Our Galaxy Is Falling, and No Single Thing Is Pulling
A 2026 study says the Great Attractor is not the dominant pull on our galaxy group. Here is why its 620 km/s fall has no single cause, and why it may never arrive.
By Allan
CosmosIn60CosmosIn60A 2026 study says the Great Attractor is not the dominant pull on our galaxy group. Here is why its 620 km/s fall has no single cause, and why it may never arrive.
The most famous gravitational landmark in our corner of the universe may be less a place than a side effect of how the map was drawn. The Milky Way really is falling: its galaxy group moves at about 620 km/s relative to the afterglow of the Big Bang. But the name that has carried that fall for nearly forty years, the Great Attractor, has lost its job. A study published in February 2026 concludes that it is "not a dynamically dominant structure" and that no single attractor is likely to explain our motion. The tempting headline is a debunking. The stronger claim, and the one this article defends, is that "what is pulling us?" was the wrong question all along: our speed is a running total across every scale of the universe, and a total has no address.
The Local Group, the small family of galaxies that holds the Milky Way and Andromeda, moves at 627 ± 22 km/s relative to the cosmic microwave background, according to COBE's first-year sky maps. Three decades later, the 2026 paper works from about 620 km/s, comfortably inside that old error bar. That is roughly 2.2 million km/h, or 1.4 million mph.
Motion through the afterglow makes one side of the sky slightly warmer and the opposite side slightly cooler, an imbalance COBE measured at 3.365 thousandths of a degree. Nothing in the result names a destination. It is a speedometer reading with no road sign attached.
Why does nobody feel 1.4 million mph? Because this is free fall. All objects fall with the same acceleration, so a galaxy and everything inside it fall together. The motion is gravitational, a peculiar velocity stacked on top of cosmic expansion, and the only way to detect it is to look outward at that ancient light.
The Great Attractor emerged in the late 1980s, when Dressler and colleagues measured elliptical galaxies streaming across the sky and Lynden-Bell's team fitted an attractor model to the pattern, placing it roughly 200 million light-years away. (Distances here convert the 2026 paper's units using its own h = 0.681.)
The direction could hardly have been worse. In that paper's words, "a large portion of the Great Attractor region lies behind the Zone of Avoidance," the band of sky where the Milky Way's own stars and dust get in the way. In 1999, Anthony Fairall called the overdensity "largely hidden by the southern Milky Way" and named the Norma cluster "the most likely candidate" for its unseen core.
Put those together and you have a legend's recipe: an invisible target, a singular name, and real galaxy motions pointing roughly its way. By 2013, an ESA/Hubble caption could say the region "dominates our region of the Universe." That is exactly the claim now under review.
So is the Great Attractor a black hole, or some single hidden object? No. The 2026 paper treats it as "a moderate overdensity embedded within a multiscale flow." Its reconstruction places the convergence point near the cluster Abell 3565 and, "in contrast to previous studies," leaves Norma outside the basin altogether. It also moves that point to about 27 degrees off the Milky Way's plane, against about 9 degrees in the 1988 fit. Much of the region is still buried behind our galaxy's dust, but the drain itself may not be hiding there.
In 2014, Brent Tully's team measured distances and motions for more than 8,000 galaxies and drew our home basin the way a hydrologist draws a watershed. They called it Laniakea, Hawaiian for "immense heaven": about 500 million light-years across, with a predicted 100,000 galaxies. That map cast the Great Attractor as the bottom of the valley.
The 2026 team traced streamlines through 80 "digital twins" of the nearby universe and found that the endpoint depends on how much the velocity field is blurred. Keep detail down to about 10 million light-years and a streamline ends at Virgo. Blur across 10 to 20 million and the classical Great Attractor appears. Blur across roughly 40 million and it runs on toward Shapley. The drains barely move. What changes, in the paper's phrase, is that "the 'mountain ranges' that separate watersheds are sensitive to smoothing." Soften them enough and basins merge.
Anyone who has asked where a raindrop in Pittsburgh ends up knows this shape. The Allegheny, the Ohio, the Mississippi, and the Gulf of Mexico are all correct, at different zoom levels. "Where is the Milky Way going?" behaves the same way. Virgo is the creek, the Great Attractor a river, and Shapley perhaps the sea.
That sea may be larger than the old chart showed. A 2024 Nature Astronomy study built on the motions of 56,000 galaxies put a 60 percent probability on the Milky Way belonging to a basin potentially 10 times Laniakea's volume, centered on the Shapley concentration. The 2026 paper reads that result as only "a slight preference." Our cosmic home address is, for now, close to a 60-40 call, and co-author Ehsan Kourkchi's caveat hangs over it: "our cosmic surveys may not yet be large enough to map the full extent of these immense basins."
