Astronomers Identify Previously Unknown 'Loki' Galaxy Hidden Inside the Milky Way

The Milky Way is not a pristine creation. It was assembled, piece by piece, from smaller galaxies consumed over billions of years. Now, a team of astronomers says they have found the chemical fingerprints of one such devoured galaxy — previously unrecognized — embedded deep within our own galactic plane. They call it Loki.

Published in the Monthly Notices of the Royal Astronomical Society in April 2026, the study led by Federico Sestito at the Strasbourg Astronomical Observatory presents evidence that 20 ancient, metal-poor stars orbiting near the Sun share a common origin in a now-dissolved dwarf galaxy with a baryonic mass of roughly 1.4 billion solar masses [1][2]. If confirmed, Loki would rank among the more significant early building blocks of the Milky Way — and raise pointed questions about how many other ghost galaxies remain undetected inside our own.

What Loki Is — and What It Isn't

Loki is not a galaxy you can point a telescope at. It no longer exists as a coherent structure. Instead, the proposal rests on a group of 20 very metal-poor stars (with iron abundances less than one hundredth of the Sun's, or [Fe/H] ≤ −2.0) found within roughly 2 kiloparsecs of the solar neighborhood [3]. These stars orbit within the galactic plane — the flat disk of stars, gas, and dust that defines the Milky Way's visible structure.

What makes them unusual is twofold. First, most very metal-poor stars in the Milky Way are found in the halo — the diffuse, roughly spherical cloud of old stars surrounding the disk. Finding a concentration of them in the plane is unexpected [1]. Second, these 20 stars share strikingly similar chemical compositions despite traveling on both prograde orbits (moving in the same direction as the disk's rotation) and retrograde orbits (moving against it), all with high orbital eccentricities between 0.5 and 0.9 [3].

The estimated baryonic mass of approximately 1.4 × 10⁹ solar masses places Loki in the range of classical dwarf galaxies, comparable in scale to the Small Magellanic Cloud [3]. For context, this is far smaller than the Gaia-Enceladus merger, which contributed an estimated 50 billion solar masses of stars, gas, and dark matter to the Milky Way roughly 8–11 billion years ago [4][5]. But it exceeds the estimated stellar mass of the Sagittarius Dwarf (~400 million solar masses in stars) and dwarfs the Sequoia progenitor (~50 million solar masses) [6][7].

Source: Compiled from Helmi (2020), Kruijssen et al. (2020), Naidu et al. (2021), Sestito et al. (2026)

Data as of Apr 1, 2026CSV

The Chemical Case for a Separate Origin

The core of the argument rests on detailed chemical abundance measurements. Using the ESPaDOnS spectrograph on the Canada-France-Hawaii Telescope (CFHT), operating at a spectral resolution of approximately 68,000 across wavelengths from 3,700 to 10,500 angstroms, Sestito's team measured abundances of more than 20 elements in each star [3].

The elements span several nucleosynthetic families:

• Alpha-capture elements (magnesium, silicon, calcium, titanium) produced mainly in core-collapse supernovae
• Odd-Z elements (sodium, aluminum, potassium, scandium, vanadium)
• Iron-peak elements (chromium, manganese, cobalt, nickel, zinc)
• Neutron-capture elements (strontium, yttrium, zirconium, barium, lanthanum, neodymium, europium), produced through rapid and slow neutron-capture processes [3]

The key finding: these stars show enrichment patterns consistent with high-energy supernovae (hypernovae), fast-rotating massive stars, and neutron star mergers — but no significant contribution from Type Ia supernovae, which are thermonuclear explosions of white dwarf stars [1][2]. Type Ia supernovae take hundreds of millions of years to begin contributing to a galaxy's chemical inventory because they require the evolution and death of intermediate-mass stars. Their absence in these stars' chemical fingerprints indicates that Loki's star formation was extremely brief — likely lasting only 0.1 to 1.0 billion years — before the system was torn apart and absorbed [3].

