Circling the Milky Way at distances of a few hundred thousand light-years are galaxies so faint they were nicknamed ghost galaxies when Hubble first pinned down their properties in 2012. Leo IV is one of them. It contains perhaps a few thousand stars scattered across a region wider than some open clusters in our own disk, and almost every one of those stars is older than 13 billion years. When Hubble measured their ages, the picture that emerged was strange and specific: star formation in these little satellites appears to have switched off at nearly the same moment in cosmic history, more than 13 billion years ago, as if someone had thrown a single galaxy-wide breaker.
That moment lines up with the era astronomers call reionisation, when the first generations of stars flooded the early universe with ultraviolet light and burned off the neutral hydrogen fog that had filled space since the Big Bang. The ghost galaxies look like small witnesses to that event, frozen in the state they were in when the lights came on.
What a ghost galaxy actually is
The term covers a specific class of ultra-faint dwarf galaxies orbiting the Milky Way. Leo IV, Hercules and Ursa Major I are three that Hubble studied in detail. Each holds only a few thousand stars. Each sits several hundred thousand light-years from the Sun. Each is so dim that against the background sky it looks like a slight over-density of pinpricks rather than a galaxy.
The Artful Age
A weekly letter on aging well, family across generations, and the creative life after the kids leave home.
They were, as reporting on the 2012 Hubble study noted, described by the researchers as fossils of the early universe — small enough and quiet enough that whatever shut them down 13 billion years ago never got the chance to be overwritten by later star formation.
Bigger galaxies keep making stars. The Milky Way still forms new stars in its spiral arms. The ghost galaxies have not made a new star since the universe was less than a billion years old.
How you date a galaxy that stopped growing
The trick is a colour-magnitude diagram. Plot every star in a galaxy by its brightness and its colour, and the pattern that emerges depends almost entirely on the population’s age and chemistry. A young cluster shows bright blue stars still burning hydrogen fast. An old one shows only the faint, long-lived red dwarfs and a narrow band of stars turning off the main sequence at a very specific brightness — the point where the more massive stars have already exhausted their cores and moved on.
Hubble’s precision on that turnoff point is what makes the age measurement work. In the ghost galaxies, the turnoff sits at a brightness consistent with stars more than 13 billion years old, with formation histories that appear to end abruptly. No younger population sits above the turnoff. No second burst rebuilt what shut off.
The same technique — reading ages and metal content out of Hubble photometry — is now the standard tool for reconstructing the Milky Way’s early history. A team led by Davide Massari at the Bologna astrophysics observatory used it on 39 globular clusters to trace an even earlier merger, a dwarf galaxy called Low-energy-Kraken-Heracles that fell into the young Milky Way approximately 12 billion years ago.
The reionisation fingerprint
For the first few hundred million years after the Big Bang, the universe was full of neutral hydrogen gas — cold, dark, transparent to nothing in particular. Then the first stars ignited. Their ultraviolet radiation stripped electrons off hydrogen atoms across cosmic distances, and by roughly 500 to 900 million years after the Big Bang, the fog had cleared.
The physics of what that did to small galaxies is straightforward. A dwarf galaxy with only a few million solar masses of gas has a shallow gravitational well. It cannot hold onto gas that has been heated to tens of thousands of degrees. When the reionising ultraviolet background swept through, it boiled the star-making fuel out of the smallest galaxies faster than they could turn it into stars.
Big galaxies kept their gas. Small ones did not. The ghost galaxies sit right at the mass where that cutoff should bite, and the fact that they all appear to shut down at roughly the same cosmic moment is what points at reionisation as the mechanism. It is an interpretation rather than a direct measurement — you cannot photograph the ultraviolet fog turning off — but the timing and the mass scale line up with what the models predict.
Why the Milky Way’s satellites matter for cosmology
The stars in Leo IV, Hercules and Ursa Major I are chemically primitive. They formed from gas that had barely been enriched by earlier generations, so their iron content is a small fraction of the Sun’s. A star made almost entirely of hydrogen and helium, with a whisper of heavier elements, is a chemical time capsule. It records the composition of the universe at the moment its parent cloud collapsed.
