The Inferred Milky Way Dark Matter Distribution Isn't Spherical

Measuring matter dynamics outside the plane of spiral galaxies is critical 

Rotations curves of, and gravitational accelerations of, matter in the vicinity of spiral galaxies that is above or below the galactic plane where most of the ordinary matter in these galaxies is found, is critical to distinguishing between competing dark matter particle and gravity or fifth force based explanations of dark matter phenomena (or hybrids of the two paradigms like self-interacting dark matter).

These theories have been formulated and fine tuned to reproduce the dynamics of stars in the plane of spiral galaxies where they are much easier to observe and measure, and good data has been available for many decades. But because good data has not been available for the dynamics of stars outside the galactic plane of spiral galaxies, different models formulated to explain dark matter phenomena differ considerably in what they predict about that.

Measuring matter dynamics outside the plane of spiral galaxies is hard and has only recently become a viable possibility

But until very recently our astrophysical observations provided us with little data and limited accuracy outside the galactic plane of spiral galaxies with various kinds of "telescopes" for a variety of reasons. 

In the case of the Milky Way, the main problems have been that the density of stars to observe outside the galactic plane of the Milky Way is much lower than in or near the thin galactic disk where most of its stars are found, and the complication that as observers who are inside the Milky Way, the vantage point of our observations is obstructed by dense stars in the galactic plane or otherwise non-optimal.

In the case of other galaxies, one of the main problems have been that it is hard to determine if a particular star is in the galactic plane or above (or below) that plane unless we have a close to edge on view of the galaxy, that measuring rotation curves is hard with a true edge on view. Another problem is that the resolution of our view of a galaxy gets worse as the galaxy gets more distant which is especially a concern outside the plane of a spiral galaxy where the density of the stars we are trying to observe is low. And, when looking at another galaxy it is particularly hard to tell if a star outside the main plane of the galaxy is really part of the same gravitationally bound system, or is millions of megaparsecs away from it in the foreground of our observation of that galaxy.

Fortunately, we live in an era where we have an abundance of riches when it comes to astronomy observations, producing a torrent of data from extremely powerful telescopes like the Gaia space observatory (a telescope in orbit around Earth). The data from this space telescope is used by the Gaia collaboration's network of over 400 scientists and engineers funded by the European Space Agency (ESA) to build the most accurate 3D map of the Milky Way ever constructed.

As the paper below explains in its abstract, Gaia's measurement uncertainties are less than 5% for the vertical velocities of stars that it observes in the Milky Way, and are less than 20% for the vertical accelerations that it measures. 

These uncertainties may not seem all that great to someone unfamiliar with the details of galaxy scale astronomy observations. But in that subfield of astronomy,  uncertainties as low as 28% (i.e. 0.1 dex) are the considered good, and relative uncertainties on the order of 50%-100% are common place, so Gaia's measurements are gold standards of precision by comparison.

Comparing models

Both simple cold dark matter models (with their spherically symmetrical NFW dark matter particle halos) and MOND (even in its relativistic generalizations) predict dark matter phenomena are spherically symmetrical, which makes these theories mathematically much more tractable. 

Indeed, coming up with any kind of dark matter particle model without either (1) self-interactions more complex than a simple scalar field, or (2) interactions in excess of ordinary gravitational interactions with ordinary matter, that do not form spherical or nearly spherical dark matter halos, is extremely challenging and could very well be impossible (although I'm not aware of any analytically constructed "no go" theorem to that effect).

But not all explanations of dark matter phenomena predict spherically symmetric effects, and inferred dark matter halo shapes from prior observations have tended to favor non-spherical, rugby ball shaped inferred distributions of dark matter particles (even though theoretically, it has been challenging to come up with dark matter particle theories that reproduce these shapes).

Some gravity based explanations of dark matter phenomena, like the one described by Deur, also propose non-spherical dark matter phenomena in spiral galaxies. In Deur's approach dark matter phenomena arise from non-linear self-interactions within gravitational fields that manifest in, and only in, non-spherical matter distributions like those found in spiral disk disk galaxies. 

In Deur's analysis, in spiral galaxies, the pull of gravity towards the galactic center is stronger than the Newtonian expectation in the direction of rays from the galactic center in the galactic plane (especially at larger radii), while it is weaker than the Newtonian expectation in the vertical direction relative to the galactic plane (an effect which accounts, at least in part, for dark energy phenomena between galaxies).