Critically, the team found that the chemical dispersion among the 20 stars was narrower than that of either the halo or the bulge at the same metallicity [2]. The neutron-capture element ratios ([Sr, Ba, Eu/Fe]) resemble those found in classical dwarf galaxies rather than in stars formed natively within the Milky Way [3]. This chemical coherence, across stars on opposing orbital directions, is the strongest piece of evidence that they share a common birthplace.

An Ancient Merger — Among the Earliest

When did Loki merge with the Milky Way? The chemical evidence constrains the answer: very early. The absence of Type Ia supernova enrichment and the ultra-low metallicities point to a population that formed and was disrupted before the Milky Way had fully assembled its disk, likely more than 10 billion years ago [2][5].

The Milky Way began coalescing roughly 10–11 billion years ago. Its major known mergers occurred in a rough sequence: Kraken arrived first (~11 billion years ago), followed by the Helmi Streams progenitor and Sequoia, then the massive Gaia-Enceladus event (~8–11 billion years ago), and most recently the Sagittarius Dwarf, whose tidal disruption is still ongoing [5][8]. Loki, if real, would slot into the earliest phase of this timeline — a proto-galactic building block consumed during or before the Kraken era.

Cosmological zoom-in simulations run by the team supported this scenario. The simulations showed that an in-plane infall of a single system during early galactic assembly could scatter its stars across a wide range of angular momenta, producing the observed mix of prograde and retrograde orbits [3]. In other words, what appears chaotic today is consistent with a single coherent system being violently disrupted as the proto-Milky Way was still taking shape.

Loki has reached the most advanced stage of tidal disruption: it is essentially dissolved. There is no remaining core, no identifiable stellar stream, no spatial overdensity. Only the shared chemistry of its scattered stars preserves its identity [1][3].

The Steelman Case Against Loki

The discovery team is upfront about the limitations of their evidence, and the question mark in their paper's title — "An ancient system hidden in the Galactic plane?" — is deliberate [3].

The most significant objection is the small sample size. Twenty stars is a thin foundation for claiming the existence of an entire galaxy. The authors acknowledge this directly, noting that their estimates of carbon-enhanced metal-poor (CEMP) star fractions carry large uncertainties at this sample size [3].

A second concern involves selection effects. The initial sample was drawn from the LAMOST spectroscopic survey, which may be biased against carbon-enhanced stars due to how it classifies spectra [3]. If the selection systematically excludes certain stellar types, the apparent chemical coherence could be an artifact of the sample rather than a genuine physical signal.

Third, there is the question of whether the prograde and retrograde stars genuinely share a common origin or come from separate accretion events that happen to have similar chemistry. The team argues that the mass implied by their chemical evolution models is consistent with a single system rather than two, but this depends on modeling assumptions [2][3].

A broader alternative explanation is that Loki's stars represent the "low-energy tail of Gaia-Sausage-Enceladus" — debris from the well-established major merger that happened to settle into planar orbits [3]. The authors consider this possibility but argue that the chemical differences they observe, particularly in neutron-capture elements, distinguish Loki from known Gaia-Enceladus populations.

No independent research group has yet published a verification or refutation of the Loki hypothesis. The study awaits the scrutiny that comes with larger datasets from upcoming spectroscopic surveys, particularly WEAVE (the William Herschel Telescope Enhanced Area Velocity Explorer) and 4MOST (the 4-meter Multi-Object Spectroscopic Telescope), both of which will provide homogeneous chemical abundances for millions of stars [1][3].

What Loki Means for the Milky Way's History

If confirmed, Loki adds a previously unknown chapter to the Milky Way's assembly narrative. The current consensus holds that the galaxy's stellar halo was built primarily from a handful of major mergers — Gaia-Enceladus being the dominant contributor — supplemented by numerous smaller accretion events [5][8]. The known accreted populations include Gaia-Enceladus, Kraken (also called Heracles), the Helmi Streams, Sequoia, Thamnos, Sagittarius, Pontus, Arjuna, and I'itoi [6][8].

