The Invisible Ocean We're Floating In
Cosmology · Dark Matter · Local Universe
The Invisible Ocean We're Floating In
For nearly a century, astronomers couldn't explain why nearby galaxies flee the Milky Way instead of falling toward it. New simulations finally have the answer — and it's stranger than anyone expected: our galaxy floats inside a vast, flat sheet of dark matter, tens of millions of light-years wide, sandwiched between two enormous cosmic voids.
The universe has a dirty secret. Most of what it contains — roughly 27% of everything that exists — is invisible. It emits no light. It absorbs no light. It passes through ordinary matter like a ghost through a wall. No telescope, no matter how large, has ever directly seen it. And yet we know it is there, because without it, galaxies would fly apart, the cosmic web would not exist, and the universe as we see it simply could not have formed.
We call it dark matter. And now, for the first time, astronomers have mapped where it sits in our own cosmic neighbourhood — and found something extraordinary. The Milky Way is not floating in a roughly uniform sphere of dark matter, as most models assumed. It is embedded in an enormous flat sheet of the stuff, stretching tens of millions of light-years in every direction through a plane, with vast empty voids looming above and below. We do not sit at the centre of a symmetrical halo. We sit at the centre of a cosmic pancake.
The discovery, published in Nature Astronomy in January 2026 by an international team led by PhD graduate Ewoud Wempe and Professor Amina Helmi at the Kapteyn Institute in Groningen, answers a puzzle that has haunted astronomy for nearly a hundred years. It is the first detailed map of dark matter's distribution and motion in the region immediately surrounding the Milky Way and the Andromeda galaxy. And it opens a new window onto the most fundamental and mysterious substance in the universe.
ILLUSTRATION: The Milky Way and Andromeda galaxy (Local Group, centre) embedded in the cosmic dark matter sheet. The sheet extends tens of millions of light-years in the plane; vast empty voids above and below contain almost no matter. Surrounding galaxies drift outward (green arrows) because the gravitational pull of the Local Group is counterbalanced by mass distributed further out in the sheet. Credit: Illustration based on Wempe et al. (2026), Nature Astronomy.
Hubble's Puzzle and a Century of Confusion
The mystery starts with Edwin Hubble. In 1929, the American astronomer published a discovery that reshaped cosmology forever: virtually every galaxy in the observable universe is moving away from us, and the further away it is, the faster it recedes. This was the observational evidence for the expanding universe — the cornerstone of Big Bang cosmology. If everything is flying apart today, everything must have been closer together in the past, converging on a single explosive origin.
But there was a problem in our own backyard. The Milky Way and the Andromeda galaxy are not following the cosmic expansion — they are falling toward each other, on course for a merger in about 4.5 billion years. This makes sense: they are close enough and massive enough that their mutual gravity overcomes the expansion. Together with about fifty smaller galaxies, they form the Local Group — a gravitationally bound cluster immune to the Hubble flow.
The puzzle was what happened just outside the Local Group. Galaxies at slightly greater distances — beyond the gravitational grip of the Milky Way and Andromeda — should be moderately retarded by the Local Group's gravity: moving away, yes, but somewhat slower than the pure Hubble expansion would predict. The Local Group is massive — about 4 to 5 trillion solar masses — so it should exert a meaningful gravitational drag on nearby galaxies.
It didn't. The 31 galaxies in the region just beyond the Local Group were moving away in almost perfect accordance with the pure Hubble flow, as though the Local Group's gravity barely affected them at all. For decades, every model that tried to account for this — by adjusting the Local Group's mass, by tweaking the expansion rate, by modifying dark matter distributions — failed to reproduce what was observed. Something fundamental was missing.
Building a Virtual Twin of the Universe
Ewoud Wempe, a doctoral student at the Kapteyn Astronomical Institute, took a different approach from all previous attempts. Instead of starting with assumptions about the mass distribution and checking if they fit observations, he asked a much more open question: what initial conditions in the early universe could have produced exactly the cosmic neighbourhood we see today?