Deur's analysis is also supported by another key data point that corroborates astronomy observations that infer non-spherical dark matter particle distributions in dark matter particle paradigms. He has observed that the relative proportion of matter in a galaxy that is made of luminous stars to inferred dark matter is strongly correlated in elliptical galaxies, with the extent to which the elliptical galaxy is not perfectly spherical.

New, high quality data shows that the Milky Way's inferred dark matter halo is not spherical 

Gaia has assembled new data with record breaking accuracy and sample sizes on the rotational velocities and accelerations of stars in the Milky Way based upon their polar coordinates (i.e. their distance from the Galactic center and their distance from the plane of the Milky Way spiral disk). This data, was compiled by the Gaia collaboration, and was analyzed and reported in a pre-print released today of an accepted for publication astronomy paper.

The new paper's analysis strongly favors inferred dark matter particle distributions which are not spherically symmetric. Instead, it strongly favors the inference in a dark matter particle paradigm of a flattened disk-like configuration around the ordinary matter of the Milky Way.

The Gaia data generically rules out all dark matter particle explanations of dark matter phenomena, gravity or fifth force based explanations,  and hybrid explanations (like self-interacting dark matter models), that predict spherically symmetric dark matter phenomena effects. 

This is a huge deal because most of the leading explanations of dark matter phenomena are spherically symmetric, and all of those models are now definitively ruled out.

The paper

We derive both the mid-plane and off-plane rotation curves, v(c)(R,z), and the vertical acceleration, a(z)(R,z), of the Milky Way (MW) using Gaia~DR3 data over the ranges of vertical heights z∈(−2,2) kpc and galactocentric distances R∈(8.5,14) kpc where the velocity components are determined with high precision, i.e., with an error <5%. In contrast, the vertical acceleration a(z)(R,z) is dominated by model-dependent systematics, with uncertainties of up to ∼20%. This level of accuracy allows us to place stringent constraints on the geometry of the MW's dark matter (DM) distribution, as the vertical gradients of the gravitational potential attain their maximum within this range of radial and vertical distances corresponding to the characteristic scales of the disk. 

We find that models including the observed stellar components together with a spherical DM halo fail to reproduce both the pronounced variation of v(c)(R,z) with height and the observed behavior of a(z)(R,z). 

In particular, spherical halos with a scale radius of rs∼15 kpc contribute negligibly to the off-plane rotation curve and vertical acceleration in the inner disk, leaving these features primarily determined by the stellar mass distribution. 

Conversely, models in which DM is confined to a flattened, disk-like configuration predict substantial contributions to both v(c)(R,z) and a(z)(R,z), resulting in a markedly better agreement with the data. We conclude that disk-like DM distributions are strongly favored over spherical halo models. 

Forthcoming Gaia data releases will enable even more stringent tests of the geometry and distribution of the MW's DM component.

Francesco Sylos Labini, Roberto Capuzzo-Dolcetta, "Constraining the Geometry of Galactic Dark Matter with Gaia Data Release 3" arXiv:2606.12548 (June 10, 2026) (accepted for publication in The Astrophysical Journal) (emphasis added in abstract).

The body text of the conclusion further explains that:

Our results show that the DM disk model provides a significantly better agreement with the data than the standard Navarro–Frenk–White (NFW) halo profile. 

In particular, spherical halos with characteristic scale radii of order ∼ 10 kpc contribute only marginally to the off-plane rotation curve and to the vertical acceleration within the inner disk, leaving these quantities predominantly determined by the distribution of the stellar mass. As a consequence, halo-based models systematically underestimate the measured vertical accelerations and fail to reproduce the observed decline of the rotation curve at intermediate heights. 

In contrast, models in which the DM is confined to a flattened, disk-like configuration predict substantial contributions to both the radial and vertical components of the gravitational field, leading to a markedly improved agreement with the observed trends of v(c)(R,z) and a(z)(R,z). This improvement is particularly evident at low to intermediate heights (|z| ≲ 2 kpc), where the vertical acceleration inferred from the data cannot be explained by the baryonic components alone. 

The success of the DM disk model arises from its geometry: a flattened mass distribution naturally enhances the vertical component of the gravitational potential without requiring an excessive total mass, and simultaneously reproduces the modest decline of the circular velocity with increasing z. These results strongly suggest that a significant fraction of the MW’s dark matter is distributed in a disk-like structure rather than in a quasi-spherical halo. 

Forthcoming Gaia data releases, offering improved statistics and reduced systematic uncertainties in stellar kinematics, will enable more stringent and spatially extended tests of the geometry of the Galaxy’s dark matter component, potentially allowing one to constrain its vertical and radial scale lengths with unprecedented precision.