Two billion years after the Big Bang, the young Milky Way was not growing in isolation. Hubble ages of 39 globular clusters now point to a buried dwarf galaxy — Low-energy-Kraken-Heracles — swallowed long before the famous Gaia-Sausage-Enceladus crash.
 |
| About 11.8 billion years ago, a dwarf galaxy known as LKH merged with the early Milky Way. Image credit: NASA / ESA / Joseph Olmsted, STScI. |
Key Takeaways by Planet Today
The claim: A peer-reviewed Nature Astronomy study, released online on 17 August 2026, reports three distinct age-metallicity sequences among Milky Way globular clusters. One sequence is the proto-Milky Way, one is Gaia-Sausage-Enceladus, and a third is assigned to an earlier merger named Low-energy-Kraken-Heracles (LKH).
The timing: NASA and ESA describe the event as about 11.8 billion years ago, roughly two billion years after the Big Bang and about 1.8 billion years before Gaia-Sausage-Enceladus. The paper also places the accretion near 12.3 billion years ago at redshift greater than 4, depending on the reference clock.
The mass: LKH is estimated at about 5 × 108 solar masses in stars — similar, within uncertainties, to Gaia-Sausage-Enceladus — and a substantial fraction of the much smaller proto-Milky Way.
The method: Hubble photometry of 39 inner globular clusters, combined with Gaia dynamics and statistical age-metallicity modeling, is what separates the third sequence from in-situ clusters and from GSE debris.
The debate it tries to close: Earlier work split the same inner-Galaxy fossils among names such as the Low-energy group, Kraken, and Heracles, and some papers argued those clusters were born inside the Milky Way. The new paper says they are one accreted progenitor.
The implication: If the third sequence holds, the earliest “bricks” of the Galaxy include stars that formed outside it. That matters for how the inner halo, bulge, and early disk were built — and for how often large galaxies grew by eating neighbors rather than by quiet in-situ star formation.
What is still open: Absolute ages, exact mass, and whether LKH was one galaxy or more than one merger without surviving clusters remain model-dependent. The authors themselves note other, smaller progenitors cannot be ruled out.
A house whose first bricks came from next door
“Our home is the Milky Way galaxy, but we do not know how our house was built.” That line, from lead author Davide Massari of INAF’s Astrophysics and Space Science Observatory in Bologna, is the honest starting point. The Milky Way is a large spiral with hundreds of billions of stars. It did not arrive that way. It grew by forming stars from its own gas and by collecting stars, gas, and dark matter from other galaxies.
The recent chapters of that story are comparatively well known. The Sagittarius dwarf has been falling in for more than six billion years and is still being shredded. About ten billion years ago, a more violent event — Gaia-Sausage-Enceladus, or GSE — dumped stars onto highly radial orbits and helped reshape the Galactic disk. Between those two, smaller mergers left streams and lumps. The harder question has always been the first two billion years, when the proto-Milky Way was smaller, denser, and easier to confuse with the galaxies it absorbed.
On 17 August 2026, Nature Astronomy published Massari and colleagues’ case that a third, earlier sequence is now visible in globular-cluster ages and metallicities. They name the progenitor Low-energy-Kraken-Heracles, or LKH, after three earlier research threads that had already circled the same idea. NASA summarized the result the same day; ESA and Hubble issued matching releases; Reuters, Sci.News, IFLScience, and ABC followed on 18 August. The latest official image note from NASA, dated 19 August 2026, still frames the collision as an artist’s concept of a dwarf galaxy meeting a young Milky Way about 12 billion years ago.
That is the news. The rest is how the evidence was built, what mainstream coverage emphasized, where the older scientific argument sat, and what the paper does not yet prove.
What Hubble actually measured
Globular clusters are dense balls of tens of thousands to a few million stars. Many are among the oldest stellar systems in the Galaxy. Because they formed quickly and then evolved as relatively closed populations, their color-magnitude diagrams can be turned into ages, and their spectra into metallicities — the fraction of elements heavier than helium. Plot age against metallicity and you get an age-metallicity relation, or AMR. Different galaxies enrich at different rates. If you can measure relative ages well enough, clusters born in different systems can fall onto different sequences even after those systems have been torn apart.
The team used Hubble imaging of 39 globular clusters in the inner roughly 20,000 light-years of the Milky Way, the region where the oldest merger debris should still be concentrated. Chiara Zerbinati, a co-author at the University of Bologna, put the instrumental point simply: Hubble’s resolution and depth let them measure age and metal content with unusual precision. Gaia then supplied the dynamical context — energies and orbits — that sort clusters into in-situ, GSE-like, and “low-energy” families.
