A star that blinked before it should have

On the tenth of March 1977 several groups pointed telescopes at a ninth-magnitude star in Libra, expecting to watch it disappear behind Uranus. The point of the exercise was the planet’s atmosphere: the way a star fades as it sets behind an atmosphere is a temperature and pressure profile, and there was no other way to get one.

Thirty-five minutes before the planet was due, the star blinked. Then it blinked again, and three more times. Nobody had predicted anything, and the immediate suspicion was equipment — a cloud, a tracking glitch, a guide error. What settled it was what happened after the planet had passed: five more blinks, in reverse order, at very nearly the same times before and after closest approach.

The whole discovery is contained in that one sentence, and it is worth noticing how little it needed. No image was taken. Nothing was resolved. The instrument was a photometer recording a single number many times a second, and what it recorded was a star getting fainter and then brighter again, ten times in an hour.

Why symmetry is the argument #

A dip in a light curve can be a great many things. A thin cloud, a detector artefact, a passing satellite, an unrelated variable star, an error in the telescope drive: all of them produce a brief loss of light, and none is easy to exclude from a single record.

What none of them produces is a mirror image. If the obscuring material is arranged in circles concentric with the planet, the sight line crosses each circle twice — once approaching, once receding — at radii that are equal by construction. So the times are symmetric about closest approach, the depths are the same to within the noise, and the number of events matches.

The probability of five accidental dips arranging themselves that way is negligible, and the argument requires no model of anything. It is a geometric consistency test, and it is the reason a discovery that looked like an instrument fault on the night was secure within hours.

Tests of that shape recur throughout this collection and they share a structure: an accidental explanation would have to reproduce a pattern it has no reason to know about. A caustic crossing in a microlensing light curve is believed for the same reason, and so is a transit that repeats on a strict period. What makes them convincing is not the signal-to-noise of any single feature but the improbability of the arrangement.

The same test then does the rest of the work. Each pair of matched events gives a radius; the radii from different observing sites give the ring plane’s orientation; and the residual departures from perfect symmetry are the measurement — because a ring that is slightly eccentric, or slightly inclined, produces asymmetries whose pattern says which.

What the technique resolves #

An occultation’s resolution is not set by any aperture. The sight line sweeps across the ring plane at a speed set by the geometry — typically tens of kilometres a second — so a detector sampling at a kilohertz resolves structures tens of metres across.

That is four or five orders of magnitude finer than any image, and it is why occultations remained the primary tool for ring structure even after spacecraft arrived.

The reason is worth stating in general terms, because it applies far beyond rings. An imaging measurement’s resolution is limited by diffraction at the aperture, and improving it costs aperture. A timing measurement’s resolution is limited by how fast the geometry changes and how fast the detector reads, and improving it costs neither. Wherever a measurement can be reformulated as a timing problem it usually should be — which is the same trade that makes a mirror a poor instrument next to a pair of separated ones for certain questions.

The fringes are not only a limit. A diffraction pattern from a point source is sharp, and the pattern actually recorded is that sharp pattern convolved with the angular size of the star projected to the ring — so the degree of blurring is a measurement of the star’s diameter, and a bright nearby star makes a worse ring measurement than a distant one. Choosing occultation targets involves preferring small hot stars for exactly this reason.

Turning a light curve into a profile #

The conversion from time to radius is where the precision is won or lost.

What is recorded is a count rate against time. Converting it needs the star’s apparent path across the planet, which needs the star’s position, the planet’s ephemeris, and the observer’s location — and an error of a tenth of an arcsecond in the star’s position is hundreds of kilometres at Uranus, which is a catastrophe for a measurement aiming at hundreds of metres.

The way out is bootstrapping. Ring features of known radius — sharp edges, previously measured rings — are used to solve for the geometry, and the rest of the profile is then measured against them. The ring system becomes its own ruler, and absolute radii are established slowly, from the events with the best astrometry, while relative structure is measured superbly from every event.

That division — excellent relative precision, mediocre absolute accuracy — is the signature of every differential measurement, and it decides which questions can be asked. How wide a ring is, how far apart two rings are, and whether a ring has moved between two epochs are all answerable to metres. Where the ring is, in the planet’s own coordinate system, is answerable to kilometres, and it took decades and a spacecraft to do better.