The team used a method called BORG — Bayesian Origin Reconstruction from Galaxies — combined with a powerful cosmological simulation code, to work backward from today's observations to the early universe. They seeded the simulation with tiny quantum fluctuations matching the oldest light we can detect: the Cosmic Microwave Background (CMB), the thermal afterglow of the Big Bang, observed with extraordinary precision by the Planck satellite. From those early fluctuations, gravity acted for 13.8 billion years, producing the structures we see today.
But the simulation had to satisfy multiple simultaneous constraints: reproduce the mass, position, and velocity of the Milky Way and Andromeda; reproduce the positions and velocities of all 31 galaxies just outside the Local Group; and be consistent with the broader Lambda Cold Dark Matter (ΛCDM) model — the standard cosmological framework. The algorithm searched through every possible configuration of the early universe that could satisfy all these requirements simultaneously.
"We are exploring all possible local configurations of the early universe that ultimately could lead to the Local Group. It is great that we now have a model that is consistent with the current cosmological model on one hand and with the dynamics of our local environment on the other."
— Ewoud Wempe, lead author, Kapteyn Institute, University of GroningenThe simulation found hundreds of valid configurations — 169 distinct "virtual twins" of the Local Group. When the team looked at what all these successful solutions had in common, the answer was striking and unexpected: in every single case, the mass distribution around the Local Group was dramatically flattened. Not a sphere. Not a rough ellipsoid. A flat sheet, extending tens of millions of light-years through a plane, with enormous near-empty voids looming above and below.
The Sheet and the Voids: Why Galaxies Flee
The discovery of the flat sheet immediately explained the century-old puzzle — with elegant, almost geometric simplicity.
Consider a galaxy sitting in the sheet, some distance from the Local Group. Gravitationally, it is pulled toward the Local Group by the mass of the Milky Way and Andromeda. But it is also pulled away from the Local Group by all the mass in the sheet further along the plane — the vast dark matter distributed throughout the sheet beyond the galaxy's position. These two gravitational pulls largely cancel. The galaxy is left almost free to follow the Hubble flow unimpeded.
The voids provide the second piece of the explanation. Above and below the sheet, where you might expect matter to be falling inward toward the Local Group — adding to its gravitational influence — the voids are nearly empty. There are almost no galaxies there to be pulled in. The Local Group's gravity has nothing to grip. The infall from above and below that would normally be detectable in the motions of nearby galaxies simply does not occur, because there is almost no matter there to fall.
CROSS-SECTION: Why galaxies in the sheet drift away despite the Local Group's gravity. In-plane galaxies are pulled toward the Local Group (red) but equally pulled outward by mass further along the sheet (blue) — the two cancel, leaving galaxies to follow the Hubble expansion freely. Infall from the voids above and below (which would add to the Local Group's gravitational influence) does not occur because the voids are nearly empty. Credit: Author's illustration based on Wempe et al. (2026).
This two-part mechanism — gravitational cancellation within the sheet, and the absence of infall from the voids — produces exactly the calm, orderly Hubble flow that observers see. The simulations reproduced the positions and velocities of all 31 surrounding galaxies with remarkable accuracy. After nearly a century of failed attempts, the local Hubble flow is finally explained.
The bonus prediction: The simulations predicted something the researchers had not explicitly asked for — the flat distribution of galaxies in the Local Supercluster (the broader region containing thousands of galaxies on scales of hundreds of millions of light-years). This large-scale flattening is a known observational fact. The model predicted it automatically, without being told it existed, as an emergent consequence of the dark matter sheet structure. This independent confirmation strengthens confidence that the model describes real physical structure rather than a statistical artefact.
DARK MATTER: The Complete Guide to the Universe's Greatest Mystery
The discovery of the Milky Way's dark matter sheet is, at its core, a story about dark matter — the invisible scaffolding on which the visible universe is built. Before we can fully appreciate what Wempe and Helmi have found, we need to understand what dark matter is, why we believe it exists, what it might be made of, and why it remains, despite decades of searching, the most elusive substance in the cosmos.
