2354+015, had been flagged two years earlier as a candidate high-redshift radio galaxy because the optical light from its host galaxy dropped sharply between the green and red filter bands of the Subaru telescope’s Hyper Suprime-Cam. The September observations, reported by Barbara Balmaverde of the INAF Astrophysical Observatory of Turin and four colleagues in a paper uploaded to the arXiv preprint server on 23 September 2026, produced a single bright asymmetric emission line at 7220 Angstroms with an equivalent width of 900 Angstroms and a flux of just under one ten-thousandth of a billionth of an erg per square centimeter per second. That signature, confirmed by a second emission line at the wavelength expected for NV at the same redshift, fixed the source’s redshift at 4.946, a value that places the galaxy as it looked when the universe was 1.1 billion years old and the cosmic microwave background was still warm.

The result makes TXS 2354+015 the most powerful radio galaxy currently known. The team measured its intrinsic radio luminosity at a rest-frame frequency of 500 megahertz, a value long used as the standard reference for comparing radio galaxies because the emission is uncontaminated by synchrotron self-absorption at the source’s redshift. They found 6.2 times ten to the twenty-ninth watts per hertz, almost twice the previous record, held since 2008 by TN J0924-2201 at 3.22 times ten to the twenty-ninth. The new source edges past a threshold that had stood for nearly two decades, and the gain comes not from a marginal improvement but from a clean factor of two.

High-redshift radio galaxies matter because they are the most extreme manifestations of accretion onto a supermassive black hole that the early universe produces, and because the energy the jets pump into the surrounding gas regulates star formation in the host galaxy and balances the cooling of the intracluster medium in its neighbourhood. Numerical simulations of how massive galaxies grow require this feedback to match the observed galaxy luminosity function, so finding the brightest examples at the highest redshifts is a way to test how aggressive that feedback could have been at the time galaxies were assembling their first bulges. The candidate selection is also where the new paper departs from tradition: earlier searches for radio galaxies beyond redshift three relied on the ultra-steep-spectrum technique, which is biased toward one subpopulation, and the team argues that selecting candidates by their optical dropout in the red filter of the Hyper Suprime-Cam Subaru Strategic Program, the Lyman-break method, catches high-redshift radio galaxies whose flat radio spectra would slip past the steepness cut. TXS 2354+015 itself has a radio spectral slope of 0.94, well short of the steepness criterion of 1.3 that traditional searches enforce, and the authors present it as the prototype of an entire missed population.

The follow-up spectroscopy was conducted at the VLT under programme 116.28LA.001 on 18 September 2025, and the radio data were assembled from two large surveys rather than a single observation. The team used the FORS2 spectrograph with the GRIS600RI grism and GG435 blocking filter, a setup that delivers wavelength coverage from 5120 to 8450 Angstroms at a resolution of about 300 kilometres per second; they placed the slit so that it included both the target and a brighter pivot star four arcseconds to the northwest, then took two 1218-second exposures. The extracted spectrum showed a continuum below the detection threshold on the red side of the line and a sharply asymmetric emission line whose blue edge drops within a single resolution element, the canonical signature of the Lyman-alpha line from a galaxy whose neutral hydrogen is being eaten away by the intergalactic medium at that epoch, and a second emission feature at the wavelength expected for doubly ionised nitrogen at the same redshift sealed the identification. The 150 megahertz TIFR GMRT Sky Survey, produced with India’s Giant Metrewave Radio Telescope, gave the low-frequency anchor at 25 arcsecond resolution and 17.5 millijansky sensitivity, while the Karl G. Jansky Very Large Array Sky Survey gave the high-frequency anchor at 3 gigahertz, 2.5 arcsecond resolution, and 70 microjansky noise. The team cross-matched the two catalogs, looked for HSC-SSP optical counterparts within 1.5 arcseconds of the VLA positions, and selected candidates with the Lyman-break dropout signature in the red filter. The candidate that survived was TXS 2354+015, with a radio flux density bright enough to ensure a robust identification and a clean offset of just 0.2 arcseconds between the VLA and optical positions. The 3 GHz image itself, retrieved from the HTML build of the paper, shows a single compact source on a flat background, the canonical appearance of an unresolved high-redshift radio core.