Three sequences appeared. One follows the proto-Milky Way: clusters that formed inside the main progenitor and chemically evolved with it. One matches GSE. Between them sits an intermediate sequence. Those clusters are older than the GSE set but younger than the in-situ set at a given metallicity. The only clean explanation the authors accept is a separate dwarf galaxy that was accreted about 1.8 billion years before GSE and dumped most of its stars and clusters inside about six kiloparsecs of the Galactic center.
“These are the clusters that were born in LKH, and they tell us when that galaxy was devoured by ours, and how massive it was.” — Chiara Zerbinati, University of Bologna
Twelve of fifteen clusters previously placed in the so-called Low-energy group land on that third sequence in the new accounting. The stellar mass assigned to LKH is about 500 million solar masses, written in the paper as ≃ 5 × 108 M⊙, comparable to GSE within the quoted uncertainties. Relative to today’s Milky Way that is a small bite. Relative to the proto-Galaxy it is not. The authors argue the mass ratio was higher than GSE’s, likely above 0.2–0.3, which would make the event one of the most consequential mergers the Galaxy ever experienced.
Primary data and methods are public in outline. Many cluster catalogs come from the Hubble Space Telescope UV Globular Cluster Survey (HUGS). Age estimates and dynamical associations are posted by the Bologna group. The statistical fitting code is provided as supplementary material with the Nature Astronomy paper. An earlier arXiv version, 2601.18896, was submitted in January 2026 and updated to the accepted text on 17 August.
What the official agencies said
NASA’s Hubble mission write-up, dated 17 August 2026, is the cleanest public statement of the result: Hubble has “definitive evidence” of a dwarf-galaxy merger about 11.8 billion years ago, extending the known merger timeline 1.8 billion years further back. ESA’s parallel release stresses Gaia’s role in separating the cluster populations. Both agencies use the LKH name and the 500-million-solar-mass stellar figure. Both quote Massari’s line about the first significant batch of bricks.
Reuters, on 18 August, added a useful human translation of the violence. Massari told the agency that from the point of view of stars the merger was rather peaceful — no star-to-star crashes — but from the point of view of gas it was more “explosive,” likely triggering a burst of star formation when LKH’s gas hit the proto-Milky Way’s. That is an inference, not a direct Hubble movie of the collision, and it should be read as such.
Mainstream science outlets stayed close to the press kit. Sci.News republished the core narrative on 18 August. IFLScience walked readers through why globular clusters are useful clocks. Gizmodo and The Brighter Side of News framed it as a “lost chapter” and a “lost galaxy buried in ancient star clusters.” ABC News in Australia used an artist impression and noted that future telescopes will keep rewriting the family tree. None of these pieces claimed the result was beyond debate. They reported a Nature Astronomy paper backed by NASA and ESA.
For readers who follow space science on this site, the same observational logic — old light, careful chemistry, and the difference between a press image and a measured sequence — shows up whenever laboratories and agencies recast everyday materials as engineered systems. See, for example, Planet Today’s science reporting on lab-grown and CRISPR shifts in chocolate manufacturing, and the broader Planet Today science and news desk.
The older argument the paper says it settles
This is not a story that appeared from nowhere in August 2026. For several years, Galactic archaeologists have argued over whether the inner Galaxy hides one extra massive accretion event or a messy pile of smaller ones.
In 2019, Massari and others used Gaia to isolate a “low-energy” group of globular clusters. In 2020, Kruijssen and collaborators, using the E-MOSAICS simulations, argued that those clusters matched a predicted early merger they had already named Kraken — possibly the most major merger the Milky Way ever had, even if it was still formally a minor merger by stellar-mass ratio. Later, a chemo-dynamical stellar population in the inner Galaxy was named Heracles. Some papers treated Kraken, Heracles, and the Low-energy group as overlapping descriptions of one object. Others treated them as mixed bags. Still others said at least some of those clusters and stars formed in situ, inside the proto-Milky Way, and should not be counted as accreted at all.
The new paper’s language is unusually direct for a journal article. It says the identification of a third merger “puts to rest earlier debates.” It names LKH in honor of those prior works and then argues that LKH is a cleaner, low-metallicity version of Heracles and that most Low-energy and Kraken members belong to it. In the authors’ telling, the original Kraken definition had become an implausible leftover bin for everything that was neither Sagittarius nor GSE. Precise relative ages, they say, cut the bin into a real progenitor.
