Rings and a large family of satellites are features common to all four giant planets. There are variations in emphasis and scale, but the similarities between each ring-satellite system outweigh their differences.

Ring-satellite systems

Most outer satellites of each giant planet travel in eccentric orbits,usually in the opposite direction to the spin on their planet.Furthermore, many of these orbits are inclined at 〉30° relative to their planet’s equator. The typically eccentric, retrograde, and inclined nature of their orbits earns these bodies the name‘irregular satellites’, quite apart from the fact that (being at most about 100 kilometres, and more often only a few kilometres,across) they have far too little gravity to pull themselves into spherical shapes. The irregular satellites are the most numerous class: at the last count, Jupiter had 55, with orbital semi-major axes ranging from 105 to 400 Jupiter radii; Saturn had 38, with orbits from 184 to 417 Saturn radii; Uranus had 9, 167–818Uranus radii; and Neptune had 6, 223–1,954 Neptune radii.The ‘regular satellites’ are the large ones in near-circular prograde orbits, much closer to their planets, and with very low inclinations. Jupiter has 4 (the ones discovered by Galileo) whose orbital semi-major axes range from 5.9 to 26.3 Jupiter radii.These are substantial worlds, and geologically have much in common with the terrestrial planets, though of course they do not satisfy the IAU defi nition of a planet. Saturn has 8 (all but one considerably smaller than Jupiter’s, and orbiting at 3–59 Saturn radii), and Uranus has 5 (at 5–23 Uranus radii). Neptune has one large satellite, Triton, orbiting at 14 Neptune radii, that would be regarded as ‘regular’, except for its retrograde orbit. An important characteristic, shared by all regular satellites (including Triton), is that tidal forces have such a grip on them that they are in synchronous rotation, rotating once per orbit, so that (like the Earth’s Moon) they keep the same face towards their planet.

Closer still, we fi nd irregular-shaped lumps of debris that it is convenient to distinguish as ‘inner moonlets’. These have circular,prograde, equatorial orbits. So do the particles that make up the rings, and, given that some inner moonlets’ orbits lie within the rings, there is probably no fundamental difference between a large ring particle and a small inner moonlet. Jupiter has only 4 known inner moonlets, but Saturn has 14, counting 7 whose orbits lie among those of its innermost regular satellites. Uranus has 13 and Neptune 6.

The width and number of rings varies from planet to planet,Saturn’s being by far the most spectacular, but in general their thickness is no more than a few tens of kilometres. Mostly, they are closer to their planet than a distance known as the ‘Roche limit’, a boundary within which any large body should be ripped apart by tidal forces. Most rings are regarded as debris left over from the tidal disruption of a satellite or comet that strayed too close to the planet, but some less substantial rings are demonstrably supplied from nearby satellites by particles vented actively into space or thrown up by impacts.

Saturn’s rings are made of ice and refl ect about 80% of the sunlight falling on them. Despite their prominent appearance(Figure 3), the material in them would suffi ce only to make a body about 100 kilometres in diameter if it could all be gathered together. Although individual ring particles have not been imaged directly, the rate at which the rings cool when the shadow of their planet falls across them shows that Saturn’s rings are mostly particles between about a centimetre and 5 metres in size. In contrast, Jupiter’s much less substantial rings are made largely of micrometre-sized particles that are also much less refl ective than the bright icy lumps in Saturn’s rings. The ring-material at Uranus and Neptune refl ects sunlight poorly (like Jupiter’s ring-material)but is mostly centimetres to metres in size (like Saturn’s ringmaterial).

