Galaxies and the cosmic web: galaxy types, the Local Group and large-scale structure

At a glance

A galaxy is a gravitationally bound system of stars, gas, dust and dark matter. Galaxies are not scattered at random: they gather into groups and clusters, which in turn line up along filaments and sheets around enormous, nearly empty voids. Astronomers call this pattern the cosmic web. The table collects a few reference numbers used throughout this page.

Sources and dates: NASA Hubble release of 31 May 2012 (page updated January 2025); University of Helsinki release on Sawala et al., June 2025; McConnachie, Astronomical Journal 2012; University of Hawaiʻi release on Tully et al., Nature, 4 September 2014; Berkeley Lab release, 19 March 2025; ESA release, 19 March 2025. The km/s figure is our conversion of NASA’s miles-per-hour value.

Sources: [4] [5] [6] [7] [9] [11]

Galaxy types: the Hubble–de Vaucouleurs scheme

Edwin Hubble developed a classification scheme for galaxies in 1926, based on their appearance on photographic plates. Ellipticals are smooth, featureless systems labelled E0 to E7 according to how flattened they look, from round (E0) to strongly elongated (E7). Spirals have a central bulge and a disc with arms; they are labelled Sa, Sb and Sc as the bulge becomes less prominent and the arms more open, and barred spirals carry a “B” (SBa, SBb, SBc). Lenticulars (S0) sit between the two: they have a disc but no obvious arms. Galaxies that fit none of these shapes are classed as irregular. ESA notes that the scheme is now regarded as somewhat too simple, but that its basic ideas still hold.

Gérard de Vaucouleurs revised and extended the sequence in 1959. He added the late stages Sd and Sm, made the bar a three-step “family” (SA without a bar, SAB with a weak bar, SB with a strong bar) and introduced “varieties” that record whether the arms start from an inner ring (r) or directly from the centre or bar ends (s). The intermediate SAB class is regarded as one of the hallmarks of his system. Irregulars of type Im form the last normal stage of the revised sequence.

Sources: [1] [2]

Colour and shape: the red sequence and the blue cloud

Morphology is only one way of sorting galaxies. When the Sloan Digital Sky Survey (SDSS) measured colours for a large sample of nearby galaxies, a clear two-humped distribution appeared. Baldry and colleagues (2004) showed that, across a wide range of luminosities, the colour distribution of SDSS galaxies is well described by the sum of two populations, one red and one blue, each with its own colour–magnitude relation. Remarkably, this split emerges without any morphological cut.

The two populations broadly track how galaxies live. The red sequence is dominated by ellipticals and lenticulars whose stars are mostly old and in which little new star formation is taking place. The blue cloud is dominated by spirals and irregulars that still hold cold gas and keep forming young, hot, blue stars. Galaxies in between, sometimes called the “green valley”, are often interpreted as systems whose star formation is shutting down. Why and how fast that “quenching” happens (gas stripping in clusters, feedback from a central black hole, or simply running out of fuel) remains one of the central questions of galaxy evolution; our galaxy evolution dossier treats it in depth.

Sources: [2] [3]

The Local Group and the future of Andromeda

The Milky Way belongs to a small association of galaxies called the Local Group. Its two giants are the Milky Way and the Andromeda galaxy (M31); the spiral M33 in Triangulum is another prominent member. Around them orbit dozens of dwarf galaxies, including the Large and Small Magellanic Clouds. McConnachie’s widely used catalogue lists more than 100 galaxies with distance estimates within 3 megaparsecs of the Sun, a volume that reaches out to, but does not include, the next nearest groups. New faint dwarfs are still being found in deep imaging surveys, so the membership count is not final.

Andromeda lies about 2.5 million light years away and is one of the few large galaxies moving towards us: its light is blueshifted. In 2012 NASA announced that Hubble measurements of Andromeda’s sideways motion pointed to a head-on collision in about 4 billion years, followed by a merger into a single elliptical galaxy roughly 6 billion years from now.

A 2025 study led by Till Sawala (University of Helsinki), published in Nature Astronomy, revisited the question with Gaia and Hubble data and 100,000 simulations that also included M33 and the Large Magellanic Cloud. Only slightly more than half of the simulated futures ended in a merger within 10 billion years, and a direct collision within 4–5 billion years occurred in only about 2% of them. M33 pulls Andromeda towards us, while the Large Magellanic Cloud deflects the Milky Way’s motion; the uncertainty in Andromeda’s sideways velocity dominates the result. The fate of the Local Group is therefore an open question rather than a settled prediction.

Sources: [4] [5] [6]

Groups, clusters, superclusters and the cosmic web

Above the scale of groups come clusters, which contain hundreds to thousands of galaxies bound in a deep gravitational well filled with hot, X-ray-emitting gas. The nearest large one is the Virgo Cluster. Clusters and groups are in turn embedded in superclusters. In 2014 R. Brent Tully and colleagues used the motions of galaxies, not just their positions, to draw the boundary of our home supercluster: the region inside which galaxy flows converge. They named it Laniakea, “immense heaven” in Hawaiian. It spans about 500 million light years and contains about 100,000 galaxies with a total mass of roughly 10¹⁷ Suns.

On the largest scales, matter is arranged in a network. Dense nodes (clusters) are connected by filaments, which are bordered by flattened walls or sheets, and between them lie voids. Analyses of cosmological simulations, such as that of Cautun and colleagues (2014), find that clusters and filaments are the most prominent components of this web, while voids and sheets take up most of the volume. The web is the visible imprint of tiny density fluctuations in the early Universe, amplified over billions of years by gravity, mostly the gravity of dark matter. Those seed fluctuations are seen directly in the cosmic microwave background (see the Big Bang and cosmic history).