What Is Dark Matter? The Basic Picture
Dark matter is matter that does not emit, absorb, or scatter electromagnetic radiation of any wavelength — not light, not radio waves, not X-rays, not gamma rays. It is invisible to every kind of telescope we have ever built. And yet, by our best current measurements, it makes up approximately 27% of the total energy-matter content of the universe. Ordinary matter — the atoms, molecules, stars, gas, and dust that make up everything we can see — constitutes only about 5%. Dark energy, the mysterious force driving the universe's accelerating expansion, makes up the remaining 68%.
In other words: everything you can see, touch, measure directly, or make a telescope to observe — every star, every galaxy, every nebula, every planet, every atom — is only 5% of what exists. We are made of cosmic minority material. The rest is dark.
Ordinary (baryonic) matter: ~5%
Dark matter: ~27%
Dark energy: ~68%
Source: Planck satellite CMB measurements (2018)
How Do We Know It Exists?
We have never directly detected a dark matter particle. Every piece of evidence for dark matter is gravitational — we see the effects of its gravity on the things we can see. The evidence comes from multiple independent directions, and it is overwhelming.
Galaxy rotation curves were the first convincing evidence. In the 1970s, astronomer Vera Rubin and her colleague Kent Ford made careful measurements of how fast stars in spiral galaxies orbit the galactic centre. Newton's laws predict that stars far from the bright core — where most of the visible mass is concentrated — should orbit more slowly, just as Neptune orbits the Sun more slowly than Mercury does. Instead, Rubin found that rotation speeds in the outer regions of galaxies remained roughly constant, or even increased slightly, far beyond where the visible matter ended. The only explanation was that galaxies are embedded in vast, spherical halos of invisible mass — dark matter halos that extend far beyond the visible disc of stars and provide the extra gravitational pull needed to keep the outer stars moving so fast.
Gravitational lensing provides a completely independent confirmation. Einstein's general relativity predicts that any mass — visible or invisible — will bend the path of light passing near it. By observing how light from distant galaxies is distorted and bent by galaxy clusters in the foreground, astronomers can map the mass distribution of the cluster, including the invisible component. In every case, the total mass inferred from lensing far exceeds the visible mass. Clusters of galaxies contain roughly five to ten times as much dark matter as ordinary matter.
The Bullet Cluster provided perhaps the most direct evidence. When two galaxy clusters collided, the ordinary matter (hot gas, visible in X-rays) was slowed and piled up in the middle by electromagnetic interactions. But the dark matter halos passed through each other almost unimpeded — they are collisionless, interacting only gravitationally — and separated from the gas. Gravitational lensing maps showed the mass concentrated in two separated clumps corresponding to the dark matter halos, while the X-ray gas sat between them. Dark matter and ordinary matter had been physically separated and photographed in their separated state.
The Cosmic Microwave Background provides the most precise measurements. The pattern of tiny temperature fluctuations in the CMB — the oldest light in the universe — is exquisitely sensitive to the proportions of ordinary matter, dark matter, and dark energy in the early universe. The measurements from the Planck satellite fit a cosmological model with ~27% dark matter with extraordinary precision, completely independently of any galaxy observation.
Large-scale structure provides the final line of evidence. The universe is not smoothly distributed. It is organised into a vast cosmic web of filaments and sheets of galaxies, with enormous voids between them — the very voids detected in the new Groningen study. This structure could not have formed from the tiny fluctuations in the early universe without dark matter acting as a gravitational seed, pulling ordinary matter into its potential wells and enabling the first galaxies and clusters to form. Without dark matter, the universe would be far smoother, with far less structure, and galaxies like the Milky Way might not exist at all.
What Might Dark Matter Be Made Of?
This is the deepest unsolved question in cosmology — and arguably in all of physics. We know dark matter exists. We know roughly how much there is. We know how it is distributed (imperfectly). But we do not know what it is. Dozens of candidate particles and objects have been proposed over the decades, ranging from the theoretically elegant to the exotic. Here is the current landscape.
