A Census Rewritten from Space

A single image from space has rewritten our census of one of the universe's densest neighborhoods.

The Euclid space telescope, operational since 2023, has detected 1,100 dwarf galaxy candidates in the Perseus cluster—a massive gravitational assembly some 72 million light-years from Earth. Of these, 606 appear to be entirely new discoveries. The findings emerge from Euclid's Early Release Observations, which combined unprecedented depth, spatial resolution, and field of view to reveal structures that have eluded ground-based surveys for decades.

Why Dwarfs Matter

Dwarf galaxies are the universe's quiet majority. They contribute little to the total light of galaxy clusters but dominate by sheer number. Defined loosely as systems with stellar masses below a billion suns, they range from compact, star-forming irregulars to diffuse, quenched ellipticals. Some host nuclei. Some trail streams of stars. Others appear to float as ghostly smudges, their surfaces so faint they vanish against the sky.

Understanding them matters because they trace the cluster's assembly history, its dark matter distribution, and the environmental forces that strip galaxies of their gas and stars.

Breaking Through the Limits

Previous surveys of Perseus—using ground telescopes like the William Herschel and WIYN—had catalogued several hundred dwarfs. But those observations were limited. Either they covered small areas, or they struggled to distinguish faint dwarfs from background galaxies and image artifacts.

Euclid changed that. Its VIS instrument captured red-optical light at 0.16 arcsecond resolution, while its Near Infrared Spectrometer and Photometer added three near-infrared bands. Together, they reached surface brightness limits near 30 magnitudes per square arcsecond, well below the threshold where many dwarfs begin to fade into noise.

The Human Eye Meets Machine Precision

Seven team members visually inspected the images using a custom annotation tool. They marked candidates, flagged nuclei, assessed morphologies, and noted the presence of globular clusters—compact star clusters that orbit galaxies like satellites. The process was collaborative and iterative. Candidates flagged by at least five of the seven classifiers made it into the final catalogue. Those with lower agreement were set aside.

A Detailed Demographic Portrait

The results paint a detailed demographic portrait. Ninety-six percent of the sample are dwarf ellipticals, quenched systems devoid of star formation. Fifty-three percent host nuclei, dense stellar cores at their centers. Twenty-six percent are globular cluster-rich, surrounded by swarms of ancient star clusters. Six percent show signs of disturbance—tidal tails, asymmetries, warped structures—evidence of ongoing or recent interactions.

Eighty-five candidates meet the definition of ultra-diffuse galaxies: effective radii exceeding 1.5 kiloparsecs and central surface brightnesses fainter than 24 magnitudes per square arcsecond in the g band. That's roughly 8 percent of the sample, a higher fraction than found in some other clusters. The diffuse end of the dwarf parameter space is more densely populated here than in earlier studies, likely because Euclid's combination of depth and resolution allowed clearer separation of genuine dwarfs from contaminants.

The Nuclei Tell a Story

Nucleated dwarfs follow expected trends. They tend to be brighter, larger, rounder, and have flatter surface brightness profiles than their non-nucleated counterparts. The nucleated fraction drops sharply with decreasing luminosity: 81 percent of bright dwarf ellipticals host nuclei, but only 33 percent of faint ones do.

This gradient aligns with models in which nuclei grow through the infall and merger of globular clusters, a process more efficient in larger, denser systems.

Globular Clusters as Cosmic Tracers

Globular clusters themselves emerged as both tracers and curiosities. Automated detection combined with visual inspection identified cluster populations around the dwarfs. The team estimated the total number of clusters per galaxy, corrected for background contamination and incompleteness.

They found that the specific frequency—clusters per unit galaxy luminosity—ranges between the values measured in the Virgo and Coma clusters, suggesting Perseus is dynamically intermediate: neither as evolved as Coma nor as active as Virgo. Dwarfs located near the east-west strip, where most of the cluster's brightest galaxies reside, tend to have higher specific frequencies. A handful of dwarfs show complex nuclear structures—compact cores with tidal features, possibly witnessing the migration or merger of massive clusters.

A Cluster Still Assembling

The spatial distribution of the dwarfs is asymmetric. The main isodensity center lies about 110 kiloparsecs west of the brightest cluster galaxy, NGC 1275. A secondary peak appears farther west. This double structure mirrors the distributions of globular clusters and intracluster light, both of which also shift westward.

The alignment suggests these offsets are real, not artifacts of data reduction. They point to recent merger activity, consistent with earlier observations that Perseus is still assembling.

Ultra-diffuse galaxies show an even stronger spatial asymmetry. Roughly half cluster in the western overdensity, while the rest scatter more uniformly. There are hints that their distribution tracks the positions of bright galaxies, suggesting they may have joined the cluster as part of infalling groups. Whether they formed within the cluster or accreted from outside remains unclear.

Six Hundred New Worlds

Comparison with earlier catalogues reveals Euclid's gains. The study matched 494 of its dwarfs with previously known systems. The remaining 606 are new. Most of the new detections are faint and small, benefiting from Euclid's depth and the ability to resolve structure.

Some prior surveys missed them simply because their imaging was shallower. Others struggled to separate dwarfs from artifacts or background galaxies without high-resolution confirmation. Euclid resolved both problems.

A Method That Works

The work also demonstrates a method. Visual classification, though labor-intensive, proved reliable. Agreement between classifiers hovered near 80 to 90 percent, and comparison with spectroscopic cluster membership confirmed an accuracy rate in the same range.

The few mismatches mostly involved objects at the dwarf-galaxy boundary or galaxies at slightly higher redshift whose absolute magnitudes would make them implausibly large if they were truly at the Perseus distance.

The Road Ahead

Looking ahead, the Euclid Wide Survey will image roughly 14,000 square degrees of sky. If this single 0.7 square degree field yielded 1,100 dwarfs, the full survey should detect hundreds of thousands. That scale demands automation. Machine learning techniques, trained on catalogues like this one, will need to replicate the classifiers' ability to distinguish dwarfs from contaminants while handling noise, crowding, and artifacts. The Perseus field will serve as a benchmark.

The findings also raise questions. Why do some dwarfs host multiple nuclei? Are these mergers in progress, or compact stellar disks seen from particular angles? Why do globular cluster populations vary so much at fixed luminosity? What determines whether a dwarf becomes ultra-diffuse? And what fraction of the faint, disturbed systems are actually background galaxies masquerading as cluster members?

What Comes Next

Answers will require follow-up. Spectroscopy can confirm distances and measure velocities. Near-infrared imaging from missions like Roman or JWST could provide color information to better separate cluster members from interlopers and constrain ages and metallicities. Deeper Euclid observations of other clusters will reveal whether Perseus is typical or exceptional.

For now, the Perseus dwarf catalogue stands as proof of concept. Euclid can detect, classify, and characterize faint galaxies across a range of environments and masses. It can resolve their nuclei, count their globular clusters, and trace their spatial distributions. It can do all of this simultaneously, in a single observation, without the need for extensive ground-based follow-up.

The faint end of the galaxy luminosity function is no longer hidden.

Credit & Disclaimer: This article is a popular science summary written to make peer-reviewed research accessible to a broad audience. All scientific facts, findings, and conclusions presented here are drawn directly and accurately from the original research paper. Readers are strongly encouraged to consult the full research article for complete data, methodologies, and scientific detail. The article can be accessed through https://doi.org/10.1051/0004-6361/202450799