19.A 5,000-kilometre-wide view of part of Saturn’s ring system,seen by Cassini on 27 July 2009. At this scale, the curvature of therings round the planet (out of view to the right) is scarcelydiscernible. The rings refl ect most sunlight where particles are mostdensely packed, and black space shows through in particle-free gaps.Pan, a 28-kilometre-diameter shepherd satellite, can be seenorbiting in the widest gap. As well as sweeping most of this gap clear,Pan infl uences narrow and discontinuous rings within the gap. Theexceptional length of Pan’s shadow on the rings to its right is becausethis image was recorded when the Sun lay very close to the plane ofthe rings

Orbital resonances lead to a complex gravitational interplay between rings and the inner moonlets that orbit among them( Figure 19 ). Those are often called ‘shepherd satellites’ because some sweep clear many of the gaps in the rings, and others form,deform, and maintain narrow rings with orbits just within or just beyond their own.

In general, rings occur closer to their planet than the regular satellites, but Saturn is an exception in that it also has a diffuse outer ring of dark, dusty material centred around the orbit of Phoebe, one of its innermost irregular satellites. The material in this ring, which was discovered in 2009 using a space-based infrared telescope, is presumably being supplied from Phoebe in some way yet to be understood.

Remarkable satellites

There was a time when pretty much everyone expected even the largest of the outer planets’ satellites to be dreary objects. Ancient ice-balls, heavily pock-marked by impacts, they would record the outer Solar System’s bombardment history, but would be of no further interest unless you wanted to study mutual orbital evolution. That was the standard view until 2 March 1979, when Stanton Peale, working at the University of California, published(with two colleagues) a paper pointing out that the exact 2:1orbital resonance between Jupiter’s innermost Galilean satellites,Io and Europa, ought to result in so much tidal distortion of Io’s shape that its interior should be molten. From estimates of density plus spectroscopic analysis of their surfaces, it was already known that Io has a rocky crust, unlike the other satellites that are predominantly ice. To suggest a molten interior inside a rocky body (where the melting temperature is so much higher) was a particularly bold step. Few might have believed this claim if Voyager 1 had not fl own past a few days later and transmitted pictures of explosively erupting volcanoes, topped by300-kilometre-high eruption plumes.

Although tidal heating of Io is by far the strongest, the same process affects various other satellites, and many more bear signs of ancient episodes of tidal heating. This makes them varied, and intriguing to geologists. They don’t mind that in most of them only the core is made of rock, surrounded by a thick mantle of ice,with perhaps a chemically distinct icy crust at the surface. Under the low surface temperatures prevailing in the outer Solar System(ranging from –140 °C for Jupiter’s satellites down to –235 °C for Neptune’s satellites), the mechanical properties and melting behaviour of the ice are very closely analogous to how rock behaves in the inner Solar System. In other words, those bodies have both the behaviour and structure of a terrestrial planet, with rock in place of iron in the core, and ice instead of rock in the crust and mantle.

Io is an exception in being ice-free with a rocky crust and mantle surrounding an iron core, and would be classifi able as a terrestrial planet if it was orbiting the Sun instead of Jupiter. Europa is a hybrid, having a structure like Io buried below 100–150kilometres of ice. Here I describe both of those and some of the other satellites that fascinate me most, concentrating on the more active examples, though even the crater-pocked ice-balls have turned out to be more interesting than the dull globes formerly imagined.

Io

Io is only slightly bigger (3,642 kilometres in diameter) and denser than our own Moon, but the two could hardly be more different. Io’s terrain is resurfaced so rapidly by volcanic processes that not a single impact crater is to be seen, despite the fact that the effect of Jupiter’s gravity in focusing stray projectiles inwards must mean that Io is struck more often than Jupiter’s heavily cratered satellites Ganymede and Callisto that orbit beyond Europa. When the fi rst Voyager 1 colour close-ups of Io were studied in 1979, its yellow hue led many to suppose that the lobate lava fl ows that could be recognized on its surface were made of sulfur. However, it is now accepted that Io’s volcanism is molten silicate material (true ‘rock’). The temperatures recorded in the heart of erupting volcanic vents are well in excess of 1,000 °C,despite the intense cold beyond the active areas. The gas that escapes to drive explosive eruptions such as those in Figure 20 is mostly sulfur dioxide (whereas on Earth, it would be mostly water vapour), and both sulfur and sulfur dioxide condense on the surface as ‘frost’ that imparts the colour to Io.