The combined picture is more striking than any single measurement. The host galaxy, as seen by Subaru in the i-band, has a half-light diameter of about 5 kiloparsecs, several times larger than the typical 1.5 kiloparsec radius of a z equals five galaxy in deep-field samples, and the Lyman-alpha emission line is spatially extended over a full width at half maximum of roughly 6 kiloparsecs, consistent with the giant Lyman-alpha nebulae that are known to surround powerful high-redshift radio galaxies and that are aligned with the radio jet axis. The radio position angle, measured from a marginal Gaussian fit to the VLASS image, is 71 degrees, while the slit was oriented at minus 30 degrees, almost perpendicular to the jet, so the team’s measurement likely underestimates the true extent of the nebula. The stellar mass the team infers from the host’s mid-infrared luminosity carries an unusually long list of caveats: the WISE W1 band at 3.4 micrometres captures several emission lines at this redshift, with the brightest being the doubly ionised oxygen line at 5007 Angstroms rest frame, and extrapolating the well-known correlation between radio power and that line’s luminosity from low-redshift samples suggests a host stellar mass near 2 times ten to the twelfth solar masses, a value that would place the galaxy at the very top of the local luminosity function, but a younger stellar population would reduce the estimate by a factor of ten and better spectral energy distribution coverage is required before the number can be trusted.

The discovery will not stay unique for long. The combination of the Lyman-break dropout selection with the cross-matched radio catalogs is now a template that can be reapplied to the rest of the HSC-SSP wide field, a 1400-square-degree footprint that the team estimates contains hundreds of similar candidates. Each one will need VLT or Keck spectroscopy to confirm its redshift, but the path from candidate to confirmed high-redshift radio galaxy is now routine rather than rare, and outside coverage has already framed TXS 2354+015 as the prototype of a population that the older ultra-steep-spectrum searches missed. The implication is that the radio-loud active galactic nuclei population at z greater than four is larger, flatter-spectrum, and more energetic than the existing sample suggested, and that the brightest examples may have been hiding in plain sight in the same wide-area optical surveys that astronomers have been using to find high-redshift galaxies for the past decade. Whether TXS 2354+015 is itself unique or merely the first of a new class is the question the next round of spectroscopy will answer.

Almost 320 light-years away, in the direction of the constellation Libra, a giant planet swings around its star in just under three days, glowing red and orange because the side that faces the star is hot enough to vaporize rock. WASP-189b is the textbook example of a class that astronomers call ultra-hot Jupiters, planets that look like Jupiter in size and bulk composition but orbit so close to their stars that their dayside temperatures climb above 2,500 kelvin, well past the vaporization point of minerals such as magnesium, silicon, and iron. New observations of WASP-189b with the Immersion Grating Infrared Spectrograph (IGRINS), mounted on the Gemini South telescope in Chile, have produced the first simultaneous measurement of gaseous magnesium and silicon in any exoplanet atmosphere, and the ratio between the two elements in the planet turns out to match the ratio in its host star. The result, published by Jorge Antonio Sanchez of Arizona State University and an international team in Nature Communications on 18 February 2026 and highlighted in a NOIRLab press release on 13 October 2026, is the first direct evidence for a long-standing assumption that astronomers have used since exoplanet science began (Sanchez et al. 2026, Nature Communications; NOIRLab press release noirlab2609).

The assumption in question is simple to state and powerful in practice. Planets form from the same protoplanetary disk of gas and dust as their host star, and the chemistry of that disk is set by the chemistry of the parent molecular cloud. The expectation, therefore, is that the relative proportions of the major rock-forming elements in a planet should match the relative proportions of those same elements in the star. In the Solar System this expectation has been borne out within a factor of about two for the rocky planets, and the small departures from a perfect match are useful constraints on the violent early history of the inner Solar System. For every other planetary system astronomers can detect, the question had to be inferred. Stellar spectra are easy to come by, because stars are bright, but exoplanet spectra are not, because exoplanets are faint, sit right next to a much brighter source of light, and are usually moving quickly through the background. Until the new WASP-189b measurement, no team had ever detected two rocky-element species in a single exoplanet at the same time, so the assumption that exoplanet chemistry tracks stellar chemistry was reasonable but unverified (SciTechDaily; ScienceDaily).

Why the magnesium-to-silicon ratio, in particular? The two elements sit at the heart of rock-forming mineralogy. Magnesium and silicon together build the silicates that make up most of Earth’s mantle, the olivine and pyroxene minerals that dominate the rocky bodies of the inner Solar System, and the iron-magnesium-silicate matrix that is the default material from which any rocky planet is expected to condense. The Mg/Si ratio of a planet, more than its absolute abundance of either element, sets the dominant mineral phases it can carry, which in turn control how easy it is for a planet to retain plate tectonics, generate a magnetic dynamo, and outgas volatiles that contribute to oceans and atmosphere. The same ratio in a star, measurable from high-resolution stellar spectra, has therefore been used for two decades as a proxy for the mineralogy of any rocky planets in that system, even ones whose spectra are beyond current reach. The WASP-189b result is the first observational check on that proxy, and it passes (NOIRLab Nuggets: Decoding Planet Origins with Gemini).

