What If Dark Matter Doesn't Exist? Modified Gravity vs Invisible Mass

For decades, astronomers have detected mysterious gravitational effects: galaxies rotate too fast, clusters cohere too strongly, light bends around them more than visible mass should explain. The standard answer is that there's invisible "dark matter" providing extra gravity. The unstated alternative is that maybe gravity itself works differently than we think at the largest scales.

Thought experiment. This article explores a hypothetical scenario for educational and speculative purposes. The events described are not predictions and should not be interpreted as warnings, advocacy, or instructions.

Modified gravity theories — sometimes grouped under the umbrella term MOND (Modified Newtonian Dynamics) or its relativistic descendants — try to explain galaxy behavior without invoking new particles. The mainstream view still favors dark matter, but the question is genuinely open among working physicists, not settled the way, say, the shape of the Earth is settled. Here's the case for and against, drawn from the actual research literature rather than pop-science shorthand.

Key takeaways

• "Dark matter" is inferred entirely from gravitational effects; no dark matter particle has ever been directly detected in a laboratory.
• Fritz Zwicky first proposed unseen mass in 1933 studying the Coma Cluster; Vera Rubin's rotation curve measurements in the 1970s made the case hard to dismiss.
• MOND, proposed by Mordehai Milgrom in 1983, modifies Newton's second law at very low accelerations and predicts individual galaxy rotation curves with a single universal constant.
• The 2006 Bullet Cluster observation is the single hardest piece of evidence for modified gravity to explain, since it shows gravitational mass spatially separated from visible mass.
• Direct-detection experiments like LZ and XENONnT have ruled out large swaths of the parameter space for WIMP dark matter without finding a particle.
• A 2023–2024 study of Gaia wide binary stars by Kyu-Hyun Chae reported a gravitational anomaly at low acceleration consistent with MOND's predicted "external field effect" — a genuinely new and still-debated line of evidence.

What dark matter explains

Before evaluating alternatives, it helps to understand exactly what problem physicists were trying to solve when they first invoked invisible mass. The case for dark matter isn't a single observation — it's a stack of independent lines of evidence that all point toward roughly the same conclusion.

The Coma Cluster and Fritz Zwicky (1933)

Swiss astronomer Fritz Zwicky studied the Coma Cluster, a collection of roughly a thousand galaxies, and found that the galaxies were moving so fast relative to each other that the cluster's visible mass could not gravitationally hold it together. Zwicky calculated the cluster needed to be hundreds of times denser than its luminous matter suggested, and coined the term "dunkle Materie" — dark matter — for the missing mass. His idea was largely ignored for nearly four decades because the measurement techniques of the time weren't precise enough to convince skeptics.

Vera Rubin and galaxy rotation curves (1970s)

Vera Rubin and Kent Ford measured how fast stars orbit at different distances from the centers of spiral galaxies. Newtonian physics predicts that stars far from a galaxy's center, where there's less visible mass, should orbit more slowly, the same way outer planets orbit the Sun more slowly than Mercury does. Instead, Rubin and Ford found that rotation curves stay flat: stars at the edge of a galaxy orbit about as fast as stars closer in. That flatness is only possible if there's a large amount of extra mass distributed in a halo well beyond the visible disk.

Galaxy cluster dynamics

The same basic problem Zwicky found in Coma shows up in essentially every galaxy cluster ever measured in detail. Galaxies within a cluster move fast enough that the cluster's visible mass, on its own, would not be sufficient to keep it gravitationally bound over cosmic timescales.

Gravitational lensing

Massive objects bend light passing near them, a direct prediction of general relativity confirmed as far back as the 1919 solar eclipse expedition. Galaxy clusters bend background light more than their visible mass should produce, and the amount of bending maps cleanly onto the same "extra mass" inferred from rotation curves and cluster dynamics.

Cosmic microwave background

The cosmic microwave background (CMB) is the relic radiation left over from about 380,000 years after the Big Bang, and its tiny temperature fluctuations encode the universe's composition. The European Space Agency's Planck satellite measured these fluctuations in exquisite detail, and the pattern of peaks in the CMB power spectrum matches a model with roughly 26.8% dark matter, 4.9% ordinary matter, and 68.2% dark energy — the standard Lambda-CDM cosmology.

