The GD-1 Gap: An Invisible Thing Hit a Star Stream

Stretch a thread of stars across the northern sky, from Ursa Major toward Cancer, so thin and so straight it looks drawn with a ruler. That's GD-1, one of the slenderest things anyone has ever mapped in our galaxy. For almost its whole length it does exactly what a well-behaved river of stars should do: flow in a clean, narrow line.

Then you reach one spot, and the line breaks. The ribbon thins to a gap — and just off to the side, a stray tuft of stars hangs where no stars belong. Something punched straight through it. Here's the part that keeps astronomers up at night: whatever did the punching has never been seen.

What we actually know

Two astronomers, Carl Grillmair and Odysseas Dionatos, found GD-1 in 2006, hiding in plain sight. It showed up as a faint, 63-degree-long band of stars in star-count maps from the Sloan Digital Sky Survey (Grillmair & Dionatos 2006, ApJ Letters). And it is absurdly thin — about half a degree wide on the sky, a real width of only some 70 parsecs, which makes it at least 100 times longer than it is wide.

So what is it? The stars are old and metal-poor (an iron abundance of roughly [Fe/H] = −2.2), which is the giveaway. GD-1 is almost certainly the corpse of an ancient globular cluster, a tight ball of stars that the Milky Way's tides caught, stretched, and pulled into one long thread (Li et al. 2018, ApJ). Look for the cluster it came from and you find nothing: no surviving globular cluster anywhere has the right orbit, speed, and chemistry. The original is simply gone, unravelled completely into the line we see.

And that thinness is the whole point. A stream this fine is a precision instrument. Every star in it rides nearly the same orbit, so any nudge from outside leaves a mark you can read — a scar in the starlight. In 2018, Adrian Price-Whelan and Ana Bonaca went looking for those scars. They combined the European Space Agency's Gaia Data Release 2 astrometry with Pan-STARRS photometry, and the marks snapped into focus. Their map stretched the known stream another 20 degrees and exposed clear thin patches — gaps — along its length, plus two odd clumps of stars sitting just off the main track. They named them the "spur" and the "blob" (Price-Whelan & Bonaca 2018, ApJ Letters).

A gap with a spur right beside it. What makes that exact pairing? The next year, Bonaca teamed with David Hogg, Price-Whelan, and Charlie Conroy to run it down. Their answer was a hit-and-run: a close, fast gravitational fly-by. Picture a dense object screaming through the stream — it grabs the nearby stars, drags them off course, and in one motion scoops out an empty patch while flinging a spur of stars to the side. That double signature is hard to fake. Fitting the geometry, they boxed in the intruder: somewhere between about 10^5.5 and 10^8 solar masses (roughly 300,000 to 100 million Suns), and — this is the killer detail — packed into a scale radius under about 20 parsecs. Their best-fit model dated the collision to around 495 million years ago, with most workable answers landing inside the last billion years (Bonaca et al. 2019, ApJ).

Then they went hunting for the culprit, suspect by suspect. No known globular cluster swings near the impact site at the right moment. No catalogued dwarf galaxy fits. A giant molecular cloud drifting in the disk? Ruled out — those clouds are far too puffy to land a blow this sharp. The data wanted something small, dense, and dark.

The question nobody can answer

Here's the whole mystery in one breath: the thing that hit GD-1 left a clean gravitational fingerprint, matches nothing we can see, and is denser than our best theory of dark matter says it has any right to be.

That last bit is the sharp edge, so sit with it. In the standard cold dark matter (CDM) picture, the Milky Way's halo should be crawling with thousands of invisible dark-matter "subhalos." One of those, at the inferred mass, barrelling through the stream — that's a tidy explanation. Except the perturber has to be far too compact. Its required density sits roughly 2–3 sigma above what CDM subhalos of that mass are supposed to reach (Bonaca et al. 2019). So we're left with three uncomfortable doors. Either GD-1 was clipped by a freakishly dense member of an expected crowd. Or by something stranger than we've catalogued. Or our whole picture of dark matter on small scales is missing a piece. As of mid-2026, nobody has caught the object by any other method. Its identity is still wide open.

The leading guesses

Everything below this line is informed speculation, and I'll flag it as such. The facts above are solid. The explanations that follow are rival ideas, and not one of them is confirmed.

A dark-matter subhalo — the front-runner. The first guess is still the favorite: the thing is a clump of dark matter with few stars or none, precisely the kind of dark substructure CDM cosmology says should haunt galactic halos. If that's the answer, GD-1 would hand us one of the first times anyone has caught a starless dark-matter clump by its gravity alone — which would be thrilling. The snag never goes away: it's too dense.

