What It Takes to See the Universe

On September 3, the team controlling BepiColombo waited for a machine 200 million kilometers away to tell them that it had successfully taken itself apart. The Mercury Transfer Module, which had carried two scientific spacecraft through nearly eight years of interplanetary travel, was being discarded. After separation, the remaining spacecraft had to change its orientation, reconfigure itself, and report home. A preliminary radio signal offered an early indication of success. Full confirmation reached the control room in Darmstadt at 15:49 Central European Summer Time. ESA published its account on September 7. ESA mission report

The delay was partly a matter of distance. Light takes about eleven minutes to cross 200 million kilometers. But the nearly two-hour interval between the planned separation and full confirmation also included the spacecraft’s own sequence of operations. The people on Earth could prepare, simulate, and watch. They could not reach into the machinery when the moment came.

Watch: BepiColombo separation broadcast, September 3, 2026, 2 hours 31 minutes. The official replay includes live mission control footage and expert commentary. Credit: European Space Agency (ESA), with third-party material credited in the original programme. Linked in full; ESA restricts modifications and reuse of excerpts.

This was one of the more tangible developments in a week whose largest scientific question concerned something nobody can put on a test stand. A new analysis of exploding stars strengthened a hint that dark energy might change over time. Radio flashes helped researchers investigate where galaxies have pushed their gas. Old X-ray observations revealed unfamiliar objects, while Hubble and Webb found small bodies preserving clues to the construction of the solar system. Beneath these very different results lay a shared difficulty: building a reliable connection between something that happens far away and something we can justifiably say about it.

The trouble with a standard candle

The dark-energy result began with an attempt to make old measurements agree about their own meaning. Ryan Camilleri and the Supernovae Unite collaboration brought together two major collections of Type Ia supernovae, Pantheon+ and DES-SN5YR, producing a consistently processed sample of 2,884 probable explosions. The work was submitted as a preprint on September 4 and revised on September 9. Its importance lies partly in the patient work concealed by the word “combined.” Observations made with different instruments and analyzed through different procedures cannot simply be poured into a larger spreadsheet and treated as equivalent. Supernovae Unite cosmology paper

Type Ia supernovae are useful because their brightness can be standardized well enough to estimate their distances. Compare how luminous an explosion is inferred to be with how bright it looks from Earth, and one can estimate how far away it is. Measure how much the expansion of the universe has stretched that light toward redder wavelengths, and the explosion becomes a marker in the history of cosmic expansion. Repeated across many distances, these measurements helped establish that the expansion is accelerating. Dark energy is the name given to whatever is driving that acceleration. Its simplest description, the cosmological constant, assigns empty space an energy density that stays constant as the universe expands. NASA explanation of supernova distance measurements

The measuring stick, however, is an exploding star. Its brightness requires corrections, and those corrections depend on such things as how its light changes over time, its color, and the properties of the galaxy in which it exploded. Even after the usual adjustments, supernova brightness correlates with the stellar mass of the host galaxy. A companion analysis led by Jaemyoung Lee remeasured those galaxy masses consistently. It showed that changing the mass estimates can materially shift the statistical case for evolving dark energy in part of the sample. Small astronomical housekeeping decisions can reach surprisingly far into cosmology. Host-galaxy analysis

When the unified supernova measurements were combined with observations of the cosmic microwave background and large-scale patterns in galaxy positions, a model allowing dark energy to evolve fit better than the standard cosmological constant. The revised analysis put the preference at 3.1 to 3.3 standard deviations, depending on the statistical estimator. It also reduced the allowed area for two parameters describing dark energy’s evolution by roughly 30 percent relative to the comparison combination identified in the paper. Revised cosmological results

That is reason to investigate, not a discovery announcement. In the same paper, Bayesian model comparison found only weak evidence for the evolving alternative. These conclusions ask different questions. A more flexible model can fit a particular dataset better; the additional question is whether that improvement is persuasive after accounting for the range of possibilities the model permits. The answer depends on the statistical framework and assumptions. A significance quoted in standard deviations is not the probability that the proposed explanation is true.

There is no need to choose between excitement and skepticism here. A cleaner dataset makes the discrepancy more interesting, while the sensitivity to calibration makes independent measurements more valuable. Both Unite papers remain preprints, and the cosmology paper says its distance and likelihood products will become public upon acceptance. The useful next development would be another way of measuring cosmic history that encounters different sources of error.