WASP-189b is a useful target for this measurement for a reason that has nothing to do with its bulk composition and everything to do with its orbit. The host star, catalogued as HD 133112, is a hot blue-white A-type star about 2.7 times the mass of the Sun. The planet is tidally locked, with the same hemisphere permanently facing the star, and its 2.7-day orbit takes it through a particularly favorable alignment called secondary eclipse, when the planet passes behind the star as seen from Earth. During secondary eclipse the planet’s own emission drops out of the combined spectrum and the star’s emission drops in, and a careful subtraction of one from the other yields the planet’s dayside spectrum in isolation. The IGRINS spectrograph, built by a consortium led by the University of Texas at Austin, Korea Astronomy and Space Science Institute, and the Korea Institute of Science and Technology, was a visiting instrument on Gemini South from 2022 to 2023 and was chosen because its near-infrared coverage and spectral resolving power are high enough to separate the planet’s individual atomic and molecular lines from the much brighter signal of the host star. The team collected a series of transit and eclipse observations of WASP-189b with IGRINS at Gemini South, then used high-resolution cross-correlation techniques to extract the magnesium and silicon lines from the planet’s contribution. The matched Mg/Si ratio was the central result (Sanchez et al. 2026, Nature Communications).

The instrumentation choice was a constraint as much as a convenience. Ground-based telescopes can deliver the spectral resolution that this kind of work requires only in narrow wavelength windows where Earth’s atmosphere happens to be transparent, and the near-infrared windows around 1.5 and 2.3 micrometers are the most productive because they sit in absorption bands of gaseous magnesium and silicon at the temperatures found in ultra-hot Jupiter atmospheres. The IGRINS spectrograph covers both windows simultaneously, which is what made the simultaneous detection possible. A more common approach in the field is to use a space telescope such as Hubble or JWST, but the trade-off is that those instruments do not have the same spectral resolving power and cannot easily separate the planet’s line from the star’s line for atomic species at high temperature. The success of the IGRINS run on Gemini South, and the fact that the instrument’s successor IGRINS-2 has now been commissioned as a facility-class instrument on Gemini North in Hawaiʻi, is a sign that high-resolution ground-based spectroscopy is on its way to becoming a standard tool for this kind of measurement, with more targets and more elements in the coming years.

A practical consequence of the result is that the long list of rocky exoplanet host stars whose Mg/Si ratios have already been measured, mostly by ground-based stellar surveys, can now be used with more confidence to predict the mineralogy of any rocky worlds in those systems. The assumption is no longer just a reasonable extrapolation from Solar System chemistry. It is a tested prediction, validated for one of the hottest and most alien gas giants in the catalogues, and almost certainly applicable, to a close approximation, to cooler and more temperate planets that no instrument yet exists to measure directly. For astrobiologists in particular the result closes a methodological gap: the standard practice of using a star’s elemental makeup to estimate the geochemistry of its planets, and through that the suitability of those planets for plate tectonics, magnetic fields, and long-term habitability, is now backed by direct observation rather than by an unverified assumption.

The IGRINS-2 commissioning on Gemini North, and the parallel development of other high-resolution spectrographs for the upcoming extremely large telescopes, suggest that the WASP-189b measurement is the first of a series rather than a one-off. The next steps are likely to involve measuring additional ultra-hot Jupiters with comparable precision, extending the technique to cooler gas giants and Neptune-class planets where the chemistry is more complex, and eventually applying the same kind of cross-correlation analysis to rocky planets around small stars, where the same Mg/Si assumption will need to be tested under different conditions. None of that happens quickly, and the WASP-189b result will remain the cleanest single demonstration of the principle for some time, but the door it opens is the same one that exoplanet science has been pushing on for the past two decades: turning stellar spectra, which are easy to obtain, into statements about planetary interiors, which are not.

When a radio source was first catalogued three decades ago as VLSS J0318.9+5755, it looked ordinary enough: a faint filament at low galactic latitude, originally mistaken for a pulsar and shelved. Then a team led by Andrea Botteon at Italy’s National Institute for Astrophysics (INAF) took a closer look with the LOFAR Two-metre Sky Survey, and the source turned out to be doing something the radio sky almost never shows. Curling from top to bottom in a single continuous line, it traces out a shape that looks, unmistakably, like a treble clef. In a paper published Wednesday in Astronomy & Astrophysics, the team combined new radio observations with NASA’s Chandra X-ray Observatory and the German-built SRG/eROSITA X-ray telescope to trace the formation to two events that happened hundreds of millions of years apart: an old outburst from a supermassive black hole, then a galaxy cluster collision that re-energised the fossil plasma. The NASA press release that landed on 14 October 2026 calls the shape a case of pareidolia – the human habit of seeing familiar patterns in random data – which is fair, since the resemblance is the most striking thing about it but says nothing about the physics that built it.

