Show Summary Details

p. 604. The moons of giant planetsfree

  • David A. Rothery


The giant planets Jupiter, Saturn, Uranus, and Neptune each have an extensive entourage of moons: small inner moonlets, closest to the planets, mostly less than a few tens of kilometres in radius and irregular in shape; next are large regular satellites exceeding about 200 km in radius; and then there are the irregular satellites mostly less than a few tens of kilometres in radius. ‘The moons of giant planets’ describes these different types of moons, the space missions to find them, their orbital resonance and tidal heating, as well as the spectacular and complex rings and shepherd moons of Saturn and the other giant planets.

The giant planets Jupiter, Saturn, Uranus, and Neptune each have an extensive entourage of moons. Naturally, the first moons to be discovered were the largest, and are often called the ‘regular satellites’, but they are only part of the story. Not all the moons of giant planets can be neatly pigeonholed, but the overall situation is as follows.

Closest to the planet are small inner moonlets, mostly less than a few tens of kilometres in radius and irregular in shape. They are closely associated with the planet’s ring system and their orbits are circular, lie in the planet’s equatorial plane, and have radii less than about three times that of the planet itself.

Next are large regular satellites exceeding about 200 km in radius, which is large enough for their own gravity to have pulled them into near-spherical shapes, a condition described as ‘hydrostatic equilibrium’. Their orbits are only slightly less circular than those of the inner moonlets, and have radii up to twenty or thirty times that of the planet. These too lie pretty close to the plane of the planet’s equator.

Finally there are the irregular satellites, mostly less than a few tens of kilometres in radius. The term refers both to their irregularity in shape and to their orbits, which can be strongly p. 61elliptical and are usually considerably inclined relative to the planet’s equator. They extend to about 400 times the radius of Jupiter and Saturn, over 800 times the radius of Uranus, and nearly 2,000 times the radius of Neptune.

Inner moonlets and all regular satellites except for Neptune’s Triton travel round their orbits in the same direction that their planet rotates, which is described as ‘prograde’ motion. Most irregular satellites, as well as having inclined orbits, travel round their orbits in the direction opposite to their planet’s spin. This is described as ‘retrograde’ motion, and has implications for these moons’ origins.


Generally speaking, large moons are not rocky bodies. In the 1950s, telescopes fitted with spectrometers to measure the characteristics of reflected sunlight and trained on the larger moons of the giant planets began to reveal the presence of frozen water on most surfaces. This was not really surprising, because these bodies are a long way from the Sun, and mean surface temperatures are about −160°C for the moons of Jupiter, −180°C at Saturn, −200°C at Uranus, and −235°C at Neptune. At such extremely low temperatures, ice is mechanically very strong and behaves like rock; it can sustain craters, cliffs, and mountains without flowing like a glacier would on Earth.

Just as importantly, the temperature was also very low when these moons were forming. Jupiter, five times further from the Sun than is the Earth, lies beyond the ‘ice line’. Temperatures here were low enough to allow water to condense directly into ice from the gas cloud surrounding the young Sun. Bodies that formed beyond the ice line generally contain more ice than rock, because in the cloud of gas and dust from which the Solar System formed the elements required to make water (hydrogen and oxygen) were more abundant than the key ingredients of rock (silicon and various p. 62metallic elements plus oxygen). Where hydrogen could form solid compounds, it did so, so rock dominates only inside the ice line.

Carbon and nitrogen are common elements too, and these went to make up other varieties of ice that condensed further from the Sun. Possibly at Saturn, and certainly at Uranus and beyond, the ice is not just water but is mixed with frozen ammonia (NH3), methane (CH4), carbon monoxide (CO), and (at Neptune) even frozen nitrogen (N2). In the giant planets much of this, especially water, occurs as ice in their interiors below a thick gassy envelope mainly of hydrogen and helium, but the moons have too little gravity to have collected a lot of gas, so ices dominate.

The abundance of ice explains the low densities of most moons. The tables in the Appendix list densities of between 1,000 and 2,000 kg per cubic metre for most regular satellites of the giant planets. A rocky body should have a density of more than 3,000 kg per cubic metre, whereas water ice has a density of 1,000 kg per cubic metre (other ices are even less dense). Thus the lower its bulk density, the more ice and the less rock a moon contains.