The Sagittarius lead. In 2020, Bonaca and colleagues added high-resolution spectroscopy, clocking how fast the stars move toward and away from us to pin down the perturber's orbit. That single measurement slashed the patch of sky the object could be hiding in — and here's the intriguing bit, the surviving orbits land right on top of the present-day position and motion of debris from the Sagittarius dwarf galaxy. Maybe, the authors suggest, the culprit is a globular cluster or a dark subhalo riding along with the Sagittarius system (Bonaca et al. 2020, ApJ Letters). A promising lead, not a verdict — no specific Sagittarius object has been fingered.

Dark matter that collapsed in on itself. A 2025 study led by Xingyu Zhang and Hai-Bo Yu at UC Riverside found a way to make a dark clump dense enough to do the job. Suppose dark matter particles can bounce off each other — "self-interacting" dark matter. Then a subhalo can tip into a runaway "gravothermal core collapse," cramming its center far tighter than any ordinary CDM halo would. Their N-body simulations show a collapsed object like that could hit the very density GD-1 seems to demand (Zhang et al. 2025, ApJ Letters; arXiv:2409.19493). Elegant. But it leans on a property of dark matter no one has yet shown is real.

A plain old object we just can't see. The cautious bet: maybe the perturber is ordinary matter after all — faint, or already scattered. A very compact star cluster, fully torn apart, whose orbit nobody has managed to trace back. The mass-and-size limits make this a tight squeeze, but the picture keeps sharpening. Fresh data on GD-1's structure could still resurrect a boring answer — or kill it for good.

For now the gap in GD-1 sits as one of astronomy's most maddening near-misses. We can model the thing that made it. We can roughly locate it. We can even weigh it. We just can't find it. The stream remembers the collision perfectly. We're still trying to learn the name of what hit it — and somewhere out there in the halo, if the theorists are right, thousands more of these dark intruders are drifting, unseen, waiting for the next thin stream to give one away.

Sources & Further Reading

• Grillmair & Dionatos (2006), ApJ Letters — original discovery of GD-1: https://iopscience.iop.org/article/10.1088/0004-637X/643/1/L17
• Price-Whelan & Bonaca (2018), ApJ Letters — Gaia reveals the spur, blob, and gaps: https://iopscience.iop.org/article/10.3847/2041-8213/aad7b5
• Bonaca, Hogg, Price-Whelan & Conroy (2019), ApJ — "The Spur and the Gap in GD-1": https://iopscience.iop.org/article/10.3847/1538-4357/ab2873
• Bonaca et al. (2020), ApJ Letters — spectroscopy localizing the perturber near Sagittarius: https://iopscience.iop.org/article/10.3847/2041-8213/ab800c
• Li et al. (2018), ApJ — GD-1 as the relic of an old metal-poor globular cluster: https://ui.adsabs.harvard.edu/abs/2018ApJ...869..122L/abstract
• Zhang, Yu et al. (2025), ApJ Letters / UC Riverside — core-collapsed self-interacting dark matter interpretation: https://www.physics.ucr.edu/news/2025/01/23/study-gd-1-stellar-streams-distinctive-features-caused-self-interacting-dark-matter and preprint https://arxiv.org/abs/2409.19493

Sources & further reading

• Grillmair & Dionatos 2006, ApJ Letters 643, L17 — https://iopscience.iop.org/article/10.1088/0004-637X/643/1/L17
• Price-Whelan & Bonaca 2018, ApJ Letters 863, L20 — https://iopscience.iop.org/article/10.3847/2041-8213/aad7b5
• Bonaca, Hogg, Price-Whelan & Conroy 2019, ApJ 880, 38 — https://iopscience.iop.org/article/10.3847/1538-4357/ab2873
• Bonaca et al. 2020, ApJ Letters 892, L37 — https://iopscience.iop.org/article/10.3847/2041-8213/ab800c
• Li et al. 2018, ApJ 869, 122 — https://ui.adsabs.harvard.edu/abs/2018ApJ...869..122L/abstract
• Zhang, Yu et al. 2025, ApJ Letters 978, L23 (UCR release) — https://www.physics.ucr.edu/news/2025/01/23/study-gd-1-stellar-streams-distinctive-features-caused-self-interacting-dark-matter
• Zhang et al. 2024/2025 preprint, arXiv:2409.19493 — https://arxiv.org/abs/2409.19493