The Universe in Three Dimensions

What this map does and does not know

The radial axis is honest and the angles are honest, but they are not honest in the same way, and the difference is drawn rather than hidden.

1. A dot means we know which way it lies. The direction comes from the name resolver at Strasbourg — SIMBAD, then NED — not from anybody's memory, and every position on the page traces back to a catalogue entry.
2. A ring means we know how far but not which way, so the map draws the whole sphere that the distance allows. The Sloan and CfA2 Great Walls are rings because they are tens of degrees across and have no single position to resolve. Structures we are inside are rings too, drawn at half their extent, which is where their edge actually is from here.
3. The radius is the logarithm of distance, so shapes are wrong on purpose. A sphere is still a sphere and directions are exact, but anything with real extent is squashed toward us — the near side of a supercluster is pushed out and the far side pulled in. This is the only projection in which the Oort Cloud and the horizon appear in one picture, and the cost is that no volume here is the right shape.
4. The planets are orbits, not places, and they go round the centre rather than round you. A planet's direction changes through the year and its distance from us changes with it, so a dot would state a position that is not a fact; what does not change is its distance from the Sun. Zoom in far enough and the label at the centre stops reading here and reads the Sun — this is the one point on the map where the origin and the observer come apart, and the Earth is the third ring out. Each orbit is drawn as a circle at its mean distance, in the plane of the ecliptic: eccentricity is not drawn (Mercury's is the only one you would notice — it really runs between 0.31 and 0.47 AU), and neither are the inclinations, which are 7.0° for Mercury, 3.4° for Venus and under 2.5° for the rest. The asteroid and Kuiper belts are bands, because a belt has no orbit and a single ring would claim one.
5. Everything is drawn as it is now, not as it looks. Light from the far rings left before the Earth existed. A map of where things are and a map of what you see diverge completely past the Local Group, and this is the first.
6. Distance is comoving, matching the ladder. Redshift is not converted to a light-travel time anywhere on this page.
7. Four objects move, in three dimensions. Each carries a measured galactocentric velocity vector, not just a speed towards us, so the map carries its position forward and re-derives both its distance and its direction — they cross the sky as well as approach. Within the span each is drawn for, Andromeda shifts about 7° and the others 5-6°. The Sun is held at today's place: its own 230-million-year lap of the galaxy would swing the entire sky and tell you nothing about the Local Group.
8. The clock is logarithmic, spanning one million years to seven billion, for the same reason the radial axis is: the Large Cloud rounds the galaxy every couple of billion years while Andromeda takes four to arrive, and a satellite’s pericentre passage lasts a few tens of millions. Everything else on the map holds still, because nothing else in the catalogue has a measured velocity yet — add one in Notion and it starts moving.
9. Only Andromeda is actually falling towards us. Each mover also carries the size of its sideways motion, and the comparison is the point. M31: 110 km/s towards us against 17 sideways — near enough a head-on fall. M33: 44 against 230. The Large Cloud: 64 against 314. The Small Cloud: 7 against 217, which is a galaxy going almost purely round. The Local Group is not collapsing inward; it is in orbit, and M31 is the exception rather than the illustration. This map can draw only the radial part, which is why the other three hardly move on it while travelling at hundreds of kilometres a second.
10. This is now an integration, not an extrapolation. Earlier versions carried each object along a straight line, which cannot describe an orbit — satellites froze after a few tens of millions of years and Andromeda arrived late. The page now integrates the mutual gravity of the Milky Way and everything with a measured velocity, stepping a leapfrog through a galaxy modelled as a Hernquist bulge-and-disc inside an NFW halo. The Clouds complete real orbits — the Large Cloud swings between about 26 and 217 kpc, four times over eight billion years — and Andromeda's fall accelerates as it should.
11. The halo mass is a slider because nobody knows it. The Milky Way's dark halo is uncertain by roughly a factor of two, and the arrival of Andromeda turns on it: from about 4.4 billion years at 0.5×10¹² solar masses to 3.3 at 2.5×10¹². Move it and watch. The rotation speed at the Sun is printed beside it as the check — the model returns 208 to 239 km/s across that range, against a measured 229, which is precisely why the rotation curve alone cannot pin the halo down. The number this page shows you is a consequence of a parameter you can see and change, which is the honest way to show a prediction that depends on one.
12. Friction, and what it changes. A satellite ploughing through the halo leaves an overdense wake behind it, and the wake pulls back — Chandrasekhar drag, which scales as the satellite's mass squared and so matters for the Large Cloud and for Andromeda and hardly at all for the Small. Turn it off with the button and Andromeda swings past at closest approach and climbs back out: the merger does not happen. Turn it on and the same pass is deeper, and the orbit decays into one. Drag is what makes a merger a merger, and it is the clearest thing on this page to toggle.
13. What it still does not model. Tidal deformation and mass loss, once galaxies are close enough to stop being points; the halo's own response beyond taking up the momentum; and any mass outside these five bodies. The Small Cloud's track is the least trustworthy — it is bound to the Large Cloud in reality, and five bodies with 5 kpc of softening handle that pair poorly.