Missions to moons

There would be much less to say were it not for space probes that have visited the giant planets and their moons. Exploration began with fly-bys (missions that flew past the planet) but has moved on to the stage of orbital tours in the case of Jupiter and Saturn, which have each had a mission that orbited the planet for several years and that was able to make repeated close fly-bys of at least the regular satellites. Close fly-bys of moons enable detailed imaging, and usually take the probe close enough to see how the moon affects the strong magnetic field surrounding the planet and to detect whether the moon also has its own magnetic field. The size of the slight deflection to a probe’s trajectory as it passes close to a moon enables the moon’s mass to be determined. Knowing the moon’s size, it is then easy to work out its density.

p. 63The story begins with NASA’s Pioneer 10 that flew past Jupiter in December 1973, and Pioneer 11 that flew past Jupiter in December 1974 and then Saturn in September 1979. These were concerned mostly with the planets’ atmospheres and magnetic fields, and collected little data about their moons.

It was NASA’s two Voyager probes that really opened our eyes to the moons. Voyager 1 flew through the Jupiter system in March 1979 and through the Saturn system in November 1980. In August 2012 it became the first space probe to cross the heliopause, where the solar wind fails, and to enter interstellar space. Voyager 2 made fly-bys of all four giant planets: Jupiter in July 1979, Saturn in August 1981, Uranus in January 1986, and Neptune in August 1989. It remains the only probe to have visited Uranus or Neptune.

NASA’s Galileo mission went into orbit around Jupiter in December 1995. After dropping an entry probe into Jupiter it toured the moons for eight years until it ran out of manoeuvring propellant and was allowed to crash into the planet. There was a serious early set back because its main communications antenna, a parabolic dish, failed to deploy. This meant that data had to be transmitted using the backup ‘low-gain’ antenna, reducing the total number of images that could be collected, but in-flight programming and data compression techniques rescued much of the science.

The joint NASA–European Space Agency (ESA) mission Cassini–Huygens arrived at Saturn in June 2004. It released the Huygens lander that parachuted to the surface of Titan in January 2005, while the Cassini orbiter began a long and complex orbital tour that is scheduled to end with entry into Saturn’s atmosphere in 2017.

Cassini flew past Jupiter in December 2000 on its way to Saturn, and for several days was able to complement the Galileo orbiter’s p. 64observations of volcanic eruptions on Io. More images of these spectacular events were provided by NASA’s New Horizons mission, which made a close pass by Jupiter in February 2007 on its way towards Pluto, which it flew past in July 2015.

Jupiter’s regular satellites

Jupiter’s four Galilean moons are the archetypal regular satellites. I consider them here as a family, reserving individual treatment for Chapter 5. They are shown together in Figure 9, cut away to reveal their internal structures. These were deduced mainly from clues to their internal density distribution achieved by Galileo fly-bys together with measurements by Voyager and Galileo of the interplay between each moon and Jupiter’s magnetic field. The latter shows that Jupiter’s magnetic field induces a field within Europa and Callisto, most likely achieved by electrical conduction in a salty internal ocean. Ganymede has a fairly strong magnetic field of its own, which may be generated by convection currents acting like a dynamo in a liquid iron sulfide outer zone of its core, as happens inside Mercury and the Earth. Io’s magnetic field has been less well characterized, and we cannot be certain whether it results from motion in a fluid core or is an induced field with a relatively shallow source. Three are differentiated bodies, in which the denser material has been able to segregate inwards to form a core, but Callisto lacks a strong internal density gradient, showing that it is only weakly differentiated.

9. Cutaway views to show the inferred internal structures of Jupiter’s regular satellites, shown to scale. Io, Europa, and Ganymede are shown with an iron-rich core, surrounded by rock. The outer layers of Europa, Ganymede, and Callisto are ice, and the dark layers inside are liquid water. Callisto’s interior is shown as an undifferentiated mixture of rock and ice.

Io is the densest moon in the Solar System and is the only regular satellite to lack surface ice. It can be thought of as a larger (and more active) version of our own Moon. It has a rocky surface, stained yellow and red by sulfur compounds distributed by ongoing volcanic eruptions.

Europa is a smaller (and less active) version of Io, buried by water, which is solid near the surface (ice) and liquid at depth where it forms a global ocean. Europa is nearly as dense as the Moon, and p. 65this shows that the shell composed of water (ice plus liquid) is only about 100 km thick. Europa’s surface ice is able to fracture and migrate relative to the interior, but neither Europa nor any other moon shows behaviour closely similar to the formation and migration of tectonic plates as on Earth.

