How we simulate Baily's beads

What you see in the simulator

The simulator reproduces, for your location, the real geometry of the eclipse: the C1–C4 contacts, the solar crescent and Baily's beads — the last points of photosphere shining through the valleys of the lunar limb just before and after totality.

Bead positions and timings are not decorative: each bead corresponds to a real valley of the Moon's limb, and its instant derives from the lunar topographic profile and the apparent Sun–Moon track seen from the chosen point. The C2 and C3 contacts are corrected with that same profile: effective totality is usually a few seconds shorter than for a “smooth” Moon, and the panel shows it.

Beads are not drawn: they emerge

The simulator occults the photosphere with the real lunar-limb profile. Where the relief rises, light is cut; where there is a valley, a thread of photosphere — a fraction of an arcsecond — still gets through. That thread is the bead.

That is why the pattern changes from one location to another and from one tenth of a second to the next: it depends only on geometry, not on a pre-recorded animation.

Data and references

The computation relies on the same data families used by the eclipse-contact literature:

• Lunar topography: SLDEM2015 (SELENE/Kaguya + LRO) and LOLA (Lunar Reconnaissance Orbiter) models.
• Ephemerides: precision planetary and lunar series from NASA JPL, with lunar orientation from the SPICE kernels.
• Eclipse frame: Besselian elements from NASA's eclipse canon (Espenak).
• Limb profile: the lunar edge is sampled in ~1,800 directions around the disc, with the libration of the eclipse instant.
• Corona: classic radial brightness profile (Baumbach / van de Hulst). Chromosphere depicted at its real thickness (~3.5″).

Photographic exposure (EV)

The exposure control reproduces what a real camera captures: the relative luminances of photosphere, chromosphere, prominences and corona differ by several orders of magnitude, so no single exposure can show everything at once. Short exposures reveal beads and chromosphere; long ones, the corona — as in a real photographic sequence.

The fine structure of the corona and the prominences is illustrative; their relative brightness and response to exposure do follow the physical model.

Coverage by zones

The limb profile and the beads are computed for about a hundred reference locations across the totality band. Any other point uses the nearest reference (typically within 50 km, always under 120 km): at those distances the profile barely changes, because lunar libration varies very slowly with observer position.

Contact times, however, are always computed for your exact coordinates, and the beads are re-anchored to those contacts. The simulator's “computed for…” label tells you which profile reference is in use.

Declared limits

• The individual brightness of each bead is indicative: position and instant are computed; fine photometry (seeing, diffraction, equipment optics) is not modelled.
• Corona and prominences: illustrative structure — the real corona of 12 August 2026 will depend on that day's solar activity.
• The on-screen limb relief is exaggerated (and adjustable): at true scale it is thinner than one pixel.

We contrast the simulator's behaviour with the contact literature and with photographically verified reference simulators.