What the rings turned out to be #

The Uranian rings were unlike Saturn’s in every respect that mattered, and the differences were read straight off the occultation profiles.

They are narrow — most are under ten kilometres wide, against Saturn’s tens of thousands — and they have sharp edges. They are eccentric: the widest, the epsilon ring, varies from twenty kilometres wide at one end of its orbit to a hundred at the other, and its radius varies by eight hundred kilometres, which was determined from the asymmetries in exactly the way described above. And they are dark, reflecting a few per cent of the light that falls on them, which is why no image had shown them.

The eccentricity is the detail that turned out to matter most, because it is not an ordinary orbital eccentricity. Every particle in the epsilon ring shares very nearly the same orientation of its own eccentric orbit, and it keeps sharing it despite the differential precession that ought to smear the orientations apart within a few centuries. Something keeps the ring’s apsides aligned, and the candidate — the ring’s own self-gravity, feeble as it is — is the only self-gravitating structure in the solar system whose entire existence is inferred from a set of light curves.

A narrow ring is a puzzle. Left alone, a ring spreads: collisions transfer angular momentum outwards and the material diffuses. A ring ten kilometres wide should spread to hundreds within a few thousand years, so something is confining it. The prediction for Uranus was therefore that small shepherd moons should exist, and two were found flanking the epsilon ring when a spacecraft arrived nine years later. That is about as clean a case as the subject offers: a structure discovered by accident, a dynamical problem it posed, a specific prediction, and a confirmation by an entirely different instrument.

The same technique on an atmosphere #

The observation the 1977 teams thought they were making is worth describing, because it is the other half of what occultations do.

The atmospheric application has produced results the ring one cannot: the discovery of Pluto’s atmosphere in 1988, the detection of its subsequent pressure changes, and the temperature profiles of Triton and Titan. All of them come from watching a star fade for a few seconds.

Three ring systems found the same way #

The Uranian discovery was the first and it was not the last, and the sequence is a good demonstration of how a technique matures.

Neptune’s rings were found in the mid-1980s by the same method and were harder, because the events were only sometimes symmetric: a dip on one side of the planet and nothing on the other. That asymmetry, which would have been a fatal objection to the Uranus result, was here the finding — the material is confined into arcs occupying a fraction of the circumference rather than into complete rings, and it took a spacecraft image to confirm it. The arcs are held by a resonance with a nearby moon, and the confinement of material in longitude as well as in radius is still not fully accounted for.

Then, three decades later, rings were found around two small bodies in the outer solar system: a centaur and a dwarf planet, objects a few hundred kilometres across whose gravity is feeble. Both were found by occultation, both by the symmetry test, and neither was predicted by anything. Nothing in the accepted picture of ring confinement obviously works at that scale, and the discoveries opened a question rather than closing one.

The pattern across all three is the same. A technique built for one purpose, sensitive to structure far finer than anything else available, and therefore the first to see whatever happens to be there. Occultations found rings around three of the four giant planets’ worth of bodies not because anybody was looking for rings, but because a photometer sampling quickly is an instrument with almost no prior about what it might record.

A gravity field read off a ring #

The rings turned out to be an instrument pointed at the planet, and the measurement was made years before anything visited it.

A ring particle’s orbit precesses because the planet is not a point mass: its oblateness makes the orbit’s apsides rotate at a rate set by the low-order harmonics of the gravity field. For an eccentric ring whose apsides are locked together, the whole ring precesses at that rate — and the rate is measurable, because the ring’s orientation is determined afresh at every occultation.

Two rings at different radii precess at different rates, and the difference constrains the harmonics. With several eccentric and inclined rings spanning a range of radii, the second and fourth zonal harmonics can both be solved for.

That was done for Uranus through the early 1980s, from occultation data alone, and the values obtained were confirmed when a spacecraft flew past in 1986 and measured the field by tracking. The agreement was good, which is a striking thing: a planet’s internal mass distribution, inferred from the orientation of a set of ten-kilometre-wide bands of dark material, and checked against a Doppler measurement of a spacecraft’s trajectory.

There is a further quantity in the same data. The precession rate depends on the harmonics and on the planet’s rotation period, which for Uranus was uncertain by hours before the spacecraft arrived — so the ring measurement was a constraint on the rotation of a body whose surface shows no features to track.