Ganymede is the most massive moon in the Solar System and is the only differentiated Galilean moon in which the internal p. 66pressure is sufficient to compact H2O-ice into denser crystalline structures. Its surface ice is the normal sort, known as Ice I, but at greater depths the calculated pressure is enough to compress it into phases known as Ice III, then Ice V, and finally Ice VI. Figure 9 shows a complex internal structure proposed for Ganymede in 2014. According to this model, each ice phase has an underlying liquid layer, giving the appearance of a multi-layered sandwich. If these liquid layers exist, they are probably quite strong brines, containing salts dissolved out of the rocky interior that keep them liquid at temperatures too cold for pure water to melt.

The density of the Galilean moons decreases outwards from Jupiter. This is a strong clue to their origin, and suggests that they grew from a cloud of debris around the young Jupiter (in much the same way as the planets grew around the Sun), and that the heat radiated by Jupiter was sufficient to starve Io, and to partially starve Europa, of the water that was available to moons further out. Debris around Jupiter would have shared Jupiter’s rotation, which would naturally result in these having moons prograde orbits close to the planet’s equatorial plane, as indeed we see.

Orbital resonance and tidal heating

When the Voyager missions were being planned, the regular moons were expected to be fairly dull places. It was argued that, being largely icy and relatively small, the amount of heat-producing radioactive elements contained in any interior rock would be too little for there to have been any internally driven activity that could have affected their surfaces during the past three or four billion years. They would therefore be ‘dead’ worlds, heavily scarred by impact craters like the lunar highlands. Rock or ice, it doesn’t matter: impacts cause craters that look much the same.

However, it turns out that Io has a surface so young that no impact craters at all have been seen there, and Europa has very few. There are plenty of impact craters on Ganymede and Callisto, p. 67some of which you should be able to make out on Figure 9, but Ganymede has tracts of paler, younger terrain cutting across its surface. So, as well as an outward trend of decreasing density, the Galilean moons have an outward trend of increasing surface age.

The nearer moons are not actually any younger, but they have been resurfaced more recently by geological activity. The explanation lies in their orbits. Tidal friction has long since slowed down their spin, so they have synchronous rotation matching their orbital periods. It has also made their orbits become much more circular than the Moon’s, because they are orbiting a much more massive planet with stronger gravity.

In an elliptical orbit, libration would displace a moon’s tidal bulges to and fro about their mean position. The distortion of the interior to allow this to happen must add heat by means of internal friction, and would encourage the interior to become differentiated, if the even more powerful tidal heating before its rotation became synchronous had not already done so. If you want to experience the efficacy of internal frictional heating, try bending a wire coat hanger to and fro, and then (carefully!) touch the bent part to your lip.

If Jupiter had only one moon, the planet’s pull would have taken less than a hundred million years to force the moon’s orbit into an exactly circular shape, whereupon there would be no more libration and no tidal heating. However, as you know Jupiter has four regular satellites. While Voyager 1 was speeding towards Jupiter, the American Stan Peale (1937–2015) and colleagues were computing the effect that membership of a family of moons can have on tidal heating.

The orbital periods of Europa and Ganymede are exactly twice and four times that of Io, so that for every four circuits of Jupiter made by Io, Europa completes exactly two and Ganymede exactly one. This is a situation described as ‘orbital resonance’, and has p. 68been brought about by mutual gravitational interactions between the moons. This means that Io passes Europa at exactly the same point in its orbit every time, and Europa passes Ganymede at exactly the same point in its orbit. The repeated slight gravitational tug between moons has prevented their orbits from becoming exactly circular. If you saw them drawn on paper they would look like circles, but there is enough departure from circularity for the orbital speed to vary so that the tidal bulges migrate slightly to and fro, and swell and contract in height, heating their interiors in doing so.

Peale and his colleagues calculated the amount of heating that this process should produce inside Io. Having made the best assumptions they could about Io’s internal structure and strength, they published their findings in the 2 March 1979 issue of the journal Science, writing ‘dissipation of tidal energy in Jupiter’s satellite Io is likely to have melted a major fraction of the mass. Consequences of a largely molten interior may be evident in pictures of Io’s surface returned by Voyager 1.’ This is the most remarkable example of timing in planetary science known to me; images recorded by Voyager 1 six days later revealed erupting volcanoes on Io, and a total lack of surviving impact craters.