“Some past studies have argued that the earliest phases of our galaxy’s evolution were defined by stars born only in our galaxy. Here, we have shown that stars born in external galaxies also need to be considered.” — Davide Massari
That is the closest thing this story has to an “alternative” camp, and it is not a media camp. It is an intramural scientific one. There is, as of 20 August 2026, no serious alternative-media narrative that Hubble faked a dwarf galaxy or that LKH is a cover for something else. The disagreement that matters is technical: how many progenitors, how to assign clusters, whether aluminum or other abundance ratios can separate accreted from in-situ stars in the metal-poor regime, and how much weight to give age-metallicity sequences versus orbits and chemistry alone.
Readers should keep that distinction. A settled debate inside a paper is not the same as a debate that every specialist now considers closed. Independent teams will re-fit the same 39 clusters, add clusters Hubble is still observing, and test whether a single AMR sequence can be produced by two smaller galaxies whose clusters have similar ages. The authors already concede they cannot rule out other, less massive progenitors that left no surviving globular clusters or whose clusters were fully disrupted.
How a number becomes 11.8, or 12.3, or “two billion years after the Big Bang”
Press offices need one date. The paper uses several related clocks, and they are not in conflict if you watch the wording.
GSE is anchored near 10 to 10.5 billion years ago. The new sequence is measured as about 1.8 billion years older than that GSE reference. Subtract and you land near 12.3 billion years ago, at redshift greater than 4. NASA and ESA instead say about 11.8 billion years ago, or just two billion years after the Big Bang. Both descriptions are in the public record. The difference is mostly how one rounds look-back time and which absolute age zero-point is used. The robust claim in the paper is the relative one: this event is clearly earlier than GSE and later than the oldest in-situ clusters in the sample.
That relative ranking is why Hubble mattered. Ground-based ages were not precise enough, in this team’s view, to split the third sequence from the other two. Once the sequences separate, the mass estimate follows from how many clusters sit on the intermediate track and from scaling relations that connect globular-cluster populations to the stellar mass of their parent galaxy. Those relations have scatter. A factor of two either way would not erase the event, but it would change whether LKH was “GSE’s twin” or a somewhat smaller cousin.
Why an early, heavy merger changes the story of the inner Galaxy
Galaxy-formation theory in the standard cosmological picture already expects large galaxies to grow hierarchically. The new result does not invent mergers. It moves a massive one into the first couple of billion years, when the proto-Milky Way was still compact, and it parks most of the debris in the inner six kiloparsecs.
That location is the point. GSE stars are famous for their sausage-shaped velocity distribution in the halo. LKH, if the assignment is right, sank deeper. Its stars and clusters are now mixed into the inner halo and the regions that feed the bulge. Chemical models that assume the metal-poor inner Galaxy is almost purely in-situ would then be missing a large external contribution. Simulations that treat the first major merger as GSE would be starting the clock too late. Dark-matter and gas brought in with LKH would have arrived while the disk was still assembling, not after a mature spiral was already in place.
Massari’s gas-versus-stars comment sits here. Stars in a dwarf galaxy can be added with relatively little star-star violence. Gas can shock, compress, and form new stars that are neither purely LKH nor purely proto-Milky Way. Those “in-between” stars would look chemically mixed. Finding them — or failing to find them — is a test the community can run with large spectroscopic surveys already on the sky.
The result also sharpens a quieter accounting question. Even if Kraken/LKH was the most major merger the Milky Way ever had, earlier simulation papers still found that the Galaxy grew most of its stellar mass in situ. A 5 × 108 M⊙ dwarf is a lot of stars for the early universe and not a lot compared with the Galaxy today. Both statements can be true. The drama is the mass ratio at the time, not the fraction of the present-day disk.
What “definitive” does and does not mean
NASA used the word definitive. The journal paper is more careful in the fine print and more confident in the abstract. Both can be read without conspiracy and without credulity.
What is strong: a homogeneous Hubble age set; a statistical AMR split into three sequences; a spatial concentration of the intermediate clusters toward the inner Galaxy; and a naming scheme that maps onto earlier, independent hints. What is not a photograph: no one has an image of LKH as a separate galaxy. The artist’s concept released by NASA, ESA, and Joseph Olmsted at STScI is an illustration of a collision, not a snapshot from 12 billion years ago. What remains model-dependent: the conversion from cluster counts to stellar mass, the absolute age scale, the assumption that one sequence equals one progenitor, and the chemical boundary between Heracles-like accreted stars and in-situ stars at the lowest metallicities.
