In the vast space surrounding the Milky Way's disk drift long ribbons of stars known as "stellar streams"—the remnants of star clusters once captured by the galaxy and stretched thin by tidal forces. Just as the solar system orbits the galactic center, these streams trace their own paths around the galaxy. For years, astronomers have interpreted irregularities in these otherwise thin, smooth ribbons—gaps, kinks, and spurs—as scars left by collisions with invisible clumps of dark matter.
But a new paper published August 27, 2026, in The Astrophysical Journal (Volume 1008, Number 1, article 91) by Arpit Arora and colleagues at the University of Washington (DOI: 10.3847/1538-4357/ae89af) presents simulation results that call for a fundamental refinement of that assumption. In a simulation environment that completely excluded small dark-matter clumps (subhalos), the team tracked roughly 15,000 globular-cluster streams over 5 billion years. Although the published materials do not give an exact breakdown, complex morphological disturbances arose naturally in about three-quarters of all streams. Only about 70 streams remained perfectly smooth throughout the simulation.
This result does not prove that dark matter doesn't exist. Rather, it quantifies a "complexity floor"—the baseline level of structural noise that a galaxy's own gravitational field generates on its own—and calls for a fundamental reconsideration of the common practice of treating observed stream irregularities as straightforward evidence of dark matter.
Scars long attributed to dark matter
Dark matter makes up the majority of the universe's mass and forms the scaffolding on which galaxies grow, yet because it neither emits nor absorbs electromagnetic radiation, its physical nature remains unknown. Nora Shipp, assistant professor at the University of Washington and co-author of the study, notes: "Dark matter makes up most of the mass in the universe and forms the scaffolding on which galaxies grow, but we still don't know what it is. The Milky Way is one of the best laboratories we have for figuring that out, and stellar streams are among the most sensitive tools within it."
Theoretically, when a globular cluster is torn apart by a galaxy's gravity, it should retain a single, smooth, symmetric linear structure as long as it remains in a uniform gravitational potential. If the Milky Way's dark matter halo harbors countless compact subhalos on the order of (solar masses), these would deliver gravitational kicks as they cross a stream, locally disturbing the arrangement of stars.
Astronomers have long recorded irregularities in real observational data as evidence of subhalo encounters. A prime example is the "GD-1" stream in the Milky Way's halo. GD-1 exhibits a "spur"—a protrusion jutting outward from the main body of the stream—along with localized "gaps" where star density drops. Prior studies, including Bonaca et al. (2019), interpreted these features as caused by encounters with compact dark-matter subhalos. Similarly, the sharp kinks observed in the ATLAS and Aliqa Uma streams have been cited as evidence of close encounters with unseen perturbing objects.
These inferences rested on the assumption that the host galaxy's gravitational potential would not significantly disturb a stream's local width or density. But real galaxies are not uniform spheres. A rotating bar structure sits at the center, spiral arms wind through the disk, and asymmetric gravitational responses from past mergers with dwarf galaxies constantly ripple through the system.
15,000 tracked streams with dark-matter clumps deliberately removed
To isolate the effect of the host galaxy's own structure on stellar streams, Arpit Arora and colleagues designed a controlled simulation experiment. From the Latte suite of the FIRE-2 cosmological hydrodynamic simulations, they selected four galaxy halos comparable in mass to the Milky Way (m12i, m12f, m12m, and m12b). These four host galaxies span a wide range of formation histories, from ones that evolved in relative isolation to ones that experienced major mergers comparable to the Large Magellanic Cloud (LMC) or the Sagittarius dwarf spheroidal galaxy.
The team deliberately excluded dark matter subhalos and giant molecular clouds (GMCs) from these galaxies. The gravitational field within the simulation was modeled as a time-dependent potential using a Basis Function Expansion (BFE). This approach accurately reproduces time-varying features—the bar, spiral arms, disk asymmetries, and halo-wide fluctuations from satellite interactions—while, by design, smoothing away and eliminating any small-scale, localized perturbers such as subhalos.
Within this virtual host-galaxy environment, the team placed roughly 15,000 globular-cluster streams as populations of test particles and computed their orbital evolution over 5 billion years. The initial orbits of the streams were set with pericenter distances (the closest approach to the galactic center) ranging from 10 to 30 kpc, and eccentricities spanning nearly circular to highly elongated elliptical orbits—comprehensively covering the range of orbital parameters actually observed for Milky Way streams. Snapshots were taken at a time resolution of about 25 Myr, and the cumulative velocity error from interpolation was kept below 0.1% of orbital velocity.
Lead author Arpit Arora describes the results: "In our simulations, the host galaxy alone produced exactly the same kinds of irregularities seen in real stellar streams. Now that we can predict what the host galaxy contributes on its own, we can start to isolate the piece that dark matter is responsible for."
Only 70 streams remained smooth—a striking result
The simulation results showed that even with dark matter subhalos completely absent, about 75% of the roughly 15,000 streams (the published materials do not give a specific breakdown) developed complex morphological changes. The structures that emerged included deviations from the orbital path, kinks, spur-like protrusions, branching, density gaps, clumpy structures, and even cocoon-like envelopes enveloping the entire stream. After 5 billion years of evolution, only about 70 streams (roughly 0.47%) remained perfectly smooth and featureless.
To objectively evaluate stream morphology, the research team introduced three quantitative diagnostic metrics:
- Off-track structure: The presence of stellar extensions or protrusions extending outward from the central line of the main orbit.