Orbital resonance is a complex situation. While in a state of resonance, the amount of forced eccentricity (and hence the rate of tidal heating) can wax and wane, and also moons can drift in an out of resonance over millions of years. Present-day tidal heating accounts for Europa’s young surface, and past episodes of tidal heating can be invoked for Ganymede and several regular satellites of other giant planets.

Other regular satellites

The regular satellite families of the other giant planets lack the clear outward trend of decreasing density seen at Jupiter, and, with the exception of Saturn’s Titan, the individual moons are p. 69smaller. However, the orbital characteristics at Saturn and Uranus are similar; almost circular orbits in their planet’s equatorial plane, even at Uranus where some catastrophe long ago tipped the whole system on to its side (Uranus’ rotation axis is tilted at 97.9° relative to its orbit round the Sun). At present there is orbital resonance between some of Saturn’s moons but none among Uranus’ moons, though some surfaces bear signs of tidal heating suggesting that resonances occurred in the past.

Saturn has seven regular satellites (see Appendix). Working outwards these are Mimas, Enceladus, Tethys, Dione, Rhea, Titan, and Iapetus. Mimas is in 2:1 resonance with Tethys, though this seems to have produced no effective tidal heating. Enceladus is in 2:1 resonance with Dione, and this must power the eruption plumes that Cassini discovered near Enceladus’ south pole that are discussed in Chapter 5.

Uranus has five regular satellites: Miranda, Ariel, Umbriel, Titania, and Oberon. Of these, Ariel and Titania are crossed by large fractures best attributed to ancient tidal heating. Miranda has a complex surface history, possibly the result of a previous 3:1 orbital resonance with Umbriel. It is slightly non-spherical (its radii are 240, 234.2, and 232.9 km) and some might regard it as Uranus’ outermost inner moonlet.

These all probably formed around their planets in a similar manner to that proposed for Jupiter’s Galilean moons, but the same cannot be said for Neptune’s largest moon, Triton (not to be confused with the similarly named Titan at Saturn). This has an inclined and retrograde orbit. There is no known way in which Triton could have formed alongside Neptune and ended up in such an orbit, so it was probably captured by Neptune from elsewhere.

Most likely Triton began as a member of the ‘Kuiper belt’, a family of icy bodies, from 1,500 km downwards in size, that orbit the p. 70Sun near Neptune’s orbit and beyond. If this is correct, Triton’s originally independent orbit round the Sun must have, on one occasion, brought it close enough to Neptune to become captured. Capture is very difficult to achieve, because there is too much momentum to dispose of. Usually the approaching body would just swing past the planet, or (rarely) collide with it. However, if the incoming object is actually double (a primary object and a moon), one can be captured and the other can be flung away faster than it arrived, with the total momentum of the system being conserved.

Such double objects can in fact be seen today in the Kuiper belt. Pluto and its large moon Charon are a prime example. Triton is slightly bigger than Pluto, but we do not know whether or not it was the larger or smaller component of the supposed double object that arrived at Neptune. Whether double or single, Triton’s arrival and capture would have scattered any pre-existing regular satellites, which must have been lost to the Kuiper belt or destroyed in mutual collisions.

Trojan moons

When a smaller body orbits a larger one, there are two stable points where an even smaller body can reside or about which it can oscillate. These occur 60° ahead and 60° behind the orbiting body, sharing the same orbit about the larger body. Technically these are called Lagrangian points, and are designated as L4 (ahead) and L5 (trailing). However, they are often referred to as the ‘Trojan points’ because there are groups of asteroids named after characters from the Trojan war that share Jupiter’s orbit round the Sun, clustered around its L4 and L5 points.

Saturn has four small moons in leading and trailing Trojan relationships with two of its regular satellites, Tethys and Dione. Calypso, Telesto, and Helene, which are between 10 and 20 km in radius, were discovered telescopically in the 1980s and 1990s. The p. 71smallest, Polydeuces, which is less than 3 km across, was discovered on images taken by Cassini in 2004.

Images from Cassini (Figure 10) have revealed some surprising aspects to the surfaces of these Trojan moons. They have rather few impact craters, and so may be relatively young (this could still mean they are more than a billion years old). The figure shows the side of Helene that faces away from Saturn and its rings; the Saturn-facing side has more craters. Calypso’s surface is the most reflective in the entire Solar System, but this is not apparent in Figure 10 as the images have been processed to show each surface equally well. This may be a result of Calypso sweeping up ice crystals erupted from Enceladus. Both Calypso and Helene show p. 72curious gullies on their surfaces. There is no likely liquid that would be stable under their surface conditions. Some kind of dry avalanche process may be responsible, but it is a mystery how this would work on a body whose surface gravity is only 0.02 per cent as strong as the Earth’s.