Io lies within a belt of charged particles confi ned by Jupiter’s magnetic fi eld. The radiation there is so intense that NASA’s Galileo mission controllers did not allow the spacecraft to make repeat close fl y-bys of Io, so only a small fraction of Io’s globe was imaged well enough to show details below a few hundred metres in size. On the most detailed images, the pixels are only10 metres across, and even on those no impact craters have been found.

If Io’s present-day rate of volcanism is representative of the long term, then its entire crust and mantle must have been recycled many times over. Covering of older surfaces by lava fl ows and fall-out from eruption plumes, amounting to a globally averaged burial rate of a couple of centimetres per year, obscures impact craters too rapidly for any to be apparent. If Io ever had an outer layer of ice, volcanic activity has long since vaporized it, allowing it to be lost to space, because Io’s gravity is too weak to hold on to water vapour or other light gas. What a fantastic place for a volcanologist to visit, if only the harsh radiation environment did not make Io’s surface so thoroughly inimical to human exploration.

20.Top: part of the crescent view of Io seen by the Pluto-bound NewHorizons mission passing Jupiter in March 2007. The plume from avolcanic vent at a site called Tvashtar caldera on the night-side rises300 kilometres so that its upper part is in sunlight. An incandescentglow can be seen at its source, and the shadowed lower part of theplume is faintly illuminated by light refl ected from Jupiter. Bottom:A 250-kilometre-wide view of Tvashtar seen eight years earlier by theGalileo orbiter. Sunlight is from the left. The darkest material is recentlava fl ows, and the east–west bright streak near the upper left isincandescent lava being erupted from a volcanic fi ssure

Europa

Europa (3,130 kilometres in diameter) is my favourite. Voyager images from fl y-bys in 1980 and 1981 showed its surface looking like a cracked eggshell, with very few impact craters to be seen.Clearly, tidal heating was somehow refashioning Europa’s icy outer layer, though not so rapidly as Io. Higher-resolution imaging by the Galileo mission revealed a complex surface history, and led to an unusually bitter controversy. It was already well known that Europa’s surface is predominantly water-ice, and the globe’s overall density shows that its icy carapace has to be about 100–150kilometres thick, overlying a denser, rocky interior. However,density arguments cannot distinguish between solid ice and liquid water. The surface ice is strong and brittle, thanks to its low temperature. The controversy that emerged was over the state of the ‘ice’ below the surface. Was it frozen all the way down to the rock or was the lower part liquid, capped by a fl oating ice shell?

The latter requires a greater rate of internal tidal heating coupled with the exotic concept of a global ocean of liquid water below the ice. So far as I am concerned, evidence from images such as Figure 21 makes it clear that the ice is generally thin, only a few kilometres, and so must be fl oating on water. However, for several years of Galileo ’s orbital tour of the Jupiter system, a powerful lobby group on the imaging team persisted in trying to explain the surface features as a result of processes driven by solid-state convection in the thick ice layer.

What is now the generally accepted basis of Europa’s geology is best explained by reference to Figure 21 . This shows numerous high-standing ‘rafts’ of ice, bounded by 100-metre cliffs. The surfaces of the rafts are characterized by a pattern of ridges and grooves, running in a variety of directions. Between the rafts, the texture is more jumbled, and lacks a clear pattern. There are large expanses of Europa (beyond this region) that have not been broken into rafts and where the surface pattern is uninterrupted ridges and grooves. The rafts in Figure 21 are clearly broken fragments of this sort of terrain. The ridge and groove pattern is caused by the opening and closing of cracks, probably on a tidal cycle coincident with Europa’s 3.6-day orbital period. Globally,only a few cracks would be active at any one time. When an active crack is opened (to a width of perhaps just a metre or so), water is drawn up from below. The water temporarily exposed to the cold vacuum of space at the top of the crack simultaneously boils and freezes, but pretty soon becomes covered by slush. When the crack closes, some slush is squeezed out onto the surface, forming a ridge above the closed crack. The next time the crack opens, the ridge is split, and is added to by more slush when the crack closes again. A few years of opening and closing would suffi ce to surround central grooves by ridges of the size we see. Eventually,each crack seals permanently, but a new crack will begin to operate somewhere else, and so the pattern is built up, giving the ridge and groove terrain covering much of Europa, an appearance that has been likened to that of a ball of string.