| Candidate | What It Is | Key Properties | Status (2026) |
|---|---|---|---|
| WIMPs (Weakly Interacting Massive Particles) | Hypothetical particles ~10–1000× proton mass, interacting via weak nuclear force | Predicted by supersymmetry; "WIMP miracle" gives correct relic density naturally | ◆ Constrained but not ruled out — strong limits from LUX-ZEPLIN, XENON-nT |
| Axions | Ultra-light particles originally proposed to solve the strong CP problem in QCD | Extremely light (~10⁻⁵ eV); form a coherent quantum field rather than individual particles | ★ Leading candidate — ADMX, CASPEr experiments actively searching |
| Sterile Neutrinos | Hypothetical heavier cousins of ordinary neutrinos that don't interact via weak force | Right-handed; mass range keV to GeV; could explain neutrino mass via seesaw mechanism | ◆ Possible — X-ray line at 3.5 keV seen in galaxy clusters remains unexplained |
| Primordial Black Holes (PBHs) | Black holes formed in the very early universe before any stars existed | Could span wide mass range; no Standard Model extension required | ◆ Constrained — micro-lensing and gravitational wave observations limit mass windows |
| MACHOs (Massive Compact Halo Objects) | Brown dwarfs, white dwarfs, neutron stars — compact objects in galaxy halos | Ordinary matter; detectable via microlensing surveys | ✗ Largely ruled out — cannot account for most dark matter |
| Fuzzy Dark Matter | Ultra-ultra-light axion-like particles (~10⁻²² eV) with huge de Broglie wavelength | Quantum wave nature visible on galactic scales; suppresses small-scale structure | ◆ Active research area — may solve small-scale structure problems |
| Kaluza-Klein particles | Extra-dimensional particles predicted by string theory / extra-dimensional models | Properties similar to WIMPs; arise naturally in theories with compact extra dimensions | ◆ Theoretical — no dedicated searches yet |
| Mirror Matter | A complete copy of the Standard Model with opposite parity, interacting with itself but not ordinary matter | Entire shadow universe; would form "mirror stars" invisible to us | ◆ Speculative but self-consistent — hard to rule out directly |
How Dark Matter Structures the Universe: The Cosmic Web
Perhaps the most important thing about dark matter is not what it is, but what it does. Dark matter is the scaffolding of the universe. It is the invisible architecture on which all visible structure — every galaxy, every cluster, every filament — was built.
In the very early universe, matter was distributed almost uniformly. Tiny quantum fluctuations — quantum uncertainty made manifest at the moment of inflation — created slightly denser and slightly less dense regions. In ordinary matter alone, these fluctuations would have been smoothed out by radiation pressure before they could grow. But dark matter — which does not interact with radiation — began collapsing under its own gravity long before ordinary matter could. It formed the first structures: dark matter halos.
Ordinary matter, released from radiation pressure when the universe cooled sufficiently (about 380,000 years after the Big Bang), fell into these pre-formed dark matter potential wells. Stars formed at their centres. Galaxies assembled around those stars. Galaxies grouped into clusters along dark matter filaments. The result — visible in the largest surveys of the universe — is the cosmic web: a vast three-dimensional network of filaments and sheets of galaxies surrounding enormous empty voids. The structure the Groningen team found around the Milky Way — the cosmic dark matter sheet with voids above and below — is a local, intimate manifestation of this same process that has been operating across the entire universe for 13.8 billion years.
The standard cosmological model, Lambda Cold Dark Matter (ΛCDM), posits that dark matter consists of "cold" (non-relativistic, slow-moving) particles that clump under gravity to form halos and filaments. The "cold" property is essential: "hot" dark matter (fast-moving, like neutrinos) would stream out of overdense regions before structures could form, producing a universe too smooth to match observations. Every major cosmological observation — the CMB, galaxy surveys, baryon acoustic oscillations, gravitational lensing maps — is consistent with ΛCDM. Despite two decades of effort to find alternatives, no rival model matches the data as well.
The Direct Detection Challenge
If dark matter fills the galaxy — if we are swimming through a sea of it at every moment — why haven't we detected it directly? The answer is that dark matter's defining property is precisely its refusal to interact with ordinary matter through any force except gravity. Gravity is extraordinarily weak at the particle level. Detecting the rare interaction between a dark matter particle and a detector nucleus requires extraordinary technology and extreme isolation from any background radiation.
