Kosmos 482

Four days after Venera 8 left for Venus, its twin stood on the pad at Baikonur. The second Venera-72 spacecraft, vehicle V-72 No. 671, lifted off on 31 March 1972 at 07:02:33 Moscow time, 04:02:33 UT. It never left Earth. The escape stage shut down early, the Soviet press gave the stranded craft the routine name Kosmos 482, and its Venus lander spent the next 53 years and 40 days circling the planet it had been built to leave. It came down on 10 May 2025. When, to the minute, and where are still open.

Each of the pair was a flyby bus carrying a lander, to be released on approach to Venus. Lavochkin had rebuilt the 1972 lander around what Venera 7 found two years earlier, designing it for 105 atmospheres and 493 °C instead of 150 atmospheres and 540 °C and saving 38.5 kg. A new IOV-72 photometer was added because this pair was aimed at the day side, and the next generation of landers would need to know whether there was enough light for pictures.

Scans of a pre-launch review, held by the Russian State Archive of Scientific-Technical Documentation (RGANTD) and published by RussianSpaceWeb, set out the plan: a day-side landing about 1,000 km from the terminator and at least 30 minutes of work on the ground. The same pages record a trade that would matter within weeks. To save mass, the telemetry equipment had been taken off the 2VL escape stage, and the military acceptance office warned that without it, in an emergency, there would not be enough information.

The first three stages did their job, and the fourth, a Blok L variant that Lavochkin calls NVL and the commission papers 2VL, lit for Venus. It stopped after 125 seconds. RussianSpaceWeb gives the planned firing as 243 seconds; McDowell puts the cutoff 2 minutes into a 4 minute burn. Spacecraft and stage were left on a long ellipse of roughly 210 by 9,800 km, where the readings of the Great Soviet Encyclopedia yearbook for 1973, RussianSpaceWeb, McDowell and Langbroek differ only slightly.

As with other Soviet failures that reached orbit, the announcement gave a Kosmos number and said nothing of Venus. McDowell and RIA Novosti both write that it was to have been Venera 9, and RIA's 2018 account says the Kosmos name was used so as not to report a failure. Venera 8 flew on and landed on 22 July 1972; its lander, the one Kosmos 482 had duplicated, sent data for 50 minutes (63 in NSSDCA) after touchdown.

What stopped the engine is less settled than many summaries suggest. Lavochkin says only that an accident of the NVL stage cost the station its trajectory, and Roscosmos wrote in 2025 of an upper-stage malfunction. Asif Siddiqi's NASA chronology blames a failure in the on-board timer, and the Russian text of Mikhail Marov's account, on the epizodsspace archive, an erroneous command from the timer. RussianSpaceWeb cites reports that the stage fired earlier than planned; Langbroek (2022) wrote of an apparently wrongly set timer; RIA Novosti (2025) says the stage ran short for an unknown reason. None of the sources gathered here shows the analysis behind the timer explanation, and the stage flew without the telemetry the review had said would be needed to diagnose a failure.

Some of the launch hardware fell within days. Around 1 a.m. local time on 3 April 1972, people on New Zealand's South Island reported flaring objects in the sky, and titanium spheres about 38 cm across turned up south of Ashburton: three and a fourth five weeks later according to Langbroek, five according to RussianSpaceWeb (one of them 25 cm) and RIA Novosti (2025). Their origin is disputed. RussianSpaceWeb links them to the Block L stage; Langbroek argues for the escape stage's ullage unit, which re-entered on 2 April and which his modelling puts over Ashburton at about the right time, and notes that the Block L itself stayed in orbit until 1983.

The stranded spacecraft was still working four weeks later. On 28 April 1972 the State Commission, chaired by A. G. Mrykin, ordered Lavochkin and the GOGU flight-control group to separate the descent craft from the orbital compartment of V-72 No. 671 and then stop work with the vehicle, on a date to be agreed later. The decision gives no reason; RussianSpaceWeb, which published it, suggests a test or a way to reduce the mass that would eventually fall. Declassified on 10 August 2016, it did not reach the public, RussianSpaceWeb says, until 2022.

Western trackers saw the result without knowing what it was. Langbroek's orbit analysis puts the separation in mid-June 1972, some 80 to 85 days after launch, with a parting speed of about 8.5 m/s; McDowell's catalogue gives 15 June with a question mark, and the Russian text of Marov's account the end of June. The new object became number 6073, 1972-023E. RussianSpaceWeb says the US military catalogued it in June 1972; Langbroek dates its first catalogue entry to July.

The bus, 1972-023A, had far more drag area for its mass and re-entered on 5 May 1981. Object 6073, a dense ball, sank far more slowly. Catalogued as debris for decades, it was labelled the separated descent sphere by McDowell in September 2000 and by NORAD in June 2002. Langbroek later gathered the evidence: radar cross-sections of 0.72 square metres (the median in archived CSpOC data) and 1.3 (LeoLabs), and his own photometry, whose steady brightness matches a one-metre white sphere.

Not everyone accepted that the object was the lander alone. RussianSpaceWeb compared a 2011 image by the Dutch astrophotographer Ralf Vandebergh with the assembled spacecraft and suggested the lander was still attached to the bus. Langbroek's 2022 modelling showed that anything carrying much of the bus would have come down decades earlier, while the observed decay matched a one-metre sphere of about 495 kg. RussianSpaceWeb now calls the 2011 impression likely an illusion, and reads the elongated shape in Vandebergh's April 2025 images as possibly a partly deployed parachute, trailing insulation or an imaging artefact. Nobody has settled which.