Large-scale structure

Computer simulations that seed the early universe with cold dark matter produce a cosmic web of filaments and voids that closely matches what galaxy surveys actually observe. Simulations run with ordinary matter alone, and no dark matter, don't reproduce that structure — gravity from visible matter simply doesn't clump fast enough after the Big Bang to build the web we see today.

In August 2006, astronomers observed the Bullet Cluster — two galaxy clusters that collided roughly 150 million years ago — and found, via gravitational lensing at 8-sigma statistical significance, that most of the gravitational mass was spatially offset from the hot, visible gas. It's the closest thing cosmology has to smoking-gun evidence for particle dark matter, and it remains the hardest single observation for modified gravity theories to explain.

The dark matter explanation ties all five of these observations together with one concept: there's invisible, non-luminous mass, roughly five times more abundant than ordinary matter, that interacts through gravity but not (or barely) through light or the other known forces.

The alternative: maybe gravity is wrong

Every one of the observations above assumes that Newtonian gravity and general relativity work the same way at galactic and cosmic scales as they do in the solar system, where they've been tested to extraordinary precision. Modified gravity theories challenge that assumption specifically at the very low accelerations found at galaxy edges, an acceleration regime we've never been able to test directly on Earth or within the solar system.

MOND (Modified Newtonian Dynamics)

Israeli physicist Mordehai Milgrom proposed MOND in 1983. Instead of adding invisible mass, MOND modifies the relationship between force and acceleration at accelerations below a specific threshold, denoted a₀ and measured at approximately 1.2 × 10⁻¹⁰ meters per second squared. Above that threshold — anywhere in the solar system, on Earth, in ordinary lab experiments — gravity behaves exactly as Newton and Einstein described. Below it, in the outskirts of galaxies where accelerations are tiny, gravity is effectively boosted, producing flat rotation curves without any extra mass.

Think of it like a speedometer that reads accurately at highway speeds but starts under-reporting once you slow to a crawl. MOND says our gravitational "speedometer" — Newton's law — might be miscalibrated at the gravitational equivalent of a crawl, and nobody had ever built an instrument sensitive enough to notice until galaxy rotation curves came along.

Why MOND was taken seriously

MOND's headline success is the radial acceleration relation, published by Stacy McGaugh, Federico Lelli, and James Schombert in a 2016 Physical Review Letters paper. They compared 2,693 individual points across 153 galaxies of wildly different sizes, masses, and gas content, and found a tight, near-universal relationship between the acceleration a galaxy's visible matter should produce and the acceleration actually observed. The scatter in that relationship is small enough that McGaugh's team called it "a natural law for rotating galaxies." MOND predicts this relationship directly, with one constant, from first principles. Standard dark matter models can reproduce it too, but only after tuning the dark matter halo profile of each individual galaxy — the relationship isn't an automatic prediction of dark matter the way it is of MOND.

Why MOND is controversial

MOND, in its original 1983 form, is non-relativistic — it doesn't fit inside Einstein's general relativity, the framework that correctly predicts everything from GPS satellite timing to gravitational waves. It's also, in a sense, empirical rather than explanatory: it describes how gravity behaves at low acceleration without offering a physical mechanism for why it should behave that way. That's a real scientific liability. A theory that fits the data without explaining the underlying physics invites the criticism that it's just a curve-fit dressed up as a law.

TeVeS (Tensor-Vector-Scalar gravity)

Physicist Jacob Bekenstein addressed MOND's relativistic problem in 2004 by proposing TeVeS, a covariant (relativity-compatible) formulation that reproduces MOND's galactic predictions while also handling gravitational lensing, which the original non-relativistic MOND couldn't do at all. TeVeS adds extra vector and scalar fields on top of Einstein's tensor field, at the cost of significant extra theoretical complexity.

Emergent gravity (Verlinde)

Dutch theoretical physicist Erik Verlinde proposed a different route in the 2010s: that gravity isn't a fundamental force at all but an emergent phenomenon arising from the thermodynamics and entropy of quantum information, similar to how "temperature" emerges from the average behavior of countless individual molecules rather than being a fundamental property of any single one. Verlinde's framework has reproduced some galaxy rotation curve data without invoking dark matter particles, though it remains far less developed and far less tested than either MOND or standard cold dark matter.