Direct detection experiments operate in deep underground laboratories to shield them from cosmic rays. The detectors — typically tonnes of ultra-pure liquid xenon, germanium crystals, or similar materials — are cooled to near absolute zero and monitored for any vibration caused by a dark matter particle collision. The LUX-ZEPLIN experiment in the Homestake Mine in South Dakota, and the XENON-nT experiment in Italy's Gran Sasso tunnel, have set the most stringent limits on WIMP interactions, finding no convincing signal. This has progressively constrained the viable WIMP parameter space — which does not rule WIMPs out, but pushes the allowed region into ranges harder and harder to test.
The axion search is pursued differently. Axions, if they exist, can be converted into photons in a strong magnetic field. The ADMX experiment at the University of Washington uses a superconducting magnet and microwave cavity to detect this conversion, scanning through a range of axion masses. The sensitivity is now entering the region where theoretically preferred axions should be found, if they exist at the predicted mass. No signal yet — but the search is now probing theoretically motivated territory for the first time.
Could We Be Wrong About Dark Matter?
A minority of physicists argue that dark matter does not exist, and that the observations attributed to it can be explained by modifying the laws of gravity. The leading alternative, MOND (Modified Newtonian Dynamics), proposed by Moti Milgrom in 1983, replaces Newton's law of gravity with one that transitions to a different regime at very low accelerations. MOND successfully predicts galaxy rotation curves from visible mass alone — without any dark matter — and has had several striking observational confirmations.
However, MOND fails badly at scales above individual galaxies. It cannot explain the CMB power spectrum. It cannot explain cluster lensing. It particularly struggles with the Bullet Cluster, where dark matter and ordinary matter are spatially separated. Various relativistic extensions of MOND (TeVeS, AQUAL, etc.) have been proposed to address these failures, but none has matched the overall success of ΛCDM. The discovery of the dark matter sheet around the Milky Way — which explains the local Hubble flow within the standard ΛCDM framework, where MOND cannot — adds to the weight of evidence against modified gravity alternatives.
What the Simulation Actually Did: The Technical Story
Understanding the full significance of Wempe and Helmi's result requires appreciating what the simulation actually accomplished. Previous attempts to model the Local Group's environment had started with assumed mass distributions — spherical halos, uniform fields, various prescriptions — and checked whether they reproduced observations. All failed. The new approach was fundamentally different.
The BORG algorithm is a form of constrained simulation: it does not assume a mass distribution and check it, but searches the entire vast space of possible initial conditions to find those that produce the observed present-day universe. Starting from CMB fluctuations, the algorithm evolves the universe forward in time using gravitational physics, checks the result against the observed positions and velocities of 31 galaxies, and iteratively refines the initial conditions until the match is as good as possible. The process was run 169 times with different random seeds, producing 169 independent "virtual twins" of the Local Group.
The team produced 169 distinct realizations of a Local Group analogue using parameters drawn from cosmic microwave background data and the motions of 31 nearby galaxies. Results revealed a sheet-like dark matter structure extending beyond 10 megaparsecs, or roughly 30 million light-years. Its central plane contains a density roughly twice the cosmic average, with almost empty voids above and below.
The power of running 169 simulations is statistical: any feature that appears in all or most of them is a robust prediction, not a coincidence of a single run. The flat sheet appeared in every successful simulation. It was not a quirk of one particular initial condition — it was the only type of environment consistent with everything we observe about our cosmic neighbourhood.
1. Gravitational cancellation within the sheet: For galaxies in the plane of the sheet, the Local Group's inward gravitational pull is counteracted by mass distributed further out in the same plane. The pulls nearly cancel, leaving the galaxy free to follow the Hubble expansion.
2. Absence of infall from the voids: Above and below the sheet, enormous empty voids contain almost no galaxies. The infall of matter toward the Local Group from these directions — which would normally add to its gravitational braking — simply does not occur because there is almost nothing there to fall inward.
The First Dark Matter Map of Our Neighbourhood
Beyond explaining the Hubble flow puzzle, the study represents something historically significant: the first assessment of the distribution and velocity of dark matter in the region surrounding the Milky Way and the Andromeda Galaxy.