The reason the Treble Clef took so long to pin down is geography, not distance. The source sits at galactic latitude b ≈ 0.5°, almost directly behind the plane of the Milky Way. Stars, dust, and gas in our own galaxy block the optical light from anything that lies in that band, and the same obscuration makes it hard to identify the host. Optical surveys simply cannot reach that patch of sky, which is why this strip of celestial real estate has its own name among extragalactic astronomers: the Zone of Avoidance. Surveys that look through other parts of the electromagnetic spectrum have no such problem. X-rays punch through the galactic gas easily, and radio waves ignore it almost entirely, so a combination of LOFAR for the radio skeleton and Chandra and eROSITA for the hot intracluster medium turns out to be exactly the right pair of tools for the Treble Clef. Once that combination was applied, the source fell into place.

What the data shows is a textbook example of a radio phoenix. A radio phoenix is fossil plasma – relativistic electrons and magnetic fields that were puffed out by a previous activity, lost most of their energy through synchrotron and inverse-Compton cooling, and went dark at radio wavelengths. They sit in clusters for tens to hundreds of millions of years as invisible reservoirs of stored energy, and then something comes along to re-energise them. In the Treble Clef, that something is the merger of two galaxy clusters. The collision drove through the intracluster gas and generated shock waves, the same kind of weak shocks that are responsible for sonic booms from supersonic aircraft. When those shock fronts swept over the fossil plasma cloud, they re-accelerated the electrons and the cloud lit back up at radio wavelengths. The Treble Clef is the fossil turned visible again, and the merger shock is the trigger.

The host cluster, named after its coordinates in the VLSS catalogue, sits in the Zone of Avoidance but is otherwise a perfectly ordinary massive merging cluster. The Chandra and eROSITA X-ray observations show a roughly oval cloud of multimillion-degree gas, hundreds of galaxies, and a complex dynamical state. To pull the merger signature out of the X-ray image, the team used a residual technique: they subtracted the symmetric, oval-shaped hot gas that any relaxed cluster would have, and looked at what was left. The leftover emission was brightest at the left and right edges of the cluster and faintest in the middle, exactly what you would expect if the collision had been along the left-right axis. That residual image is what the new Chandra composite is built around. The radio image, taken from LOFAR’s LoTSS and LoLSS surveys plus new uGMRT follow-up, was then laid on top in red. The X-rays go in blue and purple, the infrared background from the United Kingdom Infrared Telescope goes in white, and the result is the magenta-on-blue image released on 14 October.

The shape itself is not a symbol that the universe was trying to draw. The vertical stroke of the clef is a long filament of old radio plasma that points roughly along the merger axis, and the loops at the bottom are short curls drawn into the plasma by gas motions in the merging intracluster medium. As the two clusters passed through each other, their gravitational potentials dragged the intracluster gas into a turbulent, asymmetric state, and the turbulence carved the curled loops out of what would otherwise have been a fairly straight filament. The resemblance to a treble clef is incidental. What is not incidental is that the structure is a remarkably clean laboratory for a specific question in plasma astrophysics: how do fossil relativistic plasmas respond to weak shocks, and what does the answer tell us about the energy budget of merging clusters.

The spectrum of the radio source tells its own story. Between 400 and 650 MHz, the radio spectrum is extremely steep, with a spectral index steeper than α = 4 in places, which means the source is unusually dim at high frequencies and bright at low frequencies. That is exactly the signature of an old, revived plasma cloud rather than a fresh injection from an active galactic nucleus. A new outburst from a supermassive black hole would have a flatter spectrum and would not be confined to a single, coherent filament. The steepness, combined with the morphology, is what convinced the team that this is a phoenix and not something else.

For the broader study of cluster mergers, the Treble Clef is interesting for two reasons. The first is that it sits in the Zone of Avoidance, where systematic surveys of radio halos and phoenixes have been patchy because optical and near-infrared work is hard. Adding a clean, well-characterised example to the catalogue is a reminder that the radio sky keeps surprises waiting behind the Milky Way. The second is that the paper reports a tentative radio halo near the cluster centre, the diffuse radio glow that is the canonical signature of a recent merger, which would make this cluster a textbook laboratory for studying the link between mergers, turbulence, and particle re-acceleration. The Botteon team plans to follow up with high-resolution, very-low-frequency observations using LOFAR 2.0, the upgraded array now being commissioned in Europe, which will resolve the curled loops at sub-arcsecond scales and constrain how the plasma is being re-accelerated.