10. The Solar System’s four known Trojan moons, imaged by Cassini and shown at approximately correct relative scale. Helene has a mean radius of 17 km. Polydeuces has never been seen in sufficient close up to reveal surface details. Helene and Polydeuces are Trojan co-orbitals of Dione, whereas Telesto and Calypso are co-orbitals of Tethys.

Saturn is the only planet known to have Trojan moons, but this may be a result of observational bias. The Galileo orbiter didn’t search for them at Jupiter, which could have small ones like Polydeuces. Voyager 2 didn’t have chance to look for any at Uranus and Neptune, where even Calypso-sized Trojans would be hard to spot telescopically from Earth.

Irregular satellites

There are more known irregular satellites than any other class of moons. Jupiter has seventy-one. The inner seven are in prograde orbits up to 238 Jupiter radii (Jupiter’s equatorial radius, 71,492 km) in size, and all but one of the others are all in retrograde orbits extending out to 400 Jupiter radii. This is a consequence of the differential long-term stability of orbits relative to the size of a planet’s Hill sphere (the range out to which the planet’s gravity outcompetes the Sun’s gravity). Prograde orbits are stable over billions of years out to only about half the Hill sphere radius, whereas retrograde orbits can be stable out to about two-thirds of the Hill sphere radius. Jupiter’s Hill sphere is about 740 Jupiter radii in size.

Jupiter’s largest and third-closest irregular satellite, Himalia, is 85 km in radius and was discovered as long ago as 1904, but most have been found by dedicated telescope surveys since the year 2000. The smallest are only about 1 km in radius. Even Himalia is only seven pixels across in the best Galileo image, so little is known of any of them.

Three other prograde irregular moons have orbits similar to Himalia (at about 160 Jupiter radii), and they all reflect sunlight p. 73in the same way as carbonaceous asteroids, so it is suggested that all four are fragments of a carbonaceous asteroid that broke up on capture by Jupiter.

Three groups defined by common reflectance characteristics and similar orbital radii, eccentricities, and inclinations are recognized among Jupiter’s retrograde irregular satellites. They are each named after their largest member: the Ananke group has orbits near 297 Jupiter radii, the thirteen-strong Carme group near 327 Jupiter radii, and the seven-strong Pasiphae group near 330 Jupiter radii. These are believed to be fragments of asteroids of other types. Many of Jupiter’s irregular satellites, including more than twenty beyond the Pasiphae group, have no known families. Each could be either a small captured asteroid or comet nucleus.

Quite when these moons were captured is unknown. It would have been much easier to achieve very soon after Jupiter had formed, because it could then have had an extended diffuse atmosphere to provide the necessary drag to slow the incoming objects down sufficiently to become captured. Nor is it clear when the parent objects for each related group broke up. It could have been during the capture process, or afterwards because of a collision.

None of Jupiter’s irregular satellites shows synchronous rotation. They are too small and too far from Jupiter for tidal forces to be effective, and this is borne out by the few examples where rotation periods have been measured and shown to be only a few hours, in contrast to orbital periods of hundreds of days. On the other hand they feel the Sun’s pull so strongly that the shapes and inclinations of their orbits can vary markedly in only a few years.

The irregular satellites of the other giant planets follow a similar pattern to Jupiter’s, and their origins are probably similar. The most distant moons are all retrograde, but closer to the planet p. 74prograde and retrograde moons are intermingled rather than being neatly segregated like they are at Jupiter.

Saturn has thirty-nine known irregular moons counting Hyperion (Figure 11), a 180 × 133 × 103 km radius moon that has an orbit p. 75between those of the outer two regular satellites, Titan and Iapetus. Hyperion is unique among known moons in that its rotation is chaotic. Not only is its rotation period variable, but even its axis of rotation changes as it tumbles along. Its density is only half that of solid ice, and it probably has a porous, rubbly interior. It has a low albedo, suggesting a surface dusting of dark particles which is typical of this part of the Saturnian system.