Fernando Aguado-Agelet, a co-author at the University of Vigo and the University of La Laguna, noted that Hubble is still observing globular clusters that have never been studied this way. That is the right next sentence. If the third sequence survives a larger sample, LKH becomes part of the standard merger tree. If the sequence smears out, the “one massive event” reading weakens and the older multi-progenitor picture returns.
For a sense of how quickly adjacent Hubble science is moving, ESA/Hubble’s 2026 archive already includes a 13 July release on a stellar-mass black hole in Omega Centauri and a 23 June result on a galaxy clearing its neighborhood only 1.4 billion years after the Big Bang. The LKH paper is one tile in a larger effort to time-stamp the first few billion years with objects we can still resolve inside our own Galaxy.
A short timeline of a long construction project
About 13.8 billion years ago, the Big Bang. Within roughly two billion years, the proto-Milky Way is already forming globular clusters and, if this paper is right, is massive enough to swallow LKH. Around 12.3 to 11.8 billion years ago, LKH’s stars and clusters settle into the inner Galaxy. About 10 billion years ago, GSE arrives and rearranges the disk and halo. Smaller mergers continue. More than six billion years ago, Sagittarius begins its long disruption, which is still underway. In the last decade, Gaia maps the motions; Hubble refines the ages; Nature Astronomy, on 17 August 2026, publishes the three-sequence claim.
That is a construction history, not a myth. It is also incomplete. Streams without clusters, clusters without a unique chemical tag, and dark-matter-only subhalos will not all show up in an AMR plot. The authors say so. Any popular account that treats LKH as the last missing brick is doing the press-release version of the work, not the paper’s.
How to read the next papers
Three tests will matter more than another round of headlines.
First, extra clusters. If newly observed inner globular clusters fall on the same intermediate sequence, the case hardens. If they fill the gaps between sequences, the three-track picture blurs.
Second, field stars. Heracles was defined in stars, not only in clusters. If large spectroscopic samples show a stellar population with LKH’s age-metallicity track, the same energy, and the same inner radial concentration, the merger is no longer a cluster-only argument.
Third, independent ages. JWST and, later, Roman can attack cluster and field-star chronologies with different systematics. A result that appears only in one Hubble reduction pipeline is a method. A result that survives another observatory is a fact about the Galaxy.
Until then, the honest position is the Swiss one. Mainstream agencies and a peer-reviewed journal have presented a coherent, sourced case that a GSE-class dwarf merged with the proto-Milky Way about 1.8 billion years before GSE. Earlier literature had already proposed pieces of that event under other names and had also proposed that some of those pieces were born at home. The new work argues the pieces belong together and came from outside. Readers can follow the data releases rather than the adjectives.
Related reading on this site, where laboratory claims and long historical records get the same sourcing treatment: the archaeological report on a mass grave of infants in ancient Israel, and the ongoing Science stream at Planet Today.
Sources
- D. Massari et al., “Evidence of a massive accretion event 1.8 billion years before the Gaia-Sausage-Enceladus merger,” Nature Astronomy, published online 17 August 2026. DOI: 10.1038/s41550-026-02931-5
- arXiv accepted version: https://arxiv.org/abs/2601.18896
- NASA Hubble: Hubble Solves Merger Mystery From Milky Way’s Early Years (17 August 2026)
- ESA: Hubble & Gaia solve our galaxy’s merger mystery (17 August 2026)
- ESA/Hubble science release heic2611: Hubble solves merger mystery from Milky Way’s early years
- Sci.News: Milky Way Swallowed Dwarf Galaxy Nearly 11.8 Billion Years Ago (18 August 2026)
- Reuters: The Milky Way devoured a smaller galaxy about 11.8 billion years ago (18 August 2026)
- Illustration credit: NASA / ESA / Joseph Olmsted (STScI)
Original source
Primary news write-up used as the assignment seed: Sci.News, “Milky Way Swallowed Dwarf Galaxy Nearly 11.8 Billion Years Ago,” 18 August 2026, https://www.sci.news/astronomy/low-energy-kraken-heracles-galaxy-15000.html. Underlying research article: Massari et al., Nature Astronomy, 17 August 2026.
Disclaimer for fact-checkers: This article reports a peer-reviewed study and contemporaneous NASA/ESA communications. “Definitive evidence” is the agencies’ phrase for a third age-metallicity sequence assigned to one accreted progenitor. Cluster ages, the conversion from clusters to stellar mass, and the decision to treat Kraken, Heracles, and the Low-energy group as a single galaxy remain scientific inferences with published uncertainties. No claim here requires hidden data, and no claim here is immune to a larger cluster sample or a different age calibration.