Other modified gravity frameworks

Beyond MOND, TeVeS, and emergent gravity, physicists have proposed f(R) gravity (which modifies the mathematical function governing spacetime curvature in Einstein's field equations), scalar-tensor theories, and conformal gravity, among others. Each attempts a different mathematical route to the same goal: match cosmological observations without a new particle.

Where modified gravity succeeds

Individual galaxy rotation curves

This is MOND's strongest ground. Across galaxies ranging from tiny dwarf spheroidals to giant spirals, a single acceleration constant reproduces observed rotation curves with an accuracy that dark matter models, which require fitting a separate halo profile per galaxy, generally can't match without extra tuning.

The Tully-Fisher relation

Astronomers R. Brent Tully and J. Richard Fisher discovered in 1977 that a spiral galaxy's total luminosity correlates tightly with its rotation speed raised to roughly the fourth power. MOND predicts this relationship as a direct mathematical consequence of its acceleration law. In the standard dark matter picture, the Tully-Fisher relation has to be added as an empirical observation rather than derived from the theory's basic assumptions.

Consistency across a wide mass range

MOND uses the same acceleration constant, a₀, for galaxies that differ in mass by factors of a million or more. That kind of scale-independence with a single free parameter is unusual for any physical theory and is one reason MOND continues to draw serious research attention four decades after Milgrom first proposed it.

The Gaia wide binary star anomaly

The newest and arguably most interesting evidence for modified gravity doesn't come from galaxies at all. In 2023, Sejong University physicist Kyu-Hyun Chae analyzed roughly 26,500 wide binary star systems — pairs of stars orbiting each other at large separations — using precision position data from the European Space Agency's Gaia spacecraft. He found that when the gravitational acceleration between the two stars drops below about one nanometer per second squared, their orbital dynamics deviate from Newtonian predictions by a factor of roughly 1.4, matching the boost predicted by a MOND-type theory called AQUAL, including MOND's specific prediction of an "external field effect" from the Milky Way's own gravity. Follow-up analyses published through 2024 and 2025 have largely confirmed the anomaly, though the interpretation — whether it's truly new physics, an artifact of unresolved multiple-star systems in the catalog, or a data quality issue — remains actively debated among specialists.

Where modified gravity struggles

Galaxy clusters

MOND performs dramatically worse at cluster scales than at galaxy scales. Even applying Milgrom's low-acceleration boost to a cluster's visible mass leaves a substantial gravitational shortfall — clusters appear to need roughly twice the mass MOND alone can supply, whatever that missing mass turns out to be.

The Bullet Cluster

The Bullet Cluster is widely cited as the toughest single observation for modified gravity. When the two clusters collided, the hot intracluster gas — which makes up most of a cluster's ordinary, visible mass — collided, slowed down, and heated up, the same way two clouds of sand would grind against each other and pile up in the middle. The gravitational mass, mapped by Clowe et al. through weak lensing in a 2006 study, kept moving largely undisturbed and separated cleanly from the gas. A pure gravity modification, tied directly to visible mass, has no obvious way to produce lensing centered on empty space where the modification's own source (the gas) isn't located. Dark matter explains the offset naturally: it passed through the collision as a nearly collisionless cloud while the gas did not.

Cosmic microwave background fine structure

The detailed pattern of acoustic peaks in the CMB power spectrum, the same data set that gives the 26.8% dark matter figure from Planck, is difficult for pure modified gravity theories to reproduce without effectively smuggling back in something that behaves like dark matter in the early universe. TeVeS and its successors have struggled on this front more than on galactic rotation curves.

Large-scale cosmic structure

Cold dark matter simulations correctly predict the observed distribution and clustering timescale of galaxies across billions of light-years. Modified gravity, applied on its own without any additional invisible mass component, generally can't grow structure fast enough after the Big Bang to match what large galaxy surveys actually see.

What the evidence says, on balance

Chandra X-ray Observatory scientists described the Bullet Cluster in a 2006 press release as "direct proof" of dark matter — a strong claim, and one that modified-gravity researchers dispute, but it illustrates how seriously mainstream astrophysics treats that single observation relative to everything else in the debate.