11. Hyperion (above) and Phoebe (below) seen at similar scales.

Phoebe (Figure 11) is 109 × 109 × 102 km in radius and is the largest and closest of Saturn’s retrograde irregular moons. Phoebe’s orbit lies at 548 Saturn radii. This is too far out to be visited by Cassini after it had achieved orbit, so Cassini’s approach to Saturn was timed to enable it to make a close pass by Phoebe on its way in, at a range of 2,000 km, making it the best-imaged example of all its kind. Cassini revealed a cratered surface and detected water ice, carbon dioxide ice, and clay minerals. Phoebe has an extremely low albedo, only 0.06, which may be because methane ice has been stripped of some of its hydrogen (long exposure to the Sun’s ultraviolet radiation can do this), allowing the carbon atoms to link together as a black tarry goo. Phoebe is a good candidate to be a captured centaur—a class of icy asteroid found mostly beyond the orbit of Saturn. In 2009 infrared telescope observations revealed that Phoebe orbits within a diffuse but very broad (twenty times thicker than Saturn itself) belt of dust, thought to have been knocked off Phoebe’s surface by micrometeorite impacts.

Saturn has two orbital groups of prograde irregular satellites. The individuals in one are given Inuit names such as Siarnaq, and in the other Gallic names such as Albiorix. Each group could be the remains of a larger moon destroyed by collision. Apart from Phoebe, Saturn’s retrograde irregular satellites have Norse names, and include groupings that could each represent fragments of the same captured asteroid.

Irregular moons of Uranus and Neptune are challenging objects even for the best modern telescopes. Uranus has nine, discovered p. 76in the period 1997–2003. They have retrograde orbits except for Margaret, which has the most eccentric orbit of any planet’s moon. The largest is Sycorax, about 75 km in radius, whereas the smallest known have radii of about 10 km. There are no close orbital groupings, and each is probably an individually captured object.

Only six irregular moons of Neptune are known: three prograde and three retrograde. The largest, Nereid, has a radius of 170 km. It was discovered in 1949, and the others in 2002–3. The outermost examples, Psamathe (20 km radius) and Neso (30 km radius), are in eccentric retrograde orbits at mean distances of 1,885 and 1,954 Neptune radii. This is a vast distance (they take over 9,000 days to orbit the planet), but these orbits are stable because Neptune’s Hill sphere is larger than Jupiter’s, thanks to its greater distance from the Sun.

Nereid may be a large surviving remnant of a regular satellite that was catastrophically destroyed (maybe in the Triton capture event). It has been shown to have water ice on its surface, with an imposed low albedo thanks to a darkening agent such as carbon, in which respect it resembles some of the regular satellites of Uranus. It has been proposed that one other irregular satellite, Halimede, could be a smaller fragment from the same body, but Neptune’s other irregular moons are probably individually captured objects.

Inner moonlets

Inner moonlets can be very small, as their name implies, and their proximity to the glare of their planet makes them harder to detect by telescope than irregular satellites. There is a good reason why large moons are not found close to their planets, articulated by the French astronomer Édouard Roche (1820–83) who calculated the distance from each planet at which the difference between the planet’s tidal pull on the moon’s near p. 77and farsides would exceed the moon’s own gravity. At this distance, commonly referred to as the ‘Roche limit’, a fluid or loosely consolidated body would be pulled apart, though the internal strength of a solid body allows it to approach closer before it disintegrates.

Most inner moonlets orbit within their planet’s Roche limit, and are probably fragments of larger moons ripped apart by tides. Some of the more distant, and larger, examples may have originated as regular satellites battered by collisions.

Only four inner moonlets of Jupiter are known, of which only the largest, Amalthea, had been discovered before Voyager. The best Galileo images of these are included in Figure 12.

12. A compilation of the best-imaged inner moonlets, shown at approximately correct relative scales (Pandora is 100 km from end to end). For Jupiter (Galileo images) and Saturn (Cassini images), moonlets are arranged with closest to the left. Two views of Atlas are included: equatorial (above) and southern hemisphere (below). Puck and Proteus images are from Voyager 2.

Saturn has eight known inner moonlets within the orbit of Mimas, and there are three others (only about 1 km in radius) between Mimas and Enceladus. Of those three, only Methone has been seen at close quarters by Cassini, and has a surprisingly smooth egg shape. The best Cassini images of Saturn’s conventional inner moonlets are included in Figure 12, and most show much better detail than provided by Galileo at Jupiter. Saturn’s closest inner moonlets, Pan and Atlas, have ridges running round their equators, giving them curious ‘flying saucer’ shapes. Their orbits follow gaps in Saturn’s ring system, and the equatorial ridges are probably swept up ring material.