The short version is that a source first catalogued in the 1990s as a faint low-frequency radio blip, almost written off as a pulsar, turned out to be a clean case study of how a galaxy cluster merger can light up the fossil plasma from an old black hole outburst. Chandra and eROSITA saw the cluster, LOFAR and uGMRT drew the clef, and the team from INAF put the story together. The clef shape is the eye-catcher, but the physics it illustrates is the real story, and the cluster is now on the short list for the next generation of low-frequency radio follow-up.

For most of the history of exoplanet science, planets that orbit their host stars in the wrong direction were curiosities reserved for the rare hot-Jupiter systems where gravitational chaos had clearly been at work. A team led by doctoral student Yann Carteret at the University of Geneva has now reported the first retrograde planet ever found around a small, cool red dwarf star, and the result has no obvious explanation. The planet, GJ 3090 b, is a sub-Neptune 73 light-years away in the southern constellation Phoenix, and it traces an orbit that runs opposite to the rotation of its host. For an M-dwarf system, that is new ground.

Why this matters comes down to a single number. The angle between a planet’s orbital plane and the equator of its star, called Psi by the people who measure these things, sits at roughly 136 degrees for GJ 3090 b. Anything over 90 degrees counts as retrograde. In our own Solar System, that angle is between three and seven degrees for every planet, and Mercury nudges a little higher but stays comfortably prograde. The handful of other misaligned systems spotted over the last fifteen years each had a massive companion on hand, another giant planet or a stellar partner whose gravity could tilt or flip a smaller neighbor’s orbit after the fact. GJ 3090 does not. The second planet in the system is too small to do that kind of work, and there is no sign of a stellar companion. Whatever put this world on its current path had to do so without the usual accomplices, which means the textbook explanations of misaligned hot Jupiters do not apply. The find, published in Astronomy & Astrophysics, forces the field to consider mechanisms that would otherwise be filed under unlikely.

The story begins with TESS, NASA’s Transiting Exoplanet Survey Satellite, which flagged GJ 3090 b’s shadow back in one of its southern-sector campaigns. A transit is the easy part. The hard part is figuring out which way the planet is moving when it crosses the star, because that is what tells you whether the orbit is prograde or retrograde. The team watched five separate transits with the NIRPS infrared spectrograph, an instrument built in Geneva and installed on ESO’s 3.6-metre telescope at La Silla Observatory in Chile. NIRPS is tuned to the near-infrared, which is exactly the part of the spectrum where a cool, dim red dwarf like GJ 3090 puts out most of its light, and where telluric absorption from Earth’s own atmosphere is least annoying.

What the team was after is a phenomenon called the Rossiter-McLaughlin effect. As a star rotates, the hemisphere spinning toward Earth is slightly blueshifted, and the hemisphere spinning away is slightly redshifted, from the regular Doppler motion of the stellar surface. When a planet transits, it blocks a small patch of that rotating surface. If the planet moves across the star in the same direction as the star’s spin, the planet first covers a blue patch and the star appears briefly redder, then it covers a red patch and the star appears bluer, leaving a characteristic bump-dip pattern in the radial-velocity curve during the transit. For GJ 3090 b, the team saw the order reversed on every one of the five transits, with the red patch covered first. The orbital obliquity they derived is 136 degrees, in line with the inverted transit shape, and consistent across nights. There is no ambiguity left in the geometry, and that is what makes the result clean.

The science and engineering of the result sit at the same intersection that has made red dwarfs the most-studied stellar hosts over the past decade. M dwarfs are smaller and cooler than the Sun, so a sub-Neptune-sized planet produces a deeper transit signal and a larger radial-velocity wobble per unit mass than the same planet would around a Sun-like star, but the wobble from a 2.2-Earth-radius, 4.5-Earth-mass world with a 2.9-day period is still small and the transit still short. NIRPS’s design wavelength range of roughly 0.95 to 1.8 micrometres lines up with the infrared peak of a cool red dwarf, and its stability across a single transit, and across five transits taken on different nights, is what allowed the team to distinguish a genuine 136-degree obliquity from a smaller angle buried in noise. Carteret and his co-authors, including Léna Parc and Vincent Bourrier at UNIGE, point out that GJ 3090 b is now the smallest exoplanet to have its orbital obliquity mapped around any M dwarf. The technical milestone is as important as the discovery itself, because it means the technique works on the kinds of planets that are most common in the galaxy.

