The S-Cluster of Sagittarius A*

What you are looking at

Within roughly 0.04 parsec of the Galactic Center lies the S-cluster: a swarm of mostly young, B-type main-sequence stars on tightly bound, highly eccentric orbits about an invisible mass of 4.15 million Suns. Their orbits are the cleanest dynamical proof we have that the dark mass is a single supermassive black hole, Sagittarius A*. This map takes the published orbital elements for each star and integrates the two-body Kepler problem in three dimensions, so the ellipses you see are the real measured orbits, not illustrations.

The reference star is S2 (also S0-2), whose complete 16-year orbit has been tracked since the 1990s. Its 2018 pericenter passage — a 120-AU approach at nearly 3% of the speed of light — delivered the first detections of gravitational redshift and Schwarzschild precession around a supermassive black hole. The faint, fast inner stars S62 and S4711–S4715 were reported later from independent imaging; they probe even smaller radii but their orbits remain debated, so they are drawn separately and dashed.

The default “sky plane” vantage looks straight down the line of sight, recovering the familiar overlapping-ellipse figure seen in the discovery papers. Rotate the scene to appreciate that these orbits are randomly oriented in three dimensions — the S-cluster is not a disk.

What is exact here, and what is not

Accurate. The orbital elements (a, e, i, Ω, ω, T_p, P) are the published best-fit values, used unmodified. Each orbit is integrated as a true two-body Kepler ellipse with Sgr A* at the focus; the derived quantities check out against the literature — S2 returns P = 16.0 yr, pericenter 119 AU, apocenter 1934 AU, pericenter epoch 2018.3, and a peak speed of ~7,700 km/s (2.6% c), while S4714 reaches ~8% c. The projected (on-sky) shapes, sizes, eccentricities, and orientations of the ellipses are the directly measured quantities and are reproduced faithfully.

Simplified, by design. These are Newtonian two-body orbits: they omit the relativistic Schwarzschild precession (small but real — measured for S2 by the GRAVITY Collaboration in 2020) and any perturbation from extended mass inside the cluster, so over many orbits the ellipses do not slowly rotate as the true orbits do. The absolute three-dimensional handedness (which lobe lies in front of the sky plane) follows the standard astrometric Ω/i/ω convention; the on-sky projection — what telescopes actually measure — is unaffected by that choice.

Brightness is faithful. Every S-star lies at essentially the same distance (R₀ = 8.18 kpc), so its relative apparent brightness is set entirely by its K-band magnitude. Marker luminance and size both track that magnitude on the logarithmic (perceptual) scale astronomers use, spanning the catalogue from K ≈ 10 (bright) to K ≈ 18.5 (faint) — so a dim dot really is a dim star as seen from Earth, not an arbitrary choice.

Sgr A* is not a bright object. Despite its four-million-solar-mass scale, the black hole accretes far below its Eddington limit and is a faint near-infrared source — quiescent K ≈ 16.7, dimmer than most stars on this map, brightening only briefly during flares. It is therefore drawn as a dim reddish point at that true magnitude, with a thin reticle ring marking the dynamical centre. Its event horizon (~0.08 AU in radius; the EHT shadow is ~0.43 AU across) is thousands of times smaller than one pixel here, so the ring is a locator, not a disk; nothing about the central marker should be read as a luminous body.

Symbolic, not to scale. Marker sizes are fixed on-screen symbols for legibility; a real S-star is only ~0.05 AU across — far below a pixel — so the dots flag positions, never physical radii.

Contested. The six faint fast stars (S62, S4711–S4715, dashed) come from independent imaging by Peißker et al. Their orbits — including the headline ~8% c speed of S4714 — are debated; the GRAVITY Collaboration has argued that some, such as S62, may be confusion between known stars rather than genuine new sources. They are kept in a separate, visually distinct layer so they are never confused with the well-determined sample.