Bursts that find what is missing

Fast radio bursts offer a particularly unusual tool for investigating a related problem. These brief flashes arrive at Earth after passing through enormous stretches of tenuous plasma. Lower radio frequencies are delayed more than higher ones. The frequency-dependent delay measures the accumulated free-electron content along the route. A burst lasting milliseconds can therefore carry information about matter that is difficult to see directly. Research on radio bursts as probes of ordinary matter

In a paper published in Nature Astronomy on September 8, Kritti Sharma and colleagues used 114 localized bursts to infer fluctuations in the distribution of ordinary matter and the effects of galactic feedback. Their measurements are especially sensitive to relatively nearby groups and clusters of galaxies. The resulting constraints were already competitive with earlier measurements from the Atacama Cosmology Telescope and eROSITA. Sharma and colleagues

The difficulty they are addressing begins with the fact that galaxies rearrange their surroundings. Energy released by stars and accreting black holes can heat gas and push it outward. Cosmologists also study matter by measuring the slight gravitational distortion of distant galaxy images. To interpret that distortion precisely, they need to know how ordinary matter has been redistributed. Otherwise, an imperfect account of gas can complicate a search for subtler effects involving neutrinos or the dark sector.

The radio bursts do not by themselves identify new dark-energy behavior. They offer an independent constraint on the gas that makes other measurements harder to interpret. The sample is still modest, and translating electron delays into a matter distribution requires modeling. But the authors have released their data and a reproduction package. The important prospect is a new way to check an astrophysical correction that precision cosmology increasingly needs to get right. Paper and replication resources

A population that escaped notice

Sometimes the obstacle is more basic: an entire population of objects has escaped recognition. Mustafa Muhibullah, Jimmy Irwin, and Rosanne Di Stefano searched Chandra’s archive for sources emitting exceptionally low-energy X-rays. Their work identified 84 objects across six nearby galaxies. The September 9 release described them as “hypersoft” X-ray sources, with emission concentrated at energies below about 0.3 kiloelectronvolts. The discovery came from asking a new question of observations already collected. Chandra science release

The sources probably include several kinds of compact stellar systems: white dwarfs, neutron stars, or black holes drawing material from companions. In such systems, falling gas can become hot enough to radiate at high energies. But the observations do not yet identify each source uniquely. Much of their inferred output falls in extreme ultraviolet light, which intervening gas absorbs before it can reach us. Astronomers are reconstructing an obscured part of the spectrum from the portion they can detect.

One possibility connects these objects back to the supernova measurements. If some are white dwarfs retaining material drawn from companions, they could help researchers investigate routes toward Type Ia explosions. None has been established as such a progenitor. Their potential contribution to ionizing surrounding gas is another question for follow-up observations. The immediate achievement is narrower and still substantial: researchers have found objects whose behavior demands an explanation, in an archive that had already been available for years.

Counting what is too small to see

Far beyond Neptune, another investigation used old and new observations together to examine a different kind of survivor. The Kuiper Belt contains icy remnants of the material from which the planets formed. Its larger members are much easier to find than its smaller ones, creating an awkward selection effect. A history reconstructed mainly from large survivors can miss what happened to the much more elusive small bodies.

The new Webb search addressed that problem by shifting and stacking images. An object too faint to stand out in one exposure can become detectable when many exposures are added together. But an orbiting body moves between exposures. Researchers must account for that motion so that the object’s weak signal accumulates in one place instead of smearing across the image. In a field covering only 0.05 square degrees, the search found 27 very faint trans-Neptunian objects. Thirteen were also recovered in Hubble observations, extending the available information about their colors and orbits. NASA overview of the two studies

Their brightness implies sizes reaching roughly 5 to 10 kilometers, depending on the assumed reflectivity. That qualification is fundamental. A small bright surface and a larger dark surface can send the same amount of light toward a telescope. These are inferred diameters, not measurements taken across resolved disks. Even so, the survey reaches a population that has been exceptionally difficult to study directly.

The researchers found fewer very small objects than a simple extrapolation from larger ones would suggest. Small members of the dynamically cold population, whose orbits are relatively orderly, also followed the narrow color relationship seen among larger members. Studies led by Marielle Eduardo and Anastasia Morgan therefore bring both abundance and surface properties into the discussion of how these bodies formed and survived. The results support the persistence of primordial compositional differences, but the field is small and the history remains open. Finding fewer fragments could reflect how planetesimals formed, how collisions processed them, or some combination of the two.

Getting there is its own problem

Every one of these observations depends on another kind of progress, easier to photograph but no less demanding to interpret. On September 5, Isar Aerospace’s Spectrum reached orbit from Andøya in Norway on its second flight. Its first attempt, in 2025, had lasted about thirty seconds. This time the vehicle completed staging, reached orbital velocity, and released satellite payloads. The flight also met the orbital milestone in ESA’s European Launcher Challenge. Isar Aerospace flight report, ESA assessment

Watch: Spectrum launch coverage and photographs, hosted by ESA. Footage and photographs: Isar Aerospace. Linked to the publisher’s original presentation; reproduction rights remain with the credited owner.

