Decades of Old Sky Photographs Sharpen the Orbit of a Newly Found World

On the night of Dec. 30, 1937, a telescope at Boyden Station in South Africa held its shutter open for an hour on a crowded patch of the southern Milky Way. On the glass plate it produced, cataloged as Harvard A19892, sits a small round smudge, a whisker from the spot where a world that nobody would notice for another 89 years should have been.

That smudge is the oldest and the weakest link in a chain that Joseph Murtagh, of the University of Washington's DiRAC Institute, and 10 colleagues have assembled for 2026 ED138, the body announced in October 2026 as the first dwarf planet candidate found by the NSF-DOE Vera C. Rubin Observatory. Their paper is a preprint, submitted to The Astrophysical Journal Letters on Nov. 2, 2026, and not yet peer-reviewed. The authors are also blunter about that plate than most people quoting it are likely to be. "Its flux is, however, likely dominated by noise, so we treat this position as marginal and report no magnitude," they write. Folding it into the orbit fit barely moves the answer, and "none of our following results depend on it."

What does the work is newer glass. Murtagh and colleagues report finding 2026 ED138 on 12 photographic plates taken between 1954 and 2000, at Palomar, at the UK Schmidt telescope in Australia and at two European Southern Observatory telescopes in Chile. Two of the co-authors re-measured every one of those detections separately, as a check that a faint blur really was the object and not a star. Near-infrared images from the VISTA telescope, a handful of SkyMapper frames and a decade of brightness measurements from the ATLAS survey carry the record forward to the present.

It had stayed hidden for a plain reason. 2026 ED138 lies far to the south and close to the plane of the Milky Way, and the crush of stars there defeats the software that normally picks moving objects out of survey images. Most searches of the outer solar system have avoided the region altogether. Going back and looking anyway buys a long arc: the record now runs from Rubin's discovery images back to the Boyden plate, and the archival positions improve the long-term prediction of where the object will be by a factor of five.

Fifteen chances to catch a shadow

That matters most for occultations, the brief moments when a body crosses in front of a star and its shadow sweeps over the Earth. Because 2026 ED138 sits against the plane of the Milky Way, bright stars are thick along its path. Searching for events between 2027 and 2030, the authors found 15 involving stars bright enough for the small portable telescopes that amateur networks use. Six of those shadows cross populated land, taking in West Africa and northern South America, Chile and Argentina, Namibia and South Africa, northern Australia and Hawaii.

An occultation is the most direct measurement there is of a distant body's size and shape, which is most of what it would take to confirm it as a dwarf planet. The archive does not make those predictions certain, though. The formal uncertainties the team computes are tighter, but they attach a warning to all of them: "the true path uncertainties may reach a few hundred kilometers until Gaia DR4 and continued Rubin astrometry refine the orbit." The long arc also does little for the earliest events, whose paths are largely set by Rubin's own 2026 data. What it buys is the long game.

A viewing angle that will not come back for a century

The second thing old plates buy is geometry that cannot be reproduced. 2026 ED138 takes centuries to circle the Sun, so our line of sight to it creeps round only slightly each year. Between 1954 and today, measured in the object's own frame, it has turned by 87 degrees. A flattened body seen first more pole-on and later more edge-on would have brightened or dimmed noticeably across that swing. Fitted plate by plate, the archival brightnesses do not: they stay consistent with the brightness the object shows today.

The conclusion drawn from that constancy is carefully bounded. Over 72 years of plate brightnesses, shapes flattened to 0.6 of their width or less are excluded for at least half of all possible spin-axis directions, and a Haumea-like elongated shape for about 90% of them. No shape is ruled out for every orientation, because a spin axis pointing the right way would hide the effect entirely. The team also reports a loose thread rather than smoothing it over: the two Palomar plates come out noticeably brighter than the rest; they cannot tell whether that is a calibration quirk or a real change in the object, and so they give the shape constraint both with and without them.

Two more results follow from the same archive. Near-infrared colors from VISTA put 2026 ED138 in the moderately red class of outer solar system bodies, redder than Haumea and lacking the water-ice signature that Haumea shares with the family of fragments it left behind, which supports the discovery team's conclusion that it is not one of them. There is no sign either of the methane frost that coats Eris. And the combined modern measurements give a tentative rotation period near 10.4 hours.

A method, not just a result

The longest view in the paper is a simulation rather than an observation. Murtagh and colleagues ran hundreds of copies of the orbit a billion years forward and backward in time, and found 2026 ED138 slipping into and out of a resonance with Neptune, a rhythm in which the two orbital periods hold a fixed ratio. It sticks there for tens of millions of years at a stretch, but "it rarely remains permanently captured." Its present arrangement, they conclude, is probably temporary.

How much of this is settled? The object itself is not in question: it was announced by the Rubin Observatory, registered with the Minor Planet Center, and recovered independently in five observing runs of NASA's TESS satellite by a team led by Te Han, although Matthew J. Holman, one of the object's two discoverers, is a co-author on both papers. The archival results are a different matter. The 1937 trace, the brightness record, the shape limits and the colors rest on this one unreviewed paper and nothing else.

The authors are explicit that the method is the point. The discovery team had already noticed the SkyMapper and ATLAS data without using them, so those positions and those brightness measurements appear here for the first time, and the same sweep could be run within days of any new find in the crowded band of the Milky Way. They name returning comets, active Centaurs (which are icy bodies among the giant planets), and the hypothesized Planet Nine, much of whose unsearched territory lies in exactly that band and whose predicted brightness at the bright end of its range would put it within reach of a single Schmidt plate.