What remains genuinely puzzling is the missing companion. In every previously known misaligned multiplanet system, the configuration can be traced to gravitational interactions with a much larger body, either a second giant planet in the same system or a binary star whose orbital plane pulled things off course over hundreds of millions of years. GJ 3090 has a second planet, but it is not massive enough to do that kind of work, and there is no sign of a stellar companion. The team’s favoured hypothesis is a primordial disk flip, the idea that after the original protoplanetary disk dissipated, a second, misaligned disk of gas and dust fell onto the star from the surrounding interstellar medium. If that second disk arrived at the right angle, the planets that formed inside it would inherit its orientation rather than the star’s. There is precedent in so-called Peter Pan disks, which survive far longer than standard disk-evolution models predict, around other young stars. GJ 3090 b might be a snapshot of what happens when one of those long-lived disks finally does form planets, and the planets turn out to be running around the wrong way, in the wrong direction.

The takeaway is that this is the kind of measurement that quietly resets how a discovery category gets built. Until now, retrograde exoplanets were a hot-Jupiter phenomenon with a tidy dynamical explanation. Now there is a sub-Neptune on a 2.9-day orbit around a small red dwarf, and the dynamical explanation has fallen away. NIRPS will keep taking data, and the same technique can be applied to any of the dozens of transiting M-dwarf planets already on the books. Carteret’s team plans to extend the survey, and at least one follow-up is already in the queue. Whether GJ 3090 b turns out to be a one-off oddity or the first example of a new population of backwards travelers around small stars is the question that the next year or two of observations will answer.

Roughly 300 light-years from Earth, in a part of the Milky Way that no catalogue has bothered to name, two objects circle each other every 86 minutes and 39 seconds. One is a small star, a red dwarf about 85 times the mass of Jupiter. The other is a brown dwarf about 25 times Jupiter’s mass, an object too heavy to be a planet yet too light to fuse hydrogen the way stars do. They orbit so close together that the entire configuration would fit inside the diameter of the Sun. And the smaller one has been quietly leaking material onto the larger one, at a rate slow enough that the larger object will be feeding on its neighbour for billions of years. The system, named ZTF J0440+2325, is described in a paper appearing today in Nature Astronomy by Aaron Householder, Kevin Burdge, and collaborators at MIT and partner institutions, and reported in a parallel MIT News feature by Jennifer Chu. It is the first confirmed case of stable, long-lived mass transfer from a substellar object onto a main-sequence star, and it pulls a tidy assumption about what happens to planets and brown dwarfs that end up too close to a star.

The tidy assumption is the swallow-engulfment model. When a planet, a brown dwarf, or any other substellar object drifts close enough to its host star, the standard picture is that the host tidally strips or disrupts it on a short timescale. The engulfed material gets swallowed and incorporated into the star, the companion disappears, and the binary turns into a single object with a slightly higher mass. This is the fate that astronomers have routinely inferred from observations of close white-dwarf and neutron-star binaries, and it is the fate that theory predicts for the Earth when the Sun leaves the main sequence in five billion years and expands into a red giant. The brown dwarf in ZTF J0440+2325 was supposed to be on the same conveyor belt, dragged inward by tidal friction and consumed on a timescale of thousands to millions of years. Instead, it is doing something much slower and stranger. It is leaking material onto its companion at a rate of about one hundred-thousandth of an Earth mass per year, a number the team works out in the Nature Astronomy paper using the photometric wobble of the binary and the assumption that the donor is overflowing its Roche lobe along the L1 Lagrange point. That is about 40 million dump trucks of material per year by the team’s own analogy, a quantity that sounds large but is a tiny fraction of the brown dwarf itself. The star will not run out of food before the Galaxy ages out from under it.

What the team saw first was the triangular waveform. The Zwicky Transient Facility, a wide-field optical time-domain survey that uses a camera on the 48-inch Samuel Oschin Telescope at Palomar Observatory in California, had been picking up a strange light curve from a star in Taurus for several years. Supernova light curves are bell-shaped: gradual rise, peak, gradual fade. Buried in the ZTF archive was a signal that did not look like a bell at all. It looked like a triangle, repeating on a period of about 87 minutes, with no counterpart at other wavelengths and no obvious explanation. Burdge, an assistant professor of physics at MIT who had been searching the ZTF data for periodic variables, noticed it during a search for a different kind of system. He originally suspected a “black widow binary,” the colloquial name for a millisecond pulsar that is slowly eroding a low-mass stellar companion with its particle wind and radiation. In a black widow, the optical light curve wobbles because the companion is being whipped around the much heavier neutron star, and the orbit usually shows a clean Doppler signature of that motion. The ZTF light curve did not wobble in the same way, which is what tipped the team off that the two objects were not a David-and-Goliath pair but a more balanced system in which neither dominates. Householder, a graduate student in MIT’s Department of Earth, Atmospheric and Planetary Sciences, reanalysed the ZTF photometry and combined it with follow-up time-series spectroscopy. The wobble was small, the spectrum showed hydrogen and helium emission lines characteristic of an M dwarf, and the orbital fit locked the masses at roughly 85 and 25 Jupiter masses. The orbit fit inside the Sun.