21.A 42-kilometre-wide close-up of part of the Conamara Chaosregion of Europa, where ‘melt-through’ from the underlying ocean hasallowed rafts of ice to drift apart before the area refroze. Sunlight isfrom the right

In Figure 21 , ‘ball of string’ terrain has been disrupted by the other great process that affects Europa. This is ‘melt-through’, and results in a jumbled mixture of broken ice rafts described as‘chaos’. Under a future chaos region, the ocean becomes unusually warm – maybe because of silicate volcanic eruptions on the ocean fl oor – and the base of the surface shell of ice gradually melts, so the ice becomes thinner. Eventually, melting reaches the surface,and rafts (or fl oes) of ice break away from the exposed edges of ice shelf and drift into the exposed ocean. Any exposed water would refreeze pretty quickly, and perhaps it is better to think of kilometre-thick ice rafts nudging their way into a sea covered by icy slush, rather than into truly open water like the summer thaw of pack ice in the Earth’s Arctic ocean. In the north-west part of Figure 21 , you can see the way many of the rafts originally fi tted together, because they have not drifted far apart and their ‘ball of string’ textures can be matched.

After the temporary heat excess dies away, the ocean refreezes and the rafts cease to drift. The ice of the refrozen sea surface and beneath the rafts begins to thicken again. When the refrozen area is suffi ciently thick and brittle, new cracks may open, and a new generation of ‘ball of string’ texture begins to overprint the whole region. In Figure 21 , there is a young crack, fl anked by a narrow ridge on either side, running diagonally across. It looks unremarkable where it crosses rafts, but you can tell that it must be young because it cuts the refrozen sea lying between the rafts.

If this story is even remotely correct, then there are some very thought-provoking implications. Chemical reactions with the underlying rock would make the ocean salty – though the most abundant dissolved salt might be magnesium sulfate rather than sodium chloride as in Earth’s oceans. Any such ocean overlying tidally heated rock provides a habitat for life equivalent to where life is believed to have begun on Earth. Lack of sunlight is no hindrance, because the ‘primary producers’ at the base of the local food chain would derive their energy from the chemical imbalance supplied to the ocean at submarine hot springs (hydrothermal vents). Such life is described as chemosynthetic, as opposed to photosynthetic. On Earth’s ocean fl oors, the hottest vents are called ‘black smokers’ because of the plume of metal sulfi de particles that forms when the vent fl uid mixes into the ocean water. These vents are surprising oases of life, where communities of organisms (including some as advanced as shrimps and crabs)feed on chemosynthetic microbes that gain their energy by converting carbon dioxide into methane. If life on Earth began in such a setting, why not also on Europa?

Life sealed below ice that is normally kilometres thick would be extremely challenging to fi nd, requiring Europa landers to drill or melt a hole through the ice in order to launch a robotic submarine probe that could home in on a ‘black smoker’ plume. However,such an ambitious mission may not be necessary if the ridges either side of a young crack are built of slush squeezed up from the ocean. While a crack is open, it could provide a niche for photosynthetic life such as plants or (more reasonably) marine algae. Like life on Earth, these could have evolved from chemosynthetic forebears. Radiation would render the top few centimetres of the exposed water column uninhabitable, but there would be enough sunlight for photosynthesis in the next few metres. If there are primary producers (plants, algae) living off sunlight, there could be animals feeding on them. To fi nd out, the fi rst step is to investigate a sample from the ridge squeezed out of a crack.