A ring is a set of test particles at known radii, observed for decades, which is a better instrument for a gravity field than most spacecraft trajectories, and it costs nothing to keep observing.

What it is used for now #

Occultation prediction was historically the limiting step: the events are narrow, the shadow track across the Earth is a few thousand kilometres wide at most, and predicting where it falls requires the star’s position and the body’s ephemeris to better than the track width.

Both improved by orders of magnitude with all-sky astrometric surveys, and the field changed character as a result. Predictions that used to be uncertain by thousands of kilometres are now good to tens, so campaigns can be organised in advance, with portable telescopes deployed across the predicted path. Occultations by small trans-Neptunian objects — events lasting under a second, by bodies too faint to image — are now caught routinely, and they are the only source of sizes and shapes for that population. A single well-observed occultation gives a diameter to a few kilometres for an object whose size would otherwise be inferred from a brightness and an assumed albedo, which is a factor-of-two argument at best.

What the particles are, from two colours #

A single occultation at one wavelength gives an optical depth profile. Observing the same event at several wavelengths gives something more: a constraint on the sizes of the particles doing the blocking.

The physics is the same that governs any absorbing medium. Particles much larger than the wavelength block geometrically, so the optical depth is the same at every colour. Particles comparable with the wavelength diffract, and the extinction becomes wavelength-dependent in a way that depends on the size distribution.

For the Uranian rings the optical depth is very nearly grey across the visible and into the near infrared, which says the particles are large — centimetres to metres — with very little fine material. That is unusual: most ring systems contain a substantial population of dust, produced by collisions and by micrometeoroid bombardment, and the dust is what makes a ring visible in forward-scattered light.

The absence of dust is a dynamical statement. Fine material is removed quickly by drag against the planet’s extended atmosphere and by radiation pressure, so a dust-poor ring is either young or has an efficient removal mechanism. For Uranus the atmosphere extends far enough that drag is the likely answer.

Radio occultations, where the spacecraft’s own transmitter is the source, extend the same argument to centimetre wavelengths and constrain the largest particles directly. Combining optical and radio results for Saturn’s rings gives a particle size distribution running from centimetres to about ten metres, with a power-law slope near three — which is what a collisional cascade produces.

There is a further use of the same comparison that has nothing to do with particle sizes. A ring’s opacity depends on the angle at which the sight line crosses it, because a slanted path traverses more material — so comparing occultations at different ring-opening angles measures the ring’s thickness as well as its surface density. For Saturn’s main rings that comparison gives a vertical extent of tens of metres, which is the flattest structure of its size anywhere in the solar system: a sheet a hundred thousand kilometres across and as thick as a building.

Getting that number from any other kind of observation would require resolving a ten-metre feature at ten astronomical units, which is not a measurement anybody will make. What the occultation supplies instead is a ratio of two opacities at two geometries, and the thickness falls out of the ratio.

Two wavelengths turn a measurement of how much material there is into a measurement of what size it comes in, and neither observation on its own says anything about the second question.

Where the picture stops #

A single chord is a line, not a map. One station gives one cut across the system and cannot distinguish a circular ring from an arc, or a round body from an elongated one seen end-on. The Uranian discovery needed three sites and, in the end, many events over years to establish which rings were eccentric.

Diffraction is only invertible when the signal is coherent. For a radio occultation, where the source is a spacecraft transmitter, the fringe pattern can be undone by processing and the resolution improves by an order of magnitude. For starlight it cannot, and the Fresnel scale is a genuine floor.

And the profile is an optical depth along a slanted path. What the light curve measures is the integral of the extinction along the line of sight through the ring, which depends on the ring’s inclination to the sight line as well as on its actual density — so comparing two events at different geometries requires a model of how the particles are arranged, and at high densities that model is the largest uncertainty.

One more scale height covers an atmosphere far more extended than the essay’s own case.

Where this ladder goes next #

Later rungs on this anchor: the eccentric narrow ring and the self-gravity that keeps its edges aligned; shepherding, and how few moons are actually needed; the inversion of a diffraction pattern for a coherent source; the atmospheric occultation as a thermal profile, and the central flash that reveals an atmosphere’s oblateness; and the rings found around small bodies, which have no obvious shepherds and which the confinement arguments do not yet account for.