The simulation work clinched the picture. Once the team had a mass, a radius, and an orbital geometry, the next question was whether the brown dwarf was overflowing its Roche lobe, the teardrop-shaped region of space around a body in a binary inside which material is bound to that body. If the brown dwarf were inside its Roche lobe, no mass transfer would be possible; the two objects would simply orbit forever without exchanging material. If the brown dwarf were overfilling the lobe, material would spill through the L1 point and rain down on the companion’s surface. The team ran a set of smoothed-particle hydrodynamics simulations with the measured masses, the measured orbital separation, and a brown-dwarf radius tuned to match the photometric brightness. The simulated test particles do not orbit the brown dwarf indefinitely; they stream across the L1 point and pummel the surface of the red dwarf at high speed. The team describes the geometry as more like an asteroid hitting the Moon than the usual accretion disk around a black hole or a neutron star, where material spirals inward through a flattened disk before settling onto the central object. In ZTF J0440+2325, the brown dwarf is too small for its own gravity to capture the inflowing material into a disk, so the stream lands directly on the surface and heats a roughly circular hot spot at the impact point. That hot spot is what produces the triangular waveform. As the brown dwarf orbits, the hot spot rotates in and out of view, and the system brightens when the hot spot faces Earth and dims when it faces away. The peak of the triangle is the moment of direct face-on view, when the impact fireball is brightest.

The mass-transfer rate is low enough that the system is genuinely stable on astrophysical timescales. Most binary mass transfer is unstable on some level: the donor shrinks as it loses mass, the orbit responds by widening or shrinking in ways that either accelerate or choke off the transfer, and the system either spirals toward a common-envelope merger or settles onto a longer-period detached configuration. What the simulations show for ZTF J0440+2325 is the rarer case in which the mass-loss timescale of the brown dwarf and the angular-momentum loss timescale of the orbit are tuned just right that the rate stays roughly constant for hundreds of thousands of years and may hold for the entire remaining lifetime of the brown dwarf, on the order of billions of years. The brown dwarf, as a substellar object, cannot fuse hydrogen and so does not have a main-sequence lifetime in the usual stellar sense, but it has a thermal-cooling timescale, and the team argues that the binary will outlive that timescale many times over. The brown dwarf will not be eaten quickly. It will be eaten slowly, continuously, and quietly, and the Galaxy will be a different place by the time the meal is finished.

What the team plans to look for next is more of the same. The triangle-shaped light curve is a distinctive signature, and the ZTF archive contains years of survey photometry that have not been systematically searched for it. The paper names a second candidate, ZTF J1444+4820, with an even shorter 67-minute period and a similar wobble signature. If that system confirms, the slow-eating configuration stops being a one-off and becomes a routine endpoint of binary evolution when the masses and separation land in roughly the right place. The triangular waveform, the small wobble, and the absence of a Doppler signature that would betray a heavy compact object are the three ingredients of the search. The same configuration, in principle, applies to small stars and giant planets, and the team plans to look for those as well. The conventional picture of close planetary companions has to share the page with a quieter alternative: a slow, steady meal that outlasts planets.

On 11 October 2026, NASA’s Perseverance rover rolled past a line nobody had crossed before. After 299.6 feet (91.31 meters) of driving on its 2,006th Martian day, the rover’s odometer read 28.10 miles (45.22 kilometers), nudging it 0.04 miles past the 28.06-mile (45.16-kilometer) total that Opportunity, its predecessor, had built up across nearly fifteen years of service on Mars. The new figure makes Perseverance the farthest-driven vehicle ever operated on the surface of another world, with margin enough that the gap will only widen as the rover keeps climbing the western rim of Jezero Crater. The number is small in absolute terms, a Sunday stroll in a city park would cover more ground, but on Mars the steps add up slowly, and the record has stood since 2018.

The reason this matters has less to do with the odometer than with what the odometer is measuring. Setting a distance record on another planet is not a vanity project. It is a budget statement: how much ground a mission can cover in a fixed number of sols, how many distinct outcrops the science team can visit, and how many rock and regolith samples the rover can collect before its hardware gives out. Perseverance has collected twenty-nine samples already, from an ancient lakebed at the bottom of Jezero, from a delta fan built by a long-vanished river, and now from the rim itself, where the geology changes character. Each new mile opens another valley, another ridge, another chance to find traces of ancient water chemistry that a stationary lander would never see. The original mission specification called for no more than 12.4 miles of driving. The team has now more than doubled the contract.