For the first time, we have a quantitative picture of where the dark matter is in our cosmic neighbourhood, how it is moving, how dense it is in different locations, and how it connects to the broader large-scale structure of the universe. The sheet's central plane has a dark matter density roughly twice the cosmic average. The voids flanking it have densities perhaps 5–10% of the cosmic average. The transition between sheet and void is steep — a dramatic change in density over a relatively short distance.
This matters because it gives us a map. Not a map of visible galaxies — we have had those for decades — but a map of the invisible 85% of matter that dominates the gravitational landscape. Future observations can test this map: the study anticipates strong inflows from above and below the dark matter sheet, with peculiar velocities exceeding 100 kilometres per second. If we find isolated dwarf galaxies at high latitudes above or below the sheet, moving inward at these velocities, the model will be confirmed from a completely independent direction.
"I am excited to see that, based purely on the motions of galaxies, we can determine a mass distribution that corresponds to the positions of galaxies within and just outside the Local Group."
— Professor Amina Helmi, co-author, Kapteyn Institute, University of GroningenThe Discovery Timeline
Edwin Hubble publishes the law of cosmic expansion: virtually all galaxies are receding from the Milky Way. The local Hubble flow puzzle is born — why does the Local Group not slow down nearby galaxies?
Vera Rubin and Kent Ford demonstrate flat galaxy rotation curves, providing compelling evidence for dark matter halos around galaxies. The scale of dark matter's dominance becomes clear.
Large-scale galaxy surveys begin mapping the cosmic web — the filaments, sheets, and voids of the universe on scales of hundreds of millions of light-years. The void structure flanking the Local Group (the "Local Void") is catalogued but its relationship to local galaxy motions remains unexplained.
The Bullet Cluster observation provides what many call the "smoking gun" for dark matter's existence — photographing dark matter halos separated from ordinary matter after a galactic collision.
The Planck satellite releases its final CMB maps, precisely measuring dark matter's cosmic abundance at 26.8% of the total energy-matter content. The ΛCDM cosmological model is confirmed to extraordinary precision.
Wempe, Helmi, and collaborators from Germany, France, and Sweden run 169 constrained cosmological simulations using the BORG algorithm, mapping every possible local dark matter configuration consistent with observations.
The paper "The mass distribution in and around the Local Group" is published in Nature Astronomy. The cosmic dark matter sheet is confirmed. The local Hubble flow puzzle, nearly a century old, is solved.
Follow-up observations search for dwarf galaxies at high latitudes above and below the sheet (the predicted infall zones). Next-generation surveys — the Rubin Observatory, Euclid satellite — will extend the dark matter map to greater distances and higher precision.
What This Means: The Bigger Picture
The Groningen discovery is important on several levels. Most immediately, it solves a concrete, quantitative puzzle in observational astronomy that has resisted solution for nearly a century. That alone would make it a significant result. But it also does something more profound: it demonstrates that the local universe — our immediate cosmic neighbourhood, the region we know best — is not a uniform, well-mixed environment but a structured, geometrically organised place, shaped by the invisible dark matter whose distribution we are only now beginning to map directly.
The idea that dark matter organises into sheets and filaments fits with the broader picture of the cosmic web, the large-scale structure of the universe. Simulations show matter collapsing along preferred directions, forming flattened regions and elongated strands over immense distances. What Wempe and Helmi have found is not an anomaly but a local instance of a universal process — the same dark matter architecture that shapes clusters and superclusters of galaxies on the largest scales is operating, in a more intimate way, in the tens of millions of light-years immediately surrounding us.
We have known for some time that we live in a cosmic web. Now we know precisely where, in that web, we sit: at the centre of a flat sheet of dark matter, pressed between two vast silences.
What comes next: The model predicts that galaxies above and below the sheet — in the voids — should show strong inward peculiar velocities exceeding 100 km/s as they fall toward the sheet's gravitational potential. No such galaxies have yet been systematically observed at high supergalactic latitudes within 5 megaparsecs. The Rubin Observatory's Legacy Survey of Space and Time (LSST), which began full operations in 2025, will map faint dwarf galaxies across the entire sky to unprecedented depth. If Wempe and Helmi's dark matter sheet is real — and the simulations are very strong evidence that it is — the void dwarfs should be found, falling in, exactly as predicted.
Comments
Post a Comment