Sources: [5] [7] [8]

How redshift surveys map the web

A picture of the sky is two-dimensional. To see the web in three dimensions, astronomers need a distance for every galaxy, and for large samples the practical route is redshift. The expansion of the Universe stretches light on its way to us, so spectral lines appear at longer wavelengths; the fractional shift is called z. Through the Hubble–Lemaître law and a cosmological model, z can be turned into a distance and a lookback time (the calibration of that conversion is the subject of our cosmic distance ladder page).

Redshift surveys measure spectra for millions of galaxies. SDSS pioneered this at scale; its fifth phase, SDSS-V, published its twentieth data release (DR20) in July 2026. The Dark Energy Spectroscopic Instrument (DESI) collects light through 5,000 optical fibres at once. Its first main-survey data release (DR1, 19 March 2025) contains spectra of 18.7 million objects observed between May 2021 and June 2022, and the collaboration aims to record more than 50 million galaxies and quasars. DESI’s cosmology analyses use baryon acoustic oscillations (BAO): a preferred separation between galaxies left over from sound waves in the early Universe, which acts as a standard ruler at every redshift.

ESA’s Euclid space telescope approaches the same problem from imaging. It measures the shapes of very large numbers of galaxies to map the weak gravitational lensing caused by dark matter, and adds near-infrared spectra. Its first “quick” release (Q1, 19 March 2025) covered 63.1 square degrees in three deep fields and included a classification catalogue of more than 380,000 galaxies and some 500 strong-lens candidates.

Sources: [9] [10] [11]

Lab: from redshift to distance

Choose a redshift to see how far away a galaxy is today, how long its light has travelled and how old the Universe was when it set out. Try the Virgo Cluster, the quasar 3C 273 and the galaxy MoM-z14, then switch the Hubble constant between the Planck and SH0ES values to see how much the answers move.

The calculator uses a simplified flat ΛCDM model with parameters close to Planck 2018 (see the method note inside the lab). At very low redshift, galaxies’ own motions through space (peculiar velocities) are comparable to the expansion velocity, so a “redshift distance” to the Virgo Cluster is only indicative.

Masses, dark-matter halos and active galaxies

The stars we see are only a small part of a galaxy’s mass. Rotation speeds in the outer discs of spirals, the velocities of galaxies in clusters, hot cluster gas and gravitational lensing all indicate that each galaxy sits inside a much larger and more massive dark-matter halo. In the standard picture, small halos formed first and merged into larger ones, and galaxies grew within them by accreting gas and by mergers. The evidence and the leading candidates for dark matter are discussed on our dark matter and dark energy page.

At the centre of most massive galaxies lies a supermassive black hole. When gas falls onto it, the surrounding accretion disc can outshine the whole host galaxy; the system is then called an active galactic nucleus. The most luminous examples are quasars. The first to be identified, 3C 273 in Virgo, was recognised in the early 1960s by Maarten Schmidt and Bev Oke; its light has taken some 2.5 billion years to reach us. Energy released by active nuclei is one of the leading explanations for why massive galaxies stop forming stars, linking black holes to the colour bimodality described above.

Sources: [3] [14]

Current research (2024–2026)

• DESI DR2 (March 2025). BAO measurements from more than 14 million galaxies and quasars in three years of data. Combined with cosmic microwave background data, a model in which dark energy evolves with time is preferred over a cosmological constant at 3.1σ, and at 2.8–4.2σ when supernova samples are added, depending on the sample. The BAO-only parameters are in mild (2.3σ) tension with those from the CMB. This is a hint, not yet a discovery; see cosmic history for context.
• Euclid DR1 schedule (as of October 2026). ESA’s update of 15 June 2026 splits the first cosmology release into two stages: “DR1-Foundation” in November 2026, with raw and calibrated images, catalogues and spectra for about 1,900 deg², and the complete DR1, including the higher-level galaxy-clustering and weak-lensing products, in mid-2027.
• The Local Group’s future (June 2025). Sawala and colleagues found only about a 50% chance of a Milky Way–Andromeda merger in the next 10 billion years.
• The earliest galaxies. JWST spectroscopy confirmed MoM-z14 at redshift z = 14.44, seen about 280 million years after the Big Bang (Naidu et al., 2025). The abundance of such luminous early galaxies is more than 100 times larger than pre-JWST consensus models predicted, a central puzzle for galaxy-formation theory; see our James Webb page.

Sources: [6] [10] [12] [13]

Going further: data and tools

Much of the data behind this page is public. DESI’s releases (spectra, redshift catalogues and documentation) are served from the DESI data portal. SDSS spectra and images can be queried through the SDSS website and its SkyServer tools. Euclid releases are distributed through the ESA Euclid science archive. For any named galaxy, the NASA/IPAC Extragalactic Database (NED) collects published redshifts, distances and classifications. A good first exercise is to download the redshifts of galaxies in a thin slice of sky and plot them as a “pie diagram”: walls, filaments and voids appear immediately.

Known vs open questions

Well established

• Galaxies fall into broad morphological classes, and nearby galaxies split into a red, mostly passive population and a blue, star-forming one.
• The Milky Way is part of the Local Group, which lies within the Laniakea supercluster.
• Matter is organised into a cosmic web of nodes, filaments, walls and voids, shaped mainly by the gravity of dark matter.
• Most massive galaxies host a central supermassive black hole; quasars are its most luminous phase.

Open questions

• What quenches star formation, and how quickly?
• Why are luminous galaxies so abundant at z > 10?
• Will the Milky Way and Andromeda merge, and when?
• Is dark energy constant, as DESI hints it may not be?

Related on this site

• Research dossier: galaxy evolution
• Research dossier: cosmology
• The Milky Way
• Dark matter and dark energy
• The Big Bang and cosmic history
• The cosmic distance ladder
• Euclid
• Cosmic scales tool