The most powerful all-solid rocket goes three-for-three off Shanghai, and the star-mapper that missed its orbit and remade the distance scale

01 What happened

Ignition came at 10:54 a.m. Beijing time, or 02:54 UTC, from a maritime platform in the waters off Shanghai.[1] Chinese state media called it Gravity-1's first open-sea launch and the country's first commercial sea launch staged in the East China Sea off the Yangtze River delta, farther offshore than the near-shore Yellow Sea drops the vehicle used before.[2] The rocket carried nine spacecraft: six Dongpo Earth-observation satellites for the operator Micro-Nano Star, plus Xiguang-2 01 and Tianyi-49 for remote-sensing work, and a technology demonstrator, Zidingxiang-3, testing an ultra-flat satellite bus and its attitude-control hardware.[3] All nine reached their assigned orbits.[1]

This was the third Gravity-1 to fly, after the January 2024 debut that lofted three Yunyao meteorological satellites and a second flight in October 2025.[3] Three launches, three clean insertions, is a rare early record for a solid-fuel vehicle this large.[1]

02 In numbers

Gravity-1 stands about 30 metres tall on a 4.2-metre fairing and masses roughly 400 tonnes fuelled, with sources putting the figure between 400 and 405 tonnes.[4][3] Its seven solid rocket motors produce about 600 tonnes-force at liftoff, near 5,900 kilonewtons, and lift 6.5 tonnes to low orbit or 4.2 tonnes to a 500-kilometre sun-synchronous orbit.[4][3] A published price near 39 million dollars per flight is what makes a nine-satellite rideshare pencil out.[4] That the vehicle is the world's most powerful launcher running only solid motors is the headline figure, and every other number is downstream of that choice.[1]

03 Why it's physically hard

A solid motor is a cast block of rubbery propellant with the oxidiser already mixed in, so once it lights it burns to depletion with no throttle, no shutdown, and no restart.[4] That simplicity costs performance: a solid delivers a specific impulse, the seconds of thrust a kilogram of propellant buys, of only about 250 to 285 seconds, where a good liquid engine reaches well past 340.[3] With no throttle to trim the burn, the rocket must fly a fixed thrust profile and steer purely by thrust-vector control, swivelling each motor's nozzle to point the acceleration where the guidance wants it.

Gravity-1 splits its seven motors into four strap-on boosters lit on the deck and a three-stage core whose motors are air-lit in sequence as the stack climbs.[4] Air-lighting means igniting a fresh motor high in near-vacuum after the one beneath it burns out, with no margin to relight if the igniter misfires, so the staging clock has to be right the first time. Hitting a 500-kilometre sun-synchronous orbit, one whose roughly 97-degree inclination makes its orbital plane precess in step with the Sun, demands tight burnout accuracy that a solid cannot buy with throttle; it has to come from consistent propellant grain manufacturing and a precise final attitude-control trim.

Launching from the sea adds its own tax. The barge pitches and rolls in the swell, yet the vehicle's inertial guidance must be aligned to a known attitude at the instant of release, and the range downrange has to be clear open water rather than a surveyed corridor.[2] Moving the operation into the deeper East China Sea, off a dense coastline, is a step up in that bookkeeping.[2]

04 What to watch

OrienSpace has signalled more Gravity-1 commercial batches through the rest of 2026 and, more consequentially, the maiden flight of Gravity-2, a larger, liquid-fuelled and partly reusable rocket the company is aiming to fly later this year.[3][1] That vehicle is the real test of the business: solids prove cadence and reliability, but recovering a kerosene-oxygen first stage is what would put a Chinese startup on Falcon-9 economics. next milestone Gravity-2's first launch.