This is the fourth chapter in a thirty-year American experiment in mobile Mars geology, and each rover in the sequence has reset what counts as routine. Sojourner, the microwave-oven-sized cart that landed with Mars Pathfinder in 1997, rolled about 330 feet (100 meters) during an 83-sol mission before its solar panels gave up. Spirit and Opportunity arrived in 2004 with ninety-sol prime missions, and Opportunity, in particular, refused to stop. It crossed the first extraterrestrial marathon in 2015, logged 5,111 sols of operation, and racked up the 28.06-mile total that Perseverance just surpassed. Spirit covered 4.8 miles before getting stuck in soft soil and succumbing to a Martian winter in 2010. Curiosity, which landed in Gale Crater in 2012, is still working and has logged more than 23.65 miles (38.07 kilometers). The pattern is the same each time: longer missions, longer drives, and a better understanding of what the rovers can be trusted to do on their own. Perseverance also has the lunar record in its rear-view mirror, having already overtaken the 22.2 miles covered by the Apollo 17 astronauts in 1972 and the 24.2 miles racked up by the Soviet Union’s Lunokhod 2 rover in 1973.

The deeper reason Perseverance is pulling away is software. A navigation system called AutoNav lets the rover think while it drives, building a three-dimensional map of hazards on the fly and picking its own path between waypoints chosen by rover planners on Earth. The version flying on Perseverance is orders of magnitude sharper than the original AutoNav that flew on Spirit and Opportunity: where Sojourner evaluated the path ahead with twenty measurement points, Perseverance processes six hundred thousand of them each time it looks. Roughly ninety percent of Perseverance’s total driving distance has come under AutoNav, compared with about five percent for the earlier generation. The rover has covered as much as 1,351 feet (412 meters) in a single sol under AutoNav, a figure that would have been unthinkable during the Opportunity era. Steven Lee, the project manager for Curiosity and Perseverance at NASA’s Jet Propulsion Laboratory, called the milestone a privilege to be part of, noting that the original mission specification capped driving at 12.4 miles and that all mobility systems remain in good shape for many more years of record-setting. Mark Maimone, deputy team chief for robotic operations, said the team is not finished improving Perseverance’s driving: newer software teaches the rover to minimize steering, roll more, and cross-check its position against orbital maps mid-drive. The next engineering target is glare in the navigation cameras, which still forces the rover to stop driving by mid-afternoon whenever the Sun is ahead, and teaching it to look past those artifacts would give every future Mars rover a longer working day. NASA Science blog (12 Oct 2026), NASA JPL news (11 Oct 2026)

The science payoff from the extra mileage is already showing up. The record-breaking sol took place inside a strip of terrain nicknamed Lac de Charmes on the western rim of Jezero, an area of the crater wall the science team picked specifically because the rocks there are some of the oldest Perseverance will ever see, pieces of crust old enough that their Earthly equivalents have been erased by erosion and plate tectonics. Mars has no plate tectonics, so the equivalent record is still readable on the surface. The rover is heading toward an outcrop named Idubi, where it will look for evidence of ancient hydrothermal systems, the kind of mineral-rich, hot-water environments that on this planet often host the most interesting microbial communities. The earlier leg of the mission already produced one potential biosignature, at a rock called Cheyava Falls in the ancient river delta, and Lac de Charmes is a chance to test whether similar chemistry shows up in a completely different part of the crater. NASA approved Perseverance’s first mission extension this month, a commitment that will carry the rim work through roughly the end of 2028 and, after Lac de Charmes, toward Rainbow Falls, a 328-foot-wide (100-meter-wide) block of ancient crust that was shoved into place by the same impact that created Jezero in the first place. The image that the team released to mark the milestone is a navigation-camera frame from the record sol: a stretch of rocky Martian ground under a dusty sky, with fresh rover tracks curving off into the distance. It is a working picture, the kind of frame the navigation team uses to plot the next day’s drive, not a glamour shot from Mastcam-Z, and that feels right for what is being celebrated. As long as the wheels turn and the software holds up, the number on the dashboard will keep climbing, and the science team will keep pulling rocks out of rocks that have been sitting on the surface of Mars for four billion years. NASA Science blog (12 Oct 2026), NASA Mars 2020 mission page, NASA Photojournal PIA26812 (13 Oct 2026), Sci.News (Oct 2026)