That is why the ghost galaxies get compared to other cosmic fossils. Smithsonian Magazine has covered the hunt for chemically primitive galaxies at the very edge of the observable universe, where James Webb is spotting objects whose light has been travelling for more than 13 billion years. The ghost galaxies offer the same information without the redshift — they are right here, orbiting the Milky Way, and their stars can be studied one at a time.
Hubble has been running that programme in parallel with its work on the earliest galaxies visible at high redshift, including a compact, chemically young galaxy the telescope has imaged in the distant universe. The nearby ghosts and the distant infants tell the same story from different angles.
How this connects to the Milky Way’s own bones
The satellites are one part of a larger reconstruction. Astronomers are also piecing together which stars in the Milky Way itself were born here and which arrived through mergers. The Massari team’s discovery of Low-energy-Kraken-Heracles, covered by Gizmodo, pushed the record of Milky Way mergers back to a point when the universe was only about 1.5 billion years old.
LKH was a substantial dwarf galaxy. It was comparable in scale to Gaia-Sausage-Enceladus, the better-known dwarf that merged with the Milky Way roughly 10 billion years ago and helped puff up the thick disk of stars surrounding the flat spiral. Before LKH, there was thought to be only the Milky Way’s own gas. After LKH, there was a Milky Way built partly from someone else.
The ghost galaxies are what happens to dwarfs that never got eaten. They stayed in wide orbits, kept their handful of ancient stars, and drifted on for 13 billion years without making anything new.
What Hubble actually sees
Looking at a Hubble image of Leo IV is an odd experience. Most of the frame is background galaxies — distant spirals, ellipticals, the usual deep-field crowd. Leo IV itself is barely visible as a slight thickening in the density of foreground stars. You have to know exactly where to point the telescope and integrate for a long time.
Individual stars in the ghost galaxies are faint enough that ground-based telescopes struggle to resolve them. Hubble’s angular resolution — set by a 2.4-metre primary mirror above the atmosphere — is what makes it possible to pick out the main-sequence turnoff cleanly enough to date the population. As the Bologna team put it in describing their own use of Hubble on globular clusters, the combination of high resolution and depth is what allows ages and metallicities to be measured with the precision needed to separate populations that have been mixed together for billions of years.
The instrument that carried much of this work is the Advanced Camera for Surveys, installed on Hubble by shuttle astronauts and later repaired. It has been photographing dwarf galaxies for two decades. Similar deep-imaging campaigns continue on other targets, with Hubble still producing new galaxy images as it moves through its fourth decade of operation.
What it feels like, standing under it
The ghost galaxies are technically visible in the sky in the sense that they are above the horizon at times — Leo IV sits in the direction of the constellation Leo, Hercules in Hercules, Ursa Major I in the Great Bear. You will not see them with your eyes, or a backyard telescope, or a very good backyard telescope. They are too diffuse. Their combined light is spread across too much sky.
But the stars in them are the same stars that were burning when the last of the cosmic hydrogen fog was being cleared away. If you could somehow stand on a planet in Leo IV — assuming there is one, which nobody knows — the sky would be almost empty. No dust lanes. No bright young clusters. A scatter of red dwarfs, a few old giants, and beyond them, the Milky Way as a distant patch of light hundreds of thousands of light-years away.
Curiosity about deep time is one of the things that shapes a life, and this is deep time at close range. The photons Hubble caught from Leo IV’s ancient stars left them a few hundred thousand years ago. The stars themselves were already 13 billion years old when they emitted that light. They were burning when the universe was younger than a decent oak tree is now, in Earth years, if you compress the scales enough to make the comparison land.
They are still burning. The breaker that shut off star formation did not shut off the stars that were already alive. Red dwarfs of the right mass can live for a trillion years, comfortably longer than the current age of the universe. The ghost galaxies will still be there, still faint, still chemically primitive, long after the Milky Way finishes its slow collision with Andromeda and settles into whatever comes next.
What Hubble caught, in three faint patches of sky, was the moment the first stars finished clearing the fog — recorded not in a spectrum from the edge of the observable universe, but in the ages of a few thousand stars orbiting quietly overhead.