Dark matter wins more tests

Across the full range of observational tests — clusters, CMB structure, large-scale galaxy distribution, the Bullet Cluster — the standard Lambda-CDM cosmology, built on cold dark matter plus dark energy, fits the data better and more consistently than any current modified gravity theory. That's the main reason it remains the majority position among cosmologists.

Modified gravity wins on individual galaxies

For rotation curves of specific galaxies and for the radial acceleration relation, MOND-based approaches often outperform dark matter models that haven't been separately tuned for each galaxy. This asymmetry — modified gravity strong at galaxy scale, weak at cluster and cosmic scale — is itself one of the more interesting open puzzles in the field.

Hybrid approaches

Some researchers have proposed combining the two: a small amount of dark matter (sometimes hot or sterile-neutrino dark matter rather than the standard cold variety) plus a MOND-like modification to gravity. These hybrid models add complexity and free parameters, and critics note that a theory flexible enough to fit everything risks explaining nothing. But they remain an active area of published research rather than a fringe idea.

The state of play is genuinely unsettled

Neither side has a clean, unambiguous win across every scale. Most cosmologists favor dark matter because it fits more of the evidence with fewer scale-dependent adjustments; a meaningful minority of physicists, including MOND's original architect Milgrom, continue publishing peer-reviewed work defending modified gravity as the better long-run explanation.

Why most physicists still prefer dark matter

Several reasons go beyond simple observational fit.

Consideration	Favors dark matter	Favors modified gravity
Theoretical foundation	Extends known particle physics (adds a new particle type)	Requires modifying general relativity, one of the best-tested theories in science
Explanatory unity	One new entity explains galaxies, clusters, CMB, and large-scale structure	Different acceleration/tuning behavior needed at galaxy vs. cluster scale
Testability	Direct detection experiments (LZ, XENONnT) can in principle find or rule out specific particle candidates	Hard to test independently of the astrophysical systems it's meant to explain
Bullet Cluster	Naturally explained by collisionless mass passing through the collision	No agreed mechanism for lensing centered away from visible mass
Galaxy rotation curves	Requires per-galaxy halo tuning	Single constant fits most galaxies well
CMB power spectrum	Matches Planck data precisely with cold dark matter	Struggles without adding dark-matter-like terms

Table: Dark matter vs. modified gravity — where each argument is strongest, based on peer-reviewed cosmological and astrophysical evidence through 2025.

Conservatism has scientific value

General relativity has survived over a century of increasingly precise tests, from Mercury's orbit to the 2015 direct detection of gravitational waves by LIGO. Modifying a theory with that track record requires unusually strong, unambiguous evidence, and physicists are — correctly, in the view of most working cosmologists — reluctant to abandon it on the strength of galaxy-scale anomalies alone.

Particle physics candidates already exist

Independent theoretical frameworks developed to solve entirely different problems — supersymmetry (proposed to address the mathematical stability of the Higgs boson mass) and the axion (proposed to solve the "strong CP problem" in nuclear physics) — both predict particles with roughly the right properties to be dark matter. That's a point in dark matter's favor: it isn't a theory invented just to explain rotation curves; it connects to independently motivated physics.

Direct detection is, in principle, decisive

If a dark matter particle exists and interacts even weakly with ordinary matter, deep-underground detectors like LZ (LUX-ZEPLIN, in South Dakota) and XENONnT (in Italy) could eventually catch one colliding with an atomic nucleus. LZ's 2024 results, based on 280 days of data, set the tightest limits yet on WIMP (weakly interacting massive particle) dark matter without detecting one; XENONnT's 2025 analysis, covering a 3.1 tonne-year exposure, pushed those limits further and, notably, became sensitive enough to detect the faint background signal from solar neutrinos — a milestone physicists call reaching the "neutrino floor." No experiment has found a dark matter particle. Modified gravity has no equivalent independent laboratory test.

What if dark matter really doesn't exist?

If decades of underground and collider searches keep coming up empty, several outcomes remain on the table.

Modified gravity turns out to be correct

Physicists would need a mathematically complete, relativistic theory that matches MOND's galaxy-scale success while also explaining clusters, the Bullet Cluster, and the CMB power spectrum without invoking anything dark-matter-like. Nobody has built that theory yet.