The forecasts converged slowly. In July 2018 the astronomer and space historian Pavel Shubin told RIA Novosti the station would return within four to seven years, and RIA reported his graph as pointing to about 2023 to 2025. McDowell and Langbroek both rejected press claims of a 2019 re-entry, McDowell saying early to mid 2020s. Langbroek, writing in May 2022 when the orbit was 198 by 1,957 km, gave a window from mid-2024 to mid-2027 and leaned towards its early part.

On 7 May 2025 Roscosmos said the craft would enter the dense atmosphere on 9 or 10 May. ESA's Space Debris Office opened a live forecast the same day; by the evening of 9 May it had narrowed to 06:37 UTC on 10 May, plus or minus 3.28 hours. On 8 May the TIRA radar at Fraunhofer FHR in Germany caught the capsule tumbling. On 10 May radar in Germany tracked it at about 04:30 and 06:04 UTC (ESA credits Fraunhofer FHR with the last detection, EU SST names TIRA); it was not seen at the next expected pass, 07:32 UTC.

Everything after 06:04 is modelling. Roscosmos, quoting a TsNIIMash calculation, put entry into the dense atmosphere at 06:24 UTC (09:24 Moscow time), with no uncertainty stated. ESA's aftercast was 06:16 UTC plus or minus 0.36 hours. EU SST, counting re-entry at 80 km, said the craft decayed within its final window, 06:04:48 UTC plus or minus 20 minutes. The Aerospace Corporation's last forecast was 06:29 plus or minus two hours, and the TU Delft Tudat model run by Marco Langbroek and Dominic Dirkx gave 06:40 plus or minus 91 minutes. McDowell's catalogue lists 06:34 with a question mark, his newsletter of 23 May 2025 about 06:38. The US Space Force's CSpOC prediction, 05:32 UTC plus or minus 12 minutes, cannot be right: the lander was still in orbit over Germany at 06:04. Read with that sighting, the EU SST window ends at about 06:25, and the Roscosmos time sits just inside it. Langbroek and Dirkx judge about 06:15 to 06:45 UTC most probable.

The place is even less certain: at that speed, as Langbroek and Dirkx put it, one minute of time is almost 500 km of ground track. Roscosmos named two separate points, entry into the dense atmosphere 560 km west of Middle Andaman Island and a fall into the Indian Ocean west of Jakarta. McDowell's catalogue gives 102 E, 10 S, with a question mark. Langbroek and Dirkx's nominal points for the other times scatter from England (EU SST) and southwest Asia (ESA's aftercast) to the seas south of Australia (their own aftercast and CSpOC's), with the Roscosmos time at 11 N 88 E and the Aerospace Corporation's at 5 S 99 E; RussianSpaceWeb reported the Space Force estimate as over the Pacific. The likely area, Langbroek and Dirkx add, was in daylight and mostly sea, which would explain the lack of ground observations. ESA's last word, on 13 May 2025, was that no precise time or location had been identified and no sightings or impacts on the ground reported.

Whether the lander survived is a matter of judgement, and judgements differ. ESA wrote beforehand that it might survive; EU SST afterwards that it most likely reached the ground almost intact. Langbroek and Dirkx called intact survival very likely and estimated an impact at 65 to 70 m/s. Shubin, described by RIA Novosti in 2025 as a mathematician at the Space Research Institute (IKI RAN), told it the lander would very probably survive a fall into water or sand and stay sealed even a kilometre deep. His IKI colleague Natan Eismont told Izvestia the opposite (in an article of 30 April 2025 that dates his remarks to 20 April): the craft would almost completely burn up, leaving at most heat-resistant parts such as gas bottles and engine pieces. The case for survival rests on the design: the lander was built to enter Venus's atmosphere at 11.6 km/s, and its twin took up to 335 g doing so, while Shubin put the Earth entry speed at about 8 km/s. Langbroek, though, had judged in 2022 that the parachute was extremely unlikely to work after 50 years in space.

Even the mass has several readings. The literature usually gives 495 kg, as did ESA; Roscosmos said under 500 kg, EU SST approximately 500, RussianSpaceWeb 480, and McDowell's catalogue 475, with 495 as a queried total. Langbroek and Dirkx found that 480 kg best reproduced the 53-year decay but noted that a 3 percent change in drag coefficient or area would do the same, so the fit cannot pin the mass down. Design limits vary too: Lavochkin and Siddiqi give 105 atmospheres and 493 °C, while ESA and Shubin give 100 atmospheres and up to 300 g.

Kosmos 482 had a forerunner. In August 1970 the twin of Venera 7 was stranded the same way and announced as Kosmos 359; Siddiqi records that its Blok NVL engine ignited late and cut off after 25 seconds, and that the spacecraft re-entered on 6 November 1970. Its descent sphere, listed separately by McDowell as 1970-065D, came down from a 206 by 836 km orbit on 6 October 1971. Kosmos 482's longer burn threw its payload almost 10,000 km out, and that apogee kept its lander up for 53 years where the Kosmos 359 sphere lasted about 13 months.

Nearly eighteen months after the fall, the picture has not changed. Checks made for this page on 4 November 2026 found no report of a sighting or of recovered hardware, and McDowell's catalogue, updated on 2 November 2026, still carries question marks on both the time and the place. What survives with certainty is paper: a flight plan for a lander that never went, a warning about a stage without telemetry, and a short order to cut the lander loose and stop work.