Rubin Observatory Saw Interstellar Comet 3I/ATLAS Before Anyone Knew It Existed

Ten days before anyone announced the discovery of an interstellar comet, the NSF-DOE Vera C. Rubin Observatory had already photographed it.

On July 1, 2025, the ATLAS survey spotted a faint moving object. Astronomers soon confirmed that it was 3I/ATLAS, also designated C/2025 N1, the third known object from outside the solar system. Rubin had been imaging the same area during commissioning and science validation. In a preprint posted to arXiv on July 17, 2025, Colin Orion Chandler and a large team report recovering the comet in Rubin images taken from June 21 through July 7.

Interstellar objects are small bodies that are not gravitationally bound to the Sun. They are passing through our planetary system after being ejected from planetary systems elsewhere. The first two known examples were 1I/'Oumuamua, which showed no visible coma and had a puzzling non-gravitational acceleration, and 2I/Borisov, which was an active comet. 3I/ATLAS is the third. Each one offers a sample of material from another star system, though the sample is brief and faint.

A comet hiding in commissioning data

Rubin's 8.4-meter Simonyi Survey Telescope and its 3.2-gigapixel LSSTCam can cover a wide field deeply. But at the time, Rubin was still being commissioned. The automated Solar System Processing pipelines were not yet running on these images, so the team used custom and partly manual tools to find and measure the comet. They measured the object's position and brightness on 37 images in the r, i, and z filters. Typical astrometric precision was about 20 milliarcseconds, with a systematic offset near 100 milliarcseconds, and photometric precision was about 10 millimagnitudes.

Rubin detected a dust coma around 3I/ATLAS as early as June 21, ten days before the comet was announced. A coma is the fuzzy cloud of dust and gas that surrounds an active comet. The team compared the spread of light around 3I/ATLAS with nearby stars and found excess emission beyond the sharp stellar profile. The coma's measured radial extent grew from about 6,520 kilometers on June 21 to about 9,380 kilometers on July 2. Because the comet was seen nearly head-on, the authors treat those sizes as conservative lower limits. The coma level also appeared to rise, though the increase is modest and uncertain.

A sunward tail and a size ceiling

From the brightness and an estimated ratio of coma to nucleus, the team derived a V-band, or visible-light, absolute magnitude of H_V = 13.7 ± 0.2. Absolute magnitude here is a measure of intrinsic brightness corrected to a standard distance. That corresponds to an equivalent effective nucleus radius of about 5.6 ± 0.7 kilometers. That is smaller than an earlier upper limit of about 10 kilometers from ZTF precovery data. The authors caution that their size is an upper limit: they assumed a spherically symmetric, steady-state coma, equal albedos for dust and nucleus, and a coma-to-nucleus ratio that is likely underestimated. A direct view with Hubble or JWST would help confirm the size.

Most comet dust tails point away from the Sun, pushed by radiation pressure. 3I/ATLAS showed a short tail pointing roughly sunward. Simple dust modeling for isotropic emission predicts a different orientation, so the authors say this strongly implies anisotropic dust emission. That could mean anisotropic dust release, or it could involve gas jets. Yet optical spectroscopy on July 3 detected no gas. Without knowing the nucleus's shape, no firm conclusion is possible. Similar sunward features have been seen in other distant active bodies, including comet Bernardinelli-Bernstein.

No wobble in an hour, and a preview of the Rubin era

On UT July 3 (the local night of July 2), a long sequence of images gave the team a chance to look for rapid brightness changes. They found no detectable variability on hourly timescales, ruling out variations larger than about 0.1 magnitude over less than an hour. That flat behavior could mean the nucleus is nearly spherical, its pole points toward Earth, or a steady coma is damping the light curve. The new astrometry also found no evidence so far for significant non-gravitational acceleration, unlike 1I/'Oumuamua. The orbit fits did show a small systematic offset in position, possibly caused by the asymmetric coma, but the team says it cannot rule out an instrument-related effect. They advise caution for users needing precision better than about 100 milliarcseconds.

The paper also uses simulations to ask what Rubin might have done with fully operational software. If Rubin's Solar System Processing pipelines had been running in real time, the team estimates that the commissioning observations would have been sufficient to identify 3I/ATLAS as a moving object. Looking ahead, simulations of the rest of the Science Validation survey predict roughly 100 additional Rubin exposures through August 22, 2025, if the planned strategy continues. Those are forecasts, not observed detections. They also depend on assumptions about the comet's brightness and activity.

The discovery has implications for the broader population of interstellar objects. The first two known objects were small, which suggested a steep size distribution. A larger body like 3I/ATLAS disfavors that simple picture and hints that the distribution may be wavy or piecewise, as seen in some solar system populations. But the number density of interstellar objects remains poorly constrained. Rubin's Legacy Survey of Space and Time is expected to find many more, with estimates ranging from about 5 to 50 over ten years.

The work is a preprint, so it has not yet completed peer review. The observations were made during a commissioning period, when data quality and software were still changing. The size estimate is an upper limit, the coma increase is a lower limit, and the mass-loss rate depends on unknown grain sizes, ranging from about 10 to 100 kilograms per second for plausible assumptions. Even so, a comet whose nucleus is estimated to be no larger than about 5.6 kilometers in radius was already in Rubin's data, waiting to be found. The paper's simulations suggest that, once Rubin's LSST is operational, such objects may be identified by Rubin itself rather than waiting for another survey to announce them.