Dark matter exists but is far harder to detect than assumed

Some candidates, like the QCD axion, interact so weakly with ordinary matter that current detector technology may simply be inadequate, not wrong in principle. New detector designs (haloscopes tuned to axion mass ranges) are actively being built and deployed.

Dark matter has properties nobody has modeled correctly yet

It's possible dark matter interacts only through gravity and never through the weak nuclear force, which would make it essentially undetectable by every experiment built so far, all of which assume some non-gravitational interaction to register a signal.

Something stranger is going on

A minority of theorists point to more exotic possibilities: primordial black holes formed in the early universe, extra spatial dimensions that only gravity can access, or quantum gravity effects nobody has a working framework for yet. None of these currently have strong observational support, but none are ruled out either.

The debate will most likely be settled by continued direct detection experiments, next-generation cosmological surveys, or, less probably but not impossible, by an unexpected theoretical breakthrough that unifies the strengths of both camps.

What's coming next

Continued underground searches

LZ and XENONnT are both scheduled for extended data-taking runs through the late 2020s, and next-generation detectors like DARWIN are already in the design phase, aiming to push WIMP sensitivity down to the neutrino floor across a wider mass range.

Better cosmological data

The European Space Agency's Euclid mission (launched 2023) and NASA's Nancy Grace Roman Space Telescope, along with the ground-based CMB-S4 project, are all designed to map dark matter's gravitational fingerprint with far greater precision than Planck, potentially distinguishing between dark matter and modified gravity models at cosmological scales for the first time.

Galaxy and stellar surveys

The Dark Energy Spectroscopic Instrument (DESI) and the Square Kilometre Array (SKA), currently under construction, will map galaxy positions and velocities across much larger volumes of sky, providing more statistical power to test the radial acceleration relation and cluster dynamics simultaneously.

Follow-up on the Gaia wide binary anomaly

Because Chae's 2023 wide binary result is still contested on methodological grounds — critics point to the difficulty of excluding unresolved multiple-star systems from the sample — expect several independent re-analyses of Gaia data over the next few years, which could either strengthen or substantially weaken the case for a real low-acceleration gravitational anomaly.

Theoretical refinement

Milgrom himself continued publishing refinements to MOND's mathematical foundations as recently as 2023, and researchers are actively working on relativistic completions that might reconcile MOND's galaxy-scale success with its cluster-scale and cosmological weaknesses.

Common misconceptions

"Dark matter is just a fudge factor"

Dark matter isn't a free parameter tuned separately for every observation. Its properties — total abundance, clustering behavior, and its role in structure formation — are constrained simultaneously by rotation curves, cluster dynamics, gravitational lensing, and the CMB, and those independent constraints agree with each other to a degree that would be a remarkable coincidence if dark matter weren't real in some form.

"Modified gravity is a fringe idea"

MOND papers appear regularly in mainstream peer-reviewed journals, including Physical Review D and The Astrophysical Journal. Milgrom, Bekenstein, McGaugh, and Verlinde are established physicists at respected institutions, not outsiders. It's a minority position within cosmology, but a scientifically serious one.

"We've definitively ruled out modified gravity"

We haven't. Modified gravity is disfavored by the weight of the current evidence, particularly cluster-scale and cosmological data, but no single observation has falsified it outright, and active research continues to try to close the gap between MOND's galaxy-scale success and its cluster-scale struggles.

"Dark matter has been directly detected"

Indirectly, through its gravitational effects, yes — repeatedly and across independent methods. Directly, meaning a dark matter particle registering an unambiguous signal in a detector, no experiment has achieved that as of 2026, despite LZ and XENONnT pushing sensitivity to unprecedented levels.

"Modified gravity is simpler because it avoids new particles"

It avoids a new particle but typically introduces new fields (as in TeVeS's added vector and scalar fields) and additional free parameters to make relativistic versions work. Whether that counts as "simpler" than a single new particle species depends on how you weigh theoretical parsimony against the number of moving parts a theory needs.

Frequently asked questions

Q: How much dark matter is there in the universe?

