A fixed-epoch snapshot of every small body in the Rocks in Space catalog.
Every known asteroid and comet, placed where it was at one instant: . Nothing here is animated — this is a still photograph of the solar system, not an ephemeris.
Orbital elements come from the JPL Small-Body Database. Each object is advanced to the snapshot epoch by two-body Kepler motion and converted to heliocentric ecliptic coordinates. All three conic types are solved rather than skipped: Newton's method on Kepler's equation for elliptical orbits, its hyperbolic analogue for interstellar and unbound objects, and Barker's equation for near-parabolic comets, which carry no usable semi-major axis.
Checked against JPL Horizons. For the 93.6% of the catalog whose orbit solution is
already referenced to this exact epoch, the fractional position error is about
2×10-4, set mostly by the database publishing elements to
four significant figures. Objects with older solutions drift further: a short-period
body on a fifteen-year-old epoch can land ~0.05 AU away, because two-body
propagation carries it through a dozen revolutions with no planetary perturbations.
That is invisible in a cloud of 1.5 million points and unusable for pointing a
telescope. Positions are stored quantised to about 200,000 km — smaller than one pixel
at any zoom that takes in more than a few AU.
The 42,062 near-Earth asteroids are the Atira, Aten, Apollo and Amor groups: everything with a perihelion inside 1.3 AU. Isolate them and the shape of the population is the point — they are not a belt but a diffuse swarm crossing the inner solar system, which is precisely why they are hard to find and why they matter. Most were fed inward from the main belt through resonances, so the population is continuously resupplied rather than primordial.
This group counts asteroids only. JPL's own near-Earth flag also takes in 208 short-period comets, which are in the comet group here instead.
Each population can be labelled with a leader line running to a small circle inside the cloud. The circle sits at that group's median heliocentric distance, measured from the same positions being drawn — 2.7 AU for the main belt, 41 AU for the comets, which is high because a thousand Kreutz sungrazer fragments are strung out along one very long orbit. For those four the direction carries no meaning: the populations are close to axisymmetric, so the angle is chosen only to keep the labels from piling up.
The two Trojan callouts are different, and their direction is the whole point. Jupiter's Trojans are labelled separately at L4 and L5, and those anchors are not placed at 60° by hand: the Trojans are split on which side of Jupiter they sit and each side's median longitude is measured, which comes out at +63° and −61°. Only 100 of the 16,356 fall outside either camp. Nothing in this viewer's arithmetic knows what a Lagrange point is, so that structure is an emergent result — and it doubles as a check on the orbit solver. Both camps belong to the "Other non-main-belt" population, so they switch off with it.
The three comet fragment families are labelled the same way, and they are the white clumps far out of the plane: Kreutz sungrazers (1,210 fragments, 30% of every comet in the catalog), Marsden & Kracht (93, related to 96P/Machholz) and 73P/Schwassmann-Wachmann 3 (76, which broke up in view of modern telescopes). Each is one disintegrated parent whose pieces share an orbit and differ mainly in when they reach perihelion, so at a fixed epoch they sit strung along a common track rather than spread around it. That is why they read as discrete knots and not as a haze. These are real clumps — about 95% of each family lies within 15 AU of its own centre — so unlike the four populations, the direction of the circle is meaningful, and the circle is placed on an actual member of the family rather than at an average position. They belong to the comet population and switch off with it.
Any callout is hidden when its group is off, when its anchor is off screen, or when there is nowhere to put the text that is not already taken.
The five bodies the IAU has actually accepted as dwarf planets — Ceres, Pluto, Haumea, Makemake and Eris — are drawn in gold with their orbits, and every one of them is also a point in the cloud, because they are ordinary catalogued small bodies. Their positions come from the same Kepler solve as everything else, so a label can never disagree with its own dot.
A second, dimmer tier holds eight candidates — Gonggong, Quaoar, Sedna, Orcus, Salacia, Varuna, Ixion and Varda. The first few are widely thought to qualify; for the rest the sizes are uncertain enough to leave it genuinely unsettled. None has been formally designated, so they are kept visually subordinate and off by default: "there are five dwarf planets" is the defensible sentence. Switch on the candidates and Sedna's orbit gives the scale of the problem — perihelion 76 AU, aphelion beyond 1,000 AU, one circuit every 12,700 years.
Turn off the asteroids and two narrow arcs of comets stand out, far from everything else. They are not an artefact: each is a family of fragments from a single disintegrated comet, all still on nearly the same orbit and differing mainly in when they reach perihelion — so at any one instant they lie strung out along one common track.
The bright arc is the Kreutz sungrazers: 1,210 objects, thirty percent of every comet in this catalog, on a shared orbit with perihelion just 0.005 AU from the Sun's centre and inclination 144°. Its members include the Great March Comet of 1843 and the Great Southern Comets of 1880 and 1887; most of the rest were found by the SOHO spacecraft's coronagraph as they fell in to die. The fainter arc is the Marsden and Kracht groups, 83 objects related to comet 96P/Machholz. A third, tighter knot between 1 and 5 AU is 73 catalogued fragments of 73P/Schwassmann-Wachmann 3, which broke apart in view of us.
Positions are honest, brightness is not. The main belt outnumbers the comets 350 to one, so if every object were drawn with the same opacity the smaller populations would be invisible next to it. Each group's opacity is therefore scaled up by the square root of its size deficit, capped at 6×. Densities are comparable within a population but not between them — use the checkboxes to judge one group at a time.
The Sun is drawn as a glowing marker at the origin, and its size on screen is fixed rather than physical — at this scale the real Sun is far smaller than one pixel, and every dot in this view is likewise many orders of magnitude larger than the body it stands for. The glow marks the centre of the reference frame; it is not a measurement of anything.
That slider thins the main belt only; the other three populations are always drawn in full. The belt is 93.6% of the 1,557,369 points, so it is both the whole of the performance cost and the only group dense enough that thinning it reveals structure underneath. The others are cheap to draw and are already opacity-boosted to stay visible beside the belt, so subsampling them would only make them fade. Points are shuffled within each population when the file is built, so a reduced belt is an unbiased random sample of it and not, say, the first 10% by designation.
The main belt is a torus, not a ring: tilt to a side view and its thickness is the spread of orbital inclinations. The gaps you can pick out from directly above are carved by orbital resonances with Jupiter. The two dense knots leading and trailing Jupiter by 60° are the Trojan camps, and they are in the non-main-belt group.
Small-body elements: NASA/JPL Small-Body Database. Planet elements: JPL Horizons. Research & development, not peer-reviewed. rocksinspace.org