The next big NASA-ESA collaboration in outer Solar System exploration is likely to be a mission to the Jupiter system. Its primary goal will be to verify the existence of Europa’s ocean,using ice-penetrating radar and by measuring the amount of tidal fl exing (which would be only about a metre in the case of ‘thick’ice resting on bedrock, but about 30 metres for a ‘thin’ ice shell fl oating on an ocean). Sadly, a lander cannot be contemplated yet,but there will at least be high-resolution spectroscopy from orbit to look for biogenic molecules in the ridge material.

Enceladus

It would be much easier to fi nd biomarkers if Europan ice could be sampled without having to go down to the surface. Enceladus,a satellite of Saturn, offers just that opportunity. It is only 504kilometres in diameter and has too low a density to contain much rock. Voyager showed it to be a strange little world, heavily cratered in parts but elsewhere apparently lacking in craters. The higher-resolution images transmitted by Cassini , which began an orbital survey of the Saturn system in 2004, shows a surface cut by many families of cracks (though rather unlike the ‘ball of string’regions of Europa). It also discovered jets of icy crystals venting to space from cracks near the south pole ( Figure 22 ). Fortunately,Cassini carried a mass spectrometer intended for study of ions and neutral particles, so the spacecraft’s trajectory was adjusted to allow it to fl y through the plume and capture some samples. These were found to contain water, methane, ammonia, carbon monoxide, and carbon dioxide. There were also probably some simple organic molecules, though that is a chemical term denoting carbon atoms linked together and does not imply a biological origin. If the plumes had been known in advance, instruments better suited for detection of biomarkers might have been included in Cassini ’s payload.

22.Two Cassini images of Enceladus. Left: an overexposed crescentview, showing a plume extending at least 100 kilometres above thesurface. Right: an oblique view across part of Enceladus cut by severalfamilies of cracks, like those from which the plume is known tooriginate. A few small impact craters (too small for Voyager to see)show that this particular region is probably no longer active

Almost certainly, tidal heating (driven by 2:1 orbital resonance with Saturn’s next-but-one satellite Dione) drives the crack formation and provides the impetus for the plumes. However, no one expected Enceladus to be so active, and this is particularly baffl ing given that its similar-sized neighbour Mimas is an archetypical cratered ice-ball showing no history of activity. It is unlikely that Enceladus has a global ocean hidden below its surface, but there may be pods of liquid water beneath the plume sources. Liquid water is good for life, but the availability of nutrients within Enceladus is surely much more restricted than within a large body such as Europa, so Enceladus does not seem such a promising habitat.

Titan

Titan is Saturn’s only satellite that rivals Jupiter’s Galilean satellites in scale (5,150 kilometres in diameter). Voyager showed it only as a fuzzy orange ball, because – alone among satellites – it has a dense atmosphere. This is 97% nitrogen but is made opaque by methane and its photochemical derivatives that turn the stratosphere into an opaque smog. Titan has a crust and mantle made of ice (mostly water-ice) occupying the outer one-third of Titan’s radius and overlying a rocky core. There could be an iron inner core, in which case, to balance out the average global density, the base of the icy mantle would have to be deeper. Titan’s rotation period is affected by seasonal winds, showing us that the lithosphere must be decoupled from the interior, most likely by an internal ocean. This could be mostly water or a mixture of water and ammonia (which can remain liquid at a considerably lower temperature than pure water). Most models place it as a layer within the icy mantle, rather than situated immediately on top of the internal rock.