01 The story

Parallax is the small angular shift a star appears to make against the far background as Earth swings from one side of its orbit to the other, and it is the one direct rung on the ladder of cosmic distance. The angle is tiny. A milliarcsecond is a thousandth of an arcsecond, about the width of a human hair seen from ten kilometres. Measuring it from the ground means fighting the atmosphere, which smears stellar images and had held reliable trigonometric distances to within roughly 40 parsecs of the Sun.[4] The idea behind Hipparcos was to escape the air entirely and let a spinning telescope tie the whole sky together.[3]

It launched on 8 August 1989 on an Ariane 4, sharing the ride with a broadcast satellite, and reached a geostationary transfer orbit as planned.[2] Then the Mage-2 apogee kick motor, the solid rocket meant to circularise it into a fixed slot 36,000 kilometres up, refused to fire.[2] Repeated attempts failed. The satellite was left on a long ellipse with a perigee near 210 kilometres and an apogee around 36,000, diving through the Van Allen radiation belts on every loop.[5] The observing plan had assumed a stable, distant vantage. It was gone.

Rather than write the mission off, controllers at ESOC rebuilt it around the orbit they had. The single planned ground station became three, at Odenwald in Germany, Perth in Australia, and NASA's Goldstone in California, to keep contact through an ellipse that no longer parked itself over one dish.[1] Operating procedures were rewritten from scratch.[1] Radiation gnawed at the electronics, and for months the satellite pointed itself on two working gyroscopes instead of three.[1] It kept observing for about three and a half years, and the spacecraft fell silent on 15 August 1993, four years after its August 1989 launch.[1] The early forecast had been a catalogue perhaps ten times worse than the specification.[5] The delivered catalogue held 118,218 stars at roughly one milliarcsecond, about twice as good as the original two-milliarcsecond goal.[2]

02 The hard part

Ground-based parallaxes are relative. Each shift is measured against nearby background stars that are themselves moving, so a systematic offset creeps into the zero point. Hipparcos was built to measure absolute parallax, and that demanded looking at two widely separated patches of sky at once. A beam-combining mirror fed the telescope two fields of view about 58 degrees apart, and as the satellite slowly spun, a fine grid in the focal plane turned each star's crossing into a modulated signal whose timing encoded its position.[3] Because the two fields were rigidly linked, angles measured across tens of degrees could be chained around the whole celestial sphere.

The catch was that no single measurement meant anything on its own. The final positions, parallaxes, and proper motions emerged only from a global least-squares solution that stitched millions of individual transits into one self-consistent, rigid sphere, a reduction run on the computing of the late 1980s.[3] The wrong orbit made this harder still, because passing through the radiation belts cost observing time and forced fewer looks at each star than the geometry had been designed to give.[5] That the solution still tightened to a milliarcsecond is the quiet achievement inside the loud one.

03 Why it mattered — measurably

Before Hipparcos, ground-based astrometry had delivered distances good to one percent for only a few dozen stars, and to five percent for about a hundred.[3] After it, more than 400 stars had distances to one percent and more than 7,000 to five percent, all in a single homogeneous catalogue of 118,218 entries at about one milliarcsecond.[3][2] A companion product, the Tycho Catalogue, fixed positions for 1,058,332 stars, later reprocessed into the 2,539,913-star Tycho-2.[2]

The consequence ran straight up the distance ladder. A uniform set of absolute parallaxes gave the first space-based calibration of the nearest standard candles and reset the zero point that longer-range methods depend on.[3] The record is honest about the limits, though. Hipparcos put the Pleiades cluster at 120.2 ± 1.5 parsecs, shorter than the ~130 parsecs other methods favoured, and a 2014 radio-interferometry measurement settled the true value near 136.2 ± 1.2 parsecs.[2] The discrepancy traced to small, spatially correlated errors in the satellite's scanning, a reminder that a milliarcsecond catalogue is not a perfect one. What is not in doubt is the scale of the jump: from roughly a hundred well-measured stellar distances to tens of thousands, achieved by a spacecraft that never reached the orbit it was designed for.

04 Echoes today

Hipparcos was the proof that astrometry belonged in space, and its direct heir is Gaia, launched in 2013. Gaia measures about two billion stars, with parallax errors near ten to twenty-five microarcseconds for the brighter ones, roughly a hundred times finer than Hipparcos and across tens of thousands of times as many stars.[6] Every modern map of the Milky Way's structure and motion rests on that leap, and the leap rests on a salvaged satellite on the wrong ellipse, still returning data as its gyroscopes died.