If no single thing pulls, what does? The 2026 paper effectively produces an account, set against the observed 620 km/s:
| What gets counted | What it supplies | The catch |
|---|---|---|
| Virgo supercluster alone | about 200 km/s | roughly a third of the total |
| All mass within about 190 million light-years | about half the speed | reaches only to the Great Attractor's own distance |
| Classical Great Attractor basin | 324 ± 51 km/s | direction about 20° off the modeled total |
| All mass within about 740 million light-years | 72 ± 9% of the speed | aimed 38 ± 10° away from the true direction |
Here is the fair objection to this article's thesis: 324 km/s is half the speed. Doesn't that make the Great Attractor dominant after all?
First, a basin is not an object. It is a whole watershed, about 37 quadrillion Suns on the paper's accounting. Second, the pull from that mass does not aim at its own drain. By this article's spherical arithmetic on the paper's coordinates, the two directions sit about 57 degrees apart, with error bars of 10 to 20 degrees. A watershed that tugs you far to one side of its own outlet is not behaving like an attractor in the everyday sense.
The percentage flatters the model too. "72 percent of our speed" sounds like most of the job is done, but velocity is an arrow. An arrow 72 percent as long and tilted 38 degrees leaves a missing arrow of about 380 km/s, by this article's back-of-envelope vector subtraction (rough range 300 to 460). The paper does not report that figure.
Should that alarm anyone? The authors say no: "this incomplete convergence is consistent with expectations," with the rest likely coming from mass beyond the volume and small-scale motions. They also warn that the conclusions rest on these reconstructions, the chosen smoothing, and a flat ΛCDM cosmology.
The deeper reason has the same geometry as Olbers' paradox. Brightness fades with the square of distance while each farther shell holds more stars, so every shell delivers the same light. Gravity obeys the same inverse-square law, so a lopsided distant shell can tug about as hard as a lopsided near one. In the night sky, that arithmetic predicts a blazing sky. Here, it produces a bill that settles only as the universe's lumps thin out on the largest scales.
Some entries are not made of anything. The 2017 Dipole Repeller study argued that "underdense regions push as much as overdensities attract": a near-empty region pulls less than its surroundings, so the net effect resembles a shove. In this accounting, a missing entry counts as a payment.
Will the Great Attractor swallow the Milky Way? No, and not only because it has been demoted. The Norma cluster, its traditional core, has a redshift of 4,844 ± 63 km/s, nearly eight times our fall speed. At that distance, cosmic expansion is still outrunning the fall.
The 2026 team also ran its twins forward until the universe reached ten times its present size, nearly 40 billion years from now by this article's estimate on standard cosmological parameters. Their result: "the Classical Great Attractor does not determine the Local Group's future displacement direction." The displacement points toward Virgo instead, and the trip stays unfinished. In coordinates that factor out expansion, particles near our position travel about a fifth of the way there. "None of these particles become gravitationally bound to the Virgo halo" by that point.
The arrows on a wind map show the wind right now, not the path a balloon released today will follow. Cosmic flow diagrams carry the same trap. They look like routes, but "these lines depict instantaneous velocities" in a universe still rearranging itself. The authors' harshest phrase for the Great Attractor, "an artefact of the instantaneous velocity field," is at heart a complaint about reading a weather map as an itinerary. On these scales, falling does not imply landing.
The stakes reach past one galaxy's heading, because this article's one surviving number is itself under pressure. If motion alone explained the warm and cool sides of the microwave sky, distant matter should show a matching imbalance. A 2025 Reviews of Modern Physics colloquium reports that the matter dipole agrees in direction but is "anomalously larger," at more than 5 sigma. That would not change COBE's reading. It could change what the reading means, and the review itself covers the biases and alternate interpretations still in play.
DESI's peculiar velocity catalogue, published online in July 2026, holds 98,292 early-type galaxies, roughly doubling the low-redshift Fundamental Plane distances measured to date. A May 2026 preprint, submitted for publication, draws on 163 MeerKAT-detected galaxies in the Zone of Avoidance to argue that reconstructions can push through the dust.
Our take: "Great Attractor debunked" would be the wrong headline. The overdensity is real and the region is worth mapping. What it has lost is the job title of sole cause.
Our prediction: the next decade's answer will not be a new name. Expect a budget, so many km/s from each shell and direction, with Shapley, the current favorite, recast as a large line item rather than the explanation. Until then, the honest reply to "where are we falling?" is the one a hydrologist would give: it depends how far out you look.