- Width variation: Non-uniformity in the thickness of the ribbon along its length.
- Along-track density: The distribution of stars along the stream, including the depth of local overdensities and gaps.
The analysis found that even the most well-preserved streams showed width variation along their orbits. Moreover, the angular scale of density overdensities and gaps produced by the host galaxy's gravitational perturbations alone reached about 2°. This 2° scale matches the 1°–5° range that conventional theoretical models had predicted would form from collisions with dark-matter subhalos of $10^6$–$10^8 \text{M}_\odot$.
| Evaluation criterion | Conventional assumption (smooth potential) | Observed features (e.g., GD-1) | This simulation's result (no subhalos) | Implication for dark matter searches |
|---|---|---|---|---|
| Overall stream morphology | A single smooth line of uniform width | Spurs, kinks, cocoon-like structures | Spurs and kinks arose naturally in about 75% of all streams | Optical shape alone cannot distinguish host-galaxy effects from subhalo effects |
| Width uniformity | Constant thickness maintained along the entire length | Local widening and narrowing present | Even the smoothest streams showed width variation | Width variation cannot be attributed to subhalo tidal effects with certainty |
| Density gaps | Continuous, uniform linear star density | Clear density dips on 1°–5° angular scales | The host galaxy alone produced gaps at roughly 2° scales | Gap detection alone cannot be used to infer subhalo mass |
| Survival rate (remaining smooth) | Smoothness maintained absent external collisions | Nearly all known streams show disturbances | Only about 70 of 15,000 streams (~0.47%) remained smooth | The mere presence of disturbances is not evidence of dark matter |
The single biggest factor determining a stream's degree of morphological disturbance was its pericenter distance. In the simulations, a distance of about 15 kpc served as the dividing line between relatively smooth streams and heavily disturbed ones. Streams whose pericenters fell inside 15 kpc repeatedly passed close to the dense disk structure and the rotating bar, experiencing intense gravitational perturbations that tore them apart and produced complex irregularities. By contrast, streams on nearly circular orbits with pericenters beyond about 20 kpc experienced little disturbance from the host galaxy and retained the smoothest morphology.
A control experiment described in Section 6 of the paper, comparing results against a static, axisymmetric gravitational potential, confirms that this time-varying disk structure is the primary driver of disturbances in streams passing through the inner halo.
Identical optical shapes—and a path forward through kinematics
In the paper, Arora and colleagues state explicitly that "based on photometric morphology alone—the apparent shape projected on the sky—perturbations caused by the host galaxy and those caused by dark-matter subhalos are, in principle, indistinguishable." The fact that shapes strikingly similar to GD-1's observed spur and the kinks seen in the ATLAS and Aliqa Uma streams emerged in simulations that assumed no dark-matter collisions whatsoever demands a fundamental reassessment of the conventional approach of inferring subhalo mass and spatial distribution purely from static images of streams.
However, this study does not demonstrate that dark-matter subhalos don't exist, nor that streams are useless for dark matter searches. What the research establishes is a "complexity floor"—a baseline noise level that must always be subtracted when extracting a dark-matter signal from observational data.
The paper points to stellar kinematics as a path forward past this floor. If the proper motions (movement across the sky) and radial velocities (movement along the line of sight) of individual stars can be measured with high precision—not just their positions—it may become possible to separate the large-scale tidal flow caused by the host galaxy's gravity from the localized velocity anomalies produced when a compact mass passes directly by.
The study also comes with several caveats and limitations. First, only four host-galaxy halos were used in the simulations, which does not fully capture the statistical diversity of galaxy formation histories. Second, due to computational constraints, the stars in the streams were treated as test particles, meaning the fine structure of the clusters' own self-gravitational collapse is only approximated. Third, giant molecular clouds (GMCs) within the galactic disk were excluded, so real galaxies may introduce even more complex perturbations. These are not flaws in the physical model but deliberate constraints designed to isolate the effect of the host galaxy's large-scale structure alone.
Next-generation telescopes and the outer halo
Co-author James Davenport, a research assistant professor at the University of Washington, remarks: "Unfortunately, there's no convenient shortcut to unraveling the structure of dark matter. Streams are complicated systems, but they remain one of the most fascinating tools we have close to home for studying dark matter."
As a roadmap for future research, the team plans comparative experiments that introduce an explicit population of dark-matter subhalos into the simulation environment. By running the host-galaxy-only perturbation pattern side by side with a pattern that includes subhalo collisions, and by combining positional and kinematic (velocity) information, they aim to develop algorithms capable of statistically distinguishing between the two.
On the observational side, a major turning point is approaching. The NSF–DOE Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST), currently under construction, along with the European Space Agency's Euclid space telescope and NASA's Nancy Grace Roman Space Telescope, are expected to map hundreds of streams in the Milky Way and nearby galaxies at high resolution.
Rubin Observatory in particular—although the published materials give no specific reference distance—is expected to detect streams out to distances beyond previous limits, comprehensively capturing the population of undiscovered streams in the Milky Way's outer halo. As this study shows, streams on circular orbits more than 20 kpc from the galactic center experience little gravitational disturbance from the host galaxy and retain their original smooth shape most faithfully. In other words, streams floating in the outer halo—unobscured by the host galaxy's "complexity floor"—represent the clearest observational targets for detecting genuine traces of dark-matter subhalos.
By precisely characterizing the complex background generated by the Milky Way's own gravity, the search for dark matter is entering a new phase—one where increasingly refined theoretical models intersect with next-generation observational data.