The Cassini mission tackled the problem of seeing through to Titan’s surface in three ways: it obtained blurred but usable images of the surface in some narrow bands of the near-infrared spectrum where the smog is least opaque, it used imaging radar like the Magellan Venus orbiter to see the ground irrespective of clouds, and it carried a landing craft, named Huygens , that provided images from below the clouds during parachute descent to the surface. Titan’s surface geological processes revealed by this array of imaging techniques are superb analogues to many of the processes that occur on Earth. The crust is dominantly water-ice, very rigid and rock-like in its behaviour in Titan’s –180 °C surface environment. Huygens came to rest near the equator on a sandy plain strewn with pebbles. It looked like Mars except that both sand and pebbles were made of ice. The sand could have been wind-blown, and indeed radar images reveal vast fi elds of wind-blown sand dunes in other parts of Titan. However, the pebbles must have been transported by fl owing liquid, which, given Titan’s atmospheric composition and surface temperature, must be methane (CH4 ) or ethane(C2H6 ). As it descended, Huygens saw branching drainage channels near to the landing site, and radar imaging reveals complex valley systems in many other regions, starting in highlands where the ‘bedrock’ of icy crust is exposed and draining into lowland basins where sediment accumulates.Better than that, it found lakes of ethane-tainted liquid methane near both poles ( Figure 23 ). Some lake-beds were dry, and others had shallow or marshy fringes, and it is likely that they vary seasonally. Titan is clearly geologically active. A few deeply eroded impact craters have been recognized, and there are some sites of suspected ‘cryovolcanism’, where icy ‘magma’ is erupted analogously to terrestrial lava flows.

23.A 1,100-kilometre-long mosaic of Cassini radar images, nearTitan’s north pole. The dark areas are lakes, the largest of whichexceeds 100,000 square kilometres, 20% bigger than Lake Superior inNorth America. Dendritic drainage channels can be seen feeding thelakes. Lines of longitude have been added; blank areas are unimaged

The extent to which cryovolcanism and tectonic processes contribute to the sculpting and resurfacing of Titan’s surface is unknown. However, it is clear that erosion of bedrock (in this case,ice, of course) followed by transport and deposition of sediment are major players. Rainfall on Titan must consist of droplets of methane that, like rainfall on Earth, infi ltrates the ground and feeds springs that supply streams and rivers. The capacity of methane to react chemically with the icy ‘bedrock’, its erosive power, the rate at which it evaporates back into the atmosphere,and how long it remains there before raining out again are uncertain. All these must be factors in a methanologic cycle that mimics Earth’s hydrologic cycle. Mars had rainfall, rivers, and lakes long ago, but Titan is the only other place where they occur today. One day, we will send another probe to explore Titan more thoroughly – perhaps including a balloon to drift below the smog,with variable buoyancy so it can touch down in interesting places.Such a mission could sample the lake fl uid, and obtain pictures of waves breaking on a thoroughly alien shore.

Miranda and Ariel

Although present-day cryovolcanism on Titan remains controversial, ancient cryovolcanism cannot be doubted on two of Uranus’s fi ve regular satellites, Ariel and Miranda, where the surface temperature is –200 °C. Its effects can be seen on images sent back by Voyager 2 , which fl ew through the Uranus system in January 1986.

Ariel is the larger of the two (1,158 kilometres in diameter). It is a complex globe, whose oldest cratered terrain is cut by numerous faults bounding high-standing blocks. Most of the faults defi ne fl at-fl oored valleys of the kind denoted by the descriptor term‘chasma’. However, rather than preserving down-dropped highland surface, the fl oors of most chasmata have been covered by smooth material, or at least by something that appears smooth at the 1-kilometre resolution of the Voyager images.

Probably in the distant past (more than 2 billion years ago), tidal heating led to fracturing of Ariel’s surface and the effusion of cryovolcanic lava. This covered the fl oors of the chasmata, and in places can be seen spreading beyond them to partly bury some older impact craters. At the distance of Uranus from the Sun, its ice is expected to be a more complex cocktail than the slightly salty ice found on Jupiter’s satellites. The most likely melt to be extracted by partial melting is a 2:1 mixture of water and ammonia. This is liquid at –100 °C and so can be generated by much more modest heating than would be required to melt pure water.