Janus and Epimetheus are a unique pair: the orbit of one is only a few km bigger than the other. When the inner, faster travelling, moon catches up with the other, which happens about every four years, their mutual gravitational interaction causes them to swap orbits. The previously faster one is now in the wider orbit so it slows down, whereas the previously slower one speeds up until it catches up with its fellow four years later and the cycle repeats.

Uranus has thirteen known inner moons, of which only Puck (Figure 12) was seen in any detail by Voyager 2, which discovered p. 78it. They are all dark (low albedo) objects, probably as a result of radiation damage to methane. Most were discovered by Voyager 2, but three (including the smallest, Cupid and Mab, which are only about 5 km in radius) are more recent telescopic discoveries. All thirteen are crowded into a narrow orbital range, 1.95 to 3.82 p. 79Uranus radii, and simulations suggest that they can disturb each other’s orbits so that there is likely to be a collision sometime in the next 100 million years.

Six inner moonlets of Neptune were revealed on Voyager 2 images, and a seventh (the smallest, at about 18 km radius) was discovered by imaging with the Hubble Space Telescope. Like their equivalents at Uranus, they have low albedo, and seem to be water ice made dark by radiation-damaged methane. One of these, Proteus (Figure 12), is the largest inner moonlet in the Solar System, at 220 × 208 × 202 km radius. Its shape is a long way from hydrostatic equilibrium and probably results from collisional battering. The innermost of Neptune’s moonlets, Naiad, orbits well inside the Roche limit, and its eventual fate will probably be either to spiral into the planet’s atmosphere or to be ripped apart by tides and make a new ring.

Rings and shepherd moons

Saturn is famous for its spectacular ring system. This was first seen by Galileo, but he couldn’t make it out clearly. Huygens (in 1655) was the first to correctly interpret what he saw. The other giant planets have rings too, though these contain much less mass and so are far less spectacular.

Even Saturn’s glorious rings contain less mass than Mimas, its smallest regular satellite. They are probably very old, and represent mostly either material that was too close to the young planet to have ever been gathered into a moon, or the remains of a moon that was ripped apart when it strayed within the Roche limit. The easily visible extent of the rings (which you can see for yourself with a small telescope) consists, working outwards, of rings designated C, B, and A by the IAU. These are only a hundred metres thick and extend from about 75,000 km to about 137,000 km from the planet’s centre. That’s 1.24 to 2.27 Saturn radii, but of course only 0.24 to 1.27 Saturn radii above the cloud p. 80tops. Spectroscopy shows that these rings are mostly water ice, darkened and in most places reddened by radiation damage or dusty contaminants. The relatively slow cooling rate of the rings when they pass into Saturn’s shadow, the ability of ground-based radar to record a signal reflected from the rings, and the effect of the rings on Voyager 1’s radio transmissions show that the rings are made of chunks ranging from about one centimetre to five metres in size. Each such chunk is in orbit about the planet. It would be perverse to regard every one of them as a moon, though there is no agreed lower size limit for what can be called a moon.

Less substantial rings occur to either side of the main rings. The D ring (66,900–74,500 km) lies between the C ring and the planet, whereas other rings lie beyond the A ring. Saturn’s rings have amazingly complex internal structures, most obviously at distances from Saturn where the ring material becomes more diffuse or even vanishes completely. The widest ‘gap’ is 4,800 km wide and separates the A and B rings. It was discovered in 1675 by Giovanni Cassini and known as the Cassini Division. Its inner edge occurs at the distance where any ring particles would have twice the orbital period of Mimas, and so is readily explicable. Fine structure within the gap, at radii where particles are either concentrated or dispersed, is not fully understood.

Numerous minor gaps elsewhere in the rings defy simple orbital resonance explanation, but some are swept clear by inner moonlets whose orbits actually follow the gaps—for example, Pan and Daphnis each sweep a different gap in the A ring (Figure 13).

13. Moons among the rings. Top: Prometheus, 176 km long, travelling from right to left, making periodic disturbances in Saturn’s F ring. Middle left: Pan, 34 km long, orbiting in the Encke gap, casting long shadow thanks to grazing-incidence sunlight. Middle right: Daphnis, 8 km long, orbiting in the Keeler gap, making waves in the inner edge of the A ring. Bottom: Uranus’ rings with two moons circled, Cordelia (left) and Ophelia (right). Each is about 20 km in radius, and they appear artificially elongated in this long-exposure image. There is a lot of speckly noise, and the dark band either side of the outermost ring is an artefact.