Yet these mergers, despite being the most dramatic events in the Milky Way's history, contributed only a small fraction of the galaxy's total stellar mass — roughly the mass of the stellar halo, which is a minor component compared to the disk [6]. Loki, with its estimated 1.4 billion solar masses in baryonic matter, would be a mid-range contributor — smaller than Gaia-Enceladus but larger than most other known accreted systems [3].

More broadly, the discovery raises the question of completeness. How many other dissolved galaxies remain hidden within the Milky Way, their stars masquerading as native populations?

The Hidden Census Problem

The answer, according to multiple lines of evidence, is: many.

Supercomputer simulations from Durham University published in 2025 predicted that as many as 100 additional satellite galaxies may orbit the Milky Way at close distances, too faint to be detected with current surveys [9]. The Milky Way currently has roughly 60 confirmed companion satellite galaxies. These missing satellites would be systems stripped almost entirely of their dark matter halos by repeated passages through the Milky Way's gravitational field, rendering them extremely dim [9].

But Loki represents a different category of hidden galaxy — not one that is faint and orbiting, but one that has been entirely consumed and whose stars are now mixed into the Milky Way's own populations. Detecting these requires not better imaging but better spectroscopy: detailed chemical fingerprinting of millions of individual stars to identify groups with shared, non-native chemistry.

Source: OpenAlex

Data as of Jan 1, 2026CSV

The European Space Agency's Gaia mission, which completed its sky-scanning phase in January 2025 after a decade of observations, has been the primary driver of this field [10]. Gaia has catalogued positions and motions for nearly two billion stars, enabling the discovery of stellar streams, kinematic substructures, and chemically distinct populations that would have been invisible a generation ago [10][11]. Two major data releases are still forthcoming — Gaia Data Release 4, expected around 2026, and a final release later this decade [10].

Yet Gaia has significant blind spots. It operates primarily in optical wavelengths and struggles with the dust-dense regions of the galactic plane — precisely where Loki's stars reside. The planned GaiaNIR successor mission, which would survey the Milky Way in infrared light, is designed to probe these obscured regions [10]. Until such missions fly, the galactic plane remains the largest gap in our census.

The Tools That Made This Possible

A decade ago, this detection would have been impractical. Three developments converged to make it possible now.

First, the LAMOST survey (Large Sky Area Multi-Object Fiber Spectroscopic Telescope) in China provided the initial spectroscopic identifications of metal-poor star candidates in the galactic plane — a population that older surveys had largely ignored in favor of halo stars [3].

Second, the ESPaDOnS spectrograph at CFHT delivered the high-resolution follow-up spectra needed to measure detailed abundances across more than 20 elements per star. This level of chemical detail is necessary to distinguish genuine accreted populations from the Milky Way's native stellar chemistry [3].

Third, cosmological zoom-in simulations have reached sufficient resolution to model the dynamical evolution of individual accreted dwarf galaxies within a realistic Milky Way-like potential, allowing the team to test whether the observed orbital distribution is physically plausible [3].

All three resources — LAMOST data, CFHT observing time, and computational simulation codes — are in principle available to other research groups. Independent verification using different instruments or survey data would substantially strengthen or weaken the case for Loki.

What Comes Next

The Loki hypothesis will live or die on larger samples. The upcoming WEAVE and 4MOST spectroscopic surveys, expected to begin producing science-grade data in the coming years, will observe millions of stars with the chemical precision needed to test whether the pattern seen in 20 stars holds up across hundreds or thousands [1][3].

If more stars with Loki-like chemistry are found on planar orbits, the case strengthens considerably. If the pattern dissolves into the general scatter of halo and disk populations at larger sample sizes, Loki may prove to be a statistical fluctuation or a misidentified fragment of a known merger.

Either outcome carries scientific value. A confirmed Loki would demonstrate that the Milky Way's earliest merger history is richer than current models assume, and that dissolved galaxies can be recovered even when no spatial or kinematic coherence remains — purely through chemistry. A refuted Loki would clarify the limits of chemical tagging as a method and sharpen our understanding of the selection effects that plague small-sample studies.

For now, Loki remains what its name implies: a trickster, hiding in plain sight within the galactic plane, daring astronomers to prove it real.