Individual ‘lava’ fl ows can also be seen on Miranda, which is Uranus’s smallest regular satellite (472 kilometres in diameter).For such a tiny body, it has a remarkably diverse surface, probably more varied even than Enceladus, although the number of superimposed impact craters shows that its last activity was probably billions of years ago. Voyager 2 saw only half the globe.Half of that is heavily cratered, but is unusual in that most craters(the older ones) have smooth profi les, as if they have been mantled by something falling from above, and only younger craters are pristine. The other half of the imaged area comprises three sharp-edged terrain units described as coronae. Each is different,but they all contain complex ridged or tonally patterned terrain,including features identifi ed as cryovolcanic lava fl ows (probably water-ammonia lava, as on Ariel), and pocked by pristine craters equivalent to those in the heavily cratered terrain.

An early theory about Miranda that each corona represents a fragment from catastrophic global break-up and re-accretion has been discounted. Most likely, the coronae are sites of cryovolcanism, of which only the waning phase has left recognizable fl ow-like traces. Mantling of the older craters in the terrain beyond the coronae may demonstrate explosive eruptions,spraying icy particles into space, some of which settled, snow-like,to subdue pre-existing topography. When and why this happened,we do not know. We are unlikely to fi nd out until there is another mission to Uranus, which is not likely before mid-century.

Triton

Triton is Neptune’slargest satellite (2,706 kilometres in diameter).Its outer part is icy, but it is dense enough to have a substantial rocky core. When Voyager 2 flew past in 1989, it found polar caps of frozen nitrogen ice (previously detected spectroscopically from Earth). Like the carbon dioxide in Mars’s polar caps, these probably shrink in summer, by sublimation rather than by melting, adding their content to Triton’s thin but respectable atmosphere which is made largely of nitrogen. The stable‘bedrock’ ice forming Triton’s crust appears to be a mixture of methane, carbon dioxide, carbon dioxide, and water. There may be ammonia too, which is almost invisible to optical spectroscopy.

The best images of Triton have a resolution of about 400 metres per pixel. They reveal a geologically complex surface beyond the polar cap, including various landforms that may have been created cryovolcanically ( Figure 24 ). Impact craters occur everywhere, but not in vast numbers, and it is possible that much of the surface is less than a billion years old. Triton is also remarkable for having geysers that erupt through the polar cap, lofting dark particles to a height of about 8 kilometres. There are a few high-altitude clouds made of nitrogen crystals, analogous to cirrus clouds in our own atmosphere.

Only the south polar cap was seen by Voyager 2 , because most of the northern hemisphere was in darkness. Triton’s seasons are peculiar, resulting from a combination of Neptune’s 29.6° axial tilt added to the 21° inclination of Triton’s orbit. Further to this,Trion’s orbital plane precesses about Neptune’s axis so that a full seasonal cycle on Triton equates not to Neptune’s 164-year orbital period, but to a 688-year cycle, with 164-year subcycles superimposed. During the full cycle, the subsolar latitude on Triton ranges between 50° north and 50° south. By chance, when Voyager 2 made its fl y-by, Triton was approaching extreme southern summer, with the Sun overhead at nearly 50° south, so a large part of the northern hemisphere was in darkness and could not be seen. The sunlit southern polar cap showed signs of being in retreat, and its sublimation to gas was verifi ed by observations from Earth in 1997 showing that atmospheric pressure had doubled in the eight years since the Voyager encounter.Meanwhile, the unseen north polar cap was probably growing, as atmospheric nitrogen condensed onto the frigid surface.

24.A mosaic of Voyager images covering a 2,000-kilometre-wideregion of Triton. South is towards the top, and sunlight is coming fromthe upper right. The ragged edge of the south polar cap runs diagonallyacross the top of the image. Long, narrow, curved ridges (sulci) may befi ssures where cryovolcanic icy magma was erupted. The smooth plainsand basins in the lower left are probably expanses of cryovolcanic lava.The dimpled area in the centre and lower right is called ‘canteloupeterrain’, by visual analogy to the skin of a melon, but its origin isunknown