3I/ATLAS

The ATLAS survey's 0.5-m telescope at Río Hurtado took four 30-second images of a faint moving object on 1 July 2025. Within a day the Minor Planet Center had named it 3I/ATLAS, and S. Deen, who measured the early images, reported that it looked fuzzy⁠[1]⁠[11]. Older images turned up quickly: ATLAS's own from Sutherland on 5 June, ZTF's from mid-June, the new Rubin Observatory's from 21 June, and TESS's from as early as 7 May, when it was beyond 6 au⁠[12]⁠[13]. It was already active there, and probably had been since about 9 au⁠[14].

Its path took it close to the plane of the planets and near Mars, Venus and Jupiter, but never closer to Earth than 1.8 au⁠[8]. From August the large telescopes and space observatories took spectra: JWST, SPHEREx, Hubble, Swift, the VLT, Keck, Gemini, ALMA and others⁠[3]⁠[15]⁠[16]. In early October it passed 29 million km from Mars, and the orbiters there imaged it; ESA used their positions to improve its orbit tenfold⁠[17]⁠[18]⁠[19]⁠[20].

Perihelion came on 29 October at 1.36 au, while it was on the far side of the Sun from Earth. Coronagraphs on STEREO-A, SOHO and GOES-19 and the PUNCH mission kept watch, and saw it brighten very steeply⁠[21]⁠[22]. ESA's Juice, the nearest spacecraft, measured its water in November⁠[23]. On 19 November NASA released images from more than a dozen missions and called it a comet⁠[10]. Telescopes followed it out through the winter and spring of 2026; it passed Jupiter on 16 March, and no observation by Juno has been reported⁠[8]⁠[24]. The results kept arriving: the deuterium in its water in Nature in June 2026, the carbon and nitrogen isotopes in July, the methane deuterium at the end of October⁠[6]⁠[25]⁠[7]. JPL's latest orbit, of 21 October 2026, was made to support a new JWST programme⁠[26].

Composition. Inbound at 3.3 au JWST found carbon dioxide at 7.6 ± 0.3 times water, among the highest ratios seen in any comet⁠[3]; SPHEREx traced the CO2 coma out to at least 420,000 km⁠[15]. CN appeared near 3 au with carbon chains strongly depleted⁠[27]; ammonia's products, NH and NH2, were scarce⁠[28]; methanol was among the most abundant relative to HCN of any comet⁠[29]; and the ion tail was rich in nitrogen⁠[30]. Near perihelion the balance shifted: the coma became water-dominated, CO2 fell to about a tenth of water, and by late December CO was the most abundant gas⁠[31]⁠[23]⁠[32]. Methane was first detected in mid-December⁠[33].

Nickel and iron. From 3.1 au the VLT saw nickel atoms in the coma with no iron; iron appeared only inside 2.6 au, and the total metal output was at least ten times other comets'⁠[34]⁠[4].

Isotopes. Water's deuterium-to-hydrogen ratio is 0.95 ± 0.06 per cent by JWST and above 0.66 per cent by ALMA, some 30 times typical comets; carbon-13 is depleted in CO and CO2; methane's D/H is 3.3 per cent, 14 times comet 67P's; and the carbon and nitrogen ratios in CN are high⁠[6]⁠[35]⁠[7]⁠[25].

Dust. Its light shows a negative polarisation of -2.7 to -2.9 per cent at small phase angles, deeper than any asteroid or comet, unchanged across perihelion⁠[5]⁠[36]. Its silicates are amorphous, with no measurable crystals⁠[37]. Whether water ice was seen in the coma is disputed⁠[38]⁠[39]⁠[40].

Size and motion. Hubble first limited the nucleus to a radius below 2.8 km, then separated it from the coma: 1.3 ± 0.2 km for a comet-like albedo⁠[16]⁠[41]. Its orbit needs a small push beyond gravity, about 4.6 × 10^-8 au per day squared at 1 au, peaking a month after perihelion⁠[26]⁠[42]. Its rotation period was measured at about 16 hours before perihelion and about 7 hours after⁠[43]⁠[44]. Four radio searches found no transmitters⁠[45]⁠[46]⁠[47]⁠[48].

Most of the oddities have a natural reading. The isotopes, the scarce ammonia and the amorphous silicates fit a body that formed in a very cold disc around a metal-poor star⁠[6]⁠[49]⁠[50]. A second reading holds that cosmic rays reworked its outer 15 to 20 m over billions of years, so the CO2-rich gas seen inbound came from a crust rather than the original ice⁠[51]⁠[52]; the fall in CO2 after perihelion fits that, though it also fits ordinary heating of deeper, water-richer layers⁠[53]. The nickel without iron is best explained by metal carbonyls, which sublimate far colder than metal-bearing dust⁠[4]⁠[54]. The polarisation is fitted by porous aggregates of tiny grains⁠[36]⁠[55].

The size was argued over from September 2025, when an analysis of early positions found no extra acceleration and inferred a nucleus above 5 km⁠[56]. The acceleration was detected once the arc reached perihelion, and the Hubble measurement put the radius near 1.3 km⁠[42]⁠[41]. Balancing the push against the gas lost gives a diameter of about a kilometre⁠[57]; the exact size depends on whether carbon dioxide or water carries the push, which the orbit alone cannot tell⁠[58]⁠[59]. Water rates from the same SOHO/SWAN data differ eightfold between two analyses⁠[60]⁠[61].

The age is in dispute too. Its speed places it among an old population above 7.6 billion years in one model⁠[62] and in the thick disc in another⁠[63]; a traceback through Gaia finds no home star and favours the thin disc⁠[64], and a kinematic age of 3 to 11 billion years leaves room for an object younger than the Sun⁠[65].

The technology reading began with a preprint by Hibberd, Crowl and Loeb, who described it as largely a pedagogical exercise and noted that a path this close to the planets' plane, passing near three of them, was improbable⁠[9]. Loeb placed it at level 4 of a ten-point scale and later listed 22 features he called anomalies⁠[66]⁠[67]. Each has been answered with comet physics: the sunward anti-tail by ice grains surviving longer on the sunward side, in a paper Loeb co-wrote, and the rest point by point by Jason Wright⁠[68]⁠[69]. No radio signal was found, and by December 2025 Loeb said the object was most likely natural⁠[70].

3I is too faint now for most telescopes and cannot be reached by a spacecraft. What is left is the data already taken, and a few more observations by the largest instruments.

The size can be settled from the last Hubble visits of program 18152, beyond 5 au, when the coma was faintest⁠[71]. Which gas pushes it can be settled by refitting the full set of 8,000-odd positions with the acceleration tied to the measured CO2 and water production curves rather than to standard laws⁠[72]⁠[73]. JPL's orbit solution 55 was made to support JWST program 13145; positions far out on the outbound leg will show whether the push has stopped, as outgassing predicts⁠[26]. The eightfold disagreement over water needs both hydrogen-coma models run on the same SOHO/SWAN frames⁠[60]⁠[61]. The age will be tested again when Gaia's next data release allows a better traceback⁠[64].

The larger test is more objects like it. The Rubin Observatory, which found 3I in its own commissioning images, began its ten-year survey in 2026 and is expected to find more interstellar objects⁠[12]⁠[74]. ESA's Comet Interceptor, due to launch in late 2028 or early 2029, could fly past one if it is reachable⁠[75].