According to Planck satellite measurements of the cosmic microwave background, dark matter makes up about 26.8% of the universe's total mass-energy content, roughly five times more than ordinary (baryonic) matter's 4.9% share. The remainder, about 68.2%, is attributed to dark energy, a separate phenomenon driving the universe's accelerating expansion.

Q: What might dark matter actually be made of?

The leading particle candidates are WIMPs (weakly interacting massive particles), axions (originally proposed to solve an unrelated problem in nuclear physics), and sterile neutrinos (a hypothetical heavier cousin of the known neutrino types). None has been confirmed; each motivates a different style of detection experiment.

Q: Could dark matter just be ordinary matter we haven't found yet, like black holes or failed stars?

Searches for MACHOs (massive astrophysical compact halo objects, including black holes, brown dwarfs, and rogue planets) using gravitational microlensing surveys have ruled out ordinary compact objects as the dominant source of dark matter across most of the plausible mass range. Some room remains for primordial black holes in specific mass windows, but they can't account for all of it.

Q: Does the dark matter debate affect everyday physics or technology?

No. Dark matter's gravitational effects are negligible at planetary and solar-system scales; its influence only becomes significant across galactic distances and larger. GPS satellites, spacecraft trajectories, and everyday engineering calculations don't need to account for it.

Q: Is MOND the same thing as saying "we don't understand gravity at all"?

No. MOND proposes a specific, testable modification that only applies below a precise acceleration threshold and leaves Newtonian gravity and general relativity completely intact everywhere they've already been tested, including the entire solar system.

Q: Why does the Bullet Cluster matter so much in this debate?

It's one of the few observations where visible mass and the source of gravitational lensing are spatially separated, which is very difficult to reproduce with a gravity modification tied directly to visible mass. It doesn't rule out modified gravity outright, but it's the single hardest data point for that camp to explain.

Q: What is the "external field effect" in MOND, and why does it matter for the wide binary star result?

It's MOND's specific prediction that a small system's dynamics (like a wide binary star pair) should be influenced by the larger gravitational field it sits inside (the Milky Way's), in a way standard Newtonian gravity doesn't predict. Chae's 2023–2025 Gaia analyses reported an anomaly matching that exact prediction, which is why the result drew serious attention rather than being dismissed as noise.

Q: When might this debate actually get resolved?

Realistically within the next one to two decades. Either a dark matter particle turns up in an underground detector, next-generation surveys like Euclid and DESI find a cosmological signature that clearly favors one model over the other, or the tension between galaxy-scale and cluster-scale evidence persists long enough that a genuinely new theoretical framework becomes necessary.

Q: Has anyone proposed a theory that fully replaces dark matter, including at cluster and cosmological scales?

Not yet, convincingly. TeVeS and its successors extend MOND's reach into relativistic and lensing regimes but still struggle to match cluster dynamics and CMB fine structure as precisely as cold dark matter does. Emergent gravity, proposed by Erik Verlinde, is an earlier-stage framework with promising galaxy-scale results but far less development overall.

The takeaway

Dark matter explains a wider range of observations, at more scales, with fewer scale-dependent adjustments than any current modified gravity theory, which is why it remains the working assumption for most cosmologists. But "working assumption" isn't the same as "proven fact": no laboratory has ever caught a dark matter particle, and MOND's unreasonable success at predicting individual galaxy rotation curves with a single constant is a real anomaly that the standard model doesn't explain nearly as elegantly. The honest scientific position is that both frameworks currently explain part of the picture well and part of it poorly, and the resolution will come from evidence, not from which side sounds more conservative. Whether that evidence arrives from a WIMP finally hitting a xenon detector or from a wide binary star anomaly turning into a full relativistic theory of modified gravity, the answer will reshape how physicists understand the universe's basic architecture either way.

---

Read next:

• What If Parallel Universes Are Real? The Physics Behind the Multiverse
• What If the Sun Went Dark for One Year? Earth's Apocalypse Hour by Hour
• What If Humanity Became an Interstellar Civilisation?

Related Reading

• What If You Removed All the Empty Space from Matter?
• What If Earth Were Twice as Big?
• What If Parallel Universes Are Real? The Physics Behind the Multiverse
• What If the Sun Went Dark for One Year? Earth's Apocalypse Hour by Hour
• What If Humanity Became an Interstellar Civilisation?