Cassini images in 2013 showed a 1,200 km long, 10 km wide enhancement of material at the outer edge of the A ring, which might be concentrated around a small mass, perhaps 1 km across. This could be a new moonlet in the making, but is more likely a temporary disturbance that will disperse.

p. 81p. 82The F ring is only 30–500 km wide and lies outside the A ring. This is held in place, and given a dynamic structure, by the inner moonlets Prometheus (Figure 13) and Pandora that orbit just inside and just outside it, respectively. Inner moonlets such as these, whose intimate association with a ring system helps to define its shape, are often referred to as ‘shepherd moons’.

Beyond the F ring there are various tenuous rings and ring arcs believed to be dust knocked off the outermost inner moonlets, Aegaeon, Methone, Anthe, and Pallene, by micrometeorite bombardment, and confined or concentrated into arcs by resonance with Mimas. Further out still, at 180,000 to 480,000 km, is the diffuse E ring consisting of particles less than a micrometre across made of water ice and traces of other material erupted from Enceladus. The belt of diffuse dust that envelopes Phoebe’s orbit is sometimes referred to as an additional ring, but is far too thick for the term to be appropriate. A suggestion dating from 2008 that Saturn’s regular satellite Rhea is encircled by its own diffuse ring system has now been discounted. In fact, no moon is known to have rings, just as no moon has a moon.

Jupiter’s rings were discovered by Voyager 1. The main ring is only 6,500 km wide and 30–300 km thick. The orbit of the innermost inner moonlet, Metis, occupies a gap within this ring, and the ring’s outer edge is shepherded by the orbit of Adrastea. The ring material is an unknown reddish material, consisting of a mixture of micrometre-sized dust and larger chunks, with a total mass no more than about five times that of Adrastea and possibly much less. It would take less than 1,000 years for radiation pressure and interactions with Jupiter’s magnetic field to disperse the dust, so unless it is extraordinarily young it must be being replenished either by collisions between the larger chunks, or by micrometeorite bombardment of Metis and Adrastea.

NASA’s New Horizons mission searched in vain for unknown inner moonlets within this ring during its 2007 fly-by, so there are p. 83unlikely to be any greater than 0.3 km in radius. It did however find seven clumps of material, occupying arcs extending for 1,000 to 3,000 km.

Inward of the main ring, Jupiter has a 12,500 km thick ‘halo ring’, which seems to be dust spiralling inwards to Jupiter from the main ring. There are also two exceedingly faint ‘gossamer’ rings, which are dusty features extending outwards from the orbits of Amalthea and Thebe, the outermost inner moonlets, which are probably supplied by ejecta from micrometeorite bombardment of their surfaces.

The rings of Uranus were discovered in 1977, by telescopic observers who noted the repeated dimming of a star as the otherwise invisible rings passed across it. Most of our knowledge of the rings comes from Voyager 2 and the Hubble Space Telescope. Thirteen rings are now known, with a total mass exceeding Jupiter’s rings but much less than Saturn’s. Several are narrow and consist mostly of boulder-sized chunks of low-albedo reddish material, believed to be produced by the fragmentation of inner moonlets. These are all narrow rings, the outer and brightest of which (the epsilon ring) is shepherded by the moonlets Cordelia and Ophelia (Figure 13). It is at a mean distance of 51,150 km from the planet’s centre, but is eccentric in shape so that the distance varies by 800 km around its circumference. It is 20 km wide where closest to the planet and 100 km wide where furthest. The inner rings are even more tightly confined but have no known shepherd moons, prompting the suggestion that these formed by fragmentation no more than about 600 million years ago.

Uranus’ other rings are dusty. There are four inside the narrow rings, believed to be short-lived dust spiralling towards the planet, and two beyond—the outermost of which is spread round the orbit of the outermost inner moonlet Mab, which is probably the source of its dust.

p. 84Neptune has five rings, which are extremely dark, like those of Uranus. They are about half dust and half larger chunks. Material in the outermost ring is concentrated into several arcs. Attempts to explain this as a 42:43 resonance with the inner moonlet and ring shepherd Galatea have failed, and it is clear that there is much about rings and their relationship with moons that we do not yet understand.