The significance is an additional launch route for European customers, with a new vehicle operating from a Norwegian spaceport. Europe already had launch systems, notably those operating from French Guiana. Spectrum adds a different provider and location. For a small satellite, a dedicated launch can also offer more control over timing and destination than traveling as a secondary passenger on somebody else’s mission.

One successful insertion establishes that the system can reach orbit. It cannot yet establish the launch rate, recurring cost, or reliability of a mature service. Spectrum’s advertised ability to carry up to 1,000 kilograms to low Earth orbit is likewise a performance target that this flight alone does not demonstrate. The next constraint moves into the factory and the launch schedule: building another vehicle, reproducing the result, and doing both often enough to support customers’ plans.

An engine tested on the ground

India’s propulsion work showed an earlier point in the same progression. On September 5, engineers at Mahendragiri tested the powerhead of the SE2000 semi-cryogenic engine at conditions corresponding to its full rated thrust level. The complete test lasted 35 seconds, including five seconds at full power. The hardware incorporated the engine systems except the thrust chamber. This was therefore a demonstration of powerhead operation at the intended level, not a full engine producing its rated thrust through a flight chamber and nozzle. ISRO powerhead test report

The proposed SC120 stage would use liquid oxygen and refined kerosene to replace LVM3’s existing L110 core, which uses storable liquid propellants. Liquid hydrogen belongs to the vehicle’s cryogenic upper-stage system. The distinction matters because these are separate engineering changes with different development schedules. Validating the powerhead’s switch between propellant supplies prepares the way for longer tests; integrating and qualifying the complete engine still lies ahead. ISRO semi-cryogenic development, LVM3 architecture

A second test, on September 9, concerned hardware closer to flight. ISRO acceptance-tested the CE20 engine assigned to LVM3-M7 at an uprated thrust of 22 tonnes-force and demonstrated a liquid-oxygen tank pressurization module intended for Gaganyaan C32 stages. The mission was scheduled for the fourth quarter of 2026. A development test asks whether a design can perform as intended; a flight-acceptance test asks whether a particular piece of hardware is ready for its assigned job. Both are necessary, but they justify different expectations. ISRO CE20 test report

Even after launch, obtaining a useful view can require considerable propulsion work. India’s EOS-05 completed three orbit-raising maneuvers after its September 4 launch. The final engine firing lasted 1,247 seconds, almost twenty-one minutes, and left it in an estimated orbit ranging from 34,903 to 35,884 kilometers in altitude. Its intended geosynchronous vantage point offers persistent regional access for Earth observation. The completed burns establish an orbital milestone; they do not establish calibrated, operational imagery. EOS-05 final maneuver, Mission description

Watch: GSLV-F17 liftoff and onboard camera footage, 1 minute. Video: DOS/ISRO. The link opens the official recording.

The work that begins after launch

Europe’s MTG-I2 reached a comparable transition. ESA reported on September 9 that the weather satellite had reached its geostationary operating region and had been handed to Eumetsat the previous day. It joins an earlier imager and an atmospheric sounder, completing the first three-satellite Meteosat Third Generation set. Their combined observations are intended to improve the detail and frequency of weather monitoring over Europe and North Africa. Commissioning still has to turn a functioning spacecraft into a dependable source of measurements. ESA MTG-I2 update

That work continues beyond the satellite. Instrument responses must be understood, products validated, and observations incorporated into weather models before better hardware can reliably produce better forecasts. The same broad problem appears in the supernova catalog and the radio-burst analysis: collecting a signal creates an opportunity, while establishing what the signal means requires another body of work.

What the module left behind

BepiColombo still has months of that work ahead. Its discarded transfer module used solar electricity to accelerate xenon ions, producing a gentle thrust that could operate for long stretches of the cruise. The remaining spacecraft will use chemical propulsion to manage arrival. Mercury orbit insertion is scheduled for November 21, followed by Mio’s deployment in early December and the European orbiter’s final approach to its science orbit. Full science operations are planned for April 2027. Its transfer module has been left in an orbit around the Sun, having completed the task for which it was built. ESA arrival sequence

What remains attached is the reason for the journey: instruments intended to examine Mercury’s surface, interior, and magnetic environment. Their measurements will join the accumulated records through which people on Earth try to reconstruct places they cannot visit. Some answers will arrive promptly. Others may wait in an archive until someone asks a question the original observers had not thought to ask, or